nyslfriends Chapter 182 Spread
Chapter 182 Spread
Saturday, December 9th.
Shenzhen, Su Chen's laboratory.
The 400mm parameter extrapolation calculation has been running for over sixteen hours. Su Chen started the calculation at 3 PM yesterday, went back to his dorm for a nap, and when he arrived at the lab at 7 AM this morning, the progress bar showed 87%.
He wasn't staring at the screen. He sat at the table next to him, continuing to draft the sixth part of his paper. It was handwritten, not on a computer. He used an A4 lined notebook, a blue ballpoint pen, and his handwriting, though small, was very neat.
When I got to the third page, the computer behind me beeped.
Calculation complete.
Su Chen put down his pen and walked to the computer.
A results summary window popped up on the screen. His gaze fell directly on the most crucial line of data:
400mm wafer, third-order correction model predicts sidewall angle: 12.07° | Prediction accuracy: ±0.015°
He stared at the number for about ten seconds.
±0.015°.
The measured accuracy for 300mm is ±0.018°. The measured accuracy for 250mm is ±0.025°.
The larger the size, the higher the precision.
From 250mm to 300mm, the accuracy improved from ±0.025° to ±0.018°, an improvement of 28%.
From 300mm to 400mm, the model prediction accuracy improved from ±0.018° to ±0.015°, a 17% improvement.
The rate of improvement is decreasing, but the direction remains the same—it is still getting better.
This trend cannot be explained by engineering optimization. Engineering optimization will encounter bottlenecks at a certain point, and the improvement in accuracy will tend to plateau or even stagnate. However, the prediction given by the third-order correction model is exactly the opposite—the larger the size, the more accurately the model describes the physical process, and the higher the prediction accuracy.
The reason is simple: the larger the size, the smaller the proportion of edge effects, the purer the dominant term of thermoelastic coupling, and the more accurate the fitting of the third-order correction term.
This is not a feature he designed.
This is a characteristic of physics itself.
Su Chen saved the screenshot of the results, then opened the electronic draft of Part 6 of the paper and added a paragraph in the "Outlook" section:
"Parameter extrapolation calculations based on a third-order nonlinear extended model show that, at a 400mm wafer scale, the model's prediction accuracy can reach ±0.015°, a further improvement over the measured accuracy (±0.018°) at 300mm. This result is consistent with the accuracy improvement trend from 250mm to 300mm, further validating the core characteristic of this model: the larger the processing scale, the more accurate the theoretical prediction."
He stopped and read the passage again.
Then the last sentence was deleted and replaced with: "The physical root of this trend lies in the fact that as the processing scale increases, the relative contribution of edge effects decreases, and the dominant term of thermoelastic coupling gains stronger control over the system behavior, thus making the description of the third-order correction term more accurate."
After making the changes, he picked up his phone and sent a message to Zhou Zhiyuan:
"The 400mm calculation is complete. Prediction accuracy is ±0.015°."
Zhou Zhiyuan's reply came quickly: "Even better than 300mm?"
Su Chen: "Yes. The larger the size, the more accurate it is, and it matches expectations."
Zhou Zhiyuan: "Should this result be included in the paper?"
Su Chen thought for a moment: "Thanks. But only write the model prediction, not the experimental verification. Label it 'theoretical extrapolation,' and don't make any hints about the feasibility of the experiment."
Zhou Zhiyuan: "I understand. They're giving us data, not a roadmap."
Su Chen: "Yes."
He closed QQ, packaged and encrypted the complete 400mm calculation data, and uploaded it to the internal server. Then he sent Lin Wei a message:
"Theoretical prediction for 400mm is complete. ±0.015°. Data uploaded."
Lin Wei's reply was only two words: "Received."
December 9th, 10:00 AM.
Suzhou, headquarters of MicroPort Sensors.
Five people were sitting in the conference room. Zhou Haidong, CEO of MicroPort Sensing, sat at the head of the table, with a printed report in front of him titled "Analysis of the Latest Developments of the Vilan Microsystems Technology Alliance".
This report was compiled by the marketing department yesterday, which he had them work overtime to prepare. It wasn't long, but every piece of information made his brow furrow even more deeply.
第一条：12月5日技术分享会后，薇澜技术联盟新增6家企业申请加入，联盟规模从23家扩大至29家（含待签约）。
Article 2: Among the newly added companies, 3 are packaging companies, which can provide a total MEMS packaging capacity of approximately 5.5 million units per month.
