Colloidal suspensions of plate-like particles undergo a variety of phase transitions. The predicted isotropic/nematic transition is often pre-empted by a sol/gel transition, especially in suspensions of the most commonly used natural swelling clay montmorillonite (MMT). A number of factors, including charge interactions, flexibility and salt concentration, may contribute to this competition. In this study, the effect of surfactant adsorption on suspensions of MMT was studied using rheology, small-angle X-ray scattering, static light scattering and optical microscopy. The addition of a polyetheramine surfactant reduced the moduli of the system and shifted the sol/gel transition to a much higher clay concentration, compared with suspensions of bare clay particles. Yet, scattering data revealed no change in suspension structure on length scales up to around a micrometre. Primary aggregates remain at this length scale and no nematic phase is formed. There is, however, a change in structure at large length scales (of order 20 μm) where light scattering indicates the presence of string-like aggregates that disappear on addition of surfactant. Microscope images of dried suspensions also revealed a string-like structure. The dried strings show strong birefringence and may consist of concentric cylinders, self-assembled from clay sheets.
Colloidal suspensions of spherical particles have been studied in great detail. By contrast, non-spherical colloidal particles have received less attention but the focus is shifting towards them . Suspensions of anisotropic colloidal particles have unique properties, including the ability to form liquid crystalline phases. Gelation, however, often intervenes before a nematic phase is reached and this paper seeks to elucidate the causes for this.
The formation of a liquid crystalline phase in suspensions of anisotropic particles is explained by Onsager’s theory , which demonstrated that it is due to the gain in translational entropy overweighing the loss in orientational entropy at sufficiently high concentration. For suspensions of infinitely long rods, Onsager’s theory is exact and it has also been applied successfully to finite length rods . It is less straightforward to describe suspensions of platelets. However, the same qualitative argument holds for platelet suspensions, resulting in a nematic phase on increasing particle concentration [4,5]. For thickness/diameter (L/D) ratios ranging from L/D=0 for infinitely thin plates to L/D=0.3 for thick cut spheres, the location of the isotropic/nematic (IN) transition is nearly constant in terms of the reduced number density, 1.1 in which ρ is the number density and σ the platelet diameter [4,6–8]. For aspect ratios closer to one, the nematic phase is not expected to be stable . The IN transition is a weakly first-order transition, the concentration jump between the isotropic and nematic phases is only about 5 per cent for monodisperse platelets. In practice, polydispersity of the particles is likely to result in a widening of the coexistence gap. Nevertheless, equation (1.1) may serve as a guide to estimate the concentration required experimentally for observing a nematic phase. In the presence of electrical charges, which are ubiquitous in aqueous suspensions, the effects of electric double layer interactions need to be taken into account when calculating the effective platelet thickness. For charged spheres, the disorder-order transition concentration increases with salt concentration owing to the reduction of the effective hard sphere diameter . For charged plates, one might argue that the main effect of charge interactions is an increase in the effective platelet thickness, and hence, following equation (1.1), the IN transition would still occur at the same particle concentration. We return to this issue in the following section .
With respect to experiments, liquid crystal phases have been observed in a number of aqueous colloidal suspensions of rod-like particles, such as vanadium pentoxide , microcrystalline cellulose  and tobacco mosaic virus , as well as in non-aqueous suspensions of, for example, boehmite  or sepiolite . Langmuir  for the first time reported an IN transition in a suspension of plate-like ‘California bentonite’ particles, but unfortunately the result was not reproducible. Only recently, liquid crystal phases were observed in suspensions of synthetic plate-like particles, such as gibbsite (Al(OH)3) particles [17–19] and Ni(OH)2 particles .
Smectite clays have a hydrophilic surface and can swell in water. Aqueous suspensions of smectite clays consist of individual clay sheets of high aspect ratios, which are, therefore, expected to undergo an IN transition at a low concentration, typically a few wt%, following equation (1.1). However, the nematic phase has proved to be rather elusive in these systems. Only recently, have two smectite clays been found to display a nematic phase, over a narrow concentration range at low or medium salt concentrations: nontronite and beidellite. Michot et al. [21,22] reported a true IN transition in aqueous colloidal suspensions of (lath-shaped) nontronite particles. Later, a nematic phase was observed in suspensions of the disc-shaped clay beidellite by Paineau et al. . In both cases, further increase of clay or salt concentrations led to a gel [21–23].
