Abstract
The specific capacity and energy of batteries with
electrodes can be improved by the addition of
to the cathodes, particularly at high rates.
is a standard active material used in Li-ion batteries, and
is a highly stable Li-insertion compound with high electronic and Li-ion conductivity. Batteries with
and
cathodes are compared to those with blends of the two active materials when discharged from
at C/5 and 2C rates. The Li-ion batteries with a
blend have a disproportionate increase of 25 and 36%, respectively, in their expected specific capacity and energy relative to
when the batteries are charged and discharged at a 2C rate. X-ray diffraction of the electrodes at a high state of charge confirms that the
and
phases remain distinct in the electrodes. Analysis of the discharge curves in combination with electrochemical impedance spectroscopy shows that
lowers the electrical resistance of the electrodes when combined in parallel with the
. This simple approach may be used to improve the capacity of Li-ion batteries for high rate applications.
Many commercial Li-ion batteries use
or a partially substituted variation as the insertion material at the positive electrode. These layered structures have chemical and electrochemical stability,1, 2 relatively high capacity of approximately
, and a discharge plateau of
. Modern electronic applications increasingly demand more specific energy, motivating researchers to further increase the capacity of
. Several strategies exist for improving the capacity of
, such as using a carbon coating to increase electronic conductivity.3 In addition, metal oxide coatings, such as
, MgO, and
, and the heating methods used to apply the coatings, can enhance the capacity and capacity retention of
.4, 5 Also, coating lithium metal oxides with other active insertion materials has resulted in some increase in capacity.6, 7
We are investigating the performance of
when combined with
because ruthenium oxide and its related compounds have mechanical stability and desirable electrical properties. Hydrous
is used as an electrode material for electrochemical capacitors because of its high protonic and electronic conductivity.8
also inserts
with high Coulombic efficiency but has an IR drop to a voltage plateau of
.9
has been shown to improve the capacity of
10 and
11 as determined by cyclic voltammetry in half-cells. However, in our laboratory, we have observed that
mixed with
also has a large IR drop when assembled in Li-ion cells and cycled at a constant current.
Lithium ruthenium oxides are also attractive for Li-ion battery cathodes because they possess low electronic resistivity and exhibit facile
insertion properties.12
reversibly intercalates
with high theoretical volumetric capacity13 and has two separate intercalating phases with voltage plateaus of 3.6 and
. Despite its monoclinic symmetry,
is isostructural to the cubic, layered
and
compounds.14 Research has shown that lithium ruthenates synthesized by wet chemistry techniques yield inefficient first charge/discharge cycles but then exhibit high capacity retention with cycling.13
In this paper, we show that the advantageous attributes of both
and
can be leveraged when the two materials are combined in the positive electrode of a battery. The specific capacity of Li-ion and Li-metal cells, having
and
positive electrodes, is compared to that of cells containing physical blends of the two active materials. The effects of
insertion and deinsertion on the crystal structure of the materials, as both single component and in blends, are evaluated with X-ray diffraction (XRD) and electrochemical impedance spectroscopy (EIS). The
is synthesized in our laboratory, and characterized with scanning electron microscopy (SEM) and XRD.
Experimental
Materials
was synthesized by mixing commercially available powders of
and lithium carbonate
following previously described guidelines.13, 14 Briefly, hydrous
(Alfa Aesar) was dried at
for
to remove structural water. Afterwards, the
was ground together with an equimolar amount of
(Alfa Aesar), pelletized, and sintered in air at
for
. The pellets were then pulverized using an agate mortar and pestle and sieved to
. To confirm a consistent synthetic process, each batch of
was characterized by differential thermal analysis (TGA/DTA; Rheometric Scientific) and XRD (Cu
radiation; Bruker D8 Advance). The particle morphology was surveyed with SEM (Leo Supra 55). Commercially available
was used as the standard cathode active material.
Cathodes were made of
,
, or
-rich mixtures of the two. One blend was a 2.3:1.0
molar ratio (which is approximately a
mixture) and another was 11.5:1.0 (approximately
).
