Exchange Interaction in Quantum Dots

The Heisenberg exchange interaction between electron spins in adjacent quantum dots is the workhorse two-qubit coupling for semiconductor spin qubits. It arises microscopically from the interplay of quantum tunneling and Coulomb repulsion: when two electrons occupy neighboring dots separated by a tunable tunnel barrier, virtual hopping events — where one electron briefly visits the other’s dot — lower the energy of the spin-singlet state relative to the spin-triplet. In the simplest Hubbard model, the exchange splitting is , where is the interdot tunnel coupling and is the on-site Coulomb energy. This perturbative result holds when , the regime where the dots are singly occupied and the Heisenberg Hamiltonian faithfully describes the low-energy physics.

This note is the device-and-materials view of exchange. Use it when the question is how gate voltages create , why barrier control and detuning control behave differently, or where charge sensitivity enters. Once the charge sector has been integrated out and the question becomes what unitary an exchange pulse implements inside a logical spin subspace, switch to heisenberg-exchange-in-quantum-dots.

Routing heuristic

  • Stay in this note when the interesting variable is a control knob: tunnel coupling, detuning, barrier symmetry, residual off-state exchange, or materials-specific charge sensitivity.
  • Switch to heisenberg-exchange-in-quantum-dots when the interesting variable is a pulse-area consequence: partial-SWAP angle, logical axis in an encoded manifold, or how pairwise exchange projects into a reduced qubit subspace.
  • Read both if you need the full chain from gate voltage to logical unitary.

Three-layer abstraction ladder

  1. Electrostatics and virtual hopping: gates set tunnel coupling, detuning, and charge admixture. That is this note.
  2. Effective spin Hamiltonian: those device knobs collapse to a calibrated scalar coupling multiplying . That is heisenberg-exchange-in-quantum-dots.
  3. Logical action: pulse areas and encoded projections become -style entanglers, logical splittings, or always-on control axes in specific qubit families.

Quick split from the Hamiltonian note

If the question is…Start here?Why
How do gates or detuning knobs create and shape ?YesThis note keeps the Hubbard / device picture in view.
Why does symmetric barrier control usually beat detuning control for noise?YesThe charge admixture and story lives here.
What unitary does a calibrated exchange pulse implement?No, go to heisenberg-exchange-in-quantum-dotsThat is the pulse-area / logical-action note.
Why do singlet-triplet, exchange-only, RX, and AEON inherit different logical axes from the same pairwise coupling?No, go to heisenberg-exchange-in-quantum-dotsThe projection into encoded subspaces is the real question there.

The power of the exchange interaction lies in its purely electrical tunability. Two complementary knobs exist: barrier control adjusts the height of the potential barrier between dots, modulating directly while keeping the charge distribution symmetric; detuning control tilts the double-well potential, shifting one dot’s energy relative to the other, which changes through the effective in the denominator. Barrier control is generally preferred in modern devices because it preserves the charge symmetry point, reducing sensitivity to charge noise. In practice, can be tuned over many orders of magnitude — from effectively zero (dots fully isolated, ) to tens of GHz (strong tunnel coupling) — by adjusting a single gate voltage on a timescale of nanoseconds.

At the device level, the important comparison is how cleanly a platform can manufacture and modulate . Singlet-triplet, exchange-only, RX, and AEON all inherit the same underlying exchange primitive, but the family-level engineering question here is upstream of the logical encoding: how steeply does respond to detuning versus barrier gates, how much residual off-state exchange survives, and how much charge admixture is reintroduced when fast control is demanded? In every case, the main fidelity risks enter through (detuning sensitivity), (barrier sensitivity), electrical noise on the control lines, and incomplete suppression of the interaction when it is supposed to be idle.

The exchange interaction also appears in extended forms: superexchange couples next-nearest-neighbor dots through a virtual intermediate occupation, mediator-dot variants turn the same logic into longer-range effective coupling, and the Hubbard parameters can be engineered in Si/SiGe, GaAs/AlGaAs, donors, and Si-MOS stacks with quantitatively different and scales. Understanding and controlling at the few-percent level is the central materials and device challenge for scaling spin-qubit processors.

Key relationships

  • loss-divincenzo-qubit — original proposal using -based as the native entangling gate
  • singlet-triplet-qubit — uses to rotate between and within the two-electron subspace
  • exchange-only-qubit — all control via exchange; no magnetic field gradients needed
  • aeon-qubit — always-on exchange variant operating at a sweet spot for noise protection
  • rx-qubit — always-on three-spin exchange encoding where resonant drive addresses the exchange-defined qubit splitting
  • hybrid-qubit — combines charge-like and spin-like states; exchange sets the energy scales
  • silicon-spin-qubit — Si/SiGe and Si-MOS platforms where exchange is the primary 2-qubit mechanism
  • sqrt-swap-as-universal-gate — the gate-level consequence once exchange pulses are treated abstractly
  • heisenberg-exchange-in-quantum-dots — companion note for the post-Hubbard, Hamiltonian-first view

References