We are IntechOpen, the world's largest scientific publisher of Open Access books.

3,250+ Open access books available

106,000+

International authors and editors

112M+ Downloads

151 Countries delivered to Our authors are among the

Top 1% most cited scientists

12.2%

Contributors from top 500 universities

Selection of our books indexed in the Book Citation Index in Web of Science™ Core Collection (BKCI)

## Interested in publishing with us? Contact book.department@intechopen.com

Numbers displayed above are based on latest data collected. For more information visit www.intechopen.com

## **Meet the editor**

Professor Kahrizi completed his PhD in the field of applied solid-state physics, in 1985. Currently, he is a professor and a director of the Micro Devices and Micro-fabrications in the Department of Electrical and Computer Engineering at Concordia University, Montreal, Canada. Dr Kahrizi's research and teaching covers a wide spectrum in the fields of Material Sciences, Solid

State Devices and Nano-science, which has resulted in numerous high quality journal and conference articles related to material sciences and fundamental issues concerning micro- and nano- structures. He is the Chair of IEEE, EDS, Montreal chapter and a member of Professional Engineers of Ontario. Dr Kahrizi is also a member of the Center for Applied Research on Polymers and Composite, as well as a member of Center for Biorecognition and Biosensors.

## Contents

#### **Preface** XI



## Preface

Making microsystems at a scale level of few microns is called micromachining. Micromachining is used to fabricate three-dimensional microstructures. It is the foundation of a technology called Micro-Electro-Mechanical-Systems (MEMS). MEMS usually consist of three major parts: sensors, actuators, and an associate electronic circuitry that acts as the brain and controller of the whole system.

There are two types of micromachining. Bulk micromachining starts with a silicon wafer or other substrate, which is selectively etched using dry or wet etching techniques, laser ablation, or focused ion beams. The most common substrate in this technology is single crystal silicon. Variation in the strength of bonds along various planes in this periodic structure makes it susceptible to etching with various rates along different crystal orientations. The wet anisotropic etching of silicon in hydroxide solutions, like potassium hydroxide (KOH) or tetra methyl ammonium hydroxide (TMAH), is performed to etch silicon selectively along a specific orientation. Due to the high selective ratio, the etch rate varies along various orientations in this semiconductor, making it possible to design and fabricate many 3-D microstructures. This type of etching is inexpensive and is generally used in early, low-budget research. Although the wet etching is the most common practice in micromachining, dry etching techniques like laser ablation and focused ion beams, are also often used to produce microstructures. This technique is not only used to produce micro devices; it has now been extended to fabricate many devices at the level of nano scales.

Another micromachining technique is surface micromachining, which involves fabrication of layers (usually using standard CMOS technology) on the surface of a substrate, followed by etching of the sacrificial layers.

The purpose of this book is to introduce advances in micromachining technology. For this, we have gathered review articles related to various techniques and methods of micro/nano fabrications from esteemed researchers and scientists. The book consists of 13 chapters. The first two chapters demonstrate fabrication of several micro and nano devices using Focused Ion Beams techniques. The next five chapters are related to the application of lasers and laser ablation techniques used in bulk micromachining. Several other specialized methods and technologies are presented in the subsequent chapters. Throughout the book, each chapter gives a complete description of a specific

#### XII Preface

micromachining method, design, associate analytical works, experimental set-up, and the final fabricated devices, followed by many references related to this field of research available in other literature. Due to the multidisciplinary nature of MEMS and nanotechnology, this collection of articles can be used by scientists and researchers in the disciplines of engineering, material sciences, physics and chemistry

> **Mojtaba Kahrizi, Professor**  ECE Department,

Concordia University, Montreal, Quebec, Canada

X Preface

micromachining method, design, associate analytical works, experimental set-up, and the final fabricated devices, followed by many references related to this field of research available in other literature. Due to the multidisciplinary nature of MEMS and nanotechnology, this collection of articles can be used by scientists and researchers

**Mojtaba Kahrizi, Professor** 

ECE Department, Concordia University, Montreal, Quebec,

Canada

in the disciplines of engineering, material sciences, physics and chemistry

## **Focused Ion Beam Based Three-Dimensional Nano-Machining**

Gunasekaran Venugopal1,2,

Shrikant Saini1 and Sang-Jae Kim1,3 *1Jeju National University, Department of Mechanical Engineering, Jeju, 2Karunya University, Department of Nanosciences and Technology, Tamil Nadu, 3Jeju National University, Department of Mechatronics Engineering, Jeju, 1,3South Korea 2India* 

#### **1. Introduction**

In recent days, the micro/nano machining becomes an important process to fabricate micro/nano scale dimensional patterns or devices for many applications, especially in electrical and electronic devices. There are two kinds micro-machining in use. i) bulk micromachining, ii) surface micro-maching. In the case of bulk micromaching, the structures can be made by etching inside a substrate selectively, however, in the case of surface micromachining; the patterns can be made on the top a desired substrate. FIB machining is considered as a one of famous bulk micro-machining processes. Many fabrication methods have been applied to fabricate the devices with smaller sizes (Kim, 1999; Latyshev, 1997). However, conventional until now the size of the smallest pattern was only 2×2 μm2 was achieved with a lithography technique (Odagawa et al., 1998). Three dimensional as an alternative approach, focused-ion-beam (FIB) etching technique is the best choice for the micro/nano scale patterning. FIB 3-D etching technology is now emerged as an attractive tool for precision lithography. And it is a well recognized technique for making nanoscale stacked-junction devices, nano-ribbons and graphene based 3-D Single Electron Transistor (SET) devices.

FIB micro/nano machining is a direct etching process without the use of masking and process chemicals, and demonstrates sub-micrometer resolution. FIB etching equipments have shown potential for a variety of new applications, in the area of imaging and precision micromachining (Langford, 2001; Seliger, 1979). As a result, the FIB has recently become a popular candidate for fabricating high-quality micro-devices or high-precision microstructures (Melnagilis et al., 1998). For example, in a micro-electro-mechanical system (MEMS), this processing technique produces an ultra microscale structure from a simple sensor device, such as, the Josephson junction to micro-motors (Daniel et al., 1997). Also, the FIB processing enables precise cuts to be made with great flexibility for micro- and nanotechnology. Also, the method of fabricating three-dimensional (3-D) micro- and nanostructures on thin films and single crystals by FIB etching have been developed in order to fabricate the 3-D sensor structures (Kim, 2008, 1999).

In this chapter, the focused ion beam (FIB) based three-dimensional nano-machining will be discussed in detail in which the nano-machining procedures are focused with fabricating nanoscale stacked junctions of layered-structured materials such as graphite, Bi2Sr2Can-1CunO2n+4+x (BSCCO) family superconductor (Bi-2212, Bi-2223, etc.,) and YBa2Cu3O7 (YBCO) single crystals, etc. This work could show a potential future in further development of nanoquantum mechanical electron devices and their applications.

#### **2. Classification of machining**

Micromachining is the basic technology for fabrication of micro-components of size in the range of 1 to 500 micrometers. Their need arises from miniaturization of various devices in science and engineering, calling for ultra-precision manufacturing and micro-fabrication. Micromachining is used for fabricating micro-channels and micro-grooves in micro-fluidics applications, micro-filters, drug delivery systems, micro-needles, and micro-probes in biotechnology applications. Micro-machined components are crucial for practical advancement in Micro-electromechanical systems (MEMS), Micro-electronics (semiconductor devices and integrated circuit technology) and Nanotechnology. This kind of machining can be applicable for the bulk materials in which the unwanted portions of the materials can be removed while patterning.

In the bulk machining, the materials with the dimensions of more than in the range of micrometer or above centimetre scale are being used for the machining process. A best example for the bulk machining process is that the thread forming process on a screw or bolt, formation of metal components. Also this process can be applicable to produce 3D MEMS structures, which is now being treated as one of older techniques. This also uses anisotropic etching of single crystal silicon. For example, silicon cantilever beam for atomic force microscope (AFM).

Surface micro-machining is another new technique/process for producing MEMS structures. This uses etching techniques to pattern micro-scale structures from polycrystalline (poly) silicon, or metal alloys. Example: accelerometers, pressure sensors, micro gears and transmission, and micro mirrors etc. Micromachining has evolved greatly in the past few decades, to include various techniques, broadly classified into mask-based and tool-based, as depicted in the diagram below.

While mask-based processes can generate 2-D/2.5-D features on substrates like semiconductor chips, tools-based processes have the distinct advantage of being able to adapt to metallic and non-metallic surfaces alike, and also generate 3-D features and/or free-form sculpted surfaces. However, the challenges of achieving accuracy, precision and resolution persist.

Internationally, the race to fabricate the smallest possible component has lead to realization of sizes ever below 10 µm, even though the peak industrial requirement has been recognized at 100s of µm. Thus, the present situation is particularly advantageous for the industry that develops/fabricates nano/micron scale components.

#### **2.1 Various techniques of micromachining**

Micromachining can be done by following various techniques.


2 Micromachining Techniques for Fabrication of Micro and Nano Structures

In this chapter, the focused ion beam (FIB) based three-dimensional nano-machining will be discussed in detail in which the nano-machining procedures are focused with fabricating nanoscale stacked junctions of layered-structured materials such as graphite, Bi2Sr2Can-1CunO2n+4+x (BSCCO) family superconductor (Bi-2212, Bi-2223, etc.,) and YBa2Cu3O7 (YBCO) single crystals, etc. This work could show a potential future in further development of nano-

Micromachining is the basic technology for fabrication of micro-components of size in the range of 1 to 500 micrometers. Their need arises from miniaturization of various devices in science and engineering, calling for ultra-precision manufacturing and micro-fabrication. Micromachining is used for fabricating micro-channels and micro-grooves in micro-fluidics applications, micro-filters, drug delivery systems, micro-needles, and micro-probes in biotechnology applications. Micro-machined components are crucial for practical advancement in Micro-electromechanical systems (MEMS), Micro-electronics (semiconductor devices and integrated circuit technology) and Nanotechnology. This kind of machining can be applicable for the bulk materials in which the unwanted portions of the

In the bulk machining, the materials with the dimensions of more than in the range of micrometer or above centimetre scale are being used for the machining process. A best example for the bulk machining process is that the thread forming process on a screw or bolt, formation of metal components. Also this process can be applicable to produce 3D MEMS structures, which is now being treated as one of older techniques. This also uses anisotropic etching of single crystal silicon. For example, silicon cantilever beam for atomic

Surface micro-machining is another new technique/process for producing MEMS structures. This uses etching techniques to pattern micro-scale structures from polycrystalline (poly) silicon, or metal alloys. Example: accelerometers, pressure sensors, micro gears and transmission, and micro mirrors etc. Micromachining has evolved greatly in the past few decades, to include various techniques, broadly classified into mask-based

quantum mechanical electron devices and their applications.

**2. Classification of machining** 

materials can be removed while patterning.

and tool-based, as depicted in the diagram below.

force microscope (AFM).


#### **Photolithography**

This technique is being used in microelectronics fabrication and also used to pattern oxide/nitride/polysilicon films on silicon substrate. In this process, the basic steps involved are, photoresist development, etching, and resist removal. Photolithographic process can be described as follows:

The wafers are chemically cleaned to remove particulate matter, organic, ionic, and metallic impurities. High-speed centrifugal whirling of silicon wafers known as "Spin Coating" produces a thin uniform layer of photoresist (a light sensitive polymer) on the wafers. Photoresist is exposed to a set of lights through a mask often made of quartz. Wavelength of light ranges from 300-500 nm (UV) and X-rays (wavelengths 4-50 Angstroms). Two types of photoresist are used: (a) Positive: whatever shows, goes (b) Negative: whatever shows, stays. The photo resist characteristics after UV exposure are shown below in Fig. 1

Fig. 1. Photoresist characteristics in UV exposure

#### **Etching**

Normally etching process can be classified in to two kinds. (a) Wet etching (b) Dry etching. The wet etching process involves transport of reactants to the surface, surface reaction and transport of products from surfaces. The key ingredients are the oxidizer (e.g. H2O2, HNO3), the acid or base to dissolve the oxidized surface (e.g. H2SO4, NH4OH) and dilutent media to transport the products through (e.g. H2O). Dry etching process involves two kinds. (a) plasma based and (b) non plasma based.

#### **LIGA**

The LIGA is a German term which means LIthographie (Lithography) **G**alvanoformung (Electroforming) **A**bforming (Molding). The exact English meaning of LIGA is given in parenthesis. This process involves X-ray irradiation, resist development, electroforming and resist removal.

The detailed LIGA process description is discussed below:

	- Deep X-ray (0.01 1nm wavelength) lithography can produce high aspect ratios (1 mm high and a lateral resolution of 0.2 μm).
	- X-rays break chemical bonds in the resist; exposed resist is dissolved using wetetching process.
	- The spaces generated by the removal of the irradiated plastic material are filled with metal (e.g. Ni) using electro-deposition process.
	- Precision grinding with diamond slurry-based metal plate used to remove substrate layer/metal layer.
	- PMMA resist exposed to X-ray and removed by exposure to oxygen plasma or through wet-etching.
	- Metal mold from LIGA used for injection molding of MEMS.

LIGA Process Capability


#### **Laser ablation**

High-power laser pulses are used to evaporate matter from a target surface. In this process, a supersonic jet of particles (plume) is ejected normal to the target surface which condenses on substrate opposite to target. The ablation process takes place in a vacuum chamber either in vacuum or in the presence of some background gas. The graphical process scheme is given below in Fig.2.

Fig. 2. Laser ablation experiment.

#### **Mechanical micromachining**

4 Micromachining Techniques for Fabrication of Micro and Nano Structures

the acid or base to dissolve the oxidized surface (e.g. H2SO4, NH4OH) and dilutent media to transport the products through (e.g. H2O). Dry etching process involves two kinds. (a)

The LIGA is a German term which means LIthographie (Lithography) **G**alvanoformung (Electroforming) **A**bforming (Molding). The exact English meaning of LIGA is given in parenthesis. This process involves X-ray irradiation, resist development, electroforming and






High-power laser pulses are used to evaporate matter from a target surface. In this process, a supersonic jet of particles (plume) is ejected normal to the target surface which condenses on substrate opposite to target. The ablation process takes place in a vacuum chamber either in vacuum or in the presence of some background gas. The graphical process scheme

plasma based and (b) non plasma based.

etching process.

The detailed LIGA process description is discussed below:

mm high and a lateral resolution of 0.2 μm).

with metal (e.g. Ni) using electro-deposition process.


High aspect ratio structures: 10-50 μm with Max. height of 1-500 μm

Deep X-ray lithography and mask technology

substrate layer/metal layer.

through wet-etching.

**LIGA** 

resist removal.

Electroforming

Resist Removal

Plastic Molding

**Laser ablation** 

LIGA Process Capability

is given below in Fig.2.

Fig. 2. Laser ablation experiment.

 Surface roughness < 50 nm High accuracy < 1μm

Lithography or etching methods are not capable of making true 3D structures e.g. free form surfaces and also limited in range of materials. Mechanical machining is capable of making free form surfaces in wide range of materials. Can we scale conventional/non-traditional machining processes down to the micron level? Yes! There are two approaches used to machine micron and sub-micron scale features.


Mechanical micromachining process descriptions are given below:


#### **2.2 Focused-ion-beam (FIB) technique for nanofabrication**

The focused ion beam based nanofabrication method can be followed for the fabricating the nanoscale devices on materials based on metal and non-metallic elements, particularly the layered structure materials like graphite, Bi-2212 and YBCO which are recently attracted the world scientific community due to their interesting electrical and electronic properties reported in recent reports (Venugopal, 2011; Kim, 2001).

Graphite is considered as a well known layered-structured material in which carbon sheets are arranged in a stacked-manner with interlayer distance of 0.34 nm. Each single graphite sheet is known as a graphene layer which is now becoming as one of hot topic in the world scientific community. In the recent reports (Venugopal, 2011a, 2011b, 2011c), the fabrication of submicron and below submicron stacked junctions were carved from the bulk graphite materials using FIB 3-D etching. The interesting results were obtained in those observations that the graphite stacked-junction with in-plane area *A* of 0.25 μm2 showed nonlinear concave-like *I–V* characteristics even at 300 K; however the stack with A ≥ 1 μm2 were shown an ohmic-like *I–V* characteristic at 300 K for both low and high-current biasing. It turned into nonlinear characteristics when the temperature goes down. These results may open road to develop further graphite based nonlinear electronic devices. Further researches are being carried out to find unexplored properties of graphite nano devices fabricated using FIB micro/nano machining technology.

The focused ion beam (FIB) machining to make micro-devices and microstructures has gained more and more attention recently (Tseng, 2004). FIB can be used as a direct milling method to make microstructures without involving complicated masks and pattern transfer processes. FIB machining has advantages of high feature resolution, and imposes no limitations on fabrication materials and geometry. Focused ion beams operate in the range of 10-200 keV. As the ions penetrate the material, they loose their energy and remove substrate atoms. FIB has proven to be an essential tool for highly localized implantation doping, mixing, micromachining, controlled damage as well as ion-induced deposition. The technological challenge to fabricate nanoholes using electron beam lithography and the minimal feature size accessible by these techniques is typically limited to tens of nanometers, thus novel procedures must be devised (Zhou, 2006).

The patterning of samples using the FIB (focused ion beam) technique is a very popular technique in the field of inspection of integrated circuits and electronic devices manufactured by the semi-conductor industry or research laboratories. This is the case mainly for prototyping devices. The FIB technique allowing us to engrave materials at very low dimensions is a complement of usual lithographic techniques such as optical lithography. The main difference is that FIB allows direct patterning and therefore does not require an intermediate sensitive media or process (resist, metal deposited film, etching process). FIB allows 3D patterning of target materials using a finely focused pencil of ions having speeds of several hundreds of km s−1 at impact. Concerning the nature of the ions

most existing metals can be used in FIB technology as pure elements or in the form of alloys, although gallium (Ga+ ions) is preferred in most cases.

Many device fabrication techniques based on electron beam lithography followed by reactive-ion etching (RIE), chemical methods, and evaporation using hard Si shadow masks, and including lithography-free fabrication, have been reported. The procedures, however, are complex and yield devices with dimensions of ~5 to 50 nm, which are restricted to simple geometries. RIE creates disordered edges, and the chemical methods produce irregular shapes with distributed flakes, which are not suitable for electronic-device application.

Practically, FIB patterning can be achieved either by local surface defect generation, by ion implantation or by local sputtering. These adjustments are obtained very easily by varying the locally deposited ion fluence with reference to the sensitivity of the target and to the selected FIB processing method (Gierak, 2009). The FIB milling involves two processes: 1) Sputtering*,* ions with high energy displace and remove atoms of substrate material, and the ions lose their energy as they go into the substrate; 2) Re-deposition, the displaced substrate atoms, that have gained energy from ions through energy transfer, go through similar process as ions, sputtering other atoms, taking their vacancy, or flying out.

A focused gallium ion beam having an energy typically around 30 keV is scanned over the sample surface to create a pattern through topographical modification, deposition or sputtering. A first consequence is that, mainly because of the high ion doses required (~1018 ions cm−2) and of the limited beam particle intensity available in the probe, FIB etchingbased processes remain relatively slow. We may recall that for most materials, the material removal rate for 30 keV gallium ions is around 1–10 atoms per incident ion, corresponding to a machining rate of around 0.1–1 *μ*m3 per nC of incident ions (Gierak, 2009). The second consequence is that for most applications the spatial extension of the phenomena induced by focused ion beam irradiation constitutes a major drawback.

In addition, there have been few reports of the fabrication of nano-structured materials, nano devices, and hierarchical nano-sized patterns with a 100 nm distance using a focused ion beam (FIB). Fabrication of graphene nanoribbons and graphene-based ultracapacitors were also reported recently. The above-discussed methods were followed by the twodimensional (2D) fabrication methods and required extensive efforts to achieve precise control. Hence, a novel three-dimensional (3D) nanoscale approach to the fabrication of a stack of graphene layers via FIB etching is proposed, through which a thin graphite flake can be etched in the *c*-axis direction (stack height with a few tens of nanometers). Also the main purpose of describing graphite and other BSCCO based superconducting nanoscale devices is that these layered structured materials have shown an excellent device structures

minimal feature size accessible by these techniques is typically limited to tens of

The patterning of samples using the FIB (focused ion beam) technique is a very popular technique in the field of inspection of integrated circuits and electronic devices manufactured by the semi-conductor industry or research laboratories. This is the case mainly for prototyping devices. The FIB technique allowing us to engrave materials at very low dimensions is a complement of usual lithographic techniques such as optical lithography. The main difference is that FIB allows direct patterning and therefore does not require an intermediate sensitive media or process (resist, metal deposited film, etching process). FIB allows 3D patterning of target materials using a finely focused pencil of ions having speeds of several hundreds of km s−1 at impact. Concerning the nature of the ions most existing metals can be used in FIB technology as pure elements or in the form of alloys,

Many device fabrication techniques based on electron beam lithography followed by reactive-ion etching (RIE), chemical methods, and evaporation using hard Si shadow masks, and including lithography-free fabrication, have been reported. The procedures, however, are complex and yield devices with dimensions of ~5 to 50 nm, which are restricted to simple geometries. RIE creates disordered edges, and the chemical methods produce irregular shapes with distributed flakes, which are not suitable for electronic-device

Practically, FIB patterning can be achieved either by local surface defect generation, by ion implantation or by local sputtering. These adjustments are obtained very easily by varying the locally deposited ion fluence with reference to the sensitivity of the target and to the selected FIB processing method (Gierak, 2009). The FIB milling involves two processes: 1) Sputtering*,* ions with high energy displace and remove atoms of substrate material, and the ions lose their energy as they go into the substrate; 2) Re-deposition, the displaced substrate atoms, that have gained energy from ions through energy transfer, go through similar

A focused gallium ion beam having an energy typically around 30 keV is scanned over the sample surface to create a pattern through topographical modification, deposition or sputtering. A first consequence is that, mainly because of the high ion doses required (~1018 ions cm−2) and of the limited beam particle intensity available in the probe, FIB etchingbased processes remain relatively slow. We may recall that for most materials, the material removal rate for 30 keV gallium ions is around 1–10 atoms per incident ion, corresponding to a machining rate of around 0.1–1 *μ*m3 per nC of incident ions (Gierak, 2009). The second consequence is that for most applications the spatial extension of the phenomena induced by

In addition, there have been few reports of the fabrication of nano-structured materials, nano devices, and hierarchical nano-sized patterns with a 100 nm distance using a focused ion beam (FIB). Fabrication of graphene nanoribbons and graphene-based ultracapacitors were also reported recently. The above-discussed methods were followed by the twodimensional (2D) fabrication methods and required extensive efforts to achieve precise control. Hence, a novel three-dimensional (3D) nanoscale approach to the fabrication of a stack of graphene layers via FIB etching is proposed, through which a thin graphite flake can be etched in the *c*-axis direction (stack height with a few tens of nanometers). Also the main purpose of describing graphite and other BSCCO based superconducting nanoscale devices is that these layered structured materials have shown an excellent device structures

process as ions, sputtering other atoms, taking their vacancy, or flying out.

focused ion beam irradiation constitutes a major drawback.

nanometers, thus novel procedures must be devised (Zhou, 2006).

although gallium (Ga+ ions) is preferred in most cases.

application.

during fabrication and their electrical transport characteristics were interesting which will be useful to future works.

#### **2.2.1 Nanoscale stack fabrication by focused-ion-beam**

Using an FIB, perfect stacks can be fabricated more easily along the c-axis in thin films and single-crystal whiskers. FIB 3D etching has been recognized as a well-known method for fabricating high-precision ultra-small devices, in which etching is a direct milling process that does not involve the use of any masking and process chemicals and that demonstrates a submicrometer resolution. Thus, these our proposal is focused on the fabrication of a nanoscale stack from the layered structured materials like thin graphite flake and BSCCO, via FIB 3D etching. The detailed schematic of fabrication process is shown in Fig. 3.

The 3D etching technique is followed by tilting the substrate stage up to 90° automatically for etching thin graphite flake. We have freedom to tilt the substrate stage up to 60° and rotate up to 360°. To achieve our goal, we used sample stage that itself inclined by 60° with respect to the direction of the ion beam (fig 3a). The lateral dimensions of the sample were 0.5×0.5 μm2. The in-plane area was defined by tilting the sample stage by 30° anticlockwise with respect to the ion beam and milling along the *ab*-plane.

