Surface Chemistry and Rheology of Slurries of Kaolinite and Montmorillonite from Different Sources †

Zeta potential (ζ)-pH and yield stress (τy)-pH behaviour of a number of kaolinite and montmorillonite slurries, including CMS-sourced materials, were compared. The CMS KGa-2 and Crown kaolin with very similar elemental composition displayed almost identical ζ-pH and τy-pH behaviour. Both displayed a pHζ=0 at 3–4 where the maximum τy was located. This pHζ=0 was higher at higher ionic strength as the pH-dependent charge increased with ionic strength and the permanent structural charge being invariant. The other kaolinite slurries (Reidel and Unimin) with different composition showed different behaviour. The surface chemistry and rheological properties of CMS-sourced SWy-2 (Na-) and STx-1b (Ca-) montmorillonite and bentonite slurries were also compared. They all displayed a negative zeta potential that was insensitive to pH. STx-1b slurries required a much higher solid concentration to form a gel. Maximum τy occurred over a broad range of pH. This pH is ~5 for SWy-2, 8 for STx1b, <2 for bentonite and 12 for API bentonite. Their difference in clay mineral composition such as impurities and exchangeable cations were highlighted. The point of zero charge (pzc) of kaolinite and montmorillonite slurries obtained via Mular-Roberts pH-salt addition method did not correlate well with the pHζ=0 except for the KGa-2 and Crown kaolin.


Introduction
Kaolin mineral slurries from different sources or deposits display different rheological and surface chemical properties (Au P. and Leong Y., 2013;Au P. et al., 2014;Lagaly G., 1989;Rand B. and Melton I.E., 1977;Teh E. et al., 2009).Kaolin is a very important commodity mined all over the world.The ability to use its mineral and elemental composition to predict its slurry rheological behaviour from its surface properties would be extremely useful and beneficial when processing this clay mineral.However not all of the important factors responsible for the variation in rheological behaviour have been identified and their effects understood.Attempts have been made to identify these factors (Rand B. and Melton I.E., 1977;Teh E. et al., 2009).Factors such as Ca(II) or CaO content or Ca/Na ratio (Avadiar L. et al., 2014;Avadiar L. et al., 2015;Lagaly G., 1989) have been identified.The presence of a relatively small amount of smectite content can also have a very significant effect on the rheological behaviour of kaolinite slurries (Au P. and Leong Y., 2013).
Bentonite composed of mainly sodium montmorillonite (Na + Mt) is another important commercial clay mineral because of its many uses, for example in paper coating, catalysts, pharmaceutical products, drilling muds, as an impermeable slurry wall and nuclear waste storage barrier.Slurries of bentonite displayed highly complex time-dependent rheological properties.Na + Mt in water swelled considerably.Factors affecting its rheological behaviour are thus more numerous such as impurities, exchangeable cations and salt concentration (Chang W. and Leong Y., 2014;Goh R. et al., 2011;van Olphen H., 1955).However, despite the numerous studies, the knowledge available in the open literature is still quite confusing.Gels of Na + Mt from different sources displayed different rheological properties.According to Lagaly G. (1989), the important factors are particle structure and texture.Upon swelling different Na + Mt will disintegrate to different extents giving rise to stack-layered particles of different thickness, different particle concentration and different shape flow units producing gels.A better evaluation of the difference would be to compare the rheological properties over a wide pH range.Like kaolinite, the comparison should also include zeta potential-pH behaviour.Such a comparison would allow for a more effective evaluation of the factors responsible for the rheological variation.
Rheological and surface properties data from slurries prepared from standard or well-characterised Na + Mt and kaolinite are required for comparison and evaluation.Such clay minerals can be sourced from the Clay Mineral Society USA (CMS) and will have well-characterised compositional, physical and surface properties data (CMS, 2015).The surface chemistry and rheological property data obtained for these clay mineral slurries will form the baseline or benchmark data for other kaolin and montmorillonite slurries to compare in this search to identify the important factors and understand their effects.CMSsourced kaolin KGa-2 and Na + Mt SWy-2 and Ca 2+ Mt STx-1b were used in this study.
The complete characterization of the surface properties of the clay minerals may require the value of several components: point of zero charge (pzc) or isoelectric point (IEP) (Pradip et al., 2015), point of zero net proton charge, zeta potential-pH-ionic strength behaviour, charges (layer, tetrahedral, octahedral, unbalanced and extra Si) and cation exchanged capacity or CEC.For CMS-sourced clay minerals, most of these parameters are known and listed on their database (CMS, 2015).For correlation with rheological properties, normally only the zeta potential, including pzc, is required because the magnitude of the rheological parameters such as yield stress and viscosity, is governed by the nature and strength of the predominant interparticle force.Zeta potential is a measure of the strength of the interparticle repulsive force and is often used to define the state of the slurries; flocculated or dispersed.A high magnitude is normally associated with no or a very low yield stress.This is often not the case with some clay mineral slurries (Leong Y. et al. 2012;Au P. and Leong Y. 2013).Many kaolin slurries do obey the yield stress-DLVO model (Hunter R. and Nicol S., 1968;Teh E. et al., 2009;Au P. and Leong Y, 2013).
