Article pubs.acs.org/JPCC
Dynamical Perturbations of Tetrahydroborate Anions in LiBH4 due to Nanoconfinement in Controlled-Pore Carbon Scaffolds Nina Verdal,*,†,‡ Terrence J. Udovic,† John J. Rush,†,‡ Xiangfeng Liu,§,∥ Eric H. Majzoub,§ John J. Vajo,⊥ and Adam F. Gross⊥ †
NIST Center for Neutron Research, National Institute of Standards and Technology, 100 Bureau Drive, Gaithersburg, Maryland 20899-6102, United States ‡ Department of Materials Science and Engineering, University of Maryland, College Park, Maryland 20742-2115, United States § Center for Nanoscience and Department of Physics and Astronomy, University of Missouri-St. Louis, One University Boulevard, St. Louis, Missouri 63121-4400, United States ∥ College of Materials Science and Opto-Electronic Technology, University of Chinese Academy of Sciences, Beijing, 10049 P. R. China ⊥ HRL Laboratories, LLC, 3011 Malibu Canyon Road, Malibu, California 90265-4737, United States S Supporting Information *
ABSTRACT: Neutron vibrational spectroscopy and quasielastic neutron scattering (QENS) were used to probe the dynamical properties of BH4− anions in both bulk LiBH4 and LiBH4 confined in nanoporous carbons (NPCs) having ≤4-nmdiameter, hexagonally arranged, cylindrical pores. The BH4− torsional band of the confined LiBH4 is significantly broadened relative to that of bulk LiBH4, reflecting a disruption of the bulk crystal lattice and thus a broader distribution of BH4− rotational potentials. QENS measurements of bulk orthorhombic LiBH4 indicate a single quasielastic component yielding an activation energy for localized BH4− jump reorientation of 19.2 ± 0.8 kJ/mol, consistent with previous QENS and NMR results. At room temperature, the measurements are in good agreement with BH4− reorientational jumps about a single C2 or C3 tetrahedral symmetry axis, with evidence for multiaxis rotations emerging as the temperature increases. In contrast, the QENS spectra of the NPC-confined LiBH4 exhibit two quasielastic components, one an order of magnitude broader than the other. The narrower component is presumably associated with more slowly reorienting BH4− anions in the interior of the pores and the broader component with much more rapidly reorienting BH4− anions in the vicinity of the pore surfaces. For 4-nm pores, these components yield two corresponding activation energies for reorientation: 16 ± 1 and 10.6 ± 0.7 kJ/mol. The data suggest that both components undergo single C2- or C3-axis reorientational jumps below 330 K, albeit with one an order of magnitude faster than the other. By 400 K (which is above the bulk phase transition temperature), both reorient more diffusively around multiple axes. These results were found to be qualitatively consistent with comparative dynamical measurements of LiBH4 confined in a 13-nm-average-pore-size carbon aerogel, which exhibited a much broader pore size distribution.
1. INTRODUCTION One key barrier to the use of hydrogen fuel cells for mobile applications is the lack of a storage material that not only has high H mass and volumetric densities but also requires only modest amounts of energy to release and recharge hydrogen. The lighter alkali and alkaline-earth borohydrides are appealingly hydrogen-dense, but the removal of hydrogen requires high temperatures, and the “recharging” of hydrogen is slow and requires high pressures of hydrogen gas. The nanoconfinement of hydrogen storage materials is one approach to improving the viability of the cycling reaction: Specifically, the phase separation and agglomeration of reactants and products are limited by the size of the scaffold pore. Individual components thus diffuse over only a finite distance to react. This is especially advantageous over the © 2013 American Chemical Society
course of reaction cycling, as ball-milled materials become phase-separated into progressively larger particle sizes. Moreover, the scaffold walls disrupt crystal structures, potentially resulting in an energetically destabilized system overall, relative to the bulk crystalline state. For example, LiBH4 confined in 13nm-pore carbon aerogels releases hydrogen 50 times faster than the bulk just above the melting temperature.1 When confined in 2-nm pores, LiBH4 exhibits a featureless (amorphous-like) Xray diffraction (XRD) signature, bulk phase transitions are no longer observed, and hydrogen desorption is observed as low as 220 °C.2 As an added advantage, the smaller pores seem to Received: June 27, 2013 Revised: August 5, 2013 Published: August 6, 2013 17983
dx.doi.org/10.1021/jp4063737 | J. Phys. Chem. C 2013, 117, 17983−17995
The Journal of Physical Chemistry C
Article
inhibit the release of diborane3 (B2H6), which is not only a toxic gas but also a precursor to a stable form of boron that inhibits cycling capacity.4 One disadvantage of the addition of a nanoscaffolding material is that the resulting composite hydrogen storage system will necessarily have a reduced hydrogen density compared to the bulk borohydride. Nonetheless, the hydrogen densities of the lighter-metal borohydrides are sufficiently great that some loss might be a worthwhile sacrifice to improve the thermodynamics of the hydrogen-release reaction or to improve the kinetics of the reverse reaction. Recent dynamical studies of LiBH4 in nanoporous carbon scaffolds have indicated enhanced mobilities for the BH4− anions. From solid-state NMR measurements of LiBH4 in 13and 25-nm-average-pore-size carbon aerogels, Shane et al.35 observed enhanced translational mobilities for a fraction of the BH4− anions, presumably associated with LiBH4 interfacial regions near the pore walls, with more bulk-like translational behavior from anions in the interior regions. Their assignments were consistent with the expected increasing fraction of interfacial BH4− anions with decreasing particle size. In a later solid-state NMR study of LiBH4 in a 1.9-nm-pore ordered silica (MCM-41), Verkuijlen et al.6 observed a highly translationally mobile fraction of BH4− anions in addition to BH4− anions with more bulk-like behavior. From quasielastic neutron scattering (QENS) measurements of LiBH4 in a nanoporous graphitic carbon (10-nm average pore size), Remhof et al.7 reported enhanced reorientational mobilities for a fraction of the BH4− anions, with the remaining fraction being immobile. The origin of each fraction was not firmly established. The authors speculated that the mobile fraction could be due to BH4− anions either near the pore walls, where the strain induced in the material is largest, or in the core of the pores within an inherently non-bulk-like LiBH4 nanophase. In the first and second scenarios, the immobile fraction was equated with BH4− anions in more bulk-like core regions and with interfacial BH4− anions immobilized by interactions with the pore walls, respectively. In contrast to the mentioned NMR studies involving narrow ordered-pore silica and carbon materials, the dynamical studies involving carbon-based hosts were all somewhat complicated by the associated broad pore size distributions. In the current work, we use neutron vibrational spectroscopy (NVS) and quasielastic neutron scattering (QENS) to probe the effects of nanoconfinement in a controlled-pore carbon scaffold on the local bonding and dynamics in LiBH4. Moreover, we compare these controlled-pore results to those obtained for LiBH4 confined in a carbon aerogel with relatively larger and more varied pore sizes. Because of the uniquely large incoherent neutron scattering cross section for hydrogen, the BH4− anions become sensitive indicators of their local environment. Hence, these neutron methods enable one to determine how the BH4− phonon modes and reorientational potentials are perturbed by nanoconfinement. Any knowledge of the local structure and dynamics associated with nanoconfined LiBH4 will contribute to a more comprehensive understanding of why borohydride nanoconfinement enhances the hydrogen-cycling properties compared to those of the bulk material.
