Anal. Chem. 1998, 70, 4104-4110
Application of Continuous-Flow HPLC-Proton-Nuclear Magnetic Resonance Spectroscopy and HPLC-Thermospray-Mass Spectroscopy for the Structural Elucidation of Phototransformation Products of 2,4,6-Trinitrotoluene Markus Godejohann, Michael Astratov, Alfred Preiss,* and Karsten Levsen
Fraunhofer-Institut fu¨ r Toxikologie und Aerosolforschung, Nikolai-Fuchs-Strasse 1, D-30625 Hannover, Germany Clemens Mu 1 gge
Humboldt-Universita¨ t zu Berlin, Institut fu¨ r Chemie, Hessische Strasse 1-2, D-10115 Berlin, Germany
An aqueous solution of 2,4,6-trinitrotoluene (TNT) was irradiated by natural sunlight for a period of 1 month to generate phototransformation products of this compound. After solid-phase extraction on a poly(styrene-divinylbenzene) copolymer at pH 1, the structures of several acidic nitroaromatic compounds were identified by means of continuous-flow HPLC/1H NMR and HPLC/TSP-MS measurements of this extract. By interpretation of both NMR and MS spectra, it was even possible to characterize noncommercially available phototransformation products of TNT. The results obtained by continuous-flow HPLC/ 1H NMR were compared with those obtained by the investigation of a groundwater sample of a former ammunition site near Elsnig, Germany. The results show that several identified phototransformation products of TNT are also present in this groundwater sample. Chemicals released into the environment may undergo chemical, photochemical, and microbiological transformation processes leading to complex mixtures of organic compounds. While analyzing such environmental samples, the following difficulties are encountered: (a) the development of a common sample extraction and preparation step adapted to the chemical properties of each individual component is difficult or even impossible; (b) often the separation efficiency of the analytical method applied to such complex mixtures is insufficient; (c) in general, the detectors employed do not provide sufficient structural information necessary for the identification of unknowns; (d) for many transformation products, the corresponding reference compounds are not available. These problems are less severe for compounds amenable to GC, as the GC/MS method combines the high separation efficiency of the GC with the structure specificity of the MS, and extensive MS libraries are available for the identification of unknowns. However, the unequivocal identification of unknown polar thermally labile compounds in environmental samples is 4104 Analytical Chemistry, Vol. 70, No. 19, October 1, 1998
much more difficult. The separation efficiency of the LC method usually employed is often insufficient, while the commonly used photodiode array detector (PDA) provides limited structrual information. For these polar compounds, the use of special hyphenated techniques such as HPLC/MS or HPLC/NMR is promising. HPLC/MS has been applied repeatedly in environmental analysis, while the HPLC/NMR method was only recently and for the first time applied by us to the analysis of environmental samples.1 Both methods complement each other well: while the HPLC/MS provides important information on the molecular weight and the presence of certain functional groups, HPLC/NMR gives structural information, in particular on the constitution and the isomeric substitution of unknown molecules. Moreover, as a result of the very narrow resonance signals in modern high-field NMR, the resolution of information is also improved considerably and is comparable to that of GC/MS.2 It is of further advantage that the continuous flow HPLC/NMR method allows the quantification of analytes even in the absence of reference