K+

10240

Langmuir 2008, 24, 10240-10244

Cation Exchange at the Mineral-Water Interface: H3O+/K+ Competition at the Surface of Nano-Muscovite Geoffrey M. Bowers,*,†,§ David L. Bish,‡ and R. James Kirkpatrick†,| Department of Geology, UniVersity of Illinois Urbana-Champaign, 1301 West Green Street, Urbana, Illinois 61801, and Department of Geological Sciences, Indiana UniVersity, 1001 East 10th Street, Bloomington, Indiana 47405 ReceiVed July 4, 2008. ReVised Manuscript ReceiVed July 31, 2008 This article describes a 39K nuclear magnetic resonance (NMR) spectroscopic study of K+ displacement at the muscovite/water interface as a function of aqueous phase pH. 39K NMR spectra and T2 relaxation data for nanocrystalline muscovite wet with a solid/solution weight ratio of 1 at pH 1, 3, and 5.5 show substantial liquid-like K+ only at pH 1. At pH 3 and 5.5, all K+ appears to be associated with muscovite as inner- or outer-sphere complexes, indicating that H3O+ does not displace basal surface K+ beyond the 39K detection limit under these conditions. In our pH 1 mixture, only ∼1/3 of the initial basal surface K+ population is located more than 3-4 Å from the surface. 29Si and 27Al MAS NMR spectra and SEM images show no evidence of dissolution during the 39K experiments, consistent with the liquid-like 39K fraction originating from displaced basal surface K+. Assuming no muscovite dissolution or interlayer exchange, the K+/H3O+ ratio relevant to the solution/surface exchange equilibrium is controlled by the total amount of K+ on the surface and H3O+ in solution (K+surf/H3O+aq). These parameters, in turn, depend on the basal surface area, solution pH, and the solid/solution ratio. The results here are consistent with significant displacement of surface K+ only under conditions where the initial K+surf/H3O+aq ratio is less than approximately 1. Computational molecular models of the muscovite/water interface should account for both K+ and H3O+ in the near-surface region.

Introduction Molecularly flat basal (001) surfaces of muscovite mica (KAl2(AlSi3)O10(OH)2) are readily produced by mechanical cleavage,1 making this a commonly used model surface for experimental and computational studies of interfacial processes. In the past decade, for instance, muscovite (001) has been used to explore the interaction of organic matter,2,3 inorganic and organic cations,4-11 and biomolecules12-14 with hydrophilic oxide surfaces and in experimental and computational studies of the * To whom correspondence should be addressed. E-mail: bowersg@ msu.edu. † University of Illinois, Urbana-Champaign. ‡ Indiana University. § Currently affiliated with the Department of Chemistry, Michigan State University, East Lansing, Michigan, 48824. | Currently affiliated with the College of Natural Science, Michigan State University, East Lansing, Michigan, 48824.

(1) Gains, G. L. J. Phys. Chem. 1957, 61, 1408. (2) Tugulea, A. M.; Oliver, D. R.; Thomson, D. J.; Hawthorne, F. C. Spec. Publ. R. Soc. Chem. 2001, 273, 241–251. (3) Shevchenko, S. M.; Bailey, G. W. Supramol. Sci. 1998, 5(1-2), 143–157. (4) Park, C.; Fenter, P. A.; Nagy, K. L.; Sturchio, N. C. Phys. ReV. Lett. 2006, 97(1), 016101. (5) Lee, S. S.; Nagy, K. L.; Fenter, P.; Sturchio, N. C. Geochim. Cosmochim. Acta 2005, 69(10), A491-A491. (6) Fenter, P.; Park, C.; Sturchio, N. C. Geochim. Cosmochim. Acta 2008, 72(7), 1848–1863. (7) Lee, S. S.; Nagy, K. L.; Fenter, P. Geochim. Cosmochim. Acta 2007, 71(23), 5763–5781. (8) Pagnanelli, F.; Bornoroni, L.; Moscardini, E.; Toro, L. Chemosphere 2006, 63(7), 1063–1073. (9) Osman, M. A.; Suter, U. W. J. Colloid Interface Sci. 1999, 214(2), 400– 406. (10) Zachara, J. M.; Smith, S. C.; Liu, C. X.; McKinley, J. P.; Serne, R. J.; Gassman, P. L. Geochim. Cosmochim. Acta 2002, 66(2), 193–211. (11) Schlegel, M. L.; Nagy, K. L.; Fenter, P.; Cheng, L.; Sturchio, N. C.; Jacobsen, S. D. Geochim. Cosmochim. Acta 2006, 70(14), 3549–3565. (12) Kunstelj, K.; Federiconi, F.; Spindler, L.; Drevensek-Olenik, I. Colloid Surf., B 2007, 59(2), 120–127. (13) Rivetti, C.; Guthold, M.; Bustamante, C. J. Mol. Biol. 1996, 264(5), 919– 932. (14) Vasina, E. N.; Dejardin, P.; Rezaei, H.; Grosclaude, J.; Quiquampoix, H. Biomacromolecules 2005, 6(6), 3425–3432.

