Polycapillary optics, shaped arrays consisting of hundreds of thousands of hollow glass capillary tubes, can be used to redirect, collimate, or focus x-ray beams from conventional, laboratory-based sources. Multifiber polycapillary collimating optics were placed without system optimization into a number of diffractometer systems. Strain and texture measurements were performed. Measurements on thin films with and without the lens yielded gains of 8–100. Gains are higher in comparison with systems employing pinhole rather than one-dimensional slit collimation. The collimation also resulted in peak symmetrization, which simplified peak shape analysis.

Focused beam optics provide even larger gains. Gains of 100 have been demonstrated using polycapillary optics with spot sizes as small 20 μm. This is of particular significance for providing spatial resolution for low signal applications such as thin films. The inverse dependence of the critical angle for total external reflection on photon energy also results in suppression of high energy photons. This suppression of Bremsstrahlung can allow use of higher tube potentials to increase characteristic line emission. Background suppression from a polycapillary optic ″soller slit″ is also enhanced due to the two dimensional collimation and much smaller acceptance angle of the polycapillary optics. Polycapillary optic alignment is also faster and more convenient than the usual parafocusing geometry. The combination of background suppression, intensity gain and increased tube emission by employing polycapillary optics greatly increases the signal to noise ratio for thin film stress analysis.

This paper introduces coherent gradient sensing (CGS) as an optical, full-field, real-time, non-intrusive and non-contact technique for measurement of curvature and curvature changes in thin film and micro-mechanical structures. The technique is applied to determine components of the curvature tensor field in multilayered thin films deposited on silicon wafers. Curvature field measurements using CGS are compared with average curvatures obtained using high-resolution x-ray diffraction. Finally, examples are presented to demonstrate the capability of CGS in measuring curvature in a variety of thin film and micro-mechanical structures.

FEM simulation of micro-rotating-structures was performed for local measurement of residual stresses in thin films. A sensitivity factor is introduced, studied and tabulated from the simulation results. The residual stress can be evaluated from the rotating deflection, the lengths of rotating and fixed beams, and the sensitivity factor. The micro-structure technique was applied to measure residual stresses in both silicon nitride and polysilicon thin films, before and after rapid thermal annealing (RTA), and further confirmed by wafer curvature method. Residual stresses in polysilicon films at different RTA stages were also characterized by micro-Raman spectroscopy (MRS). The experimental results indicate that micro-rotating-structures indeed have the ability to measure spatially and locally residual stresses in MEMS thin films with appropriate sensitivities.

We report a novel method to measure Poisson's ratio v of compressively prestressed thin films. The method exploits a v-dependent mechanical instability transition of buckled long rectangular membranes under differential pressure. This allows to determine v with high accuracy, using a physical model of the plate instability. The method was used to extract v = 0.253±0.017 from 0.704±0.003 μim thick PECVD silicon nitride films.

A simple method to measure the elastic modulus and Poisson's ratio of diamond-like carbon (DLC) films deposited on Si wafer was suggested. This method involved etching a side of Si substrate using the DLC film as an etching mask. The edge of DLC overhang free from constraint of Si substrate exhibited periodic sinusoidal shape. By measuring the amplitude and the wavelength of the sinusoidal edge, we can 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 films. By comparing the biaxial elastic modulus with plane-strain modulus, E/(1 −v2), measured by nano-indentation, we could further determine the elastic modulus and Poisson's ratio, independently. This method was employed to measure the mechanical properties of DLC films deposited by C6H6 r.f. glow discharge at the deposition pressure 1.33 Pa. The elastic modulus, E, increased from 94 to 128 GPa as the negative bias voltage increased from 400 to 550 V. Poisson's ratio was estimated to be about 0.22 in this bias voltage range. For the negative bias voltages less than 400 V, however, the present method resulted in negative Poisson's ratio. The limitation of the present method was discussed.

Machine parts like rolling bearings or gears are stressed during operation in a changing mechanical strain. This causes wear by impacts and wear by rolling which is marked by the so called surface ruin. The appearance of surface fatigue is based upon structural transformation, cracking and cracking-growth processes and ends with the separation of debris particles caused by the above mentioned permanent changing strain /1/. The final stage, which is equivalent to the component failure, is the so called ‘pitting’ on the technical surface, which is characteristically named ‘surface fatigue’.

The impact tester is used for detailed research about failure mechanisms of thin films. Statements about the adherence of hard material coatings under dynamic compressive stress can be made using this test method, due to the possibility to simulate some effects of rolling strain. Therefore a hard metal ball strikes with a frequency of up to 50 Hz onto the surface. The altitude stress can be varied to get a detailed evaluation of fatigue strength under reversal strain.

Selected hard material coatings were analized after testing with the described method applying an impact force of 300 N, 500 N and 700 N. In the framework of this presentation MSIP (Magnetron-Sputter-lon-Plating) coatings on titanium- and chromium basis were used. The fatigue defects and the results of this study will be discussed depending on structure and morphology of thin films.

