Hydrogen Storage in Aromatic Carbon Ring Based Molecular

The second one includes “chemical” storage methods where hydrogen is .... complexes were calculated at the MP2 level with the aug-cc-pVTZ molecula...
0 downloads 0 Views 317KB Size
Article pubs.acs.org/JPCC

Hydrogen Storage in Aromatic Carbon Ring Based Molecular Materials Decorated with Alkali or Alkali-Earth Metals Igor V. Bodrenko,†,‡ Alexander V. Avdeenkov,*,‡,¶ Dmitri G. Bessarabov,§ Anton V. Bibikov,‡ Alexander V. Nikolaev,‡,∥ Mikhail D. Taran,⊥ and Eugene V. Tkalya‡ †

National Nanotechnology Laboratory (NNL), Istituto Nanoscienze - CNR, Via per Arnesano 16, 73100 Lecce, Italy Skobeltsyn Institute of Nuclear Physics, Lomonosov Moscow State University, Leninskie Gory 1(2), GSP-1, 119234 Moscow, Russia ¶ National Institute for Theoretical Physics (NITheP), Stellenbosch and Institute of Theoretical Physics, University of Stellenbosch, Stellenbosch 7600, South Africa § DST Hydrogen Infrastructure Center of Competence (HySA Infrastructure) and North-West University, Faculty of Natural Sciences, Private Bag X6001, Potchefstroom, 2520 South Africa ∥ Moscow Institute of Physics and Technology (State University), 141700 Dolgoprudny, Russia ⊥ Troitsk Institute for Innovation and Fusion Research, Troitsk, Russia ‡

ABSTRACT: On the basis of first-principles calculations of molecular electron structure, we discuss the strategy of modifying the carbon-based materials in order to increase their capacity to bind with molecular hydrogen. In particular, we have studied hydrogen adsorption on molecular complexes having anionic aromatic carbon-based rings stabilized by cations of alkali (Li+, Na+, K+) or alkali-earth metals (Be2+, Mg2+, Ca2+). The adsorption depends more on the properties of the cation than on the ring itself. The interaction of the H2 molecule with an electrostatic field leads to the binding of the hydrogen molecule with the strongly polarized ionic molecular complex. The number of the adsorbed molecules is driven by two factors acting in opposite directions: the binding energy, which should be larger than a 4−5 kJ/mol threshold needed to keep hydrogen molecules attached, and the area around the cation (coordination sphere), which is determined by its radius. As a compromise between these factors, we propose several promising candidates for building blocks of hydrogen storage materials, including diboratabenzene lithium, C4B2H6Li2, and diboratabenzene potassium, C4B2H6K2, which can adsorb 6 and 12 H2 molecules, correspondingly. We also discuss the possibility of linking these molecular complexes in three-dimensional structures.



INTRODUCTION Hydrogen has been and is still being widely considered as a promising medium for accumulating and then releasing large amounts of energy.1 However, the progress in hydrogen energetics is hampered by the well-known problem of effective hydrogen storage.2 The existing approaches to the hydrogen storage can be divided into four groups.2−4 The first group unifies mechanical storage technologies, that is, hydrogen storage in liquid or compressed gaseous forms. The second one includes “chemical” storage methods where hydrogen is chemically combined in the form of metal hydrides or other hydrogen-rich compounds, for example, ammonia, ethanol, etc. The absorption methods, that is, those where atomic hydrogen is incorporated directly into the bulk of the material, for example, into the interstitial sites in the crystallographic lattice structure, constitute the third group. The fourth group comprises hydrogen storage technologies that are based on the adsorption of hydrogen onto the highly fractured surface of microporous or nanostructured materials, for example, zeolites, activated carbon, carbon nanotubes, etc. In the present paper, we © 2012 American Chemical Society

consider novel materials that fall into the latter category (i.e., physical absorption) when the binding between hydrogen molecules and host material is due to nonchemical van der Waals and electrostatic forces. There are two major factors that influence the storage capacity of an adsorbent. These are the effective size of the interstices, which determines the surface of the adsorbent available for the adsorbate the so-called specific surface areaand the adsorption energy, which is the binding energy of the adsorbate molecule to the adsorbent surface. Modern nanostructured materials, such as zeolites, carbon nanotubes, activated carbon, ball-milled materials, etc., can have interstices as small as 1−10 nm in size. In that case, the specific surface area is close to its theoretical limit.5 Therefore, to further improve the properties of the material, that is, to achieve sufficiently high values of both gravimetric and volumetric hydrogen storage capacities, one has to increase the hydrogen− Received: May 31, 2012 Revised: November 6, 2012 Published: November 8, 2012 25286

