Hydrogen storage in a Ni–B nanoalloy-doped three-dimensional graphene material

Abstract

A three-dimensional graphene material was doped with Ni–B nanoalloys via a chemical reduction process. The graphene doped with Ni (0.83 wt%) and B (1.09 wt%) shows the best hydrogen storage capacity of 4.4 wt% at 77 K and 106 kPa, which is better than that of pristine graphene by three times, and is also excellent among all carbon-based materials. Analysis of hydrogen isotherms and isosteric heat of adsorption suggests that doping with an appropriate amount of Ni–B nanoalloys can result in the dissociative chemisorption of hydrogen molecules by spillover to achieve the high hydrogen storage capacity.

---

Broader context

Hydrogen is regarded as a green source of energy, because it can be regenerated from renewable sources and is non-polluting. However, efficient and safe storage of hydrogen is crucial for promoting the “hydrogen economy”. In this study, a three-dimensional graphene material was doped with Ni–B nanoalloys via a chemical reduction process. The graphene doped with Ni (0.83 wt%) and B (1.09 wt%) shows the best hydrogen storage capacity of 4.4 wt% at 77 K and 106 kPa, which is better than that of pristine graphene by three times, and is also excellent among all carbon-based materials. Analysis of hydrogen isotherms and isosteric heat of adsorption suggests that doping with an appropriate amount of Ni–B nanoalloys can result in the dissociative chemisorption of hydrogen molecules by spillover to achieve the high hydrogen storage capacity. According to the results, it is shown that the Ni–B nanoalloy-doped three-dimensional graphene material may be a promising material for hydrogen storage.

1. Introduction

The rising population and increasing demand for energy supply have urged us to explore more sustainable energy resources. Hydrogen is regarded as a green source of energy, because it can be generated from renewable sources and is non-polluting. However, efficient and safe storage of hydrogen is crucial for promoting the “hydrogen economy”.^1,2 In recent decades, many types of hydrogen storage materials have been developed and investigated, which include hydrogen storage alloys,^3–6 metal nitrides and imides,⁷ammonia borane,⁸ and porous materials such as carbon nanotubes (CNTs),⁹zeolites,¹⁰ and metal–organic framework (MOF) materials.¹¹ Among these materials, carbon-based materials have received particular research interest due to their light weight, high surface areas and chemical stabilities.^12–14

Hydrogen spillover is the dissociative chemisorption of hydrogen on metal nanoparticles, followed by the subsequent migration of hydrogen atoms onto adjacent surface sites of a receptorvia surface diffusion.^15–17 This is an effective way to enhance hydrogen storage capacity of carbonaceous adsorbents, as it has been generally agreed that carbon adsorbents with physical adsorption alone can not reach the US Department of Energy target set for onboard transportation applications.^15,18,19

Yang et al.^20–23 and Cheng et al.²⁴ have demonstrated that hydrogen capacities of many nanostructured/porous materials can be significantly enhanced by incorporating hydrogen spillover. Wang and Yang¹⁵ have found that Ru is an effective hydrogen dissociation source while the templated carbon with an ultrahigh surface area is an ideal receptor. The Ru-doped templated carbon has attained a hydrogen storage capacity of 1.43 wt% at 298 K and 10.3 MPa. Chen and coworkers²⁵ have investigated the mechanisms of hydrogen spillover on carbon surfaces, finding that the process is an activation process in nature and involves the formation of C–H bonds which may hinder the diffusion process at moderate temperature.

The nanosized Ni–B alloy is a good catalyst for hydrogenation by combining organic compounds such as nitrobenzene and furfural.^26–28 The Ni–B catalyst is passivated by boron so that it is more stable than RANEY® nickel and not susceptible to fire when exposed to air. However, nanosized Ni–B alloys have never been reported as promoting hydrogen dissociation for hydrogen spillover. Very recently, graphene, as a relatively new kind of carbon material, has been extensively studied recently owing to its superior chemical stability, large surface-to-volume ratio and high conductivity.²⁹ In addition, applications of graphene in various areas such as sensors and solar cells have increased.^30–32 In this work, a three-dimensional graphene material (3D graphene) doped with Ni–B nanoalloys was prepared and its hydrogen adsorption/desorption properties were investigated for hydrogen storage applications.

