Introduction

The slake durability, especially for weak rocks, is an important engineering property in relation with the stability of geotechnical projects, such as slope stability, foundation failure, and underground structure safety (Dhakal et al. 2002; Kayabali et al. 2006; Moradian et al. 2010; Liu et al. 2020). It represents the degradability of rocks resulted from the hybrid process of mechanical (e.g., abrasion and exfoliation) and chemical (e.g., solution, hydration, and oxidation) breakdown (Erguler et al. 2016; Fereidooni 2016; Li et al. 2019). Carbonaceous rock, which possesses poor engineering properties such as easy disintegration, low strength, and high compression, is widely distributed in southwestern China, such as Guangxi Province, and is ubiquitous in infrastructure construction projects (Luo et al. 2020a; Mo et al. 2023). After few days of continuous rainfall, many carbonaceous rock slopes experienced disintegration and landslide failure as shown in Fig. 1, leading to serious traffic interruption and property damage. This imposes great challenges to the safety service and effective protection of engineering projects built in or upon carbonaceous rocks. Therefore, the mechanical properties and mineralogical characteristics of carbonaceous rock have attracted great attention from engineering geologists and geotechnical engineers.

Fig. 1
figure 1

Distribution map of carbonaceous rocks with various stratigraphic units in Guangxi Province, China (based on Luo et al. 2020b)

Full size image

Carbonaceous rock can be roughly classified into three categories: carbonaceous mudstone, carbonaceous shale, and carbonaceous limestone (Mendonça Filho et al. 2013; Xing et al. 2018). Usually, the carbonaceous limestone performs best in mechanical characteristics and durability, followed by the carbonaceous shale and the mudstone (Holubec 1976). The slaking property of these carbonaceous rocks is closely related to factors such as mineralogical characteristic, joint structure, slaking fluid property, and environmental stress (Dick et al. 1994; Mutlutürk et al. 2004; Gupta and Ahmed 2007; Yagiz 2011). The previous studies have demonstrated that the durability behavior of carbonaceous rock in natural environment is a coupled mechanical-thermal-hydro-chemical issue (Marino and Osouli 2012; Chen et al. 2024). The complex external environment could result in a considerable change in the mineralogy and microstructure of carbonaceous rocks. Cetin et al. (2000) indicated that both short-term and long-term chemical weathering result in a negative impact on the slaking property of sedimentary soft rocks, which has been further demonstrated by Singh et al. (2005) for carbonaceous shaly rock exposed to different acidic conditions. Luo et al. (2020b) evaluated the geological and mineralogical characteristics of carbonaceous rocks in Guangxi, China, and indicated that the mechanical behavior is closely related to the total organic carbon content (TOC). Moreover, high temperature-induced moisture loss poses a serious threat to the integrity of carbonaceous rock and accelerates its mechanical property deterioration process, which in turn trigger slope failure or foundation instability (Yang and Zhang 2021). So far, very limited studies have comprehensively and quantitatively investigated the effects of mineralogical characteristic and organic content on slaking property of carbonaceous rocks, including slake durability index and fractal dimension. The deterioration mechanism of carbonaceous rock subjected to adverse conditions incorporating the hybrid influence of mineralogy and microstructure also has not been fully understood.

This study aims to obtain insight into the slake durability behavior of carbonaceous rocks subjected to wetting-drying cycles. A brief overview on the distribution and geological characteristics of carbonaceous rocks deposited in Guangxi, China was presented. A series of slake durability tests were conducted on carbonaceous rocks with varied mineralogical components to assess the evolution of slake durability index Idi and fractal dimension D with the test cycle. The effects of clay minerals content wc and TOC on Idi and D were studied, and their empirical correlations were established based on the obtained data set. In addition, changes in microstructural characteristics of carbonaceous rocks, including spatial configuration and surface morphology, were examined in terms of scanning electron microscopy (SEM) images. According to these results, a great effort has been made to explore the durability deterioration mechanism of carbonaceous rock incorporating the hybrid influence of mineralogical characteristic and microstructure development. The outcomes of this investigation are useful in both preliminary design and engineering protection of geotechnical projects build in or upon carbonaceous rock.

