Nanoconfined NaAlH₄: prolific effects from increased surface area and pore volume

Abstract

Nanoconfinement is a promising technique to improve the properties of nanomaterials such as the kinetics for hydrogen release and uptake and the stability during cycling. Here we present a systematic study of nanoconfined NaAlH₄ in nanoporous scaffolds with increasing surface area and pore volume and almost constant pore sizes in the range of 8 to 11 nm. A resorcinol formaldehyde carbon aerogel was CO₂-activated under different conditions and provided aerogels with BET surface areas of 704, 1267 and 2246 m² g⁻¹ and total pore volumes of 0.91, 1.30 and 2.21 mL g⁻¹, respectively. Nanoconfinement of NaAlH₄ was achieved by melt infiltration and ²⁷Al MAS NMR reveals that the respective scaffolds incorporate 68, 82 and 91 wt% NaAlH₄, for the above-mentioned samples, while the remaining fraction decomposes to metallic Al indicating that increasing CO₂-activation tends to facilitate the infiltration process. The frequencies for the ²³Na and ²⁷Al MAS NMR centerband resonances from NaAlH₄ vary systematically for the infiltrated samples and are shifted towards higher frequency and become more narrow with increasing degree of CO₂ activation of the scaffolds. This new effect is attributed to increasing interactions with conduction electrons from increasingly graphite-/graphene-like scaffolds. The bulk versus nanoconfined ratio of NaAlH₄ was investigated using Rietveld refinement, revealing that the majority of added NaAlH₄ is confined inside the nanopores. The hydrogen desorption kinetics decreased with increasing surface area and the hydrogen storage capacity is more stable and decreases less during continuous hydrogen release and uptake cycles. In fact, the available amount of hydrogen (2.7 wt% H₂) was more than doubled compared to the nanoconfinement in the non-activated carbon aerogel (1.3 wt% H₂). Furthermore, it was demonstrated that Ti-functionalization of the CO₂-activated aerogels combines the high storage capacity with fast hydrogen release kinetics from NaAlH₄ which fully decomposes into Na₃AlH₆ at T ≤ 100 °C.

---

1. Introduction

Hydrogen is recognized as a promising future carrier of renewable energy and may play an important role in an environmentally friendly carbon-free energy system.¹ Utilization of hydrogen as a substitute for gasoline has the potential to completely eliminate the harmful waste particles and gases from cars, buses and truck exhaust in cities. Hydrogen offers a clean and safe alternative to fossil fuels with water as the only exhaust, which may also reduce the climatic changes we are experiencing in the environment from the exponentially increasing amounts of carbon dioxide in the atmosphere.

Solid state hydrogen storage materials such as MgH₂, LiBH₄ and NaAlH₄ offer high gravimetric (ρ_m = 7.6, 18.5 and 7.5 wt% H₂, respectively) and volumetric (ρ_v = 110, 121 and 94 g H₂ L⁻¹) hydrogen storage capacities, which are required for a compact and efficient hydrogen storage system.^2,3 However, until now solid state hydrogen storage has suffered from difficulties in combining high storage capacities in light element hydrides with hydrogen release and uptake conditions close to room temperature which is technologically desirable. Therefore, substantial research efforts have been spent on discovering new methods to facilitate hydrogen exchange in these materials under near ambient conditions.^4–6 One approach is confinement of metal hydrides in inert mesoporous scaffolds with pore sizes in the range of 2–50 nm, referred to as nanoconfinement.^7–10 The metal hydride is usually embedded in the nanoscaffold by melt infiltration or solvent mediated infiltration. The final crystallite size of the hydride may depend on the pore size diameter of the scaffold.¹¹ Previous studies have illustrated that nanoconfinement of MgH₂, LiBH₄ and NaAlH₄ in carbon based scaffolds facilitates improved hydrogen release and uptake kinetics and reversible hydrogen storage capacity and more favorable thermodynamic properties.^12–22

Bulk NaAlH₄ starts to decompose to Na₃AlH₆(s), Al(s) and H₂(g) upon melting at T_mp ∼ 183 °C and Na₃AlH₆ further decomposes to NaH(s), Al(s) and H₂(g) at T > 240 °C.²³ Sodium alanate, NaAlH₄, spontaneously infiltrates a nanoporous scaffold using an elevated hydrogen pressure to minimize decomposition reactions.^15–17 However, the melt infiltration process is often ineffective and only half of the hydride is nanoconfined while the rest decomposes to Al and presumably NaH.¹⁵ In addition, the weight penalty of the scaffold also reduces the gravimetric hydrogen content of the nanocomposite material. This has motivated the present study of CO₂ activated resorcinol formaldehyde carbon aerogels with variable surface area and pore volume utilized as nanoscaffolds. Furthermore, a recently proposed catalytic effect from the aerogel surface is also explored.^11,24,25

2. Experimental details

2.1. Sample preparation

The resorcinol formaldehyde carbon aerogel (RF-CA) was prepared by mixing 41.618 g resorcinol (Aldrich, 99%), 56.9 mL formaldehyde in water (37 wt% stabilized by ∼10 to 15% methanol, Merck), 56.6 mL of deionised water and 0.65 g Na₂CO₃ (Aldrich, 99.99%) in a beaker with continuous stirring. The pH value of the final solution was 6.24. Afterwards, the preparation was performed according to previously published procedures.^13,15,26 Selected fractions of the prepared carbon aerogel (CA) were CO₂-activated in order to increase the surface area.^27,28 Monoliths of CA were placed in an Al₂O₃ crucible, transferred to a tubular oven and heated to 950 °C (heating rate, ΔT/Δt = 6.67 °C min⁻¹) in a CO₂ flow. The temperature was kept constant during selected time intervals of 3, 5 and 5.1 hours, and afterwards, the gas flow and the furnace were turned off and the samples were allowed to cool naturally. This approach provided activated CA with variable textures, denoted as CA-3, CA-5 and CA-5.1, respectively. Activated CA-5 and CA-5.1 were treated under similar conditions but were from different synthesis batches. Longer CO₂-activation time provides larger surface areas and pore volumes due to aerogel burn off, see Table 1.^27–29 Prior to use, all CAs were activated at 400 °C in vacuum for several hours in order to remove moisture and gases from the porous structure. All subsequent handling was performed in a purified argon atmosphere in a glove box.