2. Experimental

Graphite was purchased from Bay Carbon. All other chemicals were purchased from Sigma and used without further purification. Millipore Mill-Q water was used in all experiments. Graphite oxide was synthesized from graphite through oxidation by using NaNO₃, H₂SO₄ and KMnO₄ as reported.³³Graphite oxide powder was obtained by drying at 60 °C under vacuum overnight. 3D graphene was prepared by putting the as-prepared graphite oxide powder into a glass bottle under vacuum and the system was heated at 150 °C for 45 min, as reported in our previous work.³² The highly loose, black-colored 3D graphene powder was then obtained. The 3D graphene was doped with Ni–B nanoalloys in the following procedures. Nickel acetate was used as the precursor while the 1 M NaBH₄ solution was used as the reducing agent. Firstly, 20 mg 3D graphene was dispersed into 80 mL water and then the precursor (nickel acetate) was added to the aqueous suspension under magnetic stirring. Secondly, the solution of NaBH₄ was added dropwise to the suspension. In order to assure the full reduction of nickel, excessive amount of NaBH₄ was used. After 6 h stirring, the suspension was filtered, washed and dried at 60 °C. By varying the ratio of reactants used, the composites with different content of nickel and boron were obtained.³⁴ The exact content of doped Ni and B in the samples was measured by an inductively coupled plasma emission spectrometer (ICP-ES, Perkin-Elmer Optima 3000).

To measure the hydrogen storage capacity, the samples were degassed overnight under high vacuum and 150 °C. The hydrogen adsorption was measured at 77 K and 273 K, respectively, with a pore and surface analyzer (Quantachrome Adsorb-1) which employs the standard volumetric method. To determine the isosteric heat of hydrogen adsorption, hydrogen adsorption isotherms at liquid argon temperature (87 K) were also measured.

The structure and morphology of the samples were investigated by X-ray diffraction (XRD, Rigaku D/max-2500), scanning electron microscopy (SEM, JEOL JSM 6700F field emission, 10 kV) and transmission electron microscopy (TEM, JEOL 1400, 120 kV). Besides ICP-ES, Elemental analyses were also performed using energy dispersive X-ray spectroscopy (EDX, FESEM, JSM-6700F, Japan). The specific surface area, pore size and pore volume of the samples were determined by analyzing the standard nitrogen adsorption isotherm measured at 77 K.

3. Results and discussion

The contents of doped Ni and B in three samples prepared were measured by ICP-ES. The results indicate that the content of Ni in those samples is 2.41, 1.51 and 0.83 wt%, respectively; while the content of B is 0.69. 1.39 and 1.09 wt%, respectively. The pristine graphene and three Ni–B alloy-doped samples were termed as GP, GP-Ni_2.41B_0.69, GP-Ni_1.51B_1.39 and GP-Ni_0.83B_1.09, respectively.

The XRD patterns of graphene and Ni–B-doped graphene samples are shown in Fig. 1(a). The diffraction peak at ca. 25° is due to the C(002) plane. The XRD patterns of the Ni–B-doped samples appear similar to that of pristine graphene. In other words, Ni–B alloys are hardly detected with the XRD diffraction. However, EDX spectra of GP-Ni_0.83B_1.09 clearly indicate the existence of Ni–B (Fig. 1(b)). Although the EDX spectra also show the presence of oxygen and silicon, they may exist in trace amounts in the residual oxygen functional groups on graphene or in the substrate. It has been reported that Ni–B nanometal catalysts prepared by different methods generally have an amorphous structure.³⁵ Therefore, the amorphous state and low content may be the reason why Ni–B alloys cannot be found in the XRD patterns of the Ni–B-doped samples. Fig. 2 shows the SEM images of GP and GP-Ni_0.83B_1.09. It can be seen that, compared with GP, the doped sample maintains the structural integrity but becomes more fluffy. The TEM image of GP-Ni_0.83B_1.09 shown in Fig. 3 tells that the alloy nanoparticles are distributed on the graphene sheet with particle sizes from a few nm to ∼20 nm.

Fig. 1 (a) XRD patterns of GP, GP-Ni_2.41B_0.69, GP-Ni_1.51B_1.39 and GP-Ni_0.83B_1.09; (b) EDX spectra of GP-Ni_0.83B_1.09.

