Enclosure of Sodium Tetrahydroborate (NaBH4) in Solidified Aluminosilicate Gels and Microporous Crystalline Solids for Fuel Processing

2.1. Solidified NaBH₄ aluminosilicate gel and its partial crystalline secondary products

The new material is a gel, formed from sodium-tetrahydroborate (NaBH₄), sodium-silicate, sodium-aluminate and water with the chemical composition 3 Na₂O : 2 SiO₂ : Al₂O₃ : 9-18 NaBH₄ x H₂O; 0 < x < 112. The material can be obtained by a two-step reaction process. Firstly an aluminosilicate mixture under addition of high portions of NaBH₄ has to be prepared at room temperature. Secondly this mixture has to be dried by heating between 80 and 110°C between 0.5 and 4 h. During this procedure the gel undergoes a partial stepwise alteration resulting in solidification of the final product. As an example a typical synthesis batch can easily be prepared by dissolving NaAlO₂ (Riedel-de Haen 13404) in 1.5 ml H₂O before high NaBH₄ amounts (between 100 - 850 mg) have to be added to this solution and dissolved under stirring until a clear solution arises (solution I). A second solution is prepared from 310 mg Na₂SiO₃ (Fluka 2299129) and 1.5 ml water. After total dissolution of the silicate the same amount of NaBH₄ as used for preparation of solution I is added and the mixture has to be stirred too, until a clear solution has formed (solution II). Afterwards gel precipitation starts by the dropwise addition of solution II to solution I. A pasty liquid results from this alkaline gel-boranate mixture. After this the gel is exposed to drying procedure in an oven as described above.

Directly after precipitation of the NaBH₄ gel the product has the state of an amorphous sodium aluminosilicate, containing the whole amount of BH₄-anions from the inserted sodium-tetrahydroborate. As a result of separation of NaOH during gel precipitation and alteration process, the alkalinity remains very high, thus preventing the tetrahydroborate from decomposition by hydrolysis. The subsequent process of drying at 80°C up to 110°C between 0.5 h and 4.0 h causes rapid gel hardening. According to this solidified aluminosilicate gels are converted into secondary products with a high content of NaBH₄. Gel precipitation and alteration during drying at 110°C up to two hours was followed by X-ray powder diffraction as shown in Fig. 1. The powder diagram of the pure salt NaBH₄ is inserted in Fig. 1 for comparison.

It can be seen that without further drying at enhanced temperature only short range order could be present revealing a very broad peak around 30° 2 Theta (d = 2.97 Å) and a shoulder around 48° 2 Theta. The broad peak could be related to short ranged ordered

Si-O-Al (sialate) type bonds which are typically also observed in geopolymers, i.e. when a mixture of waterglas and metakaolin is aged [26]. A significant crystallization can be seen after 30 minutes of drying reducing significantly the amount of short ranged ordered Si-O-Al bonds. The four most intensive peaks can uniquely be indexed as (110), (211), (310), (222) and (330) within a sodalite framework. Further aging at 110°C does not alter the sodalite type peaks, which remain rather broad (Fig. 1 c, d). However, the intensity related to short range ordered sialate bonds becomes strongly reduced simultaneously with the appearance of a rather sharp diffraction peak at 6.5° 2 Theta (d = 13.66 Å). The initial crystallisate in the aluminosilicate could be identified as sodalite type nanocrystals. The strong and sharp peak at 6.5° 2 Theta (13.66 Å) in the XRD pattern (Fig. 1) could be related to a “disordered” sequence obtaining a 1.5 a superstructure, a = typical lattice parameter of the sodalite. The structure may not be seen as an intermediate between sodalite and cancrinite [27]. Parallel to this growth of a sodalite type aluminosilicate the NaBH₄ phase re-crystallizes. This process of NaBH₄-recrystallization under the strong alkaline conditions within the solidified aluminosilicate gel can be seen according to the evaluation of the NaBH₄ peaks in agreement with data of PDF-9-386. Thus the complete material can be regarded as a composite material composed of NaBH₄ and sodalite-type phase embededed in or glued together by sialate bonds

A SEM photograph of the NaBH₄ gel, exposed to open conditions for 4 weeks is given in Fig. 2. Some more general features may be described for the handling of the NaBH₄-gel under open conditions. XRD pattern taken in a series up to 4 weeks are shown in Fig. 3. Compared with the powder pattern taken directly after precipitation and drying for two hours, after 10 days held under open conditions no remarkable decomposition occurs. According to the alkalinity of the sample some sodium carbonate was formed by uptake of carbon dioxide from the air. This can be seen by weak additional lines in the powder pattern, consistent with Na₂CO₃ PDF-18-1208. After 4 weeks under open conditions the intensity of all diffraction peaks becomes reduced and the samples gain a paste-like character.

The samples were further analysed by infrared (IR) absorption spectroscopy (KBr method). The tetrahedral BH₄-anion groups of the NaBH₄ crystals could be identified by strong vibration modes at 1143 (ν₄), 2286 (2 ν₄), 2241 (ν₃) and 2390 (ν₂₊ν₄) ([28-31], and more recently [32]) as shown in Fig. 4. By comparison with the spectrum of NaBH₄ the peaks related to the NaBH₄ in the aluminosilicate gel can be identified at the same positions in spectra a-d.

A significant uptake of water molecules to the gel could be seen as indicated by the increasing intensity of the peak at 1630 cm^-1 (H₂O bending) and the main peak around 3500 cm^-1 together with the broad shoulder towards lower wavenumbers. Despite this uptake of water no sign of hydrolysis reaction accompanied by formation of other borate species [33-35] at the expense of NaBH₄ can be derived from the spectra. This is a further indication for the stability of the NaBH₄ in the composite material. The peak at around 1450 cm^-1 and the smaller one at 880 cm^-1 indicate the presence of CO₃^2-anions [36, 37]. As already mentioned, formation of sodium carbonate is the result of reaction with CO₂ from the air under alkaline conditions on the gel surface. Indications of the presence of sodalite framework are given by the small peaks at at 436 and 469 cm^-1, the triplicate peak at 668, 705 and 733 cm^-1 and the contribution at 990 cm^-1 as could be realized by comparison with a typical sodalite spectrum Fig. 4. The sodalite spectrum appears to be superimposed to what has been called geopolymer type matrix, i.e. the alumosilicate gel. Similar spectra were observed during in situ investigations of sodalite crystallization from appropriate alumino-silicate gel in a KBr matrix [38].

