Influence of nickel and silicon addition on the deuterium siting … · Influence of Nickel and Silicon Addition on the Deuterium Siting and Mobility in fcc Mg−Ti Hydride Studied

Influence of nickel and silicon addition on the deuterium sitingand mobility in fcc Mg-Ti hydride studied with 2H MAS NMR.Citation for published version (APA):Manivasagam, T. G., Magusin, P. C. M. M., Iliksu, M., & Notten, P. H. L. (2014). Influence of nickel and siliconaddition on the deuterium siting and mobility in fcc Mg-Ti hydride studied with 2H MAS NMR. Journal of PhysicalChemistry C, 118(20), 10606-10615. https://doi.org/10.1021/jp500535q

DOI:10.1021/jp500535q

Document status and date:Published: 01/01/2014

Document Version:Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)

Please check the document version of this publication:

• A submitted manuscript is the version of the article upon submission and before peer-review. There can beimportant differences between the submitted version and the official published version of record. Peopleinterested in the research are advised to contact the author for the final version of the publication, or visit theDOI to the publisher's website.• The final author version and the galley proof are versions of the publication after peer review.• The final published version features the final layout of the paper including the volume, issue and pagenumbers.Link to publication

General rightsCopyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright ownersand it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, pleasefollow below link for the End User Agreement:www.tue.nl/taverne

Take down policyIf you believe that this document breaches copyright please contact us at:[email protected] details and we will investigate your claim.

Download date: 04. Sep. 2021

Influence of Nickel and Silicon Addition on the Deuterium Siting andMobility in fcc Mg−Ti Hydride Studied with 2H MAS NMRThirugnasambandam G. Manivasagam,† Pieter C. M. M. Magusin,*,‡ Merve Iliksu,†,§

and Peter H. L. Notten*,†

†Department of Chemical Engineering and Chemistry, Eindhoven University of Technology, P.O. Box 513, 5600 MB Eindhoven, TheNetherlands‡Centre for Surface Chemistry and Catalysis, KU Leuven, Kasteelpark Arenberg 23, 3001 Leuven, Belgium

*S Supporting Information

ABSTRACT: Fluorite-structured Mg−Ti hydrides are interesting forhydrogen storage applications because of their high gravimetric hydrogenstorage capacity, and improved (de)hydrogenation kinetics compared toMgH2. In the present study we have investigated the potential catalytic effectof Ni and Si as third element on the siting and mobility of electrochemicallyloaded deuterium in ball-milled Mg0.63Ti0.27Ni0.10 and Mg0.63Ti0.27Si0.10alloys. Magic angle spinning (MAS) 2H NMR reveals that Ni and Si inducenew types of deuterium sites in addition to the Mg-rich and Ti-rich sitesalready present in Mg0.65Ti0.35D1.2. 2D exchange NMR spectroscopy shows asubstantial deuterium exchange between the various types of sites, whichreflects their close interconnectivity in the crystal structure. Furthermore,the time scale and temperature dependence of the deuterium mobility havebeen quantified by 1D exchange NMR. The obtained effective residencetimes for deuterium atoms in the Mg-rich and Ti-rich nanodomains inMg0.65Ti0.35D1.2, Mg0.63Ti0.27Ni0.10D1.3, and Mg0.63Ti0.27Si0.10D1.1 at 300 K are 0.4, 0.3, and 0.8 s, respectively, and the respectiveapparent activation energies 17, 21, and 27 kJ mol−1. The addition of Ni promotes deuterium mobility inside Mg−Ti hydrides,which is in agreement with the observed catalytic effect of Ni on the electrochemical (de)hydrogenation of these materials.

■ INTRODUCTION

Hydrogen is a promising candidate as energy carrier forapplication in, for example, automotive and portable devices.1−3

One of the major challenges for introducing the hydrogentechnology in our present-day society is related to storinghydrogen in a compact and safe way suitable to serve stationaryand mobile applications.4 Hydrogen can be stored in bothmolecular and atomic form. The former can be achieved bystoring hydrogen gas at high pressures, as liquefied hydrogen atlow temperature and in porous solids, such as molecularorganic frameworks.5−7 The second volumetrically denser formis to bind atomic hydrogen in chemical hydrides, such as metalhydrides.8−10

MgH2 is an interesting material with a high gravimetrichydrogen storage capacity of 7.7 wt %. However, pure MgH2suffers from low (de)hydrogenation kinetics.11−13 MgH2 is arutile-structured compound in which the hydrogen mobility islimited.14 To overcome these kinetic limitations Mg can bemodified either by alloying with transition metals (TM) like Sc,Ti, V, Cr, and Zr, by making use of nanoparticles or by adding acatalyst.15−20 Notten et al. have investigated the hydrogenstorage properties of Mg−Sc alloys and proved the formation ofa favorable fluorite structure after hydrogenation.15,21 Accord-ing to experimental and theoretical investigations, thehomogeneous Mg−TM alloys with TM > 20 at. % tend to

adopt the fluorite structure, while at TM < 20 at. % the rutilestructure is maintained.15,22−24 The hydrogen mobility influorite-structured Mg0.65Sc0.35H2 is significantly faster than inMgH2.

25 As proven by X-ray diffraction and neutron diffraction,Mg−Sc alloys do not undergo phase segregation duringdeuteration/hydrogenation and the deuterium atoms arelocated at tetrahedral positions between the metal atoms inthe fcc lattice.26 However, a major disadvantage of using Sc isits low abundance and high cost. It is therefore desirable toreplace Sc with a more abundant, less precious metal. The mostsuitable replacements are expected to be found with theelements that form dihydrides having the fluorite structure,such as Ti, V, Zr, La, Cr, and rare earths, Hf and Ta.Titanium is an abundant element and therefore particularly

interesting to investigate as promoter of the hydrogen mobilityin Mg-based alloys. Mg and Ti have a positive enthalpy ofmixing and do therefore not form thermodynamically stablealloys. However, metastable alloys can be prepared bynonequilibrium techniques, such as co-sputtering and ball-milling.27,28 Indeed, Mg−Ti thin films show remarkablyimproved (de)hydrogenation kinetics during electrochemical

Received: January 16, 2014Revised: May 1, 2014Published: May 2, 2014

Article

pubs.acs.org/JPCC

© 2014 American Chemical Society 10606 dx.doi.org/10.1021/jp500535q | J. Phys. Chem. C 2014, 118, 10606−10615

hydrogenation, compared to pure Mg films. The improvementis due to the fact that Mg−Ti alloys with a Ti to Mg ratio >0.25adopt a fluorite crystal structure after hydrogenation, whereasMgH2 has a bct (rutile) crystal structure.29,30 Co-sputteredfilms are interesting for dedicated purposes but bulk Mg−Tipowders prepared by ball-milling31−35 would be moreinteresting for large-scale applications. However, such ball-milled Mg−Ti alloys suffer from phase segregation during gas-phase hydrogenation which unavoidably takes place at elevatedtemperatures.36 In contrast, electrochemical hydrogenation/deuteration takes place at room temperature, which preventsthe ball-milled Mg−Ti alloy from phase segregation, as we haverecently shown.37 In this way single phase fcc-structuredhydride compounds can be formed.38

