Synthesis of Nickel and Cobalt Ferrite-Doped Graphene as Efficient Catalysts for Improving the Hydrogen Storage Kinetics of Lithium Borohydride

Featuring a high hydrogen storage content of up to 20 wt%, complex metal borohydrides remain promising solid state hydrogen storage materials, with the real prospect of reversible behavior for a zero–emission economy. However, the thermodynamic barriers and sluggish kinetics are still barriers to overcome. In this context, nanoconfinement has provided a reliable method to improve the behavior of hydrogen storage materials. The present work describes the thermodynamic and kinetic enhancements of LiBH4 nanoconfined in MFe2O4 (M=Co, Ni) ferrite-catalyzed graphene host. Composites of LiBH4-catalysts were prepared by melt infiltration and investigated by X-ray diffraction, TEM, STEM-EDS and TPD. The role of ferrite additives, metal precursor treatment (Ar, Ar/H2) and the effect on hydrogen storage parameters are discussed. The thermodynamic parameters for the most promising composite LiBH4-graphene-NiFe2O4 (Ar) were investigated by Kissinger plot method, revealing an EA = 127 kJ/mol, significantly lower than that of neat LiBH4 (170 kJ/mol). The reversible H2 content of LiBH4-graphene-NiFe2O4 (Ar) after 5 a/d cycles was ~6.14 wt%, in line with DOE’s target of 5.5 wt% storage capacity, while exhibiting the lowest desorption temperature peak of 349 °C. The composites with catalysts treated in Ar have lower desorption temperature due to better catalyst dispersion than using H2/Ar.


Introduction
The safe and effective production, storage and transport of hydrogen energy remains a current challenge. Among potential hydrogen sources, solid-state hydrogen storage materials such as simple and complex metal hydrides have been extensively investigated over the past few decades [1]. A series of shortcomings plague their wide use in mainstream energy systems, such as sluggish kinetics and high thermodynamic barriers, which translate into a high practical dehydrogenation temperature. Possible means to mitigate these drawbacks refer to nanosizing, nanoconfinement, and usage of additives and catalysts, among others [2].
When nanoconfined in the pores of a proper scaffold, the behavior of borohydrides and alanates features marked improvements over their pristine counterparts [1,2]. On one hand, the particle agglomeration and growing during cycling are inhibited, which is particularly important since higher temperatures are required during a/d measurements. This strategy can guarantee a better kinetic behavior of the investigated composites, which showed increased dehydrogenation rates. On the other hand, the nanoporous scaffold facilitates gas diffusion during cycling, contributing to the overall enhanced behavior.
A key step in engineering competent nanocomposites for energy storage is to ensure the active hydrogen storage material is confined uniformly at the nanoscale, so that there are no outer-host processes like crystal growth which would adversely affect the hydrogen storage performance. The majority of previously-reported works deal with confinement of complex hydrides in previously-prepared hosts, which impairs some degree of inhomogeneity. Many nanoporous substrates were used as scaffolds: 2D-structured silica, carbon, carbon nanobowls, mesoporous carbon, and graphene [3]. Graphene nanosheets are 2D materials that have been extensively used in energy storage applications, both in pristine and catalyzed form. Among most utilized catalysts are metal fluorides, chlorides oxides Fe 3 O 4 [4] etc.
