As microsensor technology continues to grow and mature, the issues of cost and manufactureability become key issues in determining whether silicon transducer technologies are commercially viable. The dissolved wafer process is an attractive manufacturing technology for the production of low cost, high volume transducers. The process requires only 3 masking steps and the tooling for micromachining is inexpensive. Six sigma level yield is attainable and the turnaround time for a lot is 1.5-2 weeks for one work shift operation. This technology is currently being implemented in commercial production low cost inertial instruments.

A set of micromachined structures for holding optical fibers in anisotropically etched v- grooves has been produced. The structures are made of bulk silicon and etched in the same etch step as the aligning v-grooves, using the photo voltaic electrochemical etch stop technique (PHET). Most etching was performed in KOH solutions, however, a shallow investigation was made on the PHET technique in tetra methyl ammonium hydroxide. The structures were produced in a variety of shapes based on cantilever beams and doubly clamped bridges. They facilitate the mounting of optical fibers substantially and guide the fibers into position in the grooves.

This paper describes significant advancements to a low doped etch-stop technique which increases the differential etch rate of high doped to low doped silicon (R_h/l) by a factor of 4 or greater, to a value of up to R_h/l approximately equals 50:1. The objective of the research was to achieve this increase in the differential etch rate by decreasing the etch rate of the low doped silicon epilayer, resulting in the development of an exceptional technique for rapid, safe, and high-quality etching of complex micro-structures. The technique has been confirmed by the production of devices. These include both 10 micrometers thick diaphragms and a complete accelerometer structure, created fom n on n+ epitaxial samples.

We report here on the results of experiments concerning particular bonding processes potentially useful for ultimate miniaturization of microfluidic systems. Direct anodic bonding of continuous thin pyrex glass of 250 micrometers thickness to silicon substrates gives multiple, large voids in the glass. Etchback of thick glass of 1200 micrometers thickness bonded to silicon substrates gives thin continuous glass layers of 189 micrometers thickness without voids over areas of 5 cm X 12 cm. Glass was also successfully bonded to glass by thermal bonding at 800 degrees C over a 5 cm X 7 cm area. Anticipated applications include microfabricated DNA sequencing, flow injection analysis, and liquid and gas chromatography microinstruments.

In this paper, we report our results using eutectic bonding with the aluminum/germanium alloy to create high quality bonds. The results of a series of experiments conducted to optimize eutectic alloy bonding for MEMS are described. Issues discussed include surface preparation, eutectic composition, bonding apparatus and bonding conditions (temperature and time).

In order to help provide access to advanced MEMS technologies, and lower the barriers for both industry and academia, MCNC, and ARPA have developed a program which works to provide users with access to both MEMS processes and advanced integration techniques. The two distinct aspects of this program, the MUMPs and Smart MEMS, will be described in this paper. The multi-user MEMS processes (MUMPs) is an ARPA-supported program created to provide inexpensive access to MEMS technology in a multi-user environment. MUMPs is a proof-of-concept and educational tool to aid the developemnt of MEMS in the domestic community. MUMPs technologies currently include a 3-layer polysilicon surface micromachining process and LIGA processes that provide reasonable design flexibility within set guidelines. Smart MEMS is the development of advanced electronics integration techniques for MEMS through the application of flip chip technology.

The monolithic integration of micromechanical deviecs with their controlling electronics offers potential increases in performance as well as decreased cost for these devices. Analog devices has demonstrated the commercial viability of this integration by interleaving the micromechanical fabrication steps of an accelerometer with the microelectronic fabrication steps of its controlling electronics. Sandia's Microelectronics Development Laboratory has integrated the micromechanical and microelectronic processing sequences in a segregated fashion. In this CMOS-first, micromechanics-last approach, conventional aluminum metalization is replaced by tungsten metalization to allow the CMOS to withstand subsequent high-temperature processing during the micromechanical fabrication. This approach is a refinement of an approach originally developed at UC Berkeley. Specifically, the issues of yield, repeatability, and uniformity of the tungsten/CMOS approach are addressed. Also, material issues related to the development of high-temperature diffusion barriers, adhesion layers, and low-stress films are discussed. Processing and material issues associated with alternative approaches to this integration such as micromechanics-first, CMOS-last or the interleaved process are also discussed.

