Effect of cyclic wetting-drying on tensile mechanical behavior and microstructure of clay-bearing sandstone

The understanding of the weakening mechanism of tensile strength of rock subjected to cyclic wetting-drying is critical for rock engineering. Tensile strength tests were conducted on a total of 35 sandstone specimens with different wetting-drying cycles. The crack propagation process and acoustic emission characteristics were obtained through a high-speed camera and acoustic emission system. The results indicate that the tensile strength is observably reduced after cyclic wetting-drying, and the extent of the reduction is not only related to the number of wetting-drying cycle, but also closely related to the clay mineral content of the sample. In addition, as the cycles of wetting-drying increase, the effect of each single cycle on tensile strength is getting smaller and smaller until becoming constant. Moreover, the crack initiation and penetration time is prolonged as the number of wetting-dry cycle increases, which indicates that cyclic wetting-drying weakens the rock stiffness and enhances the ductility of sandstone. Meanwhile, the acoustic emission characteristics during the experiment further conrmed this phenomenon. Furthermore, through the analysis of the microstructure and mineral composition of the samples with different wetting-drying cycles, it is concluded that the main weakening mechanisms of sandstones containing clay minerals are frictional reduction, chemical and corrosive deterioration. weakening mechanisms are frictional reduction, chemical and corrosive deterioration, including the swelling of clay minerals.


Introduction
Many rocks in nature and rock engineering are subject to cyclic wetting-drying, such as rock in pumping reservoirs, coastlines, exposed slopes, etc. [1][2][3] . The mechanical properties of rock are highly susceptible to deteriorate under the in uence of cyclic wetting-drying, which is the cause of many engineering disasters [4][5] . Therefore, understanding the effects of cyclic wetting-drying on mechanical behavior of rock is critical to rock engineering in reality. An impressive number of papers have been published in recent years addressing the subject of the effects of water-rock interaction on mechanical behavior of various rocks [6][7][8][9][10][11][12][13][14][15] . Previous studies have shown that the mechanism of water-rock interaction that leads to the weakening of strength mainly includes fracture energy reduction, capillary tension decrease, pore pressure increase, fractional reduction, and chemical and corrosive deterioration [16][17][18][19][20][21][22][23][24] .
In addition to the mechanical behavior of rock under uniaxial and triaxial conditions, substantial research efforts have recently been focusing on the effects of water-rock interaction on the mechanical behavior under tensile conditions [25][26] , especially under the state of cyclic wetting-drying. For example, the fracture behavior of rock in different water conditions has been investigated, and results show that water has weakened fracture toughness and stiffness of sandstone [27][28] .It was found that the cyclic wetting-drying has an in uence on the crack propagation, fracture toughness and stiffness of sandstone. Through the experimental research on arti cial gypsum, it is found that the tensile strength of wet gypsum is about 50% lower than that of dry gypsum, but the strength under different soaking duration is not much different, and it is also found that the effect of water-rock interaction has a great in uence on the development of crack [29] . For sandstone with a low clay mineral content, experimental result shows that the tensile strength is affected by the water state of the rock, but is probably not sensitive to the length of immersion, and that cyclic wetting-drying does not necessarily reduce the tensile strength permanently. The major water-weakening mechanisms are reductions in the fracture energy and the fraction coe cient [30][31][32] . As regards the limestone with clay minerals, it is found that the clay mineral content directly affects the weakening effect of water against tensile and compressive stress [33] .
Moreover, the effect of cyclic drying and wetting on the dynamic tensile strength of rock is discussed, and a decay model considering the loading rate and the in uence of cyclic drying and wetting was proposed [34][35] . Dry wet cycle can cause cyclic expansion and contraction of hydrophilic clay minerals in soil, induce expansion to repeatedly act on soil microstructure, resulting in fatigue damage, leading to sudden reduction of soil cohesion; matrix suction growth during dehumidi cation, pore wall expansion during moisture absorption, decomposition and loss of cement will reduce soil strength [36][37] .
Despite the extensive and outstanding research, most of which indicates that the cyclic wetting-drying weakens the tensile strength of rock, the mechanism of weakening has not still clear. Regarding sandstones containing clay minerals in particular, the effect of the number of wetting-drying cycles on the tensile mechanical behavior of the rock, and how the cyclic wetting-drying changes the microstructure of rock and affects the macroscopic mechanical properties, remain unclear. Therefore, in this study, the cracking and failure processes in tension of sandstone containing clay minerals which underwent different number of cycles was investigated using high speed camera and acoustic emission (AE). Meanwhile, mineralogical composition and microstructure of specimen underwent different number of cycles was analyzed using X-ray diffraction and SEM. Moreover, the mechanism of the in uence of cyclic wetting-drying on rock mechanical behavior is discussed.

