Simultaneous determination of 11 polycyclic aromatic hydrocarbons ...

3 downloads 0 Views 84KB Size Report
26 V. L. Amszi, I. Cordero, B. Smith, S. A. Tucker, W. E. Acree, Jr., C. Yang, E. Abu-Shaqara and R. G. Harey, Appl. Spectrosc. 1992, 46,. 1156. 27 S. A. Tucker ...
Simultaneous determination of 11 polycyclic aromatic hydrocarbons (PAHs) by second-derivative synchronous spectrofluorimetry considering the possibility of quenching by some PAHs in the mixture A. Andrade Eiroa, E. Vázquez Blanco, P. López Mahía, S. Muniategui Lorenzo* and D. Prada Rodríguez Department of Analytical Chemistry, Coruña University, Zapateira, 15071 Coruña, Spain Received 5th June 1998, Accepted 7th August 1998

A method capable of determining 11 PAHs (acenaphthene, anthracene, benz[a]anthracene, benzo[a]pyrene, benzo[b]fluoranthene, benzo[k]fluoranthene, chrysene, phenanthrene, fluoranthene, indeno[1,2,3-cd]pyrene and perylene) in a mixture of 18 by second-derivative synchronous spectrofluorimetry in the constant wavelength mode was developed. It has not been possible to determine the following PAHs in the mixture: dibenz[a,h]anthracene, fluorene, indene, naphthalene, pyrene, triphenylene and 1,12-benzoperylene. The possibility of selective quenching of some molecules by others in the mixture was considered and it was found that the fluorescent signal of fluorene is quenched by indeno[1,2,3-cd]pyrene. Other papers related to complex mixtures have not taken this possibility into account; on the other hand, the identification and quantification of PAHs in such a complex mixture (18 compounds) has never been attempted before. The approach studied allows the sensitive, rapid, easy and inexpensive identification and quantification of 11 PAHs in a solution of hexane. The detection limits are between 0.01 and 0.70 ng ml21 in most cases (except for indeno[1,2,3-cd]pyrene with a detection limit of 4.95 ng ml21) and a short analysis time (four PAHs were determined in one interval alone, Dl = 95 nm), the total identification and quantification of the PAHs taking only 10 min.

Introduction Polycyclic aromatic hydrocarbons (PAHs) are organic contaminants of great importance because many of them are known mutagens and carcinogens.1,2 These pollutants are multiaromatic ring systems produced at high temperatures by the incomplete combustion of fossil fuel or as the result of synthesis by some plants and formation during natural forest and prairie fires.325 They can be found in the atmosphere, aqueous media, soil, biota, foods, etc.6,7 The most common techniques for their determination are chromatographic methods, especially high-performance liquid chromatography on a C18 reversed column (used to solve coelution problems with isomeric PAHs)8 with fluorimetric detection.9,10 This technique achieves the separation of the 16 EPA (Environmental Protection Agency) PAHs.11,12 However, chromatographic methods are expensive, time consuming and require large amounts of organic solvents. On the other hand, spectroscopic procedures have been very limited because of the great chemical similarity between the different PAHs and their low concentrations in the environment. Conventional fluorimetry is very sensitive; however, it does not allow for the analysis of multicomponent mixtures owing to overlapping of the spectral bands, which are usually broad.13 The development of synchronous spectrofluorimetry has allowed narrower peaks to be obtained, so it can be applied to multicomponent mixtures, in which we can now determine different compounds. The technique is very simple, rapid and sensitive. Synchronous fluorescence spectrometry was introduced by Lloyd in 197114 for the qualitative and quantitative analysis of machine oils. This technique consists of simultaneously varying the excitation and emission wavelengths with a fixed difference between them (in the case of the synchronous method in the constant Dl mode).15

