Anal. Chem. 1991, 63,580-585
580
Multiresidue Method for Pesticides in Drinking Water Using a Graphitized Carbon Black Cartridge Extraction and Liquid Chromatographic Analysis Antonio Di Corcia* and Marcello Marchetti Dipartimento di Chimica, Uniuersitci “La Sapienza” di Roma, Piazza Aldo Moro 5, 00185 Roma, Italy
A general liquid-solid extraction procedure for the isolation of pesticides from groundwater and drinking water for highperformance liquid chromatography (HPLC) is presented. This simple and rapid procedure involved passing a 2-L Sampie through a 250-mg graphitized carbon black (Carbopack B) cartridge at a flow rate of 150-160 mL/min. By taking advantage of the presence of positively charged active tenters on the Carbopack B surface, a stepwise elution system allowed the complete separation of base-neutral pesticides from acidic ones. Afler partial solvent removal, the components in the two fractions were separated and quantified by gradient elution, reversed-phase HPLC with ultraviolet (UV) detection. The performance of the Carbopack cartridge was compared with that of a 500-mg C 1 8 bonded silica cartridge. With the Carbopack cartridge, the grand mean measurement accuracy of the 35 pesticides considered was 95 % . With the C-18 cartridge, the grand mean measurement accuracy of the analytes was 7 6 % . Compared to the C-18 cartridge, additional advantages of using a Carbopack cartridge are that the extraction procedure is about 7 times shorter, no pH adjustment of the environmental sample is necessary for trapping acidic compounds, and one cartridge instead of two suffices to extract base-neutral and acidic pesticides, making the Carbopack cartridge more adaptable than the C-18 one for field use. The detection limits by this method of ail the pesticides considered were between 0.003 and 0.07 pg/L.
Published information shows that the total amount of pesticides for both agricultural and industrial purposes is increasing. As a consequence of the amounts used, pesticides are present in all compartments of the natural environment (1-5).
The monitoring of pesticides in environmental matrices poses substantial analytical problems as, in response to recent legislation in many countries, the demand is for methods of greater sensitivity. Ideally, deployment of few, inexpensive multiresidue (MR) methods should facilitate the rapid identification and quantification of a wide range of pesticides a t the required sensitivity limit. If necessary, validation of the presence of tentatively identified pesticides could then be followed up by more accurate methods calibrated for specific pesticides, such as those employing mass spectrometry and immunoassay techniques. Although capillary column gas chromatography (GC) remains the major determinative technique, the number of publications describing reversed-phase gradient elution high-performance liquid chromatography (HPLC) for pesticides is steadily increasing (6-11). The properties of relatively new classes of pesticides, such as phenylureas, phenoxyacids, quaternary amines, and carbamates, make them more amenable to HPLC than to GC.
* To whom correspondence should be addressed.
A traditional approach to the preparation of the water sample has been to utilize liquid-liquid extraction (LLE). For many analysts, this technique has proved cumbersome, has proved time consuming, and involves large volumes of expensive and flammable solvents. In addition, many pesticides tend to be more water soluble than their predecessors and thus less amenable to extraction with an organic solvent. All these shortcomings have led to the development of liquid-solid extraction (LSE) techniques with small sorbent cartridges that offer combined sampling and concentration, reduced sample handling, and adaptability to field use. Among the sorbents available for LSE of pollutants from water, octadecyl (C-18) bonded porous silica has become the most popular. However, a comparative study performed by Bellar and Budde (11) has shown that the extraction efficiency of a C-18 cartridge for polar pesticides in water is lower than that of the LLE technique. Carbopack B is a well-known graphitized carbon black (GCB) having a nonporous, substantially nonpolar, and homogeneous surface with an area of about 100 m2/g. This adsorbent has proved to be a