Intrinsic self-sensing cementitious composites with hybrid nanofillers exhibiting excellent piezoresistivity

1.1 Necessity and significance of structural health monitoring (SHM) in civil infrastructure

Concrete is a widely used manmade construction material worldwide in civil infrastructure, due to its favorable compressive strength and durability, well moldability, and inexpensive [1]. Nevertheless, many kinds of loads and environmental factors, such as impact load, earthquakes, carbonization, and acid and alkaline corrosion, can result in cracks, damage, and degradation and even cause the destruction of the concrete structures unavoidably. Thus, they will bring a threat to the safety of the public [2]. Therefore, SHM is important for assessing defects of concrete structures and evaluating their service conditions. In addition, SHM is necessary for facilitating early warning systems, prolonging the lifespan of civil infrastructure [3,4]. The sensory systems of SHM can collect insitu information of the concrete structures, for example, stress, strain, displacement, and cracks. The detected information can be utilized to recommend maintenance and strengthen the concrete structures [5,6].

In the previous research, quantity conventional sensing techniques, for example, optical fibers, strain gauges, piezoelectric ceramics, and shape memory alloys, have been utilized extensively in SHM systems [7]. However, the conventional sensors present obvious shortcomings, such as incompatible with concrete structures, low sensitivity, low survival rate, and expensive, which limit their widely applications [8,9]. Therefore, it is an urgent need to solve these issues of SHM sensory systems of civil infrastructure. As a result, intrinsic self-sensing cementitious composites possess excellent sensing properties and need to be developed urgently to address these drawbacks.

1.2 Origin and definition of intrinsic self-sensing cementitious composites

In 1993, the intrinsic self-sensing cementitious composites were first developed, and since then, these composites have been studied and applied extensively [10]. Intrinsic self-sensing cementitious composites can sense themselves, which are fabricated by adding electrical conduction fillers to cementitious composites. Self-sensing capacity is usually realized by electrical pathways and networks generated by electrical conduction fillers inside the cementitious composites, which possess structural together with sensing properties. Deformation, stress, strain, damage, and cracks can be sensed from testing the electrical resistivity of themselves during the loading process, which is also called piezoresistivity. Their mechanical performances and durability are not compromised or even enhanced [5,11,12,13]. The pathways and networks have been widely generated by the electrical conduction fillers based on percolation, field emission, and quantum tunnelling effect. They are changed when the composites are loaded or deformed, subjected to applied load or variation of environments. Thus, their electrical property will be changed [14,15]. The intrinsic self-sensing cementitious composites can provide timely data on the performance of concrete structures, monitoring potential defects. Different from conventional sensors, intrinsic self-sensing cementitious composites are compatible and have similar durability with concrete structures because of their nature cementitious properties [16].

1.3 Self-sensing property of the cementitious composites containing typical electrical conduction fillers

The electrical conduction fillers usually can be divided into non-metallic materials, for example, carbon fibers (CFs), graphene nanoplatelets (GNPs), carbon nanofibers (CNFs), nanocarbon blacks (NCBs), carbon nanotubes (CNTs), and metallic electrical conduction fillers as steel fibers, nickel powders, brass fibers, etc. In addition, the nano-electrical conduction fillers have the size and filling effect, which can enhance the electrical conduction together with the sensing property of cementitious composites [17,18].

GNPs are two-dimensional with multiple sheet-layer structures and a thickness of nanometers. The π electron involved in interlayer bonding imparts GNPs with excellent electrical conduction ability. Their large aspect ratio can offer plentiful nucleation positions for the hydration product. Thus, the nucleation effect, filling effect, and bridging effect modify the microstructures of the cementitious composites together. Thereby, both their mechanical performances and self-sensing properties can be enhanced [19,20,21]. Sun et al. studied the self-sensing capacity of cementitious composites containing GNPs under both static circular compression and dynamic loads. The effect of load amplitude and loading rate have been studied when these composites are subjected to compression. The results show that the cementitious composites containing 5 vol% GNPs present the best sensitivity, and their fractional change in electrical resistivity (FCR) is lowered to −15.6% exerted on compressive stress of 20 MPa. In addition, their stress sensitivity and strain sensitivity were achieved at 0.78%/MPa and 156, respectively [22]. Sevim et al. investigated the self-sensing property of cementitious composites containing GNPs with different types of surface area, thickness, and particle size. They indicated that the smaller particle size, together with the larger surface areas of the GNPs, cannot be dispersed effectively, thus presenting no self-sensing property. Nevertheless, the smaller surface areas together with the bigger particle size of the GNPs displayed a favorable self-sensing property [23]. Qi et al. studied the sensing capacity of the cementitious composites incorporating the GNPs with various particle sizes subjected to cyclic compression and tension loads and pointed out that these composites exhibit excellent self-sensing property. Additionally, the cementitious composites incorporating a bigger particle size of the GNPs have better stress sensitivity [24]. Dong et al. studied the self-sensing capacity of cementitious composites with GNPs subjected to impact load. They demonstrated that the maximum magnitude of FCR of the cementitious composites containing 1.0% GNPs ranged from 4.96 to 6.36, 8.31, and 10.97% under impact load with angles varied from 0° to 30°, 60°, and 90°, respectively [25]. Dong et al. investigated the self-sensing performance of the cementitious composites containing GNPs with a dosage of 3.0% under compression, and found that the FCR was approached to −30% [26].

CNTs are carbon-based tubular materials made of several layers of graphite at nanoscale, and they have a coaxial lamellar crimp structure. The CNTs possess excellent electrical conductivity, good mechanical performance, and well thermal conductivity. Thus, they can impart cementitious composites with better electrical conduction and self-sensing performances [27,28,29].

