Cement and Concrete Research 79 (2016) 291–300

Contents lists available at ScienceDirect

Cement and Concrete Research

j ourna l homepage: ht tp : / /ees .e lsev ie r .com/CEMCON/defau l t .asp

In-situmeasurement of viscoelastic properties of fresh cement paste by amicrorheology analyzer

Yanrong Zhang a,b, Xiangming Kong a,⁎, Liang Gao b, Zichen Lu a, Shiming Zhou c, Biqin Dong d, Feng Xing d

a Department of Civil Engineering, Tsinghua University, Beijing, 100084, Chinab Beijing Key Laboratory of Track Engineering, School of Civil Engineering, Beijing Jiaotong University, Beijing, 100044, Chinac SINOPEC Research Institute of Petroleum Engineering, Beijing, 100101, Chinad Department of Civil Engineering, Guangdong Provincial Key Laboratory of Durability for Marine Civil Engineering, Shenzhen University, Shenzhen, 518060, China

⁎ Corresponding author. Tel.: +86 10 62783703.E-mail address: kxm@mail.tsinghua.edu.cn (X. Kong).

http://dx.doi.org/10.1016/j.cemconres.2015.09.0200008-8846/© 2015 Elsevier Ltd. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 3 May 2015Accepted 30 September 2015Available online 23 October 2015

Keywords:Viscoelastic propertiesFresh cement pasteSuperplasticizerMicrorheologyNon-disturbing measurement

A microrheology analyzer was adapted to in-situ follow the development of viscoelastic properties of fresh ce-ment pastes (FCPs) for the first time. It enables a non-disturbing measurement on the FCPs through monitoringthe mean square displacement of cement particles, which gives an insight into the elastic and viscous propertiesofmaterials from amicrostructural point of view. Various parameters including elastic index,macroscopic viscos-ity index, storage modulus, loss modulus and Maxwell parameters were obtained to quantitatively analyze theviscoelastic properties of FCPs. Results indicate that these parameters show a progressive increase with time atfirst and then stay stable. The incorporation of superplasticizer significantly decreases these parameters andtheir growth rates. Moreover, superplasticizer could evidently weaken the elastic feature of the FCP due to its ef-fects of improving the dispersion of cement grains and retarding cement hydration. The effects of superplasticizerare more pronounced at lower water to cement ratio.

© 2015 Elsevier Ltd. All rights reserved.

1. Introduction

Fresh cement paste (FCP) is considered as a viscoelasticmaterial, be-cause it responds to external forces in amanner intermediately betweenan elastic solid and a viscous fluid [1,2]. Viscoelastic properties not onlyplay important roles in affecting the fluidity, consistence and workabil-ity of FCP, but also have effects on the volume stability of hardenedcement paste (HCP) [3]. Generally speaking, the viscoelastic behaviorof FCP is primarily determined by the physical condition of the solidcementitious particles in the suspension system which is related tothe dispersion of the particles, the hydration process of cement andthe addition of chemical admixtures [3]. Therefore, from an academicperspective, monitoring the development of viscoelastic properties ofFCP is beneficial to explore the microstructural development resultingfrom the cement hydration and from the particle interactions in cementpaste, and to further clarify the working mechanisms of chemicaladmixtures on the properties of FCPs.

Currently, various techniques have been employed to investigate theviscoelastic properties of FCPs, such as dynamic rheometer [1,2,4–7],electrical or ultrasonic reflection [3,8–20]. However, many of thosemethods have their intrinsic limitations and drawbacks. Special atten-tion has been paid to oscillatory shear tests (dynamic rheometer) inlow frequency by several researchers to study the rheology of cement

paste [1,2,4–7]. On the premise of limiting the value of oscillatoryshear strain and the frequency within the linear viscoelastic region ofthe material, the elastic and viscous behaviors of cement paste can becharacterized by directly measuring the loss and storage moduli usingthis technique. In addition, thismethod is capable of providing useful in-formation concerning structure or inter-particle forces [5]. However, inmost cases, to ensure that the measurements are performed within thelinear viscoelastic region is usually not an easy task. Electrical measure-ments barely establish any direct relationship with the mechanicalproperties despite the resistivity and capacitance of FCPs during cementhydration were recorded [8–10]. By monitoring shear wave reflected atnormal incidence from an interface between a buffer material and thetargeted material, ultrasonic reflection techniques were used to mea-sure the viscoelastic properties of cement paste [11–19]. Nevertheless,the accuracy of this technique in the application of cement paste atthe very early stage has remained a challenge [20].

In this study, a microrheology analyzer (Rheolaser LAB6™) wasadapted to in-situ follow the development of the viscoelastic propertiesof FCPs in the presence of superplasticizer during the early hours aftermixing. This method is based on the measurement of the mean squaredisplacement of cement grains by laser scattering, which gives an in-sight into the elastic and viscous properties of the suspension systemfromamicrostructural point of view. Various parameters including elas-tic index (EI), macroscopic viscosity index (MVI), storage modulus G′,loss modulus G″, and Maxwell parameters, the viscosity of dashpot ηand the elastic modulus of spring G, were extracted to quantitatively

292 Y. Zhang et al. / Cement and Concrete Research 79 (2016) 291–300

analyze the viscoelastic properties. On the basis of the results obtainedfrom this technique, insightful information about cement hydrationand the functions of superplasticizer could be also acquired.

