Particle size plays an important role in the electrochemical performance of cathodes for lithium-ion (Li-ion) batteries. High energy planetary ball milling of LiNi1/3Mn1/3Co1/3O2 (NMC) cathode materials was investigated as a route to reduce the particle size and improve the electrochemical performance. The effect of ball milling times, milling speeds, and composition on the structure and properties of NMC cathodes was determined. X-ray diffraction analysis showed that ball milling decreased primary particle (crystallite) size by up to 29%, and the crystallite size was correlated with the milling time and milling speed. Using relatively mild milling conditions that provided an intermediate crystallite size, cathodes with higher capacities, improved rate capabilities, and improved capacity retention were obtained within 14 μm-thick electrode configurations. High milling speeds and long milling times not only resulted in smaller crystallite sizes but also lowered electrochemical performance. Beyond reduction in crystallite size, ball milling was found to increase the interfacial charge transfer resistance, lower the electrical conductivity, and produce aggregates that influenced performance. Computations support that electrolyte diffusivity within the cathode and film thickness play a significant role in the electrode performance. This study shows that cathodes with improved performance are obtained through use of mild ball milling conditions and appropriately designed electrodes that optimize the multiple transport phenomena involved in electrochemical charge storage materials.
Lithium-ion (Li-ion) batteries with improved performance, reduced cost, and improved safety are needed for numerous applications including electric vehicles, consumer devices, and grid-level energy storage [1,2], and significant efforts are on-going to develop improved electrodes and electrolytes. Cathodes based on Li–Ni–Mn–Co–O compositions are of significant interest based on their high capacities, operating voltages, and safety [3,4]. Based on its relatively high capacity and voltage compared with other established cathode materials (e.g., LiCoO2 and LiFePO4), the composition LiNi1/3Mn1/3Co1/3O2 (abbreviated as NMC) is of particular interest as a cathode for Li-ion batteries for electric vehicles, plug-in hybrid electric vehicles, and other applications . Efforts to improve the performance of NMC cathodes have included examining the effect of material composition , substitution , and microstructure including porosity, electrode composition, and electrode tortuosity via a combination of numerical simulation and/or experiments [5,7–13].
Cathodes that are able to be rapidly charged and discharged and that provide high capacities and voltages are of particular need for electric vehicles . Previous work has shown that the capacity and voltage of conventional NMC cathodes drop significantly at high discharge rates (e.g., ≥ 2C) . Electrochemical charge storage processes involve multiple, highly coupled transport phenomena and numerous factors influence the electrode capacity, voltage, and rate capabilities including the phase, electronic conductivity of the electrode particles, interparticle contact resistance, electrode–current collector contact resistance, solid-state ion diffusion within the electrode particle, electrolyte resistance (ionic conductivity), and interfacial resistance . Prior work reported that at high rates, Li-ion diffusion within the electrode is the rate-determining step for the discharge process .
Smaller particle sizes can reduce the Li-ion diffusion length and result in improved rate capabilities. The beneficial effect of smaller particle sizes has been shown using various cathode materials [3,16,17]. Small particle-sized Li4Ti5O12 was shown to provide improved performance at high discharge rates and low temperatures . Previous work showed that NMC cathodes with smaller particle sizes prepared using a combustion reaction resulted in improved high rate capabilities . Particle size also plays a role in the reversibility. Investigation of the effect of particle size on electrochemical performance of NMC showed that the largest crystal sizes displayed the most significant structural changes due to the lower strain accommodation . Methods that can reduce the particle size of NMC and other materials are of significant interest to improve the performance of cathode materials.
