The mechanism of cesium ions immobilization in the nanometer channel of calcium silicate hydrate: a molecular dynamics study

Jinyang Jiang a, Pan Wang b and Dongshuai Hou *b
aSchool of Materials Science and Engineering, Southeast University, Nanjing, China. E-mail:
bDepartment of Civil Engineering, Qingdao University of Technology, Qingdao, China. E-mail:;

Received 10th August 2017 , Accepted 29th September 2017

First published on 29th September 2017

The cement-based matrices are preferred candidates in disposing nuclear waste due to the immobilization role of the calcium–silicate–hydrate (C–S–H) gel. To better understand the immobilization mechanism of cementitious materials, molecular dynamics was utilized to investigate the intensity distribution, local structure and dynamics properties of Cs+ ions in the vicinity of the calcium silicate surface. The strong inner-sphere adsorbed cesium ions were restricted by coordinated oxygen atoms in bridging and pair silicate tetrahedron and water molecules were fixed in the silicate channel by H-bonds network. On the other hand, the adsorption of chloride ion, repulsed by the negatively charged silicate surface, is mainly attributed to the formation of the cation–anion ionic pair near the interface. As compared with those of the solvated ions in the solution, the relaxation time of water in the hydration shell of adsorbed Cs+ is significantly increased and the diffusion coefficient of adsorbed Cs+ is dramatically reduced. Furthermore, based on the intensity profile and resident-time analysis, the adsorption capacities of monovalent cations on the C–S–H surface increase with decrease in the ionic radius, following the sequence of Na+ ≫ K+ > Cs+. This study provides a molecular-level understanding of the immobilization mechanism of different ions in the C–S–H gel pores.

1 Introduction

The release of the radioactive material through nuclear weapon testing, the accidents occurring in the nuclear industry and the radioisotope production facilities have a hazardous effect on the environment.1 It is a challenging and urgent task to manage and dispose the chemical and radioactive waste generated from the radionuclides and heavy metals.2 The fate of detrimental waste in the environment can be controlled to a safe level using cement-based materials that can immobilize the containments by adsorption and precipitation.3,4 Because of the high specific surface area, widely distributed nanometer gel pores and high surface reactivity, the cement-based materials are able to immobilize the radioactive containment.5,6 In addition, the low-cost, high density and durability of Portland cement satisfy the criteria for immobilizing the radioactive waste.7,8 The cement-based materials can be used as a physical barrier, such as the shielded cask, to isolate the radioactive waste. Chemically, the hazardous ions can also be immobilized into the cement paste during the cement hydration process.9 For either case, once the cement-based material comes in contact with groundwater, the radioactive species can leach from the cement paste. Hence, many leaching studies have been reported on the Cs+ species solution incorporated into the cementitious materials by different experimental techniques.10 The binding of Cs+ ions to the cement paste has been investigated by many researchers and was found to contribute to the immobilization of the ions.11,12 Glasser and Hong13 proposed that the binding capability of Cs+ ions depends on the chemical composition, such as the Ca/Si ratio, of the cement paste. In addition, additives or minerals such as tobermorite,14 zeolite,15 bentonite16 and pozzolanic material7 have been incorporated into the cement paste in order to improve the immobilization of solute radioactive waste. These experimental studies are essential and provide a direct guidance for the practical application of cement matrix in immobilizing radioactive waste. However, the intrinsic immobilizing mechanism, particularly for the adsorption characteristics and the transport process of radioactive species in C–S–H gel, is challenging to probe solely by experimentation.

Molecular dynamics simulation can provide detailed information about the dynamic and structural information of the complex systems at the molecular level and has been widely exploited to investigate the adsorption or dynamic properties of cesium ions in the vicinity of the mineral surface.17–20 Vasconcelos and Bunker21 have investigated the adsorption of Cs+ ions on the basal surface of kaolinite. They found that the immobilized mechanism for kaolinite is the chlorine-driven inner-sphere adsorption of cesium on the gibbsite surface at high ionic strengths. The structure and dynamics of water molecules were studied on the Cs-exchanged muscovite surfaces by Sakuma and Kawamura.22 The presence of Cs+ ions can influence the structure of water at the interface: water molecules around the adsorbed Cs+ ions on muscovite surfaces do not prefer to coordinate with Cs+ ions and the density of water oxygen near the muscovite surface decreases relative to that in a bulk state. Nakano et al.23 have investigated the behavior of Cs atoms and the adsorption sites for Cs atoms on the bentonite surface and found that the inner-sphere surface complexes coexisted with the outer-sphere surface complexes and the adsorbed sites were positioned nearby the edge of the basal oxygen hexagonal cavity on tetrahedral sheets. These studies have provided molecular insights on the immobilizing mechanism of the clay materials. However, the adsorption and transport of Cs+ ions in the cement hydrate is not the same as in case of the clay minerals due to the molecular structural differences. Fortunately, the interface between the solution and the cement hydrate has been investigated by many researchers and can guide the further study.

Andrey et al.24 applied ClayFF force field to investigate the adsorption properties of chloride ions in various concrete products and found that the adoption capability of different concrete products followed the order of F salt > calcium hydroxide > ettringite > tobermorite. Rotenberg et al.25 studied the exchange between water and ions in the intercrystalline microspores of clay minerals and the investigation results showed that the cations have the capability of exchanging with water via passing through the potential barrier, while the anions could not exchange with water. Kalinichev et al.26 showed that tobermorite, an important mineral analog of C–S–H, could hardly attract chloride ions due to its negatively charged surface. Hou et al.17,18,20 have performed a large number of studies focusing on the structural and dynamic behavior of water and chloride ions in the C–S–H gel pores. A significant number of studies have been carried out to investigate the interaction between C–S–H and chloride ions that can cause the corrosion of steel reinforcements and destroy of the cement-based material. Nevertheless, few studies have been carried out to investigate the interaction mechanism between solvated radioactive species such as cesium ions and cement hydrates.

