ebonymovies Chapter 202 Third-Order Model
Chapter 202 Third-Order Model
In early April, a quiet transformation was taking place simultaneously in the global microelectromechanical systems (MEMS) academic and industrial communities.
It didn't happen at a press conference, with no flashbulbs from reporters, no cheers in a stadium. It happened on computer screens in labs, in peer review systems of academic journals, in conference rooms at multinational corporation headquarters, and in late-night posts on social media.
But its impact is wider than any press conference.
……
April 3rd.
Microsystems Laboratory, ETH Zurich.
Associate Professor Marcus Zimmerman sat at his desk, a set of matrix lab calculation results displayed on the screen in front of him.
He spent the past two weeks recalculating a set of thermoelastic coupling experimental data from a 200mm wafer that his lab conducted last year, using the third-order model theoretical framework that Wei Lan had published on the preprint platform.
As a result, he stared at the screen for a full five minutes.
The traditional second-order model has a prediction bias of 3.7 percent for this data set. The third-order model has a prediction bias of 0.62 percent.
Six times the accuracy.
And this is only data at a 200-millimeter scale—at this scale, the contribution of the third-order term should be very small. At a 400-millimeter scale, the difference would be even more striking.
Zimmerman opened his email and sent a message to Vogt.
Email Subject: Reproduction Results – Validation of a Third-Order Model Based on Our 200mm Experimental Data
The email body is very short:
Professor Vogt:
I have completed all the reproduction calculations as suggested by you, and the relevant results are attached. A brief conclusion is as follows: Under our 200mm experimental dataset, the decision coefficient of the second-order model is 0.943, and the decision coefficient of the third-order model is improved to 0.991. Even at a process scale where the influence of the third-order term is generally considered negligible, the actual performance of this model still far exceeds expectations.
I would be very willing to participate in the follow-up seminar you mentioned earlier and report on the results of this reproduction.
Sincerely,
Marcus Zimmerman
Unbeknownst to him, at least three other research teams sent similar emails to Vogt on the same day.
A group of microelectromechanical systems researchers at Purdue University replicated the results using their pressure sensor data—the deviation was reduced by 58 percent.
An assistant professor at National Chiao Tung University in Hsinchu, Taiwan, verified this using her data on radio frequency microelectromechanical systems (MEMS) switches—the third-order correction reduced the temperature drift prediction error from over 4% to less than 1%.
A PhD student at MIT tested the data from his work on nanoscale microelectromechanical system resonators—the improvement in the third-order model at the nanoscale was not as significant as at the 400-millimeter scale, but it still resulted in a 15% to 20% improvement in accuracy.
Vogt received four emails with the results of his attempts to reproduce the results in one day. He sat in his study and printed out the four sets of results, laying them out on the table.
Four laboratories in four different countries. Four completely different types of devices. Four sets of completely independent experimental data.
All of these lead to the same conclusion: the third-order model is effective.
Vogt picked up the phone and dialed the secretary of the organizing committee for the IEEE Microelectromechanical Systems Conference.
"Regarding the follow-up workshop for the next meeting—I need to expand it. The original plan was a two-hour online workshop, but I now suggest changing it to a half-day thematic session."
What is the theme of this special session?
"Third-order nonlinear models: academic achievements and industrial applications."
After hanging up the phone, he sent an email to Su Chen.
Message body:
Dr. Su Chen:
Forgive my frankness, but the speed at which your theoretical model has spread is the fastest I've seen in my forty-year academic career. Four independent research teams have already reproduced the results, and more validation data will be available in the coming weeks. I have upgraded the originally planned follow-up seminar to a half-day conference session. I sincerely invite you and Professor Zhou Zhiyuan to be the keynote speakers for this session.
Heinrich Vogt
……
the same day.
Cyberspace.
The news section of the journal *Nature* published a short news report on its website. The title is:
A Chinese team developed a theoretical model for microelectromechanical systems, which has been validated by multiple independent experiments worldwide.
The article is short, only about 600 words. However, it mentions the triple verification from the IEEE Microelectromechanical Systems (MEMS) conference, Akira Ishikawa's independent data, and ongoing reproduction work by several research groups. The article concludes with a quote from Vogt:
"It is extremely rare in the field of microelectromechanical systems (MEMS) for a completely new theoretical model to achieve such a large-scale independent cross-validation within just a few months of its formal publication. The entire industry and academia are currently closely watching the subsequent developments of this model."
This report has been reprinted countless times.
Domestically...
Huxiu.com published a lengthy article titled "Why Did a Chinese Team's MEMS Theory Make Bosch and Infineon Uneasy?" The article meticulously recounts the entire process of developing and validating the third-order model, citing publicly available information from the IEEE Microelectromechanical Systems (MEMS) conference and evaluations from several anonymous industry professionals. The article garnered over 150,000 views within two days.
