ICMCTF2014 Session E2-1: Mechanical Properties and Adhesion

We have carried out an experimental study of buckle formation on a model system: the film is a Molybdenum overlayer with thickness ranging between 50 and 300 nm, while compressive biaxial stress (up to nearly 3 Gpa) is adjusted through the deposition conditions [1]. In addition a thin silver film (usually 10 nm) is deposited directly on the substrate, below the Mo layer: in this way a low and reproducible adhesion develops. With this system, we have obtained a wide range of buckling conditions. Beyond telephone cords we have also met with less ubiquitous morphologies, such as branching buckles. We have systematically explored the phase diagram of morphologies as a function of stress and thickness. To help understand these results, we have also modeled the formation of the buckles. Geometrical non-linearities of film buckling are taken into account within a Finite Element model, and film adhesion is included as a cohesive zone. We show that consistent predictions of the buckle morphologies are obtained provided the mode mixity dependence of interfacial toughness is included [2]. We have also demonstrated numerically that the period of the telephone cord buckle is directly connected to the mode I critical energy release rate GIc, and give some experimental evidence that in practice thin film adhesion energies can be measured quite accurately based on this observation [3].

[1] “Stress tuning in sputter-deposited MoOx films”, FAOU J.-Y. et al. Thin Solid Films 527 (2013) 222-226

[2] “How does adhesion induce the formation of telephone cord buckles ?” FAOU J.-Y et al. Phys. Rev. Lett. 108 (2012) 116102

[3] "Thin film adhesion energy measurement from telephone cord buckle wavelength" FAOU J.-Y et al., in preparation.

Interface delaminations and coating cracks are the major failure modes of diamond-coated carbide tools in machining. To study any influence to each other between the two failure modes, micro-scratch testing on diamond-coated carbide tools was conducted with normal and tangential forces as well as acoustic emission signals recorded to detect coating delaminations and crack initiations. Scratched samples were observed by optical microscopy after testing to determine the associated critical load of delaminations and cracking initiations. In addition, a 3D finite element model was developed to simulate the scratch process using cohesive elements and the extended finite element method (XFEM) for delamination and coating fracture behaviors. The cohesive elements are based on a bilinear traction-separation cohesive zone model. XFEM is applied to model crack behavior in diamond coatings with a damage criterion of maximum principal stress.

The results indicate that the critical load for coating crack initiations increases almost linearly with the increased coating thickness, while decreases linearly with the increase of coating elastic modulus. Moreover, the interface fracture energy has a negligible effect on the critical load for coating crack initiations, so does coating cracking on the critical load for coating delaminations, indicating the two failure modes are almost uncoupled. From the simulations and experiments, it is estimated that the coating fracture energy of the samples tested in this research is from 140 to 252 J/m², and the interface fracture energy is from 87 to 192 J/m².

Copper wiring systems of semiconductor devices has a risk of mechanical fracture along with the trend of further integration and miniaturization, because of many weak interfaces stacked to compose multilayered copper/dielectric systems [1]. Local adhesion strength of those interfaces would vary significantly, depending on material structures [2]. In fact, a recent study indicated that the fracture energy of the interface between a Cu line and a barrier layer was affected by both barrier composition and Cu electroplating purity, and that impurities were segregated to grain boundaries and triple points [3]. In addition, it was also reported that local material structure of copper influences the local adhesion strength [4]. These investigations suggest that grain structure of Cu line play a significant role to fluctuate local strength. For the case of micro-scale systems, the distribution of the strength is essential to statistically estimate the fracture risk of the systems because there must be the expected scatter of local strength, leading to weak spots from which cracks may extend.

In this paper, stochastic distribution of local adhesion strength at copper/dielectric interface was estimated by the new technique developed by the authors, which utilizes the FIB-SEM system with a nano-indenter [4]. Specimens including the copper/dielectric interface which is well known as the weakest interface in semiconductor devices, were fabricated by FIB as blocks of dielectric layer. Fracture loads obtained by the experiment with the nano-indenter under SEM observation were compared with interface crack extension simulation to determine the strength [5]. Furthermore, not only the strength but also the crystallographic information of copper line was obtained by using an electron back-scattering diffraction (EBSD) analysis. The correlation between the strength distribution and the grain structure was discussed, especially the impact of the grain boundary density on the scatter range of the evaluated strength. The result of examination above suggests that grain boundary weakens local adhesion strength and is the key factor to arise large scatter range of the strength in micro-scale systems.

[1] C.-C. Lee et al., IEEE Trans. Adv. Packag. 32 (2009) 53.

[2] B. Wunderle et al., Microsys. Technol. 15 (2009) 799.

