From Drug Molecules to Thermoset Shape Memory Polymers: A Machine Learning Approach

Ultraviolet (UV)-curable thermoset shape memory polymers (TSMPs) with high recovery stress but mild glass transition temperature (T_g) are highly desired for 3D/4D printing lightweight load-bearing structures and devices. However, a bottleneck is that high recovery stress usually means high T_g. For a few TSMPs with high recovery stress, their T_g values are close to the decomposition temperature, and thus, the shape memory effect cannot be triggered safely and effectively. While machine learning (ML) has served as a useful tool to discover new materials and drugs, the grand challenge of using ML to discover new TSMPs persists in the very limited data available. Here, we report an enhanced ML approach by combining the transfer learning-variational autoencoder with a weighted-vector combination method. By learning a large data set with drug molecules in a pretraining process, we were able to effectively map the TSMPs to a hidden space that is much closer to a Gaussian distribution. Through this approach, we created a large compositional space and were able to discover five new types of UV-curable TSMPs with desired properties, one of which was validated by the experiments. Our contribution includes (1) representing the features of TSMPs by drug molecules to overcome the barrier of a limited training data set and (2) developing a ML framework that is able to overcome the barrier of mapping the molar ratio information. It is shown that this approach can effectively learn TSMP features by utilizing the relatedness between the data-scarce (and biased) TSMP target and data-abundant drug source, and the result is much more accurate and more robust than the benchmark set by the support vector machine method using direct label encoding and Morgan encoding. Therefore, it is believed that this framework is a state-of-the-art study in the TSMP field. This study opens new opportunities for discovering not only new TSMPs but also other thermoset polymers.

Cyclic behaviors of amorphous shape memory polymers

Cyclic loading conditions are commonly encountered in the applications of shape memory polymers (SMPs), where the cyclic characteristics of the materials determine their performance during the service life, such as deformation resistance, shape recovery speed and shape recovery ratio. Recent studies indicate that in addition to the physical damage or some other irreversible softening effects, the viscoelastic nature could also be another possible reason for the degraded cyclic behavior of SMPs. In this paper, we explore in detail the influence of the viscoelastic properties on the cyclic tension and shape memory (SM) behavior of an epoxy based amorphous thermosetting polymer. Cyclic experiments were conducted first, which show that although the epoxy material does not have any visible damage or irreversible softening effect during deformation, it still exhibits obvious degradation in the cyclic tension and SM behaviors. A linear multi-branched model is utilized to assist in the prediction and understanding of the mechanical responses of amorphous SMPs. Parametric studies based on the applied model suggest that the shape memory performance can be improved by adjusting programming and recovery conditions, such as lowering the loading rate, increasing the programming temperature, and reducing the holding time.

A multiscale approach for modeling actuation response of polymeric artificial muscles

Artificial muscles are emerging materials in the field of smart materials with applications in aerospace, robotic, and biomedical industries. Despite extensive experimental investigations in this field, there is a need for numerical modeling techniques that facilitate cutting edge research and development. This work aims at studying an artificial muscle made of twisted Nylon 6.6 fibers that are highly cold-drawn. A computationally efficient phenomenological thermo-mechanical constitutive model is developed in which several physical properties of the artificial muscles are incorporated to minimize the trial-and-error numerical curve fitting processes. Two types of molecular chains are considered at the micro-scale level that control training and actuation processes viz. (a) helically oriented chains which are structural switches that store a twisted shape in their low temperature phase and restore their random configuration during the thermal actuation process, and (b) entropic chains which are highly drawn chains that could actuate as soon as the muscle heats up, and saturates when coil contact temperature is reached. The thermal actuation response of the muscle over working temperatures has been elaborated in the Modeling section. The performance of the model is validated by available experiments in the literature. The model may provide a design platform for future artificial muscle developments.