Polysilsesquioxanes are finding wider uses in microelectronics. However these materials are brittle and their mechanical behaviors need to be better understood. In this paper the mechanical properties of a cured methyl silsesquioxane network are studied and they are found to vary strongly with curing conditions. In the investigated curing temperature range there seems to be a correlation between the fracture toughness (KIc) and the viscoelastic transitions, which are affected by the network crosslink density. A higher crosslink density suppresses the cc and β transitions and reduces the KIc. A toughening approach is designed and its applicability is demonstrated by the increased fracture toughness.

Drying of gelatin films was studied by an in-situ cantilever beam method combined with spectroscopic ellipsometry, enabling measurement of the stress development, film shrinkage and compositional change during the drying process. Experiments are carried out under different drying conditions (such as drying extent variation, drying rate variation). The film experiences stress relaxation that indicates visco-elastic relaxation or plastic deformation happened during drying process. Plastic deformation is verified by carefully reversing the drying process, however, the plastic deformation accompanied by visco-elastic relaxation complicates the case of finding the value of the yield stress.

Tribochemical studies of the effect of lubricant bonding on the tribology of the head/disk interface (HDI) were conducted using hydrogenated (CHx) carbon disk samples coated with perfluoropolyether ZDOL lubricant. The studies involved drag tests with uncoated and carboncoated Al2O3-TiC sliders and also thermal desorption experiments in an ultra-high vacuum (UHV) tribochamber. We observed that a larger mobile lubricant portion significantly enhances the wear durability of the (head/disk interface) HDI by providing a reservoir to constantly replenish the lubricant displaced in the wear track during drag tests. In the thermal desorption tests we observed two distinct temperatures of desorption. The mobile ZDOL layer is desorbed at the lower thermal desorption temperature and the residual bonded ZDOL layer is desorbed at the higher thermal desorption temperature. We also observed that the hydrogen evolution from CHx overcoats initiates lubricant catalytic decomposition with uncoated Al2O3/TiC sliders, forming CF3 (69) and C2F5 (119). The generation of Hydroflouric acid (HF) during thermal desorption experiments provides the formation mechanism of Lewis acid, which is the necessary component for catalytic reaction causing Z-DOL lube degradation.

Nitrogen-doped carbon (CNX) overcoats, ranging from 1 to 6nm in thickness, were deposited on magnetic recording disks by a DC-sputtering process. A critical load, based on the first occurrence of coating damage, was used as a semi-quantitative measure of the mechanical strength of these overcoats. It was found that the critical load decreased in a nearly linear manner with the CNX thickness from 6nm down to ∼2nm regime. However, the 1nm thick CNX coating deviated from this trend with a significant decrease in critical load. High-resolution SEM was employed to find the critical loads as well as to reveal the details of the coating wear morphology and the CNX failure mechanism.

The elastic constants of thin Diamond-Like Carbon (DLC) films supply important information, but their measurement is difficult. Standard nanoindentation does not directly measure the elastic constants and has strong limitations particularly in the case of hard thin films on softer substrates, such as tetrahedral amorphous carbon on Si. Surface acoustic waves provide a better mean to investigate elastic properties. Surface Brillouin scattering (SBS) intrinsically probes acoustic waves of the wavelength which is appropriate to test the properties of films in the tens to hundreds of nanometers thickness range. SBS can be used to derive all the isotropic elastic constants of hard-on-soft and soft-on-hard amorphous carbon films of different kinds, with thickness down to less than 10 nm. The results help to resolve the previous uncertainties in mechanical data. The Young's modulus of tetrahedral amorphous carbon (ta-C) turns out to be lower than that of diamond, while the moduli of hydrogenated ta-C (ta-C:H) are considerably lower than those of ta-C because of the weakening effect of C-H bonding.

Amorphous diamond is a new material for surface micromachined microelectromechanical systems (MEMS) and sensors. Its strength and modulus was tested in uniaxial tension by pulling laterally on a specially designed sample with a flat tipped diamond in a nanoindenter. Several sample designs were attempted. Of those, only the single layer specimen with a 1 by 2 μm gage cross section and a fixed end rigidly attached to the substrate was successful. Tensile load was calculated by resolving the measured lateral and normal forces into the applied tensile force and frictional losses. Displacement was corrected for machine compliance using the differential stiffness method. Post-mortem examination of the samples was performed to document the failure mode. The load-displacement data from those samples that failed in the gage section was converted to stress-strain curves using carefully measured gage cross section dimensions. Mean fracture strength was found to be 8.5 ± 1.4 GPa and the modulus was 831 ± 94 GPa. Tensile results are compared to hardness and modulus measurements made using a nanoindenter.

