Hydrogen storage in carbon nanostructures via spillover

Hydrogen storage in carbon nanostructures via spillover

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 6 Available online at www.sciencedirect.com ScienceDi...

1MB Sizes 8 Downloads 67 Views

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 6

Available online at www.sciencedirect.com

ScienceDirect journal homepage: www.elsevier.com/locate/he

Review Article

Hydrogen storage in carbon nanostructures via spillover Darryl S. Pyle*, E. MacA. Gray, C.J. Webb Queensland Micro- and Nanotechnology Centre, Griffith University, Nathan, 4111 Brisbane, Australia

article info

abstract

Article history:

The addition of transition metal nanoparticles to carbon nanostructures has been shown to

Received 11 June 2016

increase the hydrogen storage capacity of carbon nanostructures by dissociating molecular

Received in revised form

hydrogen and allowing adsorption via chemical means, a process known as hydrogen

19 July 2016

spillover. This paper is an overview of experimental and theoretical studies on hydrogen

Accepted 9 August 2016

storage on transition metal doped carbon nanostructures via the spillover mechanism and

Available online xxx

the prospects for achieving practical hydrogen storage targets. The most promising materials are found to be high surface area hexagonal system carbons for which the p-

Keywords: Hydrogen storage Carbon nanostructures

conjugation is broken by well dispersed oxygen functional groups or lattice dopants. Crown Copyright © 2016 Published by Elsevier Ltd on behalf of Hydrogen Energy Publications LLC. All rights reserved.

Spillover Surface functionalization

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hydrogen spillover to carbon nanostructures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Atomic hydrogen interaction with carbon substrate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hydrogen surface diffusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hydrogen clustering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The role of transition-metals in hydrogen spillover . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nanoparticle size and total mass doped dependencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Migration of atomic H to carbon substrate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Carbon surface functionalization, dopants and defects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Surface oxygen functionalization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Lattice dopants and defects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

00 00 00 00 00 00 00 00 00 00 00 00 00

* Corresponding author. E-mail address: [email protected] (D.S. Pyle). http://dx.doi.org/10.1016/j.ijhydene.2016.08.061 0360-3199/Crown Copyright © 2016 Published by Elsevier Ltd on behalf of Hydrogen Energy Publications LLC. All rights reserved. Please cite this article in press as: Pyle DS, et al., Hydrogen storage in carbon nanostructures via spillover, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.08.061

2

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 6

Introduction One of the key roadblocks to the spread of hydrogen energy technology is the safe, efficient and reliable storage of hydrogen. Current storage methods require either high pressures for pressurized hydrogen tanks or expensive cryogenic processes for liquefied hydrogen and are not considered viable options for vehicular applications in the long term. The US Department of Energy has specified a gravimetric goal of 5.5 wt.% hydrogen storage for an on board system by 2020 and an ultimate target of 7.5 wt.% [1]. While recent years have seen a number of major motor vehicle manufacturers releasing hydrogen vehicles, all bar one model fall below the 5.5 wt.% storage target, with the exception being the Toyota Mirai, claimed to be 5.7 wt.% using 700 bar storage tanks [2], but this is still well below the ultimate target of 7.5 wt.%. It is in this context that solid-state materials, which are able to store large quantities of hydrogen via chemical and physical mechanisms, are considered a potential storage solution. As the US DOE hydrogen storage target of 5.5 wt.% applies to the total system, when the mass of additional components are considered, the reversible hydrogen storage capacity of a solid-state material will need to be significantly greater than 5.5 wt.% for use in an on-board system [3]. In order to minimise the required hydrogen storage capacity much research has focused on light weight solid-state materials. Of particular research interest over the past 10 years has been carbon with its low weight, low cost and its numerous graphite-derived polymorphs and nanostructures such as graphite [4], activated carbon [5e7], fullerenes [8], nanofibers [9], nanotubes [10,11] and more recently, graphene [12e16]. Hydrogen adsorption to carbon nanostructures is predominantly a surface-area dependent process [17] in which molecular hydrogen is physically bound to the surface via van der Waals forces. Secondary effects from adsorption in pores in activated carbons, carbon nanotubes (CNTs) [18] and other microporous carbons are also evident due to increased physisorption binding energies that arise from the overlap of attractive van der Waals forces from multiple surfaces. For moderate storage conditions, i.e. temperature of ~300 K and pressures no greater than 100 bar, carbon nanostructures are yet to be found that can reproducibly store in excess of 2 wt.% [19e22]. This is due to the low binding energies associated with physisorbed molecular hydrogen for which thermal excitations lead to desorption. To improve hydrogen storage capacities, small amounts of transition metals (TM) have been added to carbon nanostructures, either physically mixed with the carbon or directly doped via wet chemistry synthesis. For such cases, hydrogen storage enhancements have been reported to be anywhere between 0% [23,24] and 900% [25], but typically of the order of 100e300% [26e29] above that of the un-doped carbon nanostructure. The proposed mechanism by which metal-doped carbons achieve an enhanced hydrogen storage capacity is spillover, defined as ‘the transport of an active species sorbed or formed on a first surface onto another surface that does not, under the conditions, adsorb or form the active species’ [30]. In this context the active species is atomic hydrogen, the first surface is a TM nanoparticle and the second surface is the carbon nanostructure,

however the spillover mechanism has been long known in the catalysis field [30e33] and has been demonstrated in hydrogen storage of metal-organic frameworks [34e36]. Hydrogen spillover comprises three steps as shown in Fig. 1: 1) molecular hydrogen is dissociated and atomic hydrogen is bound to the surface of TM nanoparticles; 2) atomic hydrogen migrates from TM nanoparticles to the carbon surface and 3) atomic hydrogen diffuses along on the carbon surface eventually forming a stable CeH bond after some distance. Understanding these three key interactions is necessary for a full understanding of the hydrogen spillover process and whether it can be manipulated as a route to achieving a fully reversible hydrogen storage capacity in excess of 5.5 wt.% on carbon nanostructures. This review provides an overview of the theoretical and experimental studies of hydrogen storage on carbon nanostructures doped with transition metal nanoparticles in order to increase the hydrogen storage capacity by utilising the spillover mechanism. We begin with the nature of the interaction of atomic hydrogen with carbon nanostructures, followed by the role played by transition metal nanoparticles and finish with how carbon nanostructure surface functionalization, lattice dopants and defects affect the hydrogen storage capacity.

Hydrogen spillover to carbon nanostructures The nature of the carbon nanostructure is of primary concern for hydrogen spillover, as the majority of stored hydrogen is bound to the carbon substrate while a small percentage is bound to the metal nanoparticle. Experimental studies on hydrogen spillover to carbon materials have been performed on activated carbons [34,37e39], carbon nanotubes [40e48], carbon nanofibers [49e52], templated carbon [53e56], carbon nanospheres [57], fullerenes [58] and graphene-like materials [26,59e62]. The low-density and low hydrogen uptake of carbon substrates present particular difficulties [63,64] in accurately determining hydrogen storage capacity that is reproducible between laboratories [65] and may go some way to explaining the large discrepancies reported for H2 uptake in TM-doped carbon nanostructures. The capacity of a carbon

Fig. 1 e Hydrogen spillover mechanism: Step 1) dissociation of molecular hydrogen and formation of TM e H bonds; step 2) migration of atomic hydrogen to the carbon surface; step 3) diffusion of atomic hydrogen along the carbon surface.

Please cite this article in press as: Pyle DS, et al., Hydrogen storage in carbon nanostructures via spillover, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.08.061

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 6

nanostructure to store atomic hydrogen reversibly relies on a number of factors such as the accessible surface area, geometric nanostructure, defects, dopants, surface functionalities and surface coverage. A large accessible surface area provides more adsorption sites for atomic hydrogen while the combination of the other factors yields the binding energy for atomic hydrogen to the carbon nanostructure. For effective hydrogen storage via the spillover mechanism, CeH binding energies need to be sufficiently high that the reaction 2 Me  C þ H2 / 2 Me  C  H proceeds thermodynamically downhill thus the CeH binding energy must be >1=2EH2 ¼ ~2.3 eV [66]. The majority of experimental hydrogen spillover studies have been conducted as primary spillover experiments for which metal nanoparticles are doped directly to the target carbon substrate. Secondary spillover entails the physical mixing of a small percentage of a metal-doped primary carbon nanostructure, typically a commercial metal-doped activated carbon, and a large percentage of a secondary carbon substrate. Here, dissociated hydrogen atoms migrate from the metal nanoparticle to the commercial activated carbon then diffuse along the surface and onto the secondary carbon substrate. Secondary spillover materials provide an opportunity to investigate the physical and chemical properties of the secondary carbon substrate independently from the effects from the interaction between the metal nanoparticle, hydrogen and carbon substrate. Lueking et al. studied multiwall carbon nanotubes (MWNTs) physically mixed with a commercial Pt/Pd-C material in varying ratios using a mortar and pestle. The hydrogen storage capacity was found to be enhanced by a factor of 2.7 [67]. They reported that when the type of secondary carbon was varied the enhancement factor for a number of different primary metal-doped carbons was dependent upon the secondary carbon, indicating that in the context of hydrogen spillover a carbon support that yields the highest H2 physisorption should be of primary concern. A further consideration with secondary spillover materials is the contact between the primary metal-doped carbon and the secondary carbon. Novel work by Lachawiec et al. investigated the secondary spillover from commercial 5 wt.% Pd on activated carbon to a high surface area activated carbon, AX21, where D-glucose was melted into the mixture and heated for carbonisation to occur in an attempt to create carbon bridges between the primary and secondary carbons [68]. Hydrogen adsorption performed at 298 K and up to 10 MPa yielded a 100% increase in hydrogen storage from 0.8 wt.% to 1.6 wt.% over that obtained for a sample without the D-glucose bridge.

Atomic hydrogen interaction with carbon substrate In the spillover mechanism hydrogen dissociates on the surface of metal nanoparticles and migrates to the carbon substrate where hydrogen atoms bind either physically or chemically. Theoretical calculations using a DFT-dispersion functional (as pure DFT calculations are inaccurate for dispersion forces [69]) have provided binding energies for atomic hydrogen physisorbed to model graphene surfaces of between 0.03 and 0.08 eV e low enough that thermal vibrations at room temperature would cause spontaneous

3

desorption [70,71]. Further DFT-d calculations by Psofogiannakis and Froudakis demonstrated that for two isolated physisorbed hydrogen atoms, there was no activation barrier to associative desorption of a H2 molecule, leading to the conclusion that high surface coverage of atomic hydrogen in a predominantly physisorbed state is not possible [72]. While the statistical distribution in kinetic energies of the hydrogen atoms will result in binding via both physisorption and chemisorption simultaneously, for hydrogen storage on the carbon surface to occur at statistically significant quantities at ambient temperatures, chemisorption must be the dominant binding regime. For binding via chemisorption at room temperature on a model graphene surface a number of DFT studies have found binding energies in the range of ~0.6e0.8 eV and for which the adsorption site is directly above a carbon atom forming a CeH bond [66,73e75]. To enter the chemisorption well activation energies were reported between 0.13 and 0.2 eV, sufficiently low that chemisorption of free atomic hydrogen at ambient temperatures can be considered facile [72,74e76]. This activation energy arises as the graphene plane undergoes reconstruction for which the carbon chemisorbed to the hydrogen atom is puckered up out of the plane by ~0.05 Å due to sp2 e sp3 rehybridisation during formation of the s CeH bond. Given the variety of carbon nanostructures an understanding of the effect that structure, typically surface curvature, has on the binding energies of chemisorbed hydrogen is needed. Chen et al. performed DFT calculations on model graphene and the exterior of single walled carbon nanotubes (SWNTs) [77]. For graphene, binding energies of 0.82 eV were found, in agreement with previous studies, while for (9,9) and (5,5) SWNTs, binding energies of 1.07 and 1.43 eV were found respectively, indicating that higher chemisorption binding energies are obtained with smaller radius CNTs. Yang et al. used DFT and quantum mechanics/molecular mechanics (QM/MM) calculations on (5,0), (7,0) and (9,0) zig-zag SWNTs and also found that both computational methods yielded binding energies that increased with decreasing CNT radius [78]. The increased binding energies are thought to arise from CNT curvature inducing rehybridisation and a weakening of the p-conjugation via pyramidalisation. Chemically, this results in a localisation of the electron density above carbons on the outer walls giving these sites a radical nature and increasing the total energy of the system [79].

