Phase change
You need a special fluid that changes phase at droplet-forming conditions (so, if you're at one Earth atmosphere and zero degrees Celsius, water would be good).
Next, you need a second mechanism whereby the phase change is slowed down something fierce; an organic compound with very high latent heat would be good. Water already has a high latent heat of liquefaction, but the more the better.
The fluid must have a very rigid crystalline form. Not mandatory but this allows more coolness to form.
Actually, any liquid-to-solid phase change will do, not just freezing. Maybe air exposure is enough.
However, the fluid deposits on a surface in domed drops and starts cooling. Phase transition kicks in, gradually from the outside inwards, creating a core of liquid under greater pressure. As a result, the fluid starts spurting out.
It can do so gradually, and your drops will sport a single spike, or it fits and starts - and you get a sort of snowman, with a domed drop with another drop on it, and another, and another, maybe stacked vertically or wobbling every which way.
Coagulation
Same as above, but you get a higher gross factor. The "solid" coagulated phase undergoes a partial colliquation some time later, so the drop:
- starts as a normal liquid drop
- becomes opaque
- after some minutes, it starts shrinking and splits open in two to six symmetrical segments and sprouts as many smaller bubbles. The original drop is now a shriveled "hub" hidden by two to six smaller droplets. With large enough drops to begin with, the process can iterate fractally.