Assessing the reactivity of TiCl3 and TiF3 with hydrogen

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 8 ) 1 e1 3

Available online at www.sciencedirect.com

ScienceDirect journal homepage: www.elsevier.com/locate/he

Assessing the reactivity of TiCl3 and TiF3 with hydrogen S. Kang a, L.E. Klebanoff b,*, A.A. Baker a, D.F. Cowgill b, V. Stavila b, J.R.I. Lee a, M.H. Nielsen a, K.G. Ray a, Y.-S. Liu c, B.C. Wood a,** a

Lawrence Livermore National Laboratory, Livermore, CA 94551, USA Sandia National Laboratories, Livermore, CA 94551, USA c Lawrence Berkeley National Laboratory, Berkeley, CA 94720, USA b

article info

abstract

Article history:

TiCl3 and TiF3 additives are known to facilitate hydrogenation and dehydrogenation in a

Received 15 March 2018

variety of hydrogen storage materials, yet the associated mechanism remains under

Accepted 14 May 2018

debate. Here, experimental and computational studies are reported for the reactivity with

Available online xxx

hydrogen gas of bulk and ball-milled TiCl3 and TiF3 at the temperatures and pressures for which these additives are observed to accelerate reactions when added to hydrogen stor-

Keywords:

age materials. TiCl3, in either the a or d polymorphic forms and of varying crystallite size

Titanium fluoride

ranging from ~5 to 95 nm, shows no detectable reaction with prolonged exposure to

Titanium chloride

hydrogen gas at elevated pressures (~120 bar) and temperatures (up to 200
C). Similarly,

Hydrogen storage

TiF3 with varying crystallite size from ~4 to 25 nm exhibits no detectable reaction with

Additives

hydrogen gas. Post-exposure vibrational and electronic structure investigations using

Hydrogen reactivity

Fourier transform infrared spectroscopy and x-ray absorption spectroscopy confirm this analysis. Moreover, there is no significant promotion of H2 dissociation at either interior or exterior surfaces, as demonstrated by H2/D2 exchange studies on pure TiF3. The computed energy landscape confirms that dissociative adsorption of H2 on TiF3 surfaces is thermodynamically inhibited. However, Ti-based additives could potentially promote H2 dissociation at interfaces where structural and compositional varieties are expected, or else by way of subsequent chemical transformations. At interfaces, metallic states could be formed intrinsically or extrinsically, possibly enabling hydrogen-coupled electronic transfer by donating electrons. © 2018 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.

Introduction Metal hydrides are attractive hydrogen storage media due to their theoretical high gravimetric and volumetric hydrogen densities and possible reversibility [1,2]. Nevertheless, most

bulk hydrides suffer from unfavorable thermodynamics and slow kinetics. Reversible hydrogen storage materials have been based on Mg (e.g., MgH2), Al (e.g., NaAlH4, LiAlH4, AlH3), N (e.g., LiNH2/LiH), B (e.g., Mg(BH4)2, Ca(BH4)2) and C (C/Pt). The kinetics associated with the hydrogen storage reactions of

* Corresponding author. ** Corresponding author. E-mail addresses: [email protected] (L.E. Klebanoff), [email protected] (B.C. Wood). https://doi.org/10.1016/j.ijhydene.2018.05.128 0360-3199/© 2018 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved. Please cite this article in press as: Kang S, et al., Assessing the reactivity of TiCl3 and TiF3 with hydrogen, International Journal of Hydrogen Energy (2018), https://doi.org/10.1016/j.ijhydene.2018.05.128

2

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 8 ) 1 e1 3

desorption and re-hydrogenation in these materials are typically slow for reasons that are poorly understood. Empirically, it is found that TiF3 or TiCl3 mixed with the storage material at the ~2e5 mole % level acts to accelerate hydrogen storage reactions for Mg-based [3e9], Al-based [10e26], B-based [27e34], N-based [35,36] and C-based [37] materials. It is remarkable, given the difference in chemistries of Mg, Al, N, B and C, that this commonality of function would be observed for TiCl3 and TiF3, suggesting that there might be a common catalytic agent operative in these systems. The identity of the catalytic species that accelerates the hydrogen storage reactions remains unknown in almost all cases. Typically, the TiF3 or TiCl3 additives are ball-milled into the base hydrogen storage material, producing a 2e5 mole % mixture with particle sizes ~1 mm diameter or less. The additive/host composite mixture is then processed thermally, desorbing hydrogen at ~200
C or higher with subsequent rehydrogenation at elevated temperature and H2 pressure. This thermal processing introduces the possibility of activating the TiF3 or TiCl3 molecules themselves, or inducing chemical reactions between the additive and the host storage material. Prior work with these additives has identified the formation of MgF2 and TiH2 in the MgH2 system [4e6,8]; Ti2Al5, TiAl3, TiAl, alloys of the form Ti(1þx)Al(1-x), LiCl, NaCl, zero valent Ti metal and TiH2 in the Al-based materials [10,12,15,16,18,23e25]; LiF, TiB2 and TiH2 in the B-based materials [28,31]; LiCl and TiN in the Li-N-H system [35,36]; and catalytic F-C bonds in C-based materials [37]. Nonetheless, there is no direct evidence that any of these products have catalytic or promotion activity. On the contrary, when these products are intentionally introduced into their hydrogen storage hosts from the start, no significant improvement in reaction kinetics is observed. This has been found for tests of TiH2, TiN and MgF2 for MgH2 [6,8]; zero valent Ti metal, TiAl3 and TiH2 for Al-based materials [18,21,23,24]; and LiF and TiH2 in B-based materials [31]. In other cases, the catalytic inactivity of these products can be inferred because their production by thermal treatment and cycling coincides with a loss of kinetic and capacity performance, as seen in Al-based [38,39], B-based [31] and N-based materials [35]. Still one should remain open to the general possibility that the catalytic species is a reaction product between the additive and the host hydrogen storage material. Alternative mechanistic functions for the added TiCl3 or TiF3 have been proposed, including the generation of zero valent Ti “at the Al surface” [15], retention of “grain refinement” and increase in H vacancies (promoting H diffusion) caused by the presence of the additives [16], “point defect hexahydrides” with population promoted by the presence of Ti species [26], and catalytically active F-C bonds [37]. The anions F and Cl may also play a role in the kinetic promotion [7,40e42]. Finally, it remains unclear how TiF3 and TiCl3 can promote hydrogen storage in such a diverse range of materials, when the chemistry of the hosts varies so much. As part of an effort to understand the fundamentals of how these additives affect hydrogen storage reactions and how they can influence the chemistry of so many different materials systems, it is prudent to start from the simplest point possible, namely the reactivity of the additives themselves with hydrogen. Surprisingly, there have been no prior studies

of the reactivity of TiF3 and TiCl3 with hydrogen under the conditions of temperature and hydrogen pressure in which they promote hydrogen storage reactions as additives. In this paper, we investigate the reactivity of TiF3 and TiCl3 with hydrogen for commercial (as-received) materials, as well as for ball-milled TiF3 and TiCl3, in order to evaluate the influence of reduced particle size and potential material defects on possible reaction with H2. We also investigate the potential for the bulk and ball-milled TiF3 to dissociate molecular hydrogen using H2/D2 isotope exchange experiments, and structuredependent H2 dissociation and adsorption energetics using ab initio computation. The focus of these experimental and computational studies is to clarify how TiF3 or TiCl3 could affect the kinetics of hydrogen storage reactions after ball-milling with the storage material, but prior to any chemical reaction with the host material. In those systems where the combination of hydrogen and thermal processing produce chemical reaction between the additive and the host, these studies provide information on the unreacted portion that may remain after some chemical reaction has occurred. We also examine through computation how Ti-based additives could promote H2 dissociation at interfaces where structural and compositional varieties occur, or else by way of subsequent chemical transformation. In this work we make reference to “additives” rather than to “catalysts.” Whereas pure catalysts affect the kinetics of a chemical reaction without changing what products are generated or the equilibrium constant, we recognize that materials added to hydrogen storage reaction may instead react with the host material, thereby altering the reaction path and the product distribution. We refer to an additive that accelerates hydrogen storage reactions as a “catalytic agent” with this understanding in mind.

