Construction and Building Materials 134 (2017) 336–345

Contents lists available at ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier .com/locate /conbui ldmat

Enhanced mechanical properties of cement paste by hybrid grapheneoxide/carbon nanotubes

http://dx.doi.org/10.1016/j.conbuildmat.2016.12.1470950-0618/� 2016 Elsevier Ltd. All rights reserved.

⇑ Corresponding author.E-mail address: msfxli@scut.edu.cn (F. Li).

Cheng Zhou, Fangxian Li ⇑, Jie Hu, Mengmeng Ren, Jiangxiong Wei, Qijun YuSchool of Materials Science and Engineering, South China University of Technology, Guangzhou 510640, People’s Republic of China

h i g h l i g h t s

� GO was used as a dispersant to disperse CNTs.� The mechanical properties of cement paste were improved by the hybrid GO/CNTs.� The hybrid GO/CNTs could refine the pores of the matrix.

a r t i c l e i n f o

Article history:Received 16 August 2016Received in revised form 25 October 2016Accepted 24 December 2016

Keywords:Graphene oxideCarbon nanotubesDispersionCement pasteMicrostructure

a b s t r a c t

The prominent mechanical properties of carbon nanotubes (CNTs) make them to be potential candidatesfor enhancing cementitious material. However, CNTs are hard to be dispersed uniformly due to the Vander Waals’ force between them. Many methods have been attempted by researchers to improve the dis-persion of CNTs. This research investigated the use of graphene oxide (GO) as a dispersant for CNTs toovercome the obstacles. UV–vis spectroscopy and scanning electron microscopy (SEM) results revealedthat CNTs were highly dispersed by GO rather than by surfactants. More importantly, the excellent rein-forcing capabilities of the hybrid GO/CNTs were demonstrated by the enhanced fracture resistance prop-erties of the cementitious matrix. The hybrid GO/CNTs (0.02 wt% GO and 0.04 wt% CNTs by weight ofcement) with poly-carboxylate superplasticizer had a great contribution to improving the compressiveand flexural strength of cement paste by 23.9% and 16.7%, respectively, which was higher than cementpaste reinforced by CNTs or GO. Scanning electron microscopy images showed that the hybridGO/CNTs were well dispersed in the cement hydration products. Furthermore, mercury intrusionporosimetry (MIP) results indicated that the addition of the GO/CNTs promoted the refinement of poresin cement paste.

� 2016 Elsevier Ltd. All rights reserved.

1. Introduction

Cementitious materials are extensively used worldwide forbuilding and construction. However, cementitious materials aresusceptible to cracking. The cracking process within cement beginswith isolated nano-cracks, which then conjoin to form micro-cracks and in turn macro-cracks [1]. Traditional fibers (e.g. steelfiber or PVA fiber) show good crack resistance and restrain crackpropagation in the macro scale, but were invalid for delayingmicro-crack initiation. Recently, achievements in nanotechnologyhave produced nanofibers (e.g. carbon nanotubes and grapheneoxide) that could be used as reinforcements to move the reinforc-ing behavior from the macroscopic to the nano-scopic level.

Compared to traditional fibers, carbon nanotubes (CNTs)regarded as a one-dimensional tube exhibit excellent mechanicalproperties and high aspect ratio ranging from 30 to more thanmany thousands [2,3]. The unique mechanical properties makethem attractive for use as reinforcement to develop superiorcementitious composites. However, the incorporation of high con-centration CNTs in cement composites had proven to be re-agglomerate which will leads to weak bond in the microstructure[4]. CNTs tend to agglomerate, bundle together and entangle dueto strong Van der Waal’s attractive forces between particles. There-fore, dispersing CNTs uniformly in cementitious materials is prob-ably the most critical issue for a successful production ofCNTs/cement composite. Junqing Zuo et al. [5] used sodium dode-cyl benzene sulfonate (NaDDBS) as surface-active agents to obtainuniform dispersed CNTs with the help of sonication, and found thatthe cement pastes with addition of 0.5 wt% MWCNTs couldincrease in the compressive strengths by 18.4%. But the treatment

C. Zhou et al. / Construction and Building Materials 134 (2017) 336–345 337

time or energy of the ultra-sonication process had strong influenceon CNT dispersion. Shama Parveen et al. [6] reported that the flex-ural and compressive strengths of mortar increased by 7% and 19%through adding 0.1% single walled nanotube (SWCNT) with a noveldispersing agent, respectively. Some surfactants were successfullyused to disperse carbon nanomaterials in other matrices have beenreported [7]. Zoi S. Metaxathe et al. [8] also found that effective dis-persion of different length MWCNTs in water was achieved byapplying ultrasonic energy and in combination with the use of asurfactant. However, some surfactants of CNTs maybe entrap airin the cement paste. It was observed that the application of sodiumdodecyl sulfate (SDS) as a surfactant of CNTs led to a severe drop inthe strength of the hardened cement, which could be explained bythe high porosity of the samples containing SDS caused by the for-mation of foam [9]. In addition, a lot of work had been done on theamination, fluoration, and long alkyl chain grafting of the carbonnanotubes through the chemical reactions [10,11]. The chemicalfunctionalization of CNTs played an important role in improvingthe bonding of fibers and cementitious matrices. Acid treatmentof carbon nanotube had also been found to benefit dispersion ofthe CNTs in aqueous media. Su-tae Kang et al. [12] reported thatworkability of the cement composites decreased with acid treat-ment of CNTs, compressive and tensile strength improved signifi-cantly. However, the functionalized CNTs due to the chemicalmodifications methods might cause a structure damage to CNTs,which resulted in a shorter length, a smaller diameter and aroughen surface in CNTs.