Among the many yield stress τ y -DLVO force or interaction energy models (Hunter R. and Nicol S., 1968;Leong Y. and Ong B., 2003;Teh E. et al., 2010) is one based on constant surface potential for interactions between spherical particles and is given by (Scales P., et al., 1998): The number of particles per unit area is scaled to ϕ s 2 /a 2 .The model predicts a linear relationship between τ y and zeta potential squared ξ 2 at a fixed ionic strength.The critical zeta potential ξ cri characterizes the point of transition from flocculated to dispersed state where τ y = 0 (Leong Y.K. and Ong B.C., 2003).Thus Eqn. 1 is reduced to give (2) ξ cri is proportional to the square root of the Hamaker constant A H and is independent of the particle size a and concentration of the dispersion.This equation has been used to determine the Hamaker constant of pristine oxide in water (Leong Y. and Ong B., 2003;Teh E. et al., 2010).Laxton P. and Berg J. (2006), however, observed a positive slope for the linear relationship between τ y and ξ 2 of kaolinite-bentonite composite slurries.They attributed this to positive-negative charge attraction between the clay platelets.In their study, the zeta potential was varied by ionic strength instead of pH.
Apart from zeta potential, there are several techniques for characterizing the pzc of a mineral powder.One such technique is the Mular-Roberts (MR) salt addition-pH method (Mular A. and Roberts R., 1966).This method has been shown to accurately determine the pzc of pure inorganic oxides such as, Fe 2 O 3 , Al 2 O 3 , SiO 2 and TiO 2 (Alvarez-Silva M. et al., 2010;Mular A.L. and Roberts R.B., 1966).Its accuracy with clay minerals such as serpentine and chlorite phyllosilicate minerals is still questionable (Alvarez-Silva M. et al., 2010).For a pristine inorganic oxide such as rutile TiO 2 , the pzc is independent of the ionic strength of an indifferent electrolyte (Hunter R., 2003).The MR method exploited this property to determine the pzc.At pzc, the pH of the mineral suspended in 0.001 M sodium chloride concentration will remain unchanged when the salt concentration is increased to 0.01 M. At other pH levels, a difference in the pH between these two ionic strength states will be present.With pure oxides, such as alumina and silica, the surface charges are all pH-dependent.In contrast, clay minerals contain permanent structural negative charges which are pH-independent (Bolland M. et al., 1980).
Alvarez-Silva M. et al. (2010) found that the MR method for pzc determination was suitable for serpentine but not for chlorite phyllosilicate mineral.In this study, we extended their work to other clay minerals and included rheological data to correlate with the surface chemistry data.In addition, we also aim to develop an understanding of the applications and limitations of the MR method for kaolinite and montmorillonite.

Materials and mineral characterisation
The CMS KGa-2 kaolin was sourced from a deposit located in Warren County, Georgia, USA.KGa-2 is a high defect or low crystallinity kaolin with a crystallinity index of only 0.16.It was composed of 96 % kaolinite (with 3 % anatase, 1 % crandallite + trace mica and/or illite) (Chipera S. and Bish D., 2001).It has a BET surface area of 23.5 ± 0.06 m 2 g -1 and a cation exchange capacity (CEC) of only 3.3 meq/100 g.The large BET area is an indication of its high fine particle content.The elemental compositions of KGa-2 and other kaolinites, expressed in oxide content, are contained in Table 1 (CMS, 2015).The air-floated Crown kaolin (sourced from a deposit in Georgia) was provided by Active-Minerals Inc (USA) who indicated a median particle size of 0.3 μm and the crystallinity index of ~0.5.It composed 96-100 % kaolin, with 0-2 % quartz and 0-2 % TiO 2 (as anatase).
KGa-2 displayed a point of zero charge (pzc) at pH 3.5 and a point of zero net proton charge (pznpc) at 5.4 (Schroth B.K. and Sposito G., 1997).Zeta potential characterization of this kaolinite slurry produced a similar pzc at pH 3-4 (Du J. et al., 2010).
The other clay minerals used in this study were Unimin kaolin (Unimin Corp.), Riedel kaolin (Sigma-Aldrich), SWy-2 Na + Mt and STx-1b Ca 2+ Mt (both from CMS), bentonite from USA (B3378, Sigma-Aldrich) and an API bentonite (Rheochem Ltd).This API bentonite is used in the formulation of drilling fluids.The BET surface area of Unimin and Riedel kaolin were 19.9 and 9.9 m 2 g -1 respectively (Avadiar L. et al., 2014).The Reidel displayed a very low TiO 2 content.XRD data showed the presence of quartz (< 5 %) in the Unimin and-of both quartz and mica/illite (both < 5 %) in the Reidel (thus the high K 2 O content).
Powder characterization included particle sizing and particle morphology examination.The particle size distribution was measured with a laser Malvern Mastersizer Microplus.The morphology of the particles was imaged with a Zeiss 1555 VP-FESEM scanning electron microscope (SEM).The median size of KGa-2, Unimin, Riedel kaolin, Swy-2 and bentonite were found to be 1.7, 3.26, 5.70, 2.09, and 2.77 μm respectively.The SEM images of the various clay mineral powders are shown in Fig. 1(a)-(f).
