2.1 Microstructure
The microstructure development over time in LLST-1 is shown in Fig. 1a-c. Microporous hydrogel skeleton with smooth surface and spongy-like structure was formed in 10 min in LLST-1 (Fig. 1a). Unhydrated cement particles were observed to be more likely to be deposited on the hydrogel surface due to hydrogen bonding, ionic bonding, van der Waals forces26, 27. The rapidly construction of hydrogel matrix also provided steric hindrance to the free movement of cement particles, ensuring a uniform distribution of cement hydrates. With cement hydration progressing over time, the sponge-like porous structure can be maintained in LLST at 7 days, as shown in Fig. 1b and S. 1b, showing strong evidence on the template effect of hydrogel skeleton for cement hydration. Specifically, a large amount of water can be absorbed by hydrophilic groups in hydrogel skeleton via hydrogen force, which promoted the in situ dissolution of cement particles around the hydrogel skeleton. Additionally, based on the heterogeneous nucleation theory28, 29, the nucleation barrier during the formation process of cement hydrates can be reduced by the nucleation sites from hydrogel skeleton. Consequently, both of the water and nucleation sites provided by hydrogel skeleton facilitated the in situ cement hydration on the hydrogel skeleton.
After 28 days of cement hydration, as shown in Fig. 1c and S. 1c, it can be seen that a part of micropores in hydrogel skeleton were filled by uniform dispersed cement hydrates containing nanopores, leading to the construction of micro/nano-hierarchical porous structure. This was attributed to that more space can be provided for the better dispersion of calcium silicate hydrates (C-S-H) precursors in micropores, based on the two-step C-S-H nucleation theory30. In addition, it was observed that all the LLSTs showed sponge-like and hierarchical porous structures in Fig. 1d. Compared with the irregular shape of pores with more tips in Ref. (S. 1c), the sponge-like hierarchical porous structure in LLSTs helps to mitigate the stress concentration and realize uniform stress distribution31. As such, LLST exhibiting sponge-like hierarchical porous structure can be precisely developed through a rapid gelation of hydrogel as skeleton and subsequent deposition of cement hydrates as skin in order, which is expected to effectively mitigate stress concentration under mechanical load.
More specifically, the spatial distribution of micropores in 28-day LLST-1 was examined using X-CT (Fig. 2a), where different pore sizes were labeled with different colors. It is clear to see that 1–50 µm micropores dominated in LLST-1, which was in sharp contrast to the 0.5-2.0 mm macropores present in foam cement32. In addition, such obtained 3D uniform distribution of micropores in LLST-1 was expected to overcome the anisotropy mechanical properties of cement-based materials produced using ice-template method10. Nonetheless, a few pores larger than 50 µm were observed in LLST-1, attributing to the air bubbles introduced during the 10-min stirring, which suggested that the mixing parameters for hydrogels and cement mixture should be further optimized for a more precise control of micropores size in LLST.
The pore size distribution in LLST-(1–4) was measured using MIP (Fig. 2b&c) and NMR (S. 2a). Compared with Ref., a significantly higher volume of 5-100 nm nanopores was present in all LLSTs mainly due to the heterogeneous nucleation sites, uniform water distribution and prolonged supply in the hydrogel matrix. Besides, 30 ~ 49% reduction in total porosity was observed with the increasing content of cement particles in LLSTs, which provided an effective strategy for obtaining a controllable porous structure. As observed, this pioneer work using in situ self-assembly strategy is effective to develop a hierarchical porous structure containing micropores (1 ~ 50 µm) and nanopores (5 ~ 100 nm), to realize the precise control of pore size in cement-based materials.
2.2 Mechanical properties
As shown in Fig. 3a, the 28-day flexural strength of LLSTs-(1–4) were higher than that of Ref. by 30 ~ 220%. Moreover, the midspan deformation rate and fracture energy of LLSTs, two typical toughness parameters, were 190% ~ 780% and 130% ~ 1460% higher than that of Ref. (S. 3a). The improvement in flexural strength and toughness is related to the hierarchical porous structures in LLST, as discussed in section 2.1, which contributes to the higher fracture energy absorption, more homogeneous stress dispersion and better crack inhibition under mechanical load33, 34.