Table 1 Texture parameters for the carbon aerogel scaffold, activated scaffolds by heating in a flow of CO₂ and the TiCl₃-functionalized scaffold measured prior to infiltration of the indicated amounts of NaAlH₄

Sample	CO2 – time (h)	S BET (m^2 g^−1)	V meso (mL g^−1)	V tot (mL g^−1)	V micro/Vtot	D max (nm)	NaAlH4^a (wt%)	NaAlH4^a (vol%)
a Amounts of NaAlH4 used for the infiltration per weight and volume of the scaffold. b The TiCl3 content was 8.7 wt%.
CA	0	704 ± 12	0.74 ± 0.02	0.91 ± 0.02	0.19	10 ± 0.5	34.1 ± 0.02	62 ± 2
CA-3	3	1267 ± 270	0.96 ± 0.16	1.30 ± 0.27	0.28	11 ± 0.5	41.5 ± 0.02	63 ± 13
CA-5	5	2246 ± 270	1.53 ± 0.16	2.21 ± 0.27	0.29	10 ± 0.5	51.3 ± 0.02	53 ± 7
CA-5.1	5.1	2236 ± 270	1.45 ± 0.16	2.11 ± 0.27	0.29	8 ± 0.5	56.5 ± 0.02	69 ± 9
CA–Ti^b	5.1	2041 ± 246	1.30 ± 0.14	1.93 ± 0.25	0.27	8 ± 0.5	48.6 ± 0.02	55 ± 7

---

Aerogel CA-5.1 was functionalized with TiCl₃ nanoparticles according to previously published procedures.¹⁵ Initially, aerogel monoliths (0.6819 g) were submerged into a solution of 0.1961 g (22.3 wt%) TiCl₃ (Aldrich, 99.995%) in 185 mL dry acetone (Aldrich, ≥99.9%). The solution was transferred from the glove box to a Schlenk line system without air exposure and acetone was removed by evaporating under dynamic vacuum. The sample was further activated at 190 °C in dynamic vacuum for 48 hours and this procedure incorporated 8.7 wt% TiCl₃ nanoparticles in the CO₂ activated CA-5.1, which is denoted as TiCl₃ functionalized carbon aerogel, CA–Ti.

Sodium alanate was melt infiltrated by grinding monoliths of CA with NaAlH₄ (Aldrich, 90%), obtaining a homogeneous powder. Melt infiltration was performed under inert conditions in a custom-made manually operated steel hydrogenation apparatus by heating to 189 °C (ΔT/Δt = 2.6 °C min⁻¹) reaching hydrogen pressures in the range of 210 to 230 bar. The sample temperature was kept fixed at 189 °C for 15 minutes and then the sample was cooled naturally. The amount of NaAlH₄ was calculated to achieve similar pore filling of ∼60 vol% for all nanoconfined samples. A sample of bulk NaAlH₄ without CA was melted and cooled under the same conditions as described above and was used as a reference, which is denoted as Na-m.

2.2. Characterization

Characterization of the nanoporous aerogel scaffold material was performed by gas adsorption and desorption using a Nova 2200e surface area and pore size analyzer from Quantachrome. The aerogels were degassed under vacuum for several hours at 300 to 320 °C, prior to the measurements. A full absorption and desorption isotherm was measured in the pressure range of 0 to 1 p/p₀ at liquid nitrogen temperatures with nitrogen gas as the adsorbent. Data were analyzed using the t-plot method,³⁰ the Brunauer–Emmett–Teller (BET) method,³¹ and the Barrett–Joyner–Halenda (BJH) method, and the total volume was calculated from a single point at p/p₀ ∼ 1.³²

Solid-state ²³Na and ²⁷Al MAS NMR spectra were recorded on a Varian INOVA-400 (9.39 T) spectrometer, using a homebuilt X-[¹H] double-resonance MAS NMR probe for 5 mm o.d. rotors. The NMR experiments were performed at ambient temperature and employed air-tight end-capped zirconia rotors packed with the samples in an argon-filled glove box. ²⁷Al MAS NMR spectra were recorded for the freshly prepared samples whereas the ²³Na MAS NMR spectra were acquired for the same samples after one year of storage under argon at −40 °C. The spectra were obtained with spinning speeds of ν_R = 8–10.0 kHz, short ²³Na and ²⁷Al excitation pulses (0.5 μs) for an rf field strength of γB₁/2π ≈ 50 kHz, and a 2 s relaxation delay. For the ²⁷Al MAS NMR spectra, the transmitter offset was set to 800 ppm to obtain a similar excitation of the resonances from NaAlH₄ and metallic Al. The ²³Na and ²⁷Al isotropic chemical shifts are in ppm relative to 1.0 M aqueous solutions of NaCl and AlCl₃·6H₂O, respectively.