Materials and methods

Materials

According to the geological records and geological survey of Guangxi regional, carbonaceous rock is widely distributed as the main lithology or intercalation in the strata from the Qingbaikouan System to the Triassic System. The characteristic of lithologic assemblage and thickness of stratum are different from each group. Figure 2 shows the distribution map of carbonaceous rocks with different stratigraphic units in Guangxi Province, China (Luo et al. 2020b). According to the requires of highway construction project, this study focuses on the carbonaceous rock of the Devonian Luofu Formation (D2l). Due to deposited in different sedimentary environments, these widely distributed carbonaceous rocks exhibit significant discrepancy in mineralogical and lithological characteristics. In order to explore the mineral composition of this carbonaceous rock, a borehole sampling with depth of approximately 40 m was conducted at Nandan county, Hechi city, which is close to the target highway slope project site. Totally of 16 carbonaceous rock specimens corresponding to different borehole depths were tested by X-ray diffraction (XRD) analysis. The results indicated that this carbonaceous rock is mainly composed of quartz, clay minerals, calcite, dolomite, and other minerals. The average mass ratio of clay minerals is 35%, and its main component is illite. Interestingly, the other two common clay minerals: kaolinite and montmorillonite are almost undetected in these samples. The average mass ratio values of quartz and carbonate minerals are 30% and 25%, respectively, and a small amount (average value < 5%) of siderite and hematite is also detected.

Fig. 2
figure 2

Photograph of the typical carbonaceous rock slope failure in Guangxi Province, China (photograph credit: Jun-Hui Luo)

Full size image

The carbonaceous rock used for slake durability tests and deterioration mechanism analysis was collected from a highway slope project located in Hechi city, Guangxi Province, China, as shown in Fig. 2. This slope experienced a landslide after a week of continuous rainfall, and the carbonaceous rocks were gradually disintegrated in water (see Fig. 1). A total of five sampling points were selected at different heights of the slope, with two of them located at the top, one in the middle, and the rest two at the bottom, which can make the tested rock samples more diverse and representative and also to reduce the error level induced from the sampling location. The carbonaceous rock blocks were excavated from the remaining stable slope using a long-arm excavator, and then were sealed by a vinyl bag and coated on-site with wax to minimize moisture evaporation and artificial disturbance. The rock blocks were then placed into a wooden box with soft cushions to avoid vibration damage during transportation. Deionized water bought from the market was used as slaking solution to minimize the interferences from salt chemicals. The pH value and electrical conductivity of this water were 6.7 and 52 µS/cm, respectively, which were measured in the laboratory.

Sample preparation

Each fresh rock block was dewaxed and trimmed into cube shape with a side length of about 50 mm for moisture content test and density test. A 500 ± 50 g sample comprising of ten integrity rock lumps selected from the rest trimmed rock block was prepared for the slake durability test. Each rock lump was with uniform color, roughly spherical shape, 40 g to 60 g in weight, and no obvious fractures. A total of five samples from different sampling points were prepared, and were labeled as sample A, B, C, D, and E, respectively, in this study. The carbonaceous rock sample was ground into tiny powders for the X-ray photoelectron spectroscopy (XPS), XRD, and TOC analyses, where the particle size was less than 2 µm and 0.2 mm for XRD and TOC analyses, respectively. The detailed procedures of these three tests can be found in the relevant literatures (Gerin et al. 2003; Harris and Norman White 2008; Wang et al. 2008; Wang et al. 2023). For SEM analysis, one cuboid piece of approximately 5 mm×5 mm×3 mm in size with fresh sections was retrieved from the identical slake durability test samples before and after the testing procedures. The SEM samples were firstly cleaned with acetone and frozen by liquid nitrogen with a boiling point of -196 ˚C, and then placed into a freezing unit of -80 ˚C attached with a vacuum chamber for the sublimation of frozen water.

