Environ. Sci. Technol. 2001, 35, 4753-4757
Competitive Adsorption of Phosphate and Arsenate on Goethite ZHAO HONGSHAO AND ROBERT STANFORTH* Chemical and Environmental Engineering Department, National University of Singapore, 5 Engineering Drive 2, Singapore 117576
The competitive adsorption of phosphate and arsenate on goethite was investigated to better understand the bonding mechanisms for the two ions. The anions were added both simultaneously and sequentially. When added simultaneously, the two ions were adsorbed about equally, with the total surface coverage being slightly greater than for either ion alone. When added sequentially, the extent of exchange for the first ion depended on the equilibration time before the second ion was introduceds the longer the equilibration time the greater the exchange. There is a nonexchangeable fraction for both ions that is approximately equal to the initially adsorbed amount of each ion. The results suggest a two-phase reaction on the surface, with the first phase being a rapid surface complex formation on the goethite surface, followed by the slower buildup of a surface precipitate on the adsorbed layer. The exchangeable ions are in the surface precipitate. These results are incompatible with a surface complexation model (SCM) for anion adsorption on geothite and indicate that the actual reactions are more complicated than the reaction assumed in a SCM.
Introduction Orthophosphate adsorption on hydrous iron oxides such as goethite (R-FeOOH) has been widely studied and used for developing the surface complexation model (SCM) for anions (1). This interest reflects the importance of phosphate both as a plant nutrient and as a model for other strongly adsorbed oxyanions. Despite the widespread use of a surface complexation model for describing phosphate adsorption, there are still some questions regarding the actual bonding mechanism on the surface. Most adsorption studies look at the phosphate loss-from-solution and assume a relative simple bonding mechanismseither mononuclear or binuclear bonding on the surface sites. In the SCM, these sites are assumed to be in equilibrium with solution P, i.e., a rapid desorption process is assumed. However, those studies that have looked at the reactivity of the phosphate on the surface, using isotopic exchange (2, 3), desorption, or competitive adsorption (4, 5) have shown that the phosphate reactions on the surface are more complicated than the reactions envisioned in the surface complexation model. Further, kinetic studies on the reaction have suggested at least two different reactions occurring on the surface, an initial very fast reaction followed by a much slower reaction (3, 6). * Corresponding author phone: (065)874-8045; fax: (065)872-5483; e-mail:
[email protected]. 10.1021/es010890y CCC: $20.00 Published on Web 11/15/2001
2001 American Chemical Society
Suggested explanations for the second reaction on the surface include precipitation (7, 8), diffusion into surface pores or the adsorbent matrix (5, 9, 10), formation of a solid-solution on the surface (11), or coagulation of the adsorbent particles (12). In nature, more than one anion will normally be competing for the sorption sites. Thus it makes sense to investigate the competitive adsorption between different anions as well as single ion adsorption. In addition, investigations of the competition between the anions can provide insight into the reactions occurring on the surface. Hingston et al. (4) investigated the competition between phosphate and arsenate adsorbing on goethite and found that at a given total anion concentration, the amount adsorbed was greater in the mixed ion system than in the single ion samples. He suggested that there were sites on the surface that were specific for each ion as well as some nonspecific sites on which both ions could adsorb. Manning and Goldberg (13) used a constant capacitance SCM model to describe phosphate, arsenate, and molybdate competition on goethite and gibbsite. The modeled lines were close to the measured results, but the model was not completely successful in describing the competitive behavior of the ions. Manning and Goldberg also proposed that there are some sites on the goethite surface that are specific for each ion as well as other sites that can adsorb both ions. Most competitive adsorption studies have involved the simultaneous addition of the competing ions to the solution. Violante et al. (14) and Ji (15) used sequential addition of their competing ions (P and citrate and P and fluoride, respectively) to investigate competitive adsorption. Both studies found that the order of addition affected the competition. Oxalate was less able to replace phosphate on goethite when the phosphate was added before oxalate than when the two ions were added simultaneously or when oxalate was added first (14). Likewise, phosphate was less affected by fluoride when the phosphate was added first (15). In theory, competitive adsorption between two similar ions should be described by a simple partitioning model. Phosphate and arsenate have similar deprotonation constants in solution and should have similar effects on the surface charge of the solid. Assuming a bidentate bond on the surface and an uncharged surface complex (and using the two pKa model for charge development), the competition between arsenate and phosphate at pH 5 can be written as follows
