**Influence of Combined Action of Steel Fiber and MgO on Chloride Di**ff**usion Resistance of Concrete**

**Feifei Jiang 1,2,\*, Min Deng 1, Liwu Mo <sup>1</sup> and Wenqing Wu <sup>3</sup>**


Received: 24 March 2020; Accepted: 22 April 2020; Published: 24 April 2020

**Abstract:** To improve the chloride diffusion resistance and durability of concrete, a new kind of steel fiber reinforced MgO concrete (SFRMC) was made by adding steel fiber and MgO to concrete simultaneously. With steel fiber for load bearing and expansion limiting, MgO as the expander, SFRMC has both the advantages of fiber reinforced concrete and expansion concrete. The influence of steel fiber and MgO on the strength and chloride diffusion resistance of concrete was evaluated by splitting tensile test and chloride diffusion test. Mercury intrusion porosimeter (MIP) and scanning electron microscopy (SEM) were used to study the microstructure of SFRMC. The results showed that the combined action of steel fiber and MgO reduced the porosity of concrete and the chloride diffusion coefficient (CDC), which could not be achieved by steel fiber and MgO separately. In the free state, the expansion energy produced by the hydration of MgO made the concrete expand outwards. However, under the constraint of steel fiber, the expansion energy was used to tension the fiber, resulting in self-stress. In this way, compared to reference concrete RC, the tensile strength of SFRMC-1, SFRMC-2, and SFRMC-3 increased by 3.1%, 61.3%, and 64.5%, CDC decreased by 8.8%, 36.7%, and 33.1%, and the porosity decreased by 6.2%, 18.4%, and 20.6%, respectively. In addition, the SEM observations demonstrated that the interfacial transition zone (ITZ) between fiber and matrix was denser in SFRMC, which contributed to reduce the diffusion of chloride ions in the concrete.

**Keywords:** steel fiber; MgO expansive agent; split tensile strength; chloride diffusion resistance; porosity; interfacial transition zone

#### **1. Introduction**

Steel fiber reinforced concrete (SFRC) is a kind of high-performance concrete which distributes steel fiber uniformly in the matrix. Due to the restraint of steel fibers, the development of cracks is restrained, which makes SFRC have higher crack resistance and good toughness [1]. Owing to these excellent properties, SFRC has been widely used in port structures and protection structures, especially in those structures with high requirements for diffusion resistance and crack resistance [2–4]. However, these structures mentioned above are often in direct contact with chlorine ions. When chloride ions diffuse into the concrete, they will cause corrosion of the steel bar, leading to cracking and affecting the safety of the structure [5,6]. Therefore, it is of great significance to reduce the chloride diffusion coefficient (CDC) of SFRC.

In recent years; many researchers have studied the chloride diffusion resistance of fiber reinforced concrete. Vahid compared the effects of different kinds of fibers on the resistance of concrete to chloride tolerance. He found that polypropylene fiber reduced CDC; while adding steel fibers significantly increased CDC [7]. Guo added 0.15%, 0.30%, 0.45%, 0.60% of basalt fiber into concrete and tested the pore structure and the chloride diffusion resistance. The results showed that by adding basalt fiber, the chloride diffusion resistance was improved, the minor harmful pores (20–100 nm) were increased, and the serious harmful pores (>200 nm) were significantly reduced [8]. Yan explored the effects of different amounts of basalt fiber (BF; 0.05, 0.1, 0.15, and 0.2 vol%) on the chloride ion diffusion. He found that the inclusion of 0.05% BF accelerated the diffusion of chloride ions from the coral aggregate; while a dosage of BF above 0.1% could suppress the diffusion of chloride ions [9]. Mahyuddin explored the effects of different amounts of coconut fiber (0.6%, 1.2%, 1.8%, and 2.4%) on mechanical property and chloride diffusion. He found that by adding coconut fiber, compressive and flexural strengths were increased to 13% and 9%, respectively. However, in terms of durability, the chloride diffusion resistance was reduced [10]

Previous studies have confirmed that fiber can improve the mechanical properties of concrete [11–16], but there are different opinions on the chloride diffusion resistance. Some researchers think that fiber can enhance the chloride diffusion resistance of concrete, but the other researchers hold the opposite opinion. The main reason for different opinions is that fiber has both positive and negative effects in concrete. On the one hand, fiber reduces the shrinkage crack of concrete and improves the compactness of concrete. On the other hand, the fiber increases the number of ITZ between the fiber and the matrix, and the gap at the interface provides a channel for chloride diffusion. However, researchers have agreed that ITZ between fiber and matrix is the weakness of concrete, and the strength of ITZ has a great influence on the performance of concrete. To reduce the defect of ITZ, steel fiber and MgO were used simultaneously in this paper. The expansion of MgO was restrained by steel fiber, the void was filled, the porosity was reduced, and the strength of ITZ was improved.

From the view of reducing shrinkage cracks, the use of expansion agent is to use the expansion produced by hydration to compensate the shrinkage of concrete [17]. According to the differences in mineral composition of the expansive agent, the expansive agents currently sold in the market can be divided into five types: sulfide-aluminate expansive agent, lime expansive agent, iron powder expansive agent, MgO expansive agent, and compound expansive agent [18]. Among the above, MgO is one of the most excellent expansive agents. It has been widely used because of its stable hydration products, easy regulation of expansion performance, and no shrinkage in the later stage [19,20].

By studying the advantages and disadvantages of fiber and MgO, we found that steel fiber and MgO can be used together to give full play to the advantages of both materials. Different from SFRC, steel fiber not only plays a role in bridge connection, but also plays a role in limiting expansion in SFRMC. The three-dimensional distribution of steel fiber seriously restricts the expansion of MgO, changes the expansion mode of MgO, and makes the expansion from outward extension to inward extrusion, which greatly improves the compactness of SFRMC. Our previous research [21] has confirmed that SFRMC has excellent mechanical properties, but its durability has not been studied. In this paper, split tensile test, chloride diffusion test, mercury intrusion porosimeter MIP test, and scanning electron microscopy SEM test were conducted to study the combined effect of steel fiber and MgO on the tensile strength and chloride diffusion resistance of SFRMC. Through the analysis of pore structure, ITZ structure, the mechanism of SFRMC performance enhancement was studied. These data provided theoretical support for further application of SFRMC and opened up new application fields for MgO concrete and fiber reinforced concrete.

#### **2. Materials and Methods**

#### *2.1. Materials*

In this paper, the cement is Class 52.5 Ordinary Portland Cement (Shan Aluminum Cement Co., Ltd., Shandong, China). Fly ash was produced by Shenhua Huashou Power Co., Ltd in Shanghai, China. MgO was produced by Wuhan Sanyuan Special Building Materials co. LTD in Wuhan, China. The activity of MgO was 115 s and the specific surface area was 45.7 m2/g. Figure 1 shows the mineral composition of MgO. Table 1 summarizes the chemical composition of cement, Fly ash, and MgO.

**Figure 1.** X-ray diffraction pattern of MgO.



Continuously graded gravel with a size of 5–20 mm was used as the coarse aggregates. River sand with a fineness modulus of 2.94 was used as the fine aggregates. Steel fiber was produced by Zibo Shuanglian Building Materials co. LTD in China. The tensile strength of steel fiber was 520 MPa. The diameter of steel fiber was 0.58 mm and the length was 40 mm. Figure 2 shows the specific morphology of steel fiber, which is a wave shape. Table 2 shows the mix proportion of the concrete used in this paper.

**Figure 2.** Photograph of steel fiber.

#### *2.2. Experimental*

The specific process and analysis method of this paper are shown in Figure 3. Cube blocks of 150\*150\*150 mm were poured to study the split tensile strength and cylinder specimens of <sup>∅</sup>100×50 mm were poured to study the chloride diffusion resistance. After casting, the molds filled with fresh concrete were placed on the high-frequency vibration table to vibrate until no obvious bubbles escaped. After 24 h, the abrasives were removed and the concrete was transferred to the standard curing chamber (20 ± 2 °C , 95%*RH*).


**Table 2.** Mix proportion of concrete.

**Figure 3.** The specific process and analysis method.

#### 2.2.1. Split Tensile Test

The split tensile strength test was conducted with reference to China Standard GB/T 50081-2016 [22]. The main equipment was the SYE-2000 pressure testing machine with a maximum load of 2000 kN. The size of the specimens was 150 mm × 150 mm × 150 mm, and the mean value of the three specimens was taken as the tensile strength of the specimens (accurate to 0.01 Mpa). The loading rate was 0.06 Mpa/s. The tensile strength was calculated according to Formula (1). The test diagram is shown in the Figure 4:

$$f\_{\rm ls} = 2F/\pi A = 0.637F/A\_{\prime} \tag{1}$$

where *fts* is the tensile strength of concrete, in MPa; *F* is the failure load of the test piece, in N; *A* is the bearing area of the test piece, in mm2.