Testing methods

The moisture content tests were conducted as per ASTM D2216-19 (ASTM 2019), where the carbonaceous rock samples were placed into the oven with setting temperature of 110 ˚C for 24 h. The wax-coated method was employed to determine the density value of the test sample. The detailed testing procedures include: i) the fresh cubic sample was weighted immediately after the trimming process (m0), and oven-dried for 24 h, cooled in a desiccator to the room temperature (20 ± 2 ˚C), and then its dry weight (md) was determined; ii) the oven-dried sample was dipped into the liquid wax (about 60 ˚C) to generate a wax film coated on the sample surface with thickness of approximately 1 mm; iii) the wax-sealed sample was immersed into deionized water and weighted (m1). Subsequently, the sample was taken out, wiped off surface moisture with tissue, and weighted in the air (m2). The density value ρ of the tested rock sample was calculated by below equation:

$$\rho=\frac{m_0}{(m_2-m_1)-{\displaystyle\frac{(m_2-m_d)}{\rho_n}}}$$
(1)

where ρn is the density wax (ρn = 0.91 g/cm3 in this study). Quintuplicate samples were tested for moisture content and density, and the average values of the two parameters were investigated.

The slake durability tests were conducted on the carbonaceous rocks according to the standard testing method specified in the International Society for Rock Mechanics (ISRM 2007) and ASTM D4644-16 (ASTM 2016). The selected rock lumps were placed in a drum, dried in an oven with temperature of 110 ˚C for 24 h, and cooled down to the room temperature (20 ± 2 ˚C). Subsequently, the drum was assembled on the trough and was coupled with an electrical motor for rotation. The trough was then filled with deionized water to a level of 20 mm below the drum axis. The drum was rotated at a speed of 200 rpm for 10 min, after which the drum with remaining rock lumps was disassembled from the trough and subjected to oven-dried for 24 h, cooled, and weighed. During the test, the finer products of slaking passed through the drum and fell into the trough. The above-mentioned procedure was denoted as one slake durability cycle. These testing steps were iterated until the designed cycle was achieved. The cycle number was set as 6 in this study, which is the maximum value that the tested carbonaceous rock samples can resist. The slake durability index Idi of ith cycle was calculated by the following equation:

$$I_{\mathrm di}=\frac{m_{ri}}{m_d}\times100$$
(2)

where mri is the mass of oven-dried sample remaining in the drum after ith cycle (g); md is the mass of oven-dried sample before the test (g); and i is the cycle number. The value of Id2 was widely employed to evaluate the disintegration resistance of varied soft rocks (Sharma and Singh 2008). The particle size distribution test was conducted on the oven-dried samples after each slake durability cycle to calculate their fractal dimension D values.

The concept of fractal dimension was firstly developed by Mandelbrot (1975) in the specialized areas of physics and mathematics. It is customary to view dimension in the Euclidean sense, in which a perfectly straight line, an ideal plane, and an ideal sphere are respectively 1-D, 2-D, and 3-D features. Subsequently, this concept was introduced to quantitatively describe the microstructure irregularity (Liu et al. 2021), damage and fragmentation (Ding et al. 2021), and continuity (Piggott and Elsworth 1993; Yang et al. 2021) of rocks. The detailed review of the theory and determination methods for fractal dimension of various rocks can be found in the literatures (Xie and Wang 1999; Bagde et al. 2002). The relationship between the mass and particle size of the disintegrated rocks was a non-linear issue, which can be described by the Weibull distribution model as recommended by Fu et al. (2019) for the carbonaceous mudstones:

$$M(d)/{M_0}=1 - \exp [ - {(d/\sigma )^\theta }]$$
(3)

where M(d) is the cumulative mass of fragments with particle size less than d; M0 is the total mass of fragments; d is the opening size of sieve; σ is a constant dependent on the average particle size of rock fragments; and θ is a calculation parameter. When the ratio value d/σ ≤ 1, Eq. (3) can be simplified as the Tyler model, in which the fragment mass exhibits a linear relationship with particle size in the logarithmic system:

$$\left.\begin{array}{l}M(d)/M_0={(d/\sigma)}^\theta\\D=3-\theta\end{array}\right\}$$
(4)

