Source-linked AI summary
Gate Voltage Effect on Pulse Detection Efficiency of Perimeter-Gated SPADs
Hunter Guthrie, Md Sakibur Sajal, Zexi Liu, Marc Dandin
TL;DR
Pulsed optical systems need to distinguish return signals from dark counts and afterpulsing, but perimeter-gating’s photon-sensitivity trade-off had not been studied under pulsed excitation. This work measures pg-SPAD pulse and burst detection across gate and excess-bias voltages, finding reduced detection efficiency alongside narrower spurious-event spreads and proposing operating guidance.
Problem
The study addresses the unexplored impact of perimeter gating’s noise–photon-sensitivity trade-off on pulse detection in pulsed optical systems.
Method
The authors measure pulse and burst detection, dark counts, and afterpulsing in a 64 × 64 pg-SPAD array across gate and excess-bias voltages.
Results
Higher gate voltage decreases pulse and burst detection efficiency while reducing the spread of dark-carrier and afterpulse events within burst windows.
Takeaways & Limitations
Gate voltage should be optimized with activation time and excess bias to balance signal-to-noise ratio, detection efficiency, dark counts, and afterpulsing.
Abstract
from arXiv · showhide
Perimeter-gated single-photon avalanche diodes (pg-SPADs) are known for their dynamic dark noise modulation capabilities. They are reported to trade noise for photon sensitivity under continuous illumination. However, the implications of this trade-off have not heretofore been studied with pulsed optical systems. This work bridges this gap. We demonstrate that pg-SPADs fabricated in a 0.35 $μ$m standard CMOS process trade-off pulse detection efficiency for a reduction in the the spread of spurious events within a burst window. Consequently, herein, we propose guidelines for the optimal use of pg-SPADs in pulsed LIDAR applications in view of the observed trade-off.
I. INTRODUCTION
SPADs support pulsed optical sensing but suffer dark counts and afterpulsing that reduce correlation between transmitted pulses and recorded signals. This study examines whether perimeter gating trades pulse and burst sensitivity for narrower spurious-event spreads and develops usage guidance.
- SPADs are used in pulsed LIDAR, fluorescence sensing, and time-resolved imaging because of their fast timing response.
- Dark carriers and afterpulsing generate spurious detections that reduce correlation between an input pulse train and the recorded signal.
- Perimeter gating places a polysilicon gate at the diode periphery to reduce peripheral electric-field strength and prevent premature edge breakdown.
- For a fixed single-pulse detection rate, more pulses are required to guarantee burst detection when the rate is lower.
- The study uses a 64 × 64 custom pg-SPAD array fabricated in a 0.35 µm standard CMOS process to measure pulse and burst detection across gate and excess-bias voltages.
- Higher gate voltage reduces pulse and burst detection efficiency but greatly narrows the spread of dark-carrier and afterpulse events within burst windows.
II. EXPERIMENTAL SETUP
The experiment uses an independently selectable pixel in a 64×64 pg-SPAD array, multiplexed to an on-chip counter and monitored during detector integration.
- A DAQ triggers a 488 nm nanosecond pulsed laser while the pg-SPAD array is positioned close to the laser to improve photon collection.
- Each array pixel can be selected independently and multiplexed to a 12-bit on-chip output counter.
- The selected pg-SPAD is precharged above breakdown, then actively quenched and reset after each avalanche for subsequent detection.
A. Pulse Generation
The setup uses denoising circuits to deliver precisely timed laser bursts and segments counter data around the excitation window. Measurements sweep gate and excess-bias voltages to tune detection probability and dark-count behavior.
- A. Pulse Generation: Noise transients as small as 20 mV caused false laser triggering, so dedicated differential and Schmitt-trigger channels were used to preserve pulse timing and count.
- B. Detection: The counter records avalanche activity during a 4 ms active period while bursts contain 10–200 pulses, with 1,000 measurements collected per setting.
- B. Detection: Data are divided into pre-burst, burst, and post-burst windows for dark counts, signal plus background, and afterpulses, respectively.
- B. Detection: Peak detection is defined as the maximum incremental count across the full window, with measurements spanning V_G = 0–4 V and Vex = 3–5 V.