2. Theory background

2.1. Microrheology

Microrheology is a new domain of rheology to study the viscoelasticbehavior of a multi-phase system such as emulsion, suspension, gel orcolloidal dispersion at a micron length scale [21]. It refers to measurethe local deformation of a sample originated from an applied stress orthermal energy, which is directly correlated to the elastic and viscousproperties of materials [22]. In this study, measurement of fresh cementpastes was performed using a microrheology analyzer that measuresthe displacement of particles based on diffusing wave spectroscopy(DWS) [23]. This technology uses a multi speckle diffusing wavespectroscopy (MS-DWS) set-up in a backscattering configuration withvideo camera detection. The scheme of the apparatus is depicted inFig. 1 and the principle of DWS measurement is shown in Fig. 2.Unlike the traditional rheology testing methods in which a shear forceand shear strain are usually involved, this method enables a non-disturbing measurement without any disturbance to the sample, andthe same sample can be continuously monitored versus elapsed time.After the sample is placed in a cell, a fixed coherent laser beam (wave-length 658 nm) is incident upon the sample which contains scaterrers(cement grains here). The laser is multiply scattered many times bythe particles into the sample, which leads to the interfering backscatter-ingwaves. An interference image called a “speckle image” is detected bya multi-pixel detector. In dynamic mode, particle motion induces spotmovements of the speckle image [24]. The particle mobility in termsof speed and displacement is closely related with the viscoelastic prop-erties of the whole system, so is the deformation of the speckle image(Fig. 2). Fast motion of particles results in fast deformation of speckleimage and slow motion of the particles leads to a slow deformation ofthe speckle image. Based on the speckle image, a patented algorithmwas used to quantitatively characterize the deformation rate of thespeckle image and further plot the curve of the mean square displace-ment (MSD) of the scatterers versus decorrelation time tdec [25].

For a singlemeasurement of a stable colloid system, the decorrelationtime tdec is themeasurement time, which is used to follow the change ofthe speckle image [25]. From the MSD curve with respect to thedecorrelation time, one can calculate the viscoelastic moduli G′ and G″using the generalized Stokes-Einstein relation [26]:

~G sð Þ ¼ kBT

πas Δ~r2 sð ÞD E ¼ G0 þ iG″ ð1Þ

where kB is Boltzmann constant, T is temperature in kelvin, s is theLaplace frequency,which is proportional to 1/tdec, a represents the radius

of the tracer, and hΔ~r2ðsÞi denotes the Laplace transform of the MSD.Sampleswith differentmicrostructures possess different viscoelastic

properties, thereby presenting varied MSD curves. From the shapes ofMSD curves as shown in Fig. 3, viscoelastic properties of samplescould be analyzed qualitatively [21]. In the case of a purely viscous

Fig. 1. Schematic representation of the experimental microrheometer.

sample, the MSD grows linearly with the decorrelation time as the par-ticles are completely free to move in the sample and the slope of theMSD curve is associated with the viscosity of the sample. With respectto a viscoelastic sample, particles in the sample are not free to movebut constrained in a “cage” or “network structure” formed by the neigh-boring particles. Smaller size of “cage” or “network structure” bringsabout a stronger constraining effect, which is indicated by themore pro-nounced elasticity of the sample. Overall, theMSD curve of a viscoelasticsample could be divided into three periods with respect to thedecorrelation time. At the very initial decorrelation time, the particlesare free to move in the continuous medium phase, so the MSD curvedevelops linearly and the slope is mostly related to the viscosity of thedispersant medium. Then, they are blocked by their neighbors, andthe slope ofMSD curve decreases and finally theMSD reaches a plateau.This is a characteristic of the elasticity of the sample. A lower plateaumeans a “cage” with smaller size and stronger elasticity. Thus, theheight of the plateau characterizes the elastic modulus of the sample.At longer decorrelation time, the particles are able to find a way toescape from the “cage” and the MSD grows linearly again, which is acharacteristic of the macroscopic viscosity as it corresponds to themoving speed of the particles in the sample. The longer time neededby the particles to finish a displacement implies the lower particlemobility and the higher macroscopic viscosity.

The following parameters extracted from MSD curves enable tocharacterize the viscoelastic properties of samples [21].

– Elasticity index (EI) is computed from the elastic plateau value,which corresponds to the inverse of the height of the MSD plateau.

– Macroscopic viscosity index (MVI) is a global computation and cor-responds to a viscosity index at zero shear rate, which is the inverseof the slope of MSD curve in later linear scale.

– The storage modulus G′, which represents the elastic behavior orthe energy storage of the material, and the loss modulus G″,which signifies the viscous behavior or energy dissipation of thematerial, can be calculated using the generalized Stokes–Einsteinrelation [26]. For an elastic solid, the storage modulus dominatesin the material and the loss modulus is low. For a viscous liquid,the loss modulus dominates. Thus, a material can be readily iden-tified as an elastic solid or a viscous liquid by comparing thevalues of G′ and G″.

