pubs.acs.org/Langmuir © 2010 American Chemical Society
Intercalated Water in Synthetic Fluorhectorite Clay R^omulo P. Tenorio,† M. Engelsberg,*,‡,† Jon Otto Fossum,§ and Geraldo J. da Silva^ Programa de P os-Graduac-a~o em Ci^ encia de Materiais, Universidade Federal de Pernambuco, Cidade Universit aria, 50670-901, Recife, Pernambuco, Brazil, ‡Departamento de Fı´sica, Universidade Federal de Pernambuco, Cidade Universit aria, 50670-901, Recife, Pernambuco, Brazil, §Department of Physics, Norwegian University of Science and Technology, Hoegskoleringen 5, N-7491, Trondheim, Norway, and ^Departamento de Fı´sica, Universidade de Brası´lia, Asa Norte, 70910-900, Brası´lia, Distrito Federal, Brazil †
Received January 26, 2010. Revised Manuscript Received March 30, 2010 7 Li and 1H nuclear magnetic resonance together with X-ray diffraction measurements in powdered samples and pseudocrystalline films of synthetic fluorhectorite as a function of relative ambient humidity permit to address several aspects of the structure and dynamics of intercalated water molecules. The role of proton exchange as a possibly dominant mechanism of charge transport in the one-water layer regime of hydration is reexamined. The experimental results in Li-fluorhectorite support the result of molecular simulations which predict, for Li-montmorillonite, the existence of an intermediate regime, between one-water layer and two-water layer states.
I. Introduction Clay minerals are one the most abundant materials in the earth’s crust and are also present in an enormous variety of man-made products and processes.1 Given that the phenomenon of swelling, which is one of the most remarkable characteristics of smectite clays, is controlled by the hydration of charge-compensating counterions present in the interlamellar space, the structure, and dynamics of intercalated water in clays has attracted considerable interest.2 Several experimental techniques such as X-ray diffraction,3-5 neutron diffraction,6-8 nuclear magnetic resonance,9-11 and infrared absorption12,13 have been employed, not always yielding consistent conclusions,14 to study the structure and dynamics of intercalated water. In addition to these experimental studies, molecular simulations have recently been able to yield quite detailed information concerning the structure of water intercalated in the *To whom correspondence should be addressed. E-mail: mario@ df.ufpe.br. (1) Sposito, G.; Skipper, N. T.; Sutton, R.; Park, S.; Soper, A. K.; Greathouse, J. A. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 3358. (2) Sposito, G.; Prost, R. Chem. Rev. 1982, 82, 553. (3) Ferrage, E.; Lanson, B.; Malikova, N.; Planc-on, A.; Sakharov, B. A.; Drits, V. A. Chem. Mater. 2005, 17, 3499. (4) da Silva, G. J.; Fossum, J. O.; DiMasi, E.; Ma˚løy, K. J.; Lutnæs, S. B. Phys. Rev. E 2002, 66, 011303. (5) da Silva, G. J.; Fossum, J. O.; DiMasi, E.; Ma˚løy, K. J.; Lutnæs, S. B. Phys. Rev. B 2003, 67, 094114. (6) Powell, H. D.; Fischer, H. E.; Skipper, N. T. J. Phys. Chem. B 1998, 102, 10899. (7) Bordallo, H. N.; Aldridge, L. P.; Churchman, G. J.; Gates, W. P.; Telling, M. T. F.; Kiefer, K.; Fouquet, P.; Seydel, T.; Kimber, S. A. J. J. Phys. Chem. C 2008, 112, 13982. (8) Malikova, N.; Cadne, A.; Dubois, E.; Marry, V.; Durand-Vidal, S.; Turq, P.; Breu, J.; Longeville, S.; Zanotti, J.-M. J. Phys. Chem. C 2007, 111, 17603. (9) Tenorio, R. P.; Alme, L. R.; Engelsberg, M.; Fossum, J. O.; Hallwass, F. J. Phys. Chem. C 2008, 112, 575. (10) Sanz, J.; Herrero, C. P.; Serratosa, J. M. J. Phys. Chem. B 2006, 110, 7813. (11) Hougardy, J.; Stone, W. E. E.; Fripiat, J. J. J. Chem. Phys. 1976, 64, 3840. (12) Prost, R. Ann. Agron. 1975, 26, 400. (13) Suquet, H.; Prost, R.; Pezerat, H. Clay Miner. 1977, 12, 113. (14) Greathouse, J.; Sposito, G. J. Phys. Chem. B 1998, 102, 2406. (15) Tambach, T. J.; Bolhuis, P. G.; Hensen, E. J. M.; Smit, B. Langmuir 2006, 22, 1223. (16) Tambach, T. J.; Hensen, E. J. M.; Smit, B. J. Phys. Chem. B 2004, 108, 7586. (17) Hensen, E. J. M.; Smit, B. J. Phys. Chem. B 2002, 106, 12664. (18) Salles, F.; Bildstein, O.; Douillard, J.-M.; Jullien, M.; Van Damme, H. J. Phys. Chem. C 2007, 111, 13170.
