Source-linked AI summary
Recent advances and future perspectives of single-photon avalanche diodes for quantum photonics applications
Francesco Ceccarelli, Giulia Acconcia, Angelo Gulinatti, Massimo Ghioni, Ivan Rech, Roberto Osellame
TL;DR
Photonic quantum technologies rely on single photons, making single-photon detectors central to communication, computing, simulation, and sensing. This report reviews SPAD principles and recent detector progress for quantum photonics. It highlights SPAD developments across spectral regions and points to waveguide integration as a future direction.
Problem
Quantum-photonics applications need single-photon detectors, while superconducting detectors require cryogenic operation and can complicate experimental setups.
Source: §1 · p. 1, §2.3 · p. 5Method
The paper reviews SPAD operating principles and recent detector progress, comparing device characteristics with quantum-photonics requirements.
Source: §2.2 · p. 3, §2.3 · p. 5, §2.5 · p. 6Results
Silicon SPADs serve visible/NIR detection, InGaAs/InP devices have been demonstrated beyond 1000 nm, and germanium SPADs have gained attention for Ge-on-Si designs with relatively low DCR.
Source: §5 · pp. 24, 26Takeaways & Limitations
Waveguide-integrated SPADs could support room-temperature quantum PICs with many thermo-optic phase shifters, despite substantial materials, fabrication, optical, and electrical design challenges.
Source: §5 · p. 26At long wavelengths, SPAD timing response and PDE trade off; separating the substrate from the junction can limit transit time and diffusion tail at the cost of reduced PDE.
Source: §2.3 · p. 5
Abstract
from arXiv · showhide
Photonic quantum technologies promise a revolution of the world of information processing, from simulation and computing to communication and sensing, thanks to the many advantages of exploiting single photons as quantum information carriers. In this scenario, single-photon detectors play a key role. On the one hand, superconducting nanowire single-photon detectors (SNSPDs) are able to provide remarkable performance on a broad spectral range, but their applicability is often limited by the need of cryogenic operating temperatures. On the other hand, single-photon avalanche diodes (SPADs) overcome the intrinsic limitations of SNSPDs by providing a valid alternative at room temperature or slightly below. In this paper, we review the fundamental principles of the SPAD operation and we provide a thorough discussion of the recent progress made in this field, comparing the performance of these devices with the requirements of the quantum photonics applications. In the end, we conclude with our vision of the future by summarizing prospects and unbeaten paths that can open new perspectives in the field of photonic quantum information processing.
1 Introduction
Photonic quantum technologies use single photons for communication, computing, simulation, and sensing, making single-photon detection a key component. This report reviews SPAD fundamentals and recent progress, positioning them against detector alternatives and quantum-photonics needs.
1 Introduction
- Photonic quantum technologies target communication, computing, simulation, and sensing, with photons valued as information carriers and for compatibility with classical optical communications infrastructure.The introduction cites QKD and quantum repeaters among communication applications.Source: §1 · p. 1
- SNSPDs offer strong performance across wavelengths, but trade-offs prevent all reported best characteristics from appearing in one device, and cryogenic operation adds cost and complexity.The introduction also notes polarization-dependent efficiency and potential constraints on thermo-optic integrated systems.Source: §1 · p. 2
- SPADs operate at room temperature or slightly below and are used in quantum-photonics demonstrations, particularly when cryostats are unavailable or wide active areas and polarization insensitivity are needed.The report also identifies SPAD arrays as a scalable detector option.Source: §1 · p. 2
- The report reviews SPAD operating fundamentals, recent technological results with emphasis on quantum photonics, and prospects for integration in waveguide-based quantum PICs.The introduction identifies single-photon detectors as one part of systems that generate, manipulate, and detect quantum light.Source: §1 · pp. 1–3
2 SPAD Fundamentals
SPAD performance depends on device structure and operating conditions. A reverse-biased junction converts an absorbed photon into a self-sustaining avalanche that must be quenched and reset before another photon can be detected.
