Construction and Building Materials 104 (2016) 160–168
Contents lists available at ScienceDirect
Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat
Effect of mixing methods on the electrical properties of cementitious composites incorporating different carbon-based materials Ali Al-Dahawi a,b, Og˘uzhan Öztürk c, Farhad Emami d, Gürkan Yıldırım d,⇑, Mustafa Sß ahmaran d a
Department of Civil Engineering, Gaziantep University, Gaziantep, Turkey Dept. of Building and Construction Engineering, University of Technology, Baghdad, Iraq c Department of Civil Engineering, Selçuk University, Konya, Turkey d Department of Civil Engineering, Gazi University, Ankara, Turkey b
h i g h l i g h t s Different mixing methods were proposed to evenly distribute carbon-based materials. Carbon nanotubes (CNT), graphene nanoplatelets (GNP), carbon fibers (CF) were used. Performance was assessed by electrical resistivity and compressive strength results. The best method for uniform distribution was mechanical mixing with shear effect. Longer CF was the best to tailor matrix properties among all carbon-based materials.
a r t i c l e
i n f o
Article history: Received 30 August 2015 Received in revised form 8 November 2015 Accepted 9 December 2015 Available online 14 December 2015 Keywords: Self-sensing Structural health monitoring Carbon nanotubes (CNT) Graphene nanoplatelets (GNP) Carbon fibers (CF) Dispersion
a b s t r a c t Special attributes such as self-sensing ability could be included in conventional concrete-like materials, using carbon-based materials such as carbon nanotubes (CNT), graphene nanoplatelets (GNP) and carbon fibers (CF), which have recently come to the forefront due to their superior mechanical, thermal and electrical properties. However, their non-uniform distribution in cementitious systems stands in the way of taking full advantage of their benefits. To address this issue, an experimental study was undertaken, proposing several mixing methods to achieve uniform distribution. The performance of each method was assessed based on electrical resistivity (ER) and compressive strength measurements. Continuous reductions in ER measurements were observed with time, regardless of the mixing method and carbon-based materials used. Ultrasonication did not appear to be as useful as mechanical mixing with shear effect in terms of ER and compressive strength. Test results revealed that using longer CF (12 mm) in cementitious composites was the best way to tailor matrix properties in terms of electrical conductivity and compressive strength. CF was also easy to mix and has a relatively lower manufacturing cost than other electrically conductive carbon-based materials. Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction Portland cement concrete is the most frequently used construction material on earth, with yearly production of about 10 km3 . Its widespread use is evident in residential, industrial, commercial, sports, agricultural and other applications, and its success as an efficient and adaptable building material has been proven well over the past century. However, its long-term performance in structures serving under severe environmental circumstances is
⇑ Corresponding author. E-mail addresses: (G. Yıldırım).
http://dx.doi.org/10.1016/j.conbuildmat.2015.12.072 0950-0618/Ó 2015 Elsevier Ltd. All rights reserved.
becoming a subject of interest to many researchers. Mechanisms involved in reduced performance longevity are generally durability-based (e.g. alkali-silica reactivity, freeze–thaw susceptibility, rebar corrosion, sulfate attack etc.). Damage triggered and further exacerbated by the low durability characteristics of conventional concrete material necessitates maintenance and conservation or even full-scale reconstruction, leading to significant energy and raw material burdens. It is therefore critical to monitor the conditions of structures before they lose complete structural integrity . This kind of monitoring is possible with the use of sensory components such as optical fibers, strain gauges, and piezoelectric strain sensors [3–5]. However, several papers in the current litera-
A. Al-Dahawi et al. / Construction and Building Materials 104 (2016) 160–168
ture have reported that these sensors have poor durability, low sensitivity and a poor survival rate [6,7]. Piezoresistive behavior of the cementitious matrix can also be used as a tool to monitor structural health. Piezoresistivity (the dependence of electrical resistivity on the applied strain ) is achievable in concrete by incorporating modifiers into the cementitious matrices and having the concrete material itself serve as a sensor. Modifiers that make piezoresistive behavior possible are naturally conductive materials (e.g. iron-based), carbon fibers, carbon nanotubes (CNT), carbon black, etc. Despite the wide availability of conductive materials that favor the electrical properties of cementitious matrices, carbon fibers are the most commonly used because they are cheaper than other conductive carbon-based materials and contribute to mechanical properties such as flexural strength, flexural toughness, tensile strength and ductility, and reduced drying shrinkage . Given the fact that these materials perform differently in terms of conductivity and strain-sensing characteristics, it is hard to determine which one outperforms the other for a given structural application. However, it can be stated that these materials are generally at micro- or nano-scale, meaning that they have very large specific surface areas. Uniform distribution for the creation of an electrical network that can more easily capture or sense changes in applied strain in actual structures is therefore a challenge due to the agglomeration problem during mixing [10–12]. Several previous studies have made efforts to assess the dispersion quality of nano filaments in aqueous solutions [12–14], cement pastes , or cementitious composites [16,17] using a scanning electron microscope (SEM) and a transmission electron microscope (TEM). The downside of SEM and TEM methods is that they are solely dependent on visual observations to monitor good dispersion of nano filaments within a certain sample. There is