Powder X-ray diffraction was performed with a Rigaku Smartlab diffractometer using Cu wavelength (λ₁ = 1.540530 and λ₂ = 1.544310, ratio = 0.5). Samples were loaded in glass capillaries (after one year of storage under argon at −40 °C), sealed with glue and transferred to the diffractometer without air exposure. The data were analyzed using Rietveld refinements. The background X-ray scattering was described by linear interpolation between selected points, while Thompson–Cox–Hastings axial divergence symmetry profile functions were used to fit the diffraction peaks and the instrumental resolution function was determined from a PXD pattern of a LaB₆ standard. The observed diffraction peaks were modeled using two structural models of NaAlH₄ with identical structural parameters (kept fixed) and slightly different unit cell parameters (refined). The average crystallite size for bulk NaAlH₄ was kept fixed in accordance with the instrumental resolution while that for nanoconfined NaAlH₄ was refined using the Rietveld approach implemented in the Fullprof suite.³³

Temperature-Programmed Desorption Mass Spectroscopy, TPD-MS, was performed in a Setaram Sensys Evo differential scanning calorimeter (horizontal position) coupled with a Hiden Analytical quadrupole mass spectrometer (MS) under a constant flow (28 mL min⁻¹) of ultra-high purity helium (<10 ppb O₂ and H₂O, Air Products) for temperature-programmed desorption mass spectroscopy (TPD-MS) measurements. A powdered sample (<20 mg) was placed in an Al₂O₃ crucible with a lid and then encapsulated in a protective aluminum crucible. The samples were purged with helium for at least 2 hours and heated in the temperature range of 20 to 450 °C (ΔT/Δt = 1 °C min⁻¹). During the experiment the MS signals at m/z = 2, 18, and 32 were recorded in order to detect H₂, H₂O and O₂. No significant amounts of H₂O or O₂ were detected.

The cyclic stability during four hydrogen release and uptake cycles for selected samples was studied by Sieverts' measurements. The samples were transferred to a stainless steel autoclave, sealed under argon and attached to the PCTpro 2000 apparatus. Hydrogen desorption data were collected in the temperature range from RT to 200 °C, at p(H₂) = 10⁻² bar, and with the temperature kept fixed at 100 °C for 4 h, at 150–152 °C for 4 h and at 200–203 °C for 8 h (ΔT/Δt = 0.5 °C min⁻¹). Hydrogen absorption was performed at p(H₂) = 89 to 92 bar and a temperature of 160 °C (ΔT/Δt = 5 °C min⁻¹) during 10 hours and then the sample was cooled naturally to RT. All handling of NaAlH₄ containing samples, activated CA and all sample preparations were performed in argon filled gloveboxes to avoid air exposure.

3. Results and discussion

3.1. CO₂ activated resorcinol formaldehyde carbon aerogels

Carbon aerogel (CA) scaffolds can be synthesized with specific pore size distributions and a CA with a maximum value of the pore size distribution, D_max = 10 nm, was selected for this study.¹¹ The morphology of CA can be systematically designed by activation in CO₂ and scaffolds with gradually increasing specific surface areas, total pore volumes and fraction of micropores (V_micro/V_tot) are obtained, see Table 1.^27,29 Scanning electron microscopy pictures of CA and CA-5.1 after infiltration of NaAlH₄ are compared in Fig. S1 in the ESI,‡ revealing a significant visible difference on the macroscopic scale since the CO₂ activated sample appears more rough and granular. In contrast, the nano-pore size, D_max, remained less affected in accordance with previous studies, see Table 1.^27–29,34 The final morphology of activated aerogels strongly depends on the structure and composition of the starting material.³⁵ The activated CA materials were less homogeneous than those prior to activation, which led to a broader distribution in the surface area (±270 m² g⁻¹) and total pore volume (±0.16), see Table 1. This may be due to thermal gradients and gas flow gradients during the activation process (see the ESI‡). Carbon aerogel CA-5.1 was functionalized with TiCl₃ nanoparticles (sample CA–Ti) in order to study the combined effects of increased surface area and the addition of a catalyst.

3.2. Melt infiltration of NaAlH₄

Sodium alanate, NaAlH₄, was melt infiltrated to obtain nanoconfined samples with similar pore filling (∼60 vol%), see Table 1. Activated aerogels have an increased surface area and total pore volume and incorporate more NaAlH₄ and therefore contain gravimetrically increasing amounts of hydride.

The nanoconfined samples were investigated by ²⁷Al MAS NMR in order to estimate the relative amounts of NaAlH₄ and metallic Al, employing the same approach as used in earlier studies of a mixture of Al–LiAlH₄ ³⁶and of Al–NaAlH₄.¹⁵ The ²⁷Al MAS NMR spectrum (Fig. 1A) of NaAlH₄ melted and cooled in the absence of a nanoscaffold (Na-m) shows centerband resonances from NaAlH₄ (93.7 ppm) and metallic Al (1639.5 ppm) in accordance with the ²⁷Al chemical shifts and quadrupolar couplings earlier reported for these compounds.^16,37,38 In addition, spinning sidebands from the ²⁷Al satellite transitions are observed for Na-m, indicating a higher degree of local structural order for NaAlH₄ in this sample as compared to the nanoconfined infiltrated samples. Resonances from NaAlH₄ and metallic Al only are also observed for the infiltrated samples (Fig. 1B–E) with relative signal intensities that are significantly different. It is observed that the intensity for metallic Al decreases for longer CO₂ activation times. This corresponds to a more efficient melt infiltration of NaAlH₄ in the CO₂ activated CA scaffolds (Fig. 1B–E).

Fig. 1 ²⁷Al MAS NMR spectra (9.4 T) illustrating the central transition regions for Al and AlH₄⁻ in (A) a sample of bulk NaAlH₄ after melting and cooling (Na-m), (B) melt infiltrated NaAlH₄ in CA-5.1, (C) in CA-5, (D) in CA-3 and (E) in CA (not activated). The spectrum in (A) was obtained with a spinning speed of ν_R = 10.0 kHz while the remaining spectra (B–E) employed ν_R = 9.0 kHz. The asterisk in (A) indicates the centerband resonance (−42.5 ppm) from a minor impurity of Na₃AlH₆.

