Palladium nanoparticle and decorated carbon nanotube for electrochemical hydrogen storage

Palladium nanoparticle and decorated carbon nanotube for electrochemical hydrogen storage

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 7 ) 1 e6 Available online at www.sciencedirect.com ScienceDire...

1MB Sizes 0 Downloads 20 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 7 ) 1 e6

Available online at www.sciencedirect.com

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

Short Communication

Palladium nanoparticle and decorated carbon nanotube for electrochemical hydrogen storage Hamid Ghorbani Shiraz a,b,*, Mohadeseh Ghorbani Shiraz c a

School of Chemical Engineering, College of Engineering, University of Tehran, Tehran, Iran Young Researchers and Elite Club, Mashhad Branch, Islamic Azad University, Mashhad, Iran c Department of Biophysics, Faculty of Biological Science, Tarbiat Modares University, Tehran, Iran b

article info

abstract

Article history:

This study proposes a multicomponent system for hydrogen storage. An electrochemical

Received 3 January 2017

evaluation was used as a simple and accurate technique to assess the storage capacity. A

Received in revised form

porous silicon substrate was fabricated using an electrochemical anodization process and

16 February 2017

decorated with palladium nanoparticles using the electroless method. The hybrid sub-

Accepted 6 March 2017

strate underwent chemical vapor deposition for 45 min. Since the deposited palladium

Available online xxx

nanoparticles could act as potential catalysts, carbon nanotubes grew properly over hybrid structure. The final sample was obtained through post-treatment by palladium nano-

Keywords:

particles using the same electroless method. This triplet sample was characterized using

Electrochemical anodization

field emission scanning electron microscopy and X-ray diffraction. Galvanostatic charge/

Hydrogen storage capacity

discharge experiments were used to conduct electrochemical evaluations of proposed

Pd decorated nanotube

electrode. A maximum hydrogen storage capacity of 537 mAh/g (~2.05 wt.%) was achieved

Galvanostatic charge/discharge

for the triple-structure sample. The measurements demonstrate that the storage capacity

Cyclic life performance

of the triple-structure sample was reduced by a factor of 0.05% after 100 cycles. Although the obtained storage capacity is far from DOE targets, optimized structures based on the proposed electrode may be further developed as an efficient storage system. © 2017 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.

Introduction These days, depletion and negative environmental impacts of fossil fuels have prompted scientists to research alternative energy resources. As a high-content energy resource, hydrogen has the potential to play an important role in sustainable development. Unlike fossil fuels, it is a clean energy resource and does not emit pollution during

combustion operation. Hydrogen is also a renewable energy resource, and there is no concern regarding depletion. While several hydrogen production methods have been investigated to harness this potential [1,2], storage remains the primary obstacle to its practical application as an energy resource. Recently, several structures and materials have been developed to address this problem. Among these, porous medias have demonstrated a high potential; in fact, the

* Corresponding author. School of Chemical Engineering, College of Engineering, University of Tehran, Tehran, Iran. E-mail address: [email protected] (H.G. Shiraz). http://dx.doi.org/10.1016/j.ijhydene.2017.03.129 0360-3199/© 2017 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved. Please cite this article in press as: Shiraz HG, Shiraz MG, Palladium nanoparticle and decorated carbon nanotube for electrochemical hydrogen storage, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.03.129

