Performance Degradation and Chloride Ion Migration Behavior of Repaired Bonding Interfaces inSeawater-Freeze-Thaw Environment

Abstract

The bond interface is the weakest part of the repair system, and its performance is a key factor impacting the repair effectiveness of damaged concrete constructions. However, the research on the damage law and the mechanism of repair of the bonded interface in the cold region marine environment is not in-depth. In this study, the influence of polyvinyl alcohol (PVA) fibers and crystalline admixtures (CAs) on the mechanical properties and volumetric deformation performance of cementitious repair materials was researched. Furthermore, the deterioration patterns of the bond strength and chloride ion diffusion characteristics of the repair interface under the coupling of seawater-freeze-thaw cycles were investigated. Combined with the composition, micro-morphology, and micro-hardness of hydration products before and after erosion, the damage mechanism of the repaired bonding interface was revealed. The results indicate that the synergistic use of PVA fibers and CAs can significantly improve the compressive strength, bond strength and volume stability of the repair materials. The compressive strength and 40° shear strength of S0.6CA at 28 d were 101.7 MPa and 45.95 MPa, respectively. Under the seawater-freeze-thaw cycle action, the relationship between the contents of free and bound chloride ions in the bonded interface can be better fitted by the Langmuir equation. The deterioration process of the bonding interface and the penetration rate of chloride ions can be effectively delayed by PVA fiber and CAs. After 700 seawater-freeze-thaw cycles, the loss rates of bond strength and chloride diffusion coefficient of S0.6CA were reduced by 26.34% and 52.5%, respectively, compared with S0.

1. Introduction

Concrete constructions serving in the ocean environment are subjected to long-term seawater erosion, freezing and thawing cycles, loads and other factors, resulting in varying degrees of damage, which severely impacts the durability of structures [1,2,3]. In order to extend the service life of damaged concrete constructions, the utilization of high-performance repair materials for effective and durable repair has become an effective method at present [4,5,6]. However, partial repair projects failed prematurely and needed to be repaired again, resulting in a waste of construction resources [7,8]. As the weak area of the repair project, the bond interface between the repair material and the existing concrete is the easiest to be destroyed first [9,10,11]. The reason is that the hydrophilicity of the concrete substrate increases the water–binder ratio in the interface area. This leads to an increase in the amount of calcium hydroxide and ettringite in the interfacial region, and an increase in the crystal sizes with a directional distribution [12]. It is detrimental to the development of interface bond strength. In addition, air bubbles in the repair material accumulate on the surface of the concrete substrate, increasing defects such as pores and microcracks in the interface area [13]. Moreover, the difference in shrinkage between the repair mortar and the existing concrete will generate localized tensile stresses in the interface area, which is prone to inducing congenital microcracks [14].

When serving in the ocean environment, the micropores in the repaired interface region provided invasion channels for corrosive ions, resulting in a higher ion transport rate in the interface area than the repair material itself, thereby affecting the durability of the repair system [15]. In some restoration projects, even after 1~2 winters, the phenomenon of hollowing and overall peeling of the repair surface layer occurred [16]. The reason is that the repair interface was loose and porous, water was easy to enter, and the repair interface was easy to deteriorate under the action of ice expansion and salt expansion. Therefore, the strengthening of the bonding interface between the repair mortar and the existing concrete will be the key to achieving the formation of a durable repair system.

Research has shown that the bonding interface performance of cement-based repair materials can be improved by enhancing the density of the interface region and reducing crystal orientation, such as the use of interfacial agents, the selection of appropriate cement types, and the addition of admixtures [17,18,19]. Materials with secondary pozzolanic effect, including silica fume, fly ash, metakaolin, slag powder, nano-SiO₂, etc., were beneficial to reduce the content and directional distribution behavior of Ca(OH)₂ in the bonding interfacial zone [20]. Furthermore, mineral admixtures can regulate hydration products and control the phase structure in the bonded interfacial zone to make it denser and more uniform, which is a reliable method for promoting the chemical bonding force of the bonding interface [21]. D. Snoeck [22] and M.A. Yazdi [23] used bacteria to induce CaCO₃ deposition at the interface to heal defects and cracks in the bonding interfacial zone, thereby improving the mechanical meshing force of the existing concrete and the repair mortar, and enhancing the compactness and bond strength of the repair interface. Crystalline admixtures (CAs) have been confirmed to effectively enhance the mechanical performance and durability of repair materials by significantly increasing the impermeability of cementitious materials and promoting self-healing of cracks [24,25]. Therefore, CAs can repair the cracks and defects in the interface area, promote the growth of the hydration products of the interface layer into the pores of the concrete substrate, and improve the mechanical engaging force of the interface layer. It is expected to enhance the bonding performance and the resistance to ionic penetration of the repair interface.

