Influence Mechanism of the Interfacial Water Content on Adhesive Behavior in Calcium Silicate Hydrate−Silicon Dioxide Systems: Molecular Dynamics Simulations

Abstract

The performance indicators of concrete are mainly determined by the interface characteristics between cement hydration slurry and aggregates. In this study, molecular dynamics technology was used to evaluate the effect of the interfacial water content on the evolution of the interface structure, interaction energy, and mechanical properties of calcium silicate hydrate (C-S-H) and silicon dioxide (SiO₂) systems, and the weakening mechanism of the C-S-H/SiO₂ interface in a humid environment was revealed. The results showed that all stress–strain curves of C-S-H/SiO₂ were divided into the elastic stage and the failure stage. As the interfacial water layer thickened, the molecular weight of the water invading the C-S-H gradually increased, and the desorption of Ca²⁺ ions in the surface region became significant, while the amount of Ca²⁺ ions entering the water-layer region increased. The interaction energy of the C-S-H/SiO₂ progressively became larger, and the energy ratio (ER) significantly decreased; the tensile strength σc and residual strength σr of C-S-H/SiO₂ both showed a downward trend. In summary, a lower water content had a limited impact on the interfacial bonding strength, while the weakening effect enhanced with an increase in the interfacial water content. This phenomenon was also demonstrated in concrete interfacial bond strength experiments.

1. Introduction

As an abundantly utilized type of material in buildings, the industrial energy consumption of cement production has reached 40% of the total energy consumption of human production, and its carbon dioxide emissions account for 8% of the world’s total [1]. The development of high-performance cement−based material, under the same bearing capacity, can effectively reduce material consumption, which is of great significance in achieving sustainable development [2]. Concrete is mainly composed of gel, aggregate, and the interface transition zone (ITZ). The interface transition zone, which is the link between the gel and the aggregate, has a remarkable effect on the overall performance of cementitious composites [3]. The ITZ has a greater porosity than the substrate, as well as being relatively loose, leading to micro-cracks, which seriously affect the durability of the material [4]. Consequently, a large amount of research has been conducted into this problem.

The microscopic properties of concrete ITZ have mainly been evaluated using microstructure identification techniques, such as scanning electron microscopy (SEM). For instance, Varga et al. [5] used the SEM images of concrete ITZ to analyze how the interfacial wetness influenced the adhesive performance of the gel and aggregate pull-off. Shen et al. [6] applied backscattered electron (BSE) imaging and nano-indentation techniques to study ITZ change characteristics during the carbonation of concrete. Using atomic force microscopy (AFM), SEM, and nano−indentation techniques, Xiao et al. [7] discussed the subject−to−mix ratio, the types of aggregate, and the hydration period for interfacial characteristics of recycled concrete ITZ. Tian et al. [8] adopted X-ray computed tomography (XCT) to reconstruct the micromorphology of concrete, and comprehensively examine the impact of the ITZ, pore, and aggregate upon the chloride ions’ diffusivity flux. However, most of the above techniques require relatively specific specimens, and the spatial resolution cannot fully reveal the composition information of the interface microstructure.

The molecular dynamics (MD) technique is a computational simulation tool using the basis of traditional Newtonian mechanics. It has been extensively used in investigating the microstructure and characteristics of cement composites [9,10]. Sanchez et al. [11] utilized molecular dynamics methods to research the atomic interactions, structural changes, and kinetic characteristics at the interface of the C-S-H structures and the graphite structures with various functional groups. They discovered that interfacial interactions are the most predominant role of the electrostatic forces. Yang et al. [12] revealed the weakening mechanism for the adhesive performance at the interface of C-S-H and epoxy resin in a thermal environment from a nanoscale perspective, and found that an increased temperature decreases interfacial interactions. Zhou et al. [13] worked with the atomic interaction mechanism of the common aggregate calcite/silica and C-S-H, and discussed the influence of moisture on the interactions. From the above studies, it is clear that most research works have focused on C-S-H/carbon-based nanomaterial interfaces and C-S-H/polymer interfaces, while only a few studies have examined and considered the effects of the interfacial water molecule content on the adhesive properties of C-S-H/aggregate interfaces.

