Investigating possible kinetic limitations to MgB2 hydrogenation

international journal of hydrogen energy xxx (xxxx) xxx

Available online at www.sciencedirect.com

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

Investigating possible kinetic limitations to MgB2 hydrogenation Y.-S. Liu a, L.E. Klebanoff b,*, P. Wijeratne b, D.F. Cowgill b, V. Stavila b, T.W. Heo c, S. Kang c, A.A. Baker c, J.R.I. Lee c, T.M. Mattox a, K.G. Ray c, J.D. Sugar b, B.C. Wood c a

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

article info

abstract

Article history:

An investigation is reported of possible kinetic limitations to MgB2 hydrogenation. The role

Available online xxx

of HeH bond breaking, a necessary first step in the hydrogenation process, is assessed for bulk MgB2, ball-milled MgB2, as well as MgB2 mixed with Pd, Fe and TiF3 additives. The Pd

Keywords:

and Fe additives in the MgB2 material exist as dispersed metallic particles in the size range

Hydrogen storage

~5e40 nm diameter. In contrast, TiF3 reacts with MgB2 to form Ti metal, elemental B and

Magnesium diboride

MgF2, with the Ti and the MgF2 phases proximate to each other and coating the MgB2

Hydrogenation

particulates with a film of thickness ~3 nm. Sieverts studies of hydrogenation kinetics are

Kinetics

reported and compared to the rate of HeH bond breaking as measured by H-D exchange

H-D exchange

studies. The results show that HeH bond dissociation does not limit the rate of hydroge-

Surface diffusion

nation of MgB2 because HeH bond cleavage occurs rapidly compared to the initial MgB2 hydrogenation. The results also show that surface diffusion of hydrogen atoms cannot be a limiting factor for MgB2 hydrogenation. Instead, it is speculated that it is the intrinsic stability of the BeB extended hexagonal ring structure in MgB2 that hinders the hydrogenation of this material. This supposition is supported by B K-edge x-ray absorption measurements of the materials, which showed spectroscopically that the BeB ring was intact in the material systems studied. The TiF3/MgB2 system was examined further theoretically with reaction thermodynamics and phase nucleation kinetic calculations to better understand the production of Ti metal when TiB2 is thermodynamically favored. The results show that there exist physically reasonable ranges for which nucleation kinetics supersede thermodynamics in determining the reactive pathway for the TiF3/MgB2 system and perhaps for other additive systems as well. © 2019 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.

Introduction Hydrogen fuel cells are finding deeper and wider use in our technical society to provide zero-emissions electrical power.

The advantages of fuel cells versus diesel engines are higher thermal efficiency (especially at partial load), dramatically lower noise, reduced maintenance and associated costs, and emissions-free operation at the point of use [1]. In addition, the modularity of fuel cells allows for power architectures that

* Corresponding author. E-mail address: [email protected] (L.E. Klebanoff). https://doi.org/10.1016/j.ijhydene.2019.09.125 0360-3199/© 2019 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved. Please cite this article as: Liu Y-S et al., Investigating possible kinetic limitations to MgB2 hydrogenation, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.09.125

2

international journal of hydrogen energy xxx (xxxx) xxx

can be more efficiently matched to the power utilization profile of the particular application. Most major auto manufacturers have designs for light-duty fuel-cell vehicles, with Toyota, Honda and Hyundai already making these vehicles available to the public for sale or lease. Beyond light-duty vehicles, fuel-cell versions of buses have been in service already for many years [2], fuel-cell construction lighting equipment has been demonstrated [3], fuel-cell forklifts are now in routine use by the thousands [4] and the first light rail systems operated by hydrogen fuel cells are now operational [5]. In the last few years the feasibility of using hydrogen fuel cells to replace marine diesel engines in maritime applications has been shown, for example in high-speed ferries [6] and coastal research vessels [7]. A hydrogen fuel-cell ferry is currently under construction for use on the San Francisco Bay beginning in the Fall of 2019 [8]. Fuel-cell vessels represent a high-impact use of hydrogen. A typical high-speed ferry would use the equivalent amount of hydrogen as ~2000 light duty vehicles. Still, the sheer number of light-duty vehicles (~1 billion worldwide) makes hydrogen fuel-cell cars the largest technical driver of hydrogen fuel-cell technology development. The current light-duty fuel-cell cars store hydrogen as a high-pressure gas, with the hydrogen stored at either 350 bar or 700 bar in composite hydrogen tanks. These tanks provide the required hydrogen gravimetric and volumetric storage densities for the rollout of fuel-cell vehicles with acceptable range. However, if the hydrogen storage densities could be improved, an even greater range for fuel-cell cars could be realized. Liquid hydrogen (LH2) can provide higher hydrogen storage densities. However, the relatively small amount of hydrogen needed for light-duty fuel-cell vehicles, ~5 kg, makes LH2 storage difficult because of the higher heat leak rate (and subsequent boil-off) associated with small cryogenic tanks. It has been recognized for many years that solid-state storage of hydrogen has the potential to improve hydrogen storage efficiency beyond that achieved with 700 bar tanks. For example, while the volumetric storage density of 700 bar hydrogen gas is 40 g/L, the volumetric hydrogen storage density of aluminum hydride (AlH3) is 149 g/L, even higher than that of LH2 (71 g/L). The challenge for solid-state hydrogen storage materials is that they must satisfy at least 4 major criteria: 1) they must release hydrogen quickly enough when demanded by the fuel cell (~1.6 g/s for a 80 kW fuel cell); 2) the hydrogen must be very pure (as specified by the SAE J2719 standard [9]) to avoid poisoning the fuel-cell catalyst; 3) the material must release hydrogen with very little input energy (have the right thermodynamics) and 4) the material must be able to be refueled (rehydrogenated) in 3e5 min with gaseous hydrogen [10]. All known hydrogen storage materials fail to meet at least one of these requirements for the light-duty fuelcell vehicle application. For example, AlH3, which is an exemplary hydrogen storage material in many respects, forms Al metal when its hydrogen is released. However, it is essentially impossible, in the context of a hydrogen refueling station, for Al metal to be rehydrogenated back to AlH3 [11]. It has become clear to the hydrogen storage community that addressing the technical shortcomings of current hydrogen storage materials requires a full understanding of the

fundamental chemical and physical limitations to performance. With this understanding, overcoming such shortcomings will enable these materials to be used to store hydrogen for light-duty fuel-cell vehicles and other applications. Magnesium borohydride, Mg(BH4)2, is a material with a very high hydrogen capacity of 14.9 wt percent which can be achieved by the following overall reaction: Mg(BH4)2¼ 4H2 þ MgB2, with the detailed dehydrogenation pathway complicated by intermediate species. Much work has been devoted to understanding how this material releases hydrogen, as reviewed by Zavorotynska et al. [12], as well as in understanding the properties of destabilized reactive hydride composites (RHC) involving MgB2 as a product [13e15]. In comparison, there have been relatively few studies aimed at understanding how MgB2 can be hydrogenated to form Mg(BH4)2 [16e19], as reviewed by Ray et al. [17]. Although possible, the complete conversion of MgB2 to Mg(BH4)2 requires excessively high pressures (950 bar) and temperatures (400  C) and requires too much time (a day) for practical applications [19]. Ray et al. [17] investigated the initial hydrogenation of the material both experimentally and theoretically to understand the mechanism by which hydrogen first absorbs to MgB2. However, the fundamental reasons why MgB2 is so slow to rehydrogenate are not understood. Given the dramatic improvement needed in the rehydrogenation kinetics of the Mg(BH4)2/MgB2 system, we embarked on a study to examine two potential limitations to the rehydrogenation kinetics: HeH bond breaking and surface diffusion of H atoms. Our primary objective is to assess if HeH bond breakingda necessary first step in the hydrogenation of MgB2 by H2 gasdis a rate-limiting reaction step. Our approach is to introduce additives into MgB2 that are known to dissociate H2 and probe whether or not their introduction increases the rate of the initial MgB2 hydrogenation observed at the relatively modest applied hydrogen pressures of ~130 bar used in a Sieverts apparatus. For this study we have chosen three additives: Pd metal, Fe metal, and TiF3. Pd metal and Fe metal have demonstrated their ability to catalytically break the HeH bond in diverse morphological forms. Pd metal dissociates hydrogen at its (100), (110) and (111) surfaces [20e22]. The (100), (110) and (111) surfaces of Fe also readily dissociate the HeH bond [23e25]. Hydrogen can also be dissociated by thin films of both Pd [26,27] and Fe [27,28], as well as by dopants of these metals [29e31]. TiF3 is in a different class, not being an elemental metal, but nonetheless has demonstrated positive influence on the hydrogen storage kinetics of some systems as reviewed by Kang et al. [32]. Such a promotion must involve chemical transformation of TiF3, since it has been shown that TiF3 is completely inert to hydrogen under the relevant conditions of hydrogen temperature and pressure [32]. We shall see that Ti metal is produced when TiF3 is combined with MgB2 under our experimental conditions, a result we examine theoretically. Ti metal is known to dissociate hydrogen as well, whether in the form of a single-crystal surface [33,34], as a film [27] or as a dopant [31,35]. For these additives to promote HeH bond breaking, they should be introduced into the MgB2 solid preferably without oxidation or other contamination, since their oxides do not dissociate hydrogen [36e38]. Also, their structural and

Please cite this article as: Liu Y-S et al., Investigating possible kinetic limitations to MgB2 hydrogenation, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.09.125

international journal of hydrogen energy xxx (xxxx) xxx

chemical state in the MgB2 matrix must be understood. Toward these ends, we have extensively characterized the materials produced by ball-milling the additives Pd, Fe and TiF3 with MgB2 using powder x-ray diffraction (XRD), Fourier Transform Infra-red (FTIR) spectroscopy, x-ray photoelectron spectroscopy (XPS), x-ray absorption spectroscopy (XAS) and transmission electron microscopy (TEM) combined with energy-dispersive x-ray spectroscopy (EDS). With the structural/chemical aspects understood, the additive/MgB2 materials are then exposed to hydrogen in a Sieverts apparatus to assess the effect of the additive on the kinetics of MgB2 hydrogenation. To independently confirm the promotion of HeH bond breaking in these materials, we use a H-D exchange approach. While H-D exchange studies have been used before in studies of dehydrogenation [39e41], its use for hydrogenation studies is relatively new [32]. Since H-D exchange measurements can also involve H atom surface diffusion, the role of hydrogen diffusion on the MgB2 hydrogenation rate can also be assessed. Finally, we supplement these experimental studies with theoretical work clarifying the physical phenomena involved, including the size of the surface diffusion barrier for H atoms on MgB2 and its relation to the initial hydrogenation mechanism. In addition, we report results for nucleation barrier calculations that help explain the products formed when TiF3 is ball-milled with MgB2.

