The development of energy storage technique leads to comfortable life. Lithium ion battery (LIB) having high energy density is applied for mobile devices, electric vehicle and storage devices of sustainable energy. With expanding the demands of LIB, there are some problems to be solved. As one of major problems, Co concluded in the cathode material is expensive and poor resource. In order to reduce battery costs, especially, the alternative electrode material to LiCoO2 is needed. LiMnO2 and LiFePO4 are attracted as the alternative material. The electrode material cost is reduced by the usage of abundant material in resource, like Fe or Mn  . In addition, PO4 tetrahedron consisted of strongly covalent bond prevents oxygen decomposition during high charging state. Its structure makes a contribution to safety. Therefore, LiFePO4 is suitable for mass production with the demand expansion and also for safety electrode material.
LiFePO4 has olivine-type structure and belongs to poly-anion group. Padhi reported that poly-anion group material showed reversible Li ion extraction characters and olivine-type LiFePO4 had higher potential than that of the other poly-anion groups  . LiFePO4 is interesting because of its low-cost, flat-voltage characters, good cyclability and good stability. Olivine-type material, that composition expressed in LiMePO4 (Me = Fe, Mn, Co. Ni), is composed of edge-shearing PO4 tetrahedral and MeO6 octahedral. The covalence P-O bonds in the LiFePO4 structure stabilizes Fe(3d)-O(2p) anti-bond, as a result, the redox potential of Fe2+/Fe3+ increases to higher level  . The redox potential is 3.4 V vs. Li+/Li in LiFePO4 and 4.1 V vs. Li+/Li in LiMnPO4  , respectively. In the case of Co or Ni, it is expected to have higher redox voltage. However, low Li ion diffusion character is the problem for practical usage, especially for Co and Ni  . By substituting the multiply-charged ion, the improvements of the various electric behaviors were reported     . The Zr substituted to Me is efficient to prevent the degradation of capacity  . In this way, metal ion substitution is efficient to the improvement of the electric property. Padhi reported that Mn-O-Fe interactions in LiMn0.5Fe0.5PO4 set the redox energy of Mn2+/Mn3+ higher than that of Fe2+/Fe3+  . In addition, in LiMn0.4Fe0.6PO4, Yamada reported the influence of Mn in the redox mechanism  . Because LiMnxFe1-xPO4 has interesting features of two different plateaus (3.5 and 4.1 V), more energy density than that of pure LiFePO4 is obtained.
Since then, many researchers have evaluated the electrical properties of LiFePO4      . The conductivity has considered being a problem of LiFePO4. As one of the solutions for its problem, the addition of conduction assistant like carbon coating was reported   , and the capacity of LiFePO4 at high rate was also improved. On the other hand, the morphology and crystallinity of LiFePO4, which affect to the electrical property, change according to each synthesis method. The products synthesized by solid state reaction have high crystallinity relatively. The finer products synthesized by hydrothermal method are obtained, and its crystallinity tends to depend on synthesis temperature because of presence of amorphous phase or the vacancy of Li site  . This lack of Li is directly related to the electric property. It reported that the products hydrothermally synthesized at the temperature over 180˚C showed good property, when the hydrothermal conditions of olivine material were optimized   . In addition, in order to decrease these influences during electric measurement, olivine materials are heat treated in inert atmosphere. Thus, in the case of hydrothermal process, the both influence of synthesis condition and Mn addition to olivine structure should be considered.
In this study, the structure of Mn substituted LiMnxFe1-xPO4 synthesized by hydrothermal method was evaluated in detail. Substitution of larger size Mn2+ ion is expected to distort the olivine-type structure. As more details, the structural characters of MeO6 can be observed selectively from XAFS analysis with changing X-ray energy. Then, the effect of Mn addition on the local structure was also examined by XAFS.
