Japanese storage, a new hope?

In the 1980s and 1990s, Japan's semiconductor industry was booming and occupied half of the global semiconductor market. In 1989, six of the world's top ten semiconductor manufacturers were Japanese semiconductor companies. However, it is regrettable that by 2023, no Japanese semiconductor company has made it into the top ten.

Memory was once the "pillar" of Japan's semiconductor industry, laying a solid foundation for its dominant position in the global market. However, with the intensification of technological innovation and market competition, Japan has gradually lost its advantage in the storage field. At present, the DRAM market is mainly monopolized by the three giants of Samsung and SK Hynix in South Korea and Micron in the United States. The three companies together account for more than 90% of the market share. Among Japanese companies, only Kioxia (Toshiba's memory division spun off in 2021) still has a certain competitiveness in the NAND flash memory field.

In response to increasingly fierce market competition, Japanese companies have begun to turn their attention to new storage technologies, trying to find new growth points in this emerging field.

When it comes to storage, American companies were the first to take the lead. In the early 1970s, American companies such as Intel and Advanced Memory Systems were the first to launch 1K DRAM, targeting the mainframe market. With its excellent price-performance ratio, Intel quickly captured market share and took the lead.

Although Japan tried to catch up, DRAM, as a "commodity", has a short life cycle and drastic price fluctuations. From 1974 to 2000, the average life cycle of DRAM was only 2-3 years. Therefore, whoever can launch a newer product first will have control over pricing. In the process of catching up, Japanese companies often just finished copying one generation of products, and American companies have already launched the next generation. This situation has been repeated in the development of 1K, 4K to 16K DRAM. Japan can hardly shake the position of the United States in the storage and even computer fields.

The turning point came in 1975, when Japan launched an important VLSI project, namely Very Large Scale Integrated Circuits, with the support of the Japanese government. The VLSI project brought together Japan's five major semiconductor companies - Fujitsu, Hitachi, Mitsubishi Electric, NEC and Toshiba. The project did not end until 1980. It was this project that Japan obtained more than a thousand patents in various parts of the micro-machining process. It enabled Japan to take the lead in two key industries: photolithography technology and silicon crystal technology. Until now, Japan still maintains a leading position in semiconductor manufacturing-related technologies. In 2023, Japanese companies occupied 7 seats in the top 15 global semiconductor manufacturing equipment manufacturers in terms of sales. Japan has an 88% share of the global coating/developer market (Tokyo Electron, Screen Holdings), a 53% share of the silicon wafer market (Shin-Etsu Chemical, Sumco), and a 50% share of the photoresist market (Shin-Etsu, JSR, Tokyo Ohka Industry). The overseas sales ratio of major Japanese semiconductor manufacturing equipment manufacturers exceeds 80%, and many foreign semiconductor companies rely on Japanese-made equipment.

Continuing with storage, the VLSI project also laid the foundation for Japan to be the first to develop 64K DRAM. In April 1977, Nippon Telegraph and Telephone Corporation (NTT) announced that it had successfully developed the world's first 64K DRAM chip. Two years later, NEC, Hitachi, and Fujitsu launched 64K DRAM to the market almost at the same time as American companies, and Japan's storage technology caught up with the United States for the first time. In 1980, Japan took the lead in launching 256K DRAM, two years earlier than the United States.

More importantly, Japanese companies not only gradually surpassed the United States in technology, but also had significant advantages in product quality and cost. In 1980, a study by Hewlett-Packard compared the quality of American and Japanese products and found that the error rate of products produced by the best American companies was still six times higher than that of products produced by the worst Japanese companies. Japanese companies have fully automated semiconductor production lines and introduced computer control in key bonding processes, greatly improving product quality and production efficiency. In contrast, American companies tend to rely on cheap overseas labor to complete this final process.

As Japanese DRAM chips flooded the global market, memory prices plummeted, falling 60% in just one year. By around 1985, most DRAM companies in the United States, except Micron and Texas Instruments, faced bankruptcy and were forced to withdraw from the storage business. By 1987, Japan had occupied nearly 80% of the global DRAM market, becoming the hegemon of the global storage market.

Seeing that Japan's memory was booming, the United States could not sit still. Manufacturers including Micron, Intel, and National Semiconductor filed anti-dumping lawsuits against Japan. In 1986, the Reagan administration of the United States signed the "US-Japan Semiconductor Agreement" with Japan and imposed a 100% tariff on Japanese memory chips. The agreement required Japan to expand the market share of American semiconductor companies in Japan, while restricting the exports of Japanese companies. This measure directly weakened Japan's competitiveness in the memory market and became a turning point in the decline of Japan's semiconductor industry.