Wang et al. studied the self-sensing property of cementitious composites containing CNTs subjected to salt attacks. They pointed out that the self-sensing property greatly relies on the percolation density and the critical exponent of CNTs’ pathways around the percolation threshold. The relation of FCR and repeated compressive load without salt environments was “M” shape. In addition, the relation between electrical resistivity and strain was an exponential function [30]. Lee et al. studied the self-sensing capability of cementitious composites incorporating CNTs with different concentrations of 0.1 and 0.5 wt% withstand tensile load, and their strain sensitivity was reached up to 23.8. In addition, they embedded the composites as sensors at the tension position of the concrete beam subjected to flexural load. They also proposed empirical equations to evaluate the tensile strain of the concrete beam based on the results of FCR tested using the embedded sensor [31]. Yuan et al. coated CNTs on cement particles as electrical conduction fillers of CNTs@cement and developed cementitious composites with high sensitivity. The CNTs absorbed uniformly on the cement particles’ surface do not need a surfactant to disperse. Thus, the flowability and mechanical properties of the composites were not reduced. The FCR was lowered to −1.26% under cyclic compressive load. Additionally, the electrical conductivity of the composites with 0.1 wt% CNTs@cement was increased by two orders of magnitude [32]. Song et al. carried out an experiment on the self-sensing performance of cementitious composites incorporating CNTs by flexural, splitting tensile, and shear tests. The composites with 0.2 wt% CNTs showed a favorable self-sensing capacity for the initial crack, deflection harden or soften [33]. Mesquita et al. analyzed the self-sensing properties of cementitious composites incorporating CNTs and indicated that the FCR of the composites with 2.0% CNTs was reduced by −2.0% [34].

Because of NCBs present advantages as fine filling effect, stability at high thermal, low cost, and electrical resistivity is lowered to about 10⁻¹ Ω·cm; thus, NCBs become attractive electrical conduction fillers. However, NCBs can tightly adhere to each other because of strong Van der Waals forces [35,36,37,38]. Li et al. developed self-sensing cementitious composites containing NCBs with particle size of 120 nm. The contents of NCBs were varied from 5.0 to 25.0 wt%, and they demonstrated that the relation between FCR and strain withstand monotonic compression is linear. The strain sensitivity is as high as 55.28, corresponding to the NCBs’ concentration of 5.0 wt% [39]. Monteiro et al. studied the sensing performance of cementitious composites containing dosages of the NCBs of 0.0–10.0 wt% subjected to repeated compression. The composites present self-sensing property when the dosage of the NCBs is higher than 7.0 wt%, and the maximum strain sensitivity is up to 30.0, corresponding to the content of NCBs of 7.0 wt% [40]. Deng and Li indicated that the electrical resistivity of cementitious composites incorporating NCBs under tension increased with the propagation of cracks, and the increase ratio under the inelastic range is higher than that under the elastic range [41]. Monteiro et al. developed self-sensing cementitious composites by adding NCBs, and then utilized them as sensors to monitor cyclic compressive loads when temperatures ranged from 15 to 45°C. A linear and repeatable self-sensing property with a strain sensitivity ranging from 40 to 60 was found. Nevertheless, the strain sensitivity was reduced by 30% when the temperature rose to 45°C, but the linearity of the strain sensitivity versus the stress was not affected [42]. Hussain et al. demonstrated that the strain sensitivity of cementitious composites containing NCBs under both monotonic and repeated compression was increased with increasing concentration of the NCBs after it reached 1.50%, and the maximum strain sensitivity was achieved at 134.32 [43,44].

To address the disadvantages of intrinsic self-sensing cementitious composites with single electrical conduction fillers, some researchers proposed hybrid electrical conduction fillers due to their synergistic effect [45,46]. Zhang et al. used the NCBs together with CNTs to develop self-sensing cementitious composites. The NCBs and CNTs with different short and long ranges can stimulate the generation of extensive conduction pathways inside the cementitious composites; thus, more distance among the electrical conduction fillers can be changed and lead to sensitive self-sensing performance subjected to whatever repeated and monotonic compressive loads. These composites with a concentration of 2.40 vol% of the hybrid NCBs and CNTs present stable sensing properties, and the maximum magnitude of FCR is reached up to 25.4% [47,48]. Qiu et al. developed self-sensing cementitious composites by using self-assembled CNTs and NCBs. The electrical resistivity of these composites was lowered to 33 Ω·cm. The maximum magnitude FCR of the beam made up of these composites containing 1.8 vol% CNTs and NCBs was about 286.0% under monotonic flexural load. Moreover, the beam incorporation of 2.0 vol% CNTs and NCBs presented the optimum self-sensing performance and obtained the maximum sensitivity of 322.7 [49]. Jawed Roshan et al. investigated the self-sensing property of cementitious composites incorporating hybrid CNTs and GNPs with concentrations of 3.0 and 4.0%. The strain, cracks, and damage of the composites were detected under cyclic compressive load [50]. Zhang et al. developed cementitious composites with CNTs together with TiO₂. These composites exhibiting favorable self-sensing characteristics. Their maximum FCR and strain sensitivity were achieved at 84.09% and 317, respectively [51].

