Guest viewpoint: High-efficiency 1.06-micron single photon counting avalanche photodiodes
By Bruce Nyman, Mark Itzler, Xudong Jiang, and Mike Krainak
The efficient detection of single photons at 1.06 microns is of considerable interest for light detection and ranging (lidar) systems designed for remote sensing and ranging, as well as for free-space optical transmission in photon-starved applications.
Currently, silicon-based single photon avalanche diodes (SPADs) used at this wavelength have low single photon detection efficiency (1 to 2 percent) at 1.06 microns, while InGaAs/InP SPADs designed for telecommunications wavelengths near 1.5 microns have detection efficiencies of 20 percent but exhibit dark count rates that generally inhibit non-gated (free-running) operation.
To bridge this performance gap for wavelengths just beyond 1-micron, industry has developed high-performance, large-area (80 to 200-microns in diameter) InP-based SPADs optimized for operation at 1.06 microns.
LIDAR systems for remote sensing and ranging can operate in either gated or free running mode. In both modes the light source is typically a Q switched Nd:YAG laser. In gated mode the SPAD is enabled at a particular time. The gate time relative to the scattered laser pulse determines the distance to the object.
The total range is obtained by scanning the gate times. In free running mode the detector is always enabled. The arrival time of the scattered photons provides the distance information. The system performance is determined by the SPAD’s operating parameters. The key operating parameters are the dark count rate, detection efficiency, and after pulsing.
To understand the effect of the parameters on the performance, consider a SPAD operating in gated mode. Here the SPAD is biased above breakdown for a finite amount of time. During that time an avalanche will occur if either a photon is detected or if a dark count occurs. The dark count rate defines the probability that an avalanche will occur when a photon is not present. The dark counts act as noise in the system.
When a photon is present the detection efficiency specifies the probability of it being detected. After an avalanche occurs from either a photon or a dark count, there is a probability that another avalanche will occur without a photon being detected. These additional pulses, referred to as afterpulses, limit the overall system data rate. The afterpulse probability is a function of time. The probability is reduced as the hold-off time is increased. During the hold-off time the SPAD is not biased so no avalanches occur.
In free running mode the bias is on until either a dark count occurs or a photon is detected. After the avalanche occurs the bias is turned off to eliminate any afterpulses. The combination of the dark count rate and the afterpulse hold-off time will determine the amount of time actually available for detecting photons. If the dark count rate is too high and the afterpulse rate holds off time too long, there might not be any time available for actually detecting photons.
The optimized 1.06-micron InGaAsP/InP SPADs have device performance that allows for gated and free running operation. Devices have been fabricated with an 80-micron diameter. These have been packaged in both a fiber coupled TO-46 and a TO-8 package for free space operation. The TO-8 package includes a thermistor and a multi-stage TEC cooler that can cool the SPAD to –60 degrees Celsius. This package makes the device convenient to use.
Bruce Nyman, Mark Itzler, and Xudong Jiang are with Princeton Lightwave in Princeton, N.J. Mike Krainak is with the National Aeronautics and Space Administration (NASA).
