The main objectives of this study were to evaluate the effectiveness of using iron impregnated granular activated carbon (AC) to remove arsenic from water and to assess the partitioning behavior of arsenic under a variety of conditions. Iron impregnated granular activated carbon (AC-Fe) composites were prepared with different ferric (Fe+3) concentrations, ranging from 0.09 to 3.0 M. These AC-Fe composites were able to remove 92–98% of the arsenate [As(V)] and 42–65% of the arsenite [As(III)]. The composite containing the lowest iron concentration (1.54%) was the most effective at arsenic sorption. Langmuir model fit indicated that the maximum 125 mg As(V)/gFe and 98.4 mg As(III)/gFe can be sorbed by the composite. The kinetics of arsenic sorption is well explained by pseudo first-order kinetics. The arsenate removal efficiency was found to decrease with increasing solution pH, while the As(III) removal efficiency was found to increase. The background ionic strength (IS) had no significant effect of on As(V) removal, but As(III) removal increased when the IS was greater than 50 mM NaCl. Our results indicate that a small amount of iron embedded efficiently in AC may have considerable potential in removing arsenic from water.
- composite material
- heavy metal removal
- iron distribution
Arsenic contamination in groundwater has become a major problem in many parts of the world. Groundwater contamination by arsenic in Bangladesh has been recognized by the World Health Organization (WHO) as the largest mass poisoning of a population in human history. Approximately 100 million people in Bangladesh and India are affected by arsenic poisoning, and millions more are exposed to arsenic contamination in China and Latin America. Evidence has also been reported of high arsenic contamination in the environment at some locations in Canada, the USA, Greece, Germany, and Thailand, as a result of mining activities (Smedley & Kinniburgh 2002).
Adsorption of As on different types of sorbents (i.e., fly ash, different biomass, granular activated carbon (AC), activated alumina, zirconium oxide, goethite, siderite, different iron oxides and hydroxide, iron coated sand, etc.) have been studied extensively for arsenic removal (Mohan & Pittman 2007; Mondal et al. 2013). Of the various adsorbent available for arsenic removal, the most widely accepted involves the use of iron based adsorbents. A few recent studies have explored the possibility of using iron-impregnated granular activated carbon (AC-Fe) composites as an adsorbent to remove arsenic from water (Gu et al. 2005; Chen et al. 2007; Hristovski et al. 2009; Vitela-Rodriguez & Rangel-Mendez 2013).
The hypothesis of this study is to combine the adsorption active surface of Fe with the high specific surface area (SSA) of AC in the most efficient way to improve the performance of the composite as a filter media for arsenic removal. The arsenic removal capacity of AC-Fe composites is significantly influenced by the conditions under which they are prepared (Gu et al. 2005). For example, some of these studies suggested that with increase in initial Fe2+/Fe3+ concentration arsenic removal increases, however, after certain iron concentration arsenic removal capacity decreases (Gu et al. 2005; Hristovski et al. 2009). Those authors anticipated pore clogging by excess iron might cause reduction in As removal. To the best of our knowledge, none of the previously published studies has involved any systematic investigation into the distribution pattern of iron within activated carbon composites and the consequent arsenic sorption trends within the composites. With the advancement of materials it is extremely important to understand the distribution pattern of amorphous or crystalline metal within the pores of microporous media and its effect on the performance of the composite. The primary objectives of this study were therefore to determine the effective distribution of iron impregnated into AC for arsenic removal, and to evaluate the partitioning behavior of arsenic on the composite surfaces under a variety of conditions. The iron distribution patterns and arsenic sorption locations within AC-Fe composites were assessed qualitatively. The sorption behavior (isotherm and kinetics) of arsenic by AC-Fe composites was assessed separately for As(V) and As(III), since both of these arsenic species are likely to occur in natural groundwater systems. The effect of solution chemistry, i.e. of ionic strength (IS) and pH, on arsenic removal was also evaluated.
