Introduction
Sustainability is a major concern for concrete in civil engineering. Portland cement is the significant contributor to a sustainable development of infrastructure, because it is the most widely used construction material. In 2015, the amount of CO2 emission on a global scale was 39.2 billion tons of carbon per year [1]. The emission of 1044 lbs CO2 per 2205 lbs Portland cement production was reported in the US [2]. Concrete structures are exposed to several environmental conditions and are accordingly susceptible to a lot of damage. Considering the huge amount of CO2 emitted every year and high probability for concrete structures to be deteriorated, improving the durability of concrete as well as its mechanical properties are of high importance, both environmentally and economically [3–7]. Sustainability in concrete structures involves several features such as designing for long-term durability and having a minimal impact on the environment considering the structure’s life-cycle [8–10]. Long-term durability is the most important feature of the sustainability of concrete structures since prolonging their service life through enhanced durability has considerable benefits in terms of reducing the maintenance of operations and optimizing the resources [11]. Incorporation of nanomaterials has been found to be a promising approach to improving the pore structure of concrete because the modulus of elasticity, strength, and durability properties are subsequently enhanced [12–16]. Recently, carbon nano tubes (CNT) has been introduced in order to improve the mechanical and durability properties of concrete [17–19]. It has been reported that by adding up to 1 wt %. MWCNTs (Multi wall carbon nanotubes), the porosity and microstructure of cement paste can be improved and the mesopores are drastically reduced [20]. A detailed study on the effects of MWCNTs concentration and aspect ratio was conducted by Konsta-Gdoutos et al. [21]. Nanoindentation results proved that the use of MWCNTs can increase the amount of high stiffness calcium silicate hydrate (C-S-H) and decrease the porosity. Metaxa et al. used the nanoindentation technique to estimate the volume fraction of the capillary pores [22]. It was found that, MWCNTs reduce the amount of fine pores by filling the area between the C-S-H gels. Also, nanoimaging of the fracture surfaces of cement nanocomposites has shown that CNTs reinforce cement paste by bridging nanocracks and pores. Although most of the researches have reported a decisive conclusion on the porosity of the concrete samples, the ways of measuring porosity have been stated to be difficult because of the hydration products that form a tortuous structure [23]. More importantly, the results depend on the drying methods prior to the porosimetry process [24]. To the best of knowledge, no one has performed a review on the pore structure analysis of CNT reinforced cement composites with regards to mercury intrusion porosimetry (MIP) and contact angle. In this scope, this review paper involves analyzing the effect of CNT on microstructure and the pore structure of cementitious materials. However, a precise report and discussion about the pore structure of cementitous materials is directly related to the parameters associated with the pertinent experiments such as mercury intrusion porosimetry. Therefore, the first part of this paper will review the existing methods of measuring the porosity of cementitious material. In particular, the contact angle measurement will be discussed in detail in which the most effective parameters and possible errors of calculation will be presented. Future work related to porosity measurement and CNT reinforcement of cement composites is suggested in each section in order to help the readers find the gap in this field of knowledge.
