Fullscreen HTML5Med Res (??MB)HTML4

MEMS: SILICON CHIP introduced you to tiny MEMS devices in the Digital Spirit Level project back in August 2011, which used a MEMS 3-axis accelerometer. Then in May this year, we described MEMS speakers which measure just 6.7 x 4.7mm. MEMS devices are microscopic and are typically fabricated from silicon, similarly to integrated circuits, combining mechanical and electronic elements in the same tiny package. Their mechanical components are precisely formed at micrometre scales! by Dr David Maddison M EMS devices can provide many different func- clopedia Britannica could be fitted on the head of a pin. tions. These include: accelerometers and gyro- That was achieved in 1985. In the lecture, Feynman also speculated about “swallowscopes as used in smartphones and airbag systems, display projection systems, in-wheel tyre pressure ing the doctor”, the concept of a miniature surgical robot. That goal too has more recently been partly achieved; sensors, biosensors such as blood pressure monitoring devices or ‘labs on a chip’, inkjet printer heads and many see the SILICON CHIP article in August 2018 on ‘pill cams’ and related devices. more that you would likely use in everyday life. The size of the devices formed may be measured in microns (one-thousandth of a millimetre), up to millimetres. Types of MEMS devices MEMS devices are typically sensors or actuators, or MEMS extends techniques used by the semiconductor industry to fabricate mechanical components such as multiple combinations thereof. Examples of MEMS sengears, beams, levers, diaphragms, springs and combs, all at sors are: a much smaller scale than traditional devices. Electronic • mechanical (force, pressure, velocity, position, acceleration etc) components can also be incorporated within the device, • thermal (temperature, heat often on the same piece of silicon. flow etc) MEMS technology was initial• chemical (composition etc) ly developed in the early 1960s, • radiant energy (wavelength, inbut it wasn’t known by that name tensity, polarisation, optical at the time. The term microelecswitching, laser etc) tromechanical systems was first • magnetic (field intensity, flux used in a US DARPA (Defense density, direction etc) Advanced Research Projects • electrical phenomena (electric Agency) report in 1986. field sensor, charge, voltage refOne of the first times that the erence etc) miniaturisation of machines was Other devices include oscillarecognised as a desirable objectors, displays, printers, motors and tive was in 1959, when the faswitches. mous Caltech physicist Richard In this article, we will describe Feynman gave a speech entitled as many of these various types of “There’s Plenty of Room at the MEMS devices as we have space Bottom: An Invitation to Enter a to fit. New Field of Physics”. In this speech, he issued two Uses for MEMS devices Fig.1: Bill McLellan’s 1960 answer to Richard challenges: Some common applications of One was to build a tiny elec- Feynman’s challenge: an electric motor smaller MEMS devices are: tric motor, which was achieved than the head of a pin. It is less than 0.36mm • Automotive and aerospace: senin 1960, but without the break- per side, even smaller than specified. Feynman sors for airbag actuation; fluid through technology that Feyn- had hoped for a breakthrough in technology; however, this was made with conventional level and pressure sensors; navman had hoped for (see Fig.1). techniques very cleverly applied. It still won igation; motion sensors for susThe second was to shrink let- Feynman’s US$1000 prize (about AU$12,000 in pension, active suspension and ters such that the entire Ency- today’s money). Source: Caltech Archives. 14 Silicon Chip Australia’s electronics magazine siliconchip.com.au Micro icroEElectro lectroM Mechanical Systems • • • • • • stability control; brake force sensor for anti-lock brakes; tyre pressure and temperature sensor; various avionics sensors. Chemical: various types of chemical analysis. Communications: mobile phones; fibre optic switches; voltage-controlled oscillators; lasers; optical splitters, couplers, modulators, attenuators and switches; DC-toRF frequency switches; fibre optic components. Computers and electronics: hard disk heads; inkjet printer heads; optical projectors; gaming controllers. Medical: blood pressure sensors; motion sensor to monitor activity such as in heart pacemakers; biological sensor systems; implanted sensors; sensors in prosthetic devices; ‘labs on a chip’. Navigation and Earth science: accelerometers; gyroscopes; seismic motion detectors. Military: munitions guidance; arming systems for munitions; numerous other