Addition of Ti to Mg certainly improves the (de)-hydrogenation kinetics, but will not raise the partial hydrogenpressure (a thermodynamic property) in desorption applica-tions as a consequence of the high stability of TiH2 (−144 kJ·mol−1 H2) compared to that of MgH2 (−77 kJ·mol−1 H2).According to the rule of reversed stability, which states thatmore stable metal alloys form less stable hydrides,39 addition ofNi is expected to show a destabilizing effect, because Ni has anegative enthalpy of formation with both Mg and Ti. It hasbeen proven that the addition of Al and Si to Mg−Ti thin filmsincreases the equilibrium potential during discharging, which isequivalent to raising the partial hydrogen pressure.40 In thecurrent paper, we will describe the effect of adding Si and Ni toball-milled Mg−Ti alloys in the form of bulk powders.Magic angle spinning 1,2H NMR gives information about the

chemical environment and also about the hydrogen/deuteriummobility inside the host lattice. Solid-state NMR is nowadaysfrequently used to investigate hydrogen storage materials, suchas metal hydrides, alanates and borohydrides.41−44 Although ata given magnetic field the NMR frequency of 1H spins is higherthan that of 2H spins, 2H magic angle spinning (MAS) NMRspectra of metal deuterides reveal a better chemical resolutionthan 1H NMR. This is caused by the fact that the dipolecoupling between 2H nuclei is weaker than that between the 1Hnuclei.45,46 Following on our earlier investigation of deuteratedMg0.65Ti0.35,

37 we have studied the effect of Ni or Si inelectrochemically deuterated Mg0.63Ti0 .27Ni0 .10 andMg0.63Ti0.27Si0.10 compounds. For comparison, the results arepresented and discussed in a combined manner in this article.2H MAS NMR yields interesting information about thechemically different deuterium host sites and their relativeoccupancy. As a measure for the internal hydrogen mobility, wehave also investigated the deuterium dynamic-equilibriumtransfer between the various sites by use of exchange NMRspectroscopy.

■ EXPERIMENTAL SECTIONThe alloy powders were prepared by mechanical alloying usingpure magnesium (Alfa Aesar with particle size, ∼50 μm, andpurity, 98.5%), titanium (Sigma-Aldrich: ∼50 μm, 99.5%)Nickel (Alfa Aesar: ∼50 μm, 99%) and Silicon (Alfa Aesar: ∼50μm, 99%). The Mg−Ti binary alloy was prepared by millingelemental magnesium (65 at. %) and titanium (35 at. %) underArgon in a 55 mL tungsten-carbide milling vial using tungsten-carbide balls of 1 cm diameter as milling media with a ball-to-powder ratio of 30:1. The type of mill used for the mechanicalalloying process was SPEX shaker. The alloying process wasclosely monitored by observing the shift in the Mg reflectionstoward higher angle and decrease in the intensity of Ti

reflections in XRD pattern. After 28 h of milling the alloyingprocess was completed, when the resulting XRD pattern didnot change any more. A Mg−Ti solid solution was clearlyformed, which was concluded from the decrease in the hcplattice parameter compared to that of Mg upon milling.Ternary alloys were prepared by adding nickel and silicon

separately after the formation of Mg0.70Ti0.30. The millingprocess was continued for another 12 h when the ternaryelement was added. A decrease in the reflections of the ternaryelements Ni and Si was clearly observed. The reflections of theternary elements have completely disappeared after 12 h in theNi system and after 6 h in the case of Si. Finally, 5 at. % of Pdwas added to the homogeneous Mg0.65Ti0.35, Mg0.63Ti0.27Ni0.10and Mg0.63Ti0.27Si0.10 alloys and shortly milled for 2 h. In thisway, Pd forms a separate phase at the external particle surfaceacting as an (electro)catalyst for the hydrogenation/deuterationreaction and also protecting the underlying alloy fromoxidation.47 For typographic brevity we do not specify Pd0.05in the compound names. However, we do take the weightfraction of the Pd phase (expected not to contain deuteriumunder our experimental conditions) into account whenestimating the overall deuterium content x from electro-chemical measurements. The alloy powders are mixed witheither carbon black (1 wt % carbon black and 2 wt % PTFE asbinder) or silver (Ag/alloy wt. ratio of 4:1) and subsequentlypressed (pressure = 400 MPa) into pellets with 8 mm diameterprior to electrochemical deuteration. A Maccor (Model 2300)testing system was used for the deuterium loading. Theelectrochemical deuterium loading, denoted here as charging,was carried out in a conventional three-electrode electro-chemical cell which was thermostated at 25 °C, using 6 MKOH in D2O as electrolyte. A pure Pd metal rod was used ascounter electrode while Hg/HgO was used as referenceelectrode. All electrode potentials are referring to this referenceelectrode. The alloy powders were subjected to constantcurrent charging and discharging with a current density of 50mA g−1. For low current discharge 10 mA g−1 was used. Powderelectrodes made with silver resulted in well-defined voltageplateaus and lower electrochemical overpotentials. Therefore,silver was used as conducting material to study the electro-chemical (dis)charging behavior of the alloy powders (Figure1). However, the high weight and conductivity of Ag causessample-rotation instability and skin depth problems in the MASNMR experiments. To overcome these problems carbon black(CB) was used as additive, instead of Ag, for the materialsstudied with NMR. Even though the electrodes made withcarbon black reveal much higher overpotentials duringelectrochemical charging experiment than those with silver,the deuterium capacities of the CB based electrode materialsare similar to those of electrodes containing silver.37

Magic angle spinning (MAS) 2H NMR experiments wereperformed with a Bruker DMX500 spectrometer, operating at76.77 MHz for deuterium. The 2.5 mm NMR sample holderwas packed in the glovebox under argon, closed with a gastightcap and rotated in a flow of dry nitrogen inside thespectrometer. The typical sample spinning rate was between8 to 12 kHz. The deuterium weight content of the Mg−Tideuteride was estimated by comparison with the spectralintegral of the quantitative 2H NMR spectrum (interscan delayof 20 s) of sodium 3-(trimethylsilyl)propionate-2,2,3,3-d4,which was used as external reference compound with fourdeuterium atoms per molecule.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp500535q | J. Phys. Chem. C 2014, 118, 10606−1061510607

Two-dimensional exchange spectroscopy (2D Exsy) wascarried out by use of a pulse sequence with three 90° pulses of5 μs. The evolution time interval t1 between the first two pulseswas systematically incremented during the experiments. Thetime interval between the second and third pulse, called mixingtime tmix, was fixed at 0.01, 1, and 2 s.48 One-dimensionalexchange spectroscopic (1D Exsy) measurements wereperformed by selectively perturbing the polarization of thedeuterium nuclei at the Mg-rich sites resonating at 4 ppm, usinga rotor-synchronized DANTE pulse train of five pulses of 2 μs(50 kHz nutation frequency) and monitoring the polarizationafter a variable time interval tmix by means of a nonselective 90°pulse of 5 μs.49 The 1D exchange spectra were recorded for 12tmix values up to 10 s at four temperatures: 300, 318, 337, and355 K. Temperature calibration is done by measuring the peakseparation for ethylene glycol.

