Effect of Transition Metal Fluorides on the Sorption ... - ACS Publications

Gutfleisch§, Torben R. Jensen⊥, Yngve Cerenius, Jeppe Christensen∥, Maria Dolores Baró‡, Rüdiger Bormann†, Thomas Klassen†, and Marti...
0 downloads 0 Views 1MB Size
ARTICLE pubs.acs.org/JPCC

Effect of Transition Metal Fluorides on the Sorption Properties and Reversible Formation of Ca(BH4)2 Christian Bonatto Minella,*,† Sebastiano Garroni,‡ Claudio Pistidda,† R. Gosalawit-Utke,† Gagik Barkhordarian,† Carine Rongeat,§ Inge Lindemann,§ Oliver Gutfleisch,§ Torben R. Jensen,^ Yngve Cerenius, Jeppe Christensen,|| Maria Dolores Baro,‡ R€udiger Bormann,† Thomas Klassen,† and Martin Dornheim† †

)

Institute of Materials Research, Materials Technology, Helmholtz-Zentrum Geesthacht Zentrum f€ur Material- und K€ustenforschung GmbH, Max Planck Str. 1, D-21502 Geesthacht, Germany ‡ Departament de Física, Universitat Autonoma de Barcelona, 08193 Bellaterra, Spain § IFW Dresden, Institute for Metallic Materials, Helmholtzstrasse 20, D-01069 Dresden, Germany ^ Interdisciplinary Nanoscience Center (iNANO) and Department of Chemistry, University of Aarhus, Langelandsgade 140, DK-8000 Aarhus C, Denmark MAX-lab, Lund University, S-22100 Lund, Sweden

ABSTRACT: Light metal borohydrides are considered as promising materials for solid state hydrogen storage. Because of the high hydrogen content of 11.5 wt % and the rather low dehydrogenation enthalpy of 32 kJ mol-1H2, Ca(BH4)2 is considered to be one of the most interesting compounds in this class of materials. In the present work, the effect of selected TM-fluoride (TM = transition metal) additives on the reversible formation of Ca(BH4)2 was investigated by means of thermovolumetric, calorimetric, Fourier transform infrared spectroscopy, and ex situ, and in situ synchrotron radiation powder X-ray diffraction (SR-PXD) measurements. Furthermore, selected desorbed samples were analyzed by 11B{1H} solid state magic angle spinning nuclear magnetic resonance (MAS NMR). Under the conditions used in this study (145 bar H2 pressure and 350 C), TiF4 and NbF5 were the only additives causing partial reversibility. In these two cases, 11B{1H} MAS NMR analyses detected CaB6 and likely CaB12H12 in the dehydrogenation products. Elemental boron was found in the decomposition products of Ca(BH4)2 samples with VF4, TiF3, and VF3. The results indicate an important role of CaB6 for the reversible formation of Ca(BH4)2.

’ INTRODUCTION Because of higher energy efficiency compared to liquid and gaseous hydrogen storage, compact storage in a solid medium is an attractive alternative for mobile and stationary applications.1 Due to the high hydrogen density, light metal borohydrides are considered to be promising materials with the opportunity to satisfy the necessary criteria of efficiency, stability, and safety .2 Among them, Ca(BH4)2 represents a potential candidate, due to its high gravimetric (11.5 wt %) and volumetric (∼130 kg m-3) hydrogen content.3 Furthermore, the dehydrogenation enthalpy is calculated to be 32 kJ mol-1 H2, if CaH2 and CaB6 are the decomposition products, which is within the optimal range for mobile applications.4,5 In earlier studies, three different routes for the preparation of the starting material Ca(BH4)2 are proposed: 1 Barkhordarian et al. synthesized Ca(BH 4 )2 by solidgas reaction starting from CaH 2 and MgB2 at 400 C r 2011 American Chemical Society

and 350 bar H2 for 48 h forming MgH 2 as a side product. 6 2 Ronnebro et al. were able to synthesize Ca(BH4)2 with a yield of 60% from a mixture of CaH2 and CaB6 with Pd and TiCl3 by applying 700 bar H2 and temperatures of 400440 C.7 3 Rongeat et al. showed that 19% of calcium borohydride was obtained by high-pressure reactive ball milling (near room temperature) of CaH2 and CaB6 after 24 h at 140 bar H2, employing TiF3 or TiCl3 as additives. This yield was improved to 60% during further cycling of the material at 350 C and 90 bar H2 for 40 h.8

Received: August 17, 2010 Revised: December 7, 2010 Published: January 12, 2011 2497

dx.doi.org/10.1021/jp107781m | J. Phys. Chem. C 2011, 115, 2497–2504

The Journal of Physical Chemistry C

ARTICLE

Three possible decomposition pathways are described in literature:5,9 CaðBH4 Þ2 T 2=3CaH2 þ 1=3CaB6 þ 10=3H2 ð9:6wt%H2 Þ CaðBH4 Þ2 T CaH2 þ 2B þ 3H2

