Hydrogen dynamics in the low temperature phase of LiBH4 probed by quasielastic neutron scattering

Chemical Physics 427 (2013) 18–21

Contents lists available at ScienceDirect

Chemical Physics journal homepage: www.elsevier.com/locate/chemphys

Hydrogen dynamics in the low temperature phase of LiBH4 probed by quasielastic neutron scattering Arndt Remhof a,⇑, Andreas Züttel a, Timmy (A.J.) Ramirez-Cuesta b, Victoria García-Sakai b, Bernhard Frick c a

Empa, Swiss Federal Institute for Materials Science and Technology, Hydrogen and Energy, CH-8600 Dübendorf, Switzerland ISIS Facility, Rutherford Appleton Laboratory, Chilton, Didcot, Oxon OX11 0QX, United Kingdom c Institut Laue-Langevin, F-38002 Grenoble, France b

a r t i c l e

i n f o

Article history: Available online 17 August 2013 Keywords: Energy Materials Hydrogen Storage Lithium borohydride Quasielastic Neutron Scattering Dynamics Reorientations

a b s t r a c t LiBH4 contains 18.5 wt% hydrogen and undergoes a structural phase transition (orthorhombic ? hexagonal) at 381 K which is associated with a large increase in hydrogen and lithium solid-state mobility. We investigated the hydrogen dynamics in the low temperature phase of LiBH4 by quasielastic neutron scattering, including a new kind of inelastic fixed window scan (IFWS). In the temperature range from 175 to 380 K the H-dynamics is dominated by thermally activated rotational jumps of the [BH4] anion around the c3 axis with an activation energy of about 162 meV. In agreement with earlier NMR data, a second type of thermally activated motion with an activation energy of about 232 meV could be identified using the IFWS. The present study of hydrogen dynamics in LiBH4 illustrates the feasibility of using IFWS on neutron backscattering spectrometers as a probe of localised motion. Ó 2013 Elsevier B.V. All rights reserved.

1. Introduction Lithium borohydride (LiBH4) is a well explored model system for the investigation of localised hydrogen dynamics. LiBH4 forms ionic crystals consisting of negatively charged [BH4] ions and positively charged [Li]+ ions [1]. Within the [BH4] unit the hydrogen atoms surround the boron in a tetrahedral configuration. LiBH4 undergoes a solid–solid structural phase transition from an orthorhombic low-temperature structure (Pnma) to a hexagonal high-temperature structure (P63mc) at Tc = 381 K [2,3]. The hightemperature phase is associated with dynamical disorder, high [Li]+ translational mobility, rotational jumps of the [BH4] anions in the terahertz range, and strong lattice anharmonicities [4,5]. Due to its high hydrogen content of 18.5 wt%, LiBH4 is currently discussed as a light weight hydrogen storage material [6]. In addition, owing to its high lithium ion conductivity of r = 1 * 103 S/cm in the hexagonal phase at 393 K, LiBH4 is considered as a potential solid-state electrolyte for lithium-ion batteries and fuel cells [7]. In many cases, the rotational motion of the anion seems to be a prerequisite for the fast cation mobility and thereby for the superionic conductivity. The anion dynamics of LiBH4 have been investigated by quasielastic neutron scattering (QENS) [8,9] and by nuclear magnetic resonance spectroscopy (NMR) [10,11]. The QENS measurements so far focussed on the hexagonal high temperature (HT) phase of

⇑ Corresponding author. Tel.: +41 587654369; fax: +41 587654022. E-mail address: [email protected] (A. Remhof). 0301-0104/$ - see front matter Ó 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.chemphys.2013.07.017

LiBH4. In the low temperature (LT) phase of LiBH4, two types of rotational motion are observed by NMR. The two different kind of motion were attributed to two types of jump rotational motion with different jump rates. Skripov et al. measured activation energies of 182 and 251 meV, respectively. For both types of motions the jump rates reach values in the order of 1011 s1close to Tc [10]. Jimura et al. measures similar values of 183 and 238 meV [11]. They attribute the motion with the lower barrier to rotational jumps around the c3 axis. Previous QENS studies also support the c3 axis rotation [8]. Here we present QENS spectra and Fixed Window Scans on the LT phase of LiBH4. Different spectrometers with different energy resolutions and thereby with different time-windows have been used to follow the quasielastic signal from 175 K up to Tc. LiBH4 can be used as a model system to demonstrate the feasibility to investigate Inelastic Fixed Window Scans (IFWS) [12], which represent an alternative way to quickly scan the thermal behaviour of a sample in a backscattering spectrometer.

