**Integrated Control of Microfluidics – Application in Fluid Routing, Sensor Synchronization, and Real-Time Feedback Control**

Elishai Ezra, Danny Bavli and Yaakov Nahmias

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/64429

#### **Abstract**

Microfluidic applications range from combinatorial chemical synthesis to high-through‐ put screening, with platforms integrating analog perfusion components, digitally control‐ led microvalves, and a range of sensors that demand a variety of communication protocols. A comprehensive solution for microfluidic control has to support an arbitrary combination of microfluidic components and to meet the demand for easy-to-operate sys‐ tem as it arises from the growing community of unspecialized microfluidics users. It should also be an easy to modify and extendable platform, which offer an adequate com‐ putational resources, preferably without a need for a local computer terminal for in‐ creased mobility. Here we will describe several implementation of microfluidics control technologies and propose a microprocessor-based unit that unifies them. Integrated con‐ trol can streamline the generation process of complex perfusion sequences required for sensor-integrated microfluidic platforms that demand iterative operation procedures such as calibration, sensing, data acquisition, and decision making. It also enables the im‐ plementation of intricate optimization protocols, which often require significant compu‐ tational resources. System integration is an imperative developmental milestone for the field of microfluidics, both in terms of the scalability of increasingly complex platforms that still lack standardization, and the incorporation and adoption of emerging technolo‐ gies in biomedical research. Here we describe a modular integration and synchronization of a complex multicomponent microfluidic platform.

**Keywords:** Gadgeteer, integrated control, microprocessor, optimization

## **1. Introduction**

First, we will provide a general review of microfluidics control paradigms and applications, focusing on embedded control.

© 2016 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

#### **1.1. Microfluidics: Applications to control**

Microfluidic technology delivers the potential for high-throughput and high-content studies in a wide spectrum of experimental sciences, from biology to chemistry and physics. It provides the ability to precisely control experiments at the microscale [1–4], and enables easy and fast automation using computer-controlled micromechanical valves [5, 6]. Indeed, the ever-increasing desire to delve into biology at single entity level has made microfluidics remarkably relevant for the biological domain. By virtue of microfluidic's compatible length scale, compliant surface chemistry and the minuscule reagent volume usually required, microfluidics has opened up new research opportunities in the field of biological science, which even a decade ago looked utterly unattainable [7]. For example, Bhatia and colleagues used microfabricated stamps to generate a micropattern for coculturing hepatocytes and 3T3 fibroblasts in the study of cell-cell interactions [2]; Ho and colleagues used an enhanced fieldinduced dielectrophoresis trap to pattern hepatocytes and endothelial cells in a radial pattern to mimic the lobular morphology of liver tissue [8]; and Quake and colleagues developed a microfluidic device consisting of thousands of microfabricated switches for a genomic analysis at single-cell resolution [9].

Major microfluidics platforms include capillary, centrifugal, electrokinetic, and pressuredriven platforms [10]. In laboratory settings, pressure-driven platforms are dominant. Microfluidics handling in pressure-driven platforms often includes the control of three main aspects: fluid perfusion, valves control, and sensors monitoring. Fluid perfusion is mostly implemented using linear actuated devices such as syringe pumps and pressure sources. Syringe pumps are commercially available and can be usually controlled manually or via a computer using standard data protocols such as USB, RS-485, RS-232, or GPBI (IEEE-488). For example, the syringe pump PHD ULTRA by Harvard Apparatus can be controlled manually using a touch screen and embedded software that features sequence templates and method wizards, using standard data protocols. Pressure sources are often found in standard labora‐ tory settings and are also commercially available. Pressure sources can be controlled with a pressure regulator, which can be configured either manually usually by turning a knob that compresses a fine-pitched spring or by a computer. A dominant subclass of computercontrolled regulators is based on proportional converters, where a current or a voltage proportionally sets the level of the output pressure. For example, FESTO's lines of products: VPPE™ feature regulators that proportionally match the regulator output pressure to a range of voltage. The level of voltage can be controlled via a computer-connected control unit (FESTO's setpoint value module). Other companies commercialize more expensive integrated products. For example, Elveflow (powered by National Instruments™) commercializes a line of products called AF-1™ that integrates a pressure source and a computer-controlled regulator. While several companies commercialize integrated control products, they are usually configured using closed-source software, which is often limited to a specific line of products.

Since its original development in the laboratory of Prof. Stephen Quake, the microfluidic valve became the basic unit of fluid handling in pressure-driven microfluidic-based platforms and it plays a role analogous to that of the transistor in semiconductor electronics [11]. Today's conventional biological automation paradigm is being replaced by the integration of micro‐ fluidic devices with mechanical valves [12]. This technology enables the automation of experiments and has been used in applications such as protein crystallography [13], genetic analysis [14], amino acid analysis [15], high-throughput screening [12], bioreactors [16], chemical synthesis [17], and single-cell analysis [18]. The current design of a microfluidic valve uses air pressure or thermal actuation [19] to control microstructured switches fabricated from two-layered polydimethylsiloxane (PDMS) [11]. During the fabrication process, a membrane is formed where the control channel and flow channel intersect orthogonally, constructing a valve. Current state-of-the-art microfluidic devices integrate numerous microvalves (micro‐ fluidics large-scale integration (LSI)) that need to be independently or jointly controlled using individual pressure lines. The main control method of microfluidic valves is based on a branched pressure source implemented with pressure manifolds. Each manifold outlet can be digitally controlled with specialized hardware/software, which is commercially available. For example, FESTO commercialized a modular pressure manifold called MH-1 that can be configured to integrate an arbitrary number of pressure lines. Each pressure line requires a controllable digital line. The Microfluidics Foundry at Stanford University established one popular method of controlling manifolds' digital lines. They distributed a microcontroller and an integrated circuit that can use a computer to configure each of the manifold's pressure lines independently via MATLAB® or LabVIEW (https://goo.gl/dnWFKX). Other companies have commercialized specialized systems for controlling multiple pressure lines. For example, Elveflow® commercializes a line of products called OB-1TM that enable independent control of four pressure lines and FluigentTM commercializes OEM, a line of products that enable the control of up to eight pressure lines.

Digital microfluidic is emerging technology for precise control and manipulation of discrete liquid droplets, and it is based on an array of self-addressable electrodes that control the drops' surface tension. Control of such system is readily integrated since conventional pumps, valves, or channels are not involved [20]. In the past decade, digital microfluidics has been applied to a range of problems in biology, chemistry, and medicine. For example, Walker and colleagues used digital microfluidics for electrophoretic separations [21], Schmalzing and colleagues used it for DNA analysis [22], and Figeye and colleagues used it for protein/enzyme analysis [23]. Control of such microfluidics platforms is based on activate/deactivate electrodes patterned in the device.

#### **1.2. Embedded control**

**1.1. Microfluidics: Applications to control**

166 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

at single-cell resolution [9].

products.

Microfluidic technology delivers the potential for high-throughput and high-content studies in a wide spectrum of experimental sciences, from biology to chemistry and physics. It provides the ability to precisely control experiments at the microscale [1–4], and enables easy and fast automation using computer-controlled micromechanical valves [5, 6]. Indeed, the ever-increasing desire to delve into biology at single entity level has made microfluidics remarkably relevant for the biological domain. By virtue of microfluidic's compatible length scale, compliant surface chemistry and the minuscule reagent volume usually required, microfluidics has opened up new research opportunities in the field of biological science, which even a decade ago looked utterly unattainable [7]. For example, Bhatia and colleagues used microfabricated stamps to generate a micropattern for coculturing hepatocytes and 3T3 fibroblasts in the study of cell-cell interactions [2]; Ho and colleagues used an enhanced fieldinduced dielectrophoresis trap to pattern hepatocytes and endothelial cells in a radial pattern to mimic the lobular morphology of liver tissue [8]; and Quake and colleagues developed a microfluidic device consisting of thousands of microfabricated switches for a genomic analysis

Major microfluidics platforms include capillary, centrifugal, electrokinetic, and pressuredriven platforms [10]. In laboratory settings, pressure-driven platforms are dominant. Microfluidics handling in pressure-driven platforms often includes the control of three main aspects: fluid perfusion, valves control, and sensors monitoring. Fluid perfusion is mostly implemented using linear actuated devices such as syringe pumps and pressure sources. Syringe pumps are commercially available and can be usually controlled manually or via a computer using standard data protocols such as USB, RS-485, RS-232, or GPBI (IEEE-488). For example, the syringe pump PHD ULTRA by Harvard Apparatus can be controlled manually using a touch screen and embedded software that features sequence templates and method wizards, using standard data protocols. Pressure sources are often found in standard labora‐ tory settings and are also commercially available. Pressure sources can be controlled with a pressure regulator, which can be configured either manually usually by turning a knob that compresses a fine-pitched spring or by a computer. A dominant subclass of computercontrolled regulators is based on proportional converters, where a current or a voltage proportionally sets the level of the output pressure. For example, FESTO's lines of products: VPPE™ feature regulators that proportionally match the regulator output pressure to a range of voltage. The level of voltage can be controlled via a computer-connected control unit (FESTO's setpoint value module). Other companies commercialize more expensive integrated products. For example, Elveflow (powered by National Instruments™) commercializes a line of products called AF-1™ that integrates a pressure source and a computer-controlled regulator. While several companies commercialize integrated control products, they are usually configured using closed-source software, which is often limited to a specific line of

Since its original development in the laboratory of Prof. Stephen Quake, the microfluidic valve became the basic unit of fluid handling in pressure-driven microfluidic-based platforms and it plays a role analogous to that of the transistor in semiconductor electronics [11]. Today's An embedded system is a microprocessor-based control platform, built to manage and regulate a definite range of functions with a limited user interface and a self-contained on-chip memory [24]. In virtually every control-embedded system, the goal is to control an aspect of an electromechanical system—from laser microscopy to digital cameras. Microprocessors are deeply ingrained into modern day life with over 6 billion new microprocessors used each year [25]. Modern cars, for example, may have tens of microprocessors controlling different functions. One of them, the engine management system, controls the fuel mixture and ignition, alters the parameters and timing based on independent sensor-derived real-time data analysis, and sends indication status to the driver [24]. An embedded system consists of a processor, which provides the computational resources; a nonvolatile memory that contains initialization routines and software; peripherals that include sensors and input/output interfaces; software which constitutes the initialization and configuration procedures, the operating system, and the application; and algorithms, which are key constituents of the software and can range from mathematical processing to models of the external environment which are used to interpret information from external sensors. A typical architecture of an embedded control system is shown in Figure 1.

**Figure 1.** A typical architecture of an embedded control system.

The two most important complimentary technologies in embedded systems design are Application-Specific Integrated Circuits (ASICs) [26] and System on a Chip (SoC) [27]. ASIC is a customized integrated circuit, which was designed for a particular use. It can be designed to efficiently implement combinatorial logic circuits in very high density of up to 100 million gates/chip using hardware description languages. ASICs can be found in a wide spectrum of electronic devices, from custom real-time image processing [28, 29] to speech synthesis [30] and acoustic features extraction [31]. Because ASICs are custom-made, they are only available to the company, which designed them and are considered proprietary technology. Modern ASICs, include a CPU, memory controller, main memory, I/O control, and the various buses and interconnects. For example, ASIC of SoC architecture for speech recognition and speech compression may contain a microprocessor, a DSP (Digital Signal Processor, which was optimized for digital signal processing), two codecs (coder/decoder—capable of encoding or decoding a digital data streams), and input/output analog channels [32].

Another important design strategy in embedded design is the Field Programmable Gate Array (FPGA) [33]. The user configures FPGA after manufacturing (in contrary to ASICs, which are built according to a specific design layout). FPGAs contain arrays of programmable logic blocks and a hierarchy of reconfigurable interconnects that allow blocks to be combinatorially wired [34].

The development of such systems requires extensive expertise across electronic circuit design (logic, circuit, and physical), PCB layout, and system programming. A schematic of a typical design process for ASIC-based board is shown in Figure 2. While the design of ASICs and FPGAs can potentially produce extremely efficient designs, their impact is very limited because they cannot be utilized for the exploration or validation of new product concepts [35].

**Figure 2.** A typical design process of ASIC-based microprocessor.

## **2. Open-source microcontrollers**

and sends indication status to the driver [24]. An embedded system consists of a processor, which provides the computational resources; a nonvolatile memory that contains initialization routines and software; peripherals that include sensors and input/output interfaces; software which constitutes the initialization and configuration procedures, the operating system, and the application; and algorithms, which are key constituents of the software and can range from mathematical processing to models of the external environment which are used to interpret information from external sensors. A typical architecture of an embedded control system is

168 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

The two most important complimentary technologies in embedded systems design are Application-Specific Integrated Circuits (ASICs) [26] and System on a Chip (SoC) [27]. ASIC is a customized integrated circuit, which was designed for a particular use. It can be designed to efficiently implement combinatorial logic circuits in very high density of up to 100 million gates/chip using hardware description languages. ASICs can be found in a wide spectrum of electronic devices, from custom real-time image processing [28, 29] to speech synthesis [30] and acoustic features extraction [31]. Because ASICs are custom-made, they are only available to the company, which designed them and are considered proprietary technology. Modern ASICs, include a CPU, memory controller, main memory, I/O control, and the various buses and interconnects. For example, ASIC of SoC architecture for speech recognition and speech compression may contain a microprocessor, a DSP (Digital Signal Processor, which was optimized for digital signal processing), two codecs (coder/decoder—capable of encoding or

Another important design strategy in embedded design is the Field Programmable Gate Array (FPGA) [33]. The user configures FPGA after manufacturing (in contrary to ASICs, which are built according to a specific design layout). FPGAs contain arrays of programmable logic

decoding a digital data streams), and input/output analog channels [32].

shown in Figure 1.

**Figure 1.** A typical architecture of an embedded control system.

Throughout the last decade, engineers have been exploiting the notion of hardware and software abstraction (component-based architectures) to speed the development of increas‐ ingly complex control systems [36]. For example, the European Disappearing Computer Initiative explored different options for creating a modular system for prototyping and proposed a common processing and communications circuit board with a variety of ways to connect sensors and actuators [37]. Similar to hardware, software was also based on a modular approach, containing core libraries for control and communication. Simultaneously, "Wiring" was developed (http://wiring.org.co)—a microcontroller coupled with an accessible program‐ ming language (which was built upon "Processing" [38]) that targeted artists and designers. The Arduino system has taken the notion of low-cost and accessible components and tools much further. Arduino has arguably become the standard open-source platform for physical prototyping, with a vibrant developers' community and a tremendous range of supported hardware. Arduino is an open-source hardware board, designed around an 8-bit Atmel AVR microcontroller, which can be programmed with C or C++, and it was used numerous times in laboratory settings (Figure 3). For example, the Arduino Geiger was designed to implement a radiation detector, pHduino was designed to implement a pH meter, Xoscillo was designed to implement an oscilloscope, and OpenPCR was designed to perform DNA analysis [39]. Over the past few years, several additional platforms have been developed that complement the Arduino world in various ways. One example is the Raspberry Pi, which is manufactured by Newark Corporation (Figure 3). Raspberry Pi functions as a computer, commonly operated with Linux, and includes most of the components found on a regular computer: CPU, GPU, memory, USB ports, and video/audio inputs and outputs. Another example is the Gadgeteer, open-source microcontroller commercialized by Microsoft and built upon the.Net microfra‐ mework (Figure 3). The Gadgeteer FEZ Spider Mainboard features a 72 MHz 32-bit ARM7 Processor, 16 MB RAM, LCD controller, and a full support of TCP/IP Stack with SSL, HTTP, TCP, UDP, and DHCP. The Gadgeteer can be connected to Ethernet, a WIFI driver and 3D modems. It supports standard data protocols such as USB, SPI, and I2 C, and include UART (enabling communication via EIA, RS-232, RS-422, RS-485). By using expansion modules, the Gadgeteer can be used for wireless communication with Bluetooth, radio, GPS, Xbee WIFI, IR, and RFID. Importantly, the FEZ spider mainboard features 76 GPIOs (general purpose inputs/ outputs) digital lines, which can be individually addressed and configured. Another important aspect is the wide range of sensor modules, which were especially designed for the Gadgeteer.

**Figure 3.** Open-source microcontrollers.

## **3. Microfluidic control system**

Here will review several integrated control platforms for digital and pressure-driven micro‐ fluidics.

#### **3.1. Dropbot: Integrated control for digital microfluidics platforms**

As was described above, a control of a digital-microfluidic device is based on activate/ deactivate electrodes, which were patterned on the device for the direct manipulation of droplets with no need of actuating mechanical valves nor pressure sources and pumps. In 2013, Fobel and colleagues introduced the design of an open-source control and automation system termed DropBot, which enables manipulation of drop's position and velocity by driving up to 320 independent electrodes [40]. The DropBot is based on an Arduino microcontroller board and is consists of a high-voltage amplifier, high-voltage driver boards, a webcam, and a PC. The system continuously monitors the amplifier output and device impedance to maintain a stable actuation voltage and to track the position and velocity of drops. A schematic of the DropBot control system is shown in Figure 4.

**Figure 4.** Schematic of the DropBot control system.

### **3.2. Control of microfluidic routing**

hardware. Arduino is an open-source hardware board, designed around an 8-bit Atmel AVR microcontroller, which can be programmed with C or C++, and it was used numerous times in laboratory settings (Figure 3). For example, the Arduino Geiger was designed to implement a radiation detector, pHduino was designed to implement a pH meter, Xoscillo was designed to implement an oscilloscope, and OpenPCR was designed to perform DNA analysis [39]. Over the past few years, several additional platforms have been developed that complement the Arduino world in various ways. One example is the Raspberry Pi, which is manufactured by Newark Corporation (Figure 3). Raspberry Pi functions as a computer, commonly operated with Linux, and includes most of the components found on a regular computer: CPU, GPU, memory, USB ports, and video/audio inputs and outputs. Another example is the Gadgeteer, open-source microcontroller commercialized by Microsoft and built upon the.Net microfra‐ mework (Figure 3). The Gadgeteer FEZ Spider Mainboard features a 72 MHz 32-bit ARM7 Processor, 16 MB RAM, LCD controller, and a full support of TCP/IP Stack with SSL, HTTP, TCP, UDP, and DHCP. The Gadgeteer can be connected to Ethernet, a WIFI driver and 3D

(enabling communication via EIA, RS-232, RS-422, RS-485). By using expansion modules, the Gadgeteer can be used for wireless communication with Bluetooth, radio, GPS, Xbee WIFI, IR, and RFID. Importantly, the FEZ spider mainboard features 76 GPIOs (general purpose inputs/ outputs) digital lines, which can be individually addressed and configured. Another important aspect is the wide range of sensor modules, which were especially designed for the Gadgeteer.

Here will review several integrated control platforms for digital and pressure-driven micro‐

As was described above, a control of a digital-microfluidic device is based on activate/ deactivate electrodes, which were patterned on the device for the direct manipulation of

**3.1. Dropbot: Integrated control for digital microfluidics platforms**

C, and include UART

modems. It supports standard data protocols such as USB, SPI, and I2

170 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

**Figure 3.** Open-source microcontrollers.

fluidics.

**3. Microfluidic control system**

Fabrication and characterization of a microfluidic valve

Microfabrication techniques have been developed in the microelectronics industry to create complex electronic circuits with minimum feature sizes currently as small as 7 nm [41]. Techniques for creating the insulating or conducting features include physical vapor deposi‐ tion, chemical vapor deposition, sputter coating, molecular beam epitaxy, and chemical beam epitaxy [42]. Integrated circuits are created using deposition of conductive areas in a process known as photolithography [43]. Photolithography allows the creation of complex geometric patterns with small feature sizes and it is widely utilized for microfluidics and lab-on-a-chip fabrication. Photolithography is commonly performed using a light sensitive material (pho‐ toresist) and a photomask. A positive photoresist becomes soluble where it has been exposed to light, while a negative photoresist becomes insoluble where it has been exposed to light. Following exposure, devices are typically heated and treated with solvents to remove the soluble photoresist and develop a cured device. Using conventional optical techniques, the minimum feature size that can be created is limited by the diffraction limit of light. For ultraviolet (UV) light (248 nm) features as small at 70 nm can be defined. For microfluidic applications, a variation known as soft photolithography is often used. The hardened product of standard photolithography is used as a master and elastomeric PDMS is poured and cured over it. The channels' width is determined by the photomask and the channels' height is determined by the thickness of the patterned photoresist. PDMS is often chosen for chemical and biological experiments due to its low cost, high oxygen permeability, fast curing time, hydrophobicity, and low toxicity. PDMS forms an irreversible bond to glass using plasma activation, allowing its use with high magnification inverted microscope objectives. Since three-dimensional (3D) control of features is generally limited to time-consuming layer-bylayer alignment [44], most microfluidic designs are limited to two-dimensional (2D) geome‐ tries. However, recent fabrication techniques such as 3D printing aim to change that [45, 46]. For example, Bhargava and colleagues showed that 3D printing could be used to rapidly fabricate discrete microfluidic elements that can be assembled into complex 3D circuits [47].

The monolithic micromechanical valve is constructed from two PDMS replica moldings, which are bonded together using plasma activation and curing agent diffusion. One layer defines the control channels, and the other defines the flow channels. Membranes are formed at the intersection of the two layers, allowing valves actuation [11]. A micromechanical valve can be configured in two formations: "push-down" and "push-up". While in a "push-down" formation, the flow layer is aligned below the control layer, in a "push-up" formation the flow layer is aligned above the control layer (Figure 5, left). A "push-down" formation exposes the flow layer to the glass, facilitating prebonding surface modification. Actuation pressure depends on valve dimensions and membrane thickness. However, typically, "push-down" formation requires higher actuation pressure for proper switching relatively to a "push-up" formation—a fact that significantly limits the flow channel height. Performance of both valve formations is highly dependent on the flow channel cross-sectional profile [1]. Valves with square flow channel profiles leave pockets of fluid flow, causing leakage (Figure 5, left). Perfectly sealed valves require rounded flow channels, which are typically fabricated using a thermal reflow of photoresist (Figure 5, right). Generally, all resists which do not crosslink have a certain softening point and thus can be used for reflow. While most negative resists cross-link, all common positive resists do not cross-link and start to soften at approximately 100–130ºC.

Here, microfluidic valves were fabricated, actuated, and visualized using a color dye (Figure 6A). Valves were characterized by leakage and response time, which were analyzed repeatedly after 100, 7500, and 15,000 switching cycles to ensure mechanical endurance. Briefly, leakage was calculated with negatively pressurized channels, each withdraws different color dye. Response time was calculated from high frame rate recordings of the valve actuation process. Switching pressure (0% leakage) was measured as 4.5 PSI. No significant changes were noticed after 15,000 switching cycles (Figure 6B). Switches were utilized to create flow cycles of red and blue color dyes in a microfluidic channel for evaluation. The measured total response time is 28 ms, and a mechanical response time of 20 ms.

Integrated Control of Microfluidics – Application in Fluid Routing, Sensor Synchronization, and Real-Time... http://dx.doi.org/10.5772/64429 173

**Figure 5.** "push-up" and "push-down" valves formations [1].

soluble photoresist and develop a cured device. Using conventional optical techniques, the minimum feature size that can be created is limited by the diffraction limit of light. For ultraviolet (UV) light (248 nm) features as small at 70 nm can be defined. For microfluidic applications, a variation known as soft photolithography is often used. The hardened product of standard photolithography is used as a master and elastomeric PDMS is poured and cured over it. The channels' width is determined by the photomask and the channels' height is determined by the thickness of the patterned photoresist. PDMS is often chosen for chemical and biological experiments due to its low cost, high oxygen permeability, fast curing time, hydrophobicity, and low toxicity. PDMS forms an irreversible bond to glass using plasma activation, allowing its use with high magnification inverted microscope objectives. Since three-dimensional (3D) control of features is generally limited to time-consuming layer-bylayer alignment [44], most microfluidic designs are limited to two-dimensional (2D) geome‐ tries. However, recent fabrication techniques such as 3D printing aim to change that [45, 46]. For example, Bhargava and colleagues showed that 3D printing could be used to rapidly fabricate discrete microfluidic elements that can be assembled into complex 3D circuits [47].

172 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

The monolithic micromechanical valve is constructed from two PDMS replica moldings, which are bonded together using plasma activation and curing agent diffusion. One layer defines the control channels, and the other defines the flow channels. Membranes are formed at the intersection of the two layers, allowing valves actuation [11]. A micromechanical valve can be configured in two formations: "push-down" and "push-up". While in a "push-down" formation, the flow layer is aligned below the control layer, in a "push-up" formation the flow layer is aligned above the control layer (Figure 5, left). A "push-down" formation exposes the flow layer to the glass, facilitating prebonding surface modification. Actuation pressure depends on valve dimensions and membrane thickness. However, typically, "push-down" formation requires higher actuation pressure for proper switching relatively to a "push-up" formation—a fact that significantly limits the flow channel height. Performance of both valve formations is highly dependent on the flow channel cross-sectional profile [1]. Valves with square flow channel profiles leave pockets of fluid flow, causing leakage (Figure 5, left). Perfectly sealed valves require rounded flow channels, which are typically fabricated using a thermal reflow of photoresist (Figure 5, right). Generally, all resists which do not crosslink have a certain softening point and thus can be used for reflow. While most negative resists cross-link, all common positive resists do not cross-link and start to soften at approximately

Here, microfluidic valves were fabricated, actuated, and visualized using a color dye (Figure 6A). Valves were characterized by leakage and response time, which were analyzed repeatedly after 100, 7500, and 15,000 switching cycles to ensure mechanical endurance. Briefly, leakage was calculated with negatively pressurized channels, each withdraws different color dye. Response time was calculated from high frame rate recordings of the valve actuation process. Switching pressure (0% leakage) was measured as 4.5 PSI. No significant changes were noticed after 15,000 switching cycles (Figure 6B). Switches were utilized to create flow cycles of red and blue color dyes in a microfluidic channel for evaluation. The measured total response time

100–130ºC.

is 28 ms, and a mechanical response time of 20 ms.

**Figure 6.** Valves evaluation and characterization. Valves were visualized using color dyes and inspected with a micro‐ scope inspection. Leakage was quantified after 100, 7500, and 15,000 switching cycles.

#### **4. Microfluidic large-scale integration**

The incorporation of microfluidic valves in a fluidic circuits has enabled the creation of a variety of functional fluidic modules such as micromixers [48, 49], gradient generators [49], multiplexers [50, 51] and peristaltic pumps [51]. For example, Gomez and colleagues devel‐ oped a cell culture system, which enables automated culture of 96 cell chambers in parallel that feeds from 16 different inlets through an integrated mixer, a pump, and a multiplexer [52]. The 16 fluid inlets were connected in a binary tree manifold that features an equal fluidic resistance in all branches. The root of the tree was connected to a mixer, which delivers the fluid to a multiplexer for distribution to the chambers. System schematic is shown in Figure 7. Medium injection into the chambers was perfused with an on-chip peristaltic pump, which was implemented using a series of three valves. In this setup, the microfluidic valves were USB controlled with custom electronic units via pneumatic solenoid valves. A different control module, which was connected to a microscope, controlled the temperature and gas composi‐ tion. Independently, cells were imaged in two-hour intervals. A pressure-driven perfusion system was used to seed the cells and to drive reagents. A custom MATLAB application was written for automatic operation of the actuation sequences based on the user-supplied schedule.

**Figure 7.** Schematic of an automated culture of 96 cell chambers.

#### **5. Integrated control of pressure-driven microfluidic-based platforms**

#### **5.1. Design and specifications**

Pressure-driven microfluidic-based platforms are composed of discrete analog and digital perfusion components and sensors that have separate communication protocols, power requirements, and control interfaces limiting system integration. To address this, we designed a control unit based on Gadgeteer FEZ Spider mainboard containing 32-bit ARM7 microproc‐ essor and 11 MB of user available RAM, extended with the Hub AP5 board for a total of 23 control sockets [53]. To bridge between the control unit and the perfusion components, we designed a microfluidic shield that bridges variable power consumption, modulates power distribution from a standard AC source, and implements control of analog lines. Specifically, analog signal modulation was carried out using passive linear voltage dividers, featuring an output range of 0–10 V and a 7-bit SPI controlled potentiometer with volatile memory. The shield permits a rapid replacement of through-holes resistors in each voltage-dividing module, to modulate the system dynamic range. In addition, as solenoid valve actuation requires 1 W (5/10 V), we implemented a series of high current Darlington transistor arrays supporting 24 units of 1 W (5/10 V) digital lines, each line connected to two Darlington channels supplying up to 0.5 A each. Our Darlington transistors have an operating delay time of 0.15 µs and a turnoff delay of 1.8 µs enabling 0.5 MHz of switching capabilities, significantly faster than solenoid response time of 30–50 ms. System schematics is shown in Figure 8. Importantly, our system provides the user with hardware plug and play interface embedded within the Microsoft.Net microenvironment that enables rapid prototyping utilizing programs such as MATLAB for user interface design.

multiplexers [50, 51] and peristaltic pumps [51]. For example, Gomez and colleagues devel‐ oped a cell culture system, which enables automated culture of 96 cell chambers in parallel that feeds from 16 different inlets through an integrated mixer, a pump, and a multiplexer [52]. The 16 fluid inlets were connected in a binary tree manifold that features an equal fluidic resistance in all branches. The root of the tree was connected to a mixer, which delivers the fluid to a multiplexer for distribution to the chambers. System schematic is shown in Figure 7. Medium injection into the chambers was perfused with an on-chip peristaltic pump, which was implemented using a series of three valves. In this setup, the microfluidic valves were USB controlled with custom electronic units via pneumatic solenoid valves. A different control module, which was connected to a microscope, controlled the temperature and gas composi‐ tion. Independently, cells were imaged in two-hour intervals. A pressure-driven perfusion system was used to seed the cells and to drive reagents. A custom MATLAB application was written for automatic operation of the actuation sequences based on the user-supplied

174 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

schedule.

**Figure 7.** Schematic of an automated culture of 96 cell chambers.

**5.1. Design and specifications**

**5. Integrated control of pressure-driven microfluidic-based platforms**

Pressure-driven microfluidic-based platforms are composed of discrete analog and digital perfusion components and sensors that have separate communication protocols, power requirements, and control interfaces limiting system integration. To address this, we designed a control unit based on Gadgeteer FEZ Spider mainboard containing 32-bit ARM7 microproc‐ essor and 11 MB of user available RAM, extended with the Hub AP5 board for a total of 23 control sockets [53]. To bridge between the control unit and the perfusion components, we designed a microfluidic shield that bridges variable power consumption, modulates power

**Figure 8.** Schematic of the microprocessor-based control unit for pressure-driven microfluidic-based platforms.

#### **5.2. Gadgeteer microcontroller**

The.NET Gadgeteer is a rapid open-source development platform that utilizes mainboards and plug-and-play modules that maintained standard connections by Microsoft. The.NET Micro Framework combines the advantages of object-oriented programming, solderless assembly of electronics, and support for customizable physical design [54]. A.NET Gadgeteer system is composed of a mainboard containing an embedded processor and a variety of modules, which connect to the mainboard through a simple plug-and-play interface. There is a great variety of.NET Gadgeteer modules, which are currently available, including: display, camera, networking, storage, and a variety of sensors and input controls. New modules are constantly developed. The.NET Gadgeteer mainboard's sockets are numbered, and each is also labeled with one or more letters, which indicate which modules, can be plugged into it (Figure 9). The.NET Gadgeteer devices are programmed in C Sharp using the.NET Micro Framework via Visual Studio in a desktop, web or phone IDE. An intuitive visual designer and advanced auto code generation capabilities are also provided.

**Figure 9.** FEZ Spider Gadgeteer mainboard layout.

Gadgeteer mainboards expose their I/O interface through sockets. Each socket is a 10-way connector, with pins labeled 1 through 10. Mainboard sockets support one or more different types. A letter represents each socket type. When a mainboard socket is labeled with a socket type letter, it guarantees a particular set of electrical connections and interfaces on the sockets pins [55]. The full hardware layout of the integrated control system (from the IDE perspective) is presented in Figure 10. Gadgeteer mainboard and extension modules were purchased from GHI electronics (Madison Heights, MI). The control unit was composed of a 32-bit ARM7 microprocessor mounted on the Microsoft FEZ Spider Gadgeteer mainboard. Mainboard functionality was extended with Hub AP5 module that adds 9 additional sockets, a USB client Dual Power module to enable microprocessor programming, a TE35 LCA 3.5" touchscreen display module, USB and RS-232 modules that allows the control of sensors and peripherals. A serial camera L1 module was added directly for optical inspection. Finally, Ethernet ENC28 module provided a TCP/IP operation mode.

**Figure 10.** Hardware layout of the integrated control system from the IDE perspective.

#### **5.3. High-current interface circuit**

**5.2. Gadgeteer microcontroller**

auto code generation capabilities are also provided.

176 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

**Figure 9.** FEZ Spider Gadgeteer mainboard layout.

The.NET Gadgeteer is a rapid open-source development platform that utilizes mainboards and plug-and-play modules that maintained standard connections by Microsoft. The.NET Micro Framework combines the advantages of object-oriented programming, solderless assembly of electronics, and support for customizable physical design [54]. A.NET Gadgeteer system is composed of a mainboard containing an embedded processor and a variety of modules, which connect to the mainboard through a simple plug-and-play interface. There is a great variety of.NET Gadgeteer modules, which are currently available, including: display, camera, networking, storage, and a variety of sensors and input controls. New modules are constantly developed. The.NET Gadgeteer mainboard's sockets are numbered, and each is also labeled with one or more letters, which indicate which modules, can be plugged into it (Figure 9). The.NET Gadgeteer devices are programmed in C Sharp using the.NET Micro Framework via Visual Studio in a desktop, web or phone IDE. An intuitive visual designer and advanced

Gadgeteer mainboards expose their I/O interface through sockets. Each socket is a 10-way connector, with pins labeled 1 through 10. Mainboard sockets support one or more different types. A letter represents each socket type. When a mainboard socket is labeled with a socket type letter, it guarantees a particular set of electrical connections and interfaces on the sockets pins [55]. The full hardware layout of the integrated control system (from the IDE perspective) is presented in Figure 10. Gadgeteer mainboard and extension modules were purchased from GHI electronics (Madison Heights, MI). The control unit was composed of a 32-bit ARM7 microprocessor mounted on the Microsoft FEZ Spider Gadgeteer mainboard. Mainboard functionality was extended with Hub AP5 module that adds 9 additional sockets, a USB client Dual Power module to enable microprocessor programming, a TE35 LCA 3.5" touchscreen display module, USB and RS-232 modules that allows the control of sensors and peripherals.

The high-current interface circuit was designed using DipTrace software and printed in a twolayer layout at Beta LAYOUT (Aarbergen, Germany). Controller unit case was designed and fabricated in the Hebrew University workshop. Our control system required a branched voltage module that integrates numerous voltage stabilization circuits and an array of capacitors that filter high-frequency ripple voltages and spikes. The control unit is connected to a standard 220 V, 50 Hz AC power outlet, which was electromagnetically inducted to 5 V using Mean Well NES-35-5 transformer (New Taipei City, Taiwan) that powers the micro‐ processor and the digital control unit and to 26 V using Mean Well NES-2-24 transformer that powers the pressure regulators and the analog voltage module via an 18 V linear voltage regulator purchased from Toshiba (Tokyo, Japan). The 24 controllable high current digital lines were controlled by the general-purpose input output (GPIO) sockets via a series of high current TD62783 Darlington transistor arrays purchased from Toshiba. Each driver supports 3.3 V logic operation and a 5 V driving voltage. The analog control module was composed of passive linear voltage dividers, which were implemented using a passive resistor and a 7-bit SPIcontrolled MCP4131 potentiometer with volatile memory purchased from Microchip (Chan‐ dler, AZ). Gadgeteer driver library was expanded to support the SPI data protocol controlling the potentiometer.

## **6. Integrated control for combinatorial mixing and real-time sensors monitoring**

Microfluidic integrated control provides a straightforward way to automate complex proce‐ dures such as mixing, sensors monitoring, and implementation of feedback loops for execution of optimization protocols. In a recent work, we described a utilization of a microfluidic-based system, illustrated in Figure 11, to combinatorially mix samples for the creation of a target color, which was specified by the user, using an implementation of different optimization algorithms such as genetic algorithm, simulated annealing, multiple hill climbing, and random walk. Briefly, the control unit was connected to analog pressure regulators driving two 12 valve pressure manifolds that control a microfluidic switchboard and positive-pressure perfusion. The switchboard fed into an equipressure combinatorial mixer containing an inspection chamber monitored by a UART-controlled optical CCD sensor connected back to the FEZ Spider mainboard, completing the circuit. Flow is driven by positive pressure provided by a second, independently controlled pressure manifold. Pressure across the manifold is held constant by an analog regulator, resulting in an equal pressure distribution on all open ports, and a constant fluid velocity irrespective of the number and combination of inputs. The generated mixture of colors is inspected optically in the inspection chamber. Acquired data is communicated to the control unit that makes the decision regarding the next mixing stage according to the specified optimization protocol.

The growing field of interest in the integration of microfluidics further complicates integrated control with sensors [56]. A growing focus is given to microphysiometers devices that are able to noninvasively measure a metabolic parameter of living cells on a chip [57]. Most commonly, those parameters include cellular acidification, cellular adhesion, oxygen consumption, and energy metabolites such as glucose uptake and lactate production. Sensor monitoring is accomplished by specialized hardware/software via different measures, most commonly being changes in impedance, current, and electric potential. For example, Molecular Devices® commercialized Cytosensor® to measure extracellular acidification [58], as well as lactate and glucose levels [59]. Bionas® commercialized the Discovery 2500 SystemTM that continuously provides measurements of oxygen consumption and cell impedance [60]. Sensor monitoring is accomplished by specialized hardware/software via different measures, most commonly being changes in impedance, current, and electric potential [61]. Others are based on light emission, ion selective field effect transistors, and resonant frequency [62, 63]. Each of the sensing unit requires a unique hardware/software module, which is able to capture measure, record, analyze, and display it to the user. For example, Molecular Devices® offers variety of

**Figure 11.** Schematic of the integrated control circuit for combinatorial mixing and real-time sensors monitoring.

data acquisition and analysis software packages, such as the SoftMax®, which serves as the user interface to their popular sensing modules.

Recently, we utilized the switchboard and the control unit for real time monitoring of envi‐ ronmental toxicity requires evaluating the presence of unknown toxins in drinking water and soil samples. The system schematic is illustrated in Figure 12. Briefly, human Huh7 cells were seeded in a microfluidic chamber at a density of 106 cells/ml producing a confluence mono‐ layer. Cells were exposed to 100 µM of the pesticide rotenone, a mitochondrial complex I inhibitor. Perfusate was connected to a high-resistance waste syringe and a single inlet of the microfluidic switchboard for automated sampling (Figure 12A). We used the microfluidic switchboard to perfuse a sequence of buffer, air, sample, and air over an enzymatic-ampero‐ metric lactate sensor every two hours. Introduction of air acts as a diffusion barrier preventing sample contamination and providing zero-point calibration. The electrochemical sensor was connected to an embedded potentiostat that can communicate with the Gadgeteer using a UART protocol. A fully automated 16.5 hours experiment was programed allowing the derivation of a time-of-death of 5.5 hours.

## **7. Emerging control paradigms**

TD62783 Darlington transistor arrays purchased from Toshiba. Each driver supports 3.3 V logic operation and a 5 V driving voltage. The analog control module was composed of passive linear voltage dividers, which were implemented using a passive resistor and a 7-bit SPIcontrolled MCP4131 potentiometer with volatile memory purchased from Microchip (Chan‐ dler, AZ). Gadgeteer driver library was expanded to support the SPI data protocol controlling

178 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

**6. Integrated control for combinatorial mixing and real-time sensors**

mixing stage according to the specified optimization protocol.

Microfluidic integrated control provides a straightforward way to automate complex proce‐ dures such as mixing, sensors monitoring, and implementation of feedback loops for execution of optimization protocols. In a recent work, we described a utilization of a microfluidic-based system, illustrated in Figure 11, to combinatorially mix samples for the creation of a target color, which was specified by the user, using an implementation of different optimization algorithms such as genetic algorithm, simulated annealing, multiple hill climbing, and random walk. Briefly, the control unit was connected to analog pressure regulators driving two 12 valve pressure manifolds that control a microfluidic switchboard and positive-pressure perfusion. The switchboard fed into an equipressure combinatorial mixer containing an inspection chamber monitored by a UART-controlled optical CCD sensor connected back to the FEZ Spider mainboard, completing the circuit. Flow is driven by positive pressure provided by a second, independently controlled pressure manifold. Pressure across the manifold is held constant by an analog regulator, resulting in an equal pressure distribution on all open ports, and a constant fluid velocity irrespective of the number and combination of inputs. The generated mixture of colors is inspected optically in the inspection chamber. Acquired data is communicated to the control unit that makes the decision regarding the next

The growing field of interest in the integration of microfluidics further complicates integrated control with sensors [56]. A growing focus is given to microphysiometers devices that are able to noninvasively measure a metabolic parameter of living cells on a chip [57]. Most commonly, those parameters include cellular acidification, cellular adhesion, oxygen consumption, and energy metabolites such as glucose uptake and lactate production. Sensor monitoring is accomplished by specialized hardware/software via different measures, most commonly being changes in impedance, current, and electric potential. For example, Molecular Devices® commercialized Cytosensor® to measure extracellular acidification [58], as well as lactate and glucose levels [59]. Bionas® commercialized the Discovery 2500 SystemTM that continuously provides measurements of oxygen consumption and cell impedance [60]. Sensor monitoring is accomplished by specialized hardware/software via different measures, most commonly being changes in impedance, current, and electric potential [61]. Others are based on light emission, ion selective field effect transistors, and resonant frequency [62, 63]. Each of the sensing unit requires a unique hardware/software module, which is able to capture measure, record, analyze, and display it to the user. For example, Molecular Devices® offers variety of

the potentiometer.

**monitoring**

An important movement in the development of integrated control circuits for microfluidics lies within the development of field-deployable and autonomous microfluidic-based systems. For example, Chen and colleagues developed an enzyme-linked immunosorbent assay (ELISA) that can be conducted in field-deployable microfluidic device with smartphone-based imaging [64]. This point-of-care diagnostic device utilizes electrodes that can convert electric current into a microfluidic pump via gas bubble expansion during electrolytic reactions. The micropump receives power from the mobile phone and transports the analytes through the microfluidic device for ELISA. Cleary and colleagues developed a field deployable microflui‐

**Figure 12.** Flow system schematics automated perfusion sequences for sensor monitoring.

dic system for water quality monitoring. The system is composed of the microfluidic device where the mixing, reaction, and detection takes place, a phosphate chemical sensor, an optical detection system that composed of a light source and a photodiode detector and a microcon‐ troller (MSP430F449 by Texas Instruments). The microcontroller controls the analytical components, fluid handling, data acquisition and storage, and wireless communication. This movement of integration is highly specialized and specifically developed for particular applications.

Another important movement aim to replace the need of external support equipment such as mechanical pumps and pressure-actuated valves. A recent development aims to make the control and the functionality of a pump redundant by designing a light operated micropumps. Sze et al. developed this pump by utilizing light energy to activate bacteriorhodopsin and sugar transporter proteins, which create an osmotic pressure gradient and drive fluid flow [65]. Zimmermann and colleagues developed an autonomous capillary system, in which wettable capillary valves were implemented by employing an abruptly changing geometry of the flow path, and are able to delay or stop a moving liquid [66]. Kojima and colleagues also proposed a new concept for autonomous switching of valves and pumps by employing electrowetting principles [67]. While nonmechanical pumps were utilized for different applications they currently offer limited pressure generation capabilities compared to mechanical pumps.

Currently, control logic is implemented in a computer or a microprocessor, where its interface with the fluidic circulatory is via sensors or pneumatic solenoid valves. However, one emerging control paradigm aim to integrate the control logic on-chip. It is claimed that as the development of electronic logic gates simplified the construction and operation of electronic devices, pneumatic logic gates could reduce the number of external controllers [68]. A variety of different approaches were suggested in the past few years. Microfluidic logic gates were used to route fluids in complex networks and to perform simple on-chip calculations. For example, Cheow et al. developed droplet-based logic gates where the absence of a dispersed phase liquid in a continuous phase liquid represent 1 and 0, respectively. They were able to demonstrate various logic gates such as *and*, *or* and *not* [69]. Prakash et al. were able to utilize this principle to implement flip-flops, ring oscillators, ripple counters, and synchronizers, which provides an on-chip synchronized flow control mechanism [70]. Toepke et al. forward implemented programmable autonomous timers and rheostats [71]. Zaho et al. used digital microfluidics to implement logical gates with simple droplets manipulation techniques such as transporting, merging, and splitting [72]. They used their platform to compress multiple test-outcome droplets into a single droplet. A more recent development is the development of logical gates with microfluidic valves. Rhee and colleagues developed a pneumatic micro‐ processor, which was constructed from various combinations of microfluidic logic gates [68]. They developed an 8-bit pneumatic microprocessor that decodes a temporal command sequence, apply logic calculations, stores information for signal transportation and mainte‐ nance, and execute commands in target devices. Despite the great interest in such on-chip computing and control logic, their utilization is still anecdotal and not widely spread. This is probably due to the fabrication and operational complexity of such control circuits and their limited application that they are currently support.

## **8. Conclusions**

dic system for water quality monitoring. The system is composed of the microfluidic device where the mixing, reaction, and detection takes place, a phosphate chemical sensor, an optical detection system that composed of a light source and a photodiode detector and a microcon‐ troller (MSP430F449 by Texas Instruments). The microcontroller controls the analytical components, fluid handling, data acquisition and storage, and wireless communication. This movement of integration is highly specialized and specifically developed for particular

**Figure 12.** Flow system schematics automated perfusion sequences for sensor monitoring.

180 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

Another important movement aim to replace the need of external support equipment such as mechanical pumps and pressure-actuated valves. A recent development aims to make the control and the functionality of a pump redundant by designing a light operated micropumps. Sze et al. developed this pump by utilizing light energy to activate bacteriorhodopsin and sugar transporter proteins, which create an osmotic pressure gradient and drive fluid flow [65]. Zimmermann and colleagues developed an autonomous capillary system, in which wettable capillary valves were implemented by employing an abruptly changing geometry of the flow path, and are able to delay or stop a moving liquid [66]. Kojima and colleagues also proposed a new concept for autonomous switching of valves and pumps by employing electrowetting principles [67]. While nonmechanical pumps were utilized for different applications they currently offer limited pressure generation capabilities compared to mechanical pumps.

Currently, control logic is implemented in a computer or a microprocessor, where its interface with the fluidic circulatory is via sensors or pneumatic solenoid valves. However, one emerging control paradigm aim to integrate the control logic on-chip. It is claimed that as the development of electronic logic gates simplified the construction and operation of electronic devices, pneumatic logic gates could reduce the number of external controllers [68]. A variety of different approaches were suggested in the past few years. Microfluidic logic gates were used to route fluids in complex networks and to perform simple on-chip calculations. For example, Cheow et al. developed droplet-based logic gates where the absence of a dispersed phase liquid in a continuous phase liquid represent 1 and 0, respectively. They were able to demonstrate various logic gates such as *and*, *or* and *not* [69]. Prakash et al. were able to utilize

applications.

System integration is an imperative developmental milestone for the field of microfluidics, both in terms of the scalability of increasingly complex platforms that still lack standardization, as well as the incorporation and adoption of emerging technologies in biomedical research. This work describes a modular integration and synchronization of a complex microfluidic platform permitting implementation of user-independent optimization algorithms. The versatility and modularity of our open-system unit can be rapidly adopted by leading groups in the field for automation and optimization of droplet generation, inertial focusing purifica‐ tion, single-cell analysis, and large-scale integration. We also described a series of emerging control paradigms including on-chip control logic, autonomous valves and micropumps, and application-specific control integration.

## **Author details**

Elishai Ezra1,2, Danny Bavli1 and Yaakov Nahmias1,3\*

\*Address all correspondence to: ynahmias@cs.huji.ac.il

1 Center for Bioengineering, The Hebrew University, Jerusalem, Israel

2 Faculty of Engineering and Computer Science, Jerusalem College of Technology, Jerusalem, Israel

3 Department of Cell and Developmental Biology, The Hebrew University, Jerusalem, Israel

### **References**


[14] Liu, J., Hansen, C., and Quake, S. R., 2003, "Solving the "world-to-chip" interface problem with a microfluidic matrix," Analytical chemistry, 75(18), pp. 4718–4723.

**References**

[1] Fordyce, P. M., Diaz-Botia, C. A., DeRisi, J. L., and Gomez-Sjoberg, R., 2012, "System‐ atic characterization of feature dimensions and closing pressures for microfluidic

[2] Bhatia, S. N., Yarmush, M. L., and Toner, M., 1997, "Controlling cell interactions by micropatterning in co-cultures: hepatocytes and 3T3 fibroblasts," Journal of biomedi‐

[3] Cheung, Y. K., Gillette, B. M., Zhong, M., Ramcharan, S., and Sia, S. K., 2007, "Direct patterning of composite biocompatible microstructures using microfluidics," Lab on

[4] Mu, X., Zheng, W., Sun, J., Zhang, W., and Jiang, X., 2013, "Microfluidics for manipu‐

[5] Kartalov, E. P., Zhong, J. F., Scherer, A., Quake, S. R., Taylor, C. R., and Anderson, W. F., 2006, "High-throughput multi-antigen microfluidic fluorescence immunoassays,"

[6] Sanchez-Freire, V., Ebert, A. D., Kalisky, T., Quake, S. R., and Wu, J. C., 2012, "Micro‐ fluidic single-cell real-time PCR for comparative analysis of gene expression pat‐

[7] Das, T., and Chakraborty, S., 2009, "Biomicrofluidics: Recent trends and future chal‐

[8] Ho, C. T., Lin, R. Z., Chang, W. Y., Chang, H. Y., and Liu, C. H., 2006, "Rapid hetero‐ geneous liver-cell on-chip patterning via the enhanced field-induced dielectrophore‐

[9] Kalisky, T., Blainey, P., and Quake, S. R., 2011, "Genomic analysis at the single-cell

[10] Mark, D., Haeberle, S., Roth, G., von Stetten, F., and Zengerle, R., 2010, "Microfluidic lab-on-a-chip platforms: requirements, characteristics and applications," Chemical

[11] Melin, J., and Quake, S. R., 2007, "Microfluidic large-scale integration: the evolution of design rules for biological automation," Annual review of biophysics and biomo‐

[12] Thorsen, T., Maerkl, S. J., and Quake, S. R., 2002, "Microfluidic large-scale integra‐

[13] Hansen, C. L., Classen, S., Berger, J. M., and Quake, S. R., 2006, "A microfluidic de‐ vice for kinetic optimization of protein crystallization and in situ structure determi‐

nation," Journal of the American Chemical Society, 128(10), pp. 3142–3143.

valves produced via photoresist reflow," Lab on a chip, 12(21), pp. 4287–4295.

cal materials research, 34(2), pp. 189–199.

182 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

a chip, 7(5), pp. 574–579.

lating cells," Small, 9(1), pp. 9–21.

BioTechniques, 40(1), pp. 85–90.

terns," Nature protocols, 7(5), pp. 829–838.

sis trap," Lab on a chip, 6(6), pp. 724–734.

level," Annual review of genetics, 45.

Society reviews, 39(3), pp. 1153–1182.

lecular structure, 36, pp. 213–231.

tion," Science, 298(5593), pp. 580–584.

lenges," Sadhana, 34(4), pp. 573–590.


er Technology, Reliable Systems and Applications. 5th Annual European Computer Conference. Proceedings., IEEE, pp. 608–612.


tures by solvent vapour process," Proc. SPIE Advanced Lithography, International Society for Optics and Photonics, p. 905110.

[42] Jackson, M. J., 2005, Microfabrication and nanomanufacturing, CRC Press, Boca Ra‐ ton, Florida.

er Technology, Reliable Systems and Applications. 5th Annual European Computer

[29] Valle, M., Nateri, G., Caviglia, D. D., Bisio, G. M., and Briozzo, L., "An ASIC design for real-time image processing in industrial applications," Proc. European Design

[30] Yu, C., Hu, H.-T., and Lin, C.-Y., 2003, "Design and implantation of an ASIC architec‐ ture for 1.6 kbps speech synthesis," Consumer Electronics, IEEE Transactions on,

[31] Schmadecke, I., and Blume, H., "Hardware-accelerator design for energy-efficient acoustic feature extraction," Proc. Consumer Electronics (GCCE), 2013 IEEE 2nd

[32] Dong, M., Liu, J., and Liu, R., "Speech interface ASIC of SOC architecture for embed‐ ded application," Proc. Signal Processing, 2002 6th International Conference on, IEEE,

[33] Weikun, X., Huibin, Z., and Qiuli, Z., "Testing FPGA devices on an Automatic Test Equipment," Proc. Information and Automation (ICIA), 2013 IEEE International Con‐

[34] Monmasson, E., and Cirstea, M. N., 2007, "FPGA design methodology for industrial control systems—A review," Industrial Electronics, IEEE Transactions on, 54(4), pp.

[35] Hodges, S., Villar, N., Scott, J., and Schmidt, A., 2012, "A new era for ubicomp devel‐

[36] Kubitza, T., Pohl, N., Dingler, T., Schneegass, S., Weichel, C., and Schmidt, A., 2013, "Ingredients for a new wave of ubicomp products," IEEE Pervasive Computing(3),

[37] Gellersen, H., Kortuem, G., Schmidt, A., and Beigl, M., 2004, "Physical prototyping

[38] Reas, C., and Fry, B., 2007, Processing: a programming handbook for visual designers

[39] Pearce, J. M., 2012, "Materials science. Building research equipment with free, open-

[40] Fobel, R., Fobel, C., and Wheeler, A. R., 2013, "DropBot: An open-source digital mi‐ crofluidic control system with precise control of electrostatic driving force and in‐ stantaneous drop velocity measurement," Applied Physics Letters, 102(19), p. 193513.

[41] Chaudhari, A., Ghoshal, T., Shaw, M. T., Cummins, C., Borah, D., Holmes, J. D., and Morris, M. A., "Formation of sub-7 nm feature size PS-b-P4VP block copolymer struc‐

and Test Conference, 1995. ED&TC 1995, Proceedings., IEEE, pp. 385–390.

Conference. Proceedings., IEEE, pp. 608–612.

184 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

Global Conference on, IEEE, pp. 135–139.

opment," Pervasive Computing, IEEE, 11(1), pp. 5–9.

with smart-its," Pervasive Computing, IEEE, 3(3), pp. 74–82.

source hardware," Science, 337(6100), pp. 1303–1304.

49(3), pp. 731–736.

pp. 402–405.

1824–1842.

pp. 5–8.

and artists, Mit Press.

ference on, IEEE, pp. 65–70.


[68] Rhee, M., and Burns, M. A., 2009, "Microfluidic pneumatic logic circuits and digital pneumatic microprocessors for integrated microfluidic systems," Lab on a chip, 9(21), pp. 3131–3143.

[55] Monk, S., 2012, Getting started with. NET Gadgeteer, " O'Reilly Media, Inc.".

tion techniques," Journal of biomedical optics, 16(8), p. 080901.

186 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

146.

e19714.

146.

25(6), p. 065009.

bioelectronics, 16(3), pp. 195–203.

sorbent assay," Biomicrofluidics, 8(6), p. 064101.

capillary systems," Microfluid Nanofluid, 5(3), pp. 395–402.

IEEE 24th International Conference on, pp. 1173–1176.

[56] Wu, J., and Gu, M., 2011, "Microfluidic sensing: state of the art fabrication and detec‐

[57] Weltin, A., Slotwinski, K., Kieninger, J., Moser, I., Jobst, G., Wego, M., Ehret, R., and Urban, G. A., 2014, "Cell culture monitoring for drug screening and cancer research: a transparent, microfluidic, multi-sensor microsystem," Lab on a chip, 14(1), pp. 138–

[58] Dean G. Hafeman, W. P., Harden McConnell, 1988, "Light-addressable potentiomet‐

[59] Eklund Sven, T. R., Snider, Rachel, Carney Clare, Wright David, Wikswo John, Cliffel David, 2009, "Metabolic Discrimination of Select List Agents by Monitoring Cellular

[60] Alborzinia, H., Can, S., Holenya, P., Scholl, C., Lederer, E., Kitanovic, I., and Wölfl, S., 2011, "Real-Time Monitoring of Cisplatin-Induced Cell Death," PLoS ONE, 6(5), p.

[61] Weltin, A., Slotwinski, K., Kieninger, J., Moser, I., Jobst, G., Wego, M., Ehret, R., and Urban, G. A., 2014, "Cell culture monitoring for drug screening and cancer research: a transparent, microfluidic, multi-sensor microsystem," Lab on a Chip, 14(1), pp. 138–

[62] Lehmann, M., Baumann, W., Brischwein, M., Gahle, H., Freund, I., Ehret, R., Drechsler, S., Palzer, H., Kleintges, M., Sieben, U., and Wolf, B., 2001, "Simultaneous measure‐ ment of cellular respiration and acidification with a single CMOS ISFET," Biosensors &

[63] Eklund, S. E., Thompson, R. G., Snider, R. M., Carney, C. K., Wright, D. W., Wikswo, J., and Cliffel, D. E., 2009, "Metabolic discrimination of select list agents by monitoring cellular responses in a multianalyte microphysiometer," Sensors, 9(3), pp. 2117–2133.

[64] Chen, A., Wang, R., Bever, C. R., Xing, S., Hammock, B. D., and Pan, T., 2014, "Smart‐ phone-interfaced lab-on-a-chip devices for field-deployable enzyme-linked immuno‐

[65] Tsun-kay Jackie, S., Jin, L., and Prashanta, D., 2015, "Design and modeling of a light powered biomimicry micropump," Journal of Micromechanics and Microengineering,

[66] Zimmermann, M., Hunziker, P., and Delamarche, E., 2008, "Valves for autonomous

[67] Kojima, K., and Suzuki, H., "Programmed autonomous microfluidics using micro‐ valves and micropumps," Proc. Micro Electro Mechanical Systems (MEMS), 2011

Responses in a Multianalyte Microphysiometer," Sensors, 9(3), pp. 2117–2133.

ric sensor for biochemical systems.," Science, 240(4856), pp. 1182–1185.


## **Microfluidics for Soft Electronics**

Babak Taghavi, Jiantong Li, Mikael Östling and Shi Cheng

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/63376

#### **Abstract**

Microfluidics- based soft electronic systems have the potential to assist conventional rigid devices and circuits to achieve extreme levels of elasticity in wearable electronics and oth‐ er applications. The goal of employing microfluidics-based approach among other exist‐ ing methods is to enhance users comfort through fulfillment of wearable's mechanical performance requirements such as flexibility, twistability, and stretchability. This chapter presents a brief survey of different solutions for developing elastic electronics and a thor‐ ough review of the progress in microfluidics-based approaches. This chapter mainly fo‐ cuses on the description of the fabrication process, design, and measurement steps of different antennas and complex systems realized using microfluidic interconnects.

**Keywords:** Elastic antennas, liquid metal, microfluidics, stretchable electronics, wearable technology

## **1. Introduction**

In recent years, wearable electronics have received increasing attention in both academic research and industrial development. Wearable electronics are being utilized in various applications, including healthcare and medical electronic devices [1] as well as consumer electronics and robotics [2]. Human body surfaces are huge compared to the sizes of electronic components. Large-area coverage is sometimes a must-have feature, where tiny wearable gadgets are certainly not a good fit. Consider an electronic system equipped with multiple sensors, each placed at a fingertip for pressure sensing, and a radio frequency (RF) transmitter associated with an antenna to wirelessly send out the sensing data. Each sensor is connected to the core transmitter circuit by long, in the order of 10 cm, interconnects. Additionally, an antenna with a size of few centimeters is needed for RF communication. Therefore, the overall size of such an integrated electronic device can be as large as 50 –100 cm2 . Using conventional rigid printed circuit board (PCB) techniques to develop this kind of systems would inevitably

© 2016 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

lead to a clumsy gadget that will most likely give an uncomfortable experience to the wearer. As wearable devices intimately interact with users' body and skin, their physical shapes and impact on haptic feelings of wearers are very important design considerations. Human body is soft and our skin is stretchable, especially on joints, thus any external gadget that comes into seamless contact with our body should feature certain degree of flexibility and stretchability in order to facilitate natural motions [3]. Researchers have been experimenting different materials and assembly methods to create more human-friendly wearable systems. Flexibility can be achieved by implementing electronics on thin flexible substrates, ranging from polyi‐ mide- based Kapton films such as circuit boards to thin film transistors (TFTs) on device level. Such flexible electronic devices and systems can withstand moderate folding and twisting, but not stressing. Excessive straining often results in non-reversible mechanical deformation and permanent electrical failure. Achieving stretchability with maintained electronic functions is significantly more challenging. Radically new fabrication techniques and design strategies are required to realize soft electronics, capable of reaching extreme levels of reversible bendability, twistability, and stretchability.

Significant advances have been made in the emerging field of soft electronics in the past two decades. Great potential has been shown in numerous new appealing applications, e.g., bodyworn healthcare/medical sensing systems, ergonomic units, and tissue engineering constructs, where conventional rigid, bulky electronics are facing insurmountable obstacles to fulfill the demands. One of the most promising concepts is soft, rubbery electronic devices, implemented in the form of elastic structures, materials, or a combination of both [4]. Wavy silicon ribbons to realize stretchable silicon integrated circuits (ICs) have been reported [5]. Also, integration of TFTs on prestressed substrates has been proposed [6].

Unlike complex, miniaturized semiconductor- based active circuits, many passive components such as interconnects, sensors, and antennas, which occupy the major areas in large-area electronic devices and systems, can only be shrunk at the expense of efficiency, sensitivity, and functionalities. Elasticity of such large passive devices, thereby, dominates the overall flexibility and stretchability of integrated electronics. Various approaches to implement stretchable interconnects have been presented [7,8]. One typical example is meandered metallic interconnects built on flex foils, operating at different frequency ranges [9]. Flexible and stretchable interconnects, realized by incorporating low melting temperature solder alloy into microstructured channels in polydimethylsiloxane (PDMS), were first proposed by Siegel, et al. [10]. Various properties of different solder metals and alloys, including melting temper‐ ature, electrical conductivity, Brinell hardness, and material costs, were also investigated. Later, room temperature liquid metal alloy-, eutectic gallium and indium (EGaIn), filled microfluidic channels acting as elastic direct current (DC) interconnects have been demon‐ strated [11]. Additional appealing characteristics of EGaIn, such as good rheological proper‐ ties, excellent wetting on almost any surfaces, and low toxicity, make it well suited for microfluidics- based soft electronics that might be in proximity contact with human body in practical use [12]. The first stretchable electronic devices, operating at radio frequencies, are a liquid metal unbalanced loop antenna and a fluidic ultra-wideband planar inverted cone antenna (PICA). Both antennas were implemented by utilizing liquid metal alloy, galinstan, incorporated into microstructured, elastomeric channel matrices [13,14]. Similar concept was also proposed to demonstrate a stretchable dipole antenna. Modified inhomogeneous elasto‐ mer substrate was subsequently used to significantly improve elasticity of stretchable antennas [15]. A hybrid integration approach for combining active circuits on flexible PCBs and largearea stretchable passive electronics for RF radiation sensing was later reported by Cheng and Wu [16]. This integration strategy was afterwards expanded to create multi-layer, selfcontained, large-area, elastic wireless strain sensor [17]. Standalone patch antenna using serpentine channels filled with EGaIN was also published [18]. More recently, a reconfigurable fluidic serpentine antenna with varying resonant frequency in response to strain was presented [19]. Moreover, successful development of new fabrication techniques for microfluidics- based soft electronics has been demonstrated [20].

The following sections present and discuss recent advances in the emerging field of micro‐ fluidics -based soft electronics. This chapter focuses on ultracompliant, elastic electronic devices and systems, with the capability of wireless communications and remote sensing. Following sections start with an elaborate description of the fabrication process, followed by presentation of several application examples of microfluidics- based RF elastic electronics. Measurement results on foldable, twistable, stretchable antennas and integrated devices under different mechanical deformations are also shown. This chapter ends with a glimpse of some ongoing research activities at the Department of Integrated Devices and Circuits, KTH Royal Institute of Technology, Sweden, among which new advanced processing techniques and nanomaterials, e.g., inkjet-printed, high precision, graphene resistors and conductors, are being investigated to enable novel sensing functions. Furthermore, feasibilities of utilizing elastic, microfluidic RF energy harvesters for wireless charging of soft energy storage units or remote powering active stretchable electronics are discussed.

## **2. Fabrication process**

lead to a clumsy gadget that will most likely give an uncomfortable experience to the wearer. As wearable devices intimately interact with users' body and skin, their physical shapes and impact on haptic feelings of wearers are very important design considerations. Human body is soft and our skin is stretchable, especially on joints, thus any external gadget that comes into seamless contact with our body should feature certain degree of flexibility and stretchability in order to facilitate natural motions [3]. Researchers have been experimenting different materials and assembly methods to create more human-friendly wearable systems. Flexibility can be achieved by implementing electronics on thin flexible substrates, ranging from polyi‐ mide- based Kapton films such as circuit boards to thin film transistors (TFTs) on device level. Such flexible electronic devices and systems can withstand moderate folding and twisting, but not stressing. Excessive straining often results in non-reversible mechanical deformation and permanent electrical failure. Achieving stretchability with maintained electronic functions is significantly more challenging. Radically new fabrication techniques and design strategies are required to realize soft electronics, capable of reaching extreme levels of reversible bendability,

190 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

Significant advances have been made in the emerging field of soft electronics in the past two decades. Great potential has been shown in numerous new appealing applications, e.g., bodyworn healthcare/medical sensing systems, ergonomic units, and tissue engineering constructs, where conventional rigid, bulky electronics are facing insurmountable obstacles to fulfill the demands. One of the most promising concepts is soft, rubbery electronic devices, implemented in the form of elastic structures, materials, or a combination of both [4]. Wavy silicon ribbons to realize stretchable silicon integrated circuits (ICs) have been reported [5]. Also, integration

Unlike complex, miniaturized semiconductor- based active circuits, many passive components such as interconnects, sensors, and antennas, which occupy the major areas in large-area electronic devices and systems, can only be shrunk at the expense of efficiency, sensitivity, and functionalities. Elasticity of such large passive devices, thereby, dominates the overall flexibility and stretchability of integrated electronics. Various approaches to implement stretchable interconnects have been presented [7,8]. One typical example is meandered metallic interconnects built on flex foils, operating at different frequency ranges [9]. Flexible and stretchable interconnects, realized by incorporating low melting temperature solder alloy into microstructured channels in polydimethylsiloxane (PDMS), were first proposed by Siegel, et al. [10]. Various properties of different solder metals and alloys, including melting temper‐ ature, electrical conductivity, Brinell hardness, and material costs, were also investigated. Later, room temperature liquid metal alloy-, eutectic gallium and indium (EGaIn), filled microfluidic channels acting as elastic direct current (DC) interconnects have been demon‐ strated [11]. Additional appealing characteristics of EGaIn, such as good rheological proper‐ ties, excellent wetting on almost any surfaces, and low toxicity, make it well suited for microfluidics- based soft electronics that might be in proximity contact with human body in practical use [12]. The first stretchable electronic devices, operating at radio frequencies, are a liquid metal unbalanced loop antenna and a fluidic ultra-wideband planar inverted cone antenna (PICA). Both antennas were implemented by utilizing liquid metal alloy, galinstan,

twistability, and stretchability.

of TFTs on prestressed substrates has been proposed [6].

Microfluidic channel design procedure can be concisely described in three major steps, which are soft lithography, reverse molding, and liquid metal injection. Similar to PCB routings, depending on the complexity and architecture of circuits and antennas, a single or a multilayer configuration can be selected to develop an electronic system. First, fabrication processes of single-layer liquid metal- filled channel matrices embedded in PDMS elastomers are presented in this section. Subsequently, processing steps for implementing multi-layer microfluidics-based soft electronic devices are described in detail. Finally, heterogeneous integration method for connecting active circuits assembled on flexible PCBs to microstruc‐ tured fluidic channels encased in elastomeric substrates is discussed.

#### **2.1. Single-layer microfluidic, elastic channels**

Single- layer microfluidic channels are developed on bare silicon wafers, which are used as the substrate for molding and formation of elastic layers. SU8 photoresist is employed as a reverse mold for fluidic channels, and its thickness determines the height of microfluidic channels. SU8 photoresist is spin coated on a silicon wafer as shown in Figure 1(a). As discussed in the Introduction, microfluidic channels act the same as conductor traces on conventional PCBs; therefore, their patterns are designed in the similar way as routing the traces on PCBs. The designed patterns are then used to build the corresponding lithography mask. After transferring the pattern to the photoresist and removing unexposed parts, the mold is ready for casting the channels. Cross- sectional view of the developed mold on silicon is displayed in Figure 1(b). The prepared mold is then thermally stabilized at 150 o C for 30 minutes to enhance the adhesion between silicon and the remaining SU8 layer. PDMS is used as the soft substrate because of its stretchability, transparency, and relatively easy preparation process. PDMS is widely used in bioelectronics and microfluidic applications such as lab-on-a-chip. It is prepared by mixing PDMS prepolymer and cross linker (Elastosil RT601A and B, Wacker Chemie, Munich, Germany) with 9:1 mass ratio. As shown in Figure 1(c), PDMS should be poured on top of the silicon to form the substrate. In addition to the patterned structures, a thin layer of PDMS is prepared on the blank silicon wafer to form a cover layer for the reversely molded PDMS. After degassing, both patterned and blank PDMS layers are cured at 70 o C for 30 minutes to solidify the PDMS layers. Once the PDMS layers are ready, it is time to peel them off from the silicon carriers as illustrated in Figure 1(e). Before proceeding to the next step, numerous openings should be punched on the patterned structure so that the channels can be accessed for liquid metal injection through those openings. The PDMS layers are bonded together using corona discharging (ETP, Chicago, IL, USA) activation afterwards. At this stage, fabrication of the microfluidic channels is completed and the subsequent step is to inject galinstan into the channels as seen in Figure 1(g). Galinstan is a liquid metal alloy, consisting of 68.5%, Ga; 21.5%, In; and 10%, Sn. It is in liquid state from –19 to 1300 o C, and it also features a good electrical conductivity of 3.46×106 S/m and non-toxicity in comparison with mercury. Its excellent wettability on almost any surface simplifies the channel filling process and ensures good electrical continuity despite of mechanical deformation. After filling the channels, a visual inspection or a simple electrical conductivity check should be performed to ensure that all electrical connections are established as expected. All openings except feeding points for electrical measurement and interconnections with active circuits are sealed with uncured PDMS droplets to prevent any liquid metal leakage. Cross- sectional and top view of the resulting single-layer microfluidic channels filled with liquid metal in a PDMS substrate are shown in Figure 1(h) and (i).

#### **2.2. Multi-layer microfluidic, elastic channels**

Some applications necessitate multi-layer structures. Additional channel layers make it possible to implement more sophisticated structures that are not feasible to build in a singlelayer configuration. As depicted in Figure 2, fabrication steps resemble the previously presented processes for constructing a single-layer structure. Extra channel layers, however, require more lithography and bonding steps. In most applications, as shown in Section 2.3, feeding points on the same side of an elastomer substrate are often needed to facilitate circuit assembly or characterizations. For assessing the bottom liquid metal filled channels from the top surface of the PDMS substrate, an additional inlet is punched through the top and middle layers after bonding. In the liquid metal injection phase, top layer channels are filled and then

**Figure 1.** Single-layer microfluidic soft electronics fabrication process.

channels. SU8 photoresist is spin coated on a silicon wafer as shown in Figure 1(a). As discussed in the Introduction, microfluidic channels act the same as conductor traces on conventional PCBs; therefore, their patterns are designed in the similar way as routing the traces on PCBs. The designed patterns are then used to build the corresponding lithography mask. After transferring the pattern to the photoresist and removing unexposed parts, the mold is ready for casting the channels. Cross- sectional view of the developed mold on silicon is displayed

enhance the adhesion between silicon and the remaining SU8 layer. PDMS is used as the soft substrate because of its stretchability, transparency, and relatively easy preparation process. PDMS is widely used in bioelectronics and microfluidic applications such as lab-on-a-chip. It is prepared by mixing PDMS prepolymer and cross linker (Elastosil RT601A and B, Wacker Chemie, Munich, Germany) with 9:1 mass ratio. As shown in Figure 1(c), PDMS should be poured on top of the silicon to form the substrate. In addition to the patterned structures, a thin layer of PDMS is prepared on the blank silicon wafer to form a cover layer for the reversely molded PDMS. After degassing, both patterned and blank PDMS layers are cured at 70 o

30 minutes to solidify the PDMS layers. Once the PDMS layers are ready, it is time to peel them off from the silicon carriers as illustrated in Figure 1(e). Before proceeding to the next step, numerous openings should be punched on the patterned structure so that the channels can be accessed for liquid metal injection through those openings. The PDMS layers are bonded together using corona discharging (ETP, Chicago, IL, USA) activation afterwards. At this stage, fabrication of the microfluidic channels is completed and the subsequent step is to inject galinstan into the channels as seen in Figure 1(g). Galinstan is a liquid metal alloy, consisting

a good electrical conductivity of 3.46×106 S/m and non-toxicity in comparison with mercury. Its excellent wettability on almost any surface simplifies the channel filling process and ensures good electrical continuity despite of mechanical deformation. After filling the channels, a visual inspection or a simple electrical conductivity check should be performed to ensure that all electrical connections are established as expected. All openings except feeding points for electrical measurement and interconnections with active circuits are sealed with uncured PDMS droplets to prevent any liquid metal leakage. Cross- sectional and top view of the resulting single-layer microfluidic channels filled with liquid metal in a PDMS substrate are

Some applications necessitate multi-layer structures. Additional channel layers make it possible to implement more sophisticated structures that are not feasible to build in a singlelayer configuration. As depicted in Figure 2, fabrication steps resemble the previously presented processes for constructing a single-layer structure. Extra channel layers, however, require more lithography and bonding steps. In most applications, as shown in Section 2.3, feeding points on the same side of an elastomer substrate are often needed to facilitate circuit assembly or characterizations. For assessing the bottom liquid metal filled channels from the top surface of the PDMS substrate, an additional inlet is punched through the top and middle layers after bonding. In the liquid metal injection phase, top layer channels are filled and then

C for 30 minutes to

C, and it also features

C for

in Figure 1(b). The prepared mold is then thermally stabilized at 150 o

192 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

of 68.5%, Ga; 21.5%, In; and 10%, Sn. It is in liquid state from –19 to 1300 o

shown in Figure 1(h) and (i).

**2.2. Multi-layer microfluidic, elastic channels**

openings 2 and 5 are temporarily sealed by small pieces of Scotch tapes and opening 1 is encapsulated with uncured PDMS. Afterwards, the structure is flipped to the other side and galinstan is injected into the channel and both openings 3 and 4 are permanently sealed with uncured PDMS droplets. Later, the bonded substrate is flipped over and the temporary sealants are removed. After verifying all connections, the sample is cured at 85 o C for about 30 minutes.

**Figure 2.** Processing steps of multi-layer microfluidic soft electronics.

#### **2.3. Strain isolation cells**

Active circuits play an essential role in modern electronic systems. Purely passive microfluidic soft electronics without any active components can only be used in very limited application scenarios in practice. A pressing issue is how to realize active electronics in bendable, flexible, and stretchable manners. Implementing active circuits using microfluidic technologies would be ideal, but to our best knowledge, none of the existing microfluidic techniques support fabrication of high performance, basic active components, such as diodes and transistors, operating at gigahertz frequency range. A compromising solution is to heterogeneously integrate off-the-shelf packaged IC chipsets with passive microfluidic electronics. Here the biggest challenge lies in how to create reliable interconnections between rigid IC chips associated with discrete components and liquid metal filled elastic channels. Figure 3 illustrates a hybrid integration strategy, where IC chipsets as well as external lumped com‐ ponents are first assembled on a small piece of flexible PCB. Such a quasi-flexible circuit contains complex routing and most connections with high reliability. The subsequent step is to implement interconnects between flexible circuits and microfluidic elastic channels via solid metallic pins. Since much fewer connections are needed at this step, the overall reliability of such interconnects are dramatically improved. In order to address the issue of mechanical mismatch between elastomers and rigid components, strain isolation cells, the so-called local stiffening cells (LSCs), are introduced to encapsulate PCBs and metal pins with locally thickened PDMS. Although such cells are a bit more rigid than any other areas on the PDMS elastomer and inevitably degrade the overall elasticity of integrated devices to some extent, they can effectively minimize mechanical deformation around active circuits and interconnect pins when the entire devices are being folded, twisted, or stressed. Reliable galvanic contacts between metal pins and liquid metal alloy are crucial. Soldering tin plating on metallic pins is employed to ensure good electrical connections between solid and liquid metals, thanks to superior wetting properties of galinstan on tin. As depicted in Figure 3, a semi-spherical solder ball is also mounted on the bottom surface of each tin plated pin at one end to further improve the reliability of the connection to fluid metal.

**Figure 3.** Hybrid integration of flexible active circuits and fluidic, elastic passive electronics.

## **3. Microfluidics- based elastic antennas and heterogeneously integrated RF electronics**

Fabrication methods introduced in Section 2 are applied to realize different elastic RFintegrated electronic devices and systems. This section presents the design, fabrication, and measured results on standalone single-layer foldable, twistable, stretchable antennas; a fluidic soft RF radiation sensor; and a multi-layer, microfluidic, reversibly stretchable, large-area wireless strain sensor.

#### **3.1. Liquid metal stretchable unbalanced loop antenna**

and stretchable manners. Implementing active circuits using microfluidic technologies would be ideal, but to our best knowledge, none of the existing microfluidic techniques support fabrication of high performance, basic active components, such as diodes and transistors, operating at gigahertz frequency range. A compromising solution is to heterogeneously integrate off-the-shelf packaged IC chipsets with passive microfluidic electronics. Here the biggest challenge lies in how to create reliable interconnections between rigid IC chips associated with discrete components and liquid metal filled elastic channels. Figure 3 illustrates a hybrid integration strategy, where IC chipsets as well as external lumped com‐ ponents are first assembled on a small piece of flexible PCB. Such a quasi-flexible circuit contains complex routing and most connections with high reliability. The subsequent step is to implement interconnects between flexible circuits and microfluidic elastic channels via solid metallic pins. Since much fewer connections are needed at this step, the overall reliability of such interconnects are dramatically improved. In order to address the issue of mechanical mismatch between elastomers and rigid components, strain isolation cells, the so-called local stiffening cells (LSCs), are introduced to encapsulate PCBs and metal pins with locally thickened PDMS. Although such cells are a bit more rigid than any other areas on the PDMS elastomer and inevitably degrade the overall elasticity of integrated devices to some extent, they can effectively minimize mechanical deformation around active circuits and interconnect pins when the entire devices are being folded, twisted, or stressed. Reliable galvanic contacts between metal pins and liquid metal alloy are crucial. Soldering tin plating on metallic pins is employed to ensure good electrical connections between solid and liquid metals, thanks to superior wetting properties of galinstan on tin. As depicted in Figure 3, a semi-spherical solder ball is also mounted on the bottom surface of each tin plated pin at one end to further improve

194 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

the reliability of the connection to fluid metal.

**electronics**

wireless strain sensor.

**Figure 3.** Hybrid integration of flexible active circuits and fluidic, elastic passive electronics.

**3. Microfluidics- based elastic antennas and heterogeneously integrated RF**

Fabrication methods introduced in Section 2 are applied to realize different elastic RFintegrated electronic devices and systems. This section presents the design, fabrication, and measured results on standalone single-layer foldable, twistable, stretchable antennas; a fluidic soft RF radiation sensor; and a multi-layer, microfluidic, reversibly stretchable, large-area

An unbalanced loop antenna, operating at 2.4 GHz Industrial, Scientific, and Medical (ISM) band, is designed, fabricated, and characterized, using single-layer microfluidic channels filled with liquid metal. Schematic of the resulting loop antenna prototype as well as a crosssectional view of microfluidic channels are shown in Figure 4. The antenna consists of a loop radiating element and a semi-circular, meshed ground plane. The loop is grounded at one end and isolated from the ground plane at the other end for coaxial cable feeding.

**Figure 4.** Schematic of liquid metal stretchable unbalanced loop antenna: (a) overlook view, (b) cross- sectional view.

