Researchers have discovered a new way to perform “general inverse design” with reasonably high accuracy. This breakthrough paves the way for the further development of a burgeoning, fast-moving field that could eventually enable the use of machine learning to accurately select materials based on a desirable set of user-specified properties. This may be revolutionary in materials science and has significant industrial benefits and use cases.
The work was led by researchers from the Low Energy Electronic Systems (LEES) Interdisciplinary Research Group at the Singapore-MIT Research and Technology Alliance (SMART), an MIT-affiliated research institution in Singapore, along with collaborators at MIT and the National University of Singapore. Nanyang Technological University.
A major challenge in materials science and research has been the long-awaited ability to create a material or compound with a defined set of properties and properties in order to suit a particular application or use case. To address this issue, researchers have traditionally used material screening via material properties databases, which has resulted in the discovery of a limited number of compounds with user-specified functional properties. However, even with high-performance computing, the computational cost of the necessary computations is high, which prohibits comprehensive research in the theoretical material space. Thus, there is an urgent need for an alternative method that can make this ‘material prospecting’ process more comprehensive and efficient.
Enter the inverted design. As the name implies, the reverse design concept mirrors the traditional design process, allowing new materials and compounds to be “reverse engineered” by simply introducing a set of desired properties and properties and then using an optimization algorithm to generate a predictable solution. The recent emergence of reverse design has been of particular interest in the field of photonics, which is increasingly turning to unconventional techniques to circumvent the inherent challenges associated with designing smaller and more powerful devices. Existing methods involve a traditional design, in which the designer envisions a fixed form or structure as a starting point. This process is labor intensive and excludes from study a wide range of other devices of different shapes or structures, some of which may have greater potential than conventional shapes or structures.
Reverse design eliminates this problem and instead allows devices to be manufactured with the optimal or effective shape, composition, chemical composition, or other characteristics or properties. While reverse design is not new, SMART researchers have taken the technology a step further by discovering a viable method for “general” reverse design, in which the ability of reverse design is not limited to a specific set of elements or crystal structure, but is able to access a variety of of elements and crystal structures.
This breakthrough is illustrated in a paper titled “A reverse crystal representation of the general inverse design of inorganic crystals with target properties”, recently published in the journal Theme. In the paper, the team demonstrates a framework for the general inverse design (diverse in composition and structure) of inorganic crystals, called FTCP (Fourier Transformed Crystal Properties), which allows the reverse design of crystals with user-defined properties through sampling, decoding and post-processing. Most successful, the researchers demonstrated that FTCP is able to design new crystalline materials that differ from known structures – an important development in the exploration of this emerging technology with potentially revolutionary implications for materials science and industrial applications.
The algorithm developed by SMART researchers trains on more than 50,000 compounds in a materials database, then learns and generalizes the complex relationships between chemistry, structure and properties in order to predict which new compounds or materials have user-targeted properties. The algorithm predicts materials with target formation energies, bandgaps, and thermoelectric power factors, and validates these predictions through simulations through density functional theory, which in turn demonstrates a reasonable degree of accuracy.
“This is an incredibly exciting development in the field of materials research. Materials science researchers now have an efficient and comprehensive tool that allows them to discover and create new compounds and materials simply by introducing the desired properties,” says Tonio Bonassisi, principal investigator at LEES and professor of mechanical engineering at MIT. for technology.
S. Isaac P. Tian, a graduate student at NUS and co-first author on the paper adds, “In the next step of this journey, an important milestone will be to improve the algorithm to be able to better predict stability and fabricability. These are exciting challenges the SMART team is working on. It is currently being resolved with collaborators in Singapore and globally.”
Zekun Ren, lead author and postdoctoral researcher at LEES, says, “The goal of finding more efficient and effective ways to create materials or compounds with user-defined properties has long been a focus of materials science researchers. Our work demonstrates a viable solution beyond specialized reverse design, This allows researchers to explore potential materials of varying composition and composition and thus enable the creation of a much wider range of compounds. This is a pioneering example of successful generic inverted design, and we hope to build on this success in additional research efforts.”
The research is conducted by SMART and supported by the National Research Foundation (NRF) in Singapore within the Campus Program for Research Excellence and Technology Enterprise (CREATE).
SMART’s LEES multidisciplinary research group creates new integrated circuit technologies that lead to increased functionality, lower power consumption, and higher performance for electronic systems. In the future, these integrated circuits will influence applications in wireless communications, power electronics, LED lighting, and displays. LEES has a vertically integrated research team with expertise in materials, devices and circuits, consisting of several individuals with professional experience in the semiconductor industry. This ensures that research is targeted to meet the needs of the semiconductor industry, both within Singapore and globally.
SMART was created by MIT and NRF in 2007. SMART is the first entity in CREATE developed by NRF. SMART acts as a think tank and innovation hub for research interactions between MIT and Singapore, conducting cutting-edge research in areas of mutual interest. SMART currently includes an Innovation Center and five interdisciplinary research groups: Antimicrobial Resistance, Critical Analytics for Custom Medicine Manufacturing, Disruptive and Sustainable Technologies for Agricultural Precision, Future Urban Mobility, and LEES.
.