TiD2: Deuterium NMR - The

Mar 18, 2015 - RCB Hydrides, LLC, 117 Miami Avenue, Franklin, Ohio 45005, United States. ∥ Department of Metallurgical Engineering, University of Ut...
3 downloads 14 Views 357KB Size
Article pubs.acs.org/JPCC

Detection of Fluorite-Structured MgD2/TiD2: Deuterium NMR Samuel B. Emery,† Eric G. Sorte,†,‡ Robert C. Bowman, Jr.,§ Z. Zak Fang,∥ Chai Ren,∥ Eric H. Majzoub,⊥ and Mark S. Conradi*,† †

Department of Physics, Washington University, One Brookings Drive, St. Louis, Missouri 63130, United States Now at Department of Chemistry, Georgetown University, 37th and O Streets NW, Washington, DC, 20057, United States § RCB Hydrides, LLC, 117 Miami Avenue, Franklin, Ohio 45005, United States ∥ Department of Metallurgical Engineering, University of Utah, 135 South 1460 East, Room 412, Salt Lake City, Utah 84112-0114, United States ⊥ Center for Nanoscience and Department of Physics and Astronomy, University of Missouri-St. Louis, One University Boulevard, St. Louis, Missouri 63121, United States ‡

ABSTRACT: Deuterium magic-angle spinning NMR spectra of MgD2/TiD2 composites are presented. After high-energy, high-pressure ball-milling of an 80:20 Mg:Ti hydride, the material was H → D exchanged using high-pressure D2 gas. This material displayed the deuterium NMR spinning sideband pattern of the MgD2 component that is typical of rutile-structured MgD2. After heat treatment at 450 °C under high D2 pressure, the sidebands became much weaker, indicating that most D atoms were now in sites of local cubic symmetry. This is in agreement with a previous proposal of a nearly fully Mg−Ti segregated, lattice coherent fluorite (fcc) structure of the composite. Because H(D) diffusion is expected to be more rapid in the fluorite structure, compared to the rutile form, this may be important to the kinetics of hydrogen storage and release from MgH2/TiH2 composites.



INTRODUCTION Notten and co-workers developed Mg−Sc alloys and their hydrides for service as metal-hydride battery electrodes.1−3 These alloys displayed much greater hydrogen capacities per unit mass than existing LaNi5Hx-derived electrodes. Provided the Sc fraction x (in Mg(1−x)Scx or its hydride) exceeded x = 0.2, the alloy showed much faster electrochemical kinetics and had the fluorite (cubic) crystal structure.2,3 A hydrogen NMR study4 confirmed this picture, finding a slow hydrogen atomic hopping rate ωH in ordinary5 rutile-structured6 MgH2, with a large activation energy. By comparison, fluorite-structured Mg0.65Sc0.35H2.2 showed a much larger ωH for all the hydrogen atoms at a given temperature and a correspondingly smaller activation energy for the diffusive hopping motion.4 Thus, it appears that faster H diffusion in MgH2 will result from conditions yielding f luorite MgH2 (i.e., same crystal structure as CaF2). The high cost of Sc makes it unsuitable for large-scale applications. Thus, attention turned to Mg−Ti composites;1,7 unfortunately, Ti and Mg are not mutually soluble in bulk. However, thin f ilms of Mg(1−x)-Tix and their hydrides (i.e., (MgH2)1−x-(TiH2)x) show intriguing behavior.8,9 An important study found their EXAFS data and the optical−frequency dielectric properties to be consistent with a nearly fully segregated composite of Ti or TiH2 nodules within a Mg or MgH2 matrix.9 Crucially, for the X-ray diffraction of this material, the Ti/TiH2 nodules are lattice coherent with the © 2015 American Chemical Society

matrix, due to a near-match of lattice parameters of the constituents.9 For the hydride composite thin films, both the MgH2 and TiH2 are in the fluorite structure. For bulk MgH2−TiH2 composites, the situation is less clear. Yet, it is the bulk composites that are most relevant to hydrogen storage at all but the smallest scales. Although there is some limited theoretical10−14 and experimental15−18 evidence suggesting the existence of stable fluorite-crystal-structured MgH2 under special circumstances, no clear proof of fluorite MgH2 has yet been reported in bulk material. We report here deuterium (2D) magic-angle spinning NMR of MgD2-TiD2 bulk composites. For heavily heat-treated bulk material, the MgD2 is observed to have D atoms in sites of nearly cubic symmetry, consistent with the fully segregated, lattice-coherent fluorite structure first observed in thin films.9



MAS NMR

Magic-angle spinning (MAS) NMR involves rotation of the sample at a high frequency fs (2−100 kHz) about an axis inclined to the static magnetic field B0 by angle θ = 54.7° (chosen to satisfy 3cos2θ − 1 = 0). Provided the spinning is sufficiently fast compared to the line-broadening, any linebroadening interaction with this angular dependence will timeReceived: February 10, 2015 Revised: March 17, 2015 Published: March 18, 2015 7656

