Improvement of the kinetics of hydrogen release from ammonia borane confined in silica aerogel

1 1 2 3 4 5 Improvement of the kinetics of hydrogen release from ammonia 6 borane confined in silica aerogel 7 8 9 10 Miriam Rueda, Luis Miguel Sanz-Moral, José Juan Segovia, Ángel Martín* 11 Department of Chemical Engineering and Environmental Technology - University of 12 Valladolid 13 TERMOCAL Research Group – University of Valladolid 14 15 16 c/ Doctor Mergelina s/n 47011 Valladolid (Spain) 17 Tel: +34 983423174, e-mail: [email protected] (Á. Martín) 18 19 20 2 Improvement of the kinetics of hydrogen release from ammonia 21 borane confined in silica aerogel 22 Miriam Rueda1, Luis Miguel Sanz-Moral1, José Juan Segovia2, Ángel Martín1* 23 1Department of Chemical Engineering and Environmental Technology - University of 24 Valladolid, c/ Doctor Mergelina s/n 47011 Valladolid (Spain) 25 2TERMOCAL Research Group – University of Valladolid, c/ Paseo del Cauce 59 47011 26 Valladolid (Spain) 27 Tel: +34 983423174, e-mail: [email protected] (Á. Martín) 28 29 Abstract 30 Ammonia borane is a promising hydrogen storage material due to its high gravimetric 31 capacity (19.6 % wt), but it also presents limitations such as a slow hydrogen release with 32 a long induction time, a difficult regeneration, or the formation of foams and gaseous by33 products during thermolysis. Previous studies have shown that by nanoconfinement of 34 ammonia borane within a porous support some of these limitations can be overcome due 35 to the reduction and stabilization of ammonia borane particle size. However, this effect 36 was only observed with moderate ammonia borane loadings, as with higher loadings the 37 pores of the support became obstructed. In this work, silica aerogels produced by CO2 38 drying, with pore volumes up to 2 cm3/g, have been used to confine ammonia borane. 39 The influence of the amount of ammonia borane loaded on the aerogel support on the 40 thermal and structural properties of the material has been analyzed. It has been found that 41 more than 60 wt% of ammonia borane can be effectively stored in the pores of the 42 aerogel support. The resulting material shows faster hydrogen release kinetics by 43 thermolysis at 80ºC, due to a significant reduction in the mea size of ammonia borane 44 3 after confinement and the participation of SiOH and SiOSi groups of silica aerogel in the 45 decomposition mechanism. 46 47 Keywords: solid state hydrogen storage, ammonia borane, silica aerogel, 48 nanoconfined, supercritical carbon dioxide. 49 50 1. Introduction 51 As fossil fuel reserves are increasingly limited and their use constitute a constant source 52 of greenhouse gases and other environmental problems, the development of alternative 53 energy sources is attracting a considerable attention [1, 2]. In contrast with fossil fuels, 54 which can be easily stored and used when needed, the production of energy from most 55 renewable sources is variable and it cannot be directly controlled. A possible solution 56 for this limitation could be to use hydrogen (H2) as an energy vector, according to the 57 approach commonly known as ‘hydrogen economy’ or ‘hydrogen society’ [3]: 58 Hydrogen can be produced from water (by electrolysis, thermal decomposition, 59 thermochemical processes, photolysis etc.) using renewable energy sources, and stored 60 until needed. Furthermore, the efficiency of the combustion of hydrogen (by 61 combustion in internal combustion engines, catalytic combustion or fuel cells) is high, 62 and it is one of the most environmentally favorable fuels, as it produces nearly zero 63 gaseous emissions. 64 At large scales, hydrogen can be transported through pipelines (gas H2) or tankers 65 (liquid H2), and it is a good energy vector with an energy density of 33 kwh/kg, 66 containing three times more energy than any hydrocarbon fuel on a weight basis [4]. 67 However, the storage of hydrogen in small mobile units such as vehicles or small 68 electronic equipment is more challenging. 69 4 These challenges can be quantified analyzing the targets set by the US Department of 70 Energy for 2020 for automotive hydrogen systems: a hydrogen storage gravimetric 71 capacity of 5.5 wt% and a volumetric capacity of 0.040 kg/L, with a maximum cost of 72 333$/kg H2 stored [5]. Hydrogen storage by some of most obvious systems using 73 compressed or condensed H2 cannot fulfill these requirements due to the physical 74 properties of hydrogen. In the former case, high pressures (700 bar to reach a volumetric 75 concentration of 0.042 kg H2/L) or huge volumes are needed, and in the latter, high 76 energy consumption is unavoidable in order to maintain the required cryogenic 77 temperatures [6]. For these reasons, solid state hydrogen storage materials have been 78 intensively studied, as hydrates [7], metal-organic frameworks [8] or metallic and 79 chemical hydrides [9]. 