compounds. In this study, the combined HPLC/MS and HPLC/NMR approach is used to identify phototransformation products of 2,4,6trinitrotoluene (TNT). For the HPLC/NMR investigation, the continuous-flow mode has been employed, which has some advantages for our approach in comparison to the stop-flow technique. Only in this mode it is possible to obtain an NMR chromatogram of the sample. The contour plot of the NMR chromatogram provides a good overview of the compounds in the sample (down to the microgram range) and allows the simultaneous analysis of their chemical shifts and retention times. Therefore, this technique is particular useful for the rapid comparison of complex mixtures. (1) Godejohann, M.; Preiss, A.; Mu ¨ gge, C.; Wu ¨ nsch, G. Anal. Chem. 1997, 69, 3832-3837. (2) Preiss, A.; Levsen, K.; Humpfer, E.; Spraul, M. Fresenius J. Anal. Chem. 1996, 356, 445-451. S0003-2700(98)00292-3 CCC: $15.00
© 1998 American Chemical Society Published on Web 08/26/1998
Figure 1. Fate of TNT by means of photo- and biotransformation in munition wastewater proposed by Spanggord et al.4 Abbreviations: 2-A4,6-DNT, 2-amino-4,6-dinitrotoluene; 4-A-2,6-DNT, 4-amino-2,6-dinitrotoluene; 2,4,6-TNT, 2,4,6-trinitrotoluene; 2,4,6-TNBOH, 2,4,6-trinitrobenzyl alcohol; 1,3,5-TNB, 1,3,5-trinitrobenzene; 3,5-DNA, 3,5-dinitroaniline; 2-A-4,6-DNBA, 2-amino-4,6-dinitrobenzoic acid; 2-OH-4,6-DNBA, 2-hydroxy4,6-dinitrobenzoic acid; 3,5-DNP, 3,5-dinitrophenol; 2,4-DNBA, 2,4-dinitrobenzoic acid.
The phototransformation of TNT is a well-known process leading to highly polar, water-soluble nitroaromatic compounds.3 This fast reaction can be observed by irradiation of an aqueous solution of TNT with UV or natural sunlight: the formerly colorless solution turns immediately to pink. Since the early 1980s, the phototransformation of TNT has been investigated by several research groups in the United States. During these studies, some major transformation products were identified, first by spectroscopic investigation and then by the synthesis of the reference compounds. Spanggord et al. proposed that photolysis of TNT in ammunition wastewater leads to 2,4,6-trinitrobenzoic acid (2,4,6-TNBA) via the formation of 2,4,6-trinitrobenzyl alcohol (2,4,6-TNBOH) and 2,4,6-trinitrobenzaldehyde.4 Decarboxylation of this acid leads to 1,3,5-trinitrobenzene (1,3,5-TNB). In addition, this group found 2-amino-4,6-dinitrobenzoic acid (2-A-4,6-DNBA) and tentatively identified 2-hydroxy-4,6-dinitrobenzoic acid (2-OH(3) Burlinson, N. E.; Sitzman, M. E.; Kaplan, L. A.; Kayser, E. J. Org. Chem. 1979, 44, 3695-3698.
4,6-DNBA), both representing further important intermediate products of the phototransformation of TNT. Decarboxylation of these compounds leads to 3,5-dinitroaniline (3,5-DNA) and 3,5dinitrophenol (3,5-DNP). According to Spanggord, the loss of the amino group of 2-A4,6-DNBA results in 2,4-dinitrobenzoic acid (2,4-DNBA). Decarboxylation of this compound gives rise to 1,3-dinitrobenzene (1,3DNB). Most of these transformation products have also been found by us in high concentrations in groundwater at the former ammunition site Elsnig (Saxony, Germany).1,5 A schematic representation of the proposed fate of TNT in ammunition wastewater according to Spanggord et al.4 is given in Figure 1. (4) Spanggord, R. J.; Mabey, W. R.; Mill, T.; Chou, T.-W.; Smith, J. H.; Lee, S.; Roberts, D., Report AD-A138550. Environmental fate studies on certain ammunition wastewater constituents: Phase IV. Lagoon Model Studies U.S. Army Medical Research and Development Command, Fort Detrick, Frederick, MD, 1984. (5) Preiss, A.; Lewin, U.; Wennrich, L.; Findeisen, M.; Efer, J. Fresenius J. Anal. Chem. 1997, 357, 676-680.
