Environ. Sci. Technol. 2004, 38, 2368-2372
Arsenic Removal from Aqueous Solution via Ferrihydrite Crystallization Control WILLIAM R. RICHMOND,* MITCH LOAN,† JONATHON MORTON, AND GORDON M. PARKINSON Nanochemistry Research Institute and A. J. Parker CRC for Hydrometallurgy, Curtin University of Technology, GPO Box U 1987, Perth, 6845, Australia
Removal of arsenate anion from aqueous solution by coprecipitation with ferrihydrite has been studied under conditions in which the Fe/As ratio is maintained at a constant level, while the degree of supersaturation with respect to the iron oxide precipitate is varied. An Fe/As ratio of 12 was chosen, and supersaturation was controlled by varying the iron concentration or the pH. The relationship between supersaturation and arsenic removal was found to follow an exponential curve, with greater arsenic removal occurring at higher supersaturation ratios for each of the pH values tested. Higher supersaturation ratios were required to achieve a given level of arsenic removal at pH 7 than would be required to achieve the same level of removal at pH 3.5. The results provide important guidelines for selection of appropriate concentrations of iron(III) required for arsenic removal under various circumstances. Powder XRD analysis of the arsenate-ferrihydrite precipitates showed an increasing degree of structural order with decreasing levels of supersaturation. TEM images of the precipitates revealed that aggregates with a morphology similar to that of schwertmannite are formed in some samples at low supersaturation levels. The results described in this paper indicate that the overall efficiency of arsenic removal involves a combination of both supersaturation and pH effects, with pH controlling the affinity of arsenate for the ferrihydrite surface, and supersaturation controlling the surface area and physical properties of the ferrihydrite product.
Introduction Ferrihydrite is an iron oxy-hydroxide known to play an important role in the natural environment. Its large surface area, strong adsorptive properties, high adsorptive capacity, and low cost make it an attractive material for removal of both cationic and anionic impurities from wastewater and drinking water (1). The use of ferrihydrite to remove arsenic from hydrometallurgical process solutions and wastewaters has received a great deal of attention over a number of years (2-6). Ferrihydrite readily adsorbs arsenic(V) in the form of arsenate anion (AsO43-), but probably the most effective method of removal of arsenic from aqueous solutions is through * Corresponding author phone: +61 8 9266 3838; fax: +61 8 9266 4699; e-mail:
[email protected]. † Present address: Materials and Surface Science Institute, University of Limerick, Republic of Ireland. 2368
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 8, 2004
coprecipitation of arsenic with ferrihydrite. At relatively high concentrations of iron(III) and arsenic(V) and at low pH, coprecipitation results in the formation of iron(III) arsenate, FeAsO4‚2H2O (5). When rapid neutralization is effected in solutions in which the iron(III) concentration is high and arsenic(V) concentration is low, as is usually the case in water purification, the product is a poorly crystalline iron oxyhydroxide phase (2-line ferrihydrite) incorporating the vast majority of the arsenic impurity (2). Whether the associated arsenic in the solid phase is structurally incorporated or surface-adsorbed has been the subject of some debate (7, 8), but the resulting solid has the characteristic broad 2-line XRD pattern of ferrihydrite. There have been numerous studies of the removal of arsenic via ferrihydrite coprecipitation that prove its efficiency, but these have mainly been concerned with varying the Fe/As ratio and pH to determine the conditions that give the best removal of arsenic (2). All these studies have shown that, in the range pH 4-7, a minimum Fe/As ratio of 3:1 is required for formation of a stable arsenical ferrihydrite precipitate. However, the concept of controlling the ferrihydrite structure as a means of improving the effectiveness of arsenic removal by ferrihydrite coprecipitation has received very little attention. Controlling the nature of the ferrihydrite precipitate not only offers the possibility of optimizing arsenic removal but also of improving the stability of the precipitate with regard to rerelease of the incorporated arsenic. It is also desirable to optimize the solid-liquid separation properties of the precipitate, which in the case of ferrihydrite are notoriously poor (9). It has been shown that the degree of structural order in a ferrihydrite precipitate can be controlled by varying the manner in which ferrihydrite crystallizes, essentially by varying the hydrolysis rate at a fixed iron(III) concentration (10). This observation implies that the structural order of ferrihydrite should also be affected by varying the ferrihydrite supersaturation. Recent studies of ferrihydrite precipitation in our group have involved evaluating the effects of varying supersaturation by changing the Fe(III) ion concentration and pH in a single-stage neutralization regime (9, 11). We have found that as the Fe(III) ion concentration, and thus the degree of supersaturation, is decreased, the crystallinity of ferrihydrite increases, and intermediate phases between 2- and 6-line ferrihydrite are formed. In this study, we take a similar approach, examining the effectiveness of arsenic removal while maintaining a constant Fe/As ratio and varying the pH and iron(III) concentration to control the degree of supersaturation of ferrihydrite. The Fe/As ratio was fixed at a value known to be outside the stability region for iron(III) arsenate formation (12), so as to vary only the ferrihydrite supersaturation. This approach has not been described previously in the literature.
