Inﬂuence of Powder and Liquid Multi-Wall Carbon Nanotubes on Hydration and Dispersion of the Cementitious Composites

: Two types of multi-walled carbon nanotubes (MWCNTs), powder and liquid, were added to cementitious composites to build self-sensing concrete. To properly evaluate and quantify the e ﬀ ect of MWCNTs on electrical resistance, various tests, including isothermal conduction calorimetry, were carried out. One of primary issues of self-monitoring concrete is dispersion, so ultrasonication was used to properly mix the CNTs in the dispersion solution, and silica fume was employed to make the specimens. Scanning electron microscopy (SEM), Raman spectroscopy, and porosity analyses were performed to investigate the physical properties of the composites and to conﬁrm uniform dispersion. The distance of the electrical resistance was also measured, and the dosages and types of MWCNTs were analyzed.


Introduction
Concrete is being employed to build engineering infrastructure, causing energy and environmental issues (e.g., carbon dioxide emissions). However, compared with other materials in the construction industry such as steel, metal, and even polymers, it is better than others in terms of energy consumption and environmental problems. This is a primary reason why concrete is used for most infrastructures. The increasing demand of health-monitoring concrete is a critical global challenge of twenty-first century. In the early 1990s, many researchers and scholars experimented with self-sensing concrete by mixing functional fillers with concrete composites, including carbon fiber, graphite powder, steel slag, carbon nanotubes, carbon blocks, or a hybrid of multiple fillers [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18].
The carbon nanotube (CNT) has been utilized in a variety of industrial fields due to its outstanding mechanical properties, high thermal conductivity, low unit mass, and exceptional corrosion resistance. In the field of construction, CNTs have been added in concrete composites to create smart materials (called self-monitoring concrete) due to their excellent electrical conductivity. In general, the CNT exhibits a considerably low density of 1.3-1.4 g/cm 3 and, when added to cement concrete mixtures, the amount of CNTs does not exceed a binder weight of 2.0%. In addition, the effective dispersion of CNTs in the composites is very difficult to obtain due to van der Waals forces acting between the CNT particles. If CNTs are not properly dispersed, they are also known to limit improvements in the mechanical properties of cement composites. In order to ensure homogeneous dispersion in cement composites, there are various techniques to apply, such as ultrasonication, silica fume, water-reducing agent (KS F 2560 [33], Chemical Admixtures) was added to the cementitious composites. Detailed information on the mixture proportions used in all tests is demonstrated in Table 4. polycarboxylate-based, high-performance, water-reducing agent (KS F 2560 [33], Chemical Admixtures) was added to the cementitious composites. Detailed information on the mixture proportions used in all tests is demonstrated in Table 4.
(a) (b) Figure 1. Pictures of (a) powder type and (b) liquid type multi-wall carbon nanotubes.        Figure 2 displays the experimental equipment. To measure the heat of hydration, a cylindrical specimen with a diameter of 100 mm and a height of 200 mm was fabricated, and a thermocouple was then embedded. It was stored in an insulated cuboid box with dimensions of 300 × 300 × 400 mm 3 , and the heat of hydration was recorded for 72 h at one-minute intervals. Isothermal conduction calorimetry analysis was conducted to measure the rate of heat evolution of the cement paste for 36 h. The compressive strength of the cementitious composites was measured as per KS L ISO 679 [34] at 3, 7, 14, and 28 days. After finishing the measurements of the compressive strength tests at the age of 28 days, SEM analysis was performed with a sample collected from the specimen tested. For SEM analysis, the sample was dried at 50 °C for four hours, and the specimen was kept in a vacuum state for two hours before the test. For Raman spectroscopy analysis, a cylindrical specimen with a diameter of 70 mm and height of 10 mm was fabricated, and a Raman spectrometer (NRS-3200) was used. In order to measure the pore sizes of the composites, the method of mercury intrusion porosimetry was employed, and a sample at the age of 28 days and dried for four hours at 50 °C was employed by using the Autopore 9600IV (measurement range of 0.003-900 μm), presented in Figure 2d. The cylindrical specimen for the porosity test was 10 mm in diameter and 10 mm in height (see Figure 3). Figure 4 demonstrates the schematic of the cementitious materials for the measurement of electrical resistance. Copper plates were installed at 20 mm intervals to investigate the effect of distance. Specimens were demolded 24 h after casting. After demolding, all specimens were curried as per KS L ISO 679 [34] at a room temperature of 20 ± 1 °C and a relative humidity of 60%. Moreover, SEM and Raman spectroscopy were conducted to investigate the physical properties of the composites.   