applications listed under other categories above. Key discoveries and inventions Some key scientific discoveries and technologies that led to the development of MEMS devices are as follows, in date order. • 1745 and 1748: while modern electric motors generally use electromagnetic principles, it is possible to design a motor using electrostatic principles instead. In 1745, Benedictine monk Andrew Gordon described the “electrical whirl” and “electric chimes”, the first electrostatic mechanical devices capable of rotary and linear motion. In 1748, Benjamin Franklin invented the electric wheel, which is regarded as the first true electrostatic motor. Benjamin Franklin is often erroneously credited with the invention of electrostatic electric chimes (“Franklin Bells”), but these were invented by Gordon and used as an annunciator for his experimental lightning rod in 1752. Modern motors are electromagnetic devices as they are significantly more compact and powerful; however, for MEMS devices where it is difficult to fabricate coils to generate magnetic fields, electrostatics is often used instead. For more details, see the following videos: “Electric whirls” – https://youtu.be/6hkIGIAgxFU “Franklin’s Bells (5b1030)” – https://youtu.be/0TvvYa_ Qk6k “Electrostatic Motor” – https://youtu.be/9NkUcJBqVB4 • 1947: the first transistor was invented, paving the way for semiconductor fabrication technologies and electronic technologies that would later be used for MEMS. • 1954: the piezoresistive effect was discovered in silicon and germanium, where it is much greater than in DIY MEMS? Fig.2: Nathanson’s resonant gate transistor. It consists of a gold beam 0.1mm long and 5-10 microns thick which resonates at 5kHz. The inventor describes it as “an electrostatically excited tuning fork employing fieldeffect transistor ‘readout’.” Source: Nathanson et al., 1967, courtesy IEEE. siliconchip.com.au We saw an interesting but rather expensive book on DIY MEMS called “DIY MEMS: Fabricating Microelectromechanical Systems in Open Use Labs” by Deborah Munro from New Zealand. According to the author, MEMS devices could be fabricated in open-use facilities. You can read sample pages or buy the book at Amazon.com There is commercial software for designing MEMS layouts as well as other types of devices called “Layout Editor”. However, it can be used as a free file viewer for various microelectronics designs of any size, or as a free editor for small designs. See https://layouteditor.org/ Australia’s electronics magazine November 2020  15 Fig.3: K.E. Petersen’s electrostatically-driven torsional scanning mirror is etched from a single piece of silicon, with a reflective coating applied to the mirror’s surface. metals. This means that the material changes its resistance in response to a force. So these materials can be used to sense force, an effect now utilised by strain gauges, pressure sensors and certain accelerometers among others.   Strain gauges based on this effect were developed in 1958, with Kulite (https://kulite.com/) producing the first commercial strain gauge in 1959. They also invented the silicon pressure sensor in 1961. • 1959: Jack Kilby of Texas Instruments filed the patent for the first integrated circuit (US Patent 3138743; https://patents.google.com/patent/US3138743A/en). He and Robert Noyce (US Patent 2981877; https://patents. google.com/patent/US2981877/en) of Fairchild Semiconductor are considered the co-inventors of the integrated circuit. Fig.4: the different results achieved by bulk micromachining methods with wet and dry etching and isotropic and anisotropic processes. The dark bands represent the etch-resistant masking material. The isotropic methods undercut the mask while the anisotropic methods do not, but must be aligned with the crystal matrix. 16 Silicon Chip This led to small-scale silicon fabrication technologies which are also applicable to MEMS. • 1968: arguably the first MEMS device in terms of the modern understanding of such devices was a 1968 invention (US Patent US3413573; https://patents.google. com/patent/US3413573/en) by Harvey Nathanson. It was a resonant gate transistor comprising a mechanical resonator and a transistor (Fig.2). The purpose of this device was to act as a tuner in miniature radios. The cantilever was about 1mm long. It was created using similar techniques as are used today; a batch fabrication process in which layers of metal and insulators on a silicon substrate are alternatively shaped and undercut by etchants, etchant-resistant masks and sacrificial layers. • 1970: the first silicon accelerometer was produced by Kulite, based on piezoresistivity of silicon where it changes its