Fig. 3. FIB 3-D fabrication process (a) Scheme of the inclined plane has an angle of 60° with ion beam (where we mount sample). (b) The initial orientation of sample and sample stage. (c) Sample stage titled by 30° anticlockwise with respect to ion beam and milling along abplane. (d) The sample stage rotated by an angle of 180° and also tilted by 60° anticlockwise with respect to ion beam and milled along the c-axis.

The in-plane etching process is shown in Fig. 3(a)–(c). The out of plane or the c-axis plane was fabricated by rotating the sample stage by an angle of 180°, then tilting by 60° anticlockwise with respect to the ion beam, and milling along the c-axis direction. The schematic diagram of the fabrication process for the side-plane is shown in Fig. 3(d). The dimensions of the side-plane was W=0.5 μm, L=0.5 μm, and H=200 nm. The c-axis height length (H) of the stack was set as 200 nm. An FIB image of fabricated stack is shown in Fig. 4 in which the schematic of stack arrangement (graphene layers with interlayer distance 0.34 nm) was also shown in the inset (top right) in Fig. 4. The vertical red arrow indicates the current flow direction through the stack.

#### **2.2.2 Transport characteristics of nanoscale graphite stacks**

The electrical transport characteristics (including *ρ-T* and *I-V*) can be performed for the fabricated stack using closed-cycle refrigerator systems (CKW-21, Sumitomo) at various temperatures from 25 to 300 K with the use of the Keithley 2182A nanovoltmeters and AC & DC current source (6221). The I-V characteristics of the fabricated stack are shown in Fig.4.

Fig. 4. FIB image of the nanoscale stack fabricated on a thin graphite flake along the c-axis height of 200 nm (image scale bar is 2 μm). Inset shows the schematic diagram of stack arrangement along the c-axis. (Venugopal et al, 2011). The vertical red arrow indicates the current flow direction through the nanoscale stack. I-V characteristics at various temperatures of the fabricated nanostack are also shown (right).

The FIB ion damage effect can be avoided if the device is fabricated at a 3D angle, in which the top layer of *ab*-plane will act as a masking layer and the ion beam is exactly perpendicular to the milling surface. The expected ion damage effect was simulated using the TRIM software (Ziegler, 1996) and the fabrication parameter of etching process for the 30 keV Ga+ ions was optimized. It was found from the simulation results that the depth of ion implantation is consistent with 10 nm. Majority (*>*95%) of the Ga+ ions are expected to be implanted within 10 nm of the side walls of stack surface, with a much smaller fraction, eventually stopping at as deep as 10 nm into the surface. Therefore, the proportion of the fabricated stack affected by ion beam damage is not very large, and it does not affect the quality of graphite devices in the *c*-axis direction.

The in-plane etching process is shown in Fig. 3(a)–(c). The out of plane or the c-axis plane was fabricated by rotating the sample stage by an angle of 180°, then tilting by 60° anticlockwise with respect to the ion beam, and milling along the c-axis direction. The schematic diagram of the fabrication process for the side-plane is shown in Fig. 3(d). The dimensions of the side-plane was W=0.5 μm, L=0.5 μm, and H=200 nm. The c-axis height length (H) of the stack was set as 200 nm. An FIB image of fabricated stack is shown in Fig. 4 in which the schematic of stack arrangement (graphene layers with interlayer distance 0.34 nm) was also shown in the inset (top right) in Fig. 4. The vertical red arrow indicates the

The electrical transport characteristics (including *ρ-T* and *I-V*) can be performed for the fabricated stack using closed-cycle refrigerator systems (CKW-21, Sumitomo) at various temperatures from 25 to 300 K with the use of the Keithley 2182A nanovoltmeters and AC & DC current source (6221). The I-V characteristics of the fabricated stack are shown in Fig.4.

 Fig. 4. FIB image of the nanoscale stack fabricated on a thin graphite flake along the c-axis height of 200 nm (image scale bar is 2 μm). Inset shows the schematic diagram of stack arrangement along the c-axis. (Venugopal et al, 2011). The vertical red arrow indicates the

The FIB ion damage effect can be avoided if the device is fabricated at a 3D angle, in which the top layer of *ab*-plane will act as a masking layer and the ion beam is exactly perpendicular to the milling surface. The expected ion damage effect was simulated using the TRIM software (Ziegler, 1996) and the fabrication parameter of etching process for the 30 keV Ga+ ions was optimized. It was found from the simulation results that the depth of ion implantation is consistent with 10 nm. Majority (*>*95%) of the Ga+ ions are expected to be implanted within 10 nm of the side walls of stack surface, with a much smaller fraction, eventually stopping at as deep as 10 nm into the surface. Therefore, the proportion of the fabricated stack affected by ion beam damage is not very large, and it does not affect the

current flow direction through the nanoscale stack. I-V characteristics at various

temperatures of the fabricated nanostack are also shown (right).

quality of graphite devices in the *c*-axis direction.

current flow direction through the stack.

**2.2.2 Transport characteristics of nanoscale graphite stacks** 

By varying in-plane area (*A*) and stack height (*H*), several stacked-junctions with the dimensions of W = 1 *μ*m, L = 1 *μ*m, and H = 0.1 *μ*m (denoted as *J4*) and W = 2 *μ*m, L = 1 *μ*m, and H = 0.3 *μ*m (denoted as *J5*) were fabricated. The electrical transport characteristics were performed for these stacks and compared their results. The current-voltage (*I-V*) characteristics of the nanostack with in-plane area (*A)* of 0.25 *µ*m2 (*J*2) at various temperatures, are presented in Fig. 4. The stack showed a nonlinear concave-like *I-V* characteristics at all studied temperatures (25, 50, 110, 200, 250 and 300 K). At 300 K, the stack resistance was found as 75 Ω. The stack resistance found increases when the temperature goes down to 25 K.

The electrical characteristics of nanostack (*J*2) were analyzed and compared with bigger junctions *J4* (1 × 1 × 0.1 *µ*m3) and *J5* (2 × 1 × 0.3 *µ*m3). From the data analysis, it is clear that the stack with larger height and reduced in-plane effective area (*A*) has shown higher resistance than the stack with larger in-plane area (*A*). The *I-V* characteristics of junctions *J*4 and *J*5 at different temperatures are shown in Fig. 5 (a) and (b) respectively. A typical *c*-axis transport characteristics similar to junction *J*2 was observed. However the nonlinear *I-V* characteristics were not observed at 300 K, but ohmic like-linear behavior is observed. When the temperature goes down, this behavior is turned into curve-like nonlinear characteristics.

Fig. 5. (a) I–V characteristics of a bigger stacked-junction with A of 1 µm2 (J4) at different temperature from 25 K to 300 K. (b) I–V characteristics of another bigger junction with A of 2 × 1 × 0.3 *µ*m3 (J5) at different temperature from 25 K to 300 K. Both the junctions show ohmic like behavior at 300 K; however the same behavior turned into nonlinear characteristics when the temperature goes down (Venugopal et al, 2011).

There is a significant overlap of *I-V* curves for temperatures 110, 75 and 25 K. For graphite stacks with *A* ≥ 1 *µ*m2, there was no nonlinear *I-V* characteristics observed at 300 K even at high biasing. With a decrease of the stack size down to 0.25 *µ*m2, the junction shows clear nonlinear concave-like *I-V* characteristics for both 300 K and 25 K. Since the fabricated stack contains multiple elementary junctions along the *c*-axis, the nonlinear concave-like tunneling characteristics appeared from the *I-V* characteristics (Venugopal et al, 2011).

#### **2.2.3 Temperature dependent resistivity of nanoscale graphite stack**

Fig 6 represents the ρ–T characteristics of stacked-junction (J2). The junction J2 shows a semiconducting behavior for T > 65 K and metallic characteristics for T < 65 K. Above 65 K, thermal excitation of carriers plays a major role in semiconducting temperature dependence. However below 65 K, the interlayer hopping conduction combined with scattering of carriers by phonons can be responsible for the metallic-like temperature dependence. The ρ– T characteristics along the *ab*-plane transport are shown as inset in Fig. 6. A well understood metallic behavior was observed. This behavior is well agreed with earlier observations on caxis characteristics of bulk graphite material (Matsubara, 1990).

An electron motion parallel to its plane is not affected by the stacking faults, however, but an electron motion in the *c*-axis direction is strongly impeded by the faults. The combined effects of impurity-assisted hopping, tunneling current, and the thermal excitation of the carriers on the plane of a stack play important roles in this temperature-dependent conduction mechanism in layered structured materials such as graphite.

Fig. 6. The resistivity–temperature (ρ-T) characteristics of nanostack which shows a clear c-axis characteristics of graphite. A well agreed curve fitting to experimental data is also shown. A clear metallic behavior is observed for *ab*-plane transport of bare graphite flake which is shown as inset. (Venugopal, 2011)

#### **2.3 FIB nano fabrication on superconducting devices**

Considering Bi-family as a layered structure material, there are three compounds in the Bifamily high-temperature superconductors, differing in the type of planar CuO2 layers; single-layered Bi2Sr2CuO6+<sup>δ</sup> (Bi-2201) single crystal, double-layered Bi2Sr2CaCu2O8+<sup>δ</sup> (Bi-2212) single crystal, and triple-layered Bi2Sr2Ca2Cu3O10+<sup>δ</sup> (Bi-2223) single crystal (Saini, 2010). This Bi-family material is a one of the famous emerging material for electron tunneling devices, such as intrinsic Josephson junctions (IJJ) in layered high-*Tc*  superconductors. The spacing of consecutive copper-oxide double planes in the most anisotropic cuprate superconductors is greater than the coherence length in the out-of-plane *c*-direction. When a current flows along the *c*-direction in such a material, it therefore flows through a series array of "intrinsic" Josephson junctions (IJJs) (Kleiner, 1992). These junctions and junction arrays are showing promise for a wide variety of applications, including as voltage standards and sub-mm-wave oscillators (Wang, 2001). For sub-micron intrinsic junctions, there is an additional range of potential applications exploiting the Coulomb blockade effect, when the *Ec* is charging energy Ec ≥ EJ, KBT, where EJ is the Josephson energy & *kBT* is thermal energy. These applications include electric-field sensors

thermal excitation of carriers plays a major role in semiconducting temperature dependence. However below 65 K, the interlayer hopping conduction combined with scattering of carriers by phonons can be responsible for the metallic-like temperature dependence. The ρ– T characteristics along the *ab*-plane transport are shown as inset in Fig. 6. A well understood metallic behavior was observed. This behavior is well agreed with earlier observations on c-

An electron motion parallel to its plane is not affected by the stacking faults, however, but an electron motion in the *c*-axis direction is strongly impeded by the faults. The combined effects of impurity-assisted hopping, tunneling current, and the thermal excitation of the carriers on the plane of a stack play important roles in this temperature-dependent

Fig. 6. The resistivity–temperature (ρ-T) characteristics of nanostack which shows a clear c-axis characteristics of graphite. A well agreed curve fitting to experimental data is also shown. A clear metallic behavior is observed for *ab*-plane transport of bare graphite flake

Considering Bi-family as a layered structure material, there are three compounds in the Bifamily high-temperature superconductors, differing in the type of planar CuO2 layers; single-layered Bi2Sr2CuO6+<sup>δ</sup> (Bi-2201) single crystal, double-layered Bi2Sr2CaCu2O8+<sup>δ</sup> (Bi-2212) single crystal, and triple-layered Bi2Sr2Ca2Cu3O10+<sup>δ</sup> (Bi-2223) single crystal (Saini, 2010). This Bi-family material is a one of the famous emerging material for electron tunneling devices, such as intrinsic Josephson junctions (IJJ) in layered high-*Tc*  superconductors. The spacing of consecutive copper-oxide double planes in the most anisotropic cuprate superconductors is greater than the coherence length in the out-of-plane *c*-direction. When a current flows along the *c*-direction in such a material, it therefore flows through a series array of "intrinsic" Josephson junctions (IJJs) (Kleiner, 1992). These junctions and junction arrays are showing promise for a wide variety of applications, including as voltage standards and sub-mm-wave oscillators (Wang, 2001). For sub-micron intrinsic junctions, there is an additional range of potential applications exploiting the Coulomb blockade effect, when the *Ec* is charging energy Ec ≥ EJ, KBT, where EJ is the Josephson energy & *kBT* is thermal energy. These applications include electric-field sensors

axis characteristics of bulk graphite material (Matsubara, 1990).

which is shown as inset. (Venugopal, 2011)

**2.3 FIB nano fabrication on superconducting devices** 

conduction mechanism in layered structured materials such as graphite.

and quantum current standards (Bylander, 2005). In long arrays of junctions, Ec is enhanced by electron-electron interactions (Likharev, 1989, 1995) by a factor [C/C0]1/2, where C is the junction capacitance and C0 is the stray capacitance to ground. The large ratio C/C0 ∼ 106 for intrinsic junctions makes them particularly suited to the applications involving Coulomb blockade effects. The features of the single Cooper-pair tunneling effect from the layered structure of Bi-family as well as for YBCO will also be discussed in detail.

Superconductivity is a phenomenon when the resistance of the material becomes zero and it expels all the magnetic field below a certain temperature usually at very low temperature. The phenomenon of superconductivity was discovered in 1911 by the Dutch physicist H. Kamerlingh Onnes. The quantum application of superconductivity was introduced in 1962. B. D. Josephson discovered a tunnel junction consists of two strips of superconductors separated by an insulator where the insulator is so thin that electrons can tunnel through it known as Josephson junction.

The schematic of different types of Josephson junctions are shown below in Fig.7. S stands for superconductor, S' for a superconductor above *Tc*, N for normal metal, Se for semiconductor, and I for an insulator.

Fig. 7. The schematics of different types of superconducting devices.

The term high-temperature superconductor was first introduced in 1986 to designate the new family of cuprate-perovskite ceramic materials discovered by Johannes George Bednorz and Karl Alexander Müller [J. G. Bednorz, K. A. Mueller (1986) "Possible high *T*<sup>C</sup> superconductivity in the Ba-La-Cu-O system", *Zeitschrift für Physik B* 64 (2) 189–193 doi:10.1007/BF01303701] for which they won the Nobel Prize in Physics in the following year. Their discovery of the first high-temperature superconductor, LaBaCuO, with a transition temperature of 30 K, generated great excitement. In 1988, BSCCO (Bi2Sr2Ca*n*-1Cu*n*O2*n*+4+x, with *n*=2 being the most commonly studied compound, though *n*=1 and *n*=3 have also received significant attention) as a new class of superconductor was discovered by Maeda and coworkers [H. Maeda, Y. Tanaka, M. Fukutumi, and T. Asano (1988) "A New High-*T*c Oxide Superconductor without a Rare Earth Element" *Jpn. J. Appl. Phys.* 27 (2) L209– L210. doi:10.1143/JJAP.27.L209.] at the National Research Institute for Metals in Japan, though at the time they were unable to determine its precise composition and structure. The discovery of these high temperature superconductors gave a path for the application of the superconductivity at higher temperature.

#### **2.3.1 FIB nanomachining of Intrinsic Josephson Junctions (IJJs) on BSCCO and Y123/Pr123 multilayered thin films**

Many fabrication methods based on high-resolution patterning have been applied to develop high-*Tc* superconducting devices. Very small structures are needed in the fabrication of tunneling devices, such as intrinsic Josephson junctions (IJJ) in layered high *Tc*  superconductors Bi2Sr2CaCu2O8+δ (Bi-2212). Perfect stacks are more easily obtained in *c*-axis high-quality thin films than in *a*-axis films or single-crystal whiskers. However, the IJJ fabrication process using *c*-axis thin films and single crystals requires intricate processes and limits the junction size in mesa structures.

As per previous reports, the fabrication of IJJs by the focused ion beam (FIB) etching method using single-crystal whiskers as a base material requires some complicated processes, including turning over of the sample. As an alternative approach, in this chapter, a threedimensional IJJ fabrication method is presented using *c*-axis thin films. The fabrication steps using *c*-axis single crystal are also simplified by the *in situ* process. Here, the 3D FIB etching methods using YBCO thin films and Bi-2212 single-crystal whiskers were described as examples with a successive decrease of their in-plane area, *S*, down to a submicron scale. Also, there was a possibility to identify the features of the single Cooper-pair tunneling effect from the layered structure of Bi-2212 with very narrow interval between layers.

FIB image of a submicron stack fabricated on Bi-2212 single crystal whiskers with in-plane area of 0.4 µm × 0.4 µm and schematic of the IJJs configuration are shown in Fig. 8, in which FIB fabrication procedures followed as described in section 2.1.2.

Fig. 8. FIB image of a submicron stack fabricated on Bi-2212 single crystal whisker. The red color circular part shown in stack contains many IJJs.

FIB image of submicron stack (scale bar of 1 µm) and schematic of the Josephson junctions configuration in the submicron stack fabricated on *a*-axis oriented YBa2Cu3O7/PrBa2Cu3O7 multi layered thin films are shown in Fig.9. The arrow indicates the direction of current to observe the effect of Josephson junctions. The axial direction of thin film is shown in the expended view.

Many fabrication methods based on high-resolution patterning have been applied to develop high-*Tc* superconducting devices. Very small structures are needed in the fabrication of tunneling devices, such as intrinsic Josephson junctions (IJJ) in layered high *Tc*  superconductors Bi2Sr2CaCu2O8+δ (Bi-2212). Perfect stacks are more easily obtained in *c*-axis high-quality thin films than in *a*-axis films or single-crystal whiskers. However, the IJJ fabrication process using *c*-axis thin films and single crystals requires intricate processes and

As per previous reports, the fabrication of IJJs by the focused ion beam (FIB) etching method using single-crystal whiskers as a base material requires some complicated processes, including turning over of the sample. As an alternative approach, in this chapter, a threedimensional IJJ fabrication method is presented using *c*-axis thin films. The fabrication steps using *c*-axis single crystal are also simplified by the *in situ* process. Here, the 3D FIB etching methods using YBCO thin films and Bi-2212 single-crystal whiskers were described as examples with a successive decrease of their in-plane area, *S*, down to a submicron scale. Also, there was a possibility to identify the features of the single Cooper-pair tunneling effect from the layered structure of Bi-2212 with very narrow interval between layers. FIB image of a submicron stack fabricated on Bi-2212 single crystal whiskers with in-plane area of 0.4 µm × 0.4 µm and schematic of the IJJs configuration are shown in Fig. 8, in which

Fig. 8. FIB image of a submicron stack fabricated on Bi-2212 single crystal whisker. The red

FIB image of submicron stack (scale bar of 1 µm) and schematic of the Josephson junctions configuration in the submicron stack fabricated on *a*-axis oriented YBa2Cu3O7/PrBa2Cu3O7 multi layered thin films are shown in Fig.9. The arrow indicates the direction of current to observe the effect of Josephson junctions. The axial direction of thin film is shown in the

**2.3.1 FIB nanomachining of Intrinsic Josephson Junctions (IJJs) on BSCCO and** 

**Y123/Pr123 multilayered thin films** 

limits the junction size in mesa structures.

FIB fabrication procedures followed as described in section 2.1.2.

color circular part shown in stack contains many IJJs.

expended view.

Fig. 9. FIB image of a submicron stack fabricated on *a*-axis oriented Y123/Pr123 multi layered thin films.

#### **2.3.2 Electrical transport characteristics of Josephson junctions fabricated on multi layered thin films of Y123/Pr123**

Fig. 10 represents *R-T* characteristics of the Josephson junctions fabricated on multi layered thin films of Y123/Pr123 which shows *T*c about 83 K.

Fig. 10. *R-T* characteristics of the device show *T*c about 83 K.

*I-V* characteristics of the same device were studied without microwave irradiation at different temperature of 10, 20, and 30 K, shown in Fig. 11. As temperature decreases, the critical current density of superconducting device is increases gradually.

The above discussed nanomachining/milling techniques followed by focused ion beam 3-D technique shall be applicable to other layered-structured materials rather than graphite flake, BSCCO, YBCO and multilayered thin films, etc,. This may have great potential in future nanodevice development and applications.

Fig. 11. *I-V* characteristics of the device without microwave irradiation at different temperature of 10, 20, and 30 K. The critical current density *J*c about 2.2 X 105 *A*/cm2 is measured at 20 K.

#### **2.4 Future advances**

In the future, micromachining is destined to improve upon its shortcomings, as the various micromachining processes become accurate, reliable, versatile and cost-effective. In India, BARC has established premier micromachining and nano-finishing facilities along with state-of-the art metrology systems. On the other hand, IIT Bombay has also taken a lead in establishing tool-based micromachining facilities. Even at South Korea, the technology towards nanomachining becomes popular nowadays and the active research is now under progress through which an interesting studies may be explored in near future.

FIB technology is still relatively young compared with other semiconductor fabrication processes. One of the major challenges for all of the microfabrication and nanofabrication technologies is to downscale the feature size while maintaining a high throughput. To increase the throughput and the ability to be used in production, the milling rate of the existing FIB milling systems has to be improved. A variable-diameter beam system should be developed to provide multi-resolution milling to cope with different accuracy or tolerance requirements. It is ideal that the beam diameter can be continuously changed in situ. (Tseng, 2004). This type of system has been available for many macro-scale fabrication processes. With this system, a larger beam can be used for roughing 'cut' (milling) to increase the milling rate in regions where only lower resolution is needed. The advantages to use a heavy-duty two-lens system with improved automation should be examined with the goal to develop a system for limited production usage first. Once the high-performance FIB system is used in production, it can be a vital candidate to become the mainstream tool for the future microtechnology and nanotechnology industry.

With an increasing awareness about the advantages of manufacturing micro-components indigenously instead of importing at high costs, the researchers and industrialists are in need of the knowledge of micromachining technology.

#### **3. Conclusion**

14 Micromachining Techniques for Fabrication of Micro and Nano Structures

Fig. 11. *I-V* characteristics of the device without microwave irradiation at different temperature of 10, 20, and 30 K. The critical current density *J*c about 2.2 X 105 *A*/cm2 is

progress through which an interesting studies may be explored in near future.

for the future microtechnology and nanotechnology industry.

need of the knowledge of micromachining technology.

In the future, micromachining is destined to improve upon its shortcomings, as the various micromachining processes become accurate, reliable, versatile and cost-effective. In India, BARC has established premier micromachining and nano-finishing facilities along with state-of-the art metrology systems. On the other hand, IIT Bombay has also taken a lead in establishing tool-based micromachining facilities. Even at South Korea, the technology towards nanomachining becomes popular nowadays and the active research is now under

FIB technology is still relatively young compared with other semiconductor fabrication processes. One of the major challenges for all of the microfabrication and nanofabrication technologies is to downscale the feature size while maintaining a high throughput. To increase the throughput and the ability to be used in production, the milling rate of the existing FIB milling systems has to be improved. A variable-diameter beam system should be developed to provide multi-resolution milling to cope with different accuracy or tolerance requirements. It is ideal that the beam diameter can be continuously changed in situ. (Tseng, 2004). This type of system has been available for many macro-scale fabrication processes. With this system, a larger beam can be used for roughing 'cut' (milling) to increase the milling rate in regions where only lower resolution is needed. The advantages to use a heavy-duty two-lens system with improved automation should be examined with the goal to develop a system for limited production usage first. Once the high-performance FIB system is used in production, it can be a vital candidate to become the mainstream tool

With an increasing awareness about the advantages of manufacturing micro-components indigenously instead of importing at high costs, the researchers and industrialists are in

measured at 20 K.

**2.4 Future advances** 

In conclusion, the focused ion beam based nanomachining have been discussed in detail for the layered structured materials, BSCCO superconducting devices, YBCO based thin film devices, and a-axis oriented Y123/Pr123 multi layered thin film devices. The development of focused ion beam technology based nanomachining is one amongst many examples on how research results may have found unexpected applications in totally different application areas. This is particularly true for the FIB technology development itself that has benefited from all the previously made advances in field emission physics, charged particle optics theory or modelling and in fundamental instrumentation or applied metrology. All these advances were very quickly and efficiently integrated into FIB instruments, so that in less than one decade FIB instruments have moved out from some specialist laboratories to enter almost every modern laboratory, research institute or processing environment. This is also true for the semiconductor industry that has been almost immediately applying FIB systems for device inspection failure analysis and reverse engineering with roaring success. The FIB processing methods which we discussed in this chapter, appear now to be well suited and very promising for several diverse nanotechnology applications, and may be of major interest for future applications to spin-electronics, nano-electronics, nano-optics or

#### **4. Acknowledgment**

nanomagnetism.