Semendary [26] believed that reducing the shrinkage of the repair material itself can effectively alleviate microcracks and defects caused by stress concentration at the interface, thus improving the bonding strength of the interface. Qian [27] evaluated the crack width, initial water seepage time, and bond strength of SAC and OPC repair mortars by using truncated cone bonded specimens. The results show that OPC repair mortar was prone to cause interface cracking due to its high shrinkage, and its bond strength was inferior to that of SAC repair mortar. Additionally, the water seepage time of the SAC repair mortar was later. The addition of fibers such as steel fibers, polypropylene (PP) fibers, carbon fibers, polyvinyl alcohol (PVA) fibers, etc., can significantly reduce the difference in volume shrinkage of the repair material and the existing concrete [28,29]. Meanwhile, the bridging effect of fibers can inhibit the extension and expansion of cracks, thereby enhancing the bonding strength and impermeability of the interface [30]. Cristina [31] demonstrated that PVA fibers can significantly increase the cohesion force of the repair interface and reduce shrinkage strain and interfacial damage. In summary, due to shrinkage differences, crystal orientation, stress concentration, pore accumulation, and other factors, the repair interface is frequently the weakest area in the repair system [32,33,34]. Previous studies on the bonding interface of repair materials have mostly focused on the degradation mechanisms and strengthening techniques under the general atmospheric environment. However, there are few studies on the evolution characteristics of the bonding interfaces between the repair material and the existing concrete in the complex marine environments.

The aim of this paper is to strengthen the bonding interface of cementitious repair materials using PVA fibers and CAs, so as to enhance the durability of the bonding interface between cementitious repair materials and existing concrete in the marine environment. In this research, the damage process of the repaired bonding interface in the cold marine environment was simulated through seawater-freeze-thaw cycle coupling experiments. The impacts of SAC, PVA fibers, and CAs on the salt-freezing erosion resistance and chloride ion diffusion behavior of the repair interface were systematically investigated. Then, based on the Langmuir equation and the second law of Fick, the chloride ion binding capacity and chloride ion diffusion coefficient were calculated. Finally, the composition and microstructure of hydration products in the repair interface area were analyzed by XRD, BSEM, and micro-indentation. The enhancement mechanism of PVA fibers and CAs on the properties of the repair interface under the coupling action of marine freeze-thaw was explored. This study can provide a theoretical basis and data support for improving the service life of damaged marine concrete repair projects.

2. Materials and Methods

2.1. Materials

The cements used were ordinary Portland cement (OPC) of 52.5R grade and sulphoaluminate cement (SAC) of 42.5 grade. The specific surface area of ground granulated blast furnace slag (GGBS) of S95 grade was 516 m²/kg, and the activity index at 28 d was 102%. The chemical composition and physical performance of OPC, SAC, and GGBS are shown in

Table 1. The chemical composition of OPC, SAC, and GGBS (wt.%).

	CaO	Al2O3	SiO2	Fe2O3	MgO	SO3	K2O	Na2O	LOI
OPC	63.23	5.29	20.38	2.98	2.11	2.46	0.37	0.15	3.03
SAC	42.25	36.46	6.86	2.20	1.33	8.82	0.18	0.22	1.25
GGBS	40.81	18.46	31.55	2.42	4.13	2.05	0.24	0.13	0.21

and

Table 2. The physical performance of OPC and SAC.

	Specific Surface Area (m2/kg)	Setting Time (min)		Compressive Strength (MPa)
		Initial	Final	1 d	3 d	7 d	28 d
OPC	362	168	213	/	36.2	40.3	55.7
SAC	350	30	65	30.5	/	45.0	47.9

. The internally incorporated SKYBURG C35 crystalline admixture (CA) Produced by VELOSIT Company (Horn-Bad Meinberg, Germany) was selected. The fibers were polyvinyl alcohol (PVA) fibers with a diameter of 15 μm, a length of 3 mm, and a tensile strength of 1830 MPa. Quartz sand (S) with grain sizes of 0.2~0.5 mm and 0.5~1.0 mm was used, and the ratio was 4:6. Polycarboxylate superplasticizer (PCE) had a water reduction rate of 25%. Tap water (W) was used for the preparation of mortar specimens and corrosion solutions, and deionized water was used for the determination of chloride ion content. The chemical reagents for the preparation of simulated seawater solution and the determination of chloride ion content were all of analytical grade.

2.2. Preparation of Specimens

2.2.1. Preparation of Repaired Substrate

The 42.5 grade OPC, class I fly ash (FA), S95 grade GGBS, and quartz sand were used to prepare the repair substrates for testing the bonding properties; the mix proportion is listed in

Table 3. Mix proportion of repaired substrate (kg/m³).

