E ﬃ cient Synthesis of Alkali Borohydrides from Mechanochemical Reduction of Borates Using Magnesium–Aluminum-Based Waste

: Lithium borohydride (LiBH 4 ) and sodium borohydride (NaBH 4 ) were synthesized via mechanical milling of LiBO 2 , and NaBO 2 with Mg–Al-based waste under controlled gaseous atmosphere conditions. Following this approach, the results herein presented indicate that LiBH 4 and NaBH 4 can be formed with a high conversion yield starting from the anhydrous borates under 70 bar H 2 . Interestingly, NaBH 4 can also be obtained with a high conversion yield by milling NaBO 2 · 4H 2 O and Mg–Al-based waste under an argon atmosphere. Under optimized molar ratios of the starting materials and milling parameters, NaBH 4 and LiBH 4 were obtained with conversion ratios higher than 99.5%. Based on the collected experimental results, the inﬂuence of the milling energy and the correlation with the ﬁnal yields were also discussed.


Introduction
Tetrahydroborates, discovered in the 1940s, have been attracting the attention of the scientific community in the last decades as possible energy vectors. Although tetrahydroborates, such as LiBH 4 and NaBH 4 , are commonly used as reducing agents in organic and inorganic chemistry [1][2][3][4], their employment as potential hydrogen storage materials have also been investigated, due to their high gravimetric hydrogen densities. LiBH 4 and NaBH 4 feature very attractive gravimetric and volumetric hydrogen storage capacities, i.e., of 18.5 wt.% H 2 and 10.8 wt.% H 2 , and 113.1 kg H 2 /m 3 and 121 kg H 2 /m 3 , respectively [2][3][4]. In particular, their use as single hydrides [5][6][7][8][9][10][11][12] and their combination in the so-called reactive hydride composites (RHCs) approach were intensively studied [13][14][15][16][17][18][19][20][21][22][23][24][25][26][27][28]. These tetrahydroborates were first synthesized by Schlesinger and Brown by reacting diborane with ethyl lithium to form LiBH 4 [29] and with sodium trimethoxyborohydride to form NaBH 4 [30]. Since then, several studies attempted to improve the synthesis of these two tetrahydroborates [31][32][33][34][35][36][37][38][39]. Nowadays, Li B(OCH 3 ) 3 + 4 NaH → NaBH 4 + 3 NaOCH 3, In the literature, several works on the conversion of borates of Na and Li using high-purity MgH 2 are reported. Li et al. [32] investigated the possibility of synthesizing NaBH 4 by ball milling-dehydrated Na 2 B 4 O 7 with MgH 2 in the presence of Na-based compounds (e.g., NaOH, Na 2 CO 3 , and Na 2 O 2 ). Kojima and Haga. [34] published that a reaction yield equal to 98 % of NaBH 4 can be obtained when annealing a mixture of NaBO 2 with Mg 2 Si under 7 MPa of hydrogen pressure at 550 • C for 2 to 4 h. Kong et al. [33], Hsueh et al. [36] and Çakanyıldırım et al. [44] independently investigated the possibility of forming NaBH 4 from a mixture of MgH 2 and NaBO 2 ball milled in argon atmosphere. In their works, Kong et al., Hsueh et al., and Çakanyıldırım et al. achieved an NaBH 4 yield of above 70%. Similarly, Bilen et al. [37] achieved 90% LiBH 4 purity from the reaction of LiBO 2 with MgH 2 by ball milling. In addition, ball milling is known to be an extremely powerful and versatile technique for the treatment of waste materials [45][46][47][48][49]. These works clearly show that ball milling is a suitable method to produce LiBH 4 and NaBH 4 from mixtures of LiBO 2 or NaBO 2 and MgH 2 , respectively. However, in view of possible large-scale production of these borohydrides, due to the production costs associated with the use of high-purity MgH 2 , the above-mentioned processes are not economically feasible. In order to tackle the production cost issue, the possibility of replacing MgH 2 with cost-neutral wastes, such as Mg-Al-based alloys, was pursued in this work. The prepared samples were characterized by XRD, FT-IR, and MAS-NMR techniques. The results of this investigation are presented and thoroughly discussed in the following sections.

