Surface and Intercalation Chemistry of Polycarboxylate Copolymers inCementitious Systems

Claire Giraudeau, Jean-Baptiste d’Espinose de Lacaillerie,w and Zied Souguir

Physico-chimie des Polymeres et des Milieux Disperses, UMR 7615 CNRS, ESPCI ParisTech, 75005 Paris, France

Andre Nonat

Institut Carnot de Bourgogne, UMR 5209 CNRS Universite de Bourgogne, BP 47 870, 21078 Dijon, France

Robert J. Flatt

Sika Technology AG, Corporate Research, Tuffenwies, 8048 Zurich, Switzerland

The Ca–Al-layered double hydroxide, the so-called AFm phase,is a product of cement hydration. It is shown that the interactionof this phase with anionic polycarboxylate ether (PCE)-baseddispersant polymers is not a simple adsorption but a more com-plex intercalation phenomenon leading to the transient seques-tration of the PCE within the AFm crystallites. As a result, partof the PCE is immobilized, forming a layered organo-mineralcomposite, and does not play its role of a dispersing agent. Thisarticle presents, along general considerations on the links be-tween cement chemistry and rheology, a detailed investigation ofthe formation, structure, and stability of a pure poly(methacry-late-g-PEO)/hydrocalumite composite obtained by coprecipita-tion. The predictions of scaling laws derived from models ofconformation of comb copolymers in solution were testedagainst small-angle X-ray diffraction, transfer of populationsby double-resonance nuclear magnetic resonance, and small-an-gle neutron scattering experimental results. A model of adsorbedpolymers in a configuration of a flexible chain of hemisphericcores is proposed and appears to be compatible with the ob-served interlayer spacings in the range of several nanometers.Finally, these phases are shown to persist for several hours in thepresence of sulfate ions.

I. Introduction

THE success of cementitious systems (mortars, concrete, etc.)as construction materials is due to the unique succession of

flow and then of hardening during their hydration period. Thus,a strong solid material can be easily cast into place. Any mod-ification of the formulation of these materials must retain these

fundamental properties. Recently, high-performance concreteshave been developed by building on the idea that the macroscopicmechanical strength can be increased by reducing the water tocement ratio of the paste. However, reducing the water contentobviously increases the solid ratio of the suspension and, if noth-ing is done, can only result in a loss of fluidity (see Sidebar 1). Forthis reason, the cement industry has developed polymericdispersants (also called ‘‘water-reducing agents’’).1,2 The mosteffective of them are commonly named ‘‘superplasticizers.’’Among them, the polycarboxylate ethers (PCEs) are actuallygrafted polyelectrolytes. Their negative polycarboxylate back-bone chains interact with the positively charged cement parti-cles, while the PEO graft side chains create a steric repulsion,leading to the dispersion of cement particles (see Sidebar 2).3

Actually, the effect of PCE adjunction to cementitious materialsis not fully robust, and rheological aberrations are occasionallyobserved in the field.4 In particular, the ability of the cementpaste to flow at a given PCE dosage depends strongly on thecalcium aluminate content of the clinker (Fig. 1).5 To solve thispractical problem, diverse fundamental aspects of organo-min-eral interactions at the molecular level must be considered. Oneof them is the accidental formation of an organo-aluminatecomposite within the cement paste.6,7

At the beginning of cement hydration, tricalcium alumin-ate, C3A in the cement industry notation system, is the mostreactive phase (see Sidebar 3). A product of its hydration is theso-called AFm phase. It is proposed here that the interactionof this phase with anionic polymers is not a simple adsorptionbut a more complex intercalation phenomenon leading to thetransient sequestration of the PCE within the AFm crystal-lites. As a result, part of the PCE is immobilized and forms alayered organo-mineral composite and does not play its role asa dispersal agent (Sidebar 4, Fig. 2).

This paper presents a detailed investigation on the formation,structure, and stability of such an organo-mineral phase. It con-tains four main parts (Sections II–V) accompanied by six side-bars providing complementary information. Following anoverview of intercalation chemistry in the context of cement

Feature

D. J. Green—contributing editor

wAuthor to whom correspondence should be addressed. e-mail: jean-baptiste.despinose@espci.fr

Manuscript No. 26536. Received July 14, 2009; approved September 11, 2009.

Journal

J. Am. Ceram. Soc., 92 [11] 2471–2488 (2009)

DOI: 10.1111/j.1551-2916.2009.03413.x

r 2009 The American Ceramic Society

chemistry, a series of results are presented that demonstrate theformation of an intercalated organo-mineral phase. Then amodel derived from the conformation of comb copolymers insolution is proposed for the structure of these phases. Finally,the last section examines the stability of these phases in thepresence of sulfate ions. These results are of major importanceto establish the relevance of this phase formation to the actualcementitious systems.

II. Intercalation Chemistry and Cement Hydration

Beside calcium silicates, cement clinker contains a significantamount of calcium aluminates. These calcium aluminates hy-drate significantly faster than the silicates and form differentcrystalline phases depending on sulfate ion availability.

In the absence of sulfates, a tetracalcium aluminate hydrate(4CaO �Al2O3 � nH2O, n5 7–19) and a tricalcium aluminate hy-drate (3CaO �Al2O3 � 6H2O) are formed. The tricalcium alum-inate hydrate is the cubic mineral katoite while the tricalciumaluminate hydrate is the hexagonal mineral hydrocalumite.

In the presence of sulfates, a monosulfoaluminate hydrate(SO2 � 4CaO �Al2O3 � nH2O) and a trisulfoaluminate hydrate(3SO3 � 6CaO �Al2O3 � 32H2O) are formed. The trisulfoalumin-ate hydrate is the mineral ettringite, called AFt, while the mono-sulfoaluminate hydrate is called monosulfate.

Hydrocalumite and monosulfate are based on the same struc-tural unit made of [Ca2Al(OH)6]

1 hydrated layers (Fig. 3). They

differ only by the nature of the interlamellar anion: OH� forhydrocalumite and SO4

2� for the monosulfoaluminate. In fact,these anions are exchangeable. They can be replaced by CO3

2� toform hemi- or mono-carboaluminate, or Cl� to form the Friedelsalt for example. This group of minerals is called AFm.

Actually, these AFm cement minerals (hydrocalumite,monosulfoaluminate, carboaluminates, Friedel’s salt) belongto an even larger family well known to clay and colloid chem-ists: layered double hydroxides (LDHs) with the structuralformula [MIIMx

III(OH)6]1 � x/nAn� � yH2O (see Sidebar 5,

Figs. 4 and 5).55

When hydrating a cement paste in the presence of a super-plasticizer polyelectrolyte, we encounter a situation that is in factvery similar to the coprecipitation LDH intercalation method.As C3A dissolves, Ca(II) and Al(III) ions are released into thewater. When the superplasticizer is added to the water before-hand (the so-called immediate addition), a templated formationof hydrocalumite around the polyelectrolyte thus becomes a realpossibility that may have been overlooked. Indeed, in sucha scheme that bypasses the difficult delamination step, it isconceivable that the polyelectrolyte could end up spontaneouslyintercalated within the hydrocalumite layer in the cementpaste. This has been suggested to occur for polysulfonatesuperplasticizers in the thesis of V. Fernon more than a decadeago.6,7 Sometime later, this phenomenon was proposed to be amore general issue also affecting PCE-based superplasticizers.24

More recently, Plank and colleagues have shown that hydra-tion of C3A in the presence of polycarboxylate-g-PEO indeed

Sidebar 1. Rheology of Cementitious Materials

Cementitious materials come in all kinds of forms, ranging from highly fluid injection grouts that must be able to penetratesmall cracks, to roller compacted concretes that are so stiff that they require a Roller Compressor to pack them into place. Themost broadly used of these materials, however, is concrete, which may come in a stiff form that requires vibration for placementor in a self-levelling form that requires only minimal placing work.

Such concrete is also an archetype of a multiscale material in that the finest mineral additions such as nano-silica may rangefrom 10 to 100 nm, while the largest aggregates can be a couple of centimeters large. In between this broad size range, there iscement that is in the range of 1–100 mm and sand from about 0.1 to 3 mm.

The behavior of all these components depends very much on their size, as the forces affecting particulate suspensions such asvan der Waals, electrostatics, viscous drag, inertia, and gravity scale differently with the particle sizes.8 Equally important is thevolume fraction of these components9,10 and its relation to the maximum packing of each class of particles.11 In these materials,the interparticle forces among the finer components play a very important role in terms of rheology. In addition to cement, thisconcerns any of the possible additional ingredients such as fly-ash, slag, limestone filler, silica-fume, etc. For these particles,attractive van der Waals and electrostatic forces lead to agglomeration, which, at the macroscopic scale, translates into thedevelopment of yield stress.12 In this case, not only are the volume fraction and maximum packing of the fines important butalso the particle size distribution and the magnitude of the attractive interparticle forces.13

In the vast majority of these cementitious systems, dispersants are used to reduce or eliminate these interparticle forces (seeSidebar 2). This has a direct impact on yield stress, which is an extremely important property for the placement of thesematerials.12,14 Indeed, it is because of yield stress that ordinary concrete does not flow freely and must most often be vibratedrather than cast into place. Self-compacting behavior is, in contrast, obtained when the fine particles are dispersed by reducingor eliminating yield stress with the type of admixtures studied in this paper. Moreover, it is important to point out that yieldstress largely dictates the shape of cementitious materials once they stop flowing (the so-called stoppage phenomenon). Indeed,this occurs at shear rates approaching zero when yield stress is the rheological property of most crucial importance.14

It is also certainly for this reason that simple tests linked to stoppage have found such a broad relevance in practice. Forexample, just about each and every job site using concrete uses a slump test to characterize the workability of this material. Thisconsists in filling a cone (30 cm high, 10 cm top diameter, and 20 cm lower diameter), lifting it and measuring the decline inheight. Alternatively, for particularly fluid materials, it is the spread diameter that is measured. While various studies hadpointed to the link of this test to yield stress,11,12 it is only recently that a clear relationship could be establishedquantitatively.15,16

Because yield stress develops due to the interaction between fine particles, cement pastes are often used to examine the effectsof chemical admixtures rather than working directly on concrete, which is much more labor intensive. For this, researchersworking on admixtures often adopt extremely simple tests such as the flow test,17,18 which provides a direct access to yield stressand, consequently, to the degree of dispersion.19 More interestingly, in the frame of this paper, effects such as superplasticizerstructure,20,21 addition mode, sulfate content,22,23 etc., can be assessed and related to changes in interparticle forces.20,23

Variations in the degree of dispersion can therefore be tracked with such tests. Specifically, in the context of this paper, theformation of an organo-aluminate phase would reduce the amount of polymer available for dispersion and reduce the flow.Because the formation of such phases is expected to take place in the first seconds when water contacts the cement, measuringflow properties by varying the addition time of the polymer provides important clues as to the propensity of a polymer to beconsumed by the formation of these phases. For example, a polymer expected to intercalate extensively should yield a muchbetter flow (broader spread) when added shortly after the water as opposed to including it in the mixing water.24

2472 Journal of the American Ceramic Society—Giraudeau et al. Vol. 92, No. 11

Sidebar 2. Superplasticizers and Dispersion Mechanisms

Chemical families of superplasticizersThe introduction of dispersants in concrete was accidental. Carbon black had been added to change the color of the center lineof three-lane on-going highways. Because of this, the concrete had poor workability and a dispersant was introduced. Thehardened concrete showed properties that indicated that the cement had also been affected by the dispersant25 (and probablymeant that the water content had been reduced too).

Since then, efficient and cheap products based on such polymers have become commercially available. They are referred toeither as superplasticizers or as high-range water reducers. They must not only improve workability at low water cement ratiosand maintain it over sufficient time for proper placing but also avoid the development of undesired secondary effects such as setretardation, segregation, and excessive air entrainment.4,26 Schematically, we can classify the base polymers used today into fivefamilies that are briefly described below.

2.1. LignosulfonatesLignosulfonates were one of the first dispersants used in concrete. They are obtained from a byproduct of the paper industry,and as such, they have a broad molar mass distribution. A good review of the complex structure of these molecules has recentlybeen given by Zhor.27 For this paper, only an idealized structure is shown (Table I).