A new test approach is presented to measure the fracture toughness of thin films. The polysilicon specimen is a center-cracked panel that is 3.5 μm thick and 3 mm wide with a 100 μm long slot in the center. It is subjected to tensile loading, and the crack-opening displacement is measured by interferometry. The average toughness is 1.4 ± 0.65 MPa-m1/2.

A method for tensile testing thin films at elevated temperatures is presented. Freestanding tensile specimens were prepared by evaporating 2.1 μm of gold onto a patterned oxidized silicon wafer. Tensile tests were performed at various temperatures up to 600°C. The ultimate tensile strengths of the films decreased as temperature increased, and ductility increased with increasing temperature. Films tested at high temperatures (≥ 400 °C) exhibited significant strain rate-dependence, while films tested at low temperatures (≤ 200 °C) were found to be virtually rate-independent. At higher temperatures and strain rates, stress-enhanced grain growth was observed.

This study explores the difficulties encountered when using conventional nanoindentation techniques to measure the Young's modulus and hardness of hard films on soft substrates. In general, the indentation measurement of film/substrate systems is affected by four material properties: the Young's modulus and hardness of the film, and the Young's modulus and hardness of the substrate. For the particular case of a hard film on a soft substrate, there is a tendency for the material around the hardness impression to sink-in which results from the large difference in yielding of the two materials. In this study, a ‘model’ system consisting of NiP on annealed Cu was used to explore the behavior. This system is interesting because the film and substrate have similar Young's moduli, minimizing the elastic behavior as a variable. In contrast, the hardness of NiP is approximately 7–18 GPa, and that of the annealed copper is less than 1 GPa, providing a factor of 10 difference in the plastic flow characteristics. Experimental results indicate that standard analytical methods for determining the contact depth, hardness and Young's modulus do not work well for the case of a hard film on a soft substrate. At shallow contact depths, the measured indentation modulus is close to that of the film, but at larger depths sink-in phenomena result in an overestimation of the contact area, and an indentation modulus which is less than the Young's modulus of both the film and substrate. Atomic force microscopy (AFM) and scanning electron microscopy (SEM) provide critical details of the physical processes involved, and illustrate how the standard data analyses overestimate the true contact area.

We have developed a new technique for visualizing displacement fields of indentations in thin films. In this technique, the indented film consists of alternating layers of two different materials. One of the materials serves as a marker for visualizing the plastic flow induced by the indentation. Focused Ion Beam (FIB) milling is used to cross-section the indentation, revealing the deformed layers. This technique can be used to study how the presence of the substrate affects the plastic displacement field around the indentation. The technique is applied to a multilayered film of aluminum and titanium nitride on a silicon substrate. The titanium nitride layers are much thinner than the aluminum layers and serve the function of marker. Pile-up of the film material around the indenter and the effect of the hard substrate are easily revealed and a mechanism for pile-up is suggested. The technique also shows that the grain structure in the deformed zone around the indentation is altered profoundly.

Nanoindentation studies on metal and semiconducting surfaces often display excursions in the load-displacement curves. These displacement excursions have been attributed to phase transitions, oxide breakthrough, surface contamination effects, and dislocation nucleation under the indenter tip. We have shown recently that displacement excursions were present for nanoindentation on single crystal Au (111), (110) and (100), and were attributed to dislocation nucleation since all other phenomena were ruled out. We present our recent results that have been aimed at understanding the effects of surface modification at the nanoscale on dislocation nucleation. The effects of modifying the Au surface with electrochemically deposited metal monolayers (Pb and Ag), with an electrochemically deposited oxide monolayer and an electrochemically reconstructed surface will be presented. Hardness differences as great as a factor of 3 have been observed for these surfaces. These experiments are unique in that they were carried out under electrochemical control where strict control of the surface cleanliness can be maintained.

Continuous indentation has been used to evaluate the mechanical response of chromium and platinum films deposited on silicon and glass substrates. The traditional method of analysis of indentation data to determine modulus and hardness is sensitive to residual stress in the film, thereby suggesting that modulus measurements may be used to gauge residual stresses. Evaporated metal films, which are under residual tension, tend to show a lower modulus (using standard calculation methods) than the bulk materials. However, a sputtered platinum film exhibits an apparently higher modulus than the bulk material. These data suggest that differences between the calculated and expected modulus using indentation tests cannot be used to unequivocally determine the stress state in a thin film.

A method that is becoming increasingly common for measuring the mechanical behavior of thin films is low-load indentation testing. However, there can be complications in interpreting the results as many factors can affect hardness and moduli measurements such as surface roughness and determination of the indentation contact area. To further our understanding, the mechanical properties of thin (50 nm) films of AlN on sapphire substrates were evaluated using a scanning force microscopy (SFM) based pico-indentation device to allow imaging of the surface and indentations. Our primary emphasis was the types of problems or limitations involved in testing very thin, as deposited films in which properties are desired over indentation depths less than 50 nm.

