Anal. Chem. 2005, 77, 7679-7686
Photocatalytic Degradation-Excitation-Emission Matrix Fluorescence for Increasing the Selectivity of Polycyclic Aromatic Hydrocarbon Analyses Yoon-Chang Kim, James A. Jordan, Michelle L. Nahorniak, and Karl S. Booksh*
Department of Chemistry and Biochemistry, Arizona State University, Tempe, Arizona 85287-1604
Polycyclic aromatic hydrocarbons (PAHs) constitute a wellknown priority pollutant group due to their carcinogenic and mutagenic effects in humans.1-3 PAHs are produced in natural and anthropogenic processes, and they are found in all kinds of biological and environmental areas.4,5 Consequently, the U.S. Environmental Protection Agency (EPA) has listed a number of PAHs to be considered as priority pollutants that should be monitored in the environment.6 There is a constant need to improve existing methods for PAHs, including, but not limited to, obtaining better selectivity and limits of detection, shortening the time required for an analysis, and making instrumentation capable of bringing the measurements into the field. Hyphenated instruments (e.g., liquid chromatography-UV diode array detection (LC-DAD), gas chromatography/mass spectrometry (GC/MS), and excitationemission fluorescence) can be among the most appropriate technologies to determine PAHs.5 Sample collection and handling must be considered in PAHs’ detection with these analytical methods. Because PAHs are hydrophobic, trace concentrations in water adsorb to the walls of most sample collection vessels,
making accurate quantification difficult as well as increasing the problem of sensitivity.7 In addition, multiple studies have demonstrated that PAHs undergo fairly rapid transformations when exposed to light in an aqueous medium.8 For routine analysis, chromatographic techniques (i.e., LC-DAD and GC/MS) are suitable, because it is a relatively simple measuring step and provides good sensitivity and selectivity. This reduction in time is usually achieved in the steps of sample pretreatment, preconcentration or cleanup, but can also be obtained in the measuring step.4,9,10 Fluorescence spectroscopy, because of its high sensitivity, avoidance of consumable reagents and adaptability to field measurements, is a powerful analytical technique for the analysis of PAHs in the environment. Although the selectivity of fluorescence spectroscopy is based on the fact that relatively few compounds show intrinsic fluorescence, and emission intensity depends on two variables, excitation and emission wavelengths, the wide application of fluorescence techniques for environmental monitoring has been limited by the lack of selectivity of fluorescence spectroscopy.6 Additionally, the ability to analyze multicomponent mixtures without separation procedures is extremely useful for routine analysis. Generally, a number of PAHs have fluorescence in the same excited region. Thus, a single-wavelength measurement has limited ability for analyzing complicated multicomponent PAH samples or even for spectra of simple mixtures, which are highly overlapping. A conventional fluorescence spectrum presents fluorescence only within one spectral region and does not provide enough data to distinguish between two or more closely related molecules. This disadvantage of a conventional fluorometer can be overcome by extending the dimensionality of the fluorescence measurements (i.e., video fluorometer) when followed by chemometric techniques. In an attempt to construct the video fluorometer, Christian et al. developed a device for simultaneous recording of fluorescence data in the form of an excitation-emission matrix (EEM).11 This yields a unique computer interface and a two-dimensional multichannel detector to
(1) Santodonato, J. Chemosphere 1997, 34, 835-848. (2) Arfsten, D. P.; Schaeffer, D. J.; Mulveny, D. C. Ecotoxicol. Environ. Saf. 1996, 33, 1-24. (3) Mu ¨ ncnerova´, D.; Augustin, J. Bioresour. Technol. 1994, 48, 97-106. (4) Mastral, A. M.; Callen, M. S. Environ. Sci. Technol. 2000, 34, 3051-3057. (5) JiJi, R. D.; Andersson, G. G.; Booksh, K. S. J. Chemom. 2000, 14, 171185. (6) JiJi, R. D.; Cooper, G. A.; Booksh, K. S. Anal. Chim. Acta 1999, 397, 6172.