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 7 ) 1 e6

presence of pores in the substrate could offer large surface area and empty volume [3]. Porous silicon (PS) has received considerable attention in practical and theoretical studies [4,5]. Due to the medium affinity of PS and hydrogen [6], the presence of catalyst species associated with PS could improve storage capacity. For example, transitional metals could contribute to the efficient atomization process of hydrogen molecules [4,7], thereby facilitating the diffusion of hydrogen atoms into PS's nanoscale pores and channels. Moreover, the catalysts could improve the strength of interactions through chemisorption. Ghorbani Shiraz et al. demonstrated a high hydrogen storage capacity (HSC) for palladium (Pd)-decorated PS [8], observing that Pd nanoparticles selectively absorb hydrogen molecules [9] and improve storage capacity through atomization and chemisorption. Javan et al. studied hydrogen's interactions with a graphene-like silicon carbide (SiC) sheet doped with Pd using density functional theory [10]. They reported that hydrogen is absorbed chemically over either Pd or a silicon carbide sheet. However, some of the bonding may be attributable to the Kubas interaction, which is the hybridization of the hydrogen molecule's anti-bonding orbital with the d-orbital of the transition metal. Although Kubas interaction is one of the efficient hydrogenation process, its shortcoming is the tendency of metal species to clustering phenomenon [11]. Javan et al. reported that, in the absence of Pd, most of the interactions are physisorption. Carbonaceous nanomaterials have recently exhibited a high potential for hydrogen storage applications. According to the literature, these nanomaterials could be hydrogenated through physisorption [12,13] and spillover [14,15]. Ren et al. documented that hydrogen's interaction (physisorption) with carbon nanomaterials possesses very low binding energy, which indicates a low HSC [16]. However, high surface areas, in the case of these materials, present the potential for numerous physisorption. Absolutely, decoration of carbonaceous nanomaterials with the transition metals could strongly create numerous physisorption and chemisorption interactions [11,17]. Faye et al. used the first principle method to calculate the HSC of Pdfunctionalized graphene [4]. They reported an HSC of 3.622 wt.% for double-sided Pd-functionalized graphene and demonstrated that the active hydrogenation mechanisms were polarization and orbital hybridization. In fact, the polarization process could contribute to hydrogenation through the electric field enhancement caused by the interaction of cations with multiwall carbon nanotubes (MWCNT) and graphene. One of the simplest and most accurate methods to evaluate hydrogen storage is electrochemical hydrogenation [18]. During this process, hydrogen is produced by splitting water on the surface of a polarized electrode that then absorbs it [19]. Oberoi et al. measured the HSC of an active carbon electrode made of brown coal using the electrochemical method. They reported an HSC of 1.29 wt.% [20]. Visintin et al. reported that, in presence of Pd, electrochemical hydrogenation demonstrated a longer cycle lifetime and a higher rate of electrode discharge [21]. In this study, the triple-structure of Pd-CNT/Pd/PS/Si is introduced to improve HSC through a synergic effect.

Experimental Electrochemical anodization employs an electrochemical cell to advance the etching process through chemical reactions. An ethanol-based solution was used as an electrolyte for electrochemical anodization, in a given volume ratio (35% HF: EtOH: DI H2O; 2: 1: 2). Electrochemical anodization was conducted using direct current at a constant density of 10 mAcm2 for 10 min. Immediately after electrochemical anodization, Pd nanoparticles were deposited on the porous polysilicon using the electroless technique. This process used an aqueous solution of palladium dichloride (PdCl2) in the presence of trace amounts of hydrogen chloride (HCl) at room temperature. Following this, carbon nanotubes (CNTs) were grown over the decorated-porous substrate (Pd/PS/Si) using Chemical vapor deposition (CVD), which was performed using methane (CH4) as a feed gas with a flow rate of 80 sccm for 45 min at 980  C. Finally, the prepared samples (CNT/Pd/PS/Si) were decorated with Pd nanoparticles using the same electroless process (post-treatment).

Characterization The surface morphology was carried out by Field emission scanning electron microscopy (FESEM, Mira 3-XMU). The crystalline structure of the samples was characterized using X-ray diffraction (Philips X'pert operating with CuKa radiation (l ¼ 1.54178  A) at 40 kV/40 mA). The chemical composition of the Pd-CNT/Pd/PS/Si sample was characterized by Energy dispersive X-ray spectroscopy (EDS). The electrochemical measurement was carried out with a three-electrode system, using 3 M H2SO4 as an aqueous electrolyte; where working electrodes were PS, Pd/PS, and PdCNT/Pd/PS/Si. Platinum foil was assigned as the counter and Ag/AgCl electrode (Metrohm AG 9101 Herisau, 3 M KCl, 0.207 V versus, SHE at 25  C) was chosen as reference electrode. The reference electrode was fixed near the working electrode to minimize the ohmic drop of electrolyte. Also, the working electrode potential vs. reference electrode was monitored during this process. CP analysis was carried out using a potentiostat/galvanostat (PGSTAT204, Autolab, Echo Chemie) in a three-electrode glass cell. Galvanostatic characterizations for charge/discharge process were carried out under the current density of 500 mA/g; meanwhile, the cut-off potential of 0.5 V was adjusted. The charge operation was followed by15 min rest. Also, the discharge process was performed under current density of 500 mA/g with the cut-off potential of 0.6 V.

