Environ. Sci. Technol. 2004, 38, 5413-5419
Hydrophilic and Hydrophobic Sorption of Organic Acids by Variable-Charge Soils: Effect of Chemical Acidity and Acidic Functional Group S E U N G H U N H Y U N †,‡ A N D L I N D A S . L E E * ,† Department of Agronomy, Purdue University, West Lafayette, Indiana 47907-2054, and School of Civil Engineering, Purdue University, West Lafayette, Indiana 47907-2051
Sorption of organic acids by variable-charge soil occurs through both hydrophilic and hydrophobic sorption. In this study, the effect of chemical acidity and the type of acidic functional group on the relative contribution of hydrophilic and hydrophobic processes to sorption by a gibbsite-dominated and a kaolinite-dominated variable-charge soils was quantified by measuring sorption isotherms from different electrolytes (CaCl2, Ca(H2PO4)2, and KCl). The A1 soil is dominated by gibbsite whereas the DRC soil is primarily kaolinite. The organic acids investigated include five chlorinated phenols (pentachlorophenol, 2,3,4,6-tetrachlorophenol, 2,4,6-trichlorophenol, 2,4,5trichlorophenol, and 2,4-dichlorophenol) with pKa values ranging from 4.69 to 7.85 and two acidic herbicides (2,4-D (pKa ) 2.8) and prosulfuron (pKa ) 3.76)) that contain carboxyl and urea functional groups, respectively. Anion exchange of chlorinated phenols and prosulfuron on both variable-charge soils as well as 2,4-D sorption on the A1 soil was linearly correlated to chemical acidity. The effective positive surface charge [AEC/(AEC + CEC)] and the anionic fraction of the organic acid in solution, which are both pH-dependent, were sufficient to estimate the contribution of anion exchange to organic acid sorption except for 2,4-D sorption by DRC soil. The latter was much greater than would be predicted from the pKa of 2,4-D. Calcium bridging between silanol edge group and 2,4-D was hypothesized and corroborated by differences in sorption measured from KCl and CaCl2 solutions. For predicting contributions from hydrophobic processes, a log-log linear relationship between pH-dependent octanol-water (KowpH) and organic carbon-normalized sorption coefficients (KocpH) appeared adequate.
Introduction Many tropical and subtropical soils containing variablecharge minerals exhibit a positive charge within the environmentally relevant pH range, while several common agricultural and industrial organic acids of concern, such as acidic pesticides and chlorinated phenols, exist as anions. In * Corresponding author phone: (765)494-8612; fax: (765)496-2926; e-mail:
[email protected]. † Department of Agronomy. ‡ School of Civil Engineering. 10.1021/es0494914 CCC: $27.50 Published on Web 09/10/2004
2004 American Chemical Society
the pH range where organic anions coexist with positively charged soil surfaces, the contribution of anion exchange of organic anions must be considered in predicting sorption of organic acids. The contribution of anion exchange on organic acid sorption by variable-charge soil may vary with the magnitude of positive surface charge and the degree of organic acid dissociation, which are both pH-dependent. Variable-charge soils develop both positive and negative surface charge as a function of aqueous solution pH. At the pH equal to the point of zero net charge (PZNC), the soil surface will have the same amount of positive and negative charges, resulting in a zero net charge. At pH values lower than PZNC, the soil surface will have a net positive charge. Therefore, anion exchange of organic anions by variablecharge soil is impacted by both chemical and soil surface speciation. The fraction of an organic acid existing as an anion in aqueous solution (f1-R) can be estimated by solution pH and chemical acid dissociation constant (Ka): f1-R ) (1 + 10pKa-pH)-1. At pH ) pKa, the neutral and anionic species are in equal amounts. As pH increases for a given