Autogenous healing of hollow natural fiber (HNF)-reinforced reactive magnesium cement (RMC)

. This work investigated the autogenous healing of interface bond between hollow natural fiber (HNF) (e.g., sisal fiber) and reactive magnesium cement (RMC). Preloading -reloading test was conducted to single fiber embedded specimens, the results suggested that the interface bond can be improved by over 300% after wetting-air cycles conditioning, and hot water conditioning demonstrated greater healing efficiency compared to room temperature conditioning. Based on SEM-EDS characterization, this prominent restoration of interface bond is due to: 1) further hydration and carbonation of RMC, which partially fills the microcracks between fiber and matrix; 2) the migration and precipitation of expansive agents in lumens, which improves the contact between sisal fiber and matrix. Furthermore, the interface bond healing resulted in a significant recovery of the fiber-bridging strength of sisal fiber-reinforced RMC, the fiber-bridging strength doubled after healing compared with that of preloading.

length (single fiber pull-out test) and 15-mm-length (NFRC tensile test) short fibers. The RMC was acquired from Shanghai Yuanjiang Chemical Co., Ltd, it is composed of 95 wt.% of reactive MgO and minor impurities like CaO and SiO2; the sodium hexametaphosphate was acquired from Xilong Chemistry Ltd., which was added to modify the rheology of the fresh paste and improve the fiber dispersion.
To fabricate the single fiber specimen, the fresh RMC-based paste (water/MgO ratio of 0.6, sodium hexametaphosphate/MgO ratio of 0.02) was cast into a shallow mold with multiple 100-mm-length sisal fibers sticking out of the mold. After 24 h ambient curing, the specimen was demoulded and then cured in carbon chamber (30±1C, 85±5% relative humidity, 10% CO2 concentration) for 7 days. After which, it was cut into single fiber specimens with the thickness of 3±0.5 mm.
To prepare NFRC specimens, 4 vol.% sisal fibers (15 mm) were added into the same fresh paste used for single fiber specimen, after uniformly distribution of sisal fiber in fresh paste by mixing in a Hobart TM HL 400 mixer, the fresh composites were cast into prism molds (300  60  12 mm 3 ). After curing in ambient environment for 24 h and demoulding, the NFRC specimens were moved to carbon chamber (30±1C, 85±5% relative humidity, 10% CO2 concentration) for 7 days.

Single fiber pull out test
Single-fiber pull-out test was conducted in an electrical universal material testing machine (Lyllod, modle EZ50) equipped with a 20 N load cell and a displacementcontrolled actuator, as shown in Fig.1 a. The loading rate was 0.5 mm/min. In the test setup, the single-fiber pullout specimen was glued on a X-Y table that is for aiming the fiber to the clamp. At least eight specimens were tested for each group.
Monotonic loading and preloading-healing-reloading were applied to single fiber specimens. For monotonic loading test, the fiber was directly pulled out from MgO matrix, the load-displacement curves were recorded. For preloading-healing-reloading test, the fiber was first preloaded and stopped right away when a sudden load drop is detected, which stands for the full fiber-composites debonding [19]. Then the preloaded specimens underwent two different conditioning cycles before reloading test: 1) wetting-air cycles at room temperature (W(RT)-A); 2) wetting in 65C hot water and air at room temperature cycles (W(65C)-A), each W-A cycle consisted of 12 h in water followed by 12 h in ambient air.

Tensile test of notched NFRC
After carbonation, two notches were made at the middle part of NFRC, the width and depth of notch were 3 mm and 10 mm, respectively. Uniaxial tensile loading at the rate of 0.3 mm/min was applied to these specimens to induce single crack at the notched location. Two LVDTs were placed at a gage length of 23 mm to record the crack opening, as presented in Fig. 1b. The NFRC specimens were first preloaded until the crack width reached around 0.3 mm, then these specimens were dismounted and exposed to wetting-air cycles conditioning before reloading to failure. Fig. 1c shows the NFRC specimen after preloading, a single crack was successfully induced between two notches. Each group contains at least 4 specimens.

Microstructure characterization
The microstructure of sisal fiber, interface zone between fiber and composites and healing products in matrix crack were characterized by scanning electron mesh sand paper, respectively, then these specimens were immersed in isopropanol for 7 days to kill the hydration, followed by oven drying at 65C for 1day before examination.

Fig. 1
Setup of (a) single fiber pull out test and (b) tensile test, (c) is the optical image of NRCC specimen after preloading.

Fig. 2
shows the representative single fiber pullout P-u curve, it exhibits two characteristic stages [19]: the fiber debonding stage as P increases linearly and the fiber pullout stage where P gradually decreases to zero as u reaches the fiber embedment length (~3 mm). The sudden load drop from Pa to Pb indicates the fiber has debonded from surrounding matrix. With using the characteristic loads Pa and Pb, the chemical debonding energy Gd and frictional bond τ0 can be calculated by the following equations [19]: where the Ef, df, and Le are Young's modulus, fiber diameter, and fiber embedment length, respectively. The  The calculated interface properties from at least 4 presentative specimens (including monotonic loading and preloading-reloading tests) are summarized in Table. 1, it can be seen that the interface bond is almost double  Consecutive and hollow lumens with the diameter of about 6 m can be clearly observed, which can be pathway for CO2 and water diffusion during carbonation and healing [16]. After 20 W(RT)-A cycles conditioning as shown in Fig. 4b, the microcrack close to the surface is filled while deeper region is unhealed, this is agreement with the analysis of the recovery of chemical bond. The lumens are fully stuffed with needle-like healing products, even though the lumens in deep region were also filled, combining with the EDS mapping results, the healing products are likely to be nesquehonite [21]. It should be noted that the stuffed lumens demonstrate larger diameter (e.g., > 20 m) than the lumens before wetting-air conditioning. Natural fiber contains a great number of hydroxyl group and the side surface and cell wall are porous, this makes it hydrophilic. Therefore, the surrounding pore solution can quickly penetrate into hollow lumens when the sisal fiber contacts with fresh RMC, and the Mg 2+ would simultaneously migrate from pore solution to lumen.
Upon drying, the Mg(OH)2 precipitates in lumens and react with CO2 that diffuses from ambient atmosphere via hollow lumens to form nesquehonite. Due to the molar volume of nesquehonite (i.e., 74.8 cm 3 /mol) is prominently larger than brucite (i.e., 24.6 cm 3 /mol) [22], the formation of nesquehonite fills and enlargers the lumens, this not only effectively eliminates the lumen collapse-induced weak interface bond [10], but also remarkably promote the contact between sisal fiber and matrix and thus enhance the frictional bond and energy absorption during pulling out.

Conclusion
This work investigated the autogenous healing of HNF-RMC, including the interface bond healing between sisal fiber and matrix and tensile strength restoration of NFRC.
The single fiber pull-out tests demonstrated that the interface bond can be fully recovered and get stronger after preloading-healing, the frictional bond and energy absorption can continually grow to six times that of the control group with wetting-air cycles progresses. The increased temperature is able to expedite the recovery process, this is due to the higher temperature can promote the migration of Mg 2+ from pore solution to lumens of sisal fiber, followed by precipitation of brucite and carbonation to form expended nesquehonite, this effectively enhances the frictional bond and energy absorption. The tensile strength was improved over 100% after healing compared with preloading strength. The significantly increased interface bond and limited healing products in matrix crack confirms that the recovery of interface bond is the predominant contributor to the restoration of tensile strength.