Experimental and theoretical methods 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 MgB2 and additives where purchased from commercial sources. The ball-milled materials were produced by loading tungsten carbide (WC) mill pots with the appropriate commercially available chemicals and milling with WC balls under argon for the times indicated. Details are provided in the Supporting Information (SI). A variety of substances used in the study were either prepared or used directly from commercial sources without further purification. These include: commercial (or "bulk") MgB2; MgB2 ball-milled for 1 h and MgB2 ball-milled for 2 h. In addition, a series of [MgB2 þ additive] samples were made including: 1. MgB2 þ 4.88% Fe; where Fe powder was added to the commercial MgB2 powder and milled for 2 h. The mole fraction of Fe to MgB2 was 0.0488, or 4.88% Fe. 2. MgB2 þ 4.97% Pd; Pd powder was added to the commercial MgB2 powder and milled for 2 h. The mole fraction of Pd to MgB2 (moles Pd/moles MgB2) was 0.0497, or 4.97% Pd. 3. MgB2 þ 4.46% TiF3; where TiF3 powder was added to the commercial MgB2 powder and milled for 2 h. The mole fraction of TiF3 to MgB2 was 0.0446, or 4.46% TiF3 More details of the synthesis of these materials can be found in the SI. An extensive series of characterization studies were performed to check for sample integrity, potential contamination

3

during handling and ball-mill processing and for possible reaction products. X-ray diffraction was used to investigate crystal structures and to check for crystalline contamination or reaction products. In addition, FTIR spectroscopy was used to check for amorphous contamination and reaction products that would not be detected by XRD. Element-specific corelevel XPS and XAS was used to evaluate the local electronic structure at the B, Mg and additive sites. The details of the materials characterization instruments are provided in the SI. XAS data were simultaneously recorded in two modes: total electron yield (TEY) [42] (via measurement of the drainage current to the experiment sample) and total fluorescence yield (TFY) [43] (using a channeltron electron multiplier). All XAS spectra presented are normalized to the incident x-ray flux, I0, which was measured concurrently as a drainage current from an upstream gold mesh and scaled to the magnitude of the absorption edge step. The depth sensitivities of the XPS measurements as well as the TFY/ TEY XAS data are discussed in the SI for the various core levels. The capacity for the materials under study to dissociate H2 was assessed using a Sandia-built “in-house” apparatus that provides exposure to calibrated mixtures of H2 and D2, at ~10 Torr pressure and for temperatures from 22  C to 300  C. The breaking of HeH and D-D bonds were monitored by measuring the partial pressure of HD generated as a function of time with a residual gas analyzer (RGA). Fig. S1 in the SI shows a diagram of the H2/D2 isotope exchange experimental setup. The method of H-D exchange measurement is described in the SI. The samples were exposed to hydrogen in a Sieverts apparatus to assess the rate of hydrogen uptake. The Sieverts measurements employed a PCTPro 2000 apparatus from Setarem. The samples were loaded inside a sample holder inside the Ar glove box, with a thermocouple 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 of H absorbed with respect to the total sample weight: wt.% absorbed ¼ [mass H absorbed]/[(mass H absorbed þ mass MgB2 original sample)] x 100. To understand better the chemical reaction that was observed between MgB2 and TiF3, the reaction thermodynamics and nucleation kinetics for this system were computed by combining the NIST-JANAF thermochemical database [44], density functional theory (DFT) total energy calculations and classical nucleation theory [45]. Bulk and surface energies of involved phases were obtained from the NIST-JANAF database or literatures if available, and otherwise computed within DFT using Vienna ab initio Simulation Package [46]. More computational details are described in the SI, some of which are also directed at understanding the role of surface diffusion in the H-D exchange studies.

Please cite this article as: Liu Y-S et al., Investigating possible kinetic limitations to MgB2 hydrogenation, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.09.125

4

international journal of hydrogen energy xxx (xxxx) xxx

Results Materials characterization To fully understand the effect of the additives on the hydrogenation of MgB2, the chemical and structural nature of these additive/MgB2 systems requires detailed characterization. The SI provides full details of the XRD, FTIR, XPS, XAS and TEM studies of these systems. The essential results are summarized here.   MgB2 þ 4:88%Fe We refer to the sample made by ball milling together for 2 h in argon a mixture of MgB2 with 4.88% mole fraction of Fe metal powder as [MgB2 þ 4.88% Fe]. The starting Fe powder was characterized by XRD and FTIR to check for iron oxide contamination. The XRD results indicated no evidence for crystalline oxide contamination. However, FTIR indicated trace oxidation of Fe to amorphous Fe3O4. Neither XRD nor FTIR measurements for [MgB2 þ 4.88% Fe] showed any evidence for MgO or BeO species (e.g. B2O3) which could have been produced by an air leak in the WC milling vial. No evidence was found for FeB formation, which could be formed if the Fe metal reacted with MgB2. The detailed results are presented in the SI. The morphology and elemental composition of [MgB2 þ 4.88% Fe] was examined with TEM in combination with EDS. Characteristic TEM images are presented in Fig. 1, with more results provided in the SI. The High Angle Annular Dark Field (HAADF) image shows ~200e300 nm MgB2 particles. The elemental maps show that the particles consist of Mg and B, consistent with MgB2. The MgB2 particles are decorated with small Fe particles, with a range of particle sizes from ~12 to 40 nm in diameter, whose d-spacing is identical to the original starting Fe powder, as revealed by XRD data (see the SI). There was no evidence for atomic-scale dispersal of the Fe additive.   MgB2 þ 4:97%Pd The sample made by ball milling together a mixture of MgB2 with 4.97% mole fraction of Pd metal powder for 2 h under argon is labelled [MgB2 þ 4.97% Pd]. The starting Pd metal powder, consisting of micron-sized particulates, was synthesized as described in the SI. XRD and FTIR measurements of [MgB2 þ 4.97% Pd] showed no evidence for oxidation of the sample (e.g. MgO, B2O3, PdO). The XRD showed evidence for a small amount of Pd2B, as described in the SI. TEM was used to examine the morphological and elemental composition of [MgB2 þ 4.97% Pd]. Typical TEM images are shown in Fig. 2, with more results displayed in the SI. The HAADF and EDS images show well-defined ~300 nm diameter particles of MgB2, decorated with discrete Pd particles with ~5e12 nm diameter. The XRD of this sample showed the Pd additive has a d-spacing identical to that of

bulk Pd. Analysis of the TEM images indicated no evidence for an atomically dispersed Pd component in the MgB2 matrix.   MgB2 þ 4:46%TiF3 Characterization measurements were performed for the mixture of MgB2 with 4.46% mole fraction of TiF3 powder, milled for 2 h in an Ar-filled milling vial. We refer to this sample as [MgB2 þ 4.46% TiF3]. A full account of the characterization of this sample is given in the Appendix, with additional details in the SI. XRD and FTIR studies of [MgB2 þ 4.46% TiF3] presented in the Appendix and SI show the complete disappearance of the TiF3 reactant. Since our characterization work showed that TiF3 is stable with ball-milling (see Fig. S20 in the SI), the disappearance of TiF3 signals chemical reaction with MgB2. XAS and XPS data shown in the Appendix identifies the reaction products Ti metal, MgF2 and elemental B according to the chemical reaction: TiF3 þ 3/2MgB2 ¼ Ti þ 3/2MgF2 þ 3B. TEM was used to examine the morphological and elemental composition of [MgB2 þ 4.46% TiF3]. Typical results are show in Fig. 3, with additional TEM data provided in the SI. The TEM images of [MgB2 þ 4.46% TiF3] show mostly ~ 300 nm MgB2 particles as well as a few smaller ones. The EDS maps show that the particles consist of MgB2. There is no evidence for discrete additive particles in [MgB2 þ 4.46% TiF3], of either Ti metal or MgF2. Rather, the Ti and the F seem so be smeared out on the MgB2 base solid. This finding is consistent with the XPS data presented in the Appendix. The dispersion is not complete, as the F is generally spatially associated with the Ti, indicating that the reaction remains somewhat localized. When analyzed in conjunction with the XPS results, these TEM findings are consistent with the formation of films of Ti metal and MgF2 spread across the MgB2 particles.

