Hydrolysis of the Borohydride Anion BH4−: A 11B NMR Study Showing the Formation of Short-Living Reaction Intermediates including BH3OH−

In hydrolysis and electro-oxidation of the borohydride anion BH4−, key reactions in the field of energy, one critical short-living intermediate is BH3OH−. When water was used as both solvent and reactant, only BH3OH− is detected by 11B NMR. By moving away from such conditions and using DMF as solvent and water as reactant in excess, four 11B NMR quartets were observed. These signals were due to BH3-based intermediates as suggested by theoretical calculations; they were DMF·BH3, BH3OH−, and B2H7− (i.e., [H3B−H−BH3]− or [H4B−BH3]−). Our results shed light on the importance of BH3 stemming from BH4− and on its capacity as Lewis acid to interact with Lewis bases such as DMF, OH−, and BH4−. These findings are important for a better understanding at the molecular level of hydrolysis of BH4− and production of impurities in boranes synthesis.


Introduction
Sodium borohydride NaBH 4 in alkaline aqueous solution is a potential fuel of lowtemperature fuel cell [1]. It is regarded as an indirect fuel (i.e., H carrier) when it is hydrolyzed to produce H 2 , the as-produced H 2 then feeding a fuel cell (Equation (1)) [2]. It is regarded as a direct fuel (i.e., reductant) when it directly feeds a direct liquid fuel cell to be electro-oxidized (Equation (2) The aqueous solution has to be alkaline, that is, stabilized [4], because this is the only way to prevent spontaneous (exothermic, with an enthalpy of about −240 kJ mol −1 [5]) hydrolysis of BH 4 − from occurring extensively. In hydrolysis (Equation (1)), a metal catalyst is therefore required to accelerate the production of H 2 [6]. In electro-oxidation (Equation (2)), a metal electro-catalyst is required to promote the generation of a maximum of electrons (out of eight) [4]. However, the electro-catalyst also acts as catalyst of hydrolysis, a reaction that is in this case regarded as heterogeneous because it is detrimental to the fuel cell faradaic efficiency [7].

OH − , and
Molecules 2022, 27, 1975 2 of 11 [H] + [BHOH] − as possible short-living intermediates [9]. More recently, Guella et al. reported that, by 11 B nuclear magnetic resonance (NMR) spectroscopy, they detected only BH 4 − and B(OH) 4 − (Equation (1)) for a Pd-catalyzed hydrolysis [10]. The non-detection of other species was explained by the fact that the hydrolysis intermediates are excessively short-living in their experimental conditions. By quantum chemical calculations, Lu et al. [11] confirmed Guella et al.'s explanation and modelled a multistep process involving the following hypothetical short-living intermediates (Equation (3)): Comparable predictions were reported by Zhou et al. [12], Andrieux et al. [13], Churikov et al. [14], and Choi et al. [15] detected traces of BH 3 OH − by using 11 B NMR spectroscopy. It is therefore arguable whether BH 3 OH − , as the first short-living intermediate, directly hydrolyzes into B(OH) 4 − . This is a possible parallel pathway as suggested by Mochalov et al. [8] for example.
Similarly, Nanayakkara et al. [25] investigated the mechanism of H 2 release of BH 3 in water and the following solvent effects by using MP2 quantum calculations. One H 2 O molecule interacting with BH 3 led to an activation energy equal to 24.9 kcal mol −1 , while the energy values ranged from 29 and 32 kcal mol −1 when one H 2 O molecule interacted with BH 3 and another H 2 O molecule interacted with the H 2 O molecule bonded to BH 3 . The resulting enthalpy was estimated at 20 kcal mol −1 for the first configuration and ranged between 12 and 14 kcal mol −1 for the others.
The present study is to be seen against the background described above. Based on a systematic study using 11 B NMR spectroscopy, we attempted to detect and identify any short-living intermediates in order to gain insight and better understanding of both hydrolysis and electro-oxidation of BH 4 − . Furthermore, theoretical investigations were performed for obtaining vibrational results, determining the sensitive frequencies and estimating the energies of the different hypothetical molecular structures.

