Environ. Sci. Technol. 2010, 44, 3388–3394
Extended Triple Layer Modeling of Arsenate and Phosphate Adsorption on a Goethite-based Granular Porous Adsorbent M A S A K A Z U K A N E M A T S U , * ,† THOMAS M. YOUNG,† KEISUKE FUKUSHI,‡ PETER G. GREEN,† AND JEANNIE L. DARBY† Department of Civil and Environmental Engineering, University of California, Davis, 1 Shields Avenue, Davis, California, 95616, and Institute of Nature and Environmental Technology, Kanazawa University, Kakuma, Kanazawa, Ishikawa 920-1192, Japan
Received December 2, 2009. Revised manuscript received February 22, 2010. Accepted February 27, 2010.
The extended triple layer model (ETLM), which is consistent with spectroscopic and theoretical molecular evidence, is first systematicallytestedforitscapabilitytomodeladsorptionofarsenate and phosphate, a strong competitor, on a common goethitebased granular porous adsorptive media (Bayoxide E33 (E33)) in water treatment systems under a wide range of solution conditions.Deprotonatedbidentate-binuclear,protonatedbidentatebinuclear, and deprotonated monodentate complexes are chosen as surface species for both arsenate and phosphate. The estimated values of the ETLM parameters of arsenate for the adsorbent are close to those for pure goethite minerals previously determined by others. The ETLM predictions for arsenate and phosphate adsorption basically agree with experimental results over a wide range of pH, surface coverage, and solid concentrations. High background electrolyte concentration (i.e., I ) 0.1 M), however, was found to strongly impact arsenate and phosphate adsorption on E33 probably because of the porous structure of the adsorbent, which cannot be observed for pure goethite minerals and could not be completely modeled by the ETLM. Prediction of phosphate adsorption isotherms at higher pH were relatively poor, and this may suggest searching for alternative surface species for phosphate. Since adsorption equilibrium constants of major coexisting ions encountered in water treatment systems for goethite minerals have been estimated by others, the application of ETLM theory to this common goethite-based adsorptive media will enable us to understand how those coexisting ions macroscopically and thermodynamically interact with arsenate and phosphate in the environment of adsorptive water treatment system in a way consistent with molecular and spectroscopic evidence.
Introduction Naturally occurring arsenic in groundwater has been observed worldwide and has caused human health problems such as various types of cancer (1, 2). Arsenate (As(V)) is the * Corresponding author e-mail:
[email protected]. † University of California Davis. ‡ Kanazawa University. 3388
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 9, 2010
predominant form of arsenic in oxidized aquatic systems (3), and arsenite (As(III)) can be oxidized to arsenate by disinfectants in water treatment systems (4). The compliance for the maximum contamination level (MCL) for arsenic in drinking water (10 µg/L as As) has been required since 2006 (5), and nanostructured granular porous metal oxide adsorbents have been developed and recognized as a suitable technology to remove arsenic in small water systems (6, 7). These adsorbents have high affinities for arsenic and are packed in continuous-flow fixed bed adsorbers to remove arsenic. To understand performance and capacity of the adsorbers, particularly in the presence of competing solutes, and to predict arsenic breakthrough curves, pilot tests and rapid small scale column tests (RSSCTs) typically have been conducted (6–9). The RSSCT approach was initially developed to predict breakthrough curves of organic compounds from fixed-bed granular activated carbon (GAC) treatment systems (10). Although RSSCTs are now widely employed, the experiments are still difficult and time-consuming to conduct. Moreover, pilot testing and RSSCTs are site-specific tests, and the results will change if water quality and system configuration are changed. Breakthrough times have proven very sensitive to the presence of various constituents including silica, phosphate, etc. Thus it is desirable to develop a method to predict or estimate arsenic breakthrough curves from fixed-bed adsorbers packed with granular porous metal oxide materials in the presence of multiple competing solutes. Surface complexation models (SCMs) have been developed in the last few decades to macroscopically quantify ion adsorption onto oxides under various water quality conditions and have been recognized as useful tools (11–14). Arsenate and phosphate adsorption onto metal oxides has been extensively simulated using SCMs in previous research (8, 15–23). It has been recognized that phosphate is one of the strongest competing ions against arsenate and that its adsorption behavior is quite similar to that of arsenate (22, 24). Phosphate removal is also of interest because of its role in eutrophication, and previous research has focused on adsorption processes for phosphate removal (25). Recently, SCMs have been applied to predict adsorption equilibrium of arsenate and phosphate on commercial metal oxide adsorbents (26), and this study has implied that SCMs can be used to predict adsorption of arsenate and phosphate. However, adsorption data sets are not sufficient to conclusively demonstrate applicability of SCMs as a tool to predict arsenate adsorption on the adsorbents, and the model used in these studies does not consider recent modifications such as water dipole modification (27). One of the candidate models for predicting arsenate breakthrough curves is the pore surface diffusion model (PSDM), which was initially developed to predict breakthrough curves of organic compounds from GAC fixed-bed treatment systems (28). In the PSDM, accurate adsorption isotherms under conditions encountered in water treatment systems are required to predict adsorption breakthrough curves (29). In this research, extensive adsorption tests for arsenate and phosphate on a commercially important nanostructured granular porous ferric oxide adsorbent were conducted in single-solute systems, and the applicability of SCMs as an engineering tool to predict adsorption equilibrium of arsenate and phosphate on the adsorbent was tested. Effects of pH, ionic strength, surface coverage, and solid concentration on adsorption equilibrium of arsenate and phosphate for the adsorbent in single-solute systems are all investigated and discussed. The extended triple layer model (ETLM) (13) was 10.1021/es903658h
2010 American Chemical Society
Published on Web 03/31/2010
chosen for this study not only because it has been recognized as one of the SCMs consistent with spectroscopic and theoretical molecular evidence but that Previous ETLM studies can be helpful guide and constraints to determine surface parameters and adsorption equilibrium constants.