structure and dynamics of H2O at mineral-water interfaces.15-19 The negative structural charge of the muscovite aluminosilicate layers is balanced by surface and interlayer K+, and a comprehensive understanding of the muscovite/water interface requires that we understand the roles of K+, H2O, and H3O+ at and near the surface. Comparison of recent experimental and computational molecular modeling results shows that current knowledge about these interactions is incomplete.15,16,18 For example, in 2001 Cheng et al. used the results of surface X-ray reflectivity (SXRR) experiments to develop an atomic density profile (ADP) in the aqueous phase above a 625 mm2 muscovite (001) surface in contact with a relatively large volume (∼15 mL) of deionized H2O in equilibrium with atmospheric CO2 (pH ≈ 5.7).15 Based on the previously published suggestion that H3O+ and H2O fully replace basal surface K+ at this pH,1,20,21 they interpreted their results using a model with two structural types of ordered water (defined as H2O and/or H3O+) within 3-4 Å of the surface. More recent Monte Carlo (MC) simulations of this interface by Park and Sposito18 and subsequent molecular dynamics (MD) simulations by Wang et al.16 show a more complicated interfacial structure. Both sets of calculations suggest that some combination of surface K+, H3O+, and H2O is required for a satisfactory fit to the experimentally derived ADP. However, the MC and MD results disagree regarding the most likely position(s) of the K+ ions and where they contribute intensity to the ADP. In this publication, we provide new insight into the pHdependent displacement of K+ at the muscovite-water interface (15) Cheng, L.; Fenter, P.; Nagy, K. L.; Schlegel, M. L.; Sturchio, N. C. Phys. ReV. Lett. 2001, 8715(15), 156103. (16) Wang, J. W.; Kalinichev, A. G.; Kirkpatrick, R. J.; Cygan, R. T. J. Phys. Chem. B 2005, 109(33), 15893–15905. (17) Cantrell, W.; Ewing, G. E. J. Phys. Chem. B 2001, 105(23), 5434–5439. (18) Park, S. H.; Sposito, G. Phys. ReV. Lett. 2002, 89(8), 085501. (19) Balmer, T. E.; Christenson, H. K.; Spencer, N. D.; Heuberger, M. Langmuir 2007, (20) Pashley, R. M. J. Colloid Interface Sci. 1981, 83(2), 531–546. (21) Pashley, R. M. J. Colloid Interface Sci. 1981, 80(1), 153–162.

10.1021/la8021112 CCC: $40.75  2008 American Chemical Society Published on Web 08/21/2008

