Summary
“Machine Learning for Tailoring Organic Semiconductors” (MALTOSE) connects fundamental materials research with machine-learning (ML) techniques, focusing on the electronic properties of organic semiconductors. The aim of this innovative project is to discover and design novel materials with exciting properties, the prime example being the design of compounds for better organic photovoltaic cells, i.e., that reach higher power-conversion efficiencies and are more stable and more environmentally friendly.
The methodology relies on a deep tensor neural network, the so-called PredictNet, that is designed and trained to predict electronic properties of molecules and polymers, at a fraction of the numerical cost compared to density-functional theory (DFT) computations, not to mention experimental measurements. PredictNet will be particularly fruitful in combination with a genetic algorithm that will be developed to propose candidate compounds from crossover and mutation from previously successful compounds. MALTOSE will enable the identification and design of promising compounds, out of the immense pool of imaginable molecules and materials, for future technological applications in fields like organic photovoltaic solar cells, large-area electronic displays, flexible organic electronics, or sensors.
The project will bring together the fellow, a recognized quantum physicist and data scientist with academic and industry research experience, and a top research host institution under the supervision of a leading expert in materials science, genetic algorithms, modelling, simulation and knowledge transfer. The fellow will receive an advanced training programme in research skills and complementary non-research-oriented skills in order to enhance his future career prospects and to provide a strong basis for an independent career.
The methodology relies on a deep tensor neural network, the so-called PredictNet, that is designed and trained to predict electronic properties of molecules and polymers, at a fraction of the numerical cost compared to density-functional theory (DFT) computations, not to mention experimental measurements. PredictNet will be particularly fruitful in combination with a genetic algorithm that will be developed to propose candidate compounds from crossover and mutation from previously successful compounds. MALTOSE will enable the identification and design of promising compounds, out of the immense pool of imaginable molecules and materials, for future technological applications in fields like organic photovoltaic solar cells, large-area electronic displays, flexible organic electronics, or sensors.
The project will bring together the fellow, a recognized quantum physicist and data scientist with academic and industry research experience, and a top research host institution under the supervision of a leading expert in materials science, genetic algorithms, modelling, simulation and knowledge transfer. The fellow will receive an advanced training programme in research skills and complementary non-research-oriented skills in order to enhance his future career prospects and to provide a strong basis for an independent career.
Unfold all
/
Fold all
More information & hyperlinks
| Web resources: | https://cordis.europa.eu/project/id/883256 |
| Start date: | 01-03-2021 |
| End date: | 30-07-2024 |
| Total budget - Public funding: | 172 932,48 Euro - 172 932,00 Euro |
Cordis data
Original description
“Machine Learning for Tailoring Organic Semiconductors” (MALTOSE) connects fundamental materials research with machine-learning (ML) techniques, focusing on the electronic properties of organic semiconductors. The aim of this innovative project is to discover and design novel materials with exciting properties, the prime example being the design of compounds for better organic photovoltaic cells, i.e., that reach higher power-conversion efficiencies and are more stable and more environmentally friendly.The methodology relies on a deep tensor neural network, the so-called PredictNet, that is designed and trained to predict electronic properties of molecules and polymers, at a fraction of the numerical cost compared to density-functional theory (DFT) computations, not to mention experimental measurements. PredictNet will be particularly fruitful in combination with a genetic algorithm that will be developed to propose candidate compounds from crossover and mutation from previously successful compounds. MALTOSE will enable the identification and design of promising compounds, out of the immense pool of imaginable molecules and materials, for future technological applications in fields like organic photovoltaic solar cells, large-area electronic displays, flexible organic electronics, or sensors.
The project will bring together the fellow, a recognized quantum physicist and data scientist with academic and industry research experience, and a top research host institution under the supervision of a leading expert in materials science, genetic algorithms, modelling, simulation and knowledge transfer. The fellow will receive an advanced training programme in research skills and complementary non-research-oriented skills in order to enhance his future career prospects and to provide a strong basis for an independent career.
Status
SIGNEDCall topic
MSCA-IF-2019Update Date
28-04-2024
Images
No images available.
Geographical location(s)