Further tests of thermal stability of the new composite material including NaBH₄ crystals were carried out for heating the sample in a muffle furnace under open conditions in air. Spectra obtained before the treatment and treated for 2 h at 100°C steps up to 500°C are given in Fig. 5 (a-e). It can be seen that the NaBH₄ in the gel largely remains stable up to 300°C and then starts to decompose into borate species at temperatures between 300 and 400°C in air. A simple test of hydrogen release could be given by burning the sample initiated with a pocket lighter. The sample is burning and a glassy-like mixture of aluminosilicate and sodiumborate remains as could be identified in the IR absorption spectra, also shown in Fig. 5 (f). A more precise way for hydrogen release is the reaction in acid solutions [39]. The investigation of hydrogen release from solidified aluminosilicate gels by wetting with diluted acid is described in the following.

2.2. Hydrogen release experiments and further optimizations of storage capability

Hydrogen release experiments and optimizations of the starting compositions and solidification conditions for the hydrogen storage capability were carried out using an apparatus made from glassware as shown in Fig. 6. It consists of a 100 ml bulb, a gas syringe which measures the volume expansion related to the gas release, and the possibility of injection of liquid reactants. The bulb could also be heated and the temperature inside the bulb could be measured during the experiment. A stability check of the NaBH₄ enclosed in the aluminosilicate gel may be given in comparison to the raw NaBH₄-salt. Below 40°C NaBH₄ creates a stable hydrated form NaBH₄ 2H₂O in contact with water. This species dehydrates at about 40°C to water and NaBH₄ which leads to the uncontrollable reaction with water [32]. For a further check of the reactivity of NaBH₄-salt with water an amount of 26.2 mg of NaBH₄ was wetted with 10 ml water in the bulb and heated to 70°C. This reveals an increased gas volume to about 58.7 ml in the reaction with water. For 10 ml water in the repeated experiment the gas volume increased by about 21 ml as a reference value. Since the amount of 26.2 mg NaBH₄ could release 58.66 ml of hydrogen, as also verified in further experiments below Fig. 7 it could be estimated that about 58% of the NaBH₄ reacted at 70°C under such conditions. Repeating the same experiment for an example of aluminosilicate gel (the NaBH₄-gel_0.47, Fig. 7, Tab. 1) reveals that only 11% of the NaBH₄ enclosed in the gel reacted. This shows a significant increase in protection.

A controlled hydrogen release from the NaBH₄ enclosed in the aluminosilicate gel could be achieved by lowering the pH-value by adding diluted acid. Using an injection needle as sketched in Fig. 6 diluted acid was added through a pierceable rubberplug, so the apparatus remained gas tight. The released gas was identified as hydrogen using the hydrogen-oxygen-reaction. Due to the high alkaline character of the NaBH₄-gel CO₂ from the surrounding air is absorbed in small amounts as CO₃^2- (compare IR-Spectra Fig. 4). By adding the diluted acid the CO₂ is also released. A gas check showed, however, that the concentration of CO₂ was below the detection limit of 1000 ppm and therefore negligible. As diluted acid a 1 % solution of hydrochlorid acid was chosen. It showed the best compromise between needed volume for entire hydrogen release and reaction velocity. If higher concentrations of acid were used the gas is released too fast. To get reliable results for every gel composition at least 5 different sample masses were investigated. The volumes of added acid were subtracted from the shown volume at the gas syringe to get the pure released gas volumes. The volumes obtained were plotted for the different samples against the used sample mass. With a linear regression the volume of released hydrogen per 100 mg sample could be calculated and compared to the pure NaBH₄ salt Fig. 7, Tab. 1.

NaBH₄/solid ratios between 0.26 and 0.75 were investigated. Tab. 1 depicts the synthesis masses of the investigated samples. The experiments show a linear trend between used sample mass and released hydrogen volume (Fig. 7). The more NaBH₄ is enclosed in the aluminosilicate gel the more hydrogen can be stored. Using a NaBH₄/solid ratio of 0.75 at the synthesis the released hydrogen volume is equivalent to 72 % of the pure NaBH₄-salt. More added NaBH₄ during the synthesis lowers the protection ability of the aluminosilicate gel after solidification. The hydrogen content approaches a saturation with further added NaBH₄ mass during synthesis. Some higher amount could still be enclosed by decreasing the solidify temperature. This was investigated using NaBH₄/gel ratios of 0.47 and 0.59. At 80°C solidified samples the 0.59 NaBH₄-gel releases about 15 % more hydrogen compared to the identical synthesis, solidified at 110°C. This higher amount of hydrogen storage capacity is reached, however, at the expense of longer solidify time up to 48 hours. Drying temperatures below 40°C are not able to solidify the NaBH₄-gel even after 96 hours which makes these temperatures inefficient. Below 40°C the hydrogen release per sample mass is very low because main fractions of the sample mass consist of water.

It can be concluded that the hydrogen content of the solidified aluminosilicate gels can be varied with the amount of added NaBH₄ during the synthesis. Till now the highest ratio of NaBH₄ per solid reactants in the synthesis is about 0.75 of gels solidified at 110°C. Some higher amount could still be enclosed decreasing the solidify temperature, however, on expanse of the solidification time.

3.1. Primary steps

The three-dimensional structure net of the sodalites is known for more than 80 years [41]. It was found as basic structure type of many zeolite related compounds up to date [42]. The general sodalite composition is Na₈[T¹T²O₄]₆X₂ where T¹ is a trivalent cation (usually Al³⁺) and T² a tetravalent cation (usually Si⁴⁺) but others like Ga³⁺ and Ge⁴⁺ can be built in during synthesis [18, 43-45]. The sodalite framework is built up by a space filling package of truncated octahedral cages (“toc-units”) formed by tetrahedral TO₄ units. Each cage is filled by a [Na₄X]³⁺ -complex with X representing a monovalent anion or anion group as for example the BH₄-anion. Those guests are enclathrated during synthesis according to the chlathralite like properties of sodalites [46]. As known for other salt-filled sodalites the thermal reactivity of the enclathrated guests can differ from the behaviour of the pure salt according to special interactions of the guest-complex with the sodalite host-framework [47-50]. The toc-unit of the sodalite structure is a common building unit in other zeolites like LTA, LSX, X and Y. Fig. 8 gives a schematic view on the sodalite framework and the framework of zeolite LTA.

