Balance between Surface Complexation and Surface Phase

---

Langmuir 1998, 14, 825-831

825

Balance between Surface Complexation and Surface Phase Transformation at the Alumina/Water Interface Erkki Laiti,† Per Persson, and Lars-Olof O ¨ hman* Department of Inorganic Chemistry, Umeå University, S-901 87 Umeå, Sweden Received April 14, 1997. In Final Form: November 25, 1997 This paper synthesizes and expands on the results of a recent series of investigations aimed at characterizing the interactions of orthophosphate, phenylphosphonate, and clodronate ions with hydrous alumina surfaces. The paper shows that γ-Al2O3 is a thermodynamically unstable substance in water, which undergoes a (surface) phase transformation into bayerite, β-Al(OH)3. Furthermore, it also shows that while phenylphosphonate ions are exclusively adsorbed via surface complexation to the alumina surfaces, clodronate ions dissolve the alumina phase and precipitate as an aluminum clodronate phase. Orthophosphate ions show a transient behavior in this respect, and the limits for, and consequences of, AlPO4(s) formation are determined via a series of chemical modeling calculations. The paper finally shows that, with respect to phenylphosphonate surface complexation, care must be taken when macroscopically derived stoichiometric compositions are used to assign microscopic surface complex structures.

Introduction In a recent series of investigations at this department, the properties of alumina/water interfaces have been studied.1-4 These investigations have included work with γ-Al2O3 and boehmite (γ-AlOOH) and cover the characterization of their aqueous stability, the acid/base properties of their hydrated surfaces, and studies of their interactions with phenylphosphonate, orthophosphate, and clodronate (dichloromethylenebis(phosphonate)) ions. Potentiometric titrations combined with adsorption measurements, that is, -log [H+] and nonbound ligand data, have been used as basic experimental techniques. Also FTIR and FT-Raman spectroscopy have been applied to validate the interpretations made from the macroscopic data and to provide structural information about the systems studied. The leading idea in these investigations has been to collect precise equilibrium data under well-controlled conditions and, if possible, to model these data in terms of surface complexation. During the course of the investigations it has, however, been recognized that the simple approach of modeling sorption processes as a formation of surface complexes can result in a misconception of the actual sorption process. In fact, several of our studies have shown that, in addition to surface complexation, phase transformations with the formation of new phases can, partially or exclusively, control the sorption process. This is an important point, since the general applicability of a macroscopically derived model definitely depends on how good the agreement is with the actual molecular process. In the present paper we will discuss the most important findings of our studies with alumina phases. The surface complexation results have been separately reported in our earlier papers.1-3 Thus, in the present paper, it is not * To whom correspondence should be addressed. † Present address: MoDo Research and Development, S-891 80 O ¨ rnsko¨ldsvik, Sweden.

(1) Laiti, E.; O ¨ hman, L.-O.; Nordin, J.; Sjo¨berg, S. J. Colloid Interface Sci. 1995, 175, 230. (2) Laiti, E.; O ¨ hman, L.-O. J. Colloid Interface Sci. 1996, 183, 441. (3) Laiti, E.; Persson, P.; O ¨ hman, L.-O. Langmuir 1996, 12, 2969. (4) Persson, P.; Laiti, E.; O ¨ hman, L.-O. J. Colloid Interface Sci. 1997, 190, 341.

