Sometimes the thorium, when it absorbs a neutron, will emit two neutrons and turn into thorium-231, which decays (though a couple more isotopes) into U-232. It can also be formed when U-233 absorbs a neutron and emits two, forming U-232.

How hard this would be to separate depends on the type of LFTR design you're using. However, you could also include some thorium-230 in the fuel mix to denature any protactinium produced, so there's really no way around having U-232 mixed with the bred U-233. Here's a blog post with more details on the entire process:

http://energyfromthorium.com/2006/10/06/denaturing-thorium-w...