Principles of flow of kaolin and bentonite dispersions

Applied Clay Science, 4 (1989) 105-123

105

Elsevier Science Publishers B.V., Amsterdam - - Printed in The Netherlands

Principles of Flow of Kaolin and Bentonite Dispersions G. L A G A L Y

Institut fi~r Anorganische Chemie der Universit(tt Kiel, Olshausenstrasse 40/60, D-2300 Kiel (B.R.D.) (Received November 28, 1988; accepted February 10, 1989)

ABSTRACT

Lagaly, G., 1989. Principles of flow of kaolin and bentonite dispersions. Appl. Clay Sci., 4: 105123. Flow of kaolin and bentonite dispersions is decisively determined by edge ( + )/face ( - ) contacts (card-houses) in an acidic medium and face ( - )/face ( - ) contacts (band-like structures) in an alkaline medium. Formation of the different networks depends on p H and Ca/Na ratio. Calcium ions promote face ( - )/face ( - ) contacts and stabilizeband-like structures. In alkaline dispersions of homoionic sodium smectites and at low salt concentration, sodium ions cause disintegration of the particles into thinner lamellae and stacks of silicatelayers which, at low solid content (about < 5 % ), move independently under applied stress (Newtonian flow). These dispersions are sensitiveto the Ca/Na ratio.Some amounts of calcium ions link the lamellae and stacks of layers to form band-like networks, and the consistency of the dispersion increases considerably. Admixed crystalline or non-crystalline materials affect the flow of clay dispersions when they interact with the clay minerals. A n example is the influence of iron oxides. Organic compounds can stabilizeor destabilize networks, which is demonstrated for surface active agents and kaolinite.

INTRODUCTION

Clays are often defined as materials with particle sizes below 2 #m. Outstanding properties of these materials, in particular flow behaviour and plasticity, would be inconceivable if the particle size were the only decisive parameter. The properties of clays are brought about by the content of clay minerals. Besides the small size of the particles several other parameters determine the rheological properties: shape of the particles, layer charge, exchangeability of cations, structure of the particle edges and edge charge density. A further strong point is that the shape of smectite particles can change with environmental or experimental conditions and can never be considered as invariable. Ulrich Hofmann attributed plasticity of clays to aggregation of clay mineral 0169-1317/89/$03.50

© 1989 Elsevier Science Publishers B.V.

106

particles to a card-house structure. The particles are linked by edge/face contacts to a three-dimensional network throughout the whole system. The conception of the card-house model (cf. Fig. 5a) was the decisive step in understanding plasticity and flow properties of clay dispersions. I would like to emphasize the significant contributions of Ulrich Hofmann on understanding rheological behaviour of clays (Hofmann and Hausdorf, 1945; Hofmann, 1961, 1962, 1964). Ulrich Hofmann died in 1986 at an age of 83 years. RHEOLOGICAL MEASUREMENTS

A clay becomes plastic when it contains a certain amount of water. The Atterberg plastic limit is the lowest water content, expressed as a percentage by weight of oven-dried clay, at which a clay can be rolled into threads without breaking into pieces. At a certain higher water content the clay begins to flow when jarred slightly. This water content, again expressed as a percentage of oven-dried clay, is the Atterberg liquid limit. The difference between the liquid limit and plastic limit, the plasticity index, describes the range of water content in which a clay is plastic (Table I). The Atterberg limits have high significance in soil mechanics and for ceramic materials (Grim, 1962). The Atterberg limits respond sensitively to changes of the experimental conditions in some cases and remain invariant in other cases. The cause is that the particle concentration in a plastic clay is very high and the particles cannot assume equilibrium positions, and the system is held under geometrical constraints. Dilution required to bring the particles in equilibrium distances deTABLEI Atterberg limits (percentage by weight of oven-dried clay) (Mtiller-Vonmoos et al., 1985 ) Clay

Cation

Kaolin Georgia, CMS-KGa-2, Source Clay Miner. Rep., Univ. Missouri

Na Ca Na "~

69 74 33

31 31 24

38 43 9

Illite Bassin du Velay, Massif Central, France

Na Ca Na*~

76 93 54

29 32 29

47 61 25

431 190

48 50

383 140

Montmorillonite Arizona, CMS-SAz- 1, Source Clay Miner. Rep., Univ. Missouri

Na Ca

Liquid limit

Plastic limit

"~Sodium form, and 0.5 g Na4P207 • 10H20 added to 100 g oven-dried clay.