2. Experiment To avoid the strong neutron absorption by the 10B isotope in natural boron, 11B enriched (99.5%) Li11BH4 (chemical purity > 98%), purchased from Katchem, was used. The material was handled solely under inert gas conditions in purified Ar or He. Quasielastic neutron scattering (QENS) measurements were carried out using the neutron spectrometer IRIS at the ISIS facility of the Rutherford Appleton Laboratory in Didcot, UK [13] and the

19

A. Remhof et al. / Chemical Physics 427 (2013) 18–21

LðTÞ / ½1  AðQÞ

sðTÞ 1 þ x2off sðTÞ2

ð1Þ

where sðTÞ ¼ s0 expðEa =kTÞ is the residence time, Ea the activation energy, k the Boltzmann constant and s0 the high temperature limit of the residence time. The benefits of IFWS and their interpretation have recently been discussed by Frick et al. [12]. At IRIS a neutron wavelength of k = 6.66 Å was chosen. The sample was filled into an annular sample container with a wall thickness of 0.5 mm and a diameter of 24 mm. The spectra were recorded in a range of scattering vectors of 0.45 < Q < 1.85 Å1 and binned into 17 different Q-groups. Energy transfers of up to ±400 leV were recorded. A typical scan was recorded within 2 h. The data reduction was carried out by using the data analysis and visualisation environment ‘‘DAVE’’ [17]. At IN16 a neutron wavelength of k = 6.271 Å was chosen. The Doppler drive allows varying the incident energy by ±15 leV. The momentum transfer ranges from 0.2 to 1.9 Å1. The samples were measured in a flat sample holder (width  height  thickness = 30  40  1 mm3) placed at a sample angle of 45° and measuring times were 4–6 h per temperature. The resulting QENS spectra were also analysed using DAVE. For both instruments, the QENS spectra were analysed by using the general purpose curve fitting utility ‘‘PAN’’, which is included in the DAVE distribution. The QENS spectra were modelled by using

three components: First, a resolution limited elastic peak using a Gaussian line shape of full width at half maximum Cel and an integrated area Iel, and second, a quasielastic broadened component arising from the reorientational motion of the BH4 units using a Lorentzian line shape of width Cqe and integrated intensity Iqe. The third component is a linear background. In all fits, the peak centres of the elastic and the quasielastic peaks were constrained to be the same. The width of the elastic line was fixed to the width of the measured elastic line of a vanadium standard sample. 3. Results and discussion Fig. 1 displays the EFWS, summed over the whole Q-range. The scans recorded at IN16 are represented by red squares; the ones recorded at IRIS are represented by blue circles. At low temperatures, the intensity decreases with temperature due to an effective Debye Waller factor which includes angular libration of the hydrogen. The different initial slopes for the different instruments are determined by the different energy width of the elastic line, i.e. by the different energy resolution. The hydrogen dynamics match roughly the instrument’s time window between 170 and 240 K for IN16 and between 240 and 360 K for IRIS. At lower temperatures, the energy resolution exceeds the quasielastic broadening, so the scattered quasielastic intensity cannot be distinguished from the elastic line. At higher temperatures, the broadening of the quasielastic line is too large to be distinguishable from a flat background. The step at 380 K corresponds to the structural phase transition. Fig. 2 displays the IFWS (symbols) together with a model curve. We will first compare the experimental IFWS (Fig 2) with the EFWS measured at IN16 (Fig. 1). At low temperatures, the IFWS displays a constant background, while the EFWS drops with increasing temperature. Around 150 K, a redistribution of intensity away from the elastic line to the inelastic part of the spectrum occurs, the elastic line broadens and the intensity of the IFWS rises. Note that the deviation from the background is noticeable around 130 K in the IFWS, while the onset of the increased loss of elastic intensity is observed from 150 K onwards, showing the higher sensitivity of the IFWS as compared to the elastic scan. The IFWS reaches is maximum around 215 K. At this temperature the EFWS almost levelled out to its equilibrium value. The IFWS is slightly asymmetric, showing a shoulder on the descending flank. The IFWS reaches the background level again above room temperature. The background level at high temperatures is lower than at low temperatures as the IFWS is carried out close to the elastic line and is 1.0