Owing to their bulky structure and their unavoidable variations in composition due to their natural origin, it is not surprisingthat lignosulfonates have low performances. This is why they are usually referred to as water reducers rather than high-rangewater reducers. On the other hand, due to their low cost, they are still largely used in many applications.

2.2. Sulfonated naphtalene formaldehyde polycondensateSulfonated naphthalene formaldehyde polycondensates (SNFC) were introduced in the 1960s. Since then, they have been largelyused and studied. Their usual representation is the poly-b-naphthalene sulfonate condensate form in Table I, although thesynthesis can lead to more complex structures.28

2.3. Sulfonated melamine formaldehyde polycondensates (SMFCs)In terms of performance, SMFCs illustrated in Table I are similar to SNFCs and are also not always obtained as linear chains.28

In the future, it is expected that legislations concerning formaldehyde will limit geographically the use of such products muchmore than SNFC due to their lower stability.

2.4. Vinyl copolymersIn the 1980s, products based on linear vinyl copolymers were introduced that typically gave higher fluidity and improved fluidityretention. An example of such a polymer is given in Table I.

2.5. PolycarboxylatesPolycarboxylate comb-copolymers are synthetic molecules that are not linear and offer many variations in structures, so thatsubstantially higher performances can be achieved. In particular, these polymers should be noted for their larger capacity ofwater reduction and extended time during which the material remains workable. This is why they greatly contributed to thegrowth of self-compacting concrete, of which they are now an essential ingredient.

Typical variations in the structure of these polymers concern the lengths of the backbone and side chains, the number of sidechains, and of course the nature of the monomers making up either the backbone or the side chains.

Dispersion mechanismsSuperplasticizers adsorb at the surface of particles and impart a repulsive interparticle force, thus reducing or eliminating theadhesion between particles in close proximity.8,24,29 The attractive forces between the cement grains and possible other powdersin the mix like silica fume, fly ash, and slag particles may originate from van derWaals,30 electrostatic forces ion correlation,31 orfrom surface charge inhomogeneities.

For many years, electrostatic repulsion was considered as the primary dispersion mechanism, following a milestone paper byDaimon and Roy.32,33 Interestingly, their conclusion was questioned in a discussion paper by Banfill,34 who stated that ‘‘theeffect of zeta potential on rheology is not proved and steric stabilization may be equally important.’’ The basic argument here isthat to prevent particles from coming into close proximity and compromising flow, electrostatics alone could not suffice incementitious systems where the ionic strength is high (�0.1M). However, it was not until the advent of polycarboxylates manyyears later that this concept gained acceptance.3,35 Indeed, these molecules have a more pronounced steric layer due to theirnonadsorbing side chains that extend from the surface in solution as illustrated in Fig. 11(a). Basic studies could demonstratesome of the expected features of these dispersants, and their conformation at interfaces have been clarified as explained inSidebar 6.36,37 It must nevertheless be stated that despite the absence of nonadsorbing side chains in polymers of generations1–3, steric hindrance probably also plays an important role, this time, however, in combination with electrostaticrepulsion.38

Finally, we note that in order for a steric dispersant to be efficient, it must first be adsorbed. Studies on inert model systemsfirst pointed out that competitive adsorption with other anions could be problematic for polyacrylates due to a reduction in theiradsorption energy as a result of side chain grafting.39 This was later confirmed in a very elegant way in cement systems byYamada and Hanehara,22 who showed that high sulfate concentrations prevented their polycarboxylate from adsorbing. Apossible implication is that in cements where the sulfate concentration is initially high and then declines, the dispersion degreeshould increase accordingly. Macroscopically, this would translate by what is referred to as delayed fluidification oroverfluidification. In fact, this was observed to increase with the degree of grafting,21,40 which points again to the role thatadsorption energy plays in this process.23,39 Sulfates also clearly affect superplasticizer action through the hydration chemistry inearly ages as discussed in the main part of this paper.

November 2009 Polycarboxylate Intercalation in Cementitious Systems 2473

results in the formation of an intercalated composite.78,79

This result is of practical importance as the existence andstability of such a composite is likely to affect the rheologicalproperties of the cement paste. It would elegantly explainthe dependence of the fluidizing effect of the PCEs on the C3Acontent of cement.

We therefore conducted a study of the formation, structure,and stability of poly(methacrylate-g-PEO)/hydrocalumite com-

posites. The hydration of cement in general and of C3A in par-ticular leads to the formation of a complex mixture of phases,some of which are transient. Thus, in order to isolate thepoly(methacrylate-g-PEO)/hydrocalumite composites, weworked not from hydration but from precipitation products.We coprecipitated the hydrocalumite in the presence of purifiedpolymers with a polydispersity index of about 2 and selected totypify PCEs’ behavior (Fig. 6 and Table III).

Table I. Representation of the Main Dispersants Used in Concrete

Generation Water reduction (%) Type of polymer

1 5–10 Lignosulfonates

O

H CO

CO

O

OH H

SO M

HOn

2 10–20 SNFC

SO Nan

SMFC

N N

NNH NH O

NH

SO Mn

3 10–25 Vinyl copolmyers

OC

NH

SO Na

COONa NO

n

(a)

OC

NH

SO Na

COONa OCOCH

n

(b)

4 20–40 Polycarboxylates

R

CH2 CH2

R

C

CO2Na

C

O Y1−x x

CH2

CHR

O

CH3

n

Y = N, OR = H, CH3

Along with the generation of products that these polymers can be associated with, a typical range of water reductions achievable without significant retardation is given.

2474 Journal of the American Ceramic Society—Giraudeau et al. Vol. 92, No. 11

III. Experimental Proof of the Existence ofPCE–Hydrocalumite Intercalation Composite

(1) Materials and Synthesis Method

(A) PCE Polymers: Polymers (courtesy of Dr. I. Sch-rober, Sika Technology AG, Zurich, Switzerland) were purifiedby ultrafiltration. The MWCO of the membrane were 5000 Da.The molecular weight distribution was evaluated by size exclu-sion chromatography. Their characteristics are presented inFig. 6 and Table III.

(B) Synthesis: In a 1 L reactor, under an inert atmo-sphere maintained by a continuous Ar flow, 0.89 g of CaO (15.9mmol) was introduced in 800 mL of ultrapure water. CaO wasdissolved and the temperature was decreased to 51C. This tem-perature was selected to stabilize the hydrocalumite phase at thecost of the cubic-phase katoite. The necessary amount of poly-mer solution was then added to achieve a final concentration of2.1 g/L (3.7 and 6.8 mequiv COO� for polymers B and D, re-spectively). Then, 200 mL of a 20 mM solution of NaAlO2 (4mmol) was introduced drop by drop into the reactor, leading tothe formation of a precipitate. The suspension was matured un-der reaction conditions during 10 days. The solid and the liquidphases were then separated by centrifugation and washed with10 mL of ultrapure water (resistivity 18.2 MO � cm, organic car-bon 3 ppb, Milli-Q Corporation, Bedford, MA). This centrifu-gation and washing process was repeated three times and thesample was finally dried by freeze drying (one night,�501C, 0.35mbar). The solid samples were stored at 51C under argon.

(2) Characterization

(A) Chemical Analysis: The evolution of the polymerconcentrations in solution was followed by total organic car-bon analysis, and the elemental Al and Ca concentrations insolution were measured by ICP-AES after filtering out the solidthrough a 0.3 mm filter. The amounts fixed by the precipitatewere inferred by difference.

(B) Powder X-ray Diffraction (XRD): The mineralphases were first identified by standard XRD using a PhilipsPW 1700 X-ray diffractometer with CuKa radiation, a graphitemonochromator, and y–2y geometry, operated at 40 kV, 40 mA,and at a scanning rate of 0.251/min. The dried samples wereplaced on glass plates. Subsequently, the low-angle (0.51–101)analyses were performed on a specially designed apparatus(LPS, Orsay, France) in capillary tubes (0.5 mm diameter)with samples rehydrated in a water-saturated atmosphere for 3

days. CuKa radiation was also used and the diffractogram wasobtained from the radius average of the diffraction image plate.

(C) TEM: Images were obtained by Christopher Plum-mer at the Laboratory of Composite and Polymer Technologyat EPFL in Lausanne (Switzerland). The instrument was a Phi-lips CM20 TEM with a LaB6 filament in the clear field, at avoltage of 200 kV. The sample under examination was the driedpolymer B composite; the powder was dispersed in epoxy resinand sliced into thin sections from the hardened epoxy block witha microtome.

(D) SEM: Experiments were performed on powders, inthe secondary electrons mode. The instrument used is a Hitachi3600N with a W filament. Powders were dispersed in acetoneand a drop of the suspension was placed on a carbon wafer andair dried. This sample was coated with gold (1 kV, 12 mA, 2min) and introduced in the microscope.

(3) Results and Discussion

Well-crystallized LDH can be easily identified by conventionalpowder XRD thanks to their strong 00l basal reflections. Thevalue of the Miller index l varies depending on the stacking or-der of the particular LDH phase under consideration. In theabsence of polymer, the precipitate had an X-ray diffractogramthat was typical of an oriented hydroxyl hydrocalumite with abasal distance of 1.04 nm (Fig. 7(a)). The sample was partlycarbonated though, probably during centrifugation and/or sam-ple handling, as evidenced by the appearance of additional basalreflections corresponding to the basal distance of 0.82 nm typicalof hemicarbonate (JCPDS 041-0221).

When the same coprecipitation was performed in the presenceof polymer, XRD showed that the samples were much less crys-tallized (Fig. 7(b) and (c)). However, the precipitate structurewas still based on hydrocalumite because characteristic in-planereflections of hydrocalumite were present between 301 and 401and between 501 and 601 (JCPDS 042-0487). These reflectionsform an indication of the formation of a Ca/Al LDH struc-ture.80 The basal reflection was not observable anymore with aconventional XRD setup. However, measurements performedusing an apparatus specially designed to collect small-angle datarevealed that the basal reflections had actually shifted to muchlower angles corresponding to a basal spacing of 8–13 nm(Fig. 8). This increase of the basal spacing could be visualizedby TEM where the propping apart of the layers upon polymerintercalation is further evidenced (Fig. 9). Subtracting the layerthickness of 0.48 nm from the basal spacing,81 the interlamellarspace increased to about 7.5 and 12.3 nm. This was proof thatthe precipitated solid was actually an intercalation product ofthe polymer between the HDL layers.

As the driving force for the composite formation is expectedto be the charge compensation of the mineral layers by thepolyelectrolyte, an important point to examine is the chargebalance of the composite according to its chemical analysis. De-termination of the total organic carbon content of the solutionafter reaction showed that 0.23 and 0.73 COO� were fixed perAl for the composites obtained from polymers B and D, respec-tively. Because each Al results in one positive charge due to thenominal layer composition ([Ca2Al(OH)6]

1), this meant that asignificant amount of negative charge must have been inducedby supplementary anions. In the case of the polymer D com-posite, it can only be OH�. In the case of the polymer B com-posite, though, monocarbonate was detected in the XRDpattern (Fig. 7(c)), and we thus cannot rule out the presenceof CO3

2� anions. Furthermore, elemental analysis of the solutionafter reaction yielded measured Ca/Al atomic ratios of 2.0170.5and 2.1470.5 for the solid obtained by coprecipitation inthe presence of polymers B and D, respectively. Therewas thus an excess of Ca versus the nominal ratio of 2 in the([Ca2Al(OH)6]

1) mineral layer. As neither portlandite nor cal-cite was apparent from the XRD pattern, one can only concludethat a significant amount of Ca21 ions was still screening thecharges of the polyelectrolyte. All in all, one could propose the

Fig. 1. Examples of polymer influence on the flow properties of cementpastes with different anhydrous cement compositions. The water to ce-ment ratio is 0.25 and flow is measured using the standard flow cone testas mentioned in Sidebar 1.

November 2009 Polycarboxylate Intercalation in Cementitious Systems 2475

Sidebar 3. Concise Coverage of Cement Composition

Portland cement is a mixture of clinker and calcium sulfate in a ratio of approximately 95%–5%. The clinker is manufacturedby calcination up to 14501C of a finely crushed and homogenized mixture of limestone and clay to form calcium silicates as wellas aluminates and calcium aluminoferrites in a ratio of approximately 80%–20%. The average mineralogical composition of thePortland cement clinker is given in Table II.42

It is a common usage in cement chemistry to write the first letter of the oxide in place of its chemical formula:

C ¼ CaO; S ¼ SiO2; A ¼ Al2O3; F ¼ Fe2O3; M ¼MgO; �S ¼ SO3; �C ¼ CO2; H ¼ H2O . . .

