A self-healing approach to cement-based materials with crystalline admixtures in normal and accelerated environmental conditions

. Cracks in cement-based materials can be a source of significant problems in civil works, especially those exposed to the action of harmful agents and moisture. Its occurrence may be associated with various physical, chemical, or mechanical factors. Self-healing phenomenon can promote cracks repair, contributing to a useful life increase through the sealing by cracking closure. This work aimed to evaluate the self-healing phenomenon in cementitious matrices with different crystalline admixtures. Mortars were produced using high initial strength cement, quartz sand, three types of crystalline additives, and 0.4 as w/c ratio. Compressive strength tests were performed at 3 days (crack opening age), 28 days, and 45 days. The environmental conditions were wet/dry cycles in controlled and accelerated climate chambers. Ultrasonic pulse and optical microscopy with image editing techniques were used to evaluate the self-healing phenomenon. The results indicate that the samples in accelerated chamber (CAR) presented the highest mechanical strength at 45 days, and the most significant crack closure to samples with Z crystalline admixture. The ultrasonic wave propagation speed analysis showed that crystalline admixtures-based samples (X, Y and Z) had the highest matrix densification. Additionally, special attention should be given to the samples preparing steps and images editing to identify the healed area for the techniques being used.


Introduction
Portland cement concrete is the second most used material by humanity and the most widely used in civil construction. Conventional concrete consists of a hydraulic binder (Portland cement), coarse and fine aggregates, such as crushed basalt stone in its various sizes and sand, respectively, and water [1]. Superplasticizers, mineral and viscosity modifiers, stabilizers, crystalline admixtures, among others, can be used in cementitious matrices to obtain specific property, either in the fresh or hardened state. It depends on where the concrete will be applied, its function (such as high loading, noise, and heat insulation), and its casting (vibrated, pumped, self-compacting). Several sources can produce cracks in concrete from the moment it first comes in contact with water during the fresh state formation. These cracks can be associated with chemical, physical or mechanical factors [2]. Cracks are one of the most common pathological evidence present in cement-based materials [3]. Cracks occurrence can be directly associated with loss of durability and a consequent decrease in the lifetime of the material. It allows aggressive agents to access the concrete and compromises mortars and concrete elements. Many methodologies can be used to promote crack closure after its occurrence. Examples of these techniques include the repair by maintenance activities, * Corresponding author: yasminwtamimi@gmail.com as well as choices during the matrix design phase. Considering this, the concept of self-healing emerges, which can partially or fully restore the tightness of cementitious matrices, without the need for human intervention after the occurrence of the crack [4]. Self-healing approaches have been widely undertaken in both Brazilian and international scenarios. The studies focus on evaluating the cracks closure, promoted by superabsorbent polymers, by crystalline admixtures, by encapsulated or unencapsulated bacteria, by an internal vascular network system, among other possibilities. This study aimed to evaluate the self-healing phenomenon in mortars with three different commercial crystalline admixtures. Therefore, mortars were produced with crack induction at three days, and were evaluated by optical microscopy and ultrasonic wave propagation speed. The samples' environmental conditions were a humid chamber, wet and dry cycles in controlled and accelerated chambers. The total period of analysis was 45 days.  [5], classifies them into two categories regarding the absence (PRAN) or presence (PRAH) of hydrostatic pressure applied to the concrete. PRA is a shortening term to Permeability-Reducing Admixtures, and the letters N and H refer to Non-hydrostatic conditions and Hydrostatic conditions, respectively. These materials patents protect the chemical composition. However, it is known that they exhibit a hydrophilic behaviour, which generates insoluble deposits in pores and cracks of the hardened matrix, increasing the density of the C-S-H and hindering the water penetration [6].