With the aim to clarify major questions connected with wet chemistry of NaBH₄ and zeolites with wider open frameworks than sodalites, experiments on impregnation possibility by interaction of zeolite A (LTA) and NaBH₄ and on direct synthesis of NaBH₄-LTA using the gel method under addition of NaBH₄ were carried out. The impregnation experiments were performed using commercial zeolite Na-LTA (Fluka-69836) and NaBH₄. The zeolite was stirred in 2 M NaOH-NaBH₄ solution at room temperature with a solid:liquid ratio of 1:20. After a treatment period of 60 minutes the solution was filtered. NaOH residues were carefully washed out of the solid followed by drying at 80°C over night. However the IR absorption spectrum does not show the presence of any BH₄-anions in the otherwise typical LTA zeolite signatures. Therefore, it can be concluded that the NaBH₄ salt or BH₄-anions do not enter nor cannot be stabilized neither in the supercages (grc) or in the toc units of LTA.

Because of this failure of BH₄-incorporation into the cavities of pre-formed zeolite LTA another experimental series was performed to test possibilities of direct formation of NaBH₄-LTA. The common gel method was used under addition of NaBH₄ salt to the gel during its precipitation. Sodium metasilicate and sodium aluminate were used for gel formation and NaBH₄ salt was added to both of the starting solutions before they were mixed to form the gel. Whereas the sodalite crystallization occurs under high alkalinity which prevents rapid hydrolysis of the inserted BH₄-anion, LTA formation needs lower alkaline solutions. Thus the experiments were performed under the Na₂O:H₂O- ratio of 1:20 M. Under those conditions of low alkalinity a partial decomposition of NaBH₄ cannot be excluded. Therefore the crystallization time was shortened neglecting an otherwise necessary further gel aging step in order to prevent NaBH₄ decomposition choosing a temperature of 100°C for 2-4 h. This higher temperature for LTA formation was selected to also accelerate the crystallization process within the short crystallization time interval.

The X-ray powder pattern of a typical synthesis product is shown in Fig. 9. The powder patterns of common zeolite LTA (Fluka) and NaBH₄-sodalite are inserted in this figure for comparison. Beside sharp peaks, consistent with the powder pattern of zeolite LTA, broad lines of sodalite can be distinguished from the powder pattern of the synthesis product.

According to this, the product can be regarded as a mixture of a lower amount of zeolite LTA and mainly nano-sized sodalite beside some short range ordered aluminosilicate units, indicated by the broad peak in the range 20° to 40° 2 Theta. The formation of the two phases zeolite LTA and sodalite can also be observed in the IR absorption spectrum in comparison to spectra of the NaBH₄-sodalite and zeolite LTA (Fluka) Fig. 10. In particular the sodalite can be identified by the triplicate sodalite “fingerprint”.

Further experiments under variation of the alkalinity, the solid/liquid ratio as well as the time of syntheses all failed to obtain NaBH₄ zeolite LTA. In each case a two phase product was observed, consisting of NaBH₄ sodalite and NaBH₄-free zeolite LTA. The reaction parameters mentioned here showed only a small influence on the mass ratio of both of these phases. The results are a hint that even a direct crystallization of NaBH₄-zeolite LTA seems to be impossible under the conditions of gel-crystallization usually used in zeolite chemistry.

3.2. Enclosure of the BH₄-anion in micro- and nano-crystalline sodalites

The route of synthesis follows certain rules in order to include the BH₄-anion in the sodalite cage and not to obtain just co-crystallization of NaBH₄ and the sodalite within the aluminosilicate matrix as described in section 2. Synthesis of NaBH₄-aluminosilicate sodalite in microcrystalline form was performed under mild hydrothermal and strong alkaline conditions (NaOH) using kaolinite as Si-Al-source. An excess of NaBH₄ salt has to be added to this solution. From wet chemical reaction behaviour of pure sodium tetrahydroborate in water it is known that the kinetics of decomposition are highly influenced by the alkalinity [51, 52, 10]. Further parameters of synthesis like the solid to liquid ratio, temperature and reaction time had to be optimized for sodalite synthesis with hydrolysis sensitive BH₄-anions. 50 ml Teflon coated steel autoclaves were used for synthesis.

After screening experiments the amounts of 1 g of kaolinite, 2 g of sodium tetrahydroborate salt and 10 ml of 16 M sodium hydroxide solution were selected for preparation of the favored reactant mixture. Crystallization was performed at a temperature of 110°C for 24 hours reaction time. The final products were washed with water and dried at 80°C for 24 hours [17]. Tetrahydroborate-sodalite nanoparticles were successfully synthesized even at lower temperature hydrothermal conditions (60°C) from high alkaline aluminosilicate gels and NaBH₄ salt [22, 23]. Preparation of basic hydrosodalite by this very simple method was first described by [53, 54] during experiments on zeolite A crystallization at very low temperatures under superalkaline conditions. Fine tuning of this gel method by [55] also yielded basic-hydrosodalite nanoparticles. Gel conditions are suitable for precipitation of salt-filled sodalites and cancrinites, too, as recently demonstrated by [56, 57] for the nitrate sodalite-cancrinite system. The use of similar gels at low temperatures under superalkaline conditions was shown to be a suitable method for NaBH₄-sodalite nanoparticle formation [22, 23] as the hydrolysis reaction of the highly moisture sensitive NaBH₄ salt is retarded under low temperature strong alkaline conditions [10, 51, 52, 58].

Details for obtaining BH₄-sodalite nanocrystaline samples are batch compositions 13 Na₂O : 2 SiO₂ : 1,5 Al₂O₃ : 5 NaBH₄ : 220 H₂O prepared from analytical grade chemicals sodium-metasilicate, sodium aluminate, NaBH₄ and sodium hydroxide solution. Syntheses were performed in a Teflon coated steel autoclave at 60°C. The final products were washed with 500 ml water and dried at 110°C for 48 h [22, 23]. A heating period of 12 h was proved to be an optimal reaction time for nanoparticle formation of suitable size and sufficient crystallinity.

The gallosilicate tetrahydroborate enclathrated sodalite was prepared by alkaline hydrothermal treatment of a solid mixture of gallium oxide, sodium silicate, NaBH₄ in 10 ml of a 6 M NaOH at 110°C for 24 h, using teflon lined autoclaves. The final product was washed with water and dried at 80°C for 24 hours [18]. Synthesis of the aluminogermanate phase was performed from a beryllonite analogous NaAlGeO₄ following [59]. This starting material was obtained from GeO₂, γ-Al₂O₃ and Na₂CO₃ heated for 12 h at 1200°C before quenched to room temperature and crystallized at 800°C for 48 hours. The sodalite was subsequently synthesized by treatment of the NaAlGeO₄ in 10 ml 4 M NaOH at 110°C for 24 hours, again using a Teflon lined autoclave and same washing procedure as for the gallosilicate sodalite [18].