	42.5 OPC	FA	GGBS	S	PCE	W
Cement paste substrate	1020	320	160	-	-	560
Mortar substrate	462	132	66	1320	0.66	231

. The paste was used for the preparation of microbonding specimens, and the mortar was used for the preparation of shear bonding specimens. The compressive strength of mortar substrate at 28 d was 63.2 MPa. The mortar substrate was cured in the standard curing room for more than 90 d and then used after its strength development had stabilized.

2.2.2. Preparation of Durability Test Specimens

The repair mortar specimens were prepared following the mix proportions shown in

Table 4. Mix proportion of repair mortars (kg/m³).

Mix ID	OPC	GGBS	SAC	PVA	CAs	S	PCE	W
S0	738	316	0	-	-	1054	4.22	242
S10	664	285	105	-	-	1054	4.22	242
S0.6	664	285	105	8.16	-	1054	6.32	242
S0.6CA	664	285	105	8.16	10.54	1054	6.32	242

. After pouring, the specimen was covered using plastic film and regularly watered for curing. The specimen was demolded after 2 d and moved into the standard curing room to continue maintenance. The specimen used for measuring the chloride ion is displayed in Figure 1. The specimen size was 70.7 mm × 70.7 mm × 70.7 mm, with repair materials on both sides and repair substrates in the middle. After 28 d of curing, the specimen surface was coated with epoxy resin, and the erosion surface was reserved. When the specimen was eroded to the specified age, powder samples were drilled along the bonding interface according to the direction of erosion depth, and the chloride amount at various penetration depths was determined. The preparation process of the specimen used to test the deterioration of bond performance is shown in Figure 2. The change in the bond strength under the erosion environment was tested by using the 45° shear bond specimen. The size of the specimen was 40 mm × 40 mm × 80 mm. The surface of the bonded specimen was wrapped with epoxy resin, and only 10 mm range at the bonding interface was reserved as the erosion surface.

2.2.3. Erosion Solution Preparation and Erosion System

The simulated seawater erosion solution was produced according to ASTM D1141-2021 [35], and simulated seawater composition is shown in

Table 5. Composition of simulated seawater (g/L).

NaCl	MgCl2·6H2O	Na2SO4	CaCl2	KCl	NaHCO3
24.53	11.12	4.09	1.17	0.69	0.20

. In order to ensure the stability of ion concentration in simulated seawater, the erosion solution was replaced once a month. In the seawater-freeze-thaw coupling test, the specimens were subjected to rapid freeze-thaw test under simulated seawater solutions. The control parameters of the freeze-thaw chamber were set so that the central control temperature of the mortar was −17 ± 2 °C and 8 ± 2 °C, respectively. Each freeze-thaw cycle was accomplished over a period of 2~4 h. The specimens need to be immersed in simulated seawater solution for 24 h before freeze-thaw cycle test. After every 50 freeze-thaw cycles, the mortars were removed and dried at 50 °C for 48 h. Then, the bond strength and chloride ion content were measured.

2.3. Experimental Methods

2.3.1. Shear Bonding Strength

The shear bond strength was measured on a compressive testing machine, and the force was applied in compressive shear. The calculation method is as shown in Equation (1), and the result is taken as the average value of three specimens.

$$\sigma ={\displaystyle \frac{\mathit{P}}{\mathit{A}}}$$

(1)

where σ is the nominal shear bond strength (MPa); P is the applied damage load (N); A is the area of the bond interface (mm²), A = 40 mm × 40$\sqrt{2}$ mm.

2.3.2. Restrained Expansion Rate

The restrained expansion rate was tested as per the China national standard (GB/T 23439-2017) [36]. After 14 d of curing in water, the samples were transferred to air (20 ± 2 °C, 60 ± 5% RH) for further curing. The sample length changes were tested at a fixed age, with a precision of 0.001 mm and three samples in each group. The restrained expansion rate was calculated according to Equation (2), and the calculation result was accurate to 0.001%.

$$\epsilon ={\displaystyle \frac{{\mathit{L}}_{ 1 }-\mathit{L}}{{\mathit{L}}_0}} \times 100$$

(2)

where ε is the restrained expansion rate at the measured age (%); L₁ and L are the length measurement values of the specimens at the measured age and the initial age, respectively (mm); L₀ is the reference length of the specimen, 140 mm.

2.3.3. Measurement of Chloride Ions in Mortar

When the mortar specimens were eroded to the corresponding age, the powder samples were drilled along the direction of erosion depth. After drying, the samples were ground and sieved to pass through a 75 μm sieve. Free chloride ion content (C_f) was determined using potentiometry using a chloride ion selective electrode. A total of 1 g of sample was immersed in 40 mL deionized water for 24 h, and then the chloride ion concentration in the soaking solution was detected through a rapid chloride ion content meter, as illustrated in Figure 3. The free chloride ion amount in the powder sample can be obtained by conversion. The total chloride ion content (C_t) was measured following the acid solubilization approach. A total of 1 g of powder sample was taken and titrated with NH₄SCN solution, and the total chloride ion amount was obtained following the test procedure in Figure 4.

