Gate-Defined Hole Spin Qubits

• Gate-defined hole spin qubits are qubits encoded in the spin states of holes confined by electrostatic gates in semiconductor quantum dots, with their Hamiltonians tuned via spin–orbit coupling.
• They leverage advanced material platforms—such as strained Ge/SiGe, Ge/Si core/shell nanowires, and silicon FinFET/MOS devices—to achieve all-electric, ultrafast, and coherent spin manipulation.
• Electrical control mechanisms like g–tensor modulation and displacement-induced fields enable sub–nanosecond qubit operations with high single–qubit fidelity, though sensitivity to charge noise remains a key challenge.

Gate-defined hole spin qubits are qubits encoded in the spin states of holes electrostatically confined in semiconductor quantum dots, typically in strained Ge/SiGe quantum wells, Ge/Si core/shell nanowires, or silicon transistor-like devices. Their defining feature is not only that local gates set occupation, tunnel barriers, and confinement, but also that the same electrostatic environment directly reshapes the spin Hamiltonian through electrically tunable spin-orbit coupling and $g$-tensor anisotropy. In that sense, “gate-defined” refers both to dot formation and to Hamiltonian engineering. The field includes single-hole qubits, exchange-coupled two-qubit systems, and encoded two-hole modalities such as $S$-$T_-$ qubits, with the common thread being strong spin-orbit-enabled all-electrical control and pronounced sensitivity to confinement details (Froning et al., 2020, Mei et al., 13 May 2026).

1. Conceptual development and early theoretical foundations

The modern subject emerged from the recognition that valence-band holes combine strong spin-orbit interaction with reduced contact hyperfine coupling. An early explicit proposal for fully electrical gate-defined control considered a single heavy-hole qubit in a planar semiconductor nanodevice, where static voltages on metal gates steer a self-trapped hole around closed loops and the Dresselhaus interaction converts that motion into deterministic spin rotations. In that framework the rotation angle is set geometrically by

$\phi = 2\pi \frac{\lambda}{\lambda_{\mathrm{SO}}},$

and a four-segment loop implements

$\hat U_{\mathrm{NOT}} = i\sigma_x.$

The same work reported sub-nanosecond theoretical gate times, with $$t_{\mathrm{NOT}}^{\mathrm{GaAs}}\approx 220~\mathrm{ps}$$, $$t_{\mathrm{NOT}}^{\mathrm{CdTe}}\approx 40~\mathrm{ps}$$, and $$t_{\mathrm{NOT}}^{\mathrm{ZnSe}}\approx 50~\mathrm{ps}$$, while keeping light-hole admixture at the level of $\sim 10^{-4}$ during transport (Szumniak et al., 2012).

A second foundational line of theory established why Ge-based hole devices are unusually attractive. In Ge/Si core/shell nanowire quantum dots, strong direct Rashba spin-orbit interaction was predicted to allow electrically switchable single-qubit control and cavity-mediated coupling, with spin-flip times shorter than $100~\mathrm{ps}$ and cavity-assisted $S$0 times below $S$1 (Kloeffel et al., 2013). In parallel, theory for gate-defined Ge hole dots in strained quantum wells argued that the top-gate electric field generically creates optimal operation points where first-order dephasing from electric noise vanishes, the electrically driven spin response is maximized, and phonon-limited relaxation is strongly suppressed at low magnetic field, with $S$2 or $S$3 depending on the Rashba channel (Wang et al., 2019).

These early works fixed several enduring themes: electrically generated spin rotations, strong but engineerable spin-orbit coupling, the search for “on” and “off” operating modes, and the idea that confinement geometry is itself a control resource rather than a passive background.

2. Material platforms and device architectures

Gate-defined hole spin qubits now span several experimentally distinct architectures. In planar strained Ge/SiGe, electrostatic gates define lateral dots inside a two-dimensional hole gas. In Ge/Si core/shell nanowires, radial confinement is provided by the heterostructure and longitudinal confinement by gates. In silicon FinFET and MOS devices, wrap-around or overlapping gates define quasi-one-dimensional or planar hole dots in a transistor-compatible geometry. Across these implementations, the active qubit degree of freedom is a hole confined by gate-controlled potentials rather than by a fixed impurity potential (Hardy et al., 2018, Froning et al., 2020, Geyer et al., 2022).

Planar strained Ge/SiGe was established as a viable host by the demonstration of single and double hole quantum dots in undoped heterostructures using a simple one-layer gate design. The strained Ge quantum well was $S$4 thick, the upper SiGe barrier $S$5, and the measured hole effective mass was $S$6, significantly lighter than the $S$7 quoted for Si/SiGe electrons. The same work emphasized peak mobility exceeding $S$8, absence of valley states, and the relevance of strong intrinsic spin-orbit coupling for later qubit operation (Hardy et al., 2018).

