Sample Preparation Based on Dynamic Ion-Exchange Solid-Phase

The newly established enrichment technique, dynamic ion-exchange solid-phase extraction (DIE-SPE), was studied for sample preparation for GC/MS analys...
1 downloads 0 Views 95KB Size
Download PDF
Anal. Chem. 2000, 72, 3077-3084

Sample Preparation Based on Dynamic Ion-Exchange Solid-Phase Extraction for GC/MS Analysis of Acidic Herbicides in Environmental Waters Nanqin Li and Hian Kee Lee*

Department of Chemistry, National University of Singapore, 3 Science Drive 3, Republic of Singapore 117543

The newly established enrichment technique, dynamic ion-exchange solid-phase extraction (DIE-SPE), was studied for sample preparation for GC/MS analysis of 16 acidic herbicides in environmental waters. C18 bonded silica was the solid-phase material used. The optimal sample pH was weakly acidic to neutral. However, for common tap water and surface water, which run pH 6-9, all the acidic herbicides except for Chloramben could be effectively extracted from a sample of 1000-mL volume without pH adjustment. The humic acid could be concurrently extracted from water, but most of it was separated from the sample by using 3 mL of 10% methanol in acetone as the eluent, which would completely elute the analytes and leave a large part of the humic acid on the cartridge. The selective elution reduced the interference of humic acid and made the DIE-SPE an effective approach for the analysis of the acidic herbicides in surface water. Comparing DIE-SPE with conventional reversedphase SPE (RP-SPE), the former gave higher recoveries for the acidic herbicides and was less affected by sample matrixes. A tandem-cartridge system combining RP- and DIE-SPE in sequence was set up for the simultaneous isolation of the acidic herbicides and removal of the interfering substances. Despite some minimal retention on the upper RP-SPE cartridge, most of the acidic herbicides could be extracted on the lower DIE-SPE cartridge with recovery over 80% except for Chloramben (50%), fenoprop (73%), MCPB (67%), and 2,4-DB (70%) when a 500-mL aqueous sample of pH 9.5 was percolated through the tandem-cartridge system. The effectiveness of the system in removing the long carbon chain fatty acids as well as the basic and neutral organic interfering substances from the sample was also demonstrated. Acidic herbicides are mainly chlorophenoxy acids and related compounds which are widely used in agriculture and forestry for weed control. When released to the global environment, they may generate certain toxicological effects on human beings and aquatic life. EPA method 515.1 for acidic herbicides in drinking water makes use of liquid-liquid extraction (LLE) followed by gas 10.1021/ac991410c CCC: $19.00 Published on Web 06/06/2000

© 2000 American Chemical Society

chromatography-electron capture detector (GC-ECD).1 Solid-phase extraction (SPE) with adsorbent materials of ion exchangers,2-6 reversed-phase bonded silica,7-12 polymers,13-16and graphitized carbon black (GCB),17-20 has been widely investigated for sample preparation of the acidic herbicides in water prior to chromatographic determination. Of the numerous organic pollutants, the acidic herbicides are among those most compatible to SPE instead of conventional LLE as the sample preparation technique for various environmental waters. This is because, in common water matrixes, those acidic herbicides are ionized and show a low tendency of adsorption on the surface of glassware. Free filtration is possible before the sample is applied to the SPE cartridge or disk and clogging of the adsorbent bed can therefore be avoided, which is often the problem with SPE methods for surface waters. For multiresidue analysis of the acidic herbicides, there are two practical problems for the existing SPE techniques. First, the hydrophilic acidic (1) Engell, K. W.; Erb, E. J.; Wesselman, R. J.; Longbottom, J. E. J. AOAC Int. 1993, 76, 1098-1112. (2) Renberg, L. Anal. Chem. 1974, 46, 459-461. (3) Richard, J. J.; Fritz, J. S. J. Chromatogr. Sci. 1980, 18, 35-38. (4) Field, J. A.; Monohan, K. Anal. Chem. 1995, 67, 3357-3362. (5) Chatfield, S. N.; Croft, M. Y.; Dang, T.; Murby, E. J.; Yu, G. Y. F.; Wells, R. J. Anal. Chem. 1995, 67, 945-951. (6) Herna´ndez-Mateos, M. A.; Pe´rez-Arribas, L. V.; Navarro-Villoslada, F.; Leo´nGonza´lez, M. E.; Polo-Die´z, L. M. J. Liq. Chromatogr. Relat. Technol. 1999, 22, 695-704. (7) Wells, M. J. M.; Michael, J. L. Anal. Chem. 1987, 59, 1739-1742. (8) Hoke, S. H.; Brueggemann, E. E.; Baxter, L. J.; Trybus, T. J. Chromatogr. 1986, 357, 429-432. (9) Chiron, S.; Papilloud, S.; Haerdi, W.; Barcelo´, D. Anal. Chem. 1995, 67, 1637-1643. (10) Balinova, A. J. Chromatogr., A 1996, 728, 319-324. (11) Vink, M.; Van der Poll, J. M. J. Chromatogr., A 1996, 733, 361-366. (12) Butz, S.; Heberer, T.; Stan, H.-J. J. Chromatogr., A 1994, 677, 63-74. (13) Hodgeson, J. W.; Collins, J.; Bashe, W. J. Chromatogr., A 1994, 659, 395401. (14) Coquart, V.; Hennion, M.-C. Sci. Total Environ. 1993, 132, 349-360. (15) Geerdink, R. B.; Graumans, A. M. B. C.; Viveen, J. J. Chromatogr. 1991, 547, 478-483. (16) Chiron, S.; Martinez, E.; Barcelo´, D. J. Chromatogr., A 1994, 665, 283293. (17) Di Corcia, A.; Marchetti, M.; Samperi, R. Anal. Chem. 1989, 61, 13631367. (18) Crescenzi, C.; Di Corcia, A.; Marchese, S.; Samperi, R. Anal. Chem. 1995, 67, 1968-1975. (19) Cappiello, A.; Famiglini, G.; Berloni, A.; Bruner, F. Environ. Sci. Technol. 1995, 29, 2295-2300. (20) Bucheli, T. D.; Gru ¨ ebler, F. C.; Mu ¨ ller, S. R.; Schwarzenbach, R. P. Anal. Chem. 1997, 69, 1569-1576.