A new process is being developed for fabricating micromechanical structures using SOI material. This paper will review the process and show preliminary data for accelerometers made using the process. The focus of micromechanical technology at the C.S. Draper Laboratory, Inc. has been on the developement of inertial sensors, gyros, and accelerometers. Many of our current devices are fabricated using a dissolved wafer process. The resultant devices from that process are boron-doped silicon structures suspended over thin-film electrodes on a Pyrex substrate. It is a particularly attractive feature of this process that thick structures can be fabricated with small electrode gaps. Although this process has led to many excellent device results, requirements for future devices give reason to explore alternative technologies such as the SOI process, which yields an all-silicon device while preserving many of these advantages. Experimental devices were fabricated using a geometry similar to accelerometers being made by the previous process. Although not all of the geometric goals were met, the results are promising. Among the expected advantages for the new process are: a better thermal expansion match between device and substrate, the ability to add on-chip electronics, better alignment, and the capability of fabricating new types of structures.

This paper presents details of fabrication and performance testing of prototype microchannel heat exchangers. The microchannel heat exchangers are being developed for advanced cooling and climate control applications, and are designed for heat loads of 100 W/cm². Bulk and surface micromachining techniques are used to fabricate the test devices. Each heat exchanger section consists of over 150 microchannels etched in silicon substrates by either chemical etching or ion milling processes. The channels are 100-micrometers deep, 100-micrometers wide, and spaced 50- to 100-micrometers apart and connected with headers. Other heat exchangers have also been fabricated in copper and aluminum using machining and ion milling processes. Process steps involved photolithographic patterning, deposition of etch masks, ion or chemical etching, electrostatic bonding of the silicon to glass, insulator deposition, lamination of silicon to metals, application of thin heater coatings with busbars, and installation of the inlet/outlet hardware and valves. Recent hear exchangers have the silicon laminated to copper substrates. Performance testing focuses on determining the performance characteristics of the microchannel heat exchangers over a wide range of flow and heat transfer conditions. The working fluid for heat transfer is restricted to water or SUVA refrigerant HCFC-124 (R-124). Testing with water is run under single-phase conditions. The tests with R-124 are run under single-and two-phase flow conditions.

Prototypes of micromachined tunable infrared optical filters are being produced. Micromachining silicon for use in these filters requires the integration of multilayer dielectric optical coatings such as ZnSe/ThF₄. These coatings are novel materials for integration with microlithographic processing. Devices were engineered and a process flow was developed to avoid contaminating processing tools with the coating. A method for patterning the coating was developed. Low-temperature bonding techniques have been explored and tested. Fabrication issues for these micromachined devices are discussed.

Extensions of the German LIGA process have brought about fabrication capability suitable for cost effective production of precision engineered components. The process attributes a low fabrication of mechanical components which are not capable of being made via conventional subtractive machining methods. Two process improvements have been responsible for this extended capability which involve the areas of thick photoresist application and planarization via precision lapping. Application of low-stress x-ray photoresist has been achieved using room temperature solvent bonding of a preformed photoresist sheet. Precision diamond lapping and polishing has provided a flexible process for the planarization of a wide variety of electroplated metals in the presence of photoresist. Exposure results from the 2.5 GeV National Synchrotron Light Source storage ring at Brooklyn National Laboratory have shown that structural heights of several millimeter and above are possible. The process capabilites are also well suited for microactuator fabrication. Linear and rotational magnetic microactuator have been constructed which use coil winding technology with LIGA fabricated coil forms. Actuator output forces of 1 milliNewton have been obtained with power dissipation on the order of milliWatts. A rotational microdynamometer system which is capable of measuring torque-speed data is also discussed.

Micromolding is a key technology for the economic production of micro-components for microsystems. It is applied in several microstructuring techniques including the LIGA process which was invented and developed at Forschungszentrum Karlsruhe. Injection molding of multiple-use LIGA tool inserts produced by deep-etch x-ray lithography and electroforming allows the economic production of components for most applications using microsystems technology. Such microstructures are produced in small and large series and commercialized by Forschungszentrum Karlsruhe and the microParts Company, Dormund, Germany, cooperating within the framework of a license agreement. Special molding machines are applied for the production of single- or multi-stepped microstructures of a few micrometers in lateral dimension and structural details in the submicrometer range. Maximum aspect ratios of several ten up to 600 are achieved. In contrast to compact disc production, the machines are equipped with a special control unit, by means of which tool temperature is often kept above the melting temperatures of the plastics processed during injection. Evacuation of the tool cavity is required for the complete filling of the microstructurized nest area of the mold. Cycle time is mainly determined by the heating and cooling of the whole molding tool. Recently, novel techniques were developed for the production of ceramic LIGA or LIGA-similar microstructures at Forschungszentrum Karlsruhe, where further development of the LIGA technique has been performed for more than a decade. Using lost plastic microstructures and sometimes even metal tools, microstructures are made of structural (e.g., aluminum oxide, zirconium oxide) and functional ceramics (e.g., PZT). Current development activities are aimed at producing lost plastic molds for metal microstructures by injection molding. Molding tests with conductively filled thermoplastics have been carried out to manufacture lost molds for e.g. spin nozzles.

