4860
J. Phys. Chem. B 2001, 105, 4860-4866
Relationship between Chemical Bonding Character and Electrochemical Performance in Nickel-Substituted Lithium Manganese Oxides Hyo-Suk Park, Seong-Ju Hwang, and Jin-Ho Choy* National Nanohybrid Materials Laboratory, School of Chemistry and Molecular Engineering, Seoul National UniVersity, Seoul 151-747, Korea ReceiVed: January 7, 2001; In Final Form: March 3, 2001
Nickel-substituted LiMn1-xNixO2 (0 e x e 0.1) layered oxides have been prepared and characterized in order to examine the effect of Ni substitution on the chemical bonding nature and electrochemical property of layered lithium manganate. From X-ray diffraction and micro-Raman spectroscopic analyses, it is found that all of the nickel-substituted compounds are crystallized in an R-NaFeO2-type layered structure with monoclinic symmetry. The electrochemical measurements demonstrate that the replacement of Mn with Ni gives rise to only a slight improvement in electrochemical performance, illustrating the fact that Ni substitution is not so effective in reducing the capacity loss of layered lithium manganate. According to X-ray absorption spectroscopic (XAS) analyses at the Mn and Ni K edges, it is evident that the oxidation state of manganese in LiMn1-xNixO2 is increased by substituting the trivalent manganese ion with a divalent nickel ion, which leads to the reduction of the Jahn-Teller (JT) distortion around the manganese ion. However, the XAS results on the delithiated/relithiated LiMn1-xNixO2 compounds reveal that, regardless of Ni content, the irreversible transition to spinel structure is caused by Li deintercalation-intercalation reactions because of an incomplete fixation of the nickel ion accompanied by displacement of the manganese ion. On the basis of the present experimental findings, it can be concluded that the capacity loss of layered lithium manganate during the cycling process is primarily due to the migration of manganese, rather than the presence of the JT distortion.
Introduction Recently, lithium manganese oxides have attracted intense research interest as prospective cathode materials in lithium secondary batteries, because they are less expensive and less toxic than lithium cobalt oxide which is being widely commercialized.1,2 In this context, extensive research has been conducted in order to develop new types of Mn-based materials as well as to improve the electrochemical performance of existing lithium manganese oxides.3-9 Among the various lithium manganate systems, layered LiMnO2 has received special attention because of its large theoretical charge capacity.3,4,10,11 However, this material also suffers from severe capacity fading like the spinel LiMn2O4, which should be related to the JahnTeller (JT) distortion induced by high spin Mn+III (3d4) ion and/ or to the migration of manganese during Li deintercalationintercalation.10 In fact, it was strongly expected that the layered LiMnO2 with its monoclinic structure exhibits more serious capacity fading compared to the spinel LiMn2O4, because all of the manganese ions in this material are JT active. A recent study on the cobalt-substituted Li(Mn1-yCoy)O2 electrode demonstrated that the substitution of only 10% Mn with Co remarkably improves the cycling characteristic of layered lithium manganate, which would be ascribed to the suppression of monoclinic distortion.12 In light of this work, we have tried to prepare nickel-substituted LiMn1-xNixO2 layered oxides, because the substitution of Mn with Ni+II is considered to be more effective in depressing JT distortion through the enhancement of Mn oxidation state, compared to substitution with Co+III. * To whom correspondence should be addressed. Tel: +82-2-880-6658. Fax: +82-2-872-9864. E-mail:
[email protected].
