Full Report BE4thyr

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________
CHAPTER 1 Introduction to MEMS

How many times when you are working on something frustratingly tiny, like your wifes wrist watch, have you said to yourself, If I could only train an ant to do this! What I would like to suggest is the possibility of training an ant to train a mite to do this. What are the possibilities of small but movable machines? They may or may not be useful, but they surely would be fun to make.
(From the talk Theres Plenty of Room at the Bottom, delivered by Richard P. Feynman at the annual meeting of the American Physical Society, Pasadena, California, December 1959.)
1.1 Introduction to Microelectronics:

The field of microelectronics began in 1948 when the first transistor was invented. This first transistor was a point-contact transistor, which became obsolete in the 1950s following the development of the bipolar junction transistor (BJT). The first modern day junction field-effect transistor (JFET) was proposed by Shockley (1952). These two types of electronic devices are at the heart of all microelectronic components, but it was the development of integrated circuits (ICs) in 1958 that spawned today's computer industry.IC technology has developed rapidly during the past 40 years. The continual improvement in silicon processing has resulted in a decreasing device size; currently, the minimum feature size is about 200 nm.
Figure 1.1 Moore's law for integrated circuits: exponential growth in the number of transistors in an 1C during the past 30 years
The resultant increase in the number of transistors contained within a single IC follows what is commonly referred to as Moore's law. Figure 1.1 shows that in just 30 years
__________________________________________________________________________________ Dept of E&CE BEC Bagalkot 1
the number of transistors in an IC has risen from about 100 in 1970 to 100 million in 2000. This is equivalent to a doubling of the number per chip every 18 months. Figure 1.1 plots a number of different common microprocessor chips on the graph and shows the clock speed rising from 100 kHz to 1000 MHz as the chip size falls. These microprocessors are of the type used in common personal computers costing about 1000 in today's prices1. Memory chips consist of transistors and capacitors; therefore, the size of dynamic random access memories (DRAM) has also followed Moore's law as a function of time. Figure 1.2 shows the increase of a standard memory chip from 1 kB in 1970 to 512 MB in 2000. If this current rate of progress is maintained, it would be possible to buy for 1000 a memory chip that has the same capacity as the human brain by 2030 and a memory chip that has the same brain capacity as everyone in the whole world combined by 2075! This phenomenal rise in the processing speed and power of chips has resulted first in a computer revolution and currently in an information revolution. Consequently, the world market value of ICs is currently worth some 250 billion euros, that is, about 250 times their processing speed in hertz.
Figure 1.2 Size of memory chips (DRAM) and minimum feature as a function of time. From Campbell (1996)
1.2 Evolution of MEMS:

The next ambitious goal is to fabricate monolithic or integrated chips that can not only sense (with microsensors) but also actuate (with microactuators), that is, to create a microsystem that encompasses the information-processing triptych. The technology employed to make such a microsystem is commonly referred to as MST. Figure 1.3 provides
an overview of MST together with some of the application areas. Work to achieve this goal started in the late 1980s, and there has been enormous effort microelectromechanical systems (MEMS) using MST. Working definition of the term MEMS:'A MEMS is a device made from extremely small parts (i.e. microparts).' Early efforts focused upon silicon technology and resulted in a number of successful micromechanical devices, such as pressure sensors and ink-jet printer nozzles. Yet, these are, perhaps, more accurately described as devices rather than as MEMS. The reason for the relatively slow emergence of a complete MEMS has been the complexity of the Figure 1.4 details some new materials for MEMS and the various microtechnologies that need to be developed.. One attractive solution to the development of MEMS is to make all the techniques compatible with silicon processing. In other words, conventional complementary metal oxide semiconductor (CMOS) processing is combined with a pre-CMOS or post-CMOS MST. The present MEMS market is relatively staid and mainly consists of some simple optical switches for the communications industry, pressure sensors, and inertia!. Sensors for the automotive industry, as shown in Figure 1.5. This current staidness contrasts with the potential for MEMS, which is enormous. Table 1.1 is taken from a recent report on the world market for MEMS devices. The major growth areas were identified as microfluidics and photonics and communications. However, there have been some exciting manufacturing process. Microstereolithography technique can be used to make a variety of three-dimensional microparts, such as microsprings, microgears, microturbines, and so on. to fabricate
Figure 1.3 Overview of microsystems technology and the elements of a MEMS chip. From Fatikow and Rembold (1997)
Figure 1.4 Some of the many fundamental techniques required to make MEMS devices. From Fatikow and Rembold (1997)
Figure 1.5 Pie chart showing the relative size of the current world MEMS market. The units shown are billions of euros
There are two major challenges facing us today: first, to develop methods that will manufacture microparts in high volume at low cost and, second, to develop microassembly techniques. To meet these challenges, certain industries have moved away from the use of silicon to the use of glasses and plastics, and we are now seeing the emergence of chips in biotechnology that include microfluidic systems, which can truly be regarded as MEMS devices.

Table 1.1 Sales in millions of euros of MEMS devices according to the System Planning Corporation Market Survey (1999)
Figure 1.6 Dimensions of microsensors, MEMS, and micromachines; they are compared with some everyday objects. The horizontal axis has a logarithmic scale. Modified from Gardner (1994)
MEMS devices associated with this development shown fig 1.6(the is compaired with nomal devices), there is a further major problem to solve, namely, miniaturisation of a suitable power source. Moving a micromachine through space requires significant energy. If it is to then do something useful, such as removing a blood clot in an artery, even more power will be required. Consequently, the future of MEMS devices may ultimately be limited by the communication link and the size of its 'battery pack!' The road to practicable micromachines appears to be long and hard but the first steps toward microsensors and MEMS devices have been taken, and this book provides an overview of these initial steps.
1.3 MEMS fabrication technologies

Microelectromechanics integrates fundamental theories (electromagnetics,
micromechanics and microelectronics), engineering practice, and fabrication technologies. Using fundamental research, MEMS can be devised, designed, and optimized. In addition to theoretical fundamentals, affordable (low-cost) high-yield fabrication technologies are required in order to make three-dimensional microscale structures, devices, and MEMS. Micromachining and high-aspect-ratio are key fabrication technologies for MEMS. Microelectromechanical systems fabrication technologies fall into three broad categories. In particular: bulk micromachining, surface micromachining and LIGA (LIGA-like) technologies. Different fabrication techniques and processes are available. In general, the fabrication processes, techniques and materials depend upon the available facilities and equipment. High-yield proprietary integrated CMOSbased MEMS-oriented industrial fabrication technologies are developed by the leading MEMS manufacturers. It is the authors goal for this book to be used for the MEMS, micro- and nanoengineering courses which may integrate laboratories and experiments. However, due to different facilities, equipment, infrastructure, distinct course structures and different number of credit-hours allocated, it is difficult to focus and provide detailed comprehensive coverage of all possible MEMS fabrication processes including packaging, calibration, and testing. There exist an infinite number of different developments, experiments, demonstrations, and laboratories in MEMS, microdevices, microstructures and microelectronics. Excellent Web sites (Case Western University, Georgia Institute of Technology, Massachusetts Institute of Technology, Stanford University, Universities of California at Berkeley, Los Angeles and Santa Barbara, University of Wisconsin Madison, etc.) support the possible fabrication technologies, equipment, facilities, and MEMS-NEMS infrastructure developments.The fitness, applicability and reliability depend upon the equipment and infrastructure, as well as MEMS needed to be made.
1.3.1. Bulk Micromachining