Cation Exchange at the Mineral-Water Interface

using 39K nuclear magnetic resonance (NMR) spectroscopy. NMR is element specific and unique in its ability to simultaneously probe the molecular-scale structure and dynamics of surface species.22-27 Because the NMR-active K isotopes (39K, 41K) exhibit poor sensitivity relative to other cationic nuclei such as 23Na and 133Cs,28 NMR has not been used extensively to study K+ behavior in phyllosilicates28,29 or at phyllosilicate-water interfaces.27 Despite recent advances in instrument technology that make such studies more feasible,28 NMR is, in general, an insensitive technique. Thus, in this study we use synthetic nanocrystalline muscovite to provide a large surface 39K population and a low concentration of paramagnetic impurities, the latter of which can lead to significant line-broadening that further limits sensitivity in natural samples. We are also limited to higher solid/solution ratios than those of other experimental studies (g1 vs ,1) because of the need to avoid potential interferences from bulk solution not in contact with the solid.22 We note that solid/solution ratios greater than 1 are common in many rocks and soils, and, thus, our experiments may more directly reflect some natural conditions than those with smaller solid/solution ratios. In our experiments, solution-like K+ occurs only when the molar ratio of total K+ on the surface to H3O+ in solution (K+surf/ H3O+aq) is ∼1 and that, within detection limits, all of the K+ remains associated with the muscovite surface when K+surf/H3O+aq is .1. Supporting scanning electron microscopy (SEM) and 27Al and 29Si magic angle spinning (MAS) NMR show no evidence of dissolution at these pHs on the microscopic or molecular-levels, indicating that the source of the solution-like K+ is displaced basal surface K+ ions. Two important implications of our results are that K+ must be considered when interpreting experimental studies of the aqueous muscovite-water interface and that accurate computational molecular modeling of processes at this interface requires a structural model including both K+ and H3O+ to balance the structural charge of the muscovite layers. Assuming no muscovite dissolution or interlayer exchange, when no K+ is present in the initial solution phase, the K+/H3O+ ratio relevant to the solution/surface exchange equilibrium is controlled by the total amount of K+ on the surface and H3O+ in solution (K+surf/H3O+aq). These parameters in turn depend on the basal surface area, solution pH, and the solid/solution ratio. The results here are consistent with significant displacement of surface K+ only under conditions where the initial total K+surf/H3O+aq ratio is less than approximately 1. This ratio must be considered when interpreting and comparing experimental results.

Materials and Methods Nanocrystalline muscovite was synthesized hydrothermally from a muscovite-composition gel30 at 400 °C and 1 kbar, conditions expected to produce primarily 2M1 muscovite.31 The gel and excess (22) Yu, P.; Kirkpatrick, R. J. Cem. Concr. Res. 2001, 31(10), 1479–1485. (23) Weiss, C. A.; Kirkpatrick, R. J.; Altaner, S. P. Geochim. Cosmochim. Acta 1990, 54(6), 1655–1669. (24) Kim, Y.; Kirkpatrick, R. J. Am. Mineral. 1998, 83(5-6), 661–665. (25) Kim, Y.; Kirkpatrick, R. J.; Cygan, R. T. Geochim. Cosmochim. Ac. 1996, 60(21), 4059–4074. (26) Kim, Y.; Kirkpatrick, R. J. Geochim. Cosmochim. Acta 1997, 61(24), 5199–5208. (27) Bowers, G. M.; Bish, D. L.; Kirkpatrick, R. J. J. Phys. Chem. C 2008, 112, 6430–6438. (28) Stebbins, J. F.; Du, L. S.; Kroeker, S.; Neuhoff, P.; Rice, D.; Frye, J.; Jakobsen, H. J. Solid State Nucl. Mag. 2002, 21(1-2), 105–115. (29) Lambert, J. F.; Prost, R.; Smith, M. E. Clays Clay Min. 1992, 40(3), 253–261. (30) Hamilton, D. L.; Henderso.Cm, Mineral. Mag. J. M. Soc. 1968, 36(282), 832. (31) Yoder, H. S.; Eugster, H. P. Geochim. Cosmochim. Acta 1955, 8, 225– 242.