80 Ammonia borane (AB) is considered as a promising chemical hydride, due to its high 81 hydrogen gravimetric capacity (19.6%wt H2) and volumetric capacity (140 g/L), 82 moderate decomposition temperature, non-toxicity [10] and stability at room 83 temperature, even in the range 50-60ºC, which is important from the point of view of 84 safety and engineering implications [11]. The thermal decomposition of neat AB 85 releases one mole of hydrogen per mole of AB in each of the following reactions (1), 86 (2) and (3) [12]. In the first decomposition step, AB releases H2 forming a complex 87 polymeric aminoborane (PAB) below its melting point (114 ºC): 88 BH3NH3 → BH2NH2 + H2 T > 120ºC (1) 89 Thereafter, PAB decomposes above 120 ºC, forming polymeric iminoborane and small 90 fractions of undesirable volatile byproducts: 91 BH2NH2 → 1/3 (BHNH)3 + H2 T > 120ºC (2) 92 (BHNH) → BN + H2 T > 500ºC (3) 93 Nevertheless, the use of this compound as hydrogen storage material also faces some 94 5 important limitations. One of them is the kinetic limitation due to the long induction 95 time needed to disrupt the dihydrogen bonding and initiate the release of hydrogen. 96 Furthermore, AB is difficult to regenerate: after thermolysis, it is not possible to restore 97 the initial AB by direct hydrogenation, and complex chemical regeneration routes 98 comprising several steps are needed. Additionally, it forms foams during thermolysis 99 that also complicate the regeneration due to the disruption of the physical structure of 100 the material. Moreover, during the decomposition process, the emission of some volatile 101 byproducts as borazine, diborane or ammonia can be released which could be poisonous 102 for downstream processes and, particularly, hydrogen fuel cells. 103 Several strategies have been tested to overcome these barriers, including the addition of 104 catalysts as silicon (Si), nickel (Ni), ruthenium (Ru), palladium (Pd) or zinc (Zn) [13105 16], confinement of ammonia borane into porous solid supports [17-21], dissolution of 106 AB in ionic liquids [22,23] or using polymers composites [24]. Regarding confinement 107 of AB, different supports have been tested: silica scaffolds as SBA-15 or MCM-48 [18], 108 metal organic frameworks (MOF) [16, 25] or carbon based materials [17] among others. 109 The amount of AB that has been successfully loaded in the support depends on their 110 structural properties (SBET and Vpores), getting a maximum amount of 50%wt using silica 111 scaffolds [17]. In all the cases that have been reported, hydrogen kinetic and 112 thermodynamic properties have been improved respect to neat hydride due to the 113 reduction in mean size. In our previous work [26], microparticles of silica aerogel were 114 used as support, getting a maximum concentration of 5%wt AB loaded in hydrophobic 115 silica aerogel. Liquid antisolvent technique was used to precipitate the hydride prior 116 supercritical carbon dioxide drying, resulting in enhanced hydrogen release kinetic 117 compared to pure AB. 118 In this work, we report the confinement of AB using silica aerogel as porous host. The 119 6 aerogel has been produced by liquid or supercritical CO2 drying, a technique that 120 enables to produce a silica material with a high pore volume, and a correspondingly 121 high potential capacity for storage of ammonia borane inside its pores. Several samples 122 with different concentrations of AB up to 60 wt% have been prepared in order to 123 analyze the influence of the loading of AB on the thermal and structural properties of 124 the material. Scanning electron microscopy, N2 adsorption isotherms, FT-IR 125 spectroscopy and X-ray diffraction have been used in order to characterize the final 126 product prior to the measurement of hydrogen release kinetics by decomposition at 127 80ºC. 128 129 2. Experimental methods 130 2.1 Materials 131 Tetramethylorthosilicate (TMOS, 98.0% purity), ammonium hydroxide (NH4OH, 28.0132 30.0% ammonia purity) and ammonia borane (AB, 97% purity) were supplied by 133 Sigma-Aldrich. Figure 1 shows a micrograph of neat AB as received. As shown in this 134 Figure, AB was constituted by agglomerated particles with sizes in the range of 100 m 135 and with a porous structure. Methanol (MeOH; 99.8% purity), n-hexane (95% purity) 136 and dry tetrahydrofuran (THF; with maximum water of 0.0075wt %) were purchased 137 from Panreac. Carbon dioxide (CO2, 99.95% purity) was supplied by Carburos 138 Metálicos S.A. 139 (FIGURE 1) 140 2.2 Preparation of AB loaded in silica aerogel microparticles 141 As presented in Figure 2, the procedure for the preparation of AB-loaded silica aerogel 142 microparticles consists of three key steps: preparation of silica gel microparticles, 143 7 addition of ammonia borane by a wet impregnation method, and drying of the AB144 loaded gel particles with pressurized carbon dioxide in order to produce the final, dry 145 AB-loaded aerogel microparticles. 