2.3.4. Microscopic Test Methods

When the paste specimens were eroded to a fixed age, the block or powder samples were obtained at a specific position according to the design scheme. Subsequently, block or powder samples were immersed in anhydrous ethanol to discontinue hydration and ground to pass through a 75 μm square sieve. After drying at 40 °C for 48 h in a vacuum oven, the crystal components of specimens were analyzed via an X-ray diffractometer (XRD), with a scanning range of 5~75°. The repaired bonding interface was observed using the backscattered mode (BSEM) of the scanning electron microscope. Before the experiment, block samples were taken at 5~10 mm from the erosion surfaces. Then, the specimens were immobilized using epoxy resin and burnished step by step on a metallographic polishing machine using 800~10,000 mesh sandpaper until the surface was glossy. During the polishing process, deionized water was used to continuously rinse. After each level of polishing, the samples were washed in anhydrous ethanol using an ultrasonic cleaner.

2.3.5. Microhardness Analysis

The block samples were taken 0~25 mm from the erosion surface. After drying, the samples were polished to 10,000 mesh by metallographic sandpaper and further polished with 12,000 mesh and 15,000 mesh polishing pastes. The microhardness was indented using a Vickers indenter from Shanghai Taiming Optical Instrument Co., Ltd. (Shanghai, China). A 40× objective lens was selected, and the eyepiece was used with an electronic microscope camera with a 10× magnification. The total magnification was 400×. The test force was 9.807 N (1000 gf), and the retention time was 15 s. Due to the difference in physical properties between the repair material and the mortar substrates, the deterioration may develop horizontally on both sides of the bonding interface in addition to the longitudinal development. Therefore, the longitudinal indentation points were selected at 2 mm, 4 mm, 6 mm, 8 mm, 12 mm, 16 mm, and 20 mm along the direction of erosion depth, and the transverse indentation points were selected at 1 mm, 2 mm and 3 mm from the bonding interface, as shown in Figure 5. Indentation should try to avoid the holes in the materials. Vickers hardness value was calculated according to Equation (3).

$$\mathit{HV}= 0.102 \times {\displaystyle \frac{2\mathit{F}\mathit{sin}\frac{\alpha }{2}}{{\mathit{d}}^{ 2}}}=0.1891 \times {\displaystyle \frac{\mathit{F}}{{\mathit{d}}^{ 2}}}$$

(3)

where HV is the Vickers hardness value; F is the experimental force (N); α is the angle of the indenter relative surface, 136°; d is the average length of the indentation diagonal (mm).

3. Results and Discussion

3.1. Mechanical and Volume Deformation Properties of Repair Mortars

3.1.1. Mechanical Properties

The compressive strength and bond strength of the repair mortar are illustrated in Figure 6a,b. In Figure 6a, the compressive strength of S10 at all ages was higher than that of S0. After adding a suitable amount of SAC, ettringite (AFt) and hydrated calcium silicate (C-S-H) gel were interspersed and filled with each other, which was conducive to optimizing the pore structure and improving the density of the system. Furthermore, the dissolution and hydration reaction process of OPC was promoted by the rapid hydration release of SAC and the consumption of hydration products such as Ca(OH)₂. After adding PVA fiber, the compressive strength of S0.6 was higher than that of S10 at 1 d, but lower than that of S10 at all other curing ages. The reason is that the pastes containing fibers tended to become entangled and agglomerate during the mixing process, trapping air in the mixture. The bubbles formed through the mixing procedure were also impeded by the fibers, making it difficult for bubbles to escape from the surface of the specimen [37]. This leads to higher porosity of the mortar. The compressive strength of S0.6CA was higher than that of S0.6 at all ages. It is explained that after adding CAs, the complex precipitation products had a certain filling effect on the pores brought about by PVA fibers, which promoted the repair mortar to be more compact, thereby weakening the adverse effect of PVA fiber on the compressive strength.