This research was supported by National Research Foundation of Korea Grant under contract numbers 2009-0087091 and 2011-0015829 through the Human Resource Training Project for Regional Innovation. A part of this research was also supported by the 2012 Jeju Sea Grant College Program funded by Ministry of Land, Transport and Maritime Affairs, Republic of Korea

#### **5. References**


## **Miniature Engineered Tapered Fiber Tip Devices by Focused Ion Beam Micromachining**

Fei Xu, Jun-long Kou, Yan-qing Lu and Wei Hu

*College of Engineering and Applied Sciences and National Laboratory of Solid State Mi-crostructures, Nanjing University, Nanjing, P. R. China* 

#### **1. Introduction**

16 Micromachining Techniques for Fabrication of Micro and Nano Structures

Latyshev, Y. I. (1997). Intrinsic Josephson Effects on Stacks Fabricated from High Quality BSCCO 2212 Single Crystal Whiskers. *Physica C.,* Vol.293, pp. 174-180. Langford, R. M. (2001). Preparation of Site Specific Transmission Electron Microscopy Plan-

Likharev, K. K. (1989). Single-electron Tunnel Junction Array: An Electrostatic Anolog of the

Likharev, K. K. (1995). Electron-electron Interaction in Linear Arays of Small Tunnel

Matsubara, K. (1990). Electrical Resistance in c-direction of Graphite. *Phys. Rev. B.,* Vol.41,

Melnagilis, J. (1998). A Review of Ion Projection Lithography. *J. Vac. Sci. Technol. B.,* Vol.16,

Odagawa, A. (1998). Characteristics of Intrinsic Josephson Junctions in Thin Stack on Bi-2223

Saini, S. (2010). Characterization of Submicron Sized Josephson Junction Fabricated in a

Seliger, R. L. (1979). High‐resolution, Ion‐beam Processes for Microstructure Fabrication. *J.* 

Tseng, A. A. (2004). Recent Developments in Micromilling using Focused Ion Beam

Venugopal, G. (2011a). Fabrication of Nanoscale Three-dimensional Graphite Stacked-

Venugopal, G. (2011b). Temperature Dependence of Transport Anisotrophy of Planar-type

Venugopal, G. (2011c). Fabrication and Characteristics of Submicron Stacked-Junctions on Thin Graphite Flakes. *J. Nanosci. Nanotechnol*. Vol.11, No.2, pp. 1405-1408. Wang, H. B. (2001). Terahertz Responses of Intrinsic Josephson Junctions in High Tc

Ziegler, J. F; Biersack, J. P. & Littmark, U. (1996). *The Stopping and Range of Ions in Solids,* 

Zhou, J.; Yang, G. (2006). *Proceedings of the 7th ICFDM 2006 International Conference on* 

*Frontiers of Design and Manufacturing,* pp. 453-458, Guangzhou, China, June 19-22,

Bi2Sr2Ca2Cu3O10+*<sup>δ</sup>* (Bi-2223) Single Crystal Whisker. *J. Supercond. Nov. Magn.* Vol.23,

junctions by Focused-ion-beam and Observation of Anomalous Transport

Graphite Nano-structures Fabricated by Focused Ion Beam. *J. Nanosci. Nanotechnol.*

No.755. doi:10.1116/1.1371317.

Junctions. *Appl. Phys. Lett.,* Vol.67, pp. 3037-3039.

Thin Films. *Jpn. J. Appl. Phys.,* Vol.37, No.1, pp. 486-491.

Technology. *J. Micromech. Microengg.,* Vol.14, pp. R15-R34.

Superconductors. *Phys. Rev. Lett.,* Vol.87, pp. 107002-107005.

*Vac. Sci. Technol. B.,* Vol.16, No.6, pp.1610-1612.

Characteristics. *Carbon,* Vol.49, No.8, pp. 2766-2772.

DOI: 10.1109/20.92566.

pp. 969-974.

pp. 811-813.

No.3, pp. 927-957.

Vol.11, No.1, pp. 296-300.

Pergamon, New York.

2006

view Specimens using a Focused Ion Beam System. *J. Vac. Sci. Technol. B.,* Vol.19,

Josephson Transmission Line. *IEEE Trans. Mag.,* Vol.25, pp. 1436.

Optical fibers have been the basis of the modern information technology since Kao and Hockham proposed glass waveguides as a practical medium for communication in 1965. A lot of different optical fiber active/passive devices including couplers, interferometers, gratings, resonators and amplifiers have been widely employed for applications on telecommunications and sensing networks (Agrawal, 2002). For a number of applications, it is important to reduce the device's size. Small size is often attractive for particular sensing applications because of some benefits such as fast response to detecting small objection with little perturbation on the object being measured. There are two steps to obtain fiber devices as small as possible. First, it is to taper or etch the fiber and reduce its diameter. A subwavelength-scale microfiber is the basic element of miniature fiber devices and subsystems (Tong et al., 2003; Brambilla et al., 2004, 2005, 2010). The second is to engineer the microfiber to realize miniature version of conventional fiber devices. There are various fabrication methods to engineer the microfiber, such as CO2 laser, femtosecond (fs) laser, HF acid etching, arc splicing and focused ion beam (FIB). Most of these techniques have the difficulties in carving the microfiber freely because of the resolution. The latest progress in FIB technique has opened a new widow for ultra-small size fiber devices. So far, FIB is the most flexible and powerful tool for patterning, cross-sectioning or functionalizing a subwavelength circular microfiber due to its small and controllable spot size and high beam current density.

FIB systems have been produced commercially for approximately thirty years, primarily for large semiconductor manufacturers. FIB systems operate in a similar fashion to a scanning electron microscope (SEM) except, rather than a beam of electrons and as the name implies, FIB systems use a finely focused beam of ions that can be operated at low beam currents for imaging or high beam currents for site specific sputtering or milling (http://en.wikipedia.org/wiki/Focused\_ion\_beam). The fine and controllable ion spot size and high beam current density are perfect for micro- and nano-fabrications with high spatial resolution (~ 10 nm). As a result, FIB has recently become a popular candidate for fabricating high-quality micro-devices or high-precision microstructures. Originally, FIB processing was used for mask repair (Liang et al., 2000), integrated circuit chip repair/modification (Liu et al., 2006), cross-sectional imaging of critical parts of semiconductor devices and sample preparation for transmission electron microscopy (Daniel et al., 1998; Hopman et al., 2008; Jeon et al., 2010). Besides these applications, FIB milling can also be used to assist carbon nanotube growth and manipulation (Hofmann et al., 2005; Deng et al., 2006), pattern magnetic data storage media (Terris et al., 2007) and structure hard-to-etch materials like SiC or LiNbO3. In the field of optoelectronics, there have been extensively studies toward utilizing the FIB as a machining tool to fabricate planar micro-optical components with low surface roughness for integrated optical circuits, for example, the end facet mirrors, ring resonators, gratings and photonic crystals (Hopman et al., 2008). Obviously, FIB processing can and in fact has been widely applied to fabricate microfiber based devices to reduce the size of fiber devices as much as possible.

In this chapter, we will review several kinds of ultra-small engineered tapered fiber tip (TFT) devices including interferometers and gratings by FIB micromachining and their characteristics and sensing applications.

#### **2. Fabrication and measurement**

Standard optical TFT is an optical microfiber with only one output or input end and a taper transition. The taper transition is connected to untapered fiber at the extremities which can easily be connected to other fiber optic components. The taper is etched or pulled from a standard single mode fiber when heated by a CO2 laser, electrical microheater or a small flame. Since the TFT is for analyte detecting rather than launching the light, it should be short enough in order to be rigid. However, too short and sharp shape results in high losses due to the poor 'adiabaticity' of the taper profile which couples light to lossy unbound modes (Love et al., 1991). During the last decade, much work has been carried out to study and optimize TFT profiles for telecom devices. Technology development allows manufacturing tapers with diameters well below 100 nm and it is possible to tailor the taper shape to an ideal profile (Brambilla & Xu, 2007). The quickest and simplest way to manufacture short TFT relies on using a commercially available pipette puller. This method is often used to manufacture fibre tips for optical tweezers and scanning near-field optical microscopy (SNOM) tips. In this chapter, we make TFTs using a commercial pipette puller (model P-2000, Sutter Instrument). The P-2000 is a microprocessor-controlled CO2 laserbased micropipette puller. The bare fibre is held on two puller stages. The P-2000 can also be used to pull tubes and optical fibres to extremely small diameters. The pipette puller has five parameters which can be adjusted to achieve the wanted profile. The fabrication process is simple, convenient and extremely fast, which usually takes less than 0.3 second. The obtained TFT is then checked under a high-magnification optical microscope. Figure 1 shows a microscope image of a typical TFT with a sharp profile.

Fig. 1. Microscope image of a typical TFT, five photographs separated by four dashed vertical lines are used to show the whole profile of the TFT. The black arrow indicates the milling location (Kou et al., 2010b). Reprinted with permission. Copyright 2010 Optical Society of America

The TFT is then coated with a thin metal layer such as aluminium (Al) or gold (Au, for exciting surface waves discussed in Section 4.2). The coating thickness is around 30 ~ 150 nm and the metal is deposited on only one side of the taper. The metal Al is used as a conductive layer to prevent gallium ion accumulation in the FIB micromachining process. Then, the Al-coated TFT is placed stably in the FIB machining chamber (Strata FIB 201, FEI Company, Ga ions) using conductive copper tape. We generally use a 30.0 kV gallium ion beam with current 60 ~ 300 pA. This enables us to make structures with high accuracy and sharp end-faces. We mill the structures from the taper end with small diameter to that with bigger diameter, because the milled part becomes non-conductive when the metal is removed by the beam. The total micro-machining process takes about 15 ~ 30 minutes depending on the size of the machined structures. Finally, the TFT is immersed in hydrochloric acid for about 15 ~ 30 minutes to totally remove the Al layer before cleaned with deionized water. In our experiment, the cavity or grating is made from a two-step process. Because there are some remains adhering onto the surfaces of the cavity after the first milling step, a second step under the same or smaller beam current is used to improve the surface smoothness.

Fig. 2. Experimental setup of an FPMI.

18 Micromachining Techniques for Fabrication of Micro and Nano Structures

semiconductor devices and sample preparation for transmission electron microscopy (Daniel et al., 1998; Hopman et al., 2008; Jeon et al., 2010). Besides these applications, FIB milling can also be used to assist carbon nanotube growth and manipulation (Hofmann et al., 2005; Deng et al., 2006), pattern magnetic data storage media (Terris et al., 2007) and structure hard-to-etch materials like SiC or LiNbO3. In the field of optoelectronics, there have been extensively studies toward utilizing the FIB as a machining tool to fabricate planar micro-optical components with low surface roughness for integrated optical circuits, for example, the end facet mirrors, ring resonators, gratings and photonic crystals (Hopman et al., 2008). Obviously, FIB processing can and in fact has been widely applied to fabricate

In this chapter, we will review several kinds of ultra-small engineered tapered fiber tip (TFT) devices including interferometers and gratings by FIB micromachining and their

Standard optical TFT is an optical microfiber with only one output or input end and a taper transition. The taper transition is connected to untapered fiber at the extremities which can easily be connected to other fiber optic components. The taper is etched or pulled from a standard single mode fiber when heated by a CO2 laser, electrical microheater or a small flame. Since the TFT is for analyte detecting rather than launching the light, it should be short enough in order to be rigid. However, too short and sharp shape results in high losses due to the poor 'adiabaticity' of the taper profile which couples light to lossy unbound modes (Love et al., 1991). During the last decade, much work has been carried out to study and optimize TFT profiles for telecom devices. Technology development allows manufacturing tapers with diameters well below 100 nm and it is possible to tailor the taper shape to an ideal profile (Brambilla & Xu, 2007). The quickest and simplest way to manufacture short TFT relies on using a commercially available pipette puller. This method is often used to manufacture fibre tips for optical tweezers and scanning near-field optical microscopy (SNOM) tips. In this chapter, we make TFTs using a commercial pipette puller (model P-2000, Sutter Instrument). The P-2000 is a microprocessor-controlled CO2 laserbased micropipette puller. The bare fibre is held on two puller stages. The P-2000 can also be used to pull tubes and optical fibres to extremely small diameters. The pipette puller has five parameters which can be adjusted to achieve the wanted profile. The fabrication process is simple, convenient and extremely fast, which usually takes less than 0.3 second. The obtained TFT is then checked under a high-magnification optical microscope. Figure 1

microfiber based devices to reduce the size of fiber devices as much as possible.

shows a microscope image of a typical TFT with a sharp profile.

Fig. 1. Microscope image of a typical TFT, five photographs separated by four dashed vertical lines are used to show the whole profile of the TFT. The black arrow indicates the milling location (Kou et al., 2010b). Reprinted with permission. Copyright 2010 Optical

characteristics and sensing applications.

**2. Fabrication and measurement** 

Society of America

125 μm

In this chapter, we mainly consider the reflected signals. The reflective spectral response of these TFT based devices are measured with a broadband source (1525 ~ 1610 nm) and an Ando AQ6317B optical spectrum analyzer (OSA) through a circulator, as shown in Fig. 2. The TFTs before milling display an ignorable reflection of less than - 100 dB over the whole broadband spectrum.

#### **3. FIB machined micro-cavity TFT interferometers**

Optical fiber interferometers have been extensively used in various sensing applications due to its advantages of versatility, linear response and relatively simple structure. In the past decades, a lot of efforts have been made to develop intrinsic and extrinsic interferometers, especially the micro-cavity Fabry-Perot interferometers (MCFPIs). MCFPIs with tens-ofmicrometer-length cavity are attractive because of the small size, large free spectrum range (FSR) and high sensitivity. The cavity can be assembled by splicing two single mode fibers (SMFs) to a hollow-core fiber (Sirkis et al., 1993), inserting a silica SMF and a multi-mode fiber into a glass capillary (Bhatia et al., 1996), or splicing a SMF and an index-guiding photonic crystal fiber together (Villatoro et al., 2009). Although many progresses have been made, people are still pursuing new micro cavity fabrication techniques to improve the cavity length precision, structure accuracy and the process repeatability. Femtosecond laser technology thus was proposed recently showing great success in micromachining fiber devices. MCFPIs can be quickly fabricated by milling a small-open hole in a SMF for liquid and gas sensing (Rao et al., 2007). However, even the fs-laser machined MCFPIs still show low fringe visibility of several dBs in liquids due to the rugged surfaces inside the cavity; what's more, it is difficult to focus the laser spot to a sub-wavelength scale due to the diffraction limit. Thus the micromachining accuracy is limited and the size of the microcavity is large (tens of micrometers). The latest progress in FIB technique opens a new widow of opportunity for ultra-small size cavity (Kou et al., 2010a, 2010b). Microcavities with nanometer-scale accuracy in a subwavelength microfiber could be fabricated by FIB, which is relatively difficult for fs laser approach. There are several typical geometries which can be realized by FIB machined-TFT as shown in Fig. 3. Among them, an open-notch in one side is the most preferred and easiest to be fabricated.

Fig. 3. Illustration of several typical geometries which can be realized by FIB machining, (a) a side open notch, (b) a hole in the middle and (c) a hole in the tip end and parallel to the fiber axis.

For geometry (a), due to the low reflectivity of the air-glass interfaces, multiple reflections have negligible contributions to the optical interference. However, a TFT consists of a SMF and a MMF in nature, without splicing. It may hold both the original single core mode and the multi-modes in the cladding at different positions. As shown in Fig. 4, we only consider two reflections I1 and I2 at the two end-faces, respectively. The fundamental LP01 mode can be coupled to high-order LP0m mode in the taper transition or be excited to high-order LP0m mode at the end-faces. I1 or I2 possibly includes LP01 or LP0m mode (Kou et al., 2010b). We also break the cavity and measure the reflection I1 at end-face 1; flat reflective spectrum without obvious interference fringes is observed. Accordingly, a reasonable assumption is to consider only one dominated mode in I1 (LP01 mode) and I2 (LP01 or LP0m mode excited when inputting I2 into end-face 1) (Kou et al., 2010b). We call this kind of device as a hybrid FP modal interferometer (FPMI). The interference spectrum can be modelled using the following two-beam optical interference equation (Kou et al., 2010b):

$$I = I\_1 + I\_2 + 2\sqrt{I\_1 I\_2} \cos(\delta + \varphi\_0) \tag{1}$$

The phase difference between two modes in I1 and I2 is (Kou et al., 2010b)

Fig. 4. Illustration of the FPMI. I1 and I2 are the reflections at end-face 1 and end-face 2 respectively; Lc is the length of the cavity. When I2 enters end-face 1, the fundamental mode is possible to be excited to a higher-order mode (Kou et al., 2010b). Reprinted with permission. Copyright 2010 Optical Society of America

$$\begin{cases} \delta = \delta\_1 + q \delta\_2 = (2\pi / \lambda)(\Delta\_1 + q \Delta\_2) \\ \Delta\_1 = 2n\_c L\_c \\ \Delta\_2 = \int (n\_2(r) - n\_i(r))dz(r) \end{cases} \tag{2}$$

and FSR is (Kou et al., 2010b)

20 Micromachining Techniques for Fabrication of Micro and Nano Structures

technology thus was proposed recently showing great success in micromachining fiber devices. MCFPIs can be quickly fabricated by milling a small-open hole in a SMF for liquid and gas sensing (Rao et al., 2007). However, even the fs-laser machined MCFPIs still show low fringe visibility of several dBs in liquids due to the rugged surfaces inside the cavity; what's more, it is difficult to focus the laser spot to a sub-wavelength scale due to the diffraction limit. Thus the micromachining accuracy is limited and the size of the microcavity is large (tens of micrometers). The latest progress in FIB technique opens a new widow of opportunity for ultra-small size cavity (Kou et al., 2010a, 2010b). Microcavities with nanometer-scale accuracy in a subwavelength microfiber could be fabricated by FIB, which is relatively difficult for fs laser approach. There are several typical geometries which can be realized by FIB machined-TFT as shown in Fig. 3. Among them, an open-notch in one

Fig. 3. Illustration of several typical geometries which can be realized by FIB machining, (a) a side open notch, (b) a hole in the middle and (c) a hole in the tip end and parallel to the

(a) (b) (c)

following two-beam optical interference equation (Kou et al., 2010b):

The phase difference between two modes in I1 and I2 is (Kou et al., 2010b)

For geometry (a), due to the low reflectivity of the air-glass interfaces, multiple reflections have negligible contributions to the optical interference. However, a TFT consists of a SMF and a MMF in nature, without splicing. It may hold both the original single core mode and the multi-modes in the cladding at different positions. As shown in Fig. 4, we only consider two reflections I1 and I2 at the two end-faces, respectively. The fundamental LP01 mode can be coupled to high-order LP0m mode in the taper transition or be excited to high-order LP0m mode at the end-faces. I1 or I2 possibly includes LP01 or LP0m mode (Kou et al., 2010b). We also break the cavity and measure the reflection I1 at end-face 1; flat reflective spectrum without obvious interference fringes is observed. Accordingly, a reasonable assumption is to consider only one dominated mode in I1 (LP01 mode) and I2 (LP01 or LP0m mode excited when inputting I2 into end-face 1) (Kou et al., 2010b). We call this kind of device as a hybrid FP modal interferometer (FPMI). The interference spectrum can be modelled using the

1 2 12 <sup>0</sup> *I I I II* 2 cos( )

(1)

side is the most preferred and easiest to be fabricated.

fiber axis.

$$\text{FSR} = \text{2n} \lambda / \delta \tag{3}$$

where q = 0 (for LP01 in I2) and 1 for (LP0m in I2); ∆1 (δ1) and ∆2 (δ2) are the optical path length difference (the phase difference) owing to the micro-cavity and the modal difference in the taper transition, respectively; n1(r) and n2(r) are the effective index of LP01 and LP0m modes, respectively, functions of local radius r(z) of the TFT at position z, which can be calculated by three-layer model of finite cladding step-profile fiber with the TFT profile r(z) which can be obtained from the microscope figure of the TFT (Kou et al., 2010b).

#### **3.1 FIB machined FPMI for temperature sensing**

The FPMI can be applied as a high-temperature sensor. Its extremely small size and especially unique structure offer great potentials for fast-response high temperature sensing particularly in small and harsh area with high temperature gradient, such as micro-flame and high temperature gas-phase/liquid-phase flow in microfluidics channel. Figure 5 shows an SEM picture of a FPMI with a micro-notch cavity from the side view and cross section after cleaving the TFT at the cavity. The end-face is very sharp and smooth. The cavity is 4.4 μm long and 5 μm high, located at the position with the local radius r = 4.6 μm.

Fig. 5. SEM image (a) of the micro-notch cavity from the side view: three arrows show the edges of the cavity at the fiber tip, (b) of the cross section with the fiber tip cleaved at the position indicated in (a) by a dash line (Kou et al., 2010b). Reprinted with permission. Copyright 2010 Optical Society of America

The reflective spectral response of this FPMI device is measured with the setup as shown in Fig. 2. The TFT without a cavity displays an ignorable reflection of less than - 100 dB over the whole broadband spectrum. Hence, the detected signal is the light reflected only at the two end-faces of the micro-cavity, and the reflection at the tip end is negligible. The interference spectra of the FPMI device at room temperatures (19 °C) are shown in Fig. 6. The spectra indicates a free spectral range (FSR) of ~ 11 nm and a fringe visibility of ~ 11 dB around 1550 nm, which is larger than some other MCFPI sensors (Choi et al., 2008), and enough for sensing application. δ1 is ~ 12π and δ2 ~ 295π for LP03 mode, and FSR ~ 10 nm, in good agreement with what we obtain in the experiment. In our calculation, λ = 1530 nm, Lc = 4.4 μm and nc = 1 (Kou et al., 2010b).

Fig. 6. Interference spectra of the FPMI device in air at different temperatures (Kou et al., 2010b) Reprinted with permission. Copyright 2010 Optical Society of America

We characterize the thermal response of the FTMI device by heating it up in a micro-furnace (FIBHEAT200, Micropyretics Heaters International Inc.) and temperature ranging from

Fig. 5. SEM image (a) of the micro-notch cavity from the side view: three arrows show the edges of the cavity at the fiber tip, (b) of the cross section with the fiber tip cleaved at the position indicated in (a) by a dash line (Kou et al., 2010b). Reprinted with permission.

The reflective spectral response of this FPMI device is measured with the setup as shown in Fig. 2. The TFT without a cavity displays an ignorable reflection of less than - 100 dB over the whole broadband spectrum. Hence, the detected signal is the light reflected only at the two end-faces of the micro-cavity, and the reflection at the tip end is negligible. The interference spectra of the FPMI device at room temperatures (19 °C) are shown in Fig. 6. The spectra indicates a free spectral range (FSR) of ~ 11 nm and a fringe visibility of ~ 11 dB around 1550 nm, which is larger than some other MCFPI sensors (Choi et al., 2008), and enough for sensing application. δ1 is ~ 12π and δ2 ~ 295π for LP03 mode, and FSR ~ 10 nm, in good agreement with what we obtain in the experiment. In our calculation, λ = 1530 nm, Lc

1540 1560 1580 1600

Fig. 6. Interference spectra of the FPMI device in air at different temperatures (Kou et al.,

We characterize the thermal response of the FTMI device by heating it up in a micro-furnace (FIBHEAT200, Micropyretics Heaters International Inc.) and temperature ranging from

2010b) Reprinted with permission. Copyright 2010 Optical Society of America

Wavelength (nm)

19 C 305 C 520 C

Copyright 2010 Optical Society of America

**(a) (b)** 

= 4.4 μm and nc = 1 (Kou et al., 2010b).



Reflection (dB)


room temperature (19 °C) to 520 °C is measured by a thermocouple (TES-1310, Type K, TES Electrical Electronic Corp.). The spectrum and temperature were recorded when both of them are stable for several minutes (Kou et al., 2010b).

