Reversible hydrogen storage in yttrium aluminum hydride

School of Materials Science and Engineer Storage Materials of Guangdong Province, Guangzhou, 510641, PR China. E-mail: 87112762 China-Australia Joint Laboratory for Energy University of Technology, Guangzhou, 5106 Max-Planck-Institut für Kohlenforschung, K der Ruhr, Germany. E-mail: felderhoff@mpi Key Laboratory for Fuel Cell Technology in G PR China † Electronic supplementary informa 10.1039/c6ta10928d Cite this: J. Mater. Chem. A, 2017, 5, 6042

Pure hydrogen is an ideal energy source for proton exchange membrane (PEM) fuel cell vehicles. However, large-scale hydrogen application in the eld of PEM fuel cells, especially for those with operating temperatures below 100 C, is limited due to the absence of a safe and effective hydrogen storage approach. 1,2 A possible solution could be a novel hybrid tank system, which combines a high-pressure tank with unstable metal hydrides. This system shows obvious advantages in terms of gravimetric and/or volumetric hydrogen density compared to high pressure, solid state or liquid hydrogen storage techniques. 3 Unstable metal hydrides for hybrid tank systems have the characteristic of high desorption plateau pressures, meaning that they don't exist under ambient conditions (room temperature, 1 bar pressure). This high desorption plateau pressure is the result of a decomposition enthalpy (DH) of the metal hydrides of less than 20 kJ mol À1 H 2 . Consequently, these unstable hydrides can be synthesized only at low temperatures and/or under very high hydrogen pressure conditions. 4 To achieve needs for future metal hydride/fuel cell applications, it is important to consider the overall system performance and not just the properties of materials. Nevertheless, large hydrogen capacity, sufficient kinetics at low temperatures and excellent long-term stability are also very important material properties. To date, several unstable metal hydrides including AB 2 -type 5-7 and V-based BCC alloys 8,9 have been investigated for a hybrid tank system. Until now, their reversible hydrogen capacity (<2 wt%) is too low to meet practical requirements. It is believed that an unstable metal hydride with a reversible capacity of 4 wt% can bring this hybrid system close to an attractive level. 10 However, novel unstable hydrides like light weight alloys or complex hydride systems with higher hydrogen capacity are more favorable.
During the past two decades, light complex aluminium hydrides like NaAlH 4 , LiAlH 4 , KAlH 4 , etc., have been intensively investigated as potential candidates for solid-state hydrogen storage due to their relatively high capacity, moderate absorption/ desorption conditions and good reversibility in the presence of catalysts. [11][12][13][14][15] For example, TiCl 3 -doped NaAlH 4 showed a high reversible capacity of $4 wt% H 2 during 100 de/rehydrogenation cycles at relatively low temperatures of 70 and 270 C. 11 On the other hand, LiAlH 4 can't be synthesized directly in the solid state from commercially available LiH and Al powders. The synthesis is only possible in an ether solution. LiAlH 4 can release more than 7 wt% H 2 with an onset dehydrogenation temperature of 80 C, and the dehydrogenated sample can be recycled in an ether solution with a retention capacity of 6.4 wt% even aer 3 cycles. 12 For these light aluminium hydrides, the hydrogen capacities of the rst decomposition steps are not sufficient, while the decomposition temperatures for the second decomposition steps are too high, which makes them inappropriate for low temperature fuel cells. 16 Besides these light aluminium hydrides, several thermodynamically unstable transition metal aluminium hydrides with a high hydrogen content like Ti(AlH 4 ) 4 (9.3 wt% H 2 ), Fe(AlH 4 ) 2 (5.8 wt% H 2 ), or Y(AlH 4 ) 3 (6.6 wt% H 2 ) are described, but have attracted less attention. 17 The general synthesis of these transition metal complex aluminium hydrides M(AlH 4 ) n (M ¼ Ti, V, Co, Mn, Fe, Cu, Zr, Nb, Ag, Ce, Ta, etc.) was carried out at very low temperatures between À110 and À80 C because the decomposition and hydrogen release start in most cases at À50 C or below. 18 For instance, Ti(AlH 4 ) 4 and Fe(AlH 4 ) 2 start to decompose slowly above À80 C and two hydrogen atoms are liberated while heating to room temperature. 19 Compounds [RE(AlH 4 ) 3 ] (RE ¼ La, Ce, Pr) already start to evolve hydrogen and form RE aluminium hydride REAlH 6 and Al metal during the ball milling process of rare earth chlorides and sodium aluminium hydride. 16 Among all these transition metal complex aluminium hydrides, the relatively high stability of TaH 2 (AlH 4 ) 2 and Y(AlH 4 ) 3 is remarkable. [20][21][22] The thermal decomposition of TaH 2 (AlH 4 ) 2 occurs in the interval of 135-195 C, and the metal-hydrogen bonds are even retained aer hydrolysis. 20,21 Y(AlH 4 ) 3 was reported to be quite stable and starts to decompose at 50 C. 22 Y(AlH 4 ) 3 has a high theoretical hydrogen content of 6.6 wt%, while until now no further information is known about its hydrogen storage properties and reversibility. In this work, in order to continue to explore new unstable high capacity hydrides (>4 wt%), Y(AlH 4 ) 3 was prepared via a mechanochemical reaction. The dehydrogenation mechanism, hydrogen storage properties, reversibility and prospect for hydrogen storage were systematically evaluated. The preliminary results of the kinetics and reversibility of this alanate are quite interesting, which encourages us to place more effort into exploring new types of unstable hydrides with favorable hydrogen storage properties.