(FeO)2PO2H + H2AsO4-s ) (FeO)2AsO2H + H2PO4-s (1) where H2AsO4-s represents the arsenate concentration at the surface. The equilibrium constant for the exchange is as follows
Kexc ) {(FeO)2AsO2H} [H2PO4-]s/
{(FeO)2PO2H}[H2AsO4-]s (2)
where {} represents the activity of the surface group. Since the incoming and outgoing ions have the same charge, the electrostatic terms relating surface and bulk solution concentrations cancel out. Kexc is the ratio of the intrinsic equilibrium constants for the surface complexation reactions of phosphate and arsenate, i.e.,
Kexc ) KAs/KP VOL. 35, NO. 24, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
(3) 9
4753
Kexc ) KAs/KP ) {(FeO)2AsO2H}[H2PO4-]/
-
{(FeO)2PO2H}[H2AsO4 ] (4) {(FeO)2AsO2H}/{(FeO)2PO2H} ) (KP[H2PO4-])/
(KAs[H2AsO4-]) (5)
If the constants are the same or very similar, as appears to be the case for arsenate and phosphate based on the single ion isotherms from the results found in this study, then Kexc is approximately 1, and the ratio on the surface reflects the ratio of the dissolved ions. Note that, in theory, the total amount on the surface at a given total anion concentration cannot exceed the amount on the surface of the more strongly bound anion at the same concentration. Adsorption Kinetics. P adsorption kinetics often follows a complicated pattern, with a rapid initial adsorption followed by a slow reaction that frequently does not reach equilibrium up to the end of the experiment (3). A simplified form of the Elovich equation, comparing surface coverage with the log of time, often describes the moderately slow reaction kinetics very well (3), although at long reaction times (e.g. months) the adsorption kinetics may deviate from an Elovich eq 12. In this study the competition between arsenate and phosphate on goethite has been investigated using both simultaneous and sequential addition, measuring the competition over time.
Materials and Methods Goethite (R-FeOOH) was prepared using the procedure of Atkinson et al. (16) as described in Stanforth (3). The same stock solution of goethite was used for all experiments. The BET surface area of the goethite is 20 m2/g. SEM analysis of the goethite showed typical elongated needles, with an average length of 1.6 mm and a width of 0.095 mm. During goethite preparation and the adsorption experiments, the samples were exposed to the atmosphere. It is probable, therefore, that some carbonate adsorption occurred on the goethite, particularly during preparation (17). Since carbon dioxide and carbonate are nearly ubiquitous in the aquatic and soil environments, conducting experiments in the presence of carbonate is more realistic than experiments in carbon dioxide-free conditions. Contact of the experimental solutions with glass was minimal, and there should be little silicate contamination of the goethite (18). Three experiments were conducted as part of this study. First, the adsorption kinetics of both arsenate and phosphate at pH values of 2.45 and 5.15 were investigated. The pH values were chosen to be in the range where both phosphate and arsenate are strongly adsorbed and to provide a range of goethite solubilities (although even at pH 2.45 the solubility of iron in a goethite suspension is only 10-8 M). Next the competitive adsorption following the simultaneous addition of equal concentrations of both ions was investigated, both at pH 5.15 using total ion concentrations of 40 and 80 mM and at a variety of pH values using 80 mM total ion concentration. Finally, the effect of sequentially adding the two ions was investigated, with the second ion added at varying times after the first ion was introduced. Sequential addition experiments were run at pH 3.0 and 5.15. Samples were analyzed and pH adjusted (if the pH was more than 0.2 units off from the target value) at varying times following the start of the experiments, including 1 h, 5 h, 24 h, 72 h, 144 h, and 288 h. Adsorption experiment samples were conducted using a solids concentration of 0.45 g/L goethite and in a 1 × 10-3 M NaCl solution to “buffer” ionic strength. The goethite stock slurry was ultrasonicated for 20 min prior to sample preparation to ensure particle separation. pH adjustment, 4754
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 24, 2001
when necessary, was done using nitric acid or sodium hydroxide (0.05-1.0 M, depending on the degree of pH adjustment necessary). Samples were agitated on a rotating shaker at 18 rpm. At appropriate times, aliquots were removed and filtered using Whatman Autovial filters with a 0.45 µm pore size PTFE membrane filters. Anion concentrations were measured on a Perkin-Elmer Optima 3000 DV inductively coupled plasma emission spectrometer. Error Analysis. Replicates of the adsorption experiments were run in order to evaluate the experimental reproducibility. For the single ion adsorption results, the mean percent deviation of the replicates was 1.1% (n ) 80 pairs) for the surface coverage results, with a standard deviation of the surface coverage percent deviations of 1.2%. For the competitive adsorption experiments, the mean percent deviation of the surface coverage results was 3.2%, (n ) 190 pairs) with a standard deviation of the sample deviation of 3.3%. Some simultaneous addition experiments were done in triplicate. For these samples, the mean of the standard deviations of the replicates was 6.45% for phosphate and 2.46% for arsenate.