Similar to CNTs, graphene oxide (GO) with a unique atom-thicktwo-dimensional structure has excellent mechanical properties.The intrinsic strength and Young’s modulus are estimated to100 GPa and 1 TPa, respectively [13]. Moreover, GO is an excellenthydrophilic material with oxygen-containing functional groupssuch as hydroxyl, carbonyl and carboxyl, which could be dispersedwell in water [14]. The unique structure and high surface area ofGO could be beneficial for improving bonding between graphenesheets and cement products. Zhu et al. [15] demonstrated thatthe incorporation of 0.05 wt% GO in the cement paste increasedthe compressive and flexural strength by 33% and 41%, respec-tively. Shenghua Lv et al. [16] reported that the cement compositesexhibited remarkable increase in tensile strength (78.6%), flexuralstrength (60.7%) and compressive strength (38.9%) comparing withthose without GO, when the content of GO was 0.03%. This hap-pened because the oxygen-containing function groups providedadsorption sites for water and cement, leading to the crystal nucle-uses for cement hydrates [17]. Zeyu Lu et al. found that the addi-tion of 0.05 wt% GO could improve the compressive and flexuralstrength of the magnesium potassium phosphate cement (MKPC)paste by 6.8% and 8.3%, respectively, compared with the freshMKPC paste [18].

So far, there have been a lot of previous researches on CNTs orGO enhanced cementitious materials. But the study on co-effects ofGO/CNTs on the mechanical behavior of cementitious materials isstill few. In fact, the hybrid GO/CNTs has been extensively studiedin the electrode and solar cells. Hejie Song et al. [19] prepared a‘‘clean” three-dimensional architecture consisting of two-dimensional reduced graphene oxide (rGO) and one-dimensionalCNTs supporting zero-dimensional Pd nanoparticles by using GOas a surfactant to disperse CNTs, which could be use as highlyactive and stable electro-catalysts. Leping Yu et al. [20] used GOsheets as the surfactant to disperse single-walled carbon nan-otubes in water, and prepared GO/CNT electrodes that could beused in solar cells. Compared to CNTs or GO, the hybrid GO/CNTsmay be better reinforcements in cement materials.

The aim of this paper is to investigate the dispersion of thehybrid GO/CNTs and the mechanical behavior of cement pastereinforced by GO/CNTs. The dispersion of CNTs used GO as a dis-

persant was characterized by Ultraviolet spectrophotometer(UV-s). And the properties of GO/CNTs/cement composites,involving Fourier transform infrared spectroscopy (FT-IR), Ramanspectra, X-ray diffraction (XRD), scanning electron microscopy(SEM) andmercury intrusion porosimetry (MIP), were investigated.

2. Experimental

2.1. Materials

Ordinary Portland cement (OPC) type 42.5 was used in this paper. Multi-wallcarbon nanotubes (MWCNT) with an outer diameter of 30–50 nm, was suppliedby Chengdu Institute of Organic Chemistry, Chinese Academy of Sciences, China.The chemical compositions of cement and the physical properties of the CNTs wereshown in Tables 1 and 2, respectively.

2.2. Preparation of GO

Graphene oxide was prepared from graphite flakes (325 mesh, Qing Dao)according to the improved Hummers’ method [21]. First of all, graphite powder(3 g) and sodium nitrate (3 g) were added to 100 ml concentrated H2SO4 in an icewater-bath (below 5 �C). Successively, KMnO4 (9 g) was gradually added under stir-ring for 30 min. Then the mixture was transferred to water-bath (35 �C) for 5 h.Deionized water (100 ml) was added into the mixture, followed by heating to98 �C for 15 min. Finally, deionized water (230 ml) and 30% H2O2 solution (5 mL)was added to terminate the oxidation reaction. And the color of the solution turnedfrom dark brown into yellow. After air cooling, the obtained sample was filtered andwashed successively with 10% HCl aqueous solution to remove metal ions, and thencentrifuged successively with warm deionized water completely until sulfate couldnot be detected with BaCl2.

2.3. Preparation of GO/CNTs/cement composite

First, GO suspension (100 ml) was exfoliated by ultrasonic (420 W) for 30 min.The concentration of MWCNTs adding in cement paste was varied from 0 to 0.08 wt% by weight of cement. And the ratio of GO to CNTs was set as 0.5. Further, corre-sponding quality of CNTs was added into the GO suspension to disperse by ultra-sonic (300 W) for another 30 min. And the ultrasonic was operated at a lowerenergy in order to prevent overheating of the hybrid suspensions, which mightdamage the structure of CNTs [22]. Based on the water to cement ratio (w/c) of0.4, extra water and the GO/CNTs suspension were added to the cement powderfor mixing. The mixing process and casting procedures of all the cement mixtureswere similar. For compressive strength tests, the cement mixtures with or withoutpoly-carboxylate superplasticizer (0.5 wt% of cement) were poured into the20 mm � 20 mm � 20 mm steel molds. Specimens with a size of25 mm � 25 mm � 140 mm were prepared for flexural strength tests. All the spec-imens were cured at a temperature of 20 �C and humidity of 98%. After demolding(24 h later), the specimens were stored in water at laboratory temperature for dif-ferent curing ages.