2. Experimental Procedure
2.1. Synthesis of LiMnxFe1-xPO4
LiOH・H2O, (NH4)2HPO4, FeSO4・7H2O, MnSO4・5H2O (Wako. Ltd.) were used as starting materials. They were dissolved in deionized water, and 1M LiOH, 0.5 M (NH4) HPO4, 0.5M FeSO4, 0.5M MnSO4 were prepared. These reagents were weighted with molar ratio Li:P:Fe:Mn = 2:1:1-x:x (x = 0 - 1) by 0.25, respectively. In order to prevent the oxidation of Fe2+ to Fe3+, distilled water was bubbled by N2 gas, and the synthesis was done under N2 atmosphere. Reagents were mixed in Teflon vessel vigorously. Teflon vessel with mixed solution was hydrothermally treated at 200˚C for 24 hours. Obtained products were filtrated, washed, and dried under vacuum for overnight.
The crystal phase of samples was identified by XRD (Ultima IV, Rigaku Co., Japan) at 2θ = 10˚ - 70˚ with scan rate of 4˚/min using CuKα radiation. The microstructure was observed by FE-SEM (S-4500, Hitachi, Japan) with applied voltage of 10 kV. The specific surface area was measured by nitrogen BET method (BELSORP-mini, microtrac-BEL, Japan) using the data of P/P0 = 10−3 - 10−2. The Rietveld refinement of structural parameter was performed by the analysis software RIETAN-FP  . The refined data range was 2θ = 10˚ - 90˚ by stepping 0.01˚. The composition of the sample was analyzed by ICP (PS-7800, Hitachi Co.). The sample was dissolved in 0.1 M nitric acid, and its solution was measured. The local structure of samples was investigated by XAFS spectra for Fe K-edge and MnK-edge. XAFS data were corrected by transmission mode using Si (111) double crystal monochrometer at BL14B2 in the SPring-8. For the XAFS measurement, the sample was prepared as pellets with the thickness varied to obtain a 0.5 - 1.0 jump at the both Fe K-edge and MnK-edge. The evaluation of XAFS data was conducted using the commercial software “REX2000” (RigakuCo. Ltd., Japan). The vibrational structure was identified by FT-IR (ALPHA-OPT, Bruker Co.) at wave vector range 400 - 4000 cm−1. For FT-IR measurements, the sample was grinded with KBr, and the powder was pressed in a mechanical press to form a translucent pellet.
3. Results and Discussion
3.1. The Microstructure and the Composition
Mn substituted LiMnxFe1-xPO4 was synthesized by hydrothermal process at 200˚C for 24 h. SEM images of the products of Mn substituted LiMnxFe1-xPO4 were shown in Figure 1. The microstructure of the products was finely plate-like
Figure 1. SEM images of the products synthesized by hydrothermal process with Mn addition ratio (a) x = 0, (b) x = 0.25, (c) x = 0.5, (d) x = 0.75, (e) x = 1.0.
with particle size of about 0.5 μm. In case of Mn addition, the same size particle was observed as LiFePO4, and a specific surface area was about 7.8 m2/g. No remarkable difference of the microstructure was observed regardless of Mn addition. The composition of the Mn added products was measured by ICP analysis and shown in Table 1. The ratio x of Mn addition amount was changed from 0 to 1.0 by 0.25. Compared to the ratio x of Mnaddition amount, the Mn/Fe ratio was provided equally, respectively. In addition, Li/(Mn + Fe) ratio was around 0.96 a little less than stoichiometric ratio. Therefore, it was thought that added Mn alternative to Fe was included in the products.
3.2. Structure Refinement of Mn Substituted Olivine Material
The crystal phases of the products were identified by XRD and the structure parameter was analyzed by Rietveld refinement. XRD results of the products were shown in Figure 2. The diffraction peaks were attributed to orthorhombic olivine-type structure, Pnma. No other crystalline peaks, which were attributed to the impurities like Fe2P or Li3PO4, were observed. With increasing the Mn addition amount, the diffraction peaks were shifted to low angle. The Fe2+ ionic radius in coordination number 6 is 61 pm, and the Mn2+ ionic radius in coordination number 6 is 81 pm  . As a reason of peak shift, the lattice spacing was expanded by substituted Mn ion into olivine-structure, of which ionic radius is larger than that of Fe2+. Therefore, it was thought that Mn ion was substituted with Fe ion. The refined lattice constant and the reliability factors (R-factors) were shown in Table 2. Refinement patterns of LiMnxFe1-xPO4 were shown in supplementary file. Mn was considered to be the substitute atom to Fe site, and its occupancy rate was set the stoichiometric value. Isotropic displacement parameter, B parameter, was set the constant value because of having strongly correlation with occupancy rate. In x = 0, R-factors was Rwp = 0.53%, Rp = 0.38%, RB = 2.71%, RF = 1.21% and S = 1.31%. According to R-factors, the calculated pattern was the good agreement with the experiment pattern. The lattice constant
Table 1. The composition of LiMnxFe1-xPO4.