Among investigated species nanoconfined in porous scaffolds are LiBH 4 [5][6][7][8][9][10][11][12][13][14][15], Ca(BH 4 ) 2 [16], or LiAlH 4 [17]. There are, however, some scarce reports where nanosized complex hydride was synthesized and showed unexpectedly high storage capacity of 12% in case of LiBH 4 for instance, without the additional support of a porous scaffold [18]. Various enhancements can be obtained in low temperature synthesis by solid-gas reaction [19], extensive study of pressure or temperature dependence of decomposition [20] and extensive reversibility studies [21]. These in-extenso studies also revealed some of the plausible intermediates occurring during dehydrogenation, such as Li 2 B 12 H 12 and possibly other crystallographic forms of LiBH 4 as well [22]. Other aspects with high practical importance such as the oxidation of borohydride groups occurring upon exposure to oxygen-containing atmosphere or functional groups grafted onto the nanoporous support have also been described [23]. Modern investigation techniques like QENS (neutron vibrational spectroscopy and quasielastic neutron scattering) have been used to probe the dynamic properties of borohydride anion, with relevance to hydrogen storage measurements [24]. Other mixed oxides have shown good catalytic activity for ammonia borane methanolysis [25,26]. Other potential strategies to improve complex hydride behavior have been reviewed recently [27], while CoFe 2 O 4 and NiFe 2 O 4 have been investigated as catalysts for complex hydrides, especially for complex hydrides like lithium borohydride LiBH 4 [28] and alkali metal alanates LiAlH 4 [29,30] and NaAlH 4 [31].
However, oftentimes the activity of nanosized catalysts dispersed in porous scaffolds yields results which are hard to predict and moreover, even differ between various research groups. A possible reason is that, even though the chemical nature of the catalyst may be similar, its dispersion within the scaffold is not homogeneous and even the chemical composition of the catalyst can be altered during HEBM (high energy ball milling). In the current work, we address the inhomogeneity issue of the catalyst dispersion inside graphene 2D sheets by an in-situ catalyst preparation (NiFe 2 O 4 , CoFe 2 O 4 ) starting from a stoichiometric mixture of corresponding metal nitrates. The thermal decomposition of metal precursors was carried out under two different conditions: in Ar flow or H 2 /Ar flow. A comparison of catalyst morphology revealed that the Ar flow treatment leads to a better catalyst dispersion inside graphene nanosheets, hence producing a more potent catalyst. Additionally, the in-situ production of MFe 2 O 4 (M=Fe, Co) stems from a bottomup approach that would ensure homogeneous catalyst dispersion and a more controllable kinetic behavior of confined hydride species. The different gas flow approaches (inert-Ar or reducing H 2 /Ar) allowed further evaluation of metal ferrites (CoFe 2 O 4 , NiFe 2 O 4 -under Ar) and intermetallics/alloys (FeNi, Fe, CoFe 2 , Fe 3 C, Co 3 C-obtained under reducing atmosphere H 2 /Ar). In all investigated cases, the ferrite catalysts showed enhanced kinetic behavior of the LiBH 4 @catalyst nanocomposites, and superior to the intermetallic catalysts obtained under reducing conditions, providing a useful starting point for further investigation of in-situ generated ferrite catalysts for hydrogen production.

Materials and Methods
Starting materials for samples preparations were: lithium borohydride, LiBH 4 3 .9H 2 O, and were used in stoichiometric amount to yield MFe 2 O 4 (M=Ni, Co). The impregnated samples were subjected to thermal treatments either in Argon (Ar) or 5% Hydrogen in Argon (further noted in the text as: H 2 Ar) gas flow (treatment temperature 620 • C, treatment time 4 h, gas flow rate 100 mL/min). The CoFe 2 O 4 /graphene and NiFe 2 O 4 /graphene catalysts synthesized in Ar or H 2 Ar gas flow were mixed with LiBH 4 in the mass ratio 1:1 by prolonged time grinding under dry inert atmosphere using pestle and mortar. Afterwards they were hydrogenated at 300 • C in hydrogen gas with 99.9999% purity under 80 atm H 2 pressure. This