*2.2.3 Thermal time-of-flight flow sensing*

Both the calorimetric and anemometric flow sensors require a calibration of the real fluid for the desired precision or metrological accuracy, as the fluidic properties will have a nonlinear response in the full dynamic range. The limited dynamic range and the accuracy would not be desirable for the precision requirements for many microfluidic applications such as drug infusion. Also, these flow sensing products

could only provide mass flowrate measurements. The microfluidic applications would appreciate additional fluidic information such as fluidic concentration, physical or even chemical properties of the fluids at the same time. To this end, thermal time-of-flight sensing technology offers much of the competitive advantages. The thermal time-of-flight sensing concept can be traced back to the late 1940s [48] and has been an interest in many subsequent research works [49–51]. The thermal timeof-flight sensing measures the heat transfer transient time as well as the responses at each sensing element. Several sensing elements can be placed downstream of the microheater. Consequently, this approach can measure additional parameters other than the flowrates [52]. The sensor works with a thermal pulse or modulated thermal wave signals. Compared to calorimetry or anemometry, the transient time-domain data are much more immutable to the background interferences. Despite the advantages, a commercially available thermal time-of-flight flow meter is not seen until the past decade [42]. One reason could be that the microheater must possess a mass as small as possible for the needed thermal response to enable the measurement scheme. In the traditional approach, such a tiny wire is extremely vulnerable for reliability in actual applications. On the other hand, pure time-offlight will only measure the flow velocity. In contrast, the other parameters require advanced and complicated electronics that are not readily accessible until recent years. Nevertheless, the sensor build and package limitation will still lead to a nonpure time-of-flight, and calibration will be required to remove those effects. On the other hand, these effects can also be used to provide additional fluidic information. For the microfluidic applications, the microheater is driven with a modulated microheater, the constant heating spot in the flow channel is therefore eliminated. The sensor outputs flow velocity as well as fluidic mass flowrate and the additional data of the fluidic properties, making the thermal time-of-flight technology an ideal approach for the desired microfluidic flow measurement applications.

**Figure 2** shows a typical structure of a micromachined thermal time-of-flight sensor chip [53]. The micromachined process has a wide spectrum of materials selection to allow the sensor with excellent thermal isolation while not sacrificing reliability. This is particularly important for the thermal time-of-flight sensing that requires a super-fast thermal response. The blue materials showed in **Figure 2(b)** can be silicon or glass substrate. The gray colored block will be for thermal isolation. For example, a 10 ~ 15 μm parylene conformal layer will provide the properties of the good material of stiffness and robustness for the application. The green-colored materials need to have good thermal conductivity while excellent surface passivation for reliability. Ideal materials include multi-layered silicon nitride or silicon carbide. Underneath the microheater and sensing elements, a cavity will enhance the thermal performance of the sensor chip. The brown-colored elements are for microheater and sensing elements. One sensing element is placed directly on the substrate to measure the environmental temperature that provides the compensation of the microheater's temperature performance and control. In the photo shown in **Figure 2(a)**, the central element has another sensing element at the proximity of the microheater, which is used to fine-tune the microheater temperature or power with those other physical properties such as thermal conductivity can be precisely acquired.

The heat transfer in the thermal time-of-flight configuration is measured by the temperature *T* with time *t* in the flow direction *x* of non-uniform temperature distribution, determined by the flow medium's conductivity and diffusivity. The working principle can be expressed as below, [52].

$$\frac{\partial T}{\partial t} = \frac{k}{\rho c} \left( \frac{\partial^2 T}{\partial \mathbf{x}^2} \right) - V\_x \left( \frac{\partial T}{\partial \mathbf{x}} \right) + \frac{Q}{\rho c} \tag{1}$$

**63**

*Microfluidic Flow Sensing Approaches*

α, and the modulated heat *Q*, [54].

**Figure 2.**

*(b) cross-section schematic.*

two or three pairs of sensing elements.

**2.3 Coriolis flow sensing**

*DOI: http://dx.doi.org/10.5772/intechopen.96096*

Where *Q* is the heat generated by the microheater that is normally modulated with a square or sine wave, *k* is the thermal conductivity, ρ is the density, *c* is the heat capacity of the flow medium, and *V*x is the fluid velocity. With one dimensional approximate, the fluid velocity, *V*x, between the microheater and the sensing element would be determined by fluid thermal conductivity *k*, thermal diffusivity

*Example of a micromachined thermal time-of-flight sensing chip: (a) optical photo of the chip, top view;* 

