Journal of Colloid and Interface Science 440 (2015) 282–291

Contents lists available at ScienceDirect

Journal of Colloid and Interface Science www.elsevier.com/locate/jcis

Surface complexation modeling of inositol hexaphosphate sorption onto gibbsite Maika Ruyter-Hooley a, Anna-Carin Larsson b, Bruce B. Johnson a, Oleg N. Antzutkin b,c, Michael J. Angove a,⇑ a

La Trobe University, P.O. Box 199, Bendigo, VIC 3552, Australia Chemistry of Interfaces, Luleå University of Technology, S-971 87 Luleå, Sweden c Department of Physics, Warwick University, CV7 4AL Coventry, UK b

a r t i c l e

i n f o

Article history: Received 26 September 2014 Accepted 25 October 2014 Available online 15 November 2014 Keywords: IP6 P MAS NMR Adsorption Phosphorous ECCM Organic

31

a b s t r a c t The sorption of Inositol hexaphosphate (IP6) onto gibbsite was investigated using a combination of adsorption experiments, 31P solid-state MAS NMR spectroscopy, and surface complexation modeling. Adsorption experiments conducted at four temperatures showed that IP6 sorption decreased with increasing pH. At pH 6, IP6 sorption increased with increasing temperature, while at pH 10 sorption decreased as the temperature was raised. 31P MAS NMR measurements at pH 3, 6, 9 and 11 produced spectra with broad resonance lines that could be de-convoluted with up to five resonances (+5, 0, 6, 13 and 21 ppm). The chemical shifts suggest the sorption process involves a combination of both outer- and inner-sphere complexation and surface precipitation. Relative intensities of the observed resonances indicate that outer-sphere complexation is important in the sorption process at higher pH, while inner-sphere complexation and surface precipitation are dominant at lower pH. Using the adsorption and 31P MAS NMR data, IP6 sorption to gibbsite was modeled with an extended constant capacitance model (ECCM). The adsorption reactions that best described the sorption of IP6 to gibbsite included two inner-sphere surface complexes and one outer-sphere complex:

BAlOH þ IP12 þ 5Hþ $ BAlðIP6 H4 Þ7 þ H2 O 6 B3AlOH þ IP12 þ 6Hþ $ BAl3 ðIP6 H3 Þ6 þ 3H2 O 6 2þ

10

B2AlOH þ IP12 þ 4Hþ $ ðBAlOH2 Þ2 ðIP6 H2 Þ 6

The inner-sphere complex involving three surface sites may be considered to be equivalent to a surface precipitate. Thermodynamic parameters were obtained from equilibrium constants derived from surface complexation modeling. Enthalpies for the formation of inner-sphere surface complexes were endothermic, while the enthalpy for the outer-sphere complex was exothermic. The entropies for the proposed sorption reactions were large and positive suggesting that changes in solvation of species play a major role in driving the sorption process. Ó 2014 Elsevier Inc. All rights reserved.

1. Introduction Up to 80% of total phosphorus in soils is present as organic compounds including inositol phosphates, nucleic acids and phospholipids [25]. The dominant contributors to this fraction are inositol phosphates, with myo-inositol hexaphosphate (IP6), a major form of phosphorus in plants, the most important source [21]. Together with inorganic phosphate IP6 has the potential to be an important contributor to phosphorus in aquatic ⇑ Corresponding author. E-mail address: [email protected] (M.J. Angove). http://dx.doi.org/10.1016/j.jcis.2014.10.065 0021-9797/Ó 2014 Elsevier Inc. All rights reserved.

environments through run-off from soils, where it can contribute to the formation of algal blooms. Previous work has shown that IP6 can accumulate in soils by sorbing strongly onto soil minerals such as iron and aluminum hydroxides [1,2,47,13,9,24,20,31]. Berg et al. [9] studied the sorption of orthophosphate, b-D-glucose-6-phosphate (G6P), adenosine 50 -triphosphate (ATP) and IP6 onto a number of soils. These authors found that, of the phosphates investigated, IP6 had the greatest affinity for the soils studied and also that IP6 sorption was relatively unaffected by the presence of other phosphorus sources. In acidic soils, Anderson et al. [2] investigated the sorption of IP6 and found that it decreased the sorption of orthophosphate

M. Ruyter-Hooley et al. / Journal of Colloid and Interface Science 440 (2015) 282–291

showing just how strongly it was sorbed. Giaveno et al. [20], found that IP6 sorption in highly weathered soils was linked to the presence of iron and aluminum oxides. The sorption of IP6 onto the separate soil minerals montmorillonite, boehmite and ferric oxide gel has been investigated by Anderson et al. [2], who observed greater IP6 sorption at low pH on each of the minerals. Celi et al. [15] investigated the effect of different background electrolytes on the sorption of IP6 onto goethite. In the prescence of CaCl2, IP6 sorption was shown to increase at high pH due to the precipitation of Ca3-phytate above pH 6. Conversely, in the prescence of K+ IP6 sorption was found to decrease with increasing pH. Investigating the sorption mechanism of organic species to soil minerals is important in gaining an understanding of how they will interact in soil environments. Whilst there have been a number of studies analysing and quantifying IP6 uptake in soils, relatively few of these have investigated the chemistry of the processes involved in the sorption. By comparing the relative amounts of orthophosphate and IP6 sorbed, Ognalaga et al. [47] suggested that four phosphate groups are involved in the sorption of IP6 onto goethite. From the results of electrophoretic mobility measurements, they also proposed that the sorption of IP6 to goethite probably involved the formation of inner-sphere surface complexes. Spectroscopic methods offer the opportunity to clarify the nature of the interaction between organic species and mineral surfaces. In particular, Fourier Transform Infrared Spectroscopy (FTIR) and Attenuated Total Reflection–Fourier Transform Infrared Spectroscopy (ATR–FTIR) have been used to investigate the nature of bonding between IP6 and mineral surfaces enabling differentiation between inner- and outer-sphere complexation, and providing a means to determine when precipitates are formed [24,31,58]. Guan et al. [24] studied the sorption of IP6 onto amorphous aluminum hydroxide using ATR–FTIR. They suggested that IP6 sorbs to aluminum hydroxide by inner-sphere complexation and that sorption facilitated the deprotonation of phosphate groups. From molecular orbital calculations they proposed that three phosphate groups interact with the surface. Their results supported the earlier work from Ognalaga et al. [47] and Celi et al. [14] who, using FTIR spectra, proposed that IP6 sorbs to goethite by inner-sphere complexation. Recently Yan et al. [58] studied the kinetics of sorption of IP6 to amorphous aluminum hydroxide. By use of ATR–FTIR, XRD, NMR, zeta potential and determination of hydroxide released as a function of IP6 sorption density and time, they found that IP6 was initially sorbed as an inner-shere complex, but then there was relatively rapid transformation into a bulk Al–IP6 precipitate. In contrast, Johnson et al. [31] investigated the sorption of IP6 onto goethite and found that sorption occurred largely by outersphere complexation in the pH range 4–10. From an analysis of the ATR–FTIR bands of the solution and sorbed IP6 spectra, they concluded that hydrogen bonding played a crucial role in the sorption of IP6 to goethite. For gibbsite surfaces the use of ATR–FTIR can be problematic as the Al–OH bonds in gibbsite and the phosphate groups of IP6 absorb radiation over a similar wavelength range. However, for aluminum oxides, oxyhydroxides and hydroxides solid-state phosphorus-31 magic-angle-spinning nuclear magnetic resonance spectroscopy (31P MAS NMR) can provide valuable information about the sorption process [44,30,35,54]. Johnson et al. [30], for example, investigated the sorption of phenyl phosphates and orthophosphate onto c-Al2O3. The phenyl phosphate was sorbed largely as an inner-sphere complex with an outer-sphere complex present only at high pH values, while precipitation of AlPO4 was a significant contributor to orthophosphate sorption. A similar result was found by Van Emmerik et al. [54] for the sorption of orthophosphate onto gibbsite, with a combination of surface complexation and surface precipitation dominating sorption. 31P MAS NMR