In this study, molecular dynamics was utilized to investigate the molecular structure, dynamics and interface behavior of cesium ions in the gel pores of calcium silicate hydrate. Other ions widely distributed in the gel pore solution of cement hydrate such as sodium, potassium and calcium ions were also studied for comparison. The density distribution, the radial distribution function (RDF), the stability of chemical bonds, the motion trajectory, and the diffusion coefficient were utilized to characterize the adsorption structural and dynamic properties of different ions in nanopores. Furthermore, the influence of the adsorption cations on the structural and dynamic behavior of chloride ions was also discussed.

2 Models and methods

2.1 Model construction

As the mineral analog of calcium–silicate–hydrate, the 11 Å tobermorite structure was chosen as the simulation model. The tobermorite supercell contained 2 × 2 × 1 crystallographic unit cells, where the (001) plane was selected as the exposed surface and the dimensions were set as follows: a = 44.64 Å, b = 44.34 Å, c = 22.77 Å, α = β = γ = 90°. According to the methodology mentioned in ref. 27, the tobermorite surface was created by cleaving the interlayer region along the (001) direction. The cleaved tobermorite interface, shown in Fig. 1a, is featured by the dreierketten distributed silicate chains that included bridging silicate tetrahedron protruded into the interlayer region and pair tetrahedron rooted in the calcium sheet. It can be clearly observed in Fig. 1a and b that a narrow and long channel distributed along y-direction exists between neighboring silicate chains. The silicate chains contained a large number of oxygen sites that made the surface negatively charged. A part of the non-bridging oxygen atoms in bridging silicate tetrahedron was protonated and calcium atoms were randomly added in the vicinity of the surface to guarantee the charge balance. The gel pores were obtained by moving both sides of the substrate into two opposite directions. The left and right substrates have dimensions (xyz) of 44.64 Å × 44.34 Å × 7 Å and 44.64 Å × 44.34 Å × 16 Å, respectively. The width of the gel pores was selected as 45 Å, which is in the range of pore sizes from 5 Å to 100 Å as mentioned by Mindess et al.28 To compare the adsorption and dynamics of different ions in the C–S–H gel pores, 7840 atoms of water molecules and 24 cesium (or sodium, potassium) ions and chloride ions were added into the interlayer. Thus, the concentration was 0.5 mol L−1 for each salt solution. It should be noted that the ions were positioned 20 Å away from the bottom of the calcium silicate substrate to eliminate the interface influence at the very beginning. The constructed solution confined in the C–S–H gel pore model is shown in Fig. 1c.
image file: c7cp05437h-f1.tif
Fig. 1 (a) yz plane projection for the tobermorite surface; (b) xy plane projection for the tobermorite surface; and (c) snapshots of initial system configurations. NaCl solution for example. Atom color code: H, white; O, red; Si, yellow; Cl, light green; Ca, green; Cs, purple. This color code is adopted in all of the following figures.

2.2 Force field and system setup

All simulation studies were performed with the Lammps package.29 The ClayFF force field, which has been proved to accurately reproduce the structural and dynamic properties of water and ions on the oxide and hydroxide materials surface, was exploited.24,30,31 The approach of steepest descent and the leapfrog algorithms with a time step of 0.001 ps were used to perform the energy minimization and the equations motions of all systems, respectively. The Hoover Canonical ensemble (NVT) was performed at 300 K using a Nose thermostat coupling method.32 Periodic boundary conditions were adopted in all three directions. The total simulation time was 4000 ps; the last 3000 ps were used for the data analysis.

3 Results and discussion

3.1 The structure properties of ions in the C–S–H gel pores

3.1.1 Density distribution. Fig. 2 records the process of solvated Cs+ and Cl ions transforming from the solution to the C–S–H interface region during 1000 ps. In the first 100 ps, the Cs+ and Cl ions diffused toward the calcium silicate surface. At around 200 ps, a part of Cs+ ions connecting with bridging silicate tetrahedron protruded in the solution and a part of Cs+ ions penetrated into the cavities near the pair silicate tetrahedrons. After 1000 ps, a significant number of Cs+ ions adsorbed on the calcium silicate substrate, which suggests the strong adsorption effect of the cations. On the other hand, it is very hard to observe a direct adsorption of Cl ions on the silicate skeleton as Cl ions preferred to move around the cations. As compared with that of Cs+ ions, the amount of Cl ions was quite lower, implying the repulse effect of the silicate surface on the anions. Interestingly, a part of the surface calcium atoms desorbed from the C–S–H surface and diffused into the solution. The cesium ions adsorption and calcium ions desorption suggest that the cations were exchanged between the liquid and solid phases.
image file: c7cp05437h-f2.tif
Fig. 2 Snapshots of CsCl solution and C–S–H interface at (a) 0, (b) 100, (c) 200, and (d) 1000 ps.

The intensity profile (Fig. 3) provides the information of the interface atomic structure quantitatively. The location of the interfacial Si–OH is defined as the solid–liquid boundary (z = 0 Å) and displayed by the black line as shown in Fig. 3. The intensity profile of oxygen in silicate chains (Os) was taken as the reference for the solid substrate: the peaks located at −1.4 Å and −3.4 Å were mainly contributed by the oxygen atoms in the bridging silicate tetrahedron and a pair of silicate tetrahedrons, respectively.

image file: c7cp05437h-f3.tif
Fig. 3 Intensity profile of solid–liquid interface atoms (a) CsCl solution; (b) KCl solution; and (c) NaCl solution.