our objective to give any detailed descriptions of experimental methods or data treatment procedures. The aim is to provide a more general perspective and to emphasize the qualitative similarities and differences observed between the studied systems. Some of the data presented in this paper have thus been reported earlier.1-4 However, in the present paper we will present new data on the thermodynamic instability of γ-Al2O3 suspended in water. New data are also reported with respect to the interactions between alumina and the clodronate ion. Finally, we present results from an extensive series of model calculations with regard to the interplay between surface complexation and AlPO4(s) formation at the aged γ-Al2O3/water interface. Experimental Section Chemicals and Analysis. In this study γ-Al2O3, manufactured by Sumitomo Chemical Co., Ltd., was used. The material is of high chemical purity, 99.995%, and has a surface area (N2 BET) of 140 m2/g. Before use, the suspensions prepared from this material were allowed to age in CO2-free 0.1 M NaCl medium for more than 1 month at 25 °C. In the preparation of clodronate solutions, disodium dihydrogen clodronate tetrahydrate (Na2C(Cl)2(PO3H)2‚4H2O) (Leiras Oy, p.a.) was used. The preparation and analysis of all other solutions followed the procedures described in ref 1. The analysis of aqueous clodronate relied on the molybdenum blue method5 after a pretreatment with potassium peroxodisulfate for 2 h at 120 °C to oxidize it to phosphoric acid. Apparatus. FTIR spectra were obtained with a Perkin-Elmer 2000 FTIR spectrometer equipped with a DTGS detector. The dry solid samples were recorded as diffuse reflectance (DR) spectra, using a Harrick diffuse reflectance unit, from 2 wt % mixtures with finely powdered KBr (Merck, IR spectroscopic grade). The samples were mixed very gently with KBr in an agate mortar and were not exposed to any elevated pressures. KBr was also used as background, and the spectra, which were the average of 500 scans, were all converted to Kubelka-Munk units. The spectrophotometer used in the analysis of phosphate was a Shimadzu UV-2100, and the XRD powder diffractogram was collected on a Rigaku, Geiger-flex, instrument. Sample Preparation. The clodronate/alumina samples were prepared by adding a sodium clodronate solution to a suspension (5) Vogel, A. I. Vogel’s Textbook of Quantitative Inorganic Analysis; Longman Inc.: London, 1987.

S0743-7463(97)00383-1 CCC: $15.00 © 1998 American Chemical Society Published on Web 01/24/1998

826 Langmuir, Vol. 14, No. 4, 1998

Figure 1. Diffuse reflectance FTIR spectra of bayerite (top) and aged γ-Al2O3 (bottom) after spectral subtraction of pure γ-Al2O3. The ordinate scale is in Kubelka-Munk units and is arbitrary. of the aged solid material to obtain the proper ligand-to-solid ratio. Thereafter, the solution pH was adjusted to 5.0 by the addition of HCl, and the suspension was left to equilibrate under stirring. Samples were collected after various periods of equilibration and the liquid-to-solid separation was accomplished by pressure filtration (Schleicher & Schuell, membrane filter 0.45 µm). Aqueous clodronate concentrations were determined as described above, and DR spectra of the solids were collected as soon as the samples were sufficiently dry to be mixed with KBr (∼1 h). This drying time leaves a significant amount of water on the solid surfaces, as evidenced from the stretching and bending vibrations of water molecules in the IR spectra.

Phase Transformation of Water-Suspended γ-Al2O3 As the aim of our studies has been to characterize surface reactions on alumina phases under equilibrium conditions, a vital prerequisite has been that the phases themselves are stable in aqueous suspension. In an early test for reversibility and reproducibility of the acid/base reactions of freshly suspended γ-Al2O3, it was noticed that these proton adsorption/desorption properties changed with time.1 Only after aging the suspensions for 1 month or more were reproducible properties registered. It was also found that FTIR data recorded for freshly prepared, and aged, γ-Al2O3 showed distinct differences. This finding suggested that the observed changes in the acid/base properties with time for freshly prepared suspensions were related to some kind of structural change of the particles. Originating from a different ongoing project, a product consisting of needle-shaped crystals was obtained from a suspension in which equimolar solutions of aqueous AlCl3 and NaOH had been mixed in a 1 to 3.5 ratio and aged for an extended time. It was found that the diffuse reflectance FTIR spectrum of this precipitate matched the spectrum obtained when subtracting the spectrum of dry γ-Al2O3 from that of water-suspended-and-aged γ-Al2O3 (cf. Figure 1). Due to the crystalline nature of this grown precipitate, it could also be analyzed by means

