Environ. Sci. Technol. 1996, 30, 1220-1226 Soil Component Interactions with 2,4-Dichlorophenoxyacetic Acid under Supercritical Fluid Conditions E . A . R O C H E T T E , * ,† J . B . H A R S H , † A N D H. H. HILL, JR.‡ Department of Crop and Soil Sciences and Department of Chemistry, Washington State University, Pullman, Washington 99164 Interest in using supercritical CO2 as an alternative to nonpolar liquid solvents to extract toxic organics from soils is growing. Unfortunately, supercritical CO2 alone is a poor solvent for many polar compounds, including acid herbicides, in soils. In this study, supercritical CO2 was modified with benzoic acid/ methanol to extract 2,4-dichlorophenoxyacetic acid (2,4D) from selected model soil components, analogs of soil components that potentially limit its extraction from soils. The components included four minerals, silica gel, sodium humate, and humic acid. These model materials were chosen to test three potential factors inhibiting 2,4-D extraction: (1) adsorption to mineral surfaces, (2) diffusion-limited release from porous materials, and (3) pH-dependent partitioning between the solid and supercritical fluid phases. High recoveries were obtained from gibbsite (100 ( 3%), goethite (91 ( 3%), and illite (88 ( 6%). Porous materials such as the silica gels and humic acid yielded lower recoveries, 70 ( 4% to 87 ( 7% and 80 ( 3%, respectively. We extracted only 11 ( 2% of the spiked 2,4-D from sodium humate. An inverse relationship existed between the pH of the solidbenzoic acid/methanol suspension and 2,4-D recovery. Overall, soil pH was the main chemical factor affecting 2,4-D recovery. Due to its porosity, pH buffering capacity, and ubiquitous occurrence, we contend organic matter will generally be the main component limiting extraction of 2,4-D from soils. Furthermore, it appears methanol enhances recovery, in part, because the protonated form of 2,4-D is favored due to the higher pKa of 2,4-D in this solvent compared to water, since the ionized form will not dissolve in a nonpolar fluid unless an ion pair is formed. Introduction Supercritical fluid extraction (SFE) with CO2 above critical temperatures and pressures has provided a clean alternative to liquid/liquid extraction techniques, especially for nonpolar pesticides in soils (1, 2). It has both industrial and analytical applications, the former including the regeneration of adsorbents used in wastewater treatment (3, 4), and, potentially, treatment of diesel oil- and PCB-contaminated soil (5). Discussions of the theory, advantages, and some of the applications of modified and unmodified supercritical fluids for environmental analysis have been described by others (6-10). Extraction of polar compounds has proved difficult and will require a better understanding of both the chemical and physical processes influencing SFE from soils. The acid herbicide 2,4-D (2,4-dichlorophenoxyacetic acid) was selected for this investigation due to its heavy agronomic, industrial, and private usage and because it represents a large class of herbicides, phenoxy acids. Unfortunately, the recommended EPA techniques (Methods 8150 or 8151) for extracting chlorinated herbicides from soil are time-consuming and require large volumes of the toxic solvents diethyl ether or methylene chloride (11, 12). A need to eliminate laboratory use of these solvents provides incentives to understand the capabilities of modified supercritical carbon dioxide as an extractant. Supercritical fluid extraction of polar compounds, such as 2,4-D, from environmental samples is enhanced by polar modifiers added to the matrix or the supercritical fluid. For instance, extractions using supercritical CO2 alone recovered less than 5% of the applied 2,4-D from soil samples (13, 14), whereas the addition of derivatizing reagents and ion-pairing reagents (14-16) and/or inorganic salts in methanol (14) as modifiers improved the SFE of 2,4-D from soils. Co-extraction of native soil organic matter compounds by ion pairing reagents may interfere with 2,4-D extraction or analysis and may warrant the investigation of alternative modifiers. Testing of various extraction schemes has been largely by trial and error, because too little is known about the mechanisms governing polar solute sorption and extractability in supercritical fluids. The objective of this research was to determine which of several common soil constituents are likely to limit 2,4-D recovery by SFE. Model soil components included representatives of common soil minerals, porous materials, and organic matter to test surface reactions, diffusion in porous materials, and pH buffering as potentially limiting factors in 2,4-D extraction. Materials and Methods We considered and tested several extraction systems for this study. For a spiked Palouse silt loam, soil recoveries of 2,4-D decreased in the following order: BF3/methanol methylation during SFE (66 ( 6%) ) a modified EPA Method 8150 (64 ( 5%) > 0.3 M CaCl2‚2H2O/methanol-modified * Present address: Department of Plant, Soil, and Entomological Sciences, University of Idaho, Moscow, ID 83844-2339; telephone 208-885-2947; fax: 208-885-7760; e-mail addrss: brochette@ marvin.csrv.uidaho.edu. † Department of Crop and Soil Sciences. ‡ Department of Chemistry. 