Skip to main content
NCBI home page
Scientific Reports logoLink to Scientific Reports
. 2013 May 23;3:1882. doi: 10.1038/srep01882

Three-dimensional metal-intercalated covalent organic frameworks for near-ambient energy storage

Fei Gao 1, Zijing Ding 1, Sheng Meng 1,a
PMCID: PMC3662009  PMID: 23698018

Abstract

A new form of nanoporous material, metal intercalated covalent organic framework (MCOF) is proposed and its energy storage property revealed. Employing density functional and thermodynamical analysis, we find that stable, chemically active, porous materials could form by stacking covalent organic framework (COF) layers with metals as a gluing agent. Metal acts as active sites, while its aggregation is suppressed by a binding energy significantly larger than the corresponding cohesive energy of bulk metals. Two important parameters, metal binding and metal-metal separation, are tuned by selecting suitable building blocks and linkers when constructing COF layers. Systematic searches among a variety of elements and organic molecules identify Ca-intercalated COF with diphenylethyne units as optimal material for H2 storage, reaching a striking gravimetric density ~ 5 wt% at near-ambient conditions (300 K, 20 bar), in comparison to < 0.1 wt% for bare COF-1 under the same condition.

Finding low cost, safe, and efficient energy storage materials has been a major challenge to develop renewable-energy based economy. Hydrogen is considered as an appealing energy carrier alternative to fossil fuels in many applications, because it is abundant, energy intensive, and pollution-free1,2,3. U.S. Department of Energy (DOE) set a target for practical H storage materials to have a gravimetric capacity of 5.5 wt% by 20174. The challenge is to dramatically increase storage capacity and reversibility at ambient conditions5. Carbon-based nanostructures and porous materials including fullerene, nanotube, and carbyne networks, have been proposed as potential H2 storage media, thanks to their high surface-to-weight ratios, good reversibility and fast kinetics6,7,8,9,10. However, most of them failed at room temperature and ambient pressure due to improper binding to H2.

To reach high and retrievable H storage at ambient conditions, a light three-dimensional (3D) material with H2 binding energies in the range of 0.2–0.6 eV/H2 is required11,12. This implies porous materials must have dense active sites for H2 binding, to which two approaches have been adopted. One is to substitute carbon by other elements such as boron/nitrogen11,13; however, it shows little effect and is difficult to apply. The other is to dope light metals including alkali, alkali earth or transition metals, into porous materials14,15. Improving metal binding by using nanostructure curvature16,17 or incorporation of sp-carbons18,19 has been also explored. It was found alkali metals have too weak interaction with the substrate and H220; whereas transition metals tend to cluster on material surface, reducing substantially H2 storage capacity21. Upon compromise, calcium is recognized as a superior coating element thanks to its delicate balance of binding strengths to base materials and H218,22,23,24. Ca-coated fullerene, Ca32C60, yields strong local electric fields on C60 and reaches a H2 uptake up to ~ 8.4 wt%22. Ca-doped carbyne networks maintain a small distance ~ 10 Å between metal adsorbates and a high H2 density of ~ 8 wt%18. However, both materials have their own flaws: their 3D assembly without affecting H2 density is either not achievable or not experimentally demonstrated; and it is difficult to control the position and stability of dopant Ca atoms18,22. Similar problems emerge in Ca-doped nanotube, graphite and graphene14,23,24.

Covalent organic framework (COF) has been recently synthesized as an important extension of carbon porous materials25,26. A variety of experimental studies27,28 show that small planar organic molecules form two-dimensional (2D) or 3D supramolecular architectures including COFs, through self-assembly of end linker groups, as schematically shown in Figure 1. Due to their adjustable porous structure and unparalleled strength, COFs are very promising for H2 storage. At low temperature 77 K and high pressure 100 bar, 2D or 3D COFs can currently store 3–7 wt% H227. Stunning H2 uptake of ~ 20 wt% at 77 K and 100 bar has been predicted in theoretical simulations for COF-108 and COF-102-3; but it deteriorates quickly to a value below 1–2 wt% at near ambient condition, for instance, at 300 K and 20 bar29,30,31. Similar trends were observed for COF-1, a simple layered COF with the phenyl group as organic unit and boronic acid as the linker (see Fig. 1), shows a H2 density of 1.28 wt% at 77 K but it decreases to ~ 0.08 wt% at 298 K and 20 bar in experiment27,32. Suitable metal doping also improves H2 storage in COFs33,34,35,36,37,38. Grand canonical Monte Carlo simulations reveal that H storage capacity in Li-coated COF-202 can reach 4.39 wt% at 298 K and 100 bar, which is three times higher than in undoped COF-20234. However, with the increase of Li concentration, Li atoms form clusters and H storage density reduces35. Using density functional theory (DFT) calculations, Wu et al. found that both Li and Ca forms clusters on COF-10, which are 0.17 and 0.20 eV/atom more stable than scattered atoms, respectively. This reduces storage density from 3 to 1.5 H2 per metal atom. Ca dopants provide a stronger bonding to H2 (energy ~ 0.21 eV) compared with Li dopants (0.12 eV), due to more charge transfer from Ca to the substrate35. As a remedy, two-step doping, namely, metal decoration on B-substituted COF fragments, was also proposed36. On all these successful 3D reference materials, H2 density is below 1–2 wt% at near ambient conditions30,31. New methods and concepts for further improvement without metal aggregation are key to develop next generation H storage media.

Figure 1. Formation of COFs.

Figure 1

Open in a new tab

(a) Schematic diagram of COF building block composed by two ending linkers and an organic unit, (b) An example of real building blocks: benzene diboronic acid (BDBA), (c) 2D COF formed by BDBA self-assembly, named COF-1. Crosses mark positions tested for metal binding.

Here we present a radically different approach to improve H storage in COFs. We explore the feasibility of using a new class of 3D network—metal-intercalated COF (MCOF), for near ambient H2 storage. Metal binding in such new class of 3D materials (see the inset of Fig. 2a) has a binding energy significantly larger than that in bulk metals, thus intrinsically blocking metal segregation. Quantum mechanical calculations and thermodynamical analysis reveal that optimized MCOF material stores 4.87 wt% of H2 at 300 K and 20 bar, greatly enhanced from the value of < 0.1 wt% for the bare COF-1 in the same condition32.

Figure 2. Partial density of states (PDOS) projected onto the COF component (black lines) and the d-orbitals of Ca (red lines) in (a) CaCOF-1 and (b) CaCOF-1 after H2 adsorption.