Materials and Methods Coprecipitation Experiments. Iron(III) sulfate solution with arsenic added as sodium arsenate (300 mL) was placed in a stirred, jacketed reactor thermostated at 85 °C. The solution was then neutralized to the appropriate pH (3.5, 5.5, or 7) by a single addition of NaOH (10 and 1 M, in the appropriate volumes). Samples (20 mL) of the suspension were taken from the impeller zone, using a syringe and tube, at 10, 30, 60, and 120 min after neutralization. The solids were collected by vacuum filtration with 0.2 µm filter membrane, then washed twice with Milli-Q water (10 mL at 85 °C). The solids were dried in air for 24 h at 50 °C. Filtrates were retained and diluted for elemental analysis. 10.1021/es0353154 CCC: $27.50
2004 American Chemical Society Published on Web 03/18/2004
FIGURE 1. (a) Residual arsenic concentration in solution vs initial iron concentration for coprecipitation experiments carried out at various levels of supersaturation and at pH 3.5, 5.5, and 7.0. (b) Residual arsenic in solution (expressed as a percentage of the initial concentration) as a function of supersaturation in coprecipitation experiments carried out at pH 3.5, 5.5, and 7.0. Elemental Analysis. Flame atomic adsorption spectrometry (AAS) was used to measure iron and arsenic in the more concentrated samples, while inductively coupled plasmamass spectrometry (ICP-MS) was used to determine arsenic at trace levels. Flame AAS was carried out using a Varian SpectraAA-10 instrument with a spectraAA-100/200 PC upgrade. ICP-MS analysis was performed using a Plasmaquad 3, ICP-MS, high-resolution interface instrument. Powder X-ray diffraction patterns were recorded on a Philips X’pert powder diffractometer, with a cobalt long fine focus X-ray tube. Samples were finely ground in a mortar and pestle prior to analysis and then loaded onto siliconwafer zero-background plates by placing a small amount of solid onto the plate and dispersing with absolute ethanol. Transmission electron micrograph (TEM) images were recorded on a JEOL 2011 transmission electron microscope operating at 200 kV. Samples for TEM were gently ground and dispersed in ethanol and then drop-cast onto 200-mesh copper grids coated with a holey carbon film. TEM images were recorded on a JEOL 2011 TEM operating at 200 kV.
