Single-crystal silicon films of sub-micrometer thicknesses were tensile tested using a thin film tensile tester with an electrostatic force grip. The tester has been newly designed for this thin film. It uses a servo-controlled balance for small-force measurement with a resolution of less than 100 νN. The specimens were fabricated using bulk- and surface-micromachining starting with a SIMOX wafer, and an epi-SOI wafer. The bulk micromachined specimens were released from a wafer by anisotropic etching from the backside of the wafer. The specimen size was 0 to 500 νm long, 20 or 50 νm wide and 0.14 νm thick. The thinner specimens were fabricated by surface micromachining and dry release process using XeF2 and vapor HF etching to avoid stiction of the specimens. The size was 0 to 600 νm long, 1 to 20 νm wide and 0.05 νm thick. These wafers were (100) oriented and the loading axis of the specimen was <110>. The tensile strength of the bulk micromachined specimens ranged from 1.4 to 4.8 GPa for twelve samples and that of the surface micromachined specimens ranged from 1.0 to 6.8 GPa for seven samples.

A systematic study of small-scale size effects has been conducted on elliptically perforated specimens with minimum radius of curvature of 1 micron. This study aimed at assessing the dependence of failure stress at the tip of a notch on varying: (a) stress concentration for constant radius of curvature, (b) radius of curvature of micro-notches relative to the material grain size and constant stress concentration. The experiments demonstrate a strong influence of notch radius on the failure strength of MEMS scale specimens, while the effect of the stress concentration factor is of rather secondary importance. The local failure strength at the tip of a notch increases when the radius of curvature becomes smaller, which is in accordance with the probabilistic nature of failure. When the notch radius becomes as small as 1 micron (only three times larger than the grain size) then a strong size effect is observed. This effect becomes moderate for larger radii of curvature, up to 8 microns (25 times the grain size), when the failure stress at the notch tip almost reaches the tensile strength recorded for 50 micron wide samples.

A class of MEMS devices, which utilizes microfabrication technology and bulk piezoelectric material, is currently being developed to produce high power density transducers. A thin-film gold-tin eutectic solder bond has been developed to bond electrically and mechanically bulk piezoelectric elements to microfabricated silicon structures in these devices. A 4.3 [.proportional]m thick multilayer film structure, consisting of a titanium adhesion layer, a platinum diffusion barrier, a gold-tin (80 wt.% Au - 20 wt.% Sn) alloy layer, and a pure gold capping layer, was sputter deposited on the piezoelectric components to be bonded. Bonding was accomplished by mating the piezoelectric components with silicon components metallized with a titanium-platinum-gold multilayer film and heating to approximately 300°C in a reducing atmosphere. The bonding technology allows thin, electrically conductive bonds to be formed between dissimilar materials with minimal amounts of applied pressure during bonding. Successful bonding has been achieved between single crystal silicon and polycrystalline lead-zirconate-titanate (PZT-5H) as well as between silicon and single crystal lead zinc niobate-lead titanate (PZN-PT). The process was optimized to produce mechanically robust, void-free bonds. The absence of voids was verified through scanning electron microscope examinations of bond cross-sections. Tensile tests conducted on representative structures indicated that the strength of the bond was limited by the strength of the titanium – PZT-5H interface.

In MEMS, the geometry and applied loads between strength testing coupon and structures for design are usually different and the strength data cannot be directly applied without modification for designing brittle MEMS structures. In this paper, a method based on equal failure probability is proposed to find the equivalent strength for design by scaling the test strength with a parameter called stress ratio. By the same approach, it is also possible to correlate MEMS material strength data obtained by different testing schemes.

A three dimensional thermal imaging system is being developed for measuring temperature profiles in MEMS-biomedical devices. These devices rely on a thermal microablation of the dead-skin layer in order to sample transdermal fluids. This is accomplished using microheaters embedded into a PDMS microchannel device. In order to determine the proper functioning as well as long-term safety of the devices, a temperature profile of the device and the skin in contact with the heaters is needed. The results of simple analytical models are used to optimize a proto- type device. Using a three-dimensional chemical imaging microscope and temperature-depend- ent fluorophores, the temperature profile in a sample can be determined quantitatively as well. We demonstrate the technique on a model sample, and discuss extension to other applications such as thermal imaging in biological systems.