2.1 Principle of Operation
- Above-breakdown reverse bias lets an absorbed photon initiate a self-sustaining avalanche; quenching and reset restore the detector, whose readout can report photon counts or arrival times.Quenching may use a high-value resistor or an active circuit; the avalanche makes the electronic output a digital pulse.Source: §2.1 · p. 3
2.2 Photon Detection Efficiency
- PDE requires photon absorption and carrier collection and triggering; antireflection coatings reduce reflection, while longer wavelengths need thicker detectors and higher overvoltage trades higher PDE against DCR, afterpulsing, and array crosstalk.Collection is certain in the depletion region but limited in neutral regions, where carriers move by diffusion.Source: §2.2 · pp. 3–4
2.3 Timing Response
- Timing jitter combines absorption-depth-dependent transit time, stochastic avalanche build-up and lateral propagation; carriers diffusing from neutral regions also produce a long IRF tail.The detector delay is the sum of transit, build-up, and propagation times, and its statistical dispersion represents timing jitter.Source: §2.3 · pp. 4–5
- Higher overvoltage reduces build-up and propagation jitter, but timing and PDE trade off, especially at long wavelengths; separating the substrate from the junction can shorten transit and diffusion tails at reduced PDE.Small devices can hinder optical alignment, while low-threshold electronics or focusing light onto a small spot can reduce timing contributions in the stated conditions.Source: §2.3 · p. 5
2.4 Dark Count Rate
- Room-temperature DCR usually comes from trap-assisted SRH generation and depends on detector volume and material defects; at low temperatures, field-driven BBT can dominate and limit cooling's benefit.Higher overvoltage may raise PDE but also increases DCR; proper electric-field design can avoid the described low-temperature BBT-dominated regime.Source: §2.4 · pp. 5–6
2.5 Afterpulsing and Maximum Count Rate
- Afterpulsing arises when avalanche-trapped carriers are later released and trigger correlated spurious events; minimizing traps and avalanche charge helps, while hold-off suppresses events at the cost of maximum count rate.Active quenching supports high count rates, and longer hold-off may be needed at low temperatures because trap release slows.Source: §2.5 · pp. 6–7
2.6 Crosstalk
- Avalanche-emitted photons can trigger other array pixels; direct optical crosstalk rises as pixel spacing shrinks, and trenches may not block indirect paths reflected from the chip bottom.Indirect optical crosstalk is less sensitive to pixel spacing than direct crosstalk.Source: §2.6 · pp. 7–8
- In timing arrays, substrate charge sharing can unpredictably shift low-threshold timing, while a thick low-doped substrate used to raise long-wavelength PDE can lengthen the IRF tail and cause delayed crosstalk.A fixed substrate potential is needed to avoid charge sharing, which can challenge low-threshold front-end design.Source: §2.6 · p. 8
3 Fabrication Technologies
This section surveys SPAD fabrication technologies by material, relating optical absorption to wavelength suitability. It covers silicon devices for 400–1000 nm alongside InGaAs/InP and Ge designs for infrared detection, including efforts toward improved speed, scaling, lower-noise detection, and silicon-photonics integration while noting performance limitations.
