Anal. Chem. 1995, 67, 1637-1643
Automated On-Line Liquid-Solid Extraction Followed by Liquid Chromatography-High-Flow Pneumatically Assisted Electrospray Mass Spectrometry for the Determination of Acidic Herbicides in Environmental Waters S. Chiron,t S. Papilloud,* W. Haerdi,* and D. Barcelo*it Department of Environmental Chemistly, CID-CSIC d Jordi Girona 18-26, 08034 Barcelona, Spain, and Department of Analytical Chemistly, Universily of Geneva, 30 quai Emest Ansemet, 1211 Geneva 4, Switzerland
Iiquid chromatography-high-flow pneumatically assisted electrospray mass spectrometrywith negative ionization was used for the determination of several acidic herbicides: benazolin, bentazone, 2,4-D, MCPA, MCPP, MCPB, and 6- and 8-hydroxybentazone. To achieve good LC separation, aciditicationof the LC eluent and subsequent postcolumn addition of a neutralization buffer are needed to avoid ion signal suppression. This method was combined with a prior automated on-line liquid-solid extraction step using an OSP-2 autosampler containing C-18 cartridges, and it was applied to the trace determination of acidic herbicides in environmental waters. The proposed method required only 50 mL of water with a limit of detection between 0.01 and 0.03 p d L , generally employing selected ion monitoring of the [M - HI- ion and extraction voltages of 20 V. Examples of the trace level determination of 8-hydroxybentamne, bentazone, and MCPA in real estuarine water samples from the Ebro River delta (Tarragona, Spain) are shown. Only when extraction voltages above 30 V were used was it possible to distinguish between 6- and 8-hydroxybentazone. Determination of the chlorophenoxy carboxylic acid herbicides khloro-2-methylphenoxyacetic acid (MCPA), 2,4dichlorophenoxyacetic acid (2,4D), 2-(4-chloro-2-methylphenoxy)propionic acid (MCPP), 4(4-chlorc-2-methylphenoxy)butyric acid (MCPB), khloro-2,3-dihydro-2-oxobenzothiazol-?-ylacetic acid (benazolin) , %isopropyl-(lH)-2,l,%benzothiadiazin-4(3H)-one 2,2-dioxide @entazone), and &and 8-hydroxybentazone in surface waters can be achieved by on-line liquid-solid extraction (LSE) techniques followed by liquid chromatographic (LC) methods with W or diode array detection However, few of these compounds lack a strong chromophore above 220 nm, and a matrix peak attributed to fulvic and humic substances present in river water appears at the beginning of the ~hromatogram,~ coeluting with the more polar compounds (e.g., benazolin and 6 and 8hydroxybentazone). Therefore, difficulties arise for the detection ' CID-CSIC.
* University of Geneva. (1) Geerdink, R B.; Graumans, A. M. B. C.; Viveen, J.J Chromatogr. 1991, 547, 478-483. (2) Coquart, V Hennion, M. C.. Sci. Totul Enuiron. 1993,132, 349-360. (3) Chiron, S.; Martinez, E Barcelo, D. J. Chromatogr. A 1994,665, 283293. 0003-2700/95/0367-1637$9.00/0 0 1995 American Chemical Society
of pesticides and TPs in water at the 0.1 pg/L limit set by the Commission of European Communities.4 Another major limitation of current LC methods for pesticide analysis in surface water is the lack of mass spectrometric confirmation. On-line LSE-LUthermospray (TSP) mass spectrometry in positive ion (PI) mode using selected ion monitoring (SIM) appears to be a very powerful tool for confirmation of pesticide traces in water with LODs between 0.01 and 0.1 pg/L after preconcentration of a 50 mL water sample using C-18 or polymeric sorbent material^.^-^ However, for the identitication of acidic herbicides, the TSP-MS negative ion (TVI) mode of operation is needed.8*9 Still, NI mode in LSE-LC/TSP-MS did not allow trace determination at the 0.1 pg/L limit of the polar acidic herbicides benazolin, and 6 and ghydroxybentazone, even after preconcentration of a 100 mL of surface water.3 Particle beam interface (PB) is even worse than TSP for the determination of acidic herbicides because of poor linearity and sensitivity.lOJ1The use of flow injection analysis PIA)/TSP-MS/MS enables determination of acidic herbicides in water after a careful adjustment of the instrumental parameters.12 LC/ESP-MS has been reported to be more sensitive than LC/TSP-MS by factor of approximately 100 in PI mode for pesticide residue a n a l y ~ i s . ~Only ~ J ~two applications were reported in the literature showing the use of NI ESP-MS in environmental analysis for the determination of sulfonated azo d ~ e s . * Recently, ~J~ high-flow ESP-MS has become commercially Fielding, M.; Barcelo, D.; Helweg, A; Galassi, S.; Torstensson, L.; van Zoonen, P.; Wolter, R; Angeletti, G. Pesticides in Ground and Drinking Water. Water Pollution Research Report 27; Commission of the European Communities: Brussels, Belgium, 1992; pp 1-136. Chiron, S.; Dupas, S.; Scribe, P Barcelo. D. J. Chromatogr. A 1994,665, 295-305. Bagheri, H.; Brower, E. R; Ghijsen, R T; Brinkman, U. A. Th. Analusis 1992,20, 475-482. Volmer, D.; Levsen, K. J Am. SOC.Mass Spectrom. 