Trapped-ion qubits

In short: Trapped-ion qubits store quantum information in the electronic or hyperfine energy levels of individual charged atoms held in vacuum by oscillating electric fields, and are controlled with precisely tuned lasers — giving them the longest coherence and highest gate fidelity of any platform, at the cost of slower gate speeds than superconducting circuits.

What a trapped-ion qubit actually is

A trapped-ion qubit is a single atom that has lost one electron, so it carries a net positive charge. That charge is the handle: because the ion is charged, electric fields can grip it and suspend it in free space inside an ultra-high-vacuum chamber (typically below 10⁻¹¹ mbar, cleaner than low-Earth orbit). The quantum information itself lives not in the ion's position but in two of its internal energy levels — two states of a valence electron or two hyperfine sub-levels of the nucleus-electron system. Popular ion species include ⁱⁱ⁷¹Yb⁺, ⁴⁰Ca⁺, and ⁹Be⁺, each chosen because it has a clean two-level system reachable with available laser wavelengths.

The reason physicists love these systems is simple: every ⁴⁰Ca⁺ ion in the universe is identical. There is no fabrication variation, no two-qubit "sorting" problem, no device-to-device spread in transition frequencies the way there is with lithographically patterned superconducting junctions. That uniformity is the single biggest structural advantage of the platform.

How the trap holds the ion

You cannot hold a charged particle in a stable position with static electric fields alone — Earnshaw's theorem forbids it. The workaround is the Paul trap (radio-frequency quadrupole trap), which applies an RF voltage, usually in the tens of MHz at a few hundred volts, to shaped electrodes. The field rotates faster than the ion can respond, producing a time-averaged "pseudopotential" that confines the ion to a point. Modern devices use surface (chip) traps: the electrodes are microfabricated on a planar substrate and the ions float roughly 50–100 µm above the surface. This is where semiconductor process knowledge meets quantum hardware — these chips are patterned with the same lithography, etch, and metallization steps used for classical ICs, and companies routinely use Cadence and Synopsys layout and electromagnetic-extraction tools, plus Siemens/Mentor Calibre for DRC, to design the electrode structures and route the many DC control lines.

Laser control: initialization, gates, and readout

Everything you do to a trapped ion, you do with light. Four jobs matter:

• Doppler cooling — a red-detuned laser removes kinetic energy until the ion's motion drops to near its quantum ground state, so the trap behaves like a clean harmonic oscillator.
• State preparation — optical pumping drives the ion into one well-defined qubit state with >99.9% reliability.
• Single-qubit gates — a focused beam (or microwave field) drives Rabi oscillations between the two levels; rotations take microseconds.
• Readout — a "cycling" laser makes one qubit state fluoresce brightly while the other stays dark. A PMT or camera counts photons: bright = 1, dark = 0, with detection fidelity often above 99.9%.

Two-qubit entangling gates exploit the shared motion of the ion chain. The Mølmer–Sørensen gate uses bichromatic laser beams to couple the qubits' internal states through a common vibrational mode (a "phonon bus"), entangling ions that never physically touch.

Coherence versus speed: the core trade-off

Trapped ions are the clear coherence champions. Because the qubit is a bare atom isolated in vacuum, it barely couples to its environment. Hyperfine qubits show coherence times (T₂) of seconds to over ten minutes with dynamical decoupling — compared with roughly 100–300 µs for a good superconducting transmon. That is a factor of a million or more.

The price is speed. A two-qubit ion gate typically takes 10–100 µs, because it must act through slow mechanical motion, while a superconducting two-qubit gate finishes in 20–60 ns — about 1,000× faster. The honest figure of merit is not raw time but the ratio of coherence to gate time (how many operations fit inside one coherence window), and here ions still win decisively.

All-to-all connectivity matters for algorithms: in a single ion chain any qubit can entangle with any other, so a circuit needs far fewer SWAP operations than on a 2D superconducting lattice — often cutting effective circuit depth substantially.

Scaling: the hard part

A single chain of 20–40 ions works well, but packing hundreds into one chain makes the vibrational modes dense and the gates slow and error-prone. The leading answer is the QCCD architecture (Quantum Charge-Coupled Device): many small trap zones on one chip, with DC electrodes used to physically shuttle ions between memory, gate, and readout regions — much as a classical CCD shuttles charge packets. Shuttling, splitting, and merging ion crystals without heating them is now routine in devices such as Quantinuum's H-series, and photonic interconnects are being developed to link separate trap chips into a network.

This is squarely an electronic-design problem as much as a physics one: routing hundreds of independent DC control lines, integrating on-chip photonics and DACs, and managing crosstalk all borrow directly from mainstream VLSI methodology. Teams doing this kind of mixed-signal and layout work often partner with specialists such as Avecas for custom chip-design services to close the gap between a physics lab prototype and a manufacturable control chip.

Where it fits in the quantum landscape

Trapped ions and superconducting circuits are the two most mature gate-model platforms, and they make opposite bets. If you want our side-by-side teardown of the competing approach, read our companion piece on superconducting qubits. For the error-correction angle — why coherence-rich, high-fidelity qubits dramatically lower the overhead of logical encoding — see our explainer on quantum error correction. If you want the full structured curriculum rather than single articles, browse the CourseTron course catalogue.

Ready to go deeper? CourseTron's Quantum Hardware Engineering track walks you from Paul-trap physics to QCCD control-chip design, with hands-on labs on gate calibration and readout. Explore the course →

Frequently asked questions

Why are trapped ions slower than superconducting qubits?

Their two-qubit gates work through the ions' shared mechanical vibration, which is inherently slow (tens of microseconds). Superconducting gates act through fast electrical couplings in nanoseconds. Ions trade speed for much longer coherence and higher fidelity.

Do trapped-ion computers need extreme cooling like superconducting ones?

Not a dilution refrigerator. The trap chip sits at or near room temperature in ultra-high vacuum; it is the ions' motion that is laser-cooled to near the ground state. Some systems use cryogenics to improve vacuum and reduce heating, but millikelvin fridges are not required.

What limits the number of qubits in a trapped-ion machine?

A single ion chain becomes unwieldy beyond a few dozen ions because the vibrational modes crowd together. Scaling relies on the QCCD approach — many trap zones with ions shuttled between them — and photonic links between chips.

Which companies build trapped-ion quantum computers?

Quantinuum (H-series) and IonQ are the best-known, alongside academic groups at NIST, Oxford, Innsbruck, and Maryland. They use species such as Yb⁺, Ba⁺, and Ca⁺.