Langmuir, Vol. 24, No. 18, 2008 10241 H2O were held in seven different 3 mm diameter gold capsules that were sealed by welding and reacted for five days in a cold-seal pressure vessel. The reaction products were examined by SEM, powder X-ray diffraction (XRD) with and without glycolation, and 27Al and 29Si MAS NMR. The unreacted gel was also characterized using 27Al and 29Si MAS NMR methods. To investigate displacement of surface K+ by H3O+, dried and unground synthetic mica powders were packed in 3 cm long, 4 mm outer diameter sodium borosilicate glass NMR tubes with each tube containing the products of a single gold capsule. One tube contained powder in equilibrium with the atmosphere (∼50% relative humidity, denoted as reacted powder in the remainder of the text). The other three samples were wet at a solid-to-solution weight ratio of 1:1 with deionized H2O in equilibrium with the atmosphere (pH ≈ 5.5) or adjusted to pH 1 or 3 using concentrated HCl. This amount of solution is just sufficient to fill the void volume between the solid particles with no liquid-rich supernatant phase. The packed NMR tubes were sealed using water-tight epoxy to prevent water loss. 39K NMR spectra were obtained using a 14.1 T narrow-bore Infinity Plus instrument at the University of Illinois and a Hahn-echo pulse sequence employing a phase cycle that cancels quadrupolar-related imperfections.32 Spectra were collected with both central-transition (CT) selective and nonselective pulse widths, the latter of which suppresses signal from rigid spins for an I ) 3/2 nucleus like 39K. Identical experiments using CT selective pulses were performed with KCl (aq) solutions to determine the limit-of-detection (LOD) for liquid-like 39K+ under these conditions. The 39K T2 relaxation dynamics were examined using quadrupolar Carr-Purcell-MeiboomGill experiments33,34 and variable dephasing time echo experiments. All experiments were performed using a recycle delay of 500 ms, sufficiently long to permit full relaxation of the liquid-like 39K (T1 ) 62 ms) and surface-sorbed/bulk 39K (T1,CT ) 66 ms). Following the suite of 39K NMR investigations, samples were removed from the sealed NMR tubes and dried in an oven at 120 °C for approximately 23 h. The dried mixtures were examined for evidence of dissolution by 29Si and 27Al MAS NMR and SEM. Additional details on the synthesis and NMR experiments are described in the Supporting Information.

Results and Discussion Material Characterization. XRD shows that the reacted samples contain primarily 2M1 muscovite mica with a small amount of 1 M muscovite and a minor poorly crystalline phase (Figure 1). Optimum Rietveld refinement of the diffraction pattern is obtained assuming predominantly 2M1 muscovite, 2.5% impurity 1 M muscovite, and an average crystallite size for these phases over three dimensions of 17.3 ( 0.5 nm. The broad diffraction “hump” between 20° and 30° 2θ is due mostly to overlapping mica reflections, although there may be some diffraction intensity from a small amount of poorly crystalline material. The XRD patterns are nearly identical before and after glycolation, indicating that there is no detectable expandable material (smectite) (Figure S1). Figure S2 in the Supporting Information shows that the XRD patterns of each individual capsule are similar to each other and to the XRD pattern of the three combined samples in Figure 1. SEM micrographs of the unground reaction products show principally featureless particles, although a few have visible platelet morphologies under higher magnification (Figure S3). The 27Al MAS NMR spectra of our muscovite are very similar to those of natural muscovite35 but show a small resonance for five-coordinate aluminum (Al[5]) near 29 ppm (Figure 2, Table (32) Kunwar, A. C.; Turner, G. L.; Oldfield, E. J. Magn. Reson. 1986, 69(1), 124–7. (33) Cheng, J. T.; Ellis, P. D. J. Phys. Chem. 1989, 93(6), 2549–55. (34) Larsen, F. H.; Jakobsen, H. J.; Ellis, P. D.; Nielsen, N. C. J. Phys. Chem. A 1997, 101(46), 8597–8606. (35) Sanz, J.; Serratosa, J. J. Am. Chem. Soc. 1984, 106, 4790–4793.

10242 Langmuir, Vol. 24, No. 18, 2008

Figure 1. X-ray diffraction pattern of three combined reacted gel samples prior to glycolation. The stick patterns reflect the ICDD reference data for 1 M (red) and 2M1 (blue) muscovite mica.

Figure 2. 27Al (left) and 29Si (right) MAS NMR spectra of the reaction product examined at pH 1 after acquisition of the 39K NMR spectra and drying (a); the atmosphere-equilibrated “raw” reaction product powder (b); and the unreacted gel used for synthesis (c). Spinning sidebands in the 27Al MAS NMR are denoted with an *.

