Trapped-Ion Qubits
The platform behind IonQ and Quantinuum, with the best coherence times in the field
By the end of this topic you'll be able to
Where superconducting qubits are entirely artificial circuits, trapped-ion qubits use actual atoms: individual ions, commonly ytterbium or calcium, suspended in free space by oscillating electric fields in a Paul trap.
Because every ion of a given species is physically identical, unlike fabricated superconducting circuits that always carry some manufacturing variation, trapped-ion qubits are naturally uniform devices — and they tend to have the longest coherence times of any current qubit platform, seconds or longer, against microseconds for superconducting qubits.
The qubit itself is encoded in two long-lived internal energy states of the ion, typically hyperfine ground-state sublevels, which are extremely well isolated from environmental noise precisely because they don't involve the ion's more fragile excited electronic states.
Before any quantum operation begins, ions are laser-cooled — first via the Doppler effect, then, for the final small step, via resolved sideband cooling — down to near their motional ground state.
Single-qubit gates are implemented with precisely tuned laser or microwave pulses driving transitions between the two hyperfine states. Two-qubit gates are where this platform's real distinctiveness appears: rather than a direct qubit-qubit interaction, ions confined together in the same trap share a collective vibrational, or phonon, mode — a shared mechanical 'wobble' of the entire chain.
The Mølmer-Sørensen gate, introduced by Mølmer and Sørensen in 1999, uses laser pulses to temporarily and virtually excite this shared motion, using it as a mediator to entangle two ions that have no other direct coupling to each other — a fundamentally different mechanism from a superconducting CNOT, which relies on near-direct physical coupling between adjacent qubits on a chip.
The price for these excellent coherence properties is speed: gate operations on trapped ions typically run slower than on superconducting circuits, and scaling to very large numbers of ions in a single trap, or reliably linking multiple traps together, remains an active engineering challenge — the central reason trapped-ion companies have so far favored fewer, higher-fidelity qubits over a 'more, noisier qubits' strategy.
Try It Yourself
A trapped-ion qubit has T2 = 1 second, and a Mølmer-Sørensen gate takes 10 μs. Using the budget rule 'keep cumulative decoherence probability under 1%,' roughly how many gates can be run in sequence before that budget is used up?
- 1Convert to consistent units: T2 = 1 s = 1,000,000 μs, and gate time = 10 μs.
- 2The cumulative decoherence probability after N gates scales roughly as N × (gate time)/T2 for N not too large.
- 3Set this equal to the 1% (0.01) budget and solve for N: N × 10/1,000,000 = 0.01, so N = 0.01 × 1,000,000/10 = 1,000.
Roughly 1,000 gates before the cumulative decoherence budget is used up — orders of magnitude more than a superconducting qubit's equivalent budget (compare the ~0.2%-per-gate figure from Superconducting Qubits), directly reflecting trapped ions' much longer T2.
Reference
| Paul trap | Physically suspends ions in a vacuum chamber | |
| Qubit encoding | Long-lived, environmentally isolated internal states | |
| Two-qubit gate | Uses a shared vibrational (phonon) mode to mediate entanglement | |
| Typical coherence time | Substantially longer than superconducting qubits |
Quick Check
Why do trapped-ion qubits tend to have much longer coherence times than superconducting qubits?
How does the Mølmer-Sørensen gate entangle two trapped ions?