resistance in response to a mechanical load. • 1977: the first capacitive pressure sensor was developed at Stanford University. • 1979: HP produced the first micromachined inkjet nozzle, “thermal inkjet technology”. • 1980: K.E. Petersen of IBM invented the electrostatically-driven torsional scanning mirror using batch photolithography and thin-film techniques. It consisted of a flat armature-like shape made and etched from a single Fig.5: how surface micromachining uses a sacrificial layer (tan), which is eventually removed, to produce a freestanding structure; in this case, a cantilever beam. Source: memsnet.org Australia’s electronics magazine siliconchip.com.au Fig.6: a MEMS wafer subassembly joined to a CMOS wafer integrated circuit subassembly using eutectic and fusion wafer bonding. A cross-section of the final result is shown at upper right, with a plan view below. This is a gyroscope assembly. Source: Allan Hilton and Dorota S. Temple. piece of silicon, in which the mirror surface had a reflective coating – see Fig.3. The silicon arms (22 and 24) attached to the mirrored surface (30) were arranged as a torsion bar and could twist in response to electrostatic forces as supplied by the electrodes mounted beneath and on either side of the longaxis centreline of the reflector portion (14 & 16). This allowed a light beam to be reflected in one direction or another. This device is now the basis of digital video projector systems (pioneered by Texas Instruments and called digital light processing [DLP]) and optical switches, for example, to switch between several optical fibres, among other applications. See https://patents.google.com/patent/US4317611/en • 1981: IBM invents the scanning tunnelling microscope (STM) that can image individual atoms on a surface using a cantilever and probe. • 1982: a MEMS-based disposable blood pressure sensor is produced by Foxboro/ICT and Honeywell, selling for US$40. • 1982: the LIGA process is invented in Germany (more details below). • 1984: the first polysilicon MEMS device is produced (Howe, Muller). Fig.7: the LIGA process for making high aspect ratio MEMS devices. The first step is at the top, and the process continues clockwise. • 1985: the atomic force microscope (AFM) is invented, based on IBM’s STM. • 1988: the first electrostatic side-drive motors (100 microns across) are made by Richard Muller et al. at UC Berkeley. • 1989: an electrostatic lateral comb drive is fabricated in polysilicon (Tang et al.). Mask F SFx+ Etch Silicon nCFx+ Deposit Polymer Polymer (nCF2) F SFx+ Etch Fig.8: a tall, high aspect ratio gear produced with LIGA technology. siliconchip.com.au Fig.9: in deep reactive ion etching, an area is etched, a polymer coating is deposited and then further etching is performed. The polymer coating is preferentially etched at the bottom and not on the sidewalls due to the dominant flow direction of the plasma etchant. Australia’s electronics magazine November 2020  17 Fig.10(a): a silicon structure formed with deep reactive ion etching (DRIE). Fig.10(b): after the structure in Fig.10(a) is modified by removing the outer pillars and sharpening the central pillar with reactive ion etching (RIE), the result is a needle for interfacing with biological cells. Source: Yael Hanein et al. • 1992: the MEMS deformable grating light modulator (GLM), also known as the grating light valve (GLV), was invented. It has uses in display technology, graphic printing, lithography and optical communications. • 1993: the first surface micromachined accelerometer, the TI ADXL50, went on sale. It was mainly used for airbag deployment systems. More on this later. • 1994: Bosch patents the process for deep reactive ion etching. • 1995: Xenon difluoride, XeF2, was demonstrated as an isotropic etchant for MEMS and used to dissolve sacrificial layers to release moving parts. It is also highly selective, meaning it will not dissolve certain materials but will fully dissolve others giving excellent design flexibility for MEMS devices. • 1999: Lucent’s “LamdaRouter” optical network switches are released, based on MEMS devices. Fabrication techniques Fig.11: a Damasko watch spring made from polycrystalline silicon, which they refer to as “Epi-PolySilicon” (EPS). The silicon is made by vapour deposition followed by deep reactive ion etching (DRIE). It has many advantages over a traditional spring such as being non-magnetic, temperature insensitive, of minimal asymmetry and with highly precise dimensions. 