Materials and Methods
Lithium metaborate (LiBO 2 , anhydrous ≥ 98% purity, Sigma Aldrich) and sodium metaborate tetrahydrate (NaBO 2 ·4H 2 O, 99% purity, Sigma Aldrich) were purchased in powder form. As suggested by the differential thermal analysis (DTA) shown in Figure S1, NaBO 2 was obtained by heating NaBO 2 ·4H 2 O up to 350 • C. The waste Mg-Al-based alloys used in this work were obtained from the in-house technical workshop at the Helmholtz-Zentrum Geesthacht in the form of swarf/chips of a few millimeter sizes ( Figure S2). These scrap particles were kept in air before the beginning of the experimental activity. The composition determined via spark emission spectrum analysis is: 76.09 wt.% Mg, 13.6 wt.% Al, 0.06 wt.% Ca, 0.13 wt.% Cu, 0.13 wt.% Mn, 0.45 wt.% Nd, 8.6 wt.% Zn, 0.24 wt.% Y, 0.7 wt.% Ag. All the specimens were prepared and handled in a glove box under continuously purified argon atmosphere (<10 ppm O 2 and H 2 O) to avoid any further oxidation of the starting materials.
In order to reduce the particles' size, the as-received waste material was milled in argon atmosphere for 2 h using a Simoloyer CM08 mill (Zoz GmbH, Wenden, Germany) in a batch of 350 g with 5 mm 100Cr6 steel balls using a ball to powder ratio (BPR) of 20:1. The material morphology was characterized by scanning electron microscopy (EvoMA10, Zeiss, Oberkochen, Germany) at the University of Pavia (Italy). In order to avoid moisture and/or oxygen contaminations during the sample preparation, a small amount of material was placed on a special Al sample holder inside a dedicated argon filled glove box (<1 ppm O 2 and H 2 O). The sample holder was then evacuated before transporting it to the SEM and was opened only after a high vacuum had been created inside the SEM chamber.
The mechanochemical reaction was conducted by ball milling a mixture of Mg-Al waste and dehydrated borates under high hydrogen pressure in a planetary mill. Since Mg accounts for 76.09 wt.% in the Mg-Al-based waste, mixtures of anhydrous LiBO 2 or NaBO 2 and Mg-Al-based waste (as designated in Table 1) were prepared with respect to the amount of Mg contained in the Mg-Al-based waste; that is, the molar ratio of borate and Mg was 1:2. For example, in this experimental study, for the synthesis of the LBOM batch, the amount of LiBO 2 and Mg-Al waste was fixed at 2.430 and 3.121 g, respectively. Similarly, the calculated amount NaBO 2 and Mg-Al waste was 2.697 and 2.779 g, respectively, for NBOM synthesis. The reaction components were put into a high-pressure vial (from Evico magnetics GmbH, Dresden, Germany). The milling process was performed under 70 bar of hydrogen with a BPR of 20:1, using a speed of 500 rpm and milling times in the range between 1 and 36 h. An attempt to synthesize NaBH 4 directly from NaBO 2 ·4H 2 O and Mg-Al-based waste was made by milling the mixture under 1 bar of argon pressure for 36 h (i.e., the amount of NaBO 2 .4H 2 O and Mg-Al waste was fixed at 2.448 and 3.558 g, respectively). For the sake of simplicity, the description and designation of the mixtures prepared and investigated in this work are reported as in Table 1. The hours of milling that the system underwent are indicated by the number following the sample name (e.g., NBOM_36 is the system NaBO 2 + Mg-Al-based waste milled for 36 h).
In order to evaluate the studied process from the thermodynamic point of view, equilibrium composition calculations were performed with the HSC Chemistry software 9.7 [50]. The calculations were done based on the thermodynamic data available for the phases involved in the syntheses. For these calculations, a hydrogen pressure of 70 bar, a vial volume of 200 cm 3 , and a temperature between 25 and 40 • C were considered (Table 2). Since Mg accounted for 76.09 wt.% in the Mg-Al-based waste, stoichiometric ratios were used with respect to the amount of Mg contained in the Mg-Al-based waste as indicated in Table 2. X-ray diffraction (XRD) analyses were carried out using a Bruker D8 Discover diffractometer (Bruker AXS GmbH, Karlsruhe, Germany) equipped with a Cu Kα radiation (λ = 1.54184 Å) X-ray source and a VÅNTEC-500 area detector. The diffraction patterns were acquired in nine steps in the 2θ range from 10 • to 90 • , with an exposure time of 300 s per step and a step size of 10 • . A small amount of powder was placed onto a sample holder and sealed with an airtight dome made of polymethylmethacrylate (PMMA), which is transparent to X-rays.
The composition of the synthesized samples was also characterized by means of the FT-IR technique (Cary 630 FTIR spectrometer, Agilent Technologies Deutschland GmbH, Waldbronn, Germany). For each measurement, the background was calibrated, a small amount of material was placed on the diamond ATR top plate, and the FT-IR spectrum was acquired in the frequency range 4000-400 cm −1 , with a spectral resolution of 4 cm −1 .