All four kaolin tested were poorly crystalline.Fig. 1(a) shows that the KGa-2 possessed a complex surface structure on its basal surface, including micro-pits, the high frequency occurrence of micro-islands, and ragged and broken edges.This was also observed in the poorly crystallised Unimin, Riedel and Crown kaolin.The overall particle shape of these high defect kaolinites should lead to higher aspect ratios (Żbik M. and Smart R., 1998).The kaolinite crystals in SEM micrographs appear as rigid particles.However the particles of Swy-2 Na + Mt and bentonite appeared as undulated flexible sheets, as seen in Figs.1(e) and 1(f).The difference in microstructure between kaolinite and smectite is due to the relatively strong hydrogen bonding between hydroxyl groups on the 1:1 kaolinite sheets (Żbik M. et al., 2010).

Rheological charaterization
The rheological parameter used was the yield stress, which was measured directly using a vane technique (Nguyen Q. and Boger D., 1983).Clay suspensions were prepared at high pHs by homogenising a mixture of clay powder and solution with a Branson Sonifier which break ups the particle aggregates by sonic vibration.Dilute suspensions were prepared in the same manner for zeta potential measurement.
Brookfield vane viscometers with different spring constants, LVDV-II + Pro and RVDV-II + Pro, were used.A range of four-blade vanes were used to cover yield stress values ranging from < 1 Pa to 1000 Pa.The suspension pH was varied by using a 10 M potassium hydroxide and a 6.61 M nitric acid solution in order to avoid excessive dilution of sample during pH change in the experiment.Yield stress measurement was conducted in the direction from high to low pH so as to avoid the dissolution of clay minerals and the precipitation of the hydrolysis products of metal ions.

Zeta potential and pzc determination
Zeta potential was measured using a Colloidal Dynamics ZetaProbe (Appel C. et al., 2003).The pzc or IEP usually corresponded to the pH of zero zeta potential.The pzc determination via the MR method of pH-salt addition measurement required preparation of a 2 wt% slurry in a 0.001 M NaCl solution at a specific pH ranging from 3 to 11.In the case of kaolinite, 1 g of material was placed in a 50 mL of a 0.001 M NaCl solution prepared and the initial pH (pH ini ) was measured.Dry NaCl solids was then added to increase the ionic strength from 0.001 to 0.1 M and the final pH (pH fin ) was recorded.When ΔpH (= pH fin -pH ini ) is zero, the pH fin is the pzc.The clay particle is positively charged when ΔpH is positive (Ndlovu B. et al., 2014).For smectite swelling clay, 0.5 g was added to a 50 ml 0.001 M NaCl solution.Measurement for swelling clay was performed 24 hours after preparation.Goh R. et al. (2011) reported that freshly prepared bentonite gels required about a day to reach surface chemical equilibrium.

Kaolinite slurries
The zeta potential-pH behaviour of KGa-2 and Crown kaolin slurries was compared.They displayed almost identical zeta potential-pH behaviour at an ionic strength of 0.1 M NaCl (Fig. 2).An isoelectric point (IEP) or pzc at pH 3-4 was observed.Both contained a very low amount of CaO and other mineral impurities such as quartz, mica and the like.
The pzc of Crown kaolin at lower ionic strength of 0.001 M NaCl occurred at a lower pH of 3.0 as shown in Fig. 2. Kaolin possesses both permanent structural (negative) charge due to isomorphic substitution and pH-dependent charge.Unlike pH-dependent charge, the amount of permanent negative charge is not dependent upon the ionic strength.For pristine oxides such as TiO 2 with only pH-dependent charges, its surface charge density at a given pH increased with ionic strength-except at the pzc (Hunter R., 2003).At the pzc, the amount of positive charges must be equal to the negative charges.With kaolin slurries, at a given pH (below pznpc) away from pznpc, the positive pH-dependent positive charge density increased with ionic strength.At low ionic strength, the pH must be further away from the pznpc for the kaolin particles to acquire sufficient positive charge to balance the permanent negative charge (Tombácz E. and Szekeres M., 2006).Thus for Crown kaolin, a lower pH is required to acquire a sufficient amount of pH-dependent positive charge to balance the permanent negative charge at low ionic strength.This is the likely explanation for the lower pzc.
The yield stress-pH behaviour shown in Fig. 3(a) for KGa-2 and 3(b) for Crown kaolin were also similar, namely, high yield stress at low pH and low yield stress at high pH.The maximum yield stress was observed to be located at or close to pzc or IEP.
The yield stress-solids volume fraction relationships shown in Fig. 4 at low pH of ~3 were almost identical.A power law model described this relationship.An exponent value of ~3 was obtained.A similar exponent value was observed at a higher pH of 7 for the Crown kaolin.The low CaO Reidel kaolin slurries also displayed a similar exponent value of 3.1 (Au P.I. et al., 2014).This exponent value was related to the fractal dimension of the microstructure in the slurries (Au P. and Leong Y., 2015;Au P.I. et al., 2015).Kranenburg C. (1994) derived the following relationship between yield stress τ y and ϕ s from scaling theory:   (1992) for creeping shear flow of fractal aggregates and gels.Using Eqn (3), the fractal dimension D f of these kaolin slurries was determined to have a value of 2.3.The predominant particle interaction configuration is faceface or stair-step (Au P. et al., 2014) with some face-face aggregates interacting via edge-face configurations (Gupta V. et al., 2011).