– When the viscoelastic properties of the sample are described bythe Maxwell model, the elastic modulus of a Hookean spring Gand the viscosity of a Newtonian dashpot η could be obtained.

2.2. Microstructure of fresh cement paste

Aswell known, FCP is a reactive solid–liquid dispersion system. Afterthe contact of cement with water, various ions are quickly dissolvedfrom the mineral phases of cement grains into the aqueous phase, con-sequently developing a heterogeneous charge distribution on the sur-face of hydrating cement grains. This mosaic structure results in theformation of flocculated cement grains that entrap a large quantity ofmixing water. Fig. 4(a) presents the schematic illustration of the initial“cage” or “network structure” in FCP. In the first minutes of cement–water contact, massive needle-like ettringite crystals are producedand cause a bridging effect among the cement grains, as shown inFig. 4(b). With the progressing cement hydration, more newly formedhydrates are produced which create new links among cement grainsand strengthen the inter-particle crosslinking, thus leading to a stronger“network structure”, as shown in Fig. 4(c). Because of the change of themicrostructure of FCP, the rheological properties of FCP continuallyevolve over hydration time.

At higher W/C, the distance between the cement grains is larger,namely the size of “cage” is larger. Thus, the “network structure” ofFCP is relatively weaker during the cement hydration, as illustrated inFig. 4(d)–(f).

Fig. 2. The principle of the multi speckle diffusing wave spectroscopy (MS-DWS) measurement and the relationship between the motion rate and the deformation rate [21].

293Y. Zhang et al. / Cement and Concrete Research 79 (2016) 291–300

The addition of superplasticizer leads to disassembly of the flocculat-ed structures and thus release of the entrapped water, by inducingsteric and/or electrostatic repulsion between cement grains upon ad-sorption of superplasticizer molecules on surface of cement grains. Inthis way, a better dispersion of cement grains is achieved by theaddition of superplasticizers (Fig. 4(h)) [27–29], which is the so-calleddispersing effect of superplasticizer. Moreover, the incorporation ofsuperplasticizers, especially polycarboxylate type (PCE) in cementitiousmaterials, retards cement hydration to some extent [29–31]. Generalobservation is that PCE retards cement hydration by delaying the mainhydration peak and decreasing the maximum hydration rate [29].Such retarding effect is usually seen in a hydrating cement paste forsome hours. On the other hand, it has been also realized that in thevery early hydration period (b20 min), PCE clearly retards the forma-tion of ettringite [31], which can be called as early retarding effect.This way, the “network structure” in FCP containing superplasticizersbecomes weaker than that in blank FCP due to the dispersingeffect and early retarding effect of superplasticizers, as described inFig. 4(i) and (j).

FCPs with different microstructures are bound to present differentviscoelastic properties, which has been extensively reported in theliteratures [17,20]. It was found that the elasticity of FCPs would contin-uously grow with time and the addition of superplasticizer enabled toreduce the elasticity of FCPs as well as its growth rate with time.

Fig. 3.MSD curves of purely viscous and viscoelastic samples.

2.3. Application of microrheology analyzer in fresh cement paste

As discussed above, the principle of the microrheology analyzer isbased on the measurement of the Brownian motion of particles. As atypical solid–liquid suspension system [32], fresh cement paste can bea suitable object to be measured by the microrheology analyzer. Thismethod may provide abundant information about the rheologicalproperties as well as the viscoelastic properties of the FCP samples. In-formation of cement hydration and setting of FCP could also be obtainedby following the evolution of those parameters provided by themicrorheology analyzer. In addition, the impacts of chemical admix-tures, including viscosity modifier and superplasticizer on propertiesof FCP could be easily investigated by this method. Compared to the tra-ditional mechanical testing methods, the most important advantage ofthis method is non-disturbing and easy operation. Therefore, the inten-tion of this study is to provide the first proof-of-concept for usingmicrorheology method in cementitious materials. The first test resultof MSD curve captured by microrheology analyzer is shown in Fig. 5.The MSD curves emerge in blue at first and then in green, and lastly inred with the progressing cement hydration. During the cement hydra-tion, the movement of MSD curves from top to bottom and from leftto right suggests an increasing elasticity of FCP with the elapsed time.This conclusion is fully consistent with the results of oscillatory rheom-eter and ultrasonic reflectionmeasurements found by other researchers[17,20]. As a consequence, it is confirmed that microrheology analyzercould be used to follow the development of the viscoelastic propertiesof FCPs. The corresponding parameters extracted from the MSD curvescould also be applied in the quantitative analysis of the viscoelasticproperties.

3. Experimental

3.1. Materials

Reference cement PI 42.5 complying with the Chinese standardGB8076-2008was used,whose composition is listed in Table 1. The con-tent of CaSO4 in the cement is 4.9%. The fineness of the cement is 2.3%and the density is 3.10 g cm−3. Fineness of cement is measured bysieving it on a sieve with mesh size of 0.08 mm. The proportion of the

Fig. 4. Schematic illustration of microstructure of FCPs (a, b, c) blank FCP at lowW/C, 0 min, 10min, and 60min, (d, e, f) blank FCP at highW/C, 0 min, 10min, and 60min and (h, i, j) FCP

with superplasticizer, 0 min, 10 min, and 60 min. ( : cement grain; : flocculated structure; : adsorption water; : free water; : entrapped water; : ettringite; :

Ca(OH)2).