S1). The detectable Al[5] resonance is probably not due to residual unreacted gel because peaks corresponding to the four- and sixcoordinate 27Al (Al[4] and Al[6]) resonances of the gel are absent. The Al[5] may be on edge sites of muscovite layers and visible due to the small particle size36,37 or in the poorly crystalline material suggested by XRD. The observed tailing of the 27Al peaks to low frequency is commonly observed for disordered materials like clays and micas28,38 and reflects a distribution of 27Al quadrupolar couplings due to the numerous local structural environments. The 27Al[4] and 27Al[6] isotropic chemical shifts are statistically identical to those of muscovite (Table S1), but the Al[4]/(Al[4]+Al[5]+Al[6]) intensity ratio of the raw powder (36) McHale, J. M.; Navrotsky, A.; Kirkpatrick, R. J. Chem. Mater. 1998, 10(4), 1083–1090. (37) Coster, D. J.; Fripiat, J. J.; Muscas, M.; Auroux, A. Langmuir 1995, 11(7), 2615–2620. (38) Xu, X.; Kirkpatrick, R. J. J. Membr. Sci. 2006, 280(1-2), 226–233.

Bowers et al.

obtained from NMR (Figure 2b) is 0.24 ( 0.04, less than expected based on the theoretical muscovite composition (0.33).39 This observation may reflect the presence of some material with a lower layer charge than muscovite, but caution is required in interpreting the 27Al peak integrals, because they may be affected by differences in the quadrupolar couplings, distribution of quadrupolar interactions, or T2 relaxation rates for different sites.35 The 29Si MAS NMR spectra of the muscovite samples (Figure 2) differ substantially from that of the unreacted gel and show that the majority of the powder is muscovite, in agreement with the XRD data. There are four resolvable 29Si NMR peak maxima in the Q3 region, a triad at approximately -82, -86, and -90 ppm representing the Q3(2Al), Q3(1Al), and Q3 (0Al) sites of muscovite,35 respectively, and a peak near -94 ppm indicating the presence of Q3(0Al) sites in layers with higher Si/Al ratios (lower tetrahedral layer charge) than ideal muscovite.40 Unique fitting of the spectra with these four peaks or with an additional peak near -88.5 ppm for Q3(1Al) sites in lower charge layers40 is not possible due to peak overlap, thus, we cannot estimate the Si/Al ratio in the tetrahedral layers accurately from the 29Si NMR results. Despite the absence of smectite reflections in the XRD pattern, the -94 ppm resonance is sharper than typically observed for highly disordered materials such as glasses and gels, and the 29Si MAS NMR spectra of the reaction products are similar to those of mixed-layer Illite-smectites.40 The lower-charge material is most likely smectite with poor long-range order and may correspond to the poorly crystallized phase suggested by XRD. The presence of a second, lower-charge layered aluminosilicate with 27Al resonances that overlap with those of muscovite is also consistent with the Al[4]/(Al[4]+Al[5]+Al[6]) intensity ratio of 0.24 described above. A small amount of low-charge phyllosilicate should not affect the 39K NMR results significantly, since it has much less K+ than muscovite. K+ Behavior. The 39K NMR results show appreciable liquidlike K+ only in the pH 1 liquid-solid mixture, indicating that all or most of the K+ remains associated with the muscovite surface as inner- or outer-sphere species under our experimental conditions. The 39K NMR spectra of the mixtures collected with central-transition selective pulse widths (π/2 pulses for rigid 39K) are similar to the spectrum of the reacted powder (Figure 3) and the room-temperature 39K spectra reported previously for phyllosilicates with low K+ mobility.27-29 The spectra of the mixtures contain a broad resonance (full width at half-height ) 107 ( 2 ppm) that tails to low frequency and has an isotropic shift of -7.2 ( 1.7 ppm. For our samples, most of the intensity for this resonance comes from inner-sphere K+ in muscovite interlayers, but a significant fraction comes from surface sites (Table 1). It is not possible to determine the specific binding environment(s) of surface K+ using our NMR results alone, because our previous 39K NMR study of the related smectite mineral hectorite shows that inner- and outer-sphere K+ within 3-4 Å of the basal surface yield similar and highly overlapped resonances.27 The only significant change in line-shape with pH is the presence of a weak, sharp feature near the reference frequency at pH 1. 39K NMR echo experiments with nonselective pulse widths show that this sharp feature arises from a K+ population with isotropic mobility and a solution-like chemical shift (-1.0 ( 1.7 ppm; Figure 3a). We assign this signal to fully solvated K+ in extra-particle H2O. This feature is not visible in the rigid-suppressed spectra of the pH 3.0 and 5.5 samples, showing that fewer than 6 × 1016 39K spins (the liquid-like limit of detection) are displaced in these samples. (39) Woessner, D. E. Am. Mineral. 1989, 74(1-2), 203–215. (40) Altaner, S. P.; Weiss, C. A.; Kirkpatrick, R. J. Nature 1988, 331(6158), 699–702.