also a high likelihood that the particles analyzed in these methods are small and therefore not representative of the whole sample. Other researchers have concentrated on the improvements in mechanical and electrical properties of cement pastes or mortars as an indication of the dispersion of electrically conductive materials. For example, Peyvandi et al.  employed the surface modification technique to facilitate the dispersion of graphite nanomaterials in aqueous media, and concluded that the addition of 0.13% of nanomaterial by weight of anhydrous cementitious materials improved the flexural strength of the cementitious matrix by up to 73%. The effects of silica fume use on the mechanical and electrical properties of cement/CNT composites were investigated by Kim et al. . The study reported that when no silica fume was used in composite material, CNT tended to agglomerate, leading to insignificant effects on compressive strength and electrical resistance, while the small amounts of silica fume helped intermix some agglomerated CNT by mechanically breaking them into smaller sizes. In their study, Sanchez and Ince  added carbon nano fibers into silica fume/cement pastes and concluded that due to poor distribution of the fibers in the cementitious composites, no marked changes took place in the compressive and splitting tensile strengths of specimens. A different study by Li et al.  concluded that the compressive and flexural strengths of cement mortar can be increased by 19% and 25%, respectively, with chemically functionalized CNT with a concentration of 0.5% by weight of cement. Based on the examples listed above, different studies have used a number of techniques for the proper distribution of a wide range of conductive materials. However, as there is no consensus on how to achieve this goal, the current experimental study was undertaken to address the issue. Electrically conductive materials from nano-scale (carbon nanotubes [CNT] and graphene nanoplatelets [GNP]) to micro-scale (carbon fibers [CF]) were used in composite material to distribute these materials in a way that would provide the highest possible electrical conductivity. This feature is vital for efficient self-sensing capability; a number of different mixing
methods were therefore used, taking into account the wide literature review and laboratory experience of the authors. In addition to electrical property measurements, compressive strength results were recorded to compare mixing methods and determine which one was superior. 2. Materials and mixture proportions 2.1. Materials The three conductive materials employed during the study were multi-walled carbon nanotubes (CNT), industrial graphene nanoplatelets (GNP) and chopped carbon fibers with 6 and 12 mm lengths and aspect ratios of 800 and 1600 (CF6 and CF12, respectively). Chemical and physical properties of CNT and GNP are provided in Table 1 and in the SEM photos in Fig. 1. CF had a tensile strength of 4200 MPa, elastic modulus of 240 GPa, elongation of 1.8%, density of nearly 1.7–2.0 g/cm3 and diameter of 7.5 lm. SEM photos of CF are also included in Fig. 1. In addition to the electrically conductive materials listed above, all mixtures produced in this study consisted of common ingredients such as CEM I 42.5R ordinary Portland cement (PC) (similar to ASTM Type I), Class-F fly ash (FA), fine silica sand with a maximum aggregate size of 0.4 mm, specific gravity of 2.60 and water absorption capacity of 0.3%, potable mixing water and polycarboxylate-ether-based high range water reducing admixture (HRWRA). In order to enhance the distribution of conductive materials, silica fume (SF) and nano-silica (NS), nano-calcite (NC) and a methylcellulose-based dispersion agent were also used in some of the mixing methods. The chemical and physical properties of PC, FA, SF, NS and NC are provided in Table 2. Particle size distributions of PC, SF, FA, and silica sand are provided in Fig. 2. 2.2. Mixture proportions Several mixing methods were used to achieve better distribution of electrically conductive carbon-based materials and obtain the best self-sensing capability in cementitious composites. During the preparation of mixtures, water to cementitious materials (PC + FA) ratio (W/CM) and fly ash to Portland cement ratio (FA/ PC) were kept constant at 0.27 and 1.2, respectively. For all mixtures incorporating CNT and GNP, carbon-based material amount was constant (0.25% of total weight of cementitious materials). In mixtures incorporating CF, the amount of CF was 0.5% by volume of the total mixture, regardless of the length of fibers used. These utilization rates for different electrically conductive materials were decided based on previously conducted studies in literature [12,20–22]. For certain mixing methods SF, NS, NC and methylcellulose were used at 5%, 1%, 1% and 0.2% of the total cementitious materials, respectively, for the enhancement of dispersion quality.
3. Dispersion methods of electrically conductive materials As mentioned above, the study used electrically conductive carbon-based materials both at nano- (CNT and GNP) and microscale (CF). In addition to the use of several conductive materials, different dispersion-enhancing materials were tested for certain mixing proportions as well. Given the significantly varying particle sizes of these materials, it was not possible to keep HRWRA amounts the same for all the mixing methods. Therefore, to ensure consistency of different mixtures, mini slump tests were performed and the quantity of HRWRA was selected based on similar flow deformation levels. In the mini-slump flow spread test, a truncated cone mold (diameter of 92 mm at the bottom, 44 mm at the top, with a height of 76 mm) was placed on a smooth plate, filled with mortar, and lifted upward. The slump flow deformation was defined as the dimension of the spread when the mortar stopped
Table 1 Physical and chemical properties of CNT and GNP. Physical properties
Diameter (nm) Length (lm) Thickness (nm) Surface area (m2/g)
10–30 10–30 – >200
5 – 50–100 13
Chemical properties Purity Oxygen content
A. Al-Dahawi et al. / Construction and Building Materials 104 (2016) 160–168
Fig. 1. SEM photos of CNT, GNP and CF.