The mechanochemical metathesis reaction is a convenient and efficient procedure for the synthesis of complex aluminium hydrides. 23 Fig. 1(a) presents the XRD patterns of the as-milled YCl 3 -3LiAlH 4 mixture and LiAlH 4 milled under the same conditions. For comparison the XRD pattern of the blank sample stage was also presented. Pure LiAlH 4 remains in a highly crystalline state aer ball milling for 6 h. For the asmilled YCl 3 -3LiAlH 4 mixture, an exchange reaction between 3LiAlH 4 and YCl 3 initiated by ball milling would theoretically lead to the formation of Y(AlH 4 ) 3 and 3LiCl, while here only the diffraction peaks of LiCl can be observed with the absence of any other phases. Moreover, it is noteworthy that the background intensity of the as-milled YCl 3 -3LiAlH 4 sample between 26 and 36 is obviously higher than that of the blank sample stage, suggesting a possible amorphous nature of Y(AlH 4 ) 3 . Actually the amorphous structure was veried aer the removal of LiCl via an extraction with diethylether and drying procedure (Fig. S1 †). This result is consistent with the observation by Kost et al.,22 who also demonstrated the amorphous nature of Y(AlH 4 ) 3 . As shown in Fig. S1, † the presence of traceable diffraction peaks of Al indicated a minimal decomposition of Y(AlH 4 ) 3 during the purication process. Most of the transition tetrahydroaluminates M(AlH 4 ) n decompose at À50 C or even below while yttrium tetrahydroaluminate is an exception. 18 The appearance of the diffraction peaks of Al metal and the increase of pressure inside the jar during the milling process implied the decomposition of unstable RE(AlH 4 ) 3 hydrides. 16 In this study, Y(AlH 4 ) 3 still stays in an amorphous state aer ball milling without the observation of any decomposition products, further conrming its relatively high stability under normal conditions. Fig. 1(b) shows the FT-IR patterns of the as-milled LiAlH 4 and YCl 3 -3LiAlH 4 sample. The as-milled LiAlH 4 shows two Al-H stretching vibration frequencies at 1785 cm À1 and 1645 cm À1 , consistent with the reported values (1757 cm À1 and 1615 cm À1 ). 24 In the ngerprint region, the bands at 885, 790, and 704 cm À1 correspond to the deformational modes; meanwhile the combination band is at 1450 cm À1 . 25 The shi of the Al-H stretching band towards a higher frequency (about 30 cm À1 ) in the spectrum of the as-milled LiAlH 4 was thought to be related to the strain effect induced by milling treatment, 26 which was also observed in the as-milled KAlH 4 (ref. 27) and LiAlH 4 under a high static pressure (GPa). 28 Distinct differences exist in the FT-IR spectra of the as-milled YCl 3 -3LiAlH 4 and LiAlH 4 due to the different chemical environments of the Al-H bonds. The as-milled YCl 3 -3LiAlH 4 sample exhibits only one stretching vibration (1800 cm À1 ) and one broad deformational band at 690 cm À1 for the Al-H bond. The peak at 1640 cm À1 corresponds to the water bending vibration. 29 No bands for the LiAlH 4 can be detected in the FT-IR spectrum of the YCl 3 -3LiAlH 4 sample, implying the complete transformation from the starting materials into the product (a mixture of Y(AlH 4 ) 3 and 3LiCl, donates as Y(AlH 4 ) 3 -3LiCl) aer 6 h of ball milling. Such a result is in good agreement with the results of XRD analysis.