Thus, the principal components of Portland cement are written as oxides:(1) Tricalcium silicate, Ca3SiO5 or 3CaO, SiO2, written C3S(2) Dicalcium silicate, Ca2SiO4 or 2CaO, SiO2, written C2S(3) Tricalcium aluminate, Ca3Al2O6 or 3CaO, Al2O3, written C3A(4) Tetracalcium aluminoferrite, Ca4Al2O10Fe2 or 4CaO, Al2O3, Fe2O3, written C4AF(5) Calcium sulfate, CaSO4 or CaO, SO3, written C�S

Cement hydration defines, in the most general way, all the reactions that occur as soon as cement is mixed with water: theoriginal anhydrous phases dissolve, leading to a supersaturated solution with respect to less soluble hydrated phases, whicheventually precipitate, thus forming the hardened cement paste (Fig. 2).

Two hydrated phases precipitate from the dissolution of anhydrous silicate: calcium silicate hydrate (C–S–H), which is themost important constituent of the hydrated cement, and calcium hydroxide (portlandite, CH).43

The hydration of the silicate anhydrous phase in a Portland cement can be summarized schematically according to

C3SþH2O! C2S2Hþ 1; 3CH

C2SþH2O! C2S2Hþ 0; 3CH

The Ca/Si ratio in C–S–H depends on the composition of its equilibrium solution. It varies between o1 and 2. In Portlandcement, this ratio is about 1.7. As the dissolution of the C3S produces more calcium ions than needed for C–S–H precipitation,the solution becomes oversaturated with respect to calcium hydroxide and portlandite forms.

Hydration of aluminate phases leads to the formation of different hydrates depending on the presence of calcium sulfate: atan early age, when there is still solid gypsum dissolving in the interstitial water, the less soluble hydrate is ettringite. This is acalcium trisulfoaluminate (called AFt in the cement community). When all the solid calcium sulfate is consumed, calciummonosulfoaluminate (monosulfate for short) is more stable.

The corresponding reactions may be represented by the following equations42:

C3Aþ 3C�SH2 þ 26H2O! C6A�S3H32

(initial formation of ettringite AFt)

2C3Aþ C6A�S3H32 þ 4H2O! 3C4A�SH12

(once the gypsum is consumed: dissolution of ettringite and precipitation of monosulfate)

C3Aþ CHþ 12H2O! C4AHx

(in the absence of sulfate anions, they are replaced within the monosulfate structure by hydroxyl anions to form the mineralhydrocalumite that can form a limited solid solution with the monosulfate).44

Tricalcium silicate is the predominant phase in Portland cement but tricalcium aluminate is the most reactive. If calciumsulfate were not added to the clinker, hydrocalumite would precipitate very rapidly and lead to a flash set of the material, whichin turn would lead to poor mechanical properties.45 With the addition of gypsum, it is ettringite (AFt) that precipitates, via aslower process than hydrocalumite precipitation. Flash set is avoided, and the normal setting occurs at a later time due to theprecipitation of C–S–H. However, in some cases, because the precipitation of the monosulfate is very fast, some monosulfatemay precipitate before ettringite. The kinetics of this process are very much influenced by the hydration state of the calciumsulfate. Indeed, owing to high temperature during grinding, gypsum can partially dehydrate leading to a hemihydrate of highersolubility. It should also be noted that more recent research indicates that the real passivation of C3A reactivity takes placemainly through adsorption on its surface of sulfates (and calcium).46

Portlandite and ettringite are generally well crystallized while the monosulfate and C–S–H especially are poorly crystallizedand have a high specific surface.

The fresh cement paste is a concentrated suspension of charged particles in an electrolytic solution. Because of the highsurface charge density of the particles and the presence of divalent ions in solution (Ca21, SO4

2�), the particle interaction forcesare dominated by attractive correlation forces, leading to the aggregation of the particles and then to a connected network. Theprecipitation of hydrates with high surface reinforces the network cohesion strength and the paste ‘‘sets’’ and hardens. As thehydration proceeds, water is consumed and the interparticle forces are then dominated by iono-covalent bonds and capillaryforces.

2476 Journal of the American Ceramic Society—Giraudeau et al. Vol. 92, No. 11

following formula for composite B:

½Ca2AlðOHÞ6�þð½COO��0:230:01Ca2þÞXn�

0:79=n

and for composite D

½Ca2AlðOHÞ6�þð½COO��0:730:14Ca2þÞOH�0:54

with X standing for hydroxyl and carbonate anions.In conclusion, the occurrence of an organo hydrocalumite

(hydroxy AFm) intercalation product was clearly evidencedduring coprecipitation of Ca and Al ions in the presence ofPCE polyelectrolytes in a lime-saturated solution, a situationsimilar to the early hydration of cement in the presence of PCEfluidizers. The composite exhibited basal spacings in the range of10 nm. The charge balance of the hydrocalumite mineral layerby the polyelectrolyte varied significantly for the two polymerstructures under study. To understand these differences, it isnecessary to examine the relationship between the structure andthe conformation of these polymers.

IV. Experimental and Theoretical Analysis of PCEMolecularConformation

(1) Physical Characterization

(A) Transfer of Populations by Double-Resonance(TRAPDOR) Magic Angle Spinning (MAS) Nuclear Mag-netic Resonance (NMR): MAS NMR experiments were per-formed on a Bruker ASX500 spectrometer at 11.74 T. Thezirconia rotor of 4 mm diameter was spun at 12 kHz. 1H MASNMR spectra were measured by a single p/2 pulse with a pulselength of 6 ms and a recycle delay of 15 s. In the 1H-27Al TRAP-DOR experiment, the dipolar dephased 1H spin echo (S0) is re-corded after 8.3 ms of evolution time (100 rotation period), andcompared with the spin echo obtained under the same condi-tions but with additional dephasing due to a continuous irradi-ation of 27Al at an on-resonance frequency during the evolutionperiod irradiation (S). The TRAPDOR spectrum, defined as

Sidebar 4. Macro-Defect-Free Cement

In 1981, a small revolution occurred in the world of cement. It was shown that the low tensile strength of cementitious materialscould be increased to levels competing with metals.47 A patent was obtained by these authors 2 years later.48 The product wascalled macro-defect free (MDF) cement.

MDF is a cement-organic composite that is processed with extremely low amounts of water and relatively high amounts ofpolymer or polymer precursors. The final product is essentially that of anhydrous cement grains bound together by a polymerand a cement hydrate–polymer interphase.49 In this sense, MDF composites may be considered in the family of organo-mineralcomposites in which the mineral part would be cementitious.

The excellent mechanical properties were initially attributed to the extremely low porosity and absence of large flaws, whichled to the name of MDF.

Calcium aluminate cement shows much better properties in the final material than ordinary Portland cement. This has beenexplained by the formation of chemical bonds between the aluminum ions and the organics-added MDF.49 The fact that betterproperties tend to be obtained with calcium aluminate cements also suggests that if organo-mineral compounds are formed,then they must be organo-aluminates, which makes them in a sense precursors to the phases examined in this paper, althoughthe polymers used are clearly different.

After its invention, MDF faced, from a technical perspective, two main technological challenges: moisture sensitivity andprocessing.

The original formulations of MDF did not retain strength when exposed to moist conditions. This could lead to strengthlosses of up to 80%. Some of this was accounted for by hydration of the cement grains. However, a detailed investigation of theproblem seems to have led to the conclusion that the polymers used swell in the presence of water.50 This swelling indicates asignificant loss of strength and allows water ingress. This in turn causes partial dissolution of the interphase between the cementand the polymer. When the sample is dried again, it recovers most, but not all of its strength.

Much work on MDF has thus been devoted to providing solutions to the moisture issue.51–53

The processing of the material is the other technological difficulty. At this stage, MDF objects are mainly prepared byextrusion or pressing and the green body then requires a heat treatment. This is probably another of the reasons why theseproducts have not found their place in the market.

Table II. Typical Oxide Composition of Portland Clinker

Component Cement notation Formula

Weight percent

in clinker

Tricalciumsilicate

(alite)

C3S Ca3SiO5 60–65

Dicalciumsilicate (belite)

C2S Ca2SiO4 10–20

tricalciumaluminate

C3A Ca3Al2O6 8–12

tetracalciumaluminoferrite

C4AF Ca4Al2O10Fe2 8–10

Fig. 2. Picture of a cement paste obtained with an optic microscopeby reflected light on a polished surface, courtesy of Dr. Baroghel-Bouny.41 (1) C3S, (2) C–S–H, (3) C2S, (4) C3A and C4AF, and (5)Ca(OH)2.

November 2009 Polycarboxylate Intercalation in Cementitious Systems 2477

Sidebar 5. Layered and Intercalated Structures

Typically, solids are held together by a network of strong iono-covalent bonds between the atoms. In layered materials,however, this network is regularly interrupted along one direction and replaced by a weaker type of bonding such aselectrostatic, van der Waals, or hydrogen bonds. As a consequence, the atoms are strongly held together only within planeswhile the cohesion of the ensemble is realized by weak forces stacking the planes together. There is a large chemical variety oflayered materials, such as graphite, molybdenum, and tin sulfide, some forms of lithium silicate and phosphate, titanates, andnumerous hydroxides of hydrolyzable cations. Being stable under natural conditions of pressure and temperature, severallayered hydroxides are ubiquitous in nature: gibbsite (Al(OH)3), brucite (Mg(OH)2), and portlandite (Ca(OH)2), to name a few.Clays form another class of naturally occurring layered material that can actually be described as derived from layeredhydroxides by condensation with one or two tetrahedral silica layers.57

Because of the weak bonding between the layers, layered materials exhibit peculiar properties that make them industriallysignificant. The layers can slip easily over each other and this makes them good lubricants (graphite, talc, MoS2). Furthermore,weak forces are sufficient to drive the reversible intercalation of foreign ions, atoms, or molecules between the layers. They canthus be used as ionic charge carriers in batteries, or as adsorbants in various applications. To illustrate this point, one canconsider the important case of the montmorillonite clay. The structure of montmorillonite is based on a layer of aluminumatoms in octahedral coordination, as in gibbsite. However, some of the hydroxyl groups from the octahedral layer are removedto allow condensation with two layers of tetrahedral silicon, one on each side (thus forming the so-called TOT structure).58

Furthermore, a variable amount of Al(III) is replaced by Mg(II), resulting in a charge imbalance driving the penetration ofcations within the interlamellar space. In turn, when montmorillonite is suspended in an ionic water solution, entropy andelectrostatic can drive, against van der Waals and hydrogen bonds, the exchange of the solution cations with the ones originallypresent between the clay layers, while hydration forces drive the penetration of water or other polar molecules. As aconsequence, montmorillonite is a cation exchanger that swells. It can even exfoliate in water when the electrostatic repulsionovercomes the van der Waals attraction between the layers.59

The structural chemistry of LDH obeys the same logic as the one of clays (Figs. 3 and 4). Their structure is still based on ahydroxyl octahedral layer, albeit made of divalent atoms in this case, for example the ones of brucite (Mg(II)) or portlandite(Ca(II)). The charge imbalance arises from the substitution of some of the divalent atoms in the layer by a trivalent one(hence the name of ‘‘double’’), leading to an excess charge. Consequently, the LDH can be viewed as cationic covalentplatelets with electroneutrality ensured by interlayer anions. A typical member of the LDH family is hydrotalcite:[MgAl2(OH)6]

1 � 1/2CO32� � yH2O. This mineral is based on an Mg(OH)2 brucite layer into which one Mg21 out of two is

replaced by Al31. Electroneutrality is provided by compensated interlayer anions such as carbonates, sulfates, nitrates, orhydroxides. A large number of combinations of hydrolyzable divalent M(II) and trivalent M(III) cations can form LDHstructures (for a relatively recent compilation of commonly reported LDH compositions, see Khan and O’Hare.56).