DOI: 10.1021/acs.jpcc.5b01409 J. Phys. Chem. C 2015, 119, 7656−7661

Article

The Journal of Physical Chemistry C average to zero.19−21 These include spin−spin dipolar interactions, electrical quadrupole interactions, anisotropic parts of chemical and Knight shifts, and the effects of magnetic susceptibility of the sample (provided the susceptibility itself is isotropic, which is generally the case). In metal hydrides, the large (∼40 kHz) H−H dipole coupling requires very fast spinning to completely average to zero. But in metal deuterides, the D−D dipole strength is approximately 20 times smaller, so that excellent line narrowing results from spinning as slow as fs = 3 kHz. Deuterium MAS NMR has been used to study YD2+x;22 ZrNiDx,23−25 Mg(1−y)ScyD2.2;26 and various Mg−Ti−D composites.27 In each case, the line narrowing resulting from MAS allowed the identification of distinct, chemically inequivalent D atoms. For MgD2 in the rutile structure,6 each D atom sits in a site of noncubic symmetry, surrounded by three nearest neighbor Mg atoms. Thus, the electric field gradient (EFG) at the D atom does not vanish. The EFG interacts with the deuteron’s electrical quadrupole moment, producing the quadrupole interaction. In rutile MgD2, the quadrupole interaction is by far the largest line broadening interaction.15,16,27 Under MAS, the quadrupole interaction shows itself as sharp spinning sidebands, resonance lines spaced from the centerband by multiples (±1, ±2, ...) of the spinning frequency fs. The creation of sharp sidebands, even when the spinning frequency is smaller than the interaction strength, is characteristic of inhomogeneous broadening interactions.28 The MAS 2D NMR spectrum of coarse-grain MgD2 is presented in Figure 1a, with fs = 9110 Hz. The glitch evident at 3.5 kHz is a baseline artifact (the spectrum was taken off-resonance and shifted back for display in Figure 1a). By contrast, MgD2 in the proposed fluorite structure has each D surrounded by four nearest neighbor metal atoms with tetrahedral (cubic) symmetry. By arguments of symmetry, the

EFG and the quadrupole interaction vanish in this locally cubic structure. We note that “locally cubic” implies a cubic arrangement of atoms (e.g., tetrahedral or octahedral) as well as no random disorder in metal-atom siting. That is, all neighbors are Mg, with no randomly occurring Ti atoms, or vice versa. The spinning sidebands in such a structure are expected to have zero or small amplitude from the quadrupole interaction. We note that, even in the absence of quadrupole couplings, other weaker anisotropic interactions such as spin− spin dipolar will generate comparatively weak spinning sidebands.



EXPERIMENTAL SECTION Samples. Samples of MgH2−TiH2 were prepared at the University of Utah with a batch size of 2 g and a 4:1 MgH2 to TiH2 ratio by moles. A custom-made ultrahigh-energy highpressure (UHEHP) planetary milling machine was used for the milling process. Each sample was loaded into a custom-made stainless steel milling canister with a brass sleeve inside to eliminate ferrous contamination. Yttrium-stabilized zirconia balls with a ball-to-powder ratio of 20:1 by volume were used and the powder was milled under 50 bar hydrogen pressure for 4 h. The sample powder was H → D exchanged at 230 °C under a D2 pressure of 100 bar over a period of 48 h. For samples without Ti, a higher temperature of 350 °C was needed for exchange. On occasion, the exchange procedure was repeated with a fresh quantity of D2 gas. Heating of the stainless steel pressure vessel was by a temperature regulated electrical heater; the sample powder was held within by an open glass tube. Eventually, some samples were further heat-treated, being held for 24 h at 450 °C under 90 bar of D2 gas (typical conditions). NMR. After H → D exchange, the sample was measured by hydrogen NMR at 2.0 T (85.03 MHz). The hydrogen signal amplitude was compared to that of unexchanged MgH2. The ratio of signal amplitudes gave the fraction of H atoms remaining in the exchanged sample: the reduction in hydrogen NMR line width gave qualitative confirmation of the exchange. The exchange fraction is not critical, but a smaller fraction of remaining H means a larger 2D signal and smaller spin−spin dipole interactions, for better MAS line-narrowing. Exchange down to 10% remaining H, as typically obtained here, is better than needed for effective 2D MAS NMR line narrowing. MAS 2D NMR used a Chemagnetics probe with a 5 mm outer diameter pencil rotor, working at 7.04 T (2D frequency 46.001 MHz). All the MAS measurements were at 22 °C. An optical tachometer measured the spinning frequency fs, which was typically stable to ±10 Hz over the duration of data acquisition. Thus, drift of fs contributed negligibly to the line widths of the spinning sidebands. The NMR spectrometer was home constructed. RF pulses of π/2 nutations were 3 μs long. Acoustic coil disease was suppressed using the blinking-180 approach,29 in which alternate acquisitions are preceded by a π rf pulse (and the acquired data inverted). The sequence is 180x/null−15 ms−90x− acquire∓. The delay of 15 ms was chosen to be short compared to T1, but longer than both the duration of the coil disease signal and the duration of any NMR FID that could be generated by an imperfect π pulse. TGA. Thermogravimetric Analysis (TGA) was utilized to characterize all the samples after they had undergone UHEHP ball milling using a Shimadzu TGA50 inside an argon atmosphere glovebox. While recording the sample mass, the

