This is a prime example of a picture being worth 1000 words, but I can’t find the damn picture I took many years ago to scan it and provide a link so you can all see just WTF I am talking about. This thing has bothered me for all those years but only now, with the advent of SDMB do I have any hope of getting an explanation. The office I worked in at the time had a standard home-style refrigerator into which we would put our brought-from-home lunches. We would also make ice cubes for drinks and sodas. This was before automatic ice makers became popular, so we simply filled and emptied the trays manually. About every two to three weeks or so, we would discover a cube with a handle growing out of it. It would be in the form of a tapering spike of ice about 1/2 inch long, about 3/16" diameter at the root and “growing” upwards at an angle of about 45 degrees from the top center of the cube. Usually this would form on only one of the cubes in the tray although occasionally other cubes would also have noticable bumps in their tops, but no spikes. I don’t remember if the trays were metal or plastic (if that would make a difference.)
My engineering sense is that the cubes freeze from their outer surfaces toward the center, so in the process, the center volume of water gets "squeezed" and of course it has no place to go but upward, but why doesn't this happen EVERY time? Why only every couple of weeks and why only one cube in the tray? If some random stimulus, such as vibration from the compressor running at a critical point in the freezing cycle was a factor, shouldn't it affect ALL the cubes in the tray? And why was the spike always at a 45 deg. angle? Any H2O hypothermic dynamicists wanna tackle this one? I'll keep looking for the picture.