Considering its high energy density, abundance, small weight, and environmental friendliness, hydrogen could provide a considerably cleaner and more sustainable society in the future (Schlapbach and Züttel, 2001). However, various challenges should be overcome to achieve the storage of hydrogen in a safe, efficient, and economic manner (Eberle et al., 2009). Metal complex hydrides composed of metal cations and complex anions (i.e., alanates, borohydrides, and amides) can store considerably more hydrogen than traditional interstitial hydrides, and thus have attracted increasing interest in recent years (Orimo et al., 2007; Jain et al., 2010). In particular, sodium alanate (NaAlH₄) is considered a promising solid medium for hydrogen storage because of its suitable thermodynamics, relatively low desorption temperature, and good reversibility (Li et al., 2013a; Liu et al., 2018). Theoretically, NaAlH₄ contains 7.5 wt% H₂ which is obtained in a three-step process.

However, only 5.6 wt% H₂ from the first two steps in the above equation can be utilized for practical applications as the decomposition of NaH occurs at temperatures over 400°C (Li et al., 2013a), too high for hydrogen storage.

Several strategies have been developed to improve the hydrogen storage properties of complex hydrides, such as catalyst doping (Frankcombe, 2012; Liu et al., 2018), cations substituting (Jain et al., 2010; Fang et al., 2011; Mo and Jiang, 2018), fabrication of reactive composites (Vajo et al., 2005; Ding et al., 2015; Mustafa et al., 2018), and nanostructuring (Ding and Shaw, 2019; Ding et al., 2019a, 2020). Recently, Shaw's group have developed a ball milling process with aerosol spraying to fabricate a nanocomposite of LiBH₄ and MgH₂ and successfully achieved the dual-tuning effects of the thermodynamics and kinetics of LiBH₄ (Ding et al., 2019b). Regarding NaAlH₄, numerous studies have shown that the addition of appropriate catalysts is crucial for a reduction in its hydrogen storage operation temperatures (Liu et al., 2018). In 1997, for the first time, Bogdanović and Schwickardi have reported a reduction (higher than 80°C) in desorption temperature of NaAlH₄ by doping 2 mol% β-TiCl₃ (Bogdanović and Schwickardi, 1997). Since then, various Ti-based additives have been introduced into NaAlH₄, particularly Ti halides and oxides, the most investigated catalysts (Frankcombe, 2012). Using TiF₃, Wang et al. have reported a release of H₂ above 2.5 wt% at 120°C (Wang et al., 2005; Kang et al., 2007). Lee et al. observed superior catalytic activity for nano-TiO₂ over TiCl₃ because a nano-TiO₂-containing NaAlH₄ has released ~3 wt% H₂ at 150°C within 10 min while only 2.5 wt% H₂ has been released from a TiCl₃-doped sample (Lee et al., 2008). Moreover, complete hydrogen release from NaAlH₄ was realized with a nano-TiO₂/C composite catalyst at 140°C within 30 min, with up to an H₂ capacity of 4.5 wt% (Liu et al., 2016).

However, the introduction of high-electronegativity anions, such as O, F, Cl, and Br, has reduced the effective hydrogen capacity because these anions tend to combine with Na and/or Al and consume the active components of hydrogen storage. Thus, methods to simultaneously achieve low dehydrogenation temperatures and high practical hydrogen capacities are required. In this regard, Ti-based compounds composed of low-electronegativity anions, such as TiN, TiC, and TiB₂, have come in sight for their catalytic activities. A reversible storage capacity of 4.9 wt% H₂ has been demonstrated within 16 cycles by doping TiN into NaAlH₄ (Bogdanovic et al., 2003). A NaAlH₄-2%TiN mixture has exhibited a capacity above 5 wt% H₂ at 250°C (Li et al., 2013b). A rod-shaped nano-TiN@C-N composite has reduced the hydrogen desorption temperature to 140°C with an H₂ capacity of 4.9 wt% (Zhang et al., 2018). Relatively high hydrogen capacity was also obtained for TiB₂-doped NaAlH₄ (Li et al., 2012a,b; Liu et al., 2014). However, only a limited reduction in dehydrogenation temperature has been attained so far. Further increase in the catalytic effectiveness of TiB₂ is still desired.

