Intermittent voltage recovery after discharge in X7R capacitors

Observed behavior

In a sample-and-hold front end, an X7R ceramic capacitor is discharged to ground and then shows a residual voltage reappearing across its terminals seconds later. This matches the documented dielectric absorption (soakage) effect, where dipole relaxation in the dielectric causes partial voltage recovery after discharge.

Constraints and uncertainty

The capacitor is discharged through a low-impedance path before measurement, yet a voltage reappears at the input of the ADC driver stage. The concern is whether this recovery can push the measurement circuit outside its acceptable input range or corrupt the next sampled value. Datasheets inconsistently specify dielectric absorption, and the magnitude varies with temperature, aging, and DC bias, so the 25°C figures may not reflect worst-case conditions.

Assuming a Class 2 dielectric like X7R, the recovered voltage could reach a few percent of the charge voltage and persist for seconds to minutes.

Is the observed residual voltage consistent with dielectric absorption, and how should the discharge protocol be adjusted to ensure the capacitor is truly at zero volts before measurement?

Would substituting a C0G/NP0 or film capacitor be expected to eliminate the recovery, and is there a standardized way to verify the effect across vendors?