1.4 Application and challenges of the intrinsic self-sensing cementitious composites

Recently, intrinsic self-sensing cementitious composites have been applied as sensors in concrete structures for SHM. Dong et al. fabricated sensors by intrinsic self-sensing cementitious composites incorporating NCBs and embedded them into a concrete beam under flexural load. Cracks and stress of this beam in the compression zone have been detected with a slow reduction and a sudden increase in the FCR tested from the embedded sensors. At the same time, the cracks and stress of the concrete beam in the tension zone have also been detected by a slow increase and a rapid jump of the FCR. The change of the FCR of the sensors are in accordance with variation of flexural stress and presents good sensitivity when the concrete beam from loading to failure [35]. Ding et al. utilized the intrinsic self-sensing cementitious composites incorporating CFs and NCBs to detect cracks in a concrete maglev girder. They first demonstrated that the CFs and CNTs can generate wide networks in cementitious composites for the synergic effects, thus resulting in the composites exhibiting favorable self-sensing performance. Additionally, these composites presented high sensitivity to monitor the propagation of cracks and damage of the maglev girder [52]. Ding et al. fabricated sensors using cementitious composites with hybrid CNTs and NCBs and analyzed their sensing properties when embedded them in a concrete column. Sensing properties of embedded sensors with non-embedded ones have been compared, indicating that the differences in Poisson’s ratio and elastic modulus between concrete and sensors can lead to different monitoring results [53]. Sun et al. assessed self-sensing performances of cementitious strain sensors with steel fibers when embedded in concrete columns and pointed out that the damage of the concrete columns can be detected with irreversible variation of the sensor’s electrical resistivity [54]. Xiao et al. demonstrated that the intrinsic self-sensing cementitious strain sensor incorporating NCBs embedded in the central position of concrete columns presented good sensitivity. However, the sensing performance was susceptible to the embedded condition of the sensor [55].

1.5 Aims of this research

As known from the above review of the intrinsic self-sensing cementitious composites, little researches have been conducted on the mechanical performances, electrical properties, and self-sensing properties of cementitious composites incorporating hybrid electrical conduction fillers of the GNPs, CNTs, and NCBs. Moreover, some of the studies demonstrated that the hybrid electrical conduction fillers possess a cooperation effect, while they mainly focus on the cooperation effect according to microstructures or deduced from electrical conductivity. The previous studies lack systematic research on the precise characteristics and mechanisms related to how the hybrid electrical conduction fillers comprised by multi-dimension contribute to the mechanical performances, electrical property, and self-sensing property.

In this article, the hybrid electrical conduction fillers of GNPs, CNTs, and NCBs with dosages changed from 0.0 to 12.0 wt% were employed to investigate intrinsic self-sensing cementitious composites. First, the mix design and electrical conductivity using DC and AC methods with two electrodes of the composites were conducted. After that, the mechanical performances were tested. Finally, the self-sensing performances of the cementitious composites were comprehensively assessed withstanding repeated and monotonic compression loads. Furthermore, the FCR and sensitivities during the loading process were discussed to explore the mechanisms of self-sensing capacity. This research can offer a theoretical basis and promote application for intrinsic self-sensing cementitious composites containing the hybrid electrical conduction fillers in SHM of civil infrastructure.

2.1 Materials

The nanoparticles graphene nanoplatelets, carbon nanotubes, and nanocarbon blacks (GCNs), including GNPs, CNTs, and NCBs, with a mass proportion of 25%:30%:45%, were used as electrical conduction fillers in the present article, and they were purchased from Sciences Organic Chemistry Co., Ltd. (Chengdu, China). Their main properties are summarized in Table 1. 42.5 R ordinary Portland cement (Henan Yong'an Cement Co. Ltd., China) with a density of 3.1 g·cm⁻³ was adopted. The fly ash purchased from Henan Yulian Energy Group Co., Ltd., China, had a density of 2.4 g·cm⁻³, which was utilized as a binder together with cement. Moreover, the fly ash could facilitate the dispersion of the GCNs and decrease the porosity of the cementitious composites [56]. A polycarboxylic superplasticizer supplied from Zhongyan Building Materials Technology Co., Ltd. (Hunan, China) with a solid concentration of 28.0% and a specific gravity of 1.1 was used to assist cementitious composites to realize desirable workability, and it also could enhance the dispersion of GCNs. The polycarboxylate superplasticizer was effective for dispersing GCNs because the anionic and non-ionic components could effectively stabilize the GCNs [57,58,59]. Tap water was used for fabricating the samples.

2.2 Sample preparation

Mixture proportion of the cementitious composites is listed in Table 2. The dosage of GCNs was determined on the mass of the binders, which was varied from 0.0 to 12.0 wt%. Numbers of samples were named as the dosage of the GCNs. For example, GCN6 represents the specimen with a content of the GCNs of 6.0 wt%. Additionally, the contents of the GCNs were designed according to the previous research combination with experimental results. As summarized in Table 3, the contents of the nanofillers of GNPs, CNTs, and NCBs incorporated in cementitious composites are in the ranges of 0.0–7.5 wt% (0.0–10.0 vol%), 0.0–2.5 wt% (0.0–5.0 vol%), and 0.0–25.0 wt%, respectively. Simultaneously, the optimum dosage of the GCNs has also been obtained in accordance with a large number of experiments, which have been conducted to explore the GCNs dispersion inside the cementitious composites, flowability, electrical conduction, mechanical performances, and piezoresistivity of the cementitious composites.

Similarly, the contents of superplasticizer and water were also in accordance with the mass of binders. The sample manufacturing process is shown in Figure 1, and the details are described as follows. All the materials were weighed using an electronic balance first, and then, the GNPs, CNTs, and NCBs were stirred in a plate. Later on, water together with superplasticizer were added to the mixer. After that, the fly ash and cement were slowly added to the mixer with stirring at low speed. Subsequently, the mixed GNPs, CNTs, and NCBs were added to the mixer when stirring at low speed within 2 min. The mixer stirred at high speed for 2 min when all materials had been added. Next, the mixture was cast into an oiled mold with dimensions of 20 mm × 20 mm × 40 mm, and two stainless steel mesh probes with a distance of 20 mm were inserted in the center of the mold. Finally, the molds with the poured mixture were vibrated for 40 s by a vibrator to reduce the air bubbles. The poured molds were maintained in a standard curing room with 20 ± 2°C and relative humidity of 90% for 24 h. After that, the specimens were demolded and cured in water for 28 days.

Figure 1

The sample manufacturing process. (a) Mixing. (b) Raw materials. (c) Pouring. (d) Vibration. (e) Curing. (f) Specimens.

