Ti4- and Ni4-Doped Defective Graphene Nanoplatelets as Efficient

Article pubs.acs.org/JPCC

Ti4- and Ni4‑Doped Defective Graphene Nanoplatelets as Efficient Materials for Hydrogen Storage C. M. Ramos-Castillo,*,† J. U. Reveles,‡,§ M. E. Cifuentes-Quintal,† R. R. Zope,§ and R. de Coss† †

Department of Applied Physics, Cinvestav-Mérida, A.P. 73 Cordemex 97310, Mérida, Yucatán, México Department of Physics, Virginia Commonwealth University, Richmond, Virginia 23284, United States § Department of Physics, University of Texas at El Paso, El Paso, Texas 79958, United States ‡

S Supporting Information *

ABSTRACT: We report a detailed theoretical investigation of the structural and electronic properties of titanium- and nickel-doped defective graphene nanoplatelets, which are shown to be efficient materials for hydrogen storage. We found that H2 bond cleavage is favored by Ti4-doped defective graphene nanoplatelets because of the strong interaction between the hydrogen 1s and titanium 3d levels that leads to the formation of metal hydrides, while H2 adsorption on Ni4-doped defective graphene favors the formation of Kubas complexes as hydrogen 1s levels only interact with the nickel 4s levels. A comparison between adsorption energies, number of H2 adsorbed molecules, and hydrogen gravimetric content shows that Ti4-doped graphene has a better performance for hydrogen storage with a notably high hydrogen gravimetric content of 3.4 wt %; than Ni4-doped graphene with a 10-fold lower gravimetric content of only 0.30 wt %. This observation can be explained by three factors: Ti is a lighter transition metal, it absorbs a larger amount H2 per metallic atom, and it presents a planar geometry that increases the coverage of the graphene layer and makes possible that all atoms in the cluster participate in the H2 adsorption. Our results support the hypothesis that a controlled introduction of defects in graphene followed by the anchoring of small metallic clusters is a feasible way to enhance the hydrogen gravimetric content of graphene nanoplatelets and to fine-tune hydrogen absorption energies to achieve a reversible operation at ambient temperature and moderates pressures, addressing one of the main challenges of a sustainable hydrogen-based economy.

1. INTRODUCTION The design of hydrogen storage materials with a gravimetric density of at least 7.5 w%, a minimum volumetric density of 0.07 kg H2/L, and a reversible operation at ambient temperature and moderate pressures is one of the main challenges that must be addressed for a sustainable hydrogenbased economy.1−3 Due to their lightweight and high surface area, carbon based materials are considered good candidates to achieve a substantial hydrogen content. However, hydrogen storage in pure carbon materials is restricted by a very low adsorption energy (physisorption) of less than 0.1 eV per H2, limiting the hydrogen load at operational temperatures.4,5 Thermodynamic estimations indicate that the adsorption energies that would lead to an efficient cyclic adsorption/ desorption process at room temperature and moderate pressures are in the range of 0.2 to 0.6 eV per hydrogen molecule.6−9 Previous experimental and theoretical studies have shown that hydrogen storage on activated carbon and graphene samples can be dramatically enhanced by functionalization with metal nanoparticles.10,11 Further, doping the porous graphitic materials with metallic atoms is viewed as a promising strategy to enhance the hydrogen uptake,12,13 as these metals have the effect of increasing the binding energy of molecular hydrogen © 2016 American Chemical Society

to the pore walls. In addition, the deposited metallic atoms can also bind several hydrogen molecules12 increasing in this way the hydrogen capacity of the material. The binding of molecular hydrogen to transition metals has been explained using the Kubas model as a donation of electronic charge from the H2 molecule to the unfilled d orbitals of transition metals like Pd, followed by a back-donation from the metal to the antibonding orbital of H2.14,15 However, there are some difficulties with the metal doping of graphitic materials: first, the metal atoms tend to aggregate to form clusters when deposited on the graphene surface,16−18 and second, often the desorption of metal− hydrogen complexes competes with the H2 desorption.16−18 Both problems could however be addressed by increasing the binding energy of the metal atoms or small metal clusters to the supporting carbon substrate, and this can indeed be achieved by anchoring the metal atoms and clusters to defects in the carbon networks of the graphitic pore walls.19 For instance, firstprinciples calculations have found that defects in graphene (such as mono- and divacancies) notably increase the adsorption energy of metal atoms and small metal clusters.20−22 Received: December 29, 2015 Published: February 22, 2016 5001

DOI: 10.1021/acs.jpcc.5b12711 J. Phys. Chem. C 2016, 120, 5001−5009

Article

The Journal of Physical Chemistry C

strength of the interaction of H2 with the metal cluster up to achieving saturation. We find that Ni4 reaches saturation after the adsorption of only three H2 molecules, meanwhile Ti4 saturates after the adsorption of six H2 molecules. The interaction between Ni4 and H2 is dominated by the formation of Kubas-type complexes similarly to the one we found for the graphene supported Pd4 cluster in our previous report,23 in contrast, Ti 4 exhibits both molecular and dissociative adsorption channels with an estimated gravimetric content up to 3.4 wt %. These results shows that Ti-decorated graphene systems are promising candidates for the designing of materials with the capacity to store H2 within the ideal range for efficient cyclic adsorption/desorption at room temperature and moderate pressures.