dx.doi.org/10.1021/jp305324p | J. Phys. Chem. C 2012, 116, 25286−25292

The Journal of Physical Chemistry C

Article

However, all of these results (except the study of Rao and Jena, ref 14) have been obtained within the density functional theory (DFT) approach to electronic structure, which is based on an approximate term for electron exchange-correlation energy. Standard DFT methods utilize either local density approximation (LDA) or semilocal approximation [generalized gradient approximation (GGA)]. Although LDA calculations show some binding between graphite sheets, they fail to provide an accurate account of van der Waals forces. GGA calculations show no relevant minimum for the graphite energy variation with lattice constants. Nowadays, there are DFT functionals that are designed to describe van der Waals forces,24,25 but it is not intrinsic to DFT. Therefore, it is more accurate to use the post-Hartree−Fock method with the Møller−Plesset perturbation scheme for correlations, which we have adopted for our studies. In the present study, by using the first-principle quantummechanical calculations, we perform the comparative analysis of the hydrogen binding capacity of the several carbon-based aromatic ionic molecular complexes and discuss their potential in serving as the building blocks for future nanostructured materials for hydrogen storage.

adsorbent binding energy. The issue has been in the focus of many recent theoretical studies (see, e.g., refs 6−11). The goal is to discover new lightweight promising compounds that can bind molecular hydrogen stronger than existing adsorbers. As a starting point in the present study, we have considered the aromatic carbon ring. It is well-known that aromatic carbon rings give rise to lightweight, chemically stable nanostructured materials, such as fullerens, nanotubes, etc., with a large specific surface area. On the other hand, pure carbon aromatic compounds can (nonchemically) bind, at most, one hydrogen molecule per aromatic ring (on each side) with the energy of 4−5 kJ/mol.12 Kim et al. demonstrated that the hydrogen uptake could be enhanced by the substitution of carbon atoms in the C36 fullerene with B or Be.7 On the other hand, it was established that transition-metal and alkali metal positive ions can nondissociatively bind several H2 molecules13−15 due to electrostatic forces. In particular, by using quantum chemical calculations, including electron correlations, Rao and Jena showed that a Li cation can adsorb at least six hydrogen molecules.14 The molecules have the binding energy of 0.202 eV (19.5 kJ/mol) and are situated around Li+ at the same distance. Sun et al. further studied the Li-coated C60 fullerene and found that 12 Li atoms are capped onto the 12 pentagons of C60. Because of the charge transfer from Li 2s to the fullerene cage, the Li atom becomes positively charged, and as a result, the Li+ cation can bind 5 hydrogen molecules, which makes 60 H2 molecules for Li12C60. The binding energy is 0.075 eV (7.2 kJ/mol) per H2 molecule. It is also important that Li atoms do not form a cluster, since the binding energy between Li and C60 is slightly higher than the cohesive energy of lithium metal. The situation is very different from titanium transition-metal atoms. In Ti12C60, 12 Ti atoms tend to form a small cluster rather than being uniformly dispersed on the C60 surface.16 Still, however, lithium ions can migrate between equivalent neutral carbon rings, thus reducing the stability of the material. The solution combining the aromatic scaffold structure and the anchored cations (such as in zeolites) may be in introducing negatively charged anionic aromatic rings into the nanostructured carbon material and balancing them with the alkali or alkali-earth cations.17 Among carbon aromatic structures, the stable negatively charged one is the fivemembered cyclopentadienyl anion (Cp), which is well-known in the form of alkali metals salts (CpLi, CpNa) or in metallocenes. The negatively charged six-membered aromatic rings may be formally obtained by the isoelectronic substitution of carbon with boron. These anions, called boratabenzenes, exhibit a clear aromatic structure and have been known in the form of alkali metal salts.18,19 The boron doping of carbon systems with subsequent Li metal atom coating has been discussed in connection with hydrogen storage in carbon nanotubes.20,21 Indeed, from the theoretical studies of refs 20 and 21, it follows that the Li binding energy is increased if a carbon nanotube is doped with B. It is worth noting that some of the present authors put forward that idea in 2006,17 independently from ref 20. The ideal BC3Li2 carbon-nanotube-based system can accommodate one hydrogen molecule per lithium. Recently, the same idea has been applied to graphene.22,23 It has been found that a boronsubstituted carbon sheet increases the Li−graphene binding energy, which leads to an enhanced hydrogen uptake (up to eight H2 molecules per Li adatom according to ref 22).