Laiti et al.

of X-ray powder diffraction. This characterization showed that all appearing diffraction lines were consistent with those of bayerite (β-Al(OH)3).6 By analogy, it was therefore concluded that bayerite was present also in the aged suspensions and that, consequently, γ-Al2O3 is not thermodynamically stable in aqueous medium but undergoes a phase transformation. Support for these findings can also be found in work done by Dyer et al.7 Their XRD data of γ-Al2O3 exposed to water for 4 months showed not only the powder diffraction characteristics of γ-Al2O3 but also signs of bayerite. Although our data cannot reveal whether the bayerite formation takes place at the γ-Al2O3 particle surfaces or if new pure β-Al(OH)3 particles are being formed, the observation that the surface acid/base properties of the material become constant after about 1 month of aging makes it more reasonable to assume that the phase transformation of γ-Al2O3 is connected to a bayerite layer formation on the particle surfaces. Also the fact that the specific area of the material, 140 m2/g, stayed practically invariant during aging, gives support for this hypothesis. This interpretation does not exclude the possibility that the ultrafine particle fraction of γ-Al2O3 is completely transformed into bayerite. In this paper the surface of the aged γ-Al2O3 is referred to as a bayerite surface, even though the core of most particles probably consists of γ-Al2O3 (cf. ref 7). In contrast to these observations, the surfaces of boehmite (γ-AlOOH) were indicated to be stable in aqueous suspension.2 No changes in acid/base properties with aging time were detected. Neither did any changes in the diffuse reflectance FTIR spectrum appear during the aging. Surface Complexation on Bayerite and Boehmite The first requirement for the use of a surface complexation model to describe sorption processes is that the adsorbates react with a restricted number of specific surface sites. Accordingly, pure surface complexation models cannot be applied if surface precipitation occurs, that is, if the sorption involves a reaction between the adsorbates and dissolving and reprecipitating metal ions of the solid. Good examples of true surface complexation are the adsorption of phenylphosphonate on bayerite and boehmite. Diffuse reflectance FTIR spectra recorded for these systems did show that adsorbed phenylphosphonate ions were structurally different from those of a synthesized aluminum phenylphoshonate precipitate.4 Moreover, IR and Raman spectra of sorbed phenylphosphonate ions varied in both band intensities and band frequencies with changing pH. This feature would not be anticipated if the sorption was caused by the formation of a new threedimensional aluminum phenylphosphonate precipitate. Further support for the occurrence of surface complexation was given by adsorption measurements which indicated that maximum phenylphosphonate adsorption was related to the proton-exchange capacity of the alumina surfaces.1,2 These observations were taken to confirm that surface complexes were formed and, hence, that the use of a surface complexation approach was valid. The specific surface sites active in complexation are surface hydroxyl groups which may interact with anions under outer-sphere electrostatic interaction or via innersphere ligand exchange.8 These hydroxyl sites also (6) Rothbauer, R.; Zigan, F.; O’Daniel, H. Z. Kristallogr. 1967, 125, 317. (7) Dyer, C.; Hendra, P. J.; Forsling, W.; Ranheimer, M. Spectrochim. Acta Part A 1993, 49, 691.

Surface Complexation and Surface Phase Transformation

interact with H+, and their concentration can be determined by a proton saturation experiment. For bayerite, this analysis gave a proton adsorption capacity of 0.24 mmol/g of solid (1.0 sites/nm2),1 and for boehmite, it gave one of 0.51 mmol/g of solid (1.7 sites/nm2).2 When the phenylphosphonate adsorption capacities were ratioed against these values, the two surfaces were shown to be significantly different. On the bayerite surface, the maximum amount of adsorbed ligand coincided with the total amount of tAlOH sites, while on the boehmite surface, the ligand adsorption capacity corresponded to only about half of the amount of tAlOH sites. The evaluation of potentiometric and adsorption data, which relied on the use of the so-called constant capacitance model,9 therefore resulted in equilibrium models containing complexes of different ligand/tAlOH stoichiometries for the two materials. The H+/bayerite/phenylphosphonic acid system1 was found to be best explained with an equilibrium model containing three surface complexes formed according to the stoichiometric reactions:

In the H+/boehmite/phenylphosphonic acid system,2 a formation of two 1:2 (H2L/tAlOH) complexes provided the best explanation for the data:

It is clear that, by considering only these ligand to proton active site stoichiometries, a formation of monodentate surface complexes at the bayerite surface and bridging surface complexes at the boehmite surface would seem to be a natural structural interpretation. However, a given ligand/proton active site stoichiometry of a surface complex does not necessarily bear implications with respect to the actual binding mode. In fact, several strong indications were found to suggest that the binding mode was monodentate in both systems. Perhaps the strongest indication was provided by the IR and Raman measurements. These showed that the adsorbed phenylphosphonate ions, on both the bayerite and the boehmite surface, were involved in protonation/ deprotonation reactions.4 Since the phenylphosphonate ion has only two proton active oxygen donors and since it forms inner-sphere complexes, these protonation/ deprotonation effects imply that, at least in the protonated complex, only one of the oxygen atoms is coordinated to the surface, i.e., a monodentate coordination. (8) Westall, J. C. In Geochemical Processes at Mineral Surfaces; Davis, J. A., Hayes, K. F., Eds.; American Chemical Society: Washington, DC, 1986; p 54. (9) Morel, F. M. M.; Yeasted, J. G.; Westall, J. C. In Adsorption of Inorganics at Solid-Liquid Interfaces; Anderson, M. A., Rubin, A. J., Eds.; Ann Arbor Science: Ann Arbor, MI, 1981; p 263.

Langmuir, Vol. 14, No. 4, 1998 827

Figure 2. Modeled distribution of phenylphosphonic acid between the surfaces of γ-AlOOH (dotted line), aged γ-Al2O3 (solid line), and the aqueous phase (dashed line) as a function of -log [H+]. The diagram was calculated for conditions described in the text.

A second, more indirect, indication for suggesting a monodentate coordination also for the deprotonated complex on the boehmite surface was found when attempts were made to propose a tentative bidentate structure for the species:

Due to the given stoiciometric constraint, these attempts all ended up in chemically unrealistic constructions. Our third piece of information pointing toward a conclusion of an equivalent coordination mode on both surfaces comes from the similarity in thermodynamic stability. Figure 2 shows the result of a model calculation in which surfaces of bayerite and boehmite were simultaneously allowed to compete for a restricted amount of phenylphosphonate ions. The relative amounts of the two surfaces were chosen so that their maximum phenylphosphonate binding capacities were equal, and the total amount of phenylphosphonate was set to correspond to 50% of the sum of these capacities. As seen in this figure, the two surfaces display very similar phenylphosphonate affinities. This strong similarity would not be expected if the binding modes in the two systems were different. On the basis of these indications, we have concluded that the main phenylphosphonate complexes on both surfaces are structurally similar and, furthermore, that the phenylphosphonate ions form predominantly monodentate complexes. To explain the lower relative adsorption capacity on boehmite, it might be proposed that it either is due to ligand-ligand repulsion and thus a steric effect or due to variation in the ratio of sites which display proton activity and sites which are Lewis acceptors toward phenylphosphonate. The monodentate interpretation might also be supported by some pure geometric considerations. Thus, while the interoxygen distance in a phosphonate moiety is close to 2.5 Å,10 the closest interaluminum distance in aluminum (hydr)oxides is approximately 2.9 Å.6,11 It can therefore be anticipated that the oxygen-oxygen distance in the phosphonate ion is possibly too short to allow for a formation of bidentate complexes via bridging between two neighboring tAlOH sites. (10) Lyxell, D.-G.; Strandberg, R. Acta Crystallogr. 1988, C44, 1535. (11) Christoph, G. G.; Corbato, C. E.; Hofmann, D. A.; Tettenhorst, R. T. Clays Clay Miner. 1979, 27, 81.