The LIGA process is being developed at Louisiana State University (LSU) to mass produce high aspect ratio microstructures. One component of the LIGA process is the manufacture of mold inserts. The progress associated with manufacturing an electroplated nickel mold insert is described in this paper. Unique features of our nickel electrodeposition procedures are discussed.

Ablation by excimer laser radiation can successfullly be used for microstructuring. However, the costs of production are still rather high due to the fact that laser processing is, in general, a serial process. Cost reduction is possible by combining excimer laser micromachining with a replication process. We call this process 'Laser-LIGA', which is a variant of the LIGA technique using ablation of polymers and photoresists by UV excimer lasers as the first structure-defining process step. Compared to deep x-ray lithography the surface quality of microstructures made by laser ablation is slightly poorer and the maximum aspect ratios (up to 10) are considerably lower too. On the other hand Laser-LIGA offers more flexibility. Almost any geometry can be realized, in many cases without a mask by direct structuration. This process is therefore suited to rapid prototyping as well as for large scale production of micro- scale devices. In this paper some applications of Laser-LIGA for different fields of microstructuring will be demonstrated including components of microfluidic systems, micro- optical devices different fields of microstructuring will be demonstrated including components of microfuidic systems, micro-optical devices and devices for medical applications. Such systems open up the potential for completely new developments at the micro scale.

A new technology called 3D UV-Microforming consisting of an advanced resist preparation process, UV lithographic steps, resist development, moulding procedures by electrodeposition, and finally stripping and cleaning for finishing the structures was developed for application in microsystem technology. It enables the low-cost fabrication of a wide variety of micro components for many different users. During resist preparation, layers up to two hundred pm thickness have been obtained until now. By using a standard UV mask aligner as an exposure tool followed by immersion development, thick resist layers up to 100 jim could be patterned in a single step on pre-processed silicon wafers. Repeated exposure and development were successfully used for structuring resist layers of up to 200 R thickness. High aspect ratios of more than 10 as well as steep edges of more than 88' could be fabricated. The resist patterns were moulded by using pulse or DC electroplating. For microsystem applications some metals and alloys were deposited. Three-dimensional micro components were fabricated as demonstrators for the new technique. It allows the use of materials with interesting properties which could not be provided by standard processes. Keywords: 3D UV-Microforming, electrochemical microfabrication, UV lithography, thick photoresist layers, high aspect ratio, steep edges, moulding by electrodeposition, sacrificial layers, 3D surface components, low-cost technology

Micromachining is often divided into two categories: bulk and surface micromachining. 'Bulk' micromachining generally refers to processes involving wet chemical etching of structures formed out of the silicon substrate and so is limited to fairly large, crude structures. 'Surface' micromachining allows intricate patterning of thin films of polysilicon and other materials to from essentially 2D layered parts (since the thickness of the parts is limited by the thickness of the deposited films). In addition to these two types of micromachining there is in fact a third type of micromachining as well, namely, 'mold' micromachining, in which the part is formed by filling a mold which was defined by photolithographic means. Historically micromachining molds have been formed in some sort of photopolymer, be it with x-ray lithography (LIGA) or more conventional UV lithography, with the aim of producing piece parts. Recently, however, several groups including ours at Sandia have independently come up with the idea of forming the mold for mechanical parts by etching into the silicon substrate itself. In Sandia's mold process, the mold is recessed into the substrate using a deep silicon trench etch, lines with a sacrificial or etch-stop layer, and then filled with any of a number of mechanical materials. The completed structures are not ejected from the mold to be used as piece parts, rather the mold is dissolved from around selected movable segments of the parts, leaving the parts anchored to the substrate. Since the mold is recessed into the substrate, the whole micromechanical structure can be formed, planarized, and integrated with standard silicon microelectronic circuits before the release etch. In addition, unlike surface-micromachined parts, the thickness of the molded parts is limited by the depth of the trench etch (typically 10- 50 micrometers ) rather than the thickness of deposited polysilicon (typically 2 micrometers ). The capability of fabricating thicker (and therefore much stiffer and more massive) parts is critical for motion-sensing structures involving large gimballed platforms, proof masses, etc. At the same time, the planarized mold technology enables the subsequent fabrication of features (for example flexible springs and flexures), much finer than those possible with bulk processes.