Analytical Chemistry, Vol. 72, No. 14, July 15, 2000 3077

herbicide components are difficult to extract or the extraction efficiency is easily affected by the sample matrixes.12 Second, the naturally occurring humic acid and small molecular organic compounds are often concurrently extracted and eluted with the analytes and thus interfere with subsequent chromatographic determination.21 A selective SPE that would remove most of the humic acid or other interfering substances is preferred for trace analysis of the acidic herbicides in surface water or other complex environmental waters. Dynamic ion-exchange solid-phase extraction (DIE-SPE) was established previously22 as the enrichment approach for ionogenic organic compounds. The process is based on the effective adsorption of long carbon chain surfactant (for example, cetyltrimethylammonium bromide, CTAB) on the surface of C18 bonded silica, where the ionic functional group of the surfactant could act as ion-exchange sites to attract the ionized analytes from water. One important characteristic of the methodology is that the DIESPE can extract organic analytes on the basis of the mechanism of ionic electrostatic interaction and hydrophobic interaction and, therefore, has apparently higher recoveries for the ionogenic components as compared to the conventional reversed-phase solidphase extraction (RP-SPE). Additionally, the extraction efficiency is less affected by the nature of the sample matrix. When DIESPE was used as an alternative to conventional ion-exchange SPE, it could be conducted without the problems of slow adsorption kinetics, the negative effect of inorganic ions, and the poor desorption of organic analytes from the solid-phase adsorbent.22 DIE-SPE was shown to be an effective enrichment approach for analysis of phenolic compounds in environmental aqueous samples.22 As an analytical method for water, DIE-SPE has more advantages for acidic herbicides than for phenolic compounds, since the former generally have low pKa values. In common environmental waters, they exist in ionic form and can be directly extracted through ion-exchange electrostatic interaction, which was demonstrated to be more efficient in attracting the ionogenic analytes than reversed-phase interaction in attracting their corresponding neutral molecules.22 In this work, the optimal extraction efficiency and selectivity conditions for isolating the acidic herbicides with DIE-SPE were investigated, while the mechanism for DIE-SPE was developed in terms of the property of the solidphase material during optimization of the procedure. Subsequently, a tandem-cartridges system combining the RP- and DIESPE was evaluated for simultaneous removal of the small-molecule interfering substances and isolation of the analytes from complex environmental waters. EXPERIMENTAL SECTION Reagents and Chemicals. Authentic acidic herbicides were obtained from several companies. They were listed as follow, along with their pKa values in parentheses. 3,5-Dichlorobenzoic acid (3.54), 4-chloro-o-tolyloxyacetic acid (MCPA, 3.13), 4-(2,4-dichlorophenoxy)butyric acid (2,4-DB, 4.95), 4-chlorophenoxyacetic acid (4-CPA, 3.56), 2,4,5-trichlorophenoxyacetic acid (2,4,5-T, 2.83), 4-amino-3,5,6-trichloropicolinic acid (picloram, 3.60), and 2-(2,4,5trichlorophenoxy)propionic acid (Fenoprop, 4.41) were from (21) Ferrer, I.; Barcelo´, D. Trends Anal. Chem. 1999, 18, 180-192. (22) Li,. N. Q.; Lee, H. K. Anal. Chem. 1997, 69, 5193-5199.

3078

Analytical Chemistry, Vol. 72, No. 14, July 15, 2000

Aldrich (Milwaukee, WI). 2,4-Dichlorophenoxyacetic acid (2,4D, 2.87) was from Fluka AG (Buchs, Switzerland). 2-(2,4-Dichlorophenoxy)propionic acid (Dichlorprop, 3.10) was from Tokyo Kasei (Tokyo, Japan). 3-Isopropyl-2,1,3-benzothiadiazinon-(4)-2,2dioxide (bentazon, 3.20), 3,6-dichloro-2-methoxybenzoic acid (dicamba, 1.90), 2-(4-chloro-2-methyl-phenoxy) propionic acid (mecoprop, 3.10), 4-(4-chloro-2-methyl-phenoxy)-butyric acid (MCPB, 6.20), 3,5,6-trichloro-2-pyridyloxyacetic acid (triclopyr, 2.68), 2,3,6trichlorophenylacetic acid (chlorfenac, 3.70), and 3-amino-2,5dichlorobenzoic acid (Chloramben, 3.40) were from Accustand (New Haven, CT). The above herbicides were weighed and dissolved in methanol for individual standard solutions of 2.0 mg/ mL. A composite working standard was prepared by mixing 200 µL of each herbicide standard solution and diluting to 50 mL with methanol. 2,4-Dichlorophenylacetic acid from Fluka AG was used as the surrogate standard. The pKa value for this compound is unavailable from direct sources, but can be speculated to be ∼2.9 from the pKa values of a series of structurally related compounds. The surrogate standard was prepared in methanol for a solution of final concentration 8.0 mg/L. Water samples were spiked with the working standard and surrogate standard before being applied to the cartridges. 4-Bromobiphenyl from BDH (Poole, Dorset, U.K.) was used as internal standard for GC/MS analysis. It was prepared in methanol at a concentration of 8.0 mg/L. The fatty acids were from an Aldrich kit, which was made up of straightchain fatty acids from C6 to C30. The standard mixture of 40.0 mg/L for each of the components from C6 to C18 was prepared in methanol. All organic solvents were of pesticide or HPLC grade from Fisher Scientific (Pittsburgh, PA) and J. T. Baker (Phillipsburg, NJ). MilliQ water was used as reagent water. For SPE, Bakerbond C18 bonded silica cartridge (500 mg, 6 mL) were purchased from J. T. Baker. CTAB, analytical reagent grade, was from Tokyo Kasei and was prepared as a 0.025 M aqueous solution before use. As the experimental samples, the tap water and surface water were taken locally. Humic acid sodium salt for the preparation of artificial water samples was from Aldrich. For sample derivatization and cleanup, N-methyl-N-nitroso-p-toluenesulfonamide (Diazald), analytical reagent grade, was from Aldrich. Carbitol was from Tokyo Kasei, purity of >99%. Potassium hydroxide, analytical reagent grade from E. Merck (Darmstadt, Germany), was prepared as a 37% aqueous solution. Anhydrous Na2SO4 from E. Merck was used as the drying reagent. Before use, the salt was heated in a muffle furnace at 550 °C for 12 h and acidified with the procedure as described.1 Glass wool and Florisil were purchased from Supelco (Bellefonte, PA). Florisil was the adsorbent for the sample cleanup and was activated at 150 °C for 24 h. Procedure. SPE was carried out under positive pressure; the apparatus was the same as that described previously.22 Before sample application, the cartridge was cleaned with 5 mL of methanol and air-dried for 1 min. Another 5 mL of methanol was used to condition the cartridge, followed by 5 mL of H2O and 5 mL (or the volume indicated) of 0.025 M CTAB solution in sequence. After sample application, the cartridge was air-dried for 10 min. The analytes on the cartridge were eluted with 3 mL of methanol/acetone (10:90). The eluate was then acidified with 10 µL of concentrated HCl and dried with 1 g of acidified anhydrous Na2SO4 packed in a disposable glass pipet (230 mm).