In Figure 6b, the bond strengths at 3 d, 7 d, and 28 d were ranked from highest to lowest as follows: S0.6CA, S0.6, S10, and S0. SAC promoted the early hydration process, increased the generation of early hydration products, and quickly filled the pores at the interface, so that to enable rapid development of bonding strength. In addition, due to the micro-expansion characteristics of SAC, the shrinkage deformation of the repair material can be alleviated, thereby reducing the number of microcracks in the bonding interface area. The shrinkage of the repair material can be restrained by PVA fibers, thus enhancing the bonding strength to a certain degree. CAs significantly improved the shear bond strength of repair mortars at different ages. On the one hand, the incorporation of CAs can promote the densification of the repair material itself. On the other hand, during the continuous curing process, the active groups in CAs played a role in healing and filling microcracks and defects at the bonding interface. Slip failure mainly occurred in shear bond specimens. A portion of the healing products generated by the CAs reaction filled the pores of the bonding interface, while another portion grew into the pores of the substrate. When slippage occurred at the bonded interface, these healing products acted as ‘key pins’ at the interface, preventing the slip of the bonded interface.

3.1.2. Volume Deformation Properties

The restrained expansion rate of repair mortar is illustrated in Figure 7. During curing in the water, due to the micro-expansion characteristics of SAC, the S10, S0.6, and S0.6CA exhibited higher expansion deformation compared to the S0. Among them, the addition of PVA fibers inhibited the expansion of mortars in water, which is caused by the constraint effect of fibers on the deformation of repair mortar. The addition of CAs can promote the expansion of the mortar. This is because under the condition of sufficient water, new healing products were continuously produced by the reaction of the active substances in CAs and filled in the pores and microcracks. In this process, a certain extrusion action occurred on the hole wall and crack, causing the expansion and deformation of the specimen volume.

After the specimens were cured in air, both PVA fibers and CAs can still inhibit the drying shrinkage deformation of mortar. The CAs promoted the densification of repair mortar, and its healing and filling effect sealed capillary pores and penetration channels, thereby alleviating the shrinkage caused by water loss inside the mortar. In the case of composite use of PVA fibers and CAs, the constraint effect of fibers on shrinkage deformation and the inhibition effect of CAs on internal water loss in mortar can be simultaneously exerted. Therefore, the shrinkage rate of S0.6CA was the lowest. Compared with S0, the restrained shrinkage rate of S0.6CA at 90 d was reduced by 81.8%, indicating that the synergistic effect of PVA fiber and CAs can significantly improve the volume stability of the repair material.

3.2. Performance Damage of Repaired Bonded Interfaces Under Seawater-Freeze-Thaw Environment

The impact of seawater-freeze-thaw coupling on bond strength is shown in Figure 8a,b. Compared with the repair material itself, there were more pores and microcracks in the bonding interface area, so it is more prone to deterioration in a harsh environment. In the salt freezing test, seawater penetrated along the bonding interface, and the pores in the interface area were in a water-filled state. During the freezing process, the expansion stress was generated, so that the repair material was gradually separated from the substrate. From Figure 8a, it can be found that after 50 freeze-thaw cycles, the shear bond strength of each test group began to decrease. After 700 freeze-thaw cycles, the residual bond strength of S0.6CA was highest. This is due to the continuous self-healing effect of CAs on microcracks and pores at the interface during freezing and thawing cycles. Furthermore, the compactness of the bonded interface region between the repair mortar added with CAs and the substrate was higher, and the resistance to seawater penetration was better. Thus, the degree of water saturation in the interfacial pores was relatively lower.

In the results of the change rate of bond strength (Figure 8b), the fastest rate of decrease in bond strength was observed for S0 and S10, with loss rates of 26.42% and 27.98%, respectively, whereas the loss rates of bond strength of S0.6 and S0.6CA were 22.63% and 19.46%, respectively. The lower bond strength of S0 is due to the large dry shrinkage of the OPC-GGBS composite materials, which tends to form shrinkage cracks at the interface. For S10, it may be because of the presence of a high number of macropores and connected pores in the repair material itself, and these negative effects are also reflected at the bonding interface. The above results indicate that the addition of PVA fibers and CAs can improve the bonding strength of the repair material under seawater-freeze-thaw coupling.

3.3. Chloride Transport Behavior in Repaired Bonded Interfaces Under Seawater-Freeze-Thaw Environment

3.3.1. Distribution of Chloride Ions

The distributions of free and total chloride ions at the bonding interface under seawater-freeze-thaw action are shown in Figure 9a–d and Figure 10a–d, respectively. Compared to the transport of the repair materials, the migration rate of chloride ions at the bonding interface was faster [15]. After 300 freeze-thaw cycles, the dispersion depth of free chloride at the bond interface of both S0 and S10 reached 8 mm, while that of S0.6 and S0.6CA was only 6 mm. After 700 freeze-thawing cycles, the diffusion depth of free chloride ions of all specimens reached 20 mm, but the C_f of S0.6CA at 20 mm was still at a low level (C_f < 0.15%). It can be concluded that although the single incorporation of fiber enhanced the deterioration resistance of the repair material, the improvement effect on the anti-chlorine ion permeation at the bonding interface was still low. When the fibers were used synergistically with CAs, it not only healed the damage in the repair material, but the healing products produced by CAs also migrated towards the interface and hindered the further penetration of chloride ions.