Hydrolysis Conditions Where H 2 O Acts as Both Reactant and Solvent
In hydrolysis and electro-oxidation conditions, the fuel is an alkaline aqueous solution of BH 4 − for which the concentration of BH 4 − is usually kept low (typically < 1 M). We therefore set our experimental conditions to be in line with such practices: the concentration of NaOH was fixed as 0.1 M and the concentration of BH 4 − (from NaBH 4 ) was chosen as 0.66 M.
In hydrolysis and electro-oxidation conditions, the reaction is catalyzed by a metal catalyst and an electro-catalyst, respectively. We selected three bulk metals such as Pd, Pt, and Au (each as a piece of metal wire). They were selected because each has been used in hydrolysis [26] and electro-oxidation [22].
In the present study and unlike in common practices [26], our objective was not to develop an active (or very active) hydrolysis catalyst. Our objective was to work with a lowly active catalyst so that the kinetics of H 2 production remains slow when analyzing the solutions by 11 B NMR spectroscopy. We thus focused on metals in bulk state, which is a state that offers the desired catalytic activity. We ensured this by performing a series of hydrolysis experiments. Typically, 2 mL of the aforementioned alkaline solution of BH 4 − Molecules 2022, 27,1975 3 of 11 (corresponding to 50 mg of NaBH 4 ) were put into contact with 16 mg of Pd, 14.5 mg of Pt, or 14.3 mg of Au at 30 • C. Regardless of the nature of the metal, it took 2 h to produce <1.6 mol H 2 per mol BH 4 − (Figure 1), that is, <53 mL H 2 (out of 132 mL for a conversion of 100%). This means a H 2 generation rate of <0.45 mL(H 2 ) min −1 that is in agreement with our need. We also ensured that, in the absence of any metal, the alkaline solution of BH 4 − was quite stable. At 30 • C, <0.1 mol H 2 per mol BH 4 − was produced in 2 h (namely, <3 mL(H 2 )). lowly active catalyst so that the kinetics of H2 production remains slow when analyzing the solutions by 11 B NMR spectroscopy. We thus focused on metals in bulk state, which is a state that offers the desired catalytic activity. We ensured this by performing a series of hydrolysis experiments. Typically, 2 mL of the aforementioned alkaline solution of BH4 − (corresponding to 50 mg of NaBH4) were put into contact with 16 mg of Pd, 14.5 mg of Pt, or 14.3 mg of Au at 30 °C. Regardless of the nature of the metal, it took 2 h to produce <1.6 mol H2 per mol BH4 − (Figure 1), that is, <53 mL H2 (out of 132 mL for a conversion of 100%). This means a H2 generation rate of <0.45 mL(H2) min −1 that is in agreement with our need. We also ensured that, in the absence of any metal, the alkaline solution of BH4 − was quite stable. At 30 °C, <0.1 mol H2 per mol BH4 − was produced in 2 h (namely, <3 mL(H2)). The hydrolysis tests were repeated to analyze the solution by 11 B NMR spectroscopy every hour. Similar to a previous study [13], we detected only three signals (examples of spectra in Figure 2; Table 1). The first main signal was a quintet at δ −41.5 ppm due to BH4 − . The second main signal was a singlet at δ +1.9 ppm evidencing the formation of B(OH)4 − (Equation (1)). There was an additional minor and almost negligible signal, a quartet at δ −12.8 ppm. It was ascribed to the short-living intermediate BH3OH − [10,15,27]. The hydrolysis tests were repeated to analyze the solution by 11 B NMR spectroscopy every hour. Similar to a previous study [13], we detected only three signals (examples of spectra in Figure 2; Table 1). The first main signal was a quintet at δ −41.5 ppm due to BH 4 − . The second main signal was a singlet at δ +1.9 ppm evidencing the formation of B(OH) 4 − (Equation (1)). There was an additional minor and almost negligible signal, a quartet at δ −12.8 ppm. It was ascribed to the short-living intermediate BH 3  lowly active catalyst so that the kinetics of H2 production remains slow when analyzing the solutions by 11 B NMR spectroscopy. We thus focused on metals in bulk state, which is a state that offers the desired catalytic activity. We ensured this by performing a series of hydrolysis experiments. Typically, 2 mL of the aforementioned alkaline solution of BH4 − (corresponding to 50 mg of NaBH4) were put into contact with 16 mg of Pd, 14.5 mg of Pt, or 14.3 mg of Au at 30 °C. Regardless of the nature of the metal, it took 2 h to produce <1.6 mol H2 per mol BH4 − (Figure 1), that is, <53 mL H2 (out of 132 mL for a conversion of 100%). This means a H2 generation rate of <0.45 mL(H2) min −1 that is in agreement with our need. We also ensured that, in the absence of any metal, the alkaline solution of BH4 − was quite stable. At 30 °C, <0.1 mol H2 per mol BH4 − was produced in 2 h (namely, <3 mL(H2)). The hydrolysis tests were repeated to analyze the solution by 11 B NMR spectroscopy every hour. Similar to a previous study [13], we detected only three signals (examples of spectra in Figure 2; Table 1). The first main signal was a quintet at δ −41.5 ppm due to BH4 − . The second main signal was a singlet at δ +1.9 ppm evidencing the formation of B(OH)4 − (Equation (1)). There was an additional minor and almost negligible signal, a quartet at δ −12.8 ppm. It was ascribed to the short-living intermediate BH3OH − [10,15,27].