18 Silicon Chip MEMS devices are made using integrated circuit fabrication techniques such as photolithography, etching and deposition etc. But enhancements and modifications of those processes are required, as well as new processes not normally used for IC fabrication. The fabrication processes for MEMS are known generally as microfabrication, and can be broadly divided into two high-level categories. Bulk micromachining, surface micromachining and the related process of wafer bonding are the standard methods. The other category is designed for structures with high aspect ratios and is known as HARMST (high aspect ratio microsystems technology). The main HARMST technologies are LIGA (a German acronym for lithography, galvanoforming moulding); silicon ion etching; and glass and hot embossing. Other, less-common fabrication methods utilise lasers, ion beams and electrical discharge machining. Fig.12: the evolution of MEMS accelerometers, from the 1991 prototype to 2004. Today, such sensors incorporate additional functions such as gyroscopes and are used for airbag inflation, vehicle stability control and vehicle rollover detection among other purposes. Australia’s electronics magazine siliconchip.com.au iPhones disabled by helium gas Fig.13: the functional sections of the ADXL50 accelerometer. Common materials used to manufacture MEMS devices are silicon, polymers, metals and ceramics. Bulk micromachining Bulk micromachining involves taking a substrate and using mechanical or chemical means to remove material. A popular chemical means involves immersing a substrate in an etchant chemical to remove material, a process akin to using ferric chloride for etching patterns on a PCB. This is called wet etching (see Fig.4). With appropriate choices of etchant, etchant temperature and substrate, the rate and preferred direction of etching can be controlled. For example, it is possible to selectively etch along certain crystal planes of a silicon substrate (anisotropic etching) or etch them all evenly at the same time (isotropic). The etching process requires that a suitable masking material, such as silicon dioxide, is used to protect those areas that are to remain. With etching, it is possible to undercut protected areas. It is also possible to dry etch using vapours or plasma instead of liquids. Surface micromachining There are many variations of surface micromachining, but they all involve a multi-stage deposition process in which a combination of both permanent and “sacrificial” layers are laid down (Fig.5). The sacrificial layers are there to support an overlying structure. Once that has been deposited, the sacrificial structure can be removed, for example by etching or dissolving it, leaving a structure such as a cantilever beam. A common sacrificial layer is PSG or phosphosilicate glass. Similar etching and dissolution processes can be applied as with bulk micromachining. Wafer bonding Wafer bonding is a process by which similar materials, such as silicon wafers, can be bonded to each other, or to dissimilar materials such as glass. The technique can be used to produce materials with a variety of desired properties, it can be used for encapsulasiliconchip.com.au This may sound like a myth, but it is true. About two years ago, a new medical MRI facility was being tested, and during testing, about 40 iPhones and Apple watch devices became disabled, but no Android devices were affected. It was thought that the machine must have emitted some type of electromagnetic pulse during testing that destroyed the phones. It was later discovered that there was a helium leak during testing which disabled the devices. iPhones, Apple watches and numerous other devices use MEMS oscillators to generate clock signals instead of traditional quartz crystal oscillators as they are cheaper and smaller. These devices are hermetically sealed in a package which contains either an inert gas or a vacuum. Changes to the gas mix or pressure inside the package can affect the oscillation frequency to such an extent that its output frequency is outside of the bounds at which the CPU or other clock-driven components will function. As helium molecules are small, it is very difficult if not practically impossible to seal the package against an infusion of helium. Therefore, the gas will diffuse through the hermetic seal, changing the atmosphere inside the device and causing its oscillation frequency to shift. This is not usually a problem as such devices are usually only exposed to the very low concentration of helium naturally present in our atmosphere. The devices returned to operation after a few days. The fact that it only affected Apple devices is because most Android devices use quartz oscillators. Apple mentions the susceptibility to helium in its documentation. The MEMS device in question is the SiTime SiT1532 and is said