338 Analyst, June, 1974, Vol. 99, pp. 338-354 The Detection and Determination of Polynuclear Aromatic Hydrocarbons by Luminescence Spectrometry Utilising the Shpol’skii Effect at 77 K BY G. F. KIRKBRIGHT AND C. G. DE LIMA (Chemistry Department, Imperial College, London, S . W.7) The luminescence emission spectra of twenty-three polynuclear aromatic hydrocarbons (PAH) have been examined in n-alkane solvents at 77 K. The Shpol’skii effect, in which narrow-band (quasi-linear) emission spectra are obtained under these conditions when a monochromator of adequate resolving power is used, is shown to be readily observed for twelve of the compounds examined in these solvents. Quasi-linear emission spectra have also been obtained in tetrahydrofuran for some of the PAH compounds examined.These emission spectra provide for unambiguous qualitative identification of PAH compounds a t trace concentrations in solution; this effect is demon- strated by identification of the compounds present in an eight-component mixture of PAH compounds. Measurement of the low-temperature quasi-linear luminescence intensity can be applied quantitatively to the determination of these compounds provided that a standard additions procedure is employed in conjunction with the use of an internal standard to ensure sufficient accuracy and precision. THE detection and determination of trace concentrations of polynuclear aromatic hydro- carbons (PAH) is of extreme importance as most of these compounds are toxic and many are carcin0genic.l The development of methods for their unambiguous identification and accurate determination in samples of air, water, foods and petroleum products and effluents is therefore necessary.Methods based on solution spectrofluorimetry and spectrophosphorimetry have been widely employed for these purposes.2-6 One of the difficulties with these techniques arises from the relatively broad-band excitation and emission spectra observed for PAH compounds in many solvents at room temperature and even at low temperature in those solvents which form optically transparent glasses that are suitable for luminescence spectro- metry with right -angle illumination. As a consequence, methods for the determination of particular PAH compounds by luminescence spectrometry under these conditions may suffer interference caused by the overlapping excitation or emission spectra of other similar compounds present in the sample.It is usually necessary, therefore, to resort to preliminary separation of the compound by chromatography or extraction before determining it by fluorimetry or phosphorimetry. In complex PAH mixtures, such separations are frequently necessary before even qualitative identification of individual compounds present can be made by these techniques. A further problem arises from the fact that most commercially available fluorescence spectrometers utilise relatively low-resolution, high-aperture monochromators to permit detection of the weak luminescence emission obtained with trace concentrations of the species to be determined. Thus, even if structured luminescence emission is present, it may be difficult to observe with the spectral resolution attainable with this type of monochromator.In 1952, Shpol’skii, Il’ina and Klimova’ reported that some aromatic compounds, when included in the crystalline matrix formed at 77 K or below by use of selected n-alkane solvents, exhibited extremely well resolved fine structure in their luminescence emission spectra. This phenomenon was confirmed by Bowen and Brocklehurst .8 The observation under these conditions of line-like structure, in which individual lines may be less than 0.1 nm in half- width, can be explained by the postulate that the solute analyte molecules become embedded in the crystalline solvent lattice formed on cooling. The solute molecules are thus held in strictly oriented positions and at low concentrations are separated by large distances so that Q SAC and the authors.KIRKBRIGHT AND DE LIMA 339 they do not interact.In contrast to the case that applies in solvents which form transparent glasses at low temperature, where the glass does not show a short-range order and the elec- tronic transitions are very sensitive to variation in the molecular field, in the crystalline solid solutions produced for PAH compounds in n-alkane solvents the solute molecules experience a well defined molecular field that gives rise to sharp-line (quasi-linear) electronic spectra.9 The spectra exhibit the vibrational frequencies of the centres in the ground state; several workers have shown that good correlation can be achieved with results obtained independently from infrared or Raman spectra.1°-12 The quasi-linear spectra of more than 100 organic compounds have been recorded and the application of the technique has been reviewed.13-17 It has been observed that the molecular dimensions of the n-alkane solvent used must be matched to those of the solute molecule in order to obtain well defined quasi-linear spectra.16 The early studies of the Shpol’skii effect indicated that measurement of the quasi-linear spectra obtained at low temperature for PAH and other compounds should provide a powerful tool for