Using synchronous fluorescence in the constant Dl mode, Vo-Dinh and Mart´ınez16 identified and quantified six PAHs without a pre-separation step in a coal liquid. The method has also been applied to the identification of compounds such as benzo[k]fluoranthene and benzo[a]pyrene in chromatographic eluates of urine samples.17 Baudot et al.18 determined PAHs in air samples collected in a factory by synchronous fluorescence spectrometry in the constant wavelength mode. They identified many PAHs but they could not quantify any of them because of the great complexity of the spectra obtained. Santana Rodr´ıguez et al.19 applied synchronous spectrofluorimetry in non-ionic micellar media to study benzo[a]pyrene, perylene and chrysene in sea-water samples which had been treated. The application of derivative techniques to luminescence spectrometry was first proposed by Green and O’Haver in 1974.20 John and Soutar21 pointed out that there is an obvious potential in combining synchronous and derivative fluorimetry to enhance minor spectral features. Several workers have used the second-derivative technique to increase the selectivity of synchronous spectroscopy for determining different compounds in samples of motor oil, waters and human serum.22–24 Several papers in which the quenching of PAHs by other compounds such as chlorine25 was described and studies in which nitromethane was evaluated as a selective quenching agent for alternant alkyl-substituted PAHs26,27 have been published recently. However, the different studies published on the simultaneous determination of several PAHs by synchronous luminescence did not describe the possibility of quenching of fluorescence of some compounds (PAHs) by others present in the mixture. In this work, synchronous fluorescence in the constant wavelength mode and derivative spectrometry were applied to the identification and quantification of several PAHs in a Analyst, 1998, 123, 2113–2117

2113

mixture of 18, taking into account the possibility of quenching of some fluorescent signals by other PAHs present in the mixture. It is very important to note that in previous studies16,18,19,28,29 small wavelength increments were applied to the determination of PAHs. However, in our case, high wavelength increments were considered for achieving the identification and quantification of the mentioned compounds in a complex mixture.

Experimental Apparatus All the spectra were obtained on a Model LS-50B computer controlled spectrofluorimeter (Perkin-Elmer, Norwalk, CT, USA) equipped with a xenon discharge light source with pulses equivalent to 20 kW for 8 ms duration and two Monk–Gillieson monochromators for excitation and emission and provided with PE-FLDM fluorescence software for acquisition and processing of spectral data. Wavelength accuracy and wavelength reproducibility were 1.0 nm and 0.5 nm, respectively. The excitation and emission slits can be varied from 2.5 to 15 nm and from 2.5 to 20 nm, respectively. Fluorescence measurements were performed using standard 1 3 1 cm quartz cells. Reagents The PAHs used, acenaphthene (Ace), anthracene (Anth), benz[a]anthracene (BaA), benzo[a]pyrene (BaP), benzo[b]fluoranthene (BbFt), benzo[k]fluoranthene (BkFt) dibenz[a,h]anthracene (DBahA), chrysene (Chry), phenanthrene (Phen), fluoranthene (Ft), fluorene (Flu), naphthalene (Naph), indene (Ind), indeno[1,2,3-cd]pyrene (IP), triphenylene (Try), perylene (Per), pyrene (Pyr) and 1,12-benzoperylene (BghiP), were provided by Supelco (Bellefonte, PA, USA) or ChemService (West Chester, PA, USA). Stock standard solutions of 1000 mg ml21 in hexane [super purity solvent from Romil (Cambridge, UK)] were prepared and used for further dilutions. Concentrations (individual component solution and in the mixture) of working solutions were 20 ng ml21 for all the PAHs, except for Phen, BkFt and IP, whose concentrations were 60, 10 and 200 ng ml21, respectively, because of the fluorescence efficiency of these molecules. Other more dilute mixture solutions were also studied. Method development The selection of the experimental conditions is very important for achieving the identification of the different PAHs in the mixture. Solvent. Hexane was chosen because it has a good extractive capacity for organic compounds from many matrices23,30 and it does not show important interferences when the extracts are studied by spectrofluorimetry,30 although it is possible to use more polar solvents which improve the fluorescence efficiency of some molecules. Wavelength intervals. This selection was made empirically. The fluorescence intensity is at a maximum when the excitation wavelength corresponds to the maximum of the excitation spectrum and when the emission wavelength corresponds to the maximum of the emission spectrum. In this case, however, this definition was not useful because greater sensitivity does not always mean high selectivity. Scans were recorded from Dl = 10 to 205 nm. Scans with Dl below 10 nm were not useful since the spectrofluorimeter 2114

