Experimental understanding of noise sources in Silicon MOS devices at cryogenic temperatures for spin qubit applications

Quantum computing embodies the promise of the future, and has the potential to revolutionize our world with its significantly accelerated computational capabilities. This nascent technology is not meant to replace today’s computers but rather solve problems intractable by current computation methods. Analogous to the classical computer the quantum computer is built using basic blocks called quantum bits and performs algorithms using quantum gates. Amongst the many platforms for realizing quantum bits is the spin of an electron in silicon, which offer many advantages such as long coherence times and compatibility with existing industrial fabrication infrastructure. However, spin quantum bits are limited by charge noise which has similarly plagued the semiconductor industry for decades.

In this work, we detail the advancement made towards understanding the physical origins of charge noise in semiconductor gate stacks. The work is centered around the characterization of the silicon metal-oxide semiconductor gate stacks on the macroscopic and microscopic level and concluded with the characterization of spin qubits using the same gate stacks. We begin this work by detailing the fabrication methods necessary for this study. The fabrication platform, which is part of imec’s quantum computing program, combines optical and e-beam lithography for high flexibility and throughput. The process is designed to be modular to allow the exploration of different spin platforms without the necessary additional overhead cost.

The experiential work detailed in this thesis is conducted on the same set of devices. We begin by examining the transport characteristics of silicon two dimensional electron gases which is a widely used metric to characterize the quality of devices. We utilize Hallbar measurements to examine the mobility limiting mechanisms of different metal-oxide semiconductor gate stacks. By changing the type of metal gate used and the oxide thickness we are able to identify certain types of physical defects that affect the device quality. Additionally, we investigate the temperature dependence of the mobility which provides further insight to the nature of the physical defects in the gate stack. Next, we characterize the quantum dot current spectroscopy noise which is a widely used method to investigate low frequency noise. We find direct correlation between transport characteristics and the magnitude of the charge noise, suggesting a common underlying physical origin between them. Additionally, we conduct a statistical analysis of the charge noise on one of the gate stacks to investigate the nature of the defects in metal-oxide semiconductor stacks. We find that dipole type defects best represent the charge noise that is experimentally observed in quantum dots. Finally, we characterize an electron spin resonance type spin qubit to experimentally confirm the correlation between transport characteristics and qubit performance.

The statistical analysis, characterization methods and results presented in this work formed a closed-loop process to enhance silicon metal-oxide semiconductor gate stacks for quantum information applications, providing novel insights into the origin of charge noise and paving the way for large-scale spin qubit processor integration using a full CMOS process.