Formation of crystals of good quality with an average size > 0.5 µm could be stated fom SEM investigations of microcrystalline tetrahydroborate sodalite Na₈[AlSiO₄]₆(BH₄)₂ (Fig. 11, left side). For the nanocrystalline sample, obtained after 12 h reaction time (Fig. 11, right side) nanocrystals are “glued together” by amorphous material to larger spherical agglomerates of about 100 nm size [22, 23].

The X-ray powder patterns of the micro- and nano-crystalline aluminosilicate-sodalite, gallosilicate sodalite and aluminogermanate sodalites are shown in Fig. 12. All diffraction peaks of the X-ray powder pattern could uniquely be indexed to pure phase sodalite within P-43n.

Rietveld refinement of the XRD-data for the micro crystalline aluminosilicate sodalite sample showed an amount up to 10 % related to a broad “background contribution” which could be peaked in the range around 30° 2 Theta. This contribution is of importance for hydrogen release from the sodalite as it contains and transports the amount of available water to the sodalite cages as shown below (section 4.1). This contribution which is related to short range ordered sialate type matrix to the sodalite crystals increases in the nanocrystalline sample. This can be seen in Fig. 12 (b) for the nanocrystalline sodalite compared to the XRD pattern of the microcrystalline sample. For this sample, obtained after 12 h reaction time, refinement of lattice constant reveals a = 8.9351(8) Å, being slightly enlarged compared with the microcrystalline phase, a = 8.9161(2) Å. An average crystal size of 25 nm was calculated for this nanocrystalline sodalite and the amount of short range ordered aluminosilicate material could be estimated to about 50 % using the “TOPAS” software [22].

The enclathration of BH₄-anions inside the cages of the micro- and nanocrystalline aluminosilicate sodalite can be seen in the IR-absorption spectra of the samples, Fig. 13. For the microcrystalline sample the aluminosilicate sodalite framework related vibrations, i.e. the six typical peaks can be seen: at about 436 and 469 cm^-1; “triplicate peaks” or ”sodalite fingerprint” at 668, 705 and 733 cm^-1 and asymmetric Si-O vibration at 987 cm^-1 [60]. The BH₄-related vibrations can be seen at 1134 cm^-1 and the characteristic triplicate peaks at 2240, 2288 and 2389 cm^-1 by direct inspection in comparison to the spectrum of NaBH₄. These peaks are very slightly shifted compared to the peak positions in NaBH₄ which could be seen only in an enlarged scale. There are also indications for H₂O contributions around 1630 cm^-1 (bending of H₂O) and in the range between 3000 and 3600 cm^-1 (H₂O stretching) related to the water content in the short ranged ordered aluminosilicate. This contribution is significantly enlarged in the nanocrystalline sample. According to this the intensities of the sodalite framework and BH₄^- contribution appear smaller. The IR spectra of gallosilicate and aluminogermanate NaBH₄ sodalite are also shown in Fig. 13, too. In both spectra the enclathrated BH₄-anions can be seen by intense absorption bands of the BH₄^- tetrahedral group as compared with the spectrum of the pure salt. The spectrum of the gallosilicate-sodalite framework shows two clear resolved maxima at 922 cm^-1 and 945 cm^-1 for the asymmetric T-O-T vibrations. For the ν_s modes two very close adjacent signals with vibrations at 642 cm^-1 and a shoulder at 624 cm^-1 as well as a peak at 556 cm^-1can be seen from Fig. 13 (c) and finally the framework bending mode was found as one sharp signal at 457 cm^-1. The spectrum of the aluminogermanate sodalite Fig. 13 d shows one asymmetric T-O-T mode at 858 cm^-1, two symmetric T-O-T vibrations (609 cm^-1 and 636 cm^-1) and a T-O deformation mode at 387 cm^-1 [18].

3.3. Crystal structure of NaBH₄ sodalites: X-ray diffraction and MAS NMR study

3.3.1. XRD

The X-ray powder patterns of the microcrystalline aluminosilicate, gallosilicate as well as aluminogermanate NaBH₄ sodalites were analyzed by Rietveld method [17]. The atomic parameters of NaCl sodalite were taken as starting values [61] with the BH₄^- group with boron in the centre of the sodalite cage instead of the Cl-anion. Hydrogen was refined on x,x,x positions, restrained to distances of 116.8 pm as found in NaBD₄ [62].

The refined positional-, displacement- and occupancy- parameters together with R-values [63], cell constant and cell volume, are collected in Tab. 2 (standard deviation for occupation factors of B and Na: 3%). In the aluminosilicate sodalite the Si-O and Al-O distances are 163(3) pm and 174(3) pm, respectively l₁ ∙a, l₂ ∙a, a = lattice parameter, Fig. 14 a) For example with the formulas as given by [50] some sodalite structure specific parameters as shown in Fig. 3.7 (a) are the tetragonal tetrahedral distortions α’_Si = 112.6°, α’’_Si = 107,9°, α’_Al = 110.8°, α’’_Al = 108.8°, the tilt angles φ_Si = 23.7° and φ_Al = 22.3°, and the Al-O-Si angle γ = 139°. The “cage filling” configuration could be worked out as illustrated in Fig. 14 b. The Na atoms have three oxygen atoms at 234(2) pm as well as three hydrogen atoms at 267(5) pm in an octahedral arrangement as nearest neighbours. Positional disorder of the hydrogen atoms according to dynamic averaging of orientational disorder is suggested here. Recent ab-initio calculations could rather closely fit to the experimentally determined structural parameters [64].

Na8[AlSiO4]6(BH4)2 : (SG P-43n: a = 8.9161(2) Å, V = 708.79(2) Å3, RWP = 0.042, RP = 0.032, RI = 0.023, Re = 0.031, d = 0.951)
Atom	P-4 3 n	occup.	x	y	z	B /102nm3
Na	8e	1.02(3)1	0.1834(15)	x	x	2.3(7)
Al	6d	1.0	¼	0	½	1.5(5)2
Si	6c	1.0	¼	½	0	1.52
O	24i	1.0	0.1391(28)	0.1487(29)	0.4390(21)	1.1(7)
B	2a	1.11	0	0	0	2.2(34)3
H	8e	1.11	0.424(16)	x	x	2.23
Na8[GaSiO4]6(BH4)2 : (SG P-43n: a = 8.9590(1) Å, V = 719.09(3) Å 3, RWP = 0.097, RP = 0.074, RB = 0.023, Re = 0.043)
Na	8e	1.02(2)1	0.1731(3)	x	x	2.7(2)
Ga	6d	1.0	¼	0	½	1.1(2)2
Si	6c	1.0	¼	½	0	1.1(2)2
O	24i	1.0	0.1336(3)	0.1513(3)	0.4310(4)	1.9(2)
B	2a	1.11	0	0	0	3.2(8)3
H	8e	1.11	0.4371(38)	x	x	3.2(8)3
Na8[AlGeO4]6(BH4)2 : (SG P-43n: a = 9.0589(2) Å, V = 743.40(6) Å 3, RWP = 0.112, RP = 0.082, RB = 0.026, Re = 0.043)
Na	8e	1.02(3)4	0.1740(4)	x	x	2.5(2)
Al	6d	1.0	¼	0	½	1.4(2)5
Ge	6c	1.0	¼	½	0	1.4(2)5
O	24i	1.0	0.1425(4)	0.1441(4)	0.4296(5)	1.6(2)
B	2a	1.14	0	0	0	3.2(8)6
H	8e	1.14	0.4401(38)	x	x	3.2(8)6

Table 2.