Hydrogenation and H-D exchange results Sieverts hydrogenation measurements were conducted for commercial (bulk) MgB2, MgB2 ball-milled for 1 h, [MgB2 þ 4.88% Fe], [MgB2 þ 4.97% Pd] and [MgB2 þ 4.46% TiF3]. The results for hydrogen uptake at hydrogen pressures ranging from 130 to 140 bar and temperatures from 362 to 377  C are shown in Fig. 4. The data for MgB2 without additives are generally consistent with prior work [17], where over the course of 100 h, only ~ 0.8 wt percent hydrogen has been added to the sample. This is a significant amount of hydrogenation, corresponding to a H/MgB2 mole ratio of 0.4. However, since full conversion of MgB2 to Mg(BH4)2 would correspond to 14.9 wt percent of hydrogen added, the level of hydrogenation seen in Fig. 4 is far below the hydrogenation potential of MgB2 and what is needed for practical use of this material for vehicular hydrogen storage. Prior work [17] assigned the initial hydrogenation to the formation of Mg(BH4)2 from MgB2 without expression of persistent intermediates. Subsequent theoretical work [47] has recently shown this direct hydrogenation to Mg(BH4)2 is favored thermodynamically for the pressures and

Please cite this article as: Liu Y-S et al., Investigating possible kinetic limitations to MgB2 hydrogenation, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.09.125

international journal of hydrogen energy xxx (xxxx) xxx

5

Fig. 1 e TEM investigation of [MgB2 þ 4.88% Fe], giving the particle size, morphology and the elemental distributions as determined by EDS. (Top): EDS results for Fe and B; (Bottom): EDS results for Mg and a HAADF image.

temperature of Fig. 4. This theoretical result contrasts with a prior calculation by Pinatel et al. [48] that found the combination of MgB2 and gaseous H2 does not react under our experimental conditions, although these authors pointed out that their results were not consistent with the available experimental data. We did not characterize the hydrogenation products in this work, since neither our focus on the roles of HeH bond breaking and surface diffusion in MgB2 hydrogenation nor the resulting conclusions required it. As can be seen from Fig. 4, the MgB2 hydrogenation rates for the MgB2 ball-milled for 1 h, [MgB2 þ 4.88% Fe], [MgB2 þ 4.97% Pd] and [MgB2 þ 4.46% TiF3] are somewhat higher than for commercial (as received) MgB2, but the improvement is very modest, about a factor of two at most, depending on the exposure time and sample. In addition, the amount of hydrogen added remains very modest, ~1 wt percent hydrogen. Thus, we find no significant improvement in the rate of hydrogenation of MgB2, despite the addition of metals (Pd, Fe and Ti) known to dissociate HeH bonds even at room temperature. Since a modest (but smaller) improvement is also seen for ball-milled MgB2 without additives, it is difficult to determine whether the relatively modest uptake improvements seen in the additive samples (which were also

made by ball milling) is attributable to a functioning of the metal additive itself or else to improvements caused by the creation of reactive defects in the MgB2 host from ball milling, although we suspect both causes of modest improvement are present. One could attempt to assign the lack of significant improvement in the MgB2 hydrogenation rate with additives to a deactivation of the additive by oxidation, or some other form of additive contamination. However, our extensive characterizations show that the Pd, Fe and Ti additives exist primarily as metals in the MgB2 matrix. There is no evidence for pathological oxidation which would inactivate the HeH bond breaking ability of these additives, or of chemical reaction to a boride which also might inactivate the additives. Furthermore, there is no evidence for oxidation of the MgB2 itself to MgO or BeO species which would inhibit the rate of hydrogenation. Rather, the characterization data, combined with the Sieverts data of Fig. 4, indicate that the metals introduced are capable of HeH bond breaking in the material. We can reasonably conclude that the lack of significant enhancement of hydrogen uptake in the data of Fig. 4 points to HeH bond breaking not being rate-limiting for MgB2 hydrogenation.

Please cite this article as: Liu Y-S et al., Investigating possible kinetic limitations to MgB2 hydrogenation, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.09.125

6

international journal of hydrogen energy xxx (xxxx) xxx

Fig. 2 e TEM investigation of [MgB2 þ 4.97% Pd], giving the particle size, morphology and the elemental distributions as determined by EDS. (Top): EDS results for Pd and B; (Bottom): EDS results for Mg and a HAADF image.

To confirm the HeH bond breaking activity of the materials, and to directly measure the rate of HeH bond breaking that is occurring, we conducted H-D exchange studies of these materials. The basic idea of the experiment is to establish a pressure (typically of order 10 Torr) of H2 and D2 over the sample and measure the rate of HD produced by the reaction sequence: H2(g) þ D2(g) / 2HD(g). This reaction does not occur in the gas phase and requires a material that can dissociate HeH and D-D bonds. If the sample can dissociate HeH and D-D bonds, then Hads and Dads atoms are produced on the surface. If the Hads and Dads atoms can diffuse across the surface and find each other, they will form HDads and then desorb into the gas-phase as HD(g) molecules via the socalled Langmuir-Hinshelwood mechanism [49]. Alternatively, it is possible for H-D exchange to occur by the initial formation of Hads and Dads, followed by reaction of Hads with D2(g) in the gas-phase (to form Dads and HD(g)) and by reaction of Dads with H2(g) (to form Hads and HD(g)). Such a mechanism is called the Eley-Rideal mechanism of exchange [49]. This alternative mechanism does not require surface diffusion for the HD exchange to occur. The SI gives an extensive discussion of these two mechanisms for producing HD on our samples. The conclusion of this analysis is that both the Langmuir-Hinshelwood and Eley-Rideal mechanisms are likely operative on MgB2.

Regardless of the mechanism for exchange, the H-D exchange reaction rate can be quantified by comparing the reaction quotient Q(HD) ¼ [HD]2/([H2]x[D2]) to the expected value at dynamic chemical equilibrium, where the brackets indicate partial pressures of the gas-phase species. Under chemical equilibrium at 22  C, Q(HD) ¼ K(HD) ¼ 4.2, decreasing slightly as temperature increases [50]. Fig. 5 displays the results of the HD exchange measurements, plotting the reaction quotient Q versus time for the pressures indicated at 200  C for the samples examined in this study. The experiments were conducted at 200  C because early scoping studies indicated that sufficient HD formation occurred on these samples at this temperature. For context, we also show results for Fe powder, which is well known to efficiently dissociate hydrogen. The data are plotted on a logarithmic Q scale for Fig. 5(a) and on a linear Q scale for Fig. 5(b). The primary conclusion from Fig. 5 is that HeH bond breaking is occurring for all samples, and on a time scale much faster than the MgB2 hydrogenation. This supports the earlier conclusion from the Sieverts measurements of Fig. 4 that HeH bond dissociation is not rate limiting for MgB2 hydrogenation. Fig. 5(a) shows that HeH bond breaking is slowest for the bulk MgB2 material, establishing equilibrium much more slowly than the other samples. It takes about 4 h for bulk MgB2 to reach Q ¼ 1.0 at 200  C. The [MgB2 þ 4.88% Fe],

Please cite this article as: Liu Y-S et al., Investigating possible kinetic limitations to MgB2 hydrogenation, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.09.125

international journal of hydrogen energy xxx (xxxx) xxx

7

Fig. 3 e TEM investigation of [MgB2 þ 4.46% TiF3], giving the particle size, morphology and the elemental distributions as determined by EDS. (Top): EDS results for B and Mg; (Middle): EDS results for Ti and F; (Bottom): HAADF image.

Fig. 4 e Hydrogen uptake measured as a function of time. The sample temperatures and nominal applied hydrogen pressures were: commercial (bulk) MgB2 (364.0  C, 130 bar), MgB2 ball milled for 1 h (364.0  C, 141 bar), [MgB2 þ 4.88% Fe] (365.1  C, 135 bar), [MgB2 þ 4.97% Pd] (342.7  C, 122 bar) and [MgB2 þ 4.46% TiF3] (346.7  C, 141 bar).

[MgB2 þ 4.97% Pd] and [MgB2 þ 4.46% TiF3] samples show dramatically improved rates of HeH and D-D bond breaking, as one would expect for samples with metal additives. For the ball milled MgB2 samples and the [MgB2 þ 4.88% Fe], [MgB2 þ 4.97% Pd] samples, the HeH bond breaking is very fast, and the reaction quotient Q approaches its value for dynamic equilibrium within 15 min at T ¼ 200  C. The somewhat slower rate observed for the [MgB2 þ 4.46% TiF3] sample is a consequence of the reduced mass used for this sample. The HD exchange was generally so fast that we decided to use only 11 mg of sample for [MgB2 þ 4.46% TiF3] (as opposed to the ~300e400 mg for the other additive samples), to slow the rate of HD exchange in order to better characterize the HD exchange kinetics for this sample. The rate of HD exchange at 200  C in Fig. 5 increases in the order: bulk MgB2 << MgB2 ball milled for 1 h < MgB2 ball milled for 2 h. This improvement is attributed to the creation of defects via ball-milling that promote HeH bond dissociation, as well as to an increase in the surface area. Induced reactive defects likely include broken BeB bonds and the creation of edge-sites with undercoordinated B atoms. Therefore, the HeH bond dissociation occurring in the [MgB2 þ 4.88% Fe], [MgB2 þ 4.97% Pd] and [MgB2 þ 4.46% TiF3] materials has two contributions: HeH bond breaking at the metallic additive

Please cite this article as: Liu Y-S et al., Investigating possible kinetic limitations to MgB2 hydrogenation, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.09.125

8

international journal of hydrogen energy xxx (xxxx) xxx

10

3

MgB2 + 4.46 % TiF3

all other samples

(a)

MgB2 + 4.46 % TiF3

1

300 °C MgB2

(a)

0.01

Q

Q

0.1

200 °C

1.5 LEK020 100C LEK020 200C

empty holder

0.001

LEK020 300C

100 °C 0.0001 0

50

100

150

0

200

0

Time (minutes) 3

LEK020 200C LEK020 300C

MgB2 + 4.97 % Pd

(b)

MgB2 BM 1 hr

Q

Q

MgB2LEK020 + 4.46 % TiF3 100C MgB2 BM 2 hrs

2

1

100 150 Time (minutes)

200

3 Fe

2.5

1.5

50

200 °C

300 °C

1.5

MgB2 + 4.88 % Fe

0.5

(b)

100 °C

MgB2 + 4.46 % TiF3

0 0

5

10

15

Time (minutes) Fig. 5 e (a) The reaction quotient Q(HD) ¼ [HD]2/([H2]x[D2]) plotted on a log ordinate scale versus time (hours) for Fe powder, bulk MgB2, MgB2 ball milled for 1 h, MgB2 ball milled for 2 h, [MgB2 þ 4.88% Fe], [MgB2 þ 4.97% Pd] and [MgB2 þ 4.46% TiF3]. The nominal H2 and D2 pressures were ~6 Torr each. The sample temperature is 200  C for all samples; (b) same as (a) only with a linear ordinate scale. The masses of the samples used are given in the SI.

sites, as well as HeH bond breaking at reactive MgB2 defect sites produced by ball milling. The kinetics of HD production increase with temperature, as shown by the H-D exchange results in Fig. 6 for [MgB2 þ 4.46% TiF3]. The SI presents an extensive discussion how this T-dependent H-D exchange data involves a generally low (~0.1 monolayer) coverage of adsorbed H or D atoms. As a result, via the Langmuir-Hinshelwood mechanism, the H and D atoms have to first diffuse on the surface to find each other before desorbing as HD. Thus, the H-D exchange measurements also show that H atom surface diffusion is occurring on a timescale much faster than the MgB2 hydrogenation of Fig. 4. Consequently, we can conclude that H atoms surface diffusion is also not rate-limiting for MgB2 hydrogenation. Our Arrhenius kinetic analysis of the HD exchange experiments for [MgB2 þ 4.46% TiF3] at different temperatures reveals the associated activation energy is within the range of

0 1

10

100 1000 Time (minutes)

10000

Fig. 6 e The reaction quotient Q(HD) ¼ [HD]2/([H2]x[D2]) plotted versus time for H-D exchange measurements for 11 mg of [MgB2 þ 4.46% TiF3]. (a) results plotted on linear ordinate and abscissa scales at 100  C, 200  C and 300  C for shorter times; (b) results plotted on a logarithmic time scale for longer times. The lines connect the experimental data points. The nominal H2 and D2 pressures are ~6 Torr each.