Materials and Methods Chemicals. All chemicals were reagent grade and used without further purification. The solutions used were prepared by diluting the stock solution to desired concentrations with deionized (DI) water. Arsenate and phosphate solutions were prepared from reagent grade Na2HAsO4 · 7H2O (Sigma Aldrich) and NaH2PO4 · H2O (Sigma Aldrich), respectively. All glassware and plastic were cleaned by soaking in 0.1 M HNO3, followed by three rinses with DI water. Adsorbent. Bayoxide E33 (E33, AdEdge Technology Inc., Norcross, GA) is a dry granular porous adsorbent that contains more than 90% goethite (R-FeOOH) (30) and is frequently employed to remove arsenic in small water systems (30). Nanoscale goethite particle are aggregated and comprise the adsorbent. The adsorbent was crushed and wet-sieved (200 × 325 U.S. standard mesh fraction, 44-75 µm) and washed with DI water to remove fine particles and impurities. The washed E33 was dried in an oven at low temperature (i.e., 40 °C) for two days to avoid any phase modification or transformation (31), and then stored in a desiccator. Dried E33 (200 × 325 U.S. standard mesh fraction) was suspended in DI water for 2 days prior to all experiments. Potentiometric Titration. Potentiometic titrations were conducted following the procedure described in Hayes et al. (32) with some modifications to determine the ETLM surface parameters such as the surface protonation and electrolyte adsorption equilibrium constants, and inner-layer capacitance (C1) (Supporting Information (SI) Figure S1). Outerlayer capacitance (C2) was determined to be 20 µF/cm2. The E33 suspension (10.0 g/L) was maintained at 20 °C and continuously stirred by an overhead mixer to prevent settling and crushing of the adsorbent particles. N2 gas was bubbled in the solution to prevent carbonate contamination throughout the titration. The ETLM parameter estimation procedure is described in detail in the SI. Media Characterization. A scanning electron microscope (XL30 FEG ESEM, FEI/Philips, Hillsboro, OR) was used to obtain images of the adsorbent, and X-ray diffraction (XDS2000, Scintag, Sunnyvale, CA) was used to analyze its mineralogical composition. The BET surface area of the adsorbent was also measured for two different particle size fractions (mesh size: 100 × 140, 200 × 325) using the N2 adsorption approach (AUTOSORB-1, Quantachrome Instruments, Boynton Beach, FL). Adsorption Edge. The experiments were performed at two solid concentrations (1.0 g/L, 0.025 g/L) to investigate whether the ETLM can successfully describe adsorption equilibrium data at very different solid concentrations. In the adsorption edge experiments at a solid concentration of 1.0 g/L, triplicates of 40 mL solutions having a wide range of pH (3-11), solute concentrations, and background NaNO3 electrolyte (0.001, 0.01, and 0.1 M) were prepared in 50 mL plastic bottles. At a solid concentration of 0.025 g/L, duplicates of 800 mL solutions having wide range of pH (3-11), solute concentrations and background NaNO3 electrolyte concentrations (0.01, 0.1 M) were prepared in 1000 mL plastic bottles. All bottles were continuously tumbled for 7 days (SI Figure S3) at 20 °C. Equilibration times for the experiments were determined during preliminary rate experiments. After the desired mixing time, the supernatant was filtered through 0.45 µm filters (Millipore Co., Bedford, MA) and analyzed for arsenate or phosphate. The amount of solute adsorbed to the adsorbent was calculated using the difference between
initial and final dissolved solute concentrations. The final pH was determined at the end of each experiment. Adsorption Isotherm. Adsorption isotherms for arsenate and phosphate in single solute systems were also performed at a solid concentration of 1.0 g/L. Triplicates of 40 mL solutions having a fixed amount of solid (40 mg) and different solute concentrations were prepared in 50 mL plastic bottles. The experiments were conducted at pH 4.0, 7.0, and 10.0 and ionic strengths of 0.01, 0.02, and 0.1 M. The solution pH was checked every 12 h and adjusted using small volumes of HNO3 or NaOH during the experiments to the target pH values (0.1 without any buffers. Ionic strength was controlled using NaNO3. All bottles were tumbled for 7 days based on preliminary adsorption kinetics experiments (SI Figure S3). Analytical Methods. The pH was measured with a glass electrode