fundamental investigation of molecular structure and for the sensitive, and extremely selective, detection and determination of these compounds.Despite the predictions of advantages to be gained by the application of the Shpol’skii effect, which were recorded in the early litera- ture, Winefordner and Lucasiewiczl8 have commented adversely on its potential and outlined possible difficulties associated with its application to quantitative analysis.These workers have also stated that few papers have presented a detailed description of the use of the technique in a specific analytical method and which reported the usual data pertaining to reproducibility, accuracy and precision, and that the availability of suitable commercial luminescence instrumentation severely limits its applications. While those papers which have been concerned with the Shpol’skii effect may in many instances lack adequate reproducibility, accuracy and precision data, a considerable number of publications have described the application of the quasi-linear luminescence measurements to both qualitative and quantitative examination of real samples.Thus Il’ina and Personovlg have applied the technique to the detection of 1,12-benzoperylene in Jurassic and Cretaceous sedimentary rocks and perylene in tertiary sediments. Dikun20 identified pyrene, 1 ,Z-benzoyyrene, 3,4-benzopyrene, perylene, 1,lZ-benzo- perylene and 1,2,7,8-dibenzanthracene in smoked fish and polluted air by using their quasi- linear spectra in n-hexane solvent at 77 K ; with the same technique, he also detected o- phenylenepyrene, 1,2,4,5-dibenzopyrene and 3,4,9,10-dibenzopyrene in polluted air. Gurov and Novikov21 identified anthracene, pyrene, 3,4-benzopyrene, 1,12-benzoperylene, perylene and coronene in soil and snow samples using the quasi-linear luminescence spectra obtained at 77 K in n-hexane solvent. Parker and Hatchard22 have applied the technique using n- octane - cyclohexane solvent to an investigation of an unusual photo-reaction of 3,4-benzo- pyrene in solutions containing polymer.Eichhoff and Kohler23 determined 3,4-benzopyrene in the atmosphere by a method based on direct measurement of the absolute intensity at 403.0 nm of its quasi-linear emission in n-heptane at 79 K. P e r ~ o n o v ~ ~ determined 3,4-benzopyrene by using the quasi-linear emission of coronene for internal standardisation in order to avoid random errors caused by variations in experimental conditions, and D i k ~ n ~ ~ devised a similar method for determining 3,4-benzopyrene by using 1,12-benzoperylene as internal standard.Muel and Lacroix26 and Jager27 utilised a standard additions procedure for the determination of 3,4-benzopyrene in cigarette smoke, alcoholic- drinks, water26 and exhaust fumes;27 these workers employed n-octane solvent at 83 or 77 K and measurement of the luminescence intensity at 403.0 nm with standard additions of 3,4-benzopyrene. Personov and Teplitskaya2* and Florovskaya, Teplitskaya and Personov29 have used the standard additions method for the determination via their quasi-linear luminescence emission of 3,4-benzopyrene, 1,12-benzoperylene and perylene in rocks and minerals of different origin. Khesina and c o - w o r k e r ~ , ~ ~ , ~ ~ in methods that involve the use of both standard additions and internal standardisation, have described the determination of 9,lO-dimethyl-1 ,Z-benzan- thracene, 1,2-benzanthracene, 1,2,5,6-dibenzanthracene, pyrene, 3,4-benzopyrene, perylene and 1,12-benzoperylene.Dikun, Krasnistkaya, Gorelova and Kalinina32 have compared the standard additions, internal standard and combined methods for the determination of 3,4- benzopyrene, using its quasi-linear luminescence emission at low temperature. In the course of the investigations of the determination of various PAH compounds described above,340 KIRKBRIGHT AND DE LIMA: DETECTION AND DETERMINATION OF [Analyst, Vol. 99 qualitative and quantitative methods for their determination in smoked f i ~ h , ~ 2 diesel engine exhausts33 and industrial Several possible difficulties that might hinder utilisation of the Shpol’skii effect for trace analysis have been reported, the first of which is the necessity to choose a suitable alkane solvent in order to observe the effect.Owing to the limited number of such solvents available it may not be possible to stimulate quasi-linear luminescence emission for some PAH compounds; conversely, this would possibly be advantageous if other compounds are to be determined in the presence of these non-emitting species. Shpol’skii, Klimova, Nersesova and Glyadkov~kii~j and Bolotnikova and N a ~ m o v a ~ ~ have investigated the influence of mole- cular aggregation and energy transfer on the intensity of quasi-linear luminescence emission. At low concentrations the spectra may be obscured owing to band emission from molecules aggregated inhomogeneously in the sample and not present in the crystalline matrix, while at high concentrations molecular aggregates excluded from the solvent lattice and which exhibit strong absorption but only weak emission may be formed.These effects might give rise to a restricted concentration range over which the luminescence emission exhibits a linear