Meanwhile, a sol/gel transition occurs instead of the expected IN transition in aqueous suspensions of some smectite clays, such as the widely used Laponite  and montmorillonite (MMT) . MMT aqueous suspensions form birefringent gels above 2 wt% , indicating some local ordering. However, no IN phase separation occurs. Laponite is a synthetic hectorite clay, with a small particle diameter (approx. 25 nm) and low polydispersity. Numerous studies have been devoted to uncover the phase behaviour [26–31] and interactions between particles in Laponite suspensions [32–35]. Some discrepancies exist between studies carried out by different groups, probably due to differences in sample ageing times and preparation methods. Interactions between Laponite particles were investigated by Pignon et al. [34,35] using scattering techniques, and two characteristic length scales were detected corresponding to small subunit clusters and large aggregates. The most recent development was reported by Ruzicka et al. , in which they used the concept of patchy particles to explain the structure evolution of Laponite aqueous suspensions made following strict protocols.
Compared to Laponite, far fewer studies have focused on suspensions of MMT, a natural smectite clay, perhaps due to its polydispersity. Early studies on MMT mainly investigated the swelling properties of the clay [36,37]. It was also found that the expected IN transition was replaced by a prevailing gel at low volume fractions [25,38,39]. The phase diagram of a Na+–MMT was studied in detail by Michot et al. . Although no IN transition was detected, flow birefringence was observed in dilute suspensions when observed between crossed polarizers. Furthermore, gel samples show a permanent birefringent texture. The effect of poly (ethylene oxide) polymer with two cationic end groups on the sol/gel transition of suspensions of MMT clays was studied by Lagaly & Ziesmer [38,39]. The addition of the polymers changed the stability of the suspensions, rather than promoting the IN transition.
The importance of number density and plate diameter for the IN transition is highlighted by equation (1.1). Other factors that are likely to affect the formation of liquid crystalline phases in plate-like clay suspensions include polydispersity, particle flexibility, particle stickiness and charge interactions. Equation (1.1) is based on the assumption that the platelets are hard and monodisperse. Particle size polydispersity was shown to be important for IN transitions both in computer simulations  and experiments [22,23], where a widened biphasic gap was predicted  and observed . Theoretical predictions by Odijk  suggest flexibility may suppress the IN transition Another factor is the particle stickiness, which could be the result of bridging interactions (in the presence of polymers) or van der Waals forces.
Electrostatic interactions need considering—especially, if the clay face and edge are oppositely charged, which would encourage the well-known house of cards structure and contribute to gel formation [42,43]. We return to this effect in detail below. Leaving the role of any edge charges aside for the moment, a key property of clay minerals is their layered structure, implying a ‘charge-sandwich’ structure. As a result, quadrupolar interactions between particles arise. Dijkstra et al. [32,33] performed computer simulations of Laponite particles modelled as infinitely thin platelets with a point quadrupole to represent the clay surface charge, counter-ions and co-ions. A resulting T-shaped configuration leads to the formation of a gel at low volume fraction, which agrees well with experiments at low salt concentrations. The point quadrupole approximation is most likely to be valid for small particle diameters and at low salt and clay concentrations, such that typical particle separations are large compared with the particle diameter. A more general approach was proposed by Rowan et al. , using a multi-pole expansion of the charge distribution. This model yields repulsive potentials for both T-shape and parallel configurations. However, for a given particle centre–centre distance, the T-shape configuration is less repulsive and thus favourable. In summary, the distribution of charges inside a clay platelet and surrounding it gives rise to an energetic preference for T-shape configurations, even without considering any particular edge–face interactions. In addition to the quadrupole interaction resulting from the charge-sandwich structure, the recent work by Fartaria et al.  suggests an in-plane dipole moment may also be present.