In addition, the specific compressive strength of LLSTs were 37% ~ 145% higher than that of Ref. and 220% ~ 1300% higher than that of conventional foam cement (Fig. 3b&c). This is because, compared with Ref. and foam cement, the sponge-like pores with hierarchical structure in LLSTs contribute to a more uniform stress dispersion and prevent the initiation and propagation of cracks during mechanical loading31, 35, 36. Therefore, a consolidation of lightweight, strong and tough cement-based materials was for the first time reported in this study, with a density of 0.75 g/cm3, specific compressive strength of 350 (kN/m2)/(kg/m3), flexural strength of 16.5 MPa, and fracture energy of 4865 KJ/m3.
As observed from the compressive strain-stress curves of Ref. and LLSTs (S. 3b&c), Ref. demonstrated a typical quasi-brittleness failure mode in which a sudden drop in stress appears after the yield point37. In contrast, LLSTs exhibited a plastic failure mode with three periods, including the elastic regime, stress plateau and high densification38 (Fig. 3d). The latter might be due to the deformation of the LLSTs pores, which consumed fracture energy during stress loading (Fig. 3e). More importantly, a strain-hardening phenomenon was also for the first time to be found in cement-based materials without fiber reinforcement, suggesting that LLSTs had a higher failure threshold and better structural integrity. Specifically, LLSTs are composed of hierarchical porous structure (micropores and nanopores) and hierarchical bond strengths (covalent bonds in cement hydrates and in hydrogel, ionic bonds and hydrogen bonds in cement hydrates/hydrogel skeleton, as illustrated in Fig. 3f), which enables the multi-scale stress dispersion/mitigation capacity subjected to mechanical loading39, 40, 41. In that case, the prior deformation of nanopores and preferential disruption of hydrogen bonds safeguard the integrity of micropores and the stability of covalent bonds, which is beneficial to raise the failure threshold.
2.3 Cement hydration characterization
FT-IR was carried out to reveal the interaction between the hydrogel and cement hydrates, as shown in Fig. 4. Firstly, the peaks at 3650 cm− 1, 3437 cm− 1 and 1650 cm− 1 in Ref., as displayed in Fig. 4a, were ascribed to the vibrations of the O-H, while the peak at 975 cm− 1 was related to the vibrations of Si-O. In comparison, FT-IR spectra of LLSTs exhibited additional peaks related to the hydrogel42 associated with the vibrations of O-H (3650 cm− 1, 3437 cm− 1), C = O (1650 cm− 1,1429 cm− 1) and N = O (1116 cm− 1, 876 cm− 1), as shown in Fig. 4b&c. Notably, the O-H peaks at 3650 cm− 1 and 3437 cm− 1 in LLSTs become wider than that in Ref., indicating that a stronger hydrogen bonding was generated in LLSTs due to the interaction between Ca and Al ions released from cement hydration with -COO− in hydrogels43, which is beneficial for the compressive enhancement due to the better interfacial compatibility.
The interactions between hydrogel and cement particles were further studied by ab initio metadynamics simulations. When the Ca ion came to the carboxyl group of hydrogels, the free energy landscape showed four local minimum (Fig. 4d). For state Ⅰ (Fig. 4g), the distance between the Ca ion and carboxyl group was ~ 6.6 Å, showing there was no interaction between these two particles. The Oc (O ions from the carboxyl group) ions were stabilized by the hydrogen bond network of the water molecules, while the Ca ion was in its hydration shell with six water ligands. After crossing a small free energy barrier of 6.1 kJ/mol, the hydrated Ca ion comes closer to the carboxyl group with the distance of ~ 4.6 Å (State Ⅱ, (Fig. 4g)). At this state, the hydrated Ca ion started to interact with the hydrogel through hydrogen bonding between its water ligands and Oc ions, making the system the most stable. Further proximity between Ca and carboxyl group with the distance of 3.1 Å (State Ⅲ, (Fig. 4g)) would make one of the water ligand of the Ca ion left its hydration shell and Ca-Oc distances be of 3.0 and 3.3 Å. The free energy of this state was approximately the same as the state B with only 2.2 kJ/mol higher than that. The state Ⅳ (Fig. 4g) with forming two Ca-Oc bonds of 2.38 and 2.69 Å was much less stable than states Ⅱ and Ⅲ, indicating that the Ca ion was inclined to form weak interactions instead of strong ionic bonding with hydrogel. This result was further confirmed by another simulation that incorporated the CV of dynamic exchange of water ligands of the Ca ion beside that of the distance between the Ca ion and Oc ions. As shown in Fig. 4e, the system was more stable when the Ca coordinates with five water ligands, and the state Ⅲ was also more stable than the state Ⅳ. The coordination environment of the Ca ion during the process of the interaction was also studied. It was shown that two peaks at six and seven coordination number arise, and the peak intensity at six was stronger. This means that the Ca ion was more likely to coordinate six water ligands when far away from the carboxyl group and five water ligands and only one Oc ion (including the situation of coordinating one Oc ion directly or sum of the two Oc ion) when approaching carboxyl group.
