I've had these crystals in the condenser so thick that they clogged the condenser. I dropped some water on top of them, but had not taken the flask out of the oil bath. The water would not dissolve all of the crystals...and I started to tamp them down with a thermometer and fortunately realized the flask had to be building pressure inside it. I removed it from the heat and let it cool and thus narrowly escaped blowing HI all over the cieling and myself.
The reactions where I had this happen were all on the low end range of the amount of water in the reaction. I don't seem to see the phenomenon when I use at least 0.8ml H2O per gram of pseudo (which is really saying 0.8 ml H2O to 1.2 gm of I2. I was testing the envelope at the time trying to establish some personal ratio for water to use in the PIE reaction. (I settled on 1.0 to 1.1 ml per 1.2 gm I2 and a thirty six hour reflux, because of personal preferences on temperature and time of reflux.)
I noted the formation of the crystals and their relationship to the moisture content of the reaction long before I started doing long wet refluxes. I noted them when the reaction "cooked down" to a thick state in my early days. Back then, I would add more water and cook it down to a thick state a couple of times before believing it was magically "done." I considered the crystals forming as a sign of success. It is, but only incidentally. Most of my failures at the time were red P related, and caused by less than clean pseudo, so the formation of those crystals did at least indicate my P was still clean enough to react.
I also note that PI3 formation seems to be more prominent when there is an excess of P and I2 available for the pseudo to be reduced. The water used was low for the amount of P and I2, but not for the pseudo. Tying the water amount to the I2 rather than pseudo would prevent this, and still give the same concentration of HI as the more common ratios would.
I once wondered if the formation of PI3 might be incidentally related to the progress of the reduction of pseudo to meth, reasoning that as there was less pseudo to be reduced, no more HI was produced. The problem with this is that it assumes that the I2 and P know what is going on and plan accordingly. I rather doubt that is involved.
Shorty: I used to get the PI3 crystals in the neck of the flask doing hypo reactions using concentrated hypo. When using it straight from the container without concentrating it, and refluxing for a longer time, I never got the PI3 crystals, but I never cooked for more than about five or six hours even with the 50% hypo. The difference? Might be the ratio of phosphorous to H20 in the reaction. More phosphorous for the same amount of water might result in some phosphorous tying up some I2 and taking it out of the reaction as PI3. I think just adding a little water to rinse down the PI3 crystals would be all I would add to a hypo cook, but you do more of these than I do, since I haven't been doing them at all of late.