3.1.1 Custom Technologies
- Reach-through SPADs offer PDE above 60% up to 800 nm and DCR down to 25 cps, but timing jitter remains above 150 ps and speed is limited to tens of Mcps.The 25-cps DCR applies to a 180-µm device; even optimized modules reach only 200–250 ps jitter when light is focused to a 10-µm spot.Source: §3.1.1 · pp. 10–11
- Thin SPADs provide jitter as low as 32 ps and count rates up to 160 Mcps, but their PDE is limited to 50% at 550 nm and 15% at 800 nm.Their thin absorption region and substrate enable sharp timing and independent-pixel arrays, demonstrated up to 8 × 8.Source: §3.1.1 · p. 11
- RE-SPADs raise PDE to 70% at 650 nm and 47% at 800 nm; deep-trench isolation enables arrays, while jitter reaches 83 ps and afterpulsing exceeds 10% above 80 Mcps.The reported 32 × 1 arrays use a 250-µm pitch.Source: §3.1.1 · pp. 11–12
- RCE-SPADs use a buried Bragg reflector to enhance PDE by up to a factor of 3 at 750–950 nm while retaining thin-SPAD timing, low-voltage operation, and array compatibility.The cavity requires wavelength tuning, and dark counts below 30 cps were demonstrated at −20 °C for 20-µm devices.Source: §3.1.1 · p. 12
3.1.2 Standard Technologies
- Standard CMOS SPADs without substrate isolation reach 72% PDE at 560 nm and about 30% at 800 nm, but slow-carrier diffusion impairs blue timing and can cause delayed crosstalk.At 654 nm, the reported timing-response FWHM is below 60 ps, although the diffusion-tail time constant is on the order of nanoseconds.Source: §3.1.2 · p. 14, Figure 3 · p. 15
- Substrate-isolated CMOS designs suppress diffusion tails: a p-i-n device showed 60–300 cps DCR and about 100-ps jitter, while a 500-µm device combined below-1% afterpulsing with 53% PDE at 450 nm.The large-area device’s PDE at 800 nm was only 5%; low afterpulsing below 1% remained un demonstrated for the p-i-n design.Source: §3.1.2 · pp. 14–15
- A 0.16-µm BCD SPAD reached about 60% PDE at 500 nm and 13% at 800 nm, with less than 30-ps jitter and a diffusion-tail time constant below 50 ps using a dedicated front end.Its custom implant steps tune the electric field and reduce dark counts and series resistance.Source: §3.1.2 · pp. 15–16, Figure 4 · p. 16
3.2.1 InGaAs/InP
- Because silicon cannot detect photons above 1000 nm, infrared applications use frequency conversion or direct-detection InGaAs/InP SPADs, which have supported fiber QKD over 307 km and gated photon-number resolution.InGaAs/InP devices also support free-running operation; their cost and complexity make them a practical network option, although SNSPDs are used for longer distances.Source: §3.2 · pp. 16–17, §3.2.1 · p. 17
- InGaAs/InP SPADs commonly use a MOCVD-grown SACM stack, with a charge layer shaping the field and an InGaAsP grading layer reducing hole trapping.The lattice-matched InGaAs absorber and InP multiplier limit band-to-band tunneling; zinc diffusion forms the junction and tailors the p-doped well to avoid premature breakdown.Source: §3.2.1 · p. 18, Figure 5 · p. 20
- Fast gating enables GHz operation but not necessarily Gcps counts; SD and sinusoidal-gate readouts suppress capacitive transients, while afterpulsing and count rate remain key constraints.Low-frequency gated devices use cryogenic temperatures and tens-of-nanoseconds gates to reduce dark counts and allow carrier detrapping; fast gating reduces avalanche charge but requires electronics to distinguish avalanches from transients.Source: §3.2.1 · p. 18, Figure 5 · p. 20
- Integrated-resistor NFADs reduce avalanche charge and reset time, but passive-quenching trade-offs limit performance; an external active quench circuit enabled 100 kcps with afterpulsing no higher than 20%.At 223 K, an NFAD achieved 30% PDE and 52–67 ps FWHM timing jitter; at 163 K it reached about 10 cps DCR with 27.7% PDE and 129 ps jitter, at 50 kcps maximum count rate and 20% afterpulsing probability.Source: §3.2.1 · p. 19
- InGaAs/InP arrays have reached 128 × 32 pixels, but optical crosstalk exceeding 80% limits density; metal-filled trenches reduced crosstalk to 37% at 60 µm pixel spacing.NFAD arrays offer a route to simpler readout, with demonstrated formats up to 8 × 8, but at the cost of reduced PDE.Source: §3.2.1 · p. 19
3.2.2 Germanium
- Ge SPADs require much lower temperatures than InGaAs/InP for comparable DCR and have lower PDE, especially at 1550 nm, but planar Ge-on-Si reached <500 kcps DCR and 35% PDE at 1310 nm.Mesa sidewall depletion raises dark-carrier generation; selective boron implantation moves the field away from sidewalls. The planar device operated at 100 K, with 310 ps FWHM jitter (175 ps for 26 µm diameter) and about 90% fabrication yield.Source: §3.2.2 · pp. 19, 21, 22
- Ge-on-Si enables integration with silicon photonics, although the first waveguide-coupled SPAD demonstrated only 5.27% overall PDE at 1310 nm.Under matched temperature and PDE conditions, Ge-on-Si also compared favorably with InGaAs/InP in afterpulsing, suggesting potential for higher count rates if operated at the same temperature.Source: §3.2.2 · p. 21
4 Waveguide SPADs: applications and outlook
Integrated detection in quantum PICs could reduce photon losses and support feed-forward, while SPADs offer a higher-temperature route compatible with reconfigurable circuits. The outlook centers on waveguide SPADs and wavelength-specific material and coupling choices.
