Abstract
Targeted liquid chromatography–tandem mass spectrometry quantifies only a few dozen fluorinated analytes and requires an authentic standard for each, so most of the organofluorine in environmental extracts is never assigned to a structure. Fluorine-19 NMR is an attractive complement because every fluorine nucleus responds independently of ionization and in proportion to its abundance, but existing
19
F libraries record chemical shift alone, in inconsistent solvents, and without the statistical search tools needed for 25 identification. Here we report a
19
F NMR spectral library of 163 fluorinated compounds—PFAS, pesticides, pharmaceuticals, and other fluorinated species—measured under three defined solvent conditions and annotated at the level of the individual fluorine nucleus with chemical shift, scalar coupling, multiplicity, and transverse relaxation time. We introduce the solvent-induced shift vector (SISV), the change in chemical shift of a resonance between solvent systems, as a descriptor that discriminates local fluorine environments that a single-solvent shift cannot resolve. Annotated resonances are mapped onto HOSE substructure codes and searched with a Bayesian framework that reports a likelihood ratio for each candidate structure together with an explicit ambiguity count, so that evidence strength and identification specificity are reported separately. Validation against synthetic mixtures establishes which descriptors contribute discriminating information and where the spectrum supports only a substructure class; application to an industrial wastewater extract demonstrates candidate prioritization under realistic, nonideal matrix conditions.