A question of vital importance and which is rarely mentioned is the effect of the nature of the transition metal on the hydrogen gravimetric content. In a recent work we have showed that a Pd 4 cluster supported on a graphene monovacancy is able to adsorb up to four H2 with moderate binding energies,23 such that the ratio between the number of adsorbed hydrogen molecules and Pd atoms on the cluster’s surface is 1. This result has been further corroborated by Lopez and co-workers for the case of Pd6 supported on a graphene monovacancy.24 Although these results are encouraging, our estimations, based on first-principles calculations, determined a maximum gravimetric content for the Pd4/graphene system of only 0.65 wt % which is in reasonable agreement with experiments,9,11 but it is still far from the ideal gravimetric content of 7.5 wt %. Experimental and theoretical studies have suggested that the main limitation of Pdx-graphene systems for hydrogen storage is not the H2 adsorption energy per se, but the metal’s atomic weight and the number of hydrogen molecules that it can adsorb up to saturation. In this sense, it is reasonable to think about using lighter transition metal clusters as an alternative to solve the problem. Nickel is a promising candidate; it weighs about half as much as Pd and is well-known for its catalytic properties. Unfortunately, previous experiments on graphene decorated with monodispersed Ni nanoparticles showed a hydrogen content of only 0.5 wt %25 at moderate pressures, a value that is similar to the one measured in Pd-decorated graphene. This result supports the idea that the effect of the metal dopant enhancing the hydrogen adsorbion would be largest for maximum dispersion, that is, when single metal atoms or very small clusters are present.15 Another potential candidate is Ti. Recently, an interesting result has been reported by Mashoff and co-workers,26 where the formation of titanium-islands on graphene as a function of defect density was investigated. When depositing titanium on pristine graphene, titanium forms islands with an average diameter of about 10 nm and an average height of a few atomic layers. In contrast, if defects are introduced in the graphene by ion bombardment, the mobility of the deposited titanium atoms is reduced, and the average diameter of the islands decreases to only 5 nm. Furthermore, these islands present a single layer monatomic height that increases the titanium surface available per unit of graphene area and present a gravimetric density between 0.75 and 2.5 wt % depending of surface coverage. These results show how the lattice engineering in graphene surfaces is a promising route to enhance their catalytic properties by maximizing the dispersion of metal dopants, and hence the surface coverage. Based on these reports we consider that studying the adsorption of hydrogen on small graphene supported metal clusters is of great relevance toward the design of novel graphene based materials with tailored properties. Motivated by the recent experiments of the H2 absorption on Ti and Ni graphene supported nanoparticles25,26 and by the challenge to develop hydrogen storage materials for practical applications, the aim of this work is then to provide a detailed description of the interaction of H2 with atomic and Ti4- and Ni4-doped defective graphene nanoplatelets. We carried out first-principles calculations to investigate the binding of Ti4 and Ni4 species to a graphene monovacancy, showing that both Ti4 and Ni4 clusters strongly bind to graphene defects. Our study of the variation of the H2 binding energy to the metal clusters shows that the successive adsorption of H2 diminishes the

2. THEORETICAL METHODS We performed a density functional theory (DFT)27 study based on first-principles calculations within the generalized gradient approximation (GGA) using the exchange-correlation functional proposed by Perdew, Burke, and Ernzerhof.28 To speed up the calculations we used norm-conserving pseudopotentials29 as implemented in the SIESTA code.30 The Gamma point is used for the Brillouin zone sampling and we employed a double-ζ basis function set with polarized orbitals,30 semicore states in the pseudopotential and basis set were included for titanium atoms,30 and a mesh cutt-off energy of 300 Ry for the grid integration in real space. The convergence criterion for energy was chosen to be 10−4 eV. The structural parameters were fully optimized within a force convergence criterion of 0.01 eV/Å. All calculations were performed using spin polarization. To speed up the self-consistency convergence, a polynomial broadening of the energy levels was performed using the method of Methfessel and Paxton.31 To investigate the van der Waals dispersion effects on the hydrogen adsorption energies additional calculations were undertaken by the DFT-D method.32 A negligible contribution of less than 0.04 eV for Ti and 0.07 for Ni was found in the calculated binding energies, and the dispersion effects were not included in most of our reported results. Density of states was calculated using a 12 × 12 × 1 k-point sampling. 3. RESULTS AND DISCUSSION 3.1. Adsorption of Single Ti and Ni Atoms on Defective Graphene. The supporting defective graphene nanoplatelets with a single monovacancy was simulated as a periodically repeated unit cell consisting of 199 carbon atoms in the xy plane (Figure 1a). This supercell size has been shown to be sufficiently large to minimize the spurious interaction between monovacancies, metal clusters, and adsorbed H2 molecules placed in different cells.23 Furthermore, the vacuum

Figure 1. (a) Top and (b) side views of the atomic structure for relaxed graphene monovacancy. Yellow circles represent carbon atoms. 5002


Article


and Ni atoms on the graphene monovacancy is 7.80 and 7.12 eV respectively, in good agreement with previous calculations of other transition metal atoms on graphene vacancies.20 This large binding energy suggests that the monovacancy is able to prevent the mobility and thus migration of the transition metal atoms on the graphene surface. The calculated values for the structural parameters h and d, and binding energies ΔEb in Figure 2 are summarized in Table 1. Next, we studied the