An interferometric system is described which can measure the dynamic performance of micromechanical structures. Modes of vibration and resonant frequencies are calculated for cantilevers and bridges which have been formed from boron-doped silicon. It is shown that a cantilever can be described by a beam model, whereas the bridge fits more appropriately to string behavior. This penomenon is explained by a cantilever, after annealing, being able to relieve the tension introduced at the doping stage. The fixed structure of the bridge is unable to accomodate this, however, and the doping-induced tension becomes the dominant factor in the structure's behavior.

This paper describes design techniques and structural components for building surface- micromachined polycrystalline silicon microelectromechanical systems (MEMS). The devices presented in this paper were fabricated thorugh the ARPA-sponsored multi-user MEMS process (MUMPS), but the ideas are applicable to other surface-micromachining polycrystalline silicon (polysilicon) processes. Specific devices are not discussed; instead, generic design and assembly techniques are presented to give novice or experienced MEMS designers new tools and ideas to improve their own designs. The topological effects which result from the fabrication of surface-micromachined polysilicon structures can be detrimental to many designs. However, these same effects can be turned into an advantage for the MEMS designer. Design techniques are presented for using conformal layer topologies to shape structures for guide rails, bossing, or to obtain closer tolerances than the design rules would seen to imply. Methods are also described for wiring hinged devices and large systems. These techniques were developed from the experience gained by the faculty and students at the Air Force Institute of Technology from the design of over 200 different devices and test structures on over 25 MUMPS die in the past two years.

High density plasmas are beginning to dominate the market for advanced anisotropic silicon etching for MEMS applications. This paper looks at the reasons behind this dominance for high etch rate, deep anisotropic etching. A discussion of anisotropic etch mechanisms highlights the need for sidewall passivation to meet these requirements. Results are presented of a novel room temperature advanced silicon etch process: >= 2 micrometers /min; >= 70:1 selectivity to resist (and >= 150:1 to oxide); up to 30:1 aspect ratio; 500 micrometers depth capability; using a non-toxic, non-corrosive environmentally acceptable fluorine-based chemistry.

Dry micromachining technology is developed for fabricating high aspect ratio Si structrues for microsensors. Two microsensor structures, including Si resonators and field emitters, will be presented in this ppaer. Released Si resonators up to 30 micrometers deep with 2 micrometers wide gap were fabricated. This is accomplished by a novel deep etch and shallow diffusion technique. High aspect ratio Si microstructures with vertical profile were first etched using an electron cyclotron resonance source, followed by a shallow B diffusion to fully convert the etched microstructures to p⁺⁺ layer. In addition, dry etching was used to form Si emitters with sharp tips and high packing density. Profile for Si emitters is controlled by erosion of the SiO₂ mask during dry etching. The ion flux and energy, controlled through coupled microwave and rf power, were used to obtain the desired etch rate and basewidth of the emitters. By increasing the pressure during etching, more vertical Si emitters were developed. Sharp emitter tips in Si with 2.2 micrometers basewidth and 11 micrometers height were fabricated and packing densities up to 1 X 10⁷ tip/cm² were achieved.

Deep silicon plasma etching provides an attractive alternative to conventional anisotropic wet etching for the fabrication of silicon micromechanical structures for sensors and actuators. In this paper an SF₆/O₂ plasma chemistry has been characterized using response surface methodology (RSM) and empirical models of the process responses (such as silicon etch rate and its uniformity, selectivity over the mask, and anisotropy) as a function of the input instrumental variables (such as RF power, system pressure, total gal flow rate, and oxygen content) are obtained. With the empirical models the process is optimized for deep silicon trench etching using a multi-objective optimization scheme. It is shown that RSM is an effective method for exploring the potential of plasma etching in this very challenging application and that deep silicon trenches (several tens of microns) can be obtained with conventional plasma etching machines if appropriate setting for the input variables are chosen as a result of process characterization and optimization.

