"Machine learning is impacting all areas of materials research," said Carl V. Thompson, director of the MIT Materials Research Laboratory.
Elsa A. Olivetti, associate professor of energy studies at Atlantic Richfield, expressed that the collaboration between humans and machine learning is becoming more common. Machine learning can guide humans in creating new materials through synthetic methods.
Porous materials, such as zeolites, are composed of interconnected or enclosed pores that form a network structure. Compared to continuous medium materials, porous materials generally have the advantages of low relative density, high strength, high surface area, light weight, good sound insulation, heat insulation, and permeability. They are excellent materials for scientific research.
Rafael Gomez Bombarelli, assistant professor at OrvieTi and Toyota Materials, collaborated to better understand a porous material called zeolite using machine learning. Zeolite, made of silicon and aluminum oxide, has a wide range of applications, from cat litter to petroleum refining.
Professor Bombarelli said, "Although engineers know that there are about 250 types of zeolites, physicists have calculated thousands of possible ways to create porous materials. You can even mine a zeolite and subject it to different pressures or heat it at different temperatures, and it will transform into a different zeolite. This variability is more valuable for specific applications."
Currently, the traditional method of manufacturing zeolites treats the crystal structure as a combination of components. However, during the analysis of zeolite transformations, more than half of the time, the original zeolite before the change and the new zeolite after the change do not have common building blocks. Bombarelli represented each zeolite framework structure as a graph and used artificial intelligence to match these graphs, discovering that certain types of transformations only occur between zeolites with the same structure. This work was developed through data mining of 2.5 million materials science journal articles by Olivetti, with the aim of discovering methods for manufacturing different inorganic materials.
Chemical synthesis is the foundation and core of disciplines such as organic chemistry, inorganic chemistry, medicinal chemistry, polymer chemistry, and materials chemistry. It usually involves a series of chemical reactions aimed at obtaining one or more products, which is crucial for high-tech materials.
Warren K. Lewis, professor of chemical engineering, and Klavs F. Jensen, professor of materials science and engineering, described a chemical synthesis system that combines AI-guided processing steps with modular reaction systems operated by robots.
According to Jensen, the AI system sifted through 12.5 million reactions and created a rule library from approximately 160,000 commonly used synthesis formulas. This machine learning method can provide suggestions to experimentalists, such as what catalysts, solvents, or reagents to use in a reaction.
The system consists of a processing tower with interchangeable reaction modules and a set of different reagents that are connected by robots for synthesis.
By predicting the synthesis plans of 15 drugs or drug-like molecules and utilizing the system to manufacture drugs, researchers demonstrated the contrast between flow reactor systems and batch processing systems. Jensen explained, "To accelerate reactions, it is inevitable that we use more aggressive conditions than batch production. The modular system consists of a processing tower with interchangeable reaction modules and a set of different reagents that are connected by robots for synthesis."
These findings were published in the journal Science. The unique combination of hardware and algorithms in the new spectrometer structure can enable a wide range of applications, from industrial process monitoring to medical imaging.
Juejun "JJ" Hu, professor of materials science and engineering, combined a silicon chip spectrometer for detecting infrared wavelengths with a newly created machine learning algorithm.
Juejun's version of the spectrometer differs from traditional ones, as it collects all the light on a single detector, maintaining the intensity of the light.
To address the trade-off between spectral resolution and signal-to-noise ratio, a new type of spectrometer called wavelength-division multiplexing spectroscopy is needed. Juejun's new spectrometer structure, called digital Fourier transform spectroscopy, integrates tunable optical switches on a silicon chip. The device works by measuring the intensity of light under different switch settings and comparing the results.
A prototype device with six switches supports a total of 64 unique optical states, providing 64 independent readings. The advantage of this new device architecture is that performance doubles with the addition of each new switch. Juejun collaborated with a researcher from MIT to develop a new algorithm called Elastic D1, which achieves a resolution of 0.2 nanometers and requires only two consecutive measurements to accurately measure light.
"We believe that the unique combination of hardware and algorithms in the new spectrometer structure can enable a wide range of applications, from industrial process monitoring to medical imaging," Juejun confidently stated during a presentation.
AI and machine learning have made their presence felt in various fields such as materials science, biology, chemistry, and optics. These applications not only advance the development of each field but also contribute to the development of new algorithms, creating a win-win situation for AI and machine learning themselves.
Reference links:
Science Journal: https://dx.doi.org/10.1126/science.aax1566
Sources: Google News, MIT Official Website, Baidu Baike, Venturebeat."
Alphabet Q1 paid subscribers reached 350M, fueled by YouTube & Google One. YouTube ad revenue slightly missed at $9.88B amid Premium growth. Total Q1 revenue hit $109.9B, beating forecasts.
Anthropic releases Claude connectors for Adobe, Blender, and Ableton, enabling AI workflow automation and app control. It also sponsors the Blender Development Foundation.
Apple introduces an App Store model splitting 12-month plans into monthly installments. On iOS 26.4+, it lowers upfront costs without reducing total price, adding billing tracking.
Google launches Aadhaar digital IDs in Wallet for India, enabling secure QR/NFC verification. A phased partner rollout targets key sectors, while passport-based credentials expand globally.
YouTube tests "Ask YouTube," an AI tool blending text and video for complex queries. Initially for US Premium users, it supports interactive follow-ups, expanding Google's AI search.
Google grants US DoD classified AI access, joining OpenAI and xAI in defense deals amid Anthropic's dispute. Contractual safeguards against surveillance and lethal weapons remain unenforceable.
Mistral AI launches Workflows, a Temporal-powered orchestration platform for secure, Python-driven enterprise AI automation featuring human-in-the-loop approvals.