Seminarium "Modeling of Complex Systems"

The increase in energy storage devices has become a challenge that demands rethinking and modifying our currenttechnologies. This problem has been augmented by the lithium dependence of the current energy storage devices, a scarceraw material. Becoming Na-ion batteries as a promising alternative, however, also faces several limitations to solve, where thestorage on the anode becomes one of the main key aspects to study. Among several evaluated materials, hard carbon as ananode presents advantages such as low price, high storage capacity, and thermal stability. But the span of local chemistrymotifs makes it harder to relate their structural properties to the mechanism of storage, ranging from defective and curvedgraphene fragments, short-range-ordered graphite-like domains, to closed/open pores. Experimental studies have proven thecorrelations of the structure with the mechanism of storage of sodium, but its detailed characterization at the atomic levelremains unclear. In another hand, computational studies have tried to study from an atomic level perspective, thethermodynamics and kinetic related to the Na/C-based systems. From DFT studies, it has been possible to characterize theadsorption, intercalation, and pore filling as isolated phenomena, but the connection and interplay of these phenomena in areal system of hard carbon is limited by the size of the system. Another approach using classical force field potentials has beenable to handle the size of the system, by presenting high thermodynamics deviations. As a novel and promising approach, theuse of universal foundational models based on machine learning trained with DFT data has emerged. But still remaining tounderstand how the transferability of these is to Na/C-based systems. In this talk, we discuss which transferability issuespresent the art-state model (MACE)[1], and how to guide the fine-tuning of these models to improve the capture of the localchemistry motifs presented in the storage of sodium on hard carbon [2, 3].[1] Transferability of Foundation MACE Models to Amorphous Chemistry: Where They Work and Where They Fail. O. I. Malyi,and A. F. Usuga. *Work submitted.
[2] From mismatch to transferability: MACE for ionic-metallic interplay in Na carbon-electrodes. A. F. Usuga, I. Radchenko, O.I. Malyi. *Work in progress.
[3] I. Radchenko, A. F. Usuga, O.I. Malyi, Kinetic limit in the close pore filling in the hard carbon electrode. *Work in progress.

The seminar will be held in hybrid mode: in room 1.40 (FUW, Pasteura 5).During the seminar the coffee and cakes are provided. Join Zoom Meeting
https://uw-edu-pl.zoom.us/j/96378632993?pwd=MVgdPR80oKaE4pjLufb2NCtg6ql4Ax.1
Meeting ID: 963 7863 2993
Passcode: 569551