Using Additives to Control the Decomposition Temperature of Sodium Borohydride

Hydrogen (H2) shows great promise as zero-carbon emission fuel, but there are several challenges to overcome in regards to storage and transportation to make it a more universal energy solution. Gaseous hydrogen requires high pressures and large volume tanks while storage of liquid hydrogen requires cryogenic temperatures; neither option is ideal due to cost and the hazards involved. Storage in the solid state presents an attractive alternative, and can meet the U.S. Department of Energy (DOE) constraints to find materials containing > 7 % H2 (gravimetric weight) with a maximum H2 release under 125 °C. While there are many candidate hydrogen storage materials, the vast majority are metal hydrides. Of the hydrides, this review focuses solely on sodium borohydride (NaBH4), which is often not covered in other hydride reviews. However, as it contains 10.6% (by weight) H2 that can release at 133 ± 3 JKmol, this inexpensive material has received renewed attention. JEPT 2020; 2(2), doi:10.21926/jept.2009 Page 2/20 NaBH4 should decompose to H2(g), Na(s), and B(s), and could be recycled into its original form. Unfortunately, metal to ligand charge transfer in NaBH4 induces high thermodynamic stability, creating a high decomposition temperature of 530 °C. In an effort make H2 more accessible at lower temperatures, researchers have incorporated additives to destabilize the structure. This review highlights metal additives that have successfully reduced the decomposition temperature of NaBH4, with temperatures ranging from 522 °C (titanium (IV) fluoride) to 379 °C (niobium (V) fluoride). We describe synthetic methods employed, chemical pathways taken, and the challenges of boron derivative formation on H2 cycling. Though no trends can be found across all additives, it is our hope that compiling the data here will enable researchers to gain a better understanding of the additives’ influence and to determine how a new system might be designed to make NaBH4 a more viable H2 fuel source.


Introduction
There is an ever increasing need to develop energy efficient materials that are environmentally friendly. Hydrogen (H 2 ) burned with oxygen is incredibly clean, and is becoming a popular zero-carbon emission fuel. Unfortunately, the storage of pure H 2 can be problematic. Storing in its gaseous state is very common, but there are serious concerns about flammability and explosive hazards. Storing H 2 in the liquid state requires ultralow, cryogenic temperatures, making storage expensive while adding the hazards, challenges and expense of cryogen handling [1,2]. There is a strong need to create a regenerative hydrogen storage material that is safe and lightweight enough for easy transportation. Thus, researchers have been exploring H 2 storage methods in the solid state, trying to meet the constraints given by the U.S. Department of Energy (DOE) [3], such as the recommendation to find materials containing > 7 % H 2 (gravimetric weight) with a maximum H 2 release rate under 125 °C.
There have been various strategies employed to develop these solid state materials. Some have focused on H 2 adsorbents, gaining an understanding of the gravimetric and volumetric capacities and binding strengths in materials such as metal organic and covalent organic frameworks [4]. Interestingly, there is often disagreement in this field on the theoretical analysis, such as calculating the H 2 capacity. Other researchers have turned their attentions toward nanostructured hydride materials, whose tiny size and large surface area make it possible to tune its properties independently of their bulk counterparts [5]. Research is ongoing, and it remains to be determined whether the required thermodynamics can be achieved to make metal hydrides a viable H 2 storage material, especially considering that hydrogenation induces mechanical stresses due to volume expansion.
Four different strategies have been applied over the years to reduce the decomposition temperatures of metal hydrides: 1) The nanoconfinement approach, where the complex size is reduced by infiltration into a nanoporous scaffold, decreases the decomposition temperature as a Page 3/20 result of an increase of the reaction kinetics [6][7][8][9]. 2) Thermodynamic destabilisation uses additives to alter the decomposition pathway, therefore reducing the decomposition temperature. These additives may include a metal hydride, a complex metal hydride, a metal, or even a metal based compound such as a metal oxide [10,11]. 3) Using a combination of scaffolds can both nanoconfine and thermodynamically destabilise the infiltrated complex metal hydride [12][13][14][15]. 4) Finally, the use of a catalyst increases the reaction kinetics of the metal hydrides, resulting in a lower decomposition temperature [16,17].
This review focuses solely on a complex metal hydride system in the metal borohydride family. Sodium borohydride (NaBH 4 ) is a stable, inexpensive solid containing 10.6% (by weight) H 2 that has been a popular material of study for energy applications [18][19][20][21][22][23]. At the start of the century, scientists attempted to make use of a hydrolysis process to make NaBH 4 a feasible H 2 source. The clean concept was to simply react NaBH 4 with water via the following reaction [24]: where the sodium metaborate (NaBO 2 ) could eventually be recycled back into NaBH 4 by using a catalyst such as magnesium hydride [25,26]. Unfortunately, this process met with multiple failures due to solubility issues and low yields, so was abandoned by the DOE in 2007 [27,28].
