Influence of Nanoconfinement on the Hydrogen Release Processes from Sodium Alanate

Sodium alanate (NaAlH4) is a prospective H2 storage material for stationary and mobile applications, as NaAlH4 contains 7.4 wt% of H2, and it is possible to do multiple H2 release and accumulation cycles. Nanoconfinement is a potential solution to enhance the H2 release properties of NaAlH4. To optimize the supporting material and the synthesis method used for the nanoconfinement of NaAlH4, a better understanding of the influence of nanoconfinement on the H2 release processes is necessary. Thus, the H2 release from bulk, purely nanoconfined, and intermediate NaAlH4 is measured at different temperature ramp rates, and the characteristic parameters for each hydrogen release process are determined. Activation energies for each process are determined using the Kissinger method, and the effect of nanoconfinement on the activation energies is analysed. The impact of nanoconfinement on the H2 release processes from NaAlH4 and the limitations of each process in case of bulk and nanoconfined NaAlH4 are presented and discussed. Nanoconfinement of NaAlH4 decreases activation energies of the initial reversible H2 release steps to between 30 and 45 kJ mol−1 and increased the activation energy of the last irreversible H2 release step to over 210 kJ mol−1.


Introduction
Hydrogen is a potential energy carrier, as H 2 can be produced directly from water through electrolysis, H 2 can be directly converted into electrical energy with up to 60% efficiency with fuel cells, and the only by-product during conversion into electricity is water [1,2]. Thus, using H 2 to accumulate cheap electricity from renewable sources during high winds and/or intense solar radiation would help balance the power grid and be a viable alternative to electric battery cars. One of the hindrances to the application of a hydrogen-based energy economy is the lack of a cheap and safe hydrogen storage technology. The current commercial standard for hydrogen storage in mobile applications is storing hydrogen in pressurized tanks of up to 700 bar [3]. As the compression of H 2 to such pressures, p, consumes at least 10% [4] of the stored energy, such high pressures have inherent hazards, and even then, the H 2 energy density is only 1.31 kWh L −1 at 25 • C, alternative H 2 storage methods are of high interest.
The storage of hydrogen in a chemically bound form inside complex metal hydrides offers the advantage of high H 2 density, storage at ambient temperature, T, and p H2 = 1 bar, and potential cyclability through the application of p H2 [5]. One of the most promising complex metal hydrides for H 2 storage is sodium alanate (NaAlH 4 ), which incorporates 7.4 wt% of H 2 . The release of H 2 from the bulk phase can be described through three decomposition steps (Equations (1)- (3)). The thermodynamical equilibrium T of decomposition (i.e., the release of H 2 ) for the first two decomposition reactions (Equations (1) and (2)) of NaAlH 4 are at comparatively low temperatures (80 and 150 • C, respectively) [6]. Alas, the release of H 2 from the bulk phase is kinetically limited in the solid phase, and thus, the first H 2 release step from bulk NaAlH 4 starts at 183 • C (Equation (1)) with the melting of NaAlH 4 . H 2 is released (from Equation (1) to Equation (3)) over a wide T range ( Figure 1 and Reference [7]). The third H 2 release step (Equation (3)) is of no use for H 2 storage as the reverse reaction has not been achieved at p H2 and T feasible for current gas storage systems. Thus, the main problems of a NaAlH 4 -based H 2 storage system are the high temperature of H 2 release caused by the kinetic limitation, the irreversibility of the third decomposition step (Equation (3)), and the segregation of decomposition products (Al, NaH, and Na 3 AlH 6 ) into separate phases, which decreases the cycling lifetime of NaAlH 4 -based H 2 storage systems. 