This study reports the results of an investigation of the mechanical properties of polytetrafluoroethylene (PTFE) thin films on silicon substrates in the 0.5 to 15 μm thickness regime using frequency specific depth-sensing indentation. All measurements were conducted at an excitation frequency of 45 Hz using a constant (1/P dP/dt) load ramp of 0.1 s−1. The modulus of the PTFE at a depth of 5% of the film thickness was measured to be approximately 1 GPa (v = 0.46) independent of film thickness. These values are somewhat higher than the values obtained from free-standing 15 μm film measurements of 0.4 GPa for the tensile modulus and 0.49 GPa for the storage modulus @ 1.1 Hz. The film hardness at these depths was observed to range between 30 and 55 MPa with no correlation observed between the hardness and respective film thickness. While reliability modeling for interconnects currently uses interlayer dielectric mechanical properties data determined from free-standing films with thicknesses of several microns, these insitu results should more closely mimic the constrained deformation that occurs during service and perhaps lead to a better understanding of the electromigration resistance of PTFE.

The AC force modulation technique in nano indentation has been extended to study the dynamic viscoelastic properties of polymer materials. The frequency and temperature dependent loss and storage modulus, the time-temperature superposition, the mechanical relaxation and the activation energy for the relaxation process can all be directly measured from nano scale contact. The results obtained for PMMA and Polyisoprene are in good agreement with the reported literature results obtained from conventional dynamic testing. The technique is likely to be of particular value for polymer thin films.

Experiments and finite element simulations have been performed to examine errors in the measurement of hardness and elastic modulus caused by pile-up when soft films deposited on hard substrates are tested by nanoindentation methods. Pile-up is exacerbated in soft-film/hardsubstrate systems by the constraint imposed on plastic deformation in the film by the relatively non-deformable substrate. To experimentally examine pile-up effects, soft aluminum films with thicknesses of240, 650, and 1700 nm were deposited on hard soda-lime glass substrates and tested by nanoindentation techniques. This system is attractive because the elastic modulus of the film and the substrate are approximately the same, but the substrate is harder than the film by a factor of about ten. Consequently, substrate influences on the indentation load-displacement behavior are manifested primarily by differences in the plastic flow characteristics alone. The elastic modulus of the film/substrate system, as measured by nanoindentation techniques, exhibits an increase with indenter penetration depth which peaks at a value approximately 30% greater than the true film modulus at a penetration depth close to the film thickness. Finite element simulation shows that this unusual behavior is caused by substrate-induced enhancement of pile-up. Finite element simulation also shows that the amount of pile-up increases with increasing penetration depth, and that the pile-up geometry depends on the work-hardening characteristics of the film. Because of these effects, nanoindentation techniques overestimate the true film hardness and elastic modulus by as much as 68% and 35%, respectively, depending on the work-hardening behavior of the film and the indenter penetration depth. The largest errors occur in non-work-hardening materials at penetration depths close to the film thickness, for which substrate-induced enhancement of pile-up is greatest.

We examine the distribution of failure times in a simple and computationally efficient, yet reasonably authentic, model of interconnect reliability that allows consideration of statistically significant samples. The model includes an approximate description of the distribution of grain sizes and texture in narrow interconnects, an effective treatment of stress evolution associated with mass transport along grain boundaries, and local relaxation of stresses due to void formation. Failure time distributions for populations of idealized structures are analyzed to aid in interpretation of model behavior.

Finite element solutions to problems of electromigration and stress-driven diffusion are presented. The numerical procedures are based on a formulation that assumes bulk diffusion, so that diffusion along grain boundaries and interfaces is not explicitly taken into account. Two cases are considered: (1) a tungsten via/aluminum line junction, and; (2) a one-dimensional case with a blocking boundary at one end of the line. The results show that line geometry can significantly affect the magnitude of the stresses that develop, although in each case the maximum hydrostatic stress in the line was proportional to the square root of time.

Copper has been deposited for filling sub-0.5 μm trenches by using electroplating. Electroplating with pulse plating conditions provides the high deposition rate (0.5–1 μm/min) and defect-free filling the 0.25 μm trenches and vias of high aspect ratio (>4:1). Enhanced copper electroplating at the trench bottom has been achieved. The median grain size of electroplated copper was measured to be about 1 jim and the lognormal standard deviation is about 0.4 μm. Strong <111> texture was observed in electroplated Cu film. Low stress of electroplated Cu films and excellent adhesion of plated Cu to sputtered Cu seed were observed.

Several under bump metallisation (UBM) schemes using Ni or CuNi alloys as the solderable layer were investigated. Cr or Ti was used as the adhesion layer. The UBM pads of different compositions were sputter deposited on silicon wafers and patterned using standard photolithographic processes. Eutectic Sn-Pb solder balls were reflowed on top of the pads. The resulting interfacial microstructures were examined by SEMIEDX analysis of cross-sectioned samples. The integrity of the UBMIsolder interface was characterized by micromechanical shear testing of flip chip test samples. Growth of intermetallic layers was found to be significantly slower in Ni and CuNi schemes compared to pure Cu. The joints on Ni and CuNi had also a better adhesion at the UBM/solder interface, and in the shear tests the fracture occurred through the solder