#### 2.2.2. Chloride Diffusion Test

Normally, the main methods for measuring chloride diffusion resistance are slow method and fast method. Among them, the slow method is too time-consuming and seldom used in practice, while the fast method can measure the chloride diffusion resistance of materials in a short time. As the most typical of the rapid method, the electric flux method was first proposed by the Portland Cement Association of the United States in 1981 and has become the most widely used in the world. The principle of the method is to use the electric field to accelerate the ion transport. The ions penetrate through the specimen under the action of DC power. In this paper, the chloride diffusion resistance

was evaluated by the flux transferred in accordance with the procedures described in ASTM C1202 [23]. The test device is shown in Figure 5.

**Figure 4.** Setup for split tensile strength test: (**a**) Schematic diagram of test; (**b**) Photo of test.

**Figure 5.** Setup for the chloride diffusion test.

Cylindrical specimens with a diameter of 100 mm and a height of 50 mm were used in the test. The solution was 0.3 *mol*/*l* NaOH and 3% NaCl. SX-DTL concrete chloride flux meter was used to record the electrification time and flux. The total electric flux of the concrete test block for 6 h was calculated according to Formula (2):

$$Q = 900 \times (l\_0 + 2 \, l\_{30} + 2 \, l\_{60} + \dots + 2 \, l\_1 + \dots + 2 \, l\_{300} + 2 \, l\_{330} + l\_{360}),\tag{2}$$

where *Q* is total flux through test block for 6 h (C); *I*<sup>0</sup> is initial flux (A), to 0.001A; *It* is flux (A) at time *t* (min), to 0.001A.

According to the Nernst–Plank equation, the relationship between chloride diffusion coefficient (CDC) and Q was established (Formula (3)):

$$\text{CDC} = 2.57765 + 0.00492 \times \text{Q.} \tag{3}$$

#### 2.2.3. MIP Test

The pore structure of concrete was analyzed by mercury intrusion porosimetry (MIP).When making the specimen, the concrete was knocked into several small test blocks with a length of about 2 mm to remove the coarse aggregate. The specimens were then soaked in anhydrous ethanol for 24 h to stop the hydration of cement and MgO. Then, the specimens were placed in a vacuum drying oven at 50 °C for drying for 12 h. Finally, the dried specimens were sealed in a plastic bag to prevent moisture from entering. During the test, the pore structure of two or three specimens was tested by MIP, and the effect of steel fibers and MgO on the pore structure was observed. By analyzing the experimental data of the pores, we could explain the variation law of tensile strength and compressive strength from the perspective of microstructure.

#### 2.2.4. SEM Test

SEM test (JSM-6510lA, Japan) was used to study the interface performance between steel fiber and substrate. All specimens used for SEM test were standardly cured for 28 days.

#### **3. Results**

#### *3.1. Failure Pattern of Concrete*

At the beginning of the split tensile test, the specimen was in the elastic stage, the surface was intact, and the deformation increased proportionally with the increase of load. With the continuous increase of the load, the vertical cracks appeared in the middle of the specimen, and then gradually extended to the bottom and the top, accompanied by the sound of splitting. At the end of test, the specimen was completely damaged. There were no obvious signs before the failure of RC and MC. After the cracks appeared on the surface, they extended rapidly and destroyed the concrete completely, showing obvious brittle failure. Finally, the specimen broke into two independent parts (Figure 6a,b). When the steel fiber was added into concrete, owing to the steel fiber limited the crack extension, the crack developed slowly before the destruction of SFRC, showing obvious ductility. Some fine cracks appeared around the long crack, and the specimen remained as a complete whole after the test, with only a small amount of block concrete falling from the surface (Figure 6c).

**Figure 6.** Failure pattern of concrete. (**a**) Reference concrete (RC); (**b**) MgO concrete (MC); (**c**) Steel fiber reinforced concrete (SFRC); (**d**) Steel fiber reinforced MgO concrete (SFRMC).

On the other hand, when the steel fiber and MgO were used simultaneously, the steel fiber restrained the expansion of MgO and generated self-stress, which improved the bite force between steel fiber and matrix, enhanced the strength of the interface between fiber and matrix, and significantly enhanced the crack resistance. Steel fiber made cracks became more evenly distributed. Instead of a single long and wide main crack in RC, the crack developed into many parallel fine cracks in SFRMC. After the test, the SFRMC surface was smooth without concrete spalling, and the specimen split but did not separate (Figure 6d).

#### *3.2. Combined Action of Steel Fiber and MgO on Split Tensile Strength*

Figure 7 shows the split tensile strength of specimens with different mix proportions. Compared to RC, MgO had a small impact on the strength of MC in the early stage, but it had a greater impact on the strength in the later stage. At 28 days, the tensile strength of MC was 4.4% higher than that of RC, and the improvement was mainly due to the hydration of MgO. Compared to RC, the tensile strength of SFRC had been significantly improved. The strength had increased by 29.7% at 28 days, which was the result of the steel fiber restrained crack extension.

**Figure 7.** Split tensile strength of concrete.

On the other hand, when MgO and steel fiber were used at the same time, the tensile strength of SFRMC continued to increase. Compared to RC, the tensile strength of SFRMC-1, SFRMC-2, and SFRMC increased by 3.1%, 61.3%, and 64.5%, respectively. The tensile strength increased with the increase of steel fibers, and the growth rate was greater when the fiber was less than 1%, and it would no longer increase significantly when it exceeded 1%. Therefore, considering the cost of materials, we recommend using 8% MgO and 1% steel fiber in constructions.

As we can see from the split tensile test, the concrete strength had been significantly improved under the combined action of steel fiber and MgO. There are two main reasons for the increase. The first is that steel fiber limited the crack extension. When the cracks came out, the tensile stress was transferred from concrete to steel fiber, which restrained the further extension of cracks. The second reason is that the expansion of MgO was restrained by steel fiber, which resulted in self-stress. In the free state, the expansion energy produced by the hydration of MgO made the concrete expand outwards. However, under the constraint of steel fiber, the expansion energy was used to tension the fiber, resulting in self-stress. In this way, the tensile strength of concrete is obviously improved.

#### *3.3. Combined Action of Steel Fiber and MgO on Chloride Di*ff*usion Resistance*

Table 3 shows the chloride diffusion coefficients (CDC) of concrete at 3, 7, 28, 60, and 180 days. As can be seen from Figure 8, with the increase in curing time, CDC of concrete with different mixing ratios gradually decreased, and the descending rate in the early stage (0–28 days) was relatively larger, while the rate in the later stage gradually decreased. The results showed that early curing of concrete had a significant effect on CDC. As the curing age increased, the cementitious materials continued to hydrate, the densification of concrete became higher, and the pores in concrete were gradually reduced, which was beneficial to reduce CDC and improve the durability. No matter how we changed the amount of fiber and MgO, the test results always obeyed this rule. Therefore, we suggest that in constructions, it is necessary to increase the time of early maintenance of concrete to improve the durability.

**Table 3.** Chloride diffusion coefficient of concrete specimens (10−<sup>9</sup> cm2/s).


**Figure 8.** Changes of chloride diffusion coefficient.

Figure 8 shows that using MgO separately had little effect on CDC. At 180 days, CDC of MC was only 1.2% lower than that of RC. Different from MgO, steel fiber had both positive and negative effects on the durability of concrete. The first was the positive effect. Steel fiber could effectively inhibit the generation of cracks and reduce CDC. The other was the negative effect. The addition of steel fiber increased the number of the interface between fiber and matrix and provided a channel for the transfer of chloride ions, which possibly improved CDC. It can be found from Table 3, when the steel fiber was less than 1%, the positive factor played a major role. CDC decreased with the increase of steel fiber. Compared to RC, CDC of SFRC, SFRMC-1, SFRMC-2 decreased by 5.5%, 8.8%, and 36.7%, respectively at 7 days. However, when the steel fiber continued to increase, the negative effect gradually dominated, and CDC of SFRMC-3 was larger than that of SFRMC-2, which indicated that when steel fiber was too

large, the negative effect caused by steel fiber was larger than the positive effect. Therefore, for SFRMC, the optimal content of steel fiber should not be more than 1%. In the later stage of curing (180 days), owing to the combined action of MgO and steel fiber, CDC of SFRMC-2 was 4.8 <sup>×</sup> <sup>10</sup>−<sup>9</sup> cm2/s, which was only 70.8% of RC, and the durability of concrete was obviously enhanced.

#### *3.4. Combined Action of Steel Fiber and MgO on Porosity of Concrete*

As an important part of concrete, the pore structure directly affected the mechanical properties and durability of concrete [24–26]. Therefore, the study on the effect of MgO and steel fiber on concrete is inseparable from the research on the pore structure. The pores can be classified into four categories: cementitious pores (<10 nm), transitional pores (10–100 nm), capillary pores (100–1000 nm), and macro pores (>1000 nm) [27]. Figure 9 shows the pore diameter distribution curve of concrete. Figure 10 shows the statistics of pore size.

**Figure 9.** Porosity distribution curve of concrete.

**Figure 10.** Statistics of concrete pore size.

Figures 9 and 10 show that compared to RC, the porosity of MC increased by 5.7%, indicating that the expansion of MgO is harmful to pore structure. The porosity of SFRC decreased by 1.5% compared to RC, and the number of large pores decreased by 15.9%, indicating that the steel fiber optimized the pore structure of concrete.

