Effect of Vanadium Catalysts on Hydrogen Evolution from NaBH₄

Abstract

NaBH₄ is a very cheap and hydrogen-rich material, as well as a potential hydrogen store. However, the high temperature of its thermal decomposition (above 530 °C) renders it inapplicable in practical use. Here, we studied the effect of addition of diverse V-containing catalysts on thermal hydrogen desorption. It turns out that mechanochemical doping of NaBH₄ with vanadium metal, its oxides, or nanoparticles lowers the temperature of pyrolysis significantly. Notably, NaBH₄ milled for 3 h with 25 wt.% V₂O₅ or VO₂ releases ca. 70% of stored hydrogen in the temperature range of ca. 370–450 °C. On the other hand, precursors and solvents used to prepare rather uniform vanadium nanoparticles (~4 nm) suspended in THF or less uniform and larger ones (~15 nm) in o- difluorobenzene have adverse effect on the purity of hydrogen evolved.

1. Introduction

Fossil fuels have been widely used as an energy source all over the world for centuries. Their intense usage has led to environmental pollution and climate change, and this has forced humanity to seek alternative, green energy sources and carriers. Hydrogen, next to none in its high gravimetric energy density, is a clean source of energy, with its combustion resulting only in water. This makes it a promising alternative to both fossil energy carriers and electric energy accumulators [1,2,3,4,5].

Sodium borohydride is one of the cheapest and most hydrogen-rich compounds known, having highly competitive volumetric hydrogen density [6]. Therefore, it is not surprising that much effort has been devoted to the possibility of using it as a hydrogen storage material. Hydrogen can be released from NaBH₄ either during its hydrolysis or upon thermal decomposition. Hydrolysis reactions in alkaline aqueous solutions take place at room temperature upon contact with a catalyst (so called Hydrogen-on-demand^® technology [7]). However, the reversibility of this process is poor, as it is associated with the necessity to recycle very stable oxoborates(III) back to BH₄⁻ species. On the other hand, the pyrolysis profile has a maximum desorption peak above 500 °C [8] (values up to 534 ± 10 °C at 1 bar of hydrogen [9] are given in the literature), which is quite high for mobile applications. Three main approaches taken to lower the decomposition temperature are catalytic enhancement of pyrolysis, tuning it via hydride mixing, and nanosizing of hydrogen-containing species [10,11,12]. Various dopants have been studied by numerous teams, and we can group them for convenience into four main categories (

Table 1. Influence of the first-period transition metal-based dopants on thermal decomposition temperature of NaBH₄ (5%_mol doping if not specified differently).

Sc [13]	Ti	V	Cr [13]	Mn [13]
ScF3—501 °C	Nano—493 °C [14] TiF3—300 °C [15] 5%mol TiF3—~300 °C [16] TiF4—522 °C [13] TiCl3 1:6—100 °C * [17] TiB2—503 °C [14] TiSiO4—489 °C [14]	VF4—499 °C [13] NaBH4 + PrF3 + VF3 3:1:0.2—417 °C [18] V3B—478 °C [14]	CrF3—500 °C	MnF3—483 °C
Fe [13]	Co	Ni	Cu [13]	Zn
FeF3—498 °C	CoF3—501 °C [13] Co3B—480 °C [14]	Nano—483 °C [14] NiF2—452 °C [13] NiF2—453 °C [14] wt.10% Ni on Si/Al2O3—449 °C [14] NiCl2—460 °C [14] Ni3B—462 °C [14]	CuF2—476 °C	ZnF2—504 °C [13] ZnF2 1:2—95 °C ** [19]

* polluted with B₂H₆ and diglyme; ** polluted with B₂H_6.

,

Table 2. Influence of the second-period transition metal-based dopants on thermal decomposition temperature of NaBH₄ (5%_mol doping if not specified differently).