1220 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 30, NO. 4, 1996 0013-936X/96/0930-1220$12.00/0 1996 American Chemical Society SFE (55 ( 2%) > 0.6 M benzoic acid/methanol-modified SFE (51 ( 1%) g 0.6 M acetic acid/methanol/acetonitrilemodified SFE (42 ( 3%) > methanol-modified SFE (26 ( 3) ) acetone-modified SFE (25 ( 5%) > hexamethyldisilazane/trimethylchlorosilane silylation during SFE (12 ( 3%) (17). Because BF3/methanol and CaCl2‚2H2O/ methanol were harmful to PEEK (polyether ether ketone) and/or stainless steel (14, 17), we selected 0.6 M benzoic acid/methanol as a modifier for this study. Chromatographic interferences were not a problem. For this study, it was necessary to use spiked samples since the materials of interest were not available from the field in a 2,4-Dcontaminated form; spiked soils do not necessarily provide an accurate representation of yields obtained from soils that have been contaminated in the field by environmental processes (18). Minerals. Gibbsite [Al(OH)3], a single-layer type clay common in highly weathered soils, was obtained from Alcoa Chemicals and used without modification. It easily passed through a 250-µm sieve. The supernatant from the deionized water/gibbsite slurry had a pH of 9.4 after 4 h of equilibration. Goethite (FeOOH), an abundant soil iron oxide mineral, and calcite (CaCO3), a common carbonate mineral, were both obtained from Minerals Unlimited (Ridgecrest, CA). The goethite was from the Cary Mine of Gogebic County, MI, and was ground to approximately 50 µm in a stainless steel shatterbox. The calcite was a very fine chalk from Bedfordshire, England, and was ground in a porcelain mortar and pestle to pass through a 250-µm sieve. Illite (K,Al,Si)4(Mg3Fe2Al2)O10(OH)2, a 2:1 clay abundant in soils of eastern Washington, was from a mine in Pennsylvanian shale from the Goose Lake Area of Illinois (19) and easily passed through a 250-µm sieve. Illite was washed with organic solvents to remove impurities that interfered with GC/ECD analysis of the supercritical fluid extracts of the unwashed illite. The supernatant from a deionized water slurry of cleaned illite had a pH of 3.9 after 4 h of equilibration. Silica Gels. Three silica gels donated by Waters Chromatography Division of Millipore (Milford, MA) had particle diameters of 15-20, 37-55, and 55-105 µm. All were porous normal-phase packing materials. Waters provided the following data with the silicas: surface areas (base silica) of 320 m2/g, total carbon less than 0.1%, pore volume of 1.0 mL g-1, and median pore diameter of 125 Å for all three silica size ranges. The silica gels served as model porous, amorphous soil inorganic components. Humic Materials. Sodium humate obtained from Aldrich Chemical Company (Milwaukee, WI) and characterized by Malcom and McCarthy (20) was used without modification. The pH of the sodium humate solution in deionized water (0.29 g of sodium humate/mL of deionized water) was 8.9. Humic acid was prepared by titrating 60 g of sodium humate dissolved in 1 L of deionized water with concentrated HCl to a pH of 0.9. The humic acid slurry was transferred to Spectrapor standard cellulose dialysis tubing (Fisher Scientific, Pittsburgh, PA), with a molecular mass cutoff of 12 000-14 000 Da and dialyzed against deionized water in a 6-L glass vessel until the electrolytic conductivity of the dialyzate dropped to 0.0003 S/m. After being dried, the humic acid was ground to pass through a 1-mm sieve. Chemicals. All organic solvents and NaCl used in this study were Baker analyzed reagent grade and were used as received. Absolute methanol was used as an SFE collection solvent and modifier. For pH measurements in methanol, HPLC-grade methanol was used. Boron trifluoride/ methanol (14%) was obtained from Alltech Associates (Deerfield, IL). Benzoic acid (99% purity) was from Aldrich Chemical Company (Milwaukee, WI). The 2,4-D acid (2,4dichlorophenoxyacetic acid; 98.6% purity), 2,4-D methyl ester (97% purity), and lindane (1,2,3,4,5,6-hexachlorocyclohexane; 99% purity) were obtained from Sigma Chemical Company (St. Louis, MO). The 2,4-D acid and 2,4-D methyl ester were stored in a refrigerated desiccator with anhydrous CaSO4 prior to use. Instrument-grade carbon dioxide was obtained from Alphagaz (San Francisco, CA). Spiking Procedure. Minerals, silica gels, and humic materials were spiked to 10.8 ppm (and 1.08 ppm for goethite) by preparing saturation pastes with deionized water solutions as described by Rhoades (21) and adding 2,4-D in acetone to the paste. The ratio of acetone to solution was 0.03 by volume. The materials were allowed to dry, with the drying time controlled to 24 h either by periodic addition of deionized water to fast-drying minerals, or by mixing the pastes (illite and humic materials) with a spatula to facilitate drying. Because the humic materials and the 9:1 mixture of gibbsite plus humic acid formed aggregates during the drying process, they were gently crushed to pass through a 1-mm sieve. The material was then homogenized, split, and placed in a freezer until use. The high concentration of 2,4-D applied to the model soil components was to account for the typically low mass fraction of a given component in soil (generally less than 0.1). In addition to spiking the silica gels with aqueous solutions of 2,4-D, subsamples of all the