Analytical Chemistry, Vol. 70, No. 19, October 1, 1998
4105
I
II
III
Figure 2. Formation and decarboxylation of 2,2′-dicarboxy-3,3′,5,5′-azoxybenzene according to Kaplan et al.6
In addition to the compounds mentioned above, Kaplan et al. also identified several carboxylated diazo- and azoxybenzenes as phototransformation products of TNT after irradiation of an aqueous TNT solution with a mercury lamp, equipped with a Pyrex filter.6 Decarboxylation of these compounds leads to the nonionic compounds II and III as shown in Figure 2. The identification of phototransformation products of TNT by liquid/liquid extraction, thin-layer chromatography, and subsequent off-line spectroscopic methods as reported by Kaplan et al.6 is very tedious and did not allow an unambiguous structure assignment in each case. We report here that these transformation products may be identified unequivocally and much more effectively by means of continuous-flow HPLC/1H NMR and HPLC/thermospray-MS after solid-phase extraction of the analytes. Transformation products of TNT were also produced by irradiation of an aqueous solution of the compound with natural sunlight. In a second step, this method was applied to the identification of the degradation products in an aqueous sample from a former ammunition hazardous waste site. Due to the fact that most of these compounds are not commercially available or are extremely labile (e.g., 2,4,6-trinitrobenzoic acid), routine measurements of the groundwater of former ammunition sites in general do not cover these phototransformation products. EXPERIMENTAL SECTION Instruments and Methods. Continuous-Flow HPLC/1H NMR. The extracts of the irradiated TNT solution and the groundwater sample were freeze-dried and diluted with the mobile phase that consisted of a methanol/D2O mixture (45%/55% v/v) buffered to a pH value of 2.3 with a dihydrogen phosphate/ (6) Kaplan, L. A.; Burlinson, N. E.; Sitzmann, M. E., Report NSWC/WOL/TR 75-152. Photochemistry of TNT: Investigation of the “Pink Water” problem, Part II. Naval Surface Weapons Center White Oak Laboratory, Silver Spring, MD, 1975.
4106 Analytical Chemistry, Vol. 70, No. 19, October 1, 1998
phosphoric acid buffer. Chromatographic separation was carried out on a Merck (Darmstadt, Germany) LiChrospher 60 RP Select B column (125-mm length, 4-mm i.d., 5-µm particle size) at a flow rate of 0.017 mL/min. The injection volume was 500 µL for the photoextract and 400 µL for the groundwater extract (injection valve with a 1-mL sample loop from Rheodyne (Cotati, CA; model 7725i). For the delivery of the mobile phase, an HPLC pump from Bischoff (Leonberg, Germany), model 2250, equipped with micropump heads was used. Pseudo-2D NMR spectra were recorded on a spectrometer from Bruker (Rheinstetten, Germany), model AMX 600, at 600.13 MHz, equipped with a 1H-13C inverse flow probe (4-mm i.d. of measuring cell with a detection volume of 120 µL). Solvent suppression was done by application of a 1D version of the NOESY pulse sequence (RD-90°-t1-90°-tm-90°-FID) with presaturation during relaxation delay (RD) and mixing time (tm) simultaneously on two frequencies. The t1 value was not incremented and was set to 3 µs. The pseudo-2D spectra were recorded within 64 rows, each row consisting of 128 free induction decays (FIDs) (sweep width 14706 Hz) collected into 32k data points with a relaxation delay of 10 s and a flip angle of 90°. Data were multiplied with an exponential function in f2 (LB ) 1 Hz) and processed with the XWINNMR software from Bruker (Rheinstetten, Germany). 1H NMR chemical shift values are referenced to methanol, which was set to 3.30 ppm. HPLC/Thermospray-MS. The mobile phase consisted of a methanol/water mixture (50/50 v/v). To perform ion-suppression chromatography, the water was fortified with 0.2% formic acid. Separation was carried out on a Merck LiChrospher 100 C18 column (250-mm length, 4-mm i. d., 5-µm particle size) at a flow rate of 0.5 mL/min. Postcolumn, an aqueous solution of 0.175 mol/L ammonium formate was added to the eluent at a flow rate of 0.5 mL/min to perform discharge-assisted buffer ionization.