146 (FIGURE 2) 147 In the first step of this procedure, hydrophilic silica alcogel was prepared using the well148 known method of hydrolysis condensation sol-gel reaction, using TMOS as precursor 149 and methanol as solvent. While with this procedure it is common to prepare large gel 150 monoliths, in this work the gel was synthetized as microparticles, in order to reduce the 151 possible heat and mass transfer resistances that could be caused by larger aerogel 152 monoliths. To do this, and according to the procedure described in a previous work [26], 153 the sol-gel reaction media was dispersed in hexane under mechanical stirring, in order to 154 obtain small droplets of TMOS in methanol dispersed within the hexane continuous 155 phase. After 10 minutes of mechanical stirring of this mixture with a two bladed axial 156 stirrer set at 600 rpm, an aqueous solution of NH4OH was added as condensation 157 catalyst, which induced the gelation of TMOS. The molar ratio used was the following: 158 1 mol TMOS: 4.4 mol MeOH: 3.3 mol H2O: 4.5 mol hexane: 0.08 mol NH4OH. As 159 methanol or water produced during the condensation reaction can induce the 160 decomposition of ammonia borane during the subsequent drying processes [26], after 2 161 hours of gelation the alcogel microparticles were retrieved and immersed in THF. Gel 162 particles were then kept during 7 days immersed in THF in a closed vessel, to let the gel 163 age and strengthen its structure. During this ageing period, the THF solvent was 164 renewed at least twice in order to remove the last traces of methanol and water. 165 After the ageing process, microparticles of alcogel are ready for wet impregnation, 166 adding a solution of ammonia borane dissolved in THF. This method has the advantage 167 that impregnation can be performed under milder temperature conditions compared to 168 8 melt infiltration, and only one impregnation step is necessary in contrast to incipient 169 impregnation methods [27]. Again, THF was used instead of methanol as solvent in 170 order to avoid methanolysis and therefore the decomposition process of the hydride not 171 only during wet impregnation (due to SiOH groups) but also during drying process [26]. 172 Different samples with different concentrations of ammonia borane were prepared, 173 adding different amounts of hydride (0-0.4g AB dissolved in 5mL of THF) to 2 g of 174 microparticles of alcogel (gel before drying, therefore with the pores filled with the 175 organic solvent) in order to study his influence on the properties of the final solid 176 product. With this, concentrations of AB in the final product ranging from 10 to 60 wt% 177 AB were obtained, where the concentration of AB is defined as presented in equation 178 (4): 179 (4) 180 The third and last step is the removal of the organic solvent in order to obtain the final, 181 dry AB/SiO2 particles. The drying method employed is a key aspect that determines the 182 textural properties of the porous support. If the solvent is removed by evaporation or 183 lyophilization, the capillary stresses associated to the formation of vapor-liquid 184 interfaces inside the pores of the support cause fractures and a partial collapse of the 185 pore structure of the material. In the case of SiO2 matrixes, the materials obtained by 186 these drying methods usually show pore volumes below 0.5 – 1.0 cm3/g. Some 187 examples are the well-known SBA-15 or MCM-41 mesoporous silica matrixes. In 188 contrast, if pressurized or supercritical carbon dioxide is used to extract the solvent, the 189 collapse of the pore structure is minimized, because under these conditions carbon 190 dioxide can be completely miscible with the organic solvent, and therefore the 191 extraction proceeds without formation of gas-liquid interfaces and without capillary 192 9 stresses. Due to this enhanced preservation of the pore structure, with this method it is 193 possible to reach pore volumes in the range 2 – 4 cm3/g [28]. Additionally, during this 194 drying process CO2 can act as antisolvent for solutes dissolved in the organic solvent, as 195 it is completely miscible with the organic solvent, but it cannot dissolve high-molecular 196 weight solutes dissolved in the organic solvent. This precipitation method is commonly 197 referred in the literature as “Gas Anti Solvent” (GAS) or “Supercritical Anti Solvent” 198 (SAS) precipitation [29,30]. In the case of this work, as ammonia borane is insoluble in 199 CO2 [26], when the THF-immersed alcogels are mixed with CO2, AB dissolved in the 200 THF that fills the pores of the alcogels quickly precipitates according to a GAS 201 precipitation mechanism, thus favoring the formation of small particles within the pores 202 of the gels. 203 In a previous work, it was observed that due the interaction of AB with the SiO2 matrix, 204 the temperature needed to initiate the decomposition of AB is drastically decreased [26]. 205 Thus, in order to avoid the thermolysis of AB, the drying process was carried out at a 206 near-ambient temperature of 25ºC and at a pressure of 100 bar, thus employing 207 pressurized, liquid CO2. As in the previous work [26], a batch drying apparatus, 208 depicted in Figure 2, was used. Using this apparatus, the alcogels immersed in THF 209 were loaded into the extraction vessel and the system was pressurized with CO2 using an 210 air-driven piston pump. The system was slowly pressurized at a rate of 0.5 bar/min in 211 order to avoid breakages in the alcogel/aerogel and mechanical stresses that could 212 damage the structural properties of the final product. Once the desired pressure and 213 temperature were reached, the recirculation pump that connects the extraction vessel 214 with the CO2 reservoir was connected, thus bubbling CO2 through the THF-immersed 215 alcogels and enhancing the extraction of the solvent. As CO2 gradually becomes 216 saturated with the solvent during the extraction process, CO2 in the system was renewed 217 16 the mapping of samples with higher concentration of AB, up to the maximum 359 concentration of 60 wt% tested in this work. 360 (FIGURE 5) 361 3.2.2 XRD patterns 362 Figure 6 shows the XRD pattern of neat ammonia borane, which matches well with 363 JCPDS reference 01-074-0894 suggesting the typical polycrystalline structure with 364 tetragonal lattice symmetry, in agreement with literature information about the structure 365 of AB at ambient temperature [42]. The crystallite size estimated using the Scherrer’s 366 equation formula is 40 nm, and the dominant sharp peak of the pattern is located at 367 23.75º, corresponding to (110) planes. After recrystallization by GAS process, slight 368 modifications in the XRD pattern can be observed at 2 = 17 and 30º. A similar 369 modification can be observed in silica-loaded AB samples, although with less defined 370 peaks due to the strong signal produced by the silica support. This modification in the 371 pattern can suggest the formation of diammoniate of diborane (DADB), an isomer of 372 AB. This compound shows some differences regarding the hydrogen release mechanism 373 compared to its isomer AB: the temperature for decomposition of DADB is about 10ºC 374 lower than that of AB, and DADB undergoes solid-phase decomposition without 375 melting or induction period even at moderate temperature, while AB suffers from a long 376 induction period prior to H2 release [43]. Therefore, the formation of this compound 377 may justify some of the observed thermal properties, as it will be discussed in the 378 following sections. 379 Nevertheless, in GAS recrystallized samples, the dominant pattern is equivalent to that 380 of neat AB, also corresponding to tetragonal crystal structure is observed, although 381 diffraction peaks are not so well defined, and the estimated crystallite size increases to 382 17 75 – 150 nm. In the case of AB loaded in silica aerogel, the peaks are merged, therefore 383 suggesting a reduction of crystallinity or an increased inhomogeneity in the properties 384 of the crystals due to the incorporation in the pores [44]. Additionally, as shown in 385 Figure 5b it can be seen that the XRD pattern of amorphous silica, characterized by a 386 broad peak around 24º, is overlaid to the dominant peaks related to AB. Therefore 387 results show that AB retains its crystalline structure after nanoconfination within the 388 pores of the aerogel, with estimated crystallite sizes in the range 100 – 200 nm, similar 389 to those obtained by recrystallization of AB by GAS process. 