Photon-Number-Resolving Detection

Updated 16 January 2026

Photon-number-resolving detection is a method that quantifies the exact number of incident photons by converting absorption events into measurable electrical signals.
It employs diverse architectures such as spatial and temporal multiplexing, transition-edge sensors, and quantum-emitter cascades to achieve high efficiency and linear response.
Advances in device design and readout strategies are driving applications in quantum metrology, quantum communication, and state engineering.

Photon-number-resolving (PNR) detection refers to the ability of a photodetector to discriminate, in a single shot, the precise number of incident photons within a given temporal or spatial mode. PNR detectors are critical in quantum optics, quantum information science, metrology, and advanced sensing, as many protocols require measurement outcomes beyond simple click/no-click discrimination. To achieve photon-number resolution, detectors must exhibit both high quantum efficiency and a response that monotonically—and preferably linearly—maps photon input number to a measurable output parameter, enabling statistical estimation or direct assignment of photon number over a specified range.

1. Physical Principles and Device Architectures

PNR detectors exploit device architectures allowing individual photon events to be separately registered or mapped to distinct output states. The principal implementations are:

Spatial or Structural Multiplexing: Nanowire-based arrays (series or parallel), where each geometrically distinct element can trigger independently upon photon absorption (Ding et al., 3 Apr 2025). Series configurations, such as 32-segment nanowires with parallel shunt resistors, convert the number of triggered segments directly into a pulse amplitude proportional to the photon count.
Temporal/Spatiotemporal Multiplexing: Arrangements wherein photons are distributed across distinct time bins or spatial channels, each monitored by single-photon avalanche detectors or nanowires. The multiplexing may be implemented via optical delay lines and/or cascaded beam splitters; the number of coincident or temporally separated "clicks" infers photon number (Cheng et al., 2022).
Distributed Absorber and Mode Engineering: Coherent absorption in multilayer nanowire stacks or phase-engineered planar arrays, where each sub-detector absorbs at a standing-wave antinode, producing uniform and deterministic absorption among the elements without splitting the optical mode (Vetlugin et al., 2022).
Quantum-Emitter Cascades: Chiral waveguide-coupled λ-type emitters, each extracting a single photon via SPRINT processes and shelving itself after detection, permitting one-by-one photon counting as the pulse propagates through the cascade (Pasharavesh et al., 11 Jul 2025).
Transition-Edge Sensors (TES): Microcalorimetric devices in which the absorption of n photons produces a pulse of amplitude proportional to the deposited energy, with sufficient energy resolution to distinguish photon numbers (Calkins et al., 2013).

2. Readout Mechanisms and Photon Number Discrimination

The linkage between the number of absorbed photons and a measurable output is implemented through various mechanisms:

Pulse Amplitude Mapping: In series-segmented nanowire detectors, each triggered segment forms a resistive hotspot, diverting a fraction of the bias current through a measurement chain, resulting in a voltage pulse whose amplitude is approximately n times the single-photon response (Ding et al., 3 Apr 2025, Jahanmirinejad et al., 2012). For fully-separated events and high linearity, the mapping is described by

$V(n) = n\,\Delta V$

with readout noise typically Gaussian.

Full Waveform or Time-Tagging: High-bandwidth readout electronics or optically-sampled waveguide modulators can resolve sub-picosecond differences in rising/falling edges of the detection pulse, with the separation correlated to the number of photons absorbed (Endo et al., 2024, Sauer et al., 2023).
Maximum-Likelihood Estimators: For a given readout observable $V$ , the most likely photon number is inferred by maximizing the conditional probability $p(V|n)$ over all n, typically implemented as $\hat{n}(V)=\mathrm{round}(V/\Delta V)$ .
Detector Tomography and POVM Reconstruction: The detectors are characterized by phase-insensitive, diagonal positive-operator-valued measure (POVM) elements in the Fock basis $\{\Pi_n\}$ , reconstructed by sending a tomographically-complete set of input states (e.g., coherent states with varying mean photon number) and solving

$O_{n,j} \approx \sum_m P_{n,m}\,I_{m,j}$

where $P_{n,m}$ is the probability that m photons lead to an n-click event (Ding et al., 3 Apr 2025, Endo et al., 2021).

