Dehydrogenation of Surface-Oxidized Mixtures of 2 LiBH 4 + Al / Additives ( TiF 3 or CeO 2 )

Research for suitable hydrogen storage materials is an important ongoing subject. LiBH4–Al mixtures could be attractive; however, several issues must be solved. Here, the dehydrogenation reactions of surface-oxidized 2LiBH4 + Al mixtures plus an additive (TiF3 or CeO2) at two different pressures are presented. The mixtures were produced by mechanical milling and handled under welding-grade argon. The dehydrogenation reactions were studied by means of temperature programmed desorption (TPD) at 400 ◦C and at 3 or 5 bar initial hydrogen pressure. The milled and dehydrogenated materials were characterized by scanning electron microscopy (SEM), X-ray diffraction (XRD), and Fourier transformed infrared spectroscopy (FT-IR) The additives and the surface oxidation, promoted by the impurities in the welding-grade argon, induced a reduction in the dehydrogenation temperature and an increase in the reaction kinetics, as compared to pure (reported) LiBH4. The dehydrogenation reactions were observed to take place in two main steps, with onsets at 100 ◦C and 200–300 ◦C. The maximum released hydrogen was 9.3 wt % in the 2LiBH4 + Al/TiF3 material, and 7.9 wt % in the 2LiBH4 + Al/CeO2 material. Formation of CeB6 after dehydrogenation of 2LiBH4 + Al/CeO2 was confirmed.

The last system, the 2LiBH 4 + Al, is of potential interest.Siegel et al. [18] anticipated, based on first-principles calculations, that the reaction: 2LiBH 4 + Al → AlB 2 + 2LiH + 3H 2 (1) would release 8.6 wt % hydrogen at 277  [19].Kang et al. achieved the release of 7.2 wt % hydrogen when the mixture was catalyzed with TiF 3 , at a pressure of 0.001 bar, 450 • C, and 3 h [20].Hansen et al. [17] demonstrated that the dehydrogenation reaction of 2LiBH 4 + 3Al occurs at p(H 2 ) = 10 −2 bar when the material is heated up to 500 • C; and that the re-hydrogenation demonstrated only partial reversibility.Other examples include the experiments of Ravnsbaek et al., where the dehydrogenation reaction of LiBH 4 + Al (1:0.5) was characterized by in-situ synchrotron radiation powder X-ray diffraction from room temperature to 500 • C and dynamic vacuum [21].All these experiments have in common a very low or vacuum dehydrogenation pressure (summarized in Table 1).The vacuum dehydrogenation pressure is inadequate for any practical application.Additionally, it is well-known that the hydrogen pressure can affect the dehydrogenation products of LiBH 4 mixtures.For example with the system 2LiBH 4 + MgH 2 → 2LiH + MgB 2 + 4H 2 , a correlation was demonstrated between the dehydrogenation pressure and re-hydrogenation [22].With the 2LiBH 4 -Al system, Yang et al. proposed that a desorption backpressure of 3 bar could contribute to the formation of AlB 2 [12]; which would improve, in principle, the reversibility of hydrogenation/dehydrogenation reactions.A study of the dehydrogenation backpressure with 2LiBH 4 -Al systems has been poorly explored.Among the additives for improving reaction kinetics or reversibility; TiF 3 is the material most commonly used, and it is almost mandatory to test TiF 3 in all new mixtures.In their part, oxides such as TiO 2 , ZrO 2 , Nb 2 O 5 or MoO 3 have resulted in successful accelerators for hydrogen desorption reactions [23,24].CeO 2 is not a commonly used additive for hydrogenation/dehydrogenation reactions.However, interactions of hydrogen with CeO 2 have been reported [25].The possible effects of CeO 2 as an additive for hydrogen storage materials deserve research.
Conventionally, hydrogen storage materials are produced and handled in a high or ultra-high purity inert atmosphere, i.e., high-purity argon.This is done to avoid the deactivation of materials caused by the formation of thick oxide films that impede hydrogen diffusion from/to the bulk of the storage material.However, some reports have expressed that the use of high-purity inert atmosphere can be relaxed [26] and that allowing some surface oxidation can be helpful to the dehydrogenation kinetics [27].
Here, the dehydrogenation behavior is presented of 2LiBH 4 + Al added with 5 wt % of TiF 3 or CeO 2 , and surface-modified (oxidized) by the effects of impurities in welding-grade argon.The dehydrogenation reactions were performed at 3 bar and 5 bar of hydrogen.The effects of the borohydride-Al mixture, additives, and surface-oxide effects are discussed.[7] p: 1 bar He T: from room temperature up to 210 • C for the doped systems and 110-250 • C for the undoped systems.