Isothermal conduction calorimetry analysis was conducted to measure the rate of heat evolution of the cement paste for 36 h. The compressive strength of the cementitious composites was measured as per KS L ISO 679 [34] at 3, 7, 14, and 28 days. After finishing the measurements of the compressive strength tests at the age of 28 days, SEM analysis was performed with a sample collected from the specimen tested. For SEM analysis, the sample was dried at 50 • C for four hours, and the specimen was kept in a vacuum state for two hours before the test. For Raman spectroscopy analysis, a cylindrical specimen with a diameter of 70 mm and height of 10 mm was fabricated, and a Raman spectrometer (NRS-3200) was used. In order to measure the pore sizes of the composites, the method of mercury intrusion porosimetry was employed, and a sample at the age of 28 days and dried for four hours at 50 • C was employed by using the Autopore 9600IV (measurement range of 0.003-900 µm), presented in Figure 2d. The cylindrical specimen for the porosity test was 10 mm in diameter and 10 mm in height (see Figure 3). Figure 4 demonstrates the schematic of the cementitious materials for the measurement of electrical resistance. Copper plates were installed at 20 mm intervals to investigate the effect of distance. Specimens were demolded 24 h after casting. After demolding, all specimens were curried as per KS L ISO 679 [34] at a room temperature of 20 ± 1 • C and a relative humidity of 60%. Moreover, SEM and Raman spectroscopy were conducted to investigate the physical properties of the composites. Isothermal conduction calorimetry analysis was conducted to measure the rate of heat evolution of the cement paste for 36 h. The compressive strength of the cementitious composites was measured as per KS L ISO 679 [34] at 3, 7, 14, and 28 days. After finishing the measurements of the compressive strength tests at the age of 28 days, SEM analysis was performed with a sample collected from the specimen tested. For SEM analysis, the sample was dried at 50 °C for four hours, and the specimen was kept in a vacuum state for two hours before the test. For Raman spectroscopy analysis, a cylindrical specimen with a diameter of 70 mm and height of 10 mm was fabricated, and a Raman spectrometer (NRS-3200) was used. In order to measure the pore sizes of the composites, the method of mercury intrusion porosimetry was employed, and a sample at the age of 28 days and dried for four hours at 50 °C was employed by using the Autopore 9600IV (measurement range of 0.003-900 μm), presented in Figure 2d. The cylindrical specimen for the porosity test was 10 mm in diameter and 10 mm in height (see Figure 3). Figure 4 demonstrates the schematic of the cementitious materials for the measurement of electrical resistance. Copper plates were installed at 20 mm intervals to investigate the effect of distance. Specimens were demolded 24 h after casting. After demolding, all specimens were curried as per KS L ISO 679 [34] at a room temperature of 20 ± 1 °C and a relative humidity of 60%. Moreover, SEM and Raman spectroscopy were conducted to investigate the physical properties of the composites.     Figure 5 presents the heat of hydration curve. Since the MWCNTs were added, regardless of whether they were powder or liquid, it was observed that the heat of the hydration decreased by approximately 10 °C to 14 °C compared with the plain cement paste. In Figure 4, MW-P1.0, MW-P2.0, and MW-L signify the powder MWCNTs, with a 1.0% mass fraction of cement and silica fume for the powder MWCNTs and a 2.0% mass fraction of cement and silica fume for the liquid MWCNTs, respectively. The primary reason for the rapid shift of the heat of hydration to the right was the high thermal conductivity of the MWCNTs. In addition, the maximum values of the heat of hydration for the plain, MW-P1.0, MW-P 2.0, and MW-L MWCNTs were measured at 14, 30, 46, and 26 h, respectively. This is not due to the direct impact of the MWCNTs, but because of condensation delay, as the amount of the chemical admixture was excessively added at 5-12%.  Figure 5 presents the heat of hydration curve. Since the MWCNTs were added, regardless of whether they were powder or liquid, it was observed that the heat of the hydration decreased by approximately 10 • C to 14 • C compared with the plain cement paste. In Figure 4, MW-P1.0, MW-P2.0, and MW-L signify the powder MWCNTs, with a 1.0% mass fraction of cement and silica fume for the powder MWCNTs and a 2.0% mass fraction of cement and silica fume for the liquid MWCNTs, respectively. The primary reason for the rapid shift of the heat of hydration to the right was the high thermal conductivity of the MWCNTs. In addition, the maximum values of the heat of hydration for the plain, MW-P1.0, MW-P 2.0, and MW-L MWCNTs were measured at 14, 30, 46, and 26 h, respectively. This is not due to the direct impact of the MWCNTs, but because of condensation delay, as the amount of the chemical admixture was excessively added at 5-12%.  Figure 6 describes the heat flow as a function of age for four different composites. In general, as the MWCNTs were added, the cumulative heat of the composites tended to decrease slightly compared with the plain cement paste. The cumulative heat decreased with the increasing amount of powder MWCNTs because the heat generated during hydration was released by the MWCNTs. However, the reduced heat was approximately 4~12 Joule/g, denoting no significant difference, and there was no effect of the MWCNTs on the strength due to their chemical reactions in the composites.

