Quantum Physics

Hong-Ou-Mandel Effect: Two Photons That Refuse to Split

Slide a beam splitter by just 16 micrometers and a steady stream of paired-photon "clicks" at two detectors abruptly vanishes to nearly zero. That plunge — the famous Hong-Ou-Mandel dip, only about 50 femtoseconds wide in time — is one of the cleanest fingerprints of quantum indistinguishability ever recorded. It was first seen in 1987 by Chung-Ki Hong, Zhe-Yu Ou, and Leonard Mandel at the University of Rochester.

The Hong-Ou-Mandel (HOM) effect is a two-photon interference phenomenon: when two identical photons enter the two input ports of a 50/50 beam splitter at the same instant, they always leave together through the same output port. They never split up. Because coincident detections require one photon in each output, the coincidence rate collapses — not because photons are absorbed, but because two indistinguishable quantum paths cancel.

  • TypeTwo-photon quantum interference
  • Discovered1987 — Hong, Ou & Mandel (Univ. of Rochester)
  • Key equationP_coinc = (1/2)[1 - exp(-(Δω·Δτ)^2)]
  • RegimeSingle-photon (Fock) states, balanced 50/50 beam splitter
  • Typical dip width~50 fs (temporal) / ~16 μm (path length)
  • Observed inSPDC photon pairs, quantum dots, trapped ions, atoms, phonons, plasmons

Interactive visualization

Press play, or step through manually. The visualization is yours to drive — try it before reading on.

Open visualization fullscreen ↗

Watch the 60-second explainer

A condensed visual walkthrough — narrated, captioned, under a minute.

The physical setup: two photons meet at a beam splitter

The HOM experiment is deceptively simple. A nonlinear crystal pumped by a UV laser undergoes spontaneous parametric down-conversion (SPDC), splitting a pump photon into a pair of lower-energy signal and idler photons that are born at the same instant and are nearly identical in frequency, polarization, and spatial mode. In the original 1987 work the pump was a 351.1 nm argon-ion line and the pairs emerged near 702 nm.

  • Each photon is sent to one of the two input ports of a 50/50 beam splitter.
  • A detector sits at each of the two output ports, wired into a coincidence counter that registers an event only when both fire within a nanosecond window.
  • A movable mirror or the beam splitter itself adjusts the relative arrival-time delay Δτ between the two photons, scanned over micrometers of path.

Classically, two independent particles hitting a 50/50 splitter should land in opposite outputs 50% of the time, giving a steady baseline coincidence rate. Quantum mechanics predicts that rate drops to zero when the photons arrive simultaneously.

The mechanism: destructive interference of two-photon amplitudes

The magic lives in the beam splitter's phase relations. A lossless 50/50 splitter transmits with amplitude t = 1/√2 and reflects with amplitude r = i/√2 (the factor of i, a π/2 phase, enforces unitarity). Label the inputs a and b and outputs c and d. Two indistinguishable photons — one in a, one in b — can reach a coincidence (one in c, one in d) two ways:

  • Both transmitted: amplitude t·t = 1/2.
  • Both reflected: amplitude r·r = (i/√2)(i/√2) = -1/2.

Because the photons are identical, these two histories are indistinguishable, so their amplitudes add: 1/2 + (-1/2) = 0. The coincidence amplitude cancels exactly. In second-quantized form, the input state a†b† transforms so the cross term vanishes, leaving only c†c† and d†d† — both photons in the same port. This is bosonic bunching: two-photon N00N-like states (|2,0> and |0,2>) at the output. No energy is lost; probability is merely redistributed away from the coincidence channel.

Key quantities and a worked example

For photons with a Gaussian spectral profile of RMS bandwidth Δω, the coincidence probability as a function of delay Δτ is:

P_coinc(Δτ) = (1/2)·[1 - V·exp(-(Δω·Δτ)²)]

where V is the visibility (0 ≤ V ≤ 1), set by how indistinguishable the photons are. At Δτ = 0 with perfect overlap (V = 1), P_coinc = 0 — the bottom of the dip. For large |Δτ|, the exponential dies and P_coinc → 1/2, the classical baseline.

  • Dip width ≈ 1/Δω, the photon coherence time. Original 1987 dip: HWHM ≈ 16 μm of path, i.e. Δτ ≈ 50 fs, matching a ~10 nm bandwidth around 702 nm.
  • Visibility: V = (C_max − C_min)/C_max. The 1987 experiment measured about 0.75; modern quantum-dot and SPDC sources exceed 0.99.
  • Energy per 702 nm photon ≈ 1.77 eV (2.83×10⁻¹⁹ J); the pump at 351 nm carries ≈ 3.53 eV, split into the two.

How it is observed, measured, and used

Experimentally you scan the delay Δτ and plot coincidences: the curve is flat at the baseline, plunges into the dip near Δτ = 0, and recovers. Modern time-resolved coincidence methods with sub-picosecond timing tags, superconducting-nanowire single-photon detectors (SNSPDs, ~15 ps jitter, >90% efficiency), and narrowband filtering routinely push visibilities past 99%.

