Anal. Chem. 2005, 77, 7415-7420
Determination of Benzotriazole Corrosion Inhibitors from Aqueous Environmental Samples by Liquid Chromatography-Electrospray Ionization-Tandem Mass Spectrometry Stefan Weiss and Thorsten Reemtsma*
Department of Water Quality Control, Technical University of Berlin, Sekr KF 4, Strasse des 17 Juni 135, 10623 Berlin, Germany
The first method for the determination of commonly used corrosion inhibitors in environmental water samples by liquid chromatography-electrospray ionization-tandem mass spectrometry is presented. Benzotriazole (BTri) and the two isomers of tolyltriazole (5- and 4-TTri) are separated in an isocratic run. By gradient elution, BTri, 4-TTri, 5-TTri, and xylyltriazole can be determined simultaneously with three benzothiazoles, but here TTri isomers coelute. The instrumental detection limit of 2 pg allows the determination of the three most important benzotriazoles from municipal wastewater and most surface waters by direct injection into the HPLC system without previous enrichment. When solid-phase extraction is employed with mean recovery rates of 95-113%, the limit of quantification for benzotriazoles range from 10 ng/L in groundwater to 25 ng/L in untreated wastewater. BTri and TTri were determined in municipal wastewater in microgram per liter concentrations. Elimination in wastewater treatment appears to be poor, and BTri and TTri can be followed through a water cycle from treated municipal wastewater through surface water to bank filtrate used for drinking water production. The TTri isomers show markedly different biodegradation behavior with 4-TTri being more stable. Benzotriazoles and benzothiazoles are polar high production volume chemicals that find broad application in various industrial processes as well as in households. The group of benzotriazoles, namely, benzotriazole (BTri) itself and the methylated tolyltriazoles (TTri, used as a technical mixture of 4- and 5-TTri), are used as corrosion inhibitors in deicing fluids for aircrafts, automotive antifreeze formulations, brake fluids, metal-cutting fluids, and industrial cooling systems.1-3 BTri but not TTri is also used for silver protection in household dishwashing agents. Estimates of total production volumes are contradictory and range from 9000 * To whom correspondence should be addressed. E-mail:
[email protected]. (1) Patsalides, E.; Robards, K. J. Chromatogr. 1985, 331, 149-160. (2) Cancilla, D. A.; Baird, J. C.; Geis, S. W.; Corsi, S. R. Environ. Toxicol. Chem. 2003, 22, 134-140. (3) Gruden, C. L.; Dow, S. M.; Hernandez, M. T. Water Environ. Res. 2001, 73, 72-79. 10.1021/ac051203e CCC: $30.25 Published on Web 10/13/2005
© 2005 American Chemical Society
tons/year worldwide (confidential source) to 9000 tons/year for the United States only.4 Benzotriazoles are polar and weakly basic (pKa 8.2-8.8) compounds3 with an only moderate tendency to partition into an organic phase (log Kow 1.23 for BTri and 1.89 for TTri4). Contrary to N-substituted 1,2,4-triazoles, the benzo-1,2,3-triazoles appear to have only limited biological activity. The acute toxicity of 5-TTri to aquatic organisms is in the low to moderate milligram per liter range,2,5 but some bioaccumulation has been reported to occur.2 A few reports exist on the detection of 5-TTri in aquatic compartments, mostly from monitoring studies: 5-TTri has been reported to occur in surface waters in median concentrations of 0.4 µg/L,6 in runoff from airports,2 and in runoff from agricultural fields irrigated with treated wastewater.7 In these studies, GC/ MS analyses after analyte enrichment by liquid-liquid extraction6-9 or after solid-phase extraction (SPE)2 were used. However, neither BTri nor 4-TTri has been considered in such investigations yet, which may partly be due to the lack of a dedicated method for the determination of all these benzotriazoles from environmental samples. Another class of nitrogenous