A commonly used technique to compute mechanical properties from indentation tests is the Oliver and Pharr method. Using dimensional analysis and finite element modeling, this paper investigates errors when the Oliver and Pharr method is used to compute thin film properties.

The cobalt chrome alloy Co-28Cr-6Mo is widely used in human joint replacement applications. These joints are highly sensitive to wear and are usually replaced after approximately ten years of use. In order to extend these implants' service lifetimes, a thin film of diamond may be applied to the implant wear surfaces by microwave plasma chemical vapor deposition (MPCVD) following MPCVD nitridation. Diamond films often delaminate from cobalt chrome due to a high thermal expansion mismatch. Additionally, under most conditions diamond films degrade into graphite by dissolution of nuclei into solvents like Co and Cr. By nitriding the cobalt chrome through MPCVD prior to diamond deposition, a usable diamond film may be achieved. It may be possible to merge both the nitriding and deposition steps into a single process, since both nitridation and deposition will be performed by MPCVD. In addition, controlled amounts of nitrogen in hydrogen/methane plasma under CVD conditions result in a nanostructured diamond coating. We have investigated the formation of both micro- and nanocrystalline diamond formation on cobalt chrome.

Pure diamondlike carbon thin films largely bonded by four-fold coordination suffer from a large internal compressive stress that gives rise to a serious adhesion problem. In this work, functionally gradient (FG) diamondlike carbon thin coatings were prepared by pulsed laser deposition in a high vacuum chamber as an alternative approach to address the adhesion problem of diamondlike films. Copper, silver and titanium were incorporated into the growing films with their concentration as a function of the distance from the substrate surface. The top of the thin coating is pure DLC of about 400 nm in thickness. The total thickness of the functionally graded superhard DLC coatings can exceed 1.0 μm without buckling. Visible micro-Raman spectroscopy was used to characterize the bonding structure of the layers which contain alloy atoms. High resolution transmission electron microscopy was employed to study the microstructure of the coatings. Nanoscale mechanical characterizations using Nanoindenter XP™ were carried out to study the mechanical behavior of the functionally gradient DLC films.

Hydrogen free amorphous carbon (a-C) and carbon nitride (a-C:N) were synthesized by means of shielded arc ion plating in which a shielding plate was inserted between a target and a substrate in order to reduce macroparticle deposition onto the substrate. Using a graphite target as a cathode, thin films of a-C and a-C:N were prepared in an arc discharge plasma of argon or nitrogen gas, respectively, at a pressure of 1 Pa. Based on nanoindentation, mechanical properties of these films were studied in relation to substrate bias voltage (VS). The a-C films prepared at VS ranging from −50 to −200 V consisted of diamond-like phase and showed hardness higher than 20 GPa with its maximum of 35 GPa at Vs = −100 V. However, hardness of the films deposited at VS < -300 V was less than 7 GPa indicating that the films were converted to graphite-like phase due to excessive ion impact in Ar plasma. Wear resistance of the a-C films was closely related to their hardness. Namely, a harder a-C film was more wear resistant. On the contrary, hardness of the a-C:N films was less dependent on VS. It remained in the range of 10 to 15 GPa and was much lower than the maximum hardness of the a-C films. Nevertheless, the wear resistance of the a-C:N films was comparable to or much better than the a-C films. In particular, the a-C:N film prepared at VS = -300 V was so wear resistant that the film showed no apparent wear when rubbed with a diamond tip less than 100 nm in tip-diameter at a contact force of 20 μN. The presence of β-C3N4 like phase characterized by a Nls XPS peak at 400.5 eV has found to be crucial for high wear resistance of the a-C:N films.

