ESP32-WROOM-32E EN pin Diode Behavior

[Widlar]

Hello -

I have ESP32-WROOM-32E in my design and have produced a few hundred devices.
I have had 2 or 3 devices break after ~year use. There with no communication from the ESP.

The ESP32 EN pin is being held low by the module on these devices. When my MCU pulls the pin HIGH with a PU resistor, I measure 0.5V on a non-functional board. When the MCU drives the EN pin low I measure 0.0V. On a functional board of the same design I measure 3.1V and 0V respectively.

The resistance to GND on this pin for a non-working and a working board is the same. In the megaOhm.
I measure an 0.6V diode on a working and non-working board.

I have cut all traces to the module and directly applied 3.3V to the pin with an external power supply. There are no other components connected to the pin. It is only the module.
I set current limiting on the 3.3V supply.
With current limited to 0.02A, I am able to get the EN pin to 1.1V.
When it is limited to 0.04A I measure 1.4V.

I have verified that 3.3V is being supplied to the module 3V3 pin on the broken boards. MCU is only capable of driving GPIO 3.3V.
There is no external access to the ESP Module, and we pass ESD testing on the device so this seems unlikely.

Has anyone seen this behavior before? What additional testing/debug would you recommend? Is there a way to replicate this failure mechanism?

Thanks-