occurring Li and B are very strong neutron absorbers. Confined samples consisted of 7Li11BH4 (purchased from Katchem8) melt-infiltrated under 60 bar H2 into highly ordered nanoporous carbon (NPC) that contained hexagonally packed columnar pores of uniform diameter. The NPC was prepared by using phenolic resins as carbon precursors and amphiphilic triblock copolymers as soft templates. Details of the synthesis and infiltration of NPC are reported elsewhere.2 The NPC used here was of two types: NPC-4 nm, with 4-nm-diameter pores, a higher surface area, and a larger pore volume, and NPC-2 nm, with 2-nm-diameter pores, a lower surface area, and a smaller pore volume. Sample LiBH4@NPC-4 nm contained 20 wt % 7 11 Li BH4 in NPC-4 nm, in which 70% of the available pore volume was filled. Sample LiBH4@NPC-2 nm contained 10 wt % 7Li11BH4 in NPC-2 nm, in which 35% of the available pore volume was filled. Generally, 70% filling is the maximum achievable because of some partial filling or pore blockage during infiltration, as discussed in more detail elsewhere.2 Very minor bulk thermodynamics features were observed in differential scanning calorimetry (DSC) measurements3 of the 70%-filled sample, which indicates that only a relatively insignificant fraction of the LiBH4 was present outside the pores. These controlled-pore samples were compared with an additional sample of 7Li11BH4 (synthesized according to the method of ref 9) melt-infiltrated under 1 bar H2 into a 13-nmaverage-pore-size carbon aerogel (CA), which inherently contained a much broader distribution of pore sizes.1 The CA was prepared using resorcinol−formaldehyde condensation as reported elsewhere.1 The sample handling was performed in a dry He glovebox. Melt infiltration was performed on a modified Schlenk line under flowing H2 gas. The LiBH4 was melted at 573 K and wicked into the aerogel pores on contact. The amount of infiltrated LiBH4 was enough to fill on the order of 20% of the available pore volume for the LiBH4@CA-13 nm sample. All neutron scattering measurements were performed at the National Institute of Standards and Technology (NIST) Center for Neutron Research. Quasielastic neutron scattering measurements were performed on a disk chopper spectrometer10 (DCS) with incident neutron wavelengths of 2.75 Å [10.8 meV, with 275 μeV full width at half-maximum (fwhm) elastic line width], 5 Å (3.3 meV, with 50.1 μeV fwhm and 105.5 μeV fwhm elastic line widths), and 8 Å (1.3 meV, 14.3 μeV fwhm elastic line width). Measurements were also performed on a high-flux backscattering spectrometer11 (HFBS) with an incident neutron wavelength of 6.27 Å (2.08 meV), momentum transfers (Q) over the range of 0.2−1.75 Å−1, and a resolution of 0.83 μeV fwhm in most detectors. The QENS spectra were reduced and analyzed using the DAVE12 software package, where the elastic feature was fitted to a delta function, the quasielastic feature to one or more Lorentzian functions, and the background to a linear baseline, all convolved with the resolution function. Measurements on LiBH4@NPC-4 nm and LiBH4@NPC-2 nm were not made at temperatures higher than 400 K because the nanoconfined samples have a propensity for hydrogen release when heated.3 Vibrational densities of states (VDOS) were measured at 4 K in neutron energy loss on the Filter-Analyzer Neutron Spectrometer13 (FANS) with pre- and postmonochromator collimations of 60′ and 40′, respectively, resulting in an instrumental resolution between 1.5 and 16 meV over the vibrational energy range (34−200 meV) presented. Prompt gamma activation analysis (PGAA) measurements of
2. EXPERIMENTAL SECTION Exclusively 7Li11BH4 was used in this study. The isotopeenriched material increases the neutron transmission considerably because 6Li and 10B isotopes present in naturally 17984
dx.doi.org/10.1021/jp4063737 | J. Phys. Chem. C 2013, 117, 17983−17995
The Journal of Physical Chemistry C
Article
the NPCs were performed on a dedicated beamline.14 The relative peak areas of carbon and hydrogen in the NPCs were compared to the relative carbon and hydrogen peak areas in a standard urea−graphite mixture containing 0.0373 atom % H.
tetrahedral 3-fold and 2-fold symmetry axes, and we abbreviate these mechanisms as C2|C3. If, however, a given hydrogen atom on the BH4− tetrahedron is able to occupy any of the other hydrogen positions on the tetrahedron (i.e., undergoing a “tetrahedral tumbling” motion) over the course of the reorientation process, the EISF can be written as
3. RESULTS AND DISCUSSION 3.1. Quasielastic Neutron Scattering. Incoherent neutron scattering is a tool that is well-suited for observing hydrogen atoms because the incoherent scattering cross section for hydrogen is more than an order of magnitude greater than those of most other atoms. Quasielastic neutron scattering probes hydrogen motions that occur on a frequency scale on the order of 108−1013 s−1, as a function of neutron energy transfer (ℏω) and Q, depending on the instrument and its configuration. In the case of covalently bound hydrogen atoms in a crystal lattice, these motions commonly involve occasional jumps over a barrier on the potential energy surface to a different, but crystallographically equivalent, molecular configuration. Quasielastic neutron scattering spectra consist of an elastic line δ(ω) and a quasielastic feature, in practice convolved with the resolution function. The quasielastic features are generally represented by one or more Lorentzian functions, Li(ω), centered at the elastic line. In general terms, the incoherent scattering function S(Q,ω) can be written as S(Q , ω) = A 0(Q ) δ(ω) +
∑ Ai(Q ) Li(ω)
EISF4C3 =
1 3 + A 0 (Q ) 4 4
EISFiso = j0 2 (QdB−H)
1 3 + 4 4
(1)
EISFHT =
0
H−H
⎪
1 [1 + j0 (QdH−H)] 2
( N3
)
N
⎛ πl ⎞⎤⎫⎞⎟ ⎬ QdH−H⎜sin ⎟⎥⎪ ⎝ N ⎠⎦⎭⎟⎠ ⎣ 3 ⎡
∑ j0 ⎢ 4 l=1
2
⎪
(7)
For N = 3, this model becomes eq 5, identical to tetrahedral tumbling. It should be noted that a variant of this model involving only partial trigonal H delocalization around the circle20,21 is also consistent with the observed EISFs for the hexagonal phase. 3.2. Bulk LiBH4. Quasielastic neutron scattering spectra were collected for bulk LiBH4 at 390, 380, 370, 350, 330, and 300 K using the DCS and at 225 and 200 K using the HFBS. Data collected at higher temperatures and analysis of those data can be found elsewhere,20 but representative EISF data at 400 K, showing good agreement with the trigonal H disorder model (eq 7) over an extended Q range using 2.75-Å incident neutrons, are shown in Figure 1. This model is consistent with very large thermal ellipsoids determined from diffraction measurements at 400 (ref 22) and 408 (ref 23) K that could be interpreted as nearly isotropic disorder. This type of diffuse rotation was suggested in the interpretation of Raman spectra24 and NMR data25 as well. A very recent molecular dynamics investigation26 of the hexagonal phase of LiBH4 suggested that the average crystal symmetry of this disordered phase is actually P63/mmc rather than P63mc, which might warrant reanalysis of previous LiBH4 structural and dynamical data. However, the trigonal disorder model mentioned earlier, incorporating only
(3)
where j0(x) is a zeroth-order Bessel function equal to sin(x)/ (x). dH−H is the jump distance between two BH4− hydrogen atoms and is related to the B−H bond length, dB−H, as dH−H = [(2√2)/√3]dB−H. Therefore, the normalized elastic scattering as a function of Q for a rotation of a BH4− tetrahedron about a single C3 axis is16,17 EISFC3 =
⎛ ⎡ ⎜ 1 ⎢ N + Nj (QdH−H)⎤⎥ ⎜ 0 ⎦ + N ⎝4⎣ 3 1
3⎧N ⎨ j (QdH−H) + + ⎪ 4⎩ 3 0
(2)
{ 13 [1 + 2j (Qd )]}
(6)
Although large thermal ellipsoids are observed for the hightemperature, hexagonal phase of LiBH4, crystallographic order is partially maintained, and the proposed BH4− reorientational mechanism20 is consistent with the diffraction results. In this case, three trigonal hydrogen atoms jump about the C3 axis of the anion (parallel to the crystallographic c axis) to multiple (N ≥ 6) positions on a circle, inducing crystallographic disorder. The hydrogen atoms diffusing about the circle are able to jump to the axial hydrogen position and back again. The EISF for this dynamic is
in which the A0(Q) term is replaced by the scattering function for a C3 reorientation on a circle15 EISFC3 =
(5)
This formalism was originally developed to describe C3 rotations about all four axes in solid methane,18 and it allows the tetrahedron to retain crystallographic order. Isotropic rotational diffusion is a mechanism that destroys the crystallographic order of the tetrahedron; that is, over time, through small stochastic jumps, the BH4− hydrogen atoms visit the entire surface of a sphere with a radius equal to the B−H bond length. The EISF observed in the presence of this mechanism is15,19
The ratio of the observed elastic peak area normalized by the area of the total observed scattering as a function of Q is the elastic incoherent structure factor (EISF) and can reveal the mechanism and geometry of the observed dynamics. A0(Q) + ∑Ai(Q) sums to unity (at a given Q value), making A0(Q) synonymous with the EISF. For rotational dynamics of a BH4− “rigid-body” anion situated within a solid lattice such as LiBH4, the EISF due to the H atoms reorienting around the central B atom is sensitive to the fundamental details of the reorientational mechanism. The case of an undistorted BH4− tetrahedron, where the reorientation of three of the four hydrogen atoms occurs about one C3 axis defined by the fourth, stationary, elastically scattering hydrogen atom, yields the following elastic incoherent scattering function EISFC3 =
1 [1 + 3j0 (QdH−H)] 4
(4)
BH4−
For a tetrahedron in which all four of the hydrogen atoms reorient about a single C2 axis, jumping a distance of dH−H, the EISF15 is, coincidentally, identical to eq 4. Therefore, QENS results are unable to distinguish between jump diffusion about 17985
dx.doi.org/10.1021/jp4063737 | J. Phys. Chem. C 2013, 117, 17983−17995
The Journal of Physical Chemistry C
Article
transition, but the vibrational peak splitting trended toward a symmetry change, complicated by anharmonicity and coupling between normal modes. According to synchrotron data,23 as the crystal warms from 300 K to the phase transition, the unit cell contracts continuously along the b axis. This could be related to the evolving dynamics present at these intermediate temperatures, although NMR analysis25 found no relationship between the 1H relaxation rate and the behavior of the lattice constant. Judging from the position of the minimum and the intensity of the EISFs as a function of Q (i.e., the observed EISFs approach the tetrahedral tumbling model, eq 5), it is likely that the low-temperature-phase dynamics is simply evolving from one-axis to multiaxis reorientations while maintaining crystallographic order. The width of the quasielastic feature increases with temperature as the jump rate increases and typically exhibits Arrhenius behavior. A fit to an Arrhenius function can yield an activation energy, Ea, for the reorientational jump. Over the temperature range of this study, the fitted line widths are primarily independent of Q, implying a localized jump mechanism. The quasielastic feature half-width at halfmaximum (hwhm) is proportional to 1/τ. The average line widths, summed over momentum transfer, were fit to the Arrhenius function 1/τ = 1/τ0 exp(−Ea/kBT), in which τ is the apparent residence time between jumps. A discontinuity can be observed in the Arrhenius plot (Figure 2) between 380 and 400
Figure 1. EISF of bulk LiBH4 collected at 330 K (diamonds, lowtemperature phase), 350 K (circles, low-temperature phase), and 400 K (stars, high-temperature phase) using 2.75-Å incident neutrons. Solid lines represent models for BH4− tetrahedron reorientation: uniaxial jumps about either a C2 or a C3 axis (eq 4, cyan) and hightemperature trigonal disorder (eq 7, maroon). Vertical error bars denote ±1σ.