At the same time, when MgO and steel fiber were used together, the porosity continued to decrease. Compared to RC, the porosity of SFRMC-1, SFRMC-2, SFRMC-3 decreased by 6.2%, 18.4%, and 20.6%, respectively. In addition, the number of large pores was significantly reduced, cementitious and excessive pores were increased, and the connectivity of pores was reduced, which was also the main reason for the improvement of the chloride diffusion resistance of SFRMC.

#### *3.5. Combined Action of Steel Fiber and MgO on Interfacial Transition Zone*

To further analyze the principle of performance increase of SFRMC, the microstructure of concrete, especially the interfacial transition zone (ITZ) between fiber and matrix, was explored. The microstructure of ITZ with different mix ratios is shown in Figure 11. Figure 11 shows that in SFRC, there were obvious long and wide gaps in ITZ due to the shrinkage of matrix. Besides, tensile stress was produced because steel fiber restricted the shrinkage of matrix. In addition, when the tensile stress was larger than the tensile strength of concrete, the vertical crack was produced, leading to a reduction in durability.

Different from SFRC, when steel fiber and MgO were used at the same time, the wide gap at the interface disappeared. This change was mainly due to the extrusion and filling effect of MgO. Owing to the expansion under the constraint of steel fiber, self-stress was generated in ITZ (Figure 12). The self-stress made the biting force and friction force greatly increased, made ITZ become dense, and the boundary become fuzzy.

**Figure 11.** Interfacial transition zone (ITZ) between the steel fiber and matrix. (**a**) SFRC, (**b**) SFRMC-2.

**Figure 12.** Self-stress caused by expansion extrusion at the interface.

#### **4. Conclusions**

Considering the defects of MgO and steel fiber when they were used separately, MgO and steel fiber were used at the same time in this paper. The influence of combined action on split tensile strength, chloride diffusion resistance, and pore structure of concrete was discussed in detail. Through detailed experimental research and theoretical analysis, the following conclusions could be drawn:

(1) For concrete, MgO and steel fiber could be used at the same time. In SFRMC, the steel fiber played the role of bearing and limiting expansion at the same time. The combination of MgO and steel fiber improved the split tensile strength and chloride diffusion resistance of concrete significantly. Compared to RC, the tensile strength of SFRMC-1, SFRMC-2, and SFRMC-3 increased by 3.1%, 61.3%, and 64.5%, and CDC decreased by 8.8%, 36.7%, and 33.1%, respectively.

(2) When steel fiber and MgO are used simultaneously, the two have a synergistic effect and the performance of the concrete is greatly improved. The combined action of steel fiber and MgO reduced the porosity of concrete, which could not be achieved by steel fiber and MgO separately. Compared to RC, the porosity of SFRMC-1, SFRMC-2, and SFRMC decreased by 6.2%, 18.4%, and 20.6%, respectively.

(3) Through the analysis of pore structure, the mechanism of performance enhancement of SFRMC was studied. Owing to the expansion of MgO was restrained by steel fiber, MgO changed from outward expansion to inward extrusion, resulting in filling and compaction effect. At the same time, self-stress was produced in ITZ, which improved the interfacial strength between steel fiber and matrix.

**Author Contributions:** Conceptualization, F.J. and M.D.; methodology, L.M.; software, F.J.; validation, F.J., M.D., L.M. and W.W.; formal analysis, M.D. and L.M.; investigation, F.J., M.D., L.M. and W.W.; resources, F.J. and M.D.; data curation, W.W.; writing—original draft preparation, F.J. and M.D.; writing—review and editing, F.J., M.D., L.M. and W.W.; visualization, F.J.; supervision, F.J., M.D., L.M. and W.W.; project administration, F.J.; funding acquisition, F.J. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Science and Technology Development Plan of Suzhou (SNG201904), National Key Research and Development Plan of China (2017YFB0309903-01) and Transportation Science and Technology Planning Project of Shandong Province (2018B37-02).

**Acknowledgments:** The authors would like to thank Zhongyang Mao from Nanjing Tech University for his precious contribution in the experiments.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Study of Bond Strength of Steel Bars in Basalt Fibre Reinforced High Performance Concrete**

#### **Piotr Smarzewski**

Department of Structural Engineering, Faculty of Civil Engineering and Architecture, Lublin University of Technology, 20-618 Lublin, Poland; p.smarzewski@pollub.pl; Tel.: +48-698-695-284

Received: 22 April 2020; Accepted: 28 May 2020; Published: 29 May 2020

**Abstract:** The paper presents the study on bond behaviour of steel bars. It reports the research conducted on local bond strength of short length specimens in high performance concrete (HPC) and basalt fibre reinforced high performance concrete (BFRHPC). In this study, the basalt fibre volume content, concrete cover, bar diameter and rib geometry are the main parameters. Further important factors are the directions of the casting and loading. Determining the effect of aforementioned main parameters on the bond strength in test series is required, in order to design reinforced HPC structures. The study of local bond strength in HPC and BFRHPC with five different basalt fibre fractions included tests of seventy-two short length specimens, using two concrete cover and two diameters of steel bars with different rib face angles. For different ranges of BFRHPC strength, relationships for bond strength with respect to the splitting tensile strength were obtained. The bond strength increased with the splitting tensile strength and compressive strength of BFRHPC specimens with the 12 mm and 16 mm bar respectively. The bond strength of BFRHPC was lower for the bar with the greater distances between the lugs on the bar.

**Keywords:** bond strength; high performance concrete; reinforcing steel bar; basalt fibre

#### **1. Introduction**

High performance concrete (HPC) has compressive strength above 80 MPa and low permeability. A serious disadvantage of this composite is brittleness, which increases with strength. Numerous researchers have revealed that steel and polypropylene fibres [1–6] or combinations of these fibres [7–10] can reduce the brittleness of HPC and significantly improve its tensile strength and fracture toughness, as well as its ductility.

Basalt fibres produced from molten basalt rock have very good strength properties, as well as high resistance to fire and alkaline environment, and at the same time are relatively cheap. These characteristics determine their use in concrete [11,12]. However, the cost in addition to the chemical and mechanical properties of basalt fibres vary, depending of the type and quality of the raw material and the production process of these fibres [13]. Nevertheless, the above-mentioned characteristics and an environmentally friendly manufacturing process [14] might determine their application in HPC structures, instead of the most commonly used steel and polypropylene fibres.

The research into basalt fibre reinforced concrete (BFRC) has largely been focused on fundamental mechanical properties, such as compressive, splitting tensile and flexural strength, as well as fracture toughness [11,12,14–18]. Fibre reinforced high performance concrete (FRHPC) is widely regarded as an excellent composite for use in sustainable construction [19–21]. However, optimum fibre dosages vary significantly in different types of concrete, such as geopolymeric concrete [5], normal strength concrete [14,17], high performance concrete [16,19–21] or ultra-high performance concrete [22–25]. Although the basalt fibre reinforced high performance concrete (BFRHPC) has good tensile strength, the rather poor ductility of this composite [19–21] means that concrete structures with basalt fibres

should contain steel or other reinforcing bars to achieve the required performance and reliability. For this reason, a good interfacial bond of BFRHPC to steel bars is an important determinant of the interaction of these two materials in the structure.

Designing reinforced HPC structures with basalt fibres requires knowledge of the bond behaviour of steel reinforcement. Due to the high compressive strength, HPC elements have small cross-sections, and for this reason, the concrete cover and the failure mode are the most important parameters. When using fibres, knowledge about their impact on the bond behaviour and the possibility of replacing transverse reinforcement is very important [26]. In construction practice, the bar diameter, rib geometry, pouring direction of concrete and load direction are also relevant.

The bond strength between concrete and reinforcing bars in short lengths depends, among other factors, on the bar diameter, concrete strength and concrete cover. The theory of partly cracked thick cylinder proposed by Tepfers is the most comprehensive approach for determining the local bond, although it does not take into account the effect of deformation properties and the geometry of ribs in reinforcing bars [27]. This theory assumes that an uncracked concrete ring confines the cracked concrete and the reinforcing bar, and is resistant to bursting stresses radiating outwards from the bar at an angle of 45◦ to the bar axis. Soretz and Holzenbein [28] reported that the bonding action is different when the rib face angle is less than 30◦, and they presented the dependence of the deformation pattern on the behaviour of local bond stress-slip as a function of the relative rib area. Darwin and Graham revealed that a bar with a smaller rib face angle results in lower bond strength [29]. Hwang et al. [30] determined the impact of silica fume on the splice strength of deformed bars embedded in high strength concrete (HSC) and noted that the bond strength of a beam at 10% cement replacement by silica fume was 15% lower than the beam bond strength with no silica fume addition. Tests carried out on beams with a long embedded length showed that the bond stresses differ significantly along the length [31]. Esfahani and Rangan [32] studied the bond strength in HPC concrete and estimated the maximum bond stress based on the results of short length specimens in which the distribution of the bond stresses at failure was almost uniform. Local bond equations were proposed in this investigation that can be used to determine the maximum bond stress at the ends of the splices. Holschemacher et al. [33] showed that the high brittleness of the composite does not adversely affect the bond behaviour of steel bars anchored in ultra-high strength concrete. The binding stiffness increased, due to the high elastic modulus and compressive strength of concrete. Eligehausen et al. [34] examined different failure modes of anchoring bars (pull-out, pry-out and splitting) in normal strength concrete. It was found that the various failure modes were influenced by different parameters, such as confinement, relative rib area, fibre addition, concrete cover and casting direction. Alkaysi and El-Tawil [35] performed pull-out tests to characterize the bond strength of a non-proprietary ultra-high performance concrete mixture, and revealed that the bond strength decreases with increased embedment length and increases with higher volume content of steel fibres.

