Gabriele Raabe

Proton exchange membrane water electrolysis (PEMWE) is a key technology for production of hydrogen at high current density and dynamic operation. However, long-term durability of PEMWE is limited by the sensitivity of perfluorosulfonic acid (PFSA) membranes to impurities in the water feed.[1,2] Even small amounts of metal cations alter membrane properties such as conductivity, water uptake, and transport phenomena due to proton-cation exchange processes on the sulfonic groups.[3,4] These cation-exchange processes not only reduce system efficiency but also influence water management, which is highly critical for membrane hydration. Despite their practical relevance, a fundamental relationship between cation contamination and water transport in PEMWE is not fully understood to date.

Lignin is a promising starting material for the future provision of aromatic platform chemicals. Its low solubility in aqueous media, however, hampers its efficient depolymerization using enzymatic approaches. Thus, the addition of cosolvents is often required to improve the lignin solubility. Herein, we demonstrate that mixtures of ethylene glycol/choline chloride (EG:ChCl)-based deep eutectic solvents (DES) and glycine buffer represent powerful solvent systems for lignin dissolution in β-etherase-catalyzed lignin depolymerization reactions. They enabled the use of high lignin concentrations (up to 30 g L^-1 at 60% v/v DES) without abolishing enzyme activity. While EG:ChCl 2:1 proved superior in solubilizing lignin, the highest volumetric yield of the desired monoaromatic depolymerization products guaiacylhydroxypropanone and syringylhydroxypropanone was achieved using 20 g L^-1 lignin and 50% v/v EG:ChCl 4:1. Moreover, detailed MD analyses on the interactions of the lignin model compound guaiacylglycerol-β-guaiacyl ether (GGE) and the solvent mixture, including also glycine as a buffer component, highlight the importance of EG and glycine for lignin solubility, while the ratio of EG and ChCl in the DES impacts the solvation structure and dynamics of GGE in the solvent mixture. Thus, our comprehensive study offers insights into the molecular determinants that affect lignin solubility and its enzymatic depolymerization.

Interface force fields (IFFs) are crucial for the atomistic modeling of PEMFC interfaces. We benchmark ten optimization algorithms and four force-field forms against DFT adsorption energies and forces and present a framework for selecting the algorithm/force-field combination. At Pt–H₂O, the best IFF yields a mean absolute error (MAE) of 0.32 eV for adsorption energy and 0.12 eV/Ų for forces, cutting energy error by 70–90% and force error by 44–48% compared to COMB3, ReaxFF, and LJ mixing. Similar improvements are observed for Pt₃Ni–H₂O and Pt₃Co–H₂O. For Pt/Pt₃Ni/Pt₃Co–O₂, energy MAEs are 0.30–0.68 eV and force MAEs are 0.42–0.85 eV/Ų; errors are larger for force components parallel to the surface, while normal components remain well captured, reflecting the challenges of describing O₂ dissociation on (111) metal surfaces using classical force fields. The framework enables data-efficient IFF development and guides the selection of force-field forms and optimization algorithms for PEMFC and interfacial simulations.