Self-healing phenomena in cement-based materials
The self-healing phenomenon is any process developed by a material itself that is capable of restoring or improving its initially required performance. This characteristic is usually associated with mechanical resistance, stiffness, and water permeability. The selfhealing phenomenon can mainly occur in two different ways, autogenous and autonomous [4,7,8]: -Autogenous: the phenomenon occurs because of components in the matrix that could be present even when not specifically engineered to promote the healing after cracking. Characteristic examples of this category include the further hydration of anhydrous cement grains and the carbonation of hydrated compounds as well as crystalline admixtures and mineral additions; -Autonomous: specific, engineered materials are applied to the matrix design to promote cracking closure, that otherwise would not be in the matrix, such as encapsulated agents, vascular network systems and biologic agents. Several factors influence the self-healing phenomenon and are associated with the matrix chemical and physical properties and its environmental exposure conditions. Additionally, the concentration of calcium ions in the matrix and its surroundings significantly influences the volume of products and the speed of the self-healing phenomenon. Calcium ions are derived from the silicates hydration reactions, and their amount depends on the type of cement [9,10]. The crack's thickness also deserves special attention. There are observations of total crack closure by autogenous healing in cracks with 0.40mm [11]. However, greater thicknesses have been closed by autonomous healing in matrices with biological agents. The crystalline additives as a self-healing agent have been the subject of several studies. Many researchers are concerned that this class of materials contributes to the cracks closure, but further observations are required to verify its effectiveness under specific exposure environments [6,8].

Materials
Mortars were used as the cementitious matrices in this study. Thus, the research program consisted of four main compositions: a control matrix without crystalline admixtures, and three compositions with commercial crystalline admixtures X, Y, and Z, identified according to their main material: CONT (control, crystalline admixtures free), CRIX (crystalline admixture X), CRIY (crystalline admixture Y), and CRIZ (crystalline admixture Z). Portland Cement type V, 40R (CP V ARI), according to NBR 16697 [12], was used as a high-strength initial cement, which is equivalent to Type III by ASTM C150/C150M [13] and CEM 1 40R by . The crystalline admixtures used are commercial products purchased in Porto Alegre, Rio Grande do Sul, Brazil. The fine aggregate used was quartz sand, with a fineness modulus equal to 1.89, ABNT NBR 7211 [15], an apparent specific mass 2.62g / cm³, and water absorption 1.30% both according to ABNT NBR 16916 [16]. The mortars mix design comes from a proportioning study conducted for other works within the Self-Healing Group / NORIE / UFRGS and are presented in Table 1 [17,18].  [20], using five specimens for each evaluated combination (four compositions, accelerated chamber and air-conditioned chamber).

Cementitious matrices production and preparation of samples and specimens
Two types of samples were produced: specimens, which were 100mm in height and 50mm in diameter, used for mechanical and ultrasonic wave propagation tests; and samples, which were 20mm height and 50mm diameter, used for the optical microscopy analysis. Samples were prepared, as shown in Figure 1. A specimen with a height of 100mm and a diameter of 50mm was divided into 5 samples of 20mm height (step I). The upper and lower portions were discarded, and the middle samples were cracked (step III) by loading in the upper and lower extremities (F), producing a diametral compressive strength. All samples with cracks greater than 0.5mm were discarded, and the remaining ones were marked in six subsequent 8mm parts along the crack (step V). The last procedure delimited the region of interest in the cracked area for the image analysis. After casting, in the hardened state, the specimens were placed in a humid chamber until the crack opening age, at three days. Two phenomena are taking place at this age: the binder hydration and the chemical reactions from crystalline admixtures. The main focus of this study was to evaluate the efficiency of crystalline admixtures, regarding the superficial crack closure and the matrix densification, evaluated by microscopy analysis and ultrasonic wave propagation tests, respectively. Additionally, two exposure environments were evaluated. After the crack opening age, the specimens and samples were exposed to six wet and dry cycles, for a total of forty-two days. In these cycles, all specimens and samples were submerged into a saturated Ca(OH)2 solution during the wetting period (3 days). This approach was necessary to prevent calcium hydroxide (from the cement hydration and crystalline admixtures chemical reactions) from leaching to the ambient. During the drying period (4 days), all compositions (CONT, CRIX, CRIY and CRYZ), were divided into two groups, each with two samples (h= 20mm, Ø= 50mm) and five specimens (h= 100mm, Ø= 50mm). The first group called CLI, stayed in an air-conditioned chamber (temperature of 23±2°C, RH = 60±3%, natural atmospheric level of carbon dioxide concentration), while the second group (CAR), stayed in an accelerated chamber (temperature of 20±2°C, RH = 70±2%, and 5% of carbon dioxide concentration). After the drying period and immediately before their immersion in a new wetting time, the optical microscopy analysis and the ultrasonic wave propagation speed tests were carried out. The purpose of this procedure was to obtain a uniform saturation degree between cement-based matrixes.

