The impact ionization peaked higher. But when we tried 12V again, the peak was higher than it was before - the detector seems to have changed state again. Before it seemed like running at 600 mK caused it to change state to stronger impact ionization - now changing the bias during a series seems like it made it change state again.

In the series above we also tried removing bias 15 minutes into the series, and saw some evidence of reverse bias (inverted events) presumably caused by buildup of space charge in the detector

To do: more temperature trials, try moving source partially over blocker to reduce rate. Maybe repeat moving source during series. Maybe repeat changing bias during series.