ð8:7wt%H2 Þ

ðAÞ ðBÞ

CaðBH4 Þ2 T 1=6CaB12 H12 þ 5=6CaH2 þ 13=6H2 ð6:3wt%H2 Þ

ðCÞ

While the enthalpy of reaction A, determined by first principle method, is 32 kJ mol-1 H2, the one for reaction B, obtained by density functional theory (DFT) calculations, is 56 kJ mol-1 H2.5 First principles method and DFT calculations predicted for pathway C a free energy value close to the one of reaction A but leading to the intermediate phase CaB12H12.5,9 The enthalpy of reaction for this pathway was calculated to be 34-36 kJ mol-1 H2.5,9 Experimentally, the thermal decomposition reaction of calcium borohydride involves two dehydrogenation steps. The first decomposition step starts around 350 C and leads to the formation of CaH2 and an unknown intermediate phase that decomposes further, in the second step, in the temperature range of 390-500 C.10 Ca(BH4)2 has several structural polymorphs. First of all, there are two low temperature modifications, R (space group F2dd) and γ (PBCA), recently indexed as orthorhombic phases.3,11 These room temperature structures transform to a high temperature phase β in the 180-300 C temperature range.11 Moreover, another polymorph called δ was found upon heating together with a not yet indexed phase, still stable at 500 C.12 The formation of Ca(BH4)2 was shown to be partially reversible by using suitable additives.7 Lately, some TM-fluorides and chlorides (TiCl3 and NbF5) have demonstrated to positively affect its partial reversible formation. Over 50% hydrogen can be reversibly absorbed when 90 bar H2 pressure and 350 C are applied for 24 h to the decomposition products of calcium borohydride catalyzed by TiCl3 or NbF5.13,14 A positive kinetic effect of TM-fluorides on MgH2 was demonstrated due to the formation of MgF2 and TM-nanoparticles inside the MgH2 powder.15 Furthermore, the H-/F- substitution within the sodium aluminum hydride turned out positively tuning the thermodynamic because the two anions are isostructural.16-18 However, a comprehensive study of the sorption and structural properties of calcium borohydride with additives was not presented so far. In this work, a scanning of the effect of several selected TM-fluorides on the hydrogen sorption properties and the reversible formation of Ca(BH4)2 was performed.

’ EXPERIMENTAL METHODS Ca(BH4)2 was obtained by drying the Ca(BH4)2-2THF adduct purchased from Sigma-Aldrich, for 1 h and 30 min at 200 C in vacuum and subsequent cooling to room temperature. Fourier transform infrared spectroscopy (FTIR) using a Bruker Quinox 55 Spectrometer applied after drying the as-received material, confirmed the full removal of the solvent and the presence of calcium borohydride only. Furthermore, FT-IR was employed to verify the presence of [BH4]- after hydrogen absorption. Five different samples were prepared adding 0.05 mol of TiF3 (unknown purity), TiF4 (purity 98%), VF3 (purity 98%), VF4 (purity 95%), and NbF5 (purity 99%), purchased by Alfa Aesar,

to Ca(BH4)2. The samples were milled in a stainless steel vial in an argon atmosphere for 1 h and 40 min using a Spex Mixer Mill (model 8000) and 14:1 as ball to powder ratio (four spheres of 3.5 g each one and 1 g of powder) All powder handling and milling was performed in an MBraun argon box with H2O and O2 levels below 10 ppm to prevent contamination. Differential scanning calorimetry as well as mass spectrometry measurements were carried out using a Netzsch STA 409 C in 150 mL/min argon flow. The samples were investigated in the range of 25-500 C for the samples with additives and up to 550 C for the pure calcium borohydride. The heating rate was for both the measurements 5 K/min. Sorption properties and kinetics were evaluated by volumetric measurements using a Sievert apparatus designed by Hydro Quebec/ HERA Hydrogen Storage System. The milled powders were desorbed by heating from room temperature (25 C) to 450 C in static vacuum (2 kPa the starting pressure value) and afterward reabsorbed at 350 C and 145 bar H2 for 20 h. Transmission X-ray diffraction measurements were performed in 0.7 mm capillaries on a Stoe Stadi P (Mo KR) in DebyeScherrer geometry. The diffractometer is equipped with a curved Ge (111) monochromator and a 6 linear position sensitive detector with a resolution of about 0.06 2θ at full width-half-maximum (fwhm). In situ Synchrotron radiation powder X-ray diffraction (SR-PXD) was carried out at the Synchrotron MAX II, Lund, Sweden, at the beamline I711 with a MAR165 CCD detector. The selected X-ray wavelengths were 1.09994 (Ca(BH4)2 þ TiF4), 1.072 (Ca(BH4)2 þ NbF5), 0.979, and 1.09801 Å in case of pure dried Ca(BH4)2 and 0.939 Å for the samples measured after milling. For comparison, all XRD data are reported referring to the scattering vector 4π sin θ/λ. The samples were introduced in a single crystal sapphire tube in an argon-filled glovebox with H2O and O2 levels below 0.1 ppm. The XRD exposure time was 30 s (15 s twice) per powder diffraction pattern. In all experiments the heating rate was 5 K/min and samples were heated up to 450 C. The MAUD software was used for the evaluation of abundance of phases by XRD patterns using Rietveld method.19 Solid state magic angle spinning (MAS) NMR spectra were obtained using a Bruker Avance 400 MHz spectrometer with a wide bore 9.4 T magnet and employing a boron-free Bruker 4 mm CPMAS probe. The spectral frequency was 128.33 MHz for the 11 B nucleus and the NMR shifts are reported in parts per million (ppm) externally referenced to BF3Et2O. The powder materials were packed into 4 mm ZrO2 rotors in an argon-filled glovebox and were sealed with tight fitting Kel-F caps. The one-dimensional (1D) 11B{1H} MAS NMR spectra were acquired after a 2.7 μs single π/2 pulse (corresponding to a radiofield strength of 92.6 kHz) and with application of a strong 1H signal decoupling by using the two-pulse phase modulation (TPPM) scheme. The spectra were recorded at a MAS spinning rate of 12 kHz. Sample spinning was performed using dry nitrogen gas. The recovery delay was set to 10 s. Spectra were acquired at 20 C (controlled by a BRUKER BCU unit).