Bulk and surface micromachining are based on the modified CMOS technology with specifically designed micromachining processes. Bulk micromachining was developed more than thirty years ago to fabricate threedimensional microstructures [11]. Bulk
micromachining of silicon uses wet and dry etching techniques in conjunction with etch masks and etch-stoplayers to develop microstructures from silicon substrates. Bulk micromachining of silicon is affordable, high-yield and well-developed technology.
Microstructures are fabricated by etching areas of the silicon substrates releasing the desired three-dimensional microstructures. The anisotropic and isotropic wet etching processes, as well as concentration dependent etching techniques, are widely used in bulk micromachining. The microstructures are formed by etching away the bulk of the silicon wafer. Bulk machining with crystallographic and dopantdependent etch processes, when combined with wafer-to-wafer bonding, produces complex three-dimensional microstructures with the desired geometry. One fabricates microstructures by etching deeply into the silicon wafer. There are several ways to etch silicon wafers. The anisotropic etching uses etchants (usually potassium hydroxide KOH, sodium hydroxide NaOH, H2N4 and ethylene-diamine-pyrocatecol EDP) that etch different crystallographic directions at different etch rates. Certain crystallographic planes (stopplanes) etch very slowly. Through anisotropic etching, three-dimensional structures (cons, pyramids, cubes and channels into the surface of the silicon wafer) are fabricated. In contrast, the isotropic etching etches all directions in the silicon wafer at same (or close) etch rate. Therefore, hemisphere and cylinder structures can be made. Deep reactive ion etching uses plasma to etch straight walled structures (cubes, rectangular, triangular, etc.). In bulk micromachining, wet and dry etching processes are widely used. Wet etching is the process of removing material by immersing the wafer in a liquid bath of the chemical etchant. Wet etchants are categorized as isotropic etchants (attack the material being etched at the same rate in all directions) and anisotropic etchants (attack the material or silicon wafer at different rates in different directions, and therefore, shapes/geometry can be precisely controlled). In other words, the isotropic etching has a uniform etch rate at all orientations, while for anisotropic etching, the etch rate depends on crystal orientation. Some etchants attack silicon at different rates depending on the concentration of the impurities in the silicon (concentration dependent etching). Isotropic etchants are available for silicon, silicon oxide, silicon nitride, polysilicon, gold, aluminum, and other commonly used materials. Since isotropic etchants attack the material at the same rate in all directions, they remove material horizontally under the etch mask (undercutting) at the same rate as they etch through the material. The hydrofluoric acid etches the silicon oxide faster than the silicon. Anisotropic etchants, which etch different crystal planes at different rates, are widely used, and the most popular anisotropic etchant is potassium hydroxide (KOH) because it is the safest one to use. The application of the concentration dependent etching can be illustrated as explained below. High levels of boron (p-type dopant) will reduce the rate at
which the doped silicon is etched in the KOH system by several orders of magnitude, stopping the etching of the boron rich silicon (as described above, the boron impurities are doped into the silicon by diffusion). Let us illustrate the technique. A thick silicon oxide mask is formed over the silicon wafer and patterned to expose the surface of the silicon wafer where the boron is to be doped. The silicon wafer is then placed in a furnace in contact with a boron diffusion source. Over a period of time boron atoms migrate into the silicon wafer. As the boron diffusion is completed, the oxide mask is stripped off. A second mask can be deposited and patterned before the wafer is immersed in the KOH system that etches the silicon that is not protected by the mask etching around the boron doped silicon. The available anisotropic etchants of silicon (ethylene-
diaminepyrocatecol, potassium hydroxide, and hydrazine) etch single-crystal silicon along the crystal planes. From CMOS technology, etch masks and etch-stop techniques are available which can be used in conjunction with silicon anisotropic etchants to selectively prevent regions of silicon from being etched. Therefore, microstructures are fabricated on a silicon substrate by combining etch masks and etch-stop patterns with different anisotropic etchants. Wet etching of silicon is used for shaping and polishing, as well as for haracterizing structural and compositional features. The fundamental etch reactions are electrochemical. Oxidation reduction is followed by dissolution of the oxidation products. The etching process is either reactionrate limited (etching process depends on the chemical reaction rate) or diffusionlimited (etching process depends on the transport of etchant by diffusion to or from the surface through the liquid). Diffusion controlled processes have lower activation energies than reaction-rate controlled processes. Therefore diffusion-controlled processes are robust (insensitive) to temperature variations. However, diffusion-controlled processes are affected by agitation which increases the supply of reactant material to the semiconductor surface, increasing the etch rate. Changes the etching conditions and parameters (temperature, etchant components, their molarity, and proportions) change the rate-limiting process. The supply of minority carriers to the semiconductor surface limits the dissolution rate in etching reactions that result in a depletion of electrons or holes. Creation of electronhole pairs on the surface (by illumination or by application of electric currents) or providing generation sites increases the etch rate. Additional factors that determine the rate of etching of crystalline semiconductors include orientation, type and concentration of doping atoms, lattice defects, and surface structure. One of the most important microfabrication features is the etching directionality. If the etch rate in the x and y directions is equal to that in the z direction, the etch process is said
to be isotropic (nondirectional). The etching of single-crystal silicon, polycrystalline, and amorphous silicon in HF, BHF, HNO3 or CH3COOH etchants (which form the so-called HNA etchant system) results in isotropic process. Etch processes which are anisotropic or directional have the etch rate in the z direction higher than the lateral (x or y direction) etch rate. An example of this etch profile is the etching of (100) single-crystal silicon in the KOH/water or ethylenediamine- pyrocatechol/water (EDP) etchants. The vertical anisotropic etching is the directional etching in which the lateral etch rate is zero (for example, etching of the 110 single-crystal silicon in the KOH system, or etching the silicon substrate by ion bombardment assisted plasma etching, e.g., reactive ion etching or ion beam milling). Isotropic etching in liquid reagents is the most widely used process for removal of damaged surfaces, creating structures in single-crystal slices, and patterning single-crystal or polycrystalline semiconductor films. For isotropic etching of silicon, the most commonly used etchants are mixtures of hydrofluoric (HF) and nitric (HNO3) acids in water or acetic acid (CH3COOH). In this cocalled HNA etchant system, after the hole injection and OH- attachment to the silicon to form Si(OH)2, hydrogen is released to form SiO2. Hydrofluoric acid is used to dissolve SiO2 to form water soluble H2SiF6. The reaction is Si + HNO3 + 6HF H2SiF6 + H2NO2 + H2O + H2. Water can be used as a diluent for this etchant. However, acetic acid CH3COOH is preferred because it controls the dissociation of the nitric acid and preserves the oxidizing power of HNO3 for a wide range of dilution (i.e., it acts as a buffer). Thus, the oxidizing power of the etchant remains almost constant. The HF-HNO3 system was examined in [12]. Figure 1.7 shows the results in the form of isoetch curves for various constituents by weight.
Figure 1.7 Isotropic etching curves for the silicon using the HF-HNO3-diluent etchant system.
It should be noted here that normally available concentrated acids are 49% and 70% for HF and HNO3, respectively. Either water (dashed line curves) or acetic acid (solid-line curves) are used as the diluent in this system. At high HF and low HNO3 concentrations, the etch rate is controlled by the concentration of HNO3. Etching tends to be difficult to initiate with the actual onset of etching highly variable. In addition, it results in relatively unstable silicon surfaces which proceed to slowly grow a layer of SiO2 over a period of time. The etch is limited by the rate of the oxidation-reduction reaction, and therefore, it tends to be orientation dependent. At low HF and high HNO3 concentrations, the etch rate is controlled by the ability of HF to remove the SiO2 as it is formed. These etches are self-passivating in that the surface is covered with a 30-50 layer of SiO2. The primary limit on the etch rate is the rate of removal of the silicon complexes by diffusion. The etching process in this region is isotropic and acts as the polishing etching. The etch system in the HF:HNO3=1:1 range is initially insensitive to the addition of diluent when the percentage of diluent is less than 10%. From 10-30%, the etch rate decreases with the addition of diluent. If diluent is greater than 30%, even small changes of diluent cause large changes in the etch rate. Anisotropic etchants of silicon, such as EDP, KOH, and hydrazine are orientation dependent. That is, they etch the different crystal orientations with different etch rates. Anisotropic etchants etch the (100) and (110) silicon crystal planes faster than the (111) crystal planes. For example, the etch rates are 500:1 for (100) versus (111) orientations, respectively [13]. Silicon dioxide, silicon nitride, and metallic thin films (chromium and gold) provide good etch masks for typical silicon anisotropic etchants. These films are used to mask areas of silicon that must be protected from etching and to define the initial geometry of the regions to be etched. Two techniques have been widely used in conjunction with silicon anisotropic etching to guarantee the etch-stop. Heavily-boron-doped silicon (so-called p+ etch-stop) is effective in stopping the etch. The pn-junction technique can be used to stop etching when one side of a reverse-biased junction-diode is etched away. Anisotropic etchants for silicon are usually alkaline solutions used at elevated temperatures. For isotropic etchants, two main reactions are oxidation of the silicon, followed by dissolution of the hydrated silica. The commonly used oxidant is H2O in aqueous alkaline systems (NaOH or KOH) [14-16], cesium hydroxide [17], hydrazine and EDP [18], quaternary ammonium hydroxides [19], or sodium silicates [20]. The most commonly used anisotropic etchants (etchant systems) for silicon are EDP with water, KOH with water or isopropyl, NaOH with water, and H2N4 with water or isopropyl (the etch rate of mask varies from 1 /min to 20 /min). The KOH - water
etching system exhibits much higher (110) to (111) etch ratios than the EDP system. The etch rate ratios of the 100-, 110- and 111-plane in the EDP are 50, 30 and 1 [21], while in the KOH system, the rate ratios are 100, 600 and 1 [16]. Therefore, the KOH system is used for groove etching in 110-plane silicon wafers. The differences in etch ratios permit deep, highaspect-ratio grooves with minimal undercutting of the mask. A disadvantage of the KOH system is that silicon dioxide (SiO2) is etched at a rate which limits its use as a mask in many applications. For microstructures requiring long etching times, the silicon nitrade (Si3N4) is the preferred masking material for the KOH system, while if the EDP system is used, masks can be made applying a variety of materials (for example, SiO2, Si3N4, Cr, or Au). The etching process is a charge-transfer mechanism, and etch rates depend on dopant type and concentration. Highly doped materials may exhibit high etch rate due to the greater availability of mobile carriers. This occurs in the HNA etching system (HF:HNO3:CH3COOH or H2O = 2:3:8), where typical etch rates are 1-3 m/min at p or n concentrations [22]. The anisotropic etchants (EDP and KOH) exhibit a different preferential etching behavior. Silicon heavily doped with boron reduces the etch rate in the range from 5 to 100 times when etching in the KOH system, and by 250 times when etching in the EDP system. Thus, the etch rate is a function of boron concentration, and the etch-stops formed by the p+ technique are less than 10 m thick. The electrochemical each-stop process, which does not require heavy doping and guarantee the possibility to create thicker etch-stop-layers because the etch-stop-layer can be grown epitaxially, is widely used. The widely implemented dry etching process in micromachining applications is reactive ion etching. In this process, ions are accelerated towards the material to be etched, and the etching reaction is enhanced in the direction of travel the ions. Reactive ion etching is an anisotropic etching process. Deep trenches and pits (up to a few tens of microns) of the specified shape with vertical walls can be etched in a variety of commonly used materials, e.g., silicon, polysilicon, silicon oxide, and silicon nitride. Compared with the anisotropic wet etching, dry etching is not limited by the crystal planes in the silicon. Figure 1.8 illustrates the anisotropic etched 400 m deep and 20 m width grooves (in 110-silicon), and threedimensional silicon structure are made using reactive ion etching.
Figure 1.8 Isotropic and reactive ion etching of silicon
1.3.2 Surface Micromachining

