Crack Healing in Cementitious Mortars Using Enzyme-Induced Carbonate Precipitation: Quantification Based on Fracture Response

A non-microbial means of carbonate precipitation to heal cracks in concrete is evaluated in this paper. Enzyme induced carbonate precipitation (EICP) that relies on plant-derived urease enzyme is used to catalyze the reaction between calcium chloride and urea to precipitate calcium carbonate. The faster rate of carbonate precipitation and the absence of microbes makes this method attractive to surface applications for crack healing of concrete. Notched mortar beams, where pre-cracking is induced, are subjected to EICP solutions with different CaCl2 concentrations and a fixed molar ratio of urea-to-CaCl2 of 1.2. X-ray diffraction and thermal analysis of samples collected from the vicinity of the notch clearly demonstrate the presence of calcium carbonate in the cracks. A flexural strength enhancement of about 33% is observed for mortars treated with an EICP solution made using 0.5 M CaCl2 as compared to the mortars that are just moist cured, while the fracture toughness doubled. The strength and fracture parameters (fracture toughness, and critical crack tip opening displacement) scale well with the carbonate content. Digital image correlation (DIC) is used to quantify the reduction in crack extension after the beams are treated with the EICP solution.


INTRODUCTION
Cracking is among the most important factors that compromise the durability of concrete structures. The ingress of deleterious agents through cracks result in corrosion of reinforcing steel, and other chemically induced durability problems, which further accelerate cracking, and exacerbate deterioration. Several strategies are used to obtain self-sealing and self-healing of cracks in concrete, including the use of epoxybased materials, acrylic resins, and silicone-based polymers [1][2][3]. In addition to crack sealing, methods of crack healing also have been investigated in great detail. In a humid environment, concrete is known to exhibit autogenous healing through the hydration of unhydrated cement, even though this process is very slow and applicable only to very narrow cracks [4]. When water and CO2 are available inside the crack, healing could occur through the carbonation of leached calcium hydroxide [5][6][7][8]. Other possible self-healing mechanisms have also been described [9][10][11][12][13]. A microencapsulation approach that incorporates a healing agent into a capsule that ruptures when confronted by a crack, and polymerizes to seal the crack, has been well reported for both concretes and other composites [14,15]. Silica shell microcapsules containing functionalized silica nano-particles have also been studied to obtain smart selfhealing concrete [16]. While microcapsules are frequently applied to obtained crack healing in polymers and composites, for concrete, most often larger capsules are embedded to release the healing agent.
Bio-mediated self-healing of concrete has been extensively investigated in the recent past. In general, a bacteria-based self-healing system consists of alkali resistant spore-forming bacteria and a substrate that is added to concrete during the mixing process. The dormant spores could be activated by water and oxygen in the event of a crack, and the organic substrate (i.e., urea) is metabolized by the active bacteria [17,18]. The basic premise behind microbially induced carbonate precipitation (MICP) is that the enzymatic hydrolysis of urea catalyzed by the microbially produced urease enzyme, in the presence of calcium ions, results in the precipitation of calcium carbonate [19][20][21][22]. The precipitated carbonates seal the cracks and enhance concrete properties. The bacterial spores and organic compounds are typically encapsulated (in polymeric shells, silica shells, lightweight aggregates etc.) in order to immobilize the bacteria and protect them before being incorporated into the concrete matrix [23,24]. MICP has also been established to be advantageous for several other engineering applications including cementation of soil to protect against liquefaction [25,26], and fugitive dust control [27]. Several studies have evaluated the microstructure of healed concrete and expounded the potential improvements in durability of concrete through MICP [17,24,[28][29][30][31][32][33]. Non-ureolytic bacteria combined with an organic calcium source as a twocomponent healing system for concrete has also been proposed [17,34].
While MICP relies on microbes to generate the urease enzyme, this study directly employs a plant-derived urease enzyme to catalyze the necessary reactions that induce carbonate precipitation in cracks in mortars, thus eliminating the need to use bacteria. Enzyme induced carbonate precipitation (EICP) is reported to have several advantages over MICP such as: (i) improved efficiency since it does not consume or compete for urea, (ii) water soluble free enzyme allows for easier in-situ application for crack control, (iii) smaller size of the solubilized urease enzyme than the microbially generated one, enabling penetration into smaller cracks, and (iv) short life of urease activity that enables faster production of carbonates and enzyme degradation that eliminates long-term impacts to concrete [35]. The rapid carbonate precipitation induced by the EICP process makes it well-suited for surface treatment of cracks in concrete, and thus attractive for infrastructural management agencies. The influence of different EICP solutions and treatment durations on the crack healing of mortars is investigated in this work. The EICP solutions if developed appropriately, can be encapsulated and used as self-healing material in concrete. Notched mortar beams are subjected to pre-cracking before subjecting them to EICP. The fracture toughness and strain energy release rates of the healed beams are evaluated to understand and quantify the efficiency of EICP in crack-healing of concrete structures. Digital image correlation (DIC) is employed to determine crack propagation rates in these beams.