■ RESULTS AND DISCUSSIONFigure 1 shows the electrochemical charging and discharging ofMg0.65Ti0.35, Mg0.63Ti0.27Ni0.10, and Mg0.63Ti0.27Si0.10. Electro-chemical deuterium loading was done by applying a constantcurrent to the working electrode. The deuterium oxide in theelectrolyte is reduced, according to

+ + ⇐ ⇒=== +− −M D O e MD ODdischarging

chargingad2

(1a)

Subsequently, MDad is absorbed to form the hydrideaccording to

⇔MD MDad abs (1b)

The constant-current charging of the three electrodematerials clearly exhibits limiting negative voltage plateauscorresponding to complete deuteration of the metal compo-

sites. No further deuterium atoms can enter the bulk phase andat more negative potentials D2 gas evolves from the surface ofthe electrode according to,

⇔ + ↑2MD 2M Dad g2( ) (2)

From a thermodynamic point of view, D2 evolution can onlytake place beyond the stability window of D2O. As proposedbefore, the amount of deuterium absorbed in the material canbe calculated from the transition between the second and thirdvoltage plateau (curves a1, b1, c1 in Figure 1a).37 Maxima in thevoltage-derivative curves (a′1, b′1, c′1) reflect the transitionsbetween different plateau regions. The electrochemical storagecapacity of Mg0.65Ti0.35 (825 mA h g−1), Mg0.63Ti0.27Ni0.10 (820mA h g−1) and Mg0.63Ti0.27Si0.10 (760 mA h g−1) corresponds to6.2, 6.2, and 5.7 wt % deuterium, respectively. The theoreticalcapacity of Mg0.65Ti0.35 (1292 mA h g−1), Mg0.63Ti0.27Ni0.10(1245 mA h g−1) and Mg0.63Ti0.27Si0.10 (1339 mA h g−1)corresponds to 9.7, 9.4, and 10.1 wt % deuterium, respectively.The theoretical capacities are based on absorption of twodeuterium atoms per metal atom. The charging curve ofMg0.63Ti0.27Ni0.10 (curve b1 in Figure 1) reveals a well-definedplateau at significantly lower overpotential than Mg0.65Ti0.35(curve a1), which can be ascribed to the catalytic effect of Ni.The charging curve of Mg0.63Ti0.27Si0.10 (c1), in contrast, showssubstantial higher overpotential compared to Mg0.65Ti0.35 andMg0.63Ti0.27Ni0.10. The discharge capacities (summation of high-and low-current discharge capacities) of Mg0.65Ti0.35D1.2,Mg0.63Ti0.27Ni0.10D1.3, and Mg0.63Ti0.27Si0.10D1.1 are calculatedto be 600 (curve a3 in Figure 1), 620 (curve b3) and 380 mA hg−1 (curve c3), respectively. The difference between therespective charging and discharging capacities is attributed tothe “irreversibly” bound deuterium in Ti-rich sites, as indicatedby the more negative formation enthalpy of TiH2 than that ofMgH2. The lower deuterium-loading potential forMg0.63Ti0.27Ni0.10 can be indicative for a relative thermodynamicdestabilization of the Mg−Ti hydride resulting from Niaddition, or improved hydrogen (de)sorption kinetics causedby Ni as a catalyst. This is advantageous for the hydrogenrelease properties. In contrast, Si addition has a negative effecton the overall kinetics, as is also evident from the much lowerreversible deuterium-storage capacity of Mg0.63Ti0.27Si0.10. Allcompounds have lower deuterium content than the theoret-ically expected D/M ratio 2. A possible explanation could bethat the Pd coating in the final ball-milling step of MgTi,MgTiSi, and MgTiNi particles is not homogeneous. Particleswithout Pd coating would not readily be hydrogenated.The nanostructures of the electrochemically hydrogenated

MgTi-based materials were investigated in more detail with 2HNMR. The spin-1 nuclei of deuterium atoms interact with localelectric field gradients, which in the fluorite crystal structurearise from metal coordination asymmetry around the D atomsat the tetrahedral interstitial sites. The interaction with thesegradients causes quadrupolar line broadening in 2H NMRspectra without sample rotation. This broadening is removedby rotating the sample at the magic angle with respect to themagnetic field. In this way, magic angle spinning (MAS)increases the chemical resolution in the 2H NMR spectra. Atintermediate sample-rotation rates, spinning sidebands arevisible. The extent of these spinning sideband patterns reflectsthe stationary 2H NMR line width, and thus the deuterium-coordination symmetry in the respective host crystal lattice.Figure 2 shows 2H MAS NMR spectra of Mg0.65Ti0.35D1.2,

Mg0.63Ti0.27Ni0.10D1.3, and Mg0.63Ti0.27Si0.10D1.1, as well as earlier

Figure 1. (a) Charging voltage (left axis) and derivative (right axis)during electrochemical deuteration of Mg0.65Ti0.35 (a1, a′1),Mg0.63Ti0.27Ni0.10 (b1, b′1), and Mg0.63Ti0.27Si0.10 (c1, c′1) at 50 mA hg−1. (b) High-current and low-current discharge of Mg0.65Ti0.35 (a2, a3),Mg0.63Ti0.27Ni0.10 (b2, b3) and Mg0.63Ti0.27Si0.10 (c2, c3) at 50 and 10 mAg−1, respectively. Low-current discharging is performed consecutivelyafter high-current discharging to extract the maximum amount ofdeuterium from the electrode. The discharge capacities reported in thetext is the summation of the high- and low-current discharge capacities.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp500535q | J. Phys. Chem. C 2014, 118, 10606−1061510608

published37 spectra of MgD2 and TiD2, for comparison. Thelatter two single-component materials were prepared by gas-phase hydrogenation, because no phase segregation can occur.The 2H NMR spectrum of MgD2 exhibits an extensive sidebandpattern over a wide frequency range (Figure 2a). This is inagreement with the 3-fold coordination of deuterium by Mgatoms in rutile MgD2. The overall deuterium-to-metal ratio D/M has been estimated from the total NMR integral and wasfound to be significantly smaller than 2. At D/M < 2 MgD2splits into a 2H NMR visible MgD2 phase and a 2H NMRinvisible MgD< 0.001 phase.

20 TiD2 has a fluorite structure withdeuterium atoms located at tetrahedral interstitial sites. Thehigh coordination symmetry results in weak sidebandssurrounding the centerband at −150 ppm for gas-phasedeuterated TiD2 (Figure 2e). The large, negative Knight shiftfor TiD2 is typical for the conductive bulk material.Figure 2 also shows the 2H NMR spectra of the

electrochemical ly deuterated Mg0 .65Ti0 . 35D1 . 2 (b),Mg0.63Ti0.27Ni0.10D1.3 (c), and Mg0.63Ti0.27Si0.10D1.1 (d). Theoverall deuterium content estimated from the total spectralintegral (5.7, 5.1, and 5.4 wt %, respectively) is in all these casesin good agreement with the electrochemically determineddeuterium content. The fact that no signal of bulk TiD2 isvisible at −150 ppm (Figure 2e) shows that no phasesegregation takes places during the low-temperature electro-chemical deuteration. Still, the signal at −73 ppm signifies thepresence of TiD2 nanodomains similar to those found in co-sputtered Mg0.65Ti0.35D1.1.

50 A centerband at 4 ppm alsoappears similar to that found for pure MgD2 which cantherefore be assigned to deuterium atoms located at Mg-richsites. The 2H MAS NMR sideband patterns in Figure 2 reflectthe local electric field at the position of the deuterium atoms inthe respective crystal structures. For the alloys the sidebandpattern extends over a narrower frequency range than for MgD2(Figure 2a). In fact, it is closely similar to that of the single-phase fluorite materials melt-cast Mg0.65Sc0.35D2.2

51 and co-sputtered Mg0.65Ti0.35D1.1.

50 XRD reflections of the currentthree electrochemically deuterated materials are too broad toprove the fluorite nanostructure, but for Mg0.65Ti0.35D1.2,confirmation comes from selected-area electron diffraction.37

The tetrahedral metal coordination of deuterium in the fluoritestructure causes weaker 2H MAS NMR sidebands than the 3-fold coordination in rutile MgD2. Despite the similartetrahedral metal coordination of deuterium atoms in thefluorite structure, the sidebands of Mg0.65Ti0.35D1.2,Mg0.63Ti0.27Ni0.10D1.3, and Mg0.63Ti0.27Si0.10D1.1 are strongerthan those of TiD2 (Figure 2e). This is ascribed to the brokensymmetry caused by mixed metal coordination and localdistortions in the crystal structure, resulting from thecopresence of different metal atoms Mg and Ti, as well as Niin Mg0.63Ti0.27Ni0.10D1.3 and Si in Mg0.63Ti0.27Si0.10D1.1.