Silicon microfabricated devices are being developed for reliable low cost sensors including accelerometers with tunnel junction readout. Using bonded silicon on insulator wafers as the starting material, novel structures are made by conventional surface micromachining followed by focused ion beam etching through 7 micrometers thick Si cantilevers at oblique angles to form submicron gaps to be closed by electrostatic actuation.

Excimer laser ablation provides the micromachining engineer with a unique tool for patterning, cutting, and structuring a wide variety of materials, including ceramics, glasses, and polymers. The short pulse (20 ns) ultra violet laser beam is used for nonthermal ablative material removal producing structures with a depth resolution of the order of 0.1 micrometers and spatial resolutions of the order of 1 micrometers or better. Careful control of laser dose (usually done using CNC systems) enables multi-level machining to be performed producing 3D microstructures which may be used directly, or as mold tools for laser-LIGA replication. This talk aims to illustrate both the possibilities, and limitations, of micormachining by excimer laser ablation, and will highlight some practical examples of structures and devices manufactured using this tool, many of which are currently in or near commercial production.

A novel approach for the fabrication of single-mode channel waveguides combined with focusing grating couplers by replication into polymer substrates is proposed and experimentally demonstrated. The concept is based on fabricating a master structure containing tall ridge patterns (about 3 micrometers high) combined with shallow, focusing grating structures (about 10 nm depth). By a micromolding technique using nickel shims, this pattern is hot embossed into a polymer substrate which is then full-area coated with a high-index dielectric waveguiding film. The focusing grating coupler is directly connected to the channel guide via a width-tapered waveguide section. An incident collimated beam can thus be coupled without the need for additional optics to the stripe waveguide formed by the film deposited on top of the ridge pattern. Results of experiments on stripe waveguides in quartz, focusing grating couplers in polycarbonate and combined channel waveguide and focusing structures in polycarbonate are presented and discussed. The feasibilty of the novel concept has been demonstrated by coupling a collimated free-space laser beam into a ridge waveguide on a replicated sample. The technology should find applications in integrated optical sensors and other low-cost integrated optical devices.

This paper presents a two step sacrifical layer etching technique used for the fabrication of surface mciromachined piezoresistive pressure sensors. The sacrificial layer itself is a sandwich structure of a thin polysilicon layer with the underetching channels and a much thicker 'buried' oxide underneath. First the polysilicon part is etched in an aqueous TMAH solution with high etch rates realizing a first shallow cavity. After rinsing, the oxide part is removed in 7:1 buffered HF. Since the oxide is etched now vertically, the process is completed within minutes. Sticking is suppressed successfully and non special drying techniques are required. The whole sensor structures could be passivated by LPCVD or PECVD layers against both etchants. Although the final depth of the cavity is 1 micrometers the sensor structure remains nearly flat. This minimizes technological problems concerning for example the piezoresistor definition or the sealing of the sensor and reduces the noise in the piezoresistor arrangement.

To develop a low-stress thin film for micromachined devices, a novel liquid-phase deposition (LPD) SiO_{2 - x}F_x technique utilizing silica-saturated H₂SiF₆ solution with H₂O addition only is proposed. Due to extremely low-temperature processing and fluorine incorporation, the stress of the LPD SiO_{2 - x}F_x film can be less than 100 MPa. In this paper, we found that the deposition parameter of H₂O addition has much efect on the stress of as-deposited LPD oxide. The stress variations with thermal cycling has also been clarified. We found that the LPD SiO_{2 - x}F_x film will be a good candidate as low-stress film for micromachined devices.

Piezoelectric thin films are very promising materials for MEMS applications because they have application flexibility and compatibility with semiconductor and micromachining processes. How to design MEMS devices with piezoelectric thin films, the mechanical characteristics, and how those characteristics can be controlled by process conditions is discussed in this paper. In addition, piezoelectric/electric characteristics must be understood. With this background, mechanical characteristics (Young's modulus and built-in stress) measurements of sputtered Pb(Zrx, Ti1 - x) O₃ thin film, one of piezoelectric materials, have been carried out using the load-deflection method. Relationships between post anneal conditions and those characteristics are discussed. It was shown from the experiment results that Young's modulus increases as anneal temperature/time increases. The maximum value was 76.6 GPa(700 degrees C/3600 sec) which is more than three times larger than that of as-depo film. Built-in stress is also affected by post anneal process and ranges from 0.04 GPa(as-depo) to 0.41 GPa(700 degrees C/60 sec). SEM observation results made it clear that it was caused by film shrinkage due to grain enlargement during anneal process.