temperature was above the melting point of LiBH 4 (270 • C) allowing an intimate mixing between lithium borohydride and the supported catalysts. The samples processing was carried out in MBraun LabStar (Garching bei Munchen, Germany) glove box under purified re-circulated Argon (<1 ppm O 2 , <1 ppm H 2 O) during all stages of sample manipulation. X-Ray diffraction investigation was performed using D-8 Advance Bruker diffractometer (Bruker, Karlsruhe, Germany) with Cu K-alpha radiation. During X-ray diffraction measurements the samples were covered with polymeric foil in order to avoid oxidation. TEM images were obtained with JEM-2100 analytical transmission electron microscope (JEOL, Tokyo, Japan) operated at 200 kV endowed with dispersive X-ray spectrometer. For TEM measurements, the samples were dispersed in hexane using high power device (VCX 750 Sonics, Newtown, CT, USA) and afterwards were deposited on grids by drying the hexane. A commercially available Sievert volumetric apparatus (Advanced Material Corporation, AMC Pittsburg, Pittsburg, KS, USA) was used for hydrogen absorption and desorption measurements at a particular temperature or with a temperature ramp rate (thermal programmed desorption TPD).  Figure 1. In all XRD diagrams the dominant contribution belongs to graphene, the peak from 2Θ: 26 • being much higher than any of the other peaks. Besides graphene, sample NFO-G-Ar contains NiFe 2 O 4 (ICDD file 04-005-6361) and FeNi (ICDD file 04-021-6318). The formation of FeNi is supported by the reducing effect of the carbon matrix. By contrast, the sample CFO-G-Ar contains mainly CoFe 2 O 4 (ICDD file 00-066-0244), but also small amounts of carbide phases Fe 3 C (ICDD file 04-008-9572) and Co 3 C (ICDD file 04-003-4355). Regarding the samples treated under reducing flux (H 2 Ar), the ferrite phase MFe 2 O 4 (M=Co, Ni) is no longer visible in XRD diffractogram, and full reduction to metallic alloy/intermetallic MFe x (x = 1,2) can be observed. The sample NFO-G-H 2 Ar contains only FeNi and Fe-bcc, besides the main contribution from graphene. As previously mentioned in the literature, the hydrogenation of NiFe 2 O 4 leads to the formation of FeNi and Fe and this observation holds true for reduction of nanoconfined nickel ferrite in the particular case when using a graphene support matrix. In the case of CoFe 2 O 4 the treatment in H 2 or H 2 Ar (hydrogen + argon) besides the main contribution from graphene. As previously mentioned in the literature, the hydrogenation of NiFe2O4 leads to the formation of FeNi and Fe and this observation holds true for reduction of nanoconfined nickel ferrite in the particular case when using a graphene support matrix. In the case of CoFe2O4 the treatment in H2 or H2Ar (hydrogen + argon) mixture generates CoFe2 (ICDD file 04-016-4643). Hence, the sample CFO-G-H2Ar contains only CoFe2 besides the main graphene contribution. The XRD peaks corresponding to NiFe2O4, CoFe2O4, FeNi and Fe are relatively wide, suggesting grains in the nanometric range. For clarity, a summary of phase identification of these catalyzed graphene supports can be found in the Table S1 (Supplementary Materials).  Figure 2 represents the X-ray diffractograms of LiBH4 mixed with CoFe2O4 and NiFe2O4 catalysts supported on graphene. The samples were extracted after five cycles of hydrogen absorption/desorption being in final re-hydrogenated (absorbed) state. During measurements the samples were covered with polymeric foil in order to avoid oxidation. The peaks from from 2Θ of about 22° and 36° belong to this foil. Expectedly, the dominant contribution from the XRD pattern belongs to graphene. Additionally, in the sample LiBH4-G were evidenced not only the peaks of LiBH4 (ICDD file 01-084-8599) but also an important contribution from Li3BO3 (ICDD file 00-018-0718) as effect of surface oxidation. After hydrogen absorbtion/desorption cycles, LiBH4 interacts with the ferrites or intermetallics contained in the samples, forming borides.    indicating very small grains. A summary of the phase identification of used catalyst (after 5 a/d cycles) is presented for clarity also in Table S1 (Supplementary Materials).