High energy ball milling has been shown to be an effective mechanochemical process to decrease particle size from the micron and nanometer-size domain for a wide range of energy-related materials [19,20]. Prior efforts have reported that high energy milling can reduce particle size  and improve the performance of LiMnPO4 , LiMn2O4 [23–26], and LiFePO4  cathodes. Previous work has investigated ball-milled NMC  and related compounds ; however, the effect of ball milling conditions on the structure and properties of NMC cathodes has not been previously reported. The conditions used for high energy ball milling are particularly important and can significantly affect material structure as supported by prior work on the effect of milling conditions on TiO2 nanoparticles . In this study, we investigated the effect of ball milling speed, time, and inclusion of carbon on the structure and properties of NMC cathodes. We report that milling conditions affect multiple material properties, which must be considered in order to obtain improved performance materials. Through the use of specific milling conditions and controlling electrode thickness, NMC cathodes with improved rate capabilities can be obtained.
LiNi1/3Mn1/3Co1/3O2 (NMC) was obtained from Targray Technology International, Inc. (Kirkland, Canada) (reported average particle size of 8–12 μm, Brunauer–Emmett–Teller (BET) surface area of 0.3–0.8 m2 g−1, and a density of 2.2 g cm−3). Carbon black (Super C-65) was obtained from TIMCAL (reported aggregate size < 1 μm, BET surface area of 62.4 m2 g−1, and a density of 160 kgm−3). Kynar Flex 2801 PVDF was supplied by Arkema, and 1.2 M LiPF6 in ethylene carbonate (EC)/ethyl methyl carbonate (EMC) (3:7 wt. %) was purchased from BASF. 1-Methyl-2-pyrrolidinone (NMP) with less than 0.005% H2O was purchased from Sigma-Aldrich. Lithium ribbon with a thickness of 0.75 mm was obtained from Sigma-Aldrich. Stainless steel CR2032 coin cell components (wave springs, cases, caps, 0.5 mm spacers, and gaskets) were acquired from Pred Materials.
High Energy Ball Milling.
The as-received NMC was ball milled with a Retsch PM 100 planetary ball mill using 50 mL zirconia jar containing the NMC material and either 5 or 10 mm zirconia balls according to details described in Table 1. Prior to ball milling, 10 g batches were loaded into the zirconia jar within an Ar-filled glovebox (O2 and H2O levels ≤ 1 ppm). The jars were placed into the glovebox antechamber and put under vacuum to seal the jar and prevent oxygen exposure during the milling process. After ball milling, the samples were removed from the milling jars and stored in the Ar-filled glovebox.
The as-received and ball-milled materials were analyzed with powder X-ray diffraction (XRD) using a D8-Focus Bragg-Brentano X-ray Powder Diffractometer, Bruker) using Cu Kα radiation and a step size of 2θ = 0.02. Crystallite size for the as-received and ball-milled materials was determined using the Scherrer equation, L = Kλ/(β cosθ), where L is the crystallite size, K is the Scherrer constant, λ is the X-ray wavelength, β is the line broadening at half the maximum intensity (FWHM), and θ is the Bragg angle. The full width at half maximum (FWHM) was determined from the experimental XRD pattern, and a Scherrer constant of 1.0 was used for the analysis. Scanning electron microscopy (SEM, Quanta 600 FEG, FEI Company) was used to characterize the sample morphology.
Electrode Fabrication and Coin Cell Preparation.
Electrode slurries consisting of 80% NMC (either as-received or ball-milled), 10% carbon black (Super C65, Timcal), 10% polyvinylidene difluoride (PVDF, Kynar 2801, Arkema), and NMP were mixed with an IKA ULTRA-TURRAX Tube Drive using the ST-20 tube and 16 5 mm glass beads until a uniform dispersion was obtained. The total dry weight of all components was 1 g. The NMC, carbon fraction, and NMP were mixed initially for 8 min followed by addition of the PVDF fraction (added as a 10% by weight PVDF in NMP solution). The complete mixture was then mixed for an additional 8 min.
