Hum... this is certainly something I didn't know about.
We know of various mechanisms of electron emission from a surface, especially metallic surfaces. Field emission and photoemission are two common examples. However, it appears that there's an emission that has no present-day explanation ... yet. This emission occurs as the temperature of a photomultiplier is cooled down below 220K, and the emission rate rises as one cools further.
This, of course, is counter-intuitive. In many experiments, we perform our measurements at very low temperature to reduce the "noise" level in our experiments. Thermal noises and also other agitation causes our data to also be "noisy". The Fermi function is also less spread out above the Fermi energy, so less number of energetic electrons are available as one approaches very low temperatures. So one would expect that a photomultiplier, which is nothing more than a cathode with an applied field, and a mechanism to multiply the electron emitted, would be less noisy as well as one cools it down. Instead, what we see here is that below a certain temperature, the number of dark counts seems to increase.
This certainly is an interesting phenomenon, and it is of interest to those in the detector physics field.