Thermally Conductive Separator With Hierarchical Nano/microstructures For Improving Thermal Management of Batteries
Thermally Conductive Separator With Hierarchical Nano/microstructures For Improving Thermal Management of Batteries
Thermally Conductive Separator With Hierarchical Nano/microstructures For Improving Thermal Management of Batteries
FULL PAPER
a
Department of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, MA 02139,
United States
b
Materials Science and Engineering, Department of Applied Physics and Applied Mathematics, Columbia
University, New York, NY 10027, United States
Received 23 October 2015; received in revised form 20 January 2016; accepted 28 January 2016
Available online 21 February 2016
KEYWORDS Abstract
Thermal conduction; Thermal management is critical to improving battery performance and suppressing thermal
Batteries; runaway. Besides developing external cooling technologies, it is important to understand and
Nanocomposite control thermal transport inside batteries. In this paper, heat transfer inside batteries is first
analyzed and the thermal conductivity of each component is measured. The results show that
low thermal conductivity of the separator is one major barrier for heat transfer in Li-ion
batteries. To improve thermal conductivity of the separator, a hierarchical nano/micro-Al2O3/
polymer separator is prepared with thermal conductivity of 1 W m 1 K 1, representing an
enhancement of 5 compared to commercial polyethylene-based separators. Modeling has
been performed to understand mechanism behind the enhancement of thermal conductivity,
which suggests that addition of nanoparticles significantly reduces thickness of polymer coating
on micron-sized Al2O3 particles and thus increase the thermal conductivity of the composite
separator. This Al2O3-based separator also has similar ionic conductivity with commercial
polymer separators. Such composite separator may have potential applications in developing
batteries with better performance and safety.
& 2016 Elsevier Ltd. All rights reserved.
n
Corresponding author at: Department of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, MA 02139,
United States.
nn
Corresponding author.
E-mail addresses: yy2664@columbia.edu (Y. Yang), gchen2@mit.edu (G. Chen).
http://dx.doi.org/10.1016/j.nanoen.2016.01.026
2211-2855/& 2016 Elsevier Ltd. All rights reserved.
302 Y. Yang et al.
High-performance batteries are important for various appli- The thermal conductivity is measured by stacking multiple
cations ranging from portable electronics, electric vehicles electrode films together to ensure that thermal resistance
and grid-level energy storage [1,2]. Thermal management of sample is one order of magnitude higher than the thermal
of state-of-the-art Li-ion batteries (LIBs) and future higher contact resistance between copper and sample. After the
energy batteries is critical to their performance and safety, environmental temperature is stabilized, the thermoelec-
especially at large scale [3–6]. The high temperature tric plate cools the bottom side of the sample while the
significantly deteriorates cycle life and it is one important heater is used to keep temperature of the top surface the
reason to trigger thermal runaway, especially for batteries same as the environment. The heater power (Q) is recorded
with high energy and power density [7–10]. Past efforts after temperature is stabilized, indicated as the average
mainly focus on modeling of external cooling technologies, power over 30 s after stabilization. Temperature data is
such as forced air and liquid cooling, to lower battery recorded by K-type thermocouple. More details can be
temperature [5,11–19], where the lumped heat transfer found in Section 2 of the supporting information.
model is widely used without considering thermal conduc-
tivity (k) of a battery itself [15–18]. Only a few references Ionic conductivity measurement
took battery thermal conductivity into account with
assumed values [5,11,19]. Improving thermal transport Ionic conductivity is measured by sandwiching the separator
inside batteries can also facilitate heat dissipation, reduce between two pieces of stainless steel with the same size and
temperature inhomogeneity and thermal stress in batteries. applying an AC voltage with amplitude of 10 mV at 50 kHz.