IN16 IRIS

0.9 0.8 0.7

I(T)/I(0)

spectrometer IN16 at the Institut Max von Laue – Paul Langevin in Grenoble, France [14]. IRIS is a time of flight spectrometer in inverted geometry. Neutrons scattered by the sample are energy analysed by means of Bragg scattering from large-area PG (0 0 2) crystal analyser array. The incident neutron flux at the sample position is approximately 5.0  107 neutrons cm2 s1 (white beam at full ISIS intensity). IN16 is a focussing backscattering spectrometer [14]. First, the desired wavelength is selected by a graphite (0 0 2) deflector, preparing a beam which is directed by a graphite (0 0 2) chopper to a spherically shaped monochromator mounted on a Doppler drive. Highly monochromatic neutrons are selected by Si-(1 1 1) monochromator crystals in exact backscattering geometry. As the monochromator undergoes a periodic sinusoidal motion, the exact wavelength (energy) selected is effectively scanned by the Doppler effect. These monochromatic neutrons are focussed onto the sample and the scattered neutrons are analysed again by Si (1 1 1) analysers in perfect backscattering. Both type of spectrometers are connected to the cold source of their respective neutron sources. The spectrometers are complementary with respect to their energy resolution. In the settings used, IRIS has a resolution of 17 leV, while IN16 has a resolution of less than one leV. Together with the data measured earlier at the TOF spectrometer FOCUS [8,15], which was operated at a resolution of 60 leV, we could follow the quasielastic broadening in the low temperature phase of LiBH4 from 175 K up to the transition temperature, over a wide energy range. Two different kinds of scans were performed. First, so called ‘‘fixed window scans’’ (FWS) were taken to define the temperatures for subsequent inelastic spectra were recorded. In these measurements the incident and the final wave vector (and thereby the energy transfer DE) is fixed. A FWS in which DE ¼  hxoff ¼ 0 is called an elastic FWS (EFWS) or just an ‘‘elastic scan’’. In case of DE ¼ hxoff –0, this type of scan is called ‘‘inelastic FWS’’ (IFWS). IFWS are very sensitive to observe a broadening of the elastic line. As soon as the broadening of the elastic line reaches the pre-set value, the intensity rises above the background level. In the case of a quasielastic bradening caused by a thermally activated jump motion, Grapengeter at al. calculated the temperature dependent IFW intensity I(T) for a temperature independent elastic incoherent scattering factor A(Q) to be [16]

0.6 0.5 0.4 0.3 0.2 0

100

200

300

400

500

T (K) Fig. 1. Elastic fixed window scans recorded at IN16 (red, bold symbols) and at IRIS (blue, open symbols). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

A. Remhof et al. / Chemical Physics 427 (2013) 18–21

Fig. 2. Inelastic fixed window scan, on LiBH4 measured on IN16 at 2 leV energy offset (red dots, upper curve). The thin markers represent the elastic background measured in the same scan during the change of the velocity direction. As the wing of the elastic line is still seen at 2 leV it was subtracted by using this elastic signal, which results in the curve (hollow symbols) to which the model 1 is fitted below the phase transition (continuous line, yellow). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

therefore still influenced by the drop of the elastic intensity. The elastic signal can be measured with a somewhat relaxed resolution in the same scan within the time where the monochromator stands nearly still and changes the direction of motion [12]. This signal is subtracted from the IFWS, resulting in equal background levels both in in the high temperature and in the low temperature regime. In Fig. 2 the uncorrected data are represented by bold symbols (red), the corrected ones by open symbols (grey). Thin markers represent the elastic signal. The corrected data can be obtained by modelling the fraction of the temperature dependent decay of the elastic wing to the total background. The chosen correction procedure can give slight differences for the background level at high temperatures, which however does not influence our data interpretation. The phase transition at 383 K can be seen in the IFWS and in the EFWS as a small step. Subsequently, the corrected data were fitted to the Grapengeter equation (Eq. (1)) as described in [12]. In Fig.2, the fit is represented by the straight line. The fit was obtained with a single set of motional parameters, fixed to Ea = 182 meV and s0 = 1.9  1014 s, corresponding to the faster of the two motions determined by 1H NMR [10]. Similar agreement can be achieved by combining both sets of motional parameters determined by 1 H NMR from reference [10] In this case the best fit is achieved by a relative weight of 95% of the faster and 5% of the slower motion. A more satisfactory fit can be achieved by treating the activation energies, pre-factors and the Debye Waller factor (DWF) as free fitting parameters. The resulting activation energies turn out to be lower ((162 ± 2) meV and (232 ± 11) meV) with respect to the values obtained by NMR [10,11], the corresponding pre factors equal (43.3 ± 4) fs and (7.2 ± 3) fs, respectively. The resulting DWF is d < u2>/dT = 6  103 Å2/K. The raw data together with the fit is displayed in Fig. 3. Inelastic spectra were recorded out at 175, 200 and 230 K at IN16, and at 240, 280, 320 and 360 K at IRIS. Fig. 4 shows an inelastic scan together with its deconvolution in the elastic and the quasielastic component on the example of the spectrum recorded at IRIS at T = 320 K and at Q = 1.53 Å1. Fig. 5 displays the quasielastic broadening Cqe between 240 and 360 K. For a given temperature, the quasielastic broadening is independent from Q, indicative of a localised motion.

