Ca(BH 4 ) 2 –Mg 2 NiH 4 : on the pathway to a Ca(BH 4 ) 2 system with a reversible hydrogen cycle

The Ca(BH 4 ) 2 –Mg 2 NiH 4 system presented here is, to the best of our knowledge, the first described Ca(BH 4 ) 2 -based hydride composite that reversibly transfers boron from the Ca-based compound(s) to the reaction partner. The ternary boride MgNi 2.5 B 2 is formed upon dehydrogenation and the formation of Ca(BH 4 ) 2 upon rehydrogenation is confirmed.

Due to their high gravimetric H 2 density, metal borohydrides 1 are promising candidates for solid state hydrogen storage. However, high thermal stabilities as well as poor reversibility of their decomposition products are the major obstacles to be overcome for technical applications. One of the most promising approaches to solve the reversibility issue of borohydrides is to combine them with selected materials to create reversible reactive compositions with lowered reaction enthalpy, the so-called reactive hydride composites (RHCs). 2,3 The system that has attracted the most attention in recent years is LiBH 4 -MgH 2 . 4 Because of its reduced desorption enthalpy compared to LiBH 4 , mixtures of hydrides based on Ca(BH 4 ) 2 were also investigated intensively. Unfortunately, all systems presented so far have suffered from sluggish hydrogen reaction kinetics and poor reversibility. One reason for the latter finding is the tendency of Ca(BH 4 ) 2 to form kinetically stable side products upon dehydrogenation that hinder a complete rehydrogenation. In the case of pure Ca(BH 4 ) 2 , several desorption paths are proposed in the literature: Ca(BH 4 ) 2 -CaH 2 + 2B + 3H 2 [5][6][7] (1) Ca(BH 4 ) 2 -2/3CaH 2 + 1/3CaB 6 + 10/3H 2 5,6,[8][9][10] (2) Ca(BH 4 ) 2 -1/6CaB 12 H 12 + 5/6CaH 2 + 13/6H 2 [9][10][11] (3) Ca(BH 4 ) 2 -CaB 2 H 6 + H 2 (intermediate reaction). 6,10,11 (4) In fact, recent reports have indicated that the decomposition of Ca(BH 4 ) 2 may follow more than one single reaction path. 6,12 In these studies, the respective decomposition products and their ratio partially depend on the applied hydrogen back-pressure as well as the dehydrogenation temperature. 6,12 Also  15 The authors rule out the possibility that the hydride composite desorbs according to eqn (5). Instead, the formation of CaB 6 was confirmed to be the cause of the partial reversibility (eqn (6)), which was also reported earlier by Kim 23 In this work, the dehydrogenation reactions of the system Ca(BH 4 ) 2 -Mg 2 NiH 4 have been investigated for the first time. Furthermore, the potential for complete rehydrogenation of the desorbed products is demonstrated. For all experiments shown in this work, an initial mixture of Ca(BH 4 ) 2 and Mg 2 NiH 4 in a molar ratio of 1 : 2.5 was prepared by 5 hours of ball milling. † To investigate the thermal dehydrogenation path, an in situ synchrotron radiation powder X-ray diffraction (SR-PXD) experiment was performed applying a temperature ramp of 5 K min À1 heating from ambient temperature to 450 1C at 1 bar H 2 (Fig. 1). The room temperature diffraction pattern exhibits the diffraction peaks of the low-temperature monoclinic phase (C12/c1) of Mg 2 NiH 4 and its high-temperature cubic phase (Fm% 3m) (weight ratio approx. 2.3 : 1). Diffraction peaks of Ca(BH 4 ) 2 can be attributed to the a-(F2dd) and g-phase (Pbca), with a weight ratio of roughly 5 : 1. The first event with rising temperature is the polymorphous change from a-/g-Ca(BH 4 ) 2 to b-Ca(BH 4 ) 2 (P% 4), which starts at about 174 1C. The second modification in the diffraction pattern is related to another polymorphous change: the monoclinic Mg 2 NiH 4 converts to the cubic phase at approx. 