Cathode slurries were cast onto 15 μm-thick Al foil (MTI Corp). Before use, the Al foil was cleaned with 1.0 M KOH in isopropanol using light circular scrubbing with a Scotch-Brite Heavy Duty Scour Pad to both remove trace oils and roughen the current collector to improve electrode adhesion. Cathode slurries were then cast onto the cleaned Al foil using a doctor blade with a gap setting of either 200 μm or 50 μm. Electrode sheets were allowed to dry under ambient conditions in a vent hood to allow the NMP to evaporate and were then dried for approximately 2 h at 70 °C. The thickness of the cast and dried electrode sheets was determined to be 25 ± 1 μm or 14 ± 1 μm for electrodes cast using a gap setting of either 200 μm or 50 μm, respectively. The average loading of the electrodes was 2.0 ± 0.1 mg cm−2 and 1.5 ± 0.2 mg cm−2 active material for 25 μm and 14 μm-thick electrodes, respectively.
From the cast electrode sheets, multiple 0.5 in. diameter cathode disks were punched out for each sheet to allow for repeatability testing. The cathode disks were weighed and dried overnight in a vacuum oven at 120 °C. The electrodes were then transferred immediately into an Ar-filled glovebox with O2 and H2O levels maintained below 1 ppm. Coin cells were prepared with the Ar-filled glovebox using the cathode disks, a Celgard 2500 separator membrane, and lithium foil counter/reference electrodes. Details of the electrode fabrication procedure and coin cell construction are furnished in our recent work .
Galvanostatic cycling was performed using an Arbin BT-2000 potentiostat/galvanostat. The cells were initially cycled five times between 4.2 and 2.8 V at a rate of 0.1C to allow for the initial solid electrolyte interface (SEI) formation (formation cycles). After the final formation charge and discharge, the electrochemical impedance spectra were obtained using a Biologic VMP3 EIS/potentiostat/galvanostat with a frequency range of 100 mHz to 1 MHz and 7.07 mV root mean square (RMS) sinusoidal perturbation voltage. The experimental impedance data were fit using Zfit software (version 10.31, Biologic) utilizing the equivalent circuit described in the text. After the initial five formation cycles and initial impedance data, the cells were cycled five times each at rates of 0.5C, 1C, and 2C, followed by 100 cycles at 1C.
Electronic Conductivity Measurements.
To obtain electronic conductivities of the as-received and ball-milled materials, approximately 200–400 mg of powder was placed in a cell (Pred Materials, HS Flat Cell) with gold-coated stainless steel contacts and a 5 kg force spring. To allow evaluation of the effect of carbon and binder on the electronic conductivity, composite samples were also prepared using 80 wt. % active (CM01 or CM02), 10 wt. % carbon black (Super C65, Timcal), and 10 wt. % PVDF, which is the same formulation used for the electrode materials. Two-point probe measurements were obtained using a constant voltage (±0.1 to 0.1 V was applied to the cell) using an Arbin BT-2000 potentiostat/galvanostat, and current was monitored until quasi-steady state was reached (∼3 min). A positive voltage was applied first, followed by a 3 min rest at open circuit potential. Then a negative voltage was applied. The current was determined from the quasi-steady state current. Resistance was determined using Ohm's law, The electrical conductivity, σelec, expressed in the units S cm−1 was obtained from the experimentally determined resistance, the area, and the measured thickness using the equation where l is the thickness of the sample in cm, R is the measured resistance in Ω, and A is the cross-sectional area of the electrodes in cm2.
Calculations based on the fully coupled electrochemical model of Gu and Wang  were utilized to determine the specific capacity and discharge voltage including various input parameters described in the text. Details on computations are provided in the Appendix.
Results and Discussion
Effect of High Energy Ball Milling Conditions on Structure.
The effect of four different milling conditions on NMC materials was evaluated: (i) milling time, (ii) milling speed, (iii) ball size/number of balls, and (iv) co-milling NMC with carbon. Shown in Table 1 are the material labels and the specific milling conditions.