partial 3-fold delocalization of three H atoms around the B−H axis of the fourth H atom with additional C2|C3 jump exchanges involving this fourth H atom, is still consistent with such a revised disordered structure. The low-temperature-phase QENS data collected at 300 K (with 5-Å incident neutrons) and 330 K (with 2.75-Å incident neutrons) using the DCS were found to be consistent with uniaxial jumps of 1.98 Å (corresponding to a B−H distance of 1.21 Å22) about either a 3-fold or 2-fold (eq 4) axis of rotation. EISF data for the latter temperature are presented in Figure 1. DFT calculations by Buchter et al.27 suggested that BH4− rotations around the 3-fold symmetry axes encounter a lower barrier than those around the 2-fold symmetry axes. The surprising observation is that the dynamics appearing at 350 K and persisting through 370 K agrees neither with the EISF for uniaxial C2|C3 jumps (eq 4) nor with the EISF for hightemperature, hexagonal-phase dynamics (eq 7), but rather with an EISF of intermediate character (Figure 1). This same intermediate EISF was also observed for LiBH4 at 380 K in another QENS study28 but was incorrectly attributed to a single 3-fold-axis jump about a circle, considering only three H atoms instead of all four tetrahedral hydrogen atoms. Because the data at 350 K (Figure 1) were collected over a sufficiently large Q range, it is clear that the jump distance (which determines the Q value of the minimum in the EISF curve) of this dynamic is affiliated with C2|C3 jumps consistent with the remainder of those observed for the low-temperature phase. Refinement of neutron powder diffraction data22 at 360 K reveals that the low-temperature structure is present as a single phase, with large thermal ellipsoids indicative of dynamic disorder. Therefore, the dynamics observed up to 370 K is likely not due to the presence of both orthorhombic and hexagonal phases. Furthermore, the NMR data25 collected between 92 and 424 K indicated consistent dynamics throughout the low-temperature phase. Neither these data nor those from another NMR study5 indicated the emergence of a high-temperature-phase dynamic in the low-temperature phase. Also, a temperature-dependent Raman study29 explored the evolution of the vibrational spectra, finding that the orthorhombic phase symmetry was maintained until the
Figure 2. Arrhenius plot of LiBH4 quasielastic line widths as a function of temperature. An activation energy of 19.2 ± 0.8 kJ/mol was determined for the observed low-temperature-phase reorientation with a τ0 value of 7 ± 2 fs. A discontinuity in the line width was observed at the transition between the low- and high-temperature phases. Vertical error bars denoting ±1σ are smaller than and obscured by the symbols.
K, at the transition between the orthorhombic and hexagonal crystal phases. A fit to the two HFBS points and five DCS points collected below the phase transition (200−380 K) yields an activation energy (Ea) of 19.2 ± 0.8 kJ/mol and a preexponential factor (τ0) of 7 ± 2 fs. A fit to only the DCS points (300−380 K) yields a slightly lower activation energy of 17 ± 1 kJ/mol, in agreement with the QENS results of Remhof et al.28 (Ea = 17.3 ± 0.3 kJ/mol) measured over a similar temperature range. NMR studies on the orthorhombic low-temperature phase of LiBH4 observed two distinct dynamics.25 These results established the presence of one dynamic with Ea = 17.6 kJ/ mol and τ0= 19 fs and the other dynamic with Ea = 24.2 kJ/mol and τ0 = 3 fs. The two motions were assumed to be due to 17986
dx.doi.org/10.1021/jp4063737 | J. Phys. Chem. C 2013, 117, 17983−17995
The Journal of Physical Chemistry C
Article
Table 1. BH4− Reorientational Activation Energies Determined by QENS and NMR Spectroscopy sample
a
Ea (kJ/mol)
τ0 (fs)
temperatures (K)
method
LiBH4
102−384
NMR spectroscopya
n/a
n/a
LiBH4 LiBH4 LiBH4@NPC-4 nm LiBH4@NPC-2 nm LiBH4@CA-13 nm LiBH4@GC-10 nm
300−380 300−380 193−400 125−400 190−360 300−430
QENSb QENS QENS QENS QENS QENSc
n/a n/a 10.6 ± 0.7 10.2 ± 0.5 12.3 ± 0.3 8.0 ± 0.3
n/a n/a 18 ± 6 24 ± 6 16 ± 3 not reported
Ea (kJ/mol) 17.6 ± 0.3 24.2 ± 0.4 17.3 ± 0.3 19.2 ± 0.8 16 ± 1 14.4 ± 0.4 14.5 ± 0.8 n/a
τ0 (fs) 19 ± 1 3.1 ± 0.2 42 ± 5 7±2 25 ± 11 29 ± 5 43 ± 15 n/a
Reference 25. bReference 28. cReference 7. GC denotes nanoporous graphitic carbon.
reorientations about the 2-fold and 3-fold rotational axes of the BH4− tetrahedron. These values are presented in Table 1. Neither the current nor the previous QENS experiments detected both dynamics suggested in NMR studies.25,30 The two activation energies determined by NMR spectroscopy are comparable to one another, and the pre-exponential factors differ by a factor of 6. Apparently, these dynamic modes occur on close enough time scales at the observed temperatures that QENS is unable to distinguish between the two. It is interesting that the activation energies determined in this and the previous QENS study28 agree more with one of the two values determined by NMR spectroscopy, and not the average. Thus, one of the two dynamics observed by NMR spectroscopy seems to dominate the temperature behavior observed by QENS. 3.3. Confined LiBH4. 3.3.1. NPC/LiBH4 Mass Ratio. To interpret the observed EISF for the infiltrated NPC samples, the relative amounts of elastic scattering from the NPC and the 7 11 Li BH4 must be determined. PGAA shows that about one residual H atom is present for every 10 C atoms of the bare NPC-4 nm sample. LiBH4@NPC-4 nm is 20 mass % 7Li11BH4 and 80 mass % NPC-4 nm. 11B MAS NMR measurements after the infiltration process indicate31 that about 60% of the LiBH4 remains intact whereas about 40% undergoes chemical transformation to non-BH4− forms of boron within the scaffold, presumably to a nonhydrogenous state, as suggested by neutron vibrational spectroscopy (see section 3.3.2). If it is further assumed that no additional hydrogenation of the NPC scaffold occurs upon LiBH4 infiltration, then based on incoherent neutron scattering cross sections, about 76% of the total incoherent scattering intensity should be from hydrogen associated with the nanoconfined LiBH4, with the remaining 24% from purely elastic scattering from the scaffold, 7Li, and 11 B. Likewise for NPC-2 nm, PGAA shows that one H atom is present for every 12 C atoms. The LiBH4@NPC-2 nm sample is 10 mass % 7Li11BH4 and 90 mass % NPC-2 nm. Because no similar 11B MAS NMR measurements were performed on the LiBH4@NPC-2 nm sample to determine the fraction of LiBH4 remaining intact upon infiltration, we assumed the same value of 60% as seen for the LiBH4@NPC-4 nm sample. Thus, under the same assumptions, it is estimated that about 63% of the total incoherent neutron scattering is from the hydrogen associated with the nanoconfined LiBH4 and the remaining 37% is again purely elastic scattering from the scaffold, 7Li, and 11 B. To determine the EISFs of the reorienting BH4− anions correctly, these extra contributions to the elastic scattering were removed from all quasielastic spectra. 3.3.2. Neutron Vibrational Spectra. A typical neutron vibrational spectrum for an empty carbon nanoscaffold at 4 K is exemplified by the NPC-4 nm sample in Figure 3, reflecting the
Figure 3. Neutron vibrational spectra of (a) LiBH4@NPC-2 nm (orange), (b) LiBH4@NPC-4 nm (green), (c) LiBH4@CA-13 nm (blue), (d) bulk LiBH4 (black), and (e) empty NPC-4 nm (gray). Vertical error bars denote ±1σ. (Note: 1 meV ≈ 8.066 cm−1.)
