We characterize in-situ the adhesion of surface micromachined polysilicon beams subject to controlled humidity ambients. Beams were freed by supercritical CO2drying. Consistent adhesion results were obtained using a post-treatment in an oxygen plasma which rendered the microbeams uniformly hydrophilic. Individual beam deformations were measured by optical interferometry after equilibration at a given relative humidity (RH). Validation of each adhesion measurement was accomplished by comparing the deformations with elasticity theory. The data indicates that adhesion increases exponentially with RH from 30% to 95%, with values from 1 mJ/m2 to 50 mJ/m2. Using the Kelvin equation, we show that the data should be independent of RH if a smooth interface is considered. By modeling a rough interface consistent with atomic force microscopy (AFM) data, the exponential trend is satisfactorily explained.

The development of polysilicon fracture mechanics specimens with characteristic dimensions comparable to those of typical microelectromechanical systems (MEMS) devices is presented. The notched cantilever specimens are fully integrated with a simultaneously microfabricated electrostatic actuator, which allows on-chip testing of the specimens without the need of an external loading device, and without any possible influences from external sources. Under monotonic loading, the average maximum tensile stress (strength) and average nominal fracture toughness were measured as 4.2 GPa and 3.5 MPa-m½ for boron-doped specimens, and 5.0 GPa and 4.0 MPa-m½ for undoped specimens. An average modulus of rupture of 3.3 GPa and average nominal toughness of 2.7 MPa-m½ were measured for specimens cracked under cyclic resonance loading. The differences between the monotonic loading and cyclic loading data are attributed to fatigue initiation of a sharp crack from the 1 ýtm radius notch. The experimental data is consistent with a critical flaw size in the fabricated devices, a, that is related to the fracture toughness Klc by Klc/a1/2=4600 MPa.

Highly hydrophobic fluorocarbon films were prepared by the vapor phase (VP) deposition method in a vacuum chamber using both liquid (3M's FC40, FC722) and solid sources (perfluorodecanoic acid (CF3(CF2)8COOH), perfluorododecane (C12F26)) on Al, Si and oxide coated wafers. The highest static contact angles of water were measured on films deposited on aluminum substrate. But relatively lower contact angles were obtained on the films on Si and oxide wafers. The advancing and receding contact angle analysis using a captive drop method showed a large contact angle hysteresis (ΔH) on the VP deposited fluorocarbon films. AFM study showed poor film coverage on the surface with large hysteresis. FTIR-ATR analysis positively revealed the stretching band of CF2 groups on the VP deposited substrates. The thermal stability of films was measured at 150°C in air and nitrogen atmospheres as a function of time. The rapid decrease of contact angles was observed on VP deposited FC and PFDA films in air. However, no decrease of contact angle on them was observed in N2.

The formation of high aspect ratio Micro-Electro-Mechanical Structures (HI-MEMS) requires good adhesion of the thick microstructures to the substrate. The interfacial bond strength between PMMA and various metal substrates (Ti, Cu, Au, and Ni), in the preliminary stages of the LIGA process, was evaluated by shear stress measurement. Adhesion promotion processes have been investigated such as texturing of the surface of Ti and Cu substrates by chemical oxidation or thermal processing of the PMMA sheet, and use of a positive resist as an interfacial adhesion promoter. Several fold increases in bond strength were obtained, the largest increase associated with a combination of substrate roughening, thermal treatment and use of adhesion promoters.

This study reports in-situ observations of the buckling evolution of microelectromechanical structures during etching of their underneath sacrificial layers. As the etching went on, the buckling pattern evolved from mode I, the sinusoidal half-waves, to mode II, the constrained sinusoidal half-waves, to mode III, the conventional mode, and finally to mode IV, the blister- like local buckling. Closed formulae were derived from theoretical analysis, and the experimental results agreed well with the theoretical ones.

