Structural analysis of calcium reactive hydride composite for solid state hydrogen storage†
aMaterials Physics, Helmholtz-Zentrum Geesthacht, Max-Planck Strasse 1, Geesthacht, SH, 21502, Germany, bMaterials Technology, Helmholtz-Zentrum Geesthacht, Max-Planck Strasse 1, Geesthacht, SH, 21502, Germany, cDepartment of Microstructure and Residual Stress Analysis, Helmholtz-Zentrum Berlin, Albert-Einstein-Strasse 15, Berlin, 12489, Germany, and dHASYLAB, DESY, Notkestrasse 85, Hamburg, D-22603, Germany
*Correspondence e-mail: email@example.com
Owing to a theoretical hydrogen storage capacity of 10.5 wt% H2, Ca(BH4)2+MgH2, the so-called calcium reactive hydride composite (Ca-RHC), has a great potential as a hydrogen storage material. However, its dehydrogenation temperature (∼623 K) is too high for any mobile applications. By addition of 10 mol% of NbF5 into Ca(BH4)2+MgH2, a decrease of the dehydrogenation onset temperature by ∼120 K is observed. In order to understand the reasons behind this desorption temperature decrement two sets of samples [Ca(BH4)2+MgH2 and Ca(BH4)2+MgH2+0.1NbF5] in different hydrogenation states, were prepared. The structural investigation of the above mentioned sets of samples by means of volumetric measurements, anomalous small-angle X-ray scattering (ASAXS) and X-ray absorption spectroscopy (XAS) is reported here. The XAS results show that after the milling procedure NbB2 is formed and remains stable upon further de/rehydrogenation cycling. The results of Nb ASAXS point to nanometric spherical NbB2 particles distributed in the hydride matrix, with a mean diameter of ∼10 nm. Results from Ca ASAXS indicate Ca-containing nanostructures in the Ca-RHC+0.1NbF5 samples to be ∼50% finer compared to those without additive. Thus, a higher reaction surface area and shorter diffusion paths for the constituents are concluded to be important contributions to the catalytic effect of an NbF5 additive on the hydrogen sorption kinetics of the Ca(BH4)2+MgH2 composite system.
Hydrogen is considered as a clean and renewable energy source for the short term. However, safe storage of hydrogen is challenging owing to its gaseous state under atmospheric conditions. The possibility to reversibly store hydrogen in metal hydrides and complex hydrides has attracted the interest of many researchers over the past decade (Schuth et al., 2004; Züttel, 2003). High hydrogen capacities, high energy efficiency and storage safety are the main arguments favoring metal hydrides and complex hydrides over pressurized gas and liquid storage at cryogenic temperatures. Lightweight complex metal hydrides are regarded as especially suitable materials for mobile hydrogen storage applications (Schlapbach & Züttel, 2001; Züttel et al., 2003). Calcium borohydride, Ca(BH4)2, is one of the most attractive candidates for hydrogen storage in mobile applications, because of its high gravimetric (11.5 wt% H2) and volumetric hydrogen capacity (130 kg m−3). The theoretical decomposition reaction enthalpy has been calculated to be 32 kJ mol−1 H2 (Miwa et al., 2006), but calcium borohydride shows poor hydrogen sorption reversibility (Kim et al., 2009) owing to the formation of stable decomposition products (Kim et al., 2012). An approach to modify the thermodynamical stability of hydrides is the concept of the so-called reactive hydride composites (RHCs) (Vajo et al., 2005; Barkhordarian et al., 2007). Based on this idea, MgH2 can be used to destabilize Ca(BH4)2. Although the reversibility of the Ca(BH4)2+MgH2 (Ca-RHC) composite system is improved compared with pure Ca(BH4)2, the dehydrogenation temperature of the system of about 623 K remains too high for current industrial applications. In the present work, we were able to decrease the dehydrogenation temperature of the Ca-RHC system down to approximately 523 K by adding 10 mol% NbF5 to the composite system. To understand the effect of NbF5 on the dehydrogenation behavior of the composite system, a comprehensive investigation of Ca-RHC and Ca-RHC+0.1NbF5 systems was performed. Special attention was paid to structural characterization of the material with respect to the hydrogen sorption processes.
Anomalous small-angle X-ray scattering (ASAXS) measurements at the K absorption edge of calcium (4.03 keV) and niobium (18.8 keV) were carried out to determine the changes in the size distributions of Ca- and Nb-containing structures before and after a hydrogenation cycle. This is the first time, to best of our knowledge, that Ca ASAXS has been successfully performed on a calcium-based hydrogen storage system. The results of this work contribute to a deeper understanding of the sorption processes in Ca-RHCs, especially the structural impact of transition metal (TM)-based additives, and their correlation with the kinetic properties of hydride systems.
