The verification of new materials’ quality for use in batteries, electronics, and pharmaceuticals is a vital but daunting task. This process previously relied heavily on extensively scanning materials with specialized instruments, which in turn, tends to reduce the speed of innovation due to its expensive and time-consuming nature. Fortunately, the application of artificial intelligence (AI) presents new opportunities for streamlining this process.
An AI tool christened SpectroGen has been developed by researchers at the Massachusetts Institute of Technology (MIT). With its cutting-edge attributes, SpectroGen can simplify and hasten the previously painstaking verification process. This AI-powered virtual spectrometer, as outlined in a study published in the journal Matter, can analyze a material’s spectrum in one mode, such as infrared, and generate an accurate reflection of that spectrum in different modalities like X-ray or Raman.
Remarkably, SpectroGen boasts a 99% match rate with spectra delivered through physical scanning, and it completes this process in under a minute. This is a significant improvement when compared to traditional methods which could take hours or several days to finish.
In the world of spectroscopy, different modalities reveal different properties of a material. For instance, molecular groups are identified by infrared spectroscopy, X-ray diffraction focuses on crystal structures, and Raman scattering reveals molecular vibrations. Conventionally, each modality required its bulky and expensive equipment. However, SpectroGen provides an innovative approach by enabling a single, less pricey instrument like an infrared scanner to generate data for various modalities.
SpectroGen has functional applications that include facilitating a manufacturing facility to scan materials using a single infrared camera. The AI tool then generates the equivalent X-ray or Raman spectra, effectively eliminating the need for multiple scanning setups. The resultant effect is a remarkable reduction in cost and time.
MIT’s interdisciplinary group led by Loza Tadesse, an assistant professor of mechanical engineering at the institute, devised SpectroGen with the primary aim of simplifying and miniaturizing the complex and expensive equipment necessary for spectral analysis. They were inspired by the potential of generative AI in simulating spectral data, which they implemented instead of the computationally intensive modeling of spectra based on atomic and chemical properties.
SpectroGen was developed using a publicly available dataset of over 6,000 mineral samples, many of which included spectral data in multiple modalities. Various hundreds of these samples were used to train the AI, enabling it to recognize the relationships between different types of spectra.
Looking forward, there are exciting potential applications for SpectroGen outside of manufacturing. The team is exploring the use of this tool in rapid disease diagnostics and environmental monitoring in healthcare and agriculture, respectively. This is part of a new project financed by Google. Tadesse is even venturing to commercialize the technology through a startup with a target to bring SpectroGen to a range of industries from pharmaceuticals to defense. “I think of it as having an agent or co-pilot, supporting researchers, technicians, pipelines, and industry,” Tadesse says.
For further details, you may want to check the original article: https://news.mit.edu/2025/checking-quality-materials-just-got-easier-new-ai-tool-1014
This website uses cookies.