The effect of thermal history on the local atomic structure of thermally grown and annealed SiO2 films is discussed. Thin films of SiO2 were prepared for study by (1) thermal oxidation of silicon at 850°C and 1050°C in a dry oxygen ambient and, (2) by subsequent rapid thermal annealing at temperatures between 850°C and 1200°C. After thermal processing, the film thickness was reduced incrementally by a wet chemical etchback technique to within 50 Å of the growth interface. Relief of intrinsic growth stress by viscous flow was investigated, at each thickness increment, on a local atomic scale by infrared spectroscopy and on a macroscopic scale by beam-bending type stress measurements. Results were analyzed in terms of a Maxwell model. Changes in film thickness and refractive index resulting from viscous flow during annealing were shown to obey a Clausius-Mossotti relationship. Local atomic strain values, calculated from shifts in the IR bond-stretching frequency, showed a linear correlation with measured stress values giving a value for a plane-strain Young's modulus of 4.7 × 1011 dynes/cm2. This compares favorably with the literature value of 8.8 × 1011 dynes/cm2.

Tensile properties of thin films may be interpreted as a structure sensitive plastic region superposed on an elastic background in a manner similar to bulk specimen tensile testing. However, the limitations of both the material and tensile instrument have not usually been tested in detail.

We report our experience with aluminum and alumina films some 100 nm thick prepared by evaporation of Al followed by anodization for the alumina film. Self-supporting films are glued to glass “jaws” of the nanotensilometer and force-elongation data recorded. Mounting thickness, glue slip-page, instrument calibration, and other possible artifacts will be examined in detail. A typical Al stress-strain curve has an initial small curved region interpreted as a mounting artifact, followed by a primarily elastic (near linear) region and increasing plastic deformation until failure occurs. Alumina films fail in a brittle manner. Characterization techniques include TEM, RBS, and other surface spectroscopies; selected examples will be reported. Strain rate and preliminary annealing data are presented with a microscopic interpretation.

In general, thin metal films are less ductile than their bulk counterparts, grain sizes are much smaller, and they may possess large stresses and unexpected impurities, but have mechanical properties that can be modelled.

We have used x-ray diffraction and picosecond transient piezoreflectance to investigate the structural and elastic properties of three distinct multilayer systems: Mo/Ni, Pt/Ni and Ti/Ni. We demonstrate that the commonly observed lattice expansion perpendicular to the film plane is not a bulk effect, but is localized at the interface between the contacting metals. Incorporating the measured interfacial expansion into a universal binding relation, we show that the measured elastic softening versus compositional repeat distance can result from the interfacial strain.

The dependence of the growth pattern of thin HgCdTe LPE layers on lattice mismatch was investigated. Epilayers grown on substrates with a lattice mismatch of more than 0.2% show island-like growth, which are aligned with the network of misfit dislocation lines. Thin epilayers grown on substrates with a lattice mismatch of less than 0.05% have mirror-like flat surfaces. It was found that the surface defects for thick epilayers also have a lattice-mismatch dependency.

During and after formation of a thin layer of titanium disilicide (TiSi2) on a silicon substrate stress is caused in several ways: Intrinsic stresses are due to the deposition process or to phase transformations and grain growth of the deposited material. Extrinsic stresses are caused by thermal effects: the difference in linear thermal expansion coefficients of the layer and the substrate respectively. Problems related to stresses can occur in semiconductor device fabrication. Stresses can deteriorate gate oxides in MOSFETs and can cause cracks in interconnect lines. Also, focusing problems in lithographic steps can occur because of wafer warpage. In this paper some examples of the different types of stress that can occur are shown and discussed. Both multilayer and self aligned Ti-Si samples have been studied: The advantage of the use of Ti-Si multilayers to produce TiSi2 is that diffusion has to proceed only over a short distance i.e., the multilayer period. So the annealing time can be short. In the self aligned silicidation process, where a layer of a titanium layer on top of a silicon substrate is annealed, the diffusion length is equal to the thickness of the Ti layer. Because longer annealing times are needed, the latter type is used to monitor stress during formation.