Figure 2

Open in a new tab

The PDOS projected onto H2 blue line, multiplied by 20 times) is also shown in (b). The inset in (a) shows the side view of CaCOF-1, and the inset in (b) shows the charge density associated with the states in the shaded region.

The paper is organized as follows. After this introduction section, we first expound the idea and important considerations in constructing MCOF in the “Results” section. The feasibility of various light metal elements intercalating in model COF layers, COF-1, is tested. Ca is identified to be the most suitable gluing metal element. To further improve H storage performance, we start COF structural optimization through the selection of organic building blocks and linkers. In optimizing organic blocks, small organic molecules as well as polycyclic aromatic hydrocarbons (PAHs) are considered and compared to the reference molecule, benzene. We also investigate the effects of linker groups, where we find –B(OH)2 linkers preserve H2 binding properties of doped organic units. Ca intercalated COF layers with diphenylethyne organic units (CaCOF-d) are identified as an outstanding H storage material. After discussing H2 storage properties in 3D CaCOF-d layers at saturation and near ambient conditions, we give a brief discussion on how MCOF can be synthesized experimentally, and present a summary on our major findings.

Results

It seems natural to combine light metals and extended COF layers for near ambient H2 storage. The two are both light and stable, and can potentially form new framework with dense active sites. In early experiments the newly synthesized COF materials are found to interact weakly with H2 molecules, leading to low storage density at ambient conditions. In the meantime, various theoretical efforts demonstrated that metal doping cannot effectively improve H2 storage due to metal clustering problem on single COF layers. We start a new approach to explore that whether metal atoms could be intercalated between COF layers to form stable 3D stacks without metal segregation. The choice of workable metal elements and COF materials should be tested to satisfy the requirements of being stable, light, and active for H2 storage.

Constructing metal intercalated COF for H2 storage

We first take one of the simplest COF materials, COF-1, whose organic unit is a benzene ring and linker boronic acid, as our representative model substrate and investigate its interaction with various metal elements. A range of light metals including Li, Na, Mg, K, Ca, and Ti are considered, since they are most commonly adopted to decorate porous nanomaterials serving as active sites for H2 binding. We first test the adsorption site of metal atoms on a single layer COF-1. We find metal atoms prefer to adsorb onto the center of the organic unit, a phenyl ring, rather than on top of B, O, or B3O3 rings. For instance, the binding energy of Ca (ECa) on all sites falls in the range of 0.001–0.008 eV, except on the center of phenyl ring, which is 0.77 eV. This site preference is a result of Ca interacting with the π orbital of phenyl ring, which is absent from the electron rich boroxine ring. As shown later, the energy for Ca binding to the phenyl group in COF is an order of magnitude larger than that on isolated benzene (0.07 eV), because the π orbitals are extended onto neighboring B atoms connected to phenyl rings in COF and Ca binding is thus enhanced by delocalized π electrons, as clearly displayed in Fig. 2b inset.

Most importantly, the Ca binding energy increases to 2.83 eV when Ca and COF-1 layers assemble into multilayered metal intercalation structure, the MCOF. The interlayer distance increases from 3.35 Å to 4.40 Å upon metal intercalation and the stacking manner might also be changed. Ca binding on boroxine rings is ~ 1 eV less stable than on phenyl rings in intercalated 3D COF layers. We expect metal intercalation takes place simultaneously or with a small barrier through the pores of a diameter of ~ 10 Å present in COF-1. The Ca binding energy is 3.37 eV if Van der Waals density functional (VDW-DF)39,40 is used to incorporate van der Waals (VDW) interactions; the additional 0.54 eV is attributed to the VDW interactions between adjacent COF layers. For comparison, the cohesive energy in bulk Ca is 1.88 eV/Ca in our calculation and the experimental value is 1.86 eV/Ca40, much smaller than Ca binding energy in COFs. The energy for a second Ca binding around the Ca on the phenyl ring center is 1.81 eV with a Ca-Ca distance of 3.28 Å; the energy is smaller than the binding energy of first Ca and the cohesive energy of Ca bulk, showing straightforwardly that Ca clustering is not favored. Vibrational analysis also confirms that no major negative frequencies occur in the vibrational spectrum of CaCOF-1. Therefore, Ca and COF-1 forms stable 3D materials (referred to as CaCOF-1) via metal intercalation.

Some other metals also work in the same way as Ca to form MCOFs. In Table 1, we list the calculated parameters including the interlayer distance (L) and metal binding energy (EM) for metal intercalation COF-1. Metal insertion into the COF layers significantly expands the equilibrium interlayer distances, from L = 3.4 Å for bare COFs25 to L = 5.4 Å for KCOF-1. The binding energies of metals to COFs vary from metal to metal, with EM for Mg being smallest (≤ 0.2 eV) and EM for Ti being largest (≥ 5.0 eV) in our calculations. This is in accord with the rather inert (active) nature of Mg (Ti) metals. However, both cases have the metal binding energy, EM, smaller than the cohesive energy in bulk metals, EC, as is the case for most metal doped nanostructures previously studied22. Surprisingly, on the other hand, Li, Na, K and Ca metals have EM significantly larger than the corresponding Ec (see Table 1). This behavior renders metal segregation in such MCOFs thermodynamically unstable. Consequently, 3D network formed in this way is a new class of stable materials without any metal clustering problems.

Table 1. Metal binding energies (EM) and interlayer distances (L) for 3D stacked MCOFs. The cohesive energies (EC) for bulk metals are also listed for comparison.

  MCOF-1 MCOF-d Bulk metal
  L (Å) EM (eV) L (Å) EM (eV) EC (eV)
Li 3.7 2.289 3.9 2.231 1.605
Na 4.6 1.408 4.7 1.490 1.048
Mg 4.3 0.057 3.7 0.202 1.296
K 5.4 1.596 5.4 1.708 0.798
Ca 4.4 2.830 4.4 2.786 1.877
Ti 3.6 5.058 3.6 4.993 5.207
Open in a new tab

The new MCOF material can be approximately viewed as reduced intercalation compounds with metal dopants in cation states. Our Bader charge analysis shows that there are about 1.42 electrons transferred from Ca to COF-1 during CaCOF-1 formation, and 0.84 electrons transferred from Li to COF-1. This new class of materials contains porous structures with stable, active metal sites, which provide extraordinary advantages for molecular adsorption and storage, in particular, for H2 storage. In this aspect, MCOF as the new class of 3D material is significantly different from earlier proposals including those employing Ca32C60 and other metal-doped materials, in that:

i) It is an extended three-dimensional framework that can be readily exploited in practical applications; while Ca32C60 is finite whose 3D assembly is not feasible without significantly reducing H2 density.

ii) MCOF is thermodynamically stable with metal binding energies greater than that in bulk metals; while on C60, Ca has a binding energy of 1.3 eV, significantly smaller than that in bulk Ca (1.88 eV)18,22, therefore metal segregation is a severe problem during 3D assembly.

iii) MCOFs have intrinsic well-ordered pores freely available for H2 diffusion during charge/discharge processes; while Ca32C60 material does not have this advantage.