Analyst, 1998, 123, 2113–2117

did not allow us to fix excitation and emission slits below 2.5 nm. It must be noted that when (le + la) > (la 2 le), the value of Is is given by the expression Is ª [le + la 2 (la 2 le)]3 where Is is the scattered light intensity, le and la are the spectral widths of the excitation and the emission monochromators, respectively, and la and le are the emission and the excitation wavelengths, respectively.30 The excitation and emission slit widths vary from 2.5 to 15 nm and from 2.5 to 20 nm, respectively. Several workers30,31 have reported that the fluorescence signal is proportional to the square of the product of the excitation and emission slit widths, and that the use of narrow slit widths implies an increase in selectivity and, in general, a decrease in sensitivity, whereas wide slit widths lead to the opposite effect. We have proved that large slit widths of the emission monochromator provide higher signals than with narrow slit widths but the resolution of the peaks becomes poorer as the slit width increases. However, for the compounds included in this work, narrow excitation slit widths provided higher fluorescence signals. The most intense and resolved peaks were obtained for excitation and emission slit widths set to 2.5 and 5.0 nm, respectively. Scan speed. The scanning speed can be selected in increments of 1 nm min21 from 10 to 1500 nm min21. Selection of the scanning speed was influenced by two important factors: reproducibility of the signal and analysis time. To ensure the best signal-to-noise performance it is necessary to select a slow scanning speed. The synchronous fluorescence spectra of the hexane solutions obtained were recorded at a rate of 240 nm min21. We have proved that rates of around 240 nm min21 are sufficient to ensure a good signal-to-noise performance without wasting time unnecessarily. On the other hand, it must be noted that although modern spectrofluorimeters provide acceptable signalto-noise performance using high scanning speeds, we must try to slow the scanning speed owing to the subsequent application of the second-derivative technique. Smoothing filter. Two types of smoothing filters, Savitzsky– Golay and binomial, were available for use. The type of filtering selected depended on which aspects of the spectra were being analyzed. We adopted binomial filtering because it offers the advantage of a good noise reduction with minimum peak distortion. Second-derivative spectra. Synchronous fluorescence spectra of the individual compounds and the mixture were recorded between 200 and 500 nm under the instrumental conditions described above. Second-derivative spectra were then generated by the data handling system of the spectrofluorimeter.

Results and discussion Once the experimental conditions had been selected, 40 synchronous spectra were recorded. The values of Dl range from 10 to 205 nm, and the increment between intervals was 5 nm. It was necessary to obtain second-derivative spectra since complete identification and quantification of the compounds using direct spectra were impossible. The second-derivative technique is a mathematical operation which eliminates constant background. The perfect coincidence of (at least) two consecutive points (a maximum and a minimum) when the second-derivative spectra of the mixture and the individual compound in the same interval are overlapped reveals the resolution of the compound (see Fig. 1).

Although it is common that in spectroscopic studies the signal of a mixture is considered to represent the addition of the different signals of all the compounds included in the mixture, it is not always so. There are cases where quenching may be considered. We noted quenching only for fluorene. With Dl = 10 nm, where Flu presents a very high fluorescent signal without any

other interference, we observed important quenching of this signal in the mixture, in contrast to its independent signal. A preliminary study of this aspect proved that the IP is the main compound capable of quenching Flu within the concentration ranges included in this work. We worked with mixtures of Flu and IP in different ratios (1 : 0.72, 1 : 1.8, 1 : 3, 1 : 3.6 and 1 : 5.8) in the linear range of Flu (0–125 ng ml21) and with low

Table 1 Resolved compounds, wavelength intervals for each compound and spectral region (minima and maxima) selected for their identification and quantification

Table 2 Detection and quantification limits of all the resolved PAHs

Spectral region Compound

Dl/nm

Maximum/nm

Minimum/nm

Ace Anth BaA BaP BbFt BkFt Chry Phen Ft IP Per

95 20 95 15 160 95 95 70 200 195 30

231.0 352.0 284.4 383.2 304.3 311.1 271.0 287.8 290.0 318.4 414.0

226.6 356.0 289.0 387.2 300.5 307.1 267.3 292.4 287.0 315.0 406.4

Compound

LODa/ng ml21

LOQa/ng ml21

RSDa (%)

Ace Anth BaA BaP BbFt BkFt Chry Phen Ft IP Per

0.09 0.05 0.11 0.10 0.12 0.01 0.04 0.70 0.20 4.95 0.02

0.80 0.30 0.36 0.40 0.80 0.05 0.09 2.30 1.20 15.90 0.09

2.3 2.4 2.5 2.0 2.0 1.6 2.3 8.0 3.0 7.0 2.9

a

LOD, limit of detection. LOQ, limit of quantification obtained according to Miller and Miller.33 RSD, relative standard deviation obtained for about 20 ng ml21 in each case.

Fig. 1 Direct spectra (A–K) and second-derivative spectra (A1–K1) of: A, acenaphthene; B, anthracene; C, benz[a]anthracene; D, benzo[a]pyrene; E, benzo[b]fluoranthene; F, benzo[k]fluoranthene; G, chrysene; H, phenanthrene; I, fluoranthene; J, indeno[1,2,3-cd]pyrene; K, perylene. Dot-dashed lines, individual compound; solid line, mixture.