Porosity and characterization technique
Porosity has a direct effect on the strength of solid materials, especially concrete [25,26], however, the ways of measuring porosity have been stated to be difficult because of the hydration products that form a tortuous structure [23] and the results depend on drying methods prior to the porosimetry process [24]. A practical way of measuring porosity and studying the pore structure for nanoscale pore sizes is mercury intrusion porosimetry with which cumulative pore volume versus applied pressure is obtained. Due to the fact that, mercury does not wet most substances and cannot penetrate into pores by capillary action, pressure is required for mercury to intrude the pores. Mercury’s high surface tension as well as its non-wetting property leads us to choose this material for the intrusion porosimetry process. Maximum pressure which is needed for mercury intrusion is related to the smallest pore size, and pore diameters of 100 µm to 2.5 nm are more likely measured by MIP [27]. Consequently, MIP is applicable for describing the pore system in cement-based materials. Several research projects have been conducted on different properties of cement paste and concrete using the MIP technique including studying the interfacial transition zone [28], the effect of admixtures [29], cement paste backfill [30], and the effect of carbon nano tubes [20,31]. The test is also applicable for defining pore size distribution which then can be related to water permeability by the Katz-Thompson equation [32,33]. It has been proved that water permeability is not totally related to the total porosity of the pastes [34] as increasing in total porosity initially shows its effect only on large pores. The pore diameter can be calculated by a force balance equation called the Washburn equation [35]. This equation is based on an assumption that pores are cylindrical which can be a source of minor error in calculations. The equation helps convert pressure values to the equivalent pore radii [36]. The average pore diameter is calculated by Eq. (1) as follows:
where V is the volume of the mercury that is used per gram for each sample (cm3/g), and A is the pore surface area of the whole sample (cm2/g). This data is suitable to be used for comparing the porosity of samples with each other since because of ink-bottle effect, the average pore diameter that is calculated with the equation is smaller than that of the real average pore diameter [37]. This equation can be rewritten by Eq. (2) as follows:
where represents the pressure of liquid (kPa), represents the pressure of gas (kPa), s represents the surface tension of liquid (mN/m), q represents the contact angle of the intruded liquid (degree), and represents the pore diameter (µm). Since this method of examination is performed in a vacuum, at first the gas pressure is zero. Eq. (2) can be rewritten as Eq. (3):
When the pressure increases, the cumulative volume of the intruded mercury is increased. From this cumulative pore volume, the pressure and the median pore diameter can be obtained by finding a point where 50% of the total volume has been reached. Surface tension is defined as a contractive tendency of the liquid surface to resist an external force. This tendency has the dimension of force per unit length or energy per unit area. MIP is applicable for describing the pore system in cement-based materials, although MIP measurement does not contain total porosity values and it is limited to the pores with sizes 0.003 µm to 375 [38]. Pores with dimension of 2.5−10 nm are called gel pores or gel capillaries. Such pores are formed within amorphous C-S-H gels and are not associated with permeability. Mesopores with a 5−5000 nm radius are called capillary pores, which are interparticle spaces and are gradually filled with hydration products. Macropores, on the other hand, are mainly formed by entrained air and insufficient compaction, and are known as a pore system. There could be pores associated with shrinkage cracks. Gel pores play an important role in creep and shrinkage but not in maintaining the strength of specimens. Capillary pores are known to be responsible for reduction in elasticity and strength [39]. It is widely known that the increase in the degree of hydration leads to a decrease in capillary pore volume. Subsequent disconnection in the link between some of the capillary pores leads to a decrease in permeability. The amount of capillary pores is responsible for the percolation transition in cement paste, and it has been proved that if capillary pores are reduced to 80%, no percolation transition would occur disregarding the water to cement ratio. Furthermore, by capillary pores’ depercolation, water permeability is reduced to 10−14 m/s [40]. In another classification of the pores, pore sizes that are over 10,000 nm are discrete air voids and have a spherical shape. These pores improve the durability of concrete due to the freeze-thawing cycles as well as salt decay. Pore size ranging between 50 nm to 10,000 nm are large capillaries that have a major effect on transmitting ions and chemical particles but have a minor effect on hydration rates [33]. Pore sizes of 10 nm to 50 nm are medium capillaries which have a major effect on permeability and usually vary because of adding mineral admixtures. All such pore sizes can be distinguished by MIP. The data that is obtained through MIP is appropriate for the use in the Washburn equation, which considers the entire pores to have entire access to the outer surface and to be in a cylindrical shape. The pore size related to the steepest slope of the curve is called the threshold or critical pore size. The Washburn equation presented in Eq. (4) estimates the diameter of such pores that are intruded by mercury at each pressuring increment.
where d is the diameter of the cylinder that is intruded on (m), γ represents mercury surface tension ranges from 0.473 to 0.485 N/m, q is the contact angle of mercury on the solid ranging from 117° to 141°, and P is the pressure (N/m2). The pressure ranges between 0.04 for hundreds of micrometers and 400 MPa for a few nanometers of pore sizes.