#### **2. Research Significance**

Designing reinforced concrete structures is relied on the bond relationship between the concrete and the steel bar. However, investigations of the interfacial bond behaviour between BFRHPC and steel bars have not yet been published, and remain a pressing need. The experimental program described in this article fills this gap and aimed to quantify the bond strength between HPC and steel reinforcement for several important design parameters. To assess the effect of different parameters on the bond, the maximum bond stress was required, which was estimated from the test results of short length specimens and an almost uniform distribution of the bond stresses at failure. A total of 72 bar pull-out tests were performed with parameters including average bar coatings (43 mm, 93 mm), nominal bar diameter (12 mm, 16 mm) and fibre volume content (0%, 1%, 1.25%, 1.5%, 1.75% and 2%). The present study is part of an investigation on the bond strength in high and ultra-high performance concrete with the application of hybrid fibres. Several relationships for the bond strength with respect to the splitting tensile strength and fibre volume fractions of BFRHPC were suggested. In addition, the equations

obtained in this study for determining the local bond strength can be used as the maximum bond stress at the ends of the splices.

#### **3. Materials and Methods**

HPC was made with general use Portland cement CEM I 52.5R (C) (CEMEX, Chełm, Poland), non-densified silica fume (SF) (Ironworks Łaziska, Łaziska Górne, Poland), tap water (W) and well-graded coarse and fine aggregates (CA, FA). The physical properties and chemical compositions of cementitious materials are summarized in Table 1.


**Table 1.** Chemical compositions and physical properties of used cement and silica fume.

Superplasticizer CX ISOFLEX 793 (CEMEX Admixtures GmbH, Salzkotten, Germany) (Sp) based on polycarboxylate ethers was used in high dosage. The 12 mm lengths of chopped basalt fibers (Holtex, Rzgów, Poland) (BF) were used. The basalt fibre bundles are flat, approximately 1 mm wide and made of 13 μm diameter filaments. Other characteristics of the fibre were as follows: density 2.7 g/cm3, modulus of elasticity 70 GPa, tensile strength 1700 MPa and elongation at break 2.5%. The fibres used in this study are shown in Figure 1a. The ribbed bars, on which the bond strength was determined, were B500SP with diameters of 12 mm and 16 mm. Axial tensile tests were carried out on three specimens for each diameter of reinforcing bars (Figure 1b).

**Figure 1.** (**a**) Fibres used in experimental work; (**b**) reinforcing bars after tensile test.

The measured yield strength, tensile strength, elastic modulus, ultimate tensile strain and minimum relative rib area of 12 mm and 16 mm were 605 MPa, 691 MPa, 203 GPa, 33%, 0.040 mm2 and 622 MPa, 716 MPa, 211 GPa, 27%, 0.056 mm<sup>2</sup> respectively. In order to measure the rib geometries, the reinforcing bars were longitudinally sliced. The rib face angle in the 12 mm and 16 mm bars was between 37◦ and 73◦, as well as 44◦ and 73◦, correspondingly. The measured distances between the ribs for the above-mentioned bars were from 3 to 6 mm and from 6 to 12 mm, respectively.

HPC specimens used in this investigation were cast with a 0.28 water-binder ratio. The silica fume had a specific surface of 15 m2/g. The basalt coarse aggregate (CA) had a maximum size of 5 mm, fineness modulus of 5.92 and compressive strength of 196 MPa. The fine quartz sand aggregate (FA) had a maximum particle size of 2 mm and fineness modulus of 1.84. The quantities used in the reference mixture were as follows: cement—670.5 kg/m3, silica fume—74.5 kg/m3, coarse aggregate—990 kg/m3, fine aggregate—500 kg/m3, water—210 L/m3, and superplasticizer—20 L/m3. The reference mixture HPC-B0 did not contain any basalt fibres. The following five contained fibres with an aspect ratio of *l*/*d* = 923 and percentage ranging from 1% to 2% were made with a reduced quartz sand amount equal to the weight of the added fibres. A summary of the all mixture types in this work is shown in Table 2.


**Table 2.** Test mixture proportions.

Two different types of deformed bars with nominal diameters of 12 mm and 16 mm were used in pull-out tested specimens. Seventy-two specimens in six series were produced and examined. All the specimens were cast parallel to the steel bar fixed vertically in the mould according to the pre-adopted position of the bar. Thirty-six specimens were made for each bar using six different basalt fibre contents of 0%, 1%, 1.25%, 1.5%, 1.75% and 2%. There were three specimens for each combination of *cy*/φ*<sup>b</sup>* and fibre content. The ratios of the side cover to the bottom cover of the concrete (*cx*/*cy*) were 1 and 2.1. The details of the pull-out test specimens and test set-up are shown in Table 3 and Figure 2.


**Figure 2.** (**a**) Test specimen details (dimensions in mm), (**b**) test set-up.

To investigate the bond behaviour of steel bars embedded in HPC and BFRHPC, a pull-out test was carried out. A pull-out load was applied using the MTS 319.25 servo-hydraulic testing machine (MTS, Eden Prairie, MN, USA) with a maximum flexural load capacity of 250 kN under displacement control, with a rate of 1 mm/min during testing. The LVDT was used to measure bond slip between the reinforcing bar and the HPC/BFRHPC at the loaded end. In all specimens, longitudinal cracking of the concrete cover occurred over the anchorage length of the reinforcing bars before failure. All the specimens failed due to splitting of concrete.

For each casting, seventy-two 100 mm × 100 mm × 100 mm HPC cubes were made to determine the compressive and splitting tensile strength. These strength tests were carried out using an Advantest 9 load-controlled universal press (CONTROLS, Milan, Italy) of 3 MN capacity. The average of six measurements was recorded as the compressive and splitting tensile strength of each HPC. Eighteen 100 mm × 100 mm × 500 mm beam specimens were made to set the flexural strength. The bending tests were subjected to three-point loading using the MTS 319.25 press (MTS, Eden Prairie, MN, USA). The beam specimens were supported on two rolls spaced at a distance of 300 mm and then were loaded at the midspan. The flexural tests used deflection as the control signal at a rate of 0.05 mm/min. The average of the flexural strength of three beam specimens was reported for each HPC.

#### **4. Results and Discussion**

#### *4.1. Hardened State Properties*

Table 4 gives the results of the compressive strength, splitting tensile strength and flexural strength for all HPC at 28 days, with standard deviations (SD) and coefficients of variation. The ratio of plain HPC strength to basalt fibre reinforced HPC strength are also presented.


**Table 4.** Mechanical properties of high performance concretes (HPCs).

The results indicate that the basalt fibre volume content influenced the compressive strength in fibre content range considered from 0% to 2%. The compressive strength of HPC was reduced by 15–22%, with an increase in the fibre volume fraction with comparison with that of HPC without fibres. This can be clarified by higher porosity in the fibre reinforced concrete due to the air infiltration during the mixing procedure of the fibres and concrete [36]. Additionally, it was observed that basalt fibres can absorb water from the cement paste, and reduce the HPC fluidity and compressive strength [19]. Although the addition of basalt fibres resulted in a decrease in compressive strength, they noticeably enhanced the toughness of the HPC under compression. The literature also suggests that the primary benefit of basalt fibres in concrete under compression is the change from a brittle failure mode to a more ductile mode [11,14,16,17].

It can be noticed that the tensile splitting and flexural strength of BFRHPC is always greater than these of HPC for the reason that basalt fibres act as crack-arrestors. Such an action may be assigned to the basalt fibres role in gradually filling micro-cracks, which leads to increased bonding in high performance concrete microstructure, and, consequently, to higher splitting tensile strength and flexural strength.

The addition of basalt fibre to normal strength concrete can significantly increase its splitting tensile and flexural strength. Jiang et al. [11] revealed that basalt fibres with 12 mm length and between 0.30% and 0.5% by volume content increased the flexural strength and splitting tensile strength of normal strength concrete by 9.58% and 24.34%, respectively. Kabay [37] obtained 10.3% increase in flexural strength when 2 kg/m<sup>3</sup> and 4 kg/m3 basalt fibres were added to normal strength concrete. Çelik and Bingöl [38] stated that the addition of basalt fibre at 0.20% volume content increase flexural strength of self-compacting concrete 11.58%. On the other hand, the highest splitting tensile strength was observed at 0.30% basalt fibre content, which increased by 12.78% according to the plain specimen.