*<sup>Q</sup>* ( *x Vt <sup>x</sup>* ) *<sup>T</sup>*

 *kt t* <sup>−</sup> = −

Therefore, if the sensor only has a microheater and a sensing element pair, the measurement will still be dependent on the flow medium properties. The microheater and the sensing elements all have the fluidic dependent response that needs to be removed for the complicated fluids. Simple calibration with the conventional fluid can be applied for the fluid measurement without losing the metrology accuracy. A micro-machining process' advantages allow placing multiple sensing elements on the same chip without adding any cost that makes it possible to have the measurement independent of the fluidic properties. The thermal time-of-flight will not be a simple flow velocity measurement. The measured changes in the amplitude are directly proportional to the heat transfer between the microheater and the sensing elements that will provide the mass flowrate similar to the calorimetric or anemometric approach per the data acquisition process. The time-domain data yield additional information, which allows the acquisition of additional fluidic thermal dynamic properties such as thermal conductivity and specific heat. In the microfluidic flow measurement, the liquid is generally non-compressible. The pressure effects of compressibility can be considered secondary. Compared to the gaseous fluids, liquid has a much large heat capacitance making the sensing element resistance-related temperature effects less pronounced. And most importantly, with the multiple sensing elements on a single chip, the measurement dynamic range can be substantially extended. A practical 7500:1 dynamic range can be achieved with

Coriolis mass flow sensing principle has been well documented, and the first commercial product was introduced to the market in 1977 by Micro Motion. It is a true mass flow sensing technology with very high precision by utilizing an exciting

 α 2

exp 4 4 (2)

π

#### **Figure 2.**

*Advances in Microfluidics and Nanofluids*

could only provide mass flowrate measurements. The microfluidic applications would appreciate additional fluidic information such as fluidic concentration, physical or even chemical properties of the fluids at the same time. To this end, thermal time-of-flight sensing technology offers much of the competitive advantages. The thermal time-of-flight sensing concept can be traced back to the late 1940s [48] and has been an interest in many subsequent research works [49–51]. The thermal timeof-flight sensing measures the heat transfer transient time as well as the responses at each sensing element. Several sensing elements can be placed downstream of the microheater. Consequently, this approach can measure additional parameters other than the flowrates [52]. The sensor works with a thermal pulse or modulated thermal wave signals. Compared to calorimetry or anemometry, the transient time-domain data are much more immutable to the background interferences. Despite the advantages, a commercially available thermal time-of-flight flow meter is not seen until the past decade [42]. One reason could be that the microheater must possess a mass as small as possible for the needed thermal response to enable the measurement scheme. In the traditional approach, such a tiny wire is extremely vulnerable for reliability in actual applications. On the other hand, pure time-offlight will only measure the flow velocity. In contrast, the other parameters require advanced and complicated electronics that are not readily accessible until recent years. Nevertheless, the sensor build and package limitation will still lead to a nonpure time-of-flight, and calibration will be required to remove those effects. On the other hand, these effects can also be used to provide additional fluidic information. For the microfluidic applications, the microheater is driven with a modulated microheater, the constant heating spot in the flow channel is therefore eliminated. The sensor outputs flow velocity as well as fluidic mass flowrate and the additional data of the fluidic properties, making the thermal time-of-flight technology an ideal

approach for the desired microfluidic flow measurement applications.

properties such as thermal conductivity can be precisely acquired.

ρ

working principle can be expressed as below, [52].

The heat transfer in the thermal time-of-flight configuration is measured by the temperature *T* with time *t* in the flow direction *x* of non-uniform temperature distribution, determined by the flow medium's conductivity and diffusivity. The

> ∂∂ ∂ = −+ ∂∂ ∂ 2

*x Tk T T Q <sup>V</sup> t cx x c*

 ρ

<sup>2</sup> (1)

**Figure 2** shows a typical structure of a micromachined thermal time-of-flight sensor chip [53]. The micromachined process has a wide spectrum of materials selection to allow the sensor with excellent thermal isolation while not sacrificing reliability. This is particularly important for the thermal time-of-flight sensing that requires a super-fast thermal response. The blue materials showed in **Figure 2(b)** can be silicon or glass substrate. The gray colored block will be for thermal isolation. For example, a 10 ~ 15 μm parylene conformal layer will provide the properties of the good material of stiffness and robustness for the application. The green-colored materials need to have good thermal conductivity while excellent surface passivation for reliability. Ideal materials include multi-layered silicon nitride or silicon carbide. Underneath the microheater and sensing elements, a cavity will enhance the thermal performance of the sensor chip. The brown-colored elements are for microheater and sensing elements. One sensing element is placed directly on the substrate to measure the environmental temperature that provides the compensation of the microheater's temperature performance and control. In the photo shown in **Figure 2(a)**, the central element has another sensing element at the proximity of the microheater, which is used to fine-tune the microheater temperature or power with those other physical

