The adsorption mechanisms of organic micropollutants on high-silica zeolites causing S-shaped adsorption isotherms An experimental and Monte Carlo simulation study

The adsorption of organic micropollutants (OMPs) on high-silica zeolites is characterized by adsorption iso- therms with various shapes. The occurrence of an S-shaped adsorption isotherm indicates the lack of adsorption a ﬃ nity for OMPs at low, environmentally relevant equilibrium concentrations. In this study, S-shaped isotherms were observed during batch experiments with 2,4,6-trichlorophenol (TCP) and FAU zeolites. This is the ﬁ rst time that an S-shaped isotherm is reported for the adsorption of OMPs on high-silica zeolites. Monte Carlo (MC) simulations in the grand-canonical ensemble were used to obtain a better understanding of the mechanism of the S-shaped adsorption isotherms. From the MC simulation results, it was observed that multiple TCP molecules were adsorbed in the supercages of the FAU zeolites. It was found that the π - π interactions between TCP molecules give rise to the adsorption of multiple TCP molecules per supercage, and thus causing an S-shaped adsorption isotherm. Simulations also revealed that water molecules were preferentially adsorbed in the supercages and sodalite cages of the FAU zeolites. FAU zeolites with a higher Al content adsorbed a higher amount of water molecules and a lower amount of TCP, and showed less pronounced S-shaped isotherms.