Results and discussion Surface morphology The microstructure and surface morphology of fabricated samples were characterized using FESEM, Figs. 1 and 2. Fig. 1 (a) shows Pd nanoparticles decorated the PS substrate before CVD: A large number of Pd nanoparticles have clearly been

Please cite this article in press as: Shiraz HG, Shiraz MG, Palladium nanoparticle and decorated carbon nanotube for electrochemical hydrogen storage, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.03.129

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 7 ) 1 e6

3

Fig. 1 e FESEM images showed surface morphology of hybrid sample (Pd/PS); (a) deposited Pd nanoparticle and partial over layer growth, (b) relative uniform distribution of Pd nanoparticles over PS substrate.

Fig. 2 e FESEM images showed surface morphology of sample; (a) grown CNTs over the hybrid substrate of Pd/PS, (b) the tip growth phenomenon, (c) functionalized CNT after post-treatment.

deposited. Moreover, the Fig. 1 (a) shows a layer-over-layer growth of Pd nanoparticles, which could be found through popcorn-like structures. Also, Fig. 1 (b) shows that the distribution of deposited Pd nanoparticles is relatively uniform over the PS substrate. Measurements showed that the Pd grain size

was distributed in the range of 5e15 nm. Ghorbani Shiraz et al. [8] demonstrated that the duration of electroless process directly effects on the deposited arrays; increased length leads to an agglomeration of Pd species, which results in larger grain size and layer-over-layer growth.

Please cite this article in press as: Shiraz HG, Shiraz MG, Palladium nanoparticle and decorated carbon nanotube for electrochemical hydrogen storage, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.03.129

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 7 ) 1 e6

Fig. 2 (a) shows the growth of CNTs over the Pd/PS/Si substrate. As with the distribution of Pd nanoparticles, the growth of MWCNTs was observed to be relatively uniform. Due to the concentrated deposition of catalysts, Pd nanoparticles have grown along the nanotubes (tip growth phenomenon) during CVD. Fig. 2 (b) displays the intensified tip growth observed in the CNT/Pd/PS/Si sample. The CNT/Pd/PS/Si sample was treated with dimethylformamide to introduce surface defects prior to posttreatment. It was demonstrated that, these defects could be potential centers for a reduction of Pd cations. As a result of the post-treatment, the growth of Pd nanoparticles over PS sidewalls was observed. The post-treatment Pd-CNT/Pd/PS/Si sample is displayed in Fig. 2 (c). The results of EDS characterization is presented in Fig. 3. The presence of an oxygen peak in the EDS measurement proves that an oxygen-containing film formed over the PdCNT/Pd/PS/Si sample. It has been demonstrated that the oxide barrier created by an oxygen-containing surface film can affect and occlude hydrogen species [8]. This may improve HSC through hydrogen bonding. Conversely, HSC is negatively affected by a phenomenon on the surface. Since the Pd peak varied across similar regions, the distribution of Pd species over the PS might not be uniform. This could cause Pd clustering on the surface of the PS [22], which would have a detrimental effect on HSC. However, metal clustering could be controlled for under the proper experimental conditions [8]. It has been demonstrated that the presence of light radiationdespecially UV radiation [23]dstimulates clustering, while a porous substrate minimizes it [23]; both approaches have been considered by this study.