Different techniques and processes for depositing and patterning thin films are used to produce complex microstructures and microdevices on the surface of silicon wafers (surface silicon micromachining) or on the surface of other substrates. Surface micromachining technology allows one to fabricate the structure as layers of thin films. This technology guarantees the fabrication of three-dimensional microdevices with high accuracy, and the surface micromachining can be called a thin film technology. Each thin film is usually limited to thickness up to 5 m which leads to fabrication of highperformance planar-type microscale structures and devices. The advantage of surface micromachining is the use of standard CMOS fabrication processes and facilities, as well as compliance with ICs. Therefore, this technology is widely used to manufacture microscale actuators and sensors (microdevices). Surface
micromachining has become the major fabrication technology in recent years because it allows one to fabricate complex three-dimensional microscale structures and devices. Surface micromachining with singlecrystal silicon, polysilicon, silicon nitride, silicon oxide, and silicon dioxide (as structural and sacrificial materials which are deposited and etched), as well as metals and alloys, is widely used to fabricate thin micromechanical structures and devices on the surface of a silicon wafer. This affordable low-cost high-yield technology is integrated with electromechanical microstructures ICs fabrication processes guaranteeing the needed microstructures-IC fabrication compatibility. Surface micromachining is based on the application of sacrificial (temporary) layers that are used to maintain subsequent layers and are removed to reveal (release) fabricated microstructures. This technology was first demonstrated for ICs, and applied to fabricate
motion microstructures in the 1980s. On the surface of a silicon wafer, thin layers of structural and sacrificial materials are deposited and patterned. Then, the sacrificial material is removed, and microelectromechanical structure or device is fabricated. Figure 1.9 illustrates a typical process sequence of the surface micromachining fabrication technology.
Figure 1.9 Surface micromachining.
Usually, the sacrificial layer is made using silicon dioxide (SiO2), phosphorous-doped silicon dioxide (PSG), or silicon nitride (Si3N4). The structural layers are then typically formed with polysilicon, metals, and alloys. The sacrificial layer is removed. In particular, after abrication of the surface microstructures and microdevices (micromachines), the silicon wafer can be wet bulk etched to form cavities below the surface components, which allows a wider range of desired motion for the device. The wet etching can be done using: Hydrofluoric acid (HF), Buffered hydrofluoric acid (BHF), Potassium hydroxide (KOH), Ethylene-diamene-pyrocatecol (EDP), Tetramethylammonium hydroxide (TMAH), Sodium hydroxide (NaOH). Surface micromachining has been widely used in commercial fabrication of MEMS and microdevices (microtransducers, actuators and sensors such as rotational/translational microservos, accelerometers, gyroscopes, etc.), and microstructures (gears, flip-chip electrostatic actuators, membranes, mirrors, etc.). As was emphasized, surface
micromachining means the fabrication of micromechanical structures and devices by deposition and etching of structural and sacrificial layers (thin films). Simple microstructures (beams, gears, membranes, etc.) and complex microdevices (actuators, motors, and sensors) are fabricated on top of a silicon substrate. The most important attractive features of the surface micromachining technology are the small microstructure dimensions and the opportunity to integrate micromechanics, microelectronics (ICs), and optics on the same chip. Using the ICs compatible batch processing, affordable, low-cost, high-yield microstructure fabrication is achieved for high volume applications. For example, to fabricate microscale gears (microgear train), a sacrificial silicon dioxide is deposited on the wafer and patterned. Then, a structural layer of polysilicon is deposited and patterned. This polysilicon layer becomes the structural microgears element. Other layers are then deposited and patterned making the rest of the microstructure (microscale gears). Etching in the hydrofluoric or buffered hydrofluoric acids removes the sacrificial layers releasing the microgear. There are three key challenges in fabrication of microstructures using surface micromachining: control and minimization of stress and stress gradient in the structural layer to avoid bending or buckling of the released microstructure; high selectivity of the sacrificial layer etchant to structural layers and silicon substrate; avoidance of stiction of the released (suspended) microstructure to the substrate. By choosing appropriate deposition and doping parameters, the stress and stress gradient in thin films can be controlled and optimized (minimized). The sacrificial layers can be etched with high selectivity against structural layer and silicon substrate using hydrofluoric or buffered hydrofluoric acids. Two methods to prevent the stiction are commonly used: (1) application of gaseous hydrofluoric acid and control of temperature using the substrate heater; (2) supercritical phase transition of carbon dioxide above the critical point (73 bar and 310C). After etching the sacrificial layer and rinsing, the rinsing liquid is exchanged by liquid carbon dioxide which is subsequently transferred in the supercritical state; thus, the phase transition liquid-gaseous is avoided, and capillary forces do not occur. The sacrificial layers are removed by the lateral etching. This selective etching can be performed using hydrofluoric acid which etches SiO2 but not single-crystal and polycrystalline silicon. Alternatively, the KOH etching system with polysilicon as the sacrificial layer and silicon nitride as the cover material can be used. If 100-oriented silicon is used, substrate etching will terminate on the 111-plane. The deposition is performed using low pressure chemical vapor deposition (LPCVD) from pure silane. The requirements on the deposition rate, thickness, and stress controls lead one to analysis of mechanical properties and film morphology. The morphological range is controlled by deposition and nucleation conditions. Calculations based on single-crystal data
and texture functions indicate that fine-grained, randomly oriented films are needed to attain isotropic mechanical properties. This requirement restricts the film growth conditions to the morphouspolycrystalline boundary or for LPCVD silicon, to the temperature region from 5750C to 6100C. The optimization of mechanical properties of thin films are achieved via flow, pressure, and temperature control. This requires fast measurement and control methods. The primary issue in the deposited thin films is the control of the builtin strain. For example, the maximum membrane deflection d for a loaded plate which is subject to built-up strain is a nonlinear function of the pressure (p), strain (e), length (l), thickness (h), Youngs module (E) and Poisson ratio ( ). In particular one has following formula
If the strain field is compressive (e < 0), bending and buckling occur. If the field is tensile, membrane deflections will be reduced for a given pressure. Compressive polysilicon can be converted to tensile polysilicon via annealing which converts thin films to the fine-grained form (this involves a volume contraction which causes the tensile field). The comprehensive database for the mechanical properties of the deposited thin films was developed in order to achieve computer-aided design and manufacturing capabilities.The beam-design surface micromachining process employs thin films of two different materials (polysilicon is typically used as the structural material, and silicon oxide as the sacrificial material). These materials are deposited and patterned. The sacrificial material is etched away to release threedimensional structure. Complex microstructures have many layers, and fabrication. . Surface micromachining is an additive fabrication technique which uses modified CMOS technology and materials (e.g., doped and undoped singlecrystal silicon and polysilicon, silicon nitride, silicon oxide, and silicon dioxide for the electrical and mechanical microstructures, and aluminium alloys for the metal connections), and involves the building of a microstructure or microdevice on top the surface of a supporting substrate. This technique complies with other CMOS technologies to fabricate ICs on a substrate. To fabricate high-performance mechanical microstructures and microdevices using the silicon and other materials, the internal stresses of thin films must be controlled. It is desirable to grow/deposit the polysilicon, silicon nitride, silicon dioxide, metals, alloys, and insulators thin films within minimum time. However, the high deposition speed results in high internal
stress in thin films, and this highly compressive internal stresses leads to the bending and buckling effects. Thus, the thin film deposition process should be controlled and optimized in order to minimize or eliminate the internal stress. For example, the stress of a polysilicon thin film can be controlled by doping it with boron, phosphorus, or arsenic. However, doped polysilicon films are rough and interfere with ICs. The stress in polysilicon can be controlled by annealing (annealing the polysilicon after deposition at elevated temperatures changes thin films to be stress free or tensile). The annealing temperature sets the films final stress. Using this method ICs can be embedded into polysilicon films through selective doping, and hydrofluoric acid will not change the mechanical properties of the material. The stress of a silicon nitride film can be controlled by regulating the deposition temperature and the silicon/nitride ratio. The stress of a silicon dioxide thin film can be controlled and minimized by changing the deposition temperature and post annealing. It is difficult to control the stress in silicon dioxide accurately. Therefore, silicon dioxide is usually not used as the structural material. Silicon dioxide is used for electric insulation or as a sacrificial layer under the polysilicon structural layer. Sacrificial layers are temporary layers which will be selectively removed later allowing partial or complete release of the structures. Silicon nitride may also be used for electronic insulation and as a sacrificial layer.
1.3.3 High-Aspect-Ratio (LIGA and LIGA-Like) Technology

There is a critical need to develop the fabrication technologies allowing one to fabricate high-aspect-ratio microstructures and microdevices. The LIGA process, which denotes LithographyGalvanoformingMolding(in German LithografieGalvanik
Abformung), is capable of producing three-dimensional microstructures of a few centimeters high with the aspect ratio (depth versus lateral dimension) of more than 100. This ratio can be achieved only through buck micromachining using wet anisotropic etching. The LIGA technology is based on the x-ray lithography which guarantees shorter wavelength (from few to ten which lead to negligible diffraction effects) and larger depth of focus compared with optical lithography. The ability to fabricate microstructures and microdevices in the centimeter range is particularly important in the actuators applications since the specifications are imposed on the rated force and torque developed by the microdevices. Due to the limited force and torque densities, the designer faces the need to increase the actuator dimensions. The LIGA and LIGA-like processes are based on deep xray lithography and electroplating of metal and alloy structures, allowing one to achieve structural heights in the centimeter range [32-35]. This type of processing expands the
material base significantly and allows the fabrication of new high-performance electromechanical microtransducers. In translational (liner) and rotational microactuators, the electromagnetic force and torque depend on the change in energy which is stored in the active volume and the energy density of material. In particular, the expression for the co-energy is used to derive the electromagnetic force and torque. High-performance actuators with maximized active volumes and minimized surface areas have been designed and fabricated using LIGA and LIGA-like technologies. In these processes, a substrate with a plating base is covered with a thick photoresist (thickness can be in the centimeter range). The photoresist is cured and exposed by x-rays from a synchrotron source (x-ray lithography). The photoresist strain, which is due adhesion, causes well-known difficulties. This problem is solved by combining surface micromachining, patterning the sacrificial layers under the plating base, and optimizing the processes. The achievable structural height of LIGA or LIGA-like fabricated structures is defined mainly by the photoresist processing. The photoresist procedures (based on solvent bonding of polymethylmethacrylate PMMA and subsequent mechanical height adjustments) have been optimized and used to produce low strain photoresist layers with thickness from 50 m to the centimeters range. Large area exposures of photoresist with thickness up to 10 cm have been achieved with x-ray masks. After electroplating, replanarization can be made through precision polishing. Figure 1.10 illustrates the basic sequential processes (steps) in LIGA technology. Here, the x-ray lithography is used to produce patterns in very thick layers of photoresist. The x-rays from a synchrotron source are shone through a special mask onto a thick photoresist layer (sensitive to x-rays) which covers a conductive substrate (step 1). This photoresist is then developed (step 2). The pattern formed is electroplated with metal (step 3). The metal structures produced can be the final product, however, it is common to produce a metal mould (step 4). This mould can then be filled with a suitable media (e.g., metal, alloy, polymer, etc.) as shown in step 5. The final structure is released (step 6). The described LIGA technology (frequently referred to as the highaspect- ratio technique) allows one to fabricate microstructures with small lateral dimensions compared with thickness. Thick and narrow microstructures guarantee high ruggedness in the direction perpendicular to the substrate and compliance in the lateral directions. For actuators, highaspect- ratio technology offers the possibility to fabricate high torque and force density microtransducers. As was emphasized, high-intensity, lowdivergence, and hard x-rays are used as the exposure source for the lithography. For exposure wavelengths , the image resolution and the depth of focus are
Due to short exposure wavelength, the desired features size is achieved. These x-rays are usually produced by a synchrotron radiation source [1, 32-35]. Polymethyl-methacrylate (PMMA) and polylactides are used as the x-ray resists because PMMA (PlexiglasTM or LuciteTM) and polylactides photoresists have high sensitivity to x-rays, thermal stability, desired absorption, as well as high resolution and resistance to chemical, ion, and plasma etching. Polyglycidyl-methacryl-atecoethylacrylate (PGMA) is used as the negative x-ray resist. The exposure wavelength varies depending upon the x-ray radiation source used. For example, the 0.2 nm x-ray wavelength allows one to transfer the pattern from the highcontrast x-ray mask into the photoresist layer with a few centimeters thickness so that the photoresist relief may be fabricated with an extremely high depth to width ratio. The sidewalls of the plated structures are vertical and smooth (polished), and therefore, they can be used as optical surfaces. Photolithography using commercially available positive photoresists and near-UV light sources can produce high-aspect-ratio plating molds. Although, in comparison to LIGA, this technique is limited in terms of thickness and aspect ratio, positive photoresists can provide a simple means of fabricating high-aspect-ratio plating molds with the conventional photolithography equipment. Positive photoresists with high transparency and high viscosity can be used to achieve coatings of 20-80 m thick. Multiple coatings are needed to obtain the thicker layers of photoresist. If contrast printing is used, edge bead removal and good contact between the mask and the substrate are important. Conditions of softbake, exposure, and development should also be modified due to the large thickness of the photoresist. Longer softbake times are preferable to remove the solvent from the photoresist, and a high exposure is necessary. In particular, the energy density needed to exposure a 30-m thick photoresist is 1500 mJ/cm2. Long development time is required in order to completely remove the resist of exposed area. Hardbake conditions must be optimized because while hardbake improves the adhesion of the photoresist and chemical resistance to the plating solution, it also causes the distortion of the photoresist. Highaspect- ratios (10 or higher) can be obtained for 20 m thin films. Dry etching, based upon reactive ion etching of polyimides to form high-aspect ratio molds is used. Electromagnetically controlled dry etching of fluorinated polyimides with Ti
or Al masks has been used for deep etching to attain high-aspect-ratio, good mask selectivity, and smooth sidewalls.
Figure 1.10 LIGA fabrication technology.
A critical part of the high-aspect-ratio processes is plating to form the metallic electromechanical microstructures in the mold. Using plating, metal is deposited from ions in a solution following the shape of the plating mold. This is the additive process, and the thickness of the plated metal can be large since the plating rate can be high. A variety of metals (Al, Au, Cu, Fe, Ni, and W) and alloys (NiCo, NiFe, and NiSi) can be deposited or codeposited. It is important that roughness (smoothness) of the reflective metal surfaces with the desired shape can be achieved even for optical applications. Electroplating and electroless
plating (reduction of the metal ions occurs by the chemical reaction between a reducing agent and metal ion on a properly activated substrate) are the commonly used plating processes.
The metal seed layer can be deposited and removed from the substrate or sacrificial layer. The plating rate and the grain size are controlled by the current density, temperature, duty cycle, etc. Usually, the fabrication of MEMS is done using copper, iron, nickel, alloys, and other materials through electrodeposition and electroless plating on the selected areas of the silicon substrates (for example, deposition of the copper microwindings and magnetic NiFe alloy thin film). The electroless plating can be conducted on a chemically activated silicon substrates without deposition on the photoresist mold [1]. The optimization of plating conditions (changing the current density, temperature, pH, waveforms, duty cycle, forward and reverse current, etc.) is critical to obtain smooth surfaces and practical plating rates and to avoid spontaneous plating.
CHAPTER 2 Sensors and Actuators

