In Situ X-ray Absorption Spectroscopic Study on LiNi0.5Mn0.5O2

Karthikeyan KaliyappanWei XiaoKeegan R. AdairTsun-Kong ShamXueliang Sun ..... J. Chupas, Peter L. Lee, Thomas Proffen, John B. Parise, and Clare P. Gr...
0 downloads 0 Views 248KB Size
Download PDF
Chem. Mater. 2003, 15, 3161-3169

3161

In Situ X-ray Absorption Spectroscopic Study on LiNi0.5Mn0.5O2 Cathode Material during Electrochemical Cycling Won-Sub Yoon,†,‡ Clare P. Grey,† Mahalingam Balasubramanian,‡ Xiao-Qing Yang,‡ and James McBreen*,‡ Department of Chemistry, State University of New York at Stony Brook, Stony Brook, New York 11794-3400, and Brookhaven National Laboratory, Upton, New York 11973 Received February 10, 2003. Revised Manuscript Received May 22, 2003

We have investigated the local electronic and atomic structure of the LiMn0.5Ni0.5O2 electrode during the first charge and discharge process using in situ X-ray absorption spectroscopy (XAS) of the Mn and Ni K-edges. The Ni K-edge structure in the XANES spectrum shifts to higher energy during charge and shifts back reversibly during discharge in the higher voltage region of ∼4 V, whereas the Mn K-edge structure does not appear to exhibit a rigid edge shift. Further Li-ion intercalation during extended discharge in the 1-V plateau leads to the reduction of Mn4+ ions suggesting that the charge compensation in this region is achieved via the reduction of Mn4+ ions to Mn2+. Mn K-edge EXAFS results indicate that a small amount of Li is found in the Ni2+/Mn4+ layers. These Li ions in the transition metal layers are primarily present in the second coordination shell of Mn and not around Ni. Ni K-edge EXAFS fitting results suggest that the oxidation process upon Li deintercalation takes place in two steps: Ni2+ to Ni3+ first, and then Ni3+ to Ni4+.

Introduction LiCoO2 is widely used as a cathode material in commercial secondary lithium batteries because of a number of advantages, including easy preparation and high theoretical specific capacity.1-3 However, the toxicity of cobalt and it’s safety issues represent some of the problems of this material. Thus, extensive research has been carried out over the past 10 years to find alternative cathode materials for LiCoO2 in lithium-ion rechargeable batteries. Recently, layer-structured lithium nickel manganese oxides have been shown to be one of the most promising alternative materials for LiCoO2 because their electrochemical and safety characteristics are comparable to or better than those of LiCoO2.4,5 Ohzuku et al. showed that lithium nickel manganese oxide represents a possible alternative to LiCoO2 for advanced lithium batteries, in terms of its operating voltage, capacity, cycleability, safety, and materials economy.4 Lu et al. reported that the Li[NixLi(1/3-2x/3)Mn(2/3-x/3)]O2 system with Ni2+ and Mn4+ ions can be cycled between 2.0 and 4.6 V to give a stable capacity of about 200, 180, or 160 mAh/g for x ) 1/3, 5/12, or 1/2, respectively, at room temperature. The DSC results showed good safety characteristics.5 Kim et al. showed * To whom correspondence should be addressed. Phone: 631-3444513. E-mail: [email protected]. † State University of New York at Stony Brook. ‡ Brookhaven National Laboratory. (1) Mizushima, K.; Jones, P. C.; Wiseman, P. C.; Goodenough, J. B. Mater. Res. Bull. 1980, 15, 783. (2) Nagaura, T.; Tozawa, K. Prog. Batteries Sol. Cells 1991, 9, 209. (3) Ozawa, K. Solid State Ionics 1994, 69, 212. (4) Ohzuku, T.; Makimura, Y. Chem. Lett. 2001, 744. (5) Lu, Z.; MacNeil, D. D.; Dahn, J. R. Electrochem. Solid-State Lett. 2001, 4, A191.