In this work, we have synthesized nickel-substituted LiMn1-xNixO2 layered oxides and characterized their physicochemical properties by using X-ray diffraction (XRD), micro-Raman spectroscopy, and electrochemical measurements. In addition, Mn and Ni K-edge X-ray absorption spectroscopic (XAS) analyses have also been carried out for LiMn1-xNixO2 and their chemically delithiated/relithiated derivatives in order to probe the effects of nickel substitution and Li deintercalationintercalation on the crystal and electronic structures of the layered lithium nickel manganates. XAS would be very suitable for this purpose because this method allows us to determine the oxidation state and local atomic arrangement of the absorbing metal ion even for poorly crystallized derivatives after cycling process.10,11,13-15 Experimental Section Sample Preparation. The polycrystalline LiMn1-xNixO2 samples were prepared by the ion exchange reaction of R-NaMn1-xNixO2 with LiBr.10,11 The precursor R-NaMn1-xNixO2 was synthesized by heating a mixture of Na2CO3, Mn2O3, and NiO (mole ratio ) 1.1:1 - x:2x) at 725-745 °C for 40 h with intermittent grindings. The ion exchange reaction was carried out by reacting R-NaMn1-xNixO2 with 10 equiv lithium bromide in an n-hexanol solution at 148 °C for 48 h. The resulting precipitate was filtered, washed with n-hexanol and methanol, and dried in a vacuum. The chemically cycled samples were obtained by performing a delithiation reaction with a 2.5 M H2SO4 aqueous solution followed by a relithiation reaction with a 1.6 M n-BuLi solution in hexane.10 Sample Characterization. The crystal structures of LiMn1-xNixO2 and their chemically cycled derivatives were studied by
10.1021/jp010079+ CCC: $20.00 © 2001 American Chemical Society Published on Web 05/03/2001
Nickel-Substituted Lithium Manganese Oxides
J. Phys. Chem. B, Vol. 105, No. 21, 2001 4861
TABLE 1: Lattice Parameters, Crystal Symmetries, Unit Cell Volumes, and Chemical Formulas of the Layered LiMn1-xNixO2 Compounds and Their Chemically Cycled Derivatives sample
a (Å)
b (Å)
c (Å)
β (deg)
Vc
chemical formula
LiMnO2 LiMn0.95Ni0.05O2 LiMn0.9Ni0.1O2 chemically cycled LiMnO2 chemically cycled LiMn0.95Ni0.05O2 chemically cycled LiMn0.9Ni0.1O2
5.443 5.443 5.445 8.234 8.236 8.240
2.810 2.811 2.812
5.392 5.397 5.399
116.012 115.985 116.000
74.12 74.23 74.29 558.25 558.60 559.56
Li1.02MnO2 Li1.01Mn0.96Ni0.04O2 Li1.01Mn0.89Ni0.11O2 Li0.36H0.25MnO2 Li0.53H0.11Mn0.97Ni0.03O2 Li0.57H0.08Mn0.94Ni0.06O2
XRD measurements using Ni-filtered Cu KR radiation with a graphite diffracted beam monochromator, and their chemical compositions were determined by atomic absorption (AA) spectrometry. For the chemically cycled products, thermogravimetric analysis (TGA) was also performed to investigate the possibility of proton intercalation during acid treatment. From the weight loss for the temperature region 150-300 °C corresponding to the dehydroxylation reaction, the number of inserted protons could be calculated.10 To confirm the crystal symmetry of nickel-substituted compounds, the Raman measurements were carried out at room temperature using a JobinYvon/Atago Bussan T64000 triple spectrometer equipped with microoptics. The samples were excited with the 514.5 nm line of an Ar+ laser. All of the present spectra were obtained by backscattering from the polycrystalline sample. The resolution of the present spectra was 3-4 cm-1. For the purpose of preventing possible thermal damage of the sample, the power of the incident laser light was maintained at less than 5 mW. After each measurement, the sample surface was thoroughly checked to remove the possibility of spectral modification caused by surface degradation. The electrochemical measurements were carried out with a cell of Li/1M LiPF6 in EC:DEC (50:50 v/v)/LiMn1-xNixO2, which was assembled in a drybox. The composite cathode was prepared by mixing thoroughly the active LiMn1-xNixO2 cathode material (85%) with 10% of acetylene black and 5% of PTFE (poly(tetrafluoroethylene)). All of the experiments were performed in a galvanostatic mode with Arbin BT 2043 multichannel galvanostat/potentiostat in the voltage range of 2.6-4.3 V. X-ray Absorption Measurement. The XAS spectra were obtained for LiMn1-xNixO2 (x ) 0, 0.05, and 0.1) and their chemically cycled derivatives by using the extended X-ray absorption fine structure (EXAFS) facility installed at the beam line 7C at the Photon Factory in Tsukuba.16 The XAS data were collected at room temperature in a transmission mode using gasionization detectors. All of the present spectra were calibrated by measuring the spectra of Mn and Ni metal foils. The data analysis for the experimental spectra was carried out by the standard procedure as reported previously.10,11 Results and Discussion Powder XRD and Chemical Analyses. The powder XRD patterns of the pristine LiMn1-xNixO2 (x ) 0, 0.05, and 0.1) compounds are shown in Figure 1, together with those of their chemically cycled derivatives. For the pristine LiMn1-xNixO2 oxides, all of the intense XRD peaks can be well indexed on the basis of the R-NaFeO2-type layered structure with a monoclinic distortion, except for some small reflections corresponding to the impurity Li2MnO3 and orthorhombic LiMnO2 phase. Upon completion of the chemical delithiation-relithiation process, the XRD patterns of the layered LiMn1-xNixO2 compounds are remarkably altered to rather simpler patterns with broad features.17 Although there is a notable variation of peak intensity and width depending on the Ni content, the overall XRD features of the cycled derivatives appear to be nearly the
Figure 1. Powder XRD patterns for the layered LiMn1-xNixO2 compounds, with x ) (a) 0, (b) 0.05, and (c) 0.1, and (d-f) the chemically cycled derivatives of (a-c).
