I recently used a standard CO2 (should be written as CO2 , but few people do so) inflator from Genuine Innovations to fill a new bicycle inner tube to about 130 pounds/square inch (psi), or 9 bar. It was the first time I used this type of unit; it's a lot easier and quicker than the hand pump, especially when you're on the road.
What was interesting was that a solid formed on the bike tube filler valve and stem as I filled it. I didn't immediately understand why, but a little thinking, a look at the vendor's web site, and cracking the old thermal-analysis textbook reminded me pretty quickly. It's all about energy, volume, expansion rates, and sublimation (no, not the kind that Sigmund Freud and his associates talked about, sorry).
This is what is going on : the gaseous CO2 is compressed to 900 psi (62 bar) and turns into a liquid, and that what is stored in the small cartridge. When it is released, the liquid expands rapidly and turns into a gas. But, basic thermal reality is that a rapidly expanding gas cools as it gives up its energy, and as a result, its temperature drops rapidly. (The standard refrigerator or air conditioner uses this principle, as well).
In this case, the temperature of the gas dropped so much and so quickly that it froze, yielding some frozen CO2 (better known as “dry ice”) on the valve. It's an example of sublimation because the gas went directly to the solid state, without an intermediate transition through the liquid phase. We see other examples of sublimation when there is warm, dry air directly above snow and ice on the ground (solid water phase): while some of the solid melts as a liquid, some of it turns directly into water vapor (gas) which we see as ground fog hovering over the solid mass.
What's the connection between this and analog or power-circuitry design? It's a reminder that thermal issues are always with us, and affect almost every design. Whether it's maintaining temperatures within in a narrow range for stability and precision in instrumentation, keeping a dissipative component from cooking itself (as a local maximum), ensuring that the overall system stays within allowed range (as a global maximum) or making sure a system and its components start up and work at low temperature, we have to factor thermal concerns into almost every design analysis. “Low-power” battery-operated designs can have their thermal issues, even if you are focused on milliwatt dissipation levels.
And you can't “cheat the heat” by allowing your hot IC to assume it has the entire PCB all to itself for heat-sinking, or that there are no other components on the board which are also sourcing heat or sinking it to the copper! ♦