A planar Nanostructure for High-Capacity Hydrogen Storage

Aug 18, 2009 - A planar sheet of carbon (graphene) or isoelectronic boron nitride has been decorated by Ti atoms for a high-capacity hydrogen storage ...
24 downloads 0 Views 2MB Size
Download PDF
15783

2009, 113, 15783–15787 Published on Web 08/18/2009

Ti-Decorated BC4N Sheet: A planar Nanostructure for High-Capacity Hydrogen Storage S. Bhattacharya,†,§ C. Majumder,‡,# and G. P. Das*,† Department of Materials Science, Indian Association for the CultiVation of Science, JadaVpur, Kolkata 700032, India, and Chemistry DiVision, Bhabha Atomic Research Centre, Mumbai 400085, India, ReceiVed: June 22, 2009; ReVised Manuscript ReceiVed: August 7, 2009

A planar sheet of carbon (graphene) or isoelectronic boron nitride has been decorated by Ti atoms for a high-capacity hydrogen storage material. Due to charge transfer, the Ti becomes cationic and helps to adsorb hydrogen in the molecular form. The principal bottleneck for using such a system for efficient H storage is the problem of metal clustering on the surface. Using the first-principles density functional theory calculations, we show that clustering of Ti atoms can be avoided by using a composite BC4N planar sheet that is obtained through chemical modification of the graphene surface by systematically replacing C atoms by B and N atoms. This Ti@BC4N system shows hydrogen storage capacity with a reasonably high gravimetric efficiency (∼8.4 wt %). Further, using ab initio molecular dynamics simulation, we show that this system is stable at 300 K, while desorption starts at ∼500-600 K, which is lower compared to that of conventional graphene or BN planar structures. 1. Introduction The safe and convenient storage of molecular hydrogen is a crucial target in the transition to a hydrogen-based energy economy.1 Among many different approaches to achieve this end,2-4 one is that of storing hydrogen in nanostructure materials. Significant attention is being focused on nanostructure materials, such as fullerenes and nanotubes especially of carbonic compounds, as possible candidate structures for hydrogen storage.5-10 These nanostructures, due to their high surface-to-bulk ratio, constitute a class of materials that presents a great potential for this clean-energy alternative. Several works on metal-decorated nanostructures for hydrogen storage have been reported in the literature.7-10 However, if we look at the practical implementation of these kinds of structures for hydrogen storage, it has a very limited success. This is because of the question of stability of most of these nanostructures5 and the relatively high desorption temperature for hydrogen storage.7 Apart from these, there are several outstanding issues that need to be resolved, namely, (a) metal clustering on the surface,8 (b) ambiguity of exohedral versus endohedral decoration of metal atoms in nanostructures,9 (c) the barrier height effect for hydrogen desorption for endohedral storage of hydrogen,5 and last (d) subcritical utilization of the fullerene or nanotube surface affecting the volumetric efficiency of hydrogen storage. Over the past few years, metal-decorated carbon nanotube (CNT), boron nitride nanotube (BN-NT), and fullerene or cage structures of carbon and B-N held the central stage of attraction for hydrogen storage in nanostructure materials.5-8,10 It is expected that the planar form of these structures will be * To whom correspondence should be addressed. Phone: +91-33-2473 4971, Ext. 202 (Office). Fax: +91-33-2473 2805. E-mail: [email protected]. † Indian Association for the Cultivation of Science. ‡ Bhabha Atomic Research Centre. § E-mail: [email protected]. # E-mail: [email protected].