In more recent years, researchers have turned to non-aqueous, solid state methods to generate H 2 from NaBH 4 . When heated, NaBH 4 releases 133 ± 3 J K −1 mol −1 of H 2 along with solids of sodium (Na) and boron (B), with a small amount of sodium borides and sodium hydride [2]. Furthermore, the bulk of the decomposed by-product is recyclable according to theoretical calculations, where the solids remaining after dehydrogenation can reform NaBH 4 via the addition of more H 2 gas: Na(s) + B(s) + 2H 2 (g)  NaBH 4 (s) Unfortunately, the metal to ligand charge transfer in NaBH 4 induces high thermodynamic stability [29], so NaBH 4 does not decompose until about 535 °C at 1 bar of H 2 [2,30]. This is significantly higher than the desired <125 °C proposed by the DOE. As a result, research efforts have turned toward developing methods to weaken the NaBH 4 structure by using additives to destabilize the bonds and decrease the temperature of decomposition, thereby making H 2 more accessible. For example, using zirconium chloride [31] or titanium fluoride [32] as additives reduces the onset temperature of dehydrogenation to about 300 °C, while cobalt, copper and nickel nanoparticles reduce this onset to as low as 200 °C [9,17,33,34]. Unfortunately, dehydrogenation starting at such low temperatures is not a guarantee that the sample will decompose and completely release large amounts of H 2 at low enough temperatures to become an acceptable fuel source. Regardless, many researchers' results show promise for the method of using additives to destabilize NaBH 4 , and there is a need for ongoing work to develop a better understanding of how additives incorporate into NaBH 4 in order to improve future materials design. Figure 1 compiles the known decomposition temperatures of NaBH 4 resulting from the use of various additives. Though there are no obvious trends for controlling decomposition through metal size, anion selection, oxidation state, etc., there is a wide range of accessible decomposition temperatures available using additives to lower the kinetics. Current methods have been able to reduce the decomposition temperature as low as 379 °C. Though this does not quite reach the targets set by DOE, NaBH 4 may still be a relevant H 2 storage material if a better additive can be found.
The aim of this review is to provide information about recent research involving various additives to improve the kinetics of NaBH 4 for hydrogen storage, focusing only on additives that have been found to successfully decrease the decomposition temperature of NaBH 4 . Note that publications reporting only the H 2 release temperature without mention of decomposition are not included in this review. Though we have made every effort to capture all relevant research, the findings here are not exhaustive. This work is organized in sections by metal additive, explaining how the various additives destabilize and reduce the decomposition temperature of NaBH 4 . All of the research presented here used ball milling (BM) as the driving force to incorporate additives with NaBH 4 . Interestingly, the experimental methods for both synthesis and decomposition analysis varied greatly between groups. The molar ratios of additives used, milling rate and total mill time, breaks taken throughout the milling process to cool the material, and in some cases post-milling heating to anneal differed , which may play a large role in why some additives cause NaBH 4 to decompose at a higher or lower temperature than expected. Furthermore, variances in the methods by which thermal decomposition was measured may have a significant influence on the values reported. Though it is not traditional to include a subsection on synthetic techniques in a review like this one, we felt it important in this instance. Though no one has yet tied together how these various methods influence the final decomposition temperature, it is our hope that having all of this data in one place will help researchers make future decisions regarding their methodology. Our ultimate goal is that this review will help researchers gain a better understanding of how to select additives to more effectively weaken the bonds in NaBH 4 , thus improving decomposition and H 2 release to make NaBH 4 a more viable hydrogen storage material.

Discussion
This review focuses on the changing decomposition temperatures of NaBH 4 when using the following additives to destabilize the NaBH 4 structure: Ti, Ni, Mn, Fe, Zr, and Na. A brief description of the synthetic methodology and a discussion of results are described below.  [30]. Decomposition temperature was measured by TGA: heated to 600 °C at 2 °C/min with Ar purge of 100 mL/min.

