b04003

8/3/2019 b04003

1/6

The Rheology of Cementitious Materials

by

Robert J. Flatt

Sika Technology

Zurich, SWITZERLAND

Nicos S. Martys

Building and Fire Research Laboratory

National Institute of Standards and Technology

Gaithersburg, MD 20899 USA

and

Lennart BergstrmInstitute for Surface Chemistry, YKI

Stockholm, SWEDEN

Reprinted from MRS Bulletin, Materials Research Society, Vol. 29, No. 5, pp. 314-318,

May 2004.

NOTE: This paper is a contribution of the National Institute of Standards and

Technology and is not subject to copyright.

8/3/2019 b04003

2/6

314 MRS BULLETIN/MAY 2004

IntroductionWhen explaining what concrete is to

students, the analogy to baking a fruitcaketurns out to be quite useful. Both materialscontain solid inclusions (aggregates in con-crete, versus fruit or nuts in a fruitcake)and a binder (cement versus dough). In

both cases, the consistency of the battermay be improved by adding water, and asa result, the porosity of the final product isincreased. While in the case of fruitcake thisporosity may be an advantage, particu-larly if the recipe calls for soaking the cakewith brandy or kirsch after baking, it canprove catastrophic to the durability of con-crete. Indeed, it is through the porous net-work of the cement paste that chemicalagents can enter concrete and cause itsdegradation. Furthermore, increased poros-ity of the matrix significantly decreasesconcrete strength.

At first glance, this may suggest that thesolution for a more durable concrete is

simply to add as little water as possible.However, while the amount of water re-quired for complete hydration is2530%of the cement mass, about twice this amountis needed to achieve sufficient workability.This is where dispersants, referred to in thisfield as superplasticizers or high-rangewater-reducing admixtures (HRWRAs)come into play. Indeed, the introduction ofsuch admixtures in concrete has enhancedthe durability, increased the strength, andimproved the workability to levels that pre-viously were unattainable.

Benefits of Using HRWRAs inConcrete

In this section, the practical advantages tothe construction sector of using high-rangewater-reducing admixtures are outlined.The mechanisms by which these advan-tages are achieved in materials terms arediscussed in the following sections of thisarticle.

Self-Leveling / Self-CompactingConcrete

Nowadays, it is possible to optimize theproperties of fresh concrete with the use ofsuperplasticizers to combine a high flowa-

bility at very low additions of water withnegligible segregation of the particles. This

type of concrete, which is usually referredto either as self-compacting concrete (SCC)or self-leveling concrete (SLC), can be castinto a frame of reinforced steel without theneed for the labor-intensive vibration usu-ally associated with concrete placing. Theintroduction of SCC in the last few decadeshas enabled the development of new con-struction technologies. For example, thespeed of construction of the 238-m-highRoppongi Hills Mori Tower in Tokyo is il-lustrated in Figure 1. In this case, the build-ing structure consists of steel tubes, 2 m indiameter, which were filled with concreteto increase dimensional stability. The use

of SCC made it possible to cast sections ofup to 100 m in height without segregationof the aggregates due to gravity.

Ecological BenefitsSuperplasticizers are able to reduce the

porosity of the final material by allowingthe concrete to become workable with lesswater. This greatly enhances the durabil-ity of the concrete, which extends the lifecycle of the infrastructure in which it isused, thereby reducing the ecological im-pact of the construction sector.1 Further-more, superplasticizers also make itpossible to substitute substantial volumes

of cement with industrial-waste materialssuch as slag, fly ash, and silica fume,which reduces the CO2 emissions associ-ated with cement production.

Architectural BenefitsConcrete, while relatively inexpensive,

is unique as a building material because itcan be cast in a wide variety of shapes andsizes. Furthermore, because the use of su-perplasticizers improves workability andmechanical properties through water re-duction (see the article by Vernet in thisissue), architects can now exploit com-pletely new designs for elegant structureswith normal load-bearing capacities.