Plasticity index

107

pends on the clay mineral. Generally, the necessary amount of water is much larger for sodium montmorillonite dispersions than for sodium kaolinites (see below). Flow curves of such diluted dispersions respond sensitively to the different types of interactions between the clay mineral particles. Flow curves show the relation between shear stress z and rate of shear ~ and reveal the frictional resistance that a fluid (liquid or dispersion) offers to an applied shearing force. The principle of rheological measurements is explained by the model in Fig. 1. Consider a fluid between two plates. Keeping one plate fixed (reference plane), a tangential force is applied to the other plate so that the plate moves with a constant velocity. The shear stress z is the force applied to unit area of the fluid. In steady state (laminar flow) subsequent layers of the fluid move parallel to each other in the direction of the applied stress. A velocity gradient is established between both plates which is called rate of shear In the ideal case (Fig. 2, Newtonian liquid) the fluid begins to flow appreciably as soon as a shear stress is applied. The shear stress r is proportional to the rate of shear: z = ~/~.The proportionality constant is the shear viscosity (or, simply, viscosity). Very diluted clay dispersions exhibit Newtonian flow. At higher concentrations, z is no longer proportional to ~, and flow is plastic (Fig. 2 ). Plastic flow can exhibit a yield stress. The dispersion begins to yield to the applied stress only above a finite stress. Plasticity occurs when a particular network is formed throughout the system, which can resist to the mechanical deformation as long as the force remains below a critical value. If the flow curve is linear at higher rates of shear (Bingham flow), the extrapolated shear stress is called Bingham yield stress zB. In some cases and when the shear stress is measured at very low rates of shear, it may be difficult to decide whether the system yields to the applied stress at very low rates of shear ( < 10 s-1) (the flow curve approximates asymptotically the z-axis) or whether a finite yield stress exists at which yielding starts abruptly (cf. Fig. 10). In the first case increasing shear stresses bring about some deformations of the network which decrease the viscosity or

velocity -- force -"1

] shearing ptane II I

distance I [

II I Jreference plane

Fig. I. Laminar flow of a fluid between two plates: shear stress z = force/area of fluid,rate of shear = ~ (velocity)/distance.

108

/

"7

- newtonian flow

c

I:': s h e a r

stresslPa~

plastic flow without a yield stress with o yield stress

O

b

c

/ / - -

shear

stress--

=

thixotropic

antithixotropic

d

e

Fig. 2. Flow curves (~ = [(T) ) for Newtonian flow (a) and plastic flow without (b) and with (c) a definite yield stress; time-dependent phenomena: thixotropic (d) and antithixotropic (e) behaviour.

consistency of the system only slightly. At high shear stresses a more or less sudden break-down of the structure reduces the viscosity strongly. Curves that are concave to the T-axis (Fig. 2 ) indicate that the (differential) viscosity increases with the rate of shear. This is called shear thickening (dilatancy). Curves concave to the y-axis reveal shear thinning, that is, decreasing differential viscosity with increasing rate of shear. A system behaves thixotropic when the apparent viscosity decreases continuously with time under shear and is subsequently recovered when the flow is discontinued. A thixotropic system (clay dispersions, paints, drilling fluids, tomato catch-up ) begins to flow under stirring or vibrating and "solidifies" at rest. Thixotropy is detected by a hysteresis loop of the flow curves {Fig. 2d). An other experimental quantity, the limiting thixotropic volume, is not often determined currently but can be very informative. It is the volume of liquid that must be added to a certain amount of the solid to obtain optimal thixotropy (Hofmann et al., 1957; cf. Lagaly, 1986a). Thixotropy occurs when the fragments of the network which has been broken under shear, need some time to be linked again to a three-dimensional network. Thermal movement drives the particles and fragments into the contact positions. Shearing at low rates can also promote restoration of the network. The system then behaves antithixotropic (Fig. 2e). In clay dispersions

109

flow is often antithixotropicat low rates of shear and changes into thixotropic at higher rates (Brandenburg and Lagaly, 1988). KAOLIN

Kaolin particles carry exchangeable gegen ions (cations) on the external basal surfaces. Like 2/1 clay minerals the sign of charge at the crystal edges changes with pH (Fig. 3). The valency of the cations, monovalent (sodium ions) or divalent (calcium ions) and the charge density of the edges have a strong influence on the rheological properties. The next five sections report data of kaolinites with sodium as gegen ions. The effect of calcium ions is discussed in a subsequent section.

Effect of solid content Dispersions of clay can be sufficiently fluid up to high solid contents. Kaolin water mixtures containing as much as 70% kaolin are used for paper coating (Grim, 1962; Jepson, 1984). Generally, the viscosity increases abruptly above 65% solid content. The influence of the solid content on the flow of sodium kaolin dispersions up to 37% (wt/wt) is shown in Fig. 4a. Flow is plastic and thixotropy occurs above 20% solid content. The yield stress increases considerably between 33 and 37% kaolin.