283

was also used by Yan et al. [58] in their study of the sorption of IP6 onto amorphous aluminum hydroxide. Their results suggested that inner-sphere complexation was the dominant sorption mechanism at pH 7, but as the pH decreased increasing amounts of Al–IP6 precipitate were evident. Surface complexation modeling can also be used to provide information on the surface complexes and sorption reactions that may occur during the sorption process. Despite the fact that adjustable parameters used in surface modeling can result in several models that are consistent with experimental data, surface complexation models have been applied to the sorption of a range of adsorbates onto mineral surfaces [26,19,46,10,56,4,22,48]. Rietra et al. [49], used CD-MUSIC to, model the sorption of phosphate onto goethite assuming both protonated and non-protonated bidentate inner-sphere surface complexes. Similarly, the sorption of phosphate onto ferrihydrite was also modeled by Antelo et al. [6] using both protonated and non-protonated bidentate inner-sphere surface complexes. Using the extended constant capacitance model (ECCM), Nilsson et al. [45] modeled the sorption of o-phthalate onto goethite using two outer-sphere complexes. Lackovic et al. [38], also used the ECCM, to model the sorption of citric acid onto goethite, kaolinite and illite with both inner- and outer-sphere surface complexes. While surface complexation modeling has been used to study the sorption of a number of organic species, as yet there has been no SCM study of IP6 sorption. This study investigates the sorption of IP6 to gibbsite, a major mineral in weathered soils, by both macroscopic and microscopic techniques and also surface complexation modeling. The information obtained from adsorption edges, isotherms, 31P MAS NMR and surface complexation modeling are used to predict possible surface reactions involved in the binding of IP6 to gibbsite. 2. Materials and methods 2.1. Reagents All solutions used in these experiments were prepared with Milli-Q water (Millipore, Bedford USA)). IP6 was in the form of phytic acid sodium salt, from corn (Sigma) which, from previous work by Johnson et al. [31], has a formula weight of 880 ± 10 g mol1 and molecular formula Na6C6H6(HPO4)65H2O. Fresh IP6 solutions were made before each experiment to reduce the effect of hydrolysis, with the same IP6 sample used for each experiment. All other chemicals were of analytical reagent grade. 2.2. Adsorbent Gibbsite was prepared by dissolving 1000 g of Al(NO3)39H2O (Chem Supply) in 3 L of Milli-Q water. The resulting solution was titrated slowly with 4.0 M NaOH under N2 until the pH reached 4.5 and was then placed in an oven at 40 °C for 2 h. The gibbsite slurry was then transferred into 75 mm cellulose acetate dialysis tubing and dialysed against Milli-Q water at 50 °C for 4 weeks. The water was changed twice daily for the first week, then changed once a day for the remaining three weeks. Finally, the gibbsite was freeze dried. The BET surface area of the gibbsite powder was measured with a Micrometrics ASAP 2000 instrument after degassing for 18 h and found to be 60.2 m2 g1. This gibbsite was used in all experiments. 2.3. Equipment All experiments were conducted in glass water-jacketed reaction vessels. Water, maintained at 4, 12, 25 or 55 °C, was circulated through the water jacket of the reaction vessel by use of a Haake K15 Circulator to maintain a constant temperature during

284

M. Ruyter-Hooley et al. / Journal of Colloid and Interface Science 440 (2015) 282–291

experiments. pH measurements were performed using an Orion Sureflow pH electrode coupled to a Metrohm model 691 pH meter. pH adjustments were made with either HCl and NaOH solutions while the pH electrode was calibrated at the reaction temperature before each experiment using 0.05 M potassium hydrogen phthalate, 0.025 M phosphate and 0.01 M borax buffers. 2.4. Adsorption experiments Adsorption experiments were carried out in duplicate with all experimental data shown in the accompanying figures. Preliminary kinetic experiments indicated that the adsorption process was complete within 15 min; hence all experiments conducted at 12, 25 and 55 °C were left for 30 min to equilibrate. Because sorption at 4 °C occurred more slowly, experiments at 4 °C were left for 24 h to equilibrate in a cool room set at 4 °C. For adsorption edge experiments, sufficient gibbsite to provide a surface area of 1000 m2 L1 was added to a glass reaction vessel containing 0.10 M NaCl, and stirred for at least 12 h to equilibrate. The pH of the equilibrated gibbsite solution was adjusted to 3.5 with HCl and an IP6 aliquot added to provide a concentration of 0.075 mM. The pH was readjusted to 3.5 and the suspension left for 30 min to equilibrate. A 5 mL sample was taken and filtered into a 10 mL test tube using 25 mm Whatman 0.22 lm Glass Microfibre Filter paper. The pH of the remaining solution was then increased by approximately 1.0 unit, left for 30 min to equilibrate, and another 5 mL sample taken and filtered into a separate 10 mL test tube. This process was repeated until 10 samples were obtained over the pH range from 3.5 to 12.5. For experiments conducted at 4 °C, before being filtered, 5 mL samples were placed in sealed centrifuge tubes and placed on an end-on-end mixer situated in a cold room at 4 °C for 24 h. Following this the samples were filtered into 10 mL test tubes. The IP6 concentration of filtered solutions was measured with a Varian Liberty 220 ICP Atomic Emission Spectrometer (ICP-AES) using the P line at k = 213.618 nm. Isotherm experiments were conducted at pH 6 and 10 following a similar procedure to that used for the adsorption edge experiments. Sufficient gibbsite to provide a surface area of 1000 m2 L1 was added to a glass reaction vessel, containing 0.10 M NaCl, and stirred for at least 12 h to equilibrate. An IP6 aliquot was added to the equilibrated gibbsite solution to provide an initial solution concentration of 6.00 lM. The pH was re-adjusted to the desired pH and left for 30 min to equilibrate. A 5 mL sample was taken and filtered into a 10 mL test tube. Another IP6 aliquot was then added, the pH adjusted, and the process repeated until 12 samples were obtained within the concentration range from 6.00 lM to 1.00 mM. Again, for isotherms conducted at 4 °C 5 mL samples were placed in sealed centrifuge tubes and put on an end-on-end mixer situated in a cold room at 4 °C for 24 h, before being filtered. Following this the samples were filtered into 10 mL test tubes and the IP6 concentration of filtered solutions analysed using ICP-AES. 2.5. Sample preparation for