Electrode fabrication
The active materials were formulated into inks using previously developed recipes.15 The cathode inks were a mixture of approximately
active cathode material, 2% Super P carbon (Ensaco), 4% KS6 carbon (Timcal), and 3% polyvinylidene fluoride (PVDF, Atofina). The carbon additives were ground with the active material or stirred in the slurry. The PVDF was dissolved separately in 1-methyl-2-pyrrolidone (NMP, Aldrich) and then combined with the active material/carbon powders to make a viscous ink for efficient deposition. For the Li-ion cells, anode inks were made from 90% mesoporous carbon microbeads (MCMB 25–28; Alumina Trading Co.), 1.5% Super P carbon, and 8.5% PVDF dissolved in NMP. An additional ink was prepared from 10% Super P carbon and 90% PVDF in NMP to serve as a conductive coating and adhesion layer to the current collectors.
The inks were deposited by hand on current collectors of aluminum and copper foils (All-Foils, Inc.) approximately
in area. The foils were pretreated by etching with
potassium hydroxide and nitric acid solutions, respectively, followed by rinsing with water and then acetone. A layer of the conductive carbon ink was then added to their surfaces. The conductive coating/adhesion layer was dried on a hot plate at the lowest heat setting for approximately
in order to obtain a homogeneous film. The anode and cathode inks were then spread onto the carbon-coated Cu and Al foils, respectively. The electrodes were subsequently dried again on the hotplate, and then vacuum-dried overnight at
before transferring to an argon-filled glove box. The electrodes were not calendered or pressed.
Electrochemical investigation
The cathode materials were evaluated in both Li-metal and Li-ion cells at room temperature. The cells with Li-metal anodes were made by pressing the cathodes against a glass fiber membrane and a Li foil using stainless steel rods in a T-shaped polypropylene connector (Swagelok,
). The liquid electrolyte was dripped onto the glass fiber prior to final assembly. The electrolyte was
in a mixture of carbonates containing ethylene carbonate, propylene carbonate, ethyl methyl carbonate, and diethyl carbonate.
Li-ion microbatteries, or small packaged cells, were assembled using the procedures described previously.15 The separator, a microporous polymer film (Celgard 2730), was sandwiched between the
/carbon/aluminum and MCMB/carbon/copper electrodes. The cathode/separator/anode sandwiches were first held together by impulse sealing them in a transparent polymer bag (Saranex SX 23-P), followed by sealing in another pouch of a trilayer polyethylene-aluminum-polyester. During the sealing process, small holes were left in the polymer seals to allow electrolyte injection. After the battery was filled with the electrolyte, the pouch was fully sealed. With this packaging process, the Li-ion batteries were protected from ambient air and gently held together with no additional pressure on the electrodes.
The electrochemical testing of Li-ion cells was conducted using a battery tester (Maccor 2300) set to cutoff voltages of 2.0 and
for discharge and charge, respectively, and at constant currents at rates ranging from 2C to C/20, corresponding to current densities of
. In this paper, we focus on experiments at 2C and C/5 rates, or discharge times of
and
, respectively. The test current densities were adjusted according to the total mass of active material(s), and the specific capacities initially measured for
and
. The weight of the active materials in the cathodes ranged from
but had little impact on the electrode appearance and specific capacity. For the blends, appropriate currents were determined by normalizing the weighted averages of the single-component specific capacities.
Impedance measurements were carried out in ambient air on packaged microbattery cells with cathodes containing
,
, or
. The cells were scanned from
with a
amplitude ac current using a frequency response analyzer and electrochemical interface (Solartron 1253/1286) controlled by commercially available software (ZPlot, Scribner). Cells were then discharged and charged at 2C or C/5 currents, with the cycles periodically interrupted for additional EIS measurement between 4.0 and
. Circular fits were applied to the curves of relevant data points, and resistance values were calculated using Zplot. The circular fits of the data were chosen based on an optimization of the tradeoff between the maximum number of data points and the minimum deviation of the fit from the data.