2H NMR spectra of the alloys electrochemically deuterated atroom temperature are shown in Figure 3. The 2H NMR

spectrum of Mg0.65Ti0.35D1.2 (Figure 3a) can be decomposedinto four overlapping line shape components with centerbandsignals positioned at 7, 4, −30, and −73 ppm (Table 1). The 4ppm component has a chemical shift close to that of Mg-richdeuterium sites in melt-cast co-sputtered Mg0.65Ti0.35D1.1 andMg0.65Sc0.35D2.2.

50,51 It is therefore assigned to deuterium with

Figure 2. 2H NMR spectrum of (a) MgD2, (b) Mg0.65Ti0.35D1.2, (c)Mg0.63Ti0.27Ni0.10D1.3, (d) Mg0.63Ti0.27Si0.10D1.1, and (e) TiD2. Spinningsidebands are marked with an asterisk. Sample rotation rates werebetween 8 and 12.5 kHz.

Figure 3. Deconvolution of 2H MAS NMR spectra of (a)Mg0.65Ti0.35D1.2, (b) Mg0.63Ti0.27Ni0.10D1.3, and (c) Mg0.63Ti0.27Si0.10D1.1at 300 K by use of the line shape components specified in Table 1

Table 1. Line Shape Components Used for Deconvolution ofthe 2H MAS NMR Spectra

line shape component shift (ppm) width (ppm)

A 7 12A′ 12 12B 4 4C −30 64D −73 49E −40 20F −47 22G −95 40

Component A for Mg0.65Ti0.35D1.2, and A′ for Mg0.63Ti0.27Ni0.10D1.3 andMg0.63Ti0.27Si0.10D1.1; component E for Mg0.63Ti0.27Ni0.10D1.3 and F forMg0.63Ti0.27Si0.10D1.1; component G for Mg0.65Ti0.35D1.2 at T ≥ 318 K.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp500535q | J. Phys. Chem. C 2014, 118, 10606−1061510609

tetrahedral MgnTi4‑n (0 ≤ n ≤ 4) coordination in thediamagnetic Mg-rich nanodomains.The assignment of the minor 7 ppm signal is less evident.

Unlike for the 4 ppm component, there are no spinningsidebands associated with this component and it may thereforebelong to deuterium atoms with tetrahedrally symmetric Mg4coordination. However, this signal remains invariant uponelectrochemical discharging,37 and it is unclear why Mg4coordinated deuterium would be irreversibly bound. Becausethe 7 ppm component has not been observed for co-sputteredand gas-phase loaded Mg0.65Ti0.35D1.1, the 7 ppm componentcould also be due to residual D2O or OD− traces left over fromthe electrochemical treatment. However, this does not explainwhy this signal does not vary in intensity between the chargedand discharged material, which were independently collectedand dried. The components at −30 and −73 ppm arecomparable to those attributed to the Ti-rich nanodomainsfound in co-sputtered Mg0.65Ti0.35D1.1.

49 Their different Knightshifts reflect the different distance of the respective deuteriumsites from the Mg−Ti interface. The “deeper” the site, the morenegative the Knight shift will be with −150 ppm as the limitingKnight shift value for bulk TiD2 (Figure 2e). The sidebandpattern in the 2H NMR spectra for Mg0.63Ti0.27Ni0.10D1.3 andMg0.63Ti0.27Si0.10D1.1 is similar to that of Mg0.65Ti0.35D1.2 and co-sputtered Mg0.65Ti0.35D1.1.

50

The 4 ppm component is also present in the spectra for boththe Ni- and Si-modified compounds (Figures 3b,c). The minordownfield component is much less pronounced in these spectracompared to the spectrum of Mg0.65Ti0.35D1.2 and has shifted to12 ppm. The −30 ppm component is still present, whereas therelative peak area of the −73 ppm component has reduced. Anew component at −40 ppm is present in 2H NMR spectrumof Mg0.63Ti0.27Ni0.10D1.3 and, for that reason, can be assigned todeuterium sites with one or more Ni neighbors. The chemicalshift is more negative than that reported for other Ni-containing hydrides, like Mg2NiHx and ZrNiDx.

52,53 Wetherefore assume the −40 ppm shift to reflect the Knightshift of deuterium sites with mixed Ti−Ni coordination (TiNi)in the Ti-rich nanodomains. Similarly, a new component at −47ppm is present in 2H NMR spectrum of Mg0.63Ti0.27Si0.10D1.1,which can be attributed to deuterium coordinated by both Tiand Si (TiSi). The new TiNi and TiSi components present in the2H NMR spectra of both Mg0.63Ti0.27Ni0.10D1.3 andMg0.63Ti0.27Si0.10D1.1 are narrower than the −30 and −73 ppmcomponents in all three spectra shown in Figure 3. With theheterogeneity of deuterium sites expected to increase as a resultof the Ni or Si additive, the actually observed smaller line widthcan be indicative for increased motional averaging caused by ahigher mobility of deuterium at these sites.The interconnectivity and deuterium self-diffusion between

different deuterium sites have been studied by 2D exchangeNMR spectroscopy (2D Exsy; Figure 4). More specifically, 2DExsy correlates the chemical shift of deuterium atoms beforeand after a selected time interval, the so-called mixing time tmix.During tmix, the deuterium atoms have the opportunity to hopbetween different sites. Deuterium atoms which are stablybound or present within a spatially separate phase do notparticipate in the exchange process. These deuterium atoms donot undergo a chemical-shift change during tmix and will giverise to peaks at the spectral diagonal only. Deuterium atomswhich are reversibly bound in a single host lattice willparticipate in the exchange process. The deuterium atoms,which exchange between different sites, such as Mg-rich and Ti-

rich sites, during the mixing time tmix will give rise to a changein chemical shift, which results in off-diagonal “cross-peaks” inthe 2D spectrum.Figure 4 shows the 2D spectra of the electrochemically

deuterated alloys for long mixing time tmix. The clearly visiblecross-peaks indicate intensive deuterium exchange betweenMg-rich and Ti-rich sites at the time scale of seconds. For shortmixing time tmix = 0.01 s, the cross-peaks are absent (FigureS1), which confirms that there is no exchange between the Mg-rich and Ti-rich sites at this time scale. At this short mixing timethe 2D exchange 2H NMR spectrum of Mg0.63Ti0.27Ni0.10D1.3reflects a strong homogeneous broadening of the −40 ppmcomponent (Figure S1a). This is probably due to the presenceof Ni, which causes shortening of the deuterium transversalrelaxation time T2. A complete exchange can be recognized in2D Exsy from the similarity between cross-section S in the 2Dspectrum and projections P onto the frequency axes. Theprojections P reflect the distribution of deuterium amongvarious sites, while a horizontal cross section is indicative of theredistribution of deuterium atoms, which are initially located atone type of sites, during tmix. The similarity between the crosssections and the projection indicates a homogeneousredistribution of deuterium atoms at the time scale probed.The projections P and cross sections S in the 2D spectra of thethree materials shown in Figure 4 are similar and thereforeillustrative for a complete exchange at the time scale of seconds.Apparently, the different deuterium sites recognized in the 2HNMR spectra are connected by diffusion pathways, asconsistent with a nanostructured composite material of Mg-rich and Ti-rich nanodomains.Figure 5 shows 2H NMR spectra of Mg0.65Ti0.35D1.2,