Table 2 Chemical composition and physical properties of Portland cement, class-F fly ash, silica fume, nano-silica and nano-calcite. Chemical composition, %
SiO2 Al2O3 Fe2O3 MgO CaO Na2O K2O Loss on ignition
20.77 5.55 3.35 2.49 61.4 0.19 0.77 2.2
57.01 20.97 4.15 1.76 9.78 2.23 1.53 1.25
91.96 1.20 0.84 1.02 0.62 0.67 1.16 1.86
0.40 0.03 0.05 0.50 55.40 0.03 0.04 43.50
99.17 0.38 0.02 0.21 0.00 0.09 0.00 –
Physical properties Specific gravity Specific surface area (m2/kg) BET (m2/kg)
3.06 325 –
2.02 290 –
0.60 19,080 –
2.70 – 7.4
0.45 – 163.2
flowing . Similar mini-slump flow deformation levels were reached for all the mixing methods (Fig. 3). 3.1. Methods for the dispersion of conductive materials at nano-scale Some of the proposed mixing methods for CNT and GNP in the current study are compatible with those described in literature [12,13,17,19,24–26], but new methods to homogenously mix conductive materials have also been proposed to determine their
Fig. 3. Workability of mortar mixtures as a result of mini-slump test.
effectiveness on the distribution of nanomaterials within the cementitious matrices. In total, eight methods were proposed to enhance the dispersion of nano-scale carbon-based conductive materials and improve the mechanical and electrical properties of the fabricated cementitious mortars. 1st method: In this method, all dry raw materials (PC, FA and silica sand) and CNT or GNP were mixed in a 5-liter-capacity mortar mixer at 100 rpm for 10 min. All of the mixing water was added into the raw materials over 10 s. After that, speed was increased to 300 rpm, all of the HRWRA was added over 30 s, and mixing continued for an additional 10 min at 300 rpm .
Fig. 2. Particle size distributions of Portland cement, class-F fly ash, silica fume and silica sand.
A. Al-Dahawi et al. / Construction and Building Materials 104 (2016) 160–168
2nd method: After adding CNT or GNP into to the entire amount of mixing water, ultrasonication was applied for 10 min at 100 Hz. Mixing of dry raw materials in the mortar mixer started with ultrasonication, and continued for 10 min at 100 rpm. After 10 min, the CNT or GNP suspension was added into the mixed raw materials over 10 s. The speed of the mortar mixer was then increased to 300 rpm and all of the HRWRA was added over 30 s. Mixing continued for an additional 10 min at 300 rpm . 3rd method: CNT or GNP was added into the entire amount of mixing water along with all the HRWRA; ultrasonication was applied for 10 min at 100 Hz. HRWRA was intended to be used as a surfactant in this method. Dry raw materials were mixed in the mortar mixer with the start of ultrasonication, continuing for 10 min at 100 rpm. After 10 min, the CNT or GNP suspension was added into the mixed raw materials over 10 s. Mixing continued for an additional 10 min at 300 rpm [12,17,25]. 4th method: In this method, nano-silica particles (1% of the total weight of cementitious materials) were used to enhance CNT or GNP dispersion. CNT or GNP, along with water and nano-silica particles were ultrasonicated for 10 min at 100 Hz. Mixing of dry raw materials started with ultrasonication and continued for 10 min at 100 rpm. After 10 min, CNT- or GNP-waternano-silica suspension was added into the dry mixture over 10 s. The speed of the mortar mixer was increased to 300 rpm, and all the HRWRA was added over 30 s. Mixing of all materials continued for an additional 10 min at 300 rpm. 5th method: In this method, nano-calcite particles (1% of the total weight of cementitious materials) were used to enhance CNT or GNP dispersion. Although ultrasonication of CNT or GNP along with water and nano-calcite particles was attempted, uniform particle distribution was not observed visually. Therefore CNT or GNP, nano-calcite and dry raw materials were mixed with the mortar mixer at 100 rpm for 10 min. After 10 min, mixing water was slowly added to the mixture over 10 s. Mixing speed was increased to 300 rpm and all of the HRWRA was added over 30 s. Mixing of all materials was continued for an additional 10 min at 300 rpm. 6th method: CNT or GNP was added into the entire amount of mixing water along with all of the HRWRA. Ultrasonication was applied for 10 min at 100 Hz. After that, a methylcellulose-based dispersion agent (0.2% of the total weight of cementitious materials) was added to the suspension, and mixing with a kitchen-type blender was continued for five additional minutes at 3000 rpm. By the time all these steps were completed, mixing of dry raw materials in the mortar mixer for 10 min at 100 rpm had been completed as well. The suspension mixed with the dispersion agent was slowly added to the raw materials in the mortar mixer over 10 s, and mixing of all materials was continued for an additional 10 min at 300 rpm . 7th method: In this method, silica fume (5% of the total weight of cementitious materials) was added to the dry raw materials and mixed in the mortar mixer with CNT or GNP at 100 rpm for 10 min. While the mixer was working, all the mixing water was added into the raw materials over 10 s. After that, speed was increased to 300 rpm, and all the HRWRA was added over 30 s. Mixing of all materials continued for an additional 10 min at 300 rpm . 8th method: CNT or GNP was mixed with the entire amount of mixing water and HRWRA with a kitchen-type blender for 15 min at 3000 rpm. By the time this step was completed, the dry raw materials had been mixed for 10 min at 100 rpm in the mortar mixer as well. The previously prepared solution was then gradually added to the raw materials over 10 s and mixing continued for an additional 10 min at 300 rpm.