Article 3: Rumors have circulated on multiple industry forums that "Velocity has successfully completed a 300mm trial etching." Although there has been no official confirmation, the direction of the rumors is highly consistent with the reactions of the alliance companies.
Article 4: It appears that a tiered cooperation system has been implemented within the alliance, with core members receiving higher levels of technology sharing and priority in supply.
Zhou Haidong put down the report and looked at the four people in front of him: Vice President Zhao Mingcheng, Marketing Director Li Xuefang, Technical Director Chen Weiguo, and Financial Director Wang Jianmin.
"Share your thoughts."
Zhao Mingcheng spoke first: "President Zhou, if these three packaging companies have indeed joined the alliance, then our supply reduction in December—"
"It's like giving Vilan a huge gift," Zhou Haidong continued, his voice calm but his tone icy. "We reduced our supply by one million units, and they suddenly had an extra five and a half million units overnight. Net effect: Vilan's production capacity increased by four and a half million units, not decreased."
The meeting room was silent for a few seconds.
Technical Director Chen Weiguo whispered, "If we continue to reduce it in January—"
"If we continue reducing them," Zhou Haidong interrupted him, "Villan won't suffer any losses. Their packaging capacity is already sufficient. Reducing our production by 500,000 units per month is insignificant to them. The only ones who will suffer are ourselves—for every 500,000 units we reduce, we'll lose 500,000 units of revenue."
CFO Wang Jianmin opened a notebook: "President Zhou, I've done the calculations. If we continue to reduce supplies as planned until April next year, that is, from four million to one million, our monthly revenue loss will be approximately—"
"No need to calculate," Zhou Haidong waved his hand. "What I want to hear isn't how much revenue we'll lose. What I want to hear is, is there any way to salvage the situation?"
The meeting room fell silent again.
Zhao Mingcheng coughed and said, "President Zhou, to be frank, if the news about 300mm is true, then the problem we're facing now isn't about reducing supply or not."
"What's the problem?"
"It's a matter of our position in Vilan's supply chain," Zhao Mingcheng said, his voice slightly hoarse. "Previously, Vilan needed us because we were their largest packaging supplier, accounting for nearly 40% of their total packaging capacity. But now, with three new packaging companies joining the alliance, even if we don't reduce our supply, our share has dropped from 40% to—"
"Less than 30%," Wang Jianmin continued, "and if Vilan's self-built packaging line goes into production as planned next May, our share will further drop to below 20%."
Zhou Haidong fell silent.
He stood up and walked to the window. The December sky in Suzhou was hazy, and the industrial park in the distance was faintly visible in the mist.
The decision to reduce supply was made three months ago. The logic at the time was clear: Vilan's investment in 300mm was extremely risky. If the etching trial failed, the alliance would waver, and Vilan's supply chain would be disrupted. The plan at the time was to increase pressure during that window, forcing Vilan to make concessions on price, and even getting some alliance companies to switch to MicroPort Sensing.
But now the 300mm version has been successful.
It wasn't a narrow success, but a success that far exceeded expectations—at least judging from the reactions of the alliance companies.
The reduced supply from MicroPort Sensing did not cause any substantial harm to Vilan; on the contrary, it became a catalyst for the three packaging companies to join the alliance.
Why did those three companies join at this time? It wasn't just because of the 300mm data; it was also because they saw a gap in Vilan's supply chain—a gap caused by reduced supply from MicroPort Sensors. They joined the alliance to fill this gap and simultaneously gain technical cooperation from Vilan.
In other words, MicroPort Sensing's supply reduction strategy has objectively helped Vilan achieve supply chain diversification.
"They've really shot themselves in the foot," Zhou Haidong muttered.
He turned back to the table and looked at the four people in front of him:
"Starting in January, normal supply resumed. Four million units disappeared."
Zhao Mingcheng was stunned for a moment: "President Zhou, this is equivalent to—"
"It's tantamount to admitting we've lost this round," Zhou Haidong interrupted him, "but continuing to reduce supply is like blocking the rest of our options. Vilan doesn't lack production capacity right now, so whether we reduce supply or not won't affect them. But if we continue to reduce supply, once their self-built packaging line goes into production next year, we won't even be able to hold onto our last 30% market share."
He paused for a moment.
"Resume supply and maintain the relationship. At least preserve our current market share. As for the future—"
He didn't continue.
Everyone in the meeting room knew what "the future" meant. In the future, when Vilan's self-built packaging line goes into production, and the alliance has more packaging options, MicroPort Sensing's position will only become lower and lower.
But at least now that supply has been restored, we can still retain 30% of the supply.
