Probing the nanoscale architecture of clay minerals

Abstract In recent years, experimental and theoretical methods have provided new insights into the size, shape, reactivity, and stability of clay minerals. Although diverse and complex, the surface chemistry of all clay minerals is defined spatially on a common scale of nanometres. This review is organized around the nanoscale architecture of clay minerals examined at several different length scales. The first, and perhaps most important, is the length scale associated with Hbonding in clay minerals. Hbonding interactions define the size and shape of 1:1 phyllosilicates and dominate the surface chemistry of many clay minerals. Structural and surface OHgroups contained within and on the surface of clay minerals provide a type of ‘molecular reporter group’ and are sensitive to subtle changes in their local environment. Examples of OH-reporter group studies in clay minerals, and the spatial scales at which they provide diagnostic information, are examined. The second length scale considered here is that associated with clay–water and clay–organic interactions. Inorganic and organic solutes can be used to explore the surface chemistry of clay minerals. Similar to the use of reporter groups, molecular probes have diagnostic properties that are sensitive to changes in their molecular environment. Clay–water interactions occur at a length scale that extends from the size of the H2O molecule (~0.3 nm) to the larger scales associated with clay-swelling (>10 nm). Similarly, clay–organic interactions are also defined, in part, on the basis of their molecular size, in addition to the type of chemical bonding interactions that take place between the organic solute and the clay surface. Examples illustrating the use of clay–water and clay–organic solute interactions as molecular probes are presented. The largest scale to be considered is that of the particles themselves, with scales that approach micrometres. Recent developments in the synthesis and characterization of ultrathin hybrid films of clay minerals provide complementary information about the nature and distribution of active sites on clay minerals, as well as providing new opportunities to exploit the surface chemistry of clay minerals in the design of functional materials.

soil and subsurface environments as well as comprising a significant fraction of the nanoparticles found in the atmosphere.
More important than their overall abundance, clay minerals represent one of the dominant environmental interfaces found in the 'pedo-lithohydro-sphere' which control and direct most physical, biological and chemical reactions. To this end, clay minerals play a uniquely prominent role in the freshly minted disciplines of nanogeoscience (Hochella, 2006(Hochella, , 2008 and the science of the critical zone (Richter, 2007;Wilding & Lin, 2006). In soil and subsurface environments, essentially all inorganic and organic solutes are attracted to, transformed on, or repelled from clay minerals. In geotechnical areas, the presence of clay minerals controls structural stability of soils and influences liquefaction. As clay minerals often represent a significant fraction of the hydrous mineral phases found in the Earth's mantle, they influence dynamic processes of water regulation in the mantle and their presence in subduction zones are thought to play a role in triggering deep-focus earthquakes. In recent years, clay-based applications in the fields of nanoscience and nanotechnology have exponentially grown research areas with the goal of producing new functional nanomaterials in which low cost-and high added value-clay minerals are attractive constituents (Ras et al., 2007b). Common to all of these disciplines are the reactions that occur at the clayÀwater interface on a scale from Å ngstroms to nanometres.
The nanometre length scales associated with clay minerals along at least one dimension is both the reason for their utility and a challenge to their characterization. Since the pioneering work of Hendricks and Pauling, details about their atomic structure and composition have been steadily revealed over the past eight decades (Hendricks, 1929;Pauling, 1930). Classic and modern synchrotron-based X-ray studies have played a prominent role in revealing detailed information about the atomic structures of clay minerals and local structures of adsorbed species with sub-Å ngstrom resolution (Brown et al., 1999(Brown et al., , 2005. However, it is the size domain slightly larger than Å ngstroms that continues to present experimental, theoretical and numerical challenges. It is at the scale of nanometres, which defines active sites on clay surfaces, where both clayÀwater surface interactions and particleÀparticle interactions occur, to name two important areas. This review will examine the surface structure and reactivity of clay minerals at several different length scales as depicted in Fig. 1. These span from the sub-Å ngstrom scale interactions of hydroxyl groups to the approximate size of particles (>1000 nm) and will rely mainly on spectroscopic observations. Three size domains are considered, beginning with the smallest length scale, that of hydrogen bonding (H bonding). The chemistry of the hydrogen atom extends far beyond its size (~0.1 nm) and its elemental abundance in the Earth's crust (~0.15%), playing a central role in aqueous geochemistry and defining the surface chemistry of clay minerals. The scale of H bonding is most commonly defined on the basis of interatomic O_O distances occurring within the range of 0.25 to 0.35 nm. The next scale to be considered is that associated with clayÀwater and clayÀorganic interactions ( Fig. 1). At these length scales, multiple molecular mechanisms are operative (including H bonding) and span from the size of the H 2 O molecule and small polar organic solutes (0.3 nm) to that of large biological molecules (e.g. proteins) and the limit of clay-water swelling (10 nm). Interactions at these smaller length scales also impact how clay particles interact with other particles. The operative length scales for these interactions extend from the thickness of a delaminated smectite particle (1 nm) to the basal dimensions of large clay particles and aggregates with dimensions that commonly exceed one mm. This review will conclude with one illustrative example of clay particle-particle interactions at this 'particle' scale ( Fig. 1).

Structural and surface hydroxyl groups
All clay minerals contain structural and surface hydroxyl OH groups. Structural OH groups are those contained within the crystal structure of the clay particle and surface OH groups are those OH groups on the external basal or edge surfaces of the clay particle. These OH groups, and the H bonding interactions in which they participate, dominate the surface chemistry of many clay minerals (Davis & Kent, 1990;Johnston & Tombacz, 2002;Sposito, 1984). For 1:1 phyllosilicates and hydrous oxides, the type and strength of the interlayer H bonds of the structural OH groups govern their size, shape and geochemical stability. In addition, all clay minerals have surface OH groups located on basal surfaces, broken edges, steps and related defect sites. Basal surface OH groups, such as the OH groups on the terminal Al octahedral sheet of kaolinite, are coordinated to metal atoms whose coordination environment is complete and are charge-neutral. These types of hydroxyl surfaces are also common to many hydrous oxides, such as gibbsite (Johnston & Tombacz, 2002;. The surface OH groups located at broken edges, steps and defect sites, on the other hand, are under-coordinated and reactive. Consequently, they generally carry either a positive or a negative charge depending on the type of metal ion they are coordinated to, and the pH of the ambient aqueous solution they are in contact with (Schindler et al., 1976;Stumm, 1992). These under-coordinated surface OH groups are among the most abundant and reactive surface functional groups found in soil and subsurface environments (Sposito, 1984).
Hydrogen bonds (H bonds) are defined on the basis of their bond length, geometry and bond strength with strong H bonds having interatomic d(O_O) distances of <0.27 nm and weak H bonds defined by distances of >0.27 nm (Libowitzky, 1999;Pauling, 1960;Pimentel & McClellan, 1960). The positions of the OH stretching and bending vibrations observed in the vibrational spectra of clay minerals (IR, Raman and inelastic neutron scattering) provide the most sensitive measure of H bonding. In particular, the vibrational frequencies associated with the stretching motion of the structural hydroxyl groups are remarkably sensitive to subtle changes in their local environment. This is illustrated in Fig. 2 by the distanceÀfrequency relation obtained from 65 minerals showing the change in n(OH) frequency versus the interatomic d(O_O) distance. These data show that a change in the interatomic distance of 0.1 nm results in a change in frequency of over 2500 cm À1 (Libowitzky, 1999). In reference to Fig. 1 Fig. 3, showing the three unique inner-surface hydroxyl groups labelled AÀC which form long interlayer hydrogen bonds with O_O distances ranging from 0.295 to 0.308 nm. For example, kaolinite is one specific polytype of the kaolin subgroup that corresponds to an overall minimum in the combined distances of the three unique interlayer O(H)_O couples. The other polytypes of kaolin subgroup minerals, dickite and nacrite, share the same fundamental layer composi-tion and structure as kaolinite. However, they differ in how subsequent layers are stacked on top of each other, with each polytype represented by a unique set of interlayer O(H)_O pairings. The position of the n(OH) bands measured using vibrational spectroscopy can distinguish among the polytypes and is useful to illustrate the sensitivity of n(OH) bands of the structural OH groups to structural differences. Fourier transform infrared (FTIR) spectra of the three polytypes obtained at room and low temperature (<20 K) are shown on the left side of Fig. 4. The slope of the distanceÀfrequency curve for kaolinite, dickite and nacrite ( Fig. 4; right side) is 4000 cm À1 /nm, or 4 cm À1 /pm. This relationship reveals that a difference in the interlayer O_O distance of 0.001 nm results in a shift in band position of 4 cm À1 Johnston & Stone, 1990). Similar distanceÀfrequency relationships have been established for gibbsite using Raman and FTIR spectroscopy (Wang & Johnston, 2000). These structural (and surface) OH groups provide a type of reporter FIG. 2. Distance frequency relationship for 125 data pairs corresponding to different classes of minerals (shown by different symbols) by plotting the position of the n(OH) band against the interatomic O_O distance (i.e. the O-H_O hydrogen bonding distance). From Libowitzky (1999), used with permission.
