Bergaya F. Handbook of Clay Science

Подождите немного. Документ загружается.

Dios Cancela, G., Huertas, F.J., Romero Taboada, E., Sanchez-Rasero, F., Hernandez La-

guna, A., 1997. Adsorption of water vapor by homoionic montmorillonites: heats of ad-

sorption and desorption. Journal of Colloid and Interface Science 185, 343–354.

Fridriksson, T., Bish, D.L., Bird, D.K., 2003a. Hydrogen-bonded water in laumontite I: X-ray

powder diffraction study of water site occupancy and structural changes in laumontite

during room-temperature isothermal hydration/dehydration. American Mineralogist 88,

277–287.

Fridriksson, T., Carey, J.W., Bish, D.L., Bird, D., Neuhoff, P.S., 2003b. Hydrogen-bonded

water in laumontite II: Experimental determination of site-speciﬁc thermodynamic prop-

erties of hydration of the W1 and W5 sites. American Mineralogist 88, 1060–1072.

Kranz, R.L., Bish, D.L., Blacic, J.D., 1989. Hydration and dehydration of zeolitic tuff from

Yucca Mountain, Nevada. Geophysical Research Letters 16, 1113–1116.

Norrish, K., 1972. Forces between clay particles. International Clay Conference Preprints 2,

3–14.

Weaver, C.E., Pollard, L.D., 1973. The Chemistry of Clay Minerals. Elsevier, Amsterdam.

Chapter 13.2: Parallels and Distinctions Between Clay Minerals and Zeolites1112

Handbook of Clay Science

Edited by F. Bergaya, B.K.G. Theng and G. Lagaly

Developments in Clay Science, Vol. 1

1113

Chapter 13.3

CEMENT HYDRATES

H. VAN DAMME

AND A. GMIRA

Ecole Supe

rieure de de Physique et de Chimie Industrielle (ESPCI-PPMD), F-75231

Paris Cedex, France

CRMD, CNRS-Universite

d’Orle

ans, F-45071 Orle

ans Cedex 2, France

Concrete is, by volume, the most widely used industrial material in the world: about

two tons per year and per person. An artiﬁcial rock that is one of the bases of developed

societies, concrete is basically a structural material (Neville, 1963). It owes its cohesion to

cement or, more exactly, to the reaction products of cement with water, called ‘hydrates’

(Taylor, 1997). Cement and its hydrates have special relationships with clays. The ﬁrst

one is that Portland cement—discovered in the 19th century by Aspdin and by far the

most common type of cement—is essentially the high-temperature reaction product of

limestone with clay. The second relationship, discussed in this short chapter, is that the

most important cement hydrates share two essential properties with smectites and lay-

ered double hydroxides (LDHs), namely, a layered structure and an electrically charged

surface. In calcium silicate hydrates, the layers are negatively charged as in smectites,

while in the calcium aluminate hydrates, they are positively charged as in LDHs.

Yet, there are also dramatic differences. Dipping a clayey rock in water may lead

to swelling and possibly to complete colloidal dispersion, whereas the same treat-

ment applied to ha rdened cement leaves it basically unchanged. Cement is even able

to set and harden under water. Current theories on the origin of this difference will

be brieﬂy reviewed.

13.3.1. PORTLAND CEMENT AND ITS HYDRATION

Portland cement is the ground form of clinker, the artiﬁcial rock obtained from

the non-equilibrium cooling of a sintered mass in the system CaO–SiO

–Al

, with

magnesium and iron impurities, starting from a mixture of clay and limestone as raw

material. Each grain is a polycrystal containing several mineral phases, the most

abundant being impure tricalcium silicate, Ca

SiO

(alite, C

S in short) and dical-

cium silicate, Ca

SiO

(belite, C

S in short) (Taylor, 1997).

The other phases, which

DOI: 10.1016/S1572-4352(05)01041-X

We will use the cementitious nomenclature throughout: C for CaO; S for SiO

; A for Al

; H for

O; and



S for SO

are liquid during the synthesis at 1450 1C, are a tricalcium aluminate, Ca

in short) and a calcium alumino ferrite, Ca

(celite, C

AF in short). Gyp-

sum, CaSO

2H

O, (C



in short) is added to ground clinker in or der to regulate

the reactivity of the aluminate phases.