Pros and cons
MIP has been used for decades for pore structure evaluation [41]. It has been stated that the thresholds that should be satisfied in MIP cannot be obtained in cement-based materials. Considering the thresholds, several research projects have been performed on MIP for cement-based materials [42–45]. It has been discussed that, the first intrusion data would not be reliable for demonstrating the pore structure. Regarding percolation theory, complex network models should be used for a pore structure analysis [46–48]. It has been stated that, some deficits are associated with MIP and exact pore size distribution cannot be obtained by the experiment [49,50]. However, by considering the thresholds related to the experiment, MIP can be implemented for cement-based materials [44], especially when comparison of the results matters [51] and/or obtaining pore structure relations is one of the objectives [44]. In the cases that comparisons of the pore structures are made, the deficits of MIP are not necessary to be mentioned as an obstacle. On the other hand, it has been discussed that the Washburn pore distribution model is not suitable for accurate modeling of cement pore size distribution since neither are all pores cylindrical, nor can the porous space be totally considered as a percolative network [52].
Some other issues associated with the MIP procedure are as follows: the first problem with the model is that it considers a continuous long shaped cylinder and multiple single pores as the same diameter, and also it is barely achievable in concrete and cement pastes [37]. Moreover, the intrusion of closed large internal pores is in doubt and pore structure in cement paste has been proved to have fractal character [53]. In fact, in mercury porosimetry, the problem that is not usually considered is that the major part of gel pores as well as closed pores will not be intruded. Further, the MIP procedure measures entry sizes but not the whole pore size, which is related to the ink bottle effect that leads to the overestimation of finer pores and under estimation of larger capillary pores. Directly or through larger pores, some pores are connected to the surface of the sample, and other pores are considered to be ink bottle pores. MIP considers the whole volume of the ink bottle pores as having the apparent diameter of the smaller pores [27]. Because of the difference in the contact angles between mercury and the solid surface during intrusion and extrusion, the ink bottle effect happens and it leads the extrusion curve to move to a coarser pore-structure [46]. This hysteresis in cement-based materials can be eliminated by choosing the intrusion and extrusion contact angle of 130° and 104°, respectively [54]. By multiplying the total volume of mercury that is intruded at the maximum pressure by the bulk density of the material, total porosity is obtained. During the intrusion, ink-bottle pores as well as continuous pores can be observed from the intrusion curve. This can be monitored through the amount of mercury that remains in both such pores after the extrusion period [55]. The volume of the mercury that remains in such pores is called ink-bottle porosity. By extracting the total porosity from the ink-bottle porosity, the effective porosity is obtained. Moreover, threshold pore diameter is calculated by dividing the highest rate of mercury intrusion by the change in diameter or pressure (dV/dD) [56]. Due to the fact that, in the MIP process, the specimen should be dried severely, the shrinkage cannot be neglected. It can be concluded that the pore volume that is measured in MIP is less than that of the real amount [57], and that none of the drying methods lead us to an accurate result since many dense regions may not be emptied of water. Consequently an underestimation of open porosity is obtained [58].
Contact angle
Because mercury does not wet most substances and cannot penetrate into pores by capillary action, pressure is required for mercury to intrude the pores. Mercury’s high surface tension as well as its non-wetting property leads us to choose this material for the intrusion porosimetry process. The intrusion and extrusion of mercury take place with different contact angles. The angle where a liquid interface meets a solid surface is called the contact angle. By this angle, the wettability of a solid surface by the liquid can be quantified. Low contact angle indicates the ability of the liquid to spread on the surface and vice versa. A contact angle of less than 90° means the liquid wets the surface, and accordingly the surface tension is lower than when the contact angle is higher than 90°. Another important role of the contact angle is to show the water imbibition of materials, especially the porous ones. In simple words, this angle is formed where the three phases of liquid, gas, and solid intersect. Such interactions that modify the contact angle are intermolecular forces, cohesion, and adhesion forces. Cohesive forces are the forces that are available between the molecules of the liquid that are similar, such as hydrogen bonds and van der Waal’s forces. Adhesive forces are the forces that emerge when dissimilar molecules, liquid, and solid materials interact. Small contact angle which is typically seen in mercury indicates stronger adhesive forces compared with cohesive forces, and accordingly, the molecules of mercury interact more with the solid surfaces. Through Young’s equation presented in Eq. (5), the shape of a liquid interface can be modeled, and the contact angle in this equation is considered a boundary condition.