X-ray diffraction Fig. 4 shows the typical X-ray diffraction (XRD) patterns for the triple-structure sample and the pristine silicon wafer. The sharp peak at ~70 (1 0 0) is ascribed to the PS substrate's crystal structure [24], while the weak peaks at 49 (2 0 0) and 57 (2 2 0) reflect the crystalline nature of the Pd structures [25]. The wide peak around 30 is attributed to CNTs [26]. An amorphous structure was detected on the PS substrate by comparing PS's

Fig. 3 e EDS measurement for investigation of chemical composition at the surface of sample; vertical axial shows the amount of chemicals.

Fig. 4 e XRD measurement of proposed sample.

XRD pattern to that of the Pd-CNT/Pd/PS/Si sample. Since the Pd nanoparticles form a non-continuous layer, the relevant peak is not so strong. This could be caused by the Pd deposition method, thickness of deposited layer, and the weak crystalline nature of Pd structures.

Electrochemistry evaluation Galvanostatic charge/discharge electrochemical measurements were employed to calculate the examined electrodes' HSCs. Charge/discharge experiments were performed using a current density of 500 mA/g. The HSC in charge/discharge curves was evaluated with the help of discharge time, using the following equation: HSC ¼

Itd m

where I, td, and m represent the charge/discharge current [mA], the discharge time [h], and the active mass [g], respectively. Fig. 5 demonstrates the first-time discharge trends for three electrodes. A maximum discharge capacity of 228 mAh/ g was achieved for PS, while Pd/PS and Pd-CNT/Pd/PS/Si obtained higher values. Since Pd [8] and carbon nanomaterials [12] could efficiently improve the amount of stored hydrogen, the hybrid and triple-structure samples offered better results. It is evident that PS has more HSC than pristine polycrystalline

Fig. 5 e Discharge capacity for PS (dot), Pd/PS (dash), and Pd-CNT/Pd/PS/Si (line).

Please cite this article in press as: Shiraz HG, Shiraz MG, Palladium nanoparticle and decorated carbon nanotube for electrochemical hydrogen storage, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.03.129

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 7 ) 1 e6

silicon (Si) [23]; this can be attributed to PS's broad surface area, which extends through its numerous open-ended pores and nanoscale channels. Moreover, it has been documented that PS has a greater affinity to hydrogen than pristine silicon [27]. Fig. 5 shows a higher discharge capacity 439 mAh/g for the Pd/PS hybrid electrode, compared to other examined electrode. This could be related to the effect of Pd nanoparticles. The active and selective absorption of hydrogen species as well as Pd nanoparticles' affinity to hydrogen through chemisorption and ionic bands could enhance discharge capacity and, consequently, HSC [8,9]. As CNTs' inner, outer, and inter-tube regions offer wide surface areas, the possibilities for physisorption improves significantly. There is also great potential for chemisorption through the Pd-decorated (those nanoparticles which deposited during post-treatment) on the outer surface of CNTs. Additionally, the possibility of chemisorption on the inner layer of CNTs through the tip growth of Pd is considerable. Moreover, the improved electrical behavior of the triplestructure sample and the electron resonance disorder in the presence of structural components in the sample's surface indicate a high potential for hydrogenation. The aforementioned effects offer a discharge capacity of approximately 537 mAh/g, which is equivalent to about 2.05 wt.%. This value is higher than that of PS and Pd/PS, which could be attributed to the synergic effect of PS, CNTs, and Pd. Fig. 6 demonstrates the cyclic life performance of the three studied electrodes (PS, Pd/PS, and Pd-CNT/Pd/PS/Si). Galvanostatic characterizations were carried out using a current density of 250 mA/g. The PS electrode's HSC was quite low in comparison to the other examined electrodes, and it demonstrated a slight variation in HSC during successive electrochemical cycles. The HSC decay began after 17 cycles, and the reduction trend reached a constant value after 91 cycles (and a 17% reduction in HSC). Generally, Pd decorated PS demonstrated a higher HSC as well as a lower reduction rate: Its HSC obtained a constant value after only 19 cycles, and the presence of Pd lowered the decay ratio to approximately 14% compared to the initial value. The highest HSC value belonged to the Pd-CNT/Pd/PS/Si triple structure. The HSC reduction met a constant value after 69 cycles; the decay ratio was approximately 0.5%. Although this was primarily caused by

Fig. 6 e Cyclic life performance for PS (dot), Pd/PS (dash), and Pd-CNT/Pd/PS/Si (line).