10.1021/jp905853x CCC: $40.75

interesting as well. In fact, significant amount of effort is being devoted to designing metal-decorated planar nanostructures for hydrogen storage.11-14 However, all of the earlier works discussed selective metal decorations on the graphene surface, which is difficult to achieve experimentally when metal atoms are sputtered on a graphene surface. In this work, we have embarked upon a case of random nonselective Ti-metal decoration on both the sides of graphene as well as the BN sheet, with a focus on its potential use for storage of H2 molecules. Here, we propose a novel conjecture to circumvent the problem of metal clustering on the surface by chemical modification of the graphene-like carbon network. We have shown here that with introduction of B and N in the pure gaphene-like carbon network, with different chemical orderings of hexagons in a B-C-N composite planar nanostructure, namely, a BC4N sheet, metal clustering can completely be prevented. Remarkably, our first-principles calculation shows that this Ti-decorated BC4N sheet can store up to 8.4 wt % of hydrogen in molecular form, which is more than the DOE (Department of Energy) target of 6 wt %. Using ab initio molecular dynamics simulation, we have shown that at 300 K, the system remains stable with all four H2 attached with Ti, while at 500 K, hydrogen gets released in molecular form from the Ti@BC4N sheet without breaking the structure. We believe that these results are useful to realize the potential of B-C-N composite planar structure as promising hydrogen storage materials. 2. Computational Methods All of the planar structures in our study have been modeled in terms of supercells of suitable dimensions in the x- and y-directions and a layer of vacuum on either side. For graphene and the BN sheet, we have taken a 32 atom cubic supercell with dimensions of 9.8 Å × 9.8 Å × 12.0 Å, while for the BC4N sheet, the supercell consists of 24 B, 96 C, and 24 N atoms in a 25.6 Å × 14.9 Å × 12.0 Å cubic cell. Electronic  2009 American Chemical Society

15784

J. Phys. Chem. C, Vol. 113, No. 36, 2009

Letters

Figure 1. (a) H storage in Ti2@graphene and (b) H storage in Ti2@BN sheet. Pink ball ) B, blue ball ) N, dark ash ball ) C, light ash ball ) Ti, white ball ) H.

structure and total energies were calculated with the Vienna ab initio simulation package (VASP),15 based on DFT.16,17 Projector-augmented wave (PAW) potentials18 were employed for the elemental constituents, namely, H, B, C, N, and Ti potentials, which contained one, three, four, five, and four valence electrons, respectively. The GGA calculation was performed with the Perdew-Wang19,20 exchange-correlation potential. The k points mesh was generated by the Monkhorst-Pack21 method, and all results were tested for convergence with respect to the mesh size (4 × 4 × 1). In all calculations, self-consistency was achieved with a 0.1 meV convergence of total energy. For highprecision calculations, we used a cutoff energy of 600 eV for the plane wave basis. For obtaining the optimized ground-state geometry,22,23 atomic forces were converged to less than 0.001 eV/Å by conjugated gradient (CG) minimization. We also performed ab initio molecular dynamics (MD) simulations using the Nose algorithm24 in order to investigate the stability of the optimized structures and desorption of hydrogen molecules from the Ti@BC4N sheet. 3. Results and Discussion 3.1. Ti@Graphene and Ti@BN Sheets for H Storage: Advantages and Shortcomings. We first consider Ti adsorption on a supercell of graphene and a BN sheet. In this context, it should be mentioned that in nanotubes or fullerenes, the endohedral versus exohedral decoration of the metal atom encounters the problem of surmounting a rather large barrier height.9 However, for a sheet, it is feasible to adsorb metal atoms on both sides of the surface with a negligible energy barrier.11,14 The optimization was carried out by placing Ti at different locations, and the results suggest that for both of the sheets, Ti prefers to be adsorbed on the center of a hexagon. Further interaction of hydrogen molecules shows that the Ti is capable of binding four H2 molecules in the case of both graphene and the BN sheet. After the adsorption of a Ti on one side, we placed another Ti on the opposite side of the sheet. The results show that the second Ti can be attached below the center of the same hexagon for both of these systems. The binding energy of the second Ti is of course less than the first Ti, as expected, but interestingly, it is also capable of storing an additional four H2 molecules in the case of graphene (Figure 1a) and two H2 molecules in the case of the BN sheet (Figure 1b). In the case of the BN sheet, further adsorption of H2 leads to its dissociation into atomic hydrogen. Finite-temperature ab initio molecular dynamics (MD) simulations (2 ps time period and a time step of 1 fs) have also been performed for Ti2@graphene + 8H2 and Ti2@BN sheet +6H2 in the temperature ranges of 0-300 and 300-800 K in steps of 100 K. Our results show that at 300 K, both of the systems are stable with all the H2 molecules attached with Ti atoms, while at around 700-800 K, the hydrogen get desorbed from

