In my discussion of "the nuclear reactor of the future", below, I realize tardily that there was another issue I missed mentioning, so I'll mention it now.
There are basically two ways to build an atomic bomb: an implosion bomb with Pu-239, or a gun bomb with U-235. There are other designs and I expect there are other fissionable metals which could be used, but for the most part they don't seem to be worth the effort since uranium and plutonium work so well, and are commonly available.
I've discussed, several times, the engineering problems of constructing an implosion bomb. Implosion is finicky and hard to get right, and a "garage bomb" would be of dubious reliability, more likely to "fizzle" or just spread Pu-239 over a few blocks than to detonate properly.
What about a gun bomb?
Uranium does not lend itself to implosion devices; it doesn't have the right properties. It does not, however, need those properties, becuase a supercritical mass can be assembled simply by firing a slug into a hole in a block of the stuff.
The block sits at the end of a barrel; the slug is at the other end, in front of some high explosive (HE). At the appropriate time, the HE is detonated, driving the slug into the block. A supercritical mass is assembled, a nuclear reaction takes place, and 30 nanoseconds later you have a full-blown nuclear detonation.
As far as engineering is concerned, most of the problems have already been solved. At its base, the gun bomb is essentially a smoothbore cannon. The HE needs to be some fast-acting stuff but it need not be even as exotic as the combination of shaped explosives needed for implosion. I bet C-4 or Semtex would be more than adequate.
There are only two problems to be solved with a gun bomb. The first is that U-235 comprises about 0.7% of all uranium; the second is that a significant mass is required to contain it long enough for the nuclear reaction to complete.
The former problem is the biggest one. Isotope separation is tricky stuff, because all isotopes have the same chemical properties. (For example, we can carbon date some things because carbon-14 is chemically the same as carbon-12. Your body has all kinds of carbon-14 in it; you're radioactive!) You can't just distill uranium the way you can, say, whiskey.
To make weapons-grade U-235, the United States basically built Oak Ridge, Tennesee, as part of tha Manhattan Project. And the facility at Oak Ridge was, at the time, the largest industrial complex ever built in the United States. It consumed a massive amount of power--it was put in Tennesee because of its proximity to the big new hydroelectric dam built by the Tennesee Valley Authority, as part of the New Deal--and required an army of workers. And it took months to concentrate enough U-235 to build one bomb.
Weapons-grade U-235 is at least 90% U-235. That means that the facility had to process 142 pounds of uranium to extract one pound of U-235...and it takes on the order of 25 lbs of U-235 to make a single bomb. So after processing about a ton of uranium, you have enough U-235 for one bomb.
Assuming that you have managed, somehow, to acquire weapons-grade uranium from somewhere, without having to build your own processing facility, then what?
I once came across a web site which described all the problems a terrorist would have with building a gun bomb; and the description of each stage of the effort ended by saying that it would probably kill you. It's a dangerous exercise and not easy to accomplish on a shoestring budget. (About the same as with making an implosion bomb, really.) I wish I had the URL for that site. I can't even remember how I got to it and it doesn't look as if I bookmarked it. Bad me.
U-235 is a beta-emitter: it emits high-speed electrons. U-238 is an alpha emitter; it emits helium nuclei. The latter you can stop with a single sheet of 20 lb bond paper or a good coat of latex paint. The former will only be stopped by an inch of wood or other similar material. Don't get too cozy with your U-235. (And, by the way, you also have to worry about the radiation from the decay products, which could be anything from alphas to gammas. So you'd better use a lead box.)
So if you've got your fissionable and you want to build a gun bomb in your house in order to teach the infidels a lesson, what do you do?
You build a massive cannon with its breech in the attic and the muzzle in the basement. You put most of your U-235 at the muzzle, precisely aligned, and then fill your basement with concrete. You put the slug up at the breech, held in place with a thin sheetmetal barrier (it'll easily punch through it when the explosives are detonated) and wired up with a good timer. Then you set the timer and haul ass out of there. If you are smart and lucky, you have only received a big enough radiation dose to lower your life expectancy by, oh, 30 years. You'll probably die of cancer, but that'll give you plenty of time to enjoy the devastation your bomb wreaks on America.
Assuming that you did everything just right, your bomb yields, oh, about 20 kilotons. And since it's a ground burst, you've effectively wasted about 40% of the bomb's energy on making a big crater. You get more fallout that way, but the area of actual nuclear devastation is rather limited. Even if your house was atop the highest hill for miles around.
And what if you didn't do everything just right?
