1. Introduction
Tensile strength is one of the soil’s basic material properties; it is too small and sensitive to measure. Although it is often ignored in strength tests of soil [
1], the tensile strength of soil plays an important role in soil failure, for example, in tensile cracks associated with collapses and landslides in the processes of slope instability, in the tensile failure caused by the toppling of ground buildings or other appendages, and in the ground cracks caused by internal stress in soil [
2,
3,
4,
5,
6]. The wetting and drying cycle of soil caused by rainfall is a common phenomenon in nature, which has a great influence on the tensile strength of the soil. Natural and engineering phenomena suggest that the tensile strength of soil should not be ignored as a little factor that does not affect its strength. Its mechanical behavior and characteristics of variation should be an important part of the strength characteristics of any given soil. Tensile strength comes from the bond and suction among soil particles, which is affected by the type of soil, dry density, and saturation [
7,
8,
9,
10].
Soil tensile strength testing can be divided into indirect methods and direct methods. Indirect methods include soil beam bending, axial fracturing, and radial fracturing [
11,
12,
13,
14]; they are mainly carried out with the assumption that the sample undergoes elastic strain before failure, which is suitable for brittle soils with high stiffness. However, for cohesive soils with plastic deformation during the test, the tensile strength needs to be obtained using a modified formula, and errors are difficult to avoid [
15]. The direct method includes uniaxial tension, triaxial tension, and so on. The two ends of the soil sample are fixed by the fixture, and the tensile strength of the soil sample is obtained by applying the tension. The friction between the test instruments and the size of the failure surface will affect the results [
8,
16]. Compared with the indirect method, the results of the direct method are more precise, with fewer assumptions. A large number of studies have been conducted to obtain the strength characteristics of soil samples, including an axial section of cylindrical unsaturated soil samples [
17], a cross-section of spindle samples [
18], and wedge samples evaluated by a self-made direct tensile tester [
1]; it has been found that dry density has a positive effect on the improvement of the tensile strength [
1,
2,
19]. Reinforcing the soil structure is an important method of improving tensile strength.
The effect of moisture content on the tensile strength of soils is different. For example, in some studies, the tensile strength decreases with an increase in moisture content [
1,
3,
9], while in some other tests, the tensile strength increases first and then decreases with an increase in moisture content [
18,
20]. The main reasons for this contradiction are as follows: (1) in addition to the limited moisture content of the soil affecting its strength, the dry and wet effect of water will also lead to the performance degradation in the soil [
21]; and (2) the effect of suction resulting from the action of water during the tensile test is not clear. Most of the existing studies on the tensile strength of soil focus on the specified dry and wet conditions. There is insufficient research on the evolution of tensile strength during the drying and wetting process [
22], and no deeper theoretical explanation is given on the formation source of tensile strength and the main control factors of changes during the drying and wetting process.
The variation in the tensile strength of the soil with the process of drying and wetting is essentially the destruction or strengthening of the soil structure [
23]. The influence of water on the strength of the structure is felt throughout the whole course of the change, and is due not only to the action of water chemistry on the cementing material, but also to the action of water tension between the soil particles [
24]. At present, the tensile strength of soil has been studied by using samples with fixed moisture content, but the formation process of the strength has been neglected, and the properties of the tensile strength have been poorly studied [
23,
25,
26].
Granite residual soil is widely distributed in tropical and subtropical South America, Africa, and Southeast Asia, and humidity affects its disintegration characteristics [
27,
28], which means this soil is sensitive to water. It is appropriate to use granite residual soil as the research object of the wetting and drying process.
In this paper, a self-made soil direct tensile strength tester was used to configure granite residual soil samples with different water contents and gradations. The variation law of tensile strength of soil samples in the process of constant humidity, wetting, and drying was studied. The source of tensile strength was analyzed by combining the basic physical properties and chemical composition of the soil. The evolution law of tensile strength and its microscopic mechanism was explained from the perspective of bond force and absorbed suction.