3NaAlH 4 (l) → Na 3 AlH 6 (s) + 2Al(s) + 3H 2 ↑ T start = 183 • C (1) Na 3 AlH 6 (s) → 3NaH(s) + Al(s) + 1 The T of H 2 release from NaAlH 4 can be lowered through nanoscaling, where the higher surface area of sub-micrometer particles promotes the release of H 2 from NaAlH 4 at a T closer to that of thermodynamical equilibrium through the elimination of kinetical limitation present in case of bulk material, i.e., diffusion of H 2 inside the bulk phase. There are two main methods for nanoscaling NaAlH 4 : reducing the average particle size through ball-milling or nanoconfining nanoscale particles unto a supporting porous material [7][8][9][10][11]. Whilst ball-milling is relatively easy to perform, and the T of H 2 release is considerably lowered during multiple hydrogenation/dehydrogenation cycles, the nanoscale particles agglomerate together at increased T, and the improved H 2 release properties are lost over multiple dehydrogenation/hydrogenation cycles [12]. By using a supporting porous material for nanoconfinement of NaAlH 4 , the starting T of H 2 release is lowered to ambient conditions [8,13]. In addition, the agglomeration of NaAlH 4 and the segregation of decomposition products during dehydrogenation into separate phases is limited by the confinement of NaAlH 4 , e.g., during dehydrogenation of bulk NaAlH 4, particles of Al form in the micrometer size. In contrast, the size of Al particles formed during the dehydrogenation of confined NaAlH 4 is lowered to tens of nanometers [13,14]. Nanoscale NaAlH 4 does not release H 2 according to Equations (1)-(3), where the direct decomposition of NaAlH 4 to NaH has been shown by Gao et al. [7], and the H 2 release curves are inherently different, whereas the H 2 released from different decomposition steps (Equations (1)-(3)) from nanoconfined NaAlH 4 are visually inseparable [7,8,13].
To fine-tune the H 2 release properties of nanoconfined NaAlH 4, it is essential to understand the energetical barrier associated with each H 2 release step. It has been shown by Baldé et al. [8] that the decrease in NaAlH 4 particle size decreases the activation energy, E a , of the decomposition process and is dependent on the particle size, where the E a decreases from 116 kJ mol −1 , in the case of bulk material, to 58 kJ mol −1 , in the case of nanoparticles from 2 to 10 nm.
In this work, the influence of nanoconfinement on the H 2 release processes from NaAlH 4 is presented in depth. Bulk NaAlH 4 , NaAlH 4 deposited as bulk material on a carbon support material, and completely nanoconfined NaAlH 4 materials are synthesized, and H 2 release curves at constant T ramp rates, β = dT/dt, are measured and analyzed. By comparing E a values of H 2 release from different phases of NaAlH 4 , i.e., bulk, deposited as bulk on a supporting material, and nanoconfined, a more complete picture of the effect of nanoconfinement on the H 2 release processes from NaAlH 4 is obtained and presented.

Synthesis of NaAlH 4 /Carbon Composites
NaAlH 4 is deposited on a high surface area, specific surface area = 1840 m 2 g −1 , microporous, most pores with widths under 2 nm, and high porosity, pore volume = 0.82 cm 3 g −1 , dry carbon material RP-20 (Kuraray, Japan) through the solution impregnation method. Tetrahydrofuran, THF (anhydrous, ≥99.9%, Sigma-Aldrich, Germany), is used as the solvent, and a 0.05 g NaAlH 4 mL THF −1 solution is used for all samples. Bulk NaAlH 4 (90%, Sigma-Aldrich, Germany) was dissolved and filtrated through a glass microfiber filter (GF/B, Whatman, UK) to remove any insoluble impurities. Samples with 5, 60, and 100 mass% of deposited NaAlH 4 are synthesized to yield the nanoconfined, bulkdeposited, and bulk NaAlH 4 , respectively, and are denominated as such. In the case of the bulk NaAlH 4 sample, recrystallization is performed at identical conditions to get rid of any impurities, but without the carbon support material. All syntheses, sample storage, and sample preparation for further analysis were performed in an Ar-filled (5.0, Linde) glovebox (MBraun LABmaster sp, Germany). A full description of the synthesis process and the characterization of the synthesized materials is provided in [13].