group that signal perturbations in their local environment. Reporter groups are part of the molecular structure of the clay particle, or occur at the external surface of the clay particle, and have diagnostic properties, such as n(OH) bands, that are influenced by changes within the clay structure or by reactions at the interface (Gale, 2000;Johnston et al., 1993). The use of structural and surface OH groups is emphasized here; however, other examples of reporter groups including NMR-active structural nuclei such as 27 Al and 29 Si (Altaner et al., 1988), or structural Fe in Mossbauer spectroscopy (Heller-Kallai & Rozenson, 1981) are commonly used as well.
Structural OH groups: layer stacking disorder of kaolin subgroup minerals Structural and surface OH groups have been used as reporter groups to study structural disorder in kaolin subgroup minerals. A typical crystal of kaolinite has a thickness of the order of 50 to 100 nm. Since the fundamental thickness of a kaolinite layer is 0.7 nm, this thickness represents approximately 70 to 200 individual layers stacked on top of each other. In well-ordered kaolinite, all the layers are stacked on top of each other in the same way. In disordered kaolins, however, layer stacking 'mistakes' occur and this has been the subject of extensive review (Bookin et al., 1989;Brindley & Robinson, 1946;Murray, 1954;Plançon & Tchoubar, 1977a,b). The only quantitative models available at present to assess layer stacking disorder in kaolin subgroup minerals are based on X-ray diffraction. X-ray methods are intrinsically insensitive to structural differences involving the H atom and are consequently limited in their ability to detect layer stacking disorder (Bookin et al., 1989;Johnston et al., 2008;Plançon, 2001;Plançon et al., 1988;Plançon & Tchoubar, 1977b). From a structural point of view, each type of 'mistake' or layer stacking disorder results in a different, unique set of interlayer O_(H)O pairings. Upon cooling kaolin sub-group minerals to near liquid-He temperatures, FTIR spectra of disordered kaolins reveal an additional set of 'disorder-indicator' bands (Bish & Johnston, 1993;Frost & Kloprogge, 2000;Prost, 1984;Prost et al., 1987). These results are in good agreement with recent high-resolution transmitted electron microscopy (HRTEM) and selected area electron diffraction (SAED) studies of kaolin subgroup minerals by Kogure and coworkers who have observed stacking faults directly (Kameda et al., 2005(Kameda et al., , 2008Kogure et al., 2010;Kogure & Inoue, 2005a,b). This is illustrated by low-temperature FTIR spectra of two Brazilian kaolins shown in Fig. 5 along with SAED images of the Capim sample .
FIG. 3. The crystal structure of kaolinite projected along [010] is shown on the left side, showing the three crystallographically distinct 'inner-surface' OH groups. An expanded view of these three OH groups, labelled AÀC, and their interlayer O_O distances in nm is shown on the right side. The crystal structure is that of Bish (1993).
Probing the nanoscale architecture of clay minerals In addition to the n(OH) bands of kaolinite, these spectra reveal the presence of n(OH) bands that are coincident with those of dickite (shown by the dashed line) and nacrite (shown by the light grey line). The FTIR spectra cannot distinguish between layer stacking disorder and the presence of discrete dickite and nacrite phases. However, the SAED pattern for this sample indicates that this type of disorder does not result from a discrete phase but rather interstratification of different types of layer stacking within a given particle. The reciprocallattice rows in the pattern (Fig. 5a) were streaked along the c* direction, indicating that this crystal contained a mixture of stacking sequences from different subfamilies of 1:1 phyllosilicates (Bailey, 1988;Kogure et al., 2001). Layer-stacking disorder is an important property of kaolin group minerals related to industrial applications. In addition, layer-stacking disorder characterizes sandstone diagenesis as kaolinite is transformed to dickite with increasing depth and temperature (Beaufort et al., 1998;Lanson et al., 2002). The combined application of low-temperature FTIR spectroscopy and high-resolution electron methods are providing new insight about layer stacking disorder in kaolin subgroup minerals over the domain of a single layer (i.e. 0.7 nm) to many layers (~50 nm À see fig. 5 in Kogure & Inoue, 2005b).

Structural OH groups: clay minerals at high pressure
The spectral changes observed in Figs 4 and 5 resulted from an overall contraction of the crystal lattice induced upon cooling (Bish, 1993;Bish & Johnston, 1993) and a concomitant change in the FIG. 4. FTIR spectra of nacrite, dickite and kaolinite at room (300 K represented by dashed lines) and low temperature (<20 K represented by solid lines) in the n(OH) region. The vertical lines for each of the polytypes shown in the upper right portion of the figure illustrate the diagnostic frequencies for each polytype. The corresponding distance frequency relationship for these three polytypes is shown on the right side with the interatomic d(O_O) distance plotted on the y axis versus the observed frequency of the corresponding n(OH) band. These curves are adapted from Johnston et al. (2008).
interlayer O_O distances . Similar structural changes can also be induced by a change in pressure. Using a high-pressure diamond anvil cell (DAC) the Raman spectra and crystal structure of dickite were obtained as a function of pressure (Dera et al., 2003;Johnston et al., 2002b). Raman spectra and crystal structures were obtained at pressures up to 6.5 GPa (65 kbar) and are shown in Fig. 6. The crystal structure of dickite determined at high pressure (Dera et al., 2003) revealed interlayer compression and a layer shift that was responsible for the pronounced changes observed in the n(OH) bands as a function of increasing pressure. The pressure dependence of the n(OH) bands signalled the presence of a structural phase change (Johnston et al., 2002b) which provided the basis for a subsequent detailed structural study (Dera et al., 2003). Although the structural changes were small, with interlayer O_O distances changing less than 0.02 nm, the shifts of the n(OH) bands were significant. Related data have been reported for gibbsite (Huang et al., 1996(Huang et al., , 1999Johnston et al., 2002b), brucite (Shinoda, 1998), and vermiculite (Holtz et al., 1993). For a review of high-pressure methods applied to clay minerals see Butler & Frost, 2006. Structural OH groups: particle size and shape analysis Details about the influence of particle size and shape on vibrational spectra of hydrous minerals are slowly coming into focus. For the past 30 years, there has been some difficulty in reconciling differences observed between IR and Raman spectra of certain types of kaolin group minerals in the n(OH) region (Johnston et al., 1985;Michaelian, 1986;Shoval et al., 1999a,b;Wiewiora et al., 1979). Because of the low symmetry of kaolin group minerals, the n(OH) bands are expected to be both IR-and Raman-active. However, a 'fifth' band appears in the Raman spectra of certain kaolinite samples that does not have an apparent IR-counterpart. Experimental insight into this problem was provided by polarized single crystal Raman and FTIR spectra obtained from oriented single crystals of kaolinite and dickite (Frost & Kloprogge, 2000;Johnston et al., , 1998Shoval et al., 2002), where polarized spectra could be obtained from crystal edges as well as normal to the face of the pseudo-hexagonal shaped crystals. However, it was not until Farmer successfully applied theoretical considerations in the form of LO-TO splitting to account for these differences (Farmer, 1998(Farmer, , 2000 FIG. 5. Low-temperature FTIR spectra of two high-HI kaolins from Brazil (Patinga & Capim). The n(OH) bands shown by the dashed and light gray vertical lines are coincident with those of dickite and nacrite. Selected-area electron diffraction (SAED) pattern from the Capim kaolin sample is shown on the left. The pattern is indicative of interstratification of different subfamilies, supposedly nacrite and kaolinite-dickite. From Johnston et al. (2008), used with permission.