294 Y. Zhang et al. / Cement and Concrete Research 79 (2016) 291–300

remaining cement with size larger than the mesh size is thus deter-mined according to the British standard BS EN 196-6:2010. The particlesize distribution of the cement is measured by using a laser particle sizeanalyzer, as shown in Fig. 6. A self-synthesized polycarboxylate (PCE)type superplasticizerwas employed, which is an aqueous solution of co-polymer of acrylic acid (AA), methyl polyethylene glycol methacrylate(MPEGMA) and 2-acrylamido-2-methylpropane sulfonic acid (AMPS)

Fig. 5.MSD curves of a blank FC

with a monomer molar ratio of 2.12:1.00:0.29 prepared via free radicalpolymerization by using ammonium persulfate (APS) as initiator at80 °C. The weight average molecular weight (Mw) of MPEGMAwas about 1300 and the average polymerization degree of the poly(ethylene oxide) was about 28. The number average molecular weightof the PCE is 3.66 × 104 and polydispersity index Mw/Mn is 2.48. Solidcontent of the PCE superplasticizer solution is 40 wt.%.

P during cement hydration.

Table 1Chemical and mineral composition of cement (wt.%).

Chemical composition Mineral composition

SiO2 Al2O3 Fe2O3 CaO MgO SO3 Na2Oeq f-CaO C3S C2S C3A C4AF

21.68 4.80 3.70 64.90 2.76 0.29 0.56 0.93 57.34 18.90 6.47 11.25

295Y. Zhang et al. / Cement and Concrete Research 79 (2016) 291–300

3.2. Sample preparation and measurements

The experimental measurements have been performed on freshcement pastes in which the mass ratios of water to cement (W/C)were respectively fixed at 0.29 and 0.4. The water contained in thesuperplasticizer solution was also included in the calculation of W/C.The mass ratio of superplasticizer to cement (Sp/C) for the pastes withW/C of 0.29 was varied in the range of 0.1% to 0.8%. It should be men-tioned that at W/C as low as 0.29, it is impossible to prepare a homoge-nous and reproducible blank cement paste due to the very poor mixingability. Therefore, themeasurement bymicrorheometer was not practi-cable. AtW/C of 0.4, the dosage of the superplasticizer varies from 0% to0.5%. The maximum dosages are both close to the critical dosages of thesuperplasticizer, at which the dispersion degree of cement grainsreaches the maximum [29].

All the fresh cement pastes were prepared in accordance with theChinese standard GB/T8077. Water and the superplasticizer solutionwere firstly added into a mixer, and then the cement was gradually in-troduced over a time span of 2min into themixer at 62 rpm. After a 10 sinterval, mixing was resumed for an additional 2 min at 125 rpm. Thefreshly mixed cement pastes were instantly subjected to the followingmeasurements.

After mixed well, the FCP was poured into a cylindrical glass cell of20 mL with a 25 mm diameter and then placed into the channels. TheMSD curves of cement grains in the FCP were continuously recordedat a constant temperature of 20 °C for 4 h. According to our previousstudy [33], the initial setting time of the cement pastes is in the rangeof 4–30 h. The corresponding parameters within 2 h were extractedfrom the MSD curves to quantitatively analyze the viscoelastic proper-ties of FCPs before initial setting.

4. Results and discussion

4.1. EI and MVI

The development of EI and MVI of FCPs with different W/Cs andSp/Cs during the first 2 h is plotted in Fig. 7. In most cases, both EI

Fig. 6. Particle size distribution curve of the cement used in this study.

and MVI gradually increase with elapsed time and finally reach a pla-teau, which means the elasticity and macroscopic viscosity increaseover time. The reason is majorly connected to the continuous cementhydration after contact of cement with water. Along with theprogressing cement hydration, a portion of free water is consumedand hydration products are formed. The newly formed hydrates, by cre-ating new links among cement grains, would increase the inter-particlecrosslinking and thus lead to a stronger “network structure”. Corre-spondingly, the elasticity and macroscopic viscosity of FCPs becomemore pronounced. After a rapid initial burst of dissolution and earlyformation of hydration products, the cement hydration steps into theinduction period in which the hydration reaction is proceeding at a rel-atively low rate. During this period, the microstructure of FCPs slowlydevelops and the parameters, EI and MVI slowly grow over the elapsedtime.

In the case of FCPs with high Sp/C, EI and MVI evolution curves startwith aflat part at a very low value, followedby a sharp increasing part. Itis interestingly noted that the length of the flat part is extended by theincrease of superplasticizer dosages and then almost keeps constantwhen Sp/C is beyond the critical dosage. The delayed fast increaseof EI and MVI by the addition of PCE must be related to theabovementioned dispersing effect and early retarding effect of PCE.The addition of PCE in FCP leads to an increased dispersion degree of ce-ment grains and depresses the early formation of ettringite crystals. Ithas been realized that in the very early hydration period (b20 min),PCE clearly retards the formation of ettringite and reduces the size ofettringite crystals [31]. In this way, PCE disassembles the flocculatedstructures in FCP, thereby lessening the inter-particle crosslinking, anddelays the establishment of the “network structure”. These two effectsof PCE, the dispersing effect and the early retarding effect lead to theflat parts of the EI and MVI curves.