Recent studies on the IN transition in suspensions of the smectite clays nontronite and beidellite suggest that clays with isomophous substitution located closer to the clay surface are more likely to undergo an IN transition [21–23]. Paineau et al.  studied the effect of charge location on IN transitions using various clays, which all consisted of three layers: one octahedral layer sandwiched between two tetrahedral layers. The surface charges of these clays originate from the isomorphous substitutions inside these layers. By measuring the osmotic pressure of suspensions of different clays, they discovered that differences in the repulsive potential strength are related to the location of charges. The tetrahedrally charged clays beidellite and nontronite, in which the isomorphous substitution occurs mainly in the tetrahedral layers, show IN transitions and also experience stronger electrostatic repulsions than the octahedrally charged clays MMT and Laponite.
In the present work, aqueous MMT suspensions are studied and a polyetheramine surfactant (Jeffamine) is used to modify the surface in an attempt to switch off any particle stickiness, screen charge interactions and promote the suspected, hidden IN transition. Previously, the adsorption mechanism of Jeffamine on MMT was elucidated . Here, its effect on the phase diagram of aqueous MMT suspensions is investigated, using rheology, synchrotron X-ray scattering, static light scattering (SLS), small-angle light scattering (SALS) and microscopy. The addition of Jeffamine shifts the sol/gel transition to a much higher clay concentration. At length scales up to 10 μm, the scattering experiments did not reveal any change in structure. There was some evidence, however, of string-like aggregates at large length scales, which disappeared on addition of Jeffamine. Yet, the effect of Jeffamine was not sufficient to allow a liquid crystalline phase to form.
Wyoming SWy2 MMT was purchased from the Clay Minerals Society source clays repository at Purdue University. MMT is a smectite clay with negative surface charge and interlayer cations. The typical composition of SWy2 MMT is (Si7.94Al0.06)(Al2.88Fe0.5Mg0.62)O20(OH)4Na0.68 . The purification and fractionation of the clay was described in detail in a previous study . Jeffamine M1000 (referred to as M1000) is a polyetheramine with a molecular structure of CH3[OCH2CH2]19[OCH2(CH3)CH]3NH2 and was provided by Huntsman. It contains 96 wt% of primary amine .
(b) Sample preparation
The samples were made by mixing stock solutions of NaCl (without or with M1000) with fractionated MMT stock suspensions. The average diameter of the MMT particles was 270 nm with an s.d. of 100 nm . Two types of samples were prepared: bare MMT and M1000 pre-treated MMT, both over a range of NaCl concentrations. The cationic exchange capacity (CEC) of MMT is 84 meq per 100 g. The amount of M1000 added was kept at the plateau level of the adsorption isotherm, corresponding to 0.4 CEC of the clay (equivalent to 0.34 mmol g−1 clay) to ensure a good surface coverage yet minimize the concentration of free M1000 in solution.
All samples were sealed in glass vials and rotated end-over-end for 24 h and then aged for a further 72 h before any measurement. Samples for rheology and a phase diagram had an MMT concentration ranging from 1 to 6 wt%. All the clay concentrations in this paper refer to the weight percentage of pure MMT. To avoid multiple scattering at high MMT concentration, dilute samples were used for SLS (0.1 wt%) as well as for SALS (0.25 wt%), and these experiments were carried out immediately after the 72 h ageing period.
Oscillation experiments were carried out at 20°C using a controlled stress Bohlin CVO instrument. Owing to the low strain response of some dilute samples, a cone and plate (CP4/40, 4° and 40 mm diameter, un-roughened) geometry was used for all samples. The oscillation experiments were carried out at low stresses and small strains to avoid wall slip. An amplitude sweep was first performed at 1 Hz with the shear stress ranging from 0.2 to 10 Pa to find the linear region where the structure of the system is not disturbed . The delay time was set to 15 s and the number of samples was 15, allowing 16 s per data point. A frequency sweep was then carried out using a stress value within the linear range at frequency from 0.01 to 10 Hz. The number of data points on each run was 30. The samples were poured out from the vials and left to rest 5 min before each measurement. The elastic modulus G′ and the viscous modulus G′′ were obtained and the following criteria were used to identify a gel: (i) G′ is obviously higher than G′′ and (ii) G′ does not vary much over the whole frequency range .