The thermal dehydrogenation properties of Y(AlH 4 ) 3 were measured by TPD, MS and DSC, and the results are shown in Fig. 2. Three endothermic dehydrogenation peaks are observed during the heating process. All these dehydrogenation peaks combined with the release of hydrogen gas. These three steps with different slopes can be distinguished in the TPD volumetric release curve (labeled with a serial number). One additional exothermic peak at $395 C can be observed in the DSC curve, besides the aforementioned three endothermic dehydrogenation stages (as shown by arrows). These results indicate that altogether four different steps are involved in the thermal decomposition process of Y(AlH 4 ) 3 . As shown in the TPD curve, hydrogen release from Y(AlH 4 ) 3 starts at around 80 C, accelerates at $120 C and rst peaks at $140 C. The rst dehydrogenation stage is nished at $170 C with a desorption capacity of $3.9 wt% H 2 . In the temperature range from 170 to 350 C, two dehydrogenation reactions with maxima at 245 C and 290 C can be observed. The quantitative measurement of these two processes delivers a hydrogen amount of $1.0 wt% and $0.7 wt%, respectively. A total hydrogen amount of $5.6 wt% H 2 can be released from the Y(AlH 4 ) 3 within the temperature range of 80-400 C. These desorbed hydrogen values were normalized to reect the weight of Y(AlH 4 ) 3 (overall hydrogen amount is 6.6 wt%) considering the theoretical ratio of Y(AlH 4 ) 3 and LiCl (1 : 3) in the ball milled product.
For further understanding thermal decomposition of Y(AlH 4 ) 3 , XRD analyses of the dehydrogenated samples at different temperatures were performed (see Fig. 3). Kost et al. proposed that, thermal decomposition of Y(AlH 4 ) 3 proceeds directly through YH 3 and AlH 3 derived on the basis of evolved hydrogen capacity and thermographic data. 22 However, here only the characteristic diffraction pattern of Al metal arises aer dehydrogenation at 120 C, and no information concerning these reported intermediates can be detected. This result suggests that the dehydrogenation Y(AlH 4 ) 3 16 Here in Y(AlH 4 ) 3 , however, no other Y-Al-H containing decomposition products can be identied even up to the temperature range between 170 C and 200 C by means of XRD, suggesting that this new YAlH x hydride may also be amorphous. We propose this YAlH x hydride to be YAlH 6 based on two reasons: (1) the yttrium element has a stable +3 oxidation state, which is more likely to result in the formation of [AlH 6 ] 3À ; (2) for the formation of the hexahydride, 50% of the hydrogen was expected to be released from the rst dehydrogenation step. 16 According to the TPD curve of Y(AlH 4 ) 3 in Fig. 2, a capacity of $3.9 wt% H 2 (59% of the total hydrogen) is liberated from the rst desorption reaction, which is in reasonable agreement with the results observed for the rst dehydrogenation step of RE(AlH 4 ) 3 (La 51%, Ce 60%, Pr 56%, and Nd 57%) with REAlH 6 as the intermediate phase. 16 With further increase of the temperature to 250 C, diffraction peaks of YH 3 show up; meanwhile the intensities of visible free Al metal obviously increase. This result demonstrates that the newly formed YAlH 6 decomposes into YH 3 , Al and H 2 during the second dehydrogenation stage. Above 250 C, YH 3 starts to transform into YH 2 and H 2 . When the temperature increases to 300 C, the diffraction peaks of YH 3 disappear and only those of YH 2 can be observed. As the temperature reaches 350 C, it is noted that the diffraction peaks of YH 2 completely disappear and the intensity of free Al metal also signicantly decreases. Meanwhile new intermetallic YAl 3 is formed, which results from the reaction between YH 2 and Al. Further heating the sample up to 400 C results in the complete dehydrogenation of YH 2 ; hence YAl 3 is dominant in the XRD pattern. According to the above analysis, Y(AlH 4 ) 3 is most probably made up of isolated tetrahedral [AlH 4 ] À , while the intermediate decomposition product consists of octahedral [AlH 6 ] 3À , which is very similar to the crystal structure of REAlH 6 . 16 Therefore, the experimental results obtained in the present case suggest a reaction mechanism as follows: First step (80-170 C): Y(AlH 4 ) 3 / YAlH 6 + 2Al + 3H 2  It must be mentioned that other intermediates of the general formula YAl x H y can't be excluded, which leads to the fact that the overall hydrogen capacity is reduced and that these intermediates may contribute to additional dehydrogenation steps.
