Cement and Concrete Research 60 (2014) 37–44

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

Generation and stability of bubbles in a cement based slurry

Pauline Petit a, Isabelle Javierre b, Pierre-Henri Jézéquel b, Anne-Laure Biance a,⁎,1

a Institut Lumière Matière, Université de Lyon, UMR5306 Université Lyon 1-CNRS, 69622 Villeurbanne, Franceb Lafarge Centre de Recherche, St. Quentin Fallavier, Isère, France

⁎ Corresponding author.E-mail address: anne-laure.biance@univ-lyon1.fr (A.-L

1 Tel.: +33 4724 48228.

http://dx.doi.org/10.1016/j.cemconres.2014.02.0080008-8846/© 2014 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 21 August 2013Received in revised form 20 February 2014Accepted 28 February 2014Available online 3 April 2014

Keywords:Foamed concreteParticles at interfacesSurfactant

In this article, the fabrication of a single stable cement bubble is investigated. To achieve this goal, the stability ofmodel particle covered bubbles isfirstly experimentally studied, by bubbling in a pool filledwithmicrometric sil-ica particles. Bubble stability is shown to be governedmainly by particle covering rate, which ismaximizedwhenparticlewetting angle prior to liquid approachesπ/2. This angle can be adjusted in situ by electrostatic adsorptionof cationic surfactant on silica if proper amount of surfactant is added in the silica suspension. The covering rate isalso shown to be governed by the time spent by the bubble in the pool, allowing us to define a timescale forparticle adsorption at the liquid/gas interface. In the end, this method is shown to be successful with othertypes of foamed granular materials such as cement, and the fabrication of a stable and fully covered solid cementbubble is for the first time demonstrated.

© 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Buildings currently account for 40% of the global primary energy con-sumption,mainly through its daily use during its life time [1], and in par-ticular through thermal regulation. In this context, designing a materialwith large insulation properties keeping a good mechanical resistanceremains a crucial challenge. Solid foams, and particularly foamedconcrete, gather these characteristics [2], which explains the renewedinterest for this material. These particular foamed materials also possessa light weight with densities between 1600 and 400 kg/m3 [2] thusreducing primary material consumption and consequently decreasingthe greenhouse-gas emissions due to the manufacture of Portland ce-ment, which today represents 5% of all human generated emissions [3].

Solid foams are constituted of air voids captured in a solid matrix,with a gas fraction which can reach up to 97% [4,5]. Foam is amultiscalematerial, whose macroscopic properties, like mechanical strength [6],are strongly related to the microstructure of the foam. For example,thermal and acoustical insulations are governed by the structure ofclosed or interconnected bubbles [6,7], an open porosity being insome situation a drawback for a good insulation. Moreover, materialdurability is strongly affected by this porosity because of fluid and iontransport [8] within the foam matrix. Consequently, to achieve aperformantmaterial, foamporositymust be controlled to obtain discon-nected air bubbles within the matrix [5,9,10]. Many factors such as

. Biance).

drainage, coarsening or rupture affect the interconnection betweenbubbles, mainly during foam generation and solidification. The firststep to achieve such a material is the possibility to create one single,closed and solidified, bubble. Then, a prerequisite is the formation of alongstanding stable concrete bubble that will in a second step solidifyvia concrete hydration reaction. The conditions required for these twoprocesses to be achieved, the stable bubble formation and its subse-quent solidification, are discussed in the following.

Indeed, a bubble, which consists of a thin liquid film separating twogas regions, is intrinsically an out-of-equilibrium system that willeventually collapse. To stabilize a bubble, amphiphile molecules suchas surfactants,which adsorb on the interface and allow spontaneous cic-atrization of some holes created by external perturbations, are added inthe liquid solution. These molecules allow to stabilize foams or bubblesfor times ranging from a few minutes to a few hours. Another type ofsurface active stabilizers are nano or micro particles, which are moreprotective as they remain attached on the interface. Indeed, the desorp-tion energy of a particle attached to a liquid/gas interface depends onthe wetting property of the solid with respect to the liquid and the gasand reads W = πRp2γlg(1 − cosθ)2, with Rp the radius of the particle,γlg the surface tension of the liquid/gas interface, and θ b 90° the contactangle [11–13]. This desorption energy is maximal around 90° and ofthousands of kT for microparticles. Thus, foam stabilized only by solidparticles can last formonths as the particle layer at the interface inhibitsfilm collapse and then bubble coalescence and/or coarsening [14]. Themain challenge of thismethod is to adsorb the particles on the interface,which can be achieved by tuning in situ the relative solid/liquidwettingproperties. Martinez et al. [14] changed the liquid properties by addinga small amount of volatile ethanol in the initial mixture that will

Table 2Physical properties of the particles, with d the density, Σ the specific surface and Rp theparticle radius.

