SicelTech wrote ...

Tyndall Labs in Cork or REM in UK sells RadFETs that have Al gates and 400nm up to 1000nm gate oxide oxides. The oxide thickness determines to a large extent the sensitivity of the device. You've found papers by Len Adams and you can also look for papers by Andrew Holmes-Siedle. The sensitivity is most directly tied to the thickness of the oxide and the bias (or E-field) across the gate. Sensitivities of 0.5mV/cGy (no bias) to 1.5mV/cGy (9V bias) are possible on these devices. The cGy unit is newer unit and is used because it's the same as the rad which was the old unit. Off the shelf devices which do not have a fairly thick oxide will not work as well and will not be as sensitive. The Adams stacked RadFET design was done using 400 and 1000nm oxides and got 84mV/cGy (rad).
Hope you operate at ZTC zero temperature coefficient (in your Adams paper). That will provide the most reliable data. The poly-Si gates work as well because the oxide is doing the trapping (oxide is a sea of dangling bonds which are oxide traps). The poly-si gates have a different threshold transition due to the different work function.

Cheers