Need for Predictability andRobustness

The benefits mentioned here produce amarket drive for higher-performance con-crete with enhanced workability. However,as performance requirements rise, the ro-

bustness of the mix design becomes a morecritical issue. This brings with it a growingneed for predictability in concrete proper-ties and the a priori selection of concretecomponents as well as their proportioning,

The Rheology ofCementitiousMaterials

Robert J.Flatt, Nicos Martys,and Lennart Bergstrm

AbstractThe introduction of a new generation of dispersants in concrete allow this material to

exhibit self-compacting properties in its fresh state and high durability and mechanical

strength in its hardened state.These properties translate into many practical advantages

for the construction field.Two of the most important are reducing the ecological impact of

this sector of industry and reducing the labor-intensive work associated with placing

ordinary concrete by vibration. In this article, it will be shown that knowledge of colloidal

science has proven essential in the development of this new generation of dispersants

for concrete. Indeed, the polymer molecules used in these dispersants are specifically

designed to induce steric repulsion between cement particles, reducing their

agglomeration and allowing high workability of fresh concrete prior to setting. While the

linkage between interparticle forces and the rheological behavior of cement pastes is still

only semiquantitative, recent advances in the modeling of concrete rheology show very

promising results in terms of handling aggregates with a wide distribution of particle

sizes and shapes. However, accurate modeling requires reliable input on the interaction

of the dispersant with the hydrating cement at the molecular level, which is identified as

a future research challenge.

Keywords:cement, colloids, construction materials, rheology, aggregates.

www.mrs.org/publications/bulletin

8/3/2019 b04003

3/6


MRS BULLETIN/MAY 2004 315

in order to match the requirements of thedesign while minimizing sensitivity to vari-ations in materials supply.

Concrete Rheology and the Roleof SuperplasticizersNature of Superplasticizers

The original introduction of superplas-ticizers into concrete was accidental. Car-

bon black had been added to concrete tochange the color of the center line of athree-lane highway. In doing so, the concretehad poor workability, and a dispersantwas introduced to counter the effect of thecarbon black. The resulting hardened con-

crete showed properties that indicatedthat the cement had been positively af-fected by the dispersant;2 the workersprobably had added less water because ofthe enhanced dispersion, and thus higherstrength was obtained in the hardened ma-terial. It is now common practice to usethese additives to improve the flowabilityand extend the working time of fresh con-crete.3,4 The use of the first generation ofsuperplasticizersfor example, sulfonatednaphthalene formaldehyde (SNF) andmodified lignosulfonates (LSs)resultedin significant improvements in the proper-ties of fresh concrete, and they are stillwidely used. However, increasing de-mands for better flowability, extendedworking time, and a reduction in concreteporosity have created a need for superplas-ticizers with improved performance.

Dispersion MechanismsSuperplasticizers adsorb at the solid

liquid interface between the particles andthe aqueous phase. There, they impart arepulsive interparticle force, thus reduc-ing or eliminating adhesion between par-ticles in close proximity.4 The attractive

forces between the cement grains, silicafume, fly ash, and slag particles may origi-nate from van der Waals5 or electrostaticforces (ion correlation, see the article byPellenq and Van Damme in this issue), orfrom surface-charge inhomogeneities.

The term electrosteric stabilization isoften used to describe how superplasticiz-ers act as dispersants. Electrosteric stabi-lization is a combination of an electrostaticdouble-layer repulsion and a steric repul-sion, where the relative importance of therespective contributions is closely related tothe polymer segment density profile at theinterface, the charge density of the poly-

mer, and the ionic strength of the solution.To prevent particles from coming intoclose proximity with each other, this forcemust be sufficiently long-range. In cemen-titious systems, where the ionic strength ishigh (0.1 mol/L), electrostatics alone donot suffice. Thus, it is expected that extend-ing the layer thickness of adsorbed super-plasticizers should improve their rheologicalperformance in such systems.6,7 This may

be achieved with comb-type superplasti-cizers in which adsorption is driven by theionic content of the backbone and thesteric layer is enhanced by grafted non-adsorbing side chains that extend into thesolution (Figure 2).