and a pH meter (S20 Seven Easy pH, Mettler Toledo, Columbus, OH). Arsenic was analyzed using inductively coupled plasma-mass spectrometry (7500i, Agilent Technologies, Wilmington, DE), and phosphate was analyzed by the colorimetric method using a flow injection analyzer (FIA) (Lachat Instruments, Milwaukee, WI). Modeling. All ETLM calculations were conducted using the computer code GEOSURF (33). Aqueous ionic activity coefficients of dissolved species were calculated using the extended Debye-Hu ¨ ckel equation. Surface species of arsenate determined by Fukushi and Sverjensky (19) were employed in the ETLM calculations. For phosphate adsorption, identical surface complexation reactions to arsenate were used following an extensive review of surface species (19), although there is ongoing discussion regarding the appropriate phosphate surface species (34, 35). Arsenate and phosphate surface species used are shown in eqs 1-3, and the adsorption equilibrium constants referring to the site-occupancy standard equilibrium constants can be written in eqs 4-6. deprotonated bidentate-binuclear complex: + 2 > SOH + H3XO04 ) (>SO)2XO2 + H + 2H2O
(1)
protonated bidentate-binuclear complex: 2 > SOH + H3XO04 ) (>SO)2XOOH + 2H2O
(2)
deprotonated monodentate complex: + > SOH + H3XO04 ) >SOXO23 + 2H + H2O
(3)
θ *K(>SO) 2XO2
)
2 a(>SO)2XO2-aH+aH 2O 2 a>SOH aH3XO40
θ ) *K(>SO) 2XOOH
θ *K>SOXO 2- ) 3
2 a(>SO)2XOOHaH 2O 2 a>SOH aH3XO40
2 a>SOXO 32-aH + aH O 2
a>SOHaH3XO40
10F(∆Ψr,1)/2.303RT
(4)
10F(∆Ψr,2)/2.303RT
(5)
10F(∆Ψr,3)/2.303RT
(6)
Here X stands for As(V) or P. Here *Kiθ represents the adsorption equilibrium constant for the formation of surface species i, and the superscript “θ” represents the siteoccupancy standard states (36, 37). The adsorption equilibrium constants are represented relative to the surface species >SOH, and this is indicated by the superscript “*”. Equations 4-6 adopt the approach taken successfully in ref 19 of approximating the activity of available bidentate surface sites by the square of the activity of available monodentate sites. Although there is significant discussion on this point in the literature, it provides a reasonably robust way of approximating this quantity in the absence of more specific mechanistic information that could help refine the assumpVOL. 44, NO. 9, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3389
tion (38). The exponential terms in eqs 4-6 correct for activity differences of ions near surface potential field (∆Ψr), where F, R, and T represent Faraday’s constant (96 485 (C/m2)), gas constant (8.314 (J/(mol · K))) and absolute temperature [K], respectively. These surface complexes are formed by ligand exchange reactions, and the reactions release one or two water dipoles (13). The water dipole(s) leaving the charged surface contribute electrical work equal to - n(Ψ0 - Ψβ), where n, Ψ0, and Ψβ stands for the number of desorbed waters per reaction, the surface potential at the 0-plane, and the surface potential at the β plane, respectively. In the ETLM, this effect is taken into account in the electrostatic term for the reaction (∆Ψr), and this surface potential field was applied in this study. ∆ψr,1 ) 2ψ0 - 3ψβ - 2(ψ0 - ψβ) ) -ψβ
(7)
∆ψr,2 ) 2ψ0 - 2ψβ - 2(ψ0 - ψβ) ) 0
(8)
∆ψr,3 ) ψ0 - 3ψβ - (ψ0 - ψβ) ) -2ψβ
(9)
The relationships of the site-occupancy standard states to the hypothetical 1.0 M standard state for the arsenate surface species are given by Fukushi and Sverjensky (19) as follows: θ 0 log* K(>SO) - ) log* K(>SO) XO- + log 2XO2 2 2
(
(NSAS)2 N‡A‡
θ 0 log* K(>SO) ) log* K(>SO) + log 2XOOH 2XOOH
θ 0 log* K>SOXO 2- ) log* K>SOXO2- + log 3 3
(
(NSAS)2 N‡A‡
( ) NSAS N‡A‡
)
CS
(10)
)
CS
(11) (12)
Where log*Ki0 are the arsenate adsorption equilibrium constants referring to the hypothetical 1.0 M standard states; Ns is the surface site density on the solid adsorbent (sites/ m2); N‡ is the standard state adsorbate species site density (sites/m2); As is the BET surface area of the solid adsorbent (m2/g); A‡ is a standard state BET surface area (m2/g); Cs is the solid concentration (g/L). In this study, values of N‡ and A‡ are selected as 10 × 1018 (sites/m2) and 10 m2/g, respectively. The adsorption equilibrium constants referring to the hypothetical 1.0 M standard state for the deprotonated bidentate-binuclear and protonated bidentate-binuclear complexes change depending on the solid concentration (36).