Experimental methods All sample handling and preparation were conducted in an Arfilled glove-box equipped with a recirculation system that keeps H2O and O2 concentrations below 0.1 ppm. Commercial grade TiF3 and TiCl3 were obtained from Sigma-Aldrich and Alfa Aesar, respectively. These materials were ball-milled to investigate the effect of reduced particle size and possible material defects and stresses on the reactivity towards hydrogen. Ball-milled material was produced by loading tungsten carbide (WC) mill pots with TiF3 or TiCl3 and milling with ½” diameter 440C stainless steel balls from SPEX Sample Prep, Inc. The ball/powder mass ratio was 18. In one case, 7/1600 diameter WC balls were used with the same ball/powder mass ratio to assess the possible role of milling contamination (principally Fe from stainless steel) on hydrogen reactivity. Milling was conducted under high-purity Ar for 2 h using a SPEX Sample Prep 8000M high-energy mixer mill. The TiF3 and TiCl3 samples (both as-received and ballmilled) were examined during hydrogen exposure with Sieverts uptake measurements to assess hydrogen reaction. Sample masses for the Sievert's measurements were in the 1.2e1.4 g range. Material characterization before and after hydrogen exposure was conducted with x-ray diffraction

Please cite this article in press as: Kang S, et al., Assessing the reactivity of TiCl3 and TiF3 with hydrogen, International Journal of Hydrogen Energy (2018), https://doi.org/10.1016/j.ijhydene.2018.05.128

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 8 ) 1 e1 3

(XRD) to assess changes in crystallite size and structure; with Fourier transform infrared (FTIR) spectroscopy to examine variations in chemistry via vibrational properties; and with element-specific x-ray absorption spectroscopy (XAS) to assess changes in the local electronic structure. We also used a H2/D2 exchange measurement apparatus to assess if TiF3 could promote hydrogen (or deuterium) dissociation. The Sieverts measurements employed a PCTPro 2000 apparatus from Setarem. For Sieverts hydrogen uptake studies, the TiF3 and TiCl3 samples were loaded inside a sample holder within the Ar glove box. A thermocouple was placed in the center of the sample holder for accurate temperature measurements during the hydrogen absorption and desorption experiments. Pressure changes during the hydrogenation of the samples were quantified with calibrated pressure transducers. Baseline measurements without any sample and with stainless steel spacers in the reactor were performed for normalization and background subtraction. Hydrogen capacity data are presented as weight percent with respect to the TiF3 or TiCl3 sample weight. XRD measurements were made using a Panalytical Empyrean diffractometer and analyzed with Reitveld refinement using the HighScore Plus software package. FTIR spectra were collected in attenuated total internal reflectance (ATR) mode using a compact Carey 630 spectrometer located within the Ar glovebox. The ATR crystal was diamond, and all FTIR spectra are reported as absorbance measurements. XAS measurements [43,44] at the Ti L2,3 edge and the F Kedge were performed at beamlines (BLs) 6.3.1.2 and 8.0.1.1 of the Advanced Light Source (ALS), Lawrence Berkeley National Laboratory (LBNL), and at the Resonant Elastic and Inelastic Xray Scattering (REIXS) beamline (BL10ID-2) of the Canadian Light Source (CLS). The photon energy was calibrated using XAS from a sample of Ti metal (aggressively abraded within a glove box to remove the surface oxide) and powdered TiO2. All experimental samples and standards were transferred to the experimental endstations of BLs 6.3.1.2 and 8.0.1.1 under inert atmosphere using a purpose-designed vacuum suitcase to limit exposure to atmospheric oxygen and water. Samples measured at the CLS were transferred using an Ar-purged glove bag with a partial pressure of <25 ppm water. XAS data were simultaneously recorded in two modes: total electron yield (TEY) [43] (via measurement of the drainage current to the experiment sample) and total fluorescence yield (TFY) [44] (using a channeltron electron multiplier). All XAS spectra presented in this paper were normalized to the incident x-ray flux, I0, which was measured concurrently as drainage current from an upstream gold mesh, and scaled to the magnitude of the absorption edge step. The capacity for as-received and ball-milled TiF3 to dissociate H2 was assessed using a Sandia-built “in-house” apparatus that provides exposure to calibrated mixtures of H2 and D2, at ~20 Torr (~0.027 bar) and for temperatures of 22
C and 200
C. The breaking of H-H or D-D bonds were monitored by measuring the partial pressure of HD as a function of time with a residual gas analyzer (RGA) which functions as a mass spectrometer. Fig. S1 in the Supporting Information (SI) shows a diagram of the H2/D2 isotope exchange experimental setup. The RGA was calibrated using pure H2 and D2 gases. Sample powders were loaded into holders in the Ar-filled glove box

3

and connected to the manifold. The holder containing the experimental sample was then evacuated slowly at room temperature, while monitoring the partial pressure of volatile gas species. The sample was first exposed to a pressure of D2, after which the valve to the sample holder was closed and the manifold was evacuated. Next, the manifold was filled with approximately the same quantity of H2 and the valve to the sample holder was opened for about 30 s to allow the gases to mix. This short mixing time was selected as a tradeoff between the longer time needed to homogenize the gas mixture and the desire to minimize H-D exchange on internal stainless steel surfaces of the manifold plumbing. After 30 s, the lower valve to the sample holder was closed and the manifold was re-evacuated. H-D exchange was monitored by cracking the lower valve to the sample holder, recording the relative partial pressures of H2, HD, and D2 with the RGA and then re-closing the valve. This procedure was used with all the TiF3 samples to minimize exposure of the manifold and RGA to evolving, contaminating F species. After accumulating data for about 3 h at room temperature (22
C), each sample was quickly (~5 min) heated to 200
C and RGA measurements were resumed. Experiments were conducted at 22
C and 200
C for up to 3.5 h. Measurements were compared to the extent of HD production in the empty sample holder, and also for HD production from a material known to induce H2 dissociation, namely powdered Fe. The low vapor pressure of TiF3 [45] allowed assessment at 200
C without unmanageable F contamination of the apparatus. TiCl3 was not investigated for H2/D2 exchange because its vapor pressure is too high at 200
C [46], which would lead to Cl contamination of the experimental system.

Theoretical methods In order to elucidate the reactivity of Ti-containing catalysts with H2 and its mechanisms, we performed a case study on pure TiF3 using first-principles calculations. The dissociative adsorption reaction energies of H2 on TiF3 surfaces were computed within Density Functional Theory (DFT) using the Perdew-Burke-Ernzerhof (PBE) functional in the generalized gradient approximation (GGA) [47] and projector-augmented wave (PAW) method [48]. Spin-polarized calculations were performed using the Vienna ab initio Simulation Package (VASP) [49] with a plane-wave energy cutoff ¼ 520 eV. An onsite interaction parameter of U  J ¼ 7.1 eV is applied on the Ti 3d orbital following Reference [50]. Before investigating interactions between H2 and TiF3, bulk and surface structures of pure TiF3 were obtained by optimizing the lattice parameters and atomic positions until per-atom energies and forces A, respectively. For converged to within 105 eV and 103 eV/
surface calculations, we used a slab method with slab thickness >20
A and vacuum thickness >15
A to ensure the convergence of surface energies. Lattice parameters and atomic positions at the center of the slab were fixed to those of bulk TiF3. In all calculations, the k-points were sampled to A3. have more than 104 k-points per
The H2 adsorption energy, DGads: ; is defined as the energy difference between H2 gas and adsorbed H2 molecule (H2,ads),