where is the solid-vapor interfacial energy, is the solid-liquid interfacial energy, is the liquid-vapor interfacial energy or the surface tension, and is the equilibrium contact angle of wetting, known as the Young contact angle. Ideally, the angle between the apparent solid surface and the tangent to the liquid interface is called the apparent contact angle. However, the actual contact angle varies due to the roughness, if any, of the solid surface. Surface roughness by definition is considered as the deviation of a real surface from its ideal form. Presented in Eq. (7), Wenzel proposed an equation between wettability and roughness in 1936 [59]. If surface roughness is considered rather than considering an ideal surface only, the solid surface will chemically be noted as hydrophobic.
where and are the actual contact angle and apparent contact angle, and r is the roughness ratio which can be calculated using Eq. (7):
where Sdr is the ratio between the interfacial and projected areas, and can be calculated according to Eq. (8):
Several factors that cause the contact angle hysteresis are the chemical heterogeneity of the surface, topographical heterogeneity of the surface, the impurities of the solution that is been absorbed on the surface, and the alteration of the surface due to the solvent.
Contact angles are categorized into two categories: static angles and dynamic angles. Static angles are measurable when the liquid stands on the surface and the boundary is stable. When the boundary tends to move, dynamic contact angles, known as advancing and receding angles, are required. The dynamic contact angle maintains the maximum and the minimum value that the static contact angle can get. In fact, the difference between the two mentioned dynamic contact angles causes the contact angle hysteresis [64,65]. The measurement of the advancing and receding contact angle of mercury, the surface of cement-based material, and the surface of graphitic reinforced cement-based material all play a significant role in obtaining pore size distribution curves. Due to the hydrophobicity of carbon nanotubes, for instance, the contact angle of carbon-nanotube-reinforced cement paste may vary from that of plain cement paste. Comparing the distribution curves obtained by the same hypothetical contract angle of 140° or 130° for both materials in such cases, does not lead us to a correct comparison. There are two ways of measuring contact angle that can be performed: (i) optical tensiometry and (ii) force tensiometry. Static and dynamic contact angles can be measured by both of the methods [66]. However, the optical method is suitable when liquid-solid interactions on heterogeneous solid surfaces are measured. The same method should be applied for the cement-based materials.
The samples should have the same specification as what will be used for the MIP experiment; therefore, they should have small dimensions, like crushed chunks of 1−2 mm3 or 1−1.5 g and then be placed in an oven at 60°C for 24 h before testing [67]. After preparing chunks for testing, the specimens should be soaked in acetone for 24 h; by doing so the hydration process will be stopped [68]. Moreover, with water remaining in the finest pores because of incomplete drying, the intrusion of mercury will be limited; however, several techniques can be implemented for drying of the material: at the temperature ranges from 50°C to 105°C in an oven, by vacuuming, by freeze-drying, or with a solvent replacement method [69]. It is worth to know that the oven-drying method in the MIP procedure damages the material structure by removing the pore water. This leads to microcracking and collapsing of fine pores, so capillary porosity changes are likely to take place. Subsequently, freeze-drying technique is advised to be implemented [38]. Although time-consuming, freeze-drying has been considered more preferable method of drying which gives the “real picture” of the microstructure [70], however, solvent replacement drying has been suggested alternatively as a less time-consuming method [71]. If none of these procedures is available, the chunks should be placed in an oven at 60°C for 24 h to 48 h to be dried until their weight reach a constant value. By doing so, one can be sure that there will be no significant amount of evaporable water in the pores that would eliminate the intrusion of mercury [68].