In this paper, we first measured thermal conductivity of The electrolyte is 1 M LiClO4 in Ethylene carbonate/Diethyl
different components in batteries and identified that the carbonate (EC:DEC) with weight ratio of 1:1.
battery separator is a major limiting factor for heat
dissipation in batteries. Then a thermally conductive
COMSOL simulation
Al2O3/polymer composite separator was developed to
improve heat dissipation in batteries. The Al2O3/polymer
In simulation of temperature rise, external heat transfer
hybrid separator contains both micron-sized and nano-sized
coefficient is assumed to be 1000 W m 1 K 1 and 20 W m 1
Al2O3 particles as the thermally conductive phase, and Poly
K 1 for forced liquid cooling and force air cooling, respec-
(vinylidene fluoride-hexafluoropropylene) (PVdF-HFP) as the
tively. The voltage loss due to internal resistance is sup-
binder. The large surface area of nanoparticles reduces
posed to be 0.6 V. The simulation time for 3 C rate is 1200 s.
thickness of polymer coating on Al2O3 and enhances thermal
The capacity of 18650 cell is set to 3.1 Ah. Liquid cooling is
conductivity of the separator. At the optimized materials
applied to all surfaces. For prismatic cells, the volumetric
loading ratio, a thermal conductivity of 1.0570.16 W m 1
energy density is set to 600 W h L 1 for the all cells. More
K 1 is reached, which is more than five times that of
details can be found in Section 4 in the supporting
commercial polyethylene/polypropylene (PE/PP)-based
information.
commercial separator. Such separators could improve heat
conduction and reduce the temperature rise of batteries in
operations. Meanwhile, the composite separator shows Results and discussions
similar ionic conductivity as commercial polyethylene-
based separators, avoiding excessive joule heating due to A typical Li-ion battery is made up of a triple-layer structure
slower ionic transport across the separator. with a porous separator sandwiched between two composite
electrodes (Figure 1a). The porous separator is typically
made of polypropylene and polyethylene [20], while two
Materials and methods electrodes are mixtures of active materials (e.g. LiCoO2 or
graphite, 80–95 wt%), carbon black (2–10 wt%) and poly-
Materials meric binder (1–10 wt%) (See Figure S1 for SEM images of
electrodes and separators). The triple-layer structure is
Micro-Al2O3 particles ( 10 μm) were purchased from Sigma assembled together to form either a roll in a cylindrical cell
Aldrich. Kynar 2801 PVdF-HFP was received from Arkema. or a cuboid in a prismatic cell (Figure 1b). Voids in
Commercial single-side coated LiCoO2 and graphite electro- electrodes and separators are filled with carbonate-based
des, separators and nano-Al2O3 particles ( 100 nm) were organic electrolyte. In battery operation, heat is generated
purchased from MTI Corporation. The Al2O3/PVdF-HFP com- throughout the cell due to joule heat and entropy change in
posite separator was prepared by dispersing micro-Al2O3 electrochemical processes [21]. The heat produced is first
particles, nano-Al2O3 particles and PVdF-HFP binders in THF conducted inside the cell through both in-plane and cross-
for 12 h and drop casting onto a flat glass substrate. A mold plane directions of the triple-layer structure, followed by
made of Aluminum is used to confine the dispersion and dissipation process outside the cell, such as air/liquid
shape the separator to the desired dimension. The typical convection (Figure 1b). The in-plane direction of the
sample size is 3 cm by 3 cm. The weight ratio of PVdF-HFP triple-layer structure corresponds to the axial direction in
to THF is fixed as 1:11. The ratio of PVdF-HFP to Al2O3 varies cylindrical cells and width/length directions in prismatic
from 20: 80 to 10: 90. After drying, the Al2O3/PVdF-HFP film cells with a thermal conductivity of k== , while the cross-
is peeled off from the glass substrate and pressed under plane direction corresponds to the radial direction in
0.1 ton for 5 min at room temperature. cylindrical cells and thickness direction in prismatic cells
Thermally conductive separator with hierarchical nano/microstructures 303
Figure 1 The structure of common Li-ion batteries and heat dissipation. (a) A typical Li-ion battery consists of three layers:
composite cathode (e.g. LiCoO2) with Al as the substrate, porous separator, and composite anode (e.g. graphite) with Cu as the
substrate. The cross-plane heat transport is mainly limited by thermal resistance of the separator layer due to its low thermal
conductivity. (b) Heat transfer in both in-plane and cross-plane directions of the triple-layer structure inside batteries, and then
dissipate through convection outside batteries.
