A Modified Polanyi-based Model for Mechanistic ... - ACS Publications

The Polanyi theory was originally developed to account for the adsorption of gas molecules to porous materials.(16) It has been widely accepted as a p...
3 downloads 0 Views 1MB Size
Download PDF
Article pubs.acs.org/est

A Modified Polanyi-based Model for Mechanistic Understanding of Adsorption of Phenolic Compounds onto Polymeric Adsorbents Bingjun Pan and Huichun Zhang* Department of Civil and Environmental Engineering, Temple University, 1947 North 12th Street, Philadelphia, Pennsylvania 19122, United States S Supporting Information *

ABSTRACT: To obtain mechanistic insight into adsorption of phenolic compounds by two representative polymeric adsorbents, XAD-4 (polystyrene) and XAD-7 (polymethacrylate), a modified Polanyi-based Dubinin-Ashtakhov (D−A) model was developed based on a unique combination of the Polanyi theory, polyparameter linear energy relationships and infinitely dilute solution in n-hexadecane as the reference state. The adsorption potential in the D−A model ε = −RTln(Cw/Csat w) was redefined by replacing the term (Cw/Csat ) with the w normalized equivalent concentration in n-hexadecane (CHD), where Cw is the aqueous equilibrium concentration and Csat w is the aqueous solubility of the solute. Using the new reference state allows quantitative comparison among various solutes. By fitting adsorption isotherms to the modified model using εHD = −RTln(CHD/10 000), a new normalizing factor (Em) was obtained to quantify the contributions of specific interactions (i.e., H-bonding, dipolar/polarizability, etc.) to the overall adsorption energy. Significant linear correlations were established between “A”, the hydrogen-bond acidity, and “Em” for the investigated compounds, suggesting that, in addition to hydrophobic interactions, hydrogen-bonding is predominantly responsible for the adsorption of phenols by XAD-4 and XAD-7. Additionally, adsorption capacity and affinity of phenolates were significantly less than those of phenols; another model was proposed to accurately predict the effect of pH on the adsorption behavior of phenols.

■

INTRODUCTION The continuous discharge of emerging contaminants such as pharmaceutical and personal care products and endocrine disrupting chemicals into aquatic environments requires the development of an effective treatment technology to remove them from drinking water and wastewater effluents.1−4 Adsorption on activated carbon is often considered as a method of choice although poor removal was observed for particularly polar and hydrophilic contaminants.2,3 As a potential alternative to activated carbon, polymeric adsorbents (i.e., resins) have been widely used to remove organic pollutants from industrial wastewaters or natural waters.5−9 Polymeric adsorbents typically have high surface area, strong mechanical rigidity, adjustable pore structure and surface polarity, and are amenable to easy regeneration under mild conditions.6,10 Understanding the molecular interactions between polymeric adsorbents and contaminants is vital to choosing an appropriate adsorbent, tailor synthesizing specific adsorbents, and choosing optimal regenerant. Generally, the primary mechanism controlling the adsorption of organic compounds by polymeric adsorbents is thought to be pore-filling due to one or a combination of the following forces:6,11 hydrophobic (i.e., nonspecific) interactions between the nonpolar moieties of the solute and those of the adsorbent © 2012 American Chemical Society

matrix; dipolar/polarizability effects (D/P) including both dipole−dipole (D−D) interactions and polarizability; π−π electron-donor−acceptor interactions (EDA), H-bonding interactions, and electrostatic interactions. The unfavorable interactions between the solute and the solvent have also been considered. However, despite a general knowledge of the possible interaction mechanisms, most studies have been confined to rather phenomenological descriptions of the adsorption process,12−15 and the involved interaction forces at the molecular level remain largely obscure. Additionally, there is no predictive model available to estimate the adsorption capacity of a given organic compound on a polymeric adsorbent. The Polanyi theory was originally developed to account for the adsorption of gas molecules to porous materials.16 It has been widely accepted as a powerful theory for the adsorption of both gas and aqueous solutes toward heterogeneous surfaces and pores such as activated carbon and carbon nanomaterials.17−19 Briefly, the adsorption potential (ε) in the Polanyi Received: Revised: Accepted: Published: 6806

March 9, 2012 May 7, 2012 May 11, 2012 May 11, 2012 dx.doi.org/10.1021/es300951g | Environ. Sci. Technol. 2012, 46, 6806−6814