When cement is mixed with water, it undergoes a dissolution reaction, generating

calcium, silicate and aluminate ions, among others, in the interstitial solution . Very

soon new products precipitate, the most important of which are calcium silicate

hydrate (CSH) and calcium hydroxide (portlandite, CH). The aluminate ions react

with calcium and sulphate ions to form ﬁrst calcium trisulphato aluminate hydrate

(ettringite) initially. When all the gypsum was consumed, calcium monosulphato

aluminate hydrate is formed as a result of the reaction of ettringite with residual

tricalcium aluminate. As the hydration reaction proceeds, more and more anhydrous

material is converted into hydrates. This leads to an overall decrease in porosity since

the molar volume of hydrates is much larger than that of the anhydrous phases

(more than twice in the case of tricalcium silicate (Powers, 1958)). The remaining

porosity, the residue of the incomplete ﬁlling of the initial intergranular porosity, is

referred to as the ‘capillary porosity’ (Fig. 13.3.1).

CSH

Portlandite

CAPILLARY

PORE

Fig. 13.3.1. SEM micrograph of a hardened cement paste showing urchin-like CSH, a port-

landite crystal and, in the middle, a residual capillary pore. Inset: well-crystallized hexagonal

portlandite crystals (courtesy Moranville).

Chapter 13.3: Cement Hydrates1114

The crystal habits developed by the hydrates are strongly dependent on hydration

conditions such as temperature, water/cement ratio and organic additives. Never-

theless, typical shapes may be recognized. For instance, CSH often grows as ﬁne

urchin needles around the parent cement particle ( Fig. 13.3.1). Portlandite crystals

may easily be recognized by their well-deﬁned hexagonal shape and large size (Fig.

13.3.1). Ettringite grows as stiff but fragile needles that may restrict the ability of the

fresh paste to ﬂow. Fi nally, the crystals of monosulphato aluminate grow as hex-

agonal platelets, like portlandite, but much smaller. All those shapes are well de-

veloped when crystal growth takes place in dilute conditions, but they may hardly be

recognizable in dense pastes.

13.3.2. STRUCTURE AND SURFACE PROPERTIES OF CEMENT

HYDRATES

The hydrates are minerals that give cohesion to hardened cement. All the hydrates

play a role in this cohesion, but in Portland cement, the most important is un-

doubtedly CSH. Unfortunately, the structure of this hydrate is still imperfectly re-

solved.

CSH is basically a non-stoichiometric compound (Taylor, 1997). The average Ca/

Si ratio in ordinary hardened cement paste is around 1.7. Local values ﬂuctuate

between 0.6 and 2 or more (Viehland et al., 1996; Zhang et al., 2000; Gatty et al.,

2001). Pure CSH may also be synthes ized directly from the oxides or hydroxides—

lime, silica and water—in dilute suspension and under mild thermal conditions. This

is the so-called pozzuolanic synthesis route, similar to what was in Roman times the

common way of preparing binders from natural lime and silica-rich volcanic ashes

(‘pozzuolanas’, from the name of the city of Pozzuoli, near Naples). The pozzuolanic

method provides greater ﬂexibility in the control of the Ca/Si ratio than the simple

mixing of cement and water. This is because the Ca/Si ratio in CSH is thermody-

namically controlled by the calcium ions activity in solution and this activity is much

easier to control in the pozzuolanic method. Thus, CSH with a Ca/Si ratio varyi ng

from 0.66 to 1.5 may be prepared in this way. To prepare CSH with higher and well-

controlled Ca/Si ratios, tricalcium silicate has to be hydrated in lime solutions su-

persaturated with respect to portlandite (Damidot et al., 1995).

Though CSH is generally poorly organized (hence the widespread use of the

expression ‘CSH gel’), it is widely recognized as having a layered crystal structure

similar to tobermorite (Fig. 13.3.2) or jennite, with a layer thickness in the nano-

meter range (Taylor, 1986; Henderson and Bailey , 1988; Richardson and Groves,

1992; Taylor, 1993; Richardson, 1999). This morphology is hardly recognizable

in real hardened cement paste but becomes evident from XRD and direct TEM

observations of pozzuolanic CSH (Fig. 13.3.2).

