Spinodoid metamaterials

A new class of metamaterials

After a decade of periodic truss-, plate-, and shell-based architectures having dominated the design of metamaterials, we introduced the new class of spinodoid metamaterials. Inspired by natural self-assembly processes, spinodoid metamaterials are a close approximation of microstructures observed during spinodal phase separation in, e.g., nanoporous metal foams, microemulsions, and polymer blends. 

Topology generation by in silico mimicking the interfacial topolgy during spinodal decomposition (Kumar et al., 2020)

Additively manufactured spinodoid topology

Spinodoid metamaterials offer several advantages over conventional metamaterials.

Spinodoid metamaterials with seamless anisotropy (from left to right): isotropic, lamellar, columnar, and cubic topologies. (Kumar et al., 2020)

Inverse design enabled by machine learning

We showed that the spinodoid design space can be integrated with an efficient and robust machine learning (ML) technique for the inverse design of (meta-)materials, in particular, uniform and functionally-graded cellular mechanical metamaterials with tailored direction-dependent (anisotropic) stiffness and density. 

We explored the application of this inverse design framework toward synthetic bones and scaffolds. Specifically, the inverse design ML model accurately predicts topologies that match the target relative density and anisotropic elastic stiffness as well as bear geometric resemblance to the natural trabecular bone topologies and achieves this, remarkably, without prior information about bone during the learning stage. 

Currently, we are collaborating with experimentalists towards combining the ML-guided invere design with two-photon lithography to design and fabricate 3D porous scaffolds with prescribed elastic properties. (Coming soon!)

ML-driven inverse design for mimicking trabecular bone (Kumar et al., 2020)

ML-driven two-scale topology optimization for compliance minimization (Zheng et al., 2021)

Toplogy optimization for lightweight spinodoid structures

We presented a two-scale topology optimization framework for the design of macroscopic bodies with an optimized elastic response, which is achieved by means of a spatially-variant spinodoid architecture on the microscale. As a departure from classical FE2-type approaches, we replaced the costly microscale homogenization by a ML-based surrogate model (using deep neural networks). We designed porous structures for minimal compliance with structural features spanning several orders of magnitude in lengthscale.


TU Delft, Mekelweg 2, Delft 2628 CD, Netherlands