dependence on solute concentration, and to non-reproducible intensity from sample to sample, i.e., poor precision. In addition, Dokunikhin, Kizel, Sapozhekov and S01oda~~ have observed that both the intensity and width of quasi-linear emission lines is dependent on the rate of freezing of sample solutions. In view of these considerations, it is surprising that the relatively large number of reports of the successful application of measurement of quasi-linear lumines- cence emission to real samples outlined above have appeared.For this reason, and in order to evaluate the potential of an apparently powerful selective technique of analysis, we have studied the Shpol’skii effect for a series of twenty-three PAH compounds in several n-alkane solvents. The technique is similar in its operation to conventional luminescence spectrometry, except for the requirement of a monochromator with moderately high resolution, and direct experiments can be undertaken in order to investigate its potential for qualitative and quantitative analysis. have been established. EXPERIMENTAL APPARATUS- A double monochromator spectrofluorimeter (American Instrument Co., Maryland) fitted with a potted RCA 1P 28 photomultiplier tube, xenon arc lamp continuum source and low-temperature sample cell accessory was employed so as to obtain preliminary spectral data at low resolution, thus facilitating the choice of excitation wavelength for higher- resolution studies in which a mercury-vapour discharge lamp source was used.Spectra were displayed on a Bryans X-Y recorder (Model 21000). The quasi-linear luminescence emission of the compounds studied was recorded using the apparatus of higher resolution. Radiation from a medium pressure mercury-vapour discharge lamp (Wotan Hg/3) was focused into a light-tight sample cell compartment by using two silica lenses of 45 mm diameter and 75 and 50-mm focal length. Interference filters of narrow band width (50 x 50 mm, with a half-band width of 14 nm at 250 nm and 30 nm at 300 nm) were inserted between the source and sample-cell compartment for selection of the excitation wavelength.The sample tubes employed were constructed from silica tubing (Spectrosil) and were 200 mm in length, of 3 mm i.d. and 1-mm wall thickness and were sealed at one end. These tubes were used with the silica Dewar flask from the low-temperature accessory of the spectrofluorimeter used for the low-resolution studies. The liquid samples were plunged into liquid nitrogen contained in the Dewar flask so as to achieve rapid freezing, and the flask was then placed in the sample cell compartment so that the incident radiation was slightly de- focused at the surface of the frozen sample.This defocusing of the incident radiation was found to give rise to more reproducible signal intensities than when the source radiation was brought to a focus at the sample surface. A coil of Nichrome wire was positioned within the sample-cell compartment so as to be adjacent to the outer wall of the Dewar flask when the latter was placed in position. This wire was heated by passing a low a.c. current through it in order to minimise frosting of that part of the Dewar flask surface which is irradiated by the source and viewed by the detection system. A scanning grating monochromator (Optica, Model CF4) with a reciprocal linearJune, 19741 POLYNUCLEAR AROMATICS BY LUMINESCENCE SPECTROMETRY 341 dispersion at the exit slit of 1.6 nm mm-1 was positioned so that its optical axis was at 90" to that of the source and sample cell.Luminescence from the sample cell was focused on to the entrance slit of the mono- chromator by using a composite biconvex silica lens (40 mm in diameter and of 35-mm focal length). An end-window photomultiplier tube (EM1 9601B) was attached at the exit slit of the monochromator and operated at 1200 V by using a Brandenburg EHT supply. The luminescence signal was recorded directly at a potentiometric chart recorder (Servoscribe, Model RE 511.20) although for some experiments a microammeter (RCA, Model WV-84C) was employed for signal registration. REAGENTS- The n-alkane solvents used were n-pentane, n-hexane, n-octane and n-decane. These solvents were of laboratory-reagent grade and were used without further purification.Cyclo- hexane (laboratory-reagent grade) was purified by percolation through silica gel (60 to 120 mesh), which had been activated overnight at 120 to 130 "C. EPA solvent [diethyl ether - iso- pentane - ethanol (5 + 5 + a)] was prepared with ether that had been dried with sodium wire, isopentane dried with sodium wire and percolated through silica gel and ethanol treated with potassium hydroxide and redistilled. Tetrahydrofuran was treated with potassium hydroxide and distilled. Other solvents for study of the matrix were used without pre-treatment. MATERIALS- Samples of pure polynuclear aromatic hydrocarbon compounds were kindly donated by Tobacco Research Council Laboratories, Harrogate, British American Tobacco Co., South- ampton, and Shell Research Ltd., Thornton Research Centre, Chester.4,9-Di-t-butylpyrene and 3,5,8,10-tetraisopropylpyrene were kindly provided by Professor Arne Berg, University of Aarhus, Denmark. PROCEDURE - Low-resolution excitation and emission spectra were recorded in EPA solvent at 77 K with the Aminco spectrofluorimeter with the low-temperature cell attachment or at room temperature with a silica sample cell (10 x 10 x 30 mm). For examination of the quasi- linear luminescence spectra stock solutions of the PAH compounds were prepared in cyclo- hexane; these solutions were diluted to 20 or 2 pg ml-l concentration with the appropriate n-alkane and cyclohexane so that the final solutions examined contained 10 per cent. V/V of cyclohexane. These solutions were transferred into the silica sample tubes, which were then introduced directly into the Dewar flask cell that contained liquid nitrogen. Rapid freezing of the sample solutions was thus obtained.After the initial vigorous boiling action of the liquid nitrogen had subsided, the luminescence emission spectrum was scanned (using a pre-selected excitation wavelength) at 6 nm min-1. RESULTS SPECTRAL CHARACTERISTICS- Cyclohexane is a generally suitable solvent for PAH compounds and its presence in concentrations up to 10 per cent. V/V in the n-alkane solvents used for low-temperature luminescence work has been demonstrated not to disturb appreciably the spectra ~ b t a i n e d . ~ ' . ~ ~ . ~ ~ As only small amounts of some of the PAH compounds examined were available, it was therefore decided to use cyclohexane for their dissolution and n-alkane - cyclohexane mixtures (90+10) for dilution of the stock PAH solutions for the examination of their luminescence spectra.For some PAH compounds quasi-linear luminescence emission is observed only at 77 K when an alkane solvent of matching molecular dimensions is employed. Thus, in a survey of the occurrence of the Shpol'skii effect for a range of twenty-two PAH compounds, it has been necessary to employ several n-alkane solvents in order to choose that which was the most suitable for each compound. An examination of earlier work26,30,31J34 revealed that n-octane had proved to be suitable as a solvent for study of the Shpol'skii effect for some PAH compounds. Initially, therefore, the low-temperature luminescence emission spectra of all twenty-two compounds were342 KIRKBRIGHT AND DE LIMA: DETECTION AND DETERMINATION OF [Analyst, VOl.99 recorded in n-octane - cyclohexane (90+ 10). Excitation wavelengths were selected from the corresponding low-temperature excitation spectra that were obtained by using the low- resolution double-monochromator spectrofluorimeter. The wavelength was chosen so as to correspond to that of the most intense low-wavelength excitation peak in order to minimise observed scattered light during the study of the emission spectrum. Of the compounds examined in n-octane - cyclohexane solvent at 77 K, triphenylene, chrysene, perylene, pyrene, 3,5,8,10- t e t raisoprop ylp yrene, 3,4-benzopyrene, 1,2- benzop yrene, 1,2,3,4-dibenzopyrene, 1,2,4,5-dibenzopyrene, 3,4 , 9,10-dibenzopyrene, 1 2-benzan t hracene and 1,2,5,6-dibenzan- thracene exhibit quasi-linear luminescence in which emission peak half-widths of 0.5 nm or less were recorded.* J U' 458.0 v) 358 0 m L 405 Wavelengthhm Fig. 1. Emission spectra at 77 K of (a) phenanthrene in n-hexane - cyclohexane; (b) triphenylene in n-octane - cyclohexane; (c) chrysene in n-octane - cyclohexane; (d) perylene in n-octane - cyclo- hexane; and (e) coronene in n-hexane The spectra observed for 2 or 20 pg ml-l solutions of these compounds in n-octane - cyclohexane at 77 K are shown in Figs. 1 to 5. The compounds 3,4,8,9-dibenzopyrene (Fig. 6), indeno[l,2,3-cd]pyrene, benzo[a]naphth0[8,1,2-~de]naphthacene (Fig. 7), S-methylchol- anthrene, 7,12-dimethyl-1,2-benzanthracene (Fig.3), 3-methylpyrene, 4,9-di-t-butylpyrene *The nomenclature used follows that of reference 1.Luminescence intensity - 392.25 408.5 41 4.0 423.75 -426.5 432.75 453.5344 KIRKBRIGHT AND DE LIMA: DETECTION AND DETERMINATION OF [Ana&Si!, VOl. 99 (Fig. 5) , phenanthrene (Fig. 1) and 9,lO-dimethylanthracene (Fig. 3) exhibited intense but broad-band luminescence emission in n-octane - cyclohexane solvent. In these spectra, typical peak half-widths greater than 1.0 nm were observed. Anthracene exhibited a very broad spectrum (Fig. 6). Some of these compounds (phenanthrene, anthracene and 9,lO- dimethylanthracene) were examined in n-hexane - cyclohexane solvent and gave somewhat narrower half-band widths in their luminescence spectra (Figs.1, 3 and 6). 3,4,8,9-Diben- zopyrene was examined also in n-decane - cyclohexane (90 + 10) and gave a more well defined spectrum than that obtained in n-octane - cyclohexane (Fig. 6). A schematic presentation of the luminescence characteristics of fifteen of these compounds appears in Fig. 8. Fig. 3. Emission spectra a t 77 K of (a) 9,lO-dimethylanthracene in n-hexane - cyclohexane; and ( b ) 1,2-benzanthracene, (c) 7,12-dimethyl-1,2-benzanthracene, (d) 3-methylcholanthrene and (e) 1,2,5,6- dibenzanthracene in n-octane - cyclohexane Most early observations of the Shpol’skii effect for PAH compounds were made utilising n-alkane solvents and it has generally been accepted that matrices formed from the straight- chain hydrocarbon solvents of suitable molecular dimensions are required in order to obtain quasi-linear emission spectra.A striking demonstration of the need to select the correct n-alkane solvent can be made by comparison of the spectrum obtained for anthracene in n-hexane - cyclohexane (Fig. 6), in which quasi-linear emission is obtained, with the spectrum for anthracene in n-octane - cyclohexane (Fig. 6), in which only a very broad emission is observed. The restriction of the effect to relatively non-polar solvents, however, would result in limited analytical utility and restrict the study of the technique to those compounds which are soluble in these solvents.June, 19741 m m m m ? ? -.