In-Situ X-ray Diffraction Study of γ-Mg(BH4)2 Decomposition - The

Jun 26, 2012 - Trans., JIM 2011, 52, 1443 ..... D. B. Ravnsbæk , E. A. Nickels , R. Černý , C. H. Olesen , W. I. F. David , P. P. Edwards , Y. Fili...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/JPCC

In-Situ X-ray Diffraction Study of γ-Mg(BH4)2 Decomposition Mark Paskevicius,*,† Mark P. Pitt,† Colin J. Webb,‡ Drew A. Sheppard,† Uffe Filsø,†,§ Evan MacA. Gray,‡ and Craig E. Buckley† †

Department of Imaging and Applied Physics, Fuels and Energy Technology Institute (FETI), Curtin University, GPO Box U1987, Perth 6845, WA, Australia ‡ Queensland Micro and Nanotechnology Centre, Griffith University, Brisbane 4111, Australia § Interdisciplinary Nanoscience Center (iNANO) and Department of Chemistry, University of Aarhus, DK-800 Aarhus, Denmark S Supporting Information *

ABSTRACT: We have studied the complex decomposition mechanism of cubic γ-Mg(BH4)2 (Ia3̅d, a = 15.7858(1) Å) by in-situ synchrotron X-ray diffraction, temperature-programmed desorption, visual observation of the melt, and Fourier transform infrared (FTIR) spectroscopy. The decomposition and release of hydrogen proceeds through eight distinct steps, including two polymorphic transitions before melting, with a new ε-Mg(BH4)2 phase at ca. 150 °C. After melting, strong changes in sample color from yellow to brown to gray are consistent with the unknown Mg−B−H phase(s) (that diffract with high d-spacing halos) in the sample changing from an average composition of MgB2H5.3 at 325 °C, to MgB2.9H3.2 at 350 °C, and to MgB4.0H3.7 by 450 °C. From 350 to 450 °C, the crystalline Mg proportion increases. No combination of previously assigned anionic BnHm species (including MgB12H12 and Mg(B3H8)2) can account for the average composition of the unknown proportion of the sample. This is supported by FTIR spectra showing an absence of terminal B−H resonances in the 2500 cm−1 region that are present for B12H12 and B3H8 anionic species. Our combined analysis strongly indicates the presence of as yet unidentified Mg−B−H phase(s) in postmelted decomposed Mg(BH4)2 samples.



been assigned as δ-Mg(BH4)2.7 Of the known experimental and modeled SiO2 phases, only the cubic γ-Mg(BH4)2 phase has an exact SiO2 structural analogue,8 with the SiO2 equivalent α- and β-Mg(BH4)2 forms remaining to be found/discovered. Of particular detriment to the practical use of metal− borohydrides is their intrinsically high thermal stability, typically resulting in hydrogen release beginning only very close to or after the sample has been melted. Similarly, attempts at rehydrogenation of thermally decomposed metal−borohydrides typically take place above the melting temperature (Tm), for example, 600 °C/350 bar of H2(g) for rehydrogenation of decomposed LiBH49 and >350 °C/800 bar for hydrogenation of MgB2.5,10,11 The comilling of catalytic additives (as either salts or nanoscopic metals or alloys) with metal−borohydrides results in low temperature (below Tm) release of hydrogen by virtue of direct chemical reaction and irreversible consumption of the metal−borohydride phase.12 Furthermore, even with catalytic additives, any remaining unreacted metal−borohydride must still be melted to release the remaining hydrogen, and any unreacted additive may then be consumed and transformed by the ionic melt, exemplified by the high reactivity of molten LiBH4.13 The identification of a catalytic substance for any of

INTRODUCTION To obtain hydrogen storage materials of suitably high gravimetric and volumetric density, research efforts have focused on the group I and II metal−borohydrides. This class of materials contain tetrahedral BH4− anions that are counterbalanced by metal cations,1 resulting in phases such as LiBH4 (18.5 wt % H) and Mg(BH4)2 (14.9 wt % H). While group I phases are typically nonreversible below their melting temperature, group II phases are predicted to be less thermodynamically stable.2 Among the borohydride family, Mg(BH4)2 possesses unique crystallographic complexity. Mg2+ is isoelectronic with Si4+, and BH4− is isoelectronic with O2−. In structural terms this implies that Mg(BH4)2 phases may assume structural isomorphs to SiO2 phases, with Mg/B atoms assuming an isostructural Si/O lattice. Indeed, the structural solutions to α-Mg(BH4)23 and β-Mg(BH4)24,5 display a corner sharing network of Mg2+[BH4−]4 tetrahedra, which results in porous open channels similar to a tetrahedral SiO4 corner sharing network. Studies of these porous networks resulted in the synthesis of a new cubic γ-Mg(BH4)2 phase,6 which was shown under compression to form an almost ideal packing network of Mg2+[BH4−]4 tetrahedra, resulting in another new tetragonal polymorph denoted δ-Mg(BH4)2. Another hightemperature form of Mg(BH4)2 has also been identified by annealing TiF4:Mg(BH4)2 composites at 250 °C and 5 bar of H2, resulting in an orthorhombic Mg(BH4)2 phase that has also © 2012 American Chemical Society

Received: March 27, 2012 Revised: June 25, 2012 Published: June 26, 2012 15231

dx.doi.org/10.1021/jp302898k | J. Phys. Chem. C 2012, 116, 15231−15240

The Journal of Physical Chemistry C

Article

Figure 1. In-situ synchrotron X-ray diffraction data (λ = 1.0332 Å) illustrating the transition of γ-Mg(BH4)2 into two different polymorphs before melting.