3. Quantitative Performance Metrics

The main metrics for evaluating PNR detectors include:

Metric	State-of-the-art Nanowire PNRD	TES	Multiplexed SNSPDs	PNR via Quantum Emitters
System Detection Efficiency	98% at 1555 nm (Ding et al., 3 Apr 2025)	90–95% (Calkins et al., 2013)	80–90% (Cheng et al., 2022)	~70–90% (theory) (Pasharavesh et al., 11 Jul 2025)
Maximum Photon Number $n_\mathrm{max}$	32 (Ding et al., 3 Apr 2025)	>20 (but slow)	100 (spatio-temp array) (Cheng et al., 2022)	n (n emitters) (Pasharavesh et al., 11 Jul 2025)
Timing Jitter	down to 40 ps (n=32)	≥1 ns	16–50 ps	~10 ns (set by bandwidth)
Dark Count Rate (DCR)	20 cps	<1 cps	<1 Hz / pixel	negligible (theory)
Count Rate	41 MHz (-3 dB SDE) (Ding et al., 3 Apr 2025)	<1 MHz	GHz / chip (Cheng et al., 2022)	MHz (detector dead time)
Fidelity $F_n$ (n-photon)	$F_2=0.874$ , $F_3=0.734$ , $F_4=0.405$ (Ding et al., 3 Apr 2025)	>0.95 (n≤5) (Calkins et al., 2013)	0.90–0.97 (n≤5, $N>10$ ) (Vetlugin et al., 2022)	$P_\mathrm{lin}$ , $P_\mathrm{nl}$ as theory (Pasharavesh et al., 11 Jul 2025)

Fidelity typically decreases with photon number due to increased overlap between the response distributions and limits set by the readout noise and device segmentation.

4. Methodological Innovations and Trade-offs

Advancing PNR performance has required several technical strategies:

Twin-Layer Nanowire on Dielectric Mirror: Near-unity absorption is achieved with a DBR stack (SiO₂/Ta₂O₅) and a sandwich of NbN/SiO₂/NbN, with the nanowire patterned in a high-fill-factor meander for maximal overlap with the optical mode (Ding et al., 3 Apr 2025).
Spatial Multiplexing via Segmentation: Dividing the active region into many independently switchable, series-shunted nanowire segments enables linear conversion of event number to electrical amplitude without requiring separate readout channels (Ding et al., 3 Apr 2025).
Full-Waveform and Pattern-Matching Readout: Time-resolved waveform analysis (pattern matching of rising edges to reference traces) or high-resolution time-tagging allows discrimination up to n=5 in single meandered nanowires without hardware segmentation (Endo et al., 2021, Sauer et al., 2023).
Impedance Taper Engineering: Impedance-matching tapers coupled to single nanowires optimize current division and enhance the dependence of output pulse amplitude on photon number (Zhu et al., 2019).
Optical Sampling Techniques: Mach–Zehnder-based optical sampling with ps-range probe pulses enables discrimination of nm-scale timing variations in SNSPD outputs, extending number-resolving capability beyond electronic bandwidth limits (Endo et al., 2024).

Trade-offs include speed versus SNR (since increasing readout bandwidth increases electronic noise), number of segments versus per-segment fidelity, and the complexity of readout or segmentation versus maximum photon number discrimination.