Y	Zr	Nb [13]	Rh [14]	Pd [14]	Ag [13]	Cd [13]
YF3—513 °C [13] YF3 3:1—423 °C [20]	ZrF4—503°C [13] ZrCl4—503 °C [21]	NbF5: 2%mol—442 °C 10%mol—379 °C 15%mol—379 °C	Rh (5 wt.%) on Al2O3—476 °C	Nano—420 °C	AgF—498 °C	CdF2—512 °C
(x)YF3 + (1 − x)GdF3 3:1 [22]: x = 2/3—449 °C x = 1/2—446 °C x = 1/3—440 °C

,

Table 3. Influence of lanthanide series metal-based dopants on thermal decomposition temperature of NaBH₄ (5%_mol doping if not specified differently).

La [23]	Ce [13]	Pr [18]	Nd [24]	Ho [25]
LaF3 3:1—396 °C La 3:1—442 °C	CeF3—506 °C CeF4—502 °C	PrF3 3:1—439 °C	NdF3 3:1—413 °C	HoF3 3:1—443 °C

and

Table 4. Influence of co-milling of other hydrogen-rich compounds with NaBH₄ on its thermal decomposition temperature (molar ratio was used if not described differently).

MH2	Mg2NiH4 [26]	NaNH2	M(AlH4)x	Li3AlH6 [27]	MBH4
M = Mg: 1:2—420 °C [28] 1:2—490 °C [29] 1:2 with 5%mol of [30]: TiF3—~470 °C TiO2—~470 °C Zr—~470 °C Si—~470 °C BCC—~470 °C M = Ca: 1:6—390 °C [31]	5:4—~350 °C	1:1—330 °C * [32] 2:1 with Co–Ni–B catalyst [33]: 1 wt.%—~300 °C 3 wt.%—~285 °C 5 wt.%—285 °C 7 wt.%—~285 °C	M = Li: 1:1 [16]: No catalyst—446 °C 5% TiF3—~300 °C M = Ca: 2NaAlH4 + Ca(BH4)2 with 5 wt.% TiF3 —400 °C [34]	1:1 (1 h)—400 °C 1:1 (24 h)—392 °C 1:2—430 °C 1:3—430 °C	M = Li 0.38:0.62: pure—350 °C [35] pure—~420 °C [36] 60%vol in carbon areogel—~333 °C [35] M = K—~472 °C [37] M = Mg—~205 °C [38] M = Li, K, Mg, and Ca (equimolar)—~330 °C [39]

* polluted with ammonia.

): (I) first-period transition metal (TM) elements and their compounds, (II) second-period TM elements and their compounds, (III) lanthanides and their compounds, and (IV) other hydrides. Additionally, composites of NaBH₄ with diverse “inert” materials such as graphene and zeolites have been prepared, and their thermal profiles have been studied (

Table 5. Thermal decomposition of NaBH₄ in its composite materials (mass ratio was used if not described differently).

Ionic Liquids [40]	Carbon	Carbon Based	Zeolites [41]	Ah-BN [42]	NaBF4 ** [34]
1%mol solution in: bmimBr—~165 °C bmimNTf2—104 °C	20% of scaffolds CMK-3 —235 °C and 380 °C * [43] ~25% of HSAG-500 [44]: infiltrated—~270 °C melted—~300 °C mixed—~320 °C	45% fluorographite—141 °C [45] graphene: 10%—370 °C [46] ?—426 °C [47]	MCM-22: 1:1—490 °C 1:2—503 °C 1:3—507 °C 1:4—508 °C SAPO-34: 1:1—483 °C 1:2—491 °C 1:3—493 °C 1:4—495 °C	1:1—399 °C	1:10 mol—468 °C 1:2 mol—305 °C

* polluted with ammonia; ** polluted with BF₃ and B₂H_6.

). Last but not the least, attempts to prepare NaBH₄ nanoparticles to decrease the thermal decomposition temperature of the pristine material are gathered in the last group (

Table 6. Influence of nanosizing of NaBH₄ on its thermal decomposition temperature.