the metal−borohydrides that enables hydrogen release below Tm and does not irreversibly consume the metal−borohydride currently does not exist. As such, hydrogen release and metal− borohydride decomposition reactions have typically been studied at temperatures well beyond Tm. After melting, the metal−borohydride begins to release hydrogen and fuses into a macroscopic lump, which typically contains as end products the group I and II metal−hydrides, and a B containing phase that is very difficult to structurally identify. A large difference in the final decomposition products is discernible dependent upon hydrogen backpressure, and spectroscopic measurements suggest the formation of closed-cage metal−BnHm type phases, which to date have not been corroborated by diffraction measurements.14 For the Mg(BH4)2 system, the exact nature of decomposition and hydrogen release is complex. It remains unclear whether Mg(BH4)2 melts, with many studies in conflict, from the earliest studies in the 1970s15 claiming (a) “melting with decomposition occurs at 305 °C when a sample is heated rapidly in a capillary, while melting fails to occur even at 320 °C when the substance is heated slowly”, and (b) “in vacuo (∼10−3 mm of Hg) a sample sublimes noticeably above 290 °C and decomposes at 350 °C”. The latest study16 states that “visual observation of the sample proved that there is no melting”. What is apparent is that after the endothermic event at ca. 288−301 °C,17−20 there is a “featureless” period in diffraction data from ca. 290−360 °C, after which MgH2 formation occurs at ca. 330−360 °C.16,19,20 The exact nature of phases in this “featureless” region is currently unclear. Similarly, the Bcontaining phases that are present from when MgH2 formation occurs at >330 °C up until MgB2 formation at ca. 580 °C are also not well characterized by diffraction methods. No crystalline B-containing phases are easily observed from ca. 290−580 °C. The occurrence of MgO during some decomposition studies16 suggests that Mg(BH4)2 samples may also initially contain oxygen or water, which can be present in the form of borates,5 producing broad halos in diffraction data. The claim of observing “amorphous MgH2”16

in the featureless period appears speculative as no such phase has ever been observed previously for MgH2. In this study we follow the decomposition of γ-Mg(BH4)2 up to 600 °C by in-situ synchrotron X-ray diffraction. We focus on the particularly challenging >290 °C temperature range in order to identify what features appear in diffraction data that can be attributed to B-containing phases, including broad halos that may be from amorphous or nanocrystalline/nanoparticulate phases. We also visually inspect the γ-Mg(BH4)2 melting process in wide-bottom glass bottles, which makes inspecting any molten state significantly easier to observe compared to samples in capillaries.



EXPERIMENTAL DETAILS γ-Mg(BH4)2 (Batch # 84096JM 97.8% purity) and LiBH4 (Batch # 21898PJ 97.4% purity) were purchased from SigmaAldrich and used without further modification. All sample handling was performed in an argon glovebox (270 °C at d = 5.1 Å is not from a solid amorphous Mg(BH4)2 phase. Rather, the d = 5.1 Å halo represents a short-lived ionic melt of Mg(BH4)2, which has been particularly difficult to visually observe and which has been contested since the earliest literature.15 It should be noted that the diffraction halo at d = 5.1 Å shifts to 4.9 Å upon cooling when the ionic melt solidifies into a stoichiometric amorphous Mg(BH4)2 phase. Figure 3A−G shows a series of photographs of the melting sequence of γ-Mg(BH4)2. Similar photographs of the melting sequence for LiBH 4 are shown in Figure 3H−N for comparison. LiBH4 and Mg(BH4)2 melt at a very similar temperature, on average, ca. 270−280 °C. It is to be noted that the reported melting temperature for LiBH4 can vary by ca. 20 °C and has been quoted in the range 268−286 °C.28−30 This variation is likely dependent on the proportion and composition of impurity phases (LiBH4 is typically 90−95% purity). A similar variation in melting temperature is reported in the earliest studies of Mg(BH4)2.15 For LiBH4, the molten state is clearly obvious, with a single phase transparent melt discernible by photograph (L). Hydrogen bubbles are also evident in (J, K). In contrast, the molten transition of γMg(BH4)2 is not immediately obvious, as there is no analogous transparent melt. However, the molten wavefront (from right to left) is clearly evident in (B, C), and the original white γMg(BH4)2 crystals can be observed to be subsumed. In (E), a large hydrogen bubble can be observed pushing the sample up, and the sample remains homogenously connected and in a fused state, indicating a molten transition. If the original white γ-Mg(BH4)2 crystals had not been subsumed and homogenized into a fused state, and if they had remained as individual powder grains and experienced only a solid−solid transition, then any hydrogen bubbles would merely have displaced these individual powder grains. This is the strongest visual evidence to date that γ-Mg(BH4)2 indeed experiences a molten transition. The molten state of γ-Mg(BH4)2 is best described as “paste-like” and appears highly viscous. By in-situ X-ray diffraction, the molten state of γ-Mg(BH4)2 presents with a broad halo at ca. d = 5.1 Å (see Figure 6A), which remains at a similar d-spacing after cooling and solidifying into amorphous Mg(BH4)2 (see ex-situ cooled samples in Figure 6B). After the completion of the molten transition, both samples were allowed to cool under their own evolved hydrogen pressure. For LiBH4, the cooled sample has fused into a continuous whitish lump in (N). The cooled γ-Mg(BH4)2 sample shows a fused lump which is yellowish-brown in color in (G). This is consistent with the earliest studies which proceeded through the melting point, where it is noted that “the residue represents a brownish mass”.15 Ex-situ γ-Mg(BH4)2 samples quenched from just above the melting point show that the color change is more gradual than evident in Figure 3. Such samples show a white to yellow to brown transition. Further heating of the brown fused lump shows a color change to gray. The nature of the γMg(BH4)2 sample after melting is discussed further in section C. We note that the inspection of the melting process for Mg(BH4)2 is particularly difficult when viewing samples in

Figure 2. Two-dimensional in-situ XRD data of β-Mg(BH4)2 during heating. Note the diffuse scattering (A) in the range of (331) and (531) reflections that resolves into sharp diffraction lines at higher temperatures (B). An integrated high-temperature diffraction pattern is also shown (C).