The resonant frequency (*f*res) of a loop antenna is a function of the loop length and can be approximated by the following Equation 1:

$$f\_{\rm res} \approx \frac{c}{2L\_{\rm loop}\sqrt{\varepsilon\_{\rm eff}}},$$

where *c* is the velocity of light in free space, *L*loop is the loop length, and *ε*eff is the effective permittivity, which is about 1 for very thin PDMS layers. According to the equation, to obtain *f*res = 2.4 GHz, the loop should be 5.6 cm in length. Numerical simulations are performed in Ansoft HFSS to verify antenna port impedance and radiation characteristics. As seen in Figure 4, numerous small cylinders are aligned along the loop-shaped channel, in which extra liquid metal alloy can be stored and released to guarantee reliable galvanic connections while the antenna is mechanically deformed. It should be noted that the presence of such liquid metal reservoirs results in slightly increased electrical length of the loop antenna so that a minor decrease in resonant frequency compared with the value calculated from the equation can be observed both in simulations and measurements. For example, if the loop length, *L*loop, equals 56.4 mm, the resonant frequency, *f*res, derived according to the equation should be 2.7 GHz. However, considering the effect of reservoirs, *f*res = 2.4 GHz is obtained in numerical simula‐ tions. Once the simulation results meet the requirements, the antenna layout is exported to print a mask on transparency for subsequent soft lithography process. The printed plastic mask is then used to transfer the antenna patterns to the SU8 photoresist and to fabricate the antenna following the fabrication steps described in Section 2.1. The resulting unbalanced loop antenna prototype is illustrated in Figure 5.

**Figure 5.** Photograph of the fabricated fluidic stretchable loop antenna.

In order to verify simulated results, the port impedance and radiation characteristics are measured using a vector network analyzer in an anechoic chamber. All experiments are first carried out with the antenna in its relaxed state. Measurement results along with the corre‐ sponding simulations are shown in Figure 6.

**Figure 6.** (a) Simulated (solid line) and measured (dotted line) reflection coefficient (S11) and (b) radiation patterns of the liquid metal- based stretchable unbalanced loop antenna. The coordinate system is depicted in Figure 4.

Mechanical performance of the microfluidic elastic antenna is tested under various deforma‐ tions as shown in Figure 7. It is observed that the unbalanced loop antenna prototype can withstand omnidirectional straining, as well as severe folding and twisting, and return to its original shape after the removal of external forces.

**Figure 7.** Photographs of the unbalanced loop antenna in (a) omnidirectionally strained, (b) twisted, and (c) folded states.

Electrical properties of the fluidic soft antenna in response to varying strains applied along its *x*- and *y*-axis are also characterized. Photographs of the elastic antenna in its relaxed and stressed states are displayed in Figure 8.

**Figure 8.** Photograph of the resulting elastic loop antenna in its (a) relaxed state, and with 40% stain along (b) *x*- and (c) *y*-axes. The corresponding coordinate system is shown in Figure 4.

Figure 9 presents measured port impedance and radiation patterns of the loop antenna in different states. The decrease in the antenna resonant frequency can be explained by an increased antenna size as a result of stretching. Antennas often exhibit wider radiation bandwidth than impedance bandwidth. This is why, measured radiation efficiency remains relatively stable despite being strained. No dramatic electrical performance degradation is detected in the experiments. It implies that no interruption in electrical connections occurs during antenna deformation. This is mainly because of the superior wettability of galinstan on the inner surfaces of elastomeric channels as well as its good electrical conductivity.

#### **3.2. Elastic electromagnetic radiation sensor**

following the fabrication steps described in Section 2.1. The resulting unbalanced loop antenna

In order to verify simulated results, the port impedance and radiation characteristics are measured using a vector network analyzer in an anechoic chamber. All experiments are first carried out with the antenna in its relaxed state. Measurement results along with the corre‐

**Figure 6.** (a) Simulated (solid line) and measured (dotted line) reflection coefficient (S11) and (b) radiation patterns of

Mechanical performance of the microfluidic elastic antenna is tested under various deforma‐ tions as shown in Figure 7. It is observed that the unbalanced loop antenna prototype can withstand omnidirectional straining, as well as severe folding and twisting, and return to its

the liquid metal- based stretchable unbalanced loop antenna. The coordinate system is depicted in Figure 4.

prototype is illustrated in Figure 5.

**Figure 5.** Photograph of the fabricated fluidic stretchable loop antenna.

196 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

sponding simulations are shown in Figure 6.

original shape after the removal of external forces.

Our daily life has been greatly benefiting from revolutionary mobile communication technol‐ ogies rapidly developed in the past decades. Billions of portable devices, equipped with wireless links, are new parts of our everyday life. People and electronic equipment and systems are connected anywhere at any time. Rising concerns on potential health issues caused by the exposure to RF radiation has been expressed by both scientists and the public. It is, thereby, worth monitoring the level of ambient electromagnetic (EM) wave radiation to assess possible

**Figure 9.** Measured (a) port impedance and (b) radiation patterns of the stretchable unbalanced loop antenna. The co‐ ordinate system is seen in Figure 4.

hazards and improve signal coverage quality in real life. In order to make it more portable, an EM radiation sensor can be developed as a wearable device. As discussed in the Introduction Section 1, better user experience can be achieved by constructing such a radiation sensor in ultracompliant, soft, and elastic manners. An RF power detector circuit assembled on a flex foil and encapsulated in an LSC is designed and interconnected to a microfluidics- based stretchable antenna to form a hybrid device, *cf.* Figure 10. Target operating frequency of the proposed radiation sensor is 900 MHz. A uniplanar rectangular unbalanced loop antenna is chosen for its ease of fabrication. The overall dimensions of the loop obtained from full wave simulations conducted in Ansoft HFSS are 86.9 mm × 88.0 mm × 1.0 mm. Ambient RF radiation is picked up from free space and directed to the input of the RF power sensing module via an elastic, fluidic antenna. The injected RF energy is then proportionally converted to DC voltages at the RF power detector IC output. The upper and lower sensing range of RF radiation energy is determined by the dynamic range of the power detection IC chipset. The chosen IC chip features a dynamic range from approximately –60 dBm to 0 dBm that is sufficient in most application scenarios. Apart from the core IC, three coupling and decoupling surface mounted device (SMD) capacitors along with an input matching inductor are needed. The entire power sensing active circuit mounted on a flexible PCB is powered externally, using a pack of four AA batteries connected serially. A light emitting diode (LED) is soldered at the voltage output of the power detector for visualization, which will be switched on when the received RF energy exceeds the pre-defined threshold. More details about the integrated RF radiation sensor can be found in Figure 10.

Prior to connecting the microfluidic antenna to the active circuit, electrical performance of both the antenna and the detector circuit is evaluated separately. Figure 11 shows measured reflection coefficient of the loop antenna in different stretched modes.

The output DC voltage of the standalone active circuit in response to varying RF input power is characterized. In this experiment, continuous wave (CW) signal with incremental power

**Figure 10.** Schematic view of the elastic EM sensor: (a) overlook view of the integrated sensor, (b) top view, and (c) cross view of the flexible active circuit.

hazards and improve signal coverage quality in real life. In order to make it more portable, an EM radiation sensor can be developed as a wearable device. As discussed in the Introduction Section 1, better user experience can be achieved by constructing such a radiation sensor in ultracompliant, soft, and elastic manners. An RF power detector circuit assembled on a flex foil and encapsulated in an LSC is designed and interconnected to a microfluidics- based stretchable antenna to form a hybrid device, *cf.* Figure 10. Target operating frequency of the proposed radiation sensor is 900 MHz. A uniplanar rectangular unbalanced loop antenna is chosen for its ease of fabrication. The overall dimensions of the loop obtained from full wave simulations conducted in Ansoft HFSS are 86.9 mm × 88.0 mm × 1.0 mm. Ambient RF radiation is picked up from free space and directed to the input of the RF power sensing module via an elastic, fluidic antenna. The injected RF energy is then proportionally converted to DC voltages at the RF power detector IC output. The upper and lower sensing range of RF radiation energy is determined by the dynamic range of the power detection IC chipset. The chosen IC chip features a dynamic range from approximately –60 dBm to 0 dBm that is sufficient in most application scenarios. Apart from the core IC, three coupling and decoupling surface mounted device (SMD) capacitors along with an input matching inductor are needed. The entire power sensing active circuit mounted on a flexible PCB is powered externally, using a pack of four AA batteries connected serially. A light emitting diode (LED) is soldered at the voltage output of the power detector for visualization, which will be switched on when the received RF energy exceeds the pre-defined threshold. More details about the integrated RF radiation sensor can

**Figure 9.** Measured (a) port impedance and (b) radiation patterns of the stretchable unbalanced loop antenna. The co‐

198 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

Prior to connecting the microfluidic antenna to the active circuit, electrical performance of both the antenna and the detector circuit is evaluated separately. Figure 11 shows measured

The output DC voltage of the standalone active circuit in response to varying RF input power is characterized. In this experiment, continuous wave (CW) signal with incremental power

reflection coefficient of the loop antenna in different stretched modes.

be found in Figure 10.

ordinate system is seen in Figure 4.

**Figure 11.** Measured reflection coefficient of the microfluidic elastic rectangular loop antenna in response to varying strain applied along its (a) *x*- and (b) *y*-axis. The coordinate system is presented in Figure 10.

level generated from a signal synthesizer is fed into the RF power detector input through an RF coaxial cable, and the LED indicator at the DC voltage output is temporally disconnected. The output voltage of the RF power detector as a function of input power is shown in Figure 12, in which fairly linear behavior is achieved over the entire dynamic range.

Subsequently, the PCB is assembled on the PDMS substrate using the proposed hybrid integration method. The fabricated elastic EM radiation sensor is shown in Figure 13.

The measurement setup illustrated in Figure 14 is utilized to demonstrate RF radiation sensing capability of the integrated sensor device in ordinary office environment. The distance from the dedicated radiation source comprising a standard horn antenna and a signal generator to the elastic RF radiation sensor is approximately 5 m.

The transmitter sends out a CW signal at 900 MHz with variable power levels into the free space. The LED indicator on the integrated stretchable sensor device is on when being exposed to high levels of radiation. The ON/OFF state of the LED indicator only depends on the level of the RF power received by the microfluidic stretchable loop antenna. Increased distance

**Figure 12.** RF power module output voltage and LED status versus input RF power.

**Figure 13.** Photograph of the resulting elastic EM radiation sensor prototype.

between the radiation source and the sensor would result in greater free space loss. Higher power is, therefore, required from the RF signal generator to turn on the LED in this case. In addition to measurements in its relaxed state, the stretchable EM radiation detector is also tested in its folded, twisted, and strained modes. Figure 15 shows that the implemented microfluidics- based soft radiation sensor prototype remains functioning regardless of deformation.

#### **3.3. Fluidic, elastic ultra-wideband antenna**

Antennas with broad operational frequency bands are sometimes required. In ultra-wideband (UWB) systems, an extremely broad frequency range from 3.1 GHz to 10.6 GHz needs to be covered using a single antenna. Implementing a UWB antenna in highly flexible, bendable,

**Figure 14.** Schematic of the demonstration setup for the implemented elastic RF radiation sensor.

**Figure 15.** Photographs of the elastic EM radiation sensor operating under deformation: with 15% elongation along its (a) *x*- and (b) *y*-axis, (c) omnidirectional stressing, and (d) severe twisting. The corresponding coordinate system can be found in Figure 10.

and stretchable formats would be of great interest. Such an antenna can also be deployed in other applications that only require a fractional bandwidth of UWB, as it is expected to feature robust electrical characteristics in some narrower frequency ranges under deformation. A nonresonant antenna, PICA, is chosen because of its uniplanar configuration and ease of fabrica‐ tion. The lower and upper operational frequencies of a PICA are determined by the height of the inverted cone and the size of the gap between the radiator and the ground, respectively. Its tapered shape gradually transforms the free space impedance of 377 Ω to 50 Ω over a wide frequency range. Full wave simulations performed in Ansoft HFSS suggest a radius of 10 mm (*R*) and a gap size of 300 µm (G) for covering the operational frequency range of 3–11 GHz. Schematic views of the liquid metal elastic PICA are seen in Figure 16. Similar to the previously presented microfluidic antennas, its ground and radiating element are realized in meshed configurations to separate the top and bottom PDMS membranes.

between the radiation source and the sensor would result in greater free space loss. Higher power is, therefore, required from the RF signal generator to turn on the LED in this case. In addition to measurements in its relaxed state, the stretchable EM radiation detector is also tested in its folded, twisted, and strained modes. Figure 15 shows that the implemented microfluidics- based soft radiation sensor prototype remains functioning regardless of

**Figure 12.** RF power module output voltage and LED status versus input RF power.

200 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

**Figure 13.** Photograph of the resulting elastic EM radiation sensor prototype.

Antennas with broad operational frequency bands are sometimes required. In ultra-wideband (UWB) systems, an extremely broad frequency range from 3.1 GHz to 10.6 GHz needs to be covered using a single antenna. Implementing a UWB antenna in highly flexible, bendable,

deformation.

**3.3. Fluidic, elastic ultra-wideband antenna**

Mechanical properties of the fabricated antenna are first evaluated. Figure 17 shows that the resulting PICA prototype is reversibly stretchable, foldable, and twistable. After removing external forces, the antenna always returns to its original shape without any visible mechanical damages.

**Figure 16.** Schematic of the microfluidics- based stretchable ultra-wide band antenna: (a) overlook view and (b) cross view of microfluidic channels in a PDMS elastomer.

**Figure 17.** Photographs of the elastic PICA under deformation: (a) 40% strain along (a) *x*- and (b) *y*-axes, (c) and (d) severe folding and twisting. The coordinate system is shown in Figure 16.

Relative stretchability as high as 100% should be within reach according the data sheet of the PDMS elastomer in use. Openings in the top PDMS membrane as well as the rigid feed cable set an upper limit of approximately 50% on stretchability. Low temperature test is also performed, using a household freezer with a lowest temperature of –24 o C. No mechanical failure can be found after cooling the antenna prototype at –24 o C for 2 hours.

Port impedance and radiation characteristics of the microfluidics- based stretchable PICA are all measured at room temperature. Measured reflection coefficient, *S11*, of the relaxed antenna is in line with the corresponding simulated result and good impedance matching (*S11*<–10 dB) is achieved between 3 GHz and 11 GHz, as seen in Figure 18.

**Figure 18.** Simulated and measured S11 of the stretchable ultra-wide band PICA in its original state.

Reflection coefficients of the stretchable fluidic PICA are also characterized in various flexed states. Experimental data presented in Figure 19 indicate that robust impedance matching is achieved even if the microfluidic PICA is strained.

**Figure 19.** Measured S11 of the flexed PICA. The coordinate system can be found in Figure 16.

**Figure 17.** Photographs of the elastic PICA under deformation: (a) 40% strain along (a) *x*- and (b) *y*-axes, (c) and (d)

**Figure 16.** Schematic of the microfluidics- based stretchable ultra-wide band antenna: (a) overlook view and (b) cross

Relative stretchability as high as 100% should be within reach according the data sheet of the PDMS elastomer in use. Openings in the top PDMS membrane as well as the rigid feed cable set an upper limit of approximately 50% on stretchability. Low temperature test is also

Port impedance and radiation characteristics of the microfluidics- based stretchable PICA are all measured at room temperature. Measured reflection coefficient, *S11*, of the relaxed antenna is in line with the corresponding simulated result and good impedance matching (*S11*<–10 dB)

C. No mechanical

C for 2 hours.

performed, using a household freezer with a lowest temperature of –24 o

failure can be found after cooling the antenna prototype at –24 o

is achieved between 3 GHz and 11 GHz, as seen in Figure 18.

severe folding and twisting. The coordinate system is shown in Figure 16.

view of microfluidic channels in a PDMS elastomer.

202 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

Stretching the antenna along its *x*-axis leads to increased height of the inverted cone and lowered first resonant frequency. Port impedance at higher frequencies remains stable in spite of stressing, as it is mainly governed by the impedance transformer formed by the tapered shape and the small gap between the antenna radiating element and the ground plane. Excessive strains along the *x*-axis might deteriorate impedance matching performance to some extent because of significantly increased gap size and severe deformation of the impedance transformer. Elongation along the *y*-axis only slightly impacts measured reflection coefficients because of relatively small changes in the tapered shape and the gap size. Radiation patterns of the relaxed and flexed elastic PICA are measured at 2.5 GHz and 5 GHz. Figure 20 illustrates the measured data at 2.5 GHz.

In comparison with antenna gain, variations on measured radiation patterns as a result of stressing are greater. Broad beam coverage similar to an ordinary monopole antenna is

**Figure 20.** Measured radiation patterns of the relaxed and stressed PICA in the (a), (b) xz-, and (c), (d) yz-plane. The corresponding coordinate system is depicted in Figure 16.

achieved in both relaxed and flexed cases. Maximum gain of approximately 2.2 dBi is measured at 2.5 GHz. Detailed results on radiation characteristics of the resulting elastic PICA can be found in Cheng et al. [14]. Strong ripples are seen in measured radiation patterns at 5 GHz. The major cause for these fast ripples is frequency- dependent disturbance introduced by the feed cable in the experiments.

Radiation efficiency is another important figure of merit for terminal antennas featuring omnidirectional radiation patterns. A custom-designed reverberation chamber is employed for rapid radiation efficiency measurements. Experimental results at different frequencies are plotted in Figure 21.

Although measured radiation efficiency decreases a bit at 3 GHz during stretching, it still remains above 70%, an acceptable figure, better than many commercially available antennas. No significant efficiency drop is detected over the entire frequency range. This further verifies

**Figure 21.** Measured radiation efficiency of the PICA in its relaxed and stretched states. The coordinate system is pre‐ sented in Figure 16.

the robustness of electrical connections in the implemented PICA under different forms of deformation.

#### **3.4. Multifunctional, microfluidics- based patch antenna and reversibly stretchable, selfcontained, wireless strain sensor**

A natural further step in the development of the emerging field of microfluidic soft electronics is to create more sophisticated multi-layered electronic devices and systems that can poten‐ tially open up a full spectrum of new applications. This, however, poses new challenges in device design, fabrication, and integration. The fabrication process of multi-layer microfluidic, elastomeric channels proposed in Section 2.2 is utilized to implement a reversibly stretchable, large-area, wireless strain sensor. This section details the design, fabrication, assembly, and characterizations of the sensor device.

The integrated wireless strain sensor contains an RF transmitter on a flexible PCB and a multifunctional, microfluidic, stretchable patch antenna for both wireless communication and strain sensing. The fluidic patch antenna is constructed in a double-layer configuration, *cf*. Figure 22. The principle for using an elastic antenna for strain sensing is simple. Because the resonant frequency of the patch antenna in use is largely determined by its electrical length along the *x*-axis, elongation along this direction would result in greater electrical length so that it decreases its resonant frequency. In the case of a patch antenna with narrow bandwidth (or high quality factor, *Q*), considerable impedance mismatch will occur at its original resonant frequency as a result of strain along its *x*-axis. When feeding such an antenna with a 50 Ω source at a constant power level, significantly decreased radiation will be measured in the farfield. Higher strain sensing sensitivity can be attained by reducing antenna bandwidth. But decreasing bandwidth would put high demands on fabrication precision. A tradeoff between sensitivity and fabrication yield has to be made. In addition to the multifunctional antenna, the integrated sensor also contains a simplified RF transmitter, which can be externally powered by a stack of two AA batteries connected serially. Details of this integrated electronic sensor device can be found in the schematic views presented in Figure 22.

achieved in both relaxed and flexed cases. Maximum gain of approximately 2.2 dBi is measured at 2.5 GHz. Detailed results on radiation characteristics of the resulting elastic PICA can be found in Cheng et al. [14]. Strong ripples are seen in measured radiation patterns at 5 GHz. The major cause for these fast ripples is frequency- dependent disturbance introduced by the

**Figure 20.** Measured radiation patterns of the relaxed and stressed PICA in the (a), (b) xz-, and (c), (d) yz-plane. The

Radiation efficiency is another important figure of merit for terminal antennas featuring omnidirectional radiation patterns. A custom-designed reverberation chamber is employed for rapid radiation efficiency measurements. Experimental results at different frequencies are

Although measured radiation efficiency decreases a bit at 3 GHz during stretching, it still remains above 70%, an acceptable figure, better than many commercially available antennas. No significant efficiency drop is detected over the entire frequency range. This further verifies

feed cable in the experiments.

corresponding coordinate system is depicted in Figure 16.

204 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

plotted in Figure 21.

**Figure 22.** Schematic view of the proposed elastic wireless strain sensor: (a) top view and (b) cross view of microfluidic channels.

The designed patch antenna is fed by a short piece of microstrip. The spacing between the patch and the ground plane determines the bandwidth of the antenna. The greater the spacing is, the wider bandwidth the antenna features. Impedance bandwidth is also affected by dielectric losses of the substrate. Lossy dielectric materials would increase antenna bandwidth at the expense of radiation efficiency. Poor radiation efficiency should be definitely avoided, as it significantly shortens remote sensing range. Resembling microfluidic antennas reported in the previous sections, both the antenna patch and ground plane are realized in the meshed configuration. The active circuit assembled on a flex foil comprises a 1.5 GHz voltage controlled oscillator (VCO) associated with a few discrete components. The effective strain sensing area is as large as the overall size of the integrated sensor, approximately 110.0 mm × 80.0 mm. It makes the resulting wireless sensor well suited for remote, large-area strain sensing.

Mechanical properties should be taken into account in the design as well. Thinner PDMS substrates are preferred because of higher elasticity. A very thin substrate would, however, constrain the spacing between the patch and the ground plane, and lead to too narrow antenna bandwidth. Parametric analysis in numerical simulations shows that good compromise values for the overall thickness, and the spacing are 2.5 mm and 1.5 mm. A microfluidic, flexible, stretchable wireless strain sensor is implemented, following the fabrication and integration process described previously. Two openings on the top surface of the PDMS elastomer are utilized for hybrid integration with RF transmitter. A photograph of the fabricated sensor device is presented in Figure 23.

**Figure 23.** Photograph of the resulting microfluidics- based stretchable wireless strain sensor.

**Figure 22.** Schematic view of the proposed elastic wireless strain sensor: (a) top view and (b) cross view of microfluidic

206 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

The designed patch antenna is fed by a short piece of microstrip. The spacing between the patch and the ground plane determines the bandwidth of the antenna. The greater the spacing is, the wider bandwidth the antenna features. Impedance bandwidth is also affected by dielectric losses of the substrate. Lossy dielectric materials would increase antenna bandwidth at the expense of radiation efficiency. Poor radiation efficiency should be definitely avoided, as it significantly shortens remote sensing range. Resembling microfluidic antennas reported in the previous sections, both the antenna patch and ground plane are realized in the meshed configuration. The active circuit assembled on a flex foil comprises a 1.5 GHz voltage controlled oscillator (VCO) associated with a few discrete components. The effective strain sensing area is as large as the overall size of the integrated sensor, approximately 110.0 mm × 80.0 mm. It

makes the resulting wireless sensor well suited for remote, large-area strain sensing.

Mechanical properties should be taken into account in the design as well. Thinner PDMS substrates are preferred because of higher elasticity. A very thin substrate would, however, constrain the spacing between the patch and the ground plane, and lead to too narrow antenna bandwidth. Parametric analysis in numerical simulations shows that good compromise values for the overall thickness, and the spacing are 2.5 mm and 1.5 mm. A microfluidic, flexible, stretchable wireless strain sensor is implemented, following the fabrication and integration process described previously. Two openings on the top surface of the PDMS elastomer are utilized for hybrid integration with RF transmitter. A photograph of the fabricated sensor

channels.

device is presented in Figure 23.

Prior to the final integration, the standalone patch antenna and RF transmitter circuit are tested individually. Reflection coefficients of the fluidic soft patch in response to varying strains are seen in Figure 24.

**Figure 24.** Simulated and measured S11 of the standalone fluidic patch antenna in its original state, and with varying strain along *x*-axis. The coordinate system is seen in Figure 22.

Experimental results indicate a dramatic resonant frequency decrease in response to incre‐ mental strain along the *x*-axis. Significant mismatch losses are observed at the oscillation frequency, 1.46 GHz, of the VCO during stretching. Measured radiation patterns (including mismatch losses) depicted in Figure 25 show greater than 10 dB realized antenna gain drop when the patch antenna is stressed from its original state to 15% elongation along the *x*-axis. Front-to-back ratio of radiation patterns is better than 10 dB. It should be noted that the meshed ground plane acts as an effective radiation shield and excellent front-to-back ratio of higher than 10 dB implies good forward radiation.

**Figure 25.** Simulated and measured radiation patterns (including mismatch losses) of the relaxed and flexed microflui‐ dic patch antenna in the (a) xz- and (b) yz-plane. The coordinate system is shown in Figure 22.

After antenna measurements, the active circuit and the elastic antenna are heterogeneously integrated. On the other end of the wireless strain sensing system, a custom-designed receiver is developed to receive and interpret sensing data. The system diagram of the entire test setup for the elastic wireless strain sensor is illustrated in Figure 26.

**Figure 26.** Illustration of the setup for demonstrating the microfluidic, reversibly stretchable, large-area wireless strain sensor.

As shown in Figure 26, the receiver consists of numerous sub-modules for RF power detection and is assisted by a personal computer (PC) for data readout and plotting. The output of the RF power detector is a DC voltage, proportional to the input RF power level. This voltage is measured by a high resolution digital multi-meter (DMM) controlled by the PC. The receiver unit is calibrated with varying RF power around 1.5 GHz, using a high precision synthesizer prior to the actual measurements.

**Figure 27.** (a) Photograph of the demonstration of the elastic, wireless strain sensor. (b) Voltage readouts at a customdesigned PC-aided receiver.

The microfluidic, reversibly stretchable wireless strain sensor is tested in an ordinary office environment, *cf*. Figure 27. The distance between the integrated sensor and the receiver station is roughly 5 m. Repeated strain with a period of approximately 10 seconds is manually applied along the *x*-axis of the sensor. The plot in Figure 27 displays the data readout at the PC, which is in line with the manual stretch. Reversibility can also be clearly seen in the plot, as the measured voltage always returns to the original level after the removal of external strain. Such an integrated sensor can be deployed as a comfortable wearable electronic device for real-time monitoring of body motion.

## **4. Conclusion**

**Figure 25.** Simulated and measured radiation patterns (including mismatch losses) of the relaxed and flexed microflui‐

After antenna measurements, the active circuit and the elastic antenna are heterogeneously integrated. On the other end of the wireless strain sensing system, a custom-designed receiver is developed to receive and interpret sensing data. The system diagram of the entire test setup

**Figure 26.** Illustration of the setup for demonstrating the microfluidic, reversibly stretchable, large-area wireless strain

As shown in Figure 26, the receiver consists of numerous sub-modules for RF power detection and is assisted by a personal computer (PC) for data readout and plotting. The output of the RF power detector is a DC voltage, proportional to the input RF power level. This voltage is measured by a high resolution digital multi-meter (DMM) controlled by the PC. The receiver unit is calibrated with varying RF power around 1.5 GHz, using a high precision synthesizer

dic patch antenna in the (a) xz- and (b) yz-plane. The coordinate system is shown in Figure 22.

for the elastic wireless strain sensor is illustrated in Figure 26.

208 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

sensor.

prior to the actual measurements.

An overview in the emerging field of ultracompliant, soft electronics that can be twisted, compressed, and stressed into almost any arbitrary shapes has been presented. Different strategies for realizing bendable, flexible, stretchable electronic devices and systems have been briefly discussed. Microfluidics- based soft electronics, one of the most promising approaches for implemented elastomeric electronics, has been addressed in detail. Fabrication techniques, hybrid integration concepts, as well as various electronic devices built in microfluidic, elastic formats are the focus of this chapter. Appealing application examples, ranging from singlelayer standalone fluidic stretchable antennas to multi-layer, self-contained, large-area, wireless sensors, have been demonstrated.

Research efforts in incorporating new materials and novel technologies into soft electronics are ongoing. Figure 28 shows an example of a hand-shaped, soft, rubbery sensor for multipoint, remote pressure sensing. Emerging nanomaterials and fabrication processes, e.g., pressure-activated conductive rubber and inkjet-printed graphene -based flexible circuits, are introduced to create novel pressure sensors and active RF electronics. Feasibilities of imple‐ menting microfluidic, elastic EM energy harvesting devices are also being explored. Successful outcomes will pave the way for realizing batteryless, ultracompliant, wearable and implanted electronics that can be powered by ambient EM energy generated from billions of wireless devices around us.

**Figure 28.** Photograph of the stretchable multi-point, wireless pressure sensor prototype.

Microfluidic flexible, compressible, stretchable electronics together with other members in the new category of soft electronics are reshaping the world of electronics so as to the future of our daily life. The number of cell phones already outpaced the world population in 2014. It is anticipated that the total number of connected devices will grow to approximately seven times of the entire population in the coming 5 years, among which billions of soft electronic devices will be found in many new applications, including healthcare/medical monitoring, tissue engineering, bio-inspired robotics, and so on.

## **Acknowledgements**

Research efforts in incorporating new materials and novel technologies into soft electronics are ongoing. Figure 28 shows an example of a hand-shaped, soft, rubbery sensor for multipoint, remote pressure sensing. Emerging nanomaterials and fabrication processes, e.g., pressure-activated conductive rubber and inkjet-printed graphene -based flexible circuits, are introduced to create novel pressure sensors and active RF electronics. Feasibilities of imple‐ menting microfluidic, elastic EM energy harvesting devices are also being explored. Successful outcomes will pave the way for realizing batteryless, ultracompliant, wearable and implanted electronics that can be powered by ambient EM energy generated from billions of wireless

210 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

**Figure 28.** Photograph of the stretchable multi-point, wireless pressure sensor prototype.

engineering, bio-inspired robotics, and so on.

Microfluidic flexible, compressible, stretchable electronics together with other members in the new category of soft electronics are reshaping the world of electronics so as to the future of our daily life. The number of cell phones already outpaced the world population in 2014. It is anticipated that the total number of connected devices will grow to approximately seven times of the entire population in the coming 5 years, among which billions of soft electronic devices will be found in many new applications, including healthcare/medical monitoring, tissue

devices around us.

Dr. Shi Cheng currently holds a Young Research Fellow position financed by the Swedish Research Council (VR) under the contract (Avtals-ID: C0486501). Also Dr. Jiantong Li ac‐ knowledges Göran Gustafsson Foundation for the financial support through the Young Researcher Prize (No. 1415 B).

## **Author details**

Babak Taghavi1 , Jiantong Li1 , Mikael Östling1 and Shi Cheng1,2

1 KTH, Royal Institute of Technology, Sweden

2 Ericsson AB, Sweden

## **References**


## **Microfluidic Sensors and Circuits for Internet of Things Applications**

John Yan

[8] Haopeng W., Debao Z., Jianguo C. Development of a Stretchable Conductor Array with Embedded Metal Nanowires. IEEE Transactions on Nanotechnology. 2013;12(4):561

[9] Brosteaux D., Axisa F., Gonzalez M., Vanfleteren J. Design and Fabrication of Elastic Interconnections for Stretchable Electronic Circuits. Electron Device Letters, IEEE.

[10] Siegel A. C., Bruzewicz D. A., Weibel D. B., Whitesides G. M. Microsolidics: Fabrica‐ tion of Three-Dimensional Metallic Microstructures in Poly(dimethylsiloxane). Ad‐

[11] Kim H., Son C., Ziaie B. A Multiaxial Stretchable Interconnect Using Liquid-Alloy-Filled Elastomeric Microchannels. Applied Physics Letters. 2008;92(011904):1-3. DOI:

[12] Dickey M. D., Chiechi R. C., Larsen R. J., Weiss E. A., Weitz D. A., Whitesides G. M. Eutectic Gallium-Indium (EGaIn): A Liquid Metal Alloy for the Formation of Stable Structures in Microchannels at Room Temperature. Advanced Functional Materials.

[13] Cheng S., Rydberg A., Hjort K., Wu Z. Liquid Metal Stretchable Unbalanced Loop Antenna. Applied Physics Letters. 2009; 94:144103. DOI: 10.1063/1.3114381

[14] Cheng S., Wu Z., Hallbjörner P., Hjort K., Rydberg A. Foldable and Stretchable Liq‐ uid Metal Planar Inverted Cone Antenna. IEEE Transactions on Antennas and Propa‐

[15] Kubo M., Li X., Kim C., Hashimoto M., Wiley B. J., Whitesides G.M. Stretchable Mi‐ crofluidic Radiofrequency Antennas. Advanced Materials. 2010;22:2749-2752. DOI:

[16] Cheng S., Wu Z. Microfluidic Stretchable RF Electronics. Lab on a Chip.

[17] Cheng S., Wu Z. A Microfluidic, Reversibly Stretchable, Large-Area Wireless Strain Sensor. Advanced Functional Materials. 2011;21:2282–2290. DOI: 10.1002/adfm.

[18] Hayes G. J., So J. H., Qusba A., Dickey M. D., Lazzi G. Flexible Liquid Metal Alloy (EGaIn) Microstrip Patch Antenna. IEEE Transactions on Antennas and Propagation.

[19] Huang Y., Wang Y., Xiao L., Liu H., Dong W., Yin Z. Microfluidic Serpentine Anten‐ nas with Designed Mechanical Tunability. Lab on a Chip. 2014;14:4205-4212. DOI:

[20] Jeong S. H., Hagman A., Hjort K., Jobs M., Sundqvist J., Wu Z. Liquid Alloy Printing of Microfluidic Stretchable Electronics. Lab on a Chip. 2012;12:4657-4664. DOI:

vanced Materials. 2007;19:727–733. DOI: 10.1002/adma.200601787

gation. 2009;57(12):3765-3771. DOI: 10.1109/TAP.2009.2024560


10.1063/1.2829595

10.1002/adma.200904201

201002508

10.1039/c4lc00762j

10.1039/c2lc40628d

2010;10:3227-3234. DOI: 10.1039/c005159d

2012;60(5):2151-2156. DOI: 10.1109/TAP.2012.2189698

2007;28(7):552 - 554. DOI: 10.1109/LED.2007.897887

212 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

2008;18:1097-1104. DOI: 10.1002/adfm.200701216

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/64346

#### **Abstract**

As we move into the Internet of Things (IoT) and cloud computing era, the number of sensors deployed which seamlessly integrate themselves into environment is growing rapidly. These sensors should be minimally intrusive, both optically and mechanically, while providing high temporal and spatial contextual awareness of its environment. In this chapter, microfluidic sensors and circuits are presented to better bridge the physical and digital world for healthcare applications. Specifically, a discussion of cardiovascular sensing, glaucoma diagnosis and flexible tactile sensor arrays for smart skin application is presented.

**Keywords:** Circuits, digital health, internet of things, microfluidics, pervasive computing, sensors

## **1. Introduction**

As we move into the Internet of Things (IoT) and cloud computing era, the number of sensors deployed that seamlessly integrate themselves into environment is growing rapidly [1‒3]. The research and development challenges to create a smart and interconnected world necessitate a new paradigm in Internet architecture. This architecture requires the consideration of security, software platforms, ethical implications, standardization, smart sensors, and prag‐ matic business models. This chapter has honed down this broad vision to specifically focus on a few contributions from microfluidic sensors and circuits to better bridge the physical and digital world for healthcare applications.

#### **1.1. Overview of Internet of Things for healthcare**

As we move into the IoT and cloud computing era, the number of sensors deployed that seamlessly integrate themselves into environment is growing rapidly. This concept is descri‐

bed a totally interconnected world where devices of every shape and size are manufactured with "smart" capabilities that allow them to communicate and interact with other devices, exchange data, make autonomous decisions, and perform useful tasks based on preset conditions. Figure 1 shows an ecosystem of IoT's relationship with people and the home within the modern cloud computing environment. Wearable devices and sensors would be ubiqui‐ tously employed to continuously monitor health and infrastructure that would subsequently be uploaded to data centers and archived as datasets. These datasets then provide the training necessary for data scientists and physicians to make intelligent predictions based on the behavior of its clients.

**Figure 1.** Data flow of Internet of Things devices for healthcare applications.

#### **1.2. Microfluidics for IoT**

Microfluidics is a multidisciplinary field intersecting engineering, nanotechnology, physics, and chemistry with practical applications to design systems in which small volumes of fluids will be handled [4‒6]. In this chapter, we touch on the various facets of this multidisciplinary field and present applications on how microfluidic circuits and sensors can be utilized in an IoT environment. Figure 2 shows the variety of sensors and circuits for IoT healthcare applications ranging from cardiovascular sensing (to be integrated with smart-watch appli‐ cations [7]) to unpowered microfluidic pressure sensors for glaucoma diagnosis [8] to flexible

tactile sensor arrays for smart skin applications [9]. Each of these devices will be addressed in more detail throughout this chapter.

**Figure 2.** Example of IoT sensors for healthcare applications.

bed a totally interconnected world where devices of every shape and size are manufactured with "smart" capabilities that allow them to communicate and interact with other devices, exchange data, make autonomous decisions, and perform useful tasks based on preset conditions. Figure 1 shows an ecosystem of IoT's relationship with people and the home within the modern cloud computing environment. Wearable devices and sensors would be ubiqui‐ tously employed to continuously monitor health and infrastructure that would subsequently be uploaded to data centers and archived as datasets. These datasets then provide the training necessary for data scientists and physicians to make intelligent predictions based on the

214 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

behavior of its clients.

**Figure 1.** Data flow of Internet of Things devices for healthcare applications.

Microfluidics is a multidisciplinary field intersecting engineering, nanotechnology, physics, and chemistry with practical applications to design systems in which small volumes of fluids will be handled [4‒6]. In this chapter, we touch on the various facets of this multidisciplinary field and present applications on how microfluidic circuits and sensors can be utilized in an IoT environment. Figure 2 shows the variety of sensors and circuits for IoT healthcare applications ranging from cardiovascular sensing (to be integrated with smart-watch appli‐ cations [7]) to unpowered microfluidic pressure sensors for glaucoma diagnosis [8] to flexible

**1.2. Microfluidics for IoT**

#### **2. Coupled optomechanical pressure and flow measurements**

Cardiovascular disease (CVD) is estimated to affect 81 million Americans adults [10]. An important determinant of these diseases is the arterial wall stiffness [11]. This section describes the development of a simple, non-invasive, real-time detection system that utilizes pulse wave velocity and pressure pulsation measurements to estimate vascular resistance and compliance in the radial artery that can provide the physician important diagnostic information. This system is realized by utilizing a 780 nm laser Doppler velocimeter to obtain the local flow rates and a piezoelectric pressure sensor to measure the pressure pulsation at the radial artery. Here, the sensor measurement results at the proximal and distal radial arteries are presented in both the time and frequency domain. While the presented results are demonstrated utilizing tabletop instrument, ultimately this system can be completely integrated for a light-weight, portable, real-time monitoring system that can potentially be embedded into textiles for seamless monitoring of important cardiovascular signals.

Figure 3 shows an overview of the IoT device that is used to monitor the vascular impedance through coupled optomechanical pressure (P) and flow (Q) measurements whose form factor should be small enough to be integrated with a wrist watch. On top of the biomaterial (bone, skin, blood, etc.) lies an optically transparent pressure sensor with integrated optics and electronics for digital signal processing and wireless communication to the cloud. Such a device facilitates the monitoring of signatures of cardiovascular disease using personalized datasets for real-time, continuous monitoring.

**Figure 3.** IoT vascular impedance assessment through coupled optomechanical pressure and flow measurements.

To measure the arterial flow, the laser Doppler principle, consisting of a laser light source shone on the radial artery and collection of the backscattered light collected by the photodiode, can be used. The total backscattered light consists of a component from the static tissue and a component with frequency shifted light, whose shifted amount is dependent on the speed of moving blood. These two components are collected and mixed on a photodetector whose corresponding photocurrent can then be post processed to indicate the velocity of moving red blood cells. This optical detection system should utilize *λ*0 = 780 nm light with a typical output power of less than 1 mW of power. Therefore, the flow is:

$$Q = v\_{s,r} \pi r^2 = -\frac{\Delta f}{n\_{\text{bo,tissue}}} \lambda\_0 \pi r^2 \tag{1}$$

where *n*bio,tissue(780 nm) ~ 2.6.

The pressure sensor utilizes the concept tonometry to apply tension to the radial artery. The pressure sensor should be sufficiently flexible to conform to the skin and be optically trans‐ parent to the laser Doppler velocimeter system. Furthermore, it should have a sensitivity of 2 mmHg with a dynamic pressure range of 200 mmHg and an electromechanical bandwidth of 10 Hz. A detailed description of the construction and operation of such a pressure sensor built on microfluidic concepts will be described later on in this chapter.

**Figure 4.** Model of cardiovascular system.

portable, real-time monitoring system that can potentially be embedded into textiles for

Figure 3 shows an overview of the IoT device that is used to monitor the vascular impedance through coupled optomechanical pressure (P) and flow (Q) measurements whose form factor should be small enough to be integrated with a wrist watch. On top of the biomaterial (bone, skin, blood, etc.) lies an optically transparent pressure sensor with integrated optics and electronics for digital signal processing and wireless communication to the cloud. Such a device facilitates the monitoring of signatures of cardiovascular disease using personalized datasets

**Figure 3.** IoT vascular impedance assessment through coupled optomechanical pressure and flow measurements.

power of less than 1 mW of power. Therefore, the flow is:

where *n*bio,tissue(780 nm) ~ 2.6.

To measure the arterial flow, the laser Doppler principle, consisting of a laser light source shone on the radial artery and collection of the backscattered light collected by the photodiode, can be used. The total backscattered light consists of a component from the static tissue and a component with frequency shifted light, whose shifted amount is dependent on the speed of moving blood. These two components are collected and mixed on a photodetector whose corresponding photocurrent can then be post processed to indicate the velocity of moving red blood cells. This optical detection system should utilize *λ*0 = 780 nm light with a typical output

> 2 2 s,r 0 bio,tissue

l p

<sup>D</sup> = = - (1)

*<sup>f</sup> Qv r <sup>r</sup> n*

p

seamless monitoring of important cardiovascular signals.

216 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

for real-time, continuous monitoring.

To describe the cardiovascular system, one can use transmission line analog between hydraulic transmission system of compliant tubes and an electrical transmission lines as shown Figure 4. Here, the blood flow and pressure waveforms are used to provide a baseline model to validate our measurement results where the signal source represents the heart. Based on the coupled Navier-Stokes equations [12]:

$$-\frac{\partial P}{\partial \mathbf{x}} = QR + L\frac{\partial Q}{\partial t} \tag{2}$$

$$-\frac{\partial \mathbb{Q}}{\partial \mathbf{x}} = \mathbf{PG} + \mathbf{C} \frac{\partial \mathbf{P}}{\partial t} \tag{3}$$

where arterial flow can be modeled as a dispersive transmission lines that broadens and distortions as it travels downward the hydraulic line. The different frequency components of propagation are then described by the standard impedance (*Z*0) and reflection (Γ) relationships:

$$Z\_0 = \sqrt{\frac{R + j\rho\mathcal{L}}{G + j\rho\mathcal{C}}}\tag{4}$$

$$
\Gamma\_{\perp} = \frac{Z\_{\perp} - Z\_{0}}{Z\_{\perp} + Z\_{0}} \tag{5}
$$

$$
\Gamma\_{\rm S} = \frac{\mathbf{Z\_{S}} - \mathbf{Z\_{0}}}{\mathbf{Z\_{S}} + \mathbf{Z\_{0}}} \tag{6}
$$

where ΓL is the reflection coefficient at the load and ΓS is the reflection coefficient at the source. The mechanical resistance (*R*), compliance (*C*), and inductance (*L*) are governed by the following equations:

$$R = \frac{P}{Q} \tag{7}$$

$$\mathbf{C} = \frac{\partial V}{\partial P} \tag{8}$$

$$L = \frac{P}{dQ / dt} \tag{9}$$

where resistance is the ratio of pressure over flow, compliance is the change of volume of fluid (∂*P*) over the change of pressure (∂*P*) and the inductance is pressure over the change of flow with respect to time (*dQ* / *dt*).

When the pressure and flow can be modeled as a Gaussian pulse with reflected waves that occur during discontinuities of impedances:

$$P(z=0) = P\_i + P\_r = P\_i(1 + \Gamma e^{-2t\_{\text{day}}} + \Gamma^2 e^{-2(2t\_{\text{day}})} \cdots) \tag{10}$$

$$t\_{\text{delay}} = \frac{1}{\text{length}} \frac{\partial}{\partial \theta} \tag{11}$$

A sample of the measured results is shown in Figure 5 where the vascular impedance values of compliance, inductance and resistance can be obtained. Using reflection amplitudes, we can obtain the source and load impedance values. From the waveform, we not only know the vascular impedance at the point of optomechanical measurement but we can also estimate the vascular impedance at the source and load. The discontinuities along the arterial hydraulic line are a result of bifurcations that leads to reflected waves along the dispersive line. Time

L 0

<sup>+</sup> (5)

<sup>+</sup> (6)

*<sup>Q</sup>*<sup>=</sup> (7)

¶ <sup>=</sup> ¶ (8)

*dQ dt* <sup>=</sup> (9)

<sup>=</sup> (11)

= = + = +G +G L (10)

L 0 *Z Z Z Z* - G =

S 0

S 0 *Z Z Z Z* - G =

where ΓL is the reflection coefficient at the load and ΓS is the reflection coefficient at the source. The mechanical resistance (*R*), compliance (*C*), and inductance (*L*) are governed by the

*<sup>P</sup> <sup>R</sup>*

*<sup>V</sup> <sup>C</sup> P*

/ *<sup>P</sup> <sup>L</sup>*

where resistance is the ratio of pressure over flow, compliance is the change of volume of fluid (∂*P*) over the change of pressure (∂*P*) and the inductance is pressure over the change of flow

When the pressure and flow can be modeled as a Gaussian pulse with reflected waves that

delay delay <sup>2</sup> 2(2 ) <sup>2</sup> ( 0) (1 ) *t t*

1 length

A sample of the measured results is shown in Figure 5 where the vascular impedance values of compliance, inductance and resistance can be obtained. Using reflection amplitudes, we can obtain the source and load impedance values. From the waveform, we not only know the vascular impedance at the point of optomechanical measurement but we can also estimate the vascular impedance at the source and load. The discontinuities along the arterial hydraulic line are a result of bifurcations that leads to reflected waves along the dispersive line. Time

w

b

*ir i Pz P P P e e* - -

delay

*t*

L

218 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

S

following equations:

with respect to time (*dQ* / *dt*).

occur during discontinuities of impedances:

**Figure 5.** Sample of measured results from vascular impedance assessment of coupled optomechanical pressure and flow: (a) time domain; (b) frequency domain.

domain information can also be discrete Fourier transformed, whose signal can be improved with windowing functions due to the discrete sampling nature, to find spectral information.

### **3. Microfluidic sensor for point-of-care glaucoma diagnosis**

An important microfluidic sensor developed for IoT applications is for the continuous monitoring of glaucoma that an estimated 67 million people are believed to suffer worldwide [13]. Patients with glaucoma are considered "well controlled" if their mean intraocular pressure (IOP) is lower than 21 mmHg. Owing to a rapidly aging population, it is estimated that the number of open angle glaucoma cases will increase to 3.4 million in 2020, making it the second leading cause of blindness and the first leading cause of irreversible blindness in the United States. Since unregulated IOP can lead to irreversible blindness by pinching the optic nerve, as seen in Figure 6, it is of paramount importance to monitor this pressure and make low-cost, point-of-care diagnostic tools available. Various wireless techniques have been conceived to continuously monitor intraocular pressure [16‒19], which require radio frequen‐ cy (RF) power transfer to power the device. However, a slew of potential health issues can arise from long term exposure to the high RF power transfer needed to power the device. Microfluidic solutions provides an unique solution due to its low-cost and biocompatible material construct that is additionally amendible to large-scale manufacturing.

**Figure 6.** Glaucoma is a result of build up of intraocular pressure; figure adopted from [20].

The dynamical IOP measure is based on Laplace's principle where the pressure inside the hydraulic chamber (*Pt* ) is dynamically determined by loading pressure and measured by the built-in microfluidic pressure sensor. The sensing membrane deflects and measures the pressure difference (*P*s) at each contact point through the individual surface sensing element(s). Polydimethylsiloxane (PDMS) is used as the structural material to create the flexible, polymeric membrane and surface sensing for its elasticity and flexibility as well as excellent physical properties with a Young's modulus (*E*) of approximately 500 kPa [21].

The device to test the sensing principle is illustrated in Figure 7. It consists of a large, circular sensing chamber network with height *H* and radius *r*. This is followed by sensing channels of width *w* and height *h*, which act as the sensing elements. When local pressure is applied to the sensing chamber network, an internal strain is induced on the elastomer structure. This generates a pressure gradient internally within in the microfluidic network. As a result, the compression of the sensing chamber leds to an outward fluid displacement onto the sensing channel. As stress is released from the sensing chamber, the elastomer recovery properties of PDMS create a negative pressure to withdraw fluid from the sensing channel.

**Figure 7.** Schematic illustration of calibration device.

The microfluidic network is carefully designed to prevent air bubble formation at the sensing chamber during the injection process. This is done by installing an injection port before the sensing chamber network. Laplace valves at the entrance and exit of the sensing chamber are designed to prevent bubble cavitation as the fluid flows from the fluidic interconnects to the sensing chamber. To minimize compressive effects of displacement of air in the sensing channel during fluidic displacement, a large out-flow chamber is designed to have a volume a thou‐ sand times larger than the sensing channel, acting as a pressure relief conduit. A long, rectan‐ gular straight sensing channel is used to characterize the sensing principle of the device since its laminar flow profile characteristics are well understood. By optically observing the magni‐ tude of fluidic displaced, the resulting pressure on the sensing chamber can be determined.

**Figure 8.** Equivalent circuit model of transducer.

**Figure 6.** Glaucoma is a result of build up of intraocular pressure; figure adopted from [20].

220 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

properties with a Young's modulus (*E*) of approximately 500 kPa [21].

PDMS create a negative pressure to withdraw fluid from the sensing channel.

hydraulic chamber (*Pt*

**Figure 7.** Schematic illustration of calibration device.

The dynamical IOP measure is based on Laplace's principle where the pressure inside the

built-in microfluidic pressure sensor. The sensing membrane deflects and measures the pressure difference (*P*s) at each contact point through the individual surface sensing element(s). Polydimethylsiloxane (PDMS) is used as the structural material to create the flexible, polymeric membrane and surface sensing for its elasticity and flexibility as well as excellent physical

The device to test the sensing principle is illustrated in Figure 7. It consists of a large, circular sensing chamber network with height *H* and radius *r*. This is followed by sensing channels of width *w* and height *h*, which act as the sensing elements. When local pressure is applied to the sensing chamber network, an internal strain is induced on the elastomer structure. This generates a pressure gradient internally within in the microfluidic network. As a result, the compression of the sensing chamber leds to an outward fluid displacement onto the sensing channel. As stress is released from the sensing chamber, the elastomer recovery properties of

) is dynamically determined by loading pressure and measured by the

The mechano-fluidic transduction can be modeled as a circuit network as shown in Figure 8. Pressure applied to the sensing chamber with a set velocity results in a corresponding flow of fluid due to the change in the internal pressure displacement. A transformer is used to model the conversion from solid mechanical displacement of the elastomer to the fluidic displacement in the microfluidic network. The displacement amplification *A* is set by the geometry of the sensing chamber and sensing channel. The stiffness of the sensing chamber is:

$$k\_{\text{PDMS}} = \frac{\text{stress}}{\text{strain}}\tag{12}$$

According to the strain-stress relationship, the change of the micro-chamber height can be expressed as, Δ*H*/*H* = *σ*/*E*, where *E* is Young's Modulus of PDMS elastomer, and *σ* is the loading pressure. The displaced fluidic volume enters the incompressible microchannel, where the geometry (i.e., the cross-section area) determines the wetted length (Δ*l*) of the colored fluid. Herein, the sensitivity (*S*) can be calculated as

$$S = \frac{P}{\Delta l} = \frac{whE}{\pi r^2 H} \tag{13}$$

To assess the frequency response of the sensor, the microfluidic sensing system can be modeled as a first-order linear circuit, in which the micro-chamber membrane compliance *C* and the microchannel resistance *R* are:

$$\mathbf{C} = \frac{\pi r^2 H}{E} \tag{14}$$

$$R = \frac{12\,\mu\Delta l}{wh^3} \tag{15}$$

where *µ* is the viscosity of the sensing fluid. *D*<sup>f</sup> and *D*<sup>b</sup> represent the effect of surface tension and the resulting difference in internal pressure that needs to be overcome before the fluid can displace in the sensing channel. This surface tension is set by the surface properties of the material.

**Figure 9.** Optical photograph of the microfabricated device; scale bar is 5 mm.

To analyze the fluidic response, it is assumed that flow is dominated by pressure driven flow in the laminar region so the lubrication theory approximation can be used for the Navier-Stokes equation. The cutoff frequency, *fc*, is set by the fluidic resistance and capacitance [22] as follows:

$$f\_c = \frac{1}{2\pi\tau\_{\text{fl}}} = \frac{1}{2\pi RC} = \frac{1}{2\pi \left(\frac{12\,\mu\text{M}}{\omega h^3}\right) \left(\frac{\pi r^2}{E}\Delta H\right)}\tag{16}$$

To accommodate the human cornea with an approximately a diameter of 7.8 mm, the micro‐ fluidic sensing chambers and corresponding meandered sensing channels are designed on the peripheral of the contact lens at the sclera area. A realized microfluidic device for IoT appli‐ cations is shown in Figure 9. For the prototyped contact lens, the sensing channel length is over 80 mm with a sensing channel width of 20 µm. This corresponds to a dynamic range of 130 mmHg—more than sufficient for measuring the IOP. The characterization results of the device can be found in [8].

## **4. Microfluidic sensor for smart skin applications**

Ubiquitous sensing and smart skin applications that leverage flexible substrates for ultra-high sensitivity pressure sensing is of great interest to the IoT community. This is especially of interest when the sensor optical properties can be tuned to be optically transparent for a host of applications. Figure 10 shows a 3 × 3 array of optical transparent, microfluidic pressure sensors.

**Figure 10.** 3 × 3 array of optically transparent, microfluidic pressure sensor array (scale bar is 1 mm) for smart skin applications.

#### **4.1. Device operation**

<sup>2</sup> *r H <sup>C</sup> E* p

222 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

where *µ* is the viscosity of the sensing fluid. *D*<sup>f</sup>

**Figure 9.** Optical photograph of the microfabricated device; scale bar is 5 mm.

fl

pt

material.

3 <sup>12</sup> *<sup>l</sup> <sup>R</sup> wh* m

and the resulting difference in internal pressure that needs to be overcome before the fluid can displace in the sensing channel. This surface tension is set by the surface properties of the

To analyze the fluidic response, it is assumed that flow is dominated by pressure driven flow in the laminar region so the lubrication theory approximation can be used for the Navier-Stokes equation. The cutoff frequency, *fc*, is set by the fluidic resistance and capacitance [22] as follows:

3

m

*wh E*

 p (16)

è øè ø

c 2

p

To accommodate the human cornea with an approximately a diameter of 7.8 mm, the micro‐ fluidic sensing chambers and corresponding meandered sensing channels are designed on the peripheral of the contact lens at the sclera area. A realized microfluidic device for IoT appli‐

== = æ ö <sup>D</sup> æ ö ç ÷ç ÷ <sup>D</sup>

2 2 <sup>12</sup> <sup>2</sup> *<sup>f</sup> RC l r <sup>H</sup>*

 p

11 1

= (14)

<sup>D</sup> <sup>=</sup> (15)

and *D*<sup>b</sup> represent the effect of surface tension

The architecture of the microfluidic, capacitive pressure sensor is illustrated in Figure 11. It consists of a soft, micromachined elastomer to house fluid on a rigid plastic substrate. The highly deformable sensing chamber is designed to be tall and large to hold a volume of fluid much greater than the capacity of the sensing channel. Electrodes, in this case, transparent conductive oxide (TCO), are used to detect the degree of fluidic displacement as the sensing chamber deforms under applied pressures. Specifically, the large interfacial capacitance (>20 µF/cm2 ) from the TCO and room temperature ionic liquid (RTIL) is employed [23]. To prevent air bubble generation, Laplace valves are placed at the exit and entrance of the sensing chambers. Optional mechanical concentrator(s) can be integrated to the sensor for additional sensitivity.

As pressure is applied at the sensing chamber, strain is induced on the elastomer housing the microfluidic network. In turn, an internal pressure gradient between the sensing chamber and sensing channel leads to an outward fluid displacement across the sensing channel. Due to the geometry difference between the sensing chamber and sensing channel, mechanical displace‐

**Figure 11.** Microfluidic capacitive pressure sensor architecture.

ment amplification occurs as a result of the conservation of mass—a small compressive strain in the sensing chamber leads to a large displacement of fluidic across the sensing channel. Consequently, a change in the interfacial capacitance is detected across the coplanar electrodes. The causality of physics is illustrated in Figure 12, leading to a change in capacitance—where the strain on the elastomeric housing is exaggerated for illustrative purposes. When pressure is released from the sensing chamber, vacuum force is generated from the elastomer recovery, receding the fluid back to the sensing chamber and away from the coplanar electrodes. A mechanical concentrator, constructed out of rigid (~3 GPa) plastic, can subsequently be superimposed on top of the sensing chamber to further improve the sensitivity of the sensor. The concentrator serves to focus the mechanical pressure on the sensing chamber by using the area difference between the top disk area and the bottom disk that is in contact with the sensing chamber.

#### *4.1.1. Interfacial capacitance*

At the interface between RTIL and Indium Tin Oxide (ITO), there exists an electrochemical energy between *Φm* and *π*\* as shown in Figure 13a. In the inset, an example of an RTIL is shown. In this case 1-Butyl-3-Methylimidazolium Tetrafluroborate (BMImBF4) has attractive proper‐ ties of low surface tension and is readily available, consisting of an anion and a cation. In order to maintain electrostatic neutrality between the two materials, a thin layer of ions are collected at the surface. As a result, an interfacial capacitance is formed. The associated electrical circuit model is shown in Figure 13b. The equivalent circuit model of the interfacial capacitance, *C*I, can be described by the Gouy-Chapman Model [24]:

$$\frac{1}{C\_{\rm I}} = \frac{1}{C\_{\rm H}} + \frac{1}{C\_{\rm C}} = \frac{2}{\left(w - g\right)l} \left| \frac{d\_{\rm OHP}}{\varepsilon\_{0}\varepsilon\_{r}} + \frac{L\_{\rm D}}{\varepsilon\_{0}\varepsilon\_{r}\cosh\left(\frac{z\Phi\_{0}}{2LI\_{t}}\right)} \right| \equiv \frac{2}{\left(w - g\right)l} \frac{d\_{\rm OHP}}{\varepsilon\_{0}\varepsilon\_{r}}\tag{17}$$

**Figure 12.** Illustration of device operation with equivalent circuit: (a) initial pressure and capacitance; (b) measured pressure and capacitance.

where *w* is the width of microfluidic channel, *g* is the gap between the electrodes, *ε*0*ε*<sup>r</sup> / *d*OHP is the capacitance per unit area, and *l* is the length of fluid overlap from the sensing channel mouth. The charge transfer resistance, *Rct*, is a high-value resistance (~MΩ) due to tunnel current leakage can be described as

$$R\_{ct} = \frac{RT}{nF i\_0} \tag{18}$$

where *R* is the universal gas constant, *T* is the temperature, *F* is the Faraday constant, *n* is the number of electrons in the unit reaction and i0 is the current. *R*oxide and *R*bulk are the finite resistance of the electrolyte and metal.

The associated change in fringe capacitance, *C*p can be described as [25]

ment amplification occurs as a result of the conservation of mass—a small compressive strain in the sensing chamber leads to a large displacement of fluidic across the sensing channel. Consequently, a change in the interfacial capacitance is detected across the coplanar electrodes. The causality of physics is illustrated in Figure 12, leading to a change in capacitance—where the strain on the elastomeric housing is exaggerated for illustrative purposes. When pressure is released from the sensing chamber, vacuum force is generated from the elastomer recovery, receding the fluid back to the sensing chamber and away from the coplanar electrodes. A mechanical concentrator, constructed out of rigid (~3 GPa) plastic, can subsequently be superimposed on top of the sensing chamber to further improve the sensitivity of the sensor. The concentrator serves to focus the mechanical pressure on the sensing chamber by using the area difference between the top disk area and the bottom disk that is in contact with the sensing

At the interface between RTIL and Indium Tin Oxide (ITO), there exists an electrochemical energy between *Φm* and *π*\* as shown in Figure 13a. In the inset, an example of an RTIL is shown. In this case 1-Butyl-3-Methylimidazolium Tetrafluroborate (BMImBF4) has attractive proper‐ ties of low surface tension and is readily available, consisting of an anion and a cation. In order to maintain electrostatic neutrality between the two materials, a thin layer of ions are collected at the surface. As a result, an interfacial capacitance is formed. The associated electrical circuit model is shown in Figure 13b. The equivalent circuit model of the interfacial capacitance, *C*I,

( ) ( ) OHP <sup>D</sup> OHP

2 *<sup>t</sup>*

*U*

e e (17)

*d d L*

cosh

æ ö ç ÷

è ø è ø

IHG 0 r 0 0 r 0 r

e e

111 2 2

e e

*CC C w gl z w gl*

=+= + @ - - æ ö <sup>F</sup> ç ÷

chamber.

*4.1.1. Interfacial capacitance*

can be described by the Gouy-Chapman Model [24]:

**Figure 11.** Microfluidic capacitive pressure sensor architecture.

224 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

$$\mathbf{C}\_{\rm p} = \varepsilon\_0 \varepsilon\_{\rm ef} \frac{\mathbf{K}(k\_0')}{\mathbf{K}(k)} \tag{19}$$

where *K*(*k*) is complete elliptic integrals of the first kind described as

$$k\_0 = \frac{\mathcal{g}}{s+\mathcal{g}}\tag{20}$$

**Figure 13.** (a) Band diagram of the energy potential; (b) equivalent circuit model.

$$
\vec{k\_0} = \sqrt{1 - \vec{k\_0}^2} \tag{21}
$$

where *g* is the gap distance of the edge of the electrode from the symmetrical plane of the coplanar electrodes and *s* is the effective electrode distance as a result of the parasitic coplanar capacitance between the two electrodes whose value is typically much smaller than *C*I.

#### *4.1.2. Electromechanical model*

The equivalent electromechanical circuit model is shown in Figure 14. According to the strainstress relationship, the change of the micro-chamber height (*H*) can be expressed as, Δ*H*/*H* = *σ*/*E*, where *E* is Young's Modulus of PDMS elastomer, and, *σ* is the loading pressure.

As normal pressure is applied at the sensing chamber, a resulting strain is induced on the elastomer housing the microfluidic network following the simple stress-strain relationship

$$\frac{\Delta P}{E} = \frac{\Delta H}{H} \tag{22}$$

As a result, an internal pressure gradient occurs between the sensing chamber and sensing channel in the microfluidic channel creating an outward fluid displacement across the sensing channel due to conversation of mass, as exhibited by the *πr* <sup>2</sup> Δ*H* =*wh* Δ*l* relationship. By causalities in physics as previously described, a change in capacitance will be observed across the electrodes due to changes in pressure.

By inserting the stress-strain relationship and interfacial capacitance into the conversation of mass equation, the sensitivity of the device can be described by the following equation:

$$S = \frac{\partial C}{\partial P} = \frac{H}{E} \pi r^2 \frac{\mathcal{E}\_0 \mathcal{E}\_{\text{r,fluid}}}{d\_{\text{double layer}}} \left(\frac{w - g}{2}\right) \frac{1}{wh} \frac{1}{2} \tag{23}$$

where *H* is the height of the sensing chamber, *r* is the radius of the sensing chamber, *E* is the Young's modulus of the membrane, *w* is the width of the sensing channel, *g* is the electrode gap, *h* is the height of the sensing channel and *ε*0*ε*r,fluid/*d*double layer is the interfacial capacitance per unit area. This equation shows that the sensitivity of the sensor can be easily tuned by adjusting the geometries of the microfluidic network.

**Figure 14.** Equivalent electromechanical circuit model.

' 2

**Figure 13.** (a) Band diagram of the energy potential; (b) equivalent circuit model.

226 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

*4.1.2. Electromechanical model*

where *g* is the gap distance of the edge of the electrode from the symmetrical plane of the coplanar electrodes and *s* is the effective electrode distance as a result of the parasitic coplanar capacitance between the two electrodes whose value is typically much smaller than *C*I.

The equivalent electromechanical circuit model is shown in Figure 14. According to the strainstress relationship, the change of the micro-chamber height (*H*) can be expressed as, Δ*H*/*H* =

As normal pressure is applied at the sensing chamber, a resulting strain is induced on the elastomer housing the microfluidic network following the simple stress-strain relationship

As a result, an internal pressure gradient occurs between the sensing chamber and sensing channel in the microfluidic channel creating an outward fluid displacement across the sensing

causalities in physics as previously described, a change in capacitance will be observed across

By inserting the stress-strain relationship and interfacial capacitance into the conversation of mass equation, the sensitivity of the device can be described by the following equation:

*σ*/*E*, where *E* is Young's Modulus of PDMS elastomer, and, *σ* is the loading pressure.

*P H E H*

channel due to conversation of mass, as exhibited by the *πr* <sup>2</sup>

the electrodes due to changes in pressure.

0 0 *k k* = -1 (21)

D D <sup>=</sup> (22)

Δ*H* =*wh* Δ*l* relationship. By

The fluidic resistance is assumed to operate by a pressure driven flow in the laminar region so the lubrication theory approximation can be used for the Navier-Stokes equation. The fluidic resistance, defined as ratio of hydrodynamic pressure over volume flow rate, of the sensing channel as

$$R = \frac{12\,\mu}{\omega\eta\hbar^3} \left[ 1 - \frac{h}{w} \left( \frac{192}{\pi^5} \sum\_{n=1,3,5}^{v} \frac{1}{n^5} \tanh\left(\frac{n\pi w}{2h}\right) \right) \right]^{-1} \approx \frac{12\,\mu\Delta L}{\omega\eta\hbar^3} \tag{24}$$

when *h* << *w*. The fluidic capacitance is described as the ratio of chamber volume over applied hydrodynamic pressure. For the sensing chamber, this can be modeled as

$$C = \frac{\pi r^2}{E} \Delta H$$

when *h* << w. The microfluidic resistance and capacitance are believed to play an important role in the frequency response of the sensor. The fluidic capacitance is described as the ratio of chamber volume over applied hydrodynamic pressure.

Additionally, the amplification of mechanical concentrator is simply:

$$A\_{\rm m} = \left(\frac{D\_{\rm out}}{D\_{\rm in}}\right)^2\tag{26}$$

where *D*out is the upper diameter of the mechanical amplifier and *D*in is the lower diameter of the mechanical amplifier [26].

#### **4.2. Device fabrication**

Techniques to fabricate microfluidics [27, 28] have been modified for these devices. A typical fabrication process is illustrated in Figure 15. The master mold, shown in Figure 16a, is fabricated by a two-step SU-8 process on a silicon substrate. The first step consists of forming the buffer channel, sensing channel, drain channel, and the associated microfluidic intercon‐ nects to a height of 15 µm. The second step consists of forming the sensing chamber, out-flow chambers and injection port to have a height of 200 µm. The thin film ITO electrodes are patterned with a hydrochloric acid wet etch process and traditional photolithography. Next, the PDMS elastomer is fabricated with a 10:1 (base: agent) mixture to create a thick replica mold of 1 mm. This replica mold is subsequently aligned to the ITO electrode pattern and bonded onto the glass substrate through oxygen plasma pretreatment as shown in Figure 16c.

**Figure 15.** Microfabrication process.

A BD 30½ G needle is inserted into the injection port of the elastomer housing and a controlled volume of fluid is infused into the microfluidic network from a glass syringe using a syringe pump at a calibrated flow rate. Due to the small diameter of the gauge needle, the puncture

2 out

(26)

in

where *D*out is the upper diameter of the mechanical amplifier and *D*in is the lower diameter of

Techniques to fabricate microfluidics [27, 28] have been modified for these devices. A typical fabrication process is illustrated in Figure 15. The master mold, shown in Figure 16a, is fabricated by a two-step SU-8 process on a silicon substrate. The first step consists of forming the buffer channel, sensing channel, drain channel, and the associated microfluidic intercon‐ nects to a height of 15 µm. The second step consists of forming the sensing chamber, out-flow chambers and injection port to have a height of 200 µm. The thin film ITO electrodes are patterned with a hydrochloric acid wet etch process and traditional photolithography. Next, the PDMS elastomer is fabricated with a 10:1 (base: agent) mixture to create a thick replica mold of 1 mm. This replica mold is subsequently aligned to the ITO electrode pattern and bonded onto the glass substrate through oxygen plasma pretreatment as shown in Figure 16c.

A BD 30½ G needle is inserted into the injection port of the elastomer housing and a controlled volume of fluid is infused into the microfluidic network from a glass syringe using a syringe pump at a calibrated flow rate. Due to the small diameter of the gauge needle, the puncture

m

228 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

the mechanical amplifier [26].

**Figure 15.** Microfabrication process.

**4.2. Device fabrication**

*<sup>D</sup> <sup>A</sup> D* æ ö <sup>=</sup> ç ÷ è ø

**Figure 16.** (a) 3-D replica mold. (b and c) Device. (d) Schematic of mechanical amplifier. (e) Mechanical amplifier inte‐ grated with microfluidic, capacitive pressure sensor.

hole is self-sealed after withdrawal due to the elastomeric properties of the PDMS. For illustrative purposes, dyed glycerol is injected as shown in Figure 16b. After the injection of RTIL, the patterned microchannels become invisible due to the close refractive index between BMImBF4 (1.42) and the PDMS housing (1.4). The mechanical concentrator is constructed out of polystyrene, for its combination of mechanical rigidity (~3 GPa), optical transparency, lowcost and micromachinability with a programmable, CO2 laser (universal laser systems), whose schematic is illustrated in Figure 16d. Furthermore, polystyrene and PDMS can form covalent bonding through oxygen plasma treatment leading to simple integration. This is a result of the plasma, creating hydrogen bondings of silanol groups with C-OH and COOH moieties on the oxidized-rich polystyrene surface [29]. Additionally, the PDMS molds can be transferred to a suite of other plastic substrates through simple plasma assisted bonding to form an array of flexible pressure sensors [30]. The finished device is illustrated in Figure 16e.

#### **4.3. Characterization setup**

The test setup to evaluate the sensor sensitivity is shown in Figure 17. It consists of a force gauge and a step motor mounted onto an optical table. As controlled normal pressure is applied to the sensing chamber, the fluidic displacement within the sensing channel is monitored with an optical microscope. The electrical impedance spectroscopy is monitored with a precision LCR meter and Labview software. The frequency dependent double layer capacitance response is plotted in Figure 18 where increments of 1 kPa are applied to the pressure sensor. The response double layer capacitance at the ITO/BMImBF4 interface has a peak capacitance of approximately 25 µF/cm2 at 30 Hz. This high capacitance per unit area indicates a successful surface engineering to roughen the electrode surface.

**Figure 17.** Measurement setup for sensitivity.

**Figure 18.** Frequency-dependent double layer capacitance versus frequency for applied pressure in increments of 1 kPa.

Images of the optically transparent, flexible pressure sensor array are shown in Figure 19 with device dimensions of *w* = 100 µm, *g* = 20 µm, *h* = 15 µm, *r* = 500 µm, *H* = 150 µm, *t*PDMS = 500 µm, and *t*TCP = 1.5 µm. A meandered fluidic sensing channel geometry with interdigitated sensing electrodes was chosen for a compact device with a larger dynamic range over the straight channel sensing channel topology. Linear capacitive pressure analog response is measured over the dynamic range of approximately 96 kPa. Beyond the dynamic range, the capacitive response saturates. The measured capacitance versus pressure sensitivity is

**4.3. Characterization setup**

peak capacitance of approximately 25 µF/cm2

**Figure 17.** Measurement setup for sensitivity.

kPa.

The test setup to evaluate the sensor sensitivity is shown in Figure 17. It consists of a force gauge and a step motor mounted onto an optical table. As controlled normal pressure is applied to the sensing chamber, the fluidic displacement within the sensing channel is monitored with an optical microscope. The electrical impedance spectroscopy is monitored with a precision LCR meter and Labview software. The frequency dependent double layer capacitance response is plotted in Figure 18 where increments of 1 kPa are applied to the pressure sensor. The response double layer capacitance at the ITO/BMImBF4 interface has a

**Figure 18.** Frequency-dependent double layer capacitance versus frequency for applied pressure in increments of 1

Images of the optically transparent, flexible pressure sensor array are shown in Figure 19 with device dimensions of *w* = 100 µm, *g* = 20 µm, *h* = 15 µm, *r* = 500 µm, *H* = 150 µm, *t*PDMS = 500 µm, and *t*TCP = 1.5 µm. A meandered fluidic sensing channel geometry with interdigitated sensing electrodes was chosen for a compact device with a larger dynamic range over the straight channel sensing channel topology. Linear capacitive pressure analog response is measured over the dynamic range of approximately 96 kPa. Beyond the dynamic range, the capacitive response saturates. The measured capacitance versus pressure sensitivity is

indicates a successful surface engineering to roughen the electrode surface.

230 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

at 30 Hz. This high capacitance per unit area

**Figure 19.** Measured capacitance versus applied pressure at 30 Hz of single pressure sensor within array (insets show images of fluidic displacement as a result of applied pressure with measured corresponding capacitance).

approximately 0.23 nF/kPa. The experimental sensitivity agrees well with the theoretically predicted sensitivity equation through carefully characterization of the interfacial capacitance and effective Young's modulus of the sensing chamber structure.

To further improve the sensitivity of the device, mechanical amplification is investigated. With themechanicalamplification(*A*m),thenewsensitivity(*S*′)increasesbythefollowingexpression:

$$\mathcal{S}' = A\_{\rm m} \mathcal{S} \tag{27}$$

where *A*m is set by *D*out and *D*in of the mechanical amplifier as previous described in the electromechanical model. The mechanical concentrator is constructed out of polystyrene, for its combination of mechanical rigidity (~3 GPa), optical transparency, low-cost and microma‐ chinability with a programmable CO2 laser.

**Figure 20.** Measured capacitance versus pressure response at 30 Hz underlying the empirical computation model.

The measured capacitance versus pressure response are overlaid with theoretical sensitivity, at a fixed frequency of 30 Hz, is shown in Figure 20. The dynamic range of the device is set by the length of the sensing channel. Beyond the dynamic range, the capacitive response saturates, exhibited thorough measurement results. With the mechanical concentrators, a mechanical gain of approximately 19.5 is measured for *D*out = 5 mm and *D*in = 1 mm. A gain of 7.91 is measured when *D*out = 3 mm. The disparity between the theoretical and measured mechanical concentrator amplification is believed to lie in the inability for the forces to sum across the surface of the applied pressure and focus onto the sensing chamber.

#### **4.4. Time-resolved measurements**

To shed light on the relaxation time constant of the pressure sensor, time-resolved measure‐ ments are conducted. A schematic of the test setup is shown in Figure 21. A 30 Hz sine wave with 500 mVpk-pk is applied from a signal generator applied to one electrode and a ceramic capacitor with a value of 7.4 µF is connected to ground and parallel with the oscilloscope. The internal envelope detector function in the oscilloscope is used to smooth AC ripples. The data is saved on the oscilloscope and processed in Matlab®. The interface circuit does not amplify or compensate for the nonlinear characteristics of the sensor. The measured capacitance value follows the applied input pressure well indicating repeatability and negligible hysteresis.

**Figure 21.** Time-resolved test setup.

A computer-controlled step motor and a force sensor are used to apply an external pressure of 1000 Pa and 500 Pa, respectively, at 1 Hz and 2 Hz, while measuring the electrical response. The fluid is displaced in the middle-length of the sensing channel with an offset pressure of approximately 2 kPa. The measured electrical response of the mechanical input is plotted in Figure 22, indicating repeatability and negligible hysteresis of the sensor. There is an observ‐ able phase lag (*τ*) between the applied input pressure and the measured voltage response. This is believed to be of three components: (1) elastomeric PDMS recovery response (*τ*PDMS), (2) propagating velocity of the fluid (*τ*fl), and (3) the steady-state charge buildup of the interfacial capacitance (*τ*RC):

$$
\tau = \tau\_{\text{PDMS}} + \tau\_{\text{fl}} + \tau\_{\text{RC}} \tag{28}
$$

The measured capacitance versus pressure response are overlaid with theoretical sensitivity, at a fixed frequency of 30 Hz, is shown in Figure 20. The dynamic range of the device is set by the length of the sensing channel. Beyond the dynamic range, the capacitive response saturates, exhibited thorough measurement results. With the mechanical concentrators, a mechanical gain of approximately 19.5 is measured for *D*out = 5 mm and *D*in = 1 mm. A gain of 7.91 is measured when *D*out = 3 mm. The disparity between the theoretical and measured mechanical concentrator amplification is believed to lie in the inability for the forces to sum across the

To shed light on the relaxation time constant of the pressure sensor, time-resolved measure‐ ments are conducted. A schematic of the test setup is shown in Figure 21. A 30 Hz sine wave with 500 mVpk-pk is applied from a signal generator applied to one electrode and a ceramic capacitor with a value of 7.4 µF is connected to ground and parallel with the oscilloscope. The internal envelope detector function in the oscilloscope is used to smooth AC ripples. The data is saved on the oscilloscope and processed in Matlab®. The interface circuit does not amplify or compensate for the nonlinear characteristics of the sensor. The measured capacitance value follows the applied input pressure well indicating repeatability and negligible hysteresis.

A computer-controlled step motor and a force sensor are used to apply an external pressure of 1000 Pa and 500 Pa, respectively, at 1 Hz and 2 Hz, while measuring the electrical response. The fluid is displaced in the middle-length of the sensing channel with an offset pressure of approximately 2 kPa. The measured electrical response of the mechanical input is plotted in Figure 22, indicating repeatability and negligible hysteresis of the sensor. There is an observ‐ able phase lag (*τ*) between the applied input pressure and the measured voltage response. This is believed to be of three components: (1) elastomeric PDMS recovery response (*τ*PDMS), (2) propagating velocity of the fluid (*τ*fl), and (3) the steady-state charge buildup of the interfacial

PDMS fl RC

= ++ (28)

 t t

tt

surface of the applied pressure and focus onto the sensing chamber.

232 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

**4.4. Time-resolved measurements**

**Figure 21.** Time-resolved test setup.

capacitance (*τ*RC):

**Figure 22.** Time-resolved measurement results with applied input pressure on the left *y*-axis and corresponding output voltage on the right *y*-axis.

The frequency is not limited by the elastomer response, since PDMS has a frequency response in excess of 200 Hz [31]. The *τ*RC is set by the electrical properties of the device that is on the order of µs. This also is not the limiting factor where *R* ranges from 30 to 100 Ω and *C* ranges from 0 to 27 nF. This arrives to the conclusion that the dominating factor that limits the frequency performance of the pressure sensor is the fluidic response set by the viscosity. To analyze the fluidic response, the flow is assumed to be dominated by a pressure driven flow in the laminar region. Thus, the lubrication theory approximation can be used for the Navier-Stokes equation. The resulting cutoff frequency, *f*c, is then approximately set by the fluidic resistance and capacitance from the following equation:

$$f\_c = \frac{1}{2\pi\tau\_{\text{fl}}} = \frac{1}{2\pi RC} = \frac{1}{2\pi \left(\frac{12\,\mu\text{M}}{\text{w}\hbar^3}\right) \left(\frac{\pi r^2}{E}\Delta H\right)} = \frac{E^3 w^2 h^4}{24\,\mu\pi^3 r^4 H^2 \Delta P^2} \tag{29}$$

where *E* is the effective Young's modulus of the sensing chamber *w* and *h* are the width and height of the sensing channel, µ is the viscosity of the fluid (approximately 233 cP at room temperature[32]), *r* and *H* are the radius and height of the sensing chamber and Δ*P* is the pressure change over a period. A response and relaxation time is observed to be dependent on the fluidic RC time constant. This shows that the response time of the sensor can easily be tuned by adjusting the width and height of the sensing channel or the geometries of the sensing chamber. This also shows that the effective Young's modulus of the fluidic encasing will improve the frequency response, which can be tuned by adjusting different ratios of PDMS [33]. Furthermore, the magnitude of the input pressure and velocity of mechanical input also affects frequency response. The overshoot in the measured pressure indicates the viscoelastic nature of the elastomer. The underdamped measured output voltage suggests an RC time constant response corresponding with the large interfacial capacitance.

