Visible to SWIR eAPD Performance

A NASA Advanced Component Technology (ACT) program targeted development of electron avalanche photodiodes (eAPDs) that spanned the 500 nm to 2500 nm spectral range. This ACT task leveraged an existing eAPD that are used by astronomers and other NASA programs. However, the existing eAPD cut-on wavelength starts at 800 nm, due to the CdTe buffer layer through which photons have to traverse prior to reaching the HgCdTe absorber layer. This ACT task extended the eAPD detectors cut-on wavelength down to the visible wavelength range, which required the removal of the GaAs substrate and the CdTe buffer layer. The most challenging portion of the project was the development of a passivation layer at the illuminating surface that transmits photons between 500 nm and 2500 nm, while minimizing surface recombination velocity at the surface to maintain low dark current. The baseline was to demonstrate an eAPD with 500 nm cuton wavelength, with a goal of extending the cut-on wavelength into UV. A critical task on the program was development of a passivation process to drive minority carriers away from the passivant/HgCdTe interface towards the gain region. Two potential passivants were investigated. Results of the experiments showed one variant gave the best results and has been chosen as the deposition method. Progress was made in the development of visible to SWIR APDs in that QE was measured to be flat across the 500 nm to 2500 nm band but was low ~ 15% for one of the arrays. Another APD array had QE as high as 28%. The contention is that etching for the two arrays was different resulting in different HgCdTe absorber thickness values. Low QE is probably because the thickness of the absorber may be approximately equal to or longer than the diffusion length. Consequently, the thinner absorber regions result in higher QE, whereas the thicker absorber region of the array result in lower QE. Gain as a function of bias was measured and gain M ranged from ~ 25 to 50 at 10 V bias exceeding the M = 10 at 10 V requirement. Improving QE in the future will be by etching further into the HgCdTe absorber. This will reduce the distance that the carriers have to diffuse to reach the junction and should increase the QE.