The microelectronics revolution has led to increasingly complex signal-data processing Chips. This remarkably, has been associated with falling costs. Furthermore, these Processing chips are now combined with sensors and actuators to make an information processing triptych (see Figure 2.1). These developments follow the recognition in the 1980s that the price-to-performance ratio of both sensors and actuators had fallen woefully behind
Fig 2.1 Gives idea about the system
processors. Consequently, measurement systems tended to be large and, more importantly, expensive. Work therefore started to link the microelectronic technologies and use these to make silicon sensors, the so-called microsensors. 'A microsensor is a sensor that has at least one physical dimension at the submillimeter level.' This work was inspired by the vision of microsensors being manufactured in volumes at low cost and with, if necessary, integrated microelectronic circuitry. Figure 2.2 shows the relative market for ICs and microsensors in the past 10 years. It is evident that the market for microsensors lags well behind the market for ICs; nevertheless, it is worth 15 to 20 billion euros. The main cause has been the relatively stable price-performance (p/p) ratio of sensors and actuators since 1960, as illustrated in Figure 2.3.
Figure 2.2 World market for ICs and microsensors from 1990 to 2000. From various sources
Figure 2.3 Price-performance indicators for ICs, sensors, and actuators
Figure 2.4 Sensor market by application for the United Kingdom. From Gardner (1994)
This contrasts markedly with the p/p ratio of ICs, which has fallen enormously between 1960 and 2000 and is now significantly below that for sensors and actuators. As a consequence of these changes, the cost of a measurement system is, in general, dominated first by the cost of the microactuator and second by the cost of the microsensor. However, despite the cost advantages, there are several major technical advantages of making
microsensors with microsystems technology (MST); the main ones are as follows : The employment of well-established microtechnology The production of miniature sensors The production of less bulky and much lighter sensors The batch production of wafers for high volume The integration of processors
The UK marketplace for microsensors is diverse, as shown in Figure 2.4, and includes processing plants - environment and medical. However, the largest sector of the world (rather than UK) sensor market3 is currently automotive; in 1997, the sales of pressure These figures relate to the sensor market and hence exclude the larger markets for disk and ink-jet printer heads. Sensors was about 700 million euros and that for accelerometers was about 200 million euros. As the market for automotive sensors has matured, the price has fallen from 100 to 10 for a pressure sensor. In addition, the sophistication of the chips has increased and so has the level of integration. This has led to the development of 'smart' sensors. 'A smart sensor is a sensor that has part or its entire processing element integrated in a single chip.'
2.1 SENSORS:
Sensors can be classified as Thermal sensors, Radiation sensors, Mechanical sensors, Magnetic sensors, Bio(Chemical) sensors.
2.1.1 THERMAL SENSORS:

Thermal sensors are sensors that measure a primary thermal quantity, such as temperature, heat flow, or thermal conductivity. Other sensors may be based on a thermal measurement; for example, a thermal anemometer measures air flow. However, according to our classification of measured energy domain, this would be regarded as a mechanical sensor. Consequently, the most important thermal sensor is the temperature sensor. Temperature is probably the single most important device parameter of all. Almost every property of a material has significant temperature dependence. For example, in the case of a mechanical microstructure, its physical dimensions - Young's modulus, shear modulus, heat capacity, thermal conductivity, and so on - vary with operating temperature. The effect of temperature can sometimes be minimised by choosing materials with a low temperature coefficient of operation (TCO). However, when forced to use standard materials (e.g. silicon and silica), the structural design can often be modified (e.g. adding a reference device) to compensate for these undesirable effects. It is often necessary to use materials that are not based on complementary metal oxide semiconductor (CMOS), such as magnetoresistive,
chemoresistive, ferroelectric, pyroelectric; these compounds tend to possess strong temperature-dependencies6. In fact, the problem is particularly acute for chemical microsensors, as most chemical reactions are strongly temperature-dependent. Many nonthermal microsensors (and MEMS devices) have to operate either at a constant
temperature - an expensive and power-intensive option when requiring heaters or coolers, - or in a mode in which the temperature is monitored and real-time signal compensation is provided. Clearly, microdevices that possess an integrated temperature microsensor and microcontroller can automatically compensate for temperature and thus offer a superior performance to those without. This is why temperature sensors are a very important kind of sensors and are commonly found embedded in microsensors, microactuators, MEMS, and even in precision microelectronic components, such as analogue-to-digital converters. Temperature sensors again of different types depending on the properity . They are Resistive Temperature, Microsensors, Microthermocouples, Thermodiodes and Thermotransistors, SAW Temperature Sensor.
2.1.2 RADIATION SENSORS:

Radiation sensors can be classified according to the type and energy of the measurand, as illustrated in Figure 8.13, in which the energy ER in electron volts (eV) of the electromagnetic radiation is simply related to its frequency / in Hertz, and wavelength A. in meters by
where h is Planck's constant, c is the velocity of light in a vacuum (i.e. 3.0 x 108 m/s), and e is the charge on an electron (i.e. 1.60 x 10-19 C). Radiation can come in the form of particles, such as neutrons, protons, and alpha and beta particles, emitted from the decay of nuclear material. However, the detection of nuclear particles and high-energy electromagnetic radiation (i.e. gamma rays and X rays) generally requires sophisticated instrumentation that cannot be readily integrated into a miniature device. The exception to this general observation is perhaps the detection of low-energy X rays or electrons through a solid-state photoelectric detector.
Figure 2.5 Classification of radiation sensors according to their electromagnetic energy
The most common types of radiation microsensor detect electromagnetic radiation with energies or wavelengths from the ultraviolet-to-near-infrared (UV-NIR) region, which includes visible, through the NIR and thermal-infrared region and into the microwave and radio regions. The most important regions are the visible light region and the NIR region because these are the wavelengths at which signals are transmitted down fibre-optic cables in modern telecommunication systems. Readers interested in the general field of fibre-optic sensors are referred to Udd (1991) and those interested in the field of biosensing and chemical sensing are referred to Boside and Harmer (1996). The real interest to us here is whether the optical components and any optical interconnects can be integrated into a microtransducer or MEMS device (see Tabid-Azar 1995). In this book, we have concentrated on devices that have electrical outputs (sensors), electrical inputs (actuators), and on electrical interconnects (e.g. printed circuit boards (PCBs) and multichip modules (MCMs)). This was done as it reflects the bulk of past developments, but it is now becoming increasingly evident that the optical signal domain will become the more significant for both land-based signal transmission and for the operation of sensors within hazardous environments. Radiation microsensors can be distinguished by their underlying operating pri nciple, namely,
photoconductive, photovoltaic (or Photoelectric), pyroelectric, and microantenna shown in fig 2.5. We will employ this distinction to discuss these four types of radiation sensor in turn.
2.1.3 MECHANICAL SENSORS:

Mechanical microsensors are, perhaps, the most important class of microsensor because of both the large variety of different mechanical measurands and their successful application in mass markets, such as the automotive industry. Table 2.1 lists some 50 or so of the numerous possible mechanical measurands and covers not only static and kinematic parameters, such as displacement, velocity, and acceleration, but also physical properties of materials, such as density, hardness, and viscosity. Figure 2.5 shows a classification scheme for mechanical microsensors together with an example of a device type.
Table 2.1 List of mechanical measurands. Adapted from Gardner (1994)
The most important classes of mechanical microsensors to date is a subset of only six or so and these constitute the majority of the existing market for micromechanical sensors. Thus, the main measurands of mechanical microsensors are as follows in alphabetical order: Acceleration/deceleration Displacement Flow rate Force/torque Position/angle Pressure/stress Therefore, The four of the most important types of mechanical microsensors, are Pressure microsensors Microaccelerometers Microgyroscopes Flow microsensors
Figure 2.6 Classification scheme for mechanical microsensors. From Gardner (1994)
2.1.4 MAGNETIC SENSORS:

In this section, we focus on the basic principles of these different kinds of magnetic microsensors together with some examples of commercial products and research devices. Figure 2.7 shows the various kinds of magnetic microsensors that can be used to measure the magnetic flux density B.
Figure 2.7 Classification of magnetic microsensors by principle
The devices are classified here according to the form of the output signal rather than the energy domain of the input signal used earlier. Clearly, some of these magnetic microsensors are also employed within other types of sensors; for example, a Hall effect device can be used to measure the proximity of a magnet and it then becomes a mechanical (magnetic) sensor, and so on. The typical characteristics of magnetic microsensors are summarised in Table 2.2 and the details are discussed in the subsequent sections.
Table 2.2 Typical characteristics of some magnetic microsensors
2.1.5 BIO (CHEMICAL) SENSORS:

The general topic of chemical sensors is well covered in the standard literature on sensors (Madou and Morrison 1989; Gopel et al. 19891998; Taylor et al. 1996). The basic components of a bio (chemical) 19 sensor are illustrated in Figure 2.8 and comprise a chemically sensitive layer interfaced to a sensing transducer. The analyte molecules interact with the chemically sensitive layer and produce a physical change that is detected by the
transducer and are converted into an electrical output signal. The nature of this interaction is determined by the type of material used and can be either a reversible process or an irreversible reaction (see Figure 2.9). In a reversible inding reaction, the analyte is typically bound to specific sites within the sensitive layer, and when the external concentration is removed, the analyte molecules dissociate and there is no net change. An example of this would be the adsorption and desorption of an organic vapour in a polymeric material. In an irreversible reaction, the analyte undergoes a chemical reaction catalysed by the sensitive layer and therefore is consumed in the process. In this case, removal of the external analyte concentration still reverses the process but the associated time-constant may be considerably longer. The irreversible reaction is in fact more common and its sensitivity and selectivity varies with the shape and charge distributions of the analyte molecule and sensitive layer. The most selective reactions tend to be those like the key-lock mechanism that operates in a biological sensor (biosensor), such as a glucose sensor. However, the poor stability of biological materials makes them unsuitable for use in a real sensor that operates many thousands or millions of times with a lifetime of a year or more but more suited to a single measurement, that is, a disposable sensor. For this reason, we concentrate predominantly here on the field of chemical sensors; anyone interested specifically in biosensors is referred to books by Cass (1990) and Taylor and coworkers (1996).
Figure 2.8 Basic components of a bio(chemical) sensor: analyte molecules, chemically sensitive layer, and transducer. After Gardner and Bartlett (1999)
Figure 2.9 Basic mechanisms of a biochemical sensor: (a) reversible binding of the analyte A to a site at the chemically sensitive layer and (b) irreversible reaction of the analyte A at a site to produce molecule B at the chemically sensitive layer. After Gardner and Bartlett (1999)
Figure 2.10 shows the main types of bio(chemical) sensors classified according to the operating principle of the sensitive layer, that is, the signal transduced. The signals that are measured can be the change in electrical resistance (i.e. conductimetric), change in work function (i.e. potentiometric), change in the heat of reaction (i.e. calorimetric), and so on. Here, we are going to discuss the two most important types of chemical microsensor, their commercial availability, and the developments associated with current research devices.
Figure 2.10 Classification of the main types of bio(chemical) sensors. Devices listed are those that can be regarded as microsensors
2.2 Actuation Mechanisms:

Six distinct actuation mechanisms are available for the design and development of MEMS devices and sensors. Some actuation mechanisms are most suitable for RF-MEMS switches; some are best suited for mechanical resonators; some are most attractive for micropumps (MPs); and some are ideal for MEMS devices capable of monitoring the health of structures and buildings, or detecting the chemical and biological threats. The following distinct actuation mechanisms are widely used in the design and development of various RF and optical-MEMS devices best suited for commercial, industrial, military, and space sensor applications: > Electrostatic (ES) actuation > Electromagnetic actuation > Electrothermal actuation > Electrodynamic actuation > Electrochemical actuation > Piezoelectric actuation
2.2.1 Electrostatic Actuation Mechanism:

Research studies performed by the author indicate that ES actuation mechanism is widely used in the design and development of RF- and optical-MEMS devices because of high design flexibility, moderate power consumption, and high reliability. ES actuation scheme has been proven very effective in the design of RF-MEMS switches, MEMS-phase shifters, and tunable capacitors and is widely used by cellular phones, mobile phones, satellite communications systems, base transmitters, and Potential Actuation Mechanisms, Their Performance Capabilities & 43 other communication equipment. These MEMS devices contain movable elements that are set into motion using microactuators. The mechanical displacement of a freely moving structure, called a cantilever beam, is achieved through the microactuator, which is operated by an ES actuation mechanism as shown in Figure 2.11. This figure shows an ES actuation mechanism for (a) a microrelay, (b) force diagram illustrating a lumped model for the actuator, and (c) displacement as a function of applied or actuation voltage ES microactuators requiring controlled actuation or motion will be required for many MEMS components or sensors such as micropositioners, precision mechanical mirrors,
Figure 2.11 Typical elements of MEMS devices. (a) Microrelay, (b) force diagram of ES actuator, and (c) normalized beam displacement as a function of applied voltage.
force-rebalanced and resonant sensors, micromotors, MPs, valves, adaptive optics systems, and read/write actuation mechanisms for magnetic disk drive heads. Because of the scaleddown dimensions of the MEMS sensors or components, ES drives will be found most attractive for their high energy generation capabilities and for minimum fabrication costs. Vertical comb array microactuators (VCAMs) are capable of generating a large vertical force for a given geometrical area on a wafer surface. VCAM can generate a force that approaches the PP drive and is at least an order of magnitude larger than that with a lateral comb drive mechanism. In the case of VCAM beams, the force generated is approximately proportional to 2L, where L is the beam length, which is several 100 mm long. In the case of lateral comb drive, the force generated is proportional to 2t, where t is the thickness of the polysilicon layer, which is typically 23 mm. In summary, the force-generating capability from a VCAM
mechanism could be few orders of magnitude larger than that available from a lateral comb drive mechanism. ES rotary microactuators with optimum shape design offer significantly improved actuation force-generating capability. In conventional rotary ES microactuators, the forcegenerating capability is limited because of large clearance (i.e., the distance between the two facing electrodes) at the outer region from the center of rotation. To overcome the limitations of conventional rotary microactuators and to enhance the force-generating capability, a tilted configuration is developed, which can be integrated into the optimum shape. This will allow the gap between the two facing electrodes as small as possible with current fabrication technology. The optimum shape of the ES rotary microactuator involving integrated tilting configuration will increase the force-generating capability dramatically over the conventional-shaped rotary microactuator.
2.2.1.1 Electrostatic Force Computation:

The ES actuation force (Fes) is based on the Coulomb force of attraction existing between the charges of opposite polarity as shown in Figure 2.1. This force of attraction between two parallel plates (PPs) can be written as
........... 2.1 where o is the free-space permittivity (8.83 _ 10_12 F/m) and A is the plate area.
................2.2
where C is the capacitance between the plates and V is voltage in the air gap d. From Equations 2.1 and 2.2, one gets
.............2.3 It is evident from Equation 2.3 that the ES force is proportional to the plate area, proportional to the square of voltage across the plate, and inversely proportional to the square of the gap between the plates. It is important to point out that the ES force will not increase indefinitely with the increase in applied voltage (V ) but it is limited by the breakdown voltage of the air gap. The breakdown voltage in the micrometer-sized air gaps is 3*10^8 V/m, which leads to a maximum force. Assuming a plate area of 100*100 square um, an air gap of 1 um, and a voltage across the gap of 300 V, the computed ES force comes to or 4 mN (millinewton). Assuming a more practical value of applied voltage equal to 10 V, the
actuation force comes down to 0.004 mN or 4 uN, which is three orders of magnitude smaller than 4 mN at actuation voltage of 300 V.
Table 2.2 ES Forces as a Function of Various Parameters (N)
Computed values of ES forces as a function of practical actuation voltage and air gap for a plate area of 100*100 um2 are summarized in Table 2.2. It is important to mention that microrelays equipped with low quality contacts require contact forces at least in the order of 75100 uN (micronewtons) to achieve a stable actuator performance. This means that ES actuation is not suitable for microrelay applications unless a large actuation area (1000*1000 um2) or a higher actuation voltage (>100 V) is selected. ES actuation force requirements are discussed for specific MEMS device applications. ES force which require actuation voltages from 10 to 20 V across the PPs with area of 100*100 um2 are best suited for MEMS devices such as HF resonators, moderate actuation voltages from 30 to 40 V, RF-MEMS switches require high actuation voltages from 40 to 60 V. In each case, the plate area is assumed 100*100 um2 and the air gap is varied from 0 to 10 mm.
2.2.1.2 Pull-In and Pull-Out Voltage Requirements:

As mentioned earlier that the ES force will not increase indefinitely with the increase in applied or actuation voltage, but it is limited by the breakdown voltage in the air gap. The nonlinear relationship between the normalized displacement and applied voltage normalized to pull-in voltage causes instability in the cantilever beam, when the actuation or applied voltage exceeds the pull-in or snap-down voltage. In other words, the control of an electrostatically actuated cantilever armature or beam is lost owing to loss of equilibrium between the ES and spring forces. This phenomenon is quite evident from Equation 2.3 shown for the pull-in voltage showing the relationship between the ES and spring forces and whose solution yields the displacement of the movable element or cantilever armature. Note
that the expressions provided for the pull-in and pull-out voltages are only exact for a lumped spring-mass system .
2.2.1.3 Pull-In Voltage:

It is important to mention that the expression for pull-in voltage or the actuation voltage is based on Hookes law. This law is based on the assumption that the ES forces acting on the cantilever beam are the same as those in a PP capacitor. In other words, the expression for the pull-in voltage assumes that the ES force is equal to the restoring force of a mechanical spring given by Hookes law. The expression for the pull-in voltage without a dielectric cover on the lower electrode can be written as
..........2.4 where k is spring constant (N/m) d is the air gap (um) A is the plate area (um2) er is the relative dielectric constant for medium between the plates The pull-in voltage with dielectric cover can be written as
...........2.5
where d is the dielectric cover thickness, which varies between 8 and 10 percent of air gap. It is evident from Equations 2.4 and 2.5 that the pull-in voltage is proportional to the square root of spring constant, inversely proportional to the square root of relative dielectric constant, inversely proportional to the square root of plate area, and directly proportional to the cube root of air gap.

Table 2.3 Computed Values of Pull-In Voltages without a Cover and as a Function of Plate Area, Spring Constant, and Air Gap
Assuming various different parametric values, magnitudes of pull-in voltage without dielectric cover are computed, which are summarized in Table 2.3. These computed values indicate that the devices using ES actuation mechanisms will require higher applied voltage, and consequently, higher power consumption. Higher input voltage requirement is the major drawback of the ES actuation.In these calculations, it is evident that the pull-in voltage without the dielectric cover is slightly higher than that with the dielectric cover.
2.2.1.4 Pull-Out Voltage:

It is important to mention that after the pull-in occurred, the displacement becomes imaginary and the slope of the applied voltage no longer controls the displacement. Under these circumstances, the cantilever beam releases at much lower voltage called pull-out voltage (VPO) because of hysteresis experienced by the actuator (Figure 2.10). The pull-out voltage for an ES actuation is given as
.2.6 where
k is the spring constant (N/m) d is the air gap (mm) er is the relative dielectric constant or permittivity of the dielectric cover on the electrode d is the cover thickness (mm)
o is the free-space permittivity (8.85 _ 10_12 F/m) A is the plate area (mm2) Computed values of pull-out voltage as a function of various parameters are summarized in Table 2.4. The magnitudes of pull-out
Table 2.4 Pull-Out Voltages as a Function of Various ES Parameters
voltages remain extremely low compared to those for pull-in voltages, regardless of values of other parameters. As stated previously that once the pull-in has occurred, the beam or the cantilever structure releases at much lower voltage because of the presence of hysteresis effect in the ES actuator. Pull-out voltage computations using all possible permutations and combinations of parameters involved reveal that the pull-out voltage most likely will not exceed 2 V, although the pull-in voltages will never be less than 30 V for most MEMS devices requiring ES forces exceeding 50 uN. Note a force of 1 N is equal to 4.45 lb approximately.
2.2.2 Electromagnetic Actuation Mechanism:

The electromagnetic actuator is also known as the variable reluctance actuator. The electromagnetic actuator shown in Figure 2.18c consists of a movable beam, a yoke, and a driving coil comprising of excitation winding of N turns. The magnetomotive force (MMF) generated in this coil by this current is equal to product of number of turns in the coil and current flowing in it (MMF =NI). When a current I flows in the coil, a magnetic flux is set up in the yoke and the air gap. The yoke and a part of the beam are made from high permeable material and they are separated by an air gap (d ). The magnetic flux flows through the yoke and air gap as illustrated in Figure 2.18. The flux passing through magnetic circuit is approximately equal to the reluctance (Rgap), which can be written as
..2.7 where d is the air gap A is the area of the yoke close to air gap u0 is the permeability of the free space, which is equal to 1.257 _ 10_6 H/m
2.2.3 Electrothermal Actuation Mechanism:

This actuation mechanism is based on the expansion of the metal due to increase in temperature, which generates heat by an electric current (I) flowing through a resistor R as shown in Figure 2.12b. If this resistor is located in the upper region of the cantilever beam, a thermal wave occurs propagating in the thickness direction of the beam. This thermal wave creates a bending moment, which is used to actuate the cantilever beam
Figure 2.12 Actuation mechanisms best suited for microrelays.
Pull-in or snap-down does not occur in electrothermal actuation mechanism as in the case of piezoelectric actuation method. Detection of motion of the beam can be achieved either through a capacitive detection method or piezoelectric detection technique, whichever is reliable and cost effective. Major advantages of electrothermal actuation include low actuation voltage (typically from 5 to 10 V), lower construction cost of the device and simple process flow. Serious drawbacks are the slow response (typically 100300 ms), high powerdissipation, and continuous current drain in the actuated state. However, to avoid continuous current drain in aMEMSswitch in the downstate, ES clamping technique can be integrated in the switch design, which will slightly increase device cost. Because of extremely slow response, high power consumption, and continuous current drain, this
particular actuation mechanism is used only in few thermally actuated MEMS switches, RF mechanical resonators, and tunable capacitors. Because of very limited and few applications of this actuation mechanism, electrothermal actuation mechanism will not be discussed further, except for thermal bending moment calculations. It is critical to mention that this actuation method deploys a stress controlled membrane with pattern metallic contacts. This actuation mechanism permits operation at lower voltages without sacrificing the spring constant (k). The membrane structure allows fabrication of resistive RF-MEMS switches operating at MM-wave frequencies. Lower operating voltages offer significant reduction in the spring constant requirements. Thermal actuation can be a good alternative, where lower spring constant and lower operating voltages are the principal design requirements. As stated earlier that an electrothermal actuation by definition is temperature sensitive operation, thermally triggered operations are insensitive to heating below a defined temperature and they usually snap-down above this temperature. MEMS switches using thermal actuators can operate at as low as 5 V. However, the turn-off time is about 50 ms, while the turn-on time is close to 300 ms. The ON state requires going through complete heating process below a defined temperature. In case of thermally actuated RF-MEMS switch, typical isolation is better than 25 dB and the insertion loss is less than 2 dB up to 40 GHz. It is important to point out that standard thermal actuation offers linear displacement. In case where it is required to generate large rectilinear displacements and substantial restoring forces, bent-beam electrothermal actuators are best suited to meet these requirements. The simplest manifestation of bent-beam electrothermal actuator is a V-shaped structure. The displacement of the apex is dependent on the beam dimensions. The maximum displacement, which can be obtained in the absence of external force, is calculated from the
beam bending equation. In a bentbeam actuator, beam length can vary from 500 to 2000 mm, both the beam width and thickness from 4 to 6 mm, and bending angle from 38 to 128. Preliminary studies performed by the author reveal that bent-beam actuators could offer suitable compromises on displacements, actuation forces, power consumption, and actuation voltage and drive options best suited for several MEMS applications such as rotary and rectilinear microengines.
2.2.4 Electrodynamic Actuation Mechanism:

Electrodynamic actuation involves the Lorentz force acting on a conducting wire carrying current, which is placed in the external magnetic field environment. The operating principle of electrodynamic actuation is illustrated in Figure 2.13 showing the configuration of a cantilever-type actuator. The expression for the Lorentz force (FL) acting on the current (I ) can be written as
.2.8
where I is the current flowing in the wire of length L, which is placed in the external magnetic field B. In the above equation the magnetic field or flux density is expressed by
Tesla in MKS system, where 1 T is equal to 10,000 G in CGS system or equal to weber per square meter in MKS system. Computed values of Lorentz forces as a function of wire length, current flowing in the wire, and magnetic flux density are summarized in Table 2.5. These calculations indicate that the magnitude of the Lorentz force increases with the increase in wire length and in the current flowing for a given magnetic field or flux density. To achieve Lorentz forces or actuation forces in the order of 100 mN needed for AuAu contacts, one can select longer wire and minimum magnetic flux density or higher current. But higher current is not a good option under minimum power requirement. Plots of Lorentz forces or electrodynamic actuation forces as a function of current level in a 100 mm long wire and as a function of magnetic flux density are shown i Figure 2.15a and b, respectively.

Table 2.5 Lorentz Forces as a Function of Various Operational Parameters (mN)
These plots will provide several options for an actuator designer to select the most cost-effective beam design. It is important to mention that sensing of the cantilever beam motion is dependent on the change of magnetic flux linkage in the loop when the beam moves, which produces the induction voltage, thereby providing the detection signal. The actuation is strictly dependent on the current drive in the coil or wire, regardless whether an electromagnetic actuation or electrodynamic actuation mechanism is selected. In each case, the operating voltage is generally quite low ranging from 10 to 15 V. The low-voltage operation makes the electromagnetic actuation mechanism most attractive for many MEMS devices, including microrelays. The major drawbacks include continuous current drain and complex fabrication technology involving integration of coil and ferromagnetic materials, which are normally temperature sensitive.
Figure 2.13 Illustration of electrodynamic actuation mechanism. (a) Operating principle of electrodynamic actuation mechanism showing a current carrying wire experiencing the Lorentz force FL and (b) implementation of electrodynamic actuation in a microrelay.
Figure 2.14 Magnetic forces as a function current flowing in a coil having
100 turns. Figure 2.15 Magnitude of the Lorentz force (N) acting on a 100 mm long wire as a function of (a) current and (b) magnetic flux density.
2.2.5 Electrochemical Actuation Mechanism:
Research studies performed by the nanotechnology (NT) scientists at the University of Cincinnati and University of North Carolina indicate that carbon nanotube (CNT) arrays can play a key role in the design of electrochemical actuators. The studies further indicate that CNT arrays are considered smart materials because CNT have demonstrated high mechanical strength and electrical conductivity, in addition to unique piezoresistive and electrochemical sensing and actuation capabilities. Potential applications of nanotubes have spurred intensive research activities in the area of processing, device fabrication, and material characterization. Vertically aligned arrays of CNT have demonstrated well-defined properties with uniform length and diameter dimensional parameters. Thermally driven chemical vapor deposition (CVD) is best suited to grow high-density arrays on silicon substrates. Note the growth mechanism affects the length of the CNTs and is strictly dependent on the interaction between the nanoparticle and the supporting material.
2.2.6 Piezoelectric actuation:

It is important to mention that the piezoelectric actuation is based on inverse piezoelectric effect, although the piezoelectric detection is based on direct piezoelectric effect. Note a piezoelectric material provides its own internal biasing requirement, either due to absence of a center of symmetry in the case of single-crystal materials such as aluminum nitride (AIN) or zinc oxide (ZnO) or due to a permanent polarization present in ferroelectric materials such leadzirconatetitanate (PZT). The inverse piezoelectric effect is of critical importance because it offers mechanical deformation in the piezoelectric material when subjected to an external or applied electric field. Note that pull-in or snap-down does not occur in a piezoelectric actuator as seen in an ES actuation, and as a result the piezoelectric actuator does not suffer from mechanical instability. It is essential to know the critical parameters of potential piezoelectric materials for possible applications in cantilever beams for MEMS devices, which include AIN, PZT, and ZnO. Piezoelectric coefficients and constants are expressed generally in MKS system. The most important parameter is the transverse piezoelectric strain constant (d31), which is expressed in picocoulomb per newton (pC/N) or picometer per volt (pm/V). The constants eij are called piezoelectric stress constants or coefficients and are expressed in coulomb per newton (C/N) or meter per volt (m/V). The Cii constants such as C11, C22, C33, and C66 are known as electric stress constants or stiffness constants and are expressed in newton per square meter (N/m2). The constant 11 is known as the permittivity or the dielectric constant and has no unit. It is important to point out that the constant or coefficient d31 relates an electric fieldto external strain in any direction to the polar direction, where as the piezoelectric coefficient or constant
d15 relates an electric field normal to the polar axis with a shear strain in the plane containing the polar axis and signal field. Piezoelectric properties of some selected materials are listed in Table 2.6. The magnitude of transverse piezoelectric strain constant d31 is temperature sensitive. Its value for the PZT material ranges from 123 at -100C to 125 at 0C to 118 pC/N at 100C. One can expect the variations in the same parameter for other piezoelectric materials as a function of operating temperature. The value of parameter d31 for the AIN material is rounded to 3, for PTZ material to 100, and for ZnO material to 5 in some numerical examples to simplify calculations and for rapid comprehension by the readers. If some constants or coefficients are expressed in a mixture of CGS and FPS units, one can convert these constants into the kgs system using the equivalent values shown in Table 2.7. These equivalent conversion factors will help the readers in rapid comprehension of the parametric values given in different unit systems.
Table 2.6 Important Properties of Selected Piezoelectric Materials
Table 2.7 Equivalent Conversion Factors
2.3 Fabrication and Material Requirements for the Actuator:

Test tower requires the use of structural materials with high values of Youngs modulus, which can offer wide displacements of the electrochemical actuators needed for fast actuator response. The overall performance of the actuator is contingent upon the strain uniformity along the length of the beam and the actuator outer surface of the nanotubes. If all the shells of the MWCNT array could be made to actuate at the same time, the force generated by the electrochemical actuator will be maximum. Furthermore, higher the applied voltage, higher will be the increase in strain. Higher actuation voltage increases the chemical charge accumulation on the electrical interface, which will lead to a faster response. Any generation of bubbles on the nanotube surface will reduce both the actuator lifetime and reliability. Improved mechanical properties of the structural materials and rapid strain generation of the nanotube actuator are key to the reliability and efficiency of the electrochemical actuator. Compared to other actuation mechanisms, the low driving voltage of the CNT tower actuator is a major advantage for various MEMS applications such as smart structures, active catheters for medical applications, artificial muscles, MPs, and molecular Table 2.8 Performance Comparison Data of Various Actuation Mechanisms
motors, power harvesting systems, strain sensors, and nanorobots. Nanotube arrays made from high tensile strength materials can generate large forces during actuation. In summary, highly aligned MWCNT arrays offer excellent mechanical properties, high electrochemical sensitivity, high strain-generating capability, and improved electrically conductive probes best suited for many applications ranging from nanomedicine to space exploration. Performance comparison and unique capabilities of various actuation mechanisms are summarized in Table 2.8.
It is important to mention that highly aligned MWCNT arrays have been utilized in the development of smart materials, which play a key role in the design and development of biosensors, electrochemical actuators, and NT-based probes for scientific research. CNT arrays are considered as smart materials because they have high mechanical strength, enhanced electrical conductivity, and improved piezoresistive and electrochemical sensing and actuation properties. The small size as well as the unique properties of nanotubes are best suited for many applications such as environmental scanning electron microscopy, energy dispersive microscopy, mass spectroscopy, high resolution imaging sensors, smart sensors including chemical, environmental and biosensors, electrochemical actuators and NT-based sensors for weapon health, and harsh battlefield environmental monitoring applications. These NT-based sensors can identify the out-of-specification weapons including missiles and smart bombs, predict remaining useful shelf life, improve the reliability and readiness of the weaponry stockpile, and perform onboard readiness tests for critical elements of the offensive and defensive weapon systems. Note the NT-based sensors are best suited for harsh environmental conditions in military and commercial locations. These sensors can play a key role in sensing and monitoring of nuclear radiation and the outgassing of weapon propellant chemical/biological agents and toxic gases that can be potentially encountered in both military and civilian environments. Remotely located smart sensors can alert soldiers of harmful chemical and biological agents and explosives in the battlefield and its vicinity areas. The biosensor can be formed by casting suitable epoxy into a nanotube array and polishing the ends of the CNTs. The nanotube electrodes can be used to design a label-free immunosensor based on an electrochemical impedance spectroscope, and the sensor has good sensitivity. Nanoprobes are widely used for biosensors and electrophysiology applications. The electrochemical actuators are very sensitive and require very low actuation or drive voltage.
CHAPTER 3 Introduction to CoventorWare
CoventorWare is an integrated suite of tools designed to produce accurately modeled MicroElectroMechanical Sys-tems (MEMS) and microfluidics designs. CoventorWare supports both system-level and physical design approaches. The system-level approach involves use of behavioral model libraries with a high-speed system simulator. The sys-tem-level MEMS design can be used to generate a 2-D layout for physical level verification. The physical approach starts with a 2-D layout and involves building a 3-D model, generating a mesh, and simulating using FEM or BEM solvers. Custom reduced-order macromodels can be extracted for use in system simulations. Finally, the verified 2-D layout can be transferred to a foundry for fabrication. CoventorWare has numerous options, including design libraries and a variety of 3-D physics solvers. Various entry and exit points allow import and export of files from and to other third-party software. Version: 2006 License: Designer and Analyzer
3.1 Major Components:

