Instrumentation Ε. L. Wehry and Gleb Mamantov Department of Chemistry University of Tennessee Knoxville, Tenn. 37916
Matrix Isolation Spectroscopy A problem area of great contempo rary significance in analytical chemis try is the development of techniques for identification and reliable quanti tation of individual trace organic com pounds in complex samples. T h e de m a n d s imposed upon analytical meth odology by many present-day samples are staggering. For example, a "syn thetic fuel" sample, such as a shale oil or coal liquid, may contain h u n d r e d s of different organic compounds, some of which may be carcinogenic. A real istic analytical protocol for dealing with a sample of such complexity logi cally begins with the application of modern separation and fractionation procedures. Yet separation methods, however powerful, can be time con suming, particularly if it is d e m a n d e d t h a t the sample be fractionated into individual pure compounds. Even very efficient separation procedures may be unable to do more (in a realistic time) t h a n produce a series of frac tions which themselves are rather complex mixtures. Hence, even if it is coupled with separation procedures, the ultimate analytical measurement step should be capable of producing reliable qualitative and quantitative results in mixtures of considerable complexity. Optical spectroscopy (UV-visible absorption or luminescence and IR absorption) has been applied widely to 0003-2700/79/0351-643A$01.00/0 © 1979 American Chemical Society
the analysis of organic samples. Sub stantial improvements in the perfor mance of spectroscopic instrumenta tion (particularly for infrared absorp tion and molecular fluorescence) have recently been made, b u t much less at tention has been devoted to improve ments in the quality of the spectro scopic sample itself. T h u s , for exam ple, powerful Fourier transform in frared (FTIR) spectrometers, which may cost in excess of $100 000 and may be equipped with the latest in op tical and computer hardware, often are used in conjunction with IR sam ple preparation techniques which have changed little in the past 30 years and which tend to negate many of the ad vantages offered by the new instru mentation. An " i d e a l " sampling matrix for spectroscopic analysis of a multicomp o n e n t sample should possess the fol lowing characteristics: (a) For qualitative analysis, the spectrum of any particular compound should be a characteristic, reproduc ible "fingerprint" of t h a t compound. T h e spectrum should contain suffi cient information to permit unambig uous and relatively facile identifica tion of the compound, yet be relatively simple (so t h a t coincidences in the lo cation of features in the spectra of dif ferent compounds occur infrequently). (b) For both qualitative and quanti
tative analytical purposes, it is desir able t h a t both the appearance of the spectrum (i.e., A m a x values) and the "quantitative c o n s t a n t s " (e.g., e m a x or fluorescence q u a n t u m efficiency) for a given compound be totally indepen d e n t of the composition of the sample in which t h a t compound is encoun tered. This criterion necessitates t h a t the molecular environment " s e e n " by the analyte in the spectroscopic sam ple be always the same, irrespective of the nature of the original sample. (c) Finally, the optical properties of the spectroscopic medium must be as nearly "perfect" as possible. For ab sorption measurements, the medium should be t r a n s p a r e n t and weakly scattering [so that, for example, the "multiplex" and " t h r o u g h p u t " advan tages (1 ) of F T I R spectrometers can be exploited fully]. For luminescence measurements, not only should a ma trix be t r a n s p a r e n t and weakly scat tering, b u t it should be free of lumi nescent contaminants, the presence of which frequently determines the de tection limits in fluorescence or phos phorescence spectrometry when the most modern instrumentation is used (2). A truly "ideal" spectroscopic sam pling matrix probably does not exist. However, it is particularly striking t h a t the matrices conventionally used in analytical molecular spectroscopy
ANALYTICAL CHEMISTRY, VOL. 5 1 , NO. 6, MAY 1979 · 643 A
(liquid solutions, K B r discs, frozen so lutions, mulls, etc.) fall far short of ideal behavior. In contemplating this problem, we were attracted to t h e technique of matrix isolation, original ly developed (3, 4) in the mid-1950's as a technique for obtaining optical or electron spin resonance spectra of transient species (5-8). Since t h a t time, a few scattered reports, includ ing one in this JOURNAL (9), implied considerable analytical promise for matrix isolation. Nonetheless, t h e use of matrix isolation as a practical sam pling procedure for spectroscopic analysis of stable molecular species has been virtually ignored until re cently. Matrix Isolation Technique In matrix isolation (MI), sample constituents (if solids or liquids a t room t e m p e r a t u r e ) are vaporized and then mixed thoroughly with a large excess (typically a factor of 10 4 -10 8 on a mole basis) of a diluent gas. In prin ciple, any gas may be used as the dilu ent, provided t h a t it is chemically unreactive and does not absorb in the wavelength region of interest; we have found nitrogen to be a suitable matrix gas for general-purpose spectroscopic measurements. T h e resulting gaseous mixture is then deposited a t cryogenic t e m p e r a t u r e s on a suitable optical window for spectroscopic examination as a solid. T h e goal is to ensure t h a t virtually all sample molecules have matrix molecules as near neighbors. This condition is essential if criterion (b) of an "ideal" spectroscopic matrix is to be satisfied. In media such as solid nitrogen and argon, the interac tions of solute molecules with the ma trix are expected to be quite weak. Hence, perturbation of the spectro scopic behavior of a solute by the ma trix should be minimized by employ ing such media as spectroscopic "sol vents". A schematic diagram of an MI ex periment is shown in Figure 1. T h e sample is contained in a glass K n u d sen cell, which is a short piece of 7-mm glass tubing equipped a t one end with a vacuum joint and at the other end with a