A high-temperature viscoelastic stress relief technique has been investigated as a means for reducing in-plane stress encountered during zone-melt recrystallization of patterned silicon-on-insulator structures. This technique incorporates a phosphosilicate glass layer between the silicon film and the insulating substrate to provide a viscous flow mechanism for stress relief within the composite structure. The stress relaxation can bequalitatively described by a mechanical model which couples thermal expansion and viscoelastic flow. The model predicts the time constant for stress relief at high temperatures as a function of pattern size, and the results are useful as a design aid for zone-melt recrystallization experiments.

The stresses and strains induced in thermal laser processing of substrates and thin films on substrates, are obtained in terms of single integrals by solving the thermoelastic equations using a Gaussian profile laser as the heating source. This analysis is applied to silicon thin films on fused silica and sapphire substrates. In part, this study shows that defects can form in the films because the stresses induced during high temperature laser processing of silicon and similar materials can exceed the yield stress under certain experimental conditions.

A highly reliable metallization for high current bipolar integrated circuits was developed with the aid of thin film stress analysis to optimize the process. The final structure was a sandwich layer of TiW and AlCuSi alloy. Film stresses were measured for individual TiW layers as well as for the sandwich layers of TiW/AlCuSi/TiW. In this paper, it will be shown how sputter deposition pressure can drastically affect film stress. For individual TiW layers, the film stress can be varied between compressive and highly tensile by adjusting the Argon pressure. Meanwhile, the AlCuSi layers were always tensile under the normal range of sputtering pressures. By controlling the individual film stresses, a sandwich layer can be designed so that the overall structure has a very low average stress. This work will also show how a TiW capping layer has a major effect on the final average stress of the sandwich structure.

Silicon nitride films prepared by the Plasma Enhanced Chemical Vapor Deposition technique (PECVD) are widely used in microelectronics. The intrinsic stress value of the silicon nitride films is a key factor which determines their reliability. A low frequency (50 kHz) RF discharge and a N2/NH3 /SiH4 gas mixture were used to deposit silicon nitride films on 4-inch silicon wafers in a horizontal hot-wall reactor at a temperature of 350 °C.

The compressive stresses in the deposited films were found to increase linearly with the nitrogen ionization created in the glow discharge. The nitrogen ion bombardment at the film surface induces ion implantation into the deposited film which results in film expansion. The stresses were deduced by the interference fringes technique and from surface profiler measurements. Optical Emission Spectroscopy of the N2/NH3/SiH4 discharge was used for measuring the N+2 peak intensity variation with the partial pressure of the nitrogen gas.

When the nitrogen ionization was increased, a shift of the main infrared transmission peak corresponding to the Si-N bond in the silicon nitride films was observed using Fourier Transform Infrared Spectroscopy (FTIR). The Si-H and N-H infrared peaks in the films were not affected by the ion bombardment. A linear relationship was found between the wavenumber of the main infrared Si-N peak νM (835–875 cm−1) and the intrinsic stress σi of the deposited nitride films with the slope dνM/dσi −4.85×10−8 cm−1 /Pa.

Thin a-Ge films deposited by rf-magnetron sputtering undergo a change from compressive to tensile stress as the argon gas pressure is increased. Spectroscopic ellipsometry, neutron activation analysis, and a dual laser beam reflection technique were used to determine the film's void fraction content, Ar/Ge atomic percent, and stress, respectively. The results of these three measurements indicate that as the film's Ar/Ge content increases, the void fraction decreases and the intrinsic stress becomes compressive. These findings are consistent with currently accepted bombardment mechanisms.

Stress in plasma enhanced chemical vapor deposited (PECVD) SiNx films on InP has been evaluated as a function of source gases (NH3 /SiH4 or N2/SiH4) and plasma operating frequency (high, » 1 MHz or low, « 1 MHz). All films were deposited at 300°C in the same parallel-plate, radial flow plasma reactor. Levels of stress in PECVD SiNx on InP within a continuous range from moderately high tensile (∼ 5 × 109 dyne cm−2) to very high compressive (2 × 1010 dyne cm−2 ) were obtained from appropriate choices of deposition parameters. Deposition from NH3/SiH4 at high frequency produces tensile stress, of magnitude increasing with NH3/SiH4 flow ratio. Deposition from N2/SiH4 at high frequency produces zero to low compressive stress. At low frequency compressive stress is always produced; for N2/SiH4 increasing the gas flow ratio from 25:1 to 500:1 reduces the compressive stress from 1.8 X 1010 to 7 × 108 dyne cm−2. The ability to vary the stress in a dielectric film of approximately constant chemical composition over such a broad range is beneficial for assessing the effects of stress on device performance.

