Summary
A major challenge for the green transition is our inability to rationally design inorganic materials with tailor-made properties. This project will tackle this inability by transforming our understanding of chemical bonding in inorganic materials.
Understandable rules based on chemical bonds have greatly advanced chemistry but are missing for most material properties, severely limiting the rational design of materials. Until recently, quantum chemical bonding analysis of inorganic materials has only been carried out on a small scale, making it impossible to derive such rules using machine learning. In addition, quantum chemical bonding analysis primarily focuses on two-center bonds. However, multicenter bonds play a critical role in material properties: For example, multicenter bonds have been held responsible for the superhardness of boron-containing compounds and the unusual properties of phase-change materials. By significantly going beyond my recent results on two-center bonds predicting materials properties with simple machine-learning models, I propose to overcome these challenges. The overarching objective of MultiBonds is to derive understandable and universal rules based on chemical bonds for inorganic materials properties through large-scale quantum-chemical bonding analysis considering multicenter bonds. We will 1) develop and apply innovative automated quantum-chemical methods to compute, for the first time, multicenter bonding indicators on a large scale. The generated database will then be used for 2) developing novel predictive deep-learning models and 3) intuitive human-understandable rules for materials properties. As initial applications, we will focus on phase-change materials with low thermal conductivities, magnetic and hard materials, since their properties are known to be governed by multicenter bonds, and they have critical applications (e.g., as thermoelectrics and in the green transition of vehicles).
Understandable rules based on chemical bonds have greatly advanced chemistry but are missing for most material properties, severely limiting the rational design of materials. Until recently, quantum chemical bonding analysis of inorganic materials has only been carried out on a small scale, making it impossible to derive such rules using machine learning. In addition, quantum chemical bonding analysis primarily focuses on two-center bonds. However, multicenter bonds play a critical role in material properties: For example, multicenter bonds have been held responsible for the superhardness of boron-containing compounds and the unusual properties of phase-change materials. By significantly going beyond my recent results on two-center bonds predicting materials properties with simple machine-learning models, I propose to overcome these challenges. The overarching objective of MultiBonds is to derive understandable and universal rules based on chemical bonds for inorganic materials properties through large-scale quantum-chemical bonding analysis considering multicenter bonds. We will 1) develop and apply innovative automated quantum-chemical methods to compute, for the first time, multicenter bonding indicators on a large scale. The generated database will then be used for 2) developing novel predictive deep-learning models and 3) intuitive human-understandable rules for materials properties. As initial applications, we will focus on phase-change materials with low thermal conductivities, magnetic and hard materials, since their properties are known to be governed by multicenter bonds, and they have critical applications (e.g., as thermoelectrics and in the green transition of vehicles).
Unfold all
/
Fold all
More information & hyperlinks
| Web resources: | https://cordis.europa.eu/project/id/101161771 |
| Start date: | 01-01-2025 |
| End date: | 31-12-2029 |
| Total budget - Public funding: | 1 500 000,00 Euro - 1 500 000,00 Euro |
Cordis data
Original description
A major challenge for the green transition is our inability to rationally design inorganic materials with tailor-made properties. This project will tackle this inability by transforming our understanding of chemical bonding in inorganic materials.Understandable rules based on chemical bonds have greatly advanced chemistry but are missing for most material properties, severely limiting the rational design of materials. Until recently, quantum chemical bonding analysis of inorganic materials has only been carried out on a small scale, making it impossible to derive such rules using machine learning. In addition, quantum chemical bonding analysis primarily focuses on two-center bonds. However, multicenter bonds play a critical role in material properties: For example, multicenter bonds have been held responsible for the superhardness of boron-containing compounds and the unusual properties of phase-change materials. By significantly going beyond my recent results on two-center bonds predicting materials properties with simple machine-learning models, I propose to overcome these challenges. The overarching objective of MultiBonds is to derive understandable and universal rules based on chemical bonds for inorganic materials properties through large-scale quantum-chemical bonding analysis considering multicenter bonds. We will 1) develop and apply innovative automated quantum-chemical methods to compute, for the first time, multicenter bonding indicators on a large scale. The generated database will then be used for 2) developing novel predictive deep-learning models and 3) intuitive human-understandable rules for materials properties. As initial applications, we will focus on phase-change materials with low thermal conductivities, magnetic and hard materials, since their properties are known to be governed by multicenter bonds, and they have critical applications (e.g., as thermoelectrics and in the green transition of vehicles).
Status
SIGNEDCall topic
ERC-2024-STGUpdate Date
24-11-2024
Images
No images available.
Geographical location(s)
Search