Solvented Ion Stabilization * [48]	Evaporation with 10%mol of Ligands or Solvents [49]	Anti-Solvent Precipitation	Milled LiBH4 + NaCl
LiCl—495 °C MgCl2—445 °C NaI—485 °C	Hexyloamine—489 °C dodecyloamine—476 °C octodecyloamine—473 °C tertabuthylamonium bromide—464 and 482 °C tetraoctylamonium bromide—484 °C tetradecyloamonium bromide—~500 °C dodecane—478 °C dodecanothiol—464 °C tri-decylic acid—~400 and 474 °C tetrabutylphosphonium bromide—471 °C	Bare—460 and 535 °C [50] Coated with: Ni—418 °C [50] Co—350 °C [51] Cu—400 °C [51] Fe—380 °C [51] Sn—450 °C [51]	Graphene coated (ultrasonicated and dried)—400 °C [47] Milled with graphene nanoflakes—~300 °C [52] Milled 1:10 mass with [53]: graphite—109 °C graphene oxide—114 °C carbon nanotubes—116 °C

* polluted with buthylamine.

).

Some of the most spectacular results achieved via doping with the first-period TM species (Table 1) were seen for zinc(II) fluoride and titanium(III) chloride, which evolve hydrogen around 95 and 100 °C, respectively. However, as typical for the rather low temperature of thermal decomposition, H₂ gas is, in these cases, severely polluted with B₂H₆ or B₂H₆ and diglyme. For metals from the second TM period (Table 2), the lowest temperature of thermal decomposition of 379 °C was achieved for 10%_mol and 15%_mol doping with NbF₅. The third-period TM catalysts were also tested but the decrease in thermal decomposition temperature was not impressive (TaCl₅—460 °C and ReCl₃—465 °C [14]). In lanthanide series (Table 3), the best result was achieved for neodymium fluoride doping (413 °C).

Co-milling NaBH₄ with other hydrogen storage materials (Table 4), particularly protic ones, often leads to facile evolution of H₂ from a composite system. As for NaBH₄, particularly promising results were achieved for 5 wt.% Co-Ni-B doped 2NaNH₂/NaBH₄ mixture, milled in hexane environment, for which the temperature of pyrolysis with H₂ evolution dropped to a value as low as 285 °C. However, hydridic–protic stores rarely show sufficient reversibility [54,55].

If composites with diverse “inert” materials are considered (Table 5), substantial (45 wt.%) fluorographite addition resulted in the thermal decomposition point of 141 °C; an also interesting approach with the use of ionic liquid led to pure hydrogen desorption starting at 104 °C. Other composites usually yielded less appealing results.

The last group of proposed NaBH₄ modifications involves either nanostructurization or coating it with metal or graphene layers (Table 6) which could lead to a decomposition temperature as low as 350 °C for NaBH4 nanoparticles coated with cobalt.

It may be seen from the literature screening that vanadium compounds have rather seldom been used to modify the H₂-release properties NaBH₄, and only moderate results were achieved when using VF₃, VF₄, or V₃B as catalysts. This is quite disappointing in view of the fact that, from 40 to 100 °C, the γ phase of vanadium(II) hydride is known to exist in equilibrium with hydrogen gas and its metallic β form [56]. This implies a rather low temperature of H₂ release from VH₂, especially in an H₂-free atmosphere. Moreover, the process of H₂ release from VH₂ must have a rather low activation barrier, as it tends to be fast. Consequently, vanadium metal or its compounds (reduced in situ to VNPs) could act as catalysts and destabilize [BH₄]⁻ anions present in NaBH₄, thus decreasing the temperature of thermal decomposition of this material and leading to facile H₂ evolution at low temperature. Therefore, in this study, we researched doping of sodium borohydride with vanadium, vanadium oxides (V₂O₃, VO₂, V₂O₅), or vanadium nanoparticles and their effects on thermal stability of NaBH₄ hydrogen storage material.

2. Experimental

2.1. Materials

NaBH₄ (≥98%), LiAlH₄ 2.0 M solution in THF and catalysts V₂O₅ (≥99.6%), VO₂ (≥99%), and V₂O₃ (99.99%) were purchased from Sigma Aldrich (St. Louis, MO, USA). Ph₄PCl (98%) used in Ph₄PBH₄ synthesis was delivered by abcr GmbH (Karlsruhe, Germany). Solvents dichloromethane (DCM) (≥99.8% ≤50 ppm H₂O) and tetrahydrofuran (THF) (≥99.8%, ≤50 ppm H₂O) were purchased from ROTH (Karlsruhe, Germany), while o-difluorobenzene (o-DFB) (98%) was obtained from Fluorochem (Hadfield, UK). Before synthesis, DCM and THF were dried with P₂O₅ and distilled; o-DFB was degassed using a Schlenk line and LN₂. All solvents were then sealed in containers with 4 Å molecular sieves.