Apart from their chemical properties, LDH also exhibits peculiar physical properties. Just like clays, they are constituted, onthe one hand, of flexible, charged, nano- to micro-layers and, on the other, of an interlamellar space filled by water andcompensating ions. Thus, LDH share with clays some of their original usage properties: these materials act as ion exchangers,and may swell and delaminate in the appropriate solvent. For this reason, LDH are sometimes referred to as ‘‘anionic clays.’’This is to emphasize that, while clays are negatively charged and as such act as cation exchangers, LDH are anion exchangers.LDH exhibit, however, a phenomenological difference from clay: they do not swell extensively in water. The charge density isknown to be a key factor affecting the capacity of a layered mineral to swell, and eventually exfoliate, processes that can beviewed as a necessary requirement for the intercalation of bulky ions. Pellenq et al.60 have developed numerical primitive modelsdemonstrating the importance of ion correlation forces for the cohesion of layered materials, correlation forces that increasewith the surface charge. Incidentally, this is one of the explanations for the exceptional cohesion of the cement paste: it is relatedto the high charge density of its main constituent, layered calcium silicate hydrate (C–S–H). Along the same line, one canrationalize why, due to their surface charge density of 1.75 positive charge per nm2 (see Meyn et al.61), LDH structures do notspontaneously delaminate in water. Besides this, there is another kinetic constraint opposing the delamination of LDH. Becauseof the high and regularly localized charge density, the oxygen basal layer is strongly polarized, resulting in a cohesive hydrogenbond network between the sheets’ hydroxyl groups and the intercalated hydration water, keeping the LDH sheets together. Theenergy cost to the disruption of this network constitutes an insurmountable barrier to delamination in the aqueous phase,regardless of ionic strength or ion valency. For all these reasons, LDH do not spontaneously delaminate in water.

A large number of combinations of hydrolyzable divalent M(II) and trivalent M(III) cations can form LDH structures, andnumerous synthetic LDH have been produced based on a variety of divalent/trivalent couples. Industrially, the motivation forLDH synthesis is the development of mixed oxide precursors62 of new intercalated composites with original mechanical oroptical properties and of carriers for chemically or biologically active anions. In particular, for heterogeneous catalysis, LDHhave been pillared by polyoxometalates anions and complexes.63 Also, LDH-based organic/inorganic intercalation compositeshave been synthesized by anion exchange of LDH with diverse molecules from simple sulfonates64,65 and acrylates66 topolymers,67,68 biopolymers,69 or even DNA.70 However, as already stated, LDH do not swell spontaneously in water and thus,the intercalation chemistry of large organic polymers within LDH is slightly more elaborate than in the case of clays and hasbeen recently reviewed by Taviot-Gueho and Leroux.71

The distance between the LDH layers depends on the size and spatial configuration of the intercalated anions and molecules.Therefore, the conformation of the intercalation product can be conveniently followed by measuring the basal distance byXRD. In this respect, the case of linear carboxylic acids is particularly illustrative of the LDH’s chemistry of intercalation and itsstructural consequences (Fig. 5).

The impossibility to delaminate is believed to be a major obstacle to the intercalation of polyelectrolytes because of theirnanometric sizes. A recent interest in the synthesis of highly ordered, transparent, LDH–polymer intercalate has prompted thedevelopment of two preparation methods to circumvent the high delamination barrier. The first one is to delaminate beforehandthe LDH in a nonaqueous solvent such as formamide.72 The second one is to intercalate beforehand by anion exchange asurfactant whose hydrophobic tail props apart the LDH layers.73 Nevertheless, in order to by-pass the difficult delaminationstep, it is also possible to synthesize the polymer/LDH composite from precursors. In the coprecipitation method, the

2478 Journal of the American Ceramic Society—Giraudeau et al. Vol. 92, No. 11

(S0�S)/S0, corresponds to the 1H dipolar coupled to 27Al. The27Al radio frequency field amplitude was 140 kHz. Trimethylsi-lane was used as an external standard for calibrating the 1Hchemical shift.

(B) DLS: Hydrodynamic radii were determined in solu-tions of water saturated with CaO by dynamic light scattering at251C using an He–Ne ions laser (632.8 nm) from Uniphase(model 1145P). All solutions were filtrated through 0.45 mmpore-sized filters. In each case, measurements were performedfor polymer concentrations of 0.02, 0.05, and 0.1 g/L. Hydro-dynamic radii were calculated from the Stokes–Einstein equa-

tion with a diffusion coefficient obtained by extrapolation atzero concentration.

(C) Small-Angle Neutron Scattering (SANS): Themeasurements were performed at the Laboratoire Leon Brill-ouin of the CEA (Saclay, France) on the PAXY spectrometer atroom temperature (RT). The neutrons were detected with a BF3XY detector with a lateral resolution of 5� 5 mm2. The wave-length of the neutrons was 10 A. The scattering intensity wasmeasured at a detector—sample distance of 1.3 m correspondingto a regime of the scattering vector q of 2.9� 10�3 A�1oqo1.7� 10�1 A�1. Various concentrations (2, 5, 10, and 20 g/L) ofpolymers in D2O or D2O saturated by CaO (about 20 mM) wereplaced in glass cells. The polymers were dried under vacuum anddispersed in the solvent a few minutes before the measurement.

(2) Results

The Debye equation relates the SANS scattering intensity atsmall diffusion vector q to the radius of gyration Rg of the di-luted polymer through

IðqÞ ¼ A

q4R4g

e�q2R2

g � 1þ q2R2g

� �

where A is a constant including the concentration of these twospecies and the contrast factor between the deuterated solventand the hydrogenated polymer chains.82

polyelectrolyte preexists in a solution containing the metal salts constitutive of the targeted LDH. The metal salts hydrolyze toform the LDH layer around the charged polymer acting as its counter ion. A similar route is to reconstruct the LDH layersdirectly around the charged polymers by rehydrating in an aqueous solution the mixed oxide resulting from a thermal treatmentof LDH. For example, coprecipitation has been used to synthesize hydrotalcite (Mg/Al LDH) intercalated by terephatalateanions and then polyaspartate,74,75 and rehydration after thermal treatment has been used to intercalate glycerol andnaphthalene carboxylate.76,77

Fig. 3. (a) Schematic representation along the lamellae of the mono-carbonate (the carbonated form of hydrocalumite) structure. It is madeof [Ca2Al(OH)6]

1 covalent layers held together by hydrogen bonds andelectrostatic interactions with the hydration water and the charge-bal-ancing anions located in the interlamellar space. The octahedrons arecentered on Al31; the Ca21 is in sevenfold coordination. This two-di-mensional structure can be viewed as 0.48-nm-thick platelets separatedby a distance depending on the hydration level and nature of the ex-changeable anion. Water molecules and carbonate anions occupy theinterlamellar space. The basal distance (sum of layer thickness and in-terlamellar space) can be conveniently obtained from the 00(l) reflec-tions of XRD pattern. Reproduced with permission from Renaudin.54

(b) Scanning electron micrograph showing the hexagonal layer ofhydrocalumite mineral obtained by coprecipitation.

Fig. 4. Perspective view of the [Ca2Al(OH)6] (a) and [Mg2Al(OH)6](b) layers. This structure derives from the Mg(OH)2 or Ca(OH)2 layersinto which trivalent Al is substituted. The net charge of the layers istherefore positive. Reproduced with permission of the Royal Society ofChemistry.56

Fig. 5. Interlayer distance obtained from X-ray diffraction for the in-tercalation of dicarboxylates within Li–Al layered double hydroxide.Reproduced with permission of the Royal Society of Chemistry.56

Sidebar 5. Continued

November 2009 Polycarboxylate Intercalation in Cementitious Systems 2479

The radii of gyration measured by SANS in a 2 g/L solutionat equilibrium with lime, which is a solution representative ofthe synthesis conditions of the composite, are reported in TableIII within 70.5 nm for polymers B, D, and E.

Hydrodynamic radii of 572 nmwere obtained by DLS for allpolymers.

Here we draw from the large body of literature on polyelec-trolytes in solution, at interface, and on polymers in confinedgeometries. The conformation of a diluted polyelectrolyte is well

studied.83 It is controlled by two conflicting phenomena: elec-trostatic stiffening and exclusion effects. The former refers to theCoulombic interaction between charges along the polymer back-bone. The latter refers to the interaction of the monomer withsolvent molecules relative to its interaction with another mono-mer. PEO chains are neutral and their conformation is solelydictated by such solvent effects. Water is a good solvent for PEOat RT, and its chain can be described as a thermal blob whosesize is governed by excluded volume statistics according to theFlory theory.84 For a polymethacrylate chain, the situation isopposite. Water is a poor solvent for PMA and it would shrinkand phase separate if it were not extended by the repulsive elec-trostatic forces associated with the deprotonation of the acidicgroups.85 Thus, if the pH of the hydration solution is well abovethe pKa of the carboxylate groups, they are fully deprotonatedand the chain is soluble. However, if the solution is saturated incalcium ions, which are expected to pair with the carboxylategroups, the carboxylate charges are then fully screened and thepolymer, being in a bad solvent, precipitates. In a cement paste,the hydration solution is indeed very basic and saturated in cal-cium ions and it is only the PEO grafts that maintain thepoly(methacrylate-g-PEO) in solution. Borget et al.86 haveshown that the conformation of PCE follows the blob modeldeveloped by Gay and Raphael87 for comb-like polymers in agood solvent and adopts a flexible backbone worm (FBW) con-formation. This reasonable assumption is experimentally sup-ported by the fact that the measured DLS hydraulic and SANSgyration radii have similar values. Within this assumption, it ispossible to estimate the geometry of the polymer in solution ac-cording to the following equation (see Sidebar 6):

RPEO ¼ aP610

RC ¼ aP710N�

110

Rg ¼ aaP

aN

� �1=5

n35P

25N

15

The scaling factor can be obtained by adjustment with theradii of the gyration of polymers measured by SANS. This led tothe value of

aaP

aN

� �1=5

¼ 0:26� 0:04nm

And thus to the core sizes reported in Table III.If the conformation of the polymer in solution is relatively

well understood, when confined between the mineral layers, thesituation is significantly different.90 First, the conformation ofthe chains is modified by the geometric confinement leading to

Table III. Characteristics of the Poly(Methacrylate-g-PEO) Polymers and Predictions from the Blob Model (See Sidebar 5)

Polymer B D E A C

Number of nongraft PMA monomers per motif: x 2.3 5.0 8.3 3.0 6.4Number of monomers per motif: N5x11 3.3 6.0 9.3 4.0 7.4Grafting ratio t5 1/N 30% 17% 11% 25% 14%Number of PEO monomers on the graft chains 23 23 23 23 46Number of motifs per polymer chain: n 13.5 5.0 5.5 8.3 6.5Mn (g/mol) 17900 9619 10880 14941 13 134Mw (g/mol) 35604 18 619 22346 43282 26 217I5Mw/Mn 2.0 1.9 2.0 2.8 1.9Rh (DLS) (nm) 5.6 4.6 4.8 5.2 5.4Rg (SANS) (nm) 5.1 4.9 4.1RPEO (model) (nm) 1.7 1.7 1.7 1.7 2.6RC (model) (nm) 2.6 2.4 2.3 2.5 3.9Rg (model) (nm) 5.6 3.6 4.2 4.4 5.7

Fig. 6. General formula, schematic representation, and flexible back-bone worm coiled conformation according to the Gay and Raphaelmodel (see text) for two of the polycarboxylate ethers used in this study.

