Improved spin readout fidelity in the presence of noise

High-fidelity spin readout in semiconducting qubits is most commonly performed using the technique pioneered by Elzerman \emph{et. al} (2004). The higher energy spin state is identified during a readout pulse by the change in current of a nearby charge sensor caused by tunneling off and back onto the dot. However, the technique requires high signal-to-noise and low device drift to maintain acceptable fidelity. Here, we present an alternative analysis technique for the same readout scheme, which is simple to implement and which lessens the demands on obtaining a high signal-to-noise ratio. Rather than setting a threshold in the maximum of the readout trace for state identification, our simulations treat the charge sensor current like noise and effectively differentiate between f^-2 noise (excited spin state) and f^-1 or f⁰ noise (ground spin states). This is done by utilizing the B₁ bias function [1] developed for noise identification in time and frequency standards. Initial simulations suggest that this readout method increases the amount of tolerable f^-1 noise where the fidelity is above 50% by a factor of 2 in amplitude (4 in power) and the amount of tolerable f⁰ noise by at least a factor of 2 in amplitude. We also benchmark our results against the optimal readout fidelity obtained from Bayesian inference methods.

[1] BARNES, J. A., et al. (1971). "Characterization of Frequency Stability." IEEE Transactions on Instrumentation and Measurement IM-20(2): 105.