Materials and Mixtures
The cementitious mortars used in this study were prepared using a commercially available ordinary portland cement (OPC) conforming to ASTM C 150, and quartz sand. A mass-based water-to-cement ratio (w/c) of 0.40 (corresponding to a volumetric w/c of 1.26), and a sand volume fraction of 50% was used.
The quartz sand had a mean particle size of 0.6 mm. The mortar mixtures were cast in 330 mm x 76 mm x 25 mm beam molds. A 19 mm notch was cut using a diamond saw at the center span of the beams (corresponding to a notch depth-to-beam depth ratio of 0.25). All the specimens were cured in moist conditions (RH > 98%) at a temperature of 23 ± 2 o C for a period of 28 days.

EICP Solution Formulation
The EICP solution used in this study was formulated based on [35]. The molar ratio of urea to CaCl2 was maintained at 1.  Table 1.

Strength and Fracture Response Testing
The flexural strength of the notched beams was determined using standard center-point loading as per ASTM C293/293M-10, after 28 days of moist curing. Based on the obtained flexural strengths, a total of 24 beams were loaded in three-point bending to a CMOD of 0.032 mm (approximately 95% of the peak load) in a closed-loop testing machine with the crack-mouth opening displacement (CMOD), measured using a clip gauge, acting as the feedback signal. This process was intended to create a controlled single crack above the notch in these beams. All the beams had similar crack tip opening displacements (CTOD) and crack lengths above the notch, which were around 20 µm and 12 mm respectively. After this precracking procedure, the beams were subjected to the following treatment procedure: (i) six beams were returned to the moist-curing chamber, and (iii) three sets of six beams each were subjected to EICP solutions. Two sets of beams were subjected to the LC solution for 7 days and 14 days respectively, and the remaining set subjected to the HC solution for 7 days, as shown in Table 1. The sides of the notch were sealed, and the notches were filled with the respective EICP solutions. Since the EICP process is a fast reaction, stock solutions were mixed immediately prior to treating the beams. The beams were treated every day for the desired duration to enable continuous carbonate precipitation. The experimental procedure is schematically described in Figure 1.
After the respective treatment durations, the beams were again subjected to the three-point bending tests under CMOD control (Figure 2(a)). The tests were terminated at a CMOD of 0.20 mm. Both monotonic (one loading and unloading cycle), and cyclic (multiple loading and unloading cycles) tests were carried out. Digital image correlation (DIC) was used to determine the crack propagation features in the monotonic testing regime. The cyclic regime was used to determine the fracture parameters (e.g., fracture toughness) and compliance-based fracture resistance curves (R-curves). DIC is a non-contact optical method that provides full-field surface displacements of the sample using digital images taken at different time intervals during the mechanical testing. The beam surface was speckled with random patterns to provide adequate contrast for image correlation (See Figure 2(a)). A Charge-Coupled Device (CCD) camera recorded images every 5 seconds during the monotonic testing regime. A rectangular area of approximately 120 mm x 60 mm above the notch was imaged. After termination of the test, the digital images were post-processed. The correlation between the subset of images corresponding to different locations in the load-CMOD plot was determined in order to calculate the displacement fields ( Figure 2(b)) using VIC-2D software TM (Correlated Solutions). The requisite mathematical formulations can be found in [36,37]. The horizontal (u) and vertical (v) displacement fields were obtained through the minimization of the correlation coefficient [38].

