A1134 Journal of The Electrochemical Society, 151 共8兲 A1134-A1140 共2004兲

0013-4651/2004/151共8兲/A1134/7/$7.00 © The Electrochemical Society, Inc.

Combining Ab Initio Computation with Experiments for

Designing New Electrode Materials for Advanced Lithium
Batteries: LiNi1Õ3Fe1Õ6Co1Õ6Mn1Õ3O2
Ying S. Meng,a Yei Wei Wu,b Bing Joe Hwang,b,d,* Yi Li,a,c
and Gerbrand Cedera,d,*
a
Singapore-MIT Alliance, Advanced Materials for Micro- and Nano-Systems Programme, Singapore 117576
b
Department of Chemical Engineering, National Taiwan University of Science and Technology, Taipei,
Taiwan 106
c
Department of Materials Science, National University of Singapore, Lower Kent Ridge, Singapore 119260
d
Department of Materials Science and Engineering, Massachusetts Institute of Technology, Cambridge,
Massachusetts 02139, USA

An initial search with density functional theory to sort through potential cathode materials based on their Li intercalation potentials
and electronic structures was carried on LiNi1/3TM1 1/3TM2 1/3O2 systems, where TM1 is a 3⫹ transition metal (Co3⫹, Al3⫹, Fe3⫹
etc.兲 and TM2 is a 4⫹ transition metal (Ti4⫹, Zr4⫹, Mn4⫹ etc.兲. Fe substitution is found to be advantageous because among the
compounds investigated it shows the lowest voltage at the last stage of the charge. LiNi1/3Fe1/6Co1/6Mn1/3O2 was synthesized by
a sol-gel method and used to confirm that Fe substitution leads to a lower potential at the end of charge. Both X-ray photoelectron
spectroscopy and first principles electronic structure computations indicate that Ni and Fe are simultaneously oxidized in this
material. Computations further indicate that Co will only be oxidized at the very end of charge. The LiNi1/3Fe1/6Co1/6Mn1/3O2
compound synthesized at 750°C shows reversible capacity of 150 mAh/g with reasonably good capacity retention.
© 2004 The Electrochemical Society. 关DOI: 10.1149/1.1765032兴 All rights reserved.

Manuscript submitted November 3, 2003; revised manuscript received January 26, 2004. Available electronically June 17, 2004.