- aJ t 5) E C .- -I 345 Wavelengthhm Fig. 4. Emission spectra a t 77 K of some dibenzopyrenes in n-octane - cyclo- hexane: (u) 3,4,9,10- dibenzopyrene; (b) 1,2,4,5-dibenzopyrene; and (c) 1,2,3,4-di- benzop yrene A preliminary qualitative study of the use of other solvents was therefore undertaken.The model PAH compound chosen for study was coronene; this hydrocarbon exhibits a simple and well defined quasi-linear emission spectrum in n-hexane (Fig. 1). The emission spectra of solutions cor,t aining 20 pg ml-l of coronene in dioxan, pentanol, carbon tetrachloride, chloro-Luminescence intensity Luminescence intensity 382.25 384.25 t 3 9 4 - 2 5 398.0 = 378.0 399.0 Hg \ e 460.0 476.0 479.0 483.5 b n n 450.0 n 3 381.0 386.0 391.5 393.0 396-0 402.0 411.0 41 4.0 416.0 375.0 "5- 375.0 381.25 382.75 384.0 386.25 387.75 391.75 393.0 395-0 396.0 1 Hg j R 379.2 383.75,m $ (bi 520 500 440 Wavelengthhm Emission spectra at 77 K in n-octane - cyclohexane of (a) indeno[l,2,3-cd]pyrene; and (b) benzo[u]naphtho[8,1,2-cde]naphthacene 34 Wavelengthhm348 KIRKBRIGHT AND DE LIMA: DETECTION AND DETERMINATION OF [Analyst, Vol. 99 form, bromoform, diethyl ether, dimethylformamide, 1,1,2,2-tetrachloroethane and tetra- hydrofuran were examined at 77 K.Only with tetrahydrofuran was a well defined quasi- linear emission spectrum observed for coronene [Fig. 9 (a)]. The corresponding phosphores- cence emission spectrum for coronene is shown in Fig. 10. Of twenty-three PAH compounds studied, quasi-linear emission was also observed for 4,9-di-t-butylpyrene, 1,2-benzopyrene and 1,2,5,6-dibenzanthracene in tetrahydrofuran. The spectra observed for these compounds are shown in Figs.9 and 11 and the data recorded are listed in Table I. Although spectral emission band widths that were greater than expected for the Shpol’skii effect were observed for 3-methylcholanthrene, perylene and 3,4,9,10-dibenzopyrene in tetrahydrofuran as shown in Figs. 9 and 11, sharp and useful spectra were obtained for these compounds. : LDO 445 507 370 ‘I 420 Wavelengthh Fig. 9. Emission spectra at 77 K of some PAH compounds in tetrahydrofuran: (a) coronene; (b) perylene ; (c) 1,2-benzopyrene (phosphorescence spectrum) ; and (d) 1,2-benzopyrene (fluorescence spectrum) Good agreement is observed in the wavelength assignments made for the principal quasi-linear luminescence emission maxima in this work with those recorded earlier for several of the compounds studied by other ~ o r k e r s .~ ~ , ~ * - ~ ~ The wavelength reproducibility of these emission maxima and the relative freedom from overlap of the narrow-band “quasi-line” spectra compared with that obtained in solution at room temperature or in glass-forming organic solvents such as EPA at 77 K suggest that the quasi-linear spectra may be extremely useful for qualitative identification of PAH compounds. Fig. 2 shows the luminescence emission spectrum at 77 K for pyrene in EPA glass and the corresponding quasi-linear emission spectrum at 77 K in n-octane - cyclohexane. Both spectra were recorded with the high- resolution instrumentation. The gain in structure obtained by utilising the Shpol’skii effect and the possibility of less ambiguous identification of this compound in the presence of others is clearly seen.In order to demonstrate the “fingerprinting” ability of the technique, a synthetic mixture of eight PAH compounds was prepared and its low-temperature luminescence emission spectrum was recorded at 77 K in n-octane - cyclohexane solvent; Fig. 12 shows theJune, 19741 POLYNUCLEAR AROMATICS BY LUMINESCENCE SPECTROMETRY 349 570 Wavelengthhm Fig. 10. Phosphorescence emission spectra of coronene in n-hexane (broken line) and in tetra- hydrofuran (solid line) 430 435 390 450 Ln ld) Wavelengthhm Fig. 11. Emission spectra a t 77 K of some PAH compounds in tetrahydrofuran: (a) 4,9-di-t-butyl- pyrene ; (b) 1,2,5,6-dibenzanthracene ; (c) 3-methylcholanthrene ; and (d) 3,4,9,10-dibenzopyrene350 KIRKBRIGHT AND D E LIMA: DETECTION AND DETERMINATION OF [AfldySt, VOl.99 emission spectrum obtained. Each of the eight hydrocarbons present in the mixture is readily identified from the principal luminescence emission maxima observed. Even when some overlap occurs for certain principal peaks, there is sufficient information present in the minor features of the spectrum of each compound to permit its detection by using alternative peak wavelengths. TABLE I EMISSION AT 77K OF SOME PAH COMPOUNDS IN TETRAHYDROFURAN Excitation wavelength 300 nm Compound .. Wavelengths of principal emission maxima Coronene . . 