®

® OH

pH Fig. 3. Variable charge density on the crystal edges of 2/1 clay minerals as pH increases: desorption of protons and ionization of silanol groups impart an increasing negative charge density to the edges,

110

" .,~ 2000 "l-

/

2 [

3 //

/

3 ///

wt O/o

37

1@0

10

20

,~ 2000

67 mmol CoCI.z/L

26.8

1000

30

50

T/Pa

100

i

150 ~"

200 t'/Pa

100 / 50 ,

10

F 2'0 Cs/mrno[.L_l ~0

Fig. 4. Flow of sodium kaolin dispersions (kaolin from Zettlitz, Sz~nt6 and Oilde, 1973): (a) as a function of content of sodium kaolin (in percent wt/wt); (b) 40% dispersion of sodium kaolin with different amounts of CaC12; (c) Bingham yield stress rB of 40% dispersions of sodium kaolin in the presence of increasing amounts ca of CaC12.

Influence of pH Increasing pH reduces considerably the viscosity and the yield stress (Rand and Melton, 1977; Yong and Ohtsubo, 1987). In an acidic medium the edges of the kaolinite particles are positively charged, and a card-house structure is built-up by edge ( + )/face ( - ) contacts (Fig. 5a). Recharging of the edges at increasing pH breaks down the card-house. The kaolin particles are dispersed into a colloidal distribution and, at low solid content, can move independently.

Dispersing agents Agents like soda, sodium phosphates (for instance Na4P207"10H20, sodium polyphosphate) and also water glass solutions are well-known to de-

111 o

edgeI face Kartenhaus

face ! face Banderstruktur

Fig. 5. Edge/face and face/face contacts that link the particles to card-house (a) or band-like (b) structures.

T So0. ~

C

Ag Chin(] Ctoy

÷ l g Na4 P2 0~.10H2 0

/

/ to{

. ~

/

// IIf

so

/{water #

//

too

i/' " "[/Po

150

Fig. 6. "Liquefaction"of kaolin (100 g China clayin 200 ml of water) by sodiumdiphosphateand restoration of the gel strength by addition of potassium chloride (by courtesy of Prof. Dr. Max Mtiller-Vonmoos,ETH ZUrich). crease the yield stress of kaolin dispersions. As shown in Fig. 6, even Newtonian flow can be attained. Dispersing agents act in different ways: (1) They provide sodium ions to the system, and calcium gegen ions are exchanged by sodium ions. The degree of exchange is enhanced by formation of sparingly soluble calcium salts (calcium carbonate and phosphate) or by soluble calcium complexes (with polyphosphate macro-ions). (2) Edges that are positively charged, can be recharged by adsorbed multivalent anions or macro-anions (scheme I). Recharging can also occur at sites where calcium ions have not been exchanged by sodium ions (scheme II). (3} Even if the charge density at the edges is slightly negative, an anion exchange with multivalent anions can increase the number of negative edge charges (scheme III).

112

O,o =--t

net charge

scheme

+

+

I

-

II

2-

I]I

(4) The agents can increase the pH of the dispersion to an alkaline value which produces negative edge charges, too. (5) Higher amounts of polyelectrolytes like polyphosphates can impart a considerable amount of steric stabilization to the system (Lagaly, 1986b). In a very simplified picture stabilization of the colloidal dispersions is achieved by envelopes (lyospheres) consisting of macromolecules, solvent molecules and gegen ions which impede direct contacts between the particles. Steric stabilization can also be produced by hydrophilic uncharged macromolecules. Thus, addition of dispersing agents produces or increases the repulsive potential between edges and edges or faces which breaks up the card-house and "liquefies" the kaolin water mixture. The mixture of 100 g China clay in 200 ml water is a pasty mass which, after recharging of the edges by addition of 1 g Na4P2 Or" 10H20 flows as a Newtonian fluid dispersion.

Effect of salts When salt is added to the "liquefied" kaolin-water mixture, a yield stress is again produced (Fig. 6). The sodium ions form diffuse ionic layers around the kaolinite particles and create an electrostatic repulsion between the particles (cf. Norrish, 1954; van Olphen, 1977; Lagaly and Frey, 1979; Lagaly, 1986b). The thickness of the diffuse ionic layers in salt free dispersions is considerable (some thousand A). The repulsive potential keeps the particles at large distances, and the colloidal distribution is stable. When salt is added, the thickness of the diffuse layer is compressed which increases steeply the repulsion at small distances. Above a critical salt concentration (critical coagulation concentration) the steep increase of the repulsive potential occurs at distances so small that the van-derWaals attraction overcomes the electrostatic repulsion, and the total interaction between the particles becomes attractive. At attractive potentials, the particles can build up a particulate structure throughout the system which enhances viscosity or consistency and establishes a distinct yield stress. This effect of salts is clearly seen in fig. 3. of Rand and Melton (1977) at pH > 7. In the stability range of card-house structures, NaC1 addition decreases the Bingham yield stress (Rand and Melton, 1977). The positive edges and negative basal planes are not in direct contact but are separated by their diffuse ionic layers (heterocoagulation; Usui, 1973; Oshima, 1974; Oshima et al., 1987). Because of the irregular shape of most of the particles, the number of direct

113

contacts between two opposite particles is small. As the double-layer thickness decreases with increasing salt concentration, the electrostatic attraction between the edges ( + ) and faces ( - ) becomes operative only at small distances. The particles must approach more closely until they "feel" the electrostatic attraction. As the less regular shapes of the particles restrict the number of contacts at close distances, the net-attraction decreases with increasing salt concentration, and the stability of the card-house is reduced. At contact points where the density of the charges of edges ( + ) and faces ( - ) is very different, the potential may change from attractive to repulsive (Oshima, 1974).

