Not all epoxies stand up to water. 6 minute epoxy, not so much. Arctic silver is fast, thus, I would be careful to research and test.
Yes, price is insane for arctic silver-making it only appropriate for small, single projects. So, I just add copper or aluminum powder to normal epoxy. JB weld already has steel (lower conduction than aluminum, but it is something) in it. .... The copper was costly, so only got a little. Engineering toolbox has its thermal conduction way ahead of the closest second, Aluminum. The powered aluminum I got cheap, probably was termite. So, I store it in a bag, in a tin, duct taped, and inside a larger tin, far away from fire risk ideally, labeled and in a detached shed or garage. Then again, I am just a seat belt wearing, protected cell using fellow. You may feel free to store your thermite by your bed, beside your unprotected cells.
I mistakenly bought a pound of zinc oxide, on cpf recommendation, back in 2012- have used relentlessly, yet still have 80% left, while a few builds seem to have consumed most my copper. But upon, research I do not believe an oxide of any metal conducts anywhere as well as the pure metal, much less the thermite aluminum, even less, the copper. Empirically, only crude soldering iron plus finger, heat tests across joined aluminum heat sinks. But nothing I concrete enough to draw any conclusions, about which glue, or additive, is better. A direct metal, or ultra thin epoxy (to fill micro air gap) bond, is ideal. Thermal conduction of arctic silver is horrible on their own data sheet, compared with real metal. (No luck yet soldering aluminum for me, which is my holy grail. My problem is that my builds rely on the epoxy for the form. The great plains or bob Smith 6 minute, with additive seems to last 2 years, with no water exposure. I imagine jb weld standard formula would go decades, but likely would need over night to set with additive. I was able to get great planes 6 minute to set up in 10 minutes with a heavy percent thermite two days ago. I believe I recall zinc oxide slowing down set time of bob smith epoxy a bit more, bur not too much more. Again, not tested side by side, all permutations, neither with notebook. )