Introduction
Zeolites are crystalline, microporous aluminosilicates with a welldefined 3-dimensional structure, composed of tetrahedral SiO 4 and AlO 4 clusters connected to each other by shared oxygen atoms. To compensate the charge imbalance caused by the Al content of the framework, exchangeable cations (usually alkali and alkaline earth cations) are located in the cavities of the structure. Due to these intrinsic characteristics, e.g. exchangeable cations and a well-defined pore structure, zeolites are widely used as catalysts, molecular sieves, and adsorbents for air-pollution remediation, gas separation and removal of volatile organic compounds [1][2][3][4][5][6].
An important characteristic of zeolites is the ratio of the Si and Al atoms contained. This ratio determines the level of hydrophobicity in the framework. Based on the Si to Al molar ratio, three types of zeolites can be identified, the low-silica (i.e., Si/Al < 2), the medium-silica (i.e., Si/Al = 2 -5) and the high-silica (i.e., Si/Al > 5) [7,8]. Low-and medium-silica zeolites exhibit a high ion exchange capability and are widely applied as water hardness control ingredients in detergents [9]. High-silica zeolites are particularly useful for removal of organic micropollutants (OMPs) from water, since their hydrophobic nature promotes the adsorption of OMPs instead of water [10][11][12]. As the framework type of the zeolite defines its unique structural features, including pore opening sizes, and cage and channel structures, it highly affects the adsorption efficiency of OMPs [13,14]. Various families of commercially available high-silica zeolites, including the Faujasite (FAU), Mordenite (MOR), Beta (BEA) and ZSM-5 (MFI) types, have been shown to be effective adsorbents for OMP removal from water [15][16][17][18].
In the process of OMP adsorption from water on solid adsorbents, the adsorption isotherm describes the relation between the equilibrium concentration of OMPs in water and the adsorption loading of OMPs on the solid adsorbents. Giles et al. [19] have divided the types of adsorption isotherms into the four groups shown in Fig. 1. The C-shaped adsorption isotherm (Fig. 1a) describes a linear increase of OMP adsorption loading with the equilibrium concentration of OMPs in water, which is often used for a narrow range of OMP concentrations or for very low OMP concentrations, i.e., from a few ng L −1 to several µg L −1 [20]. When the adsorbent has a limited number of adsorption sites, a plateau in the adsorption isotherm appears after a specific equilibrium concentration of OMPs and above, suggesting the saturation of the adsorbent. This results in the L-shaped (Fig. 1b) and H-shaped (Fig. 1c) isotherms. The steeper increase of OMP adsorption loading in the Hshaped isotherms compared to the L-shaped ones indicates that the zeolites exhibiting H-shaped isotherms have a higher affinity for OMPs.
In some adsorption processes, the isotherms exhibit a characteristic S-shape (Fig. 1d) [21]. This shape can be attributed to the higher attraction between OMPs at the surface of the adsorbent or to a complexation reaction of the metallic species of OMPs with ligands in water [20,22]. In general, the occurrence of S-shaped isotherms indicates the unfavorable OMP adsorption at low equilibrium concentrations.
Traditionally, the adsorption of OMPs on adsorbents are mostly obtained from batch experiments. It has been observed in several experimental studies that the adsorption of various OMPs from aqueous solutions on high-silica zeolites result in different types of adsorption isotherms. The adsorption of hydroxycinnamic acids, i.e., p-coumaric acid and ferulic acid, on high-silica zeolites were studied by Simon et al. [23]. At the equilibrium concentration range of 0 -1 g L −1 , BEA zeolites showed H-shaped isotherms, while FAU zeolites gave L-shaped isotherms, indicating that BEA zeolites, having smaller pore sizes, have a higher affinity for the hydroxycinnamic acids than FAU zeolites. The plateaus in both isotherms have been attributed to the saturation of the micropores of the zeolites with hydroxycinnamic acids. Zhang et al. [24] studied 2,4,6 trichlorophenol (TCP) adsorption on FAU zeolites. At an equilibrium concentration of 0 -30 mg L −1 , a C-shaped adsorption isotherm was obtained.
Although the adsorption experiments can be used to design or optimize zeolite-based adsorbents for the removal of OMPs from water, they do not provide any information on the exact adsorption mechanisms [25]. To that end, molecular simulation provides the necessary atomistic resolution needed to promote the deeper understanding of the governing physical-chemical mechanisms in adsorption equilibrium, and to further guide adsorption experiments. During the last three decades molecular simulations are actively performed in order to support experiments in various industrial and environmental processes [26][27][28]. In particular, Monte Carlo (MC) simulations with classical force fields have been used to study adsorption processes in porous media including metal-organic frameworks [29][30][31], zeolites [32][33][34][35], and activated carbon [36,37]. Due to the high interest in zeolite structures for adsorption-based applications in energy storage [38][39][40], gas separation [41,42], and wastewater treatment [43,44], numerous MC simulation studies have been carried out to explain the adsorption mechanisms of various compounds and mixtures [45][46][47][48][49].
2,4,6 trichlorophenol (TCP) is an organic compound widely used for the production of industrial products, e.g., fungicide and herbicide. As a result, TCP has been repeatedly reported to occur in water bodies [50][51][52] as an important OMP, among others. In the present work, the adsorption of TCP from water on high-silica FAU zeolites was studied by means of both adsorption experiments and MC simulation. FAU zeolites were chosen due to their large pore size and amount of micropores, which is shown to have a high adsorption loading for various OMPs [53][54][55]. The schematic representation of the framework of FAU is shown in Fig. 2.
In batch experiments of TCP adsorption on high-silica FAU zeolites, an S-shaped adsorption isotherm was obtained. To the best of our knowledge, this is the first time that an S-shaped isotherm has been reported for the adsorption of OMPs on high-silica zeolites. In this context, MC simulations were carried out to investigate the adsorption mechanisms of the occurring S-shaped isotherm. Moreover, this is also one of the very few MC simulation studies reporting the adsorption of organics from water on high-silica zeolites [58][59][60]. To understand the adsorption mechanism, causing the S-shaped adsorption isotherm, the experimentally measured and the computed adsorption isotherms were compared qualitatively. The relation between the amount of Al content in FAU zeolites and the occurrence of S-shaped isotherms was also discussed.

Materials
TCP standards for chromatography analyses were purchased from Sigma-Aldrich, the Netherlands. The basic physiochemical properties of TCP are listed in Table 1.
High-silica zeolite powders of FAU framework were provided by suppliers. The names, suppliers and Si/Al ratios of zeolites are listed in Table 2.

Adsorption experiments
Batch adsorption experiments were conducted in demineralized water. The FAU zeolites (30 mg L −1 ) were dosed in the 100 ml TCP aqueous solution with the TCP concentrations ranging from 0 to 50 µmol L −1 . After the equilibrium time of 24 h at room temperature (25 ± 1°C), the zeolites were separated from the TCP solution by membrane filtration (0.2 µm syringe filter, Whatman SPARTAN TM ). This equilibrium time is expected to be sufficient, as shown by Zhang et al. [24], who measured the adsorption of TCP on FAU30.

High performance liquid chromatography (HPLC) analyses
The concentration of filtrated TCP solution was determined by HPLC (Shimadzu, Japan) with a C18 column (Phenomenex® KINETEX, 4.6 mm) at 30°C. HPLC-grade acetonitrile (Sigma-Aldrich, the Netherlands) and ultra-pure water (Milli-Q ultra-pure water system) were mixed as the mobile phase with a ratio of 65:35 (v:v). The flow rate of the mobile phase was 1.0 ml min −1 . The wavelength of the UV detector was set at 280 nm.

