Diffraction Studies of Hydrogenous Materials at GLAD

P. C. Trulove, D. Haworth, and R. T. Carlin, Division of Electrochemical Sciences, The Frank J. Seiler Research Laboratory

S. R. Nagel, R. L. Leheny, and N. Menon, James Franck Institute, The University of Chicago

P. Zhou and J. E. Fischer, Department of Materials Science and Engineering and Laboratory for Research on the Structure of Matter, University of Pennsylvania

N. Koura, Tokyo University of Science

A. K. Soper, Neutron Science Division, Rutherford Appleton Laboratory

A. J. G. Ellison, M.-L. Saboungi, K. Suzuya, S. Takehashi, and D. L. Price, Materials Science Division, and K. J. Volin, IPNS Division, Argonne National Laboratory

Introduction

The Glass, Liquid, and Amorphous Material Diffractometer (GLAD) is proving itself to be a first-class instrument for structural analysis of a wide variety of disordered materials, including glasses, liquids, fibers and other amorphous solids, and disordered crystalline materials, including oxides exhibiting superconductivity and giant magnetoresistance. The quality of the data appears excellent, and the consistency over a wide range of angles and wavelengths is highly satisfactory.

Many interesting disordered materials contain hydrogen, either as a primary component or as a residual constituent left over from preparation from organic precursors. In these cases, the conventional "Placzek" approach for subtracting the self scattering, based on a mass expansion in powers of neutron mass: atomic mass, is inapplicable. The problem has been addressed by Soper and collaborators (Soper and Luzar, 1993) at ISIS, who have developed a procedure for estimating the self scattering by a Chebyshev polynomial fit to the structure factor, made consistent with Krogh-Moe normalization (essentially, the requirement that the radial distribution function n(r) be zero below some minimum value of r) through a maximum-entropy method. This procedure has been implemented on the SMUG node at IPNS as a new analysis program, SUBSLFGLD. This report provides a brief description of some of the experiments performed on GLAD over the past two years on hydrogenous materials and a preliminary account of some of the results.

Neutron Diffraction Studies of an Ambient-Temperature Molten Salt

Ambient-temperature molten salts composed of mixtures of 1-ethyl-3-methylimidazolium chloride (ImCl) and hydrogen chloride (HCl) are part of an important class of nonaqueous solvents. These solvents have been used for studying the chemistry of a wide variety of both organic and inorganic solutes (Osteryoung, 1987; Hussey, 1983). Mixtures of ImCl and HCl are liquid at or below ambient temperature over a wide range of compositions. Recent spectroscopic studies (Campbell and Johnson, 1993) have shown that the anionic speciation in ImCl:HCl molten salts varies significantly with relative amounts of ImCl and HCl. When ImCl is in molar excess, the anions present are primarily Cl^- and HCl₂^-, while a molar excess of HCl gives HCl₂^- and H₂Cl₃^- as the primary anions. In the case of a large molar excess of HCl, additional polyanions of the form Cl(HCl)_n^- may be present. When HCl and ImCl are mixed in nearly equal amounts, HCl₂^- is the primary anion (however, small amounts of Cl^- and H₂Cl₃^- - may be present). Since Im⁺ is the sole cationic species, the equal molar mixtures of ImCl and HCl are often referred to as ImHCl₂.

The ring hydrogens on the 1-ethyl-3-methylimidazolium cation (Im⁺) are known to hydrogen-bond to chloride ion in basic AlCl₃:ImCl molten salts (Dieter et al., 1988; Dymeck and Stewart, 1989). The existence of anions of the form Cl(HCl)_n^- in the ImCl:HCl molten salts demonstrates the ability of HCl₂^- to act as an hydrogen-bond acceptor. Therefore, it is reasonable to conclude that HCl₂^- may hydrogen-bond to the ring hydrogens on Im⁺ in ImCl:HCl molten salts. In addition, recent solid-state crystallographic studies of similar low-melting salts indicate that Im⁺ cations tend to form p-stacks with d-d distances < 4 Å (Wilkes and Zaworotko, 1993). These and other observations suggest significant ordering of the ions in ImCl:HCl molten salts. Consequently, a better understanding of the structure of ImCl:HCl would be of considerable aid in understanding its chemical and physical properties, as well as how these properties relate to those of other, like low-melting salts.

Neutron diffraction on GLAD was employed to investigate the structure of the ImCl:HCl ambient-temperature molten salt. For the investigation of the liquid structure of ImCl:HCl, judicious deuterium substitution was employed to isolate the interactions involving hydrogen. Experiments were performed on ImCl:HCl samples with differing combinations of H-D substitution on both the imidazolium cation ring and hydrogen chloride. The diffraction differences from the hydrogen-deuterium substitution were analyzed by using programs developed at ISIS (Soper and Luzar, 1993).