AFM for Probing Properties ofAdsorbed Polymers

From the previous discussion, it is clearthat an important factor in the dispersionefficiency of superplasticizers is their ad-sorbed conformation. While computationalmethods such as molecular dynamics mayprovide insight into such phenomena inthe long term, it is currently more practi-cal to measure the interaction betweensuperplasticizer-coated surfaces directly

in relevant solution conditions. The atomicforce microscopy (AFM) method for meas-uring surface forces was first used byDucker et al.8 They attached a sphericalparticle at the tip of the cantilever and ob-tained a force-displacement curve fromthe deflection of the cantilever as a func-

tion of intersurface separation. This versa-tile method was recently extended tospherical MgO particles,9,10 which have asimilar surface chemistry as cement buthave the advantage of being much less re-active when subjected to water.11

Figure 3 shows the results from direct-force measurements between a sphericalMgO probe attached to the cantilever anda flat MgO substrate immersed in an aque-ous media at pH 10.12 It was found that theinteraction is repulsive in a simple mono-valent electrolyte (KCl), which probablycan be related to the positive charge on theMgO surfaces (Figure 3a) and low attrac-

tive van der Waals forces. However, theaddition of calcium resulted in an attraction between the surfaces that may originatefrom ion correlation forces, as described inthe article by Pellenq and Van Damme inthis issue. The addition of a comb-likecopolymer having a negatively charged

backbone with grafted poly(ethyleneoxide) (PEO) chains of a relatively shortlength (PCP1) resulted in a stronger repul-sion between the surfaces (Figure 3b). Noattraction was observed, even in a calcium-rich electrolyte.

Addition of a comb-like copolymerhaving an identical backbone to PCP1 but

with PEO side chains of a much longerlength (PCP4) results in a interparticle re-pulsion that is much more long-range (Fig-ure 3c). In addition, the effect of thispolymer is influenced relatively little by thenature of the polyelectrolyte. This indi-cates that the steric repulsion is dominat-ing and we can even get an indication of thethickness of the adsorbed layer on the sur-face. These results are currently being usedin the design of superplasticizers with anoptimal structure and can also be used forrealistic estimates of the interparticle forcesin larger-scale simulations of the rheologyof particulate suspensions, as described inthe section on Interparticle Interactions/Yield Stress.

The Role of Particle SizeDistribution

While the addition of superplasticizersgreatly improves concrete rheology throughdispersion of the finer particles, the granularnature of concrete must not be overlooked.13

Concrete contains particles of different types,spanning several orders of magnitude issize (50 nm for silica fume to50 mmfor the largest aggregates). Thus, rheology

Figure 1. Series of photographs showing the speed of construction of the 238-m-highRoppongi Hills Mori Tower in Tokyo (completed in 2003). The use of self-compactingconcrete made it possible to cast sections of up to 100 m in height without segregationof the aggregates due to gravity. (Photographs courtesy of M. Danzinger; personalcommunication).

8/3/2019 b04003

4/6



may be improved by better grading andproportioning of the components.14,15

IncompatibilitiesWhile dispersion mechanisms appear

well understood in general terms, many

details remain unresolved, in particular,cementsuperplasticizer incompatibility.3

This term refers to specific combinationsof cement and superplasticizers that showpoor rheological properties unless exces-sive amounts of polymer are added, or for

which the duration of the dispersion effectis extremely short.

These incompatibilities arise from thereactivity of cement and in particular fromthe calcium aluminate phases, whichhave the strongest initial reaction in thepresence of water. The influence of the

cement chemistry on superplasticizerperformance is well established4,16 andis attributed in part to the intercalation ofthe superplasticizers into hydrationproducts.17 The intercalated polymer islost for dispersion purposes and this de-creases efficiency.18 In other cases, theefficiency of some polycarboxylates can

be lowered because of competitive ad-sorption from sulfate ions.19

ModelingA detailed simulation of concrete rheol-

ogy, accounting for the motion of the ce-ment, sand, and aggregates, is impossible

on present-day computers. To solve thecomputational problem of the broad spanof particle sizes in concrete, a multiscale orhomogenization approach is used. Phe-nomena are modeled at a characteristiclength scale to determine an averageproperty, which is then used as an input ina simulation at a coarser scale. For ex-ample, if the viscosity of a mortar can bedetermined by simulation or experiment,that viscosity can be used as an input todetermine the viscosity of concrete.