In high temperature processing of microelectronics such as rapid thermal processing (RTP), an accurate measurement of temperature is critical in fabricating defect-free devices. Currently, temperatures during such processes are measured using thermocouples or pyrometers. However, accurate measurements with thermocouples are troublesome due to the difficulties in holding the thermocouples in place and maintaining good thermal contact. A further drawback for thermocouples is that they are highly intrusive. For pyrometers, optical interference effects and partial transparency limit their applicability and the local temperature of wafers near the electronic devices are difficult to measure using pyrometer due to relatively large spot size. Microelectromechanical systems (MEMS) technology is being used in developing innovative temperature sensor (T-MEMS), which allow an ex-situ examination of the maximum temperature reached during a thermal process by creating a permanent change in structure at high temperatures. The performance of the device relies on the thermophysical properties of the materials; specifically, the Young's modulus, thermal expansion coefficient, and the ultimate strength must be considered for silicon, and silicon nitride. Through careful experimental design and accurate modeling of their structural behavior, the high-temperature material properties can be determined using the T-MEMS and calibrated furnaces.

The silicon-framed tensile specimen design has been used in tensile tests of thin films of a variety of metals and epitaxial silicon. A piezo-actuated microtensile test device holds the specimen while the silicon frame is cut just before testing, imposes the tensile displacement, and provides voltages proportional to force and displacement. This technique is appropriate for films that are hundreds of micrometers long, tens to a few hundred micrometers wide, and from 0.3 to 15 micrometers thick. The specimen film must be amenable to lithographic patterning, must adhere well to the silicon substrate, and must resist a silicon etchant. The specimen fabrication is a bulk-micromachining process, because the silicon substrate is etched through underneath the specimen film. Uniaxial yield strength, ultimate tensile strength, and elongation to maximum load can be measured using the microtensile tester. The addition of laser illumination and digital photography allows implementation of electronic speckle pattern interferometry, for accurate measurement of local displacement. This addition allows evaluation of the tensile Young's modulus. Compared to bulk material, thin films of copper and aluminum have lower apparent Young's moduli, higher yield and ultimate tensile strengths because of their fine grain size, and lower elongation to failure. Correlation between properties measured by indentation and by tensile testing needs further study.

High aspect ratio microstructures (HARMs) with a height of hundreds of micrometers and a width of a few tens of micrometers present high promise in a number of challenging fields. At LSU, a number of applications are being developed in which nickel HARMs are electroplated on metal surfaces (mold insert fabrication for the LIGA process, HARMs on mechanical seal faces, HARMs on heat exchange surfaces, etc.). In some of these applications, the HARMs are subjected to high stresses and the mechanical properties are particularly important. These properties can be used to adjust processing parameters to optimize properties of the HARMS.

This paper presents a method for measuring the strength properties of cantilevered nickel HARMs constructed by LIGA. Experimentally measured values are reported for modulus of rupture (1280 MPa), Young's modulus (153 GPa) and Knoop hardness (500 Hk) for HARMs with an overplated base. SEM micrographs clearly indicate that failure of the beams is brittle and most frequently occurs at the interface of the beam and overplated base.

A newly developed approach to thin-film tensile testing has been utilized for evaluating ultra-thin-films of about 60-100 nm in thickness. Thin sputtered metal films deposited on 3.5 gm thick polyimide film substrates were tested using a Mesotronix nanotensiometer. Modifications to sample handling and zero strain calibration helped to improve testing reproducibility. The response of a thin-film metal was determined by subtracting the response of the polyimide carrier film from that of the metal-on-polyimide composite. Stress-strain curves for two amorphous alloys sputtered from Al-25 at% Ti and Nb-25 at% Al targets exhibited well-defined elastic regions followed by brittle behavior. The ultimate tensile strengths (UTS) were about 1200 MPa for the Al-Ti and 1700 MPa for the Nb-Al, both of which occurred near the elastic strain limit of about 2%. After recrystallization at 280°C for an hour, the Al-Ti film showed what appears to be an upper and lower yield point, in addition to strain hardening. Initial yielding occurred at a strain of about 1.2% and a stress of about 800 MPa, while the UTS of about 1000-1060 MPa occurs at strains as high as 9%. Stress relaxation measurements performed on both the amorphous and the recrystallized films showed lower values of stress relaxation at 1% strain compared to low alloy polycrystalline Al films.