In this study, the MD technique was employed to evaluate the influence of the interfacial water content on the evolution of the interface structure, the interaction energy, and the mechanical property of the C-S-H/SiO₂ system. The C-S-H/SiO₂ interfacial weakening mechanism in a humid environment was revealed. These results are crucial for understanding nanoscale gel/aggregate interactions, and for developing high-performance cementitious materials.

2. Computational Simulation Details

2.1. Molecular Model

The approach suggested by Pellenq et al. [14] to establish the amorphous C-S-H structure served as a foundation for the current study; the starting structure for the amorphous C-S-H structure was a supercell from Hamid’s proposed monoclinic Tobermorite-11 Å layered structure. The specific method was as follows. (1) The monoclinic structure of the supercell was transformed into a cubic structure, while the atomic cartesian positions and model dimensions were kept consistent. After this, the C-S-H structure dimensions in x, y, and z were 22.32 Å, 22.17 Å, and 22.77 Å, respectively, and the angles α, β, and γ were all 90°. (2) To match the actual distribution characteristics of the C-S-H silicates chain, as observed in nuclear magnetic resonance (NMR) experiments, some SiO₂ groups were randomly removed from the C-S-H structure [14]. (3) The ordered distribution water molecules in this structure were removed, and then water molecules were arbitrarily introduced into the C-S-H structure interlayer region, to reach water saturation using the grand canonical Monte Carlo absorption technique. (4) The saturated models were then relaxed by 1000 ps under the isothermal–isobaric (NPT) ensemble, nearly to equilibrium. Based on the above method, the obtained C-S-H’s molecular formula was (CaO)_1.67(SiO₂)·(H₂O)_1.68. Meanwhile, the density of the final structure (2.45 g/cm³), the structure of the silicon chain Q_n distribution (Q₀ = 11.63%, Q₁ = 67.44%, and Q₂ = 20.93%), and the average chain length (MCL = 2.62), as shown in Figure 1a, were in agreement with the previous simulations [14] and experimental results [15].

In this research, the silica crystal was utilized to simulate the aggregate, where the sizes of the lattice were x = y = 4.913 Å, z = 5.4052 Å, and the angles of the lattice were α = β = 90°, γ = 120°. To construct the ideal C-S-H/SiO₂ interface, additional processing was required for the established C-S-H structure and the selected SiO₂ crystal. Firstly, the established C-S-H structure was cut along the [001] direction, so that the z-direction surface was exposed, and the resulting model had the final dimensions x = 43.73 Å, y = 44.25 Å, and z = 47 Å. Secondly, the SiO₂ model was also cut along the [001] direction, and supercells were generated in the x- and y-axis directions, until it matched the dimensions of the C-S-H cross-section. Figure 1b displays the established SiO₂ model. At last, the C-S-H/SiO₂ interface model was completed, after the SiO₂ crystal model was placed on the C-S-H interface.

To examine the interfacial water content impact mechanism on the bonding performance of the C-S-H/SiO₂ system, the models that had varying water-layer thicknesses (0 Å, 1 Å, 3 Å, and 5 Å) were selected as the study subjects. This is shown in Figure 1d–g, where the preparation of water layers was achieved by the absorption of water molecules within a box of the rated thickness (0 Å, 1 Å, 3 Å, and 5 Å) and its density was established at 1 g/cm³, as seen in Figure 1c.

2.2. Force Field

The choice of force field type directly impacts the precision of molecular dynamics simulations. The combined empirical force field approach (ClayFF + CVFF) has been extensively used, and has proven effective for studying the interfacial properties of cementitious materials with other organic or inorganic materials [16,17]. To analyze the mechanism of interfacial water content on the bonding performance of the C-S-H/SiO₂ system, the combined empirical force field approach was utilized in the present investigation.

Both the C-S-H and interface water molecules were given the ClayFF force field. The ClayFF is a reliable general empirical force field, which can simulate the structural performance and dynamical characteristics of the C-S-H and its interfacial properties with precision and efficiency. This force field was applied successfully to the simulation of cementitious materials, and their interatomic interactions at the liquid-phase interface [18,19]. Furthermore, this force field was also employed to simulate the connection between water molecules, along with clay mineral surfaces [20,21].