2480 Journal of the American Ceramic Society—Giraudeau et al. Vol. 92, No. 11

entropic losses.91 Second, the polymer interacts with the mineralsurfaces. The intercalation product configuration thus resultsfrom the interplay between enthalpic (interactions between lay-ers, water, and polymer) and entropic balances (loss due to con-finement of the polymer and gain due to the separation of thelayers) and is difficult to predict beforehand.92

Building on the idea that the driving force for the formationof the composite is the electrostatic interaction between the neg-atively charged polyelectrolyte and the positively chargedhydrocalumite layer, it might seem reasonable to assume thatthe PCE anchors to the surface through the carboxylic groups.A first approach would then be to consider that the negativelycharged backbone is constrained by electrostatic interaction tolie flat on the positively charged mineral layer, while the PEOside-chain extends away from it in a brush or a mushroom con-formation depending on the ratio of the distance between graftchains and their length (Fig. 11(a)) according to de Gennes’theories.93 Such a configuration does not appear to be compat-ible with our XRD observations. Indeed, it would lead to aninterlayer space in the range of one or twice the PEO side-chaincharacteristic coil dimension (1.7 nm within the Gay and Ra-phael model) assuming a bilayer conformation when the mea-sured interlayer space is about three times this value. It has alsobeen observed that PEO chains can crystallize as parallel chainsin the interlayer region of dry LDH.79 However, such crystal-lization is less likely to occur in fully hydrated phases and wouldhave appeared anyhow prominently in the XRD patterns. Thus,this possibility can be ruled out in the present case.

In order to obtain more information on the conformation ofthe polymer at the surface, TRAPDOR NMR between the pro-tons of the polymer and aluminum of the inorganic layer wasperformed. The physics of this experiment need not be devel-oped here and can be found in the original paper.94 The TRAP-DOR experiment reveals which abundant spin 1/2 nuclei aredipolar coupled to a quadrupolar nucleus. It has been usedmainly to probe the topology of hydroxyl95 or fluoride groups96

on oxides. However, it is also a powerful method to probe mo-lecular interactions in polymer/oxide composites.97 In the pres-ent study, the spin 1/2 was the 1H from the polymer and thequadrupolar nucleus was the 27Al of the hydrocalumite layers.Because dipolar coupling implies through-space proximity, itrevealed which polymer protons were close to the mineral layers(Fig. 12). By comparison with the liquid-state NMR, the reso-nances between 3 and 4 ppm in the MAS proton spectra can beassigned to the methyl and ethyl protons of the PEO graftchains. To analyze the TRAPDOR spectra, it must be under-stood that only the protons of the graft chain can potentiallyappear in the MAS TRAPDOR spectra. Indeed, the backboneprotons give rise to resonances that too broad to be detected,while the mineral layer protons are fully relaxed before detectionof the TRAPDOR signal due to a short spin–spin relaxationtime T2. The TRAPDOR experiment showed that the PEO graft

Fig. 7. X-ray diffraction (XRD) powder pattern of hydrocalumite ob-tained by coprecipitation of Ca and Al without a polymer (a), withpolymer D (b), with polymer B (c), and with polymer B after addition ofsulfate (d). The lower stick pattern is the one of monosulfoaluminate(JCPDS 42-0062). The samples prepared with the other polymers pre-sented similar XRD patterns and are not shown.

Fig. 8. Small-angle X-ray diffraction powder pattern of hydrocalumiteobtained by coprecipitation of Ca and Al in the presence of the poly-mers. The high basal spacing reveals the intercalation of the polymerbetween the hydrocalumite layers. From these data, the interlayer dis-tance (di) is obtained and correlated to the size of the cores derived fromthe hemispheric model (RAC). The slope is 2.1, indicating a bilayer ad-sorption of hemispheric cores while the intercept is close to 0.6 nm, thatis, the interlayer distance in the absence of polymer.

November 2009 Polycarboxylate Intercalation in Cementitious Systems 2481

chain protons exerted a significant TRAPDOR effect. Thismeant that there was a relation of proximity between the pro-tons of the PEO graft chain and the aluminums of the minerallayer. Therefore, the grafts did not extend away from the layeras in a ‘‘brush’’ conformation, but the conformation of thewhole polymer remained in a coiled conformation reminiscentof the one in solution (more ‘‘mushroom’’ type). This was con-sistent with the chemical analysis of the composite, whichshowed that a large number of the backbone charges were stillscreened by calcium ions in the interlamellar space.

A more realistic view is thus to consider that the coiled FBWconformation is somewhat preserved on macromolecules ad-sorbed on the mineral surface, leading to a proximity of thegrafts to the mineral layers as clearly evidenced by TRAPDORNMR.

Flatt and colleagues have theoretically demonstrated andexperimentally (by atomic force microscopy) established thatthe PCEs adsorbed on charged planar surfaces adopt a con-formation compatible with a model of hemispheric cores ofsizes related to the Gay and Raphael expression through (seeSidebar 6)

RAC ¼ffiffiffi2p

RC

Such a view is compatible with the TRAPDOR observation as itresults not only in proximity with the mineral layer of thecharged polycarboxylate backbone but also of the graftedPEO coil part of the hemispheric cores. The composite wouldthen simply be a stacking of hydrocalumite layers and adsorbed

PCE hemispheric cores (Fig. 11(c)). This stacking would lead toan interlayer space scaling with twice the hemispheric radius.Indeed, considering a layer thickness of 0.48 nm for hydrocalu-mite, the basal spacing measured by XRD corresponds to aninterlayer distance not far from twice the hemispheric cores’ sizeas can be seen from the linear regression in Fig. 8. It is actuallysurprising that this model captured the experimental data sowell, considering that it is a purely entropic model that does nottake into account osmotic and electrostatic contributions. Thismeans that the chain can be considered as weakly (if at all)confined in the composite.

Another important check to validate the hemispheric coresmodel is the issues of surface and charge balances.

Let us first examine surface coverage. Knowing the amountof polymer adsorbed per [Ca2Al(OH)6]

1 unit and the radius ofthe hemispheric cores, we can obtain the relative surface cover-age of the mineral by the PCE polymer. For a monolayer, thesurface coverage would be one. However, for all the polymersstudied, we obtained a coverage between 2 and 3. Althoughsurprising at first glance, this result actually did not invalidatethe hemispheric cores model. An important clue for this is thatthis coverage ratio was relatively constant independent of thepolymer structure. This indicated that the system was behavingas if it had a surface proportional to the aluminum content ofthe solids but larger that estimated on the basis of the[Ca2Al(OH)6]

1 units. In fact, this observation could be ratio-nalized, considering that a single polymer chain develops a basalsurface larger than the surface of an individual hydrocalumiteparticle. We could estimate the average lateral extent of thehydrocalumite crystallite from the breadth of the 100 XRD re-flection (not shown) via the Scherrer equation and, in turn, es-timate the surface developed on the side of a single hydrocalu-mite particle. We measured a lateral coherence length for thecrystallites of between 5 and 9 nm corresponding to a surface oneach side of the platelets in the range of 20–80 nm2 per side ofthe hydrocalumite platelets. Presently, within the hemisphericcore model, the polymers of Table III could potentially cover asurface between about 50 and 120 nm2 per chain (this is a con-servative estimate not considering packing and charge densitymatching issues). Consequently, even for a single-chain adsorp-tion, the nominal coverage derived from the model can onlyexceed unity, as was indeed observed. The hemispheric coresmodel was thus compatible with the observed surface coverages.

Regarding the electrostatic charge balance, it is known thatwhen polyelectrolytes are adsorbed on charged surfaces, a de-crease of the charge density of the polyelectrolyte results inhigher adsorbed amounts.98 Thus, because the respective corescontain about the same number of carbon atoms (162 for B and146 for D), if the polymer charges were not screened, less poly-mers D than B would be adsorbed (expressed in weight percentcarbon). The opposite was actually observed. In our view, thisfact did not invalidate the hemispheric cores model proposedbut meant that the polymer had electrostatic interactions notonly with the mineral layer but also with the calcium ions thatplayed a strong role. In other words, these interactions werelikely to depend not only on the density of charge but also onscreening effects. The latter results in significant amounts of Cacounter-ions in the interlayer.

Finally, all the data presented here were compatible with ahemispheric cores model for the conformation of the PCE in-tercalated within the hydrocalumite interlayers.

V. Resistance to Sulfate Exchange and Relevance to CementChemistry

We have seen that PCE hydrocalumite intercalation compositeforms spontaneously in lime-saturated solutions and that thestructure of the composites can be rationalized with entropicconsiderations linked to the polymeric nature of PCE. However,in a cement paste, the hydration solution becomes rapidly rich insulfate ions as the calcium sulfate and alkali sulfates in the ce-

Fig. 9. TEM image of the polymer C poly(methacrylate-g-PEO)/hydrocalumite composite. Lines are drawn as guides to the eye. The in-terlayer spacing measured between those lines varies between 7 and 10nm.

2482 Journal of the American Ceramic Society—Giraudeau et al. Vol. 92, No. 11

ment dissolve. If the composite were not sulfate resistant, itspossible occurrence would not provide a realistic explanation forthe dependence of the rheological properties of the paste on thealuminate content. The stability of the composite was thus eval-uated in the presence of sulfate.

(1) Sulfate Competition

The study of the reactivity of the composite phase with sulfateswas performed on the polymer-AFm. At the end of the synthe-sis, a volume of a Na2SO4 concentrated solution correspondingto 2 mmol of SO4

2� was added to the composite suspension

Sidebar 6. Conformation of Comb Copolymer Dispersants

The conformational behavior of the comb copolymers has been examined in solution using light scattering and viscositymeasurements.86,88 It has also been examined by atomic force microscopy on the surface of cement,88 magnesium oxide,36,89 andmore recently C–S–H.88 Results from the latter three papers clearly indicate that, when adsorbed, the side chains of thesecopolymers are coiled and not stretched as they are usually represented. Furthermore, the much larger values of layer thicknessobtained in the early AFM experiments on cement3 can reasonably be attributed to artifacts from the reactivity of the substratesused.8 The adsorbed layer thickness is therefore on the order of a few nanometers rather than tens or even hundreds of nanometers.

In fact, it was shown that the conformation of these copolymer can be rationalized by scaling a law approach88 that buildson the work by Gay and Raphael87 for comb homo-polymers in solution. Because this treatment is of interest for analyzing thebasal spacing of our LDH, we briefly present the basis of this model and provide the main equations of interest to the problem atstake here.

We define the polymer backbone as the assemblage of n repeating structural units, each containing N monomers and one sidechain of Pmonomers. The size of the backbone and the side units is, respectively, denoted aN and aP. For a comb homopolymer, aphase diagram can be easily represented with five different types of conformations that are then defined as: decorated chain, FBW,stretched backbone worm, stretched backbone star, and flexible backbone star as shown in Fig. 10. Owing to the nature of thescaling law approach, the position of the boundaries is not exact.

The polymer of interest in this study lies mainly in the FBW regime, where the polymer can be viewed as a chain of cores, eachwith a number of side chains given by

n2C ¼P

N

aP

aN

� �2

ð1Þ

The end-to-end distance of the polymer in solution is given by

R ¼ aP

aN

� �21� 2wð Þ

2

!1=5

aP P2=5N1=5 n3=5 ð2Þ

while as for the individual cores it is

RC ¼aP

aN

1� 2wð Þ2

� �1=5

aP P7=10N�1=10 ð3Þ

where v is the Flory parameter. For the polymers of interest in this paper, we can use aP50.25 nm, aN5 0.36 nm, and v5 0.37 at251C.88

On the surface it is proposed to treat the polymer as a chain of hemispheres, each having a number of side chains denoted nAC.This choice is suggested by the fact that the negatively charged polymer backbone will want to interact with the positively surface,while the side chains that are nonadsorbing will extend away from the surface. By selecting a model of a chain of hemi-spheres, weare therefore implicitly incorporating the basic physical behavior expected from these polymers in their adsorbed state. As for thesolution case, we assume that the equality between the elastic energies of the side chains and the main chain segment in a given coredefines the number of side chains nAC in this core (hemisphere):

nAC ¼aP

aN2P

N

� �1=2

ð4Þ

The layer thickness of the adsorbed polymers is then given by

RAC ¼ 2ffiffiffi2p

1� 2wð Þ aPaN

� �1=5

aPP7=10N�1=10 ð5Þ

And the surface occupied by each molecule by

S ¼ pffiffiffi2p aNaP 2

ffiffiffi2p

1� 2wð Þ aPaN

� �2=5

P9=10N3=10n ð6Þ

A schematic representation of how this model would represent the polymer in solution and on the surface is represented in Fig.10(b).

Of particular interest is the comparison of the layer thickness predicted by Eq. (5) with the experimental data reported in themain section of this paper, more precisely with the XRD basal spacing measured at 100% RH.