1994,5, 655-675. Barcel6, D. Org. Mass Spectrom. 1989,24, 898-902. Liu, C. H.; Mattem, G. C.: Yu, X.; Rosen, R. T Rosen, J. D. j.Agn'c. Food Chem. 1991,39, 718-723. Jones, T. L.; Betowski, L D.; Lesnik, B.; Chiang, T. C; Teberg, J. E. Environ. Sci. Technol. 1991,25, 1880-1884. Betowski, L. D.; Roby, M; Pace, C. J Am. SOC.Mass Spectrom. 1992,3, 823-830. Geerdink, R B.; Kienhuis, G. M; Brinkman, U. k Th.J Chromutogr 1993, 647, 329-339. Voyksner. R D. Enuiron. Sci. Technol. 1994,28, 118A-127A. Molina, C.; Honing, M.; Barcel6, D. Anal. Chem. 1994,66, 4444-4449. Edlund, P. 0.; Lee, E. D.; Henion. J. D.; Budde, W. L. Biomed. Enuiron. M m Spectrom. 1989,18, 233-240.
Analytical Chemistry, Vol. 67, No. 9, May 1, 1995 1637
H
H
Table 1. Time-Scheduled SIM Conditions, with Preconcentrationthrough an OSP-2 C-18 Precolumn in LCESP-MS at 20 V
time (min in the LC traces) compds (m/z monitd for quant anal.) 0-23
8-HYDROXYBENTAZONE
6-HYDROXYBENTAZONE
Figure 1. Chemical structures of 6- and 8-hydroxybentazone.
available, affording a flow rate of 0.3-0.4 mL/min, which is directly amenable with conventional narrow-bore columns and LC pumps without the need of effluent splitting, and offers a lower detection limit than that previously reported for TSP-MS.14 In view of the different approaches used to monitor acidic herbicides in surface waters, the aim of this work is (a) to use NI mode ESP-MS for characterization of acidic herbicides and (b) to apply on-line LSELC/ESP-MS by using an OSP-2 autosampler and postcolumn addition of a neutralization buffer for the determination of acidic herbicides at the 0.05-5 pg/L limit in natural environmental waters. EXPERIMENTAL SECTION Chemicals. HPLCgrade water and methanol were purchased from J. T. Baker Oeventer, The Netherlands) and were passed through a 0.45 pm filter before use. Tripropylamine and formic acid were obtained from Fluka Chemie (Buchs, Switzerland). Bentazone, Ghydroxybentazone,fkhydroxybentazone,benazolin, 2,4D, MCPA, MCPP, and MCPB were purchased from Promochem (Wesel, Germany). Some of the chemical structures are shown in Figure 1. ChromatographicConditions. The eluent was delivered by a gradient system from Waters 616 pumps controlled by a Waters 600s controller from Waters Millipore (Milford, MA). Gradient elution was accomplishedusing an eluent containing 20%of solvent A (methanol) and 80%of solvent B (water, pH = 2.9 with formic acid) to 80%A-20%B in 30 min at a flow rate of 0.25 mL/min, returning to initial conditions in 5 min. The analysis involved a Lichrocart cartridge column (125 x 3 mm i.d) packed with Lichrospher 60RP select B material of 5 pm particle size from Merck (Darmstadt, Germany). Postcolumn addition of 0.1 mL/ min of tripropylamine (4 g/L methanol) was carried out using a &high-pressure pump from Knauer (Bad-Homburg, Germany) and a Valco tee. Mass Spectrometric Analysis. A VG platform ESP from Fisons Instruments (Manchester,U.K.) equipped with a Megatlow ESP probe was used. The instrument control and data processing involved the use of MassLynx s o h a r e installed in a Digital DEC PC 466. The parameters optimized for the characterization of acidic herbicides are as follows: drying nitrogen gas flow rate and ESP nitrogen nebulizing gas flow rate, 300 and 15 L/h, respectively; extraction and focus voltages, at 20 and 27 V, respectively; and source temperature, 145 "C. Other parameters were optimized as previously reported.14 Chromatograms were recorded under time-scheduled selected ion monitoring SIM conditions for quantitation, as shown in Table 1. Sample Preparation. Estuarine river waters samples from the Ebro River delta Varragona, Spain) were filtered through a (16) Bruins, A P.; Weidolf, L. 0. G.; Henion, J. D.; Budde, W. L. Anal. Chem. 1987,59,2647-2652.
