Multi-level charge fluctuations in a Si/SiGe double quantum dot device

TL;DR

This work addresses discrete multi-level charge fluctuations that contribute to drift in Si/SiGe quantum-dot qubits. It combines KDE/KLD-based drift removal with factorial hidden Markov models to extract transition rates from time-domain charge-noise data across multiple device configurations. A detailed-balance phenomenological model yields gate-lever-arm estimates linking two primary fluctuators to control electrodes, while model comparisons reveal a conditional coupling between fluctuators in a high-RP1 regime. The findings enhance understanding of charge-noise sources, enable triangulation of fluctuator locations, and inform design and operating strategies to mitigate decoherence in semiconductor spin qubits, with $μ$eV/mV-scale lever arms and region-dependent model dominance.

Abstract

Discrete charge fluctuations, routinely observed in semiconductor quantum dot devices, may contribute significantly to device drift and errors resulting from qubit miscalibration. Understanding the nature and origins of these discrete charge fluctuations may provide insights into material improvements or means of mitigating charge noise in semiconductor quantum dot devices. In this work, we measure multi-level charge fluctuations present in a Si/SiGe double quantum dot device over a range of device operating voltages and temperatures. To characterize the parameter-dependent dynamics of the underlying fluctuating degrees of freedom, we perform a detailed analysis of the measured noise timeseries. We perform algorithmically assisted drift detection and change point detection to detrend the data and remove a slow fluctuator component, as a preprocessing step. We perform model comparison on the post-processed time series between different $n$-level fluctuator ($n$LF) factorial hidden Markov models (FHMMs), finding that although at most sweep values the independent pair of 2LFs model would be preferred, in a particular region of voltage space the 4LF model outperforms the other models, indicating a conditional rate dependence between the two fluctuators. By tracking fluctuator transition rates, biases, and weights over a range of different device configurations, we estimate gate voltage and conductivity sensitivity. In particular, we fit a phenomenological, detailed balance model to the extracted independent 2LFs rate data, yielding lever arm estimates in the range of $-2 μ$eV/mV up to $4 μ$eV/mV between the two 2LFs and nearby gate electrodes. We expect that these characterization results may aid in subsequent spatial triangulation of the charge fluctuators.

Figures (13)

• Figure 1: Device. False-color scanning electron micrograph of a device that is nominally identical to the one measured here. Screening gates, accumulation gates, plunger gates, and tunneling gates are in gray, green, red, and blue, respectively. The sensor quantum dot is tuned under plunger gate RP1. The white scale bar is 100 nm.
• Figure 2: Drift and change point removal. Figure (a) shows an example of change point detection using the KLD method described in the main text. A peak finding algorithm identifies large step changes indicated by the dashed lines. Figure (b) shows an example of change point detection failure, with dashed red lines indicating change points found by the algorithm. Figure (c) shows our manual pruning and adjustment corrections to (b), where we have used solid gray lines to mark the start of a higher-signal segment and dashed red lines to indicate the end of a higher-signal segment, which is also the start of a subsequent lower-signal segment. Figures (d), (e), and (f) show the process of detrending and removing the slow fluctuator going from (d) $\rightarrow$ (e) $\rightarrow$ (f).
• Figure 3: Models. We show the breakdown of fitting different models to the 3 prominent observed levels in the data (left). The levels for the 3LF and 4LF models are set as shown. For the 2x2LF model, we show how the levels add indicating the final levels on the far right.
• Figure 4: RP1 model comparison and FHMM rate fits for the 4LF model. In (a), we show the model comparisons as a function of RP1, indicating that the 4LF model becomes better ($e > 2$) at the largest RP1 value. In (b), we display specific 4LF FHMM fit transition rate elements, $\Gamma_{01}$ and $\Gamma_{23}$, for each sweep value. These elements should be the same under a 2x2LF model. We can see a clear deviation at sweep value RP1$=0.567V$. In both figures we have grayed out rate calculations at sweep values where there is significant aliasing in the time series.
• Figure 5: Variation of the logarithm of the ratio of 2LF transition rates as a function of sweep parameter. Values inferred by FHMM analysis of experimental data are given by the points with error bars, while the model fits are given by the curves. (a) (b) (c). In all plots, the dashed vertical line corresponds to the fixed parameter value for which the other two sweeps were taken. In (a) the gray data points correspond to parameter values for which the transition rates $\Gamma_{10}^{1,2}$ are comparable to or faster than the 60 Hz sample rate of data acquisition and likely suffer from significant aliasing error. We exclude these points from the fits due to the poor estimation of the ratio of rates.