Fig. 3. Inelastic fixed window scan, on LiBH4 measured on IN16 at 2 leV energy offset (red dots, upper curve). The thin markers represent the elastic background measured in the same scan during the change of the velocity direction. As the wing of the elastic line is still seen at 2 leV it was subtracted by using this elastic signal, which results in the curve (hollow symbols). The fit (yellow continuous line) is achieved by treating the activation energies, the pre-factors and the Debye Waller factor as free parameters. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Data Fit Elastic Quasielastic Background

T = 320 K -1 Q = 1.53 Å

4

Intensity (arb. units)

20

3

2

1

0 -300

-200

-100

0

100

200

300

E(µeV) Fig. 4. QENS scan of LiBH4 recorded using the IRIS spectrometer at T = 320 K and at Q = 1.53 Å1.

For a localised motion, the total incoherent-scattering function can be written as the sum of the elastic and quasielastic contributions [18]

Sinc ðQ ; wÞ ¼ A0 ðQ ÞdðxÞ þ

X Ai ðQ ÞLi xÞ

ð2Þ

i

The quasielastic contributions can be expressed by Lorentzian functions Li where

Li ðxÞ ¼

1 1=si p ð1=si Þ2 þ x2

ð3Þ

each with its own time constant si = 2⁄/Cqe. The weight Ai(Q) of the individual Lorentzians is called the quasielastic incoherent structure factor. Analogously, the contribution of the elastically scattered neutrons to the signal is called the elastic incoherent structure factor A0(Q). The asymmetric shape of the IFWS and the NMR results [10,11] suggest the existence of two different kind of motion with individual motional parameters.

21

A. Remhof et al. / Chemical Physics 427 (2013) 18–21

0.30 333

0.25

360K 1E-9

0.15

residence time t120(s)

FWHM (meV)

0.20

320K

0.10 280K 0.05

240K

T (K) 250

200

FOCUS IRIS IN16 162 meV 232 meV

1E-10

1E-11

0.00 0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

0.003

-1 Q (Å )

Each of the two motions should contribute to the recorded neutron spectra with a Lorenzian shaped broadening of the elastic line as described by Eq. (3), weighted by the occurrence (i.e. the thermal occupation) of the respective rotational jumps. A clear separation is possible when the signals of the two motions enter the time window of the instrument at different temperatures. Then at low enough temperature the fast motion will be visible by a quasielastic broadening of the signal, while the slower motion cannot be distinguished from the elastic line. At higher temperatures, the signal of the fast motion will be flattened out and contribute to the background while only the slower motion gives rise to a quasielastic broadening. In an IFWS this would lead to two distinct maxima at different temperatures. In the present case, there is one dominating motion. The contribution of the second one leads to a slight asymmetry of the IFWS. The QENS spectra can be fitted with a single Lorentzian function, the second contribution cannot be differentiated. The conversion of the si to the residence time is model dependent. Within the LT phase, the localised motion was identified as 120° jumps of the [BH4] – units around the threefold c3 axes. The mean residence time in this case is given by

3 2

0.005

0.006

1/T (1/K)

Fig. 5. Width (FWHM) of the quasielastic components determined at IRIS.

s120 ¼ s

0.004

ð4Þ

Fig. 6 shows the Arrhenius plot of the residence times measured at FOCUS (black symbols, taken from ref. [8]), IRIS (blue symbols) and IN16 (red symbols). The error bars were estimated from the spread of the FWHM, measured at different Q, around their respective mean values. The straight lines represent the results from the fit to the IFWS. The residence times obtained from the individual QENS spectra follow roughly the Arrhenius line with an activation barrier of 162 meV, i.e. the faster of the two motions. The largest deviation is observed at 230 K, the highest temperature measured at IN 16. This point lies on the descending flank of the IFWS, where the influence of the jump motion above the 232 meV barrier contributes significantly to the signal. Consequently, this point lies in-between the two corresponding Arrhenius plots. As the residence times were calculated assuming rotational jumps of 120°, we identify the dominating, fast motion with rotational jumps around the c3 axis. Summarising, we measured the rotational dynamics of the [BH4] anion in the LT phase of LiBH4 by means of QENS. The analysis of the IFWS reveal the presence of two kind of rotational motion with activation energies of ((162 ± 2) meV and (232 ± 11) meV) and with corresponding pre factors of (43.3 ± 4) fs and (7.2 ± 3) fs, respectively. The data confirm the re-