The modification of the contact angle has been already discussed: 2% of errors for 1° difference in contact angle has been recorded which adds another issue to the mentioned problems associated with MIP [72]. While intruding mercury into the pores with a pressure over 70 MPa, fragile walls in the cement microstructure break down [73]. However, when the specific data of the porosimeter is not available, for the theoretical pore size calculation, it is recommended to use 480 mN/m as the surface tension [74] and 130° as the mercury-solid contact angle, although the variation of this value has a negligible effect on MIP results [75]. Also, the contact angle between the mercury and the pore surface (q) for cement-based materials is recommended to be assumed to be 130° [76] instead of 140° [75]. Besides, solvent dried from a 5 mm sawed or core-drilled sample should be prepared for the test [74].
The deficits of the Washburn equation of preparation procedure are minor problems that can be neglected [77]. Also, by using specimens with smaller dimensions, pore structure will become coarser, a reduction in the less accessible pores will attain, and it is estimated that there will be more valid pore size distribution [67]. MIP can be implemented to comparatively study the pore size distribution of cement-based materials, and although this technique might not be proved an exact reflection of the porosity, it provides pore size distribution in a wide pore range [52]. In other words, as long as the limitations are considered, MIP is an invaluable technique for comparative pore size distribution assessment [51]. Usually the measurement is performed in two stages: low pressure from 0 to 0.17 MPa and high pressure from 0.17 MPa to 205 MPa. Then the curves of mercury intrusion and extrusion hysteresis are obtained from each of the stages. All pores are supposed to be intruded by mercury at the end of the first stage. After removing the pressure, mercury is removed from almost all of the pores, except the ink bottle and dead end pores; thereby, the effective porosity, which is the volume of the mercury that is removed from the pores, can be measured [78]. By applying pressure for the second time, from the second intrusion curve, the pore size distribution of the effective porosity is obtained [79]. In the following, this review article covers the effect of CNT on the pore structure and microstructure of cement composite. Thus far detailed report of discussion about the pore structure of cementitous materials have been directly associated with the parameters of the pertinent experiments such as mercury intrusion porosimetry, all of which were discussed above comprehensively.
CNT
Multi-Walled Carbon Nano Tubes, called MWCNTs, are graphene cylinders that are arranged around a hollow core concentrically, whereas Single-Walled Carbon Nano Tubes, called SWCNTs, are single graphene cylinders. Ideally, CNTs are 100 times stiffer and 6 times lighter than steel, and their Young’s modulus and tensile strength are 1 TPa and 200 GPa, respectively [80,81]. The Young’s modulus of polystyrene was found to be increased about 23% when CNTs of 2.5% by vol. were added [82]. By adding 5% CNTs by cement weight to epoxy composites, the Young’s modulus and tensile strength are increased by about 55% and 17%, respectively. MWCNTs are used for self-sensing applications and health monitoring [83] and they are useful for electromagnetic shielding, too [84]. Carbon nano tubes are allotropes of carbon, have the same structure as graphite sheets, and can be considered as an additive for composite materials. They are the strongest nano fibers that have been made and they can provide electrical and thermal conductivity [85–87].