’ RESULTS AND DISCUSSION After removal of tetrahydrofuran, SR-PXD revealed that dried calcium borohydride contains a mixture of low temperature R, γ, and high temperature β-phase (Figure 1). The relative abundance, determined by Rietveld method, is 23, 7, and 70 wt %, respectively. 2498

dx.doi.org/10.1021/jp107781m |J. Phys. Chem. C 2011, 115, 2497–2504

The Journal of Physical Chemistry C

ARTICLE

Figure 1. XRD pattern of Ca(BH4)2 after THF removal: R-Ca(BH4)2 (R); β-Ca(BH4)2 (β); γ-Ca(BH4)2 (γ). Figure 3. DSC (-) and MS of hydrogen (0) curves of Ca(BH4)2 at 150 mL/min argon flow.

Figure 2. DSC curves at 150 mL/min argon flow of Ca(BH4)2 (a), Ca(BH4)2 þ TiF3 (b), Ca(BH4)2 þ TiF4 (c), Ca(BH4)2 þ VF3 (d), Ca(BH4)2 þ VF4 (e), and Ca(BH4)2 þ NbF5 (f).

The thermal decomposition of the milled samples was studied by differential scanning calorimetry at 150 mL/min argon flow and the results are shown in Figure 2. Figure 2a is the DSC curve of the nonmilled Ca(BH4)2. It shows a weak endothermic signal at 165 C corresponding to the phase transformation from low temperature phases (R, γ) to the β-phase. Mass spectrometry (Figure 3) evidences no hydrogen gas evolution during this transformation. The first hydrogen desorption step starts around 350 C, followed by the second one in the range of 390 and 500 C. The signal at 165 C related to the phase transition, present in the pure nonmilled Ca(BH4)2, does not occur in the milled samples (Figure 2, curves b-f). This is caused by the high mechanical energy provided by the milling, which transforms almost all the low temperature phases into the high temperature β-phase.13,20 As Figure 2 shows, there is a clear effect of the investigated fluoride additives on the shape and the onset temperatures of the peaks of the first and the second hydrogen desorption steps. The latter one, in particular, for all the samples, becomes broader compared to that of the nonmilled materials (curve a) and shifts to lower temperatures. The broadening can be associated with the reduced crystallite size of the materials due to the milling. Curves f and c, belonging to the samples milled with NbF5 and TiF4, respectively, show that the first desorption step takes place at significantly lower temperature than in case of the one of the nonmilled Ca(BH4)2 as well. In particular, with NbF5, desorption starts below 300 C. Such a shift in the desorption temperatures

Figure 4. XRD of the samples after ball milling: Ca(BH4)2 þ TiF4 (a), Ca(BH4)2 þ NbF5 (b), Ca(BH4)2 þ VF3 (c), Ca(BH4)2 þ TiF3 (d), Ca(BH4)2 þ VF4 (e). γ-Ca(BH4)2 (γ); β-Ca(BH4)2 (β); CaF2 (]); TiF3 (O O).

is not found in case of the tested vanadium based fluoride additives (curves d and e). In this case, the first desorption step is not shifted toward lower temperatures, but is even delayed. X-ray diffraction patterns for all the milled materials are presented in Figure 4. All the samples indicate the presence of γ-Ca(BH4)2 and β-Ca(BH4)2 in different abundances. The curves a and b, related to the samples milled with TiF4 and NbF5, besides the two polymorphs (γ and β), show reflections of CaF2 (PDF No. 35-819). Its presence hints to a reaction between the TM-fluoride and the borohydride already during milling. In contrast, in pattern d, we can still find peaks of TiF3 (ICSD No. 28783), indicating that no reaction has taken place in the vial. Patterns c and e, with VF3 and VF4, respectively, present neither traces of CaF2 nor of the TMfluorides. Figure 5 shows the desorption curves of the different samples with and without additives, obtained by thermovolumetric measurements. The figure shows that the addition of TM-fluorides changes the desorption kinetics: in pure calcium borohydride, the hydrogen starts to be released around 350 C, while in the samples with additives it begins already in the range of 125-225 C. However, DSC analyses do not show endothermic signals at such low temperatures. So far, the dissimilar behavior can only be linked to the different experimental conditions between DSCs (150 mL/min argon flow) and volumetric analysis (2 kPa the starting pressure value). 2499

dx.doi.org/10.1021/jp107781m |J. Phys. Chem. C 2011, 115, 2497–2504

The Journal of Physical Chemistry C

Figure 5. Volumetric measurements showing the desorption curves over the temperature: temperature (` black), Ca(BH4)2 (; black), Ca(BH4)2 þ NbF5 (þ red), Ca(BH4)2 þ TiF3 (f green), Ca(BH4)2 þ TiF4 (O dark yellow), Ca(BH4)2 þ VF3 (4 magenta), Ca(BH4)2 þ VF4 (0 blue). Experiments were carried out by heating the samples from room temperature up to 450 C in static vacuum (starting value 2 kPa).