Results and Discussion The coprecipitation experiments were designed in such a way as to vary the ferrihydrite supersaturation by altering the initial Fe(III) ion concentration, while maintaining a constant Fe/As molar ratio of 12. This value falls within the optimum range reported by Krause and Ettel (12) but lies well outside the stability range for iron(III) arsenate formation, and our preliminary tests showed effective removal of arsenic at this molar ratio. Our approach differs from other studies in this field that have generally varied the Fe/As ratio to optimize the arsenic removal and stability (12, 13). Supersaturation is generally expressed as the dimensionless ratio S ) c/c*, where c is the concentration of the solute and c* is the equilibrium solubility of the solute. Measurement of the equilibrium solubility of ferrihydrite is experimentally challenging, particularly in the pH range 4-10, as ferrihydrite primary particles have a diameter of 5 nm and can pass through conventional filter paperssa problem that can tend to skew the analysis toward artificially high iron concentrations (14). We carried out some preliminary experiments to determine the equilibrium solubility of ferrihydrite at 85 °C and found equilibrium Fe3+ concentrations of 1.8 × 10-5 and 9 × 10-6 mol L-1 at pH values of 3.5 and 4.0, respectively. Data over an entire pH range at 85 °C is currently not available,
TABLE 1. Estimated Supersaturation Ratios (SEst) and Arsenic Content of Ferrihydrite Solids for Coprecipitation Experiments initial [Fe3+] (mol L-1)
initial [AsO43-] (mol L-1)
0.236 0.148 0.028 0.014 0.003
0.018 0.012 0.0024 0.00119 0.00024
pH 3.5
SEst/1000 pH 5.5
pH 7.0
13 8.3 1.6 0.78 0.16
130 83 16 7.8 1.6
220 140 26 13 2.7
due in part to the complexities just mentioned but also to the propensity for ferrihydrite to transform to hematite during the long equilibration times required for such measurements. For the purpose of this study, we have estimated the supersaturation ratios using the results just mentioned, and the value 1.07 × 10-6 mol L-1 for the concentration of Fe3+ in equilibrium with 6-line ferrihydrite at pH 7, as reported by Cornell and Schwertmann (14). The estimated supersaturation ratios (SEst) for coprecipitation experiments carried out at pH 3.5, 5.5, and 7 are given in Table 1 along with the associated initial iron and arsenic concentrations. The graph presented in Figure 1a shows the effects of varying the supersaturation with respect to ferrihydrite (through varying the initial iron concentration) on the ability of ferrihydrite to remove arsenic from solution. At each pH tested, the degree of arsenic removal decreases slightly with the Fe(III) ion concentration, although this effect is not dramatic. The slopes of the three lines obtained at the different pH values are identical, and it is clear that the pH has a far more significant effect on arsenic removal than the initial iron concentration, with the residual arsenic levels in the system decreasing at lower pH values. This can be attributed to the more positive surface charge on ferrihydrite at lower pH (the IEP of ferrihydrite is about 7.8) (14), leading to stronger affinity of the arsenate anion for the ferrihydrite surface. It should be noted, however, that as the initial arsenic levels have also been varied to maintain the constant Fe/As ration of 12, the results as presented in Figure 1a do not tell the full story. In Figure 1b, the same data are plotted, but the residual arsenic in solution is represented as a percentage of the initial arsenic levels, and the linearized forms of the arsenic removal curves show that the slopes of the lines at the three different pH values are identical but that the curves VOL. 38, NO. 8, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2369
FIGURE 2. Powder XRD patterns of arsenic-ferrihydrite composites precipitated from a 0.236 mol L-1 Fe3+ solution at (a) pH 3.5, (b) pH 5.5, and (c) pH 7.0. The arrow indicates a broadening of the major ferrihydrite reflection resulting from distortion of the surface structure of the ferrihydrite crystallites as a result of arsenate adsorption during crystallization.
FIGURE 3. Powder XRD patterns of arsenic-ferrihydrite composites precipitated from an Fe3+ solution at various supersaturation ratios (S). (a) S ) 7.8 × 103, (b) S ) 1.6 × 104, (c) S ) 8.3 × 104, and (d) S ) 1.3 × 105. In each case, the pH was 5.5, and the Fe/As molar ratio was 12.