Another 1 mL of methanol was used to rinse the collecting tube and pass through the Na2SO4 drying column. The eluate was then concentrated at 45 °C with a nitrogen stream until just the disappearance of the last drop of solution. The derivatization procedure was from Edgell et al.1 It was also the standard procedure of EPA method 515.1. Before derivatization, the sample was redissolved in 0.25 mL of methanol and 4.75 mL of methyl tert-butyl ether (MTBE). Derivatization was carried out by purging the sample directly with diazomethane. The apparatus and procedure for generation of diazomethane was as described.1 After the sample turned yellow, the purging was stopped and the sample was left to stand for 30 min before the residue diazomethane was destroyed with 0.1∼0.2 g of silicic acid. Some of the CTAB was eluted out and was therefore present in the eluate. After derivatization, a white precipitate was obtained; this might be due to the CTAB. However, this observation needs to be further investigated. In any case, the presence of the precipitate did not interfere with the subsequent analysis because it could be easily eliminated. Sample cleanup was accomplished with 1 g of Florisil packed in a disposable glass pipet. The column was conditioned with 5 mL of 5% methanol in MTBE before applying the sample. The sample was transferred to the column with another disposable glass pipet, and the silicic acid was left in the sample tube. Another 5 mL of 5% methanol in MTBE was used to rinse the sample tube and passed through and eluted the analytes from the column. During the cleanup, the column was not allowed to go dry. After 100 µL of internal standard was added and well-mixed; the sample was concentrated to ∼0.5 mL under 35 °C with nitrogen stream, then transferred to a 1.5-mL sample vial, and adjusted to 1 mL with MTBE. The sample was well-mixed and ready for GC/MS analysis. GC/MS Analysis. The sample analysis was carried out with a Shimadzu QP5000 GC/MS (Tokyo, Japan) equipped with a Shimadzu AOC-17 autosampler and DB-1 fused-silica capillary column (30 m × 0.32 mm i.d., film thickness 0.25 mm, J & W Scientific, Folsom, CA). Helium was used as carrier gas, with flow rate 2.0 mL/min and split ratio 20. A 2-µL sample was injected under splitless mode with injection time of 2 min. The temperatures of the injection port and MS interface were set at 250 and 280 °C, respectively. The GC column temperature program was as follows: an initial temperature 45 °C, held for 2 min, and then 6 °C/min to 180 °C, followed by another ramp of 15 °C/min to 270 °C, held for 1 min. With the GC conditions, all the methyl derivatives of the acidic herbicides and the surrogate standard 2,4-dichlorophenylacetic acid and internal standard 4-bromobiphenyl could be well separated. For MS selected ion monitoring (SIM), the detector voltage was 1.5 kV and one ion was selected as the quantitative ion, while three ions were for confirmation. The retention times and quantitative ions are listed in Table 1. The internal standard method and peak area were for calibration, and linear calibration curves could be obtained for all the analytes, typically R2 > 0.999. The detection limits for the acidic herbicides were from 5 to 20 µg/L with 2 µL of sample injection. With 1000fold concentration, the lowest detection concentrations of the acidic herbicides in water sample were between 0.005 and 0.02 µg/L

Table 1. Retention Times and Quantitative Ions for GC/MS Analysis compounds

retention time (min)

quantitative ion (m/z)

3,5-dichlorobenzoic acid 4-CPA 2,4-dichlorophenylacetic acida dicamba mecoprop MCPA dichlorprop 2,4-D chlorfenac 4-bromobiphenyl (IS) triclopyr Chloramben fenoprop 2,4,5-T MCPB 2,4-DB bentazon picloram

15.28 17.20 17.61 17.89 18.69 18.90 19.90 20.17 20.56 21.00 21.41 22.35 22.66 23.02 23.14 24.21 24.45 25.12

204 200 218 234 228 214 248 234 254 232 271 219 282 270 101 101 212 198

a

Surrogate standard.

RESULTS AND DISCUSSION Effect of Silanols and Sample pH. All the acidic herbicides and the surrogate standard 2,4-dichlorophenylacetic acid have pKa from 1.9 (dicamba) to 6.2 (MCPB). In common tap water and surface water, which have pH values of 6-9,23 those compounds almost all exist as their anions. For LLE and RP-SPE, the acidic herbicides can only be effectively extracted as their neutral molecules after the sample is acidified. In contrast, the ionized acidic herbicides would be more easily extracted with DIE-SPE, because ion-exchange electrostatic interaction rather than the hydrophobic interaction only was more efficient for the extraction of ionogenic organic compounds, as presented previously.22 For DIE-SPE of phenolic compounds,22 sample pH and the amount of CTAB used for conditioning the cartridge were the most important factors influencing extraction efficiency. The sample pH determined the ionization status of the analytes, while the CTAB concentration was related to the number of ion-exchange sites on the solid-phase surface. The reasoning is also true for the extraction of the acidic herbicides. However, the DIE-SPE mechanism derived from the extraction of the phenolic compounds22 did not consider the impact of the residual silanols of the C18 bonded silica on the recoveries of the ionogenic analytes. As elucidated later in this work, the ionization of the silanols was related to the sample pH, and the ionized and negatively charged silanols could influence the adsorption of CTAB and the effective concentration of ion-exchange sites on the surface of the solidphase material. The acidic herbicides generally show much stronger acidity than most of the phenolic compounds; the starting pH values for ionization of the acidic herbicides are in the weakly acidic range. It was found that the pH had a very different impact on the recoveries of the acidic herbicides in comparison to the phenolic compounds. The influence of the pH on the recoveries of the two acidic herbicides, Chloramben and picloram, from 1000 mL of tap water (23) Barcelo´, D.; Hennion, M.-C. Trace Determination of Pesticides and Their Degradation Products in Water; Elsevier: Amsterdam, 1997.

