How a trapped-ion quantum computer works

Our quantum computer will store information in single calcium ions, held in empty space by electric fields and controlled with lasers. Here is how, step by step.

1

A qubit is a tiny, well-behaved two-state system

A normal computer bit is 0 or 1. A quantum bit, or qubit, can be in a blend of both at once, and several qubits can be linked so that they share a single state. That is where a quantum computer's power comes from.

The hard part is finding a qubit that nature leaves alone. A single atom is ideal: every calcium ion is identical, and one of its electrons can sit in two energy levels that last for over a second. Isolate the ion from everything else and those two levels become a very clean 0 and 1.

2

Holding an ion in empty space

An ion carries an electric charge, so electric fields can push it around. But a rule of physics called Earnshaw's theorem says no arrangement of fixed charges can hold it still. Any field that pushes it back towards the centre in one direction lets it escape in another, like a ball on a horse saddle.

left–rightup–down
The trick: flip the saddle millions of times a second. Before the ion can roll off one way, the slope reverses.

A Paul trap does exactly that, using a radio-frequency voltage of about a thousand volts oscillating 20 to 30 million times a second. On average, the ion feels a gentle bowl and stays at the centre.

Try it: how hard can you drive the trap?

The slider sets the trap strength, a number physicists call q. The line shows the ion's position over time.

Two motions appear together. The slow swing, secular motion, is the ion rolling in the average bowl. The fast jitter on top, micromotion, is the ion being shaken by the flipping field. Push q past about 0.91 and the trap throws the ion out. Our trap is designed to run at q of 0.4 or less.

3

Making the ion

We start with a small oven that releases a thin vapour of calcium atoms into the vacuum chamber. Two lasers turn one of these atoms into an ion inside the trap.

ionization limit6.11 eVelectron free, Ca+ left behind4s² 1S04s4p 1P10 eV2.93 eV423 nmtuned to calcium only< 390 nmenough energy to escapeNeutral calcium atom, energies to scale

The first laser, at 423 nm, is tuned exactly to calcium, so it lifts an electron only in calcium atoms and ignores anything else in the chamber. The second, near 390 nm, has enough energy to tear that electron away. What is left is a positive Ca+ ion, and the trap catches it.

4

Cooling it with light

A freshly made ion is hot and moves fast. To slow it down we shine a violet laser at 397 nm, tuned slightly below the colour the ion absorbs.

Because of the Doppler effect, an ion moving towards the laser sees the light shifted up to exactly its colour, absorbs it, and gets a small kick backwards. Moving away, it barely absorbs at all. Millions of kicks a second slow it to a thousandth of a degree above absolute zero.

4s 2S1/24p 2P1/24p 2P3/23d 2D3/23d 2D5/2397 nmcooling, glowfalls here1 time in 15866 nmrepump: brings it backEnergies to scale; fine-structure gaps enlarged
About once in every 15 cycles, the ion falls into the dark 3d D3/2 level instead of back to the ground state. A second laser at 866 nm lifts it back so cooling can continue.

While it is being cooled, the ion scatters violet light in all directions. A sensitive camera can see this as a single glowing dot: one atom, visible.

5

Computing: flipping the qubit with a very steady laser

Our qubit is the ion's ground state (0) and a long-lived upper state called D5/2 (1). A deep-red laser at 729 nm moves the ion between them. Shine it for the right time and the qubit flips from 0 to 1; stop halfway and it is in an equal blend of both.

4s 2S1/24p 2P1/24p 2P3/23d 2D3/23d 2D5/2729 nmflips the qubitqubit 0qubit 1lives about 1.2 sEnergies to scale; fine-structure gaps enlarged

This only works if the laser keeps a perfectly regular rhythm. A laser whose colour wobbles by more than a few hertz, out of 411 trillion, quickly scrambles the qubit.

Try it: how steady does the laser need to be?

The line shows the chance of finding the qubit in state 1 as the pulse gets longer. The slider sets how much the laser's colour wobbles (its linewidth).

To make qubits work together, we use the fact that ions in a chain push on each other. When one ion is kicked, the whole chain wobbles, and that shared motion lets a laser link the state of one ion to another.

6

Reading the answer

To read a qubit, we turn the 397 nm laser back on. If the ion is in 0, it scatters light and glows. If it is in 1, it is parked in the upper state, out of reach of the violet light, and stays dark. Bright means 0, dark means 1.

Afterwards, a laser at 854 nm lifts the ion from the upper state to a short-lived level, from which it drops straight back to 0. The ion is reset and ready for the next run.

4s 2S1/24p 2P1/24p 2P3/23d 2D3/23d 2D5/2854 nmdrops back to 0in nanosecondsEnergies to scale; fine-structure gaps enlarged

Six lasers, six jobs

LaserStageWhat it doesHow steady it must be
423 nmMake the ionFirst step of turning a calcium atom into an ion1–5 MHz
~390 nmMake the ionSecond step: knocks the electron freenot critical
397 nmCool itSlows the ion; makes it glow1–5 MHz
866 nmKeep coolingRescues the ion when it falls into a dark state1–10 MHz
729 nmComputeFlips the qubit between 0 and 11–10 Hz
854 nmResetEmpties the qubit's upper state1–5 MHz

The qubit laser has to be about a million times steadier than the others.

7

Where our lab is today

Optical table in the Delta Q lab with mirrors, mounts and an acousto-optic modulator
The first optical table in the Delta Q lab

In our first year we designed a blade-type linear Paul trap for calcium ions and worked out how to run it:

Trap size
About 1 mm from centre to electrode
Drive
1 to 1.5 kV at 20 to 30 MHz
Trap strength
q of 0.4 or less
Ion's swing
1 to 5 million times a second

We also set up the lab's power and grounding, the optical table, and an acousto-optic modulator that switches and fine-tunes laser beams. Next come machining the trap, the RF drive, the ultra-high vacuum, and the lasers, until we see our first glowing ion.

Adapted from the 2025–26 work report of Vishwa Chandrasekaran, Junior Research Fellow, Delta Q.