Probing the nanoscale architecture of clay minerals that a satisfactory explanation was achieved. Subsequently, particle size and shape effects have been modelled successfully using density functional theory (Balan et al., 2001(Balan et al., , 2005(Balan et al., , 2006(Balan et al., , 2007. The effect is a band shift related to the electrostatic charges occurring at the surface of the polarized dielectric particles. For platy crystals, bands at the TO and LO frequencies can be resolved (Farmer, 1998(Farmer, , 2000. However, a distribution of band positions resulting from the corresponding depolarization field is expected when particles are not ellipsoidal or when various particle shapes occur in the sample. Parallel experimental and theoretical developments have been successfully applied to gibbsite (Balan et al., 2006;Wang & Johnston, 2000), where particle size and shape have been shown to influence the n(OH) bands of gibbsite (Phambu et al., 2000).

Surface OH groups
Up to this point, we have focused our attention on H bonding interactions of structural OH groups which are contained within the clay structure. In addition, all clay minerals have surface OH groups at broken edges resulting from under-coordinated metal ions that have reacted with water. Depending on the nature of the metal ion, these groups can be very reactive, have high site densities (~2.2 groups/nm 2 ) and play a dominant role in the rheological behaviour of clay minerals in contact with water. From an experimental standpoint, these groups are difficult to study because they represent only a small fraction of the total hydroxyl content of a clay particle. In addition, these groups are H bonded to interfacial water and the corresponding n(OH) bands are broad and overlap with those of H 2 O. To overcome this problem a range of experimental approaches has been attempted. Nonlinear optical methods, such as second harmonic generation (SHG) or sum frequency generation (SFG), provide an elegant, although complex approach (Du et al., 1994a,b;Eisenthal, 1996;Jena & Hore, 2009). Selective deuteration methods have also been used, based on the isotopic exchange of protons by deuterons for exposed surface OH groups. Under vacuum, the resulting n(OD) are spectrally shifted by a factor of~1.4 to lower energy making them spectrally accessible and distinct from the bulk OH groups (which do not exchange with D 2 O or HOD). Selective deuteration methods have been used to characterize the surface hydroxyl groups of various clay minerals (de Donato et al., 2004;Ledoux & White, 1964;Russell et al., 1970;Wada, 1967). The most definitive experiments, however, have been conducted on hydrous oxides (Parfitt et al., 1976;Rochester & Topham, 1978;Russell et al., 1974;Sun & Doner, 1996). In the well-characterized goethite system, for example, selective deuteration studies have shown that goethite is characterized by two main types of surface OD groups. For singly coordinated Fe-OH groups, the corresponding n(OD) vibration occurs at a frequency of 2584 cm À1 in vacuo and for surface OH groups coordinated to two and three Fe atoms, the n(OD) band occurs at a frequencies of 2700À2680 cm À1 in vacuo. This method can be used to assess the involvement of surface OH (OD) groups in ligandexchange reactions. For example, exposure of goethite to arsenate in aqueous suspension resulted in the complete loss of the singly coordinated OD groups at 2584 cm À1 resulting from the formation of an inner sphere surface complex through ligand exchange (Sun & Doner, 1996). Similar results have been reported for goethite reacted with phosphate (Parfitt et al., 1976;Russell et al., 1974). The disadvantage of this approach is that the surface OD bands can only be observed in a high vacuum.
At the next larger scale shown in Fig. 1 we will consider clayÀwater and clayÀorganic interactions. The length scales associated with these interactions are not clearly defined but some reasonable boundaries can be established. For clayÀwater interactions, we assign a length scale which ranges from the diameter of the H 2 O molecule (0.3 nm) to the approximate upper limit of diffuselayer swelling of 10 nm (Lagaly, 2006;Norrish, 1954;Viani et al., 1983). The scale for clayorganic interactions considered here is based on the size of the organic solutes themselves. This extends from the size of small polar molecules (~0.4 nm) to larger biomolecules, such as proteins, with sizes that approach 5 nm. In terms of porosity, these length scales span from the 'microscopic' (<2 nm) to 'mesoscopic' (2 to 50 nm) scales defined by the IUPAC and applied to clay minerals (Salles et al., 2009;Sing et al., 1985).

Nanoconfined H 2 O in clay interlayers
The molecular interactions of water with clay minerals are critically linked to essentially all chemical and physical aspects of clay science. Since the first reported IR study of water-smectite interactions in 1937 (Buswell et al., 1937), these interactions have been the subject of intense study using a broad spectrum of sophisticated experimental and computational approaches (Michot et al., 2002;Newman, 1987;Sposito & Prost, 1982). Of interest here are the spatial scales at which clayÀwater interactions take place. The chemical mechanisms that contribute to clayÀwater interactions include H bonding, charge-dipole attraction, ligand-ligand repulsion, and van der Waals interactions. In addition, contributions can also occur through hydrolysis and redox reactions depending on the nature of the exchangeable cation. Although most of these forces operate at distances that do not exceed a few nanometres, well defined long-range ordering of clay particles does take place resulting from a 'long range attractive interparticle force' which is complex and thought to be electrostatic in nature (McBride, 1997;McBride & Baveye, 2002). The underlying mechanisms that are responsible for diffuse-layer swelling are the continued subject of debate in the literature (Laird, 2006;McBride, 1997;McBride & Baveye, 2002;Quirk, 2003).
In many respects, interfacial H 2 O molecules in clay interlayers and at the external periphery of clay particles are a fundamental 'part' of its structure. Removal of water, especially the water closely associated with the clay surface, can irreversibly change these minerals. This is the case for halloysite, for example, where removal of free and lumen water results in the irreversible collapse of the halloysite structure from a d spacing of 1.0 nm to 0.7 nm (Joussein et al., 2005). Similarly, attempts to remove all of the water from smectites exchanged with some cations (e.g. Li + , Na + , K + , Cs + and Mg 2+ ) through heating and evacuation result in their irreversible collapse resulting from the Hofmann-Klemen effect (Brindley & Ertem, 1971;Farmer & Mortland, 1966;Hofmann & Klemen, 1950;Jaynes et al., 1992;McBride et al., 1975b).
We will focus our attention here on smectites as they are among the most interesting and important clay minerals when considering clayÀwater interactions . The physical picture that has emerged from a wide range of experimental and computational studies regarding smectiteÀwater interactions is summarized as follows. The size and shape of smectite particles are very variable with shapes that vary from rhombic to subhedral lamella, hexagonal lamella to lathes and fibres (Guven, 1988;Ras et al., 2007b) and basal dimensions that vary from tens of nm, as in the case of laponite, to a few micrometres for montmorillonite. In recent years, the development of Langmuir-Blodgett methods applied to clay science has provided new insights into the size, shape and particle-size distribution of smectites. Individual smectite particles consist of multiple layers of 1 nm-thick 'fundamental particles' which are generally stacked on top of each other (Lagaly & Malberg, 1990;Schramm & Kwak, 1982) (Fig. 7). These particles have large aspect ratios and have a morphology similar to that of a torn sheet of paper (Lagaly, 2006;Ras et al., 2003). The number of particles stacked on top of each other varies as a function of the size and charge of the cation and on ionic strength. In the case of Na + and Li + at low ionic strength, the particles can be completely disarticulated into individual one nm-thick particles. For other cations, up to seven layers can be stacked on top of each other in structures referred to as tactoids as shown in Fig. 7 (Lagaly, 2006).
