Spin Qubits in Silicon
Silicon spin qubits store information in the spin of a single electron or nucleus, promising compatibility with semiconductor manufacturing.
Spin as a qubit
A spin qubit uses the two spin orientations, up and down, of a single electron (or nuclear spin) confined in a semiconductor. In a magnetic field the two states split in energy by the Zeeman effect, forming a clean two-level system. Silicon is attractive because it can be isotopically purified to silicon-28, which has no nuclear spin, giving the electron a very quiet magnetic environment and long coherence.
Where the spin lives
- Gate-defined quantum dots: electrodes on a silicon or SiGe heterostructure trap one electron in a small potential well
- Donor spins: a single phosphorus atom implanted in silicon supplies an electron and a nuclear spin, as in the Kane proposal
- MOS structures that reuse standard transistor stacks
Control and readout
Single-qubit gates use electron spin resonance (an oscillating magnetic field) or electric dipole spin resonance via a micromagnet gradient, driven by microwave pulses. Two-qubit gates use the exchange interaction: bringing two dots close lets their electron wavefunctions overlap, and pulsing that overlap performs entangling operations in nanoseconds. Readout converts spin to charge via spin-to-charge conversion (Pauli spin blockade or energy-selective tunneling) sensed by a nearby charge detector.
Promise and problems
Spin qubits are tiny, tens of nanometers, so in principle millions could fit on a chip made in existing foundries. Coherence times reach milliseconds in purified silicon. The hurdles are device-to-device variability, the need for individual tuning of many gate voltages, charge noise, and dense local wiring. Two-qubit fidelities above 99 percent have been demonstrated at small scale.
Their manufacturing lineage is the central bet: if fabrication yield and uniformity reach transistor-industry levels, spin qubits could scale where hand-tuned modalities struggle.