31

P MAS NMR spectroscopy

2.5.1. Sorbed samples All samples for 31P MAS NMR analysis of IP6 sorbed to gibbsite were produced at 25 °C. Sufficient gibbsite to provide a surface area of 1675 m2 L1 was added to a glass reaction vessel containing 0.10 M NaCl and stirred for at least 12 h to equilibrate. The larger amount of substrate was used in order to produce samples of sufficient size (ca 50–100 mg) for the NMR analysis. The pH of an equilibrated gibbsite solution was adjusted to pH 3 and an IP6 aliquot, from a stock solution, was added to provide a solution concentration of 0.095 mM. The pH was re-adjusted to 3, left for 15 min to equilibrate, and a 50 mL sample taken and placed in a sealed

centrifuge tube. The pH of the remaining solution was then increased to 6, left for 15 min to equilibrate and another 50 mL sample was taken and placed in a sealed centrifuge tube. This process was repeated to obtain samples at pH 9 and 11. The centrifuge tubes were placed on an end-on-end mixer for 24 h. After 24 h, samples were centrifuged at 12,000g for 30 min. Following centrifugation, the supernatant was removed from each sample, the pellets oven dried at 40 °C, and the resultant solid samples crushed with a mortar and pestle and placed in sealed 10 mL test tubes. Samples at pH 6 with different IP6 concentrations were obtained following a similar procedure. Sufficient gibbsite to provide a surface area of 1675 m2 L1 was added to a glass reaction vessel, containing 0.10 M NaCl, and stirred for at least 12 h to equilibrate. An IP6 aliquot was added to the equilibrated gibbsite solution to provide an initial solution concentration of 0.03 mM. The pH was re-adjusted to 6, left for 15 min to equilibrate, and a 50 mL sample was taken and placed in a sealed centrifuge tube. Another IP6 aliquot was then added to provide a solution concentration of 1.10 mM. After the pH was re-adjusted to 6 the suspension was left for 15 min to equilibrate before a 50 mL sample was taken and placed in a sealed centrifuge tube. The centrifuge tubes were placed on an end-on-end mixer for 24 h. After 24 h, samples were centrifuged at 12,000g for 30 min. Following centrifugation, the supernatant was removed from each sample and the pellets oven dried at 40 °C, crushed, and then placed in sealed 10 mL test tubes. 2.5.2. Al–IP6 precipitate An Al–IP6 precipitate was prepared at pH 6 at an initial Al:IP6 ratio of 2:1. Sufficient AlCl3 to provide an aluminum concentration of 20 mM was added to a glass reaction vessel containing 250 mL of 10 mM IP6. The pH of the solution was adjusted to 6 and left to equilibrate for 15 min. A 100 mL sample was taken and placed in two sealed 50 mL centrifuge tubes. The samples were then centrifuged at 12,000g for 15 min. Following centrifugation, the supernatant was removed from each sample and the pellets oven dried at 40 °C, crushed, and then placed in sealed 10 mL test tubes. 2.6.

31

P MAS NMR measurements

All solid state 31P MAS NMR spectra were recorded using an Agilent/Varian/Chemagnetics InfinityPlus CMX-360 NMR spectrometer. For these experiments, dried samples were packed into 4 mm ZrO2 rotors. Solid state 31P single-pulse MAS NMR spectra were recorded using a 4 mm T3 MAS probe and a 31P operating frequency of 149.69 MHz. Single-pulse with proton decoupling experiments were performed with pulse delay 12 s and pulse width 1.8 ls. Sample spinning frequencies ranged from 6 to 7 kHz. The number of acquisitions taken for each sample was sufficient to provide an acceptable signal to noise ratio. 31P isotropic chemical shift data are given with respect to 85% H3PO4 (0 ppm). 2.7. Deconvolution of

31

P MAS NMR spectra

The 31P MAS NMR spectra were deconvoluted into component peaks using PeakFit v4 for Windows [51]. The resonance line shape chosen was 50% Gaussian/50% Lorenzian. All spectra were deconvoluted into at least three peaks with fixed peak positions and widths based on findings from preliminary automatic deconvolution. Allowing peak centers and widths to vary made little difference to the overall fit. 2.8. Modeling of sorption results 2.8.1. Langmuir modeling All isotherms were modeled using the Langmuir equation which can be written as:

285

M. Ruyter-Hooley et al. / Journal of Colloid and Interface Science 440 (2015) 282–291

N¼

Nm KC 1 þ KC

Table 2 Dissociation Constants (log10(b) for IP6 (I = 0.1 M NaCl). Parameter

where N is the amount sorbed, Nm is the amount sorbed when monolayer coverage is achieved, K is the equilibrium constant for the overall sorption process and C is the equilibrium solution concentration.

2.8.2. Surface complexation modeling (SCM) Results from adsorption edges and isotherms were modeled with an ECCM using the computer program WinSGW [34]. The ECCM has been used previously to describe the adsorption of various organic anions onto a range of surfaces [45,46,38,37,4,5,28]. The ECCM has two sorption planes, a surface plane (0-plane) and an outer plane (b-plane); inner-sphere surface complexes form at the 0-plane while outer-sphere complexes occur at the b-plane. A more detailed description of the ECCM model is given by Nilsson et al. [45]. The basic charging properties of the gibbsite surface were described using only the singly coordinated Al–OH sites. Based on previous studies of gibbsite approximately 20% of the total surface area corresponds to edge sites [27,50]. With this in mind the specific surface area occupied by singly coordinated Al–OH sites in this study is 12.0 m2/g. Crystallographic data suggests that the number of singly co-ordinated edge sites on gibbsite is 8.15 sites nm2 [27]. From this information the concentration of singly co-ordinated edge sites is approximately 13 lmol m2. The concentration of singly co-ordinated sites is considerably larger than the Nm values reported here for IP6 adsorption on gibbsite, which indicates there is a substantial excess of singly co-ordinated sites available to interact with all of the IP6 added. For this reason, doubly co-ordinated sites were not included in the surface model. However, given the size of IP6, it is possible for outer-sphere complexes to form hydrogen bonds with doubly coordinated surface sites present on gibbsite edges, but, in order to minimize model complexity, this possibility was not considered. The parameters used in this model, shown in Tables 1 and 2, were obtained from previous studies. The temperature effect on surface electrostatics can be taken into account using the WinSGW program. When modeling at 4, 12 and 55 °C, the surface protonation constants of gibbsite became an adjustable parameter. Surface complexation modeling was carried out by optimizing formation constants for possible adsorption reactions. All adsorption edge and isotherm data, together with 31P MAS NMR spectroscopic information, were used to optimize formation constants for the sorption reactions.

4 °C

12 °C

25 °C

55 °C

14.73

14.53

14.00

13.15

IP6 protonation constants (pK)b L12 + H+ M LH11 9.73 HL11 + H+ M LH10 10.01 2 10 + 9 H2L + H M LH3 9.55 H3L9 + H+ M LH8 8.17 4 H4L8 + H+ M LH7 6.51 5 H5L7 + H+ M LH6 5.25 6 H6L6 + H+ M LH5 2.81 7

9.67 9.94 9.53 8.16 6.51 5.25 2.84

9.58 9.84 9.50 8.14 6.50 5.25 2.88

9.39 9.63 9.44 8.10 6.49 5.25 2.96

a

pKw

a

Lide [42]. Constants at 25 °C were obtained from De Stefano et al. [17]. Constants at 4, 12 and 55 °C were obtained using the equation log10 bij(T2) = log10 bij(T1) + DHij(1/ T1  1/T2)/2.303R [16]. Enthalpies were obtained from De Stefano et al. [18]. b

slope and y-intercept from which the enthalpy and entropy can be calculated. 3. Results 3.1. Adsorption edges and isotherms Data indicating the percent of IP6 sorbed onto gibbsite as a function of pH at four temperatures are shown in Fig. 1. For all temperatures virtually all of the IP6 added was sorbed up to about pH 10.5. Above pH 10.5 the percent sorbed decreased sharply at all temperatures, with the decrease more pronounced at higher temperatures. The lines indicate the fit of the SCM. Adsorption isotherms conducted at pH 6 and 10 at four temperatures are shown in Figs. 2 and 3, respectively. At pH 6 Fig. 2 indicates that the amount sorbed reached a plateau at about 0.50 lmol m2 at the four temperatures studied, while Fig. 3 shows a lower maximum amount sorbed at pH 10 for all temperatures. The effect of temperature on adsorption was relatively small at both pH values. At pH 6 there appears to be slightly more adsorbed at higher temperatures, while at pH 10 the reverse was found. Again, the lines represent the best fit of the SCM. In addition to the SCM, the isotherms were also modeled using the Langmuir equation. Both Nm and K values for isotherms conducted at pH 6 and 10 at the four different temperatures are shown in Table 3. At pH 6, the maximum amount sorbed was similar at all temperatures studied with all values greater than the maximum amount sorbed at pH 10. The results in Table 3 confirm that the 100

2.9. Determination of thermodynamic parameters Thermodynamic parameters were estimated by the use of the van’t Hoff equation from equilibrium constants obtained from surface complexation modeling. Parameters were obtained by plotting ln K versus 1T and using linear regression to calculate the Table 1 Parameters used in surface complexation modeling. Parameter Specific surface area occupied by „AlOH (m2 g1) Site density (sites nm2)a Capacitanceb C1 (F m2) C2 (F m2) a b

Hiemstra et al. [27]. Rosenqvist et al. [50].