Figure 1. 2D MAS (ν = 46.001 MHz, fs = 9.11 kHz) spectra for (a) coarse-grain MgH2 that has undergone H → D exchange at 350 °C under 100 bar D2 pressure, (b) MgH2−TiH2 ultrahigh-energy, highpressure ball-milled (UHEHP) and H → D exchanged at 230 °C under 100 bar D2 pressure, and (c) the material of panel b after undergoing a heat treatment at 450 °C under 90 bar D2 pressure. When compared to panels a and b, panel c shows significant reduction in the height of the MgD2 spinning sidebands (see Table 1). 7657

DOI: 10.1021/acs.jpcc.5b01409 J. Phys. Chem. C 2015, 119, 7656−7661

Article

The Journal of Physical Chemistry C sample was dehydrided into an inert carrier stream (50 mL/ min Ar), with the temperature being ramped 5 °C/min until reaching a maximum temperature of 400 °C. Generally, the weight loss from dehydriding was completed by 360 °C. XRD. X-ray diffraction spectra were collected using a Rigaku Ultima-IV in Bragg−Brentano geometry. All samples were handled in an Ar-filled glovebox with oxygen and water levels below 1 ppm. Sample powders were loaded onto glass slides and protected from air contact with a mylar film. XRD spectra were collected using a step size of 0.02° 2θ, at a scan rate of 0.4°/min.

Table 1. Relative Sideband Amplitudes for the Spectra of Figures 1 and 2

RESULTS AND DISCUSSION Coarse-Grain MgD2. The 2D MAS NMR spectrum of coarse-grain (rutile) MgD2 appears in Figure 1a and consists of a centerband near relative frequency 0 Hz and spinning sidebands. The sidebands are separated from the centerband by ±fs, ±2fs, ... with fs = 9110 Hz. Slower spinning (e.g., 5000 Hz) resulted in similar spectra but with the expected reduction in sideband spacing. This spectrum is in excellent agreement with spectra published by P. Magusin and co-workers at similar values of fs.15,16,27 In short, for rutile MgD2, the first sidebands at ±fs are nearly equal in amplitude to the centerband; this serves as a qualitative measure of the strength of the quadrupole interaction here. MgD2−TiD2. After H → D exchange of our ball-milled MgH2−TiH2 composite with 4:1 Mg:Ti, the spectrum of Figure 1b is obtained. Here, the MgD2 signal consists of a centerband plus spinning sidebands and is similar to the pure MgD2 spectrum of Figure 1a. For the as-exchanged composite, the sidebands are slightly weaker; this may signal a partial structural change or that D atoms near the MgD2−TiD2 interfaces have smaller quadrupole interactions. The peak near −7200 Hz (−156 ppm) is the well-known15,16,27 Knight-shifted (metallic) TiD2 resonance.20 Its frequency shift agrees with previous work. We note that the asymmetric TiD2 peak shape also appears in previous studies.15,16,27 We believe that the lowest frequency (to the right) part of the peak is from large, nearly bulk-like TiD2 regions, whereas the less shifted components (the sweeping tail to the left) is from smaller TiD2 regions, which we believe have less fully developed metallic character and smaller magnitude Knight shifts. The sample of Figure 1b was subsequently heat treated at 450 °C under 90 bar D2 pressure for 48 h. The spectrum of this material appears in Figure 1c. There, the TiD2 peak remains essentially unchanged, aside from its amplitude. This indicates that MgD2 has not alloyed with the TiD2 because such alloying would reduce the metallic character and magnitude of the Knight shift of the “TiD2” resonance. Additionally, the long T1 and absence of a Knight shift for the “MgD2” peaks indicate that the MgD2 in our MgD2−TiD2 composite (both after the H → D exchange and after the heat treatment) is not a metal. The MgD2 peak in Figure 1c has much smaller sidebands; for example, the ratio of the first sideband (±fs) amplitude to the centerband is much reduced from Figure 1a and b. This is the key finding of the present work: a form of MgD2 with greatly reduced sidebands, implying a form of MgD2 with D atoms in sites of nearly cubic symmetry. For all of the presented spectra, the relative sideband amplitudes are given in Table 1. The average amplitude of the (±fs) sidebands compared to the centerband is given, as is the ratio for the second sidebands (at ±2fs) to the centerband, and the ratio of second to first sidebands.