Analyst, 1998, 123, 2113–2117

2115

Table 3

Linear range and statistical parameters of all the resolved PAHs Regression straight line (y = a + bx)

Compound

Linear range/ng ml21

a

b

Standard error Number of points

Regression coefficient (r2 )

Ace Anth BaA BaP BbFt BkFt Chry Phen Ft IP Per

0–60 0–150 0–150 0–200 0–150 0–25 0–50 0–300 0–60 0–1200 0–30

31.87 ± 7.05 20.23 ± 7.49 22.56 ± 7.31 21.27 ± 4.30 5.52 ± 3.83 12.03 ± 5.37 22.25 ± 6.16 4.95 ± 3.31 2.12 ± 1.79 6.17 ± 0.98 14.65 ± 7.05

76.97 ± 0.24 22.77 ± 0.11 33.81 ± 0.10 25.81 ± 0.05 24.77 ± 0.09 177.38 ± 0.44 53.89 ± 0.25 3.89 ± 0.023 17.97 ± 0.06 0.50 ± 0.00 58.71 ± 0.29

34.95 52.14 45.80 31.32 19.56 28.44 29.90 18.81 9.61 6.01 23.39

0.9995 0.9987 0.9997 0.9997 0.9993 0.9997 0.9992 0.9977 0.9994 0.9992 0.9986

7 8 9 8 7 7 6 8 6 8 7

The linear regression coefficients approach 1.0000 in all cases. Linear ranges (in ng ml21) for the different PAHs are 0–60 for Ace, 0–150 for Anth and BaA, 0–200 for BaP, 0–150 for BbFt, 0–25 for BkFt, 0–50 for Chry, 0–300 for Phen, 0–60 for Ft, 0–1200 for IP and 0–30 for Per (see Table 3). In a preliminary study, the total absence of interferences when we applied this method to river and tap waters which had been extracted with hexane made us conclude that the synchronous constant wavelength mode is very useful for the determination of all PAHs in different water samples.

Fig. 2 Direct spectra (A) and second-derivative spectra (B) of: -.-.-., acenaphthene; -------, chrysene; ........, benz[a]anthracene; -..-..-.., benzo[k]fluoranthene; and ––––––, mixture.

concentrations of IP to avoid the filter effect32 (from 9 to 200 ng ml21) and in all cases the signal of Flu in the binary mixture was significantly smaller than that of Flu in the absence of IP (with decreases of 3.1, 4.2, 8.0, 9.0 and 15.9%, respectively) at a significance level of 95% (for a one tailed test and n = 5) (all measurements were made on the second-derivative signal). However, previous studies on the determination of PAHs by the mentioned technique in samples in which IP and Flu occur simultaneously18 have never taken into account the possibility of quenching of the Flu signal. Because of this problem, we were unable to quantify Flu in the mixture. The resolution of 11 PAHs was achieved (10 of them belong to the EPA list). The compounds which were identified, the intervals where the resolution was achieved and the maxima and minima selected to obtain the quantification are given in Table 1. In the interval Dl = 95 nm, four compounds could be resolved (Ace, Chry, BaA, and BkFt). Hence we could identify and quantify several compounds with just one scan, which shortens the analysis time (Fig. 2). To quantify all the analyte compounds and because it is impossible to make the common measurement by plotting the tangent between two consecutive maxima and measuring the height from the half-point of the tangent to the minimum between the two selected maxima,13 the height between the maximum and the consecutive minimum selected in each case was used. It should be noted that the maxima and minima selected are those used to identify each compound. The limits of detection vary between 0.01 and 4.95 ng ml21, the limits of quantification are between 0.05 and 15.90 ng ml21 and the relative standard deviations (RSDs) range from 1.6% for BkFt to 8.0% for Phen (obtained for values of concentrations close to 20 ng ml21 in all cases and n = 10) (see Table 2). 2116

Analyst, 1998, 123, 2113–2117

Conclusions The resolution of 11 PAHs (10 of them belonging to the EPA list) was achieved in the working mixture of a solution of 18 PAHs in hexane. Using eight different Dl, 11 PAHs were identified and quantified. Limits of detection of around 0.1 ng ml21 and typical RSDs around 2% at concentrations in the linear range (for n = 10) were obtained. This method proved to be rapid, inexpensive, sensitive and simple when applied to the determination of PAHs. The identification of four different PAHs in the interval Dl = 95 nm significantly shortens the analysis time, so the identification and quantification of the 11 mentioned PAHs is possible in only about 10 min. On the other hand, we found that only the fluorescent signal of Flu appears quenched by another PAH (IP).