The relative fractions of NaAlH₄ and Al can be derived from the intensities in the ²⁷Al MAS NMR spectra (Fig. 1), taking into account that the centerband resonance and first-order spinning sidebands for metallic Al correspond to the central as well as the satellite transition intensity whereas only the central transition is observed for NaAlH₄ in the melt-infiltrated samples (Fig. 1B–E). These data are summarized in Table 2 and show that the amount of NaAlH₄ infiltrated in the CA, CA-3, CA-5, CA-5.1 and CA–Ti samples increases with increasing surface area. The resonance from metallic Al is observed at 1639.5 ± 0.5 ppm for all samples and with very similar linewidths (FWHM = 3.3–3.7 ppm) for the melt infiltrated samples. Aluminium has a tendency to form larger nanoparticles outside the scaffold and may therefore interact more weakly with the scaffold.^11,15 The frequency for the centerband resonance from NaAlH₄ varies systematically for the infiltrated samples (Table 2) as illustrated in Fig. 2. The centerband resonance is shifted towards higher frequency and becomes more narrow (decreasing FWHM) with increasing degree of CO₂ activation of the scaffolds (Fig. 2D–B). These variations may reflect differences in the interaction between nanoconfined NaAlH₄ and the scaffold, potentially as a result of an increasing formation of graphite/graphene in the scaffold mediating electronic interactions between conduction electrons from graphite/graphene and the Al nucleus of the metal hydride.

Table 2 ²⁷Al chemical shifts (center of gravity, δ_cg) and line widths (FWHM) for the NaAlH₄ centerbands along with relative amounts of metallic Al and NaAlH₄ (mol%) derived from ²⁷Al MAS NMR. Average crystallite sizes (L) of nanoconfined NaAlH₄ and the estimated amount relative to bulk NaAlH₄ along with the ratio between the total amount of NaAlH₄ and Al extracted from Rietveld refinement of PXD data

Sample	δ cg (ppm)	FWHM (ppm)	Al^b (mol%)	NaAlH4^b (mol%)	NaAlH4 crystallite size L^c (nm)	Al^c (wt%)	NaAlH4^c (wt%)	Nanoconf. fraction of NaAlH4^c (wt%)
a Center of gravity for the centerband from the central transition. b Calculated from the ^27Al MAS NMR spectra in Fig. 1, considering the fact that the resonances from metallic Al include both the central and satellite transitions whereas only the central-transition intensity is observed for NaAlH4. The estimated error limits are 2–4% for the relative molar intensities. c Calculated from Rietveld refinements of PXD data, Fig. 3.
CA	95.1	11.9	32	68	19	23	77	85
CA-3	98.8	10.2	18	82	18	12	88	76
CA-5	101.0	9.0	9	91	17	8	92	97
CA-5.1	102.0	5.0	5	95	13	7	93	72
CA–Ti	100.9	6.2	4	96	—	—	—	—
Na-m	93.7	16.4	1	99	—	—	—	—

---

Fig. 2 Expansion of the spectral region for the ²⁷Al centerband resonances for NaAlH₄ in the ²⁷Al MAS NMR spectra shown in Fig. 1: (A) bulk NaAlH₄ (Na-m), (B) melt infiltrated NaAlH₄ in CA-5.1, (C) in CA-5, (D) in CA-3 and (E) in CA (not activated).

A similar effect is observed by ²³Na MAS NMR, as illustrated in Fig. 3 for the same samples (measured after one year of storage at −40 °C). NaAlH₄, bulk and confined in non-activated CA, gives a ²³Na centerband resonance at −9.8 ppm (Fig. 3A and E). A new ²³Na resonance is observed for NaAlH₄ at higher frequency (−3.4 ppm to −1.9 ppm), which increases in intensity and exhibits a decreasing line width (FWHM = 5.8 ppm to 4.5 ppm) for longer CO₂ activation times (Fig. 3D–B, respectively). This resonance is ascribed to electronic interactions between conduction electrons from an increasingly graphite/graphene-like scaffold and the Na nucleus of the metal hydride similar to the above mentioned observations by ²⁷Al MAS NMR. A deconvolution for sample CA-5.1 (the spectrum shown in Fig. 3B) of the two partly overlapping centerband resonances reveals that 79.8% of NaAlH₄ is infiltrated in activated CA whereas the remaining fraction is in less activated CA or is in a more bulk-like state. A high-frequency shift of the resonance from infiltrated NaAlH₄ was also observed in a recent ²³Na MAS NMR study of sodium alanate melt infiltrated in porous carbon.^37,39 The observation of two distinct resonances (Fig. 3B–D) reflects that ²³Na MAS NMR allows a more clear distinction of NaAlH₄ infiltrated in activated and non-activated CA. In fact, the high-frequency shifted resonance for NaAlH₄ infiltrated in activated CA observed in the ²⁷Al MAS NMR spectra (Fig. 2B–D) has a low-frequency shoulder which gives the asymmetric lineshapes with a tail to low frequency for the centerband.

Fig. 3 ²³Na MAS NMR spectra (9.4 T, ν_R = 8.0 kHz) of the central-transition region for NaAlH₄ in (A) bulk NaAlH₄ after melting and cooling (Na-m), (B) melt infiltrated NaAlH₄ in aerogel CA-5.1, (C) sample CA-5, (D) in CA-3 and (E) in CA.

The improved resolution of the resonances from nanoconfined NaAlH₄ in the ²³Na MAS NMR spectra may result from several effects such as smaller quadrupole couplings for ²³Na as compared to ²⁷Al and a stronger interaction of the ²³Na spins with the electrons of the scaffold material.

