Ni-dispersed fullerenes: Hydrogen storage and desorption properties

Weon Ho Shin and Seong Ho Yang
Department of Materials Science and Engineering, Korea Advanced Institute of Science and Technology, Kusung-dong 373-1, Yusung-gu, Daejon, Republic of Korea 305-701
William A. Goddard III
Materials and Process Simulation Center, California Institute of Technology, Pasadena, California 91125
Jeung Ku Kanga)
Department of Materials Science and Engineering, Korea Advanced Institute of Science and Technology, Kusung-dong 373-1, Yusung-gu, Daejon, Republic of Korea 305-701
(Received 10 October 2005; accepted 14 December 2005; published online 31 January 2006)

Our study shows that the H2 storage media using Ni-dispersed fullerenes could be viable alternatives to reversible hydrogen storage. It is demonstrated that a single Ni coated on the fullerene surface can store up to three H2 molecules. Consequently, at high Ni coverage, Ni-dispersed fullerenes are considered to be the novel hydrogen storage media capable of storing —6.8 wt % H2, thus exceeding the Department of Energy target (6.5 wt %) for automobile applications. Moreover, the H2 desorption activation barrier of 11.8 kcal/molH2 is ideal for many practical hydrogen storage applications. © 2006 American Institute of Physics. [DOI: 10.1063/1.2168775]

Hydrogen, which can be readily produced from renewable energy sources and mainly produces water as a byproduct, contains much larger chemical energy per mass (142 MJ kg-1) (Ref. 1) than any hydrocarbon fuel. Consequently, it is considered as an ideal fuel to solve energy and environmental problems in the earth that result from using petroleum fuels. Many hydrogen storage methods have been proposed, such as carbon-fiber reinforced high-strength containers, liquid hydrogen, chemical hydrides, and carbon nanotubes (CNTs). In particular, CNT has been suggested as the most promising nanostructure due to its cylindrical structure and high surface-to-volume ratio, which make it ideal for hydrogen storage. However, recent studies indicate that the structure of a pristine CNT is not relevant to satisfy the Department of Energy (DOE) target (6.5 wt %) at ambient conditions ideal for many practical applications. This is because the pristine CNT has only the —0.8 wt % hydrogen available at ambient conditions. However, we propose the use of nickel-doped fullerenes as the hydrogen storage media capable of exceeding the DOE target (6.5 wt %) at ambient conditions.

The recent experiment proved that Ni-doped multi-walled (MW) CNTs with outer diameters of —20 nm could release 2.8 wt % H2 at ambient conditions. We found that each Ni dispersed on the CNT surfaces has the high affinity for hydrogen by storing up to five H2 molecules at ambient conditions. On the other hand, it was also determined that the insertion of H2 into the interlayer of MWCNTs is thermody-namically unstable since it is an endothermic reaction by 24.9 kcal/mol H2. Consequently, it is considered that Ni-dispersed MWCNTs could release over the 3 wt % if Ni nanoparticles could be dispersed on the surfaces of the CNTs grown in vertical shapes with reduced numbers of inter-layers. However, the problem is that not all the techniques to control the growth directions and interlayers are available yet. On the other hand, unlike the CNT, a fullerene has al ways a single wall and its surface is isotropic; thus there is no need to reduce its interlayer and growth direction. Consequently, the fullerene is considered as a better nanostructure for Ni dispersion to maximize the hydrogen storage capacity than that for the CNT.

Recent study by Barajas-Barraza et al. showed that up to 19 H2 molecules could be stored in a C110 fullerene structure. Moreover, Narita et al. showed through a molecular dynamic simulation study that H2 is stable inside the C60 cage at ambient conditions. On the other hand, they also found that H2 could not escape from the cage due to the high activation barrier of 369.1 kcal/mol H2 for penetration through the cage; thus, using the inside of the cage as a hydrogen storage site is considered as impractical for many applications. On the other hand, the recent study showed that Ni nanoparticles dispersed give the activation barrier of 8.7 kcal/mol relevant for hydrogen desorption at ambient conditions. In addition, a theoretical study by Yildirim et al. also found that metals-dispersed nanostructures could be good hydrogen storage materials for practical applications.