Diamond films were grown over Si substrate at 1253K by the hot filament chemical vapor deposition method using CH4/H2 gas mixture, and intrinsic stresses in the film were deduced from the ex-situ curvature measurements. In order to account for the creep deformation of the Si substrate, an elastic/plastic stress and strain analysis were conducted. Results showed that intrinsic stresses were generally several times larger than the average film stresses and always positive increasing with the film thickness. For the film thickness larger than 10μm, stress relaxation by creep of the substrate became significant, and must be considered for the accurate assessment of the film stress in diamond. Later, an analysis based on the grain growth accounted for the development of intrinsic stresses reasonably well.

Microlaminate coatings are made of many alternating layers of two hard materials that, when combined in very thin layer on the nanometer scale, produce coatings with hardness that approaches diamond. In this report, we address these properties, from our investigations on the multilayer structures of titanium carbide (TIC) and diamond-like carbon (DLC) deposited on Si (100) substrates using pulsed laser deposition (PLD) technique. X-ray diffraction and Raman spectroscopy were used for the structural studies and the mechanical properties were analysed by the nano-indention technique. Microlaminate coatings of TiC/DLC and DLC/TiC coatings with varying layers were deposited on Si(100) substrates. Analysis of mechanical properties revealed that the hardness and modulus values of the multi-layers lie between those of monolithic coatings of TIC and DLC.

We have studied diamond films grown by electron cyclotron resonance-assisted chemical vapor deposition (ECR-CVD) at low pressure (1.0 Torr) and temperatures (550–700 °C). These films were grown on seeded Si (111) substrates with different diamond seed densities (0.225, 1.5, 2.3, and 3.1 × 109 nuclei/cm2). Scanning electron microscopy (SEM), X-ray diffraction (XRD) and Raman spectroscopy (RS) were employed to investigate the crystalline quality, diamond yield, and stresses developed in the films as a function of seeding density. Thermal interfacial stress, interactions across grain boundaries, and internal stress were considered in order to account for the total stress observed from the Raman band. We present correlations among seed density, relative amount of non-sp3 phase, O/C ratio, and total intrinsic stress.

Diamond wafer, considered as promising material for electronic substrate, thermal spreader, etc., has been made using deposition technology such as microwave plasma assisted CVD (MPACVD). However, high residual stress of diamond film induces bowing and throughthickness cracking in thick diamond wafer, and so hampers the economic fabrication of the wafer. Thus, to investigate the causes of bowing and cracking, the measurement of residual stress is very important. Of the residual stress, in particular, growth stress is one of the most important research topics because its effect on wafer-related problems is strong though the magnitude is relatively small. Quantification of growth stress is limited because of hightemperature plastic deformation of Si substrate. In this study, the initial growth stress without contribution of plastic deformation of substrate can be predicted through proposed beam bending stress model and numerical calculation in that plastic deformation of Si is considered. Finally, growth stress of film can be calculated from the apparent curvature of substrate measured just after deposition and the plastic curvature after diamond film removal.

It has been long recognized that the BDT in bulk materials may be associated with enhanced plastic energy dissipation. This can be achieved by either changing the state of stress (plane strain to plane stress) or by raising the test temperature (lowering the yield stress). The situation is somewhat different in thin films where the BDT can be achieved by increasing film thickness or perhaps, even in a limited temperature range, by raising the test temperature. To study the latter we use a superlayer technique with a 1 μm tungsten film on top of thin copper films bonded to SiO2/Si wafers. This involves indenting into the superlayer which stores and then releases large amounts of elastic energy into the thin film/substrate interface. Here, preliminary data on 500 nm thick Cu demonstrates more than an order of magnitude increase in fracture energy from about 10 to 200 J/m2 as the test temperature is raised from 20°C to 130°C. As the amount of plastic energy absorption would appear to be limited by film thickness, this relatively large value was unanticipated. This interfacial fracture energy translates to a stress intensity of 5 MPa-m1/2. In context of the highest possible nanocrystalline Cu yield strength, this still represents a plastic zone of nearly 30 μm. This illustrates the quandary associated with explaining such high apparent toughness values as one generally expects plasticity to be truncated by film thickness. Is this associated with:

–some artifact of assessing local stresses during nanoindentation at elevated temperature:

–extending the plastic zone in the direction of crack growth much further than the film thickness;

–a shielding mechanism from an organized dislocation array in a ductile film sandwiched between a brittle substrate and a higher yield strength superlayer;

–some plastic energy dissipation in the superlayer;

–or by enhanced mode II at higher temperatures?

