A Study on the Drying Shrinkage and Mechanical Properties of Fiber Reinforced Cement Composites Using Cellulose Nanocrystals

The CNCs used in this experiment were ‘C’ company products from Canada, and their physical properties are shown in Table 2. The raw CNCs were in powder form, as shown in Fig. 1, and SEM (Scanning Electron Microscope) analysis determined they had a spherical shape and TEM (Transmission electron Microscope) analysis showed a rod shape for dispersed CNCs.

Cao et al. (2015) reported that, while cement is curing, the hydration product forms a shell around the hydrated cement particles to slow water diffusion into the membrane, i.e., into the cement core (Cao et al. 2015).

CNCs enhance the hydration of unhydrated cement cores and affect the hydration product C-S-H (calcium silicate hydrate), resulting in increased strength (Cao et al. 2016). In this experiment, isothermal conduction calorimetry was analyzed for 7 days for plain cement using general water and cement using aqueous solutions of CNCs with mixing ratios of 0.4, 0.8, or 1.2 vol.%. Table 3 shows the variables for the isothermal conduction calorimetry analysis.

The nanoparticles have a relatively large Van der Waals attractive force, which has a greater impact than other effects, resulting in a large aggregation of particles. To eliminate this aggregation, mechanical dispersion and nanoparticle surface treatment can be used to reduce van der Waals attraction or to prevent electrostatic repulsion or steric hindrance (Kim et al. 2016). Generally, mechanical dispersion methods include ultrasonic dispersion and high pressure dispersion. In this experiment, CNCs were dispersed using ultrasonic dispersion. High-pressure dispersion is considered to be an appropriate dispersion method because the variation in the particle size distribution is small and the average diameter according to the pressure is less than the ultrasonic dispersion (Lee et al. 2019). However, high-pressure dispersion requires a specialist company that has expensive dispersing equipments to produce samples, and it is not provided sufficient to carry out the experiment. Therefore, the experiment was conducted using ultrasonic dispersion which can be used as nanomaterial and can be dispersed at the laboratory level. On the surface of CNCs, numerous hydroxyl groups exist (Mehta and Monteiro 2013), and when the CNCs were added to water, they were observed to form a gel, as shown in Fig. 2, through hydrogen bonding. In this case, as the amount of energy required for dispersion increased, and the dispersion process was carried out for a long period of time, it was determined that physical properties caused by sample damage would not be exhibited. As the solvent agent, primary purification distilled water was used.

A rectangular mold (40 × 40 × 160 mm) with a hole at both ends was used, as shown in Fig. 3, to analyze the drying shrinkage characteristics of the cementitious composites. As shown in Fig. 4, the embedded gauge was placed at the center of the specimen so that it would not be disturbed by shrinkage. The gauge was fixed with wires, and a 1-mm-thick Teflon^® sheet was placed on the bottom and both ends of the mold to prevent the movement of the specimen from being restrained by the mold (Yoo et al. 2010).

The cementitious composites were then formed according to the various parameters and covered with polyester film to prevent water evaporation and absorption. The autogenous shrinkage strain was measured for 24 h after cementitious composite formation, and the specimens were then demolded. Drying shrinkage strain was measured for 60 days under constant temperature and humidity conditions by sealing all the exposed surfaces with aluminum tape to prevent specimen drying. Table 4 lists the mixing ratios of the test specimens. The reason for the difference in the amount of superplasticizer is due to the hydration promoting characteristic of CNC, and as the amount of CNC increases, the initial setting time due to the temperature rise due to the influence of initial hydration is faster than that of plain specimen. Therefore, as the amount of CNC increases, hydration reaction amount also increases, so the amount of superplasticizer is increased to secure the fluidity.

The fiber used in the experiment was flax fiber (cellulose type natural fiber) and steel fiber (generally used for fiber reinforcement), and its physical properties are shown in Table 5.

The compressive and flexural strengths were evaluated to determine the mechanical properties of cement composites with CNCs and fiber reinforcement. The test specimens were prepared with dimensions of 50 × 50 × 50 mm for compressive strength and tested in accordance with KS L 5105. For the flexural strength, 40 × 40 × 160 mm specimens were prepared and tested according to KS F 2408. The mixing proportions were the same as those used in the drying shrinkage experiment.

Since cement hydration is an exothermic reaction, heat flow and cumulative calorific value are directly related to hydration rate and hydration degree (DOH). Therefore, micro hydration analysis was performed to analyze the hydration characteristics of cement with CNC incorporation. Figure 5 shows the results of isothermal conduction calorimetry analysis conducted for 7 days on cement with various amounts of CNCs. The first heat flow peak occurred within 1 h for all variables. However, the first heat flow peak value of cement with CNCs was 50% or more of that for plain cement (without CNCs). These results suggest that the heat of dissolution of aluminate and sulfate solution is higher than that of OPC because of the incorporation of CNCs (Lin et al. 2012). In the case of plain cement without CNCs, the secondary heat flow peak appeared at approximately 9 h; this peak also appeared at approximately 9 h for C04, and at approximately 10 h for C08 and C12. In addition, the cumulative heat released increased in the specimens containing CNCs compared with plain cement.

After cementitious composite formation, the composites expanded for approximately 24 h; after that, the shrinkage strain increased sharply up to approximately 30 days, and thereafter, it tended to gently increase. In the case of the plain specimen, the shrinkage showed a rapid increase up to early on the 7th day. However, when the CNCs were mixed, the shrinkage strain tended to increase relatively gently over the first 30 days.