Morphological and Compositional Investigation by TEM and STEM-EDS
TEM images for the samples LiBH4-G, LiBH4-NFO-G-Ar, LiBH4-NFO-G-H2Ar, LiBH4-CFO-G-Ar depict large sheets of graphene, with lateral dimensions of a few microns. The selected area diffraction patterns (SAED) of all 4 samples reveal the presence of graphene (rings corresponding to graphite interplanar distances were identified). Also the spots corresponding to LiBH4 were detected. TEM images at higher magnification revealed elongated crystalline zones with 7-14 nm thickness ( Figure 3). Measurements on the FFT of these images concluded that the nanocrystals consist of LiBH4.  In samples LiBH 4 -NFO-G-Ar and LiBH 4 -CFO-G-Ar the nanoparticles are well separated, although they seem to be more uniformly distributed in LiBH 4 -CFO-G-Ar. The sample LiBH 4 -NFO-G-H 2 Ar contains nanoparticles more likely to form large clusters with high particles densities, while in other areas these clusters are absent. Their composition was confirmed by the EDS mapping to be: Fe-Ni (for LiBH 4 -NFO-G-Ar, and LiBH 4 -NFO-G-H 2 Ar), and Fe-Co (for LiBH 4 -CFO-G-Ar), with slightly higher iron concentration.

Hydrogen Storage Property Measurements
Investigation of hydrogen desorption behavior using temperature ramp rate of 2 • C/min (thermal programmed desorption-TPD) was performed using AMC Sievert volumetric apparatus (Advanced Material Corporation, AMC Pittsburg, Pittsburg, KS, USA). The samples were manipulated in inert and dry atmosphere in glove box and transferred in the sample holder of AMC apparatus using the same facility. The samples were first degassed in vacuum at about 100 • C, but temperatures above 150 • C have been avoided during degassing due to the risk of hydrogen desorption before the envisaged release measurement for as-prepared samples. Hydrogen desorption measurements with temperature ramp rate (TPD) were performed both for first desorption ( Figure 5A) (when the fresh sample was just loaded in the device) and after five cycles of hydrogenation in order to evaluate the behavior and composition of investigated nanocomposites ( Figure 5B).
The theoretical maximum value of hydrogen amount released during LiBH 4 decomposition into LiH and B is 13.8 wt% H 2 [1,2]. In Figure 5 the experimental wt% H 2 was normalized to this value. LiBH 4 without graphene or catalyst addition released a normalized value of only 0.25 for desorption measurement with 2 • C/min up to 450 • C. For subsequent desorption, the sample was loaded with H 2 after each hydrogen desorption at 450 • C and 80 atm H 2 for prolonged time (24 h) in order to ensure that the sample reached the maximum level of absorbed hydrogen. One can observe that even though the first LiBH 4 -G decomposition is th e fastest among all samples, the same sample behaves worst at the saturated behavior after 5 a/d cycles. This confirmed the catalytic role of NiFe 2 O 4 and CoFe 2 O 4 addition to graphene, which improves the hydrogen desorption process. Hydrogen desorption with temperature ramp rate of 2 • C/min up to 450 • C is not enough to release all the hydrogen from samples. In order to achieve this goal, the desorption was allowed longer time at the final temperature of 450 • C. The corresponding desorption kinetics curves were provided in electronic Supplementary Materials for the first desorption ( Figure S1) and for the fifth desorption ( Figure S2).     in the sample holder of AMC apparatus using the same facility. The samples were first degassed in vacuum at about 100 °C, but temperatures above 150 °C have been avoided during degassing due to the risk of hydrogen desorption before the envisaged release measurement for as-prepared samples. Hydrogen desorption measurements with temperature ramp rate (TPD) were performed both for first desorption ( Figure 5A) (when the fresh sample was just loaded in the device) and after five cycles of hydrogenation in order to evaluate the behavior and composition of investigated nanocomposites ( Figure 5B).