No additional 11 B NMR signals that would be attributed to other short-living intermediates were seen. This might be explained by concentrations that are below the detection limit (ca. 1 × 10 −3 mol L −1 ) of the spectrometer. This might be also explained by low symmetry of the intermediates' structures, which would lead to broad signals of very low intensity and thus indistinguishable from the base line. It is worth mentioning that we used Gaussian 09 software to perform geometry optimization and NMR calculations for a series of possible intermediates including BH 2 (OH) 2 − and BH(OH) 3 − . We found that the signals of BH 2 (OH) 2 − and BH(OH) 3 − should be a triplet and a doublet appearing between δ −7 and δ 0 ppm, respectively.
Another possible explanation of the absence of additional 11 B NMR signals is that the experimental conditions were not suitable for detecting intermediates with a lifetime that is shorter than that of the detected BH 3 OH − . Based on our observations, we can state that the lifetime scale of BH 3 OH − is of tens of seconds, whereas it might be much shorter (e.g., microseconds scale) for the other intermediates. Yet, the hydrolysis tests described above were performed in the presence of an excess of water: we used 2 mL (mol ratio H 2 O/BH 4 − of 84) whereas about 0.1 mL (mol ratio H 2 O/BH 4 − of 4) would be enough to totally hydrolyze BH 4 − . Water acted as both reactant and solvent, and the excess of water could be a favorable context to promote extremely fast hydrolysis of short-living intermediates.

Hydrolysis Conditions Where H 2 O Is Only a Reactant
In order to move away from the conditions using water as both reactant and solvent, we drew on two ancient reports dealing with hydrolysis of BH 4 − . Modler (1)). The as-prepared solutions were analyzed by 11 B NMR spectroscopy. It is worth mentioning that in such conditions, the hydrolysis was expected to be slow. Accordingly, the solutions were analyzed every 24 h for 3 days.
The 11 B NMR spectra focusing on the δ range varying from +20 to −50 ppm ( Figure S1) showed only the quintet at δ −39.7 ppm due to BH 4 − . By zooming over the δ range varying from +20 to −30 ppm ( Figure S2), it was possible to distinguish an additional signal of very small intensity at δ −14.1 ppm, namely the quartet due to BH 3 OH − . The quartet could be seen after 24 h for the Pd-, Pt-, and Au-catalyzed solutions, and after 48 h for the uncatalyzed solution. These results highlighted that, in the stoichiometric conditions, the hydrolysis took place to a negligible extent. Another observation is that, even in the absence of a metal, hydrolysis spontaneously took place. The non-detection of B(OH) 4 − may have  4 − were practically insoluble in DMF [30] and may have precipitated; and/or, the concentration of B(OH) 4 − was below the detection limit.