The porous MCOF-1 structures with stable, active metal sites are ideal candidates for H2 storage. We employ first-principles DFT to study the interaction between MCOF and H2 molecules, to eliminate uncertainties in H2 binding configuration and energies in empirical calculations. Structure optimization with a full coverage of four intact H2 per metal atom was performed for CaCOF-1 and LiCOF-1. We found that LiCOF-1 fails to bind H2 strongly. The H2 binding energy (Eb) varies from ~ 0.05 eV/H2 to 0.02 eV/H2 for a single H2 and for 4 H2 adsorption per Li, due to the fact that Li is strongly bound to the upper and lower COF-1 layers and the interlayer spacing is rather limited (3.7 Å). On the other hand, CaCOF-1 can serve as an excellent H2 storage material with the H2 binding energy of 0.13 eV/H2 with generalized gradient approximation (GGA) exchange-correlation functional and 0.42 eV/H2 with local density approximation (LDA), corresponding to a gravimetric storage capacity of 4.54 wt% and volumetric density of 35.2 g/L at 0 K, very close to the 2017 DOE target. Using VDW-DF the H2 binding energy to CaCOF-1 is 0.18 eV/H2, a reasonable value lying between GGA and LDA numbers, which are known to underestimate and overestimate Eb, respectively11. Simpler Grimme's DFT+dispersion approach also gives reasonable values for VDW interaction energies between H2 and metal sites in metal organic frameworks41. In high-level MP2 and quantum Monte Carlo calculations a binding energy of ~ 0.2 eV was obtained for H2 binding to partially charged Ca, with basis-sets extrapolated to complete basis limit42,43. This number is close to our results. Based on the calculated H2 binding energy, it is shown later the high H density persists at near-ambient conditions. For comparison, H2 storage capacity of the COF-1 without metal intercalation was found only 1.28 wt% at 77 K and ambient pressure, and 0.08 wt% at 298 K and 20 bar in experiment32.

The nearly sixty times increase in storage density under near ambient conditions in COF-1 upon Ca intercalation is attributed to the presence of dense, active metal sites and the ideal binding energy to H2. In CaCOF-1 framework, the empty Ca 3d orbitals interact strongly with the p orbitals of COF-1, presenting large density of states (DOS) around the Fermi level, see Fig. 2. These states are saturated upon H2 binding. The energy distribution of H2 orbitals is highly overlapped with Ca 3d orbitals, suggesting H2 binding onto CaCOF-1 via a Kubas mechanism, namely, hybridization between the antibonding σ* orbital of H2 and the d-orbital of Ca3,22.

Optimizing MCOFs: the organic building blocks

Although CaCOF-1 has shown outstanding property and great promise for H2 storage, it is unclear whether this approach can be extended to other systems, or can be further improved to produce new materials satisfying DOE target of a 5.5 wt% density. We then explore a range of possible alternatives of organic building units and illuminate the systematic trends in H2 binding behaviors, in the hope to optimize MCOF building blocks for enhanced H2 storage capacity and to identify important parameters for MCOF design.

Benzene as the reference organic unit

To seek an optimal organic building block, the organic unit of COF-1, a benzene ring, was first studied as a standard reference. Ca binding to C6H6 produces a small energy, 0.07 eV/Ca, less than the binding energy of Ca to the single layer COF-1, due to limited electron conjugation in isolated benzene molecules. The first H2 adsorbs onto CaC6H6 in intact form, with H-H bond length (dHH) expanding from 0.749 to 0.785 Å. With the increase of the number of adsorbed H2 molecules (n), H2 binding energy and H-H bond length both first increase, then decrease, shown in Fig. 3. The maximum averaged H2 binding energy Eb = 0.34 eV/H2 is reached when n = 4; and the maximum dHH reaches 0.804 Å at n = 3. The maximum is not for n = 1 because of the competition between electron polarization and Pauli repulsion. We note that there is a strong correlation between average Eb and dHH, which might be used to steer optimal design of H2 storage material. PDOS analysis for 1-5 H2 adsorption onto CaC6H6 clearly indicates there is a largest mixing between H2 states and Ca d-orbitals and a high occupation of H2 antibonding states and Ca d-orbitals for n = 3 and 4. The sixth H2, initially placed close to Ca, is repelled to a distance larger than 6 Å, suggesting the system with five H2 to be saturated. The maximum of five H2 molecules offer 10 electrons, together with 2 electrons from Ca and 6 π-electrons from benzene, the 18-electron rule is nicely satisfied as proposed for H2 storage on transition metal complexes15. The average binding energy is ~ 0.29 eV/H2. For comparison, on Ca-decorated terephthalic acid, H2 binding energy was calculated to be 0.20 eV in GGA and 0.24 eV in MP2 with completed basis set42. Among the adsorbed H2 molecules, we observe that H2 prefers to adsorb in the neighborhood of Ca-C6H6, than on the top site of Ca. This is also indicated by the difference in binding energies (0.11 vs 0.06 eV) and H-H lengths (0.785 vs 0.752 Å) on the two sites, respectively. It is because the neighboring four molecules interact with Ca via the Kubas mechanism, as shown in Fig. 4; while the top H2 is bound via the long-ranged electrostatic attraction since no overlap between H2 bound state and Ca d-orbital was observed in PDOS12. Notice that in the case of Ca coated fullerene22, the maximal number of adsorbed H2 molecules per Ca is five with the same trend for Eb. The Ca binding energy increases to 0.73 eV/Ca with two Ca bound on both sides of benzene, perhaps due to the more polarized π orbitals of benzene than the case of single Ca adsorption. Detailed analysis shows that upon two Ca adsorption, there are +0.99 electrons transferred from each Ca to benzene, larger than that for a single Ca adsorption (+0.88 electrons). The H2 can be adsorbed in total, leading to H storage capacity of 11.2 wt% (0 K), with dHH = 0.785 Å; the Eb of 0.23 eV/H2 falls in the ideal range of 0.2–0.6 eV/H2.

Figure 3.

Figure 3

Open in a new tab

(a) Average H2 binding energy (Eb) and H-H bond length (dHH) as a function of the number of adsorbed H2 on Ca-C6H6. (b) Optimized geometry of ten adsorbed H2 on Ca2C6H6.

Figure 4.