The water intensity profile exhibits two pronounced shoulders positioned at −0.6 Å and −2.8 Å below the silicate hydroxyl, indicating that the cavity between bridging and pair silicate tetrahedrons provides space for water molecules to penetrate. As shown in Fig. 3a, the deeply rooted water molecules present below the surface are mainly due to the strong attraction from the calcium sheet and distributed near the bridging silicate tetrahedron by H-bonds connections. As shown in Fig. 4, the water molecules are ultra-confined between the neighboring silicate chains and defined as the channel water molecules. On the other hand, away from the silicate surface, three distinct peaks are located at 2.5, 4.9 and 7.7 Å and the intensity oscillation can extend to more than 10 Å from the surface. It is caused by different interfacial H-bonds network, layer-by-layer, contributed by the Os atoms, which will be discussed in the next section. These are defined as the surface adsorbed water molecules.

image file: c7cp05437h-f4.tif
Fig. 4 The snapshots of channel water molecules.

The penetration depth of cesium ions is slightly reduced in the silicate surface, as compared with that of the water molecules. The binomial Cs intensity distribution confirms the strong adsorption capability of the calcium silicate surface. Moreover, the sharp peak with a high intensity is located close to the intensity curve of oxygen atoms in the silicate chains. Furthermore, a broadening peak with a low intensity extends to the solution. The combined local structure analysis (Fig. 5) indicates different cationic adsorption mechanisms: the Cs+ ions captured in the vacancy in the silicate channel are relatively stable; the Cs+ ions, which only bind with the bridging tetrahedron, can frequently exchange with the solution species. Two types of the cesium ions are defined as the inner-sphere adsorbed species and the outer-sphere species. With respect to the geometrical position, the inner and outer species are distinguished by the location of the oxygen atoms in the bridging silicate tetrahedron. It should be noted that there are some differences between C–S–H gel and tobermorite crystal with respect to the morphology and structural features. The Ca/Si ratio in tobermorite 11 Å is 1, which is lower than the average value of C–S–H gel (1.7).33 Hence, in the actual C–S–H surface, the calcium concentration increases to some extent and the positive charge turns more pronounced, which contributes to the more repulsive effect. The surface calcium atoms also occupy more space, restricting the adsorption of the Cs+ ions in the surface vacant region.

image file: c7cp05437h-f5.tif
Fig. 5 The snapshots of molecular structure of the surface adsorbed Cs ions: (a) and (b) inner sphere ions; (c) and (d) outer sphere ions.

In addition, the sharp peak with a high intensity indicates the concentrated distribution of the surface calcium atoms. As compared with the Cs+ ions, the Ca2+ ions are distributed slightly farther from the silicate channel. On the other hand, different from that of cations, the intensity profile of Cl ions has no pronounced maximum in the region close to the silicate interface and only overlaps the outer layered Ca2+ and Cs+ ions. It indicates that the silicate surface has a low capability to adsorb the chloride ions due to the coulombic repulsive interaction from the negatively charged silicate oxygen atoms. The slight increase of the interface Cl intensity suggests the weak adsorption on the surface with Ca and Cs atoms because of the ionic pair formation.

Furthermore, the intensity profiles of KCl and NaCl solutions are displayed in Fig. 3b and c, respectively. Similarly, the penetrated depth of the solvated species in the silicate channel follows the order as in CsCl solution: water > K(Na)+ > Ca2+ > Cl. As compared with that of the surface adsorbed Ca2+ ions in the case of CsCl, the Ca intensity profile extends to a longer distance in the solution of NaCl and KCl. A small peak located at 2.5 Å along z-direction for the Ca intensity suggests that a part of the surface calcium atoms diffuse away from the silicate channel and transform to the outer layer structure. The dissociation of the calcium ions is attributed to the fact that other cationic ions adsorbed on the silicate sites occupy the original positions of calcium ions and the interaction between cations disturbs the interfacial arrangement of calcium atoms.

Along with the density profiles, the percentage of ions adsorbed plays additional roles for better explaining the adsorption conditions of ions. The rates of ionic adsorption are defined by the proportion of the amount of ions within the distance 5 Å from the surface to all ions as summarized in Table 1. As clearly seen from the table, the calcium silicate surface shows different adsorption capabilities on different cations, following the order: Na+ > K+ > Cs+. For the monovalent cations, the adsorption capacities are influenced by ionic radius: the smaller the ionic radius, the stronger the ionic adsorption. On the other hand, the coordination number between cations and oxygen atoms in C–S–H is another important influential factor, which will be discussed in more detail in the next section. In addition, 94% of calcium atoms are adsorbed on the silicate surface, which is quite higher than that of other species in the solution. It implies that the ions with higher valance have a higher probability to adsorb on the negatively charged silicate channel. It should be noted that the Sr2+ ion is another important radionuclide, coexisting with the Cs+ ions in the containment solution. A previous study has shown that the mechanism of immobilizing Sr2+ in the calcium silicate hydrate is mainly attributed to cationic exchange between Sr2+ and Ca2+ ions.34 The behavior of the divalent ion Sr2+ resembles the properties of Ca2+ ions in the C–S–H gel. As compared with the Cs+ ions, the Sr2+ ions can be more strongly attracted to the non-bridging oxygen sites provided by the silicate tetrahedron. The mobility of the Sr2+ can be reduced dramatically due to the strong chemical adsorption. This could explain many experimental findings such as the fact that the leaching rate and diffusivity of the Cs+ ions in the C–S–H gel are higher than those of the immobilized Sr2+ ions.15

Table 1 The rates of ionic adsorption on the C–S–H gel surface
Cation Ca2+ Anion (%)
Outer (%) Inner (%) Outer (%) Inner (%)
NaCl 34 48 69 27 17
KCl 26 40 66 34 17
CsCl 33 41 63 39 12

Contrary to the adsorption of cations, the adsorption of anions is strongly influenced by cations, the adsorption of cations could promote the binding of anions on the C–S–H gel surface. Hence, the amount of chloride adsorption is relatively lower in the cesium chloride solution than that in the other two.