METHOD OF CALCULATIONS Studies of the electron structure of molecular complexes and of the H2 binding energies were performed at several levels of the theory. The semiempirical (MNDO, AM1, and PM3) energy calculation and the unconstrained molecular geometry optimization were carried out using the PC GAMESS version of the GAMESS(US) program package.26 The restricted Hartree−Fock (RHF), the second-order Møller−Plesset (MP2), and a variant of the RHF-based complete active space second-order perturbation theory (CASPT227,28) calculations were performed with Algo-QMT program package developed by the authors.29 The package utilizes the resolution of the identity (RI) method to evaluate the electron−electron interaction integrals as described in ref 29. Two types of molecular basis sets in the Cartesian form were used: one was Pople’s split valence basis set with (6-31G**) and without (631G) polarization functions;30−32 the other was Dunning’s (augmented) correlation-consistent polarized (aug-)cc-pVXZ (X = D, T, Q, 5, 6) basis set series.33,34 The highly accurate RI basis sets were used for final calculations so that the error due to the RI approximation was negligible. The unconstrained and the constrained geometry optimization of monomers and intermolecular complexes at the ab initio level of the theory (RI-MP2) were performed with the AlgoQMT program package. The program utilizes a gradient-free local minimization method where the fulfillment of two criteria (for the energy and the gradient estimates) is required for convergency. The Hessian matrix as well as its smaller eigenvalues are estimated in the course of the optimization, and if the negative frequencies are found, the next optimization step is made along the corresponding normal mode. This procedure should prevent convergence to the transition state. Also, we have made the full normal-mode analysis (with the numerical Hessian) for the most significant final conformations and confirmed the absence of negative frequencies. During the geometry optimization, the energy accuracy of 10−6 in atomic units (a.u. or Hartree) and the gradient accuracy of 10−4 a.u. or below were imposed. The initial unconstrained geometry optimization of the monomers was carried out at the semiempirical MNDO level as 25287

dx.doi.org/10.1021/jp305324p | J. Phys. Chem. C 2012, 116, 25286−25292

The Journal of Physical Chemistry C

Article

geometries of molecular complexes were optimized. In addition to single-ring complexes, we have studied some small carbon polyaromatic compounds (e.g., naphthalene, pyrene). Several representative examples of the single-ring compounds are visualized in Figure 1. In particular, by substituting one

well as at the RI-MP2 (frozen core) level with 6-31G and 631G** molecular basis sets. The latter method provides the accuracy, which is comparable with experimental (0.05 Å for bonds and 1° for angles) values. This geometrical inaccuracy results, for example, in 0.5 kJ/mol uncertainty for the dihydrogen interaction energy. We have also found that, for some monomers (e.g., for benzene, naphthalene, etc., as well as for 1-H-boratabenzene Li salt), the geometry optimized at the MNDO level is very close to the geometry optimized within the MP2 approach with the 6-31G** basis set. Therefore, in our studies, we have used the MNDO geometry for large systems (nanostructures) like carbon nanotubes. In some cases (for example, those with Na ions), however, the MNDO geometry turned out to be incorrect, and we then used the geometry optimized within MP2 with the 6-31G** basis set. Final geometries and interaction energies of the intermolecular complexes were calculated at the MP2 level with the aug-cc-pVTZ molecular basis set (high polarization f-functions on heavy atoms, however, were excluded from the set). The basis set superposition error (BSSE) was corrected by applying the counterpoise (CP) method. The accuracy of the interaction energy for such an approach is within 1 kJ/mol. We have checked it by performing calculations with aug-cc-pVTZ, augcc-pVQZ, and aug-cc-pV5Z basis sets for some selected cases. For example, our result for the interaction energy of the hydrogen molecule with benzene is 4.6 kJ/mol, which is comparable with values from 3.9 to 4.9 kJ/mol obtained in ref 35. We have also observed that, at the MP2 6-31G** level without the BSSE correction, the energy accuracy is 1.3 kJ/mol, in comparison with the MP2 aug-cc-pVTZ(-hp) level accounting for BSSE at the same geometry. This effect is due to the mutual compensation of BSSE and the error of relatively short basis sets. However, the deviation increases if one compares the energies at different geometries that have been optimized according to these two calculation schemes. The energy difference in that case can reach 2.1 kJ/mol. In our calculations of the binding energy of one hydrogen atom with molecular complexes, we used the molecular geometries optimized without constraints. In the case of several hydrogen molecules, however, we have assumed that the monomers are rigid and allowed only optimizing their mutual positions and orientations. CASPT2 calculations were performed for some instances to study the influence of the higher-order (beyond MP2) electronic correlations on the H2 binding energy. The results indicate that the higher-order electronic correlations can change the MP2 binding energy of the hydrogen molecule by 0.8−1.3 kJ/mol. This estimate is close to the results of coupled clusters calculations for benzene with dihydrogen.35