(7) Hertz, H. S.; May, W. E.; Wise, S. A.; Chesler, S. N. Anal. Chem. 1978, 50, 428A-436A. (8) Botello, A. V.; Diaz, G. G.; Villanueva, S. F.; Salazar, S. L. Polycyclic Arom. Compd. 1993, 3, 397-404. (9) Wittkamp, B. L.; Hawthorne, S. B.; Tilotta, D. C. Anal. Chem. 1997, 69, 1197-1203. (10) Potter, D. W.; Pawliszyn, J. Environ. Sci. Technol. 1994, 28, 298-305. (11) Johnson, D. W.; Gladden, J. A.; Callis, J. B.; Christian, G. D. Rev. Sci. Instrum. 1979, 50, 118-126.
The application of photocatalysis enhancement to calibration of fluorescence excitation-emission matrixes (EEMs) with parallel factor (PARAFAC) analysis is described. In this study, three- and four-way PARAFAC analysis was employed to extract the fluorescent species’ spectra from overlapping EEMs. Time-dependent photocatalysis degradation of the polycyclic aromatic hydrocarbons (PAHs) was employed to create an additional dimension for analysis. The consequent four-dimension degradationEEM data cubes have greater selectivity for each PAH than do three-dimension EEM data cubes alone. On a scale of 0 to 1, with 0 being completely collinear spectra and 1 being orthogonal spectra, including the time-dependent measurements increased the selectivity an average of 21%, from 0.73 to 0.87.
10.1021/ac0509051 CCC: $30.25 Published on Web 10/29/2005
© 2005 American Chemical Society
Analytical Chemistry, Vol. 77, No. 23, December 1, 2005 7679
acquire fluorescence excitation and emission spectra simultaneously. Warner et al. developed procedures for qualitative12 and quantitative13 analyses with these EEMs. For quantitative multicomponent analysis, Davidson et al. have demonstrated a method of rank annihilation to determine several known species of interest in the presence of other unknown emitters.14 The improvement in selectivity of the video fluorometer has been exploited by the application of the parallel factor (PARAFAC) model with threeand four-way EEM data.6,15-18 Along with providing selectivity and sensitivity, the video fluorometer is an instrument that is capable of generating three dimensions (i.e., excitation, emission, and concentration), thus broadening the opportunity to utilize PARAFAC techniques for signal deconvolution. In PARAFAC, given a number of factors (related with the number of fluorescent species in sample), two vectors related with the excitation and emission spectra and one scalar (i.e., the concentration) are found for each factor. The video fluorometer can easily apply with adding the other scalar, time-dependent profiles as a fourth dimension. The chemical reactions involving degradation of target fluorescent species may provide a fourth dimension of analysis with selectivity appropriate for resolution by four-way PARAFAC.18 Analytical figures of merit, such as sensitivity, selectivity, limit of detection, and net analyte signal are commonly used as transferable metrics for performance characterization of instrumental methods. These common figures of merit are mathematically defined for univariate, multivariate, and multiway data instrumentation.19,20 Selectivity is defined as the ratio of the net analyte signal (NAS) and the gross analyte signal magnitudes.19,20 Although there have been competing definitions of exactly what constitutes “net” analyte signal in multiway calibration14,20-22 the definition by Messick, Kalivas, and Lang22 has been found to be the most straightforward extension of Lorber’s original definition23 of multivariate NAS.20 Simply put, the NAS of an analyte is the part of the analyte signal that is orthogonal to the signal from all interfering species, + NASA ) (I - R-aR-a )ra
(1)
where the vector ra is the two-dimension EEM profile or the threedimension EEM-time decay profile of analyte A unfolded into a vector, the matrix R-a is a collection of all other spectral profiles unfolded in the same manner as ra and appended column-wise, I (12) Warner, I. M.; Christian, G. D.; Davidson, E. R.; Callis, J. B. Anal. Chem. 1977, 49, 564-573. (13) Warner, I. M.; Davidson, E. R.; Christian, G. D. Anal. Chem. 1977, 49, 2155-2159. (14) Ho, C. N.; Christian, G. D.; Davidson, E. R. Anal. Chem. 1980, 52, 10711079. (15) Pen ˜a, A. M. d. l.; Mansilla, A. E.; Go´mez, D. G.; Olivieri A. C.; Goicoechea, H. C. Anal. Chem. 2003, 75, 2640-2646. (16) Tan, Y. X.; Jiang, J.-H.; Wu, H.-L.; Cui, H.; Yu, R.-Q. Anal. Chim. Acta 2000, 412, 195-202. (17) Moberg, L.; Robertsson, G.; Karlberg, B. Talanta 2001, 54, 161-170. (18) Nahorniak, M. L.; Cooper, G. A.; Kim, Y.-C.; Booksh, K. S. Analyst 2005, 130, 85-93. (19) Booksh, K. S.; Kowalski, B. R. Anal. Chem. 1994, 66, 782A-791A. (20) Faber, K.; Lorber, A.; Kowalski, B. R. J. Chemom. 1997, 11, 419-461. (21) Wang, Y. D.; Borgen, O. S.; Kowalski, B. R.; Gu, M.; Turecek, F. J. Chemom. 1993, 7, 117-130. (22) Messick, N. J.; Kalivas, J. H.; Lang, P. M. Anal. Chem. 1996, 68, 15721579. (23) Lorber, A. Anal. Chem. 1986, 58, 1167-1172.
