3.1. Characterization of Slurries and Electrodes
Since the pH of the slurry has a huge impact on the process, the pH evolution was carefully monitored to stay below the recommended coating limit of 8.6 [
5] to suppress possible corrosion of the Al substrate. This is also important with regard to the emergence of agglomerations within the slurry reported to occur when the pH of 10 is exceeded in the slurry mixing process as described above in
Section 2.1. Viscosity measurements showed a shear thinning behavior of 18
at zero shear rate and 4
at the shear rate applied during the coating process. Porosity and areal capacity were calculated based on the measured thicknesses and mass loadings in SL and ML electrodes. The values of the measured parameters are given in
Table 1. Coating thicknesses are slightly higher than 200
for both types of electrodes. The as-coated porosity values of the SL and ML samples are calculated to be almost identical with each other, attaining values of
and 52.6% for the SL and ML electrodes, respectively. Both coatings show the same areal capacity of
cm
−2, which was calculated assuming a 200 mA h g
−1 specific capacity for the NMC811 active material. The lower layer constitutes a 43% share of the overall bi-layer electrode thickness. On average, a porosity of 59.3% was calculated for the base layer with an areal capacity of
mA h cm
−2.
Investigating the surface of each coating, it is evident that a continuous film without defects of any kind was fabricated (
Figure 1a,c). Even before calendaring, a smooth homogeneous surface was present in both samples. This is a huge improvement considering the high thickness and aqueous processing of the electrode [
18,
22].
Analysis of cross sectional images taken via digital microscope are shown in
Figure 1b,d for single- and multi-layer electrodes, respectively. Both samples are determined to have a thickness of 190
after calendaring to 40% porosity, verifying the measurements via a thickness gauge. Slightly inhomogeneous distribution of lighter gray particles is noticeable for SL samples, but overall no severe irregularities, such as entrapped air, cracks, or delamination from the substrate are recognizable. No distinct transition line between top and bottom layer is visible for the multi-layer sample, which is evidence of a good mechanical connection of both layers. SEM images of the multi-layer sample are displayed in
Figure 2. Merging of the layers is visible in agreement with the analysis via digital microscope. Images of higher magnification reflect an intertwining behavior of the secondary NMC811 particles of the top and bottom layers, which is necessary for strong cohesion. The indistinguishability of the two layers can also be accredited to the slight “dissolution” of the surface of the bottom layer into the freshly applied electrode slurry of the upper layer during coating. Furthermore, the surface film of the prime coating absorbs the slurry of the second coating forming a diffuse transition layer. This is a key aspect in guaranteeing good mechanical linkage of both layers.
The significantly higher porosity of the lower layer also implies either a change in porosity of the layer itself due to the coating of the upper layer, or a porosity gradient with remarkably higher porosity in the vicinity of the current collector. Assuming no porosity transition of the base after the second coating, the porosity of the top layer should be 47.5%. According to both Qi et al. [
30] and Fang et al. [
31] this lower porosity of the top layer would lead to a decreased cell performance as they suggest an electrode design with an opposite porosity gradient (higher porosity at the top of the electrode with decreasing porosity in the direction of the current collector) to reduce electrode resistance and enhance performance at high C-rates by increasing Li-ion diffusivity. However, the higher observed porosity of the upper layer, as well as the observed bubbles which develop after the second coating is applied, imply that cavities inside the base layer are filled with applied slurry.
Figure 3 illustrates this mechanism. First, the slurry of the second coating comes in contact with the low porosity base layer (
Figure 3a). As the water evaporates, porosity decreases in the top layer (
Figure 3b) and the particles from the top layer migrate, and occupy the cavities in the bottom layer. This causes a mixing of both coatings (
Figure 3c) and results in an increase in porosity of the initial layer (
Figure 3d). This influences the porosity gradient to generate lower porosity values near the substrate foil and higher porosity values closer to the coating surface.
