Machine Learning with Enormous "Synthetic" Data Sets: Predicting Glass Transition Temperature of Polyimides Using Graph Convolutional Neural Networks.

In the present work, we address the problem of utilizing machine learning (ML) methods to predict the thermal properties of polymers by establishing "structure-property" relationships. Having focused on a particular class of heterocyclic polymers, namely polyimides (PIs), we developed a graph convolutional neural network (GCNN), being one of the most promising tools for working with big data, to predict the PI glass transition temperature T g as an example of the fundamental property of polymers. To train the GCNN, we propose an original methodology based on using a "transfer learning" approach with an enormous "synthetic" data set for pretraining and a small experimental data set for its fine-tuning. The "synthetic" data set contains more than 6 million combinatorically generated repeating units of PIs and theoretical values of their T g values calculated using the well-established Askadskii's quantitative structure-property relationship (QSPR) computational scheme. Additionally, an experimental data set for 214 PIs was also collected from the literature for training, fine-tuning, and validation of the GCNN. Both "synthetic" and experimental data sets are included into a PolyAskInG database (Polymer Askadskii's Intelligent Gateway). By using the PolyAskInG database, we developed GCNN which allows estimation of T g of PI with a mean absolute error (MAE) of about 20 K, which is 1.5 times lower than in the case of Askadskii QSPR analysis (33 K). To prove the efficiency and usability of the proposed GCNN architecture and training methodology for predicting polymer properties, we also employed "transfer learning" to develop alternative GCNN pretrained on proxy-characteristics taken from the popular quantum-chemical QM9 database for small compounds and fine-tuned on an experimental T g values data set from PolyAskInG database. The obtained results indicate that pretraining of GCNN on the "synthetic" polymer data set provides MAE which is almost twice as low as that in the case of using the QM9 data set in the pretraining stage (∼41 K). Furthermore, we address the questions associated with the influence of the differences in the size of the experimental and "synthetic" data sets (so-called "reality gap" problem), as well as their chemical composition on the training quality. Our results state the overall priority of using polymer data sets for developing deep neural networks, and GCNN in particular, for efficient prediction of polymer properties. Moreover, our work opens up a challenge for the theoretically supported generation of large "synthetic" data sets of polymer properties for the training of the complex ML models. The proposed methodology is rather versatile and may be generalized for predicting other properties of different polymers and copolymers synthesized through the polycondensation reaction.

Structural Ordering in SWCNT-Polyimide Nanocomposites and Its Influence on Their Mechanical Properties.

Using fully-atomistic models, tens-microseconds-long molecular-dynamic modelling was carried out for the first time to simulate the kinetics of polyimides ordering induced by the presence of single-walled carbon nanotube (SWCNT) nanofillers. Three polyimides (PI) were considered with different dianhydride fragments, namely 3,3',4,4'-biphenyltetracarboxylic dianhydride (BPDA), 2,3',3,4'-biphenyltetracarboxylic dianhydride (aBPDA), and 3,3',4,4'-oxidiphthalic dianhydride (ODPA) and same diamine 1,4-bis[4-(4-aminophenoxy)phenoxy]benzene (diamine P3). Both crystallizable PI BPDA-P3 and two amorphous polyimides ODPA-P3 and aBPDA-P3 reinforced by SWCNTs were studied. The structural properties of the nanocomposites at temperature close to the bulk polymer melting point were studied. The mechanical properties were determined for the nanocomposites cooled down to the glassy state. It was found that the SWCNT nanofiller initiates' structural ordering not only in the crystallizable BPDA-P3 but also in the amorphous ODPA-P3 samples were in agreement with previously obtained experimental results. Two stages of the structural ordering were detected in the presence of SWCNTs, namely the orientation of the planar moieties followed by the elongation of whole polymer chains. The first type of local ordering was observed on the microsecond time scale and did not lead to the change of the mechanical properties of a polymer binder in considered nanocomposites. At the end of the second stage, both BPDA-P3 and ODPA-P3 PI chains extended completely along the SWCNT surface, which in turn led to enhanced mechanical characteristics in their glassy state.