same as those of the spinel LiMn2O4, suggesting that the chemical cycling induces a structural transition from the layered to the spinel. Furthermore, the common XRD features for all of the chemically cycled LiMn1-xNixO2 compounds presented here imply that the substitution of Mn with Ni is not effective in hindering structural modification during Li deintercalationintercalation. As listed in Table 1, the lattice parameters and unit cell volumes were calculated from least-squares fitting analysis. The c-axis lattice parameter of layered LiMn1-xNixO2 is determined to be notably increased upon Ni substitution, whereas the in-plane cell parameters are not significantly changed, as observed for the case of chromium-substituted LiMn1-xCrxO2 layered oxides.11 In this regard, the increase of basal spacing is surely responsible for the expansion of cell volume induced by Ni substitution. As will be discussed in the Mn K-edge XANES section, the replacement of the Mn+III ion with a Ni+II ion gives rise to the partial formation of a Mn+IV ion. In accordance with the previous EXAFS studies for Nisubstituted spinel lithium manganates,18 it was found that both the divalent nickel and the tetravalent manganese ions in the lithium manganate lattice are commonly stabilized in the regular octahedra with d(Ni+II - O) ) 2.07 Å and d(Mn+IV - O) ) 1.90 Å, respectively. With consideration of the fact that the c axis lattice parameter of layered LiMnO2 phase is mainly dependent on equatorial metal-oxygen bond distance,11 the increase of the c axis lattice parameter upon substitution of Mn with Ni can be understood as a result of the replacement of the shorter (Mn+III - Oeq) bond (d ) 1.91 Å) with the longer (Ni+II, Mn+IV - Oeq) bond (d ) 1.99 Å ) (2.07 Å + 1.90 Å)/2). The chemical formulas of the layered LiMn1-xNixO2 compounds and their cycled derivatives estimated from AA and
4862 J. Phys. Chem. B, Vol. 105, No. 21, 2001
Figure 2. Micro-Raman spectra for the layered LiMn1-xNixO2 compounds, with x ) (a) 0, (b) 0.05, and (c) 0.1, and (d) tetragonal spinel Li2Mn2O4.
TGA analyses are summarized in Table 1, and they indicate that the nickel ion is surely incorporated into the crystal lattice of layered lithium manganate. It is also found that the H+/Li+ ratio in the chemically cycled compounds is markedly decreased with increasing Ni substitution rate, which can be interpreted as a result of the enhancement of the Mn oxidation state upon Ni substitution. That is, because the extraction of interlayer lithium ion is achieved through the disproportionation of trivalent manganese ion (2Mn+III f Mn+II(aq) + Mn+IV),19 the increase of the Mn oxidation state reduces the number of removable lithium ions by decreasing the concentration of Mn+III ions. Moreover, if we take into account the fact that a proton is incorporated into the interlayer space to relieve the electrostatic repulsion between MnO2 layers caused by Li extraction,10 the incomplete removal of interlayer lithium ions is expected to hinder the introduction of protons into the layered lattice during acid treatment, giving rise to a decrease of the H+/Li+ ratio in the delithiated/relithiated products. Micro-Raman Spectroscopic Analysis. The Raman spectra of the pristine LiMn1-xNixO2 (x ) 0, 0.05, and 0.1) compounds are presented in Figure 2 in comparison with the reference spectrum of tetragonal spinel Li2Mn2O4. As expected from the group theoretical calculation, the layered LiMnO2 with monoclinic C2/m symmetry shows three Raman peaks at 422, 481, and 603 cm-1, corresponding to three Raman-active 2Ag + Bg modes. Such a spectral feature of LiMnO2 is markedly different from that of tetragonal spinel Li2Mn2O4, which shows four phonon lines in the frequency range 200-650 cm-1. With consideration of the fact that their similar cation ordering makes it difficult to discriminate both phases by using diffraction tools,20 micro-Raman spectroscopy can be regarded as a powerful method to differentiate various lithium manganates whose crystal structures are closely related to each other. As can be seen from Figure 2, all of the present LiMn1-xNixO2 compounds exhibit nearly the same spectral characteristics, implying that the monoclinic layered structure is maintained for the substitution range of 0 e x e 0.1. However, the intensity of Raman peaks is significantly decreased with increasing Ni content, as reported previously for nickel-substituted lithium cobalt oxides.21 Such a peak suppression can be attributed to
Park et al.