Figure 2. In (a) graphene and (b) a BN sheet when fully decorated on either side by Ti atoms; each C atom in graphene and each B or N atom of the BN sheet is being shared by six Ti atoms.

the Ti@graphene and Ti@BN sheet at 62.5 and 58.0%, respectively. It may be argued that 2 ps MD is statistically insufficient to estimate the desorption temperature accurately; however, the most important outcome of this MD simulation is that hydrogen can be extracted from the system without breaking the Ti-C in the case of Ti@graphene and Ti-B or Ti-N bonds in the case of the Ti@BN sheet. Further, the time step was 1 fs in the temperature range of 100 K, which is sufficient to break the geometry of the structure even from this 2 ps MD simulation calculation, if the system is unstable. In order to estimate the gravimetric efficiency of these systems, we have attempted to fully cover both sides of the respective sheets with Ti atoms. However, due to stronger Ti-Ti interaction, Ti atoms agglomerate to form clusters on the surface. In the case of Ti@graphene, the strength of the Ti-C bond depends on the transfer of the electrons from Ti 3d/4s to C 2p, resulting in Ti+ ions. With two-sided Ti decoration (Figure 2a,b), the magnitude of charge transfer reduces due to effective sharing of a single C atom by six Ti atoms (three above the surface, three below the surface), and consequently, the strength of the Ti-C bond decreases at the cost of Ti-Ti clustering on one side of the graphene. The same scenario was found in the case of the BN sheet. On the basis of these results, we infer that these kinds of planar structures can serve as potential hydrogen storage materials provided we can successfully prevent metal clustering on the planar structure. To date, very few attempts have been made to prevent metal clustering on the surface of a nanostructure with the introduction of suitable additional dopant atoms,10 albeit at the cost of system stability and further reduction in gravimetric efficiency of the system. Another way to achieve this without compromising with system stability as well as gravimetric efficiency is by introducing B and N into the graphene network and thereby forming a B-C-N composite planar nanostructure. In particular, since BC4N nanotube has already been synthesized as a stable nanotube,25 we decided to flatten the structure and hence produced the BC4N planar structure. 3.2. Ti Decoration on a BC4N Sheet. The ground-state structure and properties of the Ti-decorated (6,6) armchair BC4N

Letters

J. Phys. Chem. C, Vol. 113, No. 36, 2009 15785 TABLE 1: BE Value of the Ti Atom for Estimation of the Preferred Hexagon on One Side of the BC4N Sheeta

Figure 3. (a) Ball and stick model of a pure BC4N sheet. (b) In the ground-state configuration of the BC4N sheet, B-N3 and N-B3 are connected by a B-N bond, which gives the structure some extra stability under high temperature.

nanotube have already been reported, and their suitability for hydrogen storage has been studied.9 Here, we have simply opened up this nanotube to realize a planar structure (Figure 3a) by selectively substituting the C atoms in graphene with B and N in such a way that the local structural units of BN3 and NB3 are linked with a B-N bond (Figure 3b). Such a BC4N sheet turns out to be semiconducting, with a band gap of 0.8 eV (Figure 4a). Moreover, our first-principles MD simulation predicts that the BC4N sheet has a high thermal stability (up to 900 K). The BC4N planar sheet has seven different kinds of

different hexagonal faces

Nhex

binding energy in eV/Ti atom

6C 1B-5C 1N-5C 2B-1N-3C 2B-2N-2C 2N-1B-3C 1B-1N-4C

18 6 6 12 12 9 6

-2.09 -2.09 -1.73 -1.96 -1.34 -1.28 -1.76

a Nhex is denoted as the total number of a particular type of hexagon in the BC4N sheet.

hexagonal sites which have different affinities for binding Ti on the surface (Table 1). To ensure the most preferred location of Ti, we have optimized the Ti on different hexagonal faces of the BC4N sheet and compared the relative stability. Subsequently, we decorated all of these preferred hexagons on one side of the BC4N sheet and repeated the same decoration with Ti atoms on the opposite side. Interestingly, for the second surface, the preferred hexagon was not the same as that of the first surface. Table 1 summarizes our results for the first Ti optimization on the first surface. We see that for the first surface, 1B-5C and 6C are the most stable hexagons. When we filled all of those hexagons of one side of the BC4N sheet with Ti atoms and started searching the preferred hexagons on the

Figure 4. Electronic structure of (a) a pure BC4N sheet, (b) a Ti@BC4N sheet, and (c) a Ti@BC4N sheet + 4H2. (d) Orbital plot of a Ti@BC4N sheet + 4H2.

