Hi Michael,
Please could you post the code that you're running during your test so I can try it on our board? Our results suggest a current drift of 22pA/°C, so I'll be interested to see your results with the 100k resistor.
Regarding the heating method, we originally found this problem when heating boards in a Thermotron climatic chamber, so it seems to be repeatable for different heating methods.
Thanks
Tim