Figure 3. Discharge capacities of the layered LiMn1-xNixO2 compounds with x ) 0 (squares), 0.05 (circles), and 0.1 (triangles). The electrochemical measurements were carried out in the potential range of 2.6-4.3 V with the applied current density of 0.5 mA/cm2.
the increase of electrical conductivity caused by Ni substitution, leading to the reduction of optical skin depth of the incident laser beam. Electrochemical Performance. The electrochemical performance of Ni-substituted LiMn1-xNixO2 has been examined to probe the effect of Ni substitution on the reversibility of Li deintercalation-intercalation in the layered lithium manganate. The discharge capacity of layered LiMn1-xNixO2 (x ) 0, 0.05, and 0.1) oxides is plotted in Figure 3 as a function of cycle number. Among the materials presented here, the LiMn0.95Ni0.05O2 compound with 5% Ni substitution rate shows the best electrochemical property. In the case of an applied current density of 0.5 mA/cm2, this compound retains a discharge capacity of ∼119 mAh/g after the first electrochemical cycle, which is slightly greater than the corresponding value for the unsubstituted LiMnO2 compound (∼110 mAh/g). However, in comparison with the previous result for the cation-substituted LiMn1-yMyO2 (M ) Cr and Co),11,12 the Ni substitution is found not to be so fruitful in improving the electrochemical performance of layered lithium manganate. In addition, we measured the discharge capacity of LiMn0.95Ni0.05O2 for the extended cycling to examine the effect of Ni substitution on the long cycling characteristics of layered lithium manganates. As shown in Figure 4, the discharge capacity of LiMn0.95Ni0.05O2 is increased to ∼120 mAh/g after the 15th cycle. It is also observed that two pseudoplateaus near the 3 and 4 V region in the discharge curve, indicative of the formation of the spinel phase,1 become evident as the charge-discharge cycle proceeds. On the basis of this experimental finding, it is concluded that the substitution of Mn with Ni cannot effectively prevent the irreversible transition from layered to spinel structure. Mn K-Edge XANES Analysis. The variation of the chemical bonding character of layered lithium manganates upon nickel substitution has been studied by using X-ray absorption nearedge structure (XANES) spectroscopy. The Mn K-edge XANES spectra for the layered LiMn1-xNixO2 (x ) 0, 0.05, and 0.1) and their chemically cycled derivatives are shown in Figure 5 in comparison with the reference spectra for cubic spinel LiMn2O4, Mn2O3, and MnO2.22 The position of the edge jump is found to be slightly higher for the nickel-substituted
Nickel-Substituted Lithium Manganese Oxides
Figure 4. Discharge profiles of the Ni-substituted LiMn0.95Ni0.05O2 compound after the first cycle (solid lines), the fifth one (dotted lines), the tenth one (dashed lines), and the fifteenth one (dot-dashed lines). The electrochemical measurements were carried out in the potential range of 2.6-4.3 V with the applied current density of 0.5 mA/cm2.
Figure 5. Mn K-edge XANES spectra for (a) the layered LiMn1-xNixO2 compounds and (b) the chemically cycled LiMn1-xNixO2 derivatives, in comparison with those for the references (c) cubic spinel LiMn2O4, (d) Mn2O3, and (e) MnO2. For the spectra a and b, the solid, dotted, and dashed lines represent the data for the samples with Ni content (x) ) 0, 0.05, and 0.1, respectively.
LiMn1-xNixO2 than for the unsubstituted LiMnO2, which suggests that the average oxidation state of manganese is surely enhanced by replacing Mn+III ion with Ni+II ion. All of the spectra presented here exhibit a small preedge peak P, which is assigned to the transition from the core 1s level to an unoccupied 3d state. Because the intensity of this peak is proportional to the distortion of local structure around the absorbing metal ion from inversion symmetry,23 the negligible intensity of this peak implies that all of the manganese ions in the samples are stabilized in the octahedral site with an inversion center. In the main-edge region, there are two strong peaks corresponding to the dipole-allowed transitions from the core 1s level to the unoccupied 4p level. On the basis of the previous simulation
J. Phys. Chem. B, Vol. 105, No. 21, 2001 4863
Figure 6. Ni K-edge XANES spectra for (a) the layered LiMn1-xNixO2 compounds and (b) the chemically cycled LiMn1-xNixO2 derivatives, in comparison with those for the references (c) cubic spinel LiMn1.8Ni0.2O4, (d) NiO, and (e) LiNiO2. For the spectra a and b, the solid and dotted lines represent the data for the samples with Ni content (x) ) 0.05 and 0.1, respectively.