Fig. 2 SEM images of GP (a) and GP-Ni_0.83B_1.09 (b).

Fig. 3 TEM image of GP-Ni_0.83B_1.09.

Fig. 4 presents hydrogen adsorption and desorption isotherms at 77 K and up to 106 kPa. All the samples have reversible hydrogen uptake but with a perceptible hysteresis in desorption isotherm. The GP exhibits a low hydrogen capacity of ∼1.4 wt%. With the doping of Ni–B nanoalloys, this capacity is considerably enhanced. GP-Ni_0.83B_1.09 reaches a maximal hydrogen storage capacity of 4.4 wt%, which is 3 times more than that of the GP and is among the top performer for carbon-based materials at the similar conditions. It is notable that the isotherm is still largely linear at the upper limit of the experimental pressure, suggesting that the sample is still way from saturation. The hydrogen capacity of the sample with higher Ni–B content (GP-Ni_2.41B_0.69) decreases to 2.6 wt%, which suggests that the excessive doping of Ni–B alloys is not beneficial or is even detrimental to the storage capacity, probably due to the increase in the bulk density or the blocking of some pores of the graphene. This result also indicates that the doped metal nano-particles function more as catalytic centers for hydrogen spillover, rather than as centers for hydrogen aggregation.

Fig. 4 Hydrogen adsorption and desorption isotherms of GP, GP-Ni_2.41B_0.69, GP-Ni_1.51B_1.39 and GP-Ni_0.83B_1.09 at 77 K (adsorption data are shown as closed shapes, desorption data as open shapes).

Next, the hydrogen adsorption isotherms on GP and GP-Ni_0.83B_1.09 were measured at 273 K and compared in Fig. 5. It is noted that the two isotherms are significantly different. The hydrogen adsorption capacity of GP is only noticeable when P/P₀ > 0.7 with a maximum capacity of 0.008 wt% at 106 kPa. Such a low capacity is due to the fact that a hydrogen molecule has a very low critical temperature and small molecular size/weight so that its capacity resulted from a purely physical adsorption process is minimal at the ambient conditions. This result is in agreement with our previous study that the hydrogen adsorption on a templated carbon with super high surface area (∼2,600 m² g⁻¹) was also low (∼1.0 wt%) at the similar conditions.³⁶ In comparison, GP-Ni_0.83B_1.09 presented a much improved capacity of 0.2 wt%, which is about 25 times higher than that of GP. It is interesting that a big hysteresis loop exists between the adsorption and desorption isotherms of hydrogen at such a high temperature (compared with its critical temperature of 33 K).

Fig. 5 Hydrogen adsorption and desorption isotherms of GP and GP-Ni_0.83B_1.09 at 273 K.

The much improved capacity as well as the large desorption hysteresis suggest that the hydrogen adsorption mechanism on the doped graphene must be different from the predominantly physical adsorption on GP. As the graphene does not possess many small micropores (with a size <0.7 nm), the micropore hysteresis from the diffusion barrier when molecules enter and exit the small pores is not likely to be the cause.³⁷ These phenomena, therefore, are attributed to the spillover.³⁸ It is suggested that the hydrogen molecules first attached to the Ni–B nanoalloys, become dissociated to hydrogen atoms and diffuse to the sites of the graphene to form C–H bond, which explains the observed high capacity. During the desorption process, hydrogen atoms dispersed on the graphene surface combine with each other to form hydrogen molecules and are released back to the bulk phase. Such a process requires hydrogen atoms to overcome certain energy (breaking C–H bond) and diffusion barriers and therefore resulted in the desorption hysteresis. Contescu et al.³⁹ reported Pd doped activated carbon fibers for hydrogen spillover, in which small desorption hysteresis was observed at the low pressure range. They found C–H bonds were formed on the Pd-doped carbon fibers and this chemical bond is responsible for the hysteresis. Chen et al.²⁵ performed molecular simulations and further proposed the following mechanism at high temperature. The C–H bond is difficult to maintain on the graphene surface at ambient temperature, while at low temperature, both C–H and physical adsorption can trap the hydrogen atoms on the graphene surface. The reverse spillover process requires the breaking of C–H bonds and the activation energy of which is comparable to the adsorption energy.