2.4. Measurement

2.4.1. Tests of dispersion of hybrid GO/CNTsUV–Vis absorption spectroscopy (Agilent Cary 60) was used to measure the sta-

bility of GO, CNTs and GO/CNTs in water at a scanning speed of 60 nm/min. The zeta(f) potentials of GO aqueous dispersions with CNTs or without were measured bythe use of a zeta sizer nano-system (Malvern Instruments). XRD and Raman spectrawas used to characterize the hybrid GO/CNTs. For the morphology observations ofGO/CNTs, the dispersion of CNTs and GO/CNTs was measured by stereo microscopyand SEM (Carl Zeiss, Germany).

2.4.2. Mechanical property and micro-structure testsThe compressive and flexural strength tests of the specimens at different curing

ages were carried out using a universal testing machine. Compressive and flexuralstrength of the specimens were performed at a constant strain rate of 2 mm/minand 0.1 mm/min, respectively. To measure the effects of GO/CNTs on themicrostructure of cement paste, SEM and MIP were used.

3. Results and discussion

3.1. Characterization of graphene oxide (GO)

The XRD pattern (Fig. 1a) of GO and graphite gave reflectionpeak at 2h = 10.3� and 2h = 26.5�, corresponding to interlayer spac-ing of 0.86 nm and 0.34 nm, respectively. The inter layer spacing

Table 1Chemical compositions of cement.

CaO SiO2 Al2O3 Fe2O3 SO3 MgO K2O TiO2 MnO Na2O SrO P2O5 NiO LOSS

65.01 20.69 5.71 4.18 2.52 0.61 0.61 0.31 0.16 0.08 0.05 0.03 0.03 3.23

Table 2Physical properties of multi-walled carbon nanotubes (MWCNTs).

Out diameter Length Purity Ash Specific surface face Bulk density

30–50 nm 10–20 lm >95% <1.5 wt% >60 m2/g 0.22 g/cm3

Fig. 1. (a) XRD spectra of graphite and graphite oxide, (b) IR of graphite and graphite oxide.

338 C. Zhou et al. / Construction and Building Materials 134 (2017) 336–345

distance (0.86 nm) markedly exceeds that of pristine graphite(0.34 nm). Moreover, the peaks intensity of GO decreased and thesharp peaks became broadening. Shenghua Lv et al. [16] suggestedthat the oxygen functional groups had penetrated into the graphiteinterlayers, thereby helping to weaken the interaction between thelayers. This was the fundamental reason why the oxidized graphitewas easy to be exfoliated to form GO under sonication. In addition,FT-IR spectroscopy measurements were used to confirm the func-tional groups onto GO sheets (Fig. 1b). The bands that appeared atabout 3400 cm�1 for graphite and GO were assigned to the AOHstretching vibration because of the existence of hydrolytic groupsand residual water [23]. An absorption peak for the C@C stretchwas observed at 1634 cm�1. And carbonyl groups (C@O) were alsopresent with peaks at 1720 cm�1, confirming the ACOOH group.This results indicated that GO could be dispersed well in waterbecause of the hydrolytic groups. The SEM images of GO revealeda typical wrinkled morphology of GO sheets (Fig. 2a and b), whichpossessed a huge surface area. And the wrinkled surface could playa role in mechanical interlocking between GO sheets and cementmatrix, thereby contributing to enhancing the interfacial loadtransfer and achieving better bond strength between GO sheetsand cement matrix.

3.2. Dispersion of hybrid GO/CNTs in solution

To test the stability of the CNT in water or GO suspensions, theupper suspension was used for UV–vis Spectrophotometry test.According to the Beer–Lambert law given as Abs ¼ ecl, where

Abs is the absorbance of the solution, e is the extinction coefficientand l = 1 cm is the path length of the light. The absorbance (Abs) ofthe solution is in proportion to the concentration of the solvent. Itis well known, however, that CNTs tend to agglomerate in aqueous,and consequently the concentration of well-dispersed CNTs isn’tconstantly equal to the dosage of CNTs in aqueous [24]. In otherwords, the better dispersion of carbon nanotubes is, the higherAbs is.

Fig. 3a illustrated that the UV–vis absorption spectrum of GOsuspension at different period of static time. GO showed the peakat 233 nm and a shoulder at 300 nm, which were correspondedto p–p⁄ transition of C@C bonds and n–p⁄ transition of the C@Obonds, respectively [25]. After different static time, the Abs forGO suspension varied negligibly at peak of 233 nm. The reason thatGO sheets could form a stable colloidal dispersion was the hydro-philic functional groups (ACOOH and AOH) and the electrostaticrepulsion. Zhen Xu et al. [26] confirmed that the intrinsic electro-static repulsive force of GO played a dominated role in the aqueousaccording to the zeta (f) potential tests. Moreover, Xiluan Wanget al. [27] found that GO dispersion was stable only as its f poten-tial <�30 mV. By testing, the f potential of GO suspension wastested by �46.7 mV, indicating that the suspension was verystable. Compared with GO, the Abs of CNTs suspension was moni-tored at different period of static time (Fig. 3b). It could be seenthat CNTs showed a characteristic absorption peak at 265 nm.The Abs of CNTs suspension varied from 0.9 (2 h) to 0.617(144 h) at the characteristic absorption peak, which sharplydecreased by �31%. And the f potential of CNTs suspension was

Fig. 2. SEM images of the GO sheets (a) 5 kx, (b) 10 kx.