The analysis of first-order differences due to deuteration of HCl₂^- gave two intrastructural peaks. These results indicate that HCl₂^- exists as an asymmetric ion in the ImHCl₂ molten salt. This is quite remarkable, considering that the HCl₂^- ion in AlCl₃ImCl molten salts appears to be linear with the hydrogen symmetrically placed between the two chlorides (Trulove and Osteryoung, 1992). Furthermore, to our knowledge, asymmetric HCl₂^- ions have only been observed in the solid state (Evans and Lo, 1966). An asymmetric HCl₂^- implies some type of interaction with an additional species. The samples of ImHCl₂ used for these experiments contained some free Cl^-. As shown in Figure 1A, this free Cl^- could weakly hydrogen-bond to the hydrogen in HCl₂^-. However, because of the low actual Cl^- concentration, very few of the species shown in Figure 1A could actually form. A more plausible reason for an asymmetric ion is the formation of a hydrogen-bond between one chloride from the HCl₂^- and a hydrogen on the imidazolium cation ring (Figure 1B). The chlorides in the HCl₂^- ion are capable of acting as hydrogen-bond acceptors, as demonstrated by the presence of anions of the type Cl(HCl)_n^- in the ImHCl₂ melts containing excess HCl (Campbell and Johnson, 1993). In addition, the imidazolium ring hydrogens have been shown to strongly hydrogen-bond to Cl^- in AlCl₃:ImCl molten salts (Dieter et al., 1988). Analysis of second-order differences due to correlations between hydrogens on the imidazolium cation ring and HCl₂^- gave a pronounced peak near r = 4 Å. The area for this peak corresponded to 2 - 3 hydrogens about a hydrogen at the origin. These results are consistent with the p-stacking of the imidazolium cations.

Structural Studies of Organic Liquids through the Glass Transition

At temperatures below a liquid’s melting point, the time scales that characterize its response to an external perturbation increase dramatically (Nagel, 1993). Extrapolations from measurements in this supercooled region suggest that for many liquids these time scales diverge at a finite temperature. Despite numerous studies to probe these slowing dynamics, the origin of this finite temperature divergence remains an outstanding problem . In contrast to the multitude of dynamical measurements, relatively few structural studies have attempted to identify changes in structure with the growing time scales. The main impediment to such work has been that the systems that show this finite temperature divergence most strongly are typically organic molecular liquids, in which significant intramolecular scattering and large hydrogen content complicate the analysis of structure data.

In experiments on supercooled D-propylene glycol (C₃D₈O₂) at GLAD, we have attempted to overcome some of these complications. We have collected diffraction data on the liquid at several temperatures from high in the liquid state to below the divergence temperature in an effort to identify structural trends we can associate with the slowing dynamics. The combination of the liquid’s light elements and large molecular weight makes the standard Placzek corrections for self scattering inappropriate for our data. In particular, application of this correction leads to much greater temperature dependence in S(Q) than the differential cross sections indicate. In the low-Q region, the good quantitative agreement of the differential cross sections from GLAD with data taken at SASI suggests that inelastic scattering plays a small role in our results, and in our analysis we have assumed a structureless, flat self scattering contribution. We have also applied a maximum entropy analysis to the data and find that the resulting structure factors closely approximate those calculated with the flat self scattering. In particular, much of the same temperature-dependent structure in S(Q) appears after application of each correction. Figure 2 shows the resulting S(Q) at our highest and lowest temperatures. The inset shows the second and third peaks, which reveal systematic changes with temperature.

We have also performed molecular dynamics simulations of the liquid in an effort to isolate contributions from intermolecular scattering to the changes in S(Q). Figure 3 shows a comparison of the simulation results with the experimental data. As shown in the inset to Figure 3, the simulations possess the same trends in temperature as the experimental S(Q). Our analysis of the simulation results demonstrates that the intramolecular structure factor shows little temperature dependence and that much of the change in temperature can be identified with intermolecular scattering. Fourier analysis of the intermolecular scattering in the simulations reveals a sharpening of pair correlations at distances between 1.5 and 4 Å at low temperatures. Therefore, we tentatively identify the changes in S(Q) in Figure 2 with an enhanced orientational correlation between nearest neighbors as the temperature decreases.