#### **4.5. Summary of microfluidic sensor for smart skin applications**

The development of ultra-high sensitivity, capacitive pressure sensors using ionic liquids is presented. These ultra-high sensitivities are achieved through three levels of amplification: (i) fluidic displacement amplification through the geometric volume difference between the large sensing chamber and small sensing channel; (ii) ultra-high, capacitance formed at the interface between the electrode-liquid surface; and (iii) mechanical concentration of the pressure onto the sensing chamber through the construction of a rigid construct relatively to the elastomeric housing. The measured results demonstrate a 2000× improvement sensitivity over traditional capacitive pressure sensors. Repeatability and hysteresis are investigated through timeresolved measurements demonstrating excellent performance. In addition to ultra-high sensitivity, the pressure sensor is constructed out of optically transparent material; here it displays a linear response and has a low-cost, simple fabrication process. Without the me‐ chanical amplifier, this sensor can be readily integrated with other lab-on-chip components constructed out of PDMS. With the addition of the mechanical amplifier, such sensors have potential applications in ultra-high sensitivity tactile sensing.

### **Author details**

John Yan\*

Address all correspondence to: john.yan@mail.npu.edu

Northwestern Polytechnic University, Fremont, CA, USA

### **References**


[8] J. J. Yan, "An Unpowered, Wireless Contact Lens Pressure Sensor for Point-Of-Care Glaucoma Diagnosis," Proceedings of 33rd IEEE EMB Conference, pp. 2522–2525, August 30th‒September 3rd 2011, Boston, MA.

fluidic displacement amplification through the geometric volume difference between the large sensing chamber and small sensing channel; (ii) ultra-high, capacitance formed at the interface between the electrode-liquid surface; and (iii) mechanical concentration of the pressure onto the sensing chamber through the construction of a rigid construct relatively to the elastomeric housing. The measured results demonstrate a 2000× improvement sensitivity over traditional capacitive pressure sensors. Repeatability and hysteresis are investigated through timeresolved measurements demonstrating excellent performance. In addition to ultra-high sensitivity, the pressure sensor is constructed out of optically transparent material; here it displays a linear response and has a low-cost, simple fabrication process. Without the me‐ chanical amplifier, this sensor can be readily integrated with other lab-on-chip components constructed out of PDMS. With the addition of the mechanical amplifier, such sensors have

[1] J. Stankovic, "Research Directions for the Internet of Things," IEEE Internet Things J.,

[2] C. Perera, A. Zaslavsky, C. Liu, M. Compton, P. Christen, Georgakopoulos, D., "Sensor search techniques for sensing as a service architecture for the Internet of Things," IEEE

[3] K. Lyytinen, Y. J. Yoo, "Issues and Challenges in Ubiquitous Computing," Communica‐

[4] G. M. Whitesides, "The Origins and the Future of Microfluidics," Nature 442, pp. 368–

[5] N. T. Nguyen, S. T. Werely, Fundamentals and Applications of Microfluidics. Nor‐

[6] T.Thorsen,S.J.Maerkl,S.R.Quake, "MicrofluidicLarge-scale Integration,"Science,vol.

[7] J. J. Yan, et al., "Vascular Impedance Assessment through Coupled Optomechanical Pressure and Flow Measurements," 11th Annual UC Systemwide Bioengineering

potential applications in ultra-high sensitivity tactile sensing.

234 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

Address all correspondence to: john.yan@mail.npu.edu

vol. 1, no. 1, pp. 3‒9, March 2014.

wood, MA: Artech House, 2002.

298, no. 5593, pp. 580‒584, October 2002.

Symposium, June 2010, Davis, CA.

373, July 2006.

Northwestern Polytechnic University, Fremont, CA, USA

Sensors J., vol. 14, no. 2, pp. 406‒420, April 2014.

tions of the ACM, vol. 45, no. 12, 62‒65, 2002.

**Author details**

John Yan\*

**References**


## **Chapter 10**

## **Electroosmotic Flow Pump**

Meng Gao and Lin Gui

[23] T. Torimoto, et al., "New Frontiers in Materials Science Opened by Ionic Liquids,"

[24] A. J. Bard, L. R. Faulkner, Electrochemical Methods: Fundamentals and Applications,

[25] H. A. Wheeler, "Transmission-Line Properties of Parallel Strips Separated by a Die‐ lectric Sheet," IEEE Trans. Microwave Theory Tech., vol. MTT-13, no. 2, 178‒185,

[26] Y.-M. L. I-Kuan Lin, Y. Liu, K.-S. Chen, X. Zhang, "Elastic and Viscoelastic Character‐ ization of Polydimethylsiloxane (PDMS) for Cell-Mechanics Applications," Mater.

[27] S. R. Quake, A. Scherer, "From Micro- to Nanofabrication with Soft Materials," Sci‐

[28] J. C. McDonald, D. C. Duffy, J. R. Anderson, D. T. Chiu, H. K. Wu, O. J. A. Schueller, G. M. Whitesides, "Fabrication of Microfluidic Systems in Poly(dimethylsiloxane),"

[29] A. Bubendorfer, et al., "Microfabrication of PDMS Microchannels using SU-8/PMMA Moldings and Their Sealing to Polystyrene Substrates," Smart Mater. Struct., vol. 16,

[30] J. J. Yan, "Optically Transparent, Flexible Pressure Sensor Array Micromachined Uti‐ lizing Plasma Assisted Bonding," Proc. SPIE MOEMS-MEMS, vol. 8248, pp.

[31] M. A. Unger, et al., "Monolithic Microfabricated Valves and Pumps by Multilayer

[32] P. G. Heiden, "Ionic Liquids in Chemical Analysis," Choice: Curr. Rev. Acad. Librar.,

[33] K. Khanafer, et al., "Effects of Strain Rate, Mixing Ratio, and Stress-Strain Definition on the Mechanical Behavior of the Polydimethylsiloxane (PDMS) Material as Related to its Biological Applications," Biomed. Microdevices, vol. 11, pp. 503‒508, April

Adv. Mater., vol. 22, pp. 1196‒1221, 19 March 2010.

236 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

2nd ed. New York: Wiley, 2001.

Res. Soc. Symp. Proc., vol. 1052, 2008.

ence, vol. 290, pp. 1536–1540, 2000.

pp. 367‒371, April 2007.

vol. 47, pp. 712‒712, 2009.

2009.

Electrophoresis, vol. 21, no. 1, pp. 27‒40, 2000.

824803-1–824803-8, 21‒26 January 2012, San Francisco, CA.

Soft Lithography," Science, vol. 288, pp. 113‒116, 7 April 2000.

1965.

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/64601

#### **Abstract**

Electroosmotic flow (EOF) pumping has been widely used to manipulate fluids such as liquid sample reagents in microfluidic systems. In this chapter, we will introduce the re‐ search progress on EOF pumps in the fields of microfluidic science and technology and briefly present their microfluidic applications in recent years. The chapter focuses on pump channel materials, electrodes, and their fabrication techniques in microfluidics.

**Keywords:** Electroosmotic flow pump, Channel material, Pump electrode, Fabrication, Microfluidic applications

## **1. Introduction**

Micropumps are the essential active components of fluid transport systems in microfluidics. They can manipulate small volumetric fluids on spatial scales, from several to a hundred microns [1-3]. Nowadays, they have been widely used in many scientific and technical fields of microfluidics, such as biological/chemical analysis and assays [4-7], liquid drug reagent injection/delivery [8-9], and microelectronic chip cooling [10].

With the rapid development of microfluidic technologies, great attention has been paid recently to miniature micropumps with compact design for microfluidic analysis and assays. Miniature micropumps can be easily integrated into microfluidic systems and enable users to achieve low-cost portable pumping devices such as disposable insulin infusion pumps. Miniaturization of pumping systems can simplify the operation of sample introduction and transport in the microfluidic platform with less manual intervention. Meanwhile, miniaturi‐ zation can greatly reduce the quantities of sample reagents and achieve microfluidic analysis or assays efficiently.

Recently, electroosmotic flow (EOF) pumps [11-12] have received extensive attention because of their ability to drive a wide range of liquid fluids and generate high pumping pressures or

© 2016 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

flow rates with continuous pulse-free flows. Besides, the EOF pumps can exhibit precise control of small volumetric fluids in microfluidic systems under finely controlled electric fields. Compared with mechanical micropumps, EOF pumps do not require moving mechanical parts inside, which can be easily integrated into pumping platforms to achieve the miniaturization of microfluidic systems. Notably, by changing the strength and direction of the electric field through the pump channels, the EOF pumps can conveniently offer bidirectional fluid flows for microfluidic systems.

The pumping flows in EOF pumps are driven by the mechanism of electroosmotic flow phenomenon [13-15]. When in contact with an uncharged liquid fluid (e.g., deionized water, aqueous solution), channel wall surfaces of PDMS, glass, PMMA, or Si can carry electrostatic charges, forming an electrical double layer nearby. The electrical double layer has a compact layer (containing immobile ions) on the channel surface and a diffuse layer (containing mobile ions) in the liquid fluid. Once an electric field is applied through the pump channel, the mobile ions in the diffuse layer move under the electric field force. As a result of the viscous effect, the moving ions will drag their surrounding fluid molecules to the same speed, forming electroosmotic flow in the pump channel. In short, the electroosmotic pumping performance of EOF pumps is fundamentally dependent on the material property of the pump channel wall and electrical field applied to the pump electrodes. The fabrication of pump channels and electrode plays a vital role in cost control of the EOF pumps. The material and fabrication of the pump channel and electrode are also important considerations in the selection of EOF pumps for microfluidic applications.

Recently, the scientific and technical research of EOF pumping in microfluidics has often focused on the pump channel material, pump electrode, and their fabrication techniques. In this chapter, we will mainly present the research progress of EOF pumps in these aspects and briefly introduce new and recent applications of EOF pumps in microfluidics.

## **2. Channel material and fabrication**

Generally, EOF pumps can roughly be divided into direct EOF pumps, porous membrane EOF pumps, and packed porous media EOF pumps, according to the type of EOF-generating pump channels.

#### **2.1. Direct electroosmotic flow (EOF) pump**

Direct electroosmotic flow (EOF) pumps utilize open pump channels to drive fluids inside. The pumping pressure or flow rate can be increased via enlarging the number of pump channels. Direct EOF pumps are extremely suitable for the introduction delivery of sample reagents containing cells, biomolecules, and larger particles.

Basically, the common direct EOF pumps are fabricated by capillaries (e.g., PMMA, fused silica capillaries), which are named open-capillary EOF pumps [16-20]. The open-capillary EOF pumps, compared with others, are simple, cheap, and easy to fabricate, because the capillaries are popular on the market. However, the capillary cannot offer high pumping pressure or flow rate for microfluidic systems. In the open-capillary EOF pump, inert solid-metal-based thin wires are often used to fabricate the pump electrodes. Normally, the outer diameter of thin wires is much larger than the inner diameter of the capillary channel. To generate electric field through the capillary channel, the thin wire electrodes have to be inserted and fastened into two fluid reservoirs connected with both ends of the capillary. The open-capillary EOF pumps are widely used as sample introduction devices to drive liquid reagents into microfluidic chip platforms.

The direct EOF pumps can also be constructed by open channels. These pumps can be described as direct open-channel EOF pumps [20-24], which are usually used to perform onchip integratable control of sample reagents in microfluidic chip systems. The open channels in these pumps are usually fabricated with photolithographic microfabrication technologies. Figure 1 shows a widely used direct EOF pump using a PDMS microchannel as an open pump channel. Two inert solid-metal wires (platinum or gold) are inserted into both inlet and outlet fluid reservoirs of the PDMS microchannel as the pump electrodes. The PDMS microchannel in this pump can be fabricated with the standard soft lithography technology, which will facilitate the integration of this EOF pump microfluidic systems. To obtain high pumping pressure or flow-rate fluid flows, the direct open-channel EOF pumps can be usually designed and fabricated with a large number of open pump channels in parallel.

**Figure 1.** Open-channel EOF pump with metal wire electrodes.

flow rates with continuous pulse-free flows. Besides, the EOF pumps can exhibit precise control of small volumetric fluids in microfluidic systems under finely controlled electric fields. Compared with mechanical micropumps, EOF pumps do not require moving mechanical parts inside, which can be easily integrated into pumping platforms to achieve the miniaturization of microfluidic systems. Notably, by changing the strength and direction of the electric field through the pump channels, the EOF pumps can conveniently offer bidirectional fluid flows

238 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

The pumping flows in EOF pumps are driven by the mechanism of electroosmotic flow phenomenon [13-15]. When in contact with an uncharged liquid fluid (e.g., deionized water, aqueous solution), channel wall surfaces of PDMS, glass, PMMA, or Si can carry electrostatic charges, forming an electrical double layer nearby. The electrical double layer has a compact layer (containing immobile ions) on the channel surface and a diffuse layer (containing mobile ions) in the liquid fluid. Once an electric field is applied through the pump channel, the mobile ions in the diffuse layer move under the electric field force. As a result of the viscous effect, the moving ions will drag their surrounding fluid molecules to the same speed, forming electroosmotic flow in the pump channel. In short, the electroosmotic pumping performance of EOF pumps is fundamentally dependent on the material property of the pump channel wall and electrical field applied to the pump electrodes. The fabrication of pump channels and electrode plays a vital role in cost control of the EOF pumps. The material and fabrication of the pump channel and electrode are also important considerations in the selection of EOF

Recently, the scientific and technical research of EOF pumping in microfluidics has often focused on the pump channel material, pump electrode, and their fabrication techniques. In this chapter, we will mainly present the research progress of EOF pumps in these aspects and

Generally, EOF pumps can roughly be divided into direct EOF pumps, porous membrane EOF pumps, and packed porous media EOF pumps, according to the type of EOF-generating pump

Direct electroosmotic flow (EOF) pumps utilize open pump channels to drive fluids inside. The pumping pressure or flow rate can be increased via enlarging the number of pump channels. Direct EOF pumps are extremely suitable for the introduction delivery of sample

Basically, the common direct EOF pumps are fabricated by capillaries (e.g., PMMA, fused silica capillaries), which are named open-capillary EOF pumps [16-20]. The open-capillary EOF pumps, compared with others, are simple, cheap, and easy to fabricate, because the capillaries

briefly introduce new and recent applications of EOF pumps in microfluidics.

for microfluidic systems.

pumps for microfluidic applications.

**2. Channel material and fabrication**

**2.1. Direct electroosmotic flow (EOF) pump**

reagents containing cells, biomolecules, and larger particles.

channels.

#### **2.2. Porous membrane electroosmotic flow (EOF) pump**

Porous membrane EOF pumps [25-28] utilize a piece of porous membrane to construct submicroscale or nanoscale pump channels within. They are miniaturized and highly integrated microfluidic pumping devices. Compared with the direct EOF pumps, the porous membrane EOF pumps under action of a large number of micro-/nanopump channels can offer high pumping pressure or flow-rate flows. The porous membranes are frequently made of glass, silica, alumina, or organic polymer (PC or PET) using the high-temperature sintering technique or etching technologies like chemical track etching, physical etching, and soft lithography. The drawback of the porous membrane EOF pump is that the sub-micro- or nanoscale pump channels in the pump cannot be used to transport cell, biochemical macromolecules, or large particle in aqueous suspensions.

Figure 2 shows a popular porous membrane EOF pump with mesh microelectrodes. In this EOF pump, the porous membrane is located between the inlet and outlet fluid reservoirs and vertically fastened to the macrofluid channel wall by both supporting frames. Two pieces of mesh microelectrodes are attached onto both sides of the membrane to reduce voltage drop and generate a high electric field through the pump channels. The pump channels embedded in the porous membrane are relatively short (from tens to several hundreds of µm). Hence, an electric field with high strength can be obtained when a low voltage is applied. In order to reduce fluid flow resistance, the micro-/nanopump channels embedded in membrane are often designed and fabricated straight from one side of the porous membrane to the other.

**Figure 2.** Porous membrane EOF pump with mesh microelectrodes.

#### **2.3. Packed porous media electroosmotic flow (EOF) pump**

Packed porous media EOF pumps [29-32], highly miniaturized and integrated microfluidic pumps, can drive high-pressure or flow-rate fluids. Similar to the porous membrane EOF pumps, the packed porous media EOF pumps have a large number of sub-micro- or nanopump channels inside. The sub-micro-/nanopump channels are usually prepared by packing submicro-/nanodielectric particles or columns into a mini-/microfluidic channel. These dielectric particles or columns can be made of fused silica, alumina, or organic polymer.

Figure 3 presents a typical example of packed porous media EOF pump with metal wire electrodes. In this EOF pump, a short mini-/microfluidic channel is used to build the pumping region with two pieces of porous membranes on both sides. The packed particles are held in place inside the fluid channel. Two metal wire electrodes are separately inserted into both inlet and outlet fluid reservoirs, paralleling to the fluid channel. Because the particles are randomly distributed inside the fluid channel, the fluid flow resistance in the pump will rise with pumping, thus leading to the reduction of pumping pressure or flow rate. To produce high flow rates or high pumping pressures, the EOF pump can be designed with a large number of parallel sub-micro-/nanopump channels. Alternatively, sub-micro-/nanodielectric columns can be introduced and packed into this pump (shown in Figure 3) to construct parallel submicro-/nanopump channels. For convenient fabrication purpose, the packed columns should be short in length compared with the fluid channel.

**Figure 3.** Packed porous media EOF pump with metal wire electrodes.

## **3. Electrode and fabrication**

Electrodes, the key components of EOF pumps, can be used to induce the driving electric field through the pump channels with applied voltage. In EOF pumps, the material and fabrication of the electrodes are vital factors in pump performance and cost control. Basically, there are two electrode types. One is contact electrode exposed to the fluid and the other is noncontact electrode separated from the fluid. This section will show detailed description of them.

#### **3.1. Contact electrode**

Figure 2 shows a popular porous membrane EOF pump with mesh microelectrodes. In this EOF pump, the porous membrane is located between the inlet and outlet fluid reservoirs and vertically fastened to the macrofluid channel wall by both supporting frames. Two pieces of mesh microelectrodes are attached onto both sides of the membrane to reduce voltage drop and generate a high electric field through the pump channels. The pump channels embedded in the porous membrane are relatively short (from tens to several hundreds of µm). Hence, an electric field with high strength can be obtained when a low voltage is applied. In order to reduce fluid flow resistance, the micro-/nanopump channels embedded in membrane are often

240 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

designed and fabricated straight from one side of the porous membrane to the other.

Packed porous media EOF pumps [29-32], highly miniaturized and integrated microfluidic pumps, can drive high-pressure or flow-rate fluids. Similar to the porous membrane EOF pumps, the packed porous media EOF pumps have a large number of sub-micro- or nanopump channels inside. The sub-micro-/nanopump channels are usually prepared by packing submicro-/nanodielectric particles or columns into a mini-/microfluidic channel. These dielectric

Figure 3 presents a typical example of packed porous media EOF pump with metal wire electrodes. In this EOF pump, a short mini-/microfluidic channel is used to build the pumping region with two pieces of porous membranes on both sides. The packed particles are held in place inside the fluid channel. Two metal wire electrodes are separately inserted into both inlet and outlet fluid reservoirs, paralleling to the fluid channel. Because the particles are randomly distributed inside the fluid channel, the fluid flow resistance in the pump will rise with pumping, thus leading to the reduction of pumping pressure or flow rate. To produce high flow rates or high pumping pressures, the EOF pump can be designed with a large number of

particles or columns can be made of fused silica, alumina, or organic polymer.

**Figure 2.** Porous membrane EOF pump with mesh microelectrodes.

**2.3. Packed porous media electroosmotic flow (EOF) pump**

Contact electrodes mainly made of solid metals are the most widely used electrodes in EOF pumps. The solid-metal-based contact electrodes are often divided into three groups, which are metal wire electrodes, membranous microelectrodes, and mesh microelectrodes.

Metal wire electrodes [33-36] are inserted into the inlet/outlet reservoirs of the pump channels in EOF pumps, as shown in Figure 13. These metal wire electrodes are the simplest type for EOF pumps, which can be bought easily. However, they are not suitable for the integration or miniaturization of the pumping devices in microfluidic systems. Due to the smaller size of metal wires as shown in Figure 3, the metal wire electrodes are not capable of generating a roughly uniform electric field throughout the whole pump channels with applied voltage. Therefore, the pump cannot offer steady flows with a uniform velocity field inside the channel.

Membranous microelectrodes [37-40] are often fabricated under the pump channel using sputtering or deposition techniques, as shown in Figure 4. They can be well miniaturized and

**Figure 4.** Membranous microelectrode used in open-channel EOF pump.

integrated into the on-chip pumping system. However, it is important to note that the membranous microelectrodes have to be fabricated separately with the pump microchannels, and they should be accurately aligned with the pump microchannels during bonding. The fabrication of these membranous microelectrodes is complex, expensive, and time-consuming.

Figure 5 presents the fabrication of the membranous microelectrodes together with the pump channels for the open-channel EOF pumps. The membranous microelectrodes are fabricated onto a glass substrate through techniques of sputtering and standard soft lithography (Figure 5 (a)), and the PDMS pump microchannel can be prepared by soft lithography technique (Figure 5 (b)). After fabrication, the PDMS pump microchannel is irreversibly bonded with the glass substrate. Since the membranous microelectrodes are located under the pump channel, the EOF pump cannot obtain a parallel electric field through the pump channel or drive uniform pumping flows. To achieve an almost uniformly distributed flow, the pump micro‐ channel needs to be designed with a relatively low high-aspect-ratio section.

Similarly, mesh microelectrodes [41-43] are miniature and integratable ones for EOF pumps, as shown in Figure 2. They are very suitable for the porous media EOF pumps, which will strengthen the miniaturization and integration of the EOF pumps into microfluidic systems. In EOF pumps, the mesh microelectrodes are usually placed and fastened on both ends of the porous pump channels. During assembly, meshes of each electrode have to be aligned with the sub-micro-/nanopump channels. Different with the membranous microelectrodes, the mesh microelectrodes can induce a roughly uniform electric field in the pump channels. They can easily offer high flows at relatively low voltages. However, they do have the same characteristics that are extremely complex and expensive in fabrication.

The contact electrodes exposed to the fluid usually give rise to a serious problem of electrolysis during pumping. Bubbles or other electrolytic products can occur at the electrode surfaces, entering the pump channels and blocking the EOFs. What's more, the joule heat will be generated in the fluid. All bring a sharp decrease in electroosmotic mobility and flow rate. Even worse, the short circuit of high-voltage supply equipment happens sometimes. The use of inert solid-metal platinum (Pt) and gold (Au) electrodes can largely reduce the electrolysis in EOF pumps. The abovementioned problems can be eliminated if the solid-metal electrodes are separated from the aqueous reagents in the pump channels.

**Figure 5.** Fabrication of open-channel EOF pump with membranous microelectrodes and PDMS-based pumping mi‐ crochannel.

#### **3.2. Noncontact electrode**

integrated into the on-chip pumping system. However, it is important to note that the membranous microelectrodes have to be fabricated separately with the pump microchannels, and they should be accurately aligned with the pump microchannels during bonding. The fabrication of these membranous microelectrodes is complex, expensive, and time-consuming. Figure 5 presents the fabrication of the membranous microelectrodes together with the pump channels for the open-channel EOF pumps. The membranous microelectrodes are fabricated onto a glass substrate through techniques of sputtering and standard soft lithography (Figure 5 (a)), and the PDMS pump microchannel can be prepared by soft lithography technique (Figure 5 (b)). After fabrication, the PDMS pump microchannel is irreversibly bonded with the glass substrate. Since the membranous microelectrodes are located under the pump channel, the EOF pump cannot obtain a parallel electric field through the pump channel or drive uniform pumping flows. To achieve an almost uniformly distributed flow, the pump micro‐

Similarly, mesh microelectrodes [41-43] are miniature and integratable ones for EOF pumps, as shown in Figure 2. They are very suitable for the porous media EOF pumps, which will strengthen the miniaturization and integration of the EOF pumps into microfluidic systems. In EOF pumps, the mesh microelectrodes are usually placed and fastened on both ends of the porous pump channels. During assembly, meshes of each electrode have to be aligned with the sub-micro-/nanopump channels. Different with the membranous microelectrodes, the mesh microelectrodes can induce a roughly uniform electric field in the pump channels. They can easily offer high flows at relatively low voltages. However, they do have the same

The contact electrodes exposed to the fluid usually give rise to a serious problem of electrolysis during pumping. Bubbles or other electrolytic products can occur at the electrode surfaces, entering the pump channels and blocking the EOFs. What's more, the joule heat will be generated in the fluid. All bring a sharp decrease in electroosmotic mobility and flow rate. Even worse, the short circuit of high-voltage supply equipment happens sometimes. The use of inert solid-metal platinum (Pt) and gold (Au) electrodes can largely reduce the electrolysis in EOF pumps. The abovementioned problems can be eliminated if the solid-metal electrodes

channel needs to be designed with a relatively low high-aspect-ratio section.

**Figure 4.** Membranous microelectrode used in open-channel EOF pump.

242 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

characteristics that are extremely complex and expensive in fabrication.

are separated from the aqueous reagents in the pump channels.

Noncontact electrodes have been developed to prevent the above problems. These noncontact electrodes often utilize nonmetal materials (e.g., polymer gel, silica, polyaniline, PDMS) as membrane layers to separate the solid-metal electrodes from the EOF pumping fluid. The membrane layers are capable of allowing ion charges to pass through but stopping water molecules, and thus bubbles and by-products from electrolysis at the electrode surfaces can be prevented from entering the EOF pump channels in these micropumps.

Gel-type salt-bridge electrode, a widely used noncontact electrode in the field of electrochem‐ istry, has been successfully fabricated for EOF pumps with bubble-free formation [22, 33]. Figure 6 shows a typical bubble-free EOF pump with this gel-type salt-bridge electrode. As shown in Figure 6, this EOF pump employs a thin gel region to protect the solid-metal wire electrode from the fluid. The wire electrode is immersed in the electrode reservoir filled with a conductive aqueous solution. When a high voltage is applied to the pump, the electrolysis can still emerge inside the pump. But, bubbles can only be generated at the wire electrode

**Figure 6.** Gel-type salt-bridge microelectrode used in open-channel EOF pump.

surfaces in both electrode reservoirs, having no effect on the fluid flows in the pump channels. In this pump, both of the two gel regions are located in the electrode reservoirs between the metal wire electrodes and the parallel pump channels. The polymer gel, a sensitive polymer material, can be used for the microscale gel regions with the normal photolithography technique. During photolithographic fabrication, the mask for both two gel regions always has to be aligned costly and accurately. As a result, the fabrication of these gel-type salt-bridge electrodes requires a complex process. Another potential problem for this electrode is that the gel region can be easily collapsed due to the poor compatibility of gel material to the electrode reservoir wall. In the worst cases, electrolysis and bubbles will also be generated in the pump channels.

Other noncontact microelectrodes, such as fused silica capillary microelectrodes [31, 44], polyaniline-wrapped aminated graphene microelectrodes [45], and Ag/Ag2O microelectrodes [46], have been successfully fabricated to work as noncontact electrodes for bubble-free EOF pumps. In fabrication, three noncontact microelectrodes can be made from the chemical synthesis or assembly method in the laboratory. Compared with the gel-type salt-bridge microelectrodes, the three microelectrodes are robust in long-time running. However, the fabrication of this kind of electrodes is also a very complex, time-consuming, and expensive process. Now, the first challenge is to develop a new noncontact electrode with a simpler and cheaper fabrication technique.

Injecting wettable liquid metal into microchannels to make noncontact electrodes should be a simpler and cheaper method for bubble-free EOF pumps. Figure 7 shows a handy liquid-metal (GaInSn)-based EOF pump fabricated in a PDMS microfluidic chip [47]. In this pump, the liquid metal is a kind of metal alloy (GaInSn), which can be easily injected into microchannels by a simple syringe. The melting point of this liquid metal is only 10.6 °C below room tem‐ perature. As shown in Figure 7, two pairs of liquid-metal electrodes are fabricated parallel to each other and vertical to the pump channel in the same horizontal plane of the microfluidic chip. These two pairs of electrodes are also designed symmetrically to both sides of the pump channel. To induce high electric field strength in the pump channel when a relatively low

**Figure 7.** Handy liquid-metal microelectrode used in open-channel EOF pump.

voltage is applied, the electrodes are placed very close to but always not in contact with the pump channel. In this pump, the PDMS gaps are designed to be ≤ 40 µm between the liquidmetal electrode channels and the pump channel. For the convenience of liquid-metal injection, the liquid-metal electrode microchannels are all designed in the ohm shape.

Figure 8 depicts the typical fabrication of the liquid-metal noncontact electrodes for the openchannel EOF pump. Compared with membranous or mesh microelectrodes (shown in Figure 5), the liquid-metal electrode channels can be easily made just in one step together with the pump channel using the same fabrication technique. Furthermore, the liquid-metal electrodes can also be easily designed and fabricated in any shape and any location in the EOF pump. Using liquid-metal-filled microchannels as noncontact electrodes can provide an efficient approach to the miniaturization and integration of EOF pumps in microfluidic systems.

## **4. Applications**

surfaces in both electrode reservoirs, having no effect on the fluid flows in the pump channels. In this pump, both of the two gel regions are located in the electrode reservoirs between the metal wire electrodes and the parallel pump channels. The polymer gel, a sensitive polymer material, can be used for the microscale gel regions with the normal photolithography technique. During photolithographic fabrication, the mask for both two gel regions always has to be aligned costly and accurately. As a result, the fabrication of these gel-type salt-bridge electrodes requires a complex process. Another potential problem for this electrode is that the gel region can be easily collapsed due to the poor compatibility of gel material to the electrode reservoir wall. In the worst cases, electrolysis and bubbles will also be generated in the pump

**Figure 6.** Gel-type salt-bridge microelectrode used in open-channel EOF pump.

244 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

Other noncontact microelectrodes, such as fused silica capillary microelectrodes [31, 44], polyaniline-wrapped aminated graphene microelectrodes [45], and Ag/Ag2O microelectrodes [46], have been successfully fabricated to work as noncontact electrodes for bubble-free EOF pumps. In fabrication, three noncontact microelectrodes can be made from the chemical synthesis or assembly method in the laboratory. Compared with the gel-type salt-bridge microelectrodes, the three microelectrodes are robust in long-time running. However, the fabrication of this kind of electrodes is also a very complex, time-consuming, and expensive process. Now, the first challenge is to develop a new noncontact electrode with a simpler and

Injecting wettable liquid metal into microchannels to make noncontact electrodes should be a simpler and cheaper method for bubble-free EOF pumps. Figure 7 shows a handy liquid-metal (GaInSn)-based EOF pump fabricated in a PDMS microfluidic chip [47]. In this pump, the liquid metal is a kind of metal alloy (GaInSn), which can be easily injected into microchannels by a simple syringe. The melting point of this liquid metal is only 10.6 °C below room tem‐ perature. As shown in Figure 7, two pairs of liquid-metal electrodes are fabricated parallel to each other and vertical to the pump channel in the same horizontal plane of the microfluidic chip. These two pairs of electrodes are also designed symmetrically to both sides of the pump channel. To induce high electric field strength in the pump channel when a relatively low

channels.

cheaper fabrication technique.

EOF pumps can offer a simple and cost-effective way to generate adequate pumping pressures and flow rates for microfluidic systems. They have been widely and successfully used in many areas of microfluidics. In this part, we will briefly introduce applications of EOF pumps in microfluidics. Based on the category of application areas, this section will be divided into 1) microfluidic delivery and actuation and 2) microelectronic thermal management.

#### **4.1. Microfluidic delivery and actuation**

Due to the simplicity of pumping components, EOF pumps have been widely used in micro‐ fluidic delivery of pure liquids or aqueous solutions. As a micrototal analysis pumping tool, the EOF pump is commonly fabricated to be disposable devices with a compact design. The online reduction of sample reagent quantities should be desirable. For microinjection delivery,

**Figure 8.** Fabrication of handy liquid-metal microelectrode for open-channel EOF pump.

in particular, the EOF pump is usually required to offer high flows. Preferentially, opencapillary EOF pumps [16-20] are used to perform the introduction of sample reagents into a microfluidic analysis or assay system. The open capillary has the ability to deliver a wide range of sample reagents such as pure liquid drug reagents and aqueous solutions containing cells, particles, or biochemical macromolecules. The porous media EOF pumps [25-32] can also be used to drive pure liquid sample reagents for the injection purpose. For on-chip microfluidic delivery, open-channel EOF pumps [20-24] are the most popular pumping devices. The reason is that the open pump channels and the pump electrodes can be easily and conveniently integrated into the microfluidic chip together with other functional components.

In most microfluidic systems, efficient mixing of sample reagents is extremely essential for improving the throughput of microfluidic assays and analysis. Many active mixing methods [48] using external actuation forces to perturb the sample reagents to enhance their diffusion have been recently developed to achieve a high mixing performance. Most of the external actuation forces are generated by mechanical moving, stirring, or vibrating. Owing to the fast electrical operation, the EOF pump has also been widely used to perform highly efficient active mixing in microfluidics. The EOF-actuated mixing utilizes electroosmotic driving forces to induce oscillatory, turbulent, or chaotic flows in the sample reagents, while a periodic electric field is applied simultaneously [49]. Several typical open-channel configurations, such as Tshaped, Y-shaped, and multi-shaped configurations, have been developed for the EOFactuated mixing [50-56]. The use of EOF pump does not require mechanical moving components and hence brings cheaper and more reliable microfluidic mixers. Besides, the EOF pump can also be used as a microactuator for focusing and separation of droplets, particles, or cells in microfluidic systems [57-58].

### **4.2. Microelectronic thermal management**

With the rapid development of MEMS technologies, the design of a miniature electronic chip with more and more functional components has become an essential demand in recent years. Consequently, power consumption is increased to maintain operation which generates great heat flux. Air-forced cooling cannot remove such high heat flux from the hot chip. Micro‐ pumping that drives liquid fluids through microchannels is an efficient approach to perform heat dissipation of electronic components [10-12].

EOF pumps have been considered in microchannel-based liquid cooling for electronic chips owing to their low power consumption and high pumping pressure. EOF pumps can work without any noise during liquid coolant pumping. Recently, a porous media-based EOF pump [59] has been successfully utilized for liquid cooling of microelectronic chips. In this micro‐ channel liquid cooling system, the pump works as an external device to drive water coolant to force thermal dissipation of the hot region in the microelectronic chip. To reduce the thermal resistance, the microchannels filled with liquid coolant are tightly attached to the hot surface of electronic chip. Since the microchannels have high surface-volume ratio available for thermal dissipation, the EOF pump is capable of removing the heat generation efficiently.

## **5. Future and prospect**

in particular, the EOF pump is usually required to offer high flows. Preferentially, opencapillary EOF pumps [16-20] are used to perform the introduction of sample reagents into a microfluidic analysis or assay system. The open capillary has the ability to deliver a wide range of sample reagents such as pure liquid drug reagents and aqueous solutions containing cells, particles, or biochemical macromolecules. The porous media EOF pumps [25-32] can also be used to drive pure liquid sample reagents for the injection purpose. For on-chip microfluidic delivery, open-channel EOF pumps [20-24] are the most popular pumping devices. The reason is that the open pump channels and the pump electrodes can be easily and conveniently

In most microfluidic systems, efficient mixing of sample reagents is extremely essential for improving the throughput of microfluidic assays and analysis. Many active mixing methods [48] using external actuation forces to perturb the sample reagents to enhance their diffusion have been recently developed to achieve a high mixing performance. Most of the external actuation forces are generated by mechanical moving, stirring, or vibrating. Owing to the fast electrical operation, the EOF pump has also been widely used to perform highly efficient active

integrated into the microfluidic chip together with other functional components.

**Figure 8.** Fabrication of handy liquid-metal microelectrode for open-channel EOF pump.

246 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

This chapter has briefly reported recent research progress of EOF pumps with emphasis on channel materials, electrodes, and their fabrication and summarized pump applications in microfluidics.

EOF pumping is commonly used in many microfluidic devices. Nowadays, it has become an increasingly popular tool to manipulate such liquid sample reagents with electric fields. The number of microfluidic applications is growing fast, and certain EOF pumping devices, like porous membrane EOF pumps (Osmotex), have been already commercially available. However, the popular use of EOF pumps in microfluidics may be limited due to the lack of high-performance pumps with cost-effective characteristics.

The EOF pump continues to be improved, which shows stable performance, rapid operation, and compact design. EOF pumps with noncontact electrodes have generated robust pumping flows without bubble formation, but there is still a lot of work to be done for improving these pumps. Further research is required to understand the basic driving mechanism of gel, silica, or PDMS-based noncontact electrodes. Research on new material design and fabrication is also an urgent need for noncontact electrodes with low-cost and simple process. Another option of liquid metal, instead of solid metal and conductive aqueous solution in electrode reservoirs of noncontact electrodes, should be an alternative solution for the water electrolysis at the solidmetal electrode surfaces.

EOF pumps do not require mechanical moving components, and they can offer excellent miniaturization potential in integrated microfluidic applications. With applied electric fields, EOF pumps are capable of performing fine control of fluids fast. In the near future, the EOF pump should be an important element in promising implantable medical devices such as drug transport or infusion pumps.

## **Acknowledgements**

This work is financially supported by the National Natural Science Foundation of China (Grant No. 51276189).

## **Author details**

Meng Gao1,2\* and Lin Gui1,2

\*Address all correspondence to: mgao@mail.ipc.ac.cn

1 Key Laboratory of Cryogenics, Technical Institute of Physics and Chemistry, Beijing, China

2 Chinese Academy of Sciences, Haidian District, Beijing, China

## **References**


[4] Brian T. Good, Christopher N. Bowman and Robert H. Davis. An effervescent reac‐ tion micropump for portable microfluidic systems. Lab on a Chip. 2006;6(5):659-666. DOI: 10.1039/B601542E

pumps. Further research is required to understand the basic driving mechanism of gel, silica, or PDMS-based noncontact electrodes. Research on new material design and fabrication is also an urgent need for noncontact electrodes with low-cost and simple process. Another option of liquid metal, instead of solid metal and conductive aqueous solution in electrode reservoirs of noncontact electrodes, should be an alternative solution for the water electrolysis at the solid-

248 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

EOF pumps do not require mechanical moving components, and they can offer excellent miniaturization potential in integrated microfluidic applications. With applied electric fields, EOF pumps are capable of performing fine control of fluids fast. In the near future, the EOF pump should be an important element in promising implantable medical devices such as drug

This work is financially supported by the National Natural Science Foundation of China (Grant

1 Key Laboratory of Cryogenics, Technical Institute of Physics and Chemistry, Beijing,

[1] D. J. Laser and J. G. Santiago. A review of micropumps. Journal of Micromechanics and Microengineering. 2004;14(6):R35–R64. DOI: 10.1088/0960-1317/14/6/R01

[2] Peter Woias. Micropumps—past, progress and future prospects. Sensors and Actua‐

[3] Brian D. Iverson and Suresh V. Garimella. Recent advances in microscale pumping technologies: a review and evaluation. Microfluidics and Nanofluidics. 2008;5(2):

tors B: Chemical. 2005;105(1):28–38. DOI: 10.1016/j.snb.2004.02.033

metal electrode surfaces.

transport or infusion pumps.

**Acknowledgements**

No. 51276189).

**Author details**

China

**References**

Meng Gao1,2\* and Lin Gui1,2

\*Address all correspondence to: mgao@mail.ipc.ac.cn

145-174. DOI: 10.1007/s10404-008-0266-8

2 Chinese Academy of Sciences, Haidian District, Beijing, China


[29] Ye Ai, Sinan E. Yalcin, Diefeng Gu, Oktay Baysal, Helmut Baumgart, Shizhi Qian and Ali Beskok. A low-voltage nano-porous electroosmotic pump. Journal of Colloid and Interface Science. 2010;350(2):465-470. DOI: 10.1016/j.jcis.2010.07.024

[17] Shaorong Liu and Purnendu K. Dasgupta. A simple means to increase absorbance detection sensitivity in capillary zone electrophoresis. Analytica Chimica Acta.

[18] Shaorong Liu and Purnendu K. Dasgupta. Sequential injection analysis in capillary format with an electroosmotic pump. Talanta. 1994;41(11):1903-1910. DOI:

[19] Shaorong Liu and Purnendu K. Dasgupta. Electroosmotically pumped capillary for‐ mat sequential injection analysis with a membrane sampling interface for gaseous analytes. Analytica Chimica Acta. 1995;308(1-3):281-285. DOI:

[20] Shaorong Liu, Qiaosheng Pu and Joann J. Lu. Electric field-decoupled electroosmotic pump for microfluidic devices. Journal of Chromatography A. 2003;1013(1-2):57-64.

[21] Amir Jahanshahi, Fabrice Axisa and Jan Vanfleteren. Fabrication of a biocompatible flexible electroosmosis micropump. Microfluidics and Nanofluidics. 2012;12(5):

[22] Tomasz Glawdel and Carolyn L. Ren. Electro-osmotic flow control for living cell analysis in microfluidic PDMS chips. Mechanics Research Communications.

[23] Iulia M. Lazar and Barry L. Karger. Multiple open-channel electroosmotic pumping system for microfluidic sample handling. Analytical Chemistry. 2002;74(24):

[24] John M. Edwards IV, Mark N. Hamblin, Hernan V. Fuentes, Bridget A. Peeni, Milton L. Lee, Adam T. Woolley and Aaron R. Hawkins. Thin film electro-osmotic pumps for biomicrofluidic applications. Biomicrofluidics. 2007;1(1):014101. DOI:

[25] Ceming Wang, Lin Wang, Xiaorui Zhu, Yugang Wang and Jianming Xue. Low-volt‐ age electroosmotic pumps fabricated from track-etched polymer membranes. Lab on

[26] Anders Brask, Jörg P. Kutter and Henrik Bruus. Long-term stable electroosmotic pump with ion exchange membranes. Lab on a Chip. 2005;5(7):730-738. DOI: 10.1039/

[27] Yu-Feng Chen, Ming-Chia Li, Yi-Hsin Hu, Wen-Jeng Chang and Chi-Chuan Wang. Low-voltage electroosmotic pumping using porous anodic alumina membranes. Mi‐ crofluidics and Nanofluidics. 2008;5(2):235-244. DOI: 10.1007/s10404-007-0242-8 [28] Anders Brask, Detlef Snakenborg, Jörg P. Kutter and Henrik Bruus. AC electroosmot‐ ic pump with bubble-free palladium electrodes and rectifying polymer membrane

1993;283(2):747-753. DOI: 10.1016/0003-2670(93)85289-V

250 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

10.1016/0039-9140(94)00145-6

10.1016/0003-2670(94)00478-5

DOI: 10.1016/S0021-9673(03)00941-5

771-777. DOI: 10.1007/s10404-011-0905-3

6259-6268. DOI: 10.1021/ac0203950

10.1063/1.2372215

B503626G

2009;36(1):75-81. DOI: 10.1016/j.mechrescom.2008.06.015

a Chip. 2012;12(9):1710-1716. DOI: 10.1039/C2LC40059F

valves. Lab on a Chip. 2006;6(2):280-288. DOI: 10.1039/B509997H


[52] Wee Yang Ng, Shireen Goh, Yee Cheong Lam, Chun Yang and Isabel Rodríguez. DCbiased AC-electroosmotic and AC-electrothermal flow mixing in microchannels. Lab on a Chip. 2009;9(6):802-809. DOI: 10.1039/B813639D

[40] Nazmul Islam and Davood Askari. Performance improvement of an AC electroos‐ motic micropump by hydrophobic surface modification. Microfluidics and Nanoflui‐

[41] J.-Y. Miao, Z.-L. Xu, X.-Y. Zhang, N. Wang, Z-Y. Yang and P. Sheng. Micropumps based on the enhanced electroosmotic effect of aluminum oxide membranes. Ad‐

[42] Z. Cao, L. Yuan, Y. F. Liu, S. Yao and L. Yobas. Microchannel plate electro-osmotic pump. Microfluidics and Nanofluidics. 2012;13(2):279-288. DOI: 10.1007/

[43] Shawn Litster, Matthew E. Suss and Juan G. Santiago. A two-liquid electroosmotic pump using low applied voltage and power. Sensors and Actuators A: Physical.

[44] Wei Wang, Congying Gu, Kyle B. Lynch, Joann J. Lu, Zhengyu Zhang and Qiaosh‐ eng Pu. High-pressure open-channel on-chip electroosmotic pump for nanoflow high performance liquid chromatography. Analytical Chemistry. 2014;86(4):1958-1964.

[45] Rudra Kumar, Kousar Jahan, Rajaram K. Nagarale and Ashutosh Sharma. Nongass‐ ing long-lasting electro-osmotic pump with polyaniline wrapped aminated graphene electrodes. ACS Applied Materials & Interfaces. 2015;7(1):593−601. DOI: 10.1021/

[46] Woonsup Shin, Jong Myung Lee, Rajaram Krishna Nagarale, Samuel Jaeho Shin and Adam Heller. A Miniature, nongassing electroosmotic pump operating at 0.5 V. The Journal of the American Chemical Society. 2011;133(8):2374-2377. DOI: 10.1021/

[47] Meng Gao and Lin Gui. A handy liquid metal based electroosmotic flow pump. Lab

[48] Chia-Yen Lee, Chin-Lung Chang, Yao-Nan Wang and Lung-Ming Fu. Microfluidic mixing: A review. International Journal of Molecular Sciences. 2011;12(5):3263-3287.

[49] Chih-Chang Chang and Ruey-Jen Yang. Electrokinetic mixing in microfluidic sys‐ tems. Microfluidics and Nanofluidics. 2007;3(5):501-525. DOI: 10.1007/

[50] Chen-li Sun and Shin-Shian Shie. Optimization of a diverging micromixer driven by periodic electroosmotics. Microsystem Technologies. 2012;18(9):1237-1245. DOI:

[51] Pin-Hsien Chiu, Chih-Chang Chang and Ruey-Jen Yang. Electrokinetic micromixing of charged and non-charged samples near nano–microchannel junction. Microflui‐

dics and Nanofluidics. 2013;14(5):839-844. DOI: 10.1007/s10404-012-1116-2

on a Chip. 2014;14(11):1866-18752. DOI: 10.1039/C4LC00111G

vanced Materials. 2007;19(23):4234-4237. DOI: 10.1002/adma.200700767

dics. 2013;14(3):627-635. DOI: 10.1007/s10404-012-1081-9

252 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

2010;163(1):311-314. DOI: 10.1016/j.sna.2010.07.008

s10404-012-0959-x

DOI: 10.1021/ac4040345

DOI: 10.3390/ijms12053263

10.1007/s00542-012-1475-3

s10404-007-0178-z

am506766e

ja110214f


**Fabrications and Applications in Energy Research**

## **Laser-Based Fabrication for Microfluidics Devices on Glass for Medical Applications**

Daniel Nieto García and Gerard O'Connor

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/64324

#### **Abstract**

We report a laser-based process for microstructuring glass materials for microfluidics ap‐ plications. The hybrid technique is composed of a nanosecond Q-Switch Nd:YVO4 laser for fabricating the initial microfluidic microstructures on soda-lime glass substrates and a thermal treatment for reshaping and improving its morphological and optical qualities. The proposed technique preserves the advantages of the laser direct-write technique in terms of design flexibility, simplicity, fast prototyping, low cost, and so on. The beam spot size, pulse overlapping, ablation threshold, debris deposition, heating temperature, and time are investigated and optimized for fabricating optimal microfluidics structures on glass. The manufactured chips for circulating tumor cells (CTCs) capture were tested with tumor cells (Hec 1A) after being functionalized with an EpCAM antibody coating. Cells were successfully arrested on the pillars after being flown through the device giving our technology a translational application in the field of cancer research.

**Keywords:** Laser microstructuring, microchannels, glass, microfluidics devices

## **1. Introduction**

Microfluidics is a fast developing engineering science that controls the flow of minute amounts of liquids or gases in a miniaturized system in an increasing number of applications such as biomedical diagnostic, microfuel cells, and cooling in microelectronics [1‒4]. The microfluidic devices are used in making rapid inroads in the modern analytical laboratory primarily because of their small physical footprint, speed and efficiency of chemical separations, and reduced reagent consumption. These applications demand transparent materials that permit high-resolution imaging, fluorescence microscopy, and also to analyze parameters such as laminar flow, mass transport driven by diffusion rather than turbulence, and constant removal of waste products [5‒7].

© 2016 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Traditionally, lab-on-a-chip devices have been manufactured by silica due to its well-under‐ stood surface chemistry and favorable micromachining techniques that are ubiquitous in the microelectronics industry. Recently, researchers have begun to utilize devices fabricated from polymer substrates as an alternative to glass, although glass may still be the preferred materials in clinical medicine, biology, and chemistry. Microchannels fabricated on glass have a growing importance in the miniaturization of microfluidic devices for chemical and biological micro‐ total analysis system.

Glass materials are commonly used due to the beneficial optical properties, surface stability, and solvent compatibility, as well as due to the straightforward and well-understood fabrica‐ tion techniques [8‒10]. The mechanical rigidity, chemical resistance, and low-permeability properties of glass, combined with their optical transparency, make it a good choice for many demanding lab-on-a-chip applications. Glass materials overcome some limitations of poly‐ mers, they are reusable and have low autofluorescence and smooth surfaces. Nevertheless, the high cost of processing glass and of the material itself limit the use of glass as disposable devices. Two different approaches challenging demands to be solved on microfluidics, highquality devices made of glass and disposable low-cost devices made of plastic.

There are large and well-developed micromachining methods for fabricating microfluidic devices on glass, the choice among them depends on the materials to be treated, and the size and shape of the required features. Embossing, injection molding, and other thermoforming techniques provide high throughput and cost, but are ineffective for glass [11]. Because lithography techniques require advanced facilities and numerous process steps, numerous researchers have explored the fabrication of channels in glass using other technologies, such as electron beam lithography, photolithography, and dry and wet etching [12‒15].

Some of these techniques provide high-quality microfluidic systems, but require sophisticated equipment located in clean rooms and with the inconvenient of production toxic waste, which in terms of microfabrication means that there are significant challenges in fabricating low-cost and reliable microchannels in glass. For example, the surface roughness in a microchannel can be caused by an inaccurate and imprecise alignment of the mask used in the fabrication process.

Laser has become increasingly important in recent years for many fields, including microop‐ tics, microelectronics, microbiology, and microchemistry. High micromachining quality with ns-pulse and fs-pulse lasers was demonstrated for direct ablation of dielectrics. Laser micro‐ machining offers a single-step method for direct writing of microchannels in glass. Using this laser ablation technique, it is possible to fabricate geometries with variable depth and high aspect ratio that cannot be achieved through traditional microlithographic techniques. Laser ablation, because of its noncontact nature, allows the micromachining and surface patterning of glass materials with minimal mechanical and thermal deformation. Due to the low cost, the ease of implementation in industry, and the wide quantity of lasers available, this technique is extremely important for machining glass.

Fabrication of microchannels on glass by laser ablation techniques has been investigated and reported using CO2, UV, and ultrashort pulse lasers [16‒19]. For transparent materials, in the visible spectral range, laser ablation should ideally be performed with ultraviolet radiation because of the linear optical absorption in this wavelength range. Alternatively, nonlinear coupling of ultrahigh-intensity laser pulses in the near-infrared (IR) range with subpicosecond duration may show advantages. The interaction between a laser beam and a material is determined by (1) laser characteristics (the wavelength, fluence or energy density, the pulse duration, repetition rate, and pulse energy); (2) the properties of the material (absorption characteristics and thermal relaxation) which is governed by the composition and structure of the material.

In this context, laser micromachining, because of its noncontact nature, offers several advan‐ tages for fabricating microchannels, including the capability to form complex shapes with minimal mechanical and thermal deformation. There are still significant challenges in fabri‐ cating low-cost and reliable microfluidic modules in transparent media. In particular, micro‐ fluidic devices for detection of circulating tumor cells have emerged as a promising minimally invasive diagnostic tool. Isolation of circulating tumor cells (CTCs) has become a central topic in cancer research where engineering and medical science converge with the common goal of capturing rare cell types in liquid biopsies as a starting point for early diagnosis and the development of point of care and single-cell analysis systems [20]. Several designs have explored in Polydimethylsiloxane (PDMS) and glass substrates leading to panoply of features and arrangements able to trap cells flowing through the device. In order to enhance the specificity and sensibility of such systems, a surface functionalization with EpCAM antibodies seemed to be the first choice for developing a micropillar coating, such antibodies are com‐ monly employed for CTC isolation.

Micropillars were some of the first features developed for studying cell behavior such as cell spreading, motility, and mechanobiology, but they present an interesting feature to be developed as a potential substrate prone to coating and exhibiting exceptional optical condi‐ tions for microscopy applications [21]. Therefore, functionalizing micropillars with a steady topographical control will be an optimal platform for a point-of-care device or lab-on-a-chip technology [22].

In this chapter, we present a laser-based technology for fabricating microfluidic microchannels and chips for circulating tumor cells capture. In particular, we report a fast and simple method for the fabrication of the microchannels on glass using an Nd:YVO4 nanosecond IR laser. The fabrication method combines two different laser routes: (1) a laser direct-write technique for fabricating the microchannels on glass and (2) a thermal treatment for reducing the damage created on the glass during the laser direct-write process. By combining these two laser process, it is possible to obtain high-quality microchannels with comparable characteristics as those obtained by other methods [23], preserving the advantages of the laser direct-write technique (flexibility in terms of design, fast prototyping, low-cost, noncontaminant, etc). In Section 2, the experimental procedure for fabricating the microchannels is presented. Section 3 analyzes the thermal treatment and morphological characterization of microchannels as a function of temperature. Section 4 presents some conclusions and remarks.

## **2. Materials and methods**

Traditionally, lab-on-a-chip devices have been manufactured by silica due to its well-under‐ stood surface chemistry and favorable micromachining techniques that are ubiquitous in the microelectronics industry. Recently, researchers have begun to utilize devices fabricated from polymer substrates as an alternative to glass, although glass may still be the preferred materials in clinical medicine, biology, and chemistry. Microchannels fabricated on glass have a growing importance in the miniaturization of microfluidic devices for chemical and biological micro‐

258 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

Glass materials are commonly used due to the beneficial optical properties, surface stability, and solvent compatibility, as well as due to the straightforward and well-understood fabrica‐ tion techniques [8‒10]. The mechanical rigidity, chemical resistance, and low-permeability properties of glass, combined with their optical transparency, make it a good choice for many demanding lab-on-a-chip applications. Glass materials overcome some limitations of poly‐ mers, they are reusable and have low autofluorescence and smooth surfaces. Nevertheless, the high cost of processing glass and of the material itself limit the use of glass as disposable devices. Two different approaches challenging demands to be solved on microfluidics, high-

There are large and well-developed micromachining methods for fabricating microfluidic devices on glass, the choice among them depends on the materials to be treated, and the size and shape of the required features. Embossing, injection molding, and other thermoforming techniques provide high throughput and cost, but are ineffective for glass [11]. Because lithography techniques require advanced facilities and numerous process steps, numerous researchers have explored the fabrication of channels in glass using other technologies, such

Some of these techniques provide high-quality microfluidic systems, but require sophisticated equipment located in clean rooms and with the inconvenient of production toxic waste, which in terms of microfabrication means that there are significant challenges in fabricating low-cost and reliable microchannels in glass. For example, the surface roughness in a microchannel can be caused by an inaccurate and imprecise alignment of the mask used in the fabrication process. Laser has become increasingly important in recent years for many fields, including microop‐ tics, microelectronics, microbiology, and microchemistry. High micromachining quality with ns-pulse and fs-pulse lasers was demonstrated for direct ablation of dielectrics. Laser micro‐ machining offers a single-step method for direct writing of microchannels in glass. Using this laser ablation technique, it is possible to fabricate geometries with variable depth and high aspect ratio that cannot be achieved through traditional microlithographic techniques. Laser ablation, because of its noncontact nature, allows the micromachining and surface patterning of glass materials with minimal mechanical and thermal deformation. Due to the low cost, the ease of implementation in industry, and the wide quantity of lasers available, this technique

Fabrication of microchannels on glass by laser ablation techniques has been investigated and reported using CO2, UV, and ultrashort pulse lasers [16‒19]. For transparent materials, in the visible spectral range, laser ablation should ideally be performed with ultraviolet radiation because of the linear optical absorption in this wavelength range. Alternatively, nonlinear coupling of ultrahigh-intensity laser pulses in the near-infrared (IR) range with subpicosecond

quality devices made of glass and disposable low-cost devices made of plastic.

as electron beam lithography, photolithography, and dry and wet etching [12‒15].

is extremely important for machining glass.

total analysis system.

In order to perform the microstructuring of the glass a Rofin model Nd:YVO4 laser was used. This is a solid-state laser, operating at 1064 nm wavelength and using a Q-Switch regime, with pulses of 20 ns and tunable repetition rate from single-shot to 200 kHz. For the thermal treatment, we used a Heraeus mufla furnace, equipped with a thermal ramp that allows us to control the heating speed of the glass.

For characterizing the microstructures and samples, we have used a PerkinElmer Lamb25 spectrometer (for measuring the transmission spectrum); a scanning electron microscope (SEM, Zeiss FESEM-ULTRA Plus) for determination of the chemical composition, a microscope Nikon MM-400 (visual inspection of the samples), and a confocal microscope SENSOFAR 2300 Plµ (topographic and surface roughness measurement). Cell culture and trapping were assessed by means of a Zeiss Microscope and Zen software.

The glass used as a substrate for fabricating the microchannels and microstructures was a commercial soda-lime glass, provided by a local supplier. The composition of this glass (O 50.25%, Na 9.08%, Mg 2.19%, Al 0.54%, Si 33.08%, Ca 4.87%) was determined by using a SEM EDX (Zeiss FESEM-ULTRA Plus).

## **3. Results**

#### **3.1. Laser microstructuring**

The laser direct-write technique for fabricating microfluidics devices and microchannels is based on the ablation of a soda-lime glass substrate with a beam laser of circular Gaussian profile, followed of a thermal treatment in a Heraeus mufla furnace. The laser setup for fabricating the micropost system consists of a Q-Switch Nd:YVO4 laser operating at 1064 nm combined with a galvanometer system for addressing the output laser beam. The beam is focused by a flat-field lens of effective focal length 100 mm that provides a uniform irradiance distribution over a working area of 80 × 80 mm2 (Figure 1). In order to perform the thermal treatment that allows us to improve the quality of the originally generated microchannels, a furnace Heraeus mufla was employed.

For optimizing the fabrication process, we have done a previous study of the mark obtained by focusing one laser shot on the glass substrate [24]. This mark was then analyzed with a confocal microscope. This initial characterization step let us to determine the optimal laser parameters used in the laser direct-write process. The microchannels morphology was analyzed in terms of shape and roughness using a confocal microscope SENSOFAR 2300 Plµ.

For determining the size of the mark made by the laser beam in the glass substrate, a crater was made in the glass after firing one shot at a fluence of 112 J/cm2 . The applied fluence was calculated by using the expression *F*th = (2*E*th)/(*πω*<sup>0</sup> 2 ), where *E*th is the energy per pulse at the onset of ablation, *ω*0 was determined by analyzing the mark obtained by the laser over the material.

Because the measured Gaussian spot size on the surface is not necessarily equal to the calculated spot size at the focal point of the lens, the ablation threshold was computed using the following equation *ϕ*th = (2*E*th)/(*πω*0), where *E*th is the laser energy at the ablation threshold Laser-Based Fabrication for Microfluidics Devices on Glass for Medical Applications http://dx.doi.org/10.5772/64324 261

**Figure 1.** Laser setup for fabricating the microfluidic systems.

pulses of 20 ns and tunable repetition rate from single-shot to 200 kHz. For the thermal treatment, we used a Heraeus mufla furnace, equipped with a thermal ramp that allows us to

For characterizing the microstructures and samples, we have used a PerkinElmer Lamb25 spectrometer (for measuring the transmission spectrum); a scanning electron microscope (SEM, Zeiss FESEM-ULTRA Plus) for determination of the chemical composition, a microscope Nikon MM-400 (visual inspection of the samples), and a confocal microscope SENSOFAR 2300 Plµ (topographic and surface roughness measurement). Cell culture and trapping were

The glass used as a substrate for fabricating the microchannels and microstructures was a commercial soda-lime glass, provided by a local supplier. The composition of this glass (O 50.25%, Na 9.08%, Mg 2.19%, Al 0.54%, Si 33.08%, Ca 4.87%) was determined by using a SEM

The laser direct-write technique for fabricating microfluidics devices and microchannels is based on the ablation of a soda-lime glass substrate with a beam laser of circular Gaussian profile, followed of a thermal treatment in a Heraeus mufla furnace. The laser setup for fabricating the micropost system consists of a Q-Switch Nd:YVO4 laser operating at 1064 nm combined with a galvanometer system for addressing the output laser beam. The beam is focused by a flat-field lens of effective focal length 100 mm that provides a uniform irradiance distribution over a working area of 80 × 80 mm2 (Figure 1). In order to perform the thermal treatment that allows us to improve the quality of the originally generated microchannels, a

For optimizing the fabrication process, we have done a previous study of the mark obtained by focusing one laser shot on the glass substrate [24]. This mark was then analyzed with a confocal microscope. This initial characterization step let us to determine the optimal laser parameters used in the laser direct-write process. The microchannels morphology was analyzed in terms of shape and roughness using a confocal microscope SENSOFAR 2300 Plµ.

For determining the size of the mark made by the laser beam in the glass substrate, a crater

onset of ablation, *ω*0 was determined by analyzing the mark obtained by the laser over the

Because the measured Gaussian spot size on the surface is not necessarily equal to the calculated spot size at the focal point of the lens, the ablation threshold was computed using the following equation *ϕ*th = (2*E*th)/(*πω*0), where *E*th is the laser energy at the ablation threshold

2

. The applied fluence was

), where *E*th is the energy per pulse at the

was made in the glass after firing one shot at a fluence of 112 J/cm2

calculated by using the expression *F*th = (2*E*th)/(*πω*<sup>0</sup>

control the heating speed of the glass.

EDX (Zeiss FESEM-ULTRA Plus).

**3.1. Laser microstructuring**

furnace Heraeus mufla was employed.

**3. Results**

material.

assessed by means of a Zeiss Microscope and Zen software.

260 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

and *ω*<sup>0</sup> is the waist of the focal spot. Its value was obtained from the slope of the plot of ln(*E*p/ *E*th) against *D*<sup>2</sup> (diameter squared) (*<sup>D</sup>* <sup>2</sup> =2*ω*<sup>0</sup> 2 *ln*( *Ep Eth* ) see [25] for details) where *E*p is the energy per pulse (Figure 2).

**Figure 2.** Linear relations between the squared diameter and the natural logarithm of the average laser fluences.

The main morphological parameters after laser ablation, such as depth and width for different number of pulses (fluence: 112 J/cm2 ), have been analyzed using a confocal microscope SENSOFAR 2300 Plµ and an optical microscope Nikon MM-400.

Figures 3 and 4 show the evolution of the damage (diameter and depth) at the substrate surface with the number of pulses (1‒10,000 pulses). Measurements of the depth and diameter of the ablated holes can also provide important information about the material response to the nanosecond laser.

**Figure 3.** Evolution of depth versus number of pulses for (a) 1‒10 pulses and (b) 10‒10,000 pulses.

As shown in Figure 3a, the depth increases with the first pulses, reaching a depth value of 8 µm after 10 pulses, however after 10 pulses (Figure 3b), the depth increases slowly, varying only 2 µm after 500 pulses and maintain almost constant till 10,000 pulses. That behavior is due to the fact that the ablation plume does not have sufficient energy to take out the material detached on the ablation process.

**Figure 4.** Evolution for diameter versus number of pulses: (a) 1‒10 pulses and (b) 10‒10,000 pulses.

From the slope obtained in Figure 4a, the evolution of diameter with the number of pulses can be deduced. The mark diameter increases slowly with the number of pulses, reaching a maximum value of depth 10 µm after 500 pulses, after this value (Figure 4b), increasing the number of pulses does not have a significant effect on the mark diameter.

For fabricating a homogeneous microchannel on glass, the optimization of the pulse overlaps, which determines the pulse per area is important to obtain quality microchannels. When the pulse overlap is high, there will be significant cracking in the microchannels edges because of the stress related with heat deposited in a small area. To obtain the optical pulse overlap, *O*d, we use the following relation:

Figures 3 and 4 show the evolution of the damage (diameter and depth) at the substrate surface with the number of pulses (1‒10,000 pulses). Measurements of the depth and diameter of the ablated holes can also provide important information about the material response to the

262 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

1 2 3 4 5 6 7 8 9 10 10 100 1000 10000

1 2 3 4 5 6 7 8 9 10 10 100 1000 10000

Number of pulses Number of pulses a) b)

From the slope obtained in Figure 4a, the evolution of diameter with the number of pulses can be deduced. The mark diameter increases slowly with the number of pulses, reaching a maximum value of depth 10 µm after 500 pulses, after this value (Figure 4b), increasing the

**Figure 4.** Evolution for diameter versus number of pulses: (a) 1‒10 pulses and (b) 10‒10,000 pulses.

number of pulses does not have a significant effect on the mark diameter.

Diameter (µm)

12

20

16

4

8

8

12

6

Number of pulses Number of pulses a) b)

As shown in Figure 3a, the depth increases with the first pulses, reaching a depth value of 8 µm after 10 pulses, however after 10 pulses (Figure 3b), the depth increases slowly, varying only 2 µm after 500 pulses and maintain almost constant till 10,000 pulses. That behavior is due to the fact that the ablation plume does not have sufficient energy to take out the material

**Figure 3.** Evolution of depth versus number of pulses for (a) 1‒10 pulses and (b) 10‒10,000 pulses.

Depth (µm) 10

nanosecond laser.

12

10

8

Diameter (µm)

detached on the ablation process.

Depth (µm)

$$\mathbf{O}\_d = \left( 1 - \frac{\nu}{2a\_0 f} \right) \tag{1}$$

where *υ* is the galvoscanner speed, *2ω*0 is the focused spot diameter, and *f* is the laser repetition rate.

The first experimental approach for obtaining uniform channels was checked at different scan speeds for getting the optimal pulses overlapping (Figure 5). The same laser parameters (10 kHz, 8 W) were used for fabricating all the lines varying only the scan speed.

**Figure 5.** Confocal images of one microchannel obtained by laser ablation of glass at 10 kHz, 8 W for scan speeds val‐ ues of (a) 10 mm/s, (b) 2 mm/s, (c) 30 mm/s, (d) 40 mm/s, (e) 50 mm/s, (f) 60 mm/s, (g) 70 mm/s, (h) 80 mm/s, (i) 90 mm/s, (j) 100 mm/s, (k) 110 mm/s, (l) 120 mm/s, (m) 130 mm/s, and (n) 140 mm/s.

As it can be appreciate in Figure 5, the best results were obtained for the channels fabricated between 40 and 80 mm/s (Figure 5d and h). At low scan speed, the high overlapping of the pulses delivers too much energy at the surface, which leads to an irregular channel. Low overlapping creates a distortion on the microchannels (Figure 5j‒n). For scan speeds higher than 80 mm/s, it can be appreciated the interaction of one pulse with the substrates (Figure 5a‒ c), which also increase the roughness average and does not allow the formation of the micro‐ channels. Figure 6 shows the pulse overlap, obtained using Eq. (1), and the roughness average measured at different scan speeds. In terms of pulse overlap (Figure 6a), values in the range of 60‒80% are needed for fabricating microchannels with good qualities. For scan speed values in the interval of 40‒80 mm/s, roughness around 55 nm is obtained (Figure 6b).

**Figure 6.** (a) Evolution of pulse overlap and (b) roughness versus scan speed in the range of 10‒140 mm/s.

Figure 7 shows a microchannel fabricated using the optimal pulse overlap obtained from the study presented above, and laser parameters of 8 W, 10 kHz and scan speed of 50 mm/s. The resulting channel exhibits a surface roughness of 54 nm with a diameter of 8 µm and height of 3 µm.

**Figure 7.** Initial microchannel fabricated by laser ablation.

In order to fabricate different microchannels configurations, a study of the evolution of depth and diameter with the number of laser passes over the same place was done (laser parameters:

8 W, 10 kHz and scan speed 50: mm/s). Figure 8 shows the evolution of depth and diameters varying the number of laser passes.

**Figure 8.** (a) Evolution of depth and (b) diameter with the number of laser passes.

overlapping creates a distortion on the microchannels (Figure 5j‒n). For scan speeds higher than 80 mm/s, it can be appreciated the interaction of one pulse with the substrates (Figure 5a‒ c), which also increase the roughness average and does not allow the formation of the micro‐ channels. Figure 6 shows the pulse overlap, obtained using Eq. (1), and the roughness average measured at different scan speeds. In terms of pulse overlap (Figure 6a), values in the range of 60‒80% are needed for fabricating microchannels with good qualities. For scan speed values

in the interval of 40‒80 mm/s, roughness around 55 nm is obtained (Figure 6b).

264 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

a) b)

**Figure 6.** (a) Evolution of pulse overlap and (b) roughness versus scan speed in the range of 10‒140 mm/s.

Figure 7 shows a microchannel fabricated using the optimal pulse overlap obtained from the study presented above, and laser parameters of 8 W, 10 kHz and scan speed of 50 mm/s. The resulting channel exhibits a surface roughness of 54 nm with a diameter of 8 µm and height

In order to fabricate different microchannels configurations, a study of the evolution of depth and diameter with the number of laser passes over the same place was done (laser parameters:

**Figure 7.** Initial microchannel fabricated by laser ablation.

of 3 µm.

It is evident from Figure 8 that the channel aspect ratio α = *h*/*d* varies with the number of passes. The diameter *d* reaches its saturation value after five laser passes, increasing just around 200 nm per laser pass and varying only 1 µm after five passes, above these values the diameter is maintained almost constant. In contrast, the ablation height *h* changes dramatically with the increase in number of pulses, after 6 passes this value remains constant. The height increases 1 µm per laser pass, up to a value of 12 µm. We can relate this behavior with the nonevacuation of the debris generated during laser ablation process that arise from the bottom of the micro‐ channels and also because to the lack of focus as the depth of the microchannel increases. The diameter of the microchannel is maintained almost constant by increasing the number of laser passes, which can be related with the diameter of the laser beam used.

One of the important challenges to overcome using a laser direct-write process for fabricating microchannels in glass is to obtain good-quality junctions. The propelled material of the subsequent channels is deposited on the existing microchannel that turns into a bad quality of the microchannel. This debris changes the average roughness, and because it is crucial for obtaining high-quality elements, the microchannels were cleaned after laser exposure using chemical etching process of hydrofluoric acid (HF). The HF acid is an etchant that attacks glasses at significant high-etch rate [16]. Commercial soda-lime glass used in this work is composed of SiO2 (73.8%), Na2O (12.7%), CaO (8.6%), MgO (4.1%), Fe2O3 (0.14%), and Al2O3 (0.1%). Etching soda-lime glass in an HF solution forms insoluble products that are believed to be mainly CaF2 and MgF2.

The etching time, concentration of the acid, and the temperature of the process are important parameters to control the HF cleaning process. In terms of handle, HF is a strong corrosive and highly toxic at higher concentrations. The etching process was performed at 10% HF concen‐ tration, which in terms of security reduces considerably percentages of toxic vapor that contaminates the work space. Figure 9 shows a SEM top view image of one microchannel before (Figure 9a) and after (Figure 9b) HF etching.

**Figure 9.** SEM image of the microchannel top surface (a) before chemical etching and (b) after 10 min in 10% HF aque‐ ous solution.

In Figure 9a, debris deposited on the top of the microchannels during the laser ablation can be observed. In Figure 9b, after chemical etching, the debris was successfully eliminated.

### **3.2. Thermal treatment**

A thermal treatment was applied into an oven to obtain the final elements and to improve the morphological properties. The samples were reflown in a Heraeus mufla furnace for 2 h at temperatures between 620°C and 670°C (steps of 10°C). The working range has been chosen to be higher than the glass transition temperature of soda-lime glass (*T*g =564°C). Figure 10 shows a 3D confocal image of microfluidic microchannels fabricated at different reflow temperatures.

As shown in Figure 10, at 620°C, there was not a significant modification in the surface shape, and at temperatures higher than 670°C, the initial shape surface profile becomes flat so that microchannels obtained by laser direct-write disappear.

This displacement of material due to the thermal treatment and the accumulation in the bottom turns into a reduction in height while the diameter is increased due to the material reflow from the top of the edges of the microchannels at the bottom of the crater. This effect allows both the thermal reflow and to fill the irregular structure of crater with material leading to an improvement on the morphological qualities.

The flow resistance for each microchannel depends on the geometry and roughness. The surface roughness on the wall of the microchannel increases when the flow rate decreases, so

contaminates the work space. Figure 9 shows a SEM top view image of one microchannel

a) b)

**Figure 9.** SEM image of the microchannel top surface (a) before chemical etching and (b) after 10 min in 10% HF aque‐

In Figure 9a, debris deposited on the top of the microchannels during the laser ablation can be observed. In Figure 9b, after chemical etching, the debris was successfully eliminated.

A thermal treatment was applied into an oven to obtain the final elements and to improve the morphological properties. The samples were reflown in a Heraeus mufla furnace for 2 h at temperatures between 620°C and 670°C (steps of 10°C). The working range has been chosen to be higher than the glass transition temperature of soda-lime glass (*T*g =564°C). Figure 10 shows a 3D confocal image of microfluidic microchannels fabricated at different reflow

As shown in Figure 10, at 620°C, there was not a significant modification in the surface shape, and at temperatures higher than 670°C, the initial shape surface profile becomes flat so that

This displacement of material due to the thermal treatment and the accumulation in the bottom turns into a reduction in height while the diameter is increased due to the material reflow from the top of the edges of the microchannels at the bottom of the crater. This effect allows both the thermal reflow and to fill the irregular structure of crater with material leading to an

The flow resistance for each microchannel depends on the geometry and roughness. The surface roughness on the wall of the microchannel increases when the flow rate decreases, so

microchannels obtained by laser direct-write disappear.

improvement on the morphological qualities.

before (Figure 9a) and after (Figure 9b) HF etching.

266 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

ous solution.

**3.2. Thermal treatment**

temperatures.

**Figure 10.** Confocal images of microchannels obtained after thermal treatment at different temperatures. Six different results at different reflow temperatures are shown: *T*<sup>1</sup> = 620°C, *T*<sup>2</sup> = 630°C, *T*<sup>3</sup> = 640°C, *T*4 = 650°C, *T*5 = 660°C, and *T*6 = 670°C.

the change of hydraulic resistance is proportional to the change in the surface roughness. For determining the quality of the microchannels fabricated at different thermal reflow tempera‐ tures, it was determined the roughness at the bottom of the channels [26]. Table 1 shows the evolution of roughness for temperatures between 620 and 670°C, taken at steps of 10°C. The purpose of thermal treatment is to reduce the roughness generated during the laser directwrite process. It is important to find a compromise between shape modification and roughness. The shape should be maintained whereas the thermal reflow should reduce the surface roughness.


**Table 1.** Comparative of roughness evolution with thermal reflow.

In order to study the microfluidics capabilities of our technique, after obtaining the best parameters for fabricating optimal microchannels, we focused on fabricating microchannels with different configurations trying to solve the main problems related with the laser ablation of glass, in particular for creating uniform and debris-free channels. Figure 11 shows some microchannels fabricated using the laser direct-write technique.

**Figure 11.** Microfluidics microchannels fabricated using the laser direct-write technique.

#### **3.3. Biochip for capturing circulating tumor cells**

In the previous work, we have demonstrated the suitability of the presented technique for fabricating a circulating tumor cell biochip [20]. The biochip is composed of a group of micropost (see Figure 11a), integrated in a microfluidics devices (see Figure 11d) for capturing circulating tumor cells. Using the aforementioned laser technique, we have fabricated an input channel for perfusing the media and two outputs channels for media removal (Figure 11d). The microfluidics devices were covered with a microscope cover slide (100 µm thickness). The manufactured chips were tested for detection of tumor cells (Hec 1A) after being functionalized with an EpCAM antibody coating. Cells were flown with media through the device allowing an interaction with the functionalized micropillars. After eluting the whole sample, chips were assessed for sensitivity and efficiency under a confocal microscope.

Fluorescently stained (DiO) CTCs were counted the cell trapping array region of the biochip using epifluorescent microscope (Zeiss Axio Vert A.1) was equipped with a 20× objective (Zeiss, numerical aperture = 0.4) and a low-light CCD camera.

Figure 12 shows the resulting cell isolation from cells flown in media. Pillars were able to trap circulating cells pumped in the device.

**Figure 12.** Cells attachment to the functionalized surface is observed in (a), (b), or (c). Cells are strongly bonded to the antibody surface and are not detached under flow conditions (white arrows show the location of the cells).

Pictures were taken along the device, there is a heterogeneous distribution for cell isolation along the posts while due to the chip dimensions few cells were found on the bottom (Figure 12c). In accordance to the surface functionalization protocol, the whole chip was coated with EpCAM antibodies, hence the bottom areas should be able to detain cells as well.

## **4. Conclusions**

In order to study the microfluidics capabilities of our technique, after obtaining the best parameters for fabricating optimal microchannels, we focused on fabricating microchannels with different configurations trying to solve the main problems related with the laser ablation of glass, in particular for creating uniform and debris-free channels. Figure 11 shows some

microchannels fabricated using the laser direct-write technique.

268 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

**Figure 11.** Microfluidics microchannels fabricated using the laser direct-write technique.

assessed for sensitivity and efficiency under a confocal microscope.

(Zeiss, numerical aperture = 0.4) and a low-light CCD camera.

circulating cells pumped in the device.

In the previous work, we have demonstrated the suitability of the presented technique for fabricating a circulating tumor cell biochip [20]. The biochip is composed of a group of micropost (see Figure 11a), integrated in a microfluidics devices (see Figure 11d) for capturing circulating tumor cells. Using the aforementioned laser technique, we have fabricated an input channel for perfusing the media and two outputs channels for media removal (Figure 11d). The microfluidics devices were covered with a microscope cover slide (100 µm thickness). The manufactured chips were tested for detection of tumor cells (Hec 1A) after being functionalized with an EpCAM antibody coating. Cells were flown with media through the device allowing an interaction with the functionalized micropillars. After eluting the whole sample, chips were

Fluorescently stained (DiO) CTCs were counted the cell trapping array region of the biochip using epifluorescent microscope (Zeiss Axio Vert A.1) was equipped with a 20× objective

Figure 12 shows the resulting cell isolation from cells flown in media. Pillars were able to trap

**3.3. Biochip for capturing circulating tumor cells**

A hybrid method for fabricating microfluidic microchannels and a biochip for CTCs on sodalime glass has been developed. The fabrication process consists of a combination of the laser direct-write technique for promoting glass structures and a thermal treatment for reshaping and or improving the morphological qualities of the generated microchannels. By applying the thermal treatment at 620°C (during 2 h), we are able to obtain high-quality microchannels maintaining the initial shape. This treatment allows us to reshape and improve the morpho‐ logical, in terms of roughness, and optical qualities, in terms of transparency, of the generated microfluidics structures. This approach is a rapid and low-cost procedure to obtain glass microstructures where a chemical modification can take place, leading to the possibility of performing surface modifications with EpCAM antibodies to successfully stop circulating tumor cells. Therefore, the presented glass platform in this chapter can provide a good starting point to develop more complex systems aimed for single-cell analysis and isolation leading to a primary culture.

## **Acknowledgements**

This work was supported by Xunta de Galicia, Spain, under Galician Programme for Research Innovation and Growth 2011‒2015 (I2C Plan). We would also like to thank to Ramiro Couceiro for his assistance with the biological trials and confocal images.

## **Author details**

Daniel Nieto García1,2\* and Gerard O'Connor2

\*Address all correspondence to: daniel.nieto@usc.es

1 Microoptics and GRIN Optics Group, Applied Physics Department, Faculty of Physics, University of Santiago de Compostela, Santiago de Compostela, Spain

2 School of Physics, National Centre for Laser Applications, National University of Ireland, University Road, Galway, Ireland

## **References**


[10] J.Y. Cheng, M.H. Yen, C.W. Wei, Y.C. Chuang and T.H. Young. Crack-free directwriting on glass using a low-power UV laser in the manufacture of a microfluidic chip Journal of Micromechanics Microengineering. 2005; 15:1147‒1156.

