UNTANGLING CATION ORDERING IN COMPLEX LITHIUM BATTERY CATHODE MATERIALS SIMULTANEOUS REFINEMENT OF X-RAY, NEUTRON AND RESONANT SCATTERING DATA 328 P.S. Whitfield, I.J. Davidson, L.M.D. Cranswick*, I.P. Swainson*, and P.W. Stephens $ Institute for Chemical Process and Environmental Technology, National Research Council Canada, 1200 Montreal Road, Ottawa, Ontario, K1A 0R6, Canada *Neutron Program for Materials Research, Chalk River Laboratories, Chalk River, Ontario, K0J 1J0, Canada Department of Physics and Astronomy, State University of New York, Stoney Brook, New York, 11794-3800, USA $ National Synchrotron Light Source, Brookhaven National Laboratory, Upton, New York, 11973, USA ABSTRACT The presence of multiple, neighbouring transition metals presents a challenge for powder diffraction when trying to locate individual elements in a structure. Combining the information in X-ray and neutron data in simultaneous refinements has historically been a powerful tool. However, increasingly complex compositions require more information and element contrast than two datasets can provide. This paper presents results demonstrating the application of resonant powder diffraction as additional datasets in simultaneous X-ray and neutron Rietveld refinements. Final results are presented for a simple layered R-3m structure with four cations, and preliminary results for a more complex monoclinic C2/m structure, again with four different cations. The more complex structure presents challenges with respect to the occupational constraints across sites with different multiplicities whilst allowing the lithium-to-transition metal ratio to refine. INTRODUCTION Lithium-ion batteries have largely replaced nickel-cadmium and nickel-metal hydride batteries for consumer electronics applications. This is due to their increased energy density, where increased functionality is being demanded from ever smaller devices. The lithium-ion cell relies on the ability of lithium ions to shuttle backwards and forwards between a cathode (usually a lithiated transition metal oxide) and anode (usually graphite). One of the more costly components of a lithium-ion battery is the cathode material, the conventional material being
This document was presented at the Denver X-ray Conference (DXC) on Applications of X-ray Analysis. Sponsored by the International Centre for Diffraction Data (ICDD). This document is provided by ICDD in cooperation with the authors and presenters of the DXC for the express purpose of educating the scientific community. All copyrights for the document are retained by ICDD. Usage is restricted for the purposes of education and scientific research. DXC Website www.dxcicdd.com ICDD Website - www.icdd.com
LiCoO 2. These materials have a layered structure for easy extraction and insertion of lithium 329 ions into the structure, and often have the R-3m symmetry (Figure 1). Although LiCoO 2 performs well as a cathode, cobalt is expensive as well as toxic, so intensive effort has been put into reducing or eliminating cobalt from the cathode materials, as well as improving performance. In common with many functional materials, a common approach is to substitute different elements into the structure to obtain the desired properties. More recently the substitutions have become more complex with multiple substitutions, which pose significant challenges. Figure 1. R-3m crystal structure of LiCoO 2 One particular substitution of interest is the addition of excess lithium, especially where the material contains significant amounts of manganese. These materials appear to possess anomalously high discharge capacities and improved thermal stability due to the presence of Mn 4+ (1). One such composition is the 1:1 solid solution of Li 2 MnO 3 and LiNi 0.75 Co 0.25 O 2, which can be written as Li 1.2 Mn 0.4 Ni 0.3 Co 0.1 O 2 (2). The excess lithium induces ordering which results in a Li 2 MnO 3 -like monoclinic C2/m unit cell (Figure 2) (3). 4h 2c Figure 2. Crystal structure of C2/m Li 2 MnO 3 showing the different sites within the structure. Rietveld analysis (4) of powder diffraction data has become a common tool for studying ordering, or confirming 2b 4g