The cryo-SEM image of microstructure of KGa-2 slurries at the maximum yield stress with a fractal dimension of 2.3 is shown in Fig. 5.The image shows that dense aggregates formed predominantly by face-face and stairstep interactions participating in a mixture of interaction configurations-between aggregates such as edge-face, edge-edge and face-face.Many of the edge-face interactions involved a particle face resting on the edge of another particle at a very acute angle.
Fig. 6 shows the MR salt addition-pH result for KGa-2 and Crown kaolin.For KGa-2, the experiment was repeated with a fresh sample.The two sets of data fitted with a third order polynomimal.For both KGa-2 and Crown kaolin the ΔpH = 0 occurred at pH 4.2.This value is similar to the pH of zero zeta potential of 4 obtained for both at the same ionic strength of 0.1 M NaCl.It seemed that good agreement in value between the two methods was achieved.Different mixing methods-spatula mixing and sonication-were evaluated for their effect on the pzc value for Crown kaolin.No significant effect was observed.
The pzc of KGa-2 obtained here was in agreement with the value in the literature (Schroth B. and Sposito G., 1997;Du F. et al., 2010).Point of zero salt effect, point of zero net proton charge (pznpc) and point of zero net charge (pzc) of KGa-2 kaolin were reported to have a value of pH 2.8, 5.4 and ~4 respectively (Schroth B. and Sposito G., 1997).The good correlation between the maximum yield stress and IEP or pzc suggested that the surface chemistry-yield stress result can act as a good reference point for comparison by other kaolinites.
The low CaO Riedel kaolin also displayed similar yield stress-pH behaviour (Fig. 7), low yield stress at high pH and high yield stress at low pH.Complete dispersion (τ y = 0) of the slurries with 30 and 40 wt% solids was observed at pH 9.These slurries' maximum yield stress, τ ymax , was located at pH 3. Its zeta potential-pH behaviour, however, did not display a pzc or IEP at low pH (Fig. 8).The trend of a decreasing negative zeta potential with decreasing pH was similar to others such as KGa-2 and Crown kaolin.The absence of pzc is probably due to the presence of mica/illite and quartz, as its main mineral impurities.Both these minerals have a very low IEP of ~2.0.The MR method, however, gave a pzc value of 4.0 for this kaolin (Fig. 9).Note that not all low CaO kaolin slurries displayed complete dispersion at high pH (Teh E. et al., 2009).
Unimin kaolin has a relatively high CaO content.Its Fig. 5 The cyro-SEM image of 15 wt% KGa-2 slurries at the maximum yield stress of pH 4.2.
slurries displayed very different yield stress-pH behaviour (Fig. 7).The τ ymax was located at pH 7-9 where the negative zeta potential is relatively large (Fig. 8).The addition of CaSO 4 was found to increase the Bingham yield stress of homoionic kaolinite slurries at alkaline pH (Lagaly G., 1989).The decreasing negative zeta potential with decreasing pH was similar to all others-Reidel, Fluka and Sigma kaolin (Teh E. et al., 2009).The magnitude of the zeta potential was smaller for the Unimin.No pH of zero zeta potential was observed.However the result of the MR method gave a high pzc value of pH 8.8 (see Fig. 9), which coincidentally was the location of the τ ymax .
The value of the τ ymax for the Unimin was much larger than that of Reidel at the same solid loading.At 30 wt% solids, the τ ymax was 40 Pa for the Unimin compared to only 20 Pa for the Riedel.This is probably due to the smaller particle size of Unimin (3.26 μm) compared to Riedel (5.70 μm) which meant a higher particle concentration in the Unimin-which should result in more particles participating in attractive interaction forming a stronger network structure (Leong Y. et al., 1995).
From the elemental composition data in Table 1, it can be seen that all four kaolin-KGa-2, Crown, Reidel and Unimin-possessed similar iron content in terms of Fe 2 O 3 and FeO.So, Fe content does not account for the differences in the yield stress-pH, zeta potential-pH and MR pzc results.The Ti content was very similar for KGa-2, Crown and Unimin.But it was very low in Reidel, 10 times smaller in concentration, and yet its yield stress-pH behaviour is very similar to that of KGa-2 and Crown kaolin.In addition, this Reidel kaolin has a very high content of K (from mica/illite impurities), at least 10 times larger than the other three kaolin.Only Unimin slurries displayed very different yield stress-pH behaviour.The only element significantly different to the others is the Ca(II) content quoted in terms of CaO.Avadiar L. et al. (2012;2104;2105) have conducted extensive studies on the effect of Ca(II) on the yield stress-pH and zeta potential-pH of kaolin, silica and alumina particles.They found that the addition of Ca(II) produced similar yield stress-pH behaviour in low CaO kaolin and that it also depressed the magnitude of the negative zeta potential.Their studies were quite conclusive in showing that CaO was the cause of the difference in yield stress-pH and zeta potential-pH behaviour.
Ca(II) shifts the maximum yield stress and pH of zero zeta potential to a higher pH (Avadiar L. et al. 2015).The adsorption of calcium ions on kaolinite surface increased in the high pH region in the form of positively charged of calcium complexes and precipitate in solution to provide positive charges to the negatively charged kaolinite surfaces (Atesok G. et al., 1988;Heidmann I. et al., 2005).Apart from that, the complex surface structure of poorly crystalline kaolinite produces a higher concentration of hydroxyl groups on the basal surfaces which provide ad-  ditional sites for the adsorption of Ca 2+ ions.At low pH, Ca 2+ ions adsorption has reduced as a result of increasing ionic strength and the competition with the dissolved Al 3+ , Si 4+ , Fe 3+ and Cu 2+ ions (Au P. et al. 2014;Heidmann I. et al., 2005;Zhou Z. and Gunter W., 1992).Therefore, Ca 2+ is the cause of the shift in the location of the maximum yield stress and the increase in the yield strength of the kaolinite suspensions.
