Combination of Transfer Learning and Chemprop Interpreter with Support of Deep Learning for the Energy Levels of Organic Photovoltaic Materials Prediction and Regulation DOI

Cong Nie,

Kuo Wang, Haixin Zhou

et al.

ACS Applied Materials & Interfaces, Journal Year: 2024, Volume and Issue: unknown

Published: Nov. 20, 2024

It is challenging to build a deep learning predictive model using traditional data mining methods due the scarcity of available data, and model's internal decision-making process often nonintuitive difficult explain. In this work, directed message passing neural network with transfer (TL) chemprop interpreter proposed improve energy levels prediction visualization for organic photovoltaic materials. The established shows best performance, coefficient determination reaching 0.787 HOMO 0.822 LUMO in small testing set after TL, compared other four models. Then, analyzes local global effects 12 molecular structures on After comprehensive analysis level nonfullerene Y-series, IT-series, materials, new IT-series derivatives are designed. 1,1-dicyano-methylene-3-indanone (IC) end group halogenation can reduce varying degrees, while IC modified by electron-withdrawing aromatic groups increase obtain relatively smaller electrostatic potential (ESP) reducing intermolecular interactions. influence side-chain modification limited. worth mentioning that predicted results match density functional theory calculations. also good generalization transferability predicting electronic This work not only provides cost-effective materials but explains bridge between structure properties.

Language: Английский

Designing Donors and Nonfullerene Acceptors for Organic Solar Cells Assisted by Machine Learning and Fragment‐Based Molecular Fingerprints DOI Open Access
Cai‐Rong Zhang, Rui Cao,

Xiao‐Meng Liu

et al.

Solar RRL, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 13, 2025

The molecular structures and properties of donor acceptor materials for organic solar cells (OSCs) determine their photovoltaic performance; however, the complex relationship between them has hindered development OSC materials. To study this, we constructed database comprising 544 non‐fullerene pairs. Based on principle minimal rings units, each molecule in is cut into different fragments defined as a new fingerprint, where bit corresponds to fragment number molecule. Accordingly, fingerprint length 234 723 bits donors acceptors, respectively. Random forest extreme tree regression (ETR) are applied predict parameters, with ETR being most effective. Through SHapley Additive exPlanations (SHAP) importance analysis, eight (10) important (acceptor) identified. Furthermore, by computing similarities that obtained from SHAP similarity exceeding 0.6 collected order design molecules. By assembling fragments, designed 21 168 D‐ π ‐A‐ ‐type 1 156 400 A‐ ‐D‐ ‐A‐type nonfullerene generating 24 478 675 200 donor–acceptor predictions using trained model, highest power conversion efficiency reaches 13.2%.

Language: Английский

Citations

0
Integration of Conductive SnO2 in Binary Organic Solar Cells with Fine-Tuned Nanostructured D18:L8-BO with Low Energy Loss for Efficient and Stable Structure by Optoelectronic Simulation DOI Creative Commons
Mohamed El Amine Boudia, Cunlu Zhao

Nanomaterials, Journal Year: 2025, Volume and Issue: 15(5), P. 368 - 368

Published: Feb. 27, 2025

Enhancing the performance of organic solar cells (OSCs) is essential for achieving sustainability in energy production. This study presents an innovative strategy that involves fine-tuning thickness bulk heterojunction (BHJ) photoactive layer at nanoscale to improve efficiency. The blend D18:L8-BO utilized capture a wide range photons while addressing challenge minimizing optical losses from low-energy photons. research incorporates SnO2 and ZnO as electron transport layers (ETLs), with PMMA functioning hole (HTL). A comprehensive analysis photon absorption, charge carrier generation, localized fluctuations, thermal stability reveals their critical role enhancing efficiency active films. Notably, introducing ETL significantly decreased modified energy, impressive 19.85% optimized 50 nm low voltage loss (ΔVoc) 0.4 V within Jsc 28 mA cm-2 by performing optoelectronic simulation employing "Oghma-Nano 8.1.015" software. In addition, SnO2-based structure conserved 88% PCE 350 K compared room temperature PCE, which describes high this structure. These results demonstrate potential methodology improving OSCs.

Language: Английский

Citations

0
Rational design and DFT-based study of non-fullerene acceptors for high-performance organic solar cells: End-cap and Core modifications for enhanced charge transfer DOI

Adeel Mubarık,

Faiza Shafiq, Xue‐Hai Ju

et al.

Computational and Theoretical Chemistry, Journal Year: 2025, Volume and Issue: unknown, P. 115209 - 115209

Published: March 1, 2025

Language: Английский

Citations

0
Hybrid optimizing optoelectronic properties: structural analysis of silicon and germanium-modified PCPDTBT polymers DOI
Amel Azazi

Optical and Quantum Electronics, Journal Year: 2025, Volume and Issue: 57(5)

Published: April 16, 2025

Language: Английский

Citations

0
Combination of Transfer Learning and Chemprop Interpreter with Support of Deep Learning for the Energy Levels of Organic Photovoltaic Materials Prediction and Regulation DOI

Cong Nie,

Kuo Wang, Haixin Zhou

et al.

ACS Applied Materials & Interfaces, Journal Year: 2024, Volume and Issue: unknown

Published: Nov. 20, 2024

It is challenging to build a deep learning predictive model using traditional data mining methods due the scarcity of available data, and model's internal decision-making process often nonintuitive difficult explain. In this work, directed message passing neural network with transfer (TL) chemprop interpreter proposed improve energy levels prediction visualization for organic photovoltaic materials. The established shows best performance, coefficient determination reaching 0.787 HOMO 0.822 LUMO in small testing set after TL, compared other four models. Then, analyzes local global effects 12 molecular structures on After comprehensive analysis level nonfullerene Y-series, IT-series, materials, new IT-series derivatives are designed. 1,1-dicyano-methylene-3-indanone (IC) end group halogenation can reduce varying degrees, while IC modified by electron-withdrawing aromatic groups increase obtain relatively smaller electrostatic potential (ESP) reducing intermolecular interactions. influence side-chain modification limited. worth mentioning that predicted results match density functional theory calculations. also good generalization transferability predicting electronic This work not only provides cost-effective materials but explains bridge between structure properties.

Language: Английский

Citations

0