Single-Particle Detection

Single-particle detection is the experimental practice of recording individual localized events: a click in a photodetector, a spot on a screen, a charge pulse in an electron multiplier, a droplet track in a chamber, or a pixel hit in a camera. It connects quantum probabilities to laboratory data because the formalism predicts distributions of outcomes, while the apparatus records events one run at a time.

Helpful background. The Born rule turns a state into outcome probabilities. POVMs extend that rule to detectors with loss, finite resolution, and general outcomes. For calibrated optical response matrices, number resolution, dark counts, dead time, and detector tomography, see quantum-optical Photon Counting.

From Beams to Single Events

Many early experiments measured continuous currents, photographic darkening, cloud tracks, or ensemble intensities. Those are still made of microscopic events, but the apparatus often averaged them before the experimenter saw the result. Modern detectors can record individual events with time stamps, positions, energies, or coincidence labels.

This changed the pedagogical picture of quantum mechanics. It became possible to say:

one run gives one localized event;
many runs build the probability distribution.

That slogan is useful, but it needs two cautions. First, a “single event” is still a detector response with finite efficiency, resolution, threshold, and background. Second, localized detections do not imply that the quantum state was a classical point-particle trajectory before detection. The detector record is a macroscopic outcome.

Detector Clicks

A detector click is an amplified physical event. For photons, the first microscopic step may be photoemission, avalanche multiplication, absorption in a superconducting nanowire, or excitation in a semiconductor. For electrons or ions, it may be secondary emission, scintillation, ionization, or charge collection. The detector turns a microscopic interaction into a macroscopic signal.

A detector model with outcomes $\{i\}$ assigns probabilities

$$p_i = \operatorname{Tr}(\rho E_i),$$

where $\rho$ is the incoming state and $\{E_i\}$ are positive effects summing to the identity. Include a no-click outcome when necessary. These effects can describe real imperfections; they are not restricted to an ideal detector. A calibrated model may incorporate:

• quantum efficiency;
• finite spatial and temporal resolution;
• dark counts or background events;
• dead time after a detection;
• saturation at high rates;
• threshold and discrimination electronics;
• calibration uncertainty.

Thus “the detector clicked” is a physical statement, not a bare mathematical primitive.

Photon Counting

Photon counting detectors are central in quantum optics, spectroscopy, astronomy, and quantum information. Common technologies include photomultiplier tubes, avalanche photodiodes, transition-edge sensors, and superconducting nanowire single-photon detectors.

For an on/off detector that registers each incident photon independently with the same efficiency $\eta$, has no dark counts, and is active at the start of the gate, the probability of no click for an $n$-photon input is $(1-\eta)^n$. If the photon-number distribution is $P_n$, then

$$P_{\rm click} = 1 - \sum_{n=0}^{\infty} (1-\eta)^n P_n.$$

With an independent dark-count probability $d$ per detection gate, a simple model is

$$P_{\rm click} = 1 - (1-d) \sum_{n=0}^{\infty} (1-\eta)^n P_n.$$

This formula is not universal detector theory. It is a useful reminder that a click probability depends on both the quantum state and detector imperfections.

Photon counting also clarifies why “intensity” is not always enough. A weak coherent state, a heralded single-photon state, and thermal light can have the same mean photon number but different count statistics and correlations.

Electron and Ion Detection

Electrons and ions are charged, so they can be detected through charge collection, secondary emission, scintillation, or position-sensitive amplification. A phosphor screen, photographic plate, microchannel plate, silicon detector, or electron multiplier converts an incoming particle into a visible spot or electrical pulse.

Electron detection made single-event interference especially vivid. In low-intensity electron interference, spots arrive one at a time. The accumulated histogram forms an interference pattern. The lesson is not that one electron literally paints the whole fringe pattern in one run. The lesson is that the probability distribution governing many localized events contains interference terms.

For an ideal measurement of a particle’s position at a fixed time, bins $\Delta_i$ have probabilities

$$p_i = \int_{\Delta_i} dx\,\lvert\psi(x)\rvert^2.$$

A particle arriving at a screen is also selected by an arrival process. The formula applies to a suitably normalized transverse detection profile; it is not a general arrival-time law for an arbitrary three-dimensional wavefunction. Real detectors require point-spread functions, thresholds, backgrounds, and efficiency maps.