vibrational density of states of the residual H atoms. The NPC spectra show intensity (albeit with broader peaks) in similar vibrational regions as hydrogen covalently bound to carbon in the large polyaromatic hydrocarbon coronene.32 These predominantly sp2-bonded H atoms were found to scatter only elastically on the quasielastic time scale and thus did not contribute to the quasielastic scattering intensity. Neutron vibrational spectra at 4 K of LiBH4@NPC-4 nm, LiBH4@NPC2 nm, LiBH4@CA-13 nm, and bulk LiBH4 are also shown in Figure 3. The spectra for the three nanoconfined materials are difference spectra, representing only contributions from the nanoconfined LiBH4, after appropriate subtraction of the spectrum of the bare carbon nanoscaffold. A comparison of the nanoconfined LiBH4 spectra with the bulk LiBH4 spectrum reveals noticeable differences in the 52 meV (∼419 cm−1) BH4− torsional bands. The NPC-nanoconfined spectra, in particular, both show a torsional band that is broadened greatly from that of the bulk, suggesting a larger distribution of BH4− rotational potentials for these nanoconfined samples. Although the overall LiBH4 vibrational signatures are consistent with an orthorhombic-like arrangement of ions for all samples at 4 K, a significant fraction of the BH4− anions in the LiBH4@NPC samples, most likely those nearer the pore surfaces, appear to be experiencing rotational potentials considerably different than those present in bulk orthorhombic LiBH4. A comparison of the spectra of LiBH4@NPC-4 nm and LiBH4@NPC-2 nm 17987
dx.doi.org/10.1021/jp4063737 | J. Phys. Chem. C 2013, 117, 17983−17995
The Journal of Physical Chemistry C
Article
indicates that the BH4− torsional band becomes even more non-bulk-like as the pore size decreases, which correlates with the expected increasing fraction of interfacial BH4− anions and decreasing fraction of interior BH4− anions with decreasing pore size. For comparison, the torsional band for LiBH4@CA13 nm, although somewhat broader than for the bulk, is less so than for the LiBH4@NPC samples. This probably reflects a larger fraction of the LiBH4 being associated with interior BH4− anions for these larger-pore samples. Nonetheless, some caution should be taken in interpreting the exact origin of the torsional band perturbations, because the stabilization of short-ranged, non-bulk-like (amorphous or crystalline) structures due to size-dependent nanoconfinement might very well occur and, by itself, lead to an apparently “perturbed” vibrational density of states from all BH4− anions, even for the pore interiors. As implied earlier, the neutron vibrational results suggest that the non-BH4− forms of boron present in the nanopores after infiltration are not particularly hydrogenous, because the spectra are dominated by the phonon modes of BH4− anions. It is not clear what form this boron takes. (Note that, as the neutron scattering cross section for 11B is relatively insignificant compared to that for H, the phonons due to nonhydrogenous boron entities will be relatively invisible in the neutron vibrational spectrum.) In addition to these neutron spectroscopic results, LiBH4 was also seen to dominate the FTIR spectrum of a similarly made sample of melt-infiltrated 4-nm NPC.3 The non-BH4− boron species might be a common occurrence for H2-assisted LiBH4 infiltration in C-based and other types of nanoporous scaffolds where O atom “contaminants” are also present, enabling various oxides to be formed.33,34 3.3.3. Rotational Dynamics for LiBH4@NPC-4 nm. The observed quasielastic scattering for LiBH4@NPC-4 nm collected between 300 and 400 K on the DCS with a time scale sensitivity between 10 ns and 1 ps fits best to two Lorentzian components, as shown in Figure 4. The line widths are roughly an order of magnitude different from one another and are quite distinguishable. Interestingly, the narrower Lorentzian component, representing more slowly reorienting
BH4− anions, is similar in width and temperature behavior to the single Lorentzian component observed for bulk LiBH4. As the two quasielastic components show dramatically different widths and somewhat distinctive temperature behaviors, it is most probable that they represent two independent populations of BH4− anions. As noted previously, QENS is unable to distinguish dynamics with very similar time scales, so each population might consist of a distribution of similar molecular motions. Presumably, the two populations result from two different environments imposed by nanoconfinement, with two different activation energies. It follows that the two activation energies for the nanoconfined samples do not correspond to the two observed in NMR studies of bulk orthorhombic LiBH4, because those were interpreted as two types of reorientation from indistinguishable BH4− anions. Instead, this system is reminiscent of the NMR study of LiBH4 confined in 13- and 25-nm-pore-size carbon aerogels35 in which a nanocrystallite of LiBH4 was modeled as two parts: a bulk-like core region of translationally slow-moving BH4− anions and an interfacial shell of fast-moving anions near the scaffold wall. Hence, we propose that the reorientationally slower and faster components observed in our QENS measurements also correspond to more bulk-like core and less bulk-like interfacial BH4− anions, respectively. If the dynamics responsible for the quasielastic scattering are indeed independent of one another, a fraction of the elastic intensity can be attributed to each component. Therefore, to generate the EISFs for LiBH4@NPC-4 nm, the elastic scattering unassociated with the BH4− anions was first removed from the total observed scattering intensity at each Q value. Thus, the remaining spectrum at each Q value was presumably the sum of the scattering spectra only from the two populations of BH4− anions. The presence of two different BH4− populations complicates the identification of individual reorientation mechanisms. In particular, even though the intensities of each quasielastic component can be determined, the corresponding EISF for each cannot be generated in a straightforward manner because the division of the total elastic scattering intensity between the two components is not known a priori (i.e., the elastic/ quasielastic intensity ratio is model-dependent). With this in mind, we found it beneficial, after fitting each QENS spectrum to an elastic component and two quasielastic components, to calculate at each Q value, a global “average” EISF, that is, the total (LiBH4) elastic scattering intensity divided by the sum of elastic and quasielastic scattering intensities. The Q dependences of the average EISFs at 330 and 400 K are shown in Figure 5. Fortunately, at 330 K, the average EISF agrees well with the C2|C3 single-axis jump mechanism (from eq 4). Because the EISF curve for this particular mechanism represents an upper limit for all reasonable reorientational mechanisms involving BH4− tetrahedra, this strongly suggests that both the slow and fast BH4− anion populations exhibit the same single-axis reorientation mechanism at this temperature. Indeed, if one of the two populations were forced to exhibit any other plausible mechanism, such as tetrahedral tumbling or rotational diffusion (with their smaller elastic scattering fractions and EISFs), then the other population would necessarily be reapportioned a larger fraction of the total elastic scattering intensity, resulting in an unreasonably large EISFs above the limiting curve of the C2|C3 single-axis jump mechanism.
Figure 4. Observed (black) quasielastic neutron scattering spectra for LiBH4@NPC-4 nm at 400 and 330 K, with the fitted function (red) consisting of a delta function (white) and two quasielastic functions (green and blue), each convolved with the instrumental resolution function. Vertical error bars denote ±1σ. 17988
dx.doi.org/10.1021/jp4063737 | J. Phys. Chem. C 2013, 117, 17983−17995
The Journal of Physical Chemistry C
Article
diffusion or trigonal disorder) results in the best agreement for the EISF of the fast component with the isotropic rotational diffusion model. The trigonal disorder model for the slow component would not seem unreasonable at 400 K, because this is the model that matches the Q dependence of the EISFs due to BH4− reorientations in bulk hexagonal LiBH4 at this temperature, and the slow component is believed to be due to BH4− anions located in the more bulk-like core regions of the pores. It is true that XRD measurements on LiBH4@NPC-4 nm, similarly to LiBH4@NPC-2 nm,2,3 show no clear evidence of long-range order or an orthorhombic-to-hexagonal phase transition. Moreover, there is no clear evidence of a transition in LiBH4@NPC-4 nm based on solid-state 1H NMR measurements of the spin−lattice relaxation time as a function of temperature.31 However, one can make the case that differential scanning calorimetry (DSC) measurements on LiBH4@NPC-4 nm,1,3 although not yielding a prominent, sharp endothermic feature due to a solid−solid phase transition as for LiBH4 nanoconfined in larger-pore systems, do in fact exhibit a rather weak, broadened endothermic feature centered at around 345 K (i.e., ∼40 K lower than the bulk transition temperature), hinting at some minor structural transformation for at least a portion of the LiBH4. Nevertheless, if there are hexagonal arrangements of Li+ and BH4− ions in the core regions of 4-nm pores at 400 K leading to trigonally disordered BH 4 − reorientations, they are probably a rather disordered collection of very short-ranged hexagonal “clusters”. Indeed, for the smaller infiltrated pores in LiBH4@NPC-2 nm, no trace of a similar endothermic feature can be observed by DSC measurements. Based on the relative scattering intensities and assumed reorientation models for the two components, the numbers of BH4− anions undergoing fast and slow reorientations are roughly equal at 330 K. This corresponds to an interfacial layer thickness of ∼0.6 nm, that is, the first one or two outer layers participating in the fast reorientation mechanism, within the 4nm pore. At 400 K, the total intensity of the fast component increases to 3 times that of the slow component. This now corresponds to an interfacial layer thickness of ∼1 nm, that is, about three outer layers participating in the fast reorientation mechanism. A schematic of this core−shell model is depicted in Figure 7. Hence, our DCS data near 300 K and above suggests that the population of anions undergoing fast reorientations increases whereas that undergoing slow reorientations decreases with temperature. These results correspond well with the absolute values and temperature dependence of the interfacial layer thickness associated with translationally more mobile BH4− anions determined from earlier NMR studies35 of LiBH4 nanoconfined in 13- and 25-nm carbon aerogels and are compared in Figure 8. Quasielastic neutron scattering spectra were also collected for LiBH4@NPC-4 nm at temperatures between 100 and 250 K on the HFBS. The quasielastic features were fit with a single Lorentzian function. Even after the NPC elastic scattering contribution had been removed, the calculated EISFs remained too high to agree with any physically reasonable jump reorientation model for this system; that is, too much of the observed scattering was elastic. Physically reasonable models for jump rotations, such as those presented in eqs 4−7, can fit the data only if it is assumed that a portion of the confined LiBH4 exhibits dynamics on a time scale that is outside the range of the instrument. This is not a bad assumption: The spectra
Figure 5. EISF from all quasielastic scattering for LiBH4@NPC-4 nm at 330 K (diamonds) and 400 K (stars). Solid lines represent models for BH4− tetrahedron reorientation: uniaxial jumps about (a) either a C2 or a C3 axis (cyan), (b) trigonal disorder (maroon), and (c) isotropic rotational diffusion (black). Vertical error bars denote ±1σ.