Hydrogen surface diffusion For practical hydrogen storage via the spillover mechanism, surface diffusion of chemisorbed hydrogen atoms must be kinetically facile to free up adsorption sites for further atomic hydrogen migration from the metal nanoparticle. Psofogiannakis & Froudakis performed DFT calculations for diffusion of a hydrogen atom bound to a pristine graphene model in a purely chemisorbed state and found an activation energy of 1.35 eV, ~0.7 eV higher than they calculated for desorption of a hydrogen atom from the surface [72]. The authors calculated an energy of 0.63 eV for desorption of a chemisorbed hydrogen atom into the physisorbed state, an energy barrier of ~0.01 eV for diffusion while in the physisorbed state and an energy barrier of ~0.13 eV to chemisorption. The authors state

Please cite this article in press as: Pyle DS, et al., Hydrogen storage in carbon nanostructures via spillover, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.08.061

4

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 6

that surface diffusion of hydrogen atoms chemically bound to a pristine graphene plane is considered to occur via atomic hydrogen desorption into the physisorbed state where the atom can diffuse freely along the surface before being once again chemisorbed to another carbon atom some distance away. However, given the very low physisorption well for atomic hydrogen compared to the energy barrier for chemisorption, the majority of physisorbed hydrogen atoms would be expected to associatively desorb as H2. Further theoretical results were obtained by Chen et al. where an activation energy of 0.78 eV was reported for diffusion of a chemisorbed hydrogen atom to the nearest neighbour carbon atom while an activation energy of 0.95 eV was reported for diffusion to a carbon in a para position thus diffusing across the aromatic ring [77]. The energy of 0.63 eV calculated for a chemisorbed hydrogen atom to desorb into a physisorbed state [72] is sufficiently low that at 298 K the mean residence time for a chemisorbed hydrogen atom is 8.6  103 s, corresponding to a process that can proceed at sufficiently high rates at ambient temperature. A study by Liu et al. on Pt-doped pristine and oxidised activated carbon reported that the activation energy for atomic hydrogen surface diffusion must be < ~0.7 eV [80]. The authors came to this energy by stating an upper limit of 1 h for the adsorption time and equating this to a random walk in one dimension, tzL2 =D, where (L ¼ 20 nm) was the maximum distance a H atom was assumed to travel based on transmission electron microscopy (TEM) images. D is the diffusion coefficient given as:   1 EA D ¼ a2 ve exp  z kT where ve is the vibrational frequency parallel to the surface, z ¼ 3 is the number of adjacent sites that a H atom can possibly jump to on a graphene surface, a ¼ ~1.5 Å is the jump distance and EA is the activation energy [81]. At room temperature the authors calculated that for EA ¼ 0.6 eV the adsorption time was ~1 min while for EA ¼ 0.8 eV the adsorption time was ~40 h thus stating an upper limit of EA ¼ ~0.7 eV for an adsorption time of ~1 h. Chen et al. further investigated the effects of carbon plane curvature by performing DFT calculations on the surface diffusion of a hydrogen atom chemisorbed to carbon nanotubes (CNTs) and found activation energies of 1.09 and 1.42 eV for (5,5) and (9,9) SWNTs respectively [77]. As seen above the curvature effects lead to increased binding energies on the outside of CNTs thus resulting in an increase in activation energies for surface diffusion such that the smaller the radius of the CNT the higher the activation energy leading to much slower diffusion kinetics.

would favour sites that are ortho and para relative to the first adsorbed hydrogen atom. Casolo et al. used DFT calculations to map the spin density of a 5 by 5 graphene lattice with one chemisorbed hydrogen atom. Fig. 2 shows the localisation of the spin density on the ortho and para sites to the adsorbed hydrogen and that the spin density decays slowly with distance from the chemisorbed hydrogen yielding enhanced adsorption sites in neighbouring carbon rings. Theoretical calculations by Pereira et al. on a two dimensional graphene surface found that this spin localisation actually decays as 1/r [82]. Casolo et al. found that while the first hydrogen atom chemisorbed with a binding energy of 0.79 eV, a second hydrogen atom chemisorbed with 1.93 and 1.89 eV to the ortho and para sites respectively. DFT calculations by Psofogiannakis & Froudakis also showed that adsorption of a second hydrogen atom favoured the ortho and para sites with much higher binding energies and found that there was no activation energy for such chemisorption [83]. These computational results have been supported experimentally by Hornekaer et al. using scanning tunnelling microscopy (STM) on hydrogen chemisorbed to a graphene layer grown on a SiC substrate [84]. The graphene was exposed to a 1600 K beam of deuterium atoms for 5 s resulting in a low coverage of chemisorbed D after which STM images clearly showed the preference for D to form dimers on the para and ortho sites. Further experimental and theoretical work by Hornekaer et al. linked the double peak found in temperatureprogrammed desorption (TPD) of hydrogen on graphite to the ortho and para sites found in STM experiments and theoretical calculations [85]. The authors annealed a hydrogen impregnated sample of highly ordered pyrolytic graphite at 525 K (between the two TPD peaks at 445 K (1.25 eV) and 560 K (1.64 eV)), after which imaging via STM showed that only the ortho-dimers were present. Therefore the lower temperature

Hydrogen clustering Carbon nanostructures are p-conjugated systems resulting in the delocalisation of electrons. Chemisorption of a hydrogen atom and formation of a CeH bond breaks this conjugation, causing the electrons to localise above the carbons of the p* sub-lattice [75]. From a chemical point of view the carbons of this sub-lattice have an unpaired p electron yielding a radical nature and as such adsorption of a second hydrogen atom

Fig. 2 e Spin density of hydrogenated graphene surface at 0.47 Å above the surface. Red/Blue lines are excess spinup/spin-down. Reproduced with permission from Ref. [75]. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Please cite this article in press as: Pyle DS, et al., Hydrogen storage in carbon nanostructures via spillover, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.08.061

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 6

peak is associated with desorption of the para-dimers while the higher temperature peak is associated with the orthodimers. Subsequent DFT calculations yielded energy barriers for desorption from the ortho and para dimers of 1.6 eV and 1.4 eV respectively [85]. Theoretical calculations by Lin et al. advanced this idea by calculating the binding energies for up to 24 chemisorbed hydrogen atoms iteratively populated onto a C54 graphene model [66]. In agreement with previous studies they found a binding energy of 0.79 eV for a single hydrogen atom and binding energies of 1.35 and 1.70 eV for two hydrogen atoms in the para and ortho configurations respectively. As can be seen in Fig. 3(a) addition of a third chemisorbed hydrogen atom reduced the overall binding energy while addition of a fourth increased the binding energy before another reduction on addition of a fifth. Finally the most energetically stable structure was obtained on addition of a sixth chemisorbed hydrogen atom resulting in an entire carbon ring being populated with chemisorbed hydrogen. Taking the chemical point of view, upon addition of every odd number of chemisorbed hydrogen atoms there is an unpaired p electron, giving a radical like state and raising the total energy. Subsequent addition of the next chemisorbed hydrogen atom utilises this unpaired p electron, eliminating the radical like state and reducing the overall energy. This oscillation of energy can be seen Fig. 3(a). Fig. 3(b) shows increasing local maxima in the binding energy for each fully populated carbon ring, asymptotically approaching εb(∞) ¼ 2.54 eV. The orange line (in the web version) at 2.30 eV corresponds to 1=2EH2 and for a cluster of 24 chemisorbed hydrogen atoms populating 6 full carbons rings the average binding energy per hydrogen exceeds this energy. For hydrogen spillover to occur in sufficient quantities for practical hydrogen storage at ambient temperature the overall reaction must be thermodynamically favourable. Thus the binding energy to the carbon surface, εb(∞), must be greater than 1=2EH2 , a condition that is met with a cluster of 24 chemisorbed hydrogen atoms. Taking this effect to its theoretical limit of every carbon atom supporting a hydrogen atom we arrive at graphane with a theoretical hydrogen storage capacity of 7.7 wt.% [86]. However, this theoretical hydrogen storage capacity does not

Fig. 3 e (a) Increase in binding energy due to sorption of nth H. (b) Increase in binding energy per H demonstrating local maxima upon completion of closed-ring systems. Reproduced with permission from Ref. [66].

5

factor in using TM nanoparticles as the source of atomic hydrogen. TM nanoparticle impregnation to the substrate would lower the number of active sites for CeH formation and in addition with the much larger mass of TM atoms, significantly reduce the gravimetric storage capacity.

The role of transition-metals in hydrogen spillover For carbon nanostructures to be effective in hydrogen storage, sufficient quantities of atomic hydrogen must be supplied with reasonable kinetics to facilitate surface diffusion. For a TM nanoparticle to play this role it must have the ability to dissociate molecular hydrogen and bind atomic hydrogen at sufficiently low energies that migration from the nanoparticle to the carbon nanostructure can occur at ambient temperatures. The H2 dissociation and atomic hydrogen bonding energetics are determined by the type of TM nanoparticle, nanoparticle size, hydrogen surface coverage and nanoparticle anchoring to the carbon substrate. Factors that need to be considered in relation to overall hydrogen storage characteristics are the mass of metal doped to the carbon nanostructure and nanoparticle dispersion. Dissociated atomic hydrogen can dissolve into the bulk of TMs forming hydrides, however the reaction is exothermic [87] and not considered to occur at ambient temperature with the exception of Pd. Atomic hydrogen has been shown to absorb into bulk and even nanoparticles [88] of Pd with no energy barrier, readily hydriding to a Pd:H ratio of 0.8. The hydriding of Pd and surface adsorption on other TMs must be considered when deducing the hydrogen uptake attributed to spillover. Such a study by Takagi et al. showed that the hydrogen storage capacity increase exhibited via doping with Pt and Pd was no higher than what was expected from Pt surface adsorption and Pd hydriding [89]. Catalysis studies have shown that Pd and Pt readily dissociate molecular hydrogen and allow the migration of atomic hydrogen to a substrate at ambient temperatures [31e33]. The majority of experimental hydrogen storage via spillover studies utilise one of these two transition metals [50,90e95]. Less commonly Ni [49,96] and Ru [29] have also been shown experimentally to exhibit hydrogen spillover to carbon nanostructures at ambient temperatures. Wang et al. doped the high-surface area activated carbon, AX-21, with 6 wt.% Ru, Pt and Ni and found increases in hydrogen storage over un-doped AX-21 of 136%, 118% and 81% respectively, at 298 K and 100 bar. Ru showed the greatest increase and the highest total hydrogen storage via spillover of the three metals at 1.3 wt.%. The authors also doped a templated carbon with the same three metals at 6 wt.% and obtained the same trend of increased hydrogen storage with Ru > Pt > Ni. Experimental TPD studies have reported H2 dissociative chemisorption energies of 0.7e0.8 eV for Pt [97e99], 0.9e1.06 eV for Ni [100,101] and ~0.9 eV for Pd [102] for crystalline surfaces. These represent a minimum in reported chemisorption energies. Theoretical studies have shown that single Pd and Pt atoms bound to a graphene surface dissociate H2 with no energy barrier and that dissociative chemisorption energies for both are similar at ~1.7 eV/H2 for the

Please cite this article in press as: Pyle DS, et al., Hydrogen storage in carbon nanostructures via spillover, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.08.061