Following the milling treatments, X-ray diffraction (XRD) and scanning electron microscopy (SEM) were used to determine the effect of different milling conditions on the long-range structure and microstructure. Powder XRD of the baseline and ball-milled samples was obtained to determine the effect of ball milling on long-range structure. Shown in Fig. 1 are XRD patterns for CM01, CM02, and CM08. All milled samples showed similar diffraction patterns, and for simplicity the XRD of the baseline NMC sample (CM01) and representative samples for milling under mild conditions (CM02) and under aggressive milling conditions (CM08) are presented in Fig. 1. The baseline NMC material (CM01) exhibited sharp diffraction peaks due to its highly crystalline structure. The diffraction lines can be indexed to a hexagonal lattice structure, crystallographic space group (space group no. 166) in an α-NaFeO2-type layered structure consisting of sheets of edge sharing MO6 octahedra (M = Ni, Mn, Co) separated by layers of lithium ions in octahedral interstitial sites [3,17,32,33]. As can be observed in Fig. 1, the milled CM02 and CM08 samples showed similar peaks and peak positions as the unmilled sample CM01, which indicates that the ball milling did not change the phase of the material. However, changes in the relative peak intensities and peak widths were observed for the milled samples compared with the baseline sample. The increasing peak width for the ball-milled samples can be seen from the exploded view of the 003 peak for CM01, CM02, and CM08 shown in Fig. 1. The width of the 003 peak was used to determine the primary particle (crystallite) sizes for the as-received and ball-milled samples using the Scherrer equation as presented in Table 1. Compared to the crystallite size of the baseline sample, high energy planetary ball milling was determined to significantly decrease the crystallite size for all ball-milled samples. The most significant decrease in crystallite size was for CM07 (29% decrease of crystal size compared with unmilled CM01). The crystallite size decreased from 94.8 ± 0.2 nm for the baseline CM01 to 68 ± 6 nm and 69 ± 6 nm for the samples milled at 650 rpm, CM07, and CM08, respectively. The correlation of the milling speed and milling time with the crystallite size is shown in Fig. 2. The data show that for the same milling time of either 30 min or 60 min, milling at a speed of 650 rpm resulted in significantly smaller crystallite sizes compared with milling at a speed of 350 rpm which is expected based on the higher force provided by the higher milling speed. For a milling speed of 350 rpm, the longer milling times (up to 120 min) resulted in smaller average crystallite sizes; however, the difference was within the margin of error for the experiment. For milling at 650 rpm, longer milling times resulted in similar average crystallite sizes within the margin of error. From the limited number of conditions evaluated in this study, neither changing ball size/number of balls nor co-mixing with carbon (with all other mixing conditions constant) resulted in substantial changes in the crystallite size.
Scanning electron microscopy (SEM) was used to determine the changes in microstructure from the ball milling process. SEM images of the as-received and ball-milled materials are presented in Fig. 3. The baseline material (CM01) consists of spherical particles with diameters in the range of 5–15 μm. To allow differentiation between the spherical particles observed in the SEM images and the crystallites, we use notation similar to recent reports [34,35] to notate the large spherical particles as secondary particles and the crystallites as primary particles. The spherical secondary particles in the unmilled CM01 sample are aggregates of smaller primary particles (crystallites) with sizes of 94.8 nm based on XRD analysis as discussed above.
The SEM images distinctly show that the milling conditions influence the relative concentration of spherical secondary particles. From the SEM analysis of the ball-milled samples, two notable trends are observed: (i) the break-up of spherical secondary particles and (ii) the formation of additional aggregate structures. The specific degree to which these occur depends on the specific milling conditions. From the SEM analysis, samples with higher milling times and/or speeds had a higher degree of break-up of spherical secondary particles. The sample with the most aggressive milling conditions, CM08, showed the most significant reduction of spherical secondary particles and the highest degree of additional aggregate particles. For lower intensity milling conditions, such as used for CM02, some spherical secondary particles remain among the smaller aggregates. The sample milled with carbon (CM04) showed formation of smaller aggregates as well as coating of the active material with carbon, as shown by the darker region surrounding the active material particle.
Effect of High Energy Ball Milling on Cell Performance.
