nRF52832 occasionally doesn't start up with coin cell battery


we have some problems at the startup for our nRF52832 board powered with coin cell battery. The hardware include nRF52832 SoC, an accelerometer (LIS3DHTR) and battery holder for CR2032. We have also an additional 10uF capacitor in parallel with battery (in addition to 4.7uF on VDD, pin 48); the battery voltage is connected directly on VDD pins (pin 48, 36, 13). We verifiead that occasionally when we put battery inside battery holder the nRF52832 doesn't start up and seems to go in a state where it draw current that discharge battery in very short time (2 days). When the nRF52832 is in this unknown state the VDD voltage is 3V (the coin cell battery voltage) but there isn't activity on TX and RX and nRF52832 is stucked (if startup procedure is finished normally, nRF52832 will execute firmware instruction normally and therefore it will go on TX/RX periodically). We verified also that when we insert the coin cell battery inside battery holder some bounce on VDD supply voltage occurred. Basing on internal coin cell resistance and capacitor of about 15uF (10uF+4.7uF) the rise time of our "power supply" from 0 to 1.7V should be in the range 10usec.-10msec. and therefore tpor,10msec=9msec. What happen if during tpor time (9msec. in our case) after 1.7V, VDD voltage drop under 1.7V due to contact-non contact from battery with battery holder? What could be the cause of the incorrect startup of the nRF52832 SoC?

Best Regards