Reinforcing Effect of Carbon Nanotubes / Surfactant Dispersions in Portland Cement Pastes

Decoupling the individual effects of multiwalled carbon nanotubes (MWCNTs) and surfactants when used as reinforcement materials in cement-based composites is aimed in this study. Powder MWCNTs were dispersed in deionized water using different types of surfactants as chemical dispersing agents and an ultrasonic tip processor. Cement pastes with carbon nanotubes additions of 0.15% bymass of cement were produced in two steps: first, theMWCNT/surfactant dispersions were combined with the mixing water, and then, cement was added andmixed until a homogeneous paste was obtained. Mechanical properties of the pastes cured at 7 days were measured, and their fracture behavior was characterized using the linear elastic finite element analysis. It was found that the reinforcing effect of MWCNT was masked by the negative effect of surfactants in the cement matrix; nevertheless, nanotubes were capable of increasing both stress and strain capacity of the composite by controlling the crack propagation process at the tip of the crack.


Introduction
e mechanism by which multiwalled carbon nanotubes (MWCNTs) are able to enhance the mechanical properties of a cement matrix is called "bridge effect."MWCNTs are capable of creating a network of bridges that transmit load across cracks and pores [1], enhancing the load distribution within the matrix.is depends heavily on the length and aspect ratio of nanotubes [2].Finite element modeling of beams under flexural stress has proposed that MWCNTs control crack propagation, allowing additional strain of the matrix by generating a sequential pullout of nanotubes while the crack develops, until complete pullout of the nanotubes is reached [3].
To guarantee adequate crack propagation control, MWCNTs must be well dispersed in the cement matrix [4], bridging the correct phases.is means that if MWCNTs agglomerate around individual hydrates instead of bridging neighboring ones, they will not have any effect on the stress distribution within the composite [5]. is is why the use of chemical dispersing agents, such as surfactants, is of prime importance to produce MWCNT-reinforced cement-based composites.Nevertheless, it is known that surfactants are able to interact with cement grains, modifying their hydration reaction [6,7].
It is not clear that what is the extent of the reinforcing efficiency of MWCNT, but it has been found that small amounts of nanotubes increase the flexural strength of cement-based matrices [8], identifying a correlation between their aspect ratio and reinforcing efficiency [9][10][11].Additional parameters such as modulus of rupture [12], fracture energy, and fracture toughness [13] have also been identified to increase with the presence of MWCNT in the matrix.
On the contrary, some literature reports have shown decreases in mechanical properties of cement-based matrices.is behavior has been attributed to agglomeration phenomena of MWCNT when used in high volumes [14], to the presence of functional groups on the surface of the nanotubes able to interact chemically with the hydration reaction [12], or even to the presence of fine aggregates in the matrix for the case of mortars [15].ese conflicting reports are still a matter open for discussion [16], and the issue is a subject of several experimental and numerical studies [17][18][19][20][21][22][23]. is work aims at quantifying the reinforcing efficiency of MWCNT/surfactant dispersions, separating the effects induced by surfactants in the mechanical properties of the matrix from those induced by MWCNT.

