Drying

www.acsaem.
org Article
Correlating Processing Conditions to Short- and Long-Range Order

in Coating and Drying Lithium-Ion Batteries
Renee M. Saraka, Samantha L. Morelly, Maureen H. Tang, and Nicolas J. Alvarez*
Cite This: ACS Appl. Energy Mater. 2020, 3, 11681−11689 Read Online
ACCESS Metrics & More Article Recommendations *

sı Supporting Information
ABSTRACT: Processing conditions of battery slurries into electrodes are known to affect final battery performance.
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
However, there is a lack of fundamental understanding of how the relationships between processing conditions, the
slurry microstructure, and the film microstructure affect electrode performance. This study determines the effects of
the coating shear rate and drying temperature on battery electrode performance via discharge capacity. We use
Downloaded via DREXEL UNIV on January 28, 2021 at 17:02:37 (UTC).
rheological measurements and energy dispersive X-ray spectroscopy (EDS) to correlate slurry and electrode
microstructures to trends in discharge capacity. The radial distribution function is used to quantify differences in the
electrode microstructure. More specifically, we show that the correlation between carbon and active material EDS
detections to be the most relevant in understanding battery performance. Electrodes with both short- and long-range
carbon/active material orders have the highest discharge capacities. This microstructure can be obtained through high
shear rates, which induce better carbon dispersion via strong hydrodynamic forces, or through high temperature
drying by preventing unwanted time-dependent structural changes after flow cessation. This analysis provides concrete
evidence for the importance of both short-range and long-range contacts between the conductive additive and active
material on battery performance.
KEYWORDS: electrode manufacturing, shear effects, drying effects, battery slurry rheology, radial-distribution function, lithium-ion battery
1. INTRODUCTION showed via EDS image analysis that higher temperatures lead
As lithium ion batteries power the growing market of to higher binder concentration at the electrode’s top surface,
consumer electronics and electric vehicles, researchers which results in reduced adhesion strength between the
continue to focus on increasing the battery energy density. electrode and current collector.17 Subsequently, Westphal and
Although advanced materials continue to improve theoretical Kwade suggested to avoid drying-induced segregation of the
capacity limits, optimization ultimately requires processing polymer binder via a soft drying temperature of 80 °C to
technology to achieve fast, homogeneous ion and electron improve mechanical elasticity and relative electrode resist-
transport. There is growing consensus that the performance of ance.18 Furthermore, Jaiser et al. suggest that lower drying
temperatures result in an ideal, homogeneous binder
battery electrodes with industrially relevant carbon concen-
distribution that provides higher discharge capacities.19
trations (<5 wt %) is effectively controlled by electron
These studies suggest that the distribution of binder is
transport.1−5 Efficient electron transport is directly related to
essential in ensuring adhesion to the current collector. While
the structure and connectivity between the three components,
binder is an essential component, it is a nonconductive additive
that is, the active material, conductive additive, and polymer
and is studied in the anode, which is not typically capacity
binder.6,7 The electrode microstructure and connectivity in
limiting. Fewer studies have addressed the effect of drying
turn strongly depend on the electrode slurry mixing, coating,
temperature on the conductive carbon distribution in cathode
and drying processes.2,3,8−15 Previously, we showed how slurry
electrodes, which could lead to more rational capacity
mixing impacts the short-range electronic contacts that control
optimization.
electrode rate capability.1 However, slurry coating and drying
The applied shear rate during coating is arguably important
also are known to impact performance. Relating these
for proper dispersion of the components in the final electrode
processes to the electrode microstructure, particularly elec-
film. Few studies have directly examined the effect of shear on
tronic connectivity, is necessary for further insight and
electrode performance. However, some studies have reported
improvements to electrode manufacturing. Several experimen-
tal and theoretical studies have shown the impact of drying
temperature on the distribution of the polymer binder. The Received: July 20, 2020
solvent removal process has been modeled16 and studied Accepted: November 2, 2020
experimentally. In graphite anodes, researchers have found that Published: November 30, 2020
binder accumulates at the surface of the anode with higher
drying temperature, consequently depleting the binder at the
current collector resulting in poor adhesion.17−19 Muller et al.
© 2020 American Chemical Society https://dx.doi.org/10.1021/acsaem.0c01305