Material Silica Cement Limestone

d (kg m−3) 2500 3110 2750Σ (m2 g−1) – 0.42 1Rp (μm) 5–40 1–50 1–50

38 P. Petit et al. / Cement and Concrete Research 60 (2014) 37–44

eventually evaporate, and then reduce the particle affinity with the liq-uid. Another common method is to modify the wetting properties ofparticles by electrostatic adsorption of oppositely-charged short-chainsurfactants on the solid surface [15–21].We investigate in the followingthe possibility of this processes on particle and cement stabilizedbubbles.

The second prerequisite is to keep a stabilized bubble during solidi-fication. Indeed, a solid thin film and then a closed bubble are obtainedonly if the hydration reaction occurs within the thin film covering thebubble: the film needs to be filled with particles but also requires a suf-ficient quantity of water for the hydration reaction to occur. Moreover,the reaction adds new destabilization factors to the foam, because ofwater movement, modification of particle properties, cement reactivityand shrinkage.

To investigate these two steps,we first study the fabrication of stablelong standing bubble in a model suspension that is constituted of silicabeadswhosewettability is changed in situ by surfactant adsorption. Theeffect of suspension wettability, concentration and bubble rising timeare investigated. The methodology employed is then applied to bubblescreated in a cement paste where the solidification process is tested. Thearticle is organized as follows: after a first part dedicated to thematerialand method employed, the fabrication of bubbles fully covered withsolid spherical silica beads is considered as a model system. Then, theconditions for obtaining a bubble stabilized with cementitious particlesand its subsequent solidification are successfully experimentallyinvestigated.

2. Materials and methods

2.1. Materials

In order to study bubble formation in a model paste, silica beads(Tecosphere) with a mean diameter of 36 μm and a standard deviationof 37 μm are suspended in an aqueous solution of surfactants at differ-ent concentrations between 5 × 10−4 and 2 times the critical micellarconcentration (CMC) of the system. Sodium dodecyl sulfate (SDS) isused as a model anionic surfactant (Sigma, 98.5%, No. L4509, CMC =8 × 10−3 mol L−1 [22]), and tetradecyltriammonium bromide(TTAB) is used as a model cationic surfactant (Sigma, 99%, No.T4762, CMC = 3.5 × 10−3 mol L−1). Glycerol is added to vary thebulk viscosity η of the suspending phase between 1 and 10.7 mPa s[23]. The solid volume fraction ϕ in the suspension is varied between26% and 52%. The reference compositions are given in Table 1. TTABand SDS are not used simultaneously and the different parameters ofthe system are varied independently. The viscosity of the suspendingfluid is measured with a capillary viscosimeter (Ubbelohde) with a pre-cision of 15% because of temperature variations and the surface tension ismeasured by theWilhelmy plate method (Nima) with a precision of 5%.

In our real cement paste, Portland cement CEM I 52.5 R ismixedwitha limestone filler (Betocarp HP Orgon de Omya) and the water/powdermass ratio is 0.25. Solutions of two different admixtures aremade in de-ionized water: polycarboxylate polymers (PCP) containing grafted PEOchains as superplasticizer, and surfactants (either SDS or TTAB). Small

Table 1Reference composition of the suspensions in thedifferent experiments: concentration of TTAB, Ssilica beads, cement and filler in the solid phase (%).

System a b c

TTAB [10−5;10−2] – 5 × 10−4

SDS – 8 × 10−4 –

PCP – – 4.8Glycerol [10;60] 10 –

ϕ 52 52 58Silica 100 100 –

Cement – – 33.7Filler – – 66.3

quantities of paste are used, so the powder is gently stirred by handduring 1min in the aqueous solution. Tables 1 and 2 summarize respec-tively the different reference compositions and solid phase physicalproperties used in our experiments.

2.2. Methods

Air bubbles are injected through an aperture at the bottomof a cylin-drical tank of diameter 26 mm initially filled with the paste. Afterdetaching, bubbles rise by gravity in tubes of height h = 3.1 cm. Asyringe pump allows to inject a controlled air volume (30 μL) at agiven injection rate (0.1 mL s−1), in order to obtain single bubbles.When the bubble reaches the interface, a liquid film is formed, whichis fully covered with particles. The movement of the particles on thebubble thin film is then observed through a stereomicroscope, asdepicted in Fig. 1. When the bubble reaches the interface, a liquid filmis formed, which is fully covered with particles. The movement of theparticles on the bubble thin film is then observed through a stereomi-croscope, as depicted in Fig. 1.

The particles rearrange in two zones, a covered one and a bare one asreported in Fig. 2. Obtained with reflected light, the stereomicroscopeimages allow to distinguish the area covered by particles from thebare area, as soon as the bubble reaches the surface. Then, the evolutionof the covering rate of particles on the bubble is determined as afunction of time. The covering rate is at equilibrium generally 15 safter the bubble reaches the surface, and a mean value of the coveringrate is measured at equilibrium for 5 to 10 different bubbles.

