You are on page 1of 8

Article

pubs.acs.org/crystal

Hydrothermal Synthesis of CoSb2O4: In Situ Powder X‑ray Diffraction,


Crystal Structure, and Electrochemical Properties
Peter Nørby, Martin Roelsgaard, Martin Søndergaard, and Bo B. Iversen*
Department of Chemistry and iNANO, Center for Materials Crystallography, Langelandsgade 140, DK-8000 Aarhus C, Denmark
*
S Supporting Information

ABSTRACT: MSb2O4 constitutes a relatively unexplored class


of multinary oxides that is traditionally synthesized by high-
temperature solid-state methods. Here, we report a facile
synthesis of CoSb2O4 under hydrothermal conditions (T =
135−300 °C, 256 bar). Using in situ synchrotron powder X-ray
diffraction (PXRD), the formation and growth of CoSb2O4
nanoparticles are followed in real time using different precursor
stoichiometries. Phase-pure CoSb2O4 can be formed at 135 °C,
although the formation mechanism changes with precursor
stoichiometry. The crystallite size can be fine-tuned between
14 and 17.5 nm under nonstoichiometric conditions, but crystallites twice as large are found in the stoichiometric case. An
activation energy of 65(12) kJ/mol is obtained for the crystallization from a nonstoichiometric precursor. Modeling of atomic
displacement parameters obtained from Rietveld refinement of multi-temperature high-resolution synchrotron PXRD data gives
a Debye temperature of 331(11) K. The thermal expansion coefficients for the material was found to be αa = 6.2(1) × 10−6 K−1
and αc = 3.1(4) × 10−6 K−1. Electrochemical measurement shows that CoSb2O4 displays a large irreversible capacity (1131 mAh/
g) on the first cycle in Li-ion half-cells and that the capacity decreases significantly in the following cycles.

■ INTRODUCTION
Transition metal antimony(III) oxides MSb2O4 (M: Mn, Fe,
size and purity with this method. In contrast, the low-
temperature hydrothermal synthesis method has proven to be
Co, Ni, Zn) belonging to the schafarzikite mineral group have extremely versatile in the synthesis of many different
been studied significantly less than many other binary oxides oxides.23−25 Here, we show for the first time that phase-pure
(e.g., perovskites).1−21 This is mostly due to the current lack of CoSb2O4 nanoparticles can be prepared via a hydrothermal
technological applications for these materials. It has previously synthesis route. In order to gain size control on the nanometer
been shown that MnSb2O4,5 CoSb2O4,19 FeSb2O4,11,17 and scale, it is important to understand the formation and growth
NiSb2O42,7,11 display antiferromagnetic ordering at low temper- mechanisms taking place during the hydrothermal synthesis. In
ature (highest TN = 79 K). More recently, their applicability as situ studies give the experimentalist unique insight into the
anode materials in Li-batteries (LIBs) has been investigated.20 formation mechanism and growth of crystalline materials
It is well-known that the performance of anode materials in synthesized under hydrothermal conditions.26−32 A range of
LIBs is correlated with materials’ characteristics such as techniques can provide information about hydro- and
nanocrystallite size, crystallinity, and phase purity.22 The Li- solvothermal syntheses under sub- and supercritical con-
ion and electronic conductivities of electrode materials for LIBs ditions.33−41 Here, in situ powder X-ray diffraction (PXRD)
are increased with a reduction in the crystallite’s size, owing to is used to study the formation and growth of nanocrystalline
faster diffusion in the smaller crystallites and access to more CoSb2O4 in real time from both a stoichiometric and
electrolyte.22 Hence, it is of key importance to develop nonstoichiometric precursor. The results are compared with
synthesis methods that allow full control of both the crystallite
those obtained from an ex situ hydrothermal synthesis
size and phase purity.
performed in an autoclave. In addition, the thermal expansion,
So far, MSb2O4 materials have been mainly synthesized by a
traditional solid-state synthesis method, where the elements are atomic displacement parameters, and Debye temperature are
mixed and heated in a sealed ampule (700 °C for 12 h).17,19 obtained from Rietveld refinement of high-resolution multi-
Westin and Nygren investigated the decomposition of M2+- temperature synchrotron PXRD data. Finally, the morphology
Sb3−-alkoxide gels to both MSb2O6 and MSb2O4. The and electrochemical properties of microcrystalline CoSb2O4
formation of a gel is similar to the first step in a hydrothermal were investigated.
synthesis, although the temperature used in the decomposition
of the gel to crystalline MSb2O4 is 620 °C,15 which is much Received: October 5, 2015
higher than typical hydrothermal synthesis temperatures (T > Revised: December 22, 2015
100 °C). Furthermore, it is difficult to control the crystallite’s Published: January 5, 2016

© 2016 American Chemical Society 834 DOI: 10.1021/acs.cgd.5b01421


Cryst. Growth Des. 2016, 16, 834−841
Crystal Growth & Design Article

Table 1. Synthesis Conditions for in Situ Data Collection at p = 256 bar


compound M(Sb)/M(Co) 135 °C 150 °C 180 °C 220 °C 300 °C
CoSb2O4 1.33 × × × ×
2.00 × × ×

