LiCoO2 concaved cuboctahedrons from symmetry-controlled topological reactions

Article abstract of DOI:10.1021/ja104852q

Morphology control of functional materials is generally performed by controlling the growth rates on selected orientations or faces. Here, we control particle morphology by crystal templating : by choosing appropriate precursor crystals and reaction conditions, we demonstrate that a material with rhombohedral symmetry - namely the layered, positive electrode material, LiCoO₂ - can grow to form a quadruple-twinned crystal with overall cubic symmetry. The twinned crystals show an unusual, concaved-cuboctahedron morphology, with uniform particle sizes of 0.5-2 μm. On the basis of a range of synthetic and analytical experiments, including solid-state NMR, X-ray powder diffraction analysis and HRTEM, we propose that these twinned crystals form via selective dissolution and an ion-exchange reaction accompanied by oxidation of a parent crystal of CoO, a material with cubic symmetry. This template crystal serves to nucleate the growth of four LiCoO₂ twin crystals and to convert a highly anisotropic, layered material into a pseudo-3-dimensional, isotropic material.

Full text of DOI:10.1021/ja104852q

Published on Web 12/13/2010
LiCoO Concaved Cuboctahedrons from Symmetry-Controlled
2
Topological Reactions
†
‡
‡
‡
,†,§
Hailong Chen, Lijun Wu, Lihua Zhang, Yimei Zhu, and Clare P. Grey*
Department of Chemistry, Stony Brook UniVersity, Stony Brook,
New York 11794-3400, United States of America; BrookhaVen National Laboratory, Upton,
New York 11973, United States of America; and Department of Chemistry, Cambridge
UniVersity, Lensfield Rd, Cambridge, CB2 1EW, United Kingdom
Received June 15, 2010; E-mail: cpg27@cam.ac.uk
Abstract: Morphology control of functional materials is generally performed by controlling the growth rates
on selected orientations or faces. Here, we control particle morphology by “crystal templating”: by choosing
appropriate precursor crystals and reaction conditions, we demonstrate that a material with rhombohedral
2
symmetrysnamely the layered, positive electrode material, LiCoO scan grow to form a quadruple-twinned
crystal with overall cubic symmetry. The twinned crystals show an unusual, concaved-cuboctahedron
morphology, with uniform particle sizes of 0.5-2 µm. On the basis of a range of synthetic and analytical
experiments, including solid-state NMR, X-ray powder diffraction analysis and HRTEM, we propose that
these twinned crystals form via selective dissolution and an ion-exchange reaction accompanied by oxidation
of a parent crystal of CoO, a material with cubic symmetry. This template crystal serves to nucleate the
growth of four LiCoO
2
twin crystals and to convert a highly anisotropic, layered material into a pseudo-3-
dimensional, isotropic material.
Introduction
Many highly orientated thin films, or single crystals, show
anisotropic chemical and physical properties, such as electronic
and ionic conductivities, due to the intrinsic anisotropy of the
crystal lattice. For example, in the field of lithium ion batteries
(
LIBs), Li ion diffusivity in both positive and negative
electrodes, are often anisotropic, the conduction direction(s)
being determined by the lowest activation barriers for Li
migration. One-dimensional Li ion diffusion occurs in the
1
,2
electrode material LiFePO
4
,
while in the commercial
electrode material, LiCoO , the diffusion is 2-D. In principle,
2
isotropic diffusion is desirable, since this results in a larger
active surface area for Li-transfer from the electrolyte,
potentially leading to improved LIB rate performance.
Although it is very difficult to control the dimensionality of
the conduction process, which is mostly determined by the
lattice, control of the morphology and the assembly of the
primary particles represent possible routes to alter the rate
performance.
2
Figure 1. SEM image of the surface of the “desert rose” LiCoO balls
3
prepared in our earlier study. Most of the thin rods/plates penetrate or are
connected to each other with a characteristic angle of about 71°, as illustrated
by the red lines in the figure.
samples made by similar methods, are not randomly assembled,
but rather are formed by LiMO plates that grow out of other
2
3
In earlier work, we showed that primary particles of the 2D
plates, with a characteristic angle of ∼71° between the connected
conductor LiCoO
into a pseudo-3-D conductor with good performance at high
rates, by the preparation of LiCoO “balls” with the so-called
desert rose morphology. These balls of LiCoO , and other
(M ) Fe, Mn, Ni, etc)
2
, with high aspect ratios, could be converted
plates or twins (see Figure 1). Not directly attempting to improve
2
the electrochemical performance of LiCoO , but as part of a
2
fundamental study initiated to explore the mechanistic and
structural origins of this geometric relationship, we now show
that with appropriate synthetic conditions, we can control the
growth of the plates to prepare an unusual isotropic, concaved
2
x 2
transition metal doped LiM1-xCo O
†
Department of Chemistry, Stony Brook University.
Brookhaven National Laboratory.
Department of Chemistry, Cambridge University.
‡
§
2
cuboctahedron-shaped form of LiCoO . Hydrothermal methods
were used in this study, instead of the low temperature molten
3
(
(
(
1) Islam, M. S.; Driscoll, D. J.; Fisher, C. A. J.; Slater, P. R. Chem.
Mater. 2005, 17, 5085–5092.
fluxes used earlier consisting of LiOH, CsOH, and KOH. The
hydrothermal synthesis approach allows for more tunable
experimental parameters, (e.g., type and ratio of the starting
materials), which are sometimes limited by the melting point
2) Nishimura, S.; Kobayashi, G.; Ohoyama, K.; Kanno, R.; Yashima,
M.; Yamada, A. Nat. Mater. 2008, 7, 707–711.
3) Chen, H. L.; Grey, C. P. AdV. Mater. 2008, 20, 2206–2211.