3.1.2 Local structure of water molecules. The intensity characteristics of the channel water and surface water discussed above were determined by H-bonds interactions between water molecules and the silicate channel. Oxygen atoms near the interface, which can develop H-bonds network with the channel and surface water molecules, can be categorized into three types: OB from Si–O–Si, ONB from Si–OH and OOH from Si–OH. As shown in Fig. 6a, with the progressively increasing distance from the channel bottom, the number of H bonds raises monotonically from 2 to 3.2 at the channel entry and slightly increase to 3.4 (bulk water value is 3.5) at around 10 Å from the surface. For channel water molecules, on average, the H bond number is 2.6: nearly 40% of H-bonds come from the Os in the silicate chains and the rest are contributed by the neighboring water molecules. Nevertheless, the water molecules at the surface have less interaction with the silicate chains than that of the channel water. The ratio of H-bonds with O atoms in the silicate chains to those with neighboring water is larger than 1[thin space (1/6-em)]:[thin space (1/6-em)]3. On average, less than 0.8 H bonds are formed with Oh. Up to 10 Å from the surface, water molecules both donate and accept 1.7 atoms to their neighboring molecules, which is close to 1.75 in bulk water.
image file: c7cp05437h-f6.tif
Fig. 6 (a) H-bonds evolution with the distance from the silicate surface; and (b) coordinated number of Ow evolution with the distance from the silicate surface.

The coordinated oxygen atoms also influence the local environment for the interface water molecules. Some coordinated oxygen atoms, even not forming H-bonds with water molecules, contribute to define the water molecule structures. As shown in Fig. 6b, on average, the total coordination number per water molecule decreases from 6.1 to 5 at the channel entry and continuously decreases to 4.2 (the value is similar to that in bulk water) at around 3 Å away from the surface.

For the channel water molecules, besides the elements forming H bonds, the coordinated atoms also include 3–4 additional ONB atoms from the silicate chains. These ONB atoms as well as the Ow atoms construct the cage to restrict the channel water molecules.

3.1.3 Local structure of ions. The radial distribution function (RDF) is another important structure parameter, which can reflect the sorption characteristics of ions on the C–S–H gel surface. The spatial correlations between cations and the oxygen atoms (Ow) in bulk water (Fig. 7a), between cations and the oxygen atoms (Os) in the silicate chains (Fig. 7b), and between cations and anions (Fig. 7c) were analyzed. As shown in Fig. 7a, the first peak of RDF for Na+–Ow, K+–Ow, and Cs+–Ow is located at 2.35 Å, 2.90 Å, and 3.20 Å, respectively, which matches well with the previous experimental results (i.e. 2.3–2.5 for Na and 2.95–3.21 for Cs).35 The width of this first peak ranks in the following order: Cs+–Ow > K+–Ow > Na+–Ow. It indicates that the large hydration shell of cesium ions are associated with more water molecules. In addition, while the sodium–oxygen RDF exhibits a second sharp peak at 4.5 Å, that for the potassium–oxygen RDF becomes weaker and for cesium–oxygen RDF it almost disappears in the medium range. For the sodium ions, the ordered arrangement of neighboring water molecules can extend at a longer distance.
image file: c7cp05437h-f7.tif
Fig. 7 Radial distribution function of (a) cation–Ow, (b) cation–Os, and (c) anion–cation.

The RDF of cations and oxygen in the silicate chains provides information on the interaction between the adsorbed ions and the calcium silicate surface. The sharp peak of Cs–O, K–O, and Na–O correlations confirms the previous observation that the adsorbed cations are mainly attributed to the coulombic attraction from the oxygen in the silicate chains. In addition, the positions of the first peak are consistent with the fact that ions with smaller hydration radii are distributed more close to the non-bridging oxygen atoms. Furthermore, as compared with the RDF for cations and water, that for cation and silicate oxygen shows multi-peaks, implying a strong correlation in a long range.

Fig. 7c exhibits a pronounced peak for the cation–chloride ions in the range from 3 Å to 4 Å. It suggests that the cations associate with the neighboring chloride ions and form the ionic pairs. The RDF of cation and chloride ions also has a pronounced peak at around 4 to 6 Å, which corresponds to the solvent separated ion pairs. The cations play an important role in bridging the silicate oxygen atoms and the chloride ions in the solution. The silicate surface adsorbed anions mainly come from the ionic pairs or water separated ionic pairs.

In addition, the coordination number, which is the total number of neighbors of a central atom in an ion, provides more quantitative structural information for the local environment surrounding the ion. The coordination number of ions is calculated via integrating the first shell of RDF; the same method was reported by Iwaida et al.36 The coordination numbers of cations–anions, cations–Ow and cations–Os in the C–S–H gel were calculated for both on the C–S–H gel surface and in the pore solution.

In the pore solution, the number of water molecules in the first hydration shell (hydration number) for Na+, K+, and Cs+ ions was 5.77, 7.12, and 9.49, respectively, which is consistent with the previous simulation results (i.e. 5.9 for Na, 7.2 for K, and 9.49 for Cs).37 As shown in Table 2, the coordination number for each type of the surface adsorbed ions is slightly larger than that for the solution. The discrepancy between the two types of ions is due to the fact that the ions on the surface can form coordination bonds with Os.