■ EXPERIMENTAL SECTION
Precursor Synthesis. SbCl3 (≥99.0%, Sigma-Aldrich), Co-
charged with a constant current of 30 mA/g between 0.05 and 1.3 V vs
Li/Li+.
Data Integration and Analysis. Prior to integration in Fit2D,44
(CH3COO)2·4H2O (≥98.0%, Sigma-Aldrich), NaOH, and HCl were the 2D diffraction patterns from the in situ experiments were masked
used as purchased. For the in situ study, solutions of SbCl3 in 4 M HCl in order to remove single-crystalline diffraction peaks from sapphire
and Co(CH3COO)2·4H2O in deionized water were utilized. The pH and the beam stop. The integrated in situ PXRD patterns were
value of the mixture was adjusted by addition of NaOH. Upon analyzed in the FullProf program package45 by sequential Rietveld
addition of NaOH, the solution changed color from light red to dark refinement. The volume-weighted crystallite size, scale factor, and
blue. In a typical synthesis, an aqueous 2.5 mL 1.00 M Co- lattice parameters were extracted from each frame in the refinement.
(CH3COO)2·4H2O solution was mixed with 2.5 mL of a 1.33 M SbCl3 The crystallite sizes were calculated based on the Scherrer equation
solution; subsequently, 3.0 mL of 8 M NaOH was added. The molar using the line profile parameters corrected for instrumental broad-
ratio between Sb and Co was modified to either 1.33 or 2.00 (Table 1) ening.46 The high-resolution multi-temperature PXRD patterns
in order to study the influence of excess Co on the formation and measured on the ex situ synthesized sample were analyzed by Rietveld
growth of the nanocrystallites. refinement in the FullProf program package.45 Details on the Rietveld
In Situ Experiments. A custom-made sapphire capillary setup was refinements can be found in the Supporting Information.


used as described in detail by Becker et al.42 The experiments were
carried out at beamline I711 at MAX-LAB (Lund, Sweden).43 The RESULTS AND DISCUSSION
precursor was injected into a thin single-crystalline sapphire capillary
(inner diameter 0.7 mm), which was subsequently pressurized at 256 Precursor and Phase Identification. The contour plot of
bar. Heating was initiated by switching a valve directing a jet of the in situ PXRD data shown in Figure 1 and the Supporting
preheated hot air onto the sapphire capillary (see Table 1 for applied
temperatures). The desired temperature can be reached within
approximately 20 s (see Supporting Information, Figure S1), and
time-resolved X-ray patterns with a time resolution of 5 s were
recorded prior to and during the heating. Each experiment was allowed
to run for ∼20 min. The wavelengths, sample-to-detector distances,
and instrumental contribution to the peak broadening were calibrated
with a NIST LaB6 standard for each experiment (the wavelength was λ
= 0.992 Å). The scattered X-rays were detected by an Oxford
Diffraction Titan CCD detector. All PXRD data are presented in Q-
space (Q = (4π sin θ)/λ).
Ex Situ Experiments. A scaled-up synthesis was performed by
mixing 10 mL of a 0.667 M Co(CH3COO)2·4H2O aqueous solution
with 10 mL of 1.334 M SbCl3 (dissolved in 4 M HCl). Under
magnetic stirring, 16 mL of 8.0 M NaOH was added. The autoclave
was filled to half the volume with the precursor (total volume: 70 mL).
The precursor was heated at 220 °C for 53 h. High-resolution PXRD
patterns on the powdered sample were measured at beamline BL44B2,
SPring-8 (Japan), from 100 to 300 K in steps of 50 K. A homogeneous
sample for the PXRD measurements was obtained by floating two
times in ethanol (96%), which was subsequently packed into a 0.1 mm
glass capillary. The scattered X-rays were collected on an image plate
detector mounted in a Debye−Scherrer camera covering an angular 2θ
range of 2−77°, resulting in a Q-range from 0.44 to 15.6 Å−1. The
sample was further characterized by scanning electron microscopy
(SEM) on a FEI Nova NanoSEM 600 and by transmission electron
microscopy (TEM) on a Phillips CM20 with an acceleration voltage of Figure 1. Contour plot of raw PXRD in situ data for the nanocrystals
200 kV. SEM preparation was done by placing the powder on synthesized at 135 °C and with (a) M(Sb)/M(Co) = 1.33 and (b)
conducting carbon tape. TEM samples were prepared by making a M(Sb)/M(Co) = 2.00. Asterisks indicate CoSb2O4.
diluted suspension of the particles in ethanol and placing a droplet on
a carbon-coated copper grid.
Li-Battery Assembly and Measurements. Half-cells of type Information (Figures S2−S9) reveals an almost instantaneous
CR2032 were assembled with Li metal foil as both counter and formation of phase-pure CoSb2O4 under hydrothermal
reference electrodes. The scaled-up product was used as the active conditions when heat is turned toward the sample. CoSb2O4
material in the working electrode, and it was mixed with Super P can be synthesized at the lowest investigated temperature (135
carbon as the conductor and poly(vinylidene difluoride) (PVDF) as °C) for both stoichiometries. The rapid formation indicates
the binder in a weight ratio of 76:12:12 and coated onto Cu-foil. The that the synthesis temperature can be reduced even further.
typical loading of a dry electrode was ∼30 μm or ∼2 mg/cm2. The
After 1000 s, the sample is phase-pure independent of the ratio
working electrodes were compressed with ∼50 kN/cm2. Two pieces of
25 μm thick porous polypropylene membranes were employed as of cobalt to antimony in the precursor. Surprisingly, there is an
separators, and the electrolyte solution was composed of 1 M LiPF6 in unidentified impurity in the synthesis performed at 180 °C
ethylene carbonate (EC) and dimethyl carbonate (DMC) in a ratio of under stoichiometric (M(Sb)/M(Co) = 2.00) conditions. It is
1:1 by volume. The cells were assembled in a glovebox with water and not present when the synthesis temperature is increased or
oxygen levels at ∼1 ppm. After assembly, the cells were discharged and decreased.
835 DOI: 10.1021/acs.cgd.5b01421
Cryst. Growth Des. 2016, 16, 834−841
Crystal Growth & Design Article