390 (FIGURE 6) 391 3.3 Thermal characterization of AB-loaded silica aerogels 392 Figure 7 shows the differential scanning calorimetry (DSC) traces of neat and 393 recrystallized AB compared to samples in which the hydride is loaded in silica aerogel. 394 In the case of the curve of neat AB, a sharp endothermic peak is observed whose onset 395 temperature (Ton=108.5 ºC) and peak temperature (Tp=110.8ºC) is dramatically reduced 396 when AB is loaded in silica aerogel (see table 2). This first peak is associated to the 397 melting point [14] or the dissociation of the intermolecular hydrogen bonding [18]. The 398 reduction or elimination in this first peak suggests that the degree of hydrogen bond in 399 the samples in which AB is embedded in silica aerogel is decreased, favoring the 400 reduction of the induction time. A similar result has been observed when AB was 401 embedded in other silica supports [18]. Regarding recrystallized AB, DSC results also 402 show reduction in the onset and peak endothermic temperatures, maintaining the shape 403 of the curve of neat AB. In this case, variations in characteristic temperatures of the 404 DSC traces can be associated to the reduction in the mean particle size achieved by 405 recrystallization of AB. As described by Varin et al [45], a reduction of particle size 406 into the subicrometric or nanometric scale is generally associated to a reduction of the 407 18 onset and peak temperatures of hydrogen evolution thermal events, due to the 408 destabilization of the material induced by the increased particle surface. Additionally, 409 the reduction of onset temperatures and induction time can be associated to the 410 formation of DADB by recrystallization suggested by XRD assays. 411 Similar results have been obtained in other works where AB is confined in different 412 supports. In the case of SBA and MCM silica supports, reductions in the onset and peak 413 temperatures to 48ºC and 100ºC have been reported [18], but as previously described 414 lower temperatures have been obtained in this work. This fact can be due to the higher 415 volume of pores of aerogel support that avoids the agglomeration of AB in meso416 channels. Therefore higher contact between the particle and the surface of the silica 417 aerogel takes place, enhancing the influence of silica surface groups on the 418 decomposition mechanism. In experiments with MOFs [42] or carbon cryogels [17, 20] 419 as supports, similar modifications in the thermal response of the material have been 420 reported. However, Srinivas et al. [16] observed a reduction of 30ºC in the onset and 421 peak temperatures of decomposition using MOFs, whereas in our case a displacement of 422 almost 70ºC is obtained. 423 Moreover, DSC results indicate a reduction of the exothermic enthalpy associated with 424 hydrogen release as the proportion silica/AB is increased. The measured enthalpy of 425 reaction for H2 release from neat AB is -24.9 kJ/mol AB, which is in good agreement 426 with results reported in literature. However, when the proportion silica/AB is increased, 427 the corresponding enthalpy of reaction changes to -10.9 kJ/mol AB for the sample with 428 60% of AB, and -5.5 kJ/mol AB for the sample with 30% of AB. A reduction of the 429 enthalpy of reaction for H2 release from AB was also observed by Gutowska et al. [40] 430 in their studies of incorporation of AB in mesoporous SBA-15 silica scaffold. However, 431 these authors report a more drastic variation of the enthalpy to -1.0 kJ/mol AB. These 432 19 authors indicate that the reason for the reduced exothermicity is the suppression of the 433 formation of boron compounds as byproducts of the PAB that is the main 434 decomposition product of AB according to reaction 1, which had the favorable 435 consequence of reducing the production of gaseous byproducts. 436 (TABLE 2) 437 (FIGURE 7) 438 This hypothesis agrees well with the results obtained in this work by TGA assays. 439 Figure 8 shows the results of TGA analysis performed on neat AB and AB-loaded silica 440 aerogel. In the case of neat AB, two important weight loss steps, which correspond to 441 the decomposition of the hydride, are observed: the first one till 129ºC corresponds to a 442 weight loss of 12.7 wt% and the second one, which finishes at 213ºC, corresponds to a 443 weight loss of 27.5 wt%. In comparison, and in agreement with the results of DSC 444 assays, TGA analysis shows that the AB confined in silica aerogel initiates its 445 decomposition at lower temperatures. It is also noticeable that in this case, weight loss 446 is not confined to sharp steps at defined temperatures, but it