Interestingly, at 400 K, the average EISFs deviate considerably from a single-axis jump mechanism, now reflecting a multiaxis reorientational behavior, somewhere between the trigonal disorder model (eq 7) and isotropic rotational diffusion (eq 6). Figure 6 illustrates the calculated “individual” EISFs for
Figure 6. Derived EISFs (blue, red, and green circles) for the fast LiBH4@NPC-4 nm dynamic at 400 K after restricting the slow dynamics to isotropic rotational diffusion, trigonal disorder, and tetrahedral tumbling models. Model EISF curves are shown for (a) tetrahedral tumbling, (b) trigonal rotational disorder (found in the hexagonal bulk LiBH4), and (c) isotropic rotational diffusion. Vertical error bars denote ±1σ.
the fast component for different assumed reorientation models (i.e., tetragonal tumbling, trigonal disorder, and isotropic rotational diffusion) for the slow component. Regardless of which model is used for the allocation of the elastic intensity for the slower component at 400 K, the EISFs of the faster component most resemble those corresponding to the isotropic rotational diffusion model (eq 6). The relative insensitivity of the EISFs for the fast component to the model used for the slow component is largely due to the fact that the total intensity of the fast component is 3 times that of the slow component. That being said, fitting the slow component to either of the more diffusive reorientational models (i.e., isotropic rotational 17989
dx.doi.org/10.1021/jp4063737 | J. Phys. Chem. C 2013, 117, 17983−17995
The Journal of Physical Chemistry C
Article
remaining elastic scattering intensity can then be attributed to the “missing” dynamic. At 125 K, the quasielastic feature is assumed to be due to single-axis reorientations of the fast interfacial BH4− anions. This is consistent with the HFBS fixedwindow scan (FWS) for LiBH4@NPC-4 nm shown in Figure S1 of the Supporting Information, where it is clear that the onset of non-bulk-like interfacial BH4− dynamics occurs by around 50 K. This is dramatically lower than the expected 150 K onset of the relatively slower bulk-like core BH4− dynamics, based on the comparative FWS for bulk LiBH4. After the appropriate amount of elastic scattering intensity had been allotted to the interfacial anions, the leftover elastic scattering was assigned to the slower core BH4− anions. Based on the relative intensities, we estimated roughly twice as many core BH4− anions as interfacial BH4− anions at this temperature, yielding a layer thickness of ∼0.42 nm. As the temperature increased, the amount of the observable dynamic scattering increased. By 200 K, the quasielastic line widths observed on the HFBS agreed well with those of the slower component observed on the DCS. This means that the quasielastic scattering for the faster component had become so broad that it disappeared into the baseline, leaving only an elastic scattering contribution. Again, based on the observed elastic and quasielastic intensities at 200 K, we estimated that about 59% of the BH4− population was slower core anions and 41% was faster interfacial anions, yielding a layer thickness of ∼0.46 nm. The observed interfacial layer thicknesses with temperature from the HFBS data are consistent the DCS data obtained at higher temperatures and are also plotted in Figure 8. It is comforting to see that the derived interface thickness approaches the thickness of a “single” mobile BH4− layer as the temperature is minimized. The activation energies Ea and pre-exponential factors τ0 for BH4− reorientation were determined from Arrhenius fits (Figure 9) to the LiBH4@NPC-4 nm quasielastic line widths (hwhm) as a function of temperature between 125 and 400 K. The activation energy for bulk reorientation (19.2 ± 0.8 kJ/ mol) is shown for comparison. The activation energies are 10.6 ± 0.7 kJ/mol for the faster component and 16 ± 1 kJ/mol for
Figure 7. Cross-sectional schematic of LiBH4 inside a 4-nm cylindrical pore in which the outermost ring represents the interfacial layer estimated from the lower-temperature QENS data (roughly 0.4 nm).
Figure 8. Interfacial layer thickness as a function of temperature determined from QENS for LiBH4@NPC-4 nm (green circles) and LiBH4@NPC-2 nm (orange circles) compared with NMR-derived values35 (triangles) for LiBH4 in carbon aerogels with average pore sizes of 13 nm (maroon) and 25 nm (blue), based on a core−shell model. The horizontal line at 0.4 nm marks the approximate crystallographic diameter of a BH4− anion. Values near and below 200 K were determined from HFBS data. Standard errors are commensurate with the scatter in the data.
collected on the HFBS include energy transfers of only ±17 μeV from the elastic line, with a resolution close to 1 μeV. If these dynamics have an order-of-magnitude difference in Lorentzian widths, it is improbable that they will be observed together at the same temperature. Slower dynamics would be observed entirely as elastic scattering, and faster dynamics would have a quasielastic feature that is larger than the available energy window, appearing as a baseline. Either scenario would lead to a higher-than-expected EISF. If it is assumed that the observed quasielastic component is due to C2|C3 single-axis jump rotations, as deduced from the DCS data at 330 K, then the associated elastic scattering can be determined, after the expected scattering from the scaffold is removed. The
Figure 9. Arrhenius plot of the observed LiBH4@NPC-4 nm (green) and LiBH4@NPC-2 nm (orange) quasielastic line widths as a function of inverse temperature. Line widths for bulk LiBH4 are shown in black for comparison. Vertical gray lines mark the phase transition temperature for bulk LiBH4 near 385 K and the possible phase transition temperature for LiBH4@NPC-4 nm near 345 K. The solid orange and green lines represent linear fits to the data. Vertical error bars denote ±1σ. 17990
dx.doi.org/10.1021/jp4063737 | J. Phys. Chem. C 2013, 117, 17983−17995
The Journal of Physical Chemistry C
Article
the slower component, as listed in Table 1. It is evident from the Arrhenius plot that the slower component is somewhat similar in behavior to bulk orthorhombic LiBH4, which makes sense if one assumes that this component is indeed associated with BH4− anions in the core regions. The lower activation energy for the faster component suggests the presence of weaker rotational potentials for the interface BH4− anions, which is consistent with such BH4− anions having a “lower coordination number” of Li+ and BH4− neighbors and possibly weaker interactions with the carbon scaffold walls. It would be interesting to see how the reorientational mobilities would be affected by changing the scaffold wall to a chemically different material such as silica. 3.3.4. Rotational Dynamics for LiBH4@NPC-2 nm. In addition to LiBH4@NPC-4 nm, we performed more limited DCS measurements on LiBH4@NPC-2 nm for comparison. We mentioned earlier that the smaller pore size led to a more perturbed BH4− torsional band compared to that of the 4-nm sample, suggesting that the ratio of non-bulk-like to bulk-like (i.e., interface to core) LiBH4 in the pores was enhanced. This is consistent with the expected increase in the interface-to-core ratio as the pore size decreases and is reflected in a relatively more perturbed FWS in Figure S1 (Supporting Information). As the LiBH4@NPC-2 nm sample had a significantly lower surface area and LiBH4 sample loading, this complicated our ability to satisfactorily analyze the data, and we were only able to obtain limited EISF data as a function of Q, which corroborated the general mechanistic trend that there are two quasielastic components and that an increase in temperature from 330 to 400 K leads to a change from single-axis to multiaxis reorientation. Despite the difficulties in performing a more detailed analysis, by summing the data over all Q values, we were able to make an Arrhenius plot for both components, and the results are also included in Figure 9 and Table 1. The activation energies were found to be 10.2 ± 0.5 kJ/mol for the fast component and 14.4 ± 0.3 kJ/mol for the slower component. The confined LiBH4@NPC-2 nm and LiBH4@ NPC-4 nm samples have equivalent activation energies for the fast dynamics within error, an indication that the LiBH4 undergoing faster reorientation experiences similar environments in both NPCs. The slower components in LiBH4@NPC2 nm and LiBH4@NPC-4 nm have slightly different activation energies. This is not unreasonable considering that the core regions in LiBH4@NPC-2 nm are relatively smaller, leading to slightly faster, somewhat more non-bulk-like mobilities. Figure 8 shows the interfacial thickness calculated from the LiBH4@NPC-2 nm QENS data averaged over all Q values as a function of temperature. These results indicate reasonable values again close to the diameter of the BH4− anion and similar to what was calculated for LiBH4@NPC-4 nm below 330 K. Yet, unlike the interfacial thicknesses derived from the LiBH4@NPC-4 nm QENS and LiBH4@CA NMR data, which both seem to start increasing significantly above around 330 K, the interfacial thickness and thus the fraction of the more mobile interfacial BH4− anions derived for LiBH4@NPC-2 nm appears to be rather constant with temperature. Verkuijlen et al.6 also observed, by solid-state NMR spectroscopy, no obvious variation in the fraction of translationally mobile BH4− anions for LiBH4 in 1.9-nm ordered-pore silica over the 233−313 K temperature range. The apparent differences in thickness behavior with NPC pore size warrant a caveat concerning the interpretation of the LiBH4@NPC-4 nm QENS results above 330 K. In particular, if some fraction of the interior BH4−