3.2. Methods for the dispersion of conductive materials at micro-scale Although several mixing methods similar to those proposed for mixtures incorporating CNT and GNP were used for mixtures with CF, results were not as good as expected. It was not easy to handle carbon fibers in water, given the high aspect ratios of CF6 and CF12 (800 and 1600, respectively). Thus, only two methods were proposed for carbon fiber dispersion in cementitious composites. CF-1st method: CF6 or CF12 were first mixed with the dry raw materials (PC, FA and silica sand) in a 5-liter-capacity mortar mixer at 100 rpm for 10 min. After slowly adding the mixing water at 100 rpm over 10 s, speed was increased to 300 rpm, all of the HRWRA was added over 30 s, and mixing of all materials was continued for an additional 10 min at 300 rpm (similar to 1st method explained above). CF-2nd method: A methylcellulose-based dispersion agent was employed to ensure uniform distribution of CF. The methylcellulose and CF were mixed with the entire amount of mixing water and HRWRA using a kitchen-type blender for 15 min at 3000 rpm. By the time the mixing with the blender was finished, the mixing of dry raw materials in a mortar mixer at 100 rpm for 10 min had also been completed. After that, the CF-methylcellulose suspension was gradually added to the dry raw materials over 10 s, and mixing of all materials in the mortar mixer was continued for an additional 10 min at 300 rpm (similar to 8th method explained above, with the exception of methylcellulose-based dispersion agent). 4. Specimen preparation and testing As mentioned previously, the performance of different mixing methods using several conductive materials at nano- and microscale was assessed by measuring electrical property (i.e. electrical resistivity) and compressive strength. To measure electrical resistivity (ER), £100 200 mm cylindrical specimens were prepared for each testing method. For compressive strength evaluation, 50 mm cubic specimens were prepared for each method. After casting was complete, cylindrical and cubic specimens were kept in their molds at 50 ± 5% RH, 23 ± 2 °C for 24 h, then removed and placed in isolated plastic bags at 95 ± 5% RH, 23 ± 2 °C until the completion of pre-determined testing ages. Electrical resistivity measurements for each mixing method were monitored at the end of 1, 7, 28, 60, 90 and 180 days, using four £100 80 mm pieces extracted from two different £100 200 mm specimens with a diamond blade saw. The reason for extracting two separate £100 80 mm cylinders from a single £100 200 mm cylinder was to account for any possible drawbacks due to trowelled and/or molded surface conditions, and to account for variations in ER measurements in different pieces of the same specimen. Six cubic specimens were used to evaluate average compressive strength after 28 days of curing. The same number of reference (Ref.) specimens without electrically conductive carbon-based materials were also prepared for all proposed test and mixing methods to compare performance. During the preparation of reference specimens 1st method of mixing was used. ER measurements were conducted with a concrete resistivity meter with uniaxial configuration, which is typical; details related to test setup can be found in Spragg et al. . In this method, £100 80 mm cylinders were inserted vertically between a set of plate electrodes. Pre-saturated wet sponges (10 mm high, 150 mm in diameter) were placed between the plates and the specimen to provide adequate electrical contact (Fig. 4). Given the possible effects on electrical measurements, sponges were pre-wetted with the same amount of water for each test. Saturated
A. Al-Dahawi et al. / Construction and Building Materials 104 (2016) 160–168 Table 3 Electrical resistivity measurements of cementitious composites using conductive materials at nano- and micro-scale based on mixing method (units in X m). Carbon nanotubes (CNT)
Fig. 4. Concrete resistivity meter and testing of a specimen.