As wettability of our suspending solution versus particles is a crucialpoint on our experiments, two types of wettability measurements areperformed, depending on the case considered. The contact angles ofthe different solutions on silica surface are determined by drop deposi-tion on a clean plane silica slide, and side view image analysis as shownin Fig. 3-top. Each value is a mean of six measurements. Wettability ofcement particles ismeasured by imbibition [24,25] of the liquid throughthe powder. A cylindrical column of 6 cm of height and 1.5 cm ofdiameter is filled with 13 g of packed cement powder. The bottom ofthis column is put in contact with a solution of PCP and SDS, as shownin Fig. 3-bottom. An estimation of powder wettability is deduced fromthe impregnation dynamics throughWashburn law, with hw the heightof column filled by liquid, dp the pore diameter, and t the time:

hw tð Þ≃ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffidptγlgcosθ

2η

sð1Þ

DS (mol L−1) andPCP (g L−1), glycerol content (wt.%), solid volume fractionϕ (%),mass of

d e f

– 5 × 10−4 –

[10−6;10−2] – [10−6;10−2]4.8 – –

– – –

58 35 35– – –

33.7 33.7 33.766.3 66.3 66.3

Syringe pump

Camera

Bubble

Stereomicroscope

Paste h ~ 3,1 cm

Gas injection

Reflectinglight

Fig. 1. Experimental set-up: a bubble is blown by the syringe-pusher in the tank contain-ing our suspension. Bubble stability is investigated by top view images through astereomicroscope.

1 mm

θ

hw(t)

hcement=6cm

liquid 1 cm

a

b

Fig. 3. Top: picture of a droplet on silica surface used for contact angle measurement.Bottom: picture of the imbibition experiment used to determine powder hydrophobicity.

39P. Petit et al. / Cement and Concrete Research 60 (2014) 37–44

This law can be quantitative if a geometrical factor is introduceddepending mainly on the permeability of the powder. In practice, forcomparison between the different solutions tested, and as the geometryof the powder is preserved in the experiments, a 6 cm liquid rise dura-tion is measured, which will vary only due to particle wettabilitymodifications.

In the end, the solidification of the bubbles is achieved in Petridishes. Post-mortem observations are also performed by SEM visualiza-tion. Then, the solidified bubbles are casted in a resin and polished sec-tions of diameter 30mm are prepared. Then, the samples are metalizedunder vacuum with a carbon evaporator. The microstructure of theobtained solid bubbles is analyzed by scanning electron microscopyimages of sample slices (FEG-quanta400, HI-vacuum mode, 15 kV,with backscatter electron sensor).

In the following, the covering rate of the rising bubble is measuredfor different compositions and surfactant concentrations of our modelsuspension of silica beads. It is shown to be correlated to particle wetta-bility. A similarmethod is then employed tomaximize the covering rateof bubble rising in a cement paste, where the cohesion of this particlelayer is then tested through solidification.

3. Results and discussion

3.1. Model system

The effect of the nature of surfactant on the evolution of the coveringrate with time is first studied with aqueous solutions and either anionic

Fig. 2. Left: Picture observed in reflected light from the stereomicroscop

surfactant (SDS, system (b) of Table 1) or cationic surfactant (TTAB,system (a) of Table 1 with 5 × 10−4 mol L−1 and containing10 wt.% of glycerol). The measurement values are reported in Fig. 4,which show that particles are expelled out of the film in less than 1 s

e. Right: Image treatment used to determine the film covering rate.

0 2 4 6 8 10 120

20

40

60

80

100

t [s]

Cov

erin

g ra

te [%

]

Fig. 4. Evolution of a bubble particle covering rate with time for suspending solutions at52% of solid volume fraction with SDS at 8 × 10−4 mol L−1 (system b of Table 1, greentriangles) and TTAB at 5 × 10−4 mol L−1 (system a of Table 1, red diamonds and bluesquares).

40 P. Petit et al. / Cement and Concrete Research 60 (2014) 37–44

in the case of SDS, while bubbles stay fully covered with TTAB duringmore than 10 s.

The contact angle of the suspending fluid containing 60wt.% of glyc-erol on the silica slide is 18 ± 6° in the case of SDS (8 × 10−4 mol L−1),while it is 32± 6° for TTAB (5× 10−4mol L−1). Indeed, TTAB has a pos-itively charged head, which can adsorb on the negative silanol groups ofthe silica surface [20], and therefore increase its contact angle.When theparticles have a larger contact angle, due to energy balance, they preferto adsorb on the interfaces and then create a protective layer which sta-bilizes bubbles provided that the contact angle remains below 90° [12].

