AF model at low temperature

[mosch]

I want to calculate the AF of low temperature (-90 degree), what model will be select?
And the arrhenius model suit for high temperature, but at low temperature it will be not comfortable.

[adam]

The Arrhenius or Eyring models are typically used when temperature is the factor (stress). Just make sure that you use absolute temperature scales such as Kelvin. In theory if temperature is the factor then the Arrhenius or Eyring models should be valid, even at low temps.

[Dennis Craggs]

This depends on the Physics of Failure. For electronic aging, the Arrhenius model applies. However, if you are considering fatigue failures of the solder used to attach ICs to a substrate, then a coffin-manson solder fatigue failures should be applied. Here the stresses are generated by the part geometry, differences in thermal expansion rates between the device and the mounting surface, and temperature changes. The temperature at which there is nearly 0 stress is at high temperatures, somewhat less than the solder melting point since there is some solder creep. The greatest stress levels occur at low temperatures since the product of the temperature change and the difference of coefficient of thermal expansion increases at low temperature.

So for the first question, "What is the Physics of Failure?" The second, "Is there a failure model?"