Avalanche noise in submicron GaAs in InP structures

This work is concerned with the measurement and interpretation of avalanche noise in avalanche photodiodes (APDs). The design of a measurement system with a higher signal to noise ratio than previous systems reported in the published literature is described. The system uses phase sensitive recovery techniques, which allow the avalanche noise power and photocurrent to be measured unambiguously from the system noise and leakage current of the device under test. Using the measurement system described, avalanche noise measurements have been performed on a range of homojunction GaAs p"+-i-n"+ and n"+-i-p"+ diodes with 'i' region widths, w, from 2.61#mu#m to 0.05#mu#m, and on a range of homojunction InP p"+-i-n"+ diodes with widths, w, ranging from 2.40#mu#m to 0.24#mu#m. The results show that, for w #<=# 1#mu#m, the dependence of excess noise factor F on multiplication does not follow the continuous noise model of McIntyre, which has become well established since its publication in 1966. Instead, a decreasing noise factor is observed as w decreases for a constant multiplication. This reduction in F occurs for both electron and hole initiated multiplication in the thinner w structures even though the ionisation coefficient ratio is close to unity. These results suggest that, in GaAs and in InP, if a thin multiplication region is used it is not always necessary for the greatest ionising carrier type to initiate multiplication for the lowest avalanche noise. To interpret the measurements, a model is used that includes the effect of dead space; the minimum distance a carrier must travel to gain the ionisation threshold energy. The dead space is found to be increasingly important in these thinner structures and restricts carriers to ionising in the latter half of their transit of the high field. This restriction makes the probability distribution function for the ionisation path length more deterministic and, in turn, reduces the overall unpredictability in the multiplication process, leading to a lower excess noise. The effect of distributed carrier generation, and so of mixed injection, in a 0.1#mu#m avalanche region is also investigated experimentally and using a model including dead space. These results show that the noise produced from mixed injection is lower than predicted by McIntyre's model, but higher than that obtained with pure electron or hole injection due to the effect of dead space. (author)Available from British Library Document Supply Centre-DSC:DXN029328 / BLDSC - British Library Document Supply CentreSIGLEGBUnited Kingdo