Studies of Shape Memory Graphene Nanostructures via Integration of Physics-based Modelling and Machine Learning

Abstract

In this contribution, we study the properties of new promising graphenebased materials with shape memory effects. While traditional shape memory alloys have been extensively studied, it is a challenge to preserve shape memory properties at the nanoscale. As a result, new materials have been explored, among which graphene oxide (GO) crystals with ordered epoxy groups where a recoverable strain of 14.5% has already been reported. We use such nanoscale GO structures as a benchmark example for our studies here. MBTR and SOAP representations are employed in a general-purpose ML model to analyze the effect of long-range interactions in GO. Finally, a physics-based ML model allows us to build interatomic potentials for 2D and 1D systems. The model predicts quantum mechanical effects due to the electronic confinement in narrow nanoribbons and shows the evolution of the local minimal energies associated with two-phase states.

Machine Learning for Shape Memory Graphene Nanoribbons and Applications in Biomedical Engineering

“…In this work, we use machine learning (ML) to study GO nanostructures, not only to reduce the computational cost involved in the estimation of the two stable phases, but also to analyze the response of GO nanoribbons subject to deformations, and the presence of vacancies and impurities. This work is an extension of [34]. Here, we show different approaches used to improve the model prediction, the approximate critical nanoribbon size for which the shape memory effect is suppressed, and its validation with actual DFT results.…”

“…However, while pushing MLIPs to their limits of the model space where we seek an efficient trade-off sampling between lower computational complexity and high resulting accuracy, problems with phase transformations have particularly serious challenges. In this paper, such problems have been addressed in the context of shape memory GRNs with the MLIP methodology for the first time (with the exception of [34], where a preliminary analysis was reported).…”

“…One of our main motivations to choose the MTP method for ML in this paper was due to the fact it works directly with atomic environments and already makes use of physical restrictions. While for other ML techniques, it would have been necessary to choose first an appropriate representation (see, e.g., [34] and Section 4 for further details), the methodology developed here in Sections 2 and 3 allows a robust trade-off between accuracy and computational cost. Not only we have demonstrated the proof-of-concept as well as the efficiency of the developed MTP algorithm on a class of important problems considered here, we also have provided a series of reproducible illustrative examples.…”

Coupled Multiphysics Modelling of Sensors for Chemical, Biomedical, and Environmental Applications with Focus on Smart Materials and Low-Dimensional Nanostructures

“…Therefore, the development of spectral sensors in this area requires stochastic mathematical models and Bayesian approaches. Recently stochastic methodologies have also been applied to the analysis of graphene nanoribbons with several machine-learning-based techniques [ 68 , 69 ]. Various carbone allotropes (graphite—3D, Graphene-2D, CNT-1D, Fullerene 0D, Diamond 3D, and graphyne [ 35 , 70 , 71 , 72 , 73 , 74 ]) can lead to instructive and potentially viable applications, with graphene-based sensors leading the way in many areas, including human health [ 75 ], which we discuss further in Section 5.1 .…”

“…It is important to emphasize that some of the developed models have to deal with nonlinearities, including strain nonlinearities and non-trivial situations in constructing multiband Hamiltonians [ 28 , 30 , 156 , 182 , 183 , 184 , 185 , 186 , 187 ], which are essential in the design of sensors based on such low-dimensional nanostructures. At the initial stage of the analysis, first-principles methods can provide good guidance, while subsequent optimization of design characteristics and properties in data-driven environments may require data assimilation technique and machine learning algorithms [ 68 , 69 , 174 ].…”