Environmental Science & Technology

Article

adsorbents. As a result, contributions from complex solute− water interactions are removed, which can significantly simplify the interpretation of the involved adsorption mechanisms between solutes and adsorbents. In addition, nonspecific interactions between an adsorbent and a solute are now accounted for in the HD-normalized concentration. Only specific interactions remain between the adsorbent and the solute. Inspired by the above normalization approach, the present study for the first time proposed a modified D−A model by combining the D−A model with the normalized n-hexadecane (HD) equivalent concentration approach. In this approach, the original adsorption potential ε = −RTln(Cw/Csat w ) was redefined as εHD = −RTln(CHD/10 000) (see details below), such that contributions from solute−water interactions can be minimized or eliminated from the normalizing factor “E” of the D−A model. Comparison of the thus obtained “E” values of different compounds enables quantifying the interaction forces between the adsorbate and the adsorbent. The proposed model was then applied to the adsorption behavior of bisphenol A, a potential endocrine disrupting chemical, and seven model compounds on two commercially available polymeric adsorbents (XAD-4, polystyrene; and XAD-7, polymethacrylate). The results are expected to yield mechanistic insight into the adsorption mechanism of phenolic contaminants by polymeric adsorbents. Modified D−A Model and Polyparameter Linear Free Energy Relationships (pp-LFERs). Using infinitely dilute solutions in HD as the reference state, Zhu and Pignatello26 converted aqueous solute concentrations to their equivalent concentrations in HD at the same activity according to

theory was defined as the energy that is required to move a molecule from the attractive force field of the solid surface to bulk solution. The adsorption potential varies with solute concentration and can be calculated by the following equation: ε = −RT ln(Cw /Cwsat)

(1)

where R (0.008314 KJ/mol·K) is the universal gas constant, T (K) is the absolute temperature, Csat w (μmol/L) is the aqueous solubility of the solute, and Cw (μmol/L) is the bulk liquidphase equilibrium concentration. When plotting the equilibrium adsorption capacity of the solute (qe, μmol/g), versus ε, a temperature-independent “characteristic curve” can be obtained for a single solute.19 In the Polanyi-Manes model (PMM),19 the molar volume of solute (Vm) was used as the normalizing factor to collapse the characteristic curves of various solutes to a single curve: ⎛ ε ⎞b log qe = log Q 0 − a⎜ ⎟ ⎝ Vm ⎠

(2)

o

where Q (μmol/g) is the maximum adsorption capacity of the solute, and a and b are the fitting parameters. Unfortunately, Vm only accounts for nonspecific interaction forces.20 eq 2 thus failed to yield a single correlation curve particularly when polar organic solutes were involved.18,21 To theoretically capture all of the interaction forces responsible for adsorption including hydrophobic interactions, EDA interactions, H-bonding interactions etc., another normalizing factor (Eo) (KJ/mol) was proposed in the PolanyiDubinin model (or Dubinin−Ashtakhov model, D−A model):20,22,23 ⎛ε⎞ log qe = log Q 0 − ⎜ ⎟ ⎝ Eo ⎠

const resin ⎞ ⎛ n ⎟ C HD = Cw(χHD,W )m exp⎜ − + ⎝ RT RT ⎠

b

(4)

Where CHD (μmol/L) represents the hydrophobic-effect normalized solute concentration in HD; χHD,W is the hexadecane-water partition coefficient in mole fraction; m and n are concentration-dependent coefficients (see “Data Analysis” for the calculation of m and n); and constResin can be obtained by

(3)

where Eo is the adsorption energy for a given solute, which was supposed to account for all the interaction forces involved in adsorption.17,20 Crittenden et al.20 demonstrated a significant improvement in predicting the adsorption capacity of eight adsorbents and 56 organic compounds by combining the D−A model with linear solvation energy relationships. Despite this limited success, ε in all Polanyi-based models is a solubilitynormalized parameter. Solubility is based on pure (or subcooled) liquid as the reference state and thus is compound specific.24,25 Normalizing hydrophobic effects by aqueous solubility would introduce the effect of solute−solute interactions at the reference state which differs for each solute and would render quantitative comparison among solutes problematic. Furthermore, normalization by aqueous solubility would introduce the effect of solute-water interactions which also differs for each solute. For example, the aqueous solubility of phenol (a dipolar compound with a H-donating substituent) is much larger than that of benzene (a nonpolar compound) mainly due to the additional H-bonding interactions between phenol and water molecules. Recently, Zhu and Pignatello26 normalized hydrophobic interactions by switching the reference state from a pure (or subcooled) liquid state to a common reference state − infinitely dilute solution in n-hexadecane (HD). HD is an inert solvent that can undergo only nonspecific interactions with all solutes, irrespective of their polarity. Using this new reference state, adsorption of solutes from water by adsorbents is conceptually transformed to adsorption of solutes from HD by water-wet

⎛ V ⎞ const resin w ⎟⎟ − 1 = ln⎜⎜ RT ⎝ Vm·ρresin ⎠

(5)

Where Vw and Vm are the molar volume of water (0.018 L/ mol) and organic solute and can be calculated based on atomic volumes and the number of bonds;27 and ρresin (g/mL) is the density of the adsorbent. Based on CHD, the adsorption potential was redefined as εHD = −RTln(C HD/10 000)

(6)

The “10 000” in eq 6 is an arbitrary scaling factor to converge the model fitting. We then got the modified D−A Model by replacing ε with εHD, ⎛ εHD ⎞bm log qe = log Q − ⎜ ⎟ ⎝ Em ⎠ 0