proceeds continuously over 447 the temperature range studied. In particular, at temperatures above 200ºC, where as 448 previously discussed neat AB does not experience any additional weight losses, a 449 continuous weight loss is still observed in the case of AB confined in aerogel. This 450 result indicates that the third step of the thermal decomposition mechanism, indicated in 451 reaction (3), also begins at lower temperatures in the case of confined AB compared to 452 neat AB. Moreover, as shown in Table 2, in both cases the total weight loss per unit 453 mass of AB in the sample is significantly larger than the maximum amount of hydrogen 454 that can stored in the compound. Similar results have been reported in [18], suggesting 455 that when AB is heated to high temperatures above 200ºC, other gases apart from 456 hydrogen are produced, as borazine, diborane, ammonia, etc. However, analyzing the 457 20 results reported in Table 2, it is noticeable that the total amount of volatile compounds 458 produced by heating up to 300ºC is reduced when AB is confined in silica aerogel. This 459 result, together with the variations in DSC assays of confined AB presented before, 460 suggest that interactions between AB and the silica support are taking place that 461 influence the decomposition mechanism of AB. Such interactions were suggested to 462 happen between AB and hydroxyl groups from the silica surface of the host by Lai et al. 463 [18]. These groups can interact with the BH3 group, loosening the covalent bond 464 between BH3 and NH3 groups of AB, thus destabilizing and promoting the 465 decomposition of the compound. Furthermore, by this interaction BH3 is kept bound to 466 the scaffold reducing the production of borazine and precluding the formation of 467 poliiminoborane 468 (FIGURE 8) 469 In figure 9, FTIR of neat AB and AB loaded in silica aerogel is shown before and after 470 dehydrogenation at 80ºC. It is observed that most of the peaks at frequencies related to 471 N-H and B-H bonds are broadened, shifted and decreased of intensity which indicates 472 the disruption of the bonds due to the release of hydrogen [13] in both samples. The 473 same behavior is observed for all the concentrations of AB loaded in silica aerogel, 474 although it is more pronounced at higher concentrations of AB. B-N band in the range 475 700-900 cm-1, which is observed in all the samples, is weakened but is still detected 476 after dehydrogenation; this fact clarifies that B-N is not disrupted and ammonia 477 formation is avoided during the decomposition [14]. On the other hand, in the sample in 478 which AB is loaded, the bonds related to silica are present without any change after 479 thermal treatment due to its stability at these conditions. 480 (FIGURE 9) 481 21 Regarding the crystallinity of the samples after thermal dehydrogenation, figure 10 482 shows XRD analyses of byproduct after isothermal dehydrogenation at 80ºC. According 483 to ICDD 00-019-0418, 2θ= 20.1º, 23.6º and 41.1º are assigned to amorphous PAB 484 (NH2BH2)5 [46]. In the case of AB/SiO2 samples, amorphous silica peak is present apart 485 from amorphous PAB byproduct. 486 (FIGURE 10) 487 3.4 Kinetics of hydrogen release by thermolysis at 80ºC 488 Figure 11 shows the kinetics of hydrogen release by thermolysis at 80ºC of neat AB 489 compared with AB confinement in silica aerogel. Results in this figure are normalized 490 reporting the amount of hydrogen released by unit mass of AB in the sample. Due to the 491 design of the cell used to measure kinetics, it was not possible to analyze samples of the 492 gas evolved during themolysis. However, it is assumed that at this temperature, the gas 493 which is released is H2 [12, 19, 41, 47] and no other volatile gases are present in the gas 494 stream in neat AB nor confined in silica aerogel. Regarding the shape of the curve in 495 neat AB, it follows a sigmoidal kinetic, typical of nucleation and growth pathway, with 496 a long induction time of more than 2 h, as reported in previous works [48]. After 497 confinement in silica, the induction time is reduced significantly. This means that the 498 silica could act as a catalyst (SiOH groups) [40] creating defect sites in the support that 499 initiate the decomposition at lower temperature. On the other hand, at lower 500 concentration of AB in the solution (prior to the drying), the hydride could precipitate in 501 the pores of the silica with lower mean size as was also suggested with XRD analyses 502 and corroborated in BET results. This suggests that smaller particles have more contact 503 with the surface of the silica and an easier way to form different bonds between the two 504 free pair of electrons of O in the Lewis base of Si-O-Si or Si-OH bond from the silica 505 22 with BH3 or NH3 from AB. As result, the intermolecular hydrogen bond is reduced, 506 following the mechanism proposed by Lai et al. [18]. 