The temperature sensitivity ST is defined as the interference wavelength shift divided by the corresponding temperature change. ST depends on temperature through the thermal expansion and/or thermo-optics effect (Choi et al., 2008; Kou et al., 2010b):

$$\begin{cases} S\_{\tau} = \frac{d\mathcal{A}}{dT} = \frac{2\pi}{\delta} \left( \frac{d\Delta\_{1}}{dT} + \frac{d\Delta\_{2}}{dT} \right) = \frac{2\pi}{\delta} \left( 2\alpha\_{\tau} L\_{c} + \frac{d\Delta\_{2}}{dT} \right) \\ \frac{d\Delta\_{2}}{dT} = \int \left[ \frac{\hat{\mathcal{C}}(n\_{1} - n\_{2})}{\hat{\mathcal{C}}n} \sigma\_{\tau} + \frac{\hat{\mathcal{C}}(n\_{1} - n\_{2})}{\hat{\mathcal{C}}r} \alpha\_{\tau} \right] d\boldsymbol{z} \end{cases} \tag{4}$$

where σT (1.1×10-5 /°C) is the thermo-optics coefficient and αT (5.5×10-7 /°C) is the thermal expansion coefficient. There are two contributions from temperature change: the temperature-induced length variation in the cavity, and the temperature-induced index variation and taper volume variation in taper transition. The first one is less than 1 pm/°C and ignorable, it agrees with the fact that those previous micro-cavity FP interferences in SMF by femtosecond laser machining are temperature-insensitive; the second one is about 10 ~ 20 pm/°C and dominates in temperature sensing (Kou et al., 2010b).

Figure 7 displays the measured interferometer wavelength shifts (Δλ) and error on temperature (T). As the temperature increases, the interferometer wavelength shifts to longer wavelength. A third-order polynomial was used to fit the wavelength shifts across the entire calibration range. The average sensitivity of the device is ~ 17 pm/°C, which is very close to the theoretical result. Higher sensitivity can be obtained by optimizing the profile of the SMF-TT or using special fiber taper with higher thermo-optics coefficient (Kou et al., 2010b).

Fig. 7. Dependence of the measured wavelength shift on temperature. The asterisk represents the measured results while the solid line is the fitting result. The inset shows the dependence of error on temperature (Kou et al., 2010b). Reprinted with permission. Copyright 2010 Optical Society of America

#### **3.2 FIB machined FPMI for refractive index sensing**

The FPMI also can be employed as a refractive index sensor. Figure 8 shows the SEM picture of another FPMI with a micro-notch cavity from the side view. The cavity is 3.50 μm long and 2.94 μm high, located at the position with the local radius r = 2.4 μm.

Fig. 8. SEM image of the micro-notch cavity from the side view (Kou et al., 2010a). Reprinted with permission. Copyright 2010 Optical Society of America

Fig. 9. Interference spectra of the MPRI device in air (solid line), acetone (dashed line) and isopropanol (dotted line), at room temperature (25 °C) (Kou et al., 2010a). Reprinted with permission. Copyright 2010 Optical Society of America

The reflective Interference spectrum is measured with the same setup as shown in Fig. 2. Figure 9 shows the interference spectra of the MPRI in air, acetone and isopropanol at room temperature (25 °C). The interference spectrum indicates a fringe visibility maximum of ~ 20 dB, which is much higher than those of typical MCFPIs in liquids.

The performance of resonant or interferometer refractive index sensors can be evaluated by using the sensitivity SR, which is defined as the magnitude in shift of the resonant wavelength divided by the change in refractive index of the analyte. The sensitivity was

The FPMI also can be employed as a refractive index sensor. Figure 8 shows the SEM picture of another FPMI with a micro-notch cavity from the side view. The cavity is 3.50 μm long

Fig. 8. SEM image of the micro-notch cavity from the side view (Kou et al., 2010a). Reprinted

**1530 1535 1540 1545 1550 1555**

**air acetone isopropanol**

**Wavelength (nm)**

Fig. 9. Interference spectra of the MPRI device in air (solid line), acetone (dashed line) and isopropanol (dotted line), at room temperature (25 °C) (Kou et al., 2010a). Reprinted with

The reflective Interference spectrum is measured with the same setup as shown in Fig. 2. Figure 9 shows the interference spectra of the MPRI in air, acetone and isopropanol at room temperature (25 °C). The interference spectrum indicates a fringe visibility maximum of ~ 20

The performance of resonant or interferometer refractive index sensors can be evaluated by using the sensitivity SR, which is defined as the magnitude in shift of the resonant wavelength divided by the change in refractive index of the analyte. The sensitivity was

**3.2 FIB machined FPMI for refractive index sensing** 

with permission. Copyright 2010 Optical Society of America

**-50**

permission. Copyright 2010 Optical Society of America

dB, which is much higher than those of typical MCFPIs in liquids.

**-40**

**Reflection (dB)**

**-30**

**-20**

and 2.94 μm high, located at the position with the local radius r = 2.4 μm.

measured by inserting the sensor in mixtures of isopropanol and acetone. These solutions were chosen with the objective of simulating aqueous solutions, having a refractive index in the region around 1.33 at a wavelength of 1550 nm. The ratio was increased by adding small calibrated quantities of isopropyl to the solution at a position far from the sensor. The refractive indexes of pure isopropyl and acetone at 1550 nm are 1.3739 and 1.3577 respectively (Wei et al., 2008).

Figure 10 displays the shifted spectral wavelength as a function of the liquid mixture refractive index. The asterisks represent the measurement results while the solid line is the best-fitting. As the refractive index increases, the spectrum shows a red-shift. The sensitivity of the device is 110 nm/RIU (refractive index unit) according to Fig. 11. Higher sensitivity can be obtained by optimizing the profile of the microfiber taper probe. Due to its small size, fiber-probe structure, all fiber connection, linear response, low-cost, easy fabrication and high sensitivity, MPRI devices are promising in various chemical and biological applications. It even may offer fantastic potentials sensing inside sub-wavelength liquid droplets, bubbles or biocells because of its unique probe structure and possible smaller size (Kou et al., 2010a).

Fig. 10. The shifted spectral wavelength as a function of the liquid mixture refractive index. The asterisks represent the measured results while the solid line is the fitting results (Kou et al., 2010a). Reprinted with permission. Copyright 2010 Optical Society of America

#### **4. FIB machined TFT micro-grating**

Since their discovery in 1978 (Hill et al., 1978), optical fiber gratings have found a variety of applications in telecom and sensing because of their relatively low cost, inherent selfreferencing and multiplexing/demultiplexing capabilities. Over the last two decades, fiber gratings including fiber Bragg gratings (FBGs) and long-period gratings (LPGs) have been manufactured mainly by modifying the core refractive index using interferometric or pointby-point techniques; most of interferometric techniques use a phase mask and an ultraviolet (UV) laser (Hill et al., 1993) (typically excimer or frequency doubled Ar+ ion) or femtosecond lasers (near IR or UV). Gratings based on surface etched corrugations have also been demonstrated in etched fibers using photolithographic techniques (Lin & Wang, 2001). However all these gratings fabricated in thick fibers have weak refractive index modulations (Δnmod ~ 10-4 - 10-3) and the related grating lengths are of the order of several millimeters. To reduce the grating length, strong refractive index modulations (Δnmod > 10-2) are necessary. Strong Δnmod can be obtained by alternating layers of different materials, one of which can be air. Although this process in normal optical fibers imposes the removal of large amounts of material (the propagating mode is confined at a depth > 50 µm from the fiber surface), in fiber tapers and tips it requires the removal of small amounts of matter because the propagating mode is confined to the silica/air interface. A few techniques have been proposed for the fabrication of gratings in microfibers, including photorefractive inscription using CO2 lasers (Xuan et al., 2009), femtosecond lasers (Martinez et al., 2005; Xuan et al., 2010) and wrapping a microfiber on a microstructured rod (Xu et al., 2009, 2010). None of them produced strong and short Bragg gratings. In some cases extra polymer coatings are needed, while in others the use of CO2 lasers implies that the grating length is still long (it only can be used to write long period gratings (LPG) or high-order FBG). As a consequence, devices based on gratings tend to have a sizeable length: typically FBGs have lengths in the order of few millimetres. FIB technique provides a powerful way to mill the microfiber with directly and flexibly without a mask and realize compact micro-gratings (tens of micrometers) with colourful structure in the surface. Such small size and unique structure grating offer great potentials for a lot of sensing applications such as high temperature and refractive index sensing with the advantages of fast-response, ability to work in harsh environments and occupying little space.

#### **4.1 FIB machined micro-grating for temperature sensing**

Figure 11 shows an SEM micrograph of an ultra-short second-order TFT micro-grating (TFT-MG). The grating has 11 shallow corrugations with period Λ = 1.1 μm, providing a total length of ~ 12 μm, two orders of magnitude shorter than FBGs fabricated in conventional optical fiber. Each notch is ~ 1.6 μm deep and ~ 0.6 μm long. The average radius at the position where the notches are located is ~ 2.7 μm. The Bragg wavelength of the grating can be calculated from λg = 2neffΛ/m, where neff is the mode effective refractive index in the equivalent unperturbed geometry, Λ is the period and m is the Bragg order. Unlike conventional circularly-symmetric FBGs, this nanostructured TFT-MG has asymmetric periodic corrugations. The modal field and neff in the nano-structured TFT can be derived analytically from the bare TFT using the method developed by W. Streifer, which considers an equivalent structure, obtained by shifting the boundary between air and silica to compensate for the different geometry. Figure 12 shows the cross-sections of an un-etched fiber, an etched fiber and the equivalent unperturbed geometry with this method, which shifts the boundary between air and silica to compensate for the different geometry (Streifer et al., 1975, 1978), respectively. The effective groove height heff of the equivalent unperturbed geometry satisfies (Kou et al., 2011a):

$$\begin{cases} (1 - \tau)(\theta\_{\text{g}} - \sin \theta\_{\text{g}} \cos \theta\_{\text{g}}) = \theta\_{\text{eff}} - \sin \theta\_{\text{eff}} \cos \theta\_{\text{eff}}\\ \theta\_{\text{g}} = \arccos \left\{ (r - h\_{\text{g}}) / r \right\} \\ \theta\_{\text{eff}} = \arccos \left\{ (r - h\_{\text{eff}}) / r \right\} \end{cases} \tag{5}$$

lasers (near IR or UV). Gratings based on surface etched corrugations have also been demonstrated in etched fibers using photolithographic techniques (Lin & Wang, 2001). However all these gratings fabricated in thick fibers have weak refractive index modulations (Δnmod ~ 10-4 - 10-3) and the related grating lengths are of the order of several millimeters. To reduce the grating length, strong refractive index modulations (Δnmod > 10-2) are necessary. Strong Δnmod can be obtained by alternating layers of different materials, one of which can be air. Although this process in normal optical fibers imposes the removal of large amounts of material (the propagating mode is confined at a depth > 50 µm from the fiber surface), in fiber tapers and tips it requires the removal of small amounts of matter because the propagating mode is confined to the silica/air interface. A few techniques have been proposed for the fabrication of gratings in microfibers, including photorefractive inscription using CO2 lasers (Xuan et al., 2009), femtosecond lasers (Martinez et al., 2005; Xuan et al., 2010) and wrapping a microfiber on a microstructured rod (Xu et al., 2009, 2010). None of them produced strong and short Bragg gratings. In some cases extra polymer coatings are needed, while in others the use of CO2 lasers implies that the grating length is still long (it only can be used to write long period gratings (LPG) or high-order FBG). As a consequence, devices based on gratings tend to have a sizeable length: typically FBGs have lengths in the order of few millimetres. FIB technique provides a powerful way to mill the microfiber with directly and flexibly without a mask and realize compact micro-gratings (tens of micrometers) with colourful structure in the surface. Such small size and unique structure grating offer great potentials for a lot of sensing applications such as high temperature and refractive index sensing with the advantages of fast-response, ability to work in harsh

Figure 11 shows an SEM micrograph of an ultra-short second-order TFT micro-grating (TFT-MG). The grating has 11 shallow corrugations with period Λ = 1.1 μm, providing a total length of ~ 12 μm, two orders of magnitude shorter than FBGs fabricated in conventional optical fiber. Each notch is ~ 1.6 μm deep and ~ 0.6 μm long. The average radius at the position where the notches are located is ~ 2.7 μm. The Bragg wavelength of the grating can be calculated from λg = 2neffΛ/m, where neff is the mode effective refractive index in the equivalent unperturbed geometry, Λ is the period and m is the Bragg order. Unlike conventional circularly-symmetric FBGs, this nanostructured TFT-MG has asymmetric periodic corrugations. The modal field and neff in the nano-structured TFT can be derived analytically from the bare TFT using the method developed by W. Streifer, which considers an equivalent structure, obtained by shifting the boundary between air and silica to compensate for the different geometry. Figure 12 shows the cross-sections of an un-etched fiber, an etched fiber and the equivalent unperturbed geometry with this method, which shifts the boundary between air and silica to compensate for the different geometry (Streifer et al., 1975, 1978), respectively. The effective groove height heff of the equivalent

(1 )( sin cos ) sin cos

 

 

 

*g g eff eff*

arccos{( ) / } arccos{( ) / }

*g g g eff eff eff*

*rh r rh r*    

(5)

environments and occupying little space.

**4.1 FIB machined micro-grating for temperature sensing** 

unperturbed geometry satisfies (Kou et al., 2011a):

  where τ is the grating duty cycle, hg is the groove height and r is the fiber radius, respectively. In our device, r = 2.7 μm, τ = 0.33 and hg = 1.6 μm, we find heff = 1.2 μm by solving Eq. 5 and neff = 1.428 by utilizing a finite element method. Thus, the Bragg wavelength is 1571 nm. It agrees well with the following experimental results.

Fig. 11. SEM picture of the nanostructured TFT-MG. The grating has 11 notches and a total length of ~ 12 μm. The notch length and depth are ~ 0.6 µm and ~ 1.6 µm, respectively. The grating period is *Λ* ~ 1.1 μm (Feng et al., 2011).

Fig. 12. The cross-sections of un-etched fiber (a), etched fiber (b) and equivalent unperturbed geometry (c), respectively. hg is the groove height and heff is effective height (Kou et al., 2011a).

The reflective spectral response of the TFT-MG in Fig. 10 is measured with the setup shown in Fig. 2. We characterize the thermal response of the TFT-MG by heating it up in a microfurnace from room temperature (20 °C) to 230 °C. The spectrum and temperature are recorded when both of them are stable for several minutes.

The interference spectra of the TFT-MG at different temperatures (23 °C, 47°C, 104°C, 153°C and 228 °C) are shown in Fig. 13. The Bragg wavelength is ~ 1570 nm, in agreement with our theoretical calculation. The spectra indicate a reflection peak-to-trough ratio around 1570 nm of ~ 10 dB at the Bragg wavelength which is achieved with as few as 11 periods and is similar with or even better than some other long length fiber gratings, enough for sensing applications.

The temperature sensitivity ST is defined as (Kou et al., 2011a):

$$S\_r = \frac{d\lambda\_g}{dT} = \frac{2}{m} \left( \sigma\_T \Lambda \frac{\partial n\_{\rm eff}}{\partial n\_{\rm silica}} + \Lambda \alpha\_T n\_{\rm eff} + r \Lambda \alpha\_T \frac{\partial n\_{\rm eff}}{\partial r} \right) \tag{6}$$

where σT (1.4×10-5 /°C) is the thermo-optic coefficient and αT (5.5×10-7 /°C) is the thermal expansion coefficient. According to our calculations, the first part is about 15 ~ 20 pm/°C and dominates in temperature sensing. Thermal expansion effect (the second and third parts of Eq. 6) contributes little to the total sensitivity (< 6%), mainly due to the low thermal

expansion coefficient of silica. Moreover, in the first part of Eq. 2, *eff silica n n* is nearly 1 and does not change much with the microfiber diameter, which means that the most efficient method to increase thermal sensitivity is to use fiber with higher thermo-optics coefficient.

Fig. 13. Dependence of the measured wavelength shift on temperature. The asterisk represents the measured results while the solid line is the linear fitting result (Feng et al., 2011).

Figure 13 displays the measured resonant wavelength shifts (Δλ) on temperature (T). As the temperature increases, the interference wavelength shifts to longer wavelength. The average sensitivity of the device is ~ 22 pm/°C, which is very close to the theoretical result, higher than or similar with previous fiber grating sensors. Higher sensitivity can be obtained by use special fiber taper with higher thermo-optics coefficient.

First-order micro-grating (m = 1) with smaller period also can be fabricated in TFT by FIB. Figure 14 shows the SEM photography of a first-order TFT-MG from the side view. The grating has shallow corrugations of period Λ = 600 nm with 61 periods. The total length is

The interference spectra of the TFT-MG at different temperatures (23 °C, 47°C, 104°C, 153°C and 228 °C) are shown in Fig. 13. The Bragg wavelength is ~ 1570 nm, in agreement with our theoretical calculation. The spectra indicate a reflection peak-to-trough ratio around 1570 nm of ~ 10 dB at the Bragg wavelength which is achieved with as few as 11 periods and is similar with or even better than some other long length fiber gratings, enough for sensing

*<sup>g</sup>* 2 *eff eff*

(6)

is nearly 1

*dn n*

*dT m n r*

where σT (1.4×10-5 /°C) is the thermo-optic coefficient and αT (5.5×10-7 /°C) is the thermal expansion coefficient. According to our calculations, the first part is about 15 ~ 20 pm/°C and dominates in temperature sensing. Thermal expansion effect (the second and third parts of Eq. 6) contributes little to the total sensitivity (< 6%), mainly due to the low thermal

and does not change much with the microfiber diameter, which means that the most efficient method to increase thermal sensitivity is to use fiber with higher thermo-optics

<sup>0</sup> <sup>50</sup> <sup>100</sup> <sup>150</sup> <sup>200</sup> <sup>250</sup> -1

Temperature (C)

Fig. 13. Dependence of the measured wavelength shift on temperature. The asterisk represents

Figure 13 displays the measured resonant wavelength shifts (Δλ) on temperature (T). As the temperature increases, the interference wavelength shifts to longer wavelength. The average sensitivity of the device is ~ 22 pm/°C, which is very close to the theoretical result, higher than or similar with previous fiber grating sensors. Higher sensitivity can be obtained by

First-order micro-grating (m = 1) with smaller period also can be fabricated in TFT by FIB. Figure 14 shows the SEM photography of a first-order TFT-MG from the side view. The grating has shallow corrugations of period Λ = 600 nm with 61 periods. The total length is

the measured results while the solid line is the linear fitting result (Feng et al., 2011).

*T TT eff T silica*

*S n r*

expansion coefficient of silica. Moreover, in the first part of Eq. 2, *eff silica n n*

experimental data

linear fitting

The temperature sensitivity ST is defined as (Kou et al., 2011a):

0

use special fiber taper with higher thermo-optics coefficient.

1

2

Wavelength shift (nm)

3

4

5

applications.

coefficient.

about 36.6 μm, which is extremely short. Every groove is 200 nm in depth, located at the position with the local radius around r = 3.25 μm. The resonant spectra of the TFT-MG at different temperatures are shown in Fig. 15. The Bragg wavelength is ~ 1550 nm, with excited higher order mode as deduced from our theoretical calculation. The spectra indicate an extinction ratio of ~ 11 dB at the Bragg wavelength which is achieved with a 36.6 um long Bragg grating. The average temperature sensitivity of the device from room temperature to around 500 °C is ~ 20 pm/°C as shown in Fig. 15 (b), which is similar with the second-order TFT-MG. It is reasonable because the main thermal contribution is from the thermo-optic effect (Kou et al., 2011a).

Fig. 14. Left: FIB picture of the TFPG with 61 periods (~ 36.6 μm in length and Λ = 600 nm). Right: magnified picture of the grating (Kou et al., 2011a). Reprinted with permission. Copyright 2010 Optical Society of America

Fig. 15. (a) Reflection spectra of the first-order TFT-MG in air at different temperatures. (b) Dependence of the measured wavelength shift on temperature. The asterisk represents the measured results while the solid line is the linear fitting result (Kou et al., 2011a). Reprinted with permission. Copyright 2010 Optical Society of America

#### **4.2 FIB machined metal-dielectric-hybrid micro-grating for refractive index sensing**

Conventional FBGs have been extensively developed to measure the temperature, pressure or stress. But it is scarcely used to measure the environmental refractive index variation because there is almost no evanescent field penetrating outside of a standard 125 μm diameter FBG. TFT-MG may overcome the drawback with the available evanescence field interacting with the outer environments. The sensitivity of a pure-silica TFT-MG with the diameter of several micrometers is about tens of nm/RIU. By inducing metal-cladding, more cladding modes are possible to be excited and higher sensitivity can be obtained, which is so called grating-assisted surface plasmon-polariton (SPP)-like grating sensor (Nemova & Kashyap, 2006).

Figure 16 shows the SEM picture of a metal-dielectric-hybrid TFT-MG (MD-TFT-MG) by FIB milling. The fabrication process is similar with those mentioned ones above. But the fiber tip is coated with a gold layer with thickness of 30 nm on one side by magnetron sputtering and it is kept all the way throughout the experiment. We choose gold due to its relatively low absorption in the infrared and inertness to oxidation when exposed in air. Then a grating is fabricated by FIB milling at the fiber tip with local radius of ~ 3 μm. The grating has shallow corrugations of period Λ = 578 nm with 17 periods. The total length is about 10 μm, which is extremely short with local radius of ~ 3 μm.

Fig. 16. SEM picture of the metal-dielectric-hybrid fiber tip grating (~ 10 μm in length and Λ = 578 nm). Right: magnified picture of the grating (Kou et al., 2011b).

Optical characterization of the MD-TFT-MG in Fig. 16 is performed using the same setup as shown in Fig. 2. Figure 17 shows the reflection spectra of the MD-TFT-MG in air, acetone, and isopropanol, respectively. The extinction ratio is about ~ 10 dB. There are several valleys and peaks with different characteristics in the spectral range of ~ 100 nm. They shift when the outer environment changes from acetone to isopropanol. However, these valleys and peaks show larger shifts at longer wavelengths, while those at shorter wavelength region shift much less and almost stop at specific wavelengths. This unique response to outer liquid refractive index comes from the fact that the reflected light can be coupled to different modes. In the micrometer-diameter metal-dielectric-hybrid TFT, several modes are probably excited with similar propagation constant because of the metal cladding. Some modes are well confined in the tip and have negligible field overlap with the liquid while some modes are not. The different valleys and peaks correspond to the coupling between these different forward and backward propagating modes, with different response properties for the outer environment.

The reflection resonant condition for the grating is:

30 Micromachining Techniques for Fabrication of Micro and Nano Structures

Figure 16 shows the SEM picture of a metal-dielectric-hybrid TFT-MG (MD-TFT-MG) by FIB milling. The fabrication process is similar with those mentioned ones above. But the fiber tip is coated with a gold layer with thickness of 30 nm on one side by magnetron sputtering and it is kept all the way throughout the experiment. We choose gold due to its relatively low absorption in the infrared and inertness to oxidation when exposed in air. Then a grating is fabricated by FIB milling at the fiber tip with local radius of ~ 3 μm. The grating has shallow corrugations of period Λ = 578 nm with 17 periods. The total length is about 10 μm, which is

Fig. 16. SEM picture of the metal-dielectric-hybrid fiber tip grating (~ 10 μm in length and

Optical characterization of the MD-TFT-MG in Fig. 16 is performed using the same setup as shown in Fig. 2. Figure 17 shows the reflection spectra of the MD-TFT-MG in air, acetone, and isopropanol, respectively. The extinction ratio is about ~ 10 dB. There are several valleys and peaks with different characteristics in the spectral range of ~ 100 nm. They shift when the outer environment changes from acetone to isopropanol. However, these valleys and peaks show larger shifts at longer wavelengths, while those at shorter wavelength region shift much less and almost stop at specific wavelengths. This unique response to outer liquid refractive index comes from the fact that the reflected light can be coupled to different modes. In the micrometer-diameter metal-dielectric-hybrid TFT, several modes are probably excited with similar propagation constant because of the metal cladding. Some modes are well confined in the tip and have negligible field overlap with the liquid while some modes

Λ = 578 nm). Right: magnified picture of the grating (Kou et al., 2011b).

