Environ. Sci. Technol. 2004, 38, 2428-2434
Adsorption of Arsenic from Water Using Activated Neutralized Red Mud HU ¨ L Y A G E N C¸ - F U H R M A N , * , † JENS CHRISTIAN TJELL,† AND DAVID MCCONCHIE‡ Environment & Resources, Technical University of Denmark, Bygningstorvet, Building 115, DK-2800 Kongens Lyngby, Denmark, and Centre for Coastal Management, Southern Cross University, P.O. Box 5125, East Lismore, New South Wales 2480, Australia
In this paper activated seawater-neutralized red mud, herein referred to as activated Bauxsol (AB), is used as a novel adsorbent for removing inorganic arsenic (As) from water. The adsorption of As onto AB is studied as a function of contact time, particle size, pH, initial As concentration, AB dosage, and temperature. Kinetic data indicate that the process pseudoequilibrates in 3 and 6 h for As(V) (arsenate) and As(III) (arsenite), respectively, and follows a pseudo-first-order rate expression. Within the range tested, the optimal pH for As(V) adsorption is 4.5, and close to 100% removal can be achieved irrespective of the initial As(V) concentration. Desorption of As(V) is greatest at pH 11.6 where a maximum of 40% can be achieved. In contrast, the optimum pH for As(III) removal is 8.5, and the removal efficiency changes with the initial As(III) concentration. The adsorption data fit the Langmuir isotherm and its linearized form well, with thermodynamic data indicating the spontaneous and endothermic nature of the process. The FITEQL (V.4) and PHREEQC (V.2) computer programs are used to predict As(V) adsorption at various pH values (based on diffuse double layer models). The modeling results fit the experimental results very well and indicate that surface complexation modeling is useful in describing the complex AB surface during the adsorption process. This study shows that As(III) needs to be oxidized to As(V) for a favorable removal using AB and that AB can be a very efficient unconventional adsorbent for removing As(V) from water.
Introduction Arsenic (As) is both ubiquitous in the environment and potentially toxic to humans. Drinking water with high As concentrations is of particular concern, because studies of chronic As exposure have shown that even small amounts of As in drinking water can cause cancer if ingested over a long period (1, 2). Consequently, the World Health Organization (WHO) has set a provisional guideline limit of 10 µg L-1 for As in drinking water (3). Arsenic can occur in water in organic or inorganic forms, but the inorganic form is more common. Inorganic As may exist in -3, +3, and +5 oxidation * Corresponding author phone: +45 45 25 16 00; fax: +45 45 93 28 50; e-mail:
[email protected]. † Technical University of Denmark. ‡ Southern Cross University. 2428
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 8, 2004
states with As(III) (arsenite) and As(V) (arsenate) being the dominant species in natural waters. The oxidation state of As in water depends primarily on the pH and redox conditions. Soil and groundwater with a low pH and redox potential will favor the formation of As(III) species, but As(V) is the thermodynamically stable form that generally predominates in oxic surface waters. Arsenite is more difficult to remove from water at near-neutral pH values, and in water treatment facilities it is usually oxidized (e.g. using hypochlorite) to As(V) before removal by coprecipitation or adsorption (4). However, there is a risk that chlorine-based oxidants can form potentially hazardous organochlorine compounds (5). Recent studies suggest that As(III) oxidation by oxygen in the presence of iron can be increased by several orders of magnitude by near-UV light (6), and there may be good potential for combining this approach with the use of AB as discussed in this paper. Arsenic removal in large scale water treatment plants usually involves coagulation with Fe or Al salts (5, 7), but processes based on adsorption and coprecipitation methods are promising because they can be used in small scale treatment plants and household systems, are easy to operate, may provide largely sludge-free operation, and may have a regeneration capability (8). A wide range of possible adsorbents is available for As removal including goethite and gibbsite (9, 10), ferrihydrite and hydrous ferric oxides (1113), hematite (14), activated red mud (15), and Bauxsol and activated Bauxsol (AB) (16-18). Caustic red mud is a byproduct of alumina production and red mud that has been neutralized by treatment