The injection volume of the pure acetonitrile extract was 20 µL, using an injection valve from Rheodyne, model 7125. The mobile phase was delivered to the column using an HPLC pump from Varian (Palo Alto, CA), model 5000. Thermospray spectra were acquired using a mass spectrometer from Finnigan MAT (San Jose´, CA), model 4500, equipped with a thermospray interface from Vestec (Houston, TX). The conditions of the TSP interface were as follows: vaporizer temperature ∼140 °C; source temperature 250 °C. The mass spectrometer was operated in the negative ion detection mode at a scan rate of 4 s/scan over the range of m/z 120-500. Raw data were processed with software provided by MasCom (Bremen, Germany). Reagents. 2,4,6-Trinitrotoluene (>99%) and deuterium oxide (99.9%) were from Promochem (Wesel, Germany). Anhydrous disodium hydrogen phosphate, phosphoric acid (85%), formic acid (99%), and ammonium formate were from Merck. Water was obtained from a Milli-Q water purification system from Millipore (Milford, MA). Methanol was purchased in HPLC grade from Mallnickrodt Baker (Deventer, The Netherlands). Safety Considerations. 2,4,6-Trinitrotoluene and related nitroaromatic compounds are toxic and in part carcinogenic and mutagenic and should therefore be handled with special care. Samples. To generate phototransformation products of TNT, 1 g of the pure material was dissolved in 10 L of Milli-Q water by application of ultrasonic irradiation. Sodium azide (500 mg) was added to this aqueous solution to prevent microbial degradation of TNT and its phototransformation products. The groundwater sample was taken from a former ammunition site near Elsnig, Saxony, Germany. To prevent microbial degradation during storage, sodium azide was added again. Procedures. The aqueous TNT solution was irradiated in an open glass vessel for a period of 1 month by natural sunlight during the summer (July-August). Prior to solid-phase extraction, the solution was passed through a paper filter. The solidphase extraction procedure for both the irradiated solution and the groundwater sample is described elsewhere.1 RESULTS AND DISCUSSION HPLC/TSP-MS. Figure 3 shows the total ion current (TIC) chromatogram obtained after the injection of 20 µL of the original extract of the irradiated TNT solution. In Figure 4, the mass spectra of the main peaks are represented. These figures show that, in addition to TNT with a retention time of ∼20 min (peak e) and a molecular mass of 227 (cf. Figure 4e), several other compounds are present in this extract. The mass spectra in Figure 4d and Figure 4f and the corresponding retention times in Figure 3 (peaks d and f) are consistent with those of a trinitrobenzene (d) and a dinitrophenol (f). At a retention time of ∼24 min (peak g), a compound with an m/z ratio and isotopic distribution of a chlorodinitrobenzene appears. This compound might be generated by a so-called “ipso” substitution, where the substituent of a nitroaromatic compound (e.g., a dinitrophenol) is replaced by another nucleophile (e.g., by chloride ions which are present at a high concentration in the aqueous phase after adjusting the pH with hydrochloric acid). However, this compound cannot be generated by phototransformation of TNT.
Figure 3. Total ion current chromatogram obtained after injection of 20 µL of the original solid-phase extract of the irradiated TNT solution. The main peaks are marked by letters according to the mass spectra in Figure 4. For experimental details, refer to the Experimental Section.