Hydrogen Storage Property Measurements
(A) (B) The theoretical maximum value of hydrogen amount released during LiBH4 decomposition into LiH and B is 13.8 wt% H2 [1,2]. In Figure 5 the experimental wt% H2 was normalized to this value. LiBH4 without graphene or catalyst addition released a normalized value of only 0.25 for desorption measurement with 2 °C/min up to 450 °C. For subsequent desorption, the sample was loaded with H2 after each hydrogen desorption at 450 °C and 80 atm H2 for prolonged time (24 h) in order to ensure that the sample reached the maximum level of absorbed hydrogen. One can observe that even though the first LiBH4-G decomposition is th e fastest among all samples, the same sample behaves worst at the saturated behavior after 5 a/d cycles. This confirmed the catalytic role of NiFe2O4 and CoFe2O4 addition to graphene, which improves the hydrogen desorption process. Hydrogen desorption with temperature ramp rate of 2 °C/min up to 450 °C is not enough to release all the hydrogen from samples. In order to achieve this goal, the desorption was allowed longer time at the final temperature of 450 °C. The corresponding desorption kinetics curves were provided in electronic Supplementary Materials for the first desorption ( Figure S1) and for the fifth desorption ( Figure S2).

Discussion
Using the TPD from Figure 5B one can derive the desorption peak temperature for all samples after 5 a/d cycles, as shown in Figure 6.

Discussion
Using the TPD from Figure 5B one can derive the desorption peak temperature for all samples after 5 a/d cycles, as shown in Figure 6. As depicted in Figure 6, after 5 a/d cycles, the maximum desorption peak temperature belongs to LiBH4-graphene sample that proved the advantages of doping graphene with cobalt or nickel ferrite in order to improve hydrogen desorption kinetics of LiBH4. In Table 1 are gathered the values of normalized hydrogen desorption amount for 1st and 5th de-hydrogenation along with desorption peak temperature for the latter. The normalized released amount values correspond both to the partial desorption (up to 450 °C with 2 °C/min temperature ramp rate when desorption is not completed due to kinetic limita- As depicted in Figure 6, after 5 a/d cycles, the maximum desorption peak temperature belongs to LiBH 4 -graphene sample that proved the advantages of doping graphene with cobalt or nickel ferrite in order to improve hydrogen desorption kinetics of LiBH 4 . In Table 1 are gathered the values of normalized hydrogen desorption amount for 1st and 5th de-hydrogenation along with desorption peak temperature for the latter. The normalized released amount values correspond both to the partial desorption (up to 450 • C with 2 • C/min temperature ramp rate when desorption is not completed due to kinetic limitations) and after extensive time, enough to complete desorption at the final temperature of 450 • C. Desorption peak temperature is lower for LiBH 4 -NFO sample compared with LiBH 4 -CFO sample for catalysts treated both in Ar and H 2 Ar flow. As one may observe, the graphene supported catalysts based on nickel ferrite proved better than that based on cobalt ferrite for improving hydrogen desorption kinetics of LiBH 4 . On the other hand, LiBH 4 mixed with catalysts heat treated in Ar flow have lower desorption peak temperature than their counterparts annealed in H 2 Ar flow both for cobalt and nickel ferrite supported on graphene. One should recall from STEM-EDS (Figure 4) that the catalysts treated in H 2 Ar preferentially formed large metal clusters whereas the ones treated in Ar were better dispersed. In Table 1 one can observe larger values obtained for the normalized hydrogen amount released in the 1st de-hydrogenation compared with 5th de-hydrogenation for all samples. This behavior is understandable due to formation of FeB, Fe 2 B, NiB and CoB, as shown in the X-ray diffractograms, diminishing the amount of LiBH 4 from the samples. However, there is a clear advantage of the boride presence, which is due to their catalytic effect. Interestingly, even for the complete desorption in the first cycle, LiBH 4 -G desorbed only 0.95 normalized amount whereas all the catalyzed samples reached the maximum theoretical allowed amount. For the 5th de-hydrogenation, the normalized amount is lower for LiBH 4 -G compared with that corresponding to catalyzed samples both for TPD up to 450 • C with 2 • C/min temperature ramp rate, and for complete desorption. This behavior proves once again the efficiency of using cobalt and nickel ferrite to produce catalyst-supported graphene scaffolds, which can effectively improve hydrogen desorption of nanoconfined LiBH 4 .