The political, social and economic consequences of climate change drastically influence the requirements of modern energy systems and its components. This includes not only energy production but also concepts and innovations for its storage, especially in magnitudes of gigawatt hours. Carnot batteries, which convert electrical energy to thermal energy using a heat pump cycle, offer a potential solution. If required, the stored energy is converted back in a power cycle. Thereby, the efficiency of each conversion cycle is highly relevant for the final round trip efficiency of the Carnot battery. For the design, simulation, and optimisation of thermodynamic cycles, sufficient knowledge of the properties of the working fluid is required. These properties are commonly described using highly accurate multiparameter equations of state (EOS) in the form of the Helmholtz energy. When mixtures are used as working fluid, additional adjustable parameters in the mixing rules of the reducing values of the residual fluid contribution are required. These parameters need to be fitted to experimental data of the mixture that is subject to measurement uncertainties. When there is only very little data available, the impact of measurement uncertainty is particularly large. This leads to uncertainties in the derived parameters, which must be taken into account when using the EOS for further calculations. The process of determining model parameters and its uncertainties given noisy measurements can be referred to as model calibration. Uncertainties in the vapor-liquid-equilibrium (VLE) data lead to a calibration setting, where both inputs and outputs are random, hence, an Error-in-Variables approach is chosen. Here, this approach is exemplarily applied to the Recuperative Two-Phase Cycle (RTPC) using non-conventional asymmetric working fluid mixtures, where it offers a particularly high added value due to the small amount of measurement data. The RTPC is a novel cycle concept (Bederna et al., 2023), which is currently under investigation to increase the efficiency of the thermodynamic cycles in Carnot batteries. The specific characteristic is the use of asymmetric refrigerant mixtures. In this work, parameters are derived for the zeotropic mixture propane-dodecane, whose experimental VLE data is provided by Gardeler et al., 2002. Therefore, we embed the problem in a Bayesian probabilistic setting and sample from the posterior distribution using the affine invariant Markov Chain Monte Carlo ensemble sampler (Goodman and Weare, 2010). We compare the posterior mean with the results of other automatic fitting methods. Finally, the impact of the variance on the second law efficiency and the optimal process parameters for a basic heat pump version of the RTPC are shown.

Lignin is a promising starting material for the future provision of aromatic platform chemicals. Its low solubility in aqueous media, however, hampers its efficient depolymerization using enzymatic approaches. Thus, the addition of cosolvents is often required to improve lignin solubility. Herein, we demonstrate that mixtures of ethylene glycol:choline chloride (EG:ChCl)-based deep eutectic solvents (DES) and glycine buffer represent powerful solvent systems for lignin dissolution in β-etherase-catalyzed lignin depolymerization reactions. They enabled the use of high lignin concentrations (up to 30 g L-1 at 60 % v/v DES) without abolishing enzyme activity. While EG:ChCl 2:1 proved superior in solubilizing lignin, the highest volumetric yield of the desired monoaromatic depolymerization products guaiacylhydroxypropanone and syringylhydroxypropanone was achieved using 20 g L-1 lignin and 50 % v/v EG:ChCl 4:1. Moreover, detailed MD analyses on the interactions of the lignin model compound guaiacylglycerol-β-guaiacyl ether (GGE) and the solvent mixture, including also glycine as buffer component, highlight the importance of EG and glycine for lignin solubility, while the ratio of EG and ChCl in the DES impacted the solvation structure and dynamics of GGE in the solvent mixture. Thus, our comprehensive study offers insights into the molecular determinants that affect lignin solubility and its enzymatic depolymerization.

Intermolecular parameters in force fields for molecular simulation studies of thermophysical properties are typically derived through an iterative fitting process to experimental data. However, this approach suffers from extremely high computational cost because of the necessity to perform molecular simulations for a multitude of parameter values. While machine learning approaches have emerged in recent years in bottom-up derivation of interatomic potentials from quantum chemical simulations, their application in top-down methods for parameterizing classical force fields is much less explored. Moreover, the few available machine learning approaches for top-down parametrizations are not yet capable of constructing efficient and localized surrogate models while simultaneously quantifying uncertainties. Here, we overcome these issues by presenting a new sequential design strategy, based on Gaussian Process surrogate models, directly tailored for a Bayesian calibration task. We employ our approach to calibrate force field parameters for the refrigerant trans-1,2-dichloroethene (R-1130(E)) with experimental vapor-liquid phase equilibria (VLE) data and achieve convergence with a low number of training points. In addition to utilizing the calibrated model for simulation studies on the VLE of R-1130(E), we also apply it for predictive molecular dynamics studies on its liquid density and viscosity, and thereby assess the impact of parameter uncertainties on the simulation results.