Influence of K2NbF7 Catalyst on the Desorption Behavior of LiAlH4

In this study, the modification of the desorption behavior of LiAlH4 by the addition of K2NbF7 was explored for the first time. The addition of K2NbF7 causes a notable improvement in the desorption behavior of LiAlH4. Upon the addition of 10 wt.% of K2NbF7, the desorption temperature of LiAlH4 was significantly lowered. The desorption temperature of the LiAlH4 + 10 wt.% K2NbF7 sample was lowered to 90°C (first-stage reaction) and 149°C (second-stage reaction). Enhancement of the desorption kinetics performance with the LiAlH4 + 10 wt.% K2NbF7 sample was substantiated, with the composite sample being able to desorb hydrogen 30 times faster than did pure LiAlH4. Furthermore, with the presence of 10 wt.% K2NbF7, the calculated activation energy values for the first two desorption stages were significantly reduced to 80 and 86 kJ/mol; 24 and 26 kJ/mol lower than the as-milled LiAlH4. After analysis of the X-ray diffraction result, it is believed that the in situ formation of NbF4, LiF, and K or K-containing phases that appeared during the heating process promoted the amelioration of the desorption behavior of LiAlH4 with the addition of K2NbF7.


INTRODUCTION
The excessive consumption of fossil fuels and the emission of carbon dioxide are the roots of environmental pollution. As a resolution to this global issue, the utilization of clean, and sustainable energy resources such as hydrogen, wind, and solar has become an inescapable need. Recently, hydrogen has received a large amount of attention as a future energy carrier. Hydrogen promises to be a clean and renewable energy carrier. Moreover, the production of hydrogen can be achieved from various resources, both renewable (e.g., solar, wind, and hydro) and non-renewable (e.g., natural gas and coal; Winter, 2009;Parra et al., 2019). Furthermore, energy production via hydrogen-oxygen reaction will only produce water as a by-product (Crabtree et al., 2004). In pursuit of the success of hydrogen as a future energy carrier, the need for an efficient and reliable storage method has become the top priority. In general, there are three forms of hydrogen storage which are: (i) compressed hydrogen gas, which requires high pressure, (ii) liquefaction, and (iii) solid-state hydrogen storage via hydrides (Dalebrook et al., 2013;Zhang et al., 2016;Barthelemy et al., 2017). Solid-state hydrogen storage has been perceived to be an efficient and favorable method because of its safety, storage requirements, and storage capacity.
Among these catalysts, previous studies have revealed that metal halides provide essential catalytic effects on the performance of LiAlH 4 . Cao et al. (2018) reported that the addition of ScCl 3 had a superior effect on the performance of lithium alanates. The desorption process of the LiAlH 4 -10 mol% ScCl 3 sample began at a lower temperature (∼120 • C), while the undoped LiAlH 4 released hydrogen from around 150 • C. Besides, the time needed to complete the dehydrogenation process was shortened with the addition of 1-10 mol% ScCl 3 . Meanwhile, Sun et al. (2008) found that NiCl 2 significantly boosted the desorption behavior of LiAlH 4 . A composite sample of LiAlH 4 -NiCl 2 demonstrated three times the desorption rate of pure LiAlH 4 , which was not able to desorb any hydrogen at 100 • C. It was believed that the LiAlH 4 -NiCl 2 sample presented this notable improvement due to the formation of Ni, which plays a vital role in accelerating the LiAlH 4 -NiCl 2 system. Another investigation of the catalytic effect of metal halides was carried out by Liu et al. (2012). They proved that a LiAlH 4 -TiCl 3 sample could release hydrogen at a lower temperature (80 • C) than the pure LiAlH 4 . Furthermore, the dehydrogenated sample had good cyclability, with the composite sample able to retain a high capacity for hydrogen (6.4 wt.%) even after completing the 3rd cycle. Moreover, Ismail et al. (2010) observed that the composite sample of LiAlH 4 -1 mol NbF 5 showed a 5-6 times faster dehydrogenation rate than the milled LiAlH 4. Additionally, the LiAlH 4 -NbF 5 composite sample had lower activation energy; 67 kJ/mol (first-stage reaction) and 77 kJ/mol (second-stage reaction), respectively. However, the improvement of LiAlH 4 through the addition of a catalyst is still lacking, and further enhancements still need to be carried out. Moreover, different catalysts will enable different effects and performances. Therefore, it is interesting to enhance the desorption performance of LiAlH 4 by the addition of other metal halides.
