leonardwarren Chapter 176 Trial Engraving
Chapter 176 Trial Engraving
2020 12 Month 2 Day.
At 3:17 PM, a SF Express delivery truck stopped downstairs at the headquarters of Hongyuan Feiniao.
Su Chen personally went downstairs to sign for the package.
A small, temperature-controlled box. A label on the outside reads "Precision Instruments/Shockproof/Do Not Invert." The shipper is Xinsheng Semiconductor, a subsidiary of Shanghai Silicon Industry Group.
Su Chen took the packaging box back to the DRIE laboratory on the third floor, put on clean gloves, and carefully opened it.
Three silicon wafers.
300mm P-type <100> crystal orientation, 625 micrometers thick. The surface has a mirror-like silver-gray luster and reflects slightly under laboratory incandescent light.
Su Chen picked up the first piece and held it up to the light to examine it.
There are no defects visible to the naked eye. The surface smoothness meets the standards.
He placed the three silicon wafers into the cleanroom storage box in sequence, and then wrote on the specification paper:
Silicon wafers arrived. 3/3 are qualified. Trial etching time confirmed: December 3rd.
Then he picked up his phone and sent a message to Zhou Zhiyuan:
"The silicon wafers have arrived. Parameter calculations will begin at 6:00 AM tomorrow."
Zhou Zhiyuan replied instantly: "I'll be online at 5:50."
Su Chen glanced at the message but didn't reply. He put his phone on the table and turned to walk towards the DRIE device.
He needs to perform one last equipment check.
He personally inspected every component: the etching chamber, the gas pipelines, the temperature control system, and the vacuum pump. This DRIE equipment was purchased from Germany by Hongyuan Feiniao last year for eight million yuan, and it is currently the most expensive piece of equipment in the laboratory.
But what it will do tomorrow is something the designers of this device never imagined—to perform high-precision MEMS etching on a 300mm scale.
According to Bosch's own technical roadmap, 300mm MEMS etching will require at least two to three years of preliminary research.
Su Chen planned to complete the etching in one trial.
After checking the equipment, Su Chen sat back down at the lab bench.
There were twenty-four sheets of draft paper spread out on the table—thirteen of them were his own, seven were derived with Zhou Zhiyuan, and four were Zhou Zhiyuan's independently completed angular symmetry dimensionality reduction method.
From 250mm to 300mm. From one person to two people. From the equivalent thermoelastic approximation method to the third-order modified model.
All theoretical preparations have been completed.
All that's left is six hours of calculations tomorrow—and that trial etching.
Su Chen glanced out the window. In early December in Shenzhen, it gets dark very quickly. The streetlights were already on.
He stood up, turned off the lab lights, and locked the door.
Back in his dorm, he ate a bowl of instant noodles, took a shower, and set his alarm for 5:30 the next day.
Then go to bed.
He slept soundly. He didn't dream.
……
December 3, 2020. 5:30 AM.
The alarm clock rang.
Su Chen opened his eyes, got up, brushed his teeth, washed his face, and changed into his lab coat.
At 5:45, he entered the DRIE laboratory.
Start the preheating program for all equipment. The etching chamber of the DRIE equipment needs fifteen minutes to reach operating temperature. The vacuum pump needs ten minutes to reach the working vacuum level.
At 5:50, he opened the calculator program on his computer.
Zhou Zhiyuan's video call was connected on time.
"Good morning." On the other end of the screen, Zhou Zhiyuan was already sitting in his office at the Institute of Mechanics, Chinese Academy of Sciences. There was a cup of coffee and two loaves of bread on his desk, indicating that he planned to stay there all day.
"Good morning," Su Chen said.
"Have you entered all the parameters?"
"All done. Three temperature gradients, etching gas ratios, and power curves. Everything is according to our final plan."
"Okay." Zhou Zhiyuan took a deep breath. "Let's begin."
Su Chen pressed the Enter key.
On the screen, the calculation program begins to run.
Progress bar shows: 0.1%.
Regularized Green's function correction - angular symmetry dimensionality reduction version - starting the calculation of the third-order nonlinear thermodynamic response of a 300mm silicon wafer under DRIE etching conditions.
Estimated completion time: 5 hours and 47 minutes.
Su Chen leaned back in his chair, watching the progress bar slowly move on the screen.
"Let's go get some breakfast," Zhou Zhiyuan said. "It's been over five hours."
"I'm not hungry."
"Then I'll eat it." Zhou Zhiyuan picked up the bread and started eating.
Su Chen did not respond. His eyes were fixed on the screen, but his gaze seemed to pass through the screen, looking into a more distant place.
……
Time passed by, second by second.
The progress bar grows extremely slowly—it moves about 0.3% per minute.
At 7:00 AM sharp, progress reached 12.8%.
At 8:00 AM sharp, progress reached 23.1%.
At 9:00 AM sharp, progress is 33.5%. Everything is normal. All intermediate results are converging.