in their xLiNi0.5Mn0.5O2‚(1 - x)Li2TiO3 systems that the electrochemically inactive Li2TiO3 component contributes to the stabilization of LiMn0.5Ni0.5O2 electrodes.6,7 However, the evolution of the local electronic and atomic structure of these materials during charge and discharge has not yet been reported. X-ray, neutron diffraction, chemical analysis, and electrochemical characterization reveal most of the structure-property relationships, but these studies are limited to the investigation of bulk properties and average atomic structures. X-ray absorption spectroscopy (XAS) can be used to examine the electronic and local structure of the cathode materials for Li rechargeable batteries.8-11 In an earlier communication, we reported a joint NMR and in situ XAS study of the LiMn0.5Ni0.5O2 electrode system.12 Here we present a more detailed analysis of the in situ XAS results to investigate the changes in the electronic and local structure around the Mn/Ni and Li atoms during the first charge and discharge processes. Differences in the coordination environments around Ni and Mn are identified, and the major charge-compensation mecha(6) Kim, J.-S.; Johnson, C. S.; Thackeray, M. M. Electrochem. Commun. 2002, 4, 205. (7) Johnson, C. S.; Kim, J.-S.; Kropf, A. J.; Kahaian, A.; Vaughey, J. T.; Thackeray, M. M. Electrochem. Commun. 2002, 4, 492. (8) Delmas, C.; Peres, J. P.; Rougier, A.; Demourgues, A.; Weill, F.; Chadwick, A.; Broussely, M.; Perton, F.; Biensan, Ph.; Willmann, P. J. Power Sources 1997, 68, 120. (9) Yoon, W.-S.; Lee, K.-K.; Kim, K.-B. J. Electrochem. Soc. 2000, 147, 2023. (10) Nakai, I.; Nakagome, T. Electrochem. Solid-State Lett. 1998, 1, 259. (11) Balasubramanian, M.; Sun, X.; Yang, X. Q.; McBreen, J. J. Electrochem. Soc. 2000, 147, 2903. (12) Yoon, W.-S.; Paik, Y.; Yang, X.-Q.; Balasubramanian, M.; McBreen, J.; Grey, C. P. Electrochem. Solid-State Lett. 2002, 5, A263.

10.1021/cm030220m CCC: $25.00 © 2003 American Chemical Society Published on Web 07/03/2003

3162

Chem. Mater., Vol. 15, No. 16, 2003

Yoon et al.

nism in the Li[Mn0.5Ni0.5]O2 electrode system during electrochemical cycling is established. Experimental Section LiMn0.5Ni0.5O2 powders were synthesized by reacting stoichiometric quantities of a coprecipitated double hydroxide of manganese and nickel with lithium hydroxide at 900 °C for 24 h in O2. Cathode specimens were prepared by mixing the LiMn0.5Ni0.5O2 powders with 10 wt % acetylene black and 10 wt % poly-vinylidene fluoride (PVDF) in n-methyl pyrrolidone (NMP) solution. LiPF6 (1 M) in a 1:1 ethyl carbonate/dimethyl carbonate (EC/DMC) solution was used as the electrolyte. The cell was assembled in an argon-filled glovebox. The detailed design of the spectroelectrochemical cell used in the in situ XAS measurement has been described elsewhere.13 XAS measurements were performed in the transmission mode at beamline 18B of the National Synchrotron Light Source (NSLS) using a Si(111) channel cut monochromator. The monochromator was detuned to 35-45% of its original intensity to eliminate the high-order harmonics. Energy calibration was carried out using the first inflection point of the spectrum of Mn and Ni metal foil as a reference (i.e., Mn K-edge ) 6539 eV and Ni K-edge ) 8333 eV). Reference spectra were simultaneously collected for each in situ spectrum using Mn or Ni metal foils. The in situ Mn and Ni K-edge XAS were obtained in two separate cells. The EXAFS data analysis was carried out using a standard procedure. The measured absorption spectrum below the preedge region was fitted to a straight line. The background contribution above the postedge region, µo(E), was fitted to a fourth order polynomial (cubic spline). The fitted polynomials were extrapolated through the total energy region and subtracted from the total absorption spectra. The backgroundsubtracted absorption spectra were normalized for the above energy region, χ(E) ) {µ(E) - o(E)}/o(E). The normalized χ(E) spectra were converted to χ(k) in k space, where k ) [8π2m(E-Eo)/h2]1/2. The χ(k) spectra were k3-weighted to magnify the small signal in the higher k space. The normalized k3-weighted EXAFS spectra, k3χ(k), were Fourier transformed (FT) in a k space of 2.5 to 11.0 Å-1 for the Mn edge and 3.0 to 14.0 Å-1 for the Ni edge to show the contribution of each bond pair. EXAFS structural parameters were obtained by nonlinear leastsquares analysis of the data using phase and amplitude functions generated from the FEFF6 code.14,15 The leastsquares fits were carried out in r space using FEFFIT. The amplitude reduction factor S02 was scaled to a fixed value of 0.69 for Mn edge and 0.84 for Ni edge, respectively, after preliminary refinements.

Figure 1. Voltage profiles of the cell during the first charge. Representative scans of the XAS measurements are indicated.

Results and Discussion To elucidate the charge compensation mechanism in this system, Mn and Ni K-edges XAS experiments are carried out during charge up to the fully charged state (∼280 mAh/g) at a slow current rate (C/50 rate), which is close to the quasi-equilibrium state. The voltage profile of the cell during the first charge for experiments performed at the Mn and Ni K-edges is shown in Figure 1. The specific capacity was calculated from the elapsed time, current, and mass of the active material in the cathode, by assuming that all the current passed was due to lithium intercalation/deintercalation reaction. The composition change during each scan was ∼3.5 mAh. Figure 2 shows some selected normalized Mn and Ni K-edge XANES spectra for the LiMn0.5Ni0.5O2 elec(13) Balasubramanian, M.; Sun, X.; Yang, X. Q.; McBreen, J. J. Power Sources 2001, 92, 1. (14) Rehr, J. J.; Zabinsky, S. I.; Albers, R. C. Phys. Rev. Lett. 1992, 69, 3397. (15) O’Day, P. A.; Rehr, J. J.; Zabinsky, S. I.; Brown, G. E., Jr. J. Am. Chem. Soc. 1994, 116, 2938.

