Degenerate “valleys” in the silicon conduction band represent a potentially dangerous degree of freedom that can compete with “spin” as the dominant quantum variable in quantum dot qubits. Scaling up to large numbers of qubits therefore requires controlling the valley energy splitting. Unfortunately, this splitting depends strongly on atomistic properties of the interface, including interface sharpness, roughness and various types of disorder, which are unavoidable. In this talk, I will describe our current understanding of the wide variability observed in valley splitting measurements, even in devices fabricated 100 nm apart on the same chip. I will show that typical strategies for mitigating this variability are often only weakly effective, and that ubiquitous alloy disorder in the SiGe quantum well barriers plays a significant role in determining the valley splitting in many cases of interest. I will conclude by suggesting alternative strategies for controlling the valley splitting in future generations of devices.

Correspondence: friesen@physics.wisc.edu