Table 2. The refined lattice constant and reliability factors of LiMnxFe1-xPO4 by Rietveld analysis.
Figure 2. XRD patterns of the products synthesized by hydrothermal process with Mn addition ratio (a) x = 0, (b) x = 0.25, (c) x = 0.5, (d) x = 0.75, (e) x = 1.0.
of x = 0 was a = 1.0338 ± 5 nm, b = 0.5995 ± 4 nm, c = 0.4696 ± 1 nm and V = 0.2910 ± 8 nm3, and it was a little larger than that of the reported products synthesized by hydrothermal reaction  . Its difference was thought because of presence of amorphous phase. In x = 1.0, the lattice volume was expanding up to V = 0.3028 ± 5 nm3. Thus, it was found that the lattice expansion was depended on Mn addition amount for LiMnxFe1-xPO4.
3.3. The Local Structure of Mn Substituted Olivine Material
The local structure of transition metal ion was characterized by XAFS analysis. The valence number of transition metal ion was identified from energy value of the absorption edge, E0, defined as maximum value of derivative of XANES spectra. E0 is located in lower energy with smaller valence number, generally. Fe K-edge and MnK-edge XANES spectra were shown in Figure 3. In both spectra of K-edge, with increasing Mn addition amounts, no difference of XANES spectra of the products was observed remarkably. In Fe K-edge XANES spectra, pre-edge peaks attributed to 1s-3d transition was observed at about 7112 eV. Then, E0 of the products was located in 7119 eV close to that of Fe(II)O. Correspondingly, in MnK-edge, pre-edge peaks and E0 was observed at 6538 eV and 6546 eV close to that of Mn(II)O, respectively. Therefore, it was thought that iron and manganese ions were existed as divalent ion. Next, the results of radius distribution function furrier transformed of Fe K-edge and MnK-edge EXAFS spectra were shown in Figure 4. As each radius distribution function showed the similar curves, and it was thought that similar local structure was reflected to those curves. Radius distribution function was curve fitted based on the following basic EXAFS formula  .
Figure 3. XANES spectra of Fe K-edge and Mn K-edge for the products synthesized by hydrothermal process with Mn addition ratio (a) x = 0, (b) x = 0.25, (c) x = 0.50, (d) x = 0.75, (e) x = 1.0.
Figure 4. Radius distribution function of Fe K-edge and Mn K-edge for the products synthesized by hydrothermal process with Mn addition ratio (a) x = 0, (b) x = 0.25, (c) x = 0.50, (d) x = 0.75, (e) x = 1.0.
where f(k) and δ(k) are scattering properties of the atoms neighboring the excited atom, N is the number of neighboring atoms, R is the distance to the neighboring atom, and σ2 is the disorder in the neighbor distance. The number of coordination atom (C. N.), the bond distance (R) and the Debye-Waller factor (σ) were shown in Table 3. From olivine-type structural model, the first proximity atom around transition metal ion is oxygen and the second is phosphorus. In Fe K-edge, the first peak was attributed to Fe-O bonds. The estimated coordination number of the first peak was 3 and the bond distance in x = 0 was 0.208 nm, and the Debye-Waller factor strongly correlated to the coordination number was 0.098. The obtained coordination number was less than 6 kinds of Fe-O bond distances in FeO6 octahedra. In A. Yamada’s report  , FeO6 octahedra with Pnma symmetry was distorted and the six Fe-O bond distances were 0.206, 0.206, 0.211, 0.221, 0.225 and 0.225 nm, respectively. The number of Fe-O bonds around 0.21 nm is 3, and it was similar to the calculated value. The other Fe-Obonds in FeO6 octahedra have about 0.23 nm of the bond distance. The peak separation attributed to these bonds was difficult due to less wave vector range to use for curvefitting. In LiMnxFe1-xPO4 compounds, Fe-O bond distance was similar value. Corresponding to Fe K-edge, the first peak was attributed to Mn-O bonds in Mn K-edge. The estimated coordination number of the first peak was 3 and the bond distance in x = 1.0 was 0.214 nm larger than the Fe-O bond distance. In LiMnxFe1-xPO4 compounds, Mn-O bond distance was similar value. As a result, it is thought that the Me-O distance unless the Mn addition amount was constant value. MeO6 octahedra and PO4 tetrahedra have the structure that a ridge shared one side, so it suggested that the angle provided by these polyhedral would be changed because of Mn addition. The substitution of larger size Mn2+ into Fe site might distort the olivine structure, especially the MeO6 octahedra.