The formation mechanism differs under the nonstoichio-


metric (M(Sb)/M(Co) = 1.33) and stoichiometric (M(Sb)/
M(Co) = 2.00) synthesis conditions (Figure 2). The

Figure 3. Crystallite size for CoSb2O4 nanoparticles (Sb/Co = 1.33) as


a function of time. The uncertainties are smaller than the marks. The
inset show the growth rate as a function of time.

Increasing the synthesis temperature to 220 °C results in fast


initial growth, and the final crystallite size (17.5 nm) is reached
within 150 s after the heat has been applied. It should be noted
that the estimated standard deviation in Figure 3 is smaller than
Figure 2. PXRD patterns at three different times for the reaction at the symbols.
135 °C with a Sb to Co ration of (a) 1.33 and (b) 2.00. (c) Standard
CoSb2O4 PXRD pattern. The precursor is amorphous for Sb/Co =
The Johnson−Mehl−Avrami equation47,48 is utilized here to
1.33 and crystalline for Sb/Co = 2.00. model the growth of nanocrystalline CoSb2O4 from an
amorphous precursor. Although the model was initially
developed for solid-state reactions, it has proven to give
stoichiometric precursor contains both crystalline orthorhom- valuable information from experiments performed under hydro-
bic Sb2O3 (Pccn) and nanocrystalline cubic Sb2O3 (Fd3̅m) (see and solvothermal conditions.49−52 Both the reaction mecha-
the Supporting Information, Figure S8). Switching the heat nism and activation energy are extractable parameters from the
toward the sample dissolves the crystalline precursor over a model. The extent of the reaction (f) is plotted as a function of
time period of approximately 100 s at 135 °C. The intensity of time after the first appearance of the first nanocrystallites (t −
the reflections of CoSb2O4 increases at the expense of a t0). The data are subsequently modeled by the equation f = 1 −
decrease in intensity of the reflections of Sb2O3. Increasing exp[−k(t − t0)n], where the extent of the reaction is defined as
synthesis temperature results in a faster dissolution of the V(t)/Vinf (Vinf is equal to the final stable nanocrystal volume at
crystalline precursor, and even at 180 °C the transformation the specific synthesis temperature). It is assumed that the
from Sb2O3 to CoSb2O4 is faster than the resolution time (5 s) crystallites are spherical in morphology, which is supported by
(see the Supporting Information). The in situ results show that the modeling of the peak shape (see Supporting Information).
in the nonstoichiometric case the precursor is amorphous, The two parameters in the equation are related to the rate
which indicates that the formation of Sb2O3 in the precursor is constant (k) and to the mechanism (n). If the value of n is in
suppressed when the relative concentration of antimony is the range from 0.54 to 0.62, then this typically corresponds to a
lowered. Since CoSb2O4 in this case is formed from an diffusion-controlled reaction mechanism, 1.0−1.24, to a zero-
amorphous precursor, it crystallizes instantaneously when heat order, first-order, or phase-boundary-controlled mechanism,
is applied. This is seen as a sharp transition in Figure 1. This and 2.0−3.0, to a nucleation- and growth-controlled mecha-
indicates a strong dependence on reactive antimony species in nism.53
solution in order to form CoSb2O4. Hence, it is advantageous Figure 4a−d shows t − t0 as a function of the extent of the
to have an amorphous precursor where the antimony atoms reaction with fits for the four different synthesis temperatures.
seem to be more loosely bonded. It can be rationalized that the The rate constant (k) and mechanism-dependent parameter
mechanism most likely is a dissolution recrystallization (n) are obtained from the fits. It can be seen that the value of n
mechanism where CoSb2O4 is formed as more antimony is decreases with temperature and varies from 0.9 to 0.4; however,
released to the solution, which in the stoichiometric and all four data points fall outside of physically meaningful ranges,
nonstoichiometric cases is from crystalline Sb2O3 and as indicated with gray in Figure 4e. It can be concluded that the
amorphous antimony material, respectively. reaction mechanism presumably changes from something that
Size Evolution and Growth Kinetics. The crystallite size is close to either zero- or first-order kinetics to a more diffusion-
is calculated from the sequential Rietveld refinement data controlled reaction mechanism with increasing temperature,
corrected for instrumental broadening. It is extracted as a albeit the specific mechanism is uncertain. It could also be
function of synthesis temperature for the nonstoichiometric caused by different reactions occurring in parallel, and further
precursor as shown in Figure 3. It should be noted that in a total scattering experiments may clarify the formation
diffraction experiment the volume-averaged crystallite size value mechanism from amorphous to crystalline CoSb2O4.
is obtained, which is inversely proportional to the broadening The activation energy (Ea) can be calculated by the
of the Bragg peaks. The inset in Figure 3 shows that the overall Arrhenius equation k = A exp(Ea/RT), and it is obtained as
growth rate of CoSb2O4 nanoparticles is almost independent of the slope when ln(k) vs 1000/T is plotted (Figure 4f). An
the synthesis temperature, although slight differences are seen activation energy of 65(12) kJ/mol is obtained for the
in how fast zero growth rate is achieved. At 135 °C, the crystallization of CoSb2O4 from an amorphous precursor.
crystallite size stabilizes at 14 nm after approximately 300 s. This is in good agreement with activation energies obtained in
836 DOI: 10.1021/acs.cgd.5b01421
Cryst. Growth Des. 2016, 16, 834−841
Crystal Growth & Design Article

