To get a good seal you will need a pre-constricted tube and will need to seal it with a hot sharp flame from a glass-blowing torch or similar. The sample tube will need to be under a good dynamic vacuum at the time (connected to a vacuum line), and the bottom of the sample will have to be frozen in liquid nitrogen, of course. With practice, you can get very good seals this way. Move the torch around the constricted part of the tube and let it collapse in on itself. itself due to the vacuum. I believe that this is the way that commercial sealed samples (e.g. test samples) are made. For liquid NH3, you will probably want to use thick walled tubes.
WARNING: While the tube is warming up, treat it as a bomb - just in case you have created a pin-hole leak and have condensed liquid oxygen from the air into the tube. And yes, we have had it happen here. Also, if you are using a solvent with a vapour pressure of 10 bar, you must handle the sample at all times with the appropriate safety equipment. Eye protection at the very minimum.
Your NMR facility manager may also have some opinions on loading pressurized samples into the instrument. I would probably insist on using an old probe that I don't care all that much about.