Force fields
The force field parameters for TCP, water and zeolites were taken from the Generalized Amber Force Field (GAFF) [61], Extended Single Point Charge (SPC/E) [62] and Clay Force Field (ClayFF) [63] models, respectively. The choice of these force fields is based on the study by Narasimhan et al. [64] in which it was shown that a similar force field combination was able to reproduce experimental results of the adsorption of paracresol and water onto MFI zeolites reasonably well. As discussed earlier, the adsorption isotherms computed from MC simulations were used only to obtain physical insight into the adsorption mechanisms. To this end, no modifications in the force fields or cross interactions parameters between the different species were applied to achieve better agreement with the experimental measurements. Moreover, exhaustive simulations using various other force field combinations was also not the scope of this study. All the force field parameters used in this study are listed in Table S1 in the Supporting Information.

Simulation details
All MC simulations were carried out using the Cassandra open source software package V1.2 [65]. In the Cassandra software package the chemical potential, which is imposed in the GCMC simulations is shifted with a value containing the partition functions of the molecules which are inserted in the simulation box. To be able to convert this shifted chemical potential to equilibrium concentration additional simulations are performed. The details of the applied method is shown in Section S2 in the Supporting Information.
Short-range van der Waals interactions were considered by the (12-6) Lennard-Jones potential (see Equation S1 in the Supporting Information). For the mixed pair potentials the Lorentz-Berthelot mixing rules [66] were used (see Equation S2 in the Supporting Information). Long-range electrostatic interactions were considered by the Ewald summation method [67] with a relative precision of 10 −5 . For the LJ and electrostatic interactions a cutoff radius of 14 Å with analytic tail correction was applied. In all simulations, 8 unit cells per simulation box (a 2x2x2 supercell) were used and periodic boundary conditions were imposed in all direction. The zeolite framework was considered rigid in all simulations. This approach is often being applied in simulations of nanoporous materials to prevent the necessity of excessive computational efforts [68]. Zeolite frameworks with specified Si/Al ratios were created based on the structures reported by Hriljac et al. [69]. The desired Si/Al ratios were achieved by randomly exchanging silicon atoms with aluminum atoms while obeying Löwnstein's rule which states that the Al -O -Al bonds are prohibited in   zeolites [69]. The number of sodium atoms were adjusted in accordance to the number of aluminum atoms to ensure the electro-neutrality of the framework. The sodium atoms were fixed at their crystallographic position in order to simplify the simulation scheme. It is important to note that the choice of fixing the sodium or allowing it to move in the zeolite framework may vastly affect the computed adsorption isotherms as shown by Calero et al. [70] in alkane adsorption onto FAU zeolites. The effect of fixing the sodium ion to the framework has not yet been studied in MC simulations of adsorption from the liquid phase. Thus, further investigation is needed to show the exact magnitude of this effect on the computed adsorption isotherms.
To obtain the adsorption isotherms, Configurational-Bias Monte Carlo (CBMC) simulations in the grand-canonical (μVT) ensemble (GCMC) were performed [67]. The following types of trial moves were used: translations (~12%), rotations (~12%), partial regrowth (~10%) and molecule exchanges with the reservoir (~66%). In all MC runs an equilibration period of 5,000,000 MC steps was performed. After the equilibration, production runs of 15,000,000 MC steps were carried out, from which ensemble averages were calculated.
The MC simulations were performed for the FAU250, FAU50, FAU40 and FAU30 frameworks. The Si/Al ratios and the number of Al atoms in the framework are listed in Table 3. The simulated zeolite frameworks cover a wider Si/Al range compared to the batch experiments. The Si/Al in simulations was in the range 5 -250, while in experiments was in the range 30 -250.