Figure 1. Single-ring compounds with bound hydrogen molecules (gray balls stand for carbon, green for boron, red for lithium, blue for sodium, pink for kalium, dark green for magnesium, and white for hydrogen): (a) benzene with two H2 molecules; (b) Cp lithium salt with four H2 molecules; (c) 1-H-boratabenzene sodium salt with five H2 molecules; (d) 1,3-H-diboratabenzene disodium salt with eight H2 molecules; (e) 1-H-boratabenzene kalium salt with seven H 2 molecules; (f) 1,3-H-diboratabenzene magnesium salt with six H2 molecules; (g) the same as on (b) but top view; (h) the same as on (c) but top view.

carbon atom with boron in the six-membered carbon aromatic ring (benzene), one obtains the 1-H-boratabenzene aromatic anion, C5BH6−, shown in Figure 1c. The diboratabenzene aromatic dianion, C4B2H62−, shown in Figure 1d, is obtained by substituting two carbon atoms with two boron atoms. As counterions, we have taken a single Na+ cation in Figure 1c and two Na+ cations in Figure 1d. In addition, we have considered five-membered anionic and dianionic aromatic molecules based on the cyclopentadienyl ion, C5H5−, abbreviated in the following as Cp. The Cp molecular complex with one Li+ cation and four bound H2 molecules is shown in Figure 1b. For comparison, in Figure 1a, we also depict a neutral aromatic ring of the benzene molecule. It is worth mentioning that C5BH6−, C4B2H62−, and C5H5− have been synthesized before, and their molecular complexes with cations, in particular, with alkali metals, are known to be stable.18,36 By optimizing the molecular geometry, we have found that, in general, the isolated anionic ring retains its planarity. The B− C and B−H valence bonds are somewhat elongated in comparison with the corresponding C−C and C−H bonds. In the case of B−C bonds, the typical value is 0.05−0.15 Å. The ionization energies of anions estimated according to the Koopmans’ theorem with the 6-31G** molecular basis set lie between 1.2 and 2.5 eV. These findings indicate that the substituted rings probably retain their aromatic character and, therefore, can form nanostructures (fullerene, graphite, carbon tubes, graphene, etc.) like the initial carbon rings. This conclusion is supported by the first-principles method of ref 37, where it has been shown that, in many cases, boron, by acquiring an extra electron, becomes similar to carbon.



RESULTS AND DISCUSSION Starting with the five- and six-membered carbon rings, we considered several molecular complexes on such a base, where one or two ring atoms are substituted with one or two boron atoms while the numbers of the electrons remain the same, as in the pure carbon rings. As a result, the ring becomes anionic and counterions are required to keep the molecular complex neutral. In the present study, we considered only counterion cations of alkali atoms (Li+, Na+, K+) and those of alkali-earth atoms (Be2+, Mg2+, Ca2+). We then performed numerous calculations of the binding energies between such complexes and hydrogen molecules, adding the latter one by one. The 25288

dx.doi.org/10.1021/jp305324p | J. Phys. Chem. C 2012, 116, 25286−25292

The Journal of Physical Chemistry C

Article

membered rings and independent of the substituted atoms of the ring. The binding of a single hydrogen molecule to any of the considered molecular complexes does not lead to its dissociation. Our calculations show that the H−H molecular bond is elongated by only 0.03−0.05 Å. The binding energy, E(H2), as well as the distance, d(I−H2), between a counterion and the bound hydrogen molecule are almost independent of the particularities of the anionic ring. However, they strongly depend on the choice of the cation; see Table 1. The energy is smaller for heavier cations and larger for double charged cations, while the distance demonstrates the inverse behavior. We then investigated how many hydrogen molecules, N(H2), can be bound to the considered modified compounds. To rule out weakly bound hydrogen molecules that can be easily detached from the complex by thermal energy fluctuations, we introduced the 4−5 kJ/mol energy threshold. This is also the binding energy of a hydrogen molecule to the pristine carbon six-membered aromatic ring, so one may consider N(H2) as a characteristic of the compound relative to the pure carbon ring. The threshold is close to lower binding energies used in other calculations (for example, the lower binding energy in ref 15 was 0.075 eV/H2 (7.2 kJ/mol)). N(H2) and the range of the distances, d*(Li+−H2), from cations to the bound hydrogen molecules are almost independent of the particularities of the anionic ring, but depend on the counterions; see Table 1. The d*(Li+−H2) distances for lithium given in Table 1 are in good correspondence with the 2.15 Å value for the Li+ complex with six H2 molecules from ref 14. It is also noteworthy that the number of adsorbed molecules for Li+ (N(H2) = 3) is exactly half of the value for the isolated Li+ cation saturated by H2 molecules.15 The 1/2 factor accounts for the fact that, in our case, the base ring blocks the lower hemisphere so that H2

