I'm not sure of having understood your problem. You mean that, after some time of taking CO2 and temperature measurements, you start to get senseless values and you have to reset the mote to get back to normal operation? That's pretty weird, could you post the code you are using?
Regarding the conversion into ppm, you have a graph of the electromotive force variation with gas concentration in the documentation of the board. You just have to subtract the voltage read from the voltage measured at 350ppm, divide by the gain set and translate that value to the graph to get the gas concentration.