A few possibilities spring to mind, although most of them may be ruled out for different reasons.
1. Carb icing in the R22 can happen during flight below 18" of Manifold Pressure even when the Carb Temp indicator is out of the of the the yellow range. Due to the governor gradually opening the throttle up very slowly, the 2 guys may not have been aware and as the butterfly valve area of the carb ices up to the point where the fuel/air mixture no longer flows the engine may quit instantly. This would cause rapid decay in RRPM. If an autorotation isn't successfully entered, the low RRPM could lead to excessive flapping that could chop off the tail boom. As the rotor slows down towards the stall the blades start to cone up more than normal due to the loss of centrifugal force, they will then tulip upwards as the helo falls. This is unlikely, due to the amount of time the instructor had in Robbies. I'd like to think that anyone with that much experience in the R22 wouldn't let that carb ice scenario happen. But !!!! happens sometimes, so it can't be ruled out.
2. Surprise engine failure(throttle chop) by the instructor was met by the wrong reaction by the student, raising the collective. This would cause low RRPM, which could lead to excessive flapping that could chop off the tail boom just like above. This one may be most likely but could be argued against due to the carb heat not being on. I'm not a fan of throttle chops myself, but I know other instructors that will give them. I don't know any instructor who won't find some sort of sneaky way to apply full carb heat before a surprise auto(throttle chop or not), especially in those meteorological conditions. Although, maybe he forgot(nerves due to first instructional flight).
3. Retreating Blade Stall, can cause a rapid pitch up and roll to the left. If the person on the controls applies too much aft cyclic to try to slow the chopper down it could cause the already low rear of the rotor disk to come in contact with the tail, thus chopping it off. Proper recovery is to lower the collective pitch of the blades and then slow the helicopter down with small rearwards inputs to the cyclic while correcting the roll. From what I hear, RBS is quite a violent experience, so the wrong large inputs might be very plausible. This scenario is probably ruled out with the fact that the blades had tuliped upwards due to rotor stall. As with all helicopters, RBS is a greater risk at higher density altitudes of which there is no mention in your case study.
I'm no expert, but that's my wee attempt!