the absence of ordering in such oxides. The increasing complexity of site occupancies in crystal structures parallels the problem of increasingly complex simultaneous equations. Two variables (or in this case cations) is quite straightforward when full occupancy is assumed, three requires more information (e.g. a neutron dataset) and four becomes significantly more complicated. Two ways of adding additional elemental contrast to the problem are to use isotopic substitution with neutron diffraction or resonant diffraction using a synchrotron. The resonant diffraction approach relies on tuning the wavelength to an absorption edge of one of the elements. The reduction in elastic scattering reduces the apparent electron density of the element, and consequently changes its apparent scattering power (5). An example of this effect for manganese is shown in Figure 3. 330 f Figure 3. Variation of scattering power f with X-ray energy for manganese. 0.6998Å is equivalent to MoKα, 1.5418 is equivalent to CuKα and 1.869Å is the wavelength chosen close to the Mn K absorption edge for this study. EXPERIMENTAL Samples with composition LiMn 1/3 Ni 1/3 Co 1/3 O 2 and Li 1.2 Mn 0.4 Ni 0.3 Co 0.1 O 2 were prepared using a sucrose-based chelation-combution process (2), similar in some regards to the Pechini method (6). The metal nitrates and/or acetates were dissolved in water and acidified to ph1 using nitric acid. Water was evaporated by heating on a hotplate until a viscous brown liquid was formed. At this stage the heating was increased to initially foam, and then char the mixture. Ultimately the carbonaceous matrix formed spontaneously combusted, producing a fine ash. This ash was heated in a furnace at 900ºC for the LiMn 1/3 Ni 1/3 Co 1/3 O 2 and 800ºC for the lithium-rich Li 1.2 Mn 0.4 Ni 0.3 Co 0.1 O 2.
The transition metal cation ratios of the samples were determined using wavelength dispersive X-ray fluorescence. These ratios were used as constraints in the refinements. No constraints were made in terms of lithium content. Laboratory X-ray diffraction data were obtained using a dual Göbel mirror, parallel beam Bruker D8 diffractometer. Data were collected over the range 10-100º 2θ using CuKα radiation with a 0.02º step size and counting time of 12 seconds. Neutron diffraction data were collected at 1.33Å on the NRC DUALSPEC C2 high-resolution neutron powder diffractometer located at the Chalk River Laboratories, Ontario, Canada. Synchrotron X-ray diffraction data were collected on the X3B1 beam-line at the National Synchrotron Light Source at Brookhaven National Laboratory. Data were collected in reflection mode on a quartz zero background holder at 0.6999 (off edge), 1.605 (~Co K edge) and 1.869 Å (~Mn K edge). The data analysis was carried out using Topas 2.1 and 3.0 beta versions. The datasets were weighted such that the contribution of the single neutron dataset was not overwhelmed by the multiple X-ray datasets. The occupational constraints were constructed such that the transition metals could float across the different sites whilst refining the lithium-to-transition metal ratio. This was more difficult for the C2/m cell as 4 cation sites are involved with two different multiplicities. To allow the site occupancies to refine simultaneously a technique previously described for GSAS was utilised where the element occupancies on a single site are split up, and each portion shared with another site (7). The differing site multiplicities were dealt with arithmetically and penalty functions used to avoid negative lithium occupancies. 331 RESULTS AND DISCUSSION The refinement of LiMn 1/3 Ni 1/3 Co 1/3 O 2 was stable and was consistent with a disordered R-3m structure with no sign of ordered superstructures (8). The overall R wp over all of the refinements was 11%. Details of the individual fits can be seen in Whitfield et al (2005) (8). The final refined structure (seen in Figure 4) showed a displacement of 2% lithium by nickel within the structure, despite the lithium-to-transition metal ratio being very close to ideal at 1:1. Refined thermal parameters were close to those found for LiNiO 2 (9). In LiMn 1/3 Ni 1/3 Co 1/3 O 2 all of the nickel is known to be in the +2 oxidation state (10). Given the similarity between the ionic radii of Li + (0.9Å) and Ni 2+ (0.83Å) it is not surprising that Ni has a preference for the Li site in this material. The situation in the lithium-rich Li 1.2 Mn 0.4 Ni 0.3 Co 0.1 O 2