To simplify the fabrication techniques and convenient application in practical engineering, as well as decrease the cost of the composites, the dispersion techniques for GCNs inside cementitious composites were only stirred by a cement paste mixer, and no other dispersion techniques were employed in this study. Nevertheless, dispersion the GCNs homogenously within the cementitious composites is a challenging work, which has a significant influence on the microstructures and performances of the composites. Therefore, it is innovative and challenging work to explore applicable and inexpensive dispersion techniques for the GCNs within cementitious composites in the future [45,63,64].

2.3 Experimental procedure

Electrical resistance of the samples was tested by two probes, DC and AC methods, using digital multimeters of Keithley 2100 (Keithley Instruments Inc., USA) and Agilent U1733C (Agilent Technologies Co., Ltd., USA), respectively. The electrical resistance was tested under seven different ages of 28, 35, 42, 49, 56, 63, and 70 days, and the photographs for testing procedures are displayed in Figure 2. In addition, the mass and its decrease ratio at different ages were measured using an electronic balance and calculated, which were used to characterize the water content variation of the specimen. The AC frequencies were varied in the range of 100 Hz to 100 kHz. Electrical resistivity was obtained from the following equation:

where ρ and R represent electrical resistivity and resistance of the sample; S is the cross-sectional area of electrodes embedded in the sample; and L represents the separation of two electrodes.

Figure 2

Electrical resistance measurements of the composites. (a) The DC testing method and (b) the AC testing method.

A universal testing equipment WDW-100C (Hualong Instrument Co., Ltd., China) was utilized to test the mechanical property and self-sensing property after the samples had cured for 28 days. The loading rate was 0.4 mm·min⁻¹, and three specimens were measured for each group. The electrical resistance was measured by a two-probe DC method when testing the self-sensing property. The test arrangements were displayed in Figure 3. Two vertical sides of the samples were attached with strain gauges to obtain the strain. Plastic films were arranged at the top and bottom of the sample to insulate the specimen away from the loading plate. Two different loading regimes include (1) cyclic loading and (2) monotonic compressive loading until failure to assess the mechanical property and the self-sensing property. The stress magnitude of the cyclic loading regime is 10 MPa. Loads applied to each specimen include six repeated progressions. For repeated loading and monotonic loading, the electrical resistance and strain were measured using the digital multimeters of Keithley 2100 and Dynamic strain measurement with type DH3820N (Donghua Testing Technology Co., Ltd, China), respectively. The load, electrical resistance, and strain were obtained using an automatic data acquisition system. The sampling rate was set as 2 Hz. The FCR can be calculated as the following equation:

where ρ t and ρ 0 denote electrical resistivity under loading and the initial electrical resistivity, respectively.

Figure 3

Testing setup of mechanical and self-sensing properties. (a) Testing setup and (b) testing of the specimen.

Stress and strain sensitivities are utilized to assess self-sensing capacity, which are obtained using the following equations, respectively:

where σ and ε represent the applied stress and strain, respectively, which correspond to the FCR .

3.1 Electrical resistivity

3.1.1 Electrical resistivity with the DC method

The mass and its decrease ratio, DC electrical resistivity, and variation ratio of DC electrical resistivity of specimens at various ages can be seen in Figure 4. As displayed in Figure 4(a) and (b) that the mass decreases as ages increase. The variation trend of the decrease ratio is decreased quickly in the first, with the age increasing from 28 to 49 days, and then it decreases slowly. The mass decrease ratio of the specimens with different contents of the GCNs and different ages is under the range of 3.0–11.4%, which reflects the variation law of the water content changes with age. The mass decrease ratio of GCN4 is decreased under the range of 4.4–11.4% with age increasing from 28 to 70 days, while the other reduction range of mass decrease ratios is within the scope of 3.0–9.1%. This indicates that the mass decrease ratio of GCN4 is a little more quickly than that of the other specimens with age increases from 28 to 70 days, which also demonstrates that the free water in the pore structures of GCN4 is easier to evaporate. The primary reasons for this phenomenon may be attributed to the microstructures of GCN4. Moreover, it is known to us that the microstructures of cementitious composites are uneven, and they present discreteness to some extent due to the properties of raw materials, manufacturing techniques, and so on [65].

Figure 4

The mass variation, DC electrical resistivity, and the corresponding variation ratio of cementitious composites under various dosages of the GCNs and different ages. (a) Mass of cementitious composites containing different dosages of the GCNs and different ages. (b) Mass decrease ratio of cementitious composites containing different dosages of the GCNs and different ages. (c) The variation of the DC electrical resistivity corresponding to different dosages of the GCNs and different ages. (d) The DC electrical resistivity corresponding to different dosages of the GCNs and different ages. (e) Decrease ratio of the DC electrical resistivity compared with the control group with different ages. (f) The variation of the DC electrical resistivity with age increases under different contents of the GCNs.

The DC electrical resistivity and its variation ratio are presented in Figure 4(c)–(f). As shown in Figure 4(c) and (d), the electrical resistivity is reduced from 2.3 × 10⁶ to 28.0 Ω·cm with an increase in the concentration of the GCNs from 0.0 to 12.0 wt% under each age. Compared with GCN0, the reduction ratio of electrical resistivity is varied from 5.7% to nearly 100.0% with an increase in the dosage of the GCNs at each age, which is shown in Figure 4(e). Particularly, the electrical resistivity reduction ratio decreases quickly with the content of the GCNs increasing from 4.0 to 10.0 wt%, while it decreases slowly when the contents of the GCNs vary within the scopes of 0.0–4.0 and 10.0–12.0 wt%. The relation of electrical resistivity and concentration of the GCNs presents evident percolation characteristics, which can be divided into three parts, as shown in Figure 4(c). The first part is the dosage of the GCNs changed within the range of 0.0–4.0 wt%. The electrical resistivity of the cementitious composites in this part is high, which is reduced from 2.3 × 10⁶ to 1.7 × 10⁶ Ω·cm; thus, it is called as insulation zone. The second part corresponds to the content of the GCNs within the range of 4.0–10.0 wt%. Electrical resistivity is reduced sharply from 1.7 × 10⁶ to 40.0 Ω·cm in this part, and it is called as percolation threshold zone. While the last part of the content of GCNs is under the scope of 10.0–12.0 wt%, the electrical resistivity is decreased sharply from 140 to 28.0 Ω·cm. The cementitious composites in this part display stabilized and lower electrical resistivity, which is named as conduction zone [1]. Thus, percolation threshold zone of the GCNs cementitious composites is under the regime of 4.0–10.0 wt%. In addition, percolation threshold zone is similar to each other when the age changes from 28 to 70 days.

