Research Overview

This research paper, presented at the 2024 International SPAD Sensor Workshop, introduces a novel series of planar InGaAsP/InP Single-Photon Avalanche Diodes (SPADs) designed for 1064 nm wavelength detection. The device features a 10 µm diameter design and utilizes a Separate Absorption-Charge Multiplication (SACM) structure combined with dual zinc (Zn) diffusion technology to form the multiplication region and guard ring (GR) areas. This architecture enables back-side illumination of the SPAD and achieves low noise and uniform photoresponse through optimized Zn diffusion depth control. 

Research Team

The research was conducted at the Swiss Federal Institute of Technology Lausanne (EPFL) by a team including Utku Karaca, Ekin Kizilkan, Claudio Bruschini, and Edoardo Charbon.

Design Highlights

The design focuses on optimizing structural parameters through TCAD simulation and implementing SACM structure with back-side illumination to achieve low noise, high photon detection probability (PDP), and minimal timing jitter.

• TCAD Simulation for Zn Diffusion Depth: To minimize noise and achieve uniform photoresponse, researchers utilized TCAD simulation to determine the optimal depth difference between shallow and deep zinc diffusion layers. The simulation results indicated that a 0.5 µm depth difference provided optimal performance, despite slightly affecting photoresponse uniformity. Figure 3 demonstrates the simulation results, while Figure 4‘s SEM imaging confirms that the fabricated devices achieve the designed Zn diffusion depth specifications.
• SACM Structure Implementation: The SPAD employs a Separate Absorption-Charge Multiplication (SACM) structure, segregating the absorption and multiplication regions for enhanced electric field distribution control and improved device performance.
• Dual Zinc Diffusion Technology: The implementation of dual zinc diffusion technology forms both the multiplication region and guard ring (GR) areas. The guard ring design effectively reduces edge breakdown probability, enhancing device reliability.
• Back-Side Illumination Design: The n-type contact layer is positioned on the top surface, enabling back-side illumination of the SPAD. This configuration eliminates top metal layer shadowing of incident light, maximizing photon detection efficiency.
• Absorption Layer Specifications: The absorption layer is designed with a thickness of 1 µm and a charge layer doping concentration exceeding 2×10¹⁷ cm^-3, ensuring complete carrier depletion within the absorption layer before reaching breakdown voltage.
  Figure 2 illustrates the electric field simulation results, confirming sufficient field strength for complete absorption layer depletion.
• Multiple Multiplication Region Thicknesses: To investigate the impact of multiplication region thickness on device performance, three variants (1.5 µm, 1.3 µm, and 0.75 µm) of 10 µm diameter SPADs were fabricated. Results indicate that thinner multiplication regions enhance PDP and reduce timing jitter, though at the cost of increased dark count rate (DCR).

Electric Field Distribution Simulation

Figure 2: TCAD simulation of the electric field distribution, demonstrating the field profile within the SPAD device at 300K. The simulation reveals peak field intensity in the multiplication region, essential for avalanche multiplication, while maintaining sufficient field strength in the charge layer for effective photogenerated carrier transport from the absorption region to the multiplication region.

Device Structure and Fabrication

Device optimization involved TCAD simulation-guided Zn diffusion depth determination to achieve low noise and uniform photoresponse. Simulations established that a 0.5 µm depth differential between shallow and deep Zn diffusion layers provided optimal performance. Scanning Electron Microscopy (SEM) imaging verified that fabricated devices achieved the designed Zn diffusion depth specifications. Additionally, researchers fabricated SPADs with varying multiplication region thicknesses (1.5 µm, 1.3 µm, and 0.75 µm), characterizing their avalanche breakdown and punch-through voltages through current-voltage (I-V) measurements.

Dark Count Rate Measurements

Figure 7: DCR measurement results for devices with different multiplication region thicknesses. Figures 7(a), (b), and (c) present DCR measurements for SPADs with multiplication region thicknesses of 1.5 µm, 1.3 µm, and 0.75 µm respectively, tested under varying temperatures with a 10 kHz gate frequency and 100 ns gate width. The data demonstrates significant DCR reduction with decreasing temperature across all devices.

