In-situ control of hole-spin driving mechanisms

TL;DR

This work demonstrates in-situ control of hole-spin qubit driving mechanisms in a silicon FinFET by routing microwave drive to different gates. Using the $g$-matrix formalism to separate gTMR and IZR contributions, the authors map the $g$-tensor, its gate-dependent derivatives, and the Rabi response, showing a switch from gTMR-dominated driving (P1) to a mixed regime with enhanced IZR when driving from a laterally offset gate (B). The results quantify the drive composition with $p_{ ext{IZR}}$ and $p_{ ext{gTMR}}$ (15%/85% for P1 and 55%/45% for B) and demonstrate a fivefold increase in IZR contribution, revealing a path to optimize speed and coherence via gate geometry. The findings highlight the practical potential of all-electrical, gate-by-gate control to tailor spin dynamics for scalable hole-spin qubit architectures and motivate future designs that exploit drive-mode switching and field orientation for enhanced qubit performance.

Abstract

Hole-spin qubits enable fast, all-electrical spin manipulation through electric-dipole spin resonance (EDSR), arising from two microscopic mechanisms rooted in their intrinsically strong spin-orbit interaction. Depending on how the electric field acts on the quantum dot, the spin can be driven either by a modulation of its g-factor or by a displacement of the wavefunction. Here, we demonstrate in-situ control over the dominant EDSR driving mechanism of a hole-spin qubit in a silicon fin field-effect transistor by applying microwave signals to two different gate electrodes, thereby tuning the orientation of the local electric field. We measure the effective g-factor, its electrical tunability, and the Rabi frequency as functions of magnetic-field orientation. Their distinct angular dependencies, analyzed using a g-matrix formalism, allow us to identify the underlying driving processes and track their relative contributions for different drive configurations. By selecting the drive electrode, we can switch from a regime dominated by g-factor modulation to one with a strong contribution from wavefunction displacement. This in-situ tunability provides direct experimental access to both spin-driving mechanisms and offers a route toward optimized spin-qubit performance.

Figures (3)

• Figure 1: Hole spin qubit anisotropies and driving mechanisms.a, False-color transmission electron microscope (TEM) image of the measured device, showing the cross-section along the fin direction. The qubits (Q1, Q2) are located beneath the plunger gates (P1, P2) and are manipulated by applying microwave signals to either P1 or the barrier gate B. The barrier voltage controls the inter-dot tunneling, and the lead gates (L1, L2) accumulate the hole reservoirs. The widths of the B and P gates are $\unit[{\simeq}\,{20}]{nm}$. b, Cross-sectional TEM image of a co-fabricated device, highlighting the nearly triangular fin shape. c, 3D representation of the $g$-tensor, showing a large anisotropy within the $xy$-plane. d, Rabi chevron pattern of Q1 under P1 drive, demonstrating a Rabi frequency of $f_\mathrm{R}=20$ MHz on resonance. Data were taken at $B=0.136$ T, $\phi = 0^\circ$, $\theta = 70^\circ$ (as defined in Fig. \ref{['fig2']}), and $V_\mathrm{MW,P1}=12$ mV. e, The qubit Larmor frequency $f_\mathrm{L}$ of Q1 is tuned by the voltages applied to P1 and B. Data were taken at $B=0.172$ T, $\phi = 135^\circ$, $\theta = 90^\circ$. f, g, Schematic illustrations of the gTMR and IZR driving mechanisms, respectively.
• Figure 2: Rabi drive anisotropy.a, Schematic of the three magnetic-field planes used in the experiments, shown relative to the orientation of the Si fin (light blue) and the two driving gate electrodes (dark blue, orange). b, Rabi frequency of Q1 under P1- (blue markers) and B-driving (orange markers) as a function of magnetic-field orientation. Data were taken at fixed $f_\mathrm{L}=4.5\,$GHz, $V_\mathrm{MW,P1}=12\,$mV, and $V_\mathrm{MW,B}=13.5\,$mV. Solid curves are fits based on the model described in the main text. The uncertainty in $f_\mathrm{R}$ is estimated as $\pm1$ MHz. c, Electrical longitudinal tunability of the Q1 $g$-factor as function of magnetic-field orientation for P1- (blue markers) and B-driving (orange markers). Solid curves correspond to model fits described in the main text. d, Solid and dashed curves show the transverse components of $\hat{g}'$, as defined in the main text. The 1$\sigma$ error bars in $g'$ were obtained by propagating an estimated magnetic field hysteresis of $\pm5$ mT. Gaps in the data are due to failing PSB readout for some magnetic field orientations.
• Figure 3: IZR and gTMR Rabi drive contributions.a, Calculated Rabi frequency of Q1 (first column), decomposed into the contributions from IZR (second column) and gTMR (third column) under P1-driving. The dominant contribution arises from gTMR, with $p_\mathrm{gTMR}=85\%$. b, Same analysis for B-driving. Here, IZR and gTMR contribute comparably to the total Rabi driving ($p_\mathrm{IZR}=55\%$ and $p_\mathrm{gTMR}=45\%$). Constructive (destructive) interference between the two mechanisms occurs at the red $+$ ($-$) symbols, leading to enhanced (suppressed) overall Rabi frequency.