Tobermorite exists in two forms, with basal distances of 1.1 and 1.4 nm, respec-

tively. The difference corresponds to one layer of water molecules. Only the structure

13.3.2. Structure and Surface Properties of Cement Hydrates 1115

of the 1.1 nm form was resolved. Two models were proposed, that by Hamid (1981)

and a more recent one by Merlino et al. (2001). The two models share in common a

central double plane of calcium ions 6- or 7-coordinated by oxygen ions, to which

silicate chains are grafted on each side. These composite layers, stacked along the 0 0 1

direction, are separated by Ca

ions and water molecules.

The difference between Hamid’s model and that of Merlino et al. lies in the struc-

ture and bonding of the silicate chains. In Hamid’s (1981) structure the repeating units

along the chains are so-called ‘dreierketten’ (Liebau, 1985), with two pairing tetra-

hedra linked to the CaO polyhedra by sharing edges with them, and one bridging

tetrahedron pointing away from the CaO polyhedra sheet (Fig. 13.3.2). In perfect

tobermorite, with Ca=Si ¼ 0: 66, the chains are inﬁnite and run parallel to the b axis.

All the oxygen atoms involved in the dangling bonds of the silicate chain are pro-

tonated as hydroxyl groups, so that the structure is neutral. The interlayer space is

occupied only by water molecules. No chemical bonds exist between adjacent

layers. On the contrary, in the structure proposed by Merlino et al. (2001) the

layers are linked by bridging Si–O–Si bonds, generating cavities analogous to

those found in zeolites. Recent ab initio calculations showed that Hamid’s structure

is significantly more stable (8%) than that of Merlino et al. (Gmira, 2003; Gmira

et al., 2004).

Bridging SiO

tetrahedra

Paired SiO

tetrahedra

Ca octahedra

7.32 Å

Silicate dreierketten

chain

Fig. 13.3.2. Upper left: model of CSH at Ca=Si ¼ 0:66, with the central double layer of

calcium ions, connected on each side to chains of silica tetrahedra. Lower left: TEM micro-

graph of CSH prepared by reaction of lime with amorphous silica at Ca=Si ¼ 1(Gmira et al.,

2004); Lower right: HRTEM micrograph of the hydrated region in a real, dense, CSH (Gatty

et al., 2001). AM, NC and MSOR indicate amorphous, nanocrystalline and mesoscopically

ordered regions, respectively.

Chapter 13.3: Cement Hydrates1116

Due to its ill-organized structure, no direct structural determinations can be

performed on CSH. Nevertheless, high-resolution

Si,

H and

O NMR , X-ray

absorption spectroscopy, IR and Raman spectroscopy (Col ombet et al., 199 8 )

provided a great deal of information on the structure of the layers. Up to

Ca=Si ¼ 1:5, CSH may be considered as a defect tobermorite structure (Fig. 13.3.5).

A crucial point is that the solution in equilibrium with CSH is always at very

high pH. For Ca=Si ¼ 1, the pH is already beyond 10. In the low Ca/Si regime,

the structural evolution involves the progressive replacement of the protons of

the initially doubly protonat ed bridging tetrahedra by Ca

ions that go into the

hydrated interlayer space.

At Ca=Si ¼ 1, a ﬁrst-phase transition occurs and a new type of CSH forms with a

lower degree of polymerization of the chains. The main defects are missing bridging

tetrahedra. The higher the Ca/Si ratio the larger the number of missing bridging

tetrahedra and the shorter the average chain length. This length goes from average

pentameric units at Ca=Si ¼ 1 to dimeric units at Ca=Si ¼ 1:5. Ionization of the

hydroxyl groups of the dangling bonds, generated by this chain shortening, main-

tains the negative charge of the layers an d the number of Ca

ions in the interlayer

space. At Ca=Si ¼ 1:5, a second-phase transition seems to take place. The structure

of CSH in this high Ca=Sið41:5Þ regime remains very controversial. Three main

models were proposed. In the solid-solution model, the CSH is considered to be a

solid solution of CSH with Ca=Sio1:5 and portlandite, Ca(OH)

