LEAST-SQUARES ANALYSIS OF CONGESTED SPECTRA

1. J. Xu, M. H. Mendenhall, and J. Tellinghuisen, J. Chem. Phys. 93 , 5281 (1990). 2. X. Zheng, S. Fei, M. C. Heaven, and J. Tellinghuisen, J. Mol. Spectrosc. 149 , 399 (1991).Author Institution: Department of Chemistry, Vanderbilt UniversityIn many situations nature and instrumental limitations conspire to present the spectroscopist with spectra that defy resolution into individual assignable transitions. Nonetheless, such spectra may frequently be recorded as rather large sets of precise data---for example from array deteetors or from computer-controlled scanning experiments. Their information content can be very high, even though their analysis by conventional ``measure, assign, and fit” procedures may be next to impossible. Such cases have normally been handled by trial-and-error methods of band contour analysis. In recent experience with a range of congested spectra problems, I have found that band contour analysis can be accomplished within the framework of fairly standard nonlinear least-squares fitting procedures, often yielding remarkably precise estimates of unknown parameters governing the spectra. Examples to be discussed include cases where the spectroscopic parameters are known and the spectra are used to extract information about population distributions (CN $B\rightarrow X$ and OH A $\rightarrow$ X emission following electron-stimulated desorption from alkali $halides^{1}$), and cases where both the population distributions and the spectroscopic parameters are unknown (laser-excitation spectra of diatomic halogens in a free-jet $expansion^{2}$). These methods are currently being extended to densitometer traces of photographic spectra