Metal halides, especially fluorides, are known to be highly effective catalysts for solid-state materials (Sulaiman et al., 2016;Yap et al., 2017;Youn et al., 2017). A number of researchers have reported that niobium fluoride exhibits a notable effect on the hydrogenation behavior of metal hydrides and the complex hydrides. A study conducted by Luo et al. (2008) revealed that the addition of 2 mol of NbF 5 led to faster desorption kinetics for the MgH 2 -NbF 5 sample as compared to pure MgH 2 . At 573 K, the MgH 2 -NbF 5 sample could desorb 4.7 wt.% of hydrogen, while the pristine MgH 2 desorbed almost no hydrogen. Other than that, Kou et al. (2014) added NbF 5 to LiBH 4 and demonstrated notable improvement on the desorption performance of LiBH 4 . In comparison to the milled LiBH 4 , which started to desorbed hydrogen at >400 • C, the composite sample of LiBH 4 -NbF 5 had a lower desorption temperature, 60 • C. Moreover, Wang and colleagues (Wang et al., 2020) showed that a composite of Mg(BH 4 ) 2 -doped NbF 5 possessed the best dehydrogenation performance, with the ability to release hydrogen at low temperature (120 • C), as compared to amorphous Mg(BH 4 ) 2 (126.9 • C), and pristine Mg(BH 4 ) 2 (282.7 • C) . Meanwhile, Cheng et al. (2018) demonstrated that upon the addition of NbF 5 , the composite of 4LiBH 4 -MgH 2 -Al exhibited excellent kinetics and reversibility performance. It took <4 h to achieve 90% of the total amount of hydrogen desorption. Other than that, Xiao et al. (2012) proved that the performance of LiBH 4 /MgH 2 was significantly improved with the addition of NbF 5 . Here, the addition of NbF 5 not only had reduced the onset decomposition temperature but also improved the dehydrogenation and absorption rates.
On the other hand, potassium (K) is another well-known additive for hydrogen storage systems. Wang et al. (2009) demonstrated that the addition of K significantly boosted the desorption process of Mg(NH 2 ) 2 /LiH by reducing the overall reaction temperature. Furthermore, Dong et al. (2014) revealed that superior results for the hydrogenation performance of the LiH-NH 3 system were obtained by the addition of various potassium compounds.
In respect to this matter, it is interesting to mix niobium fluoride with potassium as a ternary compound in the form of K 2 NbF 7 and to study its potential catalytic effect. To date, no studies have been conducted using doped K 2 NbF 7 as a catalyst for LiAlH 4. Moreover, previous studies reported that K 2 NbF 7 enables a remarkable improvement in the hydrogen storage performance of MgH 2 . Thus, it is of great interest to explore the influence of K 2 NbF 7 on the desorption performances of LiAlH 4 . It is anticipated that the addition of K 2 NbF 7 will have notable effects on the desorption and kinetic performances of LiAlH 4 .

EXPERIMENTAL DETAILS
Commercial powders of LiAlH 4 (purity 95%) and K 2 NbF 7 (purity 98%) were obtained from Sigma Aldrich and were used without any modification. To minimize exposure to oxygen and water moisture, the samples were prepared and handled in the Ar-filled Mbraun Unilab glove box. In this study, 10 wt.% of K 2 NbF 7 was mechanically milled together with LiAlH 4 to explore its effect on the desorption behavior of LiAlH 4 . The milling process was done in a planetary ball mill (NQM-0.4) for 1 h, starting with 0.5 h of milling, followed by 6 min of rest time, and then another 0.5 h of milling in a different rotation direction at a speed of 400 rpm. The samples were placed in a hardened stainless-steel jar with four stainless balls, each 1 cm in size. The ratio of the balls to the weight of the powder was 40:1. For comparison purposes, the as-received LiAlH 4 was treated under the same conditions.