11681 ACS Appl. Energy Mater. 2020, 3, 11681−11689
ACS Applied Energy Materials www.acsaem.org Article
the effect of shear on physical properties of electrodes in order Scanning electron microscopy (SEM) and EDS maps are
to maximize the area of the useful electrode film.14,20,21 For studied to correlate battery performance to component
example, Bitsch et al. utilized capillary suspensions to improve distribution in cross-sectional images of the electrode films
electrodes at low shear and keep slurry viscosity constant at through a radial distribution function. The radial distribution
high shear. The additives used in their study decreased function defines the probability of finding a molecule at a radial
capillary spreading, decreased the area affected by edge effects, distance from another molecule of interest. It is historically
and increased the coating edge thickness.14 Using slot die used to understand interactions between molecules in solids,
coating, the authors also conclude that increasing the shear rate liquids, and gases, for example, defining order in crystalline
increased the coating edge thickness.20 It is well-known that lattice structures. We propose that the radial distribution
slot-die coated films outperform doctor blade coated electro- function applied to EDS maps can be used to quantify short-
des. One explanation for this is the difference in shear rates. range and long-range pathways between carbon and the active
Slot die coating achieves shear rates on the order 103 s−1 and material. Furthermore, we show that there are direct
shear strains of order 10.20 However, the effect of shear still correlations between short-/long-range pathways and electrode
needs to be quantified. performance.
The electrode microstructure is expected to be a strong
function of the shear applied during coating. Industrially 2. METHODS
relevant battery slurries form colloidal gels in the absence of 2.1. Electrode Film Composition and Mixing. Electrode films
dry-mixing.1,22 There is a well-characterized nonlinear depend- were made with 95 wt % active material of LiNi0.33Mn0.33Co0.33O2
ence of the colloidal gel microstructure on the applied shear (NMC) (Toda America, NM3100) with an average particle size of 10
rate.23−26 Eberle et al. showed that hydrodynamic forces μm and a Brunauer−Emmett−Teller (BET) area of 0.5 m2/g.
induced in situ cluster reorganization depending on the applied Conductive additive consisting of 2.5 wt % CB (MTI, Super C65)
shear rate, suggesting that shear rates of sufficient magnitude with a BET area of 62.6 m2/g was used. The active material and
will increase particle dispersion and connectivity during film conductive additive were added to a solution of 2.5 wt %
formation.24 Furthermore, Colombo et al. evaluated the polyvinylidene difluoride (PVDF) (Arkema, MW = 380k) in 3.405
mL of 1-methyl-2-pyrrolidinone (NMP), which was mixed separately
evolution of the colloidal gel microstructure after low shear, in a Thinky planetary mixer at 1800 rpm for 10 min. The PVDF
suggesting that selective aggregate breakdown is occurring solution and particles were combined and mixed in the same device
even with extremely low shear rates.26 In contrast, Varadan and for 7.5 min at 1800 rpm. The final slurry composition remained
Solomon showed that their colloidal gel showed no changes in constant through all trials at 95% active material, 2.5% CB, and 2.5%
aggregate size at shear rates of 120 s−1, especially close to the PVDF binder.
gel point temperature.27 These conflicting results clearly show 2.2. Electrode Film Coating and Drying. Slurries were coated
that shear-induced changes in the microstructure depend on onto an aluminum current collector via an automatic coater with a
the type of gel and the magnitude of the shear rate. doctor blade attachment of depth 4.33 mm (0.17 in.) and 150 mm
coating width (TOB Energy, TOB-TM300). Coating speeds are
The drying time is also expected to influence the electrode
adjustable between 0 and 100 mm/s with an accuracy of 10 mm/s.
structure after the cessation of shear. This is supported by the The doctor blade has an adjustable coating thickness between 0.01
literature that shows colloidal systems undergo aging and and 3.5 mm. The coating thickness was constant for all studies at 0.1
recovery after flow cessation.26−31 The mechanism for aging is mm, resulting in a constant shear strain, γ = 43.3. This strain is a result
thought to arise from particle rearrangements caused by of the gap height (coating thickness) and the width of the doctor
hydrodynamic shear forces.28 For example, Gordon et al. blade used to coat the film. Because it is important to control film
showed that colloidal gels show faster structural recovery thickness for similar ion transport between films, the induced shear
(measured via elastic modulus) for higher imposed stresses.28 strain was fixed for all coatings. The magnitude of strain used in this
study is on the same order of magnitude as slot-die coating setups.
This is supported by another study that shows a delay in the
Shear rates were measured using video analysis. Films were coated at
relaxation of the colloidal microstructure due to changes in the 300, 750, and 1300 s−1, then either placed in a fume hood to dry
network density during the applied shear flow.29 The evolution overnight at room temperature, at 80 °C using the coater’s oven
of the colloidal microstructure during shear and drying have attachment, or at 150 °C using an in-house controlled hot plate setup.
not been explicitly tied to battery electrodes during processing. Visual inspection showed drying times of >8 h, 2 h, and 5 min,
We recently showed that improved electrode performance respectively.
for carbon-starved slurry formulations is achieved via dry- 2.3. Electrode Preparation and Assembly. After the initial
mixing carbon black (CB) with an active material prior to wet- drying time, all films were further vacuum-dried at 120 °C for 12 h,
mixing. The dry-mixing induces short-range contacts between then punched into 3/8″ diameter electrodes. Individual electrodes
were compressed at 20 MPa to ensure adequate substrate adhesion
carbon and the active material to achieve improved electron using a Carver melt press at room temperature. Standard size 2032
transport.1 One important question that remains is whether half cells were fabricated in an argon atmosphere glove box with
shear and temperature during processing can improve carbon lithium metal counter electrodes, two Celgard separators, and an
and active material connectivity without dry-mixing. This study LP30 electrolyte (Gotion).
examines the combinatorial effect of the shear rate and drying 2.4. Testing. 2.4.1. Rheological Measurements. Amplitude
temperature on battery performance. Shear rates from 300 to sweeps were measured from 0.1 to 10% strain and frequency sweeps
1300 s−1 and drying temperatures between 23 and 150 °C are were measured between 0.1 and 100 rad/s at 0.1% strain on an ARES
applied to the same colloidal gel slurry formulation. These G2 (TA Instruments) rheometer. Only slurries exhibiting reprodu-
cible rheology after mixing were cast into films. Additional frequency
shear rates are intended to span low shear rates typical in sweeps using a DHR-3 (TA Instruments) were taken after applied
laboratory studies and shear rates achieved in slot-die coating. shear. To mimic the shear flow during coating, steady shear rates were
The dried films are compressed, made into a battery, and applied with a 40 mm parallel plate geometry at 300, 750, and 1300
tested for rate capability. As expected, electrode performance is s−1 for 10, 7, and 3 s, respectively. Microstructure relaxation
a strong function of the drying temperature and shear rate. experiments were conducted via frequency sweeps from 0.1 to 100
11682 https://dx.doi.org/10.1021/acsaem.0c01305
ACS Appl. Energy Mater. 2020, 3, 11681−11689
rad/s at 0.1% strain measured 0, 1, and 2 h after constant shear rate