0.15e0.3 eV (see Fig. S35 in the SI). Notably, this value is lowdsimilar to the predicted H diffusion barrier on a clean MgB2 surface [51], which is consistent with expectations from a Langmuir-Hinshelwood HD exchange mechanism. Our prior work [17] on the mechanism of the initial MgB2 hydrogenation determined that the activation barrier during hydrogenation varies but never falls below ~0.5 eVdmuch larger than our observed HD exchange barrier. This supports the notion that H atom surface diffusion does not control the MgB2 hydrogenation kinetics. The 0.5 eV barrier was associated with an intermediate “diffusive adsorption” step in which adsorbed hydrogen atoms not only diffuse along the surfaces to reactive sites, but also react with B. With the observed surface diffusion barrier being only 0.15e0.3 eV, our present analysis indicates that the 0.5 eV barrier must be a HeB reaction barrier.

Please cite this article as: Liu Y-S et al., Investigating possible kinetic limitations to MgB2 hydrogenation, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.09.125

international journal of hydrogen energy xxx (xxxx) xxx

A full discussion of the nature of the H atom surface diffusion and the mechanism of H-D exchange is given in the SI. Having determined that neither H2 dissociation nor H atom surface diffusion limits the rate of MgB2 hydrogenation, we can consider alternative possibilities. One possibility is that it is slow bulk diffusion of H in MgB2 that limits hydrogenation. Another possibility is that it is the intrinsic stability of the BeB extended hexagonal ring structure in MgB2 that hinders the hydrogenation of this material. If this latter case is true, then bulk MgB2, ball-milled MgB2, [MgB2 þ 4.88% Fe], [MgB2 þ 4.97% Pd] and [MgB2 þ 4.46% TiF3] are not rapidly hydrogenating under our test conditions because their BeB ring structure persists, even upon ball milling and with the presence of additives. A test of this supposition is to directly observe the BeB ring structure in these materials and see if they are intact. Fig. 7(a) and (b) and presents B K-edge XAS spectra collected in the TFY and TEY modes of detection, respectively, for the materials investigated in this study. The depth sensitivities of the B K-edge TFY and TEY are estimated to be ~136 nm and ~4 nm, respectively (see Table S6 in the SI). The small XAS feature in Fig. 7(a) at ~ 187 eV photon energy corresponds to photon absorption from the B K shell to the unoccupied B 2pxy state just above the Fermi level. This feature is known to originate [17] from the extended BeB ring structure in MgB2 and does not appear in materials where the BeB ring is absent, for example in B2O3. Fig. 7(a) shows that for the MgB2 sample ball-milled for 2 h, the BeB ring is still intact, although the unoccupied 2pxy feature is diminished slightly compared to that observed from the bulk commercial material. In addition, the three additive samples [MgB2 þ 4.46% TiF3], [MgB2 þ 4.88% Fe] and [MgB2 þ 4.97% Pd] also show a prominent B 2pxy absorption peak, revealing the BeB ring feature is still intact for the bulk of these samples as well. Overall, the TFY XAS data support our supposition that MgB2 materials with the BeB ring intact are slow to hydrogenate.

9

Fig. 7(b) shows that on the relatively shallow ~ 4 nm depth scale of the B K-edge TEY measurement, the B 2pxy feature is absent in these spectra. For the Fe and Pd additives, this indicates that near the surface the BeB ring is disrupted, perhaps by the existence of a defected near-surface region, or by surface and near-surface oxidation of MgB2, or both. However, it is interesting to note that the TEY spectra for the [MgB2 þ 4.88% Fe] and [MgB2 þ 4.97% Pd] samples are not identical to B2O3. For example, these TEY spectra show spectral intensity just below the intense p* oxide resonance at 194 eV, spectral intensity that is absent in the B2O3 spectrum. This suggests that it is likely a combination of both oxidation and disruption of the normal BeB ring network at the surface conspire to disrupt the B 2pxy feature in the B Kedge TEY spectra of Fig. 7(b). If our supposition is true that BeB ring disruption is required for faster hydrogenation, then it is at these surface and near-surface sites where the initial hydrogenation of MgB2 might begin. For the [MgB2 þ 4.46% TiF3] sample, the B K-edge TEY XAS data has a large contribution from the ~3 nm-thick [Ti metal/MgF2/B] product phase, which does not possess the B 2pxy feature unique to MgB2.

Theoretical considerations for the reaction of TiF3 with MgB2 The characterization of the reaction products for [MgB2 þ 4.46% TiF3] is presented in the Appendix and in the SI. The exact chemical composition of the [MgB2 þ 4.46% TiF3] material is of secondary interest, since what matters most is that the sample presents an additive/MgB2 material for which the hydrogenation occurs slowly despite the very rapid HeH bond breaking and surface diffusion occurring on the sample. These are the main results of our study: that the slow hydrogenation rate of these additive/MgB2 systems is not limited by either the rate of HeH bond breaking, or the rate of H atom surface diffusion.

Fig. 7 e B K-edge x-ray absorption data collected in the (a) TFY and (b) TEY modes for commercial (bulk) MgB2, MgB2 ballmilled for 2 h, [MgB2 þ 4.46% TiF3], [MgB2 þ 4.88% Fe] and [MgB2 þ 4.97% Pd]. XAS spectra for B2O3 standard powder are also shown for comparison. Please cite this article as: Liu Y-S et al., Investigating possible kinetic limitations to MgB2 hydrogenation, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.09.125

10

international journal of hydrogen energy xxx (xxxx) xxx

Still, it is curious that the reaction occurring between TiF3 and MgB2 corresponds to TiF3 þ 3/2MgB2 / Ti þ 3B þ 3/2MgF2, where Ti metal is formed, as opposed to the more thermodynamically favored reaction TiF3 þ 3/2 MgB2 / TiB2 þ 3/2 MgF2 þ B, where TiB2 is formed. The product Ti metal would seem to be kinetically favored over TiB2, and that the thermodynamically stable product is inaccessible under our experimental conditions. More broadly, understanding when kinetics are favored over thermodynamics would help guide the use of additives in the promotion of hydrogen storage reactions. There have been no prior studies of the reaction of TiF3 with MgB2 in the inorganic chemistry literature. However, the reaction between TiCl3 and MgB2 has been investigated [52]. When TiCl3 and MgB2 are heated to 850  C for 18 h, TiB2 and MgCl2 are formed via the reaction TiCl3 þ 3/2MgB2 / TiB2 þ 3/2MgCl2 þ B. However, when the reaction temperature is lowered to 550  C, the less thermodynamically stable Ti metal is reported to exist along with TiB2 [52], consistent with Ti metal being kinetically favored in some circumstances. A comprehensive study of the reaction between TiF3 and MgB2 over wide ranges of temperature and composition is beyond the scope of this hydrogen storage study. However, to better understand the physical origin of the unexpected formation of metallic Ti in the [MgB2 þ 4.46% TiF3] material under our experimental conditions and to improve our understanding of how additives behave in hydrogen storage systems in general, we theoretically investigated the reaction between MgB2 and TiF3 in more detail. We first assessed the thermodynamic competition between the two possible reactions: Reaction 1: TiF3 þ 3=2 MgB2 /Ti þ 3=2 MgF2 þ 3B

(1)

Reaction 2: TiF3 þ 3=2 MgB2 /TiB2 þ 3=2 MgF2 þ B

(2)

The standard reaction enthalpies (DHrxn ) are 113 kJ/mol and 392 kJ/mol for Reactions 1 and 2, respectively, based on the NIST-JANAF thermochemical database [44]. Reaction 2 is therefore far more exothermic and would be expected to be favored over Reaction 1 at any reasonable temperature, even after accounting for entropic contributions. For example, the associated configurational entropy changes (DSrxn ) for Reactions 1 and 2 are calculated to be 31:5 J=ðK ,moleÞ and 17:4 J= ðK ,moleÞ, respectively (see the SI for the calculation details)da difference that is far too low to overcome the enthalpic preference. This confirms the preference for Reaction 1 may have a kinetic origin. Given that Reactions 1 and 2 are solid-state reactions, there are two possible kinetic limitations that could account for the observationsdnamely, diffusion and nucleation. Examining diffusion first, we note that Reaction 2 requires more species to diffuse than does Reaction 1. In both reactions, F atoms and/or Mg atoms must migrate across an interface between MgB2 and TiF3 to form MgF2. However, in the case of Reaction 2, Ti and/or B atoms must also migrate across this interface in order to form TiB2. This need for additional diffusion events likely incurs an associated kinetic penalty, favoring Reaction 1 in agreement with observations. Next, we consider the phase nucleation kinetics associated with the two reactions. Within the classical nucleation theory [45], the critical nucleation barrier (DG* ) of product phases can

be derived as 4B3 =ð27A2 Þ, where A represents the driving force for nucleation, and B represents the energy penalty term associated with creating new interfaces during the reactions. The thermodynamic driving force A was derived from the reaction free energy (DHrxn  T,DSrxn ) as discussed above. The interfacial energy penalty term B was estimated by following an approach demonstrated in our previous work [53], in which surface energy contributions (s) of the constituent phases are weighted within an assumed generic microstructure model (Fig. S2 in the SI). Within this formalism, a weighting parameter (g) acts as the single unknown in the calculation of each interfacial energy term. We systematically varied the parameter g for the distinct phase boundaries (gTiF3 =TiB2 and gTiB2 =½MgF2 ;B ) in order to determine whether conditions exist for which nucleation of Reaction 1 becomes kinetically preferred (see SI for further details) by virtue of a reduced critical nucleation barrier. The temperaturedependent nucleation barriers (DG* ) for product phases for Reactions 1 and 2, as derived using different assumed values of g that lie within physically reasonable ranges, are shown in Fig. 8. Here, we have plotted our results in terms of the ratio R ¼ DG*1 =DG*2 , where R < 1 indicates Reaction 1 is favored whereas R > 1 indicates that Reaction 2 is preferred. Our results confirm that there exist regimes (blue shaded regions in Fig. 8(a) and (b)) for which Reaction 1 becomes kinetically preferred over Reaction 2 due to reduced nucleation free energy barriers. Although the exact interfacial energies cannot be determined for our [MgB2 þ 4.46% TiF3] system, this analysis proves that the thermodynamic advantage for Reaction 2 can be superseded under physically reasonable ranges of g (g  0:35 for the TiF3 =TiB2 interface or g  0:42 for the TiB2 =½MgF2 ; B interface, based on the analysis in Fig. 8). Summarizing, we propose that the observed preference for Reaction 1 has a kinetic origin. That origin can be either 1) the prohibitive need for more diffusion events for Reaction 2 which make this reaction unfavorable or 2) reduced critical nucleation barriers associated with Reaction 1, which leads to its kinetic preference.

