For me, it depends on my specific goal for the NTA run. If I want to generic reference spectra for specific compounds, I run my samples in three different NCE levels: 15, 30, 45 (individually). I’ve not found much of an issue with 30-35 for fragments.
For doing more broad NTA type runs, I run my QE in Stepped NCE levels, with 15, 30, 45 being the steps to get an average spectrum across the three levels.
For PFAS specifically, I’ve found NCE 45 tends to produce 0 – 1 fragments on the low end (m/z 80 – 99). NCE 15 tends to produce only the pseudomolecular ion, and NCE 30 tends to give me a handful of ions with a pretty big pseudomolecular ion.