Figure 4

Open in a new tab

(a) Side and (b) top view of charge density difference after Ca binding onto benzene. Isosurfaces with values of ± 0.005 e/Å3 are shown. Red and blue clouds correspond to electron depletion and accumulation, respectively. (c) Same as (b) but for both two Ca and 10 H2 binding to benzene. (d),(e) Same as (a) and (b) but for Ca binding on five-membered C5H5. (f) The H2 binding energy as a function of distance (z) between Ca and H2 on C5H5.

Comparison to other small molecules

Then a series of Ca-decorated small organic molecules including C2Hm (m = 2,4,6) and ring structures are investigated. Although some molecules are not yet directly used as COF building blocks, exploring their potential performance yields valuable information for comparison to currently available materials and provides new insights for future applications. Compared with C6H6, we find Ca binding to the first three molecules is very weak, with ECa ≤ 0.01 eV, making them unstable for hosting H2. The binding energy of Ca to neutral C5H5 is 3.24 eV, showing a strong interaction between them. This is a result of electron deficiency in C5H5: to form a stable aromatic system44, 6 π electrons are required; where C5H5 supplies only 5, with another electron from Ca. Therefore the CaC5H5 complex formed is a product of reduction-oxidation chemical reaction, namely, C5H5 is reduced by an electron from Ca to form a stable compound. Very surprisingly, the Ca-C5H5 seems to be saturated and does not bind any H2. No matter where the H2 molecule is initially placed, it is repelled to a distance > 4 Å from Ca with negligible binding energies (≤ 0.005 eV/H2), shown in Fig. 4(f). It implies the electrostatic force is dominant. This behavior is in contrast to that for Ca-C6H6, suggesting that strong interaction between Ca and the organic component does not necessarily introduce a strong binding to H2; instead, the metal-organic complexes get saturated and become inert to H2 binding. Therefore organic systems with too strong or too weak Ca binding fail to bind H2 well. A good balance between stability (Ca-binding) and reactivity (Ca-H2 interaction) is required for optimal H2 storage.

Next, we substitute carbon atoms in benzene by boron or nitrogen, to check whether H storage properties are affected. Typical structures considered and results are shown in Fig. 5 (first row) and Table 2. We found substitutional structures with more than three B deform and are unstable, thus are not listed. Ca-doped C4H6B2 and C4H4N2 show much enhancement for the first H2 adsorption, so we study multiple H2 adsorption on them. The two can both host up to five H2. The corresponding Eb = 0.22 eV and dHH = 0.771 Å for H2 on CaC4H6B2, and Eb = 0.14 eV and dHH = 0.784 Å for CaC4H4N2. While they retain similar H2 binding properties as that for Ca-C6H6, the Ca-binding is significantly improved, leading to more stabilized material and benign thermodynamical properties without metal clustering. This is due to the fact that the Ca-C4H6B2 complex is formed by chemical redox reaction, where Ca electrons are transferred to two-electron deficient C4H6B2 molecule. If we dope two Ca to C4H6B2 on both sides, ten H2 are adsorbed, with ECa = 2.56 eV/Ca, dHH = 0.771 Å, and Eb = 0.17 eV/H2, corresponding to H2 storage density of 11.5 wt% (0 K). The experimental route for B-substitution might be complicated and their stability requires further investigation. Similar results36,45,46 were observed for Ca doping in B-substituted metal-organic framework MOF-5 and COF.

Figure 5. A variety of organic molecules serving as COF building units studied in this work.

Figure 5

Open in a new tab

Cross symbols indicate stable positions for Ca binding.

Table 2. Ca binding energy (ECa), adsorbed H-H bond length (dHH) and energy (Eb) for H2 on six- and five- membered rings.

  C6H6 C4H4N2 C5H6B C4H6B2 C3H6B3 C4H4O C4H4S C4H5N
ECa (eV) 0.07 0.73 3.02 3.13 4.80 0.19 0.11 0.25
Eb (eV) 0.11 0.15 0.01 0.35 0.02 0.20 0.27 0.35
dHH (Å) 0.785 0.774 0.753 0.805 0.755 3.778 3.846 3.680
Open in a new tab

We also consider substitution of C in the five membered rings. Three existing heterocycles, furan (C4H4O), tetrahydrothiophene (C4H4S) and pyrrole (C4H5N) are tested. Ca binding energies are 0.10–0.25 eV on these structures, and the first H2 is dissociatively adsorbed on the top of Ca, with dHH ≥ 3.680 Å (Table 2). Less than two more H2 molecules can be adsorbed further, with the binding energy of 0.1–0.2 eV/H2. They are not suitable for H2 storage.

Large polycyclic aromatic hydrocarbons

Polycyclic aromatic hydrocarbons are formed when two or more benzene rings are connected in the same aromatic plane. The interaction energy between Ca and the isolated compounds increases to around 1.1 eV, shown in Table 3. In this case, the distance between adjacent Ca atoms, dCa–Ca, is an important parameter. If the distance is too small, it could lead to metal clustering or H2 bound to two Ca atoms, both reducing H storage capacity. We found on p-terphenyl (C18H14) dCa–Ca = 4.105 Å, and four H2 molecules between adjacent Ca dissociate by the strong reaction to Ca. With the distance extends to 6.847 Å on diphenylethyne, dissociation of H2 molecules disappears and all H2 are kept intact with Inline graphic. As a result, the total number of twenty H2 are adsorbed on Ca-doped diphenylethyne (five per Ca), with a capacity of 10.6 wt% (0 K). However, for these isolated molecules, Ca binding energy is smaller than its cohesive energy in bulk, the metal clustering problem might occur if using these isolated compounds for H2 storage.

Table 3. Number of adsorbed Ca atoms (NCa), average binding energies per Ca (ECa) and distances between adjacent Ca (dCa–Ca) on representative PAHs. n is the number of adsorbed intact H2 molecules.

  Naphthalene Anthracene Biphenyl p-Terphenyl Diphenylethyne
  C10H8 C14H10 C12H10 C18H14 C14H10
NCa 4 4 4 6 4
ECa (eV) 1.14 1.06 1.16 1.13 0.80
dCa–Ca (Å) 3.326 3.859 3.658 4.105 6.847
n 8 11 12 18 20
Open in a new tab

Optimizing MCOFs: effects of the linker groups

Besides the organic units, linker groups play an essential role in forming COFs via connecting organic units studied above. Two common covalent linkers, –COOH and –B(OH)2, are usually employed. The carboxylic acid groups form hydrogen-bonded assemblies28, while –B(OH)2 groups produce a six-membered ring (B3O3) after sublimation of three water molecules27. Five-membered BO2C2 ring could also form if hydroxyl groups (–OH) are readily available during COF synthesis27.