experiments.
2.4.2. Conditioning and Rate Capability Testing. Electrochemical
testing was completed using an Arbin battery cycler (Arbin
Instruments). All cells were conditioned by completing four cycles
at a rate of 0.1 C ranging between 4.3 and 3.0 V. After conditioning,
cells completed four cycles discharging at 0.5 C, 1 C, 2 C, 5 C, 10 C,
then a final 0.5 C. All charging cycles were performed at a constant 0.1
C.
2.4.3. Microscopic Imaging and Analysis. SEM images (FEI
Quanta 600 ESEM) and energy dispersive spectroscopy (EDS) maps
(EDAX Team) were measured (20 kV) on the same films used for
battery testing. The SEM and EDS images were collected on the same
region of interest and analyzed to determine component distributions.
The samples were prepared by vertically slicing electrodes in half with
a razor blade. The half of the electrode was then mounted to a 90°
angle SEM stub with a conductive carbon tape, exposing the exposed
cross section. Previous work by our group studied SEM sample prep
methods and showed that slicing from the top to bottom or bottom to
top does not cause measurable smearing of elements.32 Cross section
polishing was not used to prevent altering element distribution. SEM/
EDS images were cropped to only include the electrode cross section.
EDS imaging provides spatially evolved elemental analysis from areas
as small as 1 nm. EDS works by applying high voltage energy to a
conductive sample which causes inner shell electrons to transition to a
higher energy level and emit X-rays with characteristic energy of the
probed element. Thus, elements are identified by a peak energy.33 20
kV excitation was chosen for this study, which is 2.5 times greater
than the highest elemental Kα of nickel, 7.471 keV. EDS signals for
carbon (C), fluorine (F), aluminum (Al), nickel (Ni), cobalt (Co),
manganese (Mn), and oxygen (O) were detected during testing. For
this study, we will focus on the spatial distributions of C detections as
an indicator for the conductive additive and Mn as an indicator for the
active material.
SEM and EDS images are a result of a large depth-of-field, which
leads to a three-dimensional (3D)-like imaging of the electrode cross-
section surface. For simplicity, we developed a routine to reduce the
3D EDS maps to two-dimensional (2D) planar elemental maps. The
3D image is dissected into planes of detection intensity and reduced
to a 2D map of pixels with a set intensity range. More specifically, a
pixel intensity histogram was created for each EDS image. A distinct
population with the highest pixel intensity is considered to represent
the top layer, that is, the brighter the pixel intensity the closer its
proximity to the detector. The image was reduced to only include
pixels in the highest pixel intensity population. Using the reduced
image, the radial distribution function was calculated to coordinate
the location of carbon or manganese pixel detections. That is,
nonblack pixels in the respective EDS map, to all other carbon or
manganese detections within a given radius, r. The 2D radial
distribution function, g(r), is given by
H(r ) Figure 1. Rate capability results for (a) 23 °C drying shows low
g (r ) =
N ρπ[(r + Δr )2 − r 2] discharge capacities that decrease with the decreasing shear rate (b)
150 °C samples show nearly identical discharge capacities with
where r is the radial distance, H(r), is the pair correlation histogram varying shear, suggesting the importance of temperature (c) 1300 s−1
normalized by the 2D particle density, ρ = N/L2, and the number of results in high discharge capacities that improve slightly with
detections, N, where L is the radial length of the box given by L2 = increasing temperature with minimal variance within each sample
(Lx2 + Ly2).34 set (d) a summary of all discharge capacities at 0.5 C show the best
performance at 1300 s−1 and 150 °C.
3. RESULTS AND DISCUSSION
The rate capabilities of electrodes with variable coating shear greater are significantly reduced. For electrodes dried at the
and drying temperature are shown in Figure 1a−d. Each data lowest temperature of 23 °C (Figure 1a), discharge capacity
point represents the average discharge capacity of three to four increases monotonically with the shear rate. At the highest
electrodes with four cycles each (an average of 12−16 drying temperature of 150 °C (Figure 1b), very little effect of
capacities per C-rate). Error bars represent one standard the shear rate is observed. Figure 1a,b suggest better electrodes
deviation of the 9−12 data points. A full account of the cycles result from high shear rates and high temperatures.
contributing to one average value is shown in Figure S1. Figure 1c compares the discharge capacities at high shear
Consistent with expectations, all electrodes approach the rates (1300 s−1) for films coated at three temperatures. There
theoretical capacity at 0.1 C, while capacity at rates 2 C and is an increase in the discharge capacity with increasing
temperature. At this shear rate, the differences between