where D is the value of fractal dimension. The higher the D value, the more fragments and the higher level in disintegration of the rock sample. Since the tested carbonaceous rock samples conformed to the Tyler model, the D value in this study was calculated from Eq. (4). Figure 3 shows a calculation example of fractal dimension D for sample E after the first slake durability cycle. It can be observed that lg[M(d)/M0] exhibits a linear relationship with lg(d/σ), and based on the linear regression method, the slope value θ of the fitted line is 0.506. According to Eq. (4), the fractal dimension D value is calculated as 2.494 (D = 3–0.506) for this example. In this study all test samples were subjected to the above-mentioned calculation procedures after each slake durability cycle.

Fig. 3
figure 3

Calculation example of the fractal dimension D for the sample E after one slake durability cycle

Full size image

The TOC tests were carried out on the pulverized samples (d < 0.2 mm) using a carbon/sulfur analyzer (LECO CS-200) with a precision of 0.5 wt% (wt% is the percentage of carbon by weight). These pulverized samples were prepared from the fresh rock blocks. Prior to the TOC test, the samples were digested with dilute HCl solution with HCl-to-H2O volume ratio of 1:7 for at least 2 h at temperature 60 ~ 80 ˚C to fully remove the inorganic carbon, as per the test standard of ‘Determination of Total Organic Carbon in Sedimentary Rock’ (GB/T 19145 − 2003). Based on the above-mentioned XRD results, only illite and clinochlore were included in the determination of clay minerals content wc in this study. The SEM analyses were performed using a Quanta 250 scanning electron microscope with magnification between 100 and 10,000 times to capture the microstructure images of the test samples.

Experimental results

Mineralogical and physico-chemical properties of carbonaceous rocks

The natural density and moisture content of the investigated carbonaceous rocks were ranged from 2.66 g/cm3 to 2.73 g/cm3 and from 0.99 to 1.49%, respectively, as shown in Fig. 4. The chemical element and composition of the carbonaceous rocks respectively obtained from XPS and XRD tests are summarized in Table 1. The tested carbonaceous rocks were mainly composed of four chemical elements: carbon (C), oxygen (O), silicon (Si), and aluminum (Al), and trace amount of potassium (K) and sodium (Na) were also detected. The XRD diffractograms of the selected samples are presented in Fig. 5. It can be observed that the mineralogical characteristics of the tested samples were different from each other. For instance, sample A was mainly composed of calcite, quartz, illite, and pyrite. While pyrite was not detected in sample C, and the quantity of calcite in sample C and sample E was much less than that of sample A. It is noteworthy that illite, one of the clay minerals, was found in all samples, which would result in an adverse influence on the engineering properties of carbonaceous rock when exposed to water intrusion (Zhao et al. 2017).

Fig. 4
figure 4

Natural moisture content and density of the tested carbonaceous rocks

Full size image
Table 1 Chemical elements and oxide chemistry of the carbonaceous rocks tested
Full size table
Fig. 5
figure 5

X-ray diffractograms of the selected rock samples

Full size image

Slake durability

Due to the similar slaking behavior of the tested rock samples, sample D was selected as an example to evaluate the changes in rock lumps during the cycles. Figure 6 presents the photos of carbonaceous rock sample D with different slake durability cycles. It can be observed that as the slake durability cycle increases, the carbonaceous rock lumps are gradually disintegrated into fine particles, indicating the integrity has been seriously damaged [see Fig. 6(a) to 6(g)]. After the completion of 6 cycles, the proportion of remaining rock lumps with particle size larger than 2 mm to the total mass is less than 50%, implying that most of the fine products pass through the mesh and fall into the trough, as shown in Fig. 6(h). Besides, as compared with other types of rocks, the most special phenomenon is that the slaking solution in the trough exhibits a turbid state after 6 cycles [see Fig. 6(h)], inferring some materials from carbonaceous rocks are dissolved or suspended in water during the slake durability test. To further quantitatively evaluate the slake durability properties of carbonaceous rocks, the evolutions of slake durability index Idi for the five rock samples from different sampling points are presented in Fig. 7. A widely accepted classification method of slake durability propose by Gamble (1971) was employed in this study, in which the slake durability of varied rocks was classified into 6 levels: very high (Id2≥98%), high (98%>Id2≥95%), medium high (95%>Id2≥85%), medium (85%>Id2≥60%), low (60%>Id2≥30%), and very low (30%>Id2). It can be observed from Fig. 7 that the Idi values of all samples present a decreasing trend with an increase in test cycles. In addition, there is an obvious difference between the tested samples in terms of slake durability properties. The slake durability performance of samples A, B, and C is greatly better than that of samples D and E. For example, after completion of 6 test cycles, slake durabilities of samples A and B fall into the medium high level, sample C is medium level, while samples D and E belong to the low level. This discrepancy can mainly be attributed to the mineralogical characteristic of carbonaceous rocks, especially for the content of clay minerals (Koncagül and Santi 1999; Erguler and Ulusay 2009).