separation between the structures in adjacent unit cells was taken to be at least 20 Å in order to avoid interactions between the structure and its images, in the neighboring cells, in the perpendicular z direction. Figure 1b shows the atomic structure of the optimized graphene monovacancy. We did not observe a significant change in the structure beyond the defect site after one carbon atom was removed from the graphene sheet; however, in the vicinity of the vacancy structural distortion that breaks the symmetry occurred due to local bonding rearrangement. This reconstruction is known to generate the 5−9 isomer of the graphene surface,33 which presents an elongated bridge causing the formation of two rings, with 5 and 9 atoms, respectively. Further, no significant out of the plane atomic displacements were observed in the relaxed structure. Our calculated vacancy formation energy of 7.6 eV is in good agreement with the experimental value34 of 7.0 ± 0.5 eV, and with previous calculations which give values from 7.7 to 7.8 eV.35 We obtained a magnetic moment of 1.26 μB/cell localized in the vacancy region. Earlier studies35 have shown that the spin moment varies from 1.1 to 1.3 μB/cell depending on the size of the supercell, and that converges toward 1.0 μB as the supercell size increases. These results highlight the importance of using large supercells to correctly describe the ground state of the graphene monovacancies. For the large supercell size used in our calculations, our value of 1.26 μB is in reasonable agreement with the magnetic moment reported by Palacios and Ynduraiń for similar supercell sizes.35 A representative scheme of the adsorption of single Ti and Ni atoms on graphene monovacancy is shown in Figure 2. We

Table 1. Calculated Values of Structural Parameters (h and d), Binding Energies (ΔEb), and Magnetic Moment (M) for Single Atoms Supported on Graphene Monovacancy atom

h (Å)

d (Å)

ΔEb (eV)

M (μB)

Ti/GM Ni/GM

1.77 1.47

1.93 1.81

−7.80 −7.12

0 0

hydrogen adsorption of H2 on single atoms supported on the graphene monovacancy. The calculated adsorption energies for H2 on supported Ti and Ni are only 0.24 and 0.18 eV, respectively, indicating a weak interaction. The relaxed structures for the adsorption of H2 on graphene supported single atoms supported, Ti/GM and Ni/GM, are shown in Figure 3. Therefore, we cannot expect that single supported

Figure 3. Relaxed structures for the adsorption of H2 on single atoms supported on the graphene monovacancy (a) Ti/GM and (b) Ni/GM.

atoms have a substantial contribution for hydrogen storage. Analogous results were reported in previous studies for the case of graphene supported Pd atom.21,23 3.2. Ti4 and Ni4 Clusters Supported on Defective Graphene. Next, we studied the bonding of Ti4 and Ni4 species to the graphene monovacancy. Figure 4 presents the optimized structures for gas phase and supported Ti4 and Ni4

Figure 2. Side and top views of the relaxed structure of single atoms supported on a graphene monovacancy (GM). Yellow and purple circles represent carbon and Ti or Ni atoms, respectively. The calculated values for h and dM‑C are given in Table I.

found that both Ti and Ni form covalent bonds with the undercoordinated C atoms at the vacancy by breaking a weak C−C bond (bond length of 2.10 Å) of one pentagon in the reconstructed vacancy. As Ti and Ni atomic radii are larger than that of the carbon atom the Ti and Ni atoms displaced outward from the graphene surface, with an elevation of 1.77 Å for Ti and 1.47 for Ni, and the three C atoms around the vacancy also move out of the graphene plane about 0.6 Å. The interatomic distance between the Ti atom and three surrounding C atoms at the vacancy is 1.93 Å, while, the distance for Ni is 1.81 Å. The binding energy (ΔEb) of a Ti and Ni atom on the graphene monovacancy was calculated using eq 1: ΔE b(M n/GM) = E(M n/GM) − E(GM) − E(M n)

(1)

where M = Ti and Ni, n = 1 for the case of a single metal atom, GM represents the supporting graphene monovacancy, and E indicates total energies. The calculated binding energy for Ti

Figure 4. Top and side views of the relaxed structures for the adsorption of Ti4 and Ni4 clusters on the GM. Yellow, gray, and blue circles represent carbon, titanium, and nickel atoms, respectively. 5003


Article

The Journal of Physical Chemistry C clusters. We found both gas phase Ti4 and Ni4 present a tetrahedral geometry with a large magnetic moment of 3.87 and 3.90 μB respectively. Also shown in Figure 4 are the relaxed structures for the Ti4 and Ni4 clusters anchored to the monovacancy. The lowest energy configuration of Ni4 attached to the vacancy is a tetrahedral structure, similar to that of a free gas phase Ni4. The tetrahedron lies on a triangular face with one Ni atom at the center of the vacancy. The calculated binding energy of Ni4 at the vacancy is 1.85 eV/atom. It is interesting to note that while free Ni4 has a magnetic moment of 3.90 μB,22 upon chemisorption of the Ni4 on the graphene vacancies the magnetic moment is reduced to only of 1 μB due to the strong interaction of the cluster with the graphene monovacancy. On the other hand, the Ti4 cluster shows a more interesting behavior upon adsorption on the graphene monovacancy. Its geometry changes leading to a planar configuration and results in quenching of the total magnetic moment, with larger binding energy of 2.47 eV/atom. Determining the strength of the interaction between clusters and the graphene support is a very important factor that determines the mobility of the clusters species on the graphene surface and the possible desorption of the cluster species. Given the large values of binding energies we conclude that both Ti4 and Ni4 should remain attached to the vacancy and will neither diffuse on the surface nor be desorbed. This will avoid the possibility of clusters encountering and binding to other atoms or clusters to form larger aggregates, which limits the system’s hydrogen storage capacity.22 This hypothesis was recently confirmed by Mashoff and co-workers,26 where a reduction of the average diameter of the deposited titanium nanoislands from 10 up to 5 nm was found after defects were introduced in the graphene layer. Another important aspect of the design of materials for H2 storage concerns to the stability of the chemical bonding of the absorbed clusters. This can be quantified by the cluster cohesive energies, that is, the energy required to fragment a species into its atomic units. We calculate the cohesive energies for both gas-phase and supported clusters using eqs 2 and 3 respectively: ΔEc = [E(M n) − nE(M)]/n