Analytical Chemistry, Vol. 72, No. 14, July 15, 2000

3079

Table 2. Percentage Recoveries of DIE-SPE of Acidic Herbicides from a 10000-mL Water Sample of pH 2.0 on Cartridges Conditioned with 2, 6, and 10 mL of 0.025 M CTAB percentage recoverya

Figure 1. Plots of recoveries of DIE-SPE of Chloramben and picloram against sample pH. Sample: 1000 mL of tap water spiked with the acidic herbicides of 0.8 µg/L; the cartridges were conditioned with 2 or 4 mL of 0.025 M CTAB solution.

is shown in Figure 1; 2 and 6 mL of 0.025 M CTAB were used to condition the cartridges. Among the 16 acidic herbicides, Chloramben and picloram were the most hydrophilic; they showed the lowest recoveries with the conventional RP-SPE as well as DIESPE. In fact, except for Chloramben and picloram, all the acidic herbicides could be consistently and satisfactorily extracted with DIE-SPE from 1000 mL of tap water throughout the pH range that they were ionized. Inspection of the variation of the recoveries of Chloramben and picloram with sample pH revealed that the optimal pH for the extraction of the two compounds was weakly acidic to neutral, which was nearly the lowest pH for the two compounds to be ionized. The low recoveries at pH 2.0 could be attributed to the depression of ionization of the carboxyl groups (as well as partial protonation of the amino groups) of the two herbicides. Under the circumstance, hydrophobic interaction was the governing mechanism for the retention of the two herbicides on the cartridge, which was a weak process because of the hydrophility of the two analytes. However, the consistently decreasing recoveries of the two compounds with increasing pH from neutral to weakly basic is supposed to be a result of the effect of silanols. Residual silanols are part of the surface of reversed-phase bonded silica with typical concentrations of 5-8 µmol/m2.24 As the packing material the for solid-phase extraction cartridge, the bonded silica is generally amorphous with particles having average diameter of 40 µm and surface area of 200-600 m2/g.25 On the basis of their structural characteristics, the silanols can be categorized into three types, i.e., isolated, germinal, and vicinal. Although there is no consistency in the literature in the pKa values of the silanols measured by various methods, it is clear that the three types of silanols have different acidic strengths,26 which results in their progressive ionization and increment of the surface negative charge of the silica with sample pH. For DIE-SPE, the electrostatic interaction between CTAB and the ionized silanols leads to the reduction of effective anion-exchange sites on the solid-phase surface with sample pH when the same amount of the CTAB is used to condition the cartridges. Therefore, the recoveries of Chloramben and picloram were lowered with sample pH throughout the range they were in their anionic form. Although the recoveries of less hydrophilic components were not affected (24) Dorsey, J. G.; Cooper, W. T. Anal. Chem. 1994, 66, 857A-867A. (25) Thurman, E. M.; Mills, M. S. Solid-Phase Extraction, Principles and Practice; John Wiley and Sons: New York, 1998. (26) Nawrocki, J. J. Chromatogr., A 1997, 779, 29-71.

3080 Analytical Chemistry, Vol. 72, No. 14, July 15, 2000

compound

2 mL

6 mL

10 mL

3,5-sichlorobenzoic acid 4-CPA 2,4-dichlorophenylacetic acid dicamba mecoprop MCPA dichlorprop 2,4-D chlorfenac triclopyr Chloramben denoprop 2,4,5-T MCPB 2,4-DB bentazon picloram

103 96 98 98 95 98 92 96 96 95 22 101 98 101 107 108 84

63 80 73 67 79 85 76 83 66 78 23 82 84 90 91 37 73

45 72 63 48 70 68 66 78 47 67 20 72 79 89 86 18 55

a

Spiking level 0.8 µg/L.

by the pH with a sample volume of 1000 mL, it can be speculated that the same recoveries-pH profile for these compounds can be observed when larger volumes of sample are applied. On the other hand, at high pH the negative effect of ionized silanols on recoveries of the acidic herbicides can be compensated to some extent by using more CTAB to condition the cartridge. In fact, when 6 or 10 mL of the CTAB solution was used to condition the cartridges, the picloram could be satisfactorily extracted from the 1000-mL sample even at pH 10.0, although the recovery for Chloramben was still a little low in this case. A systematic change of the experiment conditions found that at low sample pH significant loss of the acidic herbicides appeared when a large amount of CTAB solution was used to condition the cartridges. As comparison, the recovery data of the 16 acidic herbicides from 1000 mL of tap water at pH 2.0 with the cartridges conditioned with, respectively, 2, 6, and 10 mL of 0.025 M CTAB solution are shown in Table 2. At pH 2.0, all the acidic herbicides were supposed to exist as their neutral molecules and hydrophobic interaction was responsible for their retention on the reversedphase cartridge. As can be seen, when 2 mL of 0.025 M CTAB was used to condition the cartridge, the recoveries of the acidic herbicides were not lowered by the presence of CTAB, whereas significant losses of the acidic herbicides occurred when the cartridges were conditioned with 6 and 10 mL of the 0.025 M CTAB solution. The phenomenon is supposed to be the saturation effect of the C18 bonded silica cartridge by the CTAB. In previous work,22 the pH effect on the extraction of alkyl- and chlorophenols was investigated. When sample pH was adjusted to as low as 5.0, at which the phenols were supposed to be in their neutral molecules, no significant lowering of the extraction efficiency of the analytes occurred even when 8 mL of 0.025 M CTAB solution was used to condition the cartridges. Meantime, the loss of the acidic herbicides when a larger amount of CTAB was used to condition the cartridge could be obviated by elevating the sample pH. In fact, no significant saturation effect appeared in a sample