A short introduction to the methods used in this study is given in the following. First the method of ASAXS is introduced and subsequently the theory of X-ray absorption spectroscopy (XAS). The ASAXS technique is based on the energy dependency of the atomic scattering factor fi(E) of the elements:
Here, E is the energy of the incident beam, fi0 is the atomic number Z of the element i. and are the anomalous dispersion factors of the element i, which only become significant when the energy of the incident beam is close to an absorption edge of the resonant element. Since the electron density fluctuation is proportional to fi(E)fi*(E), the observed total scattering intensity of a sample containing the element i can be separated into non-resonant (off-resonant, O), mixed resonant (OR) and resonant (R) intensity terms (Stuhrmann, 1985):
Here, q = (4π/λ)sin(θ), where λ is the wavelength and 2θ is the scattering angle of the incident beam. The non-resonant contribution is the normal SAXS intensity, the mixed resonant intensity is the contribution of resonant and non-resonant elements, and the resonant intensity is the intensity contribution of the resonant element i with the atomic scattering amplitude fi(E). For a two-phase system, the resonant scattering intensity can be extracted from equation (2) by measuring the small-angle scattering (SAXS) intensity at a minimum of three different energies adjacent to an absorption edge of the resonant element i. For more detail the reader is referred to Simon & Lyon (1994), Serimaa et al. (1996) and Goerigk et al. (2004).
The chemical state of the additive and its local environment can be characterized using XAS. An XAS spectrum is usually separated into two distinct regions: X-ray absorption near-edge spectroscopy (XANES) and extended X-ray absorption fine structure (EXAFS). The XANES region typically starts from ∼100 eV before the main absorption edge and ends at ∼30 eV thereafter. The following part of the XANES spectrum is ascribed to the EXAFS region which typically ends about 1000 eV above the main absorbing edge. This part of the spectrum (EXAFS) describes the local environment of the absorbing atom. The EXAFS region is normalized by the absorption jump at the main absorption edge and defines the fine-structure function:
Here, μ(E) is the measured absorption coefficient, μ0(E) is the background absorption function of an isolated absorber and Δμ0(E) is the absorption jump arising in the vicinity of an absorption edge of the resonant atoms in the matrix. Equation (3) was used to transform the measured μ(E) value into χ(E), which was further transformed to the k space [χ(k)], where k = [2m(E − E0)/ℏ2]1/2. Although both regions of the XAS spectrum have the same physical origin, the processes that occur in the XANES region are quantitatively not straightforward to describe (Bunker, 2010); hence XANES is routinely used as a fingerprint technique, which outlines the chemical state of the absorber in the matrix. Based on quantum mechanics, the EXAFS region is relatively easy to describe via the EXAFS equation [equation (4)] (Königsberger, 1988), which is used in this work to describe the local environment of niobium in the hydride matrix:
Here, fj(k) and δj(k) are the scattering amplitude and the phase shift of the scattering photoelectron, respectively, at the atom j. Both parameters are related to the properties of the surrounding atoms of the absorber. Nj and Rj describe, respectively, the number and distance of the atom j, is the variance of the distance of the atom j from the absorber, and S02 is the amplitude of the EXAFS oscillations. A detailed theory of XAS is given by Königsberger (1988) and Bunker (2010).
Ca(BH4)2 powder (purity >98.8%), MgH2 (purity >95%) and NbF5 (purity >99.9%) were purchased from Sigma-Aldrich and used as starting materials. Samples of Ca(BH4)2-MgH2 with 10 mol% of NbF5 (1:1) and without the additive were prepared. Both samples were ball-milled for 5 h in a SPEX 8000 mill. The powder was milled in stainless steel vials with a ball-to-powder ratio of 10:1. For both samples, magnesium hydride was pre-milled for 5 h and mixed with Ca(BH4)2 and Ca(BH4)2+0.1NbF5 and milled further for an additional 5 h. Preparation and handling of the sample during the whole procedure were carried out in a glove box under a continuously purified argon atmosphere (H2O and O2 <10 p.p.m.).
A Sieverts apparatus (fabricated by Hydro Quebec/HERA Hydrogen Storage Systems) was used to evaluate the hydrogen sorption kinetics of the samples (pure and doped Ca-RHC). It was also used to prepare samples for XAS and ASAXS investigations. Therefore, as-milled samples (am-Ca-RHC and am-Ca-RHC+0.1NbF5) were dehydrogenated (1st-des-Ca-RHC and 1st-des-Ca-RHC+0.1NbF5) and rehydrogenated (1st-abs-Ca-RHC and 1st-abs-Ca-RHC+0.1NbF5) to observe and compare the structural evolution of the pure Ca-RHC with the doped samples. The parameters for hydrogen desorption measurements were set to 648 K under vacuum (0 bar) for desorption and 623 K and 150 bar for absorption (1 bar = 100 kPa), and the temperature ramps were 3 K min−1.
