The mechanical response of a specimen incorporating thin films is dictated by a combination of fundamental mechanical parameters such as Young's moduli of the individual layers, and by configurational parameters such as adhesion strength at the interface(s), residual stress distribution and other process dependent factors. In most systems, the overall response will be dominated by the properties of the (much thicker) substrate. Failure within the individual layers, on the other hand, is dependent on the local strain distributions and can not be predicted from the substrate values alone. To better understand the mechanical response of these systems, the strain within the individual layers of the thin film system must be measured and correlated with applied stresses. Phase selectivity of X-ray stress/strain analysis techniques is well suited for this purpose. In this paper, we will review the use of the traditional x-ray stress/strain analysis methods for the determination of the mechanical properties of thin film systems.

Investigations of the mechanical behavior of materials used in magnetic recording is of considerable practical importance in order to assure reliable file performance. With this objective, we have investigated the magnitude of the internal stresses observed in four sputtered thin film materials typically used in magnetic recording. First, the magnitude of the stresses observed in sputtered carbon films with and without hydrogen were measured and the nature of stress relaxation and thermal stability in different ambicnts were investigated. Second, the magnitude of the internal stress in CoPtCr magnetic media with and without a Cr-based sublayer that enhances magnetic performance was examined. Finally, the magnitude of stresses typically observed in thicker sputtered films of alumina were investigated both at room temperature and at elevated temperatures. The magnitude of the stresses observed in these sputtered films and their influence on mechanical performance are discussed.

We have performed real time measurements of intrinsic stresses during growth by reactive dc magnetron sputtering of aluminum nitride (AlN) thin films on silicon substrates in an UHV growth chamber. An experimental setup based on laser beam reflection is constructed such that substrate curvature as well as film thickness can be continuously monitored as growth proceeds. On Si(111) substrates, stress measurements were carried out during growth of both polycrystalline and epitaxial A1N films as a function of deposition pressure. This is the first such comparative study to our knowledge for the AlN/Si system. Our room temperature measurements on polycrystalline films corroborates previous post-growth measurements. Our high temperature measurements provide evidence of large intrinsic stresses and negligible stress relaxation during epitaxial growth of AlN on Si(111). We further compared stress behavior during both room temperature and high temperature growth of AlN films on Si(111) and Si(001) substrates. Our observations indicate while intrinsic stresses during room temperature growth can be compressive or tensile depending on plasma conditions, it is tensile during late stage growth at high temperatures.

Growth stresses in thin metal films on a substrate are important for integrated circuit device fabrication and reliability. The development of growth stress in blanket CVD-WGex films deposited by GeH4 reduction of WF6 is studied by in situ wafer curvature measurements. Depending on process conditions, the growth stress varies from 600 MPa tensile in β-WGex films deposited at low temperatures (T < 400 °C) to 400 MPa compressive in α-WGex films deposited at high temperatures (T > 400 °C). At very low temperatures amorphous WGex films grow without significant stress. Development of tensile growth stress is favoured by low temperatures and high growth rates, whereas development of compressive growth stress is favoured by high temperatures and low growth rates. Although the material properties of CVD-WGex films depend on process conditions, the development of growth stress in these films shows a dependence on deposition temperature and growth rate which is similar to pure α-W films deposited by H2 reduction of WF6. The results will be discussed in view of a model which has recently been put forward to explain development of growth stress in polycrystalline metal films on a substrate.

Shape memory alloy (SMA) thin films have attracted attention due to their potential application as actuators in micromechanical systems. The stress development in Ni50Ti50 films has been measured using a cantilever beam method. It is of interest to study their transformation characteristics. We have investigated a striking change of the stress during the B2-B19 transformation. The magnitude of this stress depends on the film thickness and reflects the interface constraint. In order to obtain direct information about the interface, in situ cross-section TEM observations of NiTi/SiO2/Si heterostructures were performed at various temperatures. A layer of parent phase at the film/substrate interface was found to buffer the transformation induced stress. The origin and effect of this buffer layer was analysed based on the TEM cross-sectional observations of the formation history of the layer.

Mechanical stress measurements in long window-plated strips of permalloy have been made and show an anisotropy in the in-plane stresses. The stress parallel to the strip length is independent of the strip width and thickness, and exceeds the in-plane stress perpendicular to the strip length. The difference between the two stresses depends on the ratio of strip width to thickness, and is in agreement with finite element simulations of stress distributions in similar thin film strips which are etched out of a sheet film of the same material. This result allows calculations of intrinsic stress distributions in more complex window-plated structures. The effect of annealing on the strips is to increase the in-plane stress anisotropy as well as the overall level of stress, by an amount which increases with the annealing temperature.