discharge capacities is significantly smaller than other rates,
for example, a decrease of 20 mA h/g at 1 C compared to
Figure 1a, which shows a decrease of 70 mA h/g. Overall at 1
C and high shear rates, the intermediate and high temperature
samples show relatively high discharge capacity compared to
low shear rates, see Figure 1a. At 1 C, low temperature drying
shows a much more significant drop in discharge capacity,
suggesting the importance of fast drying. A comparison of
Figure 1b,c suggest that electrode performance is significantly
more sensitive to drying temperature than the shear rate.
Figure 1d summarizes the findings by comparing the discharge
capacity at 0.5 C for all conditions. At low shear rates (black
line), the discharge capacity increases monotonically with
increasing drying temperature. Intermediate shear rates (red
line) show a surprising nonmonotonic increase in battery
performance with increasing temperature; a minimum in
discharge capacity is achieved at 80 °C. An additional study at
500 s−1 (green line) was conducted to validate the minimum.
High shear rates (blue line) exhibit high discharge capacities,
and have significantly less variation with temperature. It is clear
from Figure 1d that high shear rates allow for high discharge
capacities at any temperature. This shear rate is on the same
order of magnitude as that applied during slot die coating,20
which is a process known to make better performing
electrodes. The highest temperature shows a constant high
discharge capacity. There are many possible factors that
contribute to the observed capacity differences. It is important Figure 2. Thickness (a) and porosity (b) cannot simply explain the
to consider physical properties of the electrodes as well as discharge capacity as proofed by differences between boxes 1, 2, and 3
processing-induced changes. and similarities between circles 1 and 2.
It is well-known that ion and electron transport are faster in
thinner electrodes, while ion transport increases with porosity. give a general overview of the particle packing and distribution
For low-carbon formulations such as the systems here, lower in the electrode, while EDS maps show the elemental
porosity is expected to yield better performance. Figure 2a distribution. Two sample SEM and EDS maps are shown in
shows that thickness is unable to describe all of the trends in Figure 3. SEM samples, shown in Figure 3a,d were prepared
electrode performance. Although all electrodes in this work with previously validated methods to prevent element
were coated with a constant doctor blade height of 100 μm, smearing and distribution alteration. Figure 3a−c correspond
Figure 2 shows that the final film thickness and porosity to a film processed at 300 s−1 and dried at 23 °C. Figure 3d−f
strongly vary with the applied shear rate and drying corresponds to a film processed at 1300 s−1 and dried at 23 °C.
temperature. One possible explanation for film thicknesses Recall that the higher shear rate sample (Figure 3d−f) has
greater than 100 μm is due to an increased wet film thickness significantly higher discharge capacity. Manganese (Mn) maps
caused by lubrication forces during coating.34 There are in Figure 3b,e appear to be uniform and consistent for both
significant outliers to the expected trend with thickness. For electrode films, while carbon maps in Figure 3c,f show
example, boxes 1 and 2 in Figure 2a represent electrodes qualitative differences in carbon dispersion and connectivity.
fabricated at 23 °C and dried at 750 and 300 s−1, respectively. It is well-understood that conductive additive is necessary to
The differences in porosity and thickness are ±20% and ±10 transport electrons to and from the surface of active material
μm, however, the difference in capacity is significant, 43 mA h/ particles to complete the electrochemical reaction. Therefore,
g, or 30% lower for box 2 at 0.5 C. Boxes 1 and 3 represent discharge capacity is expected to strongly depend on how
different shear rates as well as different temperatures. The connected a given active particle surface is to the carbon filling
difference in porosity and thickness are ±20% and 10 μm, interstitial sites (“short-range contacts”) and how well the
respectively, however, the capacity loss is much greater than interstitial carbon is connected to the current collector (“long-
expected (70 mA h/g). Furthermore, Figure 2b clearly shows range contacts”). Our previous results argue that neither long-
that ion transport is not limiting electrode performance. Circles range carbon connectivity, nor short-range contacts alone are
1 and 2 have very similar discharge capacities (140−130 mA enough to ensure good battery performance. Rather, both
h/g) but different thicknesses (95−120 μm) and porosity long- and short-range contacts must be present and form a
(25% lower). Overall, although variations in thickness and continuous carbon network from the surface of the active
porosity certainly impact the overall electrode performance, particles to the current collector. To quantify these long- and
they do not account for all of the variations in the discharge short-range contacts, we utilized a radial distribution function
capacity. Further analysis of the electrode microstructure is analysis to perform correlations of the locations of manganese
needed to determine how shear and drying impact local pixels to carbon pixels (Mn/C), and carbon pixels to other
connections between particles. carbon (C/C) pixels. The aim is to correlate elemental
SEM and EDS imaging of electrode cross sections were used distribution in the battery electrode to battery performance.
to gain insight into the electrode microstructure. SEM images Mn was arbitrarily chosen as a representative signal of the
Figure 3. SEM and EDS images (a) 23 °C, 300 s−1 electrode image (b) 23 °C, 300 s−1 manganese map is uniform with small open pockets (c) 23
°C, 300 s−1 carbon map shows a sporadic C distribution (d) 23 °C, 1300 s−1 electrode image (e) 23 °C, 1300 s−1 manganese map is homogenous
(f) 23 °C, 300 s−1 carbon map shows sporadic distribution with come cluster connectivity.
Figure 4. Radial distribution function coordinating carbon to manganese detections (a) 23 °C, 300 s−1 shows minimal coordination (b) 23 °C, 750
s−1 has primarily short-range contacts (c) 23 °C, 1300 s−1 has large peaks for both short- and long-range connectivity (d) 80 °C, 1300 s−1 has
multiple peaks throughout large distances, suggesting more connectivity (e) 150 °C, 1300 s−1 shows a strong presence of both short- and long-
range connectivity.
active material, but analysis based on Co or Ni would yield measured (1) as a function of EDS collection time and were
identical results. found to be independent of time, and (2) at different spatial