XAS measurements were collected at the C beamline of the DORIS III synchrotron (DESY, Hamburg, Germany). The ideal amount of powder to be used for the measurements was calculated using the program XAFSMASS (https://www.cells.es/Beamlines/CLAESS/software/xafsmass.html ). The samples were mixed with cellulose (∼50 mg) in a mortar and pressed (5 bar) into pellets of 10 mm in diameter. The pellets were fixed in an aluminium sample holder plate and sealed with Kapton tape (55 µm in thickness) to avoid oxidation of the samples. All preparation and handling of specimens were carried out in a glove box with a purified argon atmosphere. NbF5 and NbB2, purchased from Alfa Aesar with the highest purity available, were measured as reference materials. The measurements were recorded in transmission mode at the K-edge of Nb. XAS data processing and analysis were performed using the IFEFFIT software package (Ravel & Newville, 2005). For ASAXS measurements, the samples were investigated at the K-edges of Ca and Nb to trace the structural changes of the Ca- and Nb-containing phases during hydrogen release and uptake. Calcium ASAXS measurements were performed at the 7T-MPW-SAXS beamline at BESSY II [Helmholtz-Zentrum Berlin (HZB), Germany]. The samples were put into a circular hole (5 mm in diameter and 0.2 mm thick) of an aluminium sample holder and sealed with Kapton tapes to avoid any oxidation of the samples. Two sample-to-detector distances (400 and 3345 mm) were chosen in order to cover the maximal accessible q range. The ASAXS measurements were taken at four distinct energies at the calcium K-edge (see Table 1). The data were collected on a multi-wire proportional counter gas-filled area detector with a pixel size of 207 µm. During the measurements, the samples were kept under continuous vacuum (10−4 mbar) conditions to minimize parasitic scattering. The measured intensities were corrected for background (scattering contribution of the beam without sample and sample holder), for detector dead time, for detector sensitivity by measuring a scandium foil, and for transmission and flux. A 90 µm-thick glassy carbon standard was used for obtaining the scattering cross section in absolute units. A silver behenate standard was measured to calibrate the q values.
The Nb ASAXS measurements were carried out at beamline B1 of the DORIS III synchrotron. The samples were mounted in aluminium sample holders with a circular hole of 5 mm in diameter and 1 mm in thickness. All samples were sealed with Kapton tapes to avoid any change in the oxidation state of Nb-containing phases. To avoid possible sample oxidation over the measurement time and air scattering, the measurements were performed under continuous vacuum (10−4 mbar). The scattering patterns were acquired at four distinct energies (see Table 2) below the Nb K-edge at two sample-to-detector distances (885 and 3585 mm) to cover the maximum available q range. The ASAXS measurements were collected with a two-dimensional single-photon counting Pilatus 1M detector. The measured intensities were corrected for background (scattering contribution of the beam without sample and sample holder), sample absorption and incoming flux, and were converted into absolute intensity units by measuring a glassy carbon reference.
Both ASAXS beamlines were equipped with a double-crystal Si monochromator with a wavelength resolution of Δλ/λ ≃ 104. Anomalous-dispersion experiments in the soft X-ray regime have already been performed by Stuhrmann et al. (1995). The difficulties arising from absorption and scattering processes of soft X-rays are summarized by Arndt (1984) and Djinović Carugo et al. (2005). Currently, the 7T-MPW-SAXS beamline at BESSY (A. Hoell) is, to our knowledge, the only beamline in Europe that provides ASAXS experiments at low energies down to 4 keV.
The dehydrogenation measurements of the pure Ca-RHC and Ca-RHC+0.1NbF5 are plotted with the corresponding temperature curve in Fig. 1. Desorption onset capacities for both samples were normalized to the same value for a better comparison of the dehydrogenation offset temperatures (horizontal black line). Vertical lines were set to mark the dehydrogenation onset times of both samples.
As can be seen in Fig. 1, the sample without additive starts to release hydrogen after 2.1 h with the corresponding temperature of roughly 643 K. The dehydrogenation reaction of the sample with 10 mol% addition of NbF5 begins after 1.3 h with an onset temperature of approximately 523 K. The onset temperature of dehydrogenation of the sample with 10 mol% of NbF5 is therefore lowered by about 120 K, which corresponds to a time difference of 48 min, in comparison to pure Ca-RHC. XAS and ASAXS measurements were carried out to shed light on the possible mechanisms of NbF5 on the reduction of the dehydrogenation temperature of the calcium hydride composite system.
XAS measurements were performed to study the chemical state of TM-based additives. In Fig. 2, XAS curves of Ca-RHC+0.1NbF5 at different hydrogenation states are shown together with spectra of several reference compounds. Different regions of the XAS spectra were utilized to determine the valence state and the local environment of Nb in the Ca-RHC matrix. The near-edge X-ray absorption fine structure was used to determine the chemical state of NbF5 in the Ca-RHC+0.1NbF5 samples. NbF5 and NbB2 were measured as reference compounds, in order to compare their XANES structures with those of the samples. As shown in Fig. 2, the valence state of niobium is already changed during the high-energy ball-milling procedure and remains stable upon further hydrogen desorption and absorption cycles.