November 2009 Polycarboxylate Intercalation in Cementitious Systems 2483

maintained at 51C. It is important to note that, before sulfateaddition, the composite was at equilibrium with the synthesissolution and that, as a result, a significant amount of the poly-mer initially introduced remained in solution. For example, theconcentration of polymer still present in solution at the time ofthe sulfate addition was 1.6 g/L (2.8 mequiv COO�/L) and 1.2 g/L (3.9 mequiv COO�/L) for the poly(methacrylate-g-PEO)/hydrocalumite composite obtained from polymers B and D, re-spectively. The quantity of sulfates added corresponded to thequantity necessary to obtain the SO4

2�/Al31 ratio of mono-

sulfoaluminate (0.5). Sulfate concentrations were measured insolution by ICP-AES.

(2) Results and Discussion

The experimental synthesis conditions corresponded to a finalsolid concentration of [Ca2Al(OH)6]

1 in the suspension of 4.0mM, which is 4.0 mequiv/L of positive charge on the minerallayer, and an initial concentration of polymer equal to 2 g/L,which is nominally 3.7 and 6.8 mequiv/L COO� for B for D,respectively. Of this, 2.8 and 3.7 mequiv COO�/L remained insolution, which meant that only 0.23 and 0.73 mequiv COO�/[Ca2Al(OH)6]

1 were in the solid phase, respectively, for poly-mers B and D. After sulfate addition (final total amount 4.0mequiv/L), the concentrations in solutions of sulfate ions andCOO� evolved with time. This time evolution is shown inFig. 13 expressed in terms of a negative equivalent charge fixedin the solid phase. The curves are similar for all polymers andexhibit three phases: a fast removal of sulfates with little or norelease of PCE in the first minutes (first phase), then a plateaufor both the amount of sulfates and the amount of polymer insolution (second phase), and finally the fixation of sulfates andremoval of polymers to attain an equilibrium state (third phase).

In the first 3 min after sulfate addition, we can observe thatthe sulfate fixed by the solid corresponded for a small part onlyto polymer exchange because little polymer was released. It wasmainly hydroxyl exchange and ion pairing to the excess calcium

Fig. 11. Models of polymer conformation on surfaces. (a) Model ofadsorbed comb polymers with the PEO grafts extending away from thesurface in a mushroom or a brush configuration. These configurationsare not compatible with the transfer of populations by double-resonance(TRAPDOR) nuclear magnetic resonance (NMR) observations. (b)Model of adsorbed comb polymer adopting a flexible backbone wormconformation of a chain of hemispheric core at the surface. This repre-sentation derived from the blob model in solution was validated bySANS and accounts properly for the proximity of the PEO graft with theAFm layers observed by TRAPDOR NMR. (c) Model of a bilayer ofchains of hemispheric cores in the interlayer of the composites predictingan interlayer of twice the hemispheric radii as indeed observed by small-angle X-ray diffraction (see Fig. 8).

Fig. 12. One-pulse liquid-state nuclear magnetic resonance (NMR) ofCA in a lime-saturated solution, Hahn echo magic angle spinning(MAS) NMR, and transfer of populations by double-resonance (TRAP-DOR)MASNMR of the solid composites. The side-chain PEO protonsclearly exert a TRAPDOR effect and are thus in close proximity to theAl of the AFm layers.

Fig. 10. Phase diagram for comb homopolymers according to Gayand Raphael. The different domains are: decorated chains (DC), flex-ible backbone worm (FBW), stretch backbone worm (SBW), stretchedbackbone star (SBS), and flexible backbone star (FBS). RC denotes thecore size and Rp is the side chain size. Adapted with the permission ofElsevier from Gay and Raphael.87

2484 Journal of the American Ceramic Society—Giraudeau et al. Vol. 92, No. 11

that drove the fixation of the sulfate. As a result, the sulfate ionsand the PCE coexisted in the AFm phase.

After these first minutes, the amount of polymer and sulfatein the solid phase stabilized during a few hours (second phase)before the solid phase converted to monosulfate by a process ofdissolution and precipitation.99 In the case of the PCE D com-posite, there was apparent overcompensation due to the excesscalcium in the interlayer. It must be emphasize though that thisbehavior was observed at 51C and not at RT so as to stabilizethe hydrocalumite phase.

The X-ray diffractogram confirmed that the final solid phasewas indeed monosulfoaluminate (Fig. 7(d)). However, becausethe entire polymer was not released in solution at equilibrium,there was then an overcompensation of charge in both cases,especially in the case of polymer D. Because no small-angle re-flections were present in the diffractogram, it is likely that theremaining polymer in the solid phase was not intercalated intothe mineral layers but adsorbed at the surface of the particles ofmonosulfoaluminate as calcium ion pairs or charged complexes.

VI. Conclusion

The formation of intercalation composites between PCEs andhydrocalumite layers obtained by coprecipitation in a lime-sat-urated solution containing the PCE was experimentally estab-lished. This composite was able to resist sulfate exchange for

times in the range of hours. We thus propose that the PCE–AFm nanocomposite formation could be a practical key to un-derstanding the dependence of superplasticizer effects on thecement composition, noticeably its calcium aluminate content.

We were able to rationalize the composite structure using anFBW model derived from Gay and Raphael. This was substan-tiated by the observed XRD basal distances and NMR TRAP-DOR effect. We concluded that the polymer was only weaklyconfined between the hydracalumite layers, adopting the con-formation of a bilayer of hemispheric cores. However, chemicalanalysis revealed that a significant amount of calcium ions werestill associated with the polyelectrolytes in the interlamellarspaces, and the resulting charge screening resulted in variableamounts of PCE intercalated. The interaction between the car-boxylic groups of the PCE and the calcium therefore seemed tobe a key factor to predict the PCE sequestration in the cementAFM phase.

Acknowledgments

The project was funded by the Nanocem consortium and a CIFRE grant fromthe French government. We are grateful to many Nanocem members for veryvaluable input and discussions, in particular: Angelique Vichot and Xavier Guillot(ATILH), Martin Mosquet, and Christian Vernet (Lafarge). We also thank inparticular Elie Raphael (ESPCI) for helpful discussions, Irene Schober (Sika) forsupplying the polymers used in this study, Pierre Antoine Albouy for small-angleXRD characterization (Laboratoire de Physique des Solides, Orsay), LaurenceNoirez (Laboratoire Leon Brillouin, Saclay) for DNPA experiments, DaniellePerrey (Universite de Bourgogne) for the analysis of ions in solutions, and Prof. J.Plank (Technische Universitat Munchen) for a careful reading of the manuscript.Finally, we are grateful to all the members of Nanocem, and particularly KarenScrivener (EPFL), for their constant support and interest.

References

1R. J. Flatt, ‘‘Polymeric Dispersants in Concrete’’; pp. 247–94 in Polymers inParticulate Systems. Properties and Applications, Edited by V. A. Hackley, P. So-masundaran, and J. A. Lewis. Marcel Dekker, New York, 2002.

2C. Jolicoeur and M.-A. Simard, ‘‘Chemical Admixture–Cement Interactions:Phenomenology and Physico-Chemical Concepts,’’ Cem. Concr. Comp., 20 [2–3]87–101 (1998).

3H. Uchikawa, S. Hanehara, and D. Sawaki, ‘‘The Role of Steric RepulsiveForce in the Dispersion of Cement Particles in Fresh Paste Prepared with OrganicAdmixture,’’ Cem. Concr. Res., 27 [1] 37–50 (1997).

4P.-C. Aıtcin, C. Jolicoeur, and J. G. MacGregor, ‘‘Superplasticizers: How TheyWork and Why They Occasionally Don’t,’’ Concr. Int., 16 [5] 45–52 (1994).

5D. Bonen and S. L. Sarkar, ‘‘The Superplasticizer Adsorption Capacity of Ce-ment Pastes, Pore Solution Composition, and Parameters Affecting Flow Loss,’’Cem. Concr. Res., 25 [7] 1423–34 (1995).

6V. Fernon, ‘‘Etude de nouveaux solides lamellaires obtenus par coprecipitationd’hydrate aluminocalcique et de sulfonates aromatiques’’; These de doctorat,Universite d’Orleans, 1994.

7V. Fernon, A. Vichot, N. L. Goanvic, P. Colombet, F. Corazza, and U. Costa,‘‘Interaction Between Portland Cement Hydrates and Polynaphtalene Sulfonates,’’SP 173 in Fifth CANMET/ACI Conference on Superplasticizers in Concrete, Editedby M. Malhotra. American Concrete Institute, Farmington Hills, MI, 1997.

8R. J. Flatt, ‘‘Dispersant in Concrete’’; pp. 247–94 in Polymers in ParticulateSystems: Properties and Applications, Vol. 9. Surfactant Science Series, Edited byV. A. Hackley, P. Somasundaran, and J. A. Lewis. Marcel Dekker, New York,2001.

9P. Coussot and A. Ancey, ‘‘Rheophysical Classification of Concentrated Sus-pensions and Granular Pastes,’’ Phys. Rev. E, 59 [4] 4445–57 (1999).

10N. Roussel, A. Lemaıtre, R. J. Flatt, and P. Coussot, ‘‘Steady State Flow ofFresh Cement Suspensions: A Micromechanical State of Art,’’ Cem. Concr. Res.,(2009), doi:10.1016/j.cemconres.2009.08.026.

11F. de Larrard, Concrete Mixture Proportioning: A Scientific Approach. E&FNSpon, London, 1999, 421pp.

12G. H. Tattersall and P. F. G. Banfill, Rheology of Fresh Concrete. Pitman,Boston, 1983, 356pp.

13R. J. Flatt and P. Bowen, ‘‘Yodel: A Yield Stress Model for Suspensions,’’ J.Am. Ceram. Soc., 89 [4] 1244–56 (2006).

14N. Roussel, M. R. Geiker, F. Dufour, L. N. Thrane, and P. Szabo, ‘‘Com-putational Modeling of Concrete Flow: General Overview,’’ Cem. Concr. Res., 37[9] 1298–307 (2007).

15N. Roussel, C. Stefani, and R. Leroy, ‘‘FromMini-Cone Test to Abrams ConeTest: Measurement of Cement-Based Materials Yield Stress Using Slump Tests,’’Cem. Concr. Res., 35 [5] 817–22 (2005).

16N. Roussel and P. Coussot, ‘‘‘Fifty-Cent Rheometer’ for Yield Stress Mea-surements: From Slump to Spreading Flow,’’ J. Rheol., 49 [3] 705–18 (2005).

17K. Yamada, S. Ogawa, and S. Hanehara, ‘‘Controlling of the Adsorption andDispersing Force of Polycarboxylate-Type Superplasticizer by Sulfate Ion Con-centration in Aqueous Phase,’’ Cem. Concr. Res., 31 [3] 375–83 (2001).

Fig. 13. Kinetics of polymer release from the composite after sulfateaddition to its mother suspension. The solid lines correspond to the totalamount of available positive charges (including excess Ca21) and thedotted line to the charges of the hydrocalumite layer only.

November 2009 Polycarboxylate Intercalation in Cementitious Systems 2485

18K. Yamada, T. Takahashi, S. Hanehara, and M. Matsuhisa, ‘‘Effects of theChemical Structure on the Properties of Polycarboxylate-Type Superplasticizer,’’Cem. Concr. Res., 30 [2] 197–207 (2000).

19R. J. Flatt, D. Larosa, and N. Roussel, ‘‘Linking Yield Stress Measurements:Spread Test Versus Viskomat,’’ Cem. Concr. Res., 36 [1] 99–109 (2006).

20R. J. Flatt and I. Schober, ‘‘Studying Admixtures with the 5c Rheometer’’; pp.93–104 in Proceeding of 8th CANMET/ACI International Conference of Super-plasticizers and Other Chemical Admixtures in Concrete, Edited by V. M. Ma-lhotra. American Concrete Institute, Farmington Hills, MI, 2006.

21J. Zimmermann, C. Hampel, C. Kurz, I. Shober, C. Plassard, E. Lesniewska,and R. J. Flatt, ‘‘Effect of Polymer Structure on the Sulphate–PolycarboxylateCompetition’’; in Proceeding of 9thCamnet/ACI Conference on Superplasticizersand Other Chemical Admixture in Concrete, Edited by V. M. Malhotra. AmericanConcrete Institute, Farmington Hills, MI, 2009.