6

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 6

first H2 [103,104], representing a maximum in chemisorption energy. For small transition-metal nanoparticles used in hydrogen spillover materials, typically <10 nm in size, the dissociative chemisorption energies are expected to range between these values. For example, theoretical calculations by Balasubramanian et al. on Pt2 and Pt3 clusters found H2 dissociation energies ranging between 0.93 and 1.33 eV [105,106] while Okomoto, using DFT calculations, found the H2 dissociation energy for a Pt13 cluster to be 1.52 eV [107], in close agreement with those found experimentally by Liu et al. for Pt12 clusters bound to NaY zeolite [108]. Experimental values for binding energies of atomic hydrogen to common spillover TMs have been reported for crystalline surfaces as ~2.6 eV for Pt [109,110], 2.69 eV for Pd [111], 2.73 eV for Ni [112] and ~2.9 eV for Ru [113]. For single atoms and small clusters, theoretical studies have found atomic hydrogen binding energies of 3.82 eV for a Pt atom [72], ~3.0 eV for a Pt4 cluster [114] and ~3.00 and 2.85 eV for Pd13 and Pd147 respectively [115]. Nanoparticle size dependency of the adsorption and binding energies, and the effect this has on hydrogen storage via spillover, is discussed in depth in the following section. For Ni-doped carbons, experimental hydrogen spillover studies [23,116,117] have indicated spillover of hydrogen atoms occurs, although theoretical calculations have shown that single Ni atoms bound to graphene/CNT surface do not dissociate H2 [118]. Wu et al. performed DFT calculations on small pure Pt and mixed Pt and Ni clusters (Pt4, Pt3Ni, Pt2Ni2, PtNi3 and Ni4) and found that while the Pt atoms in the pure and mixed cluster readily dissociated H2, the Ni atoms only bind hydrogen in molecular form via Kubas binding [119]. Kubas binding is the binding between molecular hydrogen and supported TM atoms which entails forward charge donation of the s bond electrons in H2 to the TM atom's partially filled d orbital as well as back donation of the metal atom to the H2 anti-bonding orbital [120]. This results in a lengthening and weakening of the HeH bond and Kubas H2 bonding to the transition-metal is much weaker than MeH chemisorption but much stronger than physisorption. DFT calculations by Niu et al. showed that a neutral Ni atom could dissociatively chemisorb a single H2 molecule while a Niþ cation can bind up to 6H2 molecules without dissociation [121]. An explanation of the disagreement between experimental and theoretical results for H2 interaction with Ni nanoparticles is that for a neutral transition-metal atom interacting with H2, an electron is transferred from the metal to the H2. According to the Pauli exclusion principle, the transferred electron must occupy the anti-bonding orbital, thus dissociating the hydrogen molecule. However, for a TM ion the second ionisation potential is high and donation of an electron to H2 becomes energetically unfavourable. The binding of a Ni atom or small cluster to a carbon surface entails a charge transfer from the Ni to the carbon, giving the Ni atom or small cluster a cationic character. The larger the Ni cluster the greater the number of Ni atoms among which the loss of charge is shared, leading to the nanoparticle exhibiting a more neutral character the larger it becomes [121]. This may explain the discrepancies between the experimental hydrogen spillover studies and those theoretical studies of Ni atoms and small clusters.

Theoretical studies have also demonstrated the potential for Kubas binding in single Pd and Pt atoms supported on substrates. Contescu et al. showed via DFT calculation that for a Pd atom supported on a model graphene lattice molecular hydrogen adsorbed via Kubas binding with an activated HeH bond length of 0.86 Å [92]. Similar DFT studies by Psofogiannakis et al. for a single Pt atom bound to coronene found that adsorption of a first H2 resulted in dissociation while adsorption of a second H2 was via Kubas binding with no dissociation [72]. A plot of hydrogen storage capacity, at 298 K and 30 bar versus BET surface area from a number of reports in the literature can be found in Fig. 4. For pristine carbon nanostructures the hydrogen storage capacity increases with BET surface area, as expected for physical adsorption and as previously demonstrated [13]. However, for TM-doped carbon nanostructures there is a general increase in hydrogen adsorption but that increase is not primarily surface area dependent. Given that a larger accessible surface area would be expected to provide more sites for H atoms to chemisorb, the data suggests that TM nanoparticle characteristics play the dominant role in determining hydrogen storage via spillover.

Nanoparticle size and total mass doped dependencies The TM nanoparticle size and weight percent doped to a carbon substrate affect the hydrogen spillover kinetics and gravimetric storage capacity in a variety of ways, some counteracting. For example, as the TM nanoparticle size is increased the surface area and number of active sites available for H2 dissociation also increase, leading to faster H2 dissociation. However, if we approximate a nanoparticle as a sphere, the ratio of surface area to volume behaves as 3/r, so increasing the size also increases the number of non-active internal TM atoms relative to the active surface atoms. Given the much larger atomic mass of typical TMs relative to carbon (e.g. the Pt:C mass ratio is 16.3:1), increasing the nanoparticle size increases the mass of the sample at a faster rate than it increases the number of active surface sites, eventually leading to a decrease in the gravimetric hydrogen storage capacity. It is also known that TM atoms and very small clusters bind atomic hydrogen more strongly than the surface of the bulk metal and that the hydrogen binding energies decrease with increasing atomic hydrogen surface coverage. A summary of reports of hydrogen storage capacity for a number of pristine and TM-doped carbon nanostructures with associated BET surface areas and reversibility information is provided in Table 1. The synthesis of spillover materials is such that it is experimentally difficult to isolate the effects of nanoparticle size from the effects of changing the total mass percentage of dopant as increasing the average size of the TM nanoparticles leads to an increase in the total mass percentage doped and vice versa. Many synthesis routes involve oxidation of the carbon substrate, which as described in Section surface oxygen functionalization, also affects the H2 storage via hydrogen spillover. Further to this, an increase in the relative wt.% of dopant reduces the accessible surface area of the carbon substrate, thus reducing the number of active

Please cite this article in press as: Pyle DS, et al., Hydrogen storage in carbon nanostructures via spillover, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.08.061

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 6

7

Fig. 4 e Relationship between excess H2 adsorption (298 K and 30 bar) and BET surface area for a selection of pristine (unfilled symbols) and TM-doped (filled symbols) carbon nanostructures. Data sourced from literature [13,19,20,26,29,37,49,53,96,122e128].

adsorption sites for spillover atomic hydrogen. For example, Anson et al. doped activated carbons with 30 and 50 wt.% Pd, resulting in average nanoparticle sizes of 32 and 42 nm respectively and a corresponding decrease in accessible surface area [37,129]. Zielinski et al. doped activated carbon with 1, 5 and 10 wt.% Ni but only reported nanoparticle sizes for the 10 wt.% sample of between 10 and 40 nm and confirmed that the accessible surface area decreased with increasing wt.% of dopant [96]. Wang et al. doped a high surface area templated carbon with 3, 6 and 8 wt.% Ru and reported that the 6 wt.% Ru doped sample exhibited the highest hydrogen storage capacity, followed by the 8 wt.% then the 3 wt.% sample. A Ru nanoparticle size of 2 nm was reported for the 6 wt.% sample in addition to decreased surface area with increasing wt.% of dopant [29]. Zubizarreta et al. doped carbon nanospheres with Ni at quantities between 1.3 and 12.4 wt.% but did not report the nanoparticle size for any of the materials [57]. Zhao et al. doped an activated carbon with 1.3, 3.3, 6.7 and 10 wt.% Pd while keeping a constant nanoparticle size of ~2 nm with a 15% spread in size [130]. They reported that the hydrogen storage capacity increased with increasing wt.% of Pd over that of the pure activated carbon. The 10 wt.% Pd doped material yielded an increase of ~30%, at the low end of expectations for high surface area AC (3000 m2/g) doped with Pd. Two questions thus arise: (i) would the hydrogen storage capacity continue to increase if the amount of Pd doped exceeded 10 wt.%? and (ii) was the small size of the Pd nanoparticles (~2 nm) the reason for relatively minor increase in hydrogen storage capacity? Tsao et al. used inelastic neutron scattering (INS) difference spectra to quantify the molecular hydrogen dissociated in two Pt doped activated carbons via the reduction in the area

of the H2 rotation transition peak at ~14.7 meV [131]. The two activated carbons were doped with 3.32 and 0.79 wt.% Pt yielding average nanoparticle sizes of 4.5 and 1.8 nm respectively. By analysing the H2 rotation transition peak the authors calculated that the sample with a nanoparticle size of 4.5 nm exhibited a dissociated hydrogen to Pt atom ratio of 1.8 while the sample with nanoparticle size of 1.8 nm exhibited a ratio of 25. Therefore, the sample with the smaller Pt nanoparticles and lower Pt doped mass dissociated more than an order of magnitude more H2 per Pt atom than the sample with larger nanoparticles and a higher amount of dopant. As the sample synthesis required oxidation of the activated carbon, the authors could not conclude whether the Pt nanoparticle size or sample pre-treatment was the decisive factor in these results. Huang et al. performed a calorimetry study on Pd nanoparticles varying between 1 and 10 nm bound to SiO2 and found that the heat of chemisorption and thus the dissociative chemisorption energy decreased with increasing nanoparticle size [132]. For 1 nm nanoparticles the dissociative chemisorption energy was found to be ~1.9 eV, sharply decreasing to ~1.1 eV for 3 nm nanoparticles and ~0.85 eV for 5 nm particles after which the energies remained approximately constant for increasing size. Lee et al. performed DFT calculations on Pd13, Pd55 and Pd147 nanoparticles and found that the dissociative chemisorption energy also decreased as a function of nanoparticle size [115]. Zhou et al. performed DFT calculations on Ptn clusters for n ¼ 2e9 and found that the cluster size did not affect the average H2 dissociative chemisorption energy or hydrogen atom sequential desorption energy in any consistent fashion [114]. However the authors found that the dissociative chemisorption energy and sequential desorption energy

Please cite this article in press as: Pyle DS, et al., Hydrogen storage in carbon nanostructures via spillover, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.08.061

8

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 6

Table 1 e Summary of reports of hydrogen storage capacity in TM doped carbons. Carbon Substrate A.C. (AX-21) [91] A.C. [96] A.C. [96] A.C. [96] A.C. [134] A.C.(AX-21) [126] A.C. [130] A.C. [130] A.C. [130] A.C. [130] A.C.(MAXSORB) [37] A.C. (MAXSORB) [37] A.C.(AX-21) [29] A.C.(AX-21) [29] A.C.(AX-21) [29] A.C.(rGO secondary) [59] Templated Carbon [91] Templated Carbon [125] Templated Carbon [29] Templated Carbon [29] Templated Carbon [29] Templated Carbon [29] Templated Carbon [29] A.C. Fibre [92] A.C. Fibre [135] A.C. Fibre [49] rGO [90] rGO [136] rGO [26] rGO [26] rGO [61] rGO [61] rGO [60] Ordered Carbon [128] Ordered Carbon [128] Ordered Carbon [127] CNT [42] CNT [42] MWCNT [44] MWCNT [44] SWCNT [37] SWCNT [37]

SA (m2/g)

Initial H2 @ T ¼ 298 K (wt.%)

TM Dopant

Surface Treatment

Final H2 @ T ¼ 298 K (wt.%)

Reversibility

2834 1073 1073 1073 1200 2880 3037 3037 3037 3037 2112 2112 2850 2850 2850 3300 3798 3400 3839 3839 3839 3839 3839 1996 2017 1626 e 470 754 754 e e 146.4 825 682 953 e e 52 52 262 262

0.58 (100 bar) 0.15 (30 bar) 0.15 (30 bar) 0.15 (30 bar) 0.3 (100 bar) 0.6 (100 bar) 0.6 (80 bar) 0.6 (80 bar) 0.6 (80 bar) 0.6 (80 bar) 0.42 (90 bar) 0.42 (90 bar) 0.55 (100 bar) 0.55 (100 bar) 0.55 (100 bar) 0.5 (80 bar) 0.8 (100 bar) 0.8 (100 bar) 0.75 (100 bar) 0.75 (100 bar) 0.75 (100 bar) 0.75 (100 bar) 0.75 (100 bar) 0.19 (20 bar) 0.16 (20 bar) 1.1 (100 bar) 0.38 (50 bar) 0.75 (40 bar) 0.06 (57 bar) 0.06 (57 bar) 0.36 (40 bar) 0.55 (40 bar) 1.50 (40 bar) 0.12 (30 bar) 0.22 (30 bar) 0.5 (100 bar) 0.5 (18 bar) 0.5 (18 bar) 0.01 (20 bar) 0.01 (20 bar) 0.21 (90 bar) 0.21 (90 bar)