2. Materials and Methods
2.1. Testing Apparatus
This paper involves a situation in which the soil is in an extremely dry and wet state. It is unsuitable to test the tensile strength simply via an indirect or direct method with fixed ends and vertical tension. Therefore, a direct tensile strength test instrument for soil was made (
Figure 1). The instrument is mainly composed of three parts: a tensile mold, force sensor (Fusen Taike Sensing Technology Co. Ltd., Shenzhen, China), and loading device. To reduce the stress concentration effect, and ensure that the soil is always broken in the neck, the angle between the neck tilt of the tensile mold and the horizontal line is 20°.
The test operation proceeded as follows. The soil is first compacted in the mold, and then the displacement-controlled loading device slowly pulls one end of the mold to move slowly at a tensile rate of 2.4 mm/min until the soil sample breaks at the neck. The sensor records the peak tension when the soil sample breaks. At this time, the friction is very small and can be ignored. The slow tensile rate can also effectively eliminate the influence of displacement mutation on the results. Therefore, the tensile strength can be accurately converted according to the peak tension and fracture area.
2.2. Testing Material
The soil used is granite residual soil collected from Zhuhai City in southern China. It is widely distributed in South China, and is a commonly used engineering material in this area. The basic physical indexes tested are shown in
Table 1.
The dry density of the control sample was set to 1.38 g/cm3, and the sample with a height of 20 mm was made in three layers in the mold. Every 11.16% of saturation (moisture content of 4%) is a characteristic point until the soil sample is saturated.
Three parallel samples with the same moisture content were prepared, and the average value was taken as the tensile strength after testing. The dry or wet process is carried out only once before the specimen is tested. There are nine groups of particle gradations of soil samples. For different humidity process tensile strength tests, the soil particles used had a particle size less than 2 mm, and the particle size curve is denoted T1 in
Figure 2. For the tensile strength testing of different particle group samples in the constant humidity process, eight samples with different distributions of grain size were configured, and each sample was tested for eight or nine groups of remolded soil tensile strengths under different saturations. The samples can be divided into three categories: the T1~3 group refer to clay soil, T3~6 refer to sandy clay soil, and T7~9 refers to gravel clay soil.
The granite residual soil is tested using XRD for the mineral component. The main components of the test soil samples are quartz (78.0%), kaolin (7.0%), iron oxide (6.5), and other components (8.5%).
2.3. Operation Process
The wetting and drying process is common in natural surface soils. To simulate the whole process, distilled water was added to the sample until the target moisture content and soil was stirred evenly, then tested tensile strength of samples with different initial moisture content (
Figure 3).
Wetting process: To simulate water infiltrating from the surface of the soil and gradually penetrating the soil’s internal diffusion process, the dry density of the soil sample was controlled to be constant, and then the water was uniformly and gradually sprayed onto the surface of the sample with a sprayer. After being sprayed with a water equivalent until the target moisture content was reached, a tensile test was performed when the water on the surface of the soil sample was completely infiltrated. This is similar to the law of soil wetting under natural conditions, such as the processes of rainfall or irrigation. This test contains multiple sets of tensile tests with different target moisture content. After each test, some soil samples near the tensile fracture surface were uniformly cut, and the actual moisture content was measured using the drying method.
Drying process: To first simulate the process of water loss from the soil surface, for example, via the surface drying process after rainfall, then internal water loss, the water content of each part was reduced from inside to outside. To conform to the drying law of soil under natural conditions, a series of saturated soil samples in a slurry state were prepared, and the tensile strength was tested after drying on different target moisture contents under natural conditions. At the end of each test, soil samples near the tensile fracture surface were uniformly cut, and the actual moisture content was measured using the drying method.
The difference in water content reflects the difference in the filling of soil pores with water. The water content in this paper is set from the initial value to the saturation of the soil samples, and the water content is converted into saturation as a variable of the test process.
4. Discussion
4.1. Effect of Soil Structure on Tensile Strength
Gradation leads to a difference in mosaic-wrapping effect between coarse and fine particles of soil, which makes their tensile strength different.