15786

J. Phys. Chem. C, Vol. 113, No. 36, 2009

Figure 5. Total charge contained in different Ti atoms with the Bader partitioning rule in Ti36@graphene, a Ti36@BN sheet, and a Ti36@BC4N sheet after Bader charge analysis.

opposite side (second surface), quite expectedly, this turned out to be the 2B-1N-3C face (see table, Supporting Information), which is energetically the second lowest in Table 1. This is because when all of the 6C and 1B-5C hexagons are filled with Ti on one side, it is easier for a Ti to go toward the energetically second-lowest hexagon (Table 1) of the sheet on the opposite side. In our supercell sheet structure, we have total 18 6C, 6 1B-5C, and 12 2B-1N-3C hexagons. Therefore, all total, 36 Ti atoms can be accommodated on this BC4N sheet, and we have confirmed that this fully covered structure is stable without any clustering of Ti atoms on either of the side of the sheet. This is an important observation. The reason for such nonclustering is because of the fact that with different hexagonal combinations, the probability of Ti occupancy in all of the hexagons is not the same, and consequently, Ti can fully transfer its charge to the C/B/N atoms on its preferred hexagon to make a strong Ti-C/ Ti-B/ Ti-N bond as compared to the Ti@graphene or Ti@BN sheet. To illustrate this fact, we have performed Bader charge analysis of all of the systems, namely, Ti36@graphene, the Ti36@BN sheet, and the Ti36@BC4N sheet. The total charge associated with different Ti atoms according to Bader partitioning in Ti36@graphene, the Ti36@BN sheet, and the Ti36@BC4N sheet is shown in Figure 5. Since the Ti atom has four outermost valence electrons [4s23d2], any value of charge state less (more) than 4 indicates that the Ti becomes cationic (anionic). It is clearly evident from Figure 5 that in the case of both Ti36@graphene and the Ti36@BN sheet, the charge states of different Ti atoms fluctuate around the value 4. In contrast, for the Ti36@BC4N sheet, the charge states of all of the Ti ions are significantly lower (i.e., highly cationic state), indicating strong charge transfer from Ti to the host substrate material (i.e., the

Letters BC4N sheet). Such a consistent nonfluctuating charge state of all of the Ti ions decorating the BC4N sheet helps to prevent metal clustering on the surface, thereby facilitating efficient storage of molecular hydrogen. This reiterates our conjecture that the introduction of different chemical orderings of hexagons serves as a panacea to the problem of metal clustering. From a comparison of the partial densities of states of pure BC4N and Ti@BC4N (Figure 4a,b), we observe that the partially occupied Ti peak appears in the gap of the semiconducting BC4N. The lower energy “shoulder” of the occupied Ti-DOS coincides with a newly formed “hump” arising out of C 2p (see Figure 4b). This hybridization between Ti 3d/4s and C 2p leads to a strong Ti-C bond, as compared to the Ti-Ti bond, which, in turn, plays a crucial role to avoid Ti clustering on the surface. 3.3. Hydrogen Storage in the Ti@BC4N Sheet. Once we circumvent the problem of metal clustering of Ti atoms by different chemical ordering of hexagons, we focus our attention to study the interaction of hydrogen on it. The results reveal that like in the case of Ti@graphene or the Ti@BN sheet, each Ti atom placed on the BC4N sheet can adsorb up to four H2 molecules, irrespective of whether it is on the 6C or 1B-5C site. The sequential binding energy values of four hydrogen molecules on the Ti-adsorbed (placed on the 1B-5C hexagon) BC4N sheet are estimated to be 0.55, 0.67, 0.42, and 0.45 eV, respectively. Moreover, we note that these BE values are of similar magnitude when Ti is attached on the 6C or 2B-1N-3C hexagonal faces. It should be emphasized here that all of the binding energy values from the first to fourth hydrogen adsorption lie in the perfect energy window between physisorption and chemisorption. The large unoccupied peak of the Ti 3d/4s above the Fermi energy tells us that the Ti atom in Ti@BC4N is in the cationic state and can accept electrons from other atoms (Figure 4b). As hydrogen molecules are added one after another, the vacant d orbital of Ti gets gradually filled up, and as a result of that, the unoccupied states of Ti+ (as shown in DOS of Ti@BC4N) become depopulated in Ti@BC4N + 4H2 at the cost of the H 1s peak growth (Figure 4c). This is also apparent from the molecular orbital plot [HOMO-2] of Ti@BC4N + 4H2, where we can clearly see that we need two unoccupied energy states, one for bonding with hydrogen and the other for binding the metal to the planar sheet (Figure 4d). We next consider two Ti atoms adsorbed on the consecutive 6C hexagons (or one 6C and the other nearest 1B-5C hexagons) on one side and two Ti atoms on the nearest 2B-1N-3C hexagons on the opposite side. Now, the interaction of hydrogen with this system shows that these 4 Ti atoms can hold a total of 16 hydrogen molecules, as shown in Figure 6a. This is quite encouraging as far as molecular H2 adsorption is concerned. On the basis of these results, we estimate that with full Ti