Pt at 6 wt.% Ni at 1 wt.% Ni at 5 wt.% Ni at 10 wt.% Pt at 3 wt.% Pt at 5.6 wt.% Pd at 1.3 wt.% Pd at 3.3 wt.% Pd at 6.7 wt.% Pd at 10 wt.% Pd at 30 wt.% Pd at 60 wt.% Ru at 6 wt.% Pt at 6 wt.% Ni at 6 wt.% Pd at 7.8 wt.% Pt at 6 wt.% Pt at 6 wt.% Ru at 3 wt.% Ru at 6 wt.% Ru at 8 wt.% Pt at 6 wt.% Ni at 6 wt.% Pd at 2.6 wt.% Pd at 2 wt.% Ni Pd at 20 wt.% Pd at 20 wt.% Pd at 2.2 wt.% Pt at 49 wt.% Ni at 3 wt.% Pd at 10 wt.% Pd at 30 wt.% Ni at 5 wt.% Ni at 5 wt.% Ru at 6 wt.% Pd at 2.5 wt.% V at 2.5 wt.% Pd at 7.9 wt.% V at 5.3 wt.% Pd at 13 wt.% Pd at 32 wt.%

e e e e e e e e e e e e e e e e e e e e e e e e e F-doped Oxidised Oxidised Oxidised Oxidised Oxidised Oxidised, N-doped Oxidised, N-doped

1.1 (100 bar) 0.53 (100 bar) 0.48 (100 bar) 0.40 (100 bar) 0.48 (100 bar) 1.2 (100 bar) 0.55 (80 bar) 0.50 (80 bar) 0.48 (80 bar) 0.42 (80 bar) 0.4 (80 bar) 0.7 (80 bar) 1.3 (100 bar) 1.2 (100 bar) 1.0 (100 bar) 0.83 (100 bar) 1.35 (100 bar) 1.35 (100 bar) 1.20 (100 bar) 1.43 (100 bar) 1.30 (100 bar) 1.33 (100 bar) 1.14 (100 bar) 0.26 (20 bar) 0.23 (20 bar) 1.6 (100 bar) 1.5 (50 bar) 3.0 (50 bar) 0.15 (57 bar) 0.15 (57 bar) 0.55 (40 bar) 1.8 (40 bar) 2.2 (40 bar) 0.2 (30 bar) 0.26 (30 bar) 1.2 (100 bar) 0.64 (18 bar) 0.66 (18 bar) 0.125 (20 bar) 0.10 (20 bar) 0.16 (90 bar) 0.51 (90 bar)

90% Desorb Not Reported Not Reported Not Reported 80% Desorb 90% Desorb Not Reported Not Reported Not Reported Not Reported Not Reported Not Reported Not Reported Not Reported Not Reported 100% Desorb 93% Desorb Not Reported Not Reported Not Reported Not Reported Not Reported Not Reported 90% Desorb 90% Desorb Not Reported 100% Desorb 100% Desorb Not Reported Not Reported Not Reported 100% Desorb 100% Desorb Not Reported Not Reported 100% Desorb Not Reported Not Reported Not Reported Not Reported 90% Desorb 90% Desorb

decreased with increasing hydrogen surface coverage. For the first hydrogen molecule the average H2 dissociative chemisorption energy was found to be ~1.7 eV, regardless of Pt cluster size, while for the last hydrogen molecule this energy was 1.1 eV for a Pt2 cluster, slowly decreasing to ~0.95 eV for Pt7, which is in close agreement with the energies obtained from experimental thermal desorption spectroscopy [133]. For hydrogen atom sequential desorption an energy of ~3 eV was calculated for a Pt cluster with 4 bound hydrogen atoms regardless of cluster size while at full hydrogen coverage the energy decreased to a minimum of ~2.5 eV for Pt clusters of 4 atoms and higher. While the sequential desorption energy did not decrease beyond 2.5 eV at full hydrogen coverage as the Pt cluster size was increased, the number of hydrogen atoms at this minimum energy did increase. For example a Pt5 cluster bound two hydrogen atoms at a minimum of ~2.5 eV while a Pt8 cluster

N-doped B & N-doped e e e e e e

bound 10 hydrogen atoms at the same energy. It would be expected that as the cluster size increased the majority of bound hydrogen atoms would be at the minimum sequential desorption energy. Given the significantly higher mass of TM atoms compared to carbon atoms and the reduction in accessible surface area with the increase in wt.% doped, an understanding of the optimal nanoparticle size and wt.% doped is required to fully explore the possibility of TM-doped carbons meeting hydrogen storage targets. Synthesis routes that keep a constant nanoparticle size while increasing the total wt.% doped and vice versa need to be developed. On the theoretical side, studies to date have been conducted on isolated nanoparticles or nanoparticles bound to materials that are not carbon. Therefore there is a need to conduct computational studies on variation in nanoparticle size for nanoparticles bound to carbon.

Please cite this article in press as: Pyle DS, et al., Hydrogen storage in carbon nanostructures via spillover, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.08.061

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 6

Migration of atomic H to carbon substrate Once H2 is dissociated on the surface of a TM nanoparticle, atomic hydrogen must migrate to the carbon substrate in sufficient quantities to populate the accessible surface area. Given the strength of the binding of atomic hydrogen to TMs compared to carbon in a graphene plane, the reaction would appear to be thermodynamically unfavourable. The ability of atomic hydrogen to migrate to carbon nanostructures has therefore been the subject of much controversy given the reported experimental findings of hydrogen storage via spillover in TM-doped carbons. Psofogiannakis and Froudakis performed DFT calculations on a single Pt atom bound to coronene (7-ring graphene model), reporting CeH binding energies of ~0.8 eV in agreement with previous studies and a PteH binding energy of ~3.8 eV [72]. As a Pt atom binds atomic hydrogen ~3 eV stronger than graphene, the reaction is thermodynamically unfavourable. Given the dependence of the energy of atomic hydrogen on hydrogen surface coverage, the authors performed an energy scan for the migration of a single hydrogen atom from a Pt4 cluster at full hydrogen coverage bound to coronene and found an energy barrier of ~2.6 eV, implying that migration of atomic hydrogen could not occur under ambient conditions. The effect of large overlaps between the potential wells of carbons in a graphene plane and those of bound TM atoms/ clusters was revealed in a theoretical study by Chen et al. on a Pt6 cluster with full atomic hydrogen coverage. An energy barrier of only 0.48 eV/H atom for migration from the cluster to a C50 graphene plane was predicted [137]. While the calculated energy barrier was greatly reduced compared to the previous study by Psofogiannakis and Froudakis, the mechanics of interaction were also vastly different. Chen et al. started with a fully hydrogen saturated Pt6H24 cluster separated from the graphene plane and performed minimum energy pathway calculations by gradually moving the cluster toward the graphene plane. At an un-reported distance above the graphene plane, hydrogen atom migration occurs at a point where there is large overlap of the potential wells, resulting in a greatly reduced energy barrier that could be overcome at ambient temperature. While variability of the potential well overlap caused by thermal vibrations may be a valid argument for small clusters, the vast majority of reported experimental hydrogen spillover studies are of nanoparticles >2 nm containing hundreds or even thousands of atoms and would be expected to be bound to the carbon nanostructure at numerous points, restricting displacement owing to thermal vibrations. Thus a likely cause of the discrepancy between experimental and theoretical spillover studies is the imperfection of the carbon nanostructures introduced by oxygen surface functionalization, lattice dopants and defects during bulk synthesis of these materials.

Carbon surface functionalization, dopants and defects Most carbon nanostructures are formed from graphene planes in various geometric structures, whether they are curved as in

9

single and multi-walled carbon nanotubes and nanospheres, stacked as in nanofibers and few-layer graphene or a combination of the two in a random configuration as in activated carbon. As discussed in the previous sections, adsorption of a hydrogen atom and subsequent formation of a CeH bond breaks the p-conjugation, lowering the stability. For pristine graphene surfaces this leads to CeH binding energies that are generally weaker than TMeH binding and a reaction pathway for migration of atomic hydrogen that is endothermic and thus unlikely to occur to any significant extent at ambient temperature. The bulk synthesis methods for making carbon nanostructures are such that oxygen surface functionalization, lattice dopants and defects are unintentionally generated [138e140], breaking the p-conjugation and creating favourable adsorption sites that act as nucleation sites for both hydrogen clustering behaviour and TM nanoparticle growth [141]. Popular methods of doping carbon nanostructures with TM nanoparticles entail oxidation of the carbon surface to create nanoparticle nucleation sites, further adding surface oxygen functionalities and lattice defects [142e145]. Carbon nanostructures have also been intentionally oxidised and doped with electron donors and acceptors such as nitrogen and boron, to investigate the effect on hydrogen uptake via spillover [27,127,146,147].

Surface oxygen functionalization Wang et al. investigated the effect of oxidation of Pt-doped templated carbon and AX-21 activated carbon and reported increases in hydrogen storage capacities of 49% and 51% respectively over that of un-oxidised Pt-doped samples at 298 K and 100 bar [146]. Glow discharge was used to increase the oxygen content from 2.74% to 30.66% and 5.5%e29.15% for the AX-21 and TC respectively. Desorption was reported to be almost fully reversible, however successive hydrogen adsorption cycles produced reduced hydrogen storage capacity. For the first cycle, the H2 storage capacity of the oxidised Pt-doped templated carbon was reported to be 1.74 wt.% at 298 K and 100 bar which dropped to 1.39 wt.% on the second cycle and 1.30 wt.% for the third cycle, after which the storage capacity was maintained for successive cycles. Reduction of the oxygen content by atomic hydrogen during hydrogen sorption was demonstrated via XPS analysis which showed that after 4 cycles the surface lactone groups were reduced from 4.0% to 0.97%, while other oxygen functional groups, semiquinone (25%) and hydroxyl (6%), remained largely unchanged. In another study the authors reported an increase in the H2 storage capacity of a 10 wt.% Pd-doped AX-21 activated carbon after being oxidised to 13 wt.% oxygen content [148]. These results demonstrate that the reactivity of different forms of oxygen functional groups effect the hydrogen storage capacity of spillover materials and must be taken into consideration given that the method of oxidation favours the formation of certain oxygen functional groups [149,150]. Using KOH, Li and Lueking oxidised a high surface area AC that was physically mixed in a 9:1 ratio with commercial 5 wt.% Pt-C, resulting in an overall oxygen content of 32 wt.% [147]. Hydrogen adsorption by the oxidised 5 wt.% Pt-C/AC was reported to be 1.4 wt.%, a significant increase over the un-

Please cite this article in press as: Pyle DS, et al., Hydrogen storage in carbon nanostructures via spillover, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.08.061