results,10 the presence of the strong peak A for LiMn1-xNixO2 can be regarded as clear evidence of the existence of trivalent manganese ion stabilized in the tetragonally distorted octahedra over the present Ni-substitution range (0 e x e 0.1). The intensity of this peak A is observed to decrease with increasing Ni content, suggesting that the Ni substitution relieves the JT distortion around manganese as a result of the enhancement of the Mn oxidation state. On the other hand, the chemical delithiation-lithiation process gives rise to a blue-shift of the edge position, indicating an increase of Mn oxidation state because of the irreversibility of lithium deintercalationintercalation. This is further evidenced by the displacement of the preedge peak P toward the high-energy side. In addition, the chemical cycling process also induces remarkable spectral modifications in the main-edge region. The resulting spectra of chemically cycled samples appear to be nearly the same as that of cubic spinel LiMn2O4, which confirms the above XRD and electrochemical measurement results in which the layered structure of LiMn1-xNixO2 is changed into the spinel-type structure upon Li deintercalation-intercalation reaction. Ni K-Edge XANES Analysis. The Ni K-edge XANES spectra for the layered LiMn1-xNixO2 (x ) 0.05 and 0.1) and their chemically cycled derivatives are presented in Figure 6, together with those for the references of spinel LiMn1.8Ni0.2O4, NiO, and LiNiO2.22 The edge energies of all of the lithium nickel manganates under investigation are almost identical to that of NiO, verifying the divalent oxidation state of nickel in these compounds. Like the Mn K-edge, all of the present LiMn1-xNixO2 compounds commonly exhibit a small preedge peak P related to the quadrupole-allowed 1s f 3d transition, which indicates that the nickel ions are surely incorporated into the octahedral site. In the main-edge region, both of the pristine LiMn1-xNixO2 oxides show common spectral characteristics, suggesting that the nickel ions are stabilized in almost the same chemical environment for these compounds. After the chemical delithiation-relithiation process, the main-edge spectra of LiMn1-xNixO2 become rather similar to that of spinel LiMn1.8Ni0.2O4, indicating the fact that the local chemical environment around nickel is
4864 J. Phys. Chem. B, Vol. 105, No. 21, 2001
Park et al. TABLE 2: Results of Nonlinear Least-Squares Curve-Fitting Analysis for the Mn K-Edge EXAFS Spectra of the Layered LiMn1-xNixO2 Compounds sample LiMnO2
LiMn0.95Ni0.05O2
LiMn0.9Ni0.1O2
Figure 7. (a) Fourier transformed Mn K-edge EXAFS spectra and (b) their inverse Fourier transforms for the layered LiMn1-xNixO2 compounds with x ) (i) 0, (ii) 0.05, and (iii) 0.1. The range over which the Fourier filtering has been made is shown by the arrows. The solid lines and empty circles represent the fitted and experimental data, respectively.
changed into the spinel-like cation ordering by Li deintercalation-intercalation reaction, as in the case of manganese ion. Mn K-Edge EXAFS Analysis. The local structure of manganese and nickel ions in the layered LiMn1-xNixO2 compounds has been quantitatively determined by using EXAFS spectroscopy. The k3-weighted Mn K-edge EXAFS spectra for LiMn1-xNixO2 (x ) 0, 0.05, and 0.1) are Fourier transformed in Figure 7a. All of the layered lithium manganates exhibit three intense Fourier transform (FT) peaks at ∼1.6, ∼2.2, and ∼2.7 Å, which are attributed to the (Mn-O), (Mn-M), and (MnM) shells (M ) Mn or Ni), respectively.24 Because the splitting of (Mn-M) shells originates from the JT distortion of the trivalent manganese ion, the observation of three FT peaks for the Ni-substituted phases confirms that the monoclinic symmetry of layered lithium manganate is still maintained even after Ni substitution, as shown by the micro-Raman results. To perform nonlinear curve fitting analysis, the FT spectra were inversely Fourier transformed to k space (Figure 7b). All of the present compounds show the similar EXAFS oscillation, which can be well reproduced on the basis of R-NaFeO2-type layered structure with a monoclinic distortion. The best fitting results are compared to the experimental spectra in Figure 7b and the fitted structural parameters are summarized in Table 2. The substitution of Mn with Ni gives rise to a decrease in the (Mn-Oeq) and (Mn-Oax) bond distances, confirming the oxidation of
bond
CN
R (Å)
σ2 (10-3 × Å2)
(Mn-Oeq) (Mn-Oax) (Mn-Μ) (Mn-M)
4 2 2 4
1.91 2.32 2.80 3.04
1.69 3.32 3.00 6.12
(Mn-Oeq) (Mn-Oax) (Mn-M) (Mn-M)
4 2 2 4
1.90 2.29 2.79 3.03
2.96 4.49 3.03 8.28
(Mn-Oeq) (Mn-Oax) (Mn-M) (Mn-M)
4 2 2 4
1.91 2.25 2.80 3.03
4.13 7.76 3.68 9.06