There are problems with this approach,mainly linked to agglomeration.20 How-ever, at sufficiently high shear rates, the

evolution of an agglomerating system issimilar to that of a non-agglomerating sys-tem. For cementitious materials that ex-hibit a Bingham behavior (a linear relation

between shear stress and shear rate), plas-tic viscosity is the most relevant rheologi-cal parameter, and the slope ofexperimental flow curves can be com-pared to those of simulations.

Suspensions Model Based onDissipative Particle Dynamics

While some analytical solutions describ-ing the rheological properties of simplesuspensions exist (e.g., for very dilute sus-pensions), understanding the flow of morecomplex suspensions like cement-based ma-terialsdense suspensions and suspensionscomposed of particles with different shapes orparticles that interactremains a challenge. Amajor difficulty in modeling complex fluidslike suspensions is the tracking of boundaries

between the fluid and solid phases. Recently,a promising new computational methodcalled dissipative particle dynamics (DPD)21

has been developed for modeling complexfluid systems. Indeed, DPD may have advan-tages over other computational fluid

Figure 2. Schematic illustration of the force between two surfaces with adsorbed comb-like

copolymers, as a function of separation distance. Steric repulsion prevents van der Waalsinteractions from developing strong attractive forces (negative values).Schematicillustrations (top) show surfaces (from right to left) before, at the onset, and during theoverlap of their adsorbed layers, leading to steric hindrance that overcomes the van derWaals force.The dashed line in the graph is for a thinner adsorbed layer, which is shownbefore overlap-upon-layer-overlap in the lower illustration.The structure inset at lower rightgives the generic composition of such comb-type copolymers.

Figure 3. Atomic force microscopy colloidal probe results between MgO surfaces at pH 10:(a) without superplasticizer, (b) in the presence of PCP1 [short side chains of poly(ethyleneoxide), and (c) in the presence of PCP4 (longer side chains of PEO). The errors on suchmeasurements are about 5N/m.13

8/3/2019 b04003

5/6


MRS BULLETIN/MAY 2004 317

dynamics methods because it can describemoving boundaries without requiring re-gridding of the computational domain.22

On the surface, DPD looks similar to amolecular dynamics algorithm,23 whereparticles, subject to interatomic forces,move according to Newtons laws. How-

ever, the particles in DPD are not atomisticbut instead are a mesoscopic representationof the suspension. The interactions be-tween the particles are described by threeclasses of forces: conservative, dissipative,and random. The conservative force is acentral force, derivable from some poten-tial. The dissipative force is proportionalto the difference in velocity between par-ticles and acts to slow down their relativemotion, producing a viscous effect. Therandom force (usually based on a Gauss-ian random noise) helps reproduce thetemperature of the system while produc-ing a viscous effect. Finally, it has been

shown that DPD equations can accountfor hydrodynamic behavior consistentwith the NavierStokes equations.24,25

To model a rigid-body inclusion in afluid, a subset of the DPD particles are ini-tially assigned a location in space suchthat they approximate the shape of the ob-

ject.26 The motion of these particles is thenconstrained so that their relative positionsnever change. The total force and torqueare determined from the DPD particle in-teractions, and the rigid body moves ac-cording to the Euler equations.

Cement-based materials are usuallycomposed of particles with a broad shape

and size distribution. Figure 4 shows sometypical examples. Figure 4a is a system ofpolydisperse spheres that could corre-spond to a concrete composed of riverbedaggregates, which are usually rounder andsmoother than most aggregates. Figure 4bis based on realistic images of aggregatesacquired by x-ray microtomography of acrushed aggregate. The tomographic im-ages of aggregates can be analyzed fortheir geometrical properties by construct-ing a spherical harmonic representation oftheir shape.27 Once the aggregate imagesare incorporated into the code, we can de-termine the viscosity of the total systemrelative to the matrix fluid viscosity for agiven shear rate.

So far, we have found good agreementwith experimental studies of the plasticviscosity of fresh concrete having an ag-gregate composition similar to that usedin the simulations28 (Figure 5).

