The influence of sepiolite on the self-healing capability of concrete by means of water permeability

. This paper investigates the autogenous self-healing capability of conventional concrete (C30/37) mixtures with water encapsulated in sepiolite. Sepiolite was added in two conditions: previously saturated and dry, at a dosage of 5% by the cement weight. Sepiolite is added in substitution for the sand fraction. For these two mixes and a reference mix without sepiolite, disks of size f 100x50 mm were produced and were pre-cracked at 28 days of age by splitting test until reaching residual cracks of 300±150 µm. Crack width was measured by using an optical microscope. Self-healing was promoted after pre-cracking in certain exposures: 1) continuous water immersion at 20°C for 56 days, 2) a high humidity environment at 20°C and 95% of relative humidity for 28 days and water immersion for additional 28 days, and 3) pre-saturation for one day and 55 days in a humidity chamber. Self-healing was analyzed with water permeability by comparing the results before and after the healing stage. Afterward, chlorides’ penetration through the healed cracks was evaluated to study the possible protection provided by crack healing. Uncracked specimens were also tested as a reference for chloride penetration. Mixes were characterized by measuring compression strength at 28 days, slump, air content, and fresh density. The results show that water immersion is an adequate way to improve autogenous self-healing. Sepiolite can improve the self-healing capability of concrete with only one day of pre-saturation and then healing in high humidity conditions, especially in mixes with sepiolite introduced in saturated conditions. On the contrary, sepiolite may have resulted in a higher chloride penetration compared to samples without sepiolite.


Introduction
Autogenous healing is an intrinsic potential of cracked cementitious materials to regain partial or total initial properties of the material [1]. Autogenous healing is mainly produced by continuing hydration of the unhydrated cement particles or carbonation and has been suggested to recover the composite durability properties considerably [1]. Autonomous healing promotes healing by adding engineered materials designed for this purpose, like crystalline admixtures, superabsorbent polymers (SAP), bacteria, etc.
Due to the effect of cracks on the watertightness of an element, water permeability tests conducted in cracked conditions are frequently employed to measure the healing effectiveness [2][3][4][5]. In addition, water penetrating through the cracks into the concrete matrix may transport aggressive chemicals, such as chloride ions, to the reinforcement level. Additionally, it has been found that permeability grows rapidly for cracks ranging from 50 to 200 μm in width, whereas the growth is constant for cracks greater than 200 μm [6].
Chloride ions are recognized as one of the most aggressive agents for reinforced concrete structures due to their ability to rapidly permeate reinforced concrete that has undergone substantial cracking and degradation [7,8]. It has been reported that 10 μm is the key crack width for chloride penetration, and smaller cracks will not allow a chloride penetration [9]. For larger cracks, 60 μm has been identified in some investigations as the maximum crack width that can be completely healed in terms of chloride diffusion resistance. [10]. Also, in * Corresponding author: hedoo@upv.es another study [11] provided 13 μm as the critical crack width to avoid more chloride penetration in the short term, whereas 40 μm was reported as the critical crack opening to have no major impact on chloride penetration.
Sepiolite is a microcrystalline-hydrated magnesium silicate with the formula of Mg4Si6O15(OH)2·6H2O, which exists in fibrous, fine-particulate, and solid states. This unique structure gives sepiolite a fibrous matrix with 0.36 mm channels aligned along the fibers' longitudinal axis. Sepiolite's fiber structure imparts sorptive, colloidal/rheological, and catalytic properties with a wide range of uses [12]. It is known for its low specific gravity and high porosity. Some studies show that using 2% [13] and 10% [14] sepiolite as a cement replacement resulted in mechanical strength improvement of mortar. Sepiolite has also been used as a carrier for bacteria cells immobilized, presenting more efficiency in the self-healing efficiency [15,16]. By filling the cavities in the cement mortar, sepiolite improves the compressive and bending strengths of the mortar, resulting in positive impacts on the compressive and bending strengths [13,14].
This study compares the self-healing capability of conventional concrete (C30/37) mixtures using sepiolite added in pre-saturated and dry conditions, evaluated by means of water permeability and chloride penetration.

