N. R. Moody

Interfacial toughness measurements for thin films on substrates

Abstract There are more than 200 different methods for measuring adhesion, suggesting it to be material, geometry and even industry specific. This availability has exploded at least partly due to the arrival of dissimilar material interfaces and thin films and the ease with which microfabrication techniques apply to silicon technology. Having an eye toward those tests utilized for thin films, this paper reviews only a few of these techniques. The emphasis is on measuring thin film adhesion from the standpoint of fracture mechanics, when the film is mechanically or by other means removed from the substrate, and the amount of energy necessary for this process is calculated per unit area of the removed film. This tends to give values approaching the true work of adhesion at small thickness and greater values of the practical work of adhesion at larger thickness, all being in the 30–30,000 nm range. The resulting large range of toughnesses is shown to be dependent on the scale of plasticity achieved as controlled by film thickness, microstructure, chemistry and test temperature. While the tests reviewed largely address the measurement of elastic strain energy release rates, we also briefly address a few theoretical models which are specific to the resistance side of the delamination equation. The weight of the evidence suggests for ductile metallic films that the major extrinsic variables are film stress, extent of delamination, thickness and temperature while the major intrinsic ones are modulus, yield strength, the thermodynamic work of adhesion and one or more length scales. For some 25 film/substrate multilayers, with emphasis on Al, Au and Cu, the comparison of several theoretical models as to how the extrinsic and intrinsic variables intertwine is made.

Trapping of hydrogen to lattice defects in nickel

This paper addresses the energy associated with the trapping of hydrogen to defects in a nickel lattice. Several dislocations and grain boundaries which occur in nickel are studied. The dislocations include an edge, a screw, and a Lomer dislocation in the locked configuration, i.e. a Lomer-Cottrell lock (LCL). For both the edge and screw dislocations, the maximum trap site energy is approximately 0.1 eV occurring in the region where the lattice is in tension approximately 3-4 angstroms from the dislocation core. For the Lomer-Cottrell lock, the maximum binding energy is 0.33 eV and is located at the core of the a/6(110) dislocation. Several low-index coincident site lattice grain boundaries are investigated, specifically the Sigma 3(112), Sigma 9(221) and Sigma 11(113) tilt boundaries. The boundaries all show a maximum binding energy of approximately 0.25 eV at the tilt boundary. Relaxation of the boundary structures produces an asymmetric atomic structure for both the Sigma 3 and Sigma 9 boundaries and a symmetric structure for the Sigma 11 tilt boundary. The results of this study can be compared to recent experimental studies showing that the activation energy for hydrogen-initiated failure is approximately 0.3-0.4 eV in the Fe-based superalloy IN903. From the results of this comparison it can be concluded that the embrittlement process is likely associated with the trapping of hydrogen to grain boundaries and Lomer-Cottrell locks.

QUANTITATIVE ADHESION MEASURES OF MULTILAYER FILMS : PART I. INDENTATION MECHANICS

The mechanics for calculating the quantitative driving force of indentation-induced delamination of thin-film multilayers is presented. The solution is based on the mechanics developed by Marshall and Evans [D. B. Marshall and A. G. Evans, J. Appl. Phys. {bold 56}, 2632 (1984).] and extended to the general case of a multilayer by use of standard bending and thin-plate analyses. Presented and discussed are the specific solutions for the bilayer case that show that in the limit of zero thickness of either layer, the solution converges to the single-layer case. In the range of finite thickness, the presence of the superlayer increases the driving force relative to that possible for the original film alone and can be optimized to the experimental situation by proper choice of thickness, elastic constants, and residual stress. The companion paper {open_quotes}Quantitative adhesion measures of multilayer films: Part II. Indentation of W/Cu, W/W, Cr/W{close_quotes} discusses experimental results with copper, tungsten, and chromium thin films. {copyright} {ital 1999 Materials Research Society.}

Quantitative adhesion measures of multilayer films: Part II. Indentation of W/Cu, W/W, Cr/W

Sputtered copper and tungsten thin films both with and without tungsten and chromium superlayers were tested by using nanoindentation probing to initiate and drive delamination. The adhesion energies of the films were calculated from the induced delaminations using the analysis presented in {open_quotes}Quantitative adhesion measures of multilayer films: Part I. Indentation mechanics.{close_quotes} Copper films ranging in thickness from 150 to 1500 nm in the as-sputtered condition had measured adhesion energies ranging from 0.2 to 2 J/m{sup 2}, commensurate with the thermodynamic work of adhesion. Tungsten films ranging in thickness from 500 to 1000 nm in the as-sputtered condition had measured adhesion energies ranging from 5 to 15 J/m{sup 2}. The superlayer was shown to induce radial cracking when under residual tension, resulting in underestimation of the adhesion energy when the film was well adhered. Under conditions of weak adherence or residual compression, the superlayer provided an excellent means to induce a delamination and allowed an accurate and reasonably precise quantitative measure of thin film adhesion. {copyright} {ital 1999 Materials Research Society.}