(d) Small-angle X-ray scattering
The small-angle X-ray scattering (SAXS) data was collected from 1 wt% aqueous MMT suspensions using beamline I22 at the Diamond Light Source. The X-ray energy was 12.4 keV and two different sample-to-detector distances were used to cover q (scattering wavevector) values ranging from 0.004 to 0.4 Å−1. For the lower q range measurements, the q-axis was calibrated using rat tail tendon and for the higher q range data using silver behenate. The rat tail tendon and silver behenate calibrants were also used to define the beam centre position for each set of measurements. The datasets were normalized to the transmitted beam intensity, corrected for scattering from the background (in this case water), scaled and stitched together according to standard practice.
(e) Static light scattering
A Malvern 4800 Autosizer (Malvern Instruments, UK) was used to measure the static scattering intensity (SLS) at scattering angles from 10° to 150° at 25°C. A 10 mm diameter cylindrical quartz precision light scattering cell was filled with a dilute MMT suspension in water. The wavevector was calculated using equation (2.1), 2.1 where n0=1.33 is the refractive index of the medium, θ the scattering angle and λ=532 nm the wavelength of the incident beam. The typical length scale a probed in such a scattering experiment is inversely proportional to q, 2.2
(f) Small-angle light scattering
A homebuilt SALS apparatus with a 5 mW, λ=632.8 nm helium–neon laser was used to investigate the aggregation and phase behaviour of clay suspensions. The design of the set-up is similar to that described in previous studies by Verhaegh et al.  and Schätzel & Ackerson . A Stingray camera F125B ASG with a Tamron CCTV lens was used to capture the scattering pattern projected. AVT uniCam v. 1.1.3 software was used to operate the camera and store scattering images, with 646×482 pixels and 14 bit accuracy.
To avoid multiple scattering, all samples were prepared at low clay concentrations and the relative transmissions were all above 0.85. The scattered intensity was corrected for sample transmission, solvent background and camera dark current. The data were processed and radially averaged using ImageJ v. 1.42q software. A diffraction grating (300 lines mm−1) was used as a sample to calibrate the scattering vectors.
(a) Phase diagram of pure montmorillonite
Measurement of elastic and viscous moduli (G′ and G′′) and macroscopic observations were used to plot the phase diagram, shown in figure 1a (for the rheological data, see the electronic supplementary material). The moduli of the samples near the phase boundary were determined by rheology. The samples with very high clay concentration (5.5 wt%) were very viscous and could not be poured out from sample vials. At NaCl concentrations above 5×10−2 M, the double layer repulsion is sufficiently screened for the van der Waals attraction to be dominant and the gel formed in this region is attractive. At medium and low NaCl concentration, a gel also formed at MMT concentration above 3 wt%. This observation is in line with previous studies on MMT [25,42,49] and Laponite [26,52]. However, the clay concentration where the sol/gel transition occurred varies slightly, depending on the origin of the clay. As demonstrated in figure 1a, the sol/gel transition occurs at around 3 wt% clay in an aqueous system with bare clay and salt.
(b) Phase diagram of M1000-modified montmorillonite
M1000 adsorbs strongly to the clay surface and this has been shown to involve an ion exchange mechanism . The phase diagram changed dramatically after addition of M1000 to the system (figure 1b). The sol/gel transition clay concentration increased significantly at medium and low salt concentrations, to almost 6 wt% at 10−3 and 10−2 M NaCl, twice the gelation clay concentration in the system without M1000. The gel point at 10−4 M NaCl is also shifted to 5 wt%. At 10−1 M NaCl, a strong attractive gel is still found over the clay concentration range tested. Flocculation is observed at 1 M salt. The change in flow behaviour was observed visually and further confirmed by oscillatory viscometry (see the electronic supplementary material).