Please cite this article in press as: Kang S, et al., Assessing the reactivity of TiCl3 and TiF3 with hydrogen, International Journal of Hydrogen Energy (2018), https://doi.org/10.1016/j.ijhydene.2018.05.128

4

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 8 ) 1 e1 3

and the dissociative adsorption energy, DGdiss:ads: ; is defined as the energy difference between H2 gas and two H adatoms (2Hads). The temperature and pressure effects to the reaction free energies are accounted for by introducing experimental values from the NIST-JANAF table [51] and the ideal gas law. Details about the calculations of reaction free energies can be found in the SI. In addition to the reaction of H2 on the surface of TiF3, we calculated the driving forces to form secondary impurity phases from bulk TiF3 and TiCl3 reacting with H2 gas. The ground state impurity phases considered were obtained from the Ti-H-F(Cl) ternary phase diagrams in the Materials Project [52], and their energies at our experimental temperatures, 22
C and 200
C, were collected from the NIST-JANAF equations [51]. The H2 pressure of 120 bar was accounted for according to Equation S1.

Results and discussion In the sections that follow, we first provide a complete analysis of the possible reaction of H2 with TiCl3 and TiF3 based on FTIR and XAS, as well as thermodynamic calculations from DFT. Then, we investigate the possible catalytic promotion of H2 dissociation based on H2/D2 isotope exchange experiments on TiF3, combined with calculations of surface H2 dissociation from DFT.

Possible Ti reaction upon hydrogen exposure The discussion begins with the possible reactivity of TiCl3 and TiF3 towards hydrogen. The effects of prolonged exposure to hydrogen on both materials are considered in Fig. 1. Hydrogen exposures were performed at 120 bar pressure and 22
C, to assess the effects of exposure to H2 pressure alone, and at 120 bar and 200
C which are the typical conditions for which TiCl3 and TiF3 additives function in hydrogen storage materials. It is known from the literature that there are four polymorphs of TiCl3 (a, b, d and g) [53e57]. Fig. 1 shows

that even at these elevated pressures and temperatures, and for prolonged exposure times, a-TiCl3 (the as-received material), d-TiCl3 (produced by ball-milling) and TiF3 have negligible H2 uptake (note the exploded ordinate scale). The pressure changes in Fig. 1 correspond to only 0.1e0.2 wt % of hydrogen adsorbed to the bulk and ball-milled versions of these materials, which is close to the experimental detection limit. Figs. 2 and 3 compare and contrast FTIR data collected for a-TiCl3 (bulk), d-TiCl3 (ball-milled) and TiF3 materials, respectively, before and after the Sieverts exposures shown in Fig. 1. Fig. 2 demonstrates that hydrogen exposure at 120 bar pressure from 22 to 200
C does not modify the vibrational spectra for a-TiCl3 or d-TiCl3. This behavior is consistent with the Sieverts data of Fig. 1(a) collected during the prolonged hydrogen exposure and indicates a lack of chemical modification to either polymorph of TiCl3. The FTIR data in Fig. 3 reveal comparable behavior for TiF3, as the spectra for both the bulk and ball-milled TiF3 are largely unaffected by exposure to 120 bar H2 at 22
C and 200
C. The FTIR and Sieverts uptake data indicate that TiCl3 and TIF3 do not react with hydrogen at the conditions under which these additives accelerate kinetics of hydrogen storage materials. Further evidence for this finding is provided by XAS measurements, Fig. 4, which were performed to determine if hydrogen exposure of these materials leads to changes in the electronic structure and chemical bonding. Data were taken at the Ti L2-and L3-edges and at the F K-edge for materials exposed to hydrogen at 22
C and 200
C, and compared to unexposed samples as well as a number of experimental standards. XAS measurements were also performed at the Cl L1-, L2-, and L3-edges for the TiCl3 samples, but the signal intensity was too low for meaningful interpretation. Fig. 4(a) compares TFY XAS data at the Ti L2,3 edge for aTiCl3 (bulk) and d-TiCl3 (ball-milled) prior to exposure to hydrogen, and then exposed to hydrogen under the conditions shown in Fig. 1(a). Also shown are reference XAS TFY spectra for Ti metal and TiO2 standards. We estimate the depth

Fig. 1 e Sieverts hydrogen uptake data for (a) a-TiCl3 (bulk) and d-TiCl3 (ball-milled); and (b) TiF3. Hydrogen exposures were performed at 120 bar at room temperature (22
C) and at 200
C. All uptake curves were measured nominally at the pressures indicated, although the curves have been adjusted slightly vertically for clarity of presentation. Please cite this article in press as: Kang S, et al., Assessing the reactivity of TiCl3 and TiF3 with hydrogen, International Journal of Hydrogen Energy (2018), https://doi.org/10.1016/j.ijhydene.2018.05.128

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 8 ) 1 e1 3

5

Fig. 2 e FTIR data for (a) a-TiCl3 (bulk) and (b) d-TiCl3 (ball-milled) before and after the hydrogen exposures indicated in Fig. 1(a). The spectra have been shifted vertically to aid figure clarity; the spectrum for d-TiCl3 (before exposure) was also multiplied by 0.5 for clarity.

Fig. 3 e FTIR data for (a) bulk (as-received) and (b) ball-milled TiF3 before and after the hydrogen exposures indicated in Fig. 1(b). The spectra have been shifted vertically to aid figure clarity, but have not been scaled.

sensitivity of the Ti L2,3 edge TFY XAS data to be 220 nm and 150 nm for TiF3 and TiCl3, respectively. The XAS data indicates that the TiCl3 samples were not oxidized to TiO2 during sample handling at the synchrotron, and that there is no evidence of metallic Ti (a potential product of TiCl3 reduction by hydrogen). The Ti L2,3 edge XAS in Fig. 4(a) shows that ball milling TiCl3 produces only subtle changes in the Ti electronic structure, despite the change in polymorphism from a to d and the reduced crystallite size. These changes are consistent with formation of a small amount of surface oxide. In addition, the XAS data in Fig. 4(a) reveal that prolonged hydrogen exposure at 120 bar pressure from 22 to 200
C does not modify the local Ti electronic structure for either a-TiCl3 or d-TiCl3 as no significant changes in the XAS are observed under these conditions. This indicates a lack of reaction between TiCl3 and hydrogen under these conditions, which is consistent with the Sieverts uptake and FTIR measurements. Fig. 4(b) and (c) present TFY XAS data for bulk and ballmilled TiF3 before and after exposure to hydrogen. Similar to

TiCl3, the XAS data in Fig. 4(b) demonstrates that the bulk TiF3 samples were not oxidized to TiO2 nor formed metallic Ti before and after hydrogen exposure. Ball-milling of TiF3 does induce slight changes in the Ti L2,3 spectrum, particularly on the low photon energy side of both the L3-and L2-edges (over energies from ~456 to 458 eV and ~463e465eV), which could be indicative of the formation of Ti0. However, even this modified state shows no reactivity with hydrogen. Similarly, the results for the F K-edge in Fig. 4(c) indicate no change in the F electronic structure. Summarizing these XAS results for TiF3, exposure of TiF3 to 120 bar hydrogen between 22 and 200
C produces no change in the local Ti electronic structure (as revealed by Ti L2,3 edge XAS) or the local F electronic structure (as revealed by F K-edge XAS), to within experimental error. Thus, the XAS data indicate that the local bonding of the TiF3 does not change during exposure to hydrogen at elevated temperatures either before or after ball-milling. This is consistent with the Sieverts and FTIR measurements performed for TiF3.