Characterization of As-Milled Materials
Scanning Electron Microscopy.Some of the descriptions below contain remarks about SEM images that are shown in the Supplementary Materials.Also for comparison purposes, the SEM images of LiBH 4 and Al without milling are presented in the Supplementary Materials.Non-milled LiBH 4 is composed of large crystals embedded in an amorphous phase.Non-milled Al is composed of particles of approximately 50 µm, heavily agglomerated and forming flakes of 1 mm (Supplementary Materials). Figure 1 presents the most representative SEM images of the as-milled materials.The as-milled 2LiBH 4 + Al material, Figure 1a, formed a three-dimensional, porous structure.Interestingly the surface of 2LiBH 4 + Al is covered with crystals of approximately 2-3 µm.The material 2LiBH 4 + Al/TiF 3 , Figure 1b, also presented a three-dimensional structure.The material consisted of elongated crystals of 10 µm length and 2 µm width.Elemental analysis by energy-dispersive X-ray spectroscopy (EDS) of those crystals revealed a B-rich phase, i.e., the LiBH 4 .The addition of CeO 2 , Figure 1c, to the base material, produced an agglomerated, spherical, and compacted material of about 50 µm in diameter, with some surface formations approximating flake-shape.Here, spots of CeO 2 are clearly distinguishable from the base material.In principle, the morphology characteristics of hydrogen storage materials must be reduced particle size, low agglomeration, homogeneity of component materials, and the formation of porous structures allowing the inflow/outflow of hydrogen while maintaining good thermal conductivity.SEM images showed interesting morphologies of the LiBH 4 + Al/additives, depending on the additive material.p: not specified (argon) T: 300-550 °C Cl − to BH4 − substitution at LiBH4 2LiBH4 + Al → 2LiH + AlB2 + 3H2 57.9 kJ/mol H2 [18] 277 °C [18] Theoretical desorption temperature Dehydrogenation: p: 0.001 bar H2 T: 325 °C to 525 °C [19] H2 release of about 4 wt % Multi-step dehydrogenation reaction p: 0.001 bar T: 450 °C [20] Catalyzed with TiF3 p: dynamic vacuum T: up to 500 °C [21] Formation of LixAl1−xB2