Compressive Strength
The compressive strengths of four different composites are shown in Figure 7. Similar to the findings of previous research conducted by Oh et al. [35,36], the decreasing trend of compressive strength was exhibited as the MWCNTs were added, and the strength further decreased as the addition ratio increased. The compressive strength decreased by 10 to 16 MPa, compared with the plain cement paste at 28 days. The chemical reaction of the MWCNTs did not occur in the cement composite as it did in the heat flow results because the decrease in compressive strength resulted from the intorsion and agglomeration of the MWCNTs, caused by the van der Waals forces in the  Figure 6 describes the heat flow as a function of age for four different composites. In general, as the MWCNTs were added, the cumulative heat of the composites tended to decrease slightly compared with the plain cement paste. The cumulative heat decreased with the increasing amount of powder MWCNTs because the heat generated during hydration was released by the MWCNTs. However, the reduced heat was approximately 4~12 Joule/g, denoting no significant difference, and there was no effect of the MWCNTs on the strength due to their chemical reactions in the composites.  Figure 6 describes the heat flow as a function of age for four different composites. In general, as the MWCNTs were added, the cumulative heat of the composites tended to decrease slightly compared with the plain cement paste. The cumulative heat decreased with the increasing amount of powder MWCNTs because the heat generated during hydration was released by the MWCNTs. However, the reduced heat was approximately 4~12 Joule/g, denoting no significant difference, and there was no effect of the MWCNTs on the strength due to their chemical reactions in the composites.

Compressive Strength
The compressive strengths of four different composites are shown in Figure 7. Similar to the findings of previous research conducted by Oh et al. [35,36], the decreasing trend of compressive strength was exhibited as the MWCNTs were added, and the strength further decreased as the addition ratio increased. The compressive strength decreased by 10 to 16 MPa, compared with the plain cement paste at 28 days. The chemical reaction of the MWCNTs did not occur in the cement composite as it did in the heat flow results because the decrease in compressive strength resulted from the intorsion and agglomeration of the MWCNTs, caused by the van der Waals forces in the

