Quick summary
An electron behaves like a tiny quantum magnet. In a silicon device, electric gates confine one or a few electrons inside a quantum dot. Two spin orientations, or encoded combinations of several spins, form the qubit states. Pulses manipulate the state, neighbouring electrons interact to create two-qubit gates, and sensors convert spin information into an electrical signal.
Building the confinement
Metal gates above a semiconductor channel shape an electrostatic potential that traps electrons. The dot must hold the intended charge state while barriers control tunnelling to reservoirs and neighbouring dots. Tiny differences in interfaces, strain or impurities can change the local environment, so fabrication uniformity and tuning are central challenges.
Writing, controlling and reading
- Initialize: prepare a known spin state using energy selection or measurement.
- Control: microwave magnetic or electrically mediated signals rotate the spin, or voltage pulses manipulate encoded exchange-only qubits.
- Entangle: tune the exchange interaction between neighbouring electrons.
- Read: map spin to charge, then detect the charge with a nearby sensor.
Why silicon is attractive
Silicon benefits from mature semiconductor fabrication and can host small qubits with potentially dense integration. Isotopically enriched silicon reduces magnetic noise from nuclei, supporting longer coherence. Electrical gates also offer a path toward integration with classical control electronics.
Why scaling remains difficult
Qubits usually operate at extremely low temperatures. Thousands or millions of separate room-temperature wires would carry heat and occupy space. Cryogenic control, multiplexing and shared signal delivery can reduce wiring, but introduce power, crosstalk and calibration constraints. Dense arrays also need two-dimensional connectivity or reliable shuttling of quantum states.
Uniformity and calibration
Each quantum dot may need individual voltages and pulse calibration. Noise, charge rearrangement and device drift can alter gate performance. Automation can tune arrays, but must distinguish a stable qubit from a temporarily favourable setting. Manufacturing compatibility does not mean current fabs can produce large error-corrected processors without new process control.
Reality check
A demonstration controlling many quantum dots is not necessarily the same as running that many high-quality qubits simultaneously. Count active qubits, connected gates, fidelities, measurement performance and duration. A compact device is only one part of a computer that also needs refrigeration, control, decoding and error correction.
What would show decisive progress?
Look for repeatable manufacture across wafers, automated tuning, high-fidelity neighbouring and non-neighbouring operations, cryogenic control within the heat budget and logical error suppression as code size grows. These system results matter more than similarity to conventional transistor dimensions.
