CHINESE JOURNAL OF CHEMICAL PHYSICS VOLUME 25, NUMBER 4 AUGUST 27, 2012 ARTICLE Absorption and Structural Property of Ethanol/Water Mixture with Carbon Nanotubes Sheng-ping Du, Wen-hui Zhao, Lan-feng Yuan ∗ Hefei National Laboratory for Physical Sciences at Microscale, University of Science and Technology of China, Hefei 230026, China (Dated: Received on June 1, 2012; Accepted on June 8, 2012) Molecular dynamics (MD) simulations are performed to study the structure and adsorption of ethanol/water mixture within carbon nanotubes (CNTs). Inside the (6,6) and (10,10) CNTs, there are always almost full of ethanol molecules and hardly water molecules. Inside wider CNTs, there are some water molecules, while the ethanol mass fractions inside the CNTs are still much higher than the corresponding bulk values. A series of structural analysis for the molecules inside and outside the CNTs are performed, including the distributions of radial, axial, angular density, orientation, and the number of hydrogen bonds. The angular density distribution of the molecules in the first solvation shell outside the CNTs indicates that the methyl groups of ethanol molecules have the strongest interaction with the carbon wall, and are pinned to the centers of the hexagons of the CNTs. Based on the understanding of the microscopic mechanism of these phenomena, we propose that the CNTs prefer to contain ethanol rather than methanol. Key words: Ethanol/water, Carbon nanotube, Adsorption, Molecular dynamics simulation lations [16]. Methanol molecules confined in CNTs tend to form a single file in thinnest CNTs or helices in slightly wider tubes, and when the CNT diameter is larger than 14 Å, the distribution of methanol molecules is close to that in bulk [17]. Liu et al. studied the microscopic structure and transport behavior of methanol-water solutions within an armchair CNT [18]. They found that CNTs preferentially absorbed methanol over water molecule for an equimolar mixture of methanol and water, and demonstrated nanoimmiscibility of these molecules in CNTs. Zheng et al. studied the transport of methanol-water mixtures through CNTs under a chemical potential gradient [19]. By comparing flux through hydrophobic and hydrophilic nanotubes, they showed that solutions entered hydrophilic nanotubes more easily and the diffusion in hydrophobic pores was faster, and methanol had higher flux than water through both hydrophobic and hydrophilic nanotubes. The effects of cosolvents on hydration of interior of CNT with a small diameter have been studied by Yang et al. [20]. They found that methanol or trimethylamine N -oxide dehydrated the CNT, while urea improved hydration of the CNT interior. Recently, we have found that the CNTs showed very high adsorption selectivity of methanol over water for the methanol/water mixture by MD simulations [21]. By analyzing microscopic structures of the fluid mixtures, we find that the selective adsorption roots in a cooperative effect of the van der Waals (vdW) interaction between CNT and the methyl groups of CH3 OH molecules as well as the hydrogen bonding interaction I. INTRODUCTION Carbon nanotubes (CNTs) are discovered in 1991 [1], and have been paid more and more attention because of their unique electrical, optical, and mechanical properties [2, 3]. In the last decade, many interesting properties of matter confined in carbon nanotubes which are very different from those in bulk have been found [4, 5]. Although CNT is always hydrophobic, many studies have demonstrated that water can fill and pass through the inner space of open-ended CNTs quickly [6−12]. Water solution with CNTs is a popular system, and many special behaviors of water have been found [13]. Therefore, water-filled CNTs hold the promise for applications in nanofluidic devices and proton storage [14, 15]. The design of CNT-based nanodevices requires a thorough understanding of the interactions between nanotubes and their surrounding media, as well as the ability to control these interactions. Molecular dynamics (MD) and Monte Carlo (MC) simulations are very useful methods to study the structural and dynamical properties of fluid confined in nanotubes. For example, Shao group found that the structure of ethanol confined in single-walled carbon nanotubes (SWCNT) changed with the diameters of CNTs by MD simu- ∗ Author to whom correspondence should be addressed. E-mail: [email protected] DOI:10.1088/1674-0068/25/04/487-493 487 c °2012 Chinese Physical Society 488 Chin. J. Chem. Phys., Vol. 25, No. 4 Sheng-ping Du et al. TABLE I The Lennard-Jones parameters and partial charges for ethanol, water, and carbon nanotubes. M is a point on the bisector of the HOH angle, 0.015 nm from the oxygen atom towards the hydrogen atoms. H2 O C2 H5 OH CNT Site O H M CH3 CH2 O H C σ/µm 315.4 0.0 0.0 377.5 390.5 307.0 0.0 340.0 ε/(J/mol) 648.5 0.0 0.0 866.1 493.7 711.3 0.0 232.8 q/e 0.0 0.52 −1.04 0.000 0.265 −0.700 0.435 0.0 (a) (b) FIG. 1 Snapshots of the simulation box for CNT (6,6) (a) side view and (b) top view. among water and methanol molecules. In this work, we extend the system to ethanol/water mixture, investigating whether there is similar selectivity for ethanol. A positive answer is indeed given by our numerical experiments. Furthermore, the mass fraction (MF) of ethanol inside CNTs is even higher than that of methanol under the same conditions. II. SIMULATION DETAILS In our simulations, an armchair CNT with open ends is immersed into an enthanol aqueous solution. The CNTs and their diameters are: (6,6), 8.14 Å; (7,7), 9.49 Å; (8,8), 10.85 Å; (9,9), 12.16 Å; (10,10), 13.56 Å; (15,15), 20.34 Å; and (20,20), 27.12 Å. For the ethanol molecules, the united-atom (UA) model with the OPLS force field [22, 23] is used, which can reproduce well the properties of bulk ethanol [24]. The OPLS/UA potential approximates the ethanol molecule by four interaction sites: one for the methyl group, one for the methylene group, one for the oxygen atom, and one for the hydrogen atom. As for the water molecules, the TIP4P model [25] is used. As for the CNT, each carbon atom is taken as a neutral particle with a potential for the carbon atoms of graphite. The intermolecular interactions are described by combination of Lennard-Jones 12-6 potential and Coulombic potential: "µ ¶12 µ ¶6 # σij σij qi qj U (rij ) = 4εij (1) − + rij rij rij where rij is the distance between atoms i and j in different molecules, qi is the charge assigned to atom i, and σij and εij are the Lennard-Jones parameters with the combination rules: εij =(εi εj )1/2 , σij =(σi σj )1/2 . The parameters of all the interaction centers are shown in Table I. An ethanol/water box of 5.0 nm×5.0 nm×5.0 nm with periodic boundary condition is employed (see Fig.1). The ethanol mass fraction in bulk is taken as 20%, which corresponds to mole fraction of 9%. DOI:10.1088/1674-0068/25/04/487-493 FIG. 2 The mass fraction of ethanol in carbon nanotubes with bulk mass fraction of 20% as a function of the tube diameter. The open-ended CNT (2.5 nm in length) is placed along the z axis and fixed in the center of the mixture reservoir. To test whether the CNTs prefer ethanol to methanol, we also perform a series of simulations for methanol/ethanol mixture, where the ethanol mass fraction is taken as 10%. In these simulations, the (6,6) to (10,10) CNTs are used. The simulations are performed in NVT ensemble with the program GROMACS 3.3.1 [26]. The particle-mesh Ewald (PME) algorithm [27] is used for long range electrostatic interactions. All the simulations last 12 ns with a time step of 1 fs. The cutoff for Lennard-Jones interactions is taken as 1.1 nm. The temperature is maintained at 300 K by the Berendsen algorithm with the time constant of 0.1 ps [28]. The SHAKE algorithm [29] is used to constrain the intramolecular bonds of ethanol and water molecules. The carbon atoms of CNT are kept at the initial positions to fix the nanotube. Energies and coordinates are saved every 200 fs. III. RESULTS AND DISCUSSION A. MF of ethanol within CNTs Figure 2 gives the MF of ethanol in CNTs with different diameters. Inside the