Ye Shui
Zhang
*abc,
Josh J.
Bailey
abd,
Yige
Sun
be,
Adam M.
Boyce
ab,
Will
Dawson
ab,
Carl D.
Reynolds
bf,
Zhenyu
Zhang
ab,
Xuekun
Lu
ab,
Patrick
Grant
be,
Emma
Kendrick
bf,
Paul R.
Shearing
ab and
Dan J. L.
Brett
*ab
aElectrochemical Innovation Lab, Department of Chemical Engineering, University College London, London, WC1E 7JE, UK. E-mail: d.brett@ucl.ac.uk; yeshui.zhang@ucl.ac.uk
bThe Faraday Institution, Quad One, Harwell Science and Innovation Campus, Didcot, OX11 0RA, UK
cSchool of Engineering, University of Aberdeen, Aberdeen, AB24 3UE, UK
dSchool of Mechanical and Aerospace Engineering, Queen's University Belfast, Belfast, BT9 5AH, UK
eDepartment of Materials, University of Oxford, Parks Road, Oxford, OX1 3PH, UK
fSchool of Metallurgy and Materials, University of Birmingham, Birmingham, B15 2TT, UK
First published on 12th April 2022
The performance of lithium-ion batteries is determined by the architecture and properties of electrodes formed during manufacturing, particularly in the drying process when solvent is removed and the electrode structure is formed. Temperature is one of the most dominant parameters that influences the process, and therefore a comparison of temperature effects on both NMC622-based cathodes (PVDF-based binder) and graphite-based anodes (water-based binder) dried at RT, 60, 80, 100 and 120 °C has been undertaken. X-ray computed tomography showed that NMC622 particles concentrated at the surface of the cathode coating except when dried at 60 °C. However, anodes showed similar graphite distributions at all temperatures. The discharge capacities for the cathodes dried at 60, 80, 100 and 120 °C displayed the following trend: 60 °C < 80 °C < 100 °C < 120 °C as C-rate was increased which was consistent with the trends found in adhesion testing between 60 and 120 °C. Focused-ion beam scanning electrode microscopy and energy-dispersive X-ray spectroscopy suggested that the F-rich binder distribution was largely insensitive to temperature for cathodes. In contrast, conductivity enhancing fine carbon agglomerated on the upper surface of the active NMC particles in the cathode as temperature increased. The cathode dried at RT had the highest adhesion force of 0.015 N mm−1 and the best electrochemical rate performance. Conversely, drying temperature had no significant effect on the electrochemical performance of the anode, which was consistent with only a relatively small change in the adhesion, related to the use of lower adhesion water-based binders.
Fig. 1 The three-stage drying mechanism. (a) Stage 1 is from slurry phase to form a semi-slurry, (b) Stage 2 follows, with further removal of solvent and (c) Stage 2 ends with a compacted solid coating, shown in (d). (Yellow strings indicate the binder, pink particles indicate active material particles, black dots indicate the conductive carbon and light blue colour indicates the solvent.) Adapted with permission from ref. 6. Copyright {2021} American Chemical Society. |
Temperature is known to play an important role in the DP as a key parameter that influences drying rate. For example, high temperatures lead to binder migration (typically to the upper free surface), reducing the adhesive strength between the coating layer and the current collector (CC). This can result in the delamination of the coating from the CC, electrode shrinkage and coating component segregation;7–10 which in turn increases the internal resistance of the electrode through poor adhesion and cohesion properties7,11 and decreases cell capacity.12
Recent modelling work by Lombardo et al.13 applied a physics-based three-dimensional model to mimic additive migration during drying; this simulated the effect of drying rate on the final electrode mesostructure and the dynamics of additive migration. Tsotsas and Mujumdar14 described different types of modelling: continuum-level models based on volume-averaging, pore-network models, continuous thermomechanical models and computational fluid dynamics. Defraeye et al.15 considered convective heat and mass transfer modelling at air-porous material interfaces. Iqbal et al.16 used a coupled electrochemical–mechanical and cohesive-zone finite-element model to study mechanical failure at the interface between graphite particles and polyvinylidene fluoride (PVDF) binder in LIBs. Tirumkudulu et al.17 and Yow et al.18,19 studied the critical stress of cracking in drying latex films.