For the dispersion of total chloride ion (Figure 10a–d), after 50 freeze-thaw cycles, the C_t at the bond interface started to increase in the depth range of 0~20 mm. The deterioration of the bond interface responded to the freeze-thaw action to a high degree, because there were more connected pores on the bonding interface, which were less resistant to freeze-thaw. As soon as freezing and thawing began, chloride ions in seawater were transported deeper along the interface. Meanwhile, because of the relatively weak adhesive forces at the interface, the expansion stress generated by freezing caused the cracks at the interface to continuously expand and extend inward, accelerating the permeation process of chloride in the interface. When the number of freeze-thaw cycles was low, the cracks did not extend deeper into the interface, and the amount of chloride ions penetrating was relatively small. Moreover, the monosulfate hydrated sulfoaluminate (AFm) in the repair material reacted with free chloride ions to form Friedel′s salt, thereby reducing the concentration of free chloride ions and also sealing a portion of the permeation channel [38].

3.3.2. Binding Capacity of Chloride Ions

Figure 11 presents the relationship between the free chloride ion and total chloride ion amounts at the bonding interface. The measured results were fitted with the Langmuir adsorption equation [39], the adsorption equation is shown in Equation (4), and the relevant fitting parameters are listed in

Table 6. Fitting parameters of Langmuir isotherm model.

Fitting Parameters	S0	S10	S0.6	S0.6CA
α1	1.705	2.272	1.584	1.750
β1	25.967	24.106	13.971	14.193
R2	0.845	0.959	0.955	0.883

. When C_f > 0.30%, C_b basically no longer increased, indicating that in this concentration range, the fixation of chloride ion at the interface by hydration products was close to saturation. The reason for this is that there is a large difference in chloride binding capacity between the repair material and the substrate on both sides of the bonding interface. The chloride binding capacity of the modified repair material was higher than that of the normal substrate. When chloride ions entered the interface, more free chloride ions were combined with the repair material. The chloride binding capacity of S0.6CA was highest when C_f < 0.30%. This is because the CAs promoted the generation of more hydration products at the interface, such as C-S-H gel and Mg(OH)₂ gel, which may encapsulate chloride ions. When C_f was higher, the chloride binding ability of different test groups at the bonding interface corresponded to that of the repair material itself. The binding abilities of S10, S0.6, and S0.6CA were similar, followed by S0.

$${\mathit{C}}_{\mathit{b}}={\displaystyle \frac{{\alpha }_1{\mathit{C}}_{\mathit{f}}}{1+{\beta }_1{\mathit{C}}_{\mathit{f}}}}$$

(4)

where: α₁ and β₁ are the adsorption parameters in accordance with Langmuir adsorption equation.

3.3.3. Diffusion Coefficient of Chloride Ions

On the basis of the macro diffusion model of chloride ion in Equations (5) and (6) [40,41], the chloride diffusion coefficients at the bonding interface under seawater-freeze-thaw action were calculated by substituting the C_f measured at the bonding interface into the formula. The results are illustrated in Figure 12. The permeating speed of chloride ions at the bonding interface was much higher than that in the repair material, which is because the bonding interface was more susceptible to the expansion of ice and salt induced by the seawater-freeze-thaw environment. Once cracks occurred at the interface, chloride ions quickly penetrated along the interface to the surface of the reinforcement, causing corrosion of the steel bar. Therefore, in order to improve the resistance of the repair material to chloride penetration, the strengthening of the bonding interface cannot be neglected. After 700 freezing and thawing cycles, the chloride diffusion coefficients of S0.6CA were reduced by 52.5% and 48.2% compared to those of S0 and S10, respectively. It can be demonstrated that the continuous repair and closure of damage and diffusion channels at the interface by CAs greatly retarded the chloride ion transport rate in the bonded interface. In the early stage of the freezing test, the chloride diffusion coefficient of S0.6 was not significantly different from that of S0 and S10. However, as the frequency of freezing-thawing cycles increased, the deterioration of the repair mortars closes to the interface of S0 and S10 also intensified, leading to a further increase in the number of chloride ion channels that can penetrate along the interface. After 700 freeze-thaw cycles, the chloride diffusion coefficient of S0.6 was lower, suggesting that fiber inhibited the cracking of the repair material and promoted the self-healing effect of CAs on interfacial damage, thus reducing the number of channels available for chloride penetration.