Fig. 6
figure 6

Photos of carbonaceous rock sample D in the slake durability test: (a) carbonaceous rock slumps; (b) after 1 cycle; (c) after 2 cycles; (d) after 3 cycles; (e) after 4 cycles; (f) after 5 cycles; (g) after 6 cycles; (h) disintegrated carbonaceous rock in trough after 6 cycles

Full size image
Fig. 7
figure 7

Evolution of slake durability index Idi of the carbonaceous rock samples with slake durability cycles

Full size image

Figure 8 shows the relationship between the slake durability index and clay minerals content of the carbonaceous rocks. It is evident that the slake durability index decreases with the increasing clay minerals content, and this decreasing trend can be captured well by the following power law relationship (coefficient of correlation R2 = 0.96) based on the least square fitting method:

$$\left.\begin{array}{c}I_{d2}=311.4\cdot w_c^{-1.3}\\I_{d6}=4954.2\cdot w_c^{-1.3}\end{array}\right\}$$
(5)
Fig. 8
figure 8

Relationship between slake durability index and clay minerals content for the carbonaceous rocks in the double-logarithmic scale

Full size image

Labani and Rezaee (2015) have studied the influences of clay content on mechanical properties of organic-rich shales, and revealed that Young’s modulus value shows a decreasing trend with an increase in clay content. This demonstrates that clay minerals impose a negative effect on the durability of carbonaceous rocks, as shown in Fig. 8. It is a fact that clay minerals usually possess superior hydrophilicity and swelling performances as compared to other common rock minerals (Murray 2000). Furthermore, the swelling capacity of clay minerals is different from each other, resulting in the nonuniform tensile force in the rock matrix when immersed in water. The resultant tensile stress facilitates the development of micro-cracks, which in turn accelerates the progress of water intrusion. Subsequently, the carbonaceous rock sample is gradually broken and disintegrated from the macroscopic point of view. It is noteworthy that the fine (and/or clay) proportion in carbonaceous rocks would be eluted into slaking solution during the slake durability test [see Fig. 6(h)], which yields a negative impact on the stability of rock structure. Therefore, the slake durability of carbonaceous rocks is closely related to the content of clay minerals, especially for the ones with high swelling capacity, such as montmorillonite and illite (Torabi-Kaveh et al. 2022).

Fractal dimension

Figure 9 depicts the evolution of fractal dimension D with slake durability cycle for the five tested samples. It can be observed that D value increases evidently with an increase in slake durability cycle, and becomes almost constant as the cycle number exceeds 4. This phenomenon indicates again that the slake durability cycle deteriorates the integrity of carbonaceous rocks, and most integrity deterioration is generated in the initial stage of the cycling procedure (n < 4). Therefore, effective treatments should be carried out on carbonaceous rocks in the earlier stage of summer rainy season to prevent rapid durability deterioration in its field applications. Figure 9 also reveals that for a given slake durability cycle, the fractal dimension D value of each sample is different, and the order from low to high is found to be: sample A, B, C, D, E. This observation is consistent with the change trends of slake durability index Idi as shown in Fig. 7.