Fig. 6. Residence times of BH4 units between two adjacent 120° jumps, deduced from the QENS measurements at FOCUS (black, [8]), IRIS (blue) and IN16 (red). Solid lines represent the results obtained from the IFWS fit. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

sults from earlier NMR measurements. The dominating motion could be identified as a three-fold jump rotation of the [BH4] anion around the c3 axis. The measurements demonstrate the feasibility to use IFWS as an alternative way to investigate localised dynamics by backscattering instruments. The IFWS offers a measure of the motional parameters that otherwise are only accessible from the Arrhenius plot and thereby from time-consuming inelastic scans. The strength of the IFWS lies in the detailed and relatively rapid monitoring of the temperature evolution of a quasielastic signal, whereas spectral features can be best investigated by quasielastic spectra [12]. Acknowledgements Financial support by a Grant from Switzerland through the Swiss Contribution to the enlarged European Union is gratefully acknowledged. References [1] F. Buchter, Z. Lodziana, A. Remhof, Ph. Mauron, O. Friedrichs, A. Borgschulte, A. Züttel, Y. Filinchuk, L. Platinus, Phys. Rev. B 83 (2011) 064107. ˇ erny´, K. Yvon, J. Alloys Compd. 346 (2002) 200. [2] J.Ph. Soulié, G. Renaudin, R. C [3] M.R. Hartman, J.J. Rush, T.C. Udovic, R.C. Bowman, S.J. Hwang, J. Solid State Chem. 180 (2007) 1298. [4] S.K. Callear, E.A. Nickels, M.O. Jones, M. Matsuo, S.I. Orimo, P.P. Edwards, W.I.F. David, J. Mater. Sci. 46 (2011) 566. [5] F. Buchter, Z. Łodziana, Ph. Mauron, A. Remhof, O. Friedrichs, A. Borgschulte, A. Züttel, D. Sheptyakov, T. Strässle, A.J. Ramirez-Cuesta, Phys. Rev. B 78 (2008) 094302. [6] H.W. Li, Y. Yan, S.I. Orimo, A. Züttel, C.M. Jensen, Energies 4 (2011) 185. [7] M. Matsuo, S.I. Orimo, Adv. Energy Mater. 1 (2011) 161. [8] A. Remhof, Z. Łodziana, P. Martelli, O. Friedrichs, A. Züttel, A.V. Skripov, J.P. Embs, T. Strässle, Phys. Rev. B 81 (2010) 214304. [9] N. Verdal, T.J. Udovic, J.J. Rush, J. Phys. Chem. C 116 (2012) 1614. [10] A.V. Skripov, A.V. Soloninin, Y. Filinchuk, D. Chernyshov, J. Phys. Chem. C 112 (2008) 18701. [11] K. Jimura, S. Hayashi, J. Phys. Chem. C 116 (116) (2012) 4883. [12] B. Frick, J. Combet, L. van Eijck, Nucl. Instr. Methods A 669 (2012) 7. [13] C.J. Carlile, M.A. Adams, Phys. B 182 (1992) 431. [14] B. Frick, A. Magerl, Y. Blanc, R. Rebesco, Physica B 234–236 (1997) 1177; B. Frick, M. Gonzalez, Physica B 301 (2001) 8. [15] J. Mesot, S. Janssen, L. Holitzner, R. Hempelmann, J. Neutron Res. 3 (1996) 293. [16] H.H. Grapengeter, B. Alefeld, R. Kosfeld, R. Colloid Polym. Sci. 265 (1987) 226. [17] R.T. Azuah, L.R. Kneller, Y. Qiu, P.W.L. Tregenna-Piggott, C.M. Brown, J.R.D. Copley, R.M. Dimeo, J. Res. Natl. Inst. Stand. Technol. 114 (2009) 341. [18] P.C.H. Mitchell, S.F. Parker, A.J. Ramirez-Cuesta, J. Tomkinson, Vibrational Spectroscopy with Neutrons with Applications in Chemistry, Biology, Material Science, and Catalysis, volume 3 of Series on Neutron Techniques and Applications, World Scientific, Singapore, 2005.