sponges had no associated resistivity results; values recorded from the concrete resistivity meter were used as they are without correction regarding the resistivity of sponges. The concrete resistivity meter employed alternating current (AC) impedance technique, working with a wide range of frequencies from 1 Hz to 30 kHz to give results in less than five seconds and detect the phase angle between 0° and 180°. Despite the wide range of frequencies, however, the working frequency was set at 1 kHz throughout the study in accordance with Hou , who stated that polarization effect could be nullified with a frequency of at least 1 kHz AC current application. The concrete resistivity meter used in this study results in impedance and phase angle values, which can be converted into resistivity values using geometrical factors with the equation given below:
q ¼ Z cosðhÞ
where q, Z, h, A and L stand for resistivity (X m), electrical impedance (X), phase angle (°), cross-sectional area (m2) and length (m) of the specimen, respectively. 5. Results and discussion 5.1. Electrical resistivity of cementitious composites with carbon-based nano-materials ER measurements of cementitious composites produced with different mixing methods are presented in Table 3 for specimens incorporating electrically-conductive carbon-based materials at nano-scale (CNT and GNP). In addition to the raw data presented in Table 3, resistivity results are also shown in Fig. 5 for a clearer understanding. As is evident from both Table 3 and Fig. 5, ER values increased continuously with extended curing, regardless of mixing method and electrically-conductive material used. Electrical resistivity of concrete is generally related to microstructural elements such as porosity, pore solution and tortuosity of the pore network . Anything affecting these parameters has the potential to modify the electrical properties of concrete material. The reason for the continuously increasing ER results with prolonged curing could therefore be associated with the changes in the microstructure of the cementitious composites. Over time, it is expected that further hydration and pozzolanic reactions of fly ash will be dominant and reduce the amount of pore solution through which conductive ions are transported, as well as the number of leastresistive paths through matrix densification. Although deviations from the general behavior occurred depending on further curing age and selected mixing method, the introduction of conductive carbon-based materials at nanoscale caused the ER results of reference specimens to decrease,
Reference 1st 2nd 3rd 4th 5th 6th 7th 8th
12.4 9.6 10.1 10.4 9.6 13.4 9.5 11.0 4.2
29.2 26.6 26.9 26.2 38.1 28.9 23.2 68.9 13.0
175.3 166.0 150.3 151.3 182.2 167.5 111.5 359.0 43.1
524.5 457.2 504.8 439.5 468.0 618.3 311.3 908.5 128.7
726.3 615.8 876.6 631.5 623.7 1129.5 716.5 1212.9 207.2
1797.3 1512.5 2190.2 1483.0 1502.7 2730.4 2165.6 2347.3 497.3
Graphene nanoplatelets (GNP) Reference 12.4 29.2 1st 8.1 22.4 2nd 9.0 24.8 3rd 8.5 25.2 4th 9.4 29.7 5th 9.9 26.4 6th 6.0 22.3 7th 8.8 41.2 8th 5.5 18.0
175.3 118.8 137.0 133.1 129.6 135.5 124.2 308.4 62.3
524.5 301.0 303.0 301.0 343.8 442.9 277.9 699.3 136.5
726.3 672.3 682.6 709.6 837.3 1114.7 753.3 1311.2 221.0
1797.3 1419.2 1463.4 1802.2 2219.6 2833.5 2116.5 2288.4 519.4
Carbon fibers-6 mm (CF6) Reference 12.4 1st 8.0 2nd 7.7
29.2 12.9 16.7
175.3 33.2 48.9
524.5 78.8 94.2
726.3 126.7 200.8
1797.3 249.0 601.1
Carbon fibers-12 mm (CF12) Reference 12.4 29.2 1st 4.9 9.3 2nd 6.3 12.4
175.3 14.4 21.2
524.5 31.2 37.6
726.3 104.7 114.5
1797.3 185.5 238.3
irrespective of mixing method. For example, while the one-day average ER measurement of reference specimens was 12.4 X m, the same values for specimens incorporating GNP were less than 10 X m, going down to 5.5 X m depending on mixing method (Table 3). For specimens of the same age containing CNT, the same modality held true, although values were slightly higher than in specimens with GNP. Decrements in ER results with the inclusion of CNT and GNP were not surprising given their high specific surface areas and electrical conductivity. Table 3 also shows that all of the proposed methods (except the 8th method) where electrically conductive materials at nano-scale were incorporated into different mixing methods, GNP contributed more to electrical conductivity by lowering ER results more than CNT, which is partly favorable for self-sensing ability. As mentioned, ER measurements are strictly connected with porosity, pore solution chemistry, tortuosity, etc. . Considering the substantially higher specific surface of CNT (more than 200 m2/g) compared to GNP (nearly 13 m2/g), it is likely for specimens including CNT to exhibit higher ER values by filling the spaces among hydration products and lowering overall porosity. Moreover, there is a hypothesis that nucleation of hydration products on CNT promotes and accelerates cement hydration , which partially explains why higher ER results were obtained from specimens incorporating CNT. For lower electrical resistivity measurements to be recorded, especially in CNT-bearing specimens, it is vital that the uniform distribution of CNT particles should create a continuous path in the cementitious matrix where electrical current is transferred. Another possible explanation of the higher ER results for specimens with CNT could therefore be more aggregation and clustering of CNT particles compared to GNP. Tyson et al.  stated that based on scanning electron microscopy observations, CNTs tend to be aggregated in an alkaline environment (which is similar to that of cement paste), although they are highly dispersed, initially in aqueous form. Thus, it can be broadly stated that due to the inefficiency of proposed mixing
A. Al-Dahawi et al. / Construction and Building Materials 104 (2016) 160–168
Fig. 5. Electrical resistivity (ER) measurements of specimens incorporating (a) CNT and (b) GNP based on different mixing methods.