anions were indeed undergoing a solid−solid phase transition above 330 K to a short-range-ordered hexagonal structure, then some of the quasielastic scattering from these slower anions would become much broader because of a structure-induced enhancement in reorientational mobility (similar to what is observed for bulk LiBH4). This would cause an apparent redistribution of intensities between quasielastic components, resulting in an artificial enhancement in the interfacial thickness. If the fraction of BH4− anions converting to the hexagonal phase increased with temperature, then this, in turn, would create further “enhancements” in interfacial thickness. This hypothesis is consistent with the fact that the LiBH4@ NPC-2 nm sample is not expected to undergo a solid−solid phase transition and therefore should not show an enhanced interfacial thickness with temperature, as we observed. If the remnants of a phase transition were occurring in LiBH4@NPC-4 nm, then this might also manifest itself in the apparent Arrhenius behaviors of the two components. Upon closer scrutiny of Figure 9, one might conclude that there are indeed slight discontinuities in the Arrhenius plots of both components for LiBH4@NPC-4 nm above around 345 K, the temperature corresponding to the weak feature observed previously in DSC measurements.3 One could further argue that these discontinuities are not observed for LiBH4@NPC-2 nm, as expected. Nonetheless, because the prior DSC and NMR data and the current QENS evidence for a possible solid−solid phase transition in LiBH4@NPC-4 nm is tenuous, we have chosen to ignore the presence of such a transition during our analyses. 3.3.5. Rotational Dynamics for LiBH4@CA-13 nm. It is of interest to compare these results for controlled-pore NPCs with similar measurements on LiBH4 in carbon aerogels (CAs), which exhibit a broader array of pore sizes and were the protypical support of choice in earlier work on nanosequestration of hydrogen storage materials.1 As mentioned earlier, the BH4− torsional band for LiBH4@CA-13 nm in Figure 3 is slightly broadened with respect to that for the bulk, reflecting some perturbation due to nanoconfinement, but not as dramatically as for the smaller-pore NPCs, where the fraction of interfacial BH4− anions is expected to be higher. A previous FWS on the HFBS of this partially filled CA sample36 also confirmed the presence of enhanced BH4− reorientational mobilities compared to the bulk. QENS measurements of the LiBH4@CA-13 nm sample were performed with the DCS. After the elastic scattering contribution from the empty CA had been subtracted, the resulting QENS spectra, similarly to those of the LiBH4@NPC samples, were found to fit better to two Lorentzians. As was the case for the LiBH4@NPC-2 nm sample, because of the limited sample size and the partial LiBH4 filling, no attempt was made to perform a detailed analysis of the reorientational mechanism, although a rudimentary Q-dependent analysis of the data confirmed that the H jump distance was consistent with BH4− anion reorientations. Figure 9 shows the Arrhenius plot of the QENS data after summing over all Q values. Similarly to our observations for LiBH4@NPC-4 nm as evidenced from the line widths in the Arrhenius plot, both the fast and slow dynamics observed for the LiBH4@CA-13 nm sample appear to undergo subtle step increases in rate as the temperature is increased above 360 K, most likely reflecting an orthorhombic-to-hexagonal phase transition roughly 20−25 K lower than that in the bulk. This decrease in transition temperature is consistent with what was observed previously by 17991
dx.doi.org/10.1021/jp4063737 | J. Phys. Chem. C 2013, 117, 17983−17995
The Journal of Physical Chemistry C
Article
DSC measurements.1 Although the average pore size of the CA sample is 13 nm, the pore size distribution is much broader, and 15% of the pore volume is due to pore sizes of less than 2 nm.1 It is suggested from neutron fixed-window scans on the HFBS36 that a considerable volume fraction of the smallest pores is preferentially filled at partial LiBH4 loadings. Nonetheless, because LiBH4 in 2-nm pores does not normally exhibit an observable solid−solid phase transition, the subtle step increases in the Arrhenius plots for LiBH4@CA-13 nm suggest that some volume fraction of larger pores must also be filled. This conclusion is bolstered by Figure 10, which displays the relative areas of the slow and fast quasielastic components from Figure 11. Fast (circles) and slow (diamonds) quasielastic line widths for LiBH4@CA-13 nm (blue), along with the results of the Arrhenius fits (blue line). The data points above 360 K were excluded from the fit. The fitted function for the LiBH4@NPC-4 nm sample (dashed green line), the line widths for the bulk (black), and the line widths of LiBH4 in nanoporous graphitic carbon7 (gray circles) are shown for comparison. The gray vertical lines indicate temperatures of 385 and 360 K. Vertical error bars denote ±1σ.
transformed hexagonal-phase BH4− anions complicate matters by augmenting the scattering intensity of the faster dynamic due to the interfacial BH4− anions in both the smaller and larger pores. This behavior most likely leads to unexpected perturbations in both fitted quasielastic line widths in this temperature regime. Despite the complications from a broader pore size distribution in carbon aerogels, additional HFBS data for the LiBH4@CA-13 nm sample at 140 and 190 K, where only the fast and slow components, respectively, are within the dynamical window of the instrument, allowed us to dramatically extend the temperature range of the Arrhenius plots in Figure 11 and yielded a more credible picture of the average dynamical behavior of both components. The activation energies were found to be 12.3 ± 0.3 kJ/mol for the fast component and 14.5 ± 0.8 kJ/mol for the slow component. The comparison of the bimodal behavior of the LiBH4@CA-13 nm sample to the Arrhenius fits of the LiBH4@NPC-4 nm data in Figure 11 show similar dynamical behaviors for the slow components but a somewhat larger deviation in relative behaviors for the fast components. Dynamical differences in the latter components can reasonably be attributed to differences in the interactions with the different types of carbon pore surfaces (i.e., NPC and CA) and possibly to differences in pore shapes, sizes, and distributions. The similar behaviors of the slower components are not unexpected because they involve the more bulk-like BH4− anions in the pore interior regions away from the interface and, as such, are more impervious to differences in BH4−−pore surface interactions. Because the Arrhenius behavior for the slow component of LiBH4@NPC-2 nm (in Figure 8) appears to deviate more from that of LiBH4@NPC-4 nm than does the behavior of LiBH4@CA-13 nm, this suggests that the majority of the slow BH4− anions in LiBH4@CA-13 nm involve pores equal to or larger than 4 nm in size. 3.3.6. Comparison with Other QENS Measurements. The bimodal Arrhenius behaviors of all three samples roughly bracket the Arrhenius dependence of the single reorientationally mobile BH4− component for LiBH4 in nanoporous graphitic carbon observed by Remhof et al.7 (included in Figure 11 for comparison). In contrast, their reported immobile
Figure 10. Relative intensities of the slow (open diamonds) and fast (blue circles) LiBH4@CA-13 nm quasielastic components as a function of temperature. The vertical gray lines mark 360 and 385 K, the phase transition temperatures for CA-confined and bulk LiBH4, respectively. Vertical error bars denote ±1σ.