silica gels were analogously spiked with acetone solutions of 2,4-D. The treatment was identical except that water was excluded by drying the silicas at 105-107 °C overnight in an oven and cooling in a desiccator prior to spiking; also, the silicas were maintained in a desiccator during the spiking, drying, and splitting processes. Gravimetric analysis after spiking indicated that less than 3% acetone (by mass) remained on the silica after spiking. Supercritical Fluid Extractions. Extractions of the minerals, silica gels, and humic materials were performed with an Isco SFX-210 supercritical fluid extractor with an Isco 260D syringe pump. Lindane, used as our internal standard, was injected on the exiting filter frit of the SFE cell, and 5.0 ( 0.02 g (goethite and gibbsite), 2.50 g (illite, calcite and silica gels), or 1.00 g (humic acid and sodium humate) was placed in the 10-mL stainless steel extraction cell. The reduced mass of material for illite, calcite, and humic material prevented problems of reduced flow rate and restrictor clogging encountered when larger masses of these materials were used. For silica gels, it was not possible to fit 5 g into the cell due to its low bulk density. In all cases, the ratio of modifier to solid was 0.3 mL:1 g per modifier addition. This volume of modifier was selected since many soils are saturated with solution at 30% moisture content. After the modifier was injected onto the sample, the extraction chamber was brought to 80 °C and pressurized to 350 atm. Replicate samples were extracted in series. Static extractions were performed for 15 min; dynamic extractions followed with CO2 exiting through 30 cm of 50 µm i.d. fused silica tubing that emptied into vented, capped test tubes. One milliliter of methanol was placed in the collection tubes. Dynamic extractions lasted approximately 15 min each and were performed by passing VOL. 30, NO. 4, 1996 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 1221 20 mL of pressurized CO2 through the sample at a flow rate of 1-2 mL of pressurized CO2/min. The flow restrictor was inserted through a section of aluminum pipe, which was heated to approximately 100 °C. The extractor was allowed to vent (the remaining CO2 depressurized) for 20 min into the collection tubes after each extraction and then was completely vented through the instrument vent valves. Additional modifier (0.3 mL/g solid) was added to the cell, the restrictor was placed into a new tube containing 1 mL of methanol, and the extraction continued. Final extract volumes were adjusted to 2 mL with methanol. Esterification. All extracts were esterified with BF3/ methanol, based on the protocols of Pierce (Rockford, IL), Alltech Associates (Deerfield, IL), and Erner and Coggins (22): (a) 1-mL aliquots of the extracts were placed in 5-mL derivatization vials; (b) 1 mL of BF3/methanol was added to the extracts; the vials were capped, heated on a hot plate to 60 °C for 15 min, cooled for at least 15 min, and 1 mL of aqueous 1 M NaCl was added to each vial and shaken; (c) the samples were partitioned into a total of 1 mL of benzene using three portions of benzene; (d) the benzene layers were removed, combined, and centrifuged for approximately 5 min; and (e) the final benzene solutions were placed in vials and sealed. Standards containing lindane were prepared from 2,4-D methyl ester stocks and were treated in the same way as extracts except that 1 mL of methanol (instead of BF3/methanol) was added in step b and the standards were not heated. Also, standards were prepared from 2,4-D stocks and went through all steps used for the extracts. These standards always agreed with each other within 10%, and the latter were used for recovery calculations. Gas Chromatography. Analyses were performed with a Tracor 560 gas chromatograph, which was equipped with an 18-m DB-5 capillary column, a split-mode injection system, and an electron capture detector (ECD). Analyses of methyl esters were performed at a column temperature of 180 °C (isothermal), injector at 250 °C, and ECD at 325 °C. Surface Area Measurements. Surface areas of all of the model components were measured using the ethylene glycol monoethyl ether (EGME) method of Carter et al. (23). Samples were dried for 24 ( 2 h in an oven at 110 °C instead of P2O5 drying prior to analysis. Sample mass measurements, followed by desiccator evacuations (45 min each), were made until weight losses of EGME did not exceed 0.0002 g. A standard of freeze-dried Ca2(-SWy-1 (Wyoming bentonite; Clay Minerals Society) was analyzed with each group of four samples. The surface areas of all three silica gels, goethite, and humic acid were also determined by BET-N2 on an Autosorb-6 instrument (Quantachrome Corporation, Syosset, NY). The BET-N2 surface area values were from multipoint BET measurements. pH Measurements in Methanol. The pH of slurries of minerals, silica gels, and humic materials with methanol and benzoic acid/methanol were measured with a Radiometer GK 2321 C combination glass electrode attached to a Beckman φ45 pH meter. Probe filling solutions were aqueous saturated KCl as recommended by Radiometer for aqueous solutions and 0.08 M LiCl in methanol. Lithium chloride in methanol was recommended for titrations in methanol (24). The two filling solutions gave essentially the same pH results. Modifiers of 0.6 M benzoic acid/methanol and HPLCgrade methanol were applied to the samples in a ratio of 1222 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 30, NO. 4, 1996 FIGURE 1. SFE recoveries of 2,4-D from materials spiked with aqueous solutions of 2,4-D. The modifier was 0.6 M benzoic acid/methanol. Error bars represent absolute standard deviations for triplicate extractions. The concentration of 2,4-D (µg g-1) is indicated for goethite and was 10 µg g-1 for all other materials. The silica data are from the 37-55-µm particle diameter silica treated with 2,4-D in acetone. 