Titanium Additive Results and Discussion
Several different titanium (Ti) additives have been found to lower the decomposition temperature of NaBH 4 , including hydrides, fluorides, and metal nanoparticles. Though there have been obvious challenges with the formation of intermediates and in developing trends to understand the influence of oxidation states, the fact is that these additives have been found to influence the decomposition and dehydrogenation. Figure 1 shows that the lower decomposition temperature was obtained with TiSiO 4 (Ti +4 ) additive, at 489 °C. Humphries et al. found that metallic Ti and Ti +2 in TiB 2 decompose at higher temperatures than this (493 °C and 503 °C, respectively). The powder x-ray diffraction (PXD) patterns show that the addition of TiSiO 4 and TiB 2 did not change the diffraction patterns of NaBH 4 during the BM process [33]. However, after milling there was an additional diffraction peak observed for Ti that suggested the presence of titanium metal [38]. Likewise, Mao et al. found that the TiF 3 PXD exhibited two phases [32,38] after BM, suggesting the presence of the solid additive. Looking to the study of milled TiF 3 with Mg(BH) 4 [46], there was a destabilisation of NaBH 4 due to an interruption of its framework. Furthermore, Kalantzopoulos et al. found that the boron nuclear magnetic resonance (NMR) spectra of NaBH 4 with the TiF 3 additive contained NaBF 4 in a low ratio of 1:150 NaBF 4 :NaBH 4 . It was assumed that NaBF 4 was in the amorphous state as it was not detected by PXD. Its formation can be explained thermodynamically during the warm-up induced by BM. This compound facilitates the dehydrogenation process, which will be discussed later, as it is an intermediate of the decomposition products NaF and H 2 [38]. Unfortunately, it is difficult to compare this additive effect to the others because the decomposition temperature is unknown. These samples are all physical mixtures, but the absence of additional PXD peaks for TiSiO 4 and TiB 2 indicates their possible presence in the amorphous state. Moreover, a similar reaction to the one with TiF 3 occurred during the BM using TiF 4 , where NaBF 4 was formed [30]. Surprisingly, despite this substitution reaction, the decomposition temperature of this mixture (522°C) was only slightly lower than that of pure NaBH 4 (532 °C).
A first estimation of capacities is given by hydrogen desorption rate measurements at 440 °C for Ti (65 nm), TiH 2 and TiF 3 . Within 400 minutes, the NaBH 4 -Ti mixture desorbs 1.65 wt% of H 2 , TiH 2 desorbs 1.85 wt%, and TiF 3 releases 3.2 wt% [32]. It should be noted that in the 20:1 mixture of NaBH 4 :TiF 3 , the first release of H 2 observed was from 300 to 360 °C, with a second step at 380 °C [32]. The more additive-concentrated sample, with a 4:1 ratio of NaBH 4 :TiF 3 , had lower release temperatures, with a first step from 220 and 320 °C and a second from 320 and 420 °C. Humphries et al. reported in-situ powder PXD patterns ( Figure 2) [38]. From room temperature (RT) to 220 °C, the formation of Na 3 TiF 6 was detected, whose diffraction intensity decreased until it disappeared at 320 °C, at which point NaF was formed. Thus, the two steps decomposition process of NaBH 4 -TiF 3 are related firstly to the dehydrogenation that occurs during the Na 3 TiF 6 formation, then to the decomposition of NaBH 4 decreased by the presence of the ternary fluoride [38]. In contrast to this, Mao et al. reported TiB 2 and NaF formation during TiH 2 and TiF 3 decomposition, which was determined to be an active species, but they made no mention of the appearance of Na 3 TiF 6 . They also commented that TiF 3 had better H 2 desorption and adsorption capacities compared to TiH 2 and Ti.

Figure 2
In-situ powder XRD of NaBH 4 /TiF 3 system between 50 and 560 °C with a heating rate of 2 °C/min [38]. Reproduced from Ref. [38] with permission from the PCCP Owner Societies.
The thermogravimetric analysis (TGA) of a NaBH 4 and TiF 4 mixture indicated a mass loss of 0.3 wt% between 100 and 300 °C, and 20% between 300 and 600 °C, with H 2 desorption starting at about 470 °C [30]. It is difficult to compare TiF 4 to TiF 3 because the time exposures to temperatures were not reported, generated species were not specified in the case of TiF 4 , and the ratios used differed. However, it is noteworthy that the decomposition of NaBH 4 with TiF 4 seemed particularly efficient, which does not contradict the observation made by Kalantzopoulos et al. about the effect of oxidation state in fluoride-containing additives.
It was observed that high oxidation states generally tend to more strongly decrease melting and decomposition temperatures. Jansa et al. [30] explained that the high coordination of the metal induced a high exchange area between NaBH 4 and the anions of the additive, and the additive cation furnished some vacancies that permitted further reactions. This was noticed using both chloride and fluoride additives milled with MgH 2 [47,48]. For instance, NaBH 4 -CeF 4 decomposed at 502 °C whereas NaBH 4 -CeF 3 decomposed at 506 °C . Furthermore, NbF 5 (Nb +5 ) is notably one of the best destabilizers of NaBH 4 . However, some exceptions exist; FeF 2 and CrCl 2 more effectively reduce the decomposition temperature than FeF 3 and CrCl 3 [47]. Electronic structures of metals, depending on their oxidation state, are responsible for creating bonds during milling or heating, which influence the decomposition process. Indeed, it was highlighted earlier that TiF 3 and TiF 4 behaviours during BM, then heating, are very different. In addition, their effect on NaAlH 4 unit-cell dimensions also differ slightly, where the unit-cell parameter using TiF 3 is 5,0231 Å and using TiF 4 it is 5.0235 Å [49].
