The Development of DRAM (1)

The development of DRAM (1)
May, 2022
        

Let's start with DRAM in 1995, (why 1995? Because that was my first year in the DRAM industry). JEDEC had yet to cover DRAM specification discussion (the first DRAM standard by Jedec (JC42) was DDR in the year 2000). At that time 4Mb FPM DRAM just entered mass production. Japan vendors dominated the DRAM industry, the US vendors withdrew gradually, and South Korea vendors with strong government support went all out to catch up. Today, everyone knows that the DRAM industry is a high-capital investment, high-tech, and high-risk industry. However, at that time nobody realized that this industry would be so ruthless. In the couple of years after 1990, many DRAM top-tiered companies fell. This is brutal business competition, and only strong ones could survive. Many large enterprises and business integration with the government’s endorsement rejoined the competition and ended in failure. 

Look at the table right side, the main manufacturers of DRAM in 1995 (this year was also the first year the author entered the DRAM industry). Beware that IBM, INTEL, Qimonda (Infineon), OKI, and 5 manufacturers in Taiwan aren’t covered in this table. 

Let’s look at the DRAM history of Taiwan. In 1990, the Taiwan government started to concentrate investment in semiconductors. And in 2000, it launched the semiconductor industry plan of the two-mega-dollar two-stars, setting the goal of DRAM and TFT LCD as the priority direction. At that time, there were 7 DRAM companies in Taiwan that had started up(the table below missing Shi-Da Semiconductor, founded by Dr. Zhang Rujing in 1997). One of them, VANGUARD INTERNATIONAL, was founded by TSMC investment and later transformed the business into a high-voltage device (BCD) chip foundry. TI-ACER and Shi-Da Semiconductor were both acquired by TSMC in 1999 and transformed the business into chip foundries. PowerChip Semiconductor and PROMOS technology went bankrupt in 2012. Nanya Technology and WinBond turned into niche DRAM market after 2012.

Today, only Samsung, Micron, and SKhynix (Hyundai+LG) are still in the DRAM market, taking up 95% of the market share (about $11 billion USD). The remaining 5% is shared by Taiwan's Nanya Technology, Powerchip, Winbond Electric, and Hefei/China Changxin co.

Note that in recent years, the world's top 5 semiconductor output companies, and oligomers accounted for 95% of the DRAM market; these three are listed.

I have been engaging in DRAM technology for 27 years. I have experienced the evolution of DRAM technologies, and also witnessed that some companies have fallen because of the wrong approach to new technology alternatives, or conservation on technology step ahead. For some reason, the DRAM business has been up and down. I have experienced engineering work in various DRAM major manufacturers, and I would like to take this opportunity to share my personal experience with you. This original content from me may have errors in personal insights and omissions. I also expect that everyone will not hesitate to correct.

In the past 30 years, the technical evolution of DRAM has been a miracle, and I once thought that the evolution of the DRAM process had reached its limits at 70nm (1Gb DDR3), but it turned out that DRAM 1a (roughly 15nm, Micron) 16Gb DDR5 has been mass-produced today, and technology is most likely keeping evolving. However, in the past 30 years, the technology development selection risks are also the same as they were, and in this high-risk, highly competitive market, a single mistake in the direction of technology is enough to bring down a DRAM giant and cause tens of billions of dollars in investment losses. Seeing the process technology competition between TSMC and Samsung in recent years, Samsung has been taking radical technical jumps to try to surpass TSMC, but it has been failing and losing orders, which is very similar to the technology competition of the DRAM industry in the past 30 years, and it is very critical to choose the right timing to jump over the technology obstacles; ahead to do or behind to do cause failure. I'm going to give you a few real-life examples of the important changes and impacts of  DRAM technology, which I'll describe one by one.

• Trench capacitor or Stacked capacitor.
• Innovation of DRAM bit, 8F2 cell to 6F2 cell
• Innovation of DRAM I/O, ASYNC-> SYNC (FP->EDO->SDR)
• Innovation of Sense Amplify and cell array structure
• Innovation of cell access transistor
• Innovation of DRAM Fuse, Laser fuse -> anti-fuse
• Innovation of DRAM package
• Innovation of DRAM test process

 But I want to start with the invention of DRAM  in the 1970s, thanks to Dr. Dennard for his contribution.

To be Continued