Results and Discussion Adsorption isotherms for both arsenate (As) and orthophosphate (P) were developed at pH 2.45 and 5.15. Both anions gave similar adsorption isotherms, particularly at lower concentrations. Arsenate was slightly more strongly adsorbed at higher concentrations. Adsorption kinetics follows a twophase pattern, with a rapid initial adsorption followed by a much slower reaction. The slow reaction follows an Elovich equation (i.e. a plot of surface coverage versus log time is linear) over the time period covered. The surface coverage for the initial, pre-Elovichian step is 35-45 µmol/g at pH 5.15, depending on the ion concentration. Adsorption of both ions did not reach equilibrium within the time frame of the experiments (up to 24 days). Based on the surface area of the goethite, and assuming 5.2 µmol/m2 surface area singly bound surface hydroxyls and a binuclear bond, the maximum surface coverage should be around 52 µmol/g. The singly bound hydroxyls (the A hydroxyls) are generally considered to be the adsorption sites for anions on the goethite surface (2, 9, 10). The highest measured adsorption values were slightly higher than this value, particularly at pH 2.43 (60 and 73 µmol/g for phosphate and arsenate, respectively). Competitive Adsorption at Different pH Values. Competitive adsorption studies for the two ions added simultaneously were conducted at pH values between 2 and 11. A total ion concentration of 80 µM was chosen, and the adsorption was measured after 48 h equilibration time. The concentration was chosen to provide near maximum surface coverage. At all pH values the surface coverage in the competitive adsorption experiments were higher than the adsorption at the same concentration of the individual ions. The difference was much more pronounced at lower pH values. For example, at pH 2.45, the amount adsorbed for 80 µM of both arsenate and phosphate was around 70 µmol/g in the individual ion samples but was 96 µmol/g in the combined sample. Hingston et al. (4) found similar behavior in their studies and attributed the greater adsorption in the combined system to adsorption on sites specific for each ion. During the competitive adsorption studies, the adsorption kinetics for each ion and for the total adsorption followed an Elovich equation. Competitive Adsorption Following Sequential Addition. Most competitive adsorption studies have added the ions simultaneously, as was done above. In the natural environment, however, it is more likely that the ions will come in contact with an adsorbing solid sequentially, i.e., the solid
FIGURE 1. Phosphate and total surface coverage versus log reaction time when arsenate was added sequentially after phosphate at pH 5. Arsenate (40 µM) was added at times ranging from 1 h to 288 h after phosphate (40 µM) addition. Surface coverages for 40 µM phosphate by itself and for the simultaneous addition of 40 µM phosphate and 40 µM arsenate are also shown. The exchangeable and nonexchangeable fractions of the phosphate surface coverage are indicated.