The minimum temperature for the crystallization of amorphous TiNi on substrates is of interest in developing thin-film SME material while minimizing chemical interactions with the substrate. Using 20 micron thick free standing TiNi material annealed in a vacuum furnace, X-Ray diffraction of the thin-films indicated that the crystallization occured within 20 minutes at 510°C, 490°C and 480°C. At 450°C, crystallization kinetics were significantly slower, and the foils were fulling crystallized after annealing for 7.5 hrs. To further lower the crystallization temperature, cold working of the foil by rolling was introduced and full crystallization was observed after 7.5 hours annealing at 400°C in a cold-worked foil. Cold working and annealing at 400°C and 350°C for 7.5 hrs did not observably promote lower crystallization temperatures.

160 nm Ti-W-N layers were sputtered on Si/SiO2/Al-Si-Cu substrates with the thickness of the Al-Si-Cu layer varying between 0 and 1000 nm. As a result the apparent grain size of the Ti-W-N increased from 40 to 130 nm and the compressive deposition stress of the Ti-W-N decreased from 1.77 to 0.34 GPa. The relation between stress and the inverse grain size was found to be linear for grain sizes smaller than 130 nm. Assuming a simple model, the excess volume per unit area of grain boundary can be estimated from the slope of the fit and was found to be 0.17 nm.

Light scattering measurements can be used as a quick and non-destructive method for evaluating the microstructure of these layers.

Stresses which build up in thin films, such as tantalum, during thermal processing, can cause major reliability problems in x-ray optics and electronic applications. We have demonstrated that 50 nm to 200 nm thick sputtered beta tantalum thin films undergo repeated compressive stress increases when thermally cycled from room temperature to 400°C (at 10°C/min) and back in a purified He ambient because of low levels of oxygen gettered by the tantalum. The oxygen contamination is a result of the poor quality of the quartz annealing chamber atmospheric seal. As-deposited stress in the sputter deposited tantalum films ranges from -1 to -4 GPa. The compressive stress build up was monitored in situ and was shown to increase -0.5 GPa on average after each thermal cycle for a final value of -6 to -7 GPa after seven cycles. After being cycled thermally seven times any perturbation of the film such as a four point probe resistivity measurement can cause the film to instantaneously crack in a serpentine pattern relieving the large compressive stress. Auger electron spectroscopy depth profiling analysis indicated that the as-deposited films contained one atomic percent oxygen which increased to eight to twelve percent after seven thermal cycles accompanied by an approximate doubling in resistivity. In conclusion, the increase in oxygen concentration in tantalum thin films which occurs upon thermal cycling leads to a repetitive increase in compressive stress which could be detrimental when the films are used in x-ray or electronic applications.

Using an in situ stress measurement technique which measures stress as a function of annealing temperature, we have investigated the effect of phosphorous and boron doping of silicon dioxide glass films deposited by low-pressure chemical vapor deposition (LPCVD) and plasma-enhanced chemical vapor deposition (PECVD). The stress at room temperature is σi. Upon heating, it increases to a maximum, σm, corresponding to a temperature Tm, above which the stress is reduced to zero at a temperature T0. All these parameters plus the expansion coefficient are dependent on dopant concentrations and deposition technique.

Stress in thin films has been measured very precisely (< 10 MPa) by analysing the curvature of quartz glass substrates before and after film deposition by means of a ZYGO Mark IV interferometer system. TiO2 films of approximately 100 nm thickness were prepared by reactive evaporation (RE), reactive ion plating (IP), plasma impulse chemical vapor deposition (PICVD) and spin coating (SC). Large variations in stress are found for different coating techniques and deposition conditions. This can be correlated to differences in optical properties, film density and crystalline structure. Relaxation effects and the influence of thermal treatment are also studied. The crystallization of amorphous TiO2 during heat treatment is accompanied by significant changes in film stress. The crystal size and morphology of TiO2 films after heat treatment strongly depend on the deposition technique and process conditions.

A systematic evaluation of the stress-temperature behavior of O3-TEOS Sub-Atmospheric CVD (SACVD) oxide films used in IMD applications was performed in order to elucidate the surface dependence and mechanical stability of these films when deposited on different oxide underlayers such as PECVD TEOS-based, Silane-based, and thermally grown oxides. SACVD films were deposited at 360°C using different O3:TEOS molar flow ratios. The results obtained in this study indicate that the stress-temperature behavior of SACVD films deposited on PECVD Silane-based and TEOS-based oxides using N2O as the oxidizing agent is similar to that of the SACVD films deposited on silicon. However, a large stress hysteresis is obtained when SACVD films are deposited on PECVD TEOS-O2 and thermally grown oxide underlayers. This large stress hysteresis results from the surface dependence effects of the SACVD oxides on thermally grown and PECVD TEOS-O2 oxide underlayers. Also, this stress hysteresis can be reduced or eliminated by treating the oxide underlayer surface with an in-situ Ar, He, or N2 plasma. Furthermore, it was also found that the SACVD oxides deposited using higher O3:TEOS molar flow ratios are chemically and mechanically more stable.

Thermal mismatch between the expansion coefficients of a substrate and deposited thin film gives rise to extrinsic stress in the film. Different stress states are induced into the Co/Si thin film combination by depositing the multi-layers on Al and Si substrates respectively. Interdiffusion at 360°C is found to be influenced by the extrinsic stress in the films. The difference in Co2Si growth rate on different substrates is ascribed to the difference in composition (Co molar fraction) at the Co/Co2Si and Co2Si/Si interfaces due to different stress states.