aromatic industrial chemicals are benzothiazoles, with a S, N aromatic five ring instead of the triazole system with three nitrogens. Different derivatives of this compound class are used as vulcanization accelerators in rubber production, as biocides in paper and leather manufacturing, and also as corrosion inhibitors. An analytical method for the trace determination of six polar benzothiazoles and a first study on their occurrence in municipal wastewater have been published recently.10,11 (4) Hart, D. S.; Davis, L. C.; Erickson, L. E.; Callender, T. M. Microchem. J. 2004, 77, 9-17. (5) Cornell, J. S.; Pillard, D. A.; Hernandez, M. T. Environ. Toxicol. Chem. 2000, 19, 1465-1472. (6) Kolpin, D. W.; Furlong, E. T.; Meyer, M. T.; Thurman, E. M.; Zaugg, S. D.; Barber, L. B.; Buxton, H. T. Environ. Sci. Technol. 2002, 36, 1202-1211. (7) Pedersen, J. A.; Yeager, M. A.; Suffet, I. H. J. Agric. Food Chem. 2003, 51, 1360-1372. (8) Cordy, G. E.; Duran, N. L.; Bouwer, H.; Rice, R. C.; Furlong, E. T.; Zaugg, S. D.; Meyer, M. T.; Barber, L. B.; Kolpin, D. W. Ground Water Monit. Rem. 2004, 24, 58-69. (9) Stackelberg, P. E.; Furlong, E. T.; Meyer, M. T.; Zaugg, S. D.; Henderson, A. K.; Reissman, D. B. Sci. Total Environ. 2004, 329, 99-113. (10) Kloepfer, A.; Jekel, M.; Reemtsma, T. Environ. Sci. Technol. 2005, 39, 37923798.
Analytical Chemistry, Vol. 77, No. 22, November 15, 2005 7415
Figure 1. Structures and acronyms of the benzotriazoles and benzothiazoles included in this study. Note that BT (and also BTAH) have also been used as acronyms for benzotriazole. BTri is preferred here to clearly distinguish it from benzothiazole.
In this work, we report the first LC-MS/MS approach for the analysis of all four benzotriazoles (BTri, 5-TTri, 4-TTri, and xylyltriazole, XTri) that were expected to occur in the aquatic environment. LC-MS/MS would allow for the direct injection of aqueous samples if the sensitivity of detection is sufficiently low. It was intended to develop a method enabling us to detect members of both classes, the benzotriazoles and the benzothiazoles, and suitable to study the occurrence and behavior of these widely used industrial and household chemicals. EXPERIMENTAL SECTION Chemicals: Benzothiazole (BT; 96%), 1-H-benzotriazole (BTri; 99%), 5-methyl-benzotriazole (5-tolyltriazole, 5-TTri; 98%), 5,6dimethylbenzotriazole (xylyltriazole, XTri, 99%) and benzothiazole6-carboxylic acid (BTCA; 96%) were purchased from Sigma-Aldrich Chemie (Steinheim, Germany); 2-aminobenzothiazole (ABT; 98%) was from Fluka Chemie (Buchs, Switzerland); 2-methylthiobenzothiazole (MTBT; pure) was from Ferak (Berlin, Germany), and a technical mixture of 4- and 5-TTri was kindly provided by a manufacturer. The target analytes are displayed in Figure 1. Methanol, acetone, formic acid, and ammonium acetate were of HPLC grade. Ultrapure water was delivered by a water purification system ELGA maxima (Elga, High Wycombe Bucks, U.K.). Samples. Six 24-h composite samples of the influent and the effluent of a municipal wastewater treatment plant (WWTP) with an input of ∼30% industrial wastewater and a capacity of 240 000 m3/d at dry weather flow, involving activated sludge treatment with enhanced nutrient removal (tertiary treatment), were collected. Effluent sampling of the WWTP was delayed relative to the influent to compensate for the hydraulic retention time of the wastewater in the plant. Three surface water samples (grab samples from Lake Tegel, Landwehr Canal) and two samples from subsurface (one bank filtrate (Lake Tegel) and one groundwater not influenced by surface water) were taken in Berlin. Samples were filtered at 0.45 µm and stored frozen until analysis. Enrichment. Extraction was performed using 60-mg Oasis HLB cartridges (Waters) and an AutoTrace SPE Workstation (Zymark). Cartridges were conditioned with 5.0 mL of MeOH/ (11) Kloepfer, A.; Jekel, M.; Reemtsma, T. J. Chromatogr., A 2004, 1058, 8188.