**4.2 FIB machined metal-dielectric-hybrid micro-grating for refractive index sensing**  Conventional FBGs have been extensively developed to measure the temperature, pressure or stress. But it is scarcely used to measure the environmental refractive index variation because there is almost no evanescent field penetrating outside of a standard 125 μm diameter FBG. TFT-MG may overcome the drawback with the available evanescence field interacting with the outer environments. The sensitivity of a pure-silica TFT-MG with the diameter of several micrometers is about tens of nm/RIU. By inducing metal-cladding, more cladding modes are possible to be excited and higher sensitivity can be obtained, which is so called grating-assisted surface plasmon-polariton (SPP)-like grating sensor (Nemova &

Kashyap, 2006).

extremely short with local radius of ~ 3 μm.

$$\frac{2\pi}{\lambda\_g} [n\_f + n\_b] = \frac{2\pi}{\Lambda} \tag{7}$$

where nf and nb are the effective indices of the forward and backward modes, respectively. For simplicity, we assume a theoretical model to explain our experimental results which is simple and not perfectly matched with the experiment but can give the fundamental mechanism of the device. Within the model, the microfiber is 6 μm in diameter with uniform metal cladding (20 nm in thickness). However, the real device is much more complicated, with nonuniform metal cladding and diameter. And if an asymmetrical mode field lies mainly near the grating, leading to a larger modal overlap with the grating, it may result in a higher sensitivity. Figure 3 shows the calculation on the effective index of one cladding mode and one core mode as a function of outer liquid refractive index nl. Due to the existence of the metal layer, the cladding mode has a larger effective index (corresponding to long resonant wavelength) than that of the core mode (corresponding to short resonant wavelength) and has a larger overlap with the taper surface and the outside environment, leading to a higher sensitivity to the surrounding medium which is in coincidence with the spectra of Fig. 2.

Fig. 17. Measured reflection spectra of the FTG when immersed in acetone and isopropanol (Kou et al., 2011b).

The performance of resonant refractive index sensors can be evaluated by using sensitivity S, which is defined as the magnitude in shift of the resonant wavelength divided by the change in refractive index of the analyte. In our experiment, the sensitivity is measured by inserting the sensor in a beaker containing mixtures of isopropanol and acetone, where the isopropanol component has the following ratios: 0, 1/7, 2/7, 3/7, 4/7 5/7, 6/7, and 1 (Kou et al., 2011b). Figure 18 displays measured resonant wavelength shifts of several peaks and valleys and

fitting of this FTG on the liquid refractive index (a, b, c, d as marked in Fig. 2, a and c are

peaks, b and d are valleys). As the refractive index increases, the resonant wavelength shifts to longer wavelength. The sensitivities of different modes change severely. It can be as high as 125 nm/RIU (peak a) or as low as 7 nm/RIU (valley d). For peak a (or valley b), both the resonant wavelength and sensitivity are larger than those of peak c (or valley d). According to our theoretical calculation, we believe peak a (or valley b) corresponds to cladding mode while peak c (or valley d) is core mode. The smallest sensitivity can be further decreased to nearly zero by optimizing the tip grating profile and metal coating. Because of many different properties on the outer liquid refractive index, the metal-dielectric-hybrid FTG can be applied as a multi-parameter sensor and the index-insensitive channel can be used to simultaneously measure temperature, pressure, and so on (Kou et al., 2011b).

Fig. 18. Dependence of wavelength shift on outer liquid refractive index n1. The asterisks represent the experimental results with the solid line of linear fitting (Kou et al., 2010b).

#### **5. Conclusion**

In this chapter, FIB machined TFT based micro-devices including interferometers and gratings are demonstrated. Being a very exible, mask-less, direct write process, FIB milling is perfect for carving nanoscale geometries precisely in microfibers. Various miniature fiber devices can be realized and they show great potential in sensing with the unique geometry and size. The sensitivity such as of temperature or refractive index can't increase too much because it mainly depends on the fiber materials and size. But the ultra-small size is attractive for some special application, in particular for detecting small-size objects. Some novel geometry is possible to be realized in microfiber such as an inline-microring, a slot-microfiber etc.

#### **6. Acknowledgment**

This work is supported by National 973 program under contract No. 2010CB327803, 2012CB921803 and 2011CBA00200, NSFC program No. 11074117 and 60977039. The authors also acknowledge the support from the Priority Academic Program Development of Jiangsu (PAPD), and the Fundamental Research Funds for the Central Universities.

#### **7. References**

Agrawal, G. P. (2002). Fiber-optic communication systems. New York, Wiley-Interscience.

peaks, b and d are valleys). As the refractive index increases, the resonant wavelength shifts to longer wavelength. The sensitivities of different modes change severely. It can be as high as 125 nm/RIU (peak a) or as low as 7 nm/RIU (valley d). For peak a (or valley b), both the resonant wavelength and sensitivity are larger than those of peak c (or valley d). According to our theoretical calculation, we believe peak a (or valley b) corresponds to cladding mode while peak c (or valley d) is core mode. The smallest sensitivity can be further decreased to nearly zero by optimizing the tip grating profile and metal coating. Because of many different properties on the outer liquid refractive index, the metal-dielectric-hybrid FTG can be applied as a multi-parameter sensor and the index-insensitive channel can be used to

1.36 1.365 1.37

Fig. 18. Dependence of wavelength shift on outer liquid refractive index n1. The asterisks represent the experimental results with the solid line of linear fitting (Kou et al., 2010b).

n l

In this chapter, FIB machined TFT based micro-devices including interferometers and gratings are demonstrated. Being a very exible, mask-less, direct write process, FIB milling is perfect for carving nanoscale geometries precisely in microfibers. Various miniature fiber devices can be realized and they show great potential in sensing with the unique geometry and size. The sensitivity such as of temperature or refractive index can't increase too much because it mainly depends on the fiber materials and size. But the ultra-small size is attractive for some special application, in particular for detecting small-size objects. Some novel geometry is possible to be

This work is supported by National 973 program under contract No. 2010CB327803, 2012CB921803 and 2011CBA00200, NSFC program No. 11074117 and 60977039. The authors also acknowledge the support from the Priority Academic Program Development of Jiangsu (PAPD),

Agrawal, G. P. (2002). Fiber-optic communication systems. New York, Wiley-Interscience.

**c**

**a b**

**d**

simultaneously measure temperature, pressure, and so on (Kou et al., 2011b).

0

realized in microfiber such as an inline-microring, a slot-microfiber etc.

and the Fundamental Research Funds for the Central Universities.

0.5

1

1.5

Wavelength shift (nm)

**5. Conclusion** 

**6. Acknowledgment** 

**7. References** 

2


## **Fundamentals of Laser Ablation of the Materials Used in Microfluiducs**

Tai-Chang Chen and Robert Bruce Darling *University of Washington, USA* 

#### **1. Introduction**

34 Micromachining Techniques for Fabrication of Micro and Nano Structures

Liang, T., Stivers, A., Livengood, R., Yan, P. Y., Zhang, G. J. & Lo, F. C. (2000). Progress in

Lin, C. Y. & Wang, L. A. (2001). A wavelength- and loss-tunable band-rejection filter based

Liu, K., Soskov, A., Scipioni, L., Bassom, N., Sijbrandij, S. & Smith, G. (2006). Electrical

Love, J. D., Henry, W. M., Stewart, W. J., Black, R. J., Lacroix, S. & Gonthier, F. (1991).

Martinez, A., Khrushchev, I. Y. & Bennion, I. (2005). Thermal properties of fibre Bragg

Nemova, G. & Kashyap, R. (2006). Fiber-Bragg-grating-assisted surface plasmon-polariton

Rao, Y.-J., Deng, M., Duan, D.-W., Yang, X.-C., Zhu, T. & Cheng, G.-H. (2007). Micro Fabry-

Sirkis, J. S., Brennan, D. D., Putman, M. A., Berkoff, T. A., Kersey, A. D. & Friebele, E. J.

Streifer, W. & Hardy, A. (1978). Analysis of two-dimensional waveguides with misaligned or

Streifer, W., Scifres, D. & Burnham, R. (1975). Coupling coefficients for distributed feedback

Terris, B., Thomson, T. & Hu, G. (2007). Patterned media for future magnetic data storage.

Tong, L. M., Gattass, R. R., Ashcom, J. B., He, S. L., Lou, J. Y., Shen, M. Y., Maxwell, I. &

Villatoro, J., Finazzi, V., Coviello, G. & Pruneri, V. (2009). Photonic-crystal-fiber-enabled micro-Fabry?Perot interferometer. *Optics Letters*, Vol. 34, No. 16, (2009), pp. 2441-2443 Wei, T., Han, Y., Li, Y., Tsai, H.-L. & Xiao, H. (2008). Temperature-insensitive miniaturized

Xu, F., Brambilla, G., Feng, J. & Lu, Y.-Q. (2010). A Microfiber Bragg Grating Based on a

Xu, F., Brambilla, G. & Lu, Y. (2009). A microfluidic refractometric sensor based on gratings in optical fibre microwires. *Optics Express*, Vol. 17, No. 23, (2009), pp. 20866-20871 Xuan, H., Jin, W. & Liu, S. (2010). Long-period gratings in wavelength-scale microfibers.

Xuan, H., Jin, W. & Zhang, M. (2009). CO2 laser induced long period gratings in optical

microfibers. *Optics Express*, Vol. 17, No. 24, (2009), pp. 21882-21890

measurement. *Optics Express*, Vol. 16, No. 8, (2008), pp. 5764-5769

*& Technology B*, Vol. 18, No. 6, (2000), pp. 3216-3220

*Physics Letters*, Vol. 88, No. 12, (2006), pp. 124104

*Proceedings Journal*, Vol. 138, No. 5, (1991), pp. 343-354

sensor. *Optics Leters*, Vol. 31, No. 14, (2006), pp. 2118-2120

No. 4, (2001), pp. 332-334

Vol. 41, No. 4, (2005), pp. 176-178

(1993), pp. 1973-1975

(2010), pp. 218-220

Vol. 15, No. 21, (2007), pp. 14123-14128

*Electronics*, Vol. 11, No. 11, (1975), pp. 867-873

*Optics Letters*, Vol. 35, No. 1, (2010), pp. 85-87

*Microsystem Technologies*, Vol. 13, No. 2, (2007), pp. 189-196

guiding. *Nature*, Vol. 426, No. 6968, (2003), pp. 816-819

extreme ultraviolet mask repair using a focused ion beam. *Journal of Vacuum Science* 

on corrugated long-period fiber grating. *IEEE Photonics Technology Letters*, Vol. 13,

breakthrough effect for end pointing in 90 and 45 nm node circuit edit. *Applied* 

Tapered single-mode fibres and devices. I. Adiabaticity criteria. *Optoelectronics, IEE* 

gratings inscribed point-by-point by infrared femtosecond laser. *Electronics Letters*,

Perot interferometers in silica fibers machined by femtosecond laser. *Optics Express*,

(1993). In-line fiber ealon for strain measurement. *Optics Letters*, Vol. 18, No. 22,

curved gratings. *IEEE Journal of Quantum Electronics*, Vol. 14, No. 12, (1978), pp. 935-943

single- and double-heterostructure diode lasers. *IEEE Journal of Quantum* 

Mazur, E. (2003). Subwavelength-diameter silica wires for low-loss optical wave

fiber inline Fabry-Perot interferometer for highly sensitive refractive index

Microstructured Rod: A Proposal. *IEEE Photonics Technology Letters*, Vol. 22, No. 4,

Microfluidics falls into an intermediate range within the spectrum of applications for microfabrication techniques. The width and depth of most microfluidic channels fall in the range of 10-1000 µm, and this feature size is thus small for conventional machine tool microfabrication, but quite large for photolithographically defined etching processes of the type used within the microelectronics industry. In addition, most microfluidic channels occupy only ~10% or less of the surface area of a microfluidic device. Wet chemical or plasma etching processes to produce microfluidic devices therefore take considerable time to complete, based upon the comparatively deep depths that are required for the channels. A comparatively fast wet or dry etching rate of 1 µm/min would still require up to several hours per wafer to achieve these depths. The small surface areas that are etched within this time make conventional batch processing of wafers less attractive economically. In many cases, photolithographically defined microfluidic features with micron scale accuracy are more precise than what is required for these applications.

At high volumes, other microfabrication processes become more applicable for the manufacture of microfluidics. Roll-to-roll stamping, lamination, hot embossing, and injection molding of plastic components offer excellent accuracy, repeatability, and cost effectiveness once the non-recoverable engineering (NRE) costs of molds, dies, and master templates have been paid for. However, the cost of these NRE items is comparatively high, and in most circumstances, production volumes of >1 million parts are required to recover this cost.

For part volumes from 1 to 1 million, laser microfabrication offers an excellent balance between speed, cost, and accuracy for microfluidics. Laser micromachining is also unmatched in the breadth of different of materials that it can process. A single laser system can micromachine materials all the way from lightweight plastics and elastomers up through hard, durable metals and ceramics. This versatility makes laser micromaching extremely attractive for prototyping and development, as well as for small to medium run manufacturing.

The most common criticism of laser micromachining is that it is a serial, rather than batch process, and it is therefore too slow to be economical for high volume manufacturing. While certainly true in some instances, as a generalization, this is not always the case. The processing time per part is the sum of the beam exposure time plus the beam positioning time. For parts which require only minimal volumes of material to be removed, serial processes such as laser micromachining can indeed be extremely efficient and cost effective. Whereas older laser micromachining systems were often limited by clumsy beam positioning, modern systems incorporate high speed beam positioning and parts handling so that the overall processing time is limited more by the net beam exposure time, which for many applications can be fairly small. A good counter-example to the criticism of serial processing is chip resistor trimming, which is used for almost all 1% tolerance and better metal film chip resistors in the microelectronics industry today and which are produced in extremely high volumes, >10 billion/year.

Microfluidics is becoming increasingly used for miniaturized chemical analysis systems, such as the new generations of lab-on-a-chip applications which are rapidly being developed. The fundamental structure used in microfluidics is the flow channel, but integrated microfluidic systems also incorporate vias, T-junctions, sample wells, reaction chambers, mixers, and manifolds, along with some moving mechanical components such as valves, pumps, and injectors, and often some optical and electrical components for integrated control and sensing. Unlike wet and dry etching which must be carefully formulated to achieve the required material selectivity, laser micromachining can be used to process many different materials and structures at a time. For example, a laser can be used to cut a channel to one depth, cut a via to another depth, trim a metal trace, release a check valve structure, and weld two mating elements together all within the same mounting of the part. This illustrates one of the advantages that serial processing has over traditional batch processing of wafers. Another obvious advantage of serial laser processing is that no masking is required, greatly reducing the time and expense for design changes. Different parts can also be individually customized with virtually no extra tooling overhead. Microfluidics and laser micromachining are an excellent marriage of technologies which will prove essential for the rapid development of these applications.

This chapter will discuss the fundamentals of laser ablation in the microfabrication of microfluidic materials. After briefly describing the various types of lasers which are used for this purpose, the fundamental mechanisms of laser micromachining will be described, along with some data illustrating the performance of some state-of-the-art laser micromachining systems.

#### **1.1 Lasers for micromachining**

By far the most common laser used for industrial processing is the carbon dioxide (CO2) gas laser. This popularity comes from its unique combination of high average power, high efficiency, and rugged construction. Unlike the original glass tube style gas lasers, the modern CO2 lasers which are used for materials processing are of a hard sealed waveguide construction that use extruded aluminum RF driven electrodes to excite a CO2/N2/He gas mixture. The lasing transitions are from asymmetric to symmetric stretch modes at 10.6 µm, or from asymmetric stretch to bending modes at 9.4 µm of the CO2 molecule (Verdeyen, 1989). Within each of these vibrational modes there exist numerous rotational modes, and hundreds of lasing transitions can be supported by excitation into the parent asymmetrical stretch mode of the CO2 molecules. This large number of simultaneous lasing modes along with the efficient excitation coupling through the N2 gas is what allows CO2 lasers to achieve power levels up to 1 kW with electrical to optical conversion efficiencies of nearly 10%. CO2 lasers emit in the mid-infrared (MIR), most commonly at 10.6 µm, and they principally interact with their target materials via focused, radiant heating. They are used

processes such as laser micromachining can indeed be extremely efficient and cost effective. Whereas older laser micromachining systems were often limited by clumsy beam positioning, modern systems incorporate high speed beam positioning and parts handling so that the overall processing time is limited more by the net beam exposure time, which for many applications can be fairly small. A good counter-example to the criticism of serial processing is chip resistor trimming, which is used for almost all 1% tolerance and better metal film chip resistors in the microelectronics industry today and which are produced in

Microfluidics is becoming increasingly used for miniaturized chemical analysis systems, such as the new generations of lab-on-a-chip applications which are rapidly being developed. The fundamental structure used in microfluidics is the flow channel, but integrated microfluidic systems also incorporate vias, T-junctions, sample wells, reaction chambers, mixers, and manifolds, along with some moving mechanical components such as valves, pumps, and injectors, and often some optical and electrical components for integrated control and sensing. Unlike wet and dry etching which must be carefully formulated to achieve the required material selectivity, laser micromachining can be used to process many different materials and structures at a time. For example, a laser can be used to cut a channel to one depth, cut a via to another depth, trim a metal trace, release a check valve structure, and weld two mating elements together all within the same mounting of the part. This illustrates one of the advantages that serial processing has over traditional batch processing of wafers. Another obvious advantage of serial laser processing is that no masking is required, greatly reducing the time and expense for design changes. Different parts can also be individually customized with virtually no extra tooling overhead. Microfluidics and laser micromachining are an excellent marriage of technologies which will

This chapter will discuss the fundamentals of laser ablation in the microfabrication of microfluidic materials. After briefly describing the various types of lasers which are used for this purpose, the fundamental mechanisms of laser micromachining will be described, along with some data illustrating the performance of some state-of-the-art laser micromachining

By far the most common laser used for industrial processing is the carbon dioxide (CO2) gas laser. This popularity comes from its unique combination of high average power, high efficiency, and rugged construction. Unlike the original glass tube style gas lasers, the modern CO2 lasers which are used for materials processing are of a hard sealed waveguide construction that use extruded aluminum RF driven electrodes to excite a CO2/N2/He gas mixture. The lasing transitions are from asymmetric to symmetric stretch modes at 10.6 µm, or from asymmetric stretch to bending modes at 9.4 µm of the CO2 molecule (Verdeyen, 1989). Within each of these vibrational modes there exist numerous rotational modes, and hundreds of lasing transitions can be supported by excitation into the parent asymmetrical stretch mode of the CO2 molecules. This large number of simultaneous lasing modes along with the efficient excitation coupling through the N2 gas is what allows CO2 lasers to achieve power levels up to 1 kW with electrical to optical conversion efficiencies of nearly 10%. CO2 lasers emit in the mid-infrared (MIR), most commonly at 10.6 µm, and they principally interact with their target materials via focused, radiant heating. They are used

extremely high volumes, >10 billion/year.

systems.

**1.1 Lasers for micromachining** 

prove essential for the rapid development of these applications.

extensively for marking, engraving, drilling, cutting, welding, annealing, and heat treating an enormous variety of industrial materials (Berrie & Birkett, 1980; Crane & Brown, 1981; Crane, 1982). For micromachining applications, the long wavelength translates into a fairly large spot diameter of ~50-150 µm with a corresponding kerf width when used for through cutting.

The most common solid-state laser used in industry is the neodymium-doped yttriumaluminum-garnet, or Nd:YAG. The YAG crystal is a host for Nd3+ ions, whose lasing transitions from the excited 4F3/2 band to the energetically lower 4I11/2 band produces emission at 1.064 µm in the near-infrared (NIR) (Koechner, 1988; Kuhn, 1998). Nearly all industrial Nd:YAG lasers are now pumped by semiconductor diode lasers, usually made of GaAlAs quantum wells and tuned to emit at ~810 nm, for optimum matching to the pertinent absorption band of Nd:YAG. Semiconductor diode pumping of Nd:YAG offers much more efficient pumping with minimal energy being lost to heat, since the diode emits only into that part of the spectrum which is needed for the pumping. However, semiconductor diode pump lasers can only be made up to ~100 W, and thus these are used only for Nd:YAG lasers of low to moderate average powers. Most industrial Nd:YAG lasers are also Q-switched, usually by means of a KD\*P electrooptic intracavity modulator. When the modulator is in the non-transparent state, the pumping of the Nd:YAG rod allows the population inversion to build up to very high levels. When the modulator is rapidly switched to the transparent state, the energy stored in the inverted population is discharged at once into a single giant pulse of narrow duration and high peak power. Typical Qswitched pulse widths are in the range of ~25 ns, and with firing repetition rates of ~40 kHz, the duty cycle of a Q-switched Nd:YAG laser is ~1:1000. A ~10 W average power Nd:YAG laser can then produce pulses with peak powers of ~10 kW. This high peak power makes Qswitched Nd:YAG lasers ideally suited for nonlinear optical frequency multiplication through the use of an external cavity harmonic generating crystal such as KDP, KTP, LiNbO3, or BBO. Most commonly, the 1064 nm output from the Nd:YAG is frequency doubled to produce a green output at 532 nm. The 1064 nm output can also be frequency tripled to produce 355 nm in the near ultraviolet (UVA band), or frequency quadrupled (using a sequential pair of doublers) to 266 nm in the deep ultraviolet (UVC band). All four of these commonly available Nd:YAG output wavelengths are extremely useful for micromachining purposes (Atanasov et al., 2001; Tunna et al., 2001).

Copper vapor lasers have also proven their use in high accuracy micromachining (Knowles, 2000; Lash & Gilgenbach, 1993). Similar to the Nd:YAG, they are Q-switched systems which produce high intensity pulses of typically ~25 ns at rates of 2-50 kHz and average powers of 10-100 W. Unlike the Nd:YAG, they emit directly into the green at 511 nm and 578 nm, and thus do not require a nonlinear crystal for frequency multiplication to reach these more useful wavelengths. Copper vapor lasers also have excellent beam quality and can usually produce a diffraction-limited spot on the substrate with only simple external beam steering optics. The disadvantage of copper vapor lasers is that they tend to have shorter service life and require more maintenance than Nd:YAG lasers. Frequency multiplying crystals have now become a ubiquitous feature of commercial Nd:YAG lasers, and as a result, Nd:YAGs have largely displaced the copper vapor laser for industrial micromachining applications.

Excimer lasers have also found wide use in materials processing applications. Excimer lasers operate from a molecular transition of a rare gas-halogen excited state that is usually pumped by an electric discharge. The XeCl excimer laser, which emits at 308 nm, is prototypical of these in which a pulsed electric discharge ionizes the Xe into a Xe+ state and ionizes the Cl2 into a Cl− state. These two ions can then bind into a Xe+Cl− molecule which will loose energy through a lasing transition as it relaxes back to the XeCl state. The resulting ground state XeCl molecule readily dissociates, and these products are then recycled. Other commonly used excimer lasers are the XeF which emits at 351 nm, the KrF which emits at 249 nm, the ArF which emits at 193 nm, and the diatomic F2 which emits at 157 nm (Kuhn, 1998). Like other laser systems which are well matched to applications in materials processing, excimer lasers produce pulses of ~50 ns with repetition rates of ~100 Hz to ~10 kHz and average powers of up to a few hundred Watts. Excimer lasers are fairly efficient in their electrical to optical conversion efficiency, but their use of highly reactive halogen gases at high pressures requires significantly more servicing and maintenance than other types. One of the most important properties of excimer lasers is their ability to create a rather large spot size which can be homogenized into a high quality flat top beam profile of up to several cm in dimension. Because of this, they have been the pre-eminent source for coherent UV radiation at moderate power levels, they can be used both as a masked or a scanned exposure source, and currently they are used extensively for UV and deep UV lithography as well as several other applications in thin film recrystallization and annealing. At higher beam intensities, they can be used for surface ablation of materials, and due to the short wavelength and short pulse width, they typically produce clean, crisp features in metals, ceramics, glasses, polymers, and composites, making them adaptable for numerous micromachining applications (Gower, 2000).