manganese caused by Ni substitution. Such shortening of the manganese-oxygen bond is estimated to be more pronounced for the axial bond than for the equatorial one, highlighting the reduction of the degree of JT distortion around the manganese ion induced by Ni substitution. In contrast, the (Mn-M) bond distance (M ) Mn or Ni) is nearly independent of the Ni substitution rate, which can be understood as a compromise between the following two effects: (i) the replacement of the shorter (Mn-Mn) bond with a longer (Mn-Ni) one and (ii) the increase of Mn oxidation state leading to the decrease of average (Mn-M) bond distances. It is also found that the Debye-Waller factors are enhanced with increasing nickel concentration, manifesting a remarkable increase of structural disorder upon cation substitution. Such a phenomenon was already reported for the Cr-substituted lithium manganese oxides.11 On the other hand, the influence of the chemical delithiationrelithiation process on the local atomic arrangement around manganese has also been probed by performing Mn K-edge EXAFS analyses for the chemically cycled LiMn1-xNixO2 (x ) 0, 0.05, and 0.1). As shown in Figure 8a, there is a close similarity in the overall FT features between the cycled derivatives and the spinel LiMn2O4. The two peaks corresponding to the (Mn-O) and (Mn-M) bonding pairs are commonly observed at around 1.5 and 2.5 Å for these compounds. However, in case of the cycled derivatives, the second FT peak shows a smaller magnitude than the first FT peak, as reported for the nanocrystalline spinel sample.25 This finding demonstrates that the chemical cycling results in severe structural disorder especially for the distant coordination shells. In light of the spectral resemblance between the cycled derivatives and the spinel LiMn2O4, an attempt was made to fit the spectra of the former compounds with the cubic spinel structure, in such a way that we were able to obtain reasonable fitting results (Figure 8b). As listed in Table 3, the bond distances and DebyeWaller factors are determined to be almost independent of the Ni content, reflecting the fact that the Ni substitution has little influence on the local structure around the manganese ion in these cycled derivatives. Such experimental findings provide further support for the above conclusion that nickel substitution is ineffective in suppressing the structural transition to spineltype cation ordering upon application of the Li deintercalationintercalation process. Ni K-Edge EXAFS Analysis. The FTs of k3-weighted Ni K-edge EXAFS spectra for the layered LiMn0.9Ni0.1O2 compound and its chemically cycled derivative are plotted in Figure 9a. For both compounds, the two FT peaks corresponding to the (Ni-O) and (Ni-M) coordination shells are discernible at ∼1.6 and ∼2.6 Å, suggesting that the substituted nickel ion is
Nickel-Substituted Lithium Manganese Oxides
J. Phys. Chem. B, Vol. 105, No. 21, 2001 4865
Figure 8. (a) Fourier transformed Mn K-edge EXAFS spectra and (b) their inverse Fourier transforms for the chemically cycled LiMn1-xNixO2 compounds, with x ) (i) 0, (ii) 0.05, and (iii) 0.1, and (iv) the reference LiMn2O4 spinel. The range over which the Fourier filtering has been made is shown by the arrows. The solid lines and empty circles represent the fitted and experimental data, respectively.
TABLE 3: Results of Nonlinear Least-Squares Curve-Fitting Analysis for the Mn K-Edge EXAFS Spectra of the Chemically Cycled LiMn1-xNixO2 Compounds and Reference LiMn2O4 Spinel sample
bond
CN
R (Å)
σ2 (10-3 × Å2)
chemically cycled LiMnO2
(Mn-Ο) (Mn-Mn)
6 6
1.91 2.88
4.74 9.75
chemically cycled LiMn0.95Ni0.05O2
(Mn-Ο) (Mn-Mn)
6 6
1.91 2.87
4.71 8.81
chemically cycled LiMn0.9Ni0.1O2
(Mn-Ο) (Mn-Mn)
6 6
1.91 2.87
4.69 8.09
spinel LiMn2O4
(Mn-Ο) (Mn-Mn)
6 6
1.91 2.91
4.57 5.00
situated in a regular octahedron. Such a regular symmetry of NiO6 octahedra is surely attributed to the absence of JT distortion in divalent nickel ion with an electronic configuration of [Ar]3d.8 In this regard, we have attempted to fit the spectra of both compounds on the basis of the regular NiO6 octahedron with six neighboring metal ions separated by the same distance, in such a way that we were able to get reasonable fitting results (Figure 9b). As summarized in Table 4, the nickel ions in both compounds are found to be stabilized in the regular octahedron with an (Ni-O) bond distance of 2.06 Å, which is fully consistent with the bond length of (Ni+II-O) in Ni-substituted
Figure 9. (a) Fourier transformed Ni K-edge EXAFS spectra and (b) their inverse Fourier transforms for (i) the layered LiMn0.9Ni0.1O2 compound and (ii) its chemically cycled derivative. The range over which the Fourier filtering has been made is shown by the arrows. The solid lines and empty circles represent the fitted and experimental data, respectively.