SiO2 suggests such interpenetrating incommensurate structures are also plausible. Further clouding the exact structural nature of the β-Mg(BH4)2 phase is the original unit cell indexing.4 While an orthorhombic symmetry was previously proposed,4 no higher doubled hexagonal metric was searched for, similar to the case for α-Mg(BH4)2,3 where an orthorhombic solution (the orthorhombic α-Mg(BH4)2 unit cell is similarly dimensioned to the body-centered orthorhombic cells discussed above, but doubled on the a-axis) was initially found for the unit cell, and a final higher metric doubled hexagonal solution was found. Subtle residual positional misfit in our neutron data5 suggests that the orthorhombic Fddd unit cell is not the final solution. Single crystal studies of the β-Mg(BH4)2 phase are clearly necessary to resolve the exact nature of this diffuse scattering, and our previous revision of the average βMg(BH4)2 structure5 suggests that both neutron and X-ray data are necessary to obtain an accurate crystal structure solution that is not a false minimum with respect to H positions. B. Mg(BH4)2 Melting Process. Above 270 °C the diffraction peaks from β-Mg(BH4)2 strongly fade and a large diffraction halo presents at ca. d = 5.1 Å (11.6° 2θ in Figures 4 and 6A). This halo appears at a similar d-spacing as the amorphous Mg(BH4)2 phase (4.75−4.89 Å) that appears after milling α-Mg(BH4)2 or during cooling of annealed Mg(BH4)2 samples and has been observed in numerous studies.17,19,20,25−27 In earlier studies, crystallization of amorphous 15234

dx.doi.org/10.1021/jp302898k | J. Phys. Chem. C 2012, 116, 15231−15240

The Journal of Physical Chemistry C

Article

Figure 3. γ-Mg(BH4)2 heated from room temperature (A), illustrating a transition through a frothy melt (B−E) to a fused solid crust at higher temperature (F) and when cooled (G). A heating sequence is also displayed for LiBH4 from room temperature (H) to the molten state (I−M) and when cooled (N). The glass bottle has a diameter of 16 mm. Orange and white reflections and coloring are from the heating element of the hot air gun.

Figure 4. In-situ XRD data (λ = 1.0332 Å) of the decomposition of γ-Mg(BH4)2 into MgB2 during heating to 600 °C.

small diameter capillaries. Figure S1 shows a sequence of photographs of a γ-Mg(BH4)2 sample being heated in a 0.6 mm

i.d. sapphire capillary under a 1 bar H2 backpressure. Even under this backpressure, there is enough change in pressure 15235

dx.doi.org/10.1021/jp302898k | J. Phys. Chem. C 2012, 116, 15231−15240

The Journal of Physical Chemistry C

Article

are evident from ca. 270−360 °C. The first peak from 270 to 325 °C is consistent with the release of H from the ionic Mg(BH4)2 melt. Gas bubbles can be visually observed during the melting process (as in Figure 3). The second peak from 325 to 360 °C is consistent with the release of H from MgH2, but it is not clear whether all of the H release in the second peak can be attributed to MgH2. Figure 6A shows a zoomed region of our in-situ diffraction data from ca. 9°−16° 2θ, covering the temperature range 275−515 °C. The intense primary halo from the ionic Mg(BH4)2 melt can easily be observed with maxima at ca. d = 5.1 Å (ca. 11.6° 2θ in Figure 6A) and is clearly evident from ca. 275−320 °C. This halo progressively loses intensity, commensurate with the H loss from 270 to 325 °C corresponding to the first peak in the TPD data in Figure 5. From this, it is expected that by 325 °C the sample has completely solidified and fused with no ionic melt remaining. By 325 °C, complete solid phase segregation has occurred into MgH2 and a phase(s) containing all of the boron. Inspection of Figure 6A shows that by 325 °C an intense halo has appeared at ca. d = 6.3 Å (ca. 9° 2θ). This halo has not been observed previously in diffraction data, either because quartz capillaries have been used (which produce their own primary halo at ca. 4.2 Å, rendering the identification of high d-spacing halos very difficult) or because diffraction data have not been collected to a low enough angle (this is the most typical case). A subtle decrease/shift of the intensity of the d = 6.3 Å halo is evident from ca. 355−380 °C, commensurate with a third H release peak in the TPD data from ca. 360−390 °C. It should be noted that complete release of H from MgH2 has occurred by 360 °C, and the only feature in our diffraction data which can be correlated with the third TPD peak from ca. 360−390 °C is the halo at ca. d = 6.3 Å. This suggests that the phase producing the halo at d = 6.3 Å is composed predominantly of B and also initially contains H. A small increase in the TPD background in Figure 5 occurs by ca. 520 °C, indicating a fourth H release event before the formation of MgB2, consistent with earlier studies.16 During the existence of the ionic Mg(BH4)2 melt from ca. 275−320 °C, a large amount of hydrogen is released, ca. 1/3 of all H in the sample, and no crystalline reflections are evident. This release of H will destroy the charge balance of the ionic melt as BH4− units are depleted. It is not clear which Mg−B−H phase(s) forms from the charge imbalanced melt before 320 °C; however, it is likely the phase(s) is solid in nature, as the ionic melt halo clearly disappears. Figure 6B shows low angle diffraction data for ex-situ quenched samples from 320, 350, and 450 °C, just prior to MgH2 crystallization at 325 °C, and well past melting at 350 °C. At 320 and 350 °C, the ex-situ diffraction patterns show a very broad halo with a maxima of ca. d = 6.3 Å. The ex-situ 320 °C diffraction pattern in Figure 6 also shows an intense halo at ca. 4.9 Å from amorphous Mg(BH4)2 that has solidified during cooling due to incomplete decomposition of molten state BH4− units. (The sample has likely only reached ca. 310 °C due to the temperature difference between the thermocouple and the sample.) By 450 °C, the halo maxima has shifted to ca. d = 6.8 Å. The postmelted samples have clearly decrepitated, powdered, and are solid. The 6.3 Å halo appears at 320 °C just prior to the complete decomposition of the melt by 325 °C, suggesting a brief multiphase liquid/solid transition. Therefore, the solid Mg−B− H phase(s) that forms directly from the ionic melt is responsible for the halo at ca. d = 6.3 Å. This halo appears weak in intensity principally due to the use of 300 nm Al foil,