Sample Na-m contains only traces of Al (∼1 mol%), confirming that bulk NaAlH₄ is stable towards the conditions used for melt infiltration, in agreement with the literature.² Regardless hereof, molten NaAlH₄ partially decomposes during melt infiltration into the scaffolds, as revealed from the presence of Al. However, CO₂ activation and the higher surface area appear to stabilize molten NaAlH₄ and a lower fraction decomposes to Al (see Fig. 1). This new effect may be explained by alterations in the aerogel morphology, e.g. larger fraction of micropores, higher surface area and pore volume, or in the chemical composition. Previous studies reveal that CO₂ activation of RF-CAs increases the CA skeletal density towards that of pure graphite (e.g. ρ_RF-CA ∼ 1.89, ρ_{RF-CA, activated} ∼ 2.18 and ρ_graphite = 2.25 g cm⁻³).^27,35 Different carbon additives such as graphite, graphene, fullerene and carbon nanotubes have substantially different properties regarding catalysis of hydrogen release and uptake in NaAlH₄.^24,25 In that perspective, the observed stabilization of molten NaAlH₄ may be explained by a graphitization of the activated aerogels. The infiltration process remains similarly efficient in the titanium functionalized activated RF-CA (sample CA–Ti incorporates 96 wt% of added NaAlH₄).

The amount of bulk NaAlH₄ relative to nanoconfined NaAlH₄ after the infiltration process was investigated by Rietveld refinement of powder X-ray diffraction data, shown in Fig. 4A. The analysis reveals that the Bragg diffraction peak shapes can be modeled as a superposition of a broad contribution from nano-confined NaAlH₄ and a narrow contribution from bulk NaAlH₄ (Fig. 4B). This approach provides the ratio of nanoconfined versus bulk NaAlH₄ as well as the average particle size for nanoconfined NaAlH₄ and the amount of Al present in the sample due to decomposition of NaAlH₄, see Table 2. Only a minor fraction, 7 to 23 wt%, is decomposed to Al and a majority of the remaining NaAlH₄, 72 to 97 wt%, is successfully nanoconfined, in accordance with the ²³Na and ²⁷Al MAS NMR data. The results also reveal decreasing average crystallite sizes of ∼19, 18, 17 and 13 nm for NaAlH₄ nanoconfined in CA, CA-3, CA-5 and CA-5.1, respectively, which has maximum of the BET pore size distributions in the range of 8 to 11 nm (see Table 1). This observation indicates that NaAlH₄ may preferably crystallize in the larger pores of the carbon aerogel.

Fig. 4 (A) Powder X-ray diffraction patterns of NaAlH₄ nanoconfined in CA (bottom), CA-3 (middle) and CA-5.1 (top). (B) PXD pattern of sample NaAlH₄ nanoconfined in CA-5.1 (λ = 1.54 Å, Cu Kα). The red curve shows the raw data and the black curve shows the calculated Rietveld model. The difference plot illustrates the fit quality. Reflection markers from top to bottom show NaAlH₄ and Al. The square inset shows the diffracted intensity of nanoconfined (dashed curve) and bulk (solid curve) NaAlH₄.

The results obtained from ²⁷Al and ²³Na MAS NMR and powder X-ray diffraction for freshly prepared samples and samples stored for one year at −40 °C under argon are in accord and reveal high sample stability under the selected conditions. Minor discrepancies may be attributed to sample inhomogeneity.

3.3. Effects of surface area on the hydrogen release kinetics

The first hydrogen desorption for all samples was investigated using simultaneous temperature programmed desorption and mass spectroscopy, TPD-MS (Fig. 5). The onset temperatures for hydrogen release, T_onset, and the temperatures at which the relative H₂ desorption rate reach a maximum value, T_max, are compared in Table 3. The onset temperatures for hydrogen release are clearly increasing with increasing CO₂ activation of the CAs and the titanium functionalized scaffold reveals a T_onset, which is significantly lower and similar to a previous study using a Ti-functionalized non-activated CA.¹⁵

Fig. 5 Normalized TPD-MS data showing the relative hydrogen desorption intensities for NaAlH₄ in CA (green), CA-3 (blue), CA-5.1 (black), CA–Ti (orange) and bulk Na-m (red) recording the mass spectrometer intensity of H₂⁺ ions (m/z = 2) in the temperature range of RT to 400 °C (ΔT/Δt = 1 °C min⁻¹).

Table 3 Temperatures for the onset of hydrogen release, T_onset, and the maximum of the relative hydrogen desorption intensity, T_max, for the major hydrogen release peaks

Sample	∼Tonset/°C	T max/°C	T max/°C^b
a The TPD-MS profile is shown in the ESI, Fig. S3. b Decomposition of NaH.
CA	75	147	—
CA-3	100	163	329
CA-5^a	115	180	316
CA-5.1	115	184	322
CA–Ti	25	91 and 120	293
Na-m	200	252 and 265	349

---

The recorded T_max values for NaAlH₄ infiltrated in CA, CA-3, CA-5 and CA-5.1 are 147, 163, 180 and 184 °C, respectively, i.e. the hydrogen desorption rates clearly decrease with increasing CO₂ activation of CA. Furthermore, T_max values approach the melting point of NaAlH₄ (T_mp = 183 °C) indicating that the observed stabilization of molten NaAlH₄ during melt infiltration and nanoconfinement may be a kinetic effect, i.e. a result of the higher decomposition temperature. Thus, non-activated CA provides a higher degree of decomposition and agglomeration of Al particles outside the scaffold. The T_max values for the activated samples remain significantly lower than that for bulk NaAlH₄ (Na-m) of T_max = 252 °C. However, CA–Ti incorporates 96 wt% of the added NaAlH₄ and simultaneously has low T_max values of 91 and 120 °C owing to titanium functionalization, see Tables 2 and 3.¹⁵ The two observed low temperature maxima on the TPD-MS curve for sample CA–Ti may be due to hydrogen release from NaAlH₄ followed by Na₃AlH₆.