0.3 mL/g of sample as was done for SFE modification. The modifiers were left in contact with the sample for 15 min and diluted with 2 mL of methanol/g of sample (2:1 dilutions). The slurries were thoroughly mixed and centrifuged, and the supernatant solution was decanted for measurement. For silica gels, it was necessary to make 3:1 dilutions due to their large surface areas; however, it was possible to obtain sufficient sample from a 2:1 dilution of the 55-105-µm silica, which allowed a comparison with the 3:1 dilution. There was no significant difference at the 0.05 significance level between these dilutions. The pH was calculated by using a sodium salicylate (0.01 M)/salicylic acid (0.01 M)/methanol buffer and the following equation: pH sample ) Ebuffer (mV) - Etest solution (mV) (0.198) (temperature (K)) + buffer pH where 0.198 is a constant in units of mV K-1 (25) and was obtained by combining constants from the Bates-Guggenheim convention (26). The buffer pH was assumed equal to 7.53 as given by Dean (26). A second buffer of ammonium hydrogen oxalate (0.01 M)/oxalic acid (0.01 M)/methanol having a theoretical pH value of 5.79 (26) and a 0.0008 M HCl/methanol solution were utilized to verify the linearity of the probe response. Results Our experiments were conducted to determine the influence individual soil components may have on 2,4-D partitioning into a supercritical fluid. Soil component effects may result from interactions of 2,4-D with specific surface functional groups or from the effects of pH, surface area, and porosity, which vary from one soil component to another. Minerals. Total recoveries from experiments with 0.6 M benzoic acid/methanol-modified SFE of 2,4-D from the minerals are given in Figure 1 and Table 1. The recovery from gibbsite (100 ( 3%) is consistent with the finding that 2,4-D is weakly adsorbed by alumina (27). Recovery of 2,4-D from goethite was similar whether spiked at 10 µg g-1 (91 ( 3%) or at 1 µg g-1 (83 ( 3%). Further additions of TABLE 1 Physical Characteristics and Total 2,4-D Recoveries from Model Soil Components component goethite gibbsite calcite humic acid sodium humate illite 15-20 µm silica water acetone 37-55 µm silica water acetone 55-105 µm silica water acetone EGME surface (m2 g-1) total surface in extraction cella (m2) N2-BET surface (m2 g-1) total surface in extraction cella (m2) 5(1 5(2 23 ( 1 51 ( 1 54 ( 6 136 ( 7 241 ( 9 25 ( 5 25 ( 10 57.5 ( 2.5 51 ( 1 54 ( 6 340 ( 18 603 ( 23 0.88 4.4 0.034 0.034 271 677 376 ( 17 940 ( 43 253 633 363 ( 9 907 ( 23 305 763 total recovery ( absolute SD (%) 91 ( 3 100 ( 3 69 ( 4 80 ( 3 11 ( 2 88 ( 6 87 ( 7 70 ( 4 83 ( 6 87 ( 6 72 ( 3 81 ( 1 a Calculated using the mass of the component in the extraction cell (g) × the respective surface area (m2 g-1). The modifier volume was always 0.3 mL/g of component in the extraction cell. TABLE 2 Silica Recovery Ratios, Micropore Characteristics, and Water Contents after Spiking silica 37-55 µm water acetone 15-20 µm water acetone 55-105 µm water acetone microporeb recovery surface area micropore vol water ratioa (mL g-1) content (%) (m2 g-1) 8.5 0.0035 0.68 0.62 14.0 14.4 0.0065 0.88 0.70 31.5 19.5 0.96 0.80 0.0091 41.0 a Recovery ratio ) (recovery from 1st modifier addition/total recovery). b Combined surface area of pores less than 20 Å wide. modifier may have increased 2,4-D recovery from illite and calcite given the relatively large amounts of 2,4-D in the second extractions of these minerals. In general, higher recoveries of 2,4-D were obtained from gibbsite, goethite, illite, and calcite than from soils extracted in previous studies. Silica Gels. Total recoveries of 0.6 M benzoic acid/ methanol-modified SFE of 2,4-D from spiked silica gel are given in Table 1. Prior to spiking, the silicas all contained between 2.5 and 2.9% moisture; however, after treating the samples with aqueous spike solutions and air drying, all silicas had moisture contents greater than 10% (Table 2). The recovery ratio (the 2,4-D recovery from the first addition of modifier divided by the total 2,4-D recovery from the first and second additions of modifier) correlated with the silica moisture contents for the aqueous 2,4-D-spiked silicas [recovery ratio ) 0.011 (moisture %) + 0.537, r2 ) 0.995]. The recovery ratio is assumed to be positively related to the 2,4-D extraction rate, though each 30-min static/dynamic extraction coincides with the addition of more modifier. Since essentially no 2,4-D could be recovered from soils without modifier over equivalent periods of time in a previous study (14), the extraction rate is not strictly timedependent in these experiments and is probably more dependent on solid/fluid (especially