Table 2 The coordination number of cations on the C–S–H gel surface and in the C–S–H gel pores
Ion Na+ K+ Cs+
Inner Outer Pore Inner Outer Pore Inner Outer Pore
–Ow 4.43 5.46 5.73 3.82 6.95 6.95 4.80 8.08 9.29
–Os 1.50 0.40 0 3.75 0.62 0 4.63 0.86 0
–Cl 0 0.05 0.04 0 0.05 0.17 0 0.11 0.20
Total 5.93 5.91 5.77 7.57 7.62 7.12 9.43 9.05 9.49

Furthermore, adsorbed species were decomposed into the inner sphere and outer sphere ions and subsequently the CN were calculated. Taking CsCl solution as an example, while the silicate oxygen atoms occupied 50% of the nearest neighbors for the inner Cs+ ions, Os atoms only took up around 10% for the outer Cs+ ions of the hydration shell. In addition, the channel water molecules were the predominant neighbors of the inner adsorbed Cs+ ions and the surface water molecules mainly contributed to the outer sphere Cs+ ions of the coordinate atoms. The highly concentrated H-bonds network of the channel water molecules also strongly influenced the interfacial structure of the immobilized Cs+ ions. It implies the significant local structural discrepancy between the inner and outer adsorbed species. Table 2 also exhibits that a small number of chloride ions constitutes the nearest neighbors of the outer species and no chloride ions are observed near the inner cations. It indicates that anionic adsorption results from the ionic pairs by the outer adsorbed species. Fig. 8 clearly demonstrates the local structure differences of the Cs+ ions and their neighbors in three states. As shown in Fig. 8a, the inner adsorbed Cs+ ions, connected by oxygen atoms in two different silicate chains, are deeply rooted in the silicate channel. The moment one Cs–Os bond breaks, the Cs+ ions more probably reconnect with other oxygen sites that are rich in the silicate channel. On the other hand, the outer sphere Cs+ ions, which only connect with one Os atom, are more flexible to move around. Although the ions in the near-surface regions can form a stable coordination with the oxygen atoms in the C–S–H gel, most of the coordination is formed between ions and water molecules. The breakage of the Cs–Os bond can result in the dissociation of the Cs ions from the C–S–H surface. Hence, the immobilized effect of the C–S–H gel on the cations depends on the bond number and the bond stability of cation and oxygen. In an earlier report, the adsorption of cesium on cement mortar from aqueous solutions was studied by Volchek et al.,3 in which mortar was immersed in the solution of different initial concentrations. Based on the adsorption isotherm and activation energy analysis, they proposed that the chemisorption was the prevalent mechanism of the interaction between cesium ions and cement mortar. To further study the chemical adsorption sites of Cs+ ions in the silicate hydrate, the molecular information of the adsorbed Cs atoms has been investigated by extended X-ray absorption fine structure spectroscopy (EXAFS). The EXAFS study revealed that the inner-sphere Cs+ ions surface complexes coexisted with the outer-sphere surface complexes and the adsorbed sites were positioned nearby the edge of the basal oxygen cavity on the silicate tetrahedron sheet. The findings in our molecular dynamics study confirmed well with the results from the adsorption experiment and EXAFS analysis.23 Besides, a few number of chloride ions contributed to the neighboring cations both in the pore solution and near the calcium silicate surface.

image file: c7cp05437h-f8.tif
Fig. 8 Schematic diagram illustrating the coordination number of (a) inner sphere Cs+ species; (b) outer-sphere Cs+ species; and (c) Cs+ ions in the solution.

3.2 The dynamic property of ions in the C–S–H gel pores

3.2.1 Motion trajectory. To reflect the stability of adsorption of ions on the C–S–H gel surface, the trajectories in the xy plane of ions on the C–S–H gel surface were recorded. The black diamond contours are the surface adsorbed Si atoms, which are taken as the reference positions for ionic trajectories. The red contours are the motion trajectories of ions within 2 Å of the C–S–H surface. As shown in Fig. 9, the distribution of cationic trajectory is relatively concentrated and located around the vacant sites of silicate tetrahedra. Nevertheless, the contours of anions are much less ordered. This indicates that the C–S–H gel surface has a good immobilized capacity for cations adsorption. Furthermore, among the trajectories of these cations, the trajectories of Na+ are denser on the surface and show a stronger preference to overlap with those of the solid surface compared with those of other cations. It implies that Na+ ions are easier to absorb on the C–S–H gel surface than other cations.
image file: c7cp05437h-f9.tif
Fig. 9 The motion trajectory of ions projected in the xy plane of the C–S–H gel surface: (a) chloride ions in sodium chloride solution, (b) sodium ions in sodium chloride solution, (c) chloride ions in potassium chloride solution, (d) potassium ions in potassium chloride solution, (e) chloride ions in cesium chloride solution, and (f) cesium ions in cesium chloride solution.
3.2.2 Stability of chemical bonds. The time-correlated function (TCF)38 is utilized to quantitatively describe the strength of various chemical bonds connected between ions and atoms on the C–S–H surface. The TCF (C(t)) of the ionic–covalent bond is defined by eqn (1):
image file: c7cp05437h-t1.tif(1)
where δb(t) = b(t) − 〈b〉, b(t), which is a binary calculator, takes a value of 1 when the ionic pair (e.g. Cs–O) is bonded and otherwise 0, and 〈b〉 is the mean value of b over all simulation times and pairs. The chemical bonds break and form between the cations and atoms on the C–S–H surface, resulting in the chemical adsorption and desorption of the cations. If the chemical bond persists unchanged, TCF value will remain at the value of 1. On the contrary, the bond breakage results in decreasing TCF values; the more frequent the bond breakage, the lower the TCF value. By comparing the deviations from 1 in the TCF curves, the chemical strength of various bonds can be estimated. As shown in Fig. 10a, the chemical bond stability of the species in the pore solution is of the following order: Na–Ow > K–Ow > Cs–Ow > Cl–Ow > H-bond. It means that the relaxation time for water remaining in cations is longer than the anionic relaxation time. Also, the cations with relatively smaller hydration radii strongly associate with the surrounding water molecules than the larger cations, such as cesium ions. Comparing Fig. 10a and b, the TCF for the cation–Os reduces drastically than that for the cation–Ow, which declines dramatically. It reflects the strong binding strength from the oxygen atoms in the silicate chains.

image file: c7cp05437h-f10.tif
Fig. 10 Time correlated function for (a) cation–Ow and (b) cation–Os.