The SiO₂ aggregates were simulated, utilizing the CVFF force field. This force field’s potential functions consist of bond and non-bond interactions, and it mainly contains the calculations of bond length, bond angles, torsion angles, and dihedral angles. In short, simulating the C-S-H/SiO₂ system interaction utilizing the combined empirical force field (ClayFF + CVFF) approach was successful.

2.3. Simulation Process

The simulation process was implemented using the Large-scale Atomic/Molecular Massively Parallel Simulator (LAMMPS) software. In this study, the simulation process was separated into equilibrium state simulation and uniaxial stretching simulation phases. Firstly, the dynamic equilibrium of 500 ps was performed using the NPT ensemble. The simulation utilized a time step of 1 fs, and the Nosé–Hoover thermostat was adopted to control a temperature of 300 K, along with pressure at 0 KPa. In order to guarantee the reliability of the equilibrium analysis, relevant data were used from the last 100 ps of the equilibrium state simulation. Secondly, the uniaxial (Z direction) tensile tests were simulated, to analyze the mechanical performance and stretching behavior of different C-S-H/SiO₂ systems. The uniaxial tensile load strain rate was 0.001/ps during the tensile tests, and the maximum strain was 0.25. The system’s temperature and time steps were consistent with the equilibrium state simulation process.

3. Results and Discussion

3.1. Local Microstructure Evolution

For research into the impacts of varying interfacial water molecule contents on the C-S-H/SiO₂ system, the changes in the z-axis direction sizes of four models with varying interface water-layer thicknesses (0 Å, 1 Å, 3 Å, and 5 Å) after equilibration were counted, as shown in Figure 2. Figure 2a depicts the initial and balance sizes of the C-S-H/SiO₂ model along the z-axis for varying water-layer thicknesses. The size of the C-S-H/SiO₂ model increased gradually as the interfacial water content rose. However, the change in the C-S-H/SiO₂ model’s structural size was inconsistent with the amount by which the interfacial water layer thickened.

The differential size between the initial model and the balanced model is presented in Figure 2b. It is evident that when the interfacial water layer thickened (increasing from 0 Å to 1 Å), the variation in the structure size was less than 1 Å, and the differential size between the initial model and the balanced model was 0.66 Å. When the interfacial water layer thickened (increasing from 0 Å to 3 Å and 5 Å), the differential sizes between the initial model and the balanced model were 1.32 Å and 1.96 Å, respectively. The optimized water-layer thickness changed from 1 Å, 3 Å, and 5 Å to 0.34 Å, 1.68 Å, and 3.04 Å.

In order to disclose the mechanism by which the interfacial water molecule content affects the microstructure of C-S-H/SiO₂, this research analyzed the evolution of interfacial water molecules, and the evolution of Ca²⁺ ions at the C-S-H surface region. A detailed analysis follows.

(1)

Evolution of interface H₂O molecules

Figure 3 is the distribution of the interfacial H₂O molecules with varying contents at the X−Z planar in the C-S-H/SiO₂ model. As depicted in Figure 3a, when 1 Å-thick interfacial water molecules were introduced, the interfacial water molecules progressively invaded the C-S-H structure. Based on the Wenzel theory, water molecules could fill the solid’s pits on its rough surface, and the C-S-H surface’s hydrophilic nature allowed water molecules to diffuse on its surface [22,23]. As revealed in Figure 3b,c, when the interfacial water layer thickened (increasing from 1 Å to 5 Å), the numerous molecules of H₂O formed clusters easily, while exhibiting a high mobility [24] and diffusion coefficient [25]. As a result, the number of H₂O molecules that invaded increased gradually.

Because of the invasion of H₂O molecules, the change sizes of the C-S-H/SiO₂ model were smaller than the change in the water molecule thickness. As more H₂O molecules invaded the C-S-H, the difference between the initial and balance sizes of the model structure increased, as the interfacial water layer thickened. This result is consistent with the variation law of the structure size presented in the previous paper (as shown in Figure 2). The interfacial water molecule evolution law has a similarity to the conclusion obtained by Kai et al. [26], suggesting that water molecules can impact structure size by filling interfacial molecular voids.