Just as the 40+ years of technology developments associated with the electronic application of semiconductor fabrication processes is “morphing” into a micro-electro- mechanical systems (MEMS) technology in the past dozen years or so, so it seems may the “mature” multilayer ceramic fabrication technology associated with capacitor components and interconnect substrates for the integrated circuit industry, be morphed into MEMS – microsystems technology applications. This paper highlights work underway in Motorola Labs aimed at exploring the potential to develop 3D multilayer ceramic structures to integrate (monolithic and hybrid) multiple functions to create microsystems for wireless, energy and life science applications. By multiple functions, we refer to the ability for a microsystem to perform electronic, fluidic, thermonic, photonic and mechatronic (or actuator) based functions. Current capabilities of the multilayer ceramic materials and processes to achieve integrated functionalities for wireless applications will be described including the development of a new dielectric enabling increased performance for wireless applications. Also to be highlighted will be exploratory microscale fuel cell prototypes exploiting advances in the multilayer ceramic lamination and feature forming technologies enabling the insertion of 3D microchannels for microfluidic functions. These prototypes also feature the ability of the technology to provide thermonic functionality for microreactor devices. Feasibility of a light source that can be integrated into the technology platform hinting at photonic applications will be described. Many materials science and engineering advancements are needed to achieve the potential of this “old” but newly “morphing” technology and some of these will be noted.

Mechanical prototypes of a silicon interposer device, having high-aspect ratio straight walls, were required for experiments to qualify various packaging assembly options. The ideal solution would yield expendable, quickly made, and relatively inexpensive replica parts.

This need led to the investigation of micromachining of silicon via laser ablation. We sought to optimize the results by varying several parameters, the most influential of which was shown to be laser wavelength. Silicon wafers of thickness 500 νm were machined using laser wavelengths of 10,600 nm, 1064 nm, and 355 nm. As was predicted according to the theoretical silicon absorption profile, the best results were obtained using 355 nm wavelength light from an ultraviolet laser source. Several post-machining cleaning routines, to remove melted silicon debris, were evaluated as well. Pictorial results of both laser processing and cleaning are shown and discussed.

ICP was concurrently investigated as a viable alternative to laser machining. These cuts were markedly cleaner and straighter than those induced by laser; additionally, cut features were more well defined. Pictorial results are compared and contrasted with those of laser ablation.

Cost and time are traded for quality when choosing between laser ablation and ICP etch. Therefore, process choice should be application-dependent. For wafers with moderately complex cutout features, laser micromachining is significantly lower in cost and manufacturing time. However, the laser process does produce residual thermal stress surrounding the cut region, machined edges are relatively rough, and fine features are more difficult to produce. Laser ablation is an attractive manufacturing option for quick-turnaround prototyping as well as high-volume production; ICP is better suited to producing high-definition parts without potential for thermal stress damage.

This paper reviews bulk micromachining of Si, with particular reference to the microassembly and packaging of optical and electronic components. Taking as an example the use of 2.3 [.proportional]m thick silicon nitride microclips to hold an optic fibre in place in a silicon V-groove, the potential of thin film packaging is briefly assessed. The fabrication process is to: [i] deposit a controlled stress silicon nitride layer onto a blank (100) oriented silicon wafer by chemical vapour deposition (standard LPCVD but silicon rich); [ii] pattern the silicon nitride clip mask shape by optical lithography and reactive ion etching; and [iii] remove the underlying silicon to form a V-shaped groove with an anisotropic liquid etch process using the silicon nitride as a mask. From the measured deflection of a silicon nitride beam when a force is applied, the Young's modulus is 400 GPa. The beams are rugged and a clipping force of 10 N per metre of optic fibre length is achieved.

This approach to applying the mechanical properties of thin films is compatible with standard thin film technology, enables precise microassembly, and has great potential to reduce manufacturing costs in a wide range of sensors and microsystems. In some applications high temperature processing must be avoided, even early on in the manufacturing process. Using low temperature processing, robust clips from silicon nitride made by plasma enhanced chemical vapour deposition (PECVD and hydrogen rich) are made, but the Young's modulus of this material is typically 40 GPa and the clipping force is correspondingly small. The paper concludes by assessing the prospects for other thin film materials in mechanical MEMS and packaging.

Electro-plated gold films are used extensively in packaging of MEM sensors to make connections to signal conditioning electronics. Over the past two years, production lots of gold plated substrates, procured from various vendors, failed an accelerated aging qualification test. In this test, 0.0025 [mm] diameter aluminum wires were ultrasonically welded to the film, aged at 120°C for 48 hours, and then pulled to destruction. The criterion for passing this test was that the wires should break both before and after aging. In the defective lots, the wires lifted off of the gold film after aging. Analysis of these defective films by SEM, Auger, and TOF-SIMS suggested that residues, deposited from the plating bath, concentrated beneath the bond as the gold and aluminum reacted to form an inter-metallic compound during aging. A combination, etch and cleaning treatment was developed for defective substrates, which removed a sufficient amount of residues from the gold to pass the qualification test.