Conclusions We have reported an investigation of possible kinetic limitations to the rate of hydrogenation of MgB2, both for ball-milled MgB2 and for MgB2 ball-milled together with additives. The metals Pd, Fe and Ti, known to activate HeH bond dissociation, were introduced into MgB2 by ball milling. Pd and Fe were directly introduced as crystalline metals and decorate the MgB2 particles as discrete metallic nanoparticles. In contrast Ti metal was introduced via the reaction between TiF3 and MgB2 to form Ti metal, elemental B and MgF2. The formation of Ti metal in the MgB2/TiF3 reaction as opposed to the more thermodynamically stable TiB2 was explained by theory which suggests there exist physically reasonable regimes of nucleation interfacial energy where Ti metal formation is preferred over TiB2 formation, and that these regimes are present for the [MgB2 þ 4.46% TiF3] system. Alternatively, the kinetic preference could be rooted in the minimization of product diffusion barriers. TEM and XPS data show that the Ti metal and MgF2 are smeared out on the MgB2 particles as a film but remain proximate to each other.

Please cite this article as: Liu Y-S et al., Investigating possible kinetic limitations to MgB2 hydrogenation, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.09.125

international journal of hydrogen energy xxx (xxxx) xxx

Fig. 8 e Computed ratio (DG*1 =DG*2 ) of nucleation barriers for the products of Reaction 1 versus those of Reaction 2 as a function of temperature for different values of the interfacial energy parameter (g). Results from varying g for the TiF3 =TiB2 interface and the TiB2 =½MgF2 ; B interface independently are shown in (a) and (b), respectively; in both cases, all other interfacial energies are held at g ¼ 0:1. Sieverts-acquired MgB2 hydrogenation rates for ball-milled MgB2, [MgB2 þ 4.88% Fe], [MgB2 þ 4.97% Pd] and [MgB2 þ 4.46% TiF3] are higher than for commercial MgB2, but the improvements are very modest, about a factor of two at most. The Sieverts data, combined with the extensive characterization data, show that the metals can promote HeH bond breaking in the material, but that promotion does not dramatically improve the MgB2 hydrogenation rate. Consequently, HeH bond breaking is not rate limiting for the MgB2 hydrogenation. H-D exchange studies of the samples confirm that HeH bond breaking occurs on these materials and happens on a much shorter timescale than the Sieverts hydrogenation. This finding directly proves the HeH bond dissociation is not

11

limiting the rate of hydrogenation of MgB2. Analyses of the Tdependent H-D exchange experiments additionally show that H atom surface diffusion cannot be limiting the MgB2 hydrogenation rate because H atom surface diffusion, a mechanistic step for HD generation at the low surface coverages prevailing in the experiments, occurs very rapidly. It may be that the intrinsic stability of the BeB extended hexagonal ring structure in MgB2 hinders the hydrogenation kinetics. This supposition was supported by B K-edge TFY XAS measurements of the materials, which showed these slow-to-hydrogenate materials have intact BeB ring systems in the bulk material. The results indicate that additives chosen to promote HeH bond breaking will be ineffective in accelerating MgB2 hydrogenation. Although this has been shown here to be true for mesoscale MgB2 particles of ~300 nm diameter, we suspect it will likely be true for nanoscale MgB2 particles as well. Alternatively, it may be that additives that can disrupt the BeB ring network in MgB2, without penalizing the gravimetric capacity too much, will have greater chance of improving the hydrogenation kinetics of MgB2. An assessment of this approach is in progress. It is also possible that other limitations exist that inhibit the hydrogenation of MgB2 to Mg(BH4)2. Vajo et al. have shown [54] that electrolytes, in the form of LiBH4/KBH4 and LiI/KI/CsI eutectics, can significantly improve the hydrogenation of MgB2 to Mg(BH4)2. Although the mechanism of the improvement has not been investigated in detail, it is speculated that the electrolyte improves the dissolution of product Mg(BH4)2 as it is formed, exposing fresh MgB2 surface for reaction. The enhancement in MgB2 hydrogenation kinetics supports the conclusion presented here, that it is not the HeH bond breaking or surface diffusion of H atoms that limits the MgB2 hydrogenation kinetics. Rather, some other kinetic limitation is operative. Recently, Sugai et al. have reported that ball-milling MgB2 with THF, MgH2 and/or Mg results in a lowering of the pressure and temperature required for hydrogenation to Mg(BH4)2 [16]. The hydrogenation enhancement was attributed to the creation of defects in the MgB2 structure, which render it more susceptible to hydrogenation. As was the case with the work of Vajo and coworkers [54], this very recent work indicates that it is not the HeH bond breaking or surface diffusion of H atoms that limits MgB2 hydrogenation. Rather, it is some other intrinsic kinetic limitation associated with the MgB2 structure, as postulated here.

Acknowledgements The authors acknowledge financial support through the Hydrogen Storage Materials Advanced Research Consortium (HyMARC) 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-AC0494AL85000. Part of the work was performed under the auspices of the DOE by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344. Sandia National Laboratories is a multi-mission 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

Please cite this article as: Liu Y-S et al., Investigating possible kinetic limitations to MgB2 hydrogenation, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.09.125

12

international journal of hydrogen energy xxx (xxxx) xxx

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 DEAC02e05CH11231. Portions of the research were also performed at the Canadian Light Source (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 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. XPS work at the Molecular Foundry of the Lawrence Berkeley National Laboratory was supported by the Office of Science, Office of Basic Energy Sciences, of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231. Thanks are extended to David Prendergast of Lawrence Berkeley National Lab for helpful conversations, to U.S. Borax, Inc. for providing standard samples of K2B4O7$4H2O and NaBO2$2H2O and to Samantha Lawrence of Los Alamos National Laboratory for ongoing support. We thank the reviewers of the manuscript for very helpful and constructive feedback. 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.

Supplementary data Supplementary data to this article can be found online at https://doi.org/10.1016/j.ijhydene.2019.09.125.

Appendix Determining the Chemical Composition of the [MgB2 þ 4.46% TiF3] Sample This Appendix presents experimental data used to positively identify the reaction products formed when MgB2 was ballmilled together with TiF3 to make the [MgB2 þ 4.46% TiF3] sample. Fig. A1 presents FTIR data showing that the vibrational spectral signature for TiF3 becomes undetectable in the [MgB2 þ 4.46% TiF3] sample. This disappearance of TiF3 in [MgB2 þ 4.46% TiF3] is confirmed by XRD spectra in Fig. S24 of the SI. Such a disappearance could have a spurious cause, for example if the TiF3 material decomposes upon ball-milling, or if the TiF3 material pathologically oxidized to TiO2 in the ballmilling with MgB2, eliminating the TieF stretching vibration shown in Fig. A1. These spurious causes can be eliminated. Fig. S20 of the SI presents XRD data for TiF3 that shows it is stable to ball-milling and does not degrade. In addition,

Fig. S22 of the SI confirms that crystalline TiO2 is not produced during the preparation of [MgB2 þ 4.46% TiF3]. Rather, a reasonable conclusion from the disappearance of TiF3 in Fig. A1 is that it reacts with MgB2 to form reaction products. One possible reaction product is MgF2. Initial efforts to identify MgF2 in the [MgB2 þ 4.46% TiF3] sample were unsuccessful, although these efforts ended up providing useful information nonetheless. Fig. S23(b) in the SI compares the FTIR spectra of MgF2 standard powder with [MgB2 þ 4.46% TiF3]. No evidence is found for amorphous MgF2. However, the FTIR sensitivity to MgF2 is poor because MgF2 possesses a much weaker vibrational signature than TiF3. We estimate that the entire 4.46% mole fraction of TiF3 could have converted to MgF2 when ball-milled with MgB2 and it would be difficult to detect it

Fig. A1 e (a) FTIR spectra of the starting TiF3 powder and the [MgB2 þ 4.46% TiF3] material, emphasizing the region of TieF stretch. by FTIR. Thus, the MgF2 FTIR signal is not a useful probe of the presence of MgF2 in the [MgB2 þ 4.46% TiF3] sample. Fig. S25 in the SI presents XRD data for the [MgB2 þ 4.46% TiF3] material and a standard powder sample of MgF2. The XRD peaks for crystalline MgF2 are sharp and intense. However, there is no evidence from Fig. S25 for the formation of crystalline MgF2 in [MgB2 þ 4.46% TiF3]. Consequently, if MgF2 was formed in the [MgB2 þ 4.46% TiF3], it must be amorphous. To positively identify the reaction products, XPS and XAS measurements were performed for [MgB2 þ 4.46% TiF3] sample, the results of which are presented in Figs. A2-A7. We would like to be able to distinguish between two possible reaction routes: Reaction 1: TiF3 þ 3=2 MgB2 /Ti þ 3=2 MgF2 þ 3B

