Quantum Gate

Figure

Description

A quantum gate is a physical control operation intended to implement a target unitary transformation on one or more qubits. The circuit symbol is an abstraction: hardware realizes the gate by shaping a time-dependent Hamiltonian, and the resulting propagator must approximate the target unitary within an error budget that includes decoherence, leakage, calibration error, and crosstalk.

Single-qubit gates generate rotations of the Bloch vector. Resonant in-phase and quadrature controls set the rotation axis in the equatorial plane, detuning contributes a $z$ component, and software frame updates implement virtual $Z$ rotations without a physical pulse. A universal gate set additionally requires at least one entangling two-qubit gate; CNOT, CZ, iSWAP-family, and Mølmer–Sørensen gates are different native realizations of that nonlocal resource.

In superconducting circuits, shaped microwave pulses implement one-qubit rotations. DRAG adds a derivative quadrature to suppress leakage in weakly anharmonic transmons. Entangling gates use interactions such as cross-resonance $ZX$, tunable-coupler controlled phase, resonator-mediated exchange, or parametrically activated iSWAP-like coupling. Representative one-qubit durations range from a few nanoseconds in strongly anharmonic fluxonium to tens of nanoseconds in transmons; two-qubit durations are commonly tens to hundreds of nanoseconds.

In trapped ions, Raman or microwave fields drive hyperfine or Zeeman qubits. Two-qubit gates use spin-dependent forces coupled through shared motional modes. The Mølmer–Sørensen interaction transiently displaces motion in phase space and closes the trajectory at the end of the pulse, ideally leaving a pure spin entangler; the Cirac–Zoller construction instead maps information through resolved sidebands and a single phonon.

In semiconductor spin qubits, ESR or EDSR provides local rotations, while voltage-controlled exchange between neighboring spins produces SWAP-family or controlled-phase operations. Exchange can be extremely fast, but charge noise, magnetic gradients, spectator states, and device-to-device variability make calibration and simultaneous operation central scaling problems.

In neutral-atom processors, local or global optical and microwave fields provide one-qubit control. Rydberg excitation creates a strong conditional interaction: blockade shifts the doubly excited state and permits a controlled phase. Reported fidelities depend on the averaging domain and leakage treatment, so symmetric-subspace benchmarks should not be compared naively with full two-qubit Haar-average fidelities.

For photonic qubits, passive linear optics implements deterministic one-qubit transformations on path, polarization, time-bin, or frequency-bin encodings. Deterministic photon–photon interactions are difficult; the KLM construction uses ancilla photons, detection, feed-forward, and teleportation to promote nondeterministic entangling operations, while fusion-based and measurement-based architectures consume entangled resource states.

Gate fidelity is not a single protocol-independent quantity. Randomized benchmarking suppresses state-preparation-and-measurement sensitivity under model assumptions; interleaved RB estimates a selected gate relative to a reference decay; gate-set tomography reconstructs a self-consistent gate set; and leakage-aware or subspace benchmarks answer narrower questions. The protocol, encoded subspace, averaging ensemble, concurrency, and uncertainty must accompany any record claim.

Hamiltonian

The general controlled evolution is

$$H(t)=H_0+\sum _k{u}_k(t)H_k,U(t_g)=\mathcal{T}\exp \left[-\frac{i}{\hslash }{\int }_0^{t_g}H(t)dt\right].$$

Here $H_0$ is the drift Hamiltonian, $H_k$ are available control or interaction operators, and $u_k(t)$ are calibrated waveforms. In a frame rotating near a two-level transition and after the rotating-wave approximation, a driven qubit is represented by

$$\frac{H_{1q}(t)}{\hslash }=\frac{1}{2}\left[{\Omega }_x(t){\sigma }_x+{\Omega }_y(t){\sigma }_y+\Delta (t){\sigma }_z\right].$$

The pulse area and phase determine the rotation angle and axis. This reduction must be amended when the drive is strong, the qubit is weakly anharmonic, or counter-rotating terms are not negligible.