The composition of both the starting material containing boron and the milling products was also investigated by means of 11 B Solid State MAS-NMR using a 500-MHz ( 11 B frequency: 160.46 MHz) Bruker Avance III HD NMR spectrometer (Bruker BioSpin GmbH, Rheinstetten, Germany) equipped with a Bruker 4 mm BB/1H-19F probe. Rotation speeds in the range of 8 to 12 kHz were applied. To overcome the broad 11 B background of the standard bore probe, the vendor-supplied "zgbs" sequence was employed. The repetition time of the experiments was chosen in such a way that the sample was fully relaxed.

Energy Transfer during the Milling
Ball milling (BM) is a well-recognized technique to promote physical and chemical transformations into solid and liquid systems [51][52][53]. Among all top-down approaches, BM can be considered as one of today's most used techniques for the production of hydrogen storage materials by using different apparatuses, namely, the attritor mill, vibration mill, and planetary mill. Basically, BM consists of repeated collision events, which involve solid materials trapped between the balls and reactor vial walls. The microscopic transformations are accompanied by macroscopic evidences, which can be estimated by characterization techniques, such as powder XRD. The handling of the milling parameters and the mechanisms behind the process can lead to improvements in the performance in the synthesis of new materials. In this context, the estimation of the mechanical energy transferred to the powders during milling represents one of the most important parameters for monitoring the efficiency of the milling and defining the reproducibility of the synthesis.
The energy, P*, from the mill transferred to the powders per mass unit during the milling process in a planetary ball mill was then estimated using the model proposed by Burgio et al. [54], based on Equation (1): where ϕ is the degree of milling, N b is the number of balls, m b is the mass of balls (kg), t is the milling time (s), Ω p is the rotation speed of the plate (rad/s), ω v is the rotation velocity of vial (rad/s), r v is the vial radius (m), R p is the plate radius (m), d b is the ball diameter (m), and m p is the mass of the material (kg). The results of this calculation correlated to the results of the experimental techniques allow the energy that is transferred to the system in order to reach the maximum yield to be obtained. The influence of every different process parameter on the energy transferred during the milling processes was also recently studied in a methodical analysis based on Equation (1) for hydrogen storage materials [55].

Results and Discussions
The morphology of the Mg-Al-based waste after milling was characterized by means of the SEM technique ( Figure 1). Compared to the starting material (ESI, Figure S2), the shape and size of the milled product were considerably changed. The ribbon-like structure of several hundred micrometers, which characterized the material received from the workshop, is not visible any longer; in its place, slightly elongated particles, with an average size between 10 and 70 µm, are observed. More details about the evolution of this waste alloy as a hydrogen storage material can be found in the literature [46]. The morphology of the Mg-Al-based waste after milling was characterized by means of the SEM technique ( Figure 1). Compared to the starting material (ESI, Figure S2), the shape and size of the milled product were considerably changed. The ribbon-like structure of several hundred micrometers, which characterized the material received from the workshop, is not visible any longer; in its place, slightly elongated particles, with an average size between 10 and 70 µm, are observed. More details about the evolution of this waste alloy as a hydrogen storage material can be found in the literature [46].
The equilibrium composition calculations (refer to SI Figure S3-S6 and Table S1-S6) performed for the stoichiometric mixtures LiBO2 + 2Mg and NaBO2 + 2Mg (total amount of Mg-Al-based waste: 2.5 mol) under 70 bar H2, and NaBO2·4H2O + 6Mg (total amount of Mg-Al-based waste: 7.25 mol) under 1 bar Ar, show that the calculated values for the Gibbs free energy (ΔG°) associated to Reactions (6) to (8) are largely negative, in agreement with a previously published work [56]. The mechanochemical synthesis then leads to the formation of the respective borohydrides and MgO (s).
It is noteworthy that the formation of borohydrides did not occur (Reaction (9)) when ball milling a mixture of Al and borates under the same hydrogen pressure and milling conditions applied in the previous synthesis: Given the fact that the use of a 1:2 metaborate:magnesium stoichiometric ratio only leads to the formation of the respective borohydride plus MgO, this ratio was used for the experimental investigations. After 36 h of mechanical treatment in hydrogen atmosphere, the samples were characterized via the SEM technique and the results are displayed in Figure 2. The system NBOM_36 ( Figure 2a) appears to be constituted of particles, which occasionally agglomerate in a wide size range, whereas the sample LBOM_36 ( Figure 2b) seems to form some flakes. In both cases, the particle sizes of the powders after milling are much finer in comparison to the starting materials. The equilibrium composition calculations (refer to SI Figures S3-S6 and Tables S1-S6) performed for the stoichiometric mixtures LiBO 2 + 2Mg and NaBO 2 + 2Mg (total amount of Mg-Al-based waste: 2.5 mol) under 70 bar H 2 , and NaBO 2 ·4H 2 O + 6Mg (total amount of Mg-Al-based waste: 7.25 mol) under 1 bar Ar, show that the calculated values for the Gibbs free energy (∆G • ) associated to Reactions (6) to (8) are largely negative, in agreement with a previously published work [56]. The mechanochemical synthesis then leads to the formation of the respective borohydrides and MgO (s).