TCP adsorption on FAU zeolite
The experimental results of TCP adsorption isotherms on FAU zeolites with varying Al content are shown in Fig. 3. The adsorption isotherms of TCP on FAU250, FAU50 and FAU40 had an S shape, while the adsorption isotherm of FAU30 did not reach a plateau in the experimental concentration range. It is assumed that the observed adsorption isotherm was the beginning stage of S-shaped isotherm and the maximum adsorption loading of TCP on FAU30 will be achieved with the increase of TCP equilibrium concentration. As observed from the adsorption plateaus of S-shaped isotherms, the maximum adsorption loading of TCP on FAU250, FAU50 and FAU40 followed the order FAU250 > FAU50 > FAU40. This indicates that TCP adsorption on FAU zeolites, including the TCP loading and the shape of the adsorption isotherms, was driven by the Si/Al ratio of the zeolite. In the studied equilibrium concentration range of 0 -20 µmol L −1 the highest TCP adsorption loading on FAU30 was < 400 µmol g −1 .
In Fig. 4, the simulated adsorption isotherms of TCP on FAU zeolites with varying Al contents are shown. As shown in Fig. 4, the simulated adsorption isotherms of all five FAU zeolites, including FAU5 (which was not tested in batch experiments), were S-shaped. The adsorption loadings of TCP on FAU zeolites followed the same order as in the experiments: FAU250 > FAU50 > FAU40 > FAU30 > FAU5. As can be seen in Fig. 4, the adsorption isotherms for FAU250 and FAU50 computed from MC simulations exhibited almost identical maximum loading to the respective experimentally measured isotherms. By incorporating more Al atoms in the framework, the S-shape of the adsorption isotherms became less pronounced. This is also in line with the experimental measurements shown in Fig. 3, indicating that MC simulations captures the effect of Si/Al ratio on the shape of the adsorption isotherm.
The adsorption loading of TCP on FAU40 was underestimated in the simulation, showing a deviation of~300 mg g −1 from experimental results. The simulated adsorption isotherms of FAU30 zeolites was Sshaped with a maximum TCP loading of~1200 mg g −1 (Fig. 4), while the adsorption loading of FAU30 obtained from experiments was~400 mg g −1 , which is observed from an C-shaped isotherm without adsorption plateau (Fig. 3). The equilibrium concentration at which the adsorption step occurs in the isotherms computed from MC simulations is not as pronounced as in the experimentally measured ones, and moreover, it seems to deviate quite significantly.
The possible reason for the discrepancies between experiments and MC simulations is the force fields representing the adsorbent and adsorbates, as well as the fact that sodium ions are fixed in the FAU crystal. Since there were no prior MC simulations of the co-adsorption of water and TCP onto zeolites, the force fields representing water, TCP and zeolites were chosen based on the study by Narasimhan et al. [64], where the adsorption of water and paracresol onto MFI zeolites was studied. As stated earlier, no modification or refitting of the individual force fields was performed in the current study. However, such modifications in order to improve the accuracy of the simulations, can be realized in several ways: e.g. by optimizing the Lennard-Jones parameters or atomic charges, by explicitly including polarization effects,   and by modifying the intermolecular interactions between Al atoms and the adsorbates. Although such modifications may give a better agreement between the simulation and experimental results, their implementation requires significant computational effort which is out of the scope of this study.