Besides these compounds, three unknown, early-eluting compounds are present at high concentrations in this extract. The first compound, with a retention time of ∼5 min (peak a), has almost the same mass spectrum as a trinitrobenzene (cf. Figure 4a and d). This compound can tentatively be identified as 2,4,6TNBA, which is known to decarboxylate very easily. Therefore, the molecular ion peak at m/z 257 is not present under these ionization conditions. In this case, the base peak at m/z 213 can be assigned to the [M - CO2]- ion. Figure 4b shows the mass spectrum of peak b at a retention time of ∼7 min. This mass spectrum has the same base peak as TNT (m/z 227) but a different fragmentation pattern. The loss of carbon dioxide (-44) and nitrogen oxide (-30) confirms this compound to be a nitrobenzoic acid. The loss of carbon dioxide leads to a fragment at m/z 183. This ion is consistent with the mass of a deprotonated dinitrophenol. Taking this into account, this compound can be tentatively identified as a hydroxydinitrobenzoic acid. The mass spectrum of the third unknown phototransformation product (corresponding to peak c with a retention time of ∼8 min) is shown in Figure 4c. This compound has a molecular mass of >400 and shows the loss of carbon dioxide (-44) and nitrogen oxide (-30). The peak at m/z 421 is consistent with the [M H]- ion of II represented in Figure 2, which is formed by decarboxylation of 2,2′-dicarboxy-3,3′,5,5′-tetranitroazoxybenzene. However, it cannot be ruled out that this mass spectrum corresponds to I in Figure 2. As a result of the instability of this compound, the [M - H]- ion at m/z 465 may not be present in the TSP-MS spectrum. Rather, decarboxylation may lead to the observed peak at m/z 421. It is obvious that the information provided by mass spectrometry is not sufficient to elucidate the structure of these unknown Analytical Chemistry, Vol. 70, No. 19, October 1, 1998
4107
Figure 4. Mass spectra extracted from the TIC chromatogram represented in Figure 3. The subdivision of this figure corresponds to the assignment of the peaks shown in Figure 3. The loss of fragments is visualized by adding curved lines into the spectra. Abbreviations (nominal mass of fragments): H, hydrogen (1); O, oxygen (16); OH, hydroxyl (18); NO, nitrogen oxide (30); NO2, nitrogen dioxide (46); CO2, carbon dioxide (44); H3CO, methanolate (formation of cluster, 31); HCOO, formate (formation of cluster, 45); M, m/z ratio of the molecular ion.
compounds in the absence of reference compounds. In particular, the method does not allow the differentiation between positional isomers. 1H NMR provides complementary information, e.g., about the substitution pattern of aromatic compounds, and should therefore be used to verify the tentative compound identification by mass spectrometry. Continuous-Flow HPLC/1H NMR. Figure 5 shows the NMR chromatogram of the irradiated TNT solution (represented as a two-dimensional contour plot) obtained as described in the Experimental Section. This NMR chromatogram shows signals for residual TNT, 3,5-DNP, and 1,3,5-TNB. The procedure for the identification of these three nitroaromatic compounds (for which reference compounds are commercially available) by continuous-flow HPLC/1H NMR is given elsewhere1 and will not be repeated here. Rather, the HPLC/NMR data will be used to verify the structure assignment by HPLC/TSP-MS of those compounds for which reference compounds were not available and in particular to determine the substitution pattern at the aromatic ring. As a result of the different chromatographic conditions used for HPLC/MS and HPLC/NMR (see Experimental Section), the retention times in Figure 3 and Figure 5 cannot be compared directly. Nevertheless, the elution order is the same and the 4108 Analytical Chemistry, Vol. 70, No. 19, October 1, 1998
relative retention times are very similar for both separation methods. Moreover, it is possible to estimate the relative amount of analyte injected onto the column in both cases by considering the intensities of the signals. Therefore, a correlation between the HPLC/MS peaks in Figure 3 and the HPLC/NMR signals in Figure 5 is possible. Thus, the signal at 9.21 ppm (marked “a”) in Figure 5 belongs to a very early eluting compound, i.e., peak a in Figure 3. The observation of a singlet is consistent with the symmetric isomer 2,4,6-trinitrobenzoic acid (see corresponding mass spectrum in Figure 4a). Similar reasoning led to the conclusion that the peaks b and c in the HPLC/MS chromatogram (mass spectra in Figure 4b and c) correspond to the signals b and c in the HPLC/NMR contour plot. Thus the two meta-coupling doublets (marked “b”) in Figure 5 are consistent with the 2-hydroxy-4,6-dinitrobenzoic acid. (Coupling constants in the contour plot can be recognized by the splitting up of the peaks. The larger coupling constants of aromatic protons in the ortho position of ∼8 Hz are clearly visible whereas the smaller meta couplings of ∼2 Hz are not always completely resolved. For example, the peak of 2,4-DNP (H 5) in Figure 6 demonstrates a case in which both types of couplings are visible.) Furthermore, the four meta-coupling doublets (marked “c”) in Figure 5 are consistent with I in Figure
Figure 5. NMR chromatogram as the result of an injection of 500 µL of the freeze-dried and reconstituted extract of the irradiated TNT solution onto a 125 × 4 mm RP Select B column. See text for abbreviations. Numbers in parentheses assign the NMR signals to the protons generating this signal.
