Coincidence-to-Accidentals Ratio (CAR)

• CAR is a key metric defined as the ratio of true coincidences to accidental events, clarifying signal purity in experimental measurements.
• It involves measuring singles and coincidence rates, histogram-based background estimation, and corrections for detector dead time to extract true signals.
• High CAR values indicate robust quantum correlations, supporting accurate entanglement verification and improved quantum communication protocols.

The Coincidence-to-Accidentals Ratio (CAR) is a critical figure of merit quantifying the relative purity of signal versus background in coincidence measurements throughout photon-pair generation, quantum optics, and particle physics. CAR distinguishes the rate of true, physically correlated events (e.g., entangled photon pairs or prompt-delayed nuclear decays) from the rate of accidental, uncorrelated coincidences arising from stochastic background processes, detector artifacts, or experimental dead times. High CAR values indicate low noise floors and enhanced signal-to-noise, thus enabling stringent tests of quantum correlations, accurate source benchmarking, and precise background rejection in fundamental and applied experiments (Grieve et al., 2015, Zhao et al., 2019, Ma et al., 2017, Yu et al., 2013).

1. Definitions and Analytical Formulation

Let $C_\mathrm{obs}$ denote the observed (total) coincidence rate, $C_\mathrm{true}$ the true (signal) coincidence rate, and $C_\mathrm{acc}$ the accidental (background) coincidence rate. The canonical relationship

$C_\mathrm{obs} = C_\mathrm{true} + C_\mathrm{acc}$

directly leads to the definition

$$\mathrm{CAR} \equiv \frac{C_\mathrm{true}}{C_\mathrm{acc}}$$

or, via post-processing subtraction,

$$\mathrm{CAR} = \frac{C_\mathrm{obs} - C_\mathrm{acc}}{C_\mathrm{acc}}$$

In time-tagged histogram analysis, $C$ corresponds to the integrated count in the coincidence window (central Gaussian peak), and $A$ is the flat background estimated from side regions or off-window intervals (Ma et al., 2017, Zhao et al., 2019). In cross-correlation function formalism, the CAR is given by

$\mathrm{CAR} = \max [g^{(2)}_{SI}(\tau)] - 1$

where $g^{(2)}_{SI}(\tau)$ is the normalized cross-correlation between signal and idler channels.

A typical calculation for asynchronous detection pulses of durations $C_\mathrm{true}$0, $C_\mathrm{true}$1 and singles rates $C_\mathrm{true}$2, $C_\mathrm{true}$3 yields the "standard" low-rate accidental estimate

$C_\mathrm{true}$4

Neglect of detector dead time or recovery effects strictly limits the regime of validity (Grieve et al., 2015).

2. Experimental Methodologies and Rate Estimation

CAR is operationally determined by the following procedure:

1. Measurement of Singles and Coincidence Rates: Acquire singles rates $C_\mathrm{true}$5, $C_\mathrm{true}$6 and observed coincidence rate $C_\mathrm{true}$7 over a well-defined coincidence window.
2. Accidental Background Estimation:
   • For high-purity photon-pair sources, determine $C_\mathrm{true}$8 by histogramming time differences and averaging the background level outside the true signal peak (Zhao et al., 2019, Ma et al., 2017).
   • In particle and neutrino detection, exploit analytic models of delayed-coincidence windows and account for potential dead-time (e.g., muon veto in reactor neutrino detectors) (Yu et al., 2013).
3. Extraction of True Coincidences: Calculate $C_\mathrm{true}$9.
4. Computation of CAR: Evaluate $C_\mathrm{acc}$0.

In advanced high-rate or saturated detector regimes, corrections for finite detector recovery must be included (see Section 3).

3. Detector Effects, Dead Time, and Effective Duty Cycle Modeling

Physical detectors deviate from ideal Poissonian statistics due to recovery, saturation, and dead time effects. For passively quenched avalanche photodiodes (GM-APDs), an "effective duty cycle" ($C_\mathrm{acc}$1) is introduced to model the accessible detection window as a function of incoming flux $C_\mathrm{acc}$2 (Grieve et al., 2015):

$C_\mathrm{acc}$3

where $C_\mathrm{acc}$4 encompasses avalanche trigger probability and discriminator response as the device recovers from the previous event. The modified accidental rate then becomes

$C_\mathrm{acc}$5

This correction allows background subtraction and CAR estimation even in deep saturation (high flux), substantially extending the usable dynamic range of practical detectors.