Then, to test the optimum of this process, the bubble covering rate ismeasured for suspending liquids containing 60wt.% of glycerol and var-ious TTAB concentrations inwater (system a of Table 1). Silica beads areshown to adsorb on the liquid film when TTAB concentration remainsbetween 2 × 10−4 mol L−1 and 2 × 10−3 mol L−1, as shown in Fig. 5.Particles are expelled out of the film and covering rate collapses ifTTAB concentration is set out of this range i.e. for larger and lower con-centrations. The evolution of covering rate with concentration is com-pared to the one of the contact angle on the glass plate, as reported inthe inset of Fig. 5. The variations of the wettability versus concentrationare smoother than the ones of the covering rate, but both are correlatedand show the same qualitative tendency.

If adsorption of TTAB surfactantmolecules on silica beads is themainmechanism of wettability variations, the optimum TTAB concentration

10−4

10−3

10−2

0

20

40

60

80

100

C [mol.L−1]

Cov

erin

g ra

te [%

]

10−4 10−20

10

20

30

40

C [mol.L−1]

θ [°

]

Fig. 5.Evolution of the equilibriumcovering ratewith TTAB concentrationwith systema ofTable 1 (inset: contact angle of the suspending fluid on a silica slide for different TTABconcentrations).

can be estimated by considering that each available site of silica beadis covered by a TTAB molecule. Indeed, when silica beads are mixedwith milli-Q water, the water pH increases from 6.3 to 10, indicatingthat the silanol (SiOH) groups on the silica surface are in the state ofthe conjugate base SiO−. When TTAB is added, as related in Fig. 6, theammonium ions (TTA+) adsorb on the negative SiO− sites of the silicacreating a hydrophobic monolayer and an increased contact angle.When each silanol site interacts with one surfactant molecule, a secondlayer of surfactant can adsorb on the previous one in order to minimizethe number of aliphatic chains in contact with water: the silica surfacebecomes covered by a bilayer as depicted in Fig. 6, and the contactangle decreases again.

The minimum of wettability can be estimated considering that itoccurs when the number of silanol groups is equal to the number ofsurfactant molecules in solution:

Cmax ¼1NA

1Σsilanols

3ϕ1−ϕ

1Rp

ð2Þ

with Cmax the concentration corresponding to the maximum con-tact angle, NA the Avogadro number, 1/Σsilanol the surface chargedensity, and ϕ the solid volume fraction. Therefore, with 1/Σsilanols =0.5SiO−/nm2 [22], ϕ = 52%, and Rp∼20m , one obtains Cmax ∼ 1 ×10−4mol ⋅ L−1. This estimation is slightly lower than the maxi-mum of the experimental covering rate and contact angle, around 5 ×10−4 mol L−1 (Fig. 5). However, this rough estimation does not takeinto account the loss of entropy associated to surfactant adsorption,which limits the balance between the adsorbed and bulk species and con-sequently might explain the discrepancy observed.

Requirements for foam fabrication generally involve dynamical pa-rameters among which the duration of particle adsorption is usually acrucial limiting factor for stable foam fabrication.We investigate the ef-fect of this duration on the covering rate by varying the height h of thetank and consequently the time spent by the bubble in the paste.Fig. 7 shows the results obtained for a fixed TTAB concentration of 5× 10−4 mol L−1 and particle volume fraction ϕ = 44%. The height ofrise appears to be a crucial parameter: the covering rate increaseswith h, a plateau being observed for height below 3 cm. To disentanglethe effect of time spent in the solution with the one of the number ofparticle encountered during bubble rise, the solid content of the pasteis varied for a fixed value of h. According to Fig. 7, particles are expelledout of the film if ϕ b 35%, but are stuck on the film if ϕ N 50%. Additionalexperiments (not shown here) confirm these observations for differentTTAB concentrations, so this effect is not a consequence of the shift onCmax (Eq. (2)), but it is linked to the mechanism of bead adsorption atthe liquid/gas interface during the bubble rise. Indeed, silica beadshave to overcome an interfacial energy barrier to absorb on the inter-face. Then, the probability to have a full coverage is expected to increasewith the number of particles met during the rise.

When h orϕ aremodified, both the numberN of particlesmet by therising bubble and the rising duration t are respectively varied. Indeed,when the solid content ϕ of the paste is increased, its viscosity and con-sequently the bubble rise velocity aremodified. The number of particlesNbeads met by the bubble during the rise then reads:

Nbeads

πR2b

∼ ϕh43πR

3p

ð3Þ

and the duration of rise tr writes:

tr ¼92

hηpaste ϕð ÞgR2

b ρair−ρpaste

� � ð4Þ

where ηpaste is the viscosity of the suspension estimated by the law ofKrieger–Dougherty [26], Rb is the bubble radius, ρair is the density ofthe gas inside the bubble, and ρpaste is the density of the suspension.

silica

TTAB TTAB

silica +

+ ++ silica +

+

θ θ θ=

Fig. 6. Mechanism of surfactant adsorption at the surface of silica beads. Blue and red circles represent respectively negative and positive charges.