In particularly, there is an interesting phenomenon that electrical resistivity changes with age presents the contrary law under different content of the GCNs, as displayed in Figure 4(f). As shown in Figure 4(f), the critical dosage of the GCNs is 4.0 wt%, the electrical resistivity increases with age increasing from 28 to 70 days, when the content of the GCNs is within the scope of 0.0–4.0 wt%, and the increase ratio is within the scope of 7.4–58.3%. However, the electrical resistivity corresponding to the range of GCNs of 6.0–12.0 wt% presents the contrary changeable law, which is decreased with age increasing from 28 to 70 days, and the decrease ratio is within the scope of 5.9–52.0%.

Reasons for these performances are as follows. The GCN0, GCN2, and GCN4 present higher electrical resistivity when the dosage of the GCNs is no more than 4.0%, as shown in Figure 4(c) and (d). The distances between the neighboring electrical conductive fillers of GNPs, CNTs, and NCBs within the cementitious composites are long corresponding to the lower content of GCNs, as shown in Figure 5(a) and (b). Thus, electrical conduction pathways are hard to form by field emission, tunneling effect, and contact conduction. Meanwhile, the water content of the specimen is within the range of 3.0–11.4%, which has been demonstrated by the mass variation of the sample, as shown in Figure 4(a) and (b). Hence, ionic conduction is the primary electrical conduction way of the GCN0, GCN2, and GCN4; thus, the free water in the pores is the key factor that affects the electrical conduction of the specimen [2]. When the free water in the pores of the specimens is reduced due to the water content is reduced with age increases, thus the electrical conduction pathways in the specimen caused by the ionic conduction is decreased with increases age. Therefore, electrical resistivity increases with age increasing when the content of the GCNs is no more than 4.0%. However, electrical resistivity is reduced rapidly as the content of the GCNs is more than 6.0%, as shown in Figure 4(c) and (d). This is because the separations between the adjacent GNPs, CNTs, and NCBs within cementitious composites are shortened with an increase in the concentration of the GCNs, and some of them are even contacted with each other as the content of the GCNs increases to more than 6.0 wt%, as shown in Figure 5(c) and (d). Consequently, more electrical conduction pathways and networks have been formed, thus the tunneling effect, field emission, and contact conduction dominate the electrical conduction way of the specimen [2,66]. Hence, the free water in the pores of the specimen will prevent the formation of the electrical conduction pathways and networks. Therefore, electrical resistivity is decreased with reducing water content, accompanied by age increases, as shown in Figure 4(f).

Figure 5

The dispersion state of different dosages of the GCNs within cementitious composites. (a) GCN2, (b) GCN4, (c) GCN8, and (d) GCN10.

3.1.2 Electrical resistivity with the AC method

AC electrical resistivity and variation ratio of the cementitious composites change with the content of the GCNs, age, and testing frequency are displayed in Figure 6. As shown in Figure 6, the AC electrical resistivity is reduced with an increase in both dosage of the GCNs and testing frequency, corresponding to ages varying from 28 to 70 days. The AC electrical resistivity reduces from 1,982.0 to 28.8 Ω·cm as the concentration of GCNs increases from 0.0 to 12.0 wt% under the age of 28 days, as displayed in Figure 6(c). Additionally, the AC electrical resistivity reduces from 6,350.0 to 9.7 Ω·cm as the concentration of the GCNs increases from 0.0 to 12.0 wt% under the age of 70 days, as shown in Figure 6(i). The corresponding decrease ratios are within the ranges of 17.4–98.9% and 30.1–99.8%, respectively. Additionally, AC electrical resistivity is reduced fast when the GCNs are increased from 4.0 to 10.0 wt%, while it decreases slowly as content of the GCNs variation under the scopes of 0.0–4.0 wt% and 10.0–12.0 wt%. This phenomenon is consistent with the change law of the DC electrical resistivity against the concentration of the GCNs.

Figure 6

The AC electrical resistivity and variation ratio of the cementitious composites change with the content of the GCNs, age, and testing frequency. (a) The AC electrical resistivity variations with the content of the GCNs, age, and testing frequency. (b) The variation ratio of AC electrical resistivity changes with the content of the GCNs, age, and testing frequency. (c) The AC electrical resistivity changes with the content of the GCNs and testing frequency under the age of 28 days. (d) The AC electrical resistivity changes with the content of the GCNs and testing frequency under the age of 35 days. (e) The AC electrical resistivity changes with the content of the GCNs and testing frequency under the age of 42 days. (f) The AC electrical resistivity changes with the content of the GCNs and testing frequency under the age of 49 days. (g) The AC electrical resistivity changes with the content of the GCNs and testing frequency under the age of 56 days. (h) The AC electrical resistivity changes with the concentration of GCNs and testing frequency under the age of 63 days. (i) The AC electrical resistivity changes with the concentration of the GCNs and testing frequency under the age of 70 days.