(2)

ΔEc/GM = [E(M n/GM) − E(GM) − nE(M)]/n

(3)

Table 2. Calculated Values of Binding Energy (ΔEb), Cohesive Energy (ΔEc), and Magnetic Moment (M) for Gas Phase (gas) and Graphene Supported (GM) Ti4 and Ni4 Clustersa ΔEc(eV/atom)

M (μB)

atom

ΔEb (eV/ atom)

gas

on GM

gas

on GM

geometry

Ti4 Ni4

2.47 1.47

−2.76 −1.31

−5.64 −3.56

3.87 3.90

0 1

planar tetrahedral

a

Energy values are given in eV/atom.

Figure 5. Top and side views of the relaxed structures for the H2 adsorption on supported clusters (a) Ti4/GM and (b) Ni4/GM.

ΔEad(x H 2) = [E(x H 2 /M4 /GM) − E(M4 /GM) − xE(H 2)]/x

(4)

Here M could be Ti or Ni, GM represents the graphene monovacancy, x is the number of adsorbed H2 molecules, and E(xH2/M4/GM) is the total energy of the system formed by the x hydrogen molecules adsorbed on M4/GM. Unlike the case of weak hydrogen adsorption observed on single supported atoms, the H2 adsorption on supported clusters is far stronger with calculated energies ranging from 1.0 and 1.6 eV for Ni4 and Ti4, respectively. It is interesting to note that the difference in binding energies between Ti4 and Ni4 is reflected in the way in which the H2 adsorbs on the cluster; for Ti4 the complete dissociation of H2 is favored, while Ni4 prefers the formation of Kubas-type complexes, and this result is similar to the one we recently found for supported Pd4 cluster in which case we found a H2 adsorption energy of 1.2 eV.23 This high absorption energy for Ti4 and Ni4 is already in the range of chemisorption above of the optimal energetic window for hydrogen storage of 0.2 to 0.6 eV per H2 molecule. The calculated bond length for the adsorbed H2 ranges from 0.85 Å in the case of adsorption on Ni4, a bond length that is typical of Kubas-type complexes,14 to 2.5 Å for the case of H2 adsorbed and dissociated on Ti4, thus indicating that interaction between Ti4 cluster and the graphene monovacancy favors formation hydrides.22 The formation of hydrides in titanium clusters is very interesting as this is known to be the first step for the hydrogen spillover mechanism that only occurs with the complete dissociation of H2 molecules on catalyst’s surface.24

Here ΔEc is the cohesive energy per atom of the gas phase clusters36 and ΔEc/GM contains information about the cohesive energy of the clusters in the presence of the graphene support.37 The calculated ΔEc/GM were found to be significantly larger than ΔEc, indicating that the strong chemical interaction between clusters and the graphene monovacancy stabilizes the supported clusters with respect to its atomic gas phase components. Moreover, this stabilization is also in good agreement with the observed reduction of the spin magnetic moment as a consequence of the strong hybridization between the clusters and the monovacancy. The calculated cohesive energies are summarized in Table 2. 3.3. Hydrogen Adsorption Channels on Ti4- and Ni4Doped Defective Graphene. Following our study, we investigated the effect of anchoring the Ti4 and Ni4 clusters on its capacity to adsorb H2 molecules. Figure 5 shows the relaxed structures and adsorption energies for a single H2 molecule. The adsorption energy (ΔEad) of x hydrogen molecules on the M4−graphene system is given by eq 4: 5004


Article


Figure 6. Partial density of states for (a) (H2)3/Ti4/GM and (b) (H2)3/Ni4/GM.

To investigate the rational for the difference on the H2 absorption of Ti4/GM, i.e., complete dissociation, versus Ni4/ GM, where Kubas-type complexes are present, we calculated the partial density of states for (H2)3/Ti4/GM and (H2)3/Ni4/ GM (full density of states is given in the Supporting Information, Figure S1). As shown in Figure 6 the Ti 3d states are located below and above the Fermi level and exhibit a strong interaction with the hydrogen 1s levels which leads to the hydrogen dissociation upon absorption. On the other hand, the Ni 3d levels are located below the Fermi level and the hydrogen 1s levels only interact with the Ni 4s levels, thus explaining the formation of the Kubas-type complexes. Despite the fact that the large adsorption energies found so far can be a daunting result for the aim of hydrogen uptake; we considered that there are additional questions that must be addressed in the theoretical modeling of nanomaterials for hydrogen storage, such as What is the maximum number of H2 that can bind to a given cluster? And how is the H2 absorption energy affected by the attaching of multiple hydrogen molecules? We expect that the number of adsorbed H2 increases with increasing the size of the cluster and that at the same time, the binding strengths decrease with the grade of H2 saturation of the active site. Furthermore, if the binding energy decreases with the number of H2 molecules, the question arises whether we can have a system that presents both a large H2 intake and a binding energy that lies in the ideal range for an efficient cyclic adsorption/desorption process at room temperature and moderate pressures. These questions have been investigated in our previous study for graphene supported Pdx clusters, and now we investigate the answer to both these questions in the case of graphene supported Ti4 and Ni4 clusters. 3.4. Hydrogen Saturation of Ti4- and Ni4-Doped Defective Graphene. Understanding the key factors that determine the adsorption of a single H2 on Pd clusters allows gaining insight of such interactions. However, for a more realistic modeling the role of the active sites (Ti4 and Ni4 clusters in our case), the hydrogen saturation, and the variation in the H2 binding energy should be addressed. Saturation of the catalyst plays a central role in the study of adsorption of molecular species. With this aim we analyzed the variation in the sequential H2 absorption energies ΔEx as defined by the eq 5.