with evolved hydrogen to considerably displace the sample (sample is denoted by a red bar in the figure) in the capillary, by ca. 5 mm, effectively removing it from the heat zone and also potentially removing it from an X-ray beam. Such sample displacement renders identification of melting impossible and also highly problematic for studying the process by in-situ diffraction. We note that visual inspection of our γ-Mg(BH4)2 sample during in-situ decomposition measurements on the beamline revealed that the sample displacement was minimal, and sample remained in the beam while it melted, indicating that the sample displacement can be variable (likely dependent upon sample packing in the capillary). We note that the statement about the melting process, “visual observation of the sample proved that there is no melting”,16 is not consistent with our observations here, likely due to the use of capillaries. C. Mg(BH4)2 Decomposition Pathway above the Melting Point. The exact decomposition mechanism of Mg(BH4)2 after melting has proven very difficult to structurally decipher, with a “featureless” diffraction pattern reported from ca. 290−360 °C, after which MgH2 formation occurs at ca. 330−360 °C16,19,20 (these decomposition studies typically start with α-Mg(BH4)2). Decomposition starting from β-Mg(BH4)2 has also been reported.31 Inspection of our in-situ data series from 280 to 600 °C in Figure 4 shows that γ-Mg(BH4)2 experiences a similar postmelting decomposition mechanism to α- and β-Mg(BH4)2, with a “featureless” diffraction pattern from ca. 275−325 °C. MgH2 reflections can be observed at 325 °C, with H release and formation of elemental Mg by ca. 340 °C. After the complete release of H from MgH2 by ca. 360 °C, the diffraction pattern remains similar until MgB2 formation commences at ca. 515 °C. The proportion of MgH2/Mg from the starting 1:2 Mg:B stoichiometry that has crystallized after melting has not been previously quantified. From 275 to 515 °C, there are no obvious B-containing crystalline phases evident in our diffraction data. As such, the nature of the B-containing phase that occurs after Mg(BH4)2 melting is of particular interest. TPD data for γ-Mg(BH4)2 decomposition under a 1 bar H2 backpressure are shown in Figure 5. Two strong peaks

Figure 5. Hydrogen release from γ-Mg(BH4)2 during temperatureprogrammed desorption (TPD) at 4 °C/min under 1 bar of H2. 15236

dx.doi.org/10.1021/jp302898k | J. Phys. Chem. C 2012, 116, 15231−15240

The Journal of Physical Chemistry C

Article

Figure 6. (A) In-situ XRD data (λ = 1.0332 Å) highlighting changes in the high d-spacing background during and after Mg(BH4)2 melting. (B) Exsitu XRD data (λ = 1.54 Å) from quenched postmelted samples.

internal LaB6 intensity/weight standard22 shows the average composition of the unknown proportion of the sample has changed from ca. MgB2.9H3.2 (at 350 °C) to MgB4.0H3.7 (at 450 °C). By 450 °C, there has also been a moderate release of Mg from the unknown proportion of the sample. The total sample composition at 450 °C is represented as MgB2H1.8 → 0.5Mg + 0.5MgB4.0H3.7. This indicates that Mg is steadily released and the unknown proportion of the sample is continuously changing from ca. 325−515 °C. From 325 to 350 °C, the average composition of the unknown proportion of the sample has changed from ca. MgB2H5.3 to MgB2.9H3.2. We note that the only change in the in-situ diffraction patterns from 325 to 350 °C is the occurrence of MgH2, and the only other discernible feature is the broad halo at ca. d = 6.3 Å (the halo centered at 4.9 Å is gone by 325 °C). The formation of the broad ca. d = 6.3 Å diffraction halo as a byproduct of the charge imbalanced ionic melt suggests it is an amorphous phase (as the melt itself is disordered). Assuming a single amorphous phase for the halo at d = 6.3 Å halo implies the phase has changed composition significantly from MgB2H5.3 (just below 325 °C) to MgB2.9H3.2 (at 350 °C) but still diffracts at the same d-spacing, which is not consistent. The absence of a d-spacing shift of the halo from 325 to 350 °C suggests the ca. d = 6.3 Å halo is representative of multiple Mg−B−H phases that may be nanocrystalline or nanoparticulate in nature, and a combination of Mg−B−H phases may be able to sustain small changes in composition associated with Mg loss over a small temperature range. Because of the complex nature of diffraction data, solid state11,16,31 and solution16,32 11B nuclear magnetic resonance (NMR) studies have been utilized above and below the melting temperature for Mg(BH4)2 to aid in the identification of Bcontaining species. Three different interpretations of the decomposition mechanism are proposed. In earlier work,16 an average composition of MgB2H5.5 is reported after the melting process, which can then transform through a multistep pathway including MgBH2.5, MgB12H12, and MgB4. On the basis of