Comparing the chemical state of the samples at different hydrogenation states with the reference Nb compounds, NbB2 matches best with the XANES structure of the samples. Hence, a change of the oxidation state of niobium during milling from Nb5+ to Nb2+ can be concluded, which is also in good agreement with previous results obtained for lithium-containing RHC systems (Pranzas et al., 2011).
The EXAFS region of the XAS spectra was taken to reveal the local structure of Nb in the Ca-RHC matrix. In Fig. 3, the EXAFS spectra of the samples and the references (NbF5 and NbB2) are normalized by Δμ0 [see equation (3)] and are presented in the k2-weighted space. The EXAFS spectra of the samples are out of phase with respect to the EXAFS spectrum of NbF5 and in phase with the EXAFS spectrum of NbB2, indicating the removal of fluorine atoms and the presence of boron atoms.
The EXAFS amplitudes of the samples follow in the low-k range (<0.8 nm−1) the EXAFS amplitudes of NbB2 and they show relatively high magnitude. In the higher k region, however, this is not the case, indicating a low degree of ordering (nanocrystallite). In the lower k region, the amplitudes of the EXAFS signals of the samples grow with hydrogen release and uptake, which indicates a higher degree of ordering with the de/rehydrogenation process. To reveal the atoms surrounding the central atom (Nb), the EXAFS function of the first absorption (1st-abs) of the Ca-RHC+0.1NbF5 sample, representative for other samples, was fitted. The fitting model was based on nanocrystallite NbB2 with a hexagonal crystal structure and space group P6/mmm. All fitting procedures were carried out in R space and the results are shown in Fig. 4.
The radial distribution about the central atoms (Nb) in the Ca-RHC matrix (see Fig. 4) could not be fitted at any hydrogenation state by assuming fluorine atoms in any nearest neighbor shell, whereas the radial distribution about the central atoms could be fitted nicely (R factor = 0.033) up to the second shell by assuming hexagonal NbB2. The corresponding quantitative results are summarized in Table 3. The coordination numbers (N) were fixed to the model structure. The radial deviations (ΔR) between the fit and the model structure and their mean square disorder (σ2) in R, energy shift (ΔE) and the amplitude S02 were varied to find the best fit parameters. The R factor gives the degree of misfit of the theoretical calculated curve compared with the measured curve.
The first nearest neighbor (NN) is boron, which shows about 5% higher distance to the absorbing atom compared with the model with a disorder value of 0.5%. The second NN is Nb with a distance deviation of 0.6% and a disorder value of 0.03%. These values are somewhat smaller than those of the first NN, which should usually grow because of the thermal disorder of the material. However, Nb has about nine times higher mass than boron and therefore the thermal vibrations do not affect it as much as the boron atom. The double scattering path of B1-B1 had to be included to improve the fit parameters significantly, ensuring positive definite values of the mean square disorder. The second amplitude of Nb could not be fitted perfectly because its signal was very weak and broad, hinting at a nanocrystalline-structured NbB2 phase in the RHC. In summary, the XAS results show the absence of fluorine in the nearest neighborhood of niobium atoms and clarify the presence of the NbB2 phase. These results are in excellent accordance with other studies by Bonatto Minella et al. (2013), Deprez, Muñoz-Márquez et al. (2010) and Bösenberg et al. (2009), who also reported the formation of transition metal diborides during ball milling of complex borohydrides with transition metal fluorides, chlorides or isopropoxides. In order to determine the size distribution of NbB2 structures, Nb ASAXS measurements were performed.
ASAXS measurements were performed at the K absorption edge of niobium (18.98 keV) and calcium (4.038 keV). An energy-dependent background caused by inelastic resonant Raman scattering and fluorescence at the high q values (Sparks, 1974; Tulkki & Aberg, 1982) was subtracted from all ASAXS curves as described by Vainio et al. (2007) before further analysis. All ASAXS curves were fitted by using SASfit (http://kur.web.psi.ch/sans1/SANSsoft/sasfit.html ) assuming a polydisperse spherical particle model. Since the crystal growth processes usually yield a lognormal distribution of the particle dimensions (Bergmann & Bill, 2008), a lognormal distribution was used. The average size parameter μ and the width parameter σ of the distribution were varied to obtain the best fit to the experimental data. The simple model of polydisperse spheres used is a first approximation to the complex real RHC system. However, for a comparative study of the relative structural changes caused by different parameters in the preparation process, which is the aim of this study, the used model is sufficient. In order to characterize the Nb-containing structures in the Ca-RHC systems, ASAXS measurements at the K-edge of niobium were carried out. The total scattering curves of the as-milled Ca-RHC+0.1NbF5 sample and pure am-Ca-RHC are presented in Fig. 5(a).