2
62 9 J. AM. CHEM. SOC. 2011, 133, 262–270
10.1021/ja104852q  2011 American Chemical Society

LiCoO
2
Concaved Cuboctahedrons
A R T I C L E S
or solubility in molten flux systems. We first report the synthesis
and characterization of a first batch of cuboctahedrons (sample
A) and we then describe a detailed investigation of the
mechanisms by which they form. An optimized synthesis
protocol is devised based on this mechanistic study, leading to
the synthesis of a second sample (sample B) with an almost
1
00% yield of the cuboctahedron morphology.
Experimental Section
Synthesis. Two synthetic procedures were used in this work to
prepare the concaved cuboctahedron morphology. In the first
procedure (sample A), 0.58 g (2 mmol) Co(NO
8%) was dissolved in 5 mL of distilled water to form a clear
reddish solution. Seven grams of KOH (Sigma-Aldrich, >85%) and
g of LiOH (Sigma-Aldrich, 99%) were dissolved in 10 mL of
distilled water. The Co(NO solution was then added slowly to
3 2
) (Sigma-Aldrich,
9
1
3 2
)
the hydroxide solution while stirring. The resulting solution was
transferred to a Teflon-lined hydrothermal bomb and was heated
statically in an oven at 200 °C for 24 to 48 h. After the bomb was
cooled down to room temperature, the product was washed with
distilled water and centrifuged several times. Some of the samples
were washed with 0.1 M hydrochloric acid to remove the Co(OH)
The final washed products were dried overnight in an oven at
0 °C.
Higher yields of the concaved cuboctahedron morphology were
observed by using a second procedure, in which nanoparticles of
CoO or Co(OH) with an equivalent amount of Co (2 mmol) were
2
.
8
2
used as the Co source instead of cobalt nitrate (Sample B). The
nanoparticles of CoO were prepared by dehydration of commercial
Co(OH)
at 250 °C for one hour. Nanoparticles of Co(OH)
particle size of 20 nm) were prepared by a precipitation method:
.1 M Co(NO solution was added slowly into 0.1 M LiOH (Co:
Li ratio ) 1:2) solution while stirring. The precipitation product
was washed with distilled water and dried at 60 °C overnight. An
optimal ratio to obtain the highest yield of concaved cuboctahedrons
2
(Sigma-Aldrich) in a tube furnace under Ar atmosphere
2
(with average
0
3 2
)
Figure 2. SEM images of (a) concaved cuboctahedron Li
x 2
CoO particles;
(b) “[100] ” and (c) “[111]” views of a single particle; (d) Schematic drawing
of the particle; (e) another morphology obtained in the same sample batch;
and (f) schematic of the truncated cube shown in (e) showing the exposed
was determined to be LiOH: KOH: H
2
O: CoO (or Co(OH)
2
) 1 g:
(
100), (111), and (110) faces.
3
g: 3 mL: 150 mg (CoO) or 186 mg (Co(OH)
2
), although increasing
the KOH concentration to 7 g (keeping all other reagent concentra-
tions constant) did not alter the yield significantly. The hydrothermal
reaction was performed at 200 °C for 24-48 h, to complete the
Corp.) were assembled in an argon-filled glovebox. Each cell
contains typically about 6-8 mg of active material, separated from
the Li foil anode by a piece of Celgard separator (Celgard, Inc.,
oxidation to Co3+ and minimize the impurities, e.g. Co
3 4
O . The
6
U.S.A.). A 1 M solution of LiPF in ethylene carbonate/dimethyl
products were filtered and then washed with distilled water and
dried at 70 °C overnight.
carbonate (1:1) was used as the electrolyte. Electrochemical
experiments were carried out on a battery cycler (Arbin Instruments,
College Station, TX) in galvanostatic mode at various rates.
Characterization. Powder X-ray diffraction patterns were col-
lected either on a Rigaku MiniFlex diffractometer or at beamline
X16C at Brookhaven National Laboratory. The SEM images were
taken by using a LEO-1550 field emission scanning electron
microscope. JEM-3000F and JEM-2100F high-resolution transmis-
sion electron microscopes were used to obtain HRTEM images and
single crystal select area electron diffraction (SAED) patterns. The
simulations of SAED patterns and HRTEM were carried out using
the programs developed in BNL based on the Bloch wave and
Results and Discussion
Synthesis and Characterization of the Cuboctahedron
Particles: Sample A. Hydrothermal syntheses of LiCoO were
2
performed initially by using starting material combinations
similar to those employed in our previous molten flux method
(
3 2 2
e.g., Co(NO ) (0.58 g), LiOH (1 g), and KOH (7 g), H O (15
multislice methods, respectively.
7Li magic-angle-spinning (MAS) NMR spectroscopy was per-
mL), corresponding to molar ratios of 1:21:53: 417, respec-
3
formed on some cuboctahedron LiCoO
2
samples and charged
tively), except that CsOH was omitted in this study and water
“
desert rose” LiCoO samples (as references) with a 1.8 mm probe
2
was employed as the solvent. Surprisingly, the final product of
a 24 h reaction at 200 °C contained regular, uniform and
extremely unusual concaved cuboctahedron-shaped crystals, as
seen by SEM (Figure 2 a-c) as a minor morphology. Figure
on a CMX-200 spectrometer using a magnetic field strength of 4.7
T. A spinning speed of 30 kHz and a rotor-synchronized spin-echo
sequence (π/2-τ-π-τ-acq) were used to acquire the spectra. The
7
Li spectra were collected at an operating frequency of 77.10 MHz
2
d shows a computer-generated drawing of the particle, based
with π/2 pulses of 2.5 µs and a delay time of 0.5 s. The spectra
on the SEM images of the particles viewed from different
directions (Figure 2b-c). Each cuboctahedron particle consists
of four equivalent hexagonal plates, with angles between two
of these plates of 70.5°, so as to maintain overall cubic
symmetry. The angles between the plates are the same as those
7
were referenced to 1 M LiCl in H
2
O at 0 ppm.
sample was mixed
with poly vinylidene fluoride binder and acetylene black (6:1:3 wt
) in N-methyl pyrrolidone (NMP) to make a thick slurry. The
Electrochemistry. The as-prepared LiCoO
2
%
slurry was deposited on an aluminum foil by using a doctor-blade
method and dried at 80 °C overnight. Coin cells (CR2032, Hohsen
2
observed in our earlier syntheses of “desert rose” LiCoO (Figure
J. AM. CHEM. SOC. 9 VOL. 133, NO. 2, 2011 263

A R T I C L E S
Chen et al.
