Young got carried away and couldn't stop talking. He wanted to have a good chat with Cheng Gang about his understanding of semiconductors and the development of domestically produced semiconductor alternatives.
"Now, let's get back to our main topic: what are the challenges of 14nm lithography machines?"
"Let's talk about FinFET first. FinFET is a 3D MOSFET device structure first proposed by Professor Hu Zhengming, a famous professor at the University of California, Berkeley."
“3D MOSFETs achieve performance superior to previous generation planar transistors through their three-dimensional structure. Intel was the first to commercialize them in 2011. However, with the 14nm process, FinFETs have become less reliable than before.”
"First, there's the inescapable quantum effect; it's difficult to increase the current of transistors. In fact, for some materials, the current will be lower. For chips, current means speed."
"Then there's leakage current, which directly affects power consumption. At such a small node, it's also difficult for FinFET to control leakage current."
"Finally, there's the cost. According to Moore's Law, as the number of nodes decreases, the cost of a single transistor should decrease."
"But with the 14nm process, the complex photolithography process may actually increase the cost of a single transistor."
"Why is SMIC able to handle the 14nm process, but its stock price has remained stagnant?"
"It's because the 14nm process is too difficult that SMIC's yield rate is not high enough, so it's only half-finished."
Cheng Gang nodded and said, "No wonder SMIC no longer discloses the revenue share of 14nm and 28nm processes separately, but instead continues to expand 28nm production."
Young was very satisfied with Cheng Gang's keen insight: "Even in the two years after announcing the mass production of 14nm process, SMIC still counted the revenue of 14nm and 28nm together, and now it doesn't even disclose the revenue ratio anymore."
(After examining SMIC's financial reports from 2020 and 2021, it was found that its largest source of revenue was the 55/65nm process, while the revenue share of 14/28nm was 7.8% in the first quarter of 2020, but decreased to 6.9% in the first quarter of 2021.)
Note: This only means that SMIC's 14nm process is not as good as expected, not that SMIC has failed to develop a 14nm process.
Young continued, "Chip manufacturing is less difficult than lithography machines."
"If SMIC's 14nm process is already so difficult, then 14nm lithography machines will only be more difficult."
"The fatal problem will be encountered in the foundry process, and the lithography machine is no exception."
"A chip requires hundreds of masks and photolithography processes to complete the manufacturing process."
"Therefore, photolithography is the most important part."
"Then a very obvious problem arises: how to ensure that so many photolithography steps can be etched in the same position on the silicon wafer."
"At the 14nm process, the linewidth is only a few nanometers."
"If there is even the slightest mistake, such as a slight deviation between the two masks, the entire chip will be scrapped."
"The lithography machine needs to overcome the challenges of hundreds of nanometer-level positions on the mask, and the time required is far less optimistic than the outside world imagines."
"Even if domestically produced lithography machines are manufactured, it will take time to improve the yield rate."
"Currently, lithography machines need to guarantee a yield of 125 wafers per hour. Otherwise, semiconductor manufacturers will lose money."
“We have so much to learn here,” Young said with a pained expression.
Although Young often uses his English name, he is indeed Chinese and graduated from Shanghai Jiao Tong University.
Cheng Gang sighed: "We can only take it one step at a time. For now, we only need the 28nm process, which should meet our needs."
"Since SMIC can produce 14nm chips, it doesn't matter if the price is a bit higher."
Young echoed this sentiment, saying, "There are so many things to catch up on in the semiconductor field; it's not something that can be achieved in a short time."
"It's not at all like the media portrays it; buying a lithography machine can solve the production problem. Even with a lithography machine, adjusting the mask and training skilled workers also takes time."
Cheng Gang was more optimistic than Young: "I think we are on the right track, relying on China's vast market to cultivate a complete semiconductor industry chain."
"There are certainly many difficulties, but we can only overcome them one by one."
Young: "You're more optimistic than me. We haven't even figured out the 28nm lithography machine yet, while Europe is already on the verge of 3nm. Not to mention they've already planned out their technology roadmap from 3nm to 1nm."
"The gap is quite large."
Young had something else on his mind that he didn't say: with such a huge gap, the only way to overtake them would be to open up a completely new track and use entirely new technologies.
Otherwise, given the current state of the silicon-based semiconductor industry, Young sees no hope of overtaking others.
Cheng Gang comforted me, saying, "I used to have a similar feeling when I was in the financial industry. There were gaps between us and foreign countries in many areas, and it seemed like we had no chance to catch up."
"Because risk control is a required course in the financial industry, we always see the risks first and think about how to control them well."
"After leaving the financial sector to enter the industry, I think more about what I want to do and how to achieve my goals."
"The financial sector and the industry sector have completely different perspectives on issues."
"Of course I can't judge which perspective is correct, but I can say that I think the domestic semiconductor industry still has a very good development prospect and future."
"As for the brain-computer interface chip of Kechuang Bio, even with the 14nm process, foreign manufacturers use more advanced chips, but we are confident that we can optimize the 14nm brain-computer interface chip to be better than the foreign 7nm chip."
This chapter is not finished, please click the next page to continue reading!
Continue read on readnovelmtl.com