Figure 2. Normalized (a) Mn K-edge and (b) Ni K-edge XANES spectra during charge. The spectrum of LiNiO2 is included as a standard in (b).

trode during charge. The metal K-edge XANES spectra originate from transitions of the 1s core electron of the metal to excited vacant bound states with appropriate symmetry. Peaks in the second derivative plots shown in Figure 3 indicate that the Mn and Ni K-edge XANES possess considerable structure. Although a rigorous assignment of XANES transitions is difficult, especially

In Situ XAS of LiNi0.5Mn0.5O2 Cathode Material

Figure 3. Second derivative spectra for the normalized (a) Mn K-edge and (b) Ni K-edge XANES spectra during charge.

in the absence of polarized spectra, each feature of XANES spectra can be interpreted in terms of the local geometry and local electronic structure of the Ni/Mn ions.16-19 The weak preedge absorption, peak A, is the formally electric dipole-forbidden transition of a 1s electron to an unoccupied 3d orbital of a high spin (t2g3eg0) Mn4+ ion or a low spin (t2g6eg2) Ni2+ ion, which is partially allowed because of the pure electric quad(16) Kim, M. G.; Yo, C. H. J. Phys. Chem. B 1999, 103, 6457. (17) Pickering, I. J.; George, G. N.; Lewandowski, J. T.; Jacobson, A. J. J. Am. Chem. Soc. 1993, 115, 4137. (18) Pickering, I. J.; George, G. N. Inorg. Chem. 1995, 34, 3142. (19) Martins Alves, M. C.; Dodelet, J. P.; Guay, D.; Ladouceur, M.; Tourillon, G. J. Phys. Chem. 1992, 96, 10898.

Chem. Mater., Vol. 15, No. 16, 2003 3163

rupole coupling and/or the 3d-4p orbital mixing arising from the noncentrosymmetric environment of the slightly distorted octahedral 3a site in the rhombohedral R3m space group. The first strong absorption, peak B, is assigned to the 1s f 4p transition involving a shakedown process caused by ligand-to-metal charge transfer (LMCT). The strongest absorption, peak C, is the purely dipole-allowed 1s f 4p transition without the shakedown process. Thus, the B and C peaks in the pristine materials correspond to two final states of a 1s1c3dn+1L4p1 (n ) 3 for Mn4+ and n ) 8 for Ni2+) with a shakedown process by a LMCT and a 1s1c3dn4p1 without the shakedown process, where c and L are a 1s core hole and an oxygen 2p ligand hole, respectively. The peak B occurs as a shoulder to lower energies of peak C, since the screening of the core hole due to the charge-transfer lowers the energy of the combined transitions relative to the unscreened excitation.20 As the Li ion is deintercalated (Figure 2), the Mn XANES spectrum exhibits some changes in the shape of the edge due to changes in the Mn local environment, but does not show an entire shift to higher energy values. The energy position and the shape of these spectra are very similar to those of the Li1.2Cr0.4Mn0.4O2 electrode material, in which manganese remains as Mn4+ throughout charge and discharge.21,22 This provides clear evidence that most of Mn ions in pristine LiMn0.5Ni0.5O2 are already in the Mn4+ oxidation state and are not oxidized as a result of the Li deintercalation. In contrast, the Ni edge shifts to higher energy values on charging. The entire edge shift to the higher energy region suggests that the average oxidation state of nickel ions increases during charge. In the charging process of LiMn0.5Ni0.5O2 electrode, the Ni edge shifts substantially to higher energies until scan 42 (halfcharged state, ∼140 mAh/g). However, the Ni edge shift from scan 42 to scan 82 (fully charged state, ∼280 mAh/ g) is relatively small. This indicates that charge compensation in the first half of the charging process of the cathode material is achieved mainly via the oxidation of Ni2+ ions, but in the other half of the charging process (scan 42-82) a process in addition to the oxidation of Ni ions may occur. We will discuss this below. Close observation of the A and B peak features in Mn and Ni K-edge XANES can contribute to understanding the change of the local structure around Mn/Ni atoms during charge. Figure 4 shows the experimental preedge and fitted spectra for Mn and Ni K-edge XANES of the LiMn0.5Ni0.5O2 cathode. The preedge features were fitted by a Voigt line shape (sum of Lorenzian and Gaussian functions), and the peak position and the peak intensity were varied. With the same manner, the preedge peak features during charge were fitted and the results are listed in Table 1. Figure 5 shows the variations of the preedge peak position for Mn and Ni K-edge XANES of LiMn0.5Ni0.5O2 cathode during charge. The peak position of preedge in Ni K-edge XANES moves toward higher energy during charge, reflecting the increase of the (20) Tranquada, J. M.; Heald, S. M.; Moodenbaugh, A. R. Phys. Rev. B 1987, 36, 5263. (21) Balasubramanian, M.; McBreen, J.; Davidson, I. J.; Whitfield, P. S.; Kargina, I. J. Electrochem. Soc. 2002, 149, A176. (22) Ammundsen, B.; Paulsen, J.; Davidson, I.; Liu, R. S.; Shen, C. H.; Chen, J. M.; Jang, L. Y.; Lee, J. F. J. Electrochem. Soc. 2002, 149, A431.