Fig. 9
figure 9

Evolution of fractal dimension D with slake durability cycle for different carbonaceous rocks

Full size image

Figure 10 shows the relationship between the fractal dimension D and slake durability index Idi. A clear decreasing trend of D with the increasing Idi is observed, which can be expressed by the following exponential law relationship (R2 = 0.84):

Fig. 10
figure 10

Correlation of fractal dimension D with slake durability index Idi

Full size image
$$D=-0.0033\cdot exp(I_{\mathrm di}/21.64)+2.82$$
(6)

This relationship sheds light on the estimation of slake durability by fractal characteristics of the carbonaceous rock in the preliminary design. It should be acknowledged that this empirical model was derived from a limited data set of five carbonaceous rocks investigated in this study. Further validation with more types of carbonaceous rocks is warranted to reinforce its applicability and to incorporate mineralogical characteristics.

TOC content

The presence of organic matter in carbonaceous rocks imposes an important influence on their durability performance (Breit and Wanty 1991; Tice and Lowe 2006). TOC content is a critical parameter required to evaluate the carbon content of organic matter (Hennebert et al. 2014). Earlier studies have indicated that soft rocks with higher TOC content exhibited an inferior performance of resistance against wetting-drying cycles (Wang et al. 2017). Thus, TOC content was introduced in this study to further explore the correlation of durability with chemical composition for carbonaceous rocks. Figure 11(a) depicts the relationships between TOC content and two slake durability indexes Id2 and Id6. It can be observed that TOC content is negatively correlated with Id2 and Id6, and two similar decreasing trends of Id2 and Id6 are detected. The sample with higher TOC content has a lower value of slake durability index, which is similar to the relation of Young’s modulus and tensile strength with TOC content reported in the literature for varied shales (Labani and Rezaee 2015; Li et al. 2024). For instance, TOC contents of samples C and D are 2.25% and 3.26%, respectively, while their Id2 values are respectively 96.9% and 82.8%, and Id6 values are 88.8% and 53.1%, respectively. A small increase in TOC content can lead to an obvious deterioration of slake durability for carbonaceous rocks.

Fig. 11
figure 11

Correlations of total organic carbon (TOC) content with engineering property indexes: (a) TOC versus slake durability index Idi and (b) TOC versus fractal dimension D.

Full size image

Figure 11(b) shows the relationships between TOC content and fractal dimension D at 2nd and 6th slake durability cycles. Linear increasing trends between TOC content and D were observed, with different slope values fitted for different slake durability cycles. The increased D value indicates an increase in rock fragmentsand disintegration level of the rock samples. This observation is consistent with the correlations of TOC content with Idi as shown in Fig. 11(a). Based on the simple mathematical statistics, some empirical correlations of parameter TOC content with engineering property indexes such as Idi and D were established. These empirical equations provide an intuitive understanding of the influences of organic matter on slake durability of carbonaceous rocks.

Microstructure characteristics

Figure 12 shows the SEM images of the selected samples before and after slake durability test. The natural carbonaceous rock presents a relatively stable microstructure with rock particles tightly bonded together. Besides, a considerable amount of pores and tiny cracks between rock aggregates are observed, which provide opportunities for water intrusion [see Fig. 12(a), (c), and (e)]. The spatial configuration characteristics of carbonaceous rocks are different from each other, for example, sample A exhibits a honeycomb structure with randomly distributed rock particles [see Fig. 12(a)], while a multi-layered structure with obvious cracks is detected in sample C [see Fig. 12(e)]. The discrepancy of natural carbonaceous rocks in terms of microstructure characteristics is one of the important reasons for the varied levels of slake durability. After suffering 6 slake durability cycles, a series of changes have been undergone in the microstructure of carbonaceous rocks: i) some crystalline-like substances were produced and attached on the particle surface, as marked by the red dotted circles in figures; ii) the original stable structure was damaged, which is characterized by the increase in pore volume and pore size (Dhakal et al. 2002; Ersöz and Topal 2024); iii) the tiny cracks were well developed and the spacing of the structural layers became larger [see Fig. 12(b), (d), and (f)]. The imposed wetting-drying intrusion during the slake durability test makes the microstructure of carbonaceous rocks change from ‘stable and dense’ to ‘loose and porous’, resulting in the deterioration of slake durability (Zhao et al. 2018). Furthermore, the microstructure stability of carbonaceous rock is closely related to its mineralogical characteristic. Among the five samples investigated, sample E is of the highest content of clay minerals and TOC. Its original multi-layered structure was almost destroyed and the resultant crystalline-like substances were widely distributed in the rock matrix including the particle surfaces and the pores [see Fig. 12(f)]. It should be indicated that SEM image only provides a visual and qualitative evaluation on microstructural characteristics, and additional quantitative analyses such as mercury intrusion porosimetry (MIP) and gas injection porosimetry (GIP) are suggested to quantitatively assess the microstructural changes of carbonaceous rocks.