2, whereas II and III in Figure 2 would lead one to expect another spin-spin coupling pattern and intensity ratio for the NMR signals. Relative retention times of main compounds for both separations and the tentative assignment of spectroscopic data to the proposed structures is given in Table 1. This table shows that NMR data give unequivocal information about the substitution pattern of the aromatic ring. In addition to the HPLC/MS results, the HPLC/NMR data reveal the presence of small amounts of 2,4-DNBA and 1,3-DNB. Comparison of the compounds identified in this study with those represented in Figure 1 shows that 2-amino-4,6-dinitrobenzoic acid and 3,5-dinitroaniline were not generated under the selected experimental conditions. This might be explained by the fact that sodium azide was added to the aqueous solution to prevent microbial degradation. It is known that microbial transformation of nitroaromatic compounds leads to a reduction of the nitro group to an amino group, whereas phototransformation of nitrotoluenes leads to an oxidation of the methyl group. Therefore, it is likely that the generation of 2-amino-4,6-dinitrobenzoic acid and 3,5-dinitroaniline is a result of both photolytical and microbiological transformations. By forestalling microbial transformation, 2-hydroxy-4,6-dinitrobenzoic acid will obviously be generated as the main intermediate product of the phototransformation. Investigation of a Groundwater Sample by ContinuousFlow HPLC/1H NMR. The extract of a groundwater sample taken from a former ammunition site was prepared, measured, and represented as a contour plot in Figure 6 under exactly the same conditions as the extract of the irradiated TNT solution by
Figure 6. Low-field part of the NMR chromatogram of the groundwater extract taken from the former ammunition site in Elsnig, Germany. See text for abbreviations. PA, picric acid. Numbers in parentheses assign the NMR signals to the analyte protons generating this signal.
continuous-flow HPLC/1H NMR shown in Figure 5. Therefore, the confirmation of phototransformation products of TNT in this groundwater sample can easily be achieved by simple comparison of retention times, chemical shift values, and spin-spin coupling patterns of NMR signals in both NMR chromatograms. In Figure 6, NMR signals of neutral and acidic nitroaromatic compounds can be observed. In particular, TNT and its byproducts (e.g., mononitrotoluenes and dinitrotoluenes) are present in this extract at high concentrations. However, no 2-hydroxy-4,6-dinitrobenzoic acid and no 2,2′-dicarboxy-3,3′,5,5′-tetranitroazoxybenzene can be detected in this extract above the detection limit. Instead, 2-amino4,6-dinitrobenzoic acid and its decarboxylation product 3,5dinitroaniline are present in this sample, where microbial reduction of nitroaromatic compounds occurred, as can be shown by the formation of aminodinitrotoluenes. The confirmation of 2-amino-4,6-dinitrobenzoic acid was performed by synthesis of the reference compound by regioselective reduction of 2,4,6-trinitrobenzoic acid according to a method suggested by Bil′kis et al.7 and subsequent comparison of the NMR data of this compound with those obtained from the groundwater extract. Furthermore, the presence of several other acidic nitroaromatic compounds, e.g., 2,4-dinitrophenol, 2-amino-4-nitrobenzoic acid, 4-nitrobenzoic acid, 4-nitrophenol, and picric acid (2,4,6-trinitrophenol) might be explained by the formation by phototransformation of mono- or dinitrotoluenes or by oxidation during the production process. (7) Bil’kis, I. I.; Ustov, S. I.; Shteingarts, V. D. Izv. Sib. Otd. Akad. Nauk SSSR, Ser. Khim. Nauk 1987, 3, 111-121.