7416
Analytical Chemistry, Vol. 77, No. 22, November 15, 2005
acetone (6/4) and washed with 5 mL of ultrapure water before extraction. Volumes of 50 mL of untreated wastewater samples or 100 mL of all other water samples were extracted at a flow rate of 4.0 mL/min; the cartridges rinsed with 0.5 mL of pure water and eluted with 7 mL of MeOH/acetone (6/4). The 0.5-mL samples of ultrapure water were finally added to the extract as a keeper together with 20 µL of an internal standard solution (BTCA at 500 µg/L). The solvent was evaporated to 0.4 mL using a SpeedVac concentrator (Savant) at 38 °C; the extract was transferred into a vial and filled to 1.5 mL with pure water. Chromatography. Analysis was performed on a HP 1100 (Agilent) HPLC system consisting of a membrane degasser, a binary high-pressure pump, an automatic sampler, a column heater, and a diode array detector that was coupled to a triplestage quadrupole mass spectrometer. Eluents were H2O/MeOH 80/20 (A) and H2O/MeOH 10/90 (B), each modified with 0.1% (v/v) formic acid. The separation was performed using a 2 mm × 150 mm Pursuit Diphenyl column with 3-µm particle size (Varian, Palo Alto, CA) at 40 °C. A sample volume of 60 µL was injected. Method 1: The TTri isomers were separated and detected together with BTri in an isocratic run (10% B) with a flow rate of 0.15 mL/min. A column wash (100% B for 8 min, reconditioned for 15 min at 10% B) was inserted after each sixth analysis of real samples. Method 2: The three benzotriazoles (BTri, TTri, XTri) and three benzothiazoles (ABT, MTBT, BT) were separated by gradient elution, starting with 20% B at 0 min, increased to 65% B at 5 min, held for 2.5 min, and decreased to 20% B at 8.5 min. After 15 min, the system was ready for the next injection. The flow was set to 0.2 mL/min. With this gradient elution, the two TTri isomers coeluted. Mass Spectrometry. A Quattro LC triple-stage quadrupole mass spectrometer (Micromass, Manchester, U.K.) was used with the Z-spray interface equipped with the electrospray probe operated in the positive ion mode. Nitrogen was provided by a nitrogen generator (Whatman) and used as drying and nebulizing gas. Argon (purity 5.0, Messer-Griesheim) was used as a collision gas at a pressure of 1.1 × 10-3 mbar in the collision cell. The source was operated at a temperature of 120 °C and with a drying gas flow of 800 L h-1 and nebulizer gas flow of 90 L h-1. The desolvation temperature was 220 °C. The capillary voltage was 3.6 kV. Quantitation was performed by multiple reaction monitoring (MRM) with the conditions given in Table 1. Quantitation. For quantitation, standard addition of BTri, the technical TTri mix, and XTri before SPE at three concentration levels was performed into aliquots of one representative sample of each series of samples. The isomeric composition of the technical TTri mix and the response factors of both isomers were determined as described in the Supporting Information. BTCA was added as internal standard to follow instrumental stability. Recovery Experiments. Three aliquots of a pure water and two replicates of a groundwater sample and a bank filtrate were processed as described, with one aliquot spiked with benzotriazoles at 10 ng/L and a second aliquot spiked at 50 ng/L. The extract of the nonspiked aliquot was separated into three portions of 450 µL after extraction, and two of the aliquots were spiked with the same amounts of standards as prior to extraction. The
Table 1. Chromatographic and Mass Spectrometric Parameters and Instrumental Detection Limit (IDL; S/N g 3) for the Determination of Benzotriazoles and Benzothiazoles by LC-ESI-MS/MS in the Positive Ion Mode substance
RTa (min)
[M + H]+ (m/z)
CVb (V)
ABT
3.5
151
35
BTri
4.1/6.6
120
35
5-TTri
5.6/9.9
134
35
4-TTri
5.6/10.3
134
35
BT
7.6
136
40
XTri
8.0
148
35
MTBT
11.8
182
29
product ions (m/z)
proposed product ionc
CEd (eV)
product ion ratio
IDL (pg)
109 65 65 92 77 79 77 79 109 65 77 91 167 123
C6H5S C5H5e C5H5 C6H6N C6H5 C6H7 C6H5 C6H7 C6H5S C5H5e C6H5 C7H7 C7H5NS2 C6H5NSe
26 35 20 16 25 22 25 22 26 37 25 25 22 34
2.4
15e
3.8
2
1.8
2
1.8
2
2.2
50e
2.0
2
6.9
35e
a Retention time in gradient elution/isocratic elution (if applicable). b CV, cone voltage. c For structure proposals refer to Figure 3. d CE, collision energy. e For details refer to ref 13.