ΔEx = E(x H 2 /M4 /GM) − E((x − 1)H 2 /M4 /GM) − E(H 2)

(5)

ΔEx differs from the definition of ΔEad given in eq 4 which represents an average adsorption energy and because ΔEx is generally compared with the thermodynamic estimations of adsorption energies.9,11,21,22 On the other hand, as we have noted in our previous work23 ΔEx can be a good descriptor for the grade of saturation, and based on the definition of eq 5, a value for ΔEx > 0 indicates that the adsorption of H2 is energetically unfavorable. We then analyzed the successive addition of H2 molecules to theTi4 graphene anchored cluster. Figure 7 presents the relaxed

Figure 7. Relaxed structures for the saturation of supported Ti4/GM. (a) Geometry after the adsorption of three H2 molecules, (b) after the adsorption of six H2 molecules, and (c) after the adsorption of eight H2 molecules.

structures for three, six, and eight adsorbed H2 molecules on Ti4. It is found that when the first three H2 molecules interact with the deposited Ti4 cluster most of them bind in dissociative form (see Figure 7a), although the interaction strength between Ti4 and the first three H2 slightly depends on the number of adsorbed molecules ΔEx with x = 1 to 3, ranges from 1.6 to 1.2 eV. The adsorption of another three molecules is favored by the formation of Kubas complexes as seen in Figure 7b, with ΔEx energies between 0.50 and 0.25 eV and H−H bond lengths of 0.8 Å. It is found that Ti4 achieves saturation with 6 adsorbed 5005


Article


molecules before the saturation. Figure 9b shows the average H2 adsorption energy at the saturation point. The average H2 adsorption energy ΔEad defined in eq 4 is a good descriptor of the interaction strength of the clusters with the adsorbed molecules, and this is the reason why it is directly compared with thermodynamics estimations for the ideal adsorption energy range.11,21,22,38 Our results predicts that both Ti4 and Pd4 exhibit similar ΔEad values with average adsorption energies 0.53 and 0.56 eV/H2, respectively, while Ni4 have slightly smaller value of 0.42 eV/H2. Importantly, all atomic clusters under study, i.e., Ti4, Pd4, and Ni4, are able to attach H2 molecules with an average adsorption energy between 0.2 and 0.6 eV/H2 which is in the optimal energy range for an efficient cyclic adsorption/desorption process at room temperature and moderate pressures. 3.5. Hydrogen Gravimetric Content. In previous sections we have showed that the interaction between the graphene monovacancy and the metal clusters under study lead to H2 adsorption energies within the ideal range for practical hydrogen storage; however, it is also important to analyze the hydrogen gravimetric content (wt %). For its estimation we have used the method recently proposed by Mashoff and coworkers,26 this method is convenient as it involves parameters that can be determined from the optimized geometries. These include the maximum number of loaded hydrogen molecules per metal atom in the cluster, the surface particle density, which can be evaluated from the cluster’s interatomic distances, the fraction of metal atoms in the cluster’s surface, and the exposed metal−surface to total graphene to surface ratio. All these quantities can be evaluated directly from our first-principle calculations. The detailed description of the estimation of the wt % for the Ti4-, Ni4-, and Pd4-doped defective graphene is given in the Supporting Information. Figure 10 summarizes the calculated hydrogen gravimetric content for the systems under study, and includes a comparison with experimental reports of grapheme-decorated samples. Palladium decorated graphene is by far the most studied system for hydrogen storage, in previous experimental works the measured gravimetric content at moderated pressures and room temperatures was estimated to be 0.30% by Contescu and co-workers9 and 0.75 wt % as reported by Vinayan and co-workers.11 It is important to mention that in Contescu’s experiments the graphene samples were decorated with well dispersed palladium atoms, in fact,

molecules, as after adding two more H2 molecules a weakly physisorption is observed with ΔEx energies of less that 0.2 eV, and in which the additional H2 molecules lay more than 2.5 Å away from the Ti4 cluster (Figure 7c). In contrast to the case of the supported Ti4, in the case of Ni4 anchored on the graphene monovacancy saturation occurs after adsorbing only three H2 molecules and with the forming of Kubas type complexes. The fourth H2 molecule does not directly bind to the Ni4 cluster and remained practically unaltered with respect to isolated H2 (Figure 8). This

Figure 8. Relaxed structures for the saturation of graphene supported Ni4/GM.

unattached H2 molecule situates more than 3.0 Å away from the supported cluster atoms. In general, the adsorption of multiple H2 does distort neither the planar nor the tetrahedral geometries of the Ti4 and Ni4 in a noticeable way. However, a strong influence was again found on the hydrogen−metal interaction caused by the hybridization between the carbon surface and the metal atoms. Figure 9a presents the calculated ΔEx as a function of the number of adsorbed H2 molecules for the supported Ti4 and Ni4 clusters. We have also included our previous results for the supported Pd4 cluster.23 Since Pd and Ni belong to the same group of the periodic table, it is interesting to compare the H2 saturation of Pd4 and Ni4 clusters to evaluate the effect of the atomic radius on the hydrogen adsorption. We can see in Figure 9a that the values of ΔEx for Ni4 are slightly smaller than those calculated for Pd4. This result can explain why Pd4 reaches saturation with four H2 molecules while Ni4 only can attach up to three molecules before saturation. On the other hand, the values of ΔEx for H2 on Ti4 vary from 1.6 eV for a single H2 up to 0.1 eV for the eighth adsorbed molecule. From the calculated values for ΔEx and the atomic structure seen in Figure 6b, we can conclude that Ti4 is able to attach up to 6