Materials characterization after cycling
XRD was used to check for any cycle-based structural changes in the various blends and single-component cathodes in the metallic Li cell due to cycling. The cells were cycled five times, charged, and then slightly discharged to
, where they were held for
to monitor any unexpected voltage drops. The deintercalated cathodes were then removed from the metallic Li cell while still in the glove box and clamped between Kapton sheets in a holder for the diffractometer. The encapsulated cathodes were then removed from the glove box and analyzed in the XRD with low resolution scans, using a step size of 0.01°
and a step time of
.
Results
The XRD of our synthesized
is shown in Fig. 1 and matches well to the JCPDS pattern for
(no. 85–2000) as well as patterns in the literature.16, 17 The monoclinic structure of
and the ordered nature of its
planes have been used to explain the high reversibility of
insertion into
.14
Figure 1. Powder XRD pattern of
. The indices of the JCPDS pattern for
(no. 85–2000) are shown by annotation.
The scanning electron micrographs for
and
are shown in Fig. 2a and 2b, respectively. The
features
particles in
agglomerates while the
has
particles in
agglomerates. The comparable sizes of the two compounds should foster intimate contact with each other and the carbon while allowing for electrolyte accessibility.
Figure 2. Scanning electron micrographs at
of (a) commercially available
and (b)
prepared by heating
and
.
The second and third cycles at a C/5 rate are shown for the four different cathodes vs Li metal in Fig. 3 as a function of specific capacity. Their specific capacities (mAh/g) and specific energies (mWh/g), as derived from the second cycle discharge curve, are reported in Tables I and II. The
has a single plateau at approximately
on charge and
on discharge (Fig. 3a), values consistent with the literature. The charge-discharge curves for the
in Fig. 3b are similar to those previously described with plateaus at 3.4 and
on charge and 3.4 and
on discharge.14 These data also support prior observations that, while
has a lower voltage plateau, it has a higher specific capacity than
. As determined from the second discharge from
, the specific capacities are
for
and
for
.
Figure 3. Second and third charge-discharge cycles for positive electrodes in metallic Li cells cycled between 2.0 and
at a C/5 rate. The composition of the active material in the cathode is varied: (a)
, (b)
, (c)
and
, and (d)
and
.
The Li-metal cells with physical mixtures of the
and
(Fig. 3c and 3d) have discharge plateaus at both 3.8 and
, reflecting a combination of the characteristic plateaus exhibited by each compound. However, analysis of the data indicates a 3–6% increase in capacity over the weighted average of the single-component specific capacities. That is, instead of the expected
, the 87:13 mix has a specific capacity of
at C/5 and similarly
instead of
for the 58:42 mix; the experimental error is 2% (Table I).
Table I. Capacities measured from the second discharge cycle of Li-metal and Li-ion cells with different cathode compositions at C/5 and 2C rates.
| Cathode | Li-metal C/5 capacity (mAh/g) | Li-ion cells C/5 capacity (mAh/g) | Li-ion cells 2C capacity (mAh/g) |
|---|---|---|---|
|
|
|
|
87:13
|
|
|
|
58:42
|
|
|
|
|
|
|
|
Table II. Specific energies calculated from the second discharge cycle of Li-metal and Li-ion cells with different cathode compositions at C/5 and 2C rates.
| Cathode | Li-metal C/5 specific energy (mWh/g) | Li-ion cells C/5 specific energy (mWh/g) | Li-ion cells 2C specific energy (mWh/g) |
|---|---|---|---|
|
|
|
|
87:13
|
|
|
|
58:42
|
|
|
|
|
|
|
|
The performance of the cathodes in Li-ion cells (Fig. 4) shows a similar trend with
content as seen at C/5 with the cathodes in the Li-metal cells, exhibiting a greater discharge capacity with increasing amount of
in electroactive materials. As in the Li-metal cells, the discharge profile of the Li-ion cells with 58:42
cathode appears to be a superposition of the discharge curves of the two different active materials. Additionally, as in the Li-metal cells, the capacity of the 58:42 blend is approximately 4% higher than expected from the weight-normalized capacities of the active materials.