As shown in Fig. 8, isomorphous substitution in smectites can occur in the octahedral or tetrahedral sheet. Octahedral substitution (e.g. Mg 2+ replacing Al 3+ ) results in a charge deficit which is delocalized over~8 surface oxygen atoms on each side of the fundamental particle, whereas substitution in the tetrahedral layer (e.g. Al 3+ for Si 4+ ) results in a more localized charge only on one side of the particle. Depending on the extent of substitution, the distance between the negatively charged sites ranges from 0.9 to 1.35 nm (1.2 to 1.8 mmoles/m 2 ) (Ś rodoń & McCarty, 2008). However, because smectite layers are stacked on top of each other, the effective surface charge density is increased by a factor of two and the separation between negative charge sites is further reduced to 0.65À0.84 nm (Johnston & Tombacz, 2002;. The interfacial chemistry of smectites is defined principally by interactions that occur between or within cationÀwater clusters in the interlayer region. Although the intrinsic negative surface charge of the clay can be balanced completely by cations forming inner sphere surface complexes, the strong hydration requirements of the most commonly occurring interlayer cations (Ca 2+ , Mg 2+ , K + and Na + ) maintains some metalcoordinated H 2 O between the clay layers. H 2 O molecules are attracted to metal cations through electrostatic interactions that maximize chargedipole attraction and minimize ligand-ligand repulsion . The hydrated cations are also attracted to the clay surface through electrostatic interactions (Rotenberg et al., 2009;Sutton & Sposito, 2001;Teppen & Miller, 2006). However, the binding energies of H 2 O molecules to the cation are large and result in the partial-to-full hydration of the exchangeable cations. Estimates of the binding energy of H 2 O molecules to metal ions from the vapour phase on a per-H 2 O molecule basis range from <50 kJ mol -1 to >320 kJ mol -1 depending on the nature of the cation and on the number of H 2 O molecules present (Rao et al., 2008). Thus, the interaction of the cation with the negatively charged siloxane surface is strongly modified by the presence of water.
These metal-coordinated H 2 O molecules are distinct from bulk water in several key aspects. First, they are polarized by the metal ion they are coordinated to through strong charge-dipole interactions. Figure 9 shows a top-down view of the siloxane surface showing the approximate spacing of hydrated monovalent metal ions corresponding to a medium charge density smectite (distance between cations of 0.75 nm) for a one-layer hydrate (d spacing of 1.25 nm). The approximate molarity of cations in the interlayer region is 12 M, with 4 H 2 O molecules coordinated to each cation. At small water contents, the cations direct the H 2 O molecules into coordination sites around the metal ions. In effect, this reduces H-bonding between H 2 O molecules. Unlike electrolyte solutions, the clay surface itself functions as a rigid, planar anion. As shown in the side view (left side of Fig. 9), these H 2 O molecules are 'nanoconfined'. For the one-layer hydrate, the gallery height of the interlayer is 0.3 nm. Depending on the hydration requirements of the cation, the type and extent of isomorphous substitution, and the overall water content, the height of the gallery is very variable. Distances of up to three layers of H 2 O (0.9 nm) are common. For monovalent cations at low ionic strength on low-to-medium charge smectites, these distances can increase to 4 nm and beyond. Interlayer distances of 10 nm have been reported (Norrish, 1954;Viani et al., 1983) and in some cases, complete separation or disarticulation of the layers is possible. Nanoconfined H 2 O is a feature common to a number of different types of clay minerals (Ockwig et al., 2009).
In this context, H 2 O clusters around exchangeable cations function as a hydrophilic probe of the clay surface. The planar K(H 2 O) 4 + complex shown in Fig. 9 has an approximate footprint of 0.760.7 nm. The reactivity of the hydrated metal cluster is controlled, in part, by the ionic potential Probing the nanoscale architecture of clay minerals of the exchangeable cation. For metal ions with large ionic potentials (e.g. Fe 3+ and Al 3+ ), these hydrated clusters are a source of acidity resulting from the hydrolysis of coordinated H 2 O. From a spatial perspective, the size of the hydrophilic probe can be controlled by the charge and enthalpy of hydration of the exchangeable cation. It has been shown recently that in addition to influencing cation selectivity in the smectite interlayers, the energetics of cation hydration in the aqueous solution phase are major determinants of cation exchange behaviour (Salles et al., 2008;Teppen & Miller, 2006). As will be discussed in the clayÀorganic section, the size of these hydrophilic domains is often inversely proportion to organic solute sorption. One useful method to study clay-water interactions is to examine the properties of water itself. Vibrational spectroscopy has been successfully applied to study the interaction of water with clay surfaces Poinsignon et al., 1978;Russell & Farmer, 1964;Sposito et al., 1983;Xu et al., 2000). The major vibrational bands of H 2 O occur in two regions corresponding to the OH stretching and H-O-H bending regions (Johnston & Premachandra, 2001;Xu et al., 2000). In addition, the vibrational spectrum of H 2 O is also characterized by a broad band at~2127 cm À1 and lower frequency librational bands < 800 cm À1 (Carey & Korenowski, 1998;Venyaminov & Prenderast, 1997). In the n(OH) region of H 2 O, the n(OH) bands of the clay interfere with those of H 2 O. In addition, these bands are broad and complex due to the IR-activity of both the symmetric and asymmetric n 1 and n 3 modes. The H-O-H bending band of H 2 O (n 2 mode), in contrast, occurs in a region relatively free from any clay absorption bands and the position of this band is sensitive to the extent of hydrogen bonding between H 2 O molecules FIG. 8. Isomorphous substitution of Mg 2+ for Al 3+ in the octahedral layer (top) and Al 3+ and Si 4+ in the tetrahedral layer (bottom). Also shown are the approximate 'zones of influence' associated with this type of substitution. (Pimentel & McClellan, 1960. As hydrogen bonding is increased, the position of the n(OH) bands of H 2 O shift to smaller energy and that of the H-O-H bending mode is increased (Falk, 1984;Fukuda & Shinoda, 2008 Russell & Farmer, 1964;Xu et al., 2000). Similarly, the molar absorptivity of the H-O-H bending band of sorbed H 2 O at large water contents (12 H 2 O/Na + ) is 20 cm 2 /mmol (Xu et al., 2000) and equals that of bulk water (Venyaminov & Prenderast, 1997). At smaller water contents, similar to the situation depicted in Fig. 10 (upper left), the H 2 O molecules are strongly polarized by the cation. This is one example showing that the clay surface, and especially the proximity of the H 2 O molecules to the interlayer cations, perturbs the physiochemical properties of water. In addition, structural OH groups contained within the clay structure itself are influenced by changes in water content. Among the first to show this phenomenon were Sposito & Prost (1982) who showed that the intensity of the structural OH bending bands increased significantly at larger water contents. Subsequent studies have quantified the change in molar absorptivity and the influence of water content on their band positions (Xu et al., 2000). In the case of montmorillonite, for example, there are three types of structural OH bending bands corresponding to Al-OH-Al, Al-OH-Fe, and Al-OH-Mg groups which occur at 918, 886 and 845 cm À1 , respectively. The change in band position of these three bands is shown as a function of water content in Fig. 10 (right side). It is interesting to point out that the two structural OH bending bands whose band positions are most perturbed by lowering the water content, are associated with isomorphous substitution. Upon lowering the water content to <6 H 2 O/Na + ions, a more direct cationÀsurface Probing the nanoscale architecture of clay minerals interaction can take place resulting in the observed shifts (Xu et al., 2000). The influence of water content on layer spacing and on the structural OH bending modes of the clay are also finding application in the interaction of biological molecules with clay surfaces (Haack et al., 2008). These results from our laboratory are a few examples of how water and exchangeable cations interact with clay surfaces. An ever-increasing array of sophisticated experimental and computational techniques is being applied to clay-water systems. Examples include high-resolution X-ray reflectivity (Park et al., 2008;Schlegel et al., 2006), nonlinear optical methods (Ogawa & Kuroda, 1995), vibrational spectroscopies (Johnston & Premachandra, 2001;Johnston et al., 1992;Poinsignon et al., 1978;Ras et al., 2003;Rinnert et al., 2005;Russell & Farmer, 1964;Xu et al., 2000), NMR spectroscopy (Bowers et al., 2008;Delville, 1992) and electron spin resonance (ESR) spectroscopy (Clementz et al., 1973(Clementz et al., , 1974McBride & Mortland, 1974;McBride et al., 1975a,b,c). Studies that examine the dynamic properties of water in clay interlayers include neutron scattering (Bordallo et al., 2008;Cebula et al., 1981;Chang et al., 1997;Swenson et al., 2000), dielectric relaxation (Belarbi et al., 2007;Logsdon & Laird, 2004;Salles et al., 2008) and NMR relaxation methods (Bowers et al., 2008). These experimental efforts are being supported and extended by a wide array of increasingly sophisticated computational methods.