to be the world’s smallest (1.5mm x 0.8mm), lowest power 32.768kHz oscillator, and twice as accurate as a quartz crystal. See the video “MEMS oscillator sensitivity to helium (helium kills iPhones)” at https://youtu.be/vvzWaVvB908 Tests in that video show the device is disabled in a 2% helium environment after 30 minutes. Hydrogen molecules are slightly larger than helium molecules, and did not affect the device in that experiment. The video author also does a very interesting teardown of the MEMS device. tion purposes, or it can be used to create large multi-layered structures – see Fig.6. A variety of bonding techniques can be used such as fusion, anodic, thermocompression, eutectic, glass frit and adhesive bonding. LIGA LIGA is suitable for extremely high aspect ratio parts such as a column several millimetres tall but only 0.03mm thick (see Figs.7 & 8). LIGA works as follows: 1) A thick layer of PMMA (commonly known as Perspex or acrylic) is deposited onto an electrically conducting substrate such as silicon or metal. This PMMA is designed to be sensitive to X-rays or UV light. 2) The PMMA is exposed to X-rays or UV light via a mask and “developed” to remove unwanted material from the exposed areas. 3) Metal is deposited by an electrolytic process akin to electroplating, to fill the cavities where the PMMA was removed. 4) The PMMA is removed, such as by dissolution, leaving a free-standing metal structure. Australia’s electronics magazine November 2020  19 with a plasma as with bulk micromachining, but then the process is stopped, and the hole has an inert Teflon-like polymer layer deposited in it. The etching process then continues, but since the plasma is coming from a vertical direction, the sides of the hole are protected, while the protective layer in the bottom is removed and the substrate to be etched can then also be removed. The process is repeated until the desired hole depth is achieved. Hot embossing Fig.14: an electron micrograph of the ADXL50 single-axis accelerometer sensor. Since the X-ray source has to be a highly collimated beam from a synchrotron, this makes such a method expensive for parts fabrication. A variation of this process takes the part made and then uses it as a tool to create an impression into a polymer layer. The impression formed is then filled with metal. This moulding process can be repeated many times, reducing cost. UV LIGA is a cheaper process and doesn’t need a synchrotron source, but is only suitable for lower aspect ratio parts. Ion etching Deep reactive ion etching (DRIE) is used for making deep, high aspect ratio holes for MEMS devices. But it can also be used to fabricate other devices such as watch springs and deep trenches for capacitors in DRAM chips (see Figs.9-11). The most common process involves standard etching In hot embossing, a high aspect ratio metal part is made by another MEMS process such as LIGA with the inverse pattern of the part that is to be fabricated, and that is used as a mould to make a plastic part. Both the mould and a mouldable plastic are pressed together under vacuum to make the part. Such parts are cheap and are used in microfluidics for medical applications. For more information, see our detailed article on fluidics and microfluidics in the August 2019 issue (siliconchip. com.au/Article/11762). The first popular MEMS device The first MEMS device to obtain large-scale market acceptance was an accelerometer based on CMOS technology. It was fabricated using surface micromachining, as was the device by Nathanson mentioned earlier. The device was made by Analog Devices and called the ADXL50 (see Figs.12-14) and was released to the market once it was fully qualified, in 1993. Its application was to trigger airbags in cars (for more information on airbags, see our November 2016 article at www.siliconchip.com.au/ Article/10424). It incorporated both electronic circuitry along with micromachined structures. How forces change as objects shrink As devices shrink, the relative strength of various natural forces changes. Gravity becomes less important, but the van der Waals force (a short-range force between atoms and molecules) becomes proportionally strong. When the scale of an object changes, its volume changes by the cube of one dimension, and its surface area by the square of that dimension. At smaller scales, friction becomes more significant than inertia; heat dissipation (proportional to surface area) is more significant than heat retention (proportional to volume); and electrostatic forces are more significant than magnetic forces. As devices become smaller, they can be heated or cooled much more quickly, which is important for