Fig. 12
figure 12figure 12

Microstructure of varied carbonaceous rocks before and after slake durability test: (a) sample A, before test; (b) sample A, after test; (c) sample C, before test; (d) sample C, after test; (e) sample E, before test; (f) sample E, after test

Full size image

Discussion

The aforementioned results revealed the evolutions in mechanical durability and geometric morphology of carbonaceous rocks during the slake durability test. The intrinsic mechanism accounting for these evolutions was discussed from the point of view of mineralogy and microstructure of carbonaceous rocks. The XRD analysis results indicated that the investigated carbonaceous rocks are mainly composed of quartz, illite, calcite, dolomite, and other minerals [see Fig. (5)]. Illite is one of the common clay minerals with typical hydrophilicity and swelling properties. Under the condition of water intrusion, an uneven expansion was produced from the swelling of illite in rock matrix (Madsen and Müller-Vonmoos 1989), resulting in the accelerated development of cracks (Zengin and Erguler 2022). These well-developed cracks were beneficial to the next iteration of water intrusion. This is the main reason why the sample with higher clay minerals content exhibits a lower level of slake durability (see Fig. 8). Some soluble minerals/salts were gradually dissolved and then lost during the cycled water intrusion, which imposed a negative impact on the microstructure stability (see Fig. 12). As pointed out by Elango and Kannan (2007), the main chemical component of calcite and dolomite is calcium carbonate (CaCO3), which can undergo the following reactions with oxidative conditions:

$$\begin{array}{l}{\mathrm{CaCO}}_3+2\mathrm H^+\rightarrow\mathrm{Ca}^{2+}+{\mathrm H}_2\mathrm O+{\mathrm{CO}}_2\uparrow\\{\mathrm{CaCO}}_3+{\mathrm H}_2\mathrm O+{\mathrm{CO}}_2\rightarrow\mathrm{Ca}{({\mathrm{HCO}}_3)}_2\end{array}$$

The calcite and dolomite were gradually corroded in water, resulting in an increase in pore volume and pore size [see Fig. 12(b), (d), and (f)]. Therefore, the durability index of carbonaceous rocks presented a decreasing trend with an increase in slake durability cycle (see Fig. 7). As this chemical corrosion and dissolution further progressed, most of the reactive calcium carbonate and soluble substances were exhausted, and then the durability performance reached a stable state. Thus, it can be reasonably deduced that carbonaceous rocks with higher clay minerals content possess inferior performance of slake durability. The clay minerals content wc of samples A, B, C, D, and E are 20.36%, 21.15%, 22.04%, 29.76%, and 41.62%, respectively. These are responsible for the obvious discrepancies in slake durability index between sample A and E after 6 cycles [see Fig. (7)]. Additional chemical analysis, for instance SEM-EDS, XRD, X-ray photoelectron spectroscopy (XPS), and so forth, is suggested to be conducted to fully explore the chemical reactions in rock matrix and the genesis of newly produced substances. Generally, the presence of organic matter imposed a negative impact on the mechanical properties of rocks (Sopacı and Akgün 2015). The tested samples with high organic matter content were less resistant against the continuous mechanical impact and friction forces generated by the overturned drum during the slake durability test, and were easily crushed into rock fragments with similar particle morphology. Therefore, carbonaceous rock with higher TOC content exhibited an inferior performance of slake durability to that of lower TOC content (see Fig. 11).