Su Chen finally got up and poured himself a glass of water.
At 10:00 AM, progress was 43.9%. Zhou Zhiyuan had already finished his second cup of coffee back in Beijing.
At 11:00 AM, progress was 54.2%.
At 11:17, the calculation program popped up a warning box.
Su Chen's body tensed instantly.
He moved closer to the screen—
"The number of iterations for the third-order coupling term in region 7 has exceeded the preset threshold. Current iterations: 847. Preset upper limit: 800."
"What's going on?" Zhou Zhiyuan's voice came from the other end of the screen, sounding tense.
Su Chen didn't speak. He quickly opened the calculation log and checked the data in the seventh region.
The 300mm silicon wafer was divided into twelve regions for calculation. The seventh region is located on the inner edge of the wafer—precisely the transition region where the third-order nonlinear effect is strongest.
"It's the singularity of the boundary conditions," Su Chen said. "The temperature gradient in the seventh region exhibits a local extremum in the transition section, causing the iterative convergence of the third-order terms to slow down."
Will it affect the final result?
"No," Su Chen said after looking at the data for three seconds. "The residuals are still converging, just at a slower pace. Just adjust the iteration limit to 1200 iterations."
He manually modified the parameters to keep the calculation going.
The progress bar continues to move forward.
Zhou Zhiyuan breathed a sigh of relief. But Su Chen's expression remained unchanged—from beginning to end, it was that calm, almost indifferent focus.
At 12:03, progress is 64.7%. The iteration of the seventh region converged on the 1031st time.
Looking at the number, Su Chen noted it down in his notebook: "The third-order terms converge slowly in the transition section of the seventh region boundary. The mesh density in this region needs to be optimized later. However, this does not affect the results of this calculation."
Lunch was brought up by the front desk at Lin Wei's request—two boxed meals, one for Su Chen and one for him to eat in the afternoon.
Su Chen finished his boxed lunch in five minutes.
At 12:45, progress was 75.1%.
At 1:30, progress is 85.4%.
At 2:00 PM, progress is 91.2%.
The calculation results for all regions are converging. The predictions from the third-order correction model are being gradually output.
At 2:20 PM, progress is 95.0%.
Su Chen stood up and walked to the DRIE device.
He took the first 300mm silicon wafer from the cleanroom storage box, inspected it, and then placed it on the etching chamber stage.
Close the wall door.
Back to the computer.
2:47.
The progress bar has jumped to 100%.
Calculation complete.
A results window pops up on the screen, listing hundreds of parameters—temperature, air pressure, power, time, gas flow rate—each parameter accurate to four decimal places.
These are all the process parameters for the 300mm test etching.
"It's done." Zhou Zhiyuan's voice was a little hoarse—he had been sitting in front of the screen for nearly nine hours.
"It's done," Su Chen said.
He began entering the parameters one by one into the control panel of the DRIE device.
Temperature gradient curve – Input.
Etching gas ratio - Input.
RF power curve – input.
Cycle time parameter - input.
He checked each parameter three times.
3:07. All parameters have been entered.
Su Chen stood in front of the control panel of the DRIE device.
His right hand hovered above the start button.
"Su Chen." Zhou Zhiyuan's voice came from the other end of the screen.
"Um."
"……come on."
Su Chen did not respond.
He pressed the start button.
"Buzz—"
The DRIE equipment started up. The sound of the vacuum pump was exceptionally clear in the quiet laboratory. The gas inside the etching chamber began to be injected according to the preset ratio. The RF power slowly increased according to the calculated curve.
The etching began.
On the surface of a 300mm silicon wafer, plasma sculpts microstructures at the nanoscale. Each etching cycle cuts precise trenches in the silicon lattice—depth, width, sidewall angle—each dimension predicted by Su Chen's third-order modified model and calculated by Zhou Zhiyuan's angular symmetry dimensionality reduction method.
Su Chen stood in front of the equipment, motionless.
His eyes were fixed on the data that was changing in real time on the control panel—cavity temperature, gas pressure, RF power, etching rate—every number was within the preset range.
No exception.
Fifteen minutes have passed.
Thirty minutes have passed.
An hour has passed.
3:52. The first etching cycle is complete.
Su Chen glanced at the intermediate data. The etching depth met expectations. The sidewall roughness was within acceptable limits.
"The first round went smoothly," he told Zhou Zhiyuan.
"Okay." Zhou Zhiyuan's voice trembled slightly.
4:37. The second etching cycle is complete.
5:01 AM. The third etching cycle is complete.
5:28.
The device emitted a soft "beep".
Etching complete.
All cycles have ended. The chamber begins to cool down. The vacuum pump continues to operate, removing any remaining gas.
Su Chen waited for five minutes until the cavity temperature dropped to a safe range before opening the cavity door.
He carefully removed the 300mm silicon wafer with tweezers.