To investigate the effect of linking groups on H storage, we add –COOH and –B(OH)2 to four Ca-decorated compounds, including benzene, biphenyl, p-terphenyl and diphenylethyne. Similar to Ca-coated benzene, ten H2 molecules can be stably adsorbed on Ca2C6H4(B(OH)2)2. Slight changes in binding energy of H2 from 0.23 to 0.21 eV, and in dHH from 0.785 to 0.787 Å are observed. Ca binding is also improved by 0.203 eV, due to π electron delocalization from phenyl rings to more electron-negative linker groups. For Ca-coated biphenyl, we compare H2 adsorption on compounds with –B(OH)2 and –COOH linkers, shown in Table 4. While the former shows little difference from C12H10 in every respect, the latter enlarges Ca-Ca distance from 3.658 to 4.155 Å, leading to the number of adsorbed H2 increasing from 12 to 16. This is attributed to a stronger interaction between Ca and phenyl rings when –COOH linkers are connected to phenyl rings. Therefore, the presence of –B(OH)2 linkers preserves H storage properties of the original compound, while the –COOH groups increase the H2 capacity by elongating Ca-Ca distances.

Table 4. Average binding energy per Ca (ECa) and the distance between adjacent Ca (dCa–Ca) on biphenyl with or without linker groups. Also shown are total number of adsorbed H2 molecules (n), and their bond lengths (dHH) and binding energies per H2 (Eb).

  C12H10 C12H8(B(OH)2)2 C12H6(COOH)4
ECa (eV) 1.16 1.25 1.46
dCa–Ca (Å) 3.658 3.695 4.155
n 12 12 16
Eb (eV) 0.13 0.15 0.16
dHH (Å) 0.765 0.765 0.776
Open in a new tab

We adopt this principle to select linker groups for Ca-decorated PAHs. The Ca-Ca separation on C18H10(COOH)4 is only a little larger than that on C18H14 (from 4.105 to 4.244 Å), but H2 capacity increases greatly, from 18 to 24 H2 molecules per unit, shown in Fig. 6(a,b). On the other hand, since the Ca-Ca separation is already too large on diphenylethyne (6.8 Å), we use –B(OH)2 linkers to preserve its H2 storage property. As expected, all parameters such as ECa = 0.87 eV, Eb = 0.17 eV, dCa–Ca = 6.9 Å, and dHH = 0.783 Å are very close to the original complex. The same number of H2 molecules is adsorbed. Importantly, both systems have a desirable H2 binding energy of ~ 0.17 eV/H2. So the Ca4C14H8(B(OH)2)2 unit could potentially serve as an optimal building block promising for constructing MCOF to store H2.

Figure 6. Optimized geometry of (a) 22 H2 on Ca6C18H14, (b) 24 H2 on Ca6C18H10(COOH)4, (c) 20 H2 on Ca4C14H8(B(OH)2)2.

Figure 6

Open in a new tab

Properties of three-dimensional CaCOF-d

The optimization procedures described above identify Ca intercalated COF structure with diphenylethyne organic unit and B3O3 linkers, named as COF-d hereafter, as most promising for H2 storage. Similar to COF-1, COF-d also forms 3D MCOF structures upon intercalation with metal elements such as Ca and Li, which preserves the ideal H2 storage property in low-dimension materials. We then calculate the equilibrium structure and energy storage properties of the 3D intercalated CaCOF-d material in a rhombic unit cell with in-plane dimensions of 27.24 Å × 27.24 Å and vary its interlayer distance. The unit cell contains three branches of diphenylethyne connected to the center boroxine ring, namely, it consists of 42 carbon atoms, 24 hydrogen atoms, 6 oxygen atoms, 6 boron atoms and 6 calcium atoms before H2 adsorption (see Fig. 7). The optimized interlayer spacing is found to be 4.4 Å for the minimum energy configuration, adopting an eclipsed stacking manner to maximize interlayer interactions. We note the eclipsed stacking is chosen for all COFs with relatively large organic building blocks, with or without metal intercalation25. The Ca binding energy in CaCOF-d, with respect to isolated COF-d layer and free Ca atoms as a reference, is calculated to be 2.79 eV/Ca (see Table 1), largely exceeding the cohesive energy of 1.88 eV for bulk Ca. This indicates Ca atoms in this assembly cannot form metal clusters, similar to the case of CaCOF-1. Compared to CaCOF-1, the Ca binding energy is only slightly decreased by 0.04 eV/Ca. Interestingly, Li binding energy in LiCOF-d is also 0.06 eV/Li smaller than that in LiCOF-1, indicating the diphenylethyne unit is slightly less reactive than single phenyl groups during MCOF formation. Both CaCOF-d and LiCOF-d intercalation are stable 3D materials.

Figure 7.

Figure 7

Open in a new tab

(a) H2 storage on two-dimensional Ca-decorated COF-d layer. There are five H2 molecules per Ca adsorbed in 2D CaCOF-d. The hexagonal unit is blown up on the right panel for clarity. (b) Three-dimensional CaCOF-d structure in two views. There are four H2 molecules per Ca adsorbed in 3D CaCOF-d. For clarity H2 molecules are omitted in the left panel.

Three-dimensional stacking in CaCOF-d does not reduce much H2 storage capacity in low-dimensioned clusters and overlayers. In CaCOF-d, a single Ca can adsorb four intact H2, all parallel to the plane of the COF-d layer, as shown in Fig. 7b, with the average binding energy ~ 0.15 eV/H2 in GGA form of exchange-correlation energy and 0.42 eV/H2 in LDA. We believe the exact binding energy lies between the two values, close to the averaged value at 0.29 eV/H2, based on an empirical guideline that the binding energy averaged from LDA and GGA values is close to the precise Quantum Monte Carlo result for H2 on B-doped fullerene11. A storage capacity of 4.94 wt% at 0 K is calculated. The distance between two adjacent COF-d layers is 5.0 Å upon H2 binding, enlarged by 0.6 Å compared to H2 free materials. The estimation of H2 density excludes the free volume storage of H2 in the pores of CaCOF-d, which has a diameter of ~ 20 Å. If all possible adsorption sites are considered, this porous 3D network satisfies the 2017 DOE target for H2 storage.