To determine electronic wave functions, we use B3LYP (Ref. 10) and KMLYP (Ref. 11) density functional theory (DFT) methods with a split basis set of 6-31 G valence double zeta basis sets for Ni and on fullerene carbon atoms and 6-31 G split valence double zeta basis sets for hydrogen. The B3LYP is shown to accurately predict geometries and thermochemical data compared to generalized gradient    approximation           (GGA)    and    local    density approximation (LDA), ,1 while the KMLYP has been proven to be more accurate in predicting transition state barriers than other DFT methods. In addition, here our energies include zero point-energy (ZPE) corrections determined at fully optimized geometry with the same basis sets. We have chosen a C60 molecule with a diameter of 7.12 Е as the specific system for our study, as shown in Fig. 1. The fullerene has the two distinct bond lengths of 1.46 Е for C -C bonds and 1.40 Е for C=C bonds, which compare to the experimental values of 1.43 Е and 1.39 Е,   respectively.

First, we have explored the hydrogen storage capacity on the pristineC60 . The physical adsorption reaction by a H2 molecule on the outside surface of the fullerene was found to be unstable as it is endothermic by 0.7 kcal/mol H2, while its corresponding chemical adsorption reaction was determined to have the exothermic enthalpy of 15.2 kcal/mol H2. However, it was determined that chemical adsorption by H2 molecules on pristine fullerenes has a high activation energy of 83.1 kcal/mol H2. Consequently, using the computed partition functions for the reactants and the transition state, the reaction rate k of the canonical rate equation21 is determined by

where Г (Г) is the thermal tunneling coefficient, <2ts is the partition functional for the transition state, QA and QB are the partition functions for the reactants A and B, respectively, and AE0 is the barrier height. The resulting dependence of chemisorption   on   pristine   C60   is   given   by   Ј=1.76 X 1017 exp (-83.1kcal/mol/^r)(/? = 2.0 kcal/mol K). In this respect, chemisorption on a pristine fullerene should occur at a very high temperature instead of at room temperature, which is consistent with the experimental results of Loutfy et

Contrary to our previous study on the Ni-dispersed CNTs, our theoretical calculations find that each Ni atom dispersed on a fullerene could store up to three H2 molecules at temperatures up to room temperature, where spin-polarization effects were completely considered. Recently, Zhao et al. have also studied hydrogen storage properties on nickel-doped fullerenes. On the other hand, it was assumed that the metal is the most stable on the center of the pentagonal rings of C60. However, our first-principles calculations prove that the most stable sites for metal dispersion are edge sites between hexagonal rings instead of their assumed center of pentagonal sites. In fact, there are the four distinct sites for Ni dispersion in a C60, as shown in Fig. 1: (a) the center of a pentagonal ring, (b) the center of a hexagonal ring, (c) the edge between hexagonal and pentagonal rings, and (d) the edge between two hexagonal rings. The Gibbs energy of AG for Ni adsorption is calculated by a following equation:

where AH and AS are the enthalpy and entropy changes, respectively. The lowest enthalpy structure is found to be for the case (d), with the case (c) 10.4 kcal/mol, the case (a) 15.7 kcal/mol, and the case (b) 16.6 kcal/mol higher. Consequently, this indicates that the case (d) corresponds to the global minimum state. On the other hand, the largest diffusion barrier of Ni diffusion from the local minima in the above to the global minimum state (d) is determined to be very small (6.3 kcal/mol), thus implying that Ni in the local minimum states could easily transform to the global minimum state even at the very low temperatures larger than 109 K. In addition, the entropy difference of 0.03 kcal/mol K between the global minimum (d) and the dissociated state into a fullerene plus Ni atoms finds that the global minimum state for a Ni-dispersed fullerene could be stable up to 1176 K. The exothermic adsorption energy for Ni at the most stable position of (d) is 37.6 kcal/mol, and the distances be-

tween Ni atoms and their nearest C atom are 1.82 Е. A C60 includes two different CuC and CvC bond characters. A CvC bond site in the edge between two hexagonal ring sites is found to be more favorable for metal dispersion.