Materials
The cement used was CEM I 42.5 R-SR5 from Lafarge. The aggregates used were 0/2 and 0/4 (dMin/DMax in mm) natural sands and 8/16 coarse aggregates. All the granular constituents were added in dry conditions. A dosage of 40 kg/m³ of steel fibers (Dramix RC-65/35-BN, 35 mm length, and 0.55 mm diameter) was used to control crack width during the pre-cracking and healing stages. Superplasticizer ViscoCrete-5970 from Sika was added to achieve the desired workability.
Sepiolite porous sand with a particle size of 0.25 to 1.18 mm and a 0.64 g/ml density was added in a dosage of 5% dry weight over the cement weight as a substitution for the same mass of 0/4 fine aggregate. An additional water equivalent to 80% of the dry sepiolite mass was added to mixtures containing sepiolite. Sepiolite was introduced in two conditions, dry condition, and then additional water was incorporated as added water, labeled as CCDS, and pre-saturated labeled as CCSS. CCSS Sepiolite was used to analyze the effect of encapsulated water as extra water to promote the self-healing of concrete internally. The selfhealing proficiency of these mixes is compared with conventional concrete without sepiolite, labeled as CC. The mix design of each concrete is presented in Table 1. Concrete cylinders of f100×200 mm were cast and cured in the humidity chamber for 28 days at 20°C and relative humidity > 95%. The tests performed in the fresh state were consistent with the slump test following EN 12350-2 and the fresh air content following EN 12350-7. Each mix was characterized by compressive strength at 28 days following EN 12390-3, tested on 3 cubes of 150 mm.

Pre-cracking and crack width measurement
To evaluate self-healing, disks of size f100×50 mm, were obtained by cutting the casted cylinders f100×200 mm ( Figure 1) and were pre-cracked by splitting test to produce a residual crack of size between 100 and 450 µm at 28 days of age. The crack width was measured using a YINAMA wireless optical microscope after pre-cracking and before performing each water permeability test in all the disks. To obtain a value representative of a disk, the crack of each surface was measured in three points located at intervals of 25 mm on both surfaces ( Figure 1). Therefore, the representative crack width of a disk is the average of these six values. At least 6 specimens for each mix and healing condition, together with 4 reference uncracked disks (R samples), were tested; that sums to at least 28 samples tested per mix.

Healing conditions
After pre-cracking, the disks are divided into groups depending on their allocated healing condition. Three healing conditions were analyzed: 28 days immersion in water at 20°C (WI disks), 28 days in the humidity chamber (HC disks), and a pre-saturation for a day in the water and after keeping in the humidity chamber for 27 days (PS disks). After this initial curing period, the specimens are stored for a second 28 days of the healing period. For the second healing period, WI disks were immersed again in water, PS disks were preserved in the humidity chamber, and HC disks were immersed in water labeled as 28HC+28WI. To achieve the desired healing to perform the chloride penetration test, samples of HC were healed in water for another 28 days (28HC+56WI), and samples of PS were kept in a humidity chamber for another 28 days (84PS).

Low-pressure water permeability test
After measuring the cracks, the water permeability test was done using the pre-cracked disks. The cracks of the lateral curved surface were sealed with the resin Sikaflex 11 FC, then tubes of f100 mm with a height of 250 mm were glued to the disk using that same resin ( Figure 2). After the preparation, the tubes were filled with a 200 mm water column, and the water head was measured after 0 and 30 minutes. At the beginning of the test, to record the water head at time 0, the bottom crack was sealed with tape to prevent initial water leakage before starting the measurement. This water permeability test was performed before and after each healing period. (3 times for each specimen)