valuable stationary phase in both packed and capillary GC columns. I t has been successfully used for the LSE of various analytes from biological fluids (12-17) as well as of organochlorine insecticides (18, 19), triazine (20),and phenoxyacid (21) herbicides from water. Carbopack B cartridges proved to be more efficient than C-18 cartridges for extracting phenols (22)and chloroanilines (23) from water. With respect t o chemically bonded silica, an apparent weakness of GCB is that its surface framework is contaminated by few oxygen complexes, having a structure similar to benzpyrylium salts (24,25),able to interact so strongly with sufficiently acidic compounds that conventional solvent systems are incapable of desorbing them. This apparent weakness is, in fact, an advantage in that acidic analytes can be completely isolated from base-neutral ones by first passing through the sorbent bed a solvent system for eluting nonacidic compounds and then a suitably basified solvent mixture for desorbing acidic analytes, which are collected apart. In this way, not only extraction and concentration but also class fractionation can be achieved simultaneously by a single sorbent cartridge. Recently, a similar analytical scheme was successfully adopted for extracting and fractionating estrogens and their acidic metabolites in biological fluids (17). The purpose of this work was to explore a new approach to the development of a MR method for determining pesticides present in environmental water samples. The primary objective was to evaluate the ability of a Carbopack cartridge in rapid and quantitative extraction from large water volume pesticides, whose chemical nature may greatly vary. For this purpose, we selected 35 pesticides, which comprise a large range of polarity, and they are important environmental pollutants. EXPERIMENTAL SECTION Reagents and Chemicals. Authentic phenoxyacids were purchased from Eurobase (Milan, Italy). All other pesticides
0003-2700/91/0363-0580$02.50/00 1991 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 63, NO. 6, MARCH 15, 1991
considered in this work were from Riedel-de Haen, Seelze, West Germany. Individual standard solutions were prepared by dissolving 100 mg of each pesticide in 100 mL of methanol. A composite working standard solution of base-neutral pesticides was prepared by mixing appropriately 5G-150 pL of each pesticide standard solution and diluting to 10 mL with methanol. A second composite working standard solution was prepared by mixing 100 pL of each acidic pesticide standard solution (50 pL of that containing bromoxynil) and diluting to 10 mL with methanol. For HPLC, distilled water was further purified by passing it through a Norganic cartridge (Millipore, Bedford, MA). Acetonitrile of HPLC grade was from Baker, Deventer, Holland. All other solvents were of analytical grade (Carlo Erba, Milan, Italy). Trifluoroacetic acid (TFA) was supplied by Fluka AG, Buchs, Switzerland, and ascorbic acid was from Carlo Erba. Apparatus. Two-hundred fifty milligrams of Carbopack B (120-400-mesh size) was packed in a polypropylene tube, 6.5 x 1.4 cm i.d. Polyethylene frits, 20-pm pore size, were located above and below the sorbent bed. To avoid crushing of the Carbopack B particles, which results in a decrease of the permeability of the cartridge, the upper frit was placed gently on the sorbent bed. Before water samples were processed, the cartridge was washed with 5 mL of methylene chloride/methanol (80:20 by volume) followed by 2 mL of methanol and 15 mL of 10 g/L ascorbic acid in HC1-acidified water (pH 2). The trap was fitted into a side arm filtering flask, and liquids were forced to pass through the cartridge by vacuum from a water pump. Procedure. Aqueous samples were fortified with known volumes of both the acid and base-neutral working standard solutions. When water-containing ipochlorite was analyzed, 0.4 g of sodium sulfite/L of water was added to avoid oxidation of some pesticides and other unwelcome effects, which will be discussed later. Water samples were then agitated for 1 min and, after 10 min, poured in a glass reservoir that was connected to the sorbent cartridge. Water was forced to pass through the cartridge a t flow rates of 150-160 mL/min by reducing the pressure in the vacuum apparatus to the minimum. Just