Machine learning guided design of single-molecule magnets for magnetocaloric applications

Abstract

We present a data-driven approach to predict entropy changes (ΔS) in small magnetic fields in single-molecule magnets (SMMs) relevant to their application as magnetocaloric refrigerants. We construct a database of SMMs with a representation scheme incorporating aspects related to dimensionality, structure, local coordination environment, ideal total spin of magnetic ions, ligand type, and linking chemistry. We train machine learning models for predicting the entropy change as a function of structure and chemistry and use the models to arrive at ΔS for hypothetical molecules. We also identify key descriptors that affect the entropy change, thus providing insights into designing tailored SMMs with improved magnetocaloric properties.

Computational Discovery of Transition-metal Complexes: From High-throughput Screening to Machine Learning

“…These models reduced the computational cost to identify optimal structural properties in designing single-ion magnetic anisotropy . Using experimental data sets, LASSO models have also been developed to design SMMs with a large isothermal magnetic entropy change . On the basis of a set of 60 datamined experimentally synthesized SMMs, heuristic descriptors, such as size, number, and type of metal ions, were used as inputs to the LASSO model for both feature selection and model prediction .…”

Magnetocaloric Effect in Cu5-NIPA Molecular Magnet: A Theoretical Study

“…The molecular nanomagnets proved their potential in subkelvin cooling [19] and might be useful for on-chip cooling of nanoelectronic devices [20,21]. The quest for cooling efficiency stimulates the development of various approaches to design molecular magnets with desired properties [22][23][24][25][26]. Molecular magnets may offer record-high spin per molecule maximizing the potential span of entropy change, turning the attention to high spin clusters [27][28][29].…”

Data-driven design of B20 alloys with targeted magnetic properties guided by machine learning and density functional theory

“…An ensemble of 100 ML models was trained based on the support vector regression (SVR) method for predicting the T C as a function of three input descriptors (see "Methods" for more details). Balachandran and co-workers have demonstrated the potential of ensemble-based SVR methods for building reliable ML models from small data [27,38,39,40,41,42]. The trained models were validated using Mn 0.75 Rh 0.25 Ge B20 alloy composition as a test case [37], which was recently synthesized by Sidorov Fig.…”

“…In this work, a novel computational approach, built on the foundations of machine learning (ML) and DFT, is developed to accelerate the design of B20-based chiral magnets with improved T C . Although ML methods have been used in the past to predict the ferromagnetic Curie temperature of alloys [23,24], properties of hard permanent magnets [25], two-dimensional materials [26] and magnetic properties of single-molecule magnets [27,28], no a priori rules exist that link alloy compositions to T C for the B20 alloys.…”