Ti additives offer much in the way of improving our understanding of how additives influence the decomposition of NaBH 4 . Though the lowest decomposition temperature reported reaches only 493 °C, H 2 release has been reported in these mixed materials with an on-set temperature as low as 220 °C. There are many challenges to overcome if researchers pursue the use of Ti additives. The formation of intermediates such as Na 3 TiF 6 and NaF make the chemistry of decomposition difficult to study, and raise questions about the potential for recyclability. More controlled studies are needed to gain a solid understanding of the influence of Ti oxidation states as well. The lowest reported decomposition onset temperature was 493 °C when using 65nm Ti nanoparticles as additives. Given the growing popularity of the study of nanoparticles and the well-known fact that nano-sized particles oftentimes have different tuneable properties compared to their bulk-sized counterparts, it would be worthwhile to continue this direction of research to see if smaller nanoparticles would lower the decomposition even further. It's possible that Ti will not lower the decomposition and dehydrogenation temperatures enough to be feasible for H 2 storage and transportation, but the potential insights gained about oxidation and particle size trends suggest this is a material in need of additional study.

Ni Additive Experimental Procedures
Nickel (Ni) additives have been studied using a traditional BM technique. Humphries et. al. has reported data for traditional milling methods using five different nickel (Ni) additives; Ni nanoparticles (20nm), Ni 3 B, NiCl 2 , NiF 2 and Ni on Si/AlO 3 . A 2 molar % concentration of additives were BM at 280rpm with NaBH 4 under argon atmosphere for 1 hour at 280 rpm, resulting in the final mixed products [33]. The decomposition temperature was measured by TGA: heated to 600 °C at 2° /min with an Ar purge of 50 mL/min.

Ni Additive Results and Discussion
Unlike the Ti additive, pure nickel (Ni) nanoparticles used as traditional additives through BM do not reduce the decomposition temperature better than their halogenated counterparts. It is fortunate that BM the Ni additives with NaBH 4 does not result in the creation of many boron derivatives, which is promising for H 2 cyclability, but chloride additives do tend to create Na(BH 4 ) 1-x Cl x . Though halogens have been mentioned, it remains to be seen how the oxidation state of Ni influences decomposition.
Both halide additives, NiCl 2 and NiF 2 , appear to be the best destabilizers of nickel compounds just behind the mixed material Ni on Si/AlO 3 . NiCl 2 decreases the NaBH 4 decomposition temperature to 460 °C and NiF 2 to 453 °C. The other additives examined, Ni 3 B and Ni particles, with a much lower oxidation state (Ni 0 ), were nevertheless effective as they lowered the decomposition temperature of NaBH 4 to 462 °C and 483 °C, respectively. This temperature distribution is consistent with observations made earlier with the Ti additives with regard to oxidation states. Humphries et al.'s first PXD study affirms that no reaction took place during the BM using each Ni additive, and no borane derivatives were generated, which favours reversibility [33]. Another experiment using more concentrated samples (with 20 and 25 molar % of additive) reported substitution reactions when using chloride derivatives such as NiCl 2 and TiCl 2 . The appearance of Na(BH 4 ) 1-x Cl x during BM was brought to light by PXD and FTIR [50]. This phenomena, for which the substitution ratio depends on the transition metal of the additive, has also been observed using a NaCl additive [36], which will be discussed in the sodium (Na) section below.
According to Humphries et. al., the only gas released during heating of these mixed materials was H 2 . The group studied desorption by cycling pressure-composition-temperature (PCT) and found Ni 3 B and NiCl 2 to be the most effective Ni additives (Table 1). After the first cycle, the desorption using NiCl 2 at 514 °C reached up to 0.25 wt%, and 1.48 wt% at the second cycle, and with Ni 3 B at 460 °C the desorption was up to 1.45 wt%. It was revealed by PXD that Ni additives form Ni 3 B during heating. Given the Ni 3 B efficiency at destabilizing NaBH 4 during desorption, it was logical to observe encouraging results with other Ni components. The particular destabilizer effect of the chloride-based nickel compound could be correlated to the appearance of Na(BH 4 ) 1-x Cl x during BM, but it is not sufficient to explain its best hydrogen desorption during the second cycle. Nevertheless, NiCl 2 seems to show the most promise for H 2 storage applications, but with the disadvantage that it desorbs only at high temperatures. On the other hand, NiF 2 is advertised as being less effective but it has been analysed at a temperature 50 °C lower than that of chloride whereas NaBH 4 -NiF 2 has just a 7 °C lower decomposition temperature. Furthermore, after a few dehydrogenation/rehydrogenation cycles, formation of metallic Na from the decomposition of NaBH 4 and NaH from the hydrogenation of metallic Na were observed in each case. The general results of Humphries et al. showed that chloride and boride are better than fluoride additives in decreasing the decomposition temperature of NaBH 4 , as observed in Figure 1. It is fortunate that heating NaBH 4 -Ni mixtures produce very few boron derivatives. Though this is promising for recyclability, there are many challenges yet to overcome. For example, fluorinated Ni reduces the decomposition temperature but increases the H 2 desorption temperature. There is also a need to study NaBH 4 systems with varying Ni oxidations states, as this property may be important in lowering dehydrogenation, but has yet to be studied. Additional work is needed and encouraged to determine whether Ni additives may be the key to making H 2 more accessible in NaBH 4 .