5

the incorporation of CNTs' broad surface arrays into the triple structure, the synergic effects of Pd and PS must also be considered. In the case of all three electrodes, the first cycle related to the loading process. Following this, HSC reductions caused by the oxidation of the electrodes began [28], which continued until a constant HSC was achieved when an equilibrium was established between the loading and reduction steps. Overall, the triple structure's significant HSC and life-cycle performance suggest that it is a promising system for hydrogen storage. Further study of this system is in progress and will be presented in following reports. Our group is investigating several chemicals for HSC; the substitution of another transitional metal for PPd may result in both a higher HSC and a more cost-effective approach.

Conclusion In summary, we prepared PS as a template, using electrochemical anodization. Then, the PS was decorated by Pd nanoparticles as catalyst, using electroless technique. The hybrid substrate underwent CVD operation in order to growth of MWCNT. Finally, the sample treated with Pd nanoparticles. The fabricated sample characterized by XRD, and FESEM. Also, proposed sample, as hydrogen reservoir, was measured by electrochemical charge/discharge method. The measurements demonstrated a high potential capability of hydrogen storage for Pd-CNT/Pd/PS/Si. This multicomponent system showed the HSC of 537 mAh/g (~2.05 wt.%) which is 1.2 and 2.3 times higher than that of Pd/PS and PS, respectively.

Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

references

[1] Joshi AS, Dincer I, Reddy BV. Effects of various parameters on energy and exergy efficiencies of a solar thermal hydrogen production system. Int J Hydrogen Energy 2016;41:7997e8007. [2] Yilmaz C. Thermoeconomic modeling and optimization of a hydrogen production system using geothermal energy. Geothermics 2017;65:32e43. [3] Callini E, Aguey-Zinsou K-F, Ahuja R, Ares JR, Bals S, Bili skov N, et al. Nanostructured materials for solid-state hydrogen storage: a review of the achievement of COST Action MP1103. Int J Hydrogen Energy 2016;41:14404e28. [4] Shiraz HG, Tavakoli O. Investigation of graphene-based systems for hydrogen storage. Renew Sustain Energy Rev 2017;74:104e9. [5] Shiraz HG, Astaraei FR, Mohammadpour R. TiO2/nanoporous silicon hybrid contact for heterojunction crystalline solar cell. RSC Adv 2016;6:55046e53. [6] Zuo F, Wang L, Wu T, Zhang Z, Borchardt D, Feng P. Selfdoped Ti3þ enhanced photocatalyst for hydrogen production under visible light. JACS 2010;132:11856e7.

Please cite this article in press as: Shiraz HG, Shiraz MG, Palladium nanoparticle and decorated carbon nanotube for electrochemical hydrogen storage, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.03.129