The results of the peel testing measurements confirm significant improvement following the multi-layer approach. In comparison to SL samples with ∼45 Nm
−1, ML coatings show values of ∼66 Nm
−1 (
Figure 4). Therefore, on account of the second coating, an increase of approximately 45% in adhesion can be achieved. The reason for this vast enhancement is due to more homogeneously distributed binder particles, which is enhanced by the multi-layer processing.
3.2. Electrochemical Performance of Cells
Rate capability tests were performed to assess the electrochemical performance of both cell configurations. Results of statistically significant cells are displayed for clear visualization (
Figure 5a). Formation cycles are not presented in the graph. It is worth mentioning, that after preconditioning at C/10 for 5 cycles, the C-rate was adjusted according to the measured cell discharge capacity of cycle number 3 (these preconditioning cycles are labels as C/10* in
Figure 5a). Therefore, all cathodes are exposed to the same C-rate independent of the actual specific capacity of the cathode after the preconditioning cycles. In accordance, the ML cells show higher specific discharge capacities throughout all the tests, even though the same C-rates are applied. Especially for low current densities, cells with ML cathodes outperform the SL samples by over 20%, which is a remarkable enhancement considering that both electrodes are fabricated using the same materials. Therefore, changing the coating procedure has a positive impact on material distribution inside the coating. The beneficial effect of multi-layering is even more pronounced for lower C-rates. For thick electrodes, limiting effects on Li
+ diffusion are evident at high current densities, and resulting losses in discharge capacity can not be compensated fully by multi-layering. Therefore the difference in capacity is slightly less distinct for 1C. Approximate improvements of the discharge capacities are listed in
Table 2. When returning back to C/10, both cell configurations show initial high capacities, implying that cycling at high rates is not detrimental to their stability. Long term cycling tests show no significant decrease in specific discharge capacity after 50 cycles at C/5 for SL and ML electrodes (
Figure 5b). Moreover, the coulombic efficiency for both SL and ML electrodes is determined to be close to 100%.
Specific capacities of representative charge and discharge cycles are selected for comparison at each C-rate and are plotted in
Figure 6 against applied voltages. The specific discharge capacity data as a function of C-rate are also given in
Table 3. SL cells show values of 146 mA h g
−1, 128 mA h g
−1 and 59 mA h g
−1 for C/5, C/2, and 1C, respectively. ML cells have the overall highest specific discharge capacity of 171 mA h g
−1 at C/5, whereas at C/2 and 1C, capacities of 136 mA h g
−1 and 65 mA h g
−1 are measured. Substantial differences in discharge capacity between the two types of electrodes are present for low current densities. For single layers only the region close to the electrode/separator interface is electrochemically active, since Li ion diffusion lengths are not sufficient enough to reach underlying areas [
39]. However, in ML electrodes, areas located closer to the current collector foil also contribute to specific discharge capacity.
Figure 6 shows the voltage-capacity curves of the SL and ML electrodes at C-rates of C/5 (a), C/2 (b) and 1C (c). All of them show a sloping profile, which is characteristic for NMC cathode active materials. In all cases, the charge and discharge capacities of the cells with the ML electrodes are larger than those for the SL electrodes, despite the fact that all electrodes have the same mass of active material. At high current densities, the benefits of multi-layering on Li-ion diffusion compensate for the drawbacks accompanied by electrode thickness itself [
40]. Li-ion mobility definitely poses the most difficulties for thick electrodes to compete at high C-rates.
The potentiostatic contribution to the specific charge capacity is represented by the plateau at
in
Figure 6. Its length is directly proportional to the resistance inside the electrode during charge at constant voltage. The presence of multiple layers causes additional interfacial resistances within ML coated electrodes. Especially at C/5 (
Figure 6a), a distinct difference in potentiostatic specific charge capacity is noticeable. At low C-rate in particular, a great part of the capacity is reached due to a elongated constant voltage step during charge. Nevertheless, it is worth mentioning that at C/5 higher specific charge capacities are already reached before the constant voltage step takes place.