Ultrasonic wave propagation tests
Five specimens determined the ultrasonic wave propagation speed behaviour for each composition in each exposure environment, at the crack opening age (t3) and at the end of analysis at 45 days (t45). All specimens were placed on a testing template to take all measurements at the same point over time. Three successive readings were taken, at each age, for each specimen, and the average of the results for each point described its ultrasonic wave propagation speed. All ultrasound wave propagation tests were performed according to the NBR 8802 [21] specifications. The frequency of the 150 kHz transducers used in the analysis was according to the equipment manufacturer's specifications. Additionally, a coupling gel was applied to the specimen and to electrodes interface to improve the wave transmission and reception capacity.

Cracking closure evaluation by microscopy analysis and images processing
All samples used for microscope analysis were sectioned into six equidimensional images. One image is 8mm long, and the composition to describe an entire crack is made from six images, with a total width of 48mm. The images were taken with a Zeiss Stemi 508 microscope, with 250x maximum magnification, with an identical light exposure condition for all images at the acquisition time. The light parameters were standardized with a lighting chamber, in which 24 white LEDs were arranged above the samples, parallel to the interest region. The crack monitoring by optical microscopy occurred in two moments: in the crack opening age, day 3 (t3), and at the end of the analyses, cycle 6, 45th day (t45). The area of the cracks was quantified with ImageJ software. The procedure consists of arranging the images initial mosaic (A) in grayscale (B, 8 bit), applying the Limit tool (C), and finally erasing pixels that are not covered by the crack extent and are not the focus of interest. Thus, all black pixels that remain describe the crack by the image's histogram, Figure 2.
It should be noted that for determining the values assigned in the application of the Limit tool, the water saturation of the matrix and the samples lighting condition at the imaging time are fundamental parameters to establishing a correlation in this type of analysis.
Considering the total sample area does not change over time, and as a decrease in the cracked area is expected, the ratio between the crack pixels and the total sample surface describes a crack magnitude index.

Compressive strength
Control specimens for all compositions (CONT, CRIX, CRIY, CRZ), were kept in a humid chamber (CUM) during the entire analysis period, and Figure 3 shows their mechanical behaviour at 3 days (crack opening day) and 45 days (end of the analysis period). The compressive strength of the different compositions evaluated, after the wetting and drying cycles, in accelerated and air-conditioned chambers, at the 45 days age, is shown in Figure 3. Compared to other exposure conditions, results showed that the accelerated environment, with 5% CO2 content, provided an advantage to mechanical strength development, as expected. As for the samples kept in an air-conditioned chamber after a wetting period, for all materials used (CONT, CRIX, CRIY, and CRIZ) the mechanical strengths were higher than the average strength the specimens kept in a humid chamber. The phenomenon may be associated with a higher carbonation rate of these specimens because of the higher Ca(OH)2 content, due to the baths' saturation, in the capillary pores. Thus, the exposure environments are statistically different when the compressive strengths at 45 days are compared in an analysis of variance (ANOVA). Arndt [17], when evaluating concretes with crystalline admixtures, noted that the compressive strengths average at initial ages (3 days) were higher for control samples. However, at older ages, the behaviour becomes the opposite. Thus, concretes with crystalline admixtures do not present a higher loading capacity than control concrete in wetting (2 days) and drying (12 days) cycles. Although this study used mortar as a cementitious matrix, this behaviour was not observed. The specimens with crystalline admixtures showed greater compressive strength both at three days and at 45 days, similar to what Reddy and Ravitheja [22] observed. This behaviour can be explained because the crystalline admixtures were incorporated in the cementitious matrices as an additional material, and due to their compositions and particle size, they probably chemically reacted and acted as a filler material. Further, a slight reduction in the water/binder ratio (crystalline admixture as a binder material) can be pointed. Additionally, by comparing the efficiency of crystalline admixtures in each exposure environment and with each other in the same ambient conditions, a statistical difference cannot be established.