(Richardson and

Groves, 1992; Cong and Kirkpatrick, 1996). In the nanophase model, CSH is be-

lieved to be a nanophasic mixture of tobermorite- and jennite-related units, at

nanometer scale (Taylor, 1986, 1993; Viehland et al., 1996). In the third model, the

increase of Ca/Si above 1.5 is achieved by replacement of SiO

units by Ca(OH)

(Nonat and Lecoq, 1998).

As far as surface properties are concerned, the CSH layers should bear a mixture

of Si–OH and Si–O

–

groups. The proportion of Si–O

–

groups increases as the

Ca/Si ratio and the pH increase (the equilibrium pH of a Portland cement slurry is

close to 13). This is conﬁrmed by the decrease of the Si–OH stretching band at

3740 cm

–1

in the IR spectrum when the Ca/Si ratio increases (Yu et al., 1999 ). Thus ,

the CSH layers are intrinsically negatively charged colloids. However, due to the

strong afﬁnity of the Ca

ions for the Si–O

–

groups, charge reversal may occur.

Modelling of the titrat ion curves of CSH with lime indicates the existence of three

acid–base equilibria (Pointeau, 2000):

4SiOH þ H

O24SiO



þH

ðln K ¼7:6Þ

4SiOH þ Ca

2þ

þ H

O24SiOCa

þ H

ðln K ¼12:9Þ

4SiOH þ Ca

2þ

þ 3H

O24SiOCaOH þ 2H

ðln K ¼21:3Þ

Thus, as the Ca

concentration and pH increase, four types of surface groups

are successively formed: >SiOH, >SiO

–

, >SiOCa

and ﬁnally >SiO CaOH. The

13.3.2. Structure and Surface Properties of Cement Hydrates 1117

existence of CaOH sites at the highest Ca/Si ratios and pH is conﬁrmed by the

appearance of a band at 3600 cm

–1

in the IR spectrum (Yu et al., 1999).

Electrophoretic measurements show that the zeta potential of CSH increases

with the Ca/Si ratio (which is itself determined by the calcium concentration and

pH in the solution in equilibrium with the solid). It becomes positive as soon as the

Ca/Si ratio is larger than 0.8 (the equilibrium calcium ions concentration is then

2 mmol/L), indicating that the calcium ions compensating the layer charge are po-

tential-determining (Nachbaur et al., 1998; Terrasse-Viallis et al., 2001). At an equi-

librium calcium concentration of 20–30 mmol/L (reached in the interstitial solution

of ordinary Portland cement), corresponding to CSH wi th the highest Ca/Si ratios,

the zeta potential ranges from +10 to +20 mV. This shows that not all the surface

groups are in the neutral SiOCaOH form. A significant proportion of SiOCa

sites still exist.

Sulphate ions also play a particularly important role in the surface properties of

CSH. Addition of an alkaline sulphate to the equilibrium solution of CSH shifts the

point of zero charge to higher calcium concentrations and decreases the zeta poten-

tial. The latter may even become negative at low Ca/Si ratio (Nachbaur et al., 1998).

The other important hydrate in a mature cement paste is the tetracalcium mon-

osulphato aluminate hydrate or CmSAH, with a general formula [Ca

(OH)

]

(SO

) 15H

O. Like portlandite, CmSAH crystallizes in the form of thin hexagonal

platelets (Taylor, 1997). Its structure is that of an LDH (see Chapter13.1) consisting

of a calcium hydroxide sheet in which 1/3 of the Ca

ions are replaced by Al

ions, with charge-balancing sulphate anions and water molecules in the in-

terlayer space. In ordinary Portland cement, part of the Al

ions are replaced by

ions, as a result of isomorphous substitution in the raw clay mineral used for

clinker synthesis.

Due to substitution of Al

(or Fe

) for Ca

in their structures, aluminate

hydrates carry an intrinsic positive charge. To what extent this positive charge is

screened, or even reversed, by adsorbed anions such as sulphate or carbonate ions, is

not known.

