Development of ultra-high quality mechanical oscillators

In this thesis, ultra-coherent micro mechanical oscillators based on thin-film stoichiometric silicon nitride were developed and experimentally tested. This work gathers on all existing literature in the field. This includes theory work where a stochastic model was developed for predicting how vibrations from sub-MHz oscillators couple to the surrounding substrate and are subsequently lost. While this model hasn’t been systematically tested experimentally yet, early results look promising and succeeds in predicting at least part of the high spread in results.
Two design approaches were developed and tested in an effort to fabricate the most coherent device possible. The first approach uses topology optimization to enhance the coherence of the fundamental mode of a resonator, which is to be used in quantum optomechanics. Numerical evaluations showed a Qf product enhancement above a factor of two compared to a reference design. Experimental results showed the best devices for each design are limited by intrinsic losses. However, the large spread in the results indicated most of them were heavily affected by phonon tunneling losses, which highlighted flaws in the topology optimization implementation. The best results matched the numerical predictions making them the best fundamental mode resonators in existence with Qf = 2.8 × 10¹³.
The second design approach developed called density phononics is based on phononic crystals defined by modulating the effective density of a material. This novel membrane design was achieved by fabricating microscopic pillars. Early results showed fabrication is non-trivial, but it is expected this can be optimized greatly by changing the process steps. Nonetheless, the design approach was validated and the early batch resulted in 1.4 MHz resonators with Qf = 8 × 10¹⁴ at room temperature, which is a new record for this type of device. Furthermore, models predict quality factors above 10⁹ should be possible with continued refinement of fabrication.
Finally, two designs for both reducing and enhancing the tensile stress of two-dimensional membranes was proposed. Numerical evaluations showed the former can reduce the tensile stress by multiple orders of magnitude. This has applications in enhancing the sensitivity of thermal sensing and derived sensing schemes. The second design used for stress enhancement results in enhancements up to a factor of three. When combined with density phononics, it is expected that room temperature resonators with Qf products above 10¹⁶ should be possible.