Cation Exchange at the Mineral-Water Interface

Figure 3. 39K static echo NMR spectra of the pH 1 solid/solution mixture (a); the pH 3 solid/solution mixture (b); the pH 5.5 solid/solution mixture (c); and the atmosphereequilibrated “raw” powder (d). In each case, the top spectrum was acquired with central-transition selective pulse widths (rigid limit) and the bottom with liquid-selective pulse widths. 27Al and 29Si MAS NMR and SEM examination of the samples after acquisition of the 39K NMR spectra reveals no evidence of dissolution, suggesting that the liquid-like fraction at pH ) 1 arises from basal surface K+ displaced by H3O+ ions. The 27Al and 29Si MAS NMR spectra of the solids after collection of the 39K spectra have peak intensities and chemical shifts statistically identical to the original reacted powder (Figure 2; Tables S1, S2). In addition, there are no new 27Al and 29Si NMR peaks that would indicate precipitation of secondary phases while water was in contact with the samples, such as the precipitation of amorphous hydrous silica that occurs during acid dissolution of the phyllosilicate mineral hectorite.41 A more detailed analysis of the 39K NMR results shows that the muscovite surface has a strong affinity for K+ and that even at pH 1, where there is sufficient H3O+ to achieve full K+ displacement, approximately 2/3 of the initial exposed basal surface K+ remains associated with the surface. This result is consistent with our previous 39K NMR observations for hectorite,27 which show a strong association of K+ with the basal surfaces even though hectorite has a significantly lower layercharge than muscovite. For our muscovite, the 39K T2* relaxation

(41) Komadel, P.; Madejova, J.; Janek, M.; Gates, W. P.; Kirkpatrick, R. J.; Stucki, J. W. Clays Clay Min. 1996, 44(2), 228–236.

Langmuir, Vol. 24, No. 18, 2008 10243

measurements show only single-exponential behavior with a wellshimmed sample, suggesting that the true T2’s are similar for the solution and surface-associated 39K spins. This allows us to determine the relative population of displaced 39K spins based on the peak integrals in the echo spectra. Accounting for the 39K satellite/central transition intensity ratio of 3:242 and effectively π/4, π/2 pulse widths for the liquid-like spins, our data show that 2.1 ( 0.1% of the 39K spins have liquid-like mobility at pH 1. Assuming the entire sample is composed of muscovite, that the c-dimension of each particle is 170 Å (the average particle size determined from XRD), and that the a and b dimensions are also 170 Å, this represents approximately 1/3 of the available basal surface 39K spins. Note that these assumptions may underestimate the basal surface K+ population if the crystallites have plate-like morphologies with a- and b-dimensions larger than the cdimension. With our assumptions, at pH 1 the initial K+surf/ H3O+aq ratio is ∼1 (Table 1), suggesting that sufficient H3O+ is present to displace all of the basal K+, whereas only approximately 1/3 is displaced beyond 3-4 Å from the surface. At pH 3, the initial K+surf/H3O+aq ratio is ∼150, and if 100% of the H3O+ were to displace basal K+, the liquid-like K+ population is very close to our detectable limit of 6 × 1016 39K spins. The lack of a clear liquid-like feature in the pH 3 spectrum suggests that surface sorption of all the H3O+ ions does not occur, but we cannot rule out the displacement of fewer than 6 × 1016 39K spins. At pH 5.5, we do not expect to observe K+ displacement even if all of the H3O+ were to replace basal K+ ions. Summary and Implications. Together, the NMR and XRD results provide strong evidence that a significant fraction of the basal K+ remains near the muscovite surface as inner- or outersphere species when the initial basal K+/solution H3O+ ratio is g1 and the solid/solution ratio is relative large. In the natural environment, these conditions are most likely to apply to low porosity materials containing neutral to basic pH water and with small water fluxes. Assuming no dissolution, the initial K+surf/ H3O+aq ratio depends on the basal surface area of the muscovite particles and volume of solution. As mentioned earlier, in experiments using other analytical methods, the surface area/ solution volume ratio may be substantially larger than in our NMR experiments and the initial K+surf/H3O+aq ratio consequently much smaller than ours. For example, based on the muscovite surface area, initial solution pH, and the thickness of the water film used in the SXRR experiments of Cheng et al.,15 their initial K+surf/H3O+aq ratio was ∼0.20. Under these conditions there may have been significantly more displacement of surface K+ by H3O+ than we observed in our experiments. Unfortunately, 39K NMR may not be capable of exploring lower solid/solution ratios and lower initial K+surf/H3O+aq ratios simultaneously. Examining lower initial K+surf/H3O+aq ratios while maintaining a solid/solution ratio of approximately one would require a solution pH less than 1, meaning that dissolution is likely to be a significant concern. Reducing the solid/solution ratio will reduce the number of 39K spins within the coil, thus reducing sensitivity and making experiments prohibitively long. If the solution phase is allowed to extend beyond the coil to maintain the same basal 39K spin population used in these studies, the ability to obtain quantitative structural information and accurate relaxation rates is compromised by the diffusion of 39K spins outside the excitable volume of the coil. Using a larger diameter probe head while reducing the solid/solution ratio could help, but magnetic field inhomogeneities and an anisotropic distribution of sample could make interpretation of the results difficult. (42) Abragam, A., The principles of nuclear magnetism; Clarendon Press: Oxford, 1961; p 599.