Results and Discussion Potentiometric Titration. The potentiometric titration results for E33 are shown in SI Figure S2. An intersection point of titration curves at different background electrolyte (i.e., point of zero salt effect, PZSE) was pH 8.5. Under this condition, this intersection point corresponds to a point of zero charge (PZC). The solid lines shown in SI Figure S1 are the ETLM calculation results. The major ETLM parameters such as the surface protonation and electrolyte adsorption equilibrium constants determined from the titration results are listed in Table 1, and a complete list of the ETLM parameters is shown and compared with the ETLM parameters for goethite determined by Fukushi and Sverjensky (19) in SI Table S2. Adsorbent Characterization. The SEM images and the XRD spectra of the adsorbent are shown in SI Figures S3 and S4, respectively. Physical properties such as the BET surface area and the pore volume are listed in SI Table S2. The BET surface area of E33 is 158.1 m2/g. The BET surface area of this adsorbent has been reported to be from 120 to 200 m2/g by the manufacturer (30). Naeem et al. (39) and Trotz (40) 3390
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 9, 2010
reported that it is 128 and 151.31, respectively. Our result is close to those values. This surface area is larger than pure goethite particles. Although detailed manufacturing processes of this adsorbent are unknown, this large surface area is due to very fine goethite particles that comprise this adsorbent. This adsorbent does not fit into the empirical correlation between surface area (As) and site density referred in the other study (27, 41, 42). Further work is necessary to explain these observations. Adsorption Edge. Arsenate adsorption edge data at the solid concentration of 1.0 g/L are shown in Figure 1a-c. The adsorption data depicted in Figure 1a refers to arsenate adsorption edges over a range of surface coverage. As expected, arsenate adsorption was dependent on pH. The solid curves in Figure 1a represent the ETLM fitting results using the three surface complexes (1)-(3). The ETLM is capable of fitting the experimental results well over the entire range of surface coverage investigated. The estimated adsorption equilibrium constants referring to the siteoccupancy standard states for the reactions (Kiθ) are listed in Table 1. These adsorption equilibrium constants are compared with those of goethite determined in the previous study in SI Table S3. The adsorption data depicted in Figure 1b refers to arsenate adsorption edges at the solid concentration of 1.0 g/L at three different ionic strength conditions (0.001, 0.01, and 0.1 M). The solid lines depicted in Figure 1b are the ETLM predictions corrected for the appropriate values of background electrolyte concentrations. Interestingly the arsenate adsorption edges for E33 are more affected by ionic strength than those for pure goethite minerals especially at higher pH (above the pHZPC) (17). It has been recognized that adsorption of ion that can form inner-sphere complexes increases with increasing electrolyte concentration (43). This effect is usually attributed to changes in the electric potential in the interface, which decreases the electrostatic repulsion between the charged surface and the anion. Although the ETLM was able to predict enhanced adsorption of arsenate at higher ionic strength, the prediction could not completely capture the strong effect of ionic strength observed for E33. One of the possible reasons for this effect is the porous structure of the adsorbent. A thinner boundary layer at higher ionic strength might help ions to diffuse deeply inside the pores of the adsorbent. Further investigation would be required to investigate the cause of this effect. The predicted surface and solution speciation of arsenate as a function of pH at total arsenate concentration of 301 µM is shown in Figure 1c. For this condition, the deprotonated binuclear bidentate arsenate ((>FeO)2AsO2-) and the deprotonated monodentate arsenate (>FeOAsO32-) are predicted to dominate at lower pH and higher pH values, respectively (19). 0 - and log*K 0 According to eqs 7-9, log*K (>SO) (>SO)2AsOOH 2AsO2 are dependent on the solid concentration (Cs). The adsorption equilibrium constants referring to the hypothetical 1.0 M standard states at the lower solid concentration (0.025 g/L) were calculated using the adsorption equilibrium constants referring to the site-occupancy standard state estimated above. The adsorption data depicted in Figure 1d and e refer to arsenate adsorption edges at the solid concentration of 0.025 g/L over a range of surface coverage and ionic strength values, respectively. The ETLM parameters of 0.025 g/L estimated using the parameters in the different systems (i.e., solid concentration of 1.0 g/L) could predict the experimental results over the range of surface coverage. The relatively poor prediction at lower initial concentration and at higher pH is probably due to carbonate contamination caused by incomplete exclusion of atmospheric CO2 at high pH (15). Phosphate adsorption edge data at solid concentrations of 1.0 and 0.025 g/L are shown in Figure 2 together with the ETLM predictions. The adsorption data depicted in Figure