after the sample was passed through the column, the water pump was disconnected and the cartridge was filled with 7 mL of distilled water in order to remove sample water drops stuck on the plastic walls of the cartridge, whose presence can decrease the efficiency of the organic solvent systems used for desorbing analytes. The pump was, then, again linked to the flask and the pressure suitably adjusted to allow water to pass through the cartridge at flow rates of 10-15 mL/min. Following the passage of large water volumes, some shrinking of the sorbent bed may occur. In this case, before washing with distilled water, the upper plastic frit was pushed against the top of the sorbent bed. This expedient procedure facilitates the subsequent removal of water from the sorbent bed and enhances the effectiveness of the eluant systems, as they can permeate the sorbent bed in a more homogeneous way. After 7 mL of distilled water was passed through the trap, the major part of it was removed by pulling air through the cartridge for 1 min. Then the water pump was disconnected, a round-bottom glass vial having 1.4-cm i.d. was located below the cartridge, and the base-neutral pesticides were eluted by passing through the trap drop by drop 1 mL of methanol followed by 6 mL of methylene chloride/methanol (80:20 by volume). The last drops of this solvent mixture were collected by decreasing the pressure in the flask. The vacuum was again suitably regulated, and the fraction containing the acid analytes was collected in a second glass vial by percolating through the trap drop by drop 6 mL of methylene chloride/methanol (60:40, by volume) basified with 0.016 mol/L KOH. The last drops of this eluant system were forced to come out of the cartridge by an additional decrease of the pressure in the flask. Prior to concentration, the extract containing the acidic pesticides was acidified by adding 0.35 mL of 2% (v/v) TFA in water. The extracts containing the acid and the base-neutral pesticides were then concentrated to obtain final volumes of about 0.3 and 0.5 mL, respectively, by evaporation in a water bath at 38 "C under a gentle stream of nitrogen. After the exact final extract was measured, volumes, 25 and 40 pL respectively, of the base-neutral and acid extracts were injected into the HPLC apparatus. HPLC. Liquid chromatography was carried out with a Varian (Walmut Crrek, CA) Model 5000 equipped with a Model 2550
581
UV detector (Varian). A 25-cm X 4.6-mm4.d. column filled with 5-pm LC-18 reversed-phase packing (Supelco) was used. For separating base-neutral pesticides, the initial organic modifier (acetonitrile) percentage was 2070,which was increased linearly to 70% within 30 min. For the acidic compounds, the initial mobile-phase composition was water containing 0.0570 (v/v) TFA and acetonitrile containing 0.025% (v/v) TFA (6040, by volume). This composition was varied linearly to reach 67% acetonitrile within 20 min. In both cases, the flow rates of the mobile phases were 1.5 mL/min. Basic and neutral pesticides were detected with the UV detector set at 225 nm, while 230 nm was the wavelength selected for detecting acidic pesticides. The concentrations of the pesticides in water were calculated by measuring the peak height of each pesticide and comparing them with those obtained with standard solutions. These were prepared by taking known and appropriate volumes of the working composite standard solutions, by evaporating methanol and reconstituting the residue with 0.5 mL of water/methanol(50:50, by volume) for baseneutral extract and 0.3 mL of water/methanol (5050, by volume) acidified with 0.25% (v/v) TFA for acids.
DISCUSSION OF RESULTS T h e number of pesticides used in agriculture are many hundreds. For this study, which was regarded by us as a n approach to the development of a MR method by HPLC with UV detection of pesticides in water, we selected a limited number of analytes by using the criteria of taking into consideration those pesticides that are UV absorbers, are poorly amenable to analysis by standard GC methods, and represent classes of pesticides of wide use in agriculture. According to these criteria, Table I lists the selected pesticides together with their HPLC retention times. When using untreated Carbopack for extracting pesticides spiked in distilled