Fig. 3. Ultraviolet spectrophotometry of (a) GO (0.1 mg/mg) suspension and (b) CNTs (0.1 mg/ml) suspension.

C. Zhou et al. / Construction and Building Materials 134 (2017) 336–345 339

tested by �13.6 mV, which was agreed with the experimental datareported before [28]. These results indicated that the CNTs had aserious tendency to agglomerate in water due to the strong Vander Waals’ force.

Fig. 4. Ultraviolet spectrophotometry of hybrid CNTs/GO suspension (0.2 mg/mlCNTs and 0.1 mg/ml GO).

In order to overcome the agglomeration of CNTs, GO as a disper-sant was expected to solve this problem. Fig. 4 demonstrated thatthe Abs of the hybrid CNTs/GO significantly increased. More impor-tantly, the Abs of the colloidal hybrid CNT/GO suspensionremained nearly constant, which was monitored by UV–visabsorption over a period of 24 h. Compared to Abs of GO or CNTs,the higher Abs of the hybrid GO/CNTs implied that CNTs werehighly dispersed in GO suspension under sonication. After 24 h sta-tic, the Abs slightly decreased because a part of the hybrid CNTs/GO precipitated in a certain extent. Experiment had affirmed thateven the precipitation of the CNTs/GO could be readily re-dispersed in water by mild sonication. As was the case with CNTs,sonication alone didn’t disperse CNTs in aqueous solution. How-ever, the GO could assist the dispersion of CNTs because of thestrong electrostatic repulsions between the negatively chargedGO sheets [27]. Such an explanation could be confirmed by testingthe zeta potential of the hybrid GO/CNTs (�47.6 mV), which indi-cated that the hybrid GO/CNTs suspension was a highly stable sys-tem. A decrease of the zeta potential of GO/CNTs caused byaddition of CNTs, which meant that GO might adsorb onto theCNTs to form a steric effect and stabilize the CNTs against Vander Waals attraction [28,29].

XRD patterns and Raman spectra were employed to furtherinvestigate the interaction between GO sheets and CNTs (Fig. 5).The diffraction peak of GO at 2h of about 10.3� was much weakerthan that of the original GO (Fig. 5a). The study found that the

Fig. 5. (a) XRD patterns and (b) Raman spectra of the GO/CNT films with various weight ratios of GO and CNTs (where 1G2C noted as GO/CNTs = 1:2).

340 C. Zhou et al. / Construction and Building Materials 134 (2017) 336–345

restacking of GO sheets into a layered structure had been inhibitedby the attachment of CNTs [29]. Additionally, Fig. 5b showed theRaman spectrum of GO, CNTs and the hybrid GO/CNTs. Three peakswere observed at 1320 cm�1, 1580 cm�1, 2640 cm�1, named as Dband, G band, 2D band, respectively. The intensity ratio (ID/IG) ofthe D (ID) and G (IG) bands was a measure of the degree of disorderand average size of the sp2 domain for few-layer graphene samples[30]. The ID/IG ratio of the hybrid GO/CNTs were much higher thanthat of GO or CNTs, suggesting that the degree of disorderincreased due to the intensely interaction between GO sheetsand CNTs. The results of differences in intensity ratio (ID/IG) ofthe GO, CNTs and GO/CNTs indicated that the hybrid GO/CNTshad been prepared [31]. Consequently, XRD and Raman analysesconfirmed that the insertion of CNTs into GO sheets inhibited therestacking of the GO sheets and the degree of disorder varied inGO/CNTs, respectively. Chao Zhang et al. [32] reported that theinteraction between CNTs and GO sheets was the weak p–p stack-ing interactions. And the schematic of dispersion of hybridGO/CNTs could be described in Fig. 6.

For further observing the highly dispersed hybrid GO/CNTs,stereoscopic microscope and SEM were employed. Microscopyanalysis revealed that the initial CNTs samples were heavily entan-gled (Fig. 7a and c). In contrast, the hybrid GO/CNTs were com-pletely disentangled. CNT were uniform dispersed in GO solutionas shown in Fig. 7b. And SEM images confirmed that CNTs wasabsorbed onto the GO sheets (Fig. 7d).