41P. Petit et al. / Cement and Concrete Research 60 (2014) 37–44

The covering rates are then represented as a function of the duration ofrise tr or the number of particles met Nbeads in Fig. 7.

For our two sets of experiments (with variable tank height or vari-able solid content in the tank), the collapse of the data when plottedin function of the duration of rise assumes that it is the parameterwhich controls the bubble covering rate at the end of the rise.Moreover,these experiments allow us to define a timescale for particle adsorptionof about 0.4 s, to be compared to the timescale of foam generationprocesses.

3.2. Concrete bubbles

The same experiments of the covering rate measurements are per-formed for bubbles rising in a cement paste. Our cement based slurryis a complex material as described before, as it contains cement, fillerparticles and PCP. To disentangle the effect of cement and filler particleson the covering rate, benchmark experiments in a paste (ϕ= 50% andϕ= 45%) of cement particles alone or filler particles alone respectivelyhave been performed, always in the presence of PCP in the liquid.These experiments show that filler and cement particles have similarbehaviors, very comparable to the one of amix between the two species.

0 2 4 60

20

40

60

80

100

h [cm]

Cov

erin

g ra

te [%

]

0 2 4 6 8

x 104

0

20

40

60

80

100

N [mm−2]

Cov

erin

g ra

te [%

]

Fig. 7. Top: covering rate at equilibrium, as a function of the height h of rise (left) and of the solidthe riseN ¼ Nbeads

πR2b(left) and versus the duration of the bubble rise tr (right). Blue squares and red

respectively, data from above.

In both cases, the particles stay on the bubbles, which break as soon as aliquid thin film appears, in about 17 ± 3 s.

Then, as performed with silica beads, TTAB and SDS are tested at theconcentrations of system (c) and (d) of Table 1 with a mix of cementand filler, in order to deduce which kind of surfactant can interactwith our cementitious mix of particles. The covering rate is measuredin the presence of PCP, in order to obtain a paste comparable with themodel system concerning the solid volume fraction, but with a con-trolled rheology necessary for the bubble rise. With TTAB, a liquidsoap film is observed after 2 ± 0.5 s of particle migration, while a stablecovered bubble is obtained with SDS at 8 × 10−4 mol L−1. In the case ofcement particles, it seems that wettability modification and particleadsorption is more efficient when a negative surfactant is added in theliquid phase, in good agreementwith a positive zeta potential of cementparticle [27].

By analogy with the study on silica beads, the effect of SDS concen-tration on the bubble covering by particles is investigated with thesystem (d) of Table 1. The covering rate measured for different concen-trations 15 s after a bubble appears at the liquid surface is reported inFig. 8. The particles are stuck in the thin film for SDS concentrationsranging between 4 × 10−5 and 1 × 10−3 mol L−1, where fully coveredbubbles are observed. Moreover, the bubble remains closed after

0.2 0.3 0.4 0.5 0.60

20

40

60

80

100

φ

Cov

erin

g ra

te [%

]

0 0.2 0.4 0.6 0.80

20

40

60

80

100

t [s]

Cov

erin

g ra

te [%

]

volume contentϕ (right). Bottom: covering rate versus the number of particlesmet duringdiamonds correspond to experiments at different heights of rise and solid volume fraction

10−6

10−5

10−4

10−3

10−2

10−1

75

80

85

90

95

100

C [mol.L−1]

Cov

erin

g ra

te [%

]

10−5

10−3

10−1

0

50

100

C [mol.L−1]

Cov

erin

g ra

te [%

]

Fig. 8. Covering rate on cement bubble measured 15 s after the bubble rise for differentSDS concentrations (systems (d) of Table 1). Inset: Covering rate on cement bubble mea-sured 15 s after the bubble rise for different SDS concentrations (systems (f) of Table 1).

Table 3Rise duration in 6 cm of cement powder with solutions containing 4.8 g L−1 of PCP andvarious concentrations of TTAB.

C [mol L−1] t [min] η [mPa s] γlg [mN m−1]

5 × 10−4 16 ± 10 0.99 43.71 × 10−2 23 ± 10 1.14 36.62.5 × 10−2 20 ± 10 1.07 37

42 P. Petit et al. / Cement and Concrete Research 60 (2014) 37–44

solidification if maintained during 1 day in a closed Petri dish, as report-ed in Fig. 9. Outside of this range of SDS concentration, particle migra-tion is limited but bubble breakage is observed as soon as a bare areawithout particles grows up, usually after 22 ± 7 s, depending on thesurfactant concentration.