a fit of the QENS spectra summed over all Q values. The slow component dominates the spectra at temperatures below the phase transition, whereas the fast component dominates the spectra above the phase transition. If we reasonably assume that the apparent increase in the fast component is due to contributions from hexagonally transformed LiBH4 in the interiors of the largest filled pores, then the temperature behavior in Figure 10 indeed suggests that some of the infiltrated LiBH4 is associated with the “micropores” less than about 2 nm in size, whereas the rest is associated with larger “mesopores” manifesting clear solid−solid phase transitions. Unfortunately, estimating the ratio of LiBH4 present in the micropores and mesopores is not straightforward because it would require a detailed knowledge of all of the pore sizes involved. Therefore, the presence of such a broad distribution of filled-pore sizes and the resulting ambiguity in the origin of each quasielastic feature with temperature clouds the interpretation of activation energies determined from the Arrhenius plots in Figure 11. In other words, below 360 K, the faster and slower dynamics are likely associated with the reorientational mobilities of the interfacial and interior BH4− anions, respectively, averaged over all pore sizes involved. These mobilities could very well display some pore-size dependence, especially for pore sizes of less than 2 nm. More importantly, above 360 K, the transformation of orthorhombic LiBH4 in the interiors of the larger pores into the hexagonal phase leads to an order-of-magnitude increase in the corresponding BH4− reorientation line widths. Again, as explained earlier for the LiBH4@NPC-4 nm data, any newly 17992
dx.doi.org/10.1021/jp4063737 | J. Phys. Chem. C 2013, 117, 17983−17995
The Journal of Physical Chemistry C
Article
about 330 K, with the appearance of multiaxis reorientations above this temperature. The Arrhenius dependence of the single quasielastic component indicates a reorientational activation energy of 19.2 ± 0.8 kJ/mol, consistent with previous QENS and NMR results. Unlike the bulk species, NPC-confined LiBH4 displays two quasielastic components. The narrower, more slowly reorienting component is believed to be associated with more bulk-like BH4− anions in the pore interiors and the order-of-magnitude broader, more rapidly reorienting component with much more non-bulk-like BH4− anions near the pore surfaces. For LiBH4 in the 4-nm pore NPC, these components yield reorientational activation energies of 16 ± 1 and 10.6 ± 0.7 kJ/mol, respectively. For LiBH4 in the 2-nm pore NPC, the reorientational activation energies are 14.3 ± 0.4 and 10.2 ± 0.5 kJ/mol, respectively. Both components appear to undergo uniaxial C2 or C3 reorientational jumps below 330 K and more diffusive motions around multiple axes by 400 K. QENS measurements of LiBH4 in a 13-nm-pore carbon aerogel (with a broader pore size distribution) also suggest the presence of slow and fast quasielastic components with corresponding reorientational activation energies of 14.5 ± 0.8 and 12.3 ± 0.3 kJ/mol. All nanoconfined samples display broadened BH4− torsional phonon bands compared to that of bulk LiBH4. In closing, we emphasize that probing the details of BH4− reorientational dynamics for LiBH4 nanoconfined in carbonbased materials is complicated by various factors. For example, the infiltration process typically results in the chemical modification of some of the BH4− anions as a result of various potential interactions with the support. The intact LiBH4 within the larger pores greater than about 4 nm also tends to undergo solid−solid phase transitions at temperatures that are pore-sizedependent and as much as 35−40 K below the bulk transition temperature. Moreover, different carbon-based materials exhibit different pore geometries and pore size distributions with potentially different surface chemical properties. Nonetheless, by using both the HFBS and DCS instruments and choosing to study LiBH4 in NPCs with hexagonal arrays of small, uniformsized cylindrical pores, we were able to mitigate some of the complications and, thus, greatly improve our understanding of the effects of carbon-based nanosequestration on BH4 − reorientational mobilities over a broad temperature and dynamical range.
component has no analogue in the present study. Past NMR studies35 and our current study indicate a slower component with bulk-like dynamical behavior. The reported immobile fraction could possibly be associated with such a slower component, unresolved with their instrument line width of 200 μeV and contributing to the apparent elastic scattering. We have no definite explanation for the origin of this extra elastic scattering. The authors claimed that it was possible that other nonborohydride hydrogenous species (such as hydroxides) might be present due to partial oxidation of the samples, although care was taken during sample preparation and handling. From our own past experience with this particular nanoporous graphitic carbon, the amount of residual H atoms before infiltration is an order of magnitude less than for NPC or CA materials and should be largely invisible compared to the amount of infiltrated LiBH4. Nonetheless, the pore characteristics and/or surface chemistries of various carbon-based materials are different. For example, nanoporous graphitic carbon, unlike NPC and CA, can have slit pores more akin to intercalation-sized spacings. Such pores might very well favor Li+ and BH4− “intercalation”-like infiltration leading to much more sterically hindered anions or even BH4− chemical transformation. Indeed, there is some suggestion from X-ray Raman spectroscopy that Li might be in intercalated form upon dehydrogenation of the LiBH4 in nanoporous graphitic carbon,34 implying that at least Li becomes situated between graphitic layers at some point. Remhof et al.7 also discussed the possible implications for immobilization of infiltration-induced strains in the material. These types of forces might also dramatically affect reorientational mobility within small slit pores. Indeed, from a visual comparison of the neutron vibrational spectra for bulk LiBH4 and 15 wt % LiBH4 in the nanoporous graphitic carbon in ref 7, one can conclude, after the two spectra are rescaled to equal masses of LiBH4, that a nontrivial amount of extra H-related background intensity exists for the latter sample. Unfortunately, the origin of this H is still unresolved because we cannot establish whether or not the “immobile” hydrogen component is part of the more discrete LiBH4 density of states or part of the more diffuse background. Whatever the origin of the immobile H atoms (a slower dynamic, sterically hindered BH4− anions or hydrogenous material trapped by the nanoporous graphitic carbon as a result of the infiltration process), the relative Arrhenius behavior of the mobile BH4− component in Figure 11 suggests that it might be a one-Lorentzian average fit of two Lorentzian components with widely differing line widths, as we observed for LiBH4 in NPC and CA. Because amenability of the QENS data to a twoLorentzian analysis depends on a variety of factors such as the instrumental resolution and accessible energy range, as well as the details of the pore size and corresponding line width distributions, such an analysis might not have been warranted. What is clear is that LiBH4 nanosequestration in different carbon-based materials results in dramatically enhanced BH4− reorientational mobilities, at least for a significant fraction of the borohydride. The relationship between these complex BH4− anion mobilities and improvements in LiBH4 cycling properties and reversibility or enhanced Li+ conductivity is yet to be determined.
■
ASSOCIATED CONTENT
S Supporting Information *
Elastic scattering intensity as a function of temperature from the HFBS for LiBH4@NPC-4 nm, LiBH4@NPC-2 nm, and LiBH4. This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS The authors are grateful to Dr. Michael R. Hartman for his assistance in the preparation of the partially filled aerogel sample. This work was supported in part by DOE-EERE under Grants DE-AI-01-05EE11104, DE-EE0002978, DE-AC04-
4. CONCLUSIONS According to QENS measurements, the BH4− anions in bulk orthorhombic LiBH4 undergo predominantly uniaxial reorientational jumps about a C2 or C3 tetrahedral symmetry axis below 17993