CNTs can improve stress transfer through the cement-based materials and can provide crack bridging. The micro pore structure of concrete is less than 2 nm in diameter [22], thus, concrete compressive strength is much higher than that of its tensile and flexural strength, which is attributed to its heterogeneous pore structure. Hydration does not occur with a same rate at the whole cement structure: hydration occurs less around aggregates. Therefore, the connection between aggregates and cement paste is a susceptible area for crack initiation and propagation, which affects flexural strength as well. Fiber reinforcement is one way to improve cement composite strength [88]. Another effective way is using carbon nanotubes (CNTs) as reinforcing materials. CNTs have high strengths [89], Young moduli of about 1 TPa [90], elastic behavior [91], and improved thermal properties. CNTs can be considered as filler, produce dense materials, and reduce the potential of cracking at early ages. The yield strength of SWCNTs has been reported as 63 GPa [92]. CNTs have unique electronic behavior depending on their chirality [93]. In general, CNT as a fiber can be dissolved by a solvent and produce the composite; however, this method is not applicable in applying CNTs in cement pastes because the solvents may have detrimental effects on the cement hydration products. Therefore, one main issue with using CNTs in the matrix materials is dispersion. Another main issue is the tendency of CNTs to adhere together after purification because of their van der Waals’ attractive forces. The first method of implementing CNTs in cement paste is using water reducing admixtures. Dispersion of multiwalled CNTs (MWCNTs) in polycarboxylate or polyacrylic acid solution [31,94] and dispersion of single walled CNTs (SWCNTs) in naphthalene sulphonates [84] are proved to be the most appropriate methods of functionalization. Another method of functionalization is sonication of cement paste and CNT in isopropanol [95]. However, the potential for damaging surfaces of cement paste grains through the sonication process is the fundamental disadvantages of implementing this method, because it slows down the hydration process. Some researchers have proposed the use of surfactants along with sonication, a method in which, through a liquid medium, a piezoelectric system delivers acoustic energy to nanotubes [96,97]. In this method, nanotubes networks are broken by sonic energy and tubes are dispersed into the matrix. Dispersed tubes are then mixed into the matrix with continued sonication. Having functionalized CNTs, they achieve sufficient bonding with the matrix elements. Otherwise without this method, CNT polymer composites are susceptible to fiber pullout due to low loads. This problem has been solved by developing CNT-ceramic composites [98].
There are many contradictory findings about the effectiveness and improvement of CNTs’ mechanical strength in cement pastes [96], which can be attributed to different water/cement weight ratios, hydration periods, nanotube contents, dispersion, and functionalization issues. Although macro test results are contradictory, micro investigations have revealed that if evenly dispersed, nanotubes can improve the properties of cement paste due to crack bridging and chemical bonding [31]. Through the fourier transform-infrared (FT-IR) spectroscopy, the bonding and crack bridging have been proved to be formed between carboxylated MWCNTs and cement paste [58].
According to an investigation on the formation of SWCNT and C-S-H, C-S-H gels are shown to be formed on the surface of nanotubes; however, less C-S-H gel can be observed in areas that are still unhydrated [95]. This property, called nucleation which is observable through the whole surface of SWCNTs. This phenomena has also been observed in MWCNTs [83]. Since less electrical connectivity is obtained, it can be deduced that C-S-H gels have disconnected the connection between nanotubes [99]. The bonding between CNTs and cement particles is the main issue when CNTs are added to the mixture. CNFs, on the other hand, provide more planes with exposed edges along the surface. This characteristic improves the chemical and physical interaction.
The main liability of the CNTs/CNFs, especially, CNTs for application in mortars, is the agglomeration of the particles due to both hydrophobicity and self-attraction of the particles. Some previous research projects have been performed to solve this issue. In one study, Ozone gas with 0.3% vol. in air was used as an agent for surface treatment. It has been reported that the compressive and tensile strength of the cement paste were enhanced [100]. In another study, surface treatment by H2SO4 and HNO3 solution forms carboxyl acid groups on CNTs’ surfaces leads to the improvement of reinforcement [101]. Makar et al. sonicated CNTs in isopropanol and added cement particles. The process led the cement particles to be coated with CNTs. They found that CNTs affect the hydration process and mechanical properties at early ages [84]. Cwirzen et al. implemented surfactant admixtures for CNTs and no significant mechanical improvement was achieved. They observed weak bonds between CNTs and cement particles [102]. In another study on the effect of the surface structure of CNTs, it was found that the compressive and flexural strengths were affected with the carboxyl group functionalized MWCNTs and were increased by non-functionalized or annealed MWCNTs [103]. In another study, MWCNTs were dispersed in acetone, and a superplasticizer and a viscosity modifier were added while using an ultrasonic vibration. Konsta-Gdoutos et al. found that with ultra-sonication and surfactant, the cement paste was reinforced with CNTs at the nano scale, the porosity was decreased, and the Young’s modulus was significantly increased [104]. For CNFs it was proposed that using silica fume [105] would cause a better dispersion.