organic acid, the fraction of anion increases with more than 99% existing as an anion at pH > pKa + 2. At a given pH, as pKa decreases, the fraction of species as an anion increases. Likewise, the ionic distribution of amphoteric surface functional groups of variable-charge minerals, such as iron and aluminum oxides, can be estimated using surface dissociation constants. For example, assuming only one type of ionizable site on all possible mineral surfaces in the soil and no specifically adsorbable ions in aqueous solution, the S apparent surface acid ionization constants (Ka,app , mol/L) assuming a constant ionic strength can be expressed as follows (1, 2): S Ka1,app )
S ) Ka2,app
{SOH}(H+)
(1)
{SOH2+} {SO-}(H+) {SOH}
(2)
where {SOH2+}, {SOH}, and {SO-} are the concentrations (mol/kg) of the positively charged, neutral, and negatively charged soil surface groups, respectively, with S referring to an atom on the mineral surface such as Si, Fe, or Al, and (H+) is hydrogen ion activity (mol/L). {SO-} and {SOH2+} determine soil surface charge. The intrinsic value for surface S acidity constants (Ka,int) can be obtained from Ka,app by accounting for change in free energy during dissociation process by; S S Ka,int ) Ka,app exp
(∆ZFψ RT )
(3)
where F is the Faraday potential, ψ is the surface potential, and ∆Z is the change in surface species charge due to protonation and deprotonation of the surface species SOH. The concentrations of protonated and deprotonated surface hydroxyl groups are equal at the PZNC, where PZNC ) 0.5 S S (pKa1,int + pKa2,int ). The surface charge becomes increasingly more positive below the PZNC and increasingly negative above it as a result of the pH-dependent concentration of protonated and deprotonated surface hydroxyl groups. For a natural soil system, CEC (cation exchange capacity, cmol(-)/kg) and AEC (anion exchange capacity, cmol(-)/kg) are often used to represent the number of negative and VOL. 38, NO. 20, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5413
TABLE 1. Selected Physical and Chemical Properties of Soils soil
subgroup
pHa
clay (%)
OCb
Fec
Alc
AECd
CECd
PZNCe
major mineralogyf,g
A1 DRC
Petroferric hapludox Typic hapludox
6.1 4.7
41 81
1.38 1.34
133 54.3
24.7 6.75
0.42 0.72
1.71 3.34
5.1 3.1
Gb > Gt Kl > Gb, Gt, Hm, Qz
a 1:10 (g/mL) H O. b Organic carbon (%). c Dithionite-citrate-bicarbonate (DCB) extractable Fe or Al (g/kg). d Anion exchange capacity (cmol / 2 (+) kg) or cation exchange capacity (cmol(-)/kg) at the pH value reported. e The point of zero net charge measured by ion adsorption method. f Gb, g gibbsite [Al(OH)3]; Gt, goethite [R-FeOOH]; Hm, hematite [R-Fe2O3]; Kl, kaolinite [Al2Si2O5(OH)4]; Qz, quartz [SiO2]. Greater than signs indicate mineral amount relative to other minerals identified. The absence of a sign between minerals indicates that the relative amounts are approximately the same.
positive charges on the soil surface, respectively, at a given pH. At pH values greater than PZNC (CEC > AEC), organic anion sorption to exchange sites becomes less favored because (i) reduced anion exchange capacity by deprotonation of pH-dependent surface species and (ii) electrostatic repulsion from adjacent negative surface sites. At pH values lower than chemical pKa ({HA} > {A-}), anion exchange will decrease even though the soil surface may be more positively charged because the fraction of organic anions is reduced. Likewise, at higher pH values, although the fraction of organic anions in solution increases, the speciation of surface sites shift to be more negative. Therefore, the maximum sorption occurs where the product of the number of active sites and the concentration of anionic species is highest (3). Both hydrophobic partitioning and hydrophilic sorption (e.g., anion exchange) will contribute to organic acid sorption from 5 mM CaCl2; competition for anion exchange