Performance Characterization and Analysis

• Dark Count Rate (DCR): DCR serves as a critical performance metric, quantifying the noise signal generation in the absence of incident photons. Time-gated measurements across various temperatures revealed that SPADs with 1.5 µm multiplication region thickness exhibited a median DCR of 53 kcps at 300K under 5 Vex bias, reducing to 2.75 kcps at 225K. This temperature-dependent improvement primarily results from reduced thermal carrier generation, the dominant DCR mechanism.
• Photon Detection Probability (PDP): PDP, a key indicator of SPAD sensitivity, represents the probability of detecting incident photons.
  Figures 9(a), (b), and (c) present PDP measurements at 1060 nm for devices with multiplication region thicknesses of 1.5 µm, 1.3 µm, and 0.75 µm, conducted at 300K using a monochromator and broadband source, with 10 kHz gate frequency and 100 ns gate width. Results demonstrate superior PDP for thinner multiplication regions, with the 1.5 µm device achieving 19.5% PDP at 300K and 5 Vex, while the 0.75 µm device reached 21.5%. Notably, the 1.5 µm device achieved an impressive 36% PDP at 9 Vex.
• Timing Jitter: Timing jitter, characterizing temporal response fluctuations, significantly impacts timing resolution. Time-Correlated Single-Photon Counting (TCSPC) measurements at 300K revealed FWHM timing jitter values of 118.4 ps, 110 ps, and 84 ps for multiplication region thicknesses of 1.5 µm, 1.3 µm, and 0.75 µm respectively. The reduced timing jitter in thinner multiplication regions corresponds to shorter carrier transit times.
• Afterpulsing Probability (APP): Afterpulsing, resulting from trap-mediated carrier release following avalanche events, affects counting accuracy.
  Figure 10 displays avalanche pulse interval histograms under varying temperatures and gate frequencies with 100 ns gate width. Analysis of these histograms enabled APP characterization across operating conditions, revealing an APP of 11.1% at 300K and 500 kHz gate frequency, reducing to 5.8% at 200 kHz. These results demonstrate robust high-frequency operation capability while maintaining low afterpulsing probabilities.

Comparative Analysis and Applications

Comparative analysis with existing InGaAsP SPADs demonstrates these devices’ superior characteristics, featuring minimal footprint, enhanced PDP, and reduced timing jitter. The InGaAsP/InP SPAD platform shows promise in several key applications:

• LiDAR and Range Finding: The 1064 nm operational wavelength, exhibiting minimal scattering and absorption in long-range transmission, proves ideal for LiDAR systems in autonomous vehicles, topographical mapping, and environmental monitoring.
• Free-Space Optical Communications: The resilience of 1064 nm wavelength transmission to atmospheric effects enables robust free-space optical communication systems, particularly in satellite communications and terrestrial optical links.
• Medical Applications: Time-Gated Diffuse Correlation Spectroscopy (TG-DCS) utilizing 1064 nm wavelength enables precise blood flow velocity measurements and other biomedical sensing applications.
• High-Power Laser Applications: Integration with 1064 nm Nd:YAG laser systems, where these SPADs serve as highly sensitive detectors for various high-power laser experiments and measurements.

Research Conclusions

This research successfully developed and characterized a series of high-performance InGaAsP/InP SPADs, demonstrating exceptional performance metrics across dark count rate, photon detection efficiency, timing jitter, and afterpulsing probability. The devices show significant advantages over existing technologies in comprehensive benchmarking.

The development addresses a critical gap in near-infrared single-photon detection capabilities. While CMOS SPADs exhibit degraded photon detection probability (PDP) in the near-infrared (NIR) range, requiring alternative solutions for 1064 nm single-photon imaging such as Superconducting Nanowire Single-Photon Detectors (SNSPDs), these InGaAsP/InP SPADs provide a compelling solution. Their ability to operate efficiently at substantially higher temperatures than SNSPDs, including room temperature operation, enables scalable, compact, and cost-effective implementation.

Key achievements include:

• Superior room-temperature performance compared to alternative technologies
• Excellent scalability and integration potential
• Cost-effective solution for 1064 nm single-photon detection
• Broad application versatility across multiple fields

Published at the 2024 International SPAD Sensor Workshop
Image and Content Source:
https://www.imagesensors.org/Past%20Workshops/2024%20ISSW/Papers/R02.2.pdf