Table 3. Percentage Recoveries of Acidic Herbicides with DIE-SPE and RP-SPE from 1000 mL of Tap Water (Sample 1) and 1000 mL of Tap Water Dissolved with 5 mg of Humic Acid (Sample 2) recovery (mean + RSD%)a compounds 3,5-dichlorobenzoic acid 4-CPA 2,4-dichlorophenylacetic acid dicamba mecoprop MCPA dichlorprop 2,4-D chlorfenac triclopyr Chloramben fenoprop 2,4,5-T MCPB 2,4-DB bentazon picloram

DIE-SPE (sample

1b)

DIE-SPE (sample 2b)

RP-SPE (sample 1b)

RP-SPE (sample 2b)

97 (5.0) 84 (1.8) 101 (2.2) 104 (1.8) 102 (0.6) 103 (0.1) 101 (0.5) 100 (1.3) 101(1.8) 101 (1.5) 58 (12.6) 105 (0.9) 101 (1.5) 102 (2.7) 107 (2.5) 103 (1.9) 95 (0.7)

96 (4.0) 69 (6.8) 94 (6.0) 62 (0.6) 90 (8.5) 89 (7.5) 94 (8.3) 96 (6.5) 109 (2.0) 95 (7.4) 18 (2.1) 92 (7.4) 95 (5.6) 102 (5.3) 93 (4.0) 98 (1.6) 34 (3.3)

72 (1.5) 49 (2.8) 90 (2.1) 59 (6.8) 100 (3.3) 80 (3.7) 99 (2.9) 100 (4.4) 96 (6.3) 87 (4.3) 8 (21.7) 85 (7.5) 86 (6.7) 77 (6.2) 78 (8.8) 77 (7.1) 14 (10.3)

91 (4.4) 85 (4.2) 98 (5.4) 99 (5.7) 99 (5.5) 100 (5.2) 100 (5.4) 97 (3.9) 99 (5.7) 98 (4.6) 56 (15.5) 101 (5.2) 97 (4.7) 98 (5.3) 100 (3.0) 86 (8.8) 89 (13.4)

a Average of three measurements. For DIE-SPE, there was no sample pH adjustment; the cartridge was conditioned with 5 mL of methanol, 5 mL of water, and then 6 mL of 0.025 M CTAB. For RP-SPE, sample pH was adjusted to 2.0, and the cartridge was conditioned with 5 mL of methanol and 5 mL of water. b Spiking level 0.8 µg/L; sample 2 was filtered through Whatman glass filter (1 µm) before being spiked with the acidic herbicides.

of pH >5 even if 10 mL of 0.025 M CTAB was used to condition the cartridge. The above saturation phenomenon of the C18 cartridge by the CTAB could also be explained by the silanol effect. At high pH, the silanols were ionized and negatively charged and electrostatic interaction was an alternative approach for the retention of the CTAB on the cartridge. Conversely, under low-pH conditions, the ionization of silanols was gradually depressed; hydrophobic interaction was almost the only cause for the adsorption of CTAB on the cartridge. With the increase of the amount of CTAB used to condition the cartridge, the ability for the neutral analyte molecules to be retained on the reversed-phase material was lowered as a result of their competitive adsorption with CTAB for the hydrophobic sites on the surface of the C18 bonded silica. The saturation effect of the CTAB was also found for the nonpolar organic pollutants such as polynuclear aromatic hydrocarbons (PAHs); generally it was the smaller molecules, which showed weaker hydrophobic interaction with the reversed-phase solid phase than the heavier components. Anyway, the saturation effect is unlikely to appear in practical environmental analysis, because the pHs of normal environmental waters very seldom fall below 5.0. Method Comparison and Matrix Effect. As established above, in terms of recovery, the optimal pH for DIE-SPE of all the acidic herbicides was in the weakly acidic to neutral range. For common tap water and natural surface water, the acidic herbicides could be directly extracted from the sample without pH adjustment. Under this circumstance, the sample was not necessarily at the optimal pH, but when 6 mL of 0.025 M CTAB was used to condition the cartridge, all the acidic herbicides except for Chloramben, which showed low and possibly variable recovery with pH of samples of different sources, could be satisfactorily extracted from 1000 mL of water. On the other hand, with the conventional RP-SPE, the recoveries of the analytes in surface waters could be lowered by the sample matrixes such as the humic

substance, as found by several authors for organic pollutants of several types.27-29 For this consideration, the potential of the DIESPE for trace enrichment of the acidic herbicides from real environmental samples was evaluated by extracting 1000 mL of spiked tap water and 1000 mL of spiked tap water dissolved with 5 mg/L Aldrich humic acid, respectively. For comparison, the two samples were simultaneously extracted with the RP-SPE after being acidified to pH 2.0. The tap water dissolved with the Aldrich humic acid was used to simulate the effect of the humic acid in real surface water. Because the content and composition of the humic substance in natural surface water varied seasonally and provisionally, direct estimation of the negative effect of the humic acid on the recovery of the organic compounds with a real environmental sample may not be representative. The recoveries of the acidic herbicides from the two samples with the RP-SPE and DIE-SPE are tabulated in Table 3. As can be seen, DIE-SPE appeared to be more efficient than RP-SPE in extracting the acidic herbicides. Among the 16 acidic herbicides considered, 4-CPA, dicamba, Chloramben, and picloram could not be effectively extracted from 1000 mL of tap water with the RP-SPE; the recoveries for Chloramben and picloram were even lower than 35%. In comparison, all the components could be satisfactorily extracted from the sample with DIE-SPE despite low recovery for the Chloramben. The advantages of DIE-SPE over RP-SPE also lie in the fact that the extraction efficiency of DIESPE was little affected by the humic acid in water. Johnson et al.27 had proven that the association of the organic pollutants with the humic acid in water was attributable to the loss of analytes on the conventional RP-SPE, because the analytes associated with the humic acid were not efficiently retained by the reversed-phase (27) Johnson, W. E.; Fendinger, N. J.; Plimmer, J. R. Anal. Chem. 1991, 63, 1510-1513. (28) Jime´nez, B.; Molto´, J. C.; Font, G. Chromatographia 1995, 41, 318-324. (29) Senseman, S. A.; Lavy, T. L.; Mattice, J. D.; Gbur, E. E. Environ. Sci. Technol. 1995, 29, 2647-2653.