sites by chloride ions at this concentration will be negligible (4, 5). However, in Ca(H2PO4)2 solutions, preferential sorption of phosphate to the hydrophilic mineral surface by presumably forming specific and/or irreversible mono or binuclear complexes on the oxide surface will block anion exchange sites to organic acids (4, 6, 7). The effect of dissolved phosphate on organic acid sorption to hydrophobic domains was also previously shown to be negligible for picloram sorption to humic materials (8). Therefore, a decrease in chlorinated phenol sorption in the presence of phosphate is assumed to represent the hydrophilic contribution to sorption. This assumption proved reasonable in estimating the extent of anion exchange to pentachlorophenol and prosulfuron sorption on several variable-charged sorbents (9, 10). The fraction of hydrophilic sorption (fHphilic) was calculated as follows (9, 10):
fHphilic )
Kd,CaCl2 - Kd,Ca(H2PO4)2
(4)
Kd,CaCl2
where Kd,CaCl2 and Kd,Ca(H2PO4)2 are the linear soil-water distribution coefficients (changes in sorbed concentrations relative to solution concentrations, L/kg) of organic acids from 5 mM CaCl2 and 5 mM Ca(H2PO4)2, respectively. In a similar manner, Kd values of organic acids collected from 10 mM KCl solution (Kd,KCl) were used to determine the fraction of hydrophilic sorption that involved association with calcium (fCa):
fCa )
Kd,CaCl2 - Kd,KCl
(5)
Kd,CaCl2
Organic anions have been shown to sorb to mineral surfaces in association with divalent cations (11-14). At a given pH, organic acid sorption through hydrophobic partitioning (KpH oc , L/kgoc) was estimated as follows:
Kd,CaCl2
KpH oc (L/kg,oc) ) (1 - fHphilic) 5414
9
foc
)
Kd,Ca(H2PO4)2 foc
(6)
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 20, 2004
In previous studies, we demonstrated that anion exchange in variable-charge soils is significant for pentachlorophenol (PCP) (9) and the herbicide prosulfuron (10) and that the extent of anion exchange correlated well to the ratio of pHdependent anion exchange capacity (AEC, cmol(+)/kg) to cation exchange capacity (CEC, cmol(-)/kg) (i.e., AEC/CEC) as well as the ratio of AEC to the total number of soil surface charges (AEC + CEC) (i.e., AEC/(AEC + CEC)). In the current study, the effect of chemical acidity (pKa) and acidic functional group on contribution of anion exchange to organic acid sorption by two variable-charge soils was assessed using a series of five chlorinated phenols and two herbicides (2,4-D and prosulfuron) varying in acidity and the acidic functional group.
Materials and Methods Soils. Two variable-charge soils, A1 and DRC, collected from Brazil were used as model sorbents. According to the U.S. soil taxonomy system, these soils are Oxisols, highly weathered soils containing variable-charge minerals such as iron oxide, aluminum oxide, and kaolinite. The dominant minerals in the clay fraction of the A1 and DRC soils are gibbsite (Al(OH)3) and kaolinite (Al2Si2O5(OH)4) (Table 1). Reported PZNC values for gibbsite and kaolinite are 8 to 9 and 4 to 5, respectively (15). In previous sorption studies with pentachlorophenol and prosulfuron, a significant contribution (64-82%) of the total sorption was due to anion exchange for both the A1 and DRC soils (9, 10). The soils were airdried, passed through a 2-mm sieve, and stored in closed containers at room temperature prior to use. Selected soil characteristics including pH, organic carbon content, dithionite-citrate-bicarbonate (DCB) extractable Al and Fe, CEC, AEC, and PZNC are summarized in Table 1. Surface charge measured by the KCl saturation method (16) showed both soils have significant anion exchange capacity at their natural pH (Table 1). Surface charge estimation by the ion sorption method has been reported to be more reliable than potentiometric titration, which can overestimate variable charge resulting