As extensively demonstrated by the earlier researchers, the moisture loss-induced matric suction could be developed in the drying process for various geomaterials including soils, cemented soils, and rocks (Consoli et al. 2007; Hamid and Miller 2009; Taibi et al. 2009; Zeng et al. 2018). With the help of the resultant matric suction, the original cracks gradually penetrated into the center of rock lump and connected with each other, which resulted in a new path for water intrusion. Furthermore, the non-uniform tensile/shrink force can increase the gap between the crack walls, and even induce new cracks. In the next wetting process, matric suction decreased to a negligible level (about several tens Pa) along with the saturation of cracks and pores. Additional matric suction measurement is warranted to quantitatively evaluate its influence on slaking property of carbonaceous rock. This suggested work would be useful to further explore the deterioration mechanism of carbonaceous rock in complex external environments.

The slake durability is of great importance to assess the resistance of various rocks especially soft rocks to erosion and degradation, while the experimental preparation for the slake durability index measurement is tedious. Thus, many numerical methods such as Artificial neural networks (ANN), nonlinear regression technologies, and adaptive neuro-fuzzy methods have been employed to correlate slake durability index with other engineering property indexes, and the acceptable prediction results have been achieved (Kolay et al. 2010; Yagiz et al. 2012; Farashi et al. 2021). This study only discussed the relationships between three property indexes (i.e., wc, D, and TOC) and slake durability index for the carbonaceous rocks. Further investigation on slake durability of carbonaceous rocks needs to be carried out by using the numerical methods with more property indexes.

Conclusions

This paper investigated the slake durability properties of carbonaceous rocks with different mineralogical characteristics. The evolutions of slake durability index Idi and fractal dimension D with the test cycles were quantitatively evaluated. The empirical relationships among the property indexes: Idi, D, clay minerals content wc, and total organic carbon (TOC) content were established to account for the engineering performance stability of carbonaceous rock in adverse environments. The deterioration mechanism of carbonaceous rock was discussed from the perspectives of mineralogy and microstructure. Based on the findings portrayed in this study, the following major conclusions can be advanced:

  1. 1.

    The slake durability index Idi of carbonaceous rock tended to decrease rapidly with the increasing test cycle and then achieved a relatively stable stage when the test cycle exceeded 4. After completion of the target test cycles, the investigated carbonaceous rock samples are of obvious discrepancies in slake durability index, which was mainly attributed to the discrepancy of mineralogical characteristic. The slake durability index Id6 presented an exponential relationship with clay minerals content wc, in which the sample with a higher wc value exhibited an inferior performance of slake durability. Clay minerals content wc is a key factor contributing to the discrepancy of slake durability levels for different rock samples.

  2. 2.

    The evolution of fractal dimension D was contrary to that of Idi value during the slake durability test. The durability deterioration of carbonaceous rock was mostly taken place in the first 4 cycles. Therefore, effective treatment should be conducted at the earlier stage of wetting-drying cycles to minimize the durability deterioration of carbonaceous rock. The empirical estimation model of slake durability index with D value was developed to facilitatethe preliminary design.

  3. 3.

    Carbonaceous rock with lower TOC content possessed a better performance against disintegration. The clear correlations of TOC content with parameters Idi and D were observed according to the obtained data set in this study, which well captured the influences of organic matter on the slake durability of carbonaceous rock. The changes in microstructure, including the increase in pore volume, pore size, and layer spacing, and newly produced crystalline-like substances, provided evidence of the dissolution of minerals/salts in slaking solution and chemical reactions among minerals on the rock samples.

  4. 4.

    The mineralogy dissolution combined with moisture loss-induced matric suction in the wetting-drying processes facilitated the development of cracks within the carbonaceous rocks, which yielded the deterioration in slake durability performance. Additional micro-chemical analysis was recommended to be conducted to reveal the chemical reactions in rock matrix and the genesis of newly produced substances. The measurement of matric suction was warranted to further assess the variation in microstructure of carbonaceous rock in dry condition.