This apparent structural change for the MgD2 should not be confused with a transition to the high-pressure phase γ-MgH2. Numerous studies suggest that γ-MgH2 does not survive heating past 300 °C except under extreme pressures.30−32 Most importantly, the α-PbO2 type crystal structure (space group Pbcn) is reported for γ-MgH2.30−32 This orthorhombic lattice has D atoms coordinated with three nearest neighbor Mg atoms, so there will be a substantial EFG at the deuterium site in the γ-structure, quite different from our observations here, with low or zero EFG (fluorite). MgD2 without TiD2. A similar sample of MgD2, but without TiD2, was prepared by the same route of high-energy, high-pressure ball-milling, H → D exchange (but at 350 °C), and eventual heat treatment at 450 °C. The spectra are displayed in Figure 2b and c, along with coarse-grain MgD2 in

sample Coarse-grain MgD2 MgD2−TiD2, after exchange MgD2−TiD2 after heat treat MgD2, after exchange MgD2, after heat treat



figure

±fs/ Centerband

±2fs/ Centerband

±2fs/ ±fs

1a, 2a 1b

1.01 0.66

0.42 0.23

0.42 0.37

1c

0.22

0.09

0.42

2b 2c

0.87 0.73

0.35 0.28

0.39 0.39

Figure 2. 2D MAS (ν = 46.001 MHz, fs = 9.11 kHz) spectra for (a) the same coarse-grain MgH2 that has undergone H → D exchange at 350 °C under 100 bar D2 pressure as presented in Figure 1a, (b) MgH2 UHEHP ball-milled and H → D exchanged at 350 °C under 100 bar D2 pressure, and (c) the material of panel b after undergoing a heat treatment at 450 °C under 90 bar D2 pressure. When compared to panel b, panel c shows only small changes in the MgD2 spectrum after undergoing heat treatment (see Table 1 for comparison of relative spinning sideband peak heights for the spectra presented).

Figure 2a (same as in Figure 1a). The similarity of the sideband patterns indicates that this material without Ti remains in the rutile structure, with or without heat treatment. The fluorite structure of MgD2 is not expected without Ti, and it is not observed. We caution that the absence of the abrasive TiH2 during milling may also result in less complete milling.31,33 7658

DOI: 10.1021/acs.jpcc.5b01409 J. Phys. Chem. C 2015, 119, 7656−7661

Article

The Journal of Physical Chemistry C Relative Intensities. An important issue is the change of relative intensities (spectral areas) of MgD2 and TiD2. Figure 1b shows material H → D exchanged but not further heattreated. The relative area of MgD2 lines (centerband plus all visible sidebands) to the single somewhat broadened TiD2 line is approximately 3.1:1, close to the 4:1 expected from the original material loaded into the ball mill. We note that TiD2 shows15,16,27 vanishing spinning sidebands because it is cubicstructured, or cubic with only slight distortion.34 But Figure 1c, from material after heat treatment at 450 °C, shows much less relative Mg content, with MgD2:TiD2 about 0.5:1, based on the deuterium signals. We believe the explanation for the decrease in relative amount of MgD2 (as well as decrease in the absolute amount; note the decrease in MgD2 signal-to-noise) lies in the brass (roughly CuZn) dust created by use of the brass sleeve inside the milling vessel (see Experimental Section, above). According to Reilly and Wiswall,35 Cu reacts with MgH2 according to36,37 (n + 2)MgH2 + Cu ↔ Mg 2Cu + n MgH2 + 2H 2

Figure 3. Temperature and D2 pressures (blue symbols) for the various H → D exchanges and heat treatments undergone by the samples presented in Figures 1 and 2. In addition the equilibrium P vs T curves for the Mg−D system5,38 (black) and the Mg−Cu−H system35 (red) are shown; see reactions 1 and 2

(1)

The plateau pressure for this reaction is greater than for MgH2 ↔ Mg + H 2

(2)

In fact, from the presented data,35 we have extrapolated using the general form P = P0e−H / kT

reaction was blocked by the applied pressure. Thus, reaction 1 may play a substantial role in our results. More Distant Sidebands. It is interesting to examine the ratio of amplitudes of the ±2 sidebands to the ±1 sidebands in Table1. This ratio should decrease from Figure 1a (coarse-grain MgD2) to Figure 1c (which we have interpreted as fluorite MgD2), if the quadrupole interaction simply weakened uniformly. But the right-hand column of Table 1 shows that the amplitude ratio ±2:±1 remains nearly unchanged. The implication is clear−the spectrum in Figure 1c is a superposition of two signals: One signal has essentially only the centerband without sidebands (fluorite MgD2 component) and the other signal has the centerband and sidebands of rutile MgD2, as in Figure 1a. The heat treated sample of Figure 1c is thus a smaller amount of rutile MgD2 with a larger amount of fluorite MgD2. Hence, with growing fluorite fraction, the amplitude ratios of sidebands to centerband decrease while the ±2:±1 ratio remains constant, as observed. In this light, the asexchanged sample of Figure 1b may already have a small fraction of fluorite-structured MgD2, a yet smaller fraction of the fluorite phase may be present in the ball-milled samples of Figure 2b and c (starting with MgH2 only: exchanged and exchanged then heated treated, respectively). D2 Gas Signal. We note as a warning that the exchanged and heat treated material has a small amount of D2 gas trapped in it (contributing about 3% of the total D2 centerband signal), evidently in some kind of pores.42 At very short pulse sequence recycle delays, this D2 gas is evident, because the gas has a very short T1 (a few ms to 1 s) compared to the MgD2 (100 s or more). The D2 gas signal has no quadrupolar or dipolar static interactions, so it is devoid of spinning sidebands. Thus, it is important to acquire data at long recycle delays (as done here) where the MgD2 signal is fully recovered and, in particular, overwhelms the D2 gas signal. We note that the sample with 2D MAS spectrum in Figure 1c exhibited a long T1, on the order of tens to hundreds of seconds (compare to D2 gas with T1 less than 1 s). In addition, D2 gas exhibits in our samples a small shift of the center frequency and is narrower than the MAS centerband of MgD2. Thus, we can