References 1 2 3 4 5 6 7 8 9 10 11 12

L. C. Sander, J. High Resolut. Chromatogr. Commun., 1988, 11, 383. M. T. Oms, R. Forteza, V. Cerdá, F. García and A. L. Ramos, Int. J. Environ. Anal. Chem., 1990, 42, 1. J. C. Means, S. G. Wood, J. J. Hasset and W. L. Banwart, Environ. Sci. Technol., 1980, 14, 1524. N. T. Edwards, J. Environ. Qual., 1983, 12, 427. S. O. Baek, R. A. Field, M. E. Goldstone, P. W. Kirk, J. N. Lester and R. Perry, Water Air Soil Pollut., 1991, 60, 279. C. A. Menzie, B. B. Potocki and J. Santodonato, Environ. Sci. Technol., 1992, 26, 1278. G. Kiss, Z. Varga-Puchony and J. Hlavay, J. Chromatogr. A, 1996, 725, 261 D. Domine, J. Devillers, P. Garrigues, H. Budzinski, M. Chastrette and W. Karcher, Sci. Total Environ., 1994, 155, 9. A. A. Herod and R. G. James, J. Inst. Fuel, 1978, 51, 164. S. A. Wise, L. C. Sander and W. E. May, J. Chromatogr., 1993, 642, 329. EPA Test Method, Polynuclear Aromatic Hydrocarbons–Method 610, US Environmental Protection Agency, Environmental Monitoring and Support Laboratory, Cincinnati, OH, 1982. G. Castello and T. C. Gerbino, J. Chromatogr., 1993, 642, 351.

13 14 15 16 17 18 19 20 21 22 23 24 25

M. Salgado, C. Bosch, F. Sánchez Rojas and J. M. Cano Pavón, Quím. Anal., 1986, 5, 374. J. B. F. Lloyd, Nature Phys. Sci., 1971, 231, 64. D. J. Futoma, S. R. Smith and J. Tanaka, CRC Crit. Rev. Anal. Chem., 1982, 13, 117. T. Vo-Dinh and P. R. Martínez, Anal. Chim. Acta, 1981, 215, 13. Z.-H. Zhao and W.-Y. Quan, J. Environ. Sci. (China), 1989, 1, 109. Ph. Baudot, M. L. Viriot, J. C. André, J. Y. Jezequel and M. Lafontaine, Analusis, 1991, 9, 85. J. J. Santana Rodríguez, Z. Sosa Ferrera, A. Afonso Perera and V. González Díaz, Talanta, 1992, 39, 1611. G. L. Green and T. C. O’Haver, Anal. Chem., 1974, 46, 2191. P. John and I. Soutar, Anal. Chem., 1976, 48, 520. J. B. F. Lloyd, Analyst, 1980, 105, 97. M. J. López de Alda, S. García-Falcón, M. A. Lage-Yusty and J. Simal-Lozano, J. Assoc. Off. Anal. Chem., 1995, 78, 495. D. G. Konstantianos and P. C. Ioannou, Analyst, 1996, 121, 909. S. A. Momin and R. Narayanaswamy, Analyst, 1992, 117, 83.

26 27 28 29 30 31 32 33

V. L. Amszi, I. Cordero, B. Smith, S. A. Tucker, W. E. Acree, Jr., C. Yang, E. Abu-Shaqara and R. G. Harey, Appl. Spectrosc. 1992, 46, 1156. S. A. Tucker, W. E. Acree, Jr., J. C. Fetzer, R. G. Garvey, M. J. Tanga, P. C. Cheng and L. T. Scott, Appl. Spectrosc., 1993, 47, 715. T. Vo-Dinh, R. B. Gammage, A. R. Hawthorne and J. H. Thorngate, Environ. Sci. Technology, 1978, 12, 1297. E. L. Inman, Jr. and J. D. Winefordner, Anal. Chem., 1982, 54, 2018. J. C. Andre, Ph. Baudot, M. Bouchy and M. Niclause, Anal. Chim. Acta, 1977, 92, 369. C. L. Stevenson and T. Vo-Dinh, Appl. Spectrosc., 1993, 47, 430. D. Rendell, Fluorescence and Phosphorescence, Wiley, Chichester, 1987, pp. 105–110. J. C. Miller and J. N. Miller, Statistics for Analytical Chemistry, Ellis Horwood, Chichester, 2nd edn., 1988, pp. 103–119.

Paper 8/04281K

Analyst, 1998, 123, 2113–2117

2117