MATERIALS AND METHODS
All of the chemicals used were reagent grade and only Millipore deionized (DI) water was used throughout. Granular activated carbon (Darco 20–40 mesh: 425-850 μm size), sodium arsenate [Na2HAsO4·7 H2O >98%, used for As(V)], and sodium (meta)arsenite [NaAsO2 >90%, used for As(III)], were purchased from Sigma-Aldrich, Germany. Ferric nitrate [Fe(NO3)3·9H2O >98%] was obtained from VWR International, Germany. For pH adjustment we used reagent grade sodium hydroxide (NaOH) or hydrochloric acid (HCl); sodium chloride (NaCl) was used to prepare the background electrolyte.
Synthesis of AC-Fe composites
AC-Fe composites were synthesized using a hydrolysis method similar to that reported in a number of previous studies (Gu et al. 2005; Chen et al. 2007). A series of flasks containing 200 mL of Fe(NO3)3·9H2O solution (with a range of concentrations from 0.09 to 3.0 M) and 10 g of AC were mixed for 1 hour at 30 rpm. The mixtures were then kept for 24 hours at 25°C to allow the diffusion of iron ions into the AC pore spaces. The excess solution in each flask was then discarded and the residual mixtures heated to 110°C to start hydrolysis and subsequently dried for 24 hours. After drying, the mixtures were left to reach room temperature (25°C) and then sieved using a nylon mesh (of 250 μm size). Thereafter, they were rinsed several times with DI water to remove any excess iron ions. The resulting AC-Fe composites were then dried and stored.
Characterization of the AC and AC-Fe composites
Scanning electron microscopy–energy dispersive X-ray
Samples of AC and AC-Fe composites (with and without arsenic sorption) were spread over conductive carbon and then coated with carbon for scanning electron microscopy (SEM) analysis. Several AC-Fe grains were also sectioned in order to analyze the inner faces of the composites. The SEM images were obtained using a Zeiss DSM 982 Gemini instrument equipped with a Schottky field emitter. Energy dispersive X-ray (EDX) spectra were measured using an Apollo XP silicon drift detector. The accelerating voltages were maintained at 20 kV. The raw data measurements were recorded in weight percent.
X-ray diffraction analysis
Finely ground powder from the AC-Fe composites was used for the X-ray diffraction (XRD) analyses. Diffractograms were recorded from 5 to 80° 2θ, with increments of 0.02° 2θ and an exposure time of 0.5 seconds per step. A D2 Phaser (BRUKER) with a Cu X-ray tube which operated at 30 kV and 10 mA was used. The intensity ratio of Cu-Kα2 to Cu-Kα1 was about 0.5.
In order to determine the iron content of the AC-Fe composites, 0.1 g samples from each of six batchs (B1 to B6) were acid digested in concentrated HNO3 and 30% H2O2 (as described in EPA Method 3050B) and then analyzed using inductively coupled plasma (ICP) emission spectroscopy (iCap 6,000 series, Thermo Scientific).
Specific surface area
The SSAs of granular activated carbon and AC-Fe composites were determined from their nitrogen sorption-desorption isotherms, measured using an Autosorb-1 analyzer (Quantachrome Instrument, Germany) following the Brunauer, Emmett and Teller (BET) method.
Arsenic removal efficiency: effect of iron loading, pH and ionic strength
Arsenate [As(V)] and arsenite [As(III)] stock solutions (10 mg L−1) were prepared from sodium arsenate (Na2HAsO4·7H2O) and sodium arsenite (Na2AsO2), respectively. In our experiments the stock solutions were diluted to 1 mg L−1 As(V)/As(III), with a background IS of 0.1 mM NaCl and an stabilized pH of 4.8 ± 0.3. The batch experiments were carried out by adding 100 mL of As(V) or As(III) solution to a series of glass bottles that each contained approximately 0.1 g of AC-Fe composite. The bottles were mixed at 30 rpm using an end-over-end rotator for 24 hours at 25°C. The samples were then removed from the bottles and filtered through a 0.45 μm cellulose acetate syringe filter prior to analysis.