**Author details**

**References**

Daniel Nieto García1,2\* and Gerard O'Connor2

University Road, Galway, Ireland

2004;507:11‒26.

cal Systems, 2003;12(6):907‒912.

croengineering. 2011;21:075008.

2007‒2009.

373.

\*Address all correspondence to: daniel.nieto@usc.es

University of Santiago de Compostela, Santiago de Compostela, Spain

270 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

1 Microoptics and GRIN Optics Group, Applied Physics Department, Faculty of Physics,

2 School of Physics, National Centre for Laser Applications, National University of Ireland,

[1] D. Erickson and D. Li. Integrated microfluidic devices. Analytica Chimica Acta.

[2] Y. Cheng, K. Sugioka and K. Midorikawa. Microfluidic laser embedded in glass by three-dimensional femtosecond laser microprocessing. Optics Letters. 2004;29(17):

[3] F. Cattaneo, K. Baldwin, S. Yang, T. Krupenkine, S. Ramachandran and J.A. Rogers. Digitally tunable microfluidic optical fiber devices. Journal of Microelectromechani‐

[4] G.M. Whitesides. The origins and the future of microfluidics. Nature. 2006;442:368‒

[5] T. Rajabi, V. Huck, R. Ahrens, M.C. Apfel, S.E. Kim, S.W. Schneider and A.E. Guber. Development of a novel two-channel microfluidic system for biomedical applications

[6] C.W. Huang and G.B. Lee. A microfluidic system for automatic cell culture. Journal

[7] H. Andersson, A. Van Den Berg. Microfabrication and microfluidics for tissue engi‐

[8] M.S. Giridhar, K. Seong, A. Schülzgen, P. Khulbe, N. Peyghambarian, and M. Man‐ suripur. Femtosecond pulsed laser micromachining of glass substrates with applica‐

[9] Y. Li, S. Qu and Z. Guo. Fabrication of microfluidic devices in silica glass by waterassisted ablation with femtosecond laser pulses. Journal of Micromechanics and Mi‐

neering: state of the art and future opportunities. Lab Chip. 2004;4:98‒103.

tion to microfluidic devices", Applied Optics. 2004: 43(23):4584‒4589.

in cancer research. Biomedical Engineering. 2012; 57:921‒922.

of Micromechanics and Microengineering. 2007; 17(7):1266.


## **Microfluidics in Membraneless Fuel Cells**

Jesus A. Diaz-Real, Minerva Guerra-Balcázar, Noe Arjona, Francisco Cuevas-Muñiz, Luis Gerardo Arriaga and Janet Ledesma-García

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/64448

#### **Abstract**

[23] D. Nieto, R. Couceiro, M. Aymerich, R. Lopez-Lopez, M. Abal and M.T. Flores-Arias. A laser-based technology for fabricating a soda-lime glass based microfluidic device for circulating tumour cell capture. Colloids and Surfaces B: Biointerfaces,

[24] D. Nieto, M.T. Flores-Arias, G. O'Connor and C. Gomez-Reino. A Laser direct-write technique for fabricating microlens arrays on soda lime glass with a Nd:YVO4 laser.

[25] A. Ben-Yakar and Robert L. Byer, Femtosecond laser ablation properties of borosili‐

[26] D. Nieto, T. Delgado and M.T. Flores-Arias. Fabrication of microchannels on sodalime glass substrates with a Nd:YVO4 laser. Optics and Lasers in Engineering.

2015;134:363‒369.

2014;63:11–18.

Applied Optics. 2010;49(26):4979‒4983.

272 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

cate glass. Journal of Applied Physics. 2004;96:5316‒5323.

In the 1990s, the idea of developing miniaturized devices that integrate functions other than what normally are carried out at the laboratory level was conceived, and the socalled "lab-on-a-chip" (LOC) devices emerged as one of the most important research areas. LOC devices exhibit advantages related to the use of microfluidic channels such as small sample and reagent consumption, portability, low-power consumption, laminar flow, and higher surface area/volume ratio that enhances both thermal dissipation and electrochemical kinetics. Fuel cells are electrochemical devices that convert chemical ener‐ gy to electrical energy. These are considered as one of the greener ways to generate elec‐ tricity because typical fuel cells produce water and heat as the main reaction byproducts. The technical challenges to develop systems at the microscale and the advantages of mi‐ crofluidics exhibited an important impact on fuel cells for several reasons, mainly related to avoid inherent problems of gaseous-based fuel cells. As a result, the birth of a new type of fuel cells as microfluidic fuel cells (MFCs) took place. The first microfluidic fuel cell was reported in 2002. This MFC was operated with liquid fuel/oxidant and had the advantage of the low laminar flow generated using a "Y" microfluidic channel to sepa‐ rate the anodic and cathodic streams, resulting in an energy conversion device that did not require a physical barrier to separate both streams. This electrochemical system origi‐ nated a specific type of MFCs categorized as membraneless also called colaminar micro‐ fluidic fuel cells. Since that year, numerous works focused on the nature of fuels, oxidants and anodic/cathodic electrocatalysts, and cell designs have been reported. The limiting parameters of this kind of devices toward their use in portable applications are related to their low cell performances, small mass activity, and partial selectivity/durability of elec‐ trocatalysts. On the other hand, it has been observed that the cell design has a high effect on the cell performance due to internal cell resistances and the crossover effect. Further‐ more, current technology is growing faster than last centuries and new microfabrication technologies are always emerging, allowing the development of smaller and more power‐ ful microfluidic energy devices. In this chapter, the application of microfluidics in mem‐ braneless fuel cells is addressed in terms of evolution of cell designs of miniaturized microfluidic fuel cells as a result of new discoveries in microfabrication technology and the use of several fuels and electrocatalysts for specific and selective applications.

**Keywords:** energy conversion systems, lab-on-a-chip, membraneless, microfluidic fuel cells

© 2016 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

## **1. Introduction**

## **1.1. Microfluidic fuel cells: concept, classification, and challenges**

In 1989, R. Howe introduced the term "microelectromechanical systems" (MEMS) to describe a research field in which the machining was below the millimeter scale [1]. The rapid growth in the following years in this area generated several branches in electronics, including power sources technologies where fuel cells were comprised. However, hydrogen as the fuel for the supply of these miniaturized devices results very dangerously in the gas form and makes it complicated to store as liquid. Therefore, the use of hydrogen-containing liquid fuels such as alcohols became more attractive for portable applications due to their ease of storage and highmass energy density (6.1 kWh/kg) [2].The integration of these elements into the fuel cells introduced an important characteristic that well defines the operating principle of a micro‐ fluidic fuel cell (MFC).

Typically, an MFC is a fuel cell that operates with two liquids (one fuel and the other oxidant) that are introduced by a mechanical mean into a single microchannel at laminar flow (low Reynolds number) to establish an interface. The fuel and oxidant stream will be in contact with an anodic or a cathodic surface, respectively. Both parallel streams allow the continuous flow through the length of the channel without mixing and preventing possible fuel crossover. The low speed allows the diffusion of protons between the streams in order to perform the electrochemical processes, which replaces the proton-exchange membrane (PEM) [3]. From its first appearance in 2002 to the present day, several architectures regarding the shape of the channel, electrodes geometric arrangement, oxidant supply source, and fuels have been studied. It is also common to refer MFC as laminar flow fuel cells (LFFCs) in literature indistinctively.

#### **1.2. Microfluidic fuel cell categories**

With the recent development of MFCs, a subcategory has been created to include all these membraneless cells. Generally, fuel cells have been classified by the membrane used such as polymer electrolyte fuel cells (PEFCs), solid-oxide fuel cells (SOFCs), and others. In the case of the MFCs, two main aspects have been used to classify them and are based either on the physical phases in which fuel and oxidant streams are found [4] or the stream delivery mode onto the electrodic surface [5].

### *1.2.1. Liquid fuel/liquid oxidant*

The first categorization distinguishes reactants in liquid phase. Several fuels have been studied, including alcohols (methanol and ethanol mainly) [6–9], formic acid [10–14], glycerol [15], glucose [16, 17], dissolved hydrogen [18], V+2 [19, 20], among others. It is common to see the use of dissolved oxygen as the oxidant phase [10, 12, 21, 22], while the same is not true in the case of H2O2 [23, 24], VO2 [19, 25, 26], and KMnO4 [25, 27, 28]. An important advantage is the wide range of pH supported by this type of MFCs since the membrane is not the limiting issue. This allows the choice of supporting electrolyte to tailor the optimal operating conditions for the electrodic materials. However, a major issue in these systems is the sudden formation of bubbles due to the gas byproducts (namely, CO2) that lead to the drop of ionic conductivity, perturbation of the colaminar interface, streams crossover, or even channel blockage [4].

### *1.2.2. Liquid fuel/gas oxidant*

**1. Introduction**

fluidic fuel cell (MFC).

indistinctively.

**1.2. Microfluidic fuel cell categories**

onto the electrodic surface [5].

*1.2.1. Liquid fuel/liquid oxidant*

**1.1. Microfluidic fuel cells: concept, classification, and challenges**

274 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

In 1989, R. Howe introduced the term "microelectromechanical systems" (MEMS) to describe a research field in which the machining was below the millimeter scale [1]. The rapid growth in the following years in this area generated several branches in electronics, including power sources technologies where fuel cells were comprised. However, hydrogen as the fuel for the supply of these miniaturized devices results very dangerously in the gas form and makes it complicated to store as liquid. Therefore, the use of hydrogen-containing liquid fuels such as alcohols became more attractive for portable applications due to their ease of storage and highmass energy density (6.1 kWh/kg) [2].The integration of these elements into the fuel cells introduced an important characteristic that well defines the operating principle of a micro‐

Typically, an MFC is a fuel cell that operates with two liquids (one fuel and the other oxidant) that are introduced by a mechanical mean into a single microchannel at laminar flow (low Reynolds number) to establish an interface. The fuel and oxidant stream will be in contact with an anodic or a cathodic surface, respectively. Both parallel streams allow the continuous flow through the length of the channel without mixing and preventing possible fuel crossover. The low speed allows the diffusion of protons between the streams in order to perform the electrochemical processes, which replaces the proton-exchange membrane (PEM) [3]. From its first appearance in 2002 to the present day, several architectures regarding the shape of the channel, electrodes geometric arrangement, oxidant supply source, and fuels have been studied. It is also common to refer MFC as laminar flow fuel cells (LFFCs) in literature

With the recent development of MFCs, a subcategory has been created to include all these membraneless cells. Generally, fuel cells have been classified by the membrane used such as polymer electrolyte fuel cells (PEFCs), solid-oxide fuel cells (SOFCs), and others. In the case of the MFCs, two main aspects have been used to classify them and are based either on the physical phases in which fuel and oxidant streams are found [4] or the stream delivery mode

The first categorization distinguishes reactants in liquid phase. Several fuels have been studied, including alcohols (methanol and ethanol mainly) [6–9], formic acid [10–14], glycerol [15], glucose [16, 17], dissolved hydrogen [18], V+2 [19, 20], among others. It is common to see the use of dissolved oxygen as the oxidant phase [10, 12, 21, 22], while the same is not true in the case of H2O2 [23, 24], VO2 [19, 25, 26], and KMnO4 [25, 27, 28]. An important advantage is the wide range of pH supported by this type of MFCs since the membrane is not the limiting issue. This allows the choice of supporting electrolyte to tailor the optimal operating conditions for Oxygen is by far the most used, easily available, and low-cost oxidant for MFCs. It is well known that slow kinetics of oxygen reduction (cathodic side) greatly determine the efficiency of fuel cells and the same problem affects MFCs. To address this problematic issue, Jayashree et al. innovated an MFC by introducing the concept of an air-breathing cathode that enhanced the performance of this electrode since oxygen diffusion coefficient is four orders of magnitude higher in air than in an aqueous solution [29]. This opens the opportunity to hypothetically expand this idea to many other cells because the need of both streams is still necessary to form the colaminar interface. However, several considerations are needed to take into account during the implementation of air-breathing electrodes. Air access, atmospheric counterpres‐ sure, and drying rate in bigger electrode areas are some of the parameters that might affect the performance of these MFCs when they are scaled up.

#### *1.2.3. Flow-over stream*

One of the simplest models of MFCs involves the two distributions of the electrodes, as depicted in Figure 1: (a) a parallel arrangement (facing up or over the walls) with a vertical colaminar diffusion interface and (b) facing each (top-bottom) other with a horizontal interface. In the first case, reorientation of the interface induced by gravity is possible when there is an important difference in the densities of the streams, while for the configuration with a horizontal interface this issue is more evident, increasing the crossover stream rate [30]. Regardless of the crossover problem, the amount of electrolyte that is reacting and in intimate contact with the electrodic surface is limited to the geometry of the electrode and to the flow speed. This is due to the planar electrode configuration, and thus, only the adjacent fraction of the streams is available to react. Then, the local reactant concentration will decrease while the generated subproducts will increase. This situation, along with the low convective flow speed, will generate a depletion layer near the electrode surface and a gradient of reactant concentration. Therefore, the utilization of the electrolyte is minimal, which demands an improvement in the cell design to enhance the fuel utilization [5].

#### *1.2.4. Flow through stream*

Considering the issue on electrolyte utilization, three-dimensional porous structures for electrodes were proposed. This configuration makes use of the entry of the electrolyte though the electrode itself, forcing the intimate contact to maximize the reactant consumption by rapid convective transport, replenishing the species. Another name used to refer to this configuration is multiple inlets, and the term was first proposed by Kjeang et al. [31]. An MFC with carbon paper porous electrodes was reported by Salloum et al. that used formic acid as the fuel [27].

**Figure 1.** Schematic image of the overflow designs: (a) top-bottom electrode configuration and (b) side-by-side parallel configuration. The darker color shows the diffusion interface, while a transparent color shows the depletion layer near the electrode surface.

One important challenge that faces this type of electrodes is the electrocatalyst deposition. Since the porosities of the electrode are in a smaller scale, the classical paint-spray deposition technique cannot ensure a homogeneous dispersion in the inner electrode area. Hence, new techniques for electrode preparation needs to be further investigated.

### **2. Thermodynamic and kinetic of microfluidic fuel cells**

Microfluidics is a research field in which the studied phenomena occurs in confined structures with a characteristic dimension within 1–1000 µm [32]. At this scale, low Reynolds numbers yielding laminar flows facilitate characterization of the fluid mechanics. Therefore, viscous and capillary forces, rather than volumetric or gravitational effects, govern these systems more importantly. The behavior of the fluids in the microchannels allows us to consider them as a continuum, and thus, we can follow the continuity equation for mass conservation for a fluid in Eq. (1)

$$\frac{\partial \underline{y}}{\partial \underline{\alpha}} + \nabla \left(\rho \vec{u}\right) = 0 \tag{1}$$

If the density of the fluid remains constant, the equation is simplified by the condition of incompressible flow, ∇*u* <sup>→</sup> =0. Then, the Navier-Stokes equation is valid to describe the velocity field, *u* → :

$$
\rho \left( \frac{\partial \vec{u}}{\partial \mathbf{x}} + \vec{u} \nabla \vec{u} \right) = \rho \left( \frac{\partial \vec{u}}{\partial \mathbf{x}} \right) = -\nabla p + \mu \left| \nabla^2 \vec{u} + \vec{f} \right| \tag{2}
$$

with *p* being the pressure and *f* <sup>→</sup> a term summarizing all body forces per unit volume. Since the main difference between PEMFC and MFC systems relies on the replacement a physical membrane with the liquid-liquid interface, their electrochemical operating principles remain almost identical. Due to the large amount of reports making use of organic molecules as fuel, it is convenient to remind the fundamentals of direct alcohols fuel cells (DAFC) [33]. For the sake of understanding the thermodynamics of their process, we will use CH3OH electrooxidation on a Pt surface in acid media as an example. CH3OH can be directly oxidized to carbon dioxide, although other subproducts (namely, formaldehyde, formic acid, etc.) can be produced depending on the reaction pathway.

In an assembled MFC, two reactions occur: (1) the fuel oxidation in the anode side and (2) the electroreduction of the oxidant (oxygen, for an example) at the cathode side. The general overall reaction is presented below:


One important challenge that faces this type of electrodes is the electrocatalyst deposition. Since the porosities of the electrode are in a smaller scale, the classical paint-spray deposition technique cannot ensure a homogeneous dispersion in the inner electrode area. Hence, new

**Figure 1.** Schematic image of the overflow designs: (a) top-bottom electrode configuration and (b) side-by-side parallel configuration. The darker color shows the diffusion interface, while a transparent color shows the depletion layer near

Microfluidics is a research field in which the studied phenomena occurs in confined structures with a characteristic dimension within 1–1000 µm [32]. At this scale, low Reynolds numbers yielding laminar flows facilitate characterization of the fluid mechanics. Therefore, viscous and capillary forces, rather than volumetric or gravitational effects, govern these systems more importantly. The behavior of the fluids in the microchannels allows us to consider them as a continuum, and thus, we can follow the continuity equation for mass conservation for a fluid

<sup>0</sup> ( ) *<sup>y</sup> <sup>u</sup>*

¶ +Ñ =

<sup>2</sup>*u u u u p uf*

ç ÷ ç÷ + Ñ = = -Ñ + Ñ +

r

If the density of the fluid remains constant, the equation is simplified by the condition of

rm

<sup>→</sup> =0. Then, the Navier-Stokes equation is valid to describe the velocity

r r <sup>r</sup> r r <sup>r</sup> (2)

<sup>→</sup> a term summarizing all body forces per unit volume.

<sup>r</sup> (1)

*x*

¶

*x x*

è ø èø ¶ ¶

æ ö æö ¶ ¶

r

techniques for electrode preparation needs to be further investigated.

276 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

**2. Thermodynamic and kinetic of microfluidic fuel cells**

in Eq. (1)

the electrode surface.

field, *u* → :

incompressible flow, ∇*u*

with *p* being the pressure and *f*

The overall reaction potential can be obtained by the subtraction of the anodic thermodynamic potential reaction to the cathodic value, *E*c – *E*a = *E*cell = 1.19 V. This process has an associated thermodynamic efficiency that can be calculated by obtaining the ratio of Gibbs free energy (the maximum electrical work of the system, *∆G°*) between the enthalpy (the total available energy, *∆H°*),

$$
\varepsilon\_{\rm eq}^{\rm nv} = \begin{array}{c} -\Delta G^{\rm \circ} \end{array} / \begin{array}{c} -\Delta H^{\rm \circ} \end{array} \tag{3}
$$

Methanol has a theoretical enthalpy of combustion of –715 kJ/mol [34]. With the standard electromotive force equation, it is possible to calculate the *∆G°* at equilibrium

$$
\Delta G^{\circ} = -nF \times E\_{\text{eq}}^{\circ} = -6 \times 96,485 \times 1.19 = -688.9 \text{ kJ/mol} \tag{4}
$$

where *F* is the Faraday constant (96,485 C/mol) and *n* is the number of exchanged electrons for the electrooxidation of CH3OH. Then, the theoretical efficiency of the system is given by Eq. (5):

$$
\mathcal{L}\_{\rm eq}^{\rm rev} = -\Delta G^{\rm \circ} \;/ -\Delta H^{\rm \circ} = \;688.9 \;/\, 715 = 96.3\% \tag{5}
$$

In a real experimentation, the formation of intermediates and subproducts determines in great measure the performance of the cell. These species depend on associated factors, but not limited to the reactive surface, catalyst composition, crystalline structure, and electrode potential. Other conditions, such as pH, electrolyte temperature, fuel concentration, and reaction time/flow speed, also have to be considered [35].

The differences encountered between the theoretical and actual values for *E*cell are usually addressed to three limiting factors: (1) charge transfer overpotentials (*η*act) in either the anodic or cathodic catalyst due to slow kinetics, (2) the ohmic drop (*Rej*, resulting from the product of the electrolyte resistance, *Re*, and the recorded current density, *j*), and (3) mass transfer constrains [33]. Adding these new terms to the *E*cell equation, the resulting expression is Eq. (6):

$$E\_{\rm cell} = E\_{\rm eq}^{\prime} - \eta\_a - \eta\_c - R\_\epsilon j \tag{6}$$

With *ηa* being the anode overpotential and *ηc* the cathode overpotential, this equation considers both kinetic and mass transfer limitations. As previously mentioned, the specific reaction pathway might lead to incomplete fuel oxidation, and thus, the reaction will have a different number of exchanged electrons, *η*real. If these considerations are included to the cell efficiency expression, we have Eq. (7):

$$\begin{aligned} \left. \varepsilon\_{cell}^{CH\_3OH/O\_2} = -\frac{n \ F \ E\_{cell}}{\Delta H^o} = -\left(\frac{n\_{real} \ F}{\Delta H^o}\right) \left(\frac{n}{n}\right) \right| \frac{E\_{aq}^o}{E\_{aq}^o} = \\\ \left. \left(\frac{n\_{real}}{n}\right) \right| \left. \frac{E\_{cell}}{E\_{aq}^o} \right| \left(\frac{\Delta G^o}{\Delta H^o}\right) = \varepsilon\_F \varepsilon\_E \varepsilon\_{aq}^{rev} \end{aligned} \tag{7}$$

where *εF* is related to oxidation reaction efficiency and *εE* involves the kinetics and ohmic drop limitations. From this last expression, it is easy to understand that the only possible way to improve the cell efficiency is by decreasing the overpotentials and/or ohmic drop, and promoting the complete fuel oxidation [34].

For the ohmic loss, a factor that primarily contributes is the electrode separation (*d*) and the geometric exposed area (*A*). Thus, by increasing the ratio, *A/d*, ohmic losses are minimized. Nevertheless, ionic force in the electrolyte is significant to ohmic drop, and increasing the concentration of the supporting electrolyte reduces its negative impact. Mass transfer limita‐ tions are related to the availability of the species over the electrodic area. Then, the main issues are presented in the electrode vicinity due to the low concentrations as a consequence of the consumption of reactant species, which leads to the thickening of a depletion layer. This layer becomes a limiting mass transfer factor when fresh reactants are not efficiently replenished on the reactive surface, mainly due to slow flow rates [5]. The distributions of the species concentration are controlled by convective (stream flow) and diffusive (concentration gradi‐ ent) transport, which are described by the mass conservation equation in Eq. (8):

$$\nabla \left( -\mathbf{D}\_i \nabla \mathbf{C}\_i + \mathbf{C}\_i \vec{\mu} \right) = \mathbf{S}\_i \tag{8}$$

with *Di* the diffusion coefficient for the specie "*i*,*" i* being either the fuel or oxidant depending on the cell compartment, *Ci* the concentration of the specie "*i*," and *Si* the volumetric net rate of consumption for the specie "*i*" due to its corresponding electrochemical (oxidation or reduction) reaction. Then, *Si* is estimated directly by the current density (*i*) obtained from the corresponding electrochemical reaction and given by the Butler-Volmer in Eq. (9):

$$S\_i = \frac{\dot{\mathbf{r}}}{nF} = \frac{\dot{\mathbf{r}}\_0}{nF} \left(\frac{\mathbf{C}\_i}{\mathbf{C}\_i^\*}\right) \left[e^{a\_s f \eta} - e^{a\_c f \eta}\right] \tag{9}$$

where *io* is the exchange current, *Ci \** is the bulk concentration of specie *i*, *αa* and *αa* are the charge transfer coefficients for anodic and cathodic reactions, and *η* is the electrode overpotential. Since *Si* depends on convective effects, the efficiency of the cell can be determined dividing by the flow rate:

$$
\varepsilon\_{fuel} = \frac{\dot{i}}{nFv} \tag{10}
$$

where *ε*fuel is the fuel efficiency utilization and *v* is the fuel flow rate. Hence, higher flow speeds will supply the electrodes with electrolyte at the initial concentration preventing the increase of products in the depletion layer but decreasing *ε*fuel.

### **3. Evolution of microfluidic fuel cell designs**

The evolution of the microfluidic fuel cell is related to the pursuit of more power in less volume. Many factors have been considered including size, weight, physical resistance, and lifetime [14, 21]. Lightweight structural materials are preferred such as polymethyl methacrylate (PMMA) and poly-dimethylsiloxane (PDMS) [13, 14, 21, 36]. Moreover, UV-sensitive resists, such as SU-8, have demonstrated to be excellent candidates to build MFC pieces since they present some important advantages like the flexibility and capability to be sealed to each other by a hot-pressing technique [37].

#### **3.1. Microfabrication techniques**

potential. Other conditions, such as pH, electrolyte temperature, fuel concentration, and

The differences encountered between the theoretical and actual values for *E*cell are usually addressed to three limiting factors: (1) charge transfer overpotentials (*η*act) in either the anodic or cathodic catalyst due to slow kinetics, (2) the ohmic drop (*Rej*, resulting from the product of the electrolyte resistance, *Re*, and the recorded current density, *j*), and (3) mass transfer constrains [33]. Adding these new terms to the *E*cell equation, the resulting expression is Eq. (6):

> cell eq ––– *ac e E E Rj* h h

> > Δ Δ

*o real cell rev o o F E eq*

ent) transport, which are described by the mass conservation equation in Eq. (8):

*CH OH O eq cell real cell o oo*

> Δ Δ

With *ηa* being the anode overpotential and *ηc* the cathode overpotential, this equation considers both kinetic and mass transfer limitations. As previously mentioned, the specific reaction pathway might lead to incomplete fuel oxidation, and thus, the reaction will have a different number of exchanged electrons, *η*real. If these considerations are included to the cell efficiency

> *nFE n F n E H HE n*

æ ö æ öæ ö =- =- = ç ÷ ç ÷ç ÷ç ÷ è øè øè ø

eee

where *εF* is related to oxidation reaction efficiency and *εE* involves the kinetics and ohmic drop limitations. From this last expression, it is easy to understand that the only possible way to improve the cell efficiency is by decreasing the overpotentials and/or ohmic drop, and

For the ohmic loss, a factor that primarily contributes is the electrode separation (*d*) and the geometric exposed area (*A*). Thus, by increasing the ratio, *A/d*, ohmic losses are minimized. Nevertheless, ionic force in the electrolyte is significant to ohmic drop, and increasing the concentration of the supporting electrolyte reduces its negative impact. Mass transfer limita‐ tions are related to the availability of the species over the electrodic area. Then, the main issues are presented in the electrode vicinity due to the low concentrations as a consequence of the consumption of reactant species, which leads to the thickening of a depletion layer. This layer becomes a limiting mass transfer factor when fresh reactants are not efficiently replenished on the reactive surface, mainly due to slow flow rates [5]. The distributions of the species concentration are controlled by convective (stream flow) and diffusive (concentration gradi‐

° = (6)

*o*

*eq*

Ñ- Ñ + = ( *D C Cu S ii i i* ) <sup>r</sup> (8)

(7)

reaction time/flow speed, also have to be considered [35].

278 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

3 2 /

e

promoting the complete fuel oxidation [34].

*eq*

æ ö æ ö æ ö = = ç ÷ ç ÷ ç ÷ ç ÷ è ø è ø è ø

*n E G n E H*

expression, we have Eq. (7):

In order to obtain more energy in the same volume, more sophisticated micromachining techniques have been employed. Some techniques even employ clean room facilities. The principal considerations for the choice of one or another micromachining technique are the selection of the structural material, the size of the microstructures, and the complexity of the structures. The polymers can be machined by Xerography, micromolding, hot embossing, and micromilling. In addition, assisted-laser techniques can be employed including photolithog‐ raphy, laser ablation, or stereolithography.

Micromilling is a very simple technique performed with a CNC micromilling cutter, useful for rigid polymers (i.e., PMMA). Micromilling is limited by the size of the cutter, it is uncommon to find below the 1/64 and often very fragile. Also, flexible polymers cannot be machined using this technique.

Xurography is a fast and low-cost process. It uses a plotter cutter to draw the channels in a protective mask placed over the substrate, later a conductive layer is placed over the substrate. Finally, the mask is removed leaving the conductive layer printed in the substrate. This technique is especially useful for electrodes and microchannels [38] (see Figure 2).

 **Figure 2.** Schematics for electrodes on glass process flow based on xurography technique. (a) Pattering of vinyl film (xurography). (b) Removing of film and transfer on glass slide surface. (c) 30 nm Cr and 300 nm Au deposition. (d) Removing of film (dry liftoff) (taken from [39]).

Micromolding and hot embossing are similar techniques. A mold is used, however in micro‐ molding a liquid resin is placed in a mold with the desire shape and is dried at room temper‐ ature by heat. In hot embossing, a solid polymer is heated to soften the material and then thermoformed by heat and pressure. Usually, these kinds of techniques are popular because of their simplicity. In both cases, the mold fabrication can be performed by micromilling or a laser technique.

Laser is a versatile tool for the construction of microstructures. Laser is used with a positive or a negative photoresin in order to obtain the desired pattern in the polymer. Laser techniques require the use of a clean room. In the photolithography, a UV laser beam is used to draw a pattern in a positive photoresin, generally a photoresist mask is used to protect the UVsensitive material. Later, the sample is washed in order to remove the residues. Other alter‐ native is the fabrication of a mold in negative photoresin, and later performed the casting of a polymer shell, as PDMS [13].

The technique of laser ablation allows drawing a microchannel from inside of a transparent polymer. A nanosecond or femtosecond pulsed laser is used to generate a three-dimensional pattern since the interior of material avoiding the need to paste several layers in the same piece (Figure 3) [39].

Stereolithography (SLA) is an additive technique (Figure 3). A negative photoresin (as SU-8) is placed over the support and cured with a laser beam, making a solid polymer with the desired structure. The main disadvantage of this procedure is the high cost of the machine (several hundred of dollars) [40].

**Figure 3.** Destructive (laser ablation) and constructive (stereolithography): two different principles of microfabrication techniques. In ablation, the substrate is destroyed in order to obtain the pattern. In the stereolithography, the substrate is formed.

#### **3.2. Microfluidic fuel cells two-dimensional electrode**

Micromilling is a very simple technique performed with a CNC micromilling cutter, useful for rigid polymers (i.e., PMMA). Micromilling is limited by the size of the cutter, it is uncommon to find below the 1/64 and often very fragile. Also, flexible polymers cannot be machined using

Xurography is a fast and low-cost process. It uses a plotter cutter to draw the channels in a protective mask placed over the substrate, later a conductive layer is placed over the substrate. Finally, the mask is removed leaving the conductive layer printed in the substrate. This

**Figure 2.** Schematics for electrodes on glass process flow based on xurography technique. (a) Pattering of vinyl film (xurography). (b) Removing of film and transfer on glass slide surface. (c) 30 nm Cr and 300 nm Au deposition. (d)

Micromolding and hot embossing are similar techniques. A mold is used, however in micro‐ molding a liquid resin is placed in a mold with the desire shape and is dried at room temper‐ ature by heat. In hot embossing, a solid polymer is heated to soften the material and then thermoformed by heat and pressure. Usually, these kinds of techniques are popular because of their simplicity. In both cases, the mold fabrication can be performed by micromilling or a

Laser is a versatile tool for the construction of microstructures. Laser is used with a positive or a negative photoresin in order to obtain the desired pattern in the polymer. Laser techniques require the use of a clean room. In the photolithography, a UV laser beam is used to draw a pattern in a positive photoresin, generally a photoresist mask is used to protect the UVsensitive material. Later, the sample is washed in order to remove the residues. Other alter‐ native is the fabrication of a mold in negative photoresin, and later performed the casting of a

technique is especially useful for electrodes and microchannels [38] (see Figure 2).

280 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

this technique.

polymer shell, as PDMS [13].

laser technique.

Removing of film (dry liftoff) (taken from [39]).

We have seen a fuel cell evolution through the exhaustive analysis of literatures concerning membraneless MFCs. This evolution can be summarized at five principal stages:


The one and two stages correspond to microfluidic systems in which the anodic and cathodic streams are only in contact with the electrode surface (long and wide, Figure 4, delimited by blue arrows). The reason of this design is to consider membraneless MFCs with two-dimen‐ sional electrodes. In addition, because of fuel and oxidant flow laterally to electrodes, these are called as MFCs with flow-over electrodes. All devices that operate under this condition (lateral flow) and use small organic molecules (i.e., formic acid, methanol, ethanol, glycerol, etc.) as fuels and oxygen as oxidant have shown low-power densities, where an important limiting factor is the reduced geometrical area. With this in mind, plenty of investigations have

**Figure 4.** Contact area of fuel and oxidant streams in membraneless microfluidic fuel cells with two-dimensional elec‐ trodes: (a) "Y"-type with electrodes on channel walls [19] and (b) "Y" type with electrodes on channel floor [13].

been focused to increase the activity through enhancing the cell design, and others using supports with high surface area of different nature, such as carbon nanotubes [14, 15] and polyaniline [41]. These supports also enhance the reaction due to the improvement of electron transfer, modifying the metal-support interactions. The low concentration and diffusivity of oxygen in solution are another important parameter that limits the power density of mem‐ braneless MFCs [10]. For this reason, the two-dimensional electrode MFC has been slightly modified using porous air-breathing cathodes (Figure 5). Hence, it profits from the oxygen in air that has 10,000-fold higher diffusivity than in aqueous solution. Also, the oxygen concen‐ tration in air is higher (10 mM) than in solution (2–4 mM) [29].

**Figure 5.** Membraneless microfluidic fuel cell with porous cathode as flow-over electrode: (a) typical "Y" and "T"-like configuration and (b) "F"-like channel configuration.

#### **3.3. Microfluidic fuel cells three-dimensional electrode**

Another strategy to enhance the power density is related to the increase of the contact area between the electrocatalyst and electrocatalytic solution. Carbon paper as porous electrode has been used as diffusor in proton exchange membrane (PEM) fuel for cells many years before the first membraneless MFC was reported [42–45]. Kjeang et al. have explored the use of these porous carbon electrodes in MFC, where fuel and oxidant are circulating through them, emerging the term of "MFCs with flow through electrodes" or "three-dimensional electrode MFCs" (Figure 6) [31, 46]. These devices benefit from the high surface area of carbon paper (230 m2 g−1) and carbon nanofoam (450 m2 g−1) to enhance the cell performance.

been focused to increase the activity through enhancing the cell design, and others using supports with high surface area of different nature, such as carbon nanotubes [14, 15] and polyaniline [41]. These supports also enhance the reaction due to the improvement of electron transfer, modifying the metal-support interactions. The low concentration and diffusivity of oxygen in solution are another important parameter that limits the power density of mem‐ braneless MFCs [10]. For this reason, the two-dimensional electrode MFC has been slightly modified using porous air-breathing cathodes (Figure 5). Hence, it profits from the oxygen in air that has 10,000-fold higher diffusivity than in aqueous solution. Also, the oxygen concen‐

**Figure 4.** Contact area of fuel and oxidant streams in membraneless microfluidic fuel cells with two-dimensional elec‐ trodes: (a) "Y"-type with electrodes on channel walls [19] and (b) "Y" type with electrodes on channel floor [13].

**Figure 5.** Membraneless microfluidic fuel cell with porous cathode as flow-over electrode: (a) typical "Y" and "T"-like

Another strategy to enhance the power density is related to the increase of the contact area between the electrocatalyst and electrocatalytic solution. Carbon paper as porous electrode has been used as diffusor in proton exchange membrane (PEM) fuel for cells many years before the first membraneless MFC was reported [42–45]. Kjeang et al. have explored the use of these

tration in air is higher (10 mM) than in solution (2–4 mM) [29].

282 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

configuration and (b) "F"-like channel configuration.

**3.3. Microfluidic fuel cells three-dimensional electrode**

**Figure 6.** (a) Schematic representation of membraneless microfluidic fuel cell with flow through electrodes and (b) with air-breathing window [48, 49].

The modifications of membraneless microfluidic fuel cells between flow-over electrodes and flow through electrodes as well as the use of air as oxygen source have resulted in enhancement of the cell performance. Figure 7 illustrates the evolution of these devices by changes in contact areas as discussed earlier. Changes were observed in the current density as a function of the MFC stage. The range in the stage: (i) is in the order of 8–12 mA cm−2 and (ii) it depends on the utilized fuel. The use of porous material as the oxygen source (stage ii) in an air breathing resulted in current densities from 15 to 130 mA cm−2. The use of a flow through anode (stage (iii)) increases the current density up to 140 mA cm−2. In the case of an MFC, which works with a flow through anode and cathode, formic acid as fuel and oxygen-saturated solution as oxidant, the reported current density was of 20 mA cm−2 (stage (iv)) [47].

When the MFC with flow-through electrodes combines oxygen from air and from an oxygen‐ ated solution, the current density increases to values up to 100 mA cm−2 as tested for different fuels [48]. In the case of formic acid, current density of almost 500 mA cm−2 has been reported [49]. As summary, the performance of membraneless MFCs has shown high dependence on the way that the fuel and oxidant streams flow in the anode and cathode as well as with strategies to increase/enhance the oxygen source. The use of electrodes with higher surface areas as flow-through electrodes can be considered as an important alternative to improve the cell performance of this kind of devices, which could allow them for use as power supplier for small devices such as cellphones, cameras, etc.

**Figure 7.** Current density evolution on single membraneless microfluidic fuel cells, which employ small organic mole‐ cules as fuel as function of contact form of fuel and oxidant streams to electrode surfaces (references used are listed from left to right) [21, 16, 51, 52, 12, 53, 54, 13, 55–57, 8, 58–61].

### **4. Fuels and performance evolution**

Several research groups have worked on the development of such energy conversion devices that operate on various fuels such as glycerol, ethylene glycol, formic acid, glucose, ethanol, and methanol. Many designs have been optimized, as well as manufacturing techniques incorporating computational tools for modeling and construction photolithography equip‐ ment and numerical control systems, high-resolution components and electronic interconnect assembly, and development of electrocatalytic anodic and cathodic materials of inorganic and enzymatic nature.

In this context, we highlight the bimetallic core-shell materials (core-shell) used as anodes and cathodes for the oxidation of various fuels and the reduction of oxygen, respectively. For example, AuAg supported on nanoparticulated carbon was used for glucose electrooxidation in alkaline media [16, 50]. AuAg/C exhibited good stability, higher current density, and more negative anodic potential (ca. 150 mV) associated with this reaction than that achieved with bare Au supported on Vulcan. The performance of AuAg/C and Au/C was evaluated in a glucose laminar membraneless microfluidic fuel cell where the incorporation of Ag in the electrocatalyst contributed to increasing the AuAg/C activity using low Au loading.

With regard to recent development of cathodic electrocatalytic materials, stand-up PtAg/C alloys were employed for the oxygen reduction reaction (ORR). High selectivity for ORR was observed in the presence of glucose (100 mM) in basic media [51]. This result allowed the use of these bimetallic materials as cathode in an MFC that used glucose as fuel with an additional advantage arising from the use of less noble metal and less expensive material such as silver.

The combination of high selectivity of PtAg/C in the ORR and AuAg/C in the glucose oxidation reaction allowed the construction and evaluation of a glucose MFC with outstanding per‐ formance at zero flow condition [52]. MFC was also evaluated with a simulated body fluid solution that contained salts commonly present in the human blood plasma, reaching a power of 240 mW cm−2 at zero flow. These results envisage the incorporation of this MFC as a portable power source in lab-on-a-chip devices without the need of external pumps.

Dector and coworkers developed a small air-breathing hybrid MFC using glucose from human serum and blood as a fuel with O2 supplied from air [53]. This device showed an excellent performance (200 mW cm−2) and good stability with potential applications as power source for use in building implantable and portable biomedical devices.

A new concept of membraneless air-breathing nanofluidic fuel cells with flow-through electrodes that can be powered by several fuels (individually or mixed) in alkaline media (i.e., methanol, ethanol, glycerol, and ethylene glycol) was recently reported [48]. A novel Cu@Pd core-shell electrocatalyst synthesized by chemical method revealed morphologies of nano‐ rods particles of 10 nm in length and a typical semispherical particle with a diameter of 6 nm. The electrocatalytic properties of the rod/semispherical Cu@Pd nanoparticles supported on Vulcan carbon showed higher activity than commercial Pd/C in the electro-oxidation of each fuel and the mixture of them. The air-breathing nanofluidic fuel cells constructed with Cu@Pd/ C anode exhibited individual power densities and current densities higher than 16 mW cm−2 and100mAcm−2,respectively.Amixtureoffourfuels resultedinavoltageof 0.61Vwithcurrent and power densities of 108 mA cm−2 and 18 mW cm−2, respectively. The reported devices displayed a better performance than current state-of-the-art systems and demonstrated the feasibility of developing multiuse nanodevices that provide constant power regardless of the fuel employed.

Finally, ethylene glycol has been used as fuel in PEM (polymer exchange membrane) fuel cells operating at temperatures ranging from 60 to 130°C [54, 55]. In microfluidic fuel cells, the use of ethylene glycol (EG) as well as glycerol has been sparse, and it could be related to the low performance by activity problems of electrocatalysts that are still found in PEM fuel cells. Arjona et al. [56] published, for the first time, the use of ethylene glycol in a membraneless MFC with lateral flow. In this work, a metal mixture of AuPd supported on polyaniline was used as anode. The performance device was tested as a function of EG concentration showing a maximum performance at 2 M EG and room temperature. A cell voltage of 0.53 V and a maximum current and power density of 14.2 mA cm−2 and 1.6 mW cm−2, respectively, were achieved using oxygen from a saturated solution. The resulting cell voltage was comparable with those values reported for PEM fuel cells (0.4–0.65 V), which operate at higher tempera‐ tures (60–90°C) [54, 55]. The finding was attributed to the enhancement in the electronic properties by the metal-metal AuPd interactions as well as the effect of polyaniline, which could enhance the metal/support interactions.

## **5. Conclusions and perspectives**

**Figure 7.** Current density evolution on single membraneless microfluidic fuel cells, which employ small organic mole‐ cules as fuel as function of contact form of fuel and oxidant streams to electrode surfaces (references used are listed

Several research groups have worked on the development of such energy conversion devices that operate on various fuels such as glycerol, ethylene glycol, formic acid, glucose, ethanol, and methanol. Many designs have been optimized, as well as manufacturing techniques incorporating computational tools for modeling and construction photolithography equip‐ ment and numerical control systems, high-resolution components and electronic interconnect assembly, and development of electrocatalytic anodic and cathodic materials of inorganic and

In this context, we highlight the bimetallic core-shell materials (core-shell) used as anodes and cathodes for the oxidation of various fuels and the reduction of oxygen, respectively. For example, AuAg supported on nanoparticulated carbon was used for glucose electrooxidation in alkaline media [16, 50]. AuAg/C exhibited good stability, higher current density, and more negative anodic potential (ca. 150 mV) associated with this reaction than that achieved with bare Au supported on Vulcan. The performance of AuAg/C and Au/C was evaluated in a glucose laminar membraneless microfluidic fuel cell where the incorporation of Ag in the

With regard to recent development of cathodic electrocatalytic materials, stand-up PtAg/C alloys were employed for the oxygen reduction reaction (ORR). High selectivity for ORR was observed in the presence of glucose (100 mM) in basic media [51]. This result allowed the use of these bimetallic materials as cathode in an MFC that used glucose as fuel with an additional advantage arising from the use of less noble metal and less expensive material such as silver. The combination of high selectivity of PtAg/C in the ORR and AuAg/C in the glucose oxidation reaction allowed the construction and evaluation of a glucose MFC with outstanding per‐

electrocatalyst contributed to increasing the AuAg/C activity using low Au loading.

from left to right) [21, 16, 51, 52, 12, 53, 54, 13, 55–57, 8, 58–61].

284 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

**4. Fuels and performance evolution**

enzymatic nature.

Many reports published in the last decades have contributed to several findings in the MFC technological area. As discussed, numerous issues relevant to MFCs need further investigation and optimization.

The current trend is to develop devices that combine all three properties: power generation, energy storage, and electronic interface for a specific application. These three functions integrated in a single device with a dimension in the micro- or nanometrics have placed the technology in the category of lab-on-a-chip.

Some limiting conditions, such as potential drop and ohmic resistance, have been observed in the assembly of these MFCs. All electrical contacts and external circuit wiring are necessary to be optimized in order to reduce their impact on the MFC performance. The new machining technologies, such as three-dimensional printing, offer a precise and accurate option to develop the cell assembly parts.

In terms of electrochemistry, charge transfer is also a problem that affects many systems to date. These phenomena can be studied by properly implementing a reference electrode to provide a more detailed insight into the electrochemical processes occurring in each compart‐ ment. The sluggish kinetics is usually related to the electrocatalytic material and proper pH selection. However, temperature, fuel concentration, and other physicochemical parameters are factors to investigate with each new emerging cell design.

The development of these devices involves a multidisciplinary task in electrochemistry, nanotechnology, electronics, physics, mathematics, mechanical engineering, computer sciences, biotechnology, biomedical, and environmental engineering.

## **Acknowledgements**

The authors thanks to Mexican Council of Science and Technology CONACYT for the financial support through the projects Laboratorio Nacional (Grant LN271649) and Fronteras de la Ciencia (Grant 611).

## **Author details**

Jesus A. Diaz-Real1 , Minerva Guerra-Balcázar2 , Noe Arjona1 , Francisco Cuevas-Muñiz1 , Luis Gerardo Arriaga1 and Janet Ledesma-García1\*

\*Address all correspondence to: janet.ledesma@uaq.mx

1 Center Research and Technological Development in Electrochemistry, Parque Tecnologico Queretaro Sanfandila, Queretaro, Mexico

2 Division of Research and Postgraduate, Facultad de Ingenieria, Universidad Autonoma de Queretaro, Cerro de las Campanas, Queretaro, Mexico

## **References**

The current trend is to develop devices that combine all three properties: power generation, energy storage, and electronic interface for a specific application. These three functions integrated in a single device with a dimension in the micro- or nanometrics have placed the

Some limiting conditions, such as potential drop and ohmic resistance, have been observed in the assembly of these MFCs. All electrical contacts and external circuit wiring are necessary to be optimized in order to reduce their impact on the MFC performance. The new machining technologies, such as three-dimensional printing, offer a precise and accurate option to develop

In terms of electrochemistry, charge transfer is also a problem that affects many systems to date. These phenomena can be studied by properly implementing a reference electrode to provide a more detailed insight into the electrochemical processes occurring in each compart‐ ment. The sluggish kinetics is usually related to the electrocatalytic material and proper pH selection. However, temperature, fuel concentration, and other physicochemical parameters

The development of these devices involves a multidisciplinary task in electrochemistry, nanotechnology, electronics, physics, mathematics, mechanical engineering, computer

The authors thanks to Mexican Council of Science and Technology CONACYT for the financial support through the projects Laboratorio Nacional (Grant LN271649) and Fronteras de la

1 Center Research and Technological Development in Electrochemistry, Parque Tecnologico

2 Division of Research and Postgraduate, Facultad de Ingenieria, Universidad Autonoma de

, Noe Arjona1

, Francisco Cuevas-Muñiz1

,

technology in the category of lab-on-a-chip.

are factors to investigate with each new emerging cell design.

286 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

sciences, biotechnology, biomedical, and environmental engineering.

, Minerva Guerra-Balcázar2

\*Address all correspondence to: janet.ledesma@uaq.mx

Queretaro, Cerro de las Campanas, Queretaro, Mexico

Queretaro Sanfandila, Queretaro, Mexico

and Janet Ledesma-García1\*

the cell assembly parts.

**Acknowledgements**

Ciencia (Grant 611).

**Author details**

Jesus A. Diaz-Real1

Luis Gerardo Arriaga1


[26] Moore S, Sinton D, Erickson D, A plate-frame flow-through microfluidic fuel cell stack. J. Power Sources 2011;196:9481–9487. doi:10.1016/j.jpowsour.2011.07.024

[14] Morales-Acosta D, Rodríguez GH, Godinez LA, Arriaga LG, Performance increase of microfluidic formic acid fuel cell using Pd/MWCNTs as catalyst. J. Power Sources

[15] Dector A, Cuevas-Muñiz FM, Guerra-Balcázar M, Godínez LA, Ledesma-García J, Arriaga LG, Glycerol oxidation in a microfluidic fuel cell using Pd/C and Pd/ MWCNT anodes electrodes. Int. J. Hydrogen Energy 2013;38:12617–12622. doi:

[16] Cuevas-Muñiz FM, Guerra-Balcázar M, Castaneda F, Ledesma-García J, Arriaga LG, Performance of Au and AuAg nanoparticles supported on Vulcan in a glucose lami‐ nar membraneless microfuel cell. J. Power Sources 2011;196: 5853–5857. doi:10.1016/

[17] Galindo R, Dector A, Arriaga LG, Gutiérrez S, Herrasti P, Maghemite as a catalyst for glucose oxidation in a microfluidic fuel cell. J. Electroanal. Chem. 2012;671:38–43. doi:

[18] Cohen JL, Volpe DJ, Westly DA, Pechenik A, Abruña HD, A Dual Electrolyte H2/O2 Planar Membraneless Microchannel Fuel Cell System with Open Circuit Potentials in

[19] Ferrigno R, Stroock AD, Clark TD, Mayer M, Whitesides GM. Membraneless Vanadi‐ um Redox Fuel Cell Using Laminar Flow. J. Am. Chem. Soc. 2002;124:12930–12931.

[20] Lee JW, Kjeang E, Nanofluidic fuel cell. J. Power Sources 2013;242:472–477. doi:

[21] Choban ER, Spendelow JS, Gancs L, Wieckowski A, Kenis PJA, Membraneless lami‐ nar flow-based micro fuel cells operating in alkaline, acidic, and acidic/alkaline me‐

[22] Gonzalez-Guerrero MJ, Esquivel JP, Sánchez-Molas, Godignon P, Muñoz FX, Del Campo FJ, Giroud F, Minteer SD, Sabaté N, Membraneless glucose/O2 microfluidic enzymatic biofuel cell using pyrolyzed photoresist film electrodes. Lab Chip

[23] Hasegawa S, Shimotani K, Kishi K, Watanabe H, Electricity Generation from Decom‐ position of Hydrogen Peroxide. Electrochem. Solid-State Lett. 2005;8: A119–A121.

[24] Kjeang E, Brolo AG, Harrington DA, Djilali N, Sinton D, Hydrogen Peroxide as an Oxidant for Microfluidic Fuel Cells. J. Electrochem. Soc. 2007;154:B1220–B1226. doi:

[25] Kjeang E, Proctor BT, Brolo AG, Harrington DA, Djilali N, Sinton D, High-perform‐ ance microfluidic vanadium redox fuel cell. Electrochim. Acta 2007;52:4942–4946.

Excess of 1.4 V. Langmuir 2005;21:3544–3550. doi: 10.1021/la0479307

2010;195:1862–1865. doi:10.1016/j.jpowsour.2009.10.007

288 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

10.1016/j.ijhydene.2012.12.030

10.1016/j.jelechem.2012.02.020

j.jpowsour.2011.02.081

doi: 10.1021/ja020812q

doi:10.1149/1.1849112

10.1149/1.2784185

doi:10.1016/j.electacta.2007.01.062

10.1016/j.jpowsour.2013.05.129

dia. Electrochim. Acta 2005;50:5390–5398

2013;13:2972–2979. doi: 10.1039/C3LC50319D


[52] Arjona N, Dector A, Guerra-Balcázar M, Álvarez-Contreras L, Sabaté N, Esquivel JP, Ledesma-García J, Arriaga LG, Effect of metal content on the electrocatalytic activity of AuxPdy mixtures and their use in a glucose membraneless microfluidic fuel cell, RSC Adv. 2014;4:26158–26165. doi: 10.1039/C4RA03141E

[40] Wong KV, Hernandez A, A Review of Additive Manufacturing. ISRN Mechanical Engineering 2012;2012:208760–208770. http://dx.doi.org/10.5402/2012/208760

[41] Guerra-Balcázar M, Morales-Acosta D, Castaneda F, Ledesma-García J, Arriaga LG, Synthesis of Au/C and Au/Pani for anode electrodes in glucose microfluidic fuel cell,

[42] Miachon S, Aldebert P, Internal hidration H2/O2 100 cm2 polymer electrolyte mem‐ brane fuel cell, J. Power Sources 1995;56:31–36. doi:10.1016/0378-7753(95)80005-2 [43] Prater KB, Polymer electrolyte fuel cells: a review of recent developments, J. Power

[44] Lee WK, Ho CH, Van Zee JW, Murthy M, The effects of compression and gas diffu‐ sion layers on the performance of a PEM fuel cell, J. Power Sources 1999;84: 45-51.

[45] Glora M, Wiener M, Petricevic R, Pröbstle H, Fricke J, Integration of carbon aerogels in PEM fuel cells, J. Non-Cryst. Solids, 2001;285:283–287. doi:10.1016/

[46] Kjeang E, McKechnie J, Sinton D, Djilali N, Planar and three-dimensional microflui‐ dic fuel cell architectures based on graphite rod electrodes, J. Power Sources

[47] Arjona N, Goulet MA, Guerra-Balcázar M, Ledesma-García J, Kjeang E, Arriaga LG, Direct formic acid Microfluidic fuel cell with Pd nanocubes supported on flowthrough microporous electrodes, ECS Electrochem. Lett. 2015;4:F24–F28. doi:

[48] Maya-Cornejo J, Ortiz-Ortega E, Álvarez-Contreras L, Arjona N, Guerra-Balcázar M, Ledesma-García J, Arriaga LG, Copper-palladium core-shell as an anode in a multifuel membraneless nanofluidic fuel cell: toward a new era of small energy conver‐

[49] Ortiz-Ortega E, Goulet MA, Lee JW, Guerra-Balcázar M, Arjona N, Kjeang E, Ledes‐ ma-García J, Arriaga LG, A nanofluidic direct formic acid fuel cell with a combined flow-through and air-breathing electrode for high performance, Lab Chip

[50] Cuevas-Muñiz FM, Guerra-Balcázar M, Esquivel JP, Sabaté N, Arriaga LG, Ledesma-García J, Glucose microfluidic fuel cell based on silver bimetallic selective catalysts for on-chip applications, J. Power Sources 2012;216:297–303. doi:10.1016/j.jpowsour.

[51] Guerra-Balcázar M, Cuevas-Muñiz FM, Álvarez-Contreras L, Ledesma-García J, Ar‐ riaga LG, Evaluation of bimetallic catalyst PtAg/C as a glucose-tolerant oxygen re‐ duction cathode, J. Power Sources 2012;197:121–128. doi:10.1016/j.jpowsour.

sion devices, Chem. Commun. 2015;51:2536–2539. doi: 10.1039/C4CC08529A

Electrochem. Commun. 2010;12:864–867. doi:10.1016/j.elecom.2010.04.009

Sources 1994;51:129–144. doi:10.1016/0378-7753(94)01934-7

290 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

2007;168:379–390. doi:10.1016/j.jpowsour.2007.02.087

2014;14:4596–4598. doi:10.1039/C4LC01010H

doi:10.1016/S0378-7753(99)00298-0

S0022-3093(01)00468-9

10.1149/2.0031504eel

2012.05.101

2011.09.051


## **Microfluidics in CO2 Capture, Sequestration, and Applications**

## Taotao Fu

[63] Ma J, Gago AS, Alonso-Vante N, Performance study of platinum nanoparticles sup‐ ported onto MWCNT in a formic acid microfluidic fuel cell system, J. Electrochem.

[64] Mousavi Shaegh SA, Nguyen NT, Chan SH, Zhou W, Air-breathing membraneless laminar flow-based fuel cell with flow-through anode, Int. J. Hydrogen Energy

[65] Mousavi Shaegh SA, Nguyen NT, Chan SH, Air-breathing microfluidic fuel cell with fuel reservoir, J. Power Sources 2012;209:312–317. doi:10.1016/j.jpowsour.2012.02.115

Soc. 2013;160:F859–F866; doi:10.1149/2.101308jes

292 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

2012;37:3466-3476. doi:10.1016/j.ijhydene.2011.11.051

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/64284

#### **Abstract**

The abnormal climate change has made the reduction of CO2 emission that received worldwide attention. The integration of CO2 capture-sequestration application for en‐ hanced oil recovery (EOR) technology will be the new trend. Several scholars have ap‐ plied microfluidics in CO2 capture, oil and gas analysis, and CO2 sequestration. The mass transfer process for CO2 capture can be intensified owing to the large specific surface/ volume ratio and high contact area in microchannels. The small amount of feeding vol‐ umes of oil and gas samples and the quick response for the analysis make the microflui‐ dics a promising tool for the oil and gas analysis. Moreover, microfluidics can reveal the transport mechanism at microscale for multiphase interfacial phenomena in microchan‐ nels within porous media during the CO2 flooding process in line with the pressure, tem‐ perature, and material properties of the rock within the oil reservoir. This chapter will elaborate the progress of the application of microfluidic technology in the utilization of CO2, including the mechanism of mass transfer for CO2 in microreactors, the advantages of microfluidics in oil and gas analysis, and the fundamentals of microfluidics in CO2 flooding, oil recovery improvement, and CO2 sequestration.

**Keywords:** CO2, fluid dynamics, interfacial phenomena, mass transfer, microfluidics, multiphase flow

### **1. Introduction**

#### **1.1. Progress of the intensification and mechanism of mass transfer for CO2 capture in microreactors**

Microreactors consist of microchannels with dimensions of several hundreds of micrometers. The pressure gradient, temperature gradient, and concentration gradient augment rapidly with the decrease of the length scale in microreactors, leading to the increase of the driving forces in mass transfer and heat transfer. With the scale down to micrometers, the surface/volume ratio

© 2016 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

is enhanced usually to 10,000–50,000 m2 /m3 (100–1000 m2 /m3 in conventional equipment in chemicalengineeringprocesses),whichisbeneficialforthe intensificationofmass/heattransfer. The volume of the microreactors is significantly reduced also with the decrease of the length scale. Furthermore, the numbering-up strategy for the scale-up of microreactors is beneficial for the reactions with heat and explosion. The product properties can also be improved: strengthening of the transfer process can effectively improve the conversion, selectivity, and conversion rate of the product in the microreaction system. In addition, the microreactor is beneficial for the preparation of polymer particles, microcapsule and microemulsions due to the well-controlled property of the formation, structure, and composition of the polymer and the multiphase systems by using microreactors [1, 2]. The mass transfer process for CO2 capture in microreactors is associated with the interfacial surface area between the CO2 and the flowing liquid, contact method,reactortype and geometry, and flowing conditions as well as the liquid and gas properties [3]. Slug flow is found to be the most widely used flow pattern for CO2 in flowing liquids in microchannels among otherflow patterns such as bubbly flow, annularflow, and parallel flow. This section will review the fluid dynamics for gas-liquid two-phase flow in microchannels and the mass transfer mechanism for CO2 capture in microreactors.

#### **1.2. Pressure drop for gas-liquid two-phase flow in microchannels**

The characteristics for the pressure for gas-liquid two-phase flow in microchannels differ significantly from those in conventional large-scale pipes due to the predominated effects of surface tension forces and viscous forces at the microscale in microchannels. The prediction for the pressure drop for gas-liquid two-phase flow in microchannels includes physical model, Lockhart-Martinelli model, and homogeneous model.

The pressure drop in the liquid filled microchannels with slug bubbles Δ*P* includes Δ*PB* along the slug bubble induced by the frictional loss in the liquid film between the bubble and the channel walls, Δ*P*Caps induced by the Laplace pressure across the caps of slug bubbles, and Δ*PF* in the liquid plug. The pressure drop induced by the frictional loss in the liquid film and gutters between the bubble and the channel walls is often neglected in comparison with the other two kinds ofpressuredrops, both for bubbles in surfactant-free liquids andforthe liquidphase with surfactants whose concentration is high enough [4, 5]. However, for intermediate concentra‐ tions of surfactant in the liquid, the pressure decreases rapidly across the body of the bubble owing to the concentration gradient of surfactant along the interface, giving rise to tangential surface tension tractions to immobilize the interface. This would increase the pressure drop across thebubble.Thus,bubbleswithinthe channel caninduce additionalresistance to the flow, and the pressure drop across a single bubble within the microchannel Δ*PB* can be evaluated by using Bretherton expression [6] and Ratulowski and Chang [7] expression, by assuming that Bretherton's law is valid in square and circular cross-section:

$$\Delta P\_{\rm B} = \begin{cases} 3.58 \frac{\sigma}{r} \left( \frac{3 \mu u}{\sigma} \right)^{2/3} & \text{Ca} \le 10^{-2} \\\\ 3.58 \frac{\sigma}{r} \left( \frac{3 \mu u}{\sigma} \right)^{2/3} - 9.07 \frac{\sigma}{r} \left( \frac{3 \mu u}{\sigma} \right)^{0.95} & 10^{-2} \le \text{Ca} \le 10^{-1} \end{cases} \tag{1}$$

where *µ* is the liquid viscosity, *u* is the superficial velocity of the gas-liquid flow, *σ* is the surface tension between the liquid and gas phases, and *r* is the radius of the microchannel. Whereas the whole Laplace pressure across bubbles in microchannels is related to the number of bubbles, which is hard to obtain in industrial processes. Thus, the whole Laplace pressure is always estimated by modifying the frictional factor for the pressure drop for single liquidfilled microchannel.

is enhanced usually to 10,000–50,000 m2

294 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

/m3 (100–1000 m2

chemicalengineeringprocesses),whichisbeneficialforthe intensificationofmass/heattransfer. The volume of the microreactors is significantly reduced also with the decrease of the length scale. Furthermore, the numbering-up strategy for the scale-up of microreactors is beneficial for the reactions with heat and explosion. The product properties can also be improved: strengthening of the transfer process can effectively improve the conversion, selectivity, and conversion rate of the product in the microreaction system. In addition, the microreactor is beneficial for the preparation of polymer particles, microcapsule and microemulsions due to the well-controlled property of the formation, structure, and composition of the polymer and the multiphase systems by using microreactors [1, 2]. The mass transfer process for CO2 capture in microreactors is associated with the interfacial surface area between the CO2 and the flowing liquid, contact method,reactortype and geometry, and flowing conditions as well as the liquid and gas properties [3]. Slug flow is found to be the most widely used flow pattern for CO2 in flowing liquids in microchannels among otherflow patterns such as bubbly flow, annularflow, and parallel flow. This section will review the fluid dynamics for gas-liquid two-phase flow in

microchannels and the mass transfer mechanism for CO2 capture in microreactors.

The characteristics for the pressure for gas-liquid two-phase flow in microchannels differ significantly from those in conventional large-scale pipes due to the predominated effects of surface tension forces and viscous forces at the microscale in microchannels. The prediction for the pressure drop for gas-liquid two-phase flow in microchannels includes physical model,

The pressure drop in the liquid filled microchannels with slug bubbles Δ*P* includes Δ*PB* along the slug bubble induced by the frictional loss in the liquid film between the bubble and the channel walls, Δ*P*Caps induced by the Laplace pressure across the caps of slug bubbles, and Δ*PF* in the liquid plug. The pressure drop induced by the frictional loss in the liquid film and gutters between the bubble and the channel walls is often neglected in comparison with the other two kinds ofpressuredrops, both for bubbles in surfactant-free liquids andforthe liquidphase with surfactants whose concentration is high enough [4, 5]. However, for intermediate concentra‐ tions of surfactant in the liquid, the pressure decreases rapidly across the body of the bubble owing to the concentration gradient of surfactant along the interface, giving rise to tangential surface tension tractions to immobilize the interface. This would increase the pressure drop across thebubble.Thus,bubbleswithinthe channel caninduce additionalresistance to the flow, and the pressure drop across a single bubble within the microchannel Δ*PB* can be evaluated by using Bretherton expression [6] and Ratulowski and Chang [7] expression, by assuming that

**1.2. Pressure drop for gas-liquid two-phase flow in microchannels**

Lockhart-Martinelli model, and homogeneous model.

Bretherton's law is valid in square and circular cross-section:

s m

*B*

*<sup>r</sup> <sup>P</sup>*

<sup>ì</sup> æ ö <sup>ï</sup> ç ÷ £ <sup>ï</sup> è ø D = <sup>í</sup>

2 3 0.95

 sm

ç÷ ç÷ - £ £ ï

*u u*

3 3 3.58 9.07 10 Ca 10

 s

<sup>3</sup> 3.58 Ca 10

2 3

*r r*

î èø èø

ï æö æö

*u*

s

s

sm

2

(1)


2 1


/m3 in conventional equipment in

The frictional pressure between the liquid plug and the channel wall can be estimated by using the Hagen-Poiseuille equation at low Reynolds numbers, in which case the fluid flow in the liquid slug is deemed as laminar flow [8]:

$$
\Delta P\_{\rm F} = f \left( \frac{4}{d\_{\rm H}} \right) \left( \frac{1}{2} \rho u^2 \right) \left( L\_{\rm s} \right) \tag{2}
$$

where *d*H is the equivalent diameter of the microchanel, *ρ* is the liquid density, and *L*S is the length of the liquid plug. It should be pointed out that the pressure drop for gas-liquid twophase flow in the slug bubble flow in microchannels is therefore related to the size and number of bubbles, which can modify the occupation for each part of pressure in the whole pressure drop in microchannels. However, for bubbles flowing in fluids in nontransparent microchan‐ nels during the application, the prediction of pressure drop for gas-liquid two-phase flow becomes difficult by using the physical model, as the lengths of the bubbles and liquid plugs are hard to obtain directly. It should be noted that the volume of a bubble flowing through a microchannel can be varied with the reduction of pressure along the channel. This effect becomes more pronounced in long microchannels. Moreover, the pressure drop of bubbles flowing in fluid-filled microchannels is also affected by the cross-section of the channel. Ratulowski and Chang [7] established an empirical expression to relate the pressure drop across a bubble and the capillary number in a square cross-sectional microchannel as <sup>Δ</sup>*PB* =12.2Ca0.55, when 3 <sup>×</sup> <sup>10</sup>−<sup>3</sup> <Ca<0.1 [7, 9]. Wong et al. [10] extended Bretherton's analysis [6] to rectangular microchannels: Δ*PB* ~ *<sup>σ</sup>* /*r* ~ Ca2/3 . However, the analysis by Wong et al. [10] does not provide an explicit function form for Δ*PB* <sup>~</sup>*σ*Ca2/3 /*r*. Ca (Ca=*µu* / *<sup>σ</sup>*) is the capillary number, representing the ratio of the relative viscous force to surface tension force.

Lockhart and Martinelli proposed the prediction for gas-liquid two-phase flow in microchan‐ nels related to the pressure for a single liquid-filled microchannel by using a friction multiplier of the liquid phase, which can be expressed by the gas and liquid viscosity and density, and the volumetric fraction of the gas phase. In the homogeneous model, the gas and liquid phases are deemed to be mixed uniformly in microchannels. This model is proposed to predict the pressure drop for gas-liquid two-phase flow in conventional channels, with the prerequisite that the gas and liquid phases are mixed uniformly, i.e., the gas phase is dispersed uniformly in the liquid in the form of very tiny bubbles. However, in microchannels, this regime can only be achieved at extreme conditions. That is to say, this model is not valid in most situations in microfluidics applications.

#### **1.3. Mass transfer of gas-liquid two-phase flow in microchannels**

#### *1.3.1. Mass transfer of gas-liquid two-phase flow under various flow patterns*

Themass transfer betweengas andliquidphases inmicrochannelsdominates the flowpatterns, which usually contain bubbly flow, slug flow, churn flow, annular flow, and parallel flow.

#### *1.3.1.1. Bubbly flow*

In this flow pattern, the diameter of bubbles is less than the channel diameter, which usually occurs under low superficial velocity of the gas phase and high superficial velocity of the liquid phase. The diameter of bubbles is manipulated by the gas and liquid flow rates, physical property of the fluids, and how two phases contact. In general, the surface area between gas and liquid phases increases with the reduction of the bubble size, which is beneficial for the enhancement of the mass transfer process.

#### *1.3.1.2. Churn flow*

Churn flow happens at high superficial velocity of the gas phase, within which long gas slugs and short liquid plugs pass through the channel. It is found that the mass transfer coefficient is higher in the churn flow than that in the slug flow. Yue et al. [11] found that the mass transfer coefficient increases with the increase of the gas flow rates and liquid flow rates.

#### *1.3.1.3. Annular flow and parallel flow*

When the superficial velocity of the gas phase is relatively high and that of the liquid phase is quite low, annular flow and parallel flow occurs, depending on the geometry of the contactor for the two-phase flows. The former is likely to happen at flow-focusing junctions, while the latter for T- and Y-junctions usually. In both cases, the gas and liquid streams flow in parallel along the microchannel, and the mass transfer thereby occurs at the interface of the gas-liquid two-phase flow. The experimental results show that in falling film microreactors, the enhance‐ ment of the mass transfer rate in the CO2-monoethanolamine-H2O system can be obtained [12]. However, the Marangoni effect in the falling film microreactor seems to be weaker than that in macrosystems, signifying that the convection partially prevented by the limited size of the microchannel [12].

#### *1.3.1.4. Slug flow*

The mass transfer for slug flow in the gas-liquid two-phase flow in microchannels depends on the flow conditions, physical property of the fluids, the hydrodynamics of the fluids, and the geometry of microchannels. As this flow regime is always achieved at most of microfluidic applications, the following section will mainly focus on this flow regime.

#### *1.3.2. Mass transfer mechanism*

The key issue for mass transfer mechanism for gas-liquid two-phase flow in microchannels is to reveal the mass transfer rule across the gas-liquid interface and the influence of the lengths of the gas slug and liquid plug on the mass transfer coefficient. Bercic et al. [13] investigated the absorption of CH4 by using water in capillaries with diameters ranging from 1.5 to 3.1 mm, and found that the volumetric mass transfer coefficient *k*L*a* is predominated by the length of plugs rather than the length of bubbles when the liquid plugs are sufficient long. In their experiments, the mass transfer around the caps of the slugs dominates the whole process, and the mass transfer coefficient is proposed as

**1.3. Mass transfer of gas-liquid two-phase flow in microchannels**

296 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

*1.3.1.1. Bubbly flow*

*1.3.1.2. Churn flow*

microchannel [12].

*1.3.1.4. Slug flow*

*1.3.2. Mass transfer mechanism*

enhancement of the mass transfer process.

*1.3.1.3. Annular flow and parallel flow*

*1.3.1. Mass transfer of gas-liquid two-phase flow under various flow patterns*

Themass transfer betweengas andliquidphases inmicrochannelsdominates the flowpatterns, which usually contain bubbly flow, slug flow, churn flow, annular flow, and parallel flow.

In this flow pattern, the diameter of bubbles is less than the channel diameter, which usually occurs under low superficial velocity of the gas phase and high superficial velocity of the liquid phase. The diameter of bubbles is manipulated by the gas and liquid flow rates, physical property of the fluids, and how two phases contact. In general, the surface area between gas and liquid phases increases with the reduction of the bubble size, which is beneficial for the

Churn flow happens at high superficial velocity of the gas phase, within which long gas slugs and short liquid plugs pass through the channel. It is found that the mass transfer coefficient is higher in the churn flow than that in the slug flow. Yue et al. [11] found that the mass transfer

When the superficial velocity of the gas phase is relatively high and that of the liquid phase is quite low, annular flow and parallel flow occurs, depending on the geometry of the contactor for the two-phase flows. The former is likely to happen at flow-focusing junctions, while the latter for T- and Y-junctions usually. In both cases, the gas and liquid streams flow in parallel along the microchannel, and the mass transfer thereby occurs at the interface of the gas-liquid two-phase flow. The experimental results show that in falling film microreactors, the enhance‐ ment of the mass transfer rate in the CO2-monoethanolamine-H2O system can be obtained [12]. However, the Marangoni effect in the falling film microreactor seems to be weaker than that in macrosystems, signifying that the convection partially prevented by the limited size of the

The mass transfer for slug flow in the gas-liquid two-phase flow in microchannels depends on the flow conditions, physical property of the fluids, the hydrodynamics of the fluids, and the geometry of microchannels. As this flow regime is always achieved at most of microfluidic

The key issue for mass transfer mechanism for gas-liquid two-phase flow in microchannels is to reveal the mass transfer rule across the gas-liquid interface and the influence of the lengths

applications, the following section will mainly focus on this flow regime.

coefficient increases with the increase of the gas flow rates and liquid flow rates.

$$k\_{\rm L}a = \frac{0.1111u^{1.19}}{\left(\left(1 \cdot \varepsilon\_{\rm G}\right)\left(L\_{\rm B} + L\_{\rm S}\right)\right)^{0.97}}\tag{3}$$

where *L* B, *L* S are, respectively, the length of the bubble and liquid plug. Van Baten et al. [14, 15] divided the mass transfer coefficient *k*L*a* for mass transport process for slug bubbles to the liquid plugs in vertical capillaries into two parts: one between bubbles' caps and plugs *k*L, cap*a*cap, and the other one between cylindrical parts of slug bubbles and the liquid films between the bubble and walls *k*L,film*a*film:

$$k\_{\rm L}a = k\_{\rm L,film}a\_{\rm film} + k\_{\rm L,cap}a\_{\rm cap} \tag{4}$$

When *Fo* <0.1, both parts for the mass transfer process are important. When *Fo* >0.1, the liquid films are saturated and the mass transfer between the caps and liquid plugs predominates. The liquid side mass transfer coefficient in the film can be calculated as

$$k\_{\rm L,film} = 2\sqrt{\frac{D}{\pi t\_{\rm film}}} \frac{\ln\left(1/\Delta\right)}{1-\Delta}; \text{( $Fo < 0.1$ ), and} \tag{5}$$

$$k\_{\rm l,film} = \Im.14 \frac{D}{\delta\_{\rm film}}; \text{ (}Fo > 0.1\text{)}\tag{6}$$

where (*Fo* <sup>=</sup> *<sup>D</sup> <sup>t</sup>*film*<sup>δ</sup>* <sup>2</sup> film ),*D* is the diffusivity coefficient, and *δ*film is the film thickness *<sup>L</sup>* <sup>B</sup> <sup>−</sup> *<sup>d</sup> <sup>h</sup>*

*<sup>t</sup>*film <sup>=</sup> *<sup>L</sup>* film *<sup>U</sup>*<sup>B</sup> ≈ *<sup>U</sup>*<sup>B</sup> . Vandu et al. [16] found deviation between the experimental results and the predictions by using the expression proposed by van Baten et al. [14] for the absorption process of O2 in water. They deemed that the deviation was caused by the fact that the van Baten et al.'s expression validated for long mass transfer unit, in which situation the liquid film between the bubble and the channel walls is saturated; while in the experiments of Vandu et al. [16], the length of the liquid film is short and the velocity of the bubble is high to make the liquid film unsaturated. Vandu et al. [16] obtained the mass transfer coefficient for the bubble caps by using the classical penetration theory for mass transfer process as: *k*L,cap*a*cap =2 <sup>2</sup> *π DU*<sup>B</sup> *dh* . Mass transfer coefficient for the liquid film by using the unsteady diffusion model as *k*L,film*a*film =2 *DU*<sup>B</sup> *πt*film

$$k\_{\rm L}a = \frac{2}{\sqrt{\pi}} \sqrt{\frac{DLI\_{\rm B}}{\left(L\_{\rm B} - d\_{\rm h}\right)}} \frac{4\left(L\_{\rm B} - d\_{\rm h}\right)}{dh\left(L\_{\rm B} + L\_{\rm S}\right)} + 2\frac{\sqrt{2}}{\pi} \sqrt{\frac{DLI\_{\rm B}}{d\_{\rm h}}} \frac{4}{L\_{\rm B} + L\_{\rm S}}\tag{7}$$

This equation is available for the situation when the mass transfer in the liquid film is predo‐ minated and unsaturated at (*<sup>u</sup>* / *<sup>L</sup>* S)0.5 >3s-0.5. Another experimental fitting proposed by Yue et al. [17] is illustrated as *k*L*a* =2(*DU*<sup>B</sup> /(*L* <sup>B</sup> + *L* S))0.5(*L* <sup>B</sup> /(*L* <sup>B</sup> + *L* S))0.3 ⋅*d*H−<sup>1</sup> by performing experi‐ ments for the absorption of O2 in water in square microchannels. The experiments of Yue et al. [17] were operated under short film contact time so that the mixing between the liquid film and the liquid plug is poor. In this case, the predictions proposed by van Baten et al. [14] and Vandu et al. [16] are not applicable. It is noteworthy that these expressions are applicable for physical absorptions process with tiny mass transfer between gas and liquid phases, in which the characteristic parameters for bubble slugs and liquid plugs should be evaluated to obtain the mass transfer coefficient. This limits the applications of these predictions. Furthermore, Yue et al. [17] suggested that the quantitative information on the circulation in liquid plugs and the mixing between the liquid film and liquid plug needs to be revealed for constructing a universal model for mass transfer in bubble slug flow pattern for the gas-liquid two-phase flow in microchannels.

For practical application several predictions for the mass transfer coefficient were proposed by using either the nondimensional analysis or energy dissipative model. Yue et al. [11] correlated the mass transfer coefficient in the form of the Sherwood number *Sh* with the Schmidt number *Sc* (representing the physical property of the fluids) and the Reynolds number *Re* (representing the flow conditions) as *Sh <sup>L</sup>* ⋅ *a* ⋅*dh* =0.084Re*<sup>G</sup>* 0.213Re*<sup>L</sup>* 0.937*ScL* 0.5 for slug flow for the physical absorption of CO2 into water in rectangular microchannels. *Sh*L is the liquid Sherwood number defined by *Sh <sup>L</sup>* =*k*L*dh* / *D*CO2, representing the relative magnitude of the liquid side mass transfer coefficient *k*L to the diffusivity of CO2 in the liquid. *Sc*<sup>L</sup> is the liquid Schmidt number defined by *ScL* =*µL* / *ρ<sup>L</sup> D*CO2, representing the relative magnitude of the momentum diffusivity to the mass diffusivity. Re*G* and Re*<sup>L</sup>* are respectively the Reynolds number of the gas and liquid phases, defined by Re=*ρudh* / *µ*, representing the relative magnitude of the inertial forces to the viscous forces. These methods provide experimentally and semitheoret‐ ically the mass transfer coefficient prediction with relatively high precision only if the appli‐ cations are conducted within the range of the experimental conditions performed by Yue et al. [11]. However, this method needs a series of experiments in a wide range.

The energy dissipative model [18] correlates the mass transfer coefficient with the dissipative parameter *ς* (defined as *<sup>ς</sup>* =( <sup>Δ</sup>*pF <sup>L</sup>* )*TP u*) as the pressure drop along the channel differs for the gas-liquid two-phase flow with and without mass transfer [18]. Δ*pF* is the two-phase frictional pressure drop, *L* is the length of the microchannel, and *u* is the superficial velocity. The subscript *TP* represents the two-phase flow. Yue et al. [11] applied this function in mass transfer for the absorption of CO2 in water in microchannels and proposed a similar expression as *<sup>k</sup>*L*<sup>a</sup>* =0.0863( <sup>Δ</sup>*P*<sup>F</sup> *<sup>L</sup>* )*TP* 0.881 performs well for the relatively small mass transfer rates, in which situation the mass transfer process has little effect on the flow dynamics of the gas-liquid two-phase flow in microchannels. In addition, the frictional pressure drop needs to be obtained for predicting the mass transfer coefficient in this method, limiting its usage.