828 Langmuir, Vol. 14, No. 4, 1998

Laiti et al.

Figure 3. Percentage of nonsorbed clodronate as a function of reaction time. The suspensions were prepared to contain [clodronate]tot/[tAlOH]tot ) 2.0 and a slightly acidic starting -log [H+].

Sorption by a Phase Transformation Process In view of the findings made above, it was found of interest to investigate whether a ligand containing two phosphonate moieties, like the clodronate ion (O3PC(Cl2)PO3, L4-), would form a bridging complex by coordinating the two phosphonate groups to neighboring surface active sites. The interactions between clodronate and bayerite surfaces were monitored by means of batch experiments and FTIR spectroscopy. As in our study of H+/bayerite/ orthophosphate interactions,3 the investigation was initiated with an experiment in which clodronate removal from the aqueous phase was followed as a function of time in a slightly acidic suspension at excess ligand ([clodronate]tot/ [tAlOH]tot ) 2). The result of these measurements is presented in Figure 3 and shows that, after about 6 days of equilibration, no remaining ligand could be found in the aqueous phase. Furthermore, during the course of this time, there were no indications of plateau formation, which could indicate a possible metaequilibrium state of surface complexation. Similar results were also obtained from other batch experiments at both higher and lower [clodronate]tot/[tAlOH]tot ratios. For instance, at a 4-fold excess of clodronate over tAlOH, only minor traces of the ligand could be recorded in the aqueous phase after an equilibration time of 2 weeks. This behavior significantly differed from that of phenylphosphonate, which, when added in excess over tAlOH, stayed time-invariant in the aqueous phase, and was taken to indicate that a phase transformation, with aluminum clodronate precipitating, was controlling the removal of the ligand. To verify this macroscopically derived conclusion, a series of FTIR measurements were performed. The results from these experiments are presented in Figure 4 and show that the IR spectra of sorbed clodronate exhibit a high resemblance to the spectrum of an aluminum clodronate precipitated from aqueous solution. This observation is in good agreement with the conclusion that a new aluminum clodronate solid phase is formed in aqueous bayerite/clodronate suspensions. The spectrum collected at [clodronate]tot/[tAlOH]tot ) 4 does, however, also contain an additional minor band at 1206 cm-1. It is not unreasonable to assume that this band originates from minor amounts of a specific surface complex. As our previous studies of phenylphosphonate complexation on alumina surfaces have shown that monophosphonate groups form surface complexes at the alumina surface, it is probable that such surface complexes are initially formed also with clodronate ions. The relatively rapid formation of the aluminum clodronate precipitate does, however,

Figure 4. Diffuse reflectance FTIR spectra of clodronate sorbed at bayerite at [clodronate]tot/[tAlOH]tot ) 4 (top) and [clodronate]tot/[tAlOH]tot ) 12.7 (middle), and DR-FTIR spectrum of aluminum clodronate(s) (bottom) precipitated from a 4:2:1 mixture of aqueous OH-, disodium clodronate, and Al3+ ions. The ordinate scale is in Kubelka-Munk units and is arbitrary.