To unveil its practical advantage as a new class of ideal H storage material, Figure 8 shows the storage capacity of CaCOF-d as a function of pressure and temperature at near ambient conditions. The H2 capacity at a given temperature T and pressure P is evaluated via the relation f(occupation number of adsorbed H2 molecules) = kT∂Z/∂μ, where k is Boltzmann constant, Z is the grand partition function, and μ is the H2 chemical potential47,48. The grand partition function follows,

graphic file with name srep01882-m2.jpg

where εl is the adsorption energy per H2 molecule when the number of adsorbed molecules is l, and gl is the multiplicity (degeneracy) of the configuration for a given l. The chemical potential of H2, μ, is given by

graphic file with name srep01882-m3.jpg

where μ0(T) is standard chemical potential and P0 is standard state pressure. Here, μ and εl are negative values. The gravimetric density ρ of hydrogen storage is then obtained as,

graphic file with name srep01882-m4.jpg

Figure 8.

Figure 8

Open in a new tab

(a) H2 storage density of 3D CaCOF-d as a function of pressure at 77 K, 298 K, and 400 K. H2 binding energy around 0.29 eV is used. (b) Comparison of storage density in CaCOF-d (excluding free-volume storage) and bare COF-1 (data from experiments in Ref. 27) at 77 K.

We adopt ambient pressure μ values in the literature48. At 298 K and 20 bar, we obtain 4.87 wt% storage capacity (excluding H2 storage in the free volume), which is substantially large than experimental values for un-doped COF-1 (0.1 wt%)32. The storage capacity is higher if H2 adsorption onto B3O3 rings and in free volume is considered. Most of the stored H2 (3.2 wt%) will be released at 298 K and 1 bar (by solely decreasing the pressure), or at 400 K and 20 bar (by solely increasing the temperature), where about 1.7 wt% H2 retains. This favors near-ambient condition operation for H2 storage and release, with impressive accessible H2 uptake. Without Ca, the storage capacity is very low (see Fig. 8b), and near-ambient condition operation is not possible due to too small H2 binding energies (~ 40 meV/H2). The bulk MCOFs formed by Ca intercalation have advantages of being intrinsically three-dimensional, thermodynamically stable, and with well-ordered and adjustable pores, in contrast to previously proposed Ca32C60 and other Ca-doped nanomaterials22,23,24.

Discussion

The 3D network based on Ca-decorated carbyne was recently proposed as stable H2 storage media, where Ca binding energy ranges from 1.4 to 2.0 eV/Ca depending on Ca-Ca separation18. The ECa's are close to bulk Ca cohesive energy. However, the base material, the 3D carbyne network has yet to be synthesized in experiment. Its existence and room temperature stability (in particular chemical reactivity) remain questionable. On the other hand, crystalline 3D networks formed by stacking COFs via van der Waals forces have been explicitly demonstrated25; we expect Ca-decorated COF-1 to be readily fabricated in experiment, by Ca intercalation between weakly bound COF-1 layers at high temperatures (600 K) in Ar gases. Based on existing materials (graphite and metals), a similar approach has resulted the discovery of CaC6, a superconducting material with highest transition temperature of 11.5 K among metal-intercalated carbon materials49. Hydrogen storage on metal intercalated graphite was also explored recently50,51,52,53. Once formed, the new Ca-intercalated 3D COF material is thermodynamically stable (characterized by large Ca binding energy), and would be quite inert, except for Ca active sites. The metal sites in MCOF might be sensitive to water vapor contamination, as is the case for all other H2 storage materials with active sites; in real applications water concentration must be carefully controlled during H2 charging and discharging processes. Overall, these novel properties would favor MCOF synthesis and H2 storage applications.

In conclusion, we have designed a new class of 3D networks based on first-principles: metal-intercalated COFs, which have well-defined crystal structures and are chemically active for molecular adsorption. Metal atoms prefer to bind to aromatic organic units rather than B3O3 connecting rings in MCOFs. Li, Na, K and Ca could be intercalated into COF-1 and COF-d layers and form thermodynamically stable materials, with metal binding stronger in MCOF than in bulk metals. For H2 storage, Ca is the most suitable metal to be intercalated in the stacking structures. Important parameters, such as Ca-binding and separation, and trends on 5,6-membered rings and larger PAHs are identified. The covalent linker –B(OH)2 preserves nicely H2 storage properties, while –COOH linkers adjust Ca separation and in turn H2 store capacity. Systematic optimization of organic building units and linker groups identifies CaCOF-d as most promising for near-ambient H2 storage. The porous 3D material, CaCOF-d, reaches ~ 5 wt% H2 storage capacity under near-ambient conditions without metal clustering problems. Routes for experimental synthesis are discussed. We envisage that this new class of MCOF porous materials, with a high density of active metal sites, would prove useful in a variety of applications including gas adsorption, sensors, and air cleaning.

Methods

Our first-principles calculations were performed within the framework of density functional theory (DFT) using Vienna ab initio simulation package (VASP)54. The ultrasoft pseudopotentials55 and general gradient approximation (GGA) in Perdew-Wang form for exchange-correlation energy56 are used. For the cases of hydrogen adsorption in MCOFs, we also use local density approximation (LDA) for comparison57. Although GGA tends to underestimate H2 binding energies and LDA overestimates them, the trends for comparing relative energy differences are reliable42,58. Recently developed van der Waals density functional (VDW-DF) by Langreth et al.39 and parameterized by Klimes et al.40 is also employed to check the energies in some critical cases. The binding energies and lengths reported throughout the manuscript are based on GGA exchange-correlation functional unless otherwise specified explicitly. Spin polarization is invoked whenever necessary. We found local magnetic moments are zero for CaCOF systems and very small (< 0.3μB) for Li-COFs and variations in total energy are negligible. The supercells usually contain a vacuum layer of > 10 Å and have cell dimensions of 15 Å × 15 Å × 15 Å and 30 Å × 15 Å × 20 Å for small and large finite systems, respectively. A plane wave cutoff of 400 eV and single Gamma for k-point sampling are employed for isolated systems. For periodic systems, a rhombic 15.20 Å × 15.20 Å unit cell for COF-1 and 27.24 Å × 27.24 Å for COF-d is used within the COF plane, and the interlayer distance is carefully optimized. K-point mesh of 1 × 1 × 5 is used for calculating metal and H2 binding properties of MCOFs. All atoms are allowed to relax until the forces on each atom have magnitudes less than 0.01 eV/Å.

Author Contributions

S.M. conceived the idea and carried out theoretical analysis. F.G. performed most of the numerical simulations. Z.D. carried out some numerical simulations and theoretical analysis. S.M. and F.G. analyzed the data and wrote the paper. All authors discussed the results and commented on the manuscript.

Acknowledgments

This work is partially funded by the innovation plan (energy project cluster) and hundred-talent program of CAS, NSFC (grant 11074287 and 11222431) and MOST (grant 2012CB921403).