with seawater or other Mg- and Ca-rich brines forms the basis of Bauxsol technology; neutralization using Ca and Mg salts converts soluble caustics i.e., hydroxides, and hydroxide carbonate into solid alkalinity (19, 20). Although, the term Bauxsol can apply to red mud that has been neutralized in different ways (19) and may be used in different applications, in this study it refers specifically to red mud that has been treated with seawater and is to be used as an unconventional adsorbent to remove As(V) from aqueous solutions. Genc¸ et al. (16) previously used Bauxsol to remove As(V) from nearly neutral water. Although high removal efficiencies were achieved, the amount of adsorbent required was high. It was therefore suggested that Bauxsol could be used for pretreatment of As or that its adsorption capacity needed to be increased. Thus, the activation of Bauxsol is advocated in refs 17 and 18 using a combined acid and heat treatment method (21), and high As(V) removal efficiencies are reported whether or not competing anions (i.e. phosphate, silicate, bicarbonate and sulfate) were present. However, a detailed investigation of As removal by AB is necessary to understand the limitations and potential value of the treatment. This study investigates the As adsorption characteristics of AB using batch adsorption experiments to understand the influence of contact time, particle size, pH, initial As concentration, adsorbent dosage, and temperature on the adsorption. Moreover, surface complexation modeling is used to help explain the adsorption process. Studies of As(V) adsorption on amorphous Al- and Fe-oxide using surface complexation models have been reported (12, 22, 23), but there has been no similar study of As adsorption onto Bauxsol or red mud. Consequently, this study has three main objectives: (i) to elucidate the As adsorption characteristics of AB with batch sorption experiments; (ii) to determine the reversibility of the process; and (iii) to evaluate the possibility of using a surface complexation model to describe As(V) adsorption onto AB. 10.1021/es035207h CCC: $27.50
2004 American Chemical Society Published on Web 03/09/2004
TABLE 1. Chemical Composition of Activated Bauxsol (AB)a constituent
% (w/w)
constituent
% (w/w)
constituent
% (w/w)
constituent
% (w/w)
Fe2O3 CaO P205
46.6 0.7 0.4
Al2O3 MgO MnO
26.5 0.5 0.1
SiO2 Na2O
17.4 0.5
TiO2 K2O
6.9 0.4
a
The specific surface area (BET-N2) of AB ) 130 m2 g-1.
Materials and Methods Activated Bauxsol (AB). The red mud used in this study was obtained from the Queensland Alumina Ltd. refinery, Gladstone, Australia. The Bauxsol is prepared by mixing the red mud with sufficient seawater to achieve a pseudoequilibrium pH for the mixture of about 8.6 ( 0.2. The AB is prepared using the combined acid and heat treatment method (21), which involves refluxing the Bauxsol in HCl, adding ammonia for complete precipitation, filtering, washing with distilled water (DIW), and calcining at 500 °C for 2 h. The chemical composition and the surface area of AB are given in Table 1, where it can be seen that the AB is primarily composed of Fe- and Al-oxides. Additional details on the activation method and characteristics of AB are given elsewhere (17). Sorption and Desorption Experiments. Duplicate batches were examined to prepare adsorption curves and to investigate the characteristics of the adsorption process. All chemicals used were reagent-grade or better and were used without additional purification. Arsenate and As(III) stock solutions were prepared by dissolving Na2HAsO4‚7H2O (Sigma) and NaAsO2 (Sigma), respectively, in water to 1 g L-1 As. Secondary stock solutions of 100 and 10 mg L-1 As were prepared weekly and used to prepare test solutions with As concentrations between 0.05 mg L-1 (0.67 µM) and 16.6 mg L-1 (220.9 µM). A matrix electrolyte of 0.01 M NaCl, prepared by dissolving reagent grade NaCl in DIW, was used for all sorption experiments. Test solutions (50 mL) having various initial As concentrations were placed in 100 mL polyethylene bottles, and samples were taken from each bottle to represent As concentrations prior to adding AB at amounts required to obtain 0.4, 1.0, 2.0, 5.0, and 10.0 g L-1 adsorbent dosages. The natural pseudoequilibrium pH of the AB (AB is the sorbent subjected to acid and heat treatment) is 7.0 ( 0.1, and most of the adsorption studies were carried out at this pH. For tests at different pH conditions 2 M HCl or NaOH solutions were added as required. For all tests, the