Machine learning of atomic force microscopy images of organic solar cells

Kobayashi, Yasuhito; Miyake, Yuta; Ishiwari, Fumitaka; Ishiwata, Shintaro; Saeki, Akinori

doi:10.1039/d3ra02492j

Cited by 3 publications

(2 citation statements)

References 70 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Recently, a technique for combining AFM images into a ML model in order to determine the PCE of binary (polymer and Non‐Fullerene Acceptor molecule) BHJ‐OPVs was reported. [ 170 ] The study effectively determined vital spatial wavelengths between a large dataset and displayed an approach toward alternative coherent BHJ OPVs. A high level of skill set, experience, and precision is often required in fine tuning AFM scan parameters in order to acquire a high‐resolution AFM image.…”

Section: Applications and Emerging Opportunitiesmentioning

confidence: 99%

Advances in the Use of Atomic Force Microscopy as a Diagnostic Tool for Solar Cells Characterization: From Material Design to Device Applications

Ahia,

Meyer

2023

Physica Status Solidi (a)

View full text Add to dashboard Cite

Considerable efforts in search for an effective characterization technique for photovoltaic devices with utmost precision is on the increase. For precise analysis and tailoring of device performance, a reliable technique is vital. Atomic force microscopy is one of the leading surface analysis techniques of choice for probing surface patterns in a variety of materials with atomic precision using a cantilever. It has evolved as a reliable technique for the investigation of subatomic scale properties of materials such as photocurrent heterogeneity, electromechanical response, charge distribution, molecular weight effects, and many other material parameters. The integration of artificial intelligence hybrid algorithms in atomic force microscope for optoelectronic device fabrication and characterization has increasingly emerged to be desirable due to its reliability and effectiveness in achieving high image resolution, automated analysis, actuation, and the coupling of manufactured units with a precision down to atomic units. In this review, an investigation of topical developments in the use of atomic force microscopy as a diagnostic tool for solar cells characterization is presented with special focus on polymer solar cells, perovskite solar cells, quantum dots‐sensitized solar cells, dye‐sensitized solar cells, fullerene‐based solar cells, III‐V‐based solar cells, and silicon‐based solar cells.

show abstract

Section: Applications and Emerging Opportunitiesmentioning

confidence: 99%

Advances in the Use of Atomic Force Microscopy as a Diagnostic Tool for Solar Cells Characterization: From Material Design to Device Applications

Ahia,

Meyer

2023

Physica Status Solidi (a)

View full text Add to dashboard Cite

show abstract

“…Yiming L.'s team uses machine learning with literature data to develop a prediction model for the electrical parameters of OPVs, and use Shapley additive explanation theory to reveal the main factors affecting the performance, showing the potential of machine learning in OPVs [38]. Yasuhito K.'s team uses AFM images as input to a machine learning model to explore the effect of the bulk heterojunction structure of OPVs on the PCE, and found an important in uencing factor and two novel methods for image analysis [39]. Jiyun Z.…”

Section: Introductionmentioning

confidence: 99%

A universal platform of molecular orbital energy level prediction and molecular design for organic materials

Huang,

Peng,

Liang

et al. 2024

Preprint

View full text Add to dashboard Cite

The design and optimization of organic materials with the specific functions for organic photovoltaic cells (OPV), organic light-emitting diodes (OLED), and organic photodetectors (OPD) with the customized performance are currently the time-consuming and costly process. Therefore, a molecular orbital energy level prediction platform for organic materials is established by utilizing the eXtreme Gradient Boosting (XGBT) algorithm and Klekota-Roth fingerprint (KRFP) in this study. And the prediction performance of prediction platform for predicting the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) of organic materials is characterized, which shows the accuracy is 99.0% and 97.5%, R is 0.88 and 0.93, RMSE is 0.077 and 0.126, MAE is 0.057 and 0.090, and MAPE is 0.01 and 0.025 in the training and test datasets, respectively. More importantly, thirteen key fragments are screened and their impact on HOMO and LUMO in organic materials is analyzed. Apparently, fluoromethane fragments can reduce HOMO and raise LUMO in organic materials, while Cycopropane fragments were observed to elevate HOMO and decrease LUMO. Based on the findings, Y6 molecules is modified to design four new Y6 derivatives, including Y6-DT, Y6-TF, Y6-TDF, and Y6-DFT for adjusting bandgap of organic materials. And the value difference of HOMO or LUMO in the new designed molecules between predicted by the platform and calculated by DFT is only below 5%. It is noteworthy that the platform prediction only costs an average time of 0.1 s. Moreover, this prediction platform also verifies the reported results in OLED and OPD-related literature, showing that the predicted accuracy is higher than 88.1%, the errors are limited to within 11.9%. All of these confirm the establishment of a cost-effective universal platform with high performance for accurately predicting and regulating the energy levels in organic materials.

show abstract

Deep Learning Prediction of Triplet–Triplet Annihilation Parameters in Blue Fluorescent Organic Light‐Emitting Diodes

Lim,

Kim,

Lee

2024

Advanced Materials

View full text Add to dashboard Cite

The triplet–triplet annihilation (TTA) ratio and the rate coefficient (kTT) of TTA are key factors in estimating the contribution of triplet excitons to radiative singlet excitons in fluorescent TTA organic light‐emitting diodes. In this study, we implemented deep learning models to predict key factors from transient electroluminescence (trEL) data using new numerical equations. A new TTA model was developed that considers both polaron and exciton dynamics, enabling the distinction between prompt and delayed singlet decays with a fundamental understanding of the mechanism. In addition, deep learning models for predicting the kinetic coefficients and TTA ratio were established. After comprehensive optimization inspired by photophysics, we achieved determination coefficient values of 0.992 and 0.999 in the prediction of kTT and TTA ratio, respectively, indicating a nearly perfect prediction. The contribution of each kinetic parameter of polaron and exciton dynamics to the trEL curve was discussed using various deep learning models.This article is protected by copyright. All rights reserved

show abstract

Machine learning of atomic force microscopy images of organic solar cells

Abstract: We incorporated atomic force microscopy images of polymer : non-fullerene acceptor organic photovoltaics into machine learning, where fast Fourier transform and grey-level co-occurrence matrix were utilized to predict power conversion efficiencies.

Cited by 3 publications

References 70 publications

Advances in the Use of Atomic Force Microscopy as a Diagnostic Tool for Solar Cells Characterization: From Material Design to Device Applications

Advances in the Use of Atomic Force Microscopy as a Diagnostic Tool for Solar Cells Characterization: From Material Design to Device Applications

A universal platform of molecular orbital energy level prediction and molecular design for organic materials

Deep Learning Prediction of Triplet–Triplet Annihilation Parameters in Blue Fluorescent Organic Light‐Emitting Diodes

Contact Info

Product

Resources

About