Fast high-fidelity baseband reset of a latched state for quantum dot qubit readout

TL;DR

This work addresses slow initialization after latched readout in quantum dot qubits by introducing an on-demand active reset using baseband pulses. The method drives the system to a fast unlatching region in gate space, implementing a two-step decay: rapid reservoir unloading from the latched state to a nearby charge configuration at rate $Γ_R$, followed by an inelastic tunneling transition to the ground state at rate $1/T_1$ to complete re-initialization. They demonstrate > $99.5%$ re-initialization fidelity within $4 μs$, more than a factor of $50$ faster than the natural latch decay, and map the optimal reset region bounded by $E_{orb}^*$. The authors also analyze failure regimes caused by long-lived intermediate states and discuss applicability to L-PSB and other spin qubits, highlighting potential material strategies to further speed up reset. This approach can substantially reduce initialization and readout errors, aiding quantum error correction and scalable qubit operation.

Abstract

A common method for reading out the state of a spin qubit is by latching one logical qubit state, either $|1\rangle$ or $|0\rangle$, onto a different, metastable charge state. Such a latched state can provide a superior charge sensing signal for qubit readout, and it can have a lifetime chosen to be long enough that the charge sensed readout can be high fidelity. However, the passive reset out of latched states is inherently long, which is not desirable. In this work, we demonstrate an on-demand, high fidelity (> 99%) re-initialization of a quantum dot qubit out of a latched readout state. The method is simple to apply as it involves a single baseband voltage pulse to a specific region in the quantum dot stability diagram where the relaxation time from the latched state to the ground state is over 50 times faster. We describe the mechanism for the reset process as well as the boundaries for the optimal reset region in the qubit gate voltage space.

Figures (6)

• Figure 1: (a) SEM micrograph of a Tunnel Falls triple dot device lithographically identical to the one used in this experiment. The circuit shown is used both for time-averaged measurements and fast, single-shot measurements. For the latter, a cryogenic amplifier is mounted at the mixing chamber (MC) stage before the transimpedance amplifier (TIA) and the analog-to-digital converter (ADC), the latter of which both are at room temperature. (b) Stability diagram with an applied excitation pulse (inset) in the (4,1)-(3,2) charge configuration with tunnel rates tuned to latch onto a (3,1) charge state. The excitation pulse amplitude takes the system from point I/R (initialization/readout) to point O (operation). (c) Energy dispersion diagram of the QDHQ in the (4,1)-(3,2) charge regime with the latched state shown along with its tunable parameters. (d) A distribution of the lifetimes of many latching events showing the inherently long initialization times that come with long lived latched states. The inset reports the cumulative distribution of the latch durations as the fraction of latching events that have unlatched, $N_\text{unlatched}/N_\text{total}$, as a function of time $t$, demonstrating that a non-active reset requires waiting much longer than the minimum measurement time in order for re-initialization to occur.
• Figure 2: (a) QDHQ dispersion diagram showing how the reload pulse, shown in the inset of (b), initializes the qubit quickly out of the latched state in a fast two-step process. (b) Histogram showing the lifetimes of $\sim3000$ latching events with (orange) and without (blue) the active reset pulse shown in the inset being applied. (c) Baseband reset data overlaid on top of a stability diagram showing the region where the initialization happens the best. This region is bounded in detuning by $E_\text{orb}^*$ which was measured using pulsed gate spectroscopy (inset). Two distinct regions are indicated by the blue dot and red diamond, where an abrupt change in reset fidelity is observed. (d) Reset fidelity as a function of $t_\text{dwell}$ in both regions indicated in (c). Inset shows the model used to fit to the data in the red diamond region. (e) $T_1$ data measured in the blue dot region utilizing the pulse used in the inset. Two different curves are observed which are attributed to relaxation from either $(3,2)_g$ or $(3,2)_e$ to $(4,1)_g$. (f) Reset fidelity as a function of $t_\text{meas}$ taken for four different values of the $t_\text{wait}$ pulse parameter. The rate of change of the reset fidelity follows that of the natural decay of the latched state (fit to the red dashed line) until the reload pulse is sent at which point the rate of change significantly increases (black dashed line), showing that we can re-initialize our qubit on-demand, quickly, and with high fidelity.
• Figure 3: Pulsed-gate measurements to understand the decay pathways for $\varepsilon > E_\text{orb}^{*}$. (a)-(c) The probability of transitioning from the latched state to a (4,1) state using the pulse shown in the inset of Fig. \ref{['fgr:reset']}b for three different $t_\text{dwell}$ times. The increase in $P(4,1)$ for $\varepsilon > E_\text{orb}^{*}$ includes transitions to state $(4,1)_e$. (d) Dispersion diagram showing how the pulse shown in the upper inset can differentiate between the decay paths to $(4,1)_e$ and $(4,1)_g$. The ramp between stages II and III maps the $(4,1)_g$ state to itself, resulting in only one latching signal. That ramp maps the $(4,1)_e$ state to a $(3,2)$ state, which will then relatch, resulting in an observable second latching event. The probability of seeing a second latching event given that a first latching event, $P$(2nd$|$1st), is shown in the lower inset as a function of the dwell times for each pulse, $\tau_1$ and $\tau_2$. $P$(2nd$|$1st) reaches a value as large as 60%, indicating that the latched state often decays to $(4,1)_e$ when $\varepsilon > E_\text{orb}^{*}$.
• Figure 4: Example single-shot data used in this work. The blue trace is the raw SET data and the yellow trace is the transformed PELT data showing the abrupt changes in the average value of the SET data. The red trace is the pulse marker and shows when a pulse is applied. Visible are two example latching events, one that is long lived whose signal response is affected by the high pass filter, and a short lived latching event whose signal response isn't affected by the high pass filter.
• Figure 5: Ground and excited state tunnel coupling measurements. (a) Dispersion diagram showing which tunnel couplings are extracted. $t_{eg}$, highlighted in red, is extracted from the data in (b), and $t_{gg}$ and $t_{ge}$, highlighted in blue, are extracted by simultaneous fitting to (c) and (d). (b). $t_{eg}$ data acquired using the pulse shown in the inset. The pulse is tuned to abruptly move across the ground state anti-crossing to enter an excited state via a LZ transition. Then the ramp time across the $t_{eg}$ anti-crossing is swept. (c-d). $t_{gg}$ and $t_{ge}$ data acquired using the pulse shown in the inset. Both tunnel couplings are extracted by simultaneous fitting to both data sets.