Local Atomic Structures of Amorphous Carbons from Radial Distribution Function Analysis

Lithium-intercalated amorphous carbons are attracting enormous interest recently due to their applications as anode electrodes in lithium-ion rechargeable (rocking-chair) batteries (Dahn et al., 1995a; Zheng et al., 1995). The concept is that Li cations move back and forth between the anode and cathode during charge/discharge cycles without ever being reduced to hazardous metallic form. The prototype cells exhibit energy density four times that of conventional NiCAD batteries and much longer recycling lifetimes (as much as 3,000 cycles). The host carbons can be divided into three categories according to their structures and chemical compositions. First, graphitic carbons prepared at temperatures >2500°C with a maximum lithium capacity of, in electrochemistry terminology, 372 milliampere-hours per gram (mAA h/g) correspond to a stoichiometry LiC₆ , the physics and chemistry of which are very well studied and understood (Fischer, 1987). Its exfoliation in electrolytes results in a very short lifetime, limiting its wide usage. The second category covers vitreous carbons from pyrolyzing polyaromatic hydrocarbons or polymers at moderate temperatures (800 to 1500°C). The lithium capacity of these materials ranges from 300 to 600 mAA h/g,, as good as or better than LiC6. One interpretation involves covering up with lithium on both sides of "single" graphitic sheets (Dahn et al., 1995b). The third category includes vitreous carbons from organic precursors pyrolyzed at lower temperatures (500 to 800°C). These carbons exhibit exceptionally high lithium capacity (600-1000 mAA h/g), most of which is irreversible (i.e., covalently bonded lithium atoms), an effect due to high hydrogen residual.

Here, we address the local structures of amorphous carbons with large hydrogen concentration and the chemical bonding states of lithium, using time-of-flight (TOF) pulsed neutron diffraction. A carbon sample was prepared at Simon Fraser University, Canada (Dahn et al., 1995a) by pyrolyzing epoxy novolac resin (poly[(phenyl glycidyl either)-co- formaldehyde], Dow Chemical Corp.) cured with phthallic anhydride (Aldrich). The cured monolith was ball-milled to fine powder prior to pyrolysis, which was performed in a tube furnace at 700°C in argon for one hour. Elemental chemical analysis determined the hydrogen/carbon atomic ratio, H/C = 0.17. Electrochemical testing on a small aliquot of the sample revealed a lithium capacity of 650 mAA h/g. Neutron diffraction data were taken on GLAD. A vanadium can (7/16-in. OD) with an indium gasket seal was used to prevent moisture. Scattering from the instrument and the vanadium can was also measured and removed from the data. A graphite powder sample (Fisher, 1987) was also measured under the same conditions, as a control. Each data set was taken for eight hours. A large background was observed from the amorphous sample due to both incoherent and inelastic scattering from the hydrogen. We used the SUBSLFGLD program to remove the background and obtain the coherent scattering. No resonance was considered in the fitting procedures.

The neutron-weighted structure factor S(Q) (after removal of the background) is plotted in the inset to Figure 4. Only a few broad features can be seen due to the very amorphous nature of the sample. The radial density function (4p r² r(r)) was obtained from S(Q) via Fourier transformation; the result is plotted in Figure 4 along with that of graphite. A Lorch-weighting factor was used to remove the spurious features due to truncation at 30 Å^-1. The first peak, found at 1.44 Å, agrees very well with the C-C distance in graphite (1.42 Å). The coordination number (area under the first peak) and bond angle (from the second peak position at 2.47 Å) are 3.1 and 118°, respectively, consistent with an sp² type of bonding. The so-called intrahexagonal peak is found at 2.85 Å, twice the bond length, a signature of planar hexagons. These results indicate that this particular material is remarkably similar to graphite on the length scale < 1 nm. The decrease in peak intensities relative to those of graphite is due to small particle sizes. Real space modeling using plain graphite sheets reveals that the average size is around 10-20 Å, which would have 18% at. hydrogen atoms if they were all attached to the edges, consistent with the H/C ratio from the chemical analysis.

Neutron Diffraction Measurements of Imidazolium Chloroaluminates

Mixtures of aluminum chloride, AlCl₃, and 1-ethyl-3-methylimidazolium chloride (EMIC) have low melting points (well below room temperature), high electric conductivity, and high current density of aluminum plating. These properties are strongly affected by the ionic species and the structures of the melts. Previous investigations of these melts have indicated that the major aluminum complex is AlCl₄^- in basic melts (AlCl3<50 mol%) and Al₂Cl₇^- in acidic melts (AlCl₃>50 mol%).