However, the mix contains PCP and SDS molecules, which are bothnegatively charged, and should then be in competition concerning theadsorption on cement particles. Therefore, TTAB and SDS are testedalone at the concentrations of system (e) and (f) of Table 1. The samequalitative behavior is observedwith TTAB and SDS.With TTAB, a liquidsoap film is observed after 2 ± 0.5 s of particle migration, while a stablecovered bubble is obtainedwith SDS.Moreover, the variation of the cov-ering rate with SDS concentration shows an increase at low concentra-tions (insert of Fig. 8). The curve maximum seems also to be slightlyshifted towards a higher surfactant concentration. We can deduce thatwith or without PCP, the effects of negatively charged surfactantconcentration are qualitatively similar.

To validate the mechanism proposed for model silica beads on theconcrete material, based on particle wettability modification by surfac-tant adsorption, cement wettability with respect towater must bemea-sured. Contact angle measurement by droplet deposition method isdifficult to achieve as water drops deposed on a bed of cement particlesspread on the surface in less than 1min because of thematerial porosity

Fig. 9. Solid bubble obtained with a cement paste containing PCP and SDS at 8 ×10−4 mol L−1, after setting in a Petri dish (systems d of Table 1).

[28], even on a solidified material, i.e. a hardened and sanded cementsurface. A more suitable method to quantify the wettability is the studyof imbibition dynamics [29], as defined previously (see Section 2.2).The results of imbibition duration over 6 cm of powder are reported inTables 3 and 4 for water + PCP mixtures with different surfactanttypes (SDS or TTAB) and concentrations.

One can first notice that the rise duration is always shorter than40 min, except for two concentrations of SDS. The variations of solutionviscosity and surface tension cannot explain these variations of imbibi-tion dynamics over more than one order of magnitude. Therefore, thesurface wettability is altered by the introduction of SDS in the solution,as TTAB remains inert. It correlates with the covering rate measure-ments: positively charged particles are expelled from the film by repul-sive interactions with positive surfactants (TTAB) and by drainage,while particle surface is made partially hydrophobic by adsorption ofnegative surfactants (SDS). The effect is observed evenwith PCP in solu-tion. It shows that when negative surfactants and superplasticizers aresimultaneously mixed in the solution, an adsorption competition canoccur.We can compare this effectwith the competition between sulfateions and PCP adsorption which reduces rheological modification ofsuperplasticizer on the slurry [30].

As in the model silica particle solution, particle wettability de-pends again on surfactant concentration, here SDS. The rise dura-tion is less than 20 min for solutions of low concentration in SDS until8 × 10−4 mol L−1. At 1.6 × 10−2 mol L−1 and 4 × 10−2 mol L−1, the so-lutions have imbibed only 3 cm after 2 h. For higher concentrations, therise duration decreases again to 35 min, where adsorption of multiplelayer is invoked, as depicted in Fig. 6.

The SDS concentrations at which we obtain fully covered bubblesand atwhich imbibition is lower are not in a satisfying agreement. How-ever, these discrepancies can be attributed to differences between thetwo experiments such as the timescale involved, thenumber of particlesin contactwith the liquid or powder composition. Indeed,while amix ofcement and filler is used in the covering rate measurements, cementalone has been characterized for imbibition, because of dry powder ho-mogenization issues. Surfactant adsorption is directly related to the zetapotential of the particles, which is related to the different mineralogicphases of the cement powder composition [27]. Furthermore, cementis reactive and its properties, in particular the ionic concentrationswith-in the paste, evolvewith time. The ionic composition of the solution hasan effect on the adsorption and zeta potential because of the effects ofcharge screening [31,32]. Therefore, as the timescales of both experi-ments differ from one order of magnitude, discrepancies between thetwo methods are expected. Moreover, the critical concentration forwhich wettability modification occurs is a function, as shown in our

Table 4Rise duration in 6 cm of cement powder with solutions containing 4.8 g L−1 of PCP andvarious concentrations of SDS.

C [mol L−1] t [min] η [mPa s] γlg [mN m−1]

0 17 ± 5 1.15 58.54 × 10−5 18 ± 5 1.10 578 × 10−4 20 ± 10 1.12 52.48 × 10−3 37 ± 10 1.13 40.91.6 × 10−2 N120 1.16 40.14 × 10−2 N120 1.30 39.28 × 10−2 35 ± 5 1.51 38

Fig. 10. SEM images of solid cement bubble after 6 months of hydration.

43P. Petit et al. / Cement and Concrete Research 60 (2014) 37–44

model experiments with silica particles, of solid fraction within thepaste ϕ (Eq. (2)). In the imbibition experiments, this concentrationevolves with time (as more powder is in contact with the liquid whenthe liquid front is rising). A complete characterization of surfactant dif-fusion and adsorption kinetics would allow to define the total dynamicsof wettability variations, which is not in the scope of our paper.