from hydroxide dissolution, especially at pH > 7 and pH < 4 (17, 18). Details with regard to soil surface charge as a function of pH and clay mineral identification and characterization methods were previously reported (9). Chemicals. Five chlorinated phenols [pentachlorophenol (PCP), 2,3,4,6-tetrachlorophenol (2,3,4,6-TeCP), 2,4,6-trichlorophenol (2,4,6-TCP), 2,4,5-trichlorophenol (2,4,5-TCP), and 2,4-dichlorophenol (2,4-DCP)] and two acidic herbicides [2,4dichlorophenoxy acetic acid (2,4-D) and prosulfuron (1-(4methoxy-6-methyl-triazin-2-yl)-3-[2-(3,3,3-trifluoropropyl)phenylsulfonyl]-urea)] were selected as organic acids varying in acidity and acidic functional groups. Structures of 2,4-D, prosulfuron, and PCP as a representative of the chlorinated phenols are presented in Figure 1. The selected physical and chemical properties such as aqueous solubility (Sw), octanol/ water partition coefficient (log Kow), and acid dissociation constant (pKa) are presented in Table 2. The chlorinated phenols and 2,4-D were purchased from Aldrich Chemical Co. (Milwaukee, WI), and prosulfuron was purchased from Chem Service (West Chester, PA) with reported purity of
TABLE 2. Selected Physical and Chemical Properties of Organic Acids Investigated chemical
MW (g/mol)a
pKab
Sw (mg/L)c
log Kow,HAc
log Kow,Ad
2,4-dichlorophenoxy acetic acid (2,4-D)e
221
2.80
2.83
-0.75
prosulfuronf
419
3.76
2.79g
-0.76g
pentachlorophenol (PCP)
266
4.69h
5.24
1.59
2,3,4,6-tetrachloropheonol (2,3,4,6-TeCP) 2,4,6-trichlorophenol (2,4,6-TCP) 2,4,5- trichlorophenol (2,4,5-TCP) 2,4-dichlorophenol (2,4-DCP)
232 197 197 163
5.40 6.15 6.94 7.85
311 (pH 1) highly soluble (pH>5) 30 (pH 5.1) 3580 (pH 6.8) 3.54 (pH < 4.1)i 1153 (pH > 7)i NAj 800 1190 4500
4.42 3.72 4.19 3.23
0.75 0.00 0.09 -0.46
a Data from Verschueren (19) unless otherwise noted. b Dissociation constant from Schellenberg et al. (20) unless otherwise noted. c Log K ow for neutral species in 0.1M KCl from Schellenberg et al. (20) unless otherwise noted. d Log Kow for ionized species from Escher and Schwarzenbach e f g h (21) unless otherwise noted. Ref 22. Ref 23. Extrapolated value with Kow data at pH 5 and pH 9 from ref 23. Lee et al. (24) in 0.01 N CaCl2. i Ref 25. j Not available.
FIGURE 1. Structural formula of the organic acids discussed in this study. The asterisk indicates the proton that undergoes dissociation. >98%. Calcium chloride dihydrate was purchased from Fisher Scientific (Pittsburgh, PA). Calcium bis(dihydrogenphosphate) monobasic and ammonium fluoride were purchased from Fluka (Buchs, Switzerland). Acetonitrile at greater than 99% purity was purchased from Mallinckrodt Laboratory chemical (Phillipsburg, NJ). Sorption Studies. Sorption of all organic acids by both variable-charge soils was measured from 5 mM CaCl2 and Ca(H2PO4)2 solutions Additionally, sorption of the two herbicides was also measured from 10 mM KCl (same charge equivalents as 5 mM Ca solution). Organic acid solutions applied to soils were prepared in electrolyte solutions by diluting a methanol or acetonitrile solution containing a high organic acid concentration (2400-6200 µmol/L). The resulting methanol or acetonitrile concentrations were less than 0.5 vol %. Isotherms were measured independently for each solute and included five solute concentrations plus a zero solute control. Soils and organic acid solutions were added to preweighed 35-mL glass tubes with Teflon-lined screw caps at soil-solution ratios ranging from 1:5 to 1:20, which resulted in 20-60% sorption of the spiked chemical after a 24-h contact time. To avoid pH effects on sorption isotherms, dilute acid/base adjustments were made to achieve similar equilibrium pH in each electrolyte system near the natural soil pH. Soil slurries were rotated end-over-end (30 rpm) at 23 ( 2 °C, centrifuged at 1750g for 20-30 min, and a supernatant aliquot taken