(2)

Ca²⁺ ion evolution at the surface region of C-S-H

For the C-S-H/SiO₂ system with varying interfacial water contents, Figure 4 depicts the spatial distribution of Ca²⁺ ions at the surface region of C-S-H. It is thus clear that in the absence of a water layer (0 Å), Ca²⁺ ions were bound by the surfaces of C-S-H and SiO₂ [27], resulting in a majority of the Ca²⁺ ions being located near the system interface (Figure 4a). However, with 1 Å-thick interfacial water molecules injected, almost all of the Ca²⁺ ions were found back in the C-S-H structure (Figure 4b). This is because, with the interfacial water molecules introduced, the O atoms within the H₂O molecules readily and easily form-bonded to Ca²⁺ ions [28], resulting in a weaker binding of Ca²⁺ ions on the SiO₂ surface.

With the water-layer thickness increasing from 1 Å to 5 Å, some of the Ca²⁺ ions underwent desorption, detached from the C-S-H surface region, and entered the aqueous-layer region, as illustrated in Figure 4c,d. When the interfacial water contents were relatively large, the bond of the C-S-H to Ca²⁺ ions at the surface region was weakened, resulting in the enhancement of the movement ability of Ca²⁺ ions [27]. As the interfacial water layer thickened, the desorption of Ca²⁺ ions in the surface region became significant, and the number of Ca²⁺ ions entering the water layer region increased, leading to a decrease in the cohesion of the interface. The result was similar with the movement behavior of Ca²⁺ ions between C-S-H substrates and epoxy molecules at varying interfacial water contents, as observed by Kai et al. [26].

3.2. Analysis of Interaction Mechanisms

The correlation between two particles’ spatial positions in a particle system is known as the radial distribution function (RDF) [29]. Its physical meaning lies in the proportion of the atomic density in a system’s local area to its average density. The RDF is determined using the following equation:

$$g\left(r\right)=\frac{1}{4r^2\pi \rho \delta r}dN\left(r\right)$$

(1)

where dN(r) represents the number of atoms within a region from r to r + δr, ρ represents the system’s average density, and δr is the regional distance. Differences in system structure can lead to different values and trends of RDF, determining the system state. The RDF of an amorphous system has a wave peak only in the proximity, and tends toward 1 as the radius increases [30]. Given that the ClayFF force field employed different potential energy parameters for the interlayer calcium (Ca_h) and intralayer calcium (Ca_o) in the C-S-H [31], the Ca_h dominated the motion of Ca²⁺ ions within the model. Therefore, the RDF was analyzed in this paper by counting the Ca_h.

Figure 5 shows the RDF of the different ionic pairs for varying C-S-H/SiO₂ interfacial models. The RDF between the ionic pairs all showed apparent peaks in the near range, and the RDF values gradually tended toward 1, as the atomic spacing increased. This conforms to the short-range ordered and long-range disordered structural characteristics of amorphous matter. The peaks of RDF appeared at the same position, indicating that there was no influence from the interfacial water molecule content on the bond lengths of Ca_h-Si_CSH and Si_CSH-O_CSH. However, in the C-S-H/SiO₂ model with varying interfacial water molecule contents, there were some variations in the RDF peaks of the different ion pairs.

When the interfacial water layer thickened (introducing a 1 Å thickness), the RDF peak values of Ca_h-Si_CSH gradually became larger (Figure 5a), indicating that the quantity of Cah ions around the Si ions increased gradually within C-S-H. The Ca_h-Si_CSH RDF peak values showed a decreasing law as the water-layer thickness increased from 1 Å to 5 Å, indicating that the amount of Ca_h ions close to the Si ions in C-S-H decreased gradually. This result agrees with the Ca²⁺ distribution in the C-S-H surface region observed by the C-S-H/SiO₂ interfacial model for varying thicknesses of the water layer (Figure 4).

As the interfacial water layer thickened, the RDF peak values of the Si_CSH-O_CSH increased continuously (Figure 5b). It can be seen that an augmentation of the thickness of the water layer resulted in a higher accumulation of O atoms accumulated by Si atoms in a short range. This result demonstrates that as the water layer became thicker, more water molecules intruded into the C-S-H (Figure 3).