Identification and Quantification of Carbonate Precipitation
X-Ray Diffraction (XRD) and thermal analysis were used to ascertain the enhancement in the amount of carbonates precipitated in the crack through the EICP treatments. Small chunks were removed from the vicinity of the notch in the beams which were exposed to EICP treatment. These chunks were powdered to prepare them for XRD and thermal analysis.
XRD was carried out using a Siemens D-5000 diffractometer in a - configuration using Cu-K radiation (=1.54 Å). The samples were scanned on a rotating stage between 10 o and 70 o (2) in a continuous mode with a step scan of 0.02 o and step rate of 0.5/second. High-Score Xpert Analysis software was used for phase identification. Thermal analysis was carried out using a simultaneous thermal analyzer (Perkin Elmer STA 6000) in a nitrogen environment, at a gas flow rate of 20 ml/s. The samples were heated from ambient temperature to 995°C at a heating rate of 15°C/min. At least three replicate samples were used for thermal analysis.  the hydrated products, aided by the treatment solution that initially has a lower pH than the concrete pore solution. The calcium ions that are dissociated from calcium hydroxide, as well as those from calcium chloride which is a part of the treatment solution, reacts with the carbonate ions to form calcium carbonate. This is evident from the carbonate peaks in the DTG curves. For the control (untreated) sample, only one dominant peak corresponding to calcium carbonate is observed, around 720 o C, while an additional, more pronounced carbonate dissociation peak around 800 o C is observed for the treated samples. The thermal dissociation of calcium carbonate is influenced by the grain size [40]; the carbonate species resulting from atmospheric carbonation (generally of calcium hydroxide in the cement pastes)

Quantification of Carbonate Precipitation
dissociates much before the carbonates precipitated from the EICP reaction, which explains the presence of multiple carbonate peaks in the DTG curves.
The peak corresponding to carbonate decomposition was used in this study to quantify the amount of carbonates precipitated through the EICP treatment. Figure 4(b) shows the amount of calcium carbonate present in the vicinity of the notch in the control and treated pastes. Some of this carbonate can be attributed to atmospheric carbonation. A significant amount of calcium carbonate is precipitated in the treated mortars, resulting in crack healing, which has the potential to improve the mechanical properties, as will be discussed in the forthcoming sections. The LC-14 sample shows the highest amount of calcium carbonate, which is in line with the XRD spectra shown in Figure 3. Based on carbonate quantification as a function of EICP solution composition and treatment duration, it seems that a longer duration of treatment using a lower concentration solution is a beneficial option. The use of a higher concentration solution is found to result in slightly more carbonate precipitation over 7 days than the lower concentration solution, but the difference is not very significant. This is likely because the efficiency of urease is inhibited by high calcium chloride concentrations [35]. A magnified view of the load-CMOD response of the beams in the region of approximately 25% to 50% of the peak load is shown in the inset of Figure 5(a). Because of the presence of the pre-existing crack, and the absence of any crack-healing, the control specimen shows a lower initial load, while the treated specimens show higher loads because of the carbonate precipitation in the cracks. However, around a CMOD of 5 to 6 µm, there is a sudden drop in the load of the treated specimens, after which load again increases. The drop can be attributed to the breaking/failure of the calcite in the cracks, after which the specimen begins to take more load. Note that the drop is much larger for the HC-7 specimen, which is likely the result of differences in calcite morphology, as discussed in a later section. At a CMOD of around 10 µm, the loads in the control and HC-7 specimens become almost identical and this trend is followed for the entire load-CMOD response.