The excellent correlation between yield stress and zeta potential observed in KGa-2 and Crown kaolinite was not observed with Unimin.Although the zeta potential-pH behaviour does not explain the yield stress behaviour for the Unimin, it clearly demonstrated the effect of calcium content.The difference in magnitude of negative zeta potential between high Ca(II) Unimin and low Ca(II) Riedel kaolin was considerably larger at the high pH region.The smaller negative zeta potential of Unimin at high pH was attributed to the adsorption of calcium ions and its positively charged hydrolysis product (Avadiar L. et al., 2014;McFarlane A.J. et al., 2006).At low pH, and where calcium ions adsorption is not important, the zeta potential of both Riedel and Unimin kaolin was similar in value.This confirms the adsorption of calcium species affecting the properties and behaviour of kaolinites.
MR Salt-pH addition results in Fig. 9 showed the ΔpH = 0 was located at a particular pH (pzc for pure material) for both Riedel and Unimin kaolin.For Riedel kaolin, the ΔpH = 0 was located at a much lower pH of 4, which was close to the location of the maximum yield stress.The addition of 0.5 dwb% CaO to Reidel shifted ΔpH = 0 to a higher pH of 8.7, a value similar to that displayed by Unimin.This result showed that CaO was responsible for the high pH of ΔpH = 0.
Kaolin slurries with low CaO content such as Reidel, Sigma and Fluka, obeyed the yield stress-DLVO model (Au P. and Leong Y., 2013;Teh E. et al., 2009).High CaO content Unimin slurries were found to obey this model only after treatment with phosphate-based additives (Leong Y. et al., 2012;Shankar P. et al., 2010).The linear decrease of τ y with ξ 2 illustrated in Fig. 10 showed that the Crown kaolin slurries also obeyed the model.The ξ 2 cri which characterises the state of transition from flocculated to disperse state (Leong Y.K. and Ong B.C., 2003;Teh E.J. et al., 2010) is very high, 6000-7000 mV 2 .This does not mean its Hamaker constant in water is very high.Rather, positive edge-negative face interactions in kaolin slurries contributed to this high value.

Montmorillonite slurries
The complex rheological behaviour of Na + Mt slurries is well-known and has been the subject of numerous studies with surface chemistry effects being the main focus (Lagaly G., 1989;Liang H. et al., 2010;Packter A., 1956;Paineau E. et al., 2011;Ramos-Tejada M. et al., 2001;Sakairi N. et al., 2005).The surface chemistry parameters evaluated were pH (Kelessidis V. and Maglione R., 2008;Lagaly G., 1989;Tombácz E. and Szekeres M., 2004), salt type and concentration (Abend S. and Lagaly G., 2000;Abu-Jdayil B., 2011;Brandenburg U. and Lagaly G., 1988;Yildiz N. et al., 1999), and a range of adsorbed additives such as pyro-and poly-phosphates (Goh R. et al., 2011;Lagaly G., 1989;Yoon J. and El Mohtar C., 2015), polymers (Dolz M. et al., 2007), and surfactants (Luckham P. and Rossi S., 1999;Permien T. and Lagaly G., 1994).The interpretation of the results was complicated by the pronounced swelling and thixotropic behaviour of these slurries (Abend S. and Lagaly G., 2000;Galindo-Rosales F. and Rubio-Hernández F., 2006;Lagaly G., 1989;Lee C. et al., 2012;van Olphen H., 1955, Yoon J. andEl Mohtar C., 2013).Even the sequence of reagent addition such as KOH during gel preparation has a dramatic effect on the yield stress-pH behaviour (Au P. and Leong Y., 2013).Despite more 50 years of research on this material, there is still no consensus on the nature of the predominant interparticle forces operating in the Na + Mt gels.Van Olphen H. (1951) suggested that the space-filling gel was due to attractive interaction between the positively charged edge and the negative face forming a card-house structure and many others agreed (Brandenburg U. and Lagaly G., 1988;Lockhart N., 1980).Norrish K. (1954) and others (Callaghan I. and Ottewill R., 1974;Paineau E. et al., 2011) believed that the electrostatic double layer (repulsive) force was responsible.Reliable information on the effect of surface chemistry such as pH and ionic strength on surface forces which could be evaluated in terms of rheological properties such as yield stress, shear modulus and viscosity, may help to settle this issue.However this was not helped by its (negative) zeta potential being quite pH-insensitive (Callaghan I. and Ottewill R., 1974;Goh R. et al., 2011) while its rheological properties such as yield stress or intrinsic viscosity showing strong pH-dependence (Packter A., 1956;Goh R. et al., 2011).High concentrations (> 0.1 M) of common salts such as LiCl, NaCl, KCl and CsCl, were found to weaken the gel structure of the bentonite slurries which appeared to contradict the DLVO theory (Chang W. and Leong Y., 2014).Like the kaolinite slurries, evaluation of the trend of the yield stress-pH behaviour should be included in establishing differences among the Na + Mt slurries from difference sources rather than just comparing the rheological properties alone.Such evaluation should be accompanied by information on their zeta potential-pH behaviour.The evaluation of both the yield stress and zeta potential behaviour across a broad pH range was completed for bentonite and bentonite-kaolin composite slurries (Goh R. et al., 2011;Au P. and Leong Y., 2013).Past evaluations of rheological properties have been conducted at its natural pH, over a narrow pH range and at a small number of pH values.Packter A. (1956) evaluated the intrinsic viscosity of very dilute Na + Mt sols over a very narrow pH range from 7 to 11.5.Here we present yield stress and zeta potential data across a wide pH range from 2 to 13 for a range of montmorillonite slurries.These slurries contained a significant amount of other mineral impurities and different exchangeable cations.