A high fine-grained content can lead to stronger cementation provided by clay minerals and free oxides. At the same time, the hydrophilicity of fine-grained soil causes the soil to have stronger absorbed suction, which leads to higher tensile strength. The more coarse particles, the greater the pore volume of soil, and more loose the structure; it is difficult for absorbed suction and bond force to play a role in strength. However, it is an inaccurate view that a higher fine particle content can reinforce tensile strength. The tensile strength of the particle size composition is also lower than that of the single homogeneous soil, and it will be significantly higher than that of sand soil of the same quality.
SEM images further illustrate the influence of the distribution of grain size from a microscopic point of view. The relationship between tensile strength, affected by microstructure and particle size composition, can mainly be described as follows:
(1) When the content of large particles is low, large particles suspend in the matrix composed of fine particles to form a flocculation structure. At this time, the tensile strength mainly depends on the cementation between the clay and free oxide, and the absorbed suction provided by the liquid bridge between the clay. (2) With the increase in the content of large particles, there can be more direct contact between large particles, which constitute the skeleton of the soil. At this time, the proportion of fine particles is higher, and the pores between the skeletons are filled to form a denser structure. The tensile strength is mainly provided by the cementation of large particles connected by clay particles and free oxides, and the liquid bridge formed between particles. Due to the moderate proportion of particles with different particle sizes, the cementation between particles and the absorbed suction provided by the liquid bridge is integrated to achieve the best state, and thus the tensile strength is the highest. (3) With the continuous increase in the content of large particles, there are more contacts between large particles, which constitute the skeleton of the soil; meanwhile, the proportion of fine particles is low, the filling effect on the pores between the skeleton is weak, the overall bond force and absorbed suction of the soil are low, and the tensile strength is very low.
The intrinsic structural suction in the tensile deformation of the soil is very small, and the influence on the macroscopic tensile strength of the soil is not obvious. In addition, the intrinsic structural suction has little effect on the change in tensile strength during the drying and wetting process. Therefore, only the influence of variable structural suction (mainly bond force) and absorbed suction on tensile strength is discussed.
4.2. Effect of Soil Minerals on Tensile Strength
The bond force caused by mineral cementation is part of the structural suction. Among them, kaolin and iron oxide are the main components of cement, and quartz exists as a soil skeleton. This cement will gradually precipitate during the drying process, forming a stable bond between soil particles in the form of a ‘bridge’, and generating bond force.
The enhancement of tensile strength via a chemical reaction is mainly due to the chemical reaction of kaolin and iron oxide after water. The main form of iron oxide is free iron oxide between particles. A part of free iron oxide precipitates during drying and dehydration, encapsulates, and fills the soil skeleton to form iron cementation, and this process is partially reversible [
28]. Kaolin will prompt an ion exchange agglomeration reaction if in contact with water, which adsorb ions and impurities from the surrounding medium, but it has weak ion exchange properties, with a cation exchange capacity of about 2–5 mg/100 g.
Kaolin is strong for the connection and encapsulation of soil particles, and this ability is reversible, which caused the differences in tensile strength during the drying and wetting process. The mixture formed by the combination of kaolin and soil particles is mainly plastic colloid, which has good water stability and high cementation ability. When the soil sample turns from a saturated state to a dry state, the soil strength increases greatly. In the process from a dry state to a wet or even saturated state, a small part of kaolin softens again with water, and adsorbs on the surface of soil particles. The other part ensures that the soil still has tensile strength through its strong water stability and strong adhesion. This also explains the reason why the tensile strength of the soil during the drying process is much greater than that during the wetting process.
To verify the conjecture of the reversibility of kaolin cementation ability, the tensile strength test of a pure kaolin sample with a saturation of 20% was carried out (
Figure 10). After the test, the soil samples were directly put into the tray, poured into the distilled water to cover the soil samples for soaking, and the tray was transferred into the drying cabinet. Drying while soaking ensures that the mineral components are not lost, after drying, it was crushed and remade into tensile test samples to retest its tensile strength. After five wetting–drying cycles, it was found that the tensile strength of the kaolin remained unchanged. The experimental results can verify the conjecture that the cementation of kaolin is reversible. In addition, we found that kaolin becomes very hard at the moment of water absorption, which contributes greatly to the tensile strength of soil samples.