(7)

Here, the contribution of hydrophobic interactions has been included in εHD and thus is excluded from the new normalizing factor “Em” (KJ/mol). Polyparameter linear free energy relationships (pp-LFERs) have gained increasing attention in environmental research by quantitatively considering energy contributions of multiple molecular interactions between a solute and a sorbent.28,29 For 6807

dx.doi.org/10.1021/es300951g | Environ. Sci. Technol. 2012, 46, 6806−6814

Environmental Science & Technology

Article

Table 1. Density, BET Surface Area, BJH Pore Diameter, Pore Volume and Pore Size Distribution of XAD-4 and XAD-7 BJH pore size distribution, percent of total pore volume XAD-4 XAD-7

density (g/mL)

BET surface area (m2/g)

BJH average pore diameter, (nm)

BJH pore volume (cm3/g)

50 nm

1.09 1.25

829 495

6.19 9.75

1.25 1.12

12.4% 2.3%

86.8% 90.7%

0.8% 7.0%

Table 2. Physical−Chemical Properties of the Selected Organic Compounds

a

Sw: aqueous solubility, obtained From refs 17, 32. bData obtained from refs 17, 33, 34. cN-hexadecane-water partitioning coefficient obtained from ref 35. dCalculated from SPARC online calculator and calibrated according to the difference in the values reported by ref 35 and calculated by SPARC. eCalculated by χ = KHW·(MWH/DH)/(MWW/DW), MWH, DH and MWW, DW are molecular weight and density of n-hexadecane and water, respectively. f Vm: solute intrinsic molar volume, calculated from atomic volumes27. gData obtained from refs 35−37. hEstimated based on ref 27.

■

MATERIALS AND METHODS Adsorbents and Solutes. Two nonionic cross-linked polymeric adsorbents, Amberlite XAD-4 and XAD-7, were provided by Rohm and Haas (U.S.). Prior to use, the adsorbents were sieved with the sizes from 0.4 to 0.6 mm and subjected to intensive wash by NaOH, HCl and methanol to remove possible residue impurities (see details in Supporting Information (SI) Text S1). Specific surface areas and pore size distribution of the adsorbents were obtained by N2 adsorption at 77 K using a Micrometrics ASAP-2010C automatic analyzer, and the results indicate XAD-4 is more microporous and has a higher surface area than XAD-7 (Table 1). Solutes including bisphenol A, nitrobenzene and six substituted phenols were all purchased from Fisher Scientific (SI Text S1) and used without further purification. Listed in Table 2 are the solutes, their abbreviations and relevant properties. Adsorption Experiments. All adsorption isotherms were carried out in amber glass bottles with Teflon-lined screw caps at 23 ± 0.5 °C (see details in SI Text S2). Briefly, more than 15 experimental data points for each test compound were employed with the equilibrium concentration ranging from 10−5−10−3 to >10−1 of its solubility. The ratios of aqueous solution to solids were adjusted to achieve 25−75% adsorption of the target compounds. 0.01 M HCl and 0.1 M NaOH solutions were used to adjust the solutions pH if necessary. The pH value of isotherm experiments was typically around 7, except those for 4-NP where the pH value was kept at 4 to avoid the dissociation of 4-NP. Preliminary sorption experiments indicated no substantial difference with equilibrium time ranging from 1 to 8 days (SI Figure S1). Then, the bottles containing a mixture of solute and resin were transferred to a shaker with a thermostat under 175 rpm for either 48 h (for single-ring aromatic compounds) or 4 days (for 2-NAPH and

example, equilibrium partition (Koc) of organic compounds between water and natural organic matter has been expressed as28 log K OC = e E + sS + a A + b B + v V + c

(8)

where the excess molar refraction (E) accounts for the nonspecific interactions that include London dispersive (induced dipole−induced dipole, ID−ID) and Debye forces (dipole−induced dipole, D−ID); S accounts for the D/P effect (mainly polarizability for aromatic compounds) which has some overlap with the E term;26 A and B are descriptors for the overall hydrogen-bond acidity and basicity; V is the McGowan’s characteristic molecular volume which accounts mainly for cavity formation; and c mainly depends on volume entropy effects. The above descriptors for many chemicals are either readily available30,31 or can be estimated based on the existing methods.28 The coefficients e, s, a, b, and v in eq 8 are fitting parameters that represent the differences in the descriptors between water and the adsorbent. Zhu and Pignatello26 believed that the contribution of the HD-based hydrophobic interactions is comparable to the sum of eE and vV. Thus, by using the HD-normalized solute concentration (CHD) in the modified D−A model (eq 7), the new normalizing factor Em should only include the specific interactions responsible for adsorption:20,26 Em = s S + a A + b B + c

(9)