The desorption temperature peak for LiBH 4 -NFO-G-Ar (lithium borohydride mixed with nickel ferrite supported on graphene heat treated in Ar flow) is the lowest among all samples. For this sample, TPD measurements were performed with different temperature ramp rate of 1 K/min, 2 K/min, 4 K/min and 7 K/min up to 450 • C (see inset of Figure 7). The corresponding desorption temperature peaks were used to obtain the Kissinger plot, which allowed evaluation of the hydrogen desorption activation energy from the slope of the linear fit (Figure 7). The obtained value of 127 kJ/mol for the activation energy is significantly lower than that of bulk LiBH 4 , reported in the range 146 kJ/mol [15]-170 kJ/mol [4], which again confirms the catalytic role of nanosized metal ferrites synthesized.
samples. For this sample, TPD measurements were performed with different temperature ramp rate of 1 K/min, 2 K/min, 4 K/min and 7 K/min up to 450 °C (see inset of Figure 7). The corresponding desorption temperature peaks were used to obtain the Kissinger plot, which allowed evaluation of the hydrogen desorption activation energy from the slope of the linear fit (Figure 7). The obtained value of 127 kJ/mol for the activation energy is significantly lower than that of bulk LiBH4, reported in the range 146 kJ/mol [15]-170 kJ/mol [4], which again confirms the catalytic role of nanosized metal ferrites synthesized.

Conclusions
The present work investigated the improved efficiency of LiBH4 desorption kinetics brought about by new synthesized catalysts based on Ni-and Co-ferrites supported on graphene scaffolds. These catalysts were prepared from graphene and the constituent nitrates by performing subsequent heat treatment in argon or (hydrogen + argon) flow at

Conclusions
The present work investigated the improved efficiency of LiBH 4 desorption kinetics brought about by new synthesized catalysts based on Ni-and Co-ferrites supported on graphene scaffolds. These catalysts were prepared from graphene and the constituent nitrates by performing subsequent heat treatment in argon or (hydrogen + argon) flow at 620 • C. The composites LiBH 4 -catalysts were prepared by prolonged mixing under protective atmosphere followed by melt infiltration under hydrogen pressure. Dehydrogenation measurements confirmed that both Ni-and Co-ferrite catalyzed graphene nanocomposites show faster desorption kinetics than pristine graphene. LiBH 4 -(NiFe 2 O 4 /graphene) using nickel ferrite catalyst synthesized in Ar flow has the lowest desorption temperature peak of 349 • C, whereas the same composition, but using treatment in H 2 /Ar, presents the highest reversible storing capacity of 6.14 wt% H 2 . The Kissinger plot for LiBH 4 -(NiFe 2 O 4 /graphene/Ar) revealed an activation energy of 127 kJ/mol. The kinetic improvement can be traced to the Fe, Ni and Co borides formed during hydrogenation. The composites with catalysts treated in Ar have lower desorption temperature due to better catalyst dispersion than using H 2 /Ar, when catalyst clustering was observed by STEM-EDS measurements. The Ni-ferrite NiFe 2 O 4 was more effective than Co-ferrite CoFe 2 O 4 in lowering the decomposition temperature of catalyzed LiBH 4 composites, and also more effective than metal borides and carbides identified in the reaction mixture as potential catalysts. Additionally, the current work highlights the role of metal ferrites, metal boride and carbide phases acting as hydrogenation catalysts, providing useful information on the actual state of catalytic sites during cycling. Considering the promising results exhibited by graphene-supported MFe 2 O 4 (M=Co, Ni), the ferrite-based supports will be further investigated in catalytic de-/rehydrogenation studies of other complex hydrides and RHCs.