Characterization of As-Milled Materials
Scanning Electron Microscopy.Some of the descriptions below contain remarks about SEM images that are shown in the Supplementary Materials.Also for comparison purposes, the SEM images of LiBH4 and Al without milling are presented in the Supplementary Materials.Non-milled LiBH4 is composed of large crystals embedded in an amorphous phase.Non-milled Al is composed of particles of approximately 50 μm, heavily agglomerated and forming flakes of 1 mm (Supplementary Materials). Figure 1 presents the most representative SEM images of the as-milled materials.The as-milled 2LiBH4 + Al material, Figure 1a, formed a three-dimensional, porous structure.Interestingly the surface of 2LiBH4 + Al is covered with crystals of approximately 2-3 μm.The material 2LiBH4 + Al/TiF3, Figure 1b, also presented a three-dimensional structure.The material consisted of elongated crystals of 10 μm length and 2 μm width.Elemental analysis by energydispersive X-ray spectroscopy (EDS) of those crystals revealed a B-rich phase, i.e., the LiBH4.The addition of CeO2, Figure 1c, to the base material, produced an agglomerated, spherical, and compacted material of about 50 μm in diameter, with some surface formations approximating flakeshape.Here, spots of CeO2 are clearly distinguishable from the base material.In principle, the morphology characteristics of hydrogen storage materials must be reduced particle size, low agglomeration, homogeneity of component materials, and the formation of porous structures allowing the inflow/outflow of hydrogen while maintaining good thermal conductivity.SEM images showed interesting morphologies of the LiBH4 + Al/additives, depending on the additive material.Powder X-ray diffraction.Figure 2 shows the X-ray diffraction patterns of as-milled materials.In the three studied materials, the presence of Al is evidenced by the strong diffraction peaks.By contrast, the LiBH 4 presented attenuated peaks as as-milled materials.LiBH 4 diffraction peaks in Figure 2 are a mixture of orthorhombic (main) and hexagonal (minor) phases to a different degree in the different mixtures.The material 2LiBH 4 + Al/CeO 2 presented strong peaks of CeO 2 , in agreement with the CeO 2 particles found in the SEM images.The material added with TiF 3 did not present the characteristic peaks of TiF 3 .Some researchers have proposed the in-situ formation of Ti(BH 4 ) 3 by the reaction of LiBH 4 and TiF 3 , followed by a rapid decomposition of Ti(BH 4 ) 3 [29].Reactions to form other stable fluorine salts such as TiF 2 or AlF 3 cannot be discarded.These side reactions during ball-milling can be the reason for the absence of TiF 3 diffraction peaks.The materials also presented a small amount of aluminum oxide.Powder X-ray diffraction.Figure 2 shows the X-ray diffraction patterns of as-milled materials.In the three studied materials, the presence of Al is evidenced by the strong diffraction peaks.By contrast, the LiBH4 presented attenuated peaks as as-milled materials.LiBH4 diffraction peaks in Figure 2 are a mixture of orthorhombic (main) and hexagonal (minor) phases to a different degree in the different mixtures.The material 2LiBH4 + Al/CeO2 presented strong peaks of CeO2, in agreement with the CeO2 particles found in the SEM images.The material added with TiF3 did not present the characteristic peaks of TiF3.Some researchers have proposed the in-situ formation of Ti(BH4)3 by the reaction of LiBH4 and TiF3, followed by a rapid decomposition of Ti(BH4)3 [29].Reactions to form other stable fluorine salts such as TiF2 or AlF3 cannot be discarded.These side reactions during ballmilling can be the reason for the absence of TiF3 diffraction peaks.The materials also presented a small amount of aluminum oxide.Refinement of the Al and CeO2 phases (not including LiBH4 phases, Rwp ≈ 7) produced the following Al-crystal sizes: Al in 2LiBH4 + Al: 309.6 ± 16.7 nm; Al in 2LiBH4 + Al/TiF3: 129.0 ± 5.0 nm; Al in 2LiBH4 + Al/CeO2: 142.8 ± 9.2 nm.The CeO2 crystal size was determined as 120.3 ± 4.2 nm.All these facts point to the formation of fine mixtures of 2LiBH4 + Al/additive and an influence of the additives on the crystalline characteristics of the samples.Fourier Transformed Infrared Spectroscopy.An effect of the ball milling is the increase of the local pressure and temperature during ball-ball or ball-vial collisions that induce the decomposition of sensitive materials.In the case of the 2LiBH4 + Al mixtures, it is intended to decrease the decomposition temperature during dehydrogenation, but not to decompose during ball-milling.Thus FT-IR was performed to check the "survival" of LiBH4 after ball-milling.Borohydrides present two regions of interest B-H bending (1000-1600 cm −1 ) and H-B-H stretching (2000-2500 cm −1 ) [29,30].Both IR active modes are presented in all the as-milled samples (Supplementary Materials), indicating sufficient thermal stability during ball milling.
X-ray photoelectron spectroscopy.XPS results are presented in Figure 3. Frame 3a presents the Li 1s XPS spectra.The Li 1s XPS spectrum of LiBH4 with a clean surface was reported to present a peak at 57.1 eV; meanwhile, the oxygen-exposed samples presented a contribution (shoulder) of Li2O at 55.5 eV [27].The studied materials present the main peak in a range between 57.2 and 57.5 eV.Because of that, the main contribution of Li binding is not the Li-O interactions.Frame 3(b) presents the B 1s XPS spectra.The reported peak for pure LiBH4 is located at 188.3-188.4eV [31].After exposing LiBH4 to oxygen or moisture, a new peak located at 191. 5-191.4eV emerged and it was attributed to LiBO2 [31].The studied materials presented both peaks, the LiBH4 and the LiBO2 at 188.3-188.0eV and 191.9-191.3eV, respectively.The peak intensity ratio of the LiBH4 and LiBO2 is not homogeneous throughout the set of studied materials.The mixture of 2LiBH4 + Al presented the lowest intensity of the LiBO2 peak.Frame c of Figure 3 presents the O 1s XPS spectra; as a reference, Fourier Transformed Infrared Spectroscopy.An effect of the ball milling is the increase of the local pressure and temperature during ball-ball or ball-vial collisions that induce the decomposition of sensitive materials.In the case of the 2LiBH 4 + Al mixtures, it is intended to decrease the decomposition temperature during dehydrogenation, but not to decompose during ball-milling.Thus FT-IR was performed to check the "survival" of LiBH 4 after ball-milling.Borohydrides present two regions of interest B-H bending (1000-1600 cm −1 ) and H-B-H stretching (2000-2500 cm −1 ) [29,30].Both IR active modes are presented in all the as-milled samples (Supplementary Materials), indicating sufficient thermal stability during ball milling.
X-ray photoelectron spectroscopy.XPS results are presented in Figure 3. Frame 3a presents the Li 1s XPS spectra.The Li 1s XPS spectrum of LiBH 4 with a clean surface was reported to present a peak at 57.1 eV; meanwhile, the oxygen-exposed samples presented a contribution (shoulder) of Li 2 O at 55.5 eV [27].The studied materials present the main peak in a range between 57.2 and 57.5 eV.Because of that, the main contribution of Li binding is not the Li-O interactions.Frame 3(b) presents the B 1s XPS spectra.The reported peak for pure LiBH 4 is located at 188.3-188.4eV [31].After exposing LiBH 4 to oxygen or moisture, a new peak located at 191. 5-191.4eV emerged and it was attributed to LiBO 2 [31].The studied materials presented both peaks, the LiBH 4 and the LiBO 2 at 188.3-188.0eV and 191.9-191.3eV, respectively.The peak intensity ratio of the LiBH 4 and LiBO 2 is not homogeneous throughout the set of studied materials.The mixture of 2LiBH 4 + Al presented the lowest intensity of the LiBO 2 peak.Frame c of Figure 3 presents the O 1s XPS spectra; as a reference, the O 1s spectra at the Al raw-material was included.In that spectrum, the peak at 533.6 eV can be related to an Al 2 O 3 layer [32].The XPS spectra of LiBH 4 , 2LiBH 4 + Al/TiF 3 , and 2LiBH 4 + Al/CeO 2 presented two main peaks; one at 533.9 eV and other at 532.6 eV.This last peak can be attributed to the LiBO 2 [32].The LiBO 2 peak is very much attenuated in the 2LiBH 4 + Al, in agreement with the result at the B 1s spectra.The Al 2p spectrum of the as-received Al and as-milled 2LiBH 4 + Al/additive materials is shown in the Supplementary Materials.The (as-received) Al curve presents the characteristic metallic and oxide peaks of Al [33].Meanwhile, no signal above noise was detected for the mixtures, indicating Al segregation to sub-surface layers.