Figure 5. Changes in the unit cell parameters as a function of


crystallite size. The color of the markers indicates a different synthesis
temperature (red, 135 °C; blue, 150 °C; purple, 180 °C; and black,
220 °C).

with decreasing crystallite size has been ascribed to the


softening of lattice vibration.59 In the present study, it is not
possible to distinguish between the mechanisms involved in the
Figure 4. (a−d) Johnson−Mehl−Avrami plots for syntheses at (a)
softening of the lattice vibration because reliable values of the
135, (b) 150, (c) 180, and (d) 220 °C in the nonstoichiometric case. correlated occupancy, atomic thermal displacements, and
(e) n values obtained from the fits at different synthesis temperatures atomic positions cannot be extracted due to insufficient data
with gray areas indicating physically meaningful n values. (f) Arrhenius quality.
plot. The in situ PXRD data show that changing the ratio between
antimony and cobalt from nonstoichiometric to stoichiometric
other oxide systems synthesized under hydrothermal con- affects the formation mechanism, as shown in Figures 1 and 2.
ditions. For the formation of Fe2O3, TiO2, and α-Li2TiO3 Analysis of the stoichiometric data shows that the crystallite size
nanocrystallites, activation energies of 67(15), 66(19), and is more than doubled when the ratio is stoichiometric
66(7) kJ/mol were estimated, respectively.49,51,54 Rossetti et al., compared with that for the nonstoichiometric case (Figure
by a pseudo in situ method, found an activation energy of the 6). Furthermore, the growth is significantly different when the
apparent crystallization of PbTiO3 of 30 kJ/mol;52 however, the
initial formation was omitted in this case. The activation energy
for a phase transition in nanocrystals is a larger by a factor of 2,
e.g., the phase transition from β-chalcocite Cu2S to high
digenite Cu2−xS has an activation energy of 171(27) kJ/mol.50
Hence, the similarity found for the activation energy for the
oxides synthesized under analogous conditions is interesting,
but on the basis solely of these values, a conclusion for the
mechanism governing the formation cannot be drawn. Recent
total X-ray scattering studies of the hydrothermal formation of
nanoparticles have shown that the formation mechanism for
relatively simple oxides can be very complex: going from a
polymeric precursor to a highly disordered amorphous Figure 6. Crystallite size for CoSb2O4 nanoparticles (M(Sb)/M(Co)
precipitate before crystallizing in the desired phase.55−57 = 2.00).
The changes in unit cell parameters obtained from the
sequential Rietveld refinement are plotted as a function of
crystallite size (Figure 5). Only changes in unit cell parameters ratio is changed, as seen when comparing Figure 3 with Figure
(a − afinal, c − cfinal, and V − Vfinal) are considered, owing to the 6. In the stoichiometric case, the final crystallite size at 135 °C
lack of an internal standard in the in situ experiments. It can be is reached within the first 150 s; subsequently, the crystallite
seen that both the a and c axes decrease as the crystallite grows. size stabilizes around 34 nm. Upon an increase in the synthesis
The change in unit cell parameters is anisotropic, as the temperature, the initial crystallite size of approximately 35 nm
decrease is larger along the c axis than along the a axis. This increases to 44 and 52 nm at 180 and 300 °C, respectively.
anisotropy is also seen in the formation of SnO2 nano- Hence, the ratio of antimony to cobalt influences the final
particles.58 Furthermore, there is a clear enlargement of the unit crystallite size. Furthermore, there is both a temperature and
cell when the nanocrystallites are below 7 nm. At larger compositional dependence for the nanoparticles to grow by
crystallite sizes, the unit cell has almost relaxed to its final value, Ostwald ripening (see the Supporting Information, Figure
which indicates that the internal atomic structure changes as a S10).58 In the stoichiometric case at 135 °C, no Ostwald
function of crystallite size. The origin of this unit cell expansion ripening is observed since the scale factor can be superimposed
837 DOI: 10.1021/acs.cgd.5b01421
Cryst. Growth Des. 2016, 16, 834−841
Crystal Growth & Design Article