Analytical Chemistry, Vol. 70, No. 19, October 1, 1998
4109
Table 1. HPLC/MS and HPLC/1H NMR Data for the Identification and Structural Characterization of the Phototransformation Products of 2,4,6-Trinitrotoluene main ion m/z (tentative struct assgn)
HPLC/NMR tr,rel
resonance lines δ 1H (ppm) spin-spin coupling patternc (tentative assgn)
assgna
HPLC/MS tr,relb
a b
0.26 0.36
213 (M - CO2)227 (M - H)-
0.39 0.46
9.20 s (H1,3,5) 8.71 d (H5); 8.44 d (H3)
c
0.42
347 (M - H - CO2 - CO2 - NO)-
0.52
d e
0.66 1.00
213 (M)227 (M)-
0.63 1.00
f
1.26
184 (M)-
1.19
9.34 d (H4); 9.13 d (H6); 9.02 d (H4′); 8.80 d (H6′) 9.38 s (H2,4,6) 8.98 s (H3,5); 2.65 s (methyl group) 8.52 t (H4); 8.01 d (H2,6)
compound 2,4,6-trinitrobenzoic acid 2-hydroxy-4,6-dinitrobenzoic acid 2,2′-dicarboxy-3,3′,5,5′tetranitroazoxybenzene 1,3,5-trinitrobenzene 2,4,6-trinitrotoluene 3,5-dinitrophenol
a The assignment corresponds to that shown in Figures 3-5. b Relative retention times refer to the retention time of 2,4,6-trinitrotoluene (t r,rel ) tr,i/tr,TNT). c s, singulet; d, doublet; t, triplet.
Finally, Figure 6 reveals that there are several additional minor components in the wastewater sample, which up to now have not been identified. CONCLUSIONS The results show that HPLC/1H NMR together with HPLC/ MS data give sufficient information for the structural characterization of polar transformation products (e.g., generated by phototransformation) of xenobiotics, even if no reference compounds are commercially available. The analytical strategy applied in this examination includes the generation of phototransformation products by irradiation of an aqueous solution of the pure target compound by natural sunlight, the selective solid-phase extraction of acidic and neutral transformation products, and subsequent acquisition of MS and 1H NMR data. On-line separation prior to spectroscopic investigation is a key step of the strategy applied. Such on-line coupling
4110 Analytical Chemistry, Vol. 70, No. 19, October 1, 1998
has the advantage of high efficiency and is amenable to extremely labile compounds (e.g., trinitrobenzoic acid). Is is demonstrated that HPLC/1H NMR, applied here for the first time to the structure elucidation of phototransformation products, gives information complementary to that provided by HPLC/MS. ACKNOWLEDGMENT The authors thank Dr. Klaus Steinbach from the Philipps Universita¨t Marburg, who made the proton spectrum of 2-amino4,6-dinitrobenzoic acid available to us. Financial support (Contract 146 1063) from the federal Ministerium fu¨r Bildung, Wissenschaft, Forschung und Technologie (BMBF) Bonn, Germany, is kindly acknowledged. Received for review March 13, 1998. Accepted July 8, 1998. AC980292A