Mg0.65Ti0.27Ni0.1D1.3, and Mg0.65Ti0.27Si0.1D1.1 at varied temper-atures between 300 and 355 K, along with the deconvolution interms of the line shape components specified in Table 1. Thespectral changes are relatively small and fully reversible uponcooling. For Mg0.65Ti0.35D1.2 (Figure 5a), a relative increase ofthe −30 ppm signal component and a decrease of the 4 ppmcomponent (combined with the minor 7 ppm component) isobserved at increasing temperature. In contrast, the relativeintensity of the −73 ppm component (combined with a minor−95 ppm component above 350 K) stays approximatelyconstant. At first sight, the enhanced −30 ppm intensity mayseem to be caused by a deuterium fraction, which exchanges

Figure 4. 2D exchange spectra at 300 K of (a) Mg0.65Ti0.35D1.2, tmix = 2s, (b) Mg0.63Ti0.27Ni0.10D1.3, tmix = 1 s, and (c) Mg0.63Ti0.27Si0.10D1.1, tmix= 1 s. Projections P and cross sections S for δ1 = 4 ppm are shownabove the 2D spectra. P illustrates the overall deuterium distribution,while S reveals the redistribution of deuterium atoms initially at theMg-rich sites.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp500535q | J. Phys. Chem. C 2014, 118, 10606−1061510610

fast between Mg-rich (4 ppm) and Ti-rich (−73 ppm)nanodomains at elevated temperatures. However, the fact thatonly the 4 ppm component decreases, but not the −73 ppmcomponent, is inconsistent with such explanation. The apparentcorrelation between the decrease of the combined 7 and 4 ppmcomponents and the increase of the −30 ppm componentsuggests that at increasing temperature there is a netredistribution of deuterium atoms from the Mg-rich nano-domains to sites close to the Mg−Ti interface in the Ti-richnanodomains. This may be compared with our previousobservation for co-sputtered Mg0.65Ti0.35D1.1. At increasingtemperature, the latter material showed a similar decrease of the

4 and 7 ppm components, and a new signal without spinningsidebands appeared at −10 ppm, while the relative intensities of−29 and −68 ppm components stayed almost constant.50 Inthe present study, no temperature induced increase of the −30ppm component, nor substantial decrease of the 4 ppmcomponent is observed for the Ni-containing and Si-containingternary hydrides. The most remarkable features are theincreasing component at −40 ppm for Mg0.63Ti0.27Ni0.10D1.3

(Figure 5b) and the one at −47 ppm for Mg0.63Ti0.27Si0.10D1.1

(Figure 5c), which we have above assigned to deuterium siteswith Ti−Ni and Ti−Si coordination, respectively.

Figure 5. 2H MAS NMR spectra of (a) Mg0.65Ti0.35D1.2, (b) Mg0.63Ti0.27Ni0.10D1.3, and (c) Mg0.63Ti0.27Si0.10D1.1 at varied temperatures, ranging from300 (bottom) to 355 K (top). The spectra are deconvoluted with line shape components specified in Table 1.

Figure 6. 1D Exsy spectra at 300 K of (a) Mg0.65Ti0.35D1.2, (b) Mg0.63Ti0.27Ni0.10D1.3, and (c) Mg0.63Ti0.27Si0.10D1.1 at varied mixing times tmix (0.001,0.3, 1, 10 s). For comparison, the fully relaxed spectrum after 10 s mixing time is also shown (gray curves). (d−f) Combined peak areas IMg of theMg signal components at 4 and 7 ppm, and combined peak areas (d) ITi (−30 and −73 ppm), (e) ITi−Ni (−30, −40, −73 ppm), and (f) ITi−Si (30,−47, −73 ppm) as a function of tmix together with the total spectral intensity Itot. The curves in d−f are based on least-squares fits of the coupledbiexponential model described by eq 3.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp500535q | J. Phys. Chem. C 2014, 118, 10606−1061510611

To study the deuterium exchange as a function of mixingtime, one-dimensional (1D) Exsy was performed. The 1D Exsyexperiments comprise an initial selective polarization perturba-tion of the deuterium spins in Mg-rich sites resonating around4 ppm. This is followed by variable mixing time tmix duringwhich the deuterium atoms with initially perturbed nuclear spinpolarization at the Mg-rich sites will replace the deuteriumatoms with unperturbed polarization at the Ti-rich sites and viceversa. As a consequence of this dynamic equilibrium exchangeof deuterium atoms with opposite spin polarization, the signalintensity of the initially perturbed sites will increase and theunperturbed sites will decrease as a function of mixing time.The recovery of the signal also depends on spin−latticerelaxation, which is generally slow. Thus, even withoutdeuterium exchange the perturbed polarization will relax toits thermal equilibrium value. Therefore, the longest deuterium-exchange time scale that can be probed by use of 1D Exsy isdetermined by spin−lattice relaxation.From the series of 1D exchange spectra for the three alloys

obtained with 12 different mixing times, only the spectrameasured at tmix = 0.001, 0.3, 1, and 10 s at 300 K are shown inFigure 6, parts a−c. The spectra for tmix= 0.001 s show theinitially negative intensity IMg of deuterium atoms at the Mg-rich sites resonating at 4 ppm. This signal intensity has alreadypartly recovered after 0.3 s as a result of dynamic-equilibriumexchange with deuterium atoms with initially unperturbed spinpolarization at Ti-rich sites. Simultaneously with the fastrecovery of IMg, the combined signal intensity ITi, ITi−Ni, andITi−Si of deuterium at Ti-rich sites in Mg0.65Ti0.35D1.2 (−30 and−73 ppm), Mg0.63Ti0.27Ni0.10D1.3 (−30, −40, −73 ppm), andMg0.63Ti0.27Si0.10D1.1 (−30, −47, −73 ppm) initially decreases,indeed. After the initial intensity changes, the final restorationof all signal components occurs under influence of the spin−lattice relaxation. Beyond 1 s all signal components increase atthe same rate as the spin polarization relaxation is dominatedby spin−lattice relaxation. The 1D Exsy is analyzed in aquantitative manner by deconvoluting the spectra in terms ofGaussian−Lorentzian line shape components with fixedpositions and fixed line widths derived from the respectiveone-dimensional spectra (Figure 3, Table 1). Parts d−f ofFigure 6 illustrate how the relative peak area of deuterium atMg-rich and Ti −rich sites develop as a function of the mixingtime. Deuterium self-diffusion between Mg-rich and Ti-richnanodomains expectedly involves multiple hopping stepsbetween neighboring interstitial sites. Careful modeling of theexchange curves resulting from this multistep diffusion wouldrequire detailed assumptions about parameters, such as theshape and size (distribution) of the nanodomains, as well as therespective deuterium occupation and diffusion in the Mg-richand Ti-rich nanodomains. The relatively simple shape of theobserved 1D Exsy curves (Figure 6d−f), however, makes suchdetailed interpretation ambiguous. Phenomenologically, thecombined behavior of IMg(tmix) and ITi(tmix) turns out to be welldescribed by the coupled biexponential curves for two-siteexchange:

τ= − − − −

− +

∞ ∞I t I I t I I

t T I

( ) [{ (0) }exp( / ) { }]

exp( / )

Mg mix Mg Mg mix ex Mg Mgeq

mix 1 Mgeq

(3a)

τ= − − − −

− +

∞ ∞I t I I t I I

t T I

( ) [{ (0) }exp( / ) { }]

exp( / )Ti mix Ti Ti mix ex Ti Ti

eq

mix 1 Tieq

(3b)

where the thermal-equilibrium intensities (IMgeq and ITi

eq) areproportional to the respective densities of Mg-rich and Ti-richsites and IMg