are shifted toward higher supersaturation ratios at higher pH. That is to say, a higher supersaturation ratio is required to achieve a given level of arsenic removal at pH 7 than would be required to achieve the same level of removal at pH 3.5. These results provide important guidelines for appropriate levels of iron(III) required for arsenic removal under various circumstances. For example, while 99.9% removal of arsenic can be achieved with a supersaturation ratio of 103 at pH 3.5, we can see that the supersaturation ratio must be increased by 2 orders of magnitude to achieve similar removal at pH 7. In cases where arsenic is to be removed from mineral processing liquors, precipitation at pH 3.5 is feasible, but for purification of drinking water, it is desirable to keep the pH close to 7. However, if arsenic levels in the water being treated are already low, the degree of arsenic removal required to reduce arsenic to safe levels may be considerably less than 90%. Thus, curves of the type shown in Figure 3 provide a useful guide to the precipitation conditions required to achieve the desired final arsenic concentration across a range of pH values. Structural Characteristics of Arsenical Ferrihydrite Solids. To examine the effects of pH on the structure of the arsenic-loaded ferrihydrite precipitates, we compare in Figure 2 the XRD patterns of samples formed at different pH but with the same Fe3+ concentration. It should be noted that changes in pH over the range presented here correspond to a change in supersaturation with respect to ferrihydrite, by virtue of the pH dependence of ferrihydrite solubility, and the estimated values of the supersaturation ratio are given in Table 1. Thus, the decrease in supersaturation with pH would be expected to result in ferrihydrite precipitates displaying increasing structural order. Such a trend has indeed been observed for ferrihydrite precipitates formed in the absence of arsenic (11). In the XRD patterns presented in Figure 2, however, the opposite trend is observed, with somewhat broader peaks being evident in the patterns obtained at lower pH. We have
already noted that arsenic removal is more rapid and efficient at low pH, and we attribute this to the more positive surface charge of the ferrihydrite phase as pH is decreased. Thus, the arsenate is strongly attracted to and binds rapidly at the surface of primary ferrihydrite nuclei as they form. This rapid incorporation of arsenate might be expected to lead to less ordered structures within the ferrihydrite crystallites and may also limit the extent of aggregation and thus hinder the development of any long-range order within aggregates. A noticeable feature of these XRD patterns is the broad shoulder (indicated by the arrow in Figure 2) appearing at a slightly lower 2θ value on the edge of the major reflection of the 2-line ferrihydrite pattern. Current literature provides two opposing theories on the origin of this peak: Carlson et al. (6) suggest that this peak is due to the precipitation of a disordered iron(III) arsenate phase, while Rancourt et al. (15) postulate that the peak is due to a ferrihydrite phase containing high levels of adsorbed arsenate and is made up of extremely fine-grained particles. Broadening and shift of the major reflection can thus be attributed to the rapid adsorption of arsenate on the primary ferrihydrite crystallites retarding crystal growth, leading to distortion of the surface structure of ferrihydrite crystallites, and as the surface-tovolume ratio is extremely high in a nanoparticulate sample, this is readily observed in the XRD pattern. From the XRD patterns examined in our experiments, this interpretation seems to be the best explanation for the shoulder on the main ferrihydrite peak, as this hump also occurs in the patterns of non-arsenical ferrihydrite, although it is more noticeable in the XRD patterns of arsenical ferrihydrite. Figure 3 shows the effect of varying the initial Fe(III) concentration, and thus the supersaturation ratio, on the structural order of the arsenical ferrihydrite precipitates. In all samples shown in Figure 3, the product was precipitated at pH 5.5, and the Fe/As ratio was fixed at 12, so although the iron concentration has been varied, the composition of the products in terms of arsenic content does not vary a great deal. Indeed, elemental analysis of the solids showed
2370
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 8, 2004
FIGURE 4. TEM images of ferrihydrite product precipitated at high supersaturation ratio. (a) Well-dispersed individual ferrihydrite particles from a sample precipitated at pH 7.0, corresponding to the XRD pattern shown as c in Figure 2. (b) Large ferrihydrite aggregate seen in the same sample. (c) Magnified view of a similar aggregate showing the individual crystallites of ∼5 nm diameter.