The maximum values of split tensile and flexural strength were reached for HPC with 1.5% basalt fibre fraction. It was found that its tensile strengths compared to reference mixture increased by 50% and 106.67%, respectively. The decreases in average split tensile and flexural strength of HPC with 1.75% and 2% basalt fibre contents may be related to the variable distribution of fibres and the tendency to group them in these mixtures. Çelik and Bingöl [38] also observed a reduction in the flexural strength of self-compacting concrete with an increase in the addition of basalt fibres from 0.2% to 0.25%. They found that may be due to the fact that the low flexibility of basalt fibres caused the formation of larger voids in the concrete and deteriorated the distribution of fibres.

Concerning the tensile strength results, the flexural strength exhibited more variability and much higher results than splitting strength. Considering the CV values given in Table 4, it can be observed that the addition of basalt fibres did not increase the error indicators.

Çelik and Bingöl [38] studied normal strength concrete with basalt fibres with a volume fraction of 0.30% at 28 days. Based on micrographs analyses, it was noticed that there are voids between basalt fibres and cement paste, which reduce fibre-matrix bond and, as a consequence, rather brittle post-cracking behaviour occurs in basalt fibre reinforced composites. Branston et al. [14] also found that the density of cement on the basalt fibre surface decreases considerably after nine months. However, high performance concrete is characterised by a dense interfacial transition zone (ITZ), and behaves differently to normal strength concrete. Wu et al. [39] performed microstructure analyses of the steel fibre in ultra-high performance concrete matrix, calculated changes in porosity with the distance from the edge of the fibre and found that the porosity decreased as the distance from the edge of the fibre increased. The porosity around the fibre at 50 μm was 47% lower than the porosity at 10 μm from the fibre edge. On this basis, it can be assumed that the porosity of high performance concrete with basalt fibres will also grow with increasing fibre content, and smaller distances between the fibres, which consequently results in a decrease in HPC strength.

#### *4.2. Bond Strength Versus Fibre Content Relationship*

The average values of bond properties are summarized in Table 5. The bond strength was calculated as the achieved pull-out force divided by the initial surface area of the embedded bar by means of the relation *T*<sup>b</sup> = *Fmax*/πφ*ble* where *Fmax* is the maximum force in the bar at bond failure in the specimen. It is important to note that splitting failure mode did represent a bond failure in all the HPC specimens.

Table 5 shows the increase in nominal and normalized bond strength with increasing basalt fibre volume content up to 1.5%. This is caused by that since the deformed steel bar provides bonding through interlocking between the rebar lugs and the surrounding concrete, the bond strength increases with increasing strength of the surrounding high performance concrete [40]. In addition, the average displacement at peak load increased when the basalt fibre content increased, similar to the bond strength. The average normalized bond strength from all tested series was obtained as 1.2. The fibre content had a significant impact on this value. It can be seen that the normalized bond strength was twice as high in concrete with 2% fibre content, as compared to plain HPC.

Figure 3 compares the bond strength versus the fibre content relationships using two different *cy*/φ*<sup>b</sup>* relations for two bar diameters φ*b*, namely 12 mm and 16 mm. The error indicator denotes the standard deviation.

**Figure 3.** Variation in bond strength with fibre content for different bar diameters φ*<sup>b</sup>* and relations *cy*/φ*<sup>b</sup>* (**a**) 12 mm, (**b**) 16 mm.

The comparison of the results in Figure 3a shows that, for different values of cy/φb, the bond strengths of the 12 mm bar increase with a growing basalt fibre content, and with an increasing concrete cover. The bond strengths of HPC-B0 were 11.5 MPa with the central position of the bar in the specimen and 10.5 MPa with the bar positioned halfway between the centre and the edge of the specimen and of BFRHPC provided an improvement at each volume fraction. The bond strength improvement of the 12 mm bar located axially in the specimens and pulled-out of BFRHPC ranged from 2.6% to 78.3% at the volume fractions from 1.0% to 2.0%, as well as form 12.7% to 43.1% for the eccentric pulled-out bar from BFRHPC at the same fractions. In contrast, the bond strength of the 16 mm bar falls when the fibre content is 1–1.25% (Figure 3b). The decreases ranged between 6.9% and 13.1% for an eccentric bar and reduced with higher fibre content. A similar downward trend in the range of 5.9–19.1% was maintained for the centrally located bar. Only for the 2% fibre volume fraction a slight 3.9% increase in the bond strength was noted. In the case of concrete without fibres, the bond strength was greater for the 16 mm bar in the range of 24.3–27.4%. This can be explained by the larger rib face angle between 44◦ to 73◦ of the 16 mm bar than in the 12 mm bar with a rib face angle from 37◦ to 73◦. The lower values of bond strengths of BFRHPC for the 16 mm bar can be clarified by the larger distances between the ribs from 6 to 12 mm, compared to the 12 mm bar, in which these distances are between 3 to 6 mm. At larger distances between the lugs on the bar, the short basalt fibres could be oriented parallel to the bar between the ribs causing faster bond failure occurring, as a result of shearing off and crushing the concrete. This effect was observed for the smallest concrete cover. This indicates that the bond strength of the steel bar depends on the HPC strength. The BFRHPC compressive strength used in this study was 15–22% lower, compared to the concrete without fibres, and the splitting tensile strength was 20–50% higher.


*Vf* = volume content of fibre, *fc* = mean value of compressive strength, φ*<sup>b</sup>* = bar diameter, *le* = embedded length, *cy*, *cx* = bottom and side concrete cover, *Fmax* = maximum pull-out load, τ*<sup>b</sup>* = bond strength, δ*<sup>m</sup>* = mean value of displacement at peak load, τ*<sup>b</sup> \** = normalized bond strength, τ<sup>∗</sup> *<sup>b</sup>* = τ*b*/ *fc*.

#### *4.3. Bond Strength Versus Splitting Tensile Strength Relationship*

Figure 4 presents the bond strength versus splitting tensile strength relations for test series, using two different types of bars within a limited range of basalt fibre percentages.

The error indicators denote the standard deviations both the bond strength (0.24–0.49 MPa) and splitting tensile strength (0.15–0.29 MPa). The given relationships were proposed for basalt fibre content between 0% and 1.5%, due to problems with proper fibre distribution at 1.75% and 2% fibre volume contents, which resulted in a significant reduction of 18.8% in splitting tensile strength of BFRHPC. As can be seen in Figure 4, the values of BFRHPC bond strength for test with the 16 mm bar test are higher in the range of 4.2–32.2% than the bond strength in the 12 mm bar for the 0–1% fibre volume fraction. On the other hand, with fibre contents of 1.25–1.5%, it can be seen that the 12 mm bars were characterized by higher bond strengths in the range of 4–14.3%. It can be reported that the bond strength for the 12 mm bar increases when the BFRHPC splitting tensile strength growths. In contrast, the bond strength for the specimens with the 16 mm bar is higher when the concrete compressive strength growing (see Table 4). This is probably due to different geometry of the ribs in the 16 mm bar, as well as a different orientation of the fibres around the larger diameter bar, which may cause local sliding surfaces resulting in less values of the bond strength. Therefore, the extent of the concrete crushing in front of the ribs decreased as the compressive concrete strength increased.

(b)

**Figure 4.** Variation in bond strength with splitting tensile strength for different bar diameters φ*<sup>b</sup>* and relations *cy*/φ*<sup>b</sup>* (**a**) 12 mm, (**b**) 16 mm.

#### *4.4. Bond Strength Versus Ratio of Concrete Cover to Bar Diameter Relationship*

The values for the bond strength of BFRHPC are shown in Figure 5 as a function of *cy*/φ*<sup>b</sup>* ratio. Linear regression leads to the relationships of the concrete cover to bar diameter ratio versus bond strength given in Table 6. The test data for two different bars are grouped together in the analysis.

**Figure 5.** Variation in bond strength with *cy*/φ*<sup>b</sup> ratio* for different basalt fibre volume content.

**Table 6.** Ratio of concrete cover to bar diameter—bond strength relationship of basalt fibre reinforced high performance concrete (BFRHPC).


In general, the bond strength values for the BFRHPC specimens improved with increasing the basalt fibre content and *cy*/φ*<sup>b</sup>* ratio. For the 12 mm bars and *cy*/φ*<sup>b</sup>* ratios of 3.67 and 7.83, the bond strength increased by 27% and 73.7%, respectively, when the fibre volume content increased from 1% to 2%. For the 16 mm bars and *cy*/φ*<sup>b</sup>* ratios of 2.63 and 5.75, specimens containing 1% basalt fibres showed, in sequence, 6.6% and 22.1% lower bond strengths than specimens with 2% fibres. This seems to confirm the regularity that the bond strength depends on the quantity of fibres available to bridge any cracks formed under loading. Comparing the average bond strengths obtained for the 12 mm and 16 mm embedded bars, while maintaining a similar concrete cover and 2% fibre content, it was observed that the bond strength is up to 29.7% higher for a bar with a smaller diameter. Similar trends were noted for the remaining fibre volume fractions 1–1.75%. It can be also concluded that suggested linear relations agree very well with the test results. The recorded bond strength values differed most at the lowest fibre content of 1% and 1.25%, especially at greater concrete cover, which justifies further research.