The TPD-MS profile of bulk NaAlH₄ (Na-m without the aerogel) reveals local T_max values at 252, 265 and 348 °C assigned to hydrogen release from NaAlH₄, Na₃AlH₆ and NaH(s) forming Na(l). For nanoconfined samples the latter local maximum is usually less evident, e.g. NaAlH₄ in CA, see Fig. 5.^15,16 However, hydrogen release from NaH is clearly observed for the activated samples, NaAlH₄ in CA-3, CA-5 and CA-5.1, with local T_max values of 333, 313 and 326 °C, respectively, see Fig. 5 and S2 in the ESI.‡ This resembles the TPD-MS desorption profiles reported for NaAlH₄ melt infiltrated onto the surface of non-porous graphite.¹⁶

Furthermore, comparing the kinetics for hydrogen release and uptake in NaAlH₄ ball milled with different additives, the following trend is observed: RF-CA > activated carbons > graphite, graphene.²⁵ Therefore, the increasing T_max values for NaAlH₄ and more intense observation of NaH by TPD-MS (Fig. 5) for the CO₂-activated samples may be explained as an increasing graphitization of the scaffold.²⁷ Noteworthy, there is limited variation in the apparent crystallite sizes of NaAlH₄ in CA, CA-3, CA-5 and CA-5.1 within ∼13 to 19 nm, which is shown to have a limited effect on the kinetics.¹¹

3.4. Stability during hydrogen exchange

Sieverts' measurements were utilized to study the stability of the samples during four hydrogen release and uptake cycles, see Fig. 6. NaAlH₄ nanoconfined in CA, CA-3 and CA-5.1 releases 1.3, 1.9, and 2.7 wt% H₂ (relative to the mass of the sample), respectively, during the first desorption, corresponding to 72.1, 78.1 and 89.1% of the calculated hydrogen release relative to the used amount of NaAlH₄, see Table 4. The results reveal that a higher fraction of NaAlH₄ is successfully infiltrated in the CO₂ activated scaffolds without decomposing, in agreement with ²³Na/²⁷Al MAS NMR and PXD data, see Fig. 1–4.

Fig. 6 Sieverts' measurements showing hydrogen desorption cycles 1 to 4 for NaAlH₄ infiltrated in (A) CA, (B) CA-3, (C) CA-5.1 and (D) CA–Ti. Hydrogen desorption was performed at fixed temperatures of 100, 150, and 200 °C (ΔT/Δt = 0.5 °C min⁻¹; dashed lines). Hydrogen absorption was performed at 160 °C and p(H₂) = 90 to 93 bar for 10 hours.

Table 4 The theoretical H₂ content for sodium alanate in different scaffolds, ρ_m(H₂), considering the added amount of NaAlH₄. The measured H₂ release during the first desorption calculated relative to the mass of the sample and in percent relative to the calculated amount of hydrogen in the sample (in brackets). The hydrogen release during desorption cycles one to four is calculated relative to the added amount of NaAlH₄ and in percent relative to the first cycle (in brackets). The data are extracted from Fig. 6

Sample	ρ m(H2) wt% H2/sample^a	1.Des/sample wt% H2^b	1.Des/NaAlH4 wt% H2	2.Des/NaAlH4 wt% H2	3.Des/NaAlH4 wt% H2	4.Des/NaAlH4 wt% H2
a Calculated based on the amount of added NaAlH4, considering the purity of 90%. b Calculated from the H2 release measured by Sieverts' data. c Sieverts' data are shown in the ESI, Fig. S4.
CA	1.8	1.3 (72.1)	4.0 (100)	2.7 (65.7)	2.4 (60.3)	2.1 (51.9)
CA-3	2.2	1.9 (78.1)	4.4 (100)	3.6 (83.0)	3.4 (77.3)	3.3 (76.2)
CA-5^c	2.7	2.4 (89.2)	5.0 (100)	4.3 (85.7)	3.8 (76.9)	3.7 (74.2)
CA-5.1	3.0	2.7 (89.1)	5.0 (100)	4.6 (91.9)	4.2 (85.1)	4.0 (79.8)
CA–Ti	2.6	2.1 (80.4)	4.5 (100)	3.8 (83.4)	3.3 (73.0)	3.0 (65.8)

---

Sodium alanate in CA releases 4.0, 2.7, 2.4 and 2.1 wt% H₂/NaAlH₄ during the first, second, third, and fourth desorption cycles, i.e. 100.0, 65.7, 60.3 and 51.9% of the initial H₂ content, respectively, showing that nanoconfinement of NaAlH₄ in non-activated RF-CAs suffers from significant loss of hydrogen storage capacity in agreement with previous studies.¹⁵ In contrast, NaAlH₄ in CA-3 releases 83.0, 77.3 and 76.2% and in CA-5.1 91.9, 85.1 and 79.8% of the initial hydrogen content during desorption cycles 2, 3 and 4, respectively. Thus, the activation of RF-CAs significantly improves the stability and reversible hydrogen storage capacity of nanoconfined NaAlH₄. This is a surprising result, suggesting that more graphite-like scaffolds tend to facilitate the infiltration process, stabilize nanoconfined NaAlH₄ and preserve a higher hydrogen storage capacity over several cycles of hydrogen release and uptake. NaAlH₄ infiltrated in activated and functionalized CA–Ti scaffolds releases 2.1 wt% H₂/sample during the first desorption corresponding to 4.5 wt% H₂/NaAlH₄. Interestingly, a total of 2.90 wt% H₂/NaAlH₄ is released at T ≤ 100 °C and this amount corresponds to 64% of the total hydrogen content. During decomposition of bulk NaAlH₄, 50% of the hydrogen content is released during the initial step forming Na₃AlH₆ and 25% of the hydrogen is released during the second step when Na₃AlH₆ decomposes to NaH and Al. Therefore, the current results indicate full decomposition of NaAlH₄ at T ≤ 100 °C. NaAlH₄ in CA–Ti releases 83.4, 73.0 and 65.8% of the initial hydrogen content during desorption cycles 2, 3 and 4, respectively, see Table 4 and Fig. 5D. Noteworthy, NaAlH₄ in CA–Ti has intermediate cyclic properties, i.e. it is more stable than the non-activated samples but less stable than the activated sample.