Compressive Strength
The compressive strengths of four different composites are shown in Figure 7. Similar to the findings of previous research conducted by Oh et al. [35,36], the decreasing trend of compressive strength was exhibited as the MWCNTs were added, and the strength further decreased as the addition ratio increased. The compressive strength decreased by 10 to 16 MPa, compared with the plain cement paste at 28 days. The chemical reaction of the MWCNTs did not occur in the cement composite as it did in the heat flow results because the decrease in compressive strength resulted from the intorsion and agglomeration of the MWCNTs, caused by the van der Waals forces in the cement composite. It was suggested that if the nano-sized MWCNTs do not achieve homogeneous dispersion, they cannot lead to densification due to the micro-filler effect in the cement composite. However, the MW-L case was relatively dispersive compared with MW-P1.0 or MW-P2.0, so the decrease in strength of the liquid type was not greater than the powder type.
Appl. Sci. 2020, 10, x FOR PEER REVIEW 7 of 13 cement composite. It was suggested that if the nano-sized MWCNTs do not achieve homogeneous dispersion, they cannot lead to densification due to the micro-filler effect in the cement composite. However, the MW-L case was relatively dispersive compared with MW-P1.0 or MW-P2.0, so the decrease in strength of the liquid type was not greater than the powder type.  Figures 8 and 9 show the cumulative pore volume and pore size distribution, respectively, measured by mercury intrusion porosimetry. In Figure 8, the pore volume in the cement composites increased with the addition of MWCNTs. In theory [35,36], MWCNTs in the size range of 5-100 nm reduced the pore volume due to the filling effect of the composites. However, if homogeneous dispersion was not achieved, the pore volume increased. In particular, the pore size distribution in Figure 9 revealed that the amount of relatively large micro-pores increased with the MWCNTs, while that of small micro-pores decreased as the MWCNTs were added. This implies that the intorsion and agglomeration of the MWCNTs occurred due to van der Waals forces, and thus the compressive strength of the cement composite incorporating the MWCNTs decreased as large micro-pores appeared.   Figures 8 and 9 show the cumulative pore volume and pore size distribution, respectively, measured by mercury intrusion porosimetry. In Figure 8, the pore volume in the cement composites increased with the addition of MWCNTs. In theory [35,36], MWCNTs in the size range of 5-100 nm reduced the pore volume due to the filling effect of the composites. However, if homogeneous dispersion was not achieved, the pore volume increased. In particular, the pore size distribution in Figure 9 revealed that the amount of relatively large micro-pores increased with the MWCNTs, while that of small micro-pores decreased as the MWCNTs were added. This implies that the intorsion and agglomeration of the MWCNTs occurred due to van der Waals forces, and thus the compressive strength of the cement composite incorporating the MWCNTs decreased as large micro-pores appeared.

Porosity
Appl. Sci. 2020, 10, x FOR PEER REVIEW 7 of 13 cement composite. It was suggested that if the nano-sized MWCNTs do not achieve homogeneous dispersion, they cannot lead to densification due to the micro-filler effect in the cement composite. However, the MW-L case was relatively dispersive compared with MW-P1.0 or MW-P2.0, so the decrease in strength of the liquid type was not greater than the powder type.  Figures 8 and 9 show the cumulative pore volume and pore size distribution, respectively, measured by mercury intrusion porosimetry. In Figure 8, the pore volume in the cement composites increased with the addition of MWCNTs. In theory [35,36], MWCNTs in the size range of 5-100 nm reduced the pore volume due to the filling effect of the composites. However, if homogeneous dispersion was not achieved, the pore volume increased. In particular, the pore size distribution in Figure 9 revealed that the amount of relatively large micro-pores increased with the MWCNTs, while that of small micro-pores decreased as the MWCNTs were added. This implies that the intorsion and agglomeration of the MWCNTs occurred due to van der Waals forces, and thus the compressive strength of the cement composite incorporating the MWCNTs decreased as large micro-pores appeared.   Figure 10 shows the Raman spectroscopy specimens, and the results of the spectrum analysis using the Raman spectrometer are provided in Figure 11. The MWCNT-added specimen exhibited a peak at 1350 cm −1 , which was the same as in the graphite spectrum (called G band, with a range of 1330-1360 cm −1 ) [37]. The intensity of the spectrum was the highest (about 78) in the MW-L MWCNTs, indicating that relatively homogeneous dispersion was achieved in the cement composite, compared with the powder MWCNTs.   Figure 10 shows the Raman spectroscopy specimens, and the results of the spectrum analysis using the Raman spectrometer are provided in Figure 11. The MWCNT-added specimen exhibited a peak at 1350 cm −1 , which was the same as in the graphite spectrum (called G band, with a range of 1330-1360 cm −1 ) [37]. The intensity of the spectrum was the highest (about 78) in the MW-L MWCNTs, indicating that relatively homogeneous dispersion was achieved in the cement composite, compared with the powder MWCNTs.

