Failure of polypropylene motor-run capacitors

Over the years I've had a couple of motor-run capacitors fail. A plastic cased one had some grey gunk come out of it. Recently I had an aluminium cased polypropylene film capacitor fail, and drop to <10% of its rated capacitance. When I opened it up, there was what looked like lumps of silver grey plastic that had melted, flowed to the bottom of the casing and solidified.

I was just curious as to what is actually going wrong in these capacitors.

Also, the latest one was on the refrigerant compressor of a heat-pump tumble dryer. Would I get better capacitor life and / or lower power consumption if I arranged the the capacitor disconnected after the heat pump started?

Modern capacitors are made by sputtering metal vapor onto thin polymer film (physical vapor deposition in a vacuum chamber), cutting to size, adding connectors to left and right edges, rolling them tightly, sliding them into their case and sealing the case.

The polymer molecules in the film do move, undulate, flow slightly as they are thermal-cycled with each on/off cycle. The metal is in the 10-100nm range and film is in the 1 to 5 micrometer range.

Eventually, the very thin metal coating tends to flow across the surface of the plastic and agglomerate. The metal doesn't bond to the polymer very well. The polypropylene softens and moves at much lower temps than other options but it can be rolled (extruded) to much thinner sheets than other polymers. PC is a higher temp material but much more expensive and the film is thicker so it's difficult to get the same capacitance in the same space as your current device (or, a very difficult and expensive process to make thin PC is required and Bayer was the only company to do it and still couldn't compete with PP so they quit making the film more than 25 years ago.

Ultimately, there is no way to prevent it other than making larger capacitors that can dissipate heat better.