∞ = {IMg(0) + ITi(0)}IMgeq /(IMg

eq + ITieq) and ITi

∞ ={IMg(0) + ITi(0)}ITi

eq/(IMgeq + ITi

eq) are the signal intensities in theintermediate state when the combined initial polarizationIMg(0) + ITi(0) is homogeneously distributed over the Mg- andTi-rich sites by deuterium exchange. The use of a single spin−lattice relaxation time T1 in eq 3 tacitly assumes that anyintrinsic 2H T1 relaxation differences between the Mg-rich andTi-rich nanodomains are averaged by relatively fast deuteriumexchange. Equation 3 differs slightly from the biexponential fitmodel in our previous publication37 in that the fast decayingterm is now corrected for spin−lattice relaxation, as well.The effective deuterium residence time τex determined by

fitting the biexponential model to the data recorded at 300 Kequals 0 .4 , 0 .3 , and 0 .8 s for Mg0 . 6 5Ti0 . 3 5D1 . 2 ,Mg0.63Ti0.27Ni0.10D1.3, and Mg0.63Ti0.27Si0.10D1.1, respectively.These are the time scales at which the deuterium atomsmove in dynamic equilibrium between Mg-rich and Tinanodomains in the above three compounds.To determine the activation barrier for deuterium motion,

1D Exsy and spin−lattice relaxometry was carried out as afunction of temperature (Figure 7). Before applying eq 3 as a fit

model to the 1D exchange curves (such as the ones at 300 K inFigure 6) we have measured spin−lattice relaxation (T1) of thethree materials by use of inversion−recovery NMR (Figure 7b).2H NMR spin−lattice relaxation is controlled by fast deuteriummotions with correlation times τc in the range 10−10−10−8 s,tentatively associated with deuterium hopping betweenneighboring interstitial sites in the fcc lattice. The exponentialdecrease with inverse temperature (Figure 7b) indicates, that inthis temperature range T1

−1 is inversely proportional to τc. Thispermits extracting the activation barriers from the slopes in theArrhenius plot, as 15, 11, and 9 kJ mol−1 for Mg0.65Ti0.35D1.2,Mg0.63Ti0.27Ni0.10D1.3, and Mg0.63Ti0.27Si0.10D1.1, respectively.Addition of Ni or Si thus lowers the effective energy barriersfor the fast deuterium motions. The observed deuteriumexchange observed in 1D Exsy becomes faster with increasingtemperature (Figure 7a). For instance, for Mg0.63Ti0.27Ni0.10D1.3,τex changes from 0.3 s at 300 K to 0.07 s at 355 K. The fasterexchange at increasing temperature is consistent with actualdeuterium self-diffusion, and not, for instance, spin diffusion of

Figure 7. Arrhenius plot of (a) effective deuterium-exchange ratesτex

−1 and (b) 2H NMR spin−lattice relaxation rates 1/T1 for (●)Mg0 . 6 5T i 0 . 3 5D1 . 2 , (×) Mg0 . 6 3T i 0 . 1 7Ni 0 . 1 0D1 . 3 , and (□)Mg0.63Ti0.17Si0.10D1.1. The relaxation rates (b), obtained frominversion−recovery NMR experiments at varied temperature, wereinserted into the coupled biexponential fit (eq 3) of the 1D exchangeresults (Figure 6) at the corresponding temperatures to extract theexchange rate (a).

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp500535q | J. Phys. Chem. C 2014, 118, 10606−1061510612

deuterium magnetization. The latter mechanism based onmagnetic dipole-coupling would expectedly be weak anyway,because of the low magnetic dipole moment of deuteriumnuclei and the magic angle spinning applied. Between 300 and355 K deuterium exchange is faster in Mg0.63Ti0.27Ni0.10D1.3 andslower in Mg0.63Ti0.27Si0.10D1.1 than in Mg0.65Ti0.35D1.2. This inagreement with the improved electrochemical kinetic results forthe deuteration of the Ni alloy and the inhibited kinetic resultsfound for the Si-containing alloy.The temperature trends of the exchange rates τex

−1 forMg0.65Ti0.35D1.2, Mg0.63Ti0.27Ni0.10D1.3, and Mg0.63Ti0.27Si0.10D1.1correspond to activation energies of 17, 21, and 27 kJ mol−1,respect ive ly . The fas ter deuter ium exchange inMg0.63Ti0.27Ni0.10D1.3 can thus not be explained by lower energybarriers. For Mg0.65Ti0.35D1.2, the effective barriers for thedeuterium exchange and spin−lattice relaxation are similar.This suggests that the motions at the 10−10−10−8 s time scale,underlying T1 relaxation are the elementary steps of deuteriumexchange between at Mg-rich and Ti-rich nanodomains in thethree materials. The extracted exchange times in the order of0.1 s then indicate that 107−109 jumps are necessary to movebetween the nanodomains. With the tetrahedral deuterium sitesseparated by half a lattice constant, this seems inconsistent withnanodomains smaller than the XRD coherence length. In analmost full lattice, however, deuterium atoms may spend largepart of the time jumping back and forth between neighboringsites, so-called “flicker events”.54−56 Such “flicker events”contribute to spin−lattice relaxation, but are insufficient incausing deuterium diffusion beyond the length scale of thelattice spacing.The unequal Arrhenius slopes for T1 relaxation and

deuterium exchange for the Ni- and Si-containing materials(Figure 7), however, shows that the above picture forMg0.65Ti0.35D1.2 is not the full story. The observed deuteriumexchange between Mg-rich and Ti-rich nanodomains in thesecomplex hydride materials is probably the result of acombination of several factors, including, indeed, deuteriumhopping between neighboring interstitial sites (as probed withT1 relaxation), but also the size of the nanodomains and thenature of their interfaces, which may be affected by the additionof Ni and Si during the ball-milling prior to the electrochemicalhydrogenation.

■ CONCLUSIONIn contrast to gas phase loading of metastable MgTi alloys athigh temperatures, electrochemical deuteration/hydrogenationat room temperature does not induce phase segregation. The2H NMR spectrum of electrochemically loaded Mg0.65Ti0.35D1.2

consist of Mg-rich (4 ppm) and Ti-rich sites (−30 ppm and−75 ppm). Addition of Ni or Si leads to new deuterium siteswith partial Ni or Si coordination with respective chemicalshifts of −40 or −47 ppm. 2D Exsy shows substantialdeuterium exchange between the Mg- and Ti-rich sites at atime scale of seconds. The deuterium exchange process hasbeen quantified by use of 1D Exsy at temperatures in the rangeof 300 to 355 K. The deuterium residence time forMg0.65Ti0.35D1.2 is 0.33 s at 300 K and decreases with increasingtemperature. The addition of Ni decreases the deuteriumresidence time in Mg0.63Ti0.27Ni0.10D1.3 to 0.24 s at 300 K, whileaddition of Si increases the deuterium residence time inMg0.63Ti0.27Si0.10D1.1 to 0.6 s at 300 K. The rate of deuteriummobility is therefore significantly increased by the addition ofNi, and decreased by the addition of Si to the Mg−Ti hydride

materials. The 2H NMR results are in good agreement with theelectrochemical kinetic results. Especially, the addition of Niturned out to be favorable to improve the mobility ofdeuterium/hydrogen for application of fcc Mg−Ti hydridematerials for hydrogen storage.

■ ASSOCIATED CONTENT*S Supporting InformationFigure S1, 2D exchange spectra of Mg0.63Ti0.27Ni0.10D1.3 andMg0.65Ti0.35D1.2 recorded with tmix = 0.01 s at 300 K. Thismaterial is available free of charge via the Internet at http://pubs.acs.org.