The evolution of C(t) describes the dynamics of ion–oxygen pair structural relaxation and its relaxation time τ can be obtained by integrating the C(t) function (eqn (2)):

image file: c7cp05437h-t2.tif(2)

As listed in Table 3, in the solution without the interaction with the silicate surface, the relaxation time for water molecules remaining in the first hydration shell of Na, K, and Cs is 22.8, 13.7, and 12 ps, respectively, which are quite close to those obtained for the bulk solution (i.e. 27 ps for Na, 15 ps for K, and 10.4 ps for Cs).22 On the other hand, the relaxation time for species in the gel pores increases to 40.1, 24.1, and 21 ps, which are elongated twice. It should be noted that more than 70% of the cations are adsorbed on the silicate surface. The relaxation time for water surrounding the adsorbed cations is significantly increased, as compared with those existing in the bulk solution. This implies that once the cations are adsorbed on the calcium silicate channel, it is very hard for the water molecules associated with the adsorbed species to escape from the cations. The elongation relaxation time can be explained by the fact that the neighbors of the inner adsorbed Cs+ ions mainly come from the channel water molecules connected with the Os atoms with a strong H-bond strength.

Table 3 The residence time of ions on the C–S–H gel surface and around water (ps)
Na K Cs
Ow Solution 22.84 13.73 11.96
Gel pore 40.10 24.10 21.00
Os ≫78.6 ≫75.6 ≫61.1

Furthermore, the relaxation time of ions connected with the Os atoms is utilized to characterize the adsorption strength of different cations. It should be noted that the integration for the TCF function is based on 100 ps and the real value is significantly longer than the calculated one. For the cesium ions, the relaxation time near the C–S–H surface is more than three times longer than that in the solution, indicating that once the cesium ions are captured by the calcium silicate hydrate, it would remain in the silicate channel for a long time due to the chemical and geometrical restriction from the surrounding oxygen atoms. For the inner-sphere Cs+ ions, the average number of Os neighbor is around four and the four chemical bonds with high strength can form a “cage” and inhibit the diffusion of the Cs+ ions, only allowing rotation and vibration in the fixed positions. This result can explain the concentrated trajectories observed for both cations and anions as shown in Fig. 9.

These results give a valuable guideline for the adsorbent material design. On the one hand, due to the geometrical restriction, the nanometer pores can dramatically restrict the mobility of Cs+ ions. In this respect, the porosity and the pore size distribution should be controlled properly to restrict the leaching of Cs+ ions. The addition of some minerals such as zeolite and tobermorite can help to improve the pore structure of the cement-based materials. The finesse and high surface area of minerals are capable of reducing large pores and capillaries formed in cement pastes.15 On the other hand, the material design is suggested to focus on developing a proper silicate skeleton by changing the Ca/Si ratio. In previous experimental studies, in order to enhance the immobilizing ability, pozzolanic additives were employed to mix with the cement paste.12,15 The pozzolanic reaction between the active amorphous silica and portlandite (a cement hydrated product) produces the C–S–H gel with a low Ca/Si ratio. The experimental results showed that the binding capability and the binding strength of Cs in the cement paste enhanced with the decrease in Ca/Si ratio.1 Besides, the residence time of Na and K in the C–S–H surface is quite longer than that of Cs+ ions, indicating that the large hydration size of the cations weakens the immobilizing effect of the C–S–H gel.

3.2.3 Mean square displacement and diffusion coefficient. Mean square displacement is utilized to measure the position deviation of atoms as compared with the original position, which can reflect the mobility of atoms in different states. MSD evolution with time is defined by the following equation (eqn (3)).
MSD(t) = 〈|ri(t) − ri(0)|2(3)
In eqn (3), ri(t) is the coordinate of the ith atom at time t and ri(t) is corresponding to the initial coordinate of atom i. In Fig. 11a, the temporal MSD functions for different species in the CsCl gel pores demonstrate that the mobility of water is faster than that of both Cs+ and Cl ions. Due to a small number of Cs+ and Cl ions, the ionic MSD demonstrates a larger fluctuation. In addition, the pure CsCl solution of 0.5 mol L−1 was also simulated and the MSD for Cs+ ions was also calculated for comparison and shown in Fig. 11b. The pronounced reduction of MSD values for Cs+ ions confined in the C–S–H gel pores quantitatively elucidates the immobilization role of the calcium silicate surface on the cations. The MSD function of Cs+ ions was further decomposed into MSDx, MSDy and MSDz components that characterize the anisotropic diffusivity of the Cs+ ions. In Fig. 11c, different from the continuous increase of the MSD along x and y directions, MSDz attained a plateau after 1000 ps. Furthermore, Fig. 11d compares the temporal MSD functions of Cs, K and Na and shows that the mobility of Na+ ions is quite slower than that of K+ and Cs+ ions. This can be partly explained by the long residence time of Na+ ions remaining in the calcium silicate surface.

image file: c7cp05437h-f11.tif
Fig. 11 (a) MSD evolution with time of different species in CsCl solution; (b) MSD of Cs+ ions in the gel pores and solution; (c) MSD along x, y and z directions of Cs+ ions in the gel pores; and (d) MSD of Na+, K+ and Cs+ ions in the gel pores.

Based on the MSD evolution with time, the diffusion coefficients were calculated for different species confined in the gel pores and pure solution. As shown in Fig. 12, the diffusion coefficient of water molecules deeply rooted in the silicate channel is around 0.2 × 10−9 m2 s−1, which is significantly reduced as compared with that of bulk water. The diffusion coefficient gradually increases to 1.0 × 10−9 m2 s−1 in the surface region and reaches around 2.0 × 10−9 m2 s−1 at 10 Å from the silicate surface, which is close to the bulk water value (2.3 × 10−9 m2 s−1). As listed in Table 4, the diffusion coefficient of cations on the C–S–H gel surface is dramatically smaller than that in the pore solution. This can be easily understood by the long relaxation time of the cation–oxygen bond discussed in the previous section. As compared with that of the cations in the solution, the diffusion coefficient for Na+, K+, and Cs+ in the gel pores decline in the proportion of 54.7%, 48.3%, and 36.5%, respectively. The cationic diffusivity reduction, in the dynamic respect, reflects the immobilizing effect of the calcium silicate hydrate. The immobilization capability of the C–S–H gel on the cations is enhanced with the increase in ionic radii.