Figure 9. (a) Variation in the sequential H2 absorption energy for graphene anchored Ti4(H2)x/GM, Ni4(H2)x/GM, and Pd4(H2)x/GM clusters as a function of the number of H2 molecules. Values of ΔEx > 0 indicate that adsorption is energetically unfavorable. The values for Pd4(H2)x/GM were reported in ref 23. (b) Average adsorption energy of H2 molecules at the saturation for Ti4(H2)6/GM, Ni4(H2)3/GM, and Pd4(H2)4/GM. 5006


Article


4. CONCLUSIONS A detailed theoretical description of the atomic structure and energetics of the H2 adsorption on Ti4- and Ni4-doped defective graphene is presented. The anchoring of Ni4 clusters on the defective graphene surface does not change the atomic structure with respect to the gas phase state; however, it does reduce the total spin magnetic moment of the Ni cluster and the graphene-monovacancy as a result of a strong hybridization between the surface and the Ni atoms. In the case of Ti4 the influence of the vacancy on its structural and electronic properties is more pronounced. The anchoring quenches the total spin moment and distorts its tetrahedral geometry to adopt a planar configuration. Our calculations show large binding energies for supported clusters on the graphene monovacancy indicating that the anchoring avoids the migration of metal atoms and clusters on the graphene surface, a problem that diminishes the hydrogen storage capacity of the material. We found that doping the graphene with a single metal does not contribute significantly to hydrogen uptake. The configuration of the absorbed H2 molecule is found to depend on the metal on which it is absorbed. The H2 bond activation is favored by the Ni4 clusters as hydrogen 1s levels only interact with the Ni 4s levels, forming dihydrogen-type complexes with H−H bond lengths between 0.85−0.9 Å, while the adsorption on Ti4 exhibits a dissociative character, given that hydrogen 1s levels strongly interact with the Ti 3d levels, a factor that increases the H2 content and favors the formation of Kubas-type complexes of Ti4(H2)n/defective-graphene with n > 3. Our results show that the sequential binding energy ΔEx decreases as a function of number of adsorbed molecules, and it is in the range of 1 to 0.1 eV for Ni4 and 1.6−0.2 eV for Ti4; the graphene vacancy plays an important role in modulating the absorption energies. The analysis of the sequential energies and H2 averaged bond lengths suggests that Ni4 is able to covalently attach up to 3 molecules to form dihydrogen complexes with moderate binding energies within the ideal energy range for efficient cyclic adsorption/desorption at room temperature and moderate pressures. On the other hand, Ti4−doped graphene adsorbs up to six H2 molecules before reaching saturation also exhibiting moderate binding energies. The maximum theoretical gravimetric content of H2 for Ni4(H2)3/defective graphene has been estimated to be 0.30 wt %, while that for Ti4(H2)6/defective graphene was 3.4 wt %, which is significantly larger than the one predicted for Ni4 and Pd4 clusters. The better performance of Ti4 clusters can be related to the fact that Ti is a ligther transition metal, its absorbs a larger amount of H2 per metal atom and presents a planar geometry which makes that all cluster’s atoms participate in the adsorption and increases the graphene sheet coverage. Our results supports the hypothesis that a controlled introduction of defects in graphene together within the anchoring of small metal clusters is a feasible way to enhanced the hydrogen gravimetric content and opens the possibility of investigating other light metal clusters supported on graphene defects. Further, we contribute to the understanding of the molecular interactions between H2 and metal-doped carbon structures that have important implications for the design of promising carbon-based materials for hydrogen storage technologies. We hope that our study motivates future theoretical and experimental studies as we provide a descriptor

Figure 10. Comparison of calculated and experimental values for hydrogen gravimetric content of Ti-, Ni-, and Pd-doped graphene.

they reported that 18% of the adsorbed palladium was in the form of single atoms, and the rest in the form of PdH0.6 species, which could be explained by the adsorption of H2 on small Pd clusters. On the other hand, in Vinayan’s experiments11 the formation of Pd nanoparticles, which may indeed favor the H2 absorption, was reported. In this case the hydrogen gravimetric content was twice as large as in the experiments with welldispersed atoms. Our calculated wt % for the Pd4−doped graphene cluster of 0.64% is in good agreement with the experimental report. Furthermore, our results suggest that the gravimetric content is mainly determinate by the ratio between the surface cluster’s atoms and the maximum number of adsorbed H2 molecules, which is 1:1 for the case of Pd4(H2)4/ defective graphene. Actually, this ration is predicted to be independent of the metal cluster’s size, a conclusion supported by the recent work of López and co-workers of H2 absorption of Pd6 clusters on graphene monovacancies, were H2 saturation was reached for Pd6(H2)6, and thus exhibiting a 1:1, H2:Pd ratio.24 In the case of the Ni4-doped graphene our predicted value of 0.30 wt % is slightly smaller than the one reported by Gaboardi of 0.40 wt % for Ni nanoparticles on graphene.25 It is interesting that, despite the fact that Ni is a lighter element than Pd, both the experimental and theoretical values predicts similar values of wt %. Our calculations show that the rational of this low hydrogen gravimetric content is the lower ratio between the maximum number of adsorbed H2 to number of metal atoms, which is only 0.75:1 in the case of Ni4(H2)3/ defective-graphene, that is, smaller than the one found for the case of Pd. On the other hand, Ti4-doped graphene shows a most interesting behavior. Ti4 reaches saturation after adsorbing 6 hydrogen molecules in Ti4(H2)6/defective-graphene leading to a H2:Ti ratio of 1.5:1. Notably, we calculated a gravimetric content of 3.4 wt % for the graphene supported Ti4, which is larger than the 2.4 wt % value estimated by Mashoff and co-workers for Ti nanoislands assuming a 1:1 H2:Ti ratio and an almost complete Ti coverage of the graphene sheet with small metal islands.26 The considerably higher gravimetric content found for Ti4-doped graphene (3.4 wt %) compared to the ones for Ni4-doped graphene (0.30 wt %) and Pd4-doped graphene (0.64 wt %) can be explained by the fact that Ti is lighter than Ni and Pd, and it absorbs a larger amount H2 per atom and has also a planar geometry which makes that all cluster’s atoms participate in the adsorption and increases the graphene sheet coverage. 5007


