Computer Assignment of Elemental Compositions of Mass Spectral Peaks from Isotopic Abundances I n Ki Mun, Rengachari Venkataraghavan, and Fred W. McLafferty’ Department of Chemistry, Cornell University, Ithaca, New York 14853
The Probability Based Matching (PBM) algorithm described previously has been extended to predict the number of bromine and chlorine atoms in ions of unknown mass spectra from the relative abundances of isotopic peaks. The theoretical isotopic patterns for 36 Br/Ci combinations are matched against peak “clusters” in the unknown spectrum with a modified PBM algorithm. The confidence values of the best Br/Ci assignments are further reinforced if they are consistent with the m a s differences found for other peak clusters. I n a test using the spectra of 2670 compounds In which 814 contained bromine and chlorine, 90 % of the predictions were correct.
T h e identification of polyhalogenated compounds is of increasing importance for environmental, agricultural, food science, a n d similar problems ( I , 2). Gas chromatography/ mass spectrometry (GC/MS) is often the method of choice for such analyses because of its sensitivity and speed. GC/MS can produce hundreds of spectra per day, so t h a t any automation of t h e interpretation process can be of substantial benefit. T h e presence of chlorine or bromine atoms produces characteristic isotopic patterns in the mass spectrum; isotopic abundance ratios for possible elemental compositions have been extensively tabulated (3-51, and many aspects of methods for calculating these isotopic ratios have been described (3-10). However, these methods all require a knowledge of t h e sample‘s elemental composition, while the identification of unknown mass spectra involves the reverse problem; the peak abundances must be used to derive the elemental compositions of t h e molecular and fragment ions. This is a primary step in mass spectral interpretation, and extensive rules have been formulated t o aid t h e interpreter in this (3-5). This paper describes a computer program for t h e determination of t h e number of chlorine a n d bromine atoms in molecular and fragment peaks in unknown mass spectra which does not require prior knowledge of t h e elemental composition of t h e unknown compound, T h e possible C1-Br assignments found by the program are evaluated according to the probability that the match between the abundances expected theoretically for this composition and the experimental abundances occurred by chance, an extension of the system used for the “Probability Based Matching” (PBM) of unknown mass spectra against a reference file (11). T h e program has been designed t o accept unknown spectra whose precision of abundance measurements varies widely. For mass spectral data of higher accuracy, the error specifications used here could be tightened, which should improve t h e prediction reliability. T h e system should also be applicable to other multi-isotopic elements.
EXPERIMENTAL The presence of chlorine or bromine produces characteristic mass spectral peaks separated by two mass units, due to the abundances of 0.7540 and 0.2460 of 35Cland “Cl, and abundances of 0.5057 and 0.4943 for ‘9Br and “Br. The presence of several of these atoms produces an “isotopic cluster” (5) containing one more peak than the total number of these halogen atoms. The basic approach of the algorithm is to match each cluster of peaks
separated by two mass unit intervals in the unknown mass spectrum against the theoretically expected abundances for each of 36 possible isotopic clusters, calculating a “confidence index”, K, as done in the PBM system (11). The isotopic clusters represent all possible combinations of C1 and Br containing up to 6 of these atoms, plus Br:, Brs, and C1--Cl12. Evaluation of the algorithm utilized a data base of 28 905 different compounds [the majority from the “Registry of Mass Spectral Data” (12)] containing only the elements H, C, N, 0 , F, Si, P, S, C1, Br, and I; of these compounds, only ten contained Cl/Br combinations not included in the 36 of this list. A second routine increases the K value of a higher mass cluster assignment if its mass separation from a lower mass cluster is consistent with the differential Cl/Br assignment. The program of approximately 600 statements was written in FORTRAK Iv on a DEC PDP-11/45, on which an average determination requires less than 3 s. A relatively short main program controls input and output and calls other routines. One subroutine generates the theoretical relative abundances ( A T ) of the 36 isotopic clusters utilizing Equation 1
AT(P+ 2 x ) = a(Br
-
x
c“ i=o
Br! c1! (Br - x + i)!(x- i)! (CI - i ) ! i !
+ i ) b ( x - i ) P“’
- i, 4
(1)
(31,where Br is the number of bromine atoms, C1 is the number of chlorine atoms, P is the mass of the peak containing only 35Cl and 7gBr(the“principal peak), x is an integer (0, 1, 2, ...) S ( B r + C l ) , i is an integersx, SCl, and S x - B r . The terms a, b, p , q , are 0.5057, 0.4943, 0.7540, and 0.2460, respectively, the relative occurrence probabilities of ?Br, slBr, 35Cl,and 37Cl. The resulting 36 cluster patterns can either be stored or generated for each use of the program. PBM Calculation. The program inspects the unknown mass spectrum starting a t the high mass limit to identify peak “clusters”. A cluster is defined as a continuous series of two or more non-zero abundance peaks, each separated from the next by two mass units, with the series starting a t a minimum abundance value, going through a maximum and ending with another minimum value. MHC’ (“mass highest unknown”) is the m / e value of the most abundant peak in the cluster (if there are two or more such peaks of identical abundances, the one of lowest m / e is used) and M H T (“mass highest theoretical”) is this value for the theoretical isotope abundance pattern under consideration. The theoretical pattern for Br number of bromine atoms and Cl number of chlorine atoms is designated as (Br,C l ) ;the principal peak P should be the lowest mass peak in the cluster. Clusters whose highest peak (MHU) f& within the ranges m / e