TABLE 4: Results of Nonlinear Least-Squares Curve-Fitting Analysis for the Ni K-Edge EXAFS Spectra of the Layered LiMn0.9Ni0.1O2 and Its Chemically Cycled Derivative bond
CN
R (Å)
σ2 (10-3 × Å2)
LiMn0.9Ni0.1O2
(Ni-Ο) (Ni-M)
6 6
2.06 2.96
6.88 7.01
chemically cycled LiMn0.9Ni0.1O2
(Ni-Ο) (Ni-M)
6 6
2.06 2.93
10.01 8.53
sample
spinel LiMn1.8Ni0.2O4 (2.07 Å).18 It is also observed that the chemical cycling process leads to the marked shortening of the (Ni-M) bond as well as to the increase of Debye-Waller factors. Moreover, a closer inspection of Figure 9a reveals that there are considerable differences in the FT of both compounds over the distant R range of >3 Å. These findings highlight that the local atomic arrangement around the nickel ion becomes considerably disordered after the cycling process, as conjectured from Ni K-edge XANES results. Judging from the fact that the Li deintercalation-intercalation reaction causes only negligible changes in the local structure of chromium ion stabilized in the layered LiMn1-xCrxO2 lattice,11 we considered the nickel ion to possess relatively poor stability with respect to the cycling process compared to the chromium ion. Such a difference between nickel and chromium ions can be attributed to the
4866 J. Phys. Chem. B, Vol. 105, No. 21, 2001 smaller octahedral site preferential energy of the former ion with respect to the latter ion, resulting in an incomplete fixation of the nickel ion during the cycling process.18 Because the displacement of nickel ions promotes the migration of adjacent manganese ions by reducing the potential barrier for conduction,10 the substitution of Mn with Ni is not effective in preventing the migration of manganese ion into the interlayer lithium site, leading to structural modification to spinel-type cation ordering.26 Conclusion In the present study, we have prepared nickel-substituted LiMn1-xNixO2 (0 e x e 0.1) oxides and their chemically cycled derivatives, to probe the influence of Ni substitution on the chemical bonding nature and electrochemical property of layered lithium manganate. On the basis of the Mn and Ni K-edge XANES results, we can definitely conclude that the oxidation state of manganese in LiMn1-xNixO2 is enhanced by replacing the trivalent manganese ion with a divalent nickel ion. Moreover, the Mn K-edge EXAFS analysis clarifies that Ni substitution gives rises to the reduction of JT distortion in the layered compounds. However, in contrast to our expectation that the depression of JT distortion leads to the improvement of capacity retention, the Ni-substituted LiMn1-xNixO2 compounds exhibit nearly the same electrochemical performance as the unsubstituted LiMnO2. Such phenomena can be understood by performing XRD and XAS analyses for the delithiated-relithiated LiMn1-xNixO2 derivatives, which reveal that the nickel substitution cannot prevent the irreversible transition from layered to spinel structure during Li deintercalation-intercalation reactions. This structural modification caused by the cycling process should originate from an incomplete fixation of the nickel ion accompanied by a displacement of the manganese ion. On the basis of the present experimental findings, it can be concluded that capacity loss of layered lithium manganate during the cycling process is primarily attributed to the migration of manganese, rather than the JT distortion around manganese ion. Acknowledgment. This work was supported in part by the Korean Ministry of Science and Technology through the 1999 National Research Laboratory (NRL) project. The authors are grateful to Prof. M. Nomura for helping us to get the XAS data in the Photon Factory as well as to Prof. M. Yoshimura and Prof. M. Kakihana for helping us to get the micro-Raman data. H.-S.P. thanks the Korean Ministry of Education for the Brain Korea 21 fellowship. References and Notes (1) Thackeray, M. M. Prog. Solid State Chem. 1997, 25, 1. (2) Thackeray, M. M.; Johnson, P. J.; de Piccioto, L. A.; Bruce, P. G.; Goodenough, J. B. Mater. Res. Bull. 1984, 19, 179. (3) Armstrong, A. R.; Bruce, P. G. Nature 1996, 381, 499. (4) Capitaine, F.; Gravereau, P.; Delmas, C. Solid State Ionics 1996, 89, 197.
