A Bilayer Cathode Design Procedure for Li ion Batteries Using the Multilayer Doyle-Fuller-Newman Model (M-DFN)

Abstract

Heterogeneities in lithium ion batteries can be significant factors in electrode under utilisation and degradation while charging. Bilayer electrodes have been proposed as a convenient and scalable way to homogenise the electrode response. In this paper, the design of a bilayer cathode for Li-ion batteries composed of separate layers of lithium nickel manganese cobalt oxide (NMC622) and lithium iron phosphate (LFP) is optimised using the multilayer Doyle-Fuller-Newman (M-DFN) model. Changes to the carbon binder domain, electrolyte volume fraction, and tortuosity provided the greatest control for improving Li-ion charge mobility. The optimised bilayer design was able to charge at 3C between 0-90% SOC in 18.6 minutes, achieving 4.4 mAh/cm2. Comparing the optimal bilayer to the LFP-only electrode, the bilayer achieved 41% higher capacity. Through mechanistic physics-based modelling, it was shown that the 3C charging improvement of the optimised bilayer was achieved by enabling a more homogeneous current density distribution through the thickness of the electrode and electrolyte depletion prevention. The findings were confirmed on a high-fidelity X-ray computed tomography (CT) based microstructural model. The results illustrate how modelling can be used to rapidly search novel electrode designs

Figures (21)

• Figure 1: (a) Plasma focused ion beam/scanning electron microscope (PFIB/SEM) back scattered electron (BSE) image of a bilayer positive electrode for Li ion half-cell comprising a Li(NiMnCo)O2 (NMC) sub-layer adjacent to the separator, and a LiFePO4 (LFP) sub-layer next to the Al current collector (CC), with a scale bar of 50 $\mu$m. (b) The M-DFN model domain, comprising Li counter electrode (CE), separator (s), positive electrode sub-layer p1 (NMC in this case), positive electrode sub-layer p2 (LFP in this case), and current collector (CC). There are also spherical particles at each point through the thickness of the electrode for the additional pseudo-dimension (not shown). The model includes three distinct compartments including (i) separator, (ii) positive one, and (iii) positive two. Reproduced with permissionTredenick2024multilayer.
• Figure 2: 3C charge results with a benchmark comparison including identical specific capacity at 0.05C ($\sim$3.74 mAh/cm$^2$). Conventional half cell electrodes of NMC-only and LFP-only are compared to the bilayer, for voltage as a function of normalised capacity retention (capacity/specific capacity at 0.05C) (a). The capacity retention at 4.2 V is re-plotted in (b). Parameters are shown in Table \ref{['VariablesModel']} and \ref{['BilNMCLFP']} and the experimental data is included from previous workTredenick2024multilayer. The CT model results for the bilayer are also shown, fit to the experimental data and parameters are shown in Table \ref{['optimaltable4']}.
• Figure 3: 3C charge sensitivity analysis or benchmark comparison with parameters described in Table \ref{['TestsSA3C_2']} and the default case 4 shown in black, where p1 is NMC and p2 is LFP.
• Figure 4: The effect of the Bruggeman tortuosity factor ($b_{ \text{p2}}$) of the LFP sub-layer (a) and porosity ($\varepsilon_{ \text{p}}$) (b) on C rate charge performance in terms of capacity retention at 4.2 V. (b) is not a benchmark comparison and the simulation parameters of case 1 are 5.7 mAh and 3.7 mAh/cm$^2$; case 3 are 5.9 mAh and 3.8 mAh/cm$^2$; and case 4 are 6.1 mAh and 4 mAh/cm$^2$. In (b) the two sub-layers of the bilayer have the same porosity except in case 2 for the default.
• Figure 5: 3C charge with a range of cathode thicknesses. We consider both the capacity (a,b) and capacity retention (c,d) as the specific capacity at 0.05C varies greatly with changes in thickness. The new candidate optimal parameters are shown in Table \ref{['optimaltable2']}. Case 2 is the default, case 4 is the new candidate optimal design and thickness changes are shown in Table \ref{['Thickness3C']}. Additional thicknesses are shown in Fig. \ref{['fig:thickagain']}.