Types of aggregation In the presence of salt and in an alkaline medium clay mineral particles can aggregate in different ways: e d g e ( - ) / e d g e ( - ) , e d g e ( - ) / f a c e ( - ) and face ( - )/face ( - ). Near the point of zero charge of the edges the net-charge of the edges is very small. Relatively low salt concentrations produce attractive potentials between edges and edges or edges and faces. At certain conditions ofpH and salt concentrations edge ( - )/edge ( - ) and edge ( - )/face ( - ) contacts may then be formed. At a somewhat higher salt concentration, the potential between the surfaces also becomes attractive. The particles then aggregate face by face in a way that a linked structure throughout the system results (Vali and Bachmann, 1988). As discussed by Callaghan and Ottewill (1974), the resulting band-like structure (Fig. 5b) is more likely than a card-house structure with the small contact areas between edges and faces. The "B~indermodell" was first proposed by Weiss and Frank (1961). The nature of bank-like networks should be dependent on the position of the exchangeable cations at the external surfaces of kaolinite crystals (McBride, 1976) which can carry cations on the tetrahedral face only (Weiss and Russow, 1963) or on both external basal planes. In the last case kaolinite layers are inverted, or the external surfaces are modified by strongly adsorbed aluminasilicate species (Ferris and Jepson, 1975).

Effect of calcium ions On the basis of the classical DLVO theory colloidal dispersions are stable below the critical coagulation concentration which, for divalent cations, is in the range of 0.5-3 mmol'1-1 (Schulze-Hardy rule). However, even smaller amounts of calcium ions destabilize colloidal dispersions of clay mineral particles. Calcium kaolinite and also calcium 2/1 clay minerals can not be dispersed in water to form stable colloidal dispersions. The destabilizing effect of calcium ions is stronger than expected by the DLVO theory. The experimental observations are in good agreement with calculations of Kleijn and Oster (1982). The authors developed an electrostatic model of the

114

stability of tactoids (Fripiat et al., 1982) in which clay mineral platelets are stacked parallel at about 10-A separations. The calculations are based on the DLVO-theory but the gegen ions in the interlayer region between the particle surfaces are assumed to be in equilibrium with the bulk solution. The charge density in the interlayer region between two particle surfaces generally differs only slightly in magnitude from the surface charge density, but determines the Gibbs energy of electrostatic interactions. It follows from this model that for a typical smectite (layer charge ~0.30 eq./(Si,A1)40,o) the interaction between two particles and sodium as gegen ion is repulsive at NaC1 concentrations < 0.1 M but is attractive for calcium gegen ions at all calcium concentrations (10-:' M-1M). A further point must be considered. The DLVO-theory describes the tbrce between two approaching diffuse ionic layers. However, when two clay mineral particles approach to a separation of < 10 A, the double layers rearrange into a central layer of gegen ions so that interaction becomes attractive. For montmorillonites, a central layer of sodium gegen ions (quasi-crystalline structure, Lagaly et al., 1972) forms at NaC1 concentrations of > 0.25 M. All experimental results are in agreement with the assumption that calcium ions strongly promote formation of a central layer of gegen ions which does not break up into two opposite diffuse layers, even in pure water. The consequence is that calcium ions are enriched in the contact regions between two particles and make the contacts very stable. Again, face/face aggregation under formation of band-like structures will be preferential. The effect of calcium ions is clearly seen from the flow behaviour of kaolinite dispersions. Addition of CaSO4 to a homoionic kaolinite dispersion (9 wt%, Rand and Melton, 1977) increases the Bingham yield stress at alkaline pH. In a neutral or acidic medium, that is in the stability range of card-house structures with edge ( + )/face( - ) contacts, small amounts of calcium ions decrease the Bingham yield stress (Fig. 4c). They initiate face/face contacts as defects in the card-house structure and reduce its stability. Higher concentrations of calcium ions cause a transition of the card-house into a stable bandlike network with high yield stresses. This is in agreement with the earlier observation of Weiss and Frank ( 1961 ) that calcium kaolinite particles in contrast to sodium kaolinite are arranged in a "B~inderstruktur". With increasing amounts of CaC12 flow behaviour changes from thixotropic to antithixotropic (Fig. 4b). Higher concentrations of calcium ions promote particle-like aggregation of kaolinite crystals on the expense of band-like arrangements, a process which reduces the extent of thixotropy. Shearing counteracts this process, because it assists in forming band-like arrangements of thinner particles; the system becomes antithixotropic.