**62**

*Example of a micromachined thermal time-of-flight sensing chip: (a) optical photo of the chip, top view; (b) cross-section schematic.*

Where *Q* is the heat generated by the microheater that is normally modulated with a square or sine wave, *k* is the thermal conductivity, ρ is the density, *c* is the heat capacity of the flow medium, and *V*x is the fluid velocity. With one dimensional approximate, the fluid velocity, *V*x, between the microheater and the sensing element would be determined by fluid thermal conductivity *k*, thermal diffusivity α, and the modulated heat *Q*, [54].

$$T = \frac{Q}{4\pi k t} \exp\left(-\frac{\left(\varkappa - V\_\times t\right)^2}{4\alpha t}\right) \tag{2}$$

Therefore, if the sensor only has a microheater and a sensing element pair, the measurement will still be dependent on the flow medium properties. The microheater and the sensing elements all have the fluidic dependent response that needs to be removed for the complicated fluids. Simple calibration with the conventional fluid can be applied for the fluid measurement without losing the metrology accuracy. A micro-machining process' advantages allow placing multiple sensing elements on the same chip without adding any cost that makes it possible to have the measurement independent of the fluidic properties. The thermal time-of-flight will not be a simple flow velocity measurement. The measured changes in the amplitude are directly proportional to the heat transfer between the microheater and the sensing elements that will provide the mass flowrate similar to the calorimetric or anemometric approach per the data acquisition process. The time-domain data yield additional information, which allows the acquisition of additional fluidic thermal dynamic properties such as thermal conductivity and specific heat. In the microfluidic flow measurement, the liquid is generally non-compressible. The pressure effects of compressibility can be considered secondary. Compared to the gaseous fluids, liquid has a much large heat capacitance making the sensing element resistance-related temperature effects less pronounced. And most importantly, with the multiple sensing elements on a single chip, the measurement dynamic range can be substantially extended. A practical 7500:1 dynamic range can be achieved with two or three pairs of sensing elements.

#### **2.3 Coriolis flow sensing**

Coriolis mass flow sensing principle has been well documented, and the first commercial product was introduced to the market in 1977 by Micro Motion. It is a true mass flow sensing technology with very high precision by utilizing an exciting tube which fluid is flowing through, and the tube oscillates artificially. The changes of the tube oscillation in time and space are a direct measure of the mass flow. One advantage of the Coriolis sensing approach is that the fluid density can be simultaneously determined from the oscillation frequency of the measuring tube. But it also requires a minimum density of fluid such that the resolution of the oscillation can be registered. It also suffers a high-pressure loss and a smaller dynamic range. The high cost of the measuring tube manufacture sparked the attempt with micromachining, and the first research paper was published in 1997 [55]. With the micromachining process, the MEMS Coriolis mass flow sensor can be well applied for microfluidic flow measurements. The commercially available Coriolis meters sensors via micromachining either consists of a silicon microtube via silicon wafer fuse bonding and an integrated temperature sensor [56] or a silicon-rich silicon nitride tube coupled with a strain gauge readout [57, 58]. The micromachined Coriolis sensor using silicon nitride tube has a thin tube wall of about 1.2 μm and is much lighter than the silicon tube. Hence, the light-weighted tube would have a smaller mass than the fluid it measures that simplify the package, and leads to the possibility to measure the fluids at ambient pressure. The demonstrated Coriolis sensor could measure liquid mass flow, density, and temperature (if a temperature sensor is integrated) simultaneously. Another advantage for the MEMS Coriolis mass flow sensor is that it usually operates at a much higher resonant frequency with substantially less vibratory influences from the environments than those for the traditional Coriolis mass flow technology.

Like the MEMS thermal mass flow sensors, the micro Coriolis mass flow sensor also requires clean fluid. Even fine particles can damage or clog the sensor, considering the measuring tube's tiny channel. Besides, the sensor will not function well in liquid with high viscosities and liquid with chemical reactions. The high-speed liquid flow may also alter the performance of the sensor unless bypass configuration is applied. The superior true mass flow accuracy of a Coriolis sensor is overshadowed by its footprint, complication in the package, and cost in the manufacturing process that diminishes high volume and/or disposable applications, which would be a necessity in some microfluidic applications for cross-contamination prevention. The fluidic property independence measurement characteristics also limit its measurements only for flowrate and density. Other fluidic property measurements will require integrating additional sensing elements, further enlarging the sensor footprint, indicating an even higher cost for the final product manufacture. Therefore, although the micromachined Coriolis sensor's demonstration has been over two decades, the applications are still very limited.