22K. Yamada and S. Hanehara, ‘‘Interaction Mechanism of Cement and Su-perplasticizers—The Roles of Polymer Adsorption and Ionic Conditions of Aque-ous Phase,’’ Concr Sci Eng, 3, 135–45 (2001).

23R. J. Flatt, J. Zimmermann, C. Hampel, C. Kurz, I. Shober, C. Plassard, andE. Lesniewska, ‘‘The Role of Adsorption Energy in the Suphate–PolycarboxylateCompetition’’; in Proceeding of 9th CANMET/ACI Conference on Super-plasticizers and Other Chemical Admixtures in Concrete.

24R. J. Flatt and Y. F. Houst, ‘‘A Simplified View on Chemical Effects Per-turbing the Action of Superplasticizers,’’ Cem. Concr. Res., 31 [8] 1169–76 (2001).

25V. H. Dodson, Concrete Admixtures, pp. 42–5. Van Nostrand Reinhold, NewYork, 1990.

26V. S. Ramachandran, V. M. Malhotra, C. Jolicoeur, and N. Spiratos, Super-plasticizers: Properties and Applications in Concrete. Ministry of Public Works andGovernment Services, Ottawa, Ontario, Canada, 1998.

27J. Zhor, ‘‘Molecular Structure and Performance of Lignosulfonates in Ce-ment–Water Systems’’; PhD Thesis, University of New Brunswick, 2006.

28M. Piotte, ‘‘Caracterisation du Poly(Naphthalenesulfonate)—Influence de soncontre ion et de sa masse moleculaire sur son interaction avec le ciment’’; PhDThesis, Universite de Sherbrooke, 1993.

29J. A. Lewis, H. Matsuyama, G. Kirby, S. Morissette, and J. F. Young, ‘‘Poly-electrolyte Effects on the Rheological Properties of Concentrated Cement Sus-pensions,’’ J. Am. Ceram. Soc., 83 [8] 1905–13 (2000).

30R. J. Flatt, ‘‘Dispersion Forces in Cement Suspensions,’’ Cem. Concr. Res., 34[3] 399–408 (2004).

31R. J. Pellenq and H. Van Damme, ‘‘Why does Concrete Set: The Nature ofCohesion Forces in Hardened Cement-BasedMaterials,’’MRS Bull., 29 [5] 319–23(2004).

32M. Daimon and D. M. Roy, ‘‘Rheological Properties of Cement Mixes: I.Methods, Preliminary Experiments, and Adsorption Studies,’’ Cem. Concr. Res., 8[6] 753–64 (1978).

33M. Daimon and D. M. Roy, ‘‘Rheological Properties of Cement Mixes: II.Zeta Potential and Preliminary Viscosity Studies,’’ Cem. Concr. Res., 9 [1] 103–9(1979).

34P. F. G. Banfill, ‘‘A Discussion of the Papers ‘Rheological Properties of Ce-mentMixes’ byM. Daimon and D.M. Roy,’’ Cem. Concr. Res., 9 [6] 795–6 (1979).

35K. Yoshioka, E. Sakai, M. Daimon, and A. Kitahara, ‘‘Role of Steric Hin-drance in the Performance of Superplasticizers for Concrete,’’ J. Am. Ceram. Soc.,80 [10] 2667–71 (1997).

36A. Kauppi, K. M. Andersson, and L. Bergstrom, ‘‘Probing the Effect of Su-perplasticizer Adsorption on the Surface Forces Using the Colloidal Probe AFMTechnique,’’ Cem. Concr. Res., 35 [1] 133–40 (2005).

37A. M. Kjeldsen, R. J. Flatt, and L. Bergstrom, ‘‘Relating the MolecularStructure of Comb-Type Superplasticizers to the Compression Rheology of MgOSuspensions,’’ Cem. Concr. Res., 36 [7] 1231–9 (2006).

38R. J. Flatt, Y. F. Houst, P. Bowen, and H. Hofmann, ‘‘Electrosteric RepulsionInduced by Superplasticizers Between Cement Particles: An Overlooked Mecha-nism?’’; pp. 29–42 in Proceeding of the 6th CANMET/ACI International Confer-ence of Superplasticizers and Other Chemical Admixtures in Concrete, Edited by V.M. Malhotra. American Concrete Institute, Farmington Hills, MI, 2000.

39R. J. Flatt, Y. F. Houst, P. Bowen, H. Hofmann, J. Widmer, U. Sulser, U.Maeder, and T. A. Burge, ‘‘Interaction of Superplasticizers withModel Powders ina Highly Alkaline Medium’’; pp. 743–62 in Proceeding of the 5th CANMET/ACIInternational Conference of Superplasticizers and Other Chemical Admixtures inConcrete, Edited by V. M. Malhotra. American Concrete Institute, FarmingtonHills, MI, 1997.

40L. Regnaud, A. Nonat, S. Pourchet, B. Pellerin, P. Maitrasse, J. P. Perez, andS. Georges, ‘‘Changes in Cement Paste and Mortar Fluidity After Mixing Inducedby PCP: A Parametric Study’’; in Proceeding of the 8th CANMET/ACI Interna-tional Conference of Superplasticizers and Other Chemical Admixtures in Concrete,Edited by V. M. Malhotra. American Concrete Institute, Farmington Hills, MI,2006.

41V. Baroghel-Bouny, ‘‘Caracterisation des pates de ciment et des betons’’;These de doctorat, Ecole nationale des Ponts et Chaussees, 1994.

42H. F. W. Taylor, Cement Chemistry. Thomas Telford Publishing, London,1997, 169pp.

43P. Barret and D. Bertrandie, ‘‘Fundamental Hydration Kinetic Features of theMajor Cement Constituents—Ca3SiO5 and Beta-Ca2SiO4,’’ J. Chim. Phys.Phys.—Chim. Biol., 83 [11–12] 765–75 (1986).

44T. Matschei, B. Lothenbach, and F. P. Glasser, ‘‘The AFm Phase in PortlandCement,’’ Cem. Concr. Res., 37 [2] 118–30 (2007).

45G. Tzouvalas, N. Dermatas, and S. Tsimas, ‘‘Alternative Calcium Sulfate-Bearing Materials as Cement Retarders: Part I. Anhydrite,’’ Cem. Concr. Res., 34[11] 2113–8 (2004).

46H. Minard, S. Garrault, L. Regnaud, and A. Nonat, ‘‘Mechanisms and Pa-rameters Controlling the Tricalcium Aluminate Reactivity in the Presence of Gyp-sum,’’ Cem. Concr. Res., 37 [10] 1418–26 (2007).

47J. D. Birchall, A. J. Howard, and K. Kendall, ‘‘Flexural Strength and Porosityof Cements,’’ Nature (London), 289 [5796] 388–90 (1981).

48J. D. Birchall, A. J. Howard, K. Kendall, and J. H. Raistrick, ‘‘CementitiousCompositions and Cementitious Product of High Flexural Strength’’; US Patent4410366, Imperial Chemical Industries, 1983.

49K. Scrivener and A. Capmas, ‘‘Calcium Aluminate Cements’’; in Lea’s Chem-istry of Cement and Concrete, Edited by H. P. C. Arnold. Butterworth-Heinemann,Oxford, 1999.

50O. O. Popoola, W. M. Kriven, and J. F. Young, ‘‘Microstructural and Mi-crochemical Characterization of a Calcium Aluminate—Polymer Composite(MDF Cement),’’ J. Am. Ceram. Soc., 74 [8] 1928–33 (1991).

51J. A. Lewis and M. A. Boyer, ‘‘Effects of an Organotitanate Cross-LinkingAdditive on the Processing and Properties of Macro-Defect-Free Cement,’’ Adv.Cem. Based Mater., 2 [1] 2–7 (1995).

52G. K. D. Pushpalal, T. Kobayashi, T. Kawano, and N. Maeda, ‘‘The Pro-cessing, Properties, and Applications of Calcium Aluminate–Phenol Resin Com-posite,’’ Cem. Concr. Res., 29 [1] 121–32 (1999).

53M. Drabik, S. C. Mojumdar, and R. C. T. Slade, ‘‘Prospects of Novel Macro-Defect-Free Cements for the New Millenium,’’ Ceramics-Silkaty, 4 [2] 68–73(2002).

54G. Renaudin, ‘‘I/Etude d’un hydroxyde simple d’aluminium: la bayerite. II/Etude d’une famille d’hydroxydes doubles lamellaires d’aluminium et de calcium:les phases AFm (aluminates calciques hydrates)’’; These de doctorat, Universite deNancy I, 1998.

55X. Duan and D. G. Evans, ‘‘Layered Double Hydroxides’’; in Structure andBonding, Vol. 119, Edited by D. M. P. Mingos. Springer, Berlin, 2006.

56A. I. Khan and D. O’Hare, ‘‘Intercalation Chemistry of Layered Double Hy-droxides: Recent Developments and Applications,’’ J. Mater. Chem., 12 [11] 3191–8 (2002).

57F. Liebau, Structural Chemistry of Silicates—Structure, Bonding, and Classi-fication. Springer Verlag, Berlin, 1985.

58S. W. Bailey, ‘‘Hydrous Phyllosilicates’’; 725pp in Reviews in Mineralogy, Vol.19. Mineralogical Society of America, Washington, DC, 1988.

59R. J. M. Pellenq, N. Lequeux, and H. van Damme, ‘‘Engineering the BondingScheme in C–S–H: The Iono-Covalent Framework,’’ Cem. Concr. Res., 38 [2] 159–74 (2008).

60R. J. M. Pellenq, J. M. Caillol, and A. Delville, ‘‘Electrostatic Attraction Be-tween Two Charged Surfaces: A (N,V,T) Monte Carlo Simulation,’’ J. Phys.Chem. B, 101 [42] 8584–94 (1997).

61M. Meyn, K. Beneke, and G. Lagaly, ‘‘Anion-Exchange Reactions of LayeredDouble Hydroxides,’’ Inorg. Chem., 29 [26] 5201–7 (1990).

62B. Rebours, J.-B. d’Espinose de la Caillerie, and O. Clause, ‘‘Decoration ofNickel andMagnesium Oxide Crystallites with Spinel-Type Phases,’’ J. Am. Chem.Soc., 116 [5] 1707–17 (1994).

63V. Rives and M. A. Ulibarri, ‘‘Layered Double Hydroxides (LDH) Interca-lated with Metal Coordination Compounds and Oxometalates,’’ Coord. Chem.Rev., 181, 61–120 (1999).

64Z. P. Xu and P. S. Braterman, ‘‘Competitive Intercalation of Sulfonates intoLayered Double Hydroxides (LDHs): The Key Role of Hydrophobic Interac-tions,’’ J. Phys. Chem. C, 111 [10] 4021–6 (2007).

65L. Raki, J. J. Beaudoin, and L. Mitchell, ‘‘Layered Double Hydroxide-LikeMaterials: Nanocomposites for Use in Concrete,’’ Cem. Concr. Res., 34 [9] 1717–24 (2004).

66S. Carlino, ‘‘The Intercalation of Carboxylic Acids Into Layered Double Hy-droxides: A Critical Evaluation and Review of the DifferentMethods,’’ Solid StateIonics, 98 [1–2] 73–84 (1997).

67C. O. Oriakhi, I. V. Farr, and M. M. Lerner, ‘‘Incorporation of Poly(AcrylicAcid), Poly(Vinylsulfonate) and Poly(Styrenesulfonate) within Layered DoubleHydroxides,’’ J. Mater. Chem., 6 [1] 103–7 (1996).

68F. Leroux and C. Taviot-Gueho, ‘‘Fine Tuning Between Organic and Inor-ganic Host Structure: New Trends in Layered Double Hydroxide Hybrid Assem-blies,’’ J. Mater. Chem., 15 [35–36] 3628–42 (2005).

69F. Leroux, J. Gachon, and J.-P. Besse, ‘‘Biopolymer Immobilization Duringthe Crystalline Growth of Layered Double Hydroxide,’’ J. Solid State Chem., 177[1] 245–50 (2004).

70M. A. Thyveetil, P. V. Coveney, H. C. Greenwell, and J. L. Suter, ‘‘ComputerSimulation Study of the Structural Stability and Materials Properties of DNA-Intercalated Layered Double Hydroxides,’’ J. Am. Chem. Soc., 130 [14] 4742–56(2008).