In addition to the use of water as a molecular probe, the exchangeable cations themselves can be used to probe the clay surface. Some of the early ESR studies of Mortland, McBride and Clementz (Clementz et al., 1973(Clementz et al., , 1974  complexes in the smectite interlayer which is shown in Fig. 11 (Clementz et al., 1973). At present, nuclear magnetic resonance (NMR) methods are commonly used to study the molecular environment of NMR-active cations and a partial listing of NMR nuclei which have been successfully used in clay studies is shown in Fig. 12. In recent years, cationspecific high-resolution X-ray reflectivity methods are being applied to clay-related systems, such as the external surface of muscovite (Park et al., 2008;Schlegel et al., 2006). ClayÀorganic interactions are largely dependent on waterÀcation interactions at the clay interface. From the perspective of surface chemistry, the extent to which the clay surface is rendered hydrophilic depends on the type of cation, the surface charge density of the clay and the predominant location of isomorphous substitution in the clay lattice. Smectite surfaces generally show selectivity for organic cations and protonated organic bases (Johnston, 1996;Mortland, 1970;Theng, 1974). In broad terms, sorption of all other organic solutes can occur to some extent if the clay surface is not strongly hydrophilic. As will be developed below, sorption of semi-to-nonpolar organic solutes can occur in smectite interlayers for low charge, tetrahedrally substituted clays exchanged with less hydrated monovalent cations (e.g. K + or Cs + ). For strongly hydrated cations , the 'hydration spheres' of these cations are larger than the average separation between cations (0.65 to 0.85 nm, see Fig. 9) and the organic solutes cannot effectively compete with H 2 O for interlayer sorption sites.

C L A Y ÀO R G A N I C I N T E R A C T I O N S
Similar to the use of hydrated metal clusters as hydrophilic probes of clay minerals, the reactive sites, topography and hydrophilic-hydrophobic character of clay minerals can also be accessed through studies of organic solutes and polymers. The use of clayÀorganic materials in a wide range of disciplines including nanoscience and nanotechnology is rapidly expanding as attested by the number of publications and patents being issued each year (Dubois, 2007). A comprehensive review of clayÀorganic interactions is beyond the scope of this review and has been the subject of review (Johnston, 1996;Lagaly et al., 2006;Ras et al., 2007b). Unlike prior reviews of clay organic interactions that have focused mainly on classes of organic solutes (e.g. organic cations, bases, etc. À see Johnston, 1996;McBride, 1994;Mortland, 1970;Zielke et al., 1989), emphasis here is placed on the use of organic solutes as molecular probes of clay surfaces and is organized around the size, shape, polarity and reactivity of the organic solutes. These extend from small polar molecules used to intercalate kaolin subgroup minerals to larger more complex organic solutes.
Probing the nanoscale architecture of clay minerals Small polar molecules: kaolinite intercalates As discuss earlier, the interlayer H-bonds in kaolin subgroup minerals are responsible for their geochemical stability and predominance in very weathered soils. These bonds are shown in the [010] projections of kaolinite in Fig. 3. From an energetic standpoint, these layers are difficult to 'pry apart' and, as a result, intercalation is limited to a handful of small molecules shown in Fig. 13. These molecules are small enough to be intercalated between the layers and have polar characteristics such that favourable surface interactions can occur with both types of surfaces found in the interlayer region. In nature, halloysite (middle of Fig. 14) is the only naturally occurring intercalation complex of kaolinite and its occurrence, morphology and characterization have recently been reviewed (Joussein et al., 2005). The structure of halloysite consists of one layer of H 2 O sandwiched between each fundamental layer of kaolinite (4 H 2 O molecules per unit cell). Thus, all of the interlayer innersurface OH_O pairings are replaced by innersurface OHÀH 2 O and H 2 OÀsiloxane interactions.
As noted earlier, the n(OH) vibrations of the structural OH groups of kaolin subgroup minerals are very sensitive to changes in their local environment resulting from layer stacking disorder, as well as changes in temperature and pressure. Similarly, the n(OH) bands of kaolin subgroup minerals are strongly perturbed upon intercalation as the existing kaolinite in the interlayer. These compounds include dimethyl sulphoxide, formamides, hydrazine, and ureas (Fig. 13). As such, these molecules function as polar molecular probes of the kaolinite interlayer. The crystal structure of the kaolinite-DMSO complex with a d spacing of 1.13 nm and that of the kaolinitehydrazine complex with a d spacing of 0.95 nm are shown in Fig. 15 Thompson & Cuff, 1985). For the kaolinite-DMSO complex, the oxygen atom of the sulphonyl group interacts with the inner-surface OH groups. These molecules have the ability to interact favourably with both the siloxane and hydroxyl surface and to overcome the existing interlayer attractive forces. As molecular probes, the effective length scale here is the molecular dimension of the intercalated species and is limited to the size of the intercalates which is <0.5 nm. Although direct intercalation is limited to the

FIG. 13. Structures and electrostatic potential maps of 'primary intercalates'
Probing the nanoscale architecture of clay minerals small group of intercalates shown in Fig. 13, larger molecules can be introduced into the interlayer by first expanding the interlayer using one of the 'primary intercalates' shown in Fig. 13. These materials, however, can be used as intermediates and subsequent intercalation is possible to d spacings of 1.7 nm and beyond (Elbokl & Detellier, 2008;Letaief et al., 2008). Vibrational and NMR spectroscopy combined with X-ray diffraction and thermal analysis have been the traditional methods to study intercalation processes (Johnston et al., 1984;Ledoux & White, 1966;Olejnik et al., 1968Olejnik et al., , 1971Thompson, 1985;Thompson & Cuff, 1985) and remain among the most useful techniques. In addition to the perturbation of the kaolinite n(OH) modes, these intercalates have functional groups that are perturbed by the clay surface. Examples include the sulphonyl stretching vibration (n(S=O)) of DMSO or the vibrational modes associated with the ÀNH 2 group of hydrazine. Many different permutations of NMR spectroscopy have been used successfully to characterize these intercalation and functional materials. NMR spectroscopy provides an opportunity to examine both structural nuclei ( 29 Si, 27 Al, 17 O, 1 H and 2 H) as well as nuclei of the intercalated species (e.g. 13 C, 31 P, 1,2 H). In addition, the time scale associated with NMR spectroscopy is somewhat longer than that of vibrational spectroscopy and dephasing techniques can be used to study molecular motion in clay interlayers (Brandt et al., 2003;Elbokl & Detellier, 2008;Lipsicas et al., 1986;Tonle et al., 2007). An illustration of some of the NMR-active nuclei used to study kaolinite intercalation complexes, and related clay studies, is given in Fig. 12.