4.3. The Role of Suction in Tensile Resistance
4.3.1. Liquid Surface Tension
The absorbed suction caused by liquid surface tension will vary with the change in saturation angle. After the soil particles encounter water, in addition to the bond force generated by chemical action, the surface tension between the particles also increases the attraction between the particles, which is usually called suction stress or absorbed suction [
24,
25]. The suction between soil particles caused by the surface tension of water and its contact angle with soil particles causes compressive stress between particles, which means the soil particles tend to be close to each other, the expression of absorbed suction is Equation (1). Using the video optical contact angle measuring instrument (instrument type: OCA15EC), it was found that the contact angle is almost unchanged near 12.2° at different water contents (
Figure 11). Therefore, the absorbed suction
Sa increases at first and then decreases with the decreasing saturation angle
φ, and there is a peak when
φ =
φs.
where
ua is the mean air pressure in the pores of soil particles,
uw is the average water pressure in the pores of soil particles,
θ is the contact angle between soil particles and liquid bridge deck,
φ is the saturation angle,
f (
θ,
φ)= (1 − cos
φ) sin
φ tan(
θ +
φ), and
D is a parameter related to the relative density considering the gradation factor, which is related to particle accumulation mode, particle gradation, and sorting, generally
D = 2∼4. The pore water pressure
uw decreases nonlinearly with the increase in
φ [
29], and the pore pressure
ua is unchanged.
4.3.2. Matric Suction
The matric suction only reflects the water absorption capacity of the soil, and has a limited correlation with the tensile strength. The matric suction of samples with different saturations was measured using a soil moisture tester, and the tensile strength curves under the above three conditions were put in the same figure (
Figure 12).
The matric suction varies greatly with saturation, which is extremely large when it is close to complete drying, and extremely small when it is completely saturated. The range of variation is 10−2~106 kPa. It only coincides with the tensile strength among saturations of 55~65%, and the difference between other intervals is extremely large. The soil with higher saturation has a weaker water absorption performance, while the dry soil finds it easier to absorb water. It can be seen that the index reflects the water absorption trend of the soil, and its value reflects the strength of the water absorption capacity of the soil, which is not entirely the source of tensile strength.
It is feasible to explore the main factors controlling the tensile strength of granite residual soil during the drying and wetting process from the perspective of absorbed suction and cementation. The conclusion that the tensile strength under different saturation conditions is completely dominated by absorbed suction also shows that other factors have little effect on the tensile strength of granite residual soil in the process of drying and wetting, except for the two factors of bond force and absorbed suction.
The peak tensile strength during the drying process is 20 kPa different from that when complete drying. This difference is consistent with the peak value of the tensile strength of remolded soil under different saturation conditions; the cementing material has been finalized in the positive correlation section of the drying process, and the bond force remains unchanged. Therefore, it can be considered that the strength of the peak value of tensile strength higher than the peak value of remolded soil during the drying process is all derived from cementation, or the tensile strength of the completely dried soil is all provided by the cementing force.
4.4. Analysis of the Main Factors Controlling Tensile Strength in Each Stage
4.4.1. Interparticle State in the Initial State
The contact state of soil particles in water and cement can be divided into filling, contact, and suspension. When the sample has just been prepared, the cementation state of the remolded soil is broken, and the bond force is not effectively established. Therefore, under this condition, the bond force will not contribute to the tensile strength, and the tensile strength is mainly controlled by the absorbed suction.
Under the conditions of low saturation, the main form of water in the soil is bound to water, which is manifested as a bound water film wrapped on the surface of soil particles. At this time, there are only a few liquid bridges between soil particles, and the liquid bridge connection for transferring absorbed suction is not fully established (
Figure 13a). The tensile strength mainly depends on the direct contact between some clay particles.