In order to evaluate the relative arsenic removal efficiencies of AC-Fe composites with different iron loadings, a series of these experiments was carried out under similar conditions (as specified above) using the different AC-Fe composites (B1 to B6, Table S1, available online at http://www.iwaponline.com/ws/015/057.pdf). The effects of pH and IS on arsenic removal were also assessed using the AC-Fe (B1) composite, as this was identified as the most efficient composite for arsenic removal. This involved varying the pH of the solution from 4 to 10 (with an IS of 0.3 ± 0.2 mM), and varying the IS from 0.1 to 100 mM NaCl (with an initial pH of 7.7 ± 0.3, which stabilized at 4.8 ± 0.3 after 30 minutes), these being relevant ranges for natural water. The pH values of the systems were adjusted by adding either 0.1 N NaOH solutions or 0.1 N HCl acid.
Sorption kinetic tests
To evaluate the sorption kinetics, 100 mL of 1 mg L−1 As(V) or As(III) solution (background IS 0.1 mM NaCl and pH = 4.8 ± 0.3) was poured into a series of glass bottles, each of which contained 0.1 g of AC-Fe(B1) composite. The bottles were then mixed with end-over-end rotator at a temperature of 25°C and the samples were collected after different periods of time (5, 15, 30, 45, 60, 75, 90, 120, 180, 225, 280 and 320 minutes). A few samples were retained for longer durations (24, 48 and 72 hours) in order to ensure that a 24 hour time span was sufficient for equilibrium concentration to be achieved.
Equilibrium sorption experiments
For equilibrium sorption experiments the concentration of arsenic was maintained at 2–2.5 mg L−1 while the mass of adsorbent AC-Fe(B1) composite was varied over two orders of magnitude, ranging from 0.005 to 0.5 g. The glass bottles containing 100 mL of As(V) or As(III) solution (with a pH of 4.8 ± 0.3 and an IS of 0.1 mM NaCl) and different quantities of AC-Fe(B1) composite, were again mixed using an end-over-end rotator for 24 hours at 25°C.
All experiments were carried out in duplicate (or in triplicate, e.g., for the iron loading) and mean values are presented in the results, together with the maximum–minimum range (or standard deviation, in the case of iron loading). The filtered samples were acidified using 1% HNO3 and analyzed by ICP-emission spectroscopy (iCap 6,000 series, Thermo Scientific). The arsenic peak was detected at a wavelength of 189.04 nm, with a lower limit of detection of 2.8 μg L−1.
RESULTS AND DISCUSSION
Iron content and SSA
The iron content in the unmodified AC was negligible (<0.1%, Table S1). The percentage of iron retained in AC-Fe composites increases from 1.54 to 6.01% (wt.%) as the initial Fe+3 concentration increased from 0.09 to 3.0 M (Figure 1(b) and Table S1). The SSA of unmodified AC was 582 m2 g−1, increasing to 614 m2 g−1for the AC-Fe(B1) composite, which had lowest iron content (1.54%). This increase in SSA could be explained by the fact that (i) pore blocking was negligible for the AC-Fe(B1) composite and (ii) the available surface of the impregnated iron (hydr)oxide can contribute to an increase in the SSA of the composites. The SSA of AC-Fe composites decreased to a certain extent (542–529 m2 g−1) as the iron content increased. A few other studies have also observed that the SSA of such composites decreases as the amount of impregnated iron increases (Gu et al. 2005; Hristovski et al. 2009).
SEM-EDX and XRD
The surfaces of the AC-Fe composites vary both between and within the grains, from smooth planes to quite porous surfaces. In addition to carbon and oxygen, silicon, iron, aluminum, and sulfur were also found as natural constituents in the AC-Fe composites. However, the SEM-EDX analyses of AC-Fe composites containing arsenic indicated that the arsenic was primarily associated with iron rather than with either silicon or aluminum (Figure S1(a), S1(b), available online at http://www.iwaponline.com/ws/015/057.pdf). In most of the SEM images (and all of the images for the AC-Fe(B1) composite) the iron particles could not be resolved, even though the magnification was of the order of 30,000 times. However, in a few samples (especially of the AC-Fe(B6) composite) clusters of iron particles were identified. The small difference in XRD spectra between AC and AC-Fe composites (Figure S1(c), online at http://www.iwaponline.com/ws/015/057.pdf) suggests that the impregnated iron (hydr)oxides are X-ray amorphous.