Langmuir 2010, 26(12), 9703–9709
interlamellar space, especially in montmorillonite clays.15-18 However, detailed tests focused on experimental confirmation of some of the more recent molecular simulation predictions, and their validity in different clays and for various exchangeable counterions, appear to still be lacking. In the present work we employ nuclear magnetic resonance (NMR) spectroscopy and wide-angle X-ray scattering (WAXS) to study the structure and dynamics of water intercalated in synthetic fluorhectorite. Fluorhrectorite is a 2:1 layer silicate where a fraction of Mg2þ ions is substituted by Liþ in trioctahedral sites resulting in a structural negative charge which is compensated by exchangeable hydrated cations.19 Instead, the structure of montmorillonite, which is also 2:1 clay, exhibits a substitution of Al3þ by Mg2þ in dioctahedral sites.1 In most of our measurements the exchangeable cation was Liþ (Li-fluorhectorite), but some new data on Na-fluorhectorite are also presented to complement earlier results in this material.9 One distinctive aspect of 1H and 7Li NMR spectra as a function of temperature and relative humidity (RH) is that they permit to probe the motion of water proton pairs and counterions averaged over a time scale which is long compared to what is normally accessible by molecular simulations. Another point that can be addressed by our experimental results is the long-standing issue of the role of proton exchange as a mechanism of charge transport in clays.19-23 A predominance of proton exchange over cationic diffusion in the one-water layer regime of montmorillonite has been proposed by Fripiat,20 but more recent conductivity measurements in fluorhectorite with various types of counterions appear to contradict Fripiat’s conclusions.19 A different issue can also be addressed by our data. Recent molecular simulations have predicted that an intermediate state with a basal spacing corresponding to between a one- and a twowater layer hydrate may be formed in Li-montmorillonite,15 but (19) Kaviratna, P. D.; Pinnavaia, T. J.; Schroeder, P. A. J. Phys. Chem. Solids 1996, 57, 1897. (20) Fripiat, J. J.; Jelli, A.; Poncelet, G.; Andre, J. J. Phys. Chem. 1965, 69, 2185. (21) Poinsignon, C. Solid State Ionics 1997, 97, 399. (22) Garcı´ a, N. J.; Bazan, J. C. Solid State Ionics 1996, 92, 139. (23) Salles, F.; Devautour-Vinot, S.; Bildstein, O.; Jullien, M.; Maurin, G.; Giuntini, J.; Douillard, J.-M.; Van Damme, H. J. Phys. Chem. C 2008, 112, 14001.
Published on Web 04/14/2010
DOI: 10.1021/la100377s
9703
Article
Ten orio et al.
the evidence of such a state from existing experimental data has not been conclusive.24 From WAXS data in Li-Fht and Na-Fht, in conjunction with 7Li NMR spectra as a function of RH, we can assess the validity of the 1.5-water layer state hypothesis in Li-fluorhectorite.
II. Experimental Details The chemical composition of fluorhectorite is Mx(Mg6-xLix)F4Si8O20, where M denotes either Li or some other monovalent cation. A value x = 1.2 for the negative charge per unit cell has been reported by Kaviratna et al.19 for fluorhectorite manufactured by Corning Inc. However, the synthesis procedure for fluorhectorite has been found to be capable to somewhat modifying the composition as well as the cation exchange capability.25 The starting material for our experiments was synthetic Lifluorhectorite (Li-Fht) (Corning Inc., New York) from which Na-fluorhectorite (Na-Fht) was prepared according to the ionic exchange procedure described in ref 4. Li-Fht was suspended in deionized water and subjected to ionic exchange with NaCl, lasting several weeks, followed by a dialysis process to remove excess ions. Powdered samples were dried for 48 h at 80 C before placing them, for at least 48 h, in a controlled relative humidity atmosphere, prior to NMR measurements. To that end, several aqueous solutions of different salts with standard compositions were employed, and the resulting relative humidity (RH) values were checked against hygrometer readings. In some cases very low RH samples were employed, which were obtained by placing the sample for 48 h in a desiccator with silica gel under vacuum. In addition to powdered samples, stacked pseudocrystalline films were also examined. Approximately 1 mm thick oriented films were obtained by slow evaporation of aqueous suspensions of fluorhectorite deposited on a plastic substrate. The films, cut to the right size, could only be stacked inside the sample tube with the film normal either perpendicular or parallel to the magnetic field. 1 H and 7Li NMR spectra were obtained at a magnetic field of 7.04 T using a pulse NMR Varian UNITY plus-300 spectrometer with a spectral window of 100 kHz. The samples were placed in sealed 5 mm diameter tubes, and NMR spectra were obtained using a single π/2 pulse with acquisition time of 20 ms. The time delay was set to 1 s, and the number of transients was set to 500 scans. For the adamantane 1H spectra, all of the parameters were set as mentioned above with the exception of the time delay, which was set to 10 s. The NMR probehead temperature was controlled by a standard variable temperature module (Varian) that provided a resolution of (0.1 C. Wide-angle X-ray scattering (WAXS) measurements were performed at the Laborat orio Nacional de Luz Sincrotron (LNLS) in Brazil using a wavelength λ = 1.24 A˚. In this case, the sample was mounted on a holder and placed within a continuous flow system which maintained a fixed value of RH at the sample. To control the relative humidity in X-ray experiments, air was pumped through a saturated salt solution. The moist air was then mixed with dry air pumped through a silica gel desiccant column. Relative humidity ranging from 10% to 90% could be obtained by carefully adjusting the flow rates of moist and dry air. To further isolate the sample holder from the external ambient humidity, a beryllium metal enclosure was employed. The temperature at the sample holder was maintained at 25 C by a thermoelectric Peltier element, and the RH was continuously monitored by a hygrometer sensor placed near the sample.