Based on our experimental observations and the reported works, it is proposed that, at low temperature (77 K), hydrogen molecules were adsorbed onto graphene at relatively high concentrations (i.e. 1.4 wt%), which would speed up the dissociation process (H₂ → 2H) and enhance the concentration of hydrogen atoms on graphene. Meanwhile, the adsorption affinity of graphene surface (b = b₀(E/RT), where E is the adsorption energy and T is the temperature) is strong enough to hold hydrogen atoms to form C–H bond. Therefore, a high hydrogen adsorption capacity was observed. At high temperature (273 K), the adsorption affinity became much lower so that only a small fraction of the graphene surface presented enough energy for hydrogen atoms to be adsorbed, dissociated and form the C–H bond. This resulted in a much lower capacity (although it is still much higher than that of the GP). The big hysteresis is the reflection of the energy barrier needed for hydrogen atoms to break the C–H bond and combine together to become hydrogen molecules and escape back to the bulk phase.

To investigate the role of physical adsorption, nitrogen isotherms at 77 K were measured on GP and GP-Ni_0.83B_1.09, respectively, and the result is shown in Fig. 6. It is interesting to see that the pristine GP adsorbs more N₂ than the doped sample at the low relative pressure (P/P₀ < 0.7). This exercise proves that, with physical adsorption alone (N₂ cannot be dissociated to form C–N bonds), the adsorption capability of GP can be higher than that of the doped samples. Table 1 summarized the structural characteristics of the two samples derived from the nitrogen isotherms. It can be seen that, as a result of doping, the BET surface area decreases from 477 to 272 m² g⁻¹ while the pore volume reduces from 1.04 to 0.63 cc g⁻¹. This is in agreement with our previous discussion that doping would increase the bulk density and reduce the porosity. The pore size distributions of GP and a doped sample are presented in the inset of Fig. 6. We see that both GP and GP-Ni_0.83B_1.09 contain a very small fraction of small micropores (with the size <1.0 nm) and a very large fraction of mesopores (with the size >2.0 nm). Because it has been proved⁴⁰ that the optimal size of graphene pore for hydrogen storage should be around 0.9 nm, the high hydrogen adsorption capacity on the doped graphene must be from a process other than physical adsorption.

Fig. 6 Nitrogen isotherms measured for GP and GP-Ni_0.83B_1.09 at 77 K. Inset: Pore size distribution of GP and GP-Ni_0.83B_1.09.

Table 1 Structural properties of the GP and GP-Ni_0.83B_1.09

	BET Surface area/m^2 g^−1	Pore volume/cc g^−1	Pore size distribution/Å	Theoretical monolayer H2 capacity/wt%	Real capacity/wt%
GP	477	1.04	10–60	1.3	1.4
Doped-GP	272	0.632	10–60	0.5	4.4
Perfect Graphene	2630	—	—	4.8	--

---

The structural heterogeneity of GP is also a factor worthwhile to investigate. From Fig. 2a and 2b, the structure of 3D graphene is seen to be far from the ideally perfect graphite layers, but consists of irregular pores, corners, edges, and voids, etc. Those non-ideal structures may offer high adsorption potentials for gas adsorbates as well as for the doped metal particles. Therefore, the surface area can be drastically reduced by doping. A further examination of the pore size distributions in Fig.6 revealed that, while the volume of small micropores (<0.9 nm) is virtually unchanged, the volumes of micropores (1.0–2.0 nm) and mesopores (<5.0 nm) are significantly reduced. This reduction in pore volume is well matched by the size of the doped metal particles. As can be seen from the TEM image in Fig. 3, the metal particles present a size distribution from a few nm to ∼20 nm.

An estimation of theoretical surface coverage is also presented in Table 1 for GP, doped GP, and ideal graphene. We see that, at 77 K, if the surface of the 3D graphene is covered with a monolayer of hydrogen molecules, the hydrogen capacity would be ∼1.3 wt%, which is very close to the experimental value of 1.4 wt%. The overshoot may be due to the multilayer adsorption in some micropores, as it has been shown by Kuchta et al.⁴¹ that at 77 K the 3rd hydrogen layer can form in the mid-plane of the graphene pores with the size of ∼1.2 nm. It is interesting to see that the hydrogen capacity of the doped sample (4.4 wt%) is slightly below the theoretical limit of monolayer coverage of hydrogen atoms on ideal graphenes (with the surface area of 2,630 m² g⁻¹⁴⁰). Although nitrogen adsorption isotherms suggest that the 3D graphene is not single lattice layer in nature, it has been suggested⁴² that hydrogen atoms (with the diameter of ∼0.24 nm) can diffuse into the space between the graphite lattice layers (a distance of ∼0.34 nm).

