Chapter 452: Queqiao
Real-time dynamic compensation?
Liang Mengsong naturally knew this proposal.
Similar proposals have applications in the semiconductor field.
“We have installed a large number of sensors, but they only record temperature and vibration.
I want you to rewrite the algorithm to make them a nervous system for prediction and correction.
Directly link the key parameters of the lithography machine with environmental sensor data.
Based on yield rate data from thousands of past hours, establish a predictive model that forecasts the amount of focal drift before environmental changes affect the photoresist.
Finally, using the high-speed micro-actuator built into CG-1, dynamically fine-tune the position of the lens group at the microsecond level to offset optical distortion caused by environmental changes.
This is like a surgeon on a shaking operating table relying on neural reflexes to keep hands steady.
This way, the lithography machine no longer pursues static perfection, but perfection in dynamics.
The more unstable the environment, the more valuable our compensation system becomes.”
Lin Ran paused for a moment before saying:
“Once this dynamic system matures, it can not only activate the potential of 4nm and solve the problems you are facing now, but also significantly boost the yield rate of our 7nm production line to match TSMC’s 7nm process yield rate.”
After hearing this, Liang Mengsong immediately understood Lin Ran’s intention: not to conquer the laws of nature, but to adapt to them and harness them with data and algorithms.
“Dynamic compensation…” Liang Mengsong murmured to himself, his eyes gleaming, “Use software and algorithms to compensate for hardware shortcomings. Use our strongest information technology to solve their proud precision optics problems.”
The excited expression on his face only lasted briefly.
Soon it turned into the rigor and doubt of a craftsman.
Having immersed in the semiconductor industry for decades, Liang Mengsong certainly knew the concept of “real-time dynamic compensation,” but he was even more aware that applying this technology to lithography machines like CG-1 would increase the difficulty geometrically.
“General Manager Lin, dynamic compensation is theoretically feasible, but we all know—and you personally participated in building our production line—you know the gap between theoretical feasibility and actual engineering implementation.
Many times, the Mariana Trench lies between them.” Liang Mengsong’s tone became urgent; he had to make Lin Ran understand the huge engineering chasm.
“In actual industry, dynamic compensation has applications in the semiconductor field, but the time constants and motion amplitudes in those scenarios are completely different from the core exposure scenario of the lithography machine.”
“One of the most common applications is CMP, which we commonly call chemical mechanical planarization.
In CMP equipment, we use real-time monitoring to adjust the pressure distribution of grinding force to compensate for wafer thickness unevenness during grinding.
This process is low-speed, with adjustment frequency possibly at the second or even minute level.
The equipment has enough time to collect data, calculate corrections, and execute.
Another application scenario I can think of immediately is plasma etching.
We use a spectrometer in the etching chamber to monitor the plasma state in real time, then fine-tune gas flow or RF power to compensate for etch rate drift.
This feedback frequency may reach millisecond level, but it corrects the chemical environment, not directly correcting optical focal length.”
Liang Mengsong pointed with his finger at the seemingly existing CG-1 lithography machine in the office air: “But what are we facing now? It’s CG-1’s metasurface lens.
This system is extremely sensitive to temperature and vibration; even a tiny airflow change will cause focal drift of several to tens of nanometers within microseconds.
Our dynamic compensation system must complete the entire process from data acquisition to predictive model calculation to micro-actuator physical movement within microseconds.
This requires the actuator itself to have ultra-low latency and ultra-high precision response capability.
Actuators on traditional semiconductor equipment simply cannot meet this speed and precision requirement.
This requires not only algorithmic innovation, but also innovation at the equipment level.”
Liang Mengsong spoke quickly, jumping topics just as fast; communication between smart people should be more relaxed.
Liang Mengsong’s final words carried a strong engineering warning: “The physical properties of CG-1’s metasurface lens mean its compensation path is more complex and non-linear than traditional lithography machines.
This system is humanity’s limit attempt on DUV; its instability is inherent, not simple external interference.
We are essentially building a super complex system that can predict its own ‘imbalance’ and achieve ‘balance’ before the imbalance occurs!”
He looked at Lin Ran and summed up: “General Manager Lin, in terms of difficulty, this is even harder than our past from-scratch EDA software.
This chasm is far beyond the level of dynamic corrections like CMP or etching.”
Lin Ran nodded: “Of course, I know of course, you know, I know too.
To independently build a complete semiconductor production and manufacturing system apart from the entire Western world is itself an impossible task.
Of course it’s hard, but the problem is, this is currently our most likely path to quickly reach 5nm.
From the 1950s to now, we’ve come this way all along.”
Lin Ran’s words were filled with immense confidence; so what if it’s hard? Just do it.
He finally summed up: “Algorithm problems are mine to solve; no matter how difficult the model, I will find the right solution for you. Engineering problems are yours; what you need to do is integrate domestic existing resources and resources available from abroad, and implement it through engineering.”
Liang Mengsong stood up, extended his hand, and during the handshake, he said only one word: “Good.”
This so-called TSMC traitor would deliver a truly fatal blow to TSMC.
On the first day of 2026, Apollo Technology’s live broadcast room was different from usual; today Lin Ran was personally hosting the live broadcast.
Lin Ran sat in his office, with a plain white wall behind him, no books on it.
The desk was slightly cluttered, with various documents piled together.
Lin Ran sat in front of the camera, wearing a gray cashmere sweater that looked high-quality.
“Today is the first day of the new year, New Year’s Day. As usual, I’m only responsible for chatting and answering questions; the actual launch is handled by our astronauts and ground control center.
I now fly to Wenchang less and less.