modifier) partition coefficients. Additionally, a strong correlation between micropore surface area and volume and the moisture content of watertreated silicas was seen. Because water content introduced a complicating chemical factor, it was not possible to determine if microporosity was directly or indirectly related to the extraction rate. Extractions of 2,4-D from silicas spiked with acetone solutions of the herbicide were made to avoid the possible complications that may arise from the variable amounts of adsorbed water present in the water-treated silicas after air drying. Overall, the mean total 2,4-D recoveries from acetone-treated silicas (79%) is essentially the same as that from the water-treated silicas (81%) after two additions of modifier; however, the rate of 2,4-D recovery from the water-treated silicas was greater than from the acetone-treated silicas. Humic Materials. The results of 0.6 M benzoic acid/ methanol modified SFE of 2,4-D from humic materials are given in Figure 1 and Table 1. The quantity of 2,4-D extracted from sodium humate (11 ( 2%) was much lower than that from humic acid (80 ( 3%). Mixing humic acid with gibbsite (prior to spiking) reduced the amount of 2,4-D recovered in the first modifier addition compared to pure gibbsite. A third modifier addition to the mixture brought the total 2,4-D recovery to 97 ( 5%. Surface Areas. Results from both EGME and N2-BET surface areas are presented in Table 1 for all of the components measured. The surface area of the Ca2+saturated SWy-1 EGME standard, 691 ( 39 m2 g-1,was similar to the value reported by Madsen (28). In general, 2,4-D recoveries were poorly correlated (r2 ) 0.007) with EGME surface areas. Correlations improved to r2 ) 0.724 when calcite, sodium humate, and humic acid were removed from the data set. Poor correlations (r2 ) 0.192) between surface area and recovery were also observed with N2-BET data, which were only available for five samples. The EGME surface areas for humic acid were higher than N2-BET surface areas. Other differences between N2 and EGME surface areas were not large enough to affect overall relationships between recovery and surface areas. The surface areas of humic acid and sodium humate were low compared to literature values for soil organic matter surface areas measured with polar liquids. Bower and Gschwend (29) reported ethylene glycol surface areas VOL. 30, NO. 4, 1996 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 1223 TABLE 3 pH Values of 2:1 Mineral, Silica, and Humic Material Slurries in Modifiers, with Aqueous KCl Probe Filling Solutiona sample sodium humate silica (55-105 µm) acetone-2,4-D spike solutions aqueous-2,4-D spike solutions calcite gibbsite illite goethite humic acid benzoic acid/ methanol pH ( SD methanol pH ( SD 9.37 ( 0.01 12.01 ( 0.10 7.89 7.53 7.42 ( 0.01 (7.23 ( 0.01) 6.99 ( 0.02 6.59 ( 0.08 6.22 ( 0.11 4.74 ( 0.05 10.93 ( 0.08b 10.81 ( 0.01b 11.03 ( 0.04 11.34 ( 0.00 10.93 ( 0.08 9.62 ( 0.23 4.77 ( 0.01 a Values measured with LiCl/methanol probe filling solution are given in parentheses when they differed statistically. b Prepared as 3:1 methanol slurries. for soil organic matter of 558-803 m2 g-1, based on soil samples before and after treatment with H2O2. The humic acid and sodium humate in this study, like the other materials analyzed by the EGME technique, were ovendried at 110 °C for 24 h to remove water before the EGME treatment. This may have resulted in some collapse of the humic materials, as has been observed for air-dried humic and fulvic acids (30), and in the resulting loss of specific surface area (29). pH Experiments. The effects of SFE modifiers on pH were examined by measuring the pH of the model soil components in the presence of methanol and 0.6 M benzoic acid/methanol (Table 3). No significant differences existed between H+ activities of the three particle sizes of silica (0.05 significance level). On the other hand, the slurry of moist 55-105-µm silica in benzoic acid/methanol was 0.4 pH unit lower than that for the dry silica, so the presence of water in the silica appears to slightly lower the pH in silica/benzoic acid/methanol slurries. Discussion Physical Factors Affecting Extraction. Discerning chemically controlled processes from physically controlled processes was done in previous studies (14, 17) by varying chemical modifiers while the soil type was kept constant. Additional chemical information was obtained in this study by changing the solid phase from which 2,4-D was extracted while maintaining a constant chemical modifier to the SFE system. The latter type of experiment, involving surfaces representative of soils, inherently also changed the physical features of the system. The surface area data (Table 1) were collected to help distinguish this potential source of variability in extraction results from purely chemical effects. Recoveries of 2,4-D cannot be explained by surface area variations. The poor correlation coefficient (0.007) was due mainly to humic acid, sodium humate, and calcite, implying that chemical factors are important determinants of extractability. General relationships aside, the relatively lower recovery of 2,4-D from illite compared to gibbsite may be due to the higher surface area of illite. When silica gels were considered exclusively, relationships