The positive counterions balancing the anionic rings are located above (or below) the ring plane, as it is also illustrated in Figure 1. The characteristic distances and the binding energies associated with the considered cations are given in Table 1. The perpendicular line from the cation to the ring Table 1. Characteristic Binding Energies, Distances, and Numbers of Adsorbed H2 Molecules Calculated for Various Cationsa ion

d(I−C), Å

d(I−H2), Å

E(H2), kJ/mol

N(H2)

d*(I−H2), Å

Li+ Na+ K+ Be2+ Mg2+ Ca2+

2.1−2.3 2.5−2.7 3.0−3.2 1.7−1.9 2.2−2.3 2.6−2.7

2.1 2.5 3.1 1.6 2.2 2.6

15 8 5 50 30 15−20

3 4 6 1 5−6 6

2.2−3.2 2.6 3.2 1.6 2.2−3.9 2.7−2.9

a

d(I−C) is the distance range from the cation to different carbon atoms, d(I−H2) is the average distance from the cation to a single hydrogen molecule, E(H2) is the binding energy of the single hydrogen molecule, N(H2) is the number of bound hydrogen molecules with the energy of at least 4−5 kJ/mol, and d*(I−H2) is the distance range from the cation to the N(H2) hydrogen molecules.

plane crosses the plane close to the ring center in all cases. Notice that lighter cations (of the same charge) are bound stronger and located closer to the ring. The effect is more pronounced for doubly charged cations. Although the ring remains almost planar after binding the counterion, there are small deviations of the side groups toward the cation position, especially for doubly charged cations. The distances from the cation to the carbon atoms of the ring depend mainly on the cation. They are practically the same for the five- and the six-

Table 2. Hydrogen Binding Capacities (b) and the Number of Adsorbed Molecules (N) for Selected Compounds (See Text for Details)

25289

dx.doi.org/10.1021/jp305324p | J. Phys. Chem. C 2012, 116, 25286−25292

The Journal of Physical Chemistry C

Article

molecules can reside only on the upper hemisphere around Li+; see Figure 1. From Table 1, one also sees that the number of hydrogen molecules, N(H2), attached to the molecular complex increases, whereas the binding energy of a single H2 molecule, E(H2), decreases in going from Li+ to Na+ and then to K+. We relate this behavior to a characteristic distance (d(I−H2) or d*(I−H2)) from bound H2 molecules and the cation. Indeed, from Table 1, it follows that N(H2) ∼ d(I−H2)2 while E(H2) ∼ d(I−H2)−3. Therefore, the number of bound H2 molecules is approximately proportional to the area of the upper hemisphere around the cation, while the binding energy is due to an effective quadrupole(H 2 )−monopole(cation) interaction, which has the 1/d3 dependence. Notice that Be2+ situated very close to the anionic ring has a very small radius. As a result, it can bind only one hydrogen molecule, but the energy of binding is very high, 50 kJ/mol. The peculiarities of the adsorption of a H2 molecule on Be in nanostructures are discussed in ref 38. By considering polyaromatic carbon molecules (e.g., naphthalene, pyrene) and by substituting in them one carbon atom with boron, we have found that the molecular configuration with Li+ atop the substituted (charged) ring is approximately 70 kJ/mol lower in energy than with Li+ atop the neutral carbon ring. Therefore, the anionic ring is clearly more favorable for the Li+ cation, which, in its turn, leads to a more stable material due to the suppression of the ion migration. The hydrogen binding capability of a doped ring with the counterion(s) atop (i.e., N(H2) and d*(Li+−H2) values) in the considered polyaromatic compounds remains the same as those of the corresponding single ring, and depends on the cations (Table 1). To compare the ability of various compounds to bind hydrogen molecules, we have introduced the hydrogen binding capacity defined as b = 2NH2 /(2NH2 + μc ) ·100%

possible structure on the basis of a graphite sheet (graphene) is shown in Figure 2a. Some of the present authors put forward