To realize the low temperature (25-150 C) application of Y(AlH 4 ) 3 in hybrid tanks, only the rst dehydrogenation step can be possible. Thus isothermal desorption kinetic curves of Y(AlH 4 ) 3 in a temperature range from 80 to 140 C were studied, and the results are displayed in Fig. S3. † The wt% H 2 values are normalized to the Y(AlH 4 ) 3 percentage. Y(AlH 4 ) 3 can release $3.4 wt% H 2 within 60 min at 140 C, and ninety percent of the hydrogen can be liberated in 30 min at this temperature. The activation energy for the dehydrogenation process of Y(AlH 4 ) 3 is estimated to be 91.7 kJ mol À1 (Fig. S3d †), which is much smaller compared to that of other nanocrystalline complex metal alanates, e.g. rod Mg(AlH 4 ) 2 with 123.0 kJ mol À1 . 32 A possible explanation for the low on-set dehydrogenation temperature of 80 C could be this low kinetic barrier.
Aer dehydrogenation at 145 C, a series of recharge/ discharge experiments were performed to demonstrate the reversibility of the rst dehydrogenation step of Y(AlH 4 ) 3 . Aer each dehydrogenation, the samples were rehydrogenated at a pressure of 100 bar H 2 , and aerwards dehydrogenated in a vacuum at the same temperature. Fig. 4 shows the isothermal (a) absorption and (b) desorption kinetic curves of Y(AlH 4 ) 3 for three consecutive cycles at 145 C. As shown in Fig. 4(a), during rehydrogenation cycles an amount of $2.6 wt% H 2 can be absorbed at 145 C and the whole amount of absorbed hydrogen can be liberated during the corresponding dehydrogenation processes ( Fig. 4(b)). In general, the reversible hydrogen storage capacity at 145 C can reach 2.6 wt%, indicating the reversibility of the rst dehydrogenation step of Y(AlH 4 ) 3 . This corresponds to $75% of the theoretical hydrogen capacity of Y(AlH 4 ) 3 for the rst step. To our knowledge, this is the rst time that reversible hydrogen storage is found in any transition metal alanates. To check whether Y(AlH 4 ) 3 has a higher reversibility or not, its nal thermal decomposition product (YAl 3 ) was prepared, and was subjected to a high pressure of 300 bar. As shown in Fig. S5, † not any exothermic peaks for hydrogen absorption can be observed, indicating that the YAl 3 alloy can't absorb hydrogen in a temperature range from À40 C to 100 C and at a high pressure of 300 bar. This is also the case for the YH 3 + 3Al mixture, which cannot be hydrogenated at a pressure of 100 bar H 2 at 145 C (Fig. S6 †). These results demonstrate that only the rst dehydrogenation step of Y(AlH 4 ) 3 can be rehydrogenated. Further attempts to improve the reversibility of unstable transition complex alanates are underway in our laboratories.

Conclusions
A new complex metal aluminium hydride with the composition of Y(AlH 4 ) 3 was prepared via the mechanochemical reaction of YCl 3 + 3LiAlH 4 . Upon heating, the Y(AlH 4 ) 3 sample decomposes via a four-stage dehydrogenation process over the temperature range of 80-400 C. At 80-170 C, Y(AlH 4 ) 3 is rst decomposed into an intermediate hydride, YAlH 6 , 2Al and 3H 2 . With increasing temperature up to 250 C, YAlH 6 continues to release hydrogen to form YH 3 and additional Al metal. Upon further increasing the temperature to 300 C, YH 3 starts to decompose into YH 2 and H 2 . As the temperature reaches 350 C, the newly formed YH 2 starts to react with Al to generate YAl 3 , and this reaction proceeds completely upon further heating the sample up to 400 C. An amount of 3.4 wt% H 2 can be released from the sample within $60 min at 140 C during the rst dehydrogenation step. The apparent activation energy of the rst dehydrogenation step of Y(AlH 4 ) 3 is 92.1 kJ mol À1 . Rehydrogenation experiments indicate that the rst dehydrogenation step shows a reversibility of 75% even at a low temperature of 145 C. This is the rst example that a transition metal alanate can reversibly absorb hydrogen. Further improvements on the hydrogen storage properties of Y(AlH 4 ) 3 would make it a possible and promising candidate for hybrid tank system applications.