Hydrolysis Conditions Where H 2 O Is a Reactant in Excess
We therefore repeated the experiments while increasing the water content: the mol ratio H 2 O/BH 4 − passed from 4 to 32. Once more, we prepared four 10 mL DMF solutions of BH 4 − (1.32 M) and added 7.6 mL of alkaline (0.1 M NaOH) aqueous solution. In comparison to the experiments presented in Section 2.1, the present series used water to a lesser extent (i.e., mol ratio H 2 O/BH 4 − of 32 versus 84) and the 32 equivalents of H 2 O were dispersed in 10 mL of DMF, mitigating the hydrolysis of BH 4 − . As before, the 11 B NMR spectra ( Figure S3) mainly showed the quintet at δ −40.5 ppm due to BH 4 − , and B(OH) 4 − was not observed because of the reasons listed at the end of the previous section. In contrast to the results discussed above, the 11 B NMR spectra showed additional signals at δ < 0 ( Figure 3). This is discussed hereafter. hydrolysis took place to a negligible extent. Another observation is that, even in the absence of a metal, hydrolysis spontaneously took place. The non-detection of B(OH)4 − may have up to three explanations: the amount of H2O was too low and the H2O molecules were very diluted in DMF, which hindered interaction-reaction with BH4 − and BH3OH − ; borates including B(OH)4 − were practically insoluble in DMF [30] and may have precipitated; and/or, the concentration of B(OH)4 − was below the detection limit.

Hydrolysis Conditions where H2O Is a Reactant in Excess
We therefore repeated the experiments while increasing the water content: the mol ratio H2O/BH4 − passed from 4 to 32. Once more, we prepared four 10 mL DMF solutions of BH4 − (1.32 M) and added 7.6 mL of alkaline (0.1 M NaOH) aqueous solution. In comparison to the experiments presented in Section 2.1, the present series used water to a lesser extent (i.e., mol ratio H2O/BH4 − of 32 versus 84) and the 32 equivalents of H2O were dispersed in 10 mL of DMF, mitigating the hydrolysis of BH4 − .
As before, the 11 B NMR spectra ( Figure S3) mainly showed the quintet at δ −40.5 ppm due to BH4 − , and B(OH)4 − was not observed because of the reasons listed at the end of the previous section. In contrast to the results discussed above, the 11 B NMR spectra showed additional signals at δ < 0 ( Figure 3). This is discussed hereafter. The first of the additional signals was a quartet at δ −14.4 ppm. As for our experiments discussed above, it was ascribed to BH3OH − . The first of the additional signals was a quartet at δ −14.4 ppm. As for our experiments discussed above, it was ascribed to BH 3 OH − .
The second of the additional signals was also a quartet, centered at δ −8.9 ppm. It indicated the formation of another BH 3 -containing intermediate. The third of the additional signals appeared as a multiplet located between δ −18.5 ppm and δ −23.5 ppm. With the help of 1 H-decoupled 11 B NMR spectroscopy, we shed light on its nature. It was the result of two distinct signals peaking at δ −20.3 ppm and δ −21.8 ppm (Figure 4). By deconvolution of the signal, we found that the two signals were more likely to be two overlapping quartets, thereby indicating the formation of two other BH 3 intermediates ( Figure S4 and Table S1).
The second of the additional signals was also a quartet, centered at δ −8.9 ppm. It indicated the formation of another BH3-containing intermediate.
The third of the additional signals appeared as a multiplet located between δ −18.5 ppm and δ −23.5 ppm. With the help of 1 H-decoupled 11 B NMR spectroscopy, we shed light on its nature. It was the result of two distinct signals peaking at δ −20.3 ppm and δ −21.8 ppm (Figure 4). By deconvolution of the signal, we found that the two signals were more likely to be two overlapping quartets, thereby indicating the formation of two other BH3 intermediates ( Figure S4 and Table S1).
To summarize the above: the hydrolysis of DMF-solubilized BH4 − in the presence of 32 equivalents of H2O involved more intermediates than the only short-living intermediate BH3OH − . There were three additional intermediates and they all showed a quartet in 11 B NMR spectroscopy, indicating that they all were made up of the BH3 group. We therefore focused our efforts on attributing the aforementioned quartets to possible BH3 intermediates. We thought about any species likely to form in our conditions while exploring the open literature [31][32][33][34]. The following ones were listed ( Figure 5):

•
The complex H2O·BH3 because H2O is a Lewis base able to complex the Lewis acid BH3; • The complex DMF·BH3 because DMF is Lewis bases able to complex BH3; The pentacoordinate BH3(H2).