Mn and Fe Additive Experimental Procedures
Several groups have studied Mn and Fe as additives with NaBH 4 . Below are the methods that have been successful in reducing the NaBH 4 decomposition temperature:  Humphries et al. ball-milled 25nm Fe nanoparticles for 1h at 280 rpm in a 2 molar % proportion with NaBH 4 [33]. Decomposition temperature was measured by TGA: heated to 600 °C at 2 °/min with Ar purge of 50 mL/min.  Jansa et al. took a different approach and BM FeCl 3 with NaBH 4 in the ratio 1:4 under argon during 3 hours, with 1:100 powder to stainless balls ratio, and MnCl 2 was BM under the same conditions with an additive/NaBH 4 ratio of 1:3 [50]. Decomposition temperature was measured by TGA: heated to 600 °C at 2 °/min with Ar purge of 50 mL/min.  Kalantzopoulos et al. also incorporated Fe additives, where 4NaBH 4 :1FeF 3 and 4.8NaBH 4 :1MnF 3 were BM at 400 rpm under argon for 3 hours, under 1 bar of argon gas [38]. Decomposition temperature was determined via TGA by heating the sample between 30 and 600 °C with a heating rate of 21 °C and an argon gas flow of 50 mL/min.  Sing et al. stirred MnCl 2 with a mortar and pestle with NaBH 4 in 20ml diethyl ether for 1 hour at 35 °C under nitrogen gas until the solvent evaporates totally. MnCl 2 was used in 10 wt%, 20 wt%, 30 wt%, 40 wt%, 50 wt% proportion in NaBH 4 [39]. The decomposition temperature was measured by TGA heating from ambient to 300 °C with a heating rate of 5 °C/min under a nitrogen gas flow rate of 50 mL/min.

Mn and Fe Additive Results and Discussion
Similar to what was discussed above, manganese (Mn) and iron (Fe) additives also generate boron derivatives during the milling process, and the use of chloride-based additives produces the same substituted Na(BH 4 ) 1-x Cl x structure reported for NiCl 2 , TiCl 2 and NaCl. While these intermediates may pose problems for H 2 cycling, they also offer another means of reducing the decomposition and dehydrogenation temperatures of NaBH 4 .
Kalantzopoulos et al. found that a portion of MnF 3 was reduced into MnF 2 , which lead to the formation of NaBF 4 during milling with NaBH 4 [38]. Boron NMR showed a NaBH 4 /NaBF 4 ratio of 40:1. The same product was generated in a higher amount (10NaBH 4 :NaBF 4 ) when FeF 3 was used, but in an amorphous state undetectable by PXD. In the latter case, FeF 3 was reduced to metallic Fe, which led to the formation of NaF. The presence of NaBF 4 in the MnF 2 sample induced an overlap marked by an asymmetry of the melting peak of the differential scanning calorimetry (DSC) pattern [30]. This was not observed for the FeF 3 sample in which NaBF 4 was amorphous.
Residual gas analysis found that the sample containing Na-MnF 3 released H 2 in three steps. The hydrogen desorption started at a low temperature (135 °C) due to the high destabilization effect of MnF 2 (Figure 3) [38]. The first step occurred between 160 and 230 °C, when NaBH 4 reacted with MnF 2 to form the ternary fluoride NaMnF 3 with a release of 2.0 wt%. Then, between 230 and 280°C, 1.5 wt% was released. Finally, after 390 °C, 1.8 wt% was desorbed. For the Na-FeF 3 sample, from 161 to 350 °C, 1.7 wt% was released, with a high H 2 desorption rate between 250 and 350 °C. This was consistent with the in-situ PXD signal of NaBH 4 which started decreasing at 270 °C, and was followed by 1.5 wt% desorption between 350 and 490 °C. Given these dehydrogenation results, MnF 3 may be slightly better at destabilizing NaBH 4 than FeF 3 . However, the decomposition temperatures of NaBH 4 with MnF 3 and FeF 3 are very similar, 483 °C and 498 °C, respectively (Figure 1)
The use of Fe and Mn additives to reduce the decomposition temperature of NaBH 4 has meet with some success. Interestingly, when compared to Ti additives, the opposite trend of halogen size occurs, with the Cl having the lower decomposition temperature (reaching 437 °C with FeCl 3 and 498 °C with FeF 3 ). This prohibits researchers from drawing all-encompassing conclusions about halogens attached to metal additives. It's possible the trend here differs because of the multi-step process of dehydrogenation, or that the boron derivatives must be taken into account when attempting to draw conclusions about best additive selections. Additional research is needed to determine whether or not Mn and Fe additives may have a different outcome when incorporated as nanoparticles, and if other types of additives containing these metals might prevent the formation of the boron derivatives during BM.