III. Experimental Results a. NMR. Typical Pake doublet26 proton NMR spectra in powdered samples and in pseudocrystalline films of Li-Fht at (24) Del Pennino, U.; Mazzega, E.; Valeri, S.; Alietti, A.; Brigatti, M. F.; Poppi, L. J. Colloid Interface Sci. 1981, 84, 301. (25) Breu, J.; Seidl, W.; Stoll, A. J.; Lange, K. G.; Probst, T. U. Chem. Mater. 2001, 13, 4213.
9704 DOI: 10.1021/la100377s
Figure 1. Moisture dependence of the central peak in 1H spectra of powdered samples at 20 C in Li-Fht (top) and Na-Fht (bottom). The range of RH corresponds to the one-water layer regime. The amplitude of the largest peak in each spectrum was normalized to unity.
20 C are shown in the Supporting Information. In the latter case the magnetic field was oriented perpendicular to the film normal. The RH was kept low in the spectra by equilibrating the samples in silica gel. Unlike hectorite and montmorillonite, fluorhectorite contains no structural protons from -OH groups. Hence, the proton signal can only be attributed to intercalated water of hydration, which simplifies the analysis of the spectra. The frequency splitting of the two singularities in the Pake doublet powder pattern approximately coincides with the frequency splitting of the two peaks in the oriented film when the magnetic field is perpendicular to the film normal. As the RH increases, the amplitude of a central peak grows considerably. Figure 1 shows powder spectra in Na-Fht and LiFht as a function of RH in the one-water layer (1WL) regime, where the X-ray-determined basal spacing is ∼12 A˚. The differences in the Pake powder patterns observed in the spectra of the Li-Fht and Na-Fht could be enhanced by a right shift of ∼25 μs of the time origin in the free induction decay signals, followed by the Fourier transform. This permits to better visualize the contribution of the central peak. Figure 2 shows measured values of the Pake doublet splittings ΔfLi-Fht and ΔfNa-Fht as a function of temperature in low relative humidity (RH < 15%) samples. In the range 353 K g T g TM, above the motional narrowing temperature27 TM ≈ 208 K, the splittings can be seen to be somewhat temperature-dependent. In contrast, below TM, the line widths increase drastically and the doublet structure becomes totally masked. Figure 3 shows the RH dependence of the amount of adsorbed water in Li-Fht as determined by 1H NMR. The data were (26) Pake, G. E. J. Chem. Phys. 1948, 16, 327. (27) Abragam, A. The Principles of Nuclear Magnetism; Clarendon Press: Oxford, UK, 1961.
Langmuir 2010, 26(12), 9703–9709
Ten orio et al.
Article
Figure 2. Temperature dependence of the Pake doublet split-
Figure 4. 7Li NMR spectra in powdered Li-Fht at 20 C for
ting in H spectra of powdered samples of Na-Fht (b) and Li-Fht (1) at RH < 15%.
various values of relative humidity. The inset shows in more detail the spectra corresponding to RH = 15% and RH = 39%.