Upon charge where the voltage increases, lithium ions and electrons are removed from the structure. During discharge, lithium ions and electrons are inserted into the structure, and the voltage decreases. Electrodes were fabricated and tested to determine the effect of ball milling conditions on the electrochemical performance. Our initial tests investigated electrodes with thicknesses of 25 μm. The discharge capacities at C/10, C/5, C/2, C and 2C rates, cathode-based specific energies, and capacity retention after 100 cycles at 1C for 25 μm-thick electrodes are presented in Table 2. Figure 4 shows discharge capacities at 1C and 2C rates for the baseline and ball-milled NMC materials. Despite the smaller crystallite sizes as determined from XRD analysis, the electrochemical charge and discharge of ball-milled samples using 25 μm-thick electrodes showed lower capacities and voltages for all milling conditions compared with the baseline. The data in Fig. 4(b) clearly show the lower performance of the mildly ball-milled CM02 sample compared with the baseline CM01 at 2C rates with 25 μm-thick electrodes.
Given the smaller crystallite size from XRD and the lower electrochemical performance, we postulated that ball milling may induce other effects that significantly affected the electrochemical performance. We first considered that the interface and/or electronic conductivity may have been altered by ball milling, and both of these factors may be related to the aggregate structures observed from SEM. Electrochemical impedance spectroscopy and electrical measurements were taken to evaluate the effect of ball milling on the interface and electronic conductivity, respectively.
Electrochemical Impedance Spectroscopy Analysis.
Based on the significant influence of the electrode–electrolyte interface on performance as supported by prior work , electrochemical impedance spectroscopy (EIS) was performed of the NMC cells to determine the effect of milling on the interfacial charge transfer resistance. Since the galvanostatic data of the 25 μm-thick electrodes (Fig. 4) showed that the CM02 sample with the least aggressive milling conditions (350 rpm, 30 min) performed the best of the milled materials, this sample was further analyzed with EIS and compared with the baseline CM01 sample. Shown in Fig. 5 are the complex (Nyquist) impedance plots of NMC cells containing either the as-received sample (CM01) or the ball-milled sample (CM02). The complex impedance plot shows two regions of primary interest: (i) the high-frequency semicircle and (ii) the low-frequency tail. The high-frequency semicircle is attributed to the interfacial charge-transfer processes, and the low-frequency feature is attributed to the solid-state diffusion processes of Li-ions in the bulk phase of the active material .
Based on the observed features, the equivalent circuit shown in Fig. 5 was used for fitting the experimental electrochemical impedance data. Similar equivalent circuits have been used to describe the impedance response of cathodes in Li-ion batteries . Within this equivalent circuit, Rs describes the series resistance, which includes contact resistances and the bulk electrolyte resistance. The high-frequency semicircle can be described by a resistor (R1) in parallel with a constant phase element (Q1). The resistance R1 is attributed to the charge transfer resistance  which includes the resistances associated with ion transport through the interfacial region of the NMC cathode. The capacitive element indicates charge polarization coupled with the charge-transfer process and is attributed to a double-layer capacitance. A constant phase element was used rather than a single capacitor to account for the distribution of capacitances, which may result from the distribution of particle sizes since there may be multiple parallel resistor–capacitor elements that may not be frequency resolved. The low-frequency region exhibited features similar to Warburg element (Zw) that describes semi-infinite diffusion. However, since the low-frequency region is not entirely a straight line but also shows some curvature, the low-frequency region was fit with a combination of Warburg element, a resistor (R2), and a constant phase element (Q2). From the experimental impedance data and the equivalent circuit described above, the values for RS, R1, Q1, R2, Q2, and Zw were obtained from fitting and are presented in Table 3. From the fitted parameters, R1, R2, and Zw were all higher for the milled sample CM02 compared with baseline unmilled sample CM01. The higher charge transfer resistance R1 for CM02 suggests that milling results in changes to the interfacial charge-transfer process. The higher interfacial resistance for CM02 obtained from electrochemical impedance data correlates with the lower voltages obtained from the galvanostatic discharge results (Fig. 4). It is possible that the ball-milled samples may have a higher interfacial resistance due to the aggregates formed that have a different surface structure related to an amorphous phase at the surface which would not be observed with XRD. The lower relative degree of crystallinity of the ball-milled samples is supported by the lower relative intensity of the X-ray diffraction peaks. Further analysis is needed to determine the specific nature of the higher interfacial resistance for the ball-milled materials.