Figure 2. (a) Carbon sheet (graphene) doped with boron and lithium.17,21−23 Gray circles stand for carbon, blue for boron, red for lithium cations (Li+) above the plane, and white circles for lithium cations under the plane. (b) Cross section of boron- and lithiumdoped nanotube (1 Li per B) with adsorbed H2 molecules (3 H2 per Li).17,20,21 Gray circles stand for carbon, green for boron, red for lithium, and white for hydrogen.

this idea in refs 17 and 21. The structure has one boron atom per three carbon atoms and one lithium cation per each boron atom. Recently, DFT calculations of such structures have been performed by Park et al.22,23 Combining such sheets in a threedimensional structure, one should take care to leave enough space between the sheets to accommodate all suitable hydrogen molecules. Because a large space separation between carbon rings is available in carbon nanotubes, one can think of a material in the form of nanotubes doped with boron and coated with lithium17,20,21 shown in Figure 2b. Another possibility is to dope carbon nanomaterials, such as fullerenes, carbon nanocones, etc., or polymers and metal organic frameworks containing aromatic rings.21,39 It is also worth noting that anionic aromatic rings with counterions can form molecular crystals that are stable at ambient conditions (for example, CpLi). Such crystals can also be considered as candidates for hydrogen storage materials. For example, highly porous covalent organic frameworks (COFs) have been proposed40 as candidates for an effective hydrogen storage. Our calculations,21,41 of the hydrogen storage capacity for these substances, showed that they ensure a comparatively high gravimetric storage capacity at a pressure of 100 atm, up to 4% at room temperature and up to 18% at T = 77 K, but with a rather low volumetric storage capacity, 8 and 40 g/L, respectively. By modifying the structure of COFs with a method described above, one could increase the volumetric capacity in these and some of the similar structures by a factor of 2. For example, the modified COF (Figure 3) may provide a volumetric storage capacity up to 20 g/L at room temperature and up to 70 g/L at T = 77 K.21 There are also chemically bound (chemisorbed) low-energy states of H2 on carbon-based nanostructured materials, as we know from the previous studies.42−45 However, the activation barriers are found to be very high (>2.5 eV (240 kJ/mol)) for the pure carbon structures42 and some light-metal-doped structures.45 In a recent study,46 the significantly reduced activation barrier (1.38−4.52 kcal/mol (5.8−19 kJ/mol)) is reported for the H2 chemisorption state on B-substituted C20 fullerene (C19B). However, the doping considered in ref 46 is not isoelectronic; that is, the number of electrons in the molecule is reduced by one. As the result, the electron spin is changed, which means that the chemical activity is significantly

(1)

Here, NH2 is the number of hydrogen molecules, which a compound can bind with the energy of at least 4−5 kJ/mol, and μc is the molar mass of the compound. The introduced quantity allows one to assess (screen) various modifications of the carbon aromatic rings and, assuming them as building blocks for the prospective nanostructures, to have a guess of their hydrogen storage capacity relative to the nanostructures with pure carbon aromatic rings. In Table 2, we quote b for the best modifications of the single carbon rings considered in the present study, along with the calculated number of bound hydrogen molecules, N(H2). Notice that N(H2) per anionic ring can be 5−6 times larger than that for the neutral aromatic ring (benzene). Although heavy cations worsen the m(H2)/ m(molecular complex) mass ratio, the increase in the number of bound H2 molecules prevails and improves their hydrogen storage capacity in comparison with the neutral carbon compounds (Table 2). All the considered alkali counterions (Li+, Na+, K+) may be interesting for the modifications, as the binding capacity of each aromatic ring may be increased by 2 or 3 times. Among the alkali-earth ions, only Mg2+ deserves attention in that sense and is shown in Table 2. Be2+ is located too close to the ring and can bind only one hydrogen molecule, whereas Ca2+ can bind as many hydrogen molecules as Mg2+ does (six), but it is too heavy. Substituted anionic rings complemented with cations can be linked together, very similar to the initial carbon one.37 A 25290

dx.doi.org/10.1021/jp305324p | J. Phys. Chem. C 2012, 116, 25286−25292

The Journal of Physical Chemistry C

Article

Figure 3. Modified with 1,3-H-diboratabenzene disodium salt (see Figure 1d) carbon-based COF. An enlarged picture in the left panel shows a composite element of the framework. In the aromatic rings, boron is shown in dark color, and the ions above and under the ring are sodium.



altered. For the isoelectronic substitution discussed in the present paper, the situation looks much closer to the pristine carbon structures, and one may suggest that the activation barriers are also sufficiently high. We finally make a note that the negatively charged rings in carbon nanostructures may formally be achieved by substituting carbon with aluminum, which is situated just below boron in the periodic table and also has one unpaired electron in the outer p shell. Formally, this is not an isoelectronic substitution as Al has two additional complete core shells, 2s and 2p. However, the number of electrons in the outer, active shells is the same in B and in Al. Our calculations show that the Alsubstituted carbon anionic rings balanced by the cations may also attract hydrogen molecules, similar to the corresponding Bsubstituted rings. However, the planarity of the Al-doped ring is slightly violated, which may indicate the reduced stability. To assess the potential of the Al substitution versus the B substitution, additional studies are necessary.

