Chemical Engineering Journal xxx (xxxx) xxx Please cite this article as: Rebecca L. Gibson, Chemical Engineering Journal, https://doi.org/10.1016/j.cej.2020.127318 Available online 14 October 2020 1385-8947/© 2020 Elsevier B.V. All rights reserved. Kinetic modelling of thermal processes using a modifed Sestak-Berggren equation Rebecca L. Gibson a, b, * , Mark J.H. Simmons b , E. Hugh Stitt a , John West a , Sam K. Wilkinson a, 1 , Robert W. Gallen a a Johnson Matthey, Belasis Avenue, Billingham, Stockton-on-Tees TS23 1LB, UK b School of Chemical Engineering, University of Birmingham, Edgbaston B15 2TT, UK A R T I C L E INFO Keywords: Sestak-Berggren Solid-state reaction kinetics Model validation Temperature programmed reduction ABSTRACT This study outlines the principles of modelling the kinetics of solid-state reactions through the simultaneous ftting of multiple peak curves using the modifed Sestak-Berggren equation. This mathematical model gives an indication of the mechanism occurring and allows kinetic parameters, such as activation energy, to be estimated. This methodology is demonstrated using in silico thermo-conductivity detector (TCD) data showing the internal consistency of the Sestak-Berggren modelling approach, its applicability to noisy data and its ability to predict mechanisms occurring during a thermally induced solid state reaction. Using these in silico data it has been confrmed that this empirical model can separate overlapped peaks without a priori peak deconvolution. A rigorous statistical methodology based on the Akaike Information Criteria, is recommended to identify the op- timum number of thermal events that should be applied to a system. This modifed Sestak-Berggren model is then applied to an experimental dataset of temperature programmed reduction of a calcined cobalt on alumina catalyst precursor. This allows for the identifcation of a statistically adequate kinetic triplet for each thermal event. Recommendations on the treatment of datasets which contain “shoulders” and closely overlapped peaks are also given. 1. Introduction The characterisation of a functionalised material is an important step in understanding its performance, such as catalytic activity and selec- tivity. Extracting quantitative data from characterisation techniques could offer improved understanding and predictive modelling of in- dustrial processes such as catalyst reduction. Columbo et al. [1] and Lietti et al. [2] have already demonstrated the advantages of applying the fundamental understanding gained from such techniques to a more complex system such as a SCR monolith drive cycle model. Solid-state reaction kinetics have been a widely studied using ther- mal analysis methods, with major improvements in both experimental and computational techniques allowing more accurate kinetic parame- ters to be extracted [3]. However, the deconvolution of overlapped thermal events has historically relied on the expertise of experienced practitioners; with a judgement based on experience with equipment and the interpretation of the chemistry studied, impacting the assignment of the number of thermal events. A method to simulta- neously deconvolute a statistically signifcant number of thermal events and extract kinetic information is outlined in this paper. It is common for thermal analysis results to feature more than one peak, which makes the resolution of these peaks an important part of the data processing. In the literature, thermal events (or peaks) are commonly deconvoluted frst, before kinetic analysis can be carried out. This deconvolution is commonly carried out by ftting Gaussian distri- butions, either symmetric or asymmetric [4], but this process can introduce errors. To avoid these, the full curve would be better analysed simultaneously. It is mathematically possible to assign any number of Gaussian curves to ft the thermal analysis data [5,6] and a focus of this study is to evaluate the optimum number of thermal events to assign to a system. Fitting additional curves may improve a model’s closeness of ft, however the parameter values extracted could cease to relate to physical parameters [5]. However, the addition of curves may not improve quality of ft as this * Corresponding author at: Johnson Matthey, Belasis Avenue, Billingham, Stockton-on-Tees TS23 1LB, UK. E-mail address: Rebecca.Gibson@matthey.com (R.L. Gibson). 1 Current address: Process Systems Enterprise, Hammersmith Grove, Hammersmith, London W6 7HA, UK. Contents lists available at ScienceDirect Chemical Engineering Journal journal homepage: www.elsevier.com/locate/cej https://doi.org/10.1016/j.cej.2020.127318 Received 8 July 2020; Received in revised form 21 September 2020; Accepted 8 October 2020