71C. Taviot-Gueho and F. Leroux, ‘‘In Situ Polymerization and Intercalation ofPolymers in Layered Double Hydroxides’’; pp. 121–59 in Layered Double Hy-droxides, Vol. 119 Structure and Bonding. Springer-Verlag, Berlin, 2006.

72R. Ma, Z. Liu, L. Li, N. Iyi, and T. Sasaki, ‘‘Exfoliating Layered DoubleHydroxides in Formamide: A Method to Obtain Positively Charged Nanosheets,’’J. Mater. Chem., 16 [39] 3809–13 (2006).

73N. Iyi, Y. Ebina, and T. Sasaki, ‘‘Water-Swellable MgAlLDH (Layered Dou-ble Hydroxide) Hybrids: Synthesis, Characterization, and Film Preparation,’’ Lan-gmuir, 24 [10] 5591–8 (2008).

74M. A. Drezdzon, ‘‘Synthesis of Isopolymetalate-Pillared Hydrotalcite via Or-ganic-Anion-Pillared Precursors,’’ Inorg. Chem., 27 [25] 4628–32 (1988).

75N. T. Whilton, P. J. Vickers, and S. Mann, ‘‘Bioinorganic Clays: Synthesis andCharacterization of Amino- and Polyamino Acid Intercalated Layered DoubleHydroxides,’’ J. Mater. Chem., 7, 1623–9 (1997).

76E. D. Dimotakis and T. J. Pinnavaia, ‘‘New Route to Layered Double Hy-droxides Intercalated by Organic Anions Precursors to Polyoxometalate PillaredDerivatives,’’ Inorg. Chem., 29 [13] 2393–4 (1990).

77H. Tagaya, S. Sato, H. Morioka, J. Kadokawa, M. Karasu, and K. Chiba,‘‘Preferential Intercalation of Isomers of Naphthalenecarboxylate Ions intothe Interlayer of Layered Double Hydroxides,’’ Chem. Mater., 5 [10] 1431–3(1993).

2486 Journal of the American Ceramic Society—Giraudeau et al. Vol. 92, No. 11

78J. Plank, Z. Dai, and P. R. Andres, ‘‘Preparation and Characterization of NewCa–Al–Polycarboxylate Layered Double Hydroxides,’’ Mater. Lett., 60 [29–30]3614–7 (2006).

79J. Plank, H. Keller, P. R. Andres, and Z. M. Dai, ‘‘Novel Organo-MineralPhases Obtained by Intercalation of Maleic Anhydride–Allyl Ether CopolymersInto Layered Calcium Aluminum Hydrates,’’ Inorg. Chim. Acta, 359 [15] 4901–8(2006).

80L. Vieille, E. M. Moujahid, C. Taviot-Gueho, J. Cellier, J.-P. Besse, and F.Leroux, ‘‘In Situ Polymerization of Interleaved Monomers: A Comparative StudyBetween Hydrotalcite and Hydrocalumite Host Structures,’’ J. Phys. Chem. Solids,65 [2–3] 385–93 (2004).

81G. Renaudin, M. Francois, and O. Evrard, ‘‘Order and Disorder in the Lam-ellar Hydrated Tetracalcium Monocarboaluminate Compound,’’ Cem. Concr.Res., 29 [1] 63–9 (1999).

82J. Higgins and H. Benoit, Polymer and Neutron Scattering. Oxford Universitypress, Oxford, 1994.

83H. Schiessel and P. Pincus, ‘‘Counterion-Condensation-Induced Collapse ofHighly Charged Polyelectrolytes,’’ Macromolecules, 31 [22] 7953–9 (1998).

84B. Hammouda and D. L. Ho, ‘‘Insight Into Chain Dimensions in PEO/WaterSolutions,’’ J. Polym. Sci., Part B: Polym. Phys., 45 [16] 2196–200 (2007).

85R. Schweins, P. Lindner, and K. Huber, ‘‘Calcium Induced Shrinking ofNaPA Chains: A SANS Investigation of Single Chain Behavior,’’Macromolecules,36 [25] 9564–73 (2003).

86P. Borget, L. Galmiche, J.-F. Le Meins, and F. Lafuma, ‘‘MicrostructuralCharacterisation and Behaviour in Different Salt Solutions of Sodium Polymeth-acrylate-g-PEO Comb Copolymers,’’ Colloids Surf. A, 260 [1–3] 173–82 (2005).

87C. Gay and E. Raphael, ‘‘Comb-Like Polymers Inside Nanoscale Pores,’’ Adv.Colloid Interface Sci., 94 [1–3] 229–36 (2001).

88R. J. Flatt, I. Schober, E. Raphael, C. Plassard, and E. Lesniewska, ‘‘Con-formation of Adsorbed Comb Copolymer Dispersants,’’ Langmuir, 25 [2] 845–55(2008).

89A. Kauppi, P. F. G. Banfill, P. Bowen, L. Galmiche, Y. F. Houst, U. Mader,F. Perche, B. G. Petersen, K. Reknes, I. Shober, A. Siebold, and D. S. Swift,‘‘Improved Superplasticizers for High Performance Concrete’’; pp. 528–37 in

Proceeding of the 11th International Conference of Cement Chemistry, Durban,2003.

90E. Manias, H. Chen, R. Krishnamoorti, J. Genzer, E. J. Kramer, and E. P.Giannelis, ‘‘Intercalation Kinetics of Long Polymers in 2 nm Confinements,’’Macromolecules, 33 [21] 7955–66 (2000).

91T. Sakaue and E. Raphael, ‘‘Polymer Chains in Confined Spaces andFlow-Injection Problems: Some Remarks,’’ Macromolecules, 39 [7] 2621–8(2006).

92R. A. Vaia and E. P. Giannelis, ‘‘Lattice Model of Polymer Melt Intercalationin Organically-Modified Layered Silicates,’’ Macromolecules, 30 [25] 7990–9(1997).

93P.-G. de Gennes, ‘‘Flexible Polymers in Nanopores’’; pp. 91–105 in Polymersin Confined Environments, Edited by K. Binder. Springer Verlag, Berlin, 1999.

94C. P. Grey and A. J. Vega, ‘‘Determination of the Quadrupole Coupling-Constant of the Invisible Aluminum Spins in Zeolite HY with H1/AL27 TRAP-DOR NMR,’’ J. Am. Chem. Soc., 117 [31] 8232–42 (1995).

95T. Isobe, T. Watanabe, J.-B. d’Espinose de la Caillerie, A. P. Legrand, and D.Massiot, ‘‘Solid-State H-1 and Al-27 NMR Studies of Amorphous AluminumHydroxides,’’ J. Colloid Interface Sci., 261 [2] 320–4 (2003).

96N. A. Sanchez, J. M. Saniger, J.-B. d’Espinose de la Caillerie, A. L.Blumenfeld, and J. J. Fripiat, ‘‘Dealumination and Surface Fluorination ofH-ZSM-5 by Molecular Fluorine,’’ Microporous Mesoporous Mater., 50 [1] 41–52 (2001).

97E. Haddad, J. B. d’Espinose, A. Nossov, F. Guenneau, C. Mignon, and A.Gedeon, ‘‘Organic–Inorganic Phase Interaction in AlSBA-15 Mesoporous Solidsby Double Resonance NMR Spectroscopy’’; pp. 423–8 in Nanoporous MaterialsIII, Vol. 141. Studies in Surface Science and Catalysis, Edited by M. Jaroniec andA. Sayari. Elsevier, Amsterdam, 2002.

98O. J. Rojas, M. Ernstsson, R. D. Neuman, and P. M. Claesson, ‘‘Effect ofPolyelectrolyte Charge Density on the Adsorption and Desorption Behavior onMica,’’ Langmuir, 18 [5] 1604–12 (2002).

99C. Giraudeau-Lenain, ‘‘Interactions organo-aluminates dans les ciments. In-teraction de polymethacrylates-g-PEO dans l’hydrocalumite’’; These de Doctorat,Universite Pierre et Marie Curie, 2009. &

Claire Giraudeau received her PhDdegree in 2009 from the UniversitePierre et Marie Curie, Paris VI,where she also completed her under-graduate work. For her thesis sheworked with Dr. J.-B. d’Espinose inhis laboratory at ESPCI ParisTech,on the interactions between poly-mers and inorganic material. Speci-fically, her PhD work wasconcerned with the intercalation ofpolymer between hydrocalumite

layers. She is now a postdoctoral fellow studying the adsorptionof sulfate anions on cementitious materials at ESPCI and willbegin a postdoctoral fellowship at the Atomic Energy Authority(Commissariat a l’Energie Atomique, Saclay, France) in October2009, dealing with the synthesis of nanopowders to form high-density materials.

Jean-Baptiste d’Espinose de Lacaill-erie is an Assistant Professor at Par-isTech National Institute of Physicsand Chemistry (ESPCI ParisTech) inParis. He is a member of the PhysicalChemistry of Polymers and Dis-persed Media Laboratory (a jointlaboratory between ESPCI, the Uni-versity Pierre et Marie Curie, and theNational Centre for Scientific Re-search). He holds a geological engi-neering diploma and a MS in

Geochemistry from the National Institute of Geology (ENSGINPL) in Nancy, France (1987) and a PhD in Chemistry fromthe University of Wisconsin-Milwaukee (1992). He did his post-doctoral research at the French Petroleum Institute. He hascoauthored more than 60 publications in the area of catalysis,surface chemistry of oxides in aqueous solutions, and solid-stateNMR. His current research interests focus on the physics andchemistry of organo-mineral interactions at the nanoscale, thechemistry of cements substitutes for low carbon emission, andstray-field imaging of drying porous media.

Zied Souguir received a PhD inchemistry and polymer science in2006 from the University of Rouen,France. His PhD dissertation underDr. G. Muller focused on the che-mical modification of polysacchar-ides and the study of the chemicaland physico-chemical properties ofcolloidal systems. In 2007, he joinedthe Research Centre for the Conser-vation of Collections of the ParisNational History Museum for a

postdoctoral fellowship on the study of the degradation ofpaper at the wet–dry interface. Since 2008, he has been aNational Centre for Scientific Research postdoctoral fellow atthe Physical Chemistry of Polymers and Dispersed MediaLaboratory, working on hybrids nanoassemblies and, moreprecisely, on the formation of hybrid inorganic-polymers nano-composites and their stability.

Andre Nonat is Senior ResearchScientist appointed by the NationalCentre for Scientific Research(CNRS). He is currently workingat the University of Bourgogne atDijon (France), where he is incharge of a research department onthe reactivity and interface in mate-rials of the Institut Carnot de Bour-gogne, a joint laboratory betweenthe CNRS and the University ofBourgogne. He holds a PhD in Phy-

sical Chemistry from the University of Bourgogne (1981)and was awarded a ScD from the University of Nancy (1985).His main interest is the physical chemistry of cements and,particularly, the mechanisms of hydration and setting. Hehas published over 120 papers and is a coinventor on 4 patents.Andre Nonat is a Robert L’Hermite Medallist (1993), anaward from the International Union of Laboratories andExperts in Construction Materials, Systems and Structures(RILEM).

November 2009 Polycarboxylate Intercalation in Cementitious Systems 2487

Robert J. Flatt is the head of re-search on inorganic materials andprincipal scientist in Corporate Re-search at Sika Technology AG(Zurich, Switzerland) since 2002.Before that he was a postdoctoralfellow at Princeton University(1999–2002), obtained a PhD inmaterials science (1999) and a che-mical engineering diploma fromEPFL (1994). He has authoredmore than 60 publications in the

area of materials science of construction materials and conser-

vation of cultural heritage. Currently, his main topic of researchis the use of dispersing admixtures in concrete.

In 2003, he was awarded the RILEM Robert L’HermiteMedal for his contributions to the understanding of admixturesin cement and concrete. In 2007, he received the Ross C. Purdyaward from the American Ceramic Society for the most valuablecontribution to ceramic technical literature during the prioryear. He is a member of the cement division of the AmericanCeramic Society since 2000 and an alumni of the Young LeadersConference of the American-Swiss foundation. &

2488 Journal of the American Ceramic Society—Giraudeau et al. Vol. 92, No. 11