Although modelling studies have led to new insights into electrode DP, experimental investigations are required for definitive analysis and to validate such models. Zhang et al.2 recently reviewed the most up-to-date metrologies which have been applied, or have the potential to be applied, to study the DP of LIBs. For example, scanning electron microscopy (SEM) with energy-dispersive X-ray spectroscopy (SEM-EDS) has become a powerful technique to investigate the binder migration behaviour of LIB electrodes occurring during the DP. Jaiser et al.20 not only investigated microstructure formation in LIB electrodes during drying, but also component gradients for anode coating cross-sections at different stages of drying using EDS under cryogenic conditions. The liquid phase binder and solvent diffusion were both implicated in the binder gradients of dried electrodes by “dragging” the binder to the surface with capillary forces.20 Westphal et al.3 used SEM-EDS to detect fluorine concentration to indicate binder distribution inside an electrode coating. EDS was used to map the fluorine distribution through the electrode coating layer, showing that the binder was more concentrated at the top layer of the electrode coating. The binder distribution was also investigated within an electrode made with water-based binders, such as styrene butadiene rubber (SBR) and carboxymethyl cellulose (CMC). Sodium is the typical element in CMC which is used to map these binder distributions by EDS.21,22
To investigate the porosity and pore size distribution (PSD) of the electrode microstructures futher, multi-scale X-ray computed tomography (CT) has been applied to characterize the electrodes.23–28 Westhoff et al. created a framework for stochastic 3D modelling of the microstructure of electrodes and the structures were validated by 3D tomography.23 Danner et al.24 employed synchrotron X-ray tomography to create 3D microstructure models of electrodes for battery performance. Ebner et al.28 used synchrotron X-ray tomographic microscopy to statistically characterize the microstructure of transition metal oxide-based electrodes. The authors introduced their segmentation algorithm to be able to identify each of the particles and evaluate the PSD; the calculated PSD was consistent with the PSD obtained by experimental work by laser diffraction. Rahe et al.26 used nanoscale X-ray CT to investigate the structural change in aged automotive LIB cells; the CT images showed the morphology of the electrodes in terms of their internal porous structure. The porosity of aged anodes was reduced due to organic residues and depositions that were quantified and visualized by CT imaging and analysis. Mapping the 3D microstructure of electrodes, rather than 2D cross-sections only, is recognized as increasingly important for understanding any structural heterogeneity and its corresponding effects on battery performance. X-ray CT has also been applied in situ to determine microstructure evolution during battery degradation.4,29–31
Overall, the drying of wet electrode films is complex and known to critically influence final physico-chemical performance, but the drying dynamics are poorly understood. To better manipulate electrode structure and properties in the DP, it is useful to correlate and/or integrate multiple drying analysis methodologies to understand better the effect of each parameter. In this paper we undertake a comprehensive study of the effect of temperature on the drying of LIB electrodes (both NMC622-based cathodes with PVDF-based binder and graphite-based anodes with water-based binder, which are both commonly used in LIBs). The investigation involves several advanced characterization methods, including X-ray CT for analysing electrode active material (AM) distributions; adhesion testing of the electrode coating on CC using the axial capability of a rheometer; a focused-ion beam/scanning electrode microscopy with energy-dispersive X-ray spectroscopy (FIB-SEM-EDS) to study binder distribution in electrodes; and electrochemical analysis of electrodes dried at different temperatures. The findings provide a deeper understanding of the temperature effects during the DP of LIB electrodes, which will hopefully benefit and shorten process optimization for LIB manufacturing.
Areal fraction plots as a function of distance from the CC to the surface of the coating illustrating through-plane inhomogeneity for cathodes dried at different temperatures are shown in Fig. 3(a), (c), (e), (g) and (i). Each electrode was scanned three times for three different samples to improve the statistical robustness of the results, displaying similar AM distributions across the repeated scans. Corresponding volume renderings of the segmented electrode active particles dried at different temperatures are shown in Fig. 3(b), (d), (f), (h) and (j). The different colours of the round particles indicate the size of the particles. The particle size distribution is dominated by the NMC622 particles which will not be affected by the drying process. The average areal fractions of AM as a function of distance from the CC for cathodes dried at different temperatures are also plotted, as shown in Fig. 3(k). In Fig. 3(a), (c), (e), (g) and (i), the results show the different behavior of AM distributions for cathodes dried at different temperatures. For all samples except 60 °C, more NMC622 particles accumulated at the coating surface, as shown in Fig. 3(a), (e), (g) and (i). However, in Fig. 3(a), (c) and (e), there is an accumulation of NMC622 particle fractions near the CC for cathodes dried at RT, 60 and 80 °C. Conversely, in Fig. 3(g) and (i), there is a relatively even NMC622 particle distribution near the CC for cathode dried at 100 and 120 °C.