% Adsorption

80

60

40

20

12.03 8.15

0

3

4

5

6

7

8

9

10

11

12

13

pH 7.7 1.82

Fig. 1. Adsorption edges of 0.075 mM IP6 on 1000 m2 L1 gibbsite at four temperatures: 4 °C (D), 12 °C (N), 25 °C (s) and 55 °C (d). Lines show the best fit obtained from surface complexation modeling: the parameters are given in Tables 1, 2 and 5. All data from duplicate experiments are shown.

286

M. Ruyter-Hooley et al. / Journal of Colloid and Interface Science 440 (2015) 282–291

3.2. Solid state

1.0

Ads (µmol/m2)

P MAS NMR spectroscopy

The 31P MAS NMR spectrum of IP6, shown in Fig. 4, contains a single resonance line at 1.4 ppm. The 31P MAS NMR spectra of IP6 sorbed to gibbsite, shown in Figs. 5 and 6, have relatively broad resonance lines indicating that phosphorus atoms are found in a wider range of different chemical environments. The shape of these sorbed spectra indicate that they are made up from several resonance lines as has been found in previous studies of the sorption of orthophosphate onto solid substrates [44,43,30,35,54,41]. To further investigate the nature of the sorption of IP6 to gibbsite the spectra were deconvoluted using up to four resonance lines with centers at +5, 0, 6, 13 and 21 ppm. The resonance line

0.8

0.6

0.4

0.2

0.0 0.0

31

0.2

0.4

Conc.

0.6

0.8

(mol/m3)

Fig. 2. IP6 sorption isotherms measured on 1000 m2 L1 gibbsite at four temperatures at pH 6: 4 °C (D), 12 °C (N), 25 °C (s) and 55 °C (d). Total IP6 added ranged from 6.00 lM to 1.00 mM. Lines show the best fit obtained from surface complexation modeling: the parameters are given in Tables 1, 2 and 5. All data from duplicate experiments are shown.

1.0

Ads (µmol/m2)

0.8

0.6

20

10

0

-10

-20

ppm Fig. 4. A 31P single-pulse MAS NMR spectrum of a powder sample of IP6 used in this study.

0.4

0.2

0.0 0.0

0.2

0.4

Conc.

0.6

0.8

(mol/m3)

Fig. 3. IP6 sorption isotherms measured on 1000 m2 L1 gibbsite at four temperatures at pH 10: 4 °C (D), 12 °C (N), 25 °C (s) and 55 °C (d). Total IP6 added ranged from 6.00 lM to 1.00 mM. Lines show the best fit obtained from surface complexation modeling: the parameters are given in Tables 1, 2 and 5. All data from duplicate experiments are shown.

pH 11

maximum amount sorbed at pH 6 tended to increase somewhat with increasing temperature, while at pH 10 the maximum amount sorbed decreased with increasing temperature. K values at both pH 6 and 10 are relatively large, indicating that IP6 has a strong affinity for the gibbsite surface. At both pH 6 and 10, K values tended to increase with increasing temperature with the increase somewhat greater at the lower pH.

pH 6

pH 3

Table 3 Langmuir constants from isotherms conducted at pH 6 and 10 across four temperatures. (± standard error in the non-linear regression analysis). Temperature (°C)

pH

Nm (lmol m2)

K (m3 mol1)

4 12 25 55 4 12 25 55

6 6 6 6 10 10 10 10

0.47 ± 0.01 0.46 ± 0.03 0.52 ± 0.04 0.50 ± 0.03 0.41 ± 0.02 0.39 ± 0.01 0.39 ± 0.01 0.33 ± 0.01

139 ± 2 120 ± 3 164 ± 4 193 ± 3 120 ± 2 120 ± 2 120 ± 2 135 ± 2

Al-IP6 pH 6

20

10

0

-10

-20

-30

ppm Fig. 5. 31P single-pulse MAS NMR spectra of 0.095 mM IP6 sorbed to gibbsite at pH 3, 6, and 11 and Al–IP6 precipitate at pH 6.

287

M. Ruyter-Hooley et al. / Journal of Colloid and Interface Science 440 (2015) 282–291

Table 4 Chemical shifts and relative peak intensities for the 31P MAS NMR spectra of Al–IP6 precipitate and IP6 sorbed to gibbsite at different pH and IP6 concentration. Compound/adsorption system

pH

IP6

1.1 mM

1.4 6

0 6 13 21

10 48 34 8

0.095 mM IP6/gibbsite

3

0 6 13 21

20 50 19 11

0.095 mM IP6/gibbsite

6

0 6 13

43 41 16

0.095 mM IP6/gibbsite

9

5 0 6 13

16 37 22 25

0.095 mM IP6/gibbsite

11

5 0 6 13

30 31 17 22

0.030 mM IP6/gibbsite

6

0 6 13

41 35 24

1.10 mM IP6/gibbsite

6

0 6 13

35 44 21

0.03 mM

10

0

-10

-20

-30

ppm Fig. 6. 31P single-pulse MAS NMR spectra of 0.03 mM, 0.095 mM and 1.1 mM IP6 sorbed to gibbsite at pH 6.

pH 11

pH 3

20

10

0

-10

-20

-30

ppm Fig. 7. Deconvolution of 31P single-pulse MAS NMR spectra of 0.095 mM IP6 sorbed to gibbsite at pH 3 and 11.

positions chosen provided the best fit of the deconvoluted spectra to the measured 31P MAS NMR spectra. Fig. 7 shows deconvolutions of 31P NMR NMR spectra at pH 3 and 11. The relative intensities for each resonance line at pH values from 3 to 11 and IP6 concentrations from 0.03 to 1.10 mM are shown in Table 4. There are clear trends in the relative amounts of each resonance line as the pH increases. In particular, the total intensity of the resonances at +5 and 0 ppm increased with pH while the intensity of the resonance line at 6 ppm decreased. However, at pH 6, the relative integral resonance line intensities show little change with increasing IP6 concentration. 3.3. Modeling of adsorption results 3.3.1. Surface complexation modeling Several different sorption reactions were tested to find the best fit for the experimental data while being consistent with the 31P MAS NMR data. The model presented in Table 5, which includes two inner- and one outer-sphere complex, was found to provide the best fit to the experimental data at all temperatures and was the most consistent with the 31P MAS NMR data. The inner sphere complexes involved reaction of IP6 with either one or three surface sites while the outer-sphere complex suggested interaction with two positively charged surface sites. For this model, the inner-sphere complexes dominated sorption at low pH while at higher pH more than 50% of sorption occurred as an outer-sphere

Relative intensities (%)