By conducting multiple linear regressions between the obtained Em values and the molecular descriptors, we can estimate the relative contribution of all specific interactions to the overall adsorption free energy. 6808

dx.doi.org/10.1021/es300951g | Environ. Sci. Technol. 2012, 46, 6806−6814

Environmental Science & Technology

Article

BPA). After reaching apparent equilibrium, the experimental supernatants were withdrawn for analysis by high-performance liquid chromatography (HPLC). Less than 2% loss of the dissolved organic compounds in control bottles was observed over the duration of the experiments. Adsorption isotherms were developed based on the differences between initial and equilibrium solute concentrations. Data Analysis. Equation 4 can be written in its logarithmic form as ln C HD = ln Cw + m ln χHD,W −

const resin n + RT RT

(10)

After rearrangement, we get const resin ⎞ ⎛ ⎛ n ⎞⎟ ⎜ln C ⎟ = m ln χ + ⎜ln Cw − HD − HD, W ⎝ ⎠ ⎝ RT RT ⎠

Figure 1. 4-NP adsorption isotherms on XAD-4 at different pH; the solid and dashed lines represent the results of model fitting and model prediction by eq 13, respectively.

(11)

Where CHD is equivalent to the equilibrium adsorption capacity when there is only hydrophobic interactions between the adsorbent and the solute, that is, lnCHD = lnqe.26 For any arbitrary value of Cw, one can calculate the corresponding qe for each solute based on the original D−A model (eq 3). Then regressions can be performed on a “training set” of compounds. For each value of Cw, the term (lnCw − n/RT) is constant, but (lnCHD − constresin/RT) and (lnχHD,W) are compound specific. Since there were multiple compounds in the training set, (lnCHD − constresin/RT) and (lnχHD,W) were treated as the dependent and independent variable. The regressions of eq 11 were conducted using the analytic software SAS 9.2 to obtain m (slope) and (lnCw − n/RT) (intercept) for each value of Cw. The relationships between m, n, and Cw were then obtained by correlating the obtained m and n values with the corresponding Cw values. These relationships were then substituted into eqs 4, 6, and 7. Using a commercial software program (OriginPro 8.5), parameters Em and bm were obtained by fitting eq 7 to the converted adsorption isotherms of qe versus CHD. Correlation coefficients (R2) and mean weighted square errors (MWSE) were used to evaluate the goodness of fit,17

are the molar fractions of the nondissociated and dissociated species, respectively. Using a similar approach, the D−A model was modified by considering the effect of pH on the adsorption coefficients: qe = (Q 0,N·f N + Q 0,I

qexp i

∑i = 1 [(qiexp − qical)/qiexp]2 N−3

N + b ·f I I

(13)

where Q0,N and Q0,I (μmol/g) are the maximum adsorption capacity for phenols and phenolates; and EN,EI (KJ/mol) and bN,bI are the normalizing factor and fitting parameter for phenols and phenolates, respectively. The model parameters Q0,N, EN and bN can be obtained by fitting the original D−A model to the adsorption isotherm at pH 4 where almost all 4-NP are nondissociated ( f N ≈1, f I ≈ 0). We thus got Q0,N = 4904 μmol/g, EN =12.7 KJ/mol and bN = 1.04. By substituting these values into eq 13 and fitting to the adsorption isotherm at pH 7.15 (pH = pKa) where f N = f I = 0.5, we obtained Q0,I = 480 μmol/g, EI = 8.0 KJ/mol and bI = 1.2. Note that the removal percentages at pH 10.5 ( f N ≈ 0, f I ≈ 1) were too low ( 3-NP > 4-NP > phenol for XAD-4, and 4-CP > 3-NP ≈ 4-NP > 4-MP > phenol for XAD-7. The major difference in the two orders is the position of 4-MP. 4-MP has a large logKHW (−0.19, Table 2) and thus is more hydrophobic. We expect that the hydrophobic interactions play a more important role in the adsorption by XAD-4 than by XAD-7. After converting to the hydrophobic effect normalized isotherms, qe follows the same order of 4-NP > 3-NP > 4-CP > phenol ≈ 4-MP for both XAD-4 and XAD-7, that is, nitro group > chloride group > methyl group. This tendency is corroborated by the fitted “Em” values of the modified D−A model (Table 4), where the two nitrophenols have the largest “Em” values among the studied phenols. This observation can be well explained by the electronic effects of the ring substituents. Hammett constants indicate that nitro groups are strongly electron-withdrawing (σm = 0.73, σp = 1.25), parachlorine is weakly electron-withdrawing (σp = 0.22), and methyl is slightly electron-donating (σp = −0.16).32 The

Figure 3. Comparison of the original isotherms and the hydrophobic interaction normalized isotherms of phenols (i.e., 4-NP, 3-NP, 4-CP, 4-MP and phenol) on (A) XAD-4 and (B) XAD-7.

strongly electron-withdrawing groups will significantly increase the polarizability and the acidity of the phenols which will lead to stronger EDA and H-bond interactions with the adsorbent. Indeed, strong linear correlations can be established between Em and σ values for the simple substituted phenols: XAD‐4: Em = 1.96σ + 7.28 (N = 5, p < 0.02, R2 = 0.90)