244 1C. At about 318 1C, Mg 2 NiH 4 starts to decompose to Mg 2 Ni and hydrogen. At slightly higher temperatures, the intensity of Ca(BH 4 ) 2 diffraction peaks starts to decrease, while the peaks of four other phases occur and intensify, namely MgNi 2.5 B 2 , Mg, CaH 2 and an unknown phase. With further temperature increase, the peaks of Mg 2 Ni first grow until the maximum intensity is reached at roughly 365 1C, followed by a quick decline of peak intensity. The peaks of MgNi 2.5 B 2 , Mg and CaH 2 grow during the remaining heating period and the following isothermal region at 450 1C. In contrast, the peaks of the unknown phase only grow until the highest intensity is attained at about 410 1C, then the peak intensities reduce much slower than those of Mg 2 Ni. Moreover, it must be emphasised that the diffraction peaks of Ca(BH 4 ) 2 vanish already at about 365 1C, that is when MgNi 2.5 B 2 and the unknown phase, which potentially contains boron, are still growing. This finding strongly suggests the possibility of Mg 2 Ni reacting with the decomposition products of Ca(BH 4 ) 2 , i.e. CaB 6 , CaB x H y and/or amorphous boron. Furthermore, the sequence of peak intensity changes for the different phases suggests that the unknown phase is a yet unreported Mg-Ni-B-phase. Since the reaction kinetics for the dehydrogenation are too sluggish to monitor the complete process by means of in situ SR-PXD, additional material was desorbed at a pressure of 1 bar H 2 in a Sieverts apparatus and then characterised using ex situ laboratory Powder X-ray Diffraction (PXD). Fig. 2 shows the diffraction patterns ‡ of seven different samples. All samples were heated to their respective maximum temperatures with 3 K min À1 ; subsequently the material was kept under isothermal conditions for different times. Diffractograms with the same colour represent samples with equal total desorption time, i.e. the isothermal time was altered in such a manner that the overall time of both, heating period and isotherm, was equal for desorptions of the same set. Diffractograms of samples with a total desorption time of 150 minutes are shown in black. Red and blue diffractograms indicate total desorption times of 6 and 42 hours, respectively. The sample heated up to 350 1C for 150 min shows the peaks of Mg 2 Ni and the unknown phase besides those of MgNi 2.5 B 2 , Mg and CaH 2 . However, compared to the in situ SR-PXD pattern at 350 1C the ratios of the individual phases are different: Mg 2 Ni almost vanished whereas the peaks of the unknown phase are much more pronounced than at any temperature in the in situ experiment. Hence, it can be deduced that the reaction path leading to this unknown phase is more dominant at lower temperatures. Nevertheless, looking at the PXD patterns of the material desorbed at 350 1C with longer isotherms, the peaks of the unknown phase first decline (6 h) Fig. 1 In situ SR-PXD analysis of Ca(BH 4 ) 2 -Mg 2 NiH 4 , the sample was heated at 5 K min À1 from RT to 450 1C at 1 bar H 2 .
The quantity of hydrogen released in this desorption reaction amounts to 4.6 mass%. Fig. 3 presents results of Differential Scanning Calorimetry (DSC) combined with volumetric analysis for Ca(BH 4 ) 2 -Mg 2 NiH 4 in comparison to pure Mg 2 NiH 4 and Ca(BH 4 ) 2 (all materials ball milled using the same parameters), obtained under 1 bar H 2 and a heating rate of 5 K min À1 . The composite shows one broad endothermic desorption peak (onset at 296 1C, maximum at 331 1C) that embraces the several sub-events. This is in agreement with the in situ SR-PXD experiment which revealed that the desorption of Mg 2 NiH 4 triggers the described follow-up reactions and that the partial reactions overlap in time. Compared to pure Ca(BH 4 ) 2 (onset 353 1C, maximum 365 1C), desorption temperatures are shifted to lower values by about 55 K in the case of the composite. Thus, Mg 2 NiH 4 or, more specifically, Mg 2 Ni clearly destabilises Ca(BH 4 ) 2 . However, pure Mg 2 NiH 4 § (onset 272 1C, first maximum 282 1C) starts to dehydrogenate at temperatures lower than those for Ca(BH 4 ) 2 -Mg 2 NiH 4 . This discrepancy must be attributed to the kinetic limitations of Mg 2 NiH 4 decomposition in the RHC with respect to pure Mg 2 NiH 4 . A possible explanation could be a reduced energy transfer to Mg 2 NiH 4 upon ball milling due to the cushioning effect of Ca(BH 4 ) 2 . As a consequence, the Mg 2 NiH 4 within the composite would have higher activation barriers for desorption as compared to the pure Mg 2 NiH 4 . In comparison, the temperatures in the DSC experiments are roughly 20 K lower than in the in situ SR-PXD measurements. This deviation may be due to the particular experimental setup of SR-PXD experiments, where an inevitable gap between the specimen and the thermocouple may cause a temperature offset. At 4.55 mass%, the amount of released hydrogen agrees well with the theoretical capacity of the RHC of 4.6 mass%.