Fig. 3 Areal fraction plots as a function of distance from the CC, illustrating through-plane inhomogeneity for cathodes dried at different temperatures: (a) RT, (c) 60 °C, (e) 80 °C, (j) 100 °C and (i) 120 °C; corresponding volume renderings of the electrode active particles dried at (b) RT, (d) 60 °C, (f) 80 °C, (h) 100 °C and (i) 120 °C, where the colour scale bar indicates the different size of the AM particles; (k) average areal fraction of AM as a function of distance from the CC for cathodes dried at different drying temperatures (error bar is in Fig. S1†). |
Reconstructed 2D orthoslices and 3D volume renderings of the anodes dried at RT, 60, 80, 100 and 120 °C are shown in Fig. 4(a), (c), (e), (g) and (i). The graphite particles are clearly visible, where their irregular shape is light grey, and the pores and binder occupy the gaps between graphite particles. Cubic volume renderings of each anode are shown in Fig. 4(b), (d), (f), (h) and (j), illustrating the 3D structure of the anode coatings. As with the cathodes, the graphite particles (AM) were readily segmented versus the background due to higher X-ray attenuation versus pore space and CBD.
As with the triplicate results for cathodes shown in Fig. 3, the AM distribution derived from X-ray CT scans of the cathodes showed high repeatability. The areal fraction of anode AM (graphite) as a function of distance from the CC for anodes dried at different temperatures is plotted in Fig. 4(k). The results show similar behavior for graphite distributions for anode dried at RT, 60, 80, 100 and 120 °C, with no significant difference in AM distributions as a function of temperature. The graphite distribution apparently undulates slightly through the coating, which may due to be the irregular shape of the graphite particles. The results correlate with the electrochemical performance of the anode half-cells, as shown in Fig. 8, with no significant change in capacity. This finding is supported by Li and Wang38 who investigated binder concentration distributions in dried water-based and organic-based (PVDF-based) binder LiCoO2 electrode sheets and the physical, electrical and electrochemical properties of the electrodes. They showed the organic-based binder electrode had a non-uniform distribution of the electrode components, with greater heterogeneity than water-based binder electrode, which may arise because the ‘time-for-segregation’ for the anode is shorter than cathode drying.38 The anodes in this work were prepared with a water-based binder and the cathode was prepared with organic-based binder using NMP as the solvent. NMP has a much higher boiling point (202 °C) than water such that it takes a longer time to be fully removed by evaporation. Consequently, the reduced drying time for anodes leads to a reduced time-for-segregation, thus providing an even distribution of graphite at all investigated temperatures.
Fig. 7 Specific discharge capacity at different C-rates (C/10, C/5, C/3, C, 2C and 3C) of cathodes dried at different temperatures (RT, 60, 80, 100 and 120 °C). |
Fig. 8 Specific discharge capacity at different C-rates (C/10, C/5, C/3, C, 2C and 3C) of anodes dried at different temperatures (RT, 60, 80, 100 and 120 °C). |
The anodes dried at 80 and 100 °C gave only slightly higher adhesion force per millimeter compared with anodes dried at other temperatures. This is again consistent with Westphal et al.,3 who reported that adhesion strength is independent of mass loading at a relatively low temperature (∼80 °C) as the driving force for solvent evaporation is small. However, high temperatures have been reported to potentially lead to greater binder migration (as seen for the NMC cathodes earlier), which can cause anode delamination and result in high resistance.7–10 Electrode adhesion is strongly dependent upon the drying temperature and higher temperatures result in lower adhesion strength between the copper CC and the coating layer.7,8,39 However, the thickness of the coatings also needs to be considered, and in the case of graphite,[13] the adhesion strength can also decrease as the drying temperature increases from 80 to 110 °C with mass loadings up to 8.1 g cm−2. In these cases, the temperature played a minor role in electrode adhesion, which was more sensitive to thickness and mass loading. The adhesion results are consistent with the EDS mapping of elemental sodium across the electrode cross-sections, as shown in Fig. 6(u)–(y), which are proxy elements for the binder. Sodium is distributed comparatively evenly across the anodes, regardless of drying temperature. This may be explained by a shorter time-to-fixation, which is indicative of faster polymer entanglement, and subsequently, there is less time for binder segregation. It cannot be ignored that the binder concentration in anodes is very low which leads to less concentrated sodium maps, possibly requiring further investigation. Although higher temperatures will cause faster convection currents towards the surface as solvent is removed, the polymer matrix will also become more rapidly entangled as solvent is lost, resisting this motion, and the two competing effects approximately cancel.
