UC Berkeley

Current quantum computing platforms suffer from error rates that make it challenging to show a meaningful quantum advantage. The noise driving those errors can be broken down into two categories, those intrinsic to the quantum system used and those originating from external devices used to interface with the system. This dissertation attempts to address both cases, specifically for the trapped-ion platform.First, we attempt to tackle the issue of electric-field noise originating from the surfaces of our ion trap electrodes. It is widely known that such noise leads to high levels of motional heating and reduces the fidelity of gate operations. Previous demonstrations have shown that by changing the morphology of the surface, surface-related heating was significantly reduced. In this thesis, we present the design, construction, and implementation of a trapped-ion system that allows for in-situ argon milling of a trap surface while maintaining the optical access necessary to perform complex quantum operations. We demonstrate several rounds of milling on two ion traps and characterize the mills performance. Additionally, we discuss the impact from milling, as well as potential next steps.Second, we propose and demonstrate a method to empirically probe the spectral sensitivity of a quantum system to external noise on its control signals. While approaches for dealing with intrinsic noise are system dependent, solutions for dealing with noise on external devices are more universal. The complications come not from knowing how to handle the noise, but rather from figuring out which control signals and spectral ranges are a priority. To demonstrate this approach, we consider phase and amplitude noise on the laser fields driving a two-ion quantum gate. By recording the error rate as a function of the injected modulation frequency and power, we identify the most critical spectral ranges for achieving high-fidelity operations. Further, we show that it is not always sufficient to only consider the first-order spectral response and illustrate how to treat second-order contributions. Finally, we estimate the native error rate contributions from both types of laser noise, even though they are not the leading contributors to the overall error rate.