Fracture parameters (KIC, CTODc, and GR)
The fracture parameters considered in this study for the control and treated mortars are the critical stress intensity factor or the fracture toughness (KIC), critical crack tip opening displacement (CTODc), and the strain energy release rate (GR). KIC and CTODc were evaluated using the two-parameter fracture model (TPFM). TPFM uses loading and unloading compliances (from one loading and unloading cycle), peak load, specimen and notch geometry, and a geometry correction factor to determine the values of KIC and CTODc [41,42]. Multiple loading-unloading cycles were used to determine GR. The non-linear response in the prepeak region in a notched beam (notch depth a0) is accounted for through the consideration of an effective elastic material containing a crack length of aeff (aeff > a0). Representative load-CMOD responses under several loading-unloading cycles, used to determine the TPFM parameters and GR are shown in Figure 6.
The pre-peak slopes are higher for the treated samples, indicating a higher elastic modulus, attributable to crack healing. The LC-14 sample shows the highest peak strength and elastic modulus because of the higher carbonate content. It is also important to note that the post-peak drop is more significant for the treated samples, plausibly indicating debonding of carbonate crystals from the mortar and/or breakage of the carbonate crystals.  Figure 7(a). Carbonate precipitation fills the cracks, thereby not only enhancing the strength of the mortar, but also making the crack path more tortuous. Moreover, calcium carbonate is softer than the cement hydration products, aiding in energy dissipation. Both these aspects contribute to the enhancement of KIC of the treated mortars. Energy is also dissipated in breaking the bonds between the precipitates and the mortar, which also has the potential to increase the fracture toughness.
Treatment with EICP solution also likely results in the precipitation of carbonates in small pores adjacent to the notch, thereby resulting in further reaction product densification and pore size refinement, which also could contribute to improving the strength and toughness [42,43]. The enhancement in toughness scales with the carbonate content as shown in Figure 7(c).
CTODc represents the crack tip opening beyond which unstable crack propagation begins. The carbonate precipitation in the cracks increases the CTODc values, and thus delays the start of unstable crack propagation in the treated mortars. The trend, once again, relates well to the carbonate content as shown in Figure 7(c). In fact, the beneficial increase in CTODc due to EICP is larger than the corresponding increase for KIC, which is indicative of the energy dissipation provided by the carbonates precipitated in and around the crack. Resistance curves (R-curves) employ multiple loading-unloading cycles ( Figure 6) from the closed-loop CMOD controlled testing of notched beams to better elucidate the crack growth process in cementitious materials [44,45]. R-curves are envelopes of strain energy release rates and they represent the strain energy release rate (GR) as a function of crack extension. The fact that stable crack propagation increases the compliance of the specimen is used to extract the R-curves. GR is determined as shown in Equation 1: The first term is the elastic component and the second, the inelastic component of strain energy. Here, P is the applied load, C is the unloading compliance, t is the thickness of the specimen and a is the crack length. CMODinelastic is the residual CMOD when the sample is unloaded, which can be obtained from Figure   6. The elastic component is calculated using the rate of change of the unloading compliance with crack length, and corresponds to the energy release rate due to incremental crack growth. The inelastic component considers the rate of change of inelastic CMOD with crack length, and corresponds to permanent deformation caused due to crack opening. Both the rate components in Equation 1 are obtained by differentiating polynomial curve fits of compliance-crack length and inelastic CMOD-crack length relationships. Effective crack length (a0+a) is obtained by solving a non-linear equation that considers the beam span and geometry, unloading compliance, and the elastic modulus [46]. Crack extension (a) is extracted from each of the unloading compliances from the cyclic load -CMOD curves in Figure 6. Figure 8 shows the crack growth resistance curves for the control and treated mortars. In general, R-curves of cement-based materials comprise of a rising portion (indicating the formation of fracture process zone, the region of non-linearity at the tip of the crack) and a plateau region (indicating steady-state crack growth). It can be seen that the strain energy release rates of all the mortars are higher than that of the control mortar in the steady-state crack growth region. Thus it can be safely stated that the treated mortars require a higher strain energy for a given crack extension than the control mortar in this regime, because of the presence of energy dissipating mechanisms as discussed earlier. These trends are consistent with those of KIC and CTODc. However, in the low crack extension stage (i.e., in the pre-peak region), the HC-7 mortar shows lower GR than all the other mortars. A previous work on EICP in test tubes showed that lower concentration solutions are capable of producing calcite crystals with a more uniform size distribution. This is attributed to the slower mineral precipitation [35]. With higher concentration solutions, the precipitation is rapid and non-uniform crystals, visually appearing as agglomerations of small crystals, are formed. Note that the calcite contents are rather invariant between LC-7 and HC-7 samples as noticed from Figure 5(b). It is plausible that the presence of non-uniform, agglomerated crystals in the crack (even though the carbonate volume fraction remains roughly the same) dissociate quickly in the event of a crack driving force, forcing large crack extensions at lower strain energies. It was shown in Figure 5(a) that the drop in load for the HC-7 mortar was higher than those in LC-7 and LC-14 mortars, substantiating this deduction. However, the volume of calcite filling the cracks is roughly the same irrespective of the solution concentration (i.e., between LC-7 and HC-7 in Figure 4(b)); thus after dissociation of the grains, the energy balance becomes favorable for steady-state crack growth in this case also. The bulk fracture parameters KIC and CTODc are not able to capture this behavior since they are averaged global quantities. Thus the R-curves provide a more refined understanding of the micro-scale crack propagation response, and help explain the material behavior better. This also helps to choose the appropriate treatment methodology for crack healing. Note that at this CMOD, the treated sample also carries a higher load ( Figure 5(a)); yet the crack extension is lower. These results reinforce the results reported earlier in this paper and attests to the efficiency of EICP-based crack healing method.