As Japan's semiconductor industry declined, South Korea rose rapidly. The South Korean government adopted a support strategy similar to that of Japan, promoting the development of local semiconductor companies through policy support and financing support. Samsung and SK Hynix invested heavily in R&D and production, continuously improved process technology, reduced costs, and quickly made breakthroughs in the market.

Compared with the continuous progress of Korean companies, Japanese storage companies have gradually lost their competitive advantage due to insufficient innovation, market strategy errors and internal management problems. Many Japanese companies have withdrawn from the storage business. For example, Fujitsu has gradually withdrawn from the SRAM and DRAM markets since 2000 and stopped the production of NAND Flash and NOR Flash in 2005. In the storage market, to paraphrase a Japanese saying: America pounding rice, Japan kneading dough, and South Korea eating DRAM cake.

Faced with the decline of the traditional storage sector, Japan's storage giants have begun to seek new development directions.

The rise of new storage technologies is a major direction for the transformation of Japanese storage companies.

With the widespread application of artificial intelligence and digital transformation, the market demand for high-density, high-speed, and low-power non-volatile memory is increasing. Compared with traditional DRAM and NAND Flash, new storage technologies such as MRAM and FeRAM have shown significant advantages in speed, power consumption, durability, etc. These advantages can not only meet the growing demand for data storage, but also provide a broader development space for emerging applications such as the Internet of Things and artificial intelligence.

Among Japanese semiconductor companies, Kioxia (formerly Toshiba Memory) has attracted much attention for its layout in new storage technologies. As early as 2011, Toshiba signed a joint development agreement with South Korea's SK Hynix to jointly develop STT-MRAM (spin transfer torque magnetic random access memory). The core storage element of STT-MRAM is the magnetic tunnel junction (MTJ), whose working principle relies on changing the spin state to achieve data storage. In order to increase storage density and reduce costs, reducing the size of the MTJ has become a key technical goal for achieving high-density STT-MRAM. Although research in this field has been going on for many years, it is still in the research and development stage and has not yet been mass-produced.

The structure and working principle of STT-MRAM

(Source: TEL)

It is reported that Kioxia will showcase its latest MRAM technology at the International Electron Devices Meeting (IEDM 2024) in December 2024. The high-capacity cross-point MRAM technology jointly developed by Kioxia and SK Hynix combines selectors and magnetic tunnel junctions suitable for large capacity, as well as fine processing technology for cross-point arrays, to achieve read and write operations with a minimum unit half-pitch of 20.5 nanometers for MRAM. However, with the further miniaturization of storage cells, reliability tends to decrease. The two companies are trying to improve the performance and stability of MRAM at small sizes by developing a new readout method that exploits the transient response of the selector and reduces parasitic capacitance in the readout circuit. This potential technological breakthrough may lay the foundation for the commercialization of MRAM.

Fujitsu Semiconductor is focusing on FeRAM and ReRAM. Fujitsu started developing FeRAM as early as 1995, and Ramtron first introduced FeRAM to the market in 1996. Ramtron was a fabless company, and Fujitsu provided wafer process foundry services to Ramtron at the time. Fujitsu mass-produced FeRAM in 1999, and has been in mass production for more than 20 years, with a cumulative delivery of 4.4 billion pieces.

Source: RAMXEED

Compared with traditional storage products, FeRAM has a faster read and write speed, because traditional NOR Flash and EEPROM require an erase operation, while FeRAM writes by overwriting, so FeRAM is faster, at the nanosecond level, far exceeding NOR flash and EEPROM. In addition, read and write endurance is also one of the biggest features of FeRAM. Its read and write times are as many as 1013 or 1014, which is equivalent to unlimited times in a sense. Generally, EEPROM and NOR Flash have a limit on the number of writes. Based on these two characteristics, FeRAM has an irreplaceable absolute advantage in real-time writing, power-off protection, and other applications such as electric meters that require a relatively high number of read and write times.