10

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 6

oxidised Pt-C/AC (0.1 wt.%), however only 0.4 wt.% of the sorbed H2 was able to be desorbed. The authors speculated that hydrogen remains trapped on the sample in the form of hydroxyl groups that would require elevated temperatures to desorb. In the past few years, one of the most promising carbon nanostructures for H2 storage applications has been an oxidised few-layer graphene material that has been reported under a number of differing names such as graphene, graphene nanoplatelets, functionalized graphene and few-layer graphene, but for which the proposed nomenclature is “reduced graphene-oxide” (rGO) [151], and demonstrates H2 adsorption capacities in excess of that expected from its small surface area [13]. The synthesis of the material entails the harsh oxidation of bulk crystalline graphite followed by reduction and exfoliation by methods such as thermal [152e154], chemical [155e157], solar radiation [158], electrochemical [159e161], solvo-thermal [162,163], ball-milling [164] and microwave radiation [145,165,166]. Owing to the oxidised nature of few-layer graphene, rGO exhibits numerous sites for nanoparticle nucleation and has produced some of the largest increases in H2 storage capacity when doped with TM nanoparticles [26,90]. Vinayan et al. doped rGO with 20 wt.% Pd and reported a H2 storage capacity of 1.4 wt.% at 298 K and ~35 bar, demonstrating enhanced adsorption over that expected from Pd-hydriding but did not report on the reversibility of the material [90]. Psofogiannakis et al. doped rGO with a PdHg nano-alloy, and reported a H2 storage capacity of 2 wt.% at 298 K and 20 bar that was almost fully reversible [25]. Parambhath et al. doped rGO with 20 wt.% Pd and reported a hydrogen storage capacity of ~3 wt.% at 298 K and 40 bar and reported the material to be fully reversible [136]. Characterisation of these rGO materials shows that they are comprised of varying numbers of stacked graphene layers, so the accessible surface area is far below the theoretical limit. Further exfoliation of these layers during bulk synthesis is currently a very active research field for numerous applications including hydrogen storage. Exfoliation of these layers will lead to higher surface area and more active adsorption sites thus pushing the hydrogen storage capacity higher. Theoretical studies have been undertaken in an attempt to elucidate the origin of the increased hydrogen storage capacity that has been found experimentally after oxidation of carbon nanostructures. Lueking et al. performed DFT calculations on a C16 graphene model with two hydroxyl functional groups in neighbouring carbon rings and reported a CeH binding energy to a carbon in an ortho position of 2.66 eV, substantially higher than for a non-functionalized C16 (0.72 eV) [167]. An activation energy of 0.99 eV was reported for diffusion of a CeH bound hydrogen atom to a nearest neighbour carbon atom that is also in an ortho position to a hydroxyl group. Atomic hydrogen diffusion by this pathway could be considered feasible at ambient temperature. Psofogiannakis et al. performed a DFT study on a Pt4 cluster bound to oxidised coronene and reported that for migration of a hydrogen atom from the Pt4 cluster to a chemisorbed oxygen, forming a hydroxyl group, an activation energy of 0.4 eV was required with the overall reaction being exothermic by 0.67 eV [168]. For migration of a hydrogen atom bound to oxygen as a hydroxyl group to an oxygen of a neighbouring

epoxide group in an ortho position, an activation energy of 0.33 eV was reported. For the dissociation of a hydroxyl group followed by formation of an epoxide group and CeH bond a large activation energy of 2.75 eV was reported. The activation energies of these processes are such that migration of atomic hydrogen from TM nanoparticles to a carbon surface bound oxygen is kinetically facile and thermodynamically favourable at ambient temperatures, although successive hydrogen diffusion would only be expected to occur between surface bound oxygen groups, given the prohibitive energy barrier to CeH formation from hydroxyl groups. The large binding energies between atomic hydrogen and surface oxygen groups give rise to a question as to the mechanism by which hydrogen desorption occurs. In the above theoretical study, Psofogiannakis et al. investigated the desorption of hydrogen atoms from two hydroxyl groups in a para configuration to form a H2 molecule and two epoxide groups. They reported the reaction to be endothermic by 2.2 eV, while formation of one epoxide group and one H2O molecule was exothermic by 0.45 eV with a 0.41 eV activation energy. For the reverse spillover mechanism, atomic hydrogen migration from a hydroxyl group to a Pt4 cluster where associative desorption can occur, the overall reaction was endothermic by an unreported amount with an activation energy of 1.1 eV, so that at ambient temperature the reaction kinetics would be very slow. Therefore, formation and desorption of H2O is energetically more favourable than either associative desorption of H2 from surface bound hydroxyl groups or on the surface of TM nanoparticles via the reverse spillover mechanism, leading to degradation of the surface. As already reported, this proposed surface degradation during H2 sorption has been shown experimentally for some oxidised carbon nanostructures where successive H2 sorption cycles provided reduced H2 storage capacity. However, reduction in H2 storage capacity over successive cycles has been shown to stabilise after a few cycles, with the repeatable H2 storage capacity remaining higher than that of the unoxidised samples. Further, as already seen, some experimental studies show fully reversible H2 sorption on oxidised carbons, indicating no degradation of the material surface. Surface oxygen functional groups have also been proposed to act as nucleation sites for TM nanoparticle formation. An experimental study by Parambhath et al. showed via Fourier transform infrared spectroscopy (FTIR) and XPS analysis a reduction in surface oxygen functionalization after Pd nanoparticle impregnation [136].

Lattice dopants and defects As with oxygen surface functionalization, introduction of lattice dopants and defects into a graphene plane breaks the p-conjugation, resulting in electron localisation above carbon atoms in the ortho and para positions with respect to the lattice dopant or defect and leading to increased CeH binding energies. Typical lattice dopants used in experimental and theoretical studies are nitrogen and boron that act as electron donors and acceptors respectively. A number of experimental studies have investigated the effect that B and N doping has on carbon materials doped with TM nanoparticles, with all reporting increases in H2 storage

Please cite this article in press as: Pyle DS, et al., Hydrogen storage in carbon nanostructures via spillover, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.08.061

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 6

after B and N doping. Giraudet et al. further doped a 5 wt.% Ni doped activated carbon with 3.9 wt.% nitrogen and reported a 30% increase in hydrogen storage over the Ni-doped AC without nitrogen doping, although the overall hydrogen storage at 298 K and 30 bar was very low at 0.26 wt.% [128]. Vinayan et al. doped an rGOetype material with 10 wt.% Pd and 3 at.% nitrogen and reported a hydrogen storage capacity of ~1.8 wt.% at 298 K and 40 bar, an increase by a factor of seven over that of the un-doped rGO material [61]. Parambhath et al. also doped rGO material with 20 wt.% Pd and 7 at.% nitrogen and reported a 47% increase in hydrogen storage over the material with just Pd dopant at 298 K and 40 bar [136]. Wang et al. doped an rGO material with 0.83 wt.% Ni and 1.09 wt.% boron and reported an increase in hydrogen storage by a factor of 25 over the un-doped material with the overall hydrogen storage reported as ~1.8 wt.% at 273 K and 1 bar [27]. Yang et al. doped an activated carbon with 6 wt.% Ru, 9 at.% boron and 10 at.% nitrogen and reported an increase in hydrogen storage by a factor ~3.3 over the un-doped material with the overall hydrogen storage reported as 1.2 wt.% at 298 K and 100 bar [127]. Given the low number of experimental studies on B/N lattice dopants and the variance in carbon nanostructures, transition metal and quantity of B/N doped, a number of questions still remain, including the optimal quantity of B/N doped for each carbon nanostructures, whether B/NeH bonding is energetically favourable and does the addition of B/N create nucleation sites for CeH clustering? A number of theoretical studies on B/N lattice dopants have been conducted to elucidate the origin of the increased hydrogen storage capacities that have been reported experimentally. Wu et al. performed DFT calculations on a Pt4 cluster bound to a model graphene plane that contained varying quantities of substitutionally doped boron. The authors reported binding energies for atomic hydrogen bound to a boron atom of 1.6 eV and 1.9 eV to a carbon in the ortho position to boron, indicating that atomic hydrogen binds to neighbouring carbon atoms more favourably then directly to the boron atom [169]. They reported an increase in the binding energy of a Pt4 cluster to the surface from 1.18 eV for pristine graphene to 2.38 eV for boron doped graphene, indicating that substitutional boron doping stabilises TM nanoparticles. The authors performed an energy scan for migration of a hydrogen atom from the Pt4 cluster to the graphene plane and for the case of pristine graphene an activation energy of ~2.7 eV was reported, in agreement with previous theoretical studies, while for a single B atom doped into a five ring graphene plane the activation energy decreased to ~1.9 eV. When the B doping was increased to a BC3 ratio, the activation energy for migration of a hydrogen atom from a TM was reported to be greatly reduced (~0.65 eV) while the activation energy for surface diffusion was 0.52 eV. These activation energies are sufficiently low that atomic hydrogen migration from a TM nanoparticle followed by surface diffusion away from the nanoparticle could be considered a process that can readily occur at ambient temperature. For a graphene lattice vacancy, removal of a carbon atom leaves 3 dangling s bonds that are localised on the three carbons nearest to the vacancy, and an unpaired p electron that delocalises onto one of the two sub-lattices. The vacancy typically undergoes a JahneTeller reconstruction for which

11

two of the carbon atoms weakly bind to form a pentagon with an elongated bond length about 2 Å. Opposite this bond is the apical carbon atom that is highly reactive. DFT calculations by Casartelli et al. showed that adsorption of a first hydrogen atom proceeds without an energy barrier and a strong exothermic adsorption energy of 4.24 eV, while adsorption of a second H atom is also exothermic at ~3 eV [170]. While these adsorption energies are such that desorption of the bound atomic hydrogen would require elevated temperature, the high reactivity of the vacancy defect sites would act as nucleation sites for TM nanoparticle formation. Kim et al. performed DFT calculations on vacancy defects in a graphene plane and found binding energies for Sc, Ti and V of 7.08 eV, 9.03 eV and 8.29 eV respectively, greatly increased relative to pristine graphene (2.08 eV, 2.73 eV and 2.00 eV respectively) [141]. Given that the cohesive energies of Sc, Ti and V are 4.83 eV, 6.50 eV and 6.66 eV respectively, binding of TMs to vacancy defects is energetically favoured over TM agglomeration, resulting in smaller and more dispersed TM nanoparticles. Krasheninnikov et al. showed via DFT calculations that vacancy defects in graphene produced strain fields in the lattice to a distance of about 2 nm from the defect site, for which the carbon atoms of the pristine graphene lattice are more reactive, and speculated that these strain fields guide TM atoms to the defect sites [171]. The effect that these strain fields have on the activation energy for atomic hydrogen migration from a TM cluster or small nanoparticle bound to a vacancy defect site would be of particular interest for hydrogen storage via spillover in carbon nanostructures.

Conclusion The addition of transition metal nanoparticles to carbon nanostructures has been demonstrated through numerous experimental studies to increase the hydrogen storage capacity at ambient temperature. The hydrogen spillover mechanism has been proposed to be the origin of these increases. Early theoretical studies cast doubt on the existence of the spillover mechanism on pristine carbon nanostructures, due to the apparent lack of a thermodynamic pathway and the high activation energy for migration of atomic hydrogen to the stable p-conjugated system. However, bulk synthesised TM-doped carbon nanostructures are anything but pristine, exhibiting defects and surface oxygen groups that break the p-conjugation and create highly reactive sites which act as nucleation sites for both TM nanoparticle formation and potential atomic hydrogen clustering. Later theoretical studies taking these effects into account have demonstrated thermodynamic pathways and activation energies such that the hydrogen spillover mechanism could occur at ambient temperature. However, with the wealth of experimental results leading to a consensus that doping carbons with TM nanoparticles enhances hydrogen storage via spillover, the underlying factor remains the carbons substrate capacity to chemically store hydrogen. While the theoretical limit of one hydrogen atom per carbon, graphane, yields a storage capacity of 7.7 wt.%, the act of impregnating TM nanoparticles, as the source of atomic hydrogen, would reduce the gravimetric storage capacity to

Please cite this article in press as: Pyle DS, et al., Hydrogen storage in carbon nanostructures via spillover, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.08.061

12

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 6

below what would be required to meet the ultimate DOE system target of 7.5 wt.% and most probably the system target of 5.5 wt.%. Recent experimental studies of the introduction of oxygen functional groups and lattice dopants have pushed the hydrogen storage capacity above 3 wt.% at 298 K and 40 bar [136]. While the introduction of oxygen functionalization and lattice dopants may lead to more reactive adsorption sites via breaking of the p-conjugation, the overall number of active sites is at best unchanged and most likely reduced. Furthermore, serious questions remain as to the reversibility of these materials, with contradictory experimental results reported and theoretical studies indicating that degradation of the sample is energetically favoured over H2 desorption. Therefore, while the question of whether TM doped carbon nanomaterials could meet the DOE total system targets and provide a hydrogen storage capacity in excess of 5.5 wt.% is unanswered, the prospects do not look promising. A number of possibilities for improvement remain such as optimising TM nanoparticle size, TM wt.% and the level of oxygen functionalization or lattice dopants or a combination of both. Taking into consideration these areas for improvement, the most promising materials for further research on hydrogen spillover to carbon nanostructures, appear to be high surface area hexagonal system carbons for which the p-conjugation is broken by well dispersed oxygen functional groups or lattice dopants.