The hydrogenation performances of LiAlH 4 + 10 wt.% K 2 NbF 7 were studied with temperature-programmed desorption (TPD) using Sievert-type pressure-composition-temperature (PCT) equipment (Advanced Materials Corporation). In order to determine the initial decomposition temperature, the sample was heated from room temperature to 250 • C (heating rate: 5 • C/min). Other than that, the desorption kinetics performances were evaluated at 90 • C under 1.0 atm of pressure. The apparent activation energy, E A , was determined using differential scanning calorimetry (DSC, Mettler Toledo, DSC/TGA 1), loading 5-7 mg of the samples into a crucible and heating from 25 to 300 • C at heating ramps of 15, 20, 25, and 30 • C/min under an argon flow (50 ml/min). In terms of the morphology and phase structure characterizations, the samples were analyzed using scanning electron microscopy (SEM: JEOL JSM 6350LA), X-ray diffractometry (XRD, Rigaku Miniflex), and Fourier transform infrared (IR Shimadzu Tracer-100). Figure 1 demonstrates the TPD results of the LiAlH 4 and modified LiAlH 4 system. The results show that the as-received and as-milled LiAlH 4 have similar desorption processes that occur in two stages of desorption, as in Equations (1, 2), with 7.4 wt.% hydrogen capacity. Before the ball milling process, the first stage of desorption occurred at 147 • C, with 5 wt.% of hydrogen released. Meanwhile, the desorption process for the second stage was recorded to happen at around 175 • C, with a capacity of 2.4 wt.% of the hydrogen. After the milling process, the initial desorption temperature of the sample was similar to that of pure LiAlH 4 but with slight temperature reductions to 144 • C (first stage) and 174 • C (second stage). This phenomenon showed that the 1-h milling process had an insignificant effect on the desorption behavior of LiAlH 4 . In contrast, the addition of 10 wt.% of K 2 NbF 7 significantly decreased the decomposition temperature for both stages, to 90 and 149 • C. However, the amount of hydrogen released from the LiAlH 4 + 10 wt.% K 2 NbF 7 sample was decreased to 6.3 wt.%. This is expected due to the dead weight of K 2 NbF 7 , which does not hold any hydrogen.  Further study on the catalytic activity of K 2 NbF 7 was performed based on the desorption kinetics experiment. Figure 2 depicts a comparison of the hydrogen desorption at 90 • C for LiAlH 4 and LiAlH 4 modified by the addition of 10 wt.% K 2 NbF 7 . It is noticeable that within 120 min, the undoped LiAlH 4 was only able to desorb a small amount of hydrogen; 0.1 wt.% for the as-received LiAlH 4 and 0.4 wt.% for the as-milled LiAlH 4. Surprisingly, with the addition of 10 wt.% K 2 NbF 7 , the doped sample desorbed ∼3.2 wt.% H 2 within the same duration. This desorption rate was 30 times faster than that of the as-received LiAlH 4 . This enhancement may be correlated to the formation of surface defects and active materials through the reaction of the LiAlH 4 + 10 wt.% K 2 NbF 7 composite (Cai et al., 2016).

RESULTS AND DISCUSSION
In terms of thermal behavior, DSC experiments were conducted for the doped and un-doped LiAlH 4 samples. Figure 3 displays the DSC curves of the samples at a heating ramp of 15 • C/min. Both the doped and un-doped LiAlH 4 have two endothermic and exothermic peaks. The first exothermic peak corresponds to the reaction of LiAlH 4 with surface hydroxyl groups, while the first endothermic peak is ascribed as its melting process. The second exothermic peak is attributed to the decomposition of LiAlH 4 , as described in Equation (1), and the second endothermic peak correlates with the decomposition of Li 3 AlH 6 , as described by Equation (2). Both samples exhibit similar thermal behavior, but the peaks of the LiAlH 4 + 10 wt.% K 2 NbF 7 sample occur at a lower temperature as compared to as-milled LiAlH 4 .