In this study, the reduced 2D map considers only the cross-sections of the same electrode and found to be
brightest pixel intensities and thus represents the top cross- independent of the sampled region. The results are shown in
section layer of the electrode (i.e., closest layer to the SEM Figures S2 and S3, respectively. Control (2) in Figure S3 is
detector). This is done through analyzing pixel intensity considered a particularly important result, as it demonstrates
histograms, then setting a threshold to only include the highest that analysis of one EDS image of 100 μm × 150 μm is
intensity population. The highest intensity population is considered representative of the entire film. Even so, multiple
chosen to represent the surface elemental layer. The reduced cross-sections were analyzed under different experimental
2D maps are then analyzed via 2D radial distribution functions conditions (not shown) to ensure accuracy of the results.
between carbon/carbon (C/C) pixels and carbon/manganese Figure S4a−c show radial distribution functions for C/C
(Mn/C) pixels. Two controls were performed to ensure that correlations for all low shear rate electrodes discussed in Figure
the radial distribution results were representative of the 1. Interestingly, electrodes with very different performance, for
electrode film surface. Radial distribution functions were example, 300 s−1, 23 °C (Figure S4b) and 300 s−1, 150 °C
(Figure S4c), have very similar C/C radial distribution magnitude and frequency-independence of the storage (G′)
functions, that is, the overall electron conductivity is not and loss (G″) moduli indicate a strong colloidal gel with a
limiting battery performance. This strongly supports the spanning network of particles, consistent with previous
observations in Morelly et al., which showed similar in-plane findings.1 We propose that the modulus as a function of
film conductivity between high- and low-performing electro- time and processing history is a good indicator of carbon
des.1 connectivity in the electrode slurry. For example, a decrease in
Figure 4a−c shows the results of the Mn/C radial modulus indicates a loss of carbon network connectivity and
distribution function for low temperature (23 °C) films at the formation of carbon islands (aggregates) throughout the
three shear rates (300, 750, and 1300 s−1). The value of g(R) electrode film. Therefore, we hypothesize that a correlation
represents the probability that a C pixel is found within a
exists between modulus after shear and during drying and the
certain distance of a Mn pixel.35 Figure 4a shows one peak
radial distribution results observed in Figure 4. To this end, the
within 5 μm of any given Mn pixel. Because the average NMC
radius is 5 μm, any peak before this distance indicates carbon effect of shear on the slurry structure is monitored by first
directly in contact with the NMC particle surface (i.e., short- applying a constant shear rate for a small strain, followed by a
range contacts). Beyond 5 μm, g(R) is relatively flat, which small amplitude oscillatory shear frequency sweep immediately
indicates a relatively random distribution of carbon.36 Figure after flow cessation. Figure 5 shows the results at room
4b shows a high concentration of C around a Mn pixel temperature of angular frequency sweeps at three times (0, 1,
followed by a steep drop-off toward a random distribution and 2 h) after exposure to (a) 300, (b) 750, and (c) 1300 s−1.
around 20 μm. This radial distribution indicates primarily
short-range Mn/C contacts, while Figure 4c indicates an
increase in the order of Mn/C correlations at higher radial
distances with distinct peaks at 5, 20, and 42 μm. Because
electrode performance increases from Figure 4a,c, we therefore
observe that the best performance correlates to the electrode
with the highest short- and long-range C/Mn order. Figure 4b
suggests that short-range order is better than the flat Mn/C
g(r) observed in Figure 4a. Together Figure 4c,b support the
hypothesis that both short- and long-range C order with
respect to NMC maximizes battery performance.
Figure 4c−e show the results for high shear (1300 s−1) films
at three drying temperatures (23, 80, and 150 °C). Recall that
for all samples, different (at least two) cross sections of the
same electrode were subjected to this analysis to ensure that
these results are representative of a larger area of the electrode.
Figure 4d indicates an initial high presence of C in direct
contact with the active material surface, as well as multiple
peaks of connectivity with increasing radial distance. Similarly,
Figure 4e displays a multitude of peaks with increasing radial
distance and similar magnitude with no significant drop in
Mn/C probability for the entire range of distances. Figure 4c−
e suggest that high shear rates induce short- and long-range
order, creating shorter pathways for electrons and ions to come
together. Figure 4d,e show very similar performance and have
slightly higher discharge capacities than Figure 4c. Radial
distribution functions have historically been used to determine
the crystal lattice structure by identifying the number of
nearest neighbors, and has since been adapted for use in
amorphous solids.36−38 Many mathematical models for other
particle shapes have also been developed39 and all suggest that
this is a valid analysis of coordination by detecting nearest
neighbors. Results that display multiple peaks correlate to a
more highly ordered crystal structure,36 and for this study, the
“crystal structure” is defined as the C network, relative to Mn.
Multiple peaks correlate to a higher Mn/C order in the form of
short- and long-range contacts. The results in Figure 4c−e
strongly support the hypothesis that short- and long-range
Mn/C order leads to higher battery performance. However,
the question remains why such vastly different structures arise
from different shear and drying conditions. Figure 5. Time-dependent rheological response at room temperature
It is known that the rheological response of battery slurries is after (a) 300 s−1, where the decrease in moduli is caused by particle
primarily influenced by the conductive carbon additive densification (b) 750 s−1, observing an increase in moduli caused by
network.22,40 Figure S5 shows rheological characterization for particle dispersion (c) 1300 s−1, showing immense increases related to
the initial slurry used for all coating and drying conditions. The high dispersion of carbon particles.
The 0 h frequency sweep shows that 300 s−1 has a structure surface, and strongly suggest that fast drying and no network
similar to the structure measured before shearing (Figure S5) reorganization is more effective at increasing discharge capacity