TABLE 1. Extended Triple Layer Model (ETLM) Parameters for Proton, Electrolyte, And Arsenate and Phosphate Adsorption on Bayoxide E33a name
reaction
log K
Hypothetical 1.0 M Standard State log
K 10
>SOH + H+ ) >SOH2+
4.9
>SO- + H+ ) >SOH
-12.1
0 + log *K Na
>SOH + Na+ ) >SO-_Na+ + H+
-8.7
0 log *K N0 3
>SOH + H+ + NO3- ) SOH2+_NO3-
log K 20
8.0
Site-Occupancy Standard State 2 > SOH + H3AsO40 ) (>SO)2AsO2- + H+ + 2H2O
2.9
θ log K (>so) 2 AsOOH
2 > SOH + H3AsO40 ) (>SO)2AsOOH + 2H2O
2.4
θ 3log K >SOAsO 4
>SOH + H3AsO40 ) >SOAsO32- + 2H+ + H2O
-0.2
θ log K (>SO) 2PO2
2 > SOH + H3PO40 ) (>SO)2PO2- + H+ + 2H2O
2.3
2 > SOH + H3PO40 ) (>SO)2POOH + 2H2O
1.6
>SOH + H3PO40 ) >SOPO32- + 2H++ H2O
-0.1
log
θ K (>so) 2 AsO2
θ log K (>SO) 2POOH
θ log K >SOPO 3
a The protonation constants for arsenate and phosphate were obtained from ref (2). The properties of E33 are Ns ) 4.0 sites/nm2, As ) 158.1 m2/g (mesh 200 × 325), C1 ) 100 µF/cm2, C1 ) 20 µF/cm2, and pHzpc ) 8.5. The value of ∆p K θn is assumed to be the same as for goethite ()5.6) given by Fukushi and Sverjensky (19). The values of log K 1θ and log K 2θ are θ+ and log K Lθ- were determined by the optimization procedure as 3.6 and 5.7 and 11.3, respectively. The values of log K M 3.5, respectively. The complete list of the ETLM parameters and the relationships among those parameters are presented in SI Table S2 and S3.
2a shows phosphate adsorption edge results over a range of surface coverage. Phosphate adsorption onto the adsorbent was also dependent on pH, as expected. The solid curves in Figure 2a represent the ETLM fitting results using the three equations. 1 - 3. To the best of our knowledge, phosphate adsorption on mineral surfaces has never been analyzed by the ETLM. Although phosphate adsorption could not be predicted as accurately as arsenate (Figure 2b), the ETLM using three surface species analogous to those for arsenate produced a good prediction. The deprotonated binuclear bidentate phosphate ((>FeO)2PO2-) and the deprotonated monodentate phoshate (>FeOPO32-) are predicted to dominate at lower pH and higher pH values, respectively, and the contribution of the protonated bidentate-binuclear phosphate was found to be very small in this study (Figure 2c, f). The relatively poor prediction at lower phosphate loading and at higher pH was also observed for phosphate, and this is probably due to carbonate contamination (23). Comparison of the adsorption equilibrium constants referring to hypothetical 1.0 M standard states or site occupancy standard states can be indicators for competitive
adsorption. At the pH commonly encountered in groundwater treatment systems, the deprotonated binuclear bidentate and the deprotonated monodentate arsenate species are predominant. The adsorption equilibrium constant of deprotonated bidentate-binuclear complex of arsenate on E33 are slightly larger than that of phosphate (Table 1). This indicates arsenate binds on E33 more strongly than phosphate when the same surface species are used in the ETLM. Weaker binding of phosphate compared to arsenate was observed in the adsorption isotherm tests as shown below, and the smaller value of the adsorption equilibrium constants can explain the observation. Adsorption Isotherm. Arsenate and phosphate adsorption isotherms for E33 are compared with those for goethite determined by Antelo et al. (20) at an ionic strength of 0.01 M in Figure 3a and b, respectively. Adsorption capacity of E33, which is manufactured from goethite particles, is similar to that of goethite. The arsenate and phosphate adsorption isotherms are shown together with the corresponding ETLM predictions in Figure 4a and b, respectively. The ETLM parameters estimated from the adsorption edge data above VOL. 44, NO. 9, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3391
FIGURE 1. (a) Arsenate adsorption edge results at the solid concentration of 1.0 g/L and at three different surface coverages. The experimental results shown in Figure 1a were used to estimate the ETLM parameters shown in Table 1. The curves in b-f were calculated using the estimated ETLM parameters. (b) Arsenate adsorption edge results at ionic strength of 0.001, 0.01, and 0.1 M. (c) Arsenate surface and aqueous speciations at the solid concentration of 1.0 g/L at initial arsenate concentration of 0.301 mM calculated by the ETLM. (d) Arsenate adsorption edge results at the solid concentration of 0.025 g/L and at three different surface coverages. The hypothetical 1.0 M standard states at the solid concentration of 0.025 g/L were calculated using the parameters shown in Table 1 and eqs 7-9. (e) Arsenate adsorption edge results at ionic strength of 0.01 and 0.1 M. (f) Arsenate surface and aqueous speciation at solid concentration of 0.025 g/L and at initial arsenate concentration of 0.006 mM calculated by the ETLM.