In this study, we synthesized an amorphous-carbon-supported nanoparticulate TiB₂ (nano-TiB₂@C) by calcining a mixture of Cp₂TiCl₂ and MgB₂. The fabricated nano-TiB₂ has a size of 2–4 nm and exhibited a remarkable catalytic activity for the hydrogen storage reaction of NaAlH₄. A reduction in onset dehydrogenation temperature higher than 100°C was achieved using a 7 wt% nano-TiB₂@C, which provided a practically available hydrogen capacity of 5.04 wt%. Furthermore, almost no capacity loss was observed within 20 cycles, which is superior to the performances of reported TiB₂-modified samples. The chemical states of nano-TiB₂@C and corresponding catalytic mechanisms were analyzed.

The structure and composition of the fabricated nano-TiB₂@C were analyzed by XRD, EDS and XPS. The results are shown in Figure 1. The calcinated sample exhibited the diffraction peaks of MgCl₂ (Figure 1A). After the washing with THF, only a broad bump at 44.4° (2θ) was observed in the XRD profile. The low and broad peaks indicate low crystallization and/or small particle/grain sizes. Ti, B, and C were detected by EDS (Figure 1B). An element analysis shows that their weight ratio was ~29:14:57, corresponding to a molar ratio of Ti and B of 1:2. A Raman spectrum analysis indicates that the elemental C was in its amorphous form (Figure 1C). Three characteristic peaks of TiB₂ were observed at 260, 410, and 598 cm⁻¹ (Bača and Stelzer, 2008). The high-resolution XPS spectra (Figures 1D,E) show characteristic peaks of the Ti–B bonding at binding energies of 460.3/454.7 eV for Ti 2p and 187.6 eV for B 1s (Ding J. C. et al., 2019). Combining the XRD, EDS, and XPS results, we believe that TiB₂ and amorphous carbon were formed by calcining the mixture of Cp₂TiCl₂ and MgB₂.

TEM, HRTEM, EDS mapping, and selected-area electron diffraction (SAED) analyses were carried out. The TEM image (Figure 2A) shows a large number of black nanoparticles distributed in a gray matrix. The EDS mapping (Figure 2B) reveals that the small nanoparticles consisted of Ti and B, while the gray matrix was mainly C. The SAED pattern indicates (101), (110), and (201) planes assigned to TiB₂ (Figure 2C). The HRTEM images (Figures 2D,E) indicated an interplanar spacing of 0.204 nm, corresponding to the interplanar distance of the (101) planes of TiB₂. The particle sizes of the TiB₂ were ~2–4 nm, part of which displayed clearly hexagon structures (Figure 2D). These results reveal nanoparticulate TiB₂ well-dispersed in the amorphous carbon matrix.

The resultant nano-TiB₂@C was mixed with NaAlH₄ by ball milling to test its catalytic effectiveness. After the ball milling, all samples exhibited very similar XRD patterns, as shown in Figure 3A. With the increase in amount of nano-TiB₂@C, the diffraction intensities of the NaAlH₄ phase slightly decreased. The SEM images reveal irregular solid particles with sizes of 200 nm−2 μm for 7 wt% nano-TiB₂@C-containing sample (Figures 4A,B). The EDS mapping results indicate relatively homogenous distribution of Ti, B, and C on NaAlH₄ particles (Figures 4C–G). Although no Ti-, B-, and C-containing phases were identified by XRD, possibly owing to their amorphous forms (Figure 3A), the XPS results show Ti 2p and B 1s spectra assigned to TiB₂ with binding energies of 454.7/460.3 and 187.6 eV (Figures 4H,I), respectively, indicating the presence of TiB₂ upon the mixing with NaAlH₄. We therefore believe that nano-TiB₂@C was uniformly distributed into the NaAlH₄ matrix.