Lithium nickel manganese oxide and their derivatives are con- material LiNi1/3Fe1/6Co1/6Mn1/3O2 is designed, synthesized and char-
sidered promising candidates for future lithium ion batteries.1-3 In acterized by an approach integrating ab initio computation and ex-
these compounds, nickel can exchange two electrons so that manga- periments.
nese can remain at a 4⫹ oxidation state during the charge-discharge
cycle without any loss of theoretical capacity 共about 280 mAh/g兲 Methodology
and without the instabilities associated with more reduced states of
Mn.4 Recently, LiNi1/3Co1/3Mn1/3O2 has been shown to have very Computational.—It has been amply demonstrated that reason-
good electrochemical properties. Ohzuku et al. reported that able lithium intercalation potentials and geometrical information can
LiNi1/3Co1/3Mn1/3O2 has 200 mAh/g in the voltage range of 2.5-4.6 be obtained with first principles methods.8-10 To describe the
V with negligible capacity loss up to 30 cycles.5 Hwang et al. ob- Lix Ni1/3Fe1/6Co1/6Mn1/3O2 system, supercells with six formula units
tained reversible capacity of 188 mAh/g in the potential window of were used. As it is typical in solid state computations periodic
3 to 4.5 V.6 Shaju et al. inferred from cyclic voltammetry 共CV兲 boundaries are used, so that one effectively models a system with
results that the redox processes at 3.8 and 4.6 V correspond to the Ni, Fe Co, and Mn long-range ordered 共Fig. 1.兲 The effect of disor-
Ni2⫹/Ni4⫹ and Co3⫹/Co4⫹ couples, respectively,7 though the capac- der, present in a real system, would likely smooth the voltage curve
ity from the latter redox pair is very small in this potential window. from what is achieved computationally.
First principles calculation confirmed that Co3⫹ is only oxidized to All energies, intercalation potentials, geometries, and electronic
Co4⫹ at rather high voltage in this material.6 While first principles structure of materials in this paper were obtained using first-
methods are not accurate enough to exactly predict voltages, the principles quantum mechanics in the generalized gradient approxi-
computational results in Ref. 6 indicate that complete oxidation of mation 共GGA兲 to DFT, as implemented in the vienna ab initio simu-
lation package 共VASP兲. Ultrasoft pseudo-potentials are applied to
Co3⫹ to Co4⫹ may require potentials near 5 V in this system. An
represent the nuclei and core electrons and all structures are fully
obvious way to increase practical capacity is therefore to partially
relaxed with respect to internal and external cell parameters. The
substitute Co or Mn by other transition metals, which either bring
wave functions are expanded in plane waves with energy below 405
down the redox potential of Co, or introduce another redox couple in
eV. Brillouin zone integration of the band structure is performed
a lower voltage range than Co3⫹/Co4⫹. with a 6 ⫻ 3 ⫻ 4 mesh. All calculations are performed with spin
LiNi1/3Co1/3Mn1/3O2 illustrates the recent trend towards multi- polarization, previously demonstrated to be crucial in manganese
component transition metal oxides, which creates a large number of oxides.11 Both ferromagnetic and anti-ferromagnetic spin polariza-
possible compositional choices. The materials development includ- tion was taken into consideration. For each Li concentration x, anti-
ing synthesis, processing, characterization, and optimization can be ferromagnetic coupling among Ni, Mn, and Fe gives a lower energy
more efficient and cost effective if we can do a certain amount of than ferromagnetic coupling.
property prediction and optimization during the design stage. We Partial states of delithiation were investigated at x ⫽ 5/6, 2/3,
therefore performed an initial search with density functional theory 1/2, 1/3, and 1/6. The number of possible arrangements for Li and
共DFT兲 to evaluate the Li 共de兲intercalation voltage of potential cath- vacant sites in the supercell are 1, 6, 15, 20, 15, 6, and 1 for x
ode materials. Through first principles computation methods, we
⫽ 1, 5/6, 2/3, 1/2, 1/3, 1/6, and 0, respectively. All possible ar-
identified partial substitution of Co by Fe in LiNi1/3Co1/3Mn1/3O2 as rangements have been calculated.
a way of lowering the lithium 共de兲intercalation potential at the end
of charge. Structural energies and calculated mixing enthalpies were Experimental.—Motivated by the first principles results,
used to guide the synthesis conditions and help interpret experimen- LiNi1/3Fe1/6Co1/6Mn1/3O2 was synthesized by a sol-gel method
tal results. In this paper, we present how a new potential cathode using citric acid as a chelating agent. A stoichiometric amount
of lithium acetate (Li共CH3 COO兲•2H2 O), nickel acetate
(Ni共CH3 COO兲2•4H2 O), cobalt nitrate (Co共NO3兲2•6H2 O),
* Electrochemical Society Active Member. iron nitrate (Fe共NO3兲3•9H2 O) and manganese acetate
 Journal of The Electrochemical Society, 151 共8兲 A1134-A1140 共2004兲 A1135

Figure 1. Ni, Fe, Co Mn ordering in the supercell of

Li共Ni1/3Fe1/6Co1/6Mn1/3)O2 .