prohibit the characterization of such a metaequilibrium state, and no further experiments were therefore conducted in the system. Systems Displaying Both Surface Complexation and Phase Transformation It has been debated whether sorption in the orthophosphate/alumina systems occurs via surface complexation or formation and precipitation of AlPO4(s).12-16 In an attempt to resolve this controversy, we performed a set of time-resolved adsorption experiments on bayerite.3 Some of the data obtained are illustrated in Figure 5 and show that a plateau in the point curvature at approximately a 1:1 ratio between adsorbed phosphate and tAlOH appears after a reaction time of a few hours. The samples equilibrated for extended times (weeks to months) did, however, show a continuing significant decrease in the aqueous phosphate content. Our interpretation of these data was that the phosphate removal from the aqueous phase was initially controlled by a relatively fast formation of surface complexes with 1:1 (H3L/tAlOH) stoichiometries but that the original alumina phase slowly transformed into a secondary aluminum phosphate precipitate with time. This interpretation was also supported (12) Muljadi, D.; Posner, A. M.; Quirk, J. P. J. Soil Sci. 1966, 17, 212. (13) Ferguson, J. F.; King, T. J.sWater Pollut. Control Fed. 1977, 49, 646. (14) Veith, J. A.; Sposito, G. Soil Sci. Soc. Am. J. 1977, 41, 870. (15) Goldberg, S.; Sposito, G. Soil Sci. Soc. Am. J. 1984, 48, 772. (16) Van Riemsdijk, W. H.; Lyklema, J. J. Colloid Interface Sci. 1980, 76, 55.

Surface Complexation and Surface Phase Transformation

Figure 5. Percentage of nonbound phosphate, %L(aq), as a function of reaction time. The suspensions were prepared to contain [H2L-]tot/[tAlOH]tot ) 2.0 at a slightly acidic -log [H+].

by diffuse reflectance FTIR measurements.3 The spectra of short-term equilibrated samples showed a significant -log [H+] dependence with respect to the positions of the P-O stretching bands, a feature which diminished and even disappeared in samples equilibrated for extended periods of time at higher phosphate to tAlOH ratios. Furthermore, the IR measurements revealed that the rate of phase transformation was highly dependent on the phosphate to tAlOH ratio. Thus, IR samples prepared at a 4:1 [H2PO4-]tot/[tAlOH]tot ratio, showed signs of AlPO4(s) bands even after very short equilibration times (30 min). On the other hand, samples containing a 1:1 phosphate to surface site ratio showed only weak signs of phase transformation after 1 week of equilibration, and samples prepared at 0.25:1 ratio showed no signs of phase transformation at all, even after several weeks of equilibration. To enable a characterization of the initial surface complexation reactions, with the lowest possible effects from the surface transformation, an equilibration time of 5 h was regarded as most appropriate (cf. Figure 5A). Furthermore, the investigated data range was restricted to [H2PO4-]tot/[tAlOH]tot e 1.5. The experimental data consisted of potentiometrically determined -log [H+] values and corresponding nonadsorbed ligand concentrations. The “best” explanation to these data was given by a model containing three monodentate surface complexes, tAlLH2, tAlLH-, and tAlL2-. The complexation was found to be strong at -log [H+] < 7, while a gradual increase above this value caused the ligand to desorb from the surface (cf. Figure 6). It should be emphasized that, due to the restricted equilibration time, this model does not describe full thermodynamic stability but a metaequilibrium state after 5 h. The remaining drift in -log [H+] (0.04 units/h) was, however, very low at this point, and

Langmuir, Vol. 14, No. 4, 1998 829

Figure 6. Distribution diagrams (distribution of phosphate, FL, versus -log [H+]) calculated at the [H2L-]tot/[tAlOH]tot ratios 1.00 and 0.25 (parts A and B, respectively) at a solid concentration of 20.0 g/dm3.