The Ge/Si core/shell nanowire implementation realized a single hole-spin qubit in a nanowire placed across five narrow bottom gates. The Ge core radius was about $S$9, the Si shell thickness about $T_-$0, the gates were $T_-$1 wide with $T_-$2 pitch, and the nanowire was insulated from the gates by $T_-$3 of $T_-$4. Positive gate voltages formed a depletion-mode few-hole double quantum dot used for Pauli-spin-blockade initialization and readout, while the manipulated qubit was effectively a single-spin qubit hosted in the one-dimensional hole system (Froning et al., 2020).

Silicon transistor-like devices extend the same logic into industrially familiar hardware. In a silicon FinFET, two single-hole dots were defined under plunger gates $T_-$5 and $T_-$6, with a central barrier gate $T_-$7 controlling interdot tunnelling and with gate widths of roughly $T_-$8. The fin provided a quasi-one-dimensional channel especially favorable for direct Rashba spin-orbit interaction (Geyer et al., 2022). A related FinFET geometry was later used to show that simply changing which gate carries the microwave excitation changes the dominant EDSR mechanism without changing the qubit itself (Geyer et al., 22 Dec 2025). In planar silicon MOS, a double dot with a nearby single-hole transistor and RF reflectometry demonstrated that hole-spin qubits can also be hosted in a foundry-compatible three-layer polysilicon gate stack on natural silicon (Vorreiter et al., 1 Aug 2025).

Platform	Gate-defined structure	Representative feature
Strained Ge/SiGe	Planar lateral single and double dots	$T_-$9 (Hardy et al., 2018)
Ge/Si core/shell nanowire	Bottom-gate-defined few-hole double dot	Gate-tunable spin-orbit switch (Froning et al., 2020)
Si FinFET	Two single-hole dots under plunger gates	Anisotropic exchange two-qubit logic (Geyer et al., 2022)
Si MOS	Planar double dot with reflectometry readout	Single-qubit Clifford fidelities up to 99.83% (Vorreiter et al., 1 Aug 2025)

A comparative assessment of germanium-based modalities later concluded that gate-defined Ge hole-spin qubits presently offer the strongest combination of all-electrical control, demonstrated multiqubit operation, and scalability, while explicitly identifying charge-noise sensitivity, $\phi = 2\pi \frac{\lambda}{\lambda_{\mathrm{SO}}},$0-tensor anisotropy, and materials/interface variability as their main liabilities (Mei et al., 13 May 2026).

3. Electrical control mechanisms

The central operational advantage of hole spin qubits is all-electrical driving. In the Ge/Si nanowire experiment, microwaves applied to a gate generated an oscillating electric field $\phi = 2\pi \frac{\lambda}{\lambda_{\mathrm{SO}}},$1 that displaced the hole wave function; spin-orbit coupling converted that motion into an effective transverse magnetic field and drove electric-dipole spin resonance. The observed Rabi frequency followed

$\phi = 2\pi \frac{\lambda}{\lambda_{\mathrm{SO}}},$2

and the data showed the expected linear dependence on both microwave amplitude and $\phi = 2\pi \frac{\lambda}{\lambda_{\mathrm{SO}}},$3. Gate control was exceptionally strong: the Rabi frequency was tuned from $\phi = 2\pi \frac{\lambda}{\lambda_{\mathrm{SO}}},$4 to $\phi = 2\pi \frac{\lambda}{\lambda_{\mathrm{SO}}},$5 with millivolt-scale gate changes, and further optimization yielded $\phi = 2\pi \frac{\lambda}{\lambda_{\mathrm{SO}}},$6, corresponding to a $\phi = 2\pi \frac{\lambda}{\lambda_{\mathrm{SO}}},$7-pulse time of about $\phi = 2\pi \frac{\lambda}{\lambda_{\mathrm{SO}}},$8 (Froning et al., 2020).

