ICMCTF2012 Session E2-2: Mechanical Properties and Adhesion

A MEMS tribogauge was used for on-chip and in-situ characterization of nano-tribological phenomena (stiction, friction, and wear). The measurements were made on the sidewall surfaces on the tribogauge at the fourth structural polysilicon layer in the device. The device consists of two orthogonally oriented comb-drive mechanisms that are used for both actuation and sensing functions. One actuator applies a normal load (F_n) to a contacting surface, while the other actuator induces a tangential load (F_T). A LabVIEW controlled AD7747 capacitance sensor is used to measure the position of the interacting surfaces. This data is converted into adhesive force information. The spatial resolution of the characterization apparatus is ±10nm.

Experiments were conducted with tribogauges with and without a self-assembled monolayer (SAM) coating. The SAM coatings being explored have either a fluorocarbon tail or a hydrocarbon tail group. The tribogauge with no SAM coating is UV/Ozone cleaned to remove organic contaminants, leaving behind –OH bonds on top of the MEMS surface (native oxide, SiO₂). The tribogauge characterization includes: measurement of baseline stiction force, static and dynamic coefficient of friction, and induced stiction force calculated after specific load cycles (F_induced). The UV/Ozone treated tribogauge was used to measure the baseline stiction force (F_plasma). Additional experiments showed that the induced stiction force increases in proportion to the increase in the number of load cycles, indicating erosion of the SAM coating and topographical changes to the interacting surfaces.

The local distribution of interface strength in a large scale integrated circuit (LSI) micro structure, which was not uniform nor the same as the average value obtained with conventional macro scale specimens, was investigated by applying a recently developed evaluation technique with a sub-micron range spatial resolution.

Three dimensionally stacked interconnect structures in LSIs frequently suffer from unexpected fracture, especially at the interfaces, due to stresses arisen in many steps of fabrication process. In spite of intensive efforts to avoid such damages, it still threatens the development process to push up the risk and thus the cost. The most likely reason for this frustrating situation could be that they are designed essentially on the basis of average strength data, obtained only from macro-scale specimens with blanket films of the composing materials by applying conventional techniques such as four point bending tests. For the case of micro-scale structures, there must be expected scatters of local strength, leading to weak spots from which cracks may extend. Therefore establishment of microscopic testing method was necessary to evaluate local strength distribution of interface, i.e., to map the strength of structural components with the same range of resolutions corresponding to the actual structure dimensions in LSI.

A dual beam system with a scanning electron microscope (SEM) and a focused ion beam (FIB) is further equipped with a nano-indenter for mechanical loading. In order to evaluate the interface adhesion energy between Cu damascene lines and cap layers, which is the weakest interface in such LSI interconnect systems, specimens were fabricated by FIB as blocks of the insulation layer with the dimensions down to sub-micron range. Fracture loads obtained by the experiment with the indenter under SEM observation were compared with the elastic-plastic interface crack extension simulations to determine the bonding energy, i.e. the the toughness of interface. Furthermore, not only the toughness but also the crystallographic orientations of Cu at the points of experiment, which was expected to be a cause of difference in the strength, was mapped by using an electron beam back scatter (EBSD) analyzer installed in the system. The correlation among the toughness, crystallographic structure and configuration of interconnects was investigated in detail on the basis of those distribution maps with a sub-micron resolution, aiming at establishing a possible design scheme to avoid unexpected fracture during the fabrication process of LSI.

In the first part of our presentation, we shall identify several serious inconsistencies and methodological mistakes in the presentation of Fischer-Cripps as far as we can identify them presently from the press release on his home page and from a manuscript available to us [1]. We shall further show that the hardness of the nc-TiN/a-Si₃N₄/TiSi₂ of about 80 to 100 GPa reported in our earlier papers (see [2] for a review) is supported by our indentation measurements as well as by scanning electron micrographs of the remaining indentations. We shall further briefly outline the issue of the reproducibility of the preparation of these unique materials. The emphasis of our presentation will be, however, on the future ways which will assure the reproducible preparation of nc-TiN/a-Si₃N₄ and related nanocomposites in dedicated laboratory apparatuses as well as in large-scale industrial coating equipment, which will assure achieving high hardness, high resistance against brittle fracture, high thermal stability and oxidation resistance.