FIGURE 2. Arsenate and total surface coverage versus log reaction time when phosphate was added sequentially after arsenate at pH 5. Phosphate (40 µM) was added at times ranging from 1 h to 288 h after arsenate (40 µM) addition. Surface coverages for 40 µM arsenate by itself and for the simultaneous addition of 40 µM phosphate and 40 µM arsenate are also shown. The exchangeable and nonexchangeable fractions of the arsenate surface coverage are indicated. will be exposed to one ion first, with the second ion coming in contact with the solid at a later time. In simultaneous addition, the ions compete for the same sites as they are adsorbed from solution. In sequential addition, in contrast, the first ion must be desorbed before the second ion can compete. The competition, therefore, is slightly different between the two systems. In the studies reported here, the first ion was introduced and allowed to equilibrate for between 1 h and 288 h before the second ion was added. The concentrations of both ions were then measured for a further 288 h. Both ions were added at 40 µM. The results for the competition at pH 5 are presented in Figure 1 for phosphate as the initial ion. The results for arsenate as the initial ion are similar (Figure 2). Both the total ion adsorbed and the adsorbed concentration of the first ion are shown. The results
are shown on a log scale to facilitate presenting the data over a long time period and to linearize the adsorption data. After the addition of the second ion, the total surface coverage increases rapidly, reflecting the increase in the total anion concentration, following which the amount on the surface continues to increase in a fashion similar to the simultaneous addition samples. When arsenic is added as a competing ion, the total anion on the surface is the same as when phosphate and arsenate are added simultaneously. However, when phosphate is added as the competing ion, the total amount on the surface is less than when the two ions are added simultaneously. This suggests that at high concentrations (80 µM total ion concentration) arsenate is bound more strongly than phosphate. VOL. 35, NO. 24, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4755
TABLE 1. Total, Exchangeable, and Nonexchangeable Surface Coverage of As and P Added Sequentially, after 288 Hours Exchange Time sample both ions added at 40 µM ion added first
competing ion
P
As
As
P
P
As
As
P
surface coverage, µmol/g equilibration time prior to addition of competing ion, h
total
exchangeable
nonexchangeable
42 45 48 52 55 57 41 45 48 53 56 58
10 13 15 18 20 22 1 4 6 9 12 14
32 32 33 34 35 35 40 41 42 44 44 44
60 70 75 60 65 71
25 35 40 15 18 23
35 35 35 45 47 48
pH 5 1 5 24 72 144 288 1 5 24 72 144 288
pH 3 1 24 144 1 24 144
The results show that only a fraction of the initially adsorbed ion is replaced by the competing ion and that this replaceable fraction increases with the reaction time prior to the addition of the second ion. The nonexchangeable fraction remains relatively constant throughout the course of the experiment. Further, this nonexchangeable fraction is roughly equal to the amount of ion initially adsorbed in the “pre-Elovichian” step (as shown in Figures 1 and 2). The exchangeable ion is that adsorbed during the slow reaction, rather than that adsorbed initially. A similar pattern of competition is observed at pH 3 as at pH 5. The total amount adsorbed at pH 3 is greater than at pH 5, and this greater adsorption is in the exchangeable fraction. The amount of nonexchangeable ion remains relatively constant at both pH values (Table 1). The amount of nonexchangeable P is about 35 µmol/g, while the exchangeable P varies from 10 µmol/g (after 1 h) to 22 µmol/g (after 288 h) at pH 5 and 25 µmol/g (after 1 h) to 40 µmol/g (after 144 h) at pH 3. The difference between the total ion adsorption when P is added after As and when the two ions are added simultaneously is even more pronounced at pH 3 than at pH 5.15. When phosphate is added 288 h after arsenate in the pH 3 solutions, the P simply exchanges with As adsorbed on the solid, and the total surface coverage does not increase. If the concentration of the initially introduced ion is insufficient to completely occupy all the nonexchangeable sites, then none of the initially adsorbed ion is exchanged. Rather the second ion simply adsorbs without any exchange occurring. Both P and As were added at 10 and 20 µM, followed after 24 h by 40 µM of the competing ion. In both cases there is very little exchange between the initially adsorbed ion and the competing ion, even though the competing ion is present at a much higher concentration (the 10 µM results are shown in Figure 3). Both As and P behave similarly. When the initial ion is added at 10 µM (and almost completely adsorbed), a fraction of the second ion is adsorbed very quickly, with subsequent adsorption following an Elovich-type kinetcs. The Elovich kinetics starts when the surface coverage is slightly over 30 µmol/g. When the surface 4756
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 24, 2001
