Ion Sieving in Two-Dimensional Membranes from First Principles

Nicéphore Bonnet, Nicola Marzari · 2024

Read Ion Sieving in Two-Dimensional Membranes from First Principles by Nicéphore Bonnet, Nicola Marzari (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-13899.txt.

Introduction

Nicéphore Bonnet and Nicola Marzari set out a first-principles method for predicting how well atom-thin membranes can sieve ions from water. Such membranes could sort valuable ions, like lithium, from brine. The challenge is to compute the sorting power from basic physics rather than trial and error.

First, the method computes the energy each ion needs to cross the membrane, with water effects simulated by machine-learned dynamics. Fixes remove errors from the small simulated box, and a model handles the charge layer that forms at the surface. Heat-driven effects are added and checked against a stricter method. A final model of the filtering process turns those energies into predicted sorting rates, tested on a graphene membrane.

Catalog record

CREATORNicéphore Bonnet, Nicola Marzari
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE34 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-13899

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Ion Sieving in Two-Dimensional Membranes from First Principles Nicéphore Bonnet, Nicola Marzari Abstract A first-principles approach for calculating ion separation in solution through two-dimensional (2D) membranes is proposed and applied. Ionic energy profiles across the membrane are obtained first, where solvation effects are simulated explicitly with machine-learning molecular dynamics, electrostatic corrections are applied to remove finite-size capacitive effects, and a mean-field treatment of the charging of the electrochemical double layer is used. Entropic contributions are assessed analytically and validated against thermodynamic integration. Ionic separations are then inferred through a microkinetic model of the filtration process, accounting for steady-state charge separation effects across the membrane. The approach is applied to Li + , Na + , K + sieving through a crown-ether functionalized graphene membrane, with a case study of the mechanisms for a highly selective and efficient extraction of lithium from aqueous solutions. keywords American Chemical Society, L a T e X † † email: [email protected] † † affiliation: Theory and Simulation of Materials (THEOS), Ecole Polytechnique Fédérale de Lausanne, 1015 Lausanne, Switzerland † † abbreviations: IR,NMR,UV Keywords: 2D membranes, ion sieving, first-principles calculations, machine learning, microkinetic model, multiscale modelling, electrochemical double layer 1 Introduction Advanced membrane filtration is an important target for emerging separation technologies, and is critically needed, e.g., for sustainable water and energy management Shao2020 ; DuChanois2021 . Notably, ion selectivity is an increasingly required feature for membrane technologies Werber2016 . In water treatment and recycling, ion-selective membranes can be used for recovery of valuable resources, such as lithium from brines, rare-earth elements from industrial wastewater, or phosphate and nitrate from wastewater, or for targeted removal of undesired species, such as heavy metals DuChanois2021 ; Stringfellow2021 . Ion selectivity is equally important for advanced membranes used as half-cell separators in electrochemical technologies for low-carbon energy conversion and storage, including batteries, fuel cells, and electrolyzers. Current polymeric membranes have been unable to reach the desired separation target for many advanced water and energy applications owing to the permeability-selectivity trade-off, whereby higher selectivities entail smaller permeabilities. This trade-off has been linked to the inability to control satisfactorily the pore size distribution in polymeric membranes, which motivated the development of novel materials with a higher molecular-level control over physical and chemical properties; these include porous crystalline materials, 2D materials, and biomimetic materials DuChanois2021 ; Mounet2018 ; Zhao2024 . In parallel, optimal design and operation of ion-sieving membranes requires a more fundamental understanding of the molecular drivers for ion selectivity; notably, the interplay between size, charge, dielectric and chemical effects Szymczyk2005 ; Yuan2022 ; Wen2016 ; Razmjou2019 ; Fan2023 . Size effects include the need for the ions to rearrange or remove their hydration shell to fit within the pores, and are thus related to both solvation and dielectric effects. Charge effects occur by intrinsic charging of the membrane and/or by charge separation across the membrane, e.g. via the Donnan mechanism Fievet2014 . Finally, pores’ functionalizations may bring dramatic selectivity effects through chemical coordination, as exemplified by crown-ether graphene membranes Fang2019 ; Sahu2019 ; Guo2021 ; Lv2023 . While data-driven approaches have been successfully applied to rank the importance of selectivity drivers in existing polymeric membranes Ritt2022 , atomistic simulations are needed for a detailed mechanistic understanding and computational screening of new candidate ion-selective materials DuChanois2021 ; Zhou2020 ; Chipot2023 . Classical molecular-dynamics (MD) simulations based on traditional force fields, which were first applied to investigate the permeability and selectivity of membranes Fang2019 ; Sahu2019 ; Sint2008 ; Suk2010 ; Ruan2016 ; Smolyanitsky2018 ; Zofchak2022 ; Meng2023 , can be limited by the accuracy of the models. In first-principles simulations, alternatively, an explicit treatment of all solvent molecules may be computationally burdensome, and implicit solvent models Andreussi2012 , while useful and able to reduce the computational costs, may fail to capture specific molecular effects at the solute / solvent interface. Moreover, charge effects are embedded within the electrochemical double layer (EDL), a complex structure surrounding the membrane and often exceeding the typical size of computational cells Shin2022 . Finally, the thermodynamic potential energy surfaces of ionic species across the membrane translate into effective ionic selectivities only in the context of a dynamic description of the membrane filtration process Szymczyk2005 . The present work proposes a first-principles-based methodology for the prediction of membrane ion separation and applies it to the case of a paradigmatic 2D advanced membrane, addressing all the previous points. The energy profiles (EPs) of individual ions across the membrane pore are first determined, where solvation effects are simulated explicitly by machine-learning (ML) accelerated MD, and electrostatic correction schemes are applied to remove capacitive effects arising in finite-size simulation cells. EDL effects are added subsequently as a mean-field contribution to the ionic EP, using simplifying assumptions on the structure of the EDL above certain ionic concentrations. Entropic effects are assessed analytically and through a thermodynamic integration scheme. Finally, a microkinetic model of the ion-sieving process is developed to obtain effective ionic separations for a realistic set of operational parameters of the dynamic environment. The methodology is illustrated with the case of Li + , Na + , K + separation through 12-crown-4 ethers embedded in a graphene membrane in aqueous solution. 2 Results and Discussion 2.1 Methodological overview Figure 1 : Top (left) and side (right) view of the membrane pore and of the system setup to calculate the ion translocation energy profile; here for the case of a Li + ion solvated in water across a crown-ether functionalized graphene membrane. The translocation of an ion through a membrane is considered via the setup of Fig. 1 . Unless stated otherwise, z (Å) will denote the longitudinal distance of the ion to the membrane plane. For a small pore with an activated translocation, as considered in the present study, the ion permeance is mainly governed by a two-step process Yuan2022 : first, the formation of a pore-associated state of the permeating ion, typically as an adsorption state in the pore vicinity; second, the actual translocation event from the pore-associated state. To quantify this process, we determine the energy profile (EP) of the ion in the pore region as a function of z . The solvated EP is decomposed as E_{aq}(z)=E_{vac}(z)+E_{solv}(z) (1) where E_{vac}(z) is the EP in vacuum and E_{solv}(z) is the solvation energy. In the following, we present an approach to calculate E_{vac}(z) and E_{solv}(z) from first principles detailing the intermediate steps using the example of a solvated Li + ion, as in Fig. 1 . The simulation cell contains one pore made of a 12-crown-4 ether analogue embedded in the graphene membrane. Two cell sizes are considered: a first simulation cell, denoted as 1\times 1 , with dimensions 12.33 Å \times 12.81 Å; and a larger simulation cell, denoted as 2\times 2 , with dimensions 24.66 Å \times 25.62 Å. Total energy calculations are performed in vacuum or in the presence of explicit water. Entropic contributions are assessed analytically and via thermodynamic integration, and the resulting free energy profile is injected into a microkinetic model of the sieving process. The model incorporates EDL effects as a mean-field contribution to ionic EPs and self-consistently predicts the charge separation density across the membrane upon filtration. 2.2 Energy profile in vacuum

Read the complete text