Experiment‐Oriented Machine Learning of Polymer:Non‐Fullerene Organic Solar Cells

Abstract

Despite the capacity of conjugated materials for enhanced power conversion efficiency (PCE) of organic photovoltaics (OPV), a comprehensive survey of unexplored materials is beyond the reach of most researchers’ resources. In such instances, a data‐driven approach using machine learning (ML) is an efficient alternative; however, bridging the gap between experimental observations and data science requires a number of refinements. In this investigation, using a random forest model based on an experimental dataset, a high correlation coefficient of 0.85 is achieved for the ML of polymer and non‐fullerene small molecule acceptor OPVs and performed virtual screening of 200,932 conjugated polymers generated by the combinatorial coupling of donor and acceptor units. Further, to evaluate the effectiveness of the ML model, a series of conjugated polymers (based on benzodithiophene and thiazolothiazole) were designed, synthesized, and characterized with different alkyl chains. Among these, PBDTTzEH:IT‐4F showed a PCE of 10.10%, which is in good correspondence with ML predictions with respect to the choice of alkyl chains. Thus, the current study demonstrates how ML can be utilized for developing OPVs using a relatively small number of experimental data points (566) and screening numerous molecular structures.

Machine Learning-Assisted Development of Organic Solar Cell Materials: Issues, Analyses, and Outlooks

“…Min et al constructed an RF model to screen NFA, obtained a good r value of ∼0.70 between the experimental and predicted PCE and reported the synthesis of modified Y6 NFA (experimental PCE: 6%–15%) . We conducted an ML study of polymer:NFA based on the 566 experimental data points and screened virtually generated ∼200 000 polymers . The RF model showed an improved r of 0.85 ± 0.02, which was comparable to those obtained using other algorithms (support vector machine, SVM, r = 0.85; kernel ridge regression, KRR, r = 0.84; k -nearest neighbors, k -NN, r = 0.81; gradient boosting regression, GBR, r = 0.79) except for ANN ( r = 0.59).…”

“…The data set was updated from our previous report (the number of data points (#) = 566, the number of papers = 253, which is up to 2018; thus, Y6 of NFA is not included) by continuing the data collection from the literature (up to the middle of 2021; thus, Y6 and its analogues are included). The resultant data set with 1318 data points from 558 papers (Supporting Information) was subjected to ML.…”

Impact of substituents on the performance of small-molecule semiconductors in organic photovoltaic devices via regulating morphology

“…[163][164][165][166] We also note that the use of computer-aided approaches should be widely extended to accelerate the understanding of the structuremorphology-performance relationship of OPV materials. There have been already several studies in this context, [167][168][169][170][171][172][173] and research efforts in this area will aid achieving PCEs nearing the theoretical limit.…”

Designs and understanding of small molecule-based non-fullerene acceptors for realizing commercially viable organic photovoltaics

“…In addition, computational studies using machine learning have recently demonstrated powerful techniques for understanding and predicting the physical properties of the relevant materials and blends while reducing both the time and the cost. 203,204 Therefore, a combination of these two techniques is expected to provide a better understanding of the fundamentals in view of designing more efficient and stable OPVs. View Article Online…”