#### **5. Conclusions**

The objective of this study was to investigate the bond strength between basalt fibre reinforced high performance concrete and deformed steel bar. Simple pull-out tests were carried out at two different bar diameters, two concrete covers, two embedment lengths and six fibre volume contents. Seventy-two short length pull-out specimens were tested at 28 days. Based on this investigation, the following main conclusions were drawn:


**Funding:** This research was supported financially by the Polish Ministry of Science and Higher Education within statutory research project no. FN15/ILT/2020.

**Acknowledgments:** The financial support from the Polish Ministry of Science and Higher Education is greatly appreciated. The author would also like to thank the CEMEX Company for donating the cement and superplasticizer for this research.

**Conflicts of Interest:** The author declares no conflict of interest.

#### **References**


© 2020 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Inhibition of the Alkali-Carbonate Reaction Using Fly Ash and the Underlying Mechanism**

#### **Xin Ren, Wei Li, Zhongyang Mao and Min Deng \***

College of Materials Science and Engineering, Nanjing Tech University. Nanjing 210009, China; 201761100171@njtech.edu.cn (X.R.); 201762100015@njtech.edu.cn (W.L.); mzy@njtech.edu.cn (Z.M.)

**\*** Correspondence: dengmin@njtech.edu.cn; Tel.: +86-159-5059-1582

Received: 10 May 2020; Accepted: 3 June 2020; Published: 5 June 2020

**Abstract:** In this paper, fly ash is used to inhibit the alkali-carbonate reaction (ACR). The experimental results suggest that when the alkali equivalent (equivalent Na2Oeq) of the cement is 1.0%, the adding of 30% fly ash can significantly inhibit the expansion in low-reactivity aggregates. For moderately reactive aggregates, the expansion rate can also be reduced by adding 30% of fly ash. According to a polarizing microscope analysis, the cracks are expansion cracks mainly due to the ACR. The main mechanisms of fly ash inhibiting the ACR are that it refines the pore structure of the cement paste, and that the alkali migration rate in the curing solution to the interior of the concrete microbars is reduced. As the content of fly ash increases, the concentrations of K<sup>+</sup> and Na<sup>+</sup> and the pH value in the pore solution gradually decrease. This makes the ACR in the rocks slower, such that the cracks are reduced, and the expansion due to the ACR is inhibited.

**Keywords:** inhibit; alkali-carbonate reaction; fly ash; expansion; mechanism

#### **1. Introduction**

In 1940, Stanton [1] first discovered the alkali-aggregate reaction (AAR), which has attracted the attention of many researchers. As it is extremely destructive and difficult to repair, it has been called the "cancer" of concrete. The AAR is divided into alkali-silicate reactions (ASRs) and alkali-carbonate reactions (ACRs). The ACR is one of the main problems in the long-term durability of concrete.

According to the time sequence, the expansion mechanism of the ACR can be divided into three types. Firstly, Gillott [2] believed that the expansion is the result of an increase in the solid volume due to water absorption by the clay, where dolomitization only provides a way for clay to absorb water. Secondly, Tang and Tong [3–5] believed that, although the absolute volume of the solid phase of the alkali-dolomite reaction is reduced in theory, the rearrangement and crystallization of the reaction products in a restricted space causes the expansion and cracking of the aggregate, respectively, leading to concrete cracking. Thirdly, Katayama [6–8] believed that the ACR is the combination of harmful expansion caused by the ASR of microcrystalline quartz and harmless dolomitization. Dolomitization produces the brucite and carbonate reaction ring, but the expansion is caused by the ASR due to microcrystalline quartz. However, Chen, X [9] and Chen, B [10] used tetramethylammonium hydroxide (TMAH) as the curing solution to investigate the expansion characteristics caused only by the ACR, as TMAH does not react with SiO2. Their results showed that the ACR exists separately and contributes to the expansion. Huan Yuan [11], who also used TMAH as the curing solution, proved (by an expansion stress test and concrete microbar expansion test) that the alkali-carbonate reaction causes expansion.

The concrete durability problem caused by the AAR can be controlled by many methods, such as reducing the available alkali and adding a suitable amount of ash or chemical additives [12]. Supplementary cementitious materials (SCMS) have different effects in reducing the expansion due to

the ACR and ASR [13]. It is well-known that supplementary cementitious materials have a significant effect in inhibiting the ASR [14]. Based on the success of these materials, they have been used to prevent the deterioration caused by the ACR. According to the research of Alireza Joshaghani [15], fly ash and trass can reduce the expansion rate of the ACR. At 56 d, using 10%, 20%, and 30% fly ash can reduce the expansion rate of mortar bars by 47%, 95%, and 73%, respectively, according to the ASTM C1260 standard. According to the experimental results of concrete prisms, the inhibitory effect of trass on the ACR is slightly better than that of fly ash. In a long-term test, the optimal content of fly ash was 20% and the optimal content of trass was 30%. Shehata's [16] experiment demonstrated that the expansion of concrete prisms mixed with SCMs exceeded the threshold of 0.040% in two years and no SCM had a complete effect on the ACR in the long-term, although some types were more effective than others at reducing expansion.

Shehata [16] and Min Deng [17] have shown that using fly ash to effectively inhibit the ACR expansion of the highly reactive dolomite limestones from Kingston, Canada is difficult. However, whether the ACR of low-reactive dolomite rocks and moderately reactive dolomite rocks can be effectively inhibited has not yet been studied. At present, the related specifications stipulate that concrete works cannot use the ACR reactive rocks as aggregates, limiting the application of dolomite rocks. Research on fly ash on the ACR of low-reactive dolomite rocks and moderately reactive dolomite rocks is therefore necessary. This serves to play a guiding role in the engineering application of dolomitic aggregates.

The main purpose of this study is to examine the inhibition and mechanism of fly ash on the ACR. Although the impact of fly ash on the ASR has been well-documented, there has been little research on the impact of fly ash on the ACR. In addition, most of the studies have focused on highly reactive aggregates, such as the Kingston dolomite limestone of Canada, with no clear research on low-reactive and moderately reactive rocks. In this study, therefore, we focus on low- and moderately reactive dolomite rocks.

#### **2. Materials and Methods**

#### *2.1. Materials*

The used materials were (1) low-alkali Portland cement (Type II) obtained from the Jiangnan Cement Plant, Nanjing, China, with 0.54% equivalent Na2Oeq; (2) class F fly ash (FA) obtained from Henan; and (3) dolomitic limestones obtained from Baofuling Mountain, Shandong, China (BFL8) and Shuijingwan, Guizhou, China (SJW). The chemical compositions of the cement and fly ash are presented in Table 1. The chemical compositions of the dolomitic limestones are shown in Table 2.


**Table 1.** Chemical analysis of cement and fly ash (wt. %).

**Table 2.** Chemical analysis of aggregates (wt. %).


The microscopic appearances of the fly ash and cement observed with a scanning electron microscope (SEM; JSM-6510, JEOL Co, Tokyo, Japan) are shown in Figure 1. The fly ash is shown in (A) and the cement is shown in (B). We can see, from Figure 1, that the fly ash particles were rounded and most particles in the cement were irregularly shaped. The X-ray diffraction (XRD; SmartLab X-ray diffractometer, Rigaku Co., Tokyo, Japan) analysis results are shown in Figure 2. The XRD analysis of SJW is shown in (A) and BFL8 is shown in (B). SJW rocks mainly include calcite, quartz, dolomite and muscovite. However, BFL8 rocks only have calcite, quartz and dolomite.

**Figure 1.** Scanning electron microscope images of (**A**) fly ash and (**B**) cement.

**Figure 2.** XRD analysis of (**A**) SJW and (**B**) BFL8.

#### *2.2. Methods*

#### 2.2.1. Mortar Bars Test

A set of mortar bars (25 × 25 × 285 mm) were prepared with cement as well as dolomitic rocks with grain sizes of 0.125–4 mm, according to RILEM AAR-2 [18]. The alkali equivalent (equivalent Na2Oeq) of the cement was adjusted to 0.9%, the ratio of aggregate to cement was 1:1, and the water–cement ratio was adjusted to 0.32. The designs of the mortar bars, according to RILEM AAR-2, are shown in Table 3. All specimens were kept for 24 h in a moisturized condition for curing before demolding. Then, samples were transferred to 1 mol/L NaOH solution immediately and stored in sealed containers at 80 ◦C. When curing to a set age, remove the sample and measure the length, then calculate the expansion rate according to Equation (1):

$$\mathbf{P\_t = (L\_t - L\_0)/(L\_0 - 2\,\text{b}) \times 100\%} \tag{1}$$

where Pt is the expansion rate after t days of curing, in %; Lt is the test piece length after t days of curing, in mm; L0 is the initial length of the test piece, in mm; and b is the length of the nail embedded in the concrete, in mm.


**Table 3.** Mortar bars mix designs according to RILEM AAR-2.

2.2.2. Concrete Microbars Test

A set of concrete microbars (40 × 40 × 160 mm) were prepared with cement, as well as dolomitic rocks with grain sizes of 5–10 mm, according to RILEM AAR-5 [19]. The alkali equivalent (equivalent Na2Oeq) of the cement was adjusted to 1.5%. The ratio of aggregate to cement was 1:1, and the water–cement ratio was adjusted to 0.32. The designs of the microbars, according to RILEM AAR-5, are shown in Table 4. All specimens were kept for 24 h in a moisturized condition for curing before demolding. Then, samples were transferred to 1 mol/L NaOH solution immediately and stored in sealed containers at 80 ◦C. The expansion rate is calculated according to Equation (1).