bottles were capped tightly and shaken with a mechanical shaker at room temperature (23 ( 1 °C) until pseudoequilibrium was reached. Pseudoequilibrium contact times were found by taking subsamples from the flasks for 24 h at intervals and analyzing the supernatants for As(V) and As(III). After pseudoequilibrium was reached, the pH was measured, and the value recorded is reported as the pH of the experiment; results are reported to (0.1 pH units. After completion of the treatment, all test mixtures were centrifuged for 30 min at 4200 rpm, and the supernatants were analyzed for As using a Perkin-Elmer 5000 AAS with an online hydride generation unit. Calibration was achieved using dilutions prepared from a commercially available 1 g L-1 standard As solution. For hydride generation, 3% NaBH4 (prepared in 1% NaOH) and 1.5% HCl solutions were reacted with the samples for total As determinations, and 1% NaBH4 (prepared in 0.1% NaOH) and 0.1% HCl (to reach pH 5 to prevent As(III) oxidizing to As(V)) solutions were used for As(III) determinations. After assessing the accuracy, precision, and sensitivity under various analytical conditions, these conditions were found to give the best results for As(V) and As(III) determinations. High purity nitrogen was used as a purge gas for the hydride generation unit, and the glass reaction vessel and quartz cell were continuously purged with nitrogen to eliminate any
TABLE 2. Reactions Used in the Diffuse Double Layer Modeling and Equilibrium Constantsa intrinsic acidity constants surface hydrolysis reactions
ABb
HFOc
≡SOH + H+ ) ≡SOH2+ ≡SOH ) ≡SO- + H+
6.0 -7.8
7.29 -8.93 As(V) adsorption constants
As(V) adsorption reactions
ABd
HFOc
≡SOH + H3AsO4 ) ≡SH2AsO4 + H2O 11.09 8.67 ≡SOH + H3AsO4 ) ≡SHAsO4- + H+ + H2O 4.32 2.99 ≡SOH + H3AsO4 ) ≡SAsO42- + 2H+ + H2O -2.42 -4.7 ≡SOH + H3AsO4 ) ≡SOHAsO43- + 3H+ -10.56 -10.15 a ≡SOH is defined as 1 mol of reactive hydroxyl groups on the activated Bauxsol (AB) surface. b Reference 27. c Reference 24. dThis study.
interference from air at 194.3 nm. To determine whether the As being analyzed was present in solution or As bound to suspended particles, several samples were analyzed with and without filtration at 0.45 µm. As found in previous studies, the results showed no significant difference in As(V) concentrations. The detection limit was 2 µg L-1. Dissolved Fe was undetectable in supernatants, indicating the complete oxidation of Fe(II) and the precipitation of Fe(III) (11). In the desorption tests, As(V)-loaded (i.e. the spent) AB was leached using DIW, and the pH was adjusted to the desired value by adding 2 M HCl or NaOH (in contact with the samples) as required to allow As(V) desorption to be studied over a wide pH range. All glassware and the polyethylene vials were acidwashed for at least 12 h before each experiment using 6% HNO3, then washed with DIW four times, and dried. All samples were analyzed within 2 days of completing each test, and analysis of duplicates (or triplicates) indicated a precision better than (5%. Modeling Tools. Surface complexation models treat ion adsorption as complexation reactions and are commonly used to model the adsorption of As on a surface (9, 12, 22, 23). In this study, diffuse layer modeling (DLM) (24) is used to predict the As(V) adsorption on AB, due to the difficulty in obtaining the capacitance density (which varies with pH) when using the constant capacitance or the triple layer model. FITEQL version 4 (25) and PHREEQC version 2 (26) are used to model the experimental results. The computer program FITEQL fits equilibrium constants to experimental data using a nonlinear least squares optimization method. FITEQL was employed to calculate the As(V) adsorption constants using the relevant isotherm data, and these constants are used for all adsorption modeling unless otherwise stated. The surface reactions that are expected to be taking place at the surface are summarized in Table 2, and note that only monodendate (one-site) attachment is assumed. In this study, the modeling work is carried out with the simplifying assumption that AB is a single homogeneous surface and represented with ≡SOH (surface hydroxyl group). For comparison, the constants reported in VOL. 38, NO. 8, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2429
FIGURE 1. Lagergren plots, with reaction conditions: Ionic strength (I) ) 0.01 M NaCl, pH ) 7.0 ( 0.1, AB dosage (X) ) 5.0 g L-1, initial As concentration (C0) ) 20.37 µM for As(V) and 14.82 µM for As(III).