FIGURE 5. TEM images of ferrihydrite product precipitated at low supersaturation ratio (corresponding to XRD pattern a in Figure 3). (a) A group of large spherical aggregates with a schwertmannite-like morphology. (b) Higher magnification image showing the whiskers composed of individual ferrihydrite crystallites ∼5 nm in diameter. the arsenic content of all samples fell in the range 39-43 mg/g. Changes in the degree of crystallinity in these samples are therefore due to the change in the driving force for precipitationsat higher supersaturation ratios, the precipitation rate is higher and the product less crystalline. Conversely, as the driving force for precipitation is slowed (i.e., by reducing the initial Fe(III) ion concentration), an increase in crystal order is observed. These results are in accord with our studies of ferrihydrite precipitation in the absence of arsenate anion (9, 11). The differences in crystallinity are most obvious when the major ferrihydrite reflections in each spectrum are compared. This broad reflection becomes progressively sharper as the Fe(III) ion concentration is decreased, which seems to indicate that as the supersaturation ratio is decreased, the ordering of Fe ions in the lattice is improving as a result of the lower rate of iron hydrolysis, regardless of the presence of arsenate. The extra reflections observed in spectrum a can be considered as indicative of a greater degree of threedimensional order within the ferrihydrite lattice (10), and indeed, this pattern resembles that of schwertmannite (16), a poorly ordered iron oxyhydroxide containing sulfate anions
as a structural component. Given that the coprecipitation experiments were carried out using an iron(III) sulfate solution, the formation of schwertmannite is not unexpected, and this possibility is discussed in more detail next. TEM of Arsenical Ferrihydrite Solids. The arsenical ferrihydrite precipitates were examined by TEM to investigate whether the conditions of supersaturation had any observable effect on the morphology or aggregation behavior of the products. There did not seem to be any noticeable effects of pH changes for samples produced at similar levels of supersaturation; however, some changes in sample morphology were observed between samples produced at the extremes of the supersaturation range we studied. In Figure 4, three images are presented showing typical characteristics of the arsenical ferrihydrite product precipitated under conditions of high supersaturation. In these samples, the amorphous carbon film was generally covered in rather evenly dispersed individual ferrihydrite particles of up to 5 nm diameter, as shown in Figure 4a. Aggregates of the type shown in Figure 4b were also observed, and these appeared to be composed of the same 5 nm crystallites as were observed in the dispersed form. A general trend was VOL. 38, NO. 8, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2371
apparent that seemed to suggest that the lower the level of supersaturation, the more the particles tended to form aggregates. In samples prepared under the lowest supersaturation conditions (i.e., SEst < ∼3000: toward the bottom left corner of Table 1), some aggregates of the type shown in Figure 4 were observed, but a significant proportion of the samples were made up of very large aggregates with the pin-cushion or hedgehog morphology known to be typical of schwertmannite (16). Such an aggregate is shown in Figure 5a, with a high magnification view of the outer whiskers of the aggregate presented in Figure 5b. The aggregates do not seem to be made up of rodlike single crystals as is usually observed for schwertmannite but rather seem to be composed of individual ferrihydrite crystallites of ∼5 nm diameter aligned to form irregular whiskers. Although the morphology closely resembles that of schwertmannite, selected area electron diffraction patterns of these aggregates gave only the two diffuse rings characteristic of 2-line ferrihydrite. This is surprising, given that the powder XRD pattern of this sample (Figure 3a) showed clear signs of reflections attributable to schwertmannite, but the combined XRD and TEM results would seem to indicate that both the bulk sample and the individual aggregates contain a mixture of 2-line ferrihydrite and schwertmannite, so it is possible that the large aggregates represent some form of precursor to the typical schwertmannite morphology. Indeed, it has been our experience in similar systems that XRD patterns of the type shown in Figure 3 can be accurately modeled as mixtures of ferrihydrite and schwertmannite (11). The differences we have observed in product morphology cannot be attributed to differences in arsenate content, as we have maintained a constant Fe/As ratio, so we conclude that formation of schwertmannite, albeit as a minor component of the bulk precipitate, is favored by conditions of low supersaturation. Recent evidence from electron nanodiffraction experiments (17) has indicated that schwertmannite contains a significant amount of a cubic (maghemitelike) substructure that is also found in 2-line ferrihydrite. In light of this work, a possible interpretation of our results is that lower supersaturation promotes formation of a higher proportion of this cubic substructure and that the development of the schwertmannite hedgehog morphology arises as a consequence. The presence of schwertmannite is also in accord with the slightly lower levels of arsenic removal observed at lower supersaturation, given that schwertmannite is not known to incorporate arsenate structurally, although arsenate shows a high affinity for schwertmannite surfaces (6). In general, the results of this study suggest that adsorption of arsenate onto the surface of primary ferrihydrite nuclei is very rapid with respect to the rate of growth or aggregation processes, such that rapid hydrolysis under conditions of high supersaturation provides a product with a high surface
2372
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 8, 2004
area, and thus the capacity for higher levels of arsenate adsorption. However, the overall efficiency of arsenic removal appears to involve a combination of both supersaturation and pH effects, with pH controlling the affinity of arsenate for the ferrihydrite surface, and the supersaturation controlling the surface area and physical properties of the ferrihydrite product.
