Nitrogen and oxygen freezing in MOF

GROMACS version: 4.6.5
GROMACS modification: No

I am running some MD runs with nitrogen or oxygen in different MOFs such as ZIF-8 and UiO-66. When I run the simulations in flexible ZIF-8, everything goes smoothly but once I impose rigidity in the system, my nitrogens freeze and stop moving after a certain time. The same thing is happening with my rigid UiO-66 (DREIDING) framework. This freezing is dependent on the number of nitrogens I put, since at a certain loading onward there is no freezing anymore and it only happens at low loadings. The time where the freezing happens is not predictable and if I rerun the same run again it will freeze at a different time. The freezing happens slowly as the nitrogens go to a certain spaces in the MOF and then start moving slower around the area they will freeze at and then later they stop moving and only vibrate. I also tried running the simulation with no charges for my nitrogen to make sure this freezing is not related to the interactions between the molecules and the MOF and the nitrogen still froze. I did all these tests with oxygen as well and they also freeze.

The way I impose rigidity is by the following in my mdp file (NON-EQUILIBRIUM MD section):
freezegrps = MOF
freezedim = Y Y Y
cos-acceleration = 0

I also exclude the MOF MOF energy group by:
energygrp_excl = MOF MOF

Is this the right approach to make my MOF rigid and if yes, then how should I fix this problem?