When the dosage of superplasticizer is beyond the critical dosage,the dispersion degree of particles reaches maximum and hence thelength of the flat part keeps almost constant. On the other hand, themacroscopic viscosity of FCP is also related to the dispersion of cementgrains and the content of free water, so MVI behaves in a similar wayto EI.

Moreover, the values of EI andMVI aswell as their growth rates overtime are clearly reduced by the addition of superplasticizer as seen inFig. 7, which also originates from the effects of superplasticizer interms of improving the dispersion state of cement grains and retardingthe early cement hydration.

Compared to the case of FCPs at W/C of 0.29, EI and MVI of FCPs atW/C of 0.4 are lower and the effects of superplasticizer are less remark-able. This is a result of the larger distance among the cement grains athigher W/C. In that case, the size of “cage” is so large that it is less sen-sitive to the addition of superplasticizer although the distance can befurther enlarged by the superplasticizer. Similarly, the impact of volumefraction of solid phase on the viscosity of FCPs at higher W/C is consid-ered to beweaker [31]. Hence, the variations of EI andMVI with Sp/C atW/C of 0.4 are not as significant as the FCPs with W/C of 0.29.

4.2. G′ and G″

In order to have a better insight of the viscoelastic properties of FCPs,G′ andG″ over a range of frequencies during the first fewhourswere ob-tained. The variations of G′ and G″ versus frequency at a given elapsedtime are reported in Fig. 8. With increasing frequency, moduli generally

Fig. 7. Variations of EI and MVI of FCPs with different Sp/Cs with elapsed time (a, b) W/C = 0.29 and (c, d) W/C = 0.4.

296 Y. Zhang et al. / Cement and Concrete Research 79 (2016) 291–300

present increasing trendswith slightfluctuations. An increase of bothG′and G″ up to a nearly constant value is noticed with increasing frequen-cy at high superplasticizer dosages, which is fully compatible with theearlier findings reported in reference [6]. In addition,G′ b G″ is observedin the range of lower frequency, suggesting that the viscous behavior ispronounced in this range of frequency. Oppositely, G′ exceeds G″ in thehigher frequency, indicating that the elastic behavior is becoming thedominant feature. Furthermore, with the increase of Sp/C, the intersec-tion point of G′ curve and G″ curve gradually shifts towards left andthe moduli exhibit less frequency dependence. These phenomenareflect the paste is undergoing a significant structural change whensuperplasticizer is introduced into the cement paste.

Another interesting observation from the plot of G″ against frequen-cy in Fig. 8 is that along with the general growth of G″with frequency, apeak ofG″ appears at a certain frequency for all samples. It has beenwellunderstood [34,35] thatwhen a dynamic load is exerted to a viscoelasticmaterial, the strain does not simultaneously developwith the stress anda delayed strain response is usually observed. Lag angle is used todescribe such delayed strain–stress response. The bigger lag angleindicates the more pronounced viscoelastic feature of the material andleads to the larger loss modulus G″. On the other hand, the viscoelastic-ity of materials is highly dependent on the frequency of the dynamicload. At a certain frequency, the material exhibits the most significantviscoelastic feature, which is called the characteristic frequency of thematerial. That is to say, the lag angle of a viscoelastic material is

dependent on the loading frequency. From the location of themaximumpeak of the lag angle or the peak of the loss dynamic modulus G″, thecharacteristic frequency of the material can be determined. Thus, thecharacteristic frequency of the fresh cement pastes can be obtainedas listed in Table 2. A general observation from Table 2 is that thecharacteristic frequency of the sample decreases with the increaseof the Sp/C in cement pastes. This suggests that the addition ofsuperplasticizer certainly changes the viscoelastic behavior of thefresh cement pastes.

At one certain frequency, the variation trends of G′ and G″ with theelapsed time shown in Fig. 9 behave inmuch the sameway as the curvesin Fig. 7. Initially, G′ aswell asG″ present a linear increase over time andthen reach a plateau. This demonstrates that the main evolution of the“network structure” in FCPs occurs within a few minutes after themixing process due to the fast dissolution of clinker phases and the ini-tial formation of hydrates,whereas the structure changes less drasticallyin the next hours before setting. Due to the abovementioned dispersingeffect and the early retarding effect of superplasticizer, the moduli ofFCPs keep stable initially and then start to rise over the elapsed timeat high superplasticizer dosages. Moreover, higher Sp/C brings aboutlower moduli as well as lower growth rates.

When G′ and G″ are compared in Fig. 9, it is not hard to find that G′ isalways higher than G″ for the FCPs with low Sp/C atW/C of 0.29, whichmeans an elastic feature or solid-like behavior prevails, correspondingto the well-known gel state of cement pastes [36]. Nevertheless, in the

Fig. 8.Variations ofG′ andG″ of FCPswith different Sp/Cs versus frequency (a)W/C=0.29and (b) W/C = 0.4.

Fig. 9.Variations ofG′ andG″ of FCPswith different Sp/Cs versus elapsed time (a) (b)W/C=0.29 and (c, d) W/C = 0.4.

