In principle yes, in practice it may be difficult. the first thing that occurs to be is to use a soluble soluble paramagnetic shift reagent like a gadolinium chelate. Surface Li would be quenched while bulk would not. This method (and all others I can think of) depends on having a high enough surface to bulk ratio that the surface Li can can be detected independently, even if it cannot be resolved from the bulk signal.
Could you give an approximation of necessary surface-to-bulk ratio?
The case that you described using a paramagnetic shift reagent requires ex-situ sample preparation.
The most interesting for me would be to do an in-situ/ operando Li7-NMR. In this case however one has to be aware that Li is not only stored in bulk and surface of the insertion material but is also present in electrolyte and SEI(LiF deposits).
It seems surface/bulk discrimination has been done before, using the fast relaxation of the bulk Li relative to the surface by the use of a long relaxation delay to remove the bulk signal in situ