5. Comparison with Alternative PNR Detection Platforms

Comparative attributes across major PNR technologies:

TES: TES provides ultimate energy resolution ( $\Delta E\sim 0.35$ eV, $>20$ resolved photons), at the cost of μs-scale recovery times, high-jitter (≥1 ns), and requirement for sub-100 mK cooling and SQUID-based readout (Calkins et al., 2013, Ding et al., 3 Apr 2025).
Multiplexed SNSPD Arrays: Arrays scale PNR linearly with array size, but require multiple readout channels, introduce inter-channel losses and moderate dark-count scaling, and saturate in efficiency near 80–90% (Cheng et al., 2022).
Temporal Multiplexing: Time-multiplexed fiber loops or waveguide cascades can reach very high dynamic range (>100 photons), but at the expense of added insertion loss, lower per-photon efficiency, and more complex routing. Dead times and detector noise accumulate per channel (Ding et al., 3 Apr 2025, Limongi et al., 2024).
Distributed Coherent Absorption: Fully coherent multilayer absorbers can achieve deterministic, lossless n-photon discrimination with minimal layers for small n, but require advanced thin-film engineering and are limited in bandwidth (Vetlugin et al., 2022).
Quantum-Emitter Cascades: Theoretical models predict that cascades of chiral quantum emitters can outperform spatial-multiplexed PNR under high waveguide coupling and well-separated photon pulses, but their implementation faces significant experimental challenges (Pasharavesh et al., 11 Jul 2025).

6. Applications in Quantum Technology and Advanced Sensing

Key operational regimes leveraging PNR detectors:

Quantum Metrology and Calibration: Absolute detector calibration and sub-shot-noise quantum sensing rely on precise photon number resolution and known POVM elements (Ding et al., 3 Apr 2025).
Quantum Information Processing: Boson sampling, Gaussian boson sampling, and measurement-based photonic quantum computing require multiphoton-resolved detection to identify collision events and support conditional gate operations (Ding et al., 3 Apr 2025, Cheng et al., 2022).
Quantum Communication: PNR detectors extend the security boundary of quantum key distribution by revealing photon-number-splitting attacks and enable advanced decoy-state protocols (Cassina et al., 4 Aug 2025, Cohen et al., 2019).
Quantum State Engineering: Real-time heralding of non-Gaussian states (e.g., Schrödinger-cat, GKP resources) is enabled by high-fidelity discrimination of single and multiphoton subtraction events (Endo et al., 2024).
LIDAR and Imaging: PNR detectors coupled with photon-number thresholding offer signal-to-noise enhancements over classical intensity-based detection in high-background environments (Cohen et al., 2019).

7. Current Limitations and Outlook for Further Development

Significant progress in PNR detection has recently extended photon-number discrimination to the $\sim$ 30–100 photon regime at high efficiency and MHz–GHz rates, with system detection efficiencies up to 98% and sub-50 ps timing jitter demonstrated in series-segmented SNSPDs (Ding et al., 3 Apr 2025, Cheng et al., 2022). Persistent limitations include:

Crosstalk and Amplitude Saturation: Even with low (<0.1%) measured crosstalk, amplitude saturation imposes practical limits beyond n~30–50.
Readout Complexity and SNR: High photon-number discrimination requires low-noise, high-bandwidth amplification, and sophisticated signal processing for waveform or time-tag analysis.
Scalability of Integration: Monolithic integration of large arrays or complex multi-layer absorbers is technologically challenging, although advances in planar photonics and thin-film deposition are closing this gap (Vetlugin et al., 2022).
Extending Dynamic Range: Future hybrid schemes—combining spatial, temporal, and even spectral multiplexing—are predicted to extend practical PNR resolution to >100 photons, with ongoing efforts in high-current nanowire designs, advanced cryogenic pre-amplification, and on-chip integrated resonator enhancement (Los et al., 2024, Vetlugin et al., 2022).

Near-future regimes of several tens of resolved photons at unity efficiency, sub-50 ps jitter, and >100 MHz count rates now appear experimentally accessible, establishing PNR detection as a core resource for quantum photonics, quantum information, and quantum-enhanced measurement (Ding et al., 3 Apr 2025, Los et al., 2024).