while 750 and 1300 s−1 show significantly lower moduli than than higher shear rates. The optimum temperature for drying
the starting structure. At low shear rates, Figure 5a shows that will also need to consider physical properties, such as adhesion
the modulus decreases with time, indicating a decrease in between the electrode and current collector as discussed in the
carbon-to-carbon network connectivity. However, for 750 and works of Bauer, Schabel, and co-workers.14,17,19
1300 s−1, the modulus increases with increasing time. This
suggests that the carbon network is becoming more connected 4. CONCLUSIONS
after exposure to high shear rates and less connected (forming A combination of electrochemical testing, rheological charac-
dense aggregates) at low shear rates. terization, and EDS mapping with image analysis explains the
These results are in line with other reports in the literature. effects of shear and drying temperature on the battery
For example, the work by Eberle et al. defines a dimensionless electrode microstructure and performance. Overall, the highest
6πμs a 2γ ̇
value, M′ = to relate shear forces to particle attraction electrode discharge capacity is observed at high drying
Fmax
temperatures (150 °C) and high shear rates (1300 s−1).
forces. The results shown in Figure 5 can be explained if the Manganese (Mn) and carbon (C) connectivity, measured via
hydrodynamic shear stresses are larger than the interparticle the radial distribution function, correlates well with average
attractive forces, that is, M′ > 1 for γ̇ ≥ 750 s−1 and smaller, discharge capacity. For example, higher shear rates at low
that is, M′ < 1, for γ̇ = 300 s−1. Although measured on a drying temperatures lead to better long-range Mn/C orders
different experimental system, Eberle showed that M′ < 1 leads and better battery performance, while high temperatures yield
to the formation of dense clusters/aggregates and reduced long-range Mn/C orders and better battery performance for all
network interconnectivity, while M′ > 1 leads to increased shear rates. The results clearly indicate that temperature is
network connectivity.24 Furthermore, Colombo et al. show more important than the shear rate in achieving long-range
that gels exposed to low shear, for example, up to 100 s−1, Mn/C order, that is, better battery performance. This strongly
evolve to isotropic states with reduced moduli over 3000 s, or supports the hypothesis that short and long-range order
50 min.26 Coincidentally, similar timescales are observed in between the active material and conductive additive govern
Figure 5a. electrode performance. Additionally, there is an observed
Figure 5 helps to explain the results in Figure 4a−c and increased in electrode performance as the shear rate increases,
suggest a correlation between carbon network connectivity and which could explain the better performance of slot-die coated
long-range order observed in battery electrodes. For example, electrodes compared to electrodes made from lower shear rate
Figures 5a and 4a (γ̇ = 300 s−1) show the lowest carbon blade coating.
connectivity (via rheology) and the lowest Mn/C long-range The dependence of short- and long-range order on Mn/C
order (via radial distribution function), respectively. Figures 5c distributions is best explained using rheology of the slurry after
and 4c (γ̇ = 1300 s−1) show the highest carbon connectivity coating. Our findings, supported by the colloidal gel literature,
and highest Mn/C order, respectively. These strong show that too low of a shear rate during coating can lead to
correlations suggest that there is a critical shear rate for densification of the microstructure and a loss of long-range
which battery slurries should be processed to ensure dispersion Mn/C order. The resulting loss of Mn/C correlations inhibits
of carbon and connectivity between carbon and the active electron transport and consequently decreases discharge
material. As previously mentioned, this shear rate is on the capacity. For higher shear rates, the slurry microstructure
same order of magnitude as that applied during slot die shows increased carbon connectivity and higher long-range
coating, a process known to make better electrodes. The Mn/C order. The effects of microstructural changes after shear
observed change in the microstructure at an increased shear are minimized with increased drying temperature, which locks
rate could be responsible for the better performance of slot die the sheared structure in place. These results clearly indicate the
coated electrodes. importance of the slurry microstructure during coating and
Because the room temperature drying cases take up to 24 h drying and are expected to depend on the starting slurry
to dry, the results of Figure 5 should be a realistic microstructure.
representation of the final slurry microstructure before drying To our knowledge, this work is the first study to correlate
at room temperature. At higher temperature, drying occurs the underlying elemental connectivity to battery performance
much faster, and the results in Figure 5 may not represent the as a function of shear and temperature. The radial distribution
final structures. However, when the rate of drying is very fast, function applied to EDS elemental maps is an effective method
that is, 150 °C, the effect of structure reorganization is to quantify short- and long-range contacts in battery
expected to be significantly reduced, and the structure electrodes. This method should prove useful in future studies
immediately after flow cessation is more relevant. In fact, the to correlate processing conditions to battery performance.
■
lowest shear rate appears to have the highest electrode
performance in Figure 1d. Figure 5a shows that the zero-hour ASSOCIATED CONTENT
structure for 300 s−1 has the highest modulus and therefore * Supporting Information
sı
highest carbon network connectivity. If we assume that at high
The Supporting Information is available free of charge at
drying rates this structure is locked into place, then the
https://pubs.acs.org/doi/10.1021/acsaem.0c01305.
electrode should have both short- and long-range Mn/C
orders. This is strongly supported by the radial distribution Full account of battery cycling data; effects of EDS
functions in Figure S6a,b, which show these correlations for exposure time on the radial distribution functions after
the low shear rate dried at high temperature. These results thresholding; results of radial distribution functions for
support the importance of both temperature and shear on the multiple analysis areas of the same electrode to validate
final carbon dispersion with relation to the NMC particle conclusions; C/C radial distribution function results;