Ultrasensitive accelerometers are needed by NASA for the measurement of orbital drag. We have designed an accelerometer capable of measuring 10^-8. In this paper, a method for fabricating a bulk micromachined accelerometer which incorporates a tunneling tip is presented. To meet sensitivity sepcifications, a weak spring and large mass are needed. However, these represent a delicate mechanism and a method of protection is provided by electrostatically clamping the proof mass in a fixed position. The effectiveness of the electrostatic clamp has been measured. It is found that clamping against an acceleration of 200 g is possible with voltages as low as 30 volts.

ZnO is a well known piezoelectric material. Unfortunately, it is not easy to deposit thin films onto silicon with a high resistivity by using common deposition technologies. The use of such films is therefore strictly limited to high frequency applications. The goal of our work was to find out a new deposition technology that allows the deposition of ZnO films with a high resistivity. Furthermore we were looking for the deposition of film thicknesses in a range up to 20 micrometers for SAW-sensor and microactuator applications. The deposition of the ZnO films was carried out in a programmable RF-magnetron-sputtering-system. We sputtered from a pure zinc target with a variable gas composition that consists of argon and oxygen. We worked in an alternating mode to achieve a high resistivity of the films. After a deposition cycle at a sample temperature of about 30 degrees C with a ramp shaped power the silicon- samples were cooled during the following cycle in the gas atmosphere. The deposition rate we measured was dependent from the gas composition and the applied power in a range between 1,5 micrometers /h and 2,2 micrometers /h. We deposited films of a thickness of 20 micrometers . Between two sputtered aluminium electrodes the films had a resistivity in a range between 2^*10¹⁰(Omega) cm and 2^*10¹¹(Omega) cm. The stress of the films could be influenced by the composition of the gases. The measured minimum stresses of the films were in a range of about 180 MPa. The films were also characterized by means of XRD- measurments. We found a weak < 101 > orientation of the layers perpendicular to the surface.

The quality factor of vacuum-operated single-crystal silicon microresonators was measured to identify their important sources of mechanical loss when medium-related loss is absent. The microresonators were torsional structures consisting of beams of width approximately 0.7 micrometers , and fabricated by reactive-ion-etching from the single-crystal silicon substrate. For torsional microresonators having Q's approximately 50,000, doping-impurity and support- related losses do not seem to be significant. Rather, the reactive-ion-etched surfaces of the microstructures appear to be the dominant source of mechanical loss. This etched-surface loss can be halved by thermal oxidation, resulting in microresonators with Q's consistently 80,000- 100,000. A model for this surface-related loss is presented.

Electroformed nickel resonators were constructed and tested in the temperature range of - 40 to 110 degrees C. The temperature sensitivities of the resonant frequencies are - 150 ppm/degrees C, - 200 ppm/degrees C, and - 3000 ppm/degrees C for cantilever beams, ring structures, and clamped-clamped bridges, respectively. The built-in stress for the bridge was estimated to be approximately 2 X 10⁹ dyne/cm². No resonant frequency shift was detected after long-term (over 60 days), large amplitude vibration. This implies that the electroformed nickel is a viable material for the construction of resonant mechanical sensors.

Chemical vapor deposited diamond films have been grown successfully for a number of years, and through advances in processing technology, tooling investments, and volume, growth processes have become affordable for many commercial purposes. The remarkable properties of diamond, including hardness, thermal conductivity and optical properties, make it an ideal choice of material in many applications. A key to affordability of diamond products for commercial purposes is the ability to machine the films into desirable shapes. Because of diamond's hardness, photons are an obvious choice of 'cutting medium'. This paper will discuss the available laser sources and in what circumstances they are being used to achieve desired results. Focus will then be placed on a new Nd:YLF laser based system which was built specifically to machine CVD diamond at high speed and produce low taper angle features with UV excimer laser-like quality.