thinnest CNTs (6,6) and (7,7), there is nearly 100% ethanol. This is also reflected from c °2012 Chinese Physical Society Chin. J. Chem. Phys., Vol. 25, No. 4 Ethanol/Water Mixture with Carbon Nanotubes 489 FIG. 3 The radial density distribution of the molecules in the box for (6,6), (7,7), (8,8), and (10,10). the radial distribution function (RDF) of water which has no peak inside these CNTs (Fig.3). As the CNT diameter increases, the mass percentage of ethanol decreases gradually. When the diameter reaches 1.356 nm (the (10,10) CNT), the ethanol MF is 97%, which is still a very high ratio. Therefore, all these thin CNTs can be perfectly used to purify ethanol from water. In the two wider CNTs, the inner ethanol MF is 74.1% for (15,15) and 54.0% for (20,20), respectively, which are also much higher than the outer value 20%. Afterwards, the structural analyses will be performed mainly on the CNTs with nearly pure ethanol, i.e., (6,6) to (10,10). Comparing with the corresponding results of the methanol/water-CNT systems with the bulk methanol MF of 20% [21], we find that inner MFs of ethanol are higher than those of methanol. Therefore, it is an interesting question to ask whether the CNTs prefer to adsorb ethanol rather than methanol. To examine this, we perform similar MD simulations for methanol/ethanol mixture with ethanol MF of 10%. The ethanol MFs within the CNTs (6,6), (7,7), (8,8), (9,9), and (10,10) are 30%, 30%, 23.7%, 17%, and 10%, respectively. Thus, as we expected, the CNTs indeed prefer ethanol to methanol, although the selectivity is much lower than the selectivity of alcohol vs. water and drops quickly with the increase of the CNT diameter. B. Radial density distributions function The fluid structure within a nanotube can be well described by the atomic density distributions. From the DOI:10.1088/1674-0068/25/04/487-493 radial density distribution (RDF) (see Fig.3), we can see that there are a few water molecules in the (8,8), (9,9), and (10,10) CNTs. However, in the (6,6) and (7,7) CNTs there are full of ethanol molecules and no water molecule. In every profile, we can find two groups of density peaks which correspond to the molecules surrounding the nanotubes (inside and outside). The peak of methyl is sharper than that of hydroxyl, indicating a strong interaction between the methyl groups and the nanotube. We also find a second peak of water close to the tube center (r=0) in the (10,10) CNT. This layer structure of water within wide nanotube was also observed by Wang et al. [30] and in our work on methanol/waterCNT [21]. The distance of the nearst peak of the oxygen atom of water (OW) from the carbon wall is around 0.33 nm (close to σC−OW ), the same as the distance in the (8,8) and (9,9) CNTs. The similar distance between the wall and the peak of RDFs of water molecules [30] and oxygen molecules [31] inside various CNTs suggests that this distance has a close relationship with the parameter σC−O . Inside the (6,6) CNT, the peak of the methyl group is nearly located at r=0, closer to the center of CNT than that in the other nanotubes. Then we calculate the distance between the peak position of CH3 and the CNT wall dCH3 and that of the OH group dOH for the molecules inside the nanotubes. As we expected, the values of dCH3 are always around 0.35−0.36 nm, close to σC−CH3 (0.358 nm) except for the (6,6) CNT, which is 0.39 nm. The exception of the (6,6) CNT comes from c °2012 Chinese Physical Society 490 Chin. J. Chem. Phys., Vol. 25, No. 4 the confinement effect of this narrowest nanotube with diameter of 4.07 nm. On the other hand, the value of dOH changes greatly with the increase of diameter of CNTs. These data are the same as those for pure ethanol within CNTs by Shao et al. [16]. That is to say, the water molecules within the nanotubes (if there are any) have little effect on the distribution of ethanol. Therefore, we can ascribe the behavior of the methyl groups within nanotubes to the wall-fluid interactions, while ascribing the behavior of the hydroxyl groups to the fluid-fluid