I'm not sure how much this question strongly depends on Physics/UK, so let me detail how it is done for Computer Science in France (as far as I know, the French and British systems are quite close on this, and hard sciences tend to have a common academic process).
The PhD thesis is indeed a novel contribution to the state of the art, and demonstrates the ability of the candidate to be an autonomous full-time researcher. In many fields, it's usually required to have several publications before submitting the PhD thesis.
On the other hand, the Master thesis corresponds to an research internship done in a lab (usually the one in which the candidate will pursue her PhD). The role of the Master thesis is two-fold:
- First, it demonstrates that the candidate has acquired some techniques taught during the Master, and that she was able to use them to solve a particular research problem.
- Second, it's a first step for the candidate into the world of research. Often, the Master thesis constitutes the first chapter of the thesis.
(note that I'm talking here only about research-focused Masters, as there are also industry focused Masters, for which the thesis has a different role).
As such, the topic of the thesis is usually a "simple" research problem, defined by the advisor based on the interests of the student. By simple, I mean that even though the exact solution might not be known, the advisor is usually confident that the solution can be found in a reasonable amount of time.
Note that in France, the time constraint is quite strong, and the thesis must be written in 6 months maximum, otherwise the Master is not validated at all, and therefore the topic of the thesis is usually not an open and complex research problem.