7680
Analytical Chemistry, Vol. 77, No. 23, December 1, 2005
is an identity matrix of appropriate dimension, and the superscript + indicates the Moore-Penrose pseudoinverse. The selectivity (SEL) is, thus, the ratio of the norm of the NASA divided by the norm of ra,
SELA ) ||NASA||/||ra||
(2)
More details regarding the determination of NAS and SEL can be found in the critical review by Faber et al.20 In this paper, the NAS plays an important role in the calculation of selectivity for characterizing the difference between EEM and DEEM. The larger the NAS and, hence, the greater the selectivity, the greater the power of an instrumental method to differentiate between similar compounds. The NAS also is also a factor in the instrumental signal-to-noise ratio and the limit of detection (see equations in ref 19). The larger the NAS, the better the signalto-noise ratio and limit of detection for a particular analysis. Here, the NAS is calculated from the resolved spectral profiles following analysis by PARAFAC applied to a series of raw EEM spectra (i.e., two-dimension EEM profile or the three-dimension EEM-time decay profile). In recent years, the application of photocatalytic degradation in sensor development has received much attention.24-26 The investigations have exploited that many n-type semiconductors in aqueous solution, under light irradiation and with higher energy than that of the band gab of a semiconductor, will induce catalytic decomposition and often complete mineralization of organic compounds. The main mechanism of the photocatalytic destruction process is the photogeneration of electrons and holes in the semiconductor. In heterogeneous photocatalysis, a suspension of semiconductor particles is irradiated with natural or artificial UV light. The excitation transfer of an electron from the valence to the conduction band creates an oxidizing site (a “hole”, h+ VB) and a reducing site (an “electron”, eCB). The photogenerated hole has the potential to oxidize a variety of substrates by means of electron transfer. After the direct electron transfer between the organic molecule and the excited semiconductor or after processes mediated by adsorbed OH radicals, the organic radical that is formed may add O2, forming more oxygenated compounds on the route to CO2 formation. In many cases, titanium dioxide (TiO2) can be used as a photocatalyst for the mineralization of the PAHs because of nonphotocorrosive and nontoxic behavior in aquatic systems. TiO2 is also capable of photooxidative destruction of most organic pollutants. Additionally, it has been demonstrated that in the photocatalytic process, the PAH material is converted to carbon dioxide followed by oxygen consumption. The rate of TiO2 photocatalytic degradation of PAHs varies from one compound to another.27 In this study, an application of the photocatalysis to the excitation-emission matrix fluorometer (EEM) through the PARAFAC analysis is described. The aim of this study is to (24) Kim, Y.-C.; Lee, K.-H.; Sasaki, S.; Ikebukuro, K.; Hashimoto, K.; Karube, I. Anal. Chem. 2000, 72, 3379-3382. (25) Kim, Y.-C.; Sasaki, S.; Yano, K.; Ikebukuro, K.; Hashimoto, K.; Karube, I. Analyst 2000, 125, 1915-1918. (26) Kim, Y.-C.; Sasaki, S.; Yano, K.; Ikebukuro, K.; Hashimoto, K.; Karube, I. Anal. Chem. 2002, 74, 3858-3864. (27) Ireland, J. C.; Da`vila, B.; Moreno, H.; Fink, S. K.; Tassos, S. Chemosphere 1995, 30, 965-984.