Li et al. [31] modified MWCNTs with H2SO4 and HNO3 for application in cement-based materials. In order to treat carbon nano tubes, 100 grams of CNTs were added to 1000 mL of 3 by 1 sulfuric acid and nitric acid solution. This was performed to attach carboxylic acid to the CNT surfaces. After sonication for 3 h in a basin, the solution was mixed with distilled water with a ratio of 1︰5 by volume. After 24 h, the top of the solution was removed. This process was performed 4 times so that no residual acid was observed. The diameter and length of the CNTs were 10−30 nm and 0.5−50 µm which gave an aspect ratio of 100−1000. The water to the cement weight ratio was set to 0.45 and the specimens were examined at the age of 90 days in order to reduce the effect of unhydrated cement particles. To compare the results, untreated carbon fibers with diameters of 10−14 µm and lengths of 6 mm were examined with the same mix design that was used for CNTs. Methylcellulose with 0.4% by weight of cement along with 0.13 vol% defoamer were used for dispersing the fibers.
Functionalized MWCNTs could improve the strength of a cement matrix [31,94] because of the bond between CSH and –COOH groups available on the surface of carbon nanotubes. It was reported that annealed carbon nanotubes can cause higher compressive strength compared with functionalized carbon nanotubes with the carboxyl group [103,106]. An Increase in the compressive strength was reported after adding non-functionalized CNTs, as long as such nanoparticles were dispersed properly with other admixtures, such as fly ash [107]. CNTs were made by the chemical vapor deposition (CVD) method in the presence of nickel oxide catalyst. CNTs were first mixed with water and sonicated for 10 min, then added to the fly ash and cement particles [107]. The suitable amount of carbon nanotubes to be added to the mixture has been reported to be less than 1% of the cement weight [31,103,108]. CNTs were treated with a solution of H2SO4 and HNO3 and dispersed uniformly by ultrasonication. The electrical resistivity and pressure-sensitivity of the specimen reinforced with treated CNTs and untreated CNTs were measured. Treated CNTs were observed to be covered by C-S-H, resulting in a higher compressive strength. It was shown that adding CNTs, treated or untreated, could significantly decrease the amount of electrical resistivity and increase the compressive sensitivity. In fact, the untreated CNTs number of contact points is more than that of the treated CNTs, which results a higher electrical resistivity and lower mechanical properties [107]. Li et al. treated CNTs with H2SO4 and HNO3 solutions, and it was observed that in the cement matrix, they could play a significant role in arresting cracks, which is known as bridging effect. This effect improves flexural strength and failure strain [31].
Many researchers have measured the pore size distribution in cement composites by means of mercury intrusion porosimetry (MIP) [109–112]. Sanchez and Ince [105] reported that the pore volume within the range of 6−200 nm is increased when CNTs are used. Low amounts of CNFs reduced the durability because there were more CNF pockets. However, with higher amount of CNFs, due to the obstruction of the pores by ball shaped CNFs, the water penetration resistance was improved. The porosity was 10.8%, which was 64% lower than that of the plain cement paste, and a finer pore size distribution was observed. The amount of pores with an average diameter of more than 50 nm was 1.47%, i.e. 82% lower than that of the plain cement paste. This indicates that CNTs can act as filler for the cement composites; however, the CNF reinforced matrix’s total porosity was increased by 31% compared with the plain cement paste. In the CNF reinforced matrix, the amount of the pores with an average diameter of more than 50 nm was 9.89%, i.e., 2.7 times higher than that of the plain cement paste.
