The present study rigorously develops continuum thermal models of electrochemical capacitors (ECs) accounting for the dominant interfacial and transport phenomena. It also aims to identify design rules and modeling tools to define safe modes of operation and to develop appropriate thermal management strategies. ECs are promising electrical energy storage devices, particularly for providing high power or long cycle life. They can be divided into two categories, namely electric double layer capacitors (EDLCs) storing charge electrostatically in the electric double layer (EDL) at the electrode/electrolyte interface and pseudocapacitors using both EDL and chemical charge storage. Unfortunately, ECs generate heat during operation due to a variety of interfacial and transport phenomena. Consequently, they may experience substantial changes in temperature, leading to problems such as accelerated aging and increased self-discharge rates. EC charge storage mechanisms involve complex multiphysics and multiscale transport phenomena and this complexity has impeded the physical understanding of EC heating. This study derives rigorous, physics-based continuum models for both EDLCs and pseudocapacitors from first principles. Then, detailed numerical simulations were performed to investigate characteristic thermal behavior, to physically interpret experimental measurements from the literature, and to develop design rules.
First, thermal models were developed for EDLCs. The heat diffusion equation and associated heat generation rates were derived from first principles and coupled with the transient electrodiffusion of ions in binary and symmetric electrolyte. Irreversible Joule heating and reversible heat generation rates due to ion diffusion, steric effects, and changes in entropy of mixing in the electrolyte were formulated. The predicted temperature rise for planar EDLCs qualitatively reproduced experimental data from the literature under various charging/discharging conditions. Scaling analysis simplified this model from twelve independent design parameters to seven dimensionless similarity parameters. Scaling laws were developed for the heat generated during a charging step and for the maximum temperature oscillations under galvanostatic cycling. In addition, a first-order thermal analysis for EDLCs was developed based on the lumped-capacitance approximation and accounting for both irreversible and reversible heating. A simple analytical expression for the overall temperature rise during galvanostatic cycling was derived and scaled. This simple thermal model enables rapid estimation of temperature evolution in EDLCs without computationally intensive numerical simulations and was quantitatively validated with experimental measurements from commercial EDLC devices. Moreover, the first-principles thermal model was generalized to account for multiple ion species and/or asymmetric electrolytes. Simulations with binary and asymmetric electrolytes indicated that the irreversible Joule heating decreased with increasing valency and/or diffusion coefficient of either ion while the local reversible heating near a given electrode increased with increasing counterion valency and/or decreasing counterion diameter.
Finally, the first-principles model was extended to hybrid pseudocapacitors by accounting for redox reactions and Li+ intercalation and by rigorously deriving the associated irreversible and reversible heat generation rates. The model accounted simultaneously for charge storage by EDL formation and by faradaic reactions. Simulations were performed for a planar hybrid pseudocapacitor to investigate the electrochemical interfacial and transport phenomena as well as the thermal behavior under galvanostatic cycling. Two asymptotic regimes were identified corresponding to (i) dominant faradaic charge storage at low current and low frequency or (ii) dominant EDL charge storage at high current and high frequency. Predicted cell potential, heat generation rates, and temperature showed good qualitative agreement with experimental measurements and can be used to physically interpret experimental observations.