Figure 2. Calculated voltage curves for LiNi1/3TM1 1/3TM2 1/3O2 where

(Mn共CH3 COO兲2•4H2 O) were dissolved in distilled water and well TM1 ⫽ Al3⫹, Co3⫹, or Fe3⫹ and TM2 ⫽ Mn4⫹, Ti4⫹, or Zr4⫹.
mixed with aqueous solution of citric acid. The solution was stirred
at 60-70°C for 5-6 h to obtain a clear viscous gel. The gel was dried
in vacuum oven at 120°C for 24 h. The material was precalcined in end of charge. In agreement with previous work on Al-doping,10
two stages: at 350°C for 5 h and 450°C for 4 h, and then ground substitution of Co by Al increases the potential. As it can not
before calcining at high temperature 共750-900°C兲 at a heating rate of be oxidized beyond 3⫹, Al doping also reduces the capacity at
about 2°C/min. The powders were slowly cooled to room tempera- the end of charge. Of all the compositional modifications inves-
ture in oxygen. tigated, Fe substitution of Co seems to be advantageous because
Powder X-ray diffraction 共XRD兲 data were collected on a Rigaku it lowers the voltage profile at the last stage of the charge, com-
diffractometer with Cu K␣1 radiation (␭ ⫽ 1.5406 Å兲, operating pared to LiNi1/3Co1/3Mn1/3O2 . Hence, the specific capacity of
at 300 kV and 60 mA. To minimize the preferred orientation LiNi1/3Fe1/3Mn1/3O2 could be higher in the potential window of
effect, typical in layered lithium intercalation compounds, Vaseline 3.0-4.5 V.
is mixed with sample powders to randomize the orientation It is well known that due to the similar ion size of Fe3⫹ and Li⫹,
of the particles. Grain morphology and particle size of the Fe can partially occupy the Li-layer.12-14 In addition, unlike
LiNi1/3Fe1/6Co1/6Mn1/3O2 compounds were examined by scanning LiCoO2 and LiNiO2 , the LiFeO2 ground-state structure is not lay-
electron microscopy using a JEOL FEG-6320. ered, but a structure with symmetry I4 1 /amd. 15 According to our
Information of nickel, cobalt, iron and manganese oxidation preliminary mixing enthalpy calculations using the equation below,
states in pristine and electrochemically-charged samples were ob- the relative formation energy of LiNi1/3Fe1/3Mn1/3O2 with respect to
tained from X-ray photoelectron spectroscopy 共XPS兲. Binding ener-
LiNi1/2Mn1/2O2 and LiFeO2 is approximately zero, indicating that
gies were charge-corrected using the C1s peak 共285 eV兲. only a weak entropic driving force for mixing might exist in a com-
LiNi1/3Fe1/6Co1/6Mn1/3O2 electrodes were fabricated by mixing pound where Co is fully substituted by Fe
85:1.5:3.5:10 共w/w兲 ratio of active material, SS carbon black, KS-6
carbon and polyvinylidene fluoride 共PVDF兲, respectively, using 具 ⌬Emix ⫽ ELiNi1/3FexCo1/3⫺xMn1/3O2
N-Methyl-pyrrolidone 共NMP兲 as the solvent. The resulting slurry
was cast onto an aluminum current collector, dried under vacuum ⫺ 关 2/3ELiNi1/2Mn1/2O2 ⫹ zELiFeO2 ⫹ 共 1/3 ⫺ z兲ELiCoO2 兴 典
oven for 4 h and put into an argon filled glove box for conditioning
overnight. The electrode foils were roller-pressed to a uniform thick- 关1兴
ness of 100 ␮m and then cut into disks of 10 mm diam.
Electrochemical measurements were made using coin-type cells Therefore, we chose to substitute Co only partially by Fe and
comprising Li metal counter electrode with a 1 M solution of LiPF6 targeted the nominal composition LiNi1/3Fe1/6Co1/6Mn1/3O2 to obtain
in EC/DMC 共1:1 v/v, Merck LP30兲 as the electrolyte. The cells were pure single-phase layered material.
assembled in the argon filled glove box where both moisture and Synthesis and characterization of as-prepared material.—The
oxygen levels are less than 1 ppm. The cells were charged and structure of the LiNi1/3Fe1/6Co1/6Mn1/3O2 powders synthesized by
discharged using a Maccor battery tester at a C/10 rate 共based on sintering at 750, 800, and 850°C for 16 h were characterized using
theoretical capacity of 281 mAh/g兲 over a potential range between XRD 共Cu K␣ radiation兲, as shown in Fig. 3. XRD spectra show that
3.0 to 4.5 V. all as-prepared samples have the ␣-NaFeO2 type layered structure
Results and Discussion with space group R3-m. The two small peaks between 20 and 25° in
2␪ are from Vaseline added to the sample. The region around the
First principles study of Co/Mn substitution.—An initial search 共104兲 reflection in the XRD spectra is shown in Fig. 3b. For powders
on LiNi1/3TM1 1/3TM2 1/3O2 where TM1 is a 3⫹ transition metal synthesized at 750°C the 共104兲 peak is broadened, indicating a small
(Co3⫹, Al3⫹, Fe3⫹ etc兲 and TM2 is a 4⫹ transition metal (Ti4⫹, grain size, which was further confirmed by scanning electron mi-
Zr4⫹, Mn4⫹ etc.兲 was carried out. Figure 2 shows the potential of croscopy 共SEM兲 共Fig. 4a兲. For powders synthesized at 850°C an
some of the substituted compounds as average voltages over inter- extra peak right beside the 共104兲 peak 共see Fig. 3b兲 starts to evolve.
vals of 1/3 Li composition. The stepwise nature of the curves is We suspect that this peak is related to the formation of a cation-
therefore artificial and due to the averaging of the potential over the disordered rock-salt structure. The clear splitting of the reflections
specific composition interval. assigned to the Miller indices 共006, 102兲 and 共108, 110兲 in the XRD
Substitution of Mn by either Ti or Zr clearly increases the po- spectrum of the 750 and 800°C sample indicates a well-layered
tential and is therefore not suitable to reduce the potential at the structure.
A1136 Journal of The Electrochemical Society, 151 共8兲 A1134-A1140 共2004兲