According to Tague and Andrews [34], the last species BH3(H2) possibly acts as intermediate before the formation of BH3OH − by reaction of BH4 − and H2O. To summarize the above: the hydrolysis of DMF-solubilized BH 4 − in the presence of 32 equivalents of H 2 O involved more intermediates than the only short-living intermediate BH 3 OH − . There were three additional intermediates and they all showed a quartet in 11 B NMR spectroscopy, indicating that they all were made up of the BH 3 group.
We therefore focused our efforts on attributing the aforementioned quartets to possible BH 3 intermediates. We thought about any species likely to form in our conditions while exploring the open literature [31][32][33][34]. The following ones were listed ( Figure 5):

•
The According to Tague and Andrews [34], the last species BH 3 (H 2 ) possibly acts as intermediate before the formation of BH 3 OH − by reaction of BH 4 − and H 2 O. We then used Gaussian 09 software to perform geometry optimization and NMR calculations for each of these possible intermediates. We found the chemical shifts listed in Table 2. As observed in this table, a relatively good agreement between CASTEP and Gaussian 09 results was obtained considering the two investigated functionals (B3LYP for Gaussian 09 and PBE for CASTEP), except for BH 3 (H 2 ). In the case of this species, the impact of the dispersion could be invoked but additional calculations using DFT-D in CASTEP showed a very small influence of dispersion on the calculations. It is worth mentioning that in a previous study [31], the chemical shift of B 2 H 7 − in THF as solvent was reported to be δ −26 ppm. Similarly, using CASTEP calculations, we found comparable values ( Table 2). We also calculated the chemical shifts for the intermediates based on We then used Gaussian 09 software to perform geometry optimization and NMR calculations for each of these possible intermediates. We found the chemical shifts listed in Table 2. As observed in this table, a relatively good agreement between CASTEP and Gaussian 09 results was obtained considering the two investigated functionals (B3LYP for Gaussian 09 and PBE for CASTEP), except for BH3(H2). In the case of this species, the impact of the dispersion could be invoked but additional calculations using DFT-D in CASTEP showed a very small influence of dispersion on the calculations. It is worth mentioning that in a previous study [31], the chemical shift of B2H7 − in THF as solvent was reported to be δ −26 ppm. Similarly, using CASTEP calculations, we found comparable values ( Table 2). We also calculated the chemical shifts for the intermediates based on BH4x(OH)x − (with x = 1, 2, 3, 4), such as: BH4 − with δ −51.5 ppm; BH3OH − with δ −11.4 ppm; BH2(OH)2 − with δ +0.1 ppm; BH(OH)3 − with δ +1.1 ppm; and B(OH)4 − with δ +3.1 ppm. Going back to the results presented in Figure 3 and using the data in Table 2, we ascribed the quartets at follows. The signals at δ −8.9 ppm and δ −14.4 ppm (Figure 3) were unambiguously attributed to DMF·BH3 and BH3OH − . Because the calculated chemical shift of BH3(H2) is much different from that of remaining signals at around δ −21 ppm, we discarded its formation. We also discarded the formation of H2O·BH3 due to the absence of signals at around 0 ppm in our experimental conditions. Accordingly, the partly overlapping quartets are at δ −20.3 ppm and δ −21.8 ppm and are attributed to B2H7 − and B2H7 − in interaction with H2O. Indeed, the chemical shift for the quartet due to [B2H7·H2O] − was calculated as −28.1 ppm using Gaussian 09 and −32.6 ppm using CASTEP; these shifts were close to those calculated for B2H7 − (Table 2).