1
Figure 3. 1H NMR measurement of water adsorbed in Li-Fht clay at 20 C as a function of RH: (left) calculated number of water molecules per intercalated Liþ counterion; (right) grams of water per 100 g of dry clay.
obtained at 20 C by comparing proton spectra of samples equilibrated, for at least 48 h at each RH value, with a weighted amount of a reference substance containing a well-known number of protons per molecule. To that end, we used adamantane (C10H16) which permitted a good comparison without any changes in the spectrometer settings. From the ratio of the areas of Li-Fht to adamantane and the known chemical compositions, it is possible to determine the number of water molecules per Liþ counterion in the 1WL regime, which from Figure 3 appears to be ∼3. In spite of the presence of lithium both as an exchangeable cation and as part of the structure, relevant information can be obtained from 7Li (I = 3/2) NMR spectra in Li-Fht by focusing on the RH dependence of the spectra, which is expected to be significant only for interlamellar 7Li. Figure 4 shows a series of 7 Li spectra in powdered Li-Fht obtained at 20 C for several values of relative ambient humidity using a pulse repetition time of 1 s which is expected to enhance the signal from the more rapidly relaxing interlamellar lithium nuclei by saturating the signal from structural 7Li. This can be inferred from the bimodal spin-lattice relaxation behavior of 7Li, which consists of a fast relaxing component with a short spin-lattice relaxation time T1 and a slow relaxing component with T1 . 1 s. The short T1 component is strongly dependent upon RH and increases abruptly for RH > 44%, corresponding to the boundary of the 1WL regime. This suggests that the fast relaxing component can be identified as interlamellar 7Li nuclei with a quadrupolar Langmuir 2010, 26(12), 9703–9709
Figure 5. 7Li NMR satellite splittings at 20 C in powdered samples of Li-Fht as a function of relative humidity.
relaxation mechanism which becomes less efficient as the electric field gradient at the nucleus is reduced. 7 Li spectra in Li-Fht exhibit three well-defined regions shown in Figures 4 and 5. For RH < 40% (within region I) the spectra show a narrow, 600 Hz wide, central peak corresponding to the -1/2 f 1/2 transition and a much wider powder pattern with a satellite splitting of ∼39 kHz. In the moisture range 44% < RH < 70% (region II) the satellite splitting drops to 5-7 kHz, as shown in Figures 4 and 5, and for RH > 76% (region III) a further abrupt drop to 2.2 kHz takes place with little variation in the interval 76% e RH e 84%. Finally, for a value RH>90% the satellites and the central component collapse into a single line. The satellite splitting in pseudocrystalline films spectra coincide with the values observed in the powdered sample spectra shown in Figure 4 provided the magnetic field is aligned perpendicular to the normal of the films. Furthermore, for a magnetic field parallel to the normal, the frequency splittings increase by a factor of ∼2 (see Supporting Information), as expected for an electric field gradient with approximately cylindrical symmetry about the platelet normal. b. X-ray Scattering. An interesting parallel can be drawn between the three regions in the 7Li NMR data of Figures 4 and 5 and the three peaks found in the X-ray powder diffraction patterns from 001 basal reflections in Li-Fht (shown in the Supporting Information). Here the scattering amplitudes as a function of momentum transfer g = 4π sin(Θ)/λ for 13 values of RH, where 2Θ denotes the scattering angle and λ is the wavelength DOI: 10.1021/la100377s
9705
Article
Ten orio et al.
Figure 6. Plot of the basal spacing as as a function of relative humidity obtained from X-ray 001 reflections in Li-Fht (left) and Na-Fht (right).
of the X-rays, are shown for a decreasing humidity run. From the position of the peaks the basal distances d001 ≈ 12.15 A˚ (region I), d001 ≈ 13.84 A˚ (region II), and d001 ≈ 15.30 A˚ (region III) are obtained, as shown in Figure 6 for Li-Fht and for Na-Fht. It is worth emphasizing that 001 basal reflections from Na-Fht as a function of decreasing RH, also shown in the Supporting Information for 10 values of RH, do not exhibit the intermediate peak (region II) at g ≈ 0.454 A˚-1.
IV. Discussion a. Structure and Dynamics of Intercalated Water Molecules. Pake Doublets. The NMR spectrum of static, isolated pairs of spin I = 1/2 nuclei, such as the two protons in a water molecule, which interact via magnetic dipole-dipole interactions, consists of two lines with frequency splitting:9,26,28 Δf ðθÞ ¼ ð3=2πÞpγ
2
!
1 jrBj
3
j3 cos2 ðθÞ - 1j 2
ð1Þ
Here B r denotes the internuclear vector, θ is the angle between B r and the polarizing field BB0, and γ is the gyromagnetic ratio of the nuclei. If the internuclear vector B r is not static but rather reorients about a single axis (C*), at a rate which is fast compared to Δf(θ), the frequency splitting of the Pake doublet is reduced from the value of eq 1 to a new value given by28 Δf ðψ, jÞ ¼ ð3=2πÞpγ
2
!