Figure 1. Schematic diagram of the apparatus. 1, 75-W xenon lamp; 2, spectrometer for excitation; 3, spectrometer for emission; 4, flow cell; 5, sample chamber; 6, CCD camera; 7, photochemical column consisting of TiO2 beads; 8, UV lamp (λmax ) 365 nm); 9, reflector; 10, peristaltic pump; and 11, sample reservoir.
improve selectivity of the EEM-PARAFAC fluorometer through application of the photocatalysis to collect degradation-excitationemission matrix (DEEM) data cubes, thereby improving analyte identification and quantitation in multicomponent mixtures. Consequently, the improvement in selectivity can be derived from the inclusion of the different photocatalytic-degradation rates of PAHs as an additional scalar term into the conventional PARAFAC analysis. The fluorometer and photocatytic chamber employed here are all field-portable, battery-powered units assembled with the idea of performing simple quantitative screening analyses in the field where more sophisticated laboratory methods are unavailable. EXPERIMENTAL SECTION Reagents and of Standards Preparation. Titanium dioxide beads (ST-B11) used in this study were obtained from Ishihara Sangyo, Ltd. EPA610 polynuclear aromatic hydrocarbon mixture, and 5% dimethyldichlorosilane were purchased from Supelco. Benzo[a]anthracene (BaA), benzo[k]fluoranthene (BkF), and dibenzo[a,h]anthracene (B2ahA) were purchased from Sigma, Aldrich, and Restek, respectively. All PAHs used were reagent grade. These reagents were used without further purification and were dissolved and diluted in low acetone and anhydrous methanol (Mallinckrodtft AR) to make stock solutions. The stock solutions were wrapped in aluminum foil and kept refrigerated when not in use. The standard solutions were made by the diluting stock solutions in deionozed water to 1 ppb PAH concentration (1 ng PAH/mL). The deionozed water used was purified in a Millipore MilliQ system, and the standard solutions were prepared daily. The 16-PAH standard sample employed for optimization of the flow rate through the catalytic bed was purchased from Supelco. The stock sample ranged from 94 to 2021 µg/mL in 1:1 methanol/ methylene chloride. Via serial dilution, the 16-PAH stock solution was diluted by 1 part in 2 × 106 parts distilled water to create a solution with individual PAH concentrations between 1 ppb and 4.8 ppt. The total PAH concentration in the final 16-PAH solution was 5.88 ppb (ng/mL). All experiments were performed on the day of standard solution preparation. Instrumentation. The setup of the flow system for determination of PAHs is shown in Figure 1. The PAH solutions were propelled by a peristaltic pump (Miniplus3, Gilson) through a flow
cell and a photochemical column. The photocatalytic column was made from a quartz tube (inside diameter, 0.2 cm; outside diameter, 0.4 cm) filled with 0.6-mm-diameter TiO2 beads held in position with Teflon wool plugs. The total internal volumes of the photocatalytic reactor, cuvette, and tubing are approximately 1, 5, and 4 mL, respectively. The quartz tube and flow cell were rinsed with 5% dimethyldichlorosilane for 10-15 s and then rinsed with toluene and methanol. A UV irradiator with a 15-W UV lamp (a common water purification system for home aquariums) was located 0.5 cm from the side of the photocatalytic column in the UV illumination experiment. The EEM fluorometer consisted of a 75-W xenon lamp (XBO 75W/2, OSRAM), a sample chamber (model SC-447, Acton Research Corporation), two Spectra Pro-150 imaging spectrometers (focal length of 150 mm (f/4 optics), spectral resolution of 0.4 nm, and dispersion of 5 nm/mm, Acton Research Corporation), and a low-resolution imaging charge-coupled device (CCD) (ST-6B, Santa Barbara Instrument Group) with a 375 × 241 pixel resolution, a 23 × 27 µm pixel size, and a total area of 8.6 mm × 6.5 mm. A 75-Watt xenon