rheological frequency sweep of initial slurry structure; Batteries: A Numerical Study. J. Electrochem. Soc. 2015, 162, A1485−
and C/Mn coordination for increased drying temper- A1492.
ature (PDF) (8) Kraytsberg, A.; Ein-Eli, Y. Conveying Advanced Li-Ion Battery
■
Materials into Practice The Impact of Electrode Slurry Preparation
Skills. Adv. Energy Mater. 2016, 6, 1600655.
AUTHOR INFORMATION (9) Mohanty, D.; Hockaday, E.; Li, J.; Hensley, D. K.; Daniel, C.;
Corresponding Author Wood, D. L. Effect of Electrode Manufacturing Defects on
Nicolas J. Alvarez − Department of Chemical and Biological Electrochemical Performance of Lithium-Ion Batteries: Cognizance
Engineering, Drexel University, Philadelphia, Pennsylvania of the Battery Failure Sources. J. Power Sources 2016, 312, 70−79.
19104, United States; orcid.org/0000-0002-0976-6542; (10) Yoo, M.; Frank, C. W.; Mori, S. Interaction of Poly(Vinylidene
Email: [email protected] Fluoride) with Graphite Particles. 1. Surface Morphology of a
Composite Film and Its Relation to Processing Parameters. Chem.
Authors Mater. 2003, 15, 850−861.
Renee M. Saraka − Department of Chemical and Biological (11) Yoo, M.; Frank, C. W.; Mori, S.; Yamaguchi, S. Interaction of
Engineering, Drexel University, Philadelphia, Pennsylvania Poly(Vinylidene Fluoride) with Graphite Particles. 2. Effect of Solvent
Evaporation Kinetics and Chemical Properties of PVDF on the
19104, United States Surface Morphology of a Composite Film and Its Relation to
Samantha L. Morelly − Department of Chemical and Electrochemical Performance. Chem. Mater. 2004, 16, 1945−1953.
Biological Engineering, Drexel University, Philadelphia, (12) Nanjundaswamy, K. S.; Friend, H. D.; Kelly, C. O.; Standlee, D.
Pennsylvania 19104, United States J.; Higgins, R. L. Electrode Fabrication for Li-Ion: Processing,
Maureen H. Tang − Department of Chemical and Biological Formulations and Defects during Coating. IECEC-97 Proceedings of
Engineering, Drexel University, Philadelphia, Pennsylvania the Thirty-Second Intersociety Energy Conversion Engineering Conference
19104, United States; orcid.org/0000-0003-0037-4814 (Cat. No. 97CH6203), 1997; Vol. 1, pp 42−45.
(13) Hawley, W. B.; Li, J. Electrode Manufacturing for Lithium-Ion
Complete contact information is available at: BatteriesAnalysis of Current and next Generation Processing. J.
https://pubs.acs.org/10.1021/acsaem.0c01305 Energy Storage 2019, 25, 100862.
(14) Bitsch, B.; Dittmann, J.; Schmitt, M.; Scharfer, P.; Schabel, W.;
Author Contributions Willenbacher, N. A Novel Slurry Concept for the Fabrication of
R.M.S. performed all coating, drying, rheology, imaging and Lithium-Ion Battery Electrodes with Beneficial Properties. J. Power
battery testing measurements. S.L.M. fabricated the coin cells Sources 2014, 265, 81−90.
for battery testing. R.M.S. wrote the first draft of the (15) Lee, G.-W.; Ryu, J. H.; Han, W.; Ahn, K. H.; Oh, S. M. Effect of
manuscript. All authors contributed to manuscript edits. All Slurry Preparation Process on Electrochemical Performances of
authors have given approval to the final version of the LiCoO2 Composite Electrode. J. Power Sources 2010, 195, 6049−
manuscript. 6054.
(16) Susarla, N.; Ahmed, S.; Dees, D. W. Modeling and Analysis of
Funding Solvent Removal during Li-Ion Battery Electrode Drying. J. Power
NSF CBET-1929755. Sources 2018, 378, 660−670.
Notes (17) Muller, M.; Pfaffmann, L.; Jaiser, S.; Baunach, M.; Trouillet, V.;
The authors declare no competing financial interest. Scheiba, F.; Scharfer, P.; Schabel, W.; Bauer, W. Investigation of
■
Binder Distribution in Graphite Anodes for Lithium-Ion Batteries. J.
ACKNOWLEDGMENTS Power Sources 2017, 340, 1−5.
(18) Westphal, B. G.; Kwade, A. Critical Electrode Properties and
R.M.S. and N.J.A. would like to thank Cameron Abrams for Drying Conditions Causing Component Segregation in Graphitic
helpful discussions regarding the application of the radial Anodes for Lithium-Ion Batteries. J. Energy Storage 2018, 18, 509−
distribution function to EDS elemental maps.
■
517.
(19) Jaiser, S.; Muller, M.; Baunach, M.; Bauer, W.; Scharfer, P.;
REFERENCES Schabel, W. Investigation of Film Solidification and Binder Migration
(1) Morelly, S. L.; Alvarez, N. J.; Tang, M. H. Short-Range Contacts during Drying of Li-Ion Battery Anodes. J. Power Sources 2016, 318,
Govern the Performance of Industry-Relevant Battery Cathodes. J. 210−219.
Power Sources 2018, 387, 49−56. (20) Schmitt, M.; Scharfer, P.; Schabel, W. Slot Die Coating of
(2) Peterson, S. W.; Wheeler, D. R. Direct Measurements of Lithium-Ion Battery Electrodes: Investigations on Edge Effect Issues
Effective Electronic Transport in Porous Li-Ion Electrodes. J. for Stripe and Pattern Coatings. J. Coat. Technol. Res. 2014, 11, 57−
Electrochem. Soc. 2014, 161, A2175−A2181. 63.
(3) Bockholt, H.; Indrikova, M.; Netz, A.; Golks, F.; Kwade, A. The (21) Gutoff, E. B.; Cohen, E. D. Coating and Drying Defects:
Interaction of Consecutive Process Steps in the Manufacturing of Troubleshooting Operating Problems; John Wiley & Sons, 2006.
Lithium-Ion Battery Electrodes with Regard to Structural and (22) Bauer, W.; Notzel, D. Rheological Properties and Stability of
Electrochemical Properties. J. Power Sources 2016, 325, 140−151. NMP Based Cathode Slurries for Lithium Ion Batteries. Ceram. Int.
(4) Dominko, R.; Gaberšcě k, M.; Drofenik, J.; Bele, M.; Jamnik, J. 2014, 40, 4591−4598.
Influence of Carbon Black Distribution on Performance of Oxide (23) Russel, W. B.; Saville, D. A.; Schowalter, W. R. Colloidal
Cathodes for Li Ion Batteries. Electrochim. Acta 2003, 48, 3709−3716. Dispersions; Press, Cambridge, 1989 (accessed Apr 21, 2020).
(5) Dominko, R.; Gaberscek, M.; Drofenik, J.; Bele, M.; Pejovnik, S.; (24) Eberle, A. P. R.; Martys, N.; Porcar, L.; Kline, S. R.; George, W.
Jamnik, J. The Role of Carbon Black Distribution in Cathodes for Li L.; Kim, J. M.; Butler, P. D.; Wagner, N. J. Shear Viscosity and
Ion Batteries. J. Power Sources 2003, 119−121, 770−773. Structural Scalings in Model Adhesive Hard-Sphere Gels. Phys. Rev. E:
(6) Zheng, H.; Yang, R.; Liu, G.; Song, X.; Battaglia, V. S. Stat., Nonlinear, Soft Matter Phys. 2014, 89, 050302.
Cooperation between Active Material, Polymeric Binder and (25) Martys, N. S.; Khalil, M.; George, W. L.; Lootens, D.; Hébraud,
Conductive Carbon Additive in Lithium Ion Battery Cathode. J. P. Stress Propagation in a Concentrated Colloidal Suspension under
Phys. Chem. C 2012, 116, 4875−4882. Shear. Eur. Phys. J. E 2012, 35, 20.
(7) Cerbelaud, M.; Lestriez, B.; Videcoq, A.; Ferrando, R.; (26) Colombo, G.; Kim, S.; Schweizer, T.; Schroyen, B.; Clasen, C.;
Guyomard, D. Understanding the Structure of Electrodes in Li-Ion Mewis, J.; Vermant, J. Superposition Rheology and Anisotropy in
Rheological Properties of Sheared Colloidal Gels. J. Rheol. 2017, 61,