2.2. Equipment

A Vigor SG1200/750TS-F glovebox (Houston, TX, USA) was used for storing and preparing the samples; all reactions were performed in an inert gas (Ar) atmosphere (<10 ppm O₂, <0.05 ppm H₂O) or using a Schlenk line. Mechanochemical doping was conducted in a stainless-steel disc bowl using a Testchem vibration mill (1400 RPM). Suspensions of VNPs were firstly sonicated using Polsonic Sonic-6D in an ice-cooled water bath. Thermal decomposition analysis of the samples was investigated using a thermogravimeter (TGA) combined with differential scanning calorimeter (DSC) from Netzsch STA 409 PG (NETZSCH-Gerätebau GmbH, Selb, Germany) using Al₂O₃ crucibles. TGA/DSC chamber was constantly purged at a constant Ar (99.9999%) flow rate of 80 mL·min⁻¹. The evolved gases were analyzed with a quadrupole mass spectrometer (MS) QMS 403 C (Pfeiffer Vacuum, Aβlar, Germany), connected to the TGA/DSC device by a quartz capillary preheated to 200 °C to avoid condensation of low-boiling volatiles. Fourier-transform infrared spectroscopy (FTIR) of all solid products of doping was measured using a Vertex 80v FT-IR spectrometer (Bruker, Billerica, MA, USA). Anhydrous KBr was used as a pellet material.

2.3. Synthesis and Doping

Vanadium nanoparticles (VNPs) were synthesized via reduction of vanadium(III) compounds. The THF suspension was prepared using vanadium (III) chloride and LiAlH₄ solutions in THF. Then, 10 mL of a stoichiometric amount of reducer solution was added dropwise to 30 mL of 0.05 M solution of VCl₃ under intense stirring. In the case of VNPs suspended in o-DFB, acetonitrile-coordinated vanadium(III) bis(tetra(nonafluoro-tert-butoxy)aluminate) monochloride (VCl(Al(OC(CF₃)₃)₄)₂·5ACN) was synthesized according to [57] and used as source of vanadium and tetraphenylphosphonium borohydride (Ph₄PBH₄) as a reducer and dissolved in o-DFB. Next, 20 mL of 0.015 M reducer solution was added dropwise to 20 mL of 0.005 M solution of vanadium salt, under intense stirring. Excess of the reducer was used to ensure complete reduction of vanadium cation. Both suspensions were stirred for 24 h before use and denoted in dependence of solvent used as VNPs_THF and VNPs_O-DFB. Attempts of purification by centrifuging brought no results since VNPs could not be separated from their suspension that way and were used without further purification. The rationale behind the use of VNs is in their large specific area, which could imply fast and uniform catalysis of H₂ evolution within the entire bulk of the hydrogen store.

Doping was carried out by mixing 5 or 25 wt.% of powder catalysts (V, V₂O₃, VO₂, or V₂O₅) with NaBH₄ and milling in 100 g batches for 5 or 30 min in milling periods of 1 and 5 min respectively, altered with liquid nitrogen (LN₂) cooling to avoid overheating of samples. In addition, one V₂O₅ sample was also milled for 3 h in 20 min periods. The samples were then denoted using the Na:XY T pattern, where X stands for the catalyst used, Y is its wt.% amount, and T is the milling time; the samples were tested with TGA/DSC in 20–450 or 20–600 °C range with 5 K/min heating rate together with MS of evolved gases and compared with milled pure NaBH₄ as reference. For activation energy studies of Na:V₂O₅25%_180 min, three additional heating rates were used: 10, 1, and 0.5 K/min.