Figure 6. XRD (Mo KR = 0.7107 Å) of the samples collected at the plateau during the desorption performed at 450 C in vacuum: Ca(BH4)2 þ TiF3 (a); Ca(BH4)2 þ VF3 (b); Ca(BH4)2 þ VF4 (c); Ca(BH4)2 þ NbF5 (d); Ca(BH4)2 þ TiF4 (e); δ-Ca(BH4)2 (δ); unidentified phase (?); CaF2-xHx (9).

In case of the samples milled with TiF3, VF3, and VF4, the desorption stops for a limited time (≈20 min) in the range 360-415 C and 3-4.5 wt % desorbed hydrogen. After this incubation period, the hydrogen desorption proceeds again until the decomposition products. Such an incubation period might be caused by a necessary nucleation and growth process of an intermediate, amorphous or nanocrystalline, calcium borohydride phase as observed in other systems.21,22 This, however, would implicate that by TiF3, VF3, and VF4 the reaction path is altered if compared to pure Ca(BH4)2 where such an incubation period is not observed. XRD measurements at the plateau, performed for all the materials with additives, are reported in Figure 6. Concerning the samples with NbF5 and TiF4, curves d and e, respectively, they were collected stopping the experiment when half of their hydrogen content was desorbed. This can be approximately considered being in the plateau range. In Figure 6, for all the samples, two phases are visible: δ-Ca(BH4)2 and CaF2-xHx. Only the pattern “a” shows a further peak, indicated by a question mark, not corresponding to any of the phases included in the crystallographic ICSD database. To some extent, although some of the samples show an incubation period during desorption, the X-ray patterns do not evidence relevant differences.

ARTICLE

Figure 7. FT-IR of reabsorbed powder at 350 C and 145 bar H2 pressure: Ca(BH4)2 (reference; a); Ca(BH4)2 þ NbF5 (b); Ca(BH4)2 þ TiF3 (c); Ca(BH4)2 þ TiF4 (d); Ca(BH4)2 þ VF3 (e); Ca(BH4)2 þ VF4 (f).

As visible in Figure 5, the material doped with NbF5 and TiF4 desorbs at markedly lower temperatures than pristine and doped Ca(BH4)2. These samples (milled with NbF5 and TiF4) do not show any incubation period. This could be due to a better distribution of heterogeneous, amorphous, or nanocrystalline nucleation agents, as suggested by B€osenberg and Deprez for other systems.21,22 The positive kinetic effect of NbF5, it has been recently explained by Kim et al.24 Due to its low melting point (79 C), NbF5 melts during milling. This leads to a better dispersion of the molten NbF5 in the Ca(BH4)2.23,24 Although, the melting point of TiF4 is well above 100 C (maximum temperature attainable inside a spex mill vial during milling),25 it still shows a positive kinetic effect during reabsorption. The pure calcium borohydride desorbs 8.3 wt % hydrogen in 3.5 h under the applied conditions. This value is closer to the theoretical one reported for reaction B, predicting the formation of pure boron and CaH2 as decomposition products. The samples with additives desorb less hydrogen because of the formation of some side products like CaF2 (Figure 4, curves a and b), which is formed during milling due to an irreversible reaction between the borohydrides and the fluoride additives, and CaO, present among the decomposition products. The products of the desorption were subsequently reabsorbed at 145 bar H2 pressure and 350 C for ∼20 h. Reabsorbed powders were analyzed by FTIR again to detect whether the [BH4]anion was formed. Figure 7 shows the FTIR spectra of all the materials milled with additives together with pure Ca(BH4)2 as reference. Note that only the NbF5 (spectrum b) and TiF4 (spectrum d) additives were effective in promoting the reversible formation of calcium borohydride. The FTIR signals in the B-H stretching and bending regions are rather broadened but match those of the Ca(BH4)2 reference indicating reversibility. In contrast, none of the samples with VF3, VF4, or TiF3 additions, which show the incubation period during decomposition, show formation of B-H bonds. To our knowledge, this is the first time that the positive effect of TiF4, in promoting the reversible formation of a borohydride, is shown. To obtain quantitative information about the amount of hydrogen reversibly absorbed, subsequent desorption tests on the reabsorbed powders were performed. Figure 8 shows thermovolumetric measurements of the second hydrogen desorption cycle of the samples milled with TiF4 and NbF5. As can be seen, the sample with TiF4 desorbed 2.6 wt %, whereas the sample with 2500

dx.doi.org/10.1021/jp107781m |J. Phys. Chem. C 2011, 115, 2497–2504

The Journal of Physical Chemistry C

Figure 8. Desorption curves after reabsorption at 350 C and 145 bar H2 pressure: temperature (` black), Ca(BH4)2 þ TiF4 (O dark yellow), Ca(BH4)2 þ NbF5 (þ red). Experiments were carried out by heating the samples from room temperature up to 450 C in static vacuum (starting value 2 kPa).