lamp was focused onto an imaging spectrograph with the exit slit rotated 90°. This produced an incident spectrum such that the excitation photons were spatially dispersed vertically across the flow cell in a sample chamber. Light emitted 90° from incident was focused onto the slit of the entrance slit of a second imaging spectrograph. The excitation resolution was maintained while the emission spectra were focused onto the low-resolution imaging CCD. An entire EEM spectrum was collected on a laptop/desktop computer using KestrelSpec software from Catalina Scientific. The EEM has been adapted to run off DC power and, thus, can be field-portable. The 75-W Xe lamp requires 14 V, drew 5 A/h, and has been powered in the field for 15 h using two 12-V, 75 A/h deep cycle batteries.28 RESULTS AND DISCUSSION Operating Flow Rate and Three-Component Mixture. The 16-PAH standard was used in an effort to select a flow rate that would be useful for PAH mixtures in general, and the 3-component mixture (BaA, BkF, and B2ahA) was selected for the specific selectivity exercise described here. The effect of flow rate through the catalytic bed was determined using a set of 16 PAHs ranging from 4.8 ppt to 1.0 ppb. Figure 3 shows the time-dependent change in relative intensities with different flow rate of the 16-PAH mixing sample. The final relative intensity decreased with flow rate across the tested range of 0.19-0.70 mL/min. It can be seen that the absolute decomposition ratio of the 16-PAH mixing sample obtained using this system was ∼67% of the original fluorescence intensity once the mixture hit steady state in the cuvette. The proscribed analysis times of competing EPA methods were employed as a guide for choosing the slowest acceptable total analysis time for the field portable EEM-photolysis system. EPA method 8310 describes extraction techniques followed by liquid chromatography with ultraviolet/visible or fluorescence spectroscopy for the detection of parts-per-billion levels of certain PAHs.29 The retention times using this method range from 16 to 37 min. (28) JiJi, R. D.; Nahorniak, M. L.; Fruitman, E.; Booksh, K. S. SPIE Proc. 1999, 3856, 73-82. (29) In EPA Method 8310, Polynuclear Aromatic Hydrocarbons. 2001, Spectrum Laboratories Web site; http://www.speclab.com.
Analytical Chemistry, Vol. 77, No. 23, December 1, 2005
7681
Figure 2. Actual EEM spectra of three PAHs and the mixtures in the range 275-325 nm excitation and 340-500 nm emission. (a) Pure BaA (2 ppb); (b) pure BkF (2 ppb); (c) B2ahA (2 ppb); (d) mixture of BaA (1 ppb) and BkF (1 ppb); (e) mixture of BaA (1 ppb) and B2ahA (2 ppb); (f) mixture of BkF (1 ppb) and B2ahA (2 ppb); and (g) mixture of BaA (2 ppb), BkF (2 ppb), and B2ahA (2 ppb).
Figure 3. Influence of flow rate through catalytic reactor on degradation of PAH mixture. 16-PAH solution (4.8 ppt-1.0 ppb); CCD’s exposure time, 90 s; and working range, 275-325 nm excitation and 340-500 nm emission.
EPA method 8100 involves extraction techniques followed by gas chromatography utilizing a flame ionization detector for PAH detection.30 The retention times range from 10 to 20 min. Therefore, the slowest desired flow rate is at least 0.5 mL/min in this system to achieve near complete photocatalytic degradation 7682
Analytical Chemistry, Vol. 77, No. 23, December 1, 2005
within 15 min. Consequently, the 0.54 mL/min flow rate was employed in future samples. For the specific selectivity exercise, BaA, BkF, and B2ahA were selected among the potential PAHs because of the overlapping EEM spectra of these three analytes. BaA (Figure 2a), BkF
Figure 4. Time profiles using photocatalytic PARAFAC analysis. (a) B2ahA in a mixture of BkF and B2ahA; (b) pure B2ahA; (c) BaA in a mixture of BaA and BkF; (d) pure BaA; (e) pure BkF; (f) BkF in a mixture of BaA and BkF; and (g) BkF in a mixture of BkF and B2ahA.