3164

Chem. Mater., Vol. 15, No. 16, 2003

Figure 4. Experimental preedge and fitted (a) Mn K-edge and (b) Ni K-edge XANES spectra of the LiMn0.5Ni0.5O2 cathode.

oxidation state of Ni ion. Even though at present it is not possible to conduct a more quantitative analysis of transition metal K-edge XANES features to determine the exact oxidation state, it is clear that a higher absorption energy is required to excite the 1s core electron of the higher oxidation ion, which is more strongly bounded to less screened nucleus. In addition, the peak position of preedge in Mn K-edge XANES does not show substantial change during charge, supporting the fact that manganese remains as Mn4+. It has been reported that the intensity of the preedge absorption increases as the site symmetry of the transition metal decreases from a centrosymmetric to a noncentrosymmetric environment.23,24 As shown in Figure 6, the preedge peak intensity increases in both the Mn and Ni K-edge XANES spectra as the Li ions are deintercalated, which indicates that the local structure around the Mn/Ni atom becomes more distorted with Li-ion deintercalation. In the case of both the Mn and Ni K-edge XANES during charge, intensities of peak B decrease systematically with Li-ion deintercalation. The peak B feature can (23) Lytle, F. W.; Greegor, R. B.; Panson, A.; Phys. Rev. B 1988, 37, 1550.

Yoon et al.

be explained from the viewpoint of the local structure distortion. The LMCT contribution increases as the covalency of the bond between the metal and the ligand increases.25 The Li-ion deintercalation gives rise to the increase of the peak A intensities, which corresponds to gradual evolution of more distorted MnO6/NiO6 octahedral sites. The structurally distorted sites within the lattice lead to a mismatched orbital overlap between Mn/Ni 3d-O 2p orbitals although the average oxidation state of Ni ions increases in Li-ion deintercalated LiMn0.5Ni0.5O2. Therefore, the LMCT process occurs less readily in the Li-ion deintercalated environment. Figure 7 shows Fourier transform magnitudes of the Mn and Ni K-edge EXAFS spectra during charge. The first coordination shell consists of oxygen, whereas the peak feature due to the second coordination shell is dominated by the nickel and manganese cations. In our EXAFS study, the scattering by the Li atom is neglected because the backscattering amplitude of the photoelectron due to scattering by Li is very weak. The most significant change during charge is observed in the first coordination shell around the Ni atoms. The dramatic changes of the first coordination peaks during charge indicate that the charge compensation mainly occurs at the Ni sites and results in a significant decrease in the average Ni-O bond length. A detailed quantitative analysis of the EXAFS data is presented below. Quantitative analysis to obtain EXAFS structural parameters for both Mn and Ni was performed by fitting the first two peaks of the FT. Curve-fitting results for the Mn and Ni K-edge EXAFS of LiMn0.5Ni0.5O2 during the charging process are listed in Table 2. In all the curve fitting processes, the goodness of fit by Σ(kdata3 kmodel3)2/Σ(kdata3)2 has been estimated within the allowed error range. Coordination number (CN) of Mn-Ni/Mn was fitted for all states of charge with the constraint that CN remains the same during charge. In addition, to minimize the correlation between coordination number and disorder, the disorder parameters of the Ni-O at ∼1.90 Å and Ni-O at ∼2.06 Å were constrained to be equal. Also, a single inner potential shift parameter ∆E0 was found to be sufficient to represent both the oxygen and metal shell of atoms. The local structure of the Mn environment in the starting material contains six oxygen atoms at ∼1.92 Å. Unlike the case of the first coordination shell, the coordination number of the second shell at ∼2.93 Å is ∼5.2 ( 0.5, which is less than the calculated CN of 6 based on the nominal composition of LiMn0.5Ni0.5O2 material. This is consistent with the previous X-ray refinement results for a similar material, where some disordering between Li+ and Ni2+ was found, which was ascribed to the similar radius of the Ni2+ (0.69 Å) and Li+ (0.76 Å) cation.26,27 Furthermore, our earlier 6Li MAS NMR results showed the presence of Li in the Ni2+/Mn4+ layers, in addition to the expected sites for Li in the lithium layers.12 This must be accompanied by either Ni2+ migration into the lithium (24) Stoll, S. L.; Bornick, R. M.; Stacy, A. M.; VerNooy, P. D. Inorg. Chem. 1997, 36, 1838. (25) Iwasawa, Y. X-ray Absorption Fine Structure for Catalysts and Surfaces; World Scientific: River Edge, NJ, 1996; p72. (26) Lu, Z.; Beaulieu, L. Y.; Donaberger, R. A.; Thomas, C. L.; Dahn, J. R. J. Electrochem. Soc. 2002, 149, A778. (27) Lu Z.; Dahn, J. R. J. Electrochem. Soc. 2002, 149, A815.