Temperature Controlled Decomposition
The H 2 release curves at constant β were measured with the Autochem 2950HP (Micromeritics, USA) chemisorption analyzer. The amount of released H 2 was determined with a thermal conductivity detector, where a 50 mL min −1 N 2 (6.0, Linde) gas flow was used as the carrier gas. β from 0.5 • C min −1 to 10 • C min −1 were applied. A full description of the sample preparation and measurement routine is brought in [13]. Data reduction, data analysis, and H 2 release peak fitting were performed with the OriginPro Version 2016 (OriginLab Corporation, Northampton, MA, USA) software.

Hydrogen Release from NaAlH 4 /Carbon Composites
The bulk, bulk-deposited, and nanoconfined NaAlH 4 exhibit distinct H 2 release curve shapes (Figure 1) in the case of all β used. In the case of the bulk NaAlH 4 , a low amount of H 2 is released at~170 • C, which is just under the T for melting of bulk NaAlH 4 . The release of H 2 at~170 • C is most likely the release of H 2 from the surface layer and defect sites, for which the melting T is lower, and thus, H 2 is released before the melting of the bulk phase. Starting from T = 200 • C, H 2 is released from the bulk of the material through the decomposition steps Equations (1) and (2), and at T > 280 • C H 2 is released from the last decomposition step, Equation (3). At T > 375 • C, all of the H 2 has been released as the material has been completely decomposed.
In the case of bulk-deposited NaAlH 4, a very low amount of H 2 is already released, starting at T < 90 • C. The T at which the initial H 2 release starts depends strongly on the β applied and is released at an almost constant rate. This H 2 release step is caused by the small amount of NaAlH 4 deposited in an amorphous state (highly polycrystalline structure) and/or as separate nanosized particles during the recrystallization process. Starting from T = 165 • C, a high amount of H 2 is released over a narrow T range and is followed by a high amount of released H 2 over a wide T range up to~200 • C, after which a low amount of H 2 is released at a constant rate up to T~375 • C. These three steps correspond to NaAlH 4 decomposition: Equation (1) from the surface, Equations (1) and (2) from bulk, and Equation (3), respectively. For a better deconvolution of the H 2 release from different decomposition steps and processes, the whole H 2 release curve has been fitted with distribution functions (presented in the next section).
In the case of nanoconfined NaAlH 4 H 2 release starts already at ambient conditions, T < 23 • C in case of all applied β. H 2 is mainly released at T < 200 • C, with the shape of the H 2 release curve strongly dependent on the applied β. In addition, a very low amount of H 2 is released at T > 400 • C, which is most likely H 2 released from the decomposition of NaH. Thus, deconvolution of the H 2 release peaks through fitting is necessary to understand the decomposition processes of nanoconfined NaAlH 4 .
The amount of H 2 released, compared to the theoretical amount of H 2 in the NaAlH 4 of the sample, yields the H 2 content efficiency of the investigated materials. The average H 2 content efficiencies are~100%,~66%, and~40% for bulk, bulk-deposited, and nanoconfined, respectively. Thus, the amount of recovered H 2 from NaAlH 4 decreases with the deposition process, especially with the nanoconfinement of NaAlH 4 . The decrease in H 2 content efficiency is likely caused by the decomposition of NaAlH 4 during synthesis, where the step-by-step addition of NaAlH 4 in THF solution to the carbon support promotes the decomposition of NaAlH 4 , especially at the initial addition steps where NaAlH 4 is deposited as nanosized particles. In addition, as the H 2 release from nanoconfined NaAlH 4 begins at ambient conditions, it is very likely that some of the nanoconfined NaAlH 4 decomposes during storage at ambient conditions, i.e., at 21 • C and in Ar gas environment.