The time-dependent behaviour can complicate the comparison of the rheological properties of Na + Mt gels from different sources.The ageing gels can have many rheological states.Comparison should be conducted at the same surface chemical and structural state (Chang W and Leong Y., 2014;Sehly K. et al., 2015).Ideally the rheological characterization should be conducted after the montmorillonite gels have attained surface chemical equilibrium and their flocculated network structure broken down to an equilibrium state.Doing this allowed the rheological data from different sources measured at this specific state which was easily attainable, to be quantitatively comparable.
The particle interaction configuration in the montmorillonite gel is crucial in understanding the relationship between surface forces and rheological behaviour.However, clear SEM/TEM images of distinct montmorillonite particles interacting are not possible to capture.According to Lagaly G. (2006) Na + Mt particles are strictly not true crystals and the particles are an assemblage of disordered and bent silicate layers that do not have the regular shape of real crystals.In Ca 2+ Mt particles, only a few silicate layers formed the coherent domains.These silicate layers are separated by two layers of water molecules.
The effect of mineral additives such as barite and kaolin on the ageing and rheological behaviour of bentonite slurries have been recently investigated (Yap J. et al., 2011;Au P. and Leong Y., 2015).Bentonite displayed a very large influence on the rheological and ageing be-haviour of the composite slurries.To identify mineral impurities with a significant impact on rheological properties, more montmorillonite from various sources need to be evaluated.This paper will present surface chemistry and rheological results of well-characterised CMS source montmorillonite (Swy-2 and STx-1b) in terms of mineralogy, impurities and elemental compositions, and exchangeable cations in addition to surface properties such as BET area, CEC, layer and structural charges and others.In addition, similar types of results for other bentonite slurries with different mineral impurities are presented for comparison.One of these is an API bentonite containing a very high content of mineral impurities and having the following mineralogy: 24.5 % quartz (SiO 2 ), 1.3 % cristobalite (SiO 2 ), 1.5 % calcite (CaCO 3 ), 16.3 % kaolinite, 16.0 % sepiolite (Mg 4 Si 6 O 15 (OH) 2 •6H 2 O) and 40.4 % smectite (amorphous).Table 2 lists its elemental composition which shows a relatively high content of iron compounds.
The yield stress-pH behaviour for (a) SWy-2 Na + Mt and (b) for bentonite and API bentonite slurries is shown in Fig. 11.A characteristic feature of these clay mineral slurries is the very high yield stress of several hundred Pa at very low solid loading of a few wt%.The shape of the yield stress-pH curve of both the SWy-2 and bentonite slurries differed slightly.The Swy-2 displayed a yield stress decreasing with pH.At low pH the high yield stress showed a plateauing trend.The yield stress of the bentonite slurries displayed a minimum at pH 8 and an increasing yield stress that showed no plateau trend at low pH.Packter A. (1956) observed similar behaviour with Wyoming Na + Mt sols with its intrinsic viscosity increasing with pH from pH 7 to 11.5.Similarly, Brandenburg U. and Lagaly G. ( 1988) observed a minimum shear stress at a fixed shear rate at pH ~ 8 for Wyoming bentonite and homoionic Na + Mt slurries.The yield stress-pH behaviour of the API bentonite slurry is very different.It displayed almost no yield stress in the low pH region, between 3 to 6, and a maximum yield stress at pH 12. Its zeta potential was more pH-sensitive as shown in Fig. 12.The mineral composition showed that the montmorillonite content was relatively low at 40 % and the kaolinite and sepiolite content being relatively high, at 16.3 and 16 % respectively.Sepiolite is a magnesium silicate smectite with a fibrous morphology (Aznar A. et al., 1992).It formed stable thixotropic gel in water and was used as a rheological modifier.It is a common ingredient of drilling muds because of its rheological behaviour being insensitive to salt concentration and it also imparts high temperature stability to the muds (Galan E., 1996).The dominant exchange cation in these clay minerals was Na + which resided in the interlayer of the clay particles.In presence of water molecules, hydration of the exchangeable Na + ions occurred in the interlayer, bringing about surface charge development, which caused swelling and delamination of the particles.The delamination of smectite layered particles increased the number of primary particles and the concentration of these particles participating in attractive interactions.This is the likely cause of the much higher yield stress observed (Leong Y. et al., 1995).