Comparing the adhesion forces for both cathodes and anodes, it is obvious that the adhesion forces in cathodes are much higher than in anodes; for example, the average adhesion forces per millimetre for cathodes and anodes dried at 80 °C are 0.0034 and 0.0024 N mm−1, respectively. This finding is consistent with the literature that states that organic-based binders provide higher adhesion force than water-based binders.32,38 The inferior wetting of aqueous slurries on the CC leads to higher surface tension and lower adhesion strength, potentially reducing the cycle life of the cell.40–42 Moreover, electrodes made from water-based solvents can lead to more cracking without optimum drying protocols.42–45
In Fig. 7, and Table S3 in ESI,† the discharge performance of the cathode dried at RT has the highest specific capacity of 150.06 mA h g−1 at C/10. There are similar findings at other C-rates: cathodes dried at RT have the highest specific discharge capacities at all C-rates, compared with cathodes dried at 60, 80, 100 and 120 °C. The relatively slow drying rate at RT tends to promote a more homogeneous CBD distribution, which will enhance the electrochemical performance of the electrode. The assumption is supported by the EDS mapping of carbon and fluorine, as shown in Fig. 6 (f, k) RT, (g, l) 60, (h, m) 80, (i, n) 100 and (j, o) 120 °C, where the fluorine distribution is not significantly affected by drying temperature. While carbon tends to accumulate more on the top of the AM particles as the temperature increases (yellow circles), this segregation is more localised in length-scale, and needs firmer quantification. While there was no significant difference in cathode binder distribution according to EDS at the different temperatures, as shown in Fig. 6(k)–(o), adhesion is affected: the cathode dried at RT has the highest capacity and the highest adhesion of 0.015 N mm−1. The increased adhesion is beneficial both when assembling the cells (less likely to crack/delaminate and incur failure), and also less likely to delaminate during cycling causing poor cycle life.
As shown in Fig. 7, cathodes dried at RT showed consistently higher discharge capacities as a function of C-rate, and the discharge capacity for the cathode dried at 60, 80, 100 and 120 °C presented a trend of 60 °C < 80 °C < 100 °C < 120 °C as C-rate increases. The results are consistent with the adhesion test results (Fig. 5(a)), in that the adhesion force for cathodes dried at 60, 80, 100 and 120 °C display a similar trend of 60 °C < 80 °C < 100 °C < 120 °C. As shown in Fig. 8, the specific capacity of the anodes dried at different temperatures do not display a clear trend or major difference. The results are consistent with the adhesion force test as shown in Fig. 5(b) that the adhesion forces for anodes dried at different temperatures are similar. The results are also supported by the similar distribution of AM, as plotted in Fig. 4(h), although it may be harder to identify differences in anode adhesion due to the generally much smaller values.
X-ray CT results show diverse behavior of AM distributions for cathodes dried at different temperatures, with active NMC622 particles tending to concentrate more at the free surface of the electrode coating as drying temperature increases. Meanwhile, NMC622 particles also concentrated near the current collector when dried at relatively low temperatures of RT, 60 or 80 °C. Conversely, there is an even NMC622 particle distribution near the CC for cathode dried at 100 and 120 °C. This correlates well with the electrochemical results obtained, in that at low C-rates there is not much difference, but as C-rate increases there is a trend of 60 °C < 80 °C < 100 °C < 120 °C for discharge capacity which is consistent with the trend in adhesion force between 60 to 120 °C.
X-ray CT shows similar graphite distributions for anodes dried at each temperature, with no significant trend apparent in areal fraction as a function of distance from the CC.
Although there was no obvious difference in binder distribution obtained by FIB-SEM-EDS fluorine mapping, the EDS mapping of carbon distributions for cathodes dried at different temperatures were performed in a different way showing that carbon accumulated more on the top of the AM particles as the temperature increased.
This work also showed that cathodes dried at RT had the highest adhesion force of 0.015 N mm−1, and that the specific capacity for these cathodes were higher than those dried at higher temperatures for all C-rates.
Comparing the specific capacity for the electrode (cathode and anodes) at different C-rates, the results show that the specific discharge capacity was highest at the lowest C-rate of C/10 for both cathodes and anodes, regardless of the drying temperature.
Comparing the specific capacity for the electrode (cathodes and anodes) dried at different temperatures, the cathode dried at RT had the highest specific capacity at all C-rates.
Conversely, the anode drying temperature did not play a dominant role in determining the electrochemical performance of the anodes, such that the specific capacity of the anodes dried at different temperatures presented no clear trend. These results corresponded well with the adhesion force testing and the distribution of graphite particles.
Footnote |
† Electronic supplementary information (ESI) available: Supporting data for adhesion force tests and electrochemical testing results. See https://doi.org/10.1039/d2ta00861k |
This journal is © The Royal Society of Chemistry 2022 |