I have made Fourier maps in G-Fourier program built in Fullprof Suite and based on the Rietveld refinement of powder pattern. I have Boron-rich B13C2 compound. Refinement accepts a new position partially occupied by Boron, however, Fourier maps don't show any electron density on these positions. Instead, a negative electron density only. Although, Rietveld refinement is very stable and R-value also improves with this additional boron position. Should I take this new Boron position into account or not?