methods in distributing CNT, reaggregation of CNT in highly alkaline cementitious environments, kinking of CNT particles, and porosity of the paste, ER results were not lowered as much as in specimens with GNP. Fig. 5 and Table 3 make it clear that the 8th method is the best one in terms of lowering ER measurements for both nano-scale carbon-based materials. For example, when percentage decrements in ER results of one-day-old specimens are compared to those of reference specimens of the same age, they were 66% and 56% for CNT- and GNP-bearing specimens, respectively. Similar behaviors were observed for later ages. Although ultrasonication of nano-sized electrically-conductive materials was expected to provide a more homogenous dispersion, this was not the case. This behavior in samples created with mixing methods using ultrasonication could be due to the reaggregation of carbon-based nano materials in an alkaline environment . Mechanical mixing at a very high speed, paired with the effect of shear mixing with the kitchen-type blender used for the 8th method, resulted in more uniform distribution of nanomaterials in the presence of HRWRA and significantly lowered ER measurements, irrespective of nanomaterial type. As explained earlier, specimens incorporating GNP exhibited primarily lower ER than those incorporating CNT for all mixing methods other than the 8th. For the 8th method, however, ER values of CNT-bearing specimens were slightly lower than in the GNP-bearing ones for all testing ages, which shows that this
mixing method was the best in terms of revealing superior characteristics of CNT. Table 3 and Fig. 5 also show that in the 7th method, where SF was used for a more uniform distribution of nano-materials, the highest ER values were monitored specifically at prolonged curing ages; this finding is not in line with the literature . According to Sanchez and Ince , due to the difficulties of uniformly distributing SF particles, SF agglomerates with diameters of several hundred microns are likely to occur. These agglomerates have low pozzolanicity, and can react with the alkaline pore solution, leading to ASR (alkali-silica reaction) formation, volumetric instability and further related damage. Accordingly, further microstructural damage in SF specimens produced with the 7th method may account for the higher ER values. Another explanation could be the superior performance of SF particles in enhancing paste and mechanical properties (see compressive strength results), as opposed to helping with the dispersion of conductive nanomaterials. 5.2. Electrical resistivity of cementitious composites with carbon-based micro-materials Table 3 and Fig. 6 show ER results of specimens incorporating CF with different lengths (6 and 12 mm [CF6] and [CF12]). They demonstrate that substantial reductions took place in ER measure-
A. Al-Dahawi et al. / Construction and Building Materials 104 (2016) 160–168
ments of composite specimens with the inclusion of CF, regardless of CF type and mixing method used. For instance, the average ER result of specimens produced with the reference mixing method was 175.3 X m after 28 days of curing, while this value went down to 33.2 and 48.9 X m for the 1st and 2nd mixing methods using CF6, respectively. The same values dropped even further for the same mixing method with CF12, falling to 14.4 and 21.2 X m. The decremental trend in ER values with the use of CF with higher lengths was consistent for all testing ages, and is attributable to the higher probability of longer fibers providing a continuous conductive path through fiber-to-fiber contact along the whole length of the specimen. This observed behavior is also in line with what has been reported in literature . The rate of decrements in ER results with CF was similar to that observed in the 8th method, discussed previously. Taking into account the higher strength-toweight ratio compared to steel reinforcement and the comparably lower production cost compared to CNT, CF appears to provide the same benefits as CNT and GNP in terms of structural health monitoring through strain-sensing capability, without requiring the effort of an elaborate mixing method. Table 3 and Fig. 6 show that the 2nd method used for the uniform distribution of CF always resulted in higher ER results, excluding those obtained from one-day-old specimens incorporating CF6. Furthermore, the differences in the values became more pronounced as curing time was extended. This very clear trend can be explained by checking the details of individual mixing procedures. In the 2nd method of CF mixing, a methylcellulose-based dispersion agent was used by 0.2% of the total cementitious materials. Although methylcellulose was used with the aim of preventing the recollection of CF after dispersion, it also increased the need for HRWRA usage to ensure similar rheological properties, which unintentionally created air bubbles (Fig. 7). Given that electrical current cannot travel over the pores, specimens produced with 2nd method are likely to exhibit a higher resistance to electrical current. However, despite the increased ER results with the 2nd mixing method, CF use at the micro level was significantly more efficient than CNT and GNP use in lowering ER measurements, regardless of the changes in length. 5.3. Compressive strength results of cementitious composites with carbon-based nano-materials The 28-day-old compressive strength results of specimens incorporating electrically conductive nanomaterials (CNT and GNP) and produced with eight different mixing methods are
displayed in Fig. 8a. Results are the average of six 50 mm cubic specimens. The figure shows that the compressive strength results of CNT-bearing specimens varied between 45.3 and 54.0 MPa, while the results were between 36.7 and 60.1 MPa for the GNPbearing ones. No observable changes were recorded with the use of different mixing methods except the 6th, 7th and 8th. Fig. 8a shows that the 6th method resulted in the lowest compressive strength results, regardless of the type of conductive nanomaterial used. The probable cause of this behavior was the use of a methylcellulose-based dispersion agent, which led to high HRWRA requirement and more porous cementitious systems (Fig. 7), lowering compressive strength. In the 7th method, where silica fume was used in combination with raw materials to evenly disperse CNT and GNP, minor increments in compressive strength results occurred, although they were not very distinctive. This finding also corresponds with the previously discussed high ER measurements. However, it should be noted that the effectiveness of SF addition in attaining higher compressive strength values may be interrupted by SF agglomerates with low pozzolanic reactivity, which can react with alkalis in the pore solution and cause ASR damage on a microstructural basis , as well as non-uniform distribution of conductive materials at nano-scale. Finally, the highest compressive strength results for both CNTand GNP-bearing specimens were achieved by those produced using the 8th method. This outcome, obtained with respect to the reference specimens, also correlates with the lower ER results of specimens produced with this method. The relatively smaller particle size of nano-scale materials improve particle size distribution, resulting in high-density matrices through filler effect and flaw-bridging effect at the nano level. The high surface areas of nano-scale materials can also provide more nucleation sites, favoring the attainment of higher compressive strength. Although high compressive strength results were expected to bring about high ER results due to lower porosity, disconnected pores, lower amounts of pore solution and other related factors caused the opposite behavior. This indicates that the 8th method has a higher capability to evenly disperse conductive materials at the nano scale, without sacrificing electrical and mechanical properties. 5.4. Compressive strength results of cementitious composites with carbon-based micro-materials Fig. 8b shows the 28-day-old average compressive strength results of composite specimens incorporating CF6 and CF12,
Fig. 6. Electrical resistivity (ER) measurements of specimens incorporating carbon fibers (CF) based on different mixing methods.