The scattering curves of all Ca-RHC samples with the addition of 10 mol% NbF5 show a shoulder in the q range between 0.1 and 1 nm−1 and another one centered at ∼3 nm−1, which are not observed in the pure as-milled Ca-RHC curve. In order to emphasize the structures of the ASAXS curves of the milled Ca-RHC+0.1NbF5 sample, their intensities were weighted by qα (α = 2.7 was found to be optimal for visualization of the energy dependency of the ASAXS curves) and the result is displayed in Fig. 5(b). The q-weighted ASAXS scattering curves reveal two resonant structures, located in the lower q region (0.1–1 nm−1) and at very high q values (∼3 nm−1). However, the shoulder in the region 0.1–1 nm−1 is less energy dependent than the resonant structures distributed around 3 nm−1. Hence these q regions can be considered as two distinct Nb-containing matrix structures. Both regions were simultaneously fitted by two independent volume-weighted size distributions and their theoretical intensities were superimposed. The resultant fit and the corresponding volume-weighted size distribution function for the total scattering curve is presented in Fig. 6.
The small distribution (zoomed inset of Fig. 6b) represents the volume-weighted size distribution of Nb-containing structures with mean sphere diameters of ∼1 nm and maximum sphere diameters of ∼3 nm. The larger distribution, centered at about 4.5 nm, represents the size distribution of larger Nb-containing matrix structures showing mean sphere diameters of ∼9 nm and maximum sphere diameters of ∼24 nm. To cross check the result of the Nb size distribution, the energy-dependent (resonant) scattering of niobium was separated from the total scattering of Ca-RHC+0.1NbF5 samples using the method (not given here) described by Goerigk et al. (2004). The results obtained (mean diameters of ∼10 nm) were in good agreement with the results presented above.
ASAXS measurements were also performed at the K absorption edge of calcium (4.038 keV) to determine the Ca-containing structures in the matrix. The resonant scattering curves of the Ca-containing phase were separated by using the method described by Goerigk & Mattern (2009). Fig. 7(a) shows, exemplarily, the resonant scattering curves and their corresponding fit curves for pure Ca-RHC and Ca-RHC+10 mol% NbF5 samples in the desorbed state.
All resulting size distributions of the Ca-containing structures from the fits were normalized by their area and are presented in Fig. 7(b). A comparison of the curves of the samples with and without additive (Fig. 7b) shows that the calcium structures in the desorbed and absorbed samples with 10 mol% NbF5 are much finer. The degree of polydispersity in the desorbed and absorbed samples is also considerably lower than the corresponding pure Ca-RHC samples. The mean sizes of the calcium-containing structures after desorption and absorption in the pure Ca-RHC are approximately 54 and 68 nm, respectively. This corresponds to a relative coarsening of the mean size of Ca-containing particles of 26% in comparison to the desorbed state. The maximum sizes of the Ca structures after the first desorption and the first absorption of the pure Ca-RHC remain roughly stable and their sizes amount to ∼500 nm. Likewise the degree of polydispersity of the absorbed pure Ca-RHC sample increases slightly from σdes = 0.653 to σabs = 0.655. The average sizes in the as-milled Ca-RHC+0.1NbF5 are about 22 nm, which grow after first desorption and first absorption, respectively, to about 27 and 30 nm. This corresponds to an average size growth of 18% after the first desorption and 10% after the first absorption. The maximum structures presented in the as-milled Ca-RHC+0.1NbF5 sample are ∼80 nm, which coarsen after the first desorption and first absorption to 100 and 140 nm, respectively. This relates to coarsening degrees of 20 and 28% after the first desorption and first absorption, respectively. Also an increasing degree of polydispersity with the sorption cycle is observed for the Ca-RHC+0.1NbF5 samples (σam = 0.447 → σdes = 0.492 → σabs = 0.541).
