In the process of industrial technology advancement, research and development (R&D) is the foundation of technological innovation, and collaborative research across industry, academia, and research institutions is the only way to overcome challenging challenges. Recently, promising new industrial technologies were announced in the semiconductor and display-related small and medium-sized equipment sectors, and a forum was created to share R&D results.

The Korea Institute of Industrial Technology Evaluation and Planning (KIET) held the 3rd Industrial Technology R&D Promising Technology Presentation at the COEX Conference Room on the 23rd. This event, which gathered promising new technologies in the materials, components, and equipment fields, presented presentations in various fields such as semiconductors, displays, secondary batteries, and 3D printers.

In the semiconductor field, Kim In-cheol, a senior researcher at the Green Carbon Research Center of the Korea Research Institute of Chemical Technology, presented a ‘nanoporous membrane filter with improved impurity removal efficiency,’ and Professor Woo Ji-yong of the Department of Electronic Engineering at the Industry-Academic Cooperation Foundation of Kyungpook National University followed up with a presentation on ‘3-terminal neuromorphic synapse device and manufacturing method.’

■ Nanoporous membrane filter minimizes defect rates in CMP processes by removing impurities.

As semiconductor processes become increasingly miniaturized and line widths decrease, removing impurities from photoresist solutions is emerging as an important task for improving yield. Impurities are mainly composed of metal particles and organic aggregate particles, and these particles cause microscopic scratches and result in pattern defects.

Dr. In-cheol Kim of the Korea Research Institute of Chemical Technology explained, “It is difficult to sufficiently remove impurities from photoresist solutions with currently commercialized nylon and polyolefin membranes.”

This impurity removal research was supported by Samsung Electronics and the Ministry of Trade, Industry and Energy. The company stated that it was conducted in response to an incident in the past where Samsung Electronics had to scrap all of its imported photoresist solution due to remaining impurities.

He pointed out that while hard particles in solutions can be removed through 2-nanometer PE filters, soft particle removal is limited. Dr. Kim noted, "Even companies developing 2-nanometer filters in the United States have a soft particle removal rate of only 7%." Dr. Kim stated that the membrane filter he developed could achieve a nanoparticle removal rate of up to 97% when using nylon 66.

Dr. Kim explained the special features of membrane filter technology capable of removing soft particles, saying that the technology has reached TRL Level 5, which includes prototype production and performance evaluation.

The gel particle removal performance of a membrane for purifying a photoresist solution can be easily evaluated using palladium nanoparticles, and the water treatment membrane has a nano-sized pore structure, so it has excellent chemical resistance and permeation flux, and can be used in various water treatment devices, including semiconductor process water treatment devices.

In addition, the efficiency of removing impurities was greatly improved by manufacturing a membrane with nano-sized pores formed using a hydrophobic support layer with large pores and multiple polymers. The surface of the hydrophobic support layer could be made hydrophilic by filling the pores existing on the surface and cross-section of the hydrophobic support layer.

It is reported that this technology can be applied to △PR purification for semiconductor processes, △CMP slurry purification, △chemical-resistant filters, and △membranes and filters for water treatment in the future.

■ 3-terminal neuromorphic, CMOS process-friendly new material introduced to improve recognition rate


Existing two-terminal, next-generation memory-based neuromorphic semiconductors have limited their ability to improve recognition rates due to nonlinear changes in analog resistance. To address this, a three-terminal neuromorphic semiconductor utilizing mobile ion motion has been developed, enabling linear analog resistance changes.

Professor Woo Ji-yong of the Industry-Academic Cooperation Foundation at Kyungpook National University said, “In order to induce resistance changes, the type of mobile ion and material must be suitable for the semiconductor process, but lithium is not suitable,” and “In our three-terminal neuromorphic research, we focused our research on discovering materials suitable for the semiconductor process.”

We have found materials compatible with CMOS semiconductor processes, such as those utilizing copper mobile ions, and applied them to device structures. We have also completed performance verification of the source material, process, and device technologies that control the injection and movement of copper ions at the laboratory scale, which is TRL Level 3.

It is possible to secure ultra-low power analog information characteristics compared to S-RAM and D-RAM, which are volatile memories that only hold information of 0 and 1. Professor Woo said, “In implementing AI, S-RAM is bulky because it is made with 3 to 6 units, and D-RAM is volatile, so it loses data when there is no power, so there is a big movement to implement next-generation AI semiconductors through non-volatile memory.”

Professor Woo's next-generation non-volatile memory, developed by Professor Woo, is expected to implement artificial intelligence in a single device, offering improved performance with superior area and power efficiency. The next-generation non-volatile memory also possesses high linearity, which is expected to be advantageous in accelerating neural network learning algorithms.

It is expected that the implementation of this ultra-low-power, ultra-high-speed, real-time object recognition system will enable the implementation of autonomous vehicles with minimal battery consumption and high stability, and that it will be applicable to various smart devices such as displays, healthcare equipment, and robot sensors.

According to Gartner, the global AI semiconductor market is expected to grow at an average annual rate of over 20%, reaching $23 billion (approximately 30 trillion won) in 2020 and $70 billion (approximately 93 trillion won) in 2026. Communications accounts for more than half, or 63%, of AI semiconductor applications, followed by computing at 26%.

Meanwhile, at the presentation held that day, presentations were made on promising technologies in various fields, including: △a method for manufacturing polyether polyol with easy control of structure and molecular weight; △a separator capable of suppressing lithium dendrites and a lithium metal battery using the same; △a method for manufacturing graphene-metal composites; and △a high-resolution OLED light-emitting layer pattern technology for AR/VR.