FIGURE 2. As(III) adsorption envelope on AB.
ref 24 for hydrated ferric oxide (HFO) are also shown in Table 2. We note that the HFO constants are used for hydrous 1 and 2 scenarios. The intrinsic acidity constants K1(int) and K2(int) are taken from the study of Gu ¨ c¸ lu ¨ and Apak (27), where red mud (the raw material of AB) was used for the adsorption.
Results Effect of Contact Time. The pseudoequilibrium time for As(V) is reported as 3 h using either Bauxsol or AB (16, 17). Here, the pseudoequilibrium time for As(III) is similarly determined by taking subsamples at different time intervals for 24 h (data not shown). It is found that the removal starts shortly after the shaking starts and increases over time until a steady state is attained after roughly 6 h. The kinetic data indicates that there is no significant change in pseudoequilibrium concentration after this time up to 24 h. Lagergren’s rate equation
log(qe-q) ) log qe -
tkad 2.303
(1)
is commonly used to study the rate constant of the system (15). Here qe and q are the amounts adsorbed in µmol g-1 at the pseudoequilibrium and at a particular time within the pseudoequilibrium time (within 3 h for As(V) and 6 h for As(III), respectively), t is time in min, and kad is the rate constant in min-1. In Figure 1 log(qe-q) is plotted against t to evaluate the nature of the adsorption process; kad values calculated from the slopes of the lines are 0.022 for As(V) and 0.005 for As(III). The linear curves obtained for both As(V) and As(III) indicate the first-order nature of the adsorption process and suggest that the process depends on both the solution concentration and the number of available adsorption sites. Effect of Particle Size. To determine the effect of particle size on As(V) removal, various particle size fractions, i.e. < 45, 45-125, 125-200, and 0-200 µm, are tested in sorption experiments. Percentage removals of ≈ 96, 97, 99, and 100 have been found for these particle size ranges, respectively (experimental conditions: initial As(V) concentration ) 16 µM, adsorbent dosage ) 5 g L-1 AB, pH ) 7.0, I ) 0.01 M NaCl, and reaction time ) 3 h). The increased removal efficiency for the larger particles is very surprising because the efficiency of surface adsorption processes would normally be expected to increase as the surface area-to-volume ratio of the particles increased (i.e. finer particles were used). To help explain the observations, surface areas (for each particle size range) were determined, and it was found that the surface area slightly increases with the increasing particle size. This 2430
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 8, 2004
FIGURE 3. Modeling the As(V) adsorption envelope on activated Bauxsol (AB) using the diffuse layer model calculations. Reaction conditions: Ionic strength (I) ) 0.01 M NaCl, reaction time ) 3 h, AB dosage (X) ) 5.0 g L-1, initial As(V) concentration (C0) ) 123.07 µM. Solid or dashed lines represent the sum of As(V) surface species calculated by the model. In the legend, AB As(V) adsorption constants (AAC) and the AB intrinsic acidity constants (IAC) are used for FITEQL and PHREEQC; AB AAC and the hydrated ferric oxide (HFO) IAC are used for hydrous 1; and HFO AAC and HFO IAC are used for hydrous 2. may explain, at least in part, the greater adsorption observed using AB with a larger particle size. Moreover, both the larger and the finer particles were examined by electron microscopy, and the results (not shown) suggest that larger particles are highly porous aggregates of much finer particles and it is possible that they have more adsorption sites available than the smaller well crystalline particles. Further studies would be required to determine how much of the effect of AB particle size on As adsorption was related to differences in particle crystallinity and whether any other currently unidentified factor had an influence. In this study, the 0-200 µm particle size ranges is used because it provides the best As(V) adsorption capacity. To our knowledge, no study has reported the effect