response factors obtained by linear regression analysis of each analyte from spiking before SPE (R3) were compared to the response factors obtained by spiking into the extracts (R2). Recoveries at this concentration level could not be determined from municipal wastewater as its analyte concentrations were too high. Matrix Effects. Matrix effects for each analyte in each kind of sample were determined by comparing the respective R2 factor with the response factors obtained in pure aqueous solution (R1). For directly injected samples, the standard was spiked into an aliquot of the aqueous sample to determine R2. Method Sensitivity. Instrumental detection limits (IDL) were calculated from the signal-to-noise ratio (S/N > 3) of pure standard solutions injected onto the column. Limits of quantification for the whole method including SPE were determined by the signalto-noise ratio (S/N > 10) of the injected extracts. RESULTS AND DISCUSSION The selection of analytes covered in this study is displayed in Figure 1. Besides the three benzotriazoles known to be used as corrosion inhibitors (BTri, 4-TTri, 5-TTri) also the closely related XTri was included in this study. Additionally, three benzothiazoles were included into method development, which occur as polar metabolites of industrial chemicals that are also used as corrosion inhibitors. Chromatographic Separation. A separation of BTri and the two TTri isomers was achieved by isocratic elution (method 1) and a resolution of 0.73 was obtained for 5-TTri and 4-TTri (Figure 2a). Further decreasing the methanol content did not improve resolution of the two isomers due to peak broadening with increasing retention time. Peak assignment of the two TTri isomers was accomplished by adding the commercially available 5-TTri to the technical mix; this resulted in an increase of the first peak. If no separation of TTri isomers is required a gradient elution can be used to analyze BTri, TTri, and XTri together with three benzothiazoles in a total run time of 15 min (method 2; Figure 2b). Here BT and XTri are incompletely separated, but they can be easily distinguished by MRM detection (Table 2).
Figure 2. (a) LC-MS chromatogram of a pure standard solution of BTri, 5-TTri, and 4-TTri by isocratic elution (method 1). Isomeric ratio 5-TTri/4-TTri ) 1.3. (b) LC-MS chromatogram of a pure standard solution of benzotriazoles (BTri, 5-TTri, 4-TTri, XTri) and benzothiazoles (ABT, BT, MTBT) by gradient elution (method 2). (For MRM transitions, please refer to Table 1.)
Mass Spectrometric Detection. Contrary to N-substituted 1,2,4-triazoles, which have been intensively studied by mass spectrometry, no information was available on the fragmentation of these C-substituted 1,2,3-triazoles. Product ion spectra of the two TTri isomers are available as Supporting Information (Figure S1). The loss of molecular nitrogen to form an azirine cation is the initial fragmentation of all molecular cations that are likely ionized at the triazole ring system (Figure 3). This fragmentation corresponds to N-substituted 1,2,3-triazoles.12 Then hydrocyanic acid is expelled to form a cyclopentadienyl cation. The same product is formed from the S,N-heterocyclic benzothiazole cations (Table 113). According to the different substitution patterns of the triazoles, their further fragmentation differs. While BTri shows no further fragmentation, the methyl-substituted benzotriazoles TTri and XTri can eliminate H2, likely resulting in a conjugated (12) Santos, L. S.; Padilha, M. C.; Neto, F. R. D.; Pereira, A. D.; Menegatti, R.; Fraga, C. A. M.; Barreiro, E. J.; Eberlin, M. N. J. Mass Spectrom. 2005, 40, 815-820. (13) Reemtsma, T. Rapid Commun. Mass Spectrom. 2000, 14, 1612-1618.
Analytical Chemistry, Vol. 77, No. 22, November 15, 2005
7417
Figure 3. Fragmentation pathways of the molecular cations of BTri, TTri, and XTri and suggested product ion structures as deduced from MS/MS experiments.