A table top servohydraulic load frame equipped with a laser displacement measurement system was constructed for the mechanical characterization of LIGA fabricated electroforms. A drop-in tensile specimen geometry which includes a pattern to identify gauge length via laser scanning has proven to provide a convenient means to monitor and characterize mechanical property variations arising during processing. In addition to tensile properties, hardness and metallurgical data were obtained for nickel deposit specimens of current density varying between 20 and 80 mA/cm2 from a sulfamate based bath. Data from 80/20 nickel/iron deposits is also presented for comparison. As expected, substantial mechanical property differences from bulk metal properties are observed as well as a dependence of material strength on current density which is supported by grain size variation. While elastic modulus values of the nickel electrodeposit are near 160 GPa, yield stress values vary by over 60%. A strong orientation in the metal electrodeposits as well as variations in nucleating and growth morphology present a concern for anisotropic and geometry dependent mechanical properties within and between different LIGA components.

Techniques and procedures are described for tensile testing of polysilicon specimens that are 1.5 or 3.5 νm thick and have various widths and lengths. The specimens are fixed to the wafer at one end and have a large free end that can be gripped by electrostatic forces. This enables easy handling and testing and permits the deposition of 18 specimens on a one-centimeter square portion of a wafer. The displacement of the free end is monitored, which allows one to extract Young's modulus from the force-displacement record. Some of the wider specimens have two gold lines applied so that strain can be measured interferometrically directly on the specimen to record a stress-strain curve.

The specimens were produced at the Microelectronics Center of North Carolina (MCNC). When compared with earlier results of wider MCNC specimens that were 3.5 μm thick, the Young's modulus is smaller and the strength is slightly larger.

A structure has been developed to measure the fracture strength of thin films that allows for a large amount of statistical data to be generated from a single die. The strength of a material is determined by measuring the displacement at fracture of cantilever beams bending in-plane. This device has been used to measure the fracture strain of three different polysilicon materials: polysilicon from two different MCNC MUMPS runs and a polysilicon fabricated at the Microfabrication Laboratory at the University of California at Berkeley. The strain at fracture of the MUMPS polysilicon was measured to be 1.5% ± 0.2% for both MCNC MUMPS runs. The Berkeley polysilicon exhibited a fracture strength of 1.2% ± 0.1%, a difference from the MUMPS result that is statistically significant. Also, no statistical difference in fracture strength was measured on specimens that were super-critically carbon dioxide dried as opposed to furnace dried. In a separate experiment, the Young's modulus of the MCNC MUMPS polysilicon was found to be 173 ± 20 GPa.

Elastic modulus and Poisson's ratio of diamond-like carbon (DLC) film was measured by a simple method using DLC bridges which are free from mechanical constraint of substrate. The DLC films were deposited on Si wafer by C6H6r.f. glow discharge at the deposition pressure 1.33 Pa. Because of the high residual compressive stress of the film, the bridge exhibited a sinusoidal displacement by removing the constraint of the substrate. By measuring the amplitude with known bridge length, we could determine the strain of the film required to adhere to the substrate. Combined with independent stress measurement by laser reflection method, this method allows calculation of the biaxial elastic modulus, E/(1–v), where E is the elastic modulus and v Poisson's ratio of the DLC film. By comparing the biaxial elastic modulus with plane-strain modulus, E/(1–v 2), measured by nano-indentation, we could further determine the elastic modulus and Poisson's ratio, independently. The elastic modulus, E, increased from 87 to 133 GPa as the negative bias voltage increased from 400 to 550 V. Poisson's ratio was estimated to be about 0.20 in this bias voltage range. For the negative bias voltages less than 400 V, however, the present method resulted in negative Poisson's ratio which is physically impossible. The limitation of the present method was also discussed.

In this paper, amorphous, semi-crystalline and polycrystalline silicon films have been deposited by low pressure chemical vapour deposition (LPCVD) from silane SiH4 by ranging the deposition temperature from 555 to 635°C and the total pressure from 100 to 300 millitorrs. Films residual stresses have been determined thanks to the formula of Stoney by measurements of the wafer curvature before and after removal of the back side deposition. The influences of the different deposition parameters are reported and major stress variations are evidenced. By studying the effects of a 600°C crystallisation anneal and by comparing them to those observed for amorphous silicon films deposited from disilane Si2H6, compressive and tensile stresses are respectively related to “surface” and “volume” crystallisation phenomena. The different stress values of amorphous and polycrystalline silicon have been estimated and, according to these results, solutions are finally proposed in order to have a real control of residual stress into silicon depositions and to obtain low stress (σ ≈ 0) polysilicon films.

