You'll need a series gate resistor, as recommended by Infineon, to limit the tempFET's thyristor current if it triggers.
The bts115a is only rated for 50V, so if you expect an inductance flyback beyond this the resulting avalanche condition has to be considered. tempFET appnotes show a drain-to-gate zener-plus-diode to re-activate a shutoff gate during an overvoltage, and clamp the flyback at the zener-plus-gate-turnon voltage (this FET active-clamp trick works well for low-voltage high-gate-capacitance FETs, but should not be used for high-voltage FETs, which can break into RF oscillation). Because the active-clamp trick has you connecting stuff between high voltages on the FET's drain and its vulnerable gate, they also suggest a gate-source zener diode to prevent the voltage from ever momentarily going over the FET's 10V limit. For this purpose, a 10V zener should work fine.
Infineon also suggests choosing a lower-voltage gate zener to clamp the maximum gate voltage, in order to create a short-circuit current limit for the FET. This idea is problematic to me, on its surface. First, for any given drain current, there's a gate-voltage variation due to drain temperature. Perhaps a counter to that concern is to pick the gate voltage at high FET temperatures. Second, low-voltage zener diodes, under 5V, have notoriously poor qualities, with poor voltage tolerance and dynamic resistance, i.e. the zener voltage varies dramatically with zener current. A counter to that concern could be to use an active zener, e.g., a TL431 with two resistors. This zener can be programmed to as low as 2.5V and is fairly fast.
You mention PWM control of your fan. Keep in mind the gate series resistor required for tempFET operation, and the low gate voltages required for using the current-limit concept, will both in the end limit your PWM switching speed. There are several other ways to protect a switching FET in the event of a load short circuit.