Identification spectral libraries

S. E. Stein, D. R. Scott,/. Am. Soc. Mass Spectrom. 1994, 5, 856-866. Optimization and testing of mass spectral library seardi algorithms for compound identification. 

The spray paint can was inverted and a small amount of product was dispensed into a 20 mL glass headspace vial. The vial was immediately sealed and was incubated at 80°C for approximately 30 min. After this isothermal hold, a 0.5-mL portion of the headspace was injected into the GC/MS system. The GC-MS total ion chromatogram of the paint solvent mixture headspace is shown in Figure 15. Numerous solvent peaks were detected and identified via mass spectral library searching. The retention times, approximate percentages, and tentative identifications are shown in Table 8 for the solvent peaks. These peak identifications are considered tentative, as they are based solely on the library search. The mass spectral library search is often unable to differentiate with a high degree of confidence between positional isomers of branched aliphatic hydrocarbons or cycloaliphatic hydrocarbons. Therefore, the peak identifications in Table 8 may not be correct in all cases as to the exact isomer present (e.g., 1,2,3-cyclohexane versus 1,2,4-cyclohexane). However, the class of compound (cyclic versus branched versus linear aliphatic) and the total number of carbon atoms in the molecule should be correct for the majority of peaks. 

Mass Spectral Library[35] and the Wiley Registry of Mass Spectral Data[36] are now available to help mass spectrometrists in the identification of unknowns. 

The successful use of VIRS methods as research tools requires more than anything a published framework that will provide standards for the collection and interpretation of data, and accessible data libraries that can give examples for comparison and contrast. The spectral libraries used for mineral identification are in part public-domain, but much information remains tied to the instrument software and cannot readily be distributed. Differences in the formats used for spectral files in different libraries complicate their use in standard software and impedes information exchange. 

The El source has been the most widely used ion source over the past 60 years and continues to be the method of choice for the analysis (either qualitative or quantitative) of small- to medium-sized volatile organic compounds. The inherent reproducibility of the mass spectra has enabled the assembly of large spectral libraries. Computers associated with current generation instruments can efficiently (in a few seconds) search an unknown mass spectrum against tens of thousands of reference spectra in order to aid in the identification of an analyte. The general scheme of an El source includes the introduction of the vaporized analyte molecules into the ionization chamber, exposure of those molecules... 

Fig. 11.4. Electron ionization mass spectrum of nonanal. Unlike the previous example (toluene, Fig. 11.3), this 9-carbon alkyl aldehyde displays extensive fragmentation and a very low abundance molecular ion at mlz 142. The extensive degree of fragmentation exhibited by many compounds under El conditions makes manual interpretation complex and tedious. Consequently, computerized searches of spectral libraries find extensive use in compound identification.

Spectral similarity search is a routine method for identification of compounds, and is similar to fc-NN classification. For molecular spectra (IR, MS, NMR), more complicated, problem-specific similarity measures are used than criteria based on the Euclidean distance (Davies 2003 Robien 2003 Thiele and Salzer 2003). If the unknown is contained in the used data base (spectral library), identification is often possible for compounds not present in the data base, k-NN classification may give hints to which compound classes the unknown belongs. 

Identification involves the confirmation of a certain chemical entity from its spectrum by matching against the components of a spectral library using an appropriate measure of similarity such as the correlation coefficient, also known as the spectral match value (SMV). SMV is the cosine of the angle formed by the vectors of the spectram for the sample and the average spectrum for each product included in the library. 

In theory, if the product spectrum coincides with one in the library, then its correlation coefficient should be unity. However, the random noise associated with all spectral measurements precludes an exact match. The SMV has the advantage that it is independent of library size, which facilitates the building of libraries containing large numbers of raw materials as well as correct identifications with libraries consisting of a few spectra for a single product. 

GC/MS was the primary tool for identifying the first DBFs, and it remains an important tool for measuring and identifying new DBFs. Large mass spectral libraries (NIST and Wiley databases, which contain >200,000 spectra) enable rapid identifications. When DBFs are not present in these databases, high-resolution... 

Useful interface which Is applicable to a wide range of molecules. The volatile solvent molecules are stripped from the sample and lost in a process similar to that used in the early jet separators used in GC-MS. The heavier sample molecules enter the MS and can be ionised by the standard methods of El, PICI or NICI. Gives spectra with El fragmentation which can be referred for identification to El spectral libraries built up over many years. No solvent background thus sensitive to the 1(h g level. Solvent flow rate up to 1 ml/min, mass range up to 1000 amu... 

FIGURE 15 GC-MS chromatograms of the static ethanol extracts of Santoprene tubing materials (underivatized). The chromatograms from these two Santoprene materials were quite different from those of the silicone materials (Figure 14). IS = internal standard (dimethyl phthalate). See Table 39 for the tentative peak identifications. Only those peaks with recorded spectral library matches are noted for each sample, although retention times and patterns may suggest some additional peak identifications [78]. 

Spectral searches using a library of reference spectra can be a useful tool in identification. Search algorithms have improved over the years and now use the concept of artificial intelligence. Several software packages can be used to conduct searches in spectral libraries in which the main peaks of known compounds are encoded. The compounds offering the best matches are retained as potential candidates. Library searches involve three stages ... 

The study of fragmentation processes has led to semi-empirical rules used for compound identification. Interpretation from first principles is also used to validate results obtained by spectral library searches. 

---