(Figure 2b), and B2ahA (Figure 2c) were found to have excitation/ emission maximum centers at approximately 280 nm/390 nm, 304 nm/420 nm, and 290 nm/400 nm, respectively. Therefore, all three PAHs exhibit fluorescence in the range 275-325 nm excitation and 340-500 nm emission. For display purposes, the intense Rayleigh scattering has been truncated from the upper left corner of the image. The BkF and B2ahA spectra could not be distinguished in a binary mixture because the pure spectra of BkF and B2ahA are highly overlapped and the B2ahA is much less fluorescent than BkF (Figure 2d). From the spectra of the BaA and B2ahA binary mixture, it can be seen that the B2ahA has a red-shifted feature due to excited-state complex (exciplex) formation of BaA and B2ahA. This exciplex was found to be present only with B2ahA and BaA, because B2ahA had no shift with BkF (Figure 2e); neither did BaA with BkF. The collision-deactivation of excited states of B2ahA and BaA may occur by the close proximity of the two PAHs. The new emission was assigned to B2ahA-BaA complexes formed in guest-induced defects of the host lattice. A contour plot presenting the degree of overlap among the three analytes is shown in Figure 2g. The fluorescence due to the exciplex behavior is found on the right. Time Profiles by Photocatalytic Degradation. Photocatalytic degradation of both standards and mixtures of 1 ppb PAH solutions have shown the ability to increase the selectivity of EEMbased analysis. For every sample analyzed, a 0.54 mL/min flow rate was employed while 10 EEM spectra were collected with a 90-s integration time. Photocatalytic degradation was allowed to continue for 40 min while spectra were collected, although no
further photodegradation was observed after 20 min. The results are presented in Figure 4. The rate of photodegradation was unique for each of the three compounds in the set. In addition, the rate of degradation did not significantly change with the presence of other PAHs in the mixtures. Analysis of the collection of DEEM spectra by four-way PARAFAC analysis enabled the extraction of three reasonable PAH profiles (Figure 5) that agreed qualitatively with the EEM spectra of the 1 ppb single-component standards. A two factor model was found to work best for feature extraction of the twocomponent mixtures when the intense Rayleigh line was either truncated or assigned a weight of zero in the least-squares fit.28,31 It is important to note that extraction of the weakly fluorescent B2ahA was not possible from three-way PARAFAC of the EEM spectra but was accomplished by four-way analysis of the DEEM spectra. This is due in large part to the increased selectivity of the DEEM spectra, as compared to the EEM spectra. Reanalysis of the data employing only the first five EEM spectra in the DEEM data cube (i.e., the first 20 min of analysis) yielded qualitative equivalent results. For the selectivity study, we selected a set of correlations between each of the analytes and their mixtures, as detailed in Table 1. The improvement from EEM to DEEM in the BaA and BkF has negative values. Because the Ex/Em center of the BaA is located far enough away from the Ex/Em center of the BkF, (30) In EPA method 8100, Polynuclear Aromatic Hydrocarbons, method description sheet. 2001, Spectrum Laboratories Web site; http://www.speclab.com. (31) JiJi, R. D.; Booksh, K. S. Anal. Chem. 2000, 72, 718-724.
Analytical Chemistry, Vol. 77, No. 23, December 1, 2005
7683
Figure 5. Predicted excitation and emission profiles of (a) BaA, (b) BkF, and (c) B2ahA using only photocatalytic data of mixtures, a + b and b + c with a two-factor PARAFAC model. Table 1. Comparison of EEM and DEEM with the Selectivitya Based on PARAFAC
BaA BkF B2ahA allb
BaA
BkF
B2ahA
0.914, 0.863 (-6%) 0.700, 0.994 (42%) 0.676, 0.848 (25%)
0.914, 0.863 (-6%) N/A 0.712, 0.999 (40%) 0.886, 0.930 (5%)
0.700, 0.994 (42%) 0.712, 0.999 (40%) N/A 0.632, 0.846 (34%)
N/Ac
a SEL data [EEM, DEEM (improving rate %)] calculated from eqs 1 and 2. b Data set of all analytes and their mixtures. c N/A, not applicable.
