Advancement of interconnect technology has imposed significant challenge on interface characterization and reliability for blurred interfaces between layers. There is a need for material properties and these miniaturized length scales and assessment of reliability; including the intrinsic film fracture toughness and the interfacial fracture toughness. The nano-meter range of film thicknesses currently employed, impose significant challenges on evaluating these physical quantities and thereby impose significant challenge on the design cycle.

In this study we attempted to use a combined nano-indentation and nano-acoustic emission to qualitatively and quantitatively characterize the failure modes in ultra-thin blanket films on Si substrates or stakes of different characteristics. We have performed and analyzed an exhaustive group of testes that cove many diverge combination of film-substrate combination, provided by both Intel and IBM. When the force-indentation depth curve shows excursion, a direct measure of the total energy release rate is estimated. The collected acoustic emission signal is then used to partition the total energy into two segments, one associated with the cohesive fracture toughness of the film and the other is for the adhesive fracture toughness of the interface. The acoustic emission signal is analyzed in both the time and frequency domain to achieve such energy division. In particular, the signal time domain analysis for signal skewness, time of arrival and total energy content are employed with the proper signal to noise ratio. In the frequency domain, an expansive group of acoustic emission signals are utilized to construct the details of the power spectral density. A bank of band-pass filters are designed to sort the individual signals to those associated with adhesive interlayer cracking, cohesive channel cracking, or other system induced noise. The attenuation time and the energy content within each spectral frequency were the key elements for this sorting.

In the case wherein no excursion were present on the load-displacement curve, atypical case for adhesive failure only, the final indentation load along with the size of the blister were used to characterize the cohesive fracture energy. In the few cases wherein the film toughness was much higher than the substrate toughness, the unloading segment of the force-indentation depth curve showed a clear demarcation with a much lower slope. Such demarcation arises from the film buckling induced delamination. The delaminated buckled film unloads faster than the indentation process zone within the substrate and thereby become affected by the stored elastic strain energy within the film only. A simple model, utilizing the area under the force-indentation depth curve is derived to calculate the interfacial energy release rate of the film-substrate system. The model assessment is in good agreement with estimates done by four-point-bend testing on the same material system.

The developed experimental protocol may become useful in identifying the prominent failure mechanisms for quick screening of film-substrate, as well as in providing some quantitative measures of the adhesive and cohesive fracture toughness.