Effect of iron oxides Clays never consist of clay minerals solely. Admixed crystalline and noncrystalline materials can strongly affects the rheological behaviour. One ex-

115 ample; recently described by Yong and Ohtsubo (1987), illustrates the influence of amorphous hydrated iron oxides. These materials can obtain isoelectrical points at pH > 8 so that in a large range of pH-values heterocoagulation between positively charged iron oxides and negative faces of kaolinite particles can occur. The extent of these interactions depends on the way in which hydrated iron oxides and kaolinite particles are brought together. When both compounds collide in an alkaline medium (pH 9.5) before the dispersion is acidified, the Bingham yield stress exhibits a very sharp maximum at pH about 6. This is in the heterocoagulation region where a particular network can be stabilized consisting of kaolinite ( - ) and aggregated iron oxide particles ( + ). When kaolinite and iron oxide are brought together at low pH (pH 3.5), the kaolinite particles are covered by iron oxide and obtain a positive net-charge. Increasing pH reduces the positive edge charges so that some edge ( +_ )/face ( + ) contacts may be created. Still higher pH which renders the edges negative, should stabilize an edge ( - )/face ( + ) network. However, recharging of the iron oxides into negative particles starts, and networks are no longer formed. So, a shallow maximum of the Bingham yield stress is developed between pH 7 and pH 8.

Organic compounds Organic compounds, in particular polymers, can increase or decrease the stability of the networks. The influence of surface active agents is a typical example (Welzen et al., 1981). Cetyl trimethylammonium bromide (CTA) is added to a kaolinite dispersion held at pH 10. With increasing CTA concentration the Bingham yield value increases sharply to a maximum and then decreases again (Fig. 7a). The increase corresponds to an increasing amount of adsorbed CTA cations. The particles are linked by interpenetrating alkyl chains (Fig. 7c), which increases the Bingham yield value to a maximum. At higher CTA adsorption the surface is recharged and a repulsive potential is created between the double layers around the particles. At pH 3.3 the Bingham yield stress shows a first minimum at about 10 -~ mol CTA 1-1 and decreases sharply to zero above 10 -4 mol CTA 1-1. In an acidic medium the extrapolated shear stress is high because of the card-house with edge ( + )/face ( - ) contacts. As the CTA concentration increases, surfactant cations are adsorbed on the faces and compensate the negative surface charges. Many edge/face contacts are destroyed and the shear stress is reduced. At higher CTA concentrations the particles bristling with alkyl chains are held together by the interpenetrating alkyl chains (Fig. 7c) and the extrapolated shear stress is increased. The sharp decrease of the yield stress at still higher CTA concentrations indicates recharging of the faces by CTA cations which are adsorbed together with their gegen ions. In the presence of anionic surface active agents (sodium dodecylsulfate) the

116

~-~

100

~ ' ~ J ~

100

/

g

~ 3

H:3~

C13

20

pH=lO

~,

t

-5

log [ CcTA/mot L-~

-5

-1 J

log [cSDS/tool. E1 ]

, -1

C 0)

C .........

,

lx.

x~<-x- X-Xx-X- x'x" ..... 2 <

concentration CTA

concentrcttion

-

)

SDS

Fig. 7. Influenceof a cationic surface active agent (cetyl trimethylammoniumbromide, CTA, and sodium dodecylsulfate, SDS) on the Bingham yield stress of a 9%-dispersionof sodium kaolin, Monarch kaolin, Georgia (Welzen et al., 1981): (a) Bingham yield stress zB as a function of CTA concentration at two differentpH; (b) Bingham yield stress TBas a functionof SDS concentration at two different pH; (c) modes of interparticle interactions at different concentrations of CTA and SDS in acidic media. yield stress is very small at pH 10 because of the very low level of dodecylsulfate adsorption. At pH 3.3, the yield value caused by the card-house structure is constant up to about 10-3 mol dodecylsulfate l - ' (Fig. 7b). Adsorption of the anions on the positive edges then leads to a break-down of the card-house, and the yield value decreases steeply to zero. ATTERBERGLIMITS The type of gegen ion (Na + or Ca 2+ ) is of minor influence on the plastic limit of kaolinite (Table I). The liquid limit increases somewhat when sodium ions are replaced by calcium ions. The effect of sodium diphosphate is very pronounced; the liquid and plastic limits are reduced, and the plasticity index contracts to 9%. The illite-water system behaves almost like kaolinite-water. Because of the