Article


(12) Yildirim, T.; Ciraci, S. Titanium-Decorated Carbon Nanotubes as a Potential High-Capacity Hydrogen Storage Medium. Phys. Rev. Lett. 2005, 94, 175501−175504. (13) Lee, H.; Ihm, J.; Cohen, M. L.; Louie, S. G. Calcium-Decorated Graphene-Based Nanostructures for Hydrogen Storage. Nano Lett. 2010, 10, 793−800. (14) Kubas, G. J. Molecular Hydrogen Complexes: Coordination of Sigma Bond to Transition Metals. Acc. Chem. Res. 1988, 21, 120−128. (15) Power, P. Main-Group Elements as Transition metals. Nature 2010, 463, 171−177. (16) 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. 2005, 127, 14582−14584. (17) Krasnov, P. O.; Ding, F.; Singh, A. B.; Yakobson, B. I. Clustering of Sc on SWNT and Reduction of Hydrogen Uptake: Ab-Initio AllElectron Calculations. J. Phys. Chem. C 2007, 111, 17977−17980. (18) Chan, K. T.; Neaton, J. B.; Cohen, M. L. First-Principles Study of Metal Adatom Adsorption on Graphene. Phys. Rev. B: Condens. Matter Mater. Phys. 2008, 77, 235430−23541. (19) Kim, G.; Jhi, S. H.; Lim, S.; Park, N. Effect of Vacancy Defects in Graphene on Metal Anchoring and Hydrogen Adsorption. Appl. Phys. Lett. 2009, 94, 173102−173105. (20) Krasheninnikov, A. V.; Lehtinen, P. O.; Foster, A. S.; Pyykko, P.; Nieminen, R. M. Embedding Transition-Metal Atoms in Graphene: Structure, Bonding, and Magnetism. Phys. Rev. Lett. 2009, 102, 126807−126810. (21) López, M. J.; Cabria, I.; Fraile, S.; Alonso, J. A. Adsorption and Dissociation of Molecular Hydrogen on Palladium Clusters Supported on Graphene. J. Phys. Chem. C 2012, 116, 21179−21189. (22) López, M. J.; Cabria, I.; Alonso, J. A. Palladium Clusters Anchored on Graphene Vacancies and Their Effect On the Reversible Adsorption of Hydrogen. J. Phys. Chem. C 2014, 118, 5081−5090. (23) Ramos-Castillo, C. M.; Reveles, J. U.; Zope, R. R.; De Coss, R. Palladium Clusters Supported on Graphene Monovacancies for Hydrogen Storage. J. Phys. Chem. C 2015, 119, 8402−8409. (24) Granja, A.; Alonso, J. A.; Cabria, I.; López, M. J. Competition Between Molecular and Dissociative Adsorption of Hydrogen on Palladium Clusters Deposited on Defective graphene. RSC Adv. 2015, 5, 47945−47953. (25) Gaboardi, M.; Bliersbach, A.; Bertoni, G.; Aramini, M.; Vlahopoulou, G.; Pontiroli, D.; Mauron, P.; Magnani, G.; Salviati, G.; Zuttel, A.; Ricco, M. Decoration of Graphene with Nickel Nanoparticles: Study of the Interaction with Hydrogen. J. Mater. Chem. A 2014, 2, 1039−1046. (26) Mashoff, T.; Convertino, D.; Miseikis, V.; Coletti, C.; Piazza, V.; Tozzini, V.; Beltram, F.; Heun, S. Increasing the Active Surface of Titanium Islands on Graphene by Nitrogen Sputtering. Appl. Phys. Lett. 2015, 106, 083901−083905. (27) Kohn, W.; Hohenberg, P. Inhomogeneous Electron Gas. Phys. Rev. 1964, 136, 864−871. (28) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalizaed Gradient Aproximation Made Simple. Phys. Rev. Lett. 1996, 77, 3865−3868. (29) Troullier, N.; Martins, J. L. Efficient Pseudopotentials for PlaneWave Calculations. Phys. Rev. B: Condens. Matter Mater. Phys. 1991, 43, 1993−2006. (30) Soler, J. M.; Artacho, E.; Gale, J. D.; García, A.; Junquera, J.; Ordejón, P.; Sánchez-Portal, D. The SIESTA Method for Ab-Initio Order-N Simulation. J. Phys.: Condens. Matter 2002, 14, 2745−2779. (31) Methfessel, M.; Paxton, A. T. High-Precisión Sampling for Brillouin-Zone Integration in Metals. Phys. Rev. B: Condens. Matter Mater. Phys. 1989, 40, 3616−3621. (32) Grimme, S. Semiempirical GGA-Type Density Functional Constructed with a Long-Range Dispersion Correction. J. Comput. Chem. 2006, 27, 1787−1799. (33) Zhang, Y.-H.; Yue, L.-J.; Han, L.-F.; Chen, J.-L.; Fang, S.-M.; Jia, D.-Z.; Li, F. Tuning the Magnetic Behavior and Transport Property of Graphene by Introducing Dopant and Defect: A First Principle Study. Comput. Theor. Chem. 2011, 972, 63−67.