with the crystallite volume. Hence, the nanoparticles grow large size distribution, with the smallest and largest being
because more CoSb2O4 can be formed from the precursor in around 80 and 300 nm in one dimension, respectively. No
solution; thus, the growth and scale factor stagnate when the micrometer-sized particles were found by SEM (see the
precursor is depleted. When increasing the temperature to 180 Supporting Information, Figure S15).
°C or changing to nonstoichiometric conditions, Ostwald Rietveld refinements similar to those depicted in Figure 7
ripening is observed (see the Supporting Information, Figure were performed for the five different temperatures on the high-
S10). resolution PXRD patterns, ignoring the impurity phase (see the
Ex Situ Autoclave Study. The scaled-up autoclave- Supporting Information, Figures S12 and S13). From these
synthesized ex situ product has a small impurity present, as multi-temperature Rietveld refinements, different physical
seen by the Rietveld refinement in Figure 7. It is speculated that parameters can be extracted and evaluated as a function of
temperature, e.g., unit cell parameters. Figure 9 shows that

Figure 7. Rietveld refinement of high-resolution PXRD data measured Figure 9. Thermal evolution of the unit cell as extracted from multi-
at 100 K. It is clear from the difference curve that an impurity is temperature Rietveld refinement. The linear fits show the region where
present in the measured sample. RBragg = 4.50; RF = 2.41. the thermal expansion coefficients are calculated.

the prolonged synthesis time (53 h) might have induced


decomposition of the phase-pure CoSb2O4 since, as we have there is a linear thermal expansion of the unit cell, although the
just shown, CoSb2O4 can be synthesized in seconds from a c parameter displays a deviation from linearity when compared
similar starting solution. It should be noted that there are small with the a parameter. Similarly, the unit cell’s thermal
differences between the two methods, especially the heating expansion also shows this nonlinearity. The linear thermal
rate and pressure. It has not been possible to identify the expansion coefficients with respect to 300 K are 6.2(1) × 10−6,
impurity phase. The impurity might also explain the light blue 3.1(4) × 10−6, and 1.55(7) × 10−5 K−1 in the a and c directions
or greyish color of the material, which is in contrast with the and for the unit cell, respectively. Comparing the absolute
previous reported pinkish brown.3,19 values at 300 K obtained in this study of the a axis (8.5031(1)
Figure 8 display a TEM picture of the sample from the Å) and c axis (5.9288(1) Å) with those reported by Koyama et
autoclave synthesis, and it can be seen that at long synthesis al. (a = 8.500(1) Å and c = 5.931(1) Å)3 and de Laune et al. (a
times the particles grow into rods and cubes. The largest rods = 8.49340(9) Å and c = 5.92387(8) Å)19 reveals that the values
are around 400 nm in length and 100 nm in diameter, although are slightly larger in this study. This discrepancy might be
much smaller rods are also observable. The cubes have a similar explained by the difference in the synthesis methods used in
this study and the previous ones; although the absolute values
differ, the relative changes investigated in this study should still
be valid. None of the previously reported studies investigated
the thermal expansion of CoSb2O4. Comparing the thermal
expansion coefficients with those for MnSb2O4 shows that the
result found in this study is slightly lower but is of the same
order of magnitude.10
The atomic displacement parameters (ADPs) for all atoms
are shown in Figure 10. The best model is achieved when
antimony is allowed to vibrate anisotropically. Independent of
temperature, the oxygen atoms have the smallest ADP
(constrained to be identical for the two sites), whereas cobalt
display the largest ADP. The isotropic equivalent ADP value for
antimony is in between these two values. This is rather
surprising since de Laune et al. observed by neutron powder
diffraction that antimony has a larger ADP than cobalt.19 The
absolute values at 300 K are also slightly larger in the study
presented here. It should be noted that antimony is surrounded
by four O2− anions. The u11 for antimony is twice as large as all
Figure 8. TEM picture of an autoclave synthesis sample used as the other ADP elements, and upon extrapolation to 0 K, it
anode material in Li-ion half-cells. approaches a finite value, which could indicate that static
838 DOI: 10.1021/acs.cgd.5b01421
Cryst. Growth Des. 2016, 16, 834−841
Crystal Growth & Design Article

Electrochemical Measurements. Half-cells of type


CR2032 were measured in order to investigate the potential
of CoSb2O4 for use as an anode in LIBs. The galvanostatic
cycling of CoSb2O4 can be seen in Figure 12. The initial

Figure 10. Atomic displacement parameters (ADPs) for (a) isotropic


(Co and O) or equivalent (Sb) and (b) anisotropic for antimony as a
function of temperature. Figure 12. Potential (vs Li/Li+) vs capacity for CoSb2O4 under
galvanostatic charge/discharge at 30 mA/g between 0.05 and 1.3 V for
disorder contributes to u11. This could be a consequence of the the second, third, and tenth cycles. The inset shows the initial
lone electron pair present in antimony, but detailed analysis is discharge and charge curves.
beyond the scope of this article.60 Furthermore, there is a
decrease in electron density (void) in the center of the unit cell. discharge reached 1131 mAh/g, corresponding to ∼15 Li atoms
However, the long axis of the ellipsoid representing the reacting with CoSb2O4. This is comparable to the 1127 mAh/g
anisotropic vibration is in the ab plane, as shown in Figure 11. reported by Jibin et al.20 and concurs with the formation of
Hence, it is not in the direction of the center. Li2O and Li3Sb leaving one Li, e.g., for formation of the SEI
layer. The initial charge capacity was 417 mAh/g, which
corresponds to dealloying of slightly below two formula units of
Li3Sb. Furthermore, it should be noted that the unreacted
cobalt in the half-cells is electrochemically inactive. Future
studies might reveal how cobalt is bonded in the decomposed
compound and why this unusual inactivity is seen. On
subsequent cycling, the capacity fade was relatively fast, with
only ∼20 mAh/g reversible capacity remaining after the tenth
cycle.