Under the light, the microstructure on the surface of the silicon wafer is faintly visible—rows of neat grooves, like a city skyline in the microscopic world.
However, the precision is not visible to the naked eye.
Su Chen placed the silicon wafer onto the stage of the white light interferometer.
This is the gold standard device for measuring the precision of MEMS microstructures—it achieves nanometer-level measurement precision through the principle of optical wave interference.
He initiated the measurement procedure.
The probe of the white light interferometer began scanning the surface of the silicon wafer. The data was transmitted to the computer in real time, and a three-dimensional topographic map gradually appeared on the screen—the true shape of the microstructure on the 300mm silicon wafer.
The scan lasted approximately eight minutes.
5:41.
Measurement complete.
The final result popped up on the computer screen.
Su Chen stared at the screen.
Zhou Zhiyuan was also watching the shared screen from Beijing.
Sidewall angle: 12.07°
Accuracy deviation: ±0.018°
Sidewall roughness: Ra 2.3nm
The lab was silent for five seconds.
"...12.07 degrees Celsius." Zhou Zhiyuan's voice came from the other end of the screen, trembling noticeably. "Plus or minus 0.018 degrees Celsius. Su Chen—you did it."
Su Chen looked at the numbers on the screen.
12.07 degrees.
The target was 12.08 ± 0.03 degrees. The actual result was 12.07 ± 0.018 degrees.
The accuracy not only met the standards, but far exceeded expectations.
±0.018 degrees. This accuracy is even better than the ±0.025 degrees achieved during the 12.03 degree verification at 250 mm.
This means that the third-order correction model not only holds true at the 300mm scale, but its prediction accuracy also increases with the scale.
This is not engineering experience. This is the power of theory.
"The theory is correct," Su Chen said.
His tone was calm, as if he were stating a fact he had always known.
"The theory is correct!" Zhou Zhiyuan repeated on the screen, this time his voice much louder. "Su Chen, do you know what this means? The third-order correction model is actually more accurate on a large scale—this shows that our theoretical framework captures the correct physical essence—not an approximation, not a fit—it captures the true physical laws!"
Su Chen remained silent.
He took his phone out of his pocket and opened his chat window with Lin Wei.
Four words were typed:
"Success. 12.07."
After sending the message, he put his phone back in his pocket.
Then he went back to the lab bench, picked up a pen, and wrote on the protocol paper:
300mm trial etching v1.0 results:
Sidewall angle 12.07° | Accuracy ±0.018° | Roughness Ra 2.3nm
Silicon wafer usage: 1/3
Third-order modified model verification: passed.
Note: The third-order terms converge slowly in the transition section at the boundary of the seventh region (10^31 iterations). Further optimization of the mesh density can further improve computational efficiency.
After finishing writing, he glanced at the two remaining silicon wafers in the cleanroom storage box next to him.
It wasn't used.
It worked on the first try.
……
Thirty seconds later, Lin Wei's phone vibrated.
She was in her office discussing the agenda for the technical sharing session with Zhou Lin. Upon seeing Su Chen's message, she paused for a moment.
"Success. 12.07."
Four words. No exclamation marks. No emojis.
But Lin Wei's eyes reddened slightly at that moment.
She quickly composed herself and handed the phone to Zhou Lin.
Zhou Lin glanced at the screen, then looked up at Lin Wei. Her lips moved, but she didn't say anything.
"The technical sharing session on December 5th," Lin Wei said calmly, "The agenda needs to be changed."
"How should it be changed?"
"The original agenda was to introduce the engineering application methodology of the 250mm equivalent thermoelastic approximation method. Now—we're adding a segment."
"What part?"
Lin Wei looked out the window. The Shenzhen sky in December was so clear that there wasn't a single cloud.
"Experimental verification results of 300mm MEMS etching have been published."
Zhou Lin gasped: "Just announce it directly?"
"We will release everything directly. Data, parameters, accuracy—all will be released. However, this will only be done at technical sharing sessions and exclusively for alliance member companies. It will not be made public."
Why not allow it to be made public?
"Because now is not the right time to announce it publicly," Lin Wei said. "If we announce it now, Bosch will immediately adjust its strategy. I need to buy ourselves more time. Let Bosch continue at their pace—they think we're still struggling with 250mm. Let them continue to think that."
"But if the alliance member companies find out, won't the news leak out?"
"It will leak. But leaks take time," Lin Wei said. "From the sharing session to the news reaching Bosch's ears, it will take at least one to two weeks. One to two weeks—that's enough."
"Enough for what?"
"Enough for Hongyuan Precision's first batch of equipment to arrive and be installed and debugged. Enough for CR Microelectronics to launch its yield optimization plan."
She paused for a moment, then said:
"The 300mm news is a card I need to play. I want to play it at the best time. Now isn't the time to play it publicly—but the alliance needs to know. Because this news will make those 23 registered companies understand one thing—"
"What is it?"
"They bet right."