Speed optimized large area avalanche photodetector in standard CMOS technology for visible light communication

Abstract

This paper presents a speed-optimized large area avalanche photodetector (APD) in standard CMOS technology for visible light communication applications (VLC). Recent research efforts have reported high speed CMOS APDs with low breakdown voltage for considerably small photodiode sizes, which limits the APD usage in low cost optical receivers for VLC. The speed of a large-area APD dramatically decreases due to increased transit time of diffusive carriers in charge neutral regions. The proposed technique divides the active area into multiple subsections to decrease transit time and increase speed. A prototype 350×350 µm 2 APD is fabricated in 0.13-µm CMOS technology. The photodetector achieves a maximum gain of 7.6 K at 11 V reverse bias, showing excellent agreement with simulation results as calculated using the nonlocal impact ionization model based on recursive dead-space multiplication theory (DSMT). 2-D device level simulations validate the speed enhancement by comparing the small signal simulation results of three P+/N-well photodiodes with the same area detector composed of different number of sub-sections.

High-Speed Integrated Visible Light Communication System: Device Constraints and Design Considerations

“…This led to the hypothesis that it is better to build the APD out of a 10×10 array of 20 µm×20 µm structures, whose outputs can be summed together by connecting the structures in parallel, in order to meet the bandwidth requirements. This method was also used in the paper recently published by Ray et al [31]. The dominant source of noise will come from the TIA connected to the APD.…”

Bit Error Performance of APD and SPAD Receivers in Optical Wireless Communication

“…Via pin-photodiode APDs, with diameters of 200, 400 and 600 µm, we ended up with an 800 µm-diameter APD integrated in a 0.35 µm BiCMOS receiver [30]. A segmentation of the large-area APDs for high data rates as it was used in [32] was not necessary.…”

Exact Analytical Formula for the Excess Noise Factor for Mixed Carrier Injection Avalanche Photodiodes

“…In order to obtain an exact analytical formula for F(x), we solve the recursive integral equations from the DSMT model [10] with zero dead space to obtain the second moments of electrons and holes, z2(x) and y2(x), respectively, under mixed injection. The recursive integral equations for the second moments of Z(x) and Y(x) are equations (22) and (23) in [10], and they are expressed below: The exponent r turns out to satisfy the same characteristic equation as in (9). Upon substituting the proposed forms from (19a) and (19b) into (18a) and (18b) and applying boundary conditions 2 ( ) = 2 (0) = 1, we obtain a system of twelve linear equations with ten unknown coefficients p1, p2, p3, p4, p5, q1, q2, q3, q4, and q5.…”