It is noteworthy that the formation of borohydrides did not occur (Reaction (9)) when ball milling a mixture of Al and borates under the same hydrogen pressure and milling conditions applied in the previous synthesis: 4Al Given the fact that the use of a 1:2 metaborate:magnesium stoichiometric ratio only leads to the formation of the respective borohydride plus MgO, this ratio was used for the experimental investigations. After 36 h of mechanical treatment in hydrogen atmosphere, the samples were characterized via the SEM technique and the results are displayed in Figure 2    The specimens were characterized by the FT-IR technique to detect the possible presence of non-crystalline species. Figure 4, and Tables 3 and 4 show the infrared vibration bands observed in the starting and ball-milled materials. The vibrational spectra of borates are constantly complicated due to the capability of coordinating the three and four oxygen atoms of boron atoms to formulate either a monomer or a polymer form. The IR spectrum of the NaBO 2 sample was characterized as shown in Figure 4A(a). The band, which appeared at around 1225 cm −1 , can be assigned to the B-O stretching vibrations of BO 4 units. Another band at 1395 cm -1 is attributed to the asymmetric stretching of the B-O bond of trigonal BO 3 . These values were similarly found in [57,58]. For pure NaBH 4 ( Figure 4A(b)), the measured absorption band for the B-H bending mode is 1108 cm −1 , whereas the bands of the B-H stretching mode are found at 2208 and 2278 cm −1 . In the FT-IR spectrum acquired for the sample NBOM_1, only the absorption bands of NaBO 2 are visible. The spectra of the samples NBOM_12, NBOM_24, and NBOM_36 show the characteristic absorption bands of B-H bending modes at 1118 and 1113 cm −1 , respectively, whereas the absorption bands of B-H stretching modes are observed at 2286, 2294, and 2288 cm −1 , respectively. The NaBH 4 stretching band at 2214 cm −1 is only observed in the spectrum of NBOM_36. All the NaBH 4 FT-IR absorption bands in the ball-milled NBOM are slightly shifted to a higher frequency compared to the values of pure NaBH 4 . This effect eventually originates from the presence of other compounds intimately mixed with NaBH 4 . Similarly to NaBO 2 , the IR spectrum of LiBO 2 was recorded as in Figure 4B(a): In this case, two distinguished bands emerge due to the B-O stretching vibrations of the BO 4 groups (1140 cm −1 ) and of the trigonal BO 3 groups (1420 cm −1 ). In Figure 4B diffraction peaks attributed to the presence of NaBO2. After 12 h of milling (NBOM_12), the diffraction peaks of MgO are detected. In the diffraction patterns acquired after 24 h of milling (NBOM_24), the peaks of MgO are clearly visible. In addition, small peaks belonging to a yet unknown phase (2θ = 28.39°, 32.04°, 38.33°, and 47.24°) are also present. Interestingly, in the diffraction patterns of NBOM_36, the diffraction peaks of NaBH4 are observed together with those of MgO. In the diffraction patterns of LBOM_1, the diffraction peaks of the starting materials (Mg and LiBO2) are still visible. The patterns of LBOM_12, LBOM_24, and LBOM_36 ( Figure 3B(e)-(g)) are characterized by the presence of the diffraction peaks of MgO only. For these systems, even after 36 h of milling, it is not possible to observe the formation of crystalline LiBH4 by XRD.     To better understand the nature of the obtained products and to quantify the reaction yields, the samples were also investigated via 11 B MAS-NMR. Figure 5 shows the acquired 11 B MAS-NMR spectra for the reference and ball-milled materials. In Figure 5A, the NMR spectra of pure NaBH4 and NaBO2 and those of the milled NBOM system are shown. The spectra of all the NBOM specimens To better understand the nature of the obtained products and to quantify the reaction yields, the samples were also investigated via 11 B MAS-NMR. Figure 5 shows the acquired 11 B MAS-NMR spectra for the reference and ball-milled materials. In Figure 5A, the NMR spectra of pure NaBH 4 and NaBO 2 and those of the milled NBOM system are shown. The spectra of all the NBOM specimens ( Figure 