424.5 (s), 425-0 (m), 431.25 (w), 443.5 (vs), 444.0 (s), 450.75 (m), 451.5 (m), 452.0 (m), 453-0 (m), 472.0 (m), 482.0 (w) p : 515.0 (m), 525.0 (m), 528-0 (w), 546.0 (w), 547.0 (m), 549.25 (m), 554.5 (m), 555.75 (m), 562.0 (vs) .. 448.0 (m), 453.25 (vs), 457.25 (m), 461.0 (m), 465.0 (w), 473.75 (s), 475.0 (s), 480.0 (m), 481.25 (m), 483.0 (m), 488.0 (w) . , 374.75 (m), 376.50 (m), 378.25 (w), 381.0 (w), 382.5 (m), 386.0 (vs), 388-5 (m), 390.75 (w), 394-5 (m), 396.0 (m), 397.0 (m), 399.0 (m), 406.25 (m), 407.5 (m), 409.0 (m) p : 533.25 (s), 542-75 (m), 544.25 (m), 546.0 (w) 3,4,9,10-Dibenzopyrene 433.5 (vs), 438.75 (w), 448.5 (w), 461-25 (m), 465.5 (d,w) 3-Methylcholanthrene . . 393.25 (vs), 397.5 (w), 414.25 (m), 415.5 (m), 417.5 (m), 420.5 (m) 1,2,5,6-Dibenzanthracene 392-25 (m), 394-25 (vs), 395.0 (m), 397.5 (w), 405.75 (w), 414.5 (vw), 415-5 (w), 416.25 (m), 418-0 (m). 4,9-Di-t-butylpyrenc .. 374.75 (m), 375.5 ( s ) , 376.0 (vs), 383.25 (m), 384.0 (m), 384.5 (m), 385.25 (m), 387.0 (m), 388-5 (m), 389.25 (w), 392-5 (m), 394.25 (m), 395.5 (m), 396-0 (m), 397.0 (m), 397.5 (m) Perylene . . 1,2-Benzopyrene .. Concentration of all compounds in tetrahydrofuran : 20 pg ml-1. vs, very strong emission: s, strong; w, weak; m, medium; d, diffuse; p, phosphorescence emission. Wavelengths in italics indicate the most intense peaks. The detection limits given for the compounds studied can be defined as that concentration of hydrocarbon in the solvent employed which gives a signal to noise ratio of 2 at the wave- length of the principal quasi-linear emission. The wavelengths (nm) used and the values (pg ml-l) obtained for these compounds were: phenanthrene (345-5,O.l) ; triphenylene (462.25, 0.1) ; chrysene- (360.5, 0.05) ; perylene (451.0, 0.05) ; pyrene (371.75, 0.05) ; 3.4-benzopyrene (403.0, 0.005) ; 1,2-benzopyrene (387.75, 0.07) ; 1,2,3,4-dibenzopyrene (395.25, 0.08) ; 1,2,4,5- dibenzopyrene (395.5, 0-08) ; 3,4,8,9-dibenzopyrene (449.25, 0.04) ; 3,4,9,10-dibenzopyrene (431.5, 0.1) ; anthracene (377.0, 0.05) ; 1,2-benzanthracene (383.75, 0.1) ; 7,12-dimethyl-1,2- benzanthracene (397.75, 0.2) ; 3-methylcholanthrene (392.5, 0.2) ; 1,2,5,6-dibenzanthracene (394.25, 0.1) ; 9,lO-dimethylanthracene (405.75, 0.05) ; 3-methylpyrene (375.0, 0.2) ; 4,9-di-t- butylpyrene (375.0, 0-2) ; 3,5,8,10-tetraisopropylpyrene (379.25, 0.1) ; indeno[l,2,3-cd]pyrene (462.5, 0.3) ; benzo[rn]naphtho[8,1,2-~de]naphthacene (444.75, 0.3) ; and coronene (445.0, 0.1).These limits are estimates only and would be subject to considerable improvement if a more efficient optical arrangement for collection of the emitted radiation and more sophisti- cated detector electronics were used. QUANTITATIVE STUDIES- The variation in intensity of luminescence emission at 77 K with concentration was examined for the four dibenzopyrene compounds available to us in n-octane - cyclohexane solvent, and the intensity of the quasi-linear emission for 3,4,9,10-dibenzopyrene, 1,2,4,5- dibenzopyrene and 1,2,3,4-dibenzopyrene was measured at 43160, 395.50 and 395.25 nm, respectively. This variation was also examined for 1,2,3,4-dibenzopyrene at its broad emission peak at 470-25 nm. Measurement at this wavelength, rather than at 395.25 nm, would be advantageous in the determination of this compound in the presence of 1,2,4,5-dibenzopyrene as it avoids interference from the emission of the latter compound at 395-5 nm.The lumi- nescence growth curves obtained are plotted logarithmically in Fig. 13. For the compounds examined, the quasi-linear emission intensity is linear over a con- centration range of two orders of magnitude. Fig. 13 also shows that similar intensity, slope and range of linearity are obtained at both 395.25 and 470.25 nm for the luminescence growthJune, 19741 POLYNUCLEAR AROMATICS BY LUMINESCENCE SPECTROMETRY 351 I I I II I I I I ’ 500 g & h Wavelengthhm Emission spectra of a synthetic mixture in n-octane - cyclohexane a t 77 K: Fig. 12. (u) pyrene; (b) 1,2-benzanthracene; (6) 3-methylcholanthrene; (d) 1,2,5,6-dibenzanthracene ; (e) 1,2,4,5-dibenzopyrene; (f) 3,4-benzopyrene; (g) 3,4,9,10-dibenzopyrene ; and (h) 3,4,8,9-dibenzopyrene curve of 1,2,3,4-dibenzopyrene.Although 3,4,8,9-dibenzopyrene does not show quasi-linear luminescence emission in n-octane - cyclohexane solvent at 77 K it can be seen in Fig. 13 that a similar intensity and range of linearity is observed for the luminescence of this compound when measured at the wavelength of its more intense emission at 449.25 nm. Although it appears that direct measurement of the intensity of quasi-linear luminescence emission is capable of permitting the direct quantitative determination of compounds such as the dibenzopyrenes, in the examination of real samples several difficulties exist that lead to the requirement for the use of a standard additions procedure and the use of an internal standard.Thus when other compounds are present with the PAH compound to be deter- mined, their absorption spectra may overlap that of the analyte molecule and give rise to an “inner filter” effect and low fluorescence intensities ; energy transfer between the analyte molecule and others present in the sample may also lead to inaccurate values of intensity of quasi-linear