4 Waveguide SPADs: applications and outlook
- On-chip detection can reduce photon losses, aid miniaturization and scalability, and provide prompt sensing for feed-forward operations.Source: §4 · p. 22
- Waveguide SPADs guide photons through the depletion region; orthogonal light propagation and electric field can break the detection-efficiency/transit-time trade-off.Source: §4 · p. 22
- A silicon waveguide butt-coupled to a Ge-on-Si detector is described as a rudimentary example of monolithic waveguide-SPAD integration.Source: §4 · p. 22, Figure 6 · p. 23
- At 800 nm, silicon nitride core, silicon SPAD, and silica cladding are SOI-compatible, but coupling single-mode silicon nitride to absorbing silicon needs nontrivial design.Source: §4 · p. 24
- At 1550 nm, germanium is compatible with silicon/silica PICs but its SPAD performance needs improvement; proposed paths include GeSn absorption or hybrid integration.Source: §4 · p. 24
5 Conclusion
The review compares SPADs for quantum photonics across detection efficiency, noise, timing precision, and array crosstalk. It highlights material-specific progress and identifies waveguide-integrated SPADs as a challenging route toward room-temperature quantum photonic circuits.
5 Conclusion
- Table 1 summarizes suitable SPAD performance across high PDE, low noise, timing precision, and low crosstalk for quantum photonics applications.Silicon data are reported at room temperature, possibly with best results at −20 °C in brackets; other SPAD data use the specified measurement temperature.Source: §5 · pp. 24–25, Table 1 · p. 25
- Tens of Mcps maximum count rate and few tens of ps timing jitter accompany custom silicon SPADs; reach-through structures lead PDE, while planar thin, RE, and RCE processes enable large arrays.These devices combine high detection efficiency with low noise and afterpulsing.Source: §5 · p. 24
- Figure 7 presents proposed visible/NIR silicon SPADs butt-coupled to silicon nitride and infrared GeSn SPADs butt-coupled to silicon [239, 242].The figure shows each design in 3D and cross-section.Source: Figure 7 · p. 26
- Ge-on-Si SPADs with relatively low DCR have drawn attention as high-quality crystalline germanium layers became possible on silicon.The review notes germanium SPADs are gaining research-community attention.Source: §5 · p. 26
- Waveguide-integrated SPADs pose materials, fabrication, optical, and electrical challenges, but could support room-temperature quantum PICs with many thermo-optic phase shifters.The authors identify this as the path to fully exploit integrated quantum photonics.Source: §5 · p. 26
Conflict of Interest
The authors disclose an equity interest in Micro Photon Devices S.r.l. and state that the company did not contribute resources or personnel to the work.
Conflict of Interest
- M. G. discloses equity in Micro Photon Devices S.r.l.; the company provided no resources or personnel for this work.This is the stated conflict-of-interest disclosure.Source: Conflict of Interest · p. 26