Fig. 6. Schematic of dispersion of grap

3.3. Effect of Ca(OH)2 on the hybrid GO/CNTs solution

As it was known, a large number of ions such as Mg2+, Ca2+, Al3+,Fe3+, SO2+ and OH� existed in the cement pore solution, which wasan alkaline environment. Fan X et al. [33] found that GO couldundergo quick deoxygenation in strong alkali solutions at moder-ate temperatures. Therefore, in order to test the effect of the cal-cium hydroxide (Ca(OH)2) on the stability of the hybrid GO/CNTs,the following experiments were carried out. Firstly, the experimentfound that GO aggregated quickly after the addition of a large num-ber Ca(OH)2 and slowly sank down later (Fig. 8). Moreover, SEMimage showed the morphology of the agglomerated GO sheets,indicating that the Ca(OH)2 had a negative effect on the stabilityof GO suspension due to the deoxygenation in a strongly alkalineenvironment. Furthermore, Fig. 9 showed different suspension(e.g. GO, CNTs and the hybrid GO/CNTs) with or without Ca(OH)2,which were settled for a few minutes and 48 h after ultrasonic,respectively. Results showed that adding of Ca(OH)2 in the suspen-sions resulted in obvious stratification after 48-h static(Fig. 9d and e). However, the stratification wasn’t observed in thesuspension with poly-carboxylate superplasticizer (Fig. 9f and g).According to all these phenomenon, it revealed that the addingof Ca(OH)2 had a negative effect on the suspensions (e.g. GO andthe hybrid GO/CNTs), causing stratification due to the agglomera-tion of the GO. And poly-carboxylate superplasticizer could relievethe effect of Ca(OH)2 on GO sheets. The reason might be the steric

hene oxide and carbon nanotubes.

Fig. 7. Stereoscopic microscopes of (a) CNTs in aqueous and (b) CNTs in GO solution, SEM images of (c) CNTs and (d) CNTs/GO, respectively.

Fig. 8. Picture and SEM image of GO + Ca(OH)2, respectively.

C. Zhou et al. / Construction and Building Materials 134 (2017) 336–345 341

effect of the large molecules (poly-carboxylate superplasticizer)adsorbed on the GO sheets, which prevented the reunion of GO.

3.4. Mechanical properties of the cement paste reinforced by GO/CNTs

In order to confirm that the hybrid GO/CNTs had a positiveeffect on the mechanical of cement pastes, two group experimentswere carried out. The compressive strength of the cement pastesreinforced by the hybrid GO/CNTs cement composites with and

without poly-carboxylate superplasticizer (noted as PC) wereshowed in Fig. 10. To check for reproducibility of the results, sixsamples were tested for compressive test. In all cases except theaddition of the hybrid GO/CNTs (0.04 wt% GO and 0.08 wt% CNTs),the cement pastes at 3 and 7-day age exhibited higher compressivestrength than their plain cement paste counterpart. And at 28-dayage, the cement pastes with or without PC reinforced by 0.02 wt%GO and 0.04 wt% CNTs (noted as 2G4C), the compressive strengthenhanced by 23.8% (Fig. 10b) and 9.1% (Fig. 10a), respectively.

Fig. 9. Aqueous CNT or GO dispersion (a) GO; (b) CNTs; (c) GO/CNTs; (d) GO + Ca(OH)2; (e) GO/CNTs + Ca(OH)2; (f) PC/GO + Ca(OH)2; (g) PC/GO/CNTs + Ca(OH)2 for a fewminutes (above) and 48 h (following) static. (PC: poly-carboxylate superplasticizer).

Fig. 10. Compressive strength of cement paste with different contents of CNTs/GO, (a) GO/CNTs = 1:2 without PC; (b) GO/CNTs = 1:2 with PC. (0: control; 2G4C: 0.02 wt% GOand 0.04 wt% CNTs; PC: poly-carboxylate superplasticizer).

342 C. Zhou et al. / Construction and Building Materials 134 (2017) 336–345

The group (2G4C) with adding PC showed a higher compressivestrength than that of the group (2G4C) without adding PC, whichimplied that the dispersion of the hybrid GO/CNTs was affectedby the alkaline environment. Furthermore, the group (1G2C) withor without PC at 28-day age showed the same tendency in thecompressive strength as the group (2G4C). However, for the group(4G8C), the strength dramatically decreased due to agglomerationof the hybrid GO/CNTs at a high concentration.

Furthermore, the compressive and flexural strength for thegroups enhanced with the nanofiber (e.g. GO, CNTs and GO/CNTs)were shown in Fig. 11a and b, respectively. From the Fig. 11a, itcould be seen that the early compressive strength of groupsenhanced by GO, CNTs, and the hybrid GO/CNTs were all increased.This enhanced effects might be ascribed to the introduction ofnucleation sites provided by the nanofibers (e.g. GO or CNTs). After28 d curing, compressive strength of the samples except the groupwith adding 0.04 wt% CNTs significantly increased. These resultsindicated that the enhanced effect of the hybrid GO/CNTs on thecompressive strength of the cement composites was much higherthan GO or CNTs alone. As shown in Fig. 11b, the flexural strength

of the group (2G4C) increased by 16.7% after 28-day curing, whichwas higher than that of groups with 2G (9.6%) and 4C (�6.1%). Theincrease tendency in flexural strength was similar to the compres-sive strength. It was observed that a similar gain in the strength ofthe cement composites was obtained by Zeyu Lu et al. [34], usingthe GO and functionalized CNTs composite with the similarCNTs/GO concentration.