Sample Preparation and Experimental Procedure
All sandstone samples in this experiment were collected from Dianping Mine in Lvliang City, Shanxi Province. In order to reduce the discreteness of samples, all samples were extracted from sandstone of the same stratum. According to the recommendations of the International Society for Rock Mechanics [38] , the samples were cut into discs of 50 mm in diameter and 25 mm in thickness. As shown in Fig. 1 (a), 35 samples in total were divided into 7 groups according to the number of wetting-drying cycles. The 7 groups of samples were respectively subjected to 0, 2, 4, 10, 20, 30, and 40 wetting-drying cycles. Each circle of wetting-drying treatment was achieved by oven-drying at 105ºC for 12 hours and air-drying at room temperature 25 ºC for 12 hours, immersing the samples in water for 24 hours until saturation [39] . In order to prevent the in uence of ions in the water on the properties of the rock, the soaked water is deionized. The basic physical parameters of each group of samples are shown in Table 1. We measured the density of 7 groups of wet saturated samples in the nal state after different cycles. The density of the samples is slightly reduced after multiple wetting-drying cycles, but the regularity is not obvious. It can be seen from Table 1 that the variation of the S-wave velocity with the times of wetting-drying cycles is not obvious, and the P-wave velocity increases signi cantly with the increase of the times of wetting-drying cycles. The wave velocity is measured under the wet and saturated condition after the dry and wet cycles. The effect of water on the acoustic velocity of rock is mainly to ll the subtle cracks and pores [40] , so it will increase accordingly. As for the porosity, before 10 wetting-drying cycles, the porosity increased with the increase of wetting-drying cycles, and the rule of change after 10 times was not obvious. wetting-drying cycles. As shown in Fig. 1 (a), there are mainly three targets of the experimental observation: (1) the stress-strain curve and the variation of the tensile strength with respect to the number of wetting-drying cycles; (2) the crack propagation process and the variation of the acoustic emission characteristics with respect to the number of wetting-drying cycles; (3) the microstructure and mineral composition of the sample variation with respect to the number of wetting-drying cycles.

Experimental Setup
The system diagram of the experimental setup is shown in Fig. 1 (b). It can be seen that the experimental system mainly includes the following parts: (1) a loading system; (2) a high-speed photography system; (3) an acoustic emission system. In the Brazilian splitting test, a vertical loading rate of 0.5 mm/min was applied to the samples. Throughout the experiment, the generation and propagation of crack was recorded in real time by the high-speed camera in front of samples. Meanwhile, the AE system was used to monitor the precursor of sample rupture. In this experiment, two AE sensors were attached to the upper and lower ends of the back of samples. Moreover, a sample was selected from 5 samples of each group, and the microstructure and mineral composition of this sample were analyzed with scanning electric microscope (SEM) (TESCAN-VEGA\\LMN Scanning Electron Microscope) and X-ray diffraction (Japanese TTR III Multifunctional X-ray Diffraction Instrument) experiments.