Figure 3. 共a兲 XRD spectra of LiNi1/3Fe1/6Co1/6Mn1/3O2 synthesized at 750,

800, and 850°C 共b兲 Details of 共104兲 peak in the XRD spectra of
LiNi1/3Fe1/6Co1/6Mn1/3O2 synthesized at 750, 800, and 850°C.

The lattice parameters a and c, determined from the sample

synthesized at 750°C, are 2.8854 and 14.3221 Å, respectively. The
calculated lattice parameters are 2.9138 and 14.3690 Å, which
are slightly larger than those measured, as is often the case for
computations in the generalized gradient approximation. In
LiNi1/3Co1/3Mn1/3O2 synthesized at 900°C by the same sol-gel
method, a and c are 2.864 and 14.247 Å, respectively.6 The larger a
and c lattice parameters of LiNi1/3Fe1/6Co1/6Mn1/3O2 can be ex-
plained by the larger ion size of FeIII as compared to CoIII.
A comparison of SEM micrographs obtained for
LiNi1/3Fe1/6Co1/6Mn1/3O2 synthesized at 750, 800, and 850°C
is shown in Fig. 4. All samples have a uniform grain size and
faceted grain morphology. The grains grow from 40-50 nm at
750°C to 300-500 nm at 850°C.
Intercalation potential.—Average voltage profiles for
Lix Ni1/3Fe1/6Co1/6Mn1/3O2 (0 ⭐ x ⭐ 1) were computed from the
lowest energy lithium-vacancy arrangements in the six-formula
supercell as function of lithium compositions. The calculated
potentials are typically lower than experimental values, as is
usuually the case in standard first principles energy methods.4,16
The calculated intercalation voltage of Lix Ni1/3Fe1/6Co1/6Mn1/3O2 is
compared to that of Lix Ni1/3Co1/3Mn1/3O2 in Fig. 5. In the range
1/3 ⭐ x ⭐ 1, a calculated average voltage of 3.0-3.1 V is obtained
for Lix Ni1/3Fe1/6Co1/6Mn1/3O2 . The potential increases significantly
to 3.8-3.9 V in the range 0 ⭐ x ⭐ 1/3. Compared to
Lix Ni1/3Co1/3Mn1/3O2 , 6 共shown by the solid line in Fig. 5兲 the sub-
stitution of Fe with Co increases the average voltage slightly for
1/2 ⭐ x ⭐ 1. Most importantly, in the Fe substituted compound, the
calculated voltage at the end of charge (0 ⭐ x ⭐ 1/3) is much
lower than that of Lix Ni1/3Co1/3Mn1/3O2 . These results indicate that
Fe substitution of Co in LiNi1/3Fez Co1/3 ⫺ z Mn1/3O2 may flatten the
voltage curve and increase the experimentally attainable capacity by
lowering the potential near the end of charge.
Electrochemical characterization.—Electrodes of
LiNi1/3Fe1/6Co1/6Mn1/3O2 synthesized at 750, 800, and 850°C were
cycled at a rate of C/10 共based on 281 mAh/g total capacity兲 be-
tween 3.0 and 4.5 V. The first charge and discharge curves for each Figure 4. SEM images of as-prepared LiNi1/3Fe1/6Co1/6Mn1/3O2 materials
sample are shown in Fig. 6. Qualitatively, the potential curves are synthesized at 共a兲 750, 共b兲 800, and 共c兲 850°C.
very similar, exhibiting a relatively flat potential on charging in the
range of 3.7 to 3.9 V, and then a relatively steeply sloping curve on
discharge. The compound synthesized at 750°C shows smaller po- large first-cycle irreversible capacity is currently being investigated.
larization which could be related to its smaller grain size. There is a A preliminary study shows that the first cycle reversible capacity
significant amount of irreversible capacity after the first charge for can be increased by 20% with surface treatment of the synthesized
all three samples. As the synthesis temperature increases from 750, powders.
800, to 850°C, the first charge capacity decreases from 220 to 200- The delithiation potential of the material synthesized at 750°C is
