I read in a book about an effect called "dielectric absorption".
Basically it is an effect that each capacitor has. Depending on the dielectricum it is larger or smaller.

When a capacitor is discharged the capacitor charges up again after removing the shortcut between the 2 terminals.

Of course I had to test it myself :-)

My test setup:



The device under test is a 220µF SMD Tantal capacitor from a junkbox.

I connected a scope (10 MOhm input impedance) to the 2 terminals and charged the capacitor up to 10V.

I discharged the capacitor with a 680 Ohm resistor. As soon as 0V was reached I removed the resistor and watched the scope:



Cool, it really charges up! After a few seconds it looked like this:



So the capacitor charged "itself" to 280mV!

It does not matter how the capacitor is discharged, even a direct shortcut leads to the same result.

Apart from things like piezoelectric effects this is also something that has to be considered in some electronic circuits (e.g. integrators)...