is more complex, as the majority of the nickel has been oxidised to the low-spin +3 state, where the Ni ionic radius (0.7Å) is much closer to the other transition metals, Mn 4+ (0.67Å) and lowspin Co 3+ (0.685Å) 332 Figure 4. Structure of LiMn 1/3 Ni 1/3 Co 1/3 O 2 refined using all of the datasets simultaneously. 2% nickel displaced lithium from the lithium 3a site An example of the preliminary fits obtained with the lithium-rich material can be seen in Figure 5 for the Co-edge data. It can be seen that the anisotropically broadened superstructure reflections in the range 21-25º 2θ can be fit successfully using arbitrary broadening functions. Similar fits were obtained for the superstructure reflections in all of the datasets. Figure 5. Rietveld difference plot showing the fit to the 1.605Å Co K absorption edge data from Li 1.2 Mn 0.4 Ni 0.3 Co 0.1 O 2. The fitting of the anisotropic peak broadening of the superstructure reflections in the region 21-25º 2θ can be seen. The simultaneous refinement of Li 1.2 Mn 0.4 Ni 0.3 Co 0.1 O 2 has a much shallower minimum than LiMn 1/3 Ni 1/3 Co 1/3 O 2, as well as a much larger number of degrees of freedom in the unit cell. The preliminary results suggest that the excess lithium is situated only on the 2b transition metal site. In addition, most of the transition metal occurring on the lithium sites seems to be nickel. It
is possible that a small fraction of unoxidised Ni 2+ exists in the material. The presence of nickel in the lithium layers is often regarded as detrimental to the electrochemical performance of a cathode material. However, in cases where a very large fraction of the lithium is removed from the structure, it is possible that the presence of a pillar such as nickel could stabilise the crystal structure, and enhance both the reversibility of the intercalation/deintercalation and the discharge capacities attainable (amount of lithium that can be extracted per gram of material). This is a possibility that requires further study in these materials. 333 CONCLUSIONS The addition of resonant scattering data to a simultaneous X-ray and neutron Rietveld refinement is a powerful tool to study cation ordering in complex oxide materials. In LiMn 1/3 Ni 1/3 Co 1/3 O 2 it was demonstrated that only nickel migrated to the lithium site. In the lithium-rich material a similar picture emerges, despite the much reduced difference in ionic radii between low spin Ni 3+ and the other transition metals, and the corresponding value for Ni 2+. REFERENCES [1] Lu, Z.; Dahn, J.R., J.Electrochem.Soc., 2002, 149, A815-822. [2] Whitfield, P.S.; Niketic, S.; Davidson, I.J., J.Power Sources, 2004, in press, [3] Strobel, P.; Lambert-Andron, B., J.Solid State Chem., 1988, 75, 90-98. [4] Rietveld, H.M., Acta Cryst., 1967, 22, 151-152. [5] Cox, D.E.; Wilkinson, A.P., "Powder diffraction studies using anomalous dispersion", in Resonant Anomalous X-ray Scattering: Theory and Application, Elsevier Science, Amsterdam, 1994, 195-219 [6] Lessing, P.A., American Ceramic Society Bulletin, 1989, 68, 1002-1007. [7] Joubert, J.-M.; Cerný, R.; Latroche, M.; Percheron-Guégan, A.; Yvon, K., J.Appl.Crystallogr., 1998, 31, 327-332. [8] Whitfield, P.S.; Davidson, I.J.; Cranswick, L.M.D.; Swainson, I.P.; Stephens, P.W., Solid State Ionics, 2005, 176, 463-471. [9] Hirano, A.; Kanie, K.; Ichikawa, T.; Imanishi, N.; Takeda, Y.; Kanno, R.; Kamiyama, T.; Izumi, F., Solid State Ionics, 2002, 152-153, 207-216. [10] Shaju, K.M.; Subba Rao, G.V.; Chowdari, B.V.R., Electrochim.Acta, 2002, 48, 145-151.