This study demonstrates that activated carbon aerogels have increased surface area and pore volume and confine more NaAlH₄ than non-activated CA. In addition, molten NaAlH₄ appears to be stabilized by the higher surface area and possibly more graphite/graphene-like scaffolds, which provide a more efficient melt infiltration process. Furthermore, the stability of nanoconfined NaAlH₄ during continuous hydrogen release and uptake is remarkably improved, i.e. CO₂ activation of aerogels is an efficient method to significantly increase and preserve the hydrogen storage capacity of nanoconfined NaAlH₄. Furthermore, considering absolute (100%) pore filling of a CO₂-activated aerogel, e.g. CA-5.1 with V_tot = 2.38 cm³ g⁻¹, has the potential to incorporate 68 wt% NaAlH₄ corresponding to 3.4 wt% H₂/sample.

The hydrogen desorption kinetics decrease with increasing surface area (Fig. 5 and 6). However, this study clearly demonstrates that activated aerogels can be functionalized with TiCl₃ in order to combine the prolific properties from nanoconfinement, the efficient melt infiltration and the catalytic effect from titanium facilitating fast hydrogen release below 100 °C, see Fig. 5 and 6D. The surface area and TiCl₃ content may be further optimized for the functionalized scaffolds in order to prepare ultimate materials of nanoconfined NaAlH₄. The activated carbon aerogels may have similar prolific properties for nanoconfinement of other metal hydrides such as LiBH₄, Mg(BH₄)₂ and MgH₂ and may attract significant attention in the field of nanoconfinement. The mobility of hydrogen atoms on the surface of graphene and graphite is high and comparable to hydrogen atoms in the gas phase.²²

4. Conclusions

This work reveals that CO₂-activated carbon aerogels (CAs) have increased surface area and pore volume and incorporate more nanoconfined NaAlH₄. The NaAlH₄ melt infiltration process was significantly improved for the CO₂ activated CA, which incorporates up to 91 mol% of the added NaAlH₄ in contrast to only 52 mol% nanoconfined in as prepared CA. The available hydrogen content was increased from 1.3 wt% H₂/sample in CA to 2.7 wt% H₂/sample in the activated CA using ∼60 vol% pore filling. Considering 100% pore filling, a CO₂-activated aerogel may reach a hydrogen storage capacity of 3.4 wt% H₂/sample. Furthermore, the stability of nanoconfined NaAlH₄ over several cycles of hydrogen release and uptake was significantly improved for the CO₂ activated scaffolds. These new prolific effects may originate from a graphitization and/or chemical inertness of the CO₂-activated scaffolds. In contrast, the desorption kinetics for hydrogen release from NaAlH₄ decrease due to nanoconfinement in the CO₂-activated aerogels but remain superior compared to bulk NaAlH₄. This drawback is solved by TiCl₃ nano-particle functionalization of RF-CA scaffolds. NaAlH₄ efficiently melt infiltrates such Ti-functionalized scaffolds and shows fast H₂ desorption kinetics at T < 100 °C. Activated RF-aerogels are relatively easy to prepare, appear to be more chemically inert and may be utilized to optimize nanoconfined LiBH₄, Mg(BH₄)₂, MgH₂ and other metal hydrides.

Acknowledgements

The work was supported by CarlsbergFondet, the Danish National Research Foundation, Centre for Materials Crystallography (DNRF93), the Danish Strategic Research Council (Centre for Energy Materials and the project HyFillFast) and the Danish Council for Independent Research Natural Science (DanScatt). The access to beam time at the MAX-II synchrotron, Lund, Sweden in the research laboratory MAX-lab is gratefully acknowledged. The use of the facilities at the Instrument Centre for Solid-State NMR Spectroscopy, Aarhus University, sponsored by the Danish Council for Independent Research: Natural Sciences, the Danish Technical Science Research Council, CarlsbergFondet and Director Ib Henriksens Foundation, is acknowledged. We are grateful to the Polish Ministry of Science and Higher Education (Key Project POIG.01.03.01-14-016/08 and POIG.02.01.00-14-071/08/00).