Arsenic removal efficiency
The removal of arsenic by AC-Fe composites were shown to be significantly greater than removal by unmodified AC. AC-Fe composites were able to remove 92-98% of the As(V) and 42–65% of the As(III) from the solution, while unmodified AC could only remove 19.3% of the As(V) and 17.1% of the As(III) (Table S1). The removal of arsenic from the solution can involve a variety of mechanisms such as sorption, ion exchange, and co-precipitation. Arsenic co-precipitation is important when Fe+3 and/or Fe+2 ions are present in the solution. In this study the amount of dissolved iron in AC-Fe containing water was negligible (48–90 μg L−1). The insignificant effect which anion (Cl−) concentration had on arsenic removal suggests that the As is most likely removed through the formation of inner-sphere surface complexes with iron, rather than by ion exchange. Furthermore, the SEM-EDX analyses indicate that arsenic is present on those AC-Fe surfaces where Fe is also present, suggesting that iron (hydr)oxide is the most active site for arsenic sorption.
Adsorption experiments were carried out in oxic condition with high DO, where As(V) is more stable (Pierce & Moore 1982). As(III) removal efficiency by AC-Fe composite is much lower compared to As(V) removal efficiency (Table S1), which is in good agreement with other studies conducted with iron based adsorbents (Kim et al. 2004; Guo & Chen 2005; Gu et al. 2007). As(V) adsorption capacity decreases, whereas As(III) adsorption capacity increases with an increase in solution pH (as discussed later), which is a typical phenomena occurring due to formation of different surface complexes (Dixit & Hering 2003; Kim et al. 2004). Furthermore, the DO value during the experiment does not alter much (8.05 to 7.87 mg L−1), and the pH (∼7.7 ± 0.3) of the solution decreases uniformly until 30 minutes and then it reaches to near equilibrium (∼4.8 ± 0.3) for all the solutions tested with AC-Fe composite (Figure S2, available online at http://www.iwaponline.com/ws/015/057.pdf). Different trends in arsenic removal capacities with respect to pH and IS for the solutions containing two different species of arsenic (as discussed below), suggest that the sorption of As(III) or As(V) probably controls the chemical equilibrium between them. Thus, transformation of arsenic prior and/ or during the experiments is therefore assumed to have been minor and the solutions prepared with sodium arsenite and sodium arsenate are assumed to have had dominant proportions of As(III) and As(V), respectively.
Effect of iron loading on arsenic removal
The arsenic sorption capacities of different AC-Fe composites (synthesized with different Fe+3 concentrations) ranged from 55.4 to 11.5 mg As(V)/g Fe and from 23.8 to 7.9 mg As(III)/g Fe embedded in AC, and are presented in Figure 1(a) (and Table S1). Iron embedded in AC-Fe(B1) composite was found to have a greater arsenic [As(V) and As(III)] sorption capacity than that embedded in the other composites.
The atomic percentages (at.%) of iron and arsenic on the surfaces and within the pore spaces of AC-Fe(B1) composite (which contained the least amount of iron) and AC-Fe(B6) composite (which contained the greatest amount of iron) were analyzed and compared using EDX measurements (Figure 1(c)–(f)), in order to obtain greater insight into their distribution patterns. The amount of arsenic (in at.%) adsorbed on the composite surfaces generally increased in the proportion with the relative amount of iron content (in at.%; Figure 1(c)–(d), Clusters 1, 2 and 4). In the AC-Fe(B1) composite, sorption of As(V) was even identified within the pore spaces, but at these locations sorption of As(V) occurs at higher iron concentrations (Cluster 2, Figure 1(c)) than on the outer surface (Cluster 1, Figure 1(c)). For the AC-Fe(B6) composite no arsenic was identified in some of the inner pore spaces, even though the iron content was relatively high (Cluster 3, Figure 1(d)). In contrast, however, at a few locations on the outer surfaces of AC-Fe(B6) composites where the surfaces were smooth, the relative proportion of arsenic increases irrespective of the iron content (Cluster 5, Figure 1(d)). The uneven distribution of iron and pore blocking noted in the AC-Fe(B6) composite can probably be attributed to a reduction in the effective surface area of iron available for sorption, affecting the composite's sorption capacity for arsenic. In contrast, significantly elevated arsenic sorption by iron embedded in AC-Fe(B1) composites results from a uniform distribution of iron with the maximum surface area available for sorption. An AC-Fe composite with an initial Fe concentration of 0.09 M (AC-Fe(B1) composite) was therefore used for all subsequent experiments.