If the reduction continues, even that 30% won't be achieved.
Meeting adjourned.
December 10th, Sunday.
Zhou Zhiyuan's study at Tsinghua University, Beijing.
On the computer screen in front of Zhou Zhiyuan, the rewritten version of the second part of the thesis had reached page seventeen.
The original version, written in the JMEMS style, was only nine pages long. The rewritten version is nearly twice as long because he needed to explain the theoretical framework of the third-order nonlinear extension more clearly—not just to readers in the MEMS field, but to readers in the entire materials science field.
Reviewers for Nature Materials are not necessarily MEMS experts. They may work on thin film deposition, nanoimprinting, or any micro/nano fabrication process involving thermoelastic coupling effects.
Therefore, the theoretical part must be a self-contained system and cannot assume that the reader has background knowledge of MEMS-DRIE.
Zhou Zhiyuan wrote a sentence at the beginning of the third section:
"The third-order nonlinear extension proposed in this section has a mathematical form that does not depend on specific etching processes or material systems, but is based on a general physical framework of thermoelastic coupling effects."
This sentence was the result of his repeated deliberation. It defines the theoretical boundaries of the paper—this is not a process improvement specifically for DRIE, but a theoretical tool applicable to all thermoelastic coupling scenarios.
When he was writing at 3 PM, Su Chen sent him a message:
"The theoretical prediction for 400mm is in. ±0.015°. I've already included the extrapolated data in the draft of Part Six. Please check if the wording is appropriate."
The attachment is a passage from Part VI of the paper.
Zhou Zhiyuan clicked on it and read it once.
He then added a comment: "'The statement that the relative contribution of edge effects decreases' could be more precise. I suggest changing it to 'As the wafer area increases, the ratio of the edge region area to the total area decreases by O(1/r), resulting in a reduction in the perturbation of the overall thermal field distribution by edge effects.' This way, reviewers can see the mathematical basis."
Su Chen replied five minutes later: "Revised. Your wording is better."
Zhou Zhiyuan continued writing the second part. Around 8 PM, he stopped and sent his work for the day to Su Chen:
"Part Two is 70% complete. It should be finished by tomorrow. You need to review the mathematical derivations in Part Three; I'm unsure about a few symbols related to the upper limit of integration."
Su Chen: "Okay. I'll give you the third part by Tuesday."
December 10th, 9 PM.
Hongo Ward, Tokyo, Japan.
The light in Akira Ishikawa's study was soft.
On the table in front of him lay six printed papers, all published by Su Chen. He had read them countless times, but today he wasn't reading the papers themselves; he was doing something different.
He spread a blank A3 sheet of paper on the table, picked up a pencil, and began to draw.
The left half of the figure shows Su Chen's MEMS-DRIE technology roadmap: equivalent thermoelastic approximation method → third-order nonlinear extension → 250mm verification → 300mm (rumored).
The right half of the figure represents his research areas: MEMS vibration gyroscopes → Coriolis force detection → drive/detection mode coupling → temperature drift compensation.
He drew a dotted line between the two parts, and then wrote four words above the dotted line:
"Thermoelastic coupling."
The temperature drift problem of gyroscopes is essentially a thermoelastic coupling problem—temperature changes cause changes in the elastic modulus of the microstructure, which in turn affects the resonant frequency and detection sensitivity.
If Su Chen's third-order correction model is indeed a "universal theoretical tool"—as his paper suggests—then this model should also be effective in compensating for temperature drift in vibrating gyroscopes.
Akira Ishikawa wrote a line at the bottom of the A3 paper:
"Verification: Applying the third-order correction model to gyroscope temperature drift prediction. Required data: Complete derivation of the third-order correction term not published in Su Chen's paper."
He put down his pencil and looked at the picture.
He wasn't competing. He was doing the most instinctive thing in academia—seeing a theory that might be correct and then wanting to verify it in his own field.
The problem is that Su Chen's complete derivation of the third-order modified model has not been made public. The published paper only provides the result, not the process.
"After the paper is published."
The sentence Su Chen said in his reply to the email.
Akira Ishikawa now understood the third meaning of that statement—the new paper might reveal more theoretical details. If the new paper is submitted to a journal of the caliber of Nature Materials, reviewers will request the complete mathematical derivation.
Then he will be able to get what he needs.
Akira Ishikawa folded the A3 paper and put it in the first drawer of his desk.
Then he turned on his computer and began to organize the gyroscope temperature drift data from his lab over the past three years.
If Su Chen's model is correct, these data will be the first batch of independent validation samples.