**Table 4.** Concrete microbars mix designs according to RILEM AAR-5.


2.2.3. Concrete Prisms Test

A set of concrete prisms were prepared with cement, as well as dolomitic rocks with grain sizes of 4.75–19 mm, according to ASTM C1293 [20]. The alkali equivalent (equivalent Na2Oeq) of the cement was adjusted to 1.25%. The water–cement ratio was adjusted to 0.45. The designs of the concrete prisms, according to ASTM C1293, are shown in Table 5. All specimens were kept for 24 h in a moisturized condition for curing before demolding. Then, samples were transferred to 100% humidity at 38 ◦C. The expansion rate is calculated according to Equation (1).

**Table 5.** Concrete prisms mix designs according to ASTM C1293.


2.2.4. Measurement of the PH, Ionic Concentration and Pore Structure

Three sets of samples (20 × 20 × 20 mm) were prepared with cement and fly ash, where the cement was replaced with varying amounts of fly ash. The samples were separately cured in a 1 mol/L NaOH solution and saturated Ca(OH)2 solution at 80 ◦C. First, to determine the curing age, 5.0 g of cement paste was weighed. Second, the contents were transferred to a covered dish and broken up. Third, enough water was added to make up a volume of 200 ml. Fourth, the mixture was incubated at room temperature for 1 h, and stirred frequently. Then, the mixture was filtered with medium-quality filter paper into a 500 ml volumetric flask and washed with hot water. Finally, the K<sup>+</sup> and Na<sup>+</sup> concentrations of the pore solution were measured with a flame photometer (FP650, Aopu Co. Shanghai, China) and the pH value of the pore solution was measured with a pH meter (PHS-25, INESA Co. Shanghai, China).

Three sets of samples (20 × 20 × 20 mm) were prepared with cement and fly ash. The water–cement ratio was adjusted to 0.32. They were separately cured in 1 mol/L NaOH solution at 80 ◦C. To determine the curing age, the cement paste was taken and the porosity and pore size distribution were measured using mercury intrusion porosimetry (MIP, AutoPore W 9500, Micromeritics Co. New York, NY, USA). As shown in Table 6, four mix formulations of cement pastes were designed to comparatively investigate the influences of fly ash on the pH, ionic concentration, and pore structure of the cement pastes.


**Table 6.** Mix proportions of the cement pastes.

#### 2.2.5. Thin Section Petrography

Samples taken from the concrete microrods were cut into thin slices. Then, the slices were examined with an optical microscope to check for the presence of reaction products and expansion cracks derived from the aggregate. A polarized optical microscope with transmitted light (Optiphot-II Pol reflectometer, × 25–400, Nikon, Tokyo, Japan) and a stereomicroscope (BH-2, Nikon, Tokyo, Japan) were used for this purpose. The preparation of thin sections was carried out according to the "Thin section specimen preparation" section in [21].

#### **3. Results and Discussion**

#### *3.1. Discrimination of the Aggregate Alkali Reactive*

The results of the mortar bar test (according to RILEM AAR-2) and concrete microbar test (according to RILEM AAR–5) are shown in Figures 3 and 4, respectively.

**Figure 3.** The expansion of the mortar bars prepared according to RILEM AAR-2.

**Figure 4.** The expansion of the concrete microbars prepared according to RILEM AAR-5.

Figure 3 shows that the expansion of BFL8 and SJW at 28 d was 0.126% and 0.188%, respectively. At 28 d, their expansion was greater than 0.1%, which is indicated by the position of the dotted line in the figure. According to RILEM AAR-5, the aggregates of BFL8 and SJW both demonstrated alkali-carbonate reactions.

Figure 4 shows that the expansion of BFL8 and SJW at 14 days was 0.081%, 0.062%, respectively. The dotted line in Figure 4 is the threshold value for judging whether the aggregate demonstrates an alkali–silicate reaction. At 14 d, the expansions of BFL8 and SJW were less than 0.1%; therefore, according to RILEM AAR-2, the BFL8 and SJW aggregates had no alkali-silicate reactions

Based on Figures 3 and 4, we can see from the expansion rate that BFL8 has low reactivity and SJW has moderate reactivity. Furthermore, BFL8 and SJW only have alkali-carbonate reactions.

#### *3.2. The E*ff*ect of Fly Ash on Concrete Microbars in the Short Term*

#### 3.2.1. Alkali Equivalent (Equivalent Na2Oeq) of the Cement Adjusted to 1.5%

Figure 5 shows the expansion results of the 180-day concrete microbar test conducted on samples using different concentrations of fly ash, according to RILEM AAR-5, where the alkali equivalent (equivalent Na2Oeq) of the cement was adjusted to 1.5%. The expansion of concrete microbars prepared by BFL8 is shown in (A) and the expansion of concrete microbars prepared by SJW is shown in (B).

**Figure 5.** The expansion of the concrete microbars with fly ash: (**A**) concrete microbars prepared by BFL8 and (**B**) concrete microbars prepared by SJW.

Figure 5 shows that the expansion rates of BFL8 and SJW without fly ash were 0.126% and 0.188%, respectively, at 28 days. The BFL and SJW samples expanded about 0.398% and 0.618%, respectively, at 180 days. Compared with the concrete microbars mixed with 0% fly ash, the expansion of the concrete microbars mixed with 10%, 20%, and 30% fly ash and prepared with BFL8 decreased by 18%, 42%, and 51%, respectively, at 28 d, and by 17 %, 21%, and 36%, respectively, at 180 d. The expansion of the concrete microbars mixed with 10%, 20%, and 30% fly ash and made of SJW decreased by 29%, 43%, and 46%, respectively, at 28 d, and by 23%, 33%, and 39%, respectively, at 180 d. After curing in 80 ◦C 1 mol/L NaOH solution for 56 days, the expansion of the concrete microbars prepared with BFL8 mixed with 10%–30% fly ash was greater than 0.10%. After curing for 28 days in 1mol/L NaOH solution, the expansion of the concrete microbars prepared with SJW mixed with 10%–30% fly ash was greater than 0.10%. From this point of view, when the alkali equivalent (equivalent Na2Oeq) of the cement is 1.5%, 10%–30% fly ash cannot effectively inhibit the expansion of the low-reactivity BFL8 and the moderately reactive SJW.

3.2.2. Alkali Equivalent (Equivalent Na2Oeq) of the Cement Adjusted to 1.0%

When the alkali equivalent (equivalent Na2Oeq) of the cement was 1.5%, fly ash could not effectively inhibit the ACR. Therefore, we reduced the alkali equivalent (equivalent Na2Oeq) of the cement to study the effect when the alkali equivalent (equivalent Na2Oeq) of the cement was 1.0%, in order to determine whether fly ash can effectively inhibit the expansion caused by the ACR under such conditions.

Figure 6 shows the expansion results of a 240-day concrete microbar test conducted on samples using different concentrations of fly ash according to RILEM AAR-5, where the alkali equivalent (equivalent Na2Oeq) of the cement was 1.0%. The expansion of concrete microbars prepared by BFL8 is shown in (A) and the expansion of concrete microbars prepared by SJW is shown in (B).

**Figure 6.** The expansion of the concrete microbars with fly ash: (**A**) concrete microbars prepared by BFL8 and (**B**) concrete microbars prepared by SJW.

Figure 6 shows that the expansion rates of BFL8 and SJW without fly ash were 0.106% and 0.168%, respectively, at 28 days. These samples of BFL8 and SJW expanded by approximately 0.388% and 0.603%, respectively, at 240 days, and their expansion basically stabilized. As the content of fly ash increased, the expansion of the concrete microbars made of BFL8 and SJW gradually decreased. At 28 d and 240 d, the addition of 10%, 20%, and 30% fly ash reduced the expansion of the concrete microbars prepared with BFL8 by 35%–66% and 26%–75%, respectively. Furthermore, adding 10%, 20%, and 30% fly ash reduced the expansion of the concrete microbars prepared with SJW by 16%–66% and 19%–67%, respectively.

Based on Figure 6, we can see, from the expansion rate, that adding 10%, 20%, and 30% fly ash reduced the expansion. However, when adding 30% fly ash, the expansion of SJW still exceeded 0.1% at 90 d. For BFL8, when the alkali equivalent (equivalent Na2Oeq) of the cement was 1.0%, adding 30% fly ash controlled the expansion within 0.1%.

#### *3.3. The E*ff*ect of Fly Ash on Concrete Microbars in the Long-Term*

The expansion results of the one-year concrete prisms test conducted on samples using 30% fly ash according to ASTM C1293 are shown in Figure 7, which shows that fly ash reduced the expansion for all tested aggregates.