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 7 ) 1 e6

[7] Crabtree RH. Homogeneous transition metal catalysis of acceptorless dehydrogenative alcohol oxidation: applications in hydrogen storage and to heterocycle synthesis. Chem Rev 2017. http://dx.doi.org/10.1021/ acs.chemrev.6b00556. [8] Shiraz HG, Seyfollahi R. Hybrid system for potential room temperature hydrogen storage. Vacuum 2016;131:115e9. [9] Shiraz HG, Astaraei FR, Fardindoost S, Hosseini ZS. Decorated CNT based on porous silicon for hydrogen gas sensing at room temperature. RSC Adv 2016;6:44410e4. [10] Javan MB, Shirdel-Havar AH, Soltani A, Pourarian F. Adsorption and dissociation of H2 on Pd doped graphene-like SiC sheet. Int J Hydrogen Energy 2016;41:22886e98. [11] Faye O, Szpunar JA, Szpunar B, Beye AC. Hydrogen adsorption and storage on Palladiumefunctionalized graphene with NH-dopant: a first principles calculation. Appl Surf Sci 2017;392:362e74. [12] Du¨ndar-Tekkaya E, Yu¨ru¨m Y. Mesoporous MCM-41 material for hydrogen storage: a short review. Int J Hydrogen Energy 2016;41:9789e95. [13] Niaz S, Manzoor T, Pandith AH. Hydrogen storage: materials, methods and perspectives. Renew Sustain Energy Rev 2015;50:457e69. [14] Pyle DS, Gray EM, Webb C. Hydrogen storage in carbon nanostructures via spillover. Int J Hydrogen Energy 2016;41:19098e113. [15] Wei L, Mao Y. Enhanced hydrogen storage performance of reduced graphene oxide hybrids with nickel or its metallic mixtures based on spillover mechanism. Int J Hydrogen Energy 2016;41:11692e9. [16] Ren H, Cui C, Li X, Liu Y. A DFT study of the hydrogen storage potentials and properties of Na-and Li-doped fullerenes. Int J Hydrogen Energy 2017;42:312e21. [17] Kaur M, Pal K. An investigation for hydrogen storage capability of zirconia-reduced graphene oxide nanocomposite. Int J Hydrogen Energy 2016;41:21861e9. [18] Azizi S, Salah M, Nefzi H, Khaldi C, Sediri F, Dhahri E, et al. Structure, volumetric adsorption method and

[19] [20]

[21]

[22]

[23]

[24] [25]

[26]

[27]

[28]

electrochemical hydrogen storage properties of vanadium oxide nanotubes VOx-NTs. J Alloys Compd 2015;648:244e52. Prabhukhot Prachi R, Wagh Mahesh M, Gangal Aneesh C. Adv Energy Power 2016;4(2):11e22. Oberoi AS, Andrews J, Chaffee AL, Ciddor L. Hydrogen storage capacity of selected activated carbon electrodes made from brown coal. Int J Hydrogen Energy 2016;41:23099e108. Visintin A, Castro E, Real S, Triaca W, Wang C, Soriaga M. Electrochemical activation and electrocatalytic enhancement of a hydride-forming metal alloy modified with palladium, platinum and nickel. Electrochim Acta 2006;51:3658e67. Rahimi F. Characterization of Pd nanoparticle dispersed over porous silicon as a hydrogen sensor. J Phys D: Appl Phys 2007;40:7201. Shiraz HG, Astaraei FR, Tavakoli O, Mousavi SH, Rahimi F. The effect of a porous layer on IV characterization of a polysilicon pn junction. Silicon 2016:1e6. Lee C-Y, Tseng T-Y, Li S-Y, Lin P. Growth of zinc oxide nanowires on silicon (100). Tamkang J Sci Eng 2003;6:127e32. 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;38:3681e8. Asli N, Shamsudin M, Maryam M, Yusop S, Suriani A, Rusop M, et al. Diameter controlled of carbon nanotubes synthesized on nanoporous silicon support. In: IOP conference series: materials science and engineering. IOP Publishing; 2013. p. 012004. Carraro P, Blanco AG, Soria F, Lener G, Sapag K, Eimer G, et al. Understanding the role of nickel on the hydrogen storage capacity of Ni/MCM-41 materials. Microporous Mesoporous Mater 2016;231:31e9. Guo G, Huang H, Xue F, Liu C, Yu H, Quan X, et al. Electrochemical hydrogen storage of the graphene sheets prepared by DC arc-discharge method. Surf Coat Technol 2013;228:S120e5.

Please cite this article in press as: Shiraz HG, Shiraz MG, Palladium nanoparticle and decorated carbon nanotube for electrochemical hydrogen storage, International Journal of Hydrogen Energy (2017), http://dx.doi.org/10.1016/j.ijhydene.2017.03.129