3.5. Microstructure of the cement paste reinforced by GO/CNTs

Figs. 12–14 showed the micro morphology of the composites at28-days hydration. In the figures, Ca(OH)2, C-S-H gel, pores, cracks,CNTs, GO, and ettringite could been seen. Next to the ettringite andC-S-H gel, the CNTs embedded in the hydration products could beseen in Fig. 12. And there were some traces around the crack,where CNTs or ettringite were extracted. It indicated that partialwell dispersed CNTs with a longer effective length (the length ofCNTs available for crack bridging) [35] could be anchored by thehydration products for adding 0.02 wt% GO and 0.04 wt% CNTswith PC. And the strength gains both at early and late ages could

Fig. 11. (a) Compressive strength at different age and (b) Flexural strength at 28-day age of the CNTs/GO/cement composites.

Fig. 12. SEM images of cement paste with the hybrid GO/CNTs (2G4C with PC) at 28 d hydration.

Fig. 13. SEM images of cement paste with the hybrid GO/CNTs (2G4C without PC) at 28 d hydration.

C. Zhou et al. / Construction and Building Materials 134 (2017) 336–345 343

be attributed to effective dispersion of partial CNTs [36]. However,for the group (2G4C without PC), GO sheets were absorbed on thecalcium hydroxide crystal and CNTs clumped nearby (Fig. 13),which showed a serious entanglement and aggregation of CNTsin the hydration product. Anastasia Sobolkina et al. deemed thatthe neighbor C-S-H clusters could be bound together by CNTs in

few cases only, due to the shorting of CNTs after re-aggregation,resulting in no increase in tensile strength of the cement paste[9]. As shown in Fig. 14, extensive entanglements and re-agglomeration of CNTs appeared around the hydration productswhen adding the hybrid GO/CNTs at a higher concentration(4G8C). The results showed that the strength of the composites

Table 3MIP analysis of cement pastes.

Sample Total intruded volume (ml/g) Average pore diameter (nm) Total porosity (%) Total surface area (m2/g)

Control 0.1198 23.0 20.7329 20.9042G4C 0.1012 22.0 18.2829 18.442G 0.1058 22.6 18.5214 18.7204C 0.1022 24.2 18.2325 16.881

Note: 2G4C represents 0.02 wt% GO and 0.04 wt% CNTs.

Fig. 14. SEM images of cement paste with the hybrid GO/CNTs (4G8C with PC) at 28 d hydration.

Fig. 15. Pore size distribution of the cement paste with GO or CNTs at 28 days.(Note: 2G4C represents 0.02 wt% GO and 0.04 wt% CNTs).

344 C. Zhou et al. / Construction and Building Materials 134 (2017) 336–345

decreased with increasing of the concentration of hybrid GO/CNTs.Previous studies had shown that excessive CNTs content didn’timprove strength, and sometimes even deteriorates material prop-erties dramatically [36]. As a result, the mechanical properties ofthe cement paste with the hybrid GO/CNTs (2G4C) could beimproved. The reasons could be attributed to the following. First,the hydration product (e.g. C-S-H and Ca(OH)2) in the cement werelikely linked by the GO together to form strong interfacial adhesionby the covalent bonds [34]. Second, the advantage of high aspectratio of CNTs could be revealed to prevent the micro cracks prop-agation only, in the case of well dispersed CNTs in hardenedcement paste. The well dispersed CNTs could act as a bridge, whichhad the ability to anchor matrix on both sides of a crack to inhibitcrack face separation [37].

The pore structure of the control cement pastes and the cementcomposites at 28-day age were characterized using MIP. Table 3showed that the addition of the reinforcements (e.g. GO and thehybrid GO/CNTs) in cement paste reduced porosity and total porevolume of the pastes compared to the control. However, the sam-ple (4C) showed the biggest average pore diameter caused by theserious agglomeration of the CNTs. Fig. 15 presented the distribu-tion of pore diameters of the cement pastes with GO or CNTs. Forthe sample with adding GO or CNTs, it appeared a small numberof large pores (>1000 nm) in the paste, which had a harmful effecton the mechanical properties of the paste [38]. This results indi-cated that some agglomeration of the CNTs caused defects in thehardened paste, resulting in a number of large pores. However,the sample with GO/CNTs had a lower dV/d(log D) value and fewerlarge pores (>1000 nm) compared to the control. H.K. Kim et al.[39] inferred that the compressive strength of the CNT/cementmatrix was affected by not only dispersion of CNTs and interfacialinteraction between CNTs and hydration products, but also thetotal porosity of the cement composite. The physical contributionof carbon nanotubes could further fill in the pores between thehydration products such as calcium silicate hydrates (CSH) andettringite and refined of the pore structure. Consequently, the rea-

son for improvement in mechanical properties of the sample (e.g.2G and 2G4C) was the lower porosity [40], which suggested thatthe reinforcing capabilities of the hybrid GO/CNTs was better thanCNTs in cement paste. In this way, the additional hybrid GO/CNTs(2G4C) reduced the porosity of the cement pastes, thereby leadingto an increase in mechanical strength.