Building Interference From Events

The Double-Slit Experiment gives the cleanest logic. The amplitude at a screen coordinate is

$$\psi(x) = \psi_1(x)+\psi_2(x),$$

so the probability density is

$$\lvert\psi(x)\rvert^2 = \lvert\psi_1(x)\rvert^2 + \lvert\psi_2(x)\rvert^2 + 2\operatorname{Re} \left[ \psi_1^*(x)\psi_2(x) \right].$$

Single detections sample this distribution. A histogram after $N$ events estimates the probabilities. If the expected probability for bin $i$ is $p_i$, the expected count is

$$\mathbb E[N_i] = Np_i.$$

For independent trials with fixed $N$, the binomial or multinomial fluctuations scale like

$$\sigma_{N_i}=\sqrt{Np_i(1-p_i)}.$$

For a Poisson count with mean $\mu$, the exact standard deviation is $\sqrt\mu$. A large observed count $N$ gives the common estimated uncertainty

$$\sigma_N \simeq \sqrt N.$$

This is why single-event pictures can look random at first and structured later. The structure is in the probability distribution, not in any one detection event.

Coincidences and Correlations

Single-particle detection also made coincidence experiments possible. Two or more detectors record time-stamped events, and the experimenter studies correlations between them.

Coincidence logic is central to:

• photon antibunching;
• heralded photon experiments;
• Bell tests;
• Hanbury Brown–Twiss correlations;
• particle decay reconstruction;
• background rejection in scattering and nuclear experiments.

A coincidence is not merely two counters clicking. It requires a time window, detector efficiencies, accidental-coincidence estimates, and a model of the source. In quantum optics, normalized correlation functions such as $g^{(2)}(\tau)$ distinguish coherent, thermal, and antibunched light. In foundations experiments, coincidence and detection efficiencies are part of the loophole analysis.

Common pitfalls

One click need not mean one photon. An on/off detector can give the same outcome for one photon or several. Dark counts can also produce a click with no incident photon.

Loss and dead time affect the sample. Dividing a rate by efficiency is appropriate only in the linear, unsaturated regime. Dead time and postselection can change the measured distribution, not just the overall rate.

Clicks alone do not establish nonclassical light. Attenuated coherent light also produces individual clicks. Antibunching or suitable correlation measurements test statistics that a classical fluctuating intensity cannot reproduce; calibrate accidentals and detector artifacts before interpreting them.

Exercises

1. An on/off detector has efficiency $\eta=0.70$ and no dark counts. What is the click probability for a single-photon input?

Solution

For $n=1$, the no-click probability is $1-\eta=0.30$. Thus

$$P_{\rm click} = 1-(1-\eta) = \eta = 0.70.$$

2. A bin of an interference histogram has probability $p=0.20$ and the experiment records $N=10{,}000$ independent events. Estimate the expected count and binomial standard deviation.

Solution

The expected count is

$$\mathbb E[N_i] = Np = 2000.$$

The binomial standard deviation is

$$\sigma = \sqrt{Np(1-p)} = \sqrt{10{,}000(0.20)(0.80)} = 40.$$

3. A detector records $1250$ counts per second with the source open and $50$ dark counts per second with the source blocked. Assume negligible dead time and saturation, a stable background, and efficiency $\eta=0.60$. Estimate the incident event rate before detection.

Solution

Subtract the dark rate:

$$R_{\rm signal,detected} = 1250-50 = 1200\,\mathrm{s^{-1}}.$$

Correct for efficiency:

$$R_{\rm incident} \approx \frac{1200}{0.60} = 2000\,\mathrm{s^{-1}}.$$

4. Why does a single-electron interference experiment not show a fringe pattern after one electron?

Solution

One detection gives one localized outcome sampled from the probability distribution. The fringe pattern is a statistical structure in many outcomes. The interference term affects the probability density from which events are sampled, but a single event cannot display the whole distribution.

References

• P. Grangier, G. Roger, and A. Aspect, “Experimental Evidence for a Photon Anticorrelation Effect on a Beam Splitter,” Europhysics Letters 1, 173-179, 1986, DOI: 10.1209/0295-5075/1/4/004.
• R. H. Hadfield, “Single-Photon Detectors for Optical Quantum Information Applications,” Nature Photonics 3, 696-705, 2009, DOI: 10.1038/nphoton.2009.230.
• G. F. Knoll, Radiation Detection and Measurement, 4th ed., Wiley, 2010.
• R. Loudon, The Quantum Theory of Light, 3rd ed., Oxford University Press, 2000.
• L. Mandel and E. Wolf, Optical Coherence and Quantum Optics, Cambridge University Press, 1995.
• A. Tonomura, J. Endo, T. Matsuda, T. Kawasaki, and H. Ezawa, “Demonstration of Single-Electron Buildup of an Interference Pattern,” American Journal of Physics 57, 117-120, 1989, DOI: 10.1119/1.16104.