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Phone: +27 (0)21 8083865. Fax: +27 (0)21 8083862. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The authors are thankful to Dr. S. Nechaev for useful discussions. I.V.B. acknowledges the European Research Council (ERC) Starting Grant FP7 Project DEDOM, grant agreement no. 207441, for the support in this work.



REFERENCES

(1) Winter, C.-J.; Nitsch, J. Hydrogen as an Energy Carrier: Technologies, Systems, Economy, 1st ed.; Springer: Berlin, 1988. (2) Schlapbach, L.; Zuttel, A. Nature 2001, 414, 353−358. (3) http://pubs.acs.org/books/references.shtml. (4) http://www1.eere.energy.gov/hydrogenandfuelcells/. (5) Panella, B.; Hirscher, M.; Roth, S. Carbon 2005, 43, 2209−2214. (6) Durgun, E.; Ciraci, S.; Zhou, W.; Yildirim, T. Phys. Rev. Lett. 2006, 97, 226102. (7) Kim, Y.; Zhao, Y.; Williamson, A.; Heben, M.; Zhang, S. Phys. Rev. Lett. 2006, 96, 016102. (8) Lee, H.; Choi, W. I.; Ihm, J. Phys. Rev. Lett. 2006, 97, 056104. (9) Sabir, A. K.; Lu, W.; Roland, C.; Bernholc, J. J. Phys.: Condens. Matter 2007, 19, 086226. (10) Sun, Q.; Jena, P.; Wang, Q.; Marquez, M. J. Am. Chem. Soc. 2006, 128, 9741−9745. (11) Yildirim, T.; Ciraci, S. Phys. Rev. Lett. 2005, 94, 175501. (12) Heine, T.; Zhechkov, L.; Seifert, G. Phys. Chem. Chem. Phys. 2004, 6, 980−984. (13) Niu, J.; Rao, B.; Jena, P. Phys. Rev. Lett. 1992, 68, 2277−2280. (14) Rao, B.; Jena, P. Europhys. Lett. 1992, 20, 307−312. (15) Sun, Q.; Jena, P.; Wang, Q.; Marquez, M. J. Am. Chem. Soc. 2006, 128, 9741−9745. (16) Sun, Q.; Wang, Q.; Jena, P.; Kawazoe, Y. J. Am. Chem. Soc. 2005, 127, 14582−14583. (17) Bodrenko, I.; Tkalya, E.; Bibikov, A.; Queen, C. Method for Storing Hydrogen Using Novel Carbon Based High Capacity Storage Materials. U.S. Patent 021861-000300US, 2006. (18) Hoic, D.; Davis, W.; Fu, G. J. Am. Chem. Soc. 1995, 117, 8480− 8481. (19) Ashe, A. J., III; Shu, P. J. Am. Chem. Soc. 1971, 93, 1804−1805. (20) Wu, X.; Gao, Y.; Zeng, X. C. J. Phys. Chem. C 2008, 112, 8458− 8463. (21) Avdeenkov, A. V.; Bibikov, A. V.; Bodrenko, I. V.; Nikolaev, A. V.; Taran, M. D.; Tkalya, E. V. Russ. Phys. J. 2009, 52, 1235−1241.



CONCLUSION To recapitulate, by employing the ab initio electronic structure approach, we have studied the hydrogen adsorption on molecular complexes having anionic aromatic carbon rings doped with boron atoms and stabilized by cations of alkali or alkali-earth metals. Our results indicate that the complexes can effectively bind several hydrogen molecules. We have found that hydrogen molecules are located atop the molecular complex on the upper hemisphere about the cation (Figure 1). The plausible explanation of this conclusion is that the positive ions together with the negatively charged base generate an electrostatic field that is influenced by hydrogen molecules. The interaction of the hydrogen molecule with the electrostatic field leads to a subsequent binding of the hydrogen molecules (Table 1). The anionic aromatic substrate in this binding mechanism plays a crucial role by holding the ions in place. In fact, in the first approximation, the interaction can be considered as due to the monopole component of the cation and the quadrupole component of the hydrogen molecule. This viewpoint is supported by the data from Table 1. On the basis of our studies, we have chosen molecular complexes that show promising results and estimated their hydrogen binding capacity (Table 2). We further argue that the substituted rings can be linked together similarly37 to such well-known structures as carbon nanotubes, fullerenes, and other molecular nanostructured materials. 25291