(3)

estimating the plateau pressure of reaction 1 to be 150 bar at our heat treatment temperature of 450 °C. We note that this extrapolation will be somewhat inexact from uncertainties in the original data35 and because the hydride and deuteride systems have small, but nonzero, differences (isotope effect). Because reaction 1 should proceed left to right under our conditions (450 °C, 90 bar D2) the result of a substantial brass impurity should be Mg2Cu (note: invisible to 2D NMR) and less MgD2, as some Mg is tied up now as Mg2Cu. Indeed, TGA analysis of the as-ball-milled MgH2−TiH2 material showed a TGA signal from dehydriding approximately half as large (3%) as the 6% signal for material milled without the brass liner. That is, our material had substantial brass dust in it, reducing the TGA-recorded mass fraction of H residing as MgH2. The temperatures and pressures for the H → D exchanges and heat treatments of the samples with spectra in Figures 1 and 2 are plotted along with the equilibrium P vs T curves for both reactions 1 and 2 in Figure 3 (for reaction 2 the deuterium values are presented).5,35,38 We note that Zn reacts with MgH2 in a fashion similar to Cu; its equilibrium pressure line39 (not shown) is close to that of Mg−Cu−H (red line in Figure 3). It may be the case that our fluorite−MgD2 forms only in the presence of (relatively) extra TiD2. That is, the decrease in MgD2:TiD2 by reaction of some MgD2 with Cu may be beneficial in forming fluorite MgD2. After all, a greater relative amount of TiD2 (fluorite structured) should increase its influence over the structure of the MgD2.12,40,41 This effect, dependent on brass content to decrease the relative amount of MgD2 according to reaction 1, may also explain the absence of fluorite MgD2 for MgD2−TiD2 samples heat treated at lower temperatures: We were unsuccessful in generating the fluorite MgD2 (as demonstrated by the relative amplitudes of the sidebands) when we heat treated at only 400 °C. There, the applied pressure of 90 bar D2 (open triangle in Figure 3) exceeds the equilibrium pressure of 60 bar for reaction 1, so the 7659

DOI: 10.1021/acs.jpcc.5b01409 J. Phys. Chem. C 2015, 119, 7656−7661

Article

The Journal of Physical Chemistry C

nearly zero quadrupole interactions. In turn, this implies a nearly cubic local environment of most of the D atoms. “Locally cubic” indicates two things here: (1) a cubic crystal structure, such as fluorite, and (2) no metal atom disorder. This last implication means nearly complete Mg−Ti segregation (random Mg vs Ti occupation of metal sites would destroy the locally cubic nature of the D site). The nearly complete Mg−Ti segregation is confirmed by the invariance of the TiD2 resonance position upon heat treatment. Dissolution of MgD2 into the TiD2 would be expected to reduce the magnitude of the Knight frequency shift, but this is not observed. All of these observations are consistent with the proposal (from MgH2−TiH2 thin films) of a single, lattice coherent, fluorite structure for the two-component composite.9 Additional experimental techniques, such as inelastic neutron scattering,43 should confirm these results and aid in the understanding of this complex system. We thank an anonymous reviewer for bringing this to our attention. The formation of the fluorite MgD2 in the MgD2−TiD2 composite may involve the brass (CuZn) contamination present from our ball milling apparatus. Reaction with Cu should remove MgH2 from the sample, locking-up Mg in the form of Mg2Cu, which does not hydride. Zn from the brass is expected to behave similarly. The resulting reduced ratio of MgD2:TiD2 may increase the structural effect of the fluorite TiD2 on the remaining MgD2. We do not speculate on the geometry of contact between MgD2 and TiD2. The observation of fluorite structured MgD2 may have important implications for the kinetics of dehydriding and rehydriding, as a fluorite structure is believed to exhibit faster diffusion4 of H or D.