When the testing frequencies are increased from 100 Hz to 100 kHz, the AC electrical resistivity is reduced within the scope of 6,350–5,912 and 9.6–8.2 Ω·cm, respectively, which correspond to the contents of the GCNs of 0.0 and 12.0 wt% under the age of 70 days. The reduction ratios are within the range of 6.9–15.1%. At the same time, the AC electrical resistivity is also increased with age increases when the content of the GCNs is no more than 4.0 wt%, while it is decreased with age increases when the content of the GCNs is more than 4.0 wt%, as displayed in Figure 6(b). The scopes of the increase ratio and reduction ratio are 30.0–313.9% and 3.4–66.4%, respectively.

3.2 Mechanical property

The mechanical properties of the composites studied in the present article include the curves of stress and strain, compressive strength, peak strain, and elastic modulus. Their corresponding variation ratios are presented in Figure 7. It can be seen from Figure 7(a) that the relations of stress and strain approach linearity until the stress increases to 80% of the peak stress. And then, the stress increases with strain increasing and presents obvious nonlinearity. After the peak stress, the stress reduces sharply and nonlinear with increases of strain. The failure modes of the cementitious composites present apparent brittleness, and the ductility is improved as the concentration of the GCNs increases, as shown in Figure 8. Simultaneously, compressive strength initially increases and then reduces with increasing content of the GCNs, as displayed in Figure 7(b). The maximum compressive strength is 45.3 MPa corresponding to the dosage of the GCNs of 2.0 wt%, and it is increased by 13.3% compared with GCN0, as shown in Figure 7(e). Both the compressive strength of GCN2 and GCN4 are correspondingly higher than those of GCN0. The compressive strength decreases from 35.0 to 32.4 MPa as the concentration of the GCNs increases from 6.0 to 12.0 wt%. The corresponding reduction ratio is in the range of 12.4–19.0% compared with GCN0. Additionally, the peak strain is increased from 4,598 to 5,082 με, and the elastic modulus is decreased from 10.9 to 9.2 GPa, respectively, while the concentration of the GCNs increases from 0.0 to 12.0 wt%, as displayed in Figure 7(c) and (d). The increase ratio of peak strain and reduction ratio of the elastic modulus are within the ranges of 2.4–10.5% and 3.7–16.0%, respectively, as shown in Figure 7(e). The variation law of the peak strain and elastic modulus against the concentration of GCNs reflects that the ductility improves with increasing concentration of GCNs.

Figure 7

The mechanical performances of the cementitious composites incorporating GCNs. (a) The stress–strain curves. (b) The compressive strength. (c) The peak strain. (d) The elastic modulus. (e) Variation ratio of the compressive strength, peak strain, and elastic modulus.

Figure 8

Typical failure models: (a) GCN0, (b) GCN4, and (c) GCN10.

The mechanisms of the GCNs with various contents influencing the mechanical performances of the cementitious composites are analyzed as follows. The GCNs with content less than 4.0 wt% can disperse homogenously inside the cementitious composites, which can fill the pores and nucleation by nano-size effect during the fabrication and cement hydration processes of the cementitious composites, as shown in Figure 5(a) and (b). Moreover, the evenly distributed GNPs and CNTs can also prevent the generation and expansion of the microcracks in the cementitious composites by bridging, pull-out, and pull-off effect under external force [67]. Therefore, the mechanical performances can be improved by the evenly distributed GCNs with lower contents. However, the GCNs are easy to agglomerate with each other and hard to distribute uniformly within the cementitious composites because of the strong Van der Waals force when their dosage is higher than 4.0 wt%. The agglomerations generated by the nonhomogeneous dispersed GCNs in the cementitious composites can lead to voids and defects; it can be seen from Figure 5(c) and (d), which will result in stress concentration as cementitious composites under external force [68]. Thus, the compressive strength and elastic modulus are deteriorated. Nevertheless, it has been proven that the synergetic effect of filling, nucleation, bridging, and pull out and pull off of the GCNs can enhance the peak strain of cementitious composites [69].

3.3 Self-sensing performance

3.3.1 Self-sensing performance under cyclic load

Self-sensing performance of the GCN cementitious composites subjected to repeated load is displayed in Figure 9. To ensure the deformation of the specimen subjected to the cyclic load under the range of elastic, the amplitude of cyclic stress is selected as 10 MPa, which is lower than one-third of the minimum compressive strength of 32.4 MPa, as displayed in Figure 7(b). As presented in Figure 9, the variation law of FCR is contrary to the change law of stress and strain, which is decreased upon increase of the stress and strain and increased upon reduction of the stress and strain. The repeatability and stability of time history curves of FCR-stress/strain are improved with increasing concentration of the GCNs, as shown in Figure 9(a)–(g). Nevertheless, the electrical resistivity of GCN0 and GCN2 is increased obviously upon an increase in cyclic loading numbers. Meanwhile, the maximum amplitude of FCR, stress sensitivity, and strain sensitivity are initially increased and then decreased with increasing dosage of the GCNs. Specifically, GCN10 presents the best self-sensing property, as displayed in Figure 9(h)–(k).

Figure 9

The self-sensing property subjected to cyclic load. (a) The self-sensing property of GCN0. (b) The self-sensing property of GCN2. (c) The self-sensing property of GCN4. (d) The self-sensing property of GCN6. (e) The self-sensing property of GCN8. (f) The self-sensing property of GCN10. (g) The self-sensing property of GCN12. (h) The FCR of cementitious composites with variation of GCNs. (i) The stress sensitivity with variation of GCNs. (j) The strain sensitivity with variation of GCNs. (k) Variation of FCR, stress sensitivity, and strain sensitivity compared with the reference.