Modelling of Decomposition Processes
To better understand the decomposition processes of NaAlH 4 in different phases, e.g., bulk, bulk-deposited, and nanoconfined, the H 2 release curves were fitted with a combination of gaussian and bigaussian peak functions ( Figure 2). The bigaussian, i.e., an asymmetric gaussian profile distribution peak function which has two independent peak widths, w 1 and w 2 , at x < x c (Equation (4)) and x ≥ x c (Equation (5)), respectively, where x c is the position of the peak center where y 0 is the baseline and H is the peak height. The bigaussian profile of the H 2 release is caused by the kinetic limitation of some decomposition steps, where the increase in the H 2 release amount with the initial rise in T is quick. After reaching maximum H 2 release from the process at temperature T max , the H 2 release process continues over a wide T range as H 2 release from the decomposition process is kinetically hindered. From the fitting of the H 2 release curves, T max is obtained for each decomposition process. The range of obtained T max , determined from H 2 release curves measured at different β, is presented in Table 1 with the corresponding designations for each process. It is possible to calculate the E a of a first-order reaction using the Kissinger method [8,15] where R is the ideal gas constant, and Z is the Arrhenius pre-exponential factor. By plotting ln(β/T max 2 ) vs. 1/T max, the slope will yield −E a /R, and thus, E a can be calculated. The E a values of the deconvoluted H 2 release processes have been calculated from the slope of the Kissinger plot ( Figure 3, Table 1). In the case of bulk NaAlH 4 , all the activation energies are relatively high, >100 kJ mol −1 , as the H 2 release processes are kinetically hindered by the diffusion of H 2 out of the bulk phase and by the thermal conductivity of the bulk phase, where the H 2 release processes from NaAlH 4 are endothermic [16], and thus preventively cooling the surrounding material upon decomposition. Even though the temperature of initial H 2 release from the first decomposition step depends on the applied constant β, the low amount of H 2 released hinders proper fitting. Thus, the obtained T max does not remarkably depend on applied constant β. Therefore, the E a could not be calculated and is most likely caused by the release of H 2 from the surface of NaAlH 4 particles over a wide range of T values under the melting T of bulk NaAlH 4 . The highest E a from the bulk material is for the second peak, which corresponds to Equation (1) reaction step from the bulk of the material. This H 2 release step is over a wide T range, as the H 2 diffusion length can vary widely in large NaAlH 4 particles, and thus H 2 release from inside the NaAlH 4 particles occurs at remarkably higher T values. Simultaneously with H 2 release from the Equation (1) reaction step, H 2 release from the Equation (2) reaction step starts. The H 2 release from the Equation (2) reaction step is discernably taking place in two distinct stages, as H 2 from the surface is released first, and then the H 2 from inside the particles is released. H 2 release from the last irreversible reaction step, Equation (3), is clearly separatable and has a relatively high E a , 150 kJ mol −1 , as the decomposition products from preliminary H 2 release steps limit the kinetics of H 2 release from this step. A similar case can be seen for bulk-deposited, but the T max of all H 2 release steps are lowered, and the E a values of the last two steps are also lower. This is most likely caused by the partial confinement of the decomposition products inside the porous carbon structure, enhancing the kinetics of the H 2 release step corresponding to Equation (2). This restruc-turing and the disappearance of clear crystalline phases upon the dehydrogenation of bulk-deposited NaAlH 4 has been shown before [10,13].
In the case of nanoconfined NaAlH 4 , three separate H 2 release steps are identified under one continuous H 2 release peak, starting from ambient T and continuing up to 200 • C, where the H 2 release step starting with T max from 51 to 57 • C is discernible only at a high applied constant β. All these H 2 release steps have relatively low E a values, compared to bulk and bulk-deposited, are overlapping, and do not clearly correspond to a discrete H 2 release reaction. Baldé et al. have shown that the T max of H 2 release decreases remarkably with the decrease in the nanoparticle size [8], and Gao et al. have shown that nanosized NaAlH 4 decomposes directly into NaH [7]. Thus, the indiscernibility of the H 2 release steps of nanoconfined NaAlH 4 is very likely caused by a distribution of differently sized NaAlH 4 nanoparticles, which decompose at different T values directly into NaH, going through the Equations (1) and (2) reactions in one step. In addition, the decomposition of the nanoconfined NaAlH 4 at ambient conditions, based on H 2 content efficiency, makes the assignment of the H 2 release peaks a nontrivial task. Regardless, the calculated E a values, from 30 to 45 kJ mol −1 , of H 2 release steps discernible at all applied constant β from nanoconfined NaAlH 4 are notably lower than the value of 58 kJ mol −1 achieved for 2-10 nm NaAlH 4 particles [8]. The H 2 release step detected at the highest T, most likely from the Equation (3) reaction step, releases a low amount of H 2, and the T max of this step is 100 • C higher than that of bulk and bulk-deposited materials and has an increased E a . Thus, it is possible that nanoconfined NaAlH 4 does not decompose completely to Na and Al under the investigated conditions, as, for some reason, nanosized NaH is more stable than bulk NaH. This would improve the cyclability and usability of nanoconfined-NaAlH 4 -based H 2 storage systems, as the decomposition of NaH is irreversible and, thus, unwelcome. The increased stability of the nanoconfined NaH phase must be investigated further.