187 mAh/g, the first discharge capacity changes from 150 to 139- plotted together with the calculated potential curve. The potential
134 mAh/g. The capacity retention up to 30 cycles is reasonably difference between the calculated and experimental data is sug-
good for all samples, as demonstrated in Fig. 7. The reason for the gested to be 0.7-0.8 V.16 The correction of 0.9 V in this case was
 Journal of The Electrochemical Society, 151 共8兲 A1134-A1140 共2004兲 A1137

Figure 5. Comparison of calculated average voltage curves of

Lix Ni1/3Co1/3Mn1/3O2 and Lix Ni1/3Fe1/6Co1/6Mn1/3O2 .
Figure 7. Charge and discharge capacity vs. cycle number curves of the
LiNi1/3Fe1/6Co1/6Mn1/3O2 materials synthesized at 750, 800, and 850°C for
16 h.
added to the calculated potential to display a result that can be
compared directly with the experimental values, as shown in Fig. 8.
The experimental charge-discharge curve matches the calculated one cupied and Co-eg states are empty, indicative of Co3⫹. The Fermi
well in the range of 1/3 ⭐ x ⭐ 1. The practical specific capacity of level, EF is located between the top of occupied Co-t2g bands and
this material will likely be more than 250 mAh/g if the cell is unoccupied Fe-t2g states. There is an energy gap of about 0.3 eV
charged to approximately 4.8 V, according to the computational pre- between the unoccupied and occupied states. Although it is well
diction. know that the calculated energy gaps in GGA are typically smaller
Electronic change during charge-discharge.—To understand the than the experimental values, the comparison with the calculated
electronic changes in LiNi1/3Fe1/6Co1/6Mn1/3O2 when lithium is re- energy gap for LiNi1/3Co1/3Mn1/3O2 共0.7 V兲6 indicates that the elec-
moved, the spin polarized density of states 共DOS兲 at different tronic conductivity of LiNi1/3Fe1/6Co1/6Mn1/3O2 may be as good, if
lithium concentrations is shown in Fig. 9. Because the transition not better, as LiNi1/3Co1/3Mn1/3O2 .
metal ions occupy the octahedral sites in the sublattices of oxygen As we applied a supercell with six-formula units in this study,
ions, 3d bands of transition metal ions split into t2g and eg bands. both Ni2⫹/Ni3⫹ and Ni2⫹/Ni4⫹ are possible redox reactions. Figure
The calculated DOS, projected onto the orbitals of each transition 9a and b show that for a partially delithiated state (2/3 ⭐ x ⭐ 1),
metal are shown in Fig. 9. only the Ni2⫹/Ni3⫹ redox reaction is observed, which is con-
For all lithium compositions, the Mn-t2g and Mn-eg bands are, sistent with the previous studies on LiNi1/2Mn1/2O2 and
respectively, half filled and empty in Lix Ni1/3Fe1/6Co1/6Mn1/3O2 (0 LiNi1/3Co1/3Mn1/3O2 . 6,17,18 At x ⫽ 2/3 共Fig. 9b兲, there is an overlap
⭐ x ⭐ 1), which is consistent with a Mn4⫹ valence state. For fully between filled Ni-eg and Fe-eg states, indicating very similar redox
lithiated LiNi1/3Fe1/6Co1/6Mn1/3O2 (x ⫽ 1 in Fig. 9a兲, Ni-t2g states potentials for Ni and Fe ions. Electrons are simultaneously removed
are fully occupied and only one spin direction for the Ni-eg states is from the Ni-eg and Fe-eg bands upon further delithiation as shown in
occupied. For Fe-t2g and Fe-eg only the majority spin states are Fig. 9c. It clearly indicates that the Ni3⫹/Ni4⫹ and Fe3⫹/Fe4⫹ redox
occupied indicating high-spin Fe3⫹. The Co-t2g states are fully oc- reactions take place simultaneously. Such simultaneous redox reac-
tions of Fe and Ni have been reported in the Li共Ni,Fe兲O2 system by