A useful canonical description of a two-qubit interaction is

$$\frac{H_{\mathrm{int}}}{\hslash }=J_{xx}(t){\sigma }_x\otimes {\sigma }_x+J_{yy}(t){\sigma }_y\otimes {\sigma }_y+J_{zz}(t){\sigma }_z\otimes {\sigma }_z,$$

with platform-specific local terms and unwanted spectators added as needed. For isotropic spin exchange, with $J(t)$ expressed as an energy,

$$H_{\mathrm{ex}}(t)=\frac{J(t)}{{\hslash }^2}{\mathbf{S}}_1\cdot {\mathbf{S}}_2=\frac{J(t)}{4}\left(XX+YY+ZZ\right).$$

Defining $\theta ={\hslash }^{-1}\int J(t)dt$, $\theta =\pi /2$ produces a $\sqrt{\mathrm{SWAP}}$ up to global phase, while $\theta =\pi$ produces SWAP. A controlled-$Z$ does not follow from one $\sqrt{\mathrm{SWAP}}$ pulse alone; it requires an appropriate interaction with a $ZZ$ component or a compiled sequence containing exchange pulses and local rotations.

Motivation

• divincenzo-criteria make a high-quality universal gate set a basic requirement for a useful quantum processor.
• threshold-theorem turns control quality into a scaling question: physical errors must be sufficiently low and sufficiently well behaved for logical suppression to beat overhead.
• Native interactions determine compilation depth. resonator-as-quantum-bus and sqrt-swap-as-universal-gate illustrate two distinct routes from hardware coupling to universal circuits.
• Gate engineering exposes the central hardware trade-off: stronger, faster control reduces decoherence exposure but can increase leakage, counter-rotating error, crosstalk, heating, and calibration sensitivity.

Experimental Status

First experimental quantum logic gate — Monroe et al. (1995):

• Demonstrated a controlled operation between the internal and motional states of a single trapped ${}^9{\mathrm{Be}}^{+}$ ion, establishing the first experimental two-bit quantum logic gate.

Photonic universality with measurement — Knill, Laflamme, and Milburn (2001):

• Showed that scalable quantum computation is possible with linear optics, ancilla photons, photodetection, feed-forward, and teleportation despite nondeterministic primitive entanglers.

Leakage-aware superconducting control — Motzoi et al. (2009):

• Introduced the derivative-quadrature strategy now called DRAG for suppressing leakage and phase error in weakly nonlinear qubits.

Fast superconducting one-qubit frontier — Rower et al. (2024):

• Demonstrated commensurate-pulse control of fluxonium and measured a best average single-qubit gate fidelity of $99.99807(7)\%$; the optimized $X$ pulse was about $8.2\mathrm{ns}$.

Trapped-ion one-qubit frontier — Smith et al. (2025):

• Reported a minimum error per Clifford of $1.5(4)\times 10^{-7}$ on a ${}^{43}{\mathrm{Ca}}^{+}$ hyperfine clock qubit for gate times spanning $4.4$ to $35\mu \mathrm{s}$.

Platform-diverse entangling benchmarks — Tsai et al., Edlbauer et al., and Hughes et al. (2025):

• Neutral atoms: a Rydberg CZ fidelity of $0.9971(5)$ averaged over symmetric input states and downward-corrected for leakage.
• Silicon donor processor: one-qubit fidelity up to $99.99\%$ and a local nuclear-spin CZ fidelity of $99.90(4)\%$ in an 11-qubit device.
• Trapped ions: an electronically driven two-qubit gate error of $8.4(7)\times 10^{-5}$ without ground-state cooling, using a $225.8\mu \mathrm{s}$ smooth gate.

Balanced superconducting operation — Marxer et al. (2026):

• Demonstrated a $33\mathrm{ns}$ tunable-coupler CZ with $99.93\%$ fidelity averaged over 40 hours, alongside $>99.98\%$ one-qubit gates and $>99.94\%$ readout in the same device.