Dehydrogenation Reactions
Figure 4 presents the dehydriding reactions traced by thermal-programed control.For two frames, the sample temperature also was plotted.The dehydrogenations of 2LiBH4 + Al and 2LiBH4 + Al/TiF3 at 3 bar and 5 bar initial pressure are presented in Figure 4a.The dehydrogenations were performed up to 400 °C.The dehydrogenations occurred as multiple-step reactions.The first step with both materials was observed to start at 110 °C and produced a small hydrogen release.In the 2LiBH4 + Al material, the second step was observed at 210 °C, and a slight change in reaction rate at roughly 325 °C.The dehydrogenation reactions in the 2LiBH4 + Al material were essentially finished at 400 °C and reached a hydrogen release of −5.8 wt % for the 3 bar initial pressure reaction and −6.0 wt % for the 5 bar initial pressure reaction.The dehydrogenations of 2LiBH4 + Al/TiF3 at 3 bar and 5 bar initial pressure also are multiple-step reactions.At both initial pressures, the dehydrogenation first step onset is situated at about 110 °C.At the 3 bar dehydrogenation, the second reaction step is located at 240 °C.The last step initiated at 340 °C, after a clear-defined plateau.At 3 bar initial pressure the release of hydrogen accounted for −9.3 wt %.The situation is very different for the same material at 5 bar initial pressure, here the onset of the second dehydrogenation step was observed at 200 °C.A slowdown of the dehydrogenation rate was observed at 300 °C.Finally, the dehydrogenation finished after reaching 400 °C, releasing −6.0 wt %.This last quantity is greater than the theoretical value of −8.6 wt %, thus a release of B-compounds is probable.It means that higher initial pressure in combination with lower final temperature needs to be explored for this material.
Figure 4b presents the dehydrogenation reactions of 2LiBH4 + Al/CeO2 at 3 bar and 5 bar initial hydrogen pressure up to 400 °C.The dehydrogenations reactions are similar at both pressures and  The last step initiated at 340 • C, after a clear-defined plateau.At 3 bar initial pressure the release of hydrogen accounted for −9.3 wt %.The situation is very different for the same material at 5 bar initial pressure, here the onset of the second dehydrogenation step was observed at 200 • C. A slowdown of the dehydrogenation rate was observed at 300 • C. Finally, the dehydrogenation finished after reaching 400 • C, releasing −6.0 wt %.This last quantity is greater than the theoretical value of −8.6 wt %, thus a release of B-compounds is probable.It means that higher initial pressure in combination with lower final temperature needs to be explored for this material.

Dehydrogenation Reactions
they are also multistep reactions.The onset temperature of the first dehydrogenation step is below 100 °C.The dehydrogenation reactions presented the second step starting at 220 °C.The material 2LiBH4 + Al/CeO2 released −7.9 wt % of hydrogen in both cases.In the material 2LiBH4 + Al/CeO2, dehydrogenation at 350 °C was achieved and can be observed in the Supplementary Materials.The dehydrogenation is a multi-step reaction with a hydrogen release of −5.6 wt % at 5 bar and −6.5 wt % at 3 bar after 3 h of reaction.

Characterization of Dehydrogenated Materials
Figure 5 presents the most representative SEM images of the 2LiBH4 + Al, 2LiBH4 + Al/TiF3, and 2LiBH4 + Al/CeO2 materials dehydrogenated at 5 bar, other SEM images are available in the Supplementary Materials.In general, dehydrogenated materials presented important agglomerations of about 200 μm size.Thus an important increase in particle size after dehydrogenation was observed.The agglomerated materials were composed of an amorphous matrix with crystalline zones of cubic or hexagonal morphologies.Elemental analysis of those crystals revealed the presence of B and Al as major components; and Ti, F, Ce, and O as minor components, accordingly with the additive.An important observation is that the atomic composition of Al and B in the crystals of all the tested materials was not consistent.The elemental analysis was performed in small areas on the crystals, and the results range from roughly the expected 1:2 atomic ratio of AlB2 to 1:12 atomic ratio of AlB12.The loss of available B for the re-hydrogenation reaction is a common drawback for all borohydrides and their mixtures.B can be lost as borane compounds (BH3, B2H6, etc.), form B-clusters or other boranes that are unreactive at moderate pressures and temperature for further re-hydrogenation reaction.Elemental analysis during SEM data collection indicates good B retention with the dehydrogenated samples.
Due to the detection limit of elemental analysis with the SEM microscope, Li-compounds were not located.The SEM images of the materials with CeO2 deserve some attention; the as-milled materials presented clear CeO2 bright spots (Figure 1c); meanwhile, the dehydrogenated materials (Supplementary Materials) presented a more homogeneous distribution of Ce along the sample image.Another interesting point was a tendency to formation of crystals with higher dehydrogenation pressure.A greater quantity of crystals was observed at samples dehydrogenated at 5 bar initial pressure than at 3 bar initial pressure.The dehydrogenation is a multi-step reaction with a hydrogen release of −5.6 wt % at 5 bar and −6.5 wt % at 3 bar after 3 h of reaction.