Recently, attention has focused on treatment of kaolinite through surface modification. These methods include chemical grafting of organic molecules onto the internal surfaces of kaolinite, or replacement of smaller intercalated molecules with larger molecules. One interesting example was the insertion of p-nitroaniline into kaolinite interlayers (Kuroda et al., 1999). Because of the dissimilar surfaces present in expanded kaolinite, the intercalated p-nitroaniline adopted an acentric arrangement where the ÀNO 2 groups interacted with the inner-surface OH groups and the ÀNH 2 group could favourably interact with the siloxane surface and the resulting kaoliniteÀp-nitroaniline intercalation complex was optically active. Current efforts are focused on surface modification through grafting reactions, such as organosilyl and ethoxy groups on the interlamellar surfaces of kaolinite (Tonle et al., 2007).
Synthesis and characterization of functional materials based on intercalation of organic guest molecules into layered inorganic materials is an area of considerable interest at present in the search for novel materials for nanocomposites, sensing, catalysis, improved thermophysical properties, adsorption and related applications. In the realm of clayÀorganic hybrid materials, kaolin group minerals have some interesting applications based on the fact that unlike 2:1 phyllosilicates, kaolin subgroup minerals have a permanent dipole moment due to their structural asymmetry. Unlike 2:1 clay minerals, guest molecules intercalated into kaolin subgroup minerals interact with two distinct types of surfaces (Johnston & Tombacz, 2002;. One noteworthy aspect of kaolinite intercalates is that, in some cases, the intercalated species can partially 'key' into the siloxane ditrigonal cavity. This has been observed for kaolinite where partially collapsed halloysite-like structures with d spacings 0.84 and 0.92 have been prepared (Costanzo & Giese, 1985Costanzo et al., 1982). Similarly, the ÀNH 2 group of hydrazine has also been observed to partially key into the siloxane ditrigonal surface where the reversible partial collapse of the structure has been observed (Cruz & Franco, 2000;Johnston et al., 1987Johnston et al., , 2000Johnston & Stone, 1990). Similar phenomena have also been observed in 2:1 phyllosilicates where interlayers cations can partially penetrate into the clay surface (Fernandez et al., 1970). In all of these examples, evidence for keying into the clay surface is derived from the partial collapse of the structures observed through XRD and infrared spectroscopy where the inner OH, the structural OH group located between the sheets, is perturbed in the collapsed form. The point to be emphasized here is that, in some cases, the interaction of the organic solute with the clay surface can be observed through structural components of the clay itself.

Neutral organic contaminants (NOCs): nitroaromatic smectite interactions
The fate of neutral organic contaminants (NOCs) in soil and subsurface environments has been studied Probing the nanoscale architecture of clay minerals intensively for the past 40 years. Although NOCs include a wide range of compounds, for the sake of illustration we will focus our attention on nitroaromatic compounds (NACs). Although beyond the scope of this review, the interaction of triazines (Barriuso et al., 1994;Chappell et al., 2005;Laird et al., 1992) and carbamate compounds (De Oliveira et al., 2005;Haack et al., 2008) appear to follow similar trends and, in this context, provide functional-chemistry specific probes of clay surfaces. Haderlein and co-workers were among the first to show that certain NACs exhibited a great affinity for smectites (Haderlein & Schwarzenbach, 1993;Haderlein et al., 1996;Weissmahr et al., 1997) with wide variation in sorption depending on subtle difference between the NACs. Subsequently, molecular details about their interaction with smectites was provided through coupled sorption-spectroscopic-structural-molecular modelling studies (Johnston et al., 2004). Within this context, the following structural features are relevant. When the ÀNO 2 groups do not occupy adjacent ring positions and bulky substituents are not present, NACs adopt a planar configuration. When two and especially three such coplanar ÀNO 2 groups are present on a single ring (i.e. 1-3 or 1-4 dinitrobenzene (DNB) and 1,3,5trinitrobenzene (TNB)), sorption on smectite is significantly enhanced showing the preference of smectite surfaces for planar structures Weissmahr et al., 1997). For planar molecules, or molecules containing planar substituents, polarized FTIR methods can be used to determine the molecular orientation of the sorbed species (Ahn & Franses, 1992;Ras et al., 2007a). In the case of dinitro-o-cresol (DNOC) sorbed on montmorillonite, polarized FTIR and X-ray diffraction (XRD) data revealed that the molecular orientation of DNOC was parallel to the clay surface (Sheng et al., 2002).
inter-and intramolecular interactions and have been used as indicators of change in the local environment of the sorbed nitroaromatic compound. In particular, the positions and intensities of the symmetric and asymmetric nitrogen oxygen stretching bands (i.e. n asym (NO) and n sym (NO)) are sensitive to molecular interactions in the local environment of the ÀNO 2 groups. Analysis of these bands for a number of NACs sorbed on different types of smectites (e.g. high-charge, lowcharge, reduced-charge clays) exchanged with alkali and alkaline earth metal ions revealed that the ÀNO 2 groups interacted favourably with the exchangeable cations in the interlayer Johnston et al., , 2002aSheng et al., 2002). These interactions have been the subject of a recent review and are summarized briefly here. With respect to clay properties, greatest sorption occurs for 'lowcharge' smectites that are exchanged with weakly hydrated exchangeable cations (e.g. K + and Cs + ). NAC sorption determinants are the number of ÀNO 2 groups, planarity of the structure and presence of other substituents on the ring. The greatest interaction between NAC ÀNO 2 groups and the exchangeable cations was observed for weakly hydrated cations. As noted earlier, the enthalpy of hydration of the exchangeable cation determines the overall size of the hydrated metal cluster in the clay interlayer. Favourable stabilization energy between the NAC and exchangeable cation can occur provided that the ÀNO 2 groups can gain molecular access to the cation. Increased stabilization is also provided by the presence of additional ÀNO 2 groups. The charge density of the clay also plays a significant role. If the hydrated metal clusters in the clay interlayer are too close, such that no sorptive domains exist for the NAC, then sorption will be diminished. The region between the isomorphous substitution sites on the clay surface is termed the neutral siloxane surface. There has been some debate in the literature about the role of this surface in providing additional stabilization energy for NACs with electron density from the siloxane oxygen atoms being donated to the delocalized p-electrons of the aromatic ring Weissmahr et al., 1998;Zhu et al., 2004).

Heterocyclic organic compounds: dibenzo-pdioxin and dibenzofuran
Among the most toxic and carcinogenic compounds known to humans are the tricyclic aromatic ether compounds known as dioxins. The chlorinated dioxins, toxic end-members of the group known as persistent organic pollutants, can activate biological receptors at exceptionally small aqueous solution concentrations. As a result, regulatory limits for dioxins are orders of magnitude lower than any other non-radioactive contaminant and this has prompted considerable interest in its environmental chemistry. Through an unusual course of events, starting with an occurrence of elevated levels of dioxins in chickens (Ferrario & Byrne, 2000;Hayward & Bolger, 2005;Hayward et al., 1999), an unlikely connection between these nonpolar organic solutes and clay minerals was established. As discussed earlier, most clay minerals in general, and smectites in particular, are strongly hydrophilic. At the same time, however, recent studies have established that certain smectites have a great affinity for dibenzop-dioxin (DD) and dibenzofuran (DF) in aqueous sorption studies Rana et al., 2009). As a class of molecules, dioxins provide an interesting suite of hydrophobic probe molecules.