References

  1. Schlapbach A. & Züttel A. Hydrogen-storage materials for mobile applications. Nature 414, 353–358 (2001). [DOI] [PubMed] [Google Scholar]
  2. Steele B. C. H. & Heinzel A. Materials for fuel-cell technologies. Nature 414, 345–352 (2001). [DOI] [PubMed] [Google Scholar]
  3. Park N. et al. Progress on first-principles-based materials design for hydrogen storage. Proc. Nat'l Acad. Sci. U.S.A. 109, 19893–19899 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Multi-year research, development and demonstration plan. http://www.eere.energy.gov/hydrogenandfuelcells/mypp/pdfs/storage.pdf, see Table 3.3.1. Last accessed October 1, 2012.
  5. Yuan F. et al. Complex ammine titanium(III) borohydrides as advanced solid hydrogen-storage materials with favorable dehydrogenation properties. Chem. Mater. 24, 3370–3379 (2012). [Google Scholar]
  6. Liu C. et al. Hydrogen storage in single-walled carbon nanotubes at room temperature. Science 286, 1127–1129 (1999). [DOI] [PubMed] [Google Scholar]
  7. Patchkovskii S. et al. Graphene nanostructures as tunable storage media for molecular hydrogen. Proc. Natl. Acad. Sci. U.S.A. 102, 10439–10444 (2005). [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Pupysheva O. V., Farajian A. A. & Yakobson B. I. Fullerene nanocage capacity for hydrogen storage. Nano Lett. 8, 767–774 (2008). [DOI] [PubMed] [Google Scholar]
  9. Teprovich J. A. et al. Synthesis and characterization of a lithium-doped fullerane (Li-x-C-60-H-y) for reversible hydrogen storage. Nano Lett. 12, 582–589 (2012). [DOI] [PubMed] [Google Scholar]
  10. Yang S. J. et al. MOF-derived hierarchically porous carbon with exceptional porosity and hydrogen storage capacity. Chem. Mater. 24, 464–470 (2012). [Google Scholar]
  11. Kim Y. H., Zhao Y., Williamson A., Heben M. J. & Zhang S. B. Nondissociative adsorption of H2 molecules in light-element-doped fullerenes. Phys. Rev. Lett. 96, 016102 (2006). [DOI] [PubMed] [Google Scholar]
  12. Meng S., Kaxiras E. & Zhang Z. Y. Metal-diboride nanotubes as high-capacity hydrogen storage media. Nano Lett. 7, 663–667 (2007). [DOI] [PubMed] [Google Scholar]
  13. Lee H., Ihm J., Cohen M. L. & Louie S. G. Calcium-decorated carbon nanotubes for high-capacity hydrogen storage: First-principles calculations. Phys. Rev. B 80, 115412 (2009). [Google Scholar]
  14. Yildirim T. & Ciraci S. Titanium-decorated carbon nanotubes as a potential high-capacity hydrogen storage medium. Phys. Rev. Lett. 94, 175501 (2005). [DOI] [PubMed] [Google Scholar]
  15. Zhao Y. F., Kim Y. H., Dillon A. C., Heben M. J. & Zhang S. B. Hydrogen storage in novel organometallic buckyballs. Phys. Rev. Lett. 94, 155504 (2005). [DOI] [PubMed] [Google Scholar]
  16. Chandrakumar K. R. S., Srinivasu K. & Ghosh S. K. Nanoscale curvature-induced hydrogen adsorption in alkali metal doped carbon nanomaterials. J. Phys. Chem. C 112, 15670–15679 (2008). [Google Scholar]
  17. Chandrakumar K. R. S. & Ghosh S. K. Alkali-metal-induced enhancement of hydrogen adsorption in C60 fullerene: an ab initio study. Nano Lett. 8, 13–19 (2008). [DOI] [PubMed] [Google Scholar]
  18. Sorokin P. B., Lee H., Antipina L. Y., Singh A. K. & Yakobson B. I. Calcium-decorated carbyne networks as hydrogen storage media. Nano Lett. 11, 2660–2665 (2011). [DOI] [PubMed] [Google Scholar]
  19. Srinivasu K. & Ghosh S. K. Graphyne and graphdiyne: promising materials for nanoelec-tronics and energy storage applications. J. Phys. Chem. C 116, 5951–5956 (2012). [Google Scholar]
  20. Sun Q., Jena P., Wang Q. & Marquez M. First-principles study of hydrogen storage on Li12C60. J. Am. Chem. Soc. 128, 9741–9745 (2006). [DOI] [PubMed] [Google Scholar]
  21. Sun Q., Wang Q., Jena P. & Kawazoe Y. Clustering of Ti on a C60 surface and its effect on hydrogen storage. J. Am. Chem. Soc. 127, 14582–14583 (2005). [DOI] [PubMed] [Google Scholar]
  22. Yoon M. et al. Calcium as the superior coating metal in functionalization of carbon fullerenes for high-capacity hydrogen storage. Phys. Rev. Lett. 100, 206806 (2008). [DOI] [PubMed] [Google Scholar]
  23. Yang X., Zhang R. Q. & Ni J. Stable calcium adsorbates on carbon nanostructures: Applications for high-capacity hydrogen storage. Phys. Rev. B 79, 075431 (2009). [Google Scholar]
  24. Lee H., Ihm J., Cohen M. L. & Louie S. G. Calcium-decorated graphene-based nanostructures for hydrogen storage. Nano Lett. 10, 793–798 (2010). [DOI] [PubMed] [Google Scholar]
  25. Cote A. P. et al. Porous, crystalline, covalent organic frameworks. Science 310, 1166–1170 (2005). [DOI] [PubMed] [Google Scholar]
  26. El-Kaderi H. M. et al. Designed synthesis of 3D covalent organic frameworks. Science 316, 268–272 (2007). [DOI] [PubMed] [Google Scholar]
  27. Furukawa H. & Yaghi O. M. Storage of hydrogen, methane, and carbon dioxide in highly porous covalent organic frameworks for clean energy applications. J. Am. Chem. Soc. 131, 8875–8883 (2009). [DOI] [PubMed] [Google Scholar]
  28. Blunt M. O. et al. Guest-induced growth of a surface-based supramolecular bilayer. Nature Chem. 3, 74–78 (2011). [DOI] [PubMed] [Google Scholar]
  29. Han S. S., Furukawa H., Yaghi O. M. & Goddard W. A. Covalent organic frameworks as exceptional hydrogen storage materials. J. Am. Chem. Soc. 130, 11580–11581 (2008). [DOI] [PubMed] [Google Scholar]