Equilibrium sorption isotherm
Equilibrium sorption experiments were carried out to evaluate the solid–liquid phase partitioning of arsenic ions at equilibrium, using the AC-Fe(B1) composite as the absorbent. Sorption isotherms for As(V) and As(III) (Figure 2(a)) were compared with the Freundlich and Langmuir isotherm models, as described in the supplementary material (available online at http://www.iwaponline.com/ws/015/057.pdf).
The Langmuir isotherm model was found to better explain the As(V) (r2> 0.99) sorption pattern than the Freundlich isotherm model (r2 = 0.94), which is in agreement with other studies conducted using different types of iron-based adsorbents (Chen et al. 2007; Gu et al. 2007; Hristovski et al. 2009; Vitela-Rodriguez & Rangel-Mendez 2013). For As(III) sorption, both the Langmuir (r2 = 0.89) and Freundlich (r2 = 0.88) isotherm models demonstrated a similar quality of fit. A lower value for the Freundlich constant (1/n; Table 1) for As(V) (0.25) than for As(III) (0.68) indicates that As(V) sorption is more favorable, even at very low arsenate concentrations.
The Langmuir model fit indicates maximum sorption capacity (qmax) values for arsenic on AC-Fe(B1) surfaces are 1.93 mg As(V) (g AC-Fe)−1 and 1.52 mg As(III) (g AC-Fe)−1 (Table 1). The maximum sorption capacities for arsenic with respect to iron are 125 mg As(V) (g Fe)−1 and 98.4 mg As(III) (g Fe)−1, respectively. This is in good agreement with values reported from other studies which demonstrated arsenic sorption, either for amorphous iron oxide (approximately 40–400 mg arsenic (g Fe)−1 (Dixit & Hering 2003)), or for mesoporous alumina (47–121 mg arsenic (g alumina)−1 (Kim et al. 2004)). The calculation of Gibbs free energy (ΔG) from the Langmuir sorption isotherm constant showed that more energy is released as a result of As(V) sorption (−3.40 kJ mol−1, Table S1) than as a result of As(III) sorption (−0.39 kJ mol−1), confirming that As(V) can be more easily adsorbed on AC-Fe surfaces than As(III) (Kim et al. 2004).
The sorption behavior of arsenic was observed over a period of 320 minutes. The experimental data were compared with the pseudo first-order and pseudo second-order sorption kinetics models, and with the Webber Morris model (details of the models are provided in the supplementary material).
Pseudo first-order sorption kinetics (r2 = 0.82–0.90) can better explain the data for both As(V) and As(III) over the 320 minute period than pseudo second-order sorption kinetics (r2 = 0.45-0.75) (Figure 2(b), Figure S2(a), and Table S1). Approximately 50% of the As(V) and 25% of the As(III) were removed within the 320 minute period. Some previous studies have suggested that arsenic sorption kinetics are better explained by pseudo second-order kinetics, with much faster rates of sorption by iron-based adsorbents (Gu et al. 2007), but it is important to note that the initial arsenic concentration in those studies was 20 times lower (0.05 mg L−1) than in our own investigations (arsenic concentration = 1 mg L−1, AC-Fe dose = 1 g L−1), with a similar range of adsorbent doses (1-3 g L−1). The sorption kinetics are highly dependent on the equilibrium concentration and thus on the initial concentration of sorbate and the adsorbent dose (Ho & McKay 2000). Fitting the Webber Morris model to the dataset indicates two distinctive trends with different values for the diffusion rate (k′′t) (one from 0–30 minutes, and the other after 30 minutes; Table 1 and Figure S2(b)) for both arsenic species. Differences in the range of solution pH values within 30 minutes (pH: 7.7–4.8) and after 30 minutes (pH: 4.8–4.26, Figure S2(c)), possibly due to stabilization of the system, are likely to be responsible for these different trends. The intraparticular diffusion rate for As(V) was higher (Table 1) than that for As(III). This may be attributable to the ability of arsenate ions to diffuse more easily in micro-pores because of their small size (the radius of HAsO42−is 3.97 Å, while the radii of H2AsO4− and H3AsO4 are 4.16 Å) compared to arsenite ions (the radii of H2AsO3− and H3AsO3 are 4.8 Å) (Kim et al. 2004).
