Final Report Summary - PREDMODSIM (Predictive models and simulations in nano- and biomolecular mechanics: a multiscale approach)
In the field of carbon nanostructures, we focused on the mechanics of supported graphene, relevant in most applications. Previous experiments had shown prevalent out-of-plane features including wrinkles and blisters. Because deformation severely alters the electronic structure of graphene and hence most of its properties, these features were generally seen as defects. In this project, we have understood quantitatively and mechanistically the origin of these features. More importantly, we have put forth the idea that wrinkles and blisters can be seen as an opportunity to locally tune the properties of graphene and thus harness functionality out of these “defects”. To realize this idea, we developed strain-engineering strategies to finely control shape and strain distribution in supported graphene.
In the field of biomembranes, we have provided new tools to quantitatively understand and predict their dynamical remodeling, essential in biological function and artificial bioinspired devices. We have also developed tools to consistently compute the stress tensor from MD simulations, enabling the rational design of lipid chemistry and composition to achieve desired mechanical properties. Finally, we have understood how adhered bilayers passively regulate their shape and stress in front of mechanical stimuli by spontaneously developing out-of-plane protrusions. Strikingly, we have demonstrated that living cells adopt the same passive mechanoadaptation strategies as synthetic membranes when subject to stretch. By examining the response to stretch of cohesive sheets of cells (epithelia), we have discovered a new mechanism of epithelial hydraulic fracture.
The field of accelerated molecular dynamics, we have proposed a modeling and simulation methodology by which a complex system is parametrized in terms of essential nonlinear coordinates, identified using statistical learning techniques applied on pre-existing data. This method enables a systematic determination of collective variables in complex molecular systems. In an unanticipated development, we have used these methods to understand the motility of euglenids, a family of unicellular organisms, and to drive inspiration for soft robotics technologies.