of particle size on As removal using activated red mud (ARM) or AB, though Gupta et al. (28) report that adsorption of Pb and Cr onto ARM decreased slightly with an increasing particle size. Effect of the pH. Because AB has a pH dependent surface charge, the pH dependence of As(V) and As(III) sorption onto AB is investigated over the pH range from 4.5 to 12.5 using a fixed solution concentration, adsorbent dosage, and contact time. Arsenate and As(III) adsorption at different pH values based on experimental data and modeled curves are presented in Figures 2 and 3, where the amount of As adsorbed per weight of AB is calculated by
qe )
(C0 - Ce) X
(2)
Here qe is the concentration of the As on the adsorbent (µmol g-1), C0 is the concentration of As in the solution (µM), Ce is the concentration of As in the particulate phase (µM), and X is the dosage of AB (g L-1). It can be seen from Figures 2 and 3 that the process is pH dependent, favoring As(V) sorption at pH values below 7.0, and As(III) sorption at a pH of 8.5. The reason no further increase in the As(V) adsorption capacity is observed below pH 7.0 in Figure 3 may simply reflect the fact that there was insufficient As(V) in the original test solution to satisfy all available adsorption sites. Arsenate adsorption on AB suggests a ligand-based adsorption; i.e. there is a decrease in anion sorption at higher pHs as strong negative surface charges develop and OH- is competing for the available adsorption sites. The pH dependence of As sorption onto AB can be further understood by investigating the point of zero surface charge (pHpzc) of AB; the pHpzc is where the surface charge switches from negative at higher pHs through zero to positive at lower pH values. When the pH is below the pHpzc the solid surface is positively charged and favors the adsorption of As(V) anions (e.g. H2AsO4-, HAsO42-) due to the Coulombic attraction, but when the pH is above the pHpzc the surface of the solids is negatively charged and anion adsorption must compete with Coulombic repulsion. The intrinsic acidity constants of 6.0 (pK1) and 7.8 (pK2) (27) can be used to indirectly estimate the pHpzc of AB from
pHpzc ) 0.5 [pK1(int) + pK2(int)]
(3)
which is originally given by Stumm (29). The estimation of pHpzc according to eq 3 is 6.9 for AB, which may help explain the observed decrease in As(V) adsorption above pH 7.0. However, it must be noted that Bauxsol is a complex mixture of different minerals, each with different pHpzc values e.g. the pHpzc value of hematite is 7.8, whereas that of maghemite, gibbsite, boehmite, and quartz are 6.7, 5.0, 8.2, and 2.0-2.2, respectively (29). These minerals can have different surface charges at a given pH, and this gives AB the capacity to remove As over a wide pH range. However, there will still be a pH value (equivalent to a pHpzc value) that will mark the transition from dominantly negative surface charges on the constituent particles in AB to dominantly positive charges. The transition will be, however, much more gradual than for pure minerals. The fact that AB removes As(V) more efficiently at pH < 7.0 and that As(V) sorption decreases rapidly as the pH rises above 7.0 may reflect the importance of hematite (pHpzc about 7.8) and maghemite (pHpzc about 6.7) in the sorption process (15). The data obtained in this study agree with the findings of previous studies that show that the adsorption of As(V) on metal oxides and oxyhydroxides increases at lower pH values and gradually decreases at higher pH values (30) and that the optimal pH value (i.e. where the maximum sorption occurs) for As(III) is between 7.0 and 8.5 for red mud and amorphous oxides (15, 22). Modeling the As(V) Adsorption Envelope. The As(V) adsorption envelope on AB at the initial As(V) concentration of 123 µM is shown in Figure 3 along with the modeling results. Figure 3 shows good agreement between the experimental and the model data for FITEQL and PHREEQC, whereas there is poor agreement for two other scenarios identified as hydrous 1 and hydrous 2. The thickness of the double layer was calculated as ≈3 nm using the following simplified equation from Stumm (29)