Interparticle Interactions/YieldStress

We are currently investigating the incor-poration of interparticle interactions thataccount for agglomeration of cement par-

ticles. For example, by including an ap-proximation to the attractive van der Waalsforce expected in cementitious systems,5 itis possible to produce a Bingham-like be-havior in our suspension. To model thesteric hindrance of dense layers, where in-terpenetration is low before the van derWaals force is counteracted, a cutoff dis-tance is introduced at close separation.While more refined models need to be de-veloped, the results of these preliminary

investigations are encouraging. Ultimately,we expect to be able to use conformationaldata of adsorbed superplasticizers fromAFM within such predictions.

ConclusionsIn this article, we have highlighted two

key aspects of concrete rheology: (1) the ag-glomeration of the finer particles and therole of dispersants to counter this agglomer-ation and (2) the granular nature of concreteand the role of particle size distributions

on the rheological properties of concrete.Mastering both aspects has led to signifi-cant progress in concrete technology, rais-ing expectations for concrete propertiesand highlighting questions of robustness.Consequently, there is an increased needfor predicting concrete properties andtheir variation.

In this context, modeling can play a cru-cial role. The work presented showspromising results in its ability to accountfor the three-dimensional shapes of par-ticles. While good predictions of plasticviscosity have been obtained, more workis needed to accurately measure the role ofyield stress. Ultimately, the model shoulduse structural information about the su-perplasticizer molecule to properly evalu-ate interparticle forces. Such tools will notonly allow better use of existing materials

but will also help admixture producers de-sign superplasticizers with better per-formance and enhanced robustness.

The utility and predictive capability of themodels will improve with further advancesin our ability to quantify interparticleforces and formulate multiscale models. Inthe long term, to properly describe cement

Figure 4. Example of concrete systems: (a) system of polydisperse spheres, correspondingto a concrete composed of riverbed aggregates, which are usually rounder and smootherthan most aggregates; (b) model based on three-dimensional images of realistically shaped

particles, acquired by x-ray microtomography of crushed aggregates.

Figure 5. Comparison of relative

plastic-viscosity values from asimulation of coarse aggregate gradingsand experimental measurements usingdifferent concrete rheometers. In thefigure, Grad. #1, #2, and #3 correspondto gradations of spherical aggregatesused in the computer simulation. BML isa coaxial concrete rheometer, IBB is avane concrete rheometer, and beads inpaste corresponds to measurements ofmonosized glass beads in a cementpaste using a parallel plate rheometer.The solid line is included as a guide forthe eye.28

8/3/2019 b04003

6/6



rheology, the inclusion of other features suchas the reactivity of cement and the pertur-

bation of dispersant efficiency though inter-calation or adsorption competition fromother species needs to be taken into account.

Acknowledgments

The authors thank colleagues in theirrespective institutions for support in thefundamental investigations in this fieldand for fruitful discussions and com-ments, in particular, at Sika Technology,Dr. Norman Blank, Dr. Irene Schober, andDr. Urs Mder; at NIST, Dr. ChiaraFerraris, Dr. Edward Garboczi, Dr. SteveSatterfield, and Dr. Terrance Griffins; andat YKI, Dr. Anika Kauppi. N. Martys alsoacknowledges support from VCCTL (theVirtual Cement and Concrete Testing Lab-oratory, NIST) for modeling work. AFMresults presented here were obtainedwith EU support under the project Super-plast (Project of the 5th European Frame-work Programme, G5RD-CT-2001-00435).L. Bergstrm and R. Flatt are also gratefulto all of their colleagues within that projectfor many fruitful discussions.

References1. P.K Mehta, in Proc. of the 6th CANMET/ACIInternational Conference on Fly-Ash, Silica Fume,Slag and Natural Pozzolans in Concrete, Vol. 1, ed-ited by V.M. Malhotra (American Concrete In-stitute, Detroit, 1998) p. 1.2. V.H. Dodson, Concrete Admixtures (Van Nos-trand Reinhold, New York, 1990).