with a thermal conductivity of k┴ . To understand limiting discussed in our previous work [23] (Figure 2a), which uses
factors for heat dissipation in batteries, thermal conductiv- heat flux applied to the sample, temperature difference
ity of battery components is measured first, followed by across the sample and sample geometry to derive k┴ based
estimation of thermal resistance of different directions and on the Fourier’s law. To reduce the effect of contact
external convection. thermal resistance, multiple layers of battery electrodes
The first step in our analysis is to obtain thermal or separators are stacked together so that the thermal
conductivity of electrodes and separators by experimental resistance of sample ( 1 10 3 m2 K W 1) is much larger
measurements. In the past, Maleki et al used laser flash than contact resistance between copper plates and the
method to measure k┴ and k== of electrodes and the triple- sample ( 1 10 4 m2 K W 1). To mimic a real battery, we
layer structure at different state-of-charge [22]. Their focus on electrodes and separators saturated with diethyl
results show that k┴ of electrodes and trilayer are 3 W carbonate (DEC), one major component in LIB electrolyte.
m 1 K 1 when saturated with electrolyte, while k== is 20– No salt is added since it contributes little to thermal
30 W m 1 K 1. However, a single-layer model is applied to conductivity [24] and it is sensitive to moisture. k┴ of
multi-layer samples with distinct thermal properties in their LiCoO2 electrode, graphite electrode and separators satu-
study. In this report, k┴ of electrodes and separators are rated with DEC are measured to be 1.0670.16, 2.070.3,
measured by a differential steady state method, as and 0.1970.03 W m 1 K 1 (Table 1), which are in the same
304 Y. Yang et al.
Figure 2 Measurements and simulations of heat transfer in Li-ion batteries. a) A schematic of the differential steady state method
to measure thermal conductivity. A thermoelectric (TE) cooler cools the bottom side of a sample with a thickness of Δx while a thin
film heater keeps top surface at the same temperature as the environment, which minimizes heat loss to the environment and
improves accuracy of measured heat flux through the sample. Copper plates are used to realize uniform temperature distribution at
two ends of the sample, which are measured by thermocouples (TCs). b) and c) Calculated thermal resistances of b) the cylindrical
cell configuration and c) the prismatic cell configuration. The cylindrical cell dimension is scaled from an 18650 cell. The definition
of directions in b) and c) is the same as Figure 1b. Symbols in parenthesis indicate corresponding thermal conductivity in the
direction. d) Numerical simulation by COMSOL Multiphysics on temperature rise against thermal conductivity of separator in a four-
prismatic-cell pack and a 18650 cell, respectively. e) Corresponding temperature distribution in the cross section of an 18650 cell
with the same simulation described in (d). The cross section passes the central axis of the cylindrical cell.
temperature distribution helps reduce performance degra- resistance between particles, which is typically 10 4 10 5
dation due to thermal stress. Further enhancement of m2 K W 1, while the effective thermal resistance of an
k┴;separator beyond 1 W m 1 K 1 does not significantly reduce individual Al2O3 particle is much less, in the order of
Trise or improve temperature homogeneity, as thermal 10 μm / 35 W m 1 K 1 = 3 10 7 m2 K W 1. However,
resistance of electrodes begins to dominate. Therefore, higher thermal conductivity is still observed in samples with
1 W m 1 K 1 is set as the target for thermally conductive large portion of Al2O3, suggesting the effectiveness of
separators. However, common polymers only have k of 0.2– adding Al2O3 particles in enhancing thermal conductivity.