Characterization of Dehydrogenated Materials
Figure 5 presents the most representative SEM images of the 2LiBH 4 + Al, 2LiBH 4 + Al/TiF 3 , and 2LiBH 4 + Al/CeO 2 materials dehydrogenated at 5 bar, other SEM images are available in the Supplementary Materials.In general, dehydrogenated materials presented important agglomerations of about 200 µm size.Thus an important increase in particle size after dehydrogenation was observed.The agglomerated materials were composed of an amorphous matrix with crystalline zones of cubic or hexagonal morphologies.Elemental analysis of those crystals revealed the presence of B and Al as major components; and Ti, F, Ce, and O as minor components, accordingly with the additive.An important observation is that the atomic composition of Al and B in the crystals of all the tested materials was not consistent.The elemental analysis was performed in small areas on the crystals, and the results range from roughly the expected 1:2 atomic ratio of AlB 2 to 1:12 atomic ratio of AlB 12 .The loss of available B for the re-hydrogenation reaction is a common drawback for all borohydrides and their mixtures.B can be lost as borane compounds (BH 3 , B 2 H 6 , etc.), form B-clusters or other boranes that are unreactive at moderate pressures and temperature for further re-hydrogenation reaction.Elemental analysis during SEM data collection indicates good B retention with the dehydrogenated samples.
Due to the detection limit of elemental analysis with the SEM microscope, Li-compounds were not located.The SEM images of the materials with CeO 2 deserve some attention; the as-milled materials presented clear CeO 2 bright spots (Figure 1c); meanwhile, the dehydrogenated materials (Supplementary Materials) presented a more homogeneous distribution of Ce along the sample image.Another interesting point was a tendency to formation of crystals with higher dehydrogenation pressure.A greater quantity of crystals was observed at samples dehydrogenated at 5 bar initial pressure than at 3 bar initial pressure.FT-IR spectra of all dehydrogenated products are presented in Figure 7.This figure confirms the incomplete dehydrogenation of the materials after reactions.The dehydrogenated materials presented differences in the H-B-H bending region compared to the as-milled materials.First, a shift of the peaks to higher wavenumbers was observed, from 1100-1200 cm −1 in as-milled materials to 1400-1560 cm −1 in dehydrogenated materials.Secondly, the intensity of these peaks was different with the additive; the 2LiBH4 + Al material presented the highest peak intensity meanwhile the materials added with TiF3 and CeO2 presented reduced signal intensity.The B-H stretching region (2000-2500 cm −1 ) also presented shifts in wavenumber, however, this shift was not as marked as the H-B-H region.It amounts to approximately 75 cm −1 .The materials dehydrogenated at 5 bar presented more intense H-B-H peaks than the materials dehydrogenated at 3 bar.This result points to a different reaction mechanism.FT-IR spectra of all dehydrogenated products are presented in Figure 7.This figure confirms the incomplete dehydrogenation of the materials after reactions.The dehydrogenated materials presented differences in the H-B-H bending region compared to the as-milled materials.First, a shift of the peaks to higher wavenumbers was observed, from 1100-1200 cm −1 in as-milled materials to 1400-1560 cm −1 in dehydrogenated materials.Secondly, the intensity of these peaks was different with the additive; the 2LiBH 4 + Al material presented the highest peak intensity meanwhile the materials added with TiF 3 and CeO 2 presented reduced signal intensity.The B-H stretching region (2000-2500 cm −1 ) also presented shifts in wavenumber, however, this shift was not as marked as the H-B-H region.It amounts to approximately 75 cm −1 .The materials dehydrogenated at 5 bar presented more intense H-B-H peaks than the materials dehydrogenated at 3 bar.This result points to a different reaction mechanism.

Discussion
Milling effects.A proper integration of LiBH4 and Al in the mixtures was observed after the ballmilling.The milling of Al is not an easy process.Al has the tendency of sintering instead of dispersing unless using proper milling conditions [34].The milling conditions used here represent a good balance between the optimum milling of Al and reduced degradation of LiBH4.
Protective atmosphere effects.Customarily; the storage, production, processing, and handling of hydrogen storage materials are performed with a high purity argon atmosphere, with oxygen and moisture levels below 0.1 ppm.This would raise significantly the cost of the hydrogen storage materials if commercialization is intended.In our experiments the argon purity was not so high, the supplier-guaranteed oxygen and moisture levels were 10 ppm.This allowed surface oxidation as demonstrated by means of XPS experiments.Kato et al. [27] demonstrated that the surface oxidation of LiBH4 produces Li surface segregation, the formation of Li2O over LiBH4, reduction of diborane desorption, and enhancement of the rate of hydrogen desorption.XPS suggests the formation of LiBO2 rather than Li2O on the surface of our samples and the segregation of Al to sub-surface layers.As a result of the surface oxidation, a reduction of the dehydrogenation temperature was observed here and compared with similar materials carefully protected against surface oxidation [17][18][19][20][21][22].A reduction of the dehydrogenation activation barrier was proposed by Kang et al. [35] if LiBH4 or the intermediate LiBH donate one electron (each), to a catalyst on their surfaces.Oxygen is wellknown as an electron acceptor.Thus surface oxidation could reduce the activation barrier and help to reach thermodynamic-predicted temperatures for the dehydrogenation reaction.