1 3
jrBj
j3 cos2 ðψÞ - 1j j3 cos2 ðjÞ - 1j 2 2 ð2Þ
where j denotes the angle between the vector B r and the reorientation axis C* and ψ is the angle between C* and the magnetic field BB0. In a powdered sample, where the crystallites are randomly oriented with respect to BB0, the spectrum consists of a weighted average of spectra corresponding to all possible orientations. The resulting pattern exhibits two characteristic singularities, which have been widely employed as a tool for structural studies.29,30 The frequency splitting of the two singularities can be obtained from the same expressions of eqs 1 and 2 by setting the angles (28) Gutowsky, H. S.; Pake, G. E. J. Chem. Phys. 1950, 18, 162. (29) Engelsberg, M.; Yannoni, C. S.; Jacintha, M. A.; Dybowski, C.; Souza, R. E. J. Phys. Chem. 1994, 98, 2397. (30) Engelsberg, M.; Yannoni, C. S.; Jacintha, M. A.; Dybowski, C. J. Am. Chem. Soc. 1992, 114, 8319. (31) Halle, B.; Wennerstr€om, H. J. Chem. Phys. 1981, 75, 1928.
9706 DOI: 10.1021/la100377s
θ = π/2 or ψ = π/2, respectively.27,31 The observed frequency splittings in our powdered samples are actually found to be the same as in pseudocrystalline films, provided BB0 is perpendicular to the normal of the film. If pseudocrystalline films with BB0 parallel to the surface are employed, the measured splittings were found to increase by a factor of 2, as expected from eq 2 with ψ = 0 instead of ψ = π/2. Setting θ = π/2 in eq 1 and using the value |rB| = 1.58 A˚, obtained for water molecules in gypsum,26 yields a splitting Δf(π/2) = 45.7 kHz. Slightly different values of |rB| have been reported in other cases.10,32 For intercalated water molecules in layered chalcogenides, for example, the value |rB| = 1.63 A˚ appears to be more generally accepted,33 yielding a splitting Δf(π/2) = 41.9 kHz. Within this range of possible values of |rB|, the calculated splitting would, in any case, be substantially larger than the measured values in Na-Fht and Li-Fht shown in Figure 2. This suggests that water molecules are rapidly reorienting about spatially fixed axes at a rate which is fast compared with Δf ≈ 50 kHz. This rapid reorientation could, in part, explain the relatively narrow line widths observed for the Pake doublets which could be attributed to the averaging out of the intermolecular dipole-dipole interaction. In addition to Pake doublets, the measured 1H spectra in Li-Fht and Na-Fht shown in Figure 1 also exhibit a central peak whose origin and significance will be discussed in section IV.c. Generally speaking, it is not possible, from the frequency splitting of the Pake doublets alone, to completely characterize the motion which, for some conditions of temperature and RH, could be quite complex. However, in the one-water layer (1WL) regime of RH, where from Figure 3, the number of water molecules per exchangeable cation ranges approximately from 2 to 3 and the basal spacing is d001 ≈ 12 A˚, it is possible to check the validity of some simple models. Molecular simulations in montmorillonite predict that, in the one-water layer regime, the water molecules are oriented with one of the -OH vectors pointing along the C* axis, perpendicular to the silicate planes.15 We can now test this prediction in fluorhectorite with the additional assumption that, in the time scale of order of 20 μs, the proton-proton internuclear vector B r rapidly reorients about a single axis coinciding with the C* axis. Considering first Na-Fht, the data of Figure 2 indicate that, below TM ≈ 208 K, the motion of water molecules slows down to rates smaller than Δf ≈ 50 kHz, causing a broadening of the (32) Woessner, D. E.; Snowden, B. S. J. Chem. Phys. 1969, 50, 1516. (33) Halstead, T. K.; Schmidt, C.; Spiess, H. W.; Sch€ollhorn, R.; M€ullerWarmuth, W.; M€oller, H. J. Phys. Chem. 1988, 92, 7167.
Langmuir 2010, 26(12), 9703–9709
Ten orio et al.
Figure 7. Schematic representation of a hydrated Naþ counterion in Na-Fht. The C* axis is perpendicular to the silicate planes, and the C2 axis is along the electric dipole moment of the water molecules.