Materials and Methods
e materials used were class G high sulfate-resistance (HSR) cement, industrial-grade pristine multiwalled carbon nanotubes (MWCNTs), sodium dodecyl sulfate (SDS), cetylpyridinium chloride (CPC), Triton X-100 (TX-100), and a viscosity modifying agent (VMA).Characterization of cement and VMA was performed by X-ray diffraction (XRD) and X-ray fluorescence (XRF).XRD tests were carried out in a diffractometer equipped with a Cu radiation source (Kα, λ � 1,5418 Å), from 4 to 70 °2θ with a step size of 0.05 and an accumulation time of 30 seconds.e XRF test was carried out using an X-ray fluorescence spectrometer capable of identifying elements from Na to U. Morphological characterization of the MWCNT was carried out using scanning transmission electron microscope (STEM) in the transmission mode with 20 kV acceleration and 225,000x magnification.MWCNTs were dispersed in deionized water using an ultrasonic bath; 1 ml of this dispersion was dropped in a #300 copper Formvar mesh and dried in a desiccator before imaging.
Deionized water was used to prepare MWCNT dispersions with 0.33% solid concentration of nanotubes using the three types of surfactants in 1 mM, 10 mM, and 100 mM concentrations.Surfactants were poured into the water and stirred for 5 minutes, and then solid MWCNTs were added to the mixture.An ultrasonic tip was used to apply a total energy of 390 J/g in a 500 W ultrasonic processor set to 40% amplitude in 20 seconds on/off cycles to avoid overheating the samples.ree families of pastes were produced.A reference sample (Ref.)composed of cement and VMA at a water-tocement ratio (w/c) of 0.55 and 1.0% VMA by mass of cement.e VMA was used as a stabilizing agent to guarantee that there is no static sedimentation of the pastes during curing.e second family, named plain (P), maintained the same w/c and VMA amount and was composed of water, surfactant, cement, and VMA; surfactant type and concentration were varied.e third family, named blended (B), maintained the same proportion as the P family, but 0.15% MWCNT by mass of cement was blended into them.
Pastes were prepared by adding water, VMA, and the MWCNT/surfactant dispersions to the mixer.A 500 rpm rotation was maintained for 1 minute, cement was added, and rotation was increased to 2070 rpm during 1 minute and maintained constant for 5 minutes.Immediately after mixing, pastes were molded in 115 × 115 × 25 mm slabs, vibrated for 60 seconds in a vibrating table, and cured in a high-humidity environment for 7 days.
At the testing stage, each slab was cut with a diamond saw to produce specimens for three-point bending.e obtained samples were 115 × 25 × 25 mm beams with one 5 mm notch placed at midspan in one of the 25 mm faces.Two metal plates were glued at the edge of the notch to couple a clip gauge to measure crack mouth opening displacement (CMOD).e load application rate was varied to maintain a constant deformation rate at the CMOD of 0.03% of the initial gauge length per minute.Testing was carried out in a Shimadzu AGX 100 universal testing machine equipped with a 100 kN load cell.e experimental setup is presented in Figure 1.Flexural strength was calculated using the following equation: where P is the applied load, S is the span (90 mm), B is the beam width (25 mm), W is the beam height (25 mm), and a 0 is the notch depth (5 mm).Changes in the energy adsorption capacity (D) of the paste induced by the presence of MWCNTs were measured in the load versus deflection plots by adapting the procedure recommended by the RILEM technical committee 162 on test and design methods for steel fiber-reinforced concrete [24].A typical load versus the deflection curve used for this procedure is presented in Figure 2. D was calculated as the area under the load-deflection curve after the maximum load, and it consists of two components: the contribution 2 Advances in Civil Engineering from the cement paste (D1) and the contribution from the MWCNT (D2).D1 was approximated as a right triangle, and the slope of its hypotenuse was measured in pastes from the P family.is slope was then used to obtain D2 for the MWCNT-blended pastes (B family) by subtracting D1 from D.
Linear elastic fracture mechanics was used to characterize the in uence of MWCNTs on the fracture behavior of the paste.is was achieved by employing a two-parameter facture model for concrete [25] to interpret the results of three-point bending tests on notched beams of the rectangular section. is model is based on the fact that the critical stress intensity factor (K Ic ) cannot be used to predict accurately the growth of a crack in quasibrittle materials such as concrete due to nonlinear slow crack growth before reaching the peak load [25].e proposed model states that an e ective crack extends before failure, which occurs only when the critical stress intensity factor K S I at the tip of the e ective crack reaches a critical value K S Ic and, simultaneously, the crack tip opening displacement reaches its critical value (CTOD c ).
Compliance (C) was de ned as C P/CMOD.Two compliance parameters were used, initial compliance (C i ) and failure compliance (C u ), which correspond to the compliance at the initial crack length and at failure, respectively [26].ese parameters were obtained from experimental load versus CMOD plots from the slope of two linear ts as presented in Figure 3.
Linear elastic nite element simulations were used to obtain K S Ic , CTOD c , and E values from the two-parameter model.
e generic geometry used for the simulations is presented in Figure 4. e CMOD, CTOD, and K I curves for the actual notch beam geometries were obtained as a function of the e ective crack extension (a c − a 0 ), as shown in Figure 5.An initial simulation was performed by applying P 100 N and assuming E 10 GPa.Young's modulus (E) of the material was calculated by matching the observed C i .e CMOD c of the material was calculated by matching the observed C u .e stress intensity factor was calculated using the energy release rate given by the J integral.e e ective crack length a c was determined by matching the observed CMOD at failure in the CMOD versus a c curve.e critical values CTOD c and K S Ic were obtained directly from the simulation results at a a c .