Based on the experimental results reported above and supported by the calculations performed, we suggest that the BH4 − anions dissolved in DMF are able to react with H2O  Going back to the results presented in Figure 3 and using the data in Table 2, we ascribed the quartets at follows. The signals at δ −8.9 ppm and δ −14.4 ppm (Figure 3) were unambiguously attributed to DMF·BH 3 and BH 3 OH − . Because the calculated chemical shift of BH 3 (H 2 ) is much different from that of remaining signals at around δ −21 ppm, we discarded its formation. We also discarded the formation of H 2 O·BH 3 due to the absence of signals at around 0 ppm in our experimental conditions. Accordingly, the partly overlapping quartets are at δ −20.3 ppm and δ −21.8 ppm and are attributed to B 2 H 7 − and B 2 H 7 − in interaction with H 2 O. Indeed, the chemical shift for the quartet due to [B 2 H 7 ·H 2 O] − was calculated as −28.1 ppm using Gaussian 09 and −32.6 ppm using CASTEP; these shifts were close to those calculated for B 2 H 7 − (Table 2). Based on the experimental results reported above and supported by the calculations performed, we suggest that the BH 4 − anions dissolved in DMF are able to react with H 2 O taken in excess to form BH 3 -based intermediates such as DMF·BH 3 , BH 3 OH − , and B 2 H 7 − . These intermediates are much likely to be in equilibrium. Based on the discussions reported in [30], we thus suggest that in DMF, BH 3 OH − forms first and DMF·BH 3 and B 2 H 7 − forms from BH 3 OH − (by substitution of Lewis bases). This is illustrated in Figure 6. taken in excess to form BH3-based intermediates such as DMF·BH3, BH3OH − , and B2H7 − . These intermediates are much likely to be in equilibrium. Based on the discussions reported in [30], we thus suggest that in DMF, BH3OH − forms first and DMF·BH3 and B2H7 − forms from BH3OH − (by substitution of Lewis bases). This is illustrated in Figure 6.
In a first step, the hydrolysis conditions were such that water acted as both reactant and solvent. The H2 evolution experiments were performed as follows. Under argon, 50 mg of NaBH4 were transferred in a Schlenk tube (used as hydrolysis reactor). For the catalyzed experiments, a piece of metal wire (16.1 mg of Pd, 14.5 mg of Pt, or 14.3 mg of Au) was also transferred in the tube. The tube was sealed and the glove box was taken out, installed to our hydrolysis set-up (reactor connected to an inverted burette via a cold trap kept at 0 °C), and immersed in an oil bath at 30 °C. The hydrolysis reaction was started by injecting 2 mL of an aqueous alkaline (0.1 M NaOH) solution. In these conditions, the mol ratio H2O/BH4 − was 84. The displacement of the blue-colored liquid in the inverted burette due to the generated H2 was video monitored. The H2 evolution experiments were repeated to analyze the solution by 11 B NMR spectroscopy (Bruker Avance 400 NMR spectrometer equipped with a BBOF probe; BF3·OEt2 as reference; acetonitrile-d3 such as ≥99.8 atom % D and from Sigma-Aldrich).
In a second step, the hydrolysis conditions were modified such that water only acted as reactant. To do so, 10 mL of DMF was used as solvent of 50 mg of NaBH4. To this solution prepared under argon, a piece of metal was added to catalyze the reaction. The hydrolysis reaction was started by injecting 0.95 mL of alkaline (0.1 M NaOH) aqueous solution. The concentration of H2O in DMF was 5.291 M and the mol ratio H2O/BH4 − was about 4. The solutions were analyzed by 11 B NMR spectroscopy every 24 h for 3 days.
In a third step, the hydrolysis conditions were once again modified. They were such that the water amount in DMF was increased and the mol ratio H2O/BH4 − passed from 4 to 32. Otherwise, the solutions were prepared similarly and they were analyzed by 11 B NMR spectroscopy every 24 h for 3 days.
alyzed by Pd, Pt, or Au after 0, 24, 48, and 72 h. These spectra focus on the range between δ +20 ppm and δ −30 ppm; Figure S3: 11 B NMR spectra of the 10 mL DMF solutions of BH 4 − (1.32 M) hydrolyzed by 7.6 mL of alkaline (0.1 M NaOH), uncatalyzed or catalyzed by Pd, Pt, or Au after 0, 24, 48, and 72 h. These spectra focus on the range between δ +20 ppm and δ −50 ppm; Figure S4: Deconvolution of the multiplet located between δ −18.5 ppm and δ −23.5 ppm for the 11 Table S1: Results of the deconvolution made for the signal shown in Figure S4. The chemical shifts, Pascal's triangles, and convergences are shown; the mol files of the structures are presented in Figure 5.

Conflicts of Interest:
The authors declare no conflict of interest.