Park et al. (5) Kim, J.; Manthiram, A. Nature 1997, 390, 265. (6) Kim, S. H.; Kim, S. J.; Oh, S. M. Chem. Mater. 1999, 11, 557. (7) Gummow, R. J.; Liles, D. C.; Thackeray, M. M. Mater. Res. Bull. 1993, 28, 1249. (8) Gummow, R. J.; De Kock, A.; Thackeray, M. M. Solid State Ionics 1994, 69, 59. (9) Guohua, L.; Ikuta, H.; Uchida, T.; Wakihara, M. J. Electrochem. Soc. 1996, 143, 178. (10) Hwang, S. J.; Park, H. S.; Choy, J. H.; Campet, G. Chem. Mater. 2000, 12, 1818. (11) Hwang, S. J.; Park, H. S.; Choy, J. H.; Campet, G. J. Phys. Chem. B 2000, 104, 7273. (12) Armstrong, A. R.; Gitzendanner, R.; Robertson, A. D.; Bruce, P. G. Chem. Commun. 1998, 1833. (13) Treuil, N.; Labruge`re, C.; Menetrier, M.; Portier, J.; Campet, G.; Deshayes, A.; Frison, J. C.; Hwang, S. J.; Song, S. W.; Choy, J. H. J. Phys. Chem. B 1999, 103, 2100. (14) Ammundsen, B.; Jones, D. J.; Rozie`re, J.; Villain, F. J. Phys. Chem. B 1998, 102, 7939. (15) Nakai, I.; Nakagome, T. Electrochem. Solid-State Lett. 1998, 1, 259. (16) Oyanagi, H.; Matsushida, T.; Ito, M.; Kuroda, H. KEK Report 1984, 83, 30. (17) Such a peak broadening is attributed to the fact that the chemical cycling process frustrates severely the long-range structural order in the layered LiMn1-xNixO2 oxides. (18) Hwang, S. J.; Park, H. S.; Choy, J. H.; Campet, G. J. Phys. Chem. B 2001, 105, 335. (19) Larcher, D.; Courjal, P.; Herrera Urbina, R.; Ge´rand, B.; Blyr, A.; du Pasquier, A.; Tarascon, J. M. J. Electrochem. Soc. 1998, 145, 3392. (20) Jang, Y. I.; Huang, B.; Chiang, Y. M.; Sadoway, D. R. Electrochem. Solid State Lett. 1998, 1, 13. (21) Inaba, M.; Iriyama, Y.; Ogumi, Z.; Todzuka, Y.; Tasaka, A. J. Raman Spectrosc. 1997, 28, 613. (22) The reference LiMn2-yNiyO4 spinel compounds were prepared by solid-state reaction between Li2CO3, Mn2O3, and NiO at 750 °C. (23) Hahn, J. E.; Scott, R. A.; Hodgson, K. O.; Doniach, S.; Desjardins, S. R.; Solomon, E. I. Chem. Phys. Lett. 1982, 88, 595. (24) In the case of layered lithium manganate lattice consisting of the edge-shared MnO6 octahedra with JT distortion, a manganese ion possesses six neighboring manganese ions with two different (Mn-Mn) bond distances, that is, two nearest manganese ions ligated by equatorial oxygens (i.e., Mneq) and four next-nearest Mn ions ligated by axial oxygens (i.e., Mnax). Because two kinds of neighboring manganese ions, Mneq and Mnax, are located in the same crystallographic position (2a site), the nickel ion would be randomly substituted for both manganese ions. Moreover, from the presence of two FT peaks around 2-3 Å, it can be reasonably expected that the (Mn-Nieq) pair would show the similar bond distance to the (MnMneq) pair, rather than to the (Mn-Mnax) pair, whereas the (Mn-Niax) bond distance would be closer to the (Mn-Mnax) one. In this context, the FT peaks at ∼2.2 and ∼2.7 Å are attributable to the nearest (Mn-Mneq, Nieq) shell and the next-nearest (Mn-Mnax, Niax) shell, respectively. (25) Choy, J. H.; Kim, D. H.; Kwon, C. W.; Hwang, S. J.; Kim, Y. I. J. Power Sources 1999, 77, 1. (26) As revealed from the present Mn K-edge XANES results, the replacement of Mn+III ion with Ni+II ion gives rise to the formation of tetravalent Mn+IV ion with 3d3 electronic configuration and, hence, with a strong octahedral-site preference like Cr+III ion. However, in the course of the repeated Li deintercalation-intercalation process, the Mn+IV ion would take part in the reduction-oxidation reaction, which surely differs from the substituted Cr ion in LiMn1-xCrxO2 electrode with a fixed oxidation state. This implies that a Mn+IV ion with large octahedral site stabilization energy can also be reduced to an unstable Mn+III ion upon electrochemical cycling, and therefore, it cannot contribute effectively to the stabilization of lithium manganate lattice, compared to the trivalent chromium ion.