This paper describes the effects of ambient, thermal and mechanical stress on the adhesion of titanium (Ti) to polyimide (PI). Pull testing on the Ti/PI system shows that metal/polyimide bonding degrades when the composite is thermally cycled. A thermochemical mechanism is proposed that accounts for the interface degradation. We do not treat the adhesive interface as a discrete layer, but rather as a gradual “transition zone” between metal and polymer -- a zone that may grow, and/or change in composition and stress state, thus altering the adhesive properties. The mechanism predicts discontinuities that may develop in the transition zone. Through the use of finite element techniques, it is demonstrated that when loads and displacements are imposed on the composite containing interface discontinuities, large localized stresses develop. Such stresses could explain the experimentally observed low strength failures.

The origin and propagation of defects in novel submicron SOI structures have been investigated by transmission electron microscopy. The majority of the defect configurations observed could be explained in terms of dislocations generated as a result of stresses induced by the oxidation process. Dislocations were found to propagate into the silicon along particular {111} planes in conformity with stress-modelling. The examination of particular dislocation configurations allowed an estimation of the stresses at the silicon/ silicon dioxide interface. Defect densities were observed to depend on the geometry of the structures.

Stresses affect the structure and properties of the films significantly. In this paper, Co-Cr films were deposited by D.C. planar magnetron sputtering on rectangular glass plates of thickness varied from 0.18 to 2.00mam. The measurement of stress was performed by the bending plate technique with the detection of capacitance change. The temperature of the bending plate was measured by an Fe-Ni thin film thermocouple on the back side of the substrate. The experimental results showed that the stresses were all tensile, but in the films on the thicker substrates were smaller than that on the thinner ones. And apparently the temperature rise of the thicker substrates during deposition was lower than that of the thinner ones. The magnetic characteristics of the Co-Cr films related to the substrate thickness may be attributed to both stress and temperature difference in the films deposited on the substrates of different thickness.

Undoped polysilicon layers deposited at 620 °C onto a silicon wafer induce a compressive stress of approximately 1 × 109 dynes/cm2 resulting in a bow of the wafer. This stress can be relieved by rapid annealing at temperatures above 1000 °C. A comparison with conventional furnace annealing reveals that the stress relaxation is a weak function of time and strongly depends on temperature. Rapid thermal processing also causes immediate rearrangement of the film structure, as observed by X-ray diffraction. Although both stress and film texture change with annealing, no conclusive relationship is observed.

The mechanics of interface cracks has recently been clarified (e.g., [1]) and a number of efforts are underway to develop a mechanics of interface fracture with applications to composite materials, thin film/substrate systems, and coatings. An intrinsic feature of interfacial fracture is its mixed mode character wherein both shear and normal stresses act on the interface directly ahead of the crack. Depending on geometry and loading, the mixture of modes can range from purely normal stresses (mode I) to purely shear stresses (mode II). Toughness of an interface is characterized by critical combinations of mode I and mode II stress intensity factors (i.e., a locus of critical combinations) rather than just the single critical mode I stress intensity factor in the fracture of homogeneous materials. Equivalently, the critical energy release rate depends on the mode combination for interfacial fracture. Solutions are now available to the following problems: an interface crack in a layered structure where a very thin layer is sandwiched between two thick layers of different material [2], an interface crack between two layers of arbitrary thickness subject to arbitrary combinations of bending and stretching loads [3], and a crack in a substrate paralleling an interface between a thin film and the substrate driven by a variety of loadings including residual tension in the film [4]. These solutions can be used to analyze specimens for determining interfacial toughness and for predicting cracking in thin film or layered structures.