(14)

XAD‐7: Em = 1.27σ + 7.13 (N = 5, p < 0.05, R2 = 0.78)

(15)

Modified D−A Model Results: pp-LFERs. The hydrophobic effect normalized isotherms of the other target compounds, that is, BPA, NB, and 2-NAPH, are illustrated in SI Figure S5. The related fitting parameters based on the modified D−A model are listed in Table 4. To understand the relative contribution of all specific interactions to the overall adsorption energy, multiple linear regression was conducted between the Em values and the molecular descriptors for polarizability (S) and H-bond acidity (A) and basicity (B). The best regression model was identified (based on the lowest Mallow’s Cp value) to include A as the only independent variable (Figure 4): 6811

dx.doi.org/10.1021/es300951g | Environ. Sci. Technol. 2012, 46, 6806−6814

Environmental Science & Technology

Article

Table 4. Modified D−A Model Fitting Parameters for the Adsorption Isotherms Measured for XAD-4 and XAD-7 XAD-4 compound NB phenol 4-MP 4-CP 4-NP 3-NP 2-NAPH BPA

Em (KJ/mol) 2.26 7.04 6.75 8.06 9.37 9.18 6.19 11.53

± ± ± ± ± ± ± ±

0.06 0.05 0.014 0.15 0.50 0.35 0.24 0.21

bm 1.11 1.36 1.36 1.25 1.27 1.36 1.54 2.88

± ± ± ± ± ± ± ±

0.05 0.03 0.07 0.05 0.09 0.09 0.12 0.38

XAD-7 R2

MWSE

(E0-Em)/E0

0.999 0.999 0.998 0.999 0.998 0.999 0.998 0.996

0.003 0.006 0.010 0.022 0.011 0.012 0.085 0.025

79.5% 50.8% 55.9% 48.7% 26.2% 35.4% 62.0% 28.4%

XAD‐4: Em = (8.12 ± 0.62) × A + (2.21 ± 0.45) (16)

XAD‐7: Em = (5.63 ± 0.33) × A + (3.67 ± 0.24) (N = 8, p < 0.001, R2 = 0.98)

3.69 6.98 6.57 7.93 8.40 8.39 6.85 10.21

± ± ± ± ± ± ± ±

0.39 0.07 0.19 0.17 0.44 0.42 0.48 1.29

bm 1.07 1.25 1.19 1.19 1.14 1.18 1.30 1.50

± ± ± ± ± ± ± ±

0.11 0.03 0.05 0.04 0.06 0.07 0.08 0.24

R2

MWSE

(E0-Em)/E0

0.999 0.999 0.999 0.999 0.999 0.999 0.999 0.998

0.003 0.005 0.005 0.006 0.005 0.011 0.005 0.007

57.6% 42.3% 43.4% 39.5% 28.8% 32.9% 41.5% 20.9%

case, (Eo − Em) should represent the contribution of nonspecific interactions. As shown in Table 4, the contribution of nonspecific interactions (Eo − Em)/Eo ranges from 26 to 80% for XAD-4 and from 21 to 58% for XAD-7. It also decreases with increasing A (SI Figure S8), indicating a decreasing contribution of nonspecific interactions to the adsorption of solutes with increasing H-bond acidity. logKHW is believed to represent nonspecific interactions.26 When we attempted to obtain the relationship between (Eo − Em) and logKHW, however, the correlation was poor (R2 = 0.39 for XAD4, R2 = 0.05 for XAD-7; SI Figure S9). This finding is not surprising given that the “Eo” of the original D−A model covers the molecular interactions to various extents for different compounds. Therefore, the above obtained contributions of nonspecific interactions are only a rough estimate. Further modification of the D−A model is thus necessary to quantify not only the contributions of specific interactions, but also those of nonspecific interactions. As shown in Table 4, the solutes with larger molecules have large values of “bm”. Dobruskin45 concluded that parameter “b” of the D−A equation is only determined by the standard deviation of the micropore sizes, that is, the more homogeneous the micropores are, the larger the value of “b” will be. The pore size distribution of the available adsorption sites is more homogeneous for large molecules, since some small-sized micropores are inaccessible for them. For small molecules, especially for liquid-state sorbates, the range of the micropore sizes available for adsorption are much wider, consequently, the standard deviation of micropore sizes is larger and thus the value of “bm” is smaller. By comparison, the dependence of bm on the sizes of molecules is much weaker for XAD-7 since it has a much smaller fraction of micropores (Table 1). Environmental Implications. Understanding the molecular interactions between adsorbents and solutes is vital to choosing an appropriate adsorbent for a given organic contaminant. This is particularly the case for polymeric adsorbents that can be tailor synthesized by proper adjustment of the synthetic routes toward selective removal of a specific organic compound. Subsequently, based on the predominant molecular interaction(s), an optimal regenerant can be employed to recover valuable compounds and regenerate the exhausted adsorbent. The modified D−A model proposed in this study enables us to identify the predominant specific interaction force(s) based on the normalizing factor “Em”, whereas the nonspecific interactions have been incorporated in the redefined adsorption potential. This is a significant improvement over the original theory in that a common reference state is employed for all solutes allowing direct comparison among various solutes.