interactions. Outside the CNTs, both the ethanol and water show two groups of peaks with every CNT. For the (6.6) CNT, the first peak of OW outside the CNT is close to r=0.73 nm, and the second is at r=1.1 nm. We can say that the outer molecules form a two-layer structure. The first layer consists of a strong peak of ethanol group and a weak peak of water, which suggests that the interaction between the wall and ethanol is stronger than that between the wall and water. Besides, without the confinement effect of the CNTs, the values of dCH3 for the outer molecules are all around 0.35 nm, including that for the (6,6) CNT, 0.351 nm. The distances between the water O atoms and the tube wall are always around 0.33 nm for different CNTs, which is close to σC−OW (0.327 nm). On the other hand, the peaks of the second layer at r=1.1 nm are quite flat. This is understandable, since as the molecules are far from the carbon wall, the interaction between them becomes weak. A graphene sheet can be regarded as a CNT with diameter of infinity. So with the increase of CNT diameter, the distribution of outside molecules will become similar and converge to the case of graphene. C. Hydrogen bonds inside CNTs As in pure ethanol and water, hydrogen bonds play an important role in the behavior of ethanol/water mixture. Here we use the geometrical definition of hydrogen bond proposed by Luzar et al. [32], i.e., a hydrogen bond between two molecules should fulfill these two conditions: (i) the distance between the oxygen atoms of both molecules is smaller than a threshold distance 0.35 nm, (ii) the bond angle between the O−O direction and the molecular O−H bond of the donor is less than a threshold angle 30◦ . Figure 4 shows the average number of hydrogen bonds per molecule (hnHB i) of ethanol/water molecules within CNTs as a function of the CNT diameter. The results are close to those of pure ethanol within CNTs. The number of hydrogen bonds per molecule for CNTs with diameters less than 1.0 nm ((6,6) and (7,7)) is close to 1, nearly half of the value in bulk ethanol. This reflects that the ethanol molecules in these CNTs form single-file structures (more detailed discussion will be present in the next subsection). When the CNT diameter becomes larger, the number of hydrogen bonds per DOI:10.1088/1674-0068/25/04/487-493 Sheng-ping Du et al. FIG. 4 The average number of hydrogen bonds per molecule of the confined molecules as a function of the nonotube diameter. (6,6) (7,7) (8,8) (9,9) (10,10) FIG. 5 The snapshots of the molecules inside the CNTs (6,6), (7,7), (8,8), (9,9), and (10,10). molecule becomes close to 2, i.e., bulk-like. D. Axial density distributions Zhou et al. showed the axial density profile of the water within CNTs [33], and found that water molecules confined in the (8,8) and (9,9) CNTs have marked a wave-like pattern. Here we see that the ethanol molecules show two different ways of filling CNTs which are the “single-file mode” and “ring mode”, like the cases of pure water [33]. In our simulations (Fig.5), the ethanol molecules within the (6,6) and (7,7) CNTs form a sigle-file structure. In the (9,9) and (10,10) CNTs, the methyl groups of ethanol form ring-like structures around the nanotube, consistent with the results of c °2012 Chinese Physical Society Chin. J. Chem. Phys., Vol. 25, No. 4 Ethanol/Water Mixture with Carbon Nanotubes 491 FIG. 6 The axial density distribution of molecules inside (a) (6,6) and (b) (7,7) CNTs. FIG. 7 The orientation distribution of (a) the CH3 -CH2 bond, (b) the O−H bond, and (c) the dipole moment. Shao et al. [16]. The (8,8) CNT is a special case, where an ordered structure with two parallel single-file chains takes place. For the ethanol molecules within the (6,6) CNT, the axial density distribution is a wave-like profile with a quasi-period. In Fig.6(a) we see that there are four density peaks for both CH3 and CH2 , corresponding to four ethanol molecules. As the average number of H-bonds per molecule within the (6,6) CNT is close to 1, we can conclude that the single-file