image file: c7cp05437h-f12.tif
Fig. 12 Diffusion coefficient of water evolution with the distance from the calcium silicate surface.
Table 4 The diffusion coefficient of different cations on the C–S–H gel surface or in the C–S–H gel pores (10−9 m2 s−1)
Ow Cations
Solution Gel pore Solution Gel pore
CsCl 2.81 ± 0.01 1.80 ± 0.02 1.97 ± 0.06 1.25 ± 0.04
KCl 2.83 ± 0.01 1.71 ± 0.02 1.88 ± 0.06 0.93 ± 0.04
NaCl 2.83 ± 0.01 1.72 ± 0.02 1.57 ± 0.06 0.71 ± 0.04

4 Conclusion

Molecular dynamics simulations were carried out to investigate the molecular structure and dynamic behavior of ions on the C–S–H gel pores. This study provides a molecular-level insight on the immobilization mechanism of calcium silicate hydrate on the radioactive waste such as the Cs+ ions. Some conclusions have been made in the following aspects:

(1) The dreierketten distributed silicate chains of the C–S–H surface provide non-bridging oxygen sites to associate with the cations. Two types of adsorbed ions are observed in the C–S–H surface: the strong inner-sphere adsorbed cesium ions, on average, have 4.6 coordinated oxygen atoms in bridging and pair silicate tetrahedron and are strongly restricted in the vacancy between neighboring silicate chains; the weak outer-sphere adsorbed cesium ions can only be connected with bridging tetrahedron by one Cs–O bond. On the other hand, the adsorption of chloride ions repulsed by the negatively charged silicate surface is mainly attributed to the formation of the cation–anion ionic pairs near the interface.

(2) The diffusivity of the Cs+ ions confined in the C–S–H gel pores is reduced by 36.5% as compared with that of the species in the bulk solution. The dramatic reduction of the diffusion coefficient is mainly attributed to the strong Cs–O bond that extends the relaxation time of the ions remaining in the silicate channel.

(3) The C–S–H surface exhibits better immobilization effects on the cations with a smaller ionic radium for the monovalent ions. Structurally, the Na+ ions with a small hydration shell have a higher ability to penetrate into the surface cavity than the Cs+ and K+ ions with a larger size. Dynamically, the Na+ ions have stronger Na+–Os connections which are hard to be broken and reside in the silicate channel for a longer time.

Conflicts of interest

There are no conflicts to declare.


The financial support from National Natural science foundation of China under Grants 51508292, 51678317 and the China Ministry of Science and Technology under Grant 2015CB655100 is gratefully acknowledged.