NbF5 desorbed 3.6 wt % hydrogen, corresponding to 35 and 55% reversibility, respectively. Figure 8 also shows distinct changes of the slopes of both desorption curves, typical for multistep desorption processes. In situ SR-PXD was employed to obtain a better understanding of the reactions involved in the samples with NbF5 and TiF4. The results are reported in Figures 9 and 10, respectively. Desorption reactions were studied under static vacuum by heating from room temperature up to 450 C. Rietveld refinement of the pattern collected at 30 C indicates that the initial powder is composed of 72 wt % low temperature polymorph γ-Ca(BH4)2, 24 wt % high temperature β-Ca(BH4)2, and 4 wt % CaF2 (PDF No. 35-819). The latter one must have been formed by a reaction between Ca(BH4)2 and NbF5. This might partly explain why the total amount of hydrogen desorbed from the samples with additives is lower than in case of the pure calcium borohydride. The pattern measured at 282 C shows that there are no traces of γ-phase anymore, because it is already transformed into the β-phase. At this temperature another calcium borohydride polymorph appears, called δ-phase in a previous work.12 In that study, it was reported that hydrogen desorption from β-Ca(BH4)2 occurs in the range of 330 to 380 C, whereas δ-Ca(BH4)2 releases hydrogen at higher temperature, in the range of 380500 C.12 When TM-fluorides are added to calcium borohydride, its kinetics of desorption is modified and the thermal events are shifted. As can be seen in Figure 9, the first hydrogen desorption step takes place between 282 and 361 C and involves only the β-phase. In this range, most of the hydrogen is desorbed with some residual being present inside the δ-phase. This phase fraction grows continuously up to 361 C, as can be inferred from the increase of the intensity of its diffraction peaks. At this temperature, CaF2 is still present and coexists with the δ-Ca(BH4)2 phase. The residual hydrogen is released in the 361-450 C temperature range. CaH2 should be formed after the first desorption event, however, our measurements do not show its formation up to 361 C. At 450 C, CaO and an unidentified phase form. The latter decomposes after a short time. Later on, the last recorded XRD pattern at 450 C shows an increase of the peak intensity at the scattering vector value of 2.23. This peak corresponds to the (200) peak of the CaF2-xHx phase,26 which is formed by a reaction of CaH2 and CaF2 and can explain the absence of any CaH2 reflections. The ICSD database matches those peaks with an

ARTICLE

Figure 9. SR-PXD patterns of Ca(BH4)2 milled with NbF5. The experiment was carried out by heating in vacuum from RT up to 450 C with 5 K/min as heating rate and 10 min isotherm (at 450 C): γ-Ca(BH4)2 (γ); β-Ca(BH4)2 (β); CaF2 (]); δ-Ca(BH4)2 (δ); unidentified phase (f); CaO (X); CaF2-xHx (9).

Figure 10. SR-PXD patterns of Ca(BH4)2 milled with TiF4. The experiment was carried out by heating in vacuum from RT up to 450 C with 5 K/min as heating rate and 10 min isotherm (at 450 C): γ-Ca(BH4)2 (γ); β-Ca(BH4)2 (β); CaF2 (]); δ-Ca(BH4)2 (δ); unidentified phase (f); CaO (X); CaF2-xHx (9); CaB6 (tilted square with a center dot).

exact stoichiometry CaF0.76H1.24, however, the current experimental results are not conclusive because not all of the peaks are present. Another unexpected observation is the increase in the diffraction peak intensity of CaO. The increase in the intensity of the (200) reflection of CaF2-xHx phase and the formation of CaO takes place after decomposition of the unidentified phase which releases calcium. Even though the samples were handled in an inert atmosphere, formation of CaO could not be avoided. This might be due to a leak present in the capillary used for the experiment or to the reaction with the sapphire tube. Figure 10 shows the diffraction patterns for the sample milled with TiF4. Rietveld refinement of the diffractogram at 30 C indicates that the starting material is composed of 14 wt % γCa(BH4)2, 81 wt % β-Ca(BH4)2, and 5 wt % CaF2 (PDF No. 35819). Formation of CaF2 again causes a reduction of the total hydrogen capacity of the mixture. The pattern measured at 245 C shows the peaks of δ-Ca(BH4)2, but does not show any reflection of γ-Ca(BH4)2, which has totally transformed into β at this temperature. The sample milled with TiF4 begins its first hydrogen desorption step at 245 C. This first desorption step ends slightly above 265 C. At this temperature δ-Ca(BH4)2 coexists with CaF2 and the unidentified 2501

dx.doi.org/10.1021/jp107781m |J. Phys. Chem. C 2011, 115, 2497–2504

The Journal of Physical Chemistry C

Figure 11. Rietveld refinement of the SR-PXD pattern of Ca(BH4)2 þ TiF4 sample after decomposition (λ = 1.09994 Å). Experimental values (-) are reported together with Rietveld fitting (0).