readily distinguish the signals of D2 gas and MgD2, allowing us to conclude that the spectrum of Figure 1c had negligible D2 gas content. Additional Materials. All samples from the first MgH2− TiH2 batch behaved consistently as described above. That is, upon heat treatment to 450 °C, these samples all displayed a strong reduction of the MgD2 spinning sideband amplitudes in the deuterium MAS NMR, corresponding to a fluoritestructured MgD2. We have also tried other materials. A second batch of MgD2TiD2 generated by starting with high-energy, high-pressure ball milling (the route detailed in the Experimental Section above) did not reveal fluorite MgD2 after heat treatment at 450 °C (there was no diminution of spinning sideband amplitudes). Less aggressive mechanical treatment (ball-milling of 80 mol % MgH2−20 mol % TiH2 under 1 bar argon for short time) with WC balls inside a WC vessel (so, no brass present) followed by H → D exchange and 450 °C heat treatment also showed sidebands matching those of rutile MgD2 (so, no fluorite MgD2). TGA analysis of the second batch showed a decrease (but not elimination) in the brass contamination, compared to the first batch. Essentially, any brass present reduces the mass fraction of hydrogen released on heating. Samples from the first batch showed a strong reduction in the 2D MAS NMR intensity of MgD2:TiD2 (see Figure 1b and c) upon heat treatment, from 3.1:1 to 0.5:1, due to the reaction of MgD2 with Cu or Zn via reactions such as (1). The second batch, which did not yield fluorite MgD2, showed a milder reduction from 3.2:1 after the H → D exchange step to 2.2:1 after heat treatment. This result confirms the presence of less brass in the second batch. The less aggressively milled sample had an intensity of approximately 3.5:1 both after the H → D exchange and after heat treatment at 450 °C. These results confirm the role of brass through reactions such as (1) in consuming Mg in our system. X-ray diffraction (XRD) characterization of samples from both Utah batches was generally inconclusive. These samples were heavily milled resulting in small particle size and thus broad reflections in the XRD. Additionally, the presence of impurities (brass) complicates the diffraction pattern. However, reflections corresponding to various CuZn phases are found in the XRD patterns of the as-received material. Upon H → D exchange at 230 °C additional reflections assigned to MgCu2 and MgCuZn are observed, whereas the CuZn signal decreases noticeably. After further heat treatment to 450 °C, the reflection intensities of these nonhydride Mg phases increase and the CuZn signal is nearly absent, verifying the consumption of Mg (from MgH2) and conversion to nonhydride-bearing compounds. It is expected that a greater relative amount of TiD2 (fluorite structured) should increase its influence over the structure of the MgD2.12,40,41 This may explain the role of the brass impurity in promoting formation of fluorite MgD2 and explain why it was not seen in material with limited or absent contamination. These results suggest that preparation of MgH2−TiH2 composites with MgH2:TiH2 ≤ 1 might yield fluorite structured MgH2 in the absence of the brass impurity and possibly without the high-temperature, high-pressure heat treatment. We are pursuing this path.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The work at WU and UMSL was supported in part by U.S. DOE Basic Sciences grant DE-FG02-ER46256.The work at Utah was supported by the U.S. Department of Energy (DOE) under contract number DE-AR0000173 and National Science Foundation (Grant No. 0933778). We also thank C. Klug at the Naval Research Laboratory for high-speed proton MAS NMR.



REFERENCES

(1) Notten, P. H. L.; Ouwerkerk, M.; van Hal, H.; Beelen, D.; Keur, W.; Zhou, J.; Feil, H. High Energy Density Strategies: from HydrideForming Materials Research to Battery Integration. J. Power Sources 2004, 129, 45−54. (2) Niessen, R. A. H.; Notten, P. H. L. Hydrogen Storage in Thin Films Magnesium-Scandium Alloys. J. Alloys Compd. 2005, 404 − 406, 457−460. (3) Niessen, R. A. H.; Notten, P. H. L. Electrochemical Hydrogen Storage Characteristics of Thin Film MgX (X = Sc, Ti, V, Cr) Compounds. Electrochem. Solid-State Lett. 2005, 8, A534−A538. (4) 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. C. M.; Bowman, R. C., Jr.; Huang, S.-J.; et al. NMR to Determine Rates of Motion and Structure in Metal-Hydrides. J. Alloys Compd. 2007, 446 − 447, 499−503.



CONCLUSIONS Deuterium MAS NMR of MgD2-TiD2 composites provides evidence for D atoms in the MgD2 component with zero or 7660