for HPLC analysis. Equilibrium pH was measured immediately after screw caps were removed, and the pH value in each chemical-soil-solution combination was near targeted values with a standard deviation PZNC where sorption from both phenoxyacetic acids was negligible. Similarly, Kung and McBride (28) observed proportionally greater sorption with increasing chemical acidity by noncrystalline iron oxide (PZNC ≈ 8.5) at pH 5.4 where surfaces are predominantly positively charged: 2,4,6-TCP (6.15) > 2,4-DCP (7.85) > 3,4DCP (8.63) > 3-MCP (9.10), where the numbers in parentheses denote pKa values. Kung and McBride (28) also observed increasing sorption with increasing pH at pH ranges well below the pKa of the organic acid as a result of increasing dissociation. For example, for 2,4-DCP (pKa ) 7.85), sorption increased upon 5416
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 20, 2004
an increase in pH from 4.6 to 6.2 even though the organic anion fraction only increased from negligible to 2%. Likewise, in our current study, anion exchange of 2,4-DCP accounted for more than 30% of observed sorption even though f1-R is less than 1% (Table 3). Note that even though only a small fraction of the anion may exist in the aqueous phase, as the anion is sorbed the neutral species in solution dissociates to maintain equilibrium defined by the pH-pKa relationship. Therefore, the fraction of the chemical sorbed as an anion is a result of this coupled process similar to what was previously exemplified for cation exchange of organic amines (29). The findings from this study and previous literature studies support the fact that (i) sorption of chlorinated phenols by variably charged soil is primarily controlled by the extent of dissociation as long as anion exchange sites are not limited; and that (ii) preference of ion exchange can be accounted for by the greater free-energy shift arising from
TABLE 3. Sorption Coefficients from Linear Model Fits to Organic Acid Isotherms Measured on A1 and DRC Soils in 5 mM CaCl2 (Cl), 5 mM Ca(H2PO4)2 (P), and 10 mM KCl (K) Solutionsa Al soil
DRC soil
AECb ) 0.54, CECb ) 1.46, pHc ) 5.8, PZNCd ) 5.1
AECb ) 0.96, CECb ) 2.76, pHc ) 4.2, PZNCd ) 3.1
matrixe
Kd (L/kg)
f1-rf
fHphilicg
2,4 DCP
Cl P
1.74 (0.12)j 1.19 (0.07)
0.0088
0.32
1.93
2,4,5-TriCP
Cl P
6.97 (0.62) 4.58 (0.43)
0.068
0.34
2,4,6-TriCP
Cl P
3.14 (0.28) 1.56 (0.15)
0.31
2,3,4,6-TeCP
Cl P
9.03 (0.91) 3.38 (0.32)
PCPk
Cl P
prosulfuron
2,4-D
chemical
fCah
f1-rf
fHphilicg
6.51 (0.65) 3.52 (0.38)
0.0002
0.46
2.42
2.52
19.6 (1.4) 9.44 (0.56)
0.0018
0.52
2.85
0.50
2.06
11.1 (1.3) 5.15 (0.52)
0.011
0.53
2.59
0.72
0.63
2.38
62.9 (6.4) 27.8 (2.4)
0.059
0.56
3.31
21.0 (2.00) 6.54 (0.61)
0.93
0.69
2.67
108 (32) 38.9 (6.2)
0.24
0.64
3.46
Cl P K
0.62 (0.02) 0.11 (0.01) 0.65 (0.06)
0.99
0.82
0.0l
0.91
7.19 (0.22) 2.48 (0.17) 6.98 (0.41)
0.73
0.66
0.0l
2.26
Cl P K
4.49 (0.13) 0.26 (0.11) 4.59 (0.20)
1.00
0.94
0.0l
1.29
16.6 (0.44) 1.06 (0.03) 12.6 (0.68)
0.96
0.94
0.24l
1.87
log KocpHi
Kd (L/kg)
fCah
log KocpHi
a The coefficient of determination (r 2) for linear isotherm fit was g0.93. b Anion exchange capacity (cmol /kg) and cation exchange capacity (+) (cmol(-)/kg) at the isotherm pH estimated from curve fits exchange capacity reported from ref 9. c Averaged equilibrium isotherm pH with a standard deviation less than 0.1. d The point of zero net charge. e Cl, P, and K represent 5 mM CaCl2, 5 mM Ca(H2PO4)2, and 10 mM KCl solutions, respectively. f Fraction of organic acids existing as an anion calculated using a pKa value reported in Table 2. g Fraction of hydrophilic sorption calculated using eq 4. h Fraction of cation effect calculated using eq 5. i Organic carbon-normalized hydrophobic sorption calculated by eq 6. j Values in parentheses are the 95% confidence intervals. k Data from ref 9. l Insignificant values with a significance level R ) 0.05.