In the case of VNP doping (100 mg) of NaBH₄, 2 or 10 mL of VNPs_THF or VNPs_O-DFB, respectively, were added to the glass ampoule, which was firstly sonicated for 20 min in ice-cold bath, and then dried under a dynamic vacuum using a Schlenk line until the pressure dropped to 5 × 10⁻² mbar and for 30 min after. Samples were then transferred to a milling bowl, further milled for 30 min in the same way as powder-doped samples, and denoted Na:VNPs_THF and Na:VNPs_O-DFB, respectively.

3. Results

3.1. Synthesis of VNPs

TEM analysis of synthesized VNPs (Figure 1) showed that nanoparticles prepared in THF using LiAlH₄ are smaller and more uniform than those synthesized in o-DFB.

The particle radius distribution based on several TEM images (Figure 2) was prepared in order to estimate size and standard deviation of their size. As calculated, mean sizes of VNPs were 14.86 ± 7.71 nm and 3.57 ± 0.93 nm for o-DFB and THF suspensions, respectively. The VNPs_THF size histogram could be described with a normal distribution, being more uniform in size and smaller than VNPs_O-DFB which had various, aspherical shapes and a size histogram looking more like a lognormal distribution. Since LiAlH₄ is a much stronger reducer than Ph₄PBH₄, vanadium is reduced more quickly; therefore, nanoparticles grow to a more uniform, smaller size. Unstirred VNPs are stable and stay suspended without any sign of agglomeration for months, despite no surfactant being used.

3.2. Thermogravimetric and Spectroscopic Analysis of Doped Samples

To screen the behavior of diverse V dopants, samples with 5 wt.% of catalyst milled in five short periods of 1 min each were preliminarily studied. The TG–DSC profiles for these samples, as well as a reference sample of pristine NaBH₄ (milled in a similar way), together with the ion current for diverse mass peaks in the MS of the evolved gas, are shown jointly in Figure 3. The picked masses in MS correspond to hydrogen, water, and diborane fragmentation patterns.

There was a relatively minor effect of these short-milled dopants on the thermal decomposition of NaBH₄, with a noticeable effect seen only for VO₂ (inflection of the TG curve at 402 °C associated with mass loss of 0.6 wt.% up to 450 °C) and a more pronounced one for V₂O₅ (clear mass decrease above 398 °C, associated with the total mass loss of 1.0 wt.% up to 450 °C). At the same time, there was no visible decomposition in the milled pristine NaBH₄ sample. MS clearly indicated that these events corresponded to endothermic H₂ evolution but not associated with water or diborane.

It is worth mentioning that the small mass drift observable on a few TGA curves is a phenomenon resulting from apparatus imperfection and cannot always be avoided (despite hours-long weight stabilization). However, the sudden mass drop resulting from the decomposition reaction can be clearly distinguished; thus, it does not result in data misinterpretation in qualitative studies. Encouraged by this result, we further focused on V₂O₅-doped samples. Another three samples with increased content of vanadium pentoxide of 25 wt.% were prepared via 5, 30, and 180 min milling. As-obtained samples were inspected with FTIR, and the results were compared to those for pristine NaBH₄ and V₂O₅ (Figure 4).

No change in wavenumber at peak or in the relative intensities of diverse absorption bands characteristic for NaBH₄ (2400, 2295, 2223, and 1120 cm⁻¹) was observed upon doping with V₂O₅, independent of the milling time. However, bands at 1018, 838, 621, and 482 cm⁻¹ which are characteristic for V₂O₅ clearly decreased their intensity when the sample was milled for 30 instead of 5 min, while the band at 514 cm⁻¹ could be barely seen in the 5 min milled sample. Noticeable shifts in the 838 and 621 cm⁻¹ peaks could be observed as a function of milling time. When milled for 3 h, these bands disappeared almost completely, and a few new bands could be described: a very weak band at 732 cm⁻¹, two weak bands at 882 and 816 cm⁻¹, a moderate band at 602 cm⁻¹, and a very strong band at 436 cm⁻¹ (off the scale), together with bands characteristic for water in the range of 3000–3600 cm⁻¹. This result indicates that V₂O₅ slowly reacted with milled borohydride and underwent a progressive reduction, associated with the formation of unknown species. These bands did not originate from any known compound of vanadium, as revealed by our scrutiny. Simultaneously, X-ray diffraction patterns for the 3 h milled sample did not show the presence of any new crystalline moiety. In each case, however, there remained a substantial amount of bulk NaBH₄ in these samples as evidenced by stable B–H stretching and H–B–H bending regions, as also evidenced in X-ray diffraction patterns (not shown).