The solid bubble obtained with the composition of system (d) inTable 1with 8 × 10−4mol L−1 of SDS is observed by scanning electronicmicroscopy (SEM – Fig. 10). It shows that the solid film has a thicknessof 55± 5 μm. Therefore, the liquid film after drainage is filled with par-ticle multilayers homogeneously distributed. At larger magnification,filler (bright) and cement (gray) grains can be observed mixed withhydrated phase (CSH). These CSH phases are the signature that duringthe process, the liquid film retains a sufficient amount of liquid toallow hydration reaction to occur within the protective layer. Thesephases are mandatory to maintain good mechanical properties of thelayer. One must notice also that the observation of the solid bubblehas been performed 6 months after the beginning of the experiment,

a consequence of which is that Portlandite phase has disappearedbecause of its reactivity with ambient carbon dioxide [8].

4. Conclusion

The attachment of particles on a single bubble has been studied,firstlywith amodel system containing silica particles, and thenwith ce-mentitious particles. In the two cases, it was experimentally demon-strated that the particles stay on the liquid film if their surface ispartially hydrophobic. In order to obtain such surface properties, thesurfactants need to adsorb on the particles in a monolayer. Therefore,they can be chosenwith a charge opposite to the one of the particle sur-face, and with an intermediate concentration, slightly lower than theCMC, about 0.1 CMC. The experiment highlights also that an equilibriumstate of the particles on the bubble is reached in less than 1 min afterproduction: a cement bubble fully covered after this first phase resultsmost of the time in a closed solid bubble after setting.

The parameters governing particle adsorption dynamics at the inter-faces have also been identified, as they play a crucial role in the foamingprocess, i.e. to produce foam with particles at the industrial scale. Ourstudy shows that the interface must be in contact with the particles atleast 0.5 s to obtain, if conditions are optimized, a fully covered andstable bubble. This first demonstration of solidification of a thin cemen-titious liquid film is a first step in the achievement of a concrete foamat small solid fraction and with closed bubbles. However, in themacroscopic material, interactions of the thin filmswith foam structure(i.e. contacting channels between the bubbles, the so-called Plateauborders) must be taken into account.

Acknowledgments

Anne-Laure Biance thanks CNES for a partial funding of this projectthrough convention CNRS/CNES number 127233.

References

[1] Pathway to 2050, Energy and Climate Change, 2005.[2] K. Ramamurthy, E.K.K. Nambiar, G.I.S. Ranjani, A classification of studies on proper-

ties of foam concrete, Cement Concr. Compos. 31 (2009) 388–396.[3] I. Amato, Concrete solutions, Nature 494 (2013) 300–301.[4] P. Colombo, In praise of pores, Science 322 (5900) (2008) 381–383, http://dx.doi.

org/10.1126/science.1162962.[5] A.R. Studart, U.T. Gonzenbach, E. Tervoort, L.J. Gauckler, Processing routes to

macroporous ceramics: a review, J. Am. Ceram. Soc. 89 (6) (2006) 1771–1789,http://dx.doi.org/10.1111/j.1551-2916.2006.01044.x.

[6] L.J. Gibson, M.F. Ashby, Cellular Solids: Structure and Properties, 1999.[7] M.T. Hoang, C. Perrot, Solid films and transports in cellular foams, J. Appl. Phys. 112

(5) (2012) 054911, http://dx.doi.org/10.1063/1.4751345.[8] A. Neville, Properties of Concrete, Longman Group Limited, 1995.[9] P. Colombo, J.R. Hellmann, Ceramic foams from preceramic polymers, Mater. Res.

Innov. 6 (5–6) (2002) 260–272, http://dx.doi.org/10.1007/s10019-002-0209-z.[10] F.K. Juillerat, U.T. Gonzenbach, P. Elser, A.R. Studart, L.J. Gauckler, Microstructural

control of self-setting particle-stabilized ceramic foams, J. Am. Ceram. Soc. 94(2011) 77–83.

[11] A. Scheludko, B.V. Toshev, D.T. Bojadjev, Attachment of particles to a liquid surface(capillary theory of flotation), J. Chem. Soc. Faraday Trans. 1 (72) (1976) 2815–2828.

[12] N.D. Denkov, I.B. Ivanov, P.A. Kralchevsky, D.T. Wasan, A possible mechanism of sta-bilization of emulsions by solid particles, J. Colloid Interface Sci. 150 (2) (1992)589–593.

[13] G. Kaptay, On the equation of the maximum capillary pressure induced by solidparticles to stabilize emulsions and foams and on the emulsion stability diagrams,Colloids Surf. A: Physicochem. Eng. Aspect 282–283 (2006) 387–401.

[14] A.C. Martinez, E. Rio, G. Delon, A. Saint-Jalmes, D. Langevin, B.P. Binks, On the originof the remarkable stability of aqueous foams stabilised by nanoparticles: link withmicroscopic surface properties, Soft Matter 4 (2008) 1531–1535.