Functionally, CoventorWare has these major components: Materials Properties Database: acts as the database for the materials used in the design flow. Process Editor: emulates the process flow that will fabricate the MEMS design. Architect: simulates design configuration using a system-level approach. This approach can incorporate CoventorWare parametric models, Saber generic models, as well as user-generated models (using the macro-modeling modules SpringMM, InertiaMM, DampingMM, and FlowMM). You can refine a system model, and then extract it to the 2-D Layout Editor for device design and FEM analysis. Designer: designs a MEMS model using 2-D layout tools and automatic 3-D solid model generation. These steps are also referred to as front-end design. Meshing: creates 3-D meshed layers based on the solid model. CoventorWare has several meshing options including surface, tetrahedral, extruded, and brick meshing. The meshing step prepares the model for FEM analysis. __________________________________________________________________________________ Dept of E&CE BEC Bagalkot 47
Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________ Analyzer Modules: sets boundary conditions and choose one or more solvers to perform FEM or BEM analysis. Each solver also has managed simulation capabilities: solver data can be postprocessed by varying parameters and iterating runs, or by curve-fitting results and applying equations for additional solver runs. You can also view the simulation results in a 3-D view and use the Query function to extract additional data from solver results.
3.2 Architect:
The ARCHITECT parametric libraries have been developed to allow system-level designers to simulate and rapidly evaluate multiple design configurations using a top-down, system-level approach, and to deliver precise behavioral models to the device designer. The parametric components are building blocks with pre-defined, but variable, characteristics. These components can be used to design a MEMS device efficiently at the system level. This approach also reduces personnel training requirements because system designers need not be MEMS design experts. The system model can include the parametric library components as well as other generic components (provided in CoventorWare) to simulate the MEMS device and its interface. This level of modeling analyzes much more detailed and complex behavior than is possible even with the most extensive FEM analysis, potentially reducing the number of prototyping cycles needed to complete the product design. The ARCHITECT suite has several distinct parametric libraries: electromechanical, dampers, magneto-mechanics, optics, sensors and actuators, and fluidics.
3.3 Designer:
Designer is a full-featured design creation tool especially suited for MEMS layout. The 2-D Layout Editor features all-angle drawing and a number of curve creation tools capable of creating and fitting true curves, as well as a wide variety of polygon creation and editing tools. The true curve capability allows curved MEMS components to be automatically meshed. The Layout Editor can handle simple or complex designs, with the ability to organize designs into hierarchical structures, and to push down editing to any level in the hierarchy. It is compatible with other layout formats, such as __________________________________________________________________________________ Dept of E&CE BEC Bagalkot 48
Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________ GDS, CIF, and DXF formats. Completed designs can be used to build 3-D models, which can then be meshed for analysis with CoventorWare or various third-party solvers.
Components of Designer: Material Properties Database: A default Material Properties Database (MPD) file is provided with the software. This file contains commonly used materials for MEMS processing, along with relevant mechanical, electrical, thermal, fluidic, and other generic material property values. The Process Editor uses the MPD as a source for materials from which to choose when specifying deposition steps. Process Editor: The Process Editor provides a way of creating or editing a description of the fabrication process, which is stored in the process file. The process file provides the process information needed to create a 3-D model. An actual wafer fabrication process can be represented by a series of deposit and etch steps. Layout Editor: The Layout Editor is a tool for drawing and editing multi-layered 2-D layouts. The layers can be combined in various ways to produce the masks that will be used to fabricate a device. Model Builder and Preprocessor: After a 2-D layout is complete, the 2-D information is combined with the process information to create a 3-D solid model of the device, which is then viewed in the Preprocessor. The solid model is needed for further analysis in CoventorWare. It can also be translated into other formats for use with third-party tools for meshing and analysis.
3.4 Analyzer: AN INTRODUCTION TO 3-D PHYSICS SIMULATION

Physics simulation starts with a 3-D model created from a 2-D layout and process file, or a model imported from a third-party solid modeling tool. The model is prepared for simulation by meshing it and naming the surfaces and volumes. Boundary conditions are then applied to these named surfaces and volumes to set up the problem, and various solver settings are configured. A single or point solution may be obtained, or a parametric study may be performed. Results from the solver may be viewed in table or graph form, visualized in 3-D, or queried for additional numerical values that can be plotted. The shaded boxes represent Analyzer functionality. MemElectro: With the CoventorWare solvers, the user can run various simulations on a model created and meshed in the Preprocessor or on a model imported and converted to the CoventorWare file format. This section documents the functionality of CoventorWares electrostatic solver, MemElectro, which __________________________________________________________________________________ Dept of E&CE BEC Bagalkot 49
Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________ can compute a matrix of capacitance and conductance values or can compute forces on conductors and dielectrics for MEMS designs.
Fig 3.1 MemElectro Settings and Results Flow
MemMech: MemMech is CoventorWares mechanical solver, which computes displacement and stress results. The user applies the boundary conditions set in solver dialog, and the solver generates output. The MemMech BCs dialog allows the user to set the mechanical boundary conditions for the solver. Patches may be fixed in 3-D space, pressure loads and temperatures may be applied, air flow may be set for convection analysis, boundary conditions may be applied to entire volumes, filter parameters can be set, and special boundary conditions for harmonic analysis can be selected.
Cosolve: CoSolveEM provides the capability of coupled quasi-steady electromechanicalanalysis. A typical application is the analysis of the electrostatic actuation of restrained mechanical structures. MemElectro provides the electrostatic analysis, and MemMech provides the mechanical analysis. __________________________________________________________________________________ Dept of E&CE BEC Bagalkot 50
Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________ CoSolve uses an iterative procedure to maintain consistency between the two solutions, that is, the mechanical deformation is correct for the applied electrostatic forcing.
Fig 3.2 CoSolve Single Step Simulation Flow
CHAPTER 4 COMB-DRIVES
Comb-drives are linear motors that utilize electrostatic forces that act between two metal combs. While comb drives built at normal human scales (size) are extremely inefficient there is the potential to minimize them to microscopic or nanoscale devices where more common designs will not function. Almost all comb-drives are built on the micro or nano scale and are typically manufactured using silicon. The electrostatic forces are created when a voltage is applied between the combs causing them to attract. The force developed by the motor is the force between the two combs (which increases with voltage difference, the number of comb teeth, and the length of the teeth, and decrease as the combs are further apart). The combs are arranged so that they never touch (because then there would be no voltage difference). Typically the teeth are arranged so that they can slide past one another until each tooth occupies the slot in the opposite comb. Restoring springs, levers, and crankshafts can be added if the motor's linear operation is to be converted to rotation or other motions.
4.1 V = applied electric potential = relative permitivity of dielectric = permitivity of free space (8.85 pF/m) n = number of pairs of electrodes t = thickness of electrodes g = gap between electrodes
4.1 Comb-Drive Actuator:

Electrostatic comb drive actuators exhibit a linear behavior. The actuator consists of two comb structures where the comb fingers are partly overlapping. The static and dynamic behavior has been described by a lumped element model. The work is focused on the realization of a feedback system in order to control the actuator position.
SEM photo of comb-drive close up of shuttle actuators. (1 cm is app. 100 microns)

Fig 4.1 Close view of Comb-Drive
close up of deflected fingers
4.1.1 Capacitive MEMS actuators: Capacitive sensors and actuators are widely applied in MEMS devices. There are several solutions in use, and perhaps the most widespread one is the so called comb-drive. A comb-drive is a laterally driven capacitive actuator, which is operated by electrostatic forces. The design of a MEMS comb-drive is based on the theory of parallel-plate capacitors. The capacitance is a function of the dimension and the shape of the two electrodes. In the case of comb-drives the structure consists of a movable electrode, and one or two fixed electrodes. The physical motion changes the spacing between the capacitor electrodes.
4.2 Simple Design of Comb-Drive:
Fig 4.2 Basic Structure of 9 Finger (movable) Comb-Drive
Fig 4.3 Structure of Spring
The initial geometry simulated consists of a fixed electrode with 10 comb-fingers, and a movable electrode with 9 fingers. The length of the fingers is 55.5 m and the width is 7 m for each finger. The thickness of the structural layer is 1.5 m. The lenght of the springs is 59.5m and the initial gap is 2 m.
4.3 Design Variables and Models

Design variables of the microresonator include thirteen geometrical parameters of the shuttle mass, folded flexure, and comb drive elements, the comb-drive voltage and the
number of fingers in the comb-drive voltage and the number of fingers in the comb-drive are detailed in Figure 4.4 and listed in Table 4.1. Additionally, geometric style variables, such as the width of the anchor supports, wba and wca, are necessary to completely define the layout, but do not affect the resonator behavior. Technology driven design rules set minimum beam widths and minimum spaces between structures. Maximum beam lengths are constrained to 400 um to avoid problems with undesirable curling due to stress gradients in the structural film and possible sticking and breakage during the wet release etch. Maximum width of beam is constrained to 20 um by the limited undercut of PSG to release the structures. The shuttle axle, the shuttle yoke and the comb yoke are at least 10um wide so that, they are relatively more rigid than folded-flexure beams. The comb yoke is allowed to extend to 700um, to fill up the entire flexure length allowed for resonator, even if the comb finger occupy only a fraction of a length of the comb yoke.
Fig 4.4 Design parameters.
Table 4.1 Design variables. Upper and lower limit are in micro-meters except for N and V.
4.4 Applications:
Comb-Drive has many applications. Few of them are described here. 1.) The displacement measurement device using comb-drive actuator. 2.) Comb-Drive micro resonator. 4.4.1 The displacement measurement device using comb-drive actuator: The non-linear characteristics of the mechanical admittance in comb-drive actuators that is the same as characteristic of a crystal oscillator in which the electric inductive feature is appeared .Therefore it will be possible to make a self oscillation circuit using comb-drive actuators. Fig 4.5 shows the electrical circuit used for self oscillation of comb-drive actuators. This is a simple Colpitt type LC oscillation circuit in which the comb-drive actuator is connected to the position of inductor. Two invert amplifier composed of two transistors make a positive feedback loop. First the comb-drive probe was contacted on the area of the lower part of the pattern and fixed the base stage height. Then the stage was scanned laterally so as to cross one pattern edge. During this scanning, self-oscillator frequencies were monitored. The measured frequency change is shown Fig 4.6 A large frequency drop at scan length of 1um was observed
Fig 4.5 The electrical circuit of self oscillation. .
Fig 4.6 Results of displacement measurement.
4.4.2 Comb-Drive micro resonator: Fiber optic sensors, in which the micromechanical resonator acts as a sensitive element, have been proposed for measurement of many physical parameters, such as force, temperature, pressure and acceleration. The basic principle is that the measured parameter changes the microresonator natural frequency. The flexural vibrations of the microresonator are excited and detected by light. An all-optical approach using optical fibers for light transmission offers the electrical passivity, while the output of the sensor, being a frequency, is transmission-line-independent. The use of metallic glasses as a microresonator material opens new possibility for sensing of outer actions through the changes of the magnetic field.
CHAPTER 5 MEMMECH ANALYSIS

MemMech is CoventorWares mechanical solver, which computes displacement and stress results. The user applies the boundary conditions set in solver dialog, and the solver generates output that may be viewed in a tabular form or rendered in three dimensions over the structures domain in the Visualizer. The flow chart shown in figure provides an overview of the MemMech functions:
Fig 5.1: Flowchart of overview of MemMech
5.1 MPD files:

