Abstract
The ability to readily design novel materials with chosen functional properties on-demand represents a next frontier in materials discovery. However, thoroughly and efficiently sampling the entire design space in a computationally tractable manner remains a highly challenging task. To tackle this problem, we propose an inverse design framework (MatDesINNe) utilizing invertible neural networks which can map both forward and reverse processes between the design space and target property. This approach can be used to generate materials candidates for a designated property, thereby satisfying the highly sought-after goal of inverse design. We then apply this framework to the task of band gap engineering in two-dimensional materials, starting with MoS2. Within the design space encompassing six degrees of freedom in applied tensile, compressive and shear strain plus an external electric field, we show the framework can generate novel, high fidelity, and diverse candidates with near-chemical accuracy. We extend this generative capability further to provide insights regarding metal-insulator transition in MoS2 which are important for memristive neuromorphic applications, among others. This approach is general and can be directly extended to other materials and their corresponding design spaces and target properties.
Highlights
In materials discovery problems, it is desirable to select and test candidates which hold the most promise for satisfying a particular functional target, while maintaining as broad a diversity in the search space as possible
For our models we implemented the base (3) invertible neural networks (INN)[20] and (4) conditional invertible neural networks[21] as well as the additional steps included in the MatDesINNe framework as (5) MatDesINNe-invertible neural networks20 (INNs) and (6) MatDesINNe- cINN
CINN maintains an appreciable performance of ~0.2 eV for both non-zero gap cases, though the error still remains higher than ideal
Summary
It is desirable to select and test candidates which hold the most promise for satisfying a particular functional target, while maintaining as broad a diversity in the search space as possible To this end, a data-driven approach is often used to meet these needs, whereby materials are first rapidly screened via high-throughput experimentation or computational modeling to identify potential candidates[1,2,3]. Machine learning provides promising solutions to this problem by providing a cheaper surrogate for the computational calculations, or by producing new candidates which have a specified target property[4,5,6,7,8,9] The latter approach may involve the use of generative models, which allows for sampling within a continuous chemical or materials latent space which can map to unique and undiscovered materials[10,11,12,13,14]. More challenging to implement than discriminative models, generative modeling is highly appealing for its potential to realize the “inverse design” of materials and to efficiently “close the loop” between modeling and experiments[4,15,16,17,18]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