297Y. Zhang et al. / Cement and Concrete Research 79 (2016) 291–300

case of high Sp/C or highW/C, G′ is lower than G″ at the beginning but itgrows beyond G″ at the end. The junction between G′ and G″ curveswithin the first few minutes suggests that the rheological behavior ofFCPs transforms from a slightly liquid-like behavior to a strongly solid-like behavior. Obviously, this phenomenon is consistent with the factthat the microstructure of FCPs is undergoing a substantial changefrom a rather dispersed fluid state to a highly structured solid stateresulting from the establishment of the “network structure” due to theelectrostatic interaction between cement grains and the bridging effectof the newly formed hydrates. On the other hand, at the plateau part ofthe curve, G′ is about 20 times higher than G″ at low dosage ofsuperplasticizer whereas G′ is only twice as much as G″ at high Sp/C,suggesting that the elastic characteristic of the FCPs is weakened bythe incorporation of superplasticizer in cement pastes.

Table 2The characteristic frequency of cement pastes at varied W/Cs and Sp/Cs.

W/C 0.29 0.40

Sp/C 0.1% 0.2% 0.3% 0.4% 0.5% 0.8% 0 0.1% 0.2% 0.3% 0.5%

Frequency(10−4 Hz)

2.1 0.60 0.17 0.13 0.20 0.21 20 4.0 1.02 0.60 0.60

4.3. Maxwell model parameters

When the viscoelastic properties of FCP are described by Maxwellmodel, the parameters including the elastic modulus of spring G andthe viscosity of dashpot η could be acquired. Their variation trendswith elapsed time (Fig. 10) are in good agreement with those of EIand MVI, i.e., initially G and η present a linear increase over time andthen reach a plateau, and the inclusion of superplasticizer significantlyreduces the elasticity and viscosity of the FCPs. At lowerW/C, the effectsof superplasticizer on the Maxwell parameters are more remarkable.

The ratio of η to G, η/G is an indicator to determine which feature isdominant in a material, either elasticity or viscosity. It is shown inFig. 10(c) and (f), there are some fluctuations of η/G at the beginning,which is caused by the unstableness of G and η right after mixing.Namely, the fresh cement paste needs some minutes to establish arelatively stable structure without agitation. After that, η/G of FCPsincreases over elapsed time. That is to say, although η and G bothrise due to the cement hydration, η seems to grow more rapidly thanG. Meanwhile, η/G also presents an increasing trend with increasingSp/C. This demonstrates that the addition of superplasticizer reduces Gmore significantly than η. As mentioned previously, the elasticity isdirectly related to the “network structure” of FCP. The main effect ofsuperplasticizer is to release the entrapped water and to increase thedispersion degree of cement grains in FCP. As a result, the “networkstructure” could be strongly weakened and subsequently the elasticityof FCP sharply drops. However, the viscosity of FCP is majorly

Fig. 10. Variations of Maxwell model parameters of FCPs with different Sp/Cs versus elapsed time (a, b, c) W/C = 0.29 and (d, e, f) W/C = 0.4.

298 Y. Zhang et al. / Cement and Concrete Research 79 (2016) 291–300

determined by other factors such as the volume fraction of solid phasein the solid–liquid system, interactions between particles, the frictionsbetween particles and the viscosity of the medium etc., besides thedispersion degree of cement grains [37]. From the results above, itcould be concluded that superplasticizer has more significant effectson the elastic modulus G than the viscosity η. This is in agreement

with the previous finding that the addition of superplasticizer coulddecrease the yield stress more obviously than the viscosity [38–40].

From Figs. 9 and 10(a), (b), (d), and (e), it is also noted that the pa-rameters of fresh cement pastes, G′, G″, G and η start with a flat periodand then sharply grow during cement hydration. It is noticed that theduration of the flat period is in the range of 0–20min and is dependent

299Y. Zhang et al. / Cement and Concrete Research 79 (2016) 291–300

on the dosage of PCE in the cement pastes. Themore dosage of PCE leadsto the longer flat period and the maximum flat period is reached at thecritical dosage. It is believed that this must be again related to the dis-persing effect as well as the retarding effect on early hydration. It hasbeen well understood that in such a short period (within 20 min)after mixing, dissolution of various phases such as C3S, C3A and sulfatecarriers and the early formation of ettringite crystals are the mainprocesses involved in cement hydration. The adsorbed PCE on surfaceof cement grains may retard the initial dissolution of various clinkerphases and early precipitation of ettringite, and hence change the elec-trostatic interaction among the dispersed particles in cement pastes.

5. Conclusions

In this paper, a microrheology analyzer based on the measurementof themean square displacement of particles by laser scatteringmethodwas employed to in-situ monitor the evolution of the viscoelasticproperties of fresh cement pastes (FCPs) for thefirst time. The viscoelas-tic properties of FCPs with varied water–cement ratios (W/Cs) andsuperplasticizer dosages (Sp/Cs) in the early stages of hydration wereanalyzed. On the basis of the results above, the following conclusionscan be drawn:

1) All parameters obtained from MSD curves of FCPs, EI, MVI, G′, G″, ηand G, present a progressive increase with elapsed time initiallyand then stay stable. The addition of superplasticizer significantlydecreases the values of these parameters as well as their growthrates during the hydration of cement. For cement pastes with highdosages of superplasticizer, these parameters start with a flat periodfollowed by a sharp increase with the elapsed time, which resultsfrom the dispersing effects of superplasticizer on cement grainsand the retarding effects on early cement hydration. More remark-able effects of superplasticizer are observed at lower W/C.