Please cite this article in press as: Kang S, et al., Assessing the reactivity of TiCl3 and TiF3 with hydrogen, International Journal of Hydrogen Energy (2018), https://doi.org/10.1016/j.ijhydene.2018.05.128

6

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 8 ) 1 e1 3

Fig. 4 e (a) Total Fluorescence Yield (TFY) x-ray absorption spectroscopy (XAS) data at the Ti L2,3 edge for a-TiCl3 (bulk) and dTiCl3 (ball-milled) before and after the hydrogen exposures of Fig. 1(a). Also shown are reference XAS TFY spectra for the Ti metal and TiO2 standards. (b) TFY XAS data at the Ti L2,3 edge for bulk and ball-milled TiF3 before and after the hydrogen exposures of Fig. 1(b). Also shown in panel (b) are reference XAS TFY spectra for the Ti metal and TiO2 standards. (c) TFY XAS data at the F K-edge for bulk and ball-milled TiF3 before and after the hydrogen exposures of Fig. 1(b). For all figures, vertical dashed lines highlight key peak positions for comparison.

As further confirmation that TiCl3 and TiF3 do not undergo any chemical reactions upon exposure to H2, we computed the thermodynamics of the major possible secondary reaction products. Phase information obtained from the Materials Project [52], and the associated finite-temperature thermochemical data obtained from the NIST-JANAF tables [51] were combined with the pressure term of pure H2 gas (see Theoretical Methods for details). In particular, the formation of numerous impurity phases upon exposure to H2 gas were considered: HF(g), TiF2(g), TiF4(s), and TiH2(s) for TiF3; and HCl(g), TiCl2(s), TiCl4(s) and TiH2(s) for TiCl3. The reaction free energies for the formation of these side compounds in a balanced reaction with H2 were calculated at the experimental exposure conditions of 22
C and 200
C at 120 bar H2.

The calculated reaction free energies DG (per mole of reaction as written) are listed in Table 1. TiF3 and TiCl3 show analogous trends in reaction free energies. At the given experimental conditions, all of the reactions considered were found to be endergonic (DG > 0), indicating that there is no thermodynamic driving force to form secondary impurity phases. In general, the reaction free energies are smaller for TiCl3 than for TiF3. The least endergonic reaction between TiF3 and H2 is the formation of TiF4(s) and TiH2(s) with DG ¼ 79.5 kJ/mol at 22
C and 120 bar, while the least endergonic reaction between TiCl3 and H2 has DG ¼ 72.3 kJ/ mol at 200
C and 120 bar forming TiCl2(s) and HCl(g). Significantly higher temperatures would therefore be required to render these reactions thermodynamically

Table 1 e Calculated reaction free energies of TiF3 and TiCl3 with H2 to produce impurity phases. Free energies are reported per mole of reaction as written. Reaction free energy DG (kJ/mol) TiF3(s) þ 0.25 H2(g) / 0.75 TiF4(s) þ 0.25 TiH2(s) TiF3(s) þ 2.5 H2(g) / TiH2(s) þ 3 HF(g) TiF3(s) þ 1.5 H2(g) / 3 HF(g) þ Ti(s) TiF3(s) þ 0.5 H2(g) / 0.5 TiH2(s) þ 0.5 TiF2(g) þ F2(g) TiF3(s) þ 0.5 H2(g) / TiF2(g) þ HF(g) TiCl3(s) þ 0.25 H2(g) / 0.75 TiCl4(s) þ 0.25 TiH2(s) TiCl3(s) þ 2.5 H2(g) / TiH2(s) þ 3 HCl(g) TiCl3(s) þ 1.5 H2(g) / 3 HCl(g) þ Ti(s) TiCl3(s) þ 0.5 H2(g) / 0.5 TiH2(s) þ 0.5 TiCl2(s) þ Cl2(g) TiCl3(s) þ 0.5 H2(g) / TiCl2(s) þ HCl(g)

22
C, 120 bar

200
C, 120 bar

79.5 320.1 437.4 904.6 367.0 74.8 234.5 351.7 363.4 87.7

80.0 280.0 380.1 836.4 251.9 84.3 196.8 296.9 347.3 72.3

Please cite this article in press as: Kang S, et al., Assessing the reactivity of TiCl3 and TiF3 with hydrogen, International Journal of Hydrogen Energy (2018), https://doi.org/10.1016/j.ijhydene.2018.05.128

7

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 8 ) 1 e1 3

0.04 Fe (96 nm)

0.03

Q

accessible, as well as to overcome the additional kinetic energy barriers associated with the solid-state phase transformation to form impurity phases. We therefore confirm that TiCl3 and TiF3 should not react upon H2 exposure, and that any experimentally observed traces of secondary phase or volatile gas must arise due to the involvement of adventitious species such as water.

TiF3 Ball-milled (SS, 6 nm)

0.02 TiF3 Ball-milled (WC, 4 nm)

0.01 TiF3 ( 25 nm)

Possible kinetic enhancement of H2 dissociation

empty holder

0.00

Although there is no evidence for direct reactions involving TiCl3 or TiF3 upon hydrogen exposure, these materials could conceivably promote the dissociation of molecular H2 without forming any additional chemical species. Accordingly, this section focuses on TiF3 as a representative system for exploring the possible surface enhancement of H2 dissociation kinetics. Since dissociative adsorption of H2 most likely occurs at the Ti3þ site, it is reasonable to assume that all of the tested materials should exhibit qualitatively similar behavior as long as the local environments of the Ti atoms match. Indeed, in all tested polymorphs of TiCl3 and TiF3, the key structural motif involves Ti3þ cations that are octahedrally coordinated with halide anions. The possibility of enhanced dissociation of H2 on TiF3 is first assessed by measuring the production of HD molecules within a mixture of H2 and D2; i.e., through the reaction H2 þ D2 ¼ 2 HD. This reaction rate can be quantified by comparing the reaction quotient Q ¼ P2HD/(PH2  PD2) to the expected value at chemical equilibrium, where the Ps indicate partial pressures of the gas species indicated. Under chemical equilibrium at 22
C, Q is numerically equal to the equilibrium constant, and is ~4, decreasing slightly as temperature increases [58]. The rate of exchange on three samples of TiF3 powders was measured at 22
C and 200
C by applying ~10e15 Torr of H2 and D2 gas over the material of interest and measuring the pressures of H2, D2 and HD as a function of time, yielding the reaction quotient Q. The samples were as-received TiF3, TiF3 ball-milled for 2 h with stainless steel (SS) balls, and TiF3 ballmilled for 2 h with tungsten carbide (WC) balls. Isotope exchange rates on Fe powder and for an empty sample cell were also measured as references. The Fe sample was “as-received” commercial powder with no further treatment, which exhibited an increase in exchange rate as it was cycled, possibly due to the removal of surface oxidation. Multiple exposure cycles were not attempted on the TiF3 samples due to concerns about potential fluorine contamination of the apparatus. Small amounts of F-related contamination were observed at the ~2  107 Torr level, ~8 orders of magnitude below the pressures of H2, D2 and HD of interest, and were likely caused by reaction of TiF3 with trace water in the vacuum system. No measurable isotope exchange was observed for the three TiF3 samples during 3 h at 22
C. However, when samples were heated to 200
C, a very small isotope exchange occurred. Fig. 5 compares the exchange behavior for the TiF3 powders at 200
C, where the observed exchange reaction quotient Q is plotted versus time. The second-cycle results for the Fe powder at 200
C, plotted in Fig. 5, show that Fe is dramatically more potent than bulk TiF3 in dissociating H2, in spite of the smaller crystallite diameter of TiF3 (25 nm) compared to Fe (96 nm).

0

50

100

150

200

Time (minutes) Fig. 5 e HD exchange behavior for as-received and ballmilled TiF3 powder at 200
C compared with Fe powder (also at 200
C) and the empty holder. The y-axis reports the observed exchange reaction quotient Q defined in the text. The sample masses were as follows: Fe ¼ 1.017g, TiF3 (as-received) ¼ 0.567 g, TiF3 ball-milled with stainless steel milling balls ¼ 0.823 g, and TiF3 ball-milled with WC milling balls ¼ 0.821 g.