Atomic parameters of sodalites Na₈[AlSiO₄]₆(BH₄)₂ [17] and Na₈[GaSiO₄]₆(BH₄)₂, Na₈[AlGeO₄]₆(BH₄)₂ [18]

In the case of gallosilicate sodalite the Ga-O and Si-O distances are 161.1(3) pm and 181.9(3) pm, respectively. The boron atom is located at the centre of the sodalite cage. In a static statistical model of positional disorder the sodium atoms have three oxygen atoms at 234(4) pm as well as three hydrogen atoms at 253(1) pm as nearest neighbours i.e. in principal the same arrangement, as in aluminosilicate sodalite shown in Fig. 14 b. Positional disorder of the hydrogen atoms according to dynamic averaging of orientational disorder is suggested here, too [18]. For the aluminogermanate sodalite the Ge-O and Al-O distances read 173.0(3) pm and 174.9(3) pm. The sodium atoms in this phase have three oxygen atoms at 234(4) pm and three hydrogen atoms at 257(5) pm as nearest neighbours.

3.3.2. MAS NMR

MAS NMR measurements of nuclei the ¹H, ¹¹B and ²³Na were performed for the microcrystalline NaBH₄ sodalites with aluminosilicate-, gallosilicate- and aluminogermanate framework on a Bruker ASX 400 spectrometer. Further informations on orientational disorder, dynamics and host-guest interaction between framework atoms and BH₄-anions inside the sodalite cages can be derived from MAS NMR spectroscopy [17, 65]. The ¹¹B MAS NMR results of the NaBH₄ sodalites are shown in Fig. 15 and in Tab. 2. All spectra were recorded at 128.38 MHz (pulse duration: 0.6 μs; 100 ms pulse delay, 10000 scans were accumulated at a spinning rate of 12 kHz). Chemical shifts were determined using NaBH₄ (δ = –42.0 ppm from BF₃ Et₂O) as an external reference [65]. As a result of boron nitrite of the probehead a broad line is found in all spectra, beside a sharp narrow signal. Thus spectra after subtraction of the broad background are included in Fig. 15. The sharp signals with isotropic chemical shifts (δ_iso (¹¹B) around –49.08 ppm (see Tab. 3) are typical for boron tetrahedrally coordinated by four hydrogen atoms [66]. The full-width at half-maximum (FWHM) of the sharp lines are included in Tab. 3.

11B MAS NMR
Sodalite	δiso [ppm]	FWHM [ppm]	CQ [kHz]	ηQ
AlSi	–49.08	1.70	–	–
GaSi	–49.91	1.28	–	–
AlGe	–49.01	1.11	–	–
23Na MAS NMR
AlSi	–6.61	4.43	8.82	0
GaSi	–1.31	2.32	6.41	0
AlGe	–1.60	1.88	6.75	0.22

Table 3.

Isotropic chemical shift (δ_iso), quadrupolar coupling constant (C_Q), asymmetry parameter (η_Q), Lorentzian/Gaussian broadening (FWHM) of the ¹¹B MAS NMR spectra of NaBH₄ aluminosilicate (AlSi), gallosilicate (GaSi), aluminogermanate (AlGe) sodalite [65].

Compared with pure NaBH₄ (δ¹¹B = -42.0 ppm) a slight downfield shift of the signals can be found for all three sodalites as a result of the matrix effect of the surrounding framework atoms. The nearly equal values of the chemical shifts for the three sodalites indicate no significant influence of framework composition on the arrangement or the shape of the BH₄-anions inside the sodalite cages [65]. Threefold-coordinated non-framework boron or four-coordinated framework boron can also be excluded from the distinct chemical shift values of the sharp signal in all three cases, because ¹¹B of BO₃-units resonates in the range between 12 and 25 ppm, and BO₄-groups exhibit these shifts in the approximate range between –4 ppm and –6 ppm [66, 67].

Following [65] further informations can be derived from the signals. The sharp and narrow lineshape of the signals show almost no quadrupolar interactions due to the discreet BH₄-unit possessing a highly symmetrical environment for boron in the sodalite-cage as well as possibly fast dynamic site exchange of hydrogen. Because of a relatively low C_Q (~ 0.501 MHz) value expected for BO₄-groups, and a larger value (C_Q ~ 2.5 MHz) expected for BO₃-units [66-68] the present quadrupole coupling parameters rule out any contribution of BO₃- or BO₄-units. A small contribution at +2 ppm can be seen in the spectrum of aluminosilicate sodalite after subtracting of the probehead signal. This signal could be attributed to a small amount of B(OH)₄-anions in the sodalite cages. A similar signal at about 1.7 ppm is known in the ¹¹B MAS NMR spectrum of NaB(OH)₄-aluminosilicate sodalite [69].

The ²³Na MAS NMR signals were recorded at 105.84 MHz (pulse duration: 0.6 μs, pulse delay: 100 ms; accumulation of 5000 scans; spinning rate: 12 kHz; external reference: solid NaCl salt). The “dmfit2003” software was used for peak fitting [70]. The ²³Na MAS NMR spectra are given in Fig. 16 (left). Further informations on chemical shifts, FWHM and quadropolar parameters C_Q (quadrupole coupling constant) and η_Q (asymmetry parameter) are summarized in Tab. 3. The signals exhibit a narrow line shape typical for a single type of sodium coordination indicating nearly no quadrupole interactions according to the very small quadrupole parameters (C_Q and η_Q in Tab. 3). This implies well defined position probably caused by highly symmetric and cubic orientation of the BH₄-anions on the one side and the framework oxygen on the other side but also enables dynamic fast motion of the guest atoms within each sodalite cage leading to motional narrowing.