K ) 3.29 * 109 I1/2
(4)
where K-1 is the double layer thickness (in m) and I is the ionic strength of the solution (in M). The intrisinic acidity
constants for activated red mud (note again that red mud is the raw material of AB) are given by Gu ¨ c¸ lu ¨ and Apak (27) as 6.0 (pK1) and 7.8 (pK2) and used in the present study for AB. The As(V) adsorption constants obtained in the present study using DLM in FITEQL are summarized in Table 2 and used in both PHREEQC and FITEQL to prepare the adsorption envelopes. Note that the validity of these constants depends on the maximum number of available sites, and in this study the maximum number of available sites is determined to be 2.21 × 10-4 M based on adsorption data at pH ) 4.5 where the maximum adsorption occurs. Following (25) the value of VY (the goodness of fit of the model) is defined as
∑(Y /S )
2
VY )
i
i
(5)
(npnc) - na
where Yi is the residual for each data point i on the adsorption envelope, Si is the estimated error, np is the number of data points in the adsorption envelope, nc is the number of system components for which both the total and free concentrations are known, and na is the number of adjustable parameters in the system. This corresponds to the overall variance as a weighted sum of squares divided by the degrees of freedom (9). A goodness of fit in the range of 0.1-20 indicates good agreement between the model and experimental data, and the value of 3.44 obtained for VY shows that the DLM in FITEQL program agrees well with the experimental results. In this part of the study, two scenarios (i.e. hydrous 1 and 2) are developed to help understand the AB surface. In the case of hydrous 1 the As(V) adsorption constants for HFO are taken from Dzombak and Morell (24), and the intrinsic acidity constants are the same as those from Gu ¨ c¸ lu ¨ and Apak (27). For hydrous 2 both the As(V) adsorption and the acidity constants for HFO are taken from the database of Dzombak and Morell (24). In hydrous 1, the curve fits well only in the low and high pH range (pH < 5 or > 10). However, between pH 5 and pH 10, lower model adsorptions are obtained compared to experimental data. In hydrous 2 there is a poor agreement between the model and experimental data at all pH values, as the AB surface is modeled entirely as an HFO surface (Figure 3). The model values do not fit the experimental data over the entire pH range indicating that the AB surface is a complex surface and that there are more minerals involved in the sorption of As(V) than just Fe-oxides. The As adsorption capacity of AB is comparable or better than that of HFO depending on the experimental conditions (13). Adsorption Isotherms. The experimental data obtained for As(V) and As(III) are applied to the Langmuir isotherm, its linearized form, and Freundlich isotherm, given respectively by
qe )
(Q0bCe)
(6)
(1 + bCe)
Ce Ce 1 ) + qe (Q0b) Q0
(7)
log qe ) log K + 1/n log Ce
(8)
Here Ce is the pseudoequilibrium concentration in µM, qe is the amount adsorbed at pseudoequilibrium in µmol g-1, Q0 is the adsorption maxima in µmol g-1, and b, K, and n are isotherm constants. In this study, higher correlation coefficients indicate that the Langmuir isotherm fits the adsorption data better than the Freundlich isotherm. Moreover, AB has a limited adsorption capacity, thus the adsorption can be better defined by the Langmuir isotherm rather than by the Freundlich isotherm, as an exponentially increasing VOL. 38, NO. 8, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2431
TABLE 3. Calculated Adsorption Constants and Thermodynamic Parameters Langmuir constants concn range, µM As(V), AB
5.74-182.57 7.70-136.50 2.0-151.87 6.91-152.18 7.03-220.85 As(V), Bauxsol 1.47-160.46 11.08-194.45 0.67-40.00 0.80-32.00 As(III), AB 2.04-156.66
temp, °T
pHa
15 20 23d 50 23d 23d 23d 23d 23d 23d
7.0 ( 0.1 7.0 ( 0.1 7.0 ( 0.1 7.0 ( 0.1 4.5 ( 0.1 10 ( 0.1 7.5 ( 0.1 7.3 ( 0.1 6.3 ( 0.1 6.8 ( 0.1
Q0,