Analytical Chemistry, Vol. 72, No. 14, July 15, 2000

3081

material and most of them appeared in the effluent after the sample was applied to the cartridge. The same negative effect of humic acid also appeared in the extraction of the acidic herbicides. As seen, for the RP-SPE, the recoveries of most of the acidic herbicides were significantly lowered by humic acid dissolved in the water. In contrast, the recoveries of the acidic herbicides with DIE-SPE were little affected by the humic acid. Because of the ionization of the carboxyls and hydroxyls in the humic acid molecules, the humic acid in natural surface water is negatively charged and can be effectively extracted onto the DIE-SPE cartridges through the ion-exchange process. Consequently, the analytes associated with the humic acid can be extracted together with the freely dissolved components and then are released when the eluting solvent percolates through the cartridges; thus the negative effect of the humic acid on the extraction of the acidic herbicides can be avoided. Moreover, in the practice of RP-SPE, acidification is a necessary process for extraction of the acidic herbicides. The process however may be eliminated for the DIESPE. The significance of the improvement in operation may lie in the fact that the acidification can not only add extra work to the analytical procedure but also lead to analyte loss through potentially increased adsorption of the acidic herbicides on the wall of the glassware from the lowering of their solubility originating from the change of their ionization status. Eluent Selection and Removal of the Humic Substance. The ability of organic solvents in eluting the acidic herbicides from the DIE-SPE cartridge appeared to be related to their polarity, and the polar solvents show high eluting strength for the acidic herbicides. Initially, methanol was used as the eluent. Although the solvent was powerful in eluting the acidic herbicides, a large part of the humic acid adsorbed to the cartridge was also eluted out with the analytes and appeared in the eluate as the deep-color substance, which was problematical in later chromatographic analysis. Indeed, humic acid is the most extensive natural substance in environmental waters. Even for tap water, the characteristic color of the humic acid in the sample is obvious after 1000-fold concentration. The humic acid content in common surface water is 1-5 mg/L,30 which is much higher than that in tap water. The interference of the humic acid is the reason a number of analytical methods available for organic pollutants in tap water cannot be used for surface water. Nowadays, high-performance liquid chromatography (HPLC) and gas chromatography (GC including GC/ MS) are the dominant instrumental approaches for organic pollutant analysis. The interference of humic acid to the analysis with HPLC was presented as the huge hump at the beginning of the chromatogram, because of it the early-eluted components could be overwhelmed and the detection limits for the later-eluted components were also elevated as the result of the elevated baseline.14,31 The interferences of humic acid with GC analysis of the acidic herbicides also exist. First, the acidic herbicides are not volatile, and derivatization is needed before the GC determination. Humic acid in the sample wastes a lot of the derivatization reagent,11 because the hydroxyl and carboxyl groups in the molecules react similarly to the acidic herbicides. In the case of (30) Thurman, E. M. Organic Geochemistry of Natural Waters; Martinus-Nijhoff: Dordrecht, The Netherland, 1985. (31) Hogendoorn, E. A.; Dijkman, E.; Baumann, B.; Hidalgo, C.; Sancho, J.-V.; Hernandez, F. Anal. Chem. 1999, 71, 1111-1118.

3082

Analytical Chemistry, Vol. 72, No. 14, July 15, 2000

the derivatization approach with diazomethane, the deep orange color of humic acid almost makes the process impossible, because the “end point” for adding the derivative reagent is judged by the change of the color of the sample solution to the characteristic yellow of the diazomethane itself. Second, humic acid can foul the GC column and contaminate the detector when a large amount is introduced into the chromatographic system. The Florisil column cleanup was an effective approach for separating the humic acid before GC analysis of the acid herbicides. However, the capacity of the Florisil column in separating the humic acid is proportional to the amount of adsorbent. A column of large size is therefore needed for a sample of large humic acid content. Consequently, much organic solvent will be needed to rinse the adsorbent and elute the analytes from the column. For these reasons, a selective enrichment approach that can remove the humic substance from the sample before further sample processing and analysis is always preferable. For the acidic herbicides, it is usually not convenient because of the similar acidic nature of the analytes and the humic acid. Recently, a layered adsorbent cartridge32 and double-disk SPE system combined with anion exchanger and C18 bonded silica33 have been reported to be able to remove a large part of the humic acid from the sample for enrichment of some of neutral and basic pollutants, although the same approach has not been reported to be applicable for the acidic herbicides. In this work, it is found that the removal of the humic acid can be realized for DIE-SPE by using a mixture of methanol and acetone as the eluent. In contrast to methanol, acetone has little ability in dissolving the humic acid, while it cannot completely elute the acidic herbicides from the DIE-SPE cartridge. Mixtures of varying ratios of methanol and acetone have been tested for selectively eluting the acidic herbicides and separating the humic acid. When a 3-mL volume of the eluent was used, 10% methanol/90% acetone was the suitable composition for complete elution of the analytes and effective separation of the humic acid. Using this eluent, although a small part of the humic acid could be eluted from the cartridge, the content would not hinder the later derivatization and cleanup of the sample with the method described in the Experimental Section. By further reducing the methanol to 5%, the amount of the humic acid eluted could be lowered even more. However, some 10-15% MCPB and 2,4-DB were found not to be eluted from the cartridge, although other analytes were completely recovered under the circumstance. Tandem C18 Cartridges System. Apart from the humic acid, the potential interfering substances for chromatographic analysis of the acidic herbicides come from various fatty acids and neutral compounds in environmental waters.11,34,35 When DIE-SPE was used for the sample preparation, although part of the highly nonpolar components adsorbed to the cartridges could be retained and separated with the analytes using a polar eluting solvent, most of the organic interfering substances could be concurrently eluted and appeared in the eluate. The interference is of concern under two circumstances: (1) low selective detector is used in the instrumental system; (2) complex environmental samples such (32) Raisglid, M.; Burke, M. F. Abstracts of the Pittsburgh Conference and Exposition on Analytical Chemistry and Applied Spectroscopy. Atlanta, GA, March 1621, 1997; Abstr. 653. (33) Ferrer, I.; Barcelo´, D.;Thurman, E. M. Anal. Chem. 1999, 71, 1009-1015. (34) Schu ¨ ssler, W. Chromatographia 1990, 29, 24-30. (35) Lee, H. B.; Peart, T. E.; Carron, J. M.; Tse, H. J. Assoc. Off. Anal. Chem. 1991, 74, 835-842.

