I don't know why the transistors fail.
Do you have the freewheeling diode connected?
This is a diode across the field coil, connected so it is back-biased when the power is turned on. When you turn the power off the voltage at the coil terminal switches sign but only goes to the diode voltage drop: The current in the coil circulates through the diode and the coil, gradually decaying due to diode voltage drop and coil resistance. The bulk of the energy from the decay of the magnetic field is dissipated in the coil - at a rate proportional to the square of the current, i.e. about the same heating as if the coil had a constant current supplied which gave the desired magnetization.
If you don't include the diode, when your driver transistor turns off the inductive kick raises the voltage across it until the current keeps flowing - decaying only in proportion to the back voltage across the coil. The result is that virtually all the energy stored in the magnetic field of the coil is dissipated as heat in the transistor. Normally a switching power supply's transistor spends nearly all its time either full-on (little voltage drop) or full off (little current) and thus has very little dissipation. But without the freewheeling diode the transistor is forced into conduction with a high voltage drop and dissipates as much heat as if you had an analog constant-current regulator.
It sounds to me like your freewheeling diode is missing or open. If so you just have a transistor in series with a coil across a power supply, so the transistor gets hot whether you run it class A (analog proportional control) or attempt to run it class D (switched).