507 In the case of neat AB, 2 hours are needed to start releasing H2, and more than 4 hours 508 to get half of its content in H2 at this temperature (0.025gH2/gAB) whereas only it takes 509 22 minutes in the case of 13%AB loaded in silica aerogel. This fact shows the 510 improvement of silica aerogel as support for this chemical hydride system. 511 Regarding to recrystallized sample, it can be observed that the release profile maintains 512 the sigmoidal shape characteristic of neat AB, with a slower hydrogen release than 513 samples loaded in silica aerogel during the first 30 min of thermolysis. However, 514 compared with neat AB, with the particle size reduction achieved by GAS 515 recrystallization the induction time is drastically reduced and the release of hydrogen is 516 accelerated, to the point that after one hour an equivalent amount of hydrogen is 517 released from GAS-recrystallized AB as from AB loaded silica aerogel with a 30 wt% 518 of AB. 519 (FIGURE 11) 520 As a complement to Figure 11, Table 3 presents the total hydrogen release per unit mass 521 of solid product (AB + silica aerogel support). Results in this table clearly indicate the 522 weight penalty caused by the use of silica aerogel as porous host, as this material does 523 not contribute to the hydrogen storage capacity, thus reducing the total gravimetric 524 capacity of the material [27]. However, it can be seen that this disadvantage is 525 counterbalanced by a faster hydrogen release during the first 1-2 h of thermolysis. 526 (TABLE 3) 527 Moreover, there is a significant visual change in the morphology of neat AB in contrast 528 to AB encapsulated in silica aerogel. Figure 12 shows the different result of AB after 529 23 thermal decomposition at 80ºC. In the case of neat AB, foaming process takes place at 530 the same time that H2 is released from the hydride. Thus, even if the material was 531 micronized before thermolysis, this morphology and its associated advantages are 532 completely lost during the thermolysis and therefore in possible future hydrogen cycles, 533 if the material is regenerated [49]. However, when AB is encapsulated in silica aerogel, 534 this process is avoided obtaining particles with the same physical appearance. To 535 confirm this observation, Figure 13 presents SEM/EDX micrographs of the 536 60%AB/SiO2 sample after thermolysis. As shown in this figure, the original 537 morphology of the material is preserved after the thermolysis. Furthermore, as indicated 538 by the results of nitrogen mapping, the decomposition products of AB remain 539 homogeneously dispersed within the SiO2 aerogel matrix. Therefore it can be concluded 540 that by incorporation of AB inside the aerogel the morphological variations of the 541 material due to foaming are avoided. 542 (FIGURE 12) 543 (FIGURE 13) 544 4. Conclusions 545 Ammonia Borane has been recrystallized and nanoconfined inside the pores of silica 546 aerogel by a novel process, based on a simultaneous aerogel drying and ammonia 547 borane gas antisolvent precipitation using compressed carbon dioxide. Due to the 548 favorable textural properties of the aerogel materials obtained with this method, it has 549 been possible to load aerogels with up to 60 wt% of ammonia borane, without blocking 550 of pores and with a homogeneous dispersion of ammonia borane within the aerogel. By 551 analysis of the thermolysis process, it has been observed that by nanoconfinement the 552 temperature required to initiate the thermolysis process is reduced and the release 553 kinetics are accelerated as they do not show induction time. Furthermore, by 554 24 nanoconfination of AB, the morphological properties of the material are stabilized and 555 foaming is eliminated, which could be favorable properties for a subsequent material 556 regeneration process. The feasibility of implementation of the multi-step chemical 557 process required to re-hydrogenate AB when this compound is embedded in the silica 558 matrix remains to be tested. 559 560 Supplementary Information 561 Video 1: Thermolysis of neat ammonia borane at 80ºC. Time is accelerated by a factor 562 of 64. 563 Video 2: Thermolysis of 60%AB/SiO2 sample at 80ºC. Time is accelerated by a factor 564 of 64. 