1035−1048.
(27) Varadan, P.; Solomon, M. J. Shear-Induced Microstructural
Evolution of a Thermoreversible Colloidal Gel. Langmuir 2001, 17,
2918−2929.
(28) Gordon, M. B.; Kloxin, C. J.; Wagner, N. J. The Rheology and
Microstructure of an Aging Thermoreversible Colloidal Gel. J. Rheol.
2017, 61, 23−34.
(29) López-Barrón, C. R.; Gurnon, A. K.; Eberle, A. P. R.; Porcar, L.;
Wagner, N. J. Microstructural Evolution of a Model, Shear-Banding
Micellar Solution during Shear Startup and Cessation. Phys. Rev. E:
Stat., Nonlinear, Soft Matter Phys. 2014, 89, 042301.
(30) Hipp, J. B.; Richards, J. J.; Wagner, N. J. Structure-Property
Relationships of Sheared Carbon Black Suspensions Determined by
Simultaneous Rheological and Neutron Scattering Measurements. J.
Rheol. 2019, 63, 423−436.
(31) Hoekstra, H.; Vermant, J.; Mewis, J.; Fuller, G. G. Flow-
Induced Anisotropy and Reversible Aggregation in Two-Dimensional
Suspensions. Langmuir 2003, 19, 9134−9141.
(32) Morelly, S. L.; Gelb, J.; Iacoviello, F.; Shearing, P. R.; Harris, S.
J.; Alvarez, N. J.; Tang, M. H. Three-Dimensional Visualization of
Conductive Domains in Battery Electrodes with Contrast-Enhancing
Nanoparticles. ACS Appl. Energy Mater. 2018, 1, 4479−4484.
(33) Shindo, D.; Oikawa, T. Energy Dispersive X-Ray Spectroscopy.
In Analytical Electron Microscopy for Materials Science; Shindo, D.,
Oikawa, T., Eds.; Springer Japan: Tokyo, 2002; pp 81−102.
(34) Allen, M. P.; Tildesley, D. J. Computer Simulation of Liquids
(Oxford Science Publications): 9780198556459: Amazon.com: Books
https://www.amazon.com/Computer-Simulation-Liquids-Science-
Publications/dp/0198556454 (accessed April 27, 2020).
(35) Brostow, W. Radial Distribution Function Peaks and
Coordination Numbers in Liquids and in Amorphous Solids. Chem.
Phys. Lett. 1977, 49, 285−288.
(36) Harper, R. A.; Posner, A. S. Radial Distribution Study of Non-
Crystalline Tricalcium Phosphate. Mater. Res. Bull. 1970, 5, 129−136.
(37) Srolovitz, D.; Egami, T.; Vitek, V. Radial Distribution Function
and Structural Relaxation in Amorphous Solids. Phys. Rev. B 1981, 24,
6936−6944.
(38) Li, F.; Lannin, J. S. Radial Distribution Function of Amorphous
Carbon. Phys. Rev. Lett. 1990, 65, 1905−1908.
(39) Root, L. J.; Lovett, R. On the Interpretation of Radial
Distribution Functions Determined from Integral Equations. J. Chem.
Phys. 1991, 95, 8390−8397.
(40) Morelly, S. L.; Tang, M. H.; Alvarez, N. J. The Impotence of
Non-Brownian Particles on the Gel Transition of Colloidal
Suspensions. Polymers 2017, 9, 461.

Drying

Uploaded by

Drying

Uploaded by

www.acsaem.

Correlating Processing Conditions to Short- and Long-Range Order

ACCESS Metrics & More Article Recommendations *

© 2020 American Chemical Society https://dx.doi.org/10.1021/acsaem.0c01305

rad/s at 0.1% strain measured 0, 1, and 2 h after constant shear rate

temperature. At this shear rate, the diﬀerences between

Rheological Properties of Sheared Colloidal Gels. J. Rheol. 2017, 61,

You might also like