C. Dark Count and Afterpulsing Analysis
Dark-count and afterpulsing measurements separate background activity from burst activity and quantify correlated post-excitation noise. The section’s figures examine burst efficiency and the pulse count needed to reach peak activity.
- Laser-disabled measurements determine DCR across Vex and V_G, enabling signal-to-background normalization and isolation of illumination-induced detections.
- Afterpulsing is quantified as excess post-burst activity relative to the pre-burst DCR baseline.
- Figure 4 reports burst detection efficiency at Vex = 4 V for pulse counts N = 10–200.
- Figure 5 reports the minimum pulse count required to reach the activity peak as a function of gate voltage and excess bias, with slope representing gate-voltage sensitivity.
A. Reduction of Photon Sensitivity
Increasing perimeter gate voltage reduces pulse and burst detection efficiency, revealing a photon-sensitivity cost associated with noise suppression.
- A. Reduction of Photon Sensitivity: Increasing gate voltage magnitude decreases pulse detection efficiency during burst measurements.With Vex set to 4 V and 40 pulses, the activity peak shifts later as gate voltage increases, indicating that more incident pulses are required for burst detection.
- A. Reduction of Photon Sensitivity: ηburst is derived from burst activity after subtracting averaged pre- and post-burst noise from total counter activity, divided by the number of transmitted pulses.
- A. Reduction of Photon Sensitivity: The reduction in detection efficiency is a direct trade-off of noise reduction unless avalanche probability is increased through excess bias voltage.
- A. Reduction of Photon Sensitivity: Long-burst measurements use Vex = 4 V and 200 pulses to show perimeter-gate benefits in dark-noise and afterpulsing reduction.
- A. Reduction of Photon Sensitivity: Post-burst activity is evaluated relative to dark-count rate at different excess bias voltages.
B. Increase of Photon Sensitivity
Increasing excess bias voltage raises avalanche probability and single-pulse sensitivity, reducing the number of pulses needed to reach peak burst activity.
- B. Increase of Photon Sensitivity: Increasing excess bias voltage increases avalanche probability and psingle, thereby reducing the number of pulses required to reach peak activity.As gate voltage increases, the number of required pulses still rises, with gate-voltage sensitivity higher at higher Vex.
C. Reduction in Spurious Detections
Perimeter gating reduces spurious activity and concentrates avalanche detections into narrower time bins, while the preferred gate voltage depends on operating conditions.
- C. Reduction in Spurious Detections: In a 40 µs burst of 200 pulses, the native device produces wide detection clusters throughout the burst window.
- C. Reduction in Spurious Detections: The pg-SPAD produces a dominant delayed first cluster and tighter remaining clusters than the native device.The delayed dominant cluster is associated with reduced pulse detection probability.
- C. Reduction in Spurious Detections: Perimeter gate voltage can focus avalanche activity into narrow time bins by suppressing spurious activity for a given incoming photon rate.
- C. Reduction in Spurious Detections: Post-burst analysis compares dark counts and afterpulses with and without laser excitation as functions of perimeter gate voltage.
- C. Reduction in Spurious Detections: Higher gate voltage is preferred for longer activation times or higher excess bias voltage because it can suppress dark noise.
- C. Reduction in Spurious Detections: Gate voltage should be optimized with activation time and excess bias voltage to maximize SNR rather than being overdriven.The number of pulses can be optimized from ηburst plots using the minimum gate voltage that achieves zero DCR, while signal strength and bias voltages may require adjustment.
- C. Reduction in Spurious Detections: Future work will evaluate correlation of pulse trains and develop active adjustment of gate voltage, reverse bias, and pulse trains.
VI. CONCLUSIONS
Perimeter gating degrades pulse detection efficiency by requiring more laser pulses, but substantially reduces afterpulsing and dark count rate, producing a more defined return signal with less noise and jitter.
- Perimeter gating increases the number of laser pulses required to efficiently detect a pulse train of length N, indicating degraded pulse detection efficiency.
- The first-peak registration efficiency is degraded and shifted later in time under perimeter gating.
- Afterpulsing and dark count rate are greatly reduced, leading to a more defined return signal with less noise and jitter.