The self-sensing property of the cementitious composites can be divided into three parts according to the content of the GCNs, which is similar to the change law between electrical resistivity and concentration of the GCNs. The first part with a concentration of the GCNs of 0.0–4.0 wt% is called as insulation zone. In this part, the electrical resistivity of GCN0 and GCN2 is increased obviously upon an increase in cyclic loading numbers. Additionally, the maximum absolute FCR and stress/strain sensitivity of GCN0, GCN2, and GCN4 are lower, which are in the ranges of 0.2–4.6%, 0.02–0.46%/MPa, and 2.1–46.9, respectively, as shown in Figure 9(h)–(j). Compared with GCN0, the increase ratios of the maximum absolute FCR and stress/strain sensitivity of GCN2 and GCN4 are in the ranges of 50.0–2,200.0%, 50.0–2,200.0%, and 45.9–2,058.0%, respectively, as shown in Figure 9(k). As the concentration of the GCNs increases to the second part of 4.0–10.0 wt%, which is called the percolation threshold zone, the relations between the FCR, stress, and strain with time exhibit better repeatability and reproducibility, as shown in Figure 9(d)–(f). In addition, the maximum absolute FCR and stress/strain sensitivity of GCN6, GCN8, and GCN10 are increased to the ranges of 6.2–10.3%, 0.62–1.03%/MPa, and 61.2–95.8, respectively, as shown in Figure 9(h)–(j). Compared with GCN0, the increase ratios of the maximum absolute FCR and stress/strain sensitivity of GCN6, GCN8, and GCN10 are in the scopes of 3,000.0–5,050.0%, 3,000.0–5,050.0%, and 2,749.1–4,360.1%, respectively, as shown in Figure 9(k). In particular, GCN10 presents the best self-sensing property, and its maximum absolute FCR and stress/strain sensitivity are achieved at 10.3%, 1.03%/MPa, and 95.8, respectively. Finally, as the concentration of the GCNs increases to the third part of 10.0–12.0 wt%, which belongs to the electrical conduction zone. The FCR–stress/strain time history curve of GCN12 expresses favorable stability and repeatability. Its maximum absolute FCR and stress/strain sensitivity are 7.1%, 0.71%/MPa, and 65.0, respectively, which are correspondingly lower than those of GCN10. Compared with GCN0, the increase ratios of the maximum absolute FCR and stress/strain sensitivity of GCN12 are increased by 3,450.0, 3,450.0, and 2,923.8%, respectively.

Mechanisms of the self-sensing property under cyclic load affected by the GCNs are explored in the following. The concentration of the GCNs of GCN0, GCN2, and GCN4 is low, and they belong to the insulation zone. The separations between the electrical conduction fillers of GNPs, CNTs, and NCBs are large, as shown in Figure 5(a). Thus, the electrical conduction pathway in the cementitious composites is difficult to form. Nevertheless, the pores are nearly full of free water under the age of 28 days, which is verified in Section 3.1 that the water content of the composites is within the scope of 3.0–11.4%, as shown in Figure 4(b). Moreover, it has been demonstrated by Chung [5], Han et al. [70], and Ding et al. [2] that when the dosage of electrical conduction fillers is less than percolation threshold, together with the water content of the specimen in the range of 3.3–9.9%, water filling pores can dissolve ionic from solid states, leading to ionic conduction from connected capillary pores. At the same time, it is verified in Section 3.1 that the percolation threshold zone of GCNs cementitious composites is 4.0–10.0 wt%. Therefore, the primary conduction way of the cementitious composites is ionic conduction when the concentration of the GCNs is no more than 4.0 wt%. Because of the conduction way of ionic is related to positive and negative ions move to the opposite electrodes in the pore solution, thus increase the electrical resistivity. These are the main reasons for the electrical resistivity of GCN0 and GCN2, which is obviously increased with an increase in cyclic loading numbers. Although the separations between the electrical conduction fillers of GNPs, CNTs, and NCBs are shortened when a load is applied to the specimen, the electrical conduction pathways are still hard to form within the cementitious composites due to the initial large distances between the electrical conduction fillers of GNPs, CNTs, and NCBs, leading to tunneling effect, field emission, and contact conduction hard to take place [18]. As a result, the GCN0, GCN2, and GCN4 exhibit higher electrical resistivity and lower self-sensing performance.

The GCNs’ concentration range of GCN6, GCN8, and GCN10 is 4.0–10.0 wt%, which belongs to the percolation threshold zone. Thus, the distance between the neighboring electrical conduction fillers of GNPs, CNTs, and NCBs becomes shorter with increasing content of the GCNs. Some of the electrical conduction fillers begin to touch with each other, which can be seen from Figure 5(b). Therefore, more electrical conduction pathways are formed conveniently by the tunneling effect and field emission, together with contact conduction. The former two electrical conduction ways are the primary electrical conduction ways of the cementitious composites. As a consequence, the electrical conduction pathways are easy to change when the specimen is under compression. Consequently, the GCN6, GCN8, and GCN10 present lower electrical resistivity and the optimum self-sensing property.

As the dosage of the GCNs increases to the electrical conduction zone of 10.0–12.0 wt%, the electrical conduction fillers of GNPs, CNTs, and NCBs are closer to each other; furthermore, the distances between the adjacent electrical conduction fillers are further shortened. Some of the electrical conduction fillers are even contacted with each other, which can be seen from Figure 5(d). Thus, lots of electrical conduction networks are generated stably by the tunneling effect, field emission, and contact conduction. Thus, contact conduction is the primary electrical conduction way of the composites. However, the stable electrical conduction networks are hard to change when the specimen is subjected to compression. Consequently, GCN12 presents the lowest electrical resistivity and higher self-sensing property.