Short laser pulses, on the order of a few tens of nanoseconds, are a desirable feature for laser micromachining applications, and these can be produced with many different laser systems. As will be discussed in more detail later, the short pulse width produces nearly adiabatic heating of the substrate which allows the substrate surface temperatures to quickly reach the point of vaporization with minimal heating effects on the surrounding areas. There has been interest in laser systems which can produce even shorter pulse widths, and the foremost candidate for this has been the Ti:sapphire laser. The Ti:sapphire laser has the unique feature of being tunable over a surprisingly large fluorescence band: from ~670 nm to ~1090 nm. For efficient pumping, it needs to be optically excited in its absorption band, which is centered about 500 nm, and for which argon ion lasers and frequency doubled Nd:YAG lasers provide excellent sources (Kuhn 1998). Most Ti:sapphire lasers are configured into an optical ring resonator arrangement with a set of birefringent filters for tuning. In addition, the ring cavity usually contains a Faraday rotator and wave plates to limit the propagation to only one direction around the ring. This arrangement is well suited for wide tuning and also mode locking, through which very short pulses, on the order of a few tens of femtoseconds can be produced. Ti:sapphire lasers have thus become a key resource for spectroscopy and research on ultrafast phenomena. The Ti:sapphire laser is also capable of average powers of up to several Watts, which makes it a viable tool for micromachining. Although its operation is at longer wavelengths than those normally preferred for micromachining, its capability for tuning and producing ultrashort pulses makes it attractive for research in this area. Since it requires a pump laser of ~10 W which is already in the green, and its more complicated optical system requires more maintenance and user savvy, it is presently not a common choice for industrial micromachining applications, but this may change in the future. There are many other new laser systems under development which offer efficient generation of green light at the power levels and pulse widths required for micromachining. It is worthwhile to realize that the field of laser sources is constantly changing.

ionizes the Cl2 into a Cl− state. These two ions can then bind into a Xe+Cl− molecule which will loose energy through a lasing transition as it relaxes back to the XeCl state. The resulting ground state XeCl molecule readily dissociates, and these products are then recycled. Other commonly used excimer lasers are the XeF which emits at 351 nm, the KrF which emits at 249 nm, the ArF which emits at 193 nm, and the diatomic F2 which emits at 157 nm (Kuhn, 1998). Like other laser systems which are well matched to applications in materials processing, excimer lasers produce pulses of ~50 ns with repetition rates of ~100 Hz to ~10 kHz and average powers of up to a few hundred Watts. Excimer lasers are fairly efficient in their electrical to optical conversion efficiency, but their use of highly reactive halogen gases at high pressures requires significantly more servicing and maintenance than other types. One of the most important properties of excimer lasers is their ability to create a rather large spot size which can be homogenized into a high quality flat top beam profile of up to several cm in dimension. Because of this, they have been the pre-eminent source for coherent UV radiation at moderate power levels, they can be used both as a masked or a scanned exposure source, and currently they are used extensively for UV and deep UV lithography as well as several other applications in thin film recrystallization and annealing. At higher beam intensities, they can be used for surface ablation of materials, and due to the short wavelength and short pulse width, they typically produce clean, crisp features in metals, ceramics, glasses, polymers, and composites, making them adaptable for numerous

Short laser pulses, on the order of a few tens of nanoseconds, are a desirable feature for laser micromachining applications, and these can be produced with many different laser systems. As will be discussed in more detail later, the short pulse width produces nearly adiabatic heating of the substrate which allows the substrate surface temperatures to quickly reach the point of vaporization with minimal heating effects on the surrounding areas. There has been interest in laser systems which can produce even shorter pulse widths, and the foremost candidate for this has been the Ti:sapphire laser. The Ti:sapphire laser has the unique feature of being tunable over a surprisingly large fluorescence band: from ~670 nm to ~1090 nm. For efficient pumping, it needs to be optically excited in its absorption band, which is centered about 500 nm, and for which argon ion lasers and frequency doubled Nd:YAG lasers provide excellent sources (Kuhn 1998). Most Ti:sapphire lasers are configured into an optical ring resonator arrangement with a set of birefringent filters for tuning. In addition, the ring cavity usually contains a Faraday rotator and wave plates to limit the propagation to only one direction around the ring. This arrangement is well suited for wide tuning and also mode locking, through which very short pulses, on the order of a few tens of femtoseconds can be produced. Ti:sapphire lasers have thus become a key resource for spectroscopy and research on ultrafast phenomena. The Ti:sapphire laser is also capable of average powers of up to several Watts, which makes it a viable tool for micromachining. Although its operation is at longer wavelengths than those normally preferred for micromachining, its capability for tuning and producing ultrashort pulses makes it attractive for research in this area. Since it requires a pump laser of ~10 W which is already in the green, and its more complicated optical system requires more maintenance and user savvy, it is presently not a common choice for industrial micromachining applications, but this may change in the future. There are many other new laser systems under development which offer efficient generation of green light at the power levels and pulse widths required for micromachining. It is worthwhile to realize that the field of laser sources is constantly

micromachining applications (Gower, 2000).

changing.

In general, the lasers best suited for micromachining are those that produce short pulses of high intensity at short wavelengths. Pulse widths of less than a microsecond are needed to allow the formed plasma to extinguish in between pulses so that subsequent pulses are not scattered and absorbed. Time for the debris plume to clear takes longer, often up to tens or hundreds of milliseconds, but its optical attenuation is usually less. Concentrating the laser radiation into short pulses of high intensity also has the benefit of more adiabatic heating of the substrate, bringing its temperature up to the vaporization point before too much of the heat can diffuse vertically and laterally away from the intended ablation zone. Shorter wavelengths generally have higher absorption coefficients in most materials, and they are thus absorbed nearer to the surface where the ablation is intended to occur. Shorter wavelengths can also be focused into a proportionally smaller diffraction-limited spot, which improves both the accuracy and precision of the ablation process. Typical working spot diameters for UV lasers in the 350 nm range are ~25 m, although this is larger than the theoretical diffraction limit.

#### **2. Fundamental laser micromachining processes**

Laser micromachining includes a number of different processes which are differentiated by the feature geometry and the manner in which material is removed from the substrate (Ion, 2005; Schuöcker, 1999). Cutting in this context refers to using the beam to slice all of the way through a thin sheet of substrate material, leaving behind a kerf which extends completely through to the opposite side of the substrate. As is commonly the case in laser cutting of sheet metal, the material removed from the kerf is predominantly ejected out the opposite side. Ablating is usually taken to mean removal of material in a thin layer from one side only, giving only partial penetration into the thickness of the substrate, and the removed material must necessarily be ejected from the same side as which the laser is incident. In both cases, the newly removed material is ejected primarily through the kerf which has just previously been cut and which trails along behind the laser beam as it is moved along the tool path. Whereas cutting and ablating can create geometries of any shape, drilling refers to the creation of a nominally circular hole with minimal lateral translation of the beam, with either through or blind penetration. If the laser beam is held in one fixed position and pulsed, the process often termed percussion drilling, whereas if the beam is swept around in a circular pattern to first roughly remove the bulk material and then completed with a fine finishing pass to accurately define the perimeter, the process is called trepanning. Percussion drilling produces holes whose diameter is roughly the same as the diameter of the laser beam, while trepanning produces holes whose diameter is larger than the beam diameter. Because drilling does not produce a trailing kerf, all removed material must be ejected from the same side as which the laser beam was incident, and drilling is thus necessarily an ablative process, regardless of whether it creates a through or blind hole (Voisey et al., 2003).

The removal of material can involve both thermal and chemical processes, depending upon how the laser radiation interacts with the substrate. At longer wavelengths, the photon energy is insufficient to provide anything more than simple heating of the substrate. At sufficiently high intensities, however, the heating can be concentrated enough to first melt the substrate material within a localized zone, and then vaporize it in those areas where the laser intensity and subsequent heating is higher. The substrate material is thus removed via a transition to the gas phase, although the vaporized material is often subsequently ionized by the laser radiation, leading to a plasma and plume that can have the effect of occluding the incident beam. It is customary to identify three zones around the incident beam: the heat-affected zone or HAZ, the melt zone, and the vaporization zone. Some materials can pass directly from the solid phase into the vapor phase by sublimation, and for these the melt zone is absent. Both melting followed by vaporization or direct sublimation are purely thermal ablation processes.

At shorter wavelengths, the photon energy may reach the level of the chemical bond strength of the substrate. Laser radiation may then break these chemical bonds through direct photon absorption, leading to volatilization of the substrate into simpler compounds. For most organic polymers, this photolysis process produces mainly H2O and CO2. This occurs typically for photon energies above 3.5 eV, or for wavelengths shorter than ~350 nm, i.e. into the near UV part of the spectrum. Because the photon energy is lost to chemical bond scission, the heating effects of the beam are greatly reduced, and this regime is sometimes referred to as "cold laser machining," non-thermal ablation, or photochemical ablation. This greatly reduces the transient thermal stresses that occur as part of thermal ablation, and the result is less bowing, warping, and delamination of the substrate, as well as fewer edge melting effects which degrade feature accuracy (Yung et al., 2002). Since the peak temperature rise is greatly reduced, conductive heat flow away from the irradiation area is also reduced, and better dimensional control of the micromachined structure is obtained. There has been a general trend toward using shorter wavelength lasers for micromachining over the past two decades of development. Currently, UV lasers in the 350 to 250 nm range dominate the industrial market for the above reasons.

Thermal ablation and photochemical ablation are two ideal extremes, and laser micromachining can often involve a combination of both for any given material or composite. In addition, there are several secondary processes which can arise due to the steep temperature gradients which are produced. If the laser beam is composed of short, high-intensity pulses, as would be typical for Q-switched systems, then the adiabatic heating of the substrate can cause sufficiently high temperature gradients for which differential thermal expansion and acoustic shock can produce surface cracking or spalling of the substrate (Zhou et al., 2003). Micron-sized flakes of the substrate can be explosively ejected from this process without requiring the additional thermal energy to fully vaporize the material. This is typically more prevalent for brittle materials with low thermal conductivity, e.g. ceramics and some glasses. For materials which readily oxidize, the rapid cycle of laser heating and cooling of the melt zone can cause the formed oxide film to flake off in chips from the compressive stress that was built into the oxide during the process. This is typically more prevalent for reactive metals such as chromium, nickel, iron, and copper. Thermal spalling and oxide chipping both create debris particles which are significantly larger than the redeposition of fully vaporized substrate material. Because both thermal spalling and oxide chipping occur after the melt zone has refrozen, they leave behind a surface finish which is typically more frosted or matte in visual appearance, and microscopically cusped on a smaller scale.

Inherent to all laser micromachining is the creation of a plume of ejected material, either fully vaporized or sometimes containing micron-sized debris flakes. This plume requires time to disperse, and if the next laser pulse arrives before this takes place, the laser radiation will usually produce some degree of ionization as it is absorbed by the vapor. This ionization of the vaporized material produces a plasma which, in addition to being fairly energetic and reactive, can absorb the laser radiation further, sometimes occluding the path for the beam to reach the substrate (Eloy, 1987). This luminous plasma is what is usually responsible for the "sparkles" that mark the travel of the laser beam across the substrate. Achieving beam positioning and pulse timing to avoid the plasma and plume occlusion of the beam is a central part of tuning the recipe for any laser micromachining. This problem is generally severe in continuous wave (CW) laser micromachining, but greatly reduced for pulsed lasers which are Q-switched. While the complete plume of vaporized material usually does not have time to fully disperse in between Q-switched pulses, the more optically opaque and higher density plasma does, and laser ablation can continue onward with usually only minor attenuation. If the beam positioning is not well designed, however, the plasma and plume can become trapped into the confined spaces of the kerf, and greater time will be required for their dispersal. The most common symptom of this effect is a reduced depth of ablation for a given beam transversal rate.

#### **2.1 Ablation process models**

40 Micromachining Techniques for Fabrication of Micro and Nano Structures

by the laser radiation, leading to a plasma and plume that can have the effect of occluding the incident beam. It is customary to identify three zones around the incident beam: the heat-affected zone or HAZ, the melt zone, and the vaporization zone. Some materials can pass directly from the solid phase into the vapor phase by sublimation, and for these the melt zone is absent. Both melting followed by vaporization or direct sublimation are purely

At shorter wavelengths, the photon energy may reach the level of the chemical bond strength of the substrate. Laser radiation may then break these chemical bonds through direct photon absorption, leading to volatilization of the substrate into simpler compounds. For most organic polymers, this photolysis process produces mainly H2O and CO2. This occurs typically for photon energies above 3.5 eV, or for wavelengths shorter than ~350 nm, i.e. into the near UV part of the spectrum. Because the photon energy is lost to chemical bond scission, the heating effects of the beam are greatly reduced, and this regime is sometimes referred to as "cold laser machining," non-thermal ablation, or photochemical ablation. This greatly reduces the transient thermal stresses that occur as part of thermal ablation, and the result is less bowing, warping, and delamination of the substrate, as well as fewer edge melting effects which degrade feature accuracy (Yung et al., 2002). Since the peak temperature rise is greatly reduced, conductive heat flow away from the irradiation area is also reduced, and better dimensional control of the micromachined structure is obtained. There has been a general trend toward using shorter wavelength lasers for micromachining over the past two decades of development. Currently, UV lasers in the 350

Thermal ablation and photochemical ablation are two ideal extremes, and laser micromachining can often involve a combination of both for any given material or composite. In addition, there are several secondary processes which can arise due to the steep temperature gradients which are produced. If the laser beam is composed of short, high-intensity pulses, as would be typical for Q-switched systems, then the adiabatic heating of the substrate can cause sufficiently high temperature gradients for which differential thermal expansion and acoustic shock can produce surface cracking or spalling of the substrate (Zhou et al., 2003). Micron-sized flakes of the substrate can be explosively ejected from this process without requiring the additional thermal energy to fully vaporize the material. This is typically more prevalent for brittle materials with low thermal conductivity, e.g. ceramics and some glasses. For materials which readily oxidize, the rapid cycle of laser heating and cooling of the melt zone can cause the formed oxide film to flake off in chips from the compressive stress that was built into the oxide during the process. This is typically more prevalent for reactive metals such as chromium, nickel, iron, and copper. Thermal spalling and oxide chipping both create debris particles which are significantly larger than the redeposition of fully vaporized substrate material. Because both thermal spalling and oxide chipping occur after the melt zone has refrozen, they leave behind a surface finish which is typically more frosted or matte in visual appearance, and

Inherent to all laser micromachining is the creation of a plume of ejected material, either fully vaporized or sometimes containing micron-sized debris flakes. This plume requires time to disperse, and if the next laser pulse arrives before this takes place, the laser radiation will usually produce some degree of ionization as it is absorbed by the vapor. This ionization of the vaporized material produces a plasma which, in addition to being fairly energetic and reactive, can absorb the laser radiation further, sometimes occluding the path

to 250 nm range dominate the industrial market for the above reasons.

microscopically cusped on a smaller scale.

thermal ablation processes.

Laser ablation involves a complex interaction between optical, thermal, and chemical processes, but some simplifications can lead to models which can be useful for characterization, optimization, and troubleshooting of the process. Most such models start with the optics of a Gaussian beam and compute the conductive flow of heat from this source to find the temperature distribution, adding in the thermal effects which are needed to account for melting and vaporization of the substrate (Engin & Kirby, 1996; Kaplan, 1996; Olson & Swope, 1992). An idealized geometry is illustrated in Fig. 1 in which a circularly symmetric Gaussian laser beam is moved across the substrate at a constant speed *v* in the *+x* direction. The beam has an average power of *P0 = πrB2I0*, where *I0* is the peak intensity and *rB* is the *1/e* beam radius. The beam propagates in the *+z* direction and meets the substrate surface in the x-y plane. The situation is more easily described by using the relative coordinate *ξ = x − vt* which moves along with the laser beam.

The interaction of the laser beam with the substrate first involves absorption of the optical radiation and its conversion into heat for thermal (non-photo-chemical) ablation. Shorter wavelengths are absorbed more strongly at the surface with a higher absorption coefficient , and since this is usually ~104 cm−1 or greater, the heating is effectively concentrated at the surface of the substrate. Volumetric heating effects have been considered by Zhang, et al. (2006). The surface heating density is then

$$q(\xi, y) = (1 - R)I\_0 \exp\left(-\frac{\xi^2 + y^2}{r\_B^2}\right) \qquad \text{[ $W / m^2$ ]}\text{.]}$$

where *R* is the reflectivity loss from the surface of the substrate. The heat transfer within the substrate is entirely by conduction, so the resulting temperature field is given by a solution to the heat conduction equation (Carslaw & Jaeger, 1959)

$$\frac{\partial T}{\partial t} - D\nabla^2 T = 0\_\prime$$

where *D = κ/ρC* is the thermal diffusivity, *κ* is the thermal conductivity, *ρ* is the mass density, and *C* is the specific heat capacity. The surface heating density provides a source boundary condition for the solution of the heat conduction equation. Ashby and Easterling (1984) have shown that a close analytical approximation to the solution of this problem is given by

$$T(\xi=0,y,z,t) - T\_0 = \frac{(1-R)P\_0}{2\pi\kappa v \left[t(t+r\_\mathrm{B}^2/D)\right]^{1/2}} \exp\left(-\frac{(z+z\_0)^2}{4Dt} - \frac{y^2}{4Dt+r\_\mathrm{B}^2}\right)\zeta$$

where *T0* is the initial temperature of the substrate, and *z0* is a parameter chosen to eliminate the surface singularity as *t → 0*.

Fig. 1. Geometry and intensity and temperature profiles for laser ablation.

*(1 R)P (z z ) <sup>y</sup> T( 0,y,z,t) T exp , 4Dt 4Dt r 2 v t(t r / D)* 

where *T0* is the initial temperature of the substrate, and *z0* is a parameter chosen to eliminate

Fig. 1. Geometry and intensity and temperature profiles for laser ablation.

the surface singularity as *t → 0*.

*0 0 <sup>0</sup> 1/2 <sup>2</sup> <sup>2</sup> <sup>B</sup> <sup>B</sup>*

*2 2*

With a sufficiently large laser average power *P0* and a sufficiently slow beam traversal rate *v*, the resulting temperature field can produce first melting and then vaporization of the substrate. Three zones are commonly defined based upon the phase changes: a heat affected zone or HAZ with a radius *rH*, a melted zone with a radius *rM*, and a vaporized zone with a radius *rV*, which forms the final kerf of width *2rV*. For simplicity, the depth of cut is taken to be *L* for all three of these zones, as shown in Fig. 1. These radii are defined by the points at which the peak temperature equals the melting point *TM* or the vaporization point *TV* for the substrate material. It is important to recognize that these three radii are dependent upon the beam radius *rB*, but are not equal to it. Similarly, the radial temperature distribution is not the same as the incident Gaussian beam shape.

In addition to simply raising the temperature of the substrate material, the incident laser power must also be used to change the phase of the material, first from solid to liquid, and then from liquid to vapor, in the case of simple thermal ablation. This energy balance is an important aspect of the ablation process model, and it can be described by the following conservation of energy relation,

$$\mathbb{E}(1-R)P\_0 = \frac{2\pi\mathbf{x}L(T\_M - T\_0)}{\ln(r\_H/r\_M)} + \mathbb{2}(r\_M - r\_V)\mathbb{L}\nu\mathbf{E}\_M + \mathfrak{p}\mathbf{C}(r\_M - r\_V)\mathbb{L}\nu(T\_V - T\_M) + 2r\_V\mathbb{L}\nu\mathbf{E}\_V, \quad \mathbf{x} \in \Omega$$

where *(1 − R)P0* is the optical power that is absorbed by the substrate, *v* is the beam traversal speed, and *L* is the depth of the cut. The first term on the right hand side is the power required to bring the temperature of the substrate up to the melting point *TM* at the inner boundary of the HAZ. The thermal conductivity is *κ = ρCD*. Within the melted zone, *rM < r < rV*, additional power must be added for the melting phase transition, as well as to raise the temperature up to the vaporization point *TV* at the inner boundary of the melted zone. The rate at which the beam sweeps out new material volume to melt is *2(rM − rV)Lv* [m3/s], and the latent heat of melting is *EM* [J/m3], which together give the second term. The third term on the right hand side is the power required to support the temperature difference of *TV − TM* across the melted zone. Similarly, the fourth term on the right hand side is the additional power required to vaporize the material in the kerf, where *EV* is the latent heat of vaporization [J/m3]. The latent heats of melting and vaporization are effectively constants which subtract from the applied optical power that falls within the melted and vaporized zones.

Most notable in the energy balance equation is the direct tradeoff that exists in the last three terms between the depth of cut *L* and beam traversal speed *v*. This makes the assumption that the incident beam does not become occluded by the features or debris that the ablation process creates. For CW laser cutting the debris plume and plasma can significantly attenuate the beam, leading to a reduction in the optical power that is available for subsequent ablation. Q-switched lasers with nanosecond pulses suffer far less from this problem, since the time between laser pulses allows the plasma time to extinguish (Chang & Warner, 1996). The above energy balance equation works fairly well for cutting depths *L* of up to a few beam diameters (Yuan & Das, 2007), but when the aspect ratio of the kerf becomes extreme, *L >> 2rV*, the sidewalls of the kerf will lead to beam reflections and scattering, and the change in depth may take the beam interaction beyond its depth of focus, both of which will have the effect of reducing the available intensity and slowing down the vertical ablation rate for deeper cuts (Bang et al., 1993). The concave bottom of laser drilled holes may also defocus the beam (Vatsya et al.,2003; Zhang ei al.,2008). This basic model has been extended to include the effects present in trepanning of holes (Zeng et al., 2005), and for trepanning with annular beam profiles (Zeng et al., 2006).

Laser ablation departs somewhat from the above model when the process involves a photochemical component. In this case, a significant fraction of the photons are absorbed directly for the process of breaking chemical bonds in the substrate, and these photons do not produce direct substrate heating, as would be the case for pure thermal ablation. Process models for this situation must break the photon flux into a thermally absorbed portion and a photochemical portion. The thermal portion behaves as per the above model description, while the photochemical portion creates volatilized products in proportion to the energy density of the specific bonds which are broken. Simple energy balance arguments are useful for predicting the photochemical ablation rate, under the assumption that any left-over energy that does not directly produce photochemical ablation is directed toward substrate heating of the same region. The relative split of the incident photon flux between thermal and photochemical ablation is usually taken to be proportional to the relative absorption coefficients of the two processes. However, it must be cautioned that the appropriate absorption coefficients are themselves temperature dependent and proper modeling of the optical absorption becomes a central problem in any multi-physics simulation of ablation.

#### **2.2 Optical absorption**

The most important principle of laser micromachining is that the laser output wavelength must be one which is strongly absorbed by the material to be processed. If the material is highly transparent at the wavelength of the laser, then no optical absorption and energy transfer will take place. For semiconductors and other crystalline materials, this normally means that the photon energy must be greater than the energy bandgap. For polymers and other amorphous materials, the photon energy must be greater than the energy difference between the lowest unoccupied molecular orbital (LUMO) and the highest occupied molecular orbital (HOMO). For both of these cases, that usually entails a laser emitting in the visible or UV with photon energies of ~1 eV or greater.

#### **2.2.1 Absorption of laser radiation**

The primary interaction between laser radiation and a solid is photochemical excitation of electrons from their equilibrium states to some excited states by the absorption of photons. Some of these transitions are schematically shown in Fig. 2. Interband transitions take place when photon energy is larger than bandgap of the material. In this process, electron-hole pairs are generated. The free electrons may jump back from conduction band to valence band through thermal (dashed lines) or photochemical processes. If the photon energy is less than bandgap of the material, the energy can be absorbed by defect levels in the banbgap or produce Intraband transitions. Both transitions will induce thermal processes as electrons jump back to valence band. With higher laser light intensities, multi-photon absorption is favored, because the probability of non-linear absorption increases strongly with laser intensity. The coherent multi-photon transitions would generate electron-hole pairs similar to interband transitions. (Linde et al., 1997)

Thus, the initial electronic excitation is followed by complex secondary processes, which can be classified into thermal and photochemical processes. The type of interaction between laser radiation and the material depends on laser parameters (wavelength, pulse duration, and fluence) and on the properties of the materials (Baeuerle, 2000; Mai & Nguyen, 2002). Laser ablation (material removal) can be analyzed on the basis of photothermal (purely pyrolytic), photochemical (purely photolytic), and photophysical processes, in which both thermal and non-thermal mechanisms contribute to the overall ablation rate.

#### **2.3 Thermal process**

44 Micromachining Techniques for Fabrication of Micro and Nano Structures

been extended to include the effects present in trepanning of holes (Zeng et al., 2005), and

Laser ablation departs somewhat from the above model when the process involves a photochemical component. In this case, a significant fraction of the photons are absorbed directly for the process of breaking chemical bonds in the substrate, and these photons do not produce direct substrate heating, as would be the case for pure thermal ablation. Process models for this situation must break the photon flux into a thermally absorbed portion and a photochemical portion. The thermal portion behaves as per the above model description, while the photochemical portion creates volatilized products in proportion to the energy density of the specific bonds which are broken. Simple energy balance arguments are useful for predicting the photochemical ablation rate, under the assumption that any left-over energy that does not directly produce photochemical ablation is directed toward substrate heating of the same region. The relative split of the incident photon flux between thermal and photochemical ablation is usually taken to be proportional to the relative absorption coefficients of the two processes. However, it must be cautioned that the appropriate absorption coefficients are themselves temperature dependent and proper modeling of the optical absorption becomes a central problem in any multi-physics simulation of ablation.