$$\left\{\begin{array}{c}\mathrm{i}\mathrm{n}\mathrm{i}\mathrm{t}\mathrm{i}\mathrm{a}\mathrm{l} \mathrm{c}\mathrm{o}\mathrm{n}\mathrm{d}\mathrm{i}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}: \mathit{C}\left(\mathit{x} \ge 0, \mathit{t}=0\right)={\mathit{C}}_0\\ \mathrm{b}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{d}\mathrm{a}\mathrm{r}\mathrm{y} \mathrm{c}\mathrm{o}\mathrm{n}\mathrm{d}\mathrm{i}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}: \mathit{C}\left(\mathit{x}=0, \mathit{t} \ge 0\right)={\mathit{C}}_{\mathit{S}}\end{array}\right.$$

(5)

$$\mathit{C}\left(\mathit{x},\mathit{t}\right)={\mathit{C}}_0+\left({\mathit{C}}_{\mathit{s}}-{\mathit{C}}_0\right)·\left[1-\mathit{erf}\left({\displaystyle \frac{\mathit{x}}{2\sqrt{\mathit{D}·\mathit{t}}}}\right)\right]$$

(6)

where C is the chloride ions concentration inside the concrete (%); t is the chloride ions diffusion time (d); x is the chloride ions diffusion depth (mm); D is the diffusion coefficient of chloride ions (m²/s); C₀ and C_s are the initial and surface chloride ion concentration (%), respectively; erf is the Gaussian error function, $\mathit{erf}\left(\mathit{z}\right)={\displaystyle \frac{2}{\sqrt{\pi }}}{\int }_0^{\mathit{z}}\mathit{exp}\left(-{\beta }^2\right)\mathit{d}\beta$.

3.4. Analysis of Damage and Degradation Mechanism

3.4.1. Corrosion Products Analysis

The powder samples obtained at the interface were actually a mixture of one side of the repair material and one side of the substrate. Figure 13 exhibits the XRD pattern at the bonding interface after 700 freeze-thaw cycles. During the hydration process of the bonded specimen, a water film was easily formed at the interface, and Ca(OH)₂ was prone to be enriched and directionally distributed in the interface region [10]. Therefore, the diffraction peaks intensity of Portlandite in the sample at the interface was higher, which is also a reason for the lower bond strength of the interface area. In addition, the intensity of the diffraction peak of Friedel’s salt at the interface was also greater, which is attributed to the fact that a larger amount of Cl⁻ penetrated along the interface and produced more Friedel’s salt. The intensity of the diffraction peaks of Friedel’s salt in the samples obtained at the depth of 4~6 mm was higher than that at the depth of 16~18 mm, suggesting that the number of Cl⁻ penetration affected the quantity of Friedel’s salt. The diffraction peak intensity of Brucite in S0.6CA (4~6 mm) was lower than that in S10, indicating that the amount of Mg²⁺ in seawater permeated into S0.6CA was less. This reflects, to some extent, that S0.6CA had the superior resistance to erosive ion permeation at the bonding interface.

3.4.2. Micromorphology Analysis

The BSEM-EDS image of the bonded interface after 700 freeze-thaw cycles is illustrated in Figure 14. In S10, the number of cracks on the side of the old substrate was more than that on the side of the repair material. This is due to the lower strength and compactness of the old substrate, which is more prone to cracking under the action of salt freezing. The development of cracks on the side of the old substrate in S0.6CA was essentially the same as that in S10, whereas the number of cracks on the side of the repair material was significantly reduced compared to S10. It is attributed to the cracking resistance effect of the fibers. From Figure 14a, it can be observed that a high amount of corrosion products was enriched in the cracks at the bond interface, mainly including AFt, Mg(OH)₂ gel, and Friedel′s salt, which were produced by the reaction of seawater entering along the interface with cement hydration products. Both AFt and Friedel′s salts were strongly expansive, while the Mg(OH)₂ gels were formed through displacing Ca²⁺ from Ca(OH)₂ in the hydration products, which will lead to a looser microstructure of the cement stone. Therefore, the invasion of corrosive ions was also the main reason for the increase in microcracks in the bonding interface area.

In the results of the EDS line scans, the content of Ca and Si in the samples tended to decrease at the bonding interface, and the content on the repair material side was higher than on the old substrate side. Under the influence of seawater infiltration, calcium may be lost from C-S-H gel and Ca(OH)₂ at the interface, and there were obvious peaks in the spectrum of Mg, demonstrating that the loss of Ca²⁺ at the interface may be related to the predation of OH⁻ by Mg²⁺. The content of S, Cl, and Mg in S0.6CA was lower than that in S10, especially the content of Cl was significantly lower than that in S10, which indicates that S0.6CA had great advantages in resisting the penetration of corrosive ions in seawater. The reason is that there are fewer microcracks on one side of the repair material in S0.6CA, and the number of channels through which corrosive ions can penetrate is reduced. In addition, the active groups in CAs can continuously heal microcracks at the interface through complexation-precipitation reaction, thus reducing the permeation of corrosive ions.

