Semiconductor photodetectors are critical components in LiDAR systems, enabling precise detection of reflected light pulses for distance measurement. Among the most widely used detectors are silicon photomultipliers (SiPMs), avalanche photodiodes (APDs), and single-photon avalanche diodes (SPADs). Each offers distinct advantages in sensitivity, speed, and noise performance, making them suitable for different LiDAR applications, particularly in automotive and robotics domains where range resolution and eye safety are paramount.

SiPMs consist of an array of microcells, each operating as a Geiger-mode APD, providing high gain and photon-counting capabilities. Their key advantage lies in robustness against temperature variations and immunity to electromagnetic interference, making them ideal for automotive LiDAR operating in harsh environments. SiPMs achieve high dynamic range by summing the output of multiple microcells, allowing detection of weak signals while avoiding saturation from strong returns. Typical SiPMs for LiDAR exhibit photon detection efficiencies (PDE) exceeding 40% in the near-infrared spectrum, with timing jitter below 100 ps, enabling centimeter-level range resolution. Their digital-like output simplifies signal processing, reducing system complexity compared to analog detectors.

APDs operate in linear avalanche mode, providing moderate gain with lower noise than SiPMs. They are widely used in long-range LiDAR due to their balanced performance in sensitivity and speed. InGaAs-based APDs are common for 1550 nm LiDAR systems, as this wavelength allows higher laser power while maintaining eye safety. APDs achieve gains of 100 to 1000, with dark currents in the nanoampere range, ensuring reliable detection at ranges exceeding 200 meters. Their linear response simplifies analog signal processing for time-of-flight calculations, though temperature stabilization is often required to maintain consistent gain. APDs typically offer a PDE of 60-80% at 905 nm, with bandwidths exceeding 1 GHz, supporting sub-nanosecond timing resolution.

SPADs are single-pixel detectors operating in Geiger mode, capable of detecting individual photons with picosecond timing precision. Their extreme sensitivity makes them ideal for low-power LiDAR systems, where eye safety constraints limit transmit power. SPAD arrays, often fabricated in CMOS processes, enable solid-state LiDAR without moving parts, crucial for automotive reliability. A 32x32 SPAD array can achieve a detection latency below 50 ps, enabling millimeter-level range resolution at short to medium distances. However, SPADs suffer from afterpulsing and dead time, necessitating careful quenching circuit design. Advanced SPADs integrate time-to-digital converters (TDCs) on-chip, allowing direct time-stamping of photon arrivals for high-resolution 3D imaging.

Range resolution in LiDAR photodetectors depends on timing precision and detector bandwidth. SiPMs and SPADs excel in timing resolution due to their fast rise times and low jitter, typically achieving 5-10 cm resolution at 100 meters. APDs provide slightly lower resolution but better linearity for analog ranging techniques. Eye safety considerations favor detectors sensitive to 1550 nm wavelengths, where the eye's cornea absorbs most radiation, permitting higher permissible exposure limits. InGaAs APDs and SPADs are preferred for such systems, though their higher cost and cooling requirements pose challenges. At 905 nm, silicon-based detectors dominate due to lower cost and higher PDE, but laser power must be carefully controlled to meet Class 1 safety standards.

Automotive LiDAR imposes stringent reliability requirements, with operating temperature ranges from -40°C to 125°C. SiPMs demonstrate superior temperature stability compared to APDs, as their gain varies by less than 0.3% per °C, whereas APDs may require active temperature compensation. SPAD arrays face challenges in automotive environments due to ambient light-induced noise, necessitating narrow optical filtering and gating techniques. Robotics applications often prioritize compactness and power efficiency, favoring SPAD arrays or small SiPM modules that integrate amplification and processing circuitry.

Noise performance varies significantly among detector types. SiPMs exhibit higher dark count rates (100 kHz to 1 MHz per mm²) due to the large number of microcells, but correlated noise is minimized through careful design. APDs have lower dark currents but suffer from excess noise due to the stochastic nature of avalanche multiplication. SPADs face afterpulsing effects, where trapped charge carriers trigger false detections, requiring optimized quenching and recharge cycles. Cooling can reduce dark counts in all detector types but adds system complexity.

Integration trends show increasing adoption of detector arrays with on-chip processing. Digital SiPMs now incorporate photon counting and time-stamping at each pixel, while SPAD arrays integrate TDCs and histogramming for direct depth map generation. APD arrays remain limited by crosstalk at high gains but are evolving with smaller pixel pitches and lower capacitance. Future developments focus on improving near-infrared sensitivity, reducing power consumption, and enhancing array scalability for higher resolution LiDAR systems.

Material advancements continue to push detector performance. Silicon remains dominant for 905 nm detection, but InGaAs/InP and GaSb-based detectors are progressing for extended wavelength ranges. Quantum dot-enhanced photodetectors show promise for higher operating temperatures and tunable spectral response. System-level optimization, including matched filtering and adaptive gain control, further enhances signal-to-noise ratios in practical deployments.

In summary, SiPMs offer ruggedness and high dynamic range for automotive applications, APDs provide balanced performance for long-range systems, and SPADs enable ultra-sensitive, compact designs for robotics and short-range imaging. The choice depends on specific LiDAR requirements, with ongoing innovations addressing limitations in noise, temperature stability, and integration density. As LiDAR technology advances toward higher resolution and safety standards, photodetectors will remain a focal point for improving system performance and reliability.