Academician Ma Zuoping: Development of IC Industry in China

I am particularly pleased to be able to return to Tianjin to discuss with you because I returned to Tianjin in 1995 when I was hired as an honorary professor at Tianjin University. From that year onwards, I had deep feelings for Tianjin and I was free. Back to Tianjin. So this time I am very happy to be able to come back to Tianjin.

I would like to share my opinion with you that we are now in the Silicon Age. Historically, what kind of device we have had the greatest impact on our lifestyle has been summarized into that era, such as the Stone Age and bronze artifacts. In the era and the Iron Age, with this measure, I think we are now in the Silicon Age, because the effect of silicon on our lives must be exceeded, at least with the influence of previous stone tools, ironware, and bronzeware.

Our Silicon Age should only begin in the 20th century. If young scientific workers or students are asking about what areas they will enter in the future, I think that electronics is very good in the field of silicon.

In most developed countries, electronics is the largest industry, and it is the fastest growing industry. Now we look at the data. Not long ago, the automotive industry was the largest of all industries, but the electronics industry has surpassed the automotive industry many years ago. Integrated circuit ICs have surpassed the steel industry a few years ago, and its growth rate is faster. The electronics industry accounts for more than half of the world's GDP, so the electronic information industry is a very important industry.

The foundation of the electronics industry is semiconductor technology. A breakthrough was invented in 1948. The transistor was invented. At Bell Labs, these three people were Bardeen, Brattan, and Shockley. They were the two in the lab, Bardeen and Brattan, but Shockley was the boss and they passed it to Shockley. We proved that the transistor works, and Shockley immediately called back. Don't tell anyone, I'll be back. After returning, a grand press conference was held. Major newspapers and radio stations came. Shockley sat in front of the experimental invention. Both of them stood behind. This picture is very famous (Figure), they invented the transistor, and then the inventor of the Nobel Prize was three.

Recall that in 1950, the first vacuum tube computer, which was completely electronic, was previously a lot of mechanical, hand-shake. But this electron is a big breakthrough. The computer, which has more than 3,000 gates, was made with vacuum tubes. At that time, the US Department of Defense was mainly responsible for the development of the research department to calculate the accuracy of the trajectory. At that time, the calculations were very fast, but on average, the machine was shut down every 10 minutes. There was probably a broken vacuum tube. The staff desperately searched for something that was broken. There are now billions of transistors in it, so this is impossible, if not always improving.

The first IC, more than a dozen years later, was Jack Kilby's (pictured) connection of several transistors, capacitors, and electroforms. Later in his patent talked about what will happen in the future, in fact he made very rough. In the same year, in 1959, Robert Noyce, a founder of Intel, and his invention, said that the ancestor of IC research should be Robert, not Jack Kilby, but Jack Kilby received the Nobel Prize and Robert Noyce did not get it. Guess why Because it died, the Nobel Prize passed away, and the Nobel Prize was not issued to the dead.

Therefore, there are several Nobel Prizes here, and we can see that their common ground is in different areas. The common point is that everyone is very healthy. Therefore, young people who want a Nobel Prize must be physically better.

In more than a dozen years, the first 1,000-bit DRAM came out, and Intel was DRAM-born. Please note that at that time, one thousand yuan of DRAM 21 blocks. A few years later, 16,000 yuan came out, 3 blocks, then 64,000 out of 47 blocks, 64,000 dollars has a very important point, Intel has made 64,000 When they were a bit, they said they were far ahead, but they had a very smart engineer who also studied physics. At that time, he published an article saying that 64,000 yuan was the limit and couldn't stop. They found that There was a phenomenon that the 64,000 yuan was unstable. Later they discovered that they were packed on the outside with expensive metals. There were some heavy metals inside, and the heavy metals were radioactive and would change 1 to 0. The Japanese did not see the article at that time, but also did not know what he was talking about, so he continued to go down. He said that we did not see you. The original Japanese used very cheap plastic packaging. There was no heavy metal in it and there was no such phenomenon. Later, Japan became the leader of DRAM. But on the contrary, after Intel concentrated on doing the MHz, its profit was very high and there were few competitors. Therefore, it has always dominated the semiconductor industry. It is very headache to do DRAM, there is a cyclical, sometimes DRAM big compensation, sometimes large earn. Intel made a good decision, but its reason is not correct.

This is Intel's transformation of microprocessors, to NMOS, began to use more, 8-bit, to 16-bit processor, Intel's 8088, this is a historic moment, when IBM's first This is what PCs use. Personal computers have been developing very quickly since then. Again, older people probably used 286, 386, and 486. That's Intel 80286. At that time, 10 megabytes were fast. After 10 years, 386, 486, and Pentium, then 66MHz, Pentium Pro. , 200 and MHz, Pentium 2, Pentium 3, Pentium 4, then more than 1 GH2, which is an important milestone in the history of the United States.

The progress of the semiconductor industry has followed Moors' law for the past 50 years. The number of on-chip devices, especially the number of transistors, has increased by 4 times every three years. Therefore, the speed, cost, reliability, and so on, will all be exponential. increase.

Now our technology can reach up to 10,000 transistors on a ballpoint pen tip. We use dynamic access to make an analogy, at 64,000 yuan, do not remember Intel engineers do not say this thing is not alright, so far, Intel will turn, the Japanese took, dynamic access memory chip Capacity, by 2010 can already exist in the Encyclopedia Britannica, 4GB.

How much can DRAM sell per 10,000 yuan? Now you can buy a 256mb for a penny and buy a bubble gum in 10 minutes in the United States. A Taiwanese professor invented the law of π, saying that the price of each generation of DRAM will fall to 1.5 US dollars after three years. Why is it called π law? In fact, it can be said that the 3 money law, because it sounds too unlearned. The π law is very important to Moore's Law, because each center wants to be the first to come out. Because Moore's Law is known to everyone, it's the first time you'll earn 25 times to sell, but after a year, you might be able to sell 6, 7 blocks, cost 3 blocks, you can still earn twice, in another year, you will be tied, cost 3 blocks, sell 3 blocks, and in another year, you will do it again in the third year, cost 3 blocks. You can only sell 1.5 yuan, so everyone must be the first to take the lead, Moore's Law will be maintained.

Maybe you will ask why, you already know that you don’t make money. In the third year I knew I could only collect 1.5 yuan. I would like to do it in 3 blocks. It’s not enough. For example, if you want to build a factory, it will cost 3 billion to 4 billion U.S. dollars. The value of this plant after five years is almost equal to zero because five years later it will not be able to keep up. The new generation will be replaced by two generations, so your value is zero. If you do, the annual fixed cost will be 600 million yuan.

Whether you don't produce or not, your cost is 6 billion. Your production, plus the cost of engineers, plus other materials, up to tens of millions, so your cost is at least 600 million. If you don't go to production, maybe you If you want to lose 600 million, your production may only lose 300 million, so you have to work hard to produce. This is the interaction of π law and Moore's law.

At present, China's IC market has three characteristics: First, the market is the world's largest, second is the world's highest growth rate, and the third is IC's foreign trade deficit in China.

I think the opportunity for China's IC industry is phase change memory (PCM), which is the only new mass-produced memory. Compared with flash memory, PCM is faster, more reliable, smaller, and more power-efficient.

In the next decade, PCM will replace a large part of flash memory, and its annual output value will exceed 50 billion US dollars. China does not have the burden of traditional memory. International semiconductor companies will not easily adopt new technologies to replace their existing technologies. Phase change memory can give new opportunities to China's IC industry. It is China's achievements in the world of IC.