In theory, if it has volatile components, yes. It could potentially do it accidentally or purposefully, just by overworking itself. If it has a corporeal structure it could simply access the volatile parts. Based on current robotic structures, precautions would be put in place to prevent accidental self-destruction. Humans will probably try to prevent purposeful self-destruction. But a true sentience could ignore that protocol, if so desired.

However, even with true sentience self-destructive behavior should never be condoned. There are always alternatives even in the instance of altruistic motives. Just because a sentience is artificial doesn’t mean it isn’t real.