The TiF3 sample ball-milled with 440C stainless steel (SS) balls showed a larger capacity to dissociate hydrogen than bulk TiF3. However, given the high capacity for hydrogen dissociation on Fe, even trace Fe contamination (at the ~0.15% level) from the ball-milling process could produce the HD exchange rate seen for the TiF3 sample ball-milled with stainless steel balls. This possibility was investigated by milling TiF3 using 7/1600 diameter tungsten carbide (WC) milling balls, for the same milling time, ball/powder mass ratio and otherwise identical experimental conditions to the stainless steel ball-milling. XRD analysis of the WC-milled TiF3 gave a crystallite diameter of 4 nm, comparable with the 6 nm diameter observed using stainless steel balls. Fig. 5 shows that the HD exchange for this WC milled sample was very similar to that of bulk TiF3. The fact that accelerated HD production is not seen for the WC-milled TiF3 sample suggests that it is residual iron contamination, and not nanoscale TiF3, that is responsible for the enhanced isotope exchange seen for the stainless steel ball-milled sample. Based on the results in Fig. 5, we conclude that TiF3 negligibly promotes H2 bond dissociation, even for smaller particle sizes where significant fractions of surface sites should be accessible. Additional insights into the kinetics of H2 interaction were obtained using DFT total energy calculations. Although all of the titanium halides studied in this work have similar local chemical motifs, consisting of octahedrally coordinated TiF6 or TiCl6 units, the calculations afford the opportunity to also study the impact of the orientation and arrangement of these units. Indeed, the primary distinguishing characteristics of different polymorphs of TiF3 and TiCl3 involve the packing, arrangement, and/or orientation of the octahedral units. For instance, the low-symmetry rhombohedral R3c phase of TiF3 features TiF6 octahedra alternating between those rotated about the [111] direction and those rotated about the [1 1 1] direction, whereas the high-symmetry cubic Pm3m variant has TiF6 octahedra aligned along the a, b, and c axis directions [50,59] (see Fig. S4). Similarly, the a; b; g and d phases of TiCl3 feature different TiCl6 packing arrangements and alignments,

Please cite this article in press as: Kang S, et al., Assessing the reactivity of TiCl3 and TiF3 with hydrogen, International Journal of Hydrogen Energy (2018), https://doi.org/10.1016/j.ijhydene.2018.05.128

8

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 8 ) 1 e1 3

with d-TiCl3 exhibiting disordering of the structural layers varying from hexagonal packing of the Cl atoms to cubic packing [53,54]. DFT models were constructed based on the Pm3m structure of TiF3 and a distorted Pm3m structure featuring octahdra tilted towards either the [111] or [1 1 1] direction by 7.9
(see Fig. S4 and the SI for an illustration of both phases and additional details). This latter structure was introduced to explore the effect of octahedral orientation. The (100) surface of each structure was selected and checkerboard and striped surface configurations were constructed in which half of the surface Ti atoms are fully coordinated (6 F atom neighbors) and half are partially coordinated (5 F atom neighbors), as shown in Fig. 6. These configurations were chosen by considering the stoichiometry of the surface slab and the stability of local Ti environments (see the SI for details and calculated surface energies). Adsorption of H2 was calculated on the red “” sites in Fig. 6 according to Equation S4. The H2 molecular adsorption free energies (DGads.) at 0 K and 1 bar H2 as well as our experimental conditions, 22 and 200
C and 120 bar H2, are listed in Table 2. Other than H2 adsorption on the surfaces of the distorted Pm3m structure at 0 K and 1 bar H2, all adsorption energies are endergonic. Further, the calculated dissociative adsorption free energies were calculated based on the energy of the TiF3 surface with two H adatoms, 2Hads. This structure was constructed by locating one H atom on a red “” site in Fig. 6 and the other H atom atop a nearest-neighboring undercoordinated Ti atom. The distances between two H adatoms on the striped and checkerboard configurations are

3.80e3.96 Å and 5.37e5.66 Å, respectively, after optimization. The dissociative adsorption free energies DGdiss. ads calculated according to Equation S5 are summarized in Table 2. The high endergonic DGdiss. ads (>3 eV) indicates strongly unfavorable 2Hads adsorption on TiF3 surfaces both in the pristine Pm3m and distorted Pm3m phases, even though undercoordinated Ti sites are available. The high energy for hydrogen adsorption can be understood by considering that hydrogen adsorbs as a hydride ion (H), which requires local electron transfer from the surface. In stoichiometric TiF3, this electron is chiefly provided by Ti, resulting in oxidation of Ti3þ according to: 2Ti3þ þ H2(g) / 2Ti4þ þ 2H(ads)

(1)

This scenario was confirmed by examining the local orbital projection of the Ti charge density of the complexes in our DFT calculations, which indicated depletion of approximately one d electron from the Ti to which the hydrogen atom was bound. Because the oxidation of the Ti atom has a high energy cost in this structure, adsorption becomes unfavorable, implying no activity for H2 dissociation as found in the experiments. It is natural to consider how H2 dissociation could be activated beyond the pure TiCl3 and TiF3 phases considered up to this point. We can speculate that the energetics might be improved if the preferred þ3 charge state of Ti could be recovered. This would require additional electrons to be transferred to the system, which could occur along two possible thermodynamic pathways, as depicted in Fig. 7:

Fig. 6 e Side view (left) and top view (right) of symmetric TiF3 (100) surfaces considered in this study, shown here for the undistorted Pm3m phase. Surface Ti atoms with one dangling bond alternate with fully coordinated Ti atoms in a checkerboard structure in (a), and in a striped structure in (b). Blue and orange spheres are Ti and F atoms, respectively, with surface F atoms enlarged for clear visualization. H2 adsorption sites we investigated are marked with a red “£”. (For interpretation of the references to color/colour in this figure legend, the reader is referred to the Web version of this article.) Please cite this article in press as: Kang S, et al., Assessing the reactivity of TiCl3 and TiF3 with hydrogen, International Journal of Hydrogen Energy (2018), https://doi.org/10.1016/j.ijhydene.2018.05.128

9

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 8 ) 1 e1 3

Table 2 e Calculated reaction free energies for molecular adsorption of H2 (DGads.) and dissociative adsorption of 2H (DGdiss. ads) in eV, referenced to H2 gas, on TiF3(100) surfaces. Phase

Pm3m Pm3m Distorted Pm3m Distorted Pm3m

Surface structure

Checkerboard Striped Checkerboard Striped

At 22
C, 120 bar

At 0 K, 1 bar.

At 200
C, 120 bar

DGads.

DGdiss. ads

DGads.

DGdiss. ads

DGads.

DGdiss. ads

0.41 0.59 0.26 0.20

4.28 3.03 3.22 3.54

0.72 0.90 0.05 0.10

4.58 3.34 3.53 3.85

0.95 1.13 0.29 0.34

4.82 3.58 3.77 4.09

In pathway 1, we first have : 2Ti3þ þ H2(g) / 2Ti4þ þ 2H(ads)

(2)

2Ti4þ þ 2H(ads) þ 2e / 2Ti3þ þ 2H(ads) Pathway 2: 2Ti3þ þ 2e / 2Ti2þ

(3)

2Ti2þ þ H2(g) / 2Ti3þ þ 2H(ads) Pathways 1 and 2 differ according to whether electron transfer or hydrogen dissociation occurs first. In titanium halides, it is reasonable to assume that both pathways introduce a significant barrier due to the introduction of a charged defect (Ti4þ or Ti2þ) as an intermediate state. Alternatively, it is theoretically possible for electron transfer and hydrogen adsorption to occur simultaneously. Note that this can be considered a “hydrogen-coupled electron transfer” (HCET) reaction, in direct analogy to proton-coupled electron transfer reactions that are ubiquitous in electrocatalytic cycles: Pathway 3: 2Ti3þ þ H2 (g) þ 2e / 2Ti3þ þ 2H(ads)

(4)

If Pathway 3 could be activated, the barrier for H2 dissociation might be lowered with respect to Pathways 1 and 2dperhaps even within the range of catalytic promotion.