The nano-TiB₂@C-containing NaAlH₄ samples were subjected to TPD and volumetric measurements for qualitative and quantitative characterization of their hydrogen storage performances. Three dehydrogenation peaks were observed in the TPD curves of all samples (Figure 3B), corresponding to the three-step decomposition process of NaAlH₄ with the increase in temperature (Equation 1). Furthermore, the nano-TiB₂@C-containing samples exhibited considerable low-temperature shifts. Upon the addition of 1 wt% nano-TiB₂@C, the dehydrogenation peak associated to the first dehydrogenation step (Equation 1) shifted from 255 to 169°C, a reduction of 86°C. The increase in nano-TiB₂@C content up to 7 wt% further reduced the start and end temperatures of the first-step decomposition to 75 and 118°C, 100 and 137°C lower than those of the pristine NaAlH₄, respectively. In addition to the slightly reduced peak intensities, the shape of TPD curve was almost unchanged with the further increase in content of nano-TiB₂@C. This result indicates that 7 wt% nano-TiB₂@C was optimal for the improvements in hydrogen storage performance of NaAlH₄.

Figure 3C shows volumetric release curves of the nano-TiB₂@C-modified samples. As expected, the 7 wt%-nano-TiB₂@C-containing sample exhibited the optimal dehydrogenation properties in terms of dehydrogenation temperature and hydrogen capacity in this study. Approximately 5.04 wt% of hydrogen was released in the temperature range of 75–175°C, which is remarkably superior to the performance of previously reported TiB₂-doped NaAlH₄ and other transition-metal-catalyzed NaAlH₄ structures (Table 1) (Wang et al., 2005; Lee et al., 2008; Fan et al., 2009; Naik et al., 2009; Li et al., 2012a,b, 2013b; Liu et al., 2014). In the isothermal test, the same amount of hydrogen (5.04 wt%) was released within 50 min at 140°C (Figure 3D). In contrast, <1 wt% of hydrogen was released from the pristine NaAlH₄ under the same conditions. At 120°C, the 7 wt%-nano-TiB₂@C-containing sample could desorb 4 wt% H₂ within 30 min, showing a much faster dehydrogenation kinetics than those of well-studied TiCl₃-modified NaAlH₄ (Figure 3D) (Bogdanović and Schwickardi, 1997; Naik et al., 2009). Even at 80°C, ~3.5 wt% H₂ could be desorbed, though a period of 400 min was required. These dehydrogenation kinetics outperform those of other TiB₂-doped NaAlH₄, which released only 2.79 wt% under the same conditions (Li et al., 2012b). Using the Kissinger's method (Kissinger, 1957), the apparent activation energies (E_a) were determined to be ~82.8 and 71.8 kJ/mol for the first and second dehydrogenation of the 7 wt%-nano-TiB₂@C-containing sample, respectively (Figure 3E), which are ~ 40% lower than those of pristine NaAlH₄ (Zhang et al., 2016) and is responsible for the remarkably reduced dehydrogenation temperatures.

The dehydrogenated samples were re-hydrogenated under an H₂ pressure of 100 bar. As shown in Figure 5A, the sample containing 7 wt%-nano-TiB₂@C started to absorb hydrogen at a temperature of 30°C, 70°C lower than that of the sample without doping. The hydrogenation was completed at 100°C in the non-isothermal test. The XRD analysis indicates that NaAlH₄ was formed after the full hydrogenation (Figure 5B). Isothermal hydrogenation under an H₂ pressure of 100 bar reveals that ~5.02 wt% of hydrogen recharged into the dehydrogenated 7 wt%-nano-TiB₂@C-containing sample within 35 min at 80°C, which provided full hydrogenation (Figure 5C). At 120°C, only 20 min were required to complete full hydrogenation, which shows the considerably faster kinetics. The follow-up dehydrogenation repeatedly resulted an H₂ capacity of 5.02 wt% (Figure 5D), which shows the good reversibility.