emission compared with those expected in the absence of other compounds; the352 KIRKBRIGHT AND DE LIMA: DETECTION AND DETERMINATION OF [Analyst, Vol. 99 use of the standard additions method of analysis minimises these inaccuracies. Variation of experimental measurement conditions, such as rate of freezing or the reproducibility with which the sample cell can be placed in the optical path, can lead to poor precision in measure- ment of quasi-linear luminescence intensity.These effects can be minimised by the use of an internal standard in quantitative work. A combined standard addition - internal standard procedure was adopted for quanti- tative determination of the dibenzopyrene compounds in n-octane - cyclohexane solvent. The internal standard employed was selected so that the wavelength of its quasi-linear lumi- nescence emission did not overlap that of the quasi-linear emission of the compound to be determined. It is, of course, not possible to avoid such overlapping in the corresponding excitation spectra; it is necessary for radiation transmitted by the single filter used to excite luminescence from both analyte compound and internal standard, so that some overlap in excitation spectra must occur if the internal standard technique is to be used in this manner.1,2,4,5-Dibenzopyrene, 3,4,9,10-dibenzopyrene and 3,4,8,9-dibenzopyrene were deter- mined by measurement of their quasi-linear luminescence in n-octane - cyclohexane at 77 K at the wavelengths employed in order to construct their luminescence growth curves shown in Fig. 13. The quasi-linear emission of 3,4-benzopyrene at 403 nm was used as internal standard for the determination of 1,2,4,5-dibenzopyrene and 3,4,9,10-dibenzopyrene and a standard additions procedure was employed in order to produce the calibration graphs shown in Fig.14 (a) and (b). For the determination of 3,4,8,9-dibenzopyrene in n-octane - cyclohexane the yuasi-linear emission of 1,2-benzopyrene at 387.75 nm was used for internal standardisation ; the calibration graph obtained is shown in Fig. 14 (c). In each instance the ratio of the observed intensities for the luminescence of the analyte and internal standard are plotted against concentration. lo3 3 Concentrationh Fig. 13. Luminescence intensity zlem,ts concentration graphs for some dibenzopyrenes determined a t 77 K in n-octane - cyclohexane: 0, 3,4,8,9-dibenzopyrene; A, 3,4,9,10-dibenzopyrene; 0, 1,2,4,5- dibenzopyrene ; and [7, l12,3,4-dibenzopyrene Although the use of an internal standardisation procedure leads to improved precision by decreasing the effects of random errors in the measurement, an “inner filter” effect occurs in a manner similar to that mentioned above owing to the overlap of the excitation spectrum of the internal standard compound with that of the compound determined ; lower quasi-linear luminescence emission intensities for the analyte compound are thus obtained in the presenceJune, 19741 POLYNUCLEAR AROMATICS BY LUMINESCENCE SPECTROMETRY 353 Concentration x IOVM Working curves for (u) 1,2,4,5-, (b) 3,4,9,10- and (c) 3,4,8,9-dibenzopyrene using the com- Fig.14. bined method (addition method and internal standard) a t 77 K in n-octane - cyclohexane of the internal standard compound. Improvement in precision is obtained over the analytical working range when the internal standardisation procedure is employed, but the decrease in luminescence signal intensity that results leads to some deterioration in detection limit for the compounds studied.Table I1 shows the effect of the presence of equimolar or greater concen- trations of some other PAH compounds on the quasi-linear luminescence of the dibenzopyrene compounds excited at 300 nm in n-octane - cyclohexane. The percentage suppression or enhancement of the luminescence signal in the presence of these compounds is listed. In almost all instances suppression of quasi-linear emission of the dibenzopyrenes occurs owing to the “inner filter” effect or, possibly, by energy transfer. The enhancement of 1,2,3,4- dibenzopyrene emission at 395.5 nm in the presence of 1,2,4,5-dibenzopyrene is caused by the direct overlap of their quasi-linear emission at this wavelength; no effect is observed when the luminescence of 1,2,3,4-dibenzopyrene is measured at 470.5 nm.TABLE I1 EFFECT OF OTHER PAH COMPOUNDS ON MEASURED LUMINESCENCE INTENSITIES AT 77K FOR DIBENZOPYRENE COMPOUNDS STUDIED Results are expressed as change in luminescence intensity, per cent. Interferent 1,2,3,4- 1,2,4,5- 3,4,8,9- 3,4,9,10- 3,4- 1,2- 1,2,5,6- DBP DBP DBP DBP BP BP DBA PERe 1,2,3,4-Dibenzopyrene - +226a; -35%; -35%; -54*0a*b - +22.2a.g; - Nilb Nilb Nilb - 29.6b 1,2,4,5-Dibenzopyrene - 34.5 - -35.0 -57.2 -60.5 - - 22.9’ - 3,4,8,9-Dibenzopyrene - 37.8 - 33.4 - -114.9 -58.0 -45.5‘ - - 22.8 ; - 358d 3,4,9,10-Dibenzopyrene - 17.8 -31-5 -45.8 - -20.6 - -22.8 -31.5 - 56-26 - 32.8d - 82.5d a at 395.5 nm; b at 470-5 nm; C four-fold excess of 1,2-BP; d ten-fold excess: e perylene is not excited with the filter employed; f a t ten-fold excess there is direct interference; g a t ten-fold excess there is direct interference in the emission a t 395.5 nm and suppression (-60 per cent.) a t the 470-0 nm emission.3,4-Benzopyrene, 