  1. A. J. Allen, J. J. Thomas and H. M. Jennings, Composition and density of nanoscale calcium–silicate–hydrate in cement, Nat. Mater., 2007, 6(4), 311–316 CrossRef CAS PubMed .
  2. G. Bar-Nes, A. Katz, Y. Peled and Y. Zeiri, The mechanism of cesium immobilization in densified silica-fume blended cement pastes, Cem. Concr. Res., 2008, 38(5), 667–674 CrossRef CAS .
  3. R. Cygan, J. Greathouse, H. Heinz and A. Kalinichev, Molecular models and simulations of layered materials, J. Mater. Chem., 2009, 19(17), 2470–2481 RSC .
  4. R. Cygan, J.-J. Liang and A. Kalinichev, Molecular models of hydroxide, oxyhydroxide, and clay phases and the development of a general force field, J. Phys. Chem. B, 2004, 108(4), 1255–1266 CrossRef CAS .
  5. A. M. El-Kamash, M. R. El-Naggar and M. I. El-Dessouky, Immobilization of cesium and strontium radionuclides in zeolite-cement blends, J. Hazard. Mater., 2006, 136(2), 310–316 CrossRef CAS PubMed .
  6. F. P. Glasser and M. Atkins, Cements in radioactive waste disposal, MRS Bull., 1994, 19(12), 33–38 CrossRef CAS .
  7. S. Hong and F. Glasser, Alkali binding in cement pastes Part I. The C–S–H phase, Cem. Concr. Res., 1999, 29, 1893–1903 CrossRef CAS .
  8. D. Hou and Z. Li, Molecular dynamics study of water and ions transport in nano-pore of layered structure: A case study of tobermorite, Microporous Mesoporous Mater., 2014, 195, 9–20 CrossRef CAS .
  9. D. Hou, Z. Lu, X. Li, H. Ma and Z. Li, Reactive molecular dynamics and experimental study of graphene-cement composites: structure, dynamics and reinforcement mechanisms, Carbon, 2017, 115, 188–208 CrossRef CAS .
  10. D. Hou, Z. Lu, T. Zhao and Q. Ding, Reactive molecular simulation on the ordered crystal and disordered glass of the calcium silicate hydrate gel, Ceram. Int., 2016, 42(3), 4333–4346 CrossRef CAS .
  11. T. Iwaida, S. Nagasaki, S. Tanaka, T. Yaita and S. Tachimori, Structure alteration of CSH (calcium silicate hydrated phases) caused by sorption of caesium, Radiochim. Acta, 2002, 90(9–11), 677–681 CAS .
  12. A. Kalinichev, J. Wang and R. Kirkpatrick, Molecular dynamics modeling of the structure, dynamics and energetics of mineral–water interfaces: Application to cement materials, Cem. Concr. Res., 2007, 37(3), 337–347 CrossRef CAS .
  13. R. Kirkpatrick, A. Kalinichev and J. Wang, Molecular dynamics modelling of hydrated mineral interlayers and surfaces: structure and dynamics, Mineral. Mag., 2005, 69(3), 289–308 CrossRef CAS .
  14. T. M. Krishnamoorthy, S. N. Joshi, G. R. Doshi and R. N. Nair, Desorption kinetics of radionuclides fixed in cement matrix, Nucl. Technol., 1993, 104(3), 351–357 CrossRef CAS .
  15. S. H. Lee and J. C. Rasaiah, Molecular dynamics simulation of ion mobility. 2. Alkali metal and halide ions using the SPC/E model for water at 25 °C, J. Phys. Chem., 1996, 100(4), 1420–1425 CrossRef CAS .
  16. H. Ma and Z. Li, Realistic pore structure of Portland cement paste: experimental study and numerical simulation, Comput. Concr., 2013, 11(4), 317–336 CrossRef .
  17. H. Ma, D. Hou, J. Liu and Z. Li, Estimate the relative electrical conductivity of C–S–H gel from experimental results, Constr. Build. Mater., 2014, 71, 392–396 CrossRef .
  18. S. Mindess, J. Young and D. Darwin, Concrete, Prentice Hall, 2003 Search PubMed .
  19. M. Nakano, K. Kawamura and Y. Ichikawa, Local structural information of Cs in smectite hydrates by means of an EXAFS study and molecular dynamics simulations, Appl. Clay Sci., 2003, 23(1), 15–23 CrossRef CAS .
  20. S. Nosé, A molecular dynamics method for simulations in the canonical ensemble, Mol. Phys., 1984, 52(2), 255–268 CrossRef .
  21. S. B. Oblath, Leaching from solidified waste forms under saturated and unsaturated conditions, Environ. Sci. Technol., 1989, 23(9), 1098–1102 CrossRef CAS .
  22. H. Ohtaki and T. Radnai, Structure and dynamics of hydrated ions, Chem. Rev., 1993, 93(3), 1157–1204 CrossRef CAS .
  23. K. G. Papadokostaki and A. Savidou, Study of leaching mechanisms of caesium ions incorporated in Ordinary Portland Cement, J. Hazard. Mater., 2009, 171(1), 1024–1031 CrossRef CAS PubMed .
  24. R. J. Pellenq, A. Kushima, R. Shahsavari, K. J. Van Vliet, M. J. Buehler and S. Yip, A realistic molecular model of cement hydrates, Proc. Natl. Acad. Sci. U. S. A., 2009, 106(38), 16102–16107 CrossRef CAS PubMed .
  25. S. Plimpton, Fast parallel algorithms for short-range molecular dynamics, J. Comput. Phys., 1995, 117(1), 1–19 CrossRef CAS .
  26. R. A. Rahman, D. Z. El Abidin and H. Abou-Shady, Cesium binding and leaching from single and binary contaminant cement-bentonite matrices, Chem. Eng. J., 2014, 245, 276–287 CrossRef .
  27. B. Rotenberg, V. Marry, R. Vuilleumier, N. Malikova, C. Simon and P. Turq, Water and ions in clays: unraveling the interlayer/micropore exchange using molecular dynamics, Geochim. Cosmochim. Acta, 2007, 71(21), 5089–5101 CrossRef CAS .
  28. K. Sakr, M. Sayed and N. Hafez, Comparison studies between cement and cement–kaolinite properties for incorporation of low-level radioactive wastes, Cem. Concr. Res., 1997, 27, 1919–1926 CrossRef CAS .
  29. H. Sakuma and K. Kawamura, Structure and dynamics of water on Li+-, Na+-, K+-, Cs+-, H3O+-exchanged muscovite surfaces: a molecular dynamics study, Geochim. Cosmochim. Acta, 2011, 75, 63–81 CrossRef CAS .
  30. C. Shi and A. Fernandez-Jimenez, Stabilization/solidification of hazardous and radioactive wastes with alkali-activated cements, J. Hazard. Mater., 2006, 137(3), 1656–1663 CrossRef CAS PubMed .
  31. O. P. Shrivastava and R. Shrivastava, Cation exchange applications of synthetic tobermorite for the immobilization and solidification of cesium and strontium in cement matrix, Bull. Mater. Sci., 2000, 23(6), 515–520 CrossRef CAS .
  32. O. P. Shrivastava, T. Verma and P. K. Wattal, Intrinsic sorption potential of aluminum-substituted calcium silicate hydroxy hydrate for cesium-137, Adv. Cem. Based Mater., 1995, 2(2), 80–83 CrossRef CAS .
  33. I. F. Vasconcelos, B. A. Bunker and R. T. Cygan, Molecular dynamics modeling of ion adsorption to the basal surfaces of kaolinite, J. Phys. Chem. C, 2007, 111(18), 6753–6762 CAS .
  34. K. Volchek, M. Y. Miah, W. Kuang, Z. DeMaleki and F. H. Tezel, Adsorption of cesium on cement mortar from aqueous solutions, J. Hazard. Mater., 2011, 194, 331–337 CrossRef CAS PubMed .
  35. X. Wan, D. Hou, T. Zhao and L. Wang, Insights on molecular structure and micro-properties of alkali-activated slag materials: a reactive molecular dynamics study, Constr. Build. Mater., 2017, 139, 430–437 CrossRef CAS .
  36. M. Youssef, R. J. Pellenq and B. Yildiz, Docking 90 Sr radionuclide in cement: an atomistic modeling study, Physics and Chemistry of the Earth, Parts A/B/C, 2014, 70, 39–44 CrossRef .
  37. M. Youssef, R.-M. Pellenq and B. Yildiz, Glassy nature of water in an ultraconfining disordered material: the case of calcium–silicate–hydrate, J. Am. Chem. Soc., 2011, 133(8), 2499–2510 CrossRef CAS PubMed .
  38. Y. Zhou, D. Hou, J. Jiang and P. Wang, Chloride ions transport and adsorption in the nano-pores of silicate calcium hydrate: experimental and molecular dynamics studies, Constr. Build. Mater., 2016, 126, 991–1001 CrossRef CAS .

This journal is © the Owner Societies 2017