which is significantly attenuating the low angle portion of the diffraction data. Prior to the release of H from the second TPD peak, the Mg−B−H phase(s) that is producing the ca. d = 6.3 Å halo possesses an average composition of ca. MgB2H5.3, similar to previous reports of ca. MgB2H5.5 at this temperature.16 From our in-situ synchrotron X-ray diffraction data, the release of H from the second TPD peak from 325−360 °C is consistent with the release of H from MgH2. Dependent on the H2 backpressure on the sample, the release of H from MgH2 will typically be complete by 360 °C. Figure 6B shows diffraction data from an ex-situ sample quenched from 350 °C under its own evolved backpressure (starting backpressure of 1 bar of H2). Measurement of sample mass before and after heating/quenching allows an exact measurement of H remaining in the sample. Combining this with an intensity/ weight fraction standard then allows an accurate calculation of the average Mg−B−H composition after any TPD peak. Utilizing a NIST 660a LaB6 standard (extremely well crystallographically characterized, with no amorphous phases), we have added a known weight fraction22 of 660a LaB6 to the ex-situ 350 °C quenched sample, in order to determine the average unknown Mg−B−H composition. We obtain an average composition of MgB2.9H3.2 for the fraction of the sample represented by the broad halo at ca. d = 6.3 Å. By 350 °C, the total sample composition does not exceed MgB2H2.9. The fraction of crystalline MgH2 in our ex-situ 350 °C quenched sample is relatively small, with the average MgB2H2.9 composition distributed as MgB 2 H 2.9 → 0.3MgH 2 + 0.7MgB2.9H3.2. The total percentage of H in MgH2 is only 7.5% of the original MgB2H8 starting composition, demonstrating that the majority of H release in the second TPD peak comes from the average MgB2H5.3 phase(s) that has completely formed from the ionic melt by ca. 325 °C. The release of H from the third TPD peak at ca. 360−390 °C is consistent with a strong color change of the sample from yellow to brown and a shift of the halo from ca. d = 6.3 to 6.8 Å (by 450 °C). Quantification of the unknown composition by utilizing an 15237

dx.doi.org/10.1021/jp302898k | J. Phys. Chem. C 2012, 116, 15231−15240

The Journal of Physical Chemistry C

Article

similar solid-state studies,31 a complex and gradual transition is proposed above the melting temperature, including the series of anions [BH4]− → [B2H6]2− → [B5H9]2− or [B5H8]− → [B12H12]2−. It is notable that the solid-state 11B NMR data from two separate studies closely match,16,31 yet two different mechanisms are proposed by fitting different BnHm anionic species to very similar data. A third study32 proposes the formation of a majority Mg(B3H8)2 species after decomposition at 200 °C under vacuum for 5 weeks. It is stated that solid-state 11 B NMR data were collected;32 however, this data is not presented. In all solid-state 11B NMR data from decomposed postmelted Mg(BH4)2 samples11,16,31 a ca. −15 ppm peak is consistently evident; however, at no stage is there a peak in the −31 ppm region to indicate the formation of a Mg(B3H8)2 species. Further, we note that Mg(B3H8)2 appears white33 and not yellow or brown. The occurrence of an assigned Mg(B3H8)2 phase at ca. −31 ppm in solution 11B NMR spectra16,32 but not in the solid state11,16,31 is further clouded by reported decomposition mechanisms in solution for example for NaBH4. In iodine-based solutions, NaBH4 can decompose to NaB3H8, which can subsequently decompose to Na2B12H12, yielding a multiphase solution 11B spectrum containing B12H12 and B3H8 anionic species.34 It is important to emphasize that decomposed Mg(BH4)2 samples bubble violently when placed in water, and the odor of diborane (B2H6) is evident, consistent with the earliest reports that “The residue after thermographing is a dark brown powder, which when reacted with water liberates a gas with the odor of boro-hydrides”.15 It has already been noted that in the solid state when B2H6 is reacted with LiBH4, Li2B12H12 and Li2B10H10 are produced,35 similar to the case for NaBH4 in solution.34 On this basis, it cannot be ruled out that reactions occur in solution between decomposed Mg− BnHm species and B2H6, producing phases that are not present in the original solid-state samples. For example, the claim of Mg(B3H8)2 being “highly soluble and stable in aqueous media”32 is in conflict with the study of pure Mg(B3H8)2,33 which clearly states “protic solvent such as ethanol and water react with Mg(B3H8)2, with vigorous evolution of gas”. This combined with the absence of a Mg(B3H8)2 chemical shift in solid-state 11B NMR spectra11,16,31 of decomposed Mg(BH4)2 indicates this species is absent in the solid state and an artifact of aqueous dissolution. It is to be emphasized that a range of anionic BnHm species can exist with similar chemical shift; for example, the room temperature spectra of anionic B7H7 species exhibit 11B spectra very similar to B12H12 anionic species.36 This is further supported by the >80% conversion to boric acid reported from postmelted Mg(BH4)2 samples in aqueous solution 11B NMR data,32 strongly indicating that the majority of the B is present in as yet unidentified phases after sample melting. Other assigned species such as MgB2H631 can also be ruled out on the basis that they diffract with strongest X-ray intensity37 at significantly lower d-spacing (ca. d = 2.76 Å) than 6.3 Å. Problematic with the assignment of the [B12H12]2− species is the fact that MgB12H12 has not been able to be synthesized in single phase anhydrous form,38 in contrast to the other group I and II [B12H12]2− species. As such, the crystal structure of MgB12H12 remains to be determined, and it is not known whether this phase can exist in amorphous or nanocrystalline/nanoparticulate form and at what d-spacing in diffraction data we can expect to observe the strongest diffracted intensity. There is a significant 8 ppm change of the NMR chemical shift assigned to [B12H12]2− from 320 to 370 °C,16 indicating a