phase. This phase is exactly the same, which was observed in the diffraction pattern of Ca(BH4)2 þ NbF5 (Figure 9). The second desorption step is slower and takes place in the wider temperature range of 265-340 C. After decomposition of δ-Ca(BH4)2 as well as β-Ca(BH4)2, no traces of CaH2 were detected. The diffraction pattern measured at 340 C shows the presence of CaO and the unidentified phase. The latter decomposes at 450 C and the peak at the scattering vector value of 2.24 shows higher intensity, which hints at the formation of CaF2-xHx, a reaction product of CaH2 and CaF2. Such a nonstoichiometric Ca-F-H phase explains the absence of CaH2 reflections. As in the case of the sample with NbF5, the unidentified phase decomposes and the intensities of the peaks of CaO and CaF2-xHx increase, indicating that this unidentified phase may contain calcium. The diffractogram at 450 C, of Ca(BH4)2 milled with TiF4, suggests the presence of CaB6 among the decomposition products. Figure 11 shows the Rietveld refinement of the Ca(BH4)2 þ TiF4 sample after decomposition. The X-ray refinement depicts CaB6 as an amorphous-like curve evidencing its nanocrystalline domains. SR-PXD experiments show an identical decomposition pathway for the samples milled with TiF4 and NbF5 but still, in the latter sample, it is not clear whether CaB6 is present as decomposition product. Reactions A and B, mentioned in the Introduction, indicate that decomposition of calcium borohydride ends in boron or CaB6, but X-ray diffraction does not represent the proper tool to detect fine nanocrystalline and amorphous compounds. Therefore, 11B{1H} MAS NMR measurements were carried out to identify the nature of the final B-containing compound. 11 B{1H} MAS NMR spectra are shown in Figure 12. 11B (I = 3/2) NMR was collected at room temperature on the desorbed samples. Commercial Ca(BH4)2 by Sigma-Aldrich, CaB6 and elemental boron are included as references. Commercial Ca(BH4)2 presents two sharp lines at -30 and -32 ppm belonging to the boron atom of the [BH4]- anion. Because the starting material is composed of the two polymorphs γ and β, with different crystal structure, every peak is related to a different phase. Other experiments, that will be presented by the authors elsewhere, allow to clarify that the signal at -30 ppm corresponds to the low temperature phases R or γ (both orthorhombic), while the one at -32 ppm belongs to β-Ca(BH4)2 (tetragonal). The spectrum d (Figure 12), corresponding to Ca(BH4)2 desorbed at 450 C in vacuum, indicates the presence of a broad signal at ∼-1 ppm characteristic of elemental boron and another less intense, at -33 ppm, related to residual calcium borohydride

ARTICLE

Figure 12. 11B{1H} MAS NMR spectra at room temperature of Ca(BH4)2 purchased by Sigma-Aldrich, scale adjusted by 1/4 (a); CaB6, scale adjusted by 1/4 (b); boron, scale adjusted by 1/4 (c); Ca(BH4)2 desorbed at 450 C in vacuum (d); Ca(BH4)2 þ 0.05 mol NbF5 desorbed at 450 C in vacuum (e); Ca(BH4)2 þ 0.05 mol TiF4 desorbed at 450 C in vacuum (f); Ca(BH4)2 þ 0.05 mol VF4 desorbed at 450 C in vacuum (g). Side bands are indicated by * and ].

not reacted. Broadening in NMR signal is typical of disordered materials that isotropically distribute their chemical shifts.27 Spectrum “e” of desorbed Ca(BH4)2 þ NbF5 shows a series of three signals at ∼16, ∼-16, and ∼-30 ppm. The peak at 16 ppm corresponds to CaB6 and the one at -30 ppm evidence residual calcium borohydride. The identification of the peak at -16 ppm is more difficult. By 11B MAS NMR, Hwang et al.27 confirmed the formation of [B12H12]2- anion as intermediate compound during desorption of LiBH4 by the detection of a broad signal at -12 ppm. This value might shift in dependence either of the cation associated to the [B12H12]2- anion and the disordered nature of the phase structure.27 The formation of CaB12H12 was initially predicted by Ozolins et al.9 before, and lately Wang et al.28 determined, by first principles calculation, the existence of several CaB12H12 intermediates. Their calculations revealed that these amorphous like structures are energetically favorable during desorption of Ca(BH4)2 and lead to its limited reversibility.28 So far we could not measure CaB12H12 reference samples by 11B{1H} MAS NMR, but the recent work of Hwang and Wang, together with our experimental results, strongly suggest the presence of CaB12H12 among the decomposition products. The Ca(BH4)2 þ TiF4 sample desorbed at 450 C in vacuum shows the same series of three broad signals similar to the sample milled with NbF5, evidencing CaB6, residual Ca(BH4)2 and CaB12H12. The 11B{1H} MAS NMR spectrum of Ca(BH4)2 þ VF4 shows a profile comparable to that of desorbed pure calcium borohydride. The pattern indicates the presence of elemental boron and residual Ca(BH4)2 as can be observed by the shoulder at -34 ppm. As can be seen in Figure 7, TiF4 and NbF5 were the only additives capable to promote the reversible formation of calcium borohydride, even though partial. 11B{1H} MAS NMR shows that they lead to the same decomposition products: CaB6 and CaB12H12. In contrast, elemental boron was found as decomposition product in the samples milled with VF4 (Figure 12), TiF3, and VF3 (not reported here). These three TM-fluoride additives do not have any positive influence on the reversible formation of calcium borohydride under these applied conditions. As reported in literature, the presence of CaB6 in calcium borohydride decomposition products is beneficial for reversibility, while formation of elemental boron does not lead to reversibility.29 DFT calculations demonstrated that the energies required to break CaB6 and R-boron into isolated atoms are 6.00 and 6.20 eV/atom, respectively. Furthermore, it is well-known 2502