#### **2.4 Other microfluidic flow sensing technologies**

Flow sensors are likely the ones that can be made with the most versatile technologies and are vastly selectable to the applications. More than twenty different physical measurement principles are commercially available on the market for flow metrology. However, for microfluidics, the options are limited. Other than the commercially available thermal mass flow sensors and Coriolis flow sensors, micromachining advancement offers opportunities for many studies with a wide spectrum of technologies applied for microfluidic flow sensing. But commercialization of many of those is still in question. Some selected researches micromachined flow sensing technologies are discussed below.

#### *2.4.1 Acoustic microfluidic flow sensing*

In addition to micromachined thermal and Coriolis sensors, micromachined ultrasonic sensors are also commercially available. Ultrasonic flow sensing is one

**65**

*Microfluidic Flow Sensing Approaches*

*DOI: http://dx.doi.org/10.5772/intechopen.96096*

flow metrology is yet to be demonstrated [60].

compared to thermal sensing approaches.

*2.4.2 Differential pressure microfluidic flow sensing*

pressure and hence the accuracy of the deduced flowrate.

of the well-documented technology for flow metrology with high accuracy. By measuring the time differences of the ultrasonic signals propagating in the opposite direction of a fluidic medium, the flow speed can be accurately measured. Therefore, it has the advantage of a pure velocity measurement independent of the fluidic properties. As a sound propagation, it will not require direct contact with the fluids that it measures, or it is non-invasive, which is very attractive for microfluidic related medical applications. However, it will normally require dual transducers placed in opposite directions or at a certain angle with respect to a reflector. This prevents the reduction in footprint and cost. For the microfluidic applications, its signals reduce significantly at the low flow speed, and it is also very sensitive to the fluids where cavitation or dissolution may exist. The current commercially available ultrasonic flow meters for microfluidics have about a 50:1 dynamic range and a detection limit of 300 μL/min [59]. Some research indicated that the ultrasonic transducer could be integrated into the microfluidic channel, but the capability for

Acoustic device applications in microfluidics are mostly for fluid handling, and surface acoustic wave (SAW) sensing and actuation is another approach that can be integrated into the microfluidic channels [61]. Some efforts were also made to measure the flowrate with the SAW devices. It has been reported that a micromachined interdigital transducer (IDT) could direct the fluidic droplets via the excited acoustic streaming that is fast and material independent [62]. In another study, [63] a SAW sensor with dual symmetrical IDTs made on a 30 mm by 30 mm square quartz crystal substrate was used to measure the flowrate in a designed channel by recording the delay times and the corresponding frequencies. A close to a linear correlation between the phase shift from the delay time and flowrate was established. The SAW sensors can be independent of the fluidic properties; however, they require a much larger footprint, and temperature compensation is also complicated

Measurement of flowrate with differential pressure is one of the oldest flow sensing technologies. Micromachined differential pressure sensors have been well established and are widely available on the market at a very low cost. Most sensors are made on a silicon nitride membrane or diaphragm with the piezoresistive sensing elements at the edges of the membrane or with a capacitance measuring principle for the low differential pressures [64, 65]. The advantages of a differential pressure sensor for flow measurement are lower power consumption and relatively easy installation with fewer effects on the flow conditioning. They are also independent of the fluidic properties. The microfluidic flow regime is purely laminar, and the pressure loss is linear with the flow velocity. However, limited by its sensitivity, the measurement dynamic range of a differential pressure sensor is normally small. In particular, for microfluidic applications, the pressure drop at a tiny distance may not even generate enough sensitivity for the measurement. The dependence of the microfluid's pressure loss on the dynamic viscosity also requires a temperature sensor at the proximity for the needed compensation. Other phenomena such as cavitation or multi-phase flow will have a big impact on the measurement of the

Flow measurement with drag force is an alternative pressure-related flow sensing approach. Due to the size restriction, such a sensor does not favor being placed inside the microfluidic channel. However, in an ideally integrated microfluidic system, there will be valves and other actuators. The drag force-sensing approach could be combined with the actuation parts in the system. A typical drag force