5A(d)-(f)) are dominated by the resonance of NaBH 4 at −42.2 ppm. Additionally, the resonance of small quantities of boron oxide around 1 ppm are visible (SI, Figure S9). Upon integration of the spectra, boron oxide accounts for the 2.8%, 1.8%, and 0.3%, of the spectra-integrated intensity of NBOM_12, NBOM_24, and NBOM_36, respectively. The determination of the NaBH 4 yield followed the procedure described in reference [32]. The conversion ratio of NaBO 2 into NaBH 4 , which was calculated based on the integration of the MAS-NMR signals, is approximately 97.2%, 98.2%, and 99.7% for NBOM_12, NBOM_24, and NBOM_36, respectively. No formation of NaBH 4 was observed for NBOM_1, in agreement with the XRD and FT-IR results. Similarly, in the 11 B MAS-NMR spectrum of LBOM specimens ( Figure 5B (c)-(f)), the sharp resonance of LiBH 4 is observed at -41.4 ppm. The analysis of the signals of the MAS-NMR spectra shows that boron oxide contributes to 1.6%, 1.1%, and 0.4% of the total 11 B MAS-NMR-integrated intensity for LBOM_12, LBOM_24, and LBOM_36, respectively. This implies that the conversion yield of LiBO 2 into LiBH 4 is approximately 98.4%, 98.9%, and 99.6% for LBOM_12, LBOM_24, and LBOM_36, respectively. The 11 B MAS-NMR spectrum of LBOM_1 does not show the presence of LiBH 4 . NMR spectrum of LBOM specimens ( Figure 5B (c)-(f)), the sharp resonance of LiBH4 is observed at -41.4 ppm. The analysis of the signals of the MAS-NMR spectra shows that boron oxide contributes to 1.6%, 1.1%, and 0.4% of the total 11 B MAS-NMR-integrated intensity for LBOM_12, LBOM_24, and LBOM_36, respectively. This implies that the conversion yield of LiBO2 into LiBH4 is approximately 98.4%, 98.9%, and 99.6% for LBOM_12, LBOM_24, and LBOM_36, respectively. The 11 B MAS-NMR spectrum of LBOM_1 does not show the presence of LiBH4. Although the use of ball mills for carrying out mechanochemical driven processes is often advantageous from the point of view of the time necessary to complete the process and from the perspective of its scalability, the need to apply high gas pressure within the drum of the mill is a technical challenge that is difficult to overcome. Recently, Felderhoff et al. [60] reported the possibility of partially reversing the hydrolysis of NaBH4 by ball milling from the hydrolysis by-product NaBO2·2H2O with high-purity Mg in an argon atmosphere. As a result of their investigation, they achieved a maximum conversion yield of 68.55%. Although the use of ball mills for carrying out mechanochemical driven processes is often advantageous from the point of view of the time necessary to complete the process and from the perspective of its scalability, the need to apply high gas pressure within the drum of the mill is a technical challenge that is difficult to overcome. Recently, Felderhoff et al. [60] reported the possibility of partially reversing the hydrolysis of NaBH 4 by ball milling from the hydrolysis by-product NaBO 2 ·2H 2 O with high-purity Mg in an argon atmosphere. As a result of their investigation, they achieved a maximum conversion yield of 68.55%.
Inspired by this work, an attempt was made to synthesize NaBH 4 starting from a mixture of NaBO 2 ·4H 2 O and Mg-Al-based waste. The molar ratio between metaborate and Mg contained in the waste was 1:6, as shown in Table 2. The milling process was carried out in Ar atmosphere instead of H 2 gas. Figure 6 shows the XRD diffraction patterns and FT-IR spectra of the reference materials and of the ball-milled material. In Figure 6A(d), for the sample NBOM·H 2 O_36, besides the diffraction peaks of MgO, in agreement with the diffraction patterns of pure NaBH 4 ( Figure 6A(c)), the reflections belonging to NaBH 4 at 2θ = 25.15 • , 29.01 • , and 41.45 • are visible. In Figure 6B, the FT-IR spectra of NaBO 2 ·4H 2 O, NaBH 4 , MgO, and NBOM·H 2 O_36 are shown. The FT-IR spectrum of NBOM·H 2 O_36 ( Figure 6B(d)), similarly to that of pure NaBH 4 , shows absorption bands of the bending mode of NaBH 4 at 1114 cm −1 and of the stretching modes at 2225 and 2289 cm −1 . Therefore, the presence of NaBH 4 was confirmed by both XRD and FT-IR techniques.