Figure 3. 7Li MAS NMR spectra of (a) the first synthesis (sample A), (b)
9
0% charged stoichiometric (desert-rose) LiCoO
2
, (c) pristine desert rose
LiCoO , and (d) the second synthesis (sample B; stoichiometric, cubocta-
2
hedron LiCoO
and d.
2
). The spinning side bands are indicated by asterisks in c
1
). To the best of our knowledge, this cuboctahedron morphol-
4
ogy has only been reported once before, for CuS by Wu et al.,
4
but no detailed growth mechanism was proposed. Our crystals
are, however, much more uniform than the CuS particles and
always comprise 4 hexagonal plates, while the CuS particles
sometimes contained only 2 or 3 plates. Apart from the
geometrical simplicity and the aesthetic qualities of these
crystals, a number of scientific questions are raised, which
include the structure of the material and how and why it forms.
The X-ray diffraction (XRD) pattern of the product shows
the presence of a mixture of Co(OH)
Co(OH) being formed from the reaction of Co(NO
and CoO from the decomposition of the unstable Co(OH)
2 2
, CoO, and LiCoO ,
-
2
3 2
) and OH ,
2
Figure 4. Morphology of a concaved cuboctahedron (a) viewed along the
normal of the V1 plate. (b) Selected-area diffraction (SAED) pattern of
V1. (c) View along the [-100]V1 or the equivalent [100] V2, or [241]V3,
or [-2-4-1]V4 directions. (d-f) SAED patterns of the different twin
5
phase. After quickly washing the product by using 0.1 M HCl
to dissolve the Co(OH) phase, only CoO and LiCoO are seen
by XRD, while the cuboctahedrons are still observed by SEM.
This indicates that the cuboctahedron are either CoO or LiCoO
2
2
x 2
variants of Li CoO : (d) [-100]V1 (from red circled area); (e)[100]V2 (from
2
.
green circled area); (f)[241]V3 and [-2-4-1]V4 (blue circled area).
The acid wash was only performed on one batch solely to
exclude the possibility that the concaved octahedrons were due
similarly contains a resonance at ∼100 ppm, which is consistent
with other charged stoichiometric samples which resonate at
to the acid-soluble phase, Co(OH)
applied to the XRD pattern of the mixture (see Figure S1b in
the Supporting Information, SI). The c parameter of the LiCoO
2
. Rietveld refinement was
7
6
0-100 ppm for Li
x
CoO
2
, x < 0.75. The shift seen in these
latter samples is caused by the Knight shift, which indicates
that the sample is metallic. Although we have not analyzed the
origin of the shifts of the cuboctahedron sample in any detail,
2
phase extracted from the refinement is 14.21 Å, slightly larger
than the usual value of bulk LiCoO2, which is around 14.06
6
2
the NMR results reveals that the first batch LiCoO is not
Å. This suggests that this LiCoO
stoichiometric.
2
phase may not be
stoichiometric. The electrochemical capacity of this sample is
also low, when it is used as an electrode material in a lithium-
ion battery (∼30-50 mAh/g between 2-4.8 V), again indicating
7Li MAS NMR spectroscopy was performed to explore the
Li nonstoichiometry of the lithiated phase, since the method is
2
that this LiCoO phase is not stoichiometric.
extremely sensitive to small deviations in the stoichiometry of
7
Twinned Nanocrystals: Transmission Electron Microscopy
Study of Sample A. Structure. Although it is difficult to
determine which of the phases seen by powder XRD corre-
sponds to the cuboctahedrons in the mixture, and the size of
the cuboctahedrons is too small for single crystal XRD studies,
electron diffraction on a single particle under TEM provides
an effective method with which to investigate their structure.
A transmission electron microscopy (TEM) image of a concaved
cuboctahedron viewed along the normal direction of one
hexagonal plate (labeled as V1) is shown in Figure 4a. The
selected-area electron diffraction (SAED) pattern taken from
the V1 plate is shown in Figure 4b. This pattern contains bright
the LiCoO
2
phase. A broad resonance at 100 ppm is observed
of the sample (Figure 3a), rather than the single shift at 0 ppm
7
expected for a stoichiometric, diamagnetic LiCoO
spectrum of the sample of the “desert-rose” form of LiCoO
charged to 90% is shown for comparison (Figure 3b). This
2
sample. The
2
(
(
(
4) Wu, C. Y.; Yu, S. H.; Antonietti, M. Chem. Mater. 2006, 18, 3599–
3
601.
5) Lide, D. R. CRC Handbook of Chemistry and Physics; Section 4, 79th
Edition; CRC press: Florida, USA.
6) Okubo, M.; Hosono, E.; Kim, J.; Enomoto, M.; Kojima, N.; Kudo,
T.; Zhou, H. S.; Honma, I. J. Am. Chem. Soc. 2007, 129, 7444–7452.
(7) Menetrier, M.; Saadoune, I.; Levasseur, S.; Delmas, C. J. Mater. Chem.
1
999, 9, 1135–1140.
H x 2
spots that can be indexed with the [001] pattern of Li CoO
2
64 J. AM. CHEM. SOC. 9 VOL. 133, NO. 2, 2011

LiCoO
2
Concaved Cuboctahedrons
A R T I C L E S
(
where the subscript H denotes the index based on the hexagonal
lattice of LiCoO ). The spots are not consistent with either CoO
or Co : the lattice parameters of Li CoO and CoO are
sufficiently different to allow these two phases to be resolved,
while those due to Co and LiCoO can be distinguished based
on differences in the extinction conditions. The normal of the
hexagonal plate is [001] , while its edges are indexed as (100)
010) and (1-10) . The very weak spots in Figure 4b, indicated
2
3
O
4
x
2
O
3 4
2
H
H
,
(
H
H
by the black circles, suggest the presence of a 2a × 2a LiCoO
2
superstructure. This superstructure is tentatively ascribed to
either long-range ordering due to partial Co occupancy on the
Li site, and/or proton or vacancies on the Li site, indicating
2
some nonstoichiometry of the LiCoO particles, consistent with
the NMR results for the sample.