(c) (d)
In order to shed more light into the relative healing efficiency of different treated samples as compared to the untreated one, rate of crack extension with respect to change in CMOD (a/CMOD) is plotted at as a function of CMOD in Figure 10. The a values corresponding to all the CMOD values were determined from DIC as explained earlier. The crack extension rate is shown only for CMOD values beyond about 60% to 70% of the peak load since before this load, crack extension is rather minimal. Due to the existence of an unhealed crack, a significant amount of energy is utilized in extending the crack in the control sample at relatively lower loads as can be noticed from Figure 10. The EICP-treated samples exhibit significantly lower crack extension rate, especially at loads before and close to the peak (CMOD ≤ 0.04 mm) as compared to the control mortar, attributable to carbonate precipitation that heals the crack and thus provides higher strain energy release rates ( Figure 8). The crack extension rates are the lowest for the LC-

SUMMARY AND CONCLUSIONS
The efficiency of enzyme induced carbonate precipitation (EICP) as a crack healing methodology for concrete was evaluated in this study. Plant-derived urease enzyme was used to catalyze the reaction between calcium chloride and urea to precipitate calcium carbonate in the cracks. This non-microbial method of carbonate precipitation has several advantages over microbial induced carbonate precipitation (MICP). Two different CaCl2 concentrations at the same molar ratio of urea-to-CaCl2 (of 1.2) were used for two different treatment durations in this study. Notched mortar beams were pre-cracked using a crack mouth opening displacement (CMOD) controlled test procedure, and then subjected to either moist curing or the EICP treatment.
XRD and thermal analysis studies conclusively demonstrated the presence of calcium carbonate in the cracks of the mortar beams after EICP treatment. The carbonate contents trended as: LC-7 < HC-7 < LC-14, which also matched well with the trends in the flexural strengths of the treated samples. Detailed fracture studies were carried out on the mortars to elucidate the effectiveness of the treatment methods.
Both the parameters of the two-parameter fracture model (TPFM), viz., fracture toughness (KIC), and critical crack mouth opening displacement (CTODc) were found to increase linearly with increase in carbonate content. The treated mortars were found to require a higher strain energy for a given crack extension than the control mortar, because of the presence of energy dissipating mechanisms introduced in the cracks by the precipitated calcite crystals. Digital image correlation clearly showed that the crack extension at or close to the peak load, as well as the rate of crack extension in the pre-peak region, were significantly reduced through EICP. The experimental studies conclusively showed the benefits of EICP in healing concrete cracks. The EICP method, where the calcite precipitation is rather quick, can be applied as a surface treatment strategy to heal cracks in concrete infrastructural elements.

ACKNOWLEDGEMENTS
The authors gratefully acknowledge support for this study from the U.S. National Science Foundation