dx.doi.org/10.1021/jp4063737 | J. Phys. Chem. C 2013, 117, 17983−17995
The Journal of Physical Chemistry C
Article
(17) Yildirim, T.; Gehring, P. M.; Neumann, D. A.; Eaton, P. E.; Emrick, T. Neutron-Scattering Investigation of Molecular Reorientations in Solid Cubane. Phys. Rev. B 1999, 60, 314−321. (18) Sköld, K. Effects of Molecular Reorientation in Solid Methane on the Quasielastic Scattering of Thermal Neutrons. J. Chem. Phys. 1968, 49, 2443−2445. (19) Hempelmann, R. Quasielastic Neutron Scattering and Solid State Diffusion; Clarendon Press: Oxford, U.K., 2000. (20) Verdal, N.; Udovic, T. J.; Rush, J. J. The Nature of BH4 Reorientations in Hexagonal LiBH4. J. Phys. Chem. C 2012, 116, 1614−1618; 2012, 116, 5275−5275. (21) Verdal, N.; Udovic, T. J.; Rush, J. J.; Wu, H.; Skripov, A. V. Evolution of the Reorientational Motions of the Tetrahydroborate Anions in Hexagonal LiBH 4−LiI Solid Solution by High-Q Quasielastic Neutron Scattering. J. Phys. Chem. C 2013, 117, 12010−12018. (22) Hartman, M. R.; Rush, J. J.; Udovic, T. J.; Bowman, R. C., Jr.; Hwang, S.-J. Structure and Vibrational Dynamics of Isotopically Labeled Lithium Borohydride Using Neutron Diffraction and Spectroscopy. J. Solid State Chem. 2007, 180, 1298−1305. (23) Filinchuk, Y.; Chernyshov, D.; Cerny, R. Lightest Borohydride Probed by Synchrotron X-ray Diffraction: Experiment Calls for a New Theoretical Revision. J. Phys. Chem. C 2008, 112, 10579−10584. (24) Hagemann, H.; Gomes, S.; Renaudin, G.; Yvon, K. Raman Studies of Reorientation Motions of [BH4]− Anions in Alkali Borohydrides. J. Alloys Compd. 2004, 363, 129−132. (25) Skripov, A. V.; Soloninin, A. V.; Filinchuk, Y.; Chernyshov, D. Nuclear Magnetic Resonance Study of the Rotational Motion and the Phase Transition in LiBH4. J. Phys. Chem. C 2008, 112, 18701−18705. (26) Aeberhard, P. C.; Refson, K.; David, W. I. F. Molecular Dynamics Investigation of the Disordered Crystal Structure of Hexagonal LiBH4. Phys. Chem. Chem. Phys. 2013, 15, 8081−8087. (27) Buchter, F.; Łodziana, Z.; Mauron, P.; Remhof, A.; Friedrichs, O.; Borgschulte, A.; Züttel, A.; Sheptyakov, D.; Strassle, T.; RamirezCuesta, A. J. Dynamical Properties and Temperature Induced Molecular Disordering of LiBH4 and LiBD4. Phys. Rev. B 2008, 78, 094302. (28) Remhof, A.; Łodziana, Z.; Martelli, P.; Friedrichs, O.; Züttel, A.; Skripov, A. V.; Embs, J. P.; Strässle, T. Rotational Motion of BH4 Units in MBH4 (M = Li, Na, K) from Quasielastic Neutron Scattering and Density Functional Calculations. Phys. Rev. B 2010, 81, 214304. (29) Hagemann, H.; Filinchuk, Y.; Chernyshov, D.; van Beek, W. Lattice Anharmonicity and Structural Evolution of LiBH4: An Insight from Raman and X-ray Diffraction Experiments. Phase Transit. 2009, 82, 344−355. (30) Tsang, T.; Farrar, T. C. Nuclear Magnetic Relaxation Studies of Internal Rotations and Phase Transitions in Borohydrides of Lithium, Sodium, and Potassium. J. Chem. Phys. 1969, 50, 3498. (31) Liu, X.; Majzoub, E. H.; Stavila, V.; Bhakta, R.; Allendorf, M.; Shane, D.; Conradi, M.; Verdal, N.; Udovic, T. J.; Hwang, S.-J. Probing the Unusual Anion Mobility of LiBH4 Confined in Highly Ordered Nanoporous Carbon Frameworks via Solid State NMR and Quasielastic Neutron Scattering. J. Mater. Chem. A 2013, 1, 9935− 9941. (32) Mitchell, P. C. H.; Ramirez-Cuesta, A. J.; Parker, S. F.; Tomkinson, J. Inelastic Neutron Scattering in Spectroscopic Studies of Hydrogen on Carbon-Supported CatalystsExperimental Spectra and Computed Spectra of Model Systems. J. Mol. Struct. 2003, 651−653, 781−785. (33) Goudon, J. P.; Bernard, F.; Renouard, J.; Yvart, P. Experimental Investigation on Lithium Borohydride Hydrolysis. Int. J. Hydrogen Energy 2010, 35, 11071−11076. (34) Miedema, P. S.; Ngene, P.; van der Eerden, A. M. J.; Weng, T.C.; Nordlund, D.; Sokaras, D.; Alonso-Mori, R.; Juhin, A.; de Jongh, P. E.; de Groot, F. M. F. In Situ X-ray Raman Spectroscopy of LiBH4. Phys. Chem. Chem. Phys. 2012, 14, 5581−5587. (35) Shane, D. T.; Corey, R. L.; McIntosh, C.; Rayhel, L. H.; Bowman, R. C.; Vajo, J. J.; Gross, A. F.; Conradi, M. S. LiBH4 in
94AL85000, and DE-FC36-05GO15067. The work at the University of the Chinese Academy of Sciences was supported by “Hundred Talents Project” and the National Basic Research Program of China (973 Program, 2010CB833101). This work utilized facilities supported in part by the National Science Foundation under Agreement DMR-0944772.
■
REFERENCES
(1) Gross, A. F.; Vajo, J. J.; Atta, S. L. V.; Olson, G. L. Enhanced Hydrogen Storage Kinetics of LiBH4 in Nanoporous Carbon Scaffolds. J. Phys. Chem. C 2008, 112, 5651−5657. (2) Liu, X.; Peaslee, D.; Jost, C. Z.; Majzoub, E. H. Controlling the Decomposition Pathway of LiBH4 via Confinement in Highly Ordered Nanoporous Carbon. J. Phys. Chem. C 2010, 114, 14036−14041. (3) Liu, X.; Peaslee, D.; Jost, C. Z.; Baumann, T. F.; Majzoub, E. H. Systematic Pore-Size Effects of Nanoconfinement of LiBH4 : Elimination of Diborane Release and Tunable Behavior for Hydrogen Storage Applications. Chem. Mater. 2011, 23, 1331−1336. (4) Friedrichs, O.; Remhof, A.; Hwang, S.-J.; Zuttel, A. Role of Li2B12H12 for the Formation and Decomposition of LiBH4. Chem. Mater. 2010, 22, 3265−3268. (5) Corey, R. L.; Shane, D. T.; Bowman, R. C.; Conradi, M. S. Atomic Motions in LiBH4 by NMR. J. Phys. Chem. C 2008, 112, 18706−18710. (6) Verkuijlen, M. H. W.; Ngene, P.; de Kort, D. W.; Barré, C.; Nale, A.; van Eck, E. R. H.; van Bentum, P. J. M.; de Jongh, P. E.; Kentgens, A. P. M. Nanoconfined LiBH4 and Enhanced Mobility of Li+ and BH4− Studied by Solid-State NMR. J. Phys. Chem. C 2012, 116, 22169− 22178. (7) Remhof, A.; Mauron, P.; Züttel, A.; Embs, J. P.; Łodziana, Z.; Ramirez-Cuesta, A. J.; Ngene, P.; de Jongh, P. Hydrogen Dynamics in Nanoconfined Lithiumborohydride. J. Phys. Chem. C 2013, 117, 3789−3798. (8) The mention of all commercial suppliers in this article is for clarity. It does not imply the recommendation or endorsement by NIST of these suppliers. (9) Hartman, M. R.; Rush, J. J.; Udovic, T. J.; Bowman, R. C.; Hwang, S.-J. Structure and Vibrational Dynamics of Isotopically Labeled Lithium Borohydride Using Neutron Diffraction and Spectroscopy. J. Solid State Chem. 2008, 180, 1298−1305. (10) Copley, J. R. D.; Cook, J. C. The Disk Chopper Spectrometer at NIST: A New Instrument for Quasielastic Neutron Scattering Studies. Chem. Phys. 2003, 292, 477. (11) Meyer, A.; Dimeo, R. M.; Gehring, P. M.; Neumann, D. A. The High Flux Backscattering Spectrometer at the NIST Center for Neutron Research. Rev. Sci. Instrum. 2003, 74, 2759. (12) Azuah, R.; Kneller, L.; Qiu, Y.; Tregenna-Piggott, P. L. W.; Brown, C.; Copley, J.; Dimeo, R. DAVE: A Comprehensive Software Suite for the Reduction, Visualization, and Analysis of Low Energy Neutron Spectroscopic Data. J. Res. Natl. Inst. Stand. Technol. 2009, 114, 341−358. (13) Udovic, T. J.; Brown, C. M.; Leão, J. B.; Brand, P. C.; Jiggetts, R. D.; Zeitoun, R.; Pierce, T. A.; Peral, I.; Copley, J. R. D.; Huang, Q.; Neumann, D. A.; Fields, R. J. The Design of a Bismuth-Based Auxiliary Filter for the Removal of Spurious Background Scattering Associated with Filter-Analyzer Neutron Spectrometers. Nucl. Instrum. Methods A 2008, 588, 406−413. (14) Paul, R. L.; Lindstrom, R. M. Prompt Gamma-Ray Activation Analysis: Fundamentals and Applications. J. Radioanal. Nucl. Chem. 2000, 243, 181−189. (15) Bée, M. Quasielastic Neutron Scattering: Principles and Applications in Solid State Chemistry, Biology and Materials Science; Adam Hilger: Bristol, U.K., 1988. (16) Prager, M.; Grimm, H.; Natkaniec, I.; Nowak, D.; Unruh, T. The Dimensionality of Ammonium Reorientation in (NH4)2S2O8: The View from Neutron Spectroscopy. J. Phys.: Condens. Matter 2008, 20, 125218. 17994
dx.doi.org/10.1021/jp4063737 | J. Phys. Chem. C 2013, 117, 17983−17995
The Journal of Physical Chemistry C
Article
Carbon Aerogel Nanoscaffolds: An NMR Study of Atomic Motions. J. Phys. Chem. C 2010, 114, 4008−4014. (36) Udovic, T. J.; Verdal, N.; Rush, J. J.; Vries, D. J. D.; Hartman, M. R.; Vajo, J. J.; Gross, A. F.; Skripov, A. V. Mapping Trends in the Reorientational Mobilities of Tetrahydroborate Anions via NeutronScattering Fixed-Window Scans J. Alloys Compd., published online Feb 19, 2013, 10.1016/j.jallcom.2013.02.025.
17995
dx.doi.org/10.1021/jp4063737 | J. Phys. Chem. C 2013, 117, 17983−17995