Twinning. Nanoparticles with unusual, highly symmetric
morphologies have been reported for various functional materi-
8,9
10,11
als, including noble metals catalysts and semiconductors.
However, in these cases, the symmetry of the particle is
consistent with the symmetry of the lattice of the material,
different morphologies simply resulting from different crystal
truncations and preferred growth directions of the particles. By
Figure 5. (a) Projection of the CoO lattice along the [10-1] direction,
showing the stacking of atoms along [111] direction; (b) Projection of
2
LiCoO lattice along the b direction, showing the stacking of atoms along
[
001] direction. Oxygen, cobalt, and lithium atoms are plotted as red, blue,
and green balls, respectively.
2
contrast, LiCoO , either the stoichiometric or nonstoichiometric
versions, adopts a layered, hexagonal lattice and yet the
concaved cuboctahedrons show overall cubic symmetry. The
question that must now be answered is: how and why do
the four hexagonal plates grow together to form an isometric
crystal? In principle, twinning represents one way to break the
intrinsic symmetry of a crystal. However, it usually results in
additional sets of reflections in the diffraction patterns of the
twinned crystal, due to a loss of symmetry of the overall particle;
this is not the case for the cuboctahedrons (see Figure 4). Given
The Growth Mechanism. In order to determine how and why
the cuboctahedrons grow to form this morphology, we carefully
examined the first synthesis batch for any clues in the form of
other phases and particles that are present in this sample. Large
crystals were also seen in this batch before and after washing,
which consist of regular polygons with small pit marks on select
faces. Figure 2e shows a cropped SEM image of one of these
crystals, which are geometrically derived by truncating the edges
and corners of a cube, exposing the {100}, {111}, and {110}
faces. This is illustrated in Figure 2f, which shows a model of
the truncated cube. Although much larger in size, these cubic
crystals are fewer in number in the product mixture (see Figure
S1c in the SI for a lower magnification SEM image of the same
mixture), than the noticeably smaller particles of the other
components. The larger particles are ascribed to the cubic
2
that the lattice of LiCoO can be described by a rhombohedral
unit cell with R ≈ 89.88°, i.e., the deviation from the cubic
unit cell is small, certain classes of twin variants may be present
in which the splitting or doubling of the reflections may not be
distinguishable in the SAED pattern due to the limited set of
reciprocal vectors that we are probing. However, the existence
of the twins can be verified based on the different extinctions
between the SAED patterns along the real 3-fold and the pseudo-
j
material CoO (space group Fm3m, a ) 4.26 Å), since the
reflections of CoO in the XRD patterns are very sharp and
intense (see Figure S1b in the SI). The crystals show pronounced
selective pitting of their {100} faces to expose the {111} planes,
no pits being observed on the {111} faces. One explanation for
this selectivity is that the {111} faces are charge polarized, so
that they are more easily stabilized by electrical double layers,
making it more difficult to extract an ion from the surface; in
3
-fold axes (Figure 4c-f). The SAED patterns shown in Figure
4d-f are taken from different areas indicated by the circles and
arrows in Figure 4c, without changing the tilt angle of the
crystal. Patterns d and e can be indexed as LiCoO diffraction
2
patterns along the [100] zone axis for plate V1 and [-100] for
the second plate V2, respectively, and the c axes of V1 and V2
twin variants can therefore be determined, as labeled in the figure
2
+
contrast, the {100} faces are charge neutral, and the Co and
2-
12
(
see Supporting Information for more details). The angle
O
ions can be extracted more readily. Pitting in highly basic
between the c axes of V1 and V2, which is also the angle
between the c* axes of their reciprocal lattice, is 70.5°, consistent
with the angle between the {111} planes expected for a cubic
system. The V1 and V2 twins are, thus, related by a reflection
on the (0-14) plane. The reflections for the V3 and V4 plates
are superimposed when viewed down the [241] axes of V3
solutions occurs in many materials, and in more exceptional
cases, as seen for example in PbS cubes under select conditions,
can result in concave, regularly shaped (in this case pyramidal)
1
3
pits which alter the morphology of the whole particle. Solely
on the basis of the geometry of the two objects, a concaved
cuboctahedron can be formed directly from a cube, via pitting
to maximize the surface area of the {111} faces.
2
(
Figure 4f). On the basis of this analysis, it is clear that the
concaved cuboctahedron consists of four hexagonal plates, with
The lattices of CoO and LiCoO are similar, the rhombohedral
each plate being a twin variant. The plate normal is the c axis
distortion of the unit cell of LiCoO , with R ≈ 89.88°, being
2
(
the 3-fold axis) of the variant (i.e., the Li
x
CoO
2
plate).
extremely small. Figure 5a shows the stacking of the CoO atoms
along the [111] direction and b shows the stacking of atoms of
LiCoO along [001] direction. The stacking of the two lattices
2
(
(
8) Sun, Y. G.; Xia, Y. N. Science 2002, 298, 2176–2179.
9) Tian, N.; Zhou, Z. Y.; Sun, S. G.; Ding, Y.; Wang, Z. L. Science
2
007, 316, 732–735.
(
(
10) Manna, L.; Milliron, D. J.; Meisel, A.; Scher, E. C.; Alivisatos, A. P.
Nat. Mater. 2003, 2, 382–385.
(12) Vayssieres, L.; Keis, K.; Hagfeldt, A.; Lindquist, S. E. Chem. Mater.
2001, 13, 4395–4398.
11) Manna, L.; Scher, E. C.; Alivisatos, A. P. J. Am. Chem. Soc. 2000,
(13) Hou, Y.; Kondoh, H.; Ohta, T. Cryst. Growth Des. 2009, 9 (7), 3319-
1
22, 12700–12706.