A MEMS-based gas turbine engine is being developed for use as a button-sized portable power generator or micro-aircraft propulsion source. Power densities expected for the micro- engine require high combustor exit temperatures (1300-1700K) and very high rotor peripheral speeds (300-600m/s). These harsh operating conditions induce high stress levels in the engine structure, and thus require refractory materials with high strength. Silicon carbide has been chosen as the most promising material for use in the near future due to its high strength and chemical inertness at elevated temperatures. However, techniques for microfabricating single- crystal silicon carbide to the level of high precision needed for the micro-engine are not currently available. To circumvent this limitation and to take advantage of the well-established precise silicon microfabrication technologies, silicon-silicon carbide (SiC) hybrid turbine structures are being developed using chemical vapor deposition of poly-SiC on silicon wafers and wafer bonding processes. Residual stress control of SiC coatings is of critical importance to all the silicon-silicon carbide hybrid structure fabrication steps since a high level of residual stresses causes wafer cracking during the planarization, as well as excessive wafer bow, which is detrimental to the subsequent planarization and bonding processes. The origins of the residual stresses in CVD SiC layers have been studied. SiC layers (as thick as 30µm) with low residual stresses (on the order of several tens of MPa) have been produced by controlling CVD process parameters such as temperature and gas ratio. Wafer-level SiC planarization has been accomplished by mechanical polishing using diamond grit and bonding processes are currently under development using interlayer materials such as silicon dioxide or poly-silicon. These process development efforts will be reviewed in the context of the overall micro-engine development program.

2-νm thick structural films of polycrystalline silicon are shown to display “metal-like” stress- life fatigue behavior in room air, with failures occurring after >1011 cycles at stresses as low as half the fracture strength. Using in situ measurements of the specimen compliance and transmission electron microscopy to characterize such damage, the mechanism of thin-film silicon fatigue is deduced to be sequential oxidation and moisture-assisted cracking in the native SiO2 layer. This mechanism can also occur in bulk silicon but it is only relevant in thin films where the critical crack size for catastrophic failure can be exceeded within the oxide layer. The fatigue susceptibility of thin-film silicon is shown to be suppressed by alkene-based self- assembled monolayer coatings that prevent the formation of the native oxide.

A sensor chip has been designed and tested that uses a MEMS strip heater as both source and detector of infrared radiation. An optical cavity reflects infrared radiation back onto the source filament. Changes in reflected light intensity modify heater temperature, and the measured signal is a change in resistance. The effects of processing on electrical and thermal isolation were characterized and used to evaluate device performance. Thermally isolated, uniformly heated emitters are achieved using a backside release etch process. The fully released devices demonstrated superior electric to thermal-optical conversion, with the requisite narrow band emission for CO2 detection. Using these sensor-chips, CO2 detection was demonstrated, with projected sensitivities ≤0.1%.

This paper reports results on Heusler-type NiMnGa and CoNiGa ferromagnetic shape memory alloy films deposited by magnetron sputtering and pulsed laser deposition respectively, on silicon cantilever-type substrates.

NiMnGa FMSMA films deposited by magnetron sputtering on silicon substrates show good ferromagnetic and shape memory properties if the substrate is heated above 350°C or if the films deposited at RT are annealed above 600°C. The growth of the film is columnar. TEM examination shows typical martensitic variants while the SAED pattern suggests an orthorhombic structure. CoNiGa FMSMA films deposited at 550°C by PLD are crystalline. Mechanical spectroscopy data proves the presence of a typical martensitic transformation that develops below room temperature: the phase transitions manifest themselves by typical changes of the modulus defect and displacement vs. temperature characteristics similar to those associated with the martensitic transformation in known shape memory alloy films.

The change in the elastic modulus as a function of the magnetic field suggests that in both NiMnGa and CoNiGa films the spontaneous magnetization is clamped by the magnetic anisotropy field.