structure of ethanol molecules within the (6,6) CNT is formed by two groups of molecules with one hydrogen bond in each group and no hydrogen bond between the two groups. From the distance of the two peaks of the methyl groups in middle, we deduce that the orientation of the two molecules is head-to-head (methyl to methyl). On the other hand, the axial density profile of hydroxyl seems to have two peaks rather than four. This is because the distributions of the two hydroxyl groups forming a H-bond merge to make one peak. The density profile for the (7,7) CNT is much less structured than that of the (6,6) CNT, while we can still see a wave like structure, in which the five peaks correspond to the number of ethanol molecules within the nanotube. Inside the wider CNTs, along with increase of the number of molecules, the density profile of the molecules along the axis becomes vague and less informative. DOI:10.1088/1674-0068/25/04/487-493 E. Orientation distribution of the molecules within CNTs To understand the orientation distribution of the ethanol/water molecules within CNTs, we define α, β, and θ for the angles between the positive direction of the z-axis (tube axis) and CH3 −CH2 , O−H, and the dipole moment of the molecule, respectively. The orientation distribution of the CH3 −CH2 bond in all the CNTs are found to be in good accordance with the results of Shao et al. [16]. For the (6,6) CNT this orientation is mainly concentrated on 20◦ and 160◦ . It indicates that the CH3 −CH2 bond is likely to stay along the axis, consistent with the single-file structure of ethanol within the (6,6) CNT. As the CNT diameter increases, the orientation distribution becomes even. As for the orientation distribution of O−H, Figure 7(b) shows that there are four peaks for the (6,6) CNT. Interestingly, for the (8,8) CNT we see a high peak around 150◦ which is quite different from the other CNTs. Similar abnormality is also found by Mashl et al. [34] for confined water within the (9,9) CNT. Like the cases of pure water [30] and pure ethanol [16], the dipole orientation of the ethanol/water molecules within CNTs changes greatly with the CNT diameter. The dipole orientation of the molecules within the (8,8) CNT is mainly around 150◦ , while this distribution in other nanotubes has not much structure in other CNTs c °2012 Chinese Physical Society 492 Chin. J. Chem. Phys., Vol. 25, No. 4 Sheng-ping Du et al. FIG. 8 The angular density distribution of the molecules of the first layer outside the (a) (6,6) and (b) (9,9) CNT. 1. CH3 , 2. CH2 , 3. OW, 4. OH, the dash lines express the hexagonal centers of of CNT. except (6,6). the nanotubes prefer to contain ethanol rather than methanol. F. Angular distribution Figure 8 gives the angular distribution of the molecules in the first solvation shell outside CNT. As the axial distribution, the angle distribution also shows a wave-like structure with a period that changes with the CNT diameter. Notably, the positions of all the methyl peaks correspond to the centers of the hexagonal cells of CNT. This beautiful profile implies that there is a strong interaction between CH3 and hexagonal cell. On the other hand, the correlation between the CNT hexagon centers and the other groups (CH2 , OH, OW) is weak. This is similar to the results of methanol/water-CNTs in our previous study [21]. The angular distributions of the groups inside the CNTs are disordered. This can be explained by the following reason. When the curvature of the nanotubes is negative (i.e., outside), the distance of the centers of two consecutive hexagon cells is greater than that of graphene, so the “spatial resolution” felt by the methyl groups is enhanced, and the contrary takes place for the inner case. IV. CONCLUSION Systematic MD simulations are performed for ethanol/water mixture with CNTs. Remarkably, there is nearly pure ethanol within the narrow nanotubes from (6,6) to (10,10). This compellent character of these nanotubes makes them good instruments to separate ethanol from water. Based on a variety of structural