4. Conclusions

This study investigated the dispersion of the hybrid GO/CNTsand the mechanical behavior of cement paste reinforced by GO/CNTs. And the conclusions were drawn in the following:

GO as a dispersion agent for CNTs was much effective. By UV–vis testing, it found that the Abs of the hybrid GO/CNTs suspension

C. Zhou et al. / Construction and Building Materials 134 (2017) 336–345 345

remained nearly constant even after 24 h static. And the XRD andRaman spectra analysis confirmed that the insertion of CNTs intoGO sheets inhibited the restacking of the GO sheets, which indi-cated the interaction between GO and CNTs resulted in a muchhighly stable system. Moreover, SEM images further confirmedthat the CNTs were adsorbed on the surface of GO sheets due tothe p–p stacking interactions between them. Furthermore, thepoly-carboxylate superplasticizer was helpful to disperse thehybrid GO/CNTs in an alkaline environment. More importantly,the mechanical behavior results indirectly confirmed that the par-tial well dispersed CNTs with GO exhibited their excellent reinforc-ing capabilities. For the sample (2G4C) with or without PC, thecompressive strength at 28-day age enhanced by 23.8% and 9.1%,respectively. And the flexural strength increased by 16.7%, whichwas higher than that of samples with 2G (9.6%) and 4C (�6.1%).The MIP results suggested that the hybrid CNTs /GO played a rolein filling the pores because of their small size, resulting in a lowerporosity. Finally, the cement paste composite enhanced by thehybrid GO/CNTs was proposed.

Acknowledgements

This work was supported by the National Natural Science Foun-dation of China [Grant number 51472090], the FundamentalResearch Funds for the Central Universities and the Science andTechnology Program of Guangzhou [grant number201607010047], and the National High Technology Research andDevelopment Program [‘‘863 Program”, 2015AA034701].

References

[1] S. Parveen, S. Rana, R. Fangueiro, A review on nanomaterial dispersion,microstructure, and mechanical properties of carbon nanotube and nanofiberreinforced cementitious composites, J. Nanomater. 2013 (2013) 1–19.

[2] M.H. Al-Saleh, U. Sundararaj, A review of vapor grown carbon nanofiber/polymer conductive composites, Carbon 1 (47) (2009) 2–22.

[3] M.F.L. De Volder, S.H. Tawfick, R.H. Baughman, A.J. Hart, Carbon nanotubes:present and future commercial applications, Science 6119 (339) (2013) 535–539.

[4] M.S. Morsy, S.H. Alsayed, M. Aqel, Hybrid effect of carbon nanotube and nano-clay on physico-mechanical properties of cement mortar, Constr. Build. Mater.1 (25) (2011) 145–149.

[5] J. Zuo, W. Yao, K. Wu, Seebeck effect and mechanical properties of carbonnanotube-carbon fiber/cement nanocomposites, Fuller. Nanotubes CarbonNanostruct (2014).

[6] S. Parveen, S. Rana, R. Fangueiro, M.C. Paiva, Microstructure and mechanicalproperties of carbon nanotube reinforced cementitious composites developedusing a novel dispersion technique, Cem. Concr. Res. 73 (2015) 215–227.

[7] Y. Lu, K. Yang, D. Lin, Transport of surfactant-facilitated multiwalled carbonnanotube suspensions in columns packed with sized soil particles, Environ.Pollut. 192 (2014) 36–43.

[8] Z.S. Metaxa, J.T. Seo, M.S. Konsta-Gdoutos, M.C. Hersam, S.P. Shah, Highlyconcentrated carbon nanotube admixture for nano-fiber reinforcedcementitious materials, Cem. Concr. Compos. 5 (34) (2012) 612–617.

[9] A. Sobolkina, V. Mechtcherine, V. Khavrus, D. Maier, M. Mende, M. Ritschel,et al., Dispersion of carbon nanotubes and its influence on the mechanicalproperties of the cement matrix, Cem. Concr. Compos. 10 (34) (2012) 1104–1113.

[10] T. Xu, J. Yang, J. Liu, Q. Fu, Surface modification of multi-walled carbonnanotubes by O2 plasma, Appl. Surf. Sci. 22 (253) (2007) 8945–8951.

[11] Z. Zhang, X. Xu, Nondestructive covalent functionalization of carbonnanotubes by selective oxidation of the original defects with K2FeO4, Appl.Surf. Sci. 346 (2015) 520–527.

[12] S. Kang, J. Seo, S. Park, The characteristics of CNT/cement composites withacid-treated MWCNTs, Adv. Mater. Sci. Eng. 2015 (2015) 1–9.

[13] X. Li, W. Wei, H. Qin, Hu.Y. Hang, Co-effects of graphene oxide sheets andsingle wall carbon nanotubes on mechanical properties of cement, J. Phys.Chem. Solids 85 (2015) 39–43.

[14] J. Chen, B. Yao, C. Li, G. Shi, An improved Hummers method for eco-friendlysynthesis of graphene oxide, Carbon 64 (2013) 225–229.

[15] Z. Pan, L. He, L. Qiu, A.H. Korayem, G. Li, J.W. Zhu, et al., Mechanical propertiesand microstructure of a graphene oxide–cement composite, Cem. Concr.Compos. 58 (2015) 140–147.

[16] S. Lv, Y. Ma, C. Qiu, T. Sun, J. Liu, Q. Zhou, Effect of graphene oxide nanosheetsof microstructure and mechanical properties of cement composites, Constr.Build. Mater. 49 (2013) 121–127.

[17] C. Lin, W. Wei, Y.H. Hu, Catalytic behavior of graphene oxide for cementhydration process, J. Phys. Chem. Solids 89 (2016) 128–133.

[18] Z. Lu, D. Hou, H. Ma, T. Fan, Z. Li, Effects of graphene oxide on the propertiesand microstructures of the magnesium potassium phosphate cement paste,Constr. Build. Mater. 119 (2016) 107–112.