Fundamentally, the enhancement of the initial temperature to release hydrogen and the faster desorption kinetics rates are correlated with the energy barrier of LiAlH 4 . In this study, the decomposition activation energy (E A ) is the least possible amount of energy needed by LiAlH 4 to begin the hydrogen desorption process. Figure 4 shows DSC traces for several heating ramps (15, 20, 25, and 25 • C/min). By referring to the plots, the activation energies for both decomposition stages of the as-milled LiAlH 4 and LiAlH 4 + 10 wt.% K 2 NbF 7 samples were determined using the Kissinger analysis, as in equation (4): where β, T p , R, and A are the heating rate, peak temperature in the DSC curve, gas constant, and linear constant, respectively. The apparent activation energy was determined from the slope of ln [β / T p 2 ] versus 1000/T p , as shown in Figure 5. The activation energy was calculated based on the second exothermic (decomposition of LiAlH 4 ) and second endothermic (decomposition of Li 3 AlH 6 ) reactions. For the as-milled LiAlH 4 , the activation energy values were 104 and 112 kJ/mol for the first two stages of reaction, respectively. After the addition of 10 wt.% K 2 NbF 7 , the activation energy values dropped to 80 kJ/mol (first stage) and 86 kJ/mol (second stage), 23% lower than those of the un-doped LiAlH 4 . These results are in good agreement with other studies that prove the addition of a catalyst is able to reduce the activation energy of LiAlH 4 . Table 1 lists the activation energy from previous studies for comparison purposes. The reduction in these activation energies verifies that K 2 NbF 7 plays a major role in enhancing the desorption kinetics performance of LiAlH 4.
The morphological structures of the doped and un-doped LiAlH 4 were examined using SEM equipment. Figure 6 shows SEM images of the un-doped and doped-LiAlH 4 samples. As shown in Figure 6, the pure LiAlH 4 exhibits larger particle sizes than the milled sample. The as-received LiAlH 4 ( Figure 6A) has larger (15-40 µm), non-uniform rod-shaped particles. Furthermore, the as-received LiAlH 4 shows a uniform size distribution and consists of "blocky" particles, consistent with the report by Varin and Zbroniec (2010). Meanwhile, after 1 hour of milling, the milled LiAlH 4 ( Figure 6B) displays a reduction in particle sizes but with some agglomeration and inconsistency in particle size. Then, with the addition of 10 wt.% of K 2 NbF 7 (Figure 6C), the morphological structure of the sample was notably enhanced. The doped sample has smaller   particle sizes and is less agglomerated. This observation is in line with numerous research results that have shown a reduction of particle sizes with the addition of a catalyst Ali et al., 2018;. In this study, K 2 NbF 7 functioned as a dispersing agent that impeded the sample from agglomerating. The particle size is important because smaller particles provide more area for surface defects and additional grain boundaries (Schulz et al., 1999;Sakintuna et al., 2007;Ranjbar et al., 2009). As a consequence, the desorption kinetics of LiAlH 4 will be improved. Figure 7 presents the XRD profiles of the as-received LiAlH 4 , as-milled LiAlH 4 , and LiAlH 4 -K 2 NbF 7 sample. The XRD characterization was performed to explore the reaction process and the mechanism that operated during the milling process. Figure 7A displays the XRD pattern of the as-received LiAlH 4 and shows that only the LiAlH 4 phase was detected, which confirms the purity of the LiAlH 4 . The XRD pattern of the milled LiAlH 4 Figure 7B shows similar peaks to the as-received LiAlH 4 . This result shows that LiAlH 4 has high stability during the milling process and agrees well with a previous study (Ismail et al., 2010). Meanwhile, with the addition of 10 wt.% of K 2 NbF 7 (Figure 7C), only LiAlH 4 and Al peaks are visible and no peak of K 2 NbF 7 was detected, suggesting that the amount of catalyst was too small to be picked up by the XRD. The appearance of Al peaks indicates that a part of the LiAlH 4 had decomposed to Li 3 AlH 6 and Al (reaction 1) during the milling process in the presence of 10 wt.% K 2 NbF 7 . Surprisingly, the XRD result for the 10 wt.% K 2 NbF 7 -doped LiAlH 4 sample does not show any peaks of Li 3 AlH 6 . Additional characterization was carried out for a doped sample with 30 wt.% K 2 NbF 7 (Figure 7D). A K 2 NbF 7 peak was against not detected by the XRD for this sample. This may be because the K 2 NbF 7 is in an amorphous state. Similar phenomena were reported by previous studies, where several catalysts like TiO 2 and TiF 3 were not detected by the XRD after the milling process (Ismail et al., 2011;Zang et al., 2015). However, for the LiAlH 4 + 30 wt.% K 2 NbF 7 sample, diffraction peaks corresponding to the decomposition product, Al and Li 3 AlH 6 , were detected. Meanwhile, unlike for the LiAlH 4 + 10 wt.% K 2 NbF 7 sample, for which only peaks of Al were detected while peaks of Li 3 AlH 6 could not be discovered by XRD. Figure 8 shows the IR spectra of the as-received LiAlH 4 , asmilled LiAlH 4 , and LiAlH 4 + 10 wt.% K 2 NbF 7 in the range of 800 to 2,000 cm −1 . The FTIR characterizations were conducted to identify the presence of Li 3 AlH 6 in the 10 wt.% K 2 NbF 7doped LiAlH 4 sample. For all samples, two distinct regions of Al-H modes were detected at around 800-900 cm −1 ([AlH 4 ] − stretching modes) and 1,600-1,800 cm −1 ([AlH 4 ] − bending modes), respectively. Furthermore, with the addition of 10 wt.% K 2 NbF 7 , a weak IR absorption peak at 1,398 cm −1 was detected, which indicates the presence of Li 3 AlH 6 . This result suggests that with the addition of 10 wt.% K 2 NbF 7 , LiAlH 4 was partially decomposed to Li 3 AlH 6 and Al (reaction 1) during the milling process, consistent with the XRD results ( Figure 7C).