references

[1] Energy, U.S.D.o.. US DOE technical plan - storage. 2012. http://www.energy.gov/sites/prod/files/2015/01/f19/fcto_ myrdd_table_onboard_H2_storage_systems_doe_targets_ ldv.pdf. [2] Toyota. Fuel cell vehicle technology file. 2016. http://www. toyota-global.com/innovation/environmental_technology/ technology_file/fuel_cell_hybrid.html#h306. [3] Stetson NT, Ordaz G, Adams J, Randolph K, McWhorter S. The use of application-specific performance targets and engineering considerations to guide hydrogen storage materials development. J Alloys Compd 2013;580:S333e6. [4] Kim B-J, Park S-J. Optimization of the pore structure of nickel/graphite hybrid materials for hydrogen storage. Int J Hydrogen Energy 2011;36(1):648e53. [5] Voskuilen TG, Pourpoint TL, Dailly AM. Hydrogen adsorption on microporous materials at ambient temperatures and pressures up to 50 MPa. Adsorption 2012;18(3e4):239e49. [6] Thomas KM. Hydrogen adsorption and storage on porous materials. Catal Today 2007;120(3):389e98. [7] Zu¨ttel A, Sudan P, Mauron P, Wenger P. Model for the hydrogen adsorption on carbon nanostructures. Appl Phys A 2004;78(7):941e6. [8] Kim Y-H, Zhao Y, Williamson A, Heben MJ, Zhang S. Nondissociative adsorption of H2 molecules in lightelement-doped fullerenes. Phys Rev Lett 2006;96(1):016102. [9] Lueking AD, Yang RT, Rodriguez NM, Baker RTK. Hydrogen storage in graphite nanofibers: effect of synthesis catalyst and pretreatment conditions. Langmuir 2004;20(3):714e21. [10] Shiraishi M, Takenobu T, Kataura H, Ata M. Hydrogen adsorption and desorption in carbon nanotube systems and its mechanisms. Appl Phys A 2004;78(7):947e53.

[11] Liu C, Chen Y, Wu C-Z, Xu S-T, Cheng H-M. Hydrogen storage in carbon nanotubes revisited. Carbon 2010;48(2):452e5. [12] Zheng Q, Ji X, Gao S, Wang X. Analysis of adsorption equilibrium of hydrogen on graphene sheets. Int J Hydrogen Energy 2013;38(25):10896e902. [13] Wang L, Stuckert NR, Yang RT. Unique hydrogen adsorption properties of graphene. AIChE J 2011;57(10):2902e8. [14] Srinivas G, Zhu Y, Piner R, Skipper N, Ellerby M, Ruoff R. Synthesis of graphene-like nanosheets and their hydrogen adsorption capacity. Carbon 2010;48(3):630e5. [15] Ma L-P, Wu Z-S, Li J, Wu E-D, Ren W-C, Cheng H-M. Hydrogen adsorption behavior of graphene above critical temperature. Int J Hydrogen Energy 2009;34(5):2329e32. [16] Subrahmanyam K, Vivekchand S, Govindaraj A, Rao C. A study of graphenes prepared by different methods: characterization, properties and solubilization. J Mater Chem 2008;18(13):1517e23. [17] Pierotti R, Rouquerol J. Reporting physisorption data for gas/ solid systems with special reference to the determination of surface area and porosity. Pure Appl Chem 1985;57(4):603e19. [18] Anson A, Jagiello J, Parra JB, Sanjuan ML, Benito AM, Maser WK, et al. Porosity, surface area, surface energy, and hydrogen adsorption in nanostructured carbons. J Phys Chem B 2004;108(40):15820e6. [19] Klechikov AG, Mercier G, Merino P, Blanco S, Merino C, Talyzin AV. Hydrogen storage in bulk graphene-related materials. Microporous Mesoporous Mater 2015;210:46e51.  -Beneyto M, Sua  rez-Garcı´a F, Lozano-Castello  D, [20] Jorda  s D, Linares-Solano A. Hydrogen storage on Cazorla-Amoro chemically activated carbons and carbon nanomaterials at high pressures. Carbon 2007;45(2):293e303. [21] Ritschel M, Uhlemann M, Gutfleisch O, Leonhardt A, Graff A, Taschner C, et al. Hydrogen storage in different carbon nanostructures. Appl Phys Lett 2002;80(16):2985e7. [22] Panella B, Hirscher M, Roth S. Hydrogen adsorption in different carbon nanostructures. Carbon 2005;43(10):2209e14. [23] Lin K-Y, Tsai W-T, Chang J-K. Decorating carbon nanotubes with Ni particles using an electroless deposition technique for hydrogen storage applications. Int J Hydrogen Energy 2010;35(14):7555e62. [24] Adams BD, Ostrom CK, Chen S, Chen A. High-performance Pd-based hydrogen spillover catalysts for hydrogen storage. J Phys Chem C 2010;114(46):19875e82. [25] Psofogiannakis GM, Steriotis TA, Bourlinos AB, Kouvelos EP, Charalambopoulou GC, Stubos AK, et al. Enhanced hydrogen storage by spillover on metal-doped carbon foam: an experimental and computational study. Nanoscale 2011;3(3):933e6. [26] Huang C-C, Pu N-W, Wang C-A, Huang J-C, Sung Y, Ger M-D. Hydrogen storage in graphene decorated with Pd and Pt nano-particles using an electroless deposition technique. Sep Purif Technol 2011;82:210e5. [27] Wang Y, Guo CX, Wang X, Guan C, Yang H, Wang K, et al. Hydrogen storage in a NieB nanoalloy-doped threedimensional graphene material. Energy Environ Sci 2011;4(1):195e200.  ski M, Wojcieszak R, Monteverdi S, Mercy M, [28] Zielin Bettahar M. Hydrogen storage on nickel catalysts supported on amorphous activated carbon. Catal Commun 2005;6(12):777e83. [29] Wang L, Yang RT. Hydrogen storage properties of carbons doped with ruthenium, platinum, and nickel nanoparticles. J Phys Chem C 2008;112(32):12486e94. [30] Conner Jr WC, Falconer JL. Spillover in heterogeneous catalysis. Chem Rev 1995;95(3):759e88.

Please cite this article in press as: Pyle DS, et al., Hydrogen storage in carbon nanostructures via spillover, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.08.061

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 6

[31] Conner WC, Pajonk G, Teichner S. Spillover of sorbed species. Adv Catal 1986;34:1e79. [32] Sermon P, Bond G. Hydrogen spillover. Catal Rev 1974;8(1):211e39. [33] Teichner SJ. Recent studies in hydrogen and oxygen spillover and their impact on catalysis. Appl Catal 1990;62(1):1e10. [34] Stuckert NR, Wang L, Yang RT. Characteristics of hydrogen storage by spillover on Pt-doped carbon and catalystbridged metal organic framework. Langmuir 2010;26(14):11963e71. [35] Li Y, Yang RT. Significantly enhanced hydrogen storage in metal-organic frameworks via spillover. J Am Chem Soc 2006;128(3):726e7. [36] Luzan SM, Talyzin A. Hydrogen adsorption in Pt catalyst/ MOF-5 materials. Microporous Mesoporous Mater 2010;135(1):201e5.  n A, Lafuente E, Urriolabeitia E, Navarro R, Benito AM, [37] Anso Maser WK, et al. Hydrogen capacity of palladium-loaded carbon materials. J Phys Chem B 2006;110(13):6643e8. [38] Aksoylu AE, Madalena M, Freitas A, Pereira MFR, Figueiredo JL. The effects of different activated carbon supports and support modifications on the properties of Pt/ AC catalysts. Carbon 2001;39(2):175e85. [39] Amorim C, Keane MA. Palladium supported on structured and nonstructured carbon: a consideration of Pd particle size and the nature of reactive hydrogen. J Colloid Interface Sci 2008;322(1):196e208. [40] Park S-J, Lee S-Y. Hydrogen storage behaviors of platinumsupported multi-walled carbon nanotubes. Int J Hydrogen Energy 2010;35(23):13048e54. [41] Yoo E, Gao L, Komatsu T, Yagai N, Arai K, Yamazaki T, et al. Atomic hydrogen storage in carbon nanotubes promoted by metal catalysts. J Phys Chem B 2004;108(49):18903e7. [42] Zacharia R, Kim KY, Kibria AF, Nahm KS. Enhancement of hydrogen storage capacity of carbon nanotubes via spillover from vanadium and palladium nanoparticles. Chem Phys Lett 2005;412(4):369e75. [43] Chen C-H, Huang C-C. Enhancement of hydrogen spillover onto carbon nanotubes with defect feature. Microporous Mesoporous Mater 2008;109(1):549e59. [44] Suttisawat Y, Rangsunvigit P, Kitiyanan B, Williams M, Ndungu P, Lototskyy M, et al. Investigation of hydrogen storage capacity of multi-walled carbon nanotubes deposited with Pd or V. Int J Hydrogen Energy 2009;34(16):6669e75. [45] Bhowmick R, Rajasekaran S, Friebel D, Beasley C, Jiao L, Ogasawara H, et al. Hydrogen spillover in Pt-single-walled carbon nanotube composites: formation of stable C H bonds. J Am Chem Soc 2011;133(14):5580e6. [46] Zacharia R, Rather S-u, Hwang SW, Nahm KS. Spillover of physisorbed hydrogen from sputter-deposited arrays of platinum nanoparticles to multi-walled carbon nanotubes. Chem Phys Lett 2007;434(4):286e91. [47] Reyhani A, Mortazavi S, Mirershadi S, Moshfegh A, Parvin P, Golikand AN. Hydrogen storage in decorated multiwalled carbon nanotubes by Ca, Co, Fe, Ni, and Pd nanoparticles under ambient conditions. J Phys Chem C 2011;115(14):6994e7001. [48] Wu H, Wexler D, Liu H. Effects of different palladium content loading on the hydrogen storage capacity of doublewalled carbon nanotubes. Int J Hydrogen Energy 2012;37(7):5686e90. [49] Lee YS, Kim YH, Hong JS, Suh JK, Cho GJ. The adsorption properties of surface modified activated carbon fibers for hydrogen storages. Catal Today 2007;120(3):420e5. [50] Jain P, Fonseca DA, Schaible E, Lueking AD. Hydrogen uptake of platinum-doped graphite nanofibers and

[51]

[52]

[53]

[54]

[55]

[56]

[57]

[58]

[59]

[60]

[61]

[62]

[63]

[64]

[65]

[66]

[67]

[68]

[69]