peaks of MgO, in agreement with the diffraction patterns of pure NaBH4 (Figure 6A(c)), the reflections belonging to NaBH4 at 2θ = 25.15°, 29.01°, and 41.45° are visible. In Figure 6B, the FT-IR spectra of NaBO2·4H2O, NaBH4, MgO, and NBOM·H2O_36 are shown. The FT-IR spectrum of NBOM·H2O_36 ( Figure 6B(d)), similarly to that of pure NaBH4, shows absorption bands of the bending mode of NaBH4 at 1114 cm −1 and of the stretching modes at 2225 and 2289 cm −1 . Therefore, the presence of NaBH4 was confirmed by both XRD and FT-IR techniques.  Figure 7 shows the 11 B MAS-NMR spectra of pure NaBH4 and NBOM·H2O_36. The NaBH4 resonance of pure NaBH4 (Figure 7a) at −42.2 ppm is also observed for the sample NBOM·H2O_36 (Figure 7b). Less intense signals around 1 ppm, belonging to boron oxide, are also observed. Based on the integrated signals of the NaBH4 and boron oxide, it is possible to claim that the conversion ratio of NaBO2·4H2O to NaBH4 is >99.5%.
The synthesized borohydrides (LiBH4 and NaBH4) can be completely separated from the byproducts (mainly MgO) by an extraction process with isopropylamine and ethylenediamine (EDA) followed by a purification process [36,37], as described in SI. This further step is the object of ongoing research.  Figure 7 shows the 11 B MAS-NMR spectra of pure NaBH 4 and NBOM·H 2 O_36. The NaBH 4 resonance of pure NaBH 4 (Figure 7a) at −42.2 ppm is also observed for the sample NBOM·H 2 O_36 (Figure 7b). Less intense signals around 1 ppm, belonging to boron oxide, are also observed. Based on the integrated signals of the NaBH 4 and boron oxide, it is possible to claim that the conversion ratio of NaBO 2 ·4H 2 O to NaBH 4 is >99.5%.  (1)) to quantify the energy required for the almost full conversion of the starting materials into borohydrides and to assure the reproducibility of the production processes, mainly at the time used when scaling up the reaction. These calculations were performed for the first two systems (refer to Table 1) as evidence that the milling process can be completely characterized to avoid the cost and time required by a trial-anderror procedure with large amounts of materials. Figure 8 presents the correlation between the milling time, the total transferred energy, and the conversion ratio for NBOM and LBOM materials. All parameters, including the BPR, velocity, and mass of powders, were the same for all the samples, thus the energy dissipated at each impact was the same for the entire milling time (SI, Table S7). The milling energy per gram of powder changes with the milling time varying from 1 to 36 h. According to Equation (1), the transferred energy depends on the filling vial coefficient, which is related to the number of balls and reactor volume, as well as to the volume of powder [54]. In this study, the number of balls and reactor volume were the same for all experiments; therefore, the energy transferred was mainly related to the volume occupied by each powder sample [55]. Taking into account that the densities of the initial reagents were not so different, the energy transferred for NBOM and LBOM materials was almost the same (Figure 8). The experimental results (XRD and FT-IR) show that increasing milling time leads to higher yields of borohydrides. Therefore, according to the abovedescribed calculations, 228 Wh/g corresponding to 36 h of milling process is required for NaBH4 and LiBH4 conversion ratios over 99.7% and 99.6%, respectively. The synthesized borohydrides (LiBH 4 and NaBH 4 ) can be completely separated from the by-products (mainly MgO) by an extraction process with isopropylamine and ethylenediamine (EDA) followed by a purification process [36,37], as described in SI. This further step is the object of ongoing research.
The model developed by Burgio et al. was used (Equation (1)) to quantify the energy required for the almost full conversion of the starting materials into borohydrides and to assure the reproducibility of the production processes, mainly at the time used when scaling up the reaction. These calculations were performed for the first two systems (refer to Table 1) as evidence that the milling process can be completely characterized to avoid the cost and time required by a trial-and-error procedure with large amounts of materials. Figure 8 presents the correlation between the milling time, the total transferred energy, and the conversion ratio for NBOM and LBOM materials. All parameters, including the BPR, velocity, and mass of powders, were the same for all the samples, thus the energy dissipated at each impact was the same for the entire milling time (SI, Table S7). The milling energy per gram of powder changes with the milling time varying from 1 to 36 h. According to Equation (1), the transferred energy depends on the filling vial coefficient, which is related to the number of balls and reactor volume, as well as to the volume of powder [54]. In this study, the number of balls and reactor volume were the same for all experiments; therefore, the energy transferred was mainly related to the volume occupied by each powder sample [55]. Taking into account that the densities of the initial reagents were not so different, the energy transferred for NBOM and LBOM materials was almost the same ( Figure 8). The experimental results (XRD and FT-IR) show that increasing milling time leads to higher yields of borohydrides. Therefore, according to the above-described calculations, 228 Wh/g corresponding to 36 h of milling process is required for NaBH 4 and LiBH 4 conversion ratios over 99.7% and 99.6%, respectively. According to the results shown in Figure 8, it would also be possible to reduce the milling time to 12 h with the same conversion ratio. In this experimental study, however, the test was prolonged to 36 h to assess the effects of the milling time on the powder morphology, pureness, and nanostructure. A lower energy consumption, without compromising the yield, is nevertheless fundamental in order to estimate the parameters for scaling up the process. According to Equation (1), it is theoretically possible to increase the mass of the powder milled if all the other factors are increased accordingly, to transfer the same amount of energy in the process.