Al-IP6 precipitate 0.095 mM

20

31 P Chemical shift (ppm)

complex. The variation in the speciation of surface complexes with pH when the concentration of IP6 was 0.075 mM is shown in Fig. 8 while Fig. 9 shows the speciation of surface complexes with increasing IP6 concentration at pH 6. 3.4. Thermodynamic parameters The thermodynamic parameters, obtained from equilibrium constants derived from surface complexation modeling, are shown in Table 6. The enthalpies for the formation of inner-sphere surface complexes were found to be endothermic, while the enthalpy for the outer-sphere complex was exothermic. The entropies for all the sorption reactions were large and positive. 4. Discussion 4.1. Sorption of IP6 onto gibbsite Fig. 1 shows that virtually all of the IP6 from a 0.075 mM solution added to the 1000 m2/L gibbsite suspension was sorbed in the pH range from 3.5 to 10.5 at all temperatures studied. Above pH 10.5 the fraction sorbed tended to decrease. A similar decrease in sorption of IP6 at higher pH has been found by Guan et al. [23] and Johnson et al. [31] on amorphous aluminum hydroxide and goethite respectively, however in these studies, the decrease in sorption occurred at lower pH. This decrease in sorption at high pH most likely occurs because of changes in the electrostatic interaction between IP6 and the surface. At 25 °C, the pH at the isoelectric point, pHIEP for gibbsite prepared by the method used in this work has been determined to be 10.0 by Rosenqvist et al. [50], significantly higher than the pHIEP values of amorphous aluminum hydroxide or goethite. At pH values above 10 the gibbsite surface therefore tends to carry a net negative charge which increases with pH. This results in strong repulsion between the negative surface and IP6, which also carries a large negative charge

288

M. Ruyter-Hooley et al. / Journal of Colloid and Interface Science 440 (2015) 282–291

Table 5 Reactions and constants for IP6 sorption onto gibbsite from surface complexation modeling. Equilibrium constants listed as log10 K. (I = 0.1 M NaCl).

+a



„AlOH M AlO + H „AlOH + H+ M „AlOH+a 2 „AlOH + L12 + 5H+ M „Al(LH4)7 + H2O 12 + „3AlOH + L + 6H M „Al3(LH3)6 + 3H2O 10 „2AlOH + L12 + 4H+ M („AlOH2)2+ 2 (LH2) a

4 °C

12 °C

25 °C

55 °C

11.57 7.64 71.4 84.7 75.8

11.45 7.41 74.4 87.5 76.1

11.18 7.17 76.5 89.4 75.0

10.85 6.91 77.2 89.8 72.0

Equilibrium constant at 25 °C obtained from Karamalidis and Dzombak [33].

100

1.0

0.25 0.20

(AlOH2)22+(LH2)10-

80

0.8

Al(LH4)

Al(LH4)

Ads (µmol/m2)

% Adsorption

0.10

60

7-

Al3(LH3)6-

40

20

0

0.6

7-

0.15 Al3(LH3)

0.05

6-

2+ 10(AlOH2)2 (LH2)

0.00

Al(LH4)7-

0.4

0.2 (AlOH2)22+(LH2)10-

3

4

5

6

7

8

9

10

11

12

13

pH Fig. 8. Speciation of 0.075 mM IP6 adsorbed onto gibbsite (1000 m2 L1) predicted by surface complexation model at 25 °C.

at these pH values. Below pH 10 however, the surface will have a net positive charge which leads to a favourable electrostatic interaction with the negatively charged IP6 and thus a strong attraction between IP6 and the surface. The increase in affinity of IP6 for the surface at lower pH values was also observed in the isotherms conducted at pH 6 and 10, with the maximum amount sorbed at pH 6 significantly greater than that at pH 10 for all temperatures studied. At higher temperatures the decrease in IP6 sorption above pH 10.5 was more pronounced (Figs. 1 and 3). This is consistent with other results for anion sorption to soils and oxide surfaces, which has generally been shown to decrease as the temperature increases [8,4,5], and may be the result of an increased net negative surface charge at higher temperatures. Both Ward and Brady [55] and Angove et al. [3], using kaolinite and goethite respectively, have shown that the pH at the point of zero charge shifts to lower pH at higher temperatures resulting in an increase in negative charge at the surface at pH 10 at higher temperatures. This increase in the net negative surface charge would be expected to cause a decrease in sorption at higher temperatures due to increased electrostatic repulsion between IP6 and the surface. In contrast, Fig. 2 indicates that at pH 6, IP6 sorption increased with increasing temperature. This change in the effect of temperature with pH has been found in several previous studies including those of Ward and Brady [55] for the sorption of oxalate on kaolinite, and Angove et al. [4,5] for the sorption of mellitic acid on goethite and kaolinite. The increased sorption found at lower pH has been proposed by Angove et al. to be due to a decrease in the hydrated radius of the sorbing ion with increasing temperature. In aqueous environments charged species attract water molecules with the extent of that attraction depending on the size of the ion and its charge. At pH 6, IP6 carries a net charge of 7, which results in a relatively large hydrated radius. Hence, any decrease in hydrated radius will increase the number of IP6 molecules that can sorb onto the surface thereby increasing the maximum amount sorbed.

0.0 0.0

0.2

0.4

0.6

0.8

Conc. (mol/m3) Fig. 9. Speciation of IP6 adsorbed onto gibbsite (1000 m2 L1) at pH 6 and 25 °C predicted by surface complexation model. The inset shows the speciation in the lower concentration region of the isotherm. The amount of IP6 added ranged from 6.00 lM to 1.00 mM.

Table 6 Thermodynamic parameters for IP6 adsorption onto gibbsite from surface complexation modeling. (± standard error). Reaction

DH (kJ mol1)

DS (J K1 mol1)

„AlOH + L12 + 5H+ M „Al(LH4)7 + H2O „3AlOH + L12 + 6H+ M „Al3(LH3)6 + 3H2O 10 „2AlOH + L12 + 4H+ M („AlOH2)2+ 2 (LH2)

177 ± 9 153 ± 8 141 ± 3

2032 ± 29 2200 ± 28 953 ± 11

The Langmuir constants, K, obtained at pH 6 and 10 at all four temperatures, shown in Table 3, are quite large compared with literature values for other adsorbates [29,3,54]. The large values of K suggest strong interactions between IP6 species and the gibbsite surface. Conversely, the Nm values obtained at both pH 6 and 10, are relatively small. The maximum amount sorbed at pH 6 and 25 °C is (0.50 ± 0.03) lmol m2, which is similar to that obtained by Giaveno et al. [20] for the adsorption of IP6 on gibbsite at pH 4.5. The total amount of IP6 sorbed onto gibbsite was approximately one sixth of the maximum sorption of orthophosphate onto gibbsite found by Van Emmerik et al. [54] which is expected since the IP6 species are much larger than orthophosphate ions. The areas occupied by orthophosphate and IP6 ions have been estimated as 0.7 and 2.0 nm2, respectively [12,31]. The small Nm values indicate that bulk precipitation of aluminum phytate is unlikely to be a major sorption mechanism. If bulk precipitation was occurring, IP6 would be continually removed from solution so that we would expect upward curvature in the isotherms which would lead to larger Nm values. The initial concentrations of both aluminum and IP6 in solution are important in precipitate formation. Van Emmerik et al. [54] found that the concentration of free aluminum in solution in equilibrium with gibbsite was less than 0.02 mM in the pH range from 4 to 10

M. Ruyter-Hooley et al. / Journal of Colloid and Interface Science 440 (2015) 282–291

suggesting that little Al–IP6 precipitation is likely. The presence of strong complexing anions in solution has been found to increase the dissolution of some substrates [52,13]; therefore an almost simultaneous process involving dissolution of the substrate followed by precipitation is possible, although unlikely given the shape of the isotherms. Since the stoichiometries of sorption and dissolution/precipitation reactions are likely to be similar, it is difficult to distinguish between these processes through surface complexation modeling approaches. However 31P MAS NMR spectroscopy should provide an indication of the surface processes that occur. 4.2.