Finally, the isosteric heat of hydrogen adsorption was determined for GP and GP-Ni_0.83B_1.09 from their adsorption isotherms measured at 77 K and 87 K. The isosteric heat is defined by the Van't Hoff equation:⁴³

Q_iso/RT² = −(∂lnP/∂T)_q (1)

where Q_iso is the isosteric heat of adsorption (kJ mol⁻¹), T is the T/K, P is the adsorbate pressure, q is the adsorption amount and R is the universal gas constant.

The computed results are shown in Fig. 7. It can be seen that the adsorption heat is high at the low surface loading while decreasing with the increased loading. This phenomenon corresponds to the adsorption in small pores (which are of very small volume fraction, as shown in Fig. 6) where adsorption potential is high. When hydrogen capacity is below 7.5 mg g⁻¹, the adsorption heats of both samples are similar at ∼1 kJ mol⁻¹. At high hydrogen loading, however, the adsorption heat of the Ni–B-doped sample becomes relatively constant at ∼1 kJ mol⁻¹, while that of the pristine graphene fluctuates and decreases quickly. This provides further evidence that the adsorption mechanism is chemical in natural,⁴⁴ but assisted with physical adsorption in small pores at low surface loading. Therefore, it can be concluded that the doping of appropriate content of Ni–B nanoalloys leads to the dissociative chemisorption of hydrogen molecules by spillover, and therefore, resulting in a high hydrogen storage capacity.

Fig. 7 Isosteric heat of hydrogen sorption of GP and GP-Ni_0.83B_1.09.

4. Conclusions

3D graphene materials doped with different amount of Ni–B nanoalloys were prepared by a chemical reduction method. The sample with 0.83 wt% Ni and 1.09 wt% B shows a maximal reversible hydrogen storage capacity of 4.4 wt%, which is better than that of pristine graphene by three times, and is also excellent among all carbon-based materials. According to the analysis of BET specific surface area and pore volume, it confirms that the high hydrogen adsorption capacity comes from a process other than physical adsorption. With the analysis of hydrogen isotherms at 77 and 273 K, and isosteric heat of adsorption, it is suggested that the doping of appropriate content of Ni–B nanoalloys can lead to the dissociative chemisorption of hydrogen molecules by spillover to achieve a high hydrogen storage capacity.

Acknowledgements

The authors are grateful to the financial support by Center of Advanced Bionanosystems in Nanyang Technological University.