b, L µmol-1 mmol g-1 0.41 0.52 0.97 1.01 0.44 0.01 0.15 0.27 0.76 0.032
25.91 28.90 39.84 40.98 102.00 6.08 9.97 9.26 14.43 7.22
thermodynamic parameters
r2 b 0.98 0.98 0.95 0.96 0.98 0.97 0.99 0.95 0.99 0.99
-∆G, ∆S, ∆H, kJ mol-1 kJ mol-1 K-1 kJ mol-1 30.971 32.063 34.156 37.130 32.20 22.67 29.33 30.78 33.33 25.53
0.177 0.178 0.182 0.177 0.175
20.050
Freundlich constants
r2 b 0.99 0.96 0.95 0.96 0.99 0.96 0.98 0.98 0.96 0.93
K, 1/nc L µmol-1 (µM)/(µM)-n 5.93 6.71 17.08 16.85 63.27 1.06 1.77 1.59 6.70 0.37
0.51 0.59 0.58 0.42 1.11 0.80 0.39 0.75 0.80 0.59
a The pHs were all controlled within (0.1 of the reported values. b r2 is a correlation coefficient. c n is a dimensionless constant. d Room temperature is reported within (1 °C. Adsorbent dosage (X) is 5 g L-1 for activated Bauxsol (AB) and 10 g L-1 for Bauxsol.
FIGURE 4. Langmuir plot for As(III) adsorption on AB with initial As(III) concentration range ) 2.0-156.7 µM.
FIGURE 5. Langmuir isotherm for As(V) adsorption on AB with reaction conditions: Ionic strength (I) ) 0.01 M NaCl, reaction time ) 3 h, pH ) 7.0 ( 0.1, AB dosage (X) ) 5.0 g L-1, initial As concentration (C0) range ) 2-152.0 µM. Diffuse layer model calculations, line represent the sum of As(V) surface species calculated by the model. Model (PHREEQC) data are calculated using the AB As(V) adsorption constants, and the intrinsic acidity constants are taken from the literature (27).
adsorption is assumed in the Freundlich isotherm. Langmuir plot for As(III) and the Langmuir isotherm for As(V) are given in Figures 4 and 5. The slopes of the linearized Langmuir plots are used to calculate the adsorption constants given in Table 3. The Langmuir model (in which adsorption is limited by surface site saturation) has been effectively applied to As and metal adsorption onto red mud, which consists of a heterogeneous mixture of several minerals (15, 31, 32). The data in this study show that it can also be effectively 2432
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 8, 2004
FIGURE 6. As(V) adsorption on activated Bauxsol (AB) as a function of the AB dosage with reaction conditions: Ionic strength (I) ) 0.01 M NaCl, pH ) 7.0 ( 0.1, reaction time ) 3 h, initial As(V) concentration (C0) ) 7.75 µM . applied to As adsorption onto AB, which consists of a somewhat less complex mineral assemblage. The measured removal percentages at pH 7.0 ( 0.1 are found to be 96-99 for As(V) and 13-53 for As(III) when 5 g L-1 AB is applied. The calculated adsorption parameters (Table 3) also show that As(V) is adsorbed by AB in appreciable quantities whereas the adsorption capacity for As(III) is much lower. The As(III) may be sorbed to the AB by reaction with minerals or by specific adsorption in the AB and not necessarily by simple Coulombic attraction. The As(V) adsorption capacity of AB measured in this study is comparable to, or better than, that of goethite and gibbsite (8), hematite, activated bauxite, and activated alumina reported previously. Modeling As(V) Adsorption. The experimental and the modeling results are depicted in Figure 5, where it can be seen that the modeled values fit the experimental data well, except for the last point where the Ce is underestimated. A possible reason for the deviation of the last point could be measurement inaccuracy (the As detection limit is 2 µg L-1), but it is also possible that once the As concentrations fall below some presently undetermined value, then a decrease in the probability of an As(V) ion encountering a suitable binding site may cause some deviation from model fitted curves that are dominated by data for higher concentration conditions. Effect of the AB Dosage. The sensitivity of As(V) adsorption efficiency to adsorbent dosage was determined by testing dose rates of 0.2, 0.4, 1.0, 2.0, 5.0, and 10.0 g L-1 AB at a constant test solution As(V) concentration and pH. The results depicted in Figure 6 show an increasing As(V) removal efficiency and decreasing pseudoequilibrium solution con-
