Prevention of Leakage in Machine Learning Prediction for Polymer Composite Properties.

Machine learning (ML) has facilitated property prediction for intricate materials by integrating materials and experimental features such as processing and measurement conditions. However, ML models designed for material properties have often disregarded a common issue of "leakage," resulting in an overestimation of model performance and a decrease in model transferability. This issue can arise from biases inherent in multiple data points obtained from the same experimental group. We provide a critical examination and prevention method of leakage in property prediction for polymer composites. Our proposed method utilizes data partitioning based on the experimental group to ensure that data from the same group are not mixed in both the training and test sets. Evaluation results highlight that the conventional random partitioning unintentionally inflates ML performance through the misuse of experimental features for leaking data bias within the same experimental group rather than explaining the physical causality. In contrast, the proposed method enables the leakage-free utilization of experimental features to improve prediction accuracy while ensuring model transferability. Specifically, when integrating experimental features with polymer and filler features, the conventional method overestimates the prediction performance of electrical conductivity in reducing RMSE by 26% depending on leakage, whereas the proposed method achieves a reduction in RMSE by 5% without leakage. These findings offer valuable guidance for the effective utilization of experimental features in data-driven materials science.

Polarization change of femtosecond laser pulses in air measured by electric-field-induced second-harmonic generation.

We experimentally demonstrated the polarization change of femtosecond laser pulses in air by using electric-field-induced second-harmonic generation (E-FISHG) for the first time to our knowledge. The polarization change from linear to elliptical was observed at the laser intensity over the filamentation threshold. These results suggest that the polarization change can occur by the birefringence caused by filamentation. This phenomenon can be used for new applications such as an ultra-fast and precise three-dimensional electric field measurement by E-FISHG. In addition, E-FISHG can be an excellent tool to investigate the characteristics of femtosecond laser propagation such as filamentation.

Electric-field-induced second-harmonic generation using high-intensity femtosecond laser pulses over the observable optical breakdown threshold.

We investigated the performance of electric-field-induced second-harmonic generation (E-FISHG) by spectroscopic measurement using high-intensity femtosecond laser pulses. The second-harmonic intensity increased quadratically versus the applied electric field, as expected from the theory, up to 15 kV/cm with the laser energy up to 2.5 mJ, which is ∼5 times higher than the observable optical breakdown threshold. In addition, when the laser energy was 2.8 mJ, ∼80 times signal intensity at 0.23 mJ was obtained. These results suggest that the electric-field measurement by E-FISHG with high-intensity second harmonics is expected by using high-intensity laser pulses above the observable optical breakdown threshold. Spectroscopic measurement shows no E-FISHG of white light generated by self-phase modulation in laser-induced filament.

Multiscale modeling of charge transfer in polymers with flexible backbones.

In order to evaluate the carrier transfer properties in polymers with flexible backbones, we have proposed a simplified multi-scale modeling approach combining molecular dynamics simulations, first-principles calculations and kinetic Monte Carlo simulations. Hole transfer in amorphous polyethylene (PE) is studied as a model system. It is shown that the characteristic length scale of hole localized states in PE is comparable to the Kuhn length of PE, which is the characteristic length scale in terms of the geometric structure. Because a PE oligomer whose length is equivalent to the Kuhn length (C12H26) has a similar electronic structure to amorphous PE, C12H26 is considered as an approximate model of amorphous PE. In line with experimental findings, computed hole mobility in the amorphous PE oligomer is several orders of magnitude smaller than that in crystalline PE. We show that this difference originates from strong hole localization and large site energy variation in amorphous PE due to the conformational disorder of PE oligomer chains. The agreement between experimental and calculated hole mobilities strongly supports the rationale of our modeling approach. The modeling approach proposed in this paper is considered capable of predicting carrier mobilities in polymers with flexible backbones with reasonable computational load.