dx.doi.org/10.1021/jp107781m |J. Phys. Chem. C 2011, 115, 2497–2504

The Journal of Physical Chemistry C that boron, once formed, is hardly reversible due to a high kinetic barrier. Recently, Barkhordarian et al, showed that no formation of calcium borohydride was achieved using pure boron and CaH2 at 400 C and 350 bar H2 for 24 h.6 To some extent, in this work, seems that CaB6 has a lower kinetic barrier than pure boron. This results in the partial reversible formation of Ca(BH4)2. The 11B NMR pattern (d) shows that Ca(BH4)2 without additives presents elemental boron and residual calcium borohydride as decomposition products. No traces of CaB12H12 were detected. Calcium borohydride desorbed at 450 C in vacuum was rehydrogenated under 130 bar H2 at 350 C for 24 h, however, SR-PXD analysis does not show any Ca(BH4)2 diffraction peaks. Our experiments, performed at higher hydrogen pressures, demonstrated no reversible formation in pure Ca(BH4)2 either. This, again, is ascribed to the presence of elemental boron. TiF4 and NbF5 additives led to the formation of CaB6 and CaB12H12 as decomposition products belonging to reactions A and C, which have close reaction free energy values. The TMFluoride additives do not lead to a specific decomposition pathway for Ca(BH4)2, but part of it follows reaction A, while the other proceeds through reaction B. CaB6 represents a kinetically more favorable decomposition product for the subsequent absorption reaction while the rather stable CaB12H12 phase certainly is responsible for the limited reversibility.

’ CONCLUSIONS A comprehensive investigation of the effect of selected TMfluorides on the sorption properties of calcium borohydride was carried out. Calorimetry, FTIR, volumetric measurements, SRPXD, and solid state 11B{1H} MAS NMR were employed to study the dehydrogenation pathways and characterize the decomposition products. 11 B{1H} MAS NMR showed that Ca(BH4)2 decomposes in CaH2 and elemental boron. Rehydrogenation attempts of the decomposition products did not succeed in the reversible formation of [BH4]- at 130 bar H2 and 350 C for 24 h. When TM-fluorides were added to Ca(BH4)2, CaF2 was formed as side product contributing to reduce the hydrogen content of the mixtures. Formation of CaF2-xHx was detected in the decomposition products due to the reaction between CaH2 and CaF2. Elemental boron was formed after desorption of calcium borohydride milled with VF4, TiF3, and VF3. Rehydrogenation, under 145 bar H2 at 350 C for several hours, did not show reversible formation of calcium borohydride. CaB6 and CaB12H12 were formed after hydrogen desorption of Ca(BH4)2 milled with TiF4 and NbF5. These two compounds are products of reactions A and C that are competing decomposition pathways. Rehydrogenation, under 145 bar H2 and 350 C for several hours, shows reversible formation of calcium borohydride in these two samples. This work shows for the first time the positive effect of TiF4 in promoting such a reversible reaction. CaB6 seems to play a key role in the reversible hydrogenation reaction of Ca(BH4)2. Its positive function in promoting the reversibility compared to elemental boron, known as scarcely reversible, is the reason for the observed effects. The role of CaB12H12 could not be explained completely. Its formation might be the reason for partial reversibility. Further investigations are in progress to clarify this point.