(c) Scattering techniques
To understand the marked change in flow properties of clay suspensions, scattering techniques were used to investigate the change in suspension structure upon the adsorption of M1000. SAXS was carried out for 1 wt% MMT samples with or without M1000, with a scattering vector q ranging from 0.04 to 4 nm−1 , as shown in figure 2. The resulting intensity versus scattering vector plots superimpose perfectly and they follow a power law with an exponent of −2, characteristic for suspensions of plate-like particles.
The absence of a pronounced peak in the high-q region, frequently observed in disc-like clays and attributed to platelet stacking, shows the clay platelets are well delaminated. There is, however, a very weak, broad feature around q=0.4 nm−1 (corresponding to a length scale around 15 nm), which suggests a weak tendency for platelets to be aligned parallel to each other at that separation. Previously, it was concluded that M1000 adsorbs onto the clay surface, forming a layer of polymer coils arranged like touching mushrooms ; the SAXS data suggest there is no accompanying change in structure adopted by the clay platelets at the length scales probed, ranging from 2 to 200 nm.
SLS was performed to detect larger length-scale structures in these suspensions. In order to avoid multiple scattering owing to the refractive index difference between clay and water, very dilute suspensions were used. The scattering patterns of 0.001 to 0.1 wt% clay with 0.0001 to 0.1 M NaCl were almost the same, following a power law of I∝q−3 in the q range from 4×10−3 to 1×10−2 nm−1 (figure 2). This scaling behaviour indicates a three-dimensional aggregate was formed by individual clay particles, with a size around 1.6 μm (estimated as 2π/qc, where the roll-off wavevector qc=4×10−3 nm−1).
Bare clay suspension and M1000 clay suspension gave the same result, suggesting that the adsorbed layer of M1000 is not sufficient to avoid the formation of these aggregates.
SALS was then used to investigate a series of MMT samples to detect any large length-scale structures (figures 2 and 3). SALS was only carried out in dilute clay suspensions (0.25 wt%) to avoid multiple scattering. For bare clay suspensions at salt concentrations of 0.0001 and 0.01 M, data follow a power law of I∝q−1 in the q range from 2×10−4 to 4×10−4 nm−1 (figure 3), which indicates a tenuous string-like structure, perhaps formed by self-assembly of smaller aggregates. The exception is at 0.001 M salt, where the plot is flat over the same q range. This will be discussed below. With M1000, all the scattering plots are almost flat in this range. Therefore, the string-like structures are not present after addition of M1000. Using scattering techniques, the only difference in the suspensions with or without M1000 observed is at length scales larger than 10 μm, according to equation (2.2). Presumably, the adsorption of M1000 decreases the interaction energy between the small aggregates and breaks down the string-like structure. The small aggregates regain their mobility and the moduli of the suspension decrease accordingly. This explains why the sol/gel transition concentration almost doubles in the presence of M1000.
(d) Optical microscopy
As the SALS data gave evidence for large-scale aggregate structures, attempts were also made to visualize these using differential interface contrast (DIC) microscopy. However, no meaningful images could be obtained on wet samples. In dried samples of bare clay suspensions however, dramatic string-like structures were found, as shown in figure 4a. A similar structure was also observed by Shalkevich et al.  using environmental scanning electron microscopy on wet MMT suspensions, where it was surmised that the string-like structure was due to drying at the sample surface.
Possibly any tenuous structures observed in the SALS experiments do not provide useful contrast in microscopy, yet these structures do lead to the dramatic aggregates seen after drying. It is, therefore, interesting to note that in the presence of M1000, where SALS suggests the string-like aggregates are absent, the pattern observed on drying is also completely different (figure 4b) with no marked strands forming.