(A1)

Reaction 2: TiF3 þ 3=2 MgB2 /TiB2 þ 3=2 MgF2 þ B

(A2)

Both MgF2 and elemental B are produced by these chemical reactions, and their detection would confirm that a chemical reaction is indeed taking place between TiF3 and MgB2. Beyond basic reaction, it is the fate of Ti that allows Reactions A1 and A2 to be distinguished from each other. The appearance of Ti

Please cite this article as: Liu Y-S et al., Investigating possible kinetic limitations to MgB2 hydrogenation, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.09.125

international journal of hydrogen energy xxx (xxxx) xxx

metal would signal Reaction A1 is occurring; the appearance of TiB2 would signal that Reaction 2 is occurring. The high-resolution core-level XPS data in Fig. A2 investigate if the chemical state of Ti in the [MgB2 þ 4.46% TiF3] sample is Ti metal. In Fig. A2(a) we show the preparation of a Ti metal standard from the sputtering of a commercial Ti metal foil. Before the argon-ion etching of the foil, the Ti 2p3/2 and Ti 2p1/2 XPS features contain contributions from both Tioxide (TiO2) and Ti metal. It is important to note that the oxide film on this Ti metal foil is self-limiting. As a result, Ti metal still appears in the XPS spectrum even for this untreated Ti metal foil that had been exposed to air for years. As argonion etching proceeds in Fig. A2(a), the oxide features are readily removed, leaving pure Ti metal after only 160 s. The spectrum that remains after argon-ion etching can be used as a Ti metal standard for comparision to the Ti 2p XPS spectrum of [MgB2 þ 4.46% TiF3].

13

To be sure that a pure Ti metal XPS spectrum was obtained, we continued the argon-ion etching of the Ti foil surface until a 480 s etch had been performed, thus providing a clean Ti metal standard. The Ti 2p XPS spectrum for this clean Ti metal surface is compared to [MgB2 þ 4.46% TiF3] in Fig. A2(b). For [MgB2 þ 4.46% TiF3] there is a clear Ti 2p3/2 peak that matches the peak position for Ti metal standard. Fig. A3 shows an expanded view of the same data, demonstrating the excellent peak correspondence, indicating the presence of Ti metal product in the [MgB2 þ 4.46% TiF3] sample, consistent with Reaction A1.

Fig. A3 e High-resolution Ti 2p3/2 XPS spectra of a Ti metal foil argon-etched for 480 s (to produce a clean Ti metal surface) and of the [MgB2 þ 4.46% TiF3] sample. This plot is an expanded view of the same data shown in Fig. A2(b), emphasizing the Ti 2p3/2 peak.

Fig. A2 e (a) High-resolution Ti 2p XPS spectra of a Ti metal foil. Spectra are shown before argon-ion etching, and after 40 s and 160 s of etching; (b) High-resolution Ti 2p XPS spectra of a Ti metal foil argon-etched for 480 s (to produce a clean Ti metal surface) and from the [MgB2 þ 4.46% TiF3] sample.

Some of the Ti metal in the [MgB2 þ 4.46% TiF3] sample has been oxidized, as indicated by the simultaneous presence of the oxide peaks at ~ 459 eV and ~465 eV in Fig. A2(b). But elemental Ti also remains, which is an indication that the Ti metal produced by Reaction A1 is in the form of a film. If the Ti metal had been atomically dispersed, all of the Ti would be subject to oxidation, and no Ti metal would remain. However, in the form of a film, the Ti metal oxidized very much like the Ti foil in Fig. A2(a), eventually forming a self-limited oxide that protects the Ti metal underneath. Chu and co-workers [A1] have reported the XPS from TiB2 in magnetron sputtered films. They report data that is consistent with the Ti 2p3/2 peak for TiB2 being shifted ~0.6 eV to higher binding energy than that observed for Ti metal (see Fig. 1(a) and (b) of Ref. A1). They attributed this shift to a charge transfer from Ti to B in TiB2, with the resulting Ti atom more positive and B more negative, which is a physically reasonable attribution. Such a “chemical shift” is typical of XPS spectra, where the shift of a photoemitting atom can be related to the valence charge at the photoemitting site [A2]. The more positive in character a metal site becomes, the harder it is to pull a core-level photoelectron away from it, and a shift to higher binding energy results. If TiB2 were being formed in the [MgB2 þ 4.46% TiF3] sample, as suggested by

Please cite this article as: Liu Y-S et al., Investigating possible kinetic limitations to MgB2 hydrogenation, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.09.125

14

international journal of hydrogen energy xxx (xxxx) xxx

Reaction 2, its Ti 2p3/2 peak would appear shifted to higher binding energy from that of Ti metal by ~ 0.6 eV. Such a shift would be readily detected but is not observed in the XPS data of Figs. A2 and A3. The high-resolution core-level XPS data in Fig. A4 investigate the chemical state of Mg in the [MgB2 þ 4.46% TiF3] sample. Fig. A4(a) shows XPS data for the Mg 1s level in which the spectrum acquired for [MgB2 þ 4.46% TiF3] is compared with experimental spectra acquired during the same experimental run for crystalline MgF2 standard, and for MgB2 ball-milled for 1 h, which mimics the preparation of [MgB2 þ 4.46% TiF3] only with no TiF3 added. One can see the striking result that the Mg 1s spectrum of [MgB2 þ 4.46% TiF3] occurs at the same position as that observed from MgF2, confirming the formation of MgF2, consistent with Reaction A1. Indeed, there is no sign of the Mg 1s spectrum from unreacted MgB2 in the [MgB2 þ 4.46% TiF3] spectrum. The core-level shift between MgB2 and MgF2 reflects the higher positive charge on the Mg atom for MgF2 compared to MgB2 (reflecting the higher electron affinity of F compared to B), pushing the core-level photoemission to higher binding energy.

Fig. A4 e (a) High-resolution Mg 1s XPS spectra for [MgB2 þ 4.46% TiF3] compared to experimental Mg 1s spectra obtained for crystalline MgF2 standard and MgB2 ball milled for 1 h with no additives; (b) Same as panel (a) only for Mg 2p XPS.

Fig. A4(b) shows XPS data for the Mg 2p level in which the spectrum acquired for [MgB2 þ 4.46 % TiF3] is compared with experimental Mg 2p spectra acquired during the same experimental run for crystalline MgF2 standard, and for MgB2 ball-milled for 1 hour as a control. Again, one sees evidence for a Mg 2p component for [MgB2 þ 4.46 % TiF3] at the same binding energy as that observed from MgF2, confirming the presence of MgF2 in the sample. In addition, for the Mg 2p level, one also sees a component corresponding to MgB2 unreacted with TiF3. The appearance of only MgF2 in Fig. A4(a) for Mg 1s photoemission from [MgB2 þ 4.46% TiF3], and the presence of both MgB2 and MgF2 in Fig. A4(b) for Mg 2p photoemission from [MgB2 þ 4.46% TiF3] are consequences of the much greater surface sensitivity of the Mg 1s spectra compared to the Mg 2p spectra. Table S7 in the SI presents estimates for the depth sensitivity for the XPS spectra, where we define the depth sensitivity as including depths that contribute greater than or equal to 10% of the photoemission signal associated with the surface layer. The depth sensitivity for the Mg 2p peak (with kinetic energy of 1437 eV) is 8.4 nm, compared to a depth sensitivity of the Mg 1s peak (with kinetic energy of 182 eV) of only 3.0 nm. As a result, the appearance of a sole MgF2 spectral component in the Mg 1s XPS spectrum of [MgB2 þ 4.46% TiF3] means that the Ti metal and MgF2 phases exist as a film, and that film is at least ~ 3 nm thick. The appearance in Fig. S4(b) of both MgB2 and MgF2 components in the Mg 2p data indicates that the Ti/MgF2 product film is not so thick as to fully suppress the Mg 2p photoemission from MgB2 which has a larger depth sensitivity. Recall that Ti metal was identified by the XPS results of Fig. A2(b), and a Ti film was suggested by the TEM data for [MgB2 þ 4.46% TiF3] shown in Fig. 3 as well as by the TEM data in Fig. S31 of the SI. The inferred thickness of 3 nm is interesting. The TEM data of Fig. 3 indicate that the size of the MgB2 particles in the [MgB2 þ 4.46% TiF3] sample is ~300 nm in diameter. It seems physically reasonable to assume that these MgB2 particle would react with TiF3 from their perimeter inwards. One can calculate for a 300 nm diameter MgB2 particle what the thickness is of the reacted layer if MgB2 reacts with 4.46 mol percent of TiF3, considering that in Reaction A1, for every mole of TiF3 that reacts, 1.5 mol of MgB2 are consumed. The calculated thickness for the reacted layer is 3.5 nm, in good agreement with the inference of ~3 nm made from the Mg 1s XPS data. This confirms that the underlying unreacted MgB2 Mg 1s photoemission is not observed, because it is buried under ~3 nm of product layer, leaving only the Mg 1s photoemission from the overlying MgF2 layer to escape the solid and be detected. If we are observing photoemission exclusively from the Ti metal/MgF2 product layer, with little or no contribution from MgB2 below, one can compare the relative intensities of the Mg 1s and F 1s XPS data to assess the atomic ratios. For the [MgB2 þ 4.46% TiF3] sample, that atomic ratio is: Mg:F ¼ 1:1.4. Thus, the stoichiometry is close to that expected (1:2) for MgF2. The ability to extract atomic ratios from XPS data is semiquantitative. Our XPS analysis of the commercial MgB2 sample that had been ball-milled for 1 h gave, using the relative intensities of the Mg 1s and B 1s peaks, a Mg:B ratio of 1:1.7 where 1:2 should have been obtained.

Please cite this article as: Liu Y-S et al., Investigating possible kinetic limitations to MgB2 hydrogenation, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.09.125

international journal of hydrogen energy xxx (xxxx) xxx

We sought with XPS to detect the formation of elemental B via reactions 1 or 2. The detection of elemental B would not discriminate between the two reactions, but rather would confirm our general observations that a reaction takes place between MgB2 and 4.46% TiF3. The B 1s XPS results are shown in Figure A5.