**Figure 7.** The expansion of the concrete prisms with fly ash.

Figure 7 shows that the expansion rates of BFL8 and SJW without fly ash were equal to 0.041% and 0.046%, respectively, at 360 days. In accordance with the expansion of the concrete prisms with 30% fly ash, using fly ash reduced the expansion due to the ACR, and thus the expansion rates of BFL8 and SJW decreased by 49% and 43%, respectively. The expansion of the concrete prisms prepared with 30% FA and made of BFL8 and SJW were 0.0206% and 0.0259%, respectively, at 360 d. The expansion rates of the concrete prisms were less than 0.040%. Therefore, adding 30% fly ash can effectively reduce the expansion rate due to the ACR.

#### *3.4. The Mechanism of Fly Ash Inhibiting ACR*

#### 3.4.1. The Effect of Fly Ash on PH Value

The pH value of the cement paste pore solution cured in saturated Ca(OH)2 at 80 ◦C for 28 days is shown in Figure 8, which indicates that fly ash can significantly reduce the pH value of the pore solution and delay the ACR reaction. The ACR reaction requires a pH value of 12, according to Tang and Den [22].

**Figure 8.** The influence of fly ash on pH in the cement paste pore solution (80 ◦C, Ca(OH)2).

Adding 30% fly ash significantly decreased the pH value of the pore solution. Based on Figure 8, the pH value of the cement paste pore solution without fly ash was 12.45, and the pH of the cement paste pore solution with 30% fly ash was 12.25 at 21 days. Many researchers [23] have asserted that the concentration of OH<sup>−</sup> plays an important role in the ACR and that reducing the concentration of OHcan reduce the damage to the aggregates and inhibit the ACR.

The main reason why fly ash reduces the pH is because it plays the role of a physical diluent and reduces the alkali concentration of the pore solution [13,24]. Over the long-term, the fly ash reacts with Ca(OH)2 to reduce the pH value.

#### 3.4.2. The Effect of Fly Ash on the Ionic Migration and Pore Structure

Figure 9 shows the influence of fly ash on the concentrations of K<sup>+</sup> and Na<sup>+</sup> in the pore solution after curing in 1 mol/L NaOH solution at 80 ◦C. The concentrations of K<sup>+</sup> is shown in (A) and the concentrations of Na<sup>+</sup> is shown in (B).

**Figure 9.** Effect of fly ash on the concentrations of K<sup>+</sup> and Na<sup>+</sup> in the pore solution (80 ◦C, 1 mol/L NaOH): (**A**) the concentrations of K<sup>+</sup> and (**B**) the concentrations of Na+.

Figure 9 reveals that, as the amount of fly ash increased, the concentrations of K<sup>+</sup> and Na<sup>+</sup> in the pore solution decreased. Regardless of whether the sample was pulverized or not and mixed with fly ash, the concentration of K<sup>+</sup> ions in the pore solution gradually decreases and the concentration of Na<sup>+</sup> ions increases with the increasing curing time. Increasing the amount of fly ash slows the migration rates of Na<sup>+</sup> and K<sup>+</sup> to the cement paste in the curing solution. The ionic migration rate of the cement

paste without fly ash was the fastest. The concentration of [K<sup>+</sup> + Na+] in the pore solution was close to 0.063 mol/L at 14 days, but the concentration of [K<sup>+</sup> + Na+] in the pore solution with 30% fly ash was only 0.055 mol/L at 56 days. There was still a significant difference between the cement paste with fly ash and cement paste without fly ash, which was not offset by the external alkali content. It can be seen that adding fly ash delays the ionic migration and exchange speed of the pore solution to the curing solution.

The pore structure is an important component of the concrete microstructure. Manmohan [25] divided the pores in concrete into four sizes: <4.5, 4.5–50, 50–100, and >100 nm. Mehta believed that only pores larger than 100 nm will affect the strength and permeability of concrete. We tested the effect of fly ash on the pore structure of the cement paste in the concrete microbars, in order to analyze the effect of fly ash on ionic migration. Figure 10 shows the pore volume and pore diameter distributions of the cement paste in the concrete microbars cured for 56 days in 1 mol/L NaOH solution at 80 ◦C. The porosity of cement paste in the concrete microbars is shown in (A) and the pore size distribution of cement paste in the concrete microbars is shown in (B).

**Figure 10.** Pore structure of cement paste cured in 1 mol/L NaOH at 80 ◦C: (**A**) the porosity of cement paste in the concrete microbars and (**B**) the pore size distribution of cement paste in the concrete microbars.

Adding fly ash increased the porosity of the cement paste, reduced the harmful pores larger than 100 nm and increased the amount of micropores, effectively refining the pore structure. The porosity increased from 8% to 10% when using 30% fly ash. According to Manmohan [25], adding fly ash makes the structure of concrete microbars more compact. Fly ash not only reduces the content of the large pores, but also increases the content of small pores; thus, the pore structure of the cement paste becomes more compact, the permeability of the cement paste decreases, and the ionic migration speed becomes slower.

#### 3.4.3. Polarizing Microscope and Stereomicroscope Analysis

The thin sections of the concrete microbars cured in NaOH solutions were observed by polarizing microscopy, as shown in Figure 11. From the figure, we can see the crack characteristics of the concrete prepared with the SJW and BFL8 aggregates in the NaOH solution. The cracks were created inside the rock and extended into the cement paste. We can see many different sizes of dolomite crystals distributed at the crack. Meanwhile, no ASR gel was found in the crack. Therefore, these cracks are expansion cracks, mainly due to the ACR.

**Figure 11.** Thin section of the concrete microbars cured in NaOH solution for 180 days: (**A**) concrete microbars prepared with SJW; and (**B**) concrete microbars prepared with BFL8.

The expansion cracks of the concrete microbars cured in 1 mol/L NaOH solution were examined by stereomicroscope analysis, as shown in Figure 12. From the figure, we can see that cracks appeared in the concrete microbars without fly ash. However, the concrete microbars mixed with 30% fly ash had only some fine cracks and even no cracking. This indicates that the addition of fly ash can significantly reduce the cracking in concrete microbars.

**Figure 12.** Expansion cracks of concrete microbars prepared with the dolomitic aggregate cured in NaOH solution for 56 days: (**A**) BFL8; (**B**) BFL8 with 30% fly ash; (**C**) SJW; and (**D**) SJW with 30% fly ash.

The above analysis shows that the dolomite in the rock had undergone a dolomitization reaction, and the cracks of the concrete microbars prepared with BFL8 and SJW were in dolomite-rich areas. Due to the fly ash refining the pore structure of the cement paste and reducing the migration rate of the alkali in the curing solution to the interior of the concrete microbars, as the content of fly ash increased, the concentration of K<sup>+</sup> and Na<sup>+</sup> and the pH value in the pore solution gradually decreased. This made the ACR in the rock slower, the cracks reduce, and the ACR expansion to be inhibited.

#### **4. Conclusions**

In this experiment, concrete microbars and concrete prisms made of dolomite aggregates, fly ash, and cement were used to systematically investigate the short- and long-term effects of fly ash on the ACR and to investigate the influence of fly ash on the expansion of dolomite aggregates. From the physical macro-measurements and microstructure analyses, the following main conclusions can be drawn:

Increasing the content of fly ash in concrete microbars and concrete prisms can considerably decrease the expansion rate due to the ACR. In comparison with adding 10% and 20% fly ash, concrete microbars prepared with 30% fly ash exhibited a greatly reduced expansion rate. According to the expansion rates of concrete microbars, when the alkali equivalent (equivalent Na2Oeq) of the cement was 1.0%, for low-reactivity aggregates (such as BFL8), adding 30% fly ash could inhibit the expansion due to the ACR. However, when adding 30% fly ash to moderately reactive aggregates, such as SJW, although the expansion rate decreased, it did not decrease to less than 0.1% and cracking still occurred.

The polarizing microscope and stereomicroscope analysis results indicated that the cracks were expansion cracks, and the expansion was mainly due to the ACR. Adding 30% fly ash could significantly reduce the cracking in concrete microbars.

Through the analysis of the K<sup>+</sup> and Na<sup>+</sup> concentrations, the pH value, the ionic migration of the cement paste pore solution, and the pore structure, the main mechanism of fly ash inhibiting the ACR is that fly ash refines the pore structure of the cement paste, and the alkali migration rate in the curing solution to the interior of the concrete microbars is reduced. As the content of fly ash increases, the concentration of K<sup>+</sup> and Na<sup>+</sup> and the pH value in the pore solution gradually decrease. This causes the ACR in the rock to slow, the cracking to reduce, and the ACR expansion to be inhibited.

**Author Contributions:** Conceptualization, X.R. and M.D.; methodology, X.R.; software, X.R.; validation, X.R.,W.L., Z.M., and M.D.; formal analysis, X.R.; investigation, X.R.; resources, X.R., W.L. and Z.M.; data curation, X.R.; writing—Original draft preparation, X.R.; writing—review and editing, X.R.; visualization, X.R.; supervision, M.D.; project administration, M.D.; funding acquisition, M.D. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was funded by the National Key Research and Development Project (2016YFB0303601-2) and the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD).

**Acknowledgments:** The authors gratefully acknowledge the assistance from Jun Wang, Pengcheng Yu and Huan Yuan from NJTECH, and the staff from State Key Laboratory of Materials-Oriented Chemical Engineering.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