References

1. D. J. C. Mackay, Sustainable Energy – Without the Hot Air, UIT Cambridge Ltd., England, 2009.
2. B. Bogdanovic, R. A. Brand, A. Marjanovic, M. Schwickardi and J. Tölle, J. Alloys Compd., 2000, 302, 36–58.
3. M. Dornheim, N. Eigen, G. Barkhordarian, T. Klassen and R. Bormann, Adv. Eng. Mater., 2006, 8, 377–385.
4. J. Graetz, Chem. Soc. Rev., 2009, 38, 73–82.
5. L. H. Rude, T. K. Nielsen, D. B. Ravnsbaek, U. Bösenberg, M. B. Ley, B. Richter, L. M. Arnbjerg, M. Dornheim, Y. Filinchuk and F. Besenbacher, et al. , Phys. Status Solidi A, 2011, 208, 1754–1773.
6. U. Eberle, M. Felderhoff and F. Schüth, Angew. Chem., Int. Ed., 2009, 48, 6608–6630.
7. P. E. de Jongh and P. Adelhelm, ChemSusChem, 2010, 3, 1332–1348.
8. J. J. Vajo, Curr. Opin. Solid State Mater. Sci., 2011, 15, 52–61.
9. T. K. Nielsen, F. Besenbacher and T. R. Jensen, Nanoscale, 2011, 3, 2086–2098.
10. M. Fichtner, Phys. Chem. Chem. Phys., 2011, 13, 21186–21195.
11. T. K. Nielsen, P. Javadian, M. Polanski, F. Besenbacher, J. Bystrzycki and T. R. Jensen, J. Phys. Chem. C, 2012, 116, 21046–21051.
12. Z. Zhao-Karger, J. Hu, A. Roth, D. Wang, C. Kubel, W. Lohstroh and M. Fichtner, Chem. Commun., 2010, 46, 8353–8355.
13. T. K. Nielsen, K. Manickam, M. Hirscher, F. Besenbacher and T. R. Jensen, ACS Nano, 2009, 3, 3521–3528.
14. A. F. Gross, J. J. Vajo, S. L. Van Atta and G. L. Olson, J. Phys. Chem. C, 2008, 112, 5651–5657.
15. T. K. Nielsen, M. Polanski, D. Zasada, P. Javadian, F. Besenbacher, J. Bystrzycki, J. Skibsted and T. R. Jensen, ACS Nano, 2011, 5, 4056–4064.
16. J. Gao, P. Adelhelm, M. H. W. Verkuijlen, C. Rongeat, M. Herrich, P. J. M. van Bentum, O. Gutfleisch, A. P. M. Kentgens, K. P. de Jong and P. E. de Jongh, J. Phys. Chem. C, 2010, 114, 4675–4682.
17. R. D. Stephens, A. F. Gross, S. L. Van Atta, J. J. Vajo and F. E. Pinkerton, Nanotechnology, 2009, 20, 204018.
18. M. Paskevicius, D. A. Sheppard and C. E. Buckley, J. Am. Chem. Soc., 2010, 132, 5077–5083.
19. Z. Zhao-Karger, R. Witter, E. G. Bardaji, D. Wang, D. Cossement and M. Fichtner, J. Mater. Chem. A, 2013, 1, 3379–3386.
20. A. Remhof, P. Mauron, A. Zuttel, J. P. Embs, Z. Lodziana, A. J. Ramirez-Cuesta, P. Ngene and P. de Jongh, J. Phys. Chem. C, 2013, 117, 3789–3798.
21. M. Paskevicius, H. Y. Tian, D. A. Sheppard, C. J. Webb, M. P. Pitt, E. M. Gray, N. M. Kirby and C. E. Buckley, J. Phys. Chem. C, 2011, 115, 1757–1766.
22. H. M. Cuppen and L. Hornekaer, J. Chem. Phys., 2008, 128, 174707.
23. B. Bogdanovic and M. Schwickardi, J. Alloys Compd., 1997, 253–254, 1–9.
24. P. A. Berseth, A. G. Harter, R. Zidan, A. Blomqvist, C. M. Araújo, R. H. Scheicher, R. Ahuja and P. Jena, Nano Lett., 2009, 9, 1501–1505.
25. F. E. Pinkerton, J. Alloys Compd., 2011, 509, 8958–8964.
26. T. K. Nielsen, U. Bösenberg, R. Gosalawit, M. Dornheim, Y. Cerenius, F. Besenbacher and T. R. Jensen, ACS Nano, 2010, 4, 3903–3908.
27. C. Lin and J. A. Ritter, Carbon, 2000, 38, 849–861.
28. T. F. Baumann, M. A. Worsley, T. Y. J. Han and J. H. Satcher, J. Non-Cryst. Solids, 2008, 354, 3513–3515.
29. S. A. Al-Muhtaseb and J. A. Ritter, Adv. Mater., 2003, 15, 101–114.
30. J. H. de Boer, B. G. Linsen, T. van der Plas and G. J. Zondervan, J. Catal., 1965, 4, 649–653.
31. S. Brunauer, P. H. Emmett and E. Teller, J. Am. Chem. Soc., 1938, 60, 309–319.
32. E. P. Barrett, L. G. Joyner and P. P. Halenda, J. Am. Chem. Soc., 1951, 73, 373–380.
33. J. Rodriguez-Carvajal, FULLPROF SUITE, 2003.
34. Y. Hanzawa, K. Kaneko, R. W. Pekala and M. S. Dresselhaus, Langmuir, 1996, 12, 6167–6169.
35. M. S. Contreras, C. A. Paez, L. Zubizarreta, A. Leonard, S. Blather, C. G. Olivera-Fuentes, A. Arenillas, J. P. Pirard and N. Job, Carbon, 2010, 48, 3157–3168.
36. L. Kellberg, H. Bildsøe and H. J. Jakobsen, J. Chem. Soc., Chem. Commun., 1990, 1294–1295.
37. M. H. W. Verkuijlen, J. Gao, P. Adelhelm, P. J. M. van Bentum, P. E. de Jongh and A. P. M. Kentgens, J. Phys. Chem. C, 2010, 114, 4683–4692.
38. B. Bogdanovic, M. Felderhoff, M. Germann, M. Hartel, A. Pommerin, F. Schuth, C. Weidenthaler and B. Zibrowius, J. Alloys Compd., 2003, 350, 246–255.
39. P. Adelhelm, J. B. Gao, M. H. W. Verkuijlen, C. Rongeat, M. Herrich, P. J. M. van Bentum, O. Gutfleisch, A. P. M. Kentgens, K. P. de Jong and P. E. de Jongh, Chem. Mater., 2010, 22, 2233–2238.

---

Footnotes

† Dedicated to the memory of a colleague, friend and mentor professor Jerzy Bystrzycki who suddenly passed away much too early.

‡ Electronic supplementary information (ESI) available. See DOI: 10.1039/c3nr03538g

---