Fig. 7 e Schematic of hydrogen-coupled electron transfer (HCET). Horizontal and vertical directions correspond to hydrogen adsorption and electron transfer, and the diagonal line indicates the HCET reaction.

However, this would require an external source of electrons. Fig. 7 highlights this HCET pathway as a diagonal line. In Fig. 7, Ti2þ, Ti3þ and Ti4þ represent different oxidation states of Ti while still bound to an anion, such as for F and Cl considered in this work. The most obvious potential supply of additional electrons required for activating the HCET pathway would be from secondary metallic phases, or else from redox couples with other oxidizable atoms in the system. Indeed, if Ti3þ were to be added within another hydrogen storage material, both of these electron sources might be present in reaction during the hydrogenation/dehydrogenation cycle. For example metallic Mg is present in the hydrogenation/dehydrogenation cycle of MgH2. Similarly, metallic Al is be present in the cycling of NaAlH4. For pure TiF3 (or TiCl3), however, the possible electron sources are limited. The only remaining possibility for ready electron transfer is the introduction of metallic states within TiF3 itself. To explore the possibility of metallic states as electron donors for the HCET pathway in TiF3, we investigated the electronic structure of the pristine Pm3m and distorted Pm3m configurations, whose unit cells are shown in Fig. S4. Fig. 8(a) and 8(b) display the electronic density of states (DOS) of the pristine Pm3m and distorted Pm3m bulk materials. These figures demonstrate that the alignment of TiF6 octahedra significantly alters DOS of the bulk materials near to the Fermi level (marked with dotted lines in Fig. 8). As the TiF6 octahedra are symmetric in the bulk pristine Pm3m where all six Ti-F bond lengths are equal to 2.026
A, one Ti 3d electron occupies the triply degenerate t2g orbital near to the Fermi level, resulting in metallic nature of TiF3. On the other hand, in the distorted Pm3m structure in Fig. 8(b), the Jahn-Teller distortion induces the asymmetric geometry of TiF6 octahedra. Due to this distortion, the originally degenerated Ti t2g orbital is split into occupied singlet a1g and unoccupied doublet eg orbitals. The occupied a1g and unoccupied eg orbitals are separated by 4.64 eV, indicating that the distorted Pm3m structure in bulk is electronically insulating. Unlike the striking difference in the DOS properties of bulk phases, the DOS of surface configurations between the pristine Pm3m and distorted Pm3m become indistinguishable, and both resemble the DOS of the distorted Pm3m bulk. In Fig. 8(c) and 8(d), the DOS of surface atoms in the checkerboard surface configuration after 2Hads adsorption is displayed for the pristine and distorted Pm3m structures, respectively. The analogous DOS of surface atoms in the striped configuration is displayed in Fig. S5. Again, the Fermi levels of the surface structures are denoted by dotted lines in the figures. Both in the pristine and distorted Pm3m structures, when surface is created a gap opens between the occupied and unoccupied Ti 3d bands of approximately 2.7e3.2 eV, similar to the bulk

Please cite this article in press as: Kang S, et al., Assessing the reactivity of TiCl3 and TiF3 with hydrogen, International Journal of Hydrogen Energy (2018), https://doi.org/10.1016/j.ijhydene.2018.05.128

10

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 8 ) 1 e1 3

Fig. 8 e Electronic density of states (DOS) of bulk TiF3 per formula unit (a) in the Pm3m and (b) in the distorted Pm3m structures. Positive and negative DOS indicate spin-up and spin-down electrons. Black and blue solid lines represent Ti 3d and F 2p orbital-projected DOS, respectively. The corresponding DOS of checkerboard surface structures after 2Hads adsorption is displayed in (c) for the Pm3m and (d) distorted Pm3m structures. In (c) and (d), black, blue, and red lines represent site-projected DOS corresponding to surface Ti, surface F, and H atoms. The energy levels are with respect to the Ti 3p levels, and the vertical dotted lines indicate Fermi levels. (For interpretation of the references to color/colour in this figure legend, the reader is referred to the Web version of this article.)

distorted Pm3m. The DOS of clean surfaces without H adatoms show the analogous DOS properties to these H-adsorbed surfaces (data not shown here). Given that TiF6 octahdra undergo structural distortion on the surface due to the presence of dangling bonds, excess volume, and so on, this metal-toinsulator transition of surface TiF3 can be understood caused not by H adsorption but by the alignment, orientation, and/or distortion of TiF6 octahedra. Interestingly, the fact that the calculations find that metallic states can be introduced in the bulk material with specific symmetry and alignment of TiF6 octahdra suggests a strong coupling between the d levels of Ti and local structural deviations. These deviations are known to be thermally tunable in TiF3 [59], and could also be promoted at interfaces and solid phase boundaries due to lattice incommensurability and chemical heterogeneity. The theory results allow the conclusion that although TiF3 is intrinsically inactive towards H2 dissociation, there is reason to believe that it (or a similar Ti3þ-containing species) might promote dissociation and H adsorption via HCET if it were in contact with a material that could simultaneously supply electrons. This could be a residual metallic species formed during the reaction, for instance. The oxidation state of elements is conveniently detected with XAS, as

demonstrated in Fig. 4 where the Ti L2,3 XAS spectrum for Ti metal, TiO2 and TiF3 are presented and can be seen to be easily distinguishable from each other. Although this hypothesis requires further investigation, it could mean that a key role played by Ti-based additives in complex hydrides relates to the activation of an otherwise inaccessible HCET pathway at the junction with the active material. Similar ideas were proposed by Cui et al. [60] and Milosevic et al. [61]. In their studies, TiCl3 or VO2 were added to MgH2, and various phases involving those additives (Ti(0), Ti(2þ)H2, Ti(3þ)Cl3, and Ti(4þ)O2 [60]; and VH2 and VO2 [61] were observed. They attributed the enhanced hydrogen storage kinetics in MgH2 to the charge transfer aided by the multiple valence states of Ti or V. Our proposed HCET mechanism also suggests that Ti-containing additives could more actively involve chemical reactions with hydrogen.

Conclusions The reactivity of TiCl3 and TiF3 with hydrogen have been investigated, both in terms of reactivity at the temperatures and pressures of the measurements; and also in terms of the potential for these materials to dissociate molecular hydrogen

Please cite this article in press as: Kang S, et al., Assessing the reactivity of TiCl3 and TiF3 with hydrogen, International Journal of Hydrogen Energy (2018), https://doi.org/10.1016/j.ijhydene.2018.05.128

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 8 ) 1 e1 3

without directly reacting to form products. The theoretical results for reactivity found good agreement with the experimental observations. The study forms part of a larger effort to develop a validated theory of catalytic additives for hydrogen storage reactions. XRD, FTIR, XAS and H2/D2 exchange studies are reported for a-TiCl3, d-TiCl3 and TiF3 at the conditions of hydrogen pressure and temperature for which TiCl3 and TiF3 additives are observed to accelerate hydrogen storage reactions. TiCl3 in either the a or d polymorphic forms (with the d polymorph formed by ball-milling), or with varying crystallite size from ~5 to 95 nm shows no detectable reaction with hydrogen as examined from structural (XRD), vibrational (FTIR) or electronic structure (XAS) perspectives. Similarly, TiF3 with crystallites sizes ranging from ~5 to 25 nm shows no detectable reaction with hydrogen using the same experimental techniques. None of these forms of TiF3 act in a significant way to promote hydrogen dissociation, as demonstrated by H2/D2 exchange studies. These findings indicate that TiCl3 and TiF3 cannot accelerate hydrogen storage reactions, either by themselves or reduced to nanocrystalline material by ball-milling, due to the endergonic thermodynamic driving forces. However, the calculations provide possible insight into the role of Ti-containing additives in promoting hydrogen-coupled electron transfer when placed at the interface with another active material that can act as an electron source. The calculated electronic density of states further indicates that the alignment of TiF6 octahedra critically controls the electronic properties of TiF3 polymorphs. This implies that TiF3 must be modified by reacting and creating a different chemical species; forming reactive phase boundaries; or being mechanically strained in order to facilitate charge transfer between Ti and H, and hence to have an effect on hydrogen storage reaction kinetics.