analyses, we find that the ethanol/water mixture also forms interesting structures both inside and outside the nanotubes, including single-file and ring-like. The strong interaction between the CNT and the methyl groups of ethanol molecules is a key factor in determining these structures. We also run MD simulations for methanol/ethanol-CNT systems. As we expected, DOI:10.1088/1674-0068/25/04/487-493 V. ACKNOWLEDGMENTS This work is supported by the National Natural Science Foundation of China (No.20603032 and No.20733004), the Ministry of Science and Techology of China (No.2011CB921400), the Foundation for the Author of National Excellent Doctoral Dissertation of China (No.200736), the Fundamental Research Funds for the Central Universities (No.WK2340000006 and No.WK2060140005), and the Shanghai Supercomputer Center, the USTC-HP HPC Project, and the SCCAS. [1] S. Iijima, Nature 354, 56 (1991). [2] J. T. Sander, A. R. M. Verschueren, and C. Dekker, Nature 393, 49 (1998). [3] S. S. Wong, J. D. Harper, P. T. Lansbury Jr. and C. M. Lieber, J. Am. Chem. Soc. 120, 603 (1998). [4] K. B. Jirage, J. C. Hulteen, and C. R. Martin, Science 278, 655 (1997). [5] S. H. Joo, S. J. Choi, O. Ilwhan, J. Kwak, Z. Liu, O. Terasaki, and R. Ryoo, Nature 412, 169 (2001). [6] S. Cambre, B. Schoeters, S. Luyckx, E. Goovaerts, and W. Wenseleers, Phys. Rev. Lett. 104, 207401 (2010). [7] H. J. Wang, X. K. Xi, A. Kleinhammes, and Y. Wu, Science 322, 80 (2008). [8] G. Hummer, J. C. Rasaiah, and J. P. Noworyta, Nature 414, 188 (2001). [9] T. A. Pascal, W. A. Goddarda, and Y. Jung, Proc. Natl. Acad. Sci. USA 108, 11794 (2011). [10] J. Köfinger, G. Hummer, and C. Dellago, Phys. Chem. Chem. Phys. 13, 15403 (2011). [11] H. Kyakuno, K. Matsuda, H. Yahiro, Y. Inami, T. Fukuoka, Y. Miyata, K. Yanagi, Y. Maniwa, H. I. Kataura, T. Saito, M. Yumura, and S. Iijima, J. Chem. Phys. 134, 244504 (2011). [12] J. Su and H. Guo, ACS Nano 5, 351 (2011). c °2012 Chinese Physical Society Chin. J. Chem. Phys., Vol. 25, No. 4 [13] J. A. Thomas and A. J. H. McGanghey, J. Chem. Phys. 128, 084715 (2008). [14] J. Li, X. Gong, H. Lu, D. Li, H. Fang, and R. Zhou, Proc. Natl. Acad. Sci. U.S.A. 104, 3687 (2007). [15] D. J. Mann and M. D. Halls, Phys. Rev. Lett. 90, 195503 (2003). [16] Q. Shao, L. L. Huang, J. Zhou, L. H. Lu, L. Z. Zhang, X. H. Lu, S. Y. Jiang, K. E. Gubbins, Y. Zhu, and W. F. Shen. J. Phys. Chem. C 111, 43 (2007). [17] G. Garberoglio, J. Phys.: Condens. Matter 22, 415104 (2010). [18] Y. Liu, S. Consta, and W. A. Goddard, J. Nanosci. Nanotechnol. 10, 3834 (2010). [19] J. Zheng, E. M. Lennon, H. Tsao, Y. Sheng, and S. Jiang, J. Chem. Phys. 122, 214702 (2005). [20] L. Yang and Y. Q. Gao, J. Am. Chem. Soc. 132, 842 (2010). [21] W. H. Zhao, B. Shang, S. P. Du, L. F. Yuan, J. Yang, and X. C. Zeng, J. Chem. Phys. 137, 034501 (2012). [22] W. L. Jorgensen, J. D. Madura, and C. J. Swenson, J. Am. Chem. Soc. 106, 6638 (1984). [23] W. L. Jorgensen, J. Phys. Chem. 90, 1276 (1986). DOI:10.1088/1674-0068/25/04/487-493 Ethanol/Water Mixture with Carbon Nanotubes 493 [24] E. J. W. Wensink, A. C. Hoffmann, P. J. van Maaren, and D. van der Spoel, J. Chem. Phys. 119, 7308 (2003). [25] W. L. Jorgensen, J. Chandrasekhar, J. D. Madura, R. W. Impey, and M. L. Klein, J. Chem. Phys. 79, 926 (1983). [26] D. van der Spoel, E. Lindahl, B. Hess, G. Groenhof, A. E. Mark, and H. J. C. Berendsen, J. Comput. Chem. 26, 1701, (2005). [27] T. Darden, D. York, and L. Pedersen, J. Chem. Phys. 98, 10089 (1993). [28] H. J. C. Berendsen, J. P. M. Postma, A. DiNola, and J. R. Haak, J. Chem. Phys. 81, 3684 (1984). [29] J. R. Ryckaert, G. Ciccotti, and H. J. C. Berendsen, J. Comp. Phys. 23, 327 (1977). [30] J. Wang and J. Zhou, X. H. Lu, Phys. Chem. Chem. Phys. 6, 829 (2004). [31] K. H. Lee and S. B. Sinnott, Nano Lett. 5, 793 (2005). [32] A. Luzar and D. Chandler, Nature 379, 55 (1996). [33] X. Y. Zhou and H. J. Lu, Chin. Phys. 16, 335 (2007). [34] R. J. Mashl, S. Joseph, N. R. Aluru, and E. Jakobsson, Nano Lett. 3, 589 (2003). c °2012 Chinese Physical Society
© Copyright 2026 Paperzz