Neutron diffraction studies were carried out on GLAD to obtain the structure and configuration of the ionic species. The samples were prepared by mixing highly anhydrous AlCl₃ with fully deuterated EMIC in a glove box, where the H₂O and O₂ levels were maintained at less than 2 ppm. The melts were sealed under vacuum in quartz tubes of 1-mm thickness and inner diameter of 3 mm. Four compositions, with 46, 50, 60, and 67 mol% AlCl₃, were selected. Measurements were made on each sample at 298 K, followed by a similar measurement on an empty fused silica container of the same dimensions for purposes of instrument calibration and data normalization; measurements were also carried out on a 0.64-cm-diameter vanadium standard and with the spectrometer empty, both at 298 K. The data were analyzed with standard procedures developed at Argonne for glass and liquid diffraction data, incorporating simple corrections for multiple scattering and inelasticity effects.

Figure 5 shows the structure factors S(Q) measured for the four solutions. These are rather similar over the range Q > 5A^-1. However, they show differences at relatively low Q, in the range of 1-5 A^-1, which must be due to the interactions between the AlCl₄^- and Al₂Cl₇^- anions and the EMI⁺ cation.

A better understanding of the structure of these melts can be achieved by combining ab initio quantum chemistry calculations with the neutron diffraction data. For the 67 mol% AlCl₃ melt, the calculations were carried out on the assumption that Al₂Cl₇^- and EMI⁺ were present. The neutron diffraction pattern derived is in good agreement with the experimental one. The structure of the Al₂Cl₇^- and EMI⁺ complex is shown in Figure 6.

REFERENCES

J.L.E. Campbell and K.E. Johnson, Inorg. Chem. 32, 3809 (1993).

J.R. Dahn, A.K. Sleigh, H. Shi, B.M. Way, W.J. Weydanz, J.N. Reimers, Q. Zhong, and U. von Sacken, in: Lithium Batteries- New Materials, Developments and Perspective, G. Pistoia, Editor (Elsevier, 1995a).

J.R. Dahn, et al., Science 270, 590 (1995b). The single-sheet model was derived from the widths of the (002) and (100) peaks in the x-ray diffraction patterns, using the Scherrer equation.

K.M. Dieter, C.J. Dymeck, N.E. Heimer, J.W. Rovang, and J.S. Wilkes, J. Am. Chem. Soc. 110, 2722 (1988).

C.J. Dymeck and J.J.P. Stewart, Inorg. Chem. 28, 1472 (1989).

J.C. Evans and G.Y-S. Lo, J. Phys. Chem. 70, 11 (1966).

J.E. Fischer, in: Chemical Physics of Intercalation, A.P. Legrand and S. Flandois, Editors (Plenum, N.Y., 1987), p. 59.

C.L. Hussey, Adv. Molten Salt Chem. 5, 185 (1983).

S.R. Nagel, in: Phase Transitions and Relaxation in Systems with Competing Energy Scales, T. Riste and D. Sherrington, Editors (Kluwer Academic Press, Boston, Mass., 1993), p. 259.

R.A. Osteryoung, in: Molten Salt Chemistry, G. Mamantov and R. Marassi, Editors (Reidel Publishing, Netherlands, 1987) p. 329

A.K. Soper and A. Luzar, J. Chem. Phys. 97, 1320 (1993).

P.C. Trulove and R.A. Osteryoung, Inorg. Chem. 31, 3980 (1992).

J.S. Wilkes and M.J. Zaworotko, Supramolecular Chem. 1, 191 (1993).

T. Zheng, Y. Liu, E.W. Fuller, S. Tseng, U. von Sacken, and J.R. Dahn, J. Electrochem. Soc., submitted (1995).

FIGURE CAPTIONS

Figure 1. Possible structures for the asymmetric HCl₂^- ion.

Figure 2. The structure factor measured for D-propylene glycol (C₃D₈O₂) at 90 K (_____) and 300 K (------). The inset shows an enlargement of the second and third peaks, whose shapes change systematically with temperature.

Figure 3. The structure factor for D-propylene glycol measured (_____) at 160 K and calculated from molecular dynamics simulations of the liquid (------) at the same temperature. The inset shows the second and third peaks of the structure factor calculated from simulations at 90 K (_____) and 300 K (------). S(Q) for the simulations reveals the same trends with temperature seen in the experiment.

Figure 4. Radial distribution function (rdf, 4pr²r(r), solid curve) of the amorphous carbon containing 17 at.% hydrogen. The graphite rdf is also plotted as a reference (crosses). A Lorch-weighting factor was applied in both cases to remove spurious features due to truncation at 30 Å^-1. Inset: neutron-weighted structure S(Q). The incoherent and inelastic scattering background from hydrogen were fitted to a third-order Chebychev polynomial and removed.

Figure 5. Structure factors of fully deuterated AlCl₃-EMIC.

Figure 6. Structure and Mulliken charges of the Al₂Cl₇^--EMI⁺ complex.