The crater and magic smoke release is the best clue here. I've had parts returned with a crater hole in a past life as an applications engineer. This is a very strong indicator that you are causing a latchup condition in your opamp. This can happen by either a voltage spike on its power rail, or more commonly, by driving one of the input pins above its positive supply or below its negative supply.
This wikipeda page explains latchup, though it doesn't clearly explain the trigger mechanism. This TI white paper explains it in much greater detail.
A latchup can be triggered by a voltage spike on the power rail, or from driving the input or output beyond the power supply voltage (below GND or above V+) such that enough current flows to trigger the parasitic thyristor structure inherent in CMOS devices.
For example, lets say you have a long sense cable on the opamp inputs. You've decoupled it with ceramic capacitors to filter the noise like a thoughtful engineer. What you may not realize that hot plugging this cable will result in LC ringing (made worse by low-ESR cap like ceramic) that greatly exceeds your supply rails and trigger latchup in the circuit sensing this cable. This is a common culprit. I have scope plots of USB VBUS at the device end ringing to 9 Volt peak when plugged into a PC with a 6 foot cable. Overshoot, noise pickup, LC ringing, etc. all must be designed for when dealing with cables.
To reduce the likelyhood of latchup, you can put series resistors in your inputs. A likely culprint is the + input of the opamp in your circuit. A 1k or greater resistor should do it.
Note that it is not the input overvoltage or undervoltage that damages the device in a latchup situation. The over- or undervoltage momentarily drives current into or out of the IC input to the point that it turns on the parasitic thyristor inherent in all CMOS devices. Thyristors are current triggered. This parasitic thyristor then causes an internal short of the supply rail to GND. If the transistor happens to be small, then you might just see supply current increase but the circuit functioning normally or only somewhat impaired. (You would have to remove power to reset the latchup.) But if the transistors getting turned on are large, then a very large current will flow and will damage your IC from thermal stress.
By inserting a series resistor on IC inputs exposed to possible over/under voltage, you reduce the current below what may trigger the parasitic thyristor.
It also sounds like the opamp is connected to a high current capable power supply. You can also help alleviate the problem by putting a current limiting resistor in the supply of the opamp especially since it does not draw very much current. If the latchup occurs, the current limiting resistor will limit the supply current and prevent device damage. Also, the supply current during latchup will probably now be too small to maintain the latch condition. Note that you should put a decoupling cap on the opamp to maintain stability. So maybe 100 Ohms and 0.1 uF at the opamp. This supply series resistor would also definitely help if the latchup is being caused by voltage spikes on the power rail by limiting the latch current.
I hope that helps find your culprit, -Vince