The ¹H MAS NMR spectra were obtained at 400.13 MHz with a single pulse sequence duration of 1.5 μs (90 degree pulse length of 6.5 μs) and a recycle delay of 5-30 s. 100 scans were accumulated with a spinning rate of 14 kHz, and tetramethylsilane (TMS) was used as an external reference. The high spinning speed helps to remove any residual dipolar broadenings. The ¹H MAS NMR spectra of the sodalites, given in Fig. 16 (right) exhibit a single sharp intense line at a chemical shift of about –0.6 ppm for all three sodalites assigned to the hydrogen atoms of the BH₄-groups. This indicates that proton chemical shift is not influenced by the chemical composition of the sodalite framework. A shift in high field direction, compared with NaBH₄ salt (δ(¹H) ≈ 1.0 ppm) results from the enclathration of single BH₄-groups in contrast to their incorporation in the NaCl-type structure of the salt and its strong heterovalent bonds [65]. A further but weak signal at ~ 5.0 ppm in each spectrum could be related to water molecules enclathrated in distorted sodalite cages [71] or following the interpretation due to water incorporation in short range ordered sialate bonds. The nature of the shoulder of the main signal at about 0.6 ppm could be related to enclathrated water molecules or to B(OH)₄-impurities as observed in ¹¹B NMR signals, too. A third less intense signal at 1.2 ppm appears only in the spectrum of the aluminogermanate sodalite. It could indicate small amounts of (H₃O₂^-)-impurities in a few sodalite cages [71, 65].

4.1. Variation of cage anions of Na₈(AlSiO₄)₆(BH₄)₂ when heated

The effect of various heating rates and heating temperatures in ex situ TG experiments as well as in situ TIR measurements in dry and wet NaCl environments has been investigated for the microcrystalline NaBH₄-sodalite. Heating with a constant rate of 2°C/min to 300, 400 and 500°C (He-flowing conditions, 20 ml/min) reveals weight losses of about 0.2 %, 0.6 % and 0.9 %. This weight loss is related to dehydration and to hydrogen release. The dehydration corresponds obviously to water contained in some short range ordered sialate bonds as explained in section 3.2 (compare also below peaks denote HOH and D in Fig. 19). In experiments with heating rates of 2°C/min, 4°C/min and 6°C/min to 300°C the weight loss reads 0.2 %, 0.4 % and 0.7 %, respectively. Thereby the IR absorption intensity of BH₄ related peaks decreases about 11 % at heating rate 2°C and 7 % at heating rate 6°C/min. This shows that the faster the heating rate the lesser the loss of BH₄-anion in the sodalite cage. Therefore, the faster the heating rate the more water leaves the sample without use for hydrogen release and, accordingly, the higher the weight loss. The spectra of the samples taken after cooled down to room temperature (rate 2°C/min) from 300, 400 and 500°C, exposed to atmospheric conditions and pressed into KBr pellets are shown in Fig. 19. Compared to the unheated sample the BH₄ related intensity has decreased by about 15.5 %, 13 % and 11 % when heated to 500, 400 and 300°C, respectively. Related with the decrease in BH₄-anion concentration new peaks appear denoted in the spectra by A, B, B’ and C. This notation follows that given earlier [17, 22] relating peak A, B, B’ to anions H₃BOH^-, H₂B(OH)₂^- and HB(OH)₃^-, respectively. According to this the following reactions could be seen:

A further reaction of hydrogen release followed by two steps of dehydration which finally leads to peaks C could be suggested as:

The step eq. 4 which reveals the anion species B(OH)₄^- is not seen directly due to fast dehydration above about 400°C as known for the B(OH)₄-sodalite [72, 38]. For better comparison appropriate spectra of microcrystalline Na₈(AlSiO₄)₆(B(OH)₄)₂ and those obtained by dehydration at 200°C for Na₈(AlSiO₄)₆(BO(OH)₂)₂ and at 400°C for Na₈(AlSiO₄)₆(BO₂)₂ are shown in Fig. 20. The BO₂-anion in the sodalite cages can markedly be observed by means of the peaks at about 1958 and 2029 cm^-1 denoted as peaks C here. The anion species BO(OH)₂^- shows characteristic peaks in the range 1490-1520 cm^-1 (denoted C’ further below) and around 1150 cm^-1, beside the OH-stretching at about 3610-3620 cm^-1 compared to the OH-stretching of the B(OH)₄-anion at 3620-3640 cm^-1 [38]. According to this BO₂-anion species in Fig. 19 can readily be identified indicating that reactions eq. 4, 5 and 6 occurred.

The microcrystalline sodalite shows a sharper peak which remains present during heating (see below) with maximum at about 3620 cm^-1 denoted as D’. As discussed by [17, 22] D’ could be related to an OH^- anion in the sodalite, indicating basic or hydro-hydroxo sodalite. D’ could also be related to the formation of B-OH forms or to superimposition effects of various species. A clear distinction is hard to obtain. Another important contribution is the presence of water molecules indicated by H-O-H bending vibrations at about 1640 cm^-1 and OH stretching at 3000-3600 cm^-1 called D in Fig. 19. It could be observed that these contribution becomes partly dehydrated during heating and are in some extend related to the hydrogen release reactions. If the sample is dehydrated the hydrogen release reaction invariably stops. It could be shown that all the sodalite with all formed species remains stable below 600°C also at invariably long time. Heating above 630°C immediately leads to framework destruction. These type of thermal stabilities were investigated in detail by [19, 20].

The effect of rehydration could be used to further proceed the hydrogen release reactions as shown in Fig. 21. The sample preheated at 500°C was taken out of the thermobalance, exposed to atmospheric conditions and run again up to 500°C under He conditions in 7 further cycles. Spectra obtained after each run show a gradual decrease of BH₄-absorption intensity and increase in BO₂^- content (Fig. 21). Similar to the C peaks, the B peak also becomes a bit more pronounced, whereas the peaks B’ and A become broader and more unspecific. The framework vibrations remain largely unchanged and may depict only the changes due to the changed borate species, i.e no significant destruction of the framework is observed. Thus these experiments show the high stability of the sodalite during several heating cooling cycles. A certain amount of water content related to the matrix consisting of short range ordered sialate bonds of the sample is reloadable and can be used to continue the hydrogen release reactions. An exponential decay of the ratio BO₂^-/BH₄^- could be observed (inset in Fig. 21). There is a significant increase in intensity in the range of peak B above about 400°C, too. There is further work to do to find out whether here intracage reactions occur or if borate species outside the sodalite cages are formed which could be related with a destruction of the framework. Explanation could, however, also be given with an other anion in the cage, e.g. H₂BO^- (compare below, eq. 7, 8).