At the concentrations used here, essentially all M1000 should adsorb to the clay surface leaving very little free in solution . Therefore, the difference in structures between figure 4a,b is purely caused by the adsorption of M1000 onto the clay surface. The width of the strings in figure 4a is around 3 μm; their length ranges from 20 to 30 μm, which agrees well with the string size in suspension, derived using SALS (greater than 10 μm). Dried bare clay suspensions were also observed by polarizing microscopy (see the electronic supplementary material). The string-like structures in the bare clay sample are highly birefringent. Viewing the same structures using a retardation filter  led to the conclusion that the clay particles were aligned parallel to the strand structures observed. A suggested structure of the strands formed upon drying, therefore, is that the clay particles form concentric cylinders, as shown in figure 4c. The fact that such birefringent textures could not be observed in the wet state suggests that the well-ordered structures formed as a result of the drying process, starting from the disordered, tenuous structures evidenced using SALS. Such birefringent structures were not seen in a dried clay sample treated with M1000.
(a) Aqueous clay suspensions
Across a range of length scales, the MMT suspensions here display clusters or aggregates. One factor often invoked to explain clustering is that of the role of edge and face charges on clay platelets. In contrast to the isomorphous substitution of ions in the octahedral layer that renders the clay surface negative, polar sites at the broken edges, including octahedral Al–OH and tetrahedral Si–OH groups, make the edges conditionally charged. Si–OH and Al–OH groups can be protonated or de-protonated by reacting with H+ or OH−, respectively, depending on the pH . At pH<4 and (1:1) electrolyte concentrations above 0.01 M, the edge is positively charged and exposed to other particles owing to the depression of the electric double layer, which results in a face-to-edge configuration . In the present work, however, the clay suspensions were dialysed and at their natural pH, around 9 . Under these conditions, the edges and faces are both negative . The fact that aggregates nevertheless form, presumably, is due to the distribution of charges in and around the clay platelets, leading to quadrupole (and higher order multi-pole) interactions between particles. Van der Waals forces can lead to coagulation, but this requires salt concentrations in excess of 0.1 M.
The primary aggregate structure appears to consist of homogeneous clusters with a size around 1.6 μm and a fractal dimension df≈3 (figure 2). While the q range involved is rather limited, there is evidence for string-like structures (df≈1) at length scales around 10 μm and above, especially at salt concentrations of 10−4 and 10−2 M (yet not at 10−3 M).
This behaviour is reminiscent of that observed in a detailed study of Laponite XLG suspensions . While these particles are chemically somewhat different from MMT, and their particle diameter is much smaller, they are nevertheless similar in many respects. The particles were also studied at a pH of 9.5, where edges and faces are both negatively charged . For a salt concentration of 10−3 M, at length scales ranging from 35 nm to 1 μm, homogeneous aggregates (df=3) were found; at larger length scales, dilute samples showed particle strings (df=1), whereas more concentrated samples (over 1 wt%) gave more dense large-scale structures with df≈1.8. It was concluded that the macroscopic flow behaviour of Laponite gels was mainly governed by the nature of the large-scale aggregates formed.
Returning to MMT suspensions, one particularly relevant study is that by Shalkevich et al. . That work used Cloisite CNa (which has a CEC of 98 meq per 100 g, slightly higher than the clay used here) and, notably, samples were de-ionized, leading to a low pH of 3.7, where the particle edges are positively charged . Clusters with df≈2.6, consisting of randomly oriented individual clay particles, were found at NaCl concentrations ranging from 10−5 to 0.01 M NaCl. SALS was used to characterize larger-scale structures and mostly these could be understood based on the clusters already referred to. A sample at 2 wt% clay and 10−3 M salt, however, showed scattering following I∝q−1; that same sample also showed a strand-like structure using environmental scanning electron microscopy, ascribed as possibly due to drying out near the sample surface. The image is strikingly similar to that shown here in figure 4a, obtained on a dried sample, also under conditions where the suspension showed evidence for strands.
In summary, both for Laponite and for MMT suspensions, previous studies have also found evidence for the existence of the strand-like structures proposed here, at least for certain clay and salt concentrations; upon drying (for MMT anyway), these develop into well-defined, birefringent fibrils, probably consisting of clay platelets arranged into concentric cylinders.