3. P.C. Atcin, C. Jolicoeur, and J.G. MacGregor,Concr. Int. 16 (5) (1994) p. 45.4. V.S. Ramachandran, V.M. Malhotra, C.

Jolicoeur, and N. Spiratos,Superplasticizers: Prop-erties and Applications in Concrete,CANMET Publication MTL 97-14 (CANMET,Ottawa, 1998).5. R.J. Flatt, Cem. Concr. Res. 34 (2004) p. 399.6. R.J. Flatt, in Polymers in Particulate Systems:Properties and Applications, edited by V.A.Hackley, P. Somasundaran, and J.A. Lewis(Marcel Dekker, New York, 2001) p. 247.7. J.A. Lewis, H. Matsuyama, G. Kirby, S.Morissette, and J.F. Young,J. Am. Ceram. Soc. 83(2000) p. 1905.8. W.A. Ducker, T.J. Senden, and R.M. Pashley,Nature 353 (1991) p. 239.

9. A. Kauppi, P.F.B. Banfill, P. Bowen, L.Galmiche, Y.F. Houst, F. Lafuma, U. Mder, F.Perche, B.G. Petersen, K. Reknes, I. Schober,A. Siebold, and D. Swift, in Proc. 11th Int. Congr.on the Chemistry of Cement, Vol. 2, edited by

G. Grieve and G. Owen (Cement Association ofCanada, Ottawa, 2003) p. 528.10. A. Kauppi, K.M. Andersson, and L.Bergstrm, Probing the Effect of Superplasti-cizer Adsorption on the Surface Forces Usingthe Colloidal Probe AFM Technique, Cem.Concr. Res. submitted for publication.11. R.J. Flatt, Y.F. Houst, P. Bowen, H. Hofmann,

J. Widmer, U. Sulser, U. Maeder, and T.A.Brge, in Proc. 5th CANMET/ACI Int. Conf.on Superplasticizers and Other Chemical Admix-tures in Concrete, edited by V.M. Malhotra(American Concrete Institute, Detroit, 1997)p. 743.12. H.G. Pedersen, Particle interactions: AnAFM study of colloidal systems, PhD thesis,Technical University of Denmark, 1998.13. H. Van Damme, S. Mansoutre, P. Colombet,C. Lesaffre, and D. Picart, C.R. Physique 3 (2002)p. 229.14. T.C. Powers, The Properties of Fresh Concrete(John Wiley & Sons, New York, 1968).15. F. de Larrard, Concrete Mixture Proportioning:

A Scientific Approach (E&FN Spon, London, 1999).16. D. Bonen and S.L. Sarkar, Cem. Concr. Res.

25 (1995) p. 1423.17. V. Fernon, A. Vichot, N. Le Goanvic, P.Colombet, F. Corazza, and U. Costa, in Proc. 5thCANMET/ACI Int. Conf. on Superplasticizers andOther Chemical Admixtures in Concrete, edited byV.M. Malhotra (American Concrete Institute,Detroit, 1997) p. 225.18. R.J. Flatt and Y.F. Houst, Cem. Concr. Res. 31(2001) p. 1169.19. K. Yamada and S. Hanehara, Concr. Sci. andEng. 3 (2001) p. 135.20. R.J. Flatt, Towards a Prediction of Super-plasticized Concrete Rheology, Mater. andStruct. in press.21. P.J. Hoogerbrugge and J.M.V.A. Koelman,Europhys. Lett. 19 (1992) p. 155.22. R. Peyret and D. Taylor, Computational

Methods for Fluid Flow (Springer-Verlag, NewYork, 1983).23. M.P. Allen and D.J. Tildesley, Computer Simu-lation of Liquids (Clarendon Press, Oxford, 1987).24. P. Espaol and P. Warren, Europhys. Lett. 30(1995) p. 191.25. C. Marsh, G. Backx, and M.H. Ernst, Euro-

phys. Lett. 38 (1997) p. 441.26. J.M.V.A. Koelman and P.J. Hoogerbrugge,Europhys. Lett. 21 (1993) p. 363.27. E.J. Garboczi, Cem. Conc. Res. 32 (2002)p. 1621.28. C. Ferraris and N. Martys,J. Res. Natl. Inst.Stand. Technol. 108 (2003) p. 229.

Documents

b04003