2.1. OPERATIONAL FINGERPRINTING; A NEW CONCEPT
The classical applications of fingerprinting techniques are classification and identification of the original material, using a library of reference fingerprints. Generally in these applications, little or no chemical interpretation of the pattern is attempted or even required. Computer evaluation of pyrolysis fingerprints (pyrograms) in combination with the greatly increased speed of analysis afforded by modern Py-MS techniques has opened up important new areas of application, namely screening, quality control and process monitoring (ref. 48, 49, 63). These procedures can be regarded as "operational fingerprinting" techniques in that no library of reference spectra is required but reference samples are usually present in the batch of samples submitted for analysis. In most applications of this type, a single class of samples dominates (e.g. pure samples) and only a minor fraction of the samples belongs to one or more different classes (e.g. contaminated samples).
In our experience, operational fingerprinting automatically leads to the question: what is the biochemical nature of the observed variation or changes which make the outliers different from the central cluster. Thus, as shown in several of the above cited literature references (ref. 49, 50, 52, 53), Py-MS techniques are increasingly employed to address directly problems concerning the biochemical nature, composition and structure of the sample. The success of attempted biochemical interpretation of interesting features in pyrolysis mass spectra, whether distinguished by eye or with the aid of numerical computer techniques, depends critically on the following factors: complexity of the sample, additivity of component spectra, availability of reference spectra from standard materials, knowledge of relevant pyrolysis mechanisms and availability of ancillary analytical methods. These factors will be discussed in the following paragraphs.
Obviously a multicomponent sample generally provides a spectrum more difficult to interpret than that of a sample consisting of a single, pure component. However, with multicomponent samples - even extremely complex samples such as whole cells - often only one or two components will suffice to adequately describe the analytical problem. Therefore, if suitable control samples are available, subtraction of patterns may yield a much simpler pattern, mainly representative of the component(s) of interest.
Even when comparing multiple samples, e.g., bacterial strains differing in more than one component, factor analysis techniques may b^ called upon to reveal relatively simple factor spectra representative of the various components involved (see Chapter 6). These techniques can only be used if the component spectra are additive which, in turn, requires that the respective pyrolysis pathways are mutually non-interfering.
Only pyrolysis techniques which minimise the occurrence of bimolecular reactions and secondary reactions, e.g. recombinations between pyrolysis products, can be expected to reasonably fulfill the condition of additivity. In studies by Posthumus and Nibbering (ref. 58) the Curie-point oven pyrolysis of microgram amounts of amino acids did not cause appreciable recombination reactions. This was shown by the absence of diketopiperazine formation, a reaction that readily occurs during classical oven pyrolysis (ref. 64). In the case of 4-phenylbutanoic acid (and not for other w-phenylalkanoic acids) a recombination reaction, i.e. the formation of dibenzyl, could be observed (ref. 57). However, this reaction could be eliminated by further reducing the amount of sample to submicrogram levels (ref. 56).
When using Curie-point filament pyrolysis, additivity of spectra is readily observed for such classes of compounds as polysaccharides (refs. 65, 66) and lignins (ref. 67) and, less well established, also for lipids and proteins (ref. 68). The additivity of "component subspectra" is an important requirement for (semi)quantita-tive applications of Py-MS, e.g. in monitoring ppm concentrations of the toxic, technical polymer DEAE-dextran in poliomyelitis virus vaccine preparations (ref. 63).
The admixture of inorganic salts, and also changes in pH of the sample solutions or suspensions markedly influence the pyrolysate patterns of polar organic compounds, even if no direct organic salts can be formed. This phenomenon, further discussed in Section 4.1, and in some of the Atlas spectra, should be kept in mind when evaluating spectra of complex samples for the presence of a particular organic component. Highly reactive organic groups have been found to react with inorganic components, e.g. the methyl groups of the trimethylammonium function of choline residues or ester methyl groups (see Atlas) can react with chloride ions to form methylchloride. In fact, the latter reaction is quantitative and can be used to measure the concentration of acetylcholine in brain tissue, as described by Szilagyi et at. (ref. 69). Similar formation of methylchloride can be observed during pyrolysis of choline-containing phospholipids (ref. 70).
Until now, reference spectra have not been readily available from literature sources since the Py-MS techniques used vary too widely to allow close comparison of specta. However, from building four Curie-point Py-MS systems over the past ten years, we know that reproducibility between instruments of the same basic design is remarkably good. Moreover, as demonstrated in Chapter 5, even a definite degree of inter-laboratory reproducibility is readily achieved with comparable instruments. This has encouraged the F.O.M. Pyrolysis Centre to compile a collection of several hundred pyrolysis mass spectra of natural and synthetic compounds. A selection of these spectra is included in Part II of this volume. The purpose of this Atlas is primarily to enable the reader to evaluate the applicability of Curie-point Py-MS to different problems in the analysis of biomaterials, geopolymers, humic compounds and, to a limited extent, drugs and technical polymers. At the same time, this collection of Py-MS reference spectra enables scientists working with comparable Py-MS techniques to evaluate the degree of inter-laboratory reproducibility achievable by appropriate "tuning" of their instruments. However, as was pointed out in the Preface, this collection of spectra should primarily enable a qualitative comparison of pyrolysis mass spectra. The establishment of a library for quantitative comparisons between spectra from different instruments, e.g. for fine differentiation between bacterial strains, is beyond the present state of the art.
It should further be noted that extensive chemical interpretation of the various mass peaks in the Atlas spectra was not possible; only tentative assignments could be made on the basis of comparison with results of other techniques such as Py-HRMS (refs. 53, 71, 72) or Py-GC-MS (refs. 73, 74). The use of such data from the literature may give incorrect results, however, as pyrolysis and sample transfer conditions for different techniques are often very different. Therefore, tandem mass spectrometer systems for collisional induced dissociation (refs. 75-81) should prove invaluable for establishing the chemical identity of peaks observed in Py-MS, since this approach allows precise duplication of pyrolysis and sample transfer conditions normally used in Py-MS (see Section 2.7).
Regardless of the size of any library of pyrolysis mass spectra, the variety of biomaterials that can be investigated is so immense that often pyrolysis patterns not represented in the library will be generated. Even in these cases, a limited amount of chemical interpretation of such patterns may still be possible on the basis of insight into - or empirical knowledge of - relevant pyrolysis mechanisms (see Section 2.6.).
Was this article helpful?