change in structure/composition, or that the assigned [B12H12]2− species is transient in nature, which is consistent with the yellow to brown color change we observe in our samples. Testing of all possible combinations of the Mg−B−H phases reported by NMR16,31,32 indicates the closest multiphase combination produces an average MgB3H7 composition at 350 °C and an average MgB4H8 composition at 450 °C based on the Mg and B ratios in our samples. In both cases, this is twice the amount of H we determine. The fact that the assigned anionic BnHm species to date in the literature16,31,32 cannot reproduce our experimentally determined average composition for the unknown phase(s) at 350 and 450 °C indicates that the 11 B NMR chemical shift assignments are not correct for our system. However, it is possible that the numerous decomposition studies in the literature16,19,20 are forming different decomposition products on heating. Different atmosphere and pressure conditions during heating may lead to different phases being formed. Some studies do not clearly state these conditions, and if Mg(BH4)2 is heated under vacuum, Mg will likely vaporize before 400 °C,39 leave the sample, and change the composition and/or nature of the decomposed phases. Such a low backpressure may artifactually force the system through B-rich phases, such as BnHm compositions, which will display terminal B−H stretches in the 2500 cm−1 region, in contrast to our study here. In this respect, the solidstate 11B NMR chemical shift in the −15 ppm region is likely representative of multiple possible Mg−B−H compositions and strongly dependent on backpressure during decomposition. Our quantitative phase analysis cannot be explained by the typically reported Mg−B−H phases. This opens the possibility to as yet unidentified Mg−B−H phase(s) in the Mg(BH4)2 decomposition process. This is also strongly supported by FTIR spectra from the series of ex-situ samples shown in Figure 7. Surprisingly, FTIR data have not been previously reported

Figure 7. FTIR spectra of γ-Mg(BH4)2 and decomposed samples quenched from up to 560 °C. Spectra are scaled for clarity. 15238

dx.doi.org/10.1021/jp302898k | J. Phys. Chem. C 2012, 116, 15231−15240

The Journal of Physical Chemistry C

Article

molten transition of Mg(BH4)2 is clearly observed in a wide neck glass bottle. The “thick” viscous molten state of Mg(BH4)2 is best described as “paste-like” compared to the obvious transparent ionic melt of LiBH4. Postmelting, MgH2 formation is observed, with broad diffraction halos at >6.3 Å appearing that correspond to B containing phase(s). The broad diffraction halo at ca. d = 6.3 Å becomes stronger as the ionic melt halo at ca. d = 5.1 Å disappears. Average Mg−B−H compositions of the unknown proportion of the sample at 350 and 450 °C were determined quantitatively by XRD using an internal NIST 660a LaB6 intensity/weight standard. The unknown Mg−B−H phase(s) changes color, composition, and d-spacing between 350 and 450 °C, commensurate with gaseous H release and an increase in the proportion of crystalline Mg. No combination of the previously assigned anionic BnHm species can account for the average composition of the unknown proportion of the sample. This is further supported by FTIR spectra showing an absence of terminal B−H resonances in the region of 2500 cm−1 that would be present for B12H12 and B3H8 anionic species. Our analysis strongly indicates the presence of as yet unidentified Mg−B−H phase(s) in postmelted decomposed Mg(BH4)2 samples. We suggest that quantitative X-ray diffraction, solid-state 11B NMR, and FTIR should be used in a combinatorial fashion to resolve the ambiguous interpretation of B containing phases from any one technique by itself. It will remain challenging to unambiguously structurally characterize the as yet unidentified Mg−B−H phase(s) during Mg(BH4)2 decomposition above the melting temperature. The complex postmelting physical segregation into unknown Mg−B−H phase(s) and crystalline MgH2/Mg has previously shown strongly hindered hydrogen reabsorption;5 thus, melting of Mg(BH4)2 is clearly an undesirable process for H cycling/ reversibility. A catalytic phase to assist in B−H bond breaking below the melting temperature is necessary for Mg(BH4)2 to be considered useful for practical application and avoid postmelting complications.

for a sequence of decomposed Mg(BH4)2 samples, with the focus primarily on 11B and 1H NMR data. FTIR spectra of the starting γ-Mg(BH4)2 phase shows it is of very high purity, with no discernible B−O stretch in the 1420 cm−1 region.5 FTIR spectra for samples quenched from 320, 340, and 350 °C show that a strong resonance remains at ca. 2300−2400 cm−1 after the sample has melted, with no evidence of B−H stretches in the ca. 2500 cm−1 region, where large proportions of either metal−B12H1240 or Mg(B3H8)233 species show a terminal B−H resonance. A second strong resonance can be observed at ca. 1600 cm−1. The strong resonance at ca. 2300 cm−1 is reminiscent of BH4− vibrational modes; however, the average B:H ratio in the unknown proportion of the sample precludes such geometry/composition. A resonance at 2316 cm−1 has been interpreted for example as a Mg−H−B stretch in Mg(B3H8)233 (note that the Mg(B3H8)2 composition also produces a strong 2500 cm−1 doublet which is clearly absent in our data). The intense resonance at ca. 1600 cm−1 is reminiscent of a bridging stretch, observed previously in anionic BnHm containing species, with multiple interpretations, including metal−H41,42 and H−B−H43,44 models. Similar Mg− H bonding in the 1600 cm−1 region is seen in pure MgH245 but this band is only a minor component for the MgH2 spectra, with significantly stronger resonances observed elsewhere, and we do not expect the ca. 1600 cm−1 resonance could be attributed solely to MgH2. Similar strong resonance at ca. 1500 cm−1 can also be observed in 6-fold H-coordinated Be(BH4)242 and Al(BH4)346 (originally attributed to B−H2−metal bridges); however, as discussed above, BH4− configurations are very unlikely. A H−B−H type stretch occurs in the 1600 cm−1 region in solid B2H647 and has been explained as a puckering deformation of the B−H−B−H ring.44 The presence of stretches in both the 1600 and 2300 cm−1 regions suggests a B−H ring environment (akin to B2H6) likely along with a Mg− H−B bonding environment without typical terminal B−H bonding. As the average composition of the unknown proportion of the sample changes from ca. MgB2H5.3 to MgB2.9H3.2 from 320 to 350 °C the FTIR spectra remain very similar, but by 450 °C when the average composition of the unknown proportion of the sample is MgB4.0H3.7, the FTIR spectra shows a shift of the 2300 cm−1 resonance to a higher wavenumber with a maxima at 2400 cm−1. This shift in maxima implies a change in the B−H bonding environment, consistent with the change in composition (and yellow to brown color change) of the unknown proportion of the sample from MgB2.9H3.2 at 350 °C to MgB4.0H3.7 at 450 °C, implying at least two distinctly different Mg−B−H compounds exist at 350 and 450 °C. We note that the expected >2500 cm−1 terminal B−H stretch from the B12H12 and B3H8 anions have not appeared before MgB2 formation, indicating they do not constitute a majority intermediate in the decomposition of Mg(BH4)2 under 1 bar of H2.