Dioxins are hydrophobic, planar compounds that do not have polar functional groups. In fact, their adverse biological activity is associated with this structural feature. Activation of the aryl hydrocarbon receptor occurs through a binding site that is a hydrophobic 'slot' formed between two proteins that measures 1.4 nm61.2 nm60.5 nm (Denison & Nagy, 2003). As shown in Fig. 16, sorption of DD on smectite from aqueous suspension is largely dependent on the nature of the interlayer cation and clay type Rana et al., 2009). Sorption of DD from aqueous suspension favours low-charge, tetrahedrally substituted clays exchanged with weakly hydrated cations. Sorption on Cs-saponite is~10,000 mg of DD/kg saponite or 1% by mass. What can we learn about the clay surface from the interaction of DD with surface of Cs-saponite? First, the role of the exchangeable cation plays a critical role. Although NAC sorption was enhanced by weakly exchanged cations, in the case DD or DF, sorption is >2 or 3 orders of magnitude greater for Cs-exchanged clays. Second, sorption is much greater on tetrahedrally substituted clays (e.g. saponite) compared to smectites with octahedral substitution. When isomorphous substitution occurs in the octahedral sheet, the charge is delocalized over >6 surface oxygen atoms and this occurs on both sides of the fundamental particles (now for a total of 12 oxygen atoms). In contrast, the charge is highly localized in tetrahedrally substituted clays and this only occurs on one side (see Fig. 8). The net result is that significantly larger 'patches' of the neutral siloxane surface that are not influenced by isomorphous substitution occur in saponite and this, in turn, is responsible for the greater sorption of DD on saponite.
Similar to nitroaromatic compounds, planar molecules like DD have in-plane and out-of-plane vibrational modes. Based upon polarized FTIR measurements (Rana et al., 2009) and XRD analysis , the orientation of DD in the interlayer was tilted at an angle of~30º with respect to the clay surface. The molecular symmetry of DD is centrosymmetric; as a result, certain vibrational modes are observed only in the IR and others only in the Raman mode. When sorbed to Cs-saponite, however, the selection rules of sorbed DD are relaxed, showing activity in both Raman and IR, indicating that the molecular symmetry of DD was 'broken'. When combined with the tilted orientation, the molecular configuration of the DD in the clay interlayer was consistent with one of the etheroxygen atoms resting on a Cs + cation in the interlayer as shown in Fig. 17. In these experiments, DD and DP provide unlikely but useful probes of the smectite surface, showing a great degree of specificity for non-hydrated sorption sites on the neutral siloxane surface.

Interaction of biological molecules with clay minerals: clayÀprotein interactions
Interest in the interaction of biomolecules with minerals has significantly increased in recent years in a wide range of disciplines that span from nanobiotechnology to nanoelectronics and nanome-dicine (Bromley et al., 2007;Darder et al., 2007;Patil & Mann, 2008). For a number of years, immobilization of enzymes (bio-immobilization) on clay minerals, wherein the stability and activity or the enzyme is preserved, has been the subject of attention (Albert & Harter, 1973;Chen et al., 2008;Garwood et al., 1983;Harter & Stotzky, 1971;Mortland & Lawless, 1983). Of particular relevance to biomoleculesÀclay interactions is the dual hydrophobic-hydrophilic character of clay minerals described above and in greater detail recently by . The interaction of proteins with clay minerals has been studied for many years within the realm of soil chemistry (Albert & Harter, 1973;Boyd & Mortland, 1985;De Cristofaro & Violante, 2001;Ensminger & Gieseking, 1939, 1941, 1942Garwood et al., 1983;Harter, 1975;Harter & Stotzky, 1971;Mclaren & Peterson, 1961;Mclaren et al., 1958;Naidja et al., 1995). Many different types of proteins and enzymes are released into the soil and are retained on clay mineral surfaces (Gianfreda et al., 2002). Proteins and related biopolymers bound to clay minerals serve to regulate biological activity in soils and maintain certain aspects of microbial nutrition. The interaction of enzymes with clay surfaces modifies their catalytic activity and intercalated proteins are somewhat protected in the interlamellar region. At the same time, however, their catalytic ability is compromised because substrates must be able to enter into the interlamellar region and to follow a tortuous diffusion path in order to reach the active site.
Proteins represent a large and complex suite of structural probes to examine clay minerals. The size, shape, charge, flexibility, reactivity and hydrophilic/ hydrophobic nature of proteins are diverse. To promote this concept, we will focus our attention on the interaction of lysozyme with smectites. Lysozyme has long been viewed as a model protein to study clayÀprotein interactions because of its great affinity for smectites and ability to intercalate into clay interlayers (Albert & Harter, 1973;Harter, 1975;Harter & Stotzky, 1971;Mclaren & Peterson, 1961;Mclaren et al., 1958;Violante et al., 1995). In fact, over 50 years ago, McLaren and co-workers established that interlayer expansion of montmorillonite could be used as a calliper for the size of proteins (Mclaren & Peterson, 1961;Mclaren et al., 1958). As a probe molecule, lysozyme is a hard, compact, globular protein with an isoelectric point (IEP) of 11.1 and a mass of 14.3 kDa. Sorption of lysozyme occurs in excess of the cation exchange capacity and results in interlayer expansion to d-spacing values of >4 nm consistent with a monolayer based on the dimensions of the lysozyme molecule of 4.563.063.0 nm (Walsh et al., 1998;Wang et al., 2007). The mechanism of interaction is through cation exchange with~10 Na + ions released for every molecule of lysozyme sorbed (Albert & Harter, 1973). The IEP of lysozyme is 11.1 and each molecule has approximately ten positive charges at pH values of 5 to 6 (Huang et al., 2007;Spassov & Yan, 2008;Tanford & Wagner, 1954). The distribution of ten charges over a surface of roughly 33 nm 2 /lysozyme molecule corresponds to an average distance between charge sites of roughly 1.85 nm, or slightly greater than the distance between charge sites on the clay. Furthermore, the surface of lysozyme comprises polar sites separated by hydrophobic patches similar to the distribution of polar and hydrophobic sites on smectites (Sivozhelezov et al., 2006). Lysozyme is a fairly rigid globular protein. Prior structural and spectroscopic studies have established that its secondary structure is recalcitrant to structural modification (e.g. denaturation). Lysozyme is comprised of 129 amino acids and one measure of protein secondary and tertiary structure can be obtained through FTIR spectroscopy and analysis of the 'amide bands'. In studies of lysozyme sorbed on clay minerals, these bands indicate that the secondary structure remains intact on the clay surface (Szabo et al., 2007). However, these bands are sensitive indicators of conformational changes in larger and more flexible 'soft proteins', such as bovine serum albumin and chymotrypsin (Baron et al., 1999;Noinville et al., 2004;Servagent-Noinville et al., 2000). In addition to providing a large suite of complex molecules which function as surface/structural probes, these molecules can have a catalytic ability which opens the door for supported enzymes (Garwood et al., 1983;Mortland, 1984;Quiquampoix & Burns, 2007).
The largest scale to be examined briefly in this review is the scale of the clay particles themselves. ParticleÀparticle interactions have many effects in clay science that include molecular theories of clay swelling, nematic crystals, all aspects of clay-based/ modified rheology, liquefaction, and colloid assisted transport, to name a few. Interactions at this scale have been the subject of an excellent comprehensive review (Lagaly, 2006) and a prototypical example is given here to illustrate the relationship of interactions at this scale with the smaller scales shown in Fig. 1. The synthesis, characterization and manipulation of ultrathin hybrid films is presented here as an illustrative example of a relatively new area of particle-scale research in clay science (Ras et al., 2007b). Ultrathin hybrid films based on clay minerals are finding application in nanoscience and nanotechnology and consist of monolayer and multilayer hybrid nanofilms comprised of layers of smectite particles alternating with layers of functional molecules. In this area of research, clay particles can be organized into films through a number of methods including casting, spin coating, self-assembling (also called fuzzy assembling) or layer-by-layer deposition, and through the Langmuir-Blodgett (LB) method (Ras et al., 2007b). We limit the discussion to the application of the Langmuir-Blodgett method. The initial application of the LB technique consisted of spreading a hydrophobic organo-clay complex dissolved in an organic solvent over water and allowing the organic solvent to evaporate (Inukai et al., 1994;Kotov et al., 1994). The approach that is currently being used involves spreading a cationic amphiphile dissolved in an organic solvent over a dilute aqueous suspension of clay particles as shown in Fig. 18 (Ras et al., 2003(Ras et al., , 2004a(Ras et al., ,b,c, 2007a. The cationic head group of the amphilicic surfactant has a great affinity for the negatively charged clay surface. In addition, the Probing the nanoscale architecture of clay minerals long alkyl chain(s) of the surfactant are responsible for the molecular buoyancy of the clayÀorganic complex. In this approach, the insoluble amphilicic compounds 'pickup' elementary clay particles. The method is dependent on having dispersed individual clay particles in the aqueous solution. Upon compression in the Langmuir trough, the clayorganic complexes floating on the surface of the water are compressed to a desired surface pressure and removed from the LB trough on a solid substrate as shown in Fig. 18. A number of head groups have been used to prepare ultrathin hybrid clay films and some of these are illustrated in Fig. 19. The long alkyl chains, designated by R, are not shown in the structure. The structure and electrostatic potential map of 3,3'-dioctadecyl oxacarbocyanine (abbreviated as OXA) are shown on the left side of Fig. 20 and the OXA-smectite complex on the right. The C 16 chains of OXA provide the molecular buoyancy of the clay-organic complex at the airÀwater interface with the head group of OXA electrostatically tethered to the clay surface.