Zr Additive Experimental Procedures
Two zirconium (Zr) additives have been used in an attempt to improve the H 2 desorption of NaBH 4 :  Kumar et al. ball-milled NaBH 4 with 1 mol % ZrCl 4 using stainless steel balls under 0.1 MPa argon gas for two hours, with a 30 minute break to cool the sample half way through the milling process [31]. Decomposition temperature was measured by TGA: heated to 600 °C at 2 °C/min with an Ar purge of 50mL/min.  Jansa et al. ball-milled NaBH 4 with 2 mol % ZrF 4 at 300rpm under an argon atmosphere for one hour [30]. Decomposition temperature was measured by TGA: heated to 600 °C at 2 °C/min with an Ar purge of 100 mL/min.

Zr Additive Results and Discussion
It is possible to reduce the decomposition temperature of NaBH 4 using halogenated Zr additives. Unlike the previous additives discussed, the chlorinated ZrCl 4 decomposes NaBH 4 at a significantly lower temperature than the fluorinated ZrF 4 , with a difference of more than 100 °C. Researchers have yet to determine why the different halogens yield such drastically different outcomes, but it is promising that work has been done to determine how the BM process changes the chemistry of these additives.
Studies involving ZrCl 4 have found that this additive interacts with the surfaces of magnesium borohydride and magnesium hydride to improve H 2 release while reducing to ZrCl 3 /ZrCl 2 and nanosized Zr metal [51,52]. When added into the NaBH 4 system, a similar change of the precursor occurs. The thermodynamic calculations of Kumar et al. discovered that ZrCl 4 reacts with NaBH 4 during BM via the following two pathways [31]: 4 NaBH 4 + ZrCl 4  4NaCl + 4B + Zr + 8H 2 (g) 2 NaBH 4 + ZrCl 4  2NaCl + 2B + ZrCl 2 + 4H 2 (g) The ZrCl 4 reduces partially to ZrCl 2 and completely to metallic Zr. Kumar et. al reported that the ZrCl 4 additive reduces the crystallite size of NaBH 4 during BM, making NaBH 4 more reactive as a result of the increased surface area. Interestingly, though it was clear that ZrCl 4 reacted with NaBH 4 , no NaCl was detected by PXD as a byproduct upon completion of the BM process. This problem with detection was either due to small concentrations or else the NaCl was amorphous, so not measurable by this technique.
TGA results comparing pure NaBH 4 (p-NaBH 4 ) and ZrCl 4 -containing NaBH 4 (c-NaBH 4 ) are shown in Figure 4 [31]. The decomposition and dehydrogenation of the milled p-NaBH 4 began at 475 °C, with the sample losing more than 60% of its weight by 600 °C. This weight loss reportedly was a result of both hydrogen desorption and the evaporation of Na metal, the latter of which does not bode well for recyclability of this material. The ZrCl 4 additive, on the other hand, had a significant impact on the decomposition temperature, reducing the onset temperature to below 300 °C, with most of the H 2 content being released by 500 °C.
The Fourier-transform infrared spectroscopy (FTIR) data showed no changes between pure and ZrCl 4 -containing NaBH 4 , while x-ray photoelectron spectroscopy (XPS) confirmed that there were notable changes to the oxidation state of the Zr additive. In fact, all XPS peaks shifted to lower energy after NaBH 4 and ZrCl 2 were milled together, with energy changes of the Zr3d electrons pointing to the Zr 4+ oxidation shifting to Zr 2+ after the reaction. Overall, the data reported found that ZrCl 4 reduced to ZrCl 2 , but it is unclear if metallic Zr was formed or if ZrCl 2 was in any way responsible for the effect observed.
Unlike other metal examples listed in this review, the fluorinated ZrF 4 additive did not reduce the decomposition temperature of NaBH 4 nearly as much as the chlorinated ZrCl 4 , reaching only 503 °C compared to 400 °C. The ZrF 4 additive also lost its diffraction pattern after BM, corresponding to the formation of metallic Zr and NaBF 4 . It remains to be seen if cyclability of H 2 is possible with this complex, but given the high temperature required to decompose NaBH 4 with ZrF 4 , it is likely that this mixed material is not worth pursuing.
Boron derivatives remain an ongoing challenge in the quest to find materials that will work for H 2 cycling, and Zr additives are not immune to this problem. However, the ability of the ZrCl 4 additive to decrease the decomposition so low may be one of the keys to designing a better material. Though going nano in size doesn't always solve the problem, it would be interesting to see how Zr nanoparticles influence the decomposition of NaBH 4 , and to see if more complex Zr additives have a different outcome.