Figure 6a shows the result of analysis of drying shrinkage strain according to the volume fraction of CNCs. The final drying shrinkage strain of the plain specimen at 60 days was − 874.95 µm, and when CNCs were mixed, the drying shrinkage strain of C04 was − 666.43 µm, that of C08 was − 394.14 µm, and that of C12 was − 765.10 µm. Eventually, the drying shrinkage-reducing performance was improved by mixing CNCs. In particular, when 0.8% volume fraction of CNCs was added, the shrinkage-reducing performance was the best, showing a 55% reduction compared with the plain specimen.

Figure 6b, c show the results of analysis of shrinkage strain according to the type of fiber. The shrinkage strains of specimens mixing both CNCs and fibers did not differ significantly with the volume fraction of CNCs. However, the shrinkage-reducing performance of specimen series incorporating a 0.8% volume fraction of CNCs was the best. The final shrinkage strain of C08F1, with flax fibers, was − 460.57 µm, and that of C08S1, with steel fibers, was − 612.05 µm; the shrinkage strains were reduced by 47.4% and 30%, respectively, compared with the plain specimen. Eventually, the shrinkage-reducing performance of the specimen with flax fibers was superior to that with steel fibers.

Figure 6d reveals that, overall, the shrinkage-reducing performance of the specimens mixing both CNCs and flax fibers was the best, except for that of the C08 specimen.

The experimental results confirmed that the shrinkage-reducing performance is improved by incorporating CNCs. These results suggest that the evaporation amount of surplus water involved in hydration reaction is less than Plain due to the effect of hydration reaction promoting properties of CNCs. It can be confirmed that this experimental result is consistent with the results described in previous studies (Ardanuy et al. 2015) about the relationship between the hydration reaction and the strength development because the strength of the C08 specimen, with the highest shrinkage-reducing performance, is the highest. Also, as shown in the results of hydration heat measurement, in the case of Plain and C04 and C08, there is no difference in accumulated heat, but cumulative calorific value at C12 is more than three times higher than C04 and C08. As the CNC content increases, the hydration reaction is promoted and the amount of ettringite increases. As a result, the shrinkage rate is reduced because the expansibility is increased. However, the shrinkage ratio of C12 was rather increased. It is considered that the shrinkage reduction performance due to the incorporation of CNC is somewhat reduced because the dispersion of CNC in the cement is not properly performed due to the incorporation of a large amount of CNC (Table 6).

The results of compression testing are shown in Fig. 7 according to fiber type to evaluate the effect of fiber addition. In the specimens without CNCs, the flax fiber was not an effective compressive strength reinforcement for cement composites; rather, it reduced the compressive strength compared with the plain specimen, similar to results reported in a previous study (Kang et al. 2010). However, the steel fibers improved the compressive strength. These results were judged to be related to the fiber stiffness and density. When fibers are mixed in cement composites, the fibers replace parts of the cement composites, which means that the total density and stiffness decrease. However, steel fibers have higher density and stiffness than flax fibers, which leads to the increase in compressive strength compared with flax fiber specimens (Gwon et al. 2015).

The compressive strength properties according to CNC ratio are shown in Fig. 8 to evaluate the effect of CNC addition. The 0.8% CNC specimens showed outstanding performance. In the specimens without fibers, the 0.8% CNC specimen showed a strength improvement from 10 to 20% compared with other specimens. In the specimens with flax fibers, a similar trend was observed as in specimens without fibers. The addition of CNCs in specimens with steel fibers was not effective compared with the specimens without fibers or with flax fibers. The results suggest that 0.8% CNCs is the optimal ratio.

The comprehensive data for compression tests are shown in Fig. 9. We verified that 0.8% CNCs is the optimal ratio once more through this graph. Comparing the strength improvement ratio with or without CNCs, 0.4%, 0.8%, and 1.2% CNCs showed respective improvements of 0%, 23%, and − 8% in the specimens without fibers, 11%, 53%, and 23% in the specimens with flax fibers, and 10%, 12%, and 2% in the specimens with steel fibers. The addition of 0.4% and 1.2% CNCs led to a slight improvement in performance. However, the strength was reduced in specimens without fibers. The effect of CNCs was most pronounced in the specimens incorporating 0.8% CNCs, both without fibers and with flax fibers.

Flexural test results are as shown in Fig. 10 according to fiber type which is to evaluate the effect of fiber. The flax fiber was not an effective reinforcement material for cement composites. Rather, the flexural strength decreased compared with that of the plain specimen when flax fibers were mixed with 0.4% and 0.8% CNCs. This result means that flax fiber is not suitable to reinforce the strength of cement composites. The specimens with steel fibers indicated superior strength, regardless of the CNC ratio. The composites with steel fibers were more ductile than other specimens, except for the plain specimen (Mun et al. 2015).

The flexural results are shown in Fig. 11 according to CNC ratio. The 0.8% CNC specimens had the highest strength without reference to fiber type, with the strength decreasing in the order of 0.4% CNCs and 1.2% CNCs. The strength change according to CNC ratio was observed in the plain specimens. However, the specimens with flax fibers only indicated a slight strength change according to CNC ratio.

The comprehensive data of the flexural tests are shown in Fig. 12. The results suggest that 0.8% CNCs and steel fibers are most effective reinforcement for cement composites. The strength improvements of 0.4% CNCs and 0.8% CNCs were respectively 40% and 58% compared with the plain specimen without CNCs and without fibers. The specimen with 0.8% CNCs and steel fibers indicated a 70% strength increase compared with the plain specimen with steel fibers and a 200% increase compared with the plain specimen without fibers. The specimens using flax fibers had almost the same values, regardless of the CNC ratio.

We can therefore conclude that the optimal ratio of CNCs is a 0.8% volume fraction of cement through mechanical testing of cement composites using aqueous solutions of CNCs.