36Kr followed up with a more business-oriented analysis: "Villan Microelectronics: A Chinese Startup 'Unexpectedly' Validated by Academia." The article focuses on the industrial application prospects and potential commercial value of the third-order model.
Titanium Media took a more pointed angle: "Bosch's €1.8 million vs. Vilan's 500 million – Asymmetric Competition in the MEMS Industry." This headline was forwarded over two thousand times on Weibo, and the comment section was filled with various emotional remarks.
Semiconductor Industry Watch published a more sober analysis. The author was not He Wentao—who hasn't published a single word in the last three weeks—but an anonymous editor who identified himself as an "industry observer." The article was titled "The Industrial Implications of the Third-Order Model: Why This Is More Than Just a Paper."
The core argument of the article is that the significance of the third-order model lies not only in improving sensor accuracy, but also in providing, for the first time, a complete predictive tool from theory to mass production for thermoelastic coupling in microelectromechanical systems (MEMS). This means that future MEMS production line design can shift from "trial and error based on experience" to "theoretical guidance," significantly reducing R&D costs and timelines.
The last sentence of the article has been widely quoted:
"If traditional MEMS process design is like driving in fog, then the third-order model is like having a navigation system accurate to the millimeter. Whoever uses this navigation system first will be able to run faster and more steadily on the 400mm wafer manufacturing track."
……
On Zhihu, the discussion about Wei Lan and the third-order model continues to grow.
A new question has garnered significant attention: "What impact does the third-order model have on the lives of ordinary people?"
The most upvoted answer came from a user who claimed to work at an autonomous driving company, and wrote a 2,000-word explanation. The key points are:
"The accuracy of the gyroscope in your phone, the accelerometer in your car, and the motion sensor on your fitness tracker—all of these are affected by the thermoelastic coupling effect. A third-order model can improve the accuracy of these sensors by 30% to 60%. For you, this means more accurate navigation on your phone, safer self-driving cars, and more realistic motion data."
However, this is only a short-term impact. In the long run, when sensor accuracy reaches a new order of magnitude, many previously impossible applications will become possible—such as millimeter-level indoor positioning, nanometer-level micro-manipulation robots, and truly all-weather autonomous driving. These things are currently still in the laboratory, largely because sensor accuracy is insufficient.
The third-order model doesn't directly turn these into reality, but it opens a door.
This answer received 3,600 likes.
In the comments section, "Semiconductor Veteran 2003" finally broke his month-long silence, leaving three words:
"That's right."
This is the first time he has expressed his opinion in discussions related to Vilan.
……
April 5th.
Beijing, East China University of Science and Technology.
Su Chen sat in the laboratory, with the final proofread version of the revised manuscript in front of him.
All co-authors' revisions have been compiled. Ishikawa Akira reviewed his data section word for word and offered two minor adjustments, which Su Chen adopted. Zhou Zhiyuan conducted a final round of proofreading of the theoretical extensions. Dean Chen Qiming signed the confirmation email. Lin Wei checked the citation formats and license documents for all commercial data.
Everything is ready.
Su Chen glanced at the submission system. The revision submission button was lit up in blue, waiting for him to click it.
He didn't click immediately.
He opened the revision response letter—this letter to the journal editors and reviewers is one of the most important documents in the entire revision process because it determines the reviewers' first impression of the author's attitude towards the revision.
The letter begins with the standard format used by academic journals:
Dear Editors and Reviewers:
I sincerely thank all the reviewers for their detailed and constructive comments. These revisions have greatly improved the content and rigor of this paper.
This was followed by a point-by-point response. Each reviewer comment was followed by detailed explanations of the revisions: the corresponding changes, supplementary experimental data, and details of any new figures.
In response to the third reviewer's comment regarding "supplementing validation data from multiple independent platforms," Su Chen wrote the following:
Following the suggestions of the third reviewer, this paper supplements two sets of validation data from entirely new independent platforms: the first set comes from our supply chain partner, Zhenxin Microelectronics, using a standardized, mass-produced commercial inertial navigation sensor without process optimization for field testing; the second set is a complete, independent dataset provided by Professor Akira Ishikawa's research group at Tokyo Institute of Technology. This team used a silicon deep reactive ion etching (SRI) process on insulating substrates to fabricate a microelectromechanical system (MEMS) gyroscope for model validation, and their device fabrication process differs fundamentally from the process used in this paper's experiments. The average prediction error of both supplementary data sets is less than 1.5%, highly consistent with the original experimental conclusions, fully demonstrating the universality of this third-order model across processes and device platforms.
