Wet-like plasma etching

Researchers from Nagoya University and Hitachi developed a new etch method called wet-like plasma etching that combines the selectivity of wet etching with the controllability of dry etching.

The researchers say the technique will make it possible to etch complex structures such as metal carbides consisting of titanium (Ti) and aluminum (Al), such as TiC or TiAlC, which can be used as metal gates in gate-all-around transistors.

TiAlC is a ternary material with high hardness, high wear resistance, high melting point, and high electrochemical performance. Typically, it is etched by wet etching using hydrogen peroxide liquid mixtures. However, this process requires a long etching time to completely remove the target metals. It also runs the risk of chemically damaging the metal gate. Additionally, the liquids used can create surface tension at the atomic level, destroying important features.

The new dry etching method for metal carbides uses a floating wire-assisted vapor plasma of argon gas mixed with vapor sources of ammonium hydroxide-based mixtures at medium pressure. In the circuit, the plasma is generated by adding energy to the gas, so the additional floating wire can enhance the generation of high-density plasma. Since this process generates active radicals of H, NH, and OH from the ammonium hydroxide gas (NH4OH), the treated surface of TiAlC can be removed after surface modifications of the TiAlC film.

“This floating wire-assisted plasma technique is expected to be available for highly selective etching of metals and metal compounds used in semiconductor device fabrication,” said Kenji Ishikawa, a professor at Nagoya University. “Metal carbides are promising gate electrode materials for advanced silicon semiconductors, and our joint research group was the first in the world to succeed in chemical dry etching of non-silicon semiconductor materials. This achievement is important for the development of atomic layer-level etching technology, which has been difficult to achieve so far.”

Silicon photonic MEMS on chip

Researchers from the University of Sydney, École Polytechnique Fédérale de Lausanne (EPFL), KTH Royal Institute of Technology, Ghent University, Imec, and Tyndall National Institute developed a way to combine optics and microelectromechanical systems (MEMS) in a microchip using semiconductor manufacturing techniques.

“This is the first time that nano-electro-mechanical actuators have been integrated in a standard silicon photonics technology platform,” said Niels Quack, an associate professor in the School of Aerospace, Mechanical and Mechatronic Engineering at the University of Sydney. “It is an important step towards mature large-scale, reliable photonic circuits with integrated MEMS. This technology is being prepared for high-volume production, with potential applications in 3D imaging for autonomous vehicles or new photonic assisted computing.”

The photonic MEMS are compact, consume very little power, are fast, support a broad range of optical carrier signals, and have low optical loss, according to Quack. “The technology will advance knowledge in the field of micro- and nanofabrication, photonics and semiconductors, with a wide range of applications. These include beam steering for LIDAR 3D sensing in autonomous vehicles, programmable photonic chips, or information processing in quantum photonics.”

Neuromorphic device for handwriting recognition

Researchers from the Korea Institute of Materials Science, Chungbuk National University, and Pusan National University developed a high density, high reliability neuromorphic semiconductor device by combining two-dimensional molybdenum disulfide (MoS2) with a lithium silicate (LSO) solid-state electrolyte thin film.

A lithium-ion thin film of less than 100 nanometers was created using a vacuum sputtering deposition method. A transistor-type device was fabricated on a silicon wafer substrate. When an electric field is applied from the outside, the lithium ions in the charged lithium thin film reversibly move so that the conductivity of the channel can be precisely controlled.

The research team implemented an artificial neural network for handwriting pattern recognition using this synaptic device, which performed with a success rate of about 96.77%.

“Our next-generation neuromorphic semiconductor device does not require CPU and memory,” the researchers noted. “It can simultaneously process and stores information and learns and recognizes images such as handwriting patterns. It is expected to be applied to various low-power artificial intelligence devices such as world-class neuromorphic hardware systems, haptic devices, and vision sensors.”