It's a good thing nukes are really hard to make. If they could be fabricated as easily as mustard gas or gunpowder, we would be in a world of trouble, literally and figuratively. And there's another reason it's such a good thing. The process of 1) refining enough material for a single bomb; 2) slapping together a laboratory device using that material that will detonate under controlled conditions; and 3) squeezing that big sprawling mess into a much smaller volume that can be safely delivered by missile or bomber, leaves all kinds of telltale signs easily uncovered by an experienced team of veteran inspectors. A nation has to successfully accomplish all three of those tasks in the form of a dozen or more fission bombs, minimum, before there's anything remotely approaching an effective nuclear deterrent.
So how hard would it be for a nation to develop a basic, reliable, Hiroshima-level fission weapon capable of delivering a yield of 10 to 20 megatons kilotons of TNT without getting caught? Let's add the wrinkle of inspectors and other interested parties armed with high-tech detection gear swarming around. Join us below for a brief romp through the declassified high points of nuclear weapons design, and we'll touch on a few points therein.
The enrichment process has been covered well elsewhere. But briefly, the preferred method is to concentrate the U-235 isotope using gas centrifuges. Raw, refined uranium is less than 1 percent U-235. This material is combined with compounds of fluorine to produce uranium hexafluoride, also known as Hex. Hex is a heavy gas at the right temperature and pressure. It is injected into the first of many centrifuges in that state: Think of a very tall coffee can standing upright and spinning very fast on a vertical axis. Over time the lighter U-235 begins to separate from the slightly heavier U-238. To speed the process, the layer containing the higher concentration of U-235 can be drawn off and put into a second centrifuge for more rapid separation.
This process is repeated, over and over, from one centrifuge to the next. It takes an array of at least dozens of very large centrifuges like the U.S. uses, or hundreds to thousands of smaller ones like the Iranians or Pakistanis use, all operating over weeks or months respectively, before anything approaching weapons grade material—HEU at over 90 percent—is available in sufficient quantity to produce enough for one bomb. That kind of activity leaves a huge mess of radioactive remains inside those centrifuges.
Anyone with the right skill set and proper instruments can analyze a used centrifuge and quickly determine, from both the residue present and the condition of the interior, what approximate grade of material has recently been inside it. It can be even easier than that: If an inspector sees a line of centrifuges dead-ending into a fresh-looking cinder block wall and finds filled-in holes in the cement floor that look an awful lot like centrifuge anchor points, it's pretty obvious some have been hastily removed. There are many other giveaways, some subtle and some complicated.
But suppose a signatory nation, not that we're naming names here, somehow cheats for years without getting caught and has amassed enough material to make several bombs. Now, they're ready to take on the second Herculean task: Making a laboratory device that actually works. This is what hung up the North Koreans for quite a while. Their first attempt or two resulted in reactions that either didn't go super critical at all, or went too fast. This is what weapons designers call a fizzle. They got less than a half a kiloton of yield at first, and we still detected it! And that, my friends, is proof positive that detonating a fizzling nuke, even deep underground, in this day and age means instant detection by watchdog agencies all over the world. I'm told the seismographic signature alone is a dead giveaway. Either Iran in particular has been encircled with such detection gear for years, with spy sats and recon flights peering in from above, or my defense tax dollars have been woefully misspent.
Lastly, there is the task of turning that laboratory device into a reliable, deliverable weapon that detonates when you want it to and, most importantly, will not detonate when you don't want it to. There are a bunch of ways to do that. A couple of declassified examples are inertial tampers and neutron triggers. These add-ons not only ensure the bomb will blow up, but it is also possible to design the nuke in such a way that it won't fully detonate unless those additional devices are engaged. This means they can be separately armed, adding a layer of security, and making it tougher for the General Jack the Rippers of the world to order a surprise nuclear attack. Real world data on items like tampers or triggers—and a whole bunch of others we haven't touched on and probably don't know about—is not something you'll find freely on the Internet. There may be some websites and attention-seekers who might try and fool you into thinking they've got it, but how would you really know? Those types of devices have to be desperately developed and tested, which leaves more telltale evidence.
And even in the unlikely event that a nation manages to do all that while under scrutiny from inspectors, a nuclear deterrent isn't much of a deterrent if you keep it a fucking secret. The whole point is for enemies to know, without question, that if they attack you using conventional or nuclear weapons, they will be annihilated by nuclear retaliation. And by the way, that works! There's a reason we went into Afghanistan lock, stock and barrel, but have to move gingerly against the same assholes in Pakistan. One nation has nukes, the other does not, and we know that with great confidence because the Pakistanis were eager for us to know.
**********
Note: To make a hydrogen fusion bomb, multiplying the destructive power by two or more orders of magnitude, takes anywhere from 10 to 20 fission bombs per device, each of which must detonate at precisely the same millisecond to compress and heat the hydrogen to fusion temperatures.