The cyclic voltammograms shown in
Figure 7 were recorded to compare the reduction and oxidation reactions of SL and ML electrodes during cycling. Both electrode types show no peak at 3
, indicating the absence of Mn
3+ [
41]. Two pairs of oxidation/reduction peaks were observed for each sample. The small reduction peaks for SL and ML at around
are not visible in the CV during oxidation due to overlapping with the lower voltage reduction peaks.
Table 4 displays the potentials of oxidation/reduction peaks and corresponding polarizations for each electrode type. SL cells show oxidation peaks at V
ox1 =
and V
ox2 =
and reduction peaks at V
red1 =
and V
red2 =
, with a polarization of
V
1 =
and
V
1 =
respectively. The polarizations for ML cells are
V
1 =
for peaks at V
ox1 =
and V
red1 =
and
V
1 =
for V
ox2 =
and V
red2 =
. Comparing both electrode types, no significant differences in peak position and polarization are noticeable. However, ML samples show a larger area underneath the curve, which is in direct proportion to the capacity of the cell and is in accordance with the capacity difference shown in the rate capability test mentioned above. Besides the higher capacities, no obvious difference of the reaction kinetics were observed via CV measurements.
EIS measurements were performed to help describe the transfer phenomena inside the cells during the formation cycles. The equivalent circuit used for the fitting is displayed in
Figure 8. R
e corresponds to the bulk resistance resulting from the cell components (current collector, separator) and the electrolyte. The resistance contribution from the solid electrolyte interphase (SEI) formed on the graphite anode is fitted using R
SEI and a constant phase element Q
SEI, expressing its behavior as the non ideal capacitor. The charge transfer resistance R
ct and the double-layer capacitance represented by Q
dl show the contribution of charge transfer behaviour between the electrolyte and the electrode. The diffusion at low frequencies is represented by the Warburg element (W).
Figure 8 shows Nyquist plots of SL and ML cells during the first and second cycle at
. The intercept of Re(Z) at the high frequency region shows similar values in R
e for both samples, as shown in
Table 5 and
Figure 9a. This is expected, since all components apart from the cathodes are identical for each cell. The high to medium frequency semicircle shows the contribution of SEI formed during the first cycle on the anode side. At different stages of charge and discharge for the SL and ML electrodes the fitted data is comparable, highlighting the fact that the developed cathode in this work has no negative impact on the development of the SEI and its stability.
The mid to low frequency semicircle illustrates the charge transfer process, where a significant difference was observed between the charge transfer resistance of SL and ML electrodes. The higher specific discharge capacity, as shown in
Figure 6, for the multi-layer cells, is attributed to the lower R
ct value. Heubner et al. [
42] investigated the influence of electrode porosity and thickness on cell performance, with respect to limiting processes. They conclude that among other effects, decreased porosity leads to an increase in charge transfer resistance. Available areas for charge transfer reactions are reduced due to an increase in contacts between particles and particles with current collector foil and a reduction in the total contact area with the particles and the electrolyte [
42]. The pore size distribution in the ML electrode facilitates access of Li-ions from the electrolyte in the whole multi-layer electrode and thus lowering the R
ct.
Figure 9 displays the acquired data for both electrode types and shows that a lower R
ct is achieved and a stable performance is established throughout cycling. The short inclined line at low frequencies represents the Li diffusion into the active material. Similar values for SL and ML samples lead to the conclusion that multi-layering does not negatively influence the diffusion of the Li-ions in the electrode and the mass transport controlled region of the cathode.
The advantage of better adhesion in ML samples is not expected to be fully realized in coin cell configurations, since the high pressure inside the cells suppresses the delamination. Therefore, we anticipate a more pronounced benefit in pouch-cell configurations. Considering all electrochemical results, more homogeneous binder distribution and the resulting increase in charge transfer leads to higher discharge capacities in ML cells. In addition, better electrochemical performance roots from a porosity gradient generated inside the cathode—as suggested in literature [
30,
31].