[19] H. Song, L. Yang, Y. Tang, D. Yan, C. Liu, S. Luo, Three-dimensional nitrogen-doped reduced graphene oxide-carbon nanotubes architecture supportingultrafine palladium nanoparticles for highly efficient methanolelectrooxidation, Chem. A Eur. J. 46 (21) (2015) 16631–16638.

[20] L. Yu, D. Tune, C. Shearer, J. Shapter, Heterojunction solar cells based on siliconand composite films of graphene oxide and carbon nanotubes, ChemSusChem17SI (8) (2015) 2940–2947.

[21] D.C. Marcano, D.V. Kosynkin, J.M. Berlin, A. Sinitskii, Z. Sun, A. Slesarev, et al.,Improved Synthesis of Graphene Oxide, 2010.

[22] S. Chuah, Z. Pan, J.G. Sanjayan, C.M. Wang, W.H. Duan, Nano reinforced cementand concrete composites and new perspective from graphene oxide, Constr.Build. Mater. 73 (2014) 113–124.

[23] K. Deng, C. Li, X. Qiu, J. Zhou, Z. Hou, Electrochemical preparation,characterization and application of electrodes modified with nickel–cobalthexacyanoferrate/graphene oxide–carbon nanotubes, J. Electroanal. Chem. 755(2015) 197–202.

[24] S.J. Chen, W. Wang, K. Sagoe-Crentsil, F. Collins, X.L. Zhao, M. Majumder, et al.,Distribution of carbon nanotubes in fresh ordinary Portland cement pastes:understanding from a two-phase perspective, RSC Adv. 7 (6) (2016) 5745–5753.

[25] Z. Luo, Y. Lu, L.A. Somers, A.T.C. Johnson, High yield preparation of macroscopicgraphene oxide membranes, J. Am. Chem. Soc. 3 (131) (2009) 898–899.

[26] Z. Xu, C. Gao, Aqueous liquid crystals of graphene oxide, ACS Nano 4 (5) (2011)2908–2915.

[27] X. Wang, H. Bai, G. Shi, Size fractionation of graphene oxide sheets by pH-assisted selective sedimentation, J. Am. Chem. Soc. 16 (133) (2011) 6338–6342.

[28] L. Jiang, L. Gao, J. Sun, Production of aqueous colloidal dispersions of carbonnanotubes, J. Colloid Interface Sci. 1 (260) (2003) 89–94.

[29] L. Qiu, X. Yang, X. Gou, W. Yang, Z. Ma, G.G. Wallace, et al., Dispersing carbonnanotubes with graphene oxide in water and synergistic effects betweengraphene derivatives, Chem. A Eur. J. 35 (16) (2010) 10653–10658.

[30] O.M. Dunens, K.J. MacKenzie, A.T. Harris, Synthesis of multiwalled carbonnanotubes on fly ash derived catalysts, Environ. Sci. Technol. 20 (43) (2009)7889–7894.

[31] L. Yang, D. Yan, C. Liu, H. Song, Y. Tang, S. Luo, et al., Vertically oriented reducedgraphene oxide supported dealloyed palladium–copper nanoparticles formethanol electrooxidation, J. Power Sources 278 (2015) 725–732.

[32] C. Zhang, L. Ren, X. Wang, T. Liu, Graphene oxide-assisted dispersion of pristinemultiwalled carbon nanotubes in aqueous media, J. Phys. Chem. C 26 (114)(2010) 11435–11440.

[33] X. Fan, W. Peng, Y. Li, X. Li, S. Wang, G. Zhang, et al., Deoxygenation ofexfoliated graphite oxide under alkaline conditions: a green route to graphenepreparation, Adv. Mater. 23 (20) (2008) 4490–4493.

[34] Z. Lu, D. Hou, L. Meng, G. Sun, C. Lu, Mechanism of cement paste reinforced bygraphene oxide/carbon nanotubes composites with enhanced mechanicalproperties, RSC Adv. (2015).

[35] R. Nadiv, M. Shtein, M. Refaeli, A. Peled, O. Regev, The critical role of nanotubeshape in cement composites, Cem. Concr. Compos. 71 (2016) 166–174.

[36] S. Xu, J. Liu, Q. Li, Mechanical properties and microstructure of multi-walledcarbon nanotube-reinforced cement paste, Constr. Build. Mater. 76 (2015) 16–23.

[37] I.J. Merchant, D.E. Macphee, H.W. Chandler, R.J. Henderson, Tougheningcement-based materials through the control of interfacial bonding, Cem.Concr. Res. (2001).

[38] P. Gao, X. Lu, C. Yang, X. Li, N. Shi, S. Jin, Microstructure and pore structure ofconcrete mixed with superfine phosphorous slag and superplasticizer, Constr.Build. Mater. 5 (22) (2008) 837–840.

[39] H.K. Kim, I.W. Nam, H.K. Lee, Enhanced effect of carbon nanotube onmechanical and electrical properties of cement composites by incorporationof silica fume, Compos. Struct. 107 (2014) 60–69.

[40] E. Güneyisi, M. Gesoglu, K. Mermerdas�, Improving strength, drying shrinkage,and pore structure of concrete using metakaolin, Mater. Struct. 5 (41) (2008)937–949.