as industrial and agricultural wastewaters are extracted and analyzed. Further cleanup to separate these small molecular interfering substances from the acidic herbicides is normally through column chromatography11 or gel permeation chromatography (GPC),35 which both are laborious and time-consuming processes and are not easy to handle. Selective extraction to separate the interfering substances before isolation of the acidic herbicides is the alternative approach to reduce the later cleanup work. In this respect, the so-called two-step LLE promulgated by USEPA1 and used by other workers34 combines both the processes of separating the interfering substances and isolating the acidic herbicides. Briefly, the sample was adjusted to high pH and first extracted with organic solvent to remove the basic and neutral components, and then it was acidified and the acidic analytes were extracted. Locoto36 proposed a similar method for isolating the acidic herbicides from wastewater: the basic and neutral interfering components were separated with LLE using methylene chloride under basic conditions before the sample was acidified and the acidic herbicides isolated with SPE. In the two procedures described above, the ionization properties of the acidic herbicides had been utilized to separate the analytes from the basic and neutral compounds; that is, the acidic herbicides were little extracted by the organic solvent when they were in the ionic form. On the other hand, the ionized acidic herbicides have the same low distribution tendency toward the reversed-phase bonded materials. For example, Di Corcia et al.17 showed that when ionized the acidic herbicides had very low retention ability on the C18 cartridges. Exploiting the principles of the above procedures, a tandem-cartridge system combining RP-SPE and DIE-SPE in series was constructed as the selective sample preparation technique for the acidic herbicides. The tandem-cartridge system was set up by connecting the two C18 cartridges in series using a plastic adapter, with the upper cartridge in the RP-SPE mode to remove the interfering substance and the lower cartridge in the DIE-SPE mode to isolate the acidic herbicides. Before sample application, the two cartridges were conditioned respectively with the procedure described previously. The efficiency of the tandem-cartridge system was evaluated with respect to the recoveries of the acidic herbicides on the lower cartridge and the potential of the upper cartridge in extracting and removing the fatty acids and other interfering substance from 500 mL of surface water samples. As presented above, the DIE-SPE cartridge in extracting all the ionized acidic herbicides from 500 mL of water was proven to be efficient although the Chloramben may show some low recovery at high pH. Thus, with the tandem-cartridge system, the recoveries of the acidic herbicides are largely dependent on their loss as being retained on the upper cartridges. Apparently, low retention of the acidic herbicides on the upper cartridge could only be effected when the acidic herbicide analytes were in the anionic form. Moreover, the retention studied as the recoveries of the acidic herbicides on the reversed-phase cartridge was also found to be related to pH in the range the analytes were ionized. A comparison of the recoveries obtained by percolating 500-mL spiked water samples adjusted to different pHs of 7.0 and 10.0 through the C18 cartridge is shown in Table 4. Under the two pH values, the acidic herbicides were in their anion forms. It can be seen that the recoveries of all the acidic herbicides at pH 10.0 (36) Locoto, P. R. LC-GC 1991, 9, 460-465.

Table 4. Percentage Recoveries of Acidic Herbicide Anions on Reversed-Phase C18 Cartridgesa compounds

pH 7.0

pH 10.0

3,5-dichlorobenzoic acid 4-CPA 2,4-dichlorophenylacetic acid dicamba mecoprop MCPA dichlorprop 2,4-D chlorfenac triclopyr Chloramben fenoprop 2,4,5-T MCPB 2,4-DB bentazon picloram

5.8 14.8 1.8 ndc 9.9 18.7 10.2 30.6 1.3 12 0.8 49.1 44.3 63.3 82 0.2 6.2

2.3 2.3 2.3 nd 5.6 3.2 4.1 2.6 4.4 3.0 2.9 34.2 5.7 31.8 39.0 nd 8.0

a Conditions: sample, 500 mL of tap water spiked with the acidic herbicides; spiking level, 1.6 µg/L; the cartridges were conditioned with 5 mL of methanol and 5 mL of H2O before sample application. b nd, not determined.

were systematically lower than those at pH 7.0, which is believed to be the result of progressive ionization of the silanols of the C18 bonded silica and the electrostatic repulsion of the ionized silanols with the acidic herbicides anions. For each of the herbicide anions, their retention on the C18 cartridges was related to the molecular structure; those components having more hydrophobic functional groups showed higher recovery on the reversed-phase cartridge. As can be found, the recovery on the C18 cartridge was lower than 10% at pH 10.0 for all the acidic herbicides except for the relatively hydrophobic fenoprop, MCPB, and 2,4-DB. Therefore, a reasonably high pH is necessary to achieve high recoveries for the acidic herbicides with the tandem-cartridge extraction system. On the other hand, the potential of the upper cartridge in removing the interfering substance from the water depends on the retention ability of those compounds on the C18 cartridges. In the environmental analytical methods involving SPE, C18 bonded silica shows high efficiency in adsorbing a lot of nonpolar or moderately polar compounds such as industrial pollutants and pesticides from environmental waters. In this work, it was found that in a large concentration range the ability of C18 cartridges in isolating these neutral compounds was little affected by the sample pH as tested with several PAHs and organochlorine pesticides. It was consistent with the result of Molto´ et al.37for the polychlorinated biphenyls (PCBs). On the other hand, for trace analysis of the acidic herbicides with GC, the most important interfering substance may be the various straight- or branch-chain fatty acids,35 which exist widely in natural surface waters. The fatty acids are of many kinds and almost all have pKa close to 5.0 despite the difference in carbon chain length and structure,38 and in aqueous matrixes of pH >7.0, they exist in their anion forms. There is no report for direct extraction of fatty acids in water with reversed-phase bonded material. However, direct isolation of ionic compounds of similar structure from water with the C18 bonded (37) Molito´, J. C.; Pico´, Y.; Man ˜es, J.; Font, G. J. AOAC Int. 1992, 75, 714-719. (38) Taylor, M. A.; Princen, L. H. In Fatty Acids; Pryde, E. H., Ed.; The American Oil Chemists Society: Champaign, IL, 1979; pp 195-217.