Figure 4. Charge-discharge characteristics for cycles 1, 5, 10, and 20 of Li-ion cells at C/5 rate and electrodes containing (a)
, (b)
, (c) physical mixtures of
and
, and (d)
and
.
When current is increased from a C/5 to 2C rate, the loss of capacity and energy of the 58:42 blend is not as high as in the case of single-component electrodes, as shown by the data in Fig. 5 and Table I. The
has a discharge capacity of approximately
at the fifth cycle. The
starts with a discharge capacity of
, but then it increases to
with subsequent cycles. The 87:13 mix has a capacity of
. However, the discharge capacity at 2C for the 58:42 mix
is
, which is not only greater than the weighted average of the specific capacities but is also greater than the discharge capacity of the single-component
. In contrast, at C/5 rates, neither of the
blends in Fig. 4c and 4d exceeds the
capacity in Fig. 4b. A small addition of
to the cathode recipe, as demonstrated by the 87:13 mix, results in a slight increase in specific capacity over
for both cycling rates, 93 vs
at 2C as in Fig. 5c and 5a, and also 123 vs 120 at C/5 as in Fig. 4c and 4a. The specific capacity increase with the 87:13 blend is slightly higher at 2C than the C/5 rate.
Figure 5. Charge-discharge characteristics for cycles 1, 5, 10, and 20 of Li-ion cells at 2C rate and electrodes containing (a)
, (b)
, (c) physical mixtures of
and
, and (d)
and
.
The results of the
58:42 blend indicate that the
component makes a greater contribution to capacity at the 2C rate than at the C/5 rate. The contribution of
to capacity therefore depends on cycling rate. The improvement in capacity at this high rate can be observed visually in Fig. 6, which overlays the fifth discharge cycle for the four different electrodes and clearly shows that the 58:42 blend combines the high discharge voltage of the
with the high capacity of the
.
Figure 6. Fifth discharge profile of Li-ion cells at 2C rate with cathodes containing
(solid line),
(dash), physical mixtures of
and
(dash dot), and
(dash dot dot). Dashed lines are provided for 3.87, 3.0, and
and
as a guide for the eye.
The predicted and measured values for the specific energy and capacity are depicted by graph in Fig. 7. A linear increase in the normalized capacity of the cathodes is anticipated with the addition of the higher capacity
to the
. However, the capacity (Fig. 7a) and specific energy (Fig. 7b) of the 58:42 mixture are disproportionately large when the batteries are cycled at the 2C rate.
Figure 7. (a) Specific capacity and (b) specific energy of Li-ion cells cycled at C/5 and 2C rates as a function of
wt % in composite electrodes with
. The predicted values are calculated from the weighted fractions of the specific capacities and energies of the pure materials.
The impedance spectra in Fig. 8 show that the charged batteries with
electrodes have higher impedance than those with
or the 58:42 blend. The curvature is smaller in the case of
than those of the
and
cathodes, indicating a higher resistance. This same trend is observed when the battery impedance is tested at 3.8 and
.
Figure 8. Impedance spectra for Li-ion cells charged to approximately
with cathodes comprising single-component
(circles),
(squares), and a
and
(triangles). Circular fits are shown for comparison.
Expanded views of the XRD patterns of as-prepared and charged cathodes are shown in Fig. 9. The deintercalated electrodes of
in Fig. 9a exhibit the expected shifts in the
peaks from 45.1 to 45.5°
due to the change in the (104) reflection.18 Changes to the
structure, indicated by additional peaks near 40° in Fig. 9b, are also consistent with those previously reported.14 The rearrangement from cubic close-packed to hexagonal close-packed structures has been attributed to lithium deintercalation from the all-lithium layer of occupied octahedral sites, and is exhibited in the appearance of the (202) peak at 44.6°
. The data in Fig. 9c and 9d show that the same peak changes present in the single-component
and
cathodes upon deintercalation are also present in the composite electrodes, suggesting that there is no physical interaction between the two phases as a result of charging and discharging. That is, the respective electrochemical
insertion processes have similar effects on the structure of each material whether as a mixture or as a single component. We therefore view these electrochemical processes as distinct.