All three samples were studied with TGA–DSC–MS, and the results were compared to those for Na:V₂O₅5%_5 min obtained with a very short milling time (5 min) and low catalyst load of 5 wt.% (Figure 5).

There was a clear effect of both catalyst load and milling time on the thermal decomposition of NaBH₄/V₂O₅ conglomerates, as seen in TG–DSC–MS profiles. The weak effect seen for the Na:V₂O₅5%_5 min sample at 398 °C was enhanced and shifted to lower temperature of 390 °C for the sample Na:V₂O₅25%_5 min and further shifted to 383 °C for the sample Na:V₂O₅25%_30 min. The lowest onset of thermal decomposition was seen for the 3 h milled sample (369 °C). All these events corresponded to the sharp and complex endothermic peaks of hydrogen desorption and were adjoined by sharp H₂ release in the temperature range of 380 to 410 °C as seen in the MS. Hydrogen was free from major contamination, as additionally evidenced by the MS.

The total mass loss between the onset of thermal decomposition and 450 °C was 4.0%, 4.1%, and 5.4% for 25 wt.% doped samples milled for 5, 30, and 180 min, respectively (Figure 6). The latter value corresponds to ca. 70% of total hydrogen stored by the sample. Importantly, as indicated by MS, hydrogen was not contaminated by common impurities such as B₂H₆, other boron hydrides, and water.

The onset of thermal decomposition for doped samples (368–388 °C) was quite low as compared to the temperature of thermal decomposition for pristine NaBH₄ and even to that for the best catalyzed samples described in the literature. Actually, only TiF₃ was proven to provide an even lower decomposition onset than catalysts studied here (300 °C) [15].

The Kissinger method [58,59] was used to determine the activation energy of decomposition for NaV₂O₅25%_180 min. Three additional experiments were conducted using heating rates of 0.5, 1, and 10 K/min in addition to the default 5 K/min used before (Figure 7). The clearly observable intermediate decomposition step for 10 and 5 K/min heating rates could not be detected at lower heating rates due to its size compared to neighboring ones; thus, only the first and last (strong) peaks were taken into activation energy calculations (Figure 8).

The activation energy of the first decomposition step for doped sample decreased drastically to 65.9 ± 3 kJ/mol (peak) or 69.0 ± 2 kJ/mol (onset) in comparison with 275 kJ/mol for pristine NaBH₄ [21]. This implies a fourfold reduction of the activation energy; our result may also be compared to that of ZrCl₄-doped NaBH₄, which is characterized by an E_a of 180 kJ/mol [21]. On the other hand, the E_a value for the last decomposition step varied when calculated for onset or peak value (120.3 ± 1 and 169 ± 18 kJ/mol respectively); nevertheless, it was still remarkably smaller than the literature value for NaBH₄. It is worth noticing, however, that the first decomposition step, characterized by very small activation energy, corresponded to the vast majority of hydrogen desorbed. Interestingly, the activation energies obtained for undoped but long-milled NaBH₄ were as follows: 143.0 ± 1.6 kJ/mol (first peak) and 70.0 ± 0.9 kJ/mol (second peak). Therefore, the reduction in the activation barrier at the first stage of decomposition observed in V₂O₅-catalyzed samples was partially due to nano-amorphization of NaBH₄ and partially due to the presence of a catalyst. On the other hand, the processes taking place at the second stage of decomposition were supposedly entirely different for the V₂O₅-free and V₂O₅-rich samples, as indicated by the counterintuitive behavior of the activation energies at that stage.

