TY - JOUR
T1 - Multi-fidelity machine learning models for structure–property mapping of organic electronics
AU - Yang, Chih Hsuan
AU - Pokuri, Balaji Sesha Sarath
AU - Lee, Xian Yeow
AU - Balakrishnan, Sangeeth
AU - Hegde, Chinmay
AU - Sarkar, Soumik
AU - Ganapathysubramanian, Baskar
N1 - Funding Information:
This work was supported by the ARPA-E DIFFERENTIATE program, USA under grant DE-AR0001215 . BG, C-HY, and BP were supported in part by DoD MURI, USA 6119-ISU-ONR-2453 and NSF 1906194 . CH was supported in part by NSF, USA grants 2005804 and 1815101 . Computing support from XSEDE, USA and Iowa State University, USA is gratefully acknowledged.
Publisher Copyright:
© 2022 Elsevier B.V.
PY - 2022/10
Y1 - 2022/10
N2 - Machine learning approaches have been used with significant success in constructing, curating, and exploring relationships between microstructure and property. However, one major limitation of these approaches is the need for a significant amount of training data consisting of microstructure–property pairs. Getting property values associated with a specific microstructure typically requires deploying a detailed physics simulator which becomes resource-intensive. While using a low(er) fidelity property quantifier can offset the cost of creating the training dataset, there is a trade-off in terms of accuracy/fidelity of the estimated property. Here, we leverage the availability of low- and high- fidelity property simulators to construct a multi-fidelity mapping from microstructure to property using deep convolutional neural networks. Starting with a large dataset of morphologies representing the active layer of organic photovoltaic devices, we assimilate data from a rapid graph-based low-fidelity characterization of the morphology with limited data from a high fidelity excitonic drift-diffusion detailed physics simulator. We show that our method provides significant computational savings while maintaining competitive performance. This work can be easily extended to other applications, and we envision it as a basis for accelerated material quantification and discovery.
AB - Machine learning approaches have been used with significant success in constructing, curating, and exploring relationships between microstructure and property. However, one major limitation of these approaches is the need for a significant amount of training data consisting of microstructure–property pairs. Getting property values associated with a specific microstructure typically requires deploying a detailed physics simulator which becomes resource-intensive. While using a low(er) fidelity property quantifier can offset the cost of creating the training dataset, there is a trade-off in terms of accuracy/fidelity of the estimated property. Here, we leverage the availability of low- and high- fidelity property simulators to construct a multi-fidelity mapping from microstructure to property using deep convolutional neural networks. Starting with a large dataset of morphologies representing the active layer of organic photovoltaic devices, we assimilate data from a rapid graph-based low-fidelity characterization of the morphology with limited data from a high fidelity excitonic drift-diffusion detailed physics simulator. We show that our method provides significant computational savings while maintaining competitive performance. This work can be easily extended to other applications, and we envision it as a basis for accelerated material quantification and discovery.
KW - Deep learning
KW - Multi-fidelity data
KW - Organic electronics
KW - Photovoltaics
KW - Structure–property mapping
UR - http://www.scopus.com/inward/record.url?scp=85134759660&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85134759660&partnerID=8YFLogxK
U2 - 10.1016/j.commatsci.2022.111599
DO - 10.1016/j.commatsci.2022.111599
M3 - Article
AN - SCOPUS:85134759660
SN - 0927-0256
VL - 213
JO - Computational Materials Science
JF - Computational Materials Science
M1 - 111599
ER -