between 2,4-D recovery and surface area were expected to be free of chemical variability, whereas physical characteristics such as surface area, particle size, porosity, 1224 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 30, NO. 4, 1996 and probably pore tortuosity varied. Several important physical factors relate to diffusion path geometries: tortuosity, the presence of “dead-end” pores, and variable pore diameters (31). Janda et al. (9) point out that the rate of supercritical fluid extraction depends on the radius of the matrix particles. In this study, few if any relationships between 2,4-D recovery and the measured physical characteristics of the silicas could be found. For the most part, the differences between the physical characteristics of the silicas and the differences in 2,4-D recoveries were relatively small. The silicas may actually be very similar in their micropore geometries and, if so, 2,4-D extractability may be independent of the variable macropore characteristics. Nonetheless, there are some differences between the recoveries from the individual silicas, suggesting that there may be physical effects on 2,4-D extraction that are not yet understood. Indirectly, physical factors may have affected the chemistry of the extractions of the silicas spiked with aqueous solutions of 2,4-D. These silicas contained variable amounts of water, which increased as the micropore surface areas and volumes increased (Table 2). The retained water may have acted as a supercritical fluid modifier (to be discussed later). Chemical Factors Affecting Extraction. Previous studies with soils in which the 2,4-D recoveries changed with modifier chemistry (14, 17) provide evidence that chemical interactions among 2,4-D, the supercritical solvent, and soil are important in controlling extractability by SFE. For instance, boron trifluoride/methanol enhanced 2,4-D extraction by derivatizing the compound to an ester, which is a less polar compound than the acid form of 2,4-D and is expected to have a higher affinity for nonpolar solvents. The role of supercritical fluid modifiers (aside from derivatizing and ion pairing agents) in the extraction of polar and nonpolar compounds from solids has been attributed to modification of the polarity of the supercritical fluid (32), solubility enhancement (of dioxin) by methanol-modified supercritical CO2 (33), H-bonding-enhanced solubility (34), effects on pore diffusional resistances and desorption from soils (35), and improved wetting of soil by polar methanol and dimethyl sulfoxide (36). There have been, however, no suggestions that methanol enhances extraction efficiencies by affecting the pKa values of ionizable solutes. It has been shown that the pKa of 2,4-D is roughly 5 pH units higher in methanol than in water (pKa values of 7.6 and 2.64, respectively), favoring the protonated form in methanol solvent (37). The methanol modifier should, therefore, enhance extraction by increasing the abundance of the acid form of 2,4-D in the soil. The nonionic acid form of 2,4-D is likely to be more extractable in solvents of low polarity, such as supercritical CO2, than the ionic form. Thus, CaCl2‚2H2O/methanol modifier used in previous studies (14, 17) and benzoic acid/methanol modifier may have improved 2,4-D recovery compared to methanol alone by decreasing the pH of the soil solution prior to and during SFE. The chemical nature of the soil components may also influence 2,4-D extractability. Solution pH appears to be an important factor in the case of minerals as well as soils. Mineral surfaces, with the exception of calcite, do not appear to limit 2,4-D extraction, as the highest 2,4-D recoveries from the entire study were obtained from gibbsite, goethite, and illite. Slurries of these minerals also had relatively low benzoic acid/methanol pH values; only humic acid had a lower value than illite. Slurries of calcite in benzoic acid/ methanol maintained slightly higher pH values than the other minerals, and 2,4-D yields from calcite were the lowest. On the other hand, a small effect of 2,4-D adsorption to oxide surfaces is possible. Recovery from gibbsite is rapid and complete, whereas goethite yielded less 2,4-D and required two extractions to reach 90% recovery. Consistent with this, 2,4-D is reversibly adsorbed to goethite (38) but negatively adsorbed to aluminum oxide (27). It is tempting to conclude that the 10% of 2,4-D not recovered from goethite was adsorbed by specific surface sites; however, if sites capable of irreversibly sorbing 2,4-D existed, the 1 µg g-1 spike should have been unrecoverable. It is neither possible to invoke nor rule out surface effects on 2,4-D recovery with these data. Moisture also affects 2,4-D extractability. The recovery ratio (Table 2) correlated strongly with the moisture retained on the air-dried silicas. This suggests that water acted as an additional modifier, increasing 2,4-D affinity for the methanol/CO2, particularly in the first extraction where the water content was higher. This is supported by extractions from silicas spiked from acetone solutions, where the recovery ratios were comparable to that of the low-moisture, aqueous 2,4-D-treated silica (37-55 µm). Water also appears to slightly lower the pH of the silica/ benzoic acid/methanol slurries, as described