117

•i •

°°°

•

Fig. 8. Disintegration of smectite crystals in diluteddispersionsin the presenceof sodiumgegen

ions. higher charge density of the illite surfaces the effect of sodium ions (to make the system more fluid) is stronger. Replacing sodium by calcium ions enhances the plastic limit slightly and the liquid limit considerably. Addition of sodium diphosphate again decreases the liquid limit but does not change the plastic limit, indicating that plasticity is decisively determined by band-like aggregation. Bentonites behave quite differently. Exchange of calcium ions by sodium ions reduces the plastic limit slightly but increases the liquid limit strongly, in extreme cases up to 700% (Grim, 1962). The extreme changes are caused by delamination of the crystals in the presence of sodium ions (Fig. 8). Under test conditions this process proceeds by no means to complete disintegration into single silicate layers; the crystals delaminate into thinner lamellae at some predetermined breaking points (Fig. 13 ). At the low water content of the plastic limit the particles cannot disintegrate and, as described for illite-water mixtures, sodium ions reduce the plastic limit only slightly. The presence of montmorillonite in kaolins, even in small amounts, serves to increase the liquid limits considerably. Processes have been developed to remove or modify expanding 2/1 clay minerals, for instance, in kaolins used as coating pigment in paper (Grim, 1962; Jepson, 1984). In ceramic masses small quantities of montmorillonite or expanding mixedlayer minerals can improve rheological properties (Kromer and Schiiller, 1973). Occasionally, an amount of 1-3% of bentonite is added to enhance substantially plasticity, green strength and dry strength of ceramic bodies. BENTONITES

Flow of dispersions The viscosity of bentonite dispersions is strongly pH-dependent (Brandenburg and Lagaly, 1988). The very sharp minimum for a dispersion of ho-

118

moionic sodium montmorillonite (from Wyoming) is seen in Fig. 9a. The strong decrease of the shear stress in an acidic medium reflects the decreasing stability of the card-house based on edge ( + )/face ( - ) contacts. At the minimum, the edge charge density is very low so that the interaction between edges and faces is purely van-der-Waals attraction. Since the thin montmorillonite par-ticles and lamellae are of irregular shape, formation of close contacts is restricted to some points of very narrow distances, and attraction is not very effective. Recharging of the edges at somewhat higher pH produces a repulsion between the particles which then move independently under applied stress. Flow is Newtonian as long as the solid content is low (below 5% montmorillonite) (Fig. 10a). The viscosity increase beyond the minimum reflects the promoted disintegration of the particles in an alkaline medium. At higher contents of montmorillonite the particles are subjected to an increasing geometrical constraint. T h e y can no longer occupy equilibrium positions and must assume some ordered arrays which reduce interparticle repulsion as much as possible. Work is then required to shift the particles out of these positions, and flow becomes plastic. The shear stress of the dispersion of pristine "sodium bentonite" from Wyoming which still contains calcium ions (molar ratio N a / C a ~ 6 . 5 ) changes

tonite

No,Ca- bentonite

U b i

~

i

7J

'

91 i

..pH Fig. 9. Shear stress r (at 7 = 94.5 s - ~) of 4%-dispersions in 0.01 M NaC1 at 20 ° C as a function of pH: (a) homoionic sodium montmorillonite (Wyoming), prepared from (b); (b) pristine "sodium" b e n t o n i t e (Wyoming) (molar ratio N a / C a ~ 6.5 ).

119 pH=8:7

Q

L,..5

100

loo 0.5

1.0

1.5

v /Po b

pH=Sh

2'0

~..0

85

lO0

0,5

10

-r/Pa

15

20

Fig. 10. Flow curves of 4%-dispersions in 0.01 M NaC1 at 20 ° C and at different pH: (a) homoionic sodium montmorillonite (Wyoming); (b) pristine "sodium" bentonite (Wyoming) (molar ratio Na/Ca ~ 6.5).

with pH in a very similar way as in the absence of calcium ions. However, the card-house of Na/Ca montmorillonite is destroyed at a lower pH than that of sodium montmorillonite. Calcium ions with their strong tendency of forming face/face contacts produce defects in the edge ( + )/face ( - ) network and promote its fragmentation. The card-house then breaks down at edge charge densities which, in the absence of calcium ions, can still stabilize the network. At pH 5.4 (Fig. 10b) the dispersion of N a / C a bentonite shows no definite yield stress and flow is slightly plastic with a very small hysteresis. Increasing pH causes the particles to disintegrate. In contrast to Newtonian flow of the dispersion of homoionic sodium montmorillonite at pH = 8.7, the plastic character of the flow curve is very pronounced and the (differential) viscosity is considerably enhanced. Flow is antithixotropic at very low rates of shear and thixotropic at higher rates. The calcium ions present in the dispersion form strong face/face contacts and stabilize an extended band-like structure