Figure 6. Comparison of first charge-discharge curves of the

LiNi1/3Fe1/6Co1/6Mn1/3O2 materials synthesized at 750, 800, and 850°C for Figure 8. Comparison of experimental potential curve 共sample synthesized
16 h. at 750°C兲 with the predicted potential curve by first principles calculation.
A1138 Journal of The Electrochemical Society, 151 共8兲 A1134-A1140 共2004兲

Figure 9. Density of States of LixNi1/3Fe1/6Co1/6Mn1/3O2 at 共a兲 x ⫽ 1, 共b兲 x ⫽ 2/3, 共c兲 x ⫽ 1/3, and 共d兲 x ⫽ 0.

57
Fe Mossbauer and extended X-ray absorption fine structure 共EX- and Fe 2p edges. No obvious shifts in Co and Mn edges were ob-
AFS兲 investigations.12,14 Furthermore, upon delithiation (1/3 ⭐ x served. The results are in good agreement with the calculated change
⭐ 2/3), the Fe-eg states become empty and all four valence elec- of valence states during delithiation.
trons partially fill Fe-t2g states indicating low-spin for Fe4⫹. At x Lattice parameters and bond lengths.—The calculated lattice
⫽ 0, an electron is also pulled from the Co-t2g band 共Fig. 9d兲, parameters a & c at various lithium concentrations are depicted in
which demonstrates that the redox couple Co3⫹/Co4⫹ is activated at Fig. 10. The structural parameters of the most stable states were
the end of charge.
XPS was applied to corroborate the electronic behavior predicted
computationally. Ex situ XPS study was carried out to study the
valence shifts of Ni, Co, Fe, and Mn in LiNi1/3Fe1/6Co1/6Mn1/3O2
and in partially charged LixNi1/3Co1/6Fe1/6Mn1/3O2 (x ⬇ 1/2). The Table I. XPS binding energy for as-prepared and partially
electrodes were charged to 4.4 V. The binding energies of those charged materials.
cations in the as-prepared and partially charged compounds are tabu-
Ni Fe Co Mn
lated in Table I. Indicating by the binding energy shift of the 2p
2p3/2 /eV 2p3/2 /eV 2p3/2 /eV 2p3/2 /eV
electrons for the transition metal cations from their elemental
values,19 XPS confirms that the valence states of Ni, Fe, Co, and Mn LiNi1/3Fe1/6Co1/6Mn1/3O2 854.7 710.9 780.4 842.3
in the as-synthesized LiNi1/3Fe1/6Co1/6Mn1/3O2 are 2⫹, 3⫹, 3⫹, and 共1.7兲 共3.9兲 共1.4兲 共3.3兲
4⫹, respectively. Furthermore, as lithium is removed from the com- Lix Ni1/3Fe1/6Co1/6Mn1/3O2 855.5 711.8 780.4 842.1
pound, both Ni2⫹/Ni3⫹/Ni4⫹ and Fe3⫹/Fe4⫹ redox couples are ac- x ⬇ 1/2 共2.5兲 共4.8兲 共1.4兲 共3.1兲
tivated, revealed by an obvious shift in binding energies of Ni 2p Number in the parentheses—shift in binding energy.
 Journal of The Electrochemical Society, 151 共8兲 A1134-A1140 共2004兲 A1139