1,2-benzopyrene, 1,2,5,6-dibenzanthracene and perylene appear as 3,4-BP, 1,2-BP, 1,2,5,6-DBA and PER, respectively, in the table; DBP denotes dibenzopyrene. CONCLUSIONS Our preliminary study reported here confirms that the high selectivity claimed for the identification of PAH compounds utilising the Shpol’skii effect at 77 K is readily attained. The quasi-linear emission spectra can be used for unambiguous “fingerprinting” of these compounds in solution at trace concentrations.The wavelength assignments for the principal quasi-linear luminescence emission maxima observed agree well with those made by other workers. At present, the application of this fingerprinting technique is limited to those354 KIRKBRIGHT AND DE LIMA aromatic hydrocarbons which are soluble in n-alkanes. The ability to obtain quasi-linear emission spectra in tetrahydrofuran, however, suggests that the technique may be extended to compounds that are not soluble in these solvents and that are more polar than PAH compounds. Although it is necessary to match the molecular dimensions of solute and solvent species, thisrequirement can be used to advantage in order to improveselectivity by choice of suitable solvent if spectral overlap at major luminescence peaks is observed for some compounds.The quantitative use of the Shpol’skii effect for trace analysis requires careful calibration and use of a combined internal standard - standard additions technique so as to minimise the effects of energy transfer, the inner filter effect and experimental variables and to attain acceptable precision and accuracy. It has been sh0wn4~~~3 that the corresponding absorption spectra of PAH compounds at 77 K may be quasi-linear in character. The use of a narrow-line excitation source, such as a tunable dye laser, would therefore provide even greater selectivity and sensitivity for the analytical application of the Shpol’skii effect. One of us (C.G. de L.) thanks the University of Brasilia for study leave and UNESCO for the grant of a Fellowship.We also thank those agencies mentioned in the text for the provision of samples for examination. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. REFERENCES Clar, E., “Polycyclic Hydrocarbons,” Academic Press, London, 1964. McKay, J. F., and Latham, D. K., Analyt. Chem., 1972, 44, 2132. Van Duuren, B. L., Ibid., 1960, 32, 1436. Sawicki, E., Hauser, T. R., and Stanley, T. W., I n t . J . A i r Pollut., 1960, 2, 253. Hood, L. V., and Winefordner, J. D., Analytica Chim. Acta, 1968, 42, 199. Sauerland, H. D., and Zander, M., Erdol Kohle, 1966, 19, 502. Shpol’skii, E. V., Il’ina, A. A., and Klimova, L. A., Dokl.Akad. Nauk. S.S.S.R., 1952, 87, 935; Bowen, E. J., and Brocklehurst, B., J . Chem. SOC., 1954, 3875. Shpol’skii, E. V., Zh. Prikl. SpeRtrosk., 1967, 7, 492. Pesteil, L., and Ciais, A., C.R. Hebd. Se’anc. Acad. Sci., Paris, 1959, 249, 528. Bolotnikova, T. N., Optika Spektrosk., 1959, 7, 44. -, Ibid., 1959, 7, 217. Shpol’skii, E. V., Soviet Phys. Usp., 1959, 2, 378. -, Ibid., 1960, 3, 372. Chem. Abstr., 1953, 47, 420513. -, Ibid., 1962, 5, 522. -, Ibid., 1963, 6, 411. Parkcr, C. A. , “Photoluminescence of Solutions,” Elsevier, Amsterdam, 1968, p. 383. m7inefordner, J. D., and Lucasiewicz, R. J., Talanta, 1972, 19, 381. Il’ina, A. A., and Personov, R. I., Geochemistry, Ann Arbor, 1962, No. 11, 1089. Dikun, P. P., J . A@Z. Sfiectrosc., 1967, 6, 130. Gurov, F. I., and Novikov, Yu. V., Gig. Sanit., 1971, 36, 409. Parker, C. A., and Hatchard, C. G., Photochem. Photobiol., 1966, 5, 699. Eichhoff, H. J., and Kohler, N., 2. analyt. Chem., 1963, 197, 272. Personov, R. I., Zh. Analit. Khim., 1962, 17, 506; Chem. Abstr., 1963, 58, 24b. Dikun, P. P., Vop. Onkol., 1961, 7, 42; Chern. Abstr., 1962, 57, 65%. Muel, B., and Lacroix, G., Bull. SOC. Chim. Fr., 1960, 2139. Jager, J., Chemickk L i s t y , 1966, 60, 1184. Personov, R. I., and Teplitskaya, T. A., J . Analyt. Chem. U.S.S.R., 1965, 20, 1176. Florovskaya, V. N., Teplitskaya, T. A., and Personov, R. I., Geochemistry Int., 1966, 3, 419. Danil’tseva, G. E., and Khesina, A. Ya., J . A p p l . Spectrosc., 1966, 5, 196. Fedoseeva, G. E., and Khesina, A. Ya., Ibid., 1968, 9, 838. Dikun, P. P., Krasnistkaya, N. D., Gorelova, N. D., and Kalinina, I. A., Ibid., 1968, 8 9 254. Varshavskii, I. L., Shabad, L. M., Khesina, A. Ya., Khitrovo, S. S., Chalabov, V. G., and Pakhol’nik, Kolyadich, M. N., Khesina, A. Ya., Shefer, S. S., and Yamskova, V. P., Gig. Trud. Prof. Zabol., Shpol’skii, E. V., Klimova, L. A., Nersesova, G. N., and Glyadkovskii, V. I., Optics Spectrosc., N . Y . , Bolotnikova, T. N., and Naumova, T. M., Ibid., 1968, 24, 253. Dokunikhin, N. S., Kizel, V. A., Sapozhekov, M. N., and Soloda, S. L., Ibid., 1967, 23, 42. Ciais, A., and Pesteil, L., C.R. Hebd. Se’anc. Acad. Sci., Paris, 1959, 248, 1311. Bowen, E. J., and Brocklehurst, B., J . Chem. Soc., 1955, 4320. Shpol’skii, E. V., Klimova, L. A., and Personov, R. I., Optics Spectrosc., N . Y., 1962, 13, 188. Teplyakov, P. A., Ibid., 1963, 15, 350. Klimova, L. A., Ibid., 1963, 15, 185. Ruzevich, E. S., Ibid., 1963, 15, 191. A. I., Ibid., 1965, 2, 68. 1971, 15, 3. 1968, 24, 25. Received October 23rd, 1973 Accepted January 2nd, 1974