31

P MAS NMR spectroscopy

31 P MAS NMR spectra provide information on molecular interactions with chemical shifts correlating with the electronic environment around the 31P nucleus. Outer-sphere complexation produces relatively small positive chemical shifts relative to phosphoric acid, as the interaction with the surface involves only electrostatic and hydrogen-bonding forces while inner-sphere complexation tends to cause negative chemical shifts as a result of increased shielding around the 31P nucleus because of the formation of chemical bonds. Precipitate formation, or polydentate inner-sphere complexation, tends to produce the most negative chemical shifts as the anion shares its electrons with more than one cationic species. In general, the more ionic in character the phosphate groups, the more positive the chemical shift [53,30]. Fig. 5 shows the 31P MAS NMR spectrum for freshly prepared Al–IP6 precipitates, and also of IP6 sorbed onto gibbsite at various pH values. The deconvoluted spectrum for the precipitate (Table 4) is very similar to that for sorption of IP6 on gibbsite at pH 3 with the same four lines and similar relative percentages of each, suggesting that the IP6 sorbed at pH 3 has a similar chemical environment to that in the precipitate. However, as the pH is increased the spectra for the sorbed species have peaks centered at more positive chemical shifts, indicating that overall bonding to the phosphate groups becomes less covalent in nature as the pH at which sorption occurs is increased. The deconvoluted 31P MAS NMR spectra provide more detailed information on possible speciation of the sorbed IP6 species. Fig. 7 shows that the spectra of ‘sorbed’ IP6 from pH 3 to 11 can be deconvoluted into five peaks. Two peaks at +5 and 0 ppm have similar chemical shifts to the spectrum of IP6 solid (1.4 ppm), suggesting that they arise from phosphorous atoms with similar shielding environments to IP6. These peaks could represent outer-sphere complexation, and/or phosphate groups on sorbed IP6 that are not involved in the bonding to the gibbsite surface. The deconvoluted peak observed at 6 ppm is more significantly shifted from 1.4 ppm, suggesting more shielding, and most probably corresponds to the formation of an inner-sphere complex with the gibbsite surface. Previous 31P-MAS NMR studies on the sorption of various phosphates have reported chemical shifts of similar magnitude and have assigned these to inner-sphere complexation [30,35,54,41]. Recently, Li et al. [39] and Li et al. [40], investigated the sorption of phosphate onto boehmite using NMR techniques. Both studies, which used 31P {27Al}-REAPDOR experiments, concluded that the peak at 6 ppm corresponded to the formation of a bidentate binuclear inner-sphere complex. However, because the IP6 molecule in our study has six phosphate groups the 31P MASNMR spectra are more difficult to interpret as each IP6 species will have phosphates in chemical environments ranging from those bound to the gibbsite surface to those relatively free of surface influences. The peaks at 21 and 13 ppm suggest more significant shielding of the phosphorous groups compared to H3PO4, which is an

289

indication of even stronger association with aluminum. Similar peaks have been reported for the sorption of phosphates onto aluminum oxides, and have previously been assigned to the presence of a surface precipitate [30,41]. Surface precipitation of phosphate onto aluminum oxide surfaces would be expected to have more than one aluminum ion associated with a phosphate group which would lead to more negative chemical shifts. The resonance line at pH 3 at 21 ppm may correspond to a more ordered precipitate, perhaps from bulk precipitation. The formation of a bulk precipitate at pH 3 is possible, since the solubility of gibbsite increases at lower pH values [57]. At higher pH values the resonance line at 21 ppm is not found. That observation, together with the relatively small values of Nm at pH 6 and 10 (Table 3) compared with the available surface area and the lack of upward curvature of the sorption isotherms (Figs. 2 and 3) indicate that bulk precipitation is unlikely at pH 6 or above. Yan et al. [58] found a similar set of resonance lines at 0.5, 6.4 and 11.2 ppm for IP6 sorbed onto amorphous aluminum hydroxide and assigned these resonances to surface sorbed species (0.5 and 6.4 ppm) and Al–IP6 precipitate (11.2 ppm). In their study the integral intensity of the resonance line at 11.2 ppm increased with interaction time indicating the formation of bulk precipitate at the expense of the surface sorbed species over time. This result complemented their results for the ratio OH ions released per IP6 species sorbed onto amorphous aluminum hydroxide which increased from 1.0 after 5 h of equilibration to 5 after 20 h. However, when they investigated IP6 sorption onto the more crystalline samples a-Al2O3 and boehmite the ratio remained constant at 1.0 over the time of the experiment indicating the absence of bulk precipitate formation as is found in this work. Until recently, studies have suggested that IP6 sorption onto soil minerals involves only inner-sphere complexation [47,14,24]. However, more recent ATR–FTIR work by Johnson et al. [31] showed that IP6 sorption to goethite occurs largely through the formation of outer-sphere complexes. Outer-sphere surface complexation has also been found to be important for other highly charged organic species such as benzene carboxylates [46,11,32,4,5]. In studies of the adsorption of mellitic acid onto both goethite and kaolinite, for example, there was very little spectroscopic evidence of inner-sphere binding [32,4,5]. While sorption of IP6 to goethite may occur largely through outer-sphere complexation, the 31P MAS NMR spectra in this study and the presence of resonances in regions representing significant shielding of phosphate groups, clearly indicates the presence of inner-sphere complexes. The resonance lines at 13 and 21 ppm also suggest the presence of a surface precipitate, although in practice it is difficult to differentiate between the formation of a surface precipitate and inner-sphere complexation of IP6 to a number of AlOH groups at the surface. For simplicity, in our modeling of this sorption system we define a ‘surface precipitate’ species as an inner-sphere complex associated with more than one surface site. Changes in the relative intensities of resonances in the NMR spectra (Table 4) suggest that outer-sphere complexation plays a more important role in the sorption process at high pH, whilst the more shielded species resulting from inner-sphere complexation are dominant at lower pH. The tendency to form inner-sphere complexes at low pH and outer-sphere complexes at high pH has also been observed for the sorption of benzene carboxylates such as phthalate, pyromellitate and trimellitate onto goethite [11] and phenyl phosphate onto c-alumina [30]. While previous studies of orthophosphate sorption on oxide surfaces have reported a decrease in the relative intensity of surface precipitate with increasing pH [30,54] there is little change in the intensity of the resonances ascribed to surface precipitation with pH in this study.