To investigate the specific mechanism that is related to the enhanced desorption performance of LiAlH 4 , the dehydrogenated sample was examined using XRD. The XRD pattern for the dehydrogenated sample is depicted in Figure 9. After the dehydrogenation process at 250 • C, the main peaks observed are the LiAlH 4 dehydrogenation products, LiH and Al, which indicates complete dehydrogenation of LiAlH 4 . In addition, peaks for LiF and NbF 4 were detected after the dehydrogenation process. However, the peak of the K-containing  phase was not detected after the dehydrogenation process, potentially due to the low amount of catalyst.
Niobium fluoride is well established as a promising catalyst that plays a vital role in enhancing the hydrogenation performance of solid-state material (Luo et al., 2007;Malka et al.,  2011; Mao et al., 2013). It is reasonable to state that the NbF 4 that formed in situ after the desorption process contributes to a remarkable amelioration of the desorption behavior of LiAlH 4 . This result well-agreed with previous research that demonstrates the outstanding dehydrogenation performance of LiAlH 4 -NbF 5 FIGURE 9 | XRD pattern of dehydrogenated LiAlH 4 + 10 wt.% K 2 NbF 7 sample at 250 • C. (Ismail et al., 2010). On the other hand, the LiF formed was believed to significantly affect the hydrogenation behavior of the doped sample based on work carried out as by Gosalawit-Utke et al. (2010). Additionally, it is believed that the formation of LiF plays a similar role in the growth of LiH and Al, since LiF has a similar cubic structure (space group: Fm-3m; Y. Liu et al., 2010). Also, LiF crystallites act as nucleation sites and facilitate the growth of LiH and Al crystallites, which promotes to the change of the nucleation morphology. These two factors significantly contribute to the kinetics enhancement achieved in the doped sample. Additionally, it is believed that K or K-containing phases also play a vital role in enhancing the desorption behavior of LiAlH 4 . This was deduced based on successful previous work on the application of K as a catalyst for solid-state materials (Wang et al., 2009;Dong et al., 2014). Therefore, it can be concluded that the in situ formation of LiF, NbF 4 , and K or Kcontaining phases synergistically contributed to the amelioration of the dehydrogenation kinetics of LiAlH 4 .

CONCLUSION
K 2 NbF 7 demonstrated an excellent catalytic effect on the desorption behavior of LiAlH 4 . The initial temperatures at which LiAlH 4 + 10 wt.% K 2 NbF 7 released hydrogen, 90 and 149 • C for the first two stages, were lower than those of the as-milled LiAlH 4 (147 and 175 • C). In terms of desorption kinetics behavior, the LiAlH 4 + 10 wt.% K 2 NbF 7 released 3.2 wt.% of hydrogen within 120 min, which is 30 faster than the pure LiAlH 4 . The addition of K 2 NbF 7 significantly reduced the decomposition activation energy from 104 to 80 kJ/mol for the first stage and 112 to 86 kJ/mol for the second stage. The XRD spectra suggested that the in situ formation of LiF, NbF 4 , and K or K-containing phases acted as boosters and ameliorated the dehydrogenation behavior of LiAlH 4 . This work demonstrates that K 2 NbF 7 was has a superior catalytic effect and confers better desorption behavior to LiAlH 4 .

DATA AVAILABILITY STATEMENT
The datasets generated for this study are available on request to the corresponding author.