Appl. Phys. A 74 [Suppl.], S1089–S1091 (2002) / Digital Object Identifier (DOI) 10.1007/s003390101196 Applied Physics A Materials Science & Processing Structure determination of high-voltage LiMg δ Ni 0.5-δ Mn 1.5 O 4 spinels for Li-ion batteries F.G.B. Ooms 1 , M. Wagemaker 2, ∗ , A.A. van Well 2 , F.M. Mulder 2 , E.M. Kelder 1 , J. Schoonman 1 1 Laboratory for Inorganic Chemistry, Delft University of Technology, Julianalaan 136, 2628 BL Delft, The Netherlands 2 Interfaculty Reactor Institute, Delft University of Technology, Mekelweg 15, 2629 JB Delft, The Netherlands Received: 18 July 2001/Accepted: 24 October 2001 –  Springer-Verlag 2002 Abstract. A series of cathode materials has been synthesized with the general formula LiMg δ Ni 0.5-δ Mn 1.5 O 4 (δ = 0.00, 0.05 and 0.10). These are promising cathode materials for lithium and Li-ion batteries due to the high voltage (> 4.7V vs. Li/Li + ) and the high energy density (> 570 W h/kg). The cycling stability of these materials is strongly influenced by the method of synthesis and is particularly improved by a very low cooling rate. To study the effect of such slow cooling on the crystal structure, a detailed diffraction an- alysis was performed. Initial X-ray-diffraction (XRD) meas- urements revealed that the materials crystallize in the spinel structure, which is normally refined in the Fd( 3)m space group. Neutron-diffraction (ND) experiments, however, indi- cate space group P4 3 32 and refinements of the ND and XRD patterns result in the site occupations: Li + on 8c, Mg 2+ and Ni 2+ on 4b, Mn 4+ on 12d and O 2- on 24e and 8c. It was also found that, as a function of the Mg content, the cubic lat- tice constant increases from 8.1685 Å (δ = 0.00) to 8.1733 Å (δ = 0.10). PACS: 84.60.Dn; 61.12.Ld The Li-ion batteries available today utilize a ∼ 4 V lithium transition metal oxide as positive electrode and carbon as negative electrode. The transition-metal oxides currently used are LiCoO 2 and to a small extent LiMn 2 O 4 . Research has been directed towards new anode materials like Li 4 Ti 5 O 12 [1, 2] and LiCrTiO 4 [3, 4] with potentials of ∼ 1.5 V vs. Li/Li + . They are expected to be safer and show higher-rate charge/discharge capabilities than carbon anodes. However, combining these anode materials with the 4 V LiMeO 2 ma- terials results in ∼ 2.5 V Li-ion batteries, which voltage is rather low. A way to compensate for this loss in battery poten- tial and energy is to use a high-voltage oxide with a potential in the range of 4–5 V vs. Li/Li + . This approach has been shown by Panero et al. [5] with the system Li 4 Ti 5 O 12 vs. Li 2 Co 0.4 Fe 0.4 Mn 3.2 O 8 and, recently, by our laboratory with ∗ Corresponding author. (Fax: +31-15/2788303, E-mail: m.wagemaker@iri.tudelft.nl) the system LiCrTiO 4 vs. LiMg 0.1 Ni 0.4 Mn 1.5 O 4 [4]. It was found that very slow cooling enhanced the cycling stability of these new active materials. For a better understanding of the (differences in) performance of these materials, structural characterization is essential. 1 Experimental The range of materials LiMg δ Ni 0.5-δ Mn 1.5 O 4 (δ = 0.00, 0.05 and 0.10) has been prepared in an identical way as the LiMg 0.1 Ni 0.4 Mn 1.5 O 4 material, described in a previous pa- per [4]. The cathode materials were made with a combination of a sol–gel synthesis and solid-state processing. The pre- cursors, LiOH·H 2 O (Merck), Mg(CH 3 COO) 4 ·4H 2 O (Fluka), Ni(CH 3 COO) 2 ·4H 2 O (Aldrich) and Mn(CH 3 COO) 2 ·4H 2 O (Fluka), were ball-milled in distilled water for 15 min. The homogeneous water-based mixture was dried at low tempera- tures to a homogeneous precursor powder and subsequently calcined in air (three times) at 800 ◦ C for a total of 30 h. The heating and cooling rates for the first two calcination steps were 10 ◦ C/min. The powders obtained in each step were ball-milled in hexane for 30 min using a planetary ball mill with agate jars/balls (Fritsch Pulverisette 7). The final dwell time was 6 h to a slow cooling rate of 0.1 ◦ C/min. The characterization of the structures of the active materials was done with X-ray diffraction (XRD) (Bruker AXS D8 ad- vance, Cu-Kα, no filter) and neutron diffraction (ND) (HB3 diffractometer in Petten, The Netherlands) with an incident neutron wavelength of 0.14265 nm. The X-ray and neutron- diffraction data were simultaneously refined using the GSAS Rietveld refinement code. 2 Results The XRD patterns of the three LiMg δ Ni 0.5-δ Mn 1.5 O 4 (δ = 0.00, 0.05 and 0.10) materials are presented in Fig. 1. The materials with δ = 0.00 and δ = 0.05 were not phase- pure and contained traces of a Li x Ni 1-x O (0 < x < 0.33) phase (small XRD peak at 2θ ∼ 43.5 ◦ ). The amount of this