3123.
J. AM. CHEM. SOC. 9 VOL. 133, NO. 2, 2011 265

A R T I C L E S
Chen et al.
are very similar in terms of the closest O-O distances within
the layer and between layers (in particular, the OsO distances
defining the Li and Co octahedra and thus layers are similar)
and only differ in the nature and stacking of the metal layers.
Synthesis of the Stoichiometric LiCoO
2
Cuboctahedrons:
Sample B. The role of CoO in templating the growth of LiCoO
has not, to our knowledge, been reported before. The usual
precursors used in the solid state reactions to form LiCoO are
where the Co is already fully or
partially oxidized to Co(III). As the Li CoO cuboctahedrons
2
+
2+
2
1
7
18
The LiCoO
other (111) layer of Co by Li so that the 3-fold symmetry along
cubocta-
2
lattice can be derived from CoO by replacing every
3 4
either CoOOH or Co O ,
x
2
3
out of 4 〈111〉 directions is lost. Thus, the LiCoO
2
were not the major phase/morphology in our first sample batch,
we strove to provide additional support for the formation
mechanism by improving the yield of this morphology, using
the proposed growth mechanism as a guide with which to
optimize the reaction conditions. To this end, when commercial
(∼5 µm size) CoO was used as the Co source as a replacement
hedrons can, at least from a crystallographic point of view, be
derived by a combination of CoO lithiation and selective pitting.
+
2+
The lithiation of CoO can in principle occur via Li /Co ion
exchange of alternate layers, a process that must be accompanied
2
+
3+
by the oxidation of remaining Co in the lattice to Co for
2
+
charge balance, the oxidation of a Co ion nearby an inserted
Li ion being favored on the basis of electrostatics. The basic
3 2
for Co(NO ) in the hydrothermal synthesis, very few cuboc-
+
tahedrons and more pitted large crystals were seen in the
products, indicating that the size of the parent crystal is crucial
in controlling the morphologies formed in this reaction. To
reduce the size of parent crystal, ∼50 nm sized particles of CoO,
conditions used here should increase the stability of Co(III) over
Co(II), driving this process and hindering dissolution of the ion-
exchanged Co3 phase. Li /Co exchange is likely to be rapid
at, or close to the surface, of the crystal, but is unlikely to be
fast for the whole crystal. Thus, the Li layers at the center of
the crystal may not be fully lithiated. Furthermore, this
mechanism will be less effective for large micrometer-sized
particles. We note that there is precedence for rapid ion-
exchange reactions in nanoparticles in materials where the bulk
+
3
+
2+
2
prepared by decomposition of commercial Co(OH) , were used
as the Co source. This led, after 48 h, to the formation of LiCoO
and Co as the major and minor phases, respectively. Many
more cuboctahedrons were seen in this sample. This is consistent
2
3
O
4
+
2+
2+
with the suggestion that the Li -Co ion-exchange/Co
oxidation reaction is key to the formation of the cuboctahedron,
since this reaction should be favored for nanoparticles. The
1
4,15
reactions are slow.
A number of phenomena are thought to
be responsible for this phenomenon, including (i) the strong
particle-size dependence for ion-exchange involving diffusion
controlled reactions, (ii) the fact that the reaction fronts in ion-
exchange reactions can be similar to the size of the nanomaterial,
so that a sharp interface between the non and lithiated phases
3 4
appearance of Co O is ascribed to a second oxidation process
that competes with the lithiation-oxidation reaction under these
conditions.
Experiments were performed using Co
source, to explore the role of Co (see SI for more details),
which revealed that although the larger Co particles were
stable under the basic hydrothermal reaction conditions used
here, nanoparticles of Co converted to LiCoO . However,
no concaved cuboctahedron particles were ever observed, even
after long reaction times, indicating that Co does not appear
to serve as a template crystal for this morphology. Again this
3 4
O particles as the Co
3
O
4
(
in this system) are not observed, and (iii) the increased
3
O
4
flexibility of nanosystems to accommodate expansions/contrac-
tions of the layers.
3
O
4
2
Since the original proposed crystal template (CoO) is cubic,
3
O
4
LiCoO
2
has an equal probability of being generated with its
unique 3-fold axis (the c axis in a hexagonal lattice) along the
3 4
suggests that it is CoO, and not Co O , that is the template of
[111], [11-1], [1-11], or [-111] axes of the parent phase
(CoO). Thus, four twin variants can simultaneously grow from
the cuboctahedrons.
Both increased pH and Li concentration should favor Li ion-
exchange (reaction 1) according to the following reactions:
+
the same parent CoO crystal to generate a 4-fold twinned crystal
that, despite the symmetry of the phase that grows, can maintain
overall cubic symmetry. A more rigorous analysis of the LiCoO
twins, by using the space group theory on domain formation
2
+
-
4
CoO + 1/2O + 2Li + 6OH + H O f 2LiCoO +
2
2
2
16
2Co(OH)42- (1)
during a disorder-order structural transformation (as described
in detail in the SI), is consistent with this suggestion: LiCoO
2
j
-
2-
adopts the space group R3m, which is a subgroup of the space
4
CoO + 1/2O + 2OH + H O f Co O + Co(OH)
2 2 3 4 4
j
group adopted by CoO (Fm3m). On the basis of the orders of
(
2)
the two groups, it is straightforward to show that there will be
four twin variants with orientations between them consistent
with the observed cuboactahedrons.
or
3
CoO + 1/2O f Co O
4
(3)
2
3
Finally, it is important to note that while in principle, ion-
exchange (and oxidation) of both Co(OH)
possible, both phases being present in the reaction mixture
Figure S1a,b in the SI), it is not clear how the former phase
can act to template the symmetric cuboctahedron. Consistent
with this, in our earlier synthesis of “desert-rose” LiCoO , where
LiCoO grew from Co(OH) template crystals, no symmetric
cuboctahedrons were observed. CoOOH is excluded because
we never saw any evidence for this phase in any of the XRD
patterns extracted from the hydrothermal mixture.