FIGURE 3. Correlation between hydrophilic sorption of organic acids and chemical pKa on A1 and DRC soils. Solid line is a liner regression fit using the chlorinated phenol data, and the dotted line is an extrapolation. electrostatic interaction over hydrophobic partitioning (30, 31). Effect of Acidic Functional Group. The fraction of hydrophilic sorption of 2,4-D and prosulfuron versus pKa is shown in Figure 3 (open triangles and open squares) along with the chlorinated phenol data (filled circles). For 2,4-D and prosulfuron on the A1 soil, fHphilic increased with increasing chemical acidity (decreasing pKa) as would be predicted by extrapolation (dotted line) of the chlorinated
phenol data, and likewise for prosulfuron on the DRC soil. However, for 2,4-D on the DRC soil, hydrophilic sorption is much greater than would be predicted from the other data (Figure 3B). Stumm et al. (1) proposed that organic acids with a carboxylic group could be strongly sorbed to variable-charge minerals by forming an inner-sphere complex with the oxide surface. Such surface complexes are unlikely in our the current study given that greater than 90% of 2,4-D was easily extractable from both soils, and 2,4-D sorption on the oxidic A1 soil was sufficiently explained in terms of nonspecific anion exchange under the experimental conditions investigated. Many inorganic/organic anions have been shown to sorb to mineral surfaces in association with a divalent cation (6, 11-14, 32, 33). Mechanisms that have been suggested for cation-induced sorption of inorganic/ organic anions include the following: (i) formation of a surface complex (e.g., cation bridging) between the anion and divalent cation, reducing the repulsive force between adjacent anions; (ii) formation of a positively charged Ca2,4-D complex in solution, which is subsequently sorbed to a cation exchange site; (iii) precipitation reactions occurring at high pH; and (iv) divalent cation-induced increase in positive charges on the soil surface. In a previous study, PCP sorption was similar in CaCl2 and KCl solutions (9); therefore, the latter mechanism iv is not likely to be significant in the current study. Mechanism iii, which is a type of surface precipitation process, is also unlikely given the low solute concentrations used in this study. Mechanism ii is also unlikely due to the weak association of Ca and 2,4-D in an aqueous solution, thus formation of stable aqueous ion pairs is unlikely. To quantify the effect of calcium, 2,4-D and prosulfuron isotherms were measured in 10 mM KCl solution on both soils and compared to what was observed in the calcium systems. No difference between KCl and CaCl2 systems on PCP sorption by variable-charge soil was previously observed (9); therefore, we assumed that cation effect on sorption of other phenolic acids investigated in this study was also VOL. 38, NO. 20, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5417
FIGURE 4. Fraction of positively charged soil sites (AEC) relative to the total number of charged surface sites (AEC + CEC) and anionic chlorinated phenols (f1-r) in aqueous phase as a function of pH.
negligible. Kd values of 2,4-D and prosulfuron sorption from KCl are shown in Table 3 along with fraction of sorption induced by calcium (fCa; eq 5). For prosulfuron sorption on both soils and 2,4-D sorption on the A1 soil, Ca effects were negligible. However, 2,4-D, sorption by DRC soil was approximately 24% less in the KCl system as compared to the CaCl2 system, thereby, exemplifying Ca-induced enhancement in 2,4-D sorption by DRC soil. Upon eliminating the Ca-associated sorption contribution, the remaining hydrophilic sorption fraction for 2,4-D by the DRC soil follows the trend predicted by the line extrapolated from the chlorinated phenols (Figure 3). DRC predominantly consists of kaolinite, which will provide silanol edge sites for forming Ca bridges with the carboxylic functional group of 2,4-D, whereas the gibbsitedominated A1 soil consists of neither kaolinite nor silica in the clay (