this makes it possible to conclude that the model is close to complexation equilibrium. Thermodynamic Simulations The findings presented above made it interesting to analyze how the experimental conditions, that is, the pH and [H2PO4-]tot/[tAlOH]tot ratio, could affect the thermodynamic balance between surface complexation and aluminum phosphate formation and under which conditions the surface complexes predicted by the model are thermodynamically stable toward AlPO4(s) formation. We carried out this analysis as a series of model calculations with the computer program SOLGASWATER,17 in which surface complexation, aluminum hydrolysis, and aluminum hydroxide/phosphate precipitation were simultaneously considered. The thermodynamic data used in these calculations1,3,18-20 are given in Table 1. As seen in this table, the surface active sites are treated as a separate component by the program. This implies that, to simulate dissolution/reprecipitation reactions, an “extra” component describing the “soluble/insoluble” aluminum chemistry is added. This also implies, somewhat unrealistically, that the concentration of tAlOH stays invariant in the simulations, independent of solubilization and phase transformation reactions. In the first simulation, a small amount of Al3+ was “added” to a suspension containing a given [H2PO4-]tot/ [tAlOH]tot ratio. Depending on the suspension pH and this ratio, the aluminum ions precipitated as a phosphate (17) Eriksson, G. Anal. Chim. Acta 1979, 112, 375. (18) Dyrssen, D. Vatten 1984, 40, 3. (19) O ¨ hman, L.-O.; Martin, R. B. Clin. Chem. 1994, 40, 598. (20) Duffield, J. R.; Edwards, K.; Evans, D. A.; Morrish, D. M.; Vobe, R. A.; Williams, D. R. J. Coord. Chem. 1991, 23, 277.

830 Langmuir, Vol. 14, No. 4, 1998

Laiti et al.

Table 1. Thermodynamic Model Used to Describe the Four-Component System H+/tAlOH/H2L(orthophosphate)/Al3+ a reacting components species H+ tAlOH H2LAl3+ OHtAlOH2+ tAlOH3L HL2L3tAlLH2 tAlLHtAlL2AlOH2+ Al(OH)2+ Al(OH)3(aq) Al(OH)4Al(OH)3(s) AlL(s)

log β

H+

tAlOH

H2L-

Al3+

0 0 0 0 -13.775 7.51 -8.87 1.9 -6.71 -18.45 11.49 5.14 -1.82 -5 -10.3 -16.2 -22.2 -10.7 -0.15

1 0 0 0 -1 1 -1 1 -1 -2 1 0 -1 -1 -2 -3 -4 -3 -2

0 1 0 0 0 1 1 0 0 0 1 1 1 0 0 0 0 0 0

0 0 1 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 1

0 0 0 1 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1

Q

ref

0

1 -1

0 -1 -2

1 1 3 3 3 3 3 3 18 18 18 18 19 20

Figure 8. Modeled fraction of phase transformed orthophosphate as a function of pH and calculated at six different [H2PO4-]tot/[tAlOH]tot ratios.

a The equilibrium constants are valid at 25 °C and I ) 0.1 M, and Q denotes the charge of the surface species.

Figure 9. Illustration of the pH-increasing effect of AlPO4(s) formation. The change with time for suspensions within the field of AlPO4(s) precipitation was simulated. The starting points of the samples are marked with the symbol b. The dotted lines show the evolution in nonprecipitated ligand content and in pH during the AlPO4(s) formation.

Figure 7. Modeled thermodynamic stability of AlPO4(s). The shaded field denotes the conditions of pH and [H2PO4-]tot/ [tAlOH]tot under which formation of AlPO4(s) is predicted.

phase, as a hydroxide phase, or as a mixture of the two phases. The formation of aluminum phosphate in these calculations was interpreted as a sign of a thermodynamic instability of the surface complexes, and the area of this field was determined. The result of this mapping is illustrated in Figure 7 and does, in fact, show that the stable field of surface complexation (i.e. the unshaded area) is significantly increased at low [H2PO4-]tot/[tAlOH]tot ratios. Qualitatively, these calculations thus show full agreement with the IR findings reported above. Another important aspect to consider with regard to the AlPO4(s) formation is that this phase transformation gradually lowers the remaining [PO4]/[tAlOH] ratio of the suspension. Under the assumption that the concentration of alumina surface binding sites is left unaffected, which can be debated, this implies that, from a position within the field of AlPO4(s) formation, the composition of the aqueous phase will change until the limiting line is reached. As a consequence, the phase transformation will never be complete but will always leave a certain fraction of phosphate as stable surface complexes. In Figure 8 the fraction of the ligand bound as AlPO4(s) has been plotted as a function of pH at different [H2PO4-]tot/ [tAlOH]tot ratios. This diagram clearly illustrates that an increasing acidity strongly favors the phase transformation. The figure, however, also shows that the pH range under which surface complexation dominates over phase