2
and CoO are
the latter reaction occurring without Co2+ dissolution. We note
that reactions (1) and (2) will be limited by the solubility of
(
2-
Co(OH)
4
. Thus further reaction must proceed by the precipita-
2-
tion Co(OH)
4
2 2
as CoO, Co(OH) , or directly as LiCoO (see
2
later). It is important to note that these reactions represent a
simplified form of the many reactions and equilibria occurring
in this system; for a discussion of the separate reactions please
see the SI. However, consistent with these reactions, when the
2
2
3
2 3 4
LiOH concentration was reduced, less LiCoO and more Co O
(
14) Son, D. H.; Hughes, S. M.; Yin, Y. D.; Alivisatos, A. P. Science 2004,
(17) Pralong, V.; Delahaye-Vidal, A.; Beaudoin, B.; Gerand, B.; Tarascon,
J. M. J. Mater. Chem. 1999, 9, 955–960.
3
06, 1009–1012.
(
(
15) Dloczik, L.; Konenkamp, R. Nano Lett. 2003, 3, 651–653.
16) Wu, L. J.; Zhu, Y. M. Philos. Mag. A 1997, 76, 481–492.
(18) Shao-Horn, Y.; Hackney, S. A.; Johnson, C. S.; Kahaian, A. J.;
Thackeray, M. M. J. Solid State Chem. 1998, 140, 116–127.
2
66 J. AM. CHEM. SOC. 9 VOL. 133, NO. 2, 2011

LiCoO
2
Concaved Cuboctahedrons
A R T I C L E S
was formed and fewer or no cuboctahedrons were seen. In
contrast, on reducing the amount of water used in the reactions,
while keeping all of the other reagents constant, more cuboc-
3 4
tahedrons and less Co O were seen. This was ascribed to the
increase in pH and Li+ concentration. The additional empty
space left in the hydrothermal vessel will increase the amount
2 2
of air/O , but this additional O is not sufficient to prevent
cuboctahedron formation. On completely removing water from
the starting materials, for example by performing a molten salt
reaction in the sealed hydrothermal bomb, no concaved cub-
2-
octahedrons are seen, presumably because the Co(OH)
solubility is reduced further.
4
An analysis of the results from all our synthetic attempts
described above suggested that nano-Co(OH) might be a better
2
starting material than CoO for this reaction. This hypothesis
was based on the observation that under hydrothermal heating
at 200 °C, nano-Co(OH) will quickly decompose to form
2
nanosized CoO particles, and the hypothesis that, under
optimized conditions, the ion exchange and oxidation reactions
should occur before the newly formed CoO nanoparticles
aggregate and grow larger. To test this hypothesis, laboratory-
synthesized ∼20 nm Co(OH)
higher yields of the cuboctahedrons were observed. Using this
nano-Co(OH) along with an optimized amount of water of 3
mL (e.g., 1 g of LiOH, 3 mL of H O and 3 g of KOH with 186
mg Co(OH) as starting materials), led to essentially phase-
pure LiCoO , with the uniform cuboctahedrons representing the
2
particles were used, and much
2
2
2
2
overwhelming morphology (Figure 6a,b). The optimum water
content was found by reducing the water concentration system-
atically keeping all other reagent concentrations (used to prepare
sample A) constant and monitoring the cuboctahedron yield.
Reducing the KOH:LiOH concentration from 7 to 3 did not
alter the yield noticeably. CoO still presumably acts as the
template crystal, but it now has an even smaller particle size
and higher reactivity than the 50 nm CoO precursor also tested
Figure 6. (a) High magnification and (b) larger scale SEM image of phase-
pure LiCoO with the cuboctahedron morphology.
2
(
see earlier), since it is formed in situ by decomposition of the
0 nm Co(OH) . (See SI for further evidence for the formation
2
2
of precursor CoO crystals in this reaction). This also helps
explains the coexistence of the giant pitted CoO crystal and
the cuboctahedrons in the first batch (sample A); since the
-
2 3 2
Co(OH) phase formed from the reaction of Co(NO ) and OH
has a broad size distribution this will also result in a broad size
distribution of CoO crystals formed from Co(OH) decomposi-
tion. The smaller CoO crystals are then ion-exchanged and
oxidized forming the concave Li Co1-xO crystals, which grow
to become the micrometer size cuboctahedron LiCoO , while
2
x
2
the large CoO crystals keep growing and are only heavily pitted,
but cannot be completely converted to the fully concaved and
lithiated structure.
Characterization of the Stoichiometric Cuboctahedrons. The
high-yield cuboctahedron sample, based on a range of analyses,
has a composition that is extremely close to stoichiometric
LiCoO
XRD pattern (see Figure 7). Cell parameters of a ) b )
.814(6), c ) 14.046(5) Å for LiCoO are extracted via Rietveld
refinement (with wRp ) 8.05%.), which are within the ranges
2 3 4
and only a trace amount of Co O is detected in the
Figure 7. The XRD pattern and Rietveld refinement of the high yield
cuboctahedron LiCoO2. The black arrow indicates the characteristic reflection
of the Co O impurity.
3 4
2
2
6
surrounded by Co3+ (d ), and none of the resonances charac-
teristic of nonstoichiometry are observed (Figure 3d). Consistent
with this, no superstructure reflections were detected in the
SAED pattern of this sample (see Figure 8d), in contrast to the
SAED patterns obtained from the first batch (Figure 8b).
Although this concaved shape is not ideal for creating a high
density compact cathode with high volumetric power density,
6
expected for stoichiometric micrometric LiCoO
2
and direct
coupled plasma (DCP) analysis gives a Li:Co ratio of
7
1
.05((0.05):1. The Li MAS NMR spectrum of this sample is
shown in Figure 3d. It is very similar to the spectrum of
stoichiometric LiCoO (for example the “desert rose” form
shown in Figure 3c): the spectrum is dominated by a signal at
ppm, which is due to the diamagnetic environments of Li
2
0
J. AM. CHEM. SOC. 9 VOL. 133, NO. 2, 2011 267

A R T I C L E S
Chen et al.