Su Chen read this passage three times.
He then added an extra line at the end of the article—this was not his usual writing style, but he felt it necessary to truthfully inform the reviewers of the latest industry developments:
It should be noted that since the initial draft of this paper was submitted, at least four independent research teams worldwide have completed the replication and verification of this third-order model using their own proprietary experimental datasets. A detailed summary of the replication data can be found in Supplementary Note 3. This round of cross-validation work, spontaneously undertaken by the academic community, further corroborates the stability and universality of this theoretical model.
After finishing writing, he hesitated for a moment. Would this passage seem too deliberate a promotion? Academic paper revision responses generally don't include additional information about external industry developments that occurred after submission. But he felt the reviewers should be aware of this: this comprehensive replication is actually happening within the industry, and this theory is no longer an isolated academic paper, but a general research tool being accepted and implemented throughout the entire microelectromechanical systems (MEMS) field.
He ultimately retained this supplementary note.
He then clicked the submit button.
A system notification pops up on the screen:
Your revised manuscript has been successfully submitted. Thank you for your cooperation.
Su Chen looked at the message with a calm expression, without the slightest ripple.
He took out his phone and sent a message in the dedicated work group. The group chat number was NM-00847, and there were only five members in the group: Su Chen, Zhou Zhiyuan, Chen Qiming, Lin Wei, and Zhang Li.
"The revised manuscript has been submitted and is awaiting the results of the second round of peer review."
Zhou Zhiyuan replied instantly: Received.
Lin Wei sent a thumbs-up emoji.
Dean Chen Qiming replied: Thank you for your hard work.
Zhang Li, a collaborating researcher at China Resources Microelectronics, replied: We await good news.
Su Chen put his phone back on the table.
He sat quietly at his workstation, the only sound the low, continuous hum of the lab's air conditioner. Outside the window, the warm, bright afternoon sun of April shone through the window. Scattered shouts drifted from the campus basketball court, mingling with the dull thud of basketballs hitting the ground—a lively, bustling atmosphere.
All the results have been submitted.
Now, all that's left is to wait.
But this wait was completely different from three months ago. Three months ago, when he first submitted the draft of the paper, he only had the data developed in his own lab. Even if he was convinced that the theory was correct, he couldn't predict the evaluation from the academic community or the outside world. Now, the revised manuscript includes experimental verification data from three independent platforms, official reproduction results from four overseas research groups, and an invitation from Vogt to give a keynote speech at a special seminar.
In just three months, a completely new theory proposed by a single individual has become a widely recognized consensus across the entire industry.
Su Chen got up and walked to the window.
In April, Beijing's sky was clear and cloudless. In the distance, the newly planted ginkgo trees on campus were just sprouting tender new buds.
He recalled the message Lin Wei had sent him yesterday: the paper could be submitted to the review process. He could start planning his second paper now; Ishikawa Akira's experimental data, combined with Zhenxin Microelectronics' mass production test data, was more than enough to support a complete multi-platform model verification paper.
He only replied with one word: "Hmm".
But he was considering much more than just the second academic paper.
He thought of the Tian Shu chip. The deadline for the third optimization scheme was the end of June, and the four-month buffer period given by He Zhiqiang was already halfway through. If this paper were officially published in June, what positive impact would it have on the official evaluation of the third scheme?
He envisioned the company building its own packaging production line. With production set to officially commence in May, once the capacity bottleneck is completely alleviated, Vilan Microelectronics' monthly shipments will double. Increased shipments mean more real-world commercial testing data, which will also make the industrial application verification of the third-order model more complete and robust.
He thought of the Bosch Group. Chief expert Stein remained silent throughout the IEEE Microelectromechanical Systems (MEMS) session, a silence that spoke louder than any public statement. Bosch would eventually approach him seeking cooperation; the only variable was the form of cooperation and the terms of the negotiations.
He recalled Vogt's remark: "Your theoretical model has spread the fastest I've seen in my forty-year academic career."
Forty years in total.
Su Chen smiled very slightly, but the smile vanished in an instant, lasting less than half a second.
He then turned back to his workstation and opened the outline document for his second paper.
The title of the paper has already been decided: "Multi-platform verification of a third-order thermoelastic coupling nonlinear model of microelectromechanical systems: from laboratory test to industrial mass production adaptation".
The cursor blinked silently at the top of the document.
Su Chen's fingertips touched the paper as he began writing the main text.
The warm April sun streamed through the glass window, casting a soft golden glow on the keyboard. A gentle breeze blew outside, causing the tender new buds of the ginkgo tree to sway gently.
The spring for the domestic microelectromechanical systems (MEMS) field has truly arrived.