The reaction sequence of hydrogen release may be followed in more detail for the microcrystalline NaBH₄-sodalite in TIR experiments as shown in Fig. 22. In the lower part spectra are shown when cooled down to room temperature after heating to 200, 300, 400 and 500°C in NaCl pressed pellet. In the upper part the spectra are taken at denoted temperatures. It can be seen that peaks A grow in intensity followed by B and B’ with increasing temperature. The formation of B(OH)₄-anions may not be detected as they become dehydrated to BO(OH)₂-anions as indicated by peak C’ at temperatures between 250 and 450°C. Above this temperature the formation of BO₂-anions is indicated by the peaks C (C1, C2). In TIR experiments of the 500°C pre-reacted sample it was observed that the C peak intensitity (BO₂^-) decreases from about 200°C and increases again in intensity above about 370°C. Complementary to this the intensity related to (BO(OH)₂^-) (called C’, Fig. 22, right) increases and decreases. Additionally the BH₄^- intensity decreases above about 370°C. This indicates that in a first step the reacting water must rehydrate the BO₂^- species and finally effectively reaches the BH₄-anions in the centre of the sodalite crystals. This implies that the reactions are controlled by the diffusion through the sodalite cages. The diffusing species might mainly be OH^- and H^- via defect formation in the cages. Thus water disproportion and H₂ formation may occur at the crystal surface.

A closer view of the characteristic sodalite fingerprint peaks ν₁, ν₂, ν₃ is shown in Fig. 22 (left). A significant shift of ν₁, ν₂, ν₃ towards lower wavenumbers occurs with increasing temperature which is almost completely reversible with decreasing temperature (Fig. 23). The small deviations can be related to the changes in the cage filling species. TIR experiments were also carried out using KBr pressed pellets to demonstrate the effect of an exchange of Na-cations from the sodalite with K-cations from the matrix as indicated by the strong deviation in temperature dependance of ν₁, ν₂, ν₃ from the effect observed using NaCl pellets (Fig. 23). A significant exchange occurs above about 250°C which could be similarly demonstrated for other sodalites, too [38]. This indicates that the Na-cations become highly mobile as could also be observed in temperature dependent investigations using MAS NMR [73] and XRD structure refinements [50] on related sodalite compositions. It can be concluded that above 250°C the jump rate of Na⁺ in the cage and through the sixring windows becomes very significant which implies that the out-in jump rates of other ions could also increase.

The change in peak intensities obtained in the NaCl related TIR experiment (Fig. 22) for peaks A, B, C’ and C could be evaluated in more detail as shown in Fig. 24. The intensity of A peaks increase significantly above about 220°C, crossing over in an effective temperature independent behaviour above about 320°C. About at that temperature peak B and also C’ start to increase in intensity. The later effect shows that dehydration occurs as described by eq. 5. This implies that reaction eq. 4 occurred rather fast which could not be resolved. Species B’, i.e. eq. 3, could not be considered separately because of superposition in the spectra with peaks A and peak B. A decrease in intensity of peaks A, C’ and also B’ is observed above about 450°C where peaks C start to increase strongly in intensity. This shows that above 450°C strong dehydration occurs leading finally to the BO₂-anion in the sodalite cage. It can be suggested that part of this water is used for a further effective hydrogen release with reactions as given by eq. 1-4.

One interesting point is that the peak intensity of peak B also re-increases above about 450°C. If this effect is related to further hydrogen release reaction or to a dehydration reaction or a dry hydrogen release of the type eq. 7 or 8 [17, 22] requires further investigations.

A proof that hydrogen is released is given using a nitrate tracer reaction in TIR experiments [22, 24, 74] as shown in Fig. 25. Here some amount of KNO₃ has been added to the sample pellet which could be identified as an additional peak in the spectra (see inset in Fig. 25). With increasing temperature this peak strongly decreases in intensity above about 250°C. Repeating the TIR experiment with only KNO₃, it shows a slight decrease in intensity of the KNO₃ related peak above about 400°C due to a gradual thermal decomposition. Therefore, it can be concluded that hydrogen is released from the BH₄-sodalite sample leading to a redox reaction with KNO₃. Since this reaction cannot be achieved by hydrogen molecules (compare below, Fig. 27) it can also be concluded that hydrogen is set free in an activated form.

Another interesting point is the effect of using “wet NaCl” for the pressed pellet in the TIR experiment as this increases the amount of H₂O available for the hydrogen release reaction from the embedded BH₄-sodalite. This even leads to a complete decrease of the BH₄-content with increasing temperature. The loss in BH₄-intensity in “dry NaCl” and “wet NaCl” are compared in Fig. 26 demonstrating again the effect of the different amounts of available H₂O content for the hydrogen release reaction. Included is the observed effect for nanocrystalline BH₄-sodalite as well, also using dry NaCl and wet NaCl. Here the difference is marginal distinct since the nanocrystalline sample contains a much higher amount of water related to the higher contribution of short range ordered sialate bonds. Moreover the reaction path for hydrogen release is found at significantly reduced temperature due to the much smaller crystal size.

4.2. Experiments on regeneration of pre-reacted BH₄-sodalite

Based on the observation that the hydrogen release reactions of the BH₄-anion in the sodalite cage with H₂O could be carried out in consecutive steps it seems likely that all reactions according to eqs. 1-4 or at least some are reversible. It can be concluded that hydrogen release from BH₄-sodalite crystals is a diffusion controlled process were OH-anions diffuse inward and H-anions outward. Following this, the reinsertion reaction seems only be possible if appropriate H^-/OH^- gradients could be realized which should govern the efficiency of “reactor regeneration”. This assumption could be supported by the “nitrate reduction reaction” observed in Fig. 25. Following reference [75] such type of low temperature nitrate reduction can not be achieved by hydrogen molecules. It has been shown that hydrogen molecules need to be activated by the presence of special catalysts, e.g. Pt or Pd. This requirement could be demonstrated in TG/DTA experiments with heating/cooling runs with as received commercial KNO₃ (Merck) with and without the addition of Pt powder (Fig. 27). Using pure KNO₃ the effect of the structural phase transition can be seen at 140°C and 130°C in the heating up (20°C/min) and cooling down (20°C/min) run as endothermic and exothermic peak, respectively. There is a negligible weight loss of less than 0.5 %, which could be related to dehydration effect. The experiments were carried out using forming gas (10%H₂/90%N₂). In a second run using the same conditions but about 10 wt% of Pt powder was added to the KNO₃ sample. A weight loss of about 13 % for the KNO₃ content is observed. There is a strong exothermic peak centered around 150°C due to the reaction related to the weight loss. The endothermic peak due to the structural phase transition appears only as a very small minima in the heating up run and the reversal to the low temperature phase is absence indicating the complete chemical reaction of KNO₃. The exothermic peaks at 220°C and 180°C in the cooling run can be related to the chemical products. The results are in agreement with observations reported by [75]. Two important conclusions can be drawn: 1. The hydrogen leaving the sodalite crystals are in an activated form which is able to reduce the nitrate even in some distance to the crystals in the pressed pellet (KBr/NaCl) as used in the TIR experiments. 2. Hydrogen can be brought into an activated form at temperature already as low as 150°C in the presence of Pt. Therefore this effect could be used to increase the appropriate partial pressure of activated hydrogen to stop or even reverse the reactions eq. 1-4. Conclusively series of experiments were carried out using Pt powder in addition to the sample either brought in simply as additive to the powdered sample or even as additive in the synthesis route. Some main results may be outlined in the following.