FIGURE 2. (a) Phosphate adsorption edge results at the solid concentration of 1.0 g/L and at three different surface coverages. The experimental results shown in Figure 2a were used to estimate the ETLM parameters shown in Table 1. The curves in b-f were calculated using the estimated ETLM parameters. (b) Phosphate adsorption edge results at ionic strength of 0.001, 0.01, and 0.1 M. (c) Phosphate surface and aqueous speciations at the solid concentration of 1.0 g/L at initial phosphate concentration of 0.323 mM calculated by the ETLM. (d) Phosphate adsorption edge results at the solid concentration of 0.025 g/L and at three different surface coverages. The hypothetical 1.0 M standard states at the solid concentration of 0.025 g/L were calculated using the parameters shown in Table 1 and eqs 7-9. (e) Phosphate adsorption edge results at ionic strength of 0.01 and 0.1 M. (f) Phosphate surface and aqueous speciation at solid concentration of 0.025 g/L and at initial phosphate concentration of 0.006 mM calculated by the ETLM. 3392
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 9, 2010
FIGURE 3. Comparison of arsenate (a) and phosphate (b) adsorption isotherms for E33 determined in this study with those for goethite determined by Antelo et al at ionic strength of 0.01 M (20). The solid lines are the ETLM prediction of adsorption isotherms for E33.
FIGURE 4. Arsenate (a) and phosphate (b) adsorption isotherms on E33 at solid concentration of 1 g/L shown together with the ETLM predictions in log-log plot. The black, red, and blue lines indicate the adsorption isotherms at pH 4, 7, and 10, respectively, The thick solid, thin solid, and dashed lines indicate the adsorption isotherms at ionic strength of 0.01, 0.02, and 0.1 M, respectively. The ETLM parameters used were shown in Table 1 and SI S2, and S3. (Table 1, and SI S2 and S3) were used to predict adsorption isotherms in those figures. Wide ranges of pH and ionic strength were tested here. Adsorption isotherms are displayed together with the Freundlich fits in SI Figure S6. Although it can be seen in Figure 4a that the ionic strength effect on arsenate adsorption isotherms was not completely captured by the ETLM as stated above, the ETLM is capable of predicting arsenate and phosphate adsorption isotherms over a wide range of concentrations. The ETLM prediction is accurate at ionic strength values typically encountered in water treatment operations (i.e., 0.01-0.02 M). However, relatively poor prediction of phosphate adsorption isotherms at a pH of 10 was observed. Although the surface species of phosphate were extensively reviewed in a previous study (19), this result may suggest the need for further search to identify likely surface complexes for phosphate. This is the first study to systematically test the ETLM for its capability to predict adsorption of arsenate and phosphate on the commercially important goethite-based granular porous adsorbent, E33. Effects of pH, surface coverage, ionic strength, and solid concentration on adsorption equilibria of arsenate and phosphate were extensively investigated for this adsorbent. The ETLM could predict those effects over a wide range of conditions although the effects of ionic strength on arsenate and phosphate adsorption equilibria were not completely captured. Unlike the previous study (26), the adsorption data sets presented here are sufficient to conclusively demonstrate applicability of SCMs as a tool to predict adsorption of arsenate and phosphate on the commercial adsorbent. The ETLM was developed to be consistent with spectroscopic and molecular evidence, and this model has the flexibility to be applied to any mineral surface having various surface properties. Adsorption equilibrium constants of major coexisting ions for goethite and other mineral surfaces have
been estimated by others (42, 44), and adsorption data of those ions for goethite have been obtained by others (42, 45, 46). Using the ETLM theory, it is possible to estimate adsorption equilibrium constants of all those ions for E33 from the references despite the difference of surface properties between those goethite minerals and E33. The application of ETLM theory to this adsorbent will enable us to understand how those ions macroscopically and thermodynamically interact with each other on the common adsorbent in water treatment systems in a way consistent with spectroscopic and molecular evidence.
Acknowledgments This research was supported under Contract No. 06-55254 from the California Department of Public Health Safe Drinking Water Revolving Fund. The content is solely the responsibility of the authors and does not necessarily represent the official views of the organizations above. Analytical assistance and advice by Fred Hayes and Shweta Rao are also gratefully acknowledged.
Supporting Information Available Details on the following: (1) Experimental setup for the potentiometric titration (Figure S1); (2) Experimental results of the potentiometric titration (Figure S2); (3) Preliminary adsorption kinetics experimental results for arsenate and phosphate (Figure S3); (4) Adsorbent characterizations using the SEM (Figure S4) and the XRD (Figure S5); (4) The physical properties of E33 (Table S1); (5) The detailed ETLM parameter estimation procedure; (6) The complete list of the ETLM parameters and comparison of the parameters with those of goethite determined by others (Tables S2 and S3); (7) The arsenate and phosphate adsorption isotherms shown together with the Freundlich model (Figures S6 and S7); (8) The list of the Freundlich model parameters for each VOL. 44, NO. 9, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3393
adsorption isotherm (Table S4). This material is available free of charge via the Internet at http://pubs.acs.org.