In order to investigate the possible presence of boron-containing phases that are not detectable using X-ray diffraction -CaB 12 H 12 and B are typically nano-crystalline or amorphous 7,11,13 -11 B Magic Angle Spinning Nuclear Magnetic Resonance (MAS-NMR) spectroscopy was employed. The spectra of the as-milled material and of the material desorbed for 24 h at 450 1C and dynamic H 2 pressure of more than 1 bar can be found in Fig. 4 (black and red graphs). As can be seen, the initial material comprises only one boroncontaining phase, that is Ca(BH 4 ) 2 . More precisely, the peak at À32.2 ppm is a convolution of aand g-Ca(BH 4 ) 2 . The desorbed sample features one peak at 141.6 ppm which is assigned to MgNi 2.5 B 2 . This result is quite remarkable since it shows that none of the typical decomposition products of Ca(BH 4 ) 2 could be detected in the desorbed state. Instead, all boron seems to be bonded in the form of MgNi 2.5 B 2 . Since the formation of the kinetically stable phases CaB 12 H 12 and/or B is the main reason for the degradation of Ca(BH 4 ) 2 -based systems, the absence of these phases may allow for full reversibility of a Ca(BH 4 ) 2 -based H 2 storage material. Consequently, the reabsorption potential of the Ca(BH 4 ) 2 -Mg 2 NiH 4 composite was evaluated. The decomposition products were kept in an autoclave under hydrogen with an initial pressure of 395 bar ¶ and a temperature of 400 1C for 14 h. These high pressure hydrogenation conditions were chosen arbitrarily in an attempt to maximise the potential conversion of the decomposition products into the starting material. The NMR spectrum of the reabsorbed sample is shown in Fig. 4 (blue graph). There are two major resonance peaks: the first at 141.5 ppm can be assigned to MgNi 2.5 B 2 , the second, at À32.6 ppm, clearly belongs to Ca(BH 4 ) 2 . Since the desorbed sample only features MgNi 2.5 B 2 as the sole boron-containing phase, the Ca(BH 4 ) 2 formed must have received the boron from this ternary boride upon  absorption. Consequently, aside from its ability to act as a borondonor for the formation of LiBH 4 and NaBH 4 , MgNi 2.5 B 2 also exchanges boron reversibly with Ca(BH 4 ) 2 . The 11 B spectrum of the reabsorbed material also shows another small resonance peak at À14.7 ppm which cannot be assigned clearly. However, it can be assumed that it belongs to a (B x H y ) zÀ -phase that could be an intermediate product in Ca(BH 4 ) 2 formation. The NMR spectrum indicates a converted fraction to Ca(BH 4 ) 2 of roughly one-third. However, the reabsorption did not complete under the applied conditions since there is still a high fraction of MgNi 2.5 B 2 present in the absorbed sample. At the moment the reason for this limited reconversion is not clear yet. Since the PXD analysis of the reabsorbed sample (not presented here) only shows very small and broad Ca(BH 4 ) 2 reflexes, the Ca(BH 4 ) 2 crystallites seem to be rather small. Hence, crystallite growth appears to be restricted. Further studies will have to clarify which microstructure develops and whether the reconversion into the initial hydride is confined in some way. For instance, the reconversion may be restricted to interface areas between the desorbed reactants, and the newly formed Ca(BH 4 ) 2 may act as a diffusion barrier, leading to reduced kinetic rates and thus an incomplete rehydrogenation.
In summary, the reactive hydride composite Ca(BH 4 ) 2 -Mg 2 NiH 4 features a complete boron transfer from Ca(BH 4 ) 2 to MgNi 2.5 B 2 upon dehydrogenation. No other boron-containing phases were present among the decomposition products; this includes especially the inactive phases amorphous boron and CaB 12 H 12 . Hence, a degradation of the hydrogen storage capacity caused by these boron sinks, as observed in pure Ca(BH 4 ) 2 6,7,11 or Ca(BH 4 ) 2 -MgH 2 , [13][14][15]24 can be ruled out for this composite. Mg 2 NiH 4 destabilises Ca(BH 4 ) 2 at a hydrogen pressure of 1 bar resulting in the lowering of desorption temperatures by about 55 K as compared to pure Ca(BH 4 ) 2 . These dehydrogenation conditions might still be above the desired temperature range for technical applications. However, the absence of boron sinks must be considered as a substantial improvement of Ca(BH 4 ) 2 based RHCs. Further optimisation of the material preparation should reduce the desorption temperatures at least down to the level of Mg 2 NiH 4 . Although the system did not reabsorb entirely, it was proven that the formation of Ca(BH 4 ) 2 starting from MgNi 2.5 B 2 as boron-donor is thermodynamically and kinetically possible. Reabsorption conditions have not been optimised yet. Hence, it is likely that rehydrogenation could also be achieved at lower temperatures and pressure. Moreover, no additives were used within this study. Therefore, further improvements of the hydrogen sorption kinetics may be feasible. This work was supported by the Danish Council for Strategic Research via HyFillFast. Access to beam time at the synchrotron PETRA III at DESY, Hamburg, Germany is gratefully acknowledged.
Notes and references † The milling was performed in a SPEX 8000M Mill using stainless steel milling vials and stainless steel balls with a diameter of 10 mm. A ballto-powder ratio of 10 : 1 was employed. ‡ The background of each diffraction pattern has a broad amorphous peak between 1 and 2 Å À1 which is caused by the PMMA sample holder used to protect the material from air. § Typically, Mg 2 NiH 4 features only one single desorption peak. The presence of the double peak is a kinetic effect. A possible explanation could be a kinetic separation between nucleation and growth of Mg 2 Ni 25 as a consequence of the rather low activation energy for desorption and a rather homogeneous particle size distribution after 5 h of ball milling. ¶ Due to absorption and a small leakage the pressure dropped to approx. 180 bar at the end of the synthesis.