thermally-activated devices like some inkjet heads. Heat dissipation is not a major problem in most cases. The smaller a cantilever beam is, the lower its spring constant and the more flexible it is. 20 Silicon Chip Electrical resistance is inversely proportional to scale while capacitance changes linearly with scale and electrostatic forces change with the square of scale. Electromagnetic forces scale with the fourth power of conductor length, but for permanent magnets, the amount of strength retained is roughly linear with size (depending on their geometry and the specific application). The fact that electromagnetic forces decrease so dramatically with scale is the reason they are not commonly used in MEMS devices. An electrostatic device is preferred to an electromagnetic device, as the forces involved scale with the square of the dimension, not the fourth-power. In microfluidic devices, a reduction in radius of ten times results in a 10,000 times increase in pressure drop per unit length, due to a fourth power dependence. Consider a mirror on a MEMS device that might be used as part of an optical switch. A 50% reduction in the height, width and thickness of such a device results in the Australia’s electronics magazine torque required to rotate the mirror being reduced by a factor of 32. Beyond MEMS Beyond MEMS is NEMS or nanoelectromechanical systems. These are like MEMS devices but at the nanometre (one-millionth of a millimetre) scale. They are the next step beyond MEMS, and move into the realm of machines that can directly manipulate molecules like DNA, as in nature. As an example of a nanoscale machine from nature, consider the following simulation video by Australia’s Walter and Eliza Hall Institute of Medical Research of various processes involving DNA: www.wehi.edu. au/wehi-tv/molecular-visualisations-dna There are many other similar videos at https://www.wehi.edu.au/wehi-tv Apart from the possible future development of NEMS to manipulate DNA and other biological molecules, experimental NEMS devices are currently being made. There are unique challenges at such scales as intermolecular forces dominate. siliconchip.com.au A look at some MEMS devices There is already a vast variety of MEMS devices available. Here are just some of them – but it is simply not possible to cover all of them in the available space. Some other uses for MEMS not discussed below include blood pressure monitors, pressure monitors for other applications, pill cams, ultrasonic transducers, DNA microarrays, micropumps, flow sensors and microfluidics applications. Texas Instruments digital light processing (DLP) A scanning electron microscope image of the micromirrors on the DLP device. DLP is a MEMS video projection technology using micromirrors to direct a beam to a projected area or away from it and onto a heatsink. Toggling the micromirrors rapidly gives control of brightness from 0% up to 100%. Colours are produced either with one DLP chip and a colour wheel, or with three DLP chips and three differently coloured beams of light. A MEMS micro-mirror device, the core component of a DLP device. Each micromirror drives one pixel. The mirror is mounted on a suspension device with a torsional restoring spring. The mirror is moved by electrostatic forces from the columns at upper left and lower right. Source: Wikimedia user Egmason. MEMS accelerometers MEMS acceleroMotion meters are made of 1.3 Micron Gap many interdigitated 125 2 Micron Microns Overlap fingers, similar to thick the comb drive A single finger of a shown overleaf. typical accelerometer As the device sensor element. It is a differential capacitor experiences a where the rate of force, the capchange of the output is acitance between proportional to the force the fingers changes. experienced. Source: Analog Devices. Rotary MEMS motors Rotary MEMS motors may be driven by electrostatic or by other means. MEMS three-axis gyroscope A MEMS gyroscope is correctly known as a Coriolis vibratory gyroscope. It contains parts that vibrate in all three axes. They will tend to continue to vibrate in the same plane, but if an external rotational force is applied, the Coriolis effect causes a force to be generated between the vibrating structure and its support. This force is measured to determine the rate of rotation. Accelerometers and gyroscopes can be combined in one device, which is then known as an inertial measurement unit (IMU). A MEMS threeaxis gyroscope: (1) outer frame (2) inner frame (3) driving comb electrodes (4) parallel plate sense electrode (5) double folded beams (6) anchors (7) linear beams and (8) self-rotation ring. Source: Minh Ngoc Nguyen et al. Grating light