ASSOCIATED CONTENT

S Supporting Information *

Images of powder displacement in capillaries during heating. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*E-mail [email protected]; Tel +61 8 9266 1381; Fax +61 8 9266 2377. Notes

The authors declare no competing financial interest.





ACKNOWLEDGMENTS This research was undertaken on the SAXS/WAXS beamline at the Australian Synchrotron, Victoria, Australia. The authors thank Nigel Kirby for his assistance with WAXS measurements. We acknowledge the facilities, scientific, and technical assistance with FTIR measurements of Peter Chapman, Department of Chemistry, Curtin University, Perth, Western Australia. C.E.B. acknowledges the financial support of the Australian Research Council for ARC LIEF Grants LE0775551 and LE0989180, which enabled the ex-situ XRD studies and temperature program desorption studies to be undertaken.

CONCLUSIONS The decomposition and release of hydrogen from γ-Mg(BH4)2 is a complex reaction, proceeding through two polymorphic transitions below the melting temperature, with a new lowtemperature ε-Mg(BH4)2 phase observed from ca. 150−205 °C. After 205 °C, single phase β-Mg(BH4)2 is observed, but with significantly weak (odd, odd, odd) reflections. Attempts to quench the ε-Mg(BH4)2 phase to ambient temperature were unsuccessful and resulted in transformation to hexagonal αMg(BH4)2. β-Mg(BH4)2 can be quenched in a stable manner. A 15239

dx.doi.org/10.1021/jp302898k | J. Phys. Chem. C 2012, 116, 15231−15240

The Journal of Physical Chemistry C



Article

(29) Arnberg, L. M.; Ravnsbaek, D. B.; Filinchuk, Y.; Vang, R. T.; Cerenius, Y.; Besenbacher, F.; Jorgensen, J.-E.; Jakobsen, H. J.; Jensen, T. R. Chem. Mater. 2009, 21, 5772. (30) Cahen, S.; Eymery, J.-B.; Janot, R.; Tarascon, J.-M. J. Power Sources 2009, 189, 902. (31) Yan, Y.; Li, H.-W.; Maekawa, H.; Aoki, M.; Noritake, T.; Matsumoto, M.; Miwa, K.; Towata, S.-i.; Orimo, S.-i. Mater. Trans., JIM 2011, 52, 1443. (32) Chong, M.; Karkamkar, A.; Autrey, T.; Orimo, S.-i.; Jalisatgi, S.; Jensen, C. M. Chem. Commun. 2011, 47, 1330. (33) Kim, D. Y.; Yang, Y.; Abelson, J. R.; Girolami, G. S. Inorg. Chem. 2007, 46, 9060. (34) Geis, V.; Guttsche, K.; Knapp, C.; Scherer, H.; Uzun, R. Dalton Trans. 2009, 2009, 2687. (35) Friedrichs, O.; Remhof, A.; Hwang, S.-J.; Zuttel, A. Chem. Mater. 2010, 22, 3265. (36) Schluter, F.; Bernhardt, E. Inorg. Chem. 2011, 50, 2580. (37) Zhang, Y.; Majzoub, E. H.; Ozolins, V.; Wolverton, C. Phys. Rev. B 2010, 82, 174107. (38) Chen, X.; Lingam, H. K.; Huang, Z.; Yisgedu, T.; Zhao, J.-C.; Shore, S. G. J. Phys. Chem. Lett. 2009, 1, 201. (39) Gilbreath, W. P. The Vapour Pressure of Magnesium between 223 and 385 °C; NASA TN D-2723, 1965. (40) Muetterties, E. L.; Merrifield, R. E.; Miller, H. C.; Knoth, W. H., Jr.; Downing, J. R. J. Am. Chem. Soc. 1962, 84, 2506. (41) Goedde, D. M.; Girolami, G. S. J. Am. Chem. Soc. 2004, 126, 12230. (42) Nibler, J. W.; Shriver, D. F.; Cook, T. H. J. Chem. Phys. 1971, 54, 5257. (43) Schulenberg, N.; Wadepohl, H.; Himmel, H.-J. Angew. Chem., Int. Ed. 2011, 50, 10444. (44) Sams, R. L.; Blake, T. A.; Sharpe, S. W.; Flaud, J.-M.; Lafferty, W. J. J. Mol. Spectrosc. 1998, 191, 331. (45) Sheppard, D. A.; Paskevicius, M.; Buckley, C. E. J. Phys. Chem. C 2011, 115, 8407. (46) Price, W. C. J. Chem. Phys. 1949, 17, 1044. (47) Stitt, F. J. Chem. Phys. 1941, 9, 780.