coverage of the first ion is close to the nonexchangeable value seen in the higher concentration studies (20 µM), very little of the second ion is adsorbed initially, but the subsequent adsorption follows an Elovich pattern. The slope of the Elovich line for the second ion is very similar in both samples. The relative amounts of As and P adsorbed in the sequential addition for the low concentration samples are different than when the two ions are added simultaneously. When the ions are added simultaneously, the amounts on the surface are close to what would be expected based on the relative concentrations of the two ions, assuming similar bonding energies. With the simultaneous addition of 10 µM P and 40 µM As, for example, the percentage of the total on the surface is 18.2% and 81.8% for P and As, respectively, compared with theoretical values of 20% and 80%. When the P is added first, in contrast, P makes up 30.1% of the total, and As only 69.1%. When the initial ion is present at 20 µM the contrast is even more apparent. For the samples in which P was added first, 52.2% on the surface coverage was from P, compared with a theoretical coverage of 33.3%. The results further demonstrate that there is a fraction of the surface sites that are very rapidly occupied but which are not exchangeable once occupied. These results indicate that there are at least two bonding mechanisms on the surface. The rapidly adsorbed ion is bound irreversibly, while the more slowly adsorbed ion is exchangeable with a competing ion introduced later. The results are inconsistent with a surface complexation model, since there are several assumptions made in the SCM that are not justified. These assumptions include that the adsorbate ions on the surface are exchangeable with dissolved species, that the adsorbed ions are bound uniformly, and that the reactions are at equilibrium. Several authors have suggested a two-phase reaction on the surface, with the second reaction frequently assumed to be a slow diffusion of the adsorbed P into the goethite (either into pores in the goethite or into the crystal itself). This model is also inconsistent with the results presented here, since in such a case it would be the initially adsorbed P that would be exchangeable and the slowly adsorbed fraction would be buried in the crystal. Likewise, a solid solution model in which the P slowly migrates deeper into the goethite “solution” is inconsistent with the results, since again it would be the surface, initially adsorbed ion that would be exchangeable. One explanation that could account for the results is surface precipitation. The initially adsorbed ion is adsorbed on the surface in a nonexchangeable form (as shown by the nonexchangeability of the low surface coverage experiment). These sites then act as sorption sites for dissolved iron. The decrease in the dissolved iron concentration and the presence of the anion in solution causes more iron to dissolve from the goethite matrix, which is also adsorbed on the surface phosphate. The sorbed iron in turn sorbs more phosphate from solution to form a surface precipitate. The amorphous iron phosphate precipitate will exchange phosphate with arsenate added later to form a mixed iron phosphate/arsenate solid. The higher adsorption values observed at pH 3 than at 5 are due to the greater solubility (and hence faster dissolution) of the goethite at the low pH. The Elovichian behavior of the slowly adsorbed phosphate is due to the slow anion-induced dissolution of the goethite rather than to the kinetics of the sorption reaction itself. Phosphate and arsenate are adsorbed equally, as shown by the similarity in the adsorption values at low surface coverages, but arsenate forms a less soluble precipitate than does phosphate, as shown by the inability of phosphate to replace arsenate or achieve the same surface coverage when added after arsenate. Dzombeck and Morel (19, 20) have presented a surface precipitation model in which the adsorbate is first adsorbed as a surface complex and then forms a surface precipitate
FIGURE 3. Surface coverage versus log time for the sequential addition of 10 µM initial ion (either phosphate or arsenate) and 40 µM of the competing ion added 24 h later. The total surface coverage for both ions is also shown. pH 5.0. as the solution becomes supersaturated with respect to the precipitating solid. Their model is different from the model presented here, in several ways. First, their model assumes that the adsorption step follows a SCM and is in equilibrium with the dissolved anion. From our results, the adsorbed surface complex does not exchange with the ions in solution. Next, their model requires supersaturation of the precipitating species. Iron phosphate (as either strengite or amorphous iron phosphate) was not supersaturated in these studies. Li and Stanforth (21) have shown that adsorption can transition into precipitation at dissolved phosphate concentrations below the saturation of iron phosphate. Finally, Dzombeck and Morel do not consider the dissolution of the solid to provide a continuous supply of cations for precipitating the anion, as is envisioned in this model. The removal of phosphate or arsenate from solution (i.e. adsorption) can be modeled quite well using several of the available empirical models, including the Freundlich, Langmuir, and SCM models. However, modeling the desorption step requires a more detailed understanding of the bonding mechanisms and kinetics. Desorption will control the mobility and long-term bioavailability of the ions in the environment. The desorption of these strongly bound anions from the goethite surface involves only a portion of the total ion on the surface and is quite slow. While the results presented in this paper shed light on the reactions occurring on the surface, further work is needed before mechanistic models of the surface reactions and their kinetics can be developed.