Parallel silicon-spin control — Wu et al. (2026):

• Demonstrated primitive $X/Y$ gate fidelities above $99.99\%$ across a five-qubit silicon spin array, including an interleaved-RB $X_{\pi /2}$ fidelity of $99.9985(8)\%$ on one qubit. Fidelities remained above $99.99\%$ during simultaneous operation of up to three qubits, while five-qubit parallel operation stayed above $99.9\%$.

The 2024–2026 literature search found no later peer-reviewed result that cleanly supersedes the cited one-qubit trapped-ion record, the Tsai neutral-atom CZ benchmark, or the Marxer balanced superconducting benchmark as of this audit. Wu 2026 slightly exceeds the cited Rower fluxonium value in isolated primitive-gate fidelity and adds a stronger parallel-control result, while Rower remains the representative few-cycle speed–fidelity benchmark. Cross-platform rankings remain conditional on the benchmarking protocol and operating context.

Key Metrics

Metric	Value	Notes	Fidelity reference
1Q Clifford fidelity, trapped ion	$99.999985\%$	${}^{43}{\mathrm{Ca}}^{+}$; minimum error $1.5(4)\times 10^{-7}$; $4.4$–$35\mu \mathrm{s}$	Smith et al. 2025
1Q gate fidelity, fluxonium	$99.99807(7)\%$	Best measured average; optimized $X$ pulse $\approx 8.2\mathrm{ns}$	Rower et al. 2024
1Q gate fidelity, transmon	$99.99258(4)\%$	RB error $(7.42\pm 0.04)\times 10^{-5}$	Li et al. 2023
1Q gate fidelity, silicon donor processor	up to $99.99\%$	Best qubit in an 11-qubit atom processor	Edlbauer et al. 2025
1Q primitive $X_{\pi /2}$ fidelity, silicon quantum dot	$99.9985(8)\%$	Interleaved RB on one qubit; all five individually $>99.99\%$; up to three simultaneous qubits $>99.99\%$	Wu et al. 2026
2Q fidelity, trapped ion	$99.9916(7)\%$	Smooth gate; no ground-state cooling; $225.8\mu \mathrm{s}$	Hughes et al. 2025
2Q CZ fidelity, superconducting transmons	$99.93\%$	40-hour average; $33\mathrm{ns}$; PALEA calibration	Marxer et al. 2026
2Q CZ fidelity, neutral atoms	$99.71(5)\%$	Symmetric-input average; leakage-downward-corrected	Tsai et al. 2025
2Q CZ fidelity, silicon donor nuclei	$99.90(4)\%$	Local nuclear-spin CZ in an 11-qubit processor	Edlbauer et al. 2025

These values are not a protocol-independent league table. RB, interleaved RB, leakage-aware subspace RB, and restricted-input fidelity averages probe different channels and can differ in their treatment of leakage, SPAM, drift, concurrency, and spectators.

Scaling Considerations

• Decoherence: relaxation and dephasing act during the finite gate time. The common estimate $t_g/T_{\mathrm{coh}}$ is only an order-of-magnitude guide and does not capture non-Markovian noise or pulse-dependent exposure.
• Leakage: fast control can populate noncomputational levels or motional states. DRAG, smooth envelopes, adiabatic closure, and leakage-aware calibration target different manifestations of this problem.
• Crosstalk and spectators: residual $ZZ$, microwave or optical spillover, shared modes, and frequency crowding create context-dependent and correlated errors that isolated-pair RB can miss.
• Drift and calibration: qubit frequencies, coupling strengths, pulse transfer functions, and readout parameters drift. Long-duration averages and simultaneous benchmarking are more informative for scale than a single best calibration.
• Control hardware: waveform bandwidth, DAC resolution, phase noise, latency, cryogenic heat load, optical stability, and channel count can dominate system-level performance even when a single gate is excellent.