DOI: 10.1021/acs.jpcc.5b01409 J. Phys. Chem. C 2015, 119, 7656−7661

Article

The Journal of Physical Chemistry C

(25) Adolphi, N. L.; Badola, S.; Browder, L. A.; Bowman, R. C., Jr. Erratum: Magic-Angle Spinning NMR Study of Deuterium Site Occupancy and Dynamics in ZrNiD1.0 and ZrNiD3.0. Phys. Rev. B 2004, 69, 149901(E). (26) Srinivasan, S.; Magusin, P. C. M. M. Lightweight Hydrogen Storage Material Mg0.65Sc0.35D2 Studied with 2H and 2H-{45Sc} MAS NMR Exchange Spectroscopy. Solid State Nucl. Magn. Reson. 2011, 39, 88−98. (27) Srinivasan, S.; Magusin, P.C.M.M; van Santen, R. A.; Notten, P. H. L.; Schreuders, H.; Dam, B. Siting and Mobility of Deuterium Absorbed in Cosputtered Mg0.65Ti0.35. A MAS 2H NMR Study. J. Phys. Chem. C 2011, 115, 288−297. (28) Poole, Jr., C. P.; Farach, H. A. Relaxation in Magnetic Resonance; Academic Press: New York, 1971; pp 44−46, 124−137. (29) Doverspike, M. A.; Liu, S.-B.; Ennis, P.; Johnson, T.; Conradi, M. S. NMR in High-Pressure Phases of Solid NH3 and ND3. Phys. Rev. B 1986, 33, 14−21. (30) Bortz, M.; Bertheville, B.; Bottiger, G.; Yvon, K. Structure of the High Pressure Phase γ-MgH2 by Neutron Powder Diffraction. J. Alloys Compd. 1999, 287, L4−L6. (31) Ponthieu, M.; Cuevas, F.; Fernandez, J. F.; Laversenne, L.; Porcher, F.; Latroche, M. Structural Properties and Reversible Deuterium Loading of MgD2−TiD2 Nanocomposites. J. Phys. Chem. C 2013, 117, 18851−18862. (32) Ren, C.; Fang, Z. Z.; Zhou, C.; Lu, J.; Ren, Y.; Zhang, X.; Luo, X. In Situ X-ray Diffraction Study of Dehydrogenation of MgH2 with Ti-Based Additives. Int. J. Hydrogen Energy 2014, 39, 5868−5873. (33) Cuevas, F.; Korablov, D.; Latroche, M. Synthesis, Structural and Hydrogenation Properties of Mg-Rich MgH2-TiH2 Nanocomposites Prepared by Reactive Ball Milling under Hydrogen Gas. Phys. Chem. Chem. Phys. 2012, 14, 1200−1211. (34) Mueller, W. M.; Blackledge, J. P.; Libowitz, G. G. Metal Hydrides; Academic Press: New York, 1968; pp 336 − 353. (35) Reilly, J. J.; Wiswal, R. H. The Reaction of Hydrogen with Alloys of Magnesium and Copper. Inorg. Chem. 1967, 6, 2220−2223. (36) Karty, J.; Genossar-Grunzweig, J.; Rudman, P. S. Hydriding and Dehydriding Kinetics of Mg in a Mg/Mg2Cu Eutectic Alloy: Pressure Sweep Method. J. Appl. Phys. 1979, 50, 7200−7209. (37) Andreasen, A.; Sorensen, M. B.; Burkarl, R.; M?ller, B.; Molenbroek, A. M.; Pedersen, A. S.; Vegge, T.; Jensen, T. R. Dehydrogenation Kinetics of Air-Exposed MgH2/Mg2Cu and MgH2/ MgCu2 Studied with In Situ X-ray Powder Diffraction. Appl. Phys. A: Mater. Sci. Process. 2006, 82, 515−521. (38) Leardini, F.; Ares, J. R.; Fernandez, J. F.; Bodega, J.; Sanchez, C. An Investigation on the Thermodynamics and Kinetics of Magnesium Hydride Decomposition Based on Isotope Effects. Int. J. Hydrogen Energy 2011, 36, 8351−8357. (39) Bruzzone, G.; Costa, G.; Ferretti, M.; Olcese, G. L. Hydrogen Storage in Mg51Zn20. Int. J. Hydrogen Energy 1983, 8, 459−461. (40) Asano, K.; Enoki, H.; Akiba, E. Synthesis of Mg-Ti FCC Hydrides from Mg-Ti BCC Alloys. J. Alloys Compd. 2009, 478m, 117− 120. (41) Asano, K.; Kim, H.; Sakaki, K.; Page, K.; Hayashi, S.; Nakamura, Y.; Akiba, E. Synthesis and Structural Study of Ti-rich Mg-Ti Hydrides. J. Alloys Compd. 2014, 593, 132−136. (42) Senadheera, L.; Carl, E. M.; Ivancic, T. M.; Conradi, M. S.; Bowman, R. C.; Hwang, S.-J.; Udovic, T. J. Molecular H2 Trapped in AlH3 solid. J. Alloys Compd. 2008, 463, 1−5. (43) Schimmel, H. G.; Johnson, M. R.; Kearley, G. J.; RamirezCuesta, A. J.; Huot, J.; Mulder, F. M. Structural Information on Bill Milled Magnesium Hydride from Vibrational Spectroscopy and Abinitio Calculations. J. Alloys Compd. 2005, 393, 1−4.