Acknowledgments The authors acknowledge support through the Hydrogen Storage Materials Advanced Research Consortium of the U.S. Department of Energy (DOE), Office of Energy Efficiency and Renewable Energy, Fuel Cell Technologies Office under Contracts DE-AC52-07NA27344 and DE-AC04-94AL85000. Part of the work was performed under the auspices of the DOE by Lawrence Livermore National Laboratory under Contract DEAC52-07NA27344. Sandia National Laboratories is a multimission laboratory managed by National Technology and Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International Inc., for the DOE's National Nuclear Security Administration under contract DE-NA0003525. Portions of this research were performed on BLs 6.3.1.2 and 8.0.1.1 at the Advanced Light Source, Lawrence Berkeley National Laboratory, which is supported by the Director, Office of Science, Office of Basic Energy Sciences, of the U.S. DOE under Contract DE-AC02e05CH11231. Portions of the research described in this paper were performed at the CLS on the REIXS beamline BL10ID-2, which is supported by the Canada Foundation for Innovation, Natural Sciences and Engineering Research Council of Canada, the University of Saskatchewan, the Government of Saskatchewan, Western Economic

11

Diversification Canada, the National Research Council Canada, and the Canadian Institutes of Health Research. Additional computing resources were provided under the LLNL Institutional Computing Grand Challenge program. The views and opinions of the authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, expressed or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights.

Appendix A. Supplementary data Supplementary data related to this article can be found at https://doi.org/10.1016/j.ijhydene.2018.05.128.

references

[1] Klebanoff LE. Hydrogen storage Technology: materials and applications. Boca Raton: Taylor and Francis; December 12, 2012. [2] Klebanoff LE, Keller JO. 5 years of hydrogen storage research in the U.S. DOE metal hydride center of excellence (MHCoE). Int J Hydrogen Energy 2013;38(19):4533e76. [3] Zhou CS, Fang ZZ, Sun P. An experimental survey of additives for improving dehydrogenation properties of magnesium hydride. J Power Sources 2015;278:38e42. [4] Danaie M, Mitlin D. TEM analysis of the microstructure in TiF3-catalyzed and pure MgH2 during the hydrogen storage cycling. Acta Mater 2012;60(19):6441e56. [5] Grzech A, Lafont U, Magusin P, Mulder F. Microscopic study of TiF3 as hydrogen storage catalyst for MgH2. J Phys Chem C 2012;116(49):26027e35. [6] Ma LP, Wang P, Cheng HM. Hydrogen sorption kinetics of MgH2 catalyzed with titanium compounds. Int J Hydrogen Energy 2010;35(7):3046e50. [7] Xie L, Liu Y, Wang Y, Zheng J, Li X. Superior hydrogen storage kinetics of MgH2 nanoparticles doped with TiF3. Acta Mater 2007;55(13):4585e91. [8] Ma LP, Wang P, Cheng HM. Improving hydrogen sorption kinetics of MgH2 by mechanical milling with TiF3. J Alloy Comp 2007;432(1e2):L1e4. [9] Du AJ, Smith SC, Yao XD, Lu GQ. The role of Ti as a catalyst for the dissociation of hydrogen on a Mg(0001) surface. J Phys Chem B 2005;109(38):18037e41. [10] Liu XF, Beattie SD, Langmi HW, McGrady GS, Jensen CM. Tidoped LiAlH4 for hydrogen storage: rehydrogenation process, reaction conditions and microstructure evolution during cycling. Int J Hydrogen Energy 2012;37(13):10215e21. [11] Graetz J, Reilly J, Yartys V, Maehlen J, Bulychev B, Antonov V, et al. Aluminum hydride as a hydrogen and energy storage material: past, present and future. J Alloy Comp 2011;509:S517e28. [12] Liu SS, Sun LX, Zhang Y, Xu F, Zhang J, Chu HL, et al. Effect of ball milling time on the hydrogen storage properties of TiF3doped LiAlH4. Int J Hydrogen Energy 2009;34(19):8079e85. [13] Liu XF, McGrady GS, Langmi HW, Jensen CM. Facile cycling of Ti-doped LiAlH4 for high performance hydrogen storage. J Am Chem Soc 2009;131(14):5032.

Please cite this article in press as: Kang S, et al., Assessing the reactivity of TiCl3 and TiF3 with hydrogen, International Journal of Hydrogen Energy (2018), https://doi.org/10.1016/j.ijhydene.2018.05.128

12

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 8 ) 1 e1 3

[14] Bogdanovic B, Eberle U, Felderhoff M, Schuth F. Complex aluminum hydrides. Scripta Mater 2007;56(10):813e6. [15] Balde CP, Stil HA, van der Eerden AMJ, de Jong KP, Bitter JH. Active Ti species in TiCl3-doped NaAlH4. Mechanism for catalyst deactivation. J Phys Chem C 2007;111(6):2797e802. [16] Singh S, Eijt S, Huot J, Kockelmann W, Wagemaker M, Mulder F. The TiCl3 catalyst in NaAlH4 for hydrogen storage induces grain refinement and impacts on hydrogen vacancy formation. Acta Mater 2007;55(16):5549e57. [17] Chaudhuri S, Graetz J, Ignatov A, Reilly J, Muckerman J. Understanding the role of Ti in reversible hydrogen storage as sodium alanate: a combined experimental and density functional theoretical approach. J Am Chem Soc 2006;128(35):11404e15. [18] Resan M, Hampton M, Lomness J, Slattery D. Effects of various catalysts on hydrogen release and uptake characteristics of LiAlH4. Int J Hydrogen Energy 2005;30(13e14):1413e6. [19] Andrei CM, Walmsley JC, Brinks HW, Holmestad R, Srinivasan SS, Jensen CM, et al. Electron-microscopy studies of NaAlH4 with TiF3 additive: hydrogen-cycling effects. Appl Phys Mater Sci Process 2005;80(4):709e15. [20] Brinks HW, Fossdal A, Fonnelop JE, Hauback BC. Crystal structure and stability of LiAlD4 with TiF3 additive. J Alloy Comp 2005;397(1e2):291e5. [21] Resan M, Hampton M, Lomness J, Slattery D. Effect of TixAly catalysts on hydrogen storage properties of LiAlH4 and NaAlH4. Int J Hydrogen Energy 2005;30(13e14):1417e21. [22] Anton DL. Hydrogen desorption kinetics in transition metal modified NaAlH4. J Alloy Comp 2003;356:400e4. [23] Chen J, Kuriyama N, Xu Q, Takeshita H, Sakai T. Reversible hydrogen storage via titanium-catalyzed LiAlH4 and Li3AlH6. J Phys Chem B 2001;105(45):11214e20. [24] Jensen CM, Gross KJ. Development of catalytically enhanced sodium aluminum hydride as a hydrogen-storage material. Appl Phys Mater Sci Process 2001;72(2):213e9. [25] Bogdanovic B, Brand R, Marjanovic A, Schwickardi M, Tolle J. Metal-doped sodium aluminium hydrides as potential new hydrogen storage materials. J Alloy Comp 2000;302(1e2):36e58. [26] Palumbo O, Paolone A, Cantelli R, Jensen C, Sulic M. Fast Hvacancy dynamics during alanate decomposition by anelastic spectroscopy. Proposition of a model for Tienhanced hydrogen transport. J Phys Chem B 2006;110(18):9105e11. [27] Newhouse RJ, Stavila V, Hwang SJ, Klebanoff LE, Zhang JZ. Reversibility and improved hydrogen release of magnesium borohydride. J Phys Chem C 2010;114(11):5224e32. [28] Guo Y, Yu X, Gao L, Xia G, Guo Z, Liu H. Significantly improved dehydrogenation of LiBH4 destabilized by TiF3. Energy Environ Sci 2010;3(4):465e70. [29] Mao JF, Yu XB, Guo ZP, Liu HK, Wu Z, Ni J. Enhanced hydrogen storage performances of NaBH4-MgH2 system. J Alloy Comp 2009;479(1e2):619e23. [30] Kim JH, Jin SA, Shim JH, Cho YW. Reversible hydrogen storage in calcium borohydride Ca(BH4)2. Scripta Mater 2008;58(6):481e3. [31] Wang PJ, Ma LP, Fang ZZ, Kang XD, Wang P. Improved hydrogen storage property of Li-Mg-B-H system by milling with titanium trifluoride. Energy Environ Sci 2009;2(1):120e3. [32] Kim JH, Shim JH, Cho YW. On the reversibility of hydrogen storage in Ti- and Nb-catalyzed Ca(BH4)2. J Power Sources 2008;181(1):140e3. [33] Ronnebro E, Majzoub E. Calcium borohydride for hydrogen storage: catalysis and reversibility. J Phys Chem B 2007;111(42):12045e7. [34] Vajo JJ, Skeith SL, Mertens F. Reversible storage of hydrogen in destabilized LiBH4. J Phys Chem B 2005;109(9):3719e22.