2) G′ is always higher than G″ for the FCPs at low Sp/C and low W/C(0.29), whereas in the case of high Sp/C or high W/C (0.40), G′ islower than G″ at the beginning but it grows larger than G″ later on.This indicates that FCP is transforming from a viscous fluid to a vis-coelastic semi-solid and finally becomes an elastic solid. It is believedthat the evolution of the “network structure”, originating from thecrosslinking between cement grains due to their electrostatic inter-action and the formation of the hydration products, contributes tothis transformation.

3) The incorporation of superplasticizer evidently weakens the elas-tic characteristic of fresh cement paste. Compared to viscosity η,elastic modulus G is more significantly reduced by the additionof superplasticizer in FCPs, which is consistent with the previousfindings.

4) This investigation confirms that the non-disturbing measurementprovided by the microrheology analyzer is a novel and powerfultechnique to in-situ follow the development of the viscoelastic prop-erties of fresh cement pastes, whichmay provide abundant informa-tion about the rheological properties as well as the viscoelasticproperties of the FCP samples. Information of cement hydrationand setting of FCP could also be obtained by following the evolutionof those parameters provided by the microrheology analyzer. In ad-dition, the impacts of chemical admixtures, including viscosity mod-ifier and superplasticizer on properties of FCP could be easilyinvestigated by this method. Compared to the traditional rheologytesting methods, the most important advantages of this methodare non-disturbing and easy operation.

Acknowledgment

The financial supports from the National Natural Science Founda-tion of China (Grant No. U1301241 and 50802050) are gratefullyacknowledged.

References

[1] M.A. Schultz, L.J. Struble, Use of oscillatory shear to study flow behavior of freshcement paste, Cem. Concr. Res. 23 (1993) 273–282.

[2] L.J. Struble, M.A. Schultz, Using creep and recovery to study flow behavior of freshcement paste, Cem. Concr. Res. 23 (1993) 1369–1379.

[3] Z. Sun, T. Voigt, S.P. Shah, Rheometric and ultrasonic investigations of viscoelasticproperties of fresh Portland cement pastes, Cem. Concr. Res. 36 (2006) 278–287.

[4] M. Nehdi, S. Al Martini, Effect of temperature on oscillatory shear behavior ofPortland cement paste incorporating chemical admixtures, J. Mater. Civ. Eng. 19(2007) 1090–1100.

[5] C.T. Chen, L.J. Struble, H. Zhang, Using dynamic rheology tomeasure cement–admixtureinteractions, J. ASTM Int. 3 (2006) 1–13.

[6] A. Papo, L. Piani, Effect of various superplasticizers on the rheological properties ofPortland cement pastes, Cem. Concr. Res. 34 (2004) 2097–2101.

[7] A.W. Saak, H.M. Jennings, S.P. Shah, The influence of wall slip on yield stress and vis-coelastic measurements of cement paste, Cem. Concr. Res. 31 (2001) 205–212.

[8] B.J. Christensen, T. Coverdale, R.A. Olson, S.J. Ford, E.J. Garboczi, H.M. Jennings, T.O.Mason, Impedance spectroscopy of hydrating cement-based materials: measure-ment, interpretation, application, J. Am. Ceram. Soc. 77 (1994) 2789–2804.

[9] W.J. McCarter, T.M. Chrisp, G. Starrs, J. Blewett, Characterization and monitoring ofcement-based systems using intrinsic electrical property measurements, Cem.Concr. Res. 33 (2003) 197–206.

[10] P. Xie, P. Gu, Z. Xu, J.J. Beaudoin, A rationalized AC impedance model for microstruc-tural characterization of hydrating cement systems, Cem. Concr. Res. 23 (1993)359–367.

[11] J.R. Rapoport, J.S. Popovics, S.V. Kolluru, S.P. Shah, Using ultrasound to monitor stiff-ening process of concrete with admixtures, ACI Mater. J. 97 (2000) 675–683.

[12] K.V. Subramaniam, J. Lee, B.J. Christensen, Monitoring the setting behavior of ce-mentitious materials using one-sided ultrasonic measurements, Cem. Concr. Res.35 (2005) 850–857.

[13] K.V. Subramaniam, J. Lee, Ultrasonic measurement of early age viscoelastic proper-ties of cementitious materials, Proceedings of 16th ASCE-EMD Conference, Seattle,July 16–18, 2003.

[14] K.V. Subramaniam, J. Lee, Ultrasonic assessment of early-age changes in thematerialproperties of cementitious materials, Mater. Struct. 40 (2007) 301–309.

[15] M.I. Valič, Hydration of cementitious materials by pulse echo USWR: method, appa-ratus and application examples, Cem. Concr. Res. 30 (2000) 1633–1640.