3.3. Analysis of Interaction Energy

The bonding performance of the C-S-H/SiO₂ systems is typically assessed through the interaction energy between the C-S-H and the aggregate [13]. The interaction energy detailed formulae are as follows:

$$E_{CSH-aggregate}=E_{aggregate}+E_{CSH}-E_{aggregate+CSH}$$

(2)

where E_{aggregate+CSH} represents the C-S-H/aggregate system total energy at balance; E_aggregate represents the aggregate energy; and E_CSH represents the C-S-H energy. The positive interaction energy represents repulsion, while the negative interaction energy represents attraction. As the positive value increases, the interface becomes more unstable. The interface adhesion improves as the absolute value of the negative value increases.

In previous experimental studies, the energy ratio (ER) has frequently served as a metric to assess the ramifications of the water content on the adhesive characteristics of asphalt, in relation to the aggregate [32]. This approach derived its underpinnings from the supposition that the water susceptibility is positively correlated to the dry adhesion work, and negatively correlated to the debonding work. On this basis, Xu et al. [33] and Zhou et al. [13] evaluated the bonding performance among C-S-Hs with the aggregate, demonstrating the effectiveness of this method on the microscopic scale. The ER can be computed by Equation (3), the adhesion work among C-S-H with the aggregate is computed with Equation (4), and the work of debonding due to the addition of water to it can be calculated according to Equation (5). The detailed formulae are as follows:

$$ER=\left|W_{adhesion}/W_{debonding-water}\right|$$

(3)

$$W_{adhesion}=\Delta E_{CSH\_aggregate}/A$$

(4)

$$W_{debonding-water}=\left(\Delta E_{aggregate-water}+\Delta E_{CSH-water}\right)/A-\Delta E_{CSH-aggregate}/A$$

(5)

where W_adhesion represents the adhesion work among C-S-H structures with the aggregate structure; W_{debonding-water} represents the work required to debond the two structures when water is present in the interface region; ΔE_CSH-water represents the interaction energy of the C-S-H with water; ΔE_{aggregate-water} represents the interaction energy among the aggregate with water; ΔE_{CSH-aggregate} refers the interaction energy between the C-S-H and the aggregate; and A refers the area of the interface contact region.

Figure 6 displays the computed interaction energy and energy ratio (ER) of C-S-H/SiO₂ in varying moisture contents. From Figure 6a, it can be seen that the interaction energy of the C-S-H/SiO₂ progressively became larger as the interfacial water-layer thickened. The occurrence of water at the interface weakened the interfacial interactions, and reduced the bonding capacity. There was a tendency for the interfacial interactions to become weaker as 1 Å-thick interfacial water molecules were introduced, and the interaction energy rose from −532.8849 kcal/mol to −487.7079 kcal/mol. This phenomenon arose because the O atom in the interface water bonded easily with the Ca²⁺ ions [28,34], resulting in the bond created between the Ca²⁺ ions in C-S-H and the O in SiO₂ being destroyed (as shown in Figure 7). As more Ca²⁺ ions entered the water layer, the movement ability of Ca²⁺ ions was enhanced, and it was easier to bond with the O atoms of water. Therefore, as the water layer thickened further (growing to 5 Å), the interfacial interactions became observably weaker, and the energy increased from −487.7079 kcal/mol to −181.4997 kcal/mol. This result coincides with the Ca²⁺ ion distribution law in Figure 4c,d.

As demonstrated in Figure 6b, the energy ratio (ER) decreased significantly with the increase in the interfacial water-layer thickness. This phenomenon suggests that the interfacial bonding performance between C-S-H and SiO₂ aggregates was diminished due to the existence of water molecules. Moreover, experimental studies have confirmed this conclusion. As an example, Shi et al. [35] discovered that the deterioration of the peeling fracture energy was proportional to the amount of moisture at the bonding-layer–concrete interface.