Figure 8. TEM studies of the cuboctahedra: (a) Morphology of nonsto-
ichiometric Li
LiCoO concaved cuboctahedra (route B). The scale bars in (a) and (c) are
00 and 500 nm, respectively. Parts (b) and (d) are SAED patterns of the
particles shown in (a) and (c) along the same ([214] ) zone axes.
x 2
CoO concaved cuboctahedra (route A), and (c) stoichiometric
2
2
H
Figure 10. (a) Growth mechanism of the concaved cuboctahedron. Areas
shown in dark represent the leached planes. For clarity, only one set of
leached planes is shown for each stage and the size of the parent crystal
and the final twin crystal is not proportionally scaled: in the experiments,
the parent crystals are only ∼20-50 nm and the final twin crystals are
about 1 µm. (b) Example of an “intermediate” morphology.
the neighboring layer is oxidized to Co(III) to maintain charge
neutrality, via a mechanism that we refer to as a simultaneous
“oxidation-ion-exchange” reaction. Since Co(III) is much more
stable and Co(II) is less soluble in the basic solution, the Co(III)
layer is stabilized by this Li insertion. The Li ions continue to
exchange with Co(II) until they approach the center of the
mother CoO nanoparticle, alternating Li and Co3 layers being
+
20
2
favored by the same driving forces that favor layered LiCoO .
Figure 9. Charge and discharge curves for the first 5 cycles for high-yield
cuboctahedron LiCoO . The cycling window is 2-4.3 V. A current density
of 200 mA/g, corresponding to a rate of 1.5C was used.
Dissolution will also occur on other faces, such as the {110}
and {111} faces, always leaving behind stable {111} faces. At
2
2-
the same time, Co(II) in solution (largely Co(OH)
4
) will also
+
be oxidized to Co(III) and will reprecipitate together with Li
and oxide ions on the surface of the newly formed LiCoO -like
domains, LiCoO growing along the [001] , [010] , and [110]
directions, as observed in our previous study. This also gives
rise to the characteristic angles between the LiCoO plates seen
2
the LiCoO form yields a moderate capacity of 142 mAh/g at
2
a rate of 1.5 C, as shown in Figure 9. At lower rates, e.g., C/15,
the capacity of this material is about 135 mAh/g between 2.5
2
H
H
H
3
1
9
2
and 4.3 V, being similar to that of commercial LiCoO .
2
in both the desert rose (Figure 1) and cuboctahedron morphol-
Summary and More Detailed Analysis of the Growth
Mechanism
ogy. Note that this precipitation must occur so as to consume
2-
the Co(OH)
4
produced by ion-exchange mechanism (reaction
The proposed mechanism for the growth of the cuboctahe-
drons can now be summarized as follows: a parent (template)
nanocrystal of CoO pits from all 6 [100] directions. At the same
1). Finally, via simultaneous pitting and growth, the parent CoO
crystal gradually adopts a concaved cuboctahedron shape, as
illustrated in Figure 10a.
The real growth process will be more complicated than the
simplified case shown in Figure 10a. First, irregular intermediate
morphologies are expected before the particles finally grow into
+
+
time, due to the high Li concentration, Li starts to insert into
2
+
the CoO lattice by exchanging with Co in the {111} planes.
+
Correspondingly, upon one Li insertion/exchange, one Co in
(
19) Ohzuku, T.; Ueda, A.; Nagayama, M. J. Electrochem. Soc. 1993, 140,
(20) Wu, E. J.; Tepesch, P. D.; Ceder, G. Philos. Mag. B 1998, 77, 1039–
1
862–1870.
1047.
2
68 J. AM. CHEM. SOC. 9 VOL. 133, NO. 2, 2011

LiCoO
2
Concaved Cuboctahedrons
A R T I C L E S
2
material in a related synthesis, the final product, LiFeO , which
has a tetragonal lattice, shows a very different morphology for
the final twinned crystal, but again it is determined by the
symmetry relationships between the parent (FeO) and final
product (see Figure S5 in the SI). Although the use of seed
2
1
22
crystals (of Ag and Pt, for example) to template growth on
distinct crystal facets to control morphology has been reported
by others, the growth of cuboctahedrons is distinct from these
growth mechanisms, since the final morphology is determined
by the different symmetry of the template and the final product.
Figure 11. A schematic showing the fast growth on the edge of the LiCoO
plates, which results in the equivalent size of the four twin crystals and the
perfect overall cubic symmetry.
2
The formation of cuboctahedrons is not limited to LiCoO
and Fe, Mn and Ni-doped LiCoO concaved cuboctahedrons
have been observed when using Fe-, Mn-, and Ni-doped CoO/
Co(OH) as starting materials. The yields are lower, and more
2
the regular concaved cuboctahedron shape. An example of this
is shown in Figure 10b, where crystals that are heavily leached
but have not yet achieved a concaved cuboctahedron shape can
be seen. This sample was synthesized by using micrometer-
sized CoO particles as the cobalt source, the particles presum-
ably being too large to be completely converted to cuboctahe-
drons. Second, the parent CoO template crystals do not have to
adopt regular shapes as shown in Figure 10a, the CoO formed
2
2
careful optimization of the synthesis conditions is required for
each specific starting material combination (see Figure S4 in
the SI). The cuboctahedrons themselves are not the most ideal
morphology for a cathode material and do not lead to a high
rate electrode material. One reason for the poor rate performance
may be the pinning of the layers at one end of the crystal: this
means that the twinned particles may not be able to accom-
modate the significant changes in the interlayer spacings that
occur on cycling. We note, however, that the pinning of the
layers should help to prevent the shearing of the oxygen layers
by decomposition of an anisotropic material such as Co(OH)
for example, still leading to a perfectly symmetric cubocahedron.
