Quantitative structure-property relationship (QSPR) framework assists in rapid mining of highly Thermostable polyimides

Yu, Mengxian; Shi, Yajuan; Liu, Xiao; Jia, Qi; Wang, Qiang; Luo, Zheng‐Hong; Yan, Fangyou; Zhou, Yin‐Ning

doi:10.1016/j.cej.2023.142768

Cited by 10 publications

(6 citation statements)

References 55 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…With the advances in machine learning (ML), a surge in the exploitation of polymer informatics has emerged to predict polymer properties from the chemical structure of polymers in recent years, − ,− including traditional ML and deep learning (DL) methods. Traditional ML algorithms, such as support vector machine (SVM), ,, kernel ridge regression (KRR), Gaussian process regression (GPR), and random forest (RF), have been widely used to predict polymer properties.…”

Section: Introductionmentioning

confidence: 99%

Large-Scale Glass-Transition Temperature Prediction with an Equivariant Neural Network for Screening Polymers

Long,

Lu,

Zhang

2024

ACS Omega

View full text Add to dashboard Cite

The practically infinite chemical and morphological space of polymers makes them pervasive with applications in materials science but challenges the rational discovery of new materials with favorable properties. Polymer informatics aims to accelerate materials design through property prediction and large-scale virtual screening. In this study, a new method (Lieconv-Tg) has been developed to predict glass-transition temperature (Tg) values from repeating units of polymers based on Lieconv, which is equivariant with transformations from any specified Lie group. The introduction of equivariance allows the prediction of molecular properties from their 3D structures, independent of orientation and position. A total of 27,659 homopolymers with Tg values were collected from PolyInfo, and a standard data set containing 7166 polymers (named data set_Tg) was created for training a robust Lieconv-Tg model. Using the 3D coordinates as input, Lieconv-Tg performs better than Edge-Conditioned Convolution (ECC), and the mean absolute error (MAE) is significantly reduced by ∼6 from ∼30 to ∼24 on both the validation set and the test set, and the R 2 value for both the validation set and the test set can reach 0.90. Lieconv-Tg is thus used to screen promising candidates from a benchmark database named PI1M with 995,800 generated polymers. However, there are some implausible repeating units in PI1M. To get more reasonable candidates from PI1M, a new filtering method has been accomplished by utilizing Morgan fingerprints at the polymerization points (MF@PP) of repeating units in data set_Tg. The combination of a standard data set, Lieconv-Tg, and a more reasonable screening strategy provides new directions in materials design for polymers.

show abstract

Section: Introductionmentioning

confidence: 99%

Large-Scale Glass-Transition Temperature Prediction with an Equivariant Neural Network for Screening Polymers

Long,

Lu,

Zhang

2024

ACS Omega

View full text Add to dashboard Cite

show abstract

“…Given the large polymer design space, it is difficult, time-consuming, and ineffective to screen polymers with targeted properties (e.g., specific T g ) through experimental procedures. [13][14][15] To enable rapid polymer molecular design and high-throughput screening of ideal products prior to laboratory synthesis and analysis, data-driven alternatives, [16][17][18][19][20][21][22][23] such as the quantitative structure-property relationship (QSPR) modeling [24][25][26][27][28] and machine learning (ML) approaches [29][30][31][32][33] have been successfully used to predict the properties for diverse polymers and discover high-performance polymers.…”

Section: Introductionmentioning

confidence: 99%

Leveraging data‐driven strategy for accelerating the discovery of polyesters with targeted glass transition temperatures

He,

Yu,

Han

et al. 2024

AIChE Journal

Self Cite

View full text Add to dashboard Cite

To overcome the limitations of empirical synthesis and expedite the discovery of new polymers, this work aims to develop a data‐driven strategy for profoundly aiding in the design and screening of novel polyester materials. Initially, we collected 695 polyesters with their associated glass transition temperatures (Tgs) to develop a quantitative structure–property relationship (QSPR) model. The model underwent rigorous validation (i.e., external validation, internal validation, Y‐random, and application domain analysis) to demonstrate its robust predictive capabilities and high stability. Subsequently, by employing an in‐silico retrosynthesis strategy, over 95,000 virtual polyesters were designed, largely expanding the available space for polyester material family. External assessments were performed, highlighting good extrapolation ability of the QSPR model. Furthermore, we experimentally synthesized 10 designed polyesters with predicted Tgs covering a large temperature range from −42.52 to 103.61°C, and characterization results gave an average absolute error of 17.40°C relative to the predicted ones. It is believed that such data‐driven approach can drive future product development of polymer industry.