Effect of solution chemistry on arsenic removal
Effect of pH
The effect of pH (which ranged between 4.2 ± 0.2 and 10 ± 0.1) on As(V) and As(III) sorption was evaluated and the results are presented in Figure 2(c). The pH of the system decreased steadily until reaching a near-equilibrium value after 30 minutes (Figure S2(b)). The percentage of arsenic removed has been indicated with respect to final pH (pH measured at the end of the experiments) of the system. It was observed that As(V) removal decreased as the solution pH value increased beyond a pH value of 6.0, whereas As(III) removal increased as the solution pH increased above a pH value of 4.6. The As(III) removal efficiency eventually becomes higher than the As(V) removal efficiency between pH values of 8 and 9. Numerous studies have suggested a decreasing trend in As(V) and increasing trend in As(III) sorption with increasing solution pH by different metal-based adsorbents which can be attributed to the formation of different surface-species and interaction mechanisms (Goldberg & Johnson 2001; Dixit & Hering 2003; Kim et al. 2004; Gu et al. 2007). For example, at low pH As(V) is mostly present in the form of H2AsO4− in the solution, which can be adsorbed more effectively than HAsO42− which is likely to be present in the solution at high pH values (Kim et al. 2004). In contrast, at a high pH range As(III) is present as H2AsO3− ions, and at pH values below 9.24 a non-charged trigonal species [As(OH)3] of As(III) predominates (Stachowicz et al. 2006), which is likely to be responsible for the reduction in As(III) sorption at lower pH values.
Effect of IS
The percentages of As(V) and As(III) removed as the IS increased from 0.1 to 100 mM NaCl are presented in Figure 2(d). The As(V) removal efficiency decreased slightly (by no more than 10%) as the IS increased from 0.1 to 10 mM NaCl, but any further increase in the IS above 10 mM NaCl had a negligible effect on arsenate removal. In contrast, the As(III) removal did not vary much within an IS range of 0.1 to 50 mM NaCl but increased by approximately 18% as the IS increased from 50 to 100 mM NaCl. A few studies (Guo & Chen 2005; Vitela-Rodriguez & Rangel-Mendez 2013) have demonstrated that As(V) sorption on iron-carbon-based adsorbent surfaces is significantly reduced in the presence of certain anions (e.g. Cl−, SO42−, SiO32−, and PO43− anions) due to competition for active sorption sites. The arsenic sorption trend with respect to IS suggests that there is unlikely to be any competition between arsenic and Cl− ions for active sorption sites within this system. When the sorption of arsenic is either independent of the IS or increases with increasing IS, a more probable interaction mechanism is through inner-sphere surface complexes (Goldberg & Johnson 2001). Inner-sphere surface complexes between arsenic and iron embedded in AC therefore appears to be the most likely interaction mechanism in our system.
AC-Fe composites containing the lowest amount of iron (1.54%), embedded uniformly in activated carbon with no noticeable pore blocking, yielded the best performance for arsenic removal, with maximum adsorption capacities of 125 mg As(V) and 98.4 mg As(III) per gram of Fe. The iron distribution within the composite was found to be an important factor controlling sorption, which is crucial information for the application of metal-based nanocomposites to remove contaminants or heavy metals from water. The efficiency of As(V) removal was found to decrease as the pH increased, whereas the efficiency of As(III) removal increased significantly. The pH of the natural water can therefore be adjusted before filtering to suit the arsenic speciation, in order to maximize the total arsenic removal performance of the composite.
- First received 24 November 2014.
- Accepted in revised form 17 April 2015.
- © IWA Publishing 2015