Table 3. The coordination number (C.N.), bond distance (R) and Debye-Waller factor (σ).
3.4. The Vibrational Structure of PO4 Tetrahedra
The structural distortion by larger MeO6 octahedra was affected to nearly PO4 tetrahedral structure. Mn substituted LiMnxFe1-xPO4 has the PO4 tetrahedra with infrared absorbency. The vibrational structure of PO4 tetrahedra was measured by FT-IR, and its spectra were showed in Figure 5. According to Rulmont et al.  , IR spectra of PO4 bands were attributed in following. In the case of the products of Mn addition ratio x = 0, υ1 symmetric stretching vibration of P-O was 985.3 cm−1, and υ3 asymmetric stretching vibrations were 1053.4, 1095.9 and 1138.4 cm−1. In addition, υ2 symmetric bending of O-P-O was 474.9 cm−1, and υ4 asymmetric bending was 501.9, 552.9, 578.4 and 635.1 cm−1. With increasing the Mn addition ratio x, a part of absorption band around 1000 cm−1 was shifted to blue shift. In x = 0.25, absorption band top was 998.1 cm−1, approached to 1009.4 cm−1 in x = 1.0. The reason of this tendency was why expanded MnO6 structure was affected to the nearly PO4 structure. MeO6 octahedra in the olivine structure occurred the distortion of PO4 and MeO6 zig-zag chains.
Mn substituted LiMnxFe1-xPO4 was synthesized by hydrothermal process, and that crystal structure was in detail evaluated. The microstructure of the products was 0.5 μm size particles, and those compounds were provided equally compared to Mn addition amounts by ICP-analysis. XRD results showed that the products were attributed to orthorhombic olivine-type structure. In addition, the diffraction peaks were shifted to low angle with increasing Mn addition amounts, and it suggested that larger size Mn2+ was substituted in olivine structure, and expanded the lattice spacing. The structure parameter was refined by Rietveld analysis. The lattice constant in x = 0 was a = 1.0338 ± 5 nm, b = 0.5995 ± 4 nm and c = 0.4696 ± 1 nm, and it expanded with increasing Mn addition. The substituted larger size Mn2+ might distort the structure of olivine, especially the MeO6 octahedra. This distortion was confirmed by XAFS analysis. The atom distance of Mn-O was 0.214 nm larger than 0.208 nm of Fe-O. From FTIR, the PO4 vibrational structure was partly changed, so it was thought that MeO structure expansion was affected to the nearly PO4 structure.
Figure 5. FT-IR spectra of the producta synthesized by hydrothermal process with Mn addition ratio (a) x = 0, (b) x = 0.25, (c) x = 0.50, (d) x = 0.75, (e) x = 1.0.
Grant-in-Aid for Scientific Research on Innovative Areas “Nano Informatics” (Grant No. 16H00892) from JSPS.
Figure S1. Observed (red), calculated (dark-blue), and difference (blue) refinement patterns resulting from Rietveld analysis. Green vertical bars denote positions of Bragg reﬂections. (a) LiFePO4, (b) LiMn0.25Fe0.75PO4, (c) LiMn0.5Fe0.5PO4, (d) LiMn0.75Fe0.25PO4, (e) LiMnPO4.