13.3.3. THE MESO- AND MICRO-STRUCTURE OF CSH

The widespread use of the term ‘CSH gel’ indicates our inability to qualify the

way the individual layers and their aggregates ﬁll space. XRD techniques are of little

help as both the in-plane extension and stacking order of the layers are very limited.

High resolution transmission electron microscopy (HRTEM) shows very little local

order in ordinary cement at early stages; even in well-cured high-density samples

the coherence lengths remain small (Viehland et al., 1996; Zhang et al., 2000; Gatty

et al., 2001).

Chapter 13.3: Cement Hydrates1118

Direct observation by atomic force microscopy (AFM) of the growth of CSH on

smooth silica surfaces at the solid/lime solution interface reveals the formation of

discrete, elongated 5 nm thick, disk-shaped aggregates with a long axis of the order of

60 nm (Gaufﬁnet et al., 1998). Each aggregate consists of a few CSH layers. Globular

units of similar size (60 nm) are observed by bright ﬁeld TEM observation of

ultrathin sections of mature hydrated calcium silicate pastes (Maggion et al., 1996;

Van Damme, 2002). Although it is not known if these features are universal, the

observed size seems typical of locally compact aggregates of CSH layers.

The microstructure at larger length scales is also controversial. AFM observations

show that on silica substrates, large assemblies of the basic units form as CSH

nucleation and growth proceed (Gaufﬁnet et al., 2000). These assemblies grow late-

rally, parallel to the substrate surface, or axially, perpendicular to the surface.

Such detailed observations cannot be performed on real dense pastes, but numer-

ous models were proposed on the basis of indirect data such as surface area and

porosity determinations (Jennings et al., 1996). The most recent model (Jennings,

2000) assumes the existence of two types of aggregates, compact and loose. On the

other hand, small angle X-ray and neutron scattering (SAXS and SANS) (Craievich,

1987; Adenot et al., 1993; Winslow et al., 1995; Maggion et al., 1996), mercury

intrusion (Maggion et al., 1996), image analysis (Maggion et al., 1996) and NMR

relaxometry (Plassais, et al., 2005) provided evidence to indicate that at high degrees

of hydration the texture exhibits scale invariant (fractal) properties on scale lengths

from at least the nm to the mm scale. This does not contradict the existence of two

types of CSH aggregates. Disordered growth processes may lead either to surface or to

mass fractal structures, depending on the rate-limiting step and conﬁnement condi-

tions (Van Damme, 2002). A conjectural cartoon, inspired by the pioneering work of

Powers (1958) and TEM micrographs (Maggion et al., 1996), is shown in Fig. 13.3.3.

13.3.4. CSH AND SMECTITES

It will not escape readers of this book that CSH and smectites share a number of

features:

(i) elementary particles with a 2D structure and a layer morphology with a thick-

ness of the order of 1 nm. The lateral extension of the layers is more limited in

CSH than in natural smect ites but close to the layer size in synthetic smectite

(Laponite). As a result, the deformability of the layers, and more so of

the particles (stacked layers), in CSH is much less than in montmorillonite for

instance. Likewise, the surface of the layers, at the atomic scale, is much rougher

than the oxygen planes in clay minerals (Fig. 13.3.2). Inﬁnite chains

of dreierketten units are already very rough and their roughness may be fur-

ther ampliﬁed by defects that appear as the Ca/Si ratio increa ses (see Section

13.3.2). This would introduce a strong stacking disorder, accounting for the

quasi-amorphous character of CSH in hardened cement;

13.3.4. CSH and Smectites 1119

(ii) a negative charge and

(iii) a hydrated inter layer space.

At the same time, CSH and smectites differ as to the localization and density of

electric charge. In CSH, the charge is not due to ionic substitution in the structure

but to ionization of surface groups and to speciﬁc adsorption of calcium ions that,

as pointed out above, induce charge reversal in zeta potential measurements at high

Ca/Si ratio. It is also noteworthy that the calcium ions in CSH are essentially non-

exchangeable.