The failure modes are rather numerous, but I'm going to concentrate on the ones which you can't possibly test for.
The slug could lodge in the barrel. You might rifle your barrel to prevent this, but I'm guessing that you don't want anyone to see you rifling a 20-foot-long piece of pipe. They might ask why, and you can't afford questions from anyone. You could make your slug a "cannonball" instead; that would help, but it might adversely effect the efficiency of your bomb, lessening your yield. (Gun bombs normally use a cylindrical slug, as far as I know.)
If the slug lodges in the barrel, you probably don't get any real detonation. Your house blows up and it scatters some radioactive material over your block, but it's not really the disaster you were hoping for.
The barrel has to be long, in order to insure that the slug has enough velocity when it arrives on target. If the slug is moving too slowly, it'll vaporize before a supercritical mass is assembled, and your bomb will fizzle at best. Remember, you only have about 30 nanoseconds before the critical mass gets too hot to stay together. (You filled your basement with concrete in order to keep the critical mass together as long as possible. If you can get enough U-238 to pack around the U-235, that will help, too.) Besides, a long barrel keeps the subcritical masses separate, so that you don't get fried by radiation while assembling the stupid thing.
The barrel has to be able to contain the pressure of the detonating explosive, at least long enough for the slug to reach the target. This means you can't just go buy some iron pipe from Menard's, and it's probable that someone would notice it if you obtain a cannon or something and use it. So you have a bit of trouble with how you're going to build the barrel in the first place.
Also, it takes a lot of concrete to fill a basement. You probably won't be mixing it yourself; and if you do--again--your neighbors will probably wonder why you keep carrying 50-lb sacks of concrete into your house. It will take a lot of them. If you back a cement mixer up to the house and have 'em start pumping, well.... You could make a form around the target and make a smaller tamper, but you'd have to be very careful--you want as much mass there as possible; too little and the whole exercise is for naught, because then all you've done is find an expensive way to blow up a few houses and scatter radioactive material over a block or two.
All in all, this is why we don't worry about terrorists getting their hands on U-235. For all the problems there are with making a functional "garage bomb" with plutonium, there are just as many problems with making a functional "house bomb" with U-235.
You really need four things to make a functional weapon at home: you need to be smart, you need to be educated (and in the right things), you need to be lucky, and you need to be rich.
Smart: it will take a lot of brains to do. Not only do you have to understand the physics, but you also have to be able to assemble the materials without alerting the authorities.
Educated: to borrow a phrase, making an atomic device ain't like dusting crops. Although it is merely engineering, it is precision engineering, and lower precision means lower performance. Below a certain level you have a radiological device rather than an atomic device. You must know how to assemble a critical mass in a very narrow time frame, and you must know how to figure out what you need. This means "education" and it means education in hard science, not Marxism or the Koran.
Lucky: even if you meet the first two criteria--and they are the hardest to manage, since it's virtually impossible to arrange for yourself to be born smart and to parents who can afford to educate you--you will need luck. (And not just the luck you've already used to be born smart, and to receive a college education that allows you to build atomic weapons in your home.) You will need luck because people will notice what you are doing; some might see into your garage and wonder why on Earth you filled it with machine tools encased in ventilated plexiglas boxes. Depending on how long you're there, they may wonder why you live in such a nice house yet never seem to go out, except to buy groceries and things from Ace Hardware or Harbor Freight? A million things could go wrong; it takes time to build a bomb and even if your operational security is perfect, something could still go wrong. Your neighbor's kid borrows a geiger counter from school and finds out your house is more radioactive than its environment. Your other neighbor is a photography nut and sees that all the photos he took after you moved in have bright speckles all over them. Although the grass is mowed regularly and the paper is taken in every day, you are rarely seen outside (except for shopping trips). Or whatever.
Rich: you can't do it on a shoestring. Assuming that someone gives you the fissionables, because your politics are correct, you still have to make a precision device, and that takes time and money, because you need parts and tools that have the correct precision. If you must work a job while engaged in this endeavor, it affords an additional avenue for discovery and means it will take you that much longer to assemble your bomb.
To put it all into perspective, then, let me give the example of the person who is engaged in a legal hobby: high power rocketry. People routinely spend thousands of dollars building a rocket which can only fly at certain special events, and they spend thousands more on the rocket motors required to boost them skyward. These machines take months to build, days to prepare for flight, and frequently fail to perform as expected. And the people who build them are neither stupid nor inept, but careful people who are engineers and mechanics and other things in their daily lives.
And successfully detonating an atomic bomb is harder than launching a rocket.