This can be explained by the preferred fast growth of LiCoO
2
,
2
along [100], [010] and [110] directions and Oswald ripening:
Although the parent CoO crystal can be irregular in shape, once
it forms the prototype of the cuboctahedron, e.g., once it has
23-25
that occurs at high voltage.
We hope that an understanding
all four LiCoO
2
twin variants, no matter how different they are
the formation mechanisms of this kind of twinning and particle
assembly will be very helpful for designing future 3-D nano-
structures and mesostructures, as exemplified by the “desert
in size and shape, the four variants will grow along their own
[
(
100], [010], and [110] directions. If one of the 4 twin variant
plates) of LiCoO cuboctahedron, e.g., V2, as schematically
2
rose” LiCoO
2
morphology, which does show very good rate
drawn in Figure 11, is smaller than any of the other plate(s),
e.g., V1 (for clarity, the other two plates are not shown), then
at the region where the edge of the V2 plate contacts the V1
plate, as indicated by a black arrow in Figure 11 (left hand side),
a concaved shape or v-shape is formed. Addition of ions at this
intersection will be faster, as generally seen for growth at steps
in crystals, due to the availability of higher coordination number
sites on the surface, causing quick growth of a step away from
the intersection of the plates. Fast growth of other parts on the
edge of V2 will occur as the step moves out, away from the
twin boundary, as shown in Figure 11 (right). Meanwhile,
growth at the intersection of the two plates will continue, until
eventually V2 becomes the same size as V1. This effect applies
for any two of the plates. Thus, overall, the higher surface energy
areas at the intersections will vanish leading to four plates with
exactly the same size, as is consistently seen experimentally.
At least three factors appear to be important in the nucleation
of and growth of the cuboctahedrons. First, the ability of the
parent crystal, nanoparticulate CoO to undergo ion-exchange
performance.
Conclusions
A unique morphology for LiCoO
cuboctahedrons has been observed and its formation has been
rationalized in terms of an oxidation-ion-exchange and isotropic
growth mechanism. Near 100% yield was achieved by optimiz-
ing the starting materials and synthesis conditions according to
2
comprising of concaved
2
the proposed growth mechanism. The final product (LiCoO )
maintains its own symmetry (rhombohedral) with a twinned
assembly governed by the symmetry of the precursor (cubic).
The number of twins and their orientations can be determined
by an analysis of the space groups adopted by these two
morphologies and their supergroup, subgroup relationships. The
work not only describes a new topological reaction, but the
results also open up new strategies for designing novel morphol-
ogy of crystals by exploiting the differences in symmetry
between the precursor and the products. This suggests new
routes for tuning the morphology and symmetry of crystals by
systematic crystal engineering. Finally, the ability to grow
physically and electronically connected crystals may in the future
2
+
and oxidation of the Co to form seed nuclei of LiCoO
2
on
the CoO surface, and leaching of select faces to expose stable
3
+
3 4
surfaces. Co O , for example, which contains alternating Co
lead to strategies to improve rate, as shown earlier by the desert
rose form of LiCoO .
2
and Co2 (tetrahedral) and Co octahedral layers, does not form
an ideal template crystal because Li /Co exchange (in the
Co /Co layers) is very difficult, requiring Co mobility and
dissolution. Second, a c/a ratio of LiCoO that is close to that
+
3+
3
+
3+
2+
3+
3+
Acknowledgment. We thank James Quinn, George Hart, and
Dong Su for help with the electron microscopy data collection and
2
of a cubic material, along with the likely ability of this material
to accommodate slight variations in stoichiometry of the
Li CoO phase at the twin boundaries, allows 4 twins with
x 2
rhombohedral symmetry to grow, with minimal stress at the
interconnects of the 4 plates.
Finally, the symmetry of the parent crystal and the phase that
grow on it will control the nature of the morphology of the
final particle, and the number and types of twins formed. For
example, when nanometer sized FeO was used as starting
(
(
(
(
(
21) Fan, F. R.; Ding, Y.; Liu, D. Y.; Tian, Z. Q.; Wang, Z. L. J. Am.
Chem. Soc. 2009, 131, 12036–12037.
22) Habas, S. E.; Lee, H.; Radmilovic, V.; Somorjai, G. A.; Yang, P. Nat.
Mater. 2007, 6, 692–697.
23) Amatucci, G. G.; Tarascon, J. M.; Klein, L. C. J. Electrochem. Soc.
1996, 143, 1114–1123.
24) Van der Ven, A.; Aydinol, M. K.; Ceder, G. J. Electrochem. Soc.
1
998, 145, 2149–2155.
25) Van der Ven, A.; Aydinol, M. K.; Ceder, G.; Kresse, G.; Hafner, J.
Phys. ReV. B 1998, 58, 2975–2987.
J. AM. CHEM. SOC. 9 VOL. 133, NO. 2, 2011 269

A R T I C L E S
Chen et al.
mathematical modeling, Peter Stephens for synchrotron XRD, and
Yuanzhi Tang for DCP analysis. This work was supported by the
Assistant Secretary for Energy Efficiency and Renewable Energy,
Office of FreedomCAR and Vehicle Technologies of the U.S. DOE
under Contract No. DE-AC03-76SF00098 via Subcontract No.
data and relationship between the four twin variants; detailed
analysis of the SAED data. 2. Additional discussion and
experimental support of the growth mechanism: reactions and
equilibria in the system; investigation of the role of Co
situ formation of nano-CoO. 3. Transition metal doped LiCoO
cuboctahedrons. 4. FeO and LiFeO twinned crystals. This
3 4
O ; in
2
6517749 with the Lawrence Berkeley National Laboratory. Research
2
carried out at Brookhaven National Laboratory was supported by
the U.S. DOE under Contract No. DE-AC02-98CH10886.
material is available free of charge via the Internet at http://
pubs.acs.org.
Supporting Information Available: 1. Non-stoichiometric
LiCoO
2
: XRD and refinement for sample A; additional TEM
JA104852Q
2
70 J. AM. CHEM. SOC. 9 VOL. 133, NO. 2, 2011