Figure 4. Linear relationships between the modified D−A-model fitted “Em” values and “A”, the hydrogen-bonding donor acidity.

(N = 8, p < 0.001, R2 = 0.97)

Em (KJ/mol)

(17)

These results imply that, after removing the hydrophobic effect, the hydrogen-bonding interaction is dominant in the adsorption of phenols by XAD-4 and XAD-7. Here the phenols act as the hydrogen-bonding donor and XAD-4 or XAD-7 acts as the hydrogen-bonding acceptor. XAD-7 has abundant ester and carbonyl groups that can serve as sites for hydrogenbonding (SI Figure S6). For XAD-4, the electron-rich πsystems of the styrene-divinylbenzene matrix are the hydrogenbonding accepting sites.43,44 Yang et al.17 reported a linear relationship (R2 = 0.68) between A and the fitted “Eo” value of the original D−A model for the adsorption of anilines and phenols by MWCNTs. Comparatively, the correlation coefficient (R2) of the linear relationship between A and “Eo” for XAD-4 and XAD-7 is 0.29 and 0.69, respectively (SI Figure S7). The difference in R2 values is most likely due to the different contributions of hydrophobic interactions to the overall adsorption energy. The hydrophobic driving force is greater for adsorption by XAD-4 than by XAD-7 and MWCNTs and thus the correlation is poorer for XAD-4. The newly obtained Em only represents the specific interactions. Eo in the original D−A model was proposed to represent all the interaction forces for adsorption.20 If this is the 6812

dx.doi.org/10.1021/es300951g | Environ. Sci. Technol. 2012, 46, 6806−6814

Environmental Science & Technology

Article

(8) Deosarkar, S. P.; Pangarkar, V. G. Adsorptive separation and recovery of organics from PHBA and SA plant effluents. Sep. Purif. Technol. 2004, 38 (3), 241−254. (9) Pan, B. J.; Pan, B. C.; Zhang, W. M.; Lv, L.; Zhang, Q. X.; Zheng, S. R. Development of polymeric and polymer-based hybrid adsorbents for pollutants removal from waters. Chem. Eng. J. 2009, 151 (1−3), 19−29. (10) Pan, B. J.; Zhang, W. M.; Pan, B. C.; Qiu, H.; Zhang, Q. R.; Zhang, Q. X; Zheng, S. R. Efficient removal of aromatic sulfonates from wastewater by a recyclable polymer: 2-naphthalene sulfonate as a representative pollutant. Environ. Sci. Technol. 2008, 42 (19), 7411− 7416. (11) Li, P.; Sengupta, A. K. Genesis of selectivity and reversibility for sorption of synthetic aromatic anions onto polymeric sorbents. Environ. Sci. Technol. 1998, 32 (23), 3756−3766. (12) Pan, B. J.; Pan, B. C.; Zhang, W. M.; Zhang, Q. R.; Zhang., Q. X.; Zheng, S. R. Adsorptive removal of phenol from aqueous phase by using a porous acrylic ester polymer. J. Hazard. Mater. 2008, 157 (2− 3), 293−299. (13) Hong, C. H.; Zhang, W. M.; Pan, B. C.; Lv, L.; Han, Y. Z.; Zhang, Q. X. Adsorption and desorption hysteresis of 4-nitrophenol on a hyper-cross-linked polymer resin NDA-701. J. Hazard. Mater. 2009, 168 (2−3), 1217−1222. (14) Streat, M.; Horner, D. J. Adsorption of highly soluble herbicides from water using activated carbon and hypercrosslinked polymers. Trans IChemE 2000, 78 (B5), 363−382. (15) Pan, B.; Zhang, Q.; Meng, F.; Li, X.; Zhang, X.; Zheng, J.; Zhang, W.; Pan, B.; Chen, J. Sorption enhancement of aromatic sulfonates onto an aminated hyper-cross-linked polymer. Environ. Sci. Technol. 2005, 39, 3308−3313. (16) Polanyi, M. Adsorption of gases (vapors) by a solid non-volatile adsorbent. Verh. Dtsch. Phys. Ges. 1916, 18, 55−80. (17) Yang, K.; Wu, W. H.; Jing, Q. F.; Zhu, L. Z. Aqueous adsorption of aniline, phenol, and their substitutes by multi-walled carbon nanotubes. Environ. Sci. Technol. 2008, 42 (21), 7931−7936. (18) Yang, K.; Zhu, L. Z.; Xing, B. S. Adsorption of polycyclic aromatic hydrocarbons by carbon nanomaterials. Environ. Sci. Technol. 2006, 40 (6), 1855−1861. (19) Manes, M. In Encyclopedia of Environmental Analysis and Remediation; Meyers, R. A., Eds.; John Wiley & Sons: New York, 1998; pp 26−68. (20) Crittenden, J. C.; Sanongraj, S.; Bulloch, J. L.; Hand, D. W.; Rogers, T. N.; Speth, T. F.; Ulmer, M. Correlation of aqueous-phase adsorption isotherms. Environ. Sci. Technol. 1999, 33 (17), 2926−2933. (21) Sander, M.; Pignatello, J. J. Characterization of charcoal adsorption sites for aromatic compounds: insight drawn from single solute and bi-solute competitive experiments. Environ. Sci. Technol. 2005, 39 (6), 1606−1615. (22) Dubinin, M. M. The potential theory of adsorption of gases and vapors for adsorbents with energetically nonuniform surfaces. Chem. Rev. 1960, 60 (2), 235−241. (23) Dubinin, M. M.; Astakhov, V. A. Development of concepts of volume filling of micropores in adsorption of gases and vapors by microporous adsorbents. Izv. Akad. Nauk SSSR, Ser. Khim. 1971, 1, 5− 11. (24) Borisover, M.; Graber, E. R. Classifying NOM-organic sorbate interactions using compound transfer from an inert solvent to the hydrated sorbent. Environ. Sci. Technol. 2003, 37 (24), 5657−5664. (25) Zhu, D. Q.; Kwon, S.; Pignatello, J. J. Adsorption of single-ring organic compounds to wood charcoals prepared under different thermochemical conditions. Environ. Sci. Technol. 2005, 39 (11), 3990−3998. (26) Zhu, D. Q.; Pignatello, J. J. A concentration-dependent multiterm linear free energy relationship for sorption of organic compounds to soils based on the hexadecane dilute-solution reference state. Environ. Sci. Technol. 2005, 39 (22), 8817−8828. (27) Hickey, J. P.; Passino-reader, D. R. Linear solvation energy relationships: ″Rule of thumb″ for estimation of variable values. Environ. Sci. Technol. 1991, 25 (10), 1753−1760.