As the states Ⅱ, Ⅲ and Ⅳ contributed the most to the interaction between the Ca ion and hydrogel, the type of their interaction was investigated by the IRI analysis as shown in Figs. 4 (h)-(i), which was a novel real space function that exhibit weak interaction and chemical bond regions equally. Because the interaction in state Ⅱ was obviously due to the hydrogen bonding between the water ligands of the Ca ion and COOH group, here, the IRI analyses were only applied for the states Ⅲ and Ⅳ. In state Ⅲ, the regions between the Ca ion and Oc ions and that between two Oc ions showed the green part of the IRI isosurfaces, implying the attractive dispersion effect in Figs. 4 (h) and (i). In state Ⅳ, there was a blue region between the Ca and one of the Oc ions showing strong electrostatic attractive interaction, and orange region between the Ca ion and center of COOH group, presenting steric hindrance effect. This results were in line with our experimental results and infer based on DLVO theory that the observed adsorption of cement particles onto the hydrogel network after gelation was attributed to the van der Waals forces and steric hindrance effects between the cement particles and organic molecule chains.
Furthermore, the XRD patterns in Fig. 5a and S. 7a showed similar phase compositions in Ref. and LLSTs, including calcium hydroxide (CH), C-S-H gels, monosulfoaluminate (AFm) and ettringite (AFt), indicating that the types of cement hydrates were independent on the presence of organic hydrogels. The addition of hydrogel, however, led to an increase in the peak intensity of AFm and AFt, which might be related to the diffraction preference orientation due to the large crystal size of AFm and AFt. SEM images revealed needle-like and hexagonal plate-like cement hydrates (CH and AFm) with large crystal sizes in LLST (Fig. 5b&c), resulting from that the sponge-like porous structure offered sufficient space for the growth of these cement hydrates. The needle-like cement hydrates were expected to play similar roles as fiber in preventing the propagation of cracks and contributing to enhanced flexural strength in cement paste44.
SEM images showed that the nucleation/growth sites of cement hydrates in LLST differed from Ref. For Ref., cement hydrates tended to grow on the surface of unhydrated cement particles30. For LLST, the cement hydrates mainly grew on hydrogel skeleton (Fig. 5c and S. 7b). The growth pattern of cement hydrates in LLST can be explained by the role of polymer chains as hydration nucleation sites (due to the presence of -COO− groups) and the unique role of hydrogel for water distribution and supply, contributing to uniform dispersion of cement hydrates integrated into the skeleton of hydrogels and the homogeneous pore structure in LLST. In summary, the dramatic change in crystal size and growth position of cement hydrates played a joint role in improving the mechanical strength of LLSTs.
Evolution of hydration heat was also impacted by the unique water supply mode in LLST. Compared with Ref., LLSTs showed 10–20 hours right shift in the occurrence of the acceleration period suggesting delayed cement hydration (Fig. 5e). This may be because cement particles in LLSTs have restricted access to free water at the initial stage due to the rapid absorption of water by hydrogel (Poly(NIPAm-SA) was observed to absorb water 150 times of its weight45) inhibiting the early-age hydration. Nonetheless, these absorbed water can be slowly released for cement hydration over a prolonged time owing to the gradients in ionic concentration, the capillary force and the consumption by hydration reaction. Such unique water supply mode in LLSTs was believed to positively control the cement hydration reaction and reduce the risk of early-age thermal cracking in mass concrete (dams, nuclear power plants)46.
Specifically, hydration degree in LLSTs reached 79–87% (7 days) and 80–92% (28-day), which was 5–20% higher than that in Ref. (Fig. 5f). This observation was attributed to the unique role of hydrogel in LLSTs, serving as a uniform & controlled water supply source, as discussed in earlier sections. Normally, in conventional cement-based materials, the direct mixing method often leads to a portion of water being unable to participate in cement hydration, due to water evaporation and capillary adsorption between cement particles 47. However, in the current study, water evaporation and capillary adsorption can be fixed through the water sustained-release mode driven by hydrogel water desorption, which helps to improve the utilization rate of raw materials producing for cement-based materials. In conclusion, the revolutionized water supply mode of cement hydration, changing from directly mixing to water sustained-release by hydrogel, effectively addressed the challenge of rapid reaction in the early stage and insufficient reaction in the later stage.