  • Photon-source characterization: HOM visibility is the standard metric for how indistinguishable single-photon emitters (quantum dots, trapped ions, defect centers) really are — essential for scalable photonic quantum computing.
  • Linear-optical quantum computing: HOM interference is the two-qubit gate mechanism behind the Knill-Laflamme-Milburn scheme and boson sampling.
  • Quantum metrology: the dip position tracks Δτ with attosecond-scale sensitivity, enabling ultraprecise optical delay and biological-thickness measurements.
  • Bell-state measurements: HOM at a beam splitter is the workhorse of entanglement swapping and quantum teleportation / repeater nodes.

How it differs from classical and single-photon interference

HOM is often confused with ordinary interference, but the distinction is fundamental:

  • vs. single-photon interference (Mach-Zehnder): there, one photon interferes with itself and you see sinusoidal fringes in the single-count rate. HOM shows no fringes in singles — each detector's individual rate stays flat. The interference appears only in the correlation between two detectors.
  • vs. classical Hanbury Brown-Twiss (HBT): thermal light also bunches, but that is a statistical correlation, not amplitude cancellation; classical fields cannot drive coincidences below the 50% baseline. A HOM dip deeper than 50% visibility has no classical explanation.
  • vs. fermions: swap the bosonic photons for identical fermions (e.g. electrons in an electronic HOM) and antisymmetry flips the sign — particles anti-bunch, exiting opposite ports, producing a coincidence peak instead of a dip.

The effect requires quantized Fock states (definite photon number) — it is a genuinely non-classical, particle-statistics phenomenon.

Significance, extensions, and open questions

The HOM effect is a cornerstone of quantum optics because it turns an abstract idea — bosonic indistinguishability — into a directly measurable dip. Its visibility is the operational definition of "how identical" two photons are, and near-unity HOM interference is a prerequisite for fault-tolerant photonic quantum computers.

  • Beyond photons: HOM-type interference has now been demonstrated with plasmons, phonons, single atoms, trapped ions, and even distinguishable-color frequency-entangled photons (2025), showing the effect survives large spectral mismatch when entanglement restores indistinguishability.
  • Dispersion cancellation: HOM interferometers are intrinsically insensitive to even-order dispersion, a resource for quantum-enhanced sensing.
  • Open frontiers: pushing visibility to 99.9%+ from independent, remote sources; scaling multiphoton generalizations (three- and N-photon interference, the basis of boson sampling); and using HOM as a probe of the boson/fermion boundary in engineered anyonic systems.

Leonard Mandel, who died in 2001, is widely regarded as a founder of modern quantum optics; the 1987 dip remains one of his most cited results.

HOM two-photon interference vs. related interference phenomena
PhenomenonInterfering objectsWhat cancelsSignature
Hong-Ou-MandelTwo independent photons (bosons)Two-photon amplitudes |t,t> - |r,r>Dip in coincidences to ~0
Single-photon (Mach-Zehnder)One photon with itselfSingle-particle path amplitudesSinusoidal fringes in counts
Classical (Young's slits)Classical field with itselfField amplitudes E1 + E2Intensity interference fringes
Fermionic HOM (electrons)Two identical fermionsAntisymmetry forces anti-bunchingPhotons bunch; fermions anti-bunch
HBT (Hanbury Brown-Twiss)Two independent thermal photonsNothing (correlation, not cancellation)Bunching peak g^(2)(0) > 1

Frequently asked questions

Why do the two photons refuse to split at the beam splitter?

Two indistinguishable photons entering opposite ports can produce a coincidence (one photon in each output) via two paths: both transmitted or both reflected. The beam splitter's reflection phase makes these two amplitudes exactly equal and opposite (+1/2 and -1/2), so they cancel. Only the outcomes where both photons exit the same port survive, so the pair always bunches together.

What is the Hong-Ou-Mandel dip?

It is the sharp drop in the two-photon coincidence rate as you scan the relative arrival-time delay of the two photons. When the delay is zero and the photons are perfectly indistinguishable, coincidences fall to near zero. The dip is roughly as wide as the photons' coherence time — about 50 femtoseconds (16 micrometers of path) in the original 1987 experiment.

Who discovered the Hong-Ou-Mandel effect and when?

Chung-Ki Hong, Zhe-Yu Ou, and Leonard Mandel demonstrated it in 1987 at the University of Rochester, publishing in Physical Review Letters (volume 59, page 2044). They used photon pairs from spontaneous parametric down-conversion in a nonlinear crystal and measured the coincidence dip by scanning the beam splitter position.

What does HOM visibility tell you?

Visibility V = (C_max - C_min)/C_max quantifies how indistinguishable the two photons are. A visibility of 1 means the dip reaches zero and the photons are perfectly identical in every degree of freedom; a visibility near 0 means they are fully distinguishable and no dip appears. It is the standard benchmark for single-photon sources, with the best modern sources exceeding 0.99.

Is the Hong-Ou-Mandel effect the same as classical interference?

No. Classical (single-photon or wave) interference shows up as fringes in individual detector counts, and classical light cannot push coincidences below the 50% baseline. HOM interference appears only in the correlation between two detectors and can drop coincidences all the way to zero. A dip deeper than 50% visibility has no classical explanation and proves the light's quantum, particle-number character.

What is the Hong-Ou-Mandel effect used for?

It is central to photonic quantum technology: characterizing how indistinguishable single-photon sources are, implementing two-qubit gates in linear-optical quantum computing and boson sampling, performing Bell-state measurements for quantum teleportation and repeaters, and enabling ultra-precise optical delay and dispersion-cancelled sensing at attosecond timing resolution.