Substrate effects on indentation plastic zone development in thin soft films

Plastic zone evolution in Al–2 wt% Si metal films on silicon and sapphire substrateswas studied using nanoindentation and atomic force microscopy (AFM). AFM wasused to measure the extent of plastic pileup, which is a measure of the plastic zoneradius in the film. It was found that the plastic zone size develops in a self-similarfashion with increasing indenter penetration when normalized by the contact radius,regardless of film hardness or underlying substrate properties. This behavior was usedto develop a hardness model that uses the extent of the plastic zone radius to calculatea core region within the indenter contact that is subject to an elevated contact pressure.AFM measurements also indicated that as film thickness decreases, constraint imposedby the indenter and substrate traps the film thereby reducing the pileup volume.I. INTRODUCTIONThe popularity of nanoindentation is due in large partto its ability to probe the mechanical properties of ma-terials in a nondestructive fashion without extensivesample preparation. However, it is often difficult tomeasure film properties independent of the substrateproperties. Several solutions to this problem have beenproposed with varying degrees of success, the simplestbeing the “10% rule,” by which it is proposed that thefilm properties can be measured for indentation depthsless than 10% of the total film thickness.However, this “rule of thumb” has several deficien-cies. The rule is too restrictive for soft coatings onhard substrates;

Nanoindentation of Au and Pt/Cu thin films at elevated temperatures

This paper describes the nanoindentation technique for measuring sputter-deposited Au and Cu thin films’ mechanical properties at elevated temperatures up to 130 °C. A thin, 5-nm Pt layer was deposited onto the Cu film to prevent its oxidation during testing. Nanoindentation was then used to measure elastic modulus and hardness as a function of temperature. These tests showed that elastic modulus and hardness decreased as the test temperature increased from 20 to 130 °C. Cu films exhibited higher hardness values compared to Au, a finding that is explained by the nanocrystalline structure of the film. Hardness was converted to the yield stress using both the Tabor relationship and the inverse method (based on the Johnson cavity model). The thermal component of the yield-stress dependence followed a second-order polynomial in the temperature range tested for Au and Pt/Cu films. The decrease in yield stress at elevated temperatures accounts for the increased interfacial toughness of Cu thin films.

Recent developments in thin film adhesion measurement

Interfacial fracture energies of thin films may be calculated using many different techniques. Nanoindentation and stressed overlayers are by far the most common and more reliable of the testing techniques. They depend on mechanics-based models to calculate the interfacial fracture energy of an interface using only the site specific material properties and the dimensions of the delaminated region, either in spontaneous buckle or indentation-induced blister form. This study will focus on four adhesion measurement techniques: spontaneous buckles, stressed overlayer-induced buckles, and nanoindentation-induced blisters with and without stressed overlayers, to demonstrate that the techniques will produce similar results for the measurement of adhesion energy. Films of tungsten (W), platinum (Pt), and titanium (Ti) on SiO/sub 2/ (amorphous glass) substrates are examined and values of interfacial fracture energies reported. Results of interfacial fracture energy calculated from spontaneous buckles and indentation-induced blisters compare well to one another and values are reported for the aforementioned films.

Adhesion and acoustic emission analysis of failures in nitride films with a metal interlayer

Abstract Interfacial fracture has been induced between a tantalum nitride film with an aluminum interlayer on a sapphire substrate using nanoindentation. To identify failures for which a model calculation is valid a commercial acoustic emission sensor has been used to study the details of the failure event. The interfacial fracture energy of the system with an aluminum interlayer under the loading conditions at the crack tip is approximately 8 J/m2. Within narrow bounds, this toughness value is reproducible using three different theoretical approaches. The acoustic emission signal is used to determine a lower bound interfacial crack velocity of 5 m/s. The majority of the failure occurs at the aluminum-sapphire interface, suggesting that the fracture energy and crack velocity determined are related to the toughness of this interface and not the nitride-aluminum interface.

Determining fracture toughness of vitreous silica glass

This study demonstrates that the Chevron-notched short rod (CNSR) test technique is well suited for accurate determination of the fracture toughness for vitreous silica. A manifestation of the accuracy of this fracture toughness test technique is evident by low standard deviation of values observed from statistically significant sample sizes. Findings of this study suggest that the CNSR test method could be useful for determination of subtle changes in fracture toughness of glass caused by aggressive service environments, such as, hydrogen and/or ionizing radiation. Fracture surface morphological features of vitreous silica were revealed by atomic force microscope imaging. AFM enabled the examination of the fracture surface morphology locally with nanometer scale resolution. Short range, nanofracture mechanisms can be discerned for glass by atomic force microscopy.

The influence of inclusion spacing and microstructure on the

An approach to the ductile fracture of ultra high strength steels has been evaluated. According to this approach the critical crack tip opening displaceent, δIC, will scale withX0(RV/RI)|R0.X0 is an average inclusion spacing and (RV/RI)|R0 is the void radius divided by the radius of the inclusion nucleating the void evaluated at the average inclusion size. AF1410 was selected to test this approach because it has exceptionally high fracture toughness on aging at 510 °C and because its toughness varies markedly with aging temperature. The results from this and earlier work showed a linear relationship exists between δIC andX0(RV/RI)|R0 for values of δIC ranging from about 8 μm to 60 μm. The values of (RV/RI)|R0 for AF1410 aged at 425 °C and 510 °C differed by a factor of two. Because the yield strengths and work hardening exponents were similar for the two aging temperatures, the difference in (RV/RI)|R0 was attributed to changes in microstructure. The microstructures at the two aging temperatures differed in type and size of intra-lath carbides. Both the microstructure and inclusion spacing were found to influence the fracture toughness of AF1410. The average inclusion spacing for this heat of AF1410 was the largest of the six steels evaluated with this model. The model suggests that the large inclusion spacing is critical to the high toughness of AF1410 on aging at 510 °C. This large inclusion spacing results not from an unusually low inclusion volume fraction but from a large average inclusion size.