Discussion
Milling effects.A proper integration of LiBH 4 and Al in the mixtures was observed after the ball-milling.The milling of Al is not an easy process.Al has the tendency of sintering instead of dispersing unless using proper milling conditions [34].The milling conditions used here represent a good balance between the optimum milling of Al and reduced degradation of LiBH 4 .
Protective atmosphere effects.Customarily; the storage, production, processing, and handling of hydrogen storage materials are performed with a high purity argon atmosphere, with oxygen and moisture levels below 0.1 ppm.This would raise significantly the cost of the hydrogen storage materials if commercialization is intended.In our experiments the argon purity was not so high, the supplier-guaranteed oxygen and moisture levels were 10 ppm.This allowed surface oxidation as demonstrated by means of XPS experiments.Kato et al. [27] demonstrated that the surface oxidation of LiBH 4 produces Li surface segregation, the formation of Li 2 O over LiBH 4 , reduction of diborane desorption, and enhancement of the rate of hydrogen desorption.XPS suggests the formation of LiBO 2 rather than Li 2 O on the surface of our samples and the segregation of Al to sub-surface layers.As a result of the surface oxidation, a reduction of the dehydrogenation temperature was observed here and compared with similar materials carefully protected against surface oxidation [17][18][19][20][21][22].A reduction of the dehydrogenation activation barrier was proposed by Kang et al. [35] if LiBH 4 or the intermediate LiBH donate one electron (each), to a catalyst on their surfaces.Oxygen is well-known as an electron acceptor.Thus surface oxidation could reduce the activation barrier and help to reach thermodynamic-predicted temperatures for the dehydrogenation reaction.
Dehydrogenation kinetics.The dehydrogenations of the 2LiBH 4 + Al, 2LiBH 4 + Al/TiF 3 , and 2LiBH 4 + Al/CeO 2 materials are multi-step reactions.The multistep nature of the dehydrogenation reaction of the 2LiBH 4 + Al is shared with the dehydrogenation of LiBH 4 [36] and the RHC LiBH 4 + MgH 2 [22,37].Reports of LiBH 4 -Al dehydrogenation in several molar proportions also described a multistep mechanism for the dehydrogenation reaction [17, [19][20][21]38].In the materials presented here, the first dehydrogenation step occurred at low temperature, i.e., 100-110 • C.And the main dehydrogenation step occurred between 200-300 • C, finishing at 400 • C. The main dehydrogenation temperature interval is close to the temperature predicted by Siegel et al. for the dehydrogenation reaction of 2LiBH 4 + Al, 277 • C [18].A reduction of activation barrier can be responsible for the reduced dehydrogenation temperatures, as pointed out above.Another good point concerning all the studied materials is that the dehydrogenation is rapid, completed within 1 h, mainly during the heating period.
The effect of dehydrogenation pressure.It must be pointed out that this is the first report of dehydrogenation of 2LiBH 4 + Al/additive mixtures at non-vacuum pressures, i.e., at fuel cell compatible working pressures.Thus the possibility of using these mixtures has been demonstrated for disposable or non-rechargeable hydrogen storage applications as long as further re-hydrogenation can be proved.In general, our materials dehydrogenated at 3 bar presented a better hydrogen release quantity.Further research will be conducted to prove if re-hydrogenation is possible.
The effect of the additives.TiF 3 is a well-known additive in the hydrogen storage area, it is almost mandatory to try TiF 3 as an additive in hydrogen storage systems.Meanwhile, CeO 2 is not commonly used as an additive.CeO 2 has been anticipated to interact with hydrogen [25], and a revision of its effects as additive deserve attention.However, few examples have been published.Ceria could uptake small amounts of hydrogen below 391 • C [39].Even more, Lin et al. [39] demonstrated an interfacial effect of CeH 2.73 /CeO 2 functioning as "hydrogen pump" and reducing the hydrogen desorption temperature of MgH 2 -Mg 2 NiH 4 -CeH 2.73 /CeO 2 .In that work [39], CeH 2.73 and CeO 2 were formed during successive hydrogenation and oxidation reactions.LiBH 4 or LiBH 4 + 1/2MgH 2 were mixed with CeCl 3 , CeF 3 , or CeH 2 to improve the hydrogenation/dehydrogenation kinetics, reversibility, or to reduce dehydrogenation temperature [40][41][42][43][44].In the present work, CeO 2 demonstrated its effectiveness reducing the dehydrogenation temperature.The characterization of dehydrogenated materials by powder X-ray diffraction demonstrated the formation of CeB 6 .This boride has been observed after dehydrogenation of LiBH 4 or LiBH 4 + 1/2MgH 2 destabilized with CeCl 3 , CeF 3 , or CeH 2 [40][41][42][43][44]. On the other hand, X-ray diffraction did not show additional formation of Li or Al-oxides by liberation of the oxygen of CeO 2 .The expected formation of oxides is low: for example, a mass balance indicates that if all the oxygen of CeO 2 forms Li 2 O, the lithium oxide would be 1.7 wt %.To conclude this part, CeO 2 additive was effective in reducing the dehydrogenation temperature and producing good hydrogen release.

Sample Preparation
All reactives were purchased from Sigma-Aldrich and used without further purification.The Al was granular, with a particle size roughly of 1 mm and 99.7% purity.The LiBH 4 purity was ≥95%, meanwhile, the purity of TiF 3 and CeO 2 were 99% and 99.995% respectively.The molar ratio of LiBH 4 and Al was 2:1.The amount of the additive in each sample was 5 wt %.The mixtures of 2LiBH 4 + Al + additives were produced by mechanical milling.The milling was performed in batches of 1 gram of mixture as needed for performing reactions or characterization.The milling was performed in a planetary mill (Across-International) with a rotation of the main plate of 2400 rpm.The milling vials were machinated in stainless steel 316 L with an internal volume of 100 mL, with bolted lids.The milling balls were of yttrium-stabilized zirconium oxide (1 cm diameter).The powder to ball ratio was 1:15.The total milling time was 5 h divided into periods of 1 h milling and 10 min resting.In each cycle of milling-pause, the rotation direction of the planetary mill was inverted.The handling and storage of materials were performed inside a glove box filled with welding grade argon.

