A pulsed laser excites a quantum dot's exciton state once per period Trep. The exciton decays radiatively with lifetime τ, emitting at most one photon per pulse in the ideal case. Each photon is routed at random to detector D1 or D2 (a Hanbury Brown–Twiss beamsplitter), and every pair click is binned by delay into a coincidence histogram:
g²(τ) = ⟨n₁(t) n₂(t+τ)⟩ / (⟨n₁⟩⟨n₂⟩)
This 2D view drops the 3D beam-table entirely: every click is instead plotted on a phase wheel — its angle is the click's arrival time modulo Trep (so every period lands on top of every other period, wrapping the pulsed excitation onto a circle) and its radius grows with how many periods ago it happened, so recent clicks sit near the rim and older ones spiral inward as faint rings. A true single-photon emitter draws two separate arcs (D1, D2) that barely overlap in phase; re-excitation and background counts scatter extra dots across all phases, which is exactly what fills in the central g²(0) peak in the histogram panel on the right.
- Re-excitation — after emitting, the dot can be driven back up and emit a second photon within the same pulse, before the next pulse arrives.
- Background / dark counts — detector dark counts or stray scattered light add uncorrelated clicks that also raise the central peak.
g²(0) is estimated as the central-peak area divided by the mean area of reference side peaks (normalized to 1 for a Poissonian source); purity is reported as 1 − g²(0). Real quantum dot sources report g²(0) below 0.01–0.05; g²(0) < 0.5 is the accepted threshold for calling a source "single-photon". Drag the wheel to rotate it, scroll to zoom its radial (history-depth) scale, and hover a histogram bar for its exact delay and count.