spectra which masks the doublet structure. From the value of TM one can make a rough estimate of the activation energy Ea of this process using the Waugh-Fedin approximate method,34 which yields Ea ≈ 31 kJ mol-1. Furthermore, as the temperature is increased above TM, and the Pake doublets become sharp, the frequency splitting changes little in the range 208 K < T < 228 K and then slowly decreases, reaching a value 25% lower at 353 K. If we apply eq 2 with ψ = π/2 and j = 37.75ο we obtain, assuming a proton-proton distance of |rB| = 1.65 A˚,10 Δf(π/2, 37.75) = 17.6 kHz, which agrees with the measured value in Na-Fht just above TM (Figure 2). An angle j = 37.75 between the proton-proton vector in the water molecule and the C* axis corresponds, assuming a — HOH angle of 104.50,35 to one -OH vector pointing perpendicular to the silicate planes, as shown schematically in Figure 7. Hence, just above TM, the direction of one -OH vector in 1WL Na-Fht seems to be anchored along the normal of the silicate planes as predicted by the molecular simulations of Tambach et al. for montmorillonite.15 Furthermore, on the NMR time scale, the proton-proton vectors appear to be rapidly reorienting about the C* axis. This model is schematically depicted in Figure 7 where two water molecules are assumed to be coordinated with a Naþ cation4 with the Na-O vectors pointing along the C2 axes, coincident with the electric dipole moments of water. The water dipoles form an angle R = 52.25 with the C* axis, and the whole coordination sphere is assumed to be reorienting about C*. As the temperature is increased and the Pake doublet splitting is reduced (Figure 2), it appears that the direction of the -OH vectors begins to fluctuate within a cone about the C* axis. Although the detailed dynamics of this departure cannot be univocally inferred from the 1H spectra, one could probably characterize the process by an order parameter which decreases with increasing temperature. Alternatively, one can estimate the required angular deviation of the -OH vector from the C* axis which would produce the observed frequency splitting at T = 293 K as ∼3.9 We next consider Li-Fht where not only the proton resonance but also the 7Li (I = 3/2) resonance are revealing. First we notice from Figure 4 that, even at the lowest values of RH, a narrow (∼600 Hz FWHM) central peak is observed. Neglecting secondorder quadrupolar effects,36 the width of the -1/2 f 1/2 transition should be predominantly determined by magnetic dipoledipole interactions between 7Li and neighboring protons from water molecules. Hence, one cannot assume that Li-H distances remain unchanged on the NMR time scale, as suggested schematically in Figure 7, since this would lead to a quite broader Li-H dipolar line width than observed. We must conclude that, on the (34) Waugh, J. S.; Fedin, E. I. Sov. Phys. Solid State (Engl. Trans.) 1963, 4, 1633. (35) Matsuoka, O.; Clementi, E.; Yoshimine, M. J. Chem. Phys. 1976, 64, 1351. (36) B€ohmer, R.; Jeffrey, K. R.; Vogel, M. Prog. Nucl. Magn. Reson. Spectrosc. 2007, 50, 87.
Langmuir 2010, 26(12), 9703–9709
Article
NMR time scale, even at the lowest values of RH, the Liþ cations and water molecules are in relative motion. However, the motion of H-H internuclear vector is not random but involves a reorientation about the C* axis. The measured splitting just above the motional narrowing temperature TM = 208 K (Figure 2) is consistent with an angle j = 43.10 between the H-H vector and the C* axis, which from eq 2 yields Δf(π/2, 43.10) = 12.0 kHz, in agreement with the measured value. Hence, in the case of Li-Fht the orientation of -OH bond vectors appear to depart from the C* axis by an angle of ∼5 with an angle R = 46.90 between the C* axis and the direction C2 of the dipole moments of the water molecules. b. One and a Half Layer Li-Fht Hydrate. The molecular simulations of Tambach et al. in montmorillonite hydrates with different counterions predicted an interesting difference between lithium and sodium (or potassium).15 From the minima in the free energy as a function basal spacing, it was possible to confirm, for the counterions Naþ and Kþ, the occurrence of 1WL hydrates at basal spacings of 12.50 A˚ and that indeed 2WL hydrates were the most stable states at a basal spacing of 14.75 A˚. However, for Li-montmorillonite an additional free energy minimum was found to be present at a basal spacing 13.5 A˚, corresponding to a new state between a 1WL and a 2WL hydrate. Basal spacing values that could correspond to the above theoretical prediction have been previously measured in montmorillonite, but the experimental X-ray diffraction patterns have been attributed to structural or chemical heterogeneities in the clay samples which could cause interstratification of 1WL and 2WL hydrates, resulting in an average basal spacing.24 This could possibly explain the broad feature observed between the two main Na-Fht peaks shown in the Supporting Information. In contrast, in Li-Fht, the results of Figures 4-6 strongly support the conclusions of ref 15 and appear to confirm the existence of a 1.5WL state. For some values of RH, such as for example RH = 59%, the X-ray diffraction peak at g ≈ 0.454 A˚-1, shown in the Supporting Information, appears as a single peak without any appreciable admixture from the 1WL and 2WL peaks. Moreover, the 7Li NMR results of Figures 4 and 5 further reinforce the hypothesis of an intermediate state between 1WL (region I) and 2WL (region III). The angular dependence of the 7 Li spectra suggests that the C* axis is a principal axis of cylindrical symmetry of the electric field gradient tensor at the 7 Li site. The corresponding principal value eq of the gradient produced by surrounding atoms determines the powder pattern satellite splitting of Figure 4, through δf = (1 - γ¥)e2qQ/2h.37,38 Here eQ denotes the 7Li quadrupole moment, γ¥ is Sternheimer’s antishielding factor for Liþ, h is Planck’s constant, and e is the electronic charge. When the basal spacing and the 7Li environment change abruptly, as a consequence of increasing RH, the values of eq, and consequently of δf, are expected to follow the variations, as confirmed by the data of Figure 5. Here three regions with different satellite splittings δf, approximately corresponding to the three basal spacings of Figure 6 (right), are apparent. Moreover, the behavior in region II could not be accounted for by a static interstratification between 1WL and a 2WL states since this is not expected to cause a spatially averaged, intermediate, electric field gradient that could explain the 7Li NMR spectra in this region. It has been suggested that the dominant contribution to the electric field gradient eq at the 7Li site in hectorite may be caused (37) Cohen, M. H.; Reif, F. In Solid State Physics - Advances in Research and Applications; Seitz, F., Turnbull, D., Eds.; Academic Press: New York, 1957; Vol. 5, p 321. (38) Conard, J. ACS Symp. Ser. 1976, 34, 85.