The most important principle of laser micromachining is that the laser output wavelength must be one which is strongly absorbed by the material to be processed. If the material is highly transparent at the wavelength of the laser, then no optical absorption and energy transfer will take place. For semiconductors and other crystalline materials, this normally means that the photon energy must be greater than the energy bandgap. For polymers and other amorphous materials, the photon energy must be greater than the energy difference between the lowest unoccupied molecular orbital (LUMO) and the highest occupied molecular orbital (HOMO). For both of these cases, that usually entails a laser emitting in

The primary interaction between laser radiation and a solid is photochemical excitation of electrons from their equilibrium states to some excited states by the absorption of photons. Some of these transitions are schematically shown in Fig. 2. Interband transitions take place when photon energy is larger than bandgap of the material. In this process, electron-hole pairs are generated. The free electrons may jump back from conduction band to valence band through thermal (dashed lines) or photochemical processes. If the photon energy is less than bandgap of the material, the energy can be absorbed by defect levels in the banbgap or produce Intraband transitions. Both transitions will induce thermal processes as electrons jump back to valence band. With higher laser light intensities, multi-photon absorption is favored, because the probability of non-linear absorption increases strongly with laser intensity. The coherent multi-photon transitions would generate electron-hole

Thus, the initial electronic excitation is followed by complex secondary processes, which can be classified into thermal and photochemical processes. The type of interaction between laser radiation and the material depends on laser parameters (wavelength, pulse duration, and fluence) and on the properties of the materials (Baeuerle, 2000; Mai & Nguyen, 2002).

for trepanning with annular beam profiles (Zeng et al., 2006).

the visible or UV with photon energies of ~1 eV or greater.

pairs similar to interband transitions. (Linde et al., 1997)

**2.2.1 Absorption of laser radiation** 

**2.2 Optical absorption** 

The thermal transition of the electrons can be described by the relaxation time *<sup>T</sup>* as shown in Fig. 2. When *<sup>T</sup>* is smaller than the time required for desorption of species from the surface, *<sup>R</sup>*, a photothermal process occurs. Thus, the photothermal ablation is based on the excitation energy being instantaneously transformed into heat. Due to the rapid dissipation of the excitation and ionization energy from the electrons to the lattice, the material surface is heated rapidly and vaporized explosively with or without surface melting. This regime applies to pulsed laser ablation by infrared- (IR-) and visible- (VIS-) laser radiation, and to most cases of ultraviolet- (UV-) laser radiation with nanosecond and longer pulses. These result in relatively high ablation rates and a rough surface finish (Baeuerle, 2000; Ehlich & Tsao, 1989; Luft et al., 1996; Schubart & Otto, 1997).

Fig. 2. Schematic of different types of electronic excitation in a solid.

With moderate-to-high laser fluences, and pulse lengths of nanoseconds, screening of the incident radiation by the vapor and plasma plume becomes important. Screening of the incident laser light by absorption and scattering within the vapor plume diminishes the intensity that reaches the substrate. The ablation rate depends on photon energy, laser fluence, spot size and material properties and is in a range between 0.1 and several m/pulse to be considered as a useful machining method. Additionally, with shorter wavelengths of laser radiation, the laser-plasma interaction becomes less pronounced (Baeuerle, 2000).

#### **2.4 Photochemical process**

If *T >> <sup>R</sup>*, the laser excitation can result in direct bond scission, the electrons freed from the broken bonds will be desorbed from the surface, and the process will be photochemical in nature. With purely photochemical (non-thermal) processes, the temperature of the system remains essentially unchanged under laser irradiation. The ablation rate is relatively slow ( 1m/pulse), but high surface quality can be achieved because of the absence of surface melting and explosive evaporation of the material (Baeuerle, 2000; Mai & Nguyen, 2002).

#### **3. Nd: YAG 266 nm and 355 nm laser micromachining**

#### **3.1 Laser ablation settings**

Chen and Darling (2005, 2008) have reported sysmatic studies of laser micromachining using Nd:YAG 266 nm and 355 nm lasers recently. An Electro-Scientific Industries (ESI) model 4440 laser micromachining system with a Light Wave Enterprises 210 diode-pumped frequency-tripled (355 nm) and a Photonics Industries diode-pumped frequencyquadrupled (266 nm) Nd:YAG laser were used to micromachine the samples. Output powers of the 355 nm laser were 4.8 W at repetition rate of 10 kHz and 3.0 W at repetition rate of 20 kHz, and that of the 266 nm laser was 0.5 W at repetition rate of 5 kHz. The stage was moved up and down to adjust the z-axis to focus and de-focus the laser beam on the samples. The x-y stage allowed scan speeds from 0 to 250 mm/s. The laser scan speed and repetition rate were adjusted to control the total energy of micromachining, and the focus/defocus was adjusted by moving z-axis stage up to control the laser fluences of the laser spots as shown in Table 1.

The microfluidic materials, such as sapphire, silicon and Pyrex, were micromachined by both 266 nm and 355 nm Nd:YAG lasers. A series of 1 mm 1 mm square cavities were created by laser micromachining with various laser machining conditions. Fig. 3 and Fig. 4 depict the typical laser micromachining cavities of sapphire and silicon. The ablation square cavities were inspected by an SEM and were measured for depth using an optical microscope and a scanning profilometer. Fig. 5 shows typical measurement data using a Tencor/KLA P-15 profilometer. A certain amount of solidified molten silicon remains in the ablation area after laser machining (Dauer et al., 1999). Thus, silicon wafers were cleaned and etched using a 22 wt% KOH solution at 75ºC for 4 minutes to clear the ablation debris.


aLaser focus at 0 z position, the z position shows the distance of the stage moving up; bLaser fluence was calculated by energy per pulse/ spot size area; fluence unit = J/cm2

Table 1. The fluences of the laser versus z-stage positions

nature. With purely photochemical (non-thermal) processes, the temperature of the system remains essentially unchanged under laser irradiation. The ablation rate is relatively slow ( 1m/pulse), but high surface quality can be achieved because of the absence of surface melting and explosive evaporation of the material (Baeuerle, 2000; Mai & Nguyen, 2002).

Chen and Darling (2005, 2008) have reported sysmatic studies of laser micromachining using Nd:YAG 266 nm and 355 nm lasers recently. An Electro-Scientific Industries (ESI) model 4440 laser micromachining system with a Light Wave Enterprises 210 diode-pumped frequency-tripled (355 nm) and a Photonics Industries diode-pumped frequencyquadrupled (266 nm) Nd:YAG laser were used to micromachine the samples. Output powers of the 355 nm laser were 4.8 W at repetition rate of 10 kHz and 3.0 W at repetition rate of 20 kHz, and that of the 266 nm laser was 0.5 W at repetition rate of 5 kHz. The stage was moved up and down to adjust the z-axis to focus and de-focus the laser beam on the samples. The x-y stage allowed scan speeds from 0 to 250 mm/s. The laser scan speed and repetition rate were adjusted to control the total energy of micromachining, and the focus/defocus was adjusted by moving z-axis stage up to control the laser fluences of the

The microfluidic materials, such as sapphire, silicon and Pyrex, were micromachined by both 266 nm and 355 nm Nd:YAG lasers. A series of 1 mm 1 mm square cavities were created by laser micromachining with various laser machining conditions. Fig. 3 and Fig. 4 depict the typical laser micromachining cavities of sapphire and silicon. The ablation square cavities were inspected by an SEM and were measured for depth using an optical microscope and a scanning profilometer. Fig. 5 shows typical measurement data using a Tencor/KLA P-15 profilometer. A certain amount of solidified molten silicon remains in the ablation area after laser machining (Dauer et al., 1999). Thus, silicon wafers were cleaned and etched using a 22 wt% KOH solution at 75ºC for 4 minutes to clear the ablation debris.

 Repetition Rate (Hz) 10k (355 nm) 20k (355 nm) 5k (266 nm) 0 866 271 50.93 300 96.24 30.01 24.02 600 34.65 10.83 13.93 900 17.68 5.52 9.08 1200 10.69 3.34 6.39 1500 7.16 2.24 4.73 1800 5.13 1.60 3.65 2100 3.85 1.20 2.90

Z-positiona (m) Fluenceb

aLaser focus at 0 z position, the z position shows the distance of the stage moving up; bLaser fluence was calculated by energy per pulse/ spot size area; fluence unit = J/cm2

Table 1. The fluences of the laser versus z-stage positions

**3. Nd: YAG 266 nm and 355 nm laser micromachining** 

**3.1 Laser ablation settings** 

laser spots as shown in Table 1.

Fig. 3. The SEM image of Nd:YAG laser micromachining on sapphire

Fig. 4. SEM image of Nd:YAG laser micromachining on silicon.

Fig. 5. Depth profiles for 355 nm Nd:YAG laser micromachining of sapphire with fluence of 9.27 J/cm2 and virious scan speeds.

#### **3.2 Laser micromachining ablation rate**

The ablation rates of the laser micromachining were calculated as:

Total removed volume of the material The number of total pulses ×spot size area

Figures 6 - 8 show the plots of ablation rates as a function of laser fluences with various laser scan speeds for sapphire, silicon and Pyrex using both Nd:YAG 266 nm and 355 nm lasers. It is observed that in the cases of both sapphire and Pyrex, the 266 nm laser provides higher ablation rates than the 355 nm laser under the same micromachining conditions. On the other hand, Fig. 7 (silicon) shows the varied ablation rates of Nd:YAG 355 nm laser micromachining using 20 mm/s and 50 mm/s scan speeds. The varied result is caused by the plume screening effect on the slower scan speed laser micromachining condition . In this case, the dwell time of laser light on the surface of the silicon is longer than the time for vapor/plasma formation, which attenuates the intensity of incident laser radiation.

The threshold fluences of laser micromachining of sapphire and silicon were calculated as shown in Table 2. All samples either did not exhibit fixed laser ablation threshold values or showed surface melting phenomena. Those results indicate that a thermal process was engaged in the laser micromachining of all the materials micromachined by both 266 nm and 355 nm Nd:YAG lasers. In general, the ablation rates using Nd:YAG 266 nm laser are higher than using 355 nm laser. This is due to the 266 nm laser producing a greater photochemical component.

Fig. 5. Depth profiles for 355 nm Nd:YAG laser micromachining of sapphire with fluence of

Total removed volume of the material The number of total pulses ×spot size area

Figures 6 - 8 show the plots of ablation rates as a function of laser fluences with various laser scan speeds for sapphire, silicon and Pyrex using both Nd:YAG 266 nm and 355 nm lasers. It is observed that in the cases of both sapphire and Pyrex, the 266 nm laser provides higher ablation rates than the 355 nm laser under the same micromachining conditions. On the other hand, Fig. 7 (silicon) shows the varied ablation rates of Nd:YAG 355 nm laser micromachining using 20 mm/s and 50 mm/s scan speeds. The varied result is caused by the plume screening effect on the slower scan speed laser micromachining condition . In this case, the dwell time of laser light on the surface of the silicon is longer than the time for

The threshold fluences of laser micromachining of sapphire and silicon were calculated as shown in Table 2. All samples either did not exhibit fixed laser ablation threshold values or showed surface melting phenomena. Those results indicate that a thermal process was engaged in the laser micromachining of all the materials micromachined by both 266 nm and 355 nm Nd:YAG lasers. In general, the ablation rates using Nd:YAG 266 nm laser are higher than using 355 nm laser. This is due to the 266 nm laser producing a greater

vapor/plasma formation, which attenuates the intensity of incident laser radiation.

9.27 J/cm2 and virious scan speeds.

photochemical component.

**3.2 Laser micromachining ablation rate** 

The ablation rates of the laser micromachining were calculated as:


Table 2. The threshold fluences of laser micromachining of sapphire and silicon

Fig. 6. The ablation rates for laser micromachining versus laser fluence for sapphire with different cutting speeds using 266 nm and 355 nm Nd:YAG lasers.

Fig. 7. The ablation rates of laser micromachining versus laser fluences for silicon with different cutting speeds using 266 nm and 355 nm Nd:YAG lasers.

Fig. 8. The ablation rates for laser micromachining versus laser fluences for Pyrex with different cutting speeds using 266 nm and 355 nm Nd:YAG lasers.

#### **3.3 The ablation efficiency**

The ablation efficiency was calculated by dividing the ablation rate by the energy per pulse to normalize the ablation rate performed by the 355 nm and 266 nm Nd:YAG lasers. Figures 9 - 11 show the plots of ablation efficiency as a function of laser fluence with various scan speeds using both lasers. The results indicate that at high laser fluences, the ablation efficiencies of the 266 nm laser are better than that of the 355 nm laser for all three materials.

Figure 10 (silicon) shows that the ablation rate of 266 nm Nd:YAG laser micromachining is slower than 355 nm laser micromachining under 50 mm/s scan speed after normalizing the ablation rate by energy per pulse. The result points out that at the laser fluences higher than 10 J/cm2, the ablation efficieny of the 266 nm laser is 1.5 times faster than that of the 355 nm laser at the scan speed of 50 mm/s, and 3.2 times faster in the case of 20 mm/s as shown in Table 3.


Table 3. The comparison of Nd:YAG 266 nm and 355 nm laser ablation efficiencies to sapphire, silicon and Pyrex with laser fluence larger than 10 J/cm2.

Fig. 8. The ablation rates for laser micromachining versus laser fluences for Pyrex with

The ablation efficiency was calculated by dividing the ablation rate by the energy per pulse to normalize the ablation rate performed by the 355 nm and 266 nm Nd:YAG lasers. Figures 9 - 11 show the plots of ablation efficiency as a function of laser fluence with various scan speeds using both lasers. The results indicate that at high laser fluences, the ablation efficiencies of the 266 nm laser are better than that of the 355 nm laser for all three

Figure 10 (silicon) shows that the ablation rate of 266 nm Nd:YAG laser micromachining is slower than 355 nm laser micromachining under 50 mm/s scan speed after normalizing the ablation rate by energy per pulse. The result points out that at the laser fluences higher than 10 J/cm2, the ablation efficieny of the 266 nm laser is 1.5 times faster than that of the 355 nm laser at the scan speed of 50 mm/s, and 3.2 times faster in the case of 20 mm/s as shown in

Table 3. The comparison of Nd:YAG 266 nm and 355 nm laser ablation efficiencies to

sapphire, silicon and Pyrex with laser fluence larger than 10 J/cm2.

Sapphire Silicon Pyrex

50 mm/s 20 mm/s

9 1.5 3.2 13

different cutting speeds using 266 nm and 355 nm Nd:YAG lasers.

**3.3 The ablation efficiency** 

Ablation Efficiency 266 nm/355 nm

materials.

Table 3.

Fig. 9. Laser ablation efficiency versus laser fluences for sapphire under different scan speeds using the 266 nm and 355 nm Nd:YAG lasers.

Fig. 10. Laser ablation efficiency versus laser fluence for silicon under different scan speeds using the 266 nm and 355 nm Nd:YAG lasers.

Fig. 11. Laser ablation efficiency versus laser fluence for Pyrex under different scan speeds using the 266 nm and 355 nm Nd:YAG lasers.

#### **3.4 The ablation precision of laser micromachining**

By computing the average ablation depths and standard deviation, the depth of laser micromachining can be characterized as:

> Average depth (mean) standard error (=2.58 standard deviation/ square root(sample size));

which give 99% of the cutting depths falling into this range (Lindgren et al., 1978), and the laser machining precision is defined as,

$$\text{Precision} = \text{2} \times \text{standard error} \text{ / average depth}$$

Figure 12 shows the plot of laser machining precision as a function of laser fluence using Nd:YAG 266 nm and 355 nm lasers with different scan speeds. The results portray the Nd:YAG 266 nm laser providing better precision than the 355 nm laser, and Nd:YAG laser micromachining more generally providing better precision in the order of sapphire, silicon and then Pyrex.

#### **4. CO2 laser cutting of microfluidic plastic laminates**

CO2 lasers have become the most used laser system for industrial fabrication and materials processing. This is due to a combination of their relatively low cost, high optical power and efficiency, and robust operation over a long service life. They are routinely applied to an extremely wide range of material processing, including scribing, marking, drilling, cutting, and heat treating of metals, ceramics, and polymers. CO2 laser processing has also been

Fig. 11. Laser ablation efficiency versus laser fluence for Pyrex under different scan speeds

By computing the average ablation depths and standard deviation, the depth of laser

Average depth (mean) standard error (=2.58 standard deviation/ square root(sample size)); which give 99% of the cutting depths falling into this range (Lindgren et al., 1978), and the

Precision = 2 standard error / average depth Figure 12 shows the plot of laser machining precision as a function of laser fluence using Nd:YAG 266 nm and 355 nm lasers with different scan speeds. The results portray the Nd:YAG 266 nm laser providing better precision than the 355 nm laser, and Nd:YAG laser micromachining more generally providing better precision in the order of sapphire, silicon

CO2 lasers have become the most used laser system for industrial fabrication and materials processing. This is due to a combination of their relatively low cost, high optical power and efficiency, and robust operation over a long service life. They are routinely applied to an extremely wide range of material processing, including scribing, marking, drilling, cutting, and heat treating of metals, ceramics, and polymers. CO2 laser processing has also been

using the 266 nm and 355 nm Nd:YAG lasers.

micromachining can be characterized as:

laser machining precision is defined as,

and then Pyrex.

**3.4 The ablation precision of laser micromachining** 

**4. CO2 laser cutting of microfluidic plastic laminates** 

Fig. 12. Laser micromachining precision versus laser fluences for sapphire, silicon and Pyrex using the 266 nm and 355 nm Nd:YAG lasers.

extensively applied to the field of microfluidics, principally in the form of through-cutting of plastic laminates. A great many applications for microfluidics demand disposable cartridges for the liquid contacting elements of the system. Disposable cartridges, in turn, demand extremely low cost materials and fabrication methods, often in the range of pennies per part, to be competitive in the marketplace. One approach, which has gained great popularity over the past decade, is the construction of microfluidic cartridges from a series of laser-cut plastic laminates which are aligned and bonded together. This method of fabrication offers enormous flexibility in both the design of the microfluidic plumbing as well as the materials which are used to create it.

One example of a fairly advanced microfluidic cartridge created as a bonded stack of lasercut plastic laminates is shown in Fig. 13. (Lafleur, 2010). As illustrated, this type of microfluidic cartridge can utilize both thick, rigid layers as well as thinner, flexible layers in its construction, allowing channel thicknesses from a few mils up to several mm to be created. The layers can be aligned and bonded together using a variety of techniques, including heat fusing, heat staking, solvent welding, or through the use of adhesives which are either applied directly, or which can be a pressure-sensitive adhesive which comes on one or both sides of a given layer. The cartridge shown in Fig. 13 only uses 6 layers, but cartridges employing over 20 layers are becoming more routine (Lafleur, 2010). Common structural materials for plastic laminate microfluidics include polymethyl methacrylate (PMMA), polyethylene (PE), polycarbonate (PC), and acetate. In addition, semi-permeable membranes such as Nafion and nitrocellulose are frequently employed. As is true for other types of microfluidic systems, the control of surface hydrophobicity / hydrophilicity is of paramount concern, and plays a predominant role in the materials selection.

Fig. 13. A laser-cut plastic laminate microfluidic cartridge for carrying out an immunoassay. From Lafleur (2010).

#### **5. Discussion**

The laser ablation processes, thermal and photochemical, are determined by the materials properties. Figure 14 depicts the absorption coefficients of transparent materials, sapphire and Pyrex, and Table 4 shows some physical properties of those three materials.


\* Rough estimates of source evaporation temperatures are commonly based on the assumption that vapor pressures of 10-2 Torr must be established to produce efficient source removal rates (Maissel & Glang, 1970).

Table 4. Some physical properties of sapphire, silicon, and Pyrex (Chen & Darling, 2005, 2008)

Fig. 13. A laser-cut plastic laminate microfluidic cartridge for carrying out an immunoassay.

The laser ablation processes, thermal and photochemical, are determined by the materials properties. Figure 14 depicts the absorption coefficients of transparent materials, sapphire

Sapphire 7.8 2054 511 3 5.19 4.74 1800 Silicon 1.12 1414 326.810 2.0E6 1.07E6 1350 Pyrex 7.8 821 799.611.3 14.7 1.93 --

\* Rough estimates of source evaporation temperatures are commonly based on the assumption that vapor pressures of 10-2 Torr must be established to produce efficient source removal rates (Maissel &

Table 4. Some physical properties of sapphire, silicon, and Pyrex (Chen & Darling, 2005,

Absorption Coefficient@ 266nm(cm-1)

Absorption Coefficient@ 355nm(cm-1)

Evaporation Temp.\* (C)

and Pyrex, and Table 4 shows some physical properties of those three materials.

Bond strength (kJ/mol)

From Lafleur (2010).

Eg (eV)

Melting temp. (C)

**5. Discussion** 

Glang, 1970).

2008)

In general, the laser ablation rates of sapphire, silicon, and Pyrex micromachined by near UV (355 nm) and mid-UV (266 nm) nanosecond pulsed Nd:YAG lasers, are higher using the 266 nm laser than the 355 nm laser in the absence of plume screening effects. Under those high laser fluency micromachining conditions, non-linear optical phenomena such as multiphoton process become important, and the 266 nm laser (with photon energy = 4.66 eV) has a higher probability to induce photochemical process than the 355 nm laser (with photon energy = 3.50 eV). Therefore, the ablation rates increase more in the cases of wide bandgap materials, such as sapphire and Pyrex, than the increase in the case of narrow bandgap material, like silicon as laser fluence increasing.

Fig. 14. The absorption coefficients versus wavelength for the transparent materials tested.

Sapphire has relatively the same level of absorption at 266 nm and 355 nm, however, the 266 nm laser provides a higher ablation efficiency at a given laser fluence than the 355 nm laser caused by higher photochemical process contributing to the overall ablation. Therefore, 266 nm laser micromachining on sapphire would provide not only slighly better absorption but also higher probability of photochemical process than 355 nm laser. In the case of silicon with its narrow band gap and high absorption at both wavelengths, the ablation efficiencies are not much different between the 266 nm and 355 nm lasers.

Pyrex has a low melting temperature, a high bond strength, a low absorption coefficient, and a wide energy band gap, as shown in Table 4. This implies that a predominantly thermal process was engaged in the laser micromachining of Pyrex by the 266 nm and 355 nm lasers. However, Pyrex shows better ablation efficiency using 266 nm laser due to more photochemical process at the higher absorption coefficient and higher energy (Mai & Nguyen, 2002; Baeuerle, 2000; Lim & Mai, 2002; Craciun & Craciun, 1999; Craciun et al., 2002; Hermanns, 2000).

Laser micromachining of plastic laminates for microfluidics nearly always involves through-cutting of each layer. CO2 laser systems do not offer sufficient beam control to allow accurate machining to a prescribed depth, nor would the inhomogeneity of the plastic films support this type of machining. During the laser micromachining, plastic laminates are most often supported on mesh or grille working platens to allow the beam and the ablation debris to completely pass through to the other side without obstruction. Very thin, fragile or flexible materials, such as nitrocellulose membranes, are usually supported by a sacrificial backing piece, and for this situation, the laser micromachining reverts back to pure surface ablation with the debris exiting from the same side as which the laser was incident. The greatest issue with CO2 laser through-cutting of plastics is the degree of edge melting that occurs along the kerf. While the vaporization temperatures for most plastics are comparatively low, so are the melting temperatures, and the CO2 laser beam is both broad in diameter and deeply penetrating, all of which can combine to easily cause run-away heating of the areas surrounding the desired kerf. This is particularly a problem in CW CO2 systems. The most common approach to combating this problem is to tune the beam traversal speed to a fairly high value which produces a shallow depth of cut, and then to scan back and forth repeatedly until the full depth of cut is achieved. The time between successive passes is chosen to be greater than the time required for the substrate to cool back down to a stable working point. Through cutting of laminates does offer the advantage that larger cavities and channels can be created by simply tracing the beam around their edges and dropping out the waste as one single piece, as opposed to scanning back and forth to ablate away the entire volume. This conserves laser beam time, minimizes heating, and creates finished parts faster, with the only negative feature being the need to reliably capture the waste pieces so that they do not get caught in the remainder of the manufacturing process.