Nochaiya and Chaipanich investigated the effect of adding up to 1% of the cement weight MWCNTs on the porosity and microstructure of cement paste with a water to cement ratio of 0.5. They found that by adding CNTs that are dispersed evenly in the cement matrix, the porosity was changed and mesopores were reduced [20]. The maximum decrease was about 18% which is the difference between the plain cement paste and the specimen containing 1% wt. MWCNTs. DV/dlog(D), which indicates pore size distribution, showed no considerable difference in the pores considered micropores. The maximum reduction in the total porosity and total surface area were about 17% and 22%, respectively. CNTs were observed to be available mostly between C-S-H and CH phases [20]. Mapping of carbon in the SEM micrographs, on the other hand, showed an even dispersion of CNTs through the specimen. CNTs with a diameter less than 50 nm can fill the pores, especially mesopores, and accordingly cause a denser microstructure in the cement matrix. According to the FT-IR spectrum, with treatment by a strong oxidizing acid, groups such as carboxylic and hydroxyl were attached to the surface of CNTs properly. In fact, the improvement in the microstructure of the cement composite after adding CNTs can be due to the carboxylic acid groups, which accelerates the chemical reactions with calcium silicate hydrate [113]. Bonds of C=O of carboxylic acids were observed and the C-OH stretch mode were both identified with a large amount on the surface of CNTs. Moreover, through CNTs, the interaction causes covalent forces by which the load transfer between the cement composite was improved. No new phase was reported to be formed in the cement paste reinforced with carbon fibers [113].
Sanchez and Ince [105] measured the pore size distribution by nitrogen adsorption to consider a 6−200 nm diameter range. It is noteworthy that nitrogen adsorption is suitable for considering pore sizes in micro scale, and for mesoscale pore size distribution mercury intrusion porosimetry is required. CNFs were found to be in pockets; however, in the presence of silica fume, they could stick to the hydration products. Because of having less particle size compared to the cement particles, silica fume can disjoin the van der Waals’ interactions between the fibers. Moreover, because of the presence of sulfur on the surface of CNFs, which is due to the process of production, needle shaped ettringites were available. This indicates that by modifying CNFs’ properties, especially at the surface, the characteristics of the specimen would change. CNFs, as well as silica fume particles, were agglomerated in different sites; however, through SEM micrographs, CNFs were seen to act as bridges across the cracks and hydration products [105]. Because exposed edge planes are available along the surface of CNFs, these fibers can be considered as nucleation sites for Ca-Si phases. The presence of abnormal amounts of ettringite that have needle-like shapes could be due to sulfur that is available on the surface of CNFs. The way that the surfaces of CNFs are modified can alter the microstructure of the specimens. CNFs are hydrophobic particles and can resist the penetration of water [105].
Conclusions
In this study, the effect of CNT on the pore and microstructure of cement composite were reviewed. A thorough literature review and discussion about the pore structure of cementitous materials is related to the parameters deal with the experiments such as mercury intrusion porosimetry. The review indicates that surface treatment by H2SO4 and HNO3 solution forms carboxyl acid groups on CNTs’ surfaces that lead to the improvement of reinforcement. There is no new phase to be formed in the cement paste reinforced with carbon fibers. Compared with CNFs, CNTs show better performance with regards to the durability properties. The compressive sensitivity of the untreated CNTs composite is lower than that of the treated CNTs composite. On the one hand, the untreated CNTs composite has less electrical resistivity; on the other, the treated CNT cement composite has more pressure-sensitivity properties due to the coverage of treated CNTs by C-S-H gels. CNTs are available mostly between C-S-H and CH phases and less porosity can be obtained after the implementation of CNTs as they can act as filler as well as nucleation sites for hydration products. The accurate analysis of pore size distribution is an important factor to consider because of ink-bottle effect; the average pore diameter calculated according to the Washburn equation is smaller than that of real average pore diameter. The sample for MIP measurements should have small dimensions like crushed chunks. The reduction in the sample size leads to a reduction in the less accessible pores and accordingly more valid pore size distribution can be estimated. Freeze drying is the most preferred method of drying for MIP experiments, which gives the “real picture” of microstructure. Although MIP might not be the exact reflection of the real porosity, it provides pore size distribution in a wide pore range. In other words, as long as the limitations are considered, MIP is a valuable technique for comparative pore size and pore size distribution assessment.