The adsorption mechanisms behind the S-shaped adsorption isotherms
Since the adsorption isotherms for FAU250 computed from MC simulations exhibited almost identical maximum loading to the experimentally measured adsorption isotherms, the MC simulation results of FAU250 were further used to understand the adsorption mechanisms of S-shaped adsorption isotherm.
In Fig. 5, the average number of adsorbed TCP molecules in the supercages of FAU250 are plotted as a function of the equilibrium concentration of TCP. The number of adsorbed TCP molecules increased with the increase of TCP equilibrium concentration and the Sshaped curves is observed from Fig. 5. The S-shaped curves were qualitatively characterized by a division of three stages. At the low coverage range (equilibrium concentration < 0.15 µmol L −1 ), a minimal amount of TCP molecules (close to 0) were adsorbed on the FAU250 zeolites. There was 0 to 1 TCP molecule adsorbed in the supercages. The average number of adsorbed TCPs increased considerably at the transition stage (equilibrium concentration of 0.15 to 10 µmol L −1 ). At the stage of high coverage (equilibrium concentration > 10 µmol L −1 ), the maximum number of adsorbed TCP is observed from the plateau.
As shown in Fig. 5, the S-shaped curves are observed for both supercages without Al and supercages with Al, while the maximum number of adsorbed TCP in two types of supercages varied. A supercage provided accommodation to approximately 2 TCP molecules in Al containing cages and 2.5 TCP molecules in the none-Al containing cage, respectively. In the supercages with no Al atom, a maximum of 3 TCP molecules could be adsorbed.
The simulated arrangement of the adsorbed TCP molecules with water molecules in the supercages without Al are shown in Fig. 6. When 1 TCP molecule was adsorbed in the supercage, TCP preferably excluded the water molecules in between and was arranged close to the wall of the cage (Fig. 6a). In the supercages with 2 or 3 adsorbed TCP molecules, less water molecules were adsorbed and the TCP molecules obtained a parallel configuration, with benzene rings facing each other ( Fig. 6b and 6c).
To obtain further information about the structure of adsorbed molecules in the framework, the radial distribution function (RDF) was calculated [66]. The RDF, usually represented as g(r), defines the probability of finding a particle at distance r from another tagged particle. In Fig. 7, the RDF between the center of mass of benzene rings on the FAU250 framework at TCP equilibrium concentration of 77.6 µmol L −1 is shown. From Fig. 7, it is observed that the characteristic distance between two TCP molecules was~3.7 Å, which is a typical packing distance between ππ stacked aromatic groups (3.4 Å − 3.8 Å). This is in line with literature [71,72] as TCP molecules in aqueous solutions could experience π -π interactions, as a form of lateral interactions in the supercages. The lateral interaction considerably promoted TCP adsorption at the transition stage (Fig. 5), and led to the S-shaped adsorption isotherm. This finding is in line with the work of Walton et al. [73]. who attributed the steps in the adsorption isotherms for CO 2 on metal-organic frameworks to the sudden clustering of CO 2 molecules in the pores.
The π-π interaction could occur between two TCP molecules with C -C π electrons, originating from benzene rings of TCP molecules. The C -C π electrons of TCP molecules could shift from the center of the benzene ring due to electron donating of hydroxyl groups and electron withdrawing of chlorine atoms to the conjugated π electron. Thus, two TCP molecules could form a ππ interaction with an offset stacked conformation other than a perfect face-to-face alignment, as clearly shown from the simulation snapshots in Fig. 6. In this conformation, most of the surface of TCP molecules were covered by each other and the ππ interaction between TCP molecules is expected to be strong [74]. Although there is no explicit term to account for the π-π interactions in the force fields used, the use of fixed-point charges can reproduce these interactions [75,76] as shown by the characteristic distances and orientation of the TCP molecules in Figs. 6 and 7.