0.5 W m 1 K 1 [15,28]; thus pure polymer-based separators After the composite film is saturated with DEC, its effective
are difficult to achieve k of 1 W m 1 K 1, especially along thermal conductivity (keff ) dramatically increases to
the cross-plane direction. To address this challenge, we 0.7170.11, 0.8370.12, 1.1370.17 W m 1 K 1, for micro-
develop an inorganic/organic hybrid separator to realize Al2O3/PVdF-HFP ratio of 80:20, 85:15 and 90:10, respec-
high cross-plane thermal conductivity, where the inorganic tively, representing an enhancement of 4-5 X compared to
phase, such as Al2O3, provides pathway for efficient heat dry samples (Figure 3d). The value is also 4-5X higher than
transfer, as Al2O3 is of low cost and has high thermal commercial separators saturated with DEC as measured
conductivity of 35 W m 1 K 1 [29], and the polymeric above (0.185 W m 1 K 1). Although the sample with 90%
phase acts as binder to maintain integrity of the separator Al2O3 meets the target of 1 W m 1 K 1, the low content of
(Figure 3a). The high portion of nonflammable Al2O3 in the PVdF-HFP inside indicates that the film may have a poor
separator also helps reduce risks of thermal runaway. mechanical strength, especially under swelling condition
The separator is prepared by dispersing 10 μm Al2O3 with electrolyte presented. Ideally separators should have
particles (Sigma-Aldrich) and Poly(vinylidene fluoride-hexa- both high thermal conductivity and reasonable polymer
fluoropropylene) (PVdF-HFP) binders (Kynar 2801) in tetra- content to provide enough mechanical strength.
hydrofuran (THF) for 12 h and drop casting onto a flat glass To increase thermal conductivity of the separator while
substrate. The weight ratio of PVdF-HFP to Al2O3 varies from maintaining a reasonable polymer content, we first try to
1: 4 to 1: 9, corresponding to 1:1.8 to 1:4.1 in volume ratio. understand why thermal conductivity of the composite
After drying, the Al2O3/PVdF-HFP film is peeled off from the separator (1 W m 1 K 1) is much lower than Al2O3 itself
glass substrate and pressed under 0.1 ton for 5 min. ( 35 W m 1 K 1). The heat transfer in the composite
Figure 3b and c shows optical and SEM images of a typical electrode is simplified as two polymer-wrapped Al2O3 sphe-
sample. The size of granular Al2O3 particle is 10 μm and it rical particles in contact with each other. COMSOL simula-
is wrapped by PVdF-HFP. The film thickness can be con- tion shows that the thickness of polymer coating layer and
trolled between 30 μm and 200 μm by adjusting volume of thermal conductivity of the liquid phase are two dominant
dispersion dropped onto the glass substrate. Density data factors to impede heat transfer (Figure 4b and c), while
indicate that 50–60% of the film is filled by Al2O3 and
polymer, and the portion left is void (Table 2). The separator
shows a reasonable ionic conductivity of 0.64–0.75 mS Table 3 Properties of nano/micro-Al2O3/PVdF-HFP
cm 1, slightly lower than our measurements (0.88 mS composite separator.
cm 1) and previous results [30] of commercial separators.
We will show later that extra joule heating due to this Micro-Al2O3:nano-Al2O3 65:20 70:15 75:10 80:5 85:0
slightly lower conductivity has negligible impact on the cell (weight)
temperature rise. Density (g cm 3) 1.70 1.74 1.85 1.88 1.93
The thermal conductivity of a dry composite separator is Volumetric portion of 0.51 0.52 0.55 0.59 0.6
quite low, which are 0.1770.03, 0.2270.03 and solid in the separator
0.2670.04 W m 1 K 1 for samples with 80%, 85% and 90% Ionic conductivity 0.74 0.69 0.75 0.67 0.68
of micro-Al2O3 (Figure 3d), respectively, as the thermal (mS cm 1)
transport is mainly limited by the thermal contact
Figure 4 Thermally conductive separator with both nano- and micro-Al2O3. a) A conceptual schematic of adding Al2O3
nanoparticles to improve thermal conductivity. The addition of nanoparticles reduces thickness of polymer coating on micro-
Al2O3 and enhances thermal conductivity of the matrix. b) and c) Dependence of keff on b) the thickness of PVdF-HFP coating on
Al2O3 particles, and c) k of electrolyte.