■ CONCLUSIONS
We have demonstrated by in situ PXRD studies that phase-pure
CoSb2O4 nanoparticles can be synthesized under hydrothermal
Figure 11. Illustration of the unit cell with thermal ellipsoids shown conditions at temperatures ranging from 135 to 300 °C. The
for antimony and octahedra for the cobalt atoms. The long axes for the dependence of the starting solution on stoichiometry was
Sb ellipsoids point either in the a or b direction. investigated, and it was found that with the nonstoichiometric
precursor the nanoparticle sizes range from 14.5 to 17 nm. In
the case of a stoichiometric precursor, the sizes are twice as
large. The formation mechanism is different in the two cases, as
The lattice dynamics of the structure can be analyzed from it was found that the nonstoichiometric precursor is
the ADPs, and here we approximate that the motion of the amorphous, whereas the stoichiometric precursor is crystalline.
atoms can be represented by a Debye model: The unit cell shows a clear enlargement for nanoparticles below
7 nm. Fitting growth curves with a Johnson−Mehl−Avrami
3ℏ2T ⎡ T θD/ T x θ ⎤
Uiso(T ) = ⎢
mkBθD2 ⎣ θD
∫0 exp(x) − 1
dx + D ⎥
4T ⎦
model yields an activation of 65(12) kJ/mol for the
crystallization of CoSb2O4 from a nonstoichiometric precursor.
A scaled-up ex situ autoclave synthesis gave submicrometer
+ d2 CoSb2O4 particles, and Rietveld refinements of multi-temper-
Fitting the ADPs to this formula provides an estimate of the ature high-resolution PXRD patterns reveal that the a axis
Debye temperature (see the Supporting Information, page 12, expands linearly with temperature, whereas the c axis deviates
for details), which is also related to the average sound velocity slightly from linearity. The following linear thermal expansion
in the material. If an average atomic mass of 52.3 amu is used, coefficients with respect to 300 K were obtained: αa = 6.2(1) ×
then a Debye temperature of 331(11) K is obtained for 10−6 K−1 and αc = 3.1(4) × 10−6 K−1. Modeling the ADPs gives
CoSb2O4. If the ADPs of the individual atoms are fitted to the a Debye temperature of 331(11) K for CoSb2O4. Its potential
Debye formula, then Debye temperature for the Co, Sb, and O for use as the anode material in LIBs is not promising, as the
sublattices are 255(9), 208(2), 656(57) K, respectively. material exhibits a large irreversible capacity on the first cycle
839 DOI: 10.1021/acs.cgd.5b01421
Cryst. Growth Des. 2016, 16, 834−841
Crystal Growth & Design Article

and the reversible capacity fades substantially on the following (14) Gavarri, J. R.; Chater, R.; Ziółkowski, J. J. Solid State Chem.
cycles, resulting in a capacity of ∼20 mAh/g after the tenth 1988, 73, 305−316.
cycle. Further material improvements have to be conducted (15) Westin, G.; Nygren, M. J. Mater. Sci. 1992, 27, 1617−1625.
before CoSb2O4 can be used in battery applications. (16) Abakumov, A. M.; Rozova, M. G.; Antipov, E. V.; Hadermann,


J.; Van Tendeloo, G.; Lobanov, M. V.; Greenblatt, M.; Croft, M.;
Tsiper, E. V.; Llobet, A.; Lokshin, K. A.; Zhao, Y. Chem. Mater. 2005,
ASSOCIATED CONTENT 17, 1123−1134.
* Supporting Information
S (17) Whitaker, M. J.; Bayliss, R. D.; Berry, F. J.; Greaves, C. J. Mater.
The Supporting Information is available free of charge on the Chem. 2011, 21, 14523−14529.
ACS Publications website at DOI: 10.1021/acs.cgd.5b01421. (18) Bayliss, R. D.; Berry, F. J.; de Laune, B. P.; Greaves, C.;
Helgason, Ö .; Marco, J. F.; Thomas, M. F.; Vergara, L.; Whitaker, M. J.
Details of in situ data analysis, sequential Rietveld
J. Phys.: Condens. Matter 2012, 24, 276001.
refinement, and how the crystallite size is calculated from (19) de Laune, B. P.; Greaves, C. J. Solid State Chem. 2012, 187,
spherical harmonics; contour plots for all in situ PXRD 225−230.
data along with verification of precursor composition and (20) Jibin, A. K.; Reddy, M. V.; Subba Rao, G. V.; Varadaraju, U. V.;
Ostwald ripening; temperature profile for the in situ Chowdari, B. V. R. Electrochim. Acta 2012, 71, 227−232.
setup and examples of Rietveld refinements for two (21) Iyama, A.; Wakabayashi, Y.; Hanasaki, N.; Kimura, T. Jpn. J.
frames; details of ex situ data analysis and Debye Appl. Phys. 2014, 53, 05FB02.
temperature calculations together with representation of (22) Bruce, P. G.; Scrosati, B.; Tarascon, J.-M. Angew. Chem., Int. Ed.
the whole data range for the ex situ multitemperature 2008, 47, 2930−2946.
high-resolution data; and SEM picture of the obtained (23) Walton, R. I. Chem. Soc. Rev. 2002, 31, 230−238.
(24) Byrappa, K.; Adschiri, T. Prog. Cryst. Growth Charact. Mater.
particles synthesized ex situ (PDF)