A. Al-Dahawi et al. / Construction and Building Materials 104 (2016) 160–168
Fig. 7. Air bubbles on the surface of composite specimens produced with (a) CF-1st and (b) CF-2nd methods.
Fig. 8. Compressive strength results of specimens incorporating carbon fibers (CF) based on different mixing methods.
created using two different mixing methods. The figure indicates that the 1st mixing method was much more efficient in terms of CF dispersion than the 2nd method, irrespective of fiber type. The observed behavior has already been explained in previous sections; no further comments will be made here. It appears that using CF with greater lengths (12 mm) contributed more to the achievement of higher compressive strength results compared to reference specimens, regardless of mixing method. For example, while 51.1 and 38.4 MPa levels were reached for specimens with CF6 produced with the 1st and 2nd methods, respectively, the same values for specimens with CF12 were 65.3 and 49.4 MPa. High compressive strength results recorded in specimens with longer CF also corresponded with the lower ER results detailed in previous sections. The higher compressive strength results for CF12 over CF6 specimens can be attributed to a greater possibility of longer fibers bridging defects and/or pores in the microstructure as internal reinforcements. Given their shorter lengths, CF6 fibers are more likely to be effective only in pre-peak microcracking and may not be as influential as CF12 in addressing post-peak microcracking, resulting in more sudden losses in load-carrying capacity and lower compressive strength results.
6. Conclusions This paper has proposed different mixing methods for a better distribution of electrically-conductive materials at nano- (CNT and GNP) and micro-scale (CF6 and CF12) to take advantage of self-sensing capability for structural health monitoring. The dispersion of conductive materials in cementitious systems was assessed through electrical resistivity (ER) (after 1, 7, 28, 60, 90
and 180 days of curing) and compressive strength (after 28 days of curing) measurements. The following conclusions were drawn: ER measurements increased with time, regardless of the type of conductive material and mixing method used, due to the continuous evolution of cementitious matrices, which caused reduced overall porosity and pore solution, a disconnected pore network and so on with prolonged curing ages. Although there were variations from the general trend, depending on the conductive material and mixing method preference, ER measurements showed decrements up to a certain limit with the inclusion of conductive materials. Based on the proposed mixing methods, GNP and longer CF (12 mm) contributed more to the reduced ER results than CNT and shorter CF (6 mm). Ultrasonication of conductive materials at nano-scale did not have the anticipated impact on ER results of composite materials, although mechanical mixing with shear effect (8th method) was very influential in dispersing both CNT and GNP. The best method for uniform distribution of CF was dry mixing with the raw materials (1st method), regardless of fiber length. Use of dispersion-enhancing materials such as silica fume, nanosilica, nano-calcite and methylcellulose was not as effective as expected for both ER and compressive strength results. As in ER testing, the best methods contributing to higher compressive strength results compared to reference specimens were the 8th and 1st methods for nano- and micro-scale conductive materials, where conventional mechanical mixing was used in general. It appears that as long as good dispersion of conductive materials (regardless of size) was achieved, ER results were positively influenced (decreased for a better selfsensing ability) and mechanical properties were improved.
A. Al-Dahawi et al. / Construction and Building Materials 104 (2016) 160–168
Considering its substantial effects on both electrical and mechanical properties, as well as its low cost and easier dispersibility in cementitious systems, CF appears to be more beneficial than CNT and GNP, which are expensive, labor-intensive to manufacture and hard to mix in a uniform manner.
Acknowledgments The authors gratefully acknowledge the financial assistance of the Scientific and Technical Research Council (TUBITAK) of Turkey provided under Project: 114R043. References  E.M. Gartner, D.E. Macphee, A physico-chemical basis for novel cementitious binders, Cem. Concr. Res. 41 (2011) 736–749.  K.P. Chong, Health monitoring of civil structures, J. Intel. Mater. Syst. Struct. 9 (1998) 892–898.  H. De Backer, W. De Corte, P. Van Bogaert, A case study on strain gauge measurements on large post-tensioned concrete beams of a railway support structure, Insight Non-Destruct. Test. Condit. Mon. 45 (2003) 822–826.  J.S. Leng, D. Winter, R.A. Barnes, G.C. Mays, G.F. Fernando, Structural health monitoring of concrete cylinders using protected fibre optic sensors, Smart Mater. Struct. 15 (2006) 302–308.  S. Park, S. Ahmad, C.B. Yun, Y. Roh, Multiple crack detection of concrete structures using impedance-based structural health monitoring techniques, Exp. Mech. 46 (2006) 609–618.  Y.S. Deng, B.J. Sun, Researches on intelligent material system and its application to civil engineering, Arch. Technol. 36 (2005) 92–95.  W.A. Li, Method for structural safety monitoring of composite carbon fiber reinforced concrete, China Safety Sci. J. 15 (2005) 109–112.  S.D. Senturia, Microsystem Design, Kluwer Academic Publishers, Boston, 2001.  S. Wen, D.D.L. Chung, Piezoresistivity-based strain sensing in carbon fiberreinforced cement, ACI Mater. J. 104 (2007) 171–179.  B.M. Tyson, R.K.A. Al-Rub, A. Yazdanbakhsh, Z. Grasley, A quantitative method for analyzing the dispersion and agglomeration of nano-particles in composite materials, Compos. Part B – Eng. 42 (2011) 1395–1403.  A. Yazdanbakhsh, Z. Grasley, B. Tyson, R.K.A. Al-Rub, Dispersion quantification of inclusions in composites, Compos. Part A – Appl. Sci. 42 (2011) 75–83.  A. Sobolkina, V. Mechtcherine, V. Khavrus, D. Maier, M. Mende, M. Ritschel, et al., Dispersion of carbon nanotubes and its influence on the mechanical properties of the cement matrix, Cem. Concr. Compos. 34 (2012) 1104–1113.  O. Mendoza, G. Sierra, J.I. Tobon, Influence of super plasticizer and Ca(OH)2 on the stability of functionalized multi-walled carbon nanotubes dispersions for cement composites applications, Constr. Build. Mater. 47 (2013) 771–778.