Thermally conductive separator with hierarchical nano/microstructures 307
Figure 6 Stability of Al2O3/polymer separator against lithium metal. The composite separator is attached to lithium and soaked in
1 M LiPF6 in EC/DEC for seven days. (a) XRD pattern before and after contacting with Li. (b) Optical image of a composite separator
before and after contacting with Li. No change has been observed. It is supposed that polymer coating effectively block direct
reaction between lithium and Al2O3.
micro-Al2O3 is larger than that in uniform coating, which the polymer coating layer on Al2O3 avoids the direct contact
gives a lower keff than the prediction. The higher keff between Al2O3 and lithium. As a result, the degree of
predicted by the model also implies that further optimiza- reaction is very limited and it should have little impact on
tion of particle dispersion and polymer coating may boost thermal conductivity of the whole composite. Further
keff to 2 W m 1 K 1. investigation is needed to evaluate the long term stability
A concern in applying Al2O3/polymer separator to enhance of such composite separator.
heat dissipation is that whether the lower ionic conductivity
of Al2O3/polymer separator increases heat generation in
batteries and thus compensates its higher thermal conduc-
tivity. Tables 2 and 3 show that the ionic conductivity of Conclusion
Al2O3/polymer is 0.2 mS cm 1 lower than commercial
separators. Let us use 18650 cell as an example. The typical In summary, k┴ of battery separator is important to thermal
thickness and size of separator in a 18650 cell are 25 μm and management of batteries. A nano/micro-Al2O3/PVdF-HFP-
6 50 cm2, respectively, it is estimated that the extra over- based composite separator is developed with high thermal
potential due to lower ionic conductivity is 25 mV at 3 C conductivity ( 1 W m 1 K 1), which is 5 times that of
rate, which is only 4% of the total overpotential of 0.6 V common PE-based separators. The addition of Al2O3 nano-
and has little effect on heat generation. COMSOL simulations particles helps further enhance thermal conductivity of the
show that this extra overpotential only causes an extra composite. The mechanism is assumed to be that Al2O3
temperature rise of 0.27 K, which is only 12% of the nanoparticles help reduce thickness of polymer coating on
temperature reduction due to enhanced thermal conductiv- micro-Al2O3 and improve effective thermal conductivity of
ity of separator (2.16 K). Similarly, this extra overpotential the electrolyte. Such thermally conductive separator could
results in an extra temperature rise of 1.1 K for four- have potential applications to dissipate heat faster in
prismatic-cell pack discussed above, which corresponds to batteries and reduce temperature rise in operation, espe-
only 8.5% of the temperature reduction due to enhanced cially under external forced liquid cooling.
thermal conductivity of separator (12.9 K). Moreover, as the
porosity of Al2O3/polymer separators is similar to that of
commercial ones, we believe that ionic conductivity of such
composite separators can be further improved. Another Acknowledgment
concern is the mechanical strength of the separator when
it is soaked in electrolyte, as the organic electrolyte in Li-ion We would like to thank Daniel Kraemer for developing the
batteries could swell the separator. This issue could be differential steady-state method to measure thermal con-
addressed by replacing PVdF-HFP with other polymers with ductivity and helpful discussions. This material is based
better mechanical properties, such as polyvinyl alcohol. upon work supported as part of the Solid State Solar-
The stability of such composite separator is also tested by Thermal Energy Conversion Center (S3TEC), an Energy
attaching it to a lithium metal and soak in electrolyte (1 M Frontier Research Center funded by the U.S. Department
LiPF6 in EC/DEC) for seven days. XRD data do not show any of Energy, Office of Science, Office of Basic Energy Sciences
new peaks, and there is no obvious change in camera images under Award number DE-SC0001299/DE-FG02-09ER46577.
(Figure 6). These results indicate that no or only trace Yuan Yang thanks support from startup funding from Colum-
amount of Al2O3 has reacted with lithium. We believe that bia University (Grant no. UR007667).
Thermally conductive separator with hierarchical nano/microstructures 309
Appendix A. Supplementary material [30] Y.M. Lee, J.E. Seo, N.S. Choi, J.K. Park, Electrochimica Acta 50
(2005) 2843–2848.
Supplementary data associated with this article can be [31] G. Liu, H. Zheng, X. Song, V.S. Battaglia, J. Electrochem. Soc.
found in the online version at http://dx.doi.org/10.1016/ 159 (2012) A214–A221.
[32] D.A.G. Bruggeman, Ann. Physik 24 (1935) 636–664.
j.nanoen.2016.01.026.
[33] X.G. Jin, J.T. Wu, Z.G. Liu, J. Pan, Fluid Phase Equilibria 220
(2004) 37–40.