Carbon soot can build up inside of the isolator if it is operated under load due to arcing when the contacts opens and closes, some of the carbon soot may have provided a path between the L and N to allow arcing during the inrush of current when switching on?
Just a thought :)