Whether they are called Micro-Electro-Mechanical-Systems (MEMS)', micromechanical systems, micro machines or MicroSystems Technology (MST), these manufacturing based technologies are enabling the development and production of many exciting new products. MEMS based.manufacturing technologies are the fertile ground for discontinuous innovations which are revitalizing existing industries and creating new markets. The total market for products produced through the use ofMEMS technologies is currently less than $1 billion. Projections ofthe MEMS based marketplace vary from $8 billion 2 to $14 billion3 by the year 2000, as seen by industry analysts. Initial results from our Delphi industrial forecast suggest that the market will be in the $3 to $54 billion range by the turn ofthe century. Whether the market will be $4 or $14 billion by the end of the decade is debatable, but what is not debatable is that the marketplace is exploding, both in terms of revenue and product applications. This explosive growth will severely tax an infrastructure which barely supports the current level of activity. The emerging MEMS based product marketplace is based on a series oftechnological and business innovations. These types ofinnovations have been called radical (Walsh and Lynn 1991), architectural or revolutionary (Abernathy and Clark 1985)6and now more commonly discontinuous (Morone 1993). The role of infrastructure in the development of these emerging MEMS markets is critical. In order to understand the demands placed on the infrastructure by an emerging discontinuous innovation, it is essential to understand the discontinuous innovation process as well as the difference between continuous and discontinuous innovation. In particular, we focus on how discontinuous innovation has influenced the existing technological infrastructure and how the slowly changing infrastructure influences the established state of technology. This feed-back feed-forward loop dramatically influences the growth of new markets and the time required for new technologies and novel products to have an impact on markets. In our study we introduce a model of infrastructure development. This simple three stage model is tailored to the MEMS market and was developed by combining historical information with current research on the infrastructure for MEMS based markets.

This presentation reviews aspects relevant to anisotropic very deep plasma etching of silicon. Plasma etching of silicon to depths in excess of 500 (mu) at rates above 4 (mu) /min. allow for new self-releasing or unidirectional flexure structures. It will begin by covering a brief comparison of anisotropic plasma etching with some other alternative very deep etching processes. The impact of Alcatel's product offering ain attaining this etch technology is also reviewed, as well as some of the interdependencies in the etch process. Then the different anisotropic etch regimes will be discussed along with the characteristics and sample applications in each regime.

A new method is presented to fabricate out-of-plane microstructures using traditional planar micromachining technology. Composite LPCVD polysilicon/silicon nitride beams are fabricated to study this concept. Polysilicon films ranging from 0.5 micrometers to 1.3 micrometers , and silicon nitride films ranging from 150 to 450 nm, were used to fabricate various thickness ratios of composite out-of-plane microstructures. Upon release, these planar structures take on 3D shapes, due to the bending moment caused by inherit internal stresses in the thin films. These stress engineered 3D microstructures (SEMS) open the path to novel microstructures. This paper presents a design theory for SEMS, describes the fabrication process, and discusses the results of initial experiments.

Projection displays and microelectromechanical systems (MEMS) have evolved independently, occasionally crossing paths as early as the 1950s. But the commercially viable use of MEMS for projection displays has been illusive until the recent invention of Texas Instruments Digital Light Processing ^TM (DLP) technology. DLP technology is based on the Digital Micromirror Device^TM (DMD) microchip, a MEMS technology that is a semiconductor digital light switch that precisely controls a light source for projection display and hardcopy applications. DLP technology provides a unique business opportunity because of the timely convergence of market needs and technology advances. The world is rapidly moving to an all- digital communications and entertainment infrastructure. In the near future, most of the technologies necessary for this infrastrucutre will be available at the right performance and price levels. This will make commercially viable an all-digital chain (capture, compression, transmission, reception decompression, hearing, and viewing). Unfortunately, the digital images received today must be translated into analog signals for viewing on today's televisions. Digital video is the final link in the all-digital infrastructure and DLP technoogy provides that link. DLP technology is an enabler for digital, high-resolution, color projection displays that have high contrast, are bright, seamless, and have the accuracy of color and grayscale that can be achieved only by digital control. This paper contains an introduction to DMD and DLP technology, including the historical context from which to view their developemnt. The architecture, projection operation, and fabrication are presented. Finally, the paper includes an update about current DMD business opportunities in projection displays and hardcopy.

The rapid expansion in number and scope of research projects in the general area of micromachining technology makes the field especially interesting. Advances, particularly from universities and research institutes, suggest that increased commercial development is likely in a number of new fields. The nature of the fabrication technologies used to make these prototype parts, however, lead to difficulties in quickly reaping commercial benefits from these technolgoical advancements. This is in contrast to the nature of commercial development in the integrated circuit industry, where standardized processes result in rapid development of systems and products which use new designs.