Volumetric measurements revealed a significant impact of the NbF5 additive on the onset temperature of the hydrogen desorption and on the kinetics of the Ca-RHC system. In fact, not only does the hydrogen desorption reaction of the doped material start at a lower temperature but the overall process takes place in a much shorter period of time (Fig. 1). The calcium-containing nanostructure of the RHC was studied in detail using Ca ASAXS (Fig. 7). The results reveal that the Ca-RHC samples doped with 10 mol% NbF5 have a lower degree of polydispersity and considerably smaller Ca-containing structures compared with pure Ca-RHC (Fig. 7). After the first desorption and the first absorption, the most frequent sizes of the Ca structures in the Ca-RHC+0.1NbF5 sample are smaller by ∼50% and 44%, respectively, in comparison to those present in the pure Ca-RHC. The degree of polydispersity of the doped samples in the desorbed and absorbed states compared to the corresponding values for pure Ca-RHC samples is lower by ∼25 and ∼20%, respectively. The largest structures after the first desorption and absorption are likewise smaller in the doped Ca-RHC+0.1NbF5 in comparison to pure Ca-RHC by ∼20 and ∼28%, respectively. The relative degree of coarsening of mean particle sizes from the first desorption to the first absorption in the Ca-RHC+0.1NbF5 samples increases by 10%, whereas this value amounts to 20% in the pure Ca-RHC samples. However, the largest particle sizes in pure Ca-RHC remain, after the first absorption, roughly stable, while the corresponding structures in the Ca-RHC+0.1NbF5 sample experience a coarsening of 28%. From these results of Ca ASAXS it can be concluded that the smaller dimension range of Ca structures in Ca-RHC+0.1NbF5 is linked to a significantly higher reaction surface area of the Ca-containing structures with the gaseous hydrogen phase in comparison to the pure Ca-RHC. This leads to a higher interaction rate of hydrogen on the surface of the material. Also the contact area of Ca-containing structures with other solid phases is enhanced in comparison to pure Ca-RHC, leading to a reduction of the diffusion paths and resulting in a faster formation of the reaction products. Both effects significantly improve the hydrogen sorption kinetics. This correlation between structure and sorption behavior is in good agreement with work reported by Dornheim et al. (2007) and Pistidda et al. (2011). The XAS analyses reveal the presence of NbB2 particles in the as-milled Ca-RHC+0.1NbF5 (Figs. 2 and 3). Hence, a chemical reaction between NbF5 and Ca(BH4)2 can be concluded. By this reaction the Ca(BH4)2 particles are reduced and consequently the other calcium structures which are produced by that reaction. This in turn would explain the finer calcium structures in the cycled Ca-RHC+0.1NbF5 samples in comparison to corresponding pure Ca-RHC samples. XANES results showed that the chemical state of NbB2 remains stable during the de/rehydrogenation cycle (Fig. 2). According to the Nb ASAXS results, the NbB2 structures are nanosized (diameter ∼10 nm, Fig. 6b). These results are in agreement with those obtained by the EXAFS measurements (shown in Fig. 4), where a minimum reasonable signal of the scattered photoelectron is observed only from the second coordination shell of the absorbing atom (Nb), indicating nanosized NbB2 particles with high surface area. Because of this small size and the required vicinity to the reaction partner, it is assumed that the NbB2 nanoparticles are distributed along the grain boundaries of calcium-containing structures [see also Pranzas et al. (2011) for Li-RHC (2LiBH4+MgH2)]. Therefore, the NbB2 particles prevent the annealing and agglomeration of calcium nanocrystals/nanoparticles upon the phase transformation during the de/rehydrogenation processes (Dehouche et al., 2002; Huhn et al., 2005), which would explain the stability of the calcium nanostructures in doped Ca-RHC observed by Ca ASAXS. The distribution of transition metal boride additives in doped Li-RHC was investigated by Bösenberg et al. (2010) and Deprez, Justo et al. (2010). Their results showed, indeed, a distribution of additives mainly at the grain boundaries of the matrix, which supports the assumption made above. NbB2 is also known as a hard and inert material (Gu et al., 2008; Üçisik & Bindal, 1997; Otani et al., 1998); therefore it can act during the milling procedure as a grain refiner (Wang et al., 2008; Bobet et al., 2000; Dornheim et al., 2006; Huhn et al., 2005). A further effect of the formation of NbB2 nanoparticles during milling might be the prevention of agglomerate formation and cold-welding phenomena of the matrix (Suryanarayana, 2001; Suryanarayana & Al-Aqeeli, 2013; Rivoirard et al., 2003). This in turn would elucidate the smaller particle sizes in the Ca-RHC+0.1NbF5 structure in comparison to pure Ca-RHC and is in good agreement with the results obtained by Aguey-Zinsou et al. (2007).
Anomalous small-angle scattering (ASAXS) and X-ray absorption spectroscopy (XAS) were applied to study the catalytic impact of NbF5 additive on the Ca(BH4)2+MgH2 (Ca-RHC) reactive hydride composite.
Successful ASAXS measurements of Ca-RHC at the K absorption edge (4.03 keV) of calcium were performed. It was shown that the catalytic effect of 10 mol% of NbF5 on Ca-RHC can be attributed to the higher reaction surface area of the calcium-containing portion of the RHC matrix, and hence shorter diffusion paths owing to the presence of smaller nanostructures in the Ca-RHC+0.1NbF5 system in comparison to pure Ca-RHC.
The results of XAS measurements show that NbF5 reacts during the milling process with Ca(BH4)2 to form NbB2, which is stable upon hydrogen release and uptake. As a result of this reaction the sizes of the Ca(BH4)2 structures and their associated reaction products are reduced, as was observed by Nb and Ca ASAXS measurements. The stability of NbB2 over the hydrogen sorption cycle is shown by XANES results. EXAFS results show NbB2 structures only up to the second coordination shell, indicating the existence of nanosized particles (see Figs. 3 and 4). Nb ASAXS measurements at the K-edge of Nb supported the EXAFS results, by revealing nanosized NbB2 particles in the RHC matrix with a mean diameter of ∼10 nm. These nanoparticles are assumed to stabilize the calcium-containing structures by preventing the annealing and agglomeration processes of the calcium-containing structures over the hydrogen desorption/absorption cycle. This is a possible explanation of the lower polydispersity and smaller calcium structures in the Ca-RHC+0.1NbF5 samples upon the de/rehydrogenation procedure in comparison to the pure Ca-RHC system. In summary, NbF5 induces significant structural changes during the milling procedure as a result of its chemical reaction with Ca(BH4)2. NbB2 is formed, which remains stable upon hydrogen cycling, and it stabilizes the calcium nanostructures further over the de/rehydrogenation treatment.