The calculated a lattice parameter decreases in the range of 1/3

⭐ x ⭐ 1 by approximately 2.2% and increases slightly in the range
of 0 ⭐ x ⭐ 1/3. The lattice parameter c increases by about 4.2% in
the range of 1/3 ⭐ x ⭐ 1 and decreases for 0 ⭐ x ⭐ 1/3. The
maximum change in cell volume is only 1.6%, which implies that
effect of stress and strain in the material during the lithiation-
delithiation cycle will be very small. In comparison the volume
change of LiCoO2 is about 3 to 4 percent.20
The average bond distances between the transition metal ions
and oxygen ions at different lithium contents are plotted in Fig. 11.
The bond distances of Ni-O and Mn-O in fully lithiated
Lix Ni1/3Fe1/6Co1/6Mn1/3O2 material (x ⫽ 1) are 关 2 ⫻ 2.04 2.05 2
⫻ 2.06 2.07兴 Å, 关2.06 2.07 2.08 3 ⫻ 2.09] Å, and 关1.93 4
⫻ 1.94 1.95兴 Å, 关 3 ⫻ 1.93 1.94 2 ⫻ 1.95] Å, respectively. The
bond distance of Ni-O is much longer than that of Mn-O, indicating
Ni has oxidation state of 2⫹ in this material. Fe-O has similar bond
distances as Ni-O 关2.04 4 ⫻ 2.05 2.06兴 Å, which can be explained
by the similar size of Fe3⫹ and Ni2⫹ 关Shannon radius兴.21 Co-O has
the typical bond distance of Co3⫹ in the layered compound.22
As lithium is removed, in the range of 2/3 ⭐ x ⭐ 1, Ni2⫹ is
oxidized and the bond distances of Ni-O become shorter. The large
spread of bond lengths for Ni-O in his range is due to Jahn-Teller
distortion of NiIIIO6 octahedron. Note that this also affects the Fe-O
and Mn-O bond lengths. Such distortion disappears upon further
lithium removal: in the range of 1/3 ⭐ x ⭐ 2/3, the Fe-O 共Fig. 11c兲
and Ni-O 共Fig. 11a and b兲 bond distances reduce simultaneously,
which is in good agreement with the DOS observations in Fig. 9. In
addition, in this range (1/3 ⭐ x ⭐ 2/3) the decrease in Fe-O bonds
Figure 10. Calculated lattice parameter a & c and volume at various lithium distance changes the crystal field splitting between the eg and t2g
concentrations.
bands,23 which leads to low-spin of Fe4⫹, as mentioned previously.

Conclusions
selected.There is no significant difference in the structural param- Motivated by a series of first principles calculations on
eters among different lithium-vacancy configurations at the same LiNi1/3Co1/3Mn1/3O2 with Co or Mn substituted by other metals,
composition. LiNi1/3Fe1/6Co1/6Mn1/3O2 was synthesized by a sol-gel method. We