transformation (i.e. where less than 50% has been transformed) is relatively broad at [H2PO4-]tot/[tAlOH]tot ratios of 1 and below. An additional factor, which also diminishes the fraction of phosphate susceptible for AlPO4(s) formation, is the pH-increasing effect of phase transformation. This effect can be envisaged by formulating the corresponding reaction formula valid at near-neutral pH:

tAlLH- + Al(OH)3(s) f tAlOH + AlL(s) + H2O + OH- (1) The hydroxide ions released in this reaction will predominantly react with different surface species but also contribute to an increased suspension pH. This is illustrated in Figure 9, in which a number of originally “supersaturated” suspensions have been allowed to equilibrate with Al(OH)3(s) to reach the limiting line. As seen, the [PO4]/[tAlOH] ratios at these end points are significantly higher than expected, if an invariance in pH had been assumed. In a final simulation, the sorption isotherms of phosphate ([PO4]bound/[tAlOH]tot versus log [PO4](aq)) were constructed at a series of different pH-values, including and excluding the possibility of AlPO4(s) formation, respectively. These isotherms are presented in Figure 10 and show the anticipated behavior that, if AlPO4(s) formation is mathematically excluded (dashed lines), a surface monolayer saturation is reached at some log [PO4](aq) value which is pH-dependent. If the phase transformation is considered (solid lines), however, the sorption is not restricted to the number of surface sites but drastically increases at the starting point of AlPO4(s)

Surface Complexation and Surface Phase Transformation

Langmuir, Vol. 14, No. 4, 1998 831

Conclusions

Figure 10. Phosphate sorption isotherms ([PO4]bound/[tAlOH]tot versus log [PO4(aq)]) at a number of different pH-values, including (solid lines) and excluding (dashed lines) the possibility of AlPO4(s) formation.

formation. As also seen in the figure, this phase transformation regularly occurs before a full surface saturation has been obtained, regardless of suspension pH. The drastic increase in sorption occurring at the point of phase transformation is not in full agreement with regular experimental findings.16 Such data commonly indicate that, with an increasing free sorbate concentration, a smooth transition from surface complexation to precipitation of the sorbate will occur. To accomplish this in their surface precipitation model, Farley et al.21 assumed the existence of a solid phase with variable composition, Al(OH)3(1-x)(PO4)x(s), 0 < x < 1, but another plausible explanation for the phenomenon is that it is due to a low reaction rate of the phase transformation, that is, that it is a kinetic constraint. (21) Farley, K. J.; Dzombak, D. A.; Morel, F. M. M. J. Colloid Interface Sci. 1985, 106, 226.

From the results presented in this paper, several conclusions of more general character can be drawn. The first of these is related to our finding that γ-Al2O3 is transformed in water and points to the conclusion that it is vitally important to control the stability of the solid phase itself, especially when investigating in aqueous media surface complexation properties of materials produced at high temperatures. Second, from the significantly different interaction patterns exerted by the three ligands, it can be concluded that surface complexation studies should always give consideration to the relevant process of sorption. Third, from our finding that phenylphosphonate ions most probably coordinate to both bayerite and boehmite surfaces as a monodentate ligand, although their ratios between binding capacity and proton active site capacity differ, we conclude that care must be taken when macroscopic stoichiometries are used to assign microscopic surface complex structures. This part of our work also clearly illustrates the power of combining traditional macroscopic measurements with spectroscopic information. Finally, with regard to the interactions occurring in aqueous alumina/orthophosphate suspensions, the present paper clearly illustrates that chemical modeling is of great value to conceptually understand their time-dependent behavior, although kinetic constraints might limit their detailed information. Acknowledgment. This work forms part of a program financially supported by the Swedish Natural Science Research Council. LA970383N