3.4. Analysis of Mechanical Properties

In order to further explore the C-S-H/SiO₂ system of mechanical properties subject to the interfacial water content, uniaxial tensile simulation experiments were conducted along the direction of the Z axis in this study. In Figure 8, the C-S-H/SiO₂ system stress–strain relation curves obtained for varying water-layer thicknesses are plotted, while the variation in the tensile strength σ_c and the residual strength σ_r are discussed. The C-S-H/SiO₂ systems stress–strain curves with different water-layer thicknesses were all divided into two stages: the elasticity (OA) stage and the failure (AB) stage (Figure 8a). During the OA stage, the stress values of all C-S-H/SiO₂ systems gradually increased and reached a peak value, as the strain increased. During the AB stage, the stress dropped sharply as the tensile strain increased, and then gradually flattened out.

Figure 8b demonstrates that as the interfacial water layer thickened, the C-S-H/SiO₂ system tensile strength σ_c showed a downward trend, indicating that the water molecules presented a weakening effect on the bonding performance. Whilst 1 Å-thick interfacial water molecules were introduced, the decrease in tensile strength σ_c was smaller, and the strength value decreased from 1.31 GPa to 1.18 GPa. As the water-layer thickness increased to 3 Å and 5 Å, the tensile strength σ_c decreased significantly, and the strength values decreased to 0.92 GPa and 0.68 GPa, respectively. Visible lower water content had a limited impact on the interfacial bonding strength, while the weakening effect enhanced with the increase in the interfacial water content. This result is consistent with that obtained from the interaction energy (as shown in Figure 6). The simulation results are similar to the mechanisms by which water molecules weaken the adhesive performance of the asphalt–quartz [36], and the effect of the ITZ thickness and strain rate on the mechanical properties of C-S-H/SiO₂ systems [37].

However, the tensile strength obtained from the MD simulations is much greater than the experimental results. Gu et al. [38] examined the mechanical properties of the interface between cobblestone aggregate and mortar, and found that the tensile strength of the interface was 1.8 MPa. Jebli et al. [39] obtained a tensile strength of 1.75 MPa for the cement–aggregate interface, after a hydration time of 90 days. This large difference was attributed to the time and space constraints of the MD simulations [40]. Furthermore, the actual influencing factors, such as porosity, were not considered in this simulation.

The residual strength σ_r exhibits of the material tended to decrease in the case of increasing water-layer thickness (Figure 8b). During the event of a low water content, the C-S-H/SiO₂ system exhibited an excellent interfacial bonding performance, and the structure showed good resistance to loading, even after the structure was damaged. As the interfacial water content grew, the interfacial bonding performance of the system was weakened, causing a further reduction in the material residual strength σ_r, with the strength value decreasing from 0.37 GPa to 0.18 GPa. The simulation results are similar to the mechanisms by which water molecules weaken the adhesive performance of the C-S-H–graphene-oxide [41] and asphalt–aggregate [42] interfaces. Moreover, this phenomenon was verified in the concrete interfacial bond strength experiments by Beushausen et al. [43].

4. Conclusions

• As the interfacial water content increased, the size of the C-S-H/SiO₂ model increased gradually. As the interfacial water layer thickened (increasing from 0 Å to 5 Å), the number of water molecules invading C-S-H increased gradually, leading to an increase in the difference between the initial size and the balanced size of the model structure.
• With the injection of interfacial water molecules, the O atoms within the H₂O molecules easily bonded to Ca²⁺ ions, resulting in almost all of the Ca²⁺ ions being located back within the C-S-H. As the interfacial water layer thickened, the Ca²⁺ ion desorption in the C-S-H surface region became significant, and the number of Ca²⁺ ions entering the water layer region increased, leading to a decrease in the interface cohesion.
• As the interfacial water layer thickened, the interaction energy of the C-S-H/SiO₂ progressively became larger, and the energy ratio (ER) decreased significantly. The same conclusion has also been confirmed in experimental research.
• The C-S-H/SiO₂ system stress–strain curves with varying water-layer thicknesses were all divided into two stages: the elasticity stage and the failure stage. With an increased interfacial water-layer thickness, the tensile strength σ_c and the residual strength σ_r of the C-S-H/SiO₂ system both showed a downward trend. The weakening effect of a low water content on the interface bonding strength was limited, and as the interfacial water content increased, the weakening effect on the C-S-H/SiO₂ was enhanced. This phenomenon has been verified in concrete interfacial bond strength experiments.