In practice, however, the size of synthesis is limited by the milling apparatus available. The problem of industrial production of hydride materials has already been considered [61]. The limit to the use of larger industrial devices (where geometries and sizes are larger, masses and forces are higher, but the nature of the process is the same [62]) are controlled by the atmosphere. Non-reactive materials can easily be considered for processing in industrial machines, but when an inert or a reactive atmosphere is required, the batch size is limited by the possibility of properly sealing the milling environment in such dynamic apparatuses [53,63]. Recently, more sophisticated milling processes were developed using larger machines (up to a 100-L milling volume), where the vial is static and the atmosphere is monitored [64]. This would allow for semi-industrial synthesis of borohydrides from a mechanochemical reduction of borates, as well.
The kinetics of LiBH4 formation, determined by quantitative analysis of its solid-state NMR patterns, was obtained by plotting the LiBH4 fraction, α, as a function of the milling time ( Figure 9). The kinetic curve has a sigmoidal shape that is well represented by the empirical Equation (2) [65]: where α represents the mass fraction of LiBH4 formed during the mechanochemical reaction and k is the apparent rate constant. The best-fitted line (dark full line) allows an estimate of the k value for the LiBH4 formation process, which is equal to 5.68 10 −3 min −1 , one order of magnitude higher, for example, than the one estimated in the mechanically-induced metathesis reaction of Mg(NH2)2 (8. 39 10 −4 min −1 ) [66]. In contrast, NaBH4 was not identified for the sample after 5 h of ball milling. There might be an induction period for the NaBH4 conversion under the given conditions. It is claimed that a different mechanism seems to occur in NaBH4 formation during BM, with respect to those shown by LiBH4. For this reason, further experiments are now in progress to clarify this interesting point. According to the results shown in Figure 8, it would also be possible to reduce the milling time to 12 h with the same conversion ratio. In this experimental study, however, the test was prolonged to 36 h to assess the effects of the milling time on the powder morphology, pureness, and nanostructure. A lower energy consumption, without compromising the yield, is nevertheless fundamental in order to estimate the parameters for scaling up the process. According to Equation (1), it is theoretically possible to increase the mass of the powder milled if all the other factors are increased accordingly, to transfer the same amount of energy in the process.
In practice, however, the size of synthesis is limited by the milling apparatus available. The problem of industrial production of hydride materials has already been considered [61]. The limit to the use of larger industrial devices (where geometries and sizes are larger, masses and forces are higher, but the nature of the process is the same [62]) are controlled by the atmosphere. Non-reactive materials can easily be considered for processing in industrial machines, but when an inert or a reactive atmosphere is required, the batch size is limited by the possibility of properly sealing the milling environment in such dynamic apparatuses [53,63]. Recently, more sophisticated milling processes were developed using larger machines (up to a 100-L milling volume), where the vial is static and the atmosphere is monitored [64]. This would allow for semi-industrial synthesis of borohydrides from a mechanochemical reduction of borates, as well.
The kinetics of LiBH 4 formation, determined by quantitative analysis of its solid-state NMR patterns, was obtained by plotting the LiBH 4 fraction, α, as a function of the milling time ( Figure 9). The kinetic curve has a sigmoidal shape that is well represented by the empirical Equation (2) [65]: where α represents the mass fraction of LiBH 4 formed during the mechanochemical reaction and k is the apparent rate constant. The best-fitted line (dark full line) allows an estimate of the k value for the LiBH 4 formation process, which is equal to 5.68 10 −3 min −1 , one order of magnitude higher, for example, than the one estimated in the mechanically-induced metathesis reaction of Mg(NH 2 ) 2 (8. 39 10 −4 min −1 ) [66]. In contrast, NaBH 4 was not identified for the sample after 5 h of ball milling. There might be an induction period for the NaBH 4 conversion under the given conditions. It is claimed that a different mechanism seems to occur in NaBH 4 formation during BM, with respect to those shown by LiBH 4 . For this reason, further experiments are now in progress to clarify this interesting point.