Figure 4. (a) Matrix effects in ESI-MS detection of benzotriazoles of directly injected samples of surface water, WWTP effluent, and influent. (b) Mean recoveries of benzotriazoles from two spiking levels (10, 50 ng/L) by SPE of different waters (PW, ultrapure water; BF, bank filtrate; GW, groundwater). nd, not detectable because BTri concentration in the sample was too high. (c) Matrix effects in ESI-MS detection of benzotriazoles in SPE extracts of BF and GW.
system with an exocyclic double bond (Figure 3). This XTri fragment ion (m/z 91) can further eliminate CH2 to form m/z 77 (Figure 3). Besides this fragmentation sequence, TTri and XTri also form the tropylium (m/z 91) and the methylated tropylium (m/z 105) cation, respectively. For MRM detection, the two most intense transitions of each analyte were recorded and the ion ratio was used for confirmation (Table 1). The response of all four benzotriazoles is remarkably similar, with an IDL (S/N > 3) of 2 pg injected onto the LC column. With an injection volume of 60 µL, this results in a calculated LOD of 33 ng/L and an LOQ of ∼100 ng/L. As compared to the most sensitive method published before for 5-TTri,6 this LC-MS method brings along a 10-fold increase in sensitivity for surface water analyses without using an enrichment step. Thus, most wastewater and surface water samples, in which BTri and TTri occur at concentrations above 200 ng/L, can be analyzed by direct injection of the aqueous samples, expect for XTri, which is rarely found. Some signal suppression due to matrix effects may, however, occur. By standard addition into aqueous samples and comparison of the response factor (R2) to that of pure aqueous solutions (R1), a signal decrease to 80% was determined on average (Figure 4a). As long as no isotopically labeled reference compounds are available external sample calibration14 is recommended to compensate for matrix effects and to obtain accurate results. Extraction and Recovery. If analyte concentrations below 200 ng/L have to be quantified, analyte enrichment by SPE is (14) Stueber, M.; Reemtsma, T. Anal. Bioanal. Chem. 2004, 378, 910-916.
7418
Analytical Chemistry, Vol. 77, No. 22, November 15, 2005
necessary. The optimization and evaluation of methods consisting of SPE and LC-MS/MS require a clear distinction between recovery rates of the extraction process and matrix effects occurring during ESI-MS detection, as both influence the signal intensity. Three response factors (R) are required for this purpose: R1, the response factor of the pure standard solution, R2 the factor obtained by spiking into sample extracts, and R3 the factor obtained from spiking samples before SPE. The ratio R2/R1 is a measure for matrix effects, whereas R3/R2 describes the true recovery of the SPE. Recoveries were determined at two concentration levels (10 and 50 ng/L) and from three kinds of sample, ultrapure water, groundwater, and bank filtrate. No significant difference was observed between these two concentration levels, and average data are presented. Mean recoveries range from 95 to 113% (Figure 4b), and no significant influence of the sample matrix on the recovery could be observed. The reproducibility of the whole procedure was good with relative standard deviations of 9-15% for BTri, 7-18% for 5-TTri, 4-25% for 4-TTri, and 7-15% for XTri. The high variability of 4-TTri in pure water (Figure 4b) was due to one outlier with 138% recovery. Blank values of all analytes were below the detection limits. The same extraction method can also be used for the S,N-heterocyclic benzothiazoles.11 Recoveries from wastewater could only be determined for XTri with 95% in WWTP effluent and 110% in WWTP influents. For BTri and TTri, concentrations in treated wastewater were ∼2 orders of magnitude higher than the maximum concentration added in these experiments (50 ng/L).
Table 2. Mean Concentrations and Variability of Concentration (( Standard Deviation) of Benzotriazoles Found in Different Samples of the Berlin Region (n ) Number of Samples) concentration (µg/L) WWTP influent (n ) 6) WWTP effluent (n ) 6) Lake Tegel (n ) 2) Landwehr Canal bank filtrate (n ) 1) groundwater (n ) 1)
BTri
5-TTri
4-TTri
XTri
11.9 ( 1.2 9.6 ( 1.3 0.9 3.4 0.2