Figure 9. XRD patterns of Kapton-encapsulated electrodes before cycling (bottom) and at
(top). The represented electrodes all contain carbon, PVDF, but different active materials: (a)
, (b)
, (c)
, and (d)
.
Discussion
The addition of
to
results in a disproportionate increase in specific capacity and specific energy of
, particularly at 2C rates. The benefit of adding
to
is clearly displayed when the specific energy is calculated from integration of the discharge curves, where an expected improvement in capacity
is observed at a C/5 rate (Fig. 4d) due to the higher capacity of the
. The discharge profiles of the electrode comprising the 58:42
blend simply appear to be a combination of the
and
electrodes. An unexpected 40% increase in capacity is measured at a 2C rate (Fig. 5d), and the
significantly affects the discharge profiles of
electrodes at this rate, as illustrated in Fig. 6. The
plateau is seen in the 87:13 blend, and is pronounced in the 58:42 blend. The extended voltage plateau results in the high specific energy calculated for the 58:42 blend vs
(411 vs
). Also, the specific capacity of the
component at the
plateau of the 58:42 blend does not vary appreciably with cycling rate and is near
for the C/5 and 2C rates (Fig. 4d and 5d).
We considered a chemical interaction between the
and
as the source of the anomalies in the capacities, but we rejected such a mechanism, because 1. the XRD shows no evidence of changes to the materials structure when they are in the blended electrodes, and 2. the capacity increases with decreasing discharge time, following the opposite trend to that expected from kinetics.
We therefore propose that the increased capacity of the blended
cathode is due to the lower resistance of the
; the resistivity of
is
12 and that of
is between 1000 and
.19, 20 We explore this hypothesis first by modeling the composite cathode as two resistors in parallel. The actual cell voltage is the difference between the open-circuit potential or redox voltage and the sum of the activation polarization, concentration polarization, and IR loss. Assuming the IR loss contributes the most to the difference between real and ideal open-circuit potential, internal IR losses can be determined from the discharge profiles and by impedance spectroscopy.
First, in the discharge profiles, the resistance is calculated from discharge curves via Ohm’s law using the discharge current and the difference between the actual and ideal cell redox voltages. An ideal cell voltage of
is used for the
-containing compounds because it is considered the redox potential for
insertion in
determined by differential capacity.21 A value of
is used as the ideal cell voltage for the
. However, this applies only to a low state of charge such as beyond
because a second higher voltage plateau at 3.6 had been reported for discharge to less than
.14 The voltage of the actual cells is determined from the discharge curves at a selected capacity of
in Fig. 6. The resistances of the single-component electrodes are calculated directly from the differences between the actual and ideal voltages, and the resistance of the blends is calculated from Eq. 1 for parallel resistors

The weight percent of the materials is an adequate approximation because the densities of the oxides are similar. The blend has 58%
, and substituting the proportions by weight gives

From the data in Fig. 6, at
, the voltage drops on the curve are calculated from
for
and the blend and from
for
(Table III).
Table III. IR loss
in batteries with cathodes of
,
, and 58:42
after discharging to
(data extracted from Fig. 6), and resistance of the batteries as measured by EIS.
| Active material |
|
|
| Current
| IR loss
|
from impedance
|
|---|---|---|---|---|---|---|
| 3.87 | 3.06 | 0.81 | 125 | 6530 | 270 |
| 3.4 | 2.99 | 0.41 | 360 | 1140 | 154 |
58:42 | 3.87 | 3.48 | 0.39 | 205 | 1880 | 111 |
The assumption that the
and
serve as resistors in parallel in the blend is supported, as the IR loss
values for
and
in Table III yield a weighted resistance average of
, which is within 14% of the actual value of
measured from the voltage plateau of the blend

The materials are compared at a discharge capacity of
because the
component of the blend is still a contribution to the IR, and the relevant value of
used for
in the calculations has validity. For selected capacities closer to the cutoff of
for
, the discrepancy between calculated and measured resistances (using tabulated values from the discharge curves as in the method demonstrated in Table III and Eq. 2) is even smaller than for
. For example, with
the discrepancy between the calculated and measured resistance of the blend is only 6%.
