Disentangling Electrode Slurry Behavior Under Process Conditions Using Rheo-Impedance Spectroscopy

Understanding the interplay between rheological and electrochemical properties of battery electrode slurries is essential for optimizing manufacturing processes, ensuring electrode performance, quality, and consistency. Conventional EIS provides valuable electrical insights but fails to capture dynamic and time-dependent changes under (process) representative shear conditions. Rheo-Impedance Spectroscopy (“Rheo-IS”), which integrates rheology with impedance spectroscopy, enables simultaneous monitoring of structural and conductive network evolution during processing and its time dependent recovery. This application note aims to establish good practice for applied Rheo-IS to give insight into battery slurries, offering a robust framework for predicting behaviour during coating and improving process control.

The performance and reliability of lithium-ion battery electrodes are strongly influenced by the microstructure and conductivity of their slurries during processing. These slurries are complex, multi-component systems where conductive additives, binders, and active materials interact dynamically under shear. Traditional electrochemical impedance spectroscopy (EIS) provides valuable insights into electrical properties, but it lacks the ability to capture changes under realistic processing conditions and neglects the impact on EIS from time-dependent processes.

Rheo-IS addresses this gap by enabling simultaneous measurement of rheological and electrochemical responses under controlled shear. This combined approach is critical for disentangling the evolution of conductive networks, particle interactions, and interfacial properties during mixing, coating, and subsequent recovery. By probing these behaviors in situ, Rheo-IS provides a pathway to optimize slurry formulation and processing strategies, ensuring consistent electrode quality and improved battery performance.

Model cathodes of the following generic formulation were systematically mixed using an ARE-250 small scale mixer by Thinky, as described in text:

NMC622 (by BASF, 97 – X%) : PVDF (Solef™ 5130, 3%): CB (Super P™ Li C65 from Imerys, Y%) : Carbon Nanotubes (TUBALL™ BATT 1% SWCNT, Z%) in NMP. Where X = Y + Z (total additive)

Rheo-IS measurements were made using the TA Instruments™ Discovery™ HR20 Rheometer with Rheo-Impedance Spectroscopy Accessory kit over a frequency range of 4 Hz – 8 MHz with an applied voltage of 0.1V. A higher 50 points per decade aided accurate fitting. Flow curve measurements were made using the steady state sensing algorithm with a maximum equilibration time (5% tolerance) of 180s, and a rest period as discussed in text of 10 minutes.

Fittings were made using the RelaxIS 3 software package. A proportional weighting was applied to derive each fit using the generic equivalent circuit model in Figure 1.

The Randles circuit remains the foundational model for EIS analysis, representing solution resistance (R_s) in series with a parallel combination of charge-transfer resistance (R_ct) and double-layer capacitance (C_dl), sometimes extended with a Warburg element for diffusion. While effective for simple electrode–electrolyte interfaces, this model struggles to capture the heterogeneous and multicomponent nature of battery slurries. One approach to address this is the use of a simplified, generic model comprising of a series of parallel resistance and CPE elements applicable across a wide range of formulations, as discussed by Hioki E.E. Corporation (“Hioki”) [1]. This work found that, to minimize overcomplex fitting, a lone CPE element improves this generic model by accounting for diffusion (Figure 1).

In depth equivalent circuits, for example, proposed by Wang et. al aim for high fidelity by incorporating multiple resistance and CPE branches to represent particle-particle contact, binder effects, and distributed diffusion [2,3]. While these complex models can achieve excellent fits, they risk introducing non-physical parameters that show poor transferability between slurry systems as shown in this work.

As Figure 2 demonstrates, both discussed equivalent circuit models can accurately fit the model cathode data from this work. The more complex, 10-component, equivalent circuit model was, however, fitted with both elements which converged to 0 (R_CM) and results with a high degree of error (> 25 %). ‘Error’ here refers to a range of the goodness-of-fit metric in which the result is insensitive to variations in the parameters, calculated using the RelaxIS 3 software package: a result attributed to the over-parametrization.

Although the quality of fit (low-deviation) for the simplified model (inset) is significantly improved as compared to the parameterised literature example when applied to this work, it should be considered during analysis that this is a result again of overparameterisation with non-physical elements, i.e. elements with no basis in physical reality. This compromise is inherent to the use of a generic equivalent circuit model; however, it remains possible to extract both relationships between individual processes and elements, as well as metrics based on overall resistance or capacitance metrics such as demonstrated for static slurry EIS by Hioki [1].

Figure 3. Repeatability of sampling was assessed by loading and analyzing a typical cathode slurry (NMC 622, PVDF, CB) three separate times, showing good agreement. Measurements remained consistent both at rest (as loaded) and under shear (100 s^-1, blue lines), following the methodology described in this work.

To establish meaningful relationships using Rheo-IS, the repeatability of testing in Figure 3 was investigated and found to be reliable both at rest (red lines) and under shear (blue lines). This is contingent on good rheometer practice. Figure 1 Using the generic equivalent circuit model, clear connections can be made between the methodology and formulation of Rheo-IS and the fitted parameters, allowing quantifiable metrics to be extracted. The effect of two example sensitivities is presented here, formulation (carbon additive concentration) and slurry solid content.

For both a reduction in CB fraction (Figure 4, left) and a reduction in overall solid fraction (Figure 4, right) the evolution of the associated resistance of each feature correlates clearly with the high-frequency feature (R₁) to the bulk resistivity, anticipated to be dominated by the free-carbon in the slurry known to form conductive tendril-like structures at rest, while the associated capacitance increases; these are presented in Figure 5 [4,5]. The amalgamated feature centred at 500 Ω, however, is stable across all fittings consistent with a tentative attribution to interfacial properties (e.g., contact resistance between particles, doublelayer capacitance).

Figure 5. The evolution of fitted parameters R₁, CPE₁, with the conductive additive formulation of a model NMC622 slurry

Figure 8 visualises the extracted (fitted) response of R₁ to both evolving CNT content and shear, i.e. simulated process conditions. Representing the conductive carbon network in the slurry, at lowshear, the increasing CNT content significantly reduces R₁; to a far greater extent than the same mass of carbon black. This structure is increasingly unstable with shear, however, and with increasing shear R₁ converges, informing the behaviour of this structure during the coating process.

Figure 8. Evolution of R₁ with shear rate extracted from the data as in Figure 7 for a range of CNT contents