(or template) toward the computational discovery and design of new families of novel hydrogen storage materials that address the current challenges of a sustainable hydrogen-based economy.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpcc.5b12711. Details of the gravimetric density calculation and calculated total density of states for (H2)3/Ti4/GM and (H2)3/Ni4/GM (PDF)

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Phone: +52 999 9657177. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS Authors thank the Texas Advanced Computing Center (TACC) from the National Science Foundation (NSF; Grant No. TG-DMR090071) and the High-Performance Supercomputing Center Xiuhcoatl at Cinvestav, México for the computational resources provided. This research was supported by CONACYT Mexico Grant No. 83604 and in part by the DOE Basic Energy Science under Award DE-SC0006818 (R.R.Z.) and by Virginia Commonwealth University (J.U.R.).

■

REFERENCES

(1) Schlapbach, L.; Zuttel, A. Hydrogen-Storage Materials for Mobile Applications. Nature 2001, 414, 353−358. (2) Ogden, J. M. Hydrogen: The Fuel of the Future? Phys. Today 2002, 55 (4), 69. (3) Targets for Onboard Hydrogen Storage Systems for Light-Duty Vehicles, revision 4.0; Technical report produced by the US Department of Energy, Office of Energy Efficiency and Renewable Energy, and the FreedomCAR and Fuel Partnership:Washington, DC, 2009. http://energy.gov/sites/prod/files/2014/03/f11/targets_ onboard_hydro_storage_explanation.pdf. [Accessed 4 March 2014.] (4) Chen, P.; Wu, X.; Lin, J.; Tan, K. L. High H2 Uptake by AlkaliDoped Carbon Nanotubes Under Ambient Pressure and Moderate Temperatures. Science 1999, 285, 91−93. (5) Dillon, A. C.; Jones, K. M.; Bekkedahl, T. A.; Kiang, C. H.; Bethune, D. S.; Heben, M. J. Storage of Hydrogen in Single-Walled Carbon Nanotubes. Nature 1997, 386, 377−379. (6) Liu, C.; Fan, Y. Y.; Liu, M.; Cong, H. T.; Cheng, H. M.; Dresselhaus, M. S. Nanotubes at Room Temperature Hydrogen Storage in Single-Walled Carbon. Science 1999, 286, 1127−1129. (7) Li, J.; Furuta, T.; Goto, H.; Ohashi, T.; Fujiwara, Y.; Yip, S. Theoretical Evaluation of Hydrogen Storage Capacity in Pure Carbon Nanostructures. J. Chem. Phys. 2003, 119, 2376−2385. (8) Bhatia, S. K.; Myers, A. L. Optimum Conditions for Adsorptive Storage. Langmuir 2006, 22, 1670−1688. (9) Contescu, C. I.; van Benthem, K.; Li, S.; Bonifacio, C. S.; Pennycook, S. J.; Jena, P.; Gallego, N. C. Single Pd atoms in Activated Carbon Fibers and Their Contribution to Hydrogen Storage. Carbon 2011, 49, 4050−4058. (10) Tozzini, V.; Pellegrini, V. Prospects for Hydrogen Storage in Graphene. Phys. Chem. Chem. Phys. 2013, 15, 80−89. (11) Bhagavathi, V.; Rupali, P.; Sethupathi, N. K.; Ramaprabhu, S. Investigation of Spillover Mechanism in Palladium Decorated Hydrogen Exfoliated Functionalized Graphene. J. Phys. Chem. C 2011, 115, 15679−15685. 5008


Article

The Journal of Physical Chemistry C (34) Thrower, P. A.; Mayer, R. M. Point Defects and Self-Diffusion in Graphite. Phys. Status Solidi A 1978, 47, 11−37. (35) Palacios, J. J.; Yndurain, F. A Critical Analysis of VacancyInduced Magnetism in Mono and Bilayer Graphene. Phys. Rev. B: Condens. Matter Mater. Phys. 2012, 85, 245443−245450. (36) Medina, J.; de Coss, R.; Tapia, A.; Canto, G. Estructural, Energetic and Magnetic Properties of Small Tin (n = 1−13) Clusters; a Density Functional Study. Eur. Phys. J. B 2010, 76, 427−433. (37) Harman, J.; Wu, J.; Wei, S.; Kang, H. C.; Soon, E. GrapheneAdsorbed Fe, Co, and Ni Trimers and Tetramers: Structure, Stability, and Magnetic Moment. Phys. Rev. B: Condens. Matter Mater. Phys. 2011, 83, 205408−205425. (38) Wang, Q.; Jena, P. Density Functional Theory Study of the Interaction of Hydrogen with Li6C60. J. Phys. Chem. Lett. 2012, 3, 1084−1088.

5009


Ti4- and Ni4-Doped Defective Graphene Nanoplatelets as Efficient

Recommend Documents