Effect of wetting-drying cycles on tensile strength of rock
The stress-strain curves subject to different number of wetting-drying cycles were shown in Fig. 2. Representative and average values are shown in Fig. 3.These are examples of samples from each group. The calculation of tensile strength of rock in splitting test is expressed by the following formula. The basic schematic diagram of Brazil split can be seen in Fig. 1 (b). According to the analytical solution of elastic mechanics of plane stress problem, the compressive stress at the center of the disk is only three times of the tensile stress, so the tensile failure of the sample is not the compression failure, and then the splitting tensile strength of the sample is calculated.
It can be seen that as the number of wetting-drying cycles increases, the stress-strain curve becomes more ductile, especially after 10 cycles. This indicates that after the repeated wetting-drying process, the stiffness of sandstone is signi cantly lower than that in the initial state. Moreover, it can be seen from Fig. 2 that after two wetting-drying cycles, the peak stress drops by about 50% compared with the initial state. After that, while the peak stress changes little with the increase of cycle numbers, the strain signi cantly increased. This phenomenon indicates that the rst few wetting-drying cycles have a greater impact on the strength of rock, while the latter mainly affects the stiffness of rock. Furthermore, the tensile strength of all samples under different wetting-dry cycles was shown in Fig. 3 and Table 2. It demonstrates that the sandstone in natural state features the highest average tensile strength at 4.842 MPa with the largest variance value at 2.891. Due to the in uence of water during the wetting-drying cycles, the tensile strength of the samples is reduced, and the variance value is lower than that in the natural state. Moreover, it can be found that the tensile strength of the sandstone is not only affected by the number of wetting-drying cycles, but also closely related to the clay mineral content in the sandstone. For example, the tensile strength of samples subject to 20, 30 and 40 wetting-drying cycles is greater than the three groups of samples subject to 10 cycles, which is mainly because of the low clay mineral content of these three groups. See Table 3 for speci c clay mineral content. At the same sampling location, the clay mineral composition of the sample is different. When the sample is cut into the standard test sample size, the dry and wet cycles are carried out under the same conditions. The clay content of the rst three groups is high, and that of the last three groups is relatively low.  In addition, reduction in tensile strength of sandstone subject to different number of wetting-drying cycles was shown in Fig. 4. Figure 4 (a) presents the reduction in tensile strength caused by cumulative wetting-drying cycles, and Fig. 4 (b) presents the reduction in tensile strength caused by a single wetting-drying cycle. The average cycle strength of each group of samples decreased under different cycles. It can be seen from Fig. 4 (a) that the reduction of tensile strength of sandstone with large clay content increases with the number of wetting-drying cycles, while that of sandstone with low clay mineral content is not. This is consistent with previous research results. For example, the number of wetting-drying cycles of sandstone with clay mineral content of less than 2% has little or negligible effect on tensile strength [4] [30] . However, for the sandstones containing clay minerals such as chlorite, the reduction in tensile strength increases with the number of wetting-drying cycles [27] [41] . Furthermore, the reduction in tensile strength caused by a single wetting-drying cycle is calculated using the following equation: where σ is the reduction in tensile strength, is the average tensile strength in nature, is the average tensile strength of sandstone subjected to n times wetting-drying cycles. As shown in Fig. 4 (b), as the number of wetting-drying cycle increases, the effect of single wetting-drying cycle on tensile strength is getting smaller and smaller until becoming constant. Moreover, the effect of a single wetting-drying cycle on tensile strength can be tted by a quadratic curve.

Effect of wetting-drying cycles on crack propagation process
In order to analyze in detail the effect of the number of wetting-drying cycles on crack propagation, the images on the front surface of specimens were captured using high-speed camera. The crack propagation process of sandstone with different wetting-drying cycles was shown in Fig. 5. From the perspective of qualitative analysis. It can be seen from Fig. 5 that the crack opening and penetration time is prolonged with the increase of wetting-drying cycles. This further indicates the great in uence of the number of wetting-drying cycles on the stiffness of sandstone. In addition, as the number of wetting-drying cycles increases, besides the primary crack, secondary cracks also appear in the crack process. Moreover, the crack initiation and penetration time of all samples are counted in Fig. 6 which shows that as the number of wetting-drying cycle increases, the crack initiation time is delayed accordingly. Meanwhile, it can be seen in Fig. 6 (b) that the crack penetration time increases with the increase of wettingdrying cycles, which indicates that the rate of crack propagation becomes slower as the number of wetting-drying cycles increases.
Time is the time from the beginning of crack shooting by high-speed camera to the conversion of frame rate corresponding to the throughout picture.
In addition, the AE energy rate and cumulative AE energy during the entire experiment was monitored. The variation of AE characteristics during the loading process was shown in Fig. 7. Similar to previous studies, the AE characteristics of sandstone throughout the experiment with respect to different wetting-drying cycles can be divided into four stages [42][43][44] . The rst is the existing micro-crack or pore compaction stage. During this stage the corresponding acoustic emission energy is rare. As the number of wetting-drying cycle increases, this stage tends to be extended. The second is the elastic deformation stage, in which the corresponding acoustic emission energy increases gradually and slowly. This stage is also extended as the number of wettingdrying cycle increases. The third is the crack propagation stage, in which the corresponding acoustic emission event increases sharply due to rock rupture. The fourth is the post-peak stage. As the number of wetting-drying cycles increases, the acoustic emission energy of the sample at this stage also increases accordingly. This indicates that the wetting-drying cycle enhances the plasticity and ductility of sandstone.