Nearly all of the materials used for plastic laminate microfluidics can also be readily photochemical ablated by UV lasers, usually producing harmless H2O and CO2 gas as by products. UV laser cutting of plastics is a premier method that gives the best geometrical accuracy due to the smaller beam spot and the photochemical ablation process which produces significantly less edge melting along the kerf. However, CO2 lasers still dominate the market for this type of machining as a result of their much lower cost and ease of use as compared to UV laser systems.

#### **6. Conclusion**

This chapter discusses the fundamentals of laser ablation in the microfabrication of microfluidic materials. The removal of material involves both thermal and chemical processes, depending upon how the laser radiation interacts with the substrate. At longer wavelengths and low laser fluencies, the thermal process dominates. While the photon energy of the laser radiation is sufficiently high, the laser radiation can provide heating, with or without melting the substrate material, and then vaporize it. At shorter wavelengths, the ablation process shifts to photochemical. The photon energy of laser radiation reaches the level of the chemical bond strength of the substrate, and then breaks these chemical bonds through direct photon absorption, leading to volatilization of the substrate into simpler compounds.

In the cases of the ablation rates of sapphire, silicon, and Pyrex, micromachined by near UV and mid-UV nanosecond pulsed Nd:YAG lasers. All three materials have higher ablation efficiencies using the 266 nm laser than the 355 nm laser due to better absorption and higher probability of photochemical process using 266 nm laser. The ablation efficiencies are increased more for the case of high melting temperature or/and finite absorption materials such as sapphire and Pyrex. The increase is less for narrow band gap or/and high absorption materials such as silicon.

Laser systems can micromachine materials all the way from lightweight plastics and elastomers up through hard, durable metals and ceramics by carefully selecting laser wavelengths, pulse duration, and fluencies. This versatility makes laser micromaching extremely attractive for prototyping and development, as well as for small to medium run manufacturing.

#### **7. References**

56 Micromachining Techniques for Fabrication of Micro and Nano Structures

Laser micromachining of plastic laminates for microfluidics nearly always involves through-cutting of each layer. CO2 laser systems do not offer sufficient beam control to allow accurate machining to a prescribed depth, nor would the inhomogeneity of the plastic films support this type of machining. During the laser micromachining, plastic laminates are most often supported on mesh or grille working platens to allow the beam and the ablation debris to completely pass through to the other side without obstruction. Very thin, fragile or flexible materials, such as nitrocellulose membranes, are usually supported by a sacrificial backing piece, and for this situation, the laser micromachining reverts back to pure surface ablation with the debris exiting from the same side as which the laser was incident. The greatest issue with CO2 laser through-cutting of plastics is the degree of edge melting that occurs along the kerf. While the vaporization temperatures for most plastics are comparatively low, so are the melting temperatures, and the CO2 laser beam is both broad in diameter and deeply penetrating, all of which can combine to easily cause run-away heating of the areas surrounding the desired kerf. This is particularly a problem in CW CO2 systems. The most common approach to combating this problem is to tune the beam traversal speed to a fairly high value which produces a shallow depth of cut, and then to scan back and forth repeatedly until the full depth of cut is achieved. The time between successive passes is chosen to be greater than the time required for the substrate to cool back down to a stable working point. Through cutting of laminates does offer the advantage that larger cavities and channels can be created by simply tracing the beam around their edges and dropping out the waste as one single piece, as opposed to scanning back and forth to ablate away the entire volume. This conserves laser beam time, minimizes heating, and creates finished parts faster, with the only negative feature being the need to reliably capture the waste pieces so that they do

Nearly all of the materials used for plastic laminate microfluidics can also be readily photochemical ablated by UV lasers, usually producing harmless H2O and CO2 gas as by products. UV laser cutting of plastics is a premier method that gives the best geometrical accuracy due to the smaller beam spot and the photochemical ablation process which produces significantly less edge melting along the kerf. However, CO2 lasers still dominate the market for this type of machining as a result of their much lower cost and ease of use as

This chapter discusses the fundamentals of laser ablation in the microfabrication of microfluidic materials. The removal of material involves both thermal and chemical processes, depending upon how the laser radiation interacts with the substrate. At longer wavelengths and low laser fluencies, the thermal process dominates. While the photon energy of the laser radiation is sufficiently high, the laser radiation can provide heating, with or without melting the substrate material, and then vaporize it. At shorter wavelengths, the ablation process shifts to photochemical. The photon energy of laser radiation reaches the level of the chemical bond strength of the substrate, and then breaks these chemical bonds through direct photon absorption, leading to volatilization of the

not get caught in the remainder of the manufacturing process.

compared to UV laser systems.

substrate into simpler compounds.

**6. Conclusion** 


laser ablation of silicon. *Appl. Surf. Sci.*, Vol. 186, No. 1-4, pp. 288–292, ISSN 0169- 4332


Crane, K. C. A. & Brown, J. R. (1981). Laser-induced ablation of fibre/epoxy composites. *J.* 

Crane, K. C. A. (1982). Steady-state ablation of aluminium alloys by a CO2 laser. *J. Phys. D:* 

Dauer, S., Ehlert, A., Buttgenbach, S. (1999). Rapid prototyping of micromechanical devices

Ehrlich, D.J. & Tsao, J.Y. (Eds.). (1989). *Laser Microfabrication: Thin Film Processes and lithography*, Academic Press, ISBN-10: 0122334302 Salt Lake City, USA Eloy, J.-F. (1987). *Power Lasers*, Halsted Press / John Wiley & Sons, ISBN 0-470-20851, New

Engin, D. & Kirby, K. W. (1996). Development of an analytical model for the laser machining

Gower, M. C. (2000). Industrial applications of laser micromachining. *Optics Express*, Vol. 7,

Hermanns, C. (2000). Laser cutting of glass, *Proc. SPIE 4102 international symposium on* 

Ion, J. C., (2005). *Laser Processing of Engineering Materials*, Elsevier Butterworth-Heinemann,

Kaplan, A. F. H. (1996). An analytical model of metal cutting with a laser beam. *J. Appl.* 

Knowles, M. R. H. (2000). Micro-ablation with high power pulsed copper vapor lasers.

Koechner, W. (1988). *Solid-State Laser Engineering* (2nd Ed.), Springer-Verlag, ISBN 0-387-

Kuhn, K. J. (1998). *Laser Engineering*, Prentice Hall, ISBN 0-02-366921-7, Upper Saddle River,

Lafleur, L. K. (2010). *Design and Testing of Pneumatically Actuated Disposable Microfluidic* 

Lash J. S. & Gilgenbach, R. M. (1993). Copper vapor laser drilling of copper, iron, and

Lim, G. C., & Mai, T.-A. (2002). Laser micro-fabrications: present to future applications, *Proc.* 

Lindgren, B.W., McElrath, G.W., Berry, D.A. (1978). *Introduction to Probability and Statistics*. (4th Ed.) Macmillan Publishing Co., ISBN: 0023709006, Basingstoke UK

*Devices for the DxBox: A Point-of-Care System for Multiplexed Immunoassay Detection in the Developing World*, Ph.D. Dissertation, The University of Washington, Seattle,

titanium foils in atmospheric pressure air and argon. *Rev. Sci. Inst.*, Vol. 64, No. 11,

*SPIE 4426 Second International Symposium on Laser Precision Microfabrication*, pp. 170-

ISBN 0-7506-6079-1, Oxford, UK / Burlington, Massachusetts

*Phys.*, Vol. 79, No. 5, pp. 2198-2208, ISSN: 0021-8979

*Optics Express*, Vol. 7, No. 2, pp. 50-55, ISSN: 1094-4087

*Phys. D: Appl. Phys.*, Vol. 14, No. 12, pp. 2341-2349, ISSN: 0022-3727

*Appl. Phys.*, Vol. 15, No. 10, pp. 2093-2098, ISSN: 0022-3727

*Actuators A*, Vol. A76, No. 1-3, pp. 381–385, ISSN: 0924-4247

4332

York, New York

ISSN: 0021-8979

No. 2, pp. 56-67, ISSN: 1094-4087

[California], USA, August 2000

18747-2, New York, New York

pp. 3308-3313, ISSN: 0034-6748

176, ISBN: 9780819441379, Singapore, May 2001

New Jersey

Washington

laser ablation of silicon. *Appl. Surf. Sci.*, Vol. 186, No. 1-4, pp. 288–292, ISSN 0169-

using a Q-switched Nd:YAG laser with optional frequency doubling. *Sens.* 

of ceramic and glass-ceramic materials. *J. Appl. Phys.*, Vol. 80, No. 2. pp. 681-690,

*Inorganic Optical Materials II*, pp. 219-226, ISBN: 9780819437471, San Diego,


Zhou, M., Zeng, D. Y., Kan, J. P., Zhang, Y. K., Cai, L., Shen, Z. H., Zhang, X. R., & Zhang, S. Y., (2003). Finite element simulation of the film spallation process induced by the pulsed laser peening. *J. Appl. Phys.*, vol. 94, no. 5, pp. 2968-2975, ISSN: 0021- 8979

## **Microwave Meta-Material Absorbers Utilizing Laser Micro-Machining Technology**

Hongmin Lee *Kyonggi University, Korea* 

#### **1. Introduction**

60 Micromachining Techniques for Fabrication of Micro and Nano Structures

Zhou, M., Zeng, D. Y., Kan, J. P., Zhang, Y. K., Cai, L., Shen, Z. H., Zhang, X. R., & Zhang,

8979

S. Y., (2003). Finite element simulation of the film spallation process induced by the pulsed laser peening. *J. Appl. Phys.*, vol. 94, no. 5, pp. 2968-2975, ISSN: 0021-

> Recently, artificially structured electromagnetic (EM) materials have become an extremely active research area because of the possibility of creating materials which exhibit novel EM responses not available in nature. This includes negative refractive index (NRI), super-lens, cloaking, and more generally, coordinating transformation materials. For the most part, these composites, often called meta-materials (MTMs). The double negative (DNG) MTM structure was realized in 2000 by appropriately depositing SRRs and thin-wires on dielectric substrate. Since then, most of reported designs have a 1D or 2D geometry that responds only to one (two) electrical and magnetic components of the electromagnetic fields. Much of the work in MTM has been focused on the real parts of permittivity and permeability to enable the creation of a negative refractive index material. However, they can be manipulated to create a high performance absorber. According to effective medium theory, MTMs can be represented by the complex electric permittivity εeff (= ε′ + jε″) and magnetic permeability μeff (= μ′ + jμ″). By varying the dimensions of electric and magnetic components, it is possible to adjust permittivity and permittivity independently. Additionally, by tuning the electric and magnetic resonances a MTM can be impedance matched to free space, resulting reflectivity R = 0. The additional multiple layers or metallic back-plate will also ensure transmission T = 0. As a result, 100 % absorbance A (= 1– R – T) is theoretically possible.

> The microwave absorbers are used in military application to reduce the radar cross-section (RCS) of a conducting object and electromagnetic (EM) interference among microwave components. One of the earliest approaches for the design of EM absorber structure is based on the use of Salisbury screen. This type of absorber needs the resistive sheet and a metallic ground plane. The metallic backing plays two main roles; 1) it is used to avoid power transmission on the other side of the absorber, 2) it cancels out a reflected component that combined with the impinging wave. Recently, advancement in absorber technology has been obtained by using artificially MTMs to create a high-performance absorber for the microwave and terahertz frequency regime. In practice, it is difficult to make the absorber's electrical size small enough at low frequencies. For the design of compact microwave absorbers made by MTM complimentary pairs, we need to choose proper unit cell structures which are characterized by oppositely signed values of real parts of permittivity and permittivity. However, the absorbers are usually made with metallic backing plates in order

to avoid power transmission on the absorbers' other side, which may represent many problems for stealth applications. In order to design a metallic backplane-less absorber with double- negative MTM unit cell structure, we may refer to the sketch shown in Fig. 1. The two metallic pattern layers separated by dielectric spacer can be placed either orthogonal to EM wave propagation direction or parallel. If the two metallic pattern layers are placed orthogonal to EM wave propagation direction, as shown in Fig. 1(a), the radar cross section (RCS) of the object may increase at frequencies other than aimed design frequency bands.

(a) Patterns placed orthogonal to propagation direction (b) Patterns placed parallel to propagation direction.

Fig. 1. Sketch describing the geometry of MTM absorber unit cell (where, k is the wave propagation direction).

In order to avoid this problem, the two metallic pattern layers would need to be paced parallel to EM wave propagation direction, as shown in Fig. 1(b). In this study, the prototype absorber resonator structures were fabricated using both of a surface micromachining process technique and a standard photolithography technique.

#### **2. Design of a miniaturized meta-material microwave absorber**

#### **2.1 Double negative unit cell design**

The practical implementation of a double negative MTM unit cell involves the proper choice of both the structures with the negative real part of the permittivity and the negative real part of the permeability. A single unit cell of the proposed absorber consisted of distinct metallic elements, as shown in Fig. 2(a) and 2(b). The electric responses were provided by two symmetrically placed open complimentary split ring resonators (OCSRRs), as shown in Fig 2(a). We have constructed NRI MTM unit cell using open complimentary split ring resonator (OCSRR) and split ring resonator (SRR) arrangement. The OCSRR has been derived from two former planar resonant structures: the open split ring resonator (OSRR) and the complimentary split ring resonator (CSRR). As compared to SRR and CSRR, the electrical size of OCSRR is smaller and it can be modeled as an open parallel resonant circuit. The OCSRR is modified CSRR structure exhibiting negative permittivity and the SRR structure exhibits negative permeability. Each unit cell is printed on the two side of a FR-4 substrate. We use a double-layered structure with a SRR and two OCSRRs which are put on top of each other to make a miniaturized MTM absorber unit cell for 2 GHz frequency band.

to avoid power transmission on the absorbers' other side, which may represent many problems for stealth applications. In order to design a metallic backplane-less absorber with double- negative MTM unit cell structure, we may refer to the sketch shown in Fig. 1. The two metallic pattern layers separated by dielectric spacer can be placed either orthogonal to EM wave propagation direction or parallel. If the two metallic pattern layers are placed orthogonal to EM wave propagation direction, as shown in Fig. 1(a), the radar cross section (RCS) of the object may increase at frequencies other than aimed design frequency bands.

Fig. 1. Sketch describing the geometry of MTM absorber unit cell (where, k is the wave

In order to avoid this problem, the two metallic pattern layers would need to be paced parallel to EM wave propagation direction, as shown in Fig. 1(b). In this study, the prototype absorber resonator structures were fabricated using both of a surface micromachining

The practical implementation of a double negative MTM unit cell involves the proper choice of both the structures with the negative real part of the permittivity and the negative real part of the permeability. A single unit cell of the proposed absorber consisted of distinct metallic elements, as shown in Fig. 2(a) and 2(b). The electric responses were provided by two symmetrically placed open complimentary split ring resonators (OCSRRs), as shown in Fig 2(a). We have constructed NRI MTM unit cell using open complimentary split ring resonator (OCSRR) and split ring resonator (SRR) arrangement. The OCSRR has been derived from two former planar resonant structures: the open split ring resonator (OSRR) and the complimentary split ring resonator (CSRR). As compared to SRR and CSRR, the electrical size of OCSRR is smaller and it can be modeled as an open parallel resonant circuit. The OCSRR is modified CSRR structure exhibiting negative permittivity and the SRR structure exhibits negative permeability. Each unit cell is printed on the two side of a FR-4 substrate. We use a double-layered structure with a SRR and two OCSRRs which are put on top of each other to make a miniaturized MTM absorber unit cell for 2 GHz frequency band.

(b) Patterns placed parallel to propagation direction.

(a) Patterns placed orthogonal to propagation direction

process technique and a standard photolithography technique.

**2. Design of a miniaturized meta-material microwave absorber** 

propagation direction).

**2.1 Double negative unit cell design** 

The magnetic responses were provided by two spirals, as shown in Fig. 2(b). We created electromagnetic responses by the OCSRRs with two spirals in a parallel plane separated by a lossy dielectric substrate, as shown in Fig. 2(c). The absorber unit cell is made of a FR-4 substrate whose relative dielectric constant is εr = 4.4, and loss angle tangent tan δ = 0.025, and thickness t = 0.8 mm. The metal for metallic patterns is a copper whose conductivity is σ = 5.8 ⨉ 107S/m. By changing the geometry and the separation between the OCSRRs and the spirals the electromagnetic responses are tuned to match the impedance to free space and minimize the transmission at the aimed design frequency. Computer simulations for one unit cell are carried out using the commercial finite-difference time domain solver Microwave Studio by CST. The program simulated a single unit cell with appropriate boundary conditions, as shown in Fig. 2(c). The perfect electric conductor (PEC) boundary conditions are applied to the top and bottom walls of the waveguide, where as perfect magnetic conductor (PMC) boundary conditions are applied to the side walls of the waveguide. The other two opposite sides of the waveguide is assigned as waveguide ports. The total dimension of a cell is 7.3 mm ⨉ 7 mm ⨉ 0.8 mm. A single unit cell is placed inside a waveguide, and a vertically polarized transverse electromagnetic (TEM) wave is incident normally on the front side of port 1, as shown in Fig. 2(c). The scattering parameters of this MTM unit cell were then simulated, and the absorbance was calculated using the equation A = 1- ∣S11∣2- ∣S21∣2. The simulated magnitudes of S11 and S21 parameters are plotted in Fig. 3(a). We observe that both the reflection and transmission are very low at the resonance frequency of 2.43 GHz, which indicates a strong absorption of the EM wave energy.

 In order to express the effective permittivity and permeability of artificial material in terms of the scattering parameters, they are conventionally retrieved from scattering parameters of a unit cell under plane wave excitation [11]. The impedance parameters and ABCD parameters for two- port network can be calculated from scattering parameters using simple transformation. Then the Bloch-Floquet theorem was used to calculate the Bloch impedance ZB, and complex propagation constant �.

$$\mathbf{y} = \cos^{\text{l}}((\mathbf{Z}\mathbf{1}^{\text{l}} + \mathbf{Z}\mathbf{2})/\mathbf{Z}\mathbf{2}\mathbf{1})/\mathbf{p}.\tag{1}$$

Where p is the size of the MTM unit cell, and the Bloch impedance ZB can be expressed as

$$\mathbf{Z}\_{\text{B}} = \mathbf{B} / (\mathbf{e} \text{tr} \mathbf{v} \text{-A}). \tag{2}$$

Where, the parameter A is the voltage ratio between two ports with open-circuit at port 2 and the parameter B is the trans-admittance with short-circuit at port 2 can be expressed using impedance parameters as

$$\mathbf{A} = \mathbf{Z}\_{11}/\mathbf{Z}\_{22}, \mathbf{B} = (\mathbf{Z}\_{11}\mathbf{Z}\_{22} - \mathbf{Z}\_{21}\mathbf{2})/\mathbf{Z}\_{21}.\tag{3}$$

The effective permittivity εeff and permeability μeff can then easily calculated from Bloch impedance and propagation constant with the free space wave number k0, and wave impedance Z0 of the empty waveguide, respectively.

$$
\hbar \mathbf{r}^{\text{eff}} = (\mathbf{k} \mathbf{Z}^{\text{B}}) / (\mathbf{k}^{\text{0}} \mathbf{Z}^{\text{0}}). \tag{4}
$$

$$\mathbf{x}\_{\rm eff} = (\mathbf{y}\mathbf{Z}\_0) / \langle \mathbf{k}\_0 \mathbf{Z}\_B \rangle. \tag{5}$$

The extracted frequency dependence of the effective parameter results are plotted in Fig. 3. The real and imaginary components εeff (= ε′ - jε″) and μeff (= μ′ - jμ″) are plotted in Fig. 3(b) and (c), respectively. There is a frequency interval, in which one effective parameter is negative ε′ for OCSRRs, μ′ for sipral). Note that both the real components of the effective permittivity and permeability (ε′ and μ′ ) are negative, and the imaginary components (ε″ and μ″) are positive at the aimed design frequency of 2.43 GHz. This meets the general condition for the power flow and the phase velocity to be oppositely directed to the power flow, which is written as;

$$
\varepsilon \dot{\mathfrak{u}} \!\!\!\!\!\!\!\!\/+ \bar{\mathfrak{u}} \!\!\!\/\varepsilon \!\!\/> 0. \tag{6}
$$

(c) Single unit cell showing the direction of propagation of incident electromagnetic wave (substrate thickness t = 0.8 mm).

Fig. 2. Schematic of optimum absorber unit cell and simulation setup.

and (c), respectively. There is a frequency interval, in which one effective parameter is

μ″) are positive at the aimed design frequency of 2.43 GHz. This meets the general condition for the power flow and the phase velocity to be oppositely directed to the power flow,

(a) OCSRRs (b) Spirals

(c) Single unit cell showing the direction of propagation of incident electromagnetic wave

Fig. 2. Schematic of optimum absorber unit cell and simulation setup.

ε′ μ″+ μ′

and μ′

for sipral). Note that both the real components of the effective

) are negative, and the imaginary components (ε″ and

ε″> 0. (6)

negative ε′

which is written as;

for OCSRRs, μ′

permittivity and permeability (ε′

(substrate thickness t = 0.8 mm).

Fig. 3. Simulated results for the single absorber unit cell.

Fig. 4. Simulated absorbance of the metamaterial absorber cell.

(b) f = 2.54 GHz

Fig. 5. Simulated surface current densities in the spirals and OCSRRs.

As a result, the unit cell can be regarded as a double negative metamaterial unit cell over the frequency range 2.43-2.45 GHz. As shown in Fig. 3(c), the imaginary part of the refractive index is large (n″≈ 8) in the left-handed frequency region which means strong absorption of the EM wave energy. The simulated absorbance curve over a broader frequency range is plotted in Fig. 4. The maximum absorbance peak is 96% at 2.43 GHz, there is a secondary absorbance peak at approximately at 2.54 Hz. In order to understand the nature of this absorbance, the simulated surface current densities in the top resonator structure of spiral and the lower resonator structure of OCSRRs for 2.43 and 2.54 GHz resonances are shown in Fig. 5, respectively. For the 2.43 GHz resonance, we observe that the counter-circulating currents flow on both the spirals provide magnetic resonance, and the stronger current density takes place in both the shorted-end of the left-side OCSRRs, which provide electric resonance. In contrast, the 2.54 GHz resonance is determined by the magnetic response associated with a circulating current flowing on the right-side spiral and the electric resonance associated with the shorted-end of the right-side OCSRR. Fig. 6 shows the simulated S-parameters for the different horizontal spacing lengths g between the absorber cells and the simulated results are list in the Table 1. When the spacing between two cells is 6 mm, the arrayed cell shows good impedance matched to free space impedance and maximum absorbance.

(a) f = 2.43 GHz

(b) f = 2.54 GHz

As a result, the unit cell can be regarded as a double negative metamaterial unit cell over the frequency range 2.43-2.45 GHz. As shown in Fig. 3(c), the imaginary part of the refractive index is large (n″≈ 8) in the left-handed frequency region which means strong absorption of the EM wave energy. The simulated absorbance curve over a broader frequency range is plotted in Fig. 4. The maximum absorbance peak is 96% at 2.43 GHz, there is a secondary absorbance peak at approximately at 2.54 Hz. In order to understand the nature of this absorbance, the simulated surface current densities in the top resonator structure of spiral and the lower resonator structure of OCSRRs for 2.43 and 2.54 GHz resonances are shown in Fig. 5, respectively. For the 2.43 GHz resonance, we observe that the counter-circulating currents flow on both the spirals provide magnetic resonance, and the stronger current density takes place in both the shorted-end of the left-side OCSRRs, which provide electric resonance. In contrast, the 2.54 GHz resonance is determined by the magnetic response associated with a circulating current flowing on the right-side spiral and the electric resonance associated with the shorted-end of the right-side OCSRR. Fig. 6 shows the simulated S-parameters for the different horizontal spacing lengths g between the absorber cells and the simulated results are list in the Table 1. When the spacing between two cells is 6 mm, the arrayed cell shows good impedance matched to free space impedance and

Fig. 5. Simulated surface current densities in the spirals and OCSRRs.

maximum absorbance.

Fig. 6. Simulated results for the different horizontal spacing lengths g between the absorber cells.


Table 1. The summary of the simulated results for the unit cells array.