Conclusions
This work demonstrated that the synthesis of NaBH4 and LiBH4 from low-cost starting materials, such as metaborate compounds and Mg-Al waste, employing a common industrial method, such as ball milling, is efficiently possible. The mechanochemical synthesis allows for the use of different conditions, such as hydrogen and argon atmosphere under room temperature, both leading to high yields of conversions in the case of NaBH4. The advantage of using NaBO2·4H2O as a starting material, without any need for water elimination, improves the efficiency of the synthesis method. Experimental results (XRD, FT-IR and NMR results) confirmed that NaBH4 and LiBH4 were successfully synthesized under 70 bar H2 and room temperature by ball milling, achieving conversion efficiencies of NaBH4 and LiBH4 over 99.5%. It is interesting to emphasize that NaBH4 can be directly produced from NaBO2·4H2O plus Mg-Al-based waste under 1 bar Ar and room temperature by ball milling, and the conversion ratio of NaBO2 to NaBH4 can be as high as 99.5 %.
Supplementary Materials: The following are available online at www.mdpi.com/xxx/s1. Figure S1: DTA of NaBO2·4H2O measured at argon atmosphere from RT to 380 °C with a heating rate of 5 °C/min, Figure S2: The Mg-Al based waste as received from the workshop, Figure S3. ΔG vs. T, Figure S4. ΔG vs. T. Figure S5. ΔG vs. T. Figure S6. ΔG vs. T, Figure S7. ΔG vs. T, Figure S8. ΔG vs. T, Figure S9. An inset of NMR spectra in range of 10 ppm to −10 ppm, Figure S10. Schematic diagram of a Soxhlet extractor, Figure S11. Schematic diagram of a hydrogen evolution apparatus.

Conclusions
This work demonstrated that the synthesis of NaBH 4 and LiBH 4 from low-cost starting materials, such as metaborate compounds and Mg-Al waste, employing a common industrial method, such as ball milling, is efficiently possible. The mechanochemical synthesis allows for the use of different conditions, such as hydrogen and argon atmosphere under room temperature, both leading to high yields of conversions in the case of NaBH 4 . The advantage of using NaBO 2 ·4H 2 O as a starting material, without any need for water elimination, improves the efficiency of the synthesis method. Experimental results (XRD, FT-IR and NMR results) confirmed that NaBH 4 and LiBH 4 were successfully synthesized under 70 bar H 2 and room temperature by ball milling, achieving conversion efficiencies of NaBH 4 and LiBH 4 over 99.5%. It is interesting to emphasize that NaBH 4 can be directly produced from NaBO 2 ·4H 2 O plus Mg-Al-based waste under 1 bar Ar and room temperature by ball milling, and the conversion ratio of NaBO 2 to NaBH 4 can be as high as 99.5%.

Supplementary Materials:
The following are available online at http://www.mdpi.com/2075-4701/9/10/1061/s1. Figure S1: DTA of NaBO 2 ·4H 2 O measured at argon atmosphere from RT to 380 • C with a heating rate of 5 • C/min, Figure S2: The Mg-Al based waste as received from the workshop, Figure S3. ∆G vs. T, Figure S4. ∆G vs. T. Figure  S5. ∆G vs. T. Figure S6. ∆G vs. T, Figure S7. ∆G vs. T, Figure S8. ∆G vs. T, Figure S9. An inset of NMR spectra in range of 10 ppm to −10 ppm, Figure S10. Schematic diagram of a Soxhlet extractor, Figure S11. Schematic diagram of a hydrogen evolution apparatus. Table S1: Calculated amounts (mol%) of equilibrium species. Conditions: 25-40 • C and 70 bar H2 and 25-40 • C and 1 bar H2. The amount of H2(s) is not take into account in the calculated amounts, Table S2. Reaction under 70 bar H2 condition, Table S3. Calculated amounts (mol %) of equilibrium species. Conditions: 25-40 • C and 70 bar H2. The amount of H2 (s) is not take into account in the calculated amounts, Table S4. Reaction under 70 bar H2 condition, Table S5. Calculated amounts (mol %) of equilibrium species. Conditions: 25-40 • C and 1 bar Ar, Table S6. Reaction under the milling conditions, Table S7. Energy transferred to powder during milling.
Funding: This research was funded by the European Marie Curie ITN Action-ECOSTORE project, grant number 607040. The authors also thank CONICET (Consejo Nacional de Invetigaciones Científicas y Técnicas) and Alexander von Humboldt Foundation (Fellowship number: ARG-1187279-GF-P).