Na Additive Experimental Procedures
Several groups have reported the use of Na additives with NaBH 4, with varying results as to changes in decomposition. Below are the methods reported that were successful:  Chater et al. mixed NaBH 4 with NaNH 2 in a 1:1 molar ratio and heated at 190 °C for 12 h under argon [53]. For a NaBH 4 -NaNH 2 (1:1 molar ratio) high temperature study, NaNH 2 -NaBH 4 was milled and heated to 220 °C for 1 h, then the mixture annealed at 70 °C for 10 days to reach the low temperature phase. Decomposition temperature was measured using a homebuilt temperature programmed desorption apparatus coupled to a quadrupole mass spectrometer. The sample was heated under a 100 mL/min flow of argon up to 350 °C at a rate of 2 °C/min.  Wu et al. BM a mixture of NaNH 2 and NaBH 4 (2:1 molar ratio) under argon at 300 rpm from 1 to 32 h [40]. Decomposition temperature was measured by TG-DTA: heated to 450 °C at 10 °C/min with an Ar purge of 10 ml/min.  Olsen et al. changed concentrations using a NaCl additive. A mixture of NaBH 4 and NaCl with molar ratios 9:1, 3:1, 2:1, 1:1, 1:2, 1:3, 1:9 were ball milled under argon at 500 rpm for 24 hrs. Samples were taken out after 2, 6, and 12 h of milling [35]. Decomposition temperature was measured by DSC: heated to 600 °C, at 2 °C/min.  Ravnsbaek et al. also used NaCl as an additive. NaBH 4 with NaCl were BM with respective molar ratios of 0.5:0.5 and 0.75:0.25 under argon 60 times with 2 min of milling followed by 2 min breaks to prevent overheating. After BM, 0.5g of the mixture was transferred to an Al 2 O 3 crucible, placed inside a sealed argon filled quartz capillary, and annealed in a furnace at 300 °C for 3 days [36]. Decomposition temperature was measured by Sieverts apparatus: heated to 450 °C at 0.5 °C/min.  Rude et. al. mixed a 0.1:1 ratio of NaBF 4 to NaBH 4 , BM 15 times under argon with 2 min milling followed by 2 min breaks to cool. Decomposition temperature was measured by TGA/DSC heated to 450 °C at 5 °C/min. [54] 2.5.2 Na Additive Results and Discussion Sodium (Na) as an additive for NaBH 4 has mainly been explored in two forms, NaNH 2 and NaCl. As discussed already in this review, the use of chloride-based additives, such as NaCl, produces the same substituted Na(BH 4 ) 1-x Cl x structure reported for NiCl 2 , TiCl 2 and NaCl, while NaNH 2 has shown to produce ammonia emissions. Despite this, the low decomposition and dehydrogenation temperatures for both NaCl and NaNH 2 show promise as additives as hydrogen storage materials. NaBH 4 -NaNH 2 is a composite hydrogen storage material (composed of an amide and a borohydride) which has a theoretical hydrogen storage of more than 7 wt% and the gas product obtained after heating is mainly H 2 . NaBH 4 -NaNH 2 was studied because hydrogen generation performances of composite materials such as NaBH 4 -NaNH 2 is considered potentially superior to pure borohydrides or amides. The melting point of NaNH 2 is 210 °C and the boiling point is 405 °C. Wu et al. heated NaNH 2 to 420 °C and found that NaNH 2 decomposes in two stages. The first stage, prior to the melting point, the weight loss is only about 1.5 wt% [40]. The second stage, after the melting point, the weight loss is 11.6 wt%. Based on TG-DTA data for NaNH 2 -NaBH 4 (2:1 molar ratio), this composite's decomposition mechanism appears to be influence by NaNH 2 , also having two stages, starting at 124 °C and ending at 420 °C with a total weight loss of 7.52 wt%. Wu et al. found that the thermal decomposition of NaNH 2 _NaBH 4 (2:1) was best controlled below 400 °C [5].
Through a different study, it was found that ball milling a 1:1 ratio of NaBH 4 to NaNH 2 produced a new complex, Na 2 (BH 4 )(NH 2 ). This new compound desorbed hydrogen at about 290 °C after its melting point at about 220 °C [53]. This new compound's high-temperature and low-temperature phase was studied and from the high temperature XRD and differential thermal analysis, NaNH 2 was found to react with NaBH 4 at 157 °C with good stability up to its melting point (219 °C), with no detectable structure change from 219 °C to 300 °C. Above 300 °C, the dehydrogenation reaction took place [41].