One of the goals of nanoscience and nanotechnology is to produce new functional nanomaterials with low cost and high added value. At the same time, however, the amphiphilic surfactants provide a unique class of probe molecules to study clay surfaces. Furthermore, the resultant ultrathin hybrid clay films (UHCFs), consisting of one layer of 1 nm thick clay particles opens up new opportunities to study the surface chemistry of clay minerals and to study particle-particle interactions. The synthesis and characterization of ultrathin hybrid clay films have been the subject of a recent review (Ras et al., 2007b) and are briefly summarized here. Perhaps the most useful characterization method for studying UHCFs is AFM because it provides a direct means of visualizing the size and shape of individual 1 nm thick elementary clay particles, the organization of clay particles in the UHCF, and information about the thickness of the clay film and stacking of multiple clay particles on top of each other. Polarized FTIR methods have provided detailed insight into the orientation of the clay particles within the film, as well as on the optical density of the UHCF, information about the orientation of the amphiphilic cation on the clay surface, as well as the ordering of the alkyl chains. Fluorescence spectroscopy provides a sensitive method for studying the organization of, and interaction between, amphiphilic molecules with strong UV absorption features (e.g. cationic dyes like OXA or RhB18). Although at the limit of sensitivity, X-ray methods are being successfully applied to study UHCFs and multi-layered UHCFs. Yamagishi and co-workers have further exploited UHCFs by subsequent modification of the clay monolayers with various cationic metal complexes towards producing functional materials (Umemura et al., 2002) One such application involves the separation of chiral molecules on clay surfaces.

C O N C L U D I N G R E M A R K S
Considering the spatial length scales associated with H bonding, clayÀwater and clayÀorganic interactions provide a useful framework to better understand these complex materials. The small length scale associated with the naturally occurring FIG. 18. The hybrid film deposition process of the Langmuir-Blodgett technique (Ras et al., 2007b). nanomaterials is the basis for their profound influence of a wide range of processes. At the same time, however, understanding their structure and surface chemistry requires new tools adapted to the study of nanomaterials and interfacial processes. It is now well established that molecular spectroscopy combined with sorption, structural, and computational tools are the methods of choice. At the sub-nanometre length scale, H bonding interactions of the structural OH groups of clay minerals play a key role in defining the size, shape, geochemical stability and surface chemistry of 1:1 phyllo-silicates. In addition, these 'reporter groups' are currently being used to study particle size and shape, structural disorder, temperature and pressure effects as well as their interaction with water and organic solutes. Although difficult to study because of their small abundance relative to the total OH population, exposed surface OH groups, especially under-coordinated OH groups at the broken edges, exert a profound influence on the rheological properties of clays as well as their ability to bind metals and oxyanions. For both structural and surface OH groups, the H-bonding interactions they participate in which effectively operate at a scale of <0.4 nm, defines the size, shape and many aspects of the overall surface chemistry of clay minerals.
Hydrogen bonding, along with other types of chemical forces, also plays critical roles in determining how H 2 O molecules interact with clay minerals. H 2 O molecules are always associated with clay minerals because of their small size, expansive surface area and electrically charged nature. The complexity of clayÀwater interactions is examined on the basis of the type(s) of surface present, the electrically charged nature of the clay surface and the type of inorganic cations present. In the case of smectites, the clustering of H 2 O molecules around exchangeable cations is a type of hydrophilic molecular probe over a spatial length scale that extends from the size of H 2 O molecules (~0.3 nm) to the limit of clay swelling (~10 nm). To a large extent, water defines the surface chemistry of clay FIG. 19. Various cationic head groups used to prepare ultrathin hybrid films with clay minerals.
Probing the nanoscale architecture of clay minerals minerals. At the same time, however, these interactions play a moderating role in governing how organic solutes will interact with clay minerals. Organic solutes provide a diverse suite of molecular probes to explore clay surfaces on the basis of their size and surface chemistry. By definition, water is the prototypical hydrophilic surface probe. Polar surface interactions can also be explored using small polar organic solutes, such as the small polar compounds that can intercalate kaolin group minerals (e.g. hydrazine).
In the case of 2:1 phyllosilicates with isomorphous substitution, the siloxane surface has both a hydrophilic and a hydrophobic character. H 2 O molecules clustered around exchangeable cations provide one type hydrophilic probe of these surfaces. In addition, organic solutes with varying size, shape and polarity provide a useful set of probe molecules. In the case of neutral organic contaminants (NOCs) which have polar functional groups (e.g. nitroaromatics or triazines), these molecules can be stabilized in clay interlayers through the combined interaction of charge-dipole interactions and through van der Waals interactions with the neutral portion of the siloxane surface. For these molecules, sorption is governed by the charge density of the clay, location of charge deficit in the clay structure, charge and enthalpy of hydration of exchangeable cation, number and type of polar functional group of the organic solute as well as size and planarity of the solute. Recently, these concepts have been further extended by observation that certain strongly hydrophobic compounds have a great affinity for smectite clay minerals. In particular, dibenzo-p-dioxin provides an interesting hydrophobic probe molecule that shows a great affinity for certain smectites from aqueous suspension when exchanged with weakly hydrated monovalent cations. At the next larger scale (2 to 6 nm), complex biological molecules show a great affinity for clay minerals depending on their overall size and isoelectric point. For these types of molecules, stabilization on clay minerals is attributed to both polar and nonpolar interactions and is controlled by the overall charge and size of the protein.
The exceptionally great affinity of organic cations and protonated organic bases for clay minerals with permanent charge sites resulting from isomorphous substitution is well known and FIG. 20. The structure and electrostatic potential map of 3,3'-dioctadecyl oxacarbocyanine (OXA) is shown on the left side and the OXA-smectite complex on the right.
attributed to both the nature of the clay surface and to the thermodynamics of cation hydration. For the past 50 years, clay scientists have exploited some level of 'size matching' to develop clayÀorganic complexes, mainly in the area of organoclays. A relatively new area that is based on this same great affinity of organic cations for clay minerals is in the synthesis of ultrathin hybrid clay films. In this case, cationic head groups of varying size, shape and charge, combined with long alkyl chains, function as surfactant probe molecules. These molecules are generally of the order of about 2 nm in length but collectively they 'float' individual clay particles to the surface of the airÀwater interface and subsequent isolation in LB films. This approach is providing a new set of tools to study the organization and characterization of clay minerals.

ACKNOWLEDGMENTS
The paper is submitted as a contribution in honour of George Brown and his pioneering work in the area of clay mineralogy. The author would like express sincere appreciation to Stephen Agnew, David Bish, Stephen Boyd, Hui Li, Robin Ras, Robert Schoonheydt, Garrison Sposito, Brian Teppen and Shan-Li Wang for their friendship, scientific inspiration and valued collaborations. The project was supported, in part, by grant P42 ES004911 from the National Institute of Environmental Health Sciences (NIEHS). The contents are solely the responsibility of the author and do not necessarily represent the official views of NIEHS or NIH.