■ AUTHOR INFORMATIONCorresponding Authors*(P.C.M.M.M.) Telephone: +32 16 37 91 35. E-mail: [email protected].*(P.H.L.N.) Telephone: +31 40 247 3069. E-mail: [email protected] Address§RWTH Aachen University, Institute for Power Electronics andElectrical Drives, Jaegerstr. 17−19, D-52066 Aachen, GermanyNotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThis research work in TU/e was financially supported by theDutch Science foundation (NWO) as part of the SustainableHydrogen program of Advanced Chemical Technologies forSustainability (ACTS). The authors would like to thank BrahimMezari for NMR assistance.

■ REFERENCES(1) Murphy, J. D.; Thamsiriroj, T. What Will Fuel Transport Systemsof the Future? Mater. Today 2011, 11, 518−524.(2) Moriarty, P.; Honnery, D. Hydrogen’s Role in an UncertainEnergy Future. Int. J. Hydrogen Energy 2009, 34, 31−39.(3) Peraldo Bicelli, L. Hydrogen: A Clean Energy Source. Int. J.Hydrogen Energy 1986, 11, 555−562.(4) Schlapbach, L.; Zuttel, A. Hydrogen-Storage Materials for MobileApplications. Nature 2001, 414, 353−358.(5) Von Helmolt, R.; Eberle Gonalez, U. Fuel Cell Vehicles: Status2007. J. Power Sources 2007, 165, 833−843.(6) Sherif, S. A.; Zeytinoglu, N.; Veziroglu, T. N. Liquid hydrogen:Potential, Problems, and a Proposed Research Program. Int. J.Hydrogen Energy 1997, 22, 683−688.(7) Seayad, A.; Antonelli, D. M. Recent Advances in HydrogenStorage in Metal-Containing Inorganic Nanostructures and RelatedMaterials. Adv. Mater. 2004, 16, 765−777.(8) Sakintuna, B.; Lamari-Darkrimb, F.; Hirscher, M. Metal HydrideMaterials for Solid Hydrogen Storage: A Review. Int. J. HydrogenEnergy. 2007, 32, 1121−1140.(9) Zuttel, A.; Borgschulte, A.; Schlapbach, L. Hydrogen as a FutureEnergy Carrier; Wiley-VCH Verlag GmbH &Co. KGaA: Weinheim,Germany, 2008; p 5.(10) Guo, Z. X.; Shang, C.; Aguey-Zinsou, K. F. Materials Challengesfor Hydrogen Storage. J. Eur. Ceram. Soc. 2008, 28, 1467−1473.(11) Manivasagam, T. G.; Kiraz, K.; Notten, P. H. L. Electrochemicaland Optical Properties of Magnesium-Alloy Hydrides Reviewed.Crystals 2012, 2, 1410−1433.(12) Zuttel, A. Materials for Hydrogen Storage. Mater. Today 2003,6, 24−33.(13) Selvam, P.; Viswanathan, B.; Swamy, C. S.; Srinivasan, V.Magnesium and Magnesium Alloy Hydrides. Int. J. Hydrogen Energy1986, 11, 169−192.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp500535q | J. Phys. Chem. C 2014, 118, 10606−1061510613

(14) Er, S.; Tiwari, D.; de Wijs, G. A.; Brocks, G. Tunable HydrogenStorage in Magnesium−Transition Metal Compounds: First-PrinciplesCalculations. Phys. Rev. B 2009, 79, 1−8.(15) Niessen, R. A. H.; Notten, P. H. L. Electrochemical HydrogenStorage Characteristics of Thin Film MgX (X = Sc , Ti , V , Cr)Compounds. Solid State Lett. 2005, 8, A534−A538.(16) De Jongh, P. E. Hydrogen storage: Keeping Out the Oxygen.Nat. Mater. 2011, 10, 265−266.(17) Yamauchi, M.; Ikeda, R.; Kitagawa, H.; Takata, M. NanosizeEffects on Hydrogen Storage in Palladium. J. Phys. Chem. C 2008, 112,3294−3300.(18) Mulder, F. M.; Singh, S.; Bolhuis, S.; Eijt, S. W. H. ExtendedSolubility Limits and Nanograin Refinement in Ti/Zr Fluoride-Catalyzed MgH2. J. Phys. Chem. C 2012, 116, 2001−2012.(19) Zhou, C.; Zak Fang, C.; Ren, C.; Li, J.; Lu, J. Effect of TiIntermetallic Catalysts on Hydrogen Storage Properties of MagnesiumHydride. J. Phys. Chem. C 2013, 117, 12973−12980.(20) Schimmel, H. G.; Huot, J.; Chapon, L. C.; Tichelaar, F. D.;Mulder, F. M. Hydrogen Cycling of Niobium and Vanadium CatalyzedNanostructured Magnesium. J. Am. Chem. Soc. 2005, 127, 14348−14354.(21) Kalisvaart, W. P.; Niessen, R. A. H.; Notten, P. H. L.Electrochemical Hydrogen Storage in MgSc alloys: A ComparativeStudy Between Thin Films and Bulk Materials. J. Alloys Compd. 2006,417, 280−291.(22) Tao, S. X.; Notten, P. H. L.; van Santen, R. A.; Jansen, A. P. J.DFT Studies of Hydrogen Storage Properties of Mg0.75Ti0.25. J. AlloysCompd. 2011, 509, 210−216.(23) Er, S.; van Setten, M. J.; de Wijs, G. A.; Brocks, G. First-Principles Modelling of Magnesium Titanium Hydrides. J. Phys.:Condens. Matter 2010, 22, 074208−1−13.(24) Pauw, B. R.; Kalisvaart, W. P.; Tao, S. X.; Koper, M. T. M.;Jansen, A. P. J.; Notten, P. H. L. Cubic MgH2 Stabilized by Alloyingwith Transition Metals: A Density Functional Theory Study. ActaMater. 2008, 56, 2948−2954.(25) Conradi, M. S.; Mendenhall, M. P.; Ivancic, T. M.; Carl, E. A.;Browning, C. D.; Notten, P. H. L.; Kalisvaart, W. P.; Magusin, P. C. M.M.; Bowman, R. C., Jr.; Hwang, S.; et al. NMR to Determine Rates ofMotion and Structures in Metal-Hydrides. J. Alloys Compd. 2007,446−447, 499−503.(26) Latroche, M.; Kalisvaart, P.; Notten, P. H. L. Crystal Structureof Mg0.65Sc0.35Dx Deuterides Studied by X-ray and Neutron PowderDiffraction. J. Solid State Chem. 2006, 179, 3024−3032.(27) Murray, J. L. The Mg-Ti(Magnesium Titanium) System. Bull.Alloy Phase Diagr. 1986, 7, 245−247.(28) de Boer, F. R.; Boom, R.; Mattens, W. C. M.; Miedema, A. R.;Niessen, A. K.; Cohesion in Metals: Transition Metal Alloys; North-Holland: Amsterdam, 1988.(29) Vermeulen, P.; Niessen, R. A. H.; Notten, P. H. L. HydrogenStorage in Metastable MgyTi(1−y) Thin Films. Electrochem. Commun.2006, 8, 27−32.(30) Vermeulen, P.; Wondergem, H. J.; Graat, P. C. J.; Borsa, D. M.;Schreuders, H.; Dam, B.; Griessen, R.; Notten, P. H. L. In situElectrochemical XRD Study of (De)Hydrogenation of MgyTi(100−y)Thin Films. J. Mater. Chem. 2008, 18, 3680−3687.(31) Liang, G.; Schulz, R. Synthesis of Mg-Ti Alloy by MechanicalAlloying. J. Mater. Sci. 2003, 38, 1179−1184.(32) Rousselot, S.; Bichat, M. P.; Guay, D.; Roue, L. Structure andElectrochemical Behaviour of Metastable Mg50Ti50 Alloy Prepared byBall Milling. J. Power Sources 2008, 175, 621−624.(33) Huot, J.; Ravnsbæk, D. B.; Zhang, J.; Cuevas, F.; Latroche, M.;Jensen, T. R. Mechanochemical Synthesis of Hydrogen StorageMaterials. Prog. Mater. Sci. 2013, 58, 30−75.(34) Asano, K.; Enoki, H.; Akiba, E. Synthesis Process of Mg−TiBCC Alloys by Means of Ball Milling. J. Alloys Compd. 2009, 486,115−123.(35) Asano, K.; Enoki, H.; Akiba, E. Synthesis of Mg−Ti FCCHydrides from Mg−Ti BCC Alloys. J. Alloys Compd. 2009, 478, 117−120.