the BaA and BkF could be easily resolved by the regular EEM_PARAFAC algorithm. In the total data set (ALL, Table 1) including mixtures’ profiles, DEEM increases the average selectivity across all three analytes by 21%, as compared to regular EEM analysis. The selectivity of BaA increases by 25% to 0.848 from 0.676. BkF, easily the most spectrally unique of the three analytes, increases by only 5% from 0.886 to 0.930. The selectivity of the weakly emitting B2ahA increases by 34% from 0.632 to 0.846. B2ahA and BaA are the most heavily overlapping of the three fluorophores. How inclusion of the third order of selectivity with DEEM improves the ability of PARAFAC to resolve all three species is evident in analysis of the NAS for the DEEM data. The weakly fluorescent B2ahA is both highly overlapped with BaA and also the slowest to photodegrade. These facts are loaded into the NAS (Figure 6a) by having the first two EEM slices of the DEEM cube have a large anticorrelation factor to BaA where the BaA fluorescence is relatively the strongest. This is seen by the negative regions in the NAS. Concurrently, the NAS of B2ahA actually increases in the later time slices of the DEEM data cube, 7684 Analytical Chemistry, Vol. 77, No. 23, December 1, 2005
even though the fluorescence of B2ahA decreases as the PAH degrades. This stems from the fact that the relative B2ahA fluorescence is increasing, as compared to BaA and BkF. Concurrently, just the opposite effect is observed with the strongly overlapping BaA (Figure 6b). The NAS of BaA is largest in the first few time slices of the DEEM data cube, then as BaA more rapidly degrades, the NAS rapidly shifts to account for the B2ahA fluorescence in the later time slices. B2ahA and BaA gather much increase in the selectivity by the ability to parse the regions of more unique analysis in this method. At the same time, BkF is the least overlapping of the three fluorophores and realizes the least improvement in the selectivity by employing DEEM analysis. Thus, the NAS associated with each of the first five time slices in the DEEM changes little in shape (Figure 6c). Only the magnitude of the NAS significantly changes as the BkF degrades. Implications for Instrumental Design. The seemingly incomplete photocatalysis of the 16-PAH mixture and subsequent single component standards and binary mixtures was found to be an artifact of the instrumental design; this fluorescent baseline is assumed to be PAHs adsorbed onto the quartz cuvette place at the end of the photoreactor. Prior to UV irradiation, the entire system is flushed with methanol to flush out the previous sample then flushed with the new sample to fill the system. During this process, there is ample time for the dilute PAHs to adsorb onto the quartz walls. Final analysis of the data showed that only one cycle of the solutions through the reactor is sufficient to entirely degrade all three PAHs. Thus, the assumed rate of degradation is actually a convolved rate of mixing and degradation. Consequently, the closed loop, multipass reactor, as designed, is unnecessarily
Figure 6. (a) Net analyte signal based on DEEM analysis of B2ahA. (b) Net analyte signal based on DEEM analysis of BaA. (c) Net analyte signal based on DEEM analysis of BkF.
Analytical Chemistry, Vol. 77, No. 23, December 1, 2005
7685
complex. A linear, one-pass reactor would be sufficient to collect DEEM spectra. Analysis would be done in a segmented flow protocol in which the system is flushed and filled with the UV radiation turned off. With a flowing stream, the UV radiation would be reintroduced, and DEEM spectra, collected. The cycle would be repeated for each sample. This segmented flow analysis would be more easily miniaturized, more rapid, and would mitigate the absorbance on the cuvette prior to analysis. CONCLUDING REMARKS The methodology followed in this study is based on the photocatalytically induced kinetic changes in fluorescence profiles of selected PAHs. The EEM fluorometer shows promise for
7686
Analytical Chemistry, Vol. 77, No. 23, December 1, 2005
development of DEEM as a photocatalysis-based method for determining PAHs. The time profiles yield a dimension to the data that increases selectivity of the EEM-PARAFAC fluorescence spectroscopy. ACKNOWLEDGMENT This work was supported by funding from an NSF grant (OCE 0119999).
Received for review May 23, 2005. Accepted September 23, 2005. AC0509051
