Quantifying Strength Size Effects in Polycrystalline Silicon and Aluminum with On-chip Test Platforms

Mechanical strength of components increases as their size decreases. Optimum design of reliable systems at the micro- and nanoscales will account for such size-dependent strength. However, the dependencies of strength on size in brittle ceramic and ductile metal thin films are not well known because of limited data. Therefore, in this thesis, high throughput platforms were designed, fabricated and tested. The strength size effect for polycrystalline silicon (polysilicon) and polycrystalline aluminum thin films has been investigated by on-chip testing techniques. Polysilicon structures over a size range of 100 increase in characteristic strength from 2.7 GPa to 4.2 GPa. A Weibull function alone was insufficient to predict the strength size effect for polysilicon. After taking into account the non-uniform stress distribution in the smallest specimens, both the characteristic strength and the full strength probability distribution function was well predicted. Also, a Monte Carlo technique was developed to predict strength size effects and to assess the minimum number of tests required for accurate characterization of strength distribution. Aluminum thin film structures over a size range of 6.5 increase in average yield strength from 140 MPa to 300 MPa. Unlike macroscale specimens, these samples also exhibit significant scatter in strength. The ratio of the standard deviation in yield strength to the average yield strength is 0.06 for 1 μm thick samples, and more than doubles to 0.13 for 0.6 μm thick samples. High throughput test platforms provide an important method to assess strength and strength distribution data at the micro- and nanoscales.