DOI: 10.1021/la100377s
9707
Article
by the electric dipole moment of the three water molecules coordinated to Liþ.38 Although an exact result in Li-Fht would require a more elaborate calculation involving contributions from other neighboring atoms, an estimate of eq in the 1WL regime is possible using such a simplified model. The electric field gradient along the C* direction caused by three dipoles pB with their origins, at a distance R from 7Li, arranged in an equilateral triangle is given by eq = (9|pB|/R4)|3 cos2(R) - 1|, where a value |pB| = 1.85 D42 for the water dipole moment has been employed. Adopting Zachariase ns value 1.95 A˚ for the Li-O distance39 ˚ and a 0.109 A distance from the oxygen atom to the center of the water dipole yields R = 2.059 A˚. Furthermore, using40 γ¥ = 0.255 and41 Q = 0.041 10-24 cm2, the calculated splitting coincides with the experimental value provided an angle R = 47.2, between the dipole moments of the water molecules and the C* axis, is adopted. This agrees quite well with the value of R previously inferred from the proton spectra in Li-Fht. As RH increases, and the 1WL state changes to 1.5 WL and 2WL states, the Liþ ion is believed to change from a position close to the silica layers, where it forms an inner-sphere complex, to more symmetrically located positions near the middle of the interlayer space, where it forms an outer-sphere complex.15 The results of these changes and the different dynamics prevalent in each case appear to cause the variations in the electric field gradients shown in Figure 6. c. Central Peak in the Proton Spectra. The proton spectra in Li-Fht and Na-Fht exhibit, in addition to Pake doublets, a central peak which has also been found in proton spectra of many other systems.10,11,43,44 Inasmuch as the observed Pake doublets are assumed to originate from water molecules reorienting about a well-defined axis, it could be argued that the central peak arises from rotational diffusion of molecules about randomly oriented axes. However, from the observation that, for 2H (I = 1) spectra of intercalated D2O in Na-Fht, where the central line is absent,9 one is forced to abandon such a conclusion. It has been proposed that rapidly exchanging protons, created by the dissociation of water molecules, could explain the central peak.11 Considerable controversy still exists concerning the role of proton exchange in hydrated clays. While some authors have invoked this process as a prevalent mechanism of charge transport at low RH,20,21 others have not taken proton exchange into account in the interpretation of experimental data or in simulations.8,23 It was first suggested by Fripiat20 that proton exchange should play a dominant role in the electrical conductivity of clays at low RH, but this conclusion has been challenged. Measurements by Kaviratna et al. in synthetic fluorhectorite (Corning) with Li, Na, and Cu as counterions apparently contradict this conclusion.19 It was argued that the larger the charge/radius ratio of the intercalated cation, the larger should be the polarization effect on a coordinated water molecule and the larger the probability of dissociation. Thus, if electrical conduction were dominated by proton exchange, it should lead to a larger electrical conductivity in Li-Fht than in Na-Fht, contrary to what was observed (39) Kittel, C. Introduction to Solid State Physics; John Wiley & Sons: New York, 1966; p 105. (40) Chihara, H.; Nakamura, N. 2.6 Sternheimer Antishielding Factor for Atoms, Free Ions and Ions in Crystals; Chihara, H., Ed.; SpringerMaterials - The Landolt-B€ornsterin Database (http://www.springermaterials.com). DOI: 10.1007/ 10565418_10. (41) Differt, K.; Messer, R. J. Phys. C 1980, 13, 717. (42) Clough, S. A.; Beers, Y.; Klein, G. P.; Rothman, L. S. J. Chem. Phys. 1973, 59, 2254. (43) Kunitomo, M.; Kohmoto, T.; Fukuda, Y.; Eda, K.; Sotani, N.; Kaburagi, M. Phys. Lett. A 1995, 199, 103. (44) Alexiev, V.; Meyer zu Altenschildesche, H.; Prins, R.; Weber, T. Chem. Mater. 1999, 11, 1742.