2007, 53, 117−166.
(25) Yoshimura, M.; Byrappa, K. J. Mater. Sci. 2008, 43, 2085−2103.
AUTHOR INFORMATION (26) Michailovski, A.; Grunwaldt, J.-D.; Baiker, A.; Kiebach, R.;
Corresponding Author Bensch, W.; Patzke, G. R. Angew. Chem., Int. Ed. 2005, 44, 5643−5647.
(27) Shen, X.-F.; Ding, Y.-S.; Hanson, J. C.; Aindow, M.; Suib, S. L. J.
*E-mail: bo@chem.au.dk.
Am. Chem. Soc. 2006, 128, 4570−4571.
Funding (28) Mitchell, S.; Biswick, T.; Jones, W.; Williams, G.; O’Hare, D.
The work was supported by the Danish National Research Green Chem. 2007, 9, 373−378.
Foundation (Center for Materials Crystallography, DNRF93), (29) Cheong, S.; Watt, J.; Ingham, B.; Toney, M. F.; Tilley, R. D. J.
the Danish Research Council for Nature and Universe Am. Chem. Soc. 2009, 131, 14590−14595.
(Danscatt), and the European Community’s Seventh Frame- (30) Pienack, N.; Bensch, W. Angew. Chem., Int. Ed. 2011, 50, 2014−
work Programme (FP7/2007-2013) CALIPSO under Grant 2034.
Agreement No. 312284 (31) Ok, K. M.; Lee, D. W.; Smith, R. I.; O’Hare, D. J. Am. Chem. Soc.
2012, 134, 17889−17891.
Notes (32) Andersen, H. L.; Christensen, M. Nanoscale 2015, 7, 3481−
The authors declare no competing financial interest. 3490.

■ ACKNOWLEDGMENTS
Beamline I711 at MAX-LAB and RIKEN beamline BL4402 at
(33) Bremholm, M.; Becker-Christensen, J.; Iversen, B. B. Adv. Mater.
2009, 21, 3572−3575.
(34) Bremholm, M.; Felicissimo, M.; Iversen, B. B. Angew. Chem., Int.
Ed. 2009, 48, 4788−4791.
SPring-8 are gratefully acknowledged for beam time. Henrik L. (35) Tyrsted, C.; Becker, J.; Hald, P.; Bremholm, M.; Pedersen, J. S.;
Andersen, Dr. Espen D. Bøjesen, and Dr. Mogens Christensen Chevallier, J.; Cerenius, Y.; Iversen, S. B.; Iversen, B. B. Chem. Mater.
are thanked for assistance during the experiments at MAX-LAB.