120

throughout the whole system which imparts considerable plasticity to the system. Generally, flow of bentonite dispersions is very sensitive to the Na/Ca ratio. Thus, properties of bentonites in practical applications can be improved by changing the Na/Ca ratio (Alther, 1986 ). However, it is very important to note that the rheological properties of dispersed bentonites depend on the cation originally present in the bentonite. That is, a calcium bentonite dispersed in a sodium salt solution exhibits rheological properties different from a dispersion of the bentonite in its sodium form in a calcium salt solution, even if the Na/ Ca ratio is the same (Brandenburg and Lagaly, in preparation). Soda activation

The well-known increase of viscosity and consistency of calcium bentonite dispersions after addition of soda is a direct consequence of the opposite effects of sodium and calcium ions. When soda is added to bentonite, the aqueous dispersion becomes alkaline and sodium ions are provided that exchange interlayer cations and initiate disintegration of the particles. A part of the calcium ions is precipitated as CaCO3 but other calcium ions are enriched in the contact regions and stabilize the contacts. Increasing soda addition increases the extent of disintegration of the particles into thin lamellae which are then linked by calcium ions. A resistant, stable three-dimensional network forms

~

ma× t h l × O t r o p r

7

0.5

I /

W"

• ....

.-'~.,. - 4--:= :':~ .....

?0

~0 ,00' 200' ~ soda

...... ---

e

addition

newl.on~an f l o w anlithixotropic

---- -

i

-

~

~.ome on

~

2~0

(meq/lOOg bentonite) plastic flow fhixofroplc

Fig. 11. Effect of soda addition to 2%-dispersions of different bentonites (arrows: maximum thixotropy).

121

and the dispersion obtains a considerable consistency. Optimal are amounts of 200-500 meq. soda/100 g bentonite (Fig. 11 ). Higher amounts of sodium ions reduce the consistency because they displace calcium ions from contact regions. The enhanced salt concentration, which compresses the diffuse ionic layers, also promotes formation of thicker particles on the expense of extended networks. For the same reason maximal thixotropy is observed at the maximum of the shear stresses (Fig. 11, arrows). For instance, a 2%-dispersion of Amory bentonite shows Newtonian flow after addition of 10 meq. soda/100 g (Fig. 12). A large hysteresis loop occurs at 500 meq. soda/100 g. (The shape of the upward-curve reveals that equilibrium was not attained during the measurement. ) Further addition of soda reduces thixotropy again. Viscosity increase at optimal soda addition differs from bentonite to bentonite. Important factors are particle structure and texture. High-resolution transmission electron micrographs reveal that certain montmorillonites (for instance, Upton, Wyoming) are composed of compact and continuous stacks

/

10(

"T

/

50

meq soda/ / lOOgbentonite /

10 ,

300

500

700

10C

7 .f.,~ 10

1000

5C

100

SIX) "c/roPe

Fig. 12. Flow of 2%-dispersions of Amory bentonite in the presence of differentamounts of soda (10, 500 and 1000 meq. soda/100 g bentonite).

122

Fig. 13. Influenceof particle architecture on rheologicalproperties: existenceof breaking points. of layers, whereas other montmorillonites (Polkville, Mississippi) contain disconnected, very thin and strongly disrupted stacks of layers (cf. figs. 5-8 in Vali and KSster, 1986). Dispersed in water, the natural calcium bentonites display very similar shear stresses ( z ~ 100-120 mPa at ~ 100 s-1 for 2%dispersions, Fig. 11) because the particles are aggregated and the special nature of the particles is of minor influence. W h e n sodium ions are added, the aggregates break up into the individual particles which further disintegrate to stacks of layers of different thicknesses. The true nature of the layer-stacking in the particles then becomes very important. Fig. 13 illustrates how particles with different texture disintegrate at predetermined breaking points. As a consequence the particles in dispersions of different montmorillonites will be disintegrated to different extents so t h a t the dispersions differ in the number, thickness and shape of the flow units and, therefore, in the rheological properties. This is one of the reasons why bentonites differ so widely from each other and dismay Standards Committees. REFERENCES Alther, G.R., 1986. The effect of the exchangeablecations on the physico-chemicalproperties of Wyomingbentonites. Appl. Clay Sci., 1: 273-284. Brandenburg, U. and Lagaly,G., 1988. Rheologicalproperties of sodium montmorillonitedispersions. Appl. Clay Sci., 3: 263-279. Callaghan, I.C. and Ottewill, R.H., 1974. Interparticle forces in montmorillonite gels. Faraday Disc. Chem. Soc., 57: 110-118. Ferris, A.P. and Jepson, W.B., 1975. The exchangecapacity of kaolinite and the preparation of homoionicclays.J. ColloidInterface Sci., 51: 245-259. Fripiat, J., Cases, J., Francois, M. and Letellier, M., 1982. Thermodynamicand microdynamic behavior of water in clay suspensions and gels.J. ColloidInterface Sci., 89: 378-400. Grim, R.E., 1962. AppliedClay Mineralogy.McGraw-Hill,New York, N.Y. Hofmann, U., 1961. Geheimnissedes Tons. Bet. Dtsch. Keram. Ges., 38: 201-240. Hofmann, U., 1962. Die Tonmineraleund die Plastizitiit des Tons. Keram. Z., 14: 14-19. Hofmann,U., 1964. Oberfl~ichenladungund Rheologieder Tonminerale. Bet. Dtsch. Keram. Ges., 41: 680-686.