The resistance values were also measured for the
,
, and 58:42 blend by EIS and are reported in Table III. In EIS of two-electrode cells, the curvature of the first impedance arc can provide information about charge‐transfer resistances while the high-frequency point of the arc is due to ohmic resistance. Thus, the EIS studies can be used as a comparison for the different electrodes in this study.
As above, from the resistances measured from EIS
for
and
and using a weighted average, we predict a resistance of

which is nearly double the measured value of
The resistance of the blended cathode is clearly lower than that of the single-component electrodes, as determined both from the IR drop of the discharge curves and from EIS, thereby confirming that the two materials serve as a parallel combination of Li-ion insertion materials. The discrepancy between the measured and calculated resistances from EIS is likely a result of measuring impedance on full cells, which makes it difficult to isolate the contribution of the cathode. Still, it is a direct comparison to similar cells used in other methods of characterization in this paper.
The model for parallel resistors is further supported by the observed increase in the capacity with increasing
content of the electrodes. The 87:13 blends of
and
show a smaller response to increased rate than the 58:42 blend, presumably because the concentration of
is below the volumetric percolation threshold for connectivity throughout the electrode. This helps explain the 87:13 blended cathodes having only a slightly greater specific capacity than pure
. The 58:42 blend has enough of both the
and
for the resistors to exist in parallel, resulting in a significant change of a observed capacities.
Because the effect reported in this study relies on optimization of percolation networks or connectivity, it is likely that adjustments to the materials’ particle size, morphology, and their relative concentration (or volume) can be made to further improve the performance of the electrodes. The use of nanoparticles in the blends may also improve the performance as ionic conductivity can increase with decreasing particle size, although the electronic conductivity may decrease;8 further experimentation would be needed to determine whether the ionic or electronic resistivity is the most important factor affecting the IR drop and impedance. Although
may be acceptable for applications requiring small amounts of active material, such as microbatteries, its benefit might not be economically viable for large batteries due to the high cost of ruthenium. It may be valuable to use the approach by Kobayashi et al. and Moore et al. to assess other lithiated transition metals, such as Mn, Ti, Sn, Zr, or Mo, which are structurally equivalent to
.12, 13 Care must be taken to select combinations with a redox potential greater than
, so that the
structure is not stressed by discharge to low voltages.
Conclusions
In this paper, we have demonstrated the enhancement of the specific capacity and energy of batteries by the addition of
to
, particularly at high cycling rates. This effect is observed even though the materials are utilized discretely in the cells and have no apparent chemical interaction. We show from the IR losses in the discharge profiles and impedance spectroscopy that the blended electrodes are effective because their parallel combination lowers the cell resistance, which decreases the IR drop and increases the capacity.
Although
may not be a practical material for large batteries, it may be useful for microbattery applications, in which its high cost will have little impact. Furthermore, microbatteries are typically sought for microdevices, and having a battery with the capability to perform well at high rates may allow the use of a single battery vs a battery/capacitor combination.
In summary, this paper shows that the simple mixture of two active materials in a cathode can have a significant impact on the resistive properties, discharge voltage, and specific capacities of batteries.
Acknowledgments
The authors are grateful to the Office of Naval Research for support of this research. A.M.S. is a National Research Council postdoctoral fellow.
Naval Research Laboratory assisted in meeting the publication costs of this article.



















































