Knowledge Bank

Theoretically, it should be possible to predict the precision, P(i), for multiple determinations of an experimental peak parameter (i = height, width, and position) by least squares fit to a given peak shape from the signal-to-noise ratio (S/N) and number of data points per line width (K) of a single spectrum [1]: $P(i)=c(i)\cdot S/N\cdot k$ in which c(i) is a line shape-dependent constant [1]. Thus, a sufficiently high digital resolution (i.e., large K) can overcome a poor S/N--see Figure. For Fourier transform spectra, the best precision should thus result from infinite acquisition period $(S/N\to 0)$** For FT/NMR and FT/ion cyclotron resonance mass spectrometry, for which noise is independent of the signal strength, we find that the predicted precision falls short of the experimental value by up to an order of magnitude. Possible rationales and prospects will be discussed. We urge optical spectroscopists to test the precision formula, to see whether or not it holds for their experiments. [This work was supported by N.I.H. GM-31683, NSF CHE-8218998, SCEEE-SRAP/86-29, and The Ohio State University.] Figure 1. FT/ICR mass spectra of $N2+$, for two different acquisition periods, T, in which T = time-domain damping constant. Top: $S/N=12.3$; peak position precision = 247. Bottom: S/N = 63.6; peak position precision = 216.