13

stochastic analysis of hydrogen spillover. J Phys Chem C 2007;111(4):1788e800. Lupu D, Biris‚ AR, Mis‚an I, Jianu A, Holzhu¨ter G, Burkel E. Hydrogen uptake by carbon nanofibers catalyzed by palladium. Int J Hydrogen Energy 2004;29(1):97e102. Back C-K, Sandı´ G, Prakash J, Hranisavljevic J. Hydrogen sorption on palladium-doped sepiolite-derived carbon nanofibers. J Phys Chem B 2006;110(33):16225e31. Giasafaki D, Charalambopoulou G, Bourlinos A, Stubos A, Gournis D, Steriotis T. A hydrogen sorption study on a Pddoped CMK-3 type ordered mesoporous carbon. Adsorption 2013:1e9. Zlotea C, Cuevas F, Paul-Boncour V, Leroy E, Dibandjo P, Gadiou R, et al. Size-dependent hydrogen sorption in ultrasmall Pd clusters embedded in a mesoporous carbon template. J Am Chem Soc 2010;132(22):7720e9. Yang Y-X, Bourgeois L, Zhao C, Zhao D, Chaffee A, Webley PA. Ordered micro-porous carbon molecular sieves containing well-dispersed platinum nanoparticles for hydrogen storage. Microporous Mesoporous Mater 2009;119(1):39e46. Campesi R, Cuevas F, Gadiou R, Leroy E, Hirscher M, VixGuterl C, et al. Hydrogen storage properties of Pd nanoparticle/ carbon template composites. Carbon 2008;46(2):206e14. ndez J, Pis J, Arenillas A. Improving Zubizarreta L, Mene hydrogen storage in Ni-doped carbon nanospheres. Int J Hydrogen Energy 2009;34(7):3070e6. Saha D, Deng S. Hydrogen adsorption on Pd-and Ru-doped C60 fullerene at an ambient temperature. Langmuir 2011;27(11):6780e6. Chen C-H, Chung T-Y, Shen C-C, Yu M-S, Tsao C-S, Shi G-N, et al. Hydrogen storage performance in palladium-doped graphene/carbon composites. Int J Hydrogen Energy 2013;38(9):3681e8. Vinayan B, Sethupathi K, Ramaprabhu S. Facile synthesis of triangular shaped palladium nanoparticles decorated nitrogen doped graphene and their catalytic study for renewable energy applications. Int J Hydrogen Energy 2013;38(5):2240e50. Vinayan B, Sethupathi K, Ramaprabhu S. Hydrogen storage studies of palladium decorated nitrogen doped graphene nanoplatelets. J Nanosci Nanotechnol 2012;12(8):6608e14. Parambhath VB, Nagar R, Ramaprabhu S. Effect of nitrogen doping on hydrogen storage capacity of palladium decorated graphene. Langmuir 2012;28(20):7826e33. Blach TP, Gray EM. Sieverts apparatus and methodology for accurate determination of hydrogen uptake by light-atom hosts. J Alloys Compd 2007;446:692e7. Broom D. The accuracy of hydrogen sorption measurements on potential storage materials. Int J Hydrogen Energy 2007;32(18):4871e88. Zlotea C, Moretto P, Steriotis T. A Round Robin characterisation of the hydrogen sorption properties of a carbon based material. Int J Hydrogen Energy 2009;34(7):3044e57. Lin Y, Ding F, Yakobson BI. Hydrogen storage by spillover on graphene as a phase nucleation process. Phys Rev B 2008;78(4):041402. Lueking AD, Yang RT. Hydrogen spillover to enhance hydrogen storagedstudy of the effect of carbon physicochemical properties. Appl Catal A Gen 2004;265(2):259e68. Lachawiec AJ, Qi G, Yang RT. Hydrogen storage in nanostructured carbons by spillover: bridge-building enhancement. Langmuir 2005;21(24):11418e24. Johnson ER, Wolkow RA, DiLabio GA. Application of 25 density functionals to dispersion-bound homomolecular dimers. Chem Phys Lett 2004;394(4):334e8.

Please cite this article in press as: Pyle DS, et al., Hydrogen storage in carbon nanostructures via spillover, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.08.061

14

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 6

[70] Bonfanti M, Martinazzo R, Tantardini GF, Ponti A. Physisorption and diffusion of hydrogen atoms on graphite from correlated calculations on the H-coronene model system. J Phys Chem C 2007;111(16):5825e9. [71] Sha X, Knippenberg MT, Cooper AC, Pez GP, Cheng H. Dynamics of hydrogen spillover on carbon-based materials. J Phys Chem C 2008;112(44):17465e70. [72] Psofogiannakis GM, Froudakis GE. DFT study of the hydrogen spillover mechanism on Pt-doped graphite. J Phys Chem C 2009;113(33):14908e15. [73] Zecho T, Gu¨ttler A, Sha X, Jackson B, Ku¨ppers J. Adsorption of hydrogen and deuterium atoms on the (0001) graphite surface. J Chem Phys 2002;117:8486. [74] Jeloaica L, Sidis V. DFT investigation of the adsorption of atomic hydrogen on a cluster-model graphite surface. Chem Phys Lett 1999;300(1):157e62. [75] Casolo S, Løvvik OM, Martinazzo R, Tantardini GF. Understanding adsorption of hydrogen atoms on graphene. J Chem Phys 2009;130:054704. [76] Kerwin J, Jackson B. The sticking of H and D atoms on a graphite (0001) surface: the effects of coverage and energy dissipation. J Chem Phys 2008;128:084702. [77] Chen L, Cooper AC, Pez GP, Cheng H. Mechanistic study on hydrogen spillover onto graphitic carbon materials. J Phys Chem C 2007;111(51):18995e9000. [78] Yang FH, Lachawiec AJ, Yang RT. Adsorption of spillover hydrogen atoms on single-wall carbon nanotubes. J Phys Chem B 2006;110(12):6236e44. [79] Kostov M, Cheng H, Cooper A, Pez G. Influence of carbon curvature on molecular adsorptions in carbon-based materials: a force field approach. Phys Rev Lett 2002;89(14):146105. [80] Liu XM, Tang Y, Xu ES, Fitzgibbons TC, Larsen GS, Gutierrez HR, et al. Evidence for ambient-temperature reversible catalytic hydrogenation in Pt-doped carbons. Nano Lett 2012;13(1):137e41. [81] Xiao J, Wei J. Diffusion mechanism of hydrocarbons in zeolitesdI. Theory. Chem Eng Sci 1992;47(5):1123e41. [82] Pereira VM, Guinea F, Dos Santos JL, Peres N, Neto AC. Disorder induced localized states in graphene. Phys Rev Lett 2006;96(3):036801. [83] Psofogiannakis GM, Froudakis GE. Fundamental studies and perceptions on the spillover mechanism for hydrogen storage. Chem Commun 2011;47(28):7933e43. [84] Balog R, Jørgensen B, Wells J, Lægsgaard E, Hofmann P, Besenbacher F, et al. Atomic hydrogen adsorbate structures on graphene. J Am Chem Soc 2009;131(25):8744e5.   Xu W, Otero R, Rauls E, anin Z, [85] Hornekær L, Sljivan c Stensgaard I, et al. Metastable structures and recombination pathways for atomic hydrogen on the graphite (0001) surface. Phys Rev Lett 2006;96(15):156104. [86] Sofo JO, Chaudhari AS, Barber GD. Graphane: a twodimensional hydrocarbon. Phys Rev B 2007;75(15):153401. [87] Greeley J, Mavrikakis M. Surface and subsurface hydrogen: adsorption properties on transition metals and nearsurface alloys. J Phys Chem B 2005;109(8):3460e71. [88] Tew MW, Miller JT, van Bokhoven JA. Particle size effect of hydride formation and surface hydrogen adsorption of nanosized palladium catalysts: L3 edge vs K edge X-ray absorption spectroscopy. J Phys Chem C 2009;113(34):15140e7. [89] Takagi H, Hatori H, Yamada Y, Matsuo S, Shiraishi M. Hydrogen adsorption properties of activated carbons with modified surfaces. J Alloys Compd 2004;385(1):257e63. [90] Vinayan B, Sethupathi K, Ramaprabhu S. Investigations of hydrogen storage in palladium decorated graphene nanoplatelets. Trans Indian Inst Metals 2011;64(1e2):169e73.

[91] Wang L, Yang RT. Molecular hydrogen and spiltover hydrogen storage on high surface area carbon sorbents. Carbon 2012;50(9):3134e40. [92] Contescu CI, Van Benthem K, Li S, Bonifacio CS, Pennycook SJ, Jena P, et al. Single Pd atoms in activated carbon fibers and their contribution to hydrogen storage. Carbon 2011;49(12):4050e8. [93] Geng Z, Wang D, Zhang C, Zhou X, Xin H, Liu X, et al. Spillover enhanced hydrogen uptake of Pt/Pd doped corncob-derived activated carbon with ultra-high surface area at high pressure. Int J Hydrogen Energy 2014;39(25):13643e9. [94] Giasafaki D, Charalambopoulou G, Tampaxis C, Stubos A, Steriotis T. Hydrogen sorption properties of Pd-doped carbon molecular sieves. Int J Hydrogen Energy 2014;39(18):9830e6. [95] Dibandjo P, Zlotea C, Gadiou R, Ghimbeu CM, Cuevas F, Latroche M, et al. Hydrogen storage in hybrid nanostructured carbon/palladium materials: influence of particle size and surface chemistry. Int J Hydrogen Energy 2013;38(2):952e65.  ski M, Wojcieszak R, Monteverdi S, Mercy M, [96] Zielin Bettahar M. Hydrogen storage in nickel catalysts supported on activated carbon. Int J Hydrogen Energy 2007;32(8):1024e32. [97] Nieuwenhuys BE, Hagen D, Rovida G, Somorjai G. LEED, AES and thermal desorption studies of chemisorbed Hydrogen and hydrocarbons (C2H2, C2H4, C6H6, C6H12) on the (111) and stepped [6(111)(100)] iridium crystal surfaces; comparison with platinum. Surf Sci 1976;59(1):155e76. [98] Verheij LK, Hugenschmidt MB, Anton AB, Poelsema B, Comsa G. A molecular beam study of the interaction between hydrogen and the Pt (111) surface. Surf Sci 1989;210(1):1e26. [99] Papoian G, Nørskov JK, Hoffmann R. A comparative theoretical study of the hydrogen, methyl, and ethyl chemisorption on the Pt (111) surface. J Am Chem Soc 2000;122(17):4129e44. [100] Johnson S, Madix R. Desorption of hydrogen and carbon monoxide from Ni (100), Ni (100) p (2 2) S, and Ni (100) c (2 2) S surfaces. Surf Sci 1981;108(1):77e98. [101] Christmann K, Schober O, Ertl G, Neumann M. Adsorption of hydrogen on nickel single crystal surfaces. J Chem Phys 2003;60(11):4528e40. [102] Conrad H, Ertl G, Koch J, Latta E. Adsorption of CO on Pd single crystal surfaces. Surf Sci 1974;43(2):462e80. [103] Dag S, Ozturk Y, Ciraci S, Yildirim T. Adsorption and dissociation of hydrogen molecules on bare and functionalized carbon nanotubes. Phys Rev B 2005;72(15):155404.  pez-Corral I, Germa  n EA, Juan A, Volpe MAA, Brizuela GP. [104] Lo DFT study of hydrogen adsorption on palladium decorated graphene. J Phys Chem C 2011;115(10):4315e23. [105] Dai D, Liao D, Balasubramanian K. Potential energy surfaces for Pt3þ H2 and Pd3þ H2 systems. J Chem Phys 1995;102(19):7530e9. [106] Balasubramanian K. Potential energy surfaces for the Pt2þ H2 reaction. J Chem Phys 1991;94(2):1253e63. [107] Okamoto Y. Density-functional calculations of icosahedral M13 (M¼ Pt and Au) clusters on graphene sheets and flakes. Chem Phys Lett 2006;420(4):382e6. [108] Liu X, Dilger H, Eichel R, Kunstmann J, Roduner E. A small paramagnetic platinum cluster in an NaY zeolite: characterization and hydrogen adsorption and desorption. J Phys Chem B 2006;110(5):2013e23. [109] Richter LJ, Ho W. Vibrational spectroscopy of H on Pt (111): evidence for universally soft parallel modes. Phys Rev B 1987;36(18):9797.