In Situ XAS of LiNi0.5Mn0.5O2 Cathode Material

Chem. Mater., Vol. 15, No. 16, 2003 3165

Table 1. Preedge Peak Position and Intensity for the Mn and Ni K-edge XANES of LiMn0.5Ni0.5O2 Cathode During Charge peak A in Ni edge

peak A1 in Mn edge

peak A2 in Mn edge

scan number

position

intensity

scan number

position

intensity

position

intensity

pristine 12 22 32 42 52 72 82

8333.56 8333.83 8333.97 8334.04 8334.28 8334.43 8334.67 8334.76

0.0192 0.0234 0.0251 0.0301 0.0318 0.0343 0.0351 0.0352

pristine 11 21 31 39 49 71 83

6540.40 6540.47 6540.49 6540.42 6540.41 6540.44 6540.55 6540.47

0.0968 0.1079 0.1129 0.1186 0.1326 0.1377 0.1417 0.1463

6542.34 6542.41 6542.39 6542.31 6542.26 6542.29 6542.37 6542.26

0.0580 0.0618 0.0714 0.0787 0.0854 0.0889 0.0922 0.0974

Figure 5. Variations of preedge peak position for the Mn and Ni K-edge XANES of LiMn0.5Ni0.5O2 cathode during charge.

Figure 7. k3-Weighted Fourier transform magnitudes of the (a) Mn and (b) Ni K-edge EXAFS spectra during charge.

Figure 6. Variations of preedge peak intensity for the Mn and Ni K-edge XANES of LiMn0.5Ni0.5O2 cathode during charge.

layers, as is seen for LiNiO2, or the presence of lithium vacancies in the lithium layers (and the oxidation of a subset of the Ni2+ ions), to maintain charge balance. The lithium substitution into the transition metal layers was ascribed to the strong Coulombic driving force for charge ordering of 1+ and 4+ cations on a trigonal lattice, when they are present in a 1:2 ratio, as occurs in, for example, Li2MnO3. In the case of the Ni K-edge EXAFS spectra, features in the first peak of the FT can be ascribed to Ni2+-O,

Ni3+-O, and Ni4+-O bonds. In general, Ni-O bond lengths for octahedrally coordinated Ni ions in a variety of different compounds with the same oxidation state are very similar.28-31 On the basis of earlier studies,28-31 a Ni2+-O bond length is ∼2.07 Å and a Ni4+-O bond length is ∼1.88 Å. A Ni3+-O octahedron has two long bonds (∼2.06 Å) and four short bonds (∼1.90 Å) for a (28) Pandya, K. I.; O’Grady, W. E.; Corrigan, D. A.; McBreen, J.; Hoffman, R. W. J. Phys. Chem. 1990, 94, 21. (29) Capehart, T. W.; Corrigan, D. A.; Conell, R. S.; Pandya, K. I.; Hoffman, R. W. Appl. Phys. Lett. 1991, 58, 865. (30) Mansour, A. N.; Melendres, C. A. J. Phys. Chem. A 1998, 102, 65. (31) Mansour, A. N.; Yang, X. Q.; Sun, X.; McBreen, J.; Croguennec, L.; Delmas, C. J. Electrochem. Soc. 2000, 147, 2104.

3166

Chem. Mater., Vol. 15, No. 16, 2003

Yoon et al.

Table 2. Curve-Fitting Results for the Fourier-Filtered k3χ(k) Mn K-edge and Ni K-edge EXAFS of LiMn0.5Ni0.5O2 Cathode During Charge scan no. 1 11 21 31 39 49 71 83

2 12 22 32 42 52 72 82

shell

CN

R (Å)

Mn-O Mn-Ni/Mn Mn-O Mn-Ni/Mn Mn-O Mn-Ni/Mn Mn-O Mn-Ni/Mn Mn-O Mn-Ni/Mn Mn-O Mn-Ni/Mn Mn-O Mn-Ni/Mn Mn-O Mn-Ni/Mn