REFERENCES

(1) Rude, L. H.; Nielsen, T. K.; Ravnsbaek, D. B.; Bosenberg, U.; Ley, M. B.; Richter, B.; Arnbjerg, L. M.; Dornheim, M.; Filinchuk, Y.; Besenbacher, F.; Jensen, T. R. Phys. Status Solidi A 2011, 208, 1754. (2) Nakamori, Y.; Miwa, K.; Ninomiya, A.; Li, H.; Ohba, N.; Towata, S.-i.; Zuttel, A.; Orimo, S.-i. Phys. Rev. B 2006, 74, 045126. (3) Č erný, R.; Filinchuk, Y.; Hagemann, H.; Yvon, K. Angew. Chem., Int. Ed. 2007, 46, 5765. (4) Her, J.-H.; Stephens, P. W.; Gao, Y.; Soloveichik, G. L.; Rijssenbeek, J.; Andrus, M.; Zhao, J.-C. Acta Crystallogr., Sect. B 2007, 63, 561. (5) Pitt, M. P.; Webb, C. J.; Paskevicius, M.; Sheptyakov, D.; Buckley, C. E.; Gray, E. M. J. Phys. Chem. C 2011, 115, 22669. (6) Filinchuk, Y.; Richter, B.; Jensen, T. R.; Dmitriev, V.; Chernyshov, D.; Hagemann, H. Angew. Chem., Int. Ed. 2011, 50, 11162. (7) Amieiro-Fonseca, A.; Ellis, S. R.; Nuttal, C. J.; Hayden, B. E.; Guerin, S.; Purdy, G.; Soulie, J.-P.; Callear, S. K.; Culligan, S. D.; David, W. I. F.; Edwards, P. P.; Jones, M. O.; Johnson, S. R.; Pohl, A. H. Faraday Discuss. 2011, 151, 369. (8) Foster, M. D.; Freidrichs, O. D.; Bell, R. G.; Almeida Paz, F. A.; Klinowski, J. J. Am. Chem. Soc. 2004, 126, 9769. (9) Orimo, S.; Nakamori, Y.; Kitahara, G.; Miwa, K.; Ohba, N.; Towata, S.; Zuttel, A. J. Alloys Compd. 2005, 404−406, 427. (10) Severa, G.; Ronnebro, E.; Jensen, C. M. Chem. Commun. 2010, 46, 421. (11) Newhouse, R. J.; Stavila, V.; Hwang, S.-J.; Klebanoff, L. E.; Zhang, J. Z. J. Phys. Chem. C 2010, 114, 5224. (12) Fang, Z. Z.; Ma, L. P.; Kang, X. D.; Wang, P. J.; Wang, P.; Cheng, H. M. App. Phys. Lett. 2009, 94, 044104. (13) Mosegaard, L.; Moller, B.; Jorgensen, J.-E.; Filinchuk, Y.; Cerenius, Y.; Hanson, J. C.; Dimasi, E.; Besenbacher, F.; Jensen, T. R. J. Phys. Chem. C 2008, 112, 1299. (14) Hwang, S.-J.; Bowman, R. C.; Reiter, J. W.; Rijssenbeek; Soloveichik, G. L.; Zhao, J.-C.; Kabbour, H.; Ahn, C. C. J. Phys. Chem. C 2008, 112, 3164. (15) Konoplev, V. N.; Bakulina, V. M. Russ. Chem. Bull. 1971, 20, 136. (16) Soloveichik, G. L.; Gao, Y.; Rijssenbeek, J.; Andrus, M.; Kniajanski, S.; Bowman, R. C., Jr; Hwang, S.-J.; Zhao, J.-C. Int. J. Hydrogen Energy 2009, 34, 916. (17) Li, H. W.; Kikuchi, K.; Nakamori, Y.; Miwa, K.; Towata, S.; Orimo, S. Scr. Mater. 2007, 57, 679. (18) Chlopek, K.; Frommen, C.; Leon, A.; Zabara, O.; Fichtner, M. J. Mater. Chem. 2007, 17, 3496. (19) Hanada, N.; Chlopek, K.; Frommen, C.; Lohstroh, W.; Fichtner, M. J. Mater. Chem. 2008, 18, 2611. (20) Li, H. W.; Kikuchi, K.; Nakamori, Y.; Ohba, N.; Miwa, K.; Towata, S.; Orimo, S. Acta Mater. 2008, 56, 1342. (21) Hu, J.; Cai, W.; Li, C.; Gan, Y.; Chen, L. App. Phys. Lett. 2005, 86, 151915. (22) Bish, D. L.; Howard, S. A. J. Appl. Crystallogr. 1988, 21, 86. (23) Blanchard, D.; Maronsson, J. B.; Riktor, M. D.; Kheres, J.; Sveinbjörnsson, D.; Gil Bardají, E.; Léon, A.; Juranyi, F.; Wuttke, J.; Lefmann, K.; Hauback, B. C.; Fichtner, M.; Vegge, T. J. Phys. Chem. C 2011, 116, 2013. (24) Heaney, P. J.; Veblen, D. R. Am. Mineral. 1991, 76, 1018. (25) Li, H.-W.; Miwa, K.; Ohba, N.; Fujita, T.; Sato, T.; Yan, Y.; Towata, S.; Chen, M. W.; Orimo, S. Nanotechnology 2009, 20, 204013. (26) Pistidda, C.; Garroni, S.; Dolci, F.; Bardají, E. G.; Khandelwal, A.; Nolis, P.; Dornheim, M.; Gosalawit, R.; Jensen, T.; Cerenius, Y.; Suriñach, S.; Baró, M. D.; Lohstroh, W.; Fichtner, M. J. Alloys Compd. 2010, 508, 212. (27) Bardaji, E. G.; Hanada, N.; Zabara, O.; Fichtner, M. Int. J. Hydrogen Energy 2011, 36, 12313. (28) Zuttel, A.; Rentsch, S.; Fischer, P.; Wenger, P.; Sudan, P.; Mauron, P.; Emmenegger, C. J. Alloys Compd. 2003, 356−357, 515. 15240

dx.doi.org/10.1021/jp302898k | J. Phys. Chem. C 2012, 116, 15231−15240