Given the wide application of the Polanyi theory, we expect this modified model can be applied to adsorption of various organic compounds, particularly polar compounds whose adsorption cannot be adequately described by the original model, by a variety of porous and nonporous natural or synthetic adsorbents. The combination of the modified D−A model and pp-LFERs enables us to account for adsorption energies over a wide range of solute concentrations even when nonlinear adsorption isotherms are observed. This has a major advantage over the pp-LFERs based on equilibrium adsorption constants where only linear correlation is allowed. In this study, we found that in addition to hydrophobic interactions, H-bonding interactions are the predominant interactions in the adsorption of phenolic compounds by XAD-4 and XAD-7. Hydrophobic effects also play a more important role in the adsorption by XAD-4 than by XAD-7. Thus, XAD-4 can be employed for adsorption of more hydrophobic compounds; while XAD-7 is more appropriate for the adsorption of more hydrophilic compounds. Also, the adsorption behavior of other uninvestigated organic compounds on polymeric adsorbents can be predicted by the proposed model based on their chemical parameters including “χHD,W” and “A”. To develop a more robust predictive tool, further work is warranted to apply the model to a large number of emerging contaminants and polymeric adsorbents.

■

ASSOCIATED CONTENT

S Supporting Information *

Texts S1−S2, and nine figures is available. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Phone: (215)204-4807; fax: (215)204-4696; e-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

REFERENCES

(1) Kolpin, D. W.; Furlong, E. T.; Meyer, M. T.; Thurman, E. M.; Zaugg, S. D.; Barber, L. B.; Buxton, H. T. Pharmaceuticals, Hormones, and Other Organic Wastewater Contaminants in U.S. Streams, 1999− 2000: A National Reconnaissance. Environ. Sci. Technol. 2002, 36 (6), 1202−1211. (2) Westerhoff, P.; Yoon, Y.; Snyder, S.; Wert., E. Fate of endocrinedisruptor, pharmaceutical, and personal care product chemicals during simulated drinking water treatment processes. Environ. Sci. Technol. 2005, 39 (17), 6649−6663. (3) Adams, C.; Wang, Y.; Loftin, K.; Meyer, M. Removal of antibiotics from surface and distilled water in conventional water treatment processes. J. Environ. Eng. Div. (Am. Soc. Civ. Eng.) 2002, 128 (3), 253−260. (4) Yoon, Y.; Westerhoff, P.; Snyder, S.; Song, R.; Levine, B. A review on removal of endocrine-disrupting compounds and pharmaceuticals by drinking water treatment processes. AWWA Water Qual. Technol. Conf., Proc. 2002, 1−20. (5) Kunin, R. The use of macroreticular polymeric adsorbents for the treatment of waste effluents. Pure Appl. Chem. 1979, 46, 205−211. (6) Xu, Z. Y.; Zhang, Q. X.; Fang, H. H. P. Applications of porous resin sorbents in industrial wastewater treatment and resource recovery. Crit. Rev. Environ. Sci. Technol. 2003, 33 (4), 363−389. (7) Kennedy, D. C. Treatment of effluent from manufacture of chlorinated pesticides with a synthetic, polymeric adsorbent, Amberlite XAD-4. Environ. Sci. Technol. 1973, 7 (2), 138−141. 6813

