The kinds of isotopes you need for a nuclear chain reaction are called _fissile_ isotopes. They are isotopes that have just the right nuclear properties so that they will split when they absorb a neutron of basically any energy level, including the energy level of neutrons that are released by fission. Which is always a little tricky to explain, but what you’re basically asking for is an atom that can be split by the same neutrons that are produced by splitting atoms. If you have that, then you can have the neutrons from one atom split another atom, and so on and so on.
There are isotopes that are _fissionable_ but not _fissile_. U-238, for example, can fission, but only from very high-energy neutrons, higher-energy than the ones typically released by a fission reaction. So the odds are that if they absorb a neutron created by another fissioning atom, they won’t split, they just absorb it. Even if they do split, their neutron has a very low chance of splitting another U-238 atom. So the presence of U-238 — or any other non-fissile isotopes that absorb neutrons — will kill a chain reaction.
There are some atoms, typically very light ones, that will essentially bounce neutrons off of them, which lowers the energy of the neutrons. These are used in nuclear reactors, because decreasing the energy of a neutron increases the chance (for quantum mechanical reasons) of it splitting a fissile atom.
The exact conditions for a nuclear chain reaction are complicated and can be adjusted by a lot of variables. The long and short of it is that you need the atoms that can be split by neutrons to be near other atoms that can be split by neutrons, and you need to overcome the many ways in which a neutron creating by that splitting won’t split another splittable atom (e.g., captured by a non-splitting atom, lost to the environment, etc.). There are a few ways to do this for the purposes of nuclear weapons, and many ways to do this for the purposes of nuclear reactors.
Latest Answers