[16] T. Voigt, Y. Akkaya, S.P. Shah, Determination of early age mortar and con-crete strength by ultrasonic wave reflections, J. Mater. Civ. Eng. 15 (2003)247–254.

[17] C. Chung, J.S. Popovics, L.J. Struble, Early Age Stiffening of Cement Paste Using Ultra-sonic Wave Reflection, Transition from Liquid to Solid: Re-examining the Behaviorof Concrete at Early Ages (ACI SP-259-1), in: Kyle Riding (Ed.) American ConcreteInstitute, Farmington Hills, MI 2009, pp. 7–16.

[18] C. Chung, J.S. Popovics, L.J. Struble, Flocculation and sedimentation in suspensionsusing ultrasonic wave reflection, J. Acoust. Soc. Am. 129 (2011) 2944–2951.

[19] C.W. Chung, P. Suraneni, J.S. Popovics, L.J. Struble, W.J. Weiss, Application of ultra-sonic P-wave reflection to measure development of early-age cement-paste proper-ties, Mater. Struct. 46 (2013) 987–997.

[20] X. Wang, K.V. Subramaniam, F.B. Lin, Ultrasonic measurement of viscoelasticshear modulus development in hydrating cement paste, Ultrasonics 50 (2010)726–738.

[21] C. Tisserand, M. Fleury, L. Brunel, P. Bru, G. Meunier, Passive Microrheology forMeasurement of the Concentrated Dispersions Stability, UK Colloids 2011, Springer,Berlin Heidelberg, 2012 101–105.

[22] M.L. Gardel, M.T. Valentine, D.A. Weitz, Microrheology, Department of Physics andDivision of Engineering and Applied Sciences, Harvard University, Cambridge, MA,2005 02138.

[23] M. Bellour, M. Skouri, J.P. Munch, P. Hébraud, Brownian motion of particles embed-ded in a solution of giant micelles, Eur. Phys. J. E 8 (2002) 431–436.

[24] L. Brunel, P. Snabre, EP 1,664,744, 2003.[25] L. Brunel, Formulaction, WO2010130766, 2009.[26] T.G. Mason, Estimating the viscoelastic moduli of complex fluids using the general-

ized Stokes–Einstein equation, Rheol. Acta 39 (2000) 371–378.[27] K. Yoshioka, E. Tazawa, K. Kawai, T. Enohata, Adsorption characteristics of

superplasticizers on cement component minerals, Cem. Concr. Res. 32 (2002)1507–1513.

[28] A. Zingg, F. Winnefeld, L. Holzer, J. Pakusch, S. Becker, L. Gauckler, Adsorption ofpolyelectrolytes and its influence on the rheology, zeta potential, and microstruc-ture of various cement and hydrate phases, J. Colloid Interface Sci. 323 (2008)301–312.

[29] Y. Zhang, X. Kong, Correlations of the dispersing capability of NSF and PCE typesuperplasticizers and their impacts on cement hydration with the adsorption infresh cement pastes, Cem. Concr. Res. 69 (2015) 1–96.

[30] E. Sakai, T. Kasuga, T. Sugiyama, K. Asaga, M. Daimon, Influence of superplasticizerson the hydration of cement and the pore structure of hardened cement, Cem. Concr.Res. 36 (2006) 2049–2053.

[31] J. Cheung, A. Jeknavorian, L. Roberts, D. Silva, Impact of admixtures on the hydrationkinetics of Portland cement, Cem. Concr. Res. 41 (2011) 1289–1309.

[32] N. Roussel, A. Lemaître, R.J. Flatt, P. Coussot, Steady flow of cement suspensions: amicromechanical state of the art, Cem. Concr. Res. 40 (2010) 77–84.

[33] Y. Zhang, X. Kong, S. Hou, Study on the rheological properties of Portland cementpastes with polycarboxylate superplasticizers, Rheol. Acta 52 (2013) 707–718.

[34] K.P. Menard, Dynamic Mechanical Analysis: A Practical Introduction, CRC Press,2008.

300 Y. Zhang et al. / Cement and Concrete Research 79 (2016) 291–300

[35] Y.L. Liu, X.M. Kong, Y.R. Zhang, P.Y. Yan, Static and dynamicmechanical properties ofcement–asphalt composites, J. Mater. Civ. Eng. 25 (2012) 1489–1497.

[36] Z. Zhou, M.J. Solomon, P.J. Scales, D.V. Boger, The yield stress of concentrated floccu-lated suspensions of size distributed particles, J. Rheol. 43 (1999) 651–671.

[37] A. Chougnet, A. Audibert, M. Moan, Linear and non-linear rheological behaviour ofcement and silica suspensions. Effect of polymer addition, Rheol. Acta 46 (2007)793–802.

[38] R.J. Flatt, Towards a prediction of superplasticized concrete rheology, Mater. Struct.37 (2004) 289–300.

[39] J. Gołaszewski, J. Szwabowski, Influence of superplasticizers on rheologicalbehaviour of fresh cement mortars, Cem. Concr. Res. 34 (2004) 235–248.

[40] J.Y. Petit, E. Wirquin, Y. Vanhove, K. Khayat, Yield stress and viscosity equationsfor mortars and self-consolidating concrete, Cem. Concr. Res. 37 (2007)655–670.