water at the 0.2 pg/L level, losses of metribuzin and chloridazon were observed. No trace of these two compounds was found in the water effluent. On the other hand, recovery of these two pesticides was complete when extracting water samples with a 10 times higher concentration. Washing the cartridge with a n acidic aqueous solution of ascorbic acid prior to extraction had the effect of completely eliminating the losses a t low concentrations. In the past, polarographic measurements gave us experimental evidence for the existence on the Carbopack surface of hydroquinone groups that are to some extent in equilibrium with their oxidized forms, namely semiquinones and quinones (25). In order to ascertain whether the surface quinone groups were responsible for partial irreversible adsorption of the two compounds mentioned above and t o find a n expedient procedure suitable to eliminating this unwelcome effect, a series of experiments was performed. Results are reported in Table 11. As can be read, when any surface hydroquinone was oxidized t o quinone by pretreating the Carbopack cartridge with 200 m L of an aqueous solution of sodium hypochlorite, 4 mg/L metribuzin was completely lost and the recovery of chloridazon decreased to 15%. Under such conditions, severe losses of some of the phenylureas considered were also observed. No more anomalous effects of adsorption for the analytes considered were observed when hypochlorite-treated Carbopack particles were further washed with either a n aqueous solution of a well-known reducing agent, such as ascorbic acid, or with 50 m L of a methanolic solution of phenylthiourea, 20 mg/L. It was reported (26)that thioureas react rapidly with quinones t o form stable benzothiazoles. From these experiments, it can be reasonably concluded that quinone groups, if present on the Carbopack surface, can provoke effects of chemisorption for particular adsorbates probably via addition reactions. From the results obtained on extracting analytes from pure water by means of a previously hypochlorite-treated Carbopack cartridge, it appears that irreversible adsorption occurred also for those phenylureas that were reported to be more easily
582
ANALYTICAL CHEMISTRY, VOL. 63, NO. 6, MARCH 15, 1991
Table I. List of the Pesticides Considered in This Study and Their Liquid Chromatography Retention Times0 compd 1, oxamyl 2, methomyl 3, chloridazon 4, metoxuron 5, bromacil 6, monuron 7, cyanazine 8, metribuzin 9, carbofuran 10, atrazine 11, carbaryl 12, monolinuron 13, paraoxon 14, propachlor 15, propham 16, propanil 17, linuron 18, chloroxuron 19, chloropropham 20, fenitrothion 21, azinphos ethyl 22, parathion ethyl 23, coumaphos 24, phoxim 1, bentazon 2, bromoxynil 3, dinitro-o-cresol 4, 2,4-D 5, mecoprop fi, 2,4,5-T 7, 2,4-DB 8, MCPB 9, 2,4,5-TP 10, dinoseb 11, dinoterb
pesticide class
ret time, min
Base-Neutrals carbamate carbamate phenylpyridazinone phenylurea uracil phenylurea triazine triazine carbamate triazine carbamate phenylurea organophosphate acetanilide carbamate propioanilide phenylurea phenylurea carbamate phosphorothioate phosphorodithioate phosphorothioate phosphorothioate phosphorothioate
3.1 3.7 6.1 9.7 10.2 11.0 11.4 11.7 13.3 14.4 14.8 15.6 16.7 17.3 17.7 18.8 20.4 21.1 23.1 25.0 25.5 28.0 29.2 29.8
Acids benzothiadiazin phenol phenol phenoxyacid p henoxyacid phenoxyacid phenoxyacid phenoxyacid phenoxyacid phenol phenol
6.4 6.8 7.2 8.6 9.0 9.7 10.5 10.2 12.0 18.1 19.0
Abbreviations: 2,4-D = (2,4-dichlorophenoxy)aceticacid; 2,4,5-T = (2,4,5-trichlorophenoxy)aceticacid; 2,4-DB = (2,4-dich1orophenoxy)butyric acid; 2,4,5-TP = (2,4,5-trichlorophenoxy)Dropionic acid. Table 11. Recovery of Some Selected Pesticides Extracted from 2-L Aliquots of Drinking Water by Various Pretreated Carbopack Cartridges (Spiking Level: 0.2 pg/L of Each Pesticide) compd chloridazon metrihuzin metoxuron monuron chloroxuron
recovery for pretreatment none Ab BC Cd 62 28 91 94 90
15
n.d.e 30 42 21
97 91 96 98 94
98 93 96 99 96
Mean values were calculated from three determinations. *Washing with 200 mL of a NaClO solution, 4 mg/L. ‘The NaC10-pretreated Carbopack cartridge was further washed with 15 mL of an ascorbic acid solution, 10 g/L, acidified with HC1 (pH 2). The NaC10-pretreated Carbopack cartridge was further washed with 50 mL of a phenylthiourea methanolic solution, 20 mg/L. en.d. = not detected. (I