valve (GLV) GLV is a technology that competes with DLP for display projection. Each pixel in a display device is representing by multiple ribbons which are moved electrostatically by a distance a tiny fraction of the wavelength of light. When all the ribbons are aligned, the device acts as a mirror, and all light is directed towards the image. When the ribbons move apart, a diffraction grating is formed. In that case, only some light is directed to the image, while other light goes elsewhere. When the distance between adjacent ribbons is ¼ of the light wavelength, no light is reflected towards the image. By varying the distance between zero and ¼ wavelength, a range of brightnesses is generated. Spectrometer on chip A spectrometer for chemical analysis can be fabricated with MEMS. A MEMS spectrometer on a chip by Si-Ware Systems, on their proprietary Silicon integrated Micro-Optical Systems Technology (SiMOST) platform. MEMS atomic force microscope (AFM) AFMs are based on techniques from scanning tunnelling microscopy (STM). Today, AFM probes or even the principal parts of the device are made with MEMS technology. AFMs are capable of imaging individual molecules and sensing or manipulating individual atoms. The operating principle of an atomic force microscope. PZT refers to a piezoelectric material that can change its dimensions in response to an applied electric field. The tip on the cantilever follows the atomic profile of the surface, with its position being monitored by the deflection of the laser or by other methods. Source: Wikimedia user OverlordQ. An atomic force microscope on a chip developed at Laboratory for Dynamics and Control of Nanosystems at the University of Texas by M. G. Ruppert, A. G. Fowler, M. Maroufi and S. O. R. Moheimani. Strain gauges A MEMS strain gauge relies on the change in capacitance of interdigitated electrodes as it is extended. An electron microscope image of a MEMS electrostatic motor with false colour. The central red object is the bearing, which is surrounded by the rotor. Around the rotor are the stators which are driven with phased voltages. Source: www.mems-exchange.org siliconchip.com.au A grating light valve (GLV) from Silicon Light Machines, Inc. Australia’s electronics electronics magazine magazine Australia’s N November ovember 2020  21 Gears MEMS IR sensor Gears can be fabricated with MEMS, as seen below. Infrared sensors can use photonic sensors such as in CCD or CMOS devices, or they may sense heat such as with thermoelectric infrared sensors. Thermoelectric sensors have the advantage of lower noise and possibly lower cost than photonic sensors. Infrared radiation heats a thermocouple, producing a voltage proportional to the radiation intensity. An actual MEMS strain gauge. As the device is stretched, the capacitance changes in relation to the amount of extension. Source: Michael Suster et al., Case Western University. Optical switches A MEMS optical switch contains several optical fibre inputs and outputs, and any input can be switched to any output via the use of two MEMS tilt mirror arrays. A schematic view of a 3D optical switch. A MEMS demonstration geartrain. Such gears have been driven at 250,000rpm. Source: Sandia National Laboratories. MEMS inkjet printer heads Inkjet printer heads are a common MEMS device in everyday use. A recent development is the move from rapid heating and bulk piezoelectric materials to thin-film piezoelectric materials which are deposited as part of the MEMS fabrication process. This provides more design flexibility and lower cost. Microfluidic technology is also incorporated into inkjet printer heads. Comb drive A MEMS comb drive is a linear actuation mechanism that consists of two interlocking microscopic parts resembling hair combs. As a voltage is applied between them, the parts are drawn together by electrostatAn electron ic forces. Comb microscope image drive actuators of comb drive have been used components. With the application of as the driving elean electric field, the ments for resona- interdigitated fingers tors, electromeare drawn toward each other. When chanical filters, the electric field is optical shutters removed, silicon springs return the and device to its starting microgrippers to position. Source: name just a few Sandia National Labs. applications. Fig.36: a cross-sectional diagram of a MEMS thermoelectric IR sensor. Infrared radiation enters the device and heats the thermocouple. G represents the paths of thermal losses. Source: Dehui Xu, Yuelin Wang, Bin Xiong and Tie Li. MEMS loudspeakers and microphones MEMS loudspeakers are relatively new and were featured in the May 2020 issue of SILICON CHIP (siliconchip.com.au/ Article/14441). MEMS microphones are now commonly found in consumer devices such as smartphones, microphones with earphones, headsets etc. A cross-sectional diagram of a Philips inkjet printer head in which three MEMS wafers are bonded together. The ink is propelled via a thin-film piezoelectric driver which is also deposited during the fabrication process. Source: Philips. Exterior view and a cross-section of a TDK T4064 MEMS microphone, 2.7mm x 1.6mm x 0.89mm. The device has an ASIC (applicationspecific integrated circuit) incorporated into the housing. The diaphragm and the backplate together act as a parallel plate capacitor, and when the diaphragm, moves the capacitance changes and an electrical signal is produced. A comb driver actuator as the driver for a resonator device. Source: Wikimedia user Huseyintet. 22 22  S Silicon Chip Australia’s Australia’s electronics electronics magazine magazine siliconchip.com.au Switches Bio-MEMS MEMS switches offer the ability to switch frequencies between DC and 14GHz (Analog Devices commercial models) and have the advantage of being reliable, small (4mm x 5mm) and having low power consumption. A selection of Bio-MEMS devices follows. Smart contact lens An example of a smart contact lens, the SENSIMED Triggerfish with MEMS technology with continuous ocular monitoring for glaucoma patients. Several companies are developing MEMSbased smart contact lenses. These may have features such as autofocus, data display via Bluetooth, intraocular pressure monitoring for glaucoma etc. An RF relay which uses a comb drive as the actuator. Source: L. Almeida et al., Auburn University. Debiotech NanoPUMP This device is designed for the transdermal infusion of insulin or other substances. It is also the MEMS pump component of the JewelPUMP insulin infusion system and is connected to a reservoir with enough product for a week. It is connected to the patient via a flexible cannula. It also connects to a monitoring and control App on a smartphone. Neural probes MEMS can be used to fabricate silicon neural probes for brain research. Glucose sensor A selection of MEMS neural probes by NeuroNexus. Microneedles MEMS microneedles are fabricated to deliver medication just below the skin. An Analog Devices EVAL-DGM1304SDZ evaluation board featuring a single-pole, four-throw MEMS ADGM1304 switch as well as a calibration transmission line at the bottom. The MEMS chip is at the junction of the five RF lines. MEMS oscillators MEMS oscillators are smaller, cheaper, more temperature stable, more rugged and more power-efficient than quartz crystal oscillators. In some cases, their frequency can be programmed from 1Hz to 725MHz in 1Hz increments. They have found applications in areas such as automotive electronics and smartphones. Also, see the related panel earlier on the effects of helium. A MEMS glucose sensor designed for implanting. Source: Columbia BioMEMS Laboratory. An implantable MEMS sensor for continuous glucose monitoring is being developed (see above). Glucose enters a chamber via a semipermeable membrane and binds with a glucose-sensitive substance attached to a diaphragm. The diaphragm is made to vibrate via an external magnetic field which interacts with a magnetic permalloy attached to it. The vibrational amplitude changes according to glucose concentration. This is measured via the change in capacitance between the moving and the ground electrode. DNA nanoinjector A MEMS DNA nanoinjector invented at Brigham Young University in the USA allows scientists to inject DNA into living cells. A MEMS oscillator. The resonator beam is driven by electrostatic forces between the beam and an electrode beneath it. The dual-output Microchip DSA2311 comes in a 2.5mm x 2.0mm x 0.85mm package and each output can operate between 2.3MHz and 170MHz. Source: Microchip Technology, Inc. siliconchip.com.au A DNA nanoinjector. Australia’s electronics electronics magazine magazine Australia’s The MEMS-fabricated DebioJect intradermal injection microneedle array by Debiotech, for delivery of medications just below the surface of the skin. Virus detection MEMS plays an important role in COVID-19 testing. One test involves the partitioning and multiplication of a small amount of a patient’s viral genetic material into a much larger amount that is easier and more accurate to analyse. This way, a test result can be obtained in minutes rather than hours. Part of the MEMS Microfluidic Array Partitioning chamber of the Combinati Absolute Q platform. This is used for rapid polymerase chain reaction (PCR) analysis, for COVID-19 as well as other tests and analyses. SC N November ovember 2020  23 2020  23