  30. Klontzas E., Tylianakis E. & Froudakis G. E. Hydrogen storage in 3D covalent organic frameworks. A multiscale theoretical investigation. J. Phys. Chem. C 112, 9095–9098 (2008). [Google Scholar]
  31. Klontzas E., Tylianakis E. & Froudakis G. E. Designing 3D COFs with enhanced hydrogen storage capacity. Nano Lett. 10, 452–454 (2010). [DOI] [PubMed] [Google Scholar]
  32. Li Y. W. & Yang R. T. Hydrogen storage in metal-organic and covalent-organic frameworks by spillover. AIChE J. 54, 269–279 (2008). [Google Scholar]
  33. Choi Y. J., Lee J. W., Choi J. H. & Kanga J. K. Ideal metal-decorated three dimensional covalent organic frameworks for reversible hydrogen storage. Appl. Phys. Lett. 92, 173102 (2008). [Google Scholar]
  34. Lan J. H., Cao D. P. & Wang W. C. Li-doped and nondoped covalent organic borosilicate framework for hydrogen storage. J. Phys. Chem. C 114, 3108–3114 (2010). [Google Scholar]
  35. Wu M. M., Sun Q., Jena P. & Kawazoe Y. First-principles study of hydrogen adsorption in metal-doped COF-10. J. Chem. Phys. 133, 154706 (2010). [DOI] [PubMed] [Google Scholar]
  36. Zou X., Zhou G., Duan W. H., Choi K. & Ihm J. A chemical modification strategy for hydrogen storage in covalent organic frameworks. J. Phys. Chem. C 114, 13402–13407 (2010). [Google Scholar]
  37. Sun Y. Y., Lee K., Kim Y.-H. & Zhang S. B. Ab initio design of Ca-decorated organic frameworks for high capacity molecular hydrogen storage with enhanced binding. Appl. Phys. Lett. 95, 033109 (2009). [Google Scholar]
  38. Kalidindi S. B. et al. Metal@COFs: Covalent organic frameworks as templates for Pd nanoparticles and hydrogen storage properties of Pd@COF-102 hybrid material. Chem. Eur. J. 18, 10848–10856 (2012). [DOI] [PubMed] [Google Scholar]
  39. Dion M., Rydberg H., Schroder E., Langreth D. C. & Lundqvist B. I. Van der Waals density functional for general geometries. Phys. Rev. Lett. 92, 246401 (2004). [DOI] [PubMed] [Google Scholar]
  40. Klimes J., Bowler D. R. & Michaelides A. Van der Waals density functionals applied to solids. Phys. Rev. B 83, 195131 (2011). [Google Scholar]
  41. Sillar K., Hofmann A. & Sauer J. Ab initio study of hydrogen adsorption in MOF-5. J. Am. Chem. Soc. 131, 4143–4150 (2009). [DOI] [PubMed] [Google Scholar]
  42. Sun Y. Y. et al. Accuracy of density functional theory methods for weakly bonded systems: The case of dihydrogen binding on metal centers. Phys. Rev. B 82, 073401 (2010). [Google Scholar]
  43. Purwanto W., Krakauer H., Virgus Y. & Zhang S. W. Assessing weak hydrogen binding on Ca+ centers: An accurate many-body study with large basis sets. J. Chem. Phys. 135, 164105 (2011). [DOI] [PubMed] [Google Scholar]
  44. Srinivasu K., Chandrakumar K. R. & Ghosh S. K. Computational investigation of hydrogen adsorption by alkali-metal-doped organic molecules: role of aromaticity. Chem Phys Chem 10, 427–435 (2009). [DOI] [PubMed] [Google Scholar]
  45. Zou X. L. et al. Hydrogen storage in Ca-decorated, B-substituted metal organic framework. Int. J. Hydrogen Energy 35, 198–203 (2010). [Google Scholar]
  46. Srinivasu K. & Ghosh S. K. Tuning the metal binding energy and hydrogen storage in alkali metal decorated MOF-5 through boron doping: A theoretical investigation. J. Phys. Chem. C 115, 16984–16891 (2011). [Google Scholar]
  47. Lee H., Choi W. I. & Ihm J. Combinatorial search for optimal hydrogen-storage nanomaterials based on polymers. Phys. Rev. Lett. 97, 056104 (2006). [DOI] [PubMed] [Google Scholar]
  48. Lide D. R. Handbook of Chemistry and Physics, 285–286, CRC Press, New York, (2001). [Google Scholar]
  49. Emery N. et al. Superconductivity of bulk CaC6. Phys. Rev. Lett. 95, 087003 (2005). [DOI] [PubMed] [Google Scholar]
  50. Kim B. J. & Park S. J. Optimization of the pore structure of nickel/graphite hybrid materials for hydrogen storage. Int. J. Hydrogen Energy 36, 648–653 (2011). [Google Scholar]
  51. Miyaoka H., Ishida W., Ichikawa T. & Kojima Y. Synthesis and characterization of lithiumcarbon compounds for hydrogen storage. J. Alloys Compounds 509, 719–723 (2011). [Google Scholar]
  52. Gunawano R. The soft hydrogen storage on transition metal graphite intercalates: a density functional theory calculations. Diffusion and Defect Data Part B 170, 379 (2011). [Google Scholar]
  53. Sigal A., Rojas M. I. & Leiva E. P. M. Is Hydrogen storage possible in metal-doped graphite 2D systems in conditions found on Earth? Phys. Rev. Lett. 107, 158701 (2011). [DOI] [PubMed] [Google Scholar]
  54. Kresse G. & Hafner J. Ab. initio molecular dynamics for liquid metals. Phys. Rev. B 47, 558–561 (1993). [DOI] [PubMed] [Google Scholar]
  55. Vanderbilt D. Soft self-consistent pseudopotentials in a generalized eigenvalue formalism. Phys. Rev. B 41, 7892–7895 (1999). [DOI] [PubMed] [Google Scholar]
  56. Wang Y. & Perdew J. P. Correlation hole of the spin-polarized electron gas, with exact small-wave-vector and high-density scaling. Phys. Rev. B 44, 13298–13307 (1991). [DOI] [PubMed] [Google Scholar]
  57. Ceperley D. M. & Alder B. J. Ground state of the electron gas by a stochastic method. Phys. Rev. Lett. 45, 566–569 (1980). [Google Scholar]
  58. Srepusharawoot P. et al. Ab Initio study of molecular hydrogen adsorption in covalent organic framework-1. J. Phys. Chem. C 113, 8498–8504 (2009). [Google Scholar]

Articles from Scientific Reports are provided here courtesy of Nature Publishing Group

RESOURCES