Rational Design of High-Energy-Density Polymer Composites by Machine Learning Approach

Abstract

Rationally designing polymer composite structures, including physical parameters of nanofillers, nanofiller−matrix interface characteristics, and geometric distribution of nanofillers, is thought to be an effective approach to achieve the desired dielectric properties such as breakdown strength (E b ), permittivity (ε r ), and energy density (U e ) in wide applications. However, the work is difficult to complete through merely high-cost and timeconsuming trial-and-error experiments. A machine learning (ML) driven approach, trained on hundreds of experimentally measured data, is presented to rationally design polymer composites with desired properties. The doping scheme of nanofillers is fingerprinted with a string of characters considering the physical parameters, shape, distribution of fillers, and shell properties in core−shell structures, and then the Gaussian process regression algorithm is trained to establish the linkage between the filler doping scheme and the dielectric properties. The dielectric properties of the randomly generated tens of millions of candidate composites are calculated with the resulting ML model, and representative composites with high E b , ε r , and U e are presented. The results indicate that the effects of nanofiller permittivity and bandgap on E b and ε r follow exactly the opposite trend, hence it is difficult to simultaneously improve E b and ε r by choosing the type of nanofiller. Fortunately, the trade-off between E b and ε r can be improved by tailoring the shape, orientation, and distribution of the nanofillers, for instance, by using horizontally orientated nanosheets and orthotropic nanowires with high permittivity.

Machine learning models for estimating the AC conductivity mechanism of Edirne Kufeki stone reinforced Polypyrrole composites

“…In accordance with the data given in Figure 10a, the potential barrier heights of each sample have been calculated by Equation (20) and the variation of W b with EKS concentration has been shown in Figure 10b. As is seen from Figure 10b, all potential barrier heights of the composites are higher than that of pure PPy.…”

“…Supervised ML regression algorithm are applied to predict the dielectric properties of polymer composites for energy implementations by Reference 18. Also there are many studies in the literature that use the ML algorithms for the prediction of the dielectric properties of polymer composites and nanomaterials 19–21 …”

“…Also there are many studies in the literature that use the ML algorithms for the prediction of the dielectric properties of polymer composites and nanomaterials. [19][20][21] In the light of all scientific literature mentioned above, this study has focused on proposing supervised ML algorithms that can perform the detailed analysis of the temperature and frequency-dependent conductivity properties of the natural mineral reinforced PPy polymer matrix. Here, Kufeki stone is taken as a natural mineral additive.…”

Review of machine learning‐driven design of polymer‐based dielectrics

“…For example, 1210 experimentally measured permittivity values at different frequencies were manually collected from studies in the literature to train a ML model to instantly and accurately predict the frequency-dependent permittivity of polymers [34]. To aid accelerated discovery of polymer nanocomposites with the desired breakdown strength, permittivity and energy density, we have extracted hundreds of experimentally measured data from the literature to develop a ML model for predicting dielectric properties [26]. Owing to the difficulty in interpreting technical languages, which requires domain knowledge, the use of NLP in collecting material data is still in its infancy.…”

“…By using GPR to determine prediction value and uncertainty, an active-learning framework based on a balanced exploitation/exploration strategy was demonstrated to aid the discovery of polymers possessing high glass transition temperatures [88]. We have also adopted GPR to establish a linkage between the microstructure of nanocomposites and dielectric properties, and to discover doping schemes with desired properties [26]. A multi-fidelity information fusion approach based on GPR was proposed to solve the problems where several datasets have varying levels of accuracy [89].…”