Raman Spectroscopy
Appl. Sci. 2020, 10, x FOR PEER REVIEW 8 of 13 Figure 9. Pore size distribution. Figure 10 shows the Raman spectroscopy specimens, and the results of the spectrum analysis using the Raman spectrometer are provided in Figure 11. The MWCNT-added specimen exhibited a peak at 1350 cm −1 , which was the same as in the graphite spectrum (called G band, with a range of 1330-1360 cm −1 ) [37]. The intensity of the spectrum was the highest (about 78) in the MW-L MWCNTs, indicating that relatively homogeneous dispersion was achieved in the cement composite, compared with the powder MWCNTs.   Figure 12 presents the SEM images of the MWCNT composites. In the powder type specimen, it was obviously observed that the intorsion and agglomeration of MWCNTs took place because of van der Waals forces. In MW-P 1.0, MWCNTs were not found in the SEM image due to agglomeration, while intorsion and agglomeration, like a ball of thread, was clearly identified in the SEM image of MW-P 2.0. However, in the MW-L type, there was the small amount of MWCNT agglomeration, which was relatively dispersed within the structure. This indicates that the liquid MWCNTs can ensure better dispersibility in cement composites than the powder type.   Figure 12 presents the SEM images of the MWCNT composites. In the powder type specimen, it was obviously observed that the intorsion and agglomeration of MWCNTs took place because of van der Waals forces. In MW-P 1.0, MWCNTs were not found in the SEM image due to agglomeration, while intorsion and agglomeration, like a ball of thread, was clearly identified in the SEM image of MW-P 2.0. However, in the MW-L type, there was the small amount of MWCNT agglomeration, which was relatively dispersed within the structure. This indicates that the liquid MWCNTs can ensure better dispersibility in cement composites than the powder type.

SEM Analysis
Appl. Sci. 2020, 10, x FOR PEER REVIEW 9 of 13 Figure 11. Raman spectrum. Figure 12 presents the SEM images of the MWCNT composites. In the powder type specimen, it was obviously observed that the intorsion and agglomeration of MWCNTs took place because of van der Waals forces. In MW-P 1.0, MWCNTs were not found in the SEM image due to agglomeration, while intorsion and agglomeration, like a ball of thread, was clearly identified in the SEM image of MW-P 2.0. However, in the MW-L type, there was the small amount of MWCNT agglomeration, which was relatively dispersed within the structure. This indicates that the liquid MWCNTs can ensure better dispersibility in cement composites than the powder type.

Electric Resistance
The electrical resistance values, according to the measurement distance, are demonstrated in Figure 13. It was found that (1) as the addition ratio of the MWCNTs increased, the electrical resistance value decreased, and (2) the shorter the measurement distance, the smaller the electrical resistance value. Even though MWCNT agglomeration occurred in the cement composites, it spread throughout the entire structure, reducing the electrical resistance value. Furthermore, the electrical resistance decreased even in the liquid state specimen, denoting relatively good dispersion. This finding implies that the MWCNT-added cement composite exhibits electrical conductivity, regardless of homogeneous dispersion.

Electric Resistance
The electrical resistance values, according to the measurement distance, are demonstrated in Figure 13. It was found that (1) as the addition ratio of the MWCNTs increased, the electrical resistance value decreased, and (2) the shorter the measurement distance, the smaller the electrical resistance value. Even though MWCNT agglomeration occurred in the cement composites, it spread throughout the entire structure, reducing the electrical resistance value. Furthermore, the electrical resistance decreased even in the liquid state specimen, denoting relatively good dispersion. This finding implies that the MWCNT-added cement composite exhibits electrical conductivity, regardless of homogeneous dispersion.