10244 Langmuir, Vol. 24, No. 18, 2008

Bowers et al.

Table 1. Populations of K+, 39K Spins, and H3O+ Molecules and the Initial Basal K+/Solution H3O+ Ratios for the Muscovite-Water Mixtures Examined by NMR initial solution pH property

pH 1

basal K basal 39K+ total K+ total 39K+ total H3O+ total K+/total H3O+ initial basal K+/solution H3O+

4.9 ( 0.3 × 4.5 ( 0.3 × 1018 8.3 ( 0.3 × 1019 7.7 ( 0.3 × 1019 3.7 ( 0.1 × 1018 22 1.3

+

a

pH 3

pH 5.5

7.7 ( 0.1 × 7.1 ( 0.1 × 1018 1.3 ( 0.1 × 1020 1.2 ( 0.1 × 1020 5.2 ( 0.1 × 1016 2500 150

1018

1018

5.1 ( 0.3 × 1018 4.7 ( 0.3 × 1018 8.7 ( 0.3 × 1019 8.1 ( 0.3 × 1019 1.1 × 1014a 790 000 46 000

Solution volume not recorded in this case; number of H3O+ molecules is approximate.

As described above, NMR appears unable to differentiate innerand outer-sphere K+ on phyllosilicate surfaces. Thus, the data here cannot effectively address how much K+ contributes to the different peaks in the various models of the atomic density profile for solution at the muscovite surface.15,16,18 The results are, however, consistent with the MC and MD simulation results that show K+ contributing to the atomic density maxima within 4 Å from the surface. Our results and those of the simulations suggest that modeling of processes at the mica-H2O interface require consideration of surface-sorbed K+ and H3O+. Similarly, the K+surf/H3O+aq ratio and the solution volumes used during rinsing, preparation, and in situ measurement must be considered when interpreting and comparing the results of experimental studies. Acknowledgment. This research was supported by the United States Department of Energy Office of Basic Energy Science,

Grant No. DE-FG01-05ER05-010201. We thank Prof. Craig Lundstrom for help with the hydrothermal synthesis and Drs. P. K. Babu and S. Mukherjee for acquiring the gel 27Al and 29Si NMR spectra. We also thank Dr. Mauro Sardela for obtaining the XRD patterns of the individual gold capsules and Dr. Andrey Kalinichev for many helpful discussions about processes at solidfluid interfaces and the MD simulations of the muscovite-water interface. Supporting Information Available: Details of synthesis procedure and NMR experiments, glycolated XRD pattern, XRD patterns of the individual gold capsules, SEM photos of materials, tables detailing the 27Al and 29Si peak positions, fit, and calculation results. This material is available free of charge via the Internet at http://pubs.acs.org. LA8021112