6 5.2 (5) 6 5.2 (5) 6 5.2 (5) 6 5.2 (5) 6 5.2 (5) 6 5.2 (5) 6 5.2 (5) 6 5.2 (5)

Mn K-edge 1.921 (8) 2.925 (7) 1.919 (7) 2.919 (6) 1.915 (6) 2.910 (5) 1.910 (6) 2.896 (7) 1.905 (10) 2.887 (10) 1.907 (7) 2.887 (7) 1.907 (7) 2.879 (6) 1.900 (5) 2.875 (5)

Ni-O Ni-O Ni-Ni/Mn Ni-O Ni-O Ni-Ni/Mn Ni-O Ni-O Ni-Ni/Mn Ni-O Ni-O Ni-Ni/Mn Ni-O Ni-O Ni-Ni/Mn Ni-O Ni-O Ni-Ni/Mn Ni-O Ni-O Ni-Ni/Mn Ni-O Ni-O Ni-Ni/Mn

5.5 (4) 0.5 (4) 6 4.8 (3) 1.2 (3) 6 4.2 (3) 1.8 (3) 6 2.6 (2) 3.4 (2) 6 1.6 (4) 4.4 (4) 6 1.4 (2) 4.6 (2) 6 0.9 (5) 5.1 (5) 6 0.5 (3) 5.5 (3) 6

Ni K-edge 2.056 (4) 1.899 (3) 2.922 (5) 2.056 (4) 1.899 (3) 2.918 (4) 2.056 (4) 1.899 (3) 2.915 (5) 2.056 (4) 1.899 (3) 2.898 (5) 2.056 (4) 1.899 (3) 2.890 (6) 2.056 (4) 1.899 (3) 2.885 (4) 2.056 (4) 1.899 (3) 2.875 (6) 2.056 (4) 1.899 (3) 2.871 (3)

Ni3+ ion exhibiting a static Jahn-Teller distortion. Thus, in the final fits the Ni2+-O and long bond length of Ni3+-O were constrained to be the same, as were the Ni4+-O bond length and the short bond length of Ni3+-O because the resolution in EXAFS experiment is around 0.02 Å.15 The local structure parameters around the Ni atom, deduced for the pristine LiMn0.5Ni0.5O2 compound, consist of 0.5 ( 0.4 O neighbors at 1.899 Å and 5.5 ( 0.4 O neighbors at 2.056 Å, which suggests that the average oxidation state of Ni ion in the pristine compound is slightly higher than 2. Upon electrochemical deintercalation of Li, the number of short Ni-O bonds increased linearly from 0.5 ( 0.4 for LiMn0.5Ni0.5O2 to 4.4 ( 0.4 for Li0.5Mn0.5Ni0.5O2, while the number of long bonds decreased linearly from 5.5 ( 0.4 for LiMn0.5Ni0.5O2 to 1.6 ( 0.4 for Li0.5Mn0.5Ni0.5O2. In the last half of the charge process (from scan 42 to scan 82), however, the increase in the number of short Ni-O bonds was relatively small (from 4.4 ( 0.4 for Li0.5Mn0.5Ni0.5O2 to 5.5 ( 0.3 for Mn0.5Ni0.5O2), which is consistent with the above Ni K-edge XANES results during charge. To obtain a better understanding of the charge compensation mechanism, we considered two models with different oxidation paths for the Ni ions. It was assumed in both models that simultaneous oxidation of

σ2 (× 10-4 Å2)

∆E (eV)

R factor

37 (7) 38 (6) 37 (6) 39 (5) 36 (5) 41 (4) 40 (6) 48 (5) 43 (9) 52 (8) 46 (7) 50 (5) 49 (6) 47 (5) 45 (5) 42 (3)

12.7 (1.1)

0.00922

12.6 (0.9)

0.00903

12.5 (0.8)

0.01021

12.1 (0.9)

0.00989

11.9 (1.5)

0.01132

12.3 (1.0)

0.01768

12.0 (0.9)

0.01641

11.5 (0.7)

0.01083

43 (12) 43 (12) 50 (3) 40 (9) 40 (9) 52 (2) 51 (14) 51 (14) 55 (3) 47 (10) 47 (10) 62 (3) 48 (12) 48 (12) 62 (4) 40 (7) 40 (7) 60 (2) 40 (11) 40 (11) 51 (4) 49 (7) 49 (7) 47 (2)

7.7 (8)

0.00738

8.1 (7)

0.00732

8.9 (8)

0.00863

8.5 (7)

0.01605

9.1 (1.0)

0.01193

9.4 (6)

0.01325

9.1 (1.1)

0.02101

8.9 (6)