A few of these will be addressed in some detail with a goal of narrowing the field of the most promising candidates.

In developing an adhesion test for a microelectronics fabrication facility there are many criteria which must be met. Some of these include (i) sample prep must be simple, (ii) deformations should be elastic so the problem can be easily modeled, (iii) mechanics are ideally analytical, and (iv) the test end-point must be unambiguous and easy to obtain. A testing method in the literature which meets many of these criteria is the modified edge liftoff test (MELT). Delamination is induced through the release of strain energy stored in an elastic superlayer which results from a large mismatch in CTE between the film and substrate. In this work we consider details of the energy release rate calculation, effects of plate bending, and initial flaw size.

Classical approaches to modeling fracture have proved successful in applications for which the highly deformed region near a crack tip is small compared to any other relevant dimensions in the structure. The classical theory relies on phenomenological criteria for material failure that lack a physics-based description of the fracture process itself. Small scale, thin film structures pose difficulties for analysis by these approaches because they contain complicated geometry and many interfaces within the fracture process zone itself. Moreover, plastic flow in metal layers is often severely constrained by the surrounding structure, causing the plastic dissipation part of the overall fracture energy consumed by debonding to be a strong function of geometry. Therefore, it can no longer be regarded as an intrinsic material property. To improve the fracture characterization of these structures, one must develop a physically sound methodology capable of separating the contribution of plastic flow, and other sources of dissipation, from the work of adhesion consumed at the crack tip. In this study, we investigate the parameters affecting energy dissipation by interfacial debonding in a multilayered structure. Interlayer decohesion is modeled using the Virtual Internal Bond constitutive model. We compare our predicted variations in the macroscopic fracture energy with experimental results for varying layer geometry. We also characterize the effect of variations in material properties and other experimental uncertainties in the resulting debonding behavior.

We have developed a molecular-dynamics technique for determining the adhesion strength of the interfaces between different materials. In this technique the extended Tersoff-type potential is applied to calculate the adhesive fracture energy defined as the difference between the total potential energy of the material-connected state and that of the material-separated state. The adhesion strength of metal/dielectric interfaces as well as metal/metal interfaces is discussed based on this fracture energy. We used this technique to determine the adhesion strength of the interfaces between ULSI-interconnect materials (Al and Cu) and diffusionbarrier materials (TiN and W). The calculated adhesive fracture energy shows that the adhesion strength increases in the order: Cu/TiN, Cu/W, Al/W, and Al/TiN. Because this result was confirmed by scratch testing on the film-laminated structure, this technique is considered to be effective for determining the adhesion strength.

The practical work of adhesion has been measured in thin aluminum films as a function of film thickness and residual stress. These films were sputter deposited onto thermally oxidized silicon wafers followed by sputter deposition of a one micron thick W superlayer. The superlayer deposition parameters were controlled to produce either a compressive residual stress of 1 GPa or a tensile residual stress of 100 MPa. Nanoindentation testing was then used to induce delamination and a mechanics based model for circular blister formation was used to determine practical works of adhesion. The resulting measured works of adhesion for all films between 100 nm and 1 μm thick was 30 J/m2 regardless of superlayer stress. However, films with the compressively stressed superlayers produced larger blisters than films with tensile stressed superlayers. In addition, these films were susceptible to radial cracking producing a high variability in average adhesion values.

A new experimental technique is developed to measure the fracture toughness of a hard metal thin film deposited on a soft substrate. A pre-crack was fabricated in the thin film by using the advanced focused ion beam (FIB) milling techniques. The crack extension force was generated by means of the indentation sinking-in effect. The effect creates a bending moment and tensile stress on the hard thin film near the indentation, which promotes crack growth. The amount of crack tip blunting prior to the critical failure was measured from the FIB cross-sectioned micrographs. By using the crack tip opening displacement model (CTOD), the fracture toughness of the thin film was calculated. The results show the nickel phosphorus (NiP) thin film fracture toughness is at least 15.0MPa√m. The finite element method (FEM) was used to understand the modes of mixity near the crack tip. The results indicate the crack tip modes of mixity are dominated by the Mode I opening, provided the indentation is sufficiently far from the pre-crack or the indentation depths is small when compared with the film thickness.