Uniaxial stresses up to 0.3GPa, both compressive and tensile, can be generated in thin epitaxial layers by substrate bending. We demonstrate the feasibility of the method and calibrate the stress, measuring piezoresistance at temperatures below the impurity freeze-out, on lightly B-doped silicon MBE layers, using the known expression for a surface stress in a bent cantilever beam. The ground state energy displacement of B in Si thus obtained can vary by ±5meV, suggesting potential applications of the technique in tunable devices, involving doped semiconductors or heterostructures.

This work presents the first results of a study aimed at better understanding the elastic behaviour of hard coatings produced by various techniques. This is important also in view of the need to be able to control the level of internal stresses, particularly in PVD coatings. It is well known that in extreme cases excessive internal stress can lead to complete film destruction. We devote this paper to reactively magnetron sputtered TiN, one of the most widely used compounds. Thin TiN films of different compositions were deposited on Si substrates and characterized by SEM, AES, XRD and Brillouin light scattering.

Residual compressive stress of diamond-like carbon (DLC) films was measured by beam deflection method. DLC films were deposited on thin Si wafers using r.f. plasma decomposition of methane and benzene. Negative bias voltage of the cathode was varied from -100 to -800 V and deposition pressure from 3 to 100 mTorr. When using benzene as precursor gas, the residual stress monotonically increases as increasing . (Here, Vb is the negative bias voltage of cathode and P the deposition pressure.) In case of using methane, however, the residual stress has a maximum value at between 70 and 100 V/mTorr1/2. Because of the difference in molecular size between benzene and methane, the mean free path of ions in benzene discharge is 5 times shorter than that in methane discharge. The contrasting behavior of residual stress is discussed in terms of the difference in ion energies at the specimen surface due to the difference in mean free path. On the other hand, total hydrogen concentration decreases as increasing in both cases. This result thus shows that the total hydrogen concentration cannot be a key to understand the behavior of residual stress.

Gd-Fe films were deposited by magnetron sputtering on glass substrates. The results shoved that the stress is compressive in most cases, and a larger compressive stress exists at the film-substrate interface. The stress is independent on the film thickness when the film thickness is greater than 0.20 μm. The stress becomes tensile only when the annealing temperature attains 400ºC. Therefore, the interfacial stress and the growth stress coexist in the Gd-Fe films, and the former will not be dominant when the film thickness is greater than 0. 2μm. The interfacial stress may originate from the inhomogeneous distribution of film composition at the interface, and the origin of growth stress can be explained by the atomic peening effect. The relation between the stress and the annealing temperature can be comprehended with the amorphous structure model.

A new method for evaluating the elastic moduli and residual stresses in thin films is presented. This technique is based on measurement of the complete deflected shape of pressurized diaphragms made of the material to be evaluated. It offers important advantages over previous techniques in which only the displacement of the center of the diaphragm is measured as a function of pressure. The experimental apparatus uses a laser beam and its reflection from the sample surface to determine relative height. Measurements are made over an area of the sample by performing a two-dimensional scan of the sample under the laser beam. The most important advantage of determining the entire deformed shape of the diaphragm is that the shape is known rather than assumed. This way, the particular theory chosen to analyze the structural response does not also dictate a deformed shape based solely on the maximum deflection at the midpoint. Plate theory, as opposed to the more traditional membrane theory, is used to account for the dominance of bending behavior at low pressure and near the diaphragm edges. An energy method is employed to estimate stiffness and residual stress. The accuracy of this approximate method is directly related to how closely the assumed mode shape used in the analysis approximates the true response of the structure. Since these surface profiles are measured experimentally, the resulting property estimates are obtained with increased confidence.

The ability to measure the temperature-dependence of the hardness of thin films is useful from both an applications and a scientific standpoint. For this reason, we have designed and constructed a load- and depth-sensing indentation tester combining sub-nanometer resolution with the ability to operate over a range of temperatures, currently 150K to 400K. This paper describes the new experimental apparatus and reports preliminary data on 440C stainless steel substrates with and without a 0.75 μm ZrN coating.

Understanding of both low load Nanoindentation and Atomic Force Microscopy relies on a knowledge of the mechanical response of contacts at the few nanometre scale. To investigate this area we have developed a new type of AFM/Indenter which applies forces (<1nN up to 1mN) to a tip mounted on a cantilever. The displacement of the cantilever is detected using laser heterodyne interferometry. The combination gives a direct measurement of contact compliance at very low loads. Contacts of diameter below 100Å can be repeatably made.

We show that a water film is generally present in the contact during operation in air, and show how its size can be estimated from our data. One implication is that ambient operation of AFM in both repulsive and attractive modes may necessitate the presence of a liquid lubricating film, explaining the relative lack of damage in many AFM measurements. The estimation of contact area from stiffness is complicated by this layer. We show that the contact stresses are likely to be very high.