centrations with increasing AB dosage. This behavior implies that the sorption depends on the availability of binding sites. Thermodynamic Parameters. The adsorption of As(V) on AB as a function of temperature was investigated within the temperature range from 288 to 323 K, and the Gibbs free energy (∆G°), standard enthalpy (∆H°), and standard entropy changes (∆S°) are calculated in J mol-1 K-1 for the adsorption process using eqs 9-11, respectively:
∆G° ) RT ln
() ( )
lnb3 - lnb2 ) ∆HR°
1 b1
1 1 T1 T2
∆G° ) ∆H° - T∆S°
(9)
(10) (11)
Here b1, b2, and b3 are the Langmuir constants corresponding to temperatures at 288, 293, 297, and 323 K, R is the ideal gas constant (8.3145 J mol-1 K-1), and T is the temperature (K). Relevant data are tabulated in Table 3, and it may be that the increase in adsorption with temperature is due to the increased rate of diffusion of adsorbate molecules into the AB pores. In addition, the negative ∆G° values calculated here confirm that adsorption will be spontaneous under the conditions applied, and the decrease in ∆G° with increasing temperature implies more efficient adsorption at higher temperatures. The negative enthalpy changes (∆H°) for the adsorption indicate the endothermic nature of the process. The positive value for the entropy change (∆S°) reflects the affinity of AB toward As(V) anions in aqueous solutions and may suggest some structural changes in the adsorbent (15). The increase in adsorption with increasing temperature may also indicate that chemisorption is taking place in the system and possibly that there is some tunneling of adsorbed ions into mineral phases in the AB (33). Red mud is the raw material of Bauxsol and of AB (see Materials and Methods) and exhibits similar adsorbent behaviors to Bauxsol and AB. For example, it can be activated to form activated red mud (ARM) (31) using a similar method that is used in this study to prepare the AB. ARM is thus expected to show similar thermodynamic behavior to AB. The effect of temperature on As(V) and phosphate adsorption by ARM is studied elsewhere (15, 31), and analogous results are reported. Desorption Studies. The reversibility of the As(V) adsorption onto AB was studied at different pH values and different solution concentrations. The investigation was carried out using the spent AB (AB previously used for As(V) adsorption) following the same experimental procedures used for the adsorption experiments. Arsenate desorption with changing pH and solution concentration is presented in Figure 7, which shows that a maximum of 40% desorption takes place at pH 11.6, and suggests that NaOH could be a partly effective desorption agent. The remarkably low reversibility of As(V) adsorption using AB indicates that the mechanism governing the adsorption process cannot be ion exchange alone but may involve chemisorption and much stronger forces comparable to those leading to the formation of chemical compounds. Apak et al. (32) also reported that metal adsorption on a similar adsorbent (i.e. ARM) is irreversible and indicates the formation of inner-sphere complexes. The possible formation of inner sphere surface complexes when using Bauxsol or AB for As(V) removal is reported elsewhere (16, 17).
Discussion Arsenic adsorption onto activated Bauxsol (AB) is evaluated in batch experiments as a function of the contact time, particle size, pH, initial As concentration, adsorbent dosage and
FIGURE 7. (a) Adsorption and desorption of As(V) as a function of solution concentration at pH ) 4.5 with initial As(V) concentration range ) 7-220 µM and (b) desorption of As(V) as a function of solution pH at initial As(V) concentration (C0) ) 123.1 µM. AB and the spent AB dosage ) 5 g L-1. temperature. The results demonstrate that within the pH range studied optimum As removal is achieved at pH 4.5 for As(V) (roughly 100% removal) and 8.5 for As(III) (removal efficiency depends on the initial As(III) concentration). Arsenate can also be efficiently removed at circum-neutral pH values, and a dosage of 0.4 g L-1 AB can be enough to achieve the 10 µg L-1 (WHO provisional guideline for As) at pH 7.0 when the initial As(V) concentration is