[1a] www.ibisonline.com.au/IBIS_News/PressReleaseAug11.pdf

[1b] A.C. Fischer-Cripps et al., submitted, as quoted in Ref. [1a]

[2] S. Veprek et al., Thin Solid Films 476 (2006) 1

A reverse analysis of the Vickers indentation fracture toughness was carried out to derive a numerical expression to estimate the fracture resistance of Fe₂B layers. The boride layers were formed at the surface of AISI 1018 steels by developing the powder-pack boriding process at temperatures of 1193 and 1243 K with 4, 6 and 8 h of exposure for each temperature. From the set of experimental conditions of boriding process, Vickers indentations were performed with applied loads of 0.49, 0.98, 1.96, and 2.9 N at 15 and 30 microns from the surface of borided steels, respectively. The crack lengths created from the corners of the Vickers indentation prints were analyzed in the Palmqvist crack regime with the aid of dimensional analysis and finite element method (ABAQUS software program), considering the residual stress field generated by the indentation loads in the boride layer. The numerical expression of the fracture toughness of the Fe₂B layer was estimated by the superposition technique, where the fracture resistance is a function of the elastoplastic properties of the layer, and with different crack lengths superimposed according to the residual stress field caused by Vickers indentation loads.

The results of the fracture toughness of the Fe₂B layer estimated by the finite element method were compared with the values obtained by traditional Palmqvist crack models proposed in the literature.

In this study the mechanical behavior of FeB and Fe₂B layers formed at the surface of AISI D2 steels was estimated by Berkovich nanoindentation technique. The boriding of AISI D2 steels was developed by the powder-pack method at temperatures of 1223, 1273 and 1323 K in the range of exposure times of 3 - 7 h for each temperature. The mechanical characterization was performed related to the set of experimental parameters of boriding process and considering two procedures: first the nanoindentations were performed along the depth of surface layers with a constant load of 250 mN to determine the hardness gradient and the state of thermal residual stresses in the boride layers. In addition, applied loads in the range of 10 to 300 mN were carried out in the “pure” zone of the FeB layer at 10 microns from the surface, and in the “pure” zone of the Fe₂B layer (40 microns), respectively.

The results showed, for a constant load of 250 mN, that the state of thermal residual stresses and hardness of both FeB and Fe₂B layers are a function of the temperature and exposure time of the process, where the hardness decreases due to the presence of grain coarsening in the surface layers at a temperature of 1323 K with more than 5 h of exposure. Moreover, the presence of the indentation size effect (ISE) in the FeB and Fe₂B layers was verified in the range of applied loads, in which the apparent or real hardness was estimated by traditional models according to the boriding experimental parameters. Finally, the fracture resistance and brittleness of the boride layers was evaluated in the range of nanoidentation loads as a function of boriding temperatures and exposure times; the estimated values are in the range of 1.36-2.82 and 1.98-3.80 MPa m^1/2, with the presence of compressive stresses in the range of 162 to 1604 MPa for the FeB and Fe₂B layers, respectively.

This research was in an attempt to develop a new method without applying external stress for measuring fracture toughness of transition metal-nitride thin films. TiN thin film was selected to be the model material, owing to its well-characterized mechanical properties and appropriate elastic isotropy. At present, there has been no standard methodology or test procedure for assessing the fracture toughness of hard coatings. Previous literatures have proposed various approaches on the measurement of fracture toughness, which can be divided into two categories: stress based or energy based. However, those methods need to design special specimen geometry because of the requirement in producing valid pre-cracks, and thus the substrate effect cannot be eliminated. In addition, special stages are often needed to externally apply stress, which increases the difficulty of the test methods. TiN thin films deposited by PVD methods normally have high residual stress which can be controlled by adjusting deposition parameters and measured nondestructively. Instead of externally applying stress, the residual stress was utilized in the assessment of fracture toughness. From Griffith's criterion, energy stored in the film due to elastic mismatch strain can be released by the formation of cracks. The difference in stress states before and after crack initiated was used to evaluate the average energy release rate, from which fracture toughness can be calculated by fracture mechanics. This method involved residual stress measurement by laser curvature technique and elastic modulus measurement by nanoindentation according to ISO 14577-4:2007.

The Poisson’s ratio of single-crystal TiN was used. The results were compared with those obtained from other techniques and the strong and weak points of this method were discussed.

Recent progress in both in situ and ex situ small-scale mechanical testing methods has greatly improved our understanding of mechanical size effects in volumes from a few nanometers to a few microns. Besides the important results related to the effect of size on the strength of small structures, the ability to systematically measure the mechanical properties of small volumes through mechanical probing allows us to test samples that cannot easily be processed in bulk form, such as a thin film, a specific grain boundary or a single crystal. This talk will describe our recent results from in situ compression and tensile testing of metals with different starting defect densities and sizes to illuminate the origin of size-dependent yield strength behavior and fundamental deformation structures in nanoscale samples.