In addition, we find that the Mulliken charge on Ni is 0.51. Consequently, the resulting dipole moment of 5.15 D from a Ni-C60 cluster, compared to the case of 0.002 D for the pristine C60, is sufficient to enhance the van der Waal’s interaction between hydrogen and Ni-C60.

We also determine the Ni adsorption energies of -47.5 kcal/mol Ni and -54.5 kcal/mol Ni for Ni2 and Ni3 nanoparticles, respectively. All Ni atoms in these cases were also found to be most stable at the edges of the same hexagonal ring sites, as in the case of the first Ni. The C-Ni and Ni-Ni bond distances for C60-Ni2 are determined to be 1.82 and 2.52 Е, while the CuNi bond and Ni-Ni bond distances for C60-Ni3 are 1.88 and 2.53 Е, respectively.

Figure 2 shows the hydrogen adsorption behaviors as hydrogen molecules adsorbed in the C60-Ni1 cluster increase. The physisorption energy is calculated by a following equation:

where Htot(H2+fullerene) is the enthalpy of the fullerene-H2 system at 298 K, Htot (fullerene) is the enthalpy for the fullerene, and Htot(H2) is the enthalpy for an optimized H2 molecule. The binding enthalpy change of the first hydrogen molecule adsorbed by the nickel atom is determined to be 18.5 kcal/mol, which is lower than the result of Zhao et al. by 31.8 kcal/mol. Zhao et al. have reported that Ni could store only one hydrogen molecule. However, our calculations show that Ni is more likely to store up to three hydrogen molecules rather than only one. This discrepancy appears to be reasonable. Zhao et al. assumed that Ni is the most stable at the center of a pentagonal ring in the fullerene, while our results show the edge between two hexagonal rings for the most stable site of the Ni atom, as mentioned above. Moreover, we find that three hydrogen molecules could be adsorbed around each Ni with its exothermic enthalpy of
11.8 kcal/mol H2. It should be noted that this H2-dispersed fullerene is in a global minimum state. The bond distance between hydrogen and Ni atoms ranges from 1.62 to 1.73 Е, while H—H bonds slightly elongate with bond distances ranging from 0.79 to 0.81 Е compared to the original H-H bond distance of 0.73 Е, which indicates that the bonds between hydrogen and Ni have both physical and chemical bond characters. In C60-Ni1, the dipole moment is calculated to be 4.40 D. In the case of C60-Ni2, it is also determined that three hydrogen molecules could be adsorbed around each Ni atom. The adsorption energy for this case is determined to be -9.7 kcal/mol H2, and the bond lengths between hydrogen and Ni atoms range from 1.61 to 1.76 Е. In the C60-Ni3 cluster, each Ni atom also stores three hydrogen molecules with the enthalpy change of -7.6 kcal/mol H2 and the bond distances ranging from 1.60 to 1.77 Е. Consequently, these results indicate that the Ni-dispersed C60 could store 90 hydrogen molecules attached to 30 Ni atoms, as shown in Fig. 3. Thus, this results in the hydrogen storage capacity is up to 6.8 wt %.

In conclusion, it was found that each Ni atom adsorbed three hydrogen molecules. These results indicate that a Ni-dispersed fullerene could have 6.8 wt % for its hydrogen storage capacity, which exceeds the DOE target. This enhanced hydrogen storage capacity is attributed to the fact that three H2 could be adsorbed around each polarized Ni involving the charge transfer from fullerenes.