123 Hofmann, U. and Hausdorf, A., 1945. Uber das Sedimentvolumen und die Quellung von Bentonit. Kolloid Z., 110: 1-17. Hofmann, U., Fahn, R. and Weiss, A., 1957. Thixotropie bei Kaolinit und innerkristallineQuellung bei Montmorillonit. Kolloid Z., 151: 97-115. Jepson, W.B., 1984. Kaolins: theirproperties and uses. Phil. Trans. R. Soc. London, A, 311: 411432. Kleijn, W.B. and Oster, I.D., 1982. A model of clay swelling and tactoid formation. Clays Clay Miner., 30: 383-390. Kromer, H. and Schiiller, K.-H., 1973. Eigenschaften von Kaolinen fiir die Keramik und als Ftillstoff. Ber. Dtsch. Keram. Ges., 50: 15-19, 39-41. Lagaly, G., Stange, H. and Weiss, A., 1972.0ber quasikristalline Strukturen bei der Flockung von Montmorilloniten und die Ausbildung diffuser Ionendoppelschichten in Nitrobenzol. Kolloid Z. Z. Polym., 250: 675-682. Lagaly, G., 1986a. Anorganische Systeme - - Tonmineraldispersionen. In: W.-M. Kulicke (Herausgeber), Zum Fliessverhalten von Stoffen und Stoffgemischen. Htithig und Wepf Verlag, Basel, pp. 147-186. Lagaly, G., 1986b. Colloids. In" Ullmann's Encyclopedia of Industrial Chemistry. VCH Verlagsgesellschaft, Weinheim, Vol. A 7, pp. 341-367. Lagaly, G. and Frey, E., 1979. Selective coagulation and mixed-layer formation from sodium smectite solutions. Proc. Int. Clay Conf. Oxford 1978. Elsevier, Amsterdam, pp. 131-140; and: Selective coagulation in mixed colloidal suspensions. J. Colloid Interface Sci., 70: 46-55. McBride, M.B., 1976. Origin and partition of exchange sites in kaolinite: an ESR-study. Clays Clay Miner., 24".88-92. Mtiller-Vonmoos, M., Honold, P. and Kahr, G., 1985. Das Scherverhalten reiner Tone. Mitt. IGB, ETH Ztirich, 128: 1-21. Norrish, K., 1954. The swelling of montmorillonite. Disc, Faraday Soc., 18: 120-134. Van Olphen, H., 1977. Clay Colloid Chemistry. Wiley, New York, N.Y. Oshima, H., 1974. Diffuse double layer interaction between two parallel plates with constant surface charge density in an electrolyte solution, II. The interaction between dissimilar plates. Colloid Polym. Sci., 252: 257-267. Oshima, H., Makino, K. and Kondo, T., 1987. Electrostatic interaction between two parallel plates with surface charge layers. J. Colloid Interface Sci., 116: 196-199. Rand, B. and Melton, I.E., 1977. Particle interactions in aqueous kaolinite suspensions, I.J. Colloid Interface Sci., 60: 308-320. Sz~intd, F. and Gilde, M., 1973. Rheological properties and thixotropic behaviour of kaolinite suspensions. Acta Univ. Szegedidnsis, Acta Phys. Chem., XIX (3): 291-297. Usui, S., 1973. Interaction of electrical double layers at constant surface charge. J. Colloid Interface Sci., 44: 107-113. Vali, H. and Kiister, H.M., 1986. Expanding behaviour, structural disorder, regular and random irregular interstratification of 2 : 1 layer-silicates studied by high-resolution images of transmission electron microscopy. Clay Miner., 21" 827-859. Vali, H. and Bachmann, L., 1988. Ultrastructure and flow behaviour of colloidal clay dispersions. J. Colloid Interface Sci. In press. Weiss, A. and Frank, R., 1961.0ber den Bau der Gertiste in thixotropen Gelen. Z. Naturforsch., 16: 141-142. Weiss, A. and Russow, J., 1963.0ber die Lage der austauschbaren Kationen bei Kaolinit. Proc. Int. Clay Conf. Stockholm, 1963. Pergamon Press, London, 1: 203-213. Welzen, J.T.A.M., Stein, H.N., Stevels, J.M. and Siskens, C.A.M., 1981. The influence of surfaceactive agents on kaolinite. J. Colloid Interface Sci., 81: 455-467. Young, R.N. and Ohteubo, M., 1987. Interparticle action and theology of kaolinite-amorphous iron hydroxide (ferrihydrite) complexes. Appl. Clay Sci., 2: 63-81.