Effect of wetting-drying cycle on microstructure and strength weakening mechanisms
During the wetting-drying cycle, the microstructure of the sample will be altered due to repeated water-rock interactions. Figure 8 presents the SEM images of the sandstone with different number of wetting-drying cycles. Table 3 is included with the mineralogy after cycling. Please refer to the group number. The right column in Fig. 8 shows the overall microstructure change. For sandstone samples with clay minerals. It shows that as the number of wetting-dry cycle increases, there is no big change in large granular minerals such as quartz, but signi cant changes in clay mineral particles. Due to the action of water, the clay mineral particles gradually change from being massive, neat and dense to at, muddy and honeycomb. The right side of Fig. 8 is a partially enlarged picture. It can be seen that more inter granular cracks appear as the number of wetting-drying cycle increases. Especially when the number of wetting-drying cycles is over 10, there are micro-cracks between almost every two sandstone grains, and inter granular pore increases obviously. In summary, with the increase of the number of wetting-drying cycles, the secondary fractures and micro fractures increase, the morphological characteristics and internal structure of sandstone change to a certain degree, and the microstructure of sandstone gradually changes from being neat and dense to rough and disordered, until it becomes muddy, loose and slice particle structure. In addition, with the increase of inter granular porosity, some interstitial llings become muddy or aky, and the cementation between particles weakens. Bigger pores in the sample surface develop into micro-cracks. At the macro level, it is re ected in the lower strength parameters of the sandstone, which is similar to the prior experimental study [45] .
Previous studies have shown that the weakening mechanism of water on rock strength mainly includes fracture energy reduction, capillary tension decrease, pore pressure increase, fractional reduction, and chemical and corrosive deterioration. Regarding sandstone specimens in this experiment, the main tensile strength weakening mechanism is fraction reduction and physicochemical corrosion. It can be seen from the SEM of samples subjected to different wetting-drying cycles that as the number of wetting-drying cycle increases, the cementation between the particles weakens and the pores between the granularities increase, which seriously weakens the fractional effect. Moreover, it can be seen from Table 3 that after the sample undergoes the wetting-drying cycle, the feldspar minerals containing soluble ions such as potassium and sodium decreased or even disappeared due to hydrolysis reaction. In addition, the most important reason is that clay minerals such as montmorillonite and illite swell when they encounter water, and the original structure of sample is destroyed during repeated expansion to reduce its tensile strength.

Conclusions
This paper mainly discussed the effect of cyclic wetting-drying on tensile mechanical behavior of sandstone through experiment.
First, a total of 35 sandstone samples were divided into 7 groups, each of which was subjected to different times of arti cial wetting-drying cycles. Afterwards, the tensile strength of the sandstone specimens was measured by the Brazilian disc tests. Moreover, the crack propagation process and acoustic emission characteristics were obtained through high-speed cameras and acoustic emission monitoring systems, respectively. Based on the current experimental results, main conclusions were obtained as follows.
(1) The tensile strength of the sandstone subject to different times of wetting-drying cycles is observably reduced by 24% to 60%. (2) With the increase in number of wetting-drying cycle, the crack initiation and penetration time is prolonged gradually. Meanwhile, stress-strain curves show that the stiffness also remarkably decreases with the increase in number of wetting-drying cycle. These indicate that cyclic wetting-drying weakens the stiffness and enhances the ductility of sandstone. In addition, the acoustic emission characteristics during the experiment further con rmed this phenomenon.
(3) Regarding sandstones containing clay minerals in this work, the main weakening mechanisms are frictional reduction, chemical and corrosive deterioration, including the swelling of clay minerals.

Con ict of interest statement
The authors declared that they have no con icts of interest to this work. We declare that we do not have any commercial or associative interest that represents a con ict of interest in connection with the work submitted. Stress-strain curves of sandstone subjected to different times of wetting-drying cycles Tensile strength and clay mineral content of sandstone subjected to different times of wetting-drying cycles Crack propagation process of sandstone subjected to different times of wetting-drying cycles