References

1. R. Coontz and B. Hanson, Science, 2004, 305, 957–959.
2. V. Kalamse, N. Wadnerkar and A. Chaudhari, J. Phys. Chem. C, 2010, 114, 4704–4709.
3. J.-L. Bobet, S. Pechev, B. Chevalier and B. Darriet, J. Mater. Chem., 1999, 9, 315–318.
4. K.-F. Aguey-Zinsou and J.-R. Ares-Fernández, Energy Environ. Sci., 2010, 3, 526–543.
5. D. Moser, D. J. Bull, T. Sato, D. Noréus, D. Kyoi, T. Sakai, N. Kitamura, H. Yusa, T. Taniguchi, W. P. Kalisvaart and P. Notten, J. Mater. Chem., 2009, 19, 8150–8161.
6. X. Zhao, L. Ma, Y. Yao, Y. Ding and X. Shen, Energy Environ. Sci., 2010, 3, 1316–1321.
7. D. H. Gregory, J. Mater. Chem., 2008, 18, 2321–2330.
8. B. Peng and J. Chen, Energy Environ. Sci., 2008, 1, 479–483.
9. A. C. Dillon, K. M. Jones, T. A. Bekkedahl, C. H. Kiang, D. S. Bethune and M. J. Heben, Nature, 1997, 386, 377–379.
10. G. T. Palomino, M. R. L. Carayol and C. O. Areán, J. Mater. Chem., 2006, 16, 2884–2885.
11. D. Zhao, D. Yuan and H. C. Zhou, Energy Environ. Sci., 2008, 1, 222–235.
12. P. Kowalczyk, R. Holyst, M. Terrones and H. Terrones, Phys. Chem. Chem. Phys., 2007, 9, 1786–1792.
13. C. Liu, Y. Y. Fan, M. Liu, H. T. Cong, H. M. Cheng and M. S. Dresselhaus, Science, 1999, 286, 1127–1129.
14. M. Sevilla, R. Foulston and R. Mokaya, Energy Environ. Sci., 2010, 3, 223–227.
15. L. Wang and R. T. Yang, J. Phys. Chem. C, 2008, 112, 12486–12494.
16. S. T. Srinivas and P. K. Rao, J. Catal., 1994, 148, 470–477.
17. G. M. Pajonk, Appl. Catal., A, 2000, 202, 157–169.
18. R. T. Yang, Carbon, 2000, 38, 623–626.
19. M. Shiraishi, T. Takenobu, H. Kataura and M. Ata, Appl. Phys. A: Mater. Sci. Process., 2004, 78, 947–954.
20. A. Lueking and R. T. Yang, J. Catal., 2002, 206, 165–168.
21. A. J. Lachawiec, G. S. Qi and R. T. Yang, Langmuir, 2005, 21, 11418–11424.
22. Y. W. Li and R. T. Yang, J. Am. Chem. Soc., 2006, 128, 726–727.
23. L. Wang and R. T. Yang, Energy Environ. Sci., 2008, 1, 268–279.
24. H. Cheng, L. Chen, A. C. Cooper, X. Sha and G. P. Pez, Energy Environ. Sci., 2008, 1, 338–354.
25. L. Chen, G. Pez, A. C. Cooper and H. Cheng, J. Phys.: Condens. Matter, 2008, 20, 064223.
26. G. Seo and H. Chon, J. Catal., 1981, 67, 424–430.
27. C. Li, W. Chen and W. J. Wang, Appl. Catal., A, 1994, 119, 185–190.
28. A. Ma, W. Lu and E. Min, 2000. U.S. Patent 6,051,528.
29. C. S. Shan, H. F. Yang, J. F. Song, D. X. Han, A. Ivaska and L. Niu, Anal. Chem., 2009, 81, 2378–2382.
30. C. X. Guo, H. B. Yang, Z. M. Sheng, Z. S. Lu, Q. L. Song and C. M. Li, Angew. Chem., Int. Ed., 2010, 49, 3014–3017.
31. S. R. Ng, C. X. Guo and C. M. Li, Electroanalysis, 2010 DOI:10.1002/elan.201000344.
32. C. X. Guo, Z. S. Lu, Y. Lei and C. M. Li, Electrochem. Commun., 2010, 12, 1237–1240.
33. D. Li, M. B. Muller, S. Gilje, R. B. Kaner and G. G. Wallace, Nat. Nanotechnol., 2008, 3, 101–105.
34. C. A. Brown, J. Org. Chem., 1970, 35, 1900–1904.
35. Y. C. Liu, C. Y. Huang and Y. W. Chen, J. Nanopart. Res., 2006, 8, 223–234.
36. C. Guan, X. Zhang, K. Wang and C. Yang, Sep. Purif. Technol., 2009, 66, 565–569.
37. J. Chen, L. S. Loo and K. Wang, J. Chem. Eng. Data, 2008, 53, 2–4.
38. L. C. Jesse, J. L. C. Rowsell and O. M. Yaghi, J. Am. Chem. Soc., 2006, 128, 1304–1315.
39. C. I. Contescu, C. M. Brown, Y. Liu, V. V. Bhat and N. C. Gallego, J. Phys. Chem. C, 2009, 113, 5886–5890.
40. S. K. Bhatia and A. L. Myers, Langmuir, 2006, 22, 1688–1700.
41. B. Kuchta, L. Firlej, P. Pfeifer and C. Wexler, Carbon, 2010, 48, 223–231.
42. A. Rai, R. Schneider and M. Warrier, J. Nucl. Mater., 2008, 374, 304–312.
43. D. M. Ruthven, Principles of adsorption and adsorption processes. Wiley-Interscience: New York, 1984.
44. A. J. Lachawiec, Jr. and R. T. Yang, Langmuir, 2008, 24, 6159–6165.

---