Please cite this article in press as: Pyle DS, et al., Hydrogen storage in carbon nanostructures via spillover, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.08.061

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 6

[110] Gdowski G, Fair J, Madix R. Reactive scattering of small molecules from platinum crystal surfaces: D2CO, CH3OH, HCOOH, and the nonanomalous kinetics of hydrogen atom recombination. Surf Sci 1983;127(3):541e54. [111] Rhodin TN, Ertl G. The nature of the surface chemical bond. North-Holland Publishing Company; 1979. [112] Ho W, DiNardo N, Plummer E. Angle-resolved and variable impact energy electron vibrational excitation spectroscopy of molecules adsorbed on surfaces. J Vac Sci Technol 1980;17(1):134e40. [113] Feulner P, Menzel D. The adsorption of hydrogen on ruthenium (001): adsorption states, dipole moments and kinetics of adsorption and desorption. Surf Sci 1985;154(2):465e88. [114] Zhou C, Wu J, Nie A, Forrey RC, Tachibana A, Cheng H. On the sequential hydrogen dissociative chemisorption on small platinum clusters: a density functional theory study. J Phys Chem C 2007;111(34):12773e8. [115] Lee H-W, Chang C-M. Size effect of Pd clusters on hydrogen adsorption. J Phys Condens Matter 2011;23(4):045503. ndez J, Job N, Marco-Lozar J, Pirard J-P, [116] Zubizarreta L, Mene Pis J, et al. Ni-doped carbon xerogels for H2 storage. Carbon 2010;48(10):2722e33. [117] Kim H-S, Lee H, Han K-S, Kim J-H, Song M-S, Park M-S, et al. Hydrogen storage in Ni nanoparticle-dispersed multiwalled carbon nanotubes. J Phys Chem B 2005;109(18):8983e6. [118] Mei-Yan N, Xian-Long W, Zhi Z. Interaction of hydrogen molecules on Ni-doped single-walled carbon nanotube. Chin Phys B 2009;18(1):357. [119] Wu J, Ong SW, Kang HC, Tok ES. Hydrogen adsorption on mixed platinum and nickel nanoclusters: the influence of cluster composition and graphene support. J Phys Chem C 2010;114(49):21252e61. [120] Kubas Gregory J. Molecular hydrogen complexes: coordination of a s bond to transition metals. Acc Chem Res 1988;21(3):120e8. [121] Niu J, Rao B, Jena P. Binding of hydrogen molecules by a transition-metal ion. Phys Rev Lett 1992;68(15):2277. [122] Zheng Q, Wang X, Gao S. Adsorption equilibrium of hydrogen on graphene sheets and activated carbon. Cryogenics 2014;61:143e8. [123] Baburin IA, Klechikov A, Mercier G, Talyzin A, Seifert G. Hydrogen adsorption by perforated graphene. Int J Hydrogen Energy 2015;40(20):6594e9. [124] Cunning BV, Pyle DS, Merritt CR, Brown CL, Webb CJ, Gray EM. Hydrogen adsorption characteristics of magnesium combustion derived graphene at 77 and 293 K. Int J Hydrogen Energy 2014;39(12):6783e8. [125] Lachawiec Jr AJ, Yang RT. Isotope tracer study of hydrogen spillover on carbon-based adsorbents for hydrogen storage. Langmuir 2008;24(12):6159e65. [126] Li Y, Yang RT. Hydrogen storage on platinum nanoparticles doped on superactivated carbon. J Phys Chem C 2007;111(29):11086e94. [127] Wang L, Yang FH, Yang RT. Hydrogen storage properties of B-and N-doped microporous carbon. AIChE J 2009;55(7):1823e33. [128] Giraudet S, Zhu Z. Hydrogen adsorption in nitrogen enriched ordered mesoporous carbons doped with nickel nanoparticles. Carbon 2011;49(2):398e405.  n A, Lafuente E, Urriolabeitia E, Navarro R, Benito AM, [129] Anso Maser WK, et al. Preparation of palladium loaded carbon nanotubes and activated carbons for hydrogen sorption. J Alloys Compd 2007;436(1):294e7. sar C, [130] Zhao W, Fierro V, Zlotea C, Izquierdo M, Chevalier-Ce Latroche M, et al. Activated carbons doped with Pd nanoparticles for hydrogen storage. Int J Hydrogen Energy 2012;37(6):5072e80.

15

[131] Tsao C-S, Liu Y, Chuang H-Y, Tseng H-H, Chen T-Y, Chen CH, et al. Hydrogen spillover effect of Pt-doped activated carbon studied by inelastic neutron scattering. J Phys Chem Lett 2011;2(18):2322e5. [132] Huang S-Y, Huang C-D, Chang B-T, Yeh C-T. Chemical activity of palladium clusters: sorption of hydrogen. J Phys Chem B 2006;110(43):21783e7. [133] Christmann K, Ertl G, Pignet T. Adsorption of hydrogen on a Pt (111) surface. Surf Sci 1976;54(2):365e92. [134] Li Y, Yang RT, Liu C-j, Wang Z. Hydrogen storage on carbon doped with platinum nanoparticles using plasma reduction. Ind Eng Chem Res 2007;46(24):8277e81. [135] Contescu CI, Brown CM, Liu Y, Bhat VV, Gallego NC. Detection of hydrogen spillover in palladium-modified activated carbon fibers during hydrogen adsorption. J Phys Chem C 2009;113(14):5886e90. [136] Parambhath VB, Nagar R, Sethupathi K, Ramaprabhu S. Investigation of spillover mechanism in palladium decorated hydrogen exfoliated functionalized graphene. J Phys Chem C 2011;115(31):15679e85. [137] Chen L, Pez G, Cooper AC, Cheng H. A mechanistic study of hydrogen spillover in MoO3 and carbon-based graphitic materials. J Phys Condens Matter 2008;20(6):064223. [138] Hashimoto A, Suenaga K, Gloter A, Urita K, Iijima S. Direct evidence for atomic defects in graphene layers. Nature 2004;430(7002):870e3.  mez-Navarro C, Meyer JC, Sundaram RS, Chuvilin A, [139] Go Kurasch S, Burghard M, et al. Atomic structure of reduced graphene oxide. Nano Lett 2010;10(4):1144e8. [140] Harris PJ, Liu Z, Suenaga K. Imaging the atomic structure of activated carbon. J Phys Condens Matter 2008;20(36):362201. [141] 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(17):173102. [142] Xu C, Wang X, Zhu J. Graphene metal particle nanocomposites. J Phys Chem C 2008;112(50):19841e5. [143] Chen X, Wu G, Chen J, Chen X, Xie Z, Wang X. Synthesis of “clean” and well-dispersive Pd nanoparticles with excellent electrocatalytic property on graphene oxide. J Am Chem Soc 2011;133(11):3693e5. [144] Li Y, Fan X, Qi J, Ji J, Wang S, Zhang G, et al. Palladium nanoparticle-graphene hybrids as active catalysts for the Suzuki reaction. Nano Res 2010;3(6):429e37. [145] Hassan HM, Abdelsayed V, Abd El Rahman SK, AbouZeid KM, Terner J, El-Shall MS, et al. Microwave synthesis of graphene sheets supporting metal nanocrystals in aqueous and organic media. J Mater Chem 2009;19(23):3832e7. [146] Wang Z, Yang FH, Yang RT. Enhanced hydrogen spillover on carbon surfaces modified by oxygen plasma. J Phys Chem C 2010;114(3):1601e9. [147] Li Q, Lueking AD. Effect of surface oxygen groups and water on hydrogen spillover in Pt-doped activated carbon. J Phys Chem C 2011;115(10):4273e82. [148] Wang L, Yang FH, Yang RT, Miller MA. Effect of surface oxygen groups in carbons on hydrogen storage by spillover. Ind Eng Chem Res 2009;48(6):2920e6. [149] Dreyer DR, Park S, Bielawski CW, Ruoff RS. The chemistry of graphene oxide. Chem Soc Rev 2010;39(1):228e40. [150] Gao W, Alemany LB, Ci L, Ajayan PM. New insights into the structure and reduction of graphite oxide. Nat Chem 2009;1(5):403e8. [151] Bianco A, Cheng H-M, Enoki T, Gogotsi Y, Hurt RH, Koratkar N, et al. All in the graphene family-a recommended nomenclature for two-dimensional carbon materials. Carbon 2013;65:1e6. [152] McAllister MJ, Li J-L, Adamson DH, Schniepp HC, Abdala AA, Liu J, et al. Single sheet functionalized graphene by

Please cite this article in press as: Pyle DS, et al., Hydrogen storage in carbon nanostructures via spillover, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.08.061

16

[153]

[154]

[155]

[156]

[157]

[158]

[159]

[160]

[161]

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 6

oxidation and thermal expansion of graphite. Chem Mater 2007;19(18):4396e404. Wu Z-S, Ren W, Gao L, Liu B, Jiang C, Cheng H-M. Synthesis of high-quality graphene with a pre-determined number of layers. Carbon 2009;47(2):493e9. Marcano DC, Kosynkin DV, Berlin JM, Sinitskii A, Sun Z, Slesarev A, et al. Improved synthesis of graphene oxide. ACS Nano 2010;4(8):4806e14. Hernandez Y, Nicolosi V, Lotya M, Blighe FM, Sun Z, De S, et al. High-yield production of graphene by liquid-phase exfoliation of graphite. Nat Nanotechnol 2008;3(9):563e8. Zhang L, Liang J, Huang Y, Ma Y, Wang Y, Chen Y. Sizecontrolled synthesis of graphene oxide sheets on a large scale using chemical exfoliation. Carbon 2009;47(14):3365e8. Park S, An J, Potts JR, Velamakanni A, Murali S, Ruoff RS. Hydrazine-reduction of graphite-and graphene oxide. Carbon 2011;49(9):3019e23. Eswaraiah V, Jyothirmayee Aravind SS, Ramaprabhu S. Top down method for synthesis of highly conducting graphene by exfoliation of graphite oxide using focused solar radiation. J Mater Chem 2011;21(19):6800e3. Liu J, Notarianni M, Will G, Tiong VT, Wang H, Motta N. Electrochemically exfoliated graphene for electrode films: effect of graphene flake thickness on the sheet resistance and capacitive properties. Langmuir 2013;29(43):13307e14.  Zhen S, Koka  Poh C, Leea  Yeow EK, Gopala  Sahoo N. A Gieka green approach to the synthesis of high-quality graphene oxide flakes via electrochemical exfoliation of pencil core. RSC Adv 2013;3(29):11745e50. Su C-Y, Lu A-Y, Xu Y, Chen F-R, Khlobystov AN, Li L-J. Highquality thin graphene films from fast electrochemical exfoliation. ACS Nano 2011;5(3):2332e9.

[162] Choucair M, Thordarson P, Stride JA. Gram-scale production of graphene based on solvothermal synthesis and sonication. Nat Nanotechnol 2008;4(1):30e3. [163] Wang H, Robinson JT, Li X, Dai H. Solvothermal reduction of chemically exfoliated graphene sheets. J Am Chem Soc 2009;131(29):9910e1. [164] Zhao W, Fang M, Wu F, Wu H, Wang L, Chen G. Preparation of graphene by exfoliation of graphite using wet ball milling. J Mater Chem 2010;20(28):5817e9. [165] Siamaki AR, Khder AERS, Abdelsayed V, El-Shall MS, Gupton BF. Microwave-assisted synthesis of palladium nanoparticles supported on graphene: a highly active and recyclable catalyst for carbonecarbon cross-coupling reactions. J Catal 2011;279(1):1e11. [166] Zhu Y, Murali S, Stoller MD, Velamakanni A, Piner RD, Ruoff RS. Microwave assisted exfoliation and reduction of graphite oxide for ultracapacitors. Carbon 2010;48(7):2118e22. [167] Lueking AD, Psofogiannakis G, Froudakis GE. Atomic hydrogen diffusion on doped and chemically modified graphene. J Phys Chem C 2013;117(12):6312e9. [168] Psofogiannakis GM, Froudakis GE. DFT study of hydrogen storage by spillover on graphite with oxygen surface groups. J Am Chem Soc 2009;131(42):15133e5. [169] Wu H-Y, Fan X, Kuo J-L, Deng W-Q. DFT Study of hydrogen storage by spillover on graphene with boron substitution. J Phys Chem C 2011;115(18):9241e9. [170] Casartelli M, Casolo S, Tantardini GF, Martinazzo R. Structure and stability of hydrogenated carbon atom vacancies in graphene. Carbon 2014;77:165e74. [171] Krasheninnikov A, Nieminen R. Attractive interaction between transition-metal atom impurities and vacancies in graphene: a first-principles study. Theor Chem Acc 2011;129(3e5):625e30.

Please cite this article in press as: Pyle DS, et al., Hydrogen storage in carbon nanostructures via spillover, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.08.061