(5) Stampfer, J. F., Jr.; Holley, C. E., Jr.; Suttle, T. F. The Magnesium-Hydrogen System. J. Am. Chem. Soc. 1960, 82, 3504− 3528. (6) Ellinger, F. H.; Holley, C. E., Jr.; McInteer, B. B.; Pavone, D.; Potter, R. M.; Staritzky, E.; Zachariasen, W. H. The Preparation and Some Properties of Magnesium Hydride. J. Am. Chem. Soc. 1955, 77, 2647−2648. (7) Vermulen, P.; Niessen, R. A. H.; Notten, P. H. L. Hydrogen Storage in Metastable MgyTi(1−y) Thin Films. Electrochem. Commun. 2006, 8, 27−32. (8) Borsa, D. M.; Gremaud, R.; Baldi, A.; Schreduers, H.; Rector, J. H.; Kooi, B.; Vermeulen, P.; Notten, P. H. L.; Dam, B.; Griessen, R. Structural, Optical, and Electrical Properties of MgyTi1−yHx Thin Films. Phys. Rev. B 2007, 75, 205408. (9) Baldi, A.; Gremaud, R.; Borsa, D. M.; Balde, C. P.; van der Eerden, A. M. J.; Kruijtzer, G. L.; de Jongh, P. E.; Dam, B.; Griessen, R. Nanoscale Composition Modulations in MgyTi1−yHx Thin Film Alloys for Hydrogen Storage. Int. J. Hydrogen Energy 2009, 34, 1450−1457. (10) 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 Alloying with Transition Metals: A Density Functional Theory Study. Acta Mater. 2008, 56, 2948−2954. (11) Tao, S. X.; Notten, P. H. L.; van Santen, R. A.; Jansen, A. P. J. Density Functional Theory Studies of the Hydrogenation Properties of Mg and Ti. Phys. Rev. B 2009, 79, 144121. (12) Tao, S. X.; Notten, P. H. L.; van Santen, R. A.; Jansen, A. P. J. Fluorite Transition Metal Hydride Induced Destabilization of the MgH2 System in MgH2/TMH2 Multilayers (TM = Sc, Ti, V, Cr, Y, Zr, Nb, La, Hf). Phys. Rev. B 2010, 82, 125448. (13) Tao, S. X.; Kalisvaart, W. P.; Danaie, M.; Mitlin, D.; Notten, P. H. L.; van Santen, R. A.; Jansen, A. P. J. First Principle Study of Hydrogen Diffusion in Equilibrium Rutile, Rutile with Deformation Twins and Fluorite Polymorph of Mg Hydride. Int. J. Hydrogen Energy 2011, 36, 11802−11809. (14) 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. Alloys Compd. 2011, 509, 210−216. (15) Thirugnasambandam, G.; Magusin, P. C. M. M.; Srinivasan, S.; Krishnan, G.; Kooi, B. J.; Notten, P. H. L. Electrochemical Deuteration of Metastable MgTi Alloys: An Effective Way to Inhibit Phase Segregation. Adv. Energy Mater. 2014, 4, 1300590. (16) Srinivasan, S.; Magusin, P. C. M. M.; Kalisvaart, W. P.; Notten, P. H. L.; Cuevas, F.; Latroche, M.; van Santen, R. A. Nanostructures of Mg0.65Ti0.35Dx Studied with X-ray Diffraction, Neutron Diffraction, and Magic-Angle-Spinning 2H NMR Spectroscopy. Phys. Rev. B 2010, 81, 054107. (17) Leegwater, H.; Schut, H.; Egger, W.; Baldi, A.; Dam, B.; Eijt, S. W. H. Divacancies and the Hydrogenation of Mg−Ti Films with Short Range Chemical Order. Appl. Phys. Lett. 2010, 96, 121902. (18) Korablov, D.; Besenbacher, F.; Jensen, T. R. Ternary Compounds in the Magnesium−Titanium Hydrogen Storage System. Int. J. Hydrogen Energy 2014, 39, 9700−9708. (19) Mehring, M. High Resolution NMR Spectroscopy in Solids; Springer-Verlag: New York, 1976. (20) Slichter, C. P. Principles of Magnetic Resonance; Springer: New York, 1990. (21) Levitt, Malcom H. Spin Dynamics; Wiley: Chichester, U.K., 2008. (22) Adolphi, N. L.; Balbach, J. J.; Conradi, M. S.; Markert, J. T.; Colts, R. M.; Vajda, P. Deuterium Site Occupancy in YDx by MagicAngle Spinning NMR. Phys. Rev. B 1996, 53, 15054−15062. (23) Bowman, R. C., Jr.; Adolphi, N. L.; Hwang, S.-J.; Kulleck, J. G.; Udovic, T. J.; Huang, Q.; Wu, H. Deuterium Site Occupancy and Phase Boundaries in ZrNiDx (0.87 ≤ x ≤ 3.0). Phys. Rev. B 2006, 74, 184109. (24) Adolphi, N. L.; Badola, S.; Browder, L. A.; Bowman, R. C., Jr. Magic-Angle Spinning NMR Study of Deuterium Site Occupancy and Dynamics in ZrNiD1.0 and ZrNiD3.0. Phys. Rev. B 2001, 65, 024301. 7661

DOI: 10.1021/acs.jpcc.5b01409 J. Phys. Chem. C 2015, 119, 7656−7661