Figure 11. Calculated transition metal—oxygen bond 共TM-O兲 distances of Lix Ni1/3Fe1/6Co1/6Mn1/3O2 (0 ⭐ x ⭐ 1).
A1140 Journal of The Electrochemical Society, 151 共8兲 A1134-A1140 共2004兲

predicted and confirmed that Fe substitution would lead to a lower 3. Z. H. Lu, D. D. MacNeil, and J. R. Dahn, Electrochem. Solid-State Lett., 4, A191
potential at the end of charge. Both XPS and first principles elec- 共2001兲.
4. J. Reed, G. Ceder, and A. Van Der Ven, Electrochem. Solid-State Lett., 4, A78
tronic structure computations indicate that Ni and Fe are simulta- 共2001兲.
neously oxidized in this material. Computations further indicate that 5. N. Yabuuchi and T. Ohzuku, J. Power Sources, 119, 171 共2003兲.
Co will only be oxidized at the very end of charge. The 6. B. J. Hwang, Y. W. Tsai, D. Carlier, and G. Ceder, Chem. Mater., 15, 3676 共2003兲.
LiNi1/3Fe1/6Co1/6Mn1/3O2 compound synthesized at 750°C shows re- 7. K. M. Shaju, G. V. Subba Rao, and B. V. R. Chowdari, Electrochim. Acta, 48, 145
共2002兲.
versible capacity of 150 mAh/g with reasonably good capacity re- 8. A. Van der Ven, M. K. Aydinol, G. Ceder, G. Kresse, and J. Hafner, Phys. Rev. B,
tention. In this work, we have demonstrated that an integrated ap- 58, 2975 共1998兲.
proach of computation and experiment has produced a new 9. C. Wolverton and A. Zunger, J. Electrochem. Soc., 145, 2424 共1998兲.
electrode material with very few iteration steps in the material de- 10. G. Ceder, Y. M. Chiang, D. R. Sadoway, M. K. Aydinol, Y. I. Jang, and B. Huang,
Nature (London), 392, 694 共1998兲.
sign cycle. We believe that such direct integration of ab-initio meth- 11. S. K. Mishra and G. Ceder, Phys. Rev. B, Condens. Matter, 59, 6120 共1999兲.
ods with experimental research holds promise for significantly short- 12. A. Demourgues, in Lithium Batterie Discussion, Arcachon 共2001兲.
ening the development cycle of materials. 13. C. Delmas, M. Menetrier, L. Croguennec, I. Saadoune, A. Rougier, C. Pouillerie,
G. Prado, M. Grune, and L. Fournes, Electrochim. Acta, 45, 243 共1999兲.
Acknowledgments 14. G. Prado, L. Fournès, and C. Delmas, J. Solid State Chem., 159, 103 共2001兲.
15. T. A. Hewston and B. L. Chamberland, J. Phys. Chem. Solids, 48, 97 共1987兲.
This work was supported in part by the MRSEC Program of the 16. M. K. Aydinol, A. F. Kohan, G. Ceder, K. Cho, and J. Joannopoulos, Phys. Rev. B,
National Science Foundation under award number DMR 02-13282 56, 1354 共1997兲.
and by the Singapore-MIT Alliance. We thank NPACI for the com- 17. Y. Arachi, H. Kobayashi, S. Emura, Y. Nakata, M. Tanaka, and T. Asai, Chem. Lett.,
32, 60 共2003兲.
puting resources.
18. J. Reed and G. Ceder, Electrochem. Solid-State Lett., 5, A145 共2002兲.
MIT assisted in meeting the publication costs of this article. 19. www.srdata.nist.gov/xps
20. T. Ohzuku and A. Ueda, J. Electrochem. Soc., 141, 2972 共1994兲.
References 21. R. D. Shannon, Acta Crystallogr., 32, 751 共1976兲.
22. Y. Koyama, I. Tanaka, H. Adachi, Y. Makimura, and T. Ohzuku, J. Power Sources,
1. T. Ohzuku and Y. Makimura, Chem. Lett., 2001, 744. 119-121, 644 共2003兲.
2. Z. H. Lu, D. D. MacNeil, and J. R. Dahn, Electrochem. Solid-State Lett., 4, A200 23. B. N. Figgis and M. A. Hitchman, Ligand Field Theory and its Applications,
共2001兲. Wiley-VCH, New York 共2000兲.