Ok, let’s clear up some things very quickly before people get the wrong idea about carbon fiber.
Carbon fibers very brittle
First off, carbon fiber is indeed brittle at the atomic level. It has an amorphous crystalline structure that fails spectacularly when sheared. You can think of carbon fiber as wanna-be graphite where its planes are unable to slip past each other (the reason graphite is a great lubricant!). If you line up those planes properly you end up with a really strong fiber in the axis in which the majority of the planes line up. (fibers can have an elastic modulus of hundreds of GPA along the fiber axis with tens of GPa perpendicular to the axis)
However, that does not mean that carbon fiber is not flexible at long length scales and small diameters. You can take a mat of woven carbon fiber and crumple it up just like a sheet of paper without the mat breaking into a million pieces. This is a consequence of making the fibers small in diameter. As the diameter of the fiber decreases its elastic modulus, strength, and flexibility all increase. This increase in strength is due to the decrease in cross-sectional area limiting the number of places for a large flaw to propagate through the fiber. Another important thing is that the fibers be long (called critical length)
Now another cool property of carbon fiber is that each individual fiber is independent of its neighbors. This is very different, from say steel, where the crack is propagating through a single crystal lattice and after the initial crack has formed it is easily propagated through the rest of the crystal. You must break every single carbon fiber individually. Breaking one in no way affects the strength of the next fiber in your weave. It turns out that the matrix, see below, also helps stop crack propogation.
Once you have these nice skinny fibers you can place them in a very brittle matrix, say an epoxy, and increase the strength of the material. The epoxy is only there to transfer the load to the fibers and to keep them from bending out of axis. If you took a small piece of solid epoxy, you could break it very easily. Imagine a 1â€x12â€x1/4†piece of epoxy. Grab the ends, bend it (tensile and compression being applies) and it will break. However, if you place fibers in it that are longer than the critical fiber length, they will carry the load being applied and keep the epoxy matrix from breaking. To give you an idea, you could not break a piece of carbon fiber of those dimensions, if properly engineered, easily. Stand on it, bend it like a coat hanger, it won't break.
It seems like it would crack into a zillion pieces in a fabric like Kevlar unless it wasn't in Kevlar but bonded in the separate ceramic trauma plates.
As I pointed out, a mat of carbon fiber will not break in that fashion. Due to the strength of the carbon fiber matrix, it takes quite a bit of energy to destroy it. When you see an F1 car hit the wall and it breaks into a zillion pieces, you are seeing an engineer smile. Those cars are meant to explode like that because it dissipates huge amounts of energy. If only Nascar would pay attention!
This is also one of the reasons that you place a ceramic plate in a bulletproof vest. When it breaks it dissipates a large amount of energy by propagating cracks through the material and not into your body. You absolutely want the material to crack because it is getting rid of energy that would otherwise be used to penetrate your body. Now there is a fine line of how much cracking that you want. Too brittle and the plate crumbles and subsequent shots go right through. There is a tradeoff there that is very well thought out and it is all about energy transfer.
One of the big things Kevlar has going for it is cut resistance. I won't go into depth because I've already made this too long. Bascially, it is really hard to cut Kevlar and that makes it very attractive for vests. It is worth noting that Kevlar is DuPonts brandname for an aramid fiber, specifically a polyamide. There are other aramids that would be better suited than Kevlar but the industry doesn't seem to want to use them. Kevlar can also be spun into an incredibly tight weave of fibers without breaking the chains (i.e. reducing critical length in the case of carbon fiber).
However, as Shock pointed out nanotubes could change that. Nanotubes are pretty cool, except they are very difficult to make in long lengths. It was just recently that we have been able to make a nanotubes of considerable length (microns of length was considered incredible a few years ago). They have this horrible propensity for sticking together due to surface energy and they are usually put into a solution of soaps to keep them from sticking. The solutions have a very low weight of nanotubes for a given volume and that makes them very unattractive for commercial production methods that are employed for fibers like Kevlar. All of that is changing though!
Anyhow, just thought I’d throw out a little info there… If I missed something, please feel free to correct me.