The mpd directory stores the materials database files (.mpd). This database file stores all material properties used by the solvers during the computation phase. Material constants, thermal properties, and other attributes for each material used in a MEMS design are stored in this file. To create a solid model, the user needs an MPD file, a process file, and a layout. The MPD file, the Material Properties Database, stores characteristics of the materials used in a fabrication process. The properties associated with the materials include strength, density, conductivity, and thermal characteristics. A material is selected for each of the deposit steps
in the process. Simulation results depend on the physical and chemical properties of the materials that make up the model.
Fig 5.2 MPD file in coventor.
Table 5.1 Material properties
5.2 Process:
Process contains details about materials to used for fabrication process layered-wise, dimensions of each layer, details about photo resisters, masks and etc., The Process Editor allows users to create a flow simulating the foundry process that will fabricate the MEMS or microfluidics design. Materials, layer names, mask names, actions (deposit or etch), thickness or etch depth, photoresist and more are defined in a series of steps. The choice of materials selected in the process flow is dependent on the MPD, which stores all the materials and their associated properties needed to fully characterize the process flow. 5.2.1 Process library: When the user opens the Process Editor by selecting the create new process option, the only active window is the Pocess Library. This window is common to all process files, and uses a tree view to show the CoventorWare process steps, foundry process steps, and any user-defined process steps.
5.2.2 Process description: The Process Description window provides a spreadsheet-style view of the process steps in the current process file, including step number, step name, layer name, action, material, layer thickness, and more. Each process step is displayed in its own row.
Fig 5.3 Layers in the Fabrication process
5.3 Layout:
CoventorWare has its own 2-D drawing and editing tool, accessed from the Start Layout Editor icon to the right of the Layout file field. All layouts created in the Layout Editor proceed through a conversion process and are fully compatible with the rest of the software. How to start layout editing: Start Layout Editor Icon:When selecting files to be opened in the Layout Editor, you have these options from the icons drop-down menu: New from Process: When this option is selected, the masks specified in the process file are defined as the layers in the 2-D layout file (.cat). New from Netlist: This option becomes available if a schematic is selected in the Architect tab. This option allows the Layout Editor to extract a 2-D layout from an Architect schematic. New Blank Layout: This option allows you to create a layout without first defining a process. New from Template: This option allows you to import layer definitions from another cat file.
Open Layout: When this option is selected, the Layout Editor opens to the file specified in the file field.
Fig 5.4 Layout of Single comb drive
5.4 Preprocessor (3-D Model):

The Preprocessor is used for viewing 3-D solid models. It is also used for the naming the device surfaces and parts and generating a mesh in preparation for 3-D physics analysis. According to the dimensions in the process and layout coventor will develop a 3-dimensional model (refer figure 5.2). Model definitions: Edge: a line or a curve that connects two vertices. Face: a surface that is defined by a closed set of edges. For example, a cube has six faces. Note that the faces correspond to patches in the solver BCs windows. External face: an element face that coincides with a model face. An element is external if it has only one element adjacent to it or if it has two adjacent elements, but those elements are on different parts.
Internal face: an element face that coincides with a part or with a model face that is adjacent to two parts only. Layer: a collection of one or more parts those results from one or more identifiable steps in the fabrication process. All of the parts in a layer are made from the same material. Model entity: a generic term for any component of a solid model. For example, vertices, edges, faces, parts, and layers are all entities. Part: a volume that is defined by a closed set of faces and is made out of one material. For example, a cube can be a part. Patch: one or more faces that have the same name. Solid model: the 3-D topology of a device. A solid model consists of one or more layers. Vertex: a point in three-dimensional space. For example, a cube has eight vertices.
Fig 5.5 3-D view of single-finger comb drive 5.4.1 Geometry Browser: The Geometry Browser (refer the figure 5.5) displays the hierarchy of model and mesh entities. It has the structure of a typical Windows directory hierarchy and reveals the interactions and dependencies of the model and mesh in a selected database. The Geometry Browser shows how entities are related. There are three main folders in the Geometry Browser: Solid Model, Mesh Model, and Conductors/Dielectrics.
5.4.2 Meshing: The solid model tool uses the 2-D layout and process characterization information to build a solid model. Once the 3-D model is created, the next step in physical design is mesh generation. CoventorWare uses finite element and boundary element techniques for solving the differential equations of each physical domain in the problem. The differential equations are solved by discretizing the 3-D model into a mesh, which consists of a number of elements, each with a specified number of nodes. The mesh tool, accessed from within the Preprocessor, creates the finite elements or boundary elements necessary for simulation. When the meshing is complete, this information is exported for use by the rest of the software for additional conversion, application of boundary conditions, and computation by the solvers (Refer fig 5.6).
Mesh Definitions: Element: a volume that is defined by a connected set of element faces. Element face: a planar surface that is defined by a connected set of element edges (parabolic element faces can be non-planar). Element edge: a line of curve that connects two nodes. Mesh: a collection of mesh entities such as nodes, elements, element faces, and element edges. Mesh entity: an entity that belongs to the mesh. For example, nodes, elements, element faces, and element edges are mesh entities. Node: a point in three-dimensional space. All other mesh entities, namely element edges, element faces, and elements can be defined by a directed set of nodes. The order of nodes in the set determines the orientation of each mesh entity. Surface mesh: a mesh that descretizes the top face of a solid model. Volume mesh: a mesh that descretizes the volume enclosed by the faces of the solid model. Region: consists of one or more layers that touch each other.
Fig 5.6 3-D view of meshed comb drive
5.4.3 Preprocessor Flow: This section outlines the typical work flow for the Preprocessor. Individual users may incorporate other steps, such as partitioning or splitting regions. From the Designer tab, make sure the desired process and layout file are displayed in the appropriate field. Click on the arrow beside the Start Preprocessor icon, select New from the dropdown menu. Enter the desired model name in the dialog box, and click on OK. The Preprocessor will open when the software has finished building the solid model. From the Solid Model folder the user may choose to do any of these tasks: Add a plane, block, a wedge, or a detection screen. Use a plane or wedge to create symmetry. Use a plane or a block to define a partition. Partitioning a layer allows the user to delete unwanted features from the model or to create layers for more refined meshing. Select a part and assign material or material properties as needed.
Name faces. Name parts.
Designate a part as a conductor or dielectric. Designate a part as a solid or fluid. Remove a part from any mechanical simulation.
Select and add the layers to be meshed to the Mesh folder.
Name conductors. Right click on each mesh region and set the desired mesh type and other meshing controls. Click on the Generate Mesh icon. Save the model.
5.5 ANALYSATION OF COMB DRIVE: Till now it is seen that the designing process of comb-drive. Now we will analyze comb drive. We can analyze in Electrical domain for capacitance and voltage determination. In mechanical domain we are going to analyze displacement with respect to applied force. In Cosolver we are going to see displacement with respect to applied to voltage.
5.5.1 MEMMECH ANALYSIS: MemMech is CoventorWares mechanical solver, which computes displacement and stress results. The user applies the boundary conditions set in solver dialog, and the solver generates output that may be viewed in a tabular form or rendered in three dimensions over the structures domain in the Visualizer.
Fig 5.7 Memmech analyses of 4 finger comb drive In figure 5.7 it depicts the variation of displacement with respect to force by variation of color code. The red color shows highest displacement and blue color show less movement.
Fig 5.8 MemMech results
Figure 5.8 shows mech-Domain displacement data of comb-drive in X, Y & Z directions. Maximum and minimum displacement ranges are shown.
5.5.2 Co-solver analysis: CoSolveEM provides the capability of coupled quasi-steady electromechanical analysis. A typical application is the analysis of the electrostatic actuation of restrained mechanical structures. MemElectro provides the electrostatic analysis, and MemMech provides the mechanical analysis. CoSolve uses an iterative procedure to maintain consistency between the two solutions, that is, the mechanical deformation is correct for the applied electrostatic forcing. The setup of a CoSolve simulation requires that MemMech be set up for the mechanical analysis parameters and boundary conditions and that MemElectro be set up for the electrostatic analysis parameters and boundary conditions. So the user should set parameters for these solvers before setting CoSolve parameters.
CHAPTER 6 Results, Discussion and Conclusion

6.1 Results and Discussion:
Parametric computations returned the field distribution in the dielectric for input potentials from 10 to 40 V. Forces acting upon the set of movable fingers were computed at each increment of voltage. Figures 6(a) shows a plot of the finger displacement of the initial design, as a function of the applied voltage, while Figures 6(b) shows a plot of the stress distribution in the end region on the spring. It can be seen that the response is almost linear, as expected from a rectangular-shaped comb drive and the stress is concentrated in the weaker region of the springs. At 40 V the maximal displacement is 0.092 m.
Number of Fingers. 1 2 3 4 5 6 7 8 9
Max Voltage applied in Volts.
Max Displacement in micrometer.
31 40 20 40 40 40 40 40 40
0.00313 0.09400 0.00449 0.01640 0.18500 0.19300 0.19700 0.20200 0.20300
Table 6.1 List of no. of comb, Max Voltage applied and Max displacement.
Tables.doc
6.2 Conclusion:
In this project, fundamentals of Comb-Drive are discussed and design and simulation of a Comb-Drive actuator is carried out using CoventorWare. Applications of Comb-Drive are discussed. Experimental deformations were obtained using the CoventorWare. This project deals with the development of a methodology for modeling the Comb-Drive. In this project the Comb-Drives with different number of fingers are designed, simulated and analyzed. From this project it can be concluded that the Comb-Drive are the very accurate and sensitive devices that can be used for both sensing and actuation applications.
Reference: [1] Foundations of MEMS, Chang Liue, Pierson International Edition 2006.
[2] Simulation of Comb Drive for Fracture Lifetime Measurements, G. Somlay 1, Z.Szucs, A. Poppe, M. Rencz , Budapest University of Technology & Economics. [3] Modeling of a folded spring supporting MEMS gyroscope A Thesis Submitted to the faculty of the Worcester Polytechnic Institute, Victoria Steward.
APPENDIX Steps of Design and Simulation:

1) Process of Single finger Comb-Drive:
2) Layout of Single finger Comb-Drive:
3) Meshed 3D Model of Single finger Comb-Drive:

4) MemMech Model of Single finger Comb-Drive:

Full Report BE4thyr

Transféré par

Informations du document

Description originale:

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Full Report BE4thyr

Transféré par

Droits d'auteur :

Formats disponibles

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

CHAPTER 1 Introduction to MEMS

1.1 Introduction to Microelectronics:

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

1.2 Evolution of MEMS:

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

__________________________________________________________________________________ Dept of E&CE BEC Bagalkot 3

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

__________________________________________________________________________________ Dept of E&CE BEC Bagalkot 5

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

1.3 MEMS fabrication technologies

1.3.1. Bulk Micromachining

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

__________________________________________________________________________________ Dept of E&CE BEC Bagalkot 9

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

__________________________________________________________________________________ Dept of E&CE BEC Bagalkot 11

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

Figure 1.8 Isotropic and reactive ion etching of silicon

1.3.2 Surface Micromachining

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

Figure 1.9 Surface micromachining.

__________________________________________________________________________________ Dept of E&CE BEC Bagalkot 13

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

1.3.3 High-Aspect-Ratio (LIGA and LIGA-Like) Technology

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

Figure 1.10 LIGA fabrication technology.

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

__________________________________________________________________________________ Dept of E&CE BEC Bagalkot 20

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

CHAPTER 2 Sensors and Actuators

Fig 2.1 Gives idea about the system

__________________________________________________________________________________ Dept of E&CE BEC Bagalkot 21

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

Figure 2.3 Price-performance indicators for ICs, sensors, and actuators

__________________________________________________________________________________ Dept of E&CE BEC Bagalkot 22

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

2.1.1 THERMAL SENSORS:

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

2.1.2 RADIATION SENSORS:

__________________________________________________________________________________ Dept of E&CE BEC Bagalkot 24

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

Figure 2.5 Classification of radiation sensors according to their electromagnetic energy

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

2.1.3 MECHANICAL SENSORS:

__________________________________________________________________________________ Dept of E&CE BEC Bagalkot 26

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

2.1.4 MAGNETIC SENSORS:

Figure 2.7 Classification of magnetic microsensors by principle

__________________________________________________________________________________ Dept of E&CE BEC Bagalkot 27

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

2.1.5 BIO (CHEMICAL) SENSORS:

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

__________________________________________________________________________________ Dept of E&CE BEC Bagalkot 29

Simulation of Comb-Drive using CoventorWare __________________________________________________________________________________

__________________________________________________________________________________ Dept of E&CE BEC Bagalkot 30