dx.doi.org/10.1021/es300951g | Environ. Sci. Technol. 2012, 46, 6806−6814

Environmental Science & Technology

Article

(28) Nguyen, T. H.; Goss, K. U.; Ball, W. P. Polyparameter linear free energy relationships for estimating the equilibrium partition of organic compounds between water and the natural organic matter in soils and sediments. Environ. Sci. Technol. 2005, 39 (4), 913−924. (29) Goss, K. U.; Schwarzenbach, R. P. Linear free energy relationships used to evaluate equilibrium partitioning of organic compounds. Environ. Sci. Technol. 2001, 35 (1), 1−9. (30) Abraham, M. H. Scales of solute hydrogen-bondingTheir construction and application to physicochemical and biochemical processes. Chem. Soc. Rev. 1993, 22 (2), 73−83. (31) Pharma Algorithms Inc., 591 Indian Road, Toronto, ON, Canada M60 2C4. (32) Schwarzenbach, R. P.; Gschwend, P. M.; Imboden, D. M. Environmental Organic Chemistry, 2nd ed.; Wiley-Interscience: New York, 2002. (33) Kosky, P. G.; Silva, J. M.; Guggenheim, E. A. The aqueous phase in the interfacial synthesis of polycarbonates. Part 1. Ionic equilibria and experimental solubilities in the BPA-NaOH-H2O system. Ind. Eng. Chem. Res. 1991, 30 (3), 462−467. (34) Serjeant, E. P.; Dempsey, B. Ionization Constants of Organic Acids in Aqueous Solution; Pergamon: Oxford, 1979. (35) Abraham, M. H.; Chadha, H. S.; Whiting, G. S.; Mitchell, R. C. Hydrogen bonding. 32. An analysis of water-octanol and water-alkane partitioning and the Δlog p parameter of seiler. J. Pharm. Sci. 1994, 83 (8), 1085−1100. (36) Kamlet, M. J.; Doherty, R. M.; Abraham, M. H.; Marcus, Y.; Taft, R. W. Linear solvation energy relationships. 46. An improved equation for correlation and prediction of octanol/water partition coefficients of organic nonelectrolytes (including strong hydrogen bond donor solutes). J. Phys. Chem. 1988, 92 (18), 5244−5255. (37) Marcus, Y. Linear solvation energy relationships. Correlation and prediction of the distribution of organic solutes between water and immiscible organic solvents. J. Phys. Chem. 1991, 95 (22), 8886−8891. (38) Figueroa, R. A.; Leonard, A.; Mackay, A. A. Modeling tetracycline antibiotic sorption to clays. Environ. Sci. Technol. 2004, 38 (2), 476−483. (39) Yang, K.; Xing, B. S. Adsorption of organic compounds by carbon nanomaterials in aqueous phase: polanyi theory and its application. Chem. Rev. 2010, 110 (10), 5989−6008. (40) Daignault, S. A.; Noot, D. K.; Williams, D. T.; Huck, P. M. A review of the use of XAD resins to concentrate organic compounds in water. Water Res. 1988, 22 (7), 803−813. (41) Yang, K.; Wang, X. L.; Zhu, L. Z.; Xing, B. S. Competitive sorption of pyrene, phenanthrene, and naphthalene on multiwalled carbon nanotubes. Environ. Sci. Technol. 2006, 40 (18), 5804−5810. (42) Nguyen, T. H.; Cho, H. H.; Poster, D. L.; Ball, W. P. Evidence for a pore-filling mechanism in the adsorption of aromatic hydrocarbons to a natural wood char. Environ. Sci. Technol. 2007, 41 (4), 1212−1217. (43) Keiluweit, M.; Kleber, M. Molecular-level interactions in soils and sediments: The role of aromatic π-systems. Environ. Sci. Technol. 2009, 43 (10), 3421−3429. (44) Ringwald, S. C.; Pemberton, J. E. Adsorption interactions of aromatics and heteroaromatics with hydrated and dehydrated silica surfaces by Raman and FTIR spectroscopies. Environ. Sci. Technol. 2000, 34 (2), 259−265. (45) Dobruskin, V. K. Micropore volume filling. A condensation approximation approach as a foundation to the Dubinin-Astakhov equation. Langmuir 1998, 14 (14), 3840−3846.

6814

dx.doi.org/10.1021/es300951g | Environ. Sci. Technol. 2012, 46, 6806−6814