According to a post Mike Moore (@hms) made in the past on the TPZ, they used glycerine, water, and food coloring for the TNG hypo vials. I haven’t seen much on how well those held up though, but perhaps some auction research would shed some light.
I believe that the design I put into my CAD model for the vial (based @nicksdad’s suggestion) will have the best possible chance of preventing leakage while still retaining some modicum of screen accuracy—the screen-used ones, after all, really only needed to last a day or so, one assumes.
It is only two acrylic parts, which can be precision machined to friction fit tightly together. It isn’t just a disk that is welded on top of a cylinder with the same diameter, but it rather is a cylinder with a lip on it that actually fits into the other outer cylinder. This design means that it’s two flat surfaces mating together rather than just one. And by using only two vial parts, we’ve reduced number of seams to one (and hopefully the chances of a vial with a weld that doesn’t quite work).
The two vial parts could then be chemically welded together using solvent cement fed in around the lip when the two parts have been mated. Capillary action pulls the solvent into the joint and fuses the two parts permanently. (Anyone doing this would want to mask off the sides to prevent the solvent from marring the polished tube finish.)
That’s the theory anyway. It’s all fine and well designing these things in CAD software, but it’s all just video games until somebody actually tries making it.