#### *1.3.3. Effects on the mass transfer process*

by using the classical penetration theory for mass transfer process as: *k*L,cap*a*cap =2 <sup>2</sup>

( )

*DU L d DU k a*

<sup>2</sup> <sup>4</sup> 2 4 <sup>2</sup> *<sup>h</sup> h h*

( )

298 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

*Re* (representing the flow conditions) as *Sh <sup>L</sup>* ⋅ *a* ⋅*dh* =0.084Re*<sup>G</sup>*

*L d dh L L*


*k*L,film*a*film =2

*DU*<sup>B</sup> *πt*film

flow in microchannels.

parameter *ς* (defined as *<sup>ς</sup>* =( <sup>Δ</sup>*pF*

L

p

transfer coefficient for the liquid film by using the unsteady diffusion model as

( ) B B <sup>B</sup>

This equation is available for the situation when the mass transfer in the liquid film is predo‐ minated and unsaturated at (*<sup>u</sup>* / *<sup>L</sup>* S)0.5 >3s-0.5. Another experimental fitting proposed by Yue et al. [17] is illustrated as *k*L*a* =2(*DU*<sup>B</sup> /(*L* <sup>B</sup> + *L* S))0.5(*L* <sup>B</sup> /(*L* <sup>B</sup> + *L* S))0.3 ⋅*d*H−<sup>1</sup> by performing experi‐ ments for the absorption of O2 in water in square microchannels. The experiments of Yue et al. [17] were operated under short film contact time so that the mixing between the liquid film and the liquid plug is poor. In this case, the predictions proposed by van Baten et al. [14] and Vandu et al. [16] are not applicable. It is noteworthy that these expressions are applicable for physical absorptions process with tiny mass transfer between gas and liquid phases, in which the characteristic parameters for bubble slugs and liquid plugs should be evaluated to obtain the mass transfer coefficient. This limits the applications of these predictions. Furthermore, Yue et al. [17] suggested that the quantitative information on the circulation in liquid plugs and the mixing between the liquid film and liquid plug needs to be revealed for constructing a universal model for mass transfer in bubble slug flow pattern for the gas-liquid two-phase

For practical application several predictions for the mass transfer coefficient were proposed by using either the nondimensional analysis or energy dissipative model. Yue et al. [11] correlated the mass transfer coefficient in the form of the Sherwood number *Sh* with the Schmidt number *Sc* (representing the physical property of the fluids) and the Reynolds number

physical absorption of CO2 into water in rectangular microchannels. *Sh*L is the liquid Sherwood number defined by *Sh <sup>L</sup>* =*k*L*dh* / *D*CO2, representing the relative magnitude of the liquid side mass transfer coefficient *k*L to the diffusivity of CO2 in the liquid. *Sc*<sup>L</sup> is the liquid Schmidt number defined by *ScL* =*µL* / *ρ<sup>L</sup> D*CO2, representing the relative magnitude of the momentum diffusivity to the mass diffusivity. Re*G* and Re*<sup>L</sup>* are respectively the Reynolds number of the gas and liquid phases, defined by Re=*ρudh* / *µ*, representing the relative magnitude of the inertial forces to the viscous forces. These methods provide experimentally and semitheoret‐ ically the mass transfer coefficient prediction with relatively high precision only if the appli‐ cations are conducted within the range of the experimental conditions performed by Yue et

The energy dissipative model [18] correlates the mass transfer coefficient with the dissipative

al. [11]. However, this method needs a series of experiments in a wide range.

*<sup>L</sup>* )*TP*

B B S B S

p

*dLL*


0.213Re*<sup>L</sup>*

*u*) as the pressure drop along the channel differs for the

0.937*ScL*

0.5 for slug flow for the

*π*

*DU*<sup>B</sup> *dh* . Mass

> Several studies have been performed for the various effects on the mass transfer process in microchannels. Eskin et al. [19] did experiments to study the effects of the size and length of microchannels on the mass transfer process for gas-liquid two-phase flow in microchannels, and found that the mass transfer coefficient increases with the reduction of the cross-sectional radius of microchannels. They proposed a model to predict the mass transfer coefficient for the mass transfer process of slug bubbles with the liquid plugs in long microchannels by taking into account the dynamical flow hydrodynamics such as pressure drop, variation of bubble size, and the variation of the local velocities. Shao et al. [20] conducted experiments on mass transfer process of CO2/N2 in 0.2 M NaOH aqueous solutions and water, respectively, in 0.25– 1 mm capillaries, and found that the volumetric mass transfer coefficient for the chemical absorption process is 3–12 times compared to the physical absorption process. They also studied the effects of the length of slug bubbles and liquid plugs, the velocity of bubbles, and the size of microchannels on the mass transfer process, and found that the volumetric mass transfer coefficient ranged between 0.3 and 0.5 s–1. Sobieszuk et al. [21] measured the interfacial area in the slug flow in a microchannel by using the Danckwerts' method for the CO2 absorption from CO2/N2 mixture into KHCO3/K2CO3 buffer solutions, and provides, for the first time, the mass transfer coefficients separately for the liquid film and the liquid caps. Sobieszuk et al. [12] further found that the enhancement of the overall rate of the mass transfer depends on the gas concentrations. Ichiyanagi et al. [22] performed detailed three-dimensional measurements of the velocity and concentration distributions for CO2 dissolution process through the gas-liquid interface in microchannels by using the advanced technique confocal micron-resolution particle image velocimetry (micro-PIV) combined with laser-induced fluorescence (LIF). The LIF measurement demonstrated that the dissolved gas in the spanwise direction decreases with the increase of the Reynolds number. The molar fluxes in the streamwise direction were at least 20 times compared to those in the spanwise and depthwise directions, signifying that the enhancement of the momentum transport in the spanwise and depthwise directions plays an important role in the enhancement of the mass transfer for gasliquid slug flow in microchannels.

> Several investigations have been carried out on the dynamical mass transfer process for slug bubbles for gas-liquid two-phase flow in microchannels by taking advantage of the real-time recording of the movement of bubbles in microchannels with the help of high-speed digital

camera [23-28]. Tan et al. [23] found that bubbles formed during the mass transfer process of a CO2/N2 mixture with contact with the NaOH aqueous solution at the microfluidic T-junction are smaller than those formed without mass transfer process. The typical time for bubble formation ranges between 0.2 and 0.4s, during which the amount of mass transfer contributes to around 30–40% of the total transferred solute. The overall mass transfer coefficient during bubble formation stage is in the range of 1.4 × 10−<sup>4</sup> to 5.5 × 10−<sup>4</sup> m/s, and it increases with the elevation of the gas and liquid flow rates. It is deemed that the resistance in the gas side during the mass transfer process predominates; and the circulation in the slug bubbles significantly intensifies the mass transfer with the enhancement factor in the range of 5–15. Furthermore, they studied the effects of the angle of the junction on the mass transfer coefficient and found that the mass transfer coefficient in the liquid size achieves its maximum at 90° junction, and reaches its minimum at 150° junction.

#### *1.3.4. Intensification of the mass transfer process*

In order to intensify the mass transfer of CO2 in fluids in microreactors, several methods have been employed, such as inducing turbulence by using the third inert gas and obstacles or particles in channels, specific geometry (curved) of the channel, and active forces. Su et al. [29] found that the mass transfer process can be intensified by adding solid particles in micro‐ channels. Su et al. [30] introduced the third phase-inert gas into the liquid-liquid two-phase flow in microchannels to enhance the mass transfer coefficient. Tan et al. [24] performed experiments to study the effects of the curvature of the curved geometry for microchannels on the absorption of CO2 for a CO2/N2 mixture contacting with NaOH aqueous solution, and found that the mass transfer coefficient is greatly augmented with the decrease of the curvature radius. That is, *k*L in curved channels with the curve radius of 3 cm is nearly two times compared to that in a straight channel. They also proposed an expression to predict the mass transfer coefficient as *k*L,R / *<sup>k</sup>*L,R→*<sup>∞</sup>* =1 <sup>+</sup> <sup>91</sup>*d*<sup>H</sup> / *<sup>R</sup>*, where *k*L,R and *k*L,R→*<sup>∞</sup>* are respectively the overall mass transfer coefficient in the curved channel with curve radius of *R* and in a straight channel, and *d*H is the hydrodynamic radius of the microchannel.

The mechanism for mass transfer during CO2 bubbles flowing in microchannels needs to be explored to be manipulated according to various applications, especially more attention should be paid to the dynamics of the gas-liquid interface during the mass transfer process and provide solid foundations for the coupling of transport and reaction, scale-up, and optimization of microreactors.

## **2. Microfluidic technology in oil and gas analysis, and CO2 applications**

Recently, microfluidic technique is found to be a promising tool for the oil and gas analysis, for measurements of parameters such as Henry's coefficient, solubility of gas in the liquid phase, and the gas-liquid reaction coefficient [31]. Sell et al. [32] provided a microfluidics technique to measure the diffusion coefficient for CO2 in water and brine, with the advantage of only microliters of sample and analysis within minutes (Figure 1). However, established macroscale pressure-volume-temperature cell methods require large sample volumes, which is completed within hours or days. They also found that pressure had no significant effect on diffusion rates, supporting an assumption applied by many sequestration models. Namely, time scales for dissolution are independent of reservoir depth or *ex situ* carbonation pressures. Fadaei et al. [33] also provided a microfluidic method to rapidly measure the CO2 diffusivity in Bitumen, with the range of pressures of 1–5 MPa and at a room temperature of 21°C. They obtained the diffusion coefficients in the order of 10–10 m2 /s, agreeing well with the relevant published data using conventional methods. Compared to the traditional methods (with 0.5 L of sample within hours or days), this method requires only 10 minutes and 1 nL plug of sample.

camera [23-28]. Tan et al. [23] found that bubbles formed during the mass transfer process of a CO2/N2 mixture with contact with the NaOH aqueous solution at the microfluidic T-junction are smaller than those formed without mass transfer process. The typical time for bubble formation ranges between 0.2 and 0.4s, during which the amount of mass transfer contributes to around 30–40% of the total transferred solute. The overall mass transfer coefficient during

elevation of the gas and liquid flow rates. It is deemed that the resistance in the gas side during the mass transfer process predominates; and the circulation in the slug bubbles significantly intensifies the mass transfer with the enhancement factor in the range of 5–15. Furthermore, they studied the effects of the angle of the junction on the mass transfer coefficient and found that the mass transfer coefficient in the liquid size achieves its maximum at 90° junction, and

In order to intensify the mass transfer of CO2 in fluids in microreactors, several methods have been employed, such as inducing turbulence by using the third inert gas and obstacles or particles in channels, specific geometry (curved) of the channel, and active forces. Su et al. [29] found that the mass transfer process can be intensified by adding solid particles in micro‐ channels. Su et al. [30] introduced the third phase-inert gas into the liquid-liquid two-phase flow in microchannels to enhance the mass transfer coefficient. Tan et al. [24] performed experiments to study the effects of the curvature of the curved geometry for microchannels on the absorption of CO2 for a CO2/N2 mixture contacting with NaOH aqueous solution, and found that the mass transfer coefficient is greatly augmented with the decrease of the curvature radius. That is, *k*L in curved channels with the curve radius of 3 cm is nearly two times compared to that in a straight channel. They also proposed an expression to predict the mass transfer coefficient as *k*L,R / *<sup>k</sup>*L,R→*<sup>∞</sup>* =1 <sup>+</sup> <sup>91</sup>*d*<sup>H</sup> / *<sup>R</sup>*, where *k*L,R and *k*L,R→*<sup>∞</sup>* are respectively the overall mass transfer coefficient in the curved channel with curve radius of *R* and in a straight channel, and

The mechanism for mass transfer during CO2 bubbles flowing in microchannels needs to be explored to be manipulated according to various applications, especially more attention should be paid to the dynamics of the gas-liquid interface during the mass transfer process and provide solid foundations for the coupling of transport and reaction, scale-up, and

**2. Microfluidic technology in oil and gas analysis, and CO2 applications**

Recently, microfluidic technique is found to be a promising tool for the oil and gas analysis, for measurements of parameters such as Henry's coefficient, solubility of gas in the liquid phase, and the gas-liquid reaction coefficient [31]. Sell et al. [32] provided a microfluidics technique to measure the diffusion coefficient for CO2 in water and brine, with the advantage of only microliters of sample and analysis within minutes (Figure 1). However, established

to 5.5 × 10−<sup>4</sup> m/s, and it increases with the

bubble formation stage is in the range of 1.4 × 10−<sup>4</sup>

300 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

reaches its minimum at 150° junction.

*1.3.4. Intensification of the mass transfer process*

*d*H is the hydrodynamic radius of the microchannel.

optimization of microreactors.

**Figure 1.** Microfluidic system used to measure the CO2 diffusion coefficients in water. (a) Illustration of the experimen‐ tal setup. (b) Schematic of the CO2 diffusivity test initialization procedure. "Reprinted (adapted) with permission from Sell, A., Fadaei, H., Kim, M., Sinton, D., 2013. Measurement of CO2 diffusivity for carbon sequestration: a microfluidic approach for reservoir-specific analysis. Environmental Science & Technology 47, 71–78. Copyright (2013) American Chemical Society." (Sell et al. [32]) http://dx.doi.org/10.1021/es303319q

Lefortier et al. [34] presented a high-throughput method to rapidly measure the diffusion coefficients and solubility of CO2 in pure solvents and mixtures in realtime by taking advantage of the visualization of the variation of CO2 absorption fluids in microchannels. Sun and Cubaud [25] experimentally study the dissolution of CO2 into water, ethanol, and methanol by using the microfluidic technique (Figure 2), and found that the bubble dissolution rate depends on the inlet gas pressure and fluid pair composition. For short period of time after the contact of fluids, the bubble length decreases linearly with time, displaying a fast diffusive behavior owing to the CO2 concentration gradient localized in the thin gas-liquid interfacial region. The initial rate of the diminishing bubble size is proportional to the ratio of the diffusion coefficient to the Henry's law constant. This study shows that the gas saturation and sequestration processes can be achieved rapidly across quite short distances in microfluidic devices.

**Figure 2.** Examples of diffusive multiphase flows. (a) Weakly diffusive bubbles for CO2 in water. The ratio of the liquid volumetric flow rates to the sum of the gas and liquid flow rates is 0.23, 0.47, and 0.80 from top to bottom. (b) Strong diffusive bubbles for CO2 in methanol. The ratio of the liquid volumetric flow rates to the sum of the gas and liquid flow rates is 0.24, 0.51, and 0.68 from top to bottom. Reproduced from Sun and Cubaud [25] with permission of The Royal Society of Chemistry. http://dx.doi.org/10.1039/C1LC20348G

Abolhasani et al. [35] studied the automated microfluidic method for the rapid measurement of CO2 mass transfer and solubility in physical solvents. The variation of slug bubbles flowing along the microchannel was dynamically recorded and analyzed to obtain the solubility of CO2 in solvent. For CO2-dimethyl carbonate (DMC) system, the volumetric mass transfer coefficients ranged between 4 and 30 s–1, and Henry's constants were within the range of 6–12 MPa. Li et al. [36] presented a microfluidic method to study rapid gas-liquid reactions for the rapid acquisition of the kinetic data for the reaction by also capturing the dynamical variation of slug bubbles flowing in microchannels (Figure 3). In this work, the application of micro‐ fluidic method was utilized for systems with relatively low concentrations of the reagents and products and for the low-viscosity media. This method is expected to apply for the CO2 sequestration in the oil industry. Tumarkin et al. [37] conducted similar experiments to manipulate the solubility of CO2 in water and a 0.7 M NaCl aqueous solution, by controlling the temperature of the fluids flowing in the microchannel. Bubbles experience shrinkageexpansion-shrinkage stage, termed as "bubble breathing," when the cooling-heating-cooling system is applied to the bubble flowing system. Tumarkin et al. [38] controlled the size of CO2 bubbles by manipulating the temperature of the fluids, and functional particles were loaded at the gas-liquid interface to stabilize the bubble from coalescence. Park et al. [39] produced small CO2 bubbles with a diameter less than 8 µm in microfluidic devices by manipulating the pH value of the liquid phase, and found that the bubble size depends on the flow rates of the liquid phase and the acid-base equilibrium established in the microchannels.

Lefortier et al. [34] presented a high-throughput method to rapidly measure the diffusion coefficients and solubility of CO2 in pure solvents and mixtures in realtime by taking advantage of the visualization of the variation of CO2 absorption fluids in microchannels. Sun and Cubaud [25] experimentally study the dissolution of CO2 into water, ethanol, and methanol by using the microfluidic technique (Figure 2), and found that the bubble dissolution rate depends on the inlet gas pressure and fluid pair composition. For short period of time after the contact of fluids, the bubble length decreases linearly with time, displaying a fast diffusive behavior owing to the CO2 concentration gradient localized in the thin gas-liquid interfacial region. The initial rate of the diminishing bubble size is proportional to the ratio of the diffusion coefficient to the Henry's law constant. This study shows that the gas saturation and sequestration processes can be achieved rapidly across quite short distances in microfluidic devices.

302 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

**Figure 2.** Examples of diffusive multiphase flows. (a) Weakly diffusive bubbles for CO2 in water. The ratio of the liquid volumetric flow rates to the sum of the gas and liquid flow rates is 0.23, 0.47, and 0.80 from top to bottom. (b) Strong diffusive bubbles for CO2 in methanol. The ratio of the liquid volumetric flow rates to the sum of the gas and liquid flow rates is 0.24, 0.51, and 0.68 from top to bottom. Reproduced from Sun and Cubaud [25] with permission of The

Abolhasani et al. [35] studied the automated microfluidic method for the rapid measurement of CO2 mass transfer and solubility in physical solvents. The variation of slug bubbles flowing along the microchannel was dynamically recorded and analyzed to obtain the solubility of CO2 in solvent. For CO2-dimethyl carbonate (DMC) system, the volumetric mass transfer coefficients ranged between 4 and 30 s–1, and Henry's constants were within the range of 6–12 MPa. Li et al. [36] presented a microfluidic method to study rapid gas-liquid reactions for the rapid acquisition of the kinetic data for the reaction by also capturing the dynamical variation of slug bubbles flowing in microchannels (Figure 3). In this work, the application of micro‐ fluidic method was utilized for systems with relatively low concentrations of the reagents and products and for the low-viscosity media. This method is expected to apply for the CO2 sequestration in the oil industry. Tumarkin et al. [37] conducted similar experiments to manipulate the solubility of CO2 in water and a 0.7 M NaCl aqueous solution, by controlling the temperature of the fluids flowing in the microchannel. Bubbles experience shrinkageexpansion-shrinkage stage, termed as "bubble breathing," when the cooling-heating-cooling system is applied to the bubble flowing system. Tumarkin et al. [38] controlled the size of

Royal Society of Chemistry. http://dx.doi.org/10.1039/C1LC20348G

**Figure 3.** Reversible CO2 binding to secondary amines performed in the microfluidic reactor. An ITO glass-based heat‐ er is placed underneath region 2, and an aluminum plate is placed underneath region 1 to maintain the temperature in this region at 23°C. (b) Reaction of CO2 with a secondary amine in reaction 1, and the volume of CO2 slugs decreases with CO2 reacts with R1NHR2. (c) Release of CO2 manipulated by the increase of the temperature in reaction 2. "Re‐ printed with permission from Li, W., Liu, K., Simms, R., Greener, J., Jagadeesan, D., Pinto, S., Gunther, A., Kumacheva, E., 2012. Microfluidic study of fast gas-liquid reactions. Journal of the American Chemical Society 134, 3127–3132. Copyright (2012) American Chemical Society." (Li et al. [36]) http://dx.doi.org/10.1021/ja2101278

In summary, the microfluidic technique is a promising tool toward the measurement of the physical property of fluids involved during the CO2 application, and gas and oil analysis, with the advantage of small sample size and short analysis time. However, the dynamics and mechanisms of the fluid flow and mass transfer in microchannels needs to be highlighted to meet these applications.

## **3. Microfluidic technology in CO2 flooding, improving the oil recovery and CO2 sequestration**

With the increasing concern on environmental protection, the integration of CO2 capturesequestration-application for enhanced oil recovery technology will be the new trend in the future. Microfluidics technology, with the unique advantages in realtime visualization and quantification, can reveal the transport mechanism in the microscale for multiphase interfacial phenomena in microchannels within porous media during the CO2 capture-sequestrationapplication in line with the pressure, temperature, and material properties of the rock within the oil reservoir. These wells or salt sand underground aquifers comprise of porous media made of many tens of microchannels in a network structure, which are consistent with the conventional scale of microchannels used in the microfluidic technique [31].

#### **3.1. Fluid dynamics**

When CO2is injected into underneath wells, it becomes a supercritical fluid at the underground temperature and pressure. It is lighter than oil or brine and denser than CO2 that is under normal conditions. Driven by the pressure, superficial CO2 flows through the liquid-filled microchannel networks in the rock layer of porous media, and its flow paths are influenced by the heterogeneous structure of the porous media [40]. In microchannels, the tongue of CO2 gaseous thread can breakup into bubbles driven by the capillary instability, and the generated bubbles can be captured or move through a straight microchannel or bifurcation junctions. Of course, these CO2 bubbles will be absorbed slowly by the liquid phase [41]. In a CO2 injection process, it usually takes three or four decades or even centuries of time for the complete absorption of CO2. Although the microfluidic technology has been applied to study the CO2 flooding in recent years, these studies on the dynamics and mechanics of multiphase flow in porous media consisting of microchannel networks have been carried out in the mid-20th century by the fluid mechanics community [42, 43]. For example, the famous viscous fingering phenomenon for miscible and immiscible two-phase flow in Hele-Shaw cells (rectilinear flow or radial source flow) has attracted attention for many years including the stability and nonstability mechanism of the interface, the bubble capture and release mecha‐ nism, the rupture of the interface, and the tip streaming of droplets and bubbles [44]. The interfacial phenomenon in the basic unit of the porous media network, or the loops of microchannels, has also been explored including the interfacial stability in symmetric and asymmetric loops and bubble capture in asymmetric loops [45-48]. For example, Oxaal et al. [45] studied the displacement of a high-viscosity fluid by a low-viscosity fluid that results in viscous fingering in homogeneous porous media. They found that the interfacial dynamics was dominated by the viscous forces at high flow rates and by capillary forces at low flow rates. Lenormand et al. [49] proposed a leaking mechanism to explain the dynamic phenomena in drainage for the displacement of one fluid by another in a network consisting of many ducts. They found that the wetting fluid motion at the edges was much slower than the nonwetting fluid, which was affected by the viscosity of the latter fluid. The local phenomenon was found to be linked with the drainage pressure and imbibition pressures at the microscale, and the macroscopic effect at the network scale was found to be related to the topology of the non‐ wetting fluid at the end of the drainage. The dynamics of trapped bubbles was influenced by their size and topology.

Recently, with the development of microfluidic techniques, detailed information on the twophase interfacial phenomena in microfluidic loops has been explored. Al-Housseiny et al. [48]

provided a microfluidic tool to control the interfacial instability by changing the geometry of the device, and found that a gradient in the passage can lead to different displacement behaviors. This finding can be used to manipulate instabilities in fluid-fluid systems in microfluidic devices. The interesting and complex problem is that the fluid can penetrate evenly or not the two identical daughter channels in a network composed of two identical channels that are linearly vary in radius [46, 47]. When the surface tension predominates at low capillary numbers, this geometry can lead to the fluid to enter only one of the two branches, signifying that the uniform fluid penetration into the network is not always stable. At high capillary number Ca, the interface advances together and the fluid penetration is stable as viscous forces are dominant. The system can also exhibit an interplay between viscous and surface tension effects. In addition, the preferential flow penetration increases with the decrease of the viscosity ratio. This study suggested that the sweep efficiency could be enhanced if the channels in the network became narrower in the flow direction for flushing wetting oils by a low viscosity solvent. They also showed that the presence of an elastic boundary could lead to the suppression of the instability of the dynamics of the propagating gas-liquid interface for the Hele-Shaw displacement of a viscous liquid by a gas underneath an elastic membrane, governed by the surface tension at the gas-liquid interface due to the tapered flow geometry underneath the deflected membrane [50]. Controlling instability of fingers for fluid-fluid interface is fundamental to a wide range of applications such as flows in porous media in enhanced oil recovery processes and carbon sequestration [47]. Other examples of two-phase flows in spatially varying geometries are flows of bubbles and drops through junctions, constricted capillaries, flexible tubes, and tapered channels [4, 41, 51-53]. Bubbles and droplets can pass through, be captured, stay or leave, and even break up into small ones in spatially varying geometries. For example, bubbles can be trapped or released from a linear pore, governed by a capillary number [41]. The critical capillary number characteristic of the transition between trapping and releasing depends nonmonotonically on the bubble size. To better explore the dynamics of the gas-liquid interface in microchannels, the detailed flow-field distribution can be provided by advanced measurement methods, such as micro-PIV (Figure 4) [4].

#### **3.2. Oil recovery**

future. Microfluidics technology, with the unique advantages in realtime visualization and quantification, can reveal the transport mechanism in the microscale for multiphase interfacial phenomena in microchannels within porous media during the CO2 capture-sequestrationapplication in line with the pressure, temperature, and material properties of the rock within the oil reservoir. These wells or salt sand underground aquifers comprise of porous media made of many tens of microchannels in a network structure, which are consistent with the

When CO2is injected into underneath wells, it becomes a supercritical fluid at the underground temperature and pressure. It is lighter than oil or brine and denser than CO2 that is under normal conditions. Driven by the pressure, superficial CO2 flows through the liquid-filled microchannel networks in the rock layer of porous media, and its flow paths are influenced by the heterogeneous structure of the porous media [40]. In microchannels, the tongue of CO2 gaseous thread can breakup into bubbles driven by the capillary instability, and the generated bubbles can be captured or move through a straight microchannel or bifurcation junctions. Of course, these CO2 bubbles will be absorbed slowly by the liquid phase [41]. In a CO2 injection process, it usually takes three or four decades or even centuries of time for the complete absorption of CO2. Although the microfluidic technology has been applied to study the CO2 flooding in recent years, these studies on the dynamics and mechanics of multiphase flow in porous media consisting of microchannel networks have been carried out in the mid-20th century by the fluid mechanics community [42, 43]. For example, the famous viscous fingering phenomenon for miscible and immiscible two-phase flow in Hele-Shaw cells (rectilinear flow or radial source flow) has attracted attention for many years including the stability and nonstability mechanism of the interface, the bubble capture and release mecha‐ nism, the rupture of the interface, and the tip streaming of droplets and bubbles [44]. The interfacial phenomenon in the basic unit of the porous media network, or the loops of microchannels, has also been explored including the interfacial stability in symmetric and asymmetric loops and bubble capture in asymmetric loops [45-48]. For example, Oxaal et al. [45] studied the displacement of a high-viscosity fluid by a low-viscosity fluid that results in viscous fingering in homogeneous porous media. They found that the interfacial dynamics was dominated by the viscous forces at high flow rates and by capillary forces at low flow rates. Lenormand et al. [49] proposed a leaking mechanism to explain the dynamic phenomena in drainage for the displacement of one fluid by another in a network consisting of many ducts. They found that the wetting fluid motion at the edges was much slower than the nonwetting fluid, which was affected by the viscosity of the latter fluid. The local phenomenon was found to be linked with the drainage pressure and imbibition pressures at the microscale, and the macroscopic effect at the network scale was found to be related to the topology of the non‐ wetting fluid at the end of the drainage. The dynamics of trapped bubbles was influenced by

Recently, with the development of microfluidic techniques, detailed information on the twophase interfacial phenomena in microfluidic loops has been explored. Al-Housseiny et al. [48]

conventional scale of microchannels used in the microfluidic technique [31].

304 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

**3.1. Fluid dynamics**

their size and topology.

Haas et al. [54] presented a lab-on-a-chip method to inform oil recovery by injecting steam underground in a process known as steam-assisted gravity drainage to extract bitumen—a very viscous oil, taking advantage of the pore-scale quantification of fluid dynamics at relevant reservoir conditions and pore sizes. It is found that the characteristic size of oil-in-water emulsions generated is reduced from 150 to 6 µm and the corresponding recovery effectiveness is enhanced by 50% with the additive.

#### **3.3. CO2 sequestration**

Four mechanisms have been proposed for CO2 sequestration in saline aquifers: structural and hydrodynamic capture due to the density difference between the liquid and CO2, residual capture during the formation of the porous spaces, the solubility capture owing to the

**Figure 4.** Velocity fields in the liquid phase around a breaking bubble at a T-junction with the help of micro-PIV. Re‐ printed from Fu et al. (2014). (Reproduced with permission. Copyright John Wiley and Sons, 2014) (Fu et al. [4]) http:// dx.doi.org/10.1002/aic.14377.

dissolvedCO2 in the liquid phase, and mineral capture due to the stable carbonate formed by the reaction of CO2 and rock minerals [40]. In general, low-pressure foam micromodel studies physically show the effect of foam within the porous media. To highlight chemical interactions with the oil phase requires all phases to be at reservoir pressure. Ma et al. [55] used a micro‐ model to investigate the sweep efficiency of the surfactant CO2 foam in a heterogeneous network without oil at ambient conditions. Kumar Gunda et al. [56] designed a "reservoir-ona-chip" to represent the pore structure of a naturally oil-bearing reservoir rock to perform conventional water-flooding experiments, and observed in real time the fluid-fluid interface structure at the junctions of the porous media (Figure 5). The fluid-fluid interface is found to be trapped at some junctions by capillary trapping. Datta et al. [57] constructed velocity distributions in a fluid flow in a three-dimensional porous media by using the confocal microscopy, and they found that the velocity magnitudes and the velocity components both along and transvers to the imposed flow direction were distributed exponentially. They also observed that the pore-scale correlations in the flow were predominated by the geometry of the media, which suggested that the fluid flow through the pore space was not completely random despite the considerable complexity of it. Wu et al. [58] studied the effect of the wettability of the channel and the morphology of the network on the fluid flow for the displacement of oil by water in porous media with a microfluidic model. The former affects the structure of the water phase, while the latter determines the residual oil saturation. The networks used in this study were composed of about 600 grains separated by a highly connected network of channels with an overall porosity of 0.11–0.20. The findings suggested that the morphology of the complex network resulted in a complex flow behavior that was difficult to predict by solely on porosity, and they provided a versatile tool on the study of visualized multiphase flow behavior and displacement mechanism in porous media at the microscale, nanoscale, and pore scale.

**Figure 5.** The conceptual map for "reservoir-on-a-chip." Reproduced from Kumar Gunda et al. [56] with permission of The Royal Society of Chemistry. http://dx.doi.org/10.1039/C1LC20556K

dissolvedCO2 in the liquid phase, and mineral capture due to the stable carbonate formed by the reaction of CO2 and rock minerals [40]. In general, low-pressure foam micromodel studies physically show the effect of foam within the porous media. To highlight chemical interactions with the oil phase requires all phases to be at reservoir pressure. Ma et al. [55] used a micro‐ model to investigate the sweep efficiency of the surfactant CO2 foam in a heterogeneous network without oil at ambient conditions. Kumar Gunda et al. [56] designed a "reservoir-ona-chip" to represent the pore structure of a naturally oil-bearing reservoir rock to perform conventional water-flooding experiments, and observed in real time the fluid-fluid interface

**Figure 4.** Velocity fields in the liquid phase around a breaking bubble at a T-junction with the help of micro-PIV. Re‐ printed from Fu et al. (2014). (Reproduced with permission. Copyright John Wiley and Sons, 2014) (Fu et al. [4]) http://

306 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

dx.doi.org/10.1002/aic.14377.

Several studies show that the dynamics of the fluid-fluid interface at the microscale is influ‐ enced by the morphology and wettability of the microchannels in porous space. Berejnov et al. [59] constructed a lab-on-a-chip to study the effect of the wettability of microchannels on the selectivity and percolation patterns of multiphase flow in networks with 5000 channels at pore scale. Murison et al. [60] found that additional dissipation at small extension can be caused by the contact line pinning of gas beads for a dense packing. More attention should be paid to the effects of morphology of porous media and wettability on multiphase flow in porous spaces.

To overcome the difference between the material used for typical microfluidic devices (glass, silicone, and PDMS) and the real rock for CO2 flooding and sequestration, Song et al. [61] presented a real-rock micromodel made in a naturally mineral substrate as shown in Figure 6, to directly study the multiphase flow behavior and multicomponent interactions in real time. They demonstrated the dissolution of carbonate rock with time resulted by hydrochloric acid flow relevant to acidizing processes for reservoir stimulations. Flow and crystal orientationdirected preferential dissolution was observed. This work paves the way for future applica‐ tions of real-rock microfluidics, highlighting processes in the subsurface, to study the fundamental fluid-rock interactions and examine the effect of industrially relevant injection fluids on CO2 storage in saline aquifers.

**Figure 6.** Fabrication of the microfluidic devices made in natural calcite material. Reproduced from Song et al. [61] with permission of The Royal Society of Chemistry. http://dx.doi.org/10.1039/C4LC00608A

In general, low-pressure foam micromodel studies demonstrate the effect of foam as a physical structure within the porous media. To include representative chemical interactions with the oil phase requires all phases to be at reservoir pressure. The microfluidic technique is also tested for high pressure CO2 injection, supercritical CO2 injection, CO2-surfactant injection, and low-pressure air-surfactant injection [31]. Nguyen et al. [62] provided a microcore method to study the porosity and permeability changes for pore-scale analysis of superficial CO2 reactive transport in saline aquifers at reservoir temperature and pressure conditions of 8.4 MPa and 40°C. Kim et al. [63] presented another lab-on-a-chip approach to understand the process of salt precipitation during CO2 sequestration at pore scale. The dynamics of flow, evaporation, and salt formation within porous media could be observed in the complex networks consisting of microchannels with pore size distributions relevant to native porous media. Salt precipita‐ tion happens streamwise with a speed equivalent to 2% of the CO2 velocity. The salt formation mechanism could also be identified at the pore scale using this method: large bulk crystals forming early in the trapped brine phases on the order of the pore size; and polycrystalline aggregated structures forming late in the evaporation process on a wide range of length scale. These findings can be used to explore the blockage phenomenon and well bore dry-out strategy. Nguyen et al. [64] evaluated nanoparticle CO2 foam stability and EOR efficiency by using a micromodel approach, providing a quantitative measurement of bubble size and coalescence dynamics. In comparison to other similar cases with CO2 gas, the nanoparticlestabilized CO2 foam showed a three-fold increase in oil recovery attributed to the role of the physical pore-scale bubble structures, which are rendered very stable in the presence of nanoparticles.

### **4. Conclusions and perspectives**

To overcome the difference between the material used for typical microfluidic devices (glass, silicone, and PDMS) and the real rock for CO2 flooding and sequestration, Song et al. [61] presented a real-rock micromodel made in a naturally mineral substrate as shown in Figure 6, to directly study the multiphase flow behavior and multicomponent interactions in real time. They demonstrated the dissolution of carbonate rock with time resulted by hydrochloric acid flow relevant to acidizing processes for reservoir stimulations. Flow and crystal orientationdirected preferential dissolution was observed. This work paves the way for future applica‐ tions of real-rock microfluidics, highlighting processes in the subsurface, to study the fundamental fluid-rock interactions and examine the effect of industrially relevant injection

308 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

**Figure 6.** Fabrication of the microfluidic devices made in natural calcite material. Reproduced from Song et al. [61]

In general, low-pressure foam micromodel studies demonstrate the effect of foam as a physical structure within the porous media. To include representative chemical interactions with the oil phase requires all phases to be at reservoir pressure. The microfluidic technique is also tested for high pressure CO2 injection, supercritical CO2 injection, CO2-surfactant injection, and low-pressure air-surfactant injection [31]. Nguyen et al. [62] provided a microcore method to study the porosity and permeability changes for pore-scale analysis of superficial CO2 reactive transport in saline aquifers at reservoir temperature and pressure conditions of 8.4 MPa and

with permission of The Royal Society of Chemistry. http://dx.doi.org/10.1039/C4LC00608A

fluids on CO2 storage in saline aquifers.

In conclusion, microfluidic technique is a promising tool for CO2 capture, sequestration, and application. To further take advantage of this new technique, several challenges, needs, and future directions should be highlighted. The detailed fluid dynamics for multiphase flow in single microchannels, single-stage microfluidic divergences and doublet microchannels, and multistage microfluidic divergences and network microchannels need to be explored. The fascinating fluid dynamics at fluid-fluid interface involving surface tension at microscale is still challenging. The stability and non-stability of the fluid-fluid interface at microchannels and networks need to be emphasized to gain insights into the mechanism for fluid flow in porous media for the CO2 capture, sequestration, and applications. For example, the breakup and nonbreakup of the fluid-fluid interface and the trapping and releasing of bubbles in microchannels should be explored. Studies on the effects of the gas and liquid flow rate, pressure, temperature, the viscosity and surface tension of the liquid phase, the wettability of the channel wall, the geometry and size of microchannels, and the structural morphology of multistage microchannels on the dynamical evolution of the fluid-fluid interface should be explored.

It is urgent to conduct studies on CO2 utilization in line with the pressure, temperature, and materials for porous media in practical application at the microreactor scale and pore scale. The most challenging problem is how to evaluate and manipulate the fluid-fluid interface for both immiscible and miscible systems at the microreactor scale and pore scale. Many of porescale phenomena need to be explored, including the behavior of the wetting films, the movement of contact lines and dynamic contact angles, the stability of capillary bridges, the dynamic instability of nonwetting fluid, the propagation of fluid-fluid interface, even the phase transitions during multiphase flowing, the heat and mass transfer across interfaces, and many others.

## **Acknowledgements**

The financial supports for this project from the National Natural Science Foundation of China (21106093), the Research Fund for the Doctoral Program of Higher Education (20110032120010), and the Elite Scholar Program of Tianjin University are gratefully acknowledged.

## **Author details**

Taotao Fu1,2\*

Address all correspondence to: ttfu@tju.edu.cn

1 State Key Laboratory of Chemical Engineering, Collaborative Innovation Center of Chemical Science and Engineering, Tianjin, China

2 School of Chemical Engineering and Technology, Tianjin University, Tianjin, China

## **References**


**Acknowledgements**

acknowledged.

**Author details**

Address all correspondence to: ttfu@tju.edu.cn

310 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

Chemical Science and Engineering, Tianjin, China

Taotao Fu1,2\*

**References**

905 (2013).

Technology 35, 1184 (2012).

[4] T. Fu, Y. Ma, and H. Z. Li, AIChE Journal 60, 1920 (2014).

[6] F. P. Bretherton, Journal of Fluid Mechanics 10, 166 (1961).

Whitesides, Lab on a Chip 7, 1479 (2007).

Science 60, 5895 (2005).

Journal 51, 2428 (2005).

The financial supports for this project from the National Natural Science Foundation of China (21106093), the Research Fund for the Doctoral Program of Higher Education (20110032120010), and the Elite Scholar Program of Tianjin University are gratefully

1 State Key Laboratory of Chemical Engineering, Collaborative Innovation Center of

2 School of Chemical Engineering and Technology, Tianjin University, Tianjin, China

[1] K. S. Elvira, X. C. I. Solvas, R. C. R. Wootton, and A. J. deMello, Nature Chemistry 5,

[2] V. Hessel, I. Vural Gürsel, Q. Wang, T. Noël, and J. Lang, Chemical Engineering &

[3] M. T. Kreutzer, F. Kapteijn, J. A. Moulijn, and J. J. Heiszwolf, Chemical Engineering

[5] M. J. Fuerstman, A. Lai, M. E. Thurlow, S. S. Shevkoplyas, H. A. Stone, and G. M.

[7] J. Ratulowski and H.-C. Chang, Physics of Fluids A: Fluid Dynamics 1, 1642 (1989).

[9] W. B. Kolb and R. L. Cerro, Physics of Fluids A: Fluid Dynamics 5, 1549 (1993).

[10] H. Wong, C. J. Radke, and S. Morris, Journal of Fluid Mechanics 292, 95 (1995).

[8] M. T. Kreutzer, F. Kapteijn, J. A. Moulijn, C. R. Kleijn, and J. J. Heiszwolf, AIChE


[33] H. Fadaei, B. Scarff, and D. Sinton, Energy & Fuels 25, 4829 (2011).

312 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

(2012).

(2012).

(2011).

(1987).

(2014).

034502 (2013).

(2011).

[40] D. P. Schrag, Science 315, 812 (2007).

[42] G. I. Taylor, Journal of Fluid Mechanics 10, 161 (1961).

[43] G. M. Homsy, Annual Review of Fluid Mechanics 19, 271 (1987).

[34] S. Lefortier, P. Hamersma, A. Bardow, and M. T. Kreutzer, Lab on a Chip 12, 3387

[35] M. Abolhasani, M. Singh, E. Kumacheva, and A. Gunther, Lab on a Chip 12, 1611

[36] W. Li, K. Liu, R. Simms, J. Greener, D. Jagadeesan, S. Pinto, A. Gunther, and E. Ku‐

[37] E. Tumarkin, Z. Nie, J. I. Park, M. Abolhasani, J. Greener, B. Sherwood-Lollar, A.

[38] E. Tumarkin, J. I. Park, Z. Nie, and E. Kumacheva, Chemical Communications 47

macheva, Journal of the American Chemical Society 134, 3127 (2012).

[39] J. I. Park, Z. Nie, A. Kumachev, and E. Kumacheva, Soft Matter 6, 630 (2010).

[41] G. Dawson, S. Lee, and A. Juel, Journal of Fluid Mechanics 722, 437 (2013).

[47] T. T. Al-Housseiny and H. A. Stone, Physics of Fluids 25, 092102 (2013).

[48] T. T. Al-Housseiny, P. A. Tsai, and H. A. Stone, Nat Phys 8, 747 (2012).

[52] X. Wang, C. Zhu, T. Fu, and Y. Ma, AIChE Journal 61, 1081 (2015).

[44] H. E. Huppert and J. A. Neufeld, Annual Review of Fluid Mechanics 46, 255 (2014).

[45] U. Oxaal, M. Murat, F. Boger, A. Aharony, J. Feder, and T. Jossang, Nature 329, 32

[46] T. T. Al-Housseiny, J. Hernandez, and H. A. Stone, Physics of Fluids 26, 042110

[49] R. Lenormand, C. Zarcone, and A. Sarr, Journal of Fluid Mechanics 135, 337 (1983).

[50] T. T. Al-Housseiny, I. C. Christov, and H. A. Stone, Physical Review Letters 111,

[51] T. Fu, Y. Ma, D. Funfschilling, and H. Z. Li, Chemical Engineering Science 66, 4184

[53] X. Wang, C. Zhu, T. Fu, and Y. Ma, Chemical Engineering Science 111, 244 (2014).

[54] T. W. de Haas, H. Fadaei, U. Guerrero, and D. Sinton, Lab on a Chip 13, 3832 (2013).

Gunther, and E. Kumacheva, Lab on a Chip 11, 3545 (2011).


## **Generation and Evaporation of Microsprays**

## Chin-Tai Chen

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/64756

#### **Abstract**

This chapter aims to comprehensively review the techniques and features of micro-sprays for various applications via micro-droplet generators over decades, especially focusing on the past and present microfluidics. It is organized briefly as below. The background of current research and development about the micro-spray techniques is first introduced, followed by the generation and evaporation of spray detailed with the concentrated re‐ spects of critical requirements for materials and facilities. Then, we address the critical design issues of micro-sprayers such as the actuators and nozzles required to be satisfac‐ tory for generating a number of droplets. Subsequently, we further describe characteriza‐ tion of droplets in form of spray concerning droplet size, speed, rates, and patterns in microfluidics. Moreover, the chapter presents the proof-of-concept and commercial appli‐ cations of the micro-spraying processes, highlighting their current technical progresses and future challenges, which shall be intimately related to the droplet generation and evaporation including droplet evaporative cooling, direct printing, screen printing, nanomaterial coating, liquid nebulization, and miscellaneous employment. Finally, we draw a conclusion in the end of the chapter.

**Keywords:** Spray, droplet generation, droplet evaporation, micro-actuator, microfluidics

## **1. Introduction**

Over the past 30 years, spray droplets have been investigated and engineered due to their frequent occurrence in nature such as sea water [1] and manual generation on hot surfaces such as liquid deposition [2]. This field of science and engineering has widely demonstrated their significance on thermal and fluidic dynamics, in particular related to cooling and evaporation of spray droplets. Those numerous droplets, as being mechanically pumped through small nozzles of a continuous fluid delivery system, have been used for different respects of the evaporative cooling since then. Compared to the conventional mechanical droplet generators, another type of the modern microfluidic generators can eject the micron-

© 2016 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

sized droplets through micro-electro-mechanical actuators that typically feature light weight, small size, less noise, and high power conservation [3]. Furthermore, as bonding with an array of micro-nozzles (orifices), a number of femto- to pico-liter droplets can be simultaneously generated at high driving frequency (kilo-to mega-Hz), forming a highly directional spray away from the nozzle surface [4].

To acquire high quality sprays in many applications (e.g., evaporative cooling), generating the droplets efficiently from certain micro-nozzles to a targeted space or surface exhibits technical requirements for key parameters [1], including the velocity of the droplets, the number flux of droplets, and the Sauter mean diameter. In terms of hydrodynamic properties, a stream of micro-droplets typically experience in either case below: (1) impacting, wetting, and evapo‐ rating (and depositing) on a two-dimensional (2D) surface and (2) floating and evaporating within a three-dimensional (3D) space. In both cases, the intrinsic non-uniformity of a spray generated from a single nozzle is frequently encountered in the conventional liquid delivery system, leading to the poor performance in some applications such as evaporative cooling. For example, non-uniform temperature distribution on the wall surface may be resulted from the different droplet-size distributions (DSDs) [5]. Likewise, it is possible to encounter a complex problem of determining the efficiency of evaporative cooling occurred with non-uniform distributions of droplets [6]. Moreover, non-uniform droplets forming turbulent flows can be also caused by ambient air and pressure within the zone of the spray, thereby significantly influencing the evaporation of spray droplets [7].

On the other hand, in the recent years, many efforts of research and development in academics as well as industries have been made to facilitate the various components and devices fabricated by the spray printing (SP) techniques, including amperometric enzyme electrodes [8], organic transistor electrodes [9], metal silver films [10], counter electrodes for flexible dyesensitized solar cells [11], carbon thin films on flexible substrates [12], organic field-effect transistors and complementary inverters [13], flexible polymer solar cell modules [14], etc. Compared to the current inkjet-printing (IJP) techniques [15], this SP process is generally characterized with one additional screen (mesh) for patterning a variety of materials coated on substrate surfaces. Moreover, the screen printing by spray further provides a capability of large-area patterning on substrates without requirement of precise positioning, and therefore becomes an alternative low-cost technology consuming little material and power. Accordingly, using the SP techniques as a simple manufacturing tool, several global manufacturers have developed the material-processing equipment and peripheral utilities from the laboratory to commercial grade, for example, Legend Star International Co., Ltd. (http://www.lgs.com.tw) and Diamond-MT, Inc. (http://www.diamond-mt.com), in order to offer the customizing design services for academic research and commercial production.

In addition, with the remarkable emergence of nano-science and technology, many functional and novel materials have consequently been exploited to generate specific thin films and nanostructures using this SP method. For instance, a variety of fabrication methods for deposition of thin-film oxides (e.g., yttria-stabilized zirconia, YSZ) such as chemical vapor deposition (CVD), physical vapor deposition (PVD), electrochemical vapor deposition (EVD), spray pyrolysis, slip casting, screen printing, and so forth were comprehensively presented and discussed by Will et al. [16]. Compared to the other methods, spray pyrolysis (of a metal salt solution) can be reliably utilized to form amorphous to polycrystalline microstructures (metal oxide films) at economical cost, although post-thermal treatment is usually necessary. Using this kind of spray techniques, the continuing efforts were made for the film production of copper and hybrid nanoparticles [17], the generation of metal particles [18], and the generation of metal and metal oxide nanoparticles [19]. Furthermore, more extensive studies for spraying applications were also directed to other advanced materials including biomedical calcium phosphate [20], colloidal dispersion of alumina particles [21], and composite materials of silver nanoparticles [22]. Importantly, many of the fundamental principles (e.g., preparation of pure liquid and solution materials) and utilities (e.g., setup of a liquid delivery and spray system) were commonly adopted in another nebulization application, for example, the aerosol and pulmonary drug delivery [23–25]. Therefore, full understanding of fundamental spraying mechanism here is a substantial effort to realize the key performances for such practical applications of the droplet generation techniques.

This book chapter concisely reviews the recent development of spray techniques and applications in our daily life and industry using micro-droplet generation, particularly benefiting the public in many respects such as saving energy, cost and time. It is organized briefly as follows. First, the background of the past and present development associated with the spray techniques is introduced here in Section 1, followed by the elaboration of the spray generation and evaporation regarding various liquids and actuators in Section 2. Section 3 presents the important design issues of micro-actuators/nozzles and integration/control, basically considering the implementation of spray. Section 4 describes the general character‐ istics of droplets existing in spray such as their sizes, speeds, volumetric rates, and fluidic patterns. Section 5 demonstrates the up-to-date applications of the micro-spraying process‐ es, expressing the potential development and challenges in the future, which include droplet evaporative cooling (DEC), direct printing (DP) and screen printing (SP), nanomaterial coating (NMC), liquid nebulization (LN), and miscellaneous utilities. Finally, we draw a conclusion in Section 6.

## **2. Fundamentals of microspraying**

sized droplets through micro-electro-mechanical actuators that typically feature light weight, small size, less noise, and high power conservation [3]. Furthermore, as bonding with an array of micro-nozzles (orifices), a number of femto- to pico-liter droplets can be simultaneously generated at high driving frequency (kilo-to mega-Hz), forming a highly directional spray

316 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

To acquire high quality sprays in many applications (e.g., evaporative cooling), generating the droplets efficiently from certain micro-nozzles to a targeted space or surface exhibits technical requirements for key parameters [1], including the velocity of the droplets, the number flux of droplets, and the Sauter mean diameter. In terms of hydrodynamic properties, a stream of micro-droplets typically experience in either case below: (1) impacting, wetting, and evapo‐ rating (and depositing) on a two-dimensional (2D) surface and (2) floating and evaporating within a three-dimensional (3D) space. In both cases, the intrinsic non-uniformity of a spray generated from a single nozzle is frequently encountered in the conventional liquid delivery system, leading to the poor performance in some applications such as evaporative cooling. For example, non-uniform temperature distribution on the wall surface may be resulted from the different droplet-size distributions (DSDs) [5]. Likewise, it is possible to encounter a complex problem of determining the efficiency of evaporative cooling occurred with non-uniform distributions of droplets [6]. Moreover, non-uniform droplets forming turbulent flows can be also caused by ambient air and pressure within the zone of the spray, thereby significantly

On the other hand, in the recent years, many efforts of research and development in academics as well as industries have been made to facilitate the various components and devices fabricated by the spray printing (SP) techniques, including amperometric enzyme electrodes [8], organic transistor electrodes [9], metal silver films [10], counter electrodes for flexible dyesensitized solar cells [11], carbon thin films on flexible substrates [12], organic field-effect transistors and complementary inverters [13], flexible polymer solar cell modules [14], etc. Compared to the current inkjet-printing (IJP) techniques [15], this SP process is generally characterized with one additional screen (mesh) for patterning a variety of materials coated on substrate surfaces. Moreover, the screen printing by spray further provides a capability of large-area patterning on substrates without requirement of precise positioning, and therefore becomes an alternative low-cost technology consuming little material and power. Accordingly, using the SP techniques as a simple manufacturing tool, several global manufacturers have developed the material-processing equipment and peripheral utilities from the laboratory to commercial grade, for example, Legend Star International Co., Ltd. (http://www.lgs.com.tw) and Diamond-MT, Inc. (http://www.diamond-mt.com), in order to offer the customizing

In addition, with the remarkable emergence of nano-science and technology, many functional and novel materials have consequently been exploited to generate specific thin films and nanostructures using this SP method. For instance, a variety of fabrication methods for deposition of thin-film oxides (e.g., yttria-stabilized zirconia, YSZ) such as chemical vapor deposition (CVD), physical vapor deposition (PVD), electrochemical vapor deposition (EVD), spray pyrolysis, slip casting, screen printing, and so forth were comprehensively presented

away from the nozzle surface [4].

influencing the evaporation of spray droplets [7].

design services for academic research and commercial production.

Considering the common requirements of spray applications, a series of spraying processes should be comprehensively dealt with concerning the formation evolution of a spray: liquid supply and actuation, droplet generation, spraying, evaporating and deposition, as shown in Figure 1. Spraying droplets with different sizes (i.e., s1, s2,…, and sn) may impact and wet on surface (i.e., d1, d2, and dm) through evaporating (and m < n). As a matter of fact, every step here may involve technically serious considerations and significantly influence the performances of a spray-based system in the various applications. Those consideration factors in the successful spraying of materials and components are briefly described, discussed, and analyzed as below.

**Figure 1.** Schematic representation of spray generation and evaporation undergoing typically from one position of liq‐ uid supply and actuation (on the top) to another position of deposition on surface (at the bottom): those numerous droplets with various sizes are evaporating in air (s1, s2,…, and sn) and on surface (d1, d2,…, and dm).

#### **2.1. Liquids and actuators**

For many of applications, selecting suitable pure liquids or solutions is the most important factor in evaluating primary performances of spraying processes such as evaporative cooling efficiency and direct printing of materials. For pure liquids, their dynamic viscosities (e.g., pure water of 1.00 cp and ethanol of 1.09 cp at 20°C, where 1 cp = 1 mNs/m2 ) should be low enough for the liquid delivery (through a tubing system) and ejection (from a powered actuator). Also, their surface energies in air are typically ranging from 20 to 73 mN/m in order to (e.g., pure water of 72.86 mN/m and ethanol of 22.39 mN/m at 20°C). For liquid mixtures, the solid contents (e.g., silver and gold nanoparticles) have to be effectively dispersed in the solvent (i.e., aqueous or non-aqueous) by additional suitable surfactant (e.g., detergents), in which particle aggregation is avoided for the long-term operation [26]. The particle sizes, in general ranging from a few to hundreds of nanometers (nm), are also small to a certain extent de‐ pending on some limits of the fluid delivery pipes and exit nozzles [10].

Second, the means of actuation powerfully pumping (dispensing) the liquid out of the nozzles should be considered for different needs. For instance, some mechanically pneumatic pumps connecting to metal nozzles are able to eject high-throughput sprays robustly but noisily; therefore, they are most often exploited for evaporative cooling in large buildings and facilities [27–29]. However, at smaller scales, many new types of micro actuators such as piezoelectric [3–4,30] and electrostatic [14,31] ones are used to take advantage of precise electrical control over spray. Compared to the former (pneumatic), these microfluidic actuators may be further explored for versatile circumstances of daily life using novel and versatile materials.

Using the fluidic actuators filled with suitable liquids, a number of droplets can be therefore pumped from the small nozzles as electrically powered. In theory, those droplets generally exhibit spherical shapes in air due to intrinsic surface tension of liquid as demonstrated in Figure 1. Thus, the individual volume (*V*) and surface area (*A*) of a formed droplet are geometrically estimated in Eqs. (1) and (2):

$$V = \frac{4}{3}\pi \times \text{R}^3 = \frac{4}{3}\pi \times \left(\frac{D}{2}\right)^3 = \frac{\pi}{6} \times D^3 \tag{1}$$

$$A = 4\pi \times R^2 = 4\pi \times \left(\frac{D}{2}\right)^2 = \pi \times D^2 \tag{2}$$

where *R* is the radius and *D* is the diameter for a single drop, respectively. In general, the droplet size, volume, and surface area here rely on the size of nozzle. The conventional metal nozzles have quite diverse diameters ranging from tens of micrometers (µm) to a few milli‐ meters (mm). By contrast, the micron-sized nozzles used in micro-actuators are generally found with 10–100 µm in diameter. As a result, those droplets are generated to exhibit the corresponding volumes with tens to thousands of pico-liters (pl = 10-12 l), primarily depending on the actuation means and devices. Obviously, the smaller droplets may be also yielded by simply reducing the nozzle diameter.

In fact, except for the called drop-on-demand (DOD) operation that is often used in IJP [32,33], the uniformity of size of individual droplets is one of the crucial factors considered for precise spraying in many applications [15]. For example, the thermal performances of a spray such as critical heat flux (CHF) can be significantly influenced by the deviation from the targeted value of the droplet size [34]. Hence, in many situations, it shall be substantially coped with the droplet size distribution as well as evaporation as follows.

#### **2.2. Droplet size distribution and evaporation**

**Figure 1.** Schematic representation of spray generation and evaporation undergoing typically from one position of liq‐ uid supply and actuation (on the top) to another position of deposition on surface (at the bottom): those numerous

For many of applications, selecting suitable pure liquids or solutions is the most important factor in evaluating primary performances of spraying processes such as evaporative cooling efficiency and direct printing of materials. For pure liquids, their dynamic viscosities (e.g., pure

for the liquid delivery (through a tubing system) and ejection (from a powered actuator). Also, their surface energies in air are typically ranging from 20 to 73 mN/m in order to (e.g., pure water of 72.86 mN/m and ethanol of 22.39 mN/m at 20°C). For liquid mixtures, the solid contents (e.g., silver and gold nanoparticles) have to be effectively dispersed in the solvent (i.e., aqueous or non-aqueous) by additional suitable surfactant (e.g., detergents), in which particle aggregation is avoided for the long-term operation [26]. The particle sizes, in general ranging from a few to hundreds of nanometers (nm), are also small to a certain extent de‐

Second, the means of actuation powerfully pumping (dispensing) the liquid out of the nozzles should be considered for different needs. For instance, some mechanically pneumatic pumps

) should be low enough

droplets with various sizes are evaporating in air (s1, s2,…, and sn) and on surface (d1, d2,…, and dm).

water of 1.00 cp and ethanol of 1.09 cp at 20°C, where 1 cp = 1 mNs/m2

318 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

pending on some limits of the fluid delivery pipes and exit nozzles [10].

**2.1. Liquids and actuators**

The general droplet size distribution (DSD) in conventional sprays can be represented by classic Mugele-Evan (ME) relation as described in Eq. (3) [29,35], where *f*N(*D*) is the number (*N*) density distribution, and *i* and *j* are positive integers. Hence, the arithmetic mean diameter (AMD) (i.e., *D10*, where *i* = 1, *j* = 0) is expressed in Eq. (4). Also, the well-known Sauter mean diameter (SMD) is defined for *D32* (where *i* = 3, *j* = 2) as shown in Eq. (5) [36]:

$$D\_{ij} = \left[\frac{\int\_0^\alpha D^i f\_N(D) dD}{\int\_0^\alpha D^i f\_N(D) dD}\right]^{\frac{1}{\binom{n}{\alpha}}} \tag{3}$$

$$A \text{MD} = D\_{10} = \int\_0^1 D f\_N(D) dD = \int\_0^1 D dN \tag{4}$$

$$\text{SMD} = D\_{32} = \frac{\int\_0^{\prime\prime} D^3 f\_N(D) dD}{\int\_0^{\prime\prime} D^2 f\_N(D) dD} = \frac{\int\_0^{\prime\prime} D^3 dN}{\int\_0^{\prime\prime} D^2 dN} \tag{5}$$

As the simplest case of a uniform spray, all of the droplets have the same diameter of *Ds*, thereby giving the AMD = *D10* = *Ds*. Furthermore, in this special case, the SMD (i.e., *D32*) can be clearly represented by the ratio of the single droplet volume (*V*)/surface area (*A*): *D32* = 6×(*V*/*A*), according to previous Eqs. (1) and (2). Based on this concept of the ratio of *V*/*A*, the SMD shows particular significance in some circumstances, where the surface area plays a critical role such as droplet evaporation time in air, as shown in Eq. (6) [1]:

$$
\pi = \beta \times \frac{\pi \times \text{D}^2\_{\text{s}}}{a(T)} = \frac{\beta \times A}{a(T)} \tag{6}
$$

where *β* is the dimensionless coefficient, *Ds* is the single droplet diameter, *α* is the thermal diffusivity of the droplet, and *T* is the temperature, respectively. For example, supposed that the water with the *β* = 3.23×10-2 and *α* = 1.43×10-7 m2 /s at 25 °C, a droplet of *Ds* = 100 µm is estimated with the evaporation time of *τ* = 7.10×10-3 s (i.e., 7.1 ms), while a shorter time of *τ* = 7.10×10-5 s (i.e., 71 µs) for a droplet with a 10-time smaller diameter (i.e., *D*s = 10 µm).

#### **2.3. Spraying control**

In regard to the spraying control, it is mostly concerned with the volumetric rate of a spray. As can be seen in Figure 1, some of the droplets ejected out of nozzles (i.e., s1, s2, and sn) might reach, deposit, and grow on certain surfaces (i.e., d1, d2, and dm), even though they fulfill the evaporation (in air) as mentioned above. Therefore, real-time control over the spraying is another crucial part of the microfluidic actuation. For a pneumatic pumping/dispensing system, switching the actuation of a spray on and off is frequently completed in a period of a few seconds, in order to get rid of excessive growth of deposit.

To achieve high-speed switching, however, digital control over micro actuation can be carried out precisely for the new types of microfluidic actuators, where they can operate electronically at driving frequencies more than tens of kHz [30]. Utilizing the present computer-assisted capability over one decade, those control systems have been developed and implemented with graphical user interface (GUI) by using commercial software such as MATLAB, LabVIEW, C, C++, and Visual Basic languages.

Based on the fundamental knowledge of spray science and technology, two preliminary design issues of micro-sprayers for realizing precise droplet generation are further addressed and discussed below.

## **3. Design issues of microsprayers**

## **3.1. Micro-actuators and nozzles**

1

<sup>10</sup> 0 0 ( ) *AMD D Df D dD DdN <sup>N</sup>* == = ò ò (4)

ò ò (5)

´ ´ =´ = (6)

/s at 25 °C, a droplet of *Ds* = 100 µm is

(3)

( )

¥ -

*i j i N*

( )

1 1

( )

¥ ¥

( ) *N*

As the simplest case of a uniform spray, all of the droplets have the same diameter of *Ds*, thereby giving the AMD = *D10* = *Ds*. Furthermore, in this special case, the SMD (i.e., *D32*) can be clearly represented by the ratio of the single droplet volume (*V*)/surface area (*A*): *D32* = 6×(*V*/*A*), according to previous Eqs. (1) and (2). Based on this concept of the ratio of *V*/*A*, the SMD shows particular significance in some circumstances, where the surface area plays a critical role such

2

p

a

() () *Ds A T T*

where *β* is the dimensionless coefficient, *Ds* is the single droplet diameter, *α* is the thermal diffusivity of the droplet, and *T* is the temperature, respectively. For example, supposed that

estimated with the evaporation time of *τ* = 7.10×10-3 s (i.e., 7.1 ms), while a shorter time of *τ* =

In regard to the spraying control, it is mostly concerned with the volumetric rate of a spray. As can be seen in Figure 1, some of the droplets ejected out of nozzles (i.e., s1, s2, and sn) might reach, deposit, and grow on certain surfaces (i.e., d1, d2, and dm), even though they fulfill the evaporation (in air) as mentioned above. Therefore, real-time control over the spraying is another crucial part of the microfluidic actuation. For a pneumatic pumping/dispensing system, switching the actuation of a spray on and off is frequently completed in a period of a

To achieve high-speed switching, however, digital control over micro actuation can be carried out precisely for the new types of microfluidic actuators, where they can operate electronically at driving frequencies more than tens of kHz [30]. Utilizing the present computer-assisted capability over one decade, those control systems have been developed and implemented with

7.10×10-5 s (i.e., 71 µs) for a droplet with a 10-time smaller diameter (i.e., *D*s = 10 µm).

b

> a

*N*

*D f D dD D dN SMD D*

¥ ¥ == = ò ò

3 3 0 0 <sup>32</sup> 2 2 0 0

*D f D dD D dN*

*D f D dD*

*N*

é ù ê ú <sup>=</sup> ê ú ê ú ë û

0

ò

320 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

¥

*D f D dD <sup>D</sup>*

*ij <sup>j</sup>*

as droplet evaporation time in air, as shown in Eq. (6) [1]:

the water with the *β* = 3.23×10-2 and *α* = 1.43×10-7 m2

**2.3. Spraying control**

t b

few seconds, in order to get rid of excessive growth of deposit.

ò

0

As illustrated previously in Figure 1, the liquid supply and actuation forming a spray have to be performed at the first stage. For the purpose of effective formation of a spray as exemplified in Figure 2, the fluidic systems with different complexity can be designed and built here. Figure 2(a) shows an infrared (IR) thermography picture (NEC G100, JP) of a conventional pneumatic system (red) mainly composed of liquid supply (through tubing) and a metal nozzle (exit diameter of 80 µm, GAU PU GP-A008, TW). As being pressurized (70 kg/cm2 ) from an electric pump (at 110 V, 850 rpm), droplet jetting of a water spray with a rate of 22 ml/min was continuously formed in a conical shape (see S). Meanwhile, it generated a loud noise of 85 dB from 50 dB (due to large mechanical vibration) and high humidity of 80 RH% from 60 RH% (due to excessive growth of deposit) in the neighboring space, substantially posing a notable disadvantage in some applications of the public daily life.

To avoid the induced noise and humidity, the actuators and nozzles can be largely miniatur‐ ized and arrayed for generation of a spray at the micron scale, as demonstrated in Figure 2(b). Here, the commercial ink cartridge installed in a bubble-type inkjet printer (Canon PIXMA iP 2770, JP) is composed of a print-head with a resolution of 1,200 × 4,200 dpi (1,472 fine nozzles, droplet volume of 2 pl). After de-capping the printer and removing the original ink from the cartridge, then pure water was manually re-filled from the liquid supply of a tubing container. By virtue of a microelectronic controller of the printer, the print-head was successfully actuated in scanning motion, so that water droplet jetting of 1 ml/min from its array of nozzles was achieved (see S). Compared to the previous one, a lower noise of 60 dB as well as humidity of 60 RH% was obtained in the same environment, indicating a great improvement in the spraying performance (i.e., decrease by 25 dB and 20 RH%) because of more uniform and fine droplets.

#### **3.2. Microsystem integration and control**

As using micro-actuators and nozzles, another design issue on the implementation of a microsprayer includes the microsystem integration and control required for many situations [3]. For example, as exemplified in Figure 3(a), an electrical power supply and a signal generator were equipped and controlled here to adaptively vary the driving voltage (*vdrive,* V) and frequency (*fdrive*, kHz) for a micro-actuator (e.g., piezoelectric actuation). Consequently, droplet ejection of a spray is able to be finely tuned for optimal performance [4,30]. The droplet

**Figure 2.** Demonstration and comparison with two types of spraying components and systems: (a) typical jetting proc‐ ess of water generated from liquid supply to a metal nozzle using a mechanical pump and (b) water droplets jetting from a liquid supply to a print-head cartridge with micro-actuators and nozzles using an inkjet printer.

velocity and volumetric rate for a spray may increase as raising the values of *vdrive* and *fdrive*. In addition, the compact liquid supply system comprising a tube and a container can be inte‐ grated and packaged with this microfluidic actuator (symbolized by A), as schematically depicted in Figure 3(b). Following the way, the fluid was expected to be slowly delivered (with an inner tubing diameter of 10 mm) from the bottom inlet to the top outlet (total vertical distance of 120 mm) as powered by microelectronic control and actuation (at *vdrive* and *fdrive*).

Such microsystem integration and control was realized by using a piezoelectric actuator under a suitable condition. As demonstrated in the IR thermography of Figure 3(c), one upward water spray was continuously generated in a period of 180 s from the fluidic outlet of the actuator (at *vdrive* of 60 V, *fdrive* of 90 kHz). However, their morphologic features changed apparently over time (i.e., 30, 90, and 180 s) implying the variation in physical properties (e.g., velocity and rate) and geometric pattern (e.g., turbulent flow). Hence, those characteristics of droplets within a spray are described and analyzed next in Section 4.

**Figure 3.** Integration, packaging, and test for a microsprayer: (a) electrical power supply and signal generator for con‐ trolling the micro-actuator during spraying, (b) liquid supply container integrated with the micro-actuator, and (c) thermography of droplet jetting with the electrical and liquid supply.

## **4. Characteristics of droplets in spray**

From the view point of microfluidics, two significant respects can reflect the behaviour of a spray. One is the size and velocity of droplets, and the other is the rate and pattern of a spray. The former is primarily concerned with the kinematics of individual micro-droplets, while the latter is particularly emphasized to take into account the macrospraying phenomenon. Since either micro- or macro-behavior is important for a spray, one shall recognize their distin‐ guishing characteristics entirely. Hence, the droplets in spray can be further elaborated and characterized as follows.

#### **4.1. Size and velocity of a droplet**

velocity and volumetric rate for a spray may increase as raising the values of *vdrive* and *fdrive*. In addition, the compact liquid supply system comprising a tube and a container can be inte‐ grated and packaged with this microfluidic actuator (symbolized by A), as schematically depicted in Figure 3(b). Following the way, the fluid was expected to be slowly delivered (with an inner tubing diameter of 10 mm) from the bottom inlet to the top outlet (total vertical distance of 120 mm) as powered by microelectronic control and actuation (at *vdrive* and *fdrive*).

**Figure 2.** Demonstration and comparison with two types of spraying components and systems: (a) typical jetting proc‐ ess of water generated from liquid supply to a metal nozzle using a mechanical pump and (b) water droplets jetting

from a liquid supply to a print-head cartridge with micro-actuators and nozzles using an inkjet printer.

322 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

Such microsystem integration and control was realized by using a piezoelectric actuator under a suitable condition. As demonstrated in the IR thermography of Figure 3(c), one upward water spray was continuously generated in a period of 180 s from the fluidic outlet of the actuator (at *vdrive* of 60 V, *fdrive* of 90 kHz). However, their morphologic features changed apparently over time (i.e., 30, 90, and 180 s) implying the variation in physical properties (e.g., velocity and rate) and geometric pattern (e.g., turbulent flow). Hence, those characteristics of droplets

within a spray are described and analyzed next in Section 4.

Conventionally, the size and velocity of a droplet can be visualized and calculated from particle image velocimetry (PIV), in which an expensive digital camera is used to perform the highspeed imaging capture (frame per second, *fps*, up to 10,000) for all droplets [29]. In this way, the droplet size distribution (DSD) and the others such as AMD and SMD can be calculated accordingly. From a practical view of convenience, while using a micro actuator mentioned previously, the initial velocity (v0) of droplets on average can be simply evaluated as first-order approximation as proposed in Eq. (7), where *η* is the velocity coefficient, *hnozzle* is the nozzle (via hole) thickness, and *fdrive* is the driving frequency, respectively. Moreover, assuming that the droplets travel a distance (*sa*) under constant acceleration (*a*), this value (*v0*) can be also correlated with the final average velocity (*vf* ) as expressed physically in Eq. (8):

$$
\omega\_0 = \eta \times (h\_{\text{nozzle}} \times f\_{\text{drive}}) \tag{7}
$$

$$\mathbf{v}\_f^2 = \mathbf{v}\_0^2 + \mathbf{2} \times \mathbf{a} \times \mathbf{s}\_a \tag{8}$$

For instance, as demonstrated in Figure 4, a water spray was ejected from a piezoelectric microactuator at *fdrive* = 90 kHz, in which micro-nozzles featured an average thickness (*hnozzle*) of 40 µm and an average diameter (*dnozzle*) of 30 µm [37]. Using Eq. (7), the value of (*v0/η*) was approximated as (30 µm)×(90×103 /s) = 3.60 m/s. Thus, under normal gravity (i.e., *a* = 9.8 m/s2 ) as demonstrated in Figure 4(a), this upward spray actually traveled a maximum distance *sa* = 0.12 m to reach the final velocity *vf* = 0. From Eq. (9) as ignoring the friction of air, the square of initial velocity was approximately calculated as 2×(9.8 m/s2 )×(0.12 m) = 2.35 m2 /s2 , yielding *v*0 = 1.53 m/s and *η* = (1.53 m/s)/(3.60 m/s) = 0.425 (i.e., 42.5%); therefore, this high velocity hindered those droplets from being clearly imaged by normal optical cameras (*fps* = 30) as illustrated in Figure 4(b).

**Figure 4.** A group of droplets sprayed from a micro-actuator: (a) the upward formation of a water spray and (b) the downward droplet jetting from the micro-nozzles.

#### **4.2. Rate and pattern of a spray**

As a whole, the volumetric rate (*ψr*) of a microspray generated from the nozzles is straight‐ forwardly determined by the droplet density (flux), driving frequency, and voltage. Again, using previous case as one example, the value of *ψr* for single droplet volume of 30 pl and firing nozzle number of 100 was estimated with a maximum of (30 pl)×100×(90×103 /s) = 0.27 ml/s = 16.20 ml/min. Nevertheless, this rate might drop dramatically due to failure of some nozzle during micro-actuation as shown in the inset of Figure 4(b). More importantly, the flow pattern of a spray was actually influenced by the flight velocity of droplets in air, typically leading to an appearance of turbulence [7,38]. In terms of fluid dynamics, this phenomenon pattern is associated particularly with a dimensionless Reynolds number (Re) as described in Eq. (9) [39]:

$$\text{Re} = \frac{\rho \times v \times l}{\mu} \tag{9}$$

where *ρ* is the density of liquid, *v* is the droplet velocity, *l* is the characteristic length, and *µ* is the dynamic viscosity of liquid, respectively. For instance, Figure 5 shows PIV imaging for patterns of sprays generated from micro-nozzles (*v* = 3 m/s), in which the liquid of pure water exhibited the *ρ* = 103 kg/m3 and *µ* = 10-3 Ns/m2 . Hence, the sparse spray (*l* < 1 mm) in Figure 5(a) was estimated with a low Re (< 3,000) such that it appeared like a laminar flow, while the turbulent flow with a large Re (> 15,000) was visualized for the dense spray (*l* < 5 mm) in Figure 5(b). Therefore, taking all of the characteristics of droplets into account, a variety of spray applications are described and illustrated in the next section.

**Figure 5.** Various volume rates and microfluidic patterns of flow generated from the micro-nozzles of a micro-actuator: (a) a sparse spray and (b) a dense spray.

## **5. Applications**

<sup>0</sup> ( ) *nozzle drive v hf* =´ ´

For instance, as demonstrated in Figure 4, a water spray was ejected from a piezoelectric microactuator at *fdrive* = 90 kHz, in which micro-nozzles featured an average thickness (*hnozzle*) of 40 µm and an average diameter (*dnozzle*) of 30 µm [37]. Using Eq. (7), the value of (*v0/η*) was

as demonstrated in Figure 4(a), this upward spray actually traveled a maximum distance *sa* = 0.12 m to reach the final velocity *vf* = 0. From Eq. (9) as ignoring the friction of air, the square

*v*0 = 1.53 m/s and *η* = (1.53 m/s)/(3.60 m/s) = 0.425 (i.e., 42.5%); therefore, this high velocity hindered those droplets from being clearly imaged by normal optical cameras (*fps* = 30) as

**Figure 4.** A group of droplets sprayed from a micro-actuator: (a) the upward formation of a water spray and (b) the

As a whole, the volumetric rate (*ψr*) of a microspray generated from the nozzles is straight‐ forwardly determined by the droplet density (flux), driving frequency, and voltage. Again, using previous case as one example, the value of *ψr* for single droplet volume of 30 pl and firing nozzle number of 100 was estimated with a maximum of (30 pl)×100×(90×103 /s) = 0.27

(7)

<sup>0</sup> 2 *f a v v as* = + ´´ (8)

/s) = 3.60 m/s. Thus, under normal gravity (i.e., *a* = 9.8 m/s2

)×(0.12 m) = 2.35 m2

/s2

, yielding

)

h

324 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

2 2

of initial velocity was approximately calculated as 2×(9.8 m/s2

approximated as (30 µm)×(90×103

downward droplet jetting from the micro-nozzles.

**4.2. Rate and pattern of a spray**

illustrated in Figure 4(b).

#### **5.1. Droplet evaporative cooling**

To date, droplet evaporative cooling (DEC) has been one of alternative means for reducing temperature of objects at low cost. It has been elaborated in decreasing the hot solid surfaces with high values of CHF of 250–350 W/cm2 [2,34]. Also, many other efforts have been made in the development of the spatial cooling systems, including the mechanical draft cooling tower and air-conditioning systems [28–29,40], in order to enhance the convective heat transfer in ambient environment. Figure 6(a) illustrates the photography pictures of such spray cooling circumstances, in which the conventional pumping liquid systems were operating indoors or outdoors. Also, using micro-actuators shown in Figure 6(b), the droplet ejection of water was periodically performed (between on and off) under ambient environment to demonstrate the temperature decrease of ~3°C visualized from the IR thermography [41].

Figure 6. Evaporative cooling of droplets: (a) photography of water spray from traditional liquid supply outdoors and (b) temperature evolution of four positions (A, B, C, and D) over **Figure 6.** Evaporative cooling of droplets: (a) photography of water spray from traditional liquid supply outdoors and (b) temperature evolution of four positions (A, B, C, and D) over time within a spraying space, switching the micro‐ electronic actuator on (2 min) and off (3 min) alternatively during a period of 5 min.

time within a spraying space, switching the microelectronic actuator on (2 min) and off (3 min) alternatively during a period of 5 min. In addition to the sprays of water, there are many colour liquids such as dye inks commercially available to be further adapted for industrial applications. Likewise, those inks have been used for direct and screen printing through generation of sprays as further discussed in the following.

### **5.2. Direct and screen printing**

with high values of CHF of 250–350 W/cm2

[2,34]. Also, many other efforts have been made in

the development of the spatial cooling systems, including the mechanical draft cooling tower and air-conditioning systems [28–29,40], in order to enhance the convective heat transfer in ambient environment. Figure 6(a) illustrates the photography pictures of such spray cooling circumstances, in which the conventional pumping liquid systems were operating indoors or outdoors. Also, using micro-actuators shown in Figure 6(b), the droplet ejection of water was periodically performed (between on and off) under ambient environment to demonstrate the

(a)

(b)

Figure 6. Evaporative cooling of droplets: (a) photography of water spray from traditional liquid supply outdoors and (b) temperature evolution of four positions (A, B, C, and D) over time within a spraying space, switching the microelectronic actuator on (2 min) and off (3

In addition to the sprays of water, there are many colour liquids such as dye inks commercially available to be further adapted for industrial applications. Likewise, those inks have been used for direct and screen printing through generation of sprays as further discussed in the

**Figure 6.** Evaporative cooling of droplets: (a) photography of water spray from traditional liquid supply outdoors and (b) temperature evolution of four positions (A, B, C, and D) over time within a spraying space, switching the micro‐

min) alternatively during a period of 5 min.

following.

electronic actuator on (2 min) and off (3 min) alternatively during a period of 5 min.

temperature decrease of ~3°C visualized from the IR thermography [41].

326 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

Instead of evaporating the liquid fully, ink deposition can be applied for printing as using diluted solutions with suitable solid content [42]. As a matter of fact, many printing methods have existed to date, mainly including the contact doctor-blade printing (DBP) method [43– 46] and non-contact IJP method [8,33,47]. As compared to those DBP and IJP above, this kind of spray printing (SP) features the non-contact, high-speed, and large-area patterning capa‐ bility [9,13]. With the assistance of one additional screen (mesh), the SP technique can be efficiently applied to perform the fine and complex patterning of micro-devices such as flexible dye-sensitized solar cells, polymer solar cell modules, and so forth [11,14].

For instance, Figure 7 demonstrates the spray printing of blue dye ink on a regular paper (A4 size) using a piezoelectric actuator and a mobile platform, in which direct printing of a spiral pattern (line width ~10 mm) and screen printing of a comb-type pattern (line width ~400 µm) were simply performed via a microelectronic controller (not shown here). Furthermore, fusing the novel materials from the emergence of nano-science and technology, those synthesized nanomaterials are suitable to be deposited and coated on substrates using the SP method for direct generation of thin functional films that is further described below.

**Figure 7.** Spray printing of blue ink on a regular paper using a piezoelectric actuator: (a) direct printing of a square spiral pattern and (b) screen printing of a comb-type pattern.

#### **5.3. Nanomaterial coating**

The spray technique used as one of the deposition methods was thoroughly compared and analyzed with regard to solid oxide fuel cells (SOFC) applications by Will et al. [16], in which the formations of high-quality chemical composition films were emphasized. Besides those applications, a variety of nanomaterial coating with different chemical components were utilized in the study of spray printing (SP), which include nano-Cu particles and twocomponent Cu/Cu2O [17], Cu and Ni particles [18], biomedical CaP ceramics [20], colloidal alumina particles [21], and silver (Ag) particles [22]. As shown in Figure 8, the silver ink diluted with Ag nanoparticles was spray-printed on a flexible polyimide substrate, and then the dried film was formed after sintering. As can be seen clearly here in Figure 8(a), the grain-like boundaries were created due to insufficient adhesion of coating on substrate surface; the Ag particles typically less than 100 nm in size were also obtained as seen from the scanning electron microscope (SEM) picture in Figure 8(b). Poor electrical conductivity might be therefore expected in this case due to the non-uniformity of film morphology.

**Figure 8.** (a) Spray printing and sintering of silver (Ag) nanoparticle ink on polyimide surface and (b) magnified mor‐ phology of the printed Ag surface demonstrating the typical particle size less than 100 nm.

Obviously, it is noted here that the nanomaterial coating by spraying solutions may suffer from the weak interfacial strength between two layers of different materials. Nevertheless, com‐ pared to conventional deposition techniques such as PVD and CVD, this material coating application no doubt benefits low cost and time for processing, and therefore the interfacial problem deserves to be further improved in the future.

