TY - JOUR
T1 - Machine learned metaheuristic optimization of the bulk heterojunction morphology in P3HT:PCBM thin films
AU - Munshi, Joydeep
AU - Chen, Wei
AU - Chien, Te Yu
AU - Balasubramanian, Ganesh
N1 - Funding Information:
This material is based on the work supported by the National Science Foundation (NSF) under Award Nos. CMMI-1662435, 1662509 and 1753770 (the latter includes support from a supplement for data science related CMMI research activities). JM and GB acknowledge the allocation of leadership class computing time on Frontera (at the Texas Advanced Computing Center (TACC) in The University of Texas at Austin) to perform the simulations and support through the NSF award # OAC-2031682. Any opinions, findings, conclusions or recommendations expressed in this material are those of the authors’ and do not necessarily reflect the views of the NSF.
Publisher Copyright:
© 2020 Elsevier B.V.
PY - 2021/2/1
Y1 - 2021/2/1
N2 - We discuss results from a machine learned (ML) metaheuristic cuckoo search (CS) optimization technique that is coupled with coarse-grained molecular dynamics (CGMD) simulations to solve a materials and processing design problem for organic photovoltaic (OPV) devices. The method is employed to optimize the composition of donor and acceptor materials, and the thermal annealing temperature during the morphological evolution of a polymer blend active layer composed of poly-(3-hexylthiophene) (P3HT) and phenyl-C61-butyric acid methyl ester (PCBM), for an increased power conversion efficiency (PCE). The optimal solutions, which are in qualitative agreement with earlier experiments, identify correlation between the design variables that contributes to an enhanced material performance. The framework is extended to multi-objective design (MOCS-CGMD) to attain a Pareto optimality for the blend morphology, and enhance concurrently the exciton diffusion to charge transport probability and the ultimate tensile strength of the material. The predictions reveal that a higher annealing temperature enhances the exciton diffusion to charge transport probability, while a PCBM weight fraction between 0.4 and 0.6 increases the tensile strength of the underlying blend morphology.
AB - We discuss results from a machine learned (ML) metaheuristic cuckoo search (CS) optimization technique that is coupled with coarse-grained molecular dynamics (CGMD) simulations to solve a materials and processing design problem for organic photovoltaic (OPV) devices. The method is employed to optimize the composition of donor and acceptor materials, and the thermal annealing temperature during the morphological evolution of a polymer blend active layer composed of poly-(3-hexylthiophene) (P3HT) and phenyl-C61-butyric acid methyl ester (PCBM), for an increased power conversion efficiency (PCE). The optimal solutions, which are in qualitative agreement with earlier experiments, identify correlation between the design variables that contributes to an enhanced material performance. The framework is extended to multi-objective design (MOCS-CGMD) to attain a Pareto optimality for the blend morphology, and enhance concurrently the exciton diffusion to charge transport probability and the ultimate tensile strength of the material. The predictions reveal that a higher annealing temperature enhances the exciton diffusion to charge transport probability, while a PCBM weight fraction between 0.4 and 0.6 increases the tensile strength of the underlying blend morphology.
KW - Coarse-grained molecular dynamics
KW - Cuckoo Search
KW - Organic solar cells
KW - P3HT:PCBM
KW - Pareto frontier
KW - Support vector machines
UR - http://www.scopus.com/inward/record.url?scp=85094318183&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85094318183&partnerID=8YFLogxK
U2 - 10.1016/j.commatsci.2020.110119
DO - 10.1016/j.commatsci.2020.110119
M3 - Article
AN - SCOPUS:85094318183
VL - 187
JO - Computational Materials Science
JF - Computational Materials Science
SN - 0927-0256
M1 - 110119
ER -