Figure 6. 16H2 adsorption in two adjacent hexagons (above and below) of a BC4N sheet for two-sided 4Ti decoration (two Ti above and two Ti below). (d) Optimized structure of a fully Ti decorated BC4N sheet for 144 H2 adsorption (Ti36@BC4N + 144H2).

Letters coverage, a total of 144 H2 molecules can be adsorbed on the Ti36@BC4N sheet with a gravimetric efficiency of 8.4 wt % of hydrogen. To ascertain this conjecture, we have attempted to relax 144 H2 molecules on the Ti36@BC4N substrate (Figure 6b).26 Our calculations for the fully relaxed decorated structure indicate that the system should be capable of storing up to 8.42 wt % of hydrogen.27 After hydrogenation, we performed an ab initio molecular dynamics (MD) simulations using a Nose thermostat.24 The Ti2@BC4N sheet + 8H2 system was heated at 300, 400, 500, and 600 K and allowed to equilibrate for 2 ps at each temperature. The results show that about 75% of the hydrogen desorbs from the Ti2@BC4N sheet in the temperature range of 500-600 K. Moreover, it should be mentioned that by carrying out the MD simulation for a 2 ps duration, it is difficult to achieve a statistically meaningful temperature; however, by comparing with our earlier MD results on Ti2@graphene + 8H2 or Ti2@BN sheet + 6H2, it is suggested that the desorption temperature has been lowered for a B-C-N composite planar structure. 4. Conclusion Graphene, BN, and BC4N planar structures, decorated with a 3d transition-metal, namely, Ti, have been studied and compared from the point of view of the efficient storage of molecular hydrogen. First-principles density functional calculations reveal that utilization of both sides of the sheet increases the gravimetric as well as volumetric efficiencies for hydrogen storage, as compared to those in functionalized nanotubes and fullerenes. The main bottleneck, however, is the problem of metal clustering on the surface while going for full coverage. In this work, we have shown how this metal clustering can be avoided in a BC4N sheet by chemical manipulation, without any other external means such as doping. We conclude from our studies that the Ti36@BC4N sheet is a stable structure that can be used for efficient hydrogen storage. Furthermore, this B-C-N nanocomposite system is found to desorb H2 at temperatures lower than those in graphene and the BN sheet. Supporting Information Available: (1) Two xyz files containing the structural details of the BC4N sheet and the Ti36@BC4N sheet. (2) A table containing the B.E. values of a Ti atom attached on different hexagons in the second surface of the Ti24@BC4N sheet. This was done once the first surface was decorated with 24 Ti atoms in all of its 6C and 1B-5C hexagons. This also provides a clear justification of choosing the 2B-1N-3C hexagon for the second surface decoration, which, in turn, strongly