Although FeRAM has advantages in specific market areas, its development is also limited by two major bottlenecks: capacity and cost. The maximum capacity of FeRAM on the market is currently 8Mbit, which is far lower than the capacity requirements of mainstream memory. In addition, the production cost of FeRAM is relatively high, making it difficult to compete in large-scale applications. Currently, there are two major suppliers of FeRAM in the world: Fujitsu and Infineon. Infineon acquired the FeRAM business through the acquisition of CYPRESS (which acquired Ramtron).

In addition, Fujitsu has also made some arrangements on ReRAM (resistive random access memory). ReRAM is a non-volatile storage technology that has attracted much attention. Many people believe that it has the potential to replace NOR Flash, especially when traditional processes are approaching bottlenecks. Currently, Fujitsu is one of the few companies that has achieved mass production of ReRAM, with a maximum capacity of 12Mbit. However, according to Feng Yixin, general manager of RAMXEED (formerly Fujitsu Semiconductor), in a recent forum, to replace NOR Flash, the capacity of ReRAM needs to reach 16Mbit to 1Gb.

Japanese semiconductor equipment giant Tokyo Electron (TEL) is also promoting the industrialization of new storage. Its key research and development of thin film deposition equipment EXIM sputtering system is considered to be crucial to the production of STT-MRAM. The EXIM system integrates the latest technology of thin film deposition and has a magnetoresistance ratio of up to 230%, making the production of large-capacity storage chips possible. The launch of this equipment not only marks TEL's leading position in equipment technology, but also provides strong support for the storage transformation of Japanese semiconductor companies.

Source: TEL

TEL is also working with IMEC, a world-renowned nanoelectronics research institute in Belgium, to conduct joint research on advanced patterning technology. Through this cooperation, TEL hopes to promote the manufacturing of higher-density memory cells and solve the cost and reliability issues of new memory in the mass production process.

The market competition landscape for traditional storage, especially DRAM, has become relatively mature. New types of storage are one of the directions for Japanese companies to make breakthroughs in their niche areas.

Japan has several significant advantages in developing new storage technologies:

• Japanese companies have deep technical accumulation in the fields of materials science and precision manufacturing, which plays a key role in memory research and development. New memory technologies, such as MRAM, FeRAM, ReRAM, etc., involve complex material applications and processes, such as the development and processing of magnetic materials, ferroelectric materials and resistive materials.

• Japan can provide strong support in the field of semiconductor equipment. The manufacture of advanced new storage devices such as STT-MRAM requires high-precision thin film deposition and patterning processes. The equipment developed by TEL as mentioned above provides strong support for Japan's industrialization of new storage technologies.

• Japanese companies have long been in the leading position in traditional storage (such as DRAM and NAND Flash) and have accumulated rich experience in storage technology research and development. They also have a large number of patent reserves in the field of storage technology, which cover multiple aspects such as materials, processes and device structures. Although the traditional storage market share is occupied by Korean and American companies, Japanese companies' accumulation in storage device design, material selection and mass production technology can be directly applied to the development of new storage, lowering the technical threshold and accelerating the development progress.

Although these established companies have made some achievements in exploring new storage areas, they still face many challenges:

Technical barriers: The research and development and application of new storage technologies require a lot of capital investment and technical accumulation. How to balance the development of new technologies while maintaining the existing market share is a major challenge facing enterprises.

Market acceptance: The market promotion of new storage technologies takes time. Customers’ acceptance and adaptability to new technologies are also important factors in the successful transformation of enterprises.

Intense competition: In addition to traditional competitors, these companies also face competition from emerging storage companies around the world, which often have more flexible market strategies and innovation capabilities.

Can new storage technologies be the lifeline for Japan's semiconductor industry to return to its peak? With the exploration and promotion of new storage technologies such as MRAM, FeRAM and ReRAM, Japanese companies have shown their determination to return to the storage market. However, the market is unpredictable and technical difficulties remain. Future success depends on whether Japanese companies can maintain innovation and endurance and find their own position in the global semiconductor competition. Although the road ahead is long, in the wave of new storage, Japan's semiconductor industry is still expected to occupy a place in the future storage market.

Today is the 3936th content shared by "Semiconductor Industry Observer" for you, welcome to follow.

★ Important report on EUV lithography machine released by the United States

Silicon carbide "surge": catching up, involution, and substitution

★ Chip giants all want to “kill” engineers!

Apple , playing with advanced packaging

★ A historic moment for GPU!

★ Continental Group, developing 7nm chips

★ Zhang Zhongmou's latest interview: China will find a way to fight back

★ EUV lithography’s new “savior”