4. CAR in Practice: Empirical Results and Performance Metrics

Studies in various architectures provide quantitative assessments:

Platform	CAR (low brightness)	CAR (high brightness)	Context
LNOI SPDC waveguide (Zhao et al., 2019)	$C_\mathrm{acc}$6	$C_\mathrm{acc}$7	$C_\mathrm{acc}$8–$C_\mathrm{acc}$9/s PGR
Silicon microring SFWM (Ma et al., 2017)	$C_\mathrm{obs} = C_\mathrm{true} + C_\mathrm{acc}$0	$C_\mathrm{obs} = C_\mathrm{true} + C_\mathrm{acc}$1	Up to $C_\mathrm{obs} = C_\mathrm{true} + C_\mathrm{acc}$2/s/GHz brightness
Passively quenched GM-APDs (Grieve et al., 2015)	\multicolumn{2}{c	}{Doubling in deep saturation}	Fringe visibility improved from $C_\mathrm{obs} = C_\mathrm{true} + C_\mathrm{acc}$3 to $C_\mathrm{obs} = C_\mathrm{true} + C_\mathrm{acc}$4
Reactor ν̄ detection (Daya Bay) (Yu et al., 2013)	$C_\mathrm{obs} = C_\mathrm{true} + C_\mathrm{acc}$5	—	$C_\mathrm{obs} = C_\mathrm{true} + C_\mathrm{acc}$6 Hz, $C_\mathrm{obs} = C_\mathrm{true} + C_\mathrm{acc}$7 Hz

In SPDC and SFWM, accidentals scale as singles rate squared times the coincidence window, leading to an observed decrease in CAR with increasing pair generation rate, holding window width fixed. Extremely high CAR values ($C_\mathrm{obs} = C_\mathrm{true} + C_\mathrm{acc}$8) indicate single-mode, low-noise photon-pair sources suitable for heralded single-photon applications and entanglement verification.

5. Theoretical Models in Delayed Coincidence and Accidental Calculation

In reactor neutrino and similar experiments employing a fixed delayed window $C_\mathrm{obs} = C_\mathrm{true} + C_\mathrm{acc}$9, the analytic model yields

$$\mathrm{CAR} \equiv \frac{C_\mathrm{true}}{C_\mathrm{acc}}$$0

$$\mathrm{CAR} \equiv \frac{C_\mathrm{true}}{C_\mathrm{acc}}$$1

and

$$\mathrm{CAR} \equiv \frac{C_\mathrm{true}}{C_\mathrm{acc}}$$2

where $$\mathrm{CAR} \equiv \frac{C_\mathrm{true}}{C_\mathrm{acc}}$$3 is the single background rate, $$\mathrm{CAR} \equiv \frac{C_\mathrm{true}}{C_\mathrm{acc}}$$4 the fraction of neutron captures within $$\mathrm{CAR} \equiv \frac{C_\mathrm{true}}{C_\mathrm{acc}}$$5, and $$\mathrm{CAR} \equiv \frac{C_\mathrm{true}}{C_\mathrm{acc}}$$6, $$\mathrm{CAR} \equiv \frac{C_\mathrm{true}}{C_\mathrm{acc}}$$7 are detection efficiencies. This highlights that CAR is strongly limited by the square of the uncorrelated singles rate and window duration (Yu et al., 2013).

6. Impact on Quantum Optics and Signal-Processing Applications

CAR directly impacts the feasibility and fidelity of experiments in quantum information, including the verification of energy-time entanglement (e.g., Franson visibility), heralded single-photon generation, and quantum key distribution. A high CAR is required to suppress erroneous events below critical thresholds for Bell test violation and secure quantum communication (Ma et al., 2017, Zhao et al., 2019). In neutrino detection and other delayed-coincidence techniques, accurate modeling and maximization of CAR are essential for background subtraction and precise estimation of oscillation or interaction parameters (Yu et al., 2013).

7. Limitations, Error Propagation, and Future Directions

Dominant uncertainties in CAR arise from fluctuations in singles background rates (scaling as $$\mathrm{CAR} \equiv \frac{C_\mathrm{true}}{C_\mathrm{acc}}$$8), coincidence window selection, and, in complex systems, residual systematic errors in detector response modeling (Yu et al., 2013, Grieve et al., 2015). The implementation of effective duty cycle corrections significantly mitigates the loss of CAR in high-flux scenarios, opening expanded operational regimes for single-photon and quantum-enabled detectors.

Continued advances in detector architecture, filtering schemes, and analytic modeling will further enhance attainable CAR values, particularly as devices approach the physical limits of time resolution and dark count suppression. Novel platforms such as thin-film lithium niobate and integrated silicon photonics are closing the gap with crystal and fiber-based benchmarks, establishing new standards for scalable quantum photonic sources (Zhao et al., 2019, Ma et al., 2017).