[15] K. Velikov, F. Durst, O. Velev, Direct observation of the dynamics of latex particlesconfined inside thinning water-air films, Langmuir 14 (1998) 1148–1155.

[16] U.T. Gonzenbach, A.R. Studart, E. Tervoort, L.J. Gauckler, Stabilization of foams withinorganic colloidal particles, Langmuir 22 (26) (2006) 10983–10988, http://dx.doi.org/10.1021/la061825a.

[17] Z.-G. Cui, Y.-Z. Cui, C.-F. Cui, Z. Chen, B.P. Binks, Aqueous foams stabilized by in Situsurface activation of CaCO3 nanoparticles via adsorption of anionic surfactant,Langmuir 26 (2010) 12567–12574.

[18] C. Chuanuwatanakul, C. Tallon, D.E. Dunstan, G.V. Franks, Controlling the micro-structure of ceramic particle stabilized foams: influence of contact angle and particle

44 P. Petit et al. / Cement and Concrete Research 60 (2014) 37–44

aggregation, Soft Matter 7 (24) (2011) 11464–11474, http://dx.doi.org/10.1039/c1sm06477k.

[19] L.R. Arriaga, W. Drenckhan, A. Salonen, J.A. Rodrigues, R. Iniguez-Palomares, E. Rio,D. Langevin, On the long-term stability of foams stabilised by mixtures of nano-particles and oppositely charged short chain surfactants, Soft Matter 8 (43)(2012) 11085–11097, http://dx.doi.org/10.1039/c2sm26461g.

[20] A. Maestro, E. Guzman, E. Santini, F. Ravera, L. Liggieri, F. Ortega, R.G. Rubio, Wetta-bility of silica nanoparticle-surfactant nanocomposite interfacial layers, Soft Matter8 (3) (2012) 837–843, http://dx.doi.org/10.1039/c1sm06421e.

[21] H.J. Jin, W.Z. Zhou, J. Cao, S.D. Stoyanov, T.B.J. Blijdenstein, P.W.N. de Groot, L.N.Arnaudov, E.G. Pelan, Super stable foams stabilized by colloidal ethyl celluloseparticles, Soft Matter 8 (7) (2012) 2194–2205, http://dx.doi.org/10.1039/c1sm06518a.

[22] B. Cabane, S. Hénon, Liquides: Solutions, dispersion, émulsions, gels, Echelles,2003.

[23] N. Dorsey, Properties of Ordinary Water-substance, Reinhold, New York,1940.

[24] E.W. Washburn, The dynamics of capillary flow, Phys. Rev. 17 (3) (1921) 273–283,http://dx.doi.org/10.1103/PhysRev.17.273.

[25] P.-G. de Gennes, F. Brochard-Wyart, D. Quéré, Gouttes, bulles, perles et ondes,Echelles, 2004.

[26] I.M. Krieger, T.J. Dougherty, A mechanism for non-Newtonian flow in suspensions ofrigid spheres, Trans. Soc. Rheol. 3 (1959) 137–152.

[27] 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 (2) (2008)301–312, http://dx.doi.org/10.1016/j.jcis.2008.04.052.

[28] N.S. Klein, J. Bachmann, A. Aguado, B. Toralles-Carbonari, Evaluation of thewettability of mortar component granular materials through contact anglemeasurements, Cem. Concr. Res. 42 (12) (2012) 1611–1620, http://dx.doi.org/10.1016/j.cemconres.2012.09.001.

[29] I. Tanaka, M. Koishi, K. Shinohara, Evaluation of the wettability of spherical cementparticle surfaces using penetration rate method, Cem. Concr. Res. 32 (7) (2002)1161–1168, http://dx.doi.org/10.1016/S0008-8846(02)00758-5.

[30] K. Yamada, S. Ogawa, S. Hanehara, Controlling of the adsorption and dispersing forceof polycarboxylate-type superplasticizer by sulfate ion concentration in aqueousphase, Cem. Concr. Res. 31 (3) (2001) 375–383, http://dx.doi.org/10.1016/S0008-8846(00)00503-2.

[31] Y.F. Houst, P. Bowen, F. Perche, A. Kauppi, P. Borget, L. Galmiche, J.F. Le Meins, F.Lafuma, R.J. Flatt, I. Schober, P.F.G. Banfill, D.S. Swift, B.O. Myrvold, B.G. Petersen, K.Reknes, Design and function of novel superplasticizers for more durable highperformance concrete (superplast project), Cem. Concr. Res. 38 (10) (2008)1197–1209, http://dx.doi.org/10.1016/j.cemconres.2008.04.007.

[32] L. Ferrari, J. Kaufmann, F. Winnefeld, J. Plank, Multi-method approach to study influ-ence of superplasticizers on cement suspensions, Cem. Concr. Res. 41 (10) (2011)1058–1066, http://dx.doi.org/10.1016/j.cemconres.2011.06.010