3.4.3. Microhardness Analysis

The influence of seawater-freeze-thaw action on the Vickers hardness of repair materials is displayed in Figure 15. Before salt-freeze erosion, there was no significant difference in the initial Vickers hardness of each group. Due to the higher water–binder ratio used in the preparation process, the Vickers hardness on the old substrate side was relatively low. Therefore, the residual Vickers hardness after salt-freeze deterioration was also lower, about 30–35 kgf/mm². For the side of the repair material, after 100 freeze-thaw cycles, the decrease rate of Vickers hardness of S0 and S10 gradually accelerated, and the decrease was also greater. After 300 freezing-thawing cycles, the Vickers hardness of S0.6 and S0.6CA decreased significantly. Compared with before the freezing and thawing cycle experiment, the loss rates of Vickers hardness of S0, S10, S0.6, and S0.6CA were 39.1%, 36.4%, 23.8% and 20.0% after 700 freeze-thaw cycles, respectively. It can be concluded that the addition of fiber and CAs significantly reduced the damage and deterioration of mortars caused by seawater-freeze-thaw action.

For the old substrate side, after 700 freeze-thaw cycles, the loss rates of Vickers hardness of S0, S10, S0.6, and S0.6CA were 40.8%, 36.5%, 32.7% and 30.6%, respectively. Among them, the loss rates of Vickers hardness of the old substrate side in S0.6 and S0.6CA were relatively low, indicating that the repair material added with fiber also had a certain protective effect on the old substrate. The main reason is that the addition of fibers delays the damage deterioration process at the bonding interface, thereby reducing the chemical erosion and salt crystalline expansion damage on the old substrate side caused by the penetration of corrosive ions in the interface. The change rule of Vickers hardness in different test groups under a salt freezing environment corresponded to the damage deterioration pattern of the macroscopic properties. It also explains the enhancement mechanism of PVA fiber and CAs modification on the freeze-thaw resistance and chlorine erosion resistance of the repair materials from the micromechanical scale.

4. Conclusions

In this research, the impacts of PVA fibers and CAs on the bond strength, chloride ion binding ability, and chloride ion diffusion behavior of the repair interface under the coupling action of seawater-freeze-thaw cycles were investigated, and the mechanism of effect was clarified based on microstructure analysis. The main conclusions obtained are as follows:

(1)

The incorporation of PVA fibers and CAs can inhibit the drying shrinkage and deformation of the repair mortar, promote its compressive strength and bond strength. The compressive strength of S0.6CA was 56.2 MPa and 101.7 MPa at 1 d and 28 d, respectively, and the 45° shear bond strength was 29.13 MPa and 45.95 MPa at 3 d and 28 d, respectively. Compared with S0, the restrained shrinkage rate of S0.6CA decreased by 81.8% at 90 d.

(2)

As the number of seawater-freezing-thawing cycles increased, the bond strength of the specimens increased first and then decreased. After 700 seawater-freeze-thaw cycles, the loss rates of bond strength and Vickers hardness of S0.6CA were the minimum, which were 19.46% and 20.0%.

(3)

Under the seawater-freezing-thawing environments, the free chloride ions and bound chloride ions in the bonding interface showed a nonlinear binding relationship, which was better fitted by the Langmuir adsorption equation. After synergistically modified PVA fibers and CAs, the diffusion coefficient and diffusion depth of chloride ions at the repair interface were significantly reduced, and the chloride binding capacity was enhanced. After 700 cycles, the chloride diffusion coefficients at the repair interface of S0.6CA decreased by 52.5% and 48.2% compared to S0 and S10, respectively.

(4)

PVA fiber reduced the number of microcracks caused by freeze-thaw and salt crystallization, thus delaying the damage and degradation rate of repair materials. The incorporation of CAs on the one hand improved the permeability resistance of the repair material and its bonding interface and reduced the amount of erosion ions penetrating into the specimen. On the other hand, CAs generated healing products in the pores and cracks through complexation-precipitation reactions, which provided self-healing repair of the damage and reduced the deterioration degree of the specimen.

In this study, the bonding properties and volume stability of cementitious repair materials, as well as the durability of the repair interface under seawater-freeze-thaw environments, were successfully improved by the synergistic modification of PVA fibers and CAs. This study only evaluated the variation rule of 40° shear bond strength, and future research will explore the effects of different bond strength forms and interface treatments on the performance of the repair interface. In addition, in view of the complexity of the marine environment, the bonding properties and long-term durability of the repair interface under the multifactorial coupling of load-dry-wet-seawater-freeze-thaw cycles should be investigated, so as to further evaluate the applicability of PVAs/CAs modified repair materials in the marine environment.