9708 DOI: 10.1021/la100377s
Ten orio et al.
experimentally in fluorhectorite at low RH. On the other hand, if the intercalated cations were predominantly responsible for the electrical conduction, a smaller charge/radius, as for Naþ, would tend to weaken the electrostatic attraction to the negatively charged silica layers, therefore increasing the cation mobility and the conductivity compared to Liþ.19 Although the above arguments seemed to explain the observations in fluorhectorite at low RH, they fail to explain the behavior observed in montmorillonite where the low RH electrical conductivity of Li-montmorillonite was found to be larger than in Na-montmorillonite.22 Since the charge/radius argument should, in principle, also be applicable to this latter case, one concludes that its validity appears to be questionable and that other aspects must be taken into account. Our present results may shed some light upon the controversy on the role of proton exchange in the charge transport mechanism of Li-Fht and Na-Fht. This is suggested by the quite different amplitudes of the central peaks in the proton spectra of Li-Fht and Na-Fht at low RH. If the ratio of the area of central peak to the total area under the proton spectra of Figure 1 is considered as a measurement of water dissociation,9 then, for RH = 15%, the dissociation of water molecules in Na-Fht should be ∼6 times larger than in Li-Fht. This could explain, within factors not too different from unity, the ratio of the measured value19 of the electrical conductivity in Na-Fht to the value in Li-Fht at the same value of RH. At low RH values, proton exchange is therefore rescued as possibly the prevailing mechanism of charge transport, whereas at high RH, cationic diffusivity predominates. It appears that, unlike the case of aqueous solutions, the role of the charge/ radius ratio of the cation as the dominant factor in the dissociation of intercalated water molecules in Na-Fht and Li-Fht at low RH appears to be questionable, and other possible mechanisms45 should also be considered.
V. Conclusions NMR and WAXS measurements in powdered samples and pseudocrystalline films of Li-Fht and Na-Fht as a function of relative humidity permitted to address several aspects of the structure and dynamics of intercalated water. In particular, it is interesting to compare our results with recent molecular simulations in montmorillonite15 to assess to what extent structural differences between the two smectites may affect various properties. In the 1WL regime the motion of water molecules in fluorhectorite appears to involve a reorientation about the C* axis perpendicular to the silicate planes. Moreover, at a temperature of ∼207 K one of the water -OH vector directions in Na-Fht appears to be locked parallel the C* axis, as predicted by molecular simulations15 for both Na-montmorillonite and Li-montmorillonite. However, for Li-Fht at the same temperature, some differences with the molecular simulations are apparent. From our NMR data we conclude that, in this case, the -OH vector somewhat deviates from the C* axis direction. Furthermore, at higher temperature the reorientation axis in both Li-Fht and NaFht appears to no longer be fixed in space, but its direction appears to somewhat fluctuate within a cone about the C* direction. WAXS and 7Li NMR measurements in Li-Fht appear to confirm the existence of a 1.5WL state with a basal spacing of ∼13.84 A˚, not observable in Na-Fht. Such a state has been predicted from molecular simulations15 in Li-montmorillonite and appears to also prevail in Li-Fht but not in Na-Fht. Therefore, the existence of a 1.5WL state appears to depend mostly upon the (45) Churakov, S. V. Geochim. Cosmochim. Acta 2007, 71, 1130.
Langmuir 2010, 26(12), 9703–9709
Ten orio et al.
type of counterion rather than upon the type of smectite. However, from an experimental point of view, its existence appears to be more clearly detectable in Li-Fht than in Li-montmorillonite. The role of proton exchange as a mechanism of charge transport in Fht was reexamined. In spite of the smaller charge/ radius ratio of Naþ compared to Liþ, the larger relative weight of the central peak observed in 1H NMR spectra of Na-Fht compared to Li-Fht could be invoked as a possible mechanism of the higher electrical conductivity in Na-Fht in the 1WL regime.
Langmuir 2010, 26(12), 9703–9709
Article
Acknowledgment. We thank Alberto Colnago and Fernando Hallwass for help and Elizabeth Lindbo Hansen and Henrik Hemmen for useful discussions. This work has been supported by Conselho Nacional de Desenvolvimento Cientı´ fico e Tecnologico CNPQ (Brazil) and the Research Council of Norway (RCN). Supporting Information Available: Additional NMR spectra and X-ray data (Figures S1-S3). This material is available free of charge via the Internet at http://pubs.acs.org.
DOI: 10.1021/la100377s
9709