Novel techniques to measure the intrinsic tensile strength and intrinsic toughness of interfaces have been developed. Notable among these is a laser spallation experiment for interface strength measurement. The latter is a synthesis of two inventions, (1) nanosecond rise-time stress pulses with tailless profiles, and (2) a wide-angle interferometer for non-specular surfaces. Both these inventions are contained in United States Patent No 5,438,402: System and Method for Measuring The Interface Tensile Strength of Planar Interfaces.

Using these techniques, and high resolution transmission electron microscopy, the atomic-level structural and chemical basis for such measurements have been uncovered in several engineering and model interface systems.

Interface strength degradation as a function of relaxed interfacial structure has been established in polyimide/Si systems. The relaxation occurs on account of moisture segregation when exposed to humidity/time/temperature variables. These strength degradation charts now allow reliable prediction of electronic devices when used in conjunction with established coupled mass, heat, and mechanical stress analyses simulations. This, plus the extension of the laser spallation technique for in-situ adhesion measurement in geometrically heterogeneous electronic devices and packages have led to technology transfer to Intel Corporation (see below).

Discovery of Glass-Modified Stress Waves with Rarefaction Shocks

Observations of laser-generated stress wave profiles with rarefaction shocks (almost discontinuous or zero post-peak decay times) in glasses have been uncovered. Figure 1 shows a series of stress pulse profiles with increasing laser fluence, measured in a soda lime glass. At low stress amplitudes, the profiles are similar to that in Si and other materials, having a finite (1-2 ns) rise-time and gradual post-peak decay (14-18 ns). However, as the stress pulse amplitude exceeds a certain threshold, the rise-time of the stress wave gets longer but the post-peak stress profile starts to decay rather quickly. Ultimately a profile is attained with the post-peak stress dropping instantaneously (the drop time is within the resolution limit of our instrumentation), much like a “rarefaction shock.” The modification in the stress wave profile due to glass compared with that due to Si can be appreciated by viewing Fig 3a.

Figure 1. Varying stress pulses in soda lime glass

Figure 2 shows the pulse profiles measured in Pyrex, soda lime, quartz, and borosilicate glasses. The magnitude of the above effect was found to vary from glass to glass, but all showed the formation of the rarefaction shock.

Figure 2. Pulse profiles in various substrates

The rarefaction shock stems from an initially increased compressibility of glass under increasing pressures. The decrease in the wave speed is evident in the increase in the rise-time of the stress wave with its amplitude (Fig. 1). Thus, the formation of rarefaction shock can be thought to occur due to overcrowding of the faster moving post-peak wave profile into the significantly decelerating pulse front. The initial ramp-like behavior has been attributed to the nonlinear elastic response in glasses.

The technological importance of these pulses in measuring the tensile strength of very thin film interfaces has been demonstrated using the laser spallation experiment. Because of the rarefaction shock, glass-modified waves allow generation of substantially higher interfacial tensile stress amplitudes compared with those with finite post-peak decay profiles. This is demonstrated in Figure 3. Figure 3a shows the failure inside the Si in a Cu(1400nm)/TiN(70nm)/Si system when no glass was used. Figure 3b shows a high magnification view of failure caused by a wave generated inside the glass and transferred to the backside of the Si substrate. The arrow shows the failure to be at the Cu/TiN interface with no fracture inside the Si. An interface tensile strength value of 2.62 GPa was calculated. This is a rather high value, which was not possible to attain using the basic spallation setup for the film thicknesses tested. This is demonstrated in Figures 3 (c) and (d). Figure 3 (c) shows the measured free surface velocity profiles in the borosilicate glass and Si substrates corresponding to the failures shown in Fig. 3 (a) and (b). The interface tensile stress history corresponding to each profile is shown in Fig. 3 (d). The dramatic increase in the interface tensile stress due to the glass-modified wave is evident.