Results and Discussion
is section presents the characterization of the raw materials used to produce cement pastes blended with MWCNT/surfactant dispersions; then, the individual e ect of surfactants and MWCNT on the mechanical properties of the pastes is presented and discussed.

Materials Characterization.
Chemical and mineralogical composition of cement and VMA was characterized using XRD and XRF. e obtained results are presented in Figure 6 and Table 1.Regarding the cement, the typical anhydrous phases were found, as well as anhydrite (CaSO 4 ) and gypsum (CaSO 4 •2H 2 O) which were used as a setting time controller.Regarding the VMA, it was found that it is mainly composed of cellulose mixed with calcium and sodium chloride.An image of individual MWCNT is presented in Figure 7. Nanotubes were found to have a diameter of approximately 20 nm and typical length between 1 and 2 µm.For more details on the physical characteristic of the MWCNT, refer [27].
SDS was acquired in powder with a purity of 96% and a molecular weight of 288.38 g/mol.CPC was in powder form with 100% purity and a molecular weight of 358.0 g/mol.TX-100 was in liquid form with 98% purity and a molecular weight of 646.85 g/mol.ese characteristics are according to the manufacturer.

E ect of Surfactants in the Density of Cement Paste.
Apparent density was measured by water absorption for all the pastes from the B family after 7 days of curing; the obtained results are presented in Figure 8.It was found that density was dependent on the surfactant concentration used to disperse the MWCNT.A higher surfactant concentration led to a lower density, decreasing from 1.29 g/cm 3 for the Advances in Civil Engineering reference sample to 1.01 g/cm 3 for the 10 mM SDS concentration.
is behavior can be related to a decrease in surface tension of water caused by surfactants.Air bubbles are naturally entrapped during the mixing process, and surfactants are able to adsorb on the water-air interface of these bubbles with their hydrophobic tails towards air and their hydrophilic/polar groups towards water [28]; this stabilizes the bubbles and does not allow them to coalesce [28], which generates small nonconnected bubbles in the paste.e volume of such bubbles is signi cant enough to lower the density of the paste.

Reinforcing E ect of MWCNT.
Since surfactants play such an important role on the density of the pastes, threepoint bending tests were carried out in pairs of pastes with the same surfactant concentration, with (B) and without (P) MWCNT.e goal of this experiment was to identify the individual e ects of surfactants and MWCNT on the mechanical behavior of the matrix.Previous works by the    authors identi ed that for the sonication process described in the methodology, 10 mM, 100 mM, and 100 mM are the most e cient surfactant concentrations to disperse MWCNT in water for TX-100, SDS, and CPC, respectively [29]; thus, only these concentrations were used to produce samples for mechanical testing.Flexural strength results for all pastes are presented in Figure 9.It was found that exural strength decreased for all samples when compared to the reference, regardless of the presence of MWCNT in their composition.Nevertheless, when comparing pairs of pastes with the same surfactant type and concentration, it can be seen that MWCNTs are able to improve the exural strength of the matrix, increasing it up to 54% for the SDS case.Apparent density was also measured for the same pastes tested in three-point bending; results are presented in Figure 10.As previously found, the presence of surfactants decreased apparent density, but pastes blended with MWCNT (B) presented an increase in these parameters when compared to their plain homologous with the same surfactant type and concentration (P). is increase can be related to a lower surfactant availability to stabilize bubbles in the matrix since a portion of the surfactant will adsorb on MWCNTs during sonication [30] and will not be available to adsorb on the water-air interface to stabilize bubbles.
A clear linear correlation between apparent density and exural strength was found for each group of pastes with the same surfactant type and concentration, with and without MWCNT; this is presented in Figure 11.It can be seen that for each surfactant type, plain pastes always present the lowest density and the lowest exural strength, while