[3] R. P. Birringer et al., J. Appl. Phys. 110 (2011) 0044312.

[4] S. Kamiya et al., Surf. Coatings Technol., 215 (2013) 280.

[5] S. Kamiya et al., Journal of Materials Research 25-10 (2010) 1917.

Nanoscale metallic multilayer systems (NMM) attract attention due to their unique mechanical properties and, particularly, high tolerance to radiation damage. Specific interfaces can accommodate significant amount of implanted He atoms and prevent formation of helium bubbles, which compromise mechanical properties. On the order of a few to tens of nanometers, it is possible to find yield strength values above 1 GPa and Young’s modulus values around 100-150 GPa.

Magnetron sputtered nanoscale metallic multilayer systems of Cu/W and Zr/Nb with three different layer thicknesses (5/5, 15/15 and 30/30) and a total thickness 1 micron were deposited on silicon wafer. Structure was studied by XRD and by TEM investigation of multilayer cross-section. In case of Cu/W multilayer, the grain size was significantly lower for tungsten grains compared to that of copper; in fact, grain size of copper was often higher that the thickness of corresponding layer. Kurdjumov-Sachs interface <110> Cu // <111> W was the major interface; however, there were differences between W-Cu and Cu-W interfaces. The latter was represented in some localized spots by an amorphous, 2-4 nm thick layer, whereas the former was fully crystalline. Zr/Nb multilayer showed Pitsch-Schrader orientation relationship: <110> Zr// <110> Nb; moreover, XRD spectra suggested superlattice. Nanoindentation tests showed high hardness and Young’s modulus dependence on layer thicknesses for both studied systems. Maximum values were around (5; 108 GPa) and (6; 130 GPa) for Zr/Nb (15/15) and Cu/W (15/15), respectively. We correlated mechanical response of the multilayers with the microstructure. Moreover, fracture toughness of these multilayers was correlated with the plastic deformation volume below indenter tip, which was estimated using different approaches.

The coating/substrate interface is a common source of mechanical failure in many coated materials, e.g. tools, microelectronic devices or architectural glazing. Therefore, the characterization and optimization of the interface toughness is an important topic in thin films technology. In the current work two different experimental setups are explored to assess the interface fracture toughness of thin films: a macroscopic double cantilever beam and a microcantilever fracture toughness test. The double cantilever beam setup was combined with post-mortem glow discharge optical emission spectroscopy to identify the crack path within a multilayer stack and the micro-cantilever tests were performed inside an electron microscope to follow the crack path in-situ. Advantages and disadvantages of both techniques in terms of practicability, notching strategies etc. will be presented.

A study of the interface toughness of two industrially applied coating systems, one multilayer structure deposited by sputtering and one coating produced by electrodeposition, both containing functional thin films, will also be presented. Thin nanostructured oxide films are introduced into these coatings and the influence of these layers, their structure, thickness and roughness on the interface toughness are examined. The results of the fracture mechanics experiments indicate that the introduction of oxide interlayers has a positive effect on the adhesion of the investigated systems. Furthermore, it is shown how the variations in oxide film properties can be used to optimize the interface and coating properties.

There are a large number of factors involved in wear processes (e.g. mechanical, physical and chemical properties, surface topography, loading), making the precise theoretical and quantitative approach of wear a challenge even for “simple” tribo-systems. Many of these factors are hard to measure, may vary with time and space, and there is not yet a general theory available of how to link the basic properties with the tribological response. Several well established testing methods (e.g. pin-on-disk, fretting and impact tests) have been widely used to study treated surfaces and coatings on various substrates. However, many of these existing techniques have limitations in their ability to characterize materials, since they mainly focus on a single mode of loading and wear (e.g. only impact or sliding).

In this study, the design and evaluation of a new Dynamic Impact and Sliding Test (DIST)/Dynamic Impact and Rolling Test (DIRT) for the tribo-mechanical evaluation of surfaces under complex loading conditions is presented, where the surfaces are simultaneously subjected to impact and sliding or rolling loads. Such modes exist in many critical applications, from biomedical (e.g. hip/knee implants) to automotive applications (e.g. diesel injectors, engine valves, cam shafts), in cutting tools, general machine parts and systems, etc. Expected benefits include the time and cost effective evaluation of various surfaces and the better understanding of their peculiarities under such multi-mode loading conditions. Some of the unique design characteristics of the DIST/DIRT (e.g. combined impact and sliding or rolling testing; wear area in a single point; pre-setting of desired maximum wear depth possible; evaluation of materials’ properties and behavior in a single run) and the evaluation method are presented and discussed.