1638 Analytical Chemistry, Vol. 67, No. 9, May 1 , 7995
23-27 27-35
8hydroxybentazone (255) (192)" 6hydroxybentazone (255) benazolin (198,170) bentazone (239) 2,4-D (219, 161) MCPA (199,141) MCPP (213,141) MCPB (227,141)
m/z monitored for sample with preconcentration at 32 V.
0.45 pm membrane filter (Millipore,Bedford, MA) and were then acidified at pH = 1.5 (sulfuric acid). Samples were spiked with the different pesticides, giving final concentrations in the range 0.05-5 pg/L. An OSP-2 autosampler from Merck was connected to an L-6200A intelligent pump (Merck-Hitachi), which delivered the water sample containiig the pesticides. OSP-2 Lichrospher 100 Rp-18 cartridges of 10pm particle size were iirst conditioned by flushing with 10 mL of methanol and then 10 mL of HPLG grade water @H = 1.5 with sulfuric acid) at a flow rate of 1mL/ min. Water samples of 50 mL volume were preconcentrated through the precolumns at of flow rate of 4 mL/min. Following the preconcentration step, the OSP-2 valve was switched, and the analytes were separated in an analytical column similar to the previously described m e t h ~ d . ~The J ~ general scheme of the method used is shown in Figure 2. RESULTS AND DISCUSSION Mass Spectra Information. High-flowpneumatically assisted ESP-MS formed [M - HI- ion as a base peak for the whole group of acidic herbicides except benazolin and MCPB, which exhibited [M - COOHI- and [M - (CHz)&OOH]- ions as base peaks, respectively. The different ions formed and their relative abundances obtained by LC/ESP-MS in the NI mode are shown in Table 2. Good fragmentation patterns were noticed for the chlorophenoxy carboxylic acids and benazolin at 20 V extraction voltage. These fragment ions agree with results reported in previous papers obtained using LC in combination with direct liquid introduction18 or TSP338J9interfacing systems. Benazolin displays extensive fragmentation corresponding to [M - 451- and [M - 731-, ions which can be attributed to the losses of the carboxylic acid moiety and the NCHzCOOH group, respectively. When LC/TSP-MS was used, acidic herbicides exhibited a strong dependence on the tip temperature? whereas under LC/ESP-MS this dependence does not exist, thus allowing analysis of thermally labile compounds without degradation at the LC/MS interface, as shown for several thermally labile organophosphorus pesticides.'* The negative ion stability proved to be a complication under NI ESP-MS. The relative abundance of the ions formed varied to Chiron, S.; Femandez-Alba, A; Barcelb, D. Enoiron. Sci. Technol. 1993, 27,2352-2359. Voyksner, R D.; Bursey, J. T Pellizzari, E. D. J. Chromatogr. 1984,312, 221-235. Jones, J. T.; Betowski, L. D; Yinon, J. In Liquid Chromatography/Mass Spectrometry.Applications in Agriculture, Pharmaceutical and Enoironmental Chemhty; Brown, M. A, Ed.; ACS Symposium Series 420; American Chemical Society: Washington, DC, 1990; pp 62-74.
I
Autosampier
~
1
PUMP
I
I
Diecharge
I
1
I
ESP
MARR
__.
4-75,
+ I
r Discharye
Dmharge o i Delecloi
1
Figure 2. Schematic diagram of the automated on-line precon intration unit followed by LC postcolumn addition and ESP-MS. Table 2. Main Ions and Their Relative Abundance for Acidic Herbicides under ESPMS using NI Mode of Operationw M W
compound
m / z and tentative ID of ion($
RA
256 256 243
6hydroxybentazone Shydroxybentazone benazolin
240 200
bentazone
220
2,4D
214
MCPP
228
MCPB
255, [M - HI255, [M - HI198 [M - COOHI170, [M - NCHzCOOHI239, [M - H 1199 [M - HI141, [M - CHzCOOHJ245, [M HCOOI219, [M - HI161, [M - CHzCOOHI265, [M + HCOOI213. IM - H1141; [M - CHCH3COOHl259, [M + HCOOI227, [M - HI141, [M - (CHz)$OOHl273, [M HCOOI-
100 100 100 23 100 100 18 5 100 40 4 100 21 3 32 100 9
MCPA
+
+
Conditions: extraction voltage, 20 V; carrier stream, watermethanol (5050 v/v) with formic acid, pH = 2.9; postcolumn addition of tripropylamine (4 g/L in methanol).