The zeta potential-pH behaviour of SWy-2 Na + Mt and bentonite slurries was very similar (Fig. 12).The insensitivity of the negative zeta potential to pH ranging from 2 to 12, was clearly displayed by these two clay minerals.This insensitivity is due to the pH-insensitive permanent negative charge accounting for 90-95 % of the total charges (Duc M. et al., 2005).There was no correlation between zeta potential and the yield stress for these clay mineral slurries.This gives the impression that the DLVO particle interaction theory is being violated.As can be seen in Fig. 12, the zeta potential-pH behaviour of STx-1b Ca 2+ Mont also showed insensitivity to pH.The magnitude of the negative zeta potential is much lower than that of SWy-2 Na + Mont slurries.Addition of Ca(II) ions does reduce the magnitude of the negative zeta potential of both smectite and kaolin slurries (McFarlane A. et al., 2006;Avadiar L. et al., 2014).
As Fig. 13 illustrates, the MR method for SWy-2 and bentonite resulted in ΔpH = 0 for both, at pH ~ 9.0.Both had a relatively high CaO content of 1.68 % (SWy-2) and 1.02 % (bentonite) as shown in Table 2.A similar pH value for ΔpH = 0 was observed for kaolin with a high content of CaO (Fig. 9).Smectite (2:1) clay minerals normally possessed a higher SiO 2 content (Table 2).As the pzc of SiO 2 is located at a very low pH of 2.0, the smectite particles should also have a low pzc value.According to Reymond J. and Kolenda F. (1999) the pzc of mixed alumina and silica oxide would progress towards a lower pH as the silica content increases.For this SWy-2 slurry, Tombácz E. and Szekeres M. (2004) found no common pzc at different salt contents or no pzse (pH of zero salt effect).Even so, the edge pzc was attributed a pH value of ~ 7 (Tombácz E. and Szekeres M., 2004;Durán J. et.al., 2000;Heath D. and Tadros T., 1983) Kosmulski M. ( 2011), however, attributed a pzc of pH 6.4-7.2 to Swy-2, which was obtained by potentiometric titration at different salt concentrations and a pzc of   (2004).With Na + Mt and bentonite suspensions, there is no obvious correlation between yield stress and zeta potential.Similarly, there is no correlation between the maximum yield stress and MR "pzc".This lack of correlation between surface properties and rheological properties is not uncommon with thixotropic swelling clay.This may be due to the rheological properties being determined by the swelling behaviour.The very high yield stress at low solid concentrations suggested a very high concentration of particles being released by the clay swelling participating in attractive particle interactions forming the very strong gel structure.Many studies have determined the particle interaction configurations (EE, FF and EF) of these smectite particles in water to be a function of pH, salt concentration and Ca(II) concentration (Stawinski J. et al., 1990;Żbik M. et al., 2008).However, clear SEM or TEM images of delaminated smectite particles interactions have not been available.
The trend of an increasing yield stress with decreasing pH of SWy-2 and bentonite is a reflection of a strengthening gel structure.This enhancement could be due to a stronger attractive interaction between particles or a stronger repulsion of the overlapping double layers.The enhancement at low pH was accompanied by 3 times the increase in the conductivity for the SWy-2 slurries.This meant a thinner double layer at this pH.The double layer repulsion responsible for the caging effect Norrish K., 1954;Callaghan I. and Ottewill R., 1974;Paineau E. et al., 2011) should be weaker and thus cannot be responsible for the yield stress enhancement.A stronger particle-particle attraction can come about by the following means: i) the same particle interaction configuration but a stronger van der Waals or negative-positive charge attraction or both, ii) a change to a stronger particle-particle interaction configuration-face-face to edge-face interactions; or iii) more particles being release by the swelling and delamination process, increasing the number concentration of attractive interactions at lower pH.
The yield stress at high pH is due to the smectite particle-particle attraction being mediated by other positively charged particles and hydrolysis products of cations such as Mg(II) and Ca(II).These smectite particles are negatively charged since 90-95 % of the total charges being the permanent structural negative charge.A portion of Ca(II) content was locked in the less soluble minerals such as gypsum and plagioclase.At low acidic pH, these cations became hydrated and remained in solution, acting as an indifferent electrolyte.The pH-dependent charge density of the smectite particles increased at the same time.This could have changed the particle interaction configuration to a predominately EF interaction at low pH (van Olphen H., 1977;Heath D. and Tadros T., 1983).
The STx-1b montmorillonite suspension also displayed a pH insensitive zeta potential behaviour (Fig. 12).The magnitude of the negative zeta potential was much smaller.The predominant exchangeable cation of STx-1b was calcium (CMS, 2015).In contrast the main exchangeable cations of Swy-2 were sodium and calcium.A high proportion of these cations were the exchangeable interlayer cations.STx-1b can therefore be regarded as the Ca-form montmorillonite.The rheological behaviour of Ca 2+ Mt was known to be very different to the Na + Mt (van Olphen, H., 1955;1957).The layered particle structure of Ca 2+ Mt remained intact and the particles retained their aggregated behaviour (Lagaly, G., 1989;van Olphen H., 1957).