in the Results section for the moist 55-105-µm silica. Snyder et al. (39) found that when water was added to soil before SFE, the extraction of polar pesticides by supercritical CO2 improved. The tendency for water to enhance the recovery of polar compounds although limit the extraction of nonpolar compounds was discussed by Camel et al. (8). After two modifier additions, however, the mean 2,4-D recovery from the silicas treated with aqueous and acetone spiking solutions were equivalent (81% and 79%, respectively). Organic Matter. Humic materials in soils are among the most important buffers of soil acidity and are highly porous. Therefore, organic matter likely affects 2,4-D extractability both via physical and chemical processes. Though the Aldrich humic acid used in this study is not an exact analog of soil organic matter (20), it does have many of the above features in common with soil organic matter and allows comparison of 2,4-D extraction from a wellcharacterized organic material with that of several minerals and silica. The highest and lowest pH values in benzoic acid/ methanol were those of sodium humate and humic acid, respectively, and 2,4-D extractability differed dramatically for these two components (Tables 1 and 3). The distribution of 2,4-D between acid and base forms is difficult to quantify because the pKa’s of benzoic acid and 2,4-D and the solution pH depend on the soil moisture, methanol concentration, and CO2 pressure. Any attempt to quantitatively model this complex system without knowledge of the appropriate pKa values would be fraught with uncertainty. Nonetheless, the pHs of humic acid (4.74 ( 0.05) and sodium humate (9.37 ( 0.01) in benzoic acid/methanol slurries provide relative measures of the pH of these materials before supercritical CO2 was added to them. The pKa of 2,4-D was shown to be 7.6 in methanol (37), which is roughly 1.8 units below the pH of the sodium humate/benzoic acid/methanol slurries and almost 3 units above the pH of humic acid/benzoic acid/methanol slurries. The presence of mineral or organic surfaces may alter the reactivity of H+ or 2,4-D such that protonation exceeds that predicted from the pH of the bulk solution and the FIGURE 2. Recoveries of 2,4-D from benzoic acid/methanol-modified SFE (first modifier addition) plotted against the pH of 2:1 slurries of minerals, silica gel, and humic materials in benzoic acid/methanol. Error bars represent absolute standard deviations for triplicate extractions. The curve represents the “expected recovery” of 2,4-D based on the acid/conjugate base ratios at various pHs. The pKa of 2,4-D in methanol used for this calculation was 7.6 (37). pKa. Harter and Ahlrichs (40) showed that significant protonation and sorption of 2,4-D on Na+- and H+-saturated montmorillonite occurred at pHs 2 or more units above its pKa. This may explain recovery of as much as 11% of the 2,4-D even when the suspension pH exceeded the pKa by almost 1.8 units in the case of sodium humate. Intraorganic matter diffusion, known to be the dominant process limiting the sorption of hydrophobic organic compounds onto soils (41), could also limit the extraction of 2,4-D from humic materials. Khan (42) found rapid adsorption of 2,4-D by humic acid (in water) for the first 1-1.5 h of treatment. Continued adsorption thereafter was linearly related to the square root of treatment time and was attributed to intraparticle transport. Molecular diffusion slows sorption of organic compounds in soils, with sorption to pore walls of small pores and organic gels responsible for retarding diffusion (42). Slow diffusion through the humic matrix may explain the reduced rate of extraction from gibbsite with humic acid relative to gibbsite alone (Figure 1). If the solution pH of the soil components in the presence of the modifier is the only factor controlling the recovery of 2,4-D, then an expected 2,4-D recovery can be calculated using the pH in the modifier and the pKa of 2,4-D in methanol. An “expected recovery” curve has been plotted on Figure 2. While many components plot near the line, humic acid yielded a relatively low recovery given its pH. Slower diffusion through the 3-D organic matrix than through inorganic components, and/or a relatively lower partition coefficient for 2,4-D between humic acid and the fluid, compared to that for inorganic components, may cause this discrepancy. Almost 30% more 2,4-D was obtained when a second extraction of the humic acid was made (Figure 1). Sodium humate has such a high pH and such a low 2,4-D recovery compared to inorganic materials that diffusional limitations are difficult to discern from the effects of pH. The effects of soil organic matter both on the proportion of 2,4-D in the extractable form (protonated in this case), on 2,4-D diffusion rates, and on solid-fluid partitioning are likely to be the most important factors limiting the rate and extent of extraction of 2,4-D from soils by modified SFE. VOL. 30, NO. 4, 1996 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 1225 Acknowledgments The activities on which this paper was based were financed in part by the Department of the Interior, U.S. Geological Survey, through the State of Washington Water Research Center. The contents of this paper do not necessarily reflect the views and policies of