Acknowledgments The authors thank Mr. T. Upson for technical assistance. This research has been supported under the Australian Government’s Cooperative Research Centre (CRC) Program, through the A. J. Parker CRC for Hydrometallurgy. M.L. thanks the Australian Research Council and Pasminco Ltd. for support in the form of an Australian Postgraduate Award (Industry).
Literature Cited (1) Jambor, J. L.; Dutrizac, J. E. Chem. Rev. 1998, 98, 2549-2585. (2) Riveros, P. A.; Dutrizac, J. E.; Spencer, P. Can. Metall. Q. 2001, 40, 395-420. (3) Garcia-Sanchez, A.; Alvarez-Ayuso, E.; Rodriguez-Martin, F. Clay Miner. 2002, 37, 187-194. (4) Robins, R. G.; Nishimura, T.; Singh, P. In BUET-UNU Workshop on Technologies for Removal of Arsenic from Drinking Water; Ahmed, F., Ali, A., Adel, Z., Eds.; Dhaka, Bangladesh, 2001. (5) Robins, R. G.; Glastras, M. V. Res. Dev. Extr. Metall. 1987, 51, 223-229. (6) Carlson, L.; Bigham, J. M.; Schwertmann, U.; Kyek, A.; Wagner, F. Environ. Sci. Technol. 2002, 36, 1712-1719. (7) Robins, R. G.; Wong, P. L. M.; Nishimura, T.; Khoe, G. H.; Huang, J. C. Y. Randol Gold Forum 1992, 197-200. (8) Waychunas, G. A.; Rea, B. A.; Fuller, C. C.; Davis, J. A. Geochim. Cosmochim. Acta 1993, 57, 2251-2269. (9) Loan, M.; Parkinson, G. M.; Newman, O. G. M.; Farrow, J. B. J. Cryst. Growth 2002, 235, 482-488. (10) Schwertmann, U.; Friedl, J.; Stanjek, H. J. Colloid Interface Sci. 1999, 209, 215-223. (11) Loan, M. Unpublished results. (12) Krause, E.; Ettel, V. A. In International Symposium on Crystallization and Precipitation; Strathdee, G. L., Klein, M. O., Melis, L. A., Eds.; Pergamon Press: Saskatoon, Saskatchewan, 1987; pp 195-210. (13) Krause, E.; Ettel, V. A. In Impurity Control and Disposal; Oliver, A. J., Ed.; Canadian Institute of Mining, Metallurgy, and Petroleum, 1985; pp 5, 1-5, and 20. (14) Cornell, R. M.; Schwertmann, U. The Iron Oxides, 2nd ed.; VCH: Weinheim, 2003. (15) Rancourt, D. G.; Fortin, D.; Pilcher, T.; Thibault, P.-J.; Lamarche, G.; Morris, R. V.; Mercier, P. H. J. Am. Mineral. 2001, 86, 834851. (16) Bigham, J. M.; Carlson, L.; Murad, E. Mineral. Magn. 1994, 58, 641-648. (17) Loan, M.; Hart, R. D.; Cowley, J. M. Am. Miner. (in press).
Received for review November 26, 2003. Revised manuscript received February 5, 2004. Accepted February 16, 2004. ES0353154