Characterization of the Ball-Milled and Dehydrogenated Materials
Scanning electron microscopy (SEM) images were obtained in a JSM-IT300 microscope (JEOL, Tokio, Japan).Samples were dispersed on carbon tape over a Cu sample holder.The SEM samples were prepared inside the argon glove box and transferred to the microscope by means of a glove bag to avoid oxidation; however, slight oxidation could have been possible.SEM images were obtained by backscattered or secondary electrons and 10 kV or 20 kV of acceleration voltage, accordingly to each sample characteristics.
Powder X-ray diffraction (PXRD) characterization was performed in a BrukerB8 diffractometer (Cu Kα = 1.540598Å, Bruker AXS, Karlsruhe, Germany).The powders of as-milled and dehydrogenated materials were compacted in a dedicated sample-holder, then they were covered with a Kapton foil for protection against ambient oxygen and moisture.Data processing and phase identification were performed with Diffract Suite Eva (Version 4.2.1.10,Bruker AXS, Karlsruhe, Germany, 2016) or MAUD software (Version 2.55, Trento University, Trento, Italy, 2015).ICSD (Inorganic Crystal Structure Database, Karlsruhe) or COD (Crystallography Open Database) databases were used for phase identification.
Fourier transformed infrared spectroscopy characterization was performed in a Varian 640-IR, FT-IR Spectrometer (Agilent Technologies, Santa Clara, CA, USA).The studied materials were compacted in KBr pellets.The KBr was purchased from Sigma-Aldrich and dried just before pellet preparation.About 2.5 mg of each material was dispersed in 50 mg of dry KBr.FT-IR data was collected in attenuated total reflection (ATR) mode.
X-ray photoelectron spectroscopy analyses were performed only on the as-milled materials.XPS experiments were accomplished in an ultra-high vacuum (UHV) system Scanning XPS microprobe PHI 5000 Versa Probe II (Physical Electronics, Minneapolis, United States of America), with an Al Kα X-ray source (photon energy of 1486.6 eV) monochromatic at 25.4 W, and an multichannel detector.The surface of the samples was etched for 5 min with 1 kV Ar + at 55.56 nA•mm −2 .The XPS spectra were obtained at 45 • to the normal surface in the constant pass energy mode, E 0 = 117.40 and 11.75 eV for survey surface and high-resolution narrow scan, respectively.The peak positions were referenced to the background silver 3d 5/2 photo-peak at 368.2 eV, having an FWHM of 0.56 eV, and C 1s hydrocarbon groups at 285.0 eV, Au 4f 7/2 in 84.0 eV central peak core level position.XPS characterization was performed on the as-milled materials and raw materials NaBH 4 and Al as necessary.Powder samples were compacted in an adequate sample holder and transferred to the equipment by means of a glove bag.

Dehydrogenation Reaction
Dehydrogenations of 2LiBH 4 + Al + additive materials were performed in a Sievert's-type reactor.This reactor was designated and constructed by the research group.It consists basically of twins of a sample holder and a reference holder, a well-known-volume reservoir of H 2 for sample and a well-known-volume reservoir of H 2 for reference, high precision pressure transducers for sample and reference, and delicate control of the reservoirs, reference-holder and sample-holder temperatures.The registered temperatures and pressures versus time were converted to hydrogen release in wt % with the following formula [45]: where M(H 2 ) is the hydrogen molar mass (2.01588 g•mol −1 ).∆p= ∆p sample − ∆p reference is in bar, where ∆p sample and ∆p reference mean the actual pressure minus the initial pressure of sample and reference.At zero-time both sample and reference initial pressures are equal.This performs as a differential pressure transducer and helps reducing small (1 × 10 −3 bar) variations of the pressures caused by thermal effects.The V and T are the volume and temperature of the reservoir and sample holder, in cm 3 and Kelvin degrees respectively; m is the sample mass in g; R is the gas constant (83.14459 cm 3 •bar•K −1 •mol −1 ) and, Z fact is the hydrogen compressibility factor [46,47].It is necessary to mention that the sample holder and reference holder volume account for less than 1% of the reservoirs volume, meeting the appropriate conditions for hydrogen sorption/desorption and Sieverts law.Samples were transferred to/from the Sieverts-type reactor without oxygen contact by means of a closing valve at the sample holder.Dehydrogenations were performed by a temperature-controlled process.The dehydrogenation initial pressure was fixed manually at 3 or 5 bar, the reservoir temperature was fixed at 40 • C. The sample temperature was raised from room temperature to 350 • C or 400 • C with a heating rate of 5 or 6 • C/min.The total dehydrogenation time was 3 h.Then, the system was cooled down and the remaining hydrogen was released.The dehydrogenation reaction was marked by a significant and sudden increase of the registered pressure beyond the temperature effects.The gases, hydrogen and argon, used during the dehydrogenation experiments were of chromatographic and high purity grade.