certain extents explaining the relative standard deviation that generally exceeded 20%. One of the difficulties that has hampered progress in NI ESP-MS is electric (corona) discharge, presumably caused by electrons emanating from the sharp edges of the ESP capillary needle held at a few thousand volts negative relative to a counter electrode. A breakdown to discharge at the tip of the capillary needle occurs at much lower field strength in the case of negative polarities than for positive polarities. Once the corona discharge occurs, the efficiency of extraction the ions from the solution to the gas phase decreases dramatically. This is because the plasma itself is a good electric conductor, and the high electrical field for ion separation at the tip of the capillary disappears with the occurrence of the dicharge. For aqueous solutions, it becomes very difficult to keep stable electrospray ionization @I) in the case of negative polarities due to the higher voltage needed to electrospray the liquid. This situation makes NI ESP-MS less applicable than the PI mode as reported in refs 20 and 21. The signal instability of ESI can be reduced by adding chloroacetonitrile to the LC eluent. It is always desirable to achieve a stable signal. The use of chlorinated solvents which
act as electron scavengers in the ESP gas has been reported.20t21 By adding these solvents, the electrical (corona) discharge phenomenon is suppressed, leading to a stable ESI with finer charged liquid droplets. This observed disharge suppression has been attributed to electroncapture processes. However, the whole process is not well understood since ESI is affected by many physical properties, .e.g., viscosity, surface tension, vapor pressure, mobility of the ions, dielectric constant, solvating power toward ions at the interfacial region and in the bulk solution, among other parameters, as reported.2I Chloroacetonitriledissolved in the mobile phase (2%) has partly solved the problems encountered with the ion instability, thus reducing the RSD values to the range of 14-17% (results not reported). Nevertheless, the improvement in ion stability is obtained at the expense of sensitivity (1order of magnitude less) due to the [M Cll- adduct ion formation and to the higher intensity of the solvent cluster ions. Chlorine attachment occurred in a way similar to that in TSP-MSz2for the phenoxy acids with a relative abundance of 8% in the best case. Since the major analytical requirement of the present work is to achieve a low detection limit, chloroacetonitrilewas not used for further analysis. Another problem encountered in this work was the identification of 6 and ghydroxybentazone. We decided to increase the extraction voltage of the ESP interface in order to generate diagnostic fragmentation patterns of 6 and 8-hydroxybentazone to confirm the presence of either one or both metabolites in real water samples (see Figure 1). The extraction voltage was varied from 20 to 55 V, leading to an improvement of the structural information when higher voltages were used. In this respect, it was shown that 6 and 8-hydroxybentazone exhibited the same main fragmentation patterns [M - MI-, [M - 1071-, and [M 1491-, which can be attributed to [M - Sod-, [M - SOZ CH(CH&]-, and [M - SO2 - OCNCH(CH&]- ions, respectively. It was noticed that 8-hydroxybentazone generated the [M - 641fragment ion at extraction voltages of 28 V, whereas for 6hydroxybentazone, the same fragment appeared only at extraction voltages exceeding 35 V.
+
(20) Cole, R B.; Harrata, A K. Rapid Commun.Mass Spectrom. 1 9 9 2 , 6 , 5 3 6 539. (21) Hiraoka, K.; Kudaka, I. Rapid Commun. Mass Specfrom. 1992, 6, 265268. (22) Vreeken, R J.; Brinkman, U. A Th.; De Jong, G. J; Barcelo, D. Biomed. Mass Spectrom. 1 9 9 0 , 19, 481-492.
Analytical Chemistty, Vol. 67,No. 9,May 1, 1995
1639
Table 3. Calibration Data for Selected Pesticides and LODs using Time Scheduled SIM after Preconcentrationof 50 mL of Ebro River Water.
andyte
calibrationequation
Ghydroxybentazone y = 1.42 + 7.1lX 8-hydroxybentazone y = 1.42 + 7 . 1 H
benazolin bentazone 2,4-D
MCPA MCPP MCPB
R2
0.917 0.917 y = 0.74 4.85X 0.947 y = 1.24 + 9.15X 0.935 y = 0.99 8.04X 0.956 y = 1.87 19.62X 0.958 y = 4.71 31.99X 0.942 y = 1.88 + 14.25X 0.938
+ + + +
LOD
av RSD
(ug/L)
78 76 80 82 74 83 98 83