is it possible to use Mercurial version control to track Word or PDF files? Is there any limitation or problem?
Yes, but of course you won't be able to diff in any meaningful way. The files will therefore be treated as binary during merges.
Mercurial is perfectly capable of tracking binary files:
Mercurial generally makes no assumptions about file contents. Thus, most things in Mercurial work fine with any type of file.
Mercurial stores a binary diff regardless of the file type. The problem with PDF/Word files is that a little change to them usually causes a huge difference in their binary representation on disk. .docx Documents are stored as a zipped xml, due to the zipping a single flipped bit inside the archive can cause the zip archive to look completely different.
If you don't grow your repository too large, you probably won't experience any issues using Mercurial.