The surface charge density, s, of the layers, without adsorbed ions, is also vastly

different. No direct measurements were report ed but s may be estimated from the

structure of the reference miner al tobermorite. Assuming that all surface OH groups

are ionized, one obtains s ﬃ 3e



=nm

. This is about four times larger than in typical

montmorillonites. Interestingly, this value is of the same order of magnitude as the

surface charge density in LDHs and in the other important hydrate, CmSAH.

Fig. 13.3.3. Cartoon illustrating the possible structure of CSH at increasing magniﬁcations

from top to bottom, from the scale of the capillary voids to that of the elementary layers. The

last cartoon (bottom) illustrates the two length scales that have to be considered for under-

standing the cohesion of cement: the scale of the nanoparticles (light grey) where calcium ions

form inner sphere complexes with surface oxygen atoms, and the meso scale of the larger pores

(black areas) where fully hydrated and mobile calcium ions generate ion–ion correlation forces

(ﬁrst cartoon adapted from Powers, 1958).

Chapter 13.3: Cement Hydrates1120

13.3.5. COHESION VS. SWELLING

Bearing in mind these similarities and differences, how might we account for the

remarkable cohesion of CSH? The question may be made more general by asking

how charge density and the nature of the interlayer ions affect swelling and cohesion

of any charged layer ed material, including clay minerals and LDHs. We will only

consider the case of saturated systems, i.e., materials in which the void space is

totally ﬁlled with water and ions.

As in clay minerals, the problem should be considered at two different length

scales. One is the sub-nano scale of interlayer distances within the dense CSH par-

ticles and interparticle ‘contact’ points. The other is the meso scale of inter-wall

distances in the particle fabric (Fig. 13.3.3). A similar distinction is made in smectite/

water systems where the small-scale interlayer (‘crystalline’) swelling is usually dis-

tinguished from the larger scale ‘osmotic’ swelling. In thermodynamic terms, the

driving force for interlayer swelling is essentially the hydration enthalpy of the in-

terlayer ions and the surface oxygen atoms. The driving force for osmot ic swelling is

the entropy gain of the mobile hydrated ions when their conﬁnement space expands

(as in a perfect gas). Thus, the ideal conditions for swelling are small, very mobile,

fully hydrated and homogeneously distributed ions in the interlayer space. Any

factor that restricts the hydration and/or the mobility of the interlayer ions and/or

their homogeneous distribution in the interlayer space will also rest rict swelling.

Complementary to the driving forces for swelling, consideration must be given to

the attractive forces that need to be overcome (by the repulsive forces). Non -retarded

van der Waals interactions belong to this category. Their magnitude may be esti-

mated from the continuum theory and the Hamaker constants. A system composed

of two charged particles with like charges, separated by charge-balancing ions in

water, would also generate strongly attractive conﬁgurations due to purely electro-

static interactions (Kjellander and Marcelja, 1984, 1986; Ennis et al., 1996; Belloni

and Spalla, 1997; Gronbech-Jensen et al., 1997; Larsen and Grier, 1997). At inter-

layer distances in the nm range, these attractive forces may be considerably stronger

than the van der Waals interactions.

A clear example is the restricted swelling (in water) of Ca

-montmorillonite

as compared with the extensive swelling of the Na

-exchanged form. As Kjellander

et al. (1988a, 1988b) showed, the swelling behaviour of Ca

-montmorillonite is due

to the so-called ionic ‘correlation forces’ (see Chapter 5). In the Poisson–Boltzmann

treatment of two identically charged surfaces with an intervening electrolyte solu-

tion, the two halves of the cell are symmetrical and neutral. Thus, no electric ﬁeld is

induced by one cell onto the other. In a real system, the overwhelming majority of

instantaneous ionic conﬁgurations do not achieve this ideal situation (Fig. 13.3.4).

On average, the electric ﬁeld generated by one half of the layer system is still zero

in the other half, but for every conﬁguration there is a spatially varying ﬁeld.

The charges of the second half respond to this instantaneous ﬁeld by adopting

conﬁgurations of lower energy, different from the mean ﬁeld conﬁguration. These

13.3.5. Cohesion vs. Swelling 1121