0.01030

the Ni2+ and Ni3+ ions does not occur, and that either (i) all the Ni2+ ions are first oxidized to Ni3+ and then oxidized further to Ni4+, or (ii) all the Ni2+ ions are directly oxidized to Ni4+ without any Ni3+ as an intermediate oxidation state. Figure 8 shows a comparison between the experimental data and the two models during charge. To avoid the complexity of the oxidation process in the second half of the charge, only data from the first half of the charge were used in this figure. Considering the fact that the average oxidation state of the Ni ion in the pristine compound is slightly higher than 2, the predictions based on model (i) are closer to the experimental results. This suggests that the oxidation process upon Li deintercalation takes place in two steps: Ni2+ to Ni3+ first, and then Ni3+ to Ni4+. These experimental EXAFS results are in good agreement with theoretical work by Ceder et al., where a large distribution of Ni-O bond lengths found in Li0.5Mn0.5Ni0.5O2 were ascribed to the presence of the JahnTeller Ni3+ ion.32 The results for this system should be contrasted with the cycling behavior of the spinel Li[Ni0.5Mn1.5]O4, where no evidence for a Ni3+ intermediate was observed.33 We note, however, that these materials contain considerable disorder in the transition (32) Reed, J.; Ceder, G. Electrochem. Solid-State Lett. 2002, 5, A145.

In Situ XAS of LiNi0.5Mn0.5O2 Cathode Material

Chem. Mater., Vol. 15, No. 16, 2003 3167

Figure 8. Experimental fitting data and models for Ni oxidation reaction during charge.

metal layers. This results in a variety of local environments for lithium in these materials, as confirmed by our earlier NMR studies.12 The role that these local structure differences play in controlling the potential of the Ni2+/Ni3+ and Ni3+/Ni4+ couples has yet to be established, and any overlap between these two couples in this material will further complicate the analysis of the EXAFS data. Furthermore, the variations in electrochemical data obtained by different groups4,5 suggests that different materials with nominal stoichiometry Li0.5Mn0.5Ni0.5O2 prepared in different laboratories may contain different arrangements or ordering of the Li+, Ni2+, and Mn4+ ions. XAS studies on a wider range of lithium nickel manganates are in progress to explore these issues in greater detail. The major charge compensation mechanism during the last half of the charge process (the higher voltage region) is still not clear. Lu et al.26,27 claimed in their systematic study of Li[NixLi(1/3-2x/3)Mn(2/3-x/3)]O2 that after simultaneous extraction of both Li and O from the material, in the high-voltage region, there is an increase in the capacity of the material. In contrast, Robertson et al.34 suggested in their study of Li2MnO3 that the primary process involves exchange of Li+ by H+ even though the nonaqueous electrochemical system is not supposed to contain any H2O. Earlier work by Ceder et al.35 showed that the oxygen 2p band can participate in the electrochemical reaction, in addition to the charge compensation mechanism involving the metal ion. An O K-edge study during electrochemical cycling using soft X-ray absorption techniques to understand this is underway in our group. To investigate the electronic structure and reversibility of the LiMn0.5Ni0.5O2 system during first charge and discharge, the cells were charged from their open circuit potentials up to 4.6 V, and then discharged at a constant current rate of C/20 calculated based on the theoretical capacity (∼280 mAh/g). The voltage profiles of the cell during the first charge and discharge for experiments (33) Zhang, Q.; Bonakdarpour, A.; Zhang, M.; Gao, Y.; Dahn, J. R. J. Electrochem. Soc. 1997, 144, 205. (34) Robertson, A. D.; Bruce, P. G. Chem. Commun. 2002, 2790. (35) Ceder, G.; Chiang, Y.-M.; Sadoway, D. R.; Aydinol, M. K.; Jang, Y.-I.; Huang, B. Nature 1998, 392, 694.

Figure 9. Voltage profiles during the first charge and discharge: (a) cell for the Mn XAS, (b) cell for the Ni XAS. Every third XAS scan is marked on the curves.

performed at the Mn and Ni K-edges are shown in Figure 9. In this case the change in composition during each scan was ∼12 mAh. Two electrochemical processes during discharge are observed, the first one at a potential of approximately 4 V and a second at a lower potential of 1 V. The electrochemical behavior of our electrode at 4 V is similar to that observed in earlier reports.4,5 The lower voltage process was first reported by Johnson et al.7 Selected normalized Mn and Ni K-edge XANES spectra of the LiMn0.5Ni0.5O2 electrode during the first charge and discharge are given in Figure 10 (a) and (b), respectively. No significant energy shift can be observed in the Mn XANES spectra up to scan 25, which corresponds to a discharged state slightly above the 1-volt plateau (approximately 1.2 V). The Mn edge started to move to lower energy as the 1-volt plateau was approached (scan 26) and continued to shift as the discharge was continued. This is clearly reflected in the difference between scan 29 and End. The entire edge shift from scan 29 to End suggests reduction of Mn4+ to Mn2+, and is consistent with the edge positions of the reference compounds of MnO (Mn2+) and MnO2 (Mn4+).36 The data are not of high enough quality, due to the amorphous nature of the MnO phase, to determine whether Mn3+ is formed as an intermediate. In contrast, the Ni edge shifts gradually to lower energies on discharging at higher voltages, reaching its original position at around scan 27 (1.3 V), where it remains essentially unchanged throughout the 1-volt plateau discharge. This indicates, first, that there is some (36) Conway, E.; Qu, D.; McBreen, J. In situ and ex situ spectroelectrochemical and X-ray absorption studies on rechargeable, chemically modified and other MnO2 materials. In Synchrotron Techniques in Interfacial Electrochemistry; NATO ASI Series Vol. 432; Melendres C. A., Tadjeddine, A., Eds.; Kluwer Academic Publishers: Dordrecht, 1994; pp 311-334.

