Optimization of Multi-Layered Thin Films for Minimizing Resistive Losses in Acoustic Resonators

Metal-to-metal contact resistance is a key limitation in microfabricated devices, especially when structures are operated at high frequency or integrated in resonant systems where resistive losses directly affect performance. In this project, two-layered thin-film structures were fabricated and electrically characterized in order to investigate the influence of different metal stacks, oxide removal, and geometry on the overall electrical response. Three Metal 1 stacks were studied: Al + W, Cr + W, and Ti + Mo, while Cr + Al was used as Metal 2. First, the temporal stability of the deposited films was monitored through weekly sheet resistance measurements. The results showed only minor variations over time, below 0.7%, indicating limited additional oxidation after deposition. However, significant differences were observed between the metal stacks, especially for Cr + W, whose sheet resistance was more than twice that of Al + W and Ti + Mo. This difference cannot be explained solely by the intrinsic resistivity of the adhesion layer and is likely related to deposition-dependent film properties such as microstructure, roughness, stress, density, or interface quality because this stack has been deposited by different machine than the others. DC measurements were then performed on dedicated test structures fabricated with different ion gun etching times before Metal 2 deposition. Both 100 s and 150 s etching treatments led to a decrease in measured resistance as the number of Metal 2 jumps increased, indicating that the native oxide layer was sufficiently removed and that additional jumps created parallel conduction paths. The experimental trends were in good agreement with a simplified resistance model, confirming that the main electrical contribution of the jumps was correctly captured. Finally, RF measurements from 100 kHz to 6.5 GHz were carried out on several geometries, including straight lines, serpentines, and long serpentines. The results showed that the RF response is not only governed by the DC sheet resistance, but also by the metal stack, physical layout, and parasitic reactive contributions. Cr + W structures generally showed higher impedance and a more capacitive behavior, while Ti + Mo structures showed a clearer transition toward inductive behavior at high frequency. Equivalent circuit fitting was used to further interpret the RF response. A complete RL-RLC model was first used as a baseline, and the reliability of the extracted parameters was evaluated using 95% confidence intervals together with the RMSE of both the real and imaginary parts of the impedance. The results showed that the appropriate circuit complexity depends on the structure and metal stack. For the investigated Cr + W structures, inductive parameters were not reliably identified and simplified R-RC models were sufficient. In contrast, Ti + Mo structures required inductive contributions, with the RL-RC or RL-RLC models providing a measurable improvement in fitting a