Analytical Chemistry, Vol. 72, No. 14, July 15, 2000

3083

Table 5. Percentage Recoveries (Lower Cartridge) of Acidic Herbicides with Tandem Cartridges System for 500 mL of Spiked Surface Water

Figure 2. Plots of recoveries of RP-SPE of fatty acids against sample pH. Sample: 500 mL of MilliQ water having 0.001 M phosphate salt. Fatty acids spiking level, 2.0 µg/L.

silica was feasible and recovery was found to be related to their molecular structures.39 Generally those molecules having long carbon chain lengths are easily extracted. For example, bonded silica has been proven to be efficient for extracting linear alkyl sulfonate (LAS) surfactants from water.39 The potential of the C18 cartridges in removing the fatty acids from water were herein studied with 11 straight carbon chain fatty acids from C8 to C18 as the model compounds. In this experiment, 500 mL of reagent waters fortified with a mixture of the 11 fatty acids and having different pHs from 7 to 10 were extracted with the C18 cartridges; recoveries were obtained for all the fatty acids. For conciseness, the recoveries of fatty acids of C8-C13 were plotted against the pH, as presented in Figure 2. As we can see, although the shorter carbon chain fatty acids including C8, C9, and C10 show unexceptionally and consistently decreasing recoveries with the increasing sample pH, those having carbon chains longer than C11 (including C11 itself) can be satisfactorily extracted from 500 mL of water throughout the pH range of 7-10. In GC analysis of the acidic herbicides, the methyl silicone (for example, DB-1) and 5% phenyl methyl silicone (DB5) capillary columns are the most widely used. The fatty acids having carbon chains from C12 to C18 always have the same time windows as the acidic herbicides35 and are potential interfering substances for their analysis. Now that those fatty acids could be satisfactorily extracted by the upper C18 cartridge, their interference to GC analysis of the acidic herbicides can be reduced when the tandem-cartridge system is used as the sample preparation approach. The extraction efficiency of the tandem-cartridge system was evaluated by percolating 500 mL of surface water through the apparatus. Before extraction, the sample was spiked with the acidic herbicides, adjusted to pH 9.5, and filtered with a Whatman glass fiber filter. The recoveries obtained for each of the acidic herbicides are listed in Table 5. As can be seen, except for (39) Marcomini, A.; Di Corcia, A.; Samperi, R.; Capri, S. J. Chromatogr. 1993, 644, 59-71. (40) Pichon, V.; Chen, L.; Hennion, M.-C.; Daniel, R.; Martel, A.; Le Goffic, F.; Abian, J.; Barcelo´, D. Anal. Chem. 1995, 67, 2451-2460. (41) Bjarnason, B.; Chimuka, L.; Ramstro¨m, O. Anal. Chem. 1999, 71, 21522156.

3084 Analytical Chemistry, Vol. 72, No. 14, July 15, 2000

compounds

reca

compounds

reca

3,5-dichlorobenzoic acid 4-CPA 2,4-dichlorophenylacetic acid dicamba mecoprop MCPA dichlorprop 2,4-D chlorfenac

90 83 88 88 79 82 83 81 91

triclopyr Chloramben fenoprop 2,4,5-T MCPB 2,4-DB bentazon picloram

89 50 73 87 67 70 97 87

a Conditions: sample, 500 mL of surface water, spiking level 1.6 µg/L, adjusted to pH 9.5 and filtered through 0.45-µm PTFE membrane. The upper cartridge was conditioned with 5 mL of methanol and water; the lower cartridge was conditioned with 5 mL of methanol and 5 mL of water and then 10 mL of 0.025 M CTAB solution.

Chloramben, which is hydrophilic, and fenoprop, MCPB, and 2,4DB, which have some retention on the upper cartridge, all the acidic herbicides have recoveries near or over 80% with this extraction system. CONCLUSION The high efficiency of DIE-SPE in isolating the 16 acidic herbicides from tap water and surface water was demonstrated after the procedure was established on the basis of a wellrecognized mechanism. From a practical point of view, the distinct value of the DIE-SPE is based on two aspects. First, it showed higher recoveries for the acidic herbicides and the recoveries were little affected by the sample matrixes. Therefore, the procedure was valuable as a trace enrichment technique for some of the ionogenic and hydrophilic organic pollutants. To date, there have not been many satisfactory approaches in isolating them from aqueous sample, especially complex surface water. Second, with DIE-SPE, interference by humic substances could be reduced. With conventional SPE approaches, the presence of humic acid is problematic for the analysis of the organic pollutants in natural surface water. For this reason, recently, environmental analytical chemists have shown much interest in developing new SPE materials such as immunoadsorbents40 and molecularly imprinted polymers (MIP)41 for the purpose of selective enrichment of the objective pollutants. DIE-SPE described in this work provides another potential approach that can solve similar problems and is also simple in concept. As demonstrated here, DIE-SPE was used as the selective method, and the selectivity could be further improved by the tandem-cartridges system combining the RPSPE and DIE-SPE processes. ACKNOWLEDGMENT The authors acknowledge the financial support for this work by the National University of Singapore, and the loan of the GC/ MS system by Shimadzu Asia-Pacific Private Limited. Received for review December 9, 1999. Accepted April 19, 2000. AC991410C