Semiconductor miniaturization, atomic units smaller than nanometers
It continues with electrical issues that were not previously considered
Resistance Reduction, Need to Use Materials Other Than Silicon


Earlier this year, as the slowdown in the supply of automotive semiconductors spread throughout the semiconductor market, interest in foundries reached its peak. Governments from various countries have been courting TSMC and Samsung Electronics, the industry’s No. 1 and No. 2 companies, and each has agreed to build new fabs in Arizona and Texas, USA, respectively, or is planning to do so. The two companies are leaders in semiconductor miniaturization, and are engaged in a fierce technological competition over who will be the first to commercialize the 3nm (nanometer) process in the second half of the year.

The analysis is that semiconductor miniaturization has reached its limit. In the foundry industry, if there is an improvement in performance regardless of the actual device size, the number is lowered to reveal technological prowess starting from the 20nm process or below. The industry currently estimates that the actual device size of the 2~3nm process will be 5nm. Pat Gelsinger, CEO of Intel, who declared his entry into the foundry industry, said, “The nm-based process name does not match the actual device length.”

The limits of miniaturization have become clear. Semiconductors are products that control the flow of electrons. However, as the path becomes narrower, the probability that electrons will escape and appear somewhere else increases. This is a problem called the tunnel effect in quantum mechanics. Humans living in the world of classical mechanics are now faced with the problems of quantum mechanics that they were unaware of.

◇ Miniaturization also overturns the characteristics of existing semiconductor materials

Professor Kwon Hyuk-jun of the Department of Information and Communication Engineering at Daegu Gyeongbuk Institute of Science and Technology (DGIST) stated on the 25th in an article titled “Technology to Improve Electrical Resistance According to Semiconductor Device Scaling,” “As the physical area of semiconductor devices has decreased due to miniaturization, electrical resistance has increased, and the proportion of this resistance affecting the performance of electronic devices has greatly increased.” He also explained, “For semiconductors below 5 nm, approximately 50% of the total RC delay is affected by electrical resistance.”

Semiconductor miniaturization includes not only the elements but also the metal wiring that connects the elements. The contact area of the metal material that first comes into contact with the element and the size of the metal wiring for external connection have been reduced to the nano level. Copper has been used as the metal since 1998. Copper is abundant on Earth and has excellent electrical conductivity. However, due to the problem of electromigration, additional barrier and liner structures are required, and this thickness cannot be increased.

In addition, as the resistivity, which is a characteristic of general metal materials, increases, the problem of resistance also increases occurs. Among metals other than silver, the resistivity value of copper is on the low side, but at the nanometer level, it is higher than cobalt and ruthenium, so its electrical conductivity is also lower than these. The two materials mentioned above are receiving attention as metal wiring materials to replace copper, and research is in progress.

Professor Kwon said that electrical contact resistance from the contact between the device and the metal wiring is also a factor affecting the deterioration of the device performance. The influence of electrical resistance on the 22nm device and the 5nm device increases from 44% to 66%, or from 14% to 35%. In silicon devices, silicide is placed to reduce the electrical contact resistance by increasing it by about 10 to 20 times. However, this method has a physical limitation in that it cannot be reduced below a certain value.

The industry is developing semiconductors that replace silicides with germanium, group III-V compounds, and two-dimensional materials, or that are not silicon at all, such as silicon carbide (SiC) and gallium nitride (GaN). The time when non-silicon semiconductors become mainstream is not far off.

As semiconductors are miniaturized beyond nanometers to the atomic level, the challenge of solving problems arising from physical phenomena that are not commonly encountered, such as the mean free path of metal wiring materials, crystalline boundaries of element materials, and electron tunneling, has been presented. As the demand for and use of semiconductors increases and the pace of technological innovation accelerates, it is necessary to build semiconductor manufacturing capabilities in preparation for the mainstreaming of innovative materials and to focus on achieving EMC of products.