Quantum Gate

Description

A quantum gate is a physical operation that implements a unitary transformation $U$ on one or more qubits by applying precisely calibrated control signals — microwave pulses, laser beams, voltage waveforms, or magnetic field gradients — for a specific duration. In the abstract circuit model, gates are idealized unitary matrices; in hardware, every gate is a time-dependent Hamiltonian interaction $H(t)$ whose propagator $U=\mathcal{T}\exp \left(-\frac{i}{\hslash }{\int }_0^{t_g}H(t)dt\right)$ must approximate the target unitary to within a specified error budget.

Single-qubit gates rotate the Bloch vector about arbitrary axes and are typically the highest-fidelity operations on any platform. Physically, they correspond to resonant or near-resonant driving of the qubit transition (e.g., $\hat{H}=\frac{\Omega (t)}{2}{\hat{\sigma }}_x$ for a Rabi drive with envelope $\Omega (t)$).

Two-qubit entangling gates (CNOT, CZ, $\sqrt{\text{iSWAP}}$, etc.) create entanglement by exploiting a controllable interaction between qubits — capacitive coupling, shared motional modes, exchange interaction, or dipolar Rydberg blockade. These are typically the fidelity-limiting operations and the primary bottleneck for fault-tolerant quantum computing.

Physical Implementations by Platform

Superconducting Qubits

Single-qubit gates are driven by shaped microwave pulses at the qubit transition frequency ($\sim 4-6\text{GHz}$). DRAG (Derivative Removal by Adiabatic Gate) pulses suppress leakage to the $|2⟩$ state caused by the transmon’s weak anharmonicity, achieving gate times of $20-50\text{ns}$. Rotations about $\hat{z}$ are implemented virtually (frame updates in the control software) at zero time cost.

Two-qubit gates use several schemes:

• Cross-resonance (CR): driving one transmon at a neighbor’s frequency to produce a $ZX$ interaction (IBM’s native gate)
• Tunable coupler + CZ: flux-pulsing a tunable element to mediate a controlled-phase gate ($100-300\text{ns}$)
• Parametric gates: modulating the coupler at the difference frequency to activate $\text{iSWAP}$-like interactions

Trapped Ions

Single-qubit gates use laser-driven Raman transitions between hyperfine ground states (gate time $\sim 1-10\mu \text{s}$), or direct microwave driving of the hyperfine splitting ($\sim 4-35\mu \text{s}$ for ${}^{43}{\text{Ca}}^{+}$). All-electronic microwave control avoids laser-induced photon scattering entirely.

Two-qubit gates exploit shared motional (phonon) modes of the ion chain:

• Mølmer–Sørensen (MS) gate: bichromatic laser fields create a spin-dependent force, generating an $XX$ interaction via transient phonon excitation ($\sim 30-200\mu \text{s}$)
• Cirac–Zoller gate: sequential sideband pulses transfer information through a single phonon; historically first but slower and less robust than MS

Semiconductor Spin Qubits

Single-qubit gates rotate individual electron or hole spins via:

• ESR (Electron Spin Resonance): oscillating magnetic fields at the Larmor frequency
• EDSR (Electric Dipole Spin Resonance): oscillating electric fields coupled to spin via spin–orbit interaction or micromagnet field gradients; faster and more locally addressable than ESR

Gate times are $\sim 0.1-10\mu \text{s}$ depending on driving mechanism and Rabi frequency.

Two-qubit gates exploit the exchange interaction $H_{\text{ex}}=J(t){\hat{\mathbf{S}}}_1\cdot {\hat{\mathbf{S}}}_2$ controlled by the barrier gate voltage between neighboring quantum dots. Pulsing $J(t)$ implements $\sqrt{\text{SWAP}}$ or CZ-like operations in $\sim 1-100\text{ns}$.

Neutral Atoms (Rydberg)

Single-qubit gates use global or site-selective Raman transitions between hyperfine ground states of alkali atoms (${}^{87}\text{Rb}$, ${}^{133}\text{Cs}$) held in optical tweezer arrays. Gate times are $\sim 1-10\mu \text{s}$.

Two-qubit gates exploit the Rydberg blockade: when one atom is excited to a Rydberg state ($n\sim 50-100$), the strong dipole–dipole interaction ($\sim \text{GHz}$ at $\mu \text{m}$ separations) shifts the doubly-excited state out of resonance, implementing a conditional phase (CZ) gate. Gate times are $\sim 0.1-1\mu \text{s}$.

Photonic Qubits

Single-qubit gates are implemented with passive linear-optical elements: beam splitters (Hadamard), phase shifters (rotations about $\hat{z}$), and wave plates (arbitrary SU(2)). These are deterministic and high-fidelity.

Two-qubit gates are fundamentally challenging because photons do not naturally interact. The KLM scheme (Knill, Laflamme, Milburn 2001) achieves probabilistic entangling gates using ancilla photons, beam splitters, and post-selection on photon detection. Success probability is $\sim 1/16$ for a bare CZ, improved to near-deterministic rates with teleportation-based boosting. Measurement-based (fusion) approaches sidestep this by consuming entangled resource states.