References

Landmark gates and control methods

• C. Monroe et al., “Demonstration of a Fundamental Quantum Logic Gate,” Physical Review Letters 75, 4714 (1995).
• E. Knill, R. Laflamme, and G. J. Milburn, “A scheme for efficient quantum computation with linear optics,” Nature 409, 46 (2001).
• F. Motzoi et al., “Simple Pulses for Elimination of Leakage in Weakly Nonlinear Qubits,” Physical Review Letters 103, 110501 (2009).

Recent one-qubit benchmarks

• Z. Li et al., “Error per single-qubit gate below $10^{-4}$ in a superconducting qubit,” npj Quantum Information 9 (2023).
• D. A. Rower et al., “Suppressing Counter-Rotating Errors for Fast Single-Qubit Gates with Fluxonium,” PRX Quantum 5, 040342 (2024) — arXiv:2406.08295.
• M. C. Smith et al., “Single-qubit gates with errors at the $10^{-7}$ level,” Physical Review Letters 134, 230601 (2025) — arXiv:2412.04421.

Recent entangling and integrated benchmarks

• R. B.-S. Tsai et al., “Benchmarking and fidelity response theory of high-fidelity Rydberg entangling gates,” PRX Quantum 6, 010331 (2025) — arXiv:2407.20184.
• H. Edlbauer et al., “An 11-qubit atom processor in silicon,” Nature 648, 569 (2025) — arXiv:2506.03567.
• A. C. Hughes et al., “Trapped-ion two-qubit gates with $>99.99\%$ fidelity without ground-state cooling,” arXiv:2510.17286 (2025).
• F. Marxer et al., “Above $99.9\%$ Fidelity Single-Qubit Gates, Two-Qubit Gates, and Readout in a Single Superconducting Quantum Device,” PRX Quantum 7, 020333 (2026) — arXiv:2508.16437.
• Y.-H. Wu et al., “Simultaneous high-fidelity single-qubit gates in a spin qubit array,” Nature Physics (2026) — arXiv:2507.11918.

Linked Papers

• monroe-1995-demonstration-fundamental-logic
• knill-2001-klm
• motzoi-2009-simple-pulses-elimination
• li-2023-error-per-single
• rower-2024-suppressing-counter-rotating-errors
• smith-2024-single-qubit-gate-errors
• tsai-2025-benchmarking-rydberg-entangling-gates
• edlbauer-2025-11-qubit-atom-processor
• hughes-2025-trapped-ion-two-qubit-gates
• marxer-2026-above-99-9-fidelity
• wu-2026-simultaneous-high-fidelity-spin-array

Related Entries

• transmon — dominant superconducting qubit platform
• fluxonium — strongly anharmonic superconducting qubit with record one-qubit control
• trapped-ion-qubit — platform with the lowest reported physical gate errors
• spin-qubit — umbrella entry for semiconductor spin encodings and exchange control
• silicon-spin-qubit — CMOS-compatible quantum-dot spin implementations
• kane-qubit — donor nuclear and electron spins in silicon
• neutral-atom-qubit — optically trapped atoms with Rydberg-mediated entanglers
• rydberg-neutral-atom-qubit — Rydberg-excited encoding and control
• photonic-qubit — flying-qubit platform based on optical modes
• qubit-readout — measurement and SPAM, distinct from gate error
• circuit-qed — microwave architecture for superconducting control and coupling
• tunable-coupler — switchable superconducting two-qubit interaction
• cirac-zoller-gate — original trapped-ion entangling construction
• molmer-sorenson-gate — workhorse spin-dependent-force trapped-ion gate
• dual-rail-photonic-qubit — path-mode photonic encoding
• linear-optical-photonic-qubit — KLM-family optical processing
• exchange-only-qubit — encoded spin control using exchange interactions
• surface-code-logical-qubit — error correction whose overhead depends strongly on gate error