MCT Noiseless Avalanche

Laser Focus World published a surprising paper for us, silicon junkies. The claim is that in MCT world an avalanche multiplication photodiode may have no excess noise associated with the electron multiplication.

MCT avalanche photodiode was first demonstrated by Jeffrey Beck at DRS Infrared Technologies (Dallas, Texas) in 2001. Uniquely to MCT, the avalanche process is electron-dominated, meaning that the bandwidth is independent of the gain, in contrast silicon, in which the gain-bandwidth product remains fixed. So far, no surprises, everything is obvious.

But a group at CEA-Leti (a laboratory of the Minatec Innovation Center in Grenoble, France) discovered that MCT avalanche photodiodes have an “excess noise factor”–the noise incurred by the avalanche gain process–equal to one. That is, the MCT avalanche photodiodes act as perfect amplifiers.

There is a theory of R. J. McIntyre, who in 1966 proposed that the excess noise factor of avalanche detectors could not go below a limit of two. “I think the observation of an excess noise factor lower than the well-established theoretical McIntyre limit is the result which has been central in the development of the MCT avalanche photodiodes,” said Johan Rothman, who heads the group at CEA-Leti, “It has been a true adventure to try to convince people for whom the simplified theory has become a true quantum-mechanical limit. It is particularly surprising given the strong simplifications that were used in the McIntyre theory.”

Indeed, since my school time I used to think that avalanche multiplication inevitably elevates the noise floor and reduces the detectivity, no matter what. Now I'm surprised to see this theory challenged. I'd like to learn more about the new MCT results before throwing out the trusted and proven old theory.

Many thanks to R.C. for the interesting link.