In silicon hole devices, the drive physics is more explicitly decomposed into $\phi = 2\pi \frac{\lambda}{\lambda_{\mathrm{SO}}},$9-tensor modulation resonance and wavefunction-displacement, or iso-Zeeman, resonance. The low-energy description is

$\hat U_{\mathrm{NOT}} = i\sigma_x.$0

with electric driving encoded by $\hat U_{\mathrm{NOT}} = i\sigma_x.$1. The measured Rabi vector is

$\hat U_{\mathrm{NOT}} = i\sigma_x.$2

where $\hat U_{\mathrm{NOT}} = i\sigma_x.$3 is the spin-quantization axis and $\hat U_{\mathrm{NOT}} = i\sigma_x.$4. In a silicon FinFET, driving from the plunger gate $\hat U_{\mathrm{NOT}} = i\sigma_x.$5 yielded a regime dominated by $\hat U_{\mathrm{NOT}} = i\sigma_x.$6-TMR, with average fractional contributions $\hat U_{\mathrm{NOT}} = i\sigma_x.$7 and $\hat U_{\mathrm{NOT}} = i\sigma_x.$8. Driving from the laterally offset barrier gate $\hat U_{\mathrm{NOT}} = i\sigma_x.$9 changed the local AC field orientation and shifted the balance to $$t_{\mathrm{NOT}}^{\mathrm{GaAs}}\approx 220~\mathrm{ps}$$0 and $$t_{\mathrm{NOT}}^{\mathrm{GaAs}}\approx 220~\mathrm{ps}$$1. The maximum IZR-induced Rabi contribution increased from below $$t_{\mathrm{NOT}}^{\mathrm{GaAs}}\approx 220~\mathrm{ps}$$2 under $$t_{\mathrm{NOT}}^{\mathrm{GaAs}}\approx 220~\mathrm{ps}$$3 drive to above $$t_{\mathrm{NOT}}^{\mathrm{GaAs}}\approx 220~\mathrm{ps}$$4 under $$t_{\mathrm{NOT}}^{\mathrm{GaAs}}\approx 220~\mathrm{ps}$$5 drive (Geyer et al., 22 Dec 2025).

Alternative gate-defined control paradigms remain important. The trajectory-based proposal for heavy-hole qubits uses static gate voltages to route a moving hole along prescribed segments, so that transport along $$t_{\mathrm{NOT}}^{\mathrm{GaAs}}\approx 220~\mathrm{ps}$$6 or $$t_{\mathrm{NOT}}^{\mathrm{GaAs}}\approx 220~\mathrm{ps}$$7 implements $$t_{\mathrm{NOT}}^{\mathrm{GaAs}}\approx 220~\mathrm{ps}$$8 or $$t_{\mathrm{NOT}}^{\mathrm{GaAs}}\approx 220~\mathrm{ps}$$9 directly (Szumniak et al., 2012). A more recent curved-quantum-well proposal argued that short gate-defined dots can support ultrafast operation through a transversal electric field, with $$t_{\mathrm{NOT}}^{\mathrm{CdTe}}\approx 40~\mathrm{ps}$$0 for a representative short-dot geometry and magnetic field aligned along the well (Bosco et al., 2022). These results suggest that “electrical control” in hole qubits should not be reduced to a single microscopic mechanism; in practice it includes gate-induced $$t_{\mathrm{NOT}}^{\mathrm{CdTe}}\approx 40~\mathrm{ps}$$1-tensor modulation, displacement-induced effective magnetic fields, spin-dependent transport phases, and curvature- or geometry-enabled dipole couplings.

4. Coherence, charge noise, hyperfine physics, and sweet spots

The central challenge of gate-defined hole spin qubits is that the same spin-orbit interaction enabling fast electrical control also couples the qubit to electrical noise. The theoretical sweet-spot program addressed this directly for Ge hole dots by predicting gate-field values $$t_{\mathrm{NOT}}^{\mathrm{CdTe}}\approx 40~\mathrm{ps}$$2 at which

$$t_{\mathrm{NOT}}^{\mathrm{CdTe}}\approx 40~\mathrm{ps}$$3

so first-order dephasing from vertical electric-field noise vanishes while the EDSR matrix element is maximized. In that framework, the qubit is also first-order insensitive to in-plane electric-field fluctuations in the relevant order, and low-field relaxation remains very slow because $$t_{\mathrm{NOT}}^{\mathrm{CdTe}}\approx 40~\mathrm{ps}$$4 scales as $$t_{\mathrm{NOT}}^{\mathrm{CdTe}}\approx 40~\mathrm{ps}$$5 or $$t_{\mathrm{NOT}}^{\mathrm{CdTe}}\approx 40~\mathrm{ps}$$6 (Wang et al., 2019).