oxidable, such as metoxuron, monuron, and chloroxuron (27). This seems to indicate that another possible route of irreversible adsorption may be that of oxidative degradation caused by quinone groups contaminating the Carbopack surface. Under this hypothesis, however, this oxidation process is a selective one, as the recovery from the Carbopack cartridge of other classes of pesticides prone to oxidation, such as
Table 111. Recovery of Some Selected Acidic Pesticides from 1-L Distilled Water Samples Salted with Two Different Amounts of NaCl by the Proposed Method Compared with One Making Use of a n Anion-Exchanger Cartridge 70 recovery
compd bentazone bromoxynil dinitro-o-cresol 2,4-D
mecoprop 2,4-DB 2,4,5-TP
0.1% NaCl anion this exchanger method 72 63 72 87 76 68 73
98 98 95 93 90 99 95
1% NaCl anion this exchanger method 37 44 17 51 24 18 20
101 99 97 94 92 98 97
Mean values were calculated from two determinations. phosphorothioates and phosphorodithioates, was only slightly affected by the pretreatment of the sorbent with an oxidizing solution. From a practical point of view, before pesticides were extracted from a water sample, quinones present on the Carbopack surface were completely converted t o less reactive hydroquinones by washing the Carbopack cartridge with an aqueous solution of ascorbic acid, as reported in the Experimental Section. This one-step pretreatment was preferred t o that involving reaction of phenylthiourea with quinone groups, as the former treatment was more easily practicable than the latter one. Moreover, the precaution of avoiding a too prolonged passage of room air through the cartridge after the water sample was passed through the cartridge was taken. Finally, before the drinking water samples containing hypochlorite were extracted, this was decomposed by adding 0.4 g/L of sodium sulfite. Ideally, ion exchangers are very useful materials for selectively extracting ionogenic organic compounds from aqueous solutions. In practice, this class of sorbents has been rarely included in analytical procedures developed for analyzing acidic organic compounds in aqueous environmental samples, because their extraction efficiency is strictly dependent upon the ionic strength of the actual water samples. As reported elsewhere (17, 25), in addition t o hydroquinone/quinone groups, the graphitic framework of the Carbopack surface is also contaminated by a n oxygen complex having a chromene-like structure, which, in the presence of acidified water, is promptly rearranged to form benzpyrylium salts. These latter groups, bearing a positive charge, enable Carbopack to act as an anion exchanger as well as a nonspecific adsorbent. T h e ability of the Carbopack cartridge under study in extracting and isolating acidic organic pollutants from water samples with different ionic strengths was compared with that of a cartridge containing 200 mg of a high-capacity resin-based strong anion exchanger, such as Amberlite CG-400-11 (Fluka). This material was converted t o the hydroxide form prior to extraction. For these experiments, 1-L aliquots of distilled water were salted with two different amounts of sodium chloride and spiked with some of the acidic pesticides considered at the 5 pg/L level. Removal of the analytes from the conventional anion exchanger was performed first by washing it with 4 mL of acidified water (pH 1)and then desorbing the acidic compounds with 4 m L of water/methanol (50:50, by volume) acidified with HC1,O.l mol/L. A h r the water sample was passed through the Carbopack cartridge, the same procedure, as reported in the Experimental Section, was followed for selective elution of acidic compounds. Recovery data are reported in Table 111. It is noteworthy that no significant amount of any acidic pesticide extracted from the highly salted
ANALYTICAL CHEMISTRY, VOL. 63, NO. 6, MARCH 15, 1991
583
Table IV. Recovery of Some Selected Pesticides from the Carbopack B Cartridge by Passing through Various Eluant Systems % recovery0
compd
methomyl chloridazon
metoxuron bromacil cyanazine metribuzin atrazine propachlor
parathion ethyl azinphos ethyl
ether/hexane (18)
petroleum ether/toluene (19)
CH2C12/CH3CN(20)
(50:50, v / v )
(67:33, v / v )
(60:40, v/v)
CH2C12/MeOH (8020, V/V)
CHzClZ
92