Ultrasonic wave propagation tests
Ultrasonic wave propagation speed tests can be used to compare mass density between different cementitious matrices. This evaluation of the CONT, CRIX, CRIY, and CRIZ mortars, at 3 and 45 days, is shown in Table  2.
Obtaining the ratio between the ultrasonic wave speeds at 45 (t45) and 3 (t3) days for the exposure environment in an air-conditioned chamber -CLI (45 days) / CLI (3 days) -a slight increase is observed, equivalent to 7.34; 4.49; 7.35 and 7.62%, respectively, for CONT, CRIX, CRIY, and CRIZ mortars. The same analysis for the accelerated chamber -CAR (45 days) / CAR (3 days)shows an increase of 7.00; 3.76; 8.62, and 8.91%, respectively, for the same parameter in the CONT, CRIX, CRIY, and CRIZ specimens. The X crystalline admixture for both exposure environments produces lower densification in the cementitious matrices, to 3 and 45 days age, but also a more expressive standard deviation. However, among the other mortars, a statistical difference, by the analysis of variance method (ANOVA) cannot be established. On the other side, the evaluation of the compressive strength, shown in Figure 3, the ultrasonic wave propagation speed allowed to differentiate the efficiency of crystalline admixtures in the exposure environments. What concerns the authors is that moisture is a predominant factor for the self-healing phenomenon to occur [6,23]. Cracking closure in this study may have been improved by the high wetting period, as hypothesized by Arndt [17]. The author describes that in the constant presence of moisture, most matrices with crystalline admixtures can significantly decrease the time to promote their repair.

Superficial cracking closure evaluation by optical microscopy
The cracking closure averages of two samples for each composition are shown in Figure 4. Closure of cracks occurred for both exposure environments; however, cracks with larger initial openings (> 0.30mm) were not fully healed, which is consistent with the findings of Li et al. [24]. Huang and Ye [9] cite that concentration of calcium ions is a primary factor for the healing products nucleation. Therefore, this factor will not be discussed because the samples, in the wetting periods, were exposed to a sufficient amount of calcium hydroxide, and it is assumed that the calcium ions presence is homogeneous on the sample's surface. The difference is the carbon dioxide available in each exposure environment during the drying cycle (much grater to the CAR samples), improving the calcite precipitation and, consequently, the self-healing behaviour [25]. Furthermore, since cement hydration reactions are still being developed at the age of 3 days (the crack opening age used) in addition to the performance of crystalline admixtures, crack healing can be partially produced by hydrated products from anhydrous cement grains [26]. Concerning the crack area reduction, in both exposure conditions (CLI and CAR), the crystalline admixture's efficiency follows CRIZ> CRIX> CRIY. In the control samples, at the cracking age, CLI samples were subjected to greater damage (greater crack opening in the initial instant), but an equivalence was observed in the cracking closure area for both exposure environments. Fig. 4. Cracked area to sample area ratio (%). (above) CLI, air-conditioned chamber, and (below) CAR, accelerated chamber, contain a central part of each crack at the crack opening age (3 days) and at the analysis end (45 days).

Conclusions
Despite the increase in the use of crystalline admixtures, the little knowledge about its composition limits the approaches and the better understanding of its performance. Mortars were evaluated by microscope analysis and ultrasound wave propagation tests. Those in wet and dry cycles in an accelerated chamber (CAR), 5% CO2 content, were the ones that had the highest mechanical strength at 45 days. A higher matrices densification was identified for the ultrasonic wave propagation tests, both for the air-conditioned and accelerated chambers, for samples with crystalline admixtures. Likewise, these products' performance in the closure of cracks has given CRIZ the most significant repair capacity in both exposure environments. The analysis techniques proved to be adequate for the proposed evaluation; however, special attention should be given to the stages of sample preparation, collection, and image processing to identify the healed area.