The fact that two main peaks were observed in the thermal profile means that two different species were decomposing at each of these steps. We attempted to gain insight into the nature of intermediate decomposing species and reaction products by performing additional experiments. The initial surmise was that the last step of decomposition could originate from the part of NaBH₄ not in contact with catalyst, but where H₂ release was easier than for pristine NaBH₄ due to pronounced milling and non-amorphization of the sample. Indeed, as already mentioned, milled NaBH₄ showed two major distinct decomposition steps, similarly to the catalyzed samples. However, none of those occurred at temperatures as low as 335–425 °C since the onset of thermal decomposition for milled NaBH₄ was as high as 461 °C (not shown) and extended to over 600 °C (in agreement with literature data). Additional X-ray diffraction experiments for samples heated up to diverse temperatures failed to observe any crystalline species beyond traces of the bulk NaBH₄. Moreover, while both thermal peaks coincided with H₂ evolution, the evolved gas analysis failed to detect any impurity species such as B₂H₆ and higher boron hydrides. Therefore, at the current stage, the genuine nature of a (probably in situ formed) catalyst and that of the hydride intermediate undergoing the second stage of decomposition must remain an open question. We notice that the unequivocal detection of the catalyst nature is often nontrivial, as documented by many cases including that of Ti species-catalyzed NaAlH₄, which stands today as one of reference materials in hydrogen storage. One cannot exclude even a scenario that the presence of V₂O₅ modifies the path of the decomposition reaction so substantially that sodium is no longer a volatile product of decomposition but remains in the solid residue (e.g., as some sort of vanadate salt). All these issues go beyond the scope of the current preliminary work and remain to be thoroughly studied in the future.

Having achieved a marked reduction in the thermal decomposition temperature of NaBH₄ by the high load of V₂O₅ and a prolonged milling time, we applied the same procedure to the remaining vanadium oxides. The TG–DSC–MS profiles for all samples are jointly shown in Figure 9.

A substantial effect (and comparable to that for V₂O₅) on TG and DSC profiles can only be seen for VO₂; here, the onset of thermal decomposition was around 377 °C, but the amount of hydrogen evolved up to 450 °C was smaller than that for the V₂O₅ additive. The effects seen for V₂O₃ and V were the least pronounced.

It is important to note that the total mass loss up to 600 °C in almost all samples exceeded the total H content of the samples. This is because sodium vapor is partly released at high temperatures [17,50,60]. It is uncertain at this time what amount of sodium remained in the doped samples, as well as the chemical identity of the phase(s) containing this metal.

Having studied vanadium- and vanadium oxide-doped samples, our attention turned to vanadium nanoparticles. Samples doped with vanadium nanoparticles were denoted Na:VNPs_THF and Na:VNPs_{O- DFB}, and they were studied with TGA–DSC–MS analogously to previous samples (Figure 10). Reference curves for milled NaBH₄ were added for comparison.

Both VNP-doped samples seemingly released H₂ at a lower temperature than the reference sample, as can be judged from the MS profiles. However, the doped samples also released copious amounts of CO₂, B₂H₆, and solvent residues (only in the case of Na:VNPs_THF) at even lower temperatures, as seen in the MS and TGA profiles. These contaminations most probably came from byproducts of VNP synthesis and are certainly highly undesired for any practical application.

4. Conclusions

We studied the effect of vanadium and its oxides and nanoparticles (VNPs) on the thermal decomposition of NaBH₄. It was revealed that 3 h milling of NaBH₄ with 25 wt.% of V₂O₅ decreased the temperature of NaBH₄ decomposition by as much as 165 °C with respect to pristine sodium borohydride. VO₂ doping also decreased the temperature of NaBH₄ pyrolysis, but to a lesser degree; moreover, the hydrogen release peak was much broader than for V₂O₅ doping. The hydrogen released up to 450 °C was free from B₂H₆, other boron hydrides, and water. In contrast to expectations, vanadium metal provided nearly no improvement in the hydrogen desorption temperature of sodium borohydride. Doping with VNPs resulted in contamination of the evolved H₂ with some precursors used for their synthesis, which is a highly undesirable effect.

Extension of these studies may focus on the chemical nature of vanadium in the extensively milled and thermally decomposed samples, attempts to achieve at least partial reversibility of H₂ storage, understanding the nature of the hydride intermediate and of the fate of sodium in the catalyzed samples, and broadening of the study toward other light metal borohydrides.