3168

Chem. Mater., Vol. 15, No. 16, 2003

Figure 10. Normalized (a) Mn K-edge and (b) Ni K-edge XANES spectra during the first charge and discharge.

hysteresis in the Ni2+/ Ni3+/Ni4+ redox reactions, and, second, that the capacity obtained in the 1-volt plateau is due to the reduction of Mn4+. Figure 11 shows Fourier transform magnitudes of the Mn and Ni K-edge EXAFS spectra during first charge and discharge. Once again, the most dramatic change during charge is observed in the first coordination shell around Ni atoms, confirming that the charge compensation in the voltage region between 2 and 4.6 V is achieved mainly by the oxidation of Ni2+ to Ni3+ and then Ni4+. In the 1-volt plateau discharge, peak features in the FT remain essentially unchanged. This also supports the above Ni XANES results suggesting that the charge compensation in the 1-volt plateau is not achieved in the Ni sites. There is no substantial difference in the Mn K-edge EXAFS spectra up to scan 25 indicating electrochemical redox reaction does not occur in the Mn site in the voltage region between 2 and 4.6 V. However, the peak height of the FT of the Mn K-edge EXAFS spectrum shows a large decrease throughout the 1-volt plateau discharge. The peak amplitude of the FT is mainly related to the coordination number and local disorder around the target atom. In this compound, the decrease in FT magnitudes might result from a possible local structural distortion due to the phase transition from LiMO2 to Li2MO2 (space group P3 h ml) structure7 or the

Yoon et al.

Figure 11. k3-Weighted Fourier transform magnitudes of the (a) Mn and (b) Ni K-edge EXAFS spectra during charge.

possible displacement of metal ions into less symmetric environments. The Mn and Ni K-edge XAS results during first charge and discharge at potentials near 4 V indicate that charge compensation in the cathode material is achieved mainly via the oxidation/reduction of Ni2+ and Ni4+ ions. In contrast, the Mn ions do not appear to actively participate in the charge compensation process, remaining mostly unchanged in the Mn4+ state. This is consistent with earlier suggestions that the stability of these layered materials, and the suppression of the layered-to-spinel transformation, is at least in part due to the absence of manganese ions with oxidation states of 3+, during cycling at potentials of approximately 4 V.32 In the 1-volt plateau region, however, the charge compensation for the Li-ion intercalation process is achieved via reduction of Mn4+ to Mn2+. Conclusion In situ Mn and Ni K-edge XAS for the layered LiNi0.5Mn0.5O2 cathode material have been carried out during the first charging and discharging processes. From the observation of the Mn and Ni K-edge XANES results, it is concluded that the charge compensation during charge and discharge is achieved mainly by the oxida-

In Situ XAS of LiNi0.5Mn0.5O2 Cathode Material

tion/reduction of Ni2+ and Ni4+ ions in the higher voltage region of ∼4 V, while the manganese ions remain largely unchanged in the Mn4+ oxidation state. In the lower voltage region of ∼1 V, however, the charge compensation is achieved mainly by the reduction of Mn4+ whereas the Ni ions do not appear to actively participate in the charge compensation process. These conclusions are consistent with the EXAFS results. Our quantitative EXAFS results suggest two important observations. First, despite the composition of this sample, lithium ions replace some of the transition metals in the Ni2+/ Mn4+ layers; these lithium ions primarily contain Mn ions in their second coordination shell, as found in the local structure of Li2MnO3. Second, the oxidation process during charge takes place in two steps: Ni2+ to Ni3+ first, and then Ni3+ to Ni4+.

Chem. Mater., Vol. 15, No. 16, 2003 3169

Acknowledgment. The XAS measurements were carried out at Beam Line X18B at NSLS. The work performed at SUNY Stony Brook was supported by the Assistant Secretary for Energy Efficiency and Renewable Energy, Office of FreedomCAR and Vehicle Technologies of the U.S. Department of Energy under Contract DE-AC03-76SF00098, via subcontract 6517749 with the Lawrence Berkeley National Laboratory. The work carried out at BNL was supported by the Assistant Secretary for Energy Efficiency and Renewable Energy, Office of FreedomCAR and Vehicle Technologies of the U.S. Department of Energy under Contract DE-AC0298CH10886. CM030220M