Figure 6A shows the cyclic stability of NaAlH₄-7 wt%-nano-TiB₂@C. Here, the dehydrogenation was conducted at 140°C in vacuum, while the hydrogenation took place at 100°C under an H₂ pressure of 100 bar. After 20 cycles, the available hydrogen capacity still remained at 5.02 wt%, which shows the stable recyclability. This cycling stability is superior to that of the well-studied TiCl₃-catalyzed NaAlH₄ (Figure 6B). In addition, a small but continuous reduction in onset dehydrogenation temperature was observed in the first four cycles (Figure 6C), which reflected the activation. This might correlate to some changes in catalytic active species during the initial de-/hydrogenation cycles. Further comparison reveals that hydrogen release from the nano-TiB₂@C-containing NaAlH₄ occurred at lower temperatures than those of samples with either TiB₂ or active carbon (AC) (Figure 6D), which shows the synergistic effect of TiB₂ and C, similarly to the previous observation for NaAlH₄ co-catalyzed by NbF₅ and single-walled carbon nanotubes (Mao et al., 2011, 2012).

Figure 7A shows XRD patterns of the dehydrogenated samples containing 7 wt% nano-TiB₂@C as a function of the temperature. The results indicate that with the increase in temperature, NaAlH₄ initially decomposed to Na₃AlH₆ and Al (110–140°C), which then led to the formation of NaH and Al (155–175°C) with the hydrogen release. We therefore believe that the presence of nano-TiB₂@C did not alter the dehydrogenation course of NaAlH₄. Notably, no Ti-containing species was identified by the XRD profiles. Subsequently, high-resolution Ti 2p and B 1s XPS spectra were acquired to understand the chemical states of TiB₂ (Figures 7B,C). Upon cycling, a 2p_3/2-2p_1/2 spin–orbit doublet at 452.3/458.1 eV emerged, and then became dominant in the Ti 2p XPS spectra (Figure 7B), which can be assigned to Ti–Al bonding (Mencer et al., 1991). Further XRD analysis confirms the presence of an AlTi alloy (Figures 8A,B). The AlTi alloy surface is favorable for the dissociation and recombination of H–H and Al–H bondings (Frankcombe, 2012; Liu et al., 2018). In contrast, the XPS peaks of Ti at 454.7/460.3 eV were largely reduced. On the other hand, the characteristic XPS peak of B⁰ at 186.4 eV was also detected, which gradually increased in the initial four cycles. Thus, upon the de-/hydrogenation cycling, TiB₂ was gradually converted to Ti–Al and B, possibly reacting with NaAlH₄. This is crucial for the continuous reduction in dehydrogenation temperature of the nano-TiB₂@C-modified sample in the initial four cycles. The newly formed Ti–Al and B remained stable in the following cycles, which led to a good cycling stability, as shown in Figure 6A.

The nano-TiB₂@C-containing sample subjected to 20 cycles was used for a TEM observation. As shown in Figures 8C–F, a large number of Ti catalyst nanoparticles (sizes < 5 nm) were dispersed on the surface of the NaAlH₄ particle. Therefore, we believe that ultrasmall particles of TiB₂ as precursors facilitated the formation of ultrafine dispersive Ti–Al active species. The dispersive distribution of Ti catalysts provided the high catalytic activity for hydrogen storage in NaAlH₄, particularly for long-term cycling (Figure 4D).

In this work, nano-TiB₂@C below 5 nm was synthesized. Remarkable reduction in dehydrogenation and hydrognaiton temperatures was observed when adding 7 wt% nano-TiB₂@C to NaAlH₄. The hydorgen desorption started at a temperature of 75°C, which is lowered by 100°C compared to the pristine NaAlH₄. A practical hydrogen capacity of 5.04 wt% was determined, which was released within 50 min at 140°C. The rehydrogenation occurred at 30°C under a hydrogen pressure of 100 ba, and was completed at 100°C. Notably, no capacity loss was observed in the 20 cycles. During the initial de-/hydrogenation cycling, TiB₂ presumably reacted with NaAlH₄ and was converted to AlTi alloy and zero-valence B, which were well-dispersed on the surface of the NaAlH₄ particles, and consequently contributed to the high stable catalytic activity. These findings could facilitate the practical use of NaAlH₄ as a high-capacity reversible hydrogen storage medium.

All datasets generated for this study are included in the article/supplementary material.

XiZ and YL conceived the study and designed the experiments. XiZ, XuZ, and ZR carried out the material syntheses, characterization, and measurements. XiZ, YL, JH, MG, and HP analyzed the data. XiZ, JH, and YL wrote the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.