The results of this work allow a deeper insight into the complex hydrogen sorption properties of RHCs, especially the structural effects of the transition metal fluoride additives on the RHCs. This is an important step in the optimization of this new class of hydrogen storage materials in order to meet the technical requirements of industrial applications.
†This article will form part of a virtual special issue of the journal, presenting some highlights of the 15th International Small-Angle Scattering Conference (SAS2012). This special issue will be available in early 2014.
The authors would like to acknowledge J. A. Puszkiel [Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET)] for fruitful discussions and the financial support of the BMBF (grant No. 03BV108C). This research was carried out at the synchrotron radiation source BESSY II (Helmholtz-Zentrum Berlin) and at the synchrotron ring DORIS III at DESY, a member of the Helmholtz Association (HGF).
Aguey-Zinsou, K., Ares Fernandez, J., Klassen, T. & Bormann, R. (2007). Int. J. Hydrogen Energy, 32, 2400–2407. Web of Science CrossRef CAS
Arndt, U. W. (1984). J. Appl. Cryst. 17, 118–119. CrossRef CAS Web of Science IUCr Journals
Barkhordarian, G., Klassen, T., Dornheim, M. & Bormann, R. (2007). J. Alloys Compd. 440, L18–L21. Web of Science CrossRef CAS
Bergmann, R. B. & Bill, A. (2008). J. Cryst. Growth, 310, 3135–3138. Web of Science CrossRef CAS
Bobet, J.-L., Even, C., Nakamura, Y., Akiba, E. & Darriet, B. (2000). J. Alloys Compd. 298, 279–284. Web of Science CrossRef CAS
Bonatto Minella, C., Pellicer, E., Rossinyol, E., Karimi, F., Pistidda, C., Garroni, S., Milanese, C., Nolis, P., Baró, M. D., Gutfleisch, O., Pranzas, K. P., Schreyer, A., Klassen, T., Bormann, R. & Dornheim, M. (2013). J. Phys. Chem. C, 117, 4394–4403. Web of Science CrossRef CAS
Bösenberg, U., Kim, J. W., Gosslar, D., Eigen, N., Jensen, T. R., Bellosta von Colbe, J. M., Zhou, Y., Dahms, M., Kim, D. H., Günther, R., Cho, Y. W., Oh, K. H., Klassen, T., Bormann, R. & Dornheim, M. (2010). Acta Mater. 58, 3381–3389.
Bösenberg, U., Vainio, U., Pranzas, P. K., Bellosta Von Colbe, J. M., Goerigk, G., Welter, E., Dornheim, M., Schreyer, A. & Bormann, R. (2009). Nanotechnology, 20, 204003. Web of Science PubMed
Bunker, G. (2010). Introduction to XAFS. A Practical Guide to X-ray Absorption Fine Structure Spectroscopy. Cambridge University Press.
Cromer, D. T. (1970). J. Chem. Phys. 53, 1891. CrossRef Web of Science
Dehouche, Z., Klassen, T., Oelerich, W., Goyette, J., Bose, T. K. & Schulz, R. (2002). J. Alloys Compd. 347, 319–323. Web of Science CrossRef CAS
Deprez, E., Justo, A., Rojas, T. C., López-Cartés, C., Bonatto Minella, C., Bösenberg, U., Dornheim, M., Bormann, R. & Fernández, A. (2010). Acta Mater. 58, 5683–5694. Web of Science CrossRef CAS
Deprez, E., Muñoz-Márquez, M. A., Roldán, M. A., Prestipino, C., Palomares, F. J., Bonatto Minella, C., Bösenberg, U., Dornheim, M., Bormann, R. & Fernández, A. (2010). J. Phys. Chem. C, 114, 3309–3317. Web of Science CrossRef CAS
Djinović Carugo, K., Helliwell, J. R., Stuhrmann, H. & Weiss, M. S. (2005). J. Synchrotron Rad. 12, 410–419. Web of Science CrossRef IUCr Journals
Dornheim, M., Doppiu, S., Barkhordarian, G., Boesenberg, U., Klassen, T., Gutfleisch, O. & Bormann, R. (2007). Scr. Mater. 56, 841–846. Web of Science CrossRef CAS
Dornheim, M., Eigen, N., Barkhordarian, G., Klassen, T. & Bormann, R. (2006). Adv. Eng. Mater. 8, 377–385. Web of Science CrossRef CAS
Goerigk, G. & Mattern, N. (2009). Acta Mater. 57, 3652–3661. Web of Science CrossRef CAS
Goerigk, G., Schweins, R., Huber, K. & Ballauff, M. (2004). Europhys. Lett. 66, 331–337. Web of Science CrossRef CAS
Gu, Q., Krauss, G. & Steurer, W. (2008). Adv. Mater. 20, 3620–3626. Web of Science CrossRef CAS
Huhn, P., Dornheim, M., Klassen, T. & Bormann, R. (2005). J. Alloys Compd. 404–406, 499–502. Web of Science CrossRef CAS
Kim, Y., Hwang, S.-J., Shim, J.-H., Lee, Y.-S., Han, H. N. & Cho, Y. W. (2012). J. Phys. Chem C, 116, 4330–4334. Web of Science CrossRef CAS
Kim, Y., Reed, D., Lee, Y.-S., Lee, J. Y., Shim, J.-H., Book, D. & Cho, Y. W. (2009). J. Phys. Chem. C, 113, 5865–5871.