#### **5.4. Liquid nebulization**

For years, liquid nebulization has been generally used as an effective treatment for human airway disease such as panting. Mostly, with synthesis of different component concentrations and functions, aerosol delivery of pulmonary drug to human lungs is performed through spraying for medical therapy [48–49]. Because of the breathing need for patient therapy, those drug droplets of a spray generated from medical nebulizers typically feature an extremely small size (< 5 µm) to yield a high efficient absorption, in which the delivered amount of active drug could be as low as 20 mg for 4 h. Furthermore, micro actuators integrated with micronsized nozzles are commonly applied in the nebulizers that have been commercialized and marketed over 10 years. As demonstrated in Figure 9, the portable and hand-held nebulizers with a compact size (e.g., 6.3 cm×6.4 cm×15 cm for NH60) are available for quietly generating aerosol of different drugs such as albuterol, ipratropium, sulfate, etc. Aerosol performance on nebulization rate can be adjusted and controlled by users.

**Figure 9.** (a) Packet AirTM portable nebulizer by Microbase Inc. and (b) handheld tubeless nebulizer NH60 by Rossmax International Ltd.

#### **5.5. Miscellaneous**

with Ag nanoparticles was spray-printed on a flexible polyimide substrate, and then the dried film was formed after sintering. As can be seen clearly here in Figure 8(a), the grain-like boundaries were created due to insufficient adhesion of coating on substrate surface; the Ag particles typically less than 100 nm in size were also obtained as seen from the scanning electron microscope (SEM) picture in Figure 8(b). Poor electrical conductivity might be therefore

**Figure 8.** (a) Spray printing and sintering of silver (Ag) nanoparticle ink on polyimide surface and (b) magnified mor‐

Obviously, it is noted here that the nanomaterial coating by spraying solutions may suffer from the weak interfacial strength between two layers of different materials. Nevertheless, com‐ pared to conventional deposition techniques such as PVD and CVD, this material coating application no doubt benefits low cost and time for processing, and therefore the interfacial

For years, liquid nebulization has been generally used as an effective treatment for human airway disease such as panting. Mostly, with synthesis of different component concentrations and functions, aerosol delivery of pulmonary drug to human lungs is performed through spraying for medical therapy [48–49]. Because of the breathing need for patient therapy, those drug droplets of a spray generated from medical nebulizers typically feature an extremely small size (< 5 µm) to yield a high efficient absorption, in which the delivered amount of active drug could be as low as 20 mg for 4 h. Furthermore, micro actuators integrated with micronsized nozzles are commonly applied in the nebulizers that have been commercialized and marketed over 10 years. As demonstrated in Figure 9, the portable and hand-held nebulizers with a compact size (e.g., 6.3 cm×6.4 cm×15 cm for NH60) are available for quietly generating aerosol of different drugs such as albuterol, ipratropium, sulfate, etc. Aerosol performance on

expected in this case due to the non-uniformity of film morphology.

328 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

phology of the printed Ag surface demonstrating the typical particle size less than 100 nm.

problem deserves to be further improved in the future.

nebulization rate can be adjusted and controlled by users.

**5.4. Liquid nebulization**

In fact, all applications of the spray science and technology are quite versatile in many respects beyond those mentioned earlier [50–52]. The new development and progress in spray science and techniques can continue to be applied to more novel compound materials and microstructured devices such as those demonstrated in the solar and fuel cells. Yet, the other multidisciplines and creative ideas are being further blended into the field of sprays, perhaps yielding more fruitful applications in the future.

### **6. Concluding remarks**

Over decades, many studies of science and technology have demonstrated the sprays to serve as an effective and efficient means for generation and evaporation of droplets. Many funda‐ mental factors considered for pursuing the high-quality spraying involve appropriate selec‐ tions of working liquids, operating actuators, droplet generation and evaporation, and spraying control. Preliminary design issues of microsprays concerning micro-actuators, nozzles, system integration, and control can be overcome by understanding and realizing fundamentals of working principles. Many applications in daily life and industry that include droplet evaporative cooling (DEC), direct and screen printing (SP), nanomaterial coating (NMC), and liquid nebulization (LN) have been demonstrated using spray techniques. In the future, this knowledge of science and technologies blended with the emergence of novel materials and devices may be promoted to the next generation of sprayers with higher performance than ever.

## **Acknowledgements**

The author thanks research grant for this work partially by the Ministry of Science and Technology (MOST) under MOST 104-2221-E-151-036-MY2, Taiwan, ROC.

## **Author details**

Chin-Tai Chen\*

Address all correspondence to: chintai@kuas.edu.tw

Department of Mechanical Engineering, National Kaohsiung University of Applied Sciences, Kaohsiung, Taiwan, ROC

## **References**


[9] Jang Y, Park YD, Lim JA, Lee HS, Lee WH, Cho K. Patterning the organic electrodes of all-organic thin film transistors with a simple spray printing technique. Appl. Phys. Lett. 2006; 89: 183501.

**Acknowledgements**

**Author details**

Chin-Tai Chen\*

**References**

481-497.

263-270.

The author thanks research grant for this work partially by the Ministry of Science and

Technology (MOST) under MOST 104-2221-E-151-036-MY2, Taiwan, ROC.

330 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

Department of Mechanical Engineering, National Kaohsiung University of Applied

ings of 18th IEEE SEMI-THERM Symposium; 2002. p. 169-175.

ejector arrays. IEEE Trans. Semicond. Manuf. 2005;18: 709-714.

tics of fuel spray by a swirl type atomizer. Fuel. 2011; 90: 2367-2376.

[1] Andreas E. Time constants for the evolution of sea spray droplets. Tellus. 1990; 42B:

[2] Marcos A, Chow L, Du JH, Lei S. Spray cooling at low system pressure. In: Proceed‐

[3] Yang JC, Chien W, King M,Grosshandler WL. A simple piezoelectric droplet genera‐

[4] Demirci U, Goksenin G, Yaralioglu G, Haggstrom E, Khuri-Yakub, BT. Femtoliter to picoliter droplet generation for organic polymer deposition using single reservoir

[5] Sommerfeld M, Qiu, HH. Experimental studies of spray evaporation in turbulent

[6] Fisenko SP, Wang WN, Lenggoro IW, Okyuama K. Evaporative cooling of micronsized droplets in a low-pressure aerosol reactor. Chem. Eng. Sci. 2004; 61: 6029-6034.

[7] Li T, Nishida K, Hiroyasu H. Droplet size distribution and evaporation characteris‐

[8] Hart AL, Turner APF, Hopcroft D. On the use of screen- and ink-jet printing to pro‐ duce amperometric enzyme electrodes for lactate. Biosens. Bioelectron.1996; 11:

Address all correspondence to: chintai@kuas.edu.tw

tor.Exp. Fluids. 1997; 23: 445-447.

flow. Int. J. Heat. Fluid. Flow. 1998; 19: 10-22.

Sciences, Kaohsiung, Taiwan, ROC


[36] Semiao V, Andrade P, da GracaCarvalho M. Spray characterization: numerical pre‐ diction of Sauter mean diameter and droplet size distribution. Fuel. 1996; 75: 1707-1714.

[21] Sen D, Mazumder S, Melo JS, Khan A, Bhattyacharya S, D'Souza SF. Evaporation driven self-assembly of a colloidal dispersion during spray drying: volume fraction

[22] Park M, Im J, Shin M, Min Y,Park J, Cho H, Park S, Shim MB, Jeon S, Chung DY, Bae J, Park J, Jeong U, Kim K. Highly stretchable electric circuits from a composite mate‐ rial of silver nanoparticles and elastomeric fibres, Nat. Nanotechnol. 2012; 7: 803-809.

[23] McPeck M, Tandon R, Hughes K, Smaldone GC. Aerosol delivery during continuous

[24] Wiedmann TS, DeCastro L, Wood RW. Nebulization of nanocrystals: production of a respirable solid-in-liquid-in-air colloidal dispersion. Pharm. Res. 1997; 14: 112-116.

[25] Hertel SP, Winter G, Friess W. Protein stability in pulmonary drug delivery via nebu‐

[26] Erbil HY. Evaporation of pure liquid sessile and spherical suspended drops: a re‐

[27] Kang D, Strand RK. Modeling of simultaneous heat and mass transfer within passive down-draft evaporative cooling (PDEC) towers with spray in FLUENT. Energy

[28] Koseoglu MF. Investigation of water droplet carryover phenomena in industrial evaporative air-conditioning systems. Int. Commun. Heat Mass Transf. 2013; 47:

[29] Montazeri H, Blocken B, Hensen JLM. Evaporative cooling by water spray systems: CFD simulation, experimental validation and sensitivity analysis. Build. Environ.

[30] Chen CT, Wang HY. Droplet generation and evaporative cooling using micro piezo‐ electric actuators with ring-surrounded circular nozzles. Microsyst. Technol. 2015; 21:

[31] Rahman K, Ko JB, Khan S, Kim DS, Choi KH. Simulation of droplet generation

[32] Li R, Ashgriz N, Chandra S. Droplet generation from pulsed micro-jets. Exp. Therm.

[33] Hu BC, Weng RH, Chen CT. Silver conducting lines of dye-sensitized solar cells

[34] Sawyer ML, Jeter SM, Abdel-Khalik SI. A critical heat flux correlation for droplet im‐

[35] Mugele RA, Evans HD. Droplet size distribution in sprays. Ind. Eng. Chem. 1951; 43:

through electrostatic forces. J. Mech. Sci. Technol. 2010; 24: 307-310.

printed onto commercial building tiles. Proc. Eng. 2014: 79: 267-272.

pact cooling. Int. J. Heat Mass Transf. 1997; 40: 2123-2131.

dependent morphological transition. Langmuir. 2009; 25: 6690-6695.

nebulization. Chest. 1997; 111: 1200-1205.

332 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

lization. Adv. Drug Deliv. Rev. 2015; 93: 79-94.

view. Adv. Colloid Interface Sci. 2012; 170: 67-86.

Build. 2013; 62: 196-209.

2015; 83: 129-141.

Fluid Sci. 2008; 32: 1679-1686.

2067-2075.

1317-1324.

92-97.


#### **Overview of Materials for Microfluidic Applications Overview of Materials for Microfluidic Applications**

Emmanuel Roy, Antoine Pallandre, Bacem Zribi, Marie‐Charlotte Horny, François Damien Delapierre, Andrea Cattoni, Jean Gamby and Anne‐Marie Haghiri‐Gosnet Emmanuel Roy, Antoine Pallandre, Bacem Zribi, Marie‐Charlotte Horny, François Damien Delapierre, Andrea Cattoni, Jean Gamby and Anne‐Marie Haghiri‐Gosnet

Additional information is available at the end of the chapter Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/65773

#### **Abstract**

[50] Sazhin SS, Martynov SB, Kristyadi T, Crua C, Heikal MR. Diesel fuel spray penetra‐ tion, heating, evaporation and ignition: modelling vs. experimentation. Int. J. Eng.

[51] Li T, Nishida K, Hiroyasu H. Droplet size distribution and evaporation characteris‐

[52] Tveen-Jensen K, Gesellchen F, Wilson R, Spickett CM, Cooper JM, Pitt PR. Interfacing low-energy SAW nebulization with liquid chromatography-mass spectrometry for

tics of fuel spray by a swirl type atomizer. Fuel. 2011; 90: 2367-2376.

the analysis of biological samples. Sci. Rep. 2015; 5: 9736.

334 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

Syst. Model. Simul. 2008;1: 1-19.

For each material dedicated to microfluidic applications, inherent microfabrication and specific physico‐chemical properties are key concerns and play a dominating role in further microfluidic operability. From the first generation of inorganic glass, silicon and ceramics microfluidic devices materials, to diversely competitive polymers alternatives such as soft and rigid thermoset and thermoplastics materials, to finally various paper, biodegradable and hydrogel materials; this chapter will review their advantages and drawbacks regarding their microfabrication perspectives at both research and industrial scale. The chapter will also address, the evolution of the materials used for fabricating microfluidic chips, and will discuss the application‐oriented pros and cons regarding especially their critical strategies and properties for devices assembly and biocompati‐ bility, as well their potential for downstream biochemical surface modification are presented.

**Keywords:** microfabrication, prototyping, manufacturing, thermoplastics, thermo‐ plastics elastomers, assembly, bonding

### **1. Introduction**

Following Pasteur microbiology and Curie radiotherapy, the next coming and significant medical revolution is geared toward the foreseen convergence and integration of the following innovative science and technology components for the emergence of a new predictive, personalized, and preemptive medicine and practice. Bioengineering, microbiology, micro‐

© 2016 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2016 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

fluidics, optics, and electrochemistry are today the most significant key enabler disciplines. Their complementary features lead to complex systems and democratization and wide public access to personalized and new medical products that will promote both wealth and innovation at the same time, as the IT industry did in the last decades. This technological trend is of tremendous impacts in both sociological and economical applications. For example, the needs of *in situ* personal monitoring are crucial for the comfort and the health surveillance for our specific aging modern society, and it has become a major concern for the health care system. It offers additional opportunities to lower the hospital workloads and expense. It also promotes enhanced therapy efficiency through real‐time and accurate monitoring for proper therapy delivery. The advent of rapid molecular diagnostic systems shortening pathogenic genomic identification down to 1–2 h in an integrated manner, beyond the standard cellular culture identification that may extend from several days up to few weeks is one of the most promising paths for decentralized and personalized medicine. In addition, it is also a powerful asset to address major problems in bacterial antibiotic resistance. Beyond human health, rapid molecular diagnostic approaches deserve multiple segments from water quality control to veterinary and agronomy fields for widely accessible, deployable, and low‐cost analytical systems. The others aspects of the emergence of microfluidic polymer systems are to act as template for dedicated and addressable microenvironments, thus promoting the organogenesis of organs reconstruction and substitution and sustaining the development of novel drugs that are cost effective.

Moreover, the microfluidic science and applications are by essence multidisciplinary. For a targeted application, a seamless and holistic vision is of tremendous importance. In this chapter, we aim to provide an overview of materials used for microfluidic applications. Their selection, evaluation, and integration at both prototyping scale and toward mass production are discussed. Due to the fact that some materials and strategy might be suited for specific goals yet not fit for others, the specific drawbacks and advantages regarding their status and potentials toward a path to products are analyzed.

### **2. Materials for microfluidic applications**

From research development to microfluidic platform applications and product transfer, a holistic and seamless strategy has to consider from the start point to the end. The right selection of the material interface, which might be suited for certain targeted applications, can be detrimental for another purpose. Therefore, navigating material sciences for microfluidic technology is somehow a nebulous nightmare. These difficulties are also enhanced according to the fact that these microfluidic techniques were developed by the physics and engineering landscapes involved in information technology (IT) research and development (R&D), and therefore they exploited silicon‐based clean room technologies. The lack of an adequate material culture related to biomaterials and polymer‐based microstructure induced obvious limitations, which is a significant inhibiting factor for further industrial transfers. Besides the complexity and the interdisciplines involved in microfluidic applications, where bioengineer‐ ing, microbiology, microfluidics, optics, materials science, and electrochemistry are concerned, proper and deep understanding and paradigms of each community are a must for any achievement of a significant system. Still at its early age, microfluidic science and technology have already made successful demonstrations addressing the issues; however, this knowledge is still mostly limited to chemistry, biology, and physics researchers; and unfortunately, it has not sufficiently penetrated yet the biomedical research and clinical environments. The goal of this chapter is thus to fill the gap and promote pertinent highlights on materials performance for microfluidic applications.

Legacies from microelectronics industry and the glassware history in biomedical and chem‐ istry areas, both silicon and glass materials, constitute the initial materials for microfluidic device fabrication. Currently, materials dedicated to microfluidics can be categorized into three broad groups: inorganic, polymers, and paper. Beyond silicon and glass, inorganic materials extend over co‐fired ceramics and vitroceramics. The second polymer‐based category can be divided into two subcategories (i) thermoset materials, which are thermal or UV curable materials from a low viscosity precompound dispensed over a mold and (ii) thermoplastic materials, which are thermoformable materials amendable for rapid prototyping and manu‐ facturing. Both polymer subcategories display rigid to elastomer mechanical properties, and through adaptable formulation and enriched chemical modification, offer a broad range of physicochemical surface properties. Finally, paper microfluidics is an emerging technology based on a patterning approach, where devices drive liquid through capillary actions via wicking in a cellulose matrix.

#### **2.1. Inorganic materials: silicon, glass and ceramics**

fluidics, optics, and electrochemistry are today the most significant key enabler disciplines. Their complementary features lead to complex systems and democratization and wide public access to personalized and new medical products that will promote both wealth and innovation at the same time, as the IT industry did in the last decades. This technological trend is of tremendous impacts in both sociological and economical applications. For example, the needs of *in situ* personal monitoring are crucial for the comfort and the health surveillance for our specific aging modern society, and it has become a major concern for the health care system. It offers additional opportunities to lower the hospital workloads and expense. It also promotes enhanced therapy efficiency through real‐time and accurate monitoring for proper therapy delivery. The advent of rapid molecular diagnostic systems shortening pathogenic genomic identification down to 1–2 h in an integrated manner, beyond the standard cellular culture identification that may extend from several days up to few weeks is one of the most promising paths for decentralized and personalized medicine. In addition, it is also a powerful asset to address major problems in bacterial antibiotic resistance. Beyond human health, rapid molecular diagnostic approaches deserve multiple segments from water quality control to veterinary and agronomy fields for widely accessible, deployable, and low‐cost analytical systems. The others aspects of the emergence of microfluidic polymer systems are to act as template for dedicated and addressable microenvironments, thus promoting the organogenesis of organs reconstruction and substitution and sustaining the development of novel drugs that

336 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

Moreover, the microfluidic science and applications are by essence multidisciplinary. For a targeted application, a seamless and holistic vision is of tremendous importance. In this chapter, we aim to provide an overview of materials used for microfluidic applications. Their selection, evaluation, and integration at both prototyping scale and toward mass production are discussed. Due to the fact that some materials and strategy might be suited for specific goals yet not fit for others, the specific drawbacks and advantages regarding their status and

From research development to microfluidic platform applications and product transfer, a holistic and seamless strategy has to consider from the start point to the end. The right selection of the material interface, which might be suited for certain targeted applications, can be detrimental for another purpose. Therefore, navigating material sciences for microfluidic technology is somehow a nebulous nightmare. These difficulties are also enhanced according to the fact that these microfluidic techniques were developed by the physics and engineering landscapes involved in information technology (IT) research and development (R&D), and therefore they exploited silicon‐based clean room technologies. The lack of an adequate material culture related to biomaterials and polymer‐based microstructure induced obvious limitations, which is a significant inhibiting factor for further industrial transfers. Besides the complexity and the interdisciplines involved in microfluidic applications, where bioengineer‐ ing, microbiology, microfluidics, optics, materials science, and electrochemistry are concerned,

are cost effective.

potentials toward a path to products are analyzed.

**2. Materials for microfluidic applications**

Silicon was the first material used for microfluidics [1]. Indeed, in the mid‐80s, the microelec‐ tromechanical systems (MEMS), from which microfluidic technology is one of the branches that handle fluid, were developed by microelectronic technology. In MEMS, silicon material

**Figure 1.** (a) QuantStudion3D Digital PCR system from Life Sciences and its silicon chip, 60 µm hexagonal wells organized in a honeycomb structure. (b) Array of Si cantilevers for label‐free biodetection of illicit drugs in water. (c) Next generation sequencing HiSeq 2000v3 system developed by Illumina and the silicon chip for microbeads immobili‐ zation.

was the standard material interface. Fabrication of silicon and glass devices involves either subtractive methods (e.g., wet and dry etching) or additive methods (such as metal/dielectric/ insulate deposition). Silicon surface chemistry based on the silanol group (‐Si‐OH) is also well developed and large panels of surface biochemical modification accomplished via silanes chemistry are accessible. Silicon displays a high elastic modulus (~150 GPa) and so, in such a way, active pumping and valving integration, as well as Si brittle characteristics induced overall fragility. Therefore the narrow windows of mechanical properties are limiting factors. Trans‐ parent to IR light, but not in visible range, can cause serious issues and limitations that are obvious for biological fluorescence‐based optical detection methods and for direct fluid imaging. Those issues can be partially solved via several hybrid system approaches, where Si channels can be sealed with transparent materials, such as glass or polymers, leading to a renaissance for Si‐based detectors for microfluidic systems. However, Si microfluidic systems, where reagent storage and other embedded biochemical surface functionalization (e.g., DNA, proteins and cells patternings) are required, feature limitations when considering final assembly and package steps. Indeed, high pressure, temperature, and voltage constraints induced complex strategies for overall device integration concerns. Current approaches such as reactive ion etching (RIE) for plasma exposure in order to activate surface for sealing purposes or high temperature, high pressure parameters for anodic bonding methods are definitively not suitable when considering reagent integration, such as proteins, cells, and nucleic acids species. Even recent efforts developed by Ruchi et al. [2], who reported a low‐ temperature bonding strategy and Si‐Si interface bonding at low‐voltage direct current of about 80 V, the procedure was still performed at a processing temperature of 365°C. For the aforementioned strategy, where a hybrid system is implemented, the definite air‐tight and water‐tight sealing achievements, which are also of priority for biomedical applications considering contamination issues, suffer from the rather similar limitations. Indeed, hybrid approaches in most cases currently consist of realizing a pressure‐free soft contact with the rigid silicon part that needs either oxygen plasma exposure or thermal treatment for sealing and device completion. However, at the research level, due to the high resolution of Si nanofabrication capabilities realized by either electronic beam or nanoimprint lithography among others, and its enriched surface chemistry means for biomolecular grafting, the following highly representative examples and significant achievements for the microfluidic community are highlighted. Applications related to the high‐resolution capabilities of silicon extend over plasmonic, resonators, and microcantilevers systems. Indeed, high‐quality nano‐ optomechanical resonators exhibiting mass sensitivity at the attogram level in the flow through operating environment have already been reported [3]. Si plasmonic microarrays have been interfaced for real‐time and label‐free monitoring of biomolecule interactions of A/G with immunoglobulin G (IgG) antibody [4]. Finally, as a tremendous achievement for the Si‐based microfluidic approach, beyond the aforementioned limitations, we wish to underline the obvious advantages of the stationary Si‐based system for complex digital qPCR platforms for genomic applications. Quantitative determination of pathogenic loads with such Si platforms definitely opens an area for medical research that benefits patients. In comparison to low‐cost and single‐use POC (Point‐of‐care) devices for decentralized purposes, where challenges involved with silicon materials imposed challenging technical needs, Si interface and glass are strategic materials of interests for stationary and highly complex systems. **Figure 1(a)** presents the QuantStudio® 3D Digital PCR system and the related silicon chip developed by Life Sciences Technologies Inc. (St. Petersburg, FL, US). We also refer the readers to the silicon EWOD qPCR platform developed by Gidrol et al. [5] and for further interests to a recent review by V. Marx [6]. Finally, the silicon interface is the material of reference for the vast majority of nanomechanical biosensor [7] systems based on the cantilever approach as illustrated in **Figure 1(b)**. Recently, an array of Si cantilevers for label‐free biodetection of illicit drugs in water has been reported [8]. For next generation sequencing (NGS) purposes, **Figure 1(c)** displays the HiSeq 2000v3 system developed by Illumina Inc. (San Diego, CA, US) and the silicon chip realized for microbeads immobilization.

was the standard material interface. Fabrication of silicon and glass devices involves either subtractive methods (e.g., wet and dry etching) or additive methods (such as metal/dielectric/ insulate deposition). Silicon surface chemistry based on the silanol group (‐Si‐OH) is also well developed and large panels of surface biochemical modification accomplished via silanes chemistry are accessible. Silicon displays a high elastic modulus (~150 GPa) and so, in such a way, active pumping and valving integration, as well as Si brittle characteristics induced overall fragility. Therefore the narrow windows of mechanical properties are limiting factors. Trans‐ parent to IR light, but not in visible range, can cause serious issues and limitations that are obvious for biological fluorescence‐based optical detection methods and for direct fluid imaging. Those issues can be partially solved via several hybrid system approaches, where Si channels can be sealed with transparent materials, such as glass or polymers, leading to a renaissance for Si‐based detectors for microfluidic systems. However, Si microfluidic systems, where reagent storage and other embedded biochemical surface functionalization (e.g., DNA, proteins and cells patternings) are required, feature limitations when considering final assembly and package steps. Indeed, high pressure, temperature, and voltage constraints induced complex strategies for overall device integration concerns. Current approaches such as reactive ion etching (RIE) for plasma exposure in order to activate surface for sealing purposes or high temperature, high pressure parameters for anodic bonding methods are definitively not suitable when considering reagent integration, such as proteins, cells, and nucleic acids species. Even recent efforts developed by Ruchi et al. [2], who reported a low‐ temperature bonding strategy and Si‐Si interface bonding at low‐voltage direct current of about 80 V, the procedure was still performed at a processing temperature of 365°C. For the aforementioned strategy, where a hybrid system is implemented, the definite air‐tight and water‐tight sealing achievements, which are also of priority for biomedical applications considering contamination issues, suffer from the rather similar limitations. Indeed, hybrid approaches in most cases currently consist of realizing a pressure‐free soft contact with the rigid silicon part that needs either oxygen plasma exposure or thermal treatment for sealing and device completion. However, at the research level, due to the high resolution of Si nanofabrication capabilities realized by either electronic beam or nanoimprint lithography among others, and its enriched surface chemistry means for biomolecular grafting, the following highly representative examples and significant achievements for the microfluidic community are highlighted. Applications related to the high‐resolution capabilities of silicon extend over plasmonic, resonators, and microcantilevers systems. Indeed, high‐quality nano‐ optomechanical resonators exhibiting mass sensitivity at the attogram level in the flow through operating environment have already been reported [3]. Si plasmonic microarrays have been interfaced for real‐time and label‐free monitoring of biomolecule interactions of A/G with immunoglobulin G (IgG) antibody [4]. Finally, as a tremendous achievement for the Si‐based microfluidic approach, beyond the aforementioned limitations, we wish to underline the obvious advantages of the stationary Si‐based system for complex digital qPCR platforms for genomic applications. Quantitative determination of pathogenic loads with such Si platforms definitely opens an area for medical research that benefits patients. In comparison to low‐cost and single‐use POC (Point‐of‐care) devices for decentralized purposes, where challenges involved with silicon materials imposed challenging technical needs, Si interface and glass are

338 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

Besides Si, glass, due to its wide applications in chemistry has been adopted as a key material for lab‐on‐a‐chip fabrication. Optically transparent with excellent and low‐fluorescence background, glass also displays highly suitable chemical resistance. Thus it has emerged as a material of choice. Its transfer to microfluidics has been facilitated due to the traditional "off‐ chip" chemistry developed over several decades. Glass microfabrication involves well‐ established processes such as UV photolithography and chemical etching; however, such technologies rank relatively low when considering further manufacturing perspectives, compared to rapid thermal molding technology for polymer‐based microfluidic system. It is for this reason the glass is generally viewed as an application material. Conversely, it may be easier and potentiality less cost effective to fabricate small number of microfluidic devices in glass than making molds for replication purposes using polymer‐based systems. The chemical and thermal properties of glass enable a wide range of functions to be undertaken on the device, including the requiring aggressive and corrosive solvents, chemical agents, and others extreme temperature applications. Finally, assembly, conditioning, and on‐chip reagent storage steps suffer similar limitations to the Si material. In particular, the glass fusion bonding is extremely slow and requires very high temperature. However, for high‐pressure demanding applications associated with its strength, maintaining high level of channel integrity over operation, glass is definitively positioned at the preferred place. Compatible with electrophoretic‐based separation techniques, microfluidic glass chips have been used for numerous demonstrators and coupled to either mass spectroscopy, electrochemical, or chemiluminescence detection means among others. From peptides to vascular biomarkers and DNA identification, we refer the reader to recent comprehensive reviews on advances for glass µCE chips [9, 10]. Glass mostly used the microfabrication approach that involves standard UV photolithography in combination with either dried or wet etching protocols. Frequently used wet etching chemical solutions feature a significant isotropic etchant property, therefore, the achievement of microchannels with straight and high aspect ratio structures are challenging. Alternative etching profiles such as anisotropic or straight‐walled patterns can be done with dried RIE processes. Micromilling of glass is the second‐most used approach. It is a subtractive process, where a rotating cutting tool removes material from a workpiece. Compared to micromilling of polymers, more attention for glass is required. This is primarily due to its high hardness and low tensile strength. Most milling is done using computer numerical control system (CNC micromachining), which represents a great deal of opportunity and flexibility in terms of pattern generation. However during processing, overheating of the diamond‐ coated drilling tool has to be considered, and cooling strategies therefore are implemented. Definitively, CNC micromachining offers a high level of interest for rapid prototyping of glass microchip, we refer the readers to the recently published micromilling tutorial review [11, 12].

Illustrating the high‐performance capabilities of the glass interface for highly demanding applications, **Figure 2(a)** presents the microchip developed by Chan et al*.* [13] for the high‐ temperature synthesis of CdSe nanoparticles in nanoliter‐volume droplets in a perfluorinated fluid carrier. **Figures 2(b)** and **(c)** highlight the recent efforts by Klearia (Marcoussis, France) and Dolomite (Royston, UK) companies for the development of low‐temperature bonding protocols for embedded technologies. Finally, related to the development of NGS technologies, **Figures 2(d)** and **(e)** depict the glass chips made by Pacific Biosciences (Palo Alto, CA, USA) and Roche Diagnostics (Indianapolis, IN, USA) for their single molecule real‐time sequencing technology (SMRT) and their 454 sequencing systems, respectively. Distributed by Weill Cornell Medical Researchers (Ithaca, NY, USA), each SMRT cell is provided at a cost of \$ 400.

**Figure 2.** (a) Glass chip for high‐temperature (300°C) synthesis of CdSe nanoparticles from nanoliter‐volume droplets. (b) Array of Si cantilevers for label‐free biodetection of illicit drugs in water. (c) High demanding (500°C and 300 bar) glass chip for droplet generation form dolomite. (d) Pacific Biosciences single molecule Real‐Time Sequencing glass chip. (e) Glass PicoTiter plate for DNA sequencing systems from 454 DNA sequencing system.

Ceramic microfluidic platforms can be fabricated using low temperature co‐fired ceramic (LTCC) technology for the achievement of hermetically sealed monolithic microfluidic platforms with homogeneous surface chemistry and physical properties through a pertinent cost‐effective manner compared to Si and glass platforms. This is a well‐established technology for low‐cost and high‐volume production of portable wireless electronic applications, but yet with limited involvement into the microfluidic landscape. However, its multilayer fabrication approach allows monolithic integration of complex structures. As a result, three‐dimensional microfluidic channels and cavities have been already reported [14]. The compatibility with ink printing techniques has enabled the development of highly integrated devices that incorporate electrochemical detection as well as all the electronic components for signal and data process‐ ing [15]. However, the integration of optical detection means in such platforms still constitutes a serious issue due to the overall opacity of ceramic material. To mitigate this issue, two strategies have been implemented: (1) the integration of optical fiber and (2) the implementa‐ tion of transparent windows for localized optical analysis [16]. Recently, Microresist Technol‐ ogy Inc. (Berlin, Germany) developed a promising and organically modified ceramic Ormocer as an optically transparent and UV curable ceramic material to fix this issue.

#### **2.2. Polymers: thermoset and thermoplastic materials**

coated drilling tool has to be considered, and cooling strategies therefore are implemented. Definitively, CNC micromachining offers a high level of interest for rapid prototyping of glass microchip, we refer the readers to the recently published micromilling tutorial review [11, 12].

340 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

Illustrating the high‐performance capabilities of the glass interface for highly demanding applications, **Figure 2(a)** presents the microchip developed by Chan et al*.* [13] for the high‐ temperature synthesis of CdSe nanoparticles in nanoliter‐volume droplets in a perfluorinated fluid carrier. **Figures 2(b)** and **(c)** highlight the recent efforts by Klearia (Marcoussis, France) and Dolomite (Royston, UK) companies for the development of low‐temperature bonding protocols for embedded technologies. Finally, related to the development of NGS technologies, **Figures 2(d)** and **(e)** depict the glass chips made by Pacific Biosciences (Palo Alto, CA, USA) and Roche Diagnostics (Indianapolis, IN, USA) for their single molecule real‐time sequencing technology (SMRT) and their 454 sequencing systems, respectively. Distributed by Weill Cornell Medical Researchers (Ithaca, NY, USA), each SMRT cell is provided at a cost of \$ 400.

**Figure 2.** (a) Glass chip for high‐temperature (300°C) synthesis of CdSe nanoparticles from nanoliter‐volume droplets. (b) Array of Si cantilevers for label‐free biodetection of illicit drugs in water. (c) High demanding (500°C and 300 bar) glass chip for droplet generation form dolomite. (d) Pacific Biosciences single molecule Real‐Time Sequencing glass

Ceramic microfluidic platforms can be fabricated using low temperature co‐fired ceramic (LTCC) technology for the achievement of hermetically sealed monolithic microfluidic platforms with homogeneous surface chemistry and physical properties through a pertinent cost‐effective manner compared to Si and glass platforms. This is a well‐established technology for low‐cost and high‐volume production of portable wireless electronic applications, but yet with limited involvement into the microfluidic landscape. However, its multilayer fabrication approach allows monolithic integration of complex structures. As a result, three‐dimensional microfluidic channels and cavities have been already reported [14]. The compatibility with ink printing techniques has enabled the development of highly integrated devices that incorporate electrochemical detection as well as all the electronic components for signal and data process‐ ing [15]. However, the integration of optical detection means in such platforms still constitutes a serious issue due to the overall opacity of ceramic material. To mitigate this issue, two

chip. (e) Glass PicoTiter plate for DNA sequencing systems from 454 DNA sequencing system.

In the last 15 years, polymers have played the leading role from prototyping and manufacturing perspectives. This is because they offer a broad range of physical and surface chemical properties through adaptable formulation and enriched chemical modification. Based on their adhesiveness intrinsic properties, or activated bonding strategy, complex and 3D multilayered systems have been implemented. Considering both the cost of raw materials and manufacturing perspectives, several polymer interfaces, but none of all, are amendable to mass production processes (e.g., hot‐embossing, injection‐molding, and roll‐to‐roll). For the aforementioned thermoset subcategory, the most common materials are thermal and UV curable materials, Respectively, polydimethylsiloxane (PDMS) and the so‐called SU‐8 photoresist [17, 18]. On the thermoplastic side, the most popular materials are polycyclo‐ olefin (PCO), polycarbonate (PC), polytetrafluoroethylene (PTFE) and polystyrene (PS). Polycyclo‐olefin offers high moldability and low water uptake [19, 20]. Polycarbonate [21] has excellent material toughness properties while polytetrafluoroethylene [22] and polyimide [23, 24] feature excellent chemical resistance, electrical, and thermal properties, respectively. Polystyrene has become more and more involved for cellular‐based microfluidic systems, due to its wide applications in cell biology. Indeed, PS has provided decades of conclusions and sensitive protocol establishment, and numerous cell behaviors and functions have been determined [25]. Currently, PDMS and a dozen thermoplastic materials cover the vast majority of microfluidic research activities. The intensive use of PDMS is devoted to rapid prototyping and proof‐of‐concept demonstrations. However, despite its intrinsic drawbacks, mainly related to its cellular toxicity, molecular adsorption and absorption, gas permeability, and bonding issues, PDMS still appears as the preferred material in laboratories. Its industrial transfer is generally not overly recommended due to the cost concern. On the other hand, although thermoplastic interfaces offer the solution for product development, the entry cost in terms of required mold making as well as equipment setting limits its development. In addition, only well‐established groups have significant infrastructures and facilities to afford it. To end this section, recently introduced soft thermoplastic elastomers (sTPE) are discussed. These materials bridge the gap between PDMS and classical rigid thermoplastic materials such as PC, PCO, and PS [48–52]. The sTPE materials are low‐cost and commercially available polymers, offering transparency, biocompatibility, and flexibility compared to PDMS, and they can be rapidly thermoformed like currently used thermoplastics. Additionally, they display enhanced molding properties; indeed, low‐pressure molding can be performed (e.g., <1–2 bar), lowering down the thermomechanical properties of the required molds. Due to their elastomeric properties, they can easily be peeled off from the mold. Assembling and bonding procedures are more convenient, and cohesive sealing can be done without thermal and/or plasma treatment. This material promotes a seamless integration, promoting therefore a realistic transfer path from prototyping to feasible and realistic industrialization. The sTPE material could become a mainstream platform for microfluidic technology and POC applications.

#### *2.2.1. Thermoset*

Thermosets are liquid or solid materials at room temperature. Their molecular polymer chains cross‐link through a process called curing, when the polymer is heated or exposed to light or to others radiations. The curing step involves an irreversible chemical reaction. Therefore once formed, the thermoset parts cannot be reshaped anymore. Typical examples of thermoset polymers used in microfabrication are PDMS and SU‐8 materials.

#### *2.2.2. PDMS*

First introduced in the late 90s, PDMS [26] is the most common microfluidic substrate, a large portion (40–45%) of research papers published on microfluidic devices utilizes this material [27]. Devices' molds are formed via traditional micromachining and photolithography means and a mixture of two liquid polymer components are mixed together and then casted, cured, and finally peeled off from the master. Due to its elasticity, integration of microvalves or micropumps is possible, and complex 3D system and numerous point‐of‐care devices have been developed for research applications. PDMS displays excellent optical properties for convenient fluorescence detection and fluid imaging. Due to its gas permeability, PDMS appears as a material of choice for many but not all cellular studies.

On the other hand, four important properties of PDMS have negative impacts: (1) channel deformation due its high mechanical compliance, (2) evaporation and absorption, (3) leaching‐ out of uncrosslinked oligomers, and (4) hydrophobic recovery [25]. The compliance issue is particularly true considering cell culture experiments that require accurate control over shear force on the endothelial monolayer, the inability to account for mechanical deformation bias in data analysis and subsequent data interpretation [28]. Oxygen permeability in PDMS is three orders of magnitude higher as compared to PS, and may in fact produce a hyperoxic micro‐ environment leading to cellular stress [29, 30]. Water vapor resulting from the permeable PDMS also leads to problematic shifts on volumes, concentrations, and chemical balances [31]. PDMS is also largely prone to bulk absorption of hydrophobic compounds. Regehr et al*.* [32] have shown significant depleted estrogen levels in culture media, leading to inhibition of protein‐1 activator. Similarly, important shifts were identified in response to fluoxetine over transfected human embryonic kidney (HEK) in between PDMS and polystyrene interfaces [33]. Finally, absorption not only affects fundamental cellular biology on chip, but also drug discovery and high‐throughput screening applications. However, the PDMS abilities to provide rapid (1–4 h), easy, low‐cost (50–200 \$/kg) and straightforward accessibility make it a major leading player for prototyping uses at academic level. However, its commercial appli‐ cations are generally avoided. A noticeable exception emerges from this mainstream Eration, indeed around the multilayered PDMS pressure‐actuated crossed‐channel valve architecture initially developed by Quake et al*.* [34]; the Fluidigm company (San Francisco, CA, USA) has developed several impressive applications ranging from qPCR platforms to mass cytometer as depicted in **Figure 3(a)** and **(b)**. Such system can run 48 samples in 770 reaction chambers. Each chip shown in **Figure 3(c)** ranges at a cost from \$ 400 to 800. This price highlights the intrinsic and overall PDMS difficulties and its lengthy processing steps to tackle for low‐costs microfluidic system for single use perspective. We refer the readers to the following reviews for extended discussion related to microfluidic products development and the ambiguous positioning of PDMS material in the community for applications development [25, 35, 36].

**Figure 3.** (a) Bilayer PDMS pneumatic valving microfluidic system, liquid flow inside the vertically oriented top chan‐ nel and the bottom channel (air flow) is pressure‐actuated for clogging/or liquid motion of the fluidic layer. **(b)** and **(c)** Biomark HD system for digital qPCR from Fluidigm, which run 48 samples in 770 reactions/sample chip.

#### *2.2.3. SU‐8: an epoxy‐based material*

material could become a mainstream platform for microfluidic technology and POC

Thermosets are liquid or solid materials at room temperature. Their molecular polymer chains cross‐link through a process called curing, when the polymer is heated or exposed to light or to others radiations. The curing step involves an irreversible chemical reaction. Therefore once formed, the thermoset parts cannot be reshaped anymore. Typical examples of thermoset

First introduced in the late 90s, PDMS [26] is the most common microfluidic substrate, a large portion (40–45%) of research papers published on microfluidic devices utilizes this material [27]. Devices' molds are formed via traditional micromachining and photolithography means and a mixture of two liquid polymer components are mixed together and then casted, cured, and finally peeled off from the master. Due to its elasticity, integration of microvalves or micropumps is possible, and complex 3D system and numerous point‐of‐care devices have been developed for research applications. PDMS displays excellent optical properties for convenient fluorescence detection and fluid imaging. Due to its gas permeability, PDMS

On the other hand, four important properties of PDMS have negative impacts: (1) channel deformation due its high mechanical compliance, (2) evaporation and absorption, (3) leaching‐ out of uncrosslinked oligomers, and (4) hydrophobic recovery [25]. The compliance issue is particularly true considering cell culture experiments that require accurate control over shear force on the endothelial monolayer, the inability to account for mechanical deformation bias in data analysis and subsequent data interpretation [28]. Oxygen permeability in PDMS is three orders of magnitude higher as compared to PS, and may in fact produce a hyperoxic micro‐ environment leading to cellular stress [29, 30]. Water vapor resulting from the permeable PDMS also leads to problematic shifts on volumes, concentrations, and chemical balances [31]. PDMS is also largely prone to bulk absorption of hydrophobic compounds. Regehr et al*.* [32] have shown significant depleted estrogen levels in culture media, leading to inhibition of protein‐1 activator. Similarly, important shifts were identified in response to fluoxetine over transfected human embryonic kidney (HEK) in between PDMS and polystyrene interfaces [33]. Finally, absorption not only affects fundamental cellular biology on chip, but also drug discovery and high‐throughput screening applications. However, the PDMS abilities to provide rapid (1–4 h), easy, low‐cost (50–200 \$/kg) and straightforward accessibility make it a major leading player for prototyping uses at academic level. However, its commercial appli‐ cations are generally avoided. A noticeable exception emerges from this mainstream Eration, indeed around the multilayered PDMS pressure‐actuated crossed‐channel valve architecture initially developed by Quake et al*.* [34]; the Fluidigm company (San Francisco, CA, USA) has developed several impressive applications ranging from qPCR platforms to mass cytometer as depicted in **Figure 3(a)** and **(b)**. Such system can run 48 samples in 770 reaction chambers.

polymers used in microfabrication are PDMS and SU‐8 materials.

342 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

appears as a material of choice for many but not all cellular studies.

applications.

*2.2.1. Thermoset*

*2.2.2. PDMS*

Using classical means of photolithography or stereolithography, SU‐8 microfluidic devices can be built in a complex 3D structure out of an initial liquid resist. SU‐8 contains eight epoxy groups which undergo a very strong crosslinking upon exposure to UV light. Such composi‐ tion and process lead to highly mechanically, thermally, and chemically stable materials. However, large internal stress exists due to the process, providing thus an overall brittle characteristic, and therefore making it difficult in handling and transferring part. SU‐8 can be deposited and patterned in a range of thicknesses from nanometer to millimeter films, using lithography (either UV and e‐beam), the lateral feature resolution extends from the macro‐ scopic surface down to submicrometer. SU‐8 can be structured also from the laser ablation approach. However, as for photolithography means, it appears that both methods have limited throughput [37]. Indeed, the cycle for resin preparation and others processes are lengthy. It seems that SU‐8 is limited mainly to academic uses. For in‐depth descriptions of the SU‐8 interface for microfluidic applications, we refer the reader to a couple of comprehensive and exhaustive reviews [18, 38]. Beyond its use for direct microfluidic device fabrication, SU‐8, due to its high mechanical strength and its capabilities to create high aspect ratio structures [39] and complex 3D networks [40], is definitely a material of choice for mold master making. The SU‐8 mold has been intensively used for hot‐embossing of thermoplastic devices, or as an intermediate system for injection‐molding [41].

#### *2.2.4. Alternative thermoset material*

Besides PDMS soft thermoset material, several attempts have been introduced in order to provide alternatives for soft thermoset materials and mainly the use of fluorinated‐based polymers has been reported. The attractiveness of those solutions arises from the inertness of the perfluorinated compounds. De Simone et al*.* [42] have developed a photocurable soft perfluoropolyethers (PFPE). It exhibits low toxicity and low surface energy and displays enlarged chemical resistance. For multiphasic microfluidic environment, it might be a material of choice due to the fact that such Teflon‐like structure is both oleophobic and hydrophobic. More recently, high aspect ratio (up to 6.5) PFPE microfluidic devices have been fabricated by a direct photolithographic process. Through a mask‐assisted photopolymerization approach, the authors have successfully developed a rapid overall process of around 5 min and demon‐ strated important sealing capabilities, indeed the device can withstand a pressure up to 3.8  bar [43]. Finally, the devices have been tested with some model reactions employing organic solvents.

#### *2.2.5. Thermoplastic polymers*

#### *2.2.5.1. Rigid thermoplastics*

According to the aforementioned drawbacks of PDMS, intensive use of thermoplastic mate‐ rials, such as polycarbonate and polystyrene, is increasing. These materials are amendable for rapid thermoforming manufacturing technology in the CD and biology industries. Those platforms have been clearly identified as materials of choice for microchip research and subsequent technology transfer. Complementary polymethylmethacrylate (PMMA) and polyimide, due their favorable position in the semiconductor industry, complete the set of foreseen thermoplastic candidate. More recently, polycyclo‐olefin polymer has become another popular substrate for microfluidics. It displays high chemical resistance, low water absorption, and excellent optical transparency in the near UV range. The materials are moldable polymers when heated above their glass temperature, offering thus recyclable and reshaping perspectives. They also provide a subsequent bonding pathway. They are optically clear and commercially available, and they display slightly better solvent compatibility than PDMS. However, they are incompatible with most organic solvents, such as ketones and hydrocarbons. Currently used thermoplastic interfaces are rigid and stiff materials with Young modulus in the range of Giga Pascal. Consequently, they are not as convenient as PDMS to achieve conformal contact. Therefore, sealing strategies involve pressurized solvent‐assisted and thermal approaches to melt surfaces. In addition, the following thermoplastics: perfluor‐ oalkoxy (Teflon PFA) and fluorinated ethylenepropylene (Teflon FEP) can be used for ex‐ tremely inert microfluidic devices. Ultimately, they feature nonstick and antifouling properties. PFA has been used for high‐quality immobilization of *Escherichia coli*, *Pseudomonas*

*putida,* and *Bacillus subtilis* cells in highly dense microarray patterns [44, 45]. For more classi‐ cally used interfaces, Zhang et al*.* [46] reported sealing and chemical surface modification of an integrated monolithic PMMA microdevice for DNA purification and amplification of *E. coli*. Also, a highly integrated polystyrene microfluidic chip coupled to electrospray ionization mass spectrometry for on‐chip protein digestion and online analysis was developed [47]. One of the most important challenges faced when targeting for molding of thermoplastic parts is the realization of the master cavity featuring submicron resolution, high aspect ratio, and densely packed structure areas. Most of the materials are crystalline and/or semi‐crystalline, and they display high coefficient of thermal expansion (CTE) and high shrinkage parameters compared to amorphous ones. Additionally, due to the fact that they are rigid and often brittle, they represent sensitive challenges from the manufacturing perspectives. The high shear force resulting from the pressurized environment generates asymmetric and/or random pull‐off of plastic edges due to friction in demolding, which is downstream detrimental for sealing and bonding. The characteristics of the thermoplastic materials induce high specifications on the thermo‐mechanical properties of the master molds. Currently, the impacts on the master mold making related to the realization of such thermoplastic parts are only metallic (stainless, nickel‐ cobalt alloy, and aluminum), and from prototyping perspectives, only few epoxy molds can be employed. Even if elegant CNC laser machining and electroplating processes are available for mold making, the implementation of such master are expensive and are limited in terms of resolution, pattern density, and aspect ratio. The work suggest an overall in‐depth consid‐ eration when starting to envisage a microfluidic system development, undoubtedly a holistic approach should be taken. The overall chip constraints in terms of fabrication, sealing, packaging, thermomechanical loads, biological compatibility (both in terms of physicochem‐ ical surface properties and reagent integration), biomicrofluidic functions, microfluidic statistical, and robust properties have to be considered.

#### *2.2.5.2. Soft thermoplastic elastomers: sTPE*

and complex 3D networks [40], is definitely a material of choice for mold master making. The SU‐8 mold has been intensively used for hot‐embossing of thermoplastic devices, or as an

Besides PDMS soft thermoset material, several attempts have been introduced in order to provide alternatives for soft thermoset materials and mainly the use of fluorinated‐based polymers has been reported. The attractiveness of those solutions arises from the inertness of the perfluorinated compounds. De Simone et al*.* [42] have developed a photocurable soft perfluoropolyethers (PFPE). It exhibits low toxicity and low surface energy and displays enlarged chemical resistance. For multiphasic microfluidic environment, it might be a material of choice due to the fact that such Teflon‐like structure is both oleophobic and hydrophobic. More recently, high aspect ratio (up to 6.5) PFPE microfluidic devices have been fabricated by a direct photolithographic process. Through a mask‐assisted photopolymerization approach, the authors have successfully developed a rapid overall process of around 5 min and demon‐ strated important sealing capabilities, indeed the device can withstand a pressure up to 3.8  bar [43]. Finally, the devices have been tested with some model reactions employing organic

According to the aforementioned drawbacks of PDMS, intensive use of thermoplastic mate‐ rials, such as polycarbonate and polystyrene, is increasing. These materials are amendable for rapid thermoforming manufacturing technology in the CD and biology industries. Those platforms have been clearly identified as materials of choice for microchip research and subsequent technology transfer. Complementary polymethylmethacrylate (PMMA) and polyimide, due their favorable position in the semiconductor industry, complete the set of foreseen thermoplastic candidate. More recently, polycyclo‐olefin polymer has become another popular substrate for microfluidics. It displays high chemical resistance, low water absorption, and excellent optical transparency in the near UV range. The materials are moldable polymers when heated above their glass temperature, offering thus recyclable and reshaping perspectives. They also provide a subsequent bonding pathway. They are optically clear and commercially available, and they display slightly better solvent compatibility than PDMS. However, they are incompatible with most organic solvents, such as ketones and hydrocarbons. Currently used thermoplastic interfaces are rigid and stiff materials with Young modulus in the range of Giga Pascal. Consequently, they are not as convenient as PDMS to achieve conformal contact. Therefore, sealing strategies involve pressurized solvent‐assisted and thermal approaches to melt surfaces. In addition, the following thermoplastics: perfluor‐ oalkoxy (Teflon PFA) and fluorinated ethylenepropylene (Teflon FEP) can be used for ex‐ tremely inert microfluidic devices. Ultimately, they feature nonstick and antifouling properties. PFA has been used for high‐quality immobilization of *Escherichia coli*, *Pseudomonas*

intermediate system for injection‐molding [41].

344 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

*2.2.4. Alternative thermoset material*

solvents.

*2.2.5. Thermoplastic polymers*

*2.2.5.1. Rigid thermoplastics*

The sTPE is a class of material in which elastomeric properties of elastomer rubber (e.g., PDMS) are embodied with the ease of processing of thermoplastic materials such as PMMA, PCO, and PC. The sTPE thus bridges the gap in between thermoplastics and elastomers, enhancing the advantages of each material. Moreover, the range of sTPE Young modulus extends continuously from 0.1 MPa to 1–5 GPa. The combination of elastomeric and thermoplastic properties makes these materials potential substitutes to PDMS and/or hard TP polymers that are commonly employed in microfluidics [48]. Unlike PDMS, sTPE can be used in the form of extruded sheets that provide off‐the‐shelf availability without the need of performing any precompounding step. Extruded films can be stored over a long period of time (e.g., several years) without any notable degradation, making it possible to use the material on demand at any time (**Figure 4(a)**) [49, 50]. The sTPE materials are available at low cost, and they display optical transparency and biocompatibility for proteins, nucleic acids, and cell and tissue engineering and diagnostics [49–53]. sTPEs are block copolymers comprising different monomer sequences that are distributed randomly or statistically in domains through diblock or triblock architectures [51]. For styrenics‐based sTPE materials with low PS content (10–12%), thermodynamic incompatibility between blocks induces nanophase separation and self‐assembly of PS domains into nanometric clusters (typically 10– 30 nm in diameter) that are distributed in a three‐dimensional fashion within a hexagonal symmetry in the rubber matrix of ethylene‐butylene (EB). This morphology provides the basis of the material performance: rigid PS domains act as junction points that stabilize the polymer matrix while the EB‐dominant phase offers elastomeric properties. Moreover, size and cluster distributions promote the sTPE surface to be uniform and homogenous at the microfluidic device level [49–53]. As block copolymer materials, sTPE exhibits two glass transition temperatures corresponding to the EB soft block (Tg,EB ~ −60 to −75°C) and to the styrene rigid block (Tg,EB ~ −90 to 105°C), respectively. The negative value of Tg,EB predicts liquid‐like behavior of the materials.

**Figure 4.** (a) Photograph of an extruded flexestene foil on a roll from which pieces can be cut conveniently before use. (b) Series of SEM images illustrating the fabrication of the microfluidic flexestene device. (i) SU‐8 embossing mold used for the fabrication of the bottom flexestene membrane, (ii) Upper and (iii) lower side of the bottom sTPE membrane, (iv) SU‐8 mold used for the fabrication of the top sTPE membrane, (v) overview, and (vi) close‐up view of the top flexestene membrane. Scale bars in the insets of (i), (ii), (iii), and (iv) correspond to 50, 20, 10, and 200 mm, respectively. The images shown in (iv) and (v) were assembled from several SEM micrographs to achieve the desired field of view. (c) Photographs of an assembled 3D microfluidic immobilization after filling with solutions of a red and green dye (left), optical microscope image of the resulting red‐green pattern obtained on the central region of the microfluidic device (right). (d) high‐throughput fabrication method of sTPE multilayered microfluidic devices, (i) Optical micro‐ graph of the whole micromixer made of two layers of sTPE material bonded on a poly(cyclo)olefin polymer substrate. The device size is smaller than a centimeter square, and each valve measures 200 µm×200 µm, (ii) detailed view of a valve at both open and closed positions; and (iii) curve representing the valving cycle at 1.2 Hz, fluorescence intensity under the deflected membrane is registered, it is maximal when the valve is open and minimal when closed.

Indeed, like PDMS, which similarly displays a negative glass transition temperature, the selected sTPE forms a spontaneous and conformable close contact with flat substrates, generating tight air and water sealing [49–53]. Additionally, the soft blocks provide bonding capabilities above their glass transition, which implies that even at room temperature, the polymer chains can be reorganized according to the contact surface. The bonding strength is variable and dependent on time and temperature; and irreversible bonding was obtained at room temperature [50, 52]. Depicting the enhanced molding capabilities of sTPE interface, Brassard et al*.* [51] have demonstrated the rapid and reliable patterning of open through‐hole microstructures in sTPE material using a method based on hot‐embossing (**Figure 4(b)**). The sTPE‐based 3D microfluidic patterning device was then used for the immobilization of up to 96 different biological probes in a 10×10 array format of 50 µm×50 µm spots (**Figure 4(c)**).

Additionally, for novel tissue engineering biomaterial platform, high molding performances have been confirmed in a newly reported process. We reported a rapid microfabrication of a biocompatible sTPE sheet in an overall 3 min process operating within an ultralow applied pressure (1.6 bar) [52]. Smooth muscle cells contact guidance studies have been conducted over an array of 4‐µm‐patterned grooves [52]. For reader's information, contributing to the estab‐ lishment of the biocompatibility, the bonding and the microfabrication performance of sTPE, which are highly dependent on block copolymers formulation (molecular weight of each block of each diblock (DB) and triblock (TB) and the DB/TB ratio) and also to the additives compo‐ sition (tackifiers and processing agents); we underline the Flexstene sTPE materials perform‐ ance (InfineFlex Inc., San Diego, US and Blackholelab Inc., Paris) for the fabrication of adjacent micropillar arrays with different heights for cellular studies [53]. We also demonstrated that sTPE can be used as a rapid technique for the fabrication and assembly of pneumatically driven valves in a multilayer microfluidic device using a simple SU‐8 mold material for embossing purposes (**Figure 4(d)**) [54]. The quality of the obtained soft thermoplastic valve shows a robust behavior with an opening−closing frequency of 5 Hz. Finally, more recently [49], we demon‐ strated the implementation of a sTPE CD‐like microfluidic system for genomic assay. This device integrates all required molecular assay steps, from cellular lysis to gDNA polymerase chain reaction amplification, amplicons digestion, and microarray hybridization on a plastic support. The low‐temperature, pressure‐free assembly and bonding of sTPE material on the flat polyclo‐olefin thin substrate offer a pertinent solution for simple and efficient loading and storage of the required on‐CD board elements. This was demonstrated through the integration and the conditioning of microbeads, magnetic discs, and dried enzyme species. This work highlights a seamless strategy that promotes a feasible path to transfer from prototyping toward realistic industrialization. This work aims to establish the full and pertinent potential for sTPE centrifugal system as a mainstream microfluidic diagnostic platform for clinical molecular diagnostics, water and food safety, besides other applications.

#### **2.3. Paper, biodegradable and hydrogel materials**

#### *2.3.1. Hydrogels*

nanophase separation and self‐assembly of PS domains into nanometric clusters (typically 10– 30 nm in diameter) that are distributed in a three‐dimensional fashion within a hexagonal symmetry in the rubber matrix of ethylene‐butylene (EB). This morphology provides the basis of the material performance: rigid PS domains act as junction points that stabilize the polymer matrix while the EB‐dominant phase offers elastomeric properties. Moreover, size and cluster distributions promote the sTPE surface to be uniform and homogenous at the microfluidic device level [49–53]. As block copolymer materials, sTPE exhibits two glass transition temperatures corresponding to the EB soft block (Tg,EB ~ −60 to −75°C) and to the styrene rigid block (Tg,EB ~ −90 to 105°C), respectively. The negative value of Tg,EB predicts liquid‐like

346 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

**Figure 4.** (a) Photograph of an extruded flexestene foil on a roll from which pieces can be cut conveniently before use. (b) Series of SEM images illustrating the fabrication of the microfluidic flexestene device. (i) SU‐8 embossing mold used for the fabrication of the bottom flexestene membrane, (ii) Upper and (iii) lower side of the bottom sTPE membrane, (iv) SU‐8 mold used for the fabrication of the top sTPE membrane, (v) overview, and (vi) close‐up view of the top flexestene membrane. Scale bars in the insets of (i), (ii), (iii), and (iv) correspond to 50, 20, 10, and 200 mm, respectively. The images shown in (iv) and (v) were assembled from several SEM micrographs to achieve the desired field of view. (c) Photographs of an assembled 3D microfluidic immobilization after filling with solutions of a red and green dye (left), optical microscope image of the resulting red‐green pattern obtained on the central region of the microfluidic device (right). (d) high‐throughput fabrication method of sTPE multilayered microfluidic devices, (i) Optical micro‐ graph of the whole micromixer made of two layers of sTPE material bonded on a poly(cyclo)olefin polymer substrate. The device size is smaller than a centimeter square, and each valve measures 200 µm×200 µm, (ii) detailed view of a valve at both open and closed positions; and (iii) curve representing the valving cycle at 1.2 Hz, fluorescence intensity

under the deflected membrane is registered, it is maximal when the valve is open and minimal when closed.

Indeed, like PDMS, which similarly displays a negative glass transition temperature, the selected sTPE forms a spontaneous and conformable close contact with flat substrates, generating tight air and water sealing [49–53]. Additionally, the soft blocks provide bonding capabilities above their glass transition, which implies that even at room temperature, the polymer chains can be reorganized according to the contact surface. The bonding strength is variable and dependent on time and temperature; and irreversible bonding was obtained at

behavior of the materials.

Hydrogels display a molecular architecture analog to extracellular matrix, with water uptake properties up to 80% of its total mass. Hydrogels are highly porous and thus an excellent matrix for cellular biology studies. However, direct tissue engineering from bulky hydrogels are challenging due to the restricted depth for nutriments diffusion of around hundreds micro‐ meters [55]. Microfluidic technology, through both top‐down and bottom‐up approaches, has demonstrated its abilities to tackle this fundamental issue. From a top‐down approach, microchannels are fabricated inside the hydrogel while for the bottom‐up, hydrogels filled microchannels cavities. Matrigel™ and collagen are the mostly used animal‐derived hydro‐ gels. They promote excellent cellular adhesion and proliferation [55]. Recently, Bang et al*.* [57] engineered a 3D neural circuit in a microfluidic Matrigel hydrogel system. They had grown, aligned, and organized 3D networks of axon bundles at an average speed of 250 µm. d‐1 for a period of 6 *in vitro* days. Alternatively, alginate and agarose plant‐derived hydrogels and synthetic ones, such as polyacrylamide or polyethylene glycol (PEG), can be used [58]. Even though synthetic hydrogels slightly lack cellular adhesion compared to animal‐derived ones, they nevertheless promote higher flexibility and enriched formulation adaptability for fine tuning objectives. Hydrogel composition, structure, morphology, and rigidity have been used in a high‐throughput manner in droplet‐based microfluidics [59]. Recently, agarose hydrogels have been integrated in a microfluidic system for *E. coli* purification and concentration, and finally for fluorescence immune detection. Authors reported that 90% recovery efficiency could be achieved with a million‐fold volume reduction from 400 µL to 400 pL. For concentration of 1 × 103  cells mL–1 bacteria, approximately ten million‐fold enrichment in cell density was realized. Urine and blood clinical isolates were further tested and validated [60]. We refer the readers to follow the review for further reading [61].

#### *2.3.2. Biodegradable materials*

Biodegradable polymers for tissue engineering and drug delivery purposes display degrada‐ tion time ranging from 24 h to several months. They offer, in a microfluidic format, a promising opportunity for microstructured tissue scaffolds. Commonly used biodegradable matrices are polycaprolactone, poly(lactide‐co‐glycolide), and polyglycolic acid (PGA). Their degradation and mechanical properties are tunable, and they display minimal changes in the systemic immune responses. Their degradation products such as glycolic acid for PGA are through metabolite response absorbed by the hosted living body. Curing of these materials takes place as their dimer version polymerizes via a ring‐opening reaction under appropriated heating and catalytic steps. An excellent review related to biodegradable material properties and their microenvironment integration has been recently published [62]. Various technologies, such as printing, soft‐lithography, stereolithography, hot‐embossing, and injection‐molding methods, have been used toward integration [63, 64].

#### *2.3.3. Paper*

Paper is the most recently introduced microfluidic material. Its cellulose matrix acts to wick liquids while specifically, hydrophobically, patterning areas to avoid liquid motion. The patterned barriers define the shape (i.e., width and length) while the thickness of the paper accounts for the height. Hydrophilic wicking regions thus serve as channel networks opened to air. Paper as chip material is one of the cheapest materials, and it can be easily stacked in 3D devices [65]. The fabrication approaches can be divided into two groups. Lithographic‐ based methods, where particular coated polymer areas are removed, thus are forming the channels. Second, the printing and the cutting approaches allow direct hydrophobic barrier definition without exposure of the effective channels to any reagents. Fundamentally, paper‐ based microfluidic systems are not suitable for large‐volume samples and their applicable detection perspectives are limited owing to the intrinsic cellulose matrix properties. Recently published by Mace et al*.* [66], paper‐based diagnostic devices and their manufacturing perspectives are commented in detail. Wax printing approaches and assembly of 3D vertical flow assays are further discussed therein. Concerning printing and cutting approaches, the reported channel resolution is quite low and limited to 200 µm. Paper‐based technology, displays undoubtedly tremendous limitations considering heterogonous component integra‐ tion such as valves and other reagent storage issues. Recently, Thom et al*.* [67] have performed the integration of light‐emitting diodes (LEDs) onto paper microfluidic device for the fluores‐ cence detection of β ‐d‐galactosidase. From medium to low complexity bioassay integrations, paper‐based technology appears to be a promising pathway for portable and low‐cost platform in the future, and thus a material of choice for optimized and multiplexed lateral flow assays in the health care segment. We refer the readers to follow recently published review for further reading on paper‐based microfluidic system for bioanalytical applications [68, 69].

## **3. Conclusion**

microchannels cavities. Matrigel™ and collagen are the mostly used animal‐derived hydro‐ gels. They promote excellent cellular adhesion and proliferation [55]. Recently, Bang et al*.* [57] engineered a 3D neural circuit in a microfluidic Matrigel hydrogel system. They had grown, aligned, and organized 3D networks of axon bundles at an average speed of 250 µm. d‐1 for a period of 6 *in vitro* days. Alternatively, alginate and agarose plant‐derived hydrogels and synthetic ones, such as polyacrylamide or polyethylene glycol (PEG), can be used [58]. Even though synthetic hydrogels slightly lack cellular adhesion compared to animal‐derived ones, they nevertheless promote higher flexibility and enriched formulation adaptability for fine tuning objectives. Hydrogel composition, structure, morphology, and rigidity have been used in a high‐throughput manner in droplet‐based microfluidics [59]. Recently, agarose hydrogels have been integrated in a microfluidic system for *E. coli* purification and concentration, and finally for fluorescence immune detection. Authors reported that 90% recovery efficiency could be achieved with a million‐fold volume reduction from 400 µL to 400 pL. For concentration of

 cells mL–1 bacteria, approximately ten million‐fold enrichment in cell density was

realized. Urine and blood clinical isolates were further tested and validated [60]. We refer the

Biodegradable polymers for tissue engineering and drug delivery purposes display degrada‐ tion time ranging from 24 h to several months. They offer, in a microfluidic format, a promising opportunity for microstructured tissue scaffolds. Commonly used biodegradable matrices are polycaprolactone, poly(lactide‐co‐glycolide), and polyglycolic acid (PGA). Their degradation and mechanical properties are tunable, and they display minimal changes in the systemic immune responses. Their degradation products such as glycolic acid for PGA are through metabolite response absorbed by the hosted living body. Curing of these materials takes place as their dimer version polymerizes via a ring‐opening reaction under appropriated heating and catalytic steps. An excellent review related to biodegradable material properties and their microenvironment integration has been recently published [62]. Various technologies, such as printing, soft‐lithography, stereolithography, hot‐embossing, and injection‐molding methods,

Paper is the most recently introduced microfluidic material. Its cellulose matrix acts to wick liquids while specifically, hydrophobically, patterning areas to avoid liquid motion. The patterned barriers define the shape (i.e., width and length) while the thickness of the paper accounts for the height. Hydrophilic wicking regions thus serve as channel networks opened to air. Paper as chip material is one of the cheapest materials, and it can be easily stacked in 3D devices [65]. The fabrication approaches can be divided into two groups. Lithographic‐ based methods, where particular coated polymer areas are removed, thus are forming the channels. Second, the printing and the cutting approaches allow direct hydrophobic barrier definition without exposure of the effective channels to any reagents. Fundamentally, paper‐ based microfluidic systems are not suitable for large‐volume samples and their applicable

readers to follow the review for further reading [61].

348 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

have been used toward integration [63, 64].

1 × 103

*2.3.3. Paper*

*2.3.2. Biodegradable materials*

This chapter presents an overview of materials for microfluidic applications and their appli‐ cations in recent research. The large range of materials dedicated to microfluidics is a key component for successful microfluidic applications. The optimal selection of an adequate material platform for a targeted application is of tremendous importance and represents significant technical challenges. In a concomitant manner, this decision has to be taken accordingly an exhaustive list of requirements essentially related to the biocompatibility, the overall thermomechanical properties, the latter inherent to bonding and reagent integration. Beyond the traditional proof‐of‐concept works developed at the academic research level, another higher level of concern exists when real applications and medical research are envisioned. Two important issues therefore need to be addressed. First, a reevaluation of the biocompatibility and the overall stability of the intrinsic microfluidic performance when handling real samples. The second aspect involves the scaling‐up of each microfabrication, bonding, conditioning, and other packaging needs, and their interdependences and costs. The gap currently is wide in these aspects, and it is one of the most severe limitations for micro‐ fluidic applications. Therefore dedicated efforts are needed to tackle this issue. The introduc‐ tion of sTPE highlights a seamless strategy that promotes a feasible path transfer from prototyping toward realistic industrialization, working from the earliest research steps to the end with a unique polymer interface. Beyond, the polymer materials presented in this chapter, there is tremendous space considering the introduction of other functional polymers in the microfluidic applications. We envision that new research activities focused on conductive, piezo, and magnetically doped polymers among other polymers not only provide a fantastic opportunity for further progress and advancement but also for a new field of research and IP development. The considerations extend surely over new blends of material development for specific goals and needs at large. The high level of multidisciplinary skills required in the field is challenging for the academic community; however, such multidisciplinary nature that extends from biology/medicine, microfabrication, microfluidic materials, and electronic and optical through sensors also provide unique solutions. Definitively, the development of innovative materials will bring innovations, and our community has to act proactively in this direction for the success of microfluidic research and the real benefit for progress in health‐ related biomicrofluidic applications.

### **Author details**

Emmanuel Roy\* , Antoine Pallandre, Bacem Zribi, Marie‐Charlotte Horny, François Damien Delapierre, Andrea Cattoni, Jean Gamby and Anne‐Marie Haghiri‐Gosnet

\*Address all correspondence to: royemmanuel@live.com

Centre de Nanosciences et de Nanotechnologies, CNRS, Université Paris‐Saclay, C2N Marcoussis, France

#### **References**


Quirino, Recent advances in enhancing the sensitivity of electrophoresis and electro‐ chromatography in capillaries and microchips (2012–2014), *Electrophoresis*, 34(1), 29–54, 2013.

[10] R. A. Saylor and S. M. Lunte, A review of microdialysis coupled to microchip electro‐ phoresis for monitoring biological events, *J. Chromatogr. A.*, 1382, 48–64, 2015.

optical through sensors also provide unique solutions. Definitively, the development of innovative materials will bring innovations, and our community has to act proactively in this direction for the success of microfluidic research and the real benefit for progress in health‐

, Antoine Pallandre, Bacem Zribi, Marie‐Charlotte Horny, François Damien Delapierre, Andrea Cattoni, Jean Gamby and Anne‐Marie Haghiri‐Gosnet

Centre de Nanosciences et de Nanotechnologies, CNRS, Université Paris‐Saclay, C2N

on a silicon wafer, *IEEE Trans. Electron. Dev.*, 26(12), 1880–1886, 1979.

binding events over a wide field‐of‐view, *Sci. Rep.*, 4(6789), 2014.

subsequent multiplex qPCR, *Lab Chip*, 14, 3739–3749, 2014. [6] V. Marx, PCR: paths to sensitivity, *Nat. Methods*, 11, 241–245, 2014.

nanomechanical systems, *Chem. Soc. Rev.*, 42, 1287–1311, 2013.

[1] S. C Terry, J. H. Jerman and J. B. Angell, A gas chromatographic air analyzer fabricated

[2] R. Tiwari and S. Chandra, Low‐temperature silicon‐to‐silicon anodic bonding using sodium‐rich glass for MEMS applications, *J. Electr. Mater.*, 43(2), 555–566, 2014. [3] K. Y. Fong, M. Poot and H. X. Tang, Nano‐optomechanical resonators in microfluidics,

[4] A. F. Coskun, A. E. Cetin, B. C. Galaterra, D. A. Alvarez, H. Altung and A. Ozcan, Lens free optofluidic plasmonic sensor for real‐time and label‐free monitoring of molecular

[5] A. Rival, D. Jary, C. Delattre, Y. Fouillet, G. Castellan, A. Bellemin‐Comte and X. Gidrol, An EWOD‐based microfluidic chip for single‐cell isolation, mRNA purification and

[7] J. Tamayo, P. M. Kosaka, J. J. Ruz, Á. San Paulo and M. Calleja, Biosensors based on

[8] E. Biavardi, S. Federici, C. Tudisco, D. Menozzi, C. Massera, A. Sottini, G. G. Condorelli, P. Bergese and E. Dalcanale, Cavitand‐grafted silicon microcantilevers as a universal probe for illicit and designer drugs in water, *Angew. Chem. Int. Ed. Engl.*, 53(35), 9183–

[9] M. C. Breadmore, R. M. Tubaon, A. I. Shallan, S. C. Phung, A. S. Abdul Keyon, D. Gstoettenmayr, P. Prapatpong, A. Alhusban, L. Ranjbar, H. H. See, M. Dawod and J. P.

related biomicrofluidic applications.

\*Address all correspondence to: royemmanuel@live.com

350 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

*Nano. Lett.*, 15(9), 6116–6120, 2015.

**Author details**

Emmanuel Roy\*

Marcoussis, France

9188, 2014.

**References**


[38] A. Bertsch and P. Renaud, Special issue: 15 years of SU‐8 as MEMS material, *Microma‐ chines*, 6(6), 790–792; 2015.

[23] H. Mogi, Y. Fukushi, S. Koide, R. Sano, T. Sasaki and Y. Nishioka, Ascorbic acid fuel cell with a microchannel fabricated on flexible polyimide substrate, *IEEJ Trans. Sens.*

[24] A. Zulfiqar, A. Pfreundt, W. E. Svendsen and M. Dimak, Fabrication of polyimide based microfluidic channels for biosensor devices, *J. Micromech. Microeng.,* 25, 035022, 2015.

[25] E. Berthier, E. W. K. Young and D. Beebe, Engineers are from PDMS‐lad, biologists are

[26] E. Kim, Y. Xia and G. M. Whitesides, Polymer microstructures formed by moulding in

[28] J. W. Song and L. L. Munn, Fluid forces control endothelial sprouting, *Proc. Natl. Acad.*

[29] J. S. Gewandter, R. J. Staversky and M. A. O'Reilly, Hyperoxia augments ER‐stress‐ induced cell death independent of BiP loss, *Free Radic.. Biol. Med.*, 47, 1742–1752, 2009.

[30] Y. Tang, E. A. Scheef, Z. Gurel, C. M. Sorensen, C. R. Jefcoate and N. Sheibani*,* CYP1B1 and endothelial nitric oxide synthase combine to sustain proangiogenic functions of endothelial cells under hyperoxic stress, *Am. J. Physiol.: Cell Physiol.*, 298, C665–C678,

[31] E. Berthier, J. Warrick, H. Yu and D. Beebe, Managing evaporation for more robust microscale assays, Part 1, Volume loss in high throughtput assays, *Lab Chip*, 8, 852–859,

[32] K. J. Regehr, M. Domenech, J. T. Koepsel, K. C. Carver, S. J. Ellison‐Zelski, W. L. Murphy, L. A. Schuler, E. T. Alarid and D. Beebe, Biological implications of polydimethylsilox‐

[33] X. Su, E. W. K. Young, H. A. S. UnderKofler, T. J. Kamp, C. T. January and D. J. Beebe, Microfluidic cell culture and its application in high‐throughput drug screening:

[34] M. A. Unger, H. P. Chou, T. Thorsen, A. Sherer and S. R. Quake, Monolithic microfab‐ ricated valves and pumps by multilayer soft lithography*, Science*, 288, 113–116, 2000.

[35] M. I. Mohammed, S. Haswell and I. Gibson, Lab‐on‐a‐chip or chip‐in‐a‐lab: challenges

[36] C. D. Chin, V. Linder and K. Sia, Commercialization of microfluidic point‐of‐care

[37] G. Chantal and M. Khan, Laser processing for bio‐microfluidics applications (Part II),

of commercialization lost in translation, *Procedia Technol.*, 20, 54, 2015.

diagnostic devices, *Lab Chip*, 12, 2118–2134, 2012.

*Analyt. Bioanalyt. Chem.*, 385(8), 1362–1369, 2006.

cardiotoxicity assay for hERG channels, *J. Biomed. Screen.*, 16, 101–111, 2011.

ane‐based microfluidic cell culture, *Lab Chip*, 9, 2131–2139, 2009.

*Micromach.*, 134, 366–371, 2014.

capillaries, *Nature*, 376, 581, 1995.

*Sci.*, 108, 15342–15347, 2011.

2010.

2008.

from polystyrenia, *Lab Chip*, 12, 1224–1237, 2012.

352 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

[27] H. Becker, Famous last words, *Lab Chip*, 11(13), 2133–2134, 2011.


[66] C. R. Mace and R. N. Deraney, Manufacturing prototypes for paper‐based diagnostics devices, *Microfluid. Nanofluid.,* 16, 801–809, 2014.

[52] M. Guillemette, E. Roy, F. Auger and T. Veres, Rapid isothermal substrate microfabri‐ cation of a biocompatible thermoplastic elastomer for cellular contact guidance, *Acta.*

[53] J. Wein, S. Jian, B. Wang, Y. Tan, X. Tu, E. Roy, B. Ladoux and Y. Chen, Fabrication of adjacent micropillars arrays with different heights for cell studies, *Microelectr. Eng.*, 158,

[54] E. Roy, J. C. Galas and T. Veres, Thermoplastic elastomers for microfluidics: towards a high‐throughput fabrication method of multilayered microfluidic devices, *Lab Chip*, 11,

[55] N. W. Choi, M. Cabodi, B. Held, J. P. Gelghorn, L. J. Bonassar and A. D. Stroock, Microfluidic scaffolds for tissue engineering, *Nat. Mater.*, 6, 908–915, 2007.

[56] M. W. Tibbitt and K. S. Anseth, Hydrogels as extracellular matrix mimics for 3D cell

[57] S. Bang, S. Na, J. Kim and N. L. Jeon, Engineering‐aligned 3D neural circuit in micro‐

[58] G. Y. Huang, L. H. Zhou, Q. C. Zhang, Y. M. Chen, W. Sun, F. Xu and T. J. Lu, Microfluidic

[59] M. Chau, H. Thérien‐Aubin, Y. Li, Y. Wang, D. Velasco, E. Tumarkin, A. Ramachandran and E. Kumacheva, Microfluidic generation of composite biopolymer microgels with tunable compositions and mechanical properties, *Biomacromolecules*, 15, 2419–2425,

[60] Y. Li, X. Yan, X. Feng, J. Wang, W. Du, Y. Wang, P. Chen, L. Xiong and B. F. Liu, Agarose‐ based microfluidic device for point‐of‐care concentration and detection of pathogen,

[61] V. van Duinen, S. J. Trietsch, J. Joore, P. Vulto and T. Hankemeier, Microfluidic 3D cell culture: from tools to tissue models, *Curr. Opin. Biotech.*, 35, 118–126, 2015.

[62] D. I. Dan Cho and H. Jung Yoo, Microfabrication methods for biodegradable polymeric carriers for drug delivery system applications: a review microelectromechanical

[63] P. A. Gunatillake and R. Adhikari, Biodegradable synthetic polymers for tissue

[64] Y. Lu and S. C. Chen, Micro and nano‐fabrication of biodegradable polymers for drug

[65] H. Liu and R. M. Crooks, Three‐dimensional paper microfluidic devices assembled using the principles of origami, *J. Am. Chem. Soc.,* 133(44), 17564–17566, 2011.

hydrogels for tissue engineering, *Biofabrication*, 3(1), 012001, 2011.

*Biomater.*, 7, 2492–2498, 2011.

culture, *Biotechnol. Bioeng.*, 103, 655–663, 2009.

*Anal. Chem.*, 86(21), 10653–10659, 2014.

systems, *J. Microelectromech. Syst.*, 24(1), 10–18, 2015.

delivery, *Adv. Drug Deliv. Rev.* 56, 1621–1633, 2004.

engineering, *Eur. Cell. Mater.*, 5, 1–16, 2003.

fluidic device, *Adv. Healthc. Mater.*, 5, 159–166, 2016.

354 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

22–25, 2016

2014.

3193–3196, 2011.


**Microfluidics and Material Synthesis**

## **Synthesis of Functional Materials by Non-Newtonian Microfluidic Multiphase System**

Yong Ren, Kai Seng Koh and Yaping Zhang

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/64521

#### **Abstract**

With increasing level of polymer solution involvement in multiphase microdevice for for‐ mation of emulsion and fabrication of functional materials, it is of paramount importance to systematically understand the relevant physics of droplet formation in non-Newtonian fluids and how the material formation process may be affected due to the complex rheo‐ logical effect. The chapter aims to review and discuss the recent advances in technologies that enable fabrication and application of functional materials formed from non-Newtoni‐ an microfluidic multiphase system. Rheological behavior of polymer solutions and the mathematical models are reviewed. The influence of microstructure on rheological be‐ havior of polymer solutions and the fundamental physical phenomena driving non-New‐ tonian microfluidic multiphase system are discussed. Shear thinning and viscoelastic effect on breakup dynamics and droplet formation are presented. The microfabrication process of the device and synthesis of emulsion-templated materials with potential indus‐ trial and biochemical applications are elucidated.