Advances in Civil Engineering
MWCNT-blended pastes increase density and proportionally increase exural strength.
is indicates that an apparent reinforcing e ect from MWCNTs can be partially related to changes in density of the paste rather than to a better load transfer between cracks due to the MWCNT.
Load versus CMOD plots for all samples tested in threepoint bending are presented in Figure 12.All samples showed the same general behavior.As load was applied, there was an elastic section where deformation varied linearly with load up to the maximum load where cracking begins.is was followed immediately by a nonlinear crack growth process and a strain-softening behavior, where CMOD increases and load decreases until brakeage of the sample.No strain-hardening behavior was found in any of the samples, meaning that no multiple cracking occurred due to the presence of MWCNT in the matrix.It can be seen that the deformation at which the elastic behavior ends remains roughly constant for all the samples, regardless of the presence of surfactant or MWCNT, while the maximum deformation of the sample was increased by MWCNT, as is indicates that the enhancement in deformation induced by MWCNT is mainly in the strain-softening branch, affecting the crack opening and increasing ductility of the matrix.
Energy absorption capacity results for the pastes blended with MWCNT are presented in Table 2.It was found that up to 80% of the energy absorption capacity after peak load was contributed by the MWCNT, while the contribution from the cement matrix (D1) remained approximately constant at 0.20 N•mm.e contribution from the MWCNT (D2) varied for each type surfactant due to their different dispersion efficiencies, which depend on the polar nature and the length of the molecule [31,32].e reinforcing effect of MWCNT is expected to depend strongly on the dispersion degree of the individual nanotubes throughout the cement matrix [4].
Results obtained from the application of the twoparameter fracture model are presented in Table 3.It was found that E and K S Ic decreased with the presence of surfactants (P family), but a portion of this decrease was recovered with the addition of MWCNT (B family).Similarly to the flexural strength results, the decrease in mechanical performance of the P family can be related to a lower density of the matrix, and the partial recovery can be related to the reinforcing effect of MWCNT and to lower surfactant availability so stabilize bubbles.
Critical crack length (a c ) and CTOD c were increased both by surfactants and MWCNT.e increase caused by surfactants can be related to a lower E, which allows a higher deformation of the material before rupture.
e increase caused by MWCNT, when comparing results from P and B families of pastes, can be associated with their reinforcing effect.It should be noticed that MWCNT increased CTOD c by less than 1 µm, which is compatible with their typical length found in the microscopy imaging.

General Discussion
e results found in this work showed that when using MWCNT/surfactant, aqueous dispersions surfactants decreased the density of the paste and consequently the mechanical performance of the composite.is decrease of mechanical performance was partially recovered by the presence of MWCNT, which were able to increase the strain and energy absorption capacities of the matrix, as evidenced by all the deformations measured and calculated, that is, maximum CMOD, CTOD c , and K S Ic .When comparing all the results for pastes with the same surfactant type and concentration, with and without MWCNT, the reinforcing effect of nanotubes can be clearly identified.Such effect can be divided into three different moments: (i) during the elastic regime of the deformation, where it was found that MWCNTs are capable of increasing the modulus of elasticity of the matrix due to a combination of their reinforcing effects and a lower availability of surfactant; (ii) during the crack propagation process, where it was found that MWCNTs are capable of increasing the critical opening and therefore the effective length of the crack; and (iii) during the strain-softening regime of the deformation, where it was found that MWCNTs are capable of increasing the maximum deformation and energy absorption capacity of the matrix.No strain-hardening behavior was found in any of the samples and no multiple cracking was identified, probably due to a scale incompatibility between the nanotube length and the crack width.
e reinforcing effect of MWCNTs on the mechanical behavior of cement paste was much lower than the negative effect of surfactants; this is probably due to the lack of structures that transmit tensile load at the interfaces between MWCNT and hydration products [4].
is limits the reinforcing efficiency of MWCNTs to the amount of load that the interfaces are capable of transmitting before suffering pullout, regardless of the tensile strength of the nanotubes.

Conclusions
It can be concluded that when using MWCNT/surfactant aqueous dispersions in a cement paste, the reinforcing effect of MWCNTs is masked by the negative effect of the surfactants.An adequate comparison between pastes with the same type and amount of surfactant, with and without nanotubes, is necessary to identify the net reinforcing effect of nanotubes.e increase in energy absorption capacity due to the presence of MWCNTs in the matrix could not be related to strain-hardening or multiple cracking behaviors, and this due a scale incompatibility between the nanotube length and the crack width.Nevertheless, MWCNTs are capable of increasing both stress and strain capacity of the cement matrix by controlling the crack propagation process at the tip of the crack.

Figure 1 :
Figure 1: Experimental setup used to measure flexural strength by three-point bending.

Figure 2 :Figure 3 :
Figure 2: Typical load versus de ection curve used to calculate energy absorption capacity.

Figure 4 :
Figure 4: (a) General geometry and (b) detail of the notch and crack used in the nite element simulations (scale factor: 200).

Figure 12 :
Figure 12: Load versus CMOD for all pastes tested in three-point bending.

Table 1 :
XRF chemical composition of cement and VMA (LOI: loss on ignition).

Table 3 :
Fracture mechanics analysis results.