Electronic Conductivity Measurements.
Based on the galvanostatic cycling data that showed lower voltage and capacities for the ball-milled materials in 25 μm-thick electrodes (Fig. 4), electrical conductivity measurements were obtained to determine if ball milling resulted in changes to the electronic conductivity of the materials. Shown in Table 4 are the measured direct current (DC) electronic conductivities, σelec of powders of the as-received, unmilled sample CM01, and milled samples CM02. The unmilled sample CM01 had a measured σelec of 9.5 ± 0.7 × 10−6 S cm−1 which in a similar range as the value of 2.75 × 10−7 S cm−1 reported for a NMC material prepared from a hydrothermal synthesis route . Compared with the electronic conductivity of the as-received unmilled sample, ball milling lowered the σelec of CM02 to 1.9 ± 0.6 × 10−7 S cm−1 as presented in Table 4. The comparison of the electronic conductivity data with the SEM data suggests that new aggregates formed by ball milling have higher resistances compared to the original spherical secondary particles. The higher resistance of the aggregates is consistent with prior work that showed that interconnections between particles significantly affect the electrical properties and significantly higher electronic resistances were obtained for particulate samples compared with interconnected particulate samples . It is possible that interactions between particles could be controlled through sintering or introduction of conductive elements between the particles to result in improved electronic properties for the ball-milled material. To determine the effect of carbon and binder on the electrical conductivity, tests were performed with a composite composed of the cathode material (CM01 or CM02), carbon, and binder with an identical mass ratio (80 wt. % active, 10 wt. % carbon, 10 wt. % binder) as used for the electrodes. As presented in Table 4, the measured σelec of the composite with carbon were 0.14 ± 0.04 S cm−1 and 0.08 ± 0.04 S cm−1 for the composite of CM01 and CM02, respectively. The average σelec of the composite with CM02 was lower than the value for CM01 but was within experimental error.
Electrochemical Performance of Thin Electrodes.
The effect of using thinner electrodes on the capacity of ball-milled NMC cathode materials was investigated since electrode thickness has been shown to significantly affect the electrode performance of NMC cathodes, particularly at high rates . Shown in Fig. 6 are the discharge capacities at 1C and 2C rates for the baseline and ball-milled NMC materials for 14 μm-thick electrodes. The data for discharge capacities at C/10, C/5, C/2, C and 2C rates, cathode-based specific energies, and capacity retention for 14 μm-thick electrodes are presented in Table 5. For the 14 μm-thick electrodes, the ball-milled CM02 sample exhibit higher specific capacities and voltages at 1C and 2C rates compared with the baseline sample CM01. Using the thinner electrodes, the ball-milled CM02 sample exhibited a 2C discharge capacity that was improved by 55% compared with the baseline material. The cathode-based specific energy was improved by over 60%, and the capacity rention (1C for 100 cycles) was increased to 98% (7.5% improvement) compared with the baseline CM01 sample.
Simulations to Elucidate the Effect of Transport Phenomena on Electrode Performance.
Based on our observation of improved performance for the ball-milled material within 14 μm-thick electrodes and lower performance with 25 μm-thick electrodes, we performed computational simulations to determine the effect of transport processes on the electrode performance. Simulations, relying on our recent and prior work on physicochemical transport in lithium-ion battery electrodes [40–44], were performed to determine the effect of electrode thickness, electronic resistance of the electrode (effective electronic conductivity), cathode secondary particle radius, solid-phase diffusivity, and electrolyte diffusivity (transport of Li-ions in the electrolyte within the cathode). The simulations followed the electrochemical model of Gu and Wang , and additional details of the computational model and parameters are provided in the Appendix. Thicknesses of 25 μm and 14 μm were considered to provide agreement with our experimentally measured values. The experimentally measured electrical conductivity values for the composite electrode of 0.14 ± 0.04 and 0.08 ± 0.04 S cm−1 (Table 4) were used as simulation input. The secondary particle size values were based on observations from the SEM images (Fig. 3) and additional representative values were adopted to simulate smaller secondary particle sizes for the ball-milled materials. Values for the solid-phase diffusivity were obtained from measured diffusion coefficients for NMC cathodes . A representative electrolyte diffusivity was adopted from the literature [46,47] as an intrinsic value, which was further qualified to account for the pore-space tortuosity in the simulations.