the Department of the Interior, nor does mention of trade names or commercial products constitute their endorsement by the United States Government. The authors thank Doug Wittmer of Waters Chromatography Division of Millipore for donating the silica gels used in this study. The assistance of Keith D. Miller and Gerard P. Irzyk with the lyophilizer is gratefully acknowledged. We thank William C. Koskinen for helpful suggestions and Clarke Maxwell for performing N2-BET surface area measurements. Literature Cited (1) Lopez-Avila, V.; Dodhiwala, N. S.; Beckert, W. F. Method for the Supercritical Fluid Extraction of Soils/Sediments; Environmental Monitoring Systems Laboratory, Office of Research and Development, U.S. EPA: Las Vegas, NV, 1990; EPA/600/4-90/026. (2) Snyder, J. L.; Grob, R. L.; McNally, M. E.; Oostdyk, T. S. Anal. Chem. 1992, 64 (17), 1940-1946. (3) DeFilippi, R. P.; Krykonis, V. J.; Robey, R. J.; Modell, M. Supercritical fluid regeneration of activated carbon for adsorption of pesticides; U.S. EPA Office of Research and Development: Washington, DC, 1980; EPA/600/2-80-054. (4) Picht, R. D.; Dillman, T. R.; Burke, D. J.; DeFilippi, R. P. AIChE Symp. Ser. 1982, 78 (219), 136-149. (5) Jain, V. K. Environ. Sci. Technol. 1993, 27, 806-808. (6) Hawthorne, S. B. Anal. Chem. 1990, 62 (11), 633A-642A. (7) Hawthorne, S. B.; Miller, D. J.; Burford, M. D.; Langenfeld, J. J.; Eckert-Tilotta, S.; Louie, P. K. J. Chromatogr. 1993, 642, 301317. (8) Camel, V.; Tambute, A.; Caude, M. J. Chromatogr. 1993, 642, 263-281. (9) Janda, V.; Bartle; K. D.; Clifford, A. A. J. Chromatogr. 1993, 642, 283-299. (10) Chester, T. L.; Pinkston, J. D.; Raynie, D. E. Anal. Chem. 1994, 66 (12), 106R-130R. (11) Test Methods for Evaluating Solid Wastes; U.S. Environmental Protection Agency, SW-846, Office of Solid Waste and Emergency Response; U.S. Government Printing Office: Washington, DC, 1986. (12) Test Methods for Evaluating Solid Wastes; U.S. Environmental Protection Agency, SW-846, Proposed Update II; Office of Solid Waste and Emergency Response, U.S. Government Printing Office:Washington, DC, 1992. (13) Burk, R. C.; Kruus, P. J. Environ. Sci. Health 1990, B25 (5), 553567. (14) Rochette, E. A.; Harsh, J. B.; Hill, H. H., Jr. Talanta 1993, 40 (2), 147-155. (15) Hawthorne, S. B.; Miller, D. J.; Nivens, D. E.; White, D. C. Anal. Chem. 1992, 64 (4), 405-412. (16) Lopez-Avila, V.; Dodhiwala, N. S.; Beckert, W. F. J. Agric. Food Chem. 1993, 41, 2038-2044. 1226 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 30, NO. 4, 1996 (17) Rochette, E. A. Ph.D. Dissertation, Department of Crop and Soil Sciences, Washington State University, 1994. (18) Buford, M. B.; Hawthorne, S. B.; Miller, D. J. Anal. Chem. 1993, 65, 1497-1505. (19) Grim, R. E.; Bradley, W. F.; J. Am. Ceramic Soc. 1939, 22, 157164. (20) Malcolm, R. L.; MacCarthy, P. Environ. Sci. Technol. 1986, 20 (9), 904-911. (21) Rhoades, J. D. In Methods of Soil Analysis, 2nd ed.; Agronomy Monograph 9; Page, A. L., Miller, R. H., Keeney, D. R., Eds.; American Society of Agronomy, Inc., and Soil Science Society of America, Inc.: Madison, WI, 1982; Part 2, Chapter 10. (22) Erner, Y.; Coggins, C. W., Jr. J. Am. Soc. Hort. Sci. 1989, 114 (5), 846-850. (23) Carter, D. L.; Mortland, M. M.; Kemper, W. D. In Methods of Soil Analysis, 2nd ed.; Agronomy Monograph 9; Klute, A., Ed.; American Society of Agronomy, Inc., and Soil Science Society of America, Inc.: Madison, WI, 1986; Part 1, Chapter 16. (24) Linnet, N. Non-aqueous titrations; Radiometer A/S ST24: Copenhagen, 1968, 19 pp. (25) pH Theory and Practice; Radiometer 918-518: Copenhagen. (26) Dean, J. A. Lange’s Handbook of Chemistry, 13th ed.; McGrawHill Book Co.: New York, 1985. (27) Haque, R.; Sexton, R. J. Colloid Interface Sci. 1968, 27 (4), 8181827. (28) Madsen, F. T. In Data Handbook for Clay Minerals and other Non-Metallic Minerals; Van Olphen, H., FriPiat, J. J., Eds.; Pergamon Press: Oxford, 1979. (29) Bower, C. A.; Gschwend, F. B. Soil Sci. Soc. Am. Proc. 1952, 16, 342-345. (30) Tan, K. H. Soil Sci. Soc. Am. J. 1985, 49, 1185-1191. (31) Ball, W. P.; Roberts, P. V. Environ. Sci. Technol. 1991, 25 (7), 1237-1249. (32) Janda, V.; Steenbeke, G.; Sandra, P. J. Chromatogr. 1989, 479, 200-205. (33) Onuska, F. I.; Terry, K. A. J. High Resolut. Chromatogr. 1989, 21, 357-361. (34) Walsh, J. M.; Donahue, M. D. Fluid Phase Equilib. 1989, 52, 397-404. (35) Dooley, K. M.; Kao, C.; Gambrell, R. P.; Knopf, F. C. Ind. Eng. Chem. Res. 1987, 26, 2058-2062. (36) Liu, M. H.; Kapila, S.; Yanders, A. F.; Clevenger, T. E.; Elseewi, A. A. Chemosphere 1991, 23 (8-10), 1085-1095. (37) Lee, L. S.; Bellin, C. A.; Rodolfo, P.; Rao, P. S. C. Environ. Sci. Technol. 1993, 27, 165-171. (38) Watson, J. R.; Posner, A. M.; Quirk, J. P. J. Soil Sci. 1973, 24, 503-510. (39) Snyder, J. L.; Grob, R. L.; McNally, M. E.; Oostdyk, T. S. J. Chromatogr. Sci. 1993, 31, 183-191. (40) Harter, R. D.; Ahlrichs, J. L. Soil Sci. Soc. Am. Proc. 1969, 33, 859-863. (41) Brusseau, M. L.; Jessup, R. E.; Rao, P. C. S. Environ. Sci. Technol. 1991, 25 (1), 134-142. (42) Khan, S. U. Can. J. Soil Sci. 1973, 53, 429-434. Received for review June 20, 1995. Revised manuscript received November 9, 1995. Accepted November 14, 1995.X ES950432N X Abstract published in Advance ACS Abstracts, February 1, 1996.
© Copyright 2026 Paperzz