When the saturation increases, free water fills the pores and forms more liquid bridges between particles (
Figure 13b). The formation of liquid bridges forms a skeleton that can transfer absorbed suction between soil particles, and the tensile strength increases rapidly. When the saturation increases, and the liquid bridge advances to the surface of the soil particles. The saturation angle between the liquid bridge and the soil particles becomes larger, and
uw decreases with the increasing saturation angle. This change will be manifested in the increase in absorbed suction at the beginning.
However, as the saturation continues to increase, the saturation angle continues to increase, and the absorbed suction decreases. Therefore, the absorbed suction increases first and then decreases with the increase in saturation during the liquid bridge establishment stage, and the maximum absorbed suction corresponds to the peak tensile strength. When the saturation continues to increase to a near-saturated state, the soil particles are completely wrapped in free water, the liquid bridges are connected and disappear, the soil loses the gas phase completely, only the liquid and solid phases are left, and the structure of stable transfer absorbed suction is lost. Only a small number of pores are not filled with free water (
Figure 13c). At this time, the soil is in a more dispersed state, and under the action of water pressure, tensile strength is very low and prone to disintegration.
4.4.2. Evolution of Tensile Strength in the Wetting Process
During the wetting process, the soil is remolded soil in which cementation is not established, and the tensile strength is mainly controlled by the absorbed suction. Free water gradually penetrates and diffuses in the soil. Although this process tends to be similar to the saturation of the soil over time, it cannot reach the same state. Therefore, the wetting process can be regarded as a disorderly superposition of the three states in
Figure 14, and the highly absorbed suction state that can effectively establish the liquid bridge connection is less. Therefore, the overall strength and peak strength of the wetting process is lower than those of different saturations. The changing rate of the tensile strength with saturation is small, and the curve shows a fast-growing type. During the wetting process, the absorbed suction monotonically decreases with increasing saturation, and the soil particles do not tend to approach each other due to the change in absorbed suction.
4.4.3. Tensile Strength Evolution during the Drying Process
The tensile strength during the drying process is controlled by the bonding force and the absorbed suction. In the initial stage of drying, the free water close to the surface soil evaporates first, the saturation of the surface soil is lower than that of the interior, and gas pores appear. At this time, the cementing material is less precipitated, only a small amount exists on the surface of the soil particles, and the cementation between the particles has not yet been formed (
Figure 14a).
As the water continues to evaporate, the liquid bridge connection gradually forms, the absorbed suction increases, the soil particles approach each other under the traction of the absorbed suction, and the soil volume shrinks. At this time, the free oxides between soil particles also accelerate the precipitation and form cementation between particles (
Figure 14b). The tensile strength corresponds to the negative correlation section in
Figure 5. This trend of approaching each other first appears in the surface soil, and then the absorbed suction of the soil is not equal during the internal development and drying process. For a single soil particle, if the two sides are affected by the absorbed suction from the liquid bridge between adjacent soil particles, the force on both sides is not equal, and the soil particles will move to the side with larger absorbed suction when appearing far away from the other side of the soil particles. This phenomenon is macroscopically manifested as the dry shrinkage crack of the soil, which will significantly reduce the tensile properties of the soil. This shows that the absorbed suction is not only a contributor to the tensile strength of the soil, but also a destroyer of the tensile strength.
In the positive correlation stage, the absorbed suction has reached a maximum; if the saturation continues to decrease, the absorbed suction will begin to decrease. Free oxides continue to precipitate between soil particles to form cementation, and the bond force continues to rise, but the rate of change is slower than that of the negative correlation stage. Because the cementation is gradually formed outward from the shortest connection between particles, more free oxides need to be precipitated to increase the contact area of cementation. At this time, the increase in bond force and the decrease in absorbed suction maintain a balanced state, and the tensile strength is characterized by stable high strength. In the positive correlation stage, the water in the soil is bound to water film, only a small amount of liquid bridge exists, and the cement forms a complete skeleton between multiple soil particles (
Figure 14c). As the saturation continues to decrease, the liquid bridge continues to lose, and the bound water film becomes thinner. At this stage, the cement has been finalized, the bond force is unchanged, and the tensile strength is slightly reduced.