As shown in Table 6, the effect of five different parameters on cathode capacity and voltage was evaluated and for each parameter, other factors were maintained constant. The first three parameters evaluated, electronic conductivity, secondary particle size, and solid-phase diffusivity, were not found to significantly affect the cathode capacity and voltage within the range of values tested. In contrast, electrolyte diffusivity and thickness were found to significantly affect the electrode performance. The effect of electrolyte diffusivity and electrode thickness on the calculated specific discharge capacity and voltage is shown in Fig. 7. The results of the simulations are consistent with prior work that supported that the lower capacity of NMC cathodes at higher rates was mainly due to Li-ion diffusion within the electrode . The transport of Li-ions within the electrode is affected by the electrode porosity, the tortuosity, and the electrolyte diffusivity and transference number. Reducing the electrode thickness reduces the Li-ion diffusion distance in the liquid phase within the electrode pores.
The calculations directly support our experimental observations that showed improved capacities and voltages for the ball-milled sample using thinner electrodes. Our simulations revealed that the electrolyte diffusivity within the ball-milled cathode may be a predominant limiting mechanism, which is affected by the combined transport resistance offered by the intra- and interaggregate pore network, which is an important microstructural attribute in the NMC electrodes. Due to the reduced diffusion distance of Li-ions within the thinner electrode, the beneficial effect of smaller crystallite size from the ball milling results in improved cell performance. The simulations provide a qualitative basis for explaining the experimental observations that electrode thickness significantly impacts the performance of ball-milled material. Further work will be aimed at taking into account the effects of interfacial resistance, electrolyte diffusivity, and active material aggregation to further improve the agreement between theoretical models and experiments.
The ability to control primary particle (crystallite) size provides a route to improve the electrochemical performance of Li-ion battery cathodes. High energy planetary ball milling LiMn1/3Ni1/3Co1/3O2 cathode materials was determined to result in smaller crystallite sizes with crystallite size being directly correlated to milling time and milling speed. In addition to reducing crystallite size, high energy ball milling was found to result in the formation of secondary aggregate structures. Electrochemical impedance spectroscopy and electronic conductivity tests showed that both the interfacial charge transfer resistance and the electronic conductivity of the material were altered by the ball milling process. High energy ball milling was determined to lower the electronic conductivity of NMC which is attributed to the higher interparticle resistance within the aggregate structures. Using a ball-milled material with intermediate crystallite size achieved using mild milling conditions, NMC cathodes with higher capacities, higher voltages, and improved capacity retention compared to the base material were achieved within 14 μm-thick electrodes. Computations supported that electrode performance is significantly affected by electrolyte diffusivity and thickness and provided a basis for the experimental observations of improved performance obtained for ball-milled NMC using thinner electrodes.
The electrochemical charge storage processes involve multiple, highly coupled transport phenomena that influence the overall electrode performance including electronic conductivity within and between particles, solid-state ion diffusion, interfacial charge transfer, and diffusion of Li-ions within the cathode. This study showed that high energy ball milling impacts multiple material properties beyond the crystallite size, and through understanding and control of the local structures in the electrode and their influence on the underlying charge transport characteristics, improved performance electrodes can be achieved.
CPR gratefully acknowledges funding from the National Science Foundation (NSF) PREM (Grant No. DMR-1205670) for support of this research. PPM gratefully acknowledges financial support from NSF Grant No. 1438431 to Texas A&M University.
Appendix: Computational Modeling and Simulations
The specific parameters used for the electrochemical and material properties are described in Table 7.