J. Phys. Chem. C, Vol. 113, No. 36, 2009 15787 justifies our novel conjecture of chemically modifying the homogeneous sheet to avoid metal clustering. (3) A figure for the “Projected Density of States” on the different orbitals of Ti atoms in the Ti@BC4N sheet is also provided so that Ti-3d/4s interaction can be clearly understood. This material is available free of charge via the Internet at http:// pubs.acs.org. References and Notes (1) Schlappbach, L.; Zuttel, A. Nature 2001, 414, 353. (2) Jacoby, M. Chem. Eng. News 2005, 83, 42–47. (3) Van den Berg, A. W. C.; Arean, C. O. Chem. Commun. 2008, 668– 681. (4) Bhattacharya, S.; Guotao, W.; Chen, P.; Feng, Y. P.; Das, G. P. J. Phys. Chem. B 2008, 112, 11381. (5) Sun, Q.; Wang, Q.; Jena, P. Nano Lett. 2005, 5, 1273. (6) Ma, R.; Bando, Y.; Zhu, H.; Sato, T.; Xu, C.; Wu, D. J. Am. Chem. Soc. 2002, 124, 7672. (7) Yildirim, T.; Ciraci, S. Phys. ReV. Lett. 2005, 94, 175501. (8) Barman, S.; Sen, P.; Das, G. P. J. Phys. Chem. C 2008, 112, 9963. (9) Bhattacharya, S.; Majumder, C.; Das, G. P. J. Phys. Chem. C 2008, 112, 17487. (10) Sun, Q.; Jena, P.; Wang, Q.; Marquez, M. J. Am. Chem. Soc. 2006, 128, 9741. (11) Ataca, C.; Aktu¨rk, E.; Ciraci, S. Phys. ReV. B 2009, 79, 041406(R). (12) Yang, X.; Zhang, R. Q.; Ni, J. Phys. ReV. B 2009, 79, 075431. (13) Kim, G.; Jhi, S. Phys. ReV. B 2008, 78, 085408. (14) Ataca, C.; Aktu¨rk, E.; Ciraci, S.; Ustunel, H. Appl. Phys. Lett. 2008, 93, 043123. (15) Kresse, G.; Hafner, J. Phys. ReV. B 1994, 49, 14251. Kresse, G.; Furthmu¨ller, J. J. Comput. Mater. Sci. 1996, 6, 15. (16) Hohenberg, P.; Kohn, W. Phys. ReV. 1964, 136, B864. (17) Kohn, W.; Sham, L. Phys. ReV. 1965, 140, A1133. (18) Blo¨chl, P. E. Phys. ReV. B 1994, 50, 17953. (19) Perdew, J. P.; Wang, Y. Phys. ReV. B 1992, 45, 13244. (20) Perdew, J. P.; Chevary, J. A.; Vosko, S. H.; Jackson, K. A.; Pederson, M. R.; Sing, D. J.; Fiolhais, C. Phys. ReV. B 1992, 46, 6671. (21) Monkhorst, H. J.; Pack, J. D. Phys. ReV. B 1976, 13, 5188. (22) Press, W. H.; Flannery, B. P.; Tenkolsky, S. A.; Vetterling, W. T. Numerical Recipes; Cambridge University Press: New York, 1986; Vol. 1. (23) Pulay, P. Chem. Phys. Lett. 1980, 73, 393. (24) Nose, S. J. Chem. Phys. 1984, 81, 511. (25) Raidongia, K.; Jagadeesan, D.; Kayaly, M.; Waghmare, U. V.; Pati, S. K.; Eswaranmoorty, M.; Rao, C. N. R. J. Mater. Chem. 2008, 18, 83. (26) As the system consists of 468 atoms, this is a computationally very difficult task to have all of the 144 H2 molecules be adsorbed in the molecular state into the system after full optimization. Even after final optimization, we notice that few H2 molecules are dissociated into 2H atoms, while most of the 288 H atoms are attached close to the system in molecular form. (27) There have been attempts to prepare ultrathin graphitic sheets of thicknesses down to few atomic layers, which, in the limiting case, leads to a single layer of graphene. For example, the so-called FLG (few-layer graphene) reported by Novoselov, K. S. Science 2004, 306, 666. constitutes a few layers (two, three, or more) of graphene. In view of the experimental shortcomings in producing a single-layer planar structure, as considered in our present work, there is a possibility of reduction in gravimetric density in a multilayered planar structure.

JP905853X