The Institute for Functional Intelligent Materials (I-FIM) will be the world’s first institute dedicated to the design, synthesis, and application of Functional Intelligent Materials (FIMs). Today’s materials are predominantly defined by their properties which are independent of time and environment. In many cases, such stability is useful, as it makes the behaviour of devices based on such materials robust and predictable. However, emergent technologies would strongly benefit from adaptive materials with memory functions and with properties that change through feedback from the environment and control signals. Such materials could monitor external conditions and adapt their functionality to the new set of conditions through built-in feedback. A large subclass of such functional materials are materials out-of-equilibrium, which can change their conformation in a circular manner, producing useful work, and rectifying energy from ambient surroundings. These materials could have memory and evolve over time, in many senses acting as living matter.

These are termed Functional Intelligent Materials. They are crucial for beyond von Neumann computers (neuromorphic), machine-human interfaces, artificial organs and tissues, smart membranes, smart batteries and catalysts, to name just a few.

The global vision of I-FIM is to create Physical Formalism reinforced with Dynamic Machine Learning (called the “Framework” for short) to develop FIMs with predetermined properties and autonomous, dynamic functionalities, responsive to changing environmental conditions. I-FIM will then investigate the applications of such intelligent materials for emerging smart technological opportunities.

The goals for the Institute include:

1. Create a library of designer materials – as well as standardised recipes of their synthesis – and to develop their minimal meaningful mathematical description (reduced representations), as the building blocks of FIMs.
2. Develop the Framework (based on physical and chemical dynamic, out-of-equilibrium models, reinforced with dynamic machine learning) and a Materials Robotic Lab (MRL) that allows the prediction of FIMs behaviour, structure and synthesis pathways.
3. Devise novel smart applications based on such FIMs (neuromorphic computers, machine vision, smart membranes, smart catalysis, artificial tissue engineering, etc.).

The Approaches (organised as 3 Work Packages)

1. Work Package I (WPI) – The Materials

WP1 will develop component materials and (together with WPII) their reduced representations (also known as ‘descriptors’ in AI literature). The choice of FIM components is key to the success of the overall project, thus, materials chosen are capable of strong interaction. As the strongest hetero-interaction usually happens on the surface of the materials, I-FIM’s research will focus on materials where interfaces dominate. Materials are identified in terms of dimensionality: 0D, 1D, 2D, 3D. Heterostructured materials will be combined with living systems (bacteria, cells, etc.) to achieve extra functionality and flexibility, and derive benefits through biological evolution.

2. Work Package II (WPII) – The Framework

WPII will develop the Framework: Physical Formalism reinforced with Dynamic Machine Learning (Dynamic-ML) methodology for inverse material design. It will be tested in the large throughput MRL, which will allow the synthesis of new heterostructured materials with predetermined functionalities. The proposed theoretical framework is a novel combination of physical/chemical computational models, time-dependent ML and optimal control that aims to design materials with desired dynamical properties in the face of limited experimental data and computational throughput. Reduced representation of material components will be used (jointly with WPI) and the developed Framework, Dynamic-ML and robotic synthesis utilised to create new FIMs for specific smart applications in WPIII.

3. Work Package III (WPIII) – Smart applications

WPIII will focus on producing FIMs with predetermined properties and functionalities tuned for several smart applications. As the range of applications of such smart materials is practically endless, applications which are the most impactful and relevant to Singapore will be developed first, maintaining a healthy balance between being ambitious and realistic. These applications can be divided into electronic (e.g. neuromorphic computing, machine vision, smart sensors), energy (e.g. ambient energy harvesting, smart catalysis, nano-batteries), membranes (e.g. ion separation, water treatment, artificial tissues), life sciences, medical and healthcare (e.g. artificial neurons, machine-human interface, artificial photosynthesis).