Literature Cited (1) Nordstrom, D. Public health - Worldwide occurrences of arsenic in ground water. Science 2002, 296 (5576), 2143–2145. (2) Welch, A. H.; Stollenwerk, K. G. Arsenic in Ground Water; Kluwer Academic Publishers: Boston, 2003; p 475. (3) Myneni, S. C. B.; Traina, S. J.; Waychunas, G. A.; Logan, T. J. Experimental and theoretical vibrational spectroscopic evaluation of arsenate coordination in aqueous solutions, solids, and at mineral-water interfaces. Geochim. Cosmochim. Acta 1998, 62 (19-20), 3285–3300. (4) Frankenberger, W. T. Environmental Chemistry of Arsenic; Marcel Dekker: New York, 2002; Vol. xiii, p 391. (5) U.S. Environmental Protection Agency Arsenic in drinking water. http://www.epa.gov/safewater/arsenic/index.html. (6) Westerhoff, P.; De Haan, M.; Martindale, A.; Badruzzaman, M. Arsenic adsorptive media technology selection strategies. Water Qual. Res. J. Can. 2006, 41 (2), 171–184. (7) Seigel, M.; Aragon, A.; Zhao, H.; Everett, R.; Aragon, M.; Nocon, M.; Dwyer, B.; Marbury, J.; Kirby, C.; North, K. Pilot Test of Arsenic Adsorptive Media Treatment Technologies at Socorro Springs; Sandia National Laboratories: Albuquerque, NM, 2007. (8) Westerhoff, P.; Highfield, D.; Badruzzaman, M.; Yoon, Y. Rapid small-scale column tests for arsenate removal in iron oxide packed bed columns. J. Environ. Eng. 2005, 131 (2), 262–271. (9) Jang, M.; Min, S. H.; Kim, T. H.; Park, J. K. Removal of arsenite and arsenate using hydrous ferric oxide incorporated into naturally occurring porous diatomite. Environ. Sci. Technol. 2006, 40 (5), 1636–1643. (10) Crittenden, J. C.; Berrigan, J. K.; Hand, D. W.; Lykins, B. Design of rapid fixed-bed adsorption tests for nonconstant diffusivities. J. Environ. Eng. 1987, 113 (2), 243–259. (11) Hiemstra, T.; VanRiemsdijk, W. H. A surface structural approach to ion adsorption: The charge distribution (CD) model. J. Colloid Interface Sci. 1996, 179 (2), 488–508. (12) Dzombak, D. A.; Morel, F. Surface Complexation Modeling: Hydrous Ferric Oxide; Wiley: New York, 1990; Vol. xvii, p 339. (13) Sverjensky, D. A.; Fukushi, K. Anion adsorption on oxide surfaces: Inclusion of the water dipole in modeling the electrostatics of ligand exchange. Environ. Sci. Technol. 2006, 40 (1), 263–271. (14) Hayes, K.; Redden, G.; Ela, W.; Leckie, J. Surface complexation modelssAn evaluation of model parameter-estimation using FITEQL and oxide mineral titration data. J. Colloid Interface Sci. 1991, 142 (2), 448–469. (15) Stachowicz, M.; Hiemstra, T.; Van Riemsdijk, W. Arsenicbicarbonate interaction on goethite particles. Environ. Sci. Technol. 2007, 41 (16), 5620–5625. (16) Manning, B. A.; Goldberg, S. Modeling competitive adsorption of arsenate with phosphate and molybdate on oxide minerals. Soil Sci. Soc. Am. J. 1996, 60 (1), 121–131. (17) Goldberg, S.; Johnston, C. T. Mechanisms of arsenic adsorption on amorphous oxides evaluated using macroscopic measurements, vibrational spectroscopy, and surface complexation modeling. J. Colloid Interface Sci. 2001, 234 (1), 204–216. (18) Dixit, S.; Hering, J. G. Comparison of arsenic(V) and arsenic(III) sorption onto iron oxide minerals: Implications for arsenic mobility. Environ. Sci. Technol. 2003, 37 (18), 4182–4189. (19) Fukushi, K.; Sverjensky, D. A predictive model (ETLM) for arsenate adsorption and surface speciation on oxides consistent with spectroscopic and theoretical molecular evidence. Geochim. Cosmochim. Acta 2007, 71 (15), 3717–3745. (20) Antelo, J.; Avena, M.; Fiol, S.; Lopez, R.; Arce, F. Effects of pH and ionic strength on the adsorption of phosphate and arsenate at the goethite-water interface. J. Colloid Interface Sci. 2005, 285 (2), 476–486. (21) Gao, Y.; Mucci, A. Acid base reactions, phosphate and arsenate complexation, and their competitive adsorption at the surface of goethite in 0.7 M NaCl solution. Geochim. Cosmochim. Acta 2001, 65 (14), 2361–2378. (22) Manning, B.; Goldberg, S. Modeling competitive adsorption of arsenate with phosphate and molybdate on oxide minerals. Soil Sci. Soc. Am. J. 1996, 60 (1), 121–131. (23) Rahnemaie, R.; Hiemstra, T.; van Riemsdijk, W. Carbonate adsorption on goethite in competition with phosphate. J. Colloid Interface Sci. 2007, 315 (2), 415–425.