Fig. A5 e High-resolution B 1s XPS spectra for [MgB2 þ 4.46% TiF3] compared to experimental B 1s spectra obtained for a commercial MgB2 standard ball milled for 1 h. Comparison is also made to B2O3 standard powder. The photoelectron count rates have been normalized at the spectral peak maximum for the [MgB2 þ 4.46% TiF3] and MgB2 BM 1-h samples.

15

Fig. A5 shows that [MgB2 þ 4.46% TiF3] sample has two prominent B 1s peaks of comparable intensity at 187.08 eV and 187.80 eV. The 187.08 eV peak corresponds well to the B 1s peak for MgB2 ball-milled for 1 h. The peak at 187.80 eV, shifted 0.72 eV to higher binding energy than the MgB2 B 1s peak, corresponds well to the B 1s binding energy of 187.9 eV observed for elemental B by Foo et al. [A3] and Ong and coworkers [A4]. Note that this peak is not observed when MgB2 is ball milled without TiF3, so the peak cannot be attributed to MgB2 degradation produced by ball-milling. Also, it is unlikely to be due to oxide due to the large core-level shift of 5.9 eV produced by full oxidation to B2O3, as shown in Fig. A5. This B 1s XPS core-level is shifted to higher binding energy for elemental B compared to that for B in MgB2 because of the charge difference between negatively charged boron atoms in MgB2 and neutral elemental boron. It is harder to pull an electron away from a neutral B atom than a negatively charged B atom in MgB2, leading to a somewhat higher binding energy for the elemental B film. The presence of both elemental B and MgB2 B 1s XPS peaks in Fig. A5 is consistent with the reaction film being ~3 nm thick. The depth sensitivity of the B1s XPS peaks is estimated to be 8.0 nm (see Table S7 in the SI). Thus, the B 1s XPS peak samples the entirety of the ~ 3-nm thick reacted film layer as well the remaining ~5 nm of the unreacted MgB2 below. As a result, the presence of both peaks is expected. There also appears to be a minority component in the MgB2 BM 1-h material with a low binding energy of about 185.88 eV. This peak is greatly reduced in the [MgB2 þ 4.46% TiF3] material. The origin of this peak is unknown. Summarizing the XPS data, high-resolution core-level data for the Ti 2p indicates that Ti metal exists in the [MgB2 þ 4.46% TiF3] sample. High-resolution spectra of the Mg 1s and Mg 2p level both show the presence of MgF2. High-resolution spectra of the B 1s level indicate the formation of elemental boron.

Fig. A6 e (a) F K-edge TFY and (b) F K-edge TEY spectra for TiF3 powder, MgF2 standard powder and [MgB2 þ 4.46% TiF3].

Please cite this article as: Liu Y-S et al., Investigating possible kinetic limitations to MgB2 hydrogenation, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.09.125

16

international journal of hydrogen energy xxx (xxxx) xxx

The film thickness appears to be ~3 nm. Taken together, the results argue persuasively that Reaction 1 accounts for the reaction of MgB2 with TiF3 in the [MgB2 þ 4.46% TiF3] sample. An independent method of assessing composition is by conducting XAS measurements of the [MgB2 þ 4.46% TiF3] sample using synchrotron radiation. Fig. A6 shows F K-edge XAS measurements of the [MgB2 þ 4.46% TiF3] sample and compares them to standard powder sample of crystalline MgF2 and crystalline TiF3. Fig. A6(a) presents XAS data collected in TFY, whereas Fig. A6(b) presents data collected in the more surface sensitive TEY mode of collection. The depth sensitivities of the F K-edge TFY and TEY spectra are estimated to be ~1500 nm and ~20 nm, respectively (see Table S6 in the SI). Thus, both TFY and TEY F K-edge should be sampling the entire Ti metal/MgF2 product film thickness of ~3 nm.

XAS spectrum reflects the unoccupied band structure at the site of the core-hole, which is determined by a local ensemble of likely ~20 atoms which can be influenced by the amorphous nature of the solid. This would explain why the XAS spectrum at the F K-edge for [MgB2 þ 4.46% TiF3] in Fig. A6(a) is smoother than the corresponding F K-edge XAS TFY spectrum from the highly crystalline MgF2 standard powder. Fig. A7 shows Ti L2,3-edge TFY and TEY spectra for the [MgB2 þ 4.46% TiF3] sample, compared to several Ti standards including TiB2 and an enthusiastically scraped Ti metal foil. The depth sensitivities for the Ti L2,3-edge TFY and TEY spectra are estimated to be ~500 nm and ~10 nm, respectively as discussed in the SI. Thus, both types of XAS data are sampling the entire 3 nm of the Ti metal/MgF2 product layer. The Ti L2,3-edge TFY and TEY data for the [MgB2 þ 4.46% TiF3] sample are in good agreement with each other, as are the TFY and TEY data for Ti metal as well as for the other standard Ti samples.

Fig. A7eTi L2,3-edge XAS spectra recorded via the (a) TFY and (b) TEY detection modes for TiF3, Ti metal scraped clean by surface abrasion in the glove-box immediately prior to measurement, TiO2 standard powder, TiB2 standard powder and [MgB2 þ 4.46% TiF3]. The F K-edge TFY and TEY yield data for the [MgB2 þ 4.46% TiF3] sample are in good agreement with each other, as are the TFY and TEY data for TiF3. This agreement indicates good depth uniformity of the F-containing phases in these samples. For the MgF2 crystalline standard, there are some variations between the MgF2 TFY and TEY data whose cause is unknown. Given the greater depth sensitivity of the TFY method, providing signal down to ~1500 nm of the MgF2 surface, we will choose the MgF2 TFY spectrum as the MgF2 XAS spectral standard, as it is representative of bulk MgF2. Fig. A6(a) shows that the F K-edge XAS data of [MgB2 þ 4.46% TiF3] does not resemble that acquired from TiF3, confirming reaction between TiF3 and MgB2. The F K-edge XAS spectrum of [MgB2 þ 4.46% TiF3] corresponds well with the XAS TFY spectrum of MgF2. There are some differences, which one would expect given that the standard MgF2 powder is crystalline, whereas the MgF2 in the [MgB2 þ 4.46% TiF3] material is amorphous, as shown by a lack of diffraction in Fig. S25 in the SI. The

Fig. A7 shows that Ti L2,3-edge XAS data of [MgB2 þ 4.46% TiF3] does not resemble that acquired from TiF3, again confirming reaction between TiF3 and MgB2. Furthermore, the spectrum for [MgB2 þ 4.46% TiF3] does not resemble TiO2, or TiB2, which supports the conclusions drawn from XRD, FTIR and XPS measurements that TiO2 does not exist in the sample from oxidation, and that TiB2 is not produced by the chemical reaction. Care was taken in confirming the high-purity of the TiB2 standard powder used in making this spectral assessment (see the SI). The Ti L2,3-edge XAS data from [MgB2 þ 4.46% TiF3] most closely resembles that of clean Ti metal. This finding independently confirms the XPS results that Reaction A1 accounts for the reaction of TiF3 with MgB2 in the [MgB2 þ 4.46% TiF3] sample. While the data presented in this Appendix definitively shows that the Reaction A1 has occurred in the preparation of the [MgB2 þ 4.46% TiF3] sample, this conclusion is confined to the experimental conditions our study. A comprehensive

Please cite this article as: Liu Y-S et al., Investigating possible kinetic limitations to MgB2 hydrogenation, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.09.125

international journal of hydrogen energy xxx (xxxx) xxx

study of the reaction of TiF3 and MgB2 over a wide range of temperatures and compositions would be an excellent contribution to the inorganic chemistry literature. In addition to the characterization techniques employed in this study, the use of 11B and 19F NMR would prove valuable in such a study.

[11]

Appendix References [12]

A1. K. Chu, Y.H. Lu and Y.G. Shen, “Structural and mechanical Properties of Titanium and titanium diboride Monolayers and Ti/TiB2 multilayers,” Thin Solid Films, 516 (2008) 5313-5317. A2. U. Gelius, “Binding energies and chemical shifts in ESCA,” Phys Scr 9 (1974) 133-147. A3. W.C. Foo, J.S. Ozcomert and M. Trenary, “The oxidation of the b-rhombohedral boron (111) surface,” Surf Sci 255 (1991) 245-258. A4. C. W. Ong, H. Huang, B. Zheng, R. W. M. Kwok, Y. Y. Hui and W. M. Lau, “X-ray photoemission Spectroscopy of nonmetallic materials: electronic Structures of Boron and BxOy” J Appl Phys 95 (2004) 3527 - 3534.

[13]

[14]

[15] [16]

references [17] [1] Klebanoff L, Keller J, Fronk M, Scott P. Hydrogen conversion Technologies and automotive applications. In: Klebanoff LE, editor. Chapter 2 in hydrogen storage technology, materials and applications. Boca Raton: Taylor & Francis; 2012. p. 31. [2] Hua T, Ahluwalia R, Eudy L, Singer G, Jermer B, AsselinMiller N, Wessel S, Patterson T, Marcinkoski J. Status of hydrogen fuel cell electric buses worldwide,. J Power Sources 2014;269:975e93. [3] Klebanoff LE, Breit JS, Roe GS, Damberger T, Erbel T, Wingert S, et al. Fuel cell mobile lighting: a fuel cell market transformation project. Int J of Hydrogen Energy 2014;39:12948e72. [4] Mayyas A, Wei M, Chan SH, Lipman T. Fuel cell forklift deployment in the USA. In: Stolten D, Samsun RC, Garland N, editors. Chapter 33 in “fuel cells: data, facts and Figures. Weinheim: Wiley; 2016. p. 334. [5] The first hydrogen fuel-cell rail system for commuter service has been introduced: https://phys.org/news/2018-09germany-world-hydrogen.html. [6] Klebanoff LE, Pratt JW, Leffers CM, Sonerholm KT, Escher T, Burgard J, Ghosh S. Comparison of the greenhouse gas and criteria pollutant emissions from the SF-BREEZE high-speed fuel-cell ferry with a diesel ferry. Transp Res Part D 2017;54:250e68. [7] The feasibility of a hydrogen-powered ocean-going coastal research vessel (the “Zero-V”) has recently been detemined. A full report can be downloaded at: https://energy.sandia. gov/transportation-energy/hydrogen/markettransformation/maritime-fuel-cells/. [8] The California Air Resources Board has recently funded the construction of the “Water-Go-Round” hydrogen fuel-cell ferry. Information about the vessel can be found at: https:// watergoround.com/. [9] The SAE standard J2719 provides a hydrogen fuel quality standard for commercial proton exchange membrane (PEM) fuel-cell vehicles. [10] The U.S. Department of Energy has established technical performance targets for hydrogen storage systems for onboard storage of hydrogen for light-duty fuel-cell vehicles.