Key Metrics

Metric	Platform	SOTA Value	Gate Time	Reference
1Q fidelity	Trapped ion	99.999985%	4–35 μs	Oxford 2025, ${}^{43}{\text{Ca}}^{+}$ microwave
1Q fidelity	Superconducting	99.998%	~25 ns	Ding et al. 2025, fluxonium, MIT
1Q fidelity	Superconducting (transmon)	99.9926%	~25 ns	Z. Li et al. 2023
1Q fidelity	Spin qubit	>99.9%	~1 μs	Mądzik et al. 2022, UNSW Si donor
1Q fidelity	Neutral atom	>99.9%	~1 μs	Raman/microwave on ${}^{87}\text{Rb}$
2Q fidelity	Trapped ion	99.99%	~100 μs	IonQ 2025, EQC, R&D lab
2Q fidelity	Trapped ion (production)	99.914%	~200 μs	Quantinuum H1-1, all pairs
2Q fidelity	Superconducting (CZ)	99.95%	~60 ns	IQM 2025, PALEA protocol
2Q fidelity	Neutral atom (CZ)	99.6%	~0.5 μs	Atom Computing 2024
2Q fidelity	Spin qubit	>99%	~10–100 ns	CMOS-compatible Si, multiple groups (2024)

Fidelities are measured via randomized benchmarking (RB) or interleaved RB, which isolate gate errors from state-preparation and measurement (SPAM) errors.

Gate Error Sources

Gate infidelity $1-F$ arises from several distinct physical mechanisms:

• Decoherence during the gate: $T_1$ (energy relaxation) and $T_2$ (dephasing) processes degrade the qubit state during the finite gate duration $t_g$. The coherence-limited error scales roughly as $\sim t_g/T_2$, making fast gates essential.

• Leakage to non-computational states: in weakly anharmonic systems (transmons, $|\alpha |\sim 300\text{MHz}$), population can leak to $|2⟩$ or higher levels during fast pulses. DRAG pulses and leakage reduction protocols (e.g., IQM’s PALEA) specifically target this error.

• Calibration drift: qubit frequencies, coupling strengths, and pulse amplitudes drift on timescales of minutes to hours due to TLS fluctuations, flux noise, or thermal effects. Requires periodic recalibration and drift-tracking protocols.

• Crosstalk: in multi-qubit processors, driving one qubit inadvertently perturbs neighbors through residual coupling (always-on $ZZ$), microwave leakage, or shared control lines. This produces correlated errors that are harder to correct.

• Control imperfections: finite pulse bandwidth, DAC quantization, IQ mixer imbalance, and timing jitter all contribute systematic coherent errors.

• SPAM errors (state preparation and measurement) are distinct from gate errors but often conflated in raw process metrics. Randomized benchmarking specifically separates gate error from SPAM.

Historical Milestones

Year	Milestone	Reference
1995	First experimental CNOT gate (trapped ion, ${}^9{\text{Be}}^{+}$)	Monroe et al., PRL 75, 4714
1998	First NMR quantum gates demonstrated	Chuang et al., Nature 393, 143
2003	First solid-state two-qubit gate (superconducting charge qubits)	Yamamoto et al., Nature 425, 941
2007	Transmon qubit introduced, enabling high-fidelity SC gates	Koch et al., PRA 76, 042319
2009	DRAG pulse invented for leakage suppression	Motzoi et al., PRL 103, 110501
2014	Surface code threshold crossed in superconducting qubits	Barends et al., Nature 508, 500
2016	First $>99.9\%$ two-qubit MS gate (trapped ions)	Ballance et al., PRL 117, 060504
2023	$99.5\%$ two-qubit CZ on 60 neutral atoms simultaneously	Evered et al., Nature 622, 268
2024	Google Willow: below-threshold surface code error correction	Google QAI, Nature 638, 920
2025	Oxford: 1Q error rate $1.5\times 10^{-7}$ (trapped ion, microwave)	Sheridan et al., PRL (2025)
2025	IonQ: 99.99% 2Q fidelity (trapped ion, EQC)	Moses et al., arXiv:2510.17286
2025	IQM: 99.95% 2Q CZ fidelity (superconducting, PALEA)	Landra et al., arXiv:2508.16437

Linked Papers

• heeres-2017-implementing-universal-gate

Related Entries

• transmon — dominant superconducting qubit platform
• fluxonium — alternative SC qubit with higher anharmonicity, record 1Q fidelity
• trapped-ion-qubit — leading platform for 2Q gate fidelity
• spin-qubit — semiconductor spin qubits with ultrafast exchange gates
• silicon-spin-qubit — CMOS-compatible spin qubit implementations
• neutral-atom-qubit — Rydberg-mediated entangling gates
• rydberg-neutral-atom-qubit — Rydberg blockade mechanism for 2Q gates
• qubit-readout — measurement (distinct from gate operations)
• circuit-qed — architecture enabling SC qubit control and readout
• tunable-coupler — mediates high-fidelity 2Q gates in SC processors
• cirac-zoller-gate — original trapped-ion 2Q gate proposal
• molmer-sorenson-gate — workhorse trapped-ion entangling gate
• dual-rail-photonic-qubit — photonic encoding for linear-optical gates
• linear-optical-photonic-qubit — KLM-scheme photonic gates
• exchange-only-qubit — all-exchange spin qubit gates
• surface-code-logical-qubit — error correction enabled by high gate fidelities
• nuclear-magnetic-resonance-qubit — NMR platform; first experimental quantum algorithm demonstrations