**Keywords:** non-Newtonian fluid, microfluidic, multiphase system, functional material

## **1. Introduction**

Emulsion is mixture of two immiscible liquids, where one liquid is dispersed as droplets in another liquid that forms a continuous phase [1]. The availability of a wide range of technol‐ ogies for emulsion generation and manipulation by microfluidic multiphase system has enabled the applications of microfluidics in a plethora of fields, such as fabrication of coreshell microspheres and capsules using emulsion droplets as a template for drug encapsulation and release and generation of jets as precursors of microfibers for application in wound dressing and tissue engineering. Emulsion can be formed with aqueous/oil multiphase system or all aqueous multiphase system, both of which involves the use of Newtonian fluids or non-

© 2016 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Newtonian fluids. In Newtonian fluids such as simple organic liquids, solutions of lowmolecular-weight inorganic salts, molten metals, or salts, the shear stress at steady condition in laminar flow is linearly proportional to the shear rate, i.e., the tensors that describe the viscous stress and the strain rate are related by a constant viscosity tensor which is independent of the stress state and velocity of the flow. In contrast, non-Newtonian fluids possess such properties that flow curve of shear stress versus shear rate is nonlinear or does not pass through the origin and the apparent viscosity as defined by shear stress divided by shear rate is not constant at a given temperature and pressure, as the apparent viscosity is also influenced by shear rate, the kinematic history of the fluid element, flow conditions, or microchannel configurations. The ever-expanding applications of microfluidic multiphase technologies increasingly require the use of non-Newtonian fluids, examples of which in daily life include cement paste, chocolate, coal slurries, greases and lubricating oils, molten polymer solutions, and cosmetics and personal care products such as shampoos, shaving foams, and toothpaste. Non-Newtonian multiphase microsystem has become a subject of intense research and is widely applied in biomedical engineering, food production, and energy applications. It is hence important to understand the associated physical phenomena, in particular the way how non-Newtonian rheological effect will alter breakup dynamics and, in turn, affect droplet formation, which can be characterized by droplet shape and size, as they are closely related with droplet stability as well as optical and mechanical properties [2]. For example, the shape and size of emulsion droplets have significant impacts on the drug release kinetics when they are used as template to synthesize microcapsules or microparticles for drug delivery [3, 4]. Monodisperse droplets with precisely controlled sizes can be used to deliver an accurate dosing of contained payload such as drug, flavoring, or chemical reactants [5]. Monodispersity and size tenability are highly desired to keep droplets exhibiting constant, controlled, and predictable properties. Nevertheless, the versatility in droplet size control is challenged by complex rheological properties of non-Newtonian fluids, such as formation of undesirable satellite droplets due to bead-on-string patterns, as a result of the stretching and thinning of non-Newtonian liquid filaments. Apart from that, a number of non-Newtonian fluids such as polymeric solutions, whole blood, or protein solutions with large polymeric molecules often exhibit elastic property with shear rate-dependent viscosity due to the stretching and coiling of the polymer chains [6, 7]. Novel applications may arise from non-Newtonian rheological behaviors [8]. For instance, three-dimensional (3D) focusing of microparticles can be achieved via an approach combining inertial and elastic forces in viscoelastic solutions [9]. The elasticity of the focusing fluid can facilitate formation of smaller droplets [10].

In this chapter, we aim to summarize the main technologies for non-Newtonian microfluidic multiphase system and discuss the recent advances in technologies that enable fabricating and characterizing functional materials formed from such systems. The chapter will start with review of rheological behavior of polymer solutions followed by discussion of the fundamental physical phenomena driving non-Newtonian microfluidic multiphase system where polymer solutions are involved; non-Newtonian viscosity effect on breakup dynamics and droplet formation; the techniques employed in device fabrication; and synthesis of emulsion-templat‐ ed materials. Emphasis will be placed on synthesis of functional materials from single emulsion, double emulsion, or higher order emulsions. Finally we conclude with an outlook to the future of the field. This chapter is meant to familiarize readers who may be new to the field of non-Newtonian microfluidics, as well as those readers who are new to the field of synthesis of functional materials, and eventually bridge the knowledge gap between the two disciplinary fields, leading to novel approaches for design of functional materials with tailored behaviors for specific applications.

## **2. Rheological characteristics of polymer solutions**

Newtonian fluids. In Newtonian fluids such as simple organic liquids, solutions of lowmolecular-weight inorganic salts, molten metals, or salts, the shear stress at steady condition in laminar flow is linearly proportional to the shear rate, i.e., the tensors that describe the viscous stress and the strain rate are related by a constant viscosity tensor which is independent of the stress state and velocity of the flow. In contrast, non-Newtonian fluids possess such properties that flow curve of shear stress versus shear rate is nonlinear or does not pass through the origin and the apparent viscosity as defined by shear stress divided by shear rate is not constant at a given temperature and pressure, as the apparent viscosity is also influenced by shear rate, the kinematic history of the fluid element, flow conditions, or microchannel configurations. The ever-expanding applications of microfluidic multiphase technologies increasingly require the use of non-Newtonian fluids, examples of which in daily life include cement paste, chocolate, coal slurries, greases and lubricating oils, molten polymer solutions, and cosmetics and personal care products such as shampoos, shaving foams, and toothpaste. Non-Newtonian multiphase microsystem has become a subject of intense research and is widely applied in biomedical engineering, food production, and energy applications. It is hence important to understand the associated physical phenomena, in particular the way how non-Newtonian rheological effect will alter breakup dynamics and, in turn, affect droplet formation, which can be characterized by droplet shape and size, as they are closely related with droplet stability as well as optical and mechanical properties [2]. For example, the shape and size of emulsion droplets have significant impacts on the drug release kinetics when they are used as template to synthesize microcapsules or microparticles for drug delivery [3, 4]. Monodisperse droplets with precisely controlled sizes can be used to deliver an accurate dosing of contained payload such as drug, flavoring, or chemical reactants [5]. Monodispersity and size tenability are highly desired to keep droplets exhibiting constant, controlled, and predictable properties. Nevertheless, the versatility in droplet size control is challenged by complex rheological properties of non-Newtonian fluids, such as formation of undesirable satellite droplets due to bead-on-string patterns, as a result of the stretching and thinning of non-Newtonian liquid filaments. Apart from that, a number of non-Newtonian fluids such as polymeric solutions, whole blood, or protein solutions with large polymeric molecules often exhibit elastic property with shear rate-dependent viscosity due to the stretching and coiling of the polymer chains [6, 7]. Novel applications may arise from non-Newtonian rheological behaviors [8]. For instance, three-dimensional (3D) focusing of microparticles can be achieved via an approach combining inertial and elastic forces in viscoelastic solutions [9]. The elasticity

360 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

of the focusing fluid can facilitate formation of smaller droplets [10].

In this chapter, we aim to summarize the main technologies for non-Newtonian microfluidic multiphase system and discuss the recent advances in technologies that enable fabricating and characterizing functional materials formed from such systems. The chapter will start with review of rheological behavior of polymer solutions followed by discussion of the fundamental physical phenomena driving non-Newtonian microfluidic multiphase system where polymer solutions are involved; non-Newtonian viscosity effect on breakup dynamics and droplet formation; the techniques employed in device fabrication; and synthesis of emulsion-templat‐ ed materials. Emphasis will be placed on synthesis of functional materials from single emulsion, double emulsion, or higher order emulsions. Finally we conclude with an outlook

Polymer solutions form a class of industrially significant materials exhibiting diversity of non-Newtonian rheological properties. The system with polymer solutions exhibits a rich spectrum of interesting and complex characteristics once the polymer molecules begin to interact, entangle, and knot up with each other. For example, a macromolecule responds to external force via straightening of chains, disentangling of loose networks, uncoiling and stretching, etc., in the presence of an imposed flow or force. Brownian effects tend to randomize the flow units present in such polymeric systems. The significant factors governing the rheological behavior of a polymeric system are the molecular weight, molecular weight distribution, the structure of the molecule, their possible configurations, and the chemical composition. The use of polymer solutions in microfluidic applications requires knowledge and understanding of their complex behaviors. First, we will review classification of non-Newtonian fluids and the mathematical models. Subsequently, the influence of microstructure on rheological behavior of polymer solutions will be discussed.

For an incompressible Newtonian fluid at steady condition in simple shear flow case where a thin layer of a fluid is contained between two parallel planes which have surface area of A and are parted by a distance dy, a shear by the application of a force F is exerted on the fluid, the shear rate may be expressed as the velocity gradient in the direction perpendicular to that of the shear force, and velocity vector has only one component along *x* direction with magnitude varying only in *y* direction, the shear will be balanced by an equal and opposite internal frictional force in the fluid:

$$\frac{F}{A} = \tau\_{y\alpha} = \eta \left( -\frac{dV\_x}{dy} \right) = \eta \dot{\gamma}\_{y\alpha} \tag{1}$$

where *τ* represents shear stress, *η* the viscosity of fluid of interest, V the velocity, and *γ*˙ the shear rate. The first subscript indicates the direction normal to that of shearing surface, while the second subscript refers to the direction of the force and the flow. The negative sign on the right hand side of Eq. (1) indicates that *τ* is a measure of the resistance to motion. Viscosity *η* only depends on material of interest, as well as temperature and pressure.

The shear stress for more general 3D case of an incompressible Newtonian fluid may be expressed for the *x* plane oriented normal to the *x* direction as follows [11]:

$$\sigma\_{\rm xx} = -2\eta \frac{\partial V\_{\rm x}}{\partial \mathbf{x}} + \frac{2}{3}\eta \left(\frac{\partial V\_{\rm x}}{\partial \mathbf{x}} + \frac{\partial V\_{\rm y}}{\partial \mathbf{y}} + \frac{\partial V\_{\rm z}}{\partial \mathbf{z}}\right) \tag{2}$$

$$
\sigma\_{xy} = -\eta \left( \frac{\partial V\_x}{\partial y} + \frac{\partial V\_y}{\partial \mathbf{x}} \right) \tag{3}
$$

$$\tau\_{xz} = -\eta \left( \frac{\partial V\_x}{\partial \mathbf{z}} + \frac{\partial V\_z}{\partial \mathbf{x}} \right) \tag{4}$$

Similar sets of equations can be developed for the forces acting on the *y* and *z* planes. In each case, there are two shearing components and one normal component. The nine stress compo‐ nents of a fluid element are shown in Figure 1. *τxy* =*τyx*,,nd *τyz* =*τzy*, because of the equilibrium of a fluid element. Each component of normal stress consists of two components: isotropic pressure and a contribution from flow depending on fluid type:

$$\begin{aligned} P\_{xx} &= -p + \tau\_{xx} & \quad \text{(a)}\\ P\_{yy} &= -p + \tau\_{yy} & \quad \text{(b)}\\ P\_{zz} &= -p + \tau\_{zz} & \quad \text{(c)} \end{aligned} \tag{5}$$

where p denotes the isotropic pressure and *τxx*, *τyy*, and *τzz* represent deviatoric normal stresses of Newtonian fluids or extra stresses of non-Newtonian fluids. For an incompressible New‐ tonian fluid, it is given by

$$p = -\frac{1}{3}(P\_{xx} + P\_{yy} + P\_{zz})\tag{6}$$

**Figure 1.** Stress components in 3D flow.

Based on Eqs. (5)–(6), it can be derived that

$$
\pi\_{xx} + \pi\_{yy} + \pi\_{zz} = 0 \tag{7}
$$

For a Newtonian fluid in simple shear flow,

$$
\pi\_{xx} = \pi\_{yy} = \pi\_{zz} = 0 \tag{8}
$$

Newtonian fluid thus not only possesses a constant viscosity but also satisfies the condition of Eq. (8). Boger fluids possess constant shear viscosity but do not conform to Eq. (8); thus they must be treated as non-Newtonian fluids. Three types of non-Newtonian fluids are ubiqui‐ tously applied and widely investigated [12–14]: (1) time-independent fluids (it can be further subdivided into three types including shear thinning, viscoplastic thickening, and shear thickening), (2) time-dependent fluids, and (3) viscoelastic fluids. Since most polymer solutions possess shear-thinning and viscoelastic behavior, only these rheological properties will be discussed in this chapter.

#### **2.1. Shear thinning**

<sup>2</sup> <sup>2</sup>

*xy*

*xz*

t

pressure and a contribution from flow depending on fluid type:

tonian fluid, it is given by

**Figure 1.** Stress components in 3D flow.

t

*xx*

th

362 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

3

 h

 h

*xx xx yy yy zz zz*

=- + =- + =- +

t

t

t

where p denotes the isotropic pressure and *τxx*, *τyy*, and *τzz* represent deviatoric normal stresses of Newtonian fluids or extra stresses of non-Newtonian fluids. For an incompressible New‐

( ) <sup>1</sup>

*P p P p P p*  h

*y xx z*

*y x*

*x z*

è ø ¶ ¶ (4)

<sup>3</sup> *xx yy zz p PPP* =- + + (6)

*V V z x*

æ ö ¶ ¶ =- + ç ÷

Similar sets of equations can be developed for the forces acting on the *y* and *z* planes. In each case, there are two shearing components and one normal component. The nine stress compo‐ nents of a fluid element are shown in Figure 1. *τxy* =*τyx*,,nd *τyz* =*τzy*, because of the equilibrium of a fluid element. Each component of normal stress consists of two components: isotropic

> ( ) ( ) ( )

a b c

*V V y x*

æ ö ¶ ¶ =- + ç ÷ ¶ ¶ è ø (2)

(3)

(5)

*VV V V x xyz*

¶¶ ¶ æ ö ¶ =- + + + ç ÷ ¶ ¶¶¶ è ø

> Pseudoplasticity or shear-thinning behavior is characterized by an apparent viscosity which decreases with increasing shear rate. However, most shear-thinning polymer solutions and melts exhibit Newtonian behavior at very low and very high shear rates. The values of the apparent viscosity at very low and high shear rates are known as the zero shear viscosity, *η*0, and the infinite shear viscosity, *η∞*, respectively. Thus, the apparent viscosity of a shearthinning fluid decreases from *η*0 to *η∞* with increasing shear rate. As molecular weight of the polymer is smaller, or its molecular weight distribution becomes narrower, or as polymer concentration drops, it will extend the range of shear rate over which the apparent viscosity is constant. A selection of the more widely used viscosity models is summarized here.

#### **i.** *The power law or Ostwald de Waele model*

The relationship between shear stress and shear rate for a shear-thinning fluid can often be approximated by a straight line over a limited range of shear rate (or stress) plotted on double logarithmic coordinates. This part of the flow curve can be modeled by power law or Ostwald de Waele expression in the following form:

$$
\pi = m\dot{\chi}^u\tag{9}
$$

so the apparent viscosity is thus given by

$$
\eta = \frac{\pi}{\dot{\gamma}} = m\dot{\gamma}^{n-1} \tag{10}
$$

In these equations, *m* and *n* are two empirical curve-fitting parameters and are known as the fluid consistency coefficient and the flow behavior index, respectively. The fluid exhibits shearthinning properties, Newtonian behavior, and shear-thickening behavior when n<1, n=1, and n>1, respectively. The smaller the value of *n,* the greater is the degree of shear thinning. This model applies for only a limited range of shear rates and therefore the fitted values of *m* and *n* will depend on the range of shear rates considered. The zero and infinite shear viscosities are not taken into account.

#### **ii.** *The Carreau viscosity equation*

The Carreau viscosity model incorporates both upper and lower limiting viscosities *η*0 to *η∞* and is given by

$$\frac{\eta - \eta\_{\circ}}{\eta\_{0} - \eta\_{\circ}} = \left( 1 + \left( \lambda \dot{\gamma} \right)^{2} \right)^{\left( n - 1 \right) / 2} \tag{11}$$

where *n* and λ are two curve-fitting parameters. This model can describe shear-thinning behavior over wide ranges of shear rates but only at the expense of the added complexity of four parameters.

#### **iii.** *The Cross viscosity equation*

This model has also gained wide acceptance and, in simple shear, is written as

$$\frac{\eta - \eta\_{\text{o}}}{\eta\_{0} - \eta\_{\text{o}}} = \frac{1}{1 + k \dot{\gamma}^{\text{u}}} \tag{12}$$

This model can be reduced to the Newtonian fluid behavior as k approaches 0 or the power law model as shown in Eq. (10) when *η*≪*η*0, and *η*≫*η∞*.

#### **2.2. Viscoelastic fluids**

Such substances exhibit characteristics of both ideal fluids and elastic solids and show partial elastic recovery after deformation, as they have the ability to store and recover shear energy. Examples include polymer melts, polymer and soap solutions, and synovial fluid. The shearing motion of a viscoelastic fluid gives rise to the first and second normal stress differences N1 and N2, expressed in terms of two coefficients, *ψ*1 and *ψ*2, which for 1D flow are defined by

$$\begin{aligned} \mathcal{W}\_1 &= \frac{N\_1}{\dot{\mathcal{V}}\_{yx}}^2 & \quad \text{(a)}\\ \mathcal{W}\_2 &= \frac{N\_2}{\dot{\mathcal{V}}\_{yx}}^2 & \quad \text{(b)} \end{aligned} \tag{13}$$

At very low shear rates, the first normal stress difference, *N*1, is normally proportional to the square of shear rate, and *N*1 is larger than the shear stress *τ* at the same value of shear rate. The ratio of *N*1 to *τ* is often taken as a measure of how elastic a liquid is. When recoverable shear, *N*<sup>1</sup> / 2*τ*, is greater than 0.5, it indicates a highly elastic behavior of the fluid. The two normal stress differences are used to categorize a fluid either as purely viscous (*N*<sup>1</sup> =0) or as viscoelastic, and the magnitude of *N*1 in comparison with *τyx*, is often used as a measure of viscoelasticity. N2 is usually small, with maximum values not exceeding 20% of N1 and with an opposite sign to N1. Measurement of N2 is difficult and can be done using a special coneplate apparatus [12]. N1 is responsible for some spectacular phenomena, such as the Weissen‐ berg effect [6]. The measurement of N1 can be conducted by commercial rotational rheometers. For the flow of polymer solutions in porous media, extensional effects are often encountered and the fluid is stretched as the extent and shape of the flow passages change. Elongational viscosity *ηE* is defined as

In these equations, *m* and *n* are two empirical curve-fitting parameters and are known as the fluid consistency coefficient and the flow behavior index, respectively. The fluid exhibits shearthinning properties, Newtonian behavior, and shear-thickening behavior when n<1, n=1, and n>1, respectively. The smaller the value of *n,* the greater is the degree of shear thinning. This model applies for only a limited range of shear rates and therefore the fitted values of *m* and *n* will depend on the range of shear rates considered. The zero and infinite shear viscosities

The Carreau viscosity model incorporates both upper and lower limiting viscosities *η*0 to *η∞*

( ( ) )

<sup>1</sup> *<sup>n</sup>*

lg

where *n* and λ are two curve-fitting parameters. This model can describe shear-thinning behavior over wide ranges of shear rates but only at the expense of the added complexity of

> 1 1 *<sup>n</sup> k*

This model can be reduced to the Newtonian fluid behavior as k approaches 0 or the power

Such substances exhibit characteristics of both ideal fluids and elastic solids and show partial elastic recovery after deformation, as they have the ability to store and recover shear energy. Examples include polymer melts, polymer and soap solutions, and synovial fluid. The shearing motion of a viscoelastic fluid gives rise to the first and second normal stress differences N1 and N2, expressed in terms of two coefficients, *ψ*1 and *ψ*2, which for 1D flow are defined by

> 1 1 2

> > *yx*

g

&

=

=

*N*

*N*

y

y

(a)

(b)

2 2 2

*yx*

g

&

g

0

¥ ¥

This model has also gained wide acceptance and, in simple shear, is written as

¥ ¥

0

h h

h h

law model as shown in Eq. (10) when *η*≪*η*0, and *η*≫*η∞*.

h h

364 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

h h ( ) 1 /2 <sup>2</sup>




(13)

are not taken into account.

and is given by

four parameters.

**2.2. Viscoelastic fluids**

**ii.** *The Carreau viscosity equation*

**iii.** *The Cross viscosity equation*

$$
\eta\_E = \frac{P\_{xx} - P\_{yy}}{\dot{\varepsilon}} = \frac{\tau\_{xx} - \tau\_{yy}}{\dot{\varepsilon}} \tag{14}
$$

where *ε*˙ is the elongational rate. Elongational viscosity can only be directly measured with devices like capillary breakup extensional rheometer [15]. The Trouton ratio, *Tr*, is defined as ..

$$T\_r = \frac{\eta\_E}{\eta} \tag{15}$$

For inelastic isotropic fluids, *Tr* =3 for all values of *ε*˙ and *γ*˙, and any deviation from the value of 3 indicate the existence of viscoelasticity. The Maxwell model is one of widely used linear viscoelastic models; a mechanical analogue of this model is obtained by series combinations of a spring and a dashpot. Combining the Hooke's law of elasticity and Newton's law of viscosity, one can obtain

$$
\tau + \mathcal{X}\dot{\tau} = \eta \dot{\gamma} \tag{16}
$$

where *τ*˙ is the time derivative of *τ*, *η* is the viscosity of the dashpot fluid, and *λ* (=*µ*/ *G)* is the relaxation time, which is a characteristic of the fluid. A more solid-like behavior is obtained by considering the so-called Voigt model which is represented by the parallel arrangement of a spring and a dashpot. The fluid total extra stress (*τt*) is given as the sum of an incompressible solvent contribution having a viscosity coefficient *η<sup>s</sup>* and a polymer/additive stress contribu‐ tion *τp*, as

$$
\pi\_t = \mathcal{D}\eta\_s \left(\mathcal{U}\_{D\_\*} \mathcal{U}\mathcal{U}\_D\right) \mathcal{D} + \pi\_p \tag{17}
$$

The solvent viscosity coefficient in Eq. (18) has been made to depend on the second and third invariants (*I ID*, *II ID*) of the rate of deformation tensor D, which is defined as

$$D = \frac{1}{2} \left( \nabla V + \nabla V^T \right) \tag{18}$$

For viscoelastic fluids, *ηs* is set to zero for polymer melts or to a nonzero constant when dealing with a polymer solution based on a Newtonian solvent. Flows in viscoelastic solutions are governed by the Deborah number, De; the Weissenberg number, Wi; and the elasticity number, El. The Deborah number is defined as the ratio between the relaxation time of the fluid (*λ*) and the time of observation of the flow (tf ), such as the duration of the unsteady part of a flow:

$$De = \frac{\lambda}{t\_f} \tag{19}$$

The Weissenberg number is defined as the product of the relaxation time and a characteristic rate of deformation of the flow (V/L) and quantifies the nonlinear response of the fluid

$$\text{Wi} = \lambda \frac{V}{L} \tag{20}$$

while El represents the ratio between elastic and inertial effects

$$El = \frac{\text{Wi}}{\text{Re}} = \frac{\text{\AA}\,\eta}{\rho L^2} \tag{21}$$

where *ρ* is the density. High Wi may lead to onset of elastic instabilities of viscoelastic fluids, even under creeping flow conditions. For instance, fluids with large polymeric molecules often exhibit elastic behavior due to the stretching and coiling of the polymeric chains, which significantly enrich flow behavior [6].

#### **2.3. Microstructure effect to rheology of polymers**

The zero shear viscosity of most polymeric systems increases linearly with the molecular weight below a critical value of the molecular weight, which denotes the onset of entanglement. The zero shear viscosity scales with molecular weight more significantly with the molecular weight above this value. The rheology of polymers is also affected by the nature of chains whether these are rod-like or coils and springs. For example, the charge-induced repulsive forces along the polymer chain will straighten the convoluted chain and impart some features of rod-like systems in the case of a polyelectrolyte in a weak electrolyte solution. Long molecules are normally randomly oriented at low shear rate, while the molecules become aligned along the direction of flow when increasing the shear rate progressively. The rheo‐ logical properties are also extremely sensitive to chemical composition. In polymer industry, it is a common technique for producing new rheological properties by adding fillers. The extent of modifications of viscosity-shear rate curves for filled systems depends upon the size and shape of fillers as well as concentration [16]. Blending of polymers can also be used to improve rheological characteristics as compared to that of the constituents. Such a blend may exhibit viscosity larger than that of its constituents. Another way to obtain improved mechanical properties is to create a block copolymer. Such a molecule has long sequences of one type of polymer connected to long sequences of another polymer and exhibits elastic behavior at room temperature and ease of flow at high temperatures, leading to wide range of applications. Surfactant systems also constitute an important class of industrial materials which exhibit complex rheological behavior. The surfactant molecules have hydrophobic tails and hydro‐ philic heads. When a surfactant like soap or detergent is added to water with concentration beyond a critical value, micelles comprising several molecules begin to form. The formed structures have temporary and loose networks which are continuously broken down and reformed. The shape of micelles includes spherical, rod-like, lamellar sheets and lamellar droplets, which are subject to several parameters, including temperature, the shape of the surfactant molecule, and electrolytic nature of water.

( ) , 2 *t sDD p*

The solvent viscosity coefficient in Eq. (18) has been made to depend on the second and third

( ) <sup>1</sup>

For viscoelastic fluids, *ηs* is set to zero for polymer melts or to a nonzero constant when dealing with a polymer solution based on a Newtonian solvent. Flows in viscoelastic solutions are governed by the Deborah number, De; the Weissenberg number, Wi; and the elasticity number, El. The Deborah number is defined as the ratio between the relaxation time of the fluid (*λ*) and

*f*

The Weissenberg number is defined as the product of the relaxation time and a characteristic rate of deformation of the flow (V/L) and quantifies the nonlinear response of the fluid

> *<sup>V</sup> Wi <sup>L</sup>* <sup>=</sup> l

> > <sup>2</sup> Re

*L* lh

r

where *ρ* is the density. High Wi may lead to onset of elastic instabilities of viscoelastic fluids, even under creeping flow conditions. For instance, fluids with large polymeric molecules often exhibit elastic behavior due to the stretching and coiling of the polymeric chains, which

The zero shear viscosity of most polymeric systems increases linearly with the molecular weight below a critical value of the molecular weight, which denotes the onset of entanglement. The zero shear viscosity scales with molecular weight more significantly with the molecular weight above this value. The rheology of polymers is also affected by the nature of chains

*Wi El*

while El represents the ratio between elastic and inertial effects

*De t* l  t

= + *II III D* (17)

*<sup>T</sup> D VV* = Ñ +Ñ (18)

), such as the duration of the unsteady part of a flow:

= (19)

(20)

= = (21)

th

366 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

invariants (*I ID*, *II ID*) of the rate of deformation tensor D, which is defined as

2

the time of observation of the flow (tf

significantly enrich flow behavior [6].

**2.3. Microstructure effect to rheology of polymers**

## **3. Non-Newtonian viscosity effect on breakup dynamics and droplet formation**

Emulsion droplets can be passively formed by fluid instabilities using four of the most prevalent microfluidic geometries found from literature: coaxial [17], flow focusing [18], T junction [19], and step emulsification [20]. The dynamics of the droplet formation in micro‐ system can be characterized by a transition from dripping to jetting regime, governed by the Weber number of the dispersed phase, Wein, defining the ratio of inertial force to surface tension and the capillary number of the continuous phase, Caout, defining the ratio of viscous force to surface tension:

$$\text{W}e\_{\text{in}} = \frac{\left. \rho\_{\text{D}^{\text{p}}} L V\_{\text{D}^{\text{p}}} \right|^{2}}{\sigma} \tag{22}$$

$$\mathbb{C}a\_{out} = \frac{\eta\_{\text{CP},0} V\_{\text{CP}}}{\sigma} \tag{23}$$

where *L* refers to the characteristic dimension of the microsystem and *ηCP*,0 is the apparent viscosity of the non-Newtonian fluid of continuous phase. The subscript "0" refers to zero shear rate when a shear-dependent fluid is used, and DP refers to dispersed phase. Formation regime of droplets and its transition are also studied in glass capillary device, where dripping regime and jetting regime are dominant and their transition depends on both capillary number and Weber number [21]. Droplet formation occurs at small Caout and Wein when flows are dominated by surface tension, while jetting forms at large Caout or Wein when the viscous stress or the inertial force on the droplet is large enough to overcome surface tension. Droplets are generated after breakup of a jet at some distance downstream in this regime. Emulsification processes using T- or Y-shaped junctions in poly(dimethylsiloxane) (PDMS) microfluidic systems have been applied for bioassays or drug delivery. The scaling laws of droplet formation by cross flow in T-junction devices were investigated, leading to discovery of three droplet formation patterns: (1) at Ca<0.002 (plug flow is dominant when squeezing regime occurs because n cross-flow shear is low and cannot rupture the disperse phase into drops), (2) at 0.01<Ca<0.3 (droplets are formed with a size comparable to channel cross-sectional dimension in dripping regime because shear force is large enough to cause breakup at the T junction), and (3) transition regime between the two. Simple models for dimension prediction of droplets and plugs as functions of reciprocal Ca were suggested for the squeezing and dripping regimes, respectively [22]. The formation of plug-shaped droplets and mixing in the microchannel with a T junction were studied [23]. While the dynamics of droplet breakup has been systematically investigated in Newtonian fluids, the validity of the understanding has not been adequately confirmed in non-Newtonian fluid systems. With increasing level of polymer solution involvement, systematic understanding of the relevant physics of droplet formation in non-Newtonian fluids has become particularly important. The complex rheolog‐ ical properties of non-Newtonian fluids also affect breakup dynamics and furthermore challenge the versatility in droplet size control. For example, the stretching and/or thinning of non-Newtonian liquid filaments will lead to the formation of "bead-on-string" patterns, as the interface pinch-off is greatly hindered by the polymers' tendency to recover their equilibrium state [24, 25]. These beads can subsequently become undesirable satellite droplets, increasing the polydispersity of the resultant droplet population. These examples attest to the need for a comprehensive understanding of the role of non-Newtonian viscosity effect in droplet formation using microfluidic systems. In this section, the non-Newtonian viscosity effect on the mechanism of droplet breakup, the characteristics of droplets such as final droplet size, and the frequency of droplet formation will be presented and discussed. We will focus on shear-thinning and viscoelastic fluids in different types of multiphase microsystems such as glass capillary device and planar PDMS device.

#### **3.1. Shear-thinning effect**

#### *3.1.1. Glass microcapillary device*

The dripping-to-jetting transition under various flow conditions in a Newtonian/shearthinning multiphase microsystem was characterized [17]. A numerical model of the microcapil‐ lary co-flow device has been developed with a Newtonian fluid injected in a cylindrical capillary as the dispersed phase at a constant average velocity *VDP* and a non-Newtonian outer phase injected through the coaxial square capillary as the continuous phase at a constant average velocity *VCP* (see Figure 2). The dispersed phase was silicon oil, while the continuous phase was 2% (w/v) aqueous solution of sodium rboxymethyl cellulose (CMC) (Mw=250,000 Da, DS=1.2), which is a pseudoplastic fluid and demonstrates shear-thinning behaviors with its apparent viscosity defined by Eq.(12) using Cross model. The breakup dynamics in a non-Newtonian fluid system was compared with a Newtonian fluid system at the same Weber number and capillary number using the same microdevice. For the Newtonian two-phase flow, silicone oil was also used as the dispersed phase, while 17% (w/v) aqueous solution of polyethylene glycol (PEG) (Mw=8,000 Da) was used as the continuous phase. The dispersed phase was purged out of the orifice (see Figure 3a) with a droplet growing in size and moving downstream while still connecting with the fluid neck through the orifice via a filament (see Figure 3b). The filament gradually became thinner (see Figure 3c) and finally broke up into a droplet (see Figure 3d). The process was repeated afterward. The interface between the two phases was tracked and compared with experimental measurements with reasonable agreement.

shear rate when a shear-dependent fluid is used, and DP refers to dispersed phase. Formation regime of droplets and its transition are also studied in glass capillary device, where dripping regime and jetting regime are dominant and their transition depends on both capillary number and Weber number [21]. Droplet formation occurs at small Caout and Wein when flows are dominated by surface tension, while jetting forms at large Caout or Wein when the viscous stress or the inertial force on the droplet is large enough to overcome surface tension. Droplets are generated after breakup of a jet at some distance downstream in this regime. Emulsification processes using T- or Y-shaped junctions in poly(dimethylsiloxane) (PDMS) microfluidic systems have been applied for bioassays or drug delivery. The scaling laws of droplet formation by cross flow in T-junction devices were investigated, leading to discovery of three droplet formation patterns: (1) at Ca<0.002 (plug flow is dominant when squeezing regime occurs because n cross-flow shear is low and cannot rupture the disperse phase into drops), (2) at 0.01<Ca<0.3 (droplets are formed with a size comparable to channel cross-sectional dimension in dripping regime because shear force is large enough to cause breakup at the T junction), and (3) transition regime between the two. Simple models for dimension prediction of droplets and plugs as functions of reciprocal Ca were suggested for the squeezing and dripping regimes, respectively [22]. The formation of plug-shaped droplets and mixing in the microchannel with a T junction were studied [23]. While the dynamics of droplet breakup has been systematically investigated in Newtonian fluids, the validity of the understanding has not been adequately confirmed in non-Newtonian fluid systems. With increasing level of polymer solution involvement, systematic understanding of the relevant physics of droplet formation in non-Newtonian fluids has become particularly important. The complex rheolog‐ ical properties of non-Newtonian fluids also affect breakup dynamics and furthermore challenge the versatility in droplet size control. For example, the stretching and/or thinning of non-Newtonian liquid filaments will lead to the formation of "bead-on-string" patterns, as the interface pinch-off is greatly hindered by the polymers' tendency to recover their equilibrium state [24, 25]. These beads can subsequently become undesirable satellite droplets, increasing the polydispersity of the resultant droplet population. These examples attest to the need for a comprehensive understanding of the role of non-Newtonian viscosity effect in droplet formation using microfluidic systems. In this section, the non-Newtonian viscosity effect on the mechanism of droplet breakup, the characteristics of droplets such as final droplet size, and the frequency of droplet formation will be presented and discussed. We will focus on shear-thinning and viscoelastic fluids in different types of multiphase microsystems such as

368 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

The dripping-to-jetting transition under various flow conditions in a Newtonian/shearthinning multiphase microsystem was characterized [17]. A numerical model of the microcapil‐ lary co-flow device has been developed with a Newtonian fluid injected in a cylindrical capillary as the dispersed phase at a constant average velocity *VDP* and a non-Newtonian outer phase injected through the coaxial square capillary as the continuous phase at a constant average

glass capillary device and planar PDMS device.

**3.1. Shear-thinning effect**

*3.1.1. Glass microcapillary device*

**Figure 2.** Schematic of the computational domain of multiphase microfluidic system. The closeup view of meshing of nozzle and inlets is shown in inset. Reproduced from Ref. [17] with permission from the Royal Society of Chemistry.

**Figure 3.** Time-lapse images of jet deformation and droplet formation from simulation (upper one in each sub-figure) and experiments (lower one in each sub-figure) using a Newtonian/shear-thinning two-phase co-flow system (silicon oil as dispersed phase and CMC solution as continuous phase). The blue and red scale bars are applicable for the ex‐ perimental and simulation results, respectively. (a) The dispersed phase is purged out of the orifice; (b) the main drop‐ let is connected with the fluid neck via a thin filament; (c) the filament becomes even much thinner; (d) satellite droplet occurs after breakup of the jet. Reproduced from Ref. [17] with permission from the Royal Society of Chemistry.

Droplet size scales inversely with the capillary number of the continuous phase in a monoto‐ nous fashion in the Newtonian/Newtonian two-phase system [26]. In contrast, the correlation between the droplet size and the capillary number in the Newtonian/shear-thinning two-phase system is different. As *Caout* and the shear rate of continuous phase increase, the viscosity of continuous phase is reduced for shear-thinning fluids such as CMC, leading to formation of larger droplets [27]. However, when *Caout* increases beyond a critical value, the viscous drag can overcome the surface tension effects that would otherwise minimize the stretching of the fluid neck by drawing the fluid interfaces closer to the orifice. The fluid neck becomes stretched and elongated at high *Caout*, fluid neck will subsequently become thinner, the shear rate of the continuous phase starts to decrease, the viscosity of continuous phase will increase for a shearthinning fluid, and eventually the droplet undergoes size reduction beyond a critical *Caout*.

#### *3.1.2. PDMS device*

The effects of the generalized power law coefficient, the power law exponent and the yield stress on the mechanism of drop breakup, final drop size, and frequency of drop formation were studied using 3D volume of fluid (VOF) model in multiphase flows in T-shaped micro‐ channel where the fluids are Newtonian/power law and Newtonian/Bingham [28]. The pressure implicit with splitting of operators (PISO) algorithm was used in the transient calculations for droplet generation and deformation. The technique of piecewise-linear interface construction (PLIC) [29] that satisfies the accuracy of the surface tension calculations was adopted in this numerical model. The droplet diameter decreased and frequency increased with n ≤ . 0.9 and n ≥ . 0.92. The products of *ηcp*. and *γ*˙ did not change much from K = 0.0011 to 0.0106 Pa s. For Bingham fluid, the droplet extension increased significantly as yield stress became larger. A plug-flow region was found in flows of Bingham fluids, and the width of the region was proportional to the yield stress. With a Bingham fluid as the continuous phase, an obvious transition period was found before the droplets can be formed steadily. During this transition period, droplets coalescence was observed. Microdroplet formation in different channel geometries has been studied. For instance, the formation of droplet from an aperture was investigated under the cross-flow conditions in microchannel emulsification process [30], where an oil phase was dispersed into shear-thinning continuous-phase fluid through a microchannel wall made of apertured substrate. The dispersed oil phase first grew to a jetlike deformed droplet, indicating that the net shear effect on the droplet surface in the flow direction of a non-Newtonian continuous phase is stronger than that in Newtonian flow. The detached droplet had a diameter less than half of the droplet size in the Newtonian flow case, showing a drastic reduction in the droplet size by the non-Newtonian effect in typical dripping regime, due to higher shear stress near the wall. Simulations of a droplet passing through an axisymmetric contraction were performed using VOF algorithm [31]. When the disperse phase was shear thinning, the local viscosity of the droplet decreased as it entered the contraction, remained low while within the contraction, and increased as it exited. When the continuous phase was shear thinning and the dispersed phase was Newtonian, the droplets tended to deform less than their shear-thinning counterpart when entering the contraction.

#### **3.2. Viscoelastic effect**

Droplet size scales inversely with the capillary number of the continuous phase in a monoto‐ nous fashion in the Newtonian/Newtonian two-phase system [26]. In contrast, the correlation between the droplet size and the capillary number in the Newtonian/shear-thinning two-phase system is different. As *Caout* and the shear rate of continuous phase increase, the viscosity of continuous phase is reduced for shear-thinning fluids such as CMC, leading to formation of larger droplets [27]. However, when *Caout* increases beyond a critical value, the viscous drag can overcome the surface tension effects that would otherwise minimize the stretching of the fluid neck by drawing the fluid interfaces closer to the orifice. The fluid neck becomes stretched and elongated at high *Caout*, fluid neck will subsequently become thinner, the shear rate of the continuous phase starts to decrease, the viscosity of continuous phase will increase for a shearthinning fluid, and eventually the droplet undergoes size reduction beyond a critical *Caout*.

370 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

The effects of the generalized power law coefficient, the power law exponent and the yield stress on the mechanism of drop breakup, final drop size, and frequency of drop formation were studied using 3D volume of fluid (VOF) model in multiphase flows in T-shaped micro‐ channel where the fluids are Newtonian/power law and Newtonian/Bingham [28]. The pressure implicit with splitting of operators (PISO) algorithm was used in the transient calculations for droplet generation and deformation. The technique of piecewise-linear interface construction (PLIC) [29] that satisfies the accuracy of the surface tension calculations was adopted in this numerical model. The droplet diameter decreased and frequency increased with n ≤ . 0.9 and n ≥ . 0.92. The products of *ηcp*. and *γ*˙ did not change much from K = 0.0011 to 0.0106 Pa s. For Bingham fluid, the droplet extension increased significantly as yield stress became larger. A plug-flow region was found in flows of Bingham fluids, and the width of the region was proportional to the yield stress. With a Bingham fluid as the continuous phase, an obvious transition period was found before the droplets can be formed steadily. During this transition period, droplets coalescence was observed. Microdroplet formation in different channel geometries has been studied. For instance, the formation of droplet from an aperture was investigated under the cross-flow conditions in microchannel emulsification process [30], where an oil phase was dispersed into shear-thinning continuous-phase fluid through a microchannel wall made of apertured substrate. The dispersed oil phase first grew to a jetlike deformed droplet, indicating that the net shear effect on the droplet surface in the flow direction of a non-Newtonian continuous phase is stronger than that in Newtonian flow. The detached droplet had a diameter less than half of the droplet size in the Newtonian flow case, showing a drastic reduction in the droplet size by the non-Newtonian effect in typical dripping regime, due to higher shear stress near the wall. Simulations of a droplet passing through an axisymmetric contraction were performed using VOF algorithm [31]. When the disperse phase was shear thinning, the local viscosity of the droplet decreased as it entered the contraction, remained low while within the contraction, and increased as it exited. When the continuous phase was shear thinning and the dispersed phase was Newtonian, the droplets tended to

deform less than their shear-thinning counterpart when entering the contraction.

*3.1.2. PDMS device*

#### *3.2.1. Glass microcapillary device*

The flow patterns at different ratios of flow rate with viscoelastic continuous phase of 5% w/v PAA solution and dispersed phase of silicon oil were characterized [17]. The elastic forces from viscoelastic continuous phase help to overcome interfacial tension and thus facilitate transition to jetting at smaller magnitudes of the viscous forces. Droplets adopted a spherical or nearly spherical shape in Newtonian/Newtonian or Newtonian/shear-thinning multiphase microsystem, while the droplets experienced significant deformation in the viscoelastic non-Newtonian continuous phase, because elasticity of the suspending liquid can facilitate the deformation of the Newtonian droplets [32]. For instance, the droplets adopted an elliptical shape after breakup and relaxed into pointed shapes. Droplets became more pointed as the radius decreased and the flow rate ratio increased before it transitioned to the jetting regime. The degree of droplet deformation increased, leading to formation of droplets with more pointed shape, when elastic effect became more pronounced at higher Weissenberg number. This work helps to understand how the complex rheology behavior of viscoelastic fluid can influence the breakup dynamics and droplet formation in multiphase microfluidic system when glass capillary device is used.

#### *3.2.2. PDMS device*

Highly controlled shear in viscoelastic media can lead to droplet rupturing down to a unique size and the shear-thinning nature enhances the monodispersity [33]. The droplet formation and breakup dynamics of elastic and Newtonian fluids in T-shaped microchannels for continuous phase were investigated [34]. For both cases, reduction in droplet size was found when cross-flow shear force was increased via increase of viscosity ratio of continuous phase to dispersed phase. For non-Newtonian case, after the droplet grew to a large enough size and was pushed downstream by the cross-flow, dispersed-phase filament was formed between the junction and the connected droplet. The filaments were stretched at two different rates and broke up into secondary smaller droplets. A critical Ca number existed for the inception of monodisperse secondary drops. The obvious presence of the filament during the drop detachment process of these elastic fluids when compared to its Newtonian counterpart shows that fluid elasticity played an important role in resisting drop pinch-off, and the presence of elasticity created a significant difference in drop formation dynamics, while this contributed little to the final drop size and drop productivity. The morphology of droplets was determined by polymer molecular weight and viscosity ratio. A slight reduction in primary drop size is expected with increasing levels of elastic stress within the filament, as the mass of fluid within the filament between two primary drops is excluded from effectively draining into the primary drops at either end of the filament. Analysis of the filament dynamics during the drop detachment stage revealed that there are two distinct regions: a pre-stretch region and an exponential self-thinning region. The effect of viscoelasticity of drop and medium on drop deformation was investigated, and the strategy to control the drop shape was proposed [35]. When a Newtonian drop was suspended in a viscoelastic medium, in the narrow channel region, ellipsoid-like droplet was found. In contrast, droplet normally adopted bullet-like shape in a Newtonian/Newtonian microsystem. When a viscoelastic drop was suspended in Newtonian medium, the drop swelling extent to the cross-stream direction was enhanced at the outlet of the channel, due to the normal stress difference developed in the viscoelastic fluid.

## **4. Synthesis of functional materials using microfluidics**

#### **4.1. Fabrication of microfluidic emulsification system**

The most common microfabrication methods for creating microfluidic devices are photoli‐ thography and soft lithography. Studies in optimizing these techniques have been extensively reported [36–40]. Photolithography actually refers to the transfer of a pattern of chromium on a hard material such as glass or silica plate and is known as the most important step in the microfabrication process. The whole process of photolithography involves substrate cleaning, photolithography, metal deposition and wet/ dry etching [41]. Silicon wafer is initially heated to a temperature of 150 °C to remove any moisture that may be present on the surface. Contaminant on the surface of silicon wafer can be removed by cleaning procedure [42]. After the cleaning process, silicon wafer is spin coated with a thin and uniform layer of photoresist. The silicon wafer along with the photoresist is then placed into the ultraviolet (UV) exposure machine for alignment and exposure. The exposure of photoresist toward UV light triggers chemical reaction; thus the exposed photoresist can be removed by a solution aimed to remove unreacted photoresist. The geometric pattern can be defined onto the silicon wafer by either positive photoresist or negative photoresist. Positive photoresist becomes soluble upon exposure whereas the unexposed regions become soluble when negative photoresist is used [43]. Soft lithography is a low-cost technique to replicate the microchannel pattern using the master mould generated with photolithography technique. In most of the applications, PDMS is the preferred elastomeric materials to be used as the patterned replica which transfer the original pattern of a master by molding or printing. The flexibility of the PDMS replicas allows patterning on nonplanar structures through micro-contact printing or micro-molding. As compared to chromium mask, film masks with similar pattern not only cost 20–100 times cheaper but also with a shorter production time. The subsequent process is very similar to the photolithography process with the only difference in which the negative replica is generated by casting an uncured prepolymer of PDMS against the geometry design [36]. However, these techniques require sophisticated instrumentation and clean room that sometimes are not readily available. Consequently, rapid yet low-cost prototyping technology, as a result of impressive development of microfabrication process aimed at reducing the fabrication interval and cost needed in conventional microfabrication processes such as xurography, was proposed to construct the devices [44, 45]. In addition, most of these techniques do not require cleanroom facilities as well as sophisticated steps to produce high-quality and long-lasting micro‐ fluidic devices [46]. Xurography utilized cutting plotter with a 10 µm resolution that can directly create microstructures without photolithographic process or chemicals [47]. The design pattern was cut onto an adhesive vinyl film. This was followed by removal of undesired portions of film using a pin/ blade. The remaining film with desired design was then trans‐ ferred onto a substrate, i.e., transparent plastic with the aid of the application tape to prevent deformation of the design. The transparent plastic along with the designated film was then used to create an epoxy master mould in order to produce replicas of PDMS device. Although xurography does not possess high resolution as compared to standard lithography techniques, the accuracy of this method was falling within 10 µm of drawn dimensions and the feature variability was less than 2 µm [47].

#### **4.2. Synthesis of particles using single-emulsion template**

region, ellipsoid-like droplet was found. In contrast, droplet normally adopted bullet-like shape in a Newtonian/Newtonian microsystem. When a viscoelastic drop was suspended in Newtonian medium, the drop swelling extent to the cross-stream direction was enhanced at the outlet of the channel, due to the normal stress difference developed in the viscoelastic fluid.

The most common microfabrication methods for creating microfluidic devices are photoli‐ thography and soft lithography. Studies in optimizing these techniques have been extensively reported [36–40]. Photolithography actually refers to the transfer of a pattern of chromium on a hard material such as glass or silica plate and is known as the most important step in the microfabrication process. The whole process of photolithography involves substrate cleaning, photolithography, metal deposition and wet/ dry etching [41]. Silicon wafer is initially heated to a temperature of 150 °C to remove any moisture that may be present on the surface. Contaminant on the surface of silicon wafer can be removed by cleaning procedure [42]. After the cleaning process, silicon wafer is spin coated with a thin and uniform layer of photoresist. The silicon wafer along with the photoresist is then placed into the ultraviolet (UV) exposure machine for alignment and exposure. The exposure of photoresist toward UV light triggers chemical reaction; thus the exposed photoresist can be removed by a solution aimed to remove unreacted photoresist. The geometric pattern can be defined onto the silicon wafer by either positive photoresist or negative photoresist. Positive photoresist becomes soluble upon exposure whereas the unexposed regions become soluble when negative photoresist is used [43]. Soft lithography is a low-cost technique to replicate the microchannel pattern using the master mould generated with photolithography technique. In most of the applications, PDMS is the preferred elastomeric materials to be used as the patterned replica which transfer the original pattern of a master by molding or printing. The flexibility of the PDMS replicas allows patterning on nonplanar structures through micro-contact printing or micro-molding. As compared to chromium mask, film masks with similar pattern not only cost 20–100 times cheaper but also with a shorter production time. The subsequent process is very similar to the photolithography process with the only difference in which the negative replica is generated by casting an uncured prepolymer of PDMS against the geometry design [36]. However, these techniques require sophisticated instrumentation and clean room that sometimes are not readily available. Consequently, rapid yet low-cost prototyping technology, as a result of impressive development of microfabrication process aimed at reducing the fabrication interval and cost needed in conventional microfabrication processes such as xurography, was proposed to construct the devices [44, 45]. In addition, most of these techniques do not require cleanroom facilities as well as sophisticated steps to produce high-quality and long-lasting micro‐ fluidic devices [46]. Xurography utilized cutting plotter with a 10 µm resolution that can directly create microstructures without photolithographic process or chemicals [47]. The design pattern was cut onto an adhesive vinyl film. This was followed by removal of undesired portions of film using a pin/ blade. The remaining film with desired design was then trans‐ ferred onto a substrate, i.e., transparent plastic with the aid of the application tape to prevent

**4. Synthesis of functional materials using microfluidics**

372 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

**4.1. Fabrication of microfluidic emulsification system**

The ability to form droplets with a microfluidic device allows one to structure fluids to disperse a continuous fluid into a series of equally sized liquid spheres. These spheres can then be solidified to produce particles of the same size and shape. Monodisperse microgels consisting of cross-linked network of poly(N-isopropylacrylamide) (PNIPAm) can be formed from single emulsion (see Figure 4) [48, 49]. Such PNIPAm microgels swell and shrank reversibly in response to changes in temperature. The size change occurred around 32°C, which is close to the human body temperature. Hence, these microgels have been extensively evaluated for controlled delivery of water-soluble drugs. Low polydispersity of PNIPAm microgels is desirable for drug delivery applications as it could lead to narrow distribution of drug loading levels and uniform release kinetics. Lipids or hydrocarbons are known as natural choice of the shell material as most of them are solid at room temperature, thus enabling the robust encapsulation under ambient conditions [50].

**Figure 4.** (a) Schematic illustration of a capillary-based microfluidic device for fabricating monodisperse PNIPAm mi‐ crogels. Fluid A is an aqueous suspension containing the monomer, cross-linker, and initiator; fluid B is oil, and fluid C is the same oil as fluid B but contains a reaction accelerator that is both water and oil soluble. The accelerator diffus‐ es into the drops and polymerizes the monomers to form monodisperse microgels. Cross-sectional views at different points along the device length are shown in the second row. Reproduced with permission from [48]. Copyright (2007) by Wiley-VCH Verlag GmbH & Co. KGaA. (b) Size change of PNIPAm microgels in water triggered by changing the temperature. The scale bar in Panel b denotes 100 µm. Reproduced with permission from [49]. Copyright (2008) by the Royal Society of Chemistry.

A demonstration of the fabrication of monodisperse colloidosomes, microcapsules with a shell composed of tightly packed colloidal particles, has been made using colloidal PNIPAm microgels as building blocks [51]. An aqueous suspension of amine-functionalized submicrometer-sized PNIPAm microgels was emulsified in an oil capsule using a single-emulsion microfluidic device. Prior to emulsification, a small amount of glutaraldehyde was added to the aqueous mixture. The colloidal PNIPAm microgels assembled at the oil-water interface within the emulsion droplets due to the presence of hydrophobic isopropyl groups and hydrophilic acrylamide groups. Glutaraldehyde molecules, owing to their two reactive sites each, served as connecting links between the amine-functionalized microgels through an amine-aldehyde condensation reaction. Colloidosomes were formed via interlinking of the microgels at the oil-water interface, as shown in Figure 5a. The colloidosomes exhibited similar thermosensitive behavior with the constituent microgels. The diameter of the colloidosomes decreased by 42%, roughly equivalent to an 80% decrease in volume, when the temperature was higher than the phase-transition temperature of PNIPAm (see Figure 5b). Such material is thus very promising in applications that require targeted pulsed release of active materials.

**Figure 5.** (a) Schematic representation of a technique used for fabricating colloidosomes using PNIPAm microgel parti‐ cles as building blocks and emulsion droplets as templates. (b) Equilibrium size change of PNIPAm colloidosomes and the constituent PNIPAm microgels. Colloidosomes were dispersed in water and heated from 20 to 50 °C in fixed incre‐ ments of 2 °C and then to 55 °C. Images were captured after allowing the sample to equilibrate for 30 min at each tem‐ perature. The sample was then cooled down to 20 °C using the same temperature steps. Size-change data of the constituent PNIPAm microgels over the same temperature range were collected using dynamic light scattering. Adapt‐ ed and reproduced with permission from [51]. Copyright (2010) by American Chemical Society.

#### **4.3. Synthesis of particles using double and higher order emulsion template**

Double emulsion can also be used to template structures, but these structures have more elaborated shapes. To illustrate this templating process, thermoresponsive PNIPAm was chosen as the matrix polymer to obtain environmentally sensitive microgel particles with microshell structure [52]. To form the pre-microgel droplets, an aqueous microgel precursor solution was emulsified in a continuous oil phase. Pre-microgel droplets were loaded with inner droplets of another oil in the formation process, thereby creating a shell structure, as shown in Figure 6a. After droplet formation, the shell was gelled by thermal monomer polymerization by photochemical polymer-analogous gelation, yielding uniform PNIPAm microshells as shown in Figure 6b. Operating the device with different flow rates can produce shells with two cores, as shown in Figure 6c.

Colloidosomes of diblock copolymers have been fabricated by combining water-in-oil-inwater (W/O/W) droplet encapsulation and dewetting transition in a microcapillary device [53].

microgels as building blocks [51]. An aqueous suspension of amine-functionalized submicrometer-sized PNIPAm microgels was emulsified in an oil capsule using a single-emulsion microfluidic device. Prior to emulsification, a small amount of glutaraldehyde was added to the aqueous mixture. The colloidal PNIPAm microgels assembled at the oil-water interface within the emulsion droplets due to the presence of hydrophobic isopropyl groups and hydrophilic acrylamide groups. Glutaraldehyde molecules, owing to their two reactive sites each, served as connecting links between the amine-functionalized microgels through an amine-aldehyde condensation reaction. Colloidosomes were formed via interlinking of the microgels at the oil-water interface, as shown in Figure 5a. The colloidosomes exhibited similar thermosensitive behavior with the constituent microgels. The diameter of the colloidosomes decreased by 42%, roughly equivalent to an 80% decrease in volume, when the temperature was higher than the phase-transition temperature of PNIPAm (see Figure 5b). Such material is thus very promising in applications that require targeted pulsed release of active materials.

374 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

**Figure 5.** (a) Schematic representation of a technique used for fabricating colloidosomes using PNIPAm microgel parti‐ cles as building blocks and emulsion droplets as templates. (b) Equilibrium size change of PNIPAm colloidosomes and the constituent PNIPAm microgels. Colloidosomes were dispersed in water and heated from 20 to 50 °C in fixed incre‐ ments of 2 °C and then to 55 °C. Images were captured after allowing the sample to equilibrate for 30 min at each tem‐ perature. The sample was then cooled down to 20 °C using the same temperature steps. Size-change data of the constituent PNIPAm microgels over the same temperature range were collected using dynamic light scattering. Adapt‐

Double emulsion can also be used to template structures, but these structures have more elaborated shapes. To illustrate this templating process, thermoresponsive PNIPAm was chosen as the matrix polymer to obtain environmentally sensitive microgel particles with microshell structure [52]. To form the pre-microgel droplets, an aqueous microgel precursor solution was emulsified in a continuous oil phase. Pre-microgel droplets were loaded with inner droplets of another oil in the formation process, thereby creating a shell structure, as shown in Figure 6a. After droplet formation, the shell was gelled by thermal monomer polymerization by photochemical polymer-analogous gelation, yielding uniform PNIPAm microshells as shown in Figure 6b. Operating the device with different flow rates can produce

Colloidosomes of diblock copolymers have been fabricated by combining water-in-oil-inwater (W/O/W) droplet encapsulation and dewetting transition in a microcapillary device [53].

ed and reproduced with permission from [51]. Copyright (2010) by American Chemical Society.

shells with two cores, as shown in Figure 6c.

**4.3. Synthesis of particles using double and higher order emulsion template**

**Figure 6.** Microfluidic production of hollow microgel shells. (a) A glass microcapillary device is used to create an oilwater-oil double emulsion with a semidilute solution of cross-linkable pNIPAAm as aqueous phase. Subsequently, these droplets are cured by UV exposure as they flow through a delay capillary a few centimeters downstream (not shown). (b) pNIPAAm microshells obtained from the experiment in Panel A. (c) Double-core microshells obtained upon slight variation of the flow rates in the experiment. All scale bars denote 200 µm. Reproduced with permission from [52]. Copyright (2010) by the Royal Society of Chemistry.

The properties of the colloidosomes, for example, membrane thickness, mechanical response, permeability, and thermal stability, can be tuned by varying the block ratios of the block copolymers and the homopolymer [54]. Multicompartmental polymersomes can be generated based on encapsulation of a controlled number of single polymersomes through a reinjection method [55]. Besides using glass capillary devices, double emulsions can be formed in PDMS device. This can be achieved using two drop makers in series [56–58]. The first drop maker produced the inner drops, which were fed into the inlet of the second drop maker, which produced the outer drops. This type of double emulsification can be achieved using either cascading T junctions or flow-focus drop makers. Generally, serial flow-focus devices are used, because the symmetric injection of the middle and continuous phases helps prevent wetting of the drops on the channels, making the formation more robust. In addition to geometrical considerations like this, it is also necessary to control the wetting properties of the devices; however, in contrast to single-emulsion formation in which only uniform wetting is required, spatially patterned wettability is required in double or higher order multiple emulsification.

Wettability determines the type of emulsion formed in PDMS device channel: oil-in-water emulsions are formed in hydrophilic channels, while water-in-oil emulsions are formed in hydrophobic channels. The need to spatially control the wettability in PDMS devices has stimulated techniques to spatially modify device surface properties. With a photo-patterning approach, the devices were filled with a solution that only reacts with the channels under exposure to intense UV light [58]. Since the UV beam can be shaped using holes, slits, and lenses, this approach allows for the creation of complex wettability patterns. However, in the case of double-emulsion devices, only a very simple wettability pattern is needed, and the added complication of having to align the photo-pattern with the microchannels can make this approach unattractive in these instances. For the simple patterns needed for these devices, a different approach is available that trades versatility in the pattern shape for simplicity in the treatment process. This flow-patterning technique controls wettability by introducing a reactive solution into select portions of a device and preventing it from going using the flow of an inert blocker solution [59]. The reactive solution changes the wettability only in the treated regions, while the untreated areas retain their default wettability. By controlling how the solutions are injected, wettability patterns for W/O/W or O/W/O double emulsions can be created.

One of the advantages of PDMS-based microfluidics is the ability to customize the devices to construct sophisticated channel networks. High-order multiple emulsions can be formed by scaling up device complexity. A high-order multiple emulsion is the logical continuation of a double emulsion to larger numbers of nested droplets: A triple emulsion is a double emul‐ sion encapsulated within a droplet, while a quadruple emulsion is merely a triple emulsion inside yet another droplet. As shown in Figure 7a, double emulsions can be formed by addition of another droplet maker; together also pattern wettability, so that the inner drops were encapsulated within larger drops of an immiscible phase (see Figure 7b) [60]. The design can be further improved to make higher order emulsions such as triple, quadruple, and quintu‐ ple emulsions by addition of third, fourth, and fifth junctions, as shown in Figure 7c–e, respectively.

**Figure 7.** Serial drop maker arrays used to form multiple emulsions. A single-emulsion droplet takes one drop maker to create (a), whereas double, triple, quadruple, and quintuple emulsions take two, three, four, and five drop makers in series to create, (b), (c), (d), and (e), respectively. The wettability of the multiple emulsion devices has been patterned so that it inverts between hydrophilic and hydrophobic from one junction to the next. Reprinted with permission from [60]. Copyright (2009) by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim.

A number of polymer solutions cannot be used to synthesize particles in microfluidic devices, as challenged by their non-Newtonian properties. For instance, lipid melts such as coco glycerides tend to be viscous and stick to channel surfaces because of the amphiphilic prop‐ erties of the constituent molecules; they also have low interfacial tension with oil or aqueous carrier phases. Polymer solutions of long-chain polymers such as PNIPAm or polyurethanepolybutadienediol (pU-pBDO) can form excellent particles, but appear to be viscous and have significant viscoelastic response at the shear rates required for controlled droplet formation, thus significantly limiting the applicability. One-step double emulsification enables formation of monodisperse particles from fluids that cannot normally be used in droplet-based micro‐ fluidics, including viscous or viscoelastic polymer solutions or low interfacial tension polymer melts. For example, a robust droplet formation mechanism has been proposed and demon‐ strated with highly viscous inner jet surrounded by a Newtonian fluid in one-step double emulsification, leading to formation of equal-sized droplets (see Figure 8) [61]. It takes long time for the droplet to relax into a spherical shape because of high viscosity of the fluid. The shape of droplet can be fixed by rapid solidification such as photoinduced cross-linking. This provides alternate approach to synthesize new types of particles, such as those made from waxy lipids or semidilute entangled polymer solutions.

**Figure 8.** Demonstration of failing and succeeding emulsification using single- and one-step double emulsification, re‐ spectively. (a) Emulsification of a viscoelastic semidilute pNIPAM polymer solution and the resultant particles below. (b) Emulsification of a viscous and viscoelastic polymer solution composed of pU-pBDO and the resultant particles below. (c) Emulsification of a low interfacial tension and viscous lipid melt and the resultant particles below. All scale bars in the lower panels denote 100 µ m. Reprinted with permission from [61]. Copyright (2011) by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim.

## **5. Outlook**

regions, while the untreated areas retain their default wettability. By controlling how the solutions are injected, wettability patterns for W/O/W or O/W/O double emulsions can be

376 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

One of the advantages of PDMS-based microfluidics is the ability to customize the devices to construct sophisticated channel networks. High-order multiple emulsions can be formed by scaling up device complexity. A high-order multiple emulsion is the logical continuation of a double emulsion to larger numbers of nested droplets: A triple emulsion is a double emul‐ sion encapsulated within a droplet, while a quadruple emulsion is merely a triple emulsion inside yet another droplet. As shown in Figure 7a, double emulsions can be formed by addition of another droplet maker; together also pattern wettability, so that the inner drops were encapsulated within larger drops of an immiscible phase (see Figure 7b) [60]. The design can be further improved to make higher order emulsions such as triple, quadruple, and quintu‐ ple emulsions by addition of third, fourth, and fifth junctions, as shown in Figure 7c–e,

**Figure 7.** Serial drop maker arrays used to form multiple emulsions. A single-emulsion droplet takes one drop maker to create (a), whereas double, triple, quadruple, and quintuple emulsions take two, three, four, and five drop makers in series to create, (b), (c), (d), and (e), respectively. The wettability of the multiple emulsion devices has been patterned so that it inverts between hydrophilic and hydrophobic from one junction to the next. Reprinted with permission from

A number of polymer solutions cannot be used to synthesize particles in microfluidic devices, as challenged by their non-Newtonian properties. For instance, lipid melts such as coco glycerides tend to be viscous and stick to channel surfaces because of the amphiphilic prop‐ erties of the constituent molecules; they also have low interfacial tension with oil or aqueous carrier phases. Polymer solutions of long-chain polymers such as PNIPAm or polyurethanepolybutadienediol (pU-pBDO) can form excellent particles, but appear to be viscous and have significant viscoelastic response at the shear rates required for controlled droplet formation, thus significantly limiting the applicability. One-step double emulsification enables formation

[60]. Copyright (2009) by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim.

created.

respectively.

Functional materials such as microcapsules, microgels, colloidosomes, and microshells are ubiquitous in daily life and play crucial role in numerous industrial applications driven by microencapsulation. The materials can be obtained from the breakage of dispersed phase into droplets with controlled structure and monodispersed size in multiphase microsystem using emulsion as template. However, the practical applications of the technique are challenged as the fluid precursors must be compatible with the formation of droplets in microfluidic devices. For instance, fluids with a low viscosity, negligible viscoelastic response, and moderate interfacial tension are required to facilitate formation of droplets. If even one of these con‐ straints is not satisfied, polydisperse particles are formed or even jetting occurs. This may happen especially when one has to deal with biological fluids and polymeric solutions that contain molecules of high molecular weight. These complex fluids exhibit a non-Newtonian behavior which may considerably affect both the jetting/dripping transition and the growth of perturbations in the jetting regime. A number of novel approaches have been proposed in the last few years to fulfill research needs to achieve better control over drop size, degree of monodispersity, internal structure, and production rate. In this chapter, we have reviewed these novel microdroplet-based techniques when non-Newtonian fluids are involved. We focused on the synthesis of functional materials and how they can be linked with real industrial and biochemical demands. Despite of the rapid development, one of the major opened avenues for investigation is the material forming, through which one obtains a final solid phase, normally in the form of powders or suspensions made of microparticles, microcapsules, etc. More efforts should be given to the relevant research so that material forming can be carefully designed and well controlled to prevent unfavorable effects including coalescence, aggrega‐ tion, or shell degradation and disintegration. Attention should also be paid to the development of advanced approach to inhibit the occurrence of satellite droplets and control polydispersity when non-Newtonian fluids are used in microfluidic system. This will eventually help to achieve applications such as advanced drug delivery.

## **Acknowledgements**

This work was supported by Zhejiang Provincial Natural Science Foundation of China under Grant No. Q15E090001, Young Scientist Program from National Natural Science Foundation of China under Grant No. NSFC51506103/E0605, 2014 Zhejiang 1000-Talents Awards and Ningbo Natural Science Foundation under Grant No. 2015A610281.

## **Author details**

Yong Ren1\*, Kai Seng Koh2 and Yaping Zhang3

\*Address all correspondence to: yong.ren@nottingham.edu.cn

1 Department of Mechanical, Materials and Manufacturing Engineering, The University of Nottingham Ningbo, China

2 Department of Chemical and Environmental Engineering, The University of Nottingham Ningbo, China

3 Department of Electrical and Electronic Engineering, The University of Nottingham Ningbo, China

## **References**

the fluid precursors must be compatible with the formation of droplets in microfluidic devices. For instance, fluids with a low viscosity, negligible viscoelastic response, and moderate interfacial tension are required to facilitate formation of droplets. If even one of these con‐ straints is not satisfied, polydisperse particles are formed or even jetting occurs. This may happen especially when one has to deal with biological fluids and polymeric solutions that contain molecules of high molecular weight. These complex fluids exhibit a non-Newtonian behavior which may considerably affect both the jetting/dripping transition and the growth of perturbations in the jetting regime. A number of novel approaches have been proposed in the last few years to fulfill research needs to achieve better control over drop size, degree of monodispersity, internal structure, and production rate. In this chapter, we have reviewed these novel microdroplet-based techniques when non-Newtonian fluids are involved. We focused on the synthesis of functional materials and how they can be linked with real industrial and biochemical demands. Despite of the rapid development, one of the major opened avenues for investigation is the material forming, through which one obtains a final solid phase, normally in the form of powders or suspensions made of microparticles, microcapsules, etc. More efforts should be given to the relevant research so that material forming can be carefully designed and well controlled to prevent unfavorable effects including coalescence, aggrega‐ tion, or shell degradation and disintegration. Attention should also be paid to the development of advanced approach to inhibit the occurrence of satellite droplets and control polydispersity when non-Newtonian fluids are used in microfluidic system. This will eventually help to

378 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

This work was supported by Zhejiang Provincial Natural Science Foundation of China under Grant No. Q15E090001, Young Scientist Program from National Natural Science Foundation of China under Grant No. NSFC51506103/E0605, 2014 Zhejiang 1000-Talents Awards and

1 Department of Mechanical, Materials and Manufacturing Engineering, The University of

2 Department of Chemical and Environmental Engineering, The University of Nottingham

3 Department of Electrical and Electronic Engineering, The University of Nottingham

achieve applications such as advanced drug delivery.

Ningbo Natural Science Foundation under Grant No. 2015A610281.

and Yaping Zhang3

\*Address all correspondence to: yong.ren@nottingham.edu.cn

**Acknowledgements**

**Author details**

Ningbo, China

Ningbo, China

Yong Ren1\*, Kai Seng Koh2

Nottingham Ningbo, China


[32] Tretheway, D.C., Leal, L.G. (2001). Deformation and relaxation of Newtonian drops in planar extensional flows of a Boger Fluid, J. Non-Newtonian Fluid Mech. 99, 81– 108.

[17] Ren, Y., Liu, Z., Shum, H.C. (2015) Breakup dynamics and dripping-to-jetting transi‐ tion in a shear-thinning/Newtonian multiphase microsystem, Lab. Chip. 15, 121–134.

380 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

[18] Roberts, C.C., Rao, R.R., Loewenberg, M., Brooks, C.F., Galambos, P., Grillet, A.M., Nemer, M.B. (2012). Comparison of monodisperse droplet generation in flow-focus‐ ing devices with hydrophilic and hydrophobic surfaces, Lab. Chip. 12, 1540–1547.

[19] Wang, K., Lu, Y.C., Xu, J.H., Tan, J., Luo, G.S. (2011). Generation of micromonodis‐ persed droplets and bubbles in the capillary embedded T-junction microfluidic devi‐

[20] Couture, O., Faivre, M., Pannacci, N., Babataheri, A., Servois, V., Tabeling, P., Tanter,

[21] Utada, A.S., Fernandez-Nieves, A., Stone, H.A., Weitz, D.A. (2007). Dripping to jet‐

[22] Xu, J.H., Li, S.W., Tan, J., Luo, G.S. (2008). Correlations of droplet formation in Tjunction microfluidic device: from squeezing to dripping, Microfluid. Nanofluid. 5,

[23] Tice, J.D., Lyon, A.D., Ismagilov, R.F. (2004). Effects of viscosity on droplet formation

[24] Oliveira, M.S.N., McKinley, G.H. (2005). Iterated stretching and multiple beads-on-astring phenomena in dilute solutions of highly extensible flexible polymers, Phys.

[25] Eggers, J., Villermaux, E. (2008). Physics of liquid jets, Rep. Prog. Phys., 71, 036601.

[26] Stan, C.A., Tang, S.K.Y., Whitesides, G.M. (2009). Independent control of drop size and velocity in microfluidic flow-focusing generators using variable temperature and

[27] Taylor, G.I. (1932). The viscosity of a fluid containing small drops of another fluid,

[28] Hong, Y., Wang, F. (2007). Investigation of viscosity effect on droplet formation, Mi‐

[29] Meier, M., Yadigaroglu, G., Smith, B.L. (2002). A novel technique for including sur‐

[30] Qiu, D.M., Silva, L., Tonkovich, A.L., Arora, R. (2010). Micro-droplet formation in non-Newtonian fluid in a microchannel, Microfluid. Nanofluid. 8, 531–548.

[31] Harvie, J.E., Davidson, M.R., Cooper-White, J.J., Rudman, M. (2007). A parametric study of droplet deformation through a microfluidic contraction: shear thinning liq‐

face tension in PLIC-VOF methods, Eur. J. Mech. B Fluids. 2, 61–73.

M. (2011). Ultrasound internal tattooing, Med. Phys. 38, 1116–1123.

ting transition in coflow liquid streams, Phys. Rev. Lett. 99, 094502.

and mixing in microfluidic channels, Anal. Chim. Acta. 507, 73–77.

ces, J. AIChE. 57, 299–306.

711–717.

Fluids. 17, 071704.

flow rate, Anal. Chem. 81, 2399–2402.

Proc. R. Soc. London. A. 138, 41–48.

uids, Int. J. Multiphas. Flow. 33, 545–556.

crofluid. Nanofluid. 3, 341–346.


[60] Abate, A.R., Weitz, D.A. (2009). High-order multiple emulsions formed in poly(dime‐ thylsiloxane) microfluidics, Small. 5, 2030–2032.

[46] Cornelius, T.L. (2002). Computer Aided and Integrated Manufacturing Systems, Vol.

[47] Bartholomeusz, D.A., Boutté, R.W., Andrade, D. (2005). Xurography: rapid prototyp‐ ing of microstructures using a cutting plotter, J. Microelectromechanical Syst. 14,

[48] Chu, L.Y., Kim, J.W., Shah, R.K., Weitz, D.A. (2007). Monodisperse thermoresponsive microgels with tunable volume-phase transition kinetics, Adv. Funct. Mater. 17,

[49] Shah, R.K., Kim, J.-W., Agresti, J.J., Weitz, D.A., Chu, L.-Y. (2008). Fabrication of monodisperse thermosensitive microgels and gel capsules in microfluidic devices,

[50] Datta, S.S., Abbaspourrad, A., Amstad, E., Fan, J., Kim, S.H., Romanowsky, M., Shum, H.C., Sun, B., Utada, A.S., Windbergs, M., Zhou, S., Weitz, D.A. (2014). 25th anniversary article: double emulsion templated solid microcapsules: mechanics and

[51] Shah, R.K., Kim, J-.W., Weitz, D.A. (2010). Monodisperse stimuli-responsive colloido‐ somes by self-assembly of microgels in droplets, Langmuir. 26, 1561–1565.

[52] Seiffert, S., Weitz, D.A. (2010). Controlled fabrication of polymer microgels by poly‐ mer-analogous gelation in droplet microfluidics, Soft Matter. 6, 3184–3190.

[53] Shum, H.C., Lee, D., Yoon, I., Kodger, T., Weitz, D.A. (2008). Double emulsion tem‐

[54] Shum, H.C., Kim, J.-W., Weitz, D.A. (2008). Microfluidic fabrication of monodisperse biocompatible and biodegradable polymersomes with controlled permeability, J.

[55] Kim, S.-H., Shum, H.C., Kim, J.W., Cho, J.-C., Weitz, D.A. (2011). Multiple polymer‐ somes for programmed release of multiple components, J. Am. Chem. Soc. 133,

[56] Nisisako, T., Okushima, S., Torii, T. (2005). Controlled formulation of monodisperse double emulsions in a multiple-phase microfluidic system, Soft Matter. 1, 23–27. [57] Okushima, S., Nisisako, T., Torii, T., Higuchi, T. (2004). Controlled production of monodisperse double emulsions by two-step droplet breakup in microfluidic devi‐

[58] Abate, A.R., Krummel, A.T., Lee, D., Marquez, M., Holtze, C., Weitz, D.A. (2008). Photoreactive coating for high-contrast spatial patterning of microfluidic device

[59] Abate, A.R., Thiele, J., Weinhart, M., Weitz, D.A. (2010). Patterning microfluidic de‐

vice wettability using flow confinement, Lab Chip. 10, 1774–1776.

plated monodisperse phospholipid vesicles, Langmuir. 24, 7651–7653.

3: Optimization Methods, World Scientific, Singapore.

382 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

1364–1374.

3499–3504.

Soft Matter. 4, 2303–2309.

Am. Chem. Soc. 130, 9543–9549.

ces, Langmuir. 20, 9905–9908.

wettability, Lab Chip. 8, 2157–2160.

15165–15171.

controlled release, Adv. Mater. 26, 2205–2218.

[61] Abate, A.R., Kutsovsky, M., Seiffert, S., Windbergs, M., Pinto, L.F.V., Rotem, A., Uta‐ da A.S., Weitz, D.A. (2011). Synthesis of monodisperse microparticles from non-Newtonian polymer solutions with microfluidic devices, Adv. Mater. 23, 1757–1760.

## **High and Efficient Production of Nanomaterials by Microfluidic Reactor Approaches**

Victor Sebastian Cabeza

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/64347

#### **Abstract**

This chapter overviews different approaches for the synthesis of nanostructured materi‐ als based on alternative methodologies to the most conventional and widespread colloi‐ dal wet chemical route and with a great potential applicability to large-scale and continuous production of nanomaterials. Their major outcomes, current progress in syn‐ thesis of micro and nanostructures by using microfluidics techniques and potential appli‐ cations for the next future are reviewed throughout three different sections. Emphasis is placed on nanomaterials production basics, nanomaterials production techniques and mi‐ crofluidic reactors (types, materials, designs). The integration of nanoparticle and micro‐ reactor technologies delivers enormous possibilities for the further development of novel materials and reactors. In this chapter, recent achievements in the synthesis of nanoparti‐ cles in microfluidic reactors are stated. A variety of strategies for synthesizing inorganic and polymeric nanoparticles are presented and compared, including continuous flow, gas–liquid segmented flow and droplet-based microreactors

**Keywords:** Miclofluidics, inorganic nanomaterials, polymeric nanomaterilas, microreac‐ tors, nanoengineering

### **1. Introduction**

After more than twenty years of basic and applied research, nanotechnologies are gaining in commercial use due to their unprecedented potential. Nanomaterials are the heart of the nanotechnology, they are the key in the development of electronics, medical diagnostics and therapeutics, energy storage, cosmetics, catalysts, lubricants, pigments, healthcare. In fact, the latest update available from the inventory compiled by the Woodrow Wilson Center [1] indicates that the number of nanotechnology-based consumer products was above 1800 in 2014 and growing at a strong pace. These products are produced in more than 33 countries and can

© 2016 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

be classified into different groups: Appliances, Automotive, Cross Cutting, Electronics and Computers, Food and Beverage, Goods for Children, Health and Fitness or Home and Garden. The composition of these products could contain up to more than 49 different types of nanomaterials, from gold to liposomes. Then, it is evident that during the past two or three decades there has been an explosive increase in our ability to nanoengineer and product nanostructures and nanosystems with an good degree of control. Therefore, given the expected economic and social impact of nanotechnology products and the fact that many areas of application are still scarcely explored, it seems reasonable to predict that industrial use of nanomaterials will continue to increase in the foreseeable future.

Nanomaterials can be defined as materials development at the atomic, molecular or macro‐ molecular levels, in the length of scale of approximately 1- to 100-nm range in at least one dimension. Materials under these dimensions behave in a different mode that their bulk counterparts. The scale reduction provides special properties which are uniquely attributable to the nanoscale physical size. Consequently, it seems reasonable that the control in the synthesis procedure must be good enough to assure the properties of the nanomaterials produced. Otherwise, the quality of the products which contain those nanomaterials could be limited.

New materials properties are required as the Nanotechnology market is growing, which implies that highly sophisticated configurations at the nanoscale level need to be developed. These strict requirements on the characteristics of nanomaterials pose serious challenges to their mass production. In addition, many of the most sophisticated nanomaterials are nowa‐ days prepared in lab scale by complex, multistep batch procedures that are not amenable to large-scale production [2]. This issue has been identified as one of the main barriers for the development of nanotechnology [3] and any potential technology able to overcome these drawbacks will enable the future development of Nanotechnology.

Microscale process engineering is the science of conducting chemical or physical processes inside small confined volumes, for instance inside channels with diameters of less than 1 mm (microchannels) or other structures with sub-millimeter dimensions [4]. These processes are usually carried out as continuous production with the potential to make microscale process a key for chemical production [4]. In fact, microscale process engineering enables an accurate control in the productivity of pharmaceutical and fine chemical industry. The small dimen‐ sions where the chemical reaction occurs lead to a relatively large surface area-to-volume ratios and promotes the heat and mass transport. In addition, the possibility of fast mixing of reactants and fast heating and cooling of reaction mixtures enables precise control of the reaction parameters. These advantages are well suited in the production of nanomaterials, where an exquisite control of synthesis conditions is required to assure an excellent quality.