A piezoelectric thin film MEMS device for generating power from a novel heat engine which approaches a Carnot cycle has been developed. The structure of the underlying electrode and PZT thin film generator has been optimized for increased adhesion. Atomic force microscopy was used to track electrode grain size and roughness; generating grain sizes of approximately 100 and 200 nm in diameter and a roughness of about 14-20 nm provide substantial improvements in film adhesion over systems with smaller grains and smoother surfaces. This has led to the ability to operate the engine at frequencies between 10 and 1500 Hz. The system of interest (a fluid filled cavity sealed by a micromachined silicon membrane and the PZT film) shows increased deflections for a given pressure applied to the membrane at frequencies where the system resonates. By operating the system dynamically, it is possible to generate more than 2 V from a single generator structure.

The fracture strengths of 3.5 microns thick and 20 or 50 microns wide polysilicon specimens were measured. One set of specimens was straight in the gage section, another set had a central hole 5.0 microns in diameter, and the third set had symmetric semicircular notches 2.5 microns in radius on each side. The local maximum fracture stresses of the non- uniform specimens, as calculated from the remote fracture stress with a stress concentration factor, were higher than measured in the straight specimens. This indicates a size effect, which is presumably due to the smaller highly stressed volume or area in the non-uniform specimens.

Vickers indentation tests were carried out at room temperature with a microhardness tester on ZnO thin films deposited on (100) Si wafers. The films with thicknesses ranging from 0.202 to 1.535 νm were tested under indentation loads varying from 50 mN to 2.0 N. The experiments exhibited a film thickness dependent deformation behavior. When the film thickness is smaller than a critical value, only indentation-induced radical cracking occurred and for a given load, the radical crack size decreased with film thickness. On the other hand, indentation-induced delamination was observed if the film thickness was larger than the critical value. In this case, the radial crack, if induced, had a size smaller than the size of delamination or even smaller than the size of indentation impression. Under a same indentation load, the delamination size increased with the film thickness.

We report the fabrication of complex 3D microstructures in silica and polymethyl- methacrylate glass by a combination of femtosecond laser microfabrication and chemical wet etching techniques. It is demonstrated that fabrication of interconnected network of channels having lengths of about 200 νm, and diameters as small as 10 νm is possible due to the enhanced etching selectivity (typically 20 - 60) in the laser-irradiated regions. Thus, it becomes feasible to form 3D micro-fluidic and photonic crystal structures in transparent glass-like materials using this approach. In addition, preliminary results on microstructuring of rubber are presented.

The plasticity length scale parameters associated with rotational and stretch gradients in LIGA nickel MEMS structures plated from sulfamate bath are presented in this paper. These are obtained from micro-bend and micro-tensile experiments on LIGA nickel films with different thicknesses. The composite length-scale parameter, lc is measured to be ∼ 4.8 νm for as-plated films (25, 50, 100 and 175 νm thick). The implications of the results are then discussed for the modeling of plasticity in LIGA nickel structures.

We present the results of molecular dynamics simulations of pairs of alkylsilane monolayers on silica surfaces under shear. In particular, we investigate the effects of shear velocity on the friction for chains of 6, 8, 12, and 18 carbon atoms covalently bonded to a crystalline surface. Our studies are performed at loads close to 0.2 and 2 GPa for relative velocities of 0.2, 2.0, and 20.0 m/s. We find that for perfect (defect-free) monolayers, the effects of chain length and velocity are weak, indicating that the experimentally measured dependence of friction on these properties is primarily due to defects in the monolayer. We have investigated possible finite size effects by varying our system dimensions from 43 Å ×50 Å Å to 174 Å × 201 Å. We find that increasing the surface area by a factor of N reduces the noise in the shear stress by a factor of , and has a comparable effect to averaging the smaller system data over bins of points. This indicates that finite size effects are negligible in our simulations.

A 3D Feed horn shape MEMS antenna has some attractive features for array application, which can be used to improve microbolometer performance. Since MEMS technology have been faced many difficulties to fabrication of 3D feed horn shape MEMS antenna array itself. The purpose of this paper is to propose a new fabrication method to realize a 3D feed horn shape MEMS antenna array using a MRPBI(Mirror Reflected Parallel Beam Illuminator) system with an ultra-slow-rotated and inclined x-y-z stage. A high-aspect-ratio 300 [.proportional]m sidewalls had been fabricated using SU-8 negative photo resist. It can be demonstrated to feasibility of realize 3D feed horn shape MEMS antenna array fabrication. In order to study the effect of this novel technique, the 3D feed horn shape MEMS antenna array had been simulated with HFSS(High Frequency Structure Simulator) tools and then compared with traditional 3D theoretical antenna models. As a result, it seems possible to use a 3D feed horn shape MEMS antenna at the tera hertz band to improve microbolometer performance and optical MEMS device fabrication.