 L. Kong, X. Yin, X. Yuan, Y. Zhang, X. Liu, L. Cheng, et al., Electromagnetic wave absorption properties of graphene modified with carbon nanotube/poly (dimethyl siloxane) composites, Carbon 73 (2014) 185–193.  T. Nochaiya, A. Chaipanich, Behavior of multi-walled carbon nanotubes on the porosity and microstructure of cement-based materials, Appl. Surf. Sci. 257 (2011) 1941–1945.  F. Sanchez, C. Ince, Microstructure and macroscopic properties of hybrid carbon nanofiber/silica fume cement composites, Compos. Sci. Technol. 69 (2009) 1310–1318.  M. Saafi, K. Andrew, P.L. Tang, D. McGhon, S. Taylor, M. Rahman, et al., Multifunctional properties of carbon nanotube/fly ash geopolymeric nanocomposites, Constr. Build. Mater. 49 (2013) 46–55.  A. Peyvandi, P. Soroushian, N. Abdol, A.M. Balachandra, Surface-modified graphite nanomaterials for improved reinforcement efficiency in cementitious paste, Carbon 63 (2013) 175–186.  H.K. Kim, I.W. Nam, H.K. Lee, Enhanced effect of carbon nanotube on mechanical and electrical properties of cement composites by incorporation of silica fume, Compos. Struct. 107 (2014) 60–69.  G.Y. Li, P.M. Wang, X. Zhao, Mechanical behavior and microstructure of cement composites incorporating surface-treated multi-walled carbon nanotubes, Carbon 43 (2005) 1239–1245.  D.D.L. Chung, Piezoresistive cement-based materials for strain sensing, J. Intel. Mater. Syst. Struct. 13 (2002) 599–609.  X. Yu, E. Kwon, A carbon nanotube/cement composite with piezoresistive properties, Smart Mater. Struct. 18 (2009) 055010.  K.H. Khayat, A. Yahia, Simple field tests to characterize fluidity and washout resistance of structural cement grout, Cem. Concr. Aggr. 20 (1998) 145–156.  S. Musso, J.M. Tulliani, G. Ferro, A. Tagliaferro, Influence of carbon nanotubes structure on the mechanical behavior of cement composites, Compos. Sci. Technol. 69 (2009) 1985–1990.  V.C. Moore, M.S. Strano, E.H. Haroz, R.H. Hauge, R.E. Smalley, J. Schmidt, et al., Individually suspended single-walled carbon nanotubes in various surfactants, Nano Lett. 3 (2003) 1379–1382.  G.Y. Li, P.M. Wang, X. Zhao, Pressure-sensitive properties and microstructure of carbon nanotube reinforced cement composites, Cem. Concr. Compos. 29 (2007) 377–382.  R. Spragg, Y. Bu, K. Snyder, D. Bentz, J. Weiss, Electrical testing of cement-based materials: role of testing techniques, sample conditioning, and accelerated curing, Publication FHWA/IN/JTRP-2013/28. Joint Transportation Research Program, Indiana Department of Transportation and Purdue University, West Lafayette, Indiana, 2013, http://dx.doi.org/10.5703/1288284315230.  T.C. Hou, Wireless and electromechanical approaches for strain sensing and crack detection in FRC materials. PhD dissertation, Civil and Environmental Engineering, University of Michigan, Ann Arbor, MI, 2008.  J.M. Makar, G.W. Chan, Growth of cement hydration products on single-walled carbon nanotubes, J. Am. Ceram. Soc. 92 (2009) 1303–1310.  B.M. Tyson, R.K.A. Al-Rub, A. Yazdanbakhsh, Z. Grasley, Carbon nanotubes and carbon nanofibers for enhancing the mechanical properties of nanocomposite cementitious materials, J. Mater. Civ. Eng. 23 (2011) 1028–1035.  B. Chen, K. Wu, W. Yao, Conductivity of carbon fiber reinforced cement-based composites, Cem. Concr. Compos. 26 (2004) 291–297.