This chapter overviews different approaches for the synthesis of nanostructured materials based on the application of microreactors derived from the microscale process engineering. The basics of nanomaterials production, the synthesis approaches to produce nanomaterials in continuous mode, present state-of-the-art and potential applications of microreactors for the next future are reviewed.

#### **1.1. Overview of synthetic approaches to nanomaterials production**

be classified into different groups: Appliances, Automotive, Cross Cutting, Electronics and Computers, Food and Beverage, Goods for Children, Health and Fitness or Home and Garden. The composition of these products could contain up to more than 49 different types of nanomaterials, from gold to liposomes. Then, it is evident that during the past two or three decades there has been an explosive increase in our ability to nanoengineer and product nanostructures and nanosystems with an good degree of control. Therefore, given the expected economic and social impact of nanotechnology products and the fact that many areas of application are still scarcely explored, it seems reasonable to predict that industrial use of

Nanomaterials can be defined as materials development at the atomic, molecular or macro‐ molecular levels, in the length of scale of approximately 1- to 100-nm range in at least one dimension. Materials under these dimensions behave in a different mode that their bulk counterparts. The scale reduction provides special properties which are uniquely attributable to the nanoscale physical size. Consequently, it seems reasonable that the control in the synthesis procedure must be good enough to assure the properties of the nanomaterials produced. Otherwise, the quality of the products which contain those nanomaterials could be

New materials properties are required as the Nanotechnology market is growing, which implies that highly sophisticated configurations at the nanoscale level need to be developed. These strict requirements on the characteristics of nanomaterials pose serious challenges to their mass production. In addition, many of the most sophisticated nanomaterials are nowa‐ days prepared in lab scale by complex, multistep batch procedures that are not amenable to large-scale production [2]. This issue has been identified as one of the main barriers for the development of nanotechnology [3] and any potential technology able to overcome these

Microscale process engineering is the science of conducting chemical or physical processes inside small confined volumes, for instance inside channels with diameters of less than 1 mm (microchannels) or other structures with sub-millimeter dimensions [4]. These processes are usually carried out as continuous production with the potential to make microscale process a key for chemical production [4]. In fact, microscale process engineering enables an accurate control in the productivity of pharmaceutical and fine chemical industry. The small dimen‐ sions where the chemical reaction occurs lead to a relatively large surface area-to-volume ratios and promotes the heat and mass transport. In addition, the possibility of fast mixing of reactants and fast heating and cooling of reaction mixtures enables precise control of the reaction parameters. These advantages are well suited in the production of nanomaterials, where an exquisite control of synthesis conditions is required to assure an excellent quality.

This chapter overviews different approaches for the synthesis of nanostructured materials based on the application of microreactors derived from the microscale process engineering. The basics of nanomaterials production, the synthesis approaches to produce nanomaterials in continuous mode, present state-of-the-art and potential applications of microreactors for

nanomaterials will continue to increase in the foreseeable future.

386 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

drawbacks will enable the future development of Nanotechnology.

limited.

the next future are reviewed.

The fabrication of nanomaterials of tailored properties involves the control of size, shape, structure and composition. Fabrication techniques for nanostructures can be broadly divided into two categories: Top‐down and Bottom‐up approaches (Figure 1). Top‐down approaches consist of the miniaturization or size reduction (for instance by etching or milling) of larger structures. On the other hand, bottom‐up approaches are based on growth and self-assembly to build up nanostructures from atomic or molecular precursors. Lithographic patterning is usually the most common technique to structure bulk materials at the nanoscale. Optical lithography (also named Photolithography), has been the predominant patterning technique due to the fact that it enables the production of sub‐100‐nm patterns with the use of very short‐ wavelength light (currently 193 nm).

**Figure 1.** Scheme of the most relevant production techniques of nanomaterials

Regarding bottom‐up approaches, they offers the most realistic solution toward the fabrication of complex and functional nanomaterials since the resolution and dimension control at the nanoscale level is better than in top-down approaches. Botton‐up techniques can be classified into gas phase and wet‐chemical methods, depending on the medium at which nanoparticles form (see Figure 1). The most important gas phase approaches comprise Physical Vapour Deposition, Chemical Vapour Deposition, Laser ablation, Molecular Beam Spraying, and Spraying‐Aerosol. Physical vapor deposition (PVD) is a versatile method for preparing thin‐ film materials with structural control at the nanometer or even atomic scale by carefully monitoring the processing parameters. PVD involves the generation of vapour phase species via evaporation, sputtering, or laser ablation. Chemical vapor deposition (CVD) is a technique for thin film deposition which offers an excellent uniformity and it is amenable to large scale production. In CVD the substrate is exposed to one or more volatile precursor materials that react and/or decompose on the substrate surface to produce the desired deposit. Molecular beam epitaxy (MBE) is a technique to produce ultrathin films as high quality epitaxial layers with very sharp interfaces and good control of thickness and chemical composition. To obtain epitaxial layers of high purity, the deposition takes place in very high vacuum. Spraying is a well‐stablished method, which involves the atomization of chemical precursors into tiny droplets dispersed in a gas phase. Afterwards, the drops are combusted or evaporated to form nanoparticles or thin films. Pyrolysis is the chemical decomposition of organic materials by heating in the absence of oxygen or any other reagents. The use of a laser source induces a fast heating, in a solid target (ablation) either on the gas phase or in a liquid phase, which promotes the formation of nuclei and direct the growth of nanoparticles, for instance metal oxides.

Wet‐Chemical techniques have been considered as one of the most important techniques for fabricating low‐dimensional structures with an accurate control on the dimensions and composition. These techniques seem to be very attractive since they enable the production of a wide diversity of nanomaterials, at a low cost and with a high throughput to ease a high‐ volume production (see Figure 1). Nanomaterials can be tailored with sub-nanometer level accuracy by careful regulation of crystallization kinetics and thermodynamic parameters in liquid media under assistance of selected solvents, ligands and surfactants. Wet‐ Chemical techniques can be classified into: microemulsions, thermal decomposition, hydrothermal and solvothermal synthesis, sol‐gel method, photochemical, sonochemical and electrochemical processes.

Microemulsions are ternary/quaternary systems of water, oil and amphiphilic surfactants forming micelles. Micelles form after the surfactant concentration is above the critical micellar concentration. Microemulsions can be generated after the solubilization of organic compounds in water (oil‐in water, o/w), or hydrophilic compounds in the oil phase (water in oil, w/o). In addition, a high degree of complexity can be achieved producing multiple phase emulsions, such as w/o/w or o/w/o. Nanomaterials are usually produced by mixing two microemulsions, where each other contain a reactive chemical [5]. Once the microemulsions merge, new micelles are formed and the chemical reaction occurs. Then, micelles have the role of nano/micro reactor, confining the chemicals during the reaction. Consequently, microemulsions are a versatile preparation technique which enables an accurate control of nanoparticle properties such as size, geometry, morphology, homogeneity and surface area. Thermal decomposition is usually carried out with organometallic precursors compounds and metal complexes with the assistance of high boiling solvents. Hydrothermal and solvothermal synthesis are common methods to synthesize inorganic nanomaterials. These techniques take advantage of the high reactivity of metal precursors at elevated pressure and temperature. A good control on crystallization parameters such as pressure, temperature, time, pH and reactants concentration enable to tune the size and shape of nanomaterials. Sol-gel method it is generally used for the production of metal oxide nanomaterials and involves the hydrolysis, condensation and polymerization of metal precursors. A three-dimensional network is formed after the polime‐ rization step, forming the gel, but a calcination treatment is usually required in order to drive off the structural water and transform the metal hydroxides into metal oxides. Photochemical techniques are based on the UV-irradiation of the chemical precursors to generate active radicals which promote the NPs formation with a controlled kinetic rate. On the other hand, sonochemical approaches are associated with the absorption of ultrasound by water. Acoustic wavelengths range from 10 to 10-4cm, which is far above the molecular or atomic range. Then, sonochemical reactions occur by the phenomena of acoustic cavitation, where the liquid is compressed or expanded according to the sound field. Hot spots are formed during the collapsing of bubbles generated during acoustic cavitation, achieving local temperatures of upto 4000K [6]. These conditions make water molecules dissociate into H and OH radicals and promote the reduction of metallic ions. Finally, chemical and electrochemical reduction approaches are one of the possible and powerful options for the fabrication of new types of nanomaterials. Chemical reduction relies in the presence of a substance, named reducing agent, with an oxidation potential high enough to easily donate electrons to an electron recipient compound, which is usually a metallic salt. If this electron transfer occur, metallic ions are reduced to atoms which form nuclei and later nanoparticles according to the nuclea‐ tion-growth-ripening process [7]. However, this procedure is highly dependent from the presence of a proper reducing agent which enables the electron donation without potential interferences with the crystallization kinetics. On the other hand, electrochemical reduction processes, which involve electron transfer reactions at solid–liquid interfaces controlled by an externally applied voltage, are reliable tools for nanofabrication since the redox potential can be easily tuned according the redox potential required.

#### **1.2. Nanomaterials production basics**

react and/or decompose on the substrate surface to produce the desired deposit. Molecular beam epitaxy (MBE) is a technique to produce ultrathin films as high quality epitaxial layers with very sharp interfaces and good control of thickness and chemical composition. To obtain epitaxial layers of high purity, the deposition takes place in very high vacuum. Spraying is a well‐stablished method, which involves the atomization of chemical precursors into tiny droplets dispersed in a gas phase. Afterwards, the drops are combusted or evaporated to form nanoparticles or thin films. Pyrolysis is the chemical decomposition of organic materials by heating in the absence of oxygen or any other reagents. The use of a laser source induces a fast heating, in a solid target (ablation) either on the gas phase or in a liquid phase, which promotes the formation of nuclei and direct the growth of nanoparticles, for instance metal oxides.

388 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

Wet‐Chemical techniques have been considered as one of the most important techniques for fabricating low‐dimensional structures with an accurate control on the dimensions and composition. These techniques seem to be very attractive since they enable the production of a wide diversity of nanomaterials, at a low cost and with a high throughput to ease a high‐ volume production (see Figure 1). Nanomaterials can be tailored with sub-nanometer level accuracy by careful regulation of crystallization kinetics and thermodynamic parameters in liquid media under assistance of selected solvents, ligands and surfactants. Wet‐ Chemical techniques can be classified into: microemulsions, thermal decomposition, hydrothermal and solvothermal synthesis, sol‐gel method, photochemical, sonochemical and electrochemical

Microemulsions are ternary/quaternary systems of water, oil and amphiphilic surfactants forming micelles. Micelles form after the surfactant concentration is above the critical micellar concentration. Microemulsions can be generated after the solubilization of organic compounds in water (oil‐in water, o/w), or hydrophilic compounds in the oil phase (water in oil, w/o). In addition, a high degree of complexity can be achieved producing multiple phase emulsions, such as w/o/w or o/w/o. Nanomaterials are usually produced by mixing two microemulsions, where each other contain a reactive chemical [5]. Once the microemulsions merge, new micelles are formed and the chemical reaction occurs. Then, micelles have the role of nano/micro reactor, confining the chemicals during the reaction. Consequently, microemulsions are a versatile preparation technique which enables an accurate control of nanoparticle properties such as size, geometry, morphology, homogeneity and surface area. Thermal decomposition is usually carried out with organometallic precursors compounds and metal complexes with the assistance of high boiling solvents. Hydrothermal and solvothermal synthesis are common methods to synthesize inorganic nanomaterials. These techniques take advantage of the high reactivity of metal precursors at elevated pressure and temperature. A good control on crystallization parameters such as pressure, temperature, time, pH and reactants concentration enable to tune the size and shape of nanomaterials. Sol-gel method it is generally used for the production of metal oxide nanomaterials and involves the hydrolysis, condensation and polymerization of metal precursors. A three-dimensional network is formed after the polime‐ rization step, forming the gel, but a calcination treatment is usually required in order to drive off the structural water and transform the metal hydroxides into metal oxides. Photochemical techniques are based on the UV-irradiation of the chemical precursors to generate active

processes.

Nanotechnology is a relatively new field; therefore applications of nanoparticles offer much promise to improve in future. However, most of applications based on nanoparticles are dealing only a few nanoparticle materials such as gold, silver, silica, alumina, carbon nano‐ tubes, titanium oxide and zinc oxide. But, many other interesting nanomaterials that show unique and useful properties still remain on the laboratory bench due to slow progress in bridging laboratories with high-scale production factories [8]. The lack of development of large scale production is being considered as one of the main obstacles. For instance, carbon nanotubes (CNTs) was considered a remarkable nanomaterial due to their unique physical and chemical properties. However the number of applications of CNTs was limited at the Early 90´s because the prize was too high (more than 1000\$/gram). Nowadays, Microreaction technology is expected to have a number of advantages for chemical production [2] as the high heat and mass transfer rates possible in microfluidic systems allow reactions to be performed under more aggressive conditions with higher yields that can be achieved with conventional reactors [10].In addition, new reaction pathways considered too difficult to explore in con‐ ventional microscopic equipment could be pursued because if the microreactor fails, the small amount of chemicals released accidentally could be easily contained. Microreaction technology enables a continuous production, as well as allows the presence of integrated sensor and control units to follow the nanomaterials quality or even alert about any fails. A failed microreactor can be isolated and replaced while other parallel units continued production. Also these inherent safety characteristics could allow a production scale systems of multiple microreactors enabling a distributed point-of-use synthesis of chemicals with storage and shipping limitations, such as highly reactive and toxic intermediates [2,10]. Moreover, scaleup to production by replication of microreactors units used in the laboratory would eliminate costly redesign and pilot plant experiments, thereby shortening the development time from laboratory to commercial-scale production.

**Figure 2.** Factors to be considered in the commercial scale-up of nanomaterials production.

## **2. Microfluidic technology**

Microreactors or Microfluidic reactors constitute perhaps the enabling technology of highest potential for liquid phase synthesis of nanomaterials since they have been proposed to overcome the inherent discontinuity and reproducibility of batch reactors, the inter and intrabatch dispersion regarding the physicochemical characteristics of the product nanomaterials and the difficulties regarding scale-up [2]. In addition to their well-known enhancement of heat and mass transfer, microfluidic reactors offer a flexible operation with modular design, with the possibility of increasing the number of modules to adapt to the specific process requirements. On the other hand, the scaling-up by arraying parallel microreactors allows capital investment and market growth to be coupled, reducing financial risk, which is especially important in emerging areas such as nanotechnology [2]. This section will deal with a summary of the most representative types of microreactors used in nanomaterials produc‐ tion, the synthesis approaches and few examples of nanomaterials production in continuous fashion.

#### **2.1. Microreactors typology: From simple to complex**

Considering that a microreactor is a micro vessel designed to contain chemical reactions, its design should maximize the reaction performance, which in nanomaterials typically means achieving the highest yield to the desired particles shape, with a tunable size, as well as an homogenous size distribution. Considering the chemicals and the synthesis conditions, the function-relevant criteria, such as corrosion resistance, wettability, mechanical strength, and the temperature range in which the chemical reaction occurs and the material can be used, must be considered. In addition, some other aspects should be considered: price, availability, and workability. Nevertheless, no single material fulfils all these criteria to a completely ideal extent and it means that a compromise is usually required between the most important variables.

up to production by replication of microreactors units used in the laboratory would eliminate costly redesign and pilot plant experiments, thereby shortening the development time from

**Figure 2.** Factors to be considered in the commercial scale-up of nanomaterials production.

Microreactors or Microfluidic reactors constitute perhaps the enabling technology of highest potential for liquid phase synthesis of nanomaterials since they have been proposed to overcome the inherent discontinuity and reproducibility of batch reactors, the inter and intrabatch dispersion regarding the physicochemical characteristics of the product nanomaterials and the difficulties regarding scale-up [2]. In addition to their well-known enhancement of heat and mass transfer, microfluidic reactors offer a flexible operation with modular design, with the possibility of increasing the number of modules to adapt to the specific process requirements. On the other hand, the scaling-up by arraying parallel microreactors allows capital investment and market growth to be coupled, reducing financial risk, which is especially important in emerging areas such as nanotechnology [2]. This section will deal with a summary of the most representative types of microreactors used in nanomaterials produc‐ tion, the synthesis approaches and few examples of nanomaterials production in continuous

Considering that a microreactor is a micro vessel designed to contain chemical reactions, its design should maximize the reaction performance, which in nanomaterials typically means

laboratory to commercial-scale production.

390 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

**2. Microfluidic technology**

**2.1. Microreactors typology: From simple to complex**

fashion.

Microreactors have evolved from simple capillary tubing (Figure 3-a) to complex design integrating valves, control of operation parameters and in-situ characterization, appropriately named lab on a chip. Regarding capillary tubing microfluidic reactors, there are many choices in terms of materials (e.g., polytetrafluoroethylene (PTFE), glass, and fused silica) with an inner diameter of up to 25 µm. The most commonly used microfluidic reactors are made of polymers due to the fact that these microreactors are ease to fabricate at low cost and with an acceptable feature resolution (Figure 3). Polymer tubing made of transparent or semi-transparent material ease the optical access while the nanocrystallization reaction is carried out, which is convenient to follow the reaction advance (Figure 3-b). Among the polymer tubes, those made of PTFE are the most widely used since they can tolerate a broad range of chemicals and elevated temperatures up to 240 °C. Considering the hydrophobic nature of PTFE, it naturally resists channel fouling and blockage when aqueous streams are injected in the reactor.

The straight tubing is a simple design that suffers from mixing limitations. Such mixing limitations (slow mixing) are undesirable for the synthesis of uniform nanocrystals, especially in the case when fast reduction or decomposition of a nanoparticle precursor is involved. Segmented flow is usually arranged to overcome this limitation (this flow approach will be discussed in section 2.2). Chaotic advection can effectively accelerate mixing in slug flow by using unsteady fluid flow to stretch and fold the slugs in a channel. Figure 3-c,d show that using a pinched tubing accelerate the mixing in comparison with a straight channel. However, it is not easy to create a homogenous distribution of pinched zones. Polymer tubing is usually used to fabricate coaxial flow microsystems (Figure 3-f)). These systems are used to produce microparticles by the drop-flow approach (this concept will be discussed in sections 2.2 and 2.4). For instance to produce o/w microparticles, an aqueous solution is pumped through the inner tubing. Oil stream is pumped through the outer tubing and it exerts shear on the aqueous phase at the annular junction, causing uniform microdroplets to break away and flow down the tubing.

An alternative to tubing microfluidic reactors is the use of microfluidic chips fabricated using the popular soft lithography technique. The most commonly fabricated microsystems are made of poly(dimethyl siloxane) (PDMS) and glass. The PDMS layer is peeled off and bonded to a glass slide to seal the channels (Figure 3 e, h, i, j). Figure 3-e depicts an elaborated PDMS-Glass microsystem, where an excellent control on fluid-dynamics is achieved. The channel dimensions (length and diameter) can be tuned according to the nano-crystallization reaction requirements, inserting meandering channels to improve the reagents mixing, as well as fabricating the proper micro-channels to control the pressure drop. However, these microsys‐ tems have a limited operating temperature and pressures, as well as a poor chemical compat‐ ibility, since PDMS is swelled by most of the organic solvents. Some microreactors made of SU-8-PEEK have been reported to withstand temperatures and pressures up to 150ºC and 2 MPs, respectively [11].

Sophisticated PDMS-glass microreactors can also be fabricated to produce microparticles by the before mentioned two-phase method using a co-flowing system. Differently from tubing co-flowing microsystems, lithography enables an accurate control on the nozzle zone where the non-miscible phases are put into contact, providing a superior control over fluid-dynamics. In addition, the microparticles production unit can be coupled to a polymerization unit by UV radiation, where a spiraling and gradually widening channel enables maximum absorption of radiated UV light for an inline photo-polymerization without coalescence and clogging issues (Figure 3-g) [16]. Figure 3-h depicts a PDMS glass microfluidic reactor also designed to receive a maximum absorption of LED radiation in order to promote a photochemical nucleation and shape selectivity towards certain Ag nanostructures. Figure 3-i shows another co-flowing system made in PDMS to produce PNIPAM polymeric nanoparticles with a tuneable size by the polymerization of NIPAM monomer. Monodisperse W/O/W double emulsions were prepared in glass microcapillary device, forming phospholipid vesicles using monodisperse double emulsions with a core-shell structure as templates (Figure 3-j). Finally, MPEG-PLGA nanoparticles can be produced with a high production speed by nanoprecipitation using a parallel flow focusing PDMS-Glass microfluidic reactor [19]. The microdevice contains 100 parallel outlets to improve the mixing of polymer-solvent-antisolvent system by diffusion and achieve a mixing time shorter than the nanoprecipitation time. An excellent mixing control was achieved and monodisperse nanoparticles was produced in a mixing time shorter than 60ms. The production speed can be further improved by several orders of magnitude just by increasing the number of outlet streams [19].

Metal based microreactors are especially advantageous for processes involving high heat load and toxic chemicals, except for strong acids, due to the good chemical compatibility and thermal resistance. Metal microreactors are usually fabricated using electrical discharge machining, laser ablation or etching techniques, which allow to achieve a precise shaping down to micrometer level. Highly precise microstructes can be also generated by the well-known lithographic structuring techniques (LIGA). Some of the difficulties in using metal microreac‐ tors arose from the leaks at the joints, but these days there are have been developed several welding, soldering and joining techniques that enable to create high-strength inter-metallic bonds. Metal microfluidic reactors withstand robust handling, which is quite convenient during cleaning operations. They have been also demonstrated in the synthesis of nanomate‐ rials. Size-tunable methacrlylic nanoparticles have been successfully produced by stainlesssteel multilamination micromixer (Figure 4-a). The excellent mixing achieved in the micromixer promoted the nucleation step at a high extension and enabled to decrease the nanoparticle sizes under the ones obtained in batch reactors, saving also organic solvent. The same type of micromixer was also used to produce Au-SiO2 nanoshells in a multi-step process (Figure 4-b). The micromixers enabled to improve the control over the optical properties of these nanoparticles, compared to the conventional batch process operation. A stainless steel High and Efficient Production of Nanomaterials by Microfluidic Reactor Approaches http://dx.doi.org/10.5772/64347 393

tems have a limited operating temperature and pressures, as well as a poor chemical compat‐ ibility, since PDMS is swelled by most of the organic solvents. Some microreactors made of SU-8-PEEK have been reported to withstand temperatures and pressures up to 150ºC and 2

392 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

Sophisticated PDMS-glass microreactors can also be fabricated to produce microparticles by the before mentioned two-phase method using a co-flowing system. Differently from tubing co-flowing microsystems, lithography enables an accurate control on the nozzle zone where the non-miscible phases are put into contact, providing a superior control over fluid-dynamics. In addition, the microparticles production unit can be coupled to a polymerization unit by UV radiation, where a spiraling and gradually widening channel enables maximum absorption of radiated UV light for an inline photo-polymerization without coalescence and clogging issues (Figure 3-g) [16]. Figure 3-h depicts a PDMS glass microfluidic reactor also designed to receive a maximum absorption of LED radiation in order to promote a photochemical nucleation and shape selectivity towards certain Ag nanostructures. Figure 3-i shows another co-flowing system made in PDMS to produce PNIPAM polymeric nanoparticles with a tuneable size by the polymerization of NIPAM monomer. Monodisperse W/O/W double emulsions were prepared in glass microcapillary device, forming phospholipid vesicles using monodisperse double emulsions with a core-shell structure as templates (Figure 3-j). Finally, MPEG-PLGA nanoparticles can be produced with a high production speed by nanoprecipitation using a parallel flow focusing PDMS-Glass microfluidic reactor [19]. The microdevice contains 100 parallel outlets to improve the mixing of polymer-solvent-antisolvent system by diffusion and achieve a mixing time shorter than the nanoprecipitation time. An excellent mixing control was achieved and monodisperse nanoparticles was produced in a mixing time shorter than 60ms. The production speed can be further improved by several orders of magnitude just by

Metal based microreactors are especially advantageous for processes involving high heat load and toxic chemicals, except for strong acids, due to the good chemical compatibility and thermal resistance. Metal microreactors are usually fabricated using electrical discharge machining, laser ablation or etching techniques, which allow to achieve a precise shaping down to micrometer level. Highly precise microstructes can be also generated by the well-known lithographic structuring techniques (LIGA). Some of the difficulties in using metal microreac‐ tors arose from the leaks at the joints, but these days there are have been developed several welding, soldering and joining techniques that enable to create high-strength inter-metallic bonds. Metal microfluidic reactors withstand robust handling, which is quite convenient during cleaning operations. They have been also demonstrated in the synthesis of nanomate‐ rials. Size-tunable methacrlylic nanoparticles have been successfully produced by stainlesssteel multilamination micromixer (Figure 4-a). The excellent mixing achieved in the micromixer promoted the nucleation step at a high extension and enabled to decrease the nanoparticle sizes under the ones obtained in batch reactors, saving also organic solvent. The same type of micromixer was also used to produce Au-SiO2 nanoshells in a multi-step process (Figure 4-b). The micromixers enabled to improve the control over the optical properties of these nanoparticles, compared to the conventional batch process operation. A stainless steel

MPs, respectively [11].

increasing the number of outlet streams [19].

**Figure 3.** Examples of polymeric microreactors for nanomaterials synthesis. (a)PTFE tubing. b) PTFE tubing microreac‐ tor, where the slug dimensions are modified to tune the crystallization kinetics of MOFs nanocrystals, (reprinted with permission from ref. [12], copyright 2013 American Chemical Society). c)-d) Straight and pinched PTFE microreactor to enhance the reactants mixing during the production of Au nanocrystals, (reprinted with permission from ref. [13], copyright 2014 John Wiley and Sons). e) PDMS/Glass microreactor designed to produce SiO2 nanocystals (reprinted with permission from ref. [14], copyright 2004 American Chemical Society), f) PTFE coaxial microreactor designed to produce chitosan microparticles (reprinted with permission from ref. [15], copyright 2013 The Royal Society of Chem‐ istry) g) PDMS-glass microsystem to produce uniform microbeads loaded with quantum dots. (reprinted with permis‐ sion from ref. [16], copyright 2014 American Chemical Society). h) PDMS-glass microsystem to produce Ag nanoparticles with a controlled size and shep by LED ilumination. (reprinted with permission from ref. [17], copyright 2013 The Royal Society of Chemistry).i) PDMS-glass microsystem to produce uniform PNIPAM nanoparticles by mon‐ omer polymerization. (reprinted with permission from ref. [18], copyright 2011 WILEY-VCH Verlag GmbH & Co.), j) Glass parallel-flow focusing microsystem to produce uniform Phospholipid Vesicles. (reprinted with permission from ref. [26], copyright 2008 American Chemical Society).

spiral microreactor was used to control the production of Ag nanoparticles by gas and liquid segmented flow [22].

Silicon and glass-based microfluidic reactors are fabricated by the conventional lithographic techniques and have the advantage of high bond strengths which enable to work at high pressure. Differently from glass microreactors, silicon microreactors can be used in high temperature reactions because the high thermal conductivity of silicon. However, both materials are fragile, and these reactors do not withstand mechanical impacts. A spiral silicon- pyrex reactor was designed to have separated the mixing and reaction units and promote a homoge‐ nous nucleation after a good mixing was achieved (Figure 4-c). This reactor could control the growth of Au nanoparticles, which is a pretty fast reaction and require from an excellent mixing control. The same design of spiral microreactor was applied in the synthesis of mesoporous nanomaterials (MCM-41), decreasing the synthesis time from days (batch reactor) to minutes

and improving the size and shape control [24]. Figure 4-d depicts a silicon-based microreac‐ tor where there are integrated two mixing units to control the growth and composition of Au/Ag hetero-structures. Glass microcapillary reactors can be fabricated with a higher resolution at micrometer scale than the polymer reactors ones, this fact enables to produce a better control in the production of emulsions and polymeric nanoparticles (Figure 3-j).

Finally, this section should highlight the existence of some other types of microreactors with a considered relevance in terms of functionality. Figure 4-e illustrates a continuous three-stage silicon-based microsystem consisting of mixing, aging and sequential growth stages, operating at 65 bar and temperatures as high as 340ºC. This system is an example of complex fluiddynamic control by the injection of different precursor flow-rates and pressure drop along the different microchannels. On the other hand, silicon and glass microreactors have the advantage of easy optical access, providing a mean for a wide range of in-situ characterization techniques. Figure 4-f depicts a microreactor where the size and shape of organic crystals can be deter‐ mined by optical microscopy and the crystal phase determined by in-situ Raman spectroscopy. An advance microfluidic reactor for polymer particles production was designed using a sequence of microvalves which were digitally activated to control the composition of the polymer particles in a matter of seconds [29].

**Figure 4.** Examples of metal microreactors for nanomaterials synthesis. (a)Multilamination micromixer to produce pol‐ ymer nanoparticles ref. [20].(reprinted with permission from ref. [21], copyright 2012 Royal Society of Chemistry). b) Multi-step multilamination micromixer to produce Au-SiO2 nanoshells. (reprinted with permission from ref. [21], copyright 2012 Royal Society of Chemistry), c) Spiral microreactor to produce Au NPs.(reprinted with permission from ref. [23], copyright 2012 American Chemical Society). d) Two-stage micromixer to produce Au/Ag NPs.(reprinted with permission from ref. [25], copyright 2007 J. Michael Köhler et al). e)Three-stage microfluidic system to produce quan‐ tum dots.(reprinted with permission from ref. [27], copyright 2011 WILEY-VCH Verlag GmbH & Co). f) Microfludic system with optical access for optical microcopy and Raman spectra coupling.(reprinted with permission from ref. [28], copyright 2012 American Chemical Society).

#### **2.2. Synthesis strategies in microfluidics**

and improving the size and shape control [24]. Figure 4-d depicts a silicon-based microreac‐ tor where there are integrated two mixing units to control the growth and composition of Au/Ag hetero-structures. Glass microcapillary reactors can be fabricated with a higher resolution at micrometer scale than the polymer reactors ones, this fact enables to produce a better control

Finally, this section should highlight the existence of some other types of microreactors with a considered relevance in terms of functionality. Figure 4-e illustrates a continuous three-stage silicon-based microsystem consisting of mixing, aging and sequential growth stages, operating at 65 bar and temperatures as high as 340ºC. This system is an example of complex fluiddynamic control by the injection of different precursor flow-rates and pressure drop along the different microchannels. On the other hand, silicon and glass microreactors have the advantage of easy optical access, providing a mean for a wide range of in-situ characterization techniques. Figure 4-f depicts a microreactor where the size and shape of organic crystals can be deter‐ mined by optical microscopy and the crystal phase determined by in-situ Raman spectroscopy. An advance microfluidic reactor for polymer particles production was designed using a sequence of microvalves which were digitally activated to control the composition of the

e)

f)

**Figure 4.** Examples of metal microreactors for nanomaterials synthesis. (a)Multilamination micromixer to produce pol‐ ymer nanoparticles ref. [20].(reprinted with permission from ref. [21], copyright 2012 Royal Society of Chemistry). b) Multi-step multilamination micromixer to produce Au-SiO2 nanoshells. (reprinted with permission from ref. [21], copyright 2012 Royal Society of Chemistry), c) Spiral microreactor to produce Au NPs.(reprinted with permission from ref. [23], copyright 2012 American Chemical Society). d) Two-stage micromixer to produce Au/Ag NPs.(reprinted with permission from ref. [25], copyright 2007 J. Michael Köhler et al). e)Three-stage microfluidic system to produce quan‐ tum dots.(reprinted with permission from ref. [27], copyright 2011 WILEY-VCH Verlag GmbH & Co). f) Microfludic system with optical access for optical microcopy and Raman spectra coupling.(reprinted with permission from ref.

in the production of emulsions and polymeric nanoparticles (Figure 3-j).

394 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

polymer particles in a matter of seconds [29].

[28], copyright 2012 American Chemical Society).

c)

d)

a)

b)

Regarding synthesis strategies, microfluidic reactors can be mainly grouped into two catego‐ ries related with their flow pattern: single-phase and multi-phase flow. Continuous flow (or single-phase flow) microreactors allow facile change of experimental conditions within microseconds and usually improve the homogeneity of solution during the synthesis process in comparison with batch reactors. The operability and stability of this type of flow pattern is quite good. However, liquid flow within microchannels is strictly a laminar flow due to the small Reynolds numbers, then, the mixing is achieve predominately by molecular diffusion since the absence of turbulence. The small cross section in combination with the no-slip boundary condition at the channel walls lead to a parabolic velocity profile. This velocity profile depends on the position within the cross section and can promote a wide residence time distribution depending on the flow conditions. This fact can be a serious problem in the production of those nanomaterials whose crystallization kinetics is sensitive to the residence time distribution in the early stages of growth. If this occurs, growing nanoparticles flowing near the walls will spend significantly a long time inside the reactor, as compared to those flowing near the center (Figure 5). This performance implies that the ones flowing near the walls will be bigger in size than the ones flowing by the center. To overcome nanoparticles polydispersity, continuous flow microreactors can include a micromixing stage, where a good mixing is achieveing by reducing the diffusion lengths. Micromixers can be classified in passive and active according to the existence of mechanical agitation or external forces. Passive mixing simply creates lamination of multiple fluid streams, increasing the interfacial area for diffusion (Figure 5). A passive micromixer is illustrated in Figure 4 b, where the multi-lamination is achieved by splitting the streams and then recombining them again. Although micromixing by multi-lamination is effective, the narrow microchannels can lead to high pressure drops. This fact is even more important in high viscous fluids, which limit the operability in micro‐ systems. Active mixing is directed by applying external forces such as ultrasonic waves, electric and magnetic fields. However, the fabrication of these microsystems require special fabrication techniques [30].

Multi-phase flow facilitates the passive mixing and reduces the propensity to foul after extended operation. Multiphase flow can be also classified into gas-liquid and liquid-liquid flow regimes. In gas-liquid flow systems, it is distinguished the slug and the annular flow patterns depending on the gas and liquids superficial velocities. Annular flow occurs when a continuous gas core flow in the channel center and a liquid film at the walls. On the other hand, in slug flow, gas bubbles are located between two liquid segments (Figure 5). Viscous drag at the channel wall induces convective mixing (recirculation) within each slug. As a result, it ensures an excellent chemical homogeneity. In addition, the slug move at the same speed, this implies that the residence time distribution is narrower than with continuous flow operations. Consequently, slug or segmented flow has the potential to improve size distributions in nanocrystallization process. Then, an inert gas can be introduced in a liquid system to improve mixing and residence time characteristics.

In liquid-liquid segmented flow, the surface tension differences between immiscible streams enables to achieve a steady fluid segmentation (Figure 5). Considering the surface wettability

**Figure 5.** Comparison of the synthesis strategies in continuous microflows to produce nanomaterials: Continuous flow (with and without multilamination) and segmented flow (gas-liquid and liquid-liquid).

of the microchannel, the reactor clogging occurs usually with a high frequency in the gas-liquid slug flow since the liquid segments are flowing in contact with the microchannel wall. However, the liquid entities where the chemical reaction occurs are easier to separate in gasliquid segmented flow. On the other hand the single-phase systems enable the subsequent addition of reagents in order to operate in a multi-stage system. A sequential addition of reagents in a multi-phase system is challenging because the formation of multiple liquid entities without connectivity. This issued can be addressed by the use of a merge drop flow, but, at the end, the segments merging could affect negatively to the residence time distribution.

#### **2.3. Microfluidic synthesis of inorganic nanomaterials**

Inorganic nanoparticles have been widely used in diverse areas such as electronics, energy, textiles, biotechnology and medicine, bio-imaging and bio-sensing. There have been tremen‐ dous interests in the development of microfluidic methods for producing inorganic nanopar‐ ticles because microfluidics provide an excellent control on the nanocrystallization process. There are numerous examples of synthesis of inorganic nanomaterials in microfluidic devices [31, 32], but some of them will be discussed in the present section due their relevance and novelty.

Gold nanoparticles production has been extensively studied due to their wide variety of properties and the challenge in controlling the shape and size. Typical microfluidics recipes involve the reduction of a gold precursor in the presence of different types of ligands and stabilizers. The presence of a strong reducing agent, such as sodium borohydride, promotes fast nucleation and the production of small nanoparticles. The reduction of Au ions occurs in a time scale of seconds, following a fast kinetic nanocrystallization. However, fast kinetic reactions should be carried out if the mixing time is shorter than the reaction time, otherwise a wide distribution of nanoparticles can be achieved. As it was considered in section 2.2, flow segmentation promotes a rapid mixing and an efficient heat/mass transfer by internal recir‐ culation. But, it is required to operate with the proper segmented flow pattern and conditions in order to achieve the expected results. Consequently, the particle size distribution in fast kinetic reactions, such as Au nanoparticles production, is strongly dominated by the mixing conditions induced by internal recirculation and not only by the type of flow [33]. Figure 6 ad illustrates the effect of the residence time on the Au nanoparticle size distribution. The length of the slugs is influenced by the residence time, obtaining small slugs at smaller residence times (Figure 6-c). The slug recirculation is higher in small slug than in the long ones, promoting a fast mass transfer. So the control of the slug length is a strict requirement to decrease the particle polydispersity (Figure 6-d). On the other hand, liquid-liquid and gas-liquid improve the micromixing but the grade of mixing achieved is not comparable, even if the residence time is the same. Figure 6 e-g shows that the particle size distribution of nanoparticles produced at the same residence time is sensitive to the fluid of segmentation. Air segmentation provides a better internal mixing than silicone oil (Figure 6-h) [33]. Silicone oil presents a higher viscosity than air, reducing both the slip velocity and the circulation rates.

Microfludic reactors are able to produce complex hybrid nanostructures in a simple process that reduces the synthesis time and provide an additional level of reaction control that it is not attainable in bulk stirred reactors. Silica gold nanourchin particles with hyper-branched Au nanowires were produced in batch reactors after the sequential addition of reagents (at least 10 times) (Figure 6-m). Gold was anisotropically grown on the surface of silica nanoparticles decorated with Au seeds by maintaining a good control on the kinetics in a batch reactor (Figure 6-j). After several cycles of reagents addition, the length of the nanowires was long enough to use these hybrid nanostructures in SERS. However the exquisite control required during the sequential addition of reagents unable a reproducible production of these interest‐ ing nanostructures. The continuous production of gold nanourchin structures was achieved successfully in a PTFE microreactor using a single addition of reagents (Figure 6-i). The microreactor showed perfect performance for manipulating the reagent concentrations and promote a kinetic control of the reaction during the anisotropically growth of Au. Finally, those nanostructures showed a remarkable SERS performance, which evidences the quality of the produced nanostructures.

of the microchannel, the reactor clogging occurs usually with a high frequency in the gas-liquid slug flow since the liquid segments are flowing in contact with the microchannel wall. However, the liquid entities where the chemical reaction occurs are easier to separate in gasliquid segmented flow. On the other hand the single-phase systems enable the subsequent addition of reagents in order to operate in a multi-stage system. A sequential addition of reagents in a multi-phase system is challenging because the formation of multiple liquid entities without connectivity. This issued can be addressed by the use of a merge drop flow, but, at the end, the segments merging could affect negatively to the residence time distribution.

**Figure 5.** Comparison of the synthesis strategies in continuous microflows to produce nanomaterials: Continuous flow

Inorganic nanoparticles have been widely used in diverse areas such as electronics, energy, textiles, biotechnology and medicine, bio-imaging and bio-sensing. There have been tremen‐ dous interests in the development of microfluidic methods for producing inorganic nanopar‐ ticles because microfluidics provide an excellent control on the nanocrystallization process. There are numerous examples of synthesis of inorganic nanomaterials in microfluidic devices [31, 32], but some of them will be discussed in the present section due their relevance and

Gold nanoparticles production has been extensively studied due to their wide variety of properties and the challenge in controlling the shape and size. Typical microfluidics recipes

**2.3. Microfluidic synthesis of inorganic nanomaterials**

(with and without multilamination) and segmented flow (gas-liquid and liquid-liquid).

396 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

novelty.

Gold nanorods have been widely used in biomedical application due to their interesting optical properties. The growth comprises a two-stage process: 1) Gold seeds formation and 2) Gold addition by the secondary growth of gold seeds. The optical properties of the resulted nanorods are sensitive to synthesis variables such as seed size, temperature, mixing, reagents concentration, reducing rate and so. In addition, the synthesis of Au nanorods is usually achieved in the presence of cetyltrimethyl ammonium bromide (CTABr), a high cytotoxic ligand which requires several purification treatments after synthesis. Although there are several microfluidic approaches to produce Au nanorods and circumvent the limited repro‐ ducibility of Au nanorods production, they are based in the use of seeds and CTAB [36]. A new microfluidic approach enables the continuous production of Au nanorods in a single step (without seeds) and using a biocompatible ligand (Lysine) instead of CTAB [35]. Optical properties of the produced Au nanorods are tunable with the Lysine/Au precursor ration, obtaining nanorods with an aspect ratio of 5 and a Surface Plasmos Resonance Peak centered in 780nm. Moreover, microfluidics facilitate the manipulation of separate reagent streams, which enabled the fast screening and optimization of the synthesis conditions on demand. The Au nanorods produced by this approach were successfully applied as contrast agents in Photothermal Optical Coherence Tomography (OCT) of human breast tissue. The simplicity of this approach will made feasible that unskilled staff could in future produce this type of contrast agent on demand and out from the laboratory.

Magnetic nanoparticles (MNPs) are of significant interest due to their unique properties in terms of chemical stability, size-dependent magnetic response, biocompatibility, and low price, which make them ideal candidates in a wide-range of applications in biomedicinerelated fields: drug delivery, hyperthermia, magnetic resonance imaging (MRI), tissue engineering and repair, biosensing, and biochemical separations [37]. High quality crystalline MNPs were obtained with a narrow size distribution of mean diameter 3.6 nm and standard deviation 0.8 nm in a passively-driven capillary-based droplet reactor [38]. Reagents were fully compartmentalised within the droplets, preventing the fouling of the channel walls. In contrast, when the reaction was allowed to proceed in a single-phase mode of operation, a dark brown deposit formed on the inner surface of the main channel within minutes. Then, as it was before mentioned, segmented flow is a useful tool to prevent microfluidic reactor from clogging and promote a narrow size distribution of nanoparticles during the growth process. However the flow segmentation can provide a new function, controlling the crystal phase. That is, by controlling the atmosphere in the gas segments, the crystalline phase and size of the resulting nanoparticles can be accurately tuned [37]. Pure magnetite nanocrystals with their characteristic octahedral shape were obtained when N2 and H2 were used to segmentate the liquid stream of reagents (Figure 7-a). In fact, the reducing properties of H2 accelerates the production of high crystalline magnetite NPs in 1 minutes residence time at 100ºC.; whereas flow segmentation with N2 required 6 minutes. If O2 was used instead of N2 or H2, oxidant conditions direct the synthesis to the production of an orange colloid (Figure 7-a). The nanoparticles formed were nanoflakes with a mean size (along the longest axis) of about 30 ± 8 and 3 nm thickness which was identified as feroxyhyte (δ-FeOOH). Finally, carbon monoxide is generally considered as a poisoning agent in many catalytic studies but is also considered as a reducing agent or even a capping agent to direct the shape control in nanoparticle synthesis. When a CO-liquid segmented flow was created, it was observed that feroxyhyte was obtained at temperatures lower than 80ºC (see Figure 7-a). But these nanostructures were bigger in size than the ones produced with O2, (70 ± 12 nm and thickness of 3 nm). Then, at low temperatures CO does not act as a reducing agent, as H2 does, and instead directs the oxidation of iron hydroxides to feroxyhyte. With a further increase of temperature to 100 °C, crystalline and pure magnetite nanoparticles were obtained at 1 min of residence time, High and Efficient Production of Nanomaterials by Microfluidic Reactor Approaches http://dx.doi.org/10.5772/64347 399

ligand which requires several purification treatments after synthesis. Although there are several microfluidic approaches to produce Au nanorods and circumvent the limited repro‐ ducibility of Au nanorods production, they are based in the use of seeds and CTAB [36]. A new microfluidic approach enables the continuous production of Au nanorods in a single step (without seeds) and using a biocompatible ligand (Lysine) instead of CTAB [35]. Optical properties of the produced Au nanorods are tunable with the Lysine/Au precursor ration, obtaining nanorods with an aspect ratio of 5 and a Surface Plasmos Resonance Peak centered in 780nm. Moreover, microfluidics facilitate the manipulation of separate reagent streams, which enabled the fast screening and optimization of the synthesis conditions on demand. The Au nanorods produced by this approach were successfully applied as contrast agents in Photothermal Optical Coherence Tomography (OCT) of human breast tissue. The simplicity of this approach will made feasible that unskilled staff could in future produce this type of

Magnetic nanoparticles (MNPs) are of significant interest due to their unique properties in terms of chemical stability, size-dependent magnetic response, biocompatibility, and low price, which make them ideal candidates in a wide-range of applications in biomedicinerelated fields: drug delivery, hyperthermia, magnetic resonance imaging (MRI), tissue engineering and repair, biosensing, and biochemical separations [37]. High quality crystalline MNPs were obtained with a narrow size distribution of mean diameter 3.6 nm and standard deviation 0.8 nm in a passively-driven capillary-based droplet reactor [38]. Reagents were fully compartmentalised within the droplets, preventing the fouling of the channel walls. In contrast, when the reaction was allowed to proceed in a single-phase mode of operation, a dark brown deposit formed on the inner surface of the main channel within minutes. Then, as it was before mentioned, segmented flow is a useful tool to prevent microfluidic reactor from clogging and promote a narrow size distribution of nanoparticles during the growth process. However the flow segmentation can provide a new function, controlling the crystal phase. That is, by controlling the atmosphere in the gas segments, the crystalline phase and size of the resulting nanoparticles can be accurately tuned [37]. Pure magnetite nanocrystals with their characteristic octahedral shape were obtained when N2 and H2 were used to segmentate the liquid stream of reagents (Figure 7-a). In fact, the reducing properties of H2 accelerates the production of high crystalline magnetite NPs in 1 minutes residence time at 100ºC.; whereas flow segmentation with N2 required 6 minutes. If O2 was used instead of N2 or H2, oxidant conditions direct the synthesis to the production of an orange colloid (Figure 7-a). The nanoparticles formed were nanoflakes with a mean size (along the longest axis) of about 30 ± 8 and 3 nm thickness which was identified as feroxyhyte (δ-FeOOH). Finally, carbon monoxide is generally considered as a poisoning agent in many catalytic studies but is also considered as a reducing agent or even a capping agent to direct the shape control in nanoparticle synthesis. When a CO-liquid segmented flow was created, it was observed that feroxyhyte was obtained at temperatures lower than 80ºC (see Figure 7-a). But these nanostructures were bigger in size than the ones produced with O2, (70 ± 12 nm and thickness of 3 nm). Then, at low temperatures CO does not act as a reducing agent, as H2 does, and instead directs the oxidation of iron hydroxides to feroxyhyte. With a further increase of temperature to 100 °C, crystalline and pure magnetite nanoparticles were obtained at 1 min of residence time,

contrast agent on demand and out from the laboratory.

398 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

**Figure 6.** Gold nanoparticles produced in microfluidic reactors. TEM image of gold NPs synthesized in toluene−aque‐ ous segmented flow: a) Residence time (Rt) = 10 s, b) Rt = 40 s, c) segmented slugs generated at Rt = 10 and 40 s d) Particle size distribution diagram from AuNPs obtained in toluene − aqueous segmented flow at Rt = 40, 20, and 10 s. TEM image of gold NPs synthesized at Rt = 10 s in a : (e) silicone oil−aqueous segmented flow, (f) toluene−aqueous segmented flow, (g) air − aqueous segmented flow. (h) Particle size distribution diagram from AuNPs obtained in sili‐ cone oil/toluene/air−aqueous segmented flow at Rt = 10 s.(reprinted with permission from ref. [33], copyright 2012 American Chemical Society).i) Microfludic system to produce Au nanourchin particles :j) Scheme for the synthesis of silica–gold nanourchins formation with representative images from: k) silica NP, (l) silica decorated with Au NPs (m) Au nanourchin (reprinted with permission from ref. [34], copyright 2014 Royal Society of Chemistry). n) Absorption spectra and TEM images of the gold nanorods produced with microfluidic reactors. (reprinted with permission from ref. [35], copyright 2012 Royal Society of Chemistry).

confirming the reduction potential of CO at these conditions. Consequently the gas slug microfluidics provide a flexible, easy option to implement a process to produce customized iron oxide nanostructures.

Laboratory syntheses tend to be complex, based in multistep batch processes with yields that are typically well below the gram range. These batch processes often suffer from irreproduci‐ bility of the morphology, size, size distribution and quality. Then, implementing the continu‐ ous, or at least automated, production of nanomaterials is clearly desirable in terms of improving the product homogeneity and facilitating scale-up production. The continuous production of nanomaterials is challenging because of the difficulties involved in translating the complexity of nanomaterial synthesis into on-line operations. The production of hollow gold nanoparticles constitute a clear example of the strict requirements in the dimensional control, since their optical properties are sensitive to the nanoparticle dimensions and shell thickness. The synthesis of hollow Au nanoparticles comprises a multi-stage process: 1) Synthesis of a Co NP as a template, 2) Galvanic replacement of Co by Au+3 ions. The conven‐ tional lab scale production of hollow gold NPs deals with a reaction volume of 120 mL [39]. But if the production scale was increased 10 folders, the quality and consequently the optical properties, are not reproducible. The poor quality of the NPs in the scaled-up batch syntheses was probably caused by insufficient mixing as the reaction volume increased. This problem is even more serious if the nano-crystallization reaction is directed by a fast kinetic. A micro‐ fluidic system was proposed to overcome the production limitations of hollow gold nanopar‐ ticles, but incorporating two additional stages to yield nanostructures suitable for biomedical applications: 1) PEG functionalization to improve the biocompatibility and 2) Sterilization (see Figure 7-b). Considering the 4 implemented stages, high quality nanoparticles were produced in only 22 minutes, being the sterilization process the slowest stage (15 min). The galvanic replacement reaction is a fast process, and then it is required to get an excellent mixing of reagents to promote an even distribution of Au+3 ions over the Co templates. As can be observed in Figure 7-b, the galvanic replacement of Co by Au+3 ions was properly directed since there is no trace of Co in the NP volume, releasing the hollow structure. Consequently, the appli‐ cation of microfluidic systems on multi-stage reactions is highly convenient to preserve the quality of materials and reduce irreproducibility.

A continuous three-stage microfluidic system that separates the mixing, aging, and subsequent injection stages of InP nanocrystal synthesis was designed to optimize the synthesis conditions (Figure 4-d). The microfluidic system operates at high temperature and high pressure enabling the use of solvents such as octane operating in the supercritical regime for high diffusivity resulting in the production of high-quality InP nanocrystals in as little as 2 minutes (Figure 7 c). To control the size of InP nanocrystals, six sequential injections of chemical precursors were injected. This procedure ease the production protocol of these nanocrystals, since the sequen‐ tial addition of reagents in a batch reactor requires at high temperature of an exquisite control and accuracy.

#### **2.4. Microfluidic synthesis of polymeric nanomaterials**

Microfluidic devices have emerged as promising tools for the synthesis of polymer particles. Over conventional processes, microfluidic-assisted processes allow the production of polymer particles with an improved control over their sizes, size distributions, morphologies, and compositions. This is particularly important since the synthesis of polymeric nanoparticles by High and Efficient Production of Nanomaterials by Microfluidic Reactor Approaches http://dx.doi.org/10.5772/64347 401

Laboratory syntheses tend to be complex, based in multistep batch processes with yields that are typically well below the gram range. These batch processes often suffer from irreproduci‐ bility of the morphology, size, size distribution and quality. Then, implementing the continu‐ ous, or at least automated, production of nanomaterials is clearly desirable in terms of improving the product homogeneity and facilitating scale-up production. The continuous production of nanomaterials is challenging because of the difficulties involved in translating the complexity of nanomaterial synthesis into on-line operations. The production of hollow gold nanoparticles constitute a clear example of the strict requirements in the dimensional control, since their optical properties are sensitive to the nanoparticle dimensions and shell thickness. The synthesis of hollow Au nanoparticles comprises a multi-stage process: 1) Synthesis of a Co NP as a template, 2) Galvanic replacement of Co by Au+3 ions. The conven‐ tional lab scale production of hollow gold NPs deals with a reaction volume of 120 mL [39]. But if the production scale was increased 10 folders, the quality and consequently the optical properties, are not reproducible. The poor quality of the NPs in the scaled-up batch syntheses was probably caused by insufficient mixing as the reaction volume increased. This problem is even more serious if the nano-crystallization reaction is directed by a fast kinetic. A micro‐ fluidic system was proposed to overcome the production limitations of hollow gold nanopar‐ ticles, but incorporating two additional stages to yield nanostructures suitable for biomedical applications: 1) PEG functionalization to improve the biocompatibility and 2) Sterilization (see Figure 7-b). Considering the 4 implemented stages, high quality nanoparticles were produced in only 22 minutes, being the sterilization process the slowest stage (15 min). The galvanic replacement reaction is a fast process, and then it is required to get an excellent mixing of reagents to promote an even distribution of Au+3 ions over the Co templates. As can be observed in Figure 7-b, the galvanic replacement of Co by Au+3 ions was properly directed since there is no trace of Co in the NP volume, releasing the hollow structure. Consequently, the appli‐ cation of microfluidic systems on multi-stage reactions is highly convenient to preserve the

400 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

A continuous three-stage microfluidic system that separates the mixing, aging, and subsequent injection stages of InP nanocrystal synthesis was designed to optimize the synthesis conditions (Figure 4-d). The microfluidic system operates at high temperature and high pressure enabling the use of solvents such as octane operating in the supercritical regime for high diffusivity resulting in the production of high-quality InP nanocrystals in as little as 2 minutes (Figure 7 c). To control the size of InP nanocrystals, six sequential injections of chemical precursors were injected. This procedure ease the production protocol of these nanocrystals, since the sequen‐ tial addition of reagents in a batch reactor requires at high temperature of an exquisite control

Microfluidic devices have emerged as promising tools for the synthesis of polymer particles. Over conventional processes, microfluidic-assisted processes allow the production of polymer particles with an improved control over their sizes, size distributions, morphologies, and compositions. This is particularly important since the synthesis of polymeric nanoparticles by

quality of materials and reduce irreproducibility.

**2.4. Microfluidic synthesis of polymeric nanomaterials**

and accuracy.

**Figure 7.** Synthesis of inorganic nanomaterials in microfluidic systems: a) Gas selective crystallization of iron oxides nanoscrystals in segmented flow reactors(reprinted with permission from ref. [37], copyright 2015 American Chemical Society). b) Multi-stage production of hollow gold nanoparticles (reprinted with permission from ref. [39], copyright 2013 Royal Society of Chemistry). C) Production of InP nanocrystals in pressure and temperature condition (reprinted with permission from ref. [27], copyright 2010 WILEY-VCH Verlag GmbH & Co).

bulk methods typically lacks control over the mixing processes, which may compromise the properties of the resulting nanoparticles. Nanoparticles dimensions can be tuned by varying microfludic parameters such as flow rates, precursor composition, and mixing time. The available techniques to produce polymer nanoparticles can be classified depending on the starting state of the polymers: From dispersion of preformed polymers and from polymeriza‐ tion of monomers. The most common techniques which use a preformed polymer can be stated as: solvent evaporation method, nanoprecipitation, emulsification, salting out, dialysis and supercritical fluids. On the other hand, the current techniques available for monomer poly‐ merization can be divided into: emulsion, mini emulsion, micro emulsion, interfacial poly‐ merization and radical polymerization.

This section will deal some of the most interesting examples of polymer nanoparticles production by the most relevant techniques for the production of polymer particles within microfluidic devices: (a) the direct polymerization through continuous flow projection photolithography [40], (b) nanoprecipitation and (c) the emulsification of liquid monomers, followed by the polymerization of the subsequent monomer droplets.

Polymerization through continuous flow projection photolithography consists in the UV irradiation, through the objective of an optical microscope, of a polymer solution flowing within a microchannel (See Figure 8-a). A mask placed in the field-stop plane of the microscope allows polymerizing and the desired particle shape to the flowing monomer solution. Then, the polymerized microparticles are formed when flow was stopped in the microfluidic device. Then, particles were pushed out of the polymerization area with flow of the monomer solutions. This approach is very powerful because it allows for nearly unlimited designs, including the production of barcodes and multifunctional particles [41].

Two-dimensional hydrodynamic focusing was commonly used to produce polymer nanopar‐ ticles (ca. PLGA) by nanoprecipitation. Three-dimensional flow-focusing prevent the formation of aggregates and microreactor fouling. While these approaches have provided significant advantages, there still remain challenges. First, these microfluidic approaches using slow diffusive mixing at a low flow rate, limiting the productivity. Second, diffusive mixing does not allow the development of particles that require the assembly of precursors dispersed in different phases, (e.g., lipid) in the aqueous phase with precursors (e.g., polymer) in the organic phase such as lipid−polymer hybrid nanoparticles [42]. To overcome those limitations and develop a high-throughput and reproducible nanoparticle synthesis technol‐ ogy, new advances have been performed in improving the mixing. The application of a microvortex considerable improves the mixing and enable a controlled nanoprecipitation [42]. The use of microvortices results in a rapid mixing between phases, enabling up to 1000 times higher productivity (3 g /hour) and better size control (30-170nm) than previous methods (See Figure 8-b).

A simple and versatile coaxial turbulent jet mixer consisting of coaxial cylindrical tubes where NP precursors and non-solvent are injected through the inner and outer tubes, respective‐ ly, was fabricated to produced PLGA-PEG, Lipid vesicles and polystyrene, obtaining production rates up to 3 kg/d (see Figure 8-c) [43]. This versatile mixer provides an inherent‐ ly high NP production throughput due to operation in the turbulent regime (Re 500-3500) with device dimensions in the millimeter scale. The production rates achieved are suitable for in vivo studies, clinical trials, and industrial-scale production, while retaining the advantages of homogeneity, reproducibility, and control over NP properties (size and morphology). A strong point in using this type of microfluidic devices is the potential use in biomedical research laboratories, where it is not required specialized microfabrication facilities since this is a robust and simple device, as well as a microfluidic know-how to operate in the optimum conditions.

merization can be divided into: emulsion, mini emulsion, micro emulsion, interfacial poly‐

This section will deal some of the most interesting examples of polymer nanoparticles production by the most relevant techniques for the production of polymer particles within microfluidic devices: (a) the direct polymerization through continuous flow projection photolithography [40], (b) nanoprecipitation and (c) the emulsification of liquid monomers,

Polymerization through continuous flow projection photolithography consists in the UV irradiation, through the objective of an optical microscope, of a polymer solution flowing within a microchannel (See Figure 8-a). A mask placed in the field-stop plane of the microscope allows polymerizing and the desired particle shape to the flowing monomer solution. Then, the polymerized microparticles are formed when flow was stopped in the microfluidic device. Then, particles were pushed out of the polymerization area with flow of the monomer solutions. This approach is very powerful because it allows for nearly unlimited designs,

Two-dimensional hydrodynamic focusing was commonly used to produce polymer nanopar‐ ticles (ca. PLGA) by nanoprecipitation. Three-dimensional flow-focusing prevent the formation of aggregates and microreactor fouling. While these approaches have provided significant advantages, there still remain challenges. First, these microfluidic approaches using slow diffusive mixing at a low flow rate, limiting the productivity. Second, diffusive mixing does not allow the development of particles that require the assembly of precursors dispersed in different phases, (e.g., lipid) in the aqueous phase with precursors (e.g., polymer) in the organic phase such as lipid−polymer hybrid nanoparticles [42]. To overcome those limitations and develop a high-throughput and reproducible nanoparticle synthesis technol‐ ogy, new advances have been performed in improving the mixing. The application of a microvortex considerable improves the mixing and enable a controlled nanoprecipitation [42]. The use of microvortices results in a rapid mixing between phases, enabling up to 1000 times higher productivity (3 g /hour) and better size control (30-170nm) than previous

A simple and versatile coaxial turbulent jet mixer consisting of coaxial cylindrical tubes where NP precursors and non-solvent are injected through the inner and outer tubes, respective‐ ly, was fabricated to produced PLGA-PEG, Lipid vesicles and polystyrene, obtaining production rates up to 3 kg/d (see Figure 8-c) [43]. This versatile mixer provides an inherent‐ ly high NP production throughput due to operation in the turbulent regime (Re 500-3500) with device dimensions in the millimeter scale. The production rates achieved are suitable for in vivo studies, clinical trials, and industrial-scale production, while retaining the advantages of homogeneity, reproducibility, and control over NP properties (size and morphology). A strong point in using this type of microfluidic devices is the potential use in biomedical research laboratories, where it is not required specialized microfabrication facilities since this is a robust and simple device, as well as a microfluidic know-how to

followed by the polymerization of the subsequent monomer droplets.

402 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

including the production of barcodes and multifunctional particles [41].

merization and radical polymerization.

methods (See Figure 8-b).

operate in the optimum conditions.

**Figure 8.** Synthesis of polymeric nanomaterials in microfluidic systems: a) Direct polymerization through continuous flow projection photolithography. Schematic diagram of dot-coded particle synthesis showing polymerization across two adjacent laminar (reprinted with permission from ref. [46], copyright 2012 American Chemical Society). b) Sche‐ matic and cross section views of a three-inlet microfluidic platform generating lipid−polymer hybrid nanoparticles through controlled microvortices by nanoprecipitation. Scale bars indicate 100 nm (reprinted with permission from ref. [42], copyright 2012 American Chemical Society). c) Schematic illustration of the coaxial turbulent jet mixer for highthroughput synthesis of NPs with turbulence induced by jetting. Photograph of coaxial turbulent jet mixer fabricated from PTFE tee union tube fittings and schematic illustrations and top views of fluid flow at laminar,vortex and turbu‐ lence and turbulent jet regimes (reprinted with permission from ref. [43], copyright 2014 American Chemical Soci‐ ety).d) Schematics of a microfluidic origami chip for synthesizing monodisperse DOX-loaded PLGA nanoparticles. Different 3D geometries (arc and double spiral) obtained by manually folding the origami chip. (reprinted with per‐ mission from ref. [44], copyright 2011 WILEY-VCH Verlag GmbH & Co).

A microfuidic origami chip with different geometries that enables the production of PLGA nanoparticles loaded with doxorubicin (DOX) by rapid mixing was designed to tune the nanoparticle size from 70 to 230 nm in a single nanoprecipitation step. The origami chip can be set in a two-dimensional flat geometry or can be folded in a three-dimensional geometry such as an arc or a double spiral. The combination of hydrodynamic focusing and 3D curved microchannels can signicantly shorten the mixing distance and reduce the mixing time, which can reduce the mean nanoparticle size and size distribution (see Figure 8-d).The mixing time in the 3D double spiral configuration was approximately 46% faster the 2D flat configuration. This improvement was due the formation two counter-rotating vortices perpendicular to the flow direction that could efficiently increase the mixing after curved turns [44].

Multiple emulsion drops are useful in the production of complex microcapsules for encapsu‐ lation and sequential release of multi-component active materials while avoiding crosscontamination. Most approaches utilize sequential emulsification using a series of single drop microfluidics units, but the device fabrication requires complex procedures and a good control to synchronize the frequencies of drop generations in all drop units [45]. Figure 9-a shows a flow focusing microdevice to produce monodisperse multiple emulsion drops of high order using stable biphasic flows in confining channels. Four immiscible fluids are simultaneously introduced into the orifice in the form of a coaxial flow. This results in the formation of triple emulsion drops with the simplicity of a single emulsification process. The interesting novelty of this system is that the breakup of the interfaces is achieved in dripping or jetting modes, determined by the flow rates. The jetting breakup mode promotes the injection of different number of drops in the core of a multiple emulsions (see Figure 9-a). On the other hand, the dripping mode facilitates the production of monodisperse triple or quadruple emulsion drops with an onion like configuration (Figure 9-b).

Loading inorganic nanomaterials in polymer particles entails the formation of multifunctional entities. Batch synthesis has been the method of choice for preparing inorganic-doped microparticles due to practical advantages in productivity and accessibility, despite limitations such as poor loading, low throughput, and material loss. In addition, a significant portion of loading materials is wasted during the multistep process of synthesis, purification, loading, and washing. These shortcomings are especially problematic when expensive inorganic nanomaterials are used [16]. Figure 3-g illustrates a microfluidic reactor designed to produce uniform emulsions loaded with quantum dots using high throughput harmonic breakup of a jet in a high shear coflowing system with a well-designed nozzle. Uniform fluorescent microbeads containing about 10 wt % CdSe/ZnSe QD nanocrystals were produced with a tuned size ranged from 13µm to 100 µm through in-line photopolymerization of biocompatible prepolymer resin. Unlike the batch synthesis, there are no additional washing steps required for the removal of unreacted residues causing dissipation of materials. Then, it is clear the advantages of microfluidics in producing complex nanostructures.

## **3. Conclusions and outlook**

The characteristics of micro- and nanoparticle formulations produced by continuous flow microfluidic systems have remarkable advantages over traditional bulk methods that motivate their adoption. The exquisite control of flow and mixing conditions in microfluidics led to improved homogeneity of particle size distributions and the control of particle size in a reproducible and continuous fashion. Microfluidic systems often integrate real time analysis of the generated particles in order to tune the properties of the produced materials according to specifications. From the reported examples in this chapter, there is no doubt that microfluidic

in the 3D double spiral configuration was approximately 46% faster the 2D flat configuration. This improvement was due the formation two counter-rotating vortices perpendicular to the

Multiple emulsion drops are useful in the production of complex microcapsules for encapsu‐ lation and sequential release of multi-component active materials while avoiding crosscontamination. Most approaches utilize sequential emulsification using a series of single drop microfluidics units, but the device fabrication requires complex procedures and a good control to synchronize the frequencies of drop generations in all drop units [45]. Figure 9-a shows a flow focusing microdevice to produce monodisperse multiple emulsion drops of high order using stable biphasic flows in confining channels. Four immiscible fluids are simultaneously introduced into the orifice in the form of a coaxial flow. This results in the formation of triple emulsion drops with the simplicity of a single emulsification process. The interesting novelty of this system is that the breakup of the interfaces is achieved in dripping or jetting modes, determined by the flow rates. The jetting breakup mode promotes the injection of different number of drops in the core of a multiple emulsions (see Figure 9-a). On the other hand, the dripping mode facilitates the production of monodisperse triple or quadruple emulsion drops

Loading inorganic nanomaterials in polymer particles entails the formation of multifunctional entities. Batch synthesis has been the method of choice for preparing inorganic-doped microparticles due to practical advantages in productivity and accessibility, despite limitations such as poor loading, low throughput, and material loss. In addition, a significant portion of loading materials is wasted during the multistep process of synthesis, purification, loading, and washing. These shortcomings are especially problematic when expensive inorganic nanomaterials are used [16]. Figure 3-g illustrates a microfluidic reactor designed to produce uniform emulsions loaded with quantum dots using high throughput harmonic breakup of a jet in a high shear coflowing system with a well-designed nozzle. Uniform fluorescent microbeads containing about 10 wt % CdSe/ZnSe QD nanocrystals were produced with a tuned size ranged from 13µm to 100 µm through in-line photopolymerization of biocompatible prepolymer resin. Unlike the batch synthesis, there are no additional washing steps required for the removal of unreacted residues causing dissipation of materials. Then, it is clear the

The characteristics of micro- and nanoparticle formulations produced by continuous flow microfluidic systems have remarkable advantages over traditional bulk methods that motivate their adoption. The exquisite control of flow and mixing conditions in microfluidics led to improved homogeneity of particle size distributions and the control of particle size in a reproducible and continuous fashion. Microfluidic systems often integrate real time analysis of the generated particles in order to tune the properties of the produced materials according to specifications. From the reported examples in this chapter, there is no doubt that microfluidic

flow direction that could efficiently increase the mixing after curved turns [44].

404 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

with an onion like configuration (Figure 9-b).

**3. Conclusions and outlook**

advantages of microfluidics in producing complex nanostructures.

**Figure 9.** Synthesis of polymeric nanomaterials in microfluidic systems. a) Microfluidic capillary device for prepara‐ tion of O/W/O/W triple emulsion drops. Microcapsules containing single-, two-, three, and four-core particles. Micro‐ capsules containing a magneto-responsive core particle which align upon application of an external magnetic field. (reprinted with permission from ref. [45], copyright 2011 WILEY-VCH Verlag GmbH & Co).b) Quadruple emulsion drops generation and downstream motion (reprinted with permission from ref. [45], copyright 2011 WILEY-VCH Ver‐ lag GmbH & Co).c) Images and characteristics of the polymerized PEGDA microparticles produced by jet mode in a flow focusing microreactor and a spiral channel section for photo-polymerization. Pseudocolor fluorescent image of highly monodisperse PEGDA polymer beads with 9.9 wt % of highly loaded CdSe/ZnS core−shell nanocrystals. (re‐ printed with permission from ref. [16], copyright 2014 American Chemical Society)

devices will represent new promising tools for the production of size and shape-controlled inorganic, polymer and hybrid nanomaterials.

It is clear that microfluidics can reproduce the nanostructures synthesised in batch reactors, improving their properties and reducing the synthesis time. Furthermore, particles with new morphologies and multicomponent compositions, like janus or multicore emulsions colloid can be prepared only in such systems. However, microfluidics reactors will need to overcome some shortcomings such as fouling and low throughput production. Along with chemical stability, surface properties, and ease of fabrication, other important considerations should be taken in account. Although several nanomaterials have been produced in a high-throughput mode, the majority of the procedures reported suffer from this lack, which limits the imple‐ mentation of this technology in companies and medical centres. Then, for future microfluidic systems to become high-throughput platforms to produce high-value particulate materials beyond the lab bench, they will need to become robust, well-integrated, scalable, modular, computer controlled. Another avenue of future research will be the integration of a quality evaluation unit together with feedback control through a combination of microfluidics, robotics and automation. Consequently, continuous flow microfluidic systems will have the potential to become a standard technology to produce nano/micro-particle formulations with unprecedented homogeneity and fine control, which on the other hand are not feasible to achieve in bulk techniques

Meanwhile, the search for applications which require the use of relatively small amounts of high quality nanomaterials will be vital for the development of the field. Considering the good reproducibility and quality on nanomaterials produced using microfluidics, it is believed that microsystems will push forward new applications in promising fields such as nanomedicine, drug delivery, electronic displays, and photonics.

## **Acknowledgements**

The People Program (CIG-Marie Curie Actions, REA grant agreement no. 321642) is gratefully acknowledged.

## **Author details**

Victor Sebastian Cabeza1,2

1 Institute of Nanoscience of Aragon (INA) and Department of Chemical Engineering and Environmental Technology, University of Zaragoza, Zaragoza, Spain

2 Networking Research Center on Bioengineering, Biomaterials and Nanomedicine (CIBER-BBN), Madrid, Spain

## **References**


[3] Willems & van den Wildenberg (W&W). Roadmap Report on Nanoparticles [Inter‐ net]. November 2005. Available from: http://nanoparticles.org/pdf/PerezBaxEscola‐ no.pdf [Accessed: August-2015]

systems to become high-throughput platforms to produce high-value particulate materials beyond the lab bench, they will need to become robust, well-integrated, scalable, modular, computer controlled. Another avenue of future research will be the integration of a quality evaluation unit together with feedback control through a combination of microfluidics, robotics and automation. Consequently, continuous flow microfluidic systems will have the potential to become a standard technology to produce nano/micro-particle formulations with unprecedented homogeneity and fine control, which on the other hand are not feasible to

Meanwhile, the search for applications which require the use of relatively small amounts of high quality nanomaterials will be vital for the development of the field. Considering the good reproducibility and quality on nanomaterials produced using microfluidics, it is believed that microsystems will push forward new applications in promising fields such as nanomedicine,

The People Program (CIG-Marie Curie Actions, REA grant agreement no. 321642) is gratefully

1 Institute of Nanoscience of Aragon (INA) and Department of Chemical Engineering and

2 Networking Research Center on Bioengineering, Biomaterials and Nanomedicine (CIBER-

[1] Wilson Center| Virginia Techstr. The project on Emerging Nanotechnologies [Inter‐ net]. Available from: HeaHPPT URL}http://www.nanotechproject.org/cpi/search-

[2] Victor Sebastian, Manuel Arruebo, and Jesus Santamaria. Reaction Engineering Strat‐

egies for the Production.. ;10:835-853. DOI: 10.1002/smll.201301641

Environmental Technology, University of Zaragoza, Zaragoza, Spain

products/ [Accessed: August-2015]

achieve in bulk techniques

**Acknowledgements**

acknowledged.

**Author details**

Victor Sebastian Cabeza1,2

BBN), Madrid, Spain

**References**

drug delivery, electronic displays, and photonics.

406 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences


cells from blood using a packed bed of custom-synthesized chitosan microparticles. Journal of Material Chemistry B. 2013;1:4313-4319. DOI: 10.1039/c3tb20818d


[26] Ho Cheung Shum,Daeyeon Lee, Insun Yoon, Tom Kodger, and David A. WeitzDou‐ ble Emulsion Templated Monodisperse Phospholipid Vesicles. Langmuir. 2008;24:7651-7653. DOI: 10.1021/la801833a

cells from blood using a packed bed of custom-synthesized chitosan microparticles.

Journal of Material Chemistry B. 2013;1:4313-4319. DOI: 10.1039/c3tb20818d

Communication. 2012;49:84-86. DOI: 10.1039/c2cc35652j

2011;32:3364-3370. DOI: 10.1002/elps.201100340

408 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

s10404-013-1152-6

2012.08.039

10.1155/2007/98134

[16] Seung-Kon Lee, Jinyoung Baek, and Klavs F. JensenHigh Throughput Synthesis of Uniform Biocompatible Polymer Beads with High Quantum Dot Loading Using Mi‐ crofluidic Jet-Mode Breakup. Langmuir. ;30:2216-2222. DOI: 10.1021/la4041198 [17] Simone Silvestrini, Tommaso Carofiglio and Michele MaggShape-selective growth of silver nanoparticles under continuous flow photochemical conditions. Chemical

[18] Shuqiang Huang,Bingcheng Lin.Jianhua QinMicrofluidic synthesis of tunable poly- (N-isopropylacrylamide) microparticles via PEG adjustment. Electrophoresis.

[19] Xianjie Kang, Chunxiong Luo,Qiong Wei,Chunyang Xiong,Qian Chen,Ying Chen, Qi OuyangMass production of highly monodisperse polymeric nanoparticles by parallel flow focusing system. Microfluidics and Nanofluidics. 2013;15:337-345. DOI: 10.1007/

[20] Florence Bally, Dhiraj Kumar Garg, Christophe A. Serra, Yannick Hoarau, Nicolas Anton, Cyril Brochon, Dambarudhar Parida, Thierry Vandamme, Georges Had‐ ziioannouImproved size-tunable preparation of polymeric nanoparticles by micro‐ fluidic nanoprecipitation. Polymer. 2012;53:5045-5051. DOI: 10.1016/j.polymer.

[21] Leyre Gomez, Manuel Arruebo, Victor Sebastian, Laura Gutierrez and Jesus Santa‐ mariaFacile synthesis of SiO2–Au nanoshells in a three-stage microfluidic system. Journal of MAterials Chemistry. 2012;22:21420-21425. DOI: 10.1039/c2jm34206e [22] D.V. Ravi Kumar, B.L.V. Prasad, A.A. KulkarniSegmented flow synthesis of Ag nanoparticles in spiral microreactor:Role of continuos and disperzsed phase. Chemi‐

[23] Victor Sebastian Cabeza,Simon Kuhn, Amol A. Kulkarni, and Klavs F. JensenSize-Controlled Flow Synthesis of Gold Nanoparticles Using a Segmented Flow Microflui‐

[24] Xinqing Chen, Manuel Arruebo, King Lun YeungFlow-synthesis of mesoporous sili‐ cas and their use in the preparation of magnetic catalysts for Knoevenagel condensa‐ tion reactions. Catalysis Today. 2013;204:140-147. DOI: 10.1016/j.cattod.2012.07.017

[25] J.Michael Kohler, H. Romanus, U. Hubner, and J.WagnerFormation of Star-Like and Core-Shell AuAg Nanoparticles during Two- and Three-Step Preparation in Batch and inMicrofluidic Systems. Journal of Nanomaterials. 20017;:98134. DOI:

cal Engineering Journal. 2012;192:357-368. DOI: 10.1016/j.cej.2012.02.084

dic Platform. Langmuir. 2012;28:7007-7013. DOI: 10.1021/la205131e


[38] Kritika Kumar, Adrian M. Nightingale, Siva H. Krishnadasan, Nazila Kamaly, Mar‐ zena Wylenzinska,Arridge, Katharina Zeissler, Will R. Branford, Ecaterina Ware, An‐ drew J. deMello and John C. deMelloDirect synthesis of dextran-coated superparamagnetic iron oxide nanoparticles in a capillary-based droplet reactor.

[39] Leyre Gomez, Victor Sebastian,Silvia Irusta,Alfonso Ibarra, Manuel Arruebo and Je‐ sus SantamariaScaled-up production of plasmonic nanoparticles using microfluidics: from metal precursors to functionalized and sterilized nanoparticles. Lab on a Chip. ;

[40] Christophe A. Serra,Zhenqi ChangMicrofluidic-Assisted Synthesis of Polymer.

[41] Daniel C. Pregibon, Mehmet Toner, Patrick S. DoyleMultifunctional Encoded Parti‐ cles for High-Throughput Biomolecule Analysis. Science. 2007;315:1393-1396. DOI:

[42] YongTae Kim, Bomy Lee Chung, Mingming Ma, Willem J. M. Mulder, Zahi A. Fayad,Omid C. Farokhzad, and Robert Langer,Mass Production and Size Control of Lipid−Polymer Hybrid Nanoparticles through Controlled Microvortices. Nano Let‐

[43] Jong-Min Lim, Archana Swami, Laura M. Gilson,Sunandini Chopra,Sungyoung Choi,Jun Wu,Robert Langer,Ultra-High Throughput Synthesof Nanoparticles with Homogeneousis Size Distribution Using a Coaxial,Turbulent Jet Mixer. ACS nano. ;

[44] Jiashu Sun,‡a Yunlei Xianyu,‡a Mengmeng Li,‡a Wenwen Liu,a Lu Zhang,Dingbin Liu,a Chao Liu,b Guoqing Hub and Xingyu Jiang\*A microfluidic origami chip for synthesis of functionalized polymeric nanoparticles. Nanoscale. 2013;(5):5262-5265.

[45] Shin-Hyun Kim and David A. WeitzOne-Step Emulsification of Multiple Concentric Shells with Capillary Microfluidic Devices. Angewandte Chemie International Edi‐

[46] Su Kyung Suh, Kai Yuet, Dae Kun Hwang, Ki Wan Bong, Patrick S. Doyle, and T. Alan HattonSynthesis of Nonspherical Superparamagnetic Particles: In Situ Copreci‐ pitation of Magnetic Nanoparticles in Microgels Prepared byStop-Flow Lithography. JOurnal of the Americal Chemical Society. 2012;134:7337-7343. DOI: 10.1021/

Chem.Eng. Technol.. 2008;31(8):1099-1115. DOI: 10.1002/ceat.200800219

Journal of Material Chemistry. 2012;22:4704-4708. DOI: 10.1039/c2jm30257h

14:325-332. DOI: 10.1039/c3lc50999k

410 Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

ters. 2012;12:3587-3591. DOI: 10.1021/nl301253v

tion. 2011;50:8731-8734. DOI: 10.1002/anie.201102946.

8:6056-6065. DOI: 10.1021/nn501371n

10.1126/science.1134929

DOI: 10.1039/c3nr01289a

ja209245v

## *Edited by Xiao-Ying Yu*

Increasing innovations and applications make microfluidics a versatile choice for researchers in many disciplines. This book consists of multiple review chapters that aim to cover recent advances and new applications of microfluidics in biology, electronics, energy, and materials sciences. It provides comprehensive views of various aspects of microfluidics ranging from fundamentals of fabrication, flow control, and droplet manipulation to the most recent exploration in emerging areas such as material synthesis, imaging and novel spectroscopy, and marriage with electronics. The chapters have many illustrations showcasing exciting results. This book should be useful for those who are eager to learn more about microfluidics as well as researchers who want to pick up new concepts and developments in this fast-growing field.

Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences

Advances in Microfluidics

New Applications in Biology, Energy, and

Materials Sciences

*Edited by Xiao-Ying Yu*

Photo by Ruliz / Can Stock