(b) Surfactant-modified clays
In the current study, the adsorption of the M1000 surfactant screens part of the surface charge at pH values ranging from 10 to 11, where the edges are negatively charged. The surfactant adsorption led to significantly reduced rheological moduli of the suspensions, and the onset of gel formation shifted to almost double the original concentration at a salt concentration of 10−3 M. Nevertheless, the primary aggregate structure at length scales up to about 2 μm was still present; surfactant adsorption did, however, result in a disappearance of any string-like larger-scale structures. Dried samples also revealed no fibrils (figure 4b). A similar conclusion for Laponite suspensions was reached by Pignon et al. : that the flow behaviour is determined by the size and shape of the largest aggregate structures present in the sample.
There have been few previous studies of the flow behaviour of aqueous, organically modified MMT suspensions. Lagaly & Ziesmer used poly (ethylene oxide) polymers with a range of molecular weights, with cationic groups at both ends. This resulted in dramatic changes to the phase diagrams. High molecular weight polymers (20 000 g mol−1 or more) shifted the sol/gel transition to higher clay concentrations, but an opposite effect was observed with low molecular weight ones . By contrast, the surfactant used in the present work has a low molecular weight (around 1000 g mol−1), yet it moves the gel transition to higher concentration. Contrary to Lagaly & Ziesmer , the surfactant used here has only one anchor (amine) group, so the different behaviour seen in the previous work may be associated with the twin-anchor polymer leading to bridging interactions between clay particles.
5. Summary and outlook
Aqueous suspensions of MMT were found to form dense aggregates at length scales of order 1 μm, whereas at larger length scales, of order 20 μm, at certain salt concentrations, there was evidence for further self-assembly into string-like aggregates, in dilute samples. Adsorption of amine terminated polymer surfactant onto the clay resulted in a doubling of the concentration for onset of gel formation, and a disappearance of the string-like structures. As also concluded for Laponite by Pignon et al. , the flow behaviour appears to be determined by the nature of the largest aggregate structures present in suspension. The tendency to form homogeneous aggregates consisting of randomly oriented particle at the micrometre scale presumably is the reason that a liquid crystalline (nematic) phase has not been observed in MMT suspensions.
Paineau et al. [46,56] found that among the smectite clay family, tetrahedrally charged clays (nontronite and beidellite, both with similar CEC to that of MMT) experience stronger repulsion than octahedrally charged ones (MMT and Laponite). Only the ones with strong repulsions undergo IN transitions, as observed previously [21,23,25]. Interestingly, electrolyte concentrations do not have much effect on the location of the IN transitions in systems where they do occur, for example, in suspensions of beidellite , fluorohectorite and fluortetrasilicic mica .
Further evidence of the crucial importance of the distribution of charges in and around clay particles for the structures the particles assemble into, comes from the recent work of Fartaria et al. . They studied Laponite particles, organically modified both on the faces and edges, and dispersed into styrene. While the particles appeared well dispersed, SAXS measurements revealed the particles self-assembled into two-dimensional aggregates resembling large sheets. A satisfactory model reproducing this structure required not only quadrupole interactions (representing the charge-sandwich structure of the mineral) but also dipolar interactions, representing a charge imbalance in the plane of a particle. The quadrupolar repulsive interaction prevents the platelets from stacking into small columns. The low-dielectric constant of the suspending medium will have enhanced the significance of any charge interactions compared to aqueous suspensions. Similarly, it was found previously that while MMT can be separated into individual clay sheets (exfoliated) in toluene using polymeric stabilizer, the resulting suspensions tend to form gels rather than a nematic phase .
Clay suspensions adopt complex structures across several length scales, and understanding this is key to controlling their macroscopic phase behaviour and flow properties.
Y.C. was financially supported by a Dorothy Hodgkin Postgraduate Award (Engineering and Physical Sciences Research Council grant EP/P504368/1), co-sponsored by AkzoNobel. Jeffamines were kindly provided by Huntsman. We would like to thank Dr Cheryl Flynn for her constructive suggestions on rheology data, and Diamond Light Source Ltd and the staff of I22 for the provision of SAXS beamtime.
One contribution of 14 to a Theo Murphy Meeting Issue ‘New frontiers in anisotropic fluid–particle composites’.
- © 2013 The Author(s) Published by the Royal Society. All rights reserved.