First steps of experiments were carried out with the as synthesized BH₄-sodalite in comparison to samples with the addition of about 20 wt% Pt powder using various gases: He (99.9999), synthetic air (80%N₂/20%O₂) and forming gas (10%H₂/90%N₂). The experiments were carried out using always the same conditions with a continuous gas flow of 20 ml/min with heating rate of 4°C/min to 500°C, holding time of 10 min at 500°C, followed by cooling with 4°C/min to room temperature. Spectra of the starting sample and taken after the experiment are compared in Fig. 28. As before the heat treated samples reveal in all cases a reduction in BH₄-absorption intensity and peaks A, B, B’ and C as related to H₃B(OH)-, H₂B(OH)₂-, HB(OH)₃- and BO₂-anion species, respectively. A quantification of the intensity ratio of BH₄^-/BO₂^- and the obtained mass losses with respect to the amount of BH₄-sodalite is given in Tab. 4 These data sensitively show that the highest reaction rate is achieved under synthetic air conditions – and here to a higher extend in the presence of Pt compared to those in its absence. There is also a slightly lower weight loss in the presence of Pt indicating that here more hydrogen could have been released and less much loss to dehydration occurred. It may, however, also not be ruled out that any uptake of oxygen from the air occurred which could also be indicated by the increased intensity at the B peak position. Moreover the experiments show that the same weight losses and reaction rates BH₄^- to BO₂^- could be seen when He is used with and without Pt addition as well as forming gas only without Pt. If Pt is added the reaction rate becomes smaller and the weight loss increases. This shows that here dehydration takes place with a significantly reduced release of hydrogen. These preliminary experiments show the influence of hydrogen only in the presence of an effective activation of hydrogen which may produce a significant reduction in the concentration gradient for outward diffusion of H-anions.

Further experiments were carried out with BH₄-sodalites with Pt added into the synthesis route. X-ray diffraction pattern could prove the growth of microcrystalline BH₄-sodalite in the presence of nanocrystalline Pt. REM/EDX investigations show a rather homogeneous distribution of Pt. The IR spectra (see below) showed some weak formation of A peaks (H₃B(OH)-anion) due to the presence of Pt during synthesis. Typical results of the Pt free and Pt containing as synthesized samples treated with a hydrogen pressure of 162 bar and 3 h holding time at 250°C are shown in Fig. 29. The Pt free sample shows pronounced formation of peaks A and B related to H₃B(OH)- and H₂B(OH)₂-anions in the sodalite cages, respectively. Contrary to this the sample containing Pt shows almost no change in the infrared absorption beside only a slight increase in A peak intensity. It is interesting to note that there is no peak B’ observed. This can be explained by the too low temperature, but could also be related to the special high pressure conditions. It was observed using closed Au capsules and at 450°C for half an hour pre-reacted samples that the B’ peak completely disappeared at 300°C at autogeneous pressure. This implies a high stability for H₃B(OH)- and H₂B(OH)₂-anions but less much for H₃B(OH)-anions and also for dehydrated BO₂-anions under such conditions. As there is also no indication of BO(OH)₂-anions it may be concluded here that the peak at 3620-3640 cm^-1 could be related to B(OH)₄-type species present already in the as synthesized sample.

In Fig. 30 the result of a reinsertion experiment carried out using 160 bar H₂ at 200°C of the Pt containing sample which was pre-reacted at 400°C in air is shown [25]. It can be seen that peak A increases in intensity whereas peak B decreases.

This first step of hydrogen reinsertion concerning the reaction eq. 2 could be supported in further systematic experiments where the influence of reaction temperature and time has been investigated at 55 bar hydrogen pressure on the Pt-containing sample pre-reacted at 400°C for 1 h. The results showed that with increasing temperature at 200, 250 and 300°C the BH₄-anion concentration remains constant and the same as in the pre-reacted sample. The A peak increases in intensity by about 10%, whereas the B peak decreases by about 5%. Moreover the B(OH)₄^- related intensity did not vary within the error of estimation. Thus it is likely that under such conditions reaction eq. 2 could effectively be reversed. The evaluation of peak intensities with increasing reaction time at 250°C and 55 bar H₂ between minutes up to 6 h also showed that the BH₄-anion concentration remains rather constant also reproducing the finding of the former series. The content of H₃B(OH) anions is higher compared to the pre-reacted material in minutes and slightly increases with increasing time. This shows that a fast reversion of reaction eq. 2 can be achieved.

The reinsertion experiments have shown that a slight but significant back reaction in the content of H₂B(OH)₂-anions To H3B(OH)-anions could be achieved. A higher efficiency might be reached with a further optimisation of the catalyst function. An improved back reaction might also be expected with the help of MgH₂ addition, which could increase the concentration gradient of H^- significantly. In some preliminary experiments commercially available 30 wt% MgH₂ powder (Merck) was simply mixed with two different pre-reacted samples and the spectra were taken of these samples and those after 3 h exposure to 150 bar hydrogen pressure at 200°C. The spectra shown in Fig. 31 reveal first of all the strong contribution of MgH₂ which can be seen by direct comparison with the spectrum of MgH₂ also shown. The special pre-reaction conditions of the Pt-containing BH₄-sodalite are 4 h at 360°C (sample I) and 1 h at 450°C (sample II). The most obvious and important observations here are that in both cases some part of MgH₂ forms Mg(OH)₂ as is identified by the new sharp peak at 3699 cm^-1 for the samples treated in 150 bar hydrogen. The evaluation of the intensity of the triplicate BH₄-peaks reveal an increased intensity by about 4 % and 2 % for sample I and II, respectively. By the same time the A-peak intensity decreased by about 6.2 % in both cases. These results indicate the formation of BH4-anions in the sodalite cages according to eq. 1.