The addition of fitting the H 2 release curves to deconvolute different decomposition steps before applying the Kissinger method has clearly shown that various activation barriers are present even in the case of nanoconfinement. This increases our understanding of the fundamental processes involved in releasing H 2 from NaAlH 4 , especially when nanoconfined, which may prove vital when improving and optimizing such materials for H 2 storage applications.
The achieved activation energies for the reversible steps are considerably lower than most activation energies of mixed hydride [17][18][19], doped hydride [20][21][22][23][24][25], and nanoconfined hydride [9,26,27] systems described in the literature. Those include systems that utilize NaAlH 4 [9,19,24,25,27], but also other metal and complex hydrides [17,18,[20][21][22][23]26]. Most activation energies for bulk hydrides are well over 100 kJ mol −1 . The achieved improved energies in the referred papers are mostly larger, by at least a factor of two, than the ones presented in this paper, with relatively few coming into the range of 30-45 kJ mol −1 [22,27]. This is remarkable, taking into account the simplicity of the system and its synthesis. Doping involves adding transition metals or compounds, which often adds a synthesis step of varying complexity. Mixtures may have side reactions, which are complex to analyze and may result in unwanted products or irreversible alloys [17][18][19]. Using a scaffold material for confinement improves cycling stability [13], which is essential for practical applications. In this work, the results have been achieved using a simple commercial microporous carbon, which offers lower E a values than laboratory-made materials [26,27] and potentially higher loadings than, e.g., metal-organic frameworks [9]. Doping, ball-milling, and melt-infiltration, which even occurs naturally upon cycling above the melting temperature of NaAlH 4 , is expected to improve the material's hydrogen storage properties even further.

Conclusions
The deconvolution of H 2 release processes through curve-fitting procedures to determine T max when applying the Kissinger method significantly improved the amount and quality of the information received about the H 2 release processes in bulk, bulk-deposited, and nanoconfined NaAlH 4 . The coating of porous carbon particles with a thick layer of NaAlH 4 , bulk-deposited, is shown to decrease the T max of all H 2 release steps, whereas only the E a values of the last steps are lowered. This indicates that whilst a supporting material reduces the T at which H 2 is released, the H 2 release processes are still limited by the H 2 diffusion out of the bulk phase. In the case of nanoconfined NaAlH 4 , H 2 release starts at ambient T, and the E a of H 2 release processes is remarkably lowered in comparison to bulk and bulk-deposited NaAlH 4 . None of the H 2 release steps are limited by H 2 diffusion in the case of nanoconfined NaAlH 4 . Even though the nanoconfined NaAlH 4 did not have a discrete particle size, based on the discernability of multiple H 2 release steps, very low E a values from 30 to 45 kJ mol −1 are determined. These E a values are a significant improvement on the vast majority of previously used techniques for nanoconfinement. The H 2 release at ambient T and the very low E a of the nanoconfined NaAlH 4 , in addition to limiting the final, irreversible decomposition step, highlights the possible utility of nanoconfinement for the improvement of NaAlH 4 -based H 2 storage materials, where the critical problem is synthesizing nanoconfined composites with a high total H 2 yield per mass and/or volume.
Funding: This research was supported by the EU through the European Regional Development Fund (Centers of Excellence, TK141 2014-2020.4.01.15-0011) and by the Estonian Research Council Grant, grant number PUTJD957.

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