565 566 Acknowledgements 567 This research has been financed by the Spanish Ministry of Economy and 568 Competitiveness through project ENE2011-24547. Á. Martín thanks the Spanish 569 Ministry of Economy and Competitiveness for a Ramón y Cajal research fellowship. M. 570 Rueda thanks the University of Valladolid for a FPI predoctoral grant. L. M. Sanz 571 thanks the Spanish Ministry of Economy and Competitiveness for a FPI predoctoral 572 grant. 573 References 574 [1] M.Balat, Int. J. Hydrogen Energ. 33 (2008) 4013-4029 575 [2] M.Hook, X.Tang, Energ. Policy 52 (2013) 797-809 576 [3] S. A. Sherif, F. Barbir, T. N. Veziroglu, Electricity J. 18 (2005) 62-76 577 [4] J. Graetz, Chem. Soc. Rev. 38 (2009) 73-82 578 25 [5] DOE: US Department of Energy. Web site: http://www.doe.gov. 579 [6] L. O. Williams, D. E. Spond, Appl. Energ. 6 (1980) 99-112. 580 [7] H. P. Veluswamy, R. Kumar, P. Linga, Appl. Energ.122 (2014) 112-132. 581 [8] M. Paik Suh, H. J. Park, T. K. Prasad, D. Lim, Chem. Rev. 112 (2012) 782-835 582 [9] B.Sakintuna, F.Lamari-Darkrim, M.Hirscher, Int. J. Hydrogen Energ. 32 (2007) 583 1121-1140 584 [10] B. Peng, J. Chen, Energ. Environ. Sci. 1 (2008) 479-483 585 [11] S. D. Rassat, C. L. Aardahl, T. Autrey, R. S. Smith, Energ. Fuel. 24 (2010) 2596586 2606 587 [12] G. Wolf, J. Baumann, F. Baitalow, F. P. Hoffmann, Thermochim. Acta 343 (2000) 588 19-25 589 [13] D. Kumar, H. A. Mangalvedekar, S. K. Mahajan, Mater. Renew. Sustain. Energy 3 590 (2014) 23 591 [14] A. C. Gangal, P. Kale, R. Edla, J. Manna, P. Sharma, Int. J. Hydrogen Energ. 37 592 (2012) 6741-6748 593 [15] B. Roy, J. Manna, P. Sharma, J. Alloy Compd. 645 (2015) S234-S238. 594 [16] G. Srinivas, J. Ford, W. Zhou, T. Yildirim, Int. J. Hydrogen Energ. 37 (2012) 595 3633-3638 596 [17] A. Feaver, S. Sepehri, P. Shamberger, A. Stowe, T. Autrey, G. Cao, J. Phys. Chem. 597 B 111, (2007) 7469-7472 598 [18] S-W Lai, H-L. Lin, T.L.Yu, L-P. Lee, B-J. Weng, Int. J. Hydrogen Energ. 37 599 (2012) 14393-14404 600 [19] T.Autrey, A.Gutowska, L.Li, J.Linehan, M.Gutowski, Preprints of Papers601 American Chemical Society, Division of Fuel Chemistry 49 (2004) 150-151 602 [20] S.Sepehri, B.B.Garcia, G.Cao, J. Mater. Chem. 18 (2008) 4034603 32 Table 2. - Temperature data and weight losses of neat and recrystallized AB vs loaded 699 in silica aerogel with different concentration obtained from DSC and TGA analyses 700 respectively. 701 702 Sample Ton1 (ºC) Tp1 (ºC) Ton2 (ºC) Tp2 (ºC) Total wt loss per wt AB Neat AB 108.5 110.8 113.9 114.1 40.2 Recrystallized AB 70.5 77.9 102.4 113.4 - 60% AB/SiO2 37.6 62.3 84.8 108.9 33.0 30% AB/SiO2 39.5 68.7 83.1 99.0 - 33 Table 3. - Amount of H2 released at different times by isothermal thermolysis at 80ºC in 703 neat and recrystallized vs. AB loaded in silica aerogel with different concentration 704 15min 30min 45min 1h 2h final Sample mgH2/gtotal mgH2/gtotal mgH2/gtotal mgH2/gtotal mgH2/gtotal mgH2/gtotal neat AB 0 0 0 0 2 52 AB recrystallized 1,66 7,45 20,19 30,45 43,98 52 13%AB/SiO2 2,26 4,28 5,28 5,80 6,38 6,76 30%AB/SiO2 1,69 4,43 7,30 9,07 12,31 14,56 60%AB/SiO2 2,48 9,42 17,79 20,65 27,27 32,24 705 34 Figures 706 707 708 Figure 1. - SEM micrograph of neat ammonia borane as received 709 710 35 1. Synthesis alcogel 3. L-CO2 drying (100bar/25ºC) 2. Addition of AB solution PI TI co2 PI Liquid vent Recirculation pump CO2 buffer Extractor CO2 pump Hot air oven 711 Figure 2. - Steps for preparation of silica aerogel microparticles loaded with ammonia 712 borane. 713 714 36 715 Figure 3. FTIR spectra of a) silica aerogel b) neat AB c) recrystallized AB 716 d)30%AB/SiO2 e)60%AB/SiO2. Curves are vertically displaced for clarity. 717 718 37 719 Figure 4.-Nitrogen adsorption-desorption isotherms and BJH pore size distribution of 720 silica aerogel and AB loaded in silica with different concentration ( ) adsorption ( ) 721 desorption 722 723 38 724 Figure 5.- SEM images of A) Recrystallized AB after recrystallization in THF using 725 liquid CO2 as drying method B) 30%AB loaded in silica aerogel and mapping of sample 726 B (blue is referred to silica and red to Nitrogen) 727 728 39 729 Figure 6. –a) XRD of neat ammonia borane, recrystallized ammonia borane and AB 730 loaded in silica aerogel with different concentrations b) Amplification of XRD signal 731 showing the characteristic pattern of silica aerogel. Curves are vertically displaced for 732 clarity 733 734 40 735 Figure 7. - DSC curves of AB and AB loaded in silica aerogel with different 736 concentration. The curves are normalized according to the weight of AB, and vertically 737 displaced for clarity. 738 739 41 740 Figure 8. - TGA curves of AB and AB loaded in silica aerogel 741 742