That picture was experimentally sharpened in a planar heavy-hole Ge qubit, where the full anisotropic $$t_{\mathrm{NOT}}^{\mathrm{CdTe}}\approx 40~\mathrm{ps}$$7-tensor was measured and related to both drive and decoherence. The principal-axis anisotropy was extreme,

$$t_{\mathrm{NOT}}^{\mathrm{CdTe}}\approx 40~\mathrm{ps}$$8

and electrical spin driving was shown to arise predominantly from modulation of that anisotropic $$t_{\mathrm{NOT}}^{\mathrm{CdTe}}\approx 40~\mathrm{ps}$$9-tensor. The same work experimentally confirmed the predicted Ising-like hyperfine interaction of heavy holes, identified a hyperfine sweet plane where nuclear coupling is suppressed, and then showed that qubit coherence is ultimately limited by $$t_{\mathrm{NOT}}^{\mathrm{ZnSe}}\approx 50~\mathrm{ps}$$0 charge noise rather than by hyperfine fluctuations. In that low-field regime the reported single-qubit fidelity reached $$t_{\mathrm{NOT}}^{\mathrm{ZnSe}}\approx 50~\mathrm{ps}$$1, remained above $$t_{\mathrm{NOT}}^{\mathrm{ZnSe}}\approx 50~\mathrm{ps}$$2 at $$t_{\mathrm{NOT}}^{\mathrm{ZnSe}}\approx 50~\mathrm{ps}$$3, and the abstract quoted $$t_{\mathrm{NOT}}^{\mathrm{ZnSe}}\approx 50~\mathrm{ps}$$4 (Hendrickx et al., 2023).

The Ge/Si nanowire experiment provided a different form of coherence engineering: a gate-tunable spin-orbit switch. Sweeping a single dot-defining gate over only $$t_{\mathrm{NOT}}^{\mathrm{ZnSe}}\approx 50~\mathrm{ps}$$5 changed the Rabi frequency by about a factor of 7, changed the driven coherence time $$t_{\mathrm{NOT}}^{\mathrm{ZnSe}}\approx 50~\mathrm{ps}$$6 from $$t_{\mathrm{NOT}}^{\mathrm{ZnSe}}\approx 50~\mathrm{ps}$$7 to $$t_{\mathrm{NOT}}^{\mathrm{ZnSe}}\approx 50~\mathrm{ps}$$8, and tuned the Landé $$t_{\mathrm{NOT}}^{\mathrm{ZnSe}}\approx 50~\mathrm{ps}$$9-factor from $\sim 10^{-4}$0 to $\sim 10^{-4}$1. The corresponding extracted spin-orbit length ranged from $\sim 10^{-4}$2 down to $\sim 10^{-4}$3, assuming a heavy-hole effective mass. The free-induction coherence time there was much shorter, $\sim 10^{-4}$4, but Hahn echo improved it by roughly a factor of 25, indicating dominant low-frequency noise (Froning et al., 2020).

Foundry-fabricated silicon MOS hole qubits illustrate a complementary regime in which electrical control is slower than the nanowire record but coherence and benchmarking are already highly competitive. In natural silicon, one report gave $\sim 10^{-4}$5, $\sim 10^{-4}$6, Hahn-echo times $\sim 10^{-4}$7 and $\sim 10^{-4}$8, $\sim 10^{-4}$9, and single-qubit Clifford fidelities of $100~\mathrm{ps}$0 and $100~\mathrm{ps}$1 (Vorreiter et al., 1 Aug 2025). Taken together, these results show that gate-defined hole qubits do not occupy a single speed–coherence point; rather, they span a broad operating space in which geometry, strain, field orientation, and gate bias jointly determine the balance.

5. Readout strategies, encoded variants, and two-qubit logic

Readout in gate-defined hole systems has largely developed from double-dot architectures. In the Ge/Si nanowire qubit, initialization, control, and readout were all embedded in a transport-based Pauli-spin-blockade cycle on a few-hole double dot: the device was initialized in a blockaded triplet, pulsed to a Coulomb-blocked manipulation point for microwave control, and returned to the readout point where current flowed only after spin-to-singlet conversion (Froning et al., 2020). In planar silicon MOS, latched Pauli spin blockade with a nearby single-hole transistor and RF reflectometry enabled deterministic initialization, parity readout, and exchange-based two-qubit logic in a natural-silicon foundry device (Vorreiter et al., 1 Aug 2025).

Gate-based reflectometry has also become a characterization tool for local spin parameters. In a silicon hole double dot, the non-monotonic magnetic-field dependence of the reflected phase at an even-parity interdot transition was modeled with a singlet–triplet Hamiltonian containing dot-dependent Zeeman energies, yielding local $100~\mathrm{ps}$2-factors $100~\mathrm{ps}$3 and $100~\mathrm{ps}$4, together with $100~\mathrm{ps}$5 and a phase maximum at $100~\mathrm{ps}$6 (2206.13125). This established that gate-only dispersive sensing can extract site-dependent $100~\mathrm{ps}$7-factors without coherent spin manipulation.