2010, 22, 1814−1820.
(36) Tyrsted, C.; Jensen, K. M. Ø.; Bøjesen, E. D.; Lock, N.;
REFERENCES Christensen, M.; Billinge, S. J. L.; Iversen, B. B. Angew. Chem., Int. Ed.
(1) Gonzalo, J. A.; Cox, D. E.; Shirane, G. Phys. Rev. 1966, 147, 415− 2012, 51, 9030−9033.
418. (37) Nørby, P.; Jensen, K. M. Ø.; Lock, N.; Christensen, M.; Iversen,
(2) Witteveen, H. T. Solid State Commun. 1971, 9, 1313−1315. B. B. RSC Adv. 2013, 3, 15368−15374.
(3) Koyama, E.; Nakai, I.; Nagashima, K. Nippon Kagaku Kaishi 1979, (38) Bøjesen, E. D.; Jensen, K. M. Ø.; Tyrsted, C.; Lock, N.;
793−795. Christensen, M.; Iversen, B. B. Cryst. Growth Des. 2014, 14, 2803−
(4) Gavarri, J. R. J. Solid State Chem. 1982, 43, 12−28. 2810.
(5) Gavarri, J. R.; Calvarin, G.; Chardon, B. J. Solid State Chem. 1983, (39) Jensen, K. M. Ø.; Tyrsted, C.; Bremholm, M.; Iversen, B. B.
47, 132−142. ChemSusChem 2014, 7, 1594−1611.
(6) Gavarri, J. R.; Hewat, A. W. J. Solid State Chem. 1983, 49, 14−19. (40) Birgisson, S.; Jensen, K. M. Ø.; Christiansen, T. L.; von Bülow, J.
(7) Chater, R.; Chhor, K.; Gavarri, J. R.; Pommier, C. Mater. Res. Bull. F.; Iversen, B. B. Dalton Trans. 2014, 43, 15075−15084.
1985, 20, 1427−1434. (41) Mi, J.-L.; Shen, Y.; Becker, J.; Bremholm, M.; Iversen, B. B. J.
(8) Chater, R.; Gavarri, J. R. J. Solid State Chem. 1985, 59, 123−131. Phys. Chem. C 2014, 118, 11104−11110.
(9) Chater, R.; Gavarri, J. R.; Hewat, A. W. J. Solid State Chem. 1985, (42) Becker, J.; Bremholm, M.; Tyrsted, C.; Pauw, B.; Jensen, K. M.
60, 78−86. Ø.; Eltzholt, J.; Christensen, M.; Iversen, B. B. J. Appl. Crystallogr.
(10) Fjellvåg, H.; Kjekshus, A.; Leskelä, T.; Leskelä, M.; Hoyer, E. 2010, 43, 729−736.
Acta Chem. Scand. 1985, 39a, 389−395. (43) Cerenius, Y.; Ståhl, K.; Svensson, L. A.; Ursby, T.; Oskarsson,
(11) Chater, R.; Chhor, K.; Gavarri, J. R.; Pommier, C. Mater. Res. A.; Albertsson, J.; Liljas, A. J. Synchrotron Radiat. 2000, 7, 203−208.
Bull. 1986, 21, 703−708. (44) Hammersley, A. P.; Svensson, S. O.; Hanfland, M.; Fitch, A. N.;
(12) Chater, R.; Gavarri, J. R.; Genet, F. J. Solid State Chem. 1986, 63, Häusermann, D. High Pressure Res. 1996, 14, 235−248.
295−307. (45) Rodríguez-Carvajal, J. Phys. B 1993, 192, 55−69.
(13) Chater, R.; Gavarri, J. R.; Hewat, A. W. J. Solid State Chem. (46) Patterson, A. L. Phys. Rev. 1939, 56, 978−982.
1987, 67, 98−103. (47) Avrami, M. J. Chem. Phys. 1939, 7, 1103−1112.

840 DOI: 10.1021/acs.cgd.5b01421


Cryst. Growth Des. 2016, 16, 834−841
Crystal Growth & Design Article

(48) Johnson, W. A.; Mehl, R. F. Trans. Am. Inst. Min. Metall. Eng.
1939, 135, 416−442.
(49) Eltzholtz, J. R.; Tyrsted, C.; Jensen, K. M. Ø.; Bremholm, M.;
Christensen, M.; Becker-Christensen, J.; Iversen, B. B. Nanoscale 2013,
5, 2372−2378.
(50) Nørby, P.; Johnsen, S.; Iversen, B. B. ACS Nano 2014, 8, 4295−
4303.
(51) Andersen, H. L.; Jensen, K. M. Ø.; Tyrsted, C.; Bøjesen, E. D.;
Christensen, M. Cryst. Growth Des. 2014, 14, 1307−1313.
(52) Rossetti, G. A.; Watson, D. J.; Newnham, R. E.; Adair, J. H. J.
Cryst. Growth 1992, 116, 251−259.
(53) Hancock, J. D.; Sharp, J. H. J. Am. Ceram. Soc. 1972, 55, 74−77.
(54) Laumann, A.; Jensen, K. M. Ø.; Tyrsted, C.; Bremholm, M.;
Fehr, K. T.; Holzapfel, M.; Iversen, B. B. Eur. J. Inorg. Chem. 2011,
2011, 2221−2226.
(55) Tyrsted, C.; Lock, N.; Jensen, K. M. Ø.; Christensen, M.;
Bøjesen, E. D.; Emerich, H.; Vaughan, G.; Billinge, S. J. L.; Iversen, B.
B. IUCrJ 2014, 1, 165−171.
(56) Jensen, K. M. Ø.; Andersen, H. L.; Tyrsted, C.; Bøjesen, E. D.;
Dippel, A.-C.; Lock, N.; Billinge, S. J. L.; Iversen, B. B.; Christensen,
M. ACS Nano 2014, 8, 10704−10714.
(57) Saha, D.; Jensen, K. M. Ø.; Tyrsted, C.; Bøjesen, E. D.;
Mamakhel, A. H.; Dippel, A.-C.; Christensen, M.; Iversen, B. B. Angew.
Chem., Int. Ed. 2014, 53, 3667−3670.
(58) Jensen, K. M. Ø.; Christensen, M.; Juhás, P.; Tyrsted, C.;
Bøjesen, E. D.; Lock, N.; Billinge, S. J. L.; Iversen, B. B. J. Am. Chem.
Soc. 2012, 134, 6785−6792.
(59) Fukuhara, M. Phys. Lett. A 2003, 313, 427−430.
(60) Curti, M.; Gesing, T. M.; Murshed, M. M.; Bredow, T.;
Mendive, C. B. Z. Kristallogr. - Cryst. Mater. 2013, 228, 629−634.

841 DOI: 10.1021/acs.cgd.5b01421


Cryst. Growth Des. 2016, 16, 834−841

You might also like