Fig. 2. Water-permeability test setup
The water head reduction after 30 minutes was evaluated before the first healing period and after each healing period, was used to obtain the healing efficiency (Healing Ratio) using Eq.1.
After the second water permeability test in healed samples, the penetration of chlorides through the completely healed cracks (with no leakage) was evaluated.
A modified chloride permeability test presented in previous studies [17,18] was performed. This test provides information about the penetration of chlorides throughout the healed crack path and the matrix. The same tubes were used for this test, which were filled with a 200 mm water column containing 33 g NaCl/liter. This setup was maintained for 3 days. After 3 days, the tubes were removed, and the disks were sawed perpendicular to the direction of the crack. The sawing process was done in dry conditions to avoid contamination because transport of salt due to a wet cut. Subsequently, AgNO3 solution with a concentration of 0.1 mol/liter was sprayed on both cut surfaces. Moreover, uncracked disks were used as a reference to understand better the phenomena involved, labeled as R. In the sawed surfaces, the cracked disks showed the chloride penetration pattern presented in Figure 3; two values were measured: the penetration depth, labeled as P0 through the top surface of the disc, and the penetration through the walls of the crack, labeled as W. For measuring P0, the penetration depth in four points on both sides of the cracks is measured. Similarly, three points are measured for penetration in the crack path ( Figure 3). It was expected that the 20 mm measuring distance from the crack for both P0 and W measurements would exclude the influence of probable excessive penetration caused by the joint of the matrix's penetration and the crack path's penetration. The compressive strength at the age of 28 days of the mixes is presented in Table 1. The addition of Sepiolite in dry and saturated conditions resulted in lower compressive strength than reference mixes (as in the case of CC mixes). The decreases in compressive strength are 5% and 9% for mixes containing sepiolite in dry and saturated conditions, which are all between 37 and 41 MPa. The air content of the CC samples using dry sepiolite increased slightly, from 1.1% to 1.3%. While for the addition of sepiolite in a saturated condition, the air content of the sample was reduced to 0.5%. The expected workability decrease of the mixes with sepiolite was compensated by increasing the superplasticizer dosage ( Table 1). The slump for CCSS and CC was practically the same: between 180-200 mm. In the case of CCDS, a 60 mm slump was obtained.
The values of the Healing Ratio were analyzed depending on the initial average crack width obtained for each disk before healing. Figure 4 shows the healing ratio vs the initial crack size divided into healing conditions after the first healing period (28 days) and after the second healing period (another 28 days). Additionally, the average healing ratio in each initial crack width ranges, 100-200, 200-300, or 300-400 microns, in both healing periods is summarized in Table  2, with shaded cells indicating low-efficiency values.   The results in Figure 4 and Table 2 indicate more than 80% healing for all samples with initial cracks up to 300 µm in terms of water permeability in most samples that healed for 56 days in water immersion. However, comparing the reference samples, there is no notable difference in the healing ratio of samples healed in WI condition when using sepiolite. For all samples healed in HC condition, healing ratios for 28 days of healing were almost zero; meanwhile, additional 28 days of water immersion could promote the healing capability with variable healing values between 30 and 100%, very similar to samples in WI in the first curing period. In this case, the healing ratio of the samples' CCSS is higher than CCDS and CC.
Comparable to the HC condition, the healing ratios of the PS samples in the first 28-day period were negligible. In contrast, samples incorporating sepiolite with 56 days of healing in PS had higher healing proficiency to have healing ratios between 50 to 100% depending on initial crack width compared to samples without sepiolite. Comparing the samples of presaturated conditions in 56 days, the healing ratio of conventional concrete incorporated water encapsulated in sepiolite was higher than reference conventional samples, mainly those with SS.

Chloride penetration of healed cracks
Regarding the penetration depth through the matrix (P0), the average P0 of a reference conventional concrete healed in water immersion condition is practically the same as concrete samples with sepiolite regardless of the healing condition with 1 mm of penetration ( Table 3).
The results show that using sepiolite substantially affects the P0 value measured in samples healed in HC and PS conditions, leading to up to 3 mm lower P0 in samples using DS and SS compared to the reference samples. Samples of both mixes with sepiolite had almost the same P0 of less than 2 mm in all healing conditions. Regarding the chloride penetration through the crack path (W) increased with the crack width increment ( Figure 5 and Table 3). Moreover, samples containing sepiolite mainly mixes containing SS, exhibited a higher chloride penetrability than ones without sepiolite.   Table 3. Chloride penetration through the matrix (P0), and the crack (W) in each initial crack width range. Shaded cells indicate low-efficiency values.

Conclusion
The conclusions that can be drawn are: • The healing results in terms of water permeability for the samples healed in water immersion conditions for 28 and 56 days are higher than those healed in other conditions. Water immersion promotes self-healing detectable by water permeability. • Since there was no healing in the first 28 days for the HC and PS, the self-healing improved in the period between 28 and 56 days in several cases, even in PS samples that healed in highly humid condition. • For samples healed in HC+WI condition, the healing ratio of the sample incorporating saturated sepiolite is higher than CCDS and CC. • The sepiolite incorporation, especially in a saturated condition, shows a more apparent profit when concrete is healed in HC+WI condition and pre-saturation. In cases with cracks higher than 200 μm, the effect on the healing ratio is more notable. • The chloride penetration through the matrix is influenced by healing condition and the inclusion of sepiolite in dry and saturated conditions, mainly in HC+WI and PS conditions. • The chloride penetration through the wall of the crack increased when the crack width increased. Samples healed in the WI condition incorporating sepiolite in the saturated condition had higher penetrability to chloride. Moreover, the presence of sepiolite may influence higher chloride penetration than samples without sepiolite.