Another Na compound, NaCl, was studied by Olsen et al. as an additive for NaBH 4 . NaCl was chosen for its potential to alter the thermodynamic properties of borohydride materials through anion substitution [35]. The powder X-ray diffraction (PXD) of NaBH 4 _NaCl with a 1:1 molar ratio showed that NaCl dissolves faster in NaBH 4 than vice versa and that 12 hours of milling is sufficient to reach a single phase for mixtures containing less than 50% NaBH 4 , while more than 50% required 24 hours to reach full homogeneity.
Ravnsbaek et al. found that annealing NaBH 4 _NaCl after ball milling was better than ball milling alone. Synchrotron radiation powder X-ray diffraction (SR-PXD) data showed that the annealed NaBH 4 _NaCl dissolved NaCl into NaBH 4 at about 240 °C. It was concluded from the SR-PXD data investigating of the decomposition pathway that at higher temperatures (~415 °C), the rate of diffusion is increased, and a homogeneous solid solution is formed before the NaBH 4 -part decomposed at ~530 °C [36].
Rude et. al. combined a sodium tetrafluoroborate (NaBF 4 ) additive with NaBH 4 to make a NaBH 4 -NaBF 4 composite, using PXD to directly observe the substitution of fluorine for the BH 4 group of NaBH 4 [54]. New resonances found in 19 F magic-angle spinning NMR supported DFT calculations and confirmed the formation of a complex ion, BF 2 H 2 , after BM. The H-F exchange destabilized the NaBH 4 structure enough to reduce the decomposition temperature to as low as 300 °C. Though other research groups have used different fluoride-based metal additives, the use of Na prevents confusion of the cations in the system and made it possible to observe the direct H-F exchange. Given the dramatic decrease in decomposition it is certainly worth pursuing other fluorine-based metals to determine whether the combination of a F exchange as well as the appropriate metal might enable the NaBH 4 system to come closer to approaching DOE's guidelines for H 2 release. Furthermore, it is known that anion substitution can be readily accommodated in borohydrides [55], and there is certainly a need to pursue these additives further with the goal of decreasing the decomposition temperature of NaBH 4 .
Including the NaCl additive with NaBH 4 not only reduced the NaBH 4 decomposition to 415 °C, but yielded a H 2 release of 2.6 wt%, which is 90% of the calculated capacity. It was unfortunate that NaBO 2 formed during annealing, as this raises questions about cyclability and prevents the mixed material from reaching 100% of the calculated capacity. However, reaching 90% H 2 release is still an impressive feat and is deserving of further exploration. In contrast, NaNH 2 as an additive can dehydrogenate at 300 °C with a decomposition temperature of 400 °C, but it produces ammonia emissions, which create additional concerns [31]. It is intriguing that milling NaNH 2 with CoNiB reduces the dehydrogenation temperature to 200 °C with 5.05 wt% hydrogen release and no ammonia emissions [30,31,53]. NaBH 4 containing NaCl and NaNH 2 are promising materials compared to mixtures using other additives in this review due to their low decomposition and dehydrogenation temperatures and ability to release a large amount of their H 2 . Add to this how promising the addition of CoNiB is to these mixtures, and it is clear that Na additives deserve further exploration.

Conclusions
Controlling the decomposition temperature of NaBH 4 with additives has met with a wide range of successes, with the decomposition onset temperature ranging from 522 °C with TiF 4 to as low as 300 °C with H-F exchange in NaBH 4 -NaBF 4 . Unfortunately, researchers have yet to find a common trend to allow them to destabilize NaBH 4 closer to room temperature. Some additive studies have found that fluorides are better than chlorides at weakening NaBH 4 , especially when it is confirmed that the fluoride can substituted into the BH 4 group, or in additive mixtures involving MgH 2 . Ni and Ce additives with fluorides also create a lower decomposition temperature in NaBH 4 , but not enough work has been done to determine whether this is uniformly true for all metals. It's also unclear if there is a trend where metal nanoparticle additives reduce the decomposition more than their complimentary halogenated metals. For example, Ti nanoparticle additives create a significantly lower decomposition than TiF 4 , yet the reverse is not true when comparing Ni nanoparticles to NiF 2 .
Researchers have yet to find a reliable method to weaken NaBH 4 enough to make it a viable option for H 2 storage, and much more progress is needed to meet DOE's guidelines to release H 2 below 125 °C . It is clear that in order to make a cyclable H 2 storage material of NaBH 4 , it will be important to find additives that do not create boron derivatives during milling or annealing, a common challenge in the majority of the additives discussed in this review. Our hope is that compiling information on these additives will enable researchers to gain a better understanding of additive incorporation and to determine how a new system might be designed to make NaBH 4 a viable H 2 fuel source. It's possible that creating a NaBH 4 mixture in-situ would be better than ball milling, or that different types or sizes of nanoparticles could help catalyze NaBH 4 . There is still a lot that is unknown about NaBH 4 and additional research is needed before abandoning it as a potential energy source of the future.