ARTICLE

’ ACKNOWLEDGMENT The authors are grateful to Marie-Curie European Research Training Network (contract MRTN-CT-2006-035366/COSY) for the financial support. Dr. U. B€osenberg and Dr. T. Emmler of HZG are gratefully acknowledged by the author. The author is thankful to D. Haase (MAX-lab, I711) for the beamline support. We thank the Servei de Ressonancia Magnetica Nuclear RMN at UAB for their technical assistance. S.G. thanks P. Nolis for useful discussions. The author (M.D.B.) was partially supported by ICREA ACADEMIA award. ’ REFERENCES (1) Dornheim, M. Handbook of Hydrogen Storage. In Tailoring Reaction Enthalpies of Hydrides; Hirscher, M., Ed.; Wiley-VCH: New York, 2010. (2) http://www1.eere.energy.gov/hydrogenandfuelcells/storage/index. html. (3) Buchter, F.; yodziana, Z.; Remhof, A.; Friedrichs, O.; Borgschulte, A.; Mauron, Ph.; Z€uttel, A.; Sheptyakov, D.; Barkhordarian, G.; Bormann, R.; Chzopek, K.; Fichtner, M.; Sørby, M.; Riktor, M.; Hauback, B.; Orimo, S. J. Phys. Chem. B 2008, 112, 8042–8048. (4) Miwa, K.; Aoki, M.; Noritake, T.; Ohba, N.; Nakamori, Y.; Towata, S.; Z€uttel, A.; Orimo, S. Phys. Rev. B 2006, 74, 155122. (5) Frankcombe, T. J. J. Phys. Chem. C 2010, 114, 9503–9509. (6) Barkhordarian, G.; Klassen, T.; Dornheim, M.; Bormann, R. J. Alloys Compd. 2007, 440, L18–L21. (7) R€onnebro, E.; Majzoub, E. H. J. Phys. Chem. B 2007, 111, 12045– 12047. (8) Rongeat, C.; D’Anna, V.; Hagemann, H.; Borgschulte, A.; Z€uttel, A.; Schultz, L.; Gutfleisch, O. J. Alloys Compd. 2010, 493, 281–287. (9) Ozolins, V.; Majzoub, E. H.; Wolverton, C. J. Am. Chem. Soc. 2009, 131, 230–237. (10) Kim, J.-H.; Jin, S.-A.; Shima, J.-H.; Cho, Y. W. J. Alloys Compd. 2008, 461, L20–L22. (11) Filinchuk, Y.; R€onnebro, E. Acta Mater. 2009, 57, 732–738. (12) Riktor, M. D.; Sørby, M. H.; Chzopek, K.; Fichtner, M.; Buchter, F.; Z€uttel, A.; Hauback, B. C. J. Mater. Chem. 2007, 17, 4939–4942. (13) Kim, J.-H.; Jin, S.-A.; Shim, J.-H.; Cho, Y. W. Scr. Mater. 2008, 58, 481–483. (14) Kim, J.-H.; Shim, J.-H.; Cho, Y. W. J. Power Sources 2008, 181, 140–143. (15) Yavari, A. R.; LeMoulec, A.; de Castro, F. R.; Deledda, S.; Friedrichs, O.; Botta, W. J.; Vaughan, G.; Klassen, T.; Fernandez, A.; Kvick, Å. Scr. Mater. 2005, 52, 719–724. (16) Yin, L.-C.; Wang, P.; Kang, X.-D.; Sun, C.-H.; Cheng, H.-M. Phys. Chem. Chem. Phys. 2007, 9, 1499–1502. (17) Brinks, H. W.; Fossdal, A.; Hauback, B. C. J. Phys. Chem. C 2008, 112, 5658–5661. (18) Eigen, N.; B€osenberg, U.; Bellosta von Colbe, J.; Jensen, T. R.; Cerenius, Y.; Dornheim, M.; Klassen, T.; Bormann, R. J. Alloys Compd. 2009, 477, 76–80. (19) Lutterotti, L.; Matthies, S.; Wenk, H.-R.; Schultz, A. J.; Richardson, J. J. Appl. Phys. 1997, 81 (2), 594–600. (20) Suryanarayana, C. Prog. Mater. Sci. 2001, 46, 1. (21) B€ osenberg, U.; Doppiu, S.; Mosegaard, L.; Barkhordarian, G.; Eigen, N.; Borgschulte, A.; Jensen, T. R.; Cerenius, Y.; Gutfleisch, O.; Klassen, T.; Dornheim, M.; Bormann, R. Acta Mater. 2007, 55, 3951– 3958. (22) Deprez, E.; Mu~ noz-Marquez; Miguel, A.; Roldan; Manuel, A.; Prestipino, C.; Javier Palomares, F.; Bonatto Minella, C.; B€osenberg, U.; Dornheim, M.; Bormann, R.; Fernandez, A. J. Phys. Chem. C 2010, 114, 3309–3317. (23) Jin, S.-A.; Shim, J.-H.; Cho, Y. W.; Yi., K.-W. J. Power Sources 2007, 172, 859–862. (24) Jin, S.-A.; Shim, J.-H.; Ahn, J.-P.; Cho, Y. W.; Yi, K.-W. Acta Mater. 2007, 55, 5073–5079. 2503

dx.doi.org/10.1021/jp107781m |J. Phys. Chem. C 2011, 115, 2497–2504

The Journal of Physical Chemistry C

ARTICLE

(25) Takacs, L.; McHenry, J. S. J. Mater. Sci. 2006, 41, 5246–5249. (26) Brice, J.-F.; Courtois, A. J. Solid State Chem. 1978, 24, 381–387. (27) Hwang, S.-J.; Bowman, R. C., Jr.; Reiter, J. W.; Rijssenbeek, J.; Soloveichik, G. L.; Zhao, J.-C.; Kabbour, H.; Ahn, C. C. J. Phys. Chem. C 2008, 112, 3164–3169. (28) Wang, L.-L.; Graham, D. D.; Robertson, I. M.; Johnson, D. D. J. Phys. Chem. C 2009, 113, 20088–20096. (29) Kim, Y.; Reed, D.; Lee, Y.-S.; Lee, J. Y.; Shim, J.-H.; Book, D.; Cho, Y. W. J. Phys. Chem. C 2009, 113, 5865–5871.

2504

dx.doi.org/10.1021/jp107781m |J. Phys. Chem. C 2011, 115, 2497–2504