The yield stress-pH behaviour of the STx-1b Ca 2+ Mt suspension is shown in Fig. 14.A much higher loading of solids was required to achieve a gel or yield stress material.The solid loading required was at least 20 wt%, ~ 4 times more concentrated than the lowest yield stress gel of SWy-2 Na + Mt.The pH and magnitude of the τ ymax increased with solid loading.The τ ymax was 52, 190, 399 and 716 Pa located at pH 7.6, 8.3, 9.7 and 10.1 and for solid loadings of 20, 25, 30 and 40 wt% respectively.Unimin kaolin slurries with a relatively high CaO content also displayed a high pH of maximum yield stress of ~9.The yield stress of STx-1b slurries was smaller at low and high pH.The value was as low as zero for the 20 wt% slurries at pH 10.5 and ~17 Pa at pH 11 for the 25 wt% slurries.At pH 4 the yield stress was about 25 Pa for the 20 wt% slurry and 76 Pa for the 25 wt% slurry.This marked variation of the yield stress with pH bore no correlation with zeta potential-pH behaviour.Again the DLVO theory appeared not to explain the yield stress behaviour despite its lack of thixotropy.The higher solid loading required to form a gel is consistent with the earlier findings of H. van Fig. 14 Yield stress-pH behaviour of STx-1b (Ca 2+ Mt) slurries.Olphen (1955;1957) research on Ca 2+ Mt slurries.
The maximum yield stress-solids volume fraction relationships are shown in Fig. 15 for the montmorillonite slurries.The value of the slope is 2.2 for SWy-2 and 3.8 for STx-1b.Hence the fractal dimension D f of Na + Mt slurries is 2.1 and that of Ca 2+ Mt slurries is 2.5.The microstructure of STx-1b is thus more compact and more sphere-like in shape.A Cryo-SEM image of the STx-1b microstructure at the maximum yield stress prepared by plunged freezing of a slurry droplet is shown in Fig. 16.Connection of compacted domains in the microstructure was clearly seen.An image with such distinct particles was not available for SWy-2 slurries-a reflection of the delaminated silicate layers being in the nano-size dimension.For Unimin kaolin with more soluble Ca(II) minerals, the value of the slope was 3.6 to 3.9 (Au P. et al., 2014) similar to that obtained for the Ca 2+ Mt slurries.It was found that 25 % of the CaO in the Unimin kaolin were soluble at pH 4.

Conclusion
All kaolinite slurries with a low CaO content displayed similar yield stress-pH behaviour: high yield stress at low pH and low yield stress at high pH.High CaO content kaolinite slurries displayed maximum yield stress at high pH of ~9.Many of the mineral impurities at the level in the as-received kaolinites played a small or insignificant role in determining this yield stress-pH behaviour.It was shown here that the difference in the trend of the yield stress-pH result could be used to identify factors in the kaolinite responsible for this difference.
For high purity Crown kaolinite, the pH of zero zeta potential or pzc showed a slight dependence on ionic strength.The pH-independent permanent structural charge required a lower pH for the clay mineral to acquire sufficient pH-dependent positive charge to neutralise it at lower ionic strength.
This study reviewed that the Mular-Roberts salt addition-pH method was not suitable for determining the point of zero charge (pzc) of clay minerals.This is particularly true for clay minerals containing a high content of basic impurities such as CaO.The addition of a small amount of CaO (0.5 %) to Reidel kaolin increased its pzc from 4 to 8. All clay minerals such as kaolin, bentonite and Na-montmorillonite, with high content of CaO, ranging from 0.5 to 1.7 %, displayed ΔpH sign reversal ("pzc") at pH 8-9.
For the two low CaO Crown and KGa-2 kaolin suspensions, the maximum yield stress is located at or near the point of zero zeta potential or zero charge of pH ~4.In addition to yield stress-pH behaviour both kaolinite slurries displayed similar zeta potential-pH behaviour and an identical maximum yield stress-solid volume fraction relationship.No point of zero zeta potential was detected for another low CaO Riedel kaolin slurry.The values of the power law exponent representing the yield stress-volume fraction relationship were similar: ~3 for all three low CaO kaolin slurries.In contrast high CaO Unimin kaolin displayed a maximum yield stress at pH ~9 and no point of zero zeta potential.
The negative zeta potential of SWy-2 Na + Mt, STx-1b Ca 2+ Mt and bentonite slurries was quite pH-insensitive.No clear correlation between the yield stress and zeta potential was observed for these slurries.Maximum yield stress was detected at a low pH < 3 for SWy-2 and bentonite and at a high pH > 9 for the STx-2b slurries.The swelling property of the SWy-2 and bentonite controlled its rheological behaviour to a greater extent than surface chemistry such as pH.The difficulty of imaging Na + Mt particles via SEM or TEM suggests the swelling-liberated particles are nano-size in dimension.The high concentration of these nanoparticles can explain the demonstration of high yield stress at very low solid loading and the

Fig. 2
Fig.2Zeta potential-pH behaviour of 3.76 wt% KGa-2 and Crown kaolin slurries (pzc of 3.4) and the effect of ionic strength on Crown kaolin slurries.

Fig. 4
Fig. 4 The yield stress-solid volume fraction relationship of KGa-2 and Crown kaolin dispersions.

Fig. 10
Fig. 10The yield stress-DLVO model fit to the yield stresszeta potential data of Crown kaolin.

Fig. 11
Fig. 11The yield stress-pH behaviour of various a) montmorillonite and b) bentonite slurries.

Table 1
Elemental composition of kaolinite from different sources expressed in oxide content (wt%).

Table 2
Elemental composition of montmorillonite from different sources in terms of oxide content (wt%)