show abstract

“…The "synthetic" data set contains over 6 million combinations of PI repeat units generated and has been integrated into the PolyAskInG database (http://polycomplab.org/index.php/ru/database. html), 34 which holds great significance for the future development of polymer informatics. Zhang et al 35 collected 652 kinds of PI and used seven machine learning algorithms to build a QSPR model to predict the T g and cutoff wavelength of PI, as well as to extract key feature information on the repeating unit structure.…”

Section: Introductionmentioning

confidence: 99%

Prediction and Interpretability Study of the Glass Transition Temperature of Polyimide Based on Machine Learning with Quantitative Structure–Property Relationship (T_g-QSPR)

Zhang,

Wang,

Chai

et al. 2024

J. Phys. Chem. B

View full text Add to dashboard Cite

The glass transition temperature (T g ) is a crucial characteristic of polyimides (PIs). Developing a T g predictive model using machine learning methodologies can facilitate the design of PI structures and expedite the development process. In this investigation, a data set comprising 1257 PIs was assembled, with T g values determined using differential scanning calorimetry. 210 molecular descriptors were computed using RDKit, and subsequently, six distinct feature selection methodologies were employed to refine the descriptor set. Quantitative structure− property relationship models targeting T g (T g -QSPR) were then constructed using five ensemble learning algorithms and one deep learning algorithm. These models exhibited high predictive accuracy and robustness, with the CATBoost model demonstrating the highest accuracy, achieving a coefficient of determination of 0.823 for the test set, a mean absolute error of 20.1 °C, and a rootmean-square error of 29.0 °C. The study identified the NumRotatableBonds descriptor as particularly influential on T g , showing a negative correlation with the property. Additionally, the model's accuracy was validated using ten new PI films not included in the original data set, resulting in absolute errors ranging from 2.5 to 26.9 °C and absolute percentage error rates of 1.0−12.8%. These findings underscore the importance of utilizing extensive and diverse data sets for predictive modeling to enhance accuracy and stability. Furthermore, exploring the interpretability of the model and experimentally validating newly synthesized PIs have augmented the practical utility of the model. Under the guidance of the T g -QSPR models, it will be possible to accelerate the performance prediction and structural design of PIs in the future.

show abstract

Quantitative structure-property relationship (QSPR) framework assists in rapid mining of highly Thermostable polyimides

Cited by 10 publications

References 55 publications

Large-Scale Glass-Transition Temperature Prediction with an Equivariant Neural Network for Screening Polymers

Large-Scale Glass-Transition Temperature Prediction with an Equivariant Neural Network for Screening Polymers

Leveraging data‐driven strategy for accelerating the discovery of polyesters with targeted glass transition temperatures

Prediction and Interpretability Study of the Glass Transition Temperature of Polyimide Based on Machine Learning with Quantitative Structure–Property Relationship (T_g-QSPR)

Contact Info

Product

Resources

About

Quantitative structure-property relationship (QSPR) framework assists in rapid mining of highly Thermostable polyimides

Cited by 10 publications

References 55 publications

Large-Scale Glass-Transition Temperature Prediction with an Equivariant Neural Network for Screening Polymers

Large-Scale Glass-Transition Temperature Prediction with an Equivariant Neural Network for Screening Polymers

Leveraging data‐driven strategy for accelerating the discovery of polyesters with targeted glass transition temperatures

Prediction and Interpretability Study of the Glass Transition Temperature of Polyimide Based on Machine Learning with Quantitative Structure–Property Relationship (Tg-QSPR)

Contact Info

Product

Resources

About

Prediction and Interpretability Study of the Glass Transition Temperature of Polyimide Based on Machine Learning with Quantitative Structure–Property Relationship (T_g-QSPR)