dx.doi.org/10.1021/jp305324p | J. Phys. Chem. C 2012, 116, 25286−25292

The Journal of Physical Chemistry C

Article

(22) Park, H.-L.; Yi, S.-C.; Chung, Y.-C. Int. J. Hydrogen Energy 2010, 35, 3583−3587. (23) Park, H.-L.; Yoo, D. S.; Chung, Y.-C. Jpn. J. Appl. Phys. 2011, 50, 06GJ02. (24) Dion, M.; Rydberg, H.; Schroder, E.; Langreth, D.; Lundqvist, B. Phys. Rev. Lett. 2004, 92, 246401. (25) Langreth, D.; Dion, M.; Rydberg, H.; Schroder, E.; Hyldgaard, P.; Lundqvist, B. Int. J. Quantum Chem. 2005, 101, 599−610. (26) Schmidt, M.; Baldridge, K.; Boatz, J.; Elbert, S.; Gordon, M.; Jensen, J.; Koseki, S.; Matsunaga, N.; Nguyen, K.; Su, S.; et al. J. Comput. Chem. 1993, 14, 1347−1363. (27) Andersson, K.; Malmqvist, P.; Roos, B.; Sadlej, A.; Wolinski, K. J. Phys. Chem. 1990, 94, 5483−5488. (28) Andersson, K.; Malmqvist, P.; Roos, B. J. Chem. Phys. 1992, 96, 1218−1226. (29) Artemyev, A.; Bibikov, A.; Zayets, V.; Bodrenko, I. J. Chem. Phys. 2005, 123, 024103. (30) Harihara, P. C.; Pople, J. Theor. Chim. Acta 1973, 28, 213−222. (31) Francl, M.; Pietro, W.; Hehre, W.; Binkley, J.; Gordon, M.; Defrees, D.; Pople, J. J. Chem. Phys. 1982, 77, 3654−3665. (32) Rassolov, V.; Pople, J.; Ratner, M.; Windus, T. J. Chem. Phys. 1998, 109, 1223−1229. (33) http://www.emsl.pnl.gov/forms/basisform.html. (34) Dunning, T. J. Chem. Phys. 1989, 90, 1007−1023. (35) Hubner, O.; Gloss, A.; Fichtner, M.; Klopper, W. J. Phys. Chem. A 2004, 108, 3019−3023. (36) CpLi is a commercially available compound and can be obtained from Acros Organics. (37) Olson, J. K.; Boldyrev, A. I. Chem. Phys. Lett. 2012, 523, 83−86. (38) Lee, H.; Huang, B.; Duan, W.; Ihm, J. Appl. Phys. Lett. 2010, 96, 143120. (39) Getman, R. B.; Bae, Y.-S.; Wilmer, C. E.; Snurr, R. Q. Chem. Rev. 2012, 112, 703−723. (40) El-Kaderi, H. M.; Hunt, J. R.; Mendoza-Cortes, J. L.; Cote, A. P.; Taylor, R. E.; O’Keeffe, M.; Yaghi, O. M. Science 2007, 316, 268−272. (41) The method of calculation of the gravimetric and the volumetric storage versus temperature and pressure will be considered in detail in the forthcoming article. (42) Lee, E.; Kim, Y.; Jin, Y.; Chang, K. Phys. Rev. B 2002, 66, 073415. (43) Miura, Y.; Kasai, H.; Dino, W.; Nakanishi, H.; Sugimoto, T. J. Appl. Phys. 2003, 93, 3395−3400. (44) Wang, Z.; Yao, M.; Pan, S.; Jin, M.; Liu, B.; Zhang, H. J. Phys. Chem. C 2007, 111, 4473−4476. (45) Chen, B.; Li, B.; Chen, L. Appl. Phys. Lett. 2008, 93, 043104. (46) Tian, C.; Wang, Z.; Jin, M.; Zhao, W.; Meng, Y.; Wang, F.; Feng, W.; Liu, H.; Ding, D.; Wu, D. Chem. Phys. Lett. 2011, 511, 393−398.

25292

dx.doi.org/10.1021/jp305324p | J. Phys. Chem. C 2012, 116, 25286−25292