[35] Ichikawa T, Hanada N, Isobe S, Leng H, Fujii H. Hydrogen storage properties in Ti catalyzed Li-N-H system. J Alloy Comp 2005;404:435e8. [36] Ichikawa T, Isobe S, Hanada N, Fujii H. Lithium nitride for reversible hydrogen storage. J Alloy Comp 2004;365(1e2):271e6. [37] Chen H, Yang R. Catalytic effects of TiF3 on hydrogen spillover on Pt/carbon for hydrogen storage. Langmuir 2010;26(19):15394e8. [38] Liu X, Beattie S, Langmi H, McGrady G, Jensen C. Ti-doped LiAlH4 for hydrogen storage: rehydrogenation process, reaction conditions and microstructure evolution during cycling. Int J Hydrogen Energy 2012;37(13):10215e21. [39] Jensen C, Gross K. Development of catalytically enhanced sodium aluminum hydride as a hydrogen-storage material. Appl Phys Mater Sci Process 2001;72(2):213e9. [40] Ma LP, Kang XD, Dai HB, Liang Y, Fang ZZ, Wang PJ, et al. Superior catalytic effect of TiF3 over TiCl3 in improving the hydrogen sorption kinetics of MgH2: catalytic role of fluorine anion. Acta Mater 2009;57(7):2250e8. [41] Kang XD, Wang P, Cheng HM. Advantage of TiF3 over TiCl3 as a dopant precursor to improve the thermodynamic property of Na3AlH6. Scripta Mater 2007;56(5):361e4. [42] Yin LC, Wang P, Kang XD, Sun CH, Cheng HM. Functional anion concept: effect of fluorine anion on hydrogen storage of sodium alanate. Phys Chem Chem Phys 2007;9(12):1499e502. [43] Erbil A, Cargill GS, Frahm R, Boehme RF. Total-electronyield current measurements for near-surface extended xray absorption fine-structure. Phys Rev B 1988;37(5): 2450e64. [44] Jaklevic J, Kirby JA, Klein MP, Robertson AS, Brown GS, Eisenberger P. Fluorescence detection of EXAFS - sensitivity enhancement for dilute species and thin-films. Solid State Commun 1977;23(9):679e82. [45] Zmbov KF, Margrave JL. Mass spectrometric studies at high temperatures. XVI. Sublimation pressures for TiF3 and stabilities of TiF2(g) and TiF(g). J Phys Chem 1967;71(9):2893e5. [46] Sanderson BS, MacWood GE. Thermodynamics of the titanium chlorides. III: the sublimation pressure of titanium trichloride. J Phys Chem 1956;60(3):314e6. [47] Perdew JP, Burke K, Ernzerhof M. Generalized gradient approximation made simple. Phys Rev B 1996;77:3865e8. € chl PE. Projector augmented-wave method. Phys Rev B [48] Blo 1994;50:17953. [49] Kresse G, Furthmu¨ller J. Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set. Phys Rev B 1996;54:11169e86. [50] Perebeinos V, Vogt T. Jahn-Teller transition in TiF3 investigated using density-functional theory. Phys Rev B 2004;69(11). 115102. [51] Chase MW. NIST-JANAF Thermochemical Tables. Washington, D.C.; Woodbury, N.Y: American Institute of Physics for the National Institute of standards and Technology; 1998. [52] Materials Project. http://www.materialsgenome.org/. [53] Natta G, Allegra G, Corradini P. The different crystalline modifications of TiCl3, a catalyst component for polymerization of alpha-olefins. I: a-, b-, Y-TiCl3. II: d-TiCl3. J Polym Sci 1961;51(156):399e410. [54] Wilchinsky ZW, Looney RW, Tornovist EG. Dependence of polymerization activity on particle and crystallite dimensions in ball milled TiCl3 and TiCl3-0.33AlCl3 catalyst components. J Catal 1973;28(3):351e67. [55] Miyaoka H, Kuze T, Sano H, Mori H, Mizutani G, Ushioda S, et al. Raman Spectrum of ball-milled a-TiCl3. J Lumin 2000;87e89:709e11.

Please cite this article in press as: Kang S, et al., Assessing the reactivity of TiCl3 and TiF3 with hydrogen, International Journal of Hydrogen Energy (2018), https://doi.org/10.1016/j.ijhydene.2018.05.128

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 8 ) 1 e1 3

[56] Fraser GV, Chalmers JM, Charlton V, Cudby MEA. A vibrational spectroscopic Study of TiCl3. Solid State Commun 1977;21:933e6. [57] Kanesaka I, Yonezawa M, Kawai K, Miyatake T, Kakugo M. Vibrational spectrum of a- and d-TiCl3. Spectrochim Acta 1986;42A:1415e7. [58] Luo WF, Cowgill DF, Causey RA. Thermodynamic and kinetic characterization of hydrogen-deuterium exchange in betaphase palladium. J Phys Chem B 2009;113(39):12978e87.

13

[59] Kennedy BJ, Vogt T. Powder X-ray diffraction study of the rhombohedral to cubic phase transition in TiF3. Mater Res Bull 2002;37(1):77e83. [60] Cui J, Wang H, Liu J, Ouyang L, Zhang Q, Sun D, et al. Remarkable enhancement in dehydrogenation of MgH2 by a nano-coating of multi-valence Ti-based catalysts. J Math Chem A 2013;1:5603e11. [61] Milosevic S, Kurko S, Pasquini L, Matovic L, Vujasin R, Novakovic N, et al. J Power Sources 2016;307:481e8.

Please cite this article in press as: Kang S, et al., Assessing the reactivity of TiCl3 and TiF3 with hydrogen, International Journal of Hydrogen Energy (2018), https://doi.org/10.1016/j.ijhydene.2018.05.128