(36) Srinivasan, S.; Magusin, P. C. M. M.; Kalisvaart, W. P.; Notten,P. H. L.; Cuevas, F.; Latroche, M.; van Santen, R. A. Nanostructures ofMg0.65Ti0.35Dx Studied with X-Ray Diffraction, Neutron Diffraction,and Magic-Angle-Spinning 2H NMR Spectroscopy. Phys. Rev. B 2010,81, 054107−1−10.(37) Manivasagam, T. G.; Magusin, P. C. M. M.; Srinivasan, S.;Krishnan, G.; Kooi, B. J.; Notten, P. H. L. Electrochemical Deuterationof Metastable MgTi Alloys: An Effective Way to Inhibit PhaseSegregation. Adv. Energy. Mater. 2014, 4, 1300590-1-7.(38) Rousselot, S.; Guay, D.; Roue, L. Comparative Study on theStructure and Electrochemical Hydriding Properties of MgTi,Mg0.5Ni0.5Ti and MgTi0.5Ni0.5 Alloys Prepared by High Energy BallMilling. J. Power Sources 2011, 196, 1561−1568.(39) Miedema, A. R.; Buschow, K. H. J.; Vanmal, H. H. WhichIntermetallic Compounds of Transition Metals form Stable Hydrides ?J. Less-Common Met. 1976, 49, 463−472.(40) Vermeulen, P.; van Thiel, E. F. M. J.; Notten, P. H. L. TernaryMgTiX-Alloys: A Promising Route towards Low-Temperature, High-Capacity, Hydrogen-Storage Materials. Chem.Eur. J. 2007, 13,9892−9898.(41) Joon Choi, Y.; Lu, J.; Yong Sohn, H.; Zak Fang, Z.; Kim, C.;Bowman, R. C., Jr.; Hwang, S. J. Reaction Mechanisms in the Li3AlH6/LiBH4 and Al/LiBH4 Systems for Reversible Hydrogen Storage. Part2: Solid-State NMR Studies. J. Phys. Chem. C 2011, 115, 6048−6056.(42) Grzech, A.; Lafont, U.; Magusin, P. C. M. M.; Mulder, F. M.Microscopic Study of TiF3 as Hydrogen Storage Catalyst for MgH2. J.Phys. Chem. C 2012, 116, 26027−26035.(43) Kim, C.; Hwang, S. J.; Bowman, R. C., Jr.; Reiter, J. W.; Zan, J.A.; Kulleck, J. G.; Kabbour, H.; Majzoub, E. H.; Ozolins, V. J.LiSc(BH4)4 as a Hydrogen Storage Material: Multinuclear High-Resolution Solid-State NMR and First-Principles Density FunctionalTheory Studies. J. Phys. Chem. C 2009, 113, 9956−9968.(44) Verkuijlen, M. H. W.; Gao, J.; Adelhelm, P.; van Bentum; deJongh, P. E.; Kentgens, A. P. M. Solid-State NMR Studies of the LocalStructure of NaAlH4/C Nanocomposites at Different Stages ofHydrogen Desorption and Rehydrogenation. J. Phys. Chem. C 2010,114, 4683−4692.(45) Adolphi, N. L.; Balbach, J. J.; Conradi, M. S.; Market, J. T.;Cotts, R. M.; Vajda, P. Deuterium Site Occupancy in YDx by Magic-Angle-Spinning NMR. Phys. Rev. B 1996, 53, 15054−15062.(46) Adolphi, N. L.; Badola, S.; Browder, L. A.; Bowman, R. C., Jr.Magic Angle Spinning NMR Study of Deuterium site Occupancy andDynamics in ZrNiD1.0 and ZrNiD3.0. Phys. Rev. B 2001, 65, 024301;Erratum: . Phys. Rev. B 2001, 69, 149901E.(47) Rousselot, S.; Gazeau, A.; Guay, D.; Roue, L. Influence of Pd onthe Structure and Electrochemical Hydrogen Storage Properties ofMg50Ti50 Alloy Prepared by Ball Milling. Electrochim. Acta 2010, 55,611−619.(48) Jeener, J.; Meider, B. H.; Bachman, P.; Ernst, R. R. Investigationof Exchange Processes by Two-Dimensional NMR Spectroscopy. J.Chem. Phys. 1979, 71, 4546−4553.(49) Morris, G. A.; Freeman, R. Selective Excitation in FourierTransform Nuclear Magnetic Resonance. J. Magn. Reson. 1978, 29,433−462.(50) Srinivasan, S.; Magusin, P. C. M. M.; van Santen, R. A.; Notten,P. H. L.; Schreuders, H.; Dam, B. Siting and Mobility of DeuteriumAbsorbed in Co-sputtered Mg0.65Ti0.35. A MAS 2H NMR Study. J. Phys.Chem. C 2011, 115, 288−297.(51) Magusin, P. C. M. M.; Kalisvaart, W. P.; Notten, P. H. L.; vanSanten, R. A. Hydrogen Sites and Dynamics in Light-WeightHydrogen-Storage Material Magnesium−Scandium Hydride Inves-tigated with 1H and 2H NMR. Chem. Phys. Lett. 2008, 456, 55−58.(52) Hayashi, S.; Nowak, B.; Orimo, S.; Fujii, H. I H NMR andMagnetization Measurements of a Nanostructured Composite Ma-terial of the Mg2Ni-H System Synthesized by Reactive MechanicalGrinding. J. Alloys Compd. 1997, 256, 159−165.(53) Bowman, R. C.; Adolphi, N. L.; Hwang, S.; Kulleck, J. G.;Udovic, T. J.; Huang, Q.; Wu, H. Deuterium Site Occupancy and

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp500535q | J. Phys. Chem. C 2014, 118, 10606−1061510614

Phase Boundaries in ZrNiDx (0.87 ≤ x ≤ 3.0). Phys. Rev. B 2006, 74,184109.(54) Mason, D. R.; Rudd, R. E.; Sutton, A. P. Stochastic KineticMonte Carlo Algorithms for Long-Range Hamiltonians. Comput. Phys.Commun. 2004, 160, 140−157.(55) Rudd, R. E.; Mason, D. R.; Sutton, A. P. Lanczos and RecursionTechniques for Multiscale Kinetic Monte Carlo Simulations. Progr.Mater. Sci. 2007, 52, 319−332.(56) Jansen, A. P. J. An Introduction to Kinetic Monte Carlo Simulationsof Surface Reactions; Lecture Notes in Physics 856; Springer: Berlin,2012; Chapter 8, p 254.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp500535q | J. Phys. Chem. C 2014, 118, 10606−1061510615