3394
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 9, 2010
(24) Stachowicz, M.; Hiemstra, T.; van Riemsdijk, W. Multi-competitive interaction of As(III) and As(V) oxyanions with Ca2+, Mg2+, PO43-, and CO32- ions on goethite. J. Colloid Interface Sci. 2008, 320 (2), 400–414. (25) Blaney, L. M.; Cinar, S.; SenGupta, A. K. Hybrid anion exchanger for trace phosphate removal from water and wastewater. Water Res. 2007, 41 (7), 1603–1613. (26) Zeng, H.; Fisher, B.; Giammar, D. E. Individual and competitive adsorption of arsenate and phosphate to a high-surface-area iron oxide-based sorbent. Environ. Sci. Technol. 2008, 42 (1), 147–152. (27) Sverjensky, D.; Fukushi, K. Anion adsorption on oxide surfaces: Inclusion of the water dipole in modeling the electrostatics of ligand exchange. Environ. Sci. Technol. 2006, 40 (1), 263–271. (28) Crittenden, J. C. Mathematical Modeling of Adsorber DynamicsSingle Components and Multi-Components; The Univeristy of Michigan: Ann Arbor, 1976. (29) Weber, W. J.; Smith, E. H. Simulation and design models for adsorption processes. Environ. Sci. Technol. 1987, 21 (11), 1040– 1050. (30) AdEdge Technologies Inc. http://www.adedgetechnologies.com/ index.html. (31) Schwertmann, U.; Cornell, R. M. Iron Oxides in the Laboratory: Preparation and Characterization, 2nd ed.; Wiley-VCH: Weinheim, New York, 2000; Vol. xviii, p188. (32) Hayes, K. F.; Redden, G.; Ela, W.; Leckie, J. O. Surface complexation models: An evaluation of model parameter estimation using FITEQL and oxide mineral titration data. J. Colloid Interface Sci. 1991, 142 (2), 448–469. (33) Sahai, N.; Sverjensky, D. A. GEOSURF: A computer program for modeling adsorption on mineral surfaces from aqueous solution. Comput. Geosci. 1998, 24 (9), 853–873. (34) Kwon, K. D.; Kubicki, J. D. Molecular orbital theory study on surface complex structures of phosphates to iron hydroxides: Calculation of vibrational frequencies and adsorption energies. Langmuir 2004, 20 (21), 9249–9254. (35) Rahnemaie, R.; Hiemstra, T.; van Riemsdijk, W. H. Geometry, charge distribution, and surface speciation of phosphate on goethite. Langmuir 2007, 23 (7), 3680–3689. (36) Sverjensky, D. A. Standard states for the activities of mineral surface sites and species. Geochim. Cosmochim. Acta 2003, 67 (1), 17–28. (37) Sverjensky, D. A. Prediction of surface charge on oxides in salt solutions: Revisions for 1: 1 (M+L-) electrolytes. Geochim. Cosmochim. Acta 2005, 69 (2), 225–257. (38) Benjamin, M. Modeling the mass-action expression for bidentate adsorption. Env. Sci. Technol. 2002, 36 (3), 307–313. (39) Naeem, A.; Westerhoff, P.; Mustafa, S. Vanadium removal by metal (hydr)oxide adsorbents. Water Res. 2007, 41 (7), 1596– 1602. (40) Trotz, M. A. The Feasibility of Removing Inorganic Arsenic from Landfill Leachate via Sorption to Mineral Oxide Surfaces; University of South Florida: Tampa, FL, 2008. (41) Sverjensky, D.; Fukushi, K. A predictive model (ETLM) for As(III) adsorption and surface speciation on oxides consistent with spectroscopic data. Geochim. Cosmochim. Acta 2006, 70 (15), 3778–3802. (42) Fukushi, K.; Sverjensky, D. A surface complexation model for sulfate and selenate on iron oxides consistent with spectroscopic and theoretical molecular evidence. Geochim. Cosmochim. Acta 2007, 71 (1), 1–24. (43) Barrow, N.; Bowden, J.; Posner, A.; Quirk, J. Describing the effects of electrolyte on adsorption of phosphate by a variable charge surface. Aust. J. Soil Res. 1980, 18 (4), 395–404. (44) Sverjensky, D. A. Prediction of the speciation of alkaline earths adsorbed on mineral surfaces in salt solutions. Geochim. Cosmochim. Acta 2006, 70 (10), 2427–2453. (45) Peacock, C. L.; Sherman, D. M. Vanadium(V) adsorption onto goethite (R-FeOOH) at pH 1.5 to 12: A surface complexation model based on ab initio molecular geometries and EXAFS spectroscopy. Geochim. Cosmochim. Acta 2004, 68 (8), 1723– 1733. (46) Hiemstra, T.; Barnett, M. O.; van Riemsdijk, W. H. Interaction of silicic acid with goethite. J. Colloid Interface Sci. 2007, 310 (1), 8–17.
ES903658H