Because of NCBs present advantages as fine filling effect, stability at high thermal, low cost, and electrical resistivity is lowered by about 10⁻¹ Ω·cm; thus, NCBs become attractive electrical conduction fillers. Li et al. [4] and Hussain et al. [44] indicate that the percolation of cementitious composites containing NCBs is 1.0–2.0 wt%. For the reason that the high specific surface area, they can tightly adhere to each other and agglomerate easily within cementitious composites through the van der Waals force, resulting in defective sites inside the composites [35,36,37,38]. At the same time, the NCBs present high water absorption due to the rough surface, which can decrease the effective water-to-binder ratio of the composites. To address the disadvantages of intrinsic self-sensing cementitious composites with single electrical conduction fillers, some researchers proposed hybrid electrical conduction fillers due to their synergistic effect [45,46]. Zhang et al. used the NCBs together with CNTs to develop self-sensing cementitious composites. The NCBs and CNTs with short range and long range can stimulate the generation of extensive conduction pathways inside the cementitious composites; thus, more distance among the electrical conduction fillers can be changed and lead to sensitive self-sensing performance subjected to compression. These composites with a concentration of 2.40 vol% of the hybrid NCBs and CNTs present stable sensing properties, and the maximum magnitude of FCR is reached up to 25.4% [47,48]. Although the percolation of the GCNs in the present article is 4.0–10.0 wt%, which is higher than that in the previous research, they presented a higher single-to-noise ratio and sensitivity, as well as a simplified dispersion technique.

3.3.2 Self-sensing performance subjected to monotonic load

Figure 10 displays relations of the FCR, stress, and strain with time of the cementitious composites containing the GCNs subjected to monotonic compression load. It is illustrated in Figure 10 that FCR is reduced upon increasing the stress and strain until the stress reaches the peak value. The decrease ratio is increased upon an increase in the concentration of the GCNs. After the peak stress, the variation tendency of FCR changes with stress and strain is different, which depends on the content of the GCNs. The maximum absolute FCR and stress/strain sensitivity are initially increased and then decreased with increasing concentration of the GCNs. Similarly, GCN10 presents the best self-sensing performance subjected to the monotonic compressive load, its maximum absolute FCR and stress/sensitivity are achieved at 24.9%, 0.77%/MPa, and 49.9, respectively, and they are correspondingly increased by 54.3, 90.5, and 42.5% compared with GCN0.

Figure 10

Self-sensing performance subjected to monotonic load. (a) The time history of the FCR–stress–strain of GCN0. (b) The time history of the FCR–stress–strain of GCN2. (c) The time history of the FCR–stress–strain of GCN4. (d) The time history of the FCR–stress–strain of GCN6. (e) The time history of the FCR–stress–strain of GCN8. (f) The time history of the FCR–stress–strain of GCN10. (g) The time history of the FCR–stress–strain of GCN12. (h) The FCR variation with the content of the GCNs. (i) The stress sensitivity variations with the content of the GCNs. (j) The strain sensitivity variations with the content of the GCNs. (k) The variation ratio of FCR and stress/strain sensitivity containing different concentrations of the GCNs compared with the control group.

As shown in Figure 10(a)–(c), the FCR of GCN0, GCN2, and GCN4 is decreased slowly with an increase in stress and strain until the peak stress. And then, the FCR decreased rapidly under the peak stress in the first, after that, it nearly keeps at constant, as shown in Figure 10(a)–(c). The maximum absolute FCR and stress/strain sensitivity of GCN0, GCN2, and GCN4 are ranged in the scopes of 16.1–17.9%, 0.39–0.40%/MPa, and 35.0–38.1, respectively, as depicted in Figure 10(h)–(j). Compared with GCN0, the increased percentages of the maximum absolute FCR and stress/strain sensitivity of GCN2 and GCN4 are within the ranges of 0.0–11.5%, and 0.0–8.8%, respectively, as illustrated in Figure 10(k). However, the decrease ratio of FCR of GCN6, GCN8, GCN10, and GCN12 changes with dosage increases of the GCNs until the peak stress is more quickly than those of the GCN0 to GCN4. After the peak stress, the FCR increases as the stress and strain decrease, as shown in Figure 10(d)–(g). The maximum absolute FCR and stress/strain sensitivity of the GCN6, GCN8, GCN10, and GCN12 are in the scopes of 18.2–24.9%, 0.52–0.77%/MPa, and 37.3–49.9, respectively, as shown in Figure 10(h)–(j). Compared with GCN0, the increased percentages of the maximum absolute FCR and stress/strain sensitivity of GCN6, GCN8, GCN10, and GCN12 are within the ranges of 13.0–54.3, 29.0–90.5, and 6.7–42.5%, respectively, as illustrated in Figure 10(k). Therefore, the FCR decreases nearly linearly with increasing stress and strain until loading to 80% of the peak stress, and then it decreases nonlinearly with increasing stress and strain. The FCR is decreased and increased suddenly with different concentrations of the GCNs when loading to the peak stress. Hence, these characteristics can be applied in detecting and evaluating the stress, strain, and deformation conditions of concrete structures. Mechanisms of the GCNs that affect the self-sensing performance of the cementitious composites withstanding monotonic load resemble those of the cementitious composites subjected to repeated load as analyzed in Section 3.3.1.

The optimum FCR and stress/strain sensitivity are −24.9% and 1.0%/MPa/95.8, respectively, corresponding to the content of the GCNs of 10.0 wt%. Nevertheless, the compressive strength and elastic modulus of this composite are 32.4 MPa and 9.33 GPa, respectively, which are lowered down 19.0 and 14.7%, respectively, compared with the control group. This is because the higher content of the GCNs is prone to aggregate in cementitious composites, thus leading to defects generated in the microstructures [45]. On the one hand, a reasonable dispersion method should be explored to disperse the GCNs homogenously within the cementitious composites, thus counterbalancing the harmful effects of the higher concentration brought in. On the other hand, the mechanical performances of the self-sensing cementitious composites should be compatible with the monitoring of concrete structures. Therefore, an appropriate dosage of the GCNs needs to be designed as taking into account the fabrication technique, piezoresistivity, and mechanical performances according to the requirements of practical civil infrastructure [64].