Conclusions
The 2LiBH4 + Al/additives mixtures were prepared by ball milling.The milling conditions were optimized for the integration of Al and preservation of 2LiBH 4 .The use of additives TiF 3 and CeO 2 produced different morphologies with as-milled materials.The studied materials presented a significant reduction of the dehydrogenation temperature that can be related to the surface-oxidation.The surface oxidation was the result of the use of welding-grade argon.The dehydrogenation reactions were observed to take place in two main steps, with onsets at 100 • C and 200-300 • C. The maximum released hydrogen was 9.3 wt % in the 2LiBH 4 + Al/TiF 3 material, and 7.9 wt % in the 2LiBH 4 + Al/CeO 2 material.Formation of CeB 6 after dehydrogenation of 2LiBH 4 + Al/CeO 2 was confirmed.

Figure 4
Figure4presents the dehydriding reactions traced by thermal-programed control.For two frames, the sample temperature also was plotted.The dehydrogenations of 2LiBH 4 + Al and 2LiBH 4 + Al/TiF 3 at 3 bar and 5 bar initial pressure are presented in Figure4a.The dehydrogenations were performed up to 400 • C. The dehydrogenations occurred as multiple-step reactions.The first step with both materials was observed to start at 110 • C and produced a small hydrogen release.In the 2LiBH 4 + Al material, the second step was observed at 210 • C, and a slight change in reaction rate at roughly 325 • C. The dehydrogenation reactions in the 2LiBH 4 + Al material were essentially finished at 400 • C and reached a hydrogen release of −5.8 wt % for the 3 bar initial pressure reaction and −6.0 wt % for the 5 bar initial pressure reaction.The dehydrogenations of 2LiBH 4 + Al/TiF 3 at 3 bar and 5 bar initial pressure also are multiple-step reactions.At both initial pressures, the dehydrogenation first step onset is situated at about 110 • C. At the 3 bar dehydrogenation, the second reaction step is located at 240 • C. The last step initiated at 340 • C, after a clear-defined plateau.At 3 bar initial pressure the release of hydrogen accounted for −9.3 wt %.The situation is very different for the same material at 5 bar initial pressure, here the onset of the second dehydrogenation step was observed at 200 • C. A slowdown of the dehydrogenation rate was observed at 300 • C. Finally, the dehydrogenation finished after reaching 400 • C, releasing −6.0 wt %.This last quantity is greater than the theoretical value of −8.6 wt %, thus

Figure
Figure4bpresents the dehydrogenation reactions of 2LiBH 4 + Al/CeO 2 at 3 bar and 5 bar initial hydrogen pressure up to 400 • C. The dehydrogenations reactions are similar at both pressures and they are also multistep reactions.The onset temperature of the first dehydrogenation step is below 100 • C. The dehydrogenation reactions presented the second step starting at 220 • C. The material 2LiBH 4 + Al/CeO 2 released −7.9 wt % of hydrogen in both cases.In the material 2LiBH 4 + Al/CeO 2, dehydrogenation at 350 • C was achieved and can be observed in the Supplementary Materials.The dehydrogenation is a multi-step reaction with a hydrogen release of −5.6 wt % at 5 bar and −6.5 wt % at 3 bar after 3 h of reaction.

Figure 6 .
Figure 6.X-ray diffraction patterns of dehydrogenated materials.Frame (a) after dehydrogenation at 3 bar initial hydrogen pressure; Frame (b) after dehydrogenation at 5 bar initial hydrogen pressure.

Figure 6 .
Figure 6.X-ray diffraction patterns of dehydrogenated materials.Frame (a) after dehydrogenation at 3 bar initial hydrogen pressure; Frame (b) after dehydrogenation at 5 bar initial hydrogen pressure.

Table 1 .
Reported hydrogen desorption conditions for several LiBH 4 mixtures.

or Proposed Reaction, and Reported ∆H 0 (If Available) (kJ/mol H 2 ) Desorption Conditions p (bar) and T ( • C) Comments
Refinement of the Al and CeO 2 phases (not including LiBH 4 phases, R wp ≈ 7) produced the following Al-crystal sizes: Al in 2LiBH 4 + Al: 309.6 ± 16.7 nm; Al in 2LiBH 4 + Al/TiF 3 : 129.0 ± 5.0 nm; Al in 2LiBH 4 + Al/CeO 2 : 142.8 ± 9.2 nm.The CeO 2 crystal size was determined as 120.3 ± 4.2 nm.All these facts point to the formation of fine mixtures of 2LiBH 4 + Al/additive and an influence of the additives on the crystalline characteristics of the samples.
[33]tra at the Al raw-material was included.In that spectrum, the peak at 533.6 eV can be related to an Al2O3 layer[32].The XPS spectra of LiBH4, 2LiBH4 + Al/TiF3, and 2LiBH4 + Al/CeO2 presented two main peaks; one at 533.9 eV and other at 532.6 eV.This last peak can be attributed to the LiBO2[32].The LiBO2 peak is very much attenuated in the 2LiBH4 + Al, in agreement with the result at the B 1s spectra.The Al 2p spectrum of the as-received Al and as-milled 2LiBH4 + Al/additive materials is shown in the Supplementary Materials.The (as-received) Al curve presents the characteristic metallic and oxide peaks of Al[33].Meanwhile, no signal above noise was detected for the mixtures, indicating Al segregation to sub-surface layers.