Königsberger, D. C. (1988). X-ray Absorption, Application, Techniques of EXAFS, SEXAFS and XANES. New York: John Wiley and Sons.
Miwa, K., Aoki, M., Noritake, T., Ohba, N., Nakamori, Y., Towata, S., Züttel, A. & Orimo, S. (2006). Phys. Rev. B, 74, 155122. Web of Science CrossRef
Otani, S., Korsukova, M. & Mitsuhashi, T. (1998). J. Cryst. Growth, 194, 430–433. Web of Science CrossRef CAS
Pistidda, C., Barkhordarian, G., Rzeszutek, A., Garroni, S., Minella, C. B., Baró, M., Nolis, P., Bormann, R., Klassen, T. & Dornheim, M. (2011). Scr. Mater. 64, 1035–1038. Web of Science CrossRef CAS
Pranzas, P. K. et al. (2011). Adv. Eng. Mater. 13, 730–736. Web of Science CrossRef CAS
Ravel, B. & Newville, M. (2005). J. Synchrotron Rad. 12, 537–541. Web of Science CrossRef CAS IUCr Journals
Rivoirard, S., de Rango, P., Fruchart, D., Charbonnier, J. & Vempaire, D. (2003). J. Alloys Compd. 356–357, 622–625. Web of Science CrossRef CAS
Schlapbach, L. & Züttel, A. (2001). Nature, 414, 353–358. Web of Science CrossRef PubMed CAS
Schuth, F., Bogdanovic, B. & Felderhoff, M. (2004). Chem. Commun. 20, 2249–2258.
Serimaa, R., Eteläniemi, V., Serimaa, O., Laitalainen, T. & Bienenstock, A. (1996). J. Appl. Cryst. 29, 390–402. CrossRef CAS Web of Science IUCr Journals
Simon, J. P. & Lyon, O. (1994). Resonant Anomalous X-ray Scattering, Theory and Applications, edited by G. Materlik, C. F. Sparks & K. Fischer, pp. 305–322. Amsterdam: North Holland.
Sparks, C. (1974). Phys. Rev. Lett. 33, 262–265. CrossRef CAS Web of Science
Stuhrmann, H. B. (1985). Adv. Polym. Sci. 67, 123–163. CrossRef CAS
Stuhrmann, S., Hütsch, M., Trame, C., Thomas, J. & Stuhrmann, H. B. (1995). J. Synchrotron Rad. 2, 83–86. CrossRef CAS Web of Science IUCr Journals
Suryanarayana, C. (2001). Prog. Mater. Sci. 46, 1–184. Web of Science CrossRef CAS
Suryanarayana, C. & Al-Aqeeli, N. (2013). Prog. Mater. Sci. 58, 383–502. Web of Science CrossRef CAS
Tulkki, J. & Aberg, T. (1982). J. Phys. At. Mol. Phys. 15, L435–L440. CrossRef CAS Web of Science
Üçisik, A. & Bindal, C. (1997). Surf. Coat. Technol. 94–95, 561–565.
Vainio, U., Pirkkalainen, K., Kisko, K., Goerigk, G., Kotelnikova, N. E. & Serimaa, R. (2007). Eur. Phys. J. D, 42, 93–101. Web of Science CrossRef CAS
Vajo, J. J., Skeith, S. L. & Mertens, F. (2005). J. Phys. Chem. B, 109, 3719–3722. Web of Science CrossRef PubMed CAS
Wang, Y., Wang, H., Yang, Y. & Jiang, Q. (2008). Mater. Sci. Eng. A, 478, 9–15. Web of Science CrossRef
Züttel, A. (2003). Mater. Today, 6, 24–33.
Züttel, A., Rentsch, S., Fischer, P., Wenger, P., Sudan, P., Mauron, P. & Emmenegger, C. (2003). J. Alloys Compd. 356–357, 515–520.
© International Union of Crystallography. Prior permission is not required to reproduce short quotations, tables and figures from this article, provided the original authors and source are cited. For more information, click here.