The improvement of thickness distribution and crystallinity in ZnO thin films prepared by radio frequency (rf) planer magnetron sputtering has been studied. Optimum thickness distribution of less than ± 2.2% in a 3-inch wafer is obtained by changing the substrate angle to the ZnO target and is in accordance with cosine law. The c-axis orientation perpendicular to the silicon substrate is confirmed by x-ray diffraction. The stress of ZnO thin films is larger than 0.3GPa and its distribution is independent of the substrate angle that is set at a slant to the optimum angle for thickness distribution. These results indicate that thickness distribution of ZnO thin films heavily depends on the substrate angle, while the stress and its distribution are independent of the setting angle of the substrate. Stress distribution is attributed to the distribution of argon ions and sputtered molecules impinging a wafer.

Finite-element modeling of the residual stress due to expansion-coefficient mismatch between a 3C-SiC film and Si substrate is presented. The change in residual stress after bulk etching of the silicon substrate to create a suspended diaphragm is also presented. 1, 2, or 3 μm-thick 3C-SiC films are grown at 1600 K on a 100 mm diameter, 500 μm-thick, (100) Si substrate using atmospheric pressure chemical vapor deposition (APCVD) in a cold-wall, vertical-geometry, RF-induction-heated reactor. An axisymmetric, finite-element analysis (FEA) of the cooling and etching process is presented using the ANSYS54 finite-element package. The etching process is modeled by removing the Si elements in the etched region after cooling from film-growth to room temperature. The results show that residual stress in the diaphragm decreases from 3 to 8 percent after etching.

The primary objective of this work is to carry out a combined experimental and numerical investigation of the mechanical stress induced during the Ti-silicidation of narrow lines spaced by a Poly Buffered Local field oxide. Therefore micro-Raman spectroscopy (μRS) measurements of local mechanical stress are combined with simulations by Finite Element Modeling (FEM). Numerical determination of the local stress in the Si-substrate by means of FEM becomes more and more important when scaling down the device dimensions because the spatial resolution of μRS is limited at about 1μm and because the stress in the Si-substrate underneath silicided lines can not be detected using pRS unless the silicide is very thin. The Finite Element analysis was done using SYSTUS and involves a two dimensional plane strain thermo-elastic formulation. A good agreement was obtained between the FEM simulations and the μRS data for line widths above 1Bm making the extrapolation of the FEM to line widths smaller than I1m an attractive alternative for μRS.

The objective of this work was to develop an improved wet etch-stop technology for silicon micromachining. To establish a reference for process improvement, the diffusion process currently used to fabricate p++ Si:B etch stops was comprehensively investigated. Subsequently, a novel germanium- based epitaxial etch-stop technology was developed.

A range of techniques were used to study p++ silicon layers created by boron diffusion into (001) silicon wafers. The results revealed gradients in boron and lattice constant, as well as a graded three- dimensional dislocation array from lattice-mismatch stress. The gradients in boron concentration and dislocation density can lead to curl in micromachined structures. Although annealing steps can remove the boron gradient, a flat membrane will be a tenuous balance.

Epitaxial films of p++Si:B and strain-compensated p++Si1-xGex:B can remove composition gradients and improve process control. However, it is undesirable to depend on p-type layers doped at levels near the solubility limit to prevent etching. We have therefore developed a unique etch stop created from relaxed SiGe alloys. Etch-stop behavior quite similar to heavily boron-doped silicon has been demonstrated in undoped silicon-germanium. Neither strain nor defects are responsible for the etch-stop behavior. A model is proposed based on energy band structure and a SiOx passivation mechanism.

Raman microscopy, using a novel line focus configuration, has been used here to study boron concentration distributions and depth profiles in silicon for two different sources of dopant. Changes in the Raman phonon peak frequency for boron doped silicon have been calibrated against concentration by comparison with SIMS data and a relationship between Raman shift and lattice strain has been obtained.