The effect of water adsorption on the occurrence of S-shaped adsorption isotherms and TCP adsorption
In Fig. 8, the number of adsorbed water molecules in supercages of FAU250 is plotted as a function of the equilibrium concentration of TCP. Fig. 8 shows that the number of water molecules in the supercages of FAU250 decreased with equilibrium concentration of TCP. Al-containing cages had a higher number of adsorbed water molecules than the cages without Al. At the higher equilibrium concentration (> 77 µmol L −1 ), about 8 water molecules per cage were adsorbed in the supercages with Al atom, while approximately 5 molecules per cage were adsorbed in the supercages without Al atom. Compared to water adsorption in sodalite cages (0 -3 water molecules per cage, Fig. 10), more water molecules were adsorbed in supercages (8 -25 water molecules per cage, Fig. 8). This can be attributed to the much less space inside sodalite cages than supercages [56].
The arrangement of the adsorbed water and TCP molecules in the Al-containing supercages are shown in Fig. 9. The distance between hydrogen atoms from water and the oxygen atoms of zeolites, which are connected to Al atom, was computed to be 1.7 Å. In the Al-containing supercages with one TCP molecule adsorbed, a total number of 19 water molecules were adsorbed (Fig. 9a), while 10 water molecules were adsorbed in the supercages with 2 adsorbed TCP molecules (Fig. 9b). In the none Al-containing supercages, however, 14 water molecules were co-adsorbed along with one TCP molecule, and 3 water molecules were co-adsorbed along with two TCP molecules (Fig. 6). This is in line with the results presented in Fig. 9, where it is shown that more water molecules were adsorbed in the supercages with Al. The Alcontaining supercages with 3 adsorbed TCP molecules were not found in the simulation box, since water molecules hinder the TCP adsorption due to the limited space in the supercages.
The number of adsorbed water molecules in sodalite cages of FAU250 zeolites is plotted as a function of the equilibrium concentration of TCP in Fig. 10. As shown in Fig. 10, water adsorption in sodalite cages was not affected by the equilibrium concentration of TCP. The sodalite cages with Al atom had an average number of approximately 1.5 water molecules per cage with a variation from 1 to 2 water molecules per cage, while the sodalite cages without Al atom exhibited almost no ability for water adsorption. The RDF between the hydrogen atom from water and the oxygen atom connected to Al atom of FAU250 zeolites was calculated to determine the structures of adsorbed water molecules on FAU250 zeolite. As shown in Fig. 11a, one peak is observed at the distance of 1.7 Å, which is the typical H-bond length [77]. Thus, we can safely conclude that water molecules were being adsorbed on FAU zeolites by forming H-bonds with the framework. Narasimhan et al. [60] indicated that the existence of water clusters could weaken the interaction of water molecules with zeolites, as well as the interaction of OMPs with zeolites, which, however, would strengthen the interactions between OMPs and water. The two wide peaks at distances of 2.9 Å and 4.0 Å shown in Fig. 11a, indicate that water molecules which do not form H-bonds with the FAU framework formed water clusters.     As shown in Fig. 11b, the RDF between the O atom of adsorbed water molecules and Al atom of zeolite framework has a strong peak at distance of approximately 3.4 Å. Bolis et al. [78] suggested that water molecules were adsorbed on zeolites by interacting with the -Si-O-Sigroups of zeolites. The distance of 3.4 Å between Al atom of zeolites and water molecules, shown in Fig. 11b, clearly indicate that there was an additional driving force for water adsorption on the supercages (Fig. 8) and sodalite cages (Fig. 10) of the FAU zeolite. The additional driving forces from Al atoms might cause an extra number of water molecules to be adsorbed in the supercages (Fig. 8) and sodalite cages (Fig. 10). Due to the additional water adsorption on Al atom, FAU zeolite with higher Al content (lower Si/Al ratio) had a lower TCP adsorption loading and less pronounced S-shaped adsorption isotherms (Figs. 3 and 4). Zhang et al. [24] studied TCP adsorption on FAU zeolites with Si/Al ratio 30 and observed no plateau from the adsorption isotherm, which confirms the finding in the present study.
The supercages of FAU zeolite with large pore opening and inner sizes provided the basic conditions for the accommodation of multiple TCP molecules and their lateral interaction in the presence of water molecules, which was crucial for the occurrence of the S-shaped adsorption isotherms. In the zeolite frameworks with channels, the adsorbed OMPs preferred to be located in the channels and channel intersections, whereas water molecules preferably adsorbed in the vicinity of the OMP molecules [59,79,80]. Due to the hindering of water molecules, the lateral interaction of adsorbed OMP molecules would be negligible. Therefore, the S-shaped adsorption isotherm was not observed in previous literature, which studied OMP adsorption on zeolites frameworks with channels, e.g. BEA and MFI type zeolites [81][82][83].

Conclusions
In this study, MC simulations were used to explain the mechanism of S-shaped adsorption isotherms of TCP adsorption on high-silica FAU zeolites. This study shows that the occurrence of S-shaped adsorption isotherms was attributed to the lateral interaction of TCP, e.g., the π-π interaction. The supercages of FAU zeolites provided a possible accommodation for multiple TCP molecules as a potential condition of lateral interactions of TCP. Water molecules were adsorbed in the sodalite cages and supercages of FAU zeolites by forming H-bonds with the framework of zeolites. Compared to the none Al-containing supercages, more water molecules were stably adsorbed as water clusters in the supercages of FAU zeolites with Al content. An additional amount of adsorbed water molecules in Al-containing supercages could inhibit TCP adsorption and could prevent the occurrence of an S-shaped adsorption isotherm. FAU zeolites with higher Si/Al ratio, therefore, had evident S-shaped adsorption isotherms, and, meanwhile, achieved higher TCP adsorption loadings compared to FAU zeolites with lower Si/Al ratios. MC simulation has thus been proven to be a powerful tool to study the adsorption of OMPs in water to support experimental results. Performing MC simulation will further allow investigating the adsorption mechanisms of OMP adsorption from the view of molecular interaction between OMPs, OMP and adsorbents as well as water and adsorbents.
The unfavourable adsorption of OMPs have been widely observed from OMP adsorption on solid adsorbents. The low adsorption capacity would be followed by a steep increase of OMP adsorption capacity at a certain equilibrium concentration, showing the typical feature of Sshaped isotherms. This study suggests that the increase of OMP adsorption capacity was caused by the interaction between OMPs. Except for the π-π interaction between TCP molecules in this study, various types of OMP interactions might exist as driving forces for OMP adsorption, since the physicochemical properties of OMPs also vary.