Predicting band gap; plasma-resistant ceramics; mixing 2D and silicon materials.
Researchers from Kyoto University developed a machine learning model to predict the band gap of novel semiconductor materials.
Using data from almost 2,000 semiconductor materials, the team tested six different neural networks. They found that the incorporation of conditional generative adversarial networks (CGAN) and message passing neural networks (MPNN) increased the accuracy of band gap predictions.
“Our model enables prediction based solely on the composition of a compound,” said Katsuaki Tanabe, a professor at Kyoto University, in a statement. “The computational load of the ensemble learning model is light and can be performed within a few hours on a typical laptop PC. And we can confidently say that this method enables fast and highly accurate forecasting. We are also developing other ways of interpreting the correlation between the properties of various materials and band gaps.” [1]
Researchers from the Korea Institute of Materials Science and Pusan National University developed transparent plasma-resistant high-entropy ceramics with the potential to extend the lifespan of internal components in semiconductor etching equipment.
Plasma used in the etch process gradually wears away at the equipment’s ceramic components, leading to contamination. The new ceramic has an etch rate of 8 nm/h, resulting in fewer contamination particles and higher durability compared to sapphire and the plasma-resistant ceramic yttria.
The ceramic was created using a sintering process for high-density solid-state materials, resulting in a density of 99.9%. By controlling porosity, it is also capable of transmitting visible and infrared light. [2]
Researchers from the University at Buffalo, Central South University, Shandong Normal University, Sungkyunkwan University, TU Wien, and University of Salerno found that integrating 2D materials like molybdenum disulfide (MoS2) with silicon can improve device efficiency and provide control over how an electrical charge is injected and transported.
“The 2D material mainly affects charge injection, or how the charge enters the material, but doesn’t really affect charge collection, or how the charge exits the material,” said Huamin Li, associate professor in the Department of Electrical Engineering at UB, in a press release. “This happens regardless of the specific properties of the 2D material. So, whether you use semiconducting MoS2, semi-metal graphene or insulator h-BN (hexagonal boron nitride), they can play different roles in the charge injection, but all behave similarly when it comes to the charge collection. Essentially, the 2D material in this special condition acts almost like it’s invisible or has zero resistance for collecting charge.”
Li anticipates that complex devices like three-terminal transistors could benefit from the approach but noted that significant challenges remain in understanding and engineering charge transport where the 2D material meets the 3D material. [3]
[1] Taichi Masuda, Katsuaki Tanabe. Neural network ensembles for band gap prediction. Computational Materials Science (2024). https://dx.doi.org/10.1016/j.commatsci.2024.113327
[2] Yu-Bin Shin, Su Been Ham, Ha-Neul Kim, et al. Novel transparent high-entropy sesquioxide ceramics with high physicochemical plasma etching resistance. Journal of Advanced Ceramics (2024). https://dx.doi.org/10.26599/JAC.2024.9221013
[3] Anthony Cabanillas, Simran Shahi, Maomao Liu, et al. Enormous Out-of-Plane Charge Rectification and Conductance through Two-Dimensional Monolayers. ACS Nano 2025 19 (3), 3865-3877 https://dx.doi.org/10.1021/acsnano.4c15271
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
Leave a Reply