Rust-resistant coating for 2D semiconductors; polymeric material for data storage and encryption; quantum-secure deep learning protocol.
Researchers from Pennsylvania State University, National Yang Ming Chiao Tung University in Taiwan, Purdue University, Intel, and the Kurt J. Lesker Company developed a synthesis process to produce a rust-resistant coating with properties ideal for creating faster, more durable electronics.
“One of the biggest issues that we see in 2D semiconductor research these days is the fact that the materials oxidize quickly,” said Joshua Robinson, professor of materials science and engineering and co-corresponding author of the work in a release. “You need to ensure their long-term reliability because these are going into transistors or sensors that are supposed to last years. Right now, these materials don’t last more than a week out in the open.”
The team looked for a coating material and method that could avoid the use of water entirely and found amorphous boron nitride (a-BN), which is a non-crystalline form of boron nitride known for its high thermal stability and electrical insulation properties. It is therefore ideal for use in semiconductors to insulate components, prevent unwanted electrical currents and improve device performance, Robinson said. [1]
IIT Delhi researchers demonstrated a new polymeric material, called Cyclic Transparent Optical Polymer (CYTOP), with the potential to develop advanced electronic devices for data storage and encryption.
A study showed CYTOP can hold inserted charges for a long duration, which is not generally possible with other materials. This property can be used to write the information at nanoscale in the form of charges, which can further be read by electrostatic force microscopy (EFM) only.
“In the current study, the Atomic Force Microscope’s EFM mode was used to write and read the charges by microcantilever probe on CYTOP. Although this is a very initial step of charge writing capability at nanoscale on an electret but can be extended to the macroscale using motorized stage and probe, like lithographic patterning, which could be end-to-end encrypted,” said Prof. Ankur Goswami, Department of Materials Science and Engineering, IIT Delhi in a release. [2]
MIT researchers developed a security protocol that leverages the quantum properties of light to shield data from attackers during cloud-based computation. The technique guarantees that data sent to and from a cloud server remain secure during deep-learning computations.
The protocol encodes data into the laser light used in fiber optic communications systems to exploit the fundamental principles of quantum mechanics, so attackers can’t copy or intercept the information without detection. Further, the accuracy of the DL models is not compromised.
“Deep learning models like GPT-4 have unprecedented capabilities but require massive computational resources,” said Kfir Sulimany, an MIT postdoc in the Research Laboratory for Electronics (RLE) and lead author of a paper on this security protocol, in a release. “Our protocol enables users to harness these powerful models without compromising the privacy of their data or the proprietary nature of the models themselves.” [3]
[1] Chen, C.Y., Sun, Z., Torsi, R. et al. Tailoring amorphous boron nitride for high-performance two-dimensional electronics. Nat Commun 15, 4016 (2024). https://doi.org/10.1038/s41467-024-48429-4
[2] Singh, Shalini, Stefan AL Weber, Dhiman Mallick, and Ankur Goswami. “Determination of Surface Charge Density and Charge Mapping of CYTOP Film in Air using Electrostatic Force Microscopy.” Langmuir 40, no. 31 (2024): 16330-16337. https://pubs.acs.org/doi/10.1021/acs.langmuir.4c01504
[3] Sulimany, Kfir, Sri Krishna Vadlamani, Ryan Hamerly, Prahlad Iyengar, and Dirk Englund. “Quantum-secure multiparty deep learning.” arXiv preprint arXiv:2408.05629 (2024). https://arxiv.org/abs/2408.05629
Leave a Reply