Conclusions
From the investigation of the effects of MWCNT properties on dispersion in cementitious composites, the following findings and conclusions were obtained: 1. The heat of hydration decreased with an increase in the addition ratio of the MWCNTs because retardation occurred, due to an excessive amount of the chemical admixture. In addition, as the amount of MWCNTs increased, the amount of the chemical admixture increased, because the powder volume increased in the specimen. On the other hand, the liquid MWCNT showed further improvement in workability compared with the powder MWCNT, thus reducing the amount of the chemical admixture, and the hydration delay was not as great as that in the powder MWCNT. 2. The heat flow decreased slightly with the addition of MWCNTs, but the degree of the decrease was not significant. The MWCNTs exhibited no chemical reactions in the composites. Therefore, it was concluded that the decrease in the mechanical performance of the cementitious composite was not due to the chemical reaction, but due to the physical properties. In particular, it was due to the MWCNT agglomeration caused by van der Waals forces. 3. The compressive strength tended to decrease as MWCNTs were added. In particular, the powder type showed a greater reduction than the liquid type, and it was found that the MWCNTs did not obtain the micro-filler effect in the mixtures, but rather acted as a strength reduction factor. 4. The porosity analysis indicated that the porosity increased as MWCNTs were added, because the MWCNT agglomeration occurred due to failure to secure uniform dispersion in the cementitious composite. Thus, it existed in the form of large capillary pores, which resulted in an increase in the porosity, leading to a decrease in compressive strength. However, when the liquid MWCNT was added, the porosity further decreased compared with the powder type, which implies that the liquid type was advantageous in terms of dispersibility.

Conclusions
From the investigation of the effects of MWCNT properties on dispersion in cementitious composites, the following findings and conclusions were obtained: 1.
The heat of hydration decreased with an increase in the addition ratio of the MWCNTs because retardation occurred, due to an excessive amount of the chemical admixture. In addition, as the amount of MWCNTs increased, the amount of the chemical admixture increased, because the powder volume increased in the specimen. On the other hand, the liquid MWCNT showed further improvement in workability compared with the powder MWCNT, thus reducing the amount of the chemical admixture, and the hydration delay was not as great as that in the powder MWCNT.

2.
The heat flow decreased slightly with the addition of MWCNTs, but the degree of the decrease was not significant. The MWCNTs exhibited no chemical reactions in the composites. Therefore, it was concluded that the decrease in the mechanical performance of the cementitious composite was not due to the chemical reaction, but due to the physical properties. In particular, it was due to the MWCNT agglomeration caused by van der Waals forces. 3.
The compressive strength tended to decrease as MWCNTs were added. In particular, the powder type showed a greater reduction than the liquid type, and it was found that the MWCNTs did not obtain the micro-filler effect in the mixtures, but rather acted as a strength reduction factor. 4.
The porosity analysis indicated that the porosity increased as MWCNTs were added, because the MWCNT agglomeration occurred due to failure to secure uniform dispersion in the cementitious composite. Thus, it existed in the form of large capillary pores, which resulted in an increase in the porosity, leading to a decrease in compressive strength. However, when the liquid MWCNT was added, the porosity further decreased compared with the powder type, which implies that the liquid type was advantageous in terms of dispersibility.

5.
Through Raman spectroscopy analysis, it was revealed that the MWCNTs showed G bands (1350 cm −1 ) in their Raman spectra of graphite. This confirmed the presence or absence of MWCNTs in the composite, and the strength of the peak was highest in the liquid type of MWCNTs, presenting more homogeneous dispersion than the powder MWCNT. 6.
In the SEM analysis, agglomeration was clearly observed in the powder MWCNT, and the form of a skein of thread existed. It failed to disperse as it separated from the hydrates in the composite, which resulted in a decrease in strength, due to the presence of pores. However, the liquid MWCNT was relatively well dispersed in the overall zones. 7.
The electrical resistance value decreased as the addition ratio of MWCNTs increased. The electrical conductivity also increased and was expected to help achieve self-sensing abilities.
In relation to the degree of dispersion, according to the types of MWCNTs, the liquid type was found to be superior to the powder type. It can be confirmed that improvement in the mechanical performance of cement composites incorporating MWCNTs is determined by the degree of dispersion, and it is appropriate to use liquid MWCNTs in all areas, including workability, hydration performance, and the compressive strength of the composite. However, when the liquid MWCNT was used, the composites turned black in color. This problem requires further examination. In addition, it is necessary to address the limitations of mass production technology for dispersion and the limits of the current technology capable of dispersing the powder MWCNT up to 7%.