[18]

[19]

[20] [21]

[22]

[23]

[24] [25] [26]

[27]

[28] [29]

17

These targets can be found at: https://www.energy.gov/eere/ fuelcells/doe-technical-targets-onboard-hydrogen-storagelight-duty-vehicles. Graetz J, Wolstenholme DJ, Pez GP, Klebanoff LE, McGrady GS, Cooper AC. Development of off-board reversible hydrogen storage materials. In: Klebanoff LE, editor. Chapter 8 in “hydrogen storage technology, Materials and applications”. Boca Raton: Taylor and Francis; 2012. p. 239. Zavorotynska O, El-Kharbachi A, Deledda S, Hauback BC. Recent Progress in magnesium borohydride Mg(BH4)2: Fundamentals and Applications for energy storage. Int J Hydrogen Energy 2016;41:14387e403. Paskevicius M, Jepsen LH, Schouwink P, Cerny R, Ravnsbaek DB, Filinchuk Y, Dornheim M, Besenbacher F, Jensen TR. Metal borohydrides and derivatives-synthesis, structure and properties. Chem Soc Rev 2017;46:1565e634. Barkhordarian G, Klassen T, Dornheim M, Bormann R. Unexpected kinetic Effect of MgB2 in reactive hydride composites containing complex borohydrides. J Alloy Comp 2007;440:L18e21. Vajo JJ, Skeith SL, Mertens F. Reversible Storage of Hydrogen in destabilized LiBH4. J Phys Chem B 2005;109:3719e22. Sugai C, Kim S, Severa G, White JL, Leick N, Martinez MB, Gennett T, Stavila V, Jensen C. Kinetic Enhancement of direct Hydrogenation of MgB2 to Mg(BH4)2 upon mechanical Milling with THF, MgH2 and/or Mg. Chem Phys Chem 2019;20:1301e4. Ray KG, Klebanoff LE, Lee JRI, Stavila V, Heo TW, Shea P, et al. Elucidating the Mechanism of MgB2 initial hydrogenation Via a combined experimental-theoretical study. Phys Chem Chem Phys 2017;19:22646e58. Pistidda C, Garroni S, Dolci F, Bardaji EG, Khandelwal A, Nolis P, Dornheim M, Gosalawit R, Jensen T, Cerenius Y, Surinach S, Baro MD, Lohstroh W, Fichtner M. Synthesis of amorphous Mg(BH4)2 from MgB2 and H2 at room temperature. J Alloy Comp 2010;508:212e5. Severa G, Ronnebro E, Jensen CM. Direct Hydrogenation of magnesium Boride to magnesium borohydride: Demonstration of > 11 weight percent reversible hydrogen storage. Chem Commun 2010;46:421e3. Behm RJ, Christmann K, Ertl G. Adsorption of hydrogen on Pd(100). Surf Sci 1980;99:320e40. Behm RJ, Penka V, Cattania G-G, Christmann K, Ertl G. “Evidence for “subsurface” hydrogen on Pd(110): an intermediate between chemisorbed and dissolved species. J Chem Phys 1983;78:7486e90. Engel T, Kuipers H. A molecular-beam Investigation of the scattering, Adsorption and Absorption of H2 and D2 from/on/ in Pd(111). Surf Sci 1979;90:162e80. Moritz W, Imbihl R, Behm RJ, Ertl G, Matsushima T. Adsorption geometry of hydrogen on Fe(110). J Chem Phys 1985;83:1959e68. Bozso F, Ertl G, Grunze M, Weiss M. Chemisorption of hydrogen on iron surfaces. Appl Surf Sci 1977;1:103e19. Cavalier JC, Chornet E. Hydrogen-Deuterium exchange on iron: kinetic anisotropies. Surf Sci 1976;60:125e46. Berlowitz PJ, Goodman DW. Chemisorption of ultrathin Pd Layers on W(110) and W(100): Adsorption of H2 and CO. Langmuir 1988;4:1091e5. Wulz HG, Fromm E. Hydrogen absorption rate of titanium, lanthanum, iron, nickel, manganese and palladium films with and without oxygen precoverage at 300K. J Less Common Met 1986;118:293e301. Nahm T-U, Gomer R. The adsorption of hydrogen on W(110) and Fe covered W(110) surfaces. Surf Sci 1997;375:281e92. Abanador PM, Villagracia ARC, Arboleda Jr NB, David MY. First principle Investigation of atomic hydrogen Adsorption on Pd-doped MgB2. Philippine Science Letters 2013;6:176e81.

Please cite this article as: Liu Y-S et al., Investigating possible kinetic limitations to MgB2 hydrogenation, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.09.125

18

international journal of hydrogen energy xxx (xxxx) xxx

[30] Saha D, Deng S. Hydrogen adsorption on ordered mesoporous carbons doped with Pd, Pt, Ni and Ru. Langmuir 2009;25:12550e60. [31] Pozzo M, Alfe D. Hydrogen dissociation and diffusion on transition metal (¼ Ti, Zr, V, Fe, Ru, Co, Rh, Ni, Pd, Cu, Ag)doped Mg(0001) surfaces. Int J Hydrogen Energy 2009;34:1922e30. [32] Kang S, Klebanoff LE, Baker AA, Cowgill DF, Stavila V, Lee JRI, Nielsen MH, Ray KG, Liu Y-S, Wood BC. Assessing the Reactivity of TiCl3 and TiF3 with hydrogen. Int J Hydrogen Energy 2018;43:14507e19. [33] Teeter G, Erskine JL. Surface relaxation of Ti(0001): influence of hydrogen contamination. Phys Rev B 2000;61:13929e35. [34] Arboleda Jr NB, Kasai H, Nobuhara K, Dino WA, Nakanishi H. Dissociation and Sticking of H2 on Mg(0001), Ti(0001) and La(0001) surfaces. J. Phys. Soc. Japan 2004;73:745e8. [35] Pozzo M, Alfe D, Amieiro A, French S, Pratt A. Hydrogen dissociation and diffusion on Ni- and Ti-doped Mg(0001) surfaces. J Chem Phys 2008;128:094703 1e11. [36] Clark A. Oxides of the transition metals as catalysts. Ind Eng Chem 1953;45:1476e80. [37] Dowden DA, Mackenzie N, Trapnell BMW. HydrogenDeuterium exchange on the oxides of transition metals. Adv Catal 1957;9:65e9. € pel W, Rocker G, Feierabend R. Intrinsic Defects of [38] Go TiO2(110): Interaction with chemisorbed O2, H2, CO and CO2. Phys Rev B 1983;28:3427e38. [39] Hagemann H, D’Anna V, Rapin J-P, Yvon K. Deuteriumhydrogen Exchange in solid Mg(BH4)2. J Phys Chem C 2010;114:10045e7. [40] Borgschulte A, Zu¨ttel A, Hug P, Barkhordarian G, Eigen N, Dornheim M, Bormann R, Ramirez-Cuesta AJ. Hydrogendeuterium exchange experiments to probe the decomposition reaction of sodium alanate. Phys Chem Chem Phys 2008;10:4045e55. [41] Lohstroh W, Fichtner M. Rate limiting Steps of the phase Transformations in Ti-doped NaAlH4 Investigated with isotope exchange. Phys Rev B 2007;75. 184106-184111-6. [42] Erbil A, Cargill III GS, Frahm R, Boehme RF. Total-electronyield current measurements for near-surface extended X-ray absorption fine-structure. Phys Rev B 1988;37:2450e65.

[43] 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;23:679e82. [44] 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. [45] M. Porter DA, Easterling KE, Sherif M. Phase Transformations in Metals and alloys. CRC press; 2009 (Revised Reprint) [46] Kresse G, Furthmu¨ller J. Efficient iterative schemes for abinitio total-energy calculations using a plane-wave basis set. Phys Rev B 1996;54:11169e86. [47] Unpublished results of S. Kang and B.C. Wood, Lawrence Livermore National Laboratory. [48] Pinatel ER, Albanese E, Civalleri B, Baricco M. Thermodynamic Modelling of Mg(BH4)2. J. Alloys and Cmpds. 2015;645:S64e8. [49] Atkins PW, de Paula J. Physical chemistry: thermodynamics, structure and change. 10th ed. New York: W.H. Freeman and Company; 2014. p. 956. [50] Luo WF, Cowgill DF, Causey RA. Thermodynamic and kinetic characterization of hydrogen-deuterium exchange in betaphase palladium. J Phys Chem B 2009;113:12978e87. [51] Wang Y, Michel K, Zhang Y, Wolverton C. Thermodynamic Stability of transition Metals on the Mg-terminated MgB2(0001) Surface and their Effects on hydrogen Dissociation and diffusion. Phys Rev B 2015;91:155431 1e11. [52] Rao L, Gillan EG, Kaner RB. Rapid synthesis of transtionmetal borides by solid-state metathesis. J Mater Res 1995;10:353e61. [53] Wood BC, Stavila V, Poonyayant N, Heo TW, Ray KG, Klebanoff LE, Udovic TJ, Lee JRI, Angboonpong N, Pakawatpanurut P. Nanointerface-driven reversible hydrogen storage in the nanoconfined Li-N-H system. Adv Mat Interfaces 2017;4:1600803 1e7. [54] Vajo JJ, Tan H, Ahn CC, Addison D, Hwang S-J, White JL, Wang TC, Stavila V, Graetz J. Electrolyte-assisted hydrogen storage reactions. J Phys Chem C 2018;122:26845e50.

Please cite this article as: Liu Y-S et al., Investigating possible kinetic limitations to MgB2 hydrogenation, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.09.125