290

M. Ruyter-Hooley et al. / Journal of Colloid and Interface Science 440 (2015) 282–291

4.3. Surface complexation modeling With the 31P MAS NMR results as a guide, both adsorption edge and isotherm data at the four temperatures studied were modeled using an ECCM SCM. Throughout the modeling process the experimental data was also modeled using the CD-MUSIC model. In the CD-MUSIC model of Hiemstra and Van Riemsdijk [26], surface parameters and site densities are based on Pauling’s bond valence concepts and crystallographic data. Most of the experimental data could be fitted using the CD-MUSIC model, however, we have chosen to use the ECCM model as we wanted a robust model that could fit all experimental data, including additional data studying the effect of cadmium on IP6 sorption which will be presented in a future paper. The ECCM has previously been used to model ternary systems [36,37]. The model presented here does not aim to provide a complete description of all the surface processes that occur, but rather to find a set of reactions that adequately describe the pH, concentration and temperature dependence sorption data while being consistent with 31P MASNMR data. 31P MASNMR spectra indicated that sorption involved both outer- and inner-sphere complexation, together with surface precipitation. Many different sorption reactions were tested with IP6 interacting with between one and four surface sites, consistent with the known structure of the molecule. The model shown in Table 5 provided the best fit to the experimental data and was most consistent with 31P MASNMR information. The model shown here provides a good fit to both edge and isotherm data. Other reaction stoichiometries investigated could provide a good fit to either the edge or the isotherm data, but not to both. Models that included bulk precipitation of an Al–IP6 phase were not able to adequately fit the data when literature stability constants were used. If the precipitation solubility constant was used as an adjustable parameter convergence yielded speciation with a negligible amount of the bulk precipitate. Having ruled out bulk precipitation, we then modeled a range of stoichiometries where IP6 could be closely coordinated with Al via one or more surface sites. Given the variation in the conformation of IP6 with pH [7], its large hydrated radius, and the likelihood of repulsive forces between sorbed species, it is highly improbable that all six phosphate groups could interact with the surface. The proposed inner-sphere surface complex involved complexation to one surface site while the outer-sphere complex involved interaction with two surface sites. The complex corresponding to surface precipitation involved binding to three AlOH surface sites. Between two and four IP6 phosphate groups have been previously proposed to bind to the surface of different minerals. Celi et al. [14] and Celi et al. [13] suggested that IP6 interacts with goethite and ferrihydrite through four and two phosphate groups respectively, while Guan et al. [24] using ATR–FTIR spectroscopy and molecular orbital calculations suggested that three of the six phosphate groups were bound to amorphous aluminum hydroxide. However more recently, Yan et al. [58] measured OH ion release associated with IP6 sorption onto amorphous aluminum hydroxide and found, for the formation of inner-sphere complexes, the ratio of OH ions released per IP6 species sorbed was close to 1.0. These results, similar to our proposed model, suggest that one IP6 phosphate group might interact with one surface site. The resonances at +5 and 0 ppm are most probably due to both outer-sphere complexation and to phosphate groups that are not directly interacting with the surface. The model proposes a single 10 outer-sphere complex, [(„AlOH2)2+ ], which interacts with 2 (LH2) two „AlOH2 groups. This complex is the principal adsorbed species above pH 7. At lower pH, the inner-sphere complexes [„Al(LH4)7] and [„Al3(LH3)6] were dominant (see Fig. 8), which

is consistent with the 31P MASNMR data where peaks representing higher shielding are relatively more important at low pH. While the model predicts that there is little outer-sphere complexation below pH 5 the 31P MAS NMR deconvolution data (Table 4) at pH 3 shows the relative intensity of the resonance line at 0 ppm to be 20%. This is almost certainly due to ‘free’ phosphate groups associated with sorbed IP6. These ‘free’ phosphate groups will have electronic environments, and hence chemical shifts, not dissimilar to the ‘outer-sphere’ complexes. Table 4 suggests that the transition in the mid pH region from inner- to outer-sphere is more gradual than predicted by the model. In reality there is a range of progressively less protonated inner-sphere surface complexes that could be used in the modeling. However, inclusion of these species would increase the number of adjustable parameters and produce almost infinite range of convergence options, making the generated models less meaningful. The model appears to underestimate the proportion of the complex [„Al3(LH3)6] at higher pH and IP6 concentration in comparison with the measured 31P-MAS NMR intensity ratios (Table 4). However, there is no straightforward way to compare relative NMR intensities in this system. For each sorbed species only a fraction of the phosphate groups directly interact with the surface leaving the rest ‘free’ to a greater or lesser extent. Consequently, while the 31P MAS NMR integral intensities provide valuable information about the nature of the sorbed species and a useful guide to their relative amounts, this information must be seen as qualitative rather than quantitative. 4.4. Thermodynamic parameters The thermodynamic parameters (Table 6) obtained from surface complexation modeling indicate that the adsorption enthalpies for the inner-sphere surface complexes were endothermic, while the adsorption enthalpy for the outer-sphere complex was exothermic. Both the endothermic and exothermic enthalpies are large compared with those for other adsorbates. Angove et al. [4] also found an exothermic enthalpy change for the sorption of mellitic acid on goethite. These enthalpy results explain the increase in sorption with temperature at lower pH where innersphere (endothermic) complexes dominate while the decrease in sorption with temperature at higher pH results from the increasing importance of outer-sphere (exothermic) complexation. The entropy values for both inner- and outer-sphere complexation are large and positive, most probably because of changes in the hydration of IP6 ions on binding to the charged surface. As expected the entropy for [„Al3(LH3)6] is larger than that of [„Al(LH4)7] because water molecules must be liberated from three surface sites for the [„Al3(LH3)6] complex. The entropy 10 for [(„AlOH2)2+ ] was less than half that of the inner2 (LH2) sphere surface complexes due to the water molecules coordinated between IP6 and the surface and the relatively large charge on the LH10 species. Large entropies have also been reported for the com2 plexes involved in mellitic acid adsorption [4,5]. The entropies found for sorption of IP6 are more than double those for sorption of mellitic acid, supporting the idea that solvent effects play an important role in the adsorption of highly charged organic anions, irrespective of the enthalpies of sorption. 5. Conclusion The sorption of IP6 onto gibbsite was investigated by both macroscopic and spectroscopic techniques and also surface complexation modeling. Collectively these data allow the prediction of possible surface reactions involved in the binding of IP6 to

M. Ruyter-Hooley et al. / Journal of Colloid and Interface Science 440 (2015) 282–291

gibbsite. The pooled data indicate that sorption of IP6 occurs predominately through the formation of inner-sphere complexes or surface precipitates at lower pH. At higher pH the sorption appears to be more electrostatic in nature with outer-sphere complexation becoming important as pH increases. Sorption densities of IP6 on gibbsite are quite low compared to other oxides such as goethite, but sorption enthalpies and entropies are larger than those measured for other sorbates. The large positive entropy values suggest that changes in solvation of species play a major role in driving the sorption process. The combination of sorption experiments, 31P MAS NMR and surface complexation modeling proved to be a powerful means of predicting likely surface reactions. Each was necessary to fully understand the processes involved. Without the use of NMR spectroscopy to constrain surface species used in the model it would be difficult to choose the best candidates, while the model confirmed the likely presence of an outer-sphere complex that could not be definitely assigned from the NMR results. Guan et al. [24] and Yan et al. [58] also proposed inner-sphere complexation to be the dominant sorption mechanism for IP6 on amorphous aluminum hydroxide, as was found in this study at lower pH. However, neither found evidence of outer-sphere complexation at higher pH as is found in this work. In addition, while Yan et al. [58] proposed the formation of a bulk precipitate, our results suggest only surface precipitation. This difference most probably results from the amorphous nature of the substrate used in their study compared with the crystalline gibbsite used in our work. Future work will investigate the effect of IP6 on the sorption of metal ions on gibbsite with the aim of understanding the interactions occurring in soil and sediment systems which can affect the availability of both phosphate and metal ions to plants. References [1] G. Anderson, E.Z. Arlidge, Soil Sci. 13 (2) (1962) 216–224. [2] G. Anderson, E.G. Williams, J.O. Moir, Soil Sci. 25 (1) (1974) 51–62. [3] M.J. Angove, J.D. Wells, B.B. Johnson, J. Colloid Interface Sci. 211 (1999) 281– 290. [4] M.J. Angove, J.D. Wells, B.B. Johnson, J. Colloid Interface Sci. 296 (1) (2006) 30– 40. [5] M.J. Angove, J.D. Wells, B.B. Johnson, Langmuir 22 (9) (2006) 4208–4214. [6] J. Antelo, S. Fiol, C. Pérez, S. Mariño, F. Arce, D. Gondar, R. López, J. Colloid Interface Sci. 347 (1) (2010) 112–119. [7] L.G. Barrientos, P.N.N. Murthy, Carbohydr. Res. 296 (1996) 39–54, http:// dx.doi.org/10.1016/S0008-6215(96)00250-9. Copyright (C) 2012 American Chemical Society (ACS). All Rights Reserved. [8] N.J. Barrow, Soil Sci. 43 (1) (1992) 37–45. [9] A.S. Berg, B.C. Joern, J. Environ. Qual. 35 (5) (2006) 1855–1862. [10] J.-F. Boily, N. Nilsson, P. Persson, S. Sjoeberg, Langmuir 16 (2000) 5719–5729, http://dx.doi.org/10.1021/la991407o. Copyright (C) American Chemical Society (ACS). All Rights Reserved. [11] J.-F. Boily, P. Persson, S. Sjöberg, Geochim. Cosmochim. Acta 64 (20) (2000) 3453–3470. [12] J. Bowden, S. Nagarajah, N. Barrow, A. Posner, J. Quirk, Aust. J. Soil Res. 18 (1980) 49–60. [13] L. Celi, G. De Luca, E. Barberis, Soil Sci. 168 (7) (2003) 479–488. [14] L. Celi, S. Lamacchia, F.A. Marsan, E. Barberis, Soil Sci. 164 (8) (1999) 574–585. [15] L. Celi, M. Presta, F. Ajmore-Marsan, E. Barberis, Soil Sci. Soc. Am. J. 65 (3) (2001) 753–760. [16] F. Crea, C. Stefano, D. Milea, S. Sammartano, J. Solution Chem. 38 (1) (2009) 115–134. [17] C. De Stefano, D. Milea, S. Sammartano, J. Chem. Eng. Data 48 (2003) 114–119, http://dx.doi.org/10.1021/je020124m. Copyright (C American Chemical Society (ACS). All Rights Reserved.