Gate-defined hole qubits also include encoded two-hole modalities. In a strained Ge/SiGe double dot, an $100~\mathrm{ps}$8-$100~\mathrm{ps}$9 qubit was operated near the $S$00–$S$01 anticrossing with effective Hamiltonian

$S$02

The shared barrier gate $S$03 changed the effective $S$04-factor by roughly an order of magnitude over only $S$05, which the authors attributed to gate-induced motion through a nonuniform strain field. The same device yielded $S$06 at the anticrossing, $S$07 at large detuning, and $S$08 (Rooney et al., 2023).

Two-qubit logic has advanced particularly rapidly in hole-spin transistors and germanium double dots. In a silicon FinFET, a controlled-rotation gate was demonstrated with a conditional spin flip in approximately $S$09. The exchange interaction was tunable from about $S$10 to below a $S$11 spectroscopic resolution limit, and the interaction was shown to be a rotated, anisotropic exchange rather than a scalar Heisenberg term. A global fit gave $S$12, $S$13, $S$14, and a spin-orbit axis $S$15; the theory then projected that CNOT fidelities above $S$16 should be accessible for the extracted anisotropic-exchange regime (Geyer et al., 2022).

A distinct route used resonant exchange modulation to realize native fermionic simulation gates in Ge/SiGe hole qubits. For two single-hole dots with idle frequencies $S$17 and $S$18, a combined baseband-plus-resonant exchange pulse implemented an iSWAP gate with a $S$19 diabatic duration and $S$20 interleaved randomized-benchmarking fidelity. The same work emphasized that tomography identified decoherence, rather than calibration error, as the dominant limitation (Tsoukalas et al., 18 Jul 2025).

6. Architectural directions, unresolved limits, and current outlook

Several architectural programs now extend gate-defined hole qubits beyond nearest-neighbor exchange. One line combines Ge hole dots with bosonic modes. In a phononic-crystal-cavity design for gate-controlled Ge dots, vertical electric fields were predicted to tune $S$21-factors from $S$22 down to $S$23, while the spin–phonon coupling increased from $S$24 to $S$25; the same study projected cavity quality factors above $S$26 and phonon-mediated $S$27 values in the millisecond range (Mei et al., 16 Apr 2025). Earlier circuit-QED theory for Ge/Si nanowires had already argued that hole-spin qubits can be coupled to superconducting resonators with electrically switchable transverse coupling, again using the gate field as an on–off control variable (Kloeffel et al., 2013). A curved-quantum-well proposal extended that logic to shell-like Ge wells and predicted spin–photon interaction strengths of a few hundreds of megahertz, reaching $S$28 in aggressive but realistic scenarios, while also claiming broader suppression of charge-noise sensitivity than in conventional long-dot designs (Bosco et al., 2022).

At the same time, more recent theory has imposed a fundamental qualification on the idea of a perfect spin-orbit switch. A non-perturbative, quantum-geometric treatment of gate-defined hole qubits showed that the weak in-plane confinement itself generates spin-orbit structure through Berry connections, so the projected confinement appears as

$S$29

Because the spin-orbit interaction inherited from the parent 2DHG or nanowire and the confinement-induced spin-orbit interaction have different forms, they cannot generally be turned off simultaneously. In that sense, the theory explicitly states that perfect spin-orbit-switch functionality is spoiled by the quantum geometry of the confined band manifold (György et al., 12 Jun 2026).

Strain engineering adds a further design axis. A theory comparison of heavy-hole and light-hole gate-defined Ge qubits concluded that strain can move the system between regimes with

$S$30

For the heterostructures studied, LH qubits in Ge dots were argued to offer lower relaxation rates and higher Rabi frequencies than the HH configurations considered, especially in GeSn/Ge where tensile strain stabilizes LH confinement (Dsouza et al., 2024). This suggests that the mature heavy-hole Ge/SiGe program may eventually be complemented by LH-oriented designs optimized for different operating tradeoffs.

The resulting outlook is technically specific rather than generic. Gate-defined hole spin qubits already support ultrafast electrical control, high-fidelity single-qubit operation, anisotropic-exchange two-qubit logic, resonant exchange gates, and encoded two-hole qubits. Their principal unresolved issues are equally specific: charge-noise sensitivity tied to electrically active $S$31-tensors, strong device-to-device anisotropy from confinement and strain, and the fact that spin-orbit coupling cannot be reduced to a single tunable scalar. A comparative assessment of Ge platforms nevertheless judged gate-defined hole qubits to be the clearest current route toward scalable Ge-based processors because they already combine all-electrical control, multiqubit demonstrations, and array-compatible planar fabrication (Mei et al., 13 May 2026).