Hi Marty,
There are several factors here. Damage to the part occurs when you instantaneously apply greater than 12Vin because the ceramic input capacitors resonate with the bench supply wiring inductance and create a leading edge transient that damages the device. This is a common malady when bench testing High input voltage parts that only have ceramic capacitors. Adding an Aluminium electrolytic greater than 47uF across the ceramic input capacitor will generally resolve this problem. This electrolytic might not be required in the final implementation. The rule of thumb is that the input voltage can ring to twice the supply input voltage. It is not clear why you seem to be damaging the 42Vin rate part with only 12V input setting.
Please examine the current rating of the inductor. Perhaps it is being saturated and causing early current limit. Inductors often have several current specs, the first is a thermal rating often related to 40C rise in inductor temperature, the second is usually peak saturation current which describes 20% decrease in inductance at the current specified. If you push the inductor current beyond this saturation spec, it may explain the current limitation you've observed.
Next look at the inductance value. For 8V out, it should be 33 or 47uH. If you took a 1.2V application board, and modified it to 8V out, the inductance value needs to change also. If the inductance value is too low, then peak current is high and current limit is reached prematurely.
For your application, 8V*.65A = 5.2W. A rough estimate of the losses in the switch are as follows,
a) RDson = 1ohm
b) Duty cycle is approx = 40% (for 20Vin and 8Vout)
c) Loss in the switch is .65*.65*1.5ohm = .63375W
d) Temperature rise in the part = .63375 * 121C/W = 76.68C.
It may be that current limit has some dependency on Vin that might lower current limit.
Thanks,
Anston