291

[18] C. De Stefano, D. Milea, S. Sammartano, Thermochim. Acta 423 (1–2) (2004) 63–69. [19] J.D. Filius, T. Hiemstra, W.H. Van Riemsdijk, J. Colloid Interface Sci. 195 (2) (1997) 368–380. [20] C. Giaveno, L. Celi, R.M.A. Cessa, M. Prati, E. Bonifacio, E. Barberis, Soil Sci. 173 (10) (2008) 694–706. [21] C. Giles, B. Cade-Menun, J. Hill, Can. J. Soil Sci. 91 (3) (2011) 397–416. [22] S. Goldberg, Soil Sci. 175 (3) (2010) 105–110, http://dx.doi.org/10.1097/ SS.1090b1013e3181d3462f. [23] X.-H. Guan, G.-H. Chen, C. Shang, Soil Sci. 170 (5) (2005) 340–349. [24] X.-H. Guan, C. Shang, J. Zhu, G.-H. Chen, J. Colloid Interface Sci. 293 (2) (2006) 296–302. [25] A.F. Harrison, Soil Organic Phosphorus: A Review of World Literature, C.A.B International, United Kingdom, 1987. [26] T. Hiemstra, W.H. Van Riemsdijk, J. Colloid Interface Sci. 179 (1996) 488–508, http://dx.doi.org/10.1006/jcis.1996.0242. no. Copyright (C) 2013 American Chemical Society (ACS). All Rights Reserved. [27] T. Hiemstra, H. Yong, W.H. Van Riemsdijk, Langmuir 15 (18) (1999) 5942– 5955. [28] Y.S. Hwang, J.J. Lenhart, J. Colloid Interface Sci. 336 (1) (2009) 200–207. [29] B.B. Johnson, Environ. Sci. Technol. 24 (1) (1990) 112–118. [30] B.B. Johnson, A.V. Ivanov, O.N. Antzutkin, W. Forsling, Langmuir 18 (4) (2002) 1104–1111. [31] B.B. Johnson, E. Quill, M.J. Angove, J. Colloid Interface Sci. 367 (1) (2012) 436– 442. [32] B.B. Johnson, S. Sjöberg, P. Persson, Langmuir 20 (3) (2004) 823–828. [33] A.K. Karamalidis, D.A. Dzombak, Surface Complexation Modeling: Gibbsite, first ed., Wiley, Hoboken, 2010. [34] M. Karlsson, J. Lindgren, ‘WinSGW: SolgasWater for Windows X2’ (computer program), 2007. [35] Y. Kim, R.J. Kirkpatrick, Eur. J. Soil Sci. 55 (2) (2004) 243–251. [36] K. Lackovic, M.J. Angove, J.D. Wells, B.B. Johnson, J. Colloid Interface Sci. 257 (1) (2003) 31–40. [37] K. Lackovic, M.J. Angove, J.D. Wells, B.B. Johnson, J. Colloid Interface Sci. 269 (1) (2004) 37–45. [38] K. Lackovic, B.B. Johnson, M.J. Angove, J.D. Wells, J. Colloid Interface Sci. 267 (1) (2003) 49–59. [39] W. Li, J. Feng, K.D. Kwon, J.D. Kubicki, B.L. Phillips, Langmuir 26 (7) (2010) 4753–4761. [40] W. Li, X. Feng, Y. Yan, D.L. Sparks, B.L. Phillips, Environ. Sci. Technol. 47 (15) (2013) 8308–8315. [41] W. Li, A.-M. Pierre-Louis, K.D. Kwon, J.D. Kubicki, D.R. Strongin, B.L. Phillips, Geochim. Cosmochim. Acta 107 (2013) 252–266. [42] D.R. Lide, in: D.R. Lide (Ed.), Handbook of Chemistry and Physics, 82nd ed., CRC Press LLC, USA, 2001. [43] R. Lookman, P. Grobet, R. Merckx, W.H. Van Riemsdijk, Geoderma 80 (3–4) (1997) 369–388. [44] R. Lookman, P. Grobet, R. Merckx, K. Vlassak, Eur. J. Soil Sci. 45 (1) (1994) 37– 44. [45] N. Nilsson, P. Persson, L. Lövgren, S. Sjöberg, Geochim. Cosmochim. Acta 60 (22) (1996) 4385–4395. [46] J. Nordin, P. Persson, A. Nordin, S. Sjöberg, Langmuir 14 (13) (1998) 3655– 3662. [47] M. Ognalaga, E. Frossard, F. Thomas, Soil Sci. Soc. Am. J. 58 (2) (1994) 332–337. [48] P. Persson, T. Andersson, H. Nelson, S. Sjöberg, R. Giesler, L. Lövgren, J. Colloid Interface Sci. 386 (1) (2012) 350–358. [49] R.P.J.J. Rietra, T. Hiemstra, W.H. van Riemsdijk, Environ. Sci. Technol. 35 (16) (2001) 3369–3374. [50] J. Rosenqvist, P. Persson, S. Sjöberg, Langmuir 18 (12) (2002) 4598–4604. [51] SPSS Science Inc. PeakFit Separation and Analysis Software, v. 4.0, Chicago, 1997. [52] W. Stumm, L. Sigg, B. Sulzberger, Chemistry of the Solid-Water Interface. Processes at the Mineral-Water and Particle-Water Interface in Natural Systems, Wiley, 1992. [53] G.L. Turner, K.A. Smith, R.J. Kirkpatrick, E. Oldfieldt, J. Magn. Reson. 70 (3) (1986) 408–415 (1969). [54] T.J. Van Emmerik, D.E. Sandström, O.N. Antzutkin, M.J. Angove, B.B. Johnson, Langmuir 23 (6) (2007) 3205. [55] D.B. Ward, P.V. Brady, Clays Clay Miner. 46 (4) (1998) 453–465. [56] R. Weerasooriya, D. Aluthpatabendi, H.J. Tobschall, Colloids Surf., A 189 (1–3) (2001) 131–144. [57] J.D. Wolt, Soil Solution Chemistry: Applications to Environmental Science and Agriculture, Wiley, 1994. [58] Y. Yan, W. Li, J. Yang, A. Zheng, F. Liu, X. Feng, D.L. Sparks, Environ. Sci. Technol. 48 (12) (2014) 6735–6742.