Acknowledgments Financial support from the National University of Singapore in the form of a research scholarship for Zhao Hong Shao and a research grant (RP279-000-033-112) is gratefully acknowledged.
(2) Atkinson, R. J.; Posner, A. M.; Quirk, J. P. J. Inorg. Nucl. Chem. 1971, 34, 2201-2211. (3) Stanforth, R. Ph.D. Thesis (Water Chemistry), U. Wisc-Madison, 1981. (4) Hingston, F. J.; Posner, A. M.; Quirk, J. P. Discuss. Faraday Soc. 1971, 52, 334-342. (5) Cabrera, F.; De Arambarri, P.; Madrid, L.; Toca, C. G. Geoderma 1981, 26, 203-216. (6) Torrent, J.; Schwertmann, U.; Barron, V. Clays Clay Miner. 1992, 40, 14-21. (7) Jonasson, R. G.; Martin, R. R.; Giuliacci, M. F.; Tazaki, K. J. Chem Soc., Faraday Trans. 1 1988, 84, 2311-2315. (8) Martin, R. R.; Smart, R. St. C. Soil Sci. Soc. Am. J. 1987, 51, 54-56. (9) Strauss, R.; Brummer, G. W.; Barrow, N. J. Eur. J. Soil Sci. 1997, 48, 101-114. (10) Torrent, J.; Barron, V.; Schwertmann, U. Soil Sci. Soc. Am. J. 1990, 54, 1007-1012. (11) Miller, J. W.; Logan, T. J.; Bigham, J. M. Soil. Sci. Soc. Am. J. 1986, 50, 609-616. (12) Anderson, M. A.; Tejedor-Tejedor, M. I.; Stanforth, R. R. Environ. Sci. Technol. 1985, 19, 632-637. (13) Manning, B. A.; Goldberg, S. Soil Sci. Soc. Am. J. 1996, 60, 121131. (14) Violante, A.; Colombo, C.; Buondonno, A. Soil Sci. Soc. Am. J. 1991, 55, 65-70. (15) Ji, G. L. Acta Pedologica Sinica 1986, 23, 220-227. (16) Atkinson, R. J.; Posner, A. M.; Quirk, J. P. J. Inorg. Nucl. Chem. 1968, 30, 2371-2381. (17) Evans, T. D.; Leal, J. R.; Arnold, P. W. J. Electroanal. Chem. 1979, 105, 161-167. (18) Furlong, D. N.; Freeman, P. A.; Law, A. C. J. Colloid Interface Sci. 1981, 80, 20-31. (19) Dzombak, D. A.; Morel, F. M. M. Surface Complexation Modeling - Hydrous Ferric Oxide; J. Wiley and Sons: New York, 1990; pp 22-25. (20) Farley, K. J.; Dzombak, D. A.; Morel, F. M. M. J. Colloid Interface Sci. 1985, 106, 226-242. (21) Li, L.; Stanforth, R. J. Colloid Interface Sci. 2000, 230, 12-21.
Received for review April 23, 2001. Revised manuscript received September 10, 2001. Accepted September 18, 2001.
Literature Cited (1) Sigg, L.; Stumm, W. Colloids Surfaces 1981, 2, 101-117
ES010890Y
VOL. 35, NO. 24, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4757
