Date of Award


Degree Name

MS in Engineering - Materials Engineering


Materials Engineering


Richard Savage


Previous work done at Cal Poly has shown that thin-film nickel-titanium (NiTi) can be easily sputtered onto silicon wafers and annealed to create a crystallized shape memory alloy (SMA) film. Initial work on creating devices yielded cantilevers that were highly warped due to thin-film stress created during the sputtering process. The objective of this work was to create a thin-film NiTi SMA device that could be better characterized. A membrane was selected due to the simplicity of fabrication and testing which would also oppose the thin-film stress due to the increase in attachment points to the substrate.

Silicon wafers were etched through the majority of the thickness (~75%) creating square etch pits of varying sizes varying from 1294 µm to 4394 µm. The wafers were then sputtered with an approximate NiTi film of 5 µm followed by a thin chromium film. The chromium film would act as a diffusion barrier and prevent oxygen from diffusing into the NiTi and reacting with the titanium and forming titanium dioxide. These wafers were then annealed in a custom built vacuum annealing chamber at 550 °C for 1 hour with a pressure around 77 kPa. The chromium was then etched away followed by the remaining silicon. This left a thin membrane of shape memory NiTi which was packaged in order for characterization. The devices were glued to an aluminum substrate using polydimethylsiloxane (PDMS) and sealed with a small Tygon tube leading to the sealed chamber.

This packaged device was then able to be pressurized using a nitrogen tank and the resulting NiTi membrane deflection was measured using a profilometer. Due to the differences in elastic moduli of the room temperature phase (martensite) and the high temperature phase (austenite) a difference of deflection was expected. The austenite finish (Af) temperature of bulk NiTi films was found to be around 60 °C so the devices were tested at both room temperature and at 60 °C. After testing seven separate devices of varying sizes, a regression model was used to analyze the final data.

It was found that pressure, membrane size and theoretical versus actual deflection all affected the maximum deflection, but temperature did not. Higher pressures and larger membranes led to higher deflections as membrane deflection models from fundamental principles indicated. Some devices showed inferior performance when compared to the model due to incomplete silicon etching which caused lower deflection due to the much higher modulus of the remaining silicon. Thickness could also limit the amount of deflection measured with a thicker film leading to less deflection, but this is likely not the case due to the high uniformity of the sputtering system. Other devices showed superior performance over the model most likely due to either local delamination or lateral silicon etching. Both these would create a membrane that was larger than expected leading to a higher deflection. Unforutnaly, differential scanning calorimetry (DSC) analysis showed no shape memory behavior on a test wafer which was anneald at 550 ˚C for 1 hour. A design of experiments was conducted in order to find a heat treatment that would anneal the NiTi film and ensure that shape memory behavior could be obtained. An annealing at 650 °C for 1 hour showed a sharper and clearer Af phase transformation at around the target temperature of 60 °C. Annealing a full wafer at this temperature and time also showed that no delamination would occur which has also been linked to nonideal behavior of the NiTi membranes which has also been linked to meaningful behavior of the NiTi membranes.