### *Microfluidic Flow Sensing Approaches DOI: http://dx.doi.org/10.5772/intechopen.96096*

*Advances in Microfluidics and Nanofluids*

tube which fluid is flowing through, and the tube oscillates artificially. The changes of the tube oscillation in time and space are a direct measure of the mass flow. One advantage of the Coriolis sensing approach is that the fluid density can be simultaneously determined from the oscillation frequency of the measuring tube. But it also requires a minimum density of fluid such that the resolution of the oscillation can be registered. It also suffers a high-pressure loss and a smaller dynamic range. The high cost of the measuring tube manufacture sparked the attempt with micromachining, and the first research paper was published in 1997 [55]. With the micromachining process, the MEMS Coriolis mass flow sensor can be well applied for microfluidic flow measurements. The commercially available Coriolis meters sensors via micromachining either consists of a silicon microtube via silicon wafer fuse bonding and an integrated temperature sensor [56] or a silicon-rich silicon nitride tube coupled with a strain gauge readout [57, 58]. The micromachined Coriolis sensor using silicon nitride tube has a thin tube wall of about 1.2 μm and is much lighter than the silicon tube. Hence, the light-weighted tube would have a smaller mass than the fluid it measures that simplify the package, and leads to the possibility to measure the fluids at ambient pressure. The demonstrated Coriolis sensor could measure liquid mass flow, density, and temperature (if a temperature sensor is integrated) simultaneously. Another advantage for the MEMS Coriolis mass flow sensor is that it usually operates at a much higher resonant frequency with substantially less vibratory influences from

the environments than those for the traditional Coriolis mass flow technology.

over two decades, the applications are still very limited.

**2.4 Other microfluidic flow sensing technologies**

flow sensing technologies are discussed below.

*2.4.1 Acoustic microfluidic flow sensing*

Like the MEMS thermal mass flow sensors, the micro Coriolis mass flow sensor also requires clean fluid. Even fine particles can damage or clog the sensor, considering the measuring tube's tiny channel. Besides, the sensor will not function well in liquid with high viscosities and liquid with chemical reactions. The high-speed liquid flow may also alter the performance of the sensor unless bypass configuration is applied. The superior true mass flow accuracy of a Coriolis sensor is overshadowed by its footprint, complication in the package, and cost in the manufacturing process that diminishes high volume and/or disposable applications, which would be a necessity in some microfluidic applications for cross-contamination prevention. The fluidic property independence measurement characteristics also limit its measurements only for flowrate and density. Other fluidic property measurements will require integrating additional sensing elements, further enlarging the sensor footprint, indicating an even higher cost for the final product manufacture. Therefore, although the micromachined Coriolis sensor's demonstration has been

Flow sensors are likely the ones that can be made with the most versatile technologies and are vastly selectable to the applications. More than twenty different physical measurement principles are commercially available on the market for flow metrology. However, for microfluidics, the options are limited. Other than the commercially available thermal mass flow sensors and Coriolis flow sensors, micromachining advancement offers opportunities for many studies with a wide spectrum of technologies applied for microfluidic flow sensing. But commercialization of many of those is still in question. Some selected researches micromachined

In addition to micromachined thermal and Coriolis sensors, micromachined ultrasonic sensors are also commercially available. Ultrasonic flow sensing is one

**64**

of the well-documented technology for flow metrology with high accuracy. By measuring the time differences of the ultrasonic signals propagating in the opposite direction of a fluidic medium, the flow speed can be accurately measured. Therefore, it has the advantage of a pure velocity measurement independent of the fluidic properties. As a sound propagation, it will not require direct contact with the fluids that it measures, or it is non-invasive, which is very attractive for microfluidic related medical applications. However, it will normally require dual transducers placed in opposite directions or at a certain angle with respect to a reflector. This prevents the reduction in footprint and cost. For the microfluidic applications, its signals reduce significantly at the low flow speed, and it is also very sensitive to the fluids where cavitation or dissolution may exist. The current commercially available ultrasonic flow meters for microfluidics have about a 50:1 dynamic range and a detection limit of 300 μL/min [59]. Some research indicated that the ultrasonic transducer could be integrated into the microfluidic channel, but the capability for flow metrology is yet to be demonstrated [60].

Acoustic device applications in microfluidics are mostly for fluid handling, and surface acoustic wave (SAW) sensing and actuation is another approach that can be integrated into the microfluidic channels [61]. Some efforts were also made to measure the flowrate with the SAW devices. It has been reported that a micromachined interdigital transducer (IDT) could direct the fluidic droplets via the excited acoustic streaming that is fast and material independent [62]. In another study, [63] a SAW sensor with dual symmetrical IDTs made on a 30 mm by 30 mm square quartz crystal substrate was used to measure the flowrate in a designed channel by recording the delay times and the corresponding frequencies. A close to a linear correlation between the phase shift from the delay time and flowrate was established. The SAW sensors can be independent of the fluidic properties; however, they require a much larger footprint, and temperature compensation is also complicated compared to thermal sensing approaches.