Everyone should know about our lunar electromagnetic rail from our official announcement or news media reports.
Or rather, the name everyone prefers, the Lunar Steel Dragon; construction progress has advanced a big step again.
It can now provide higher initial speed to spacecraft, including replacing the intermediate connection nodes with new electromagnetic materials for better conductivity.
So today, we’re doing something big: during this year’s Earth-Moon window, launching a spaceship.
The last launch was just a spacecraft, an unmanned one, very light, about the same as a lunar rover, just tens of kilograms.
We only did a technical feasibility verification to ensure this path works, provides sufficient thrust, and lunar surface heat dissipation is adequate.
This time is different; this is a newly designed spaceship with a very aggressive design, which you can see from the design drawing our official Weibo posted.
Its shape is somewhat like a cone, with a wide and solid base like an inverted shallow pan, perfectly fitting the electromagnetic rail structure.
Its exterior is pure functionalism.
Because everyone knows there’s no air on the Moon, meaning no air resistance, so no need to consider aerodynamics.
I know everyone likes streamlined shapes, thinking they’re cool, but in space navigation, such designs only have aesthetic significance, human aesthetics that aliens may not appreciate.
It has a wide chassis and a rounded front end; I’ve seen comments joking that laid flat, this shape looks like a UFO.
Actually, this is for optimal magnetic coupling and acceleration stress dispersion on the rail, while the rounded front end provides the best blunt body aerodynamic shape for future Earth return.”
Lin Ran’s live broadcast room quickly surged with tens of millions of audience members from different platforms; bullet screen crowded, but Lin Ran ignored netizens’ questions, just talking to himself.
Meanwhile, many anchors were relaying, adding cooler animations like 3D animations of Lunar South Pole Shackleton Crater with AI-processed electromagnetic rails that felt more sci-fi.
There were many similar live broadcast rooms, including some with retro sci-fi effects doing relays.
But all these secondary streams combined didn’t have as many viewers as Lin Ran’s room.
Even if netizens didn’t understand magnetic coupling, stress dispersion, or aerodynamic shapes, professional anchors would at least explain a bit, but Lin Ran just kept saying what he wanted.
No way around it; he was the most dazzling tech star in China this era, far surpassing other tech entrepreneurs in the same track in every aspect.
Lin Ran continued: “The difference in this launch is that this is the first manned spaceship in human history to completely rely on electromagnetic rail for initial speed.
Of course, this time it’s carrying a full-scale model of equivalent weight, not real people.
If this launch goes smoothly, the next one will be real people.
We named the spaceship Queqiao, which conflicts with China Aerospace’s Queqiao satellite.
But no problem; I talked to them, and they said their Queqiao refers to the satellite, so not a duplicate name.
Because it does quick round trips; whenever needed, as long as within the time window, it can launch.
Mainly electricity; its heat shield is also 3D printed from moon soil materials.
The cost of one return is almost only electricity and depreciation, low to scary; our internal estimate is that for one Queqiao completing 100 round trips, the single round-trip cost is only 2 million RMB.
Outbound by reusable rocket launch, return by electromagnetic rail.
Full automatic navigation.
That means the cost price for a Moon tourism flight ticket is 2 million; make it a 5 million tourism product—anyone interested?
But I estimate Lei Zong will curse me to death inside; his trip is 1 billion, yours is 5 million—this price gap is too big; normally hard to convince yourself with ‘buy early enjoy early.’
It can accommodate two astronauts; tonight, we’ll use it to complete a quick return test from Moon to Earth.”
After the simple explanation, Lin Ran glanced at the bullet screen and picked the most common question.
Lin Ran said: “Okay, first question: ‘How does the electromagnetic rail accelerate the manned spaceship to lunar escape velocity? Won’t people be crushed?’ Great question! This is the core technical problem.
As we all know, the Moon’s escape velocity is 2.4 km/s.
Reaching this speed on a 20 km track means high acceleration load.
To protect astronauts, we use a multi-stage linear acceleration system, which is what I mentioned earlier: replacing new electromagnetic materials in different stages.
Combined with liquid buffer seats.
Simply put, acceleration is distributed across the entire track, with astronauts’ peak acceleration strictly controlled under 4G.
What is 4G? It’s the most thrilling moment on a roller coaster.
Ordinary people can fully withstand it; of course, astronauts with strict training can even more so.”
Lin Ran looked at the second question: “Many friends ask, without traditional fuel, how to do orbit change?
Not without traditional fuel, but very little, plus equipped with new Hall thruster.
We use traditional fossil fuel engines for orbit change, new Hall thruster for fine angle corrections.
Relying on these two, combined with Queqiao’s navigation system, to accurately complete two orbit changes: one from low lunar orbit to Earth transfer orbit, the other from Earth transfer orbit to Earth’s high elliptical orbit, preparing for atmospheric reentry.
Bullet screen started showing questions about spaceship return to Earth heat protection.
Lin Ran answered: “Yes, everyone knows that spaceships returning to Earth rub intensely with the atmosphere, needing a heat shield.
The coolest part of this spaceship is that its heat shield is Moon-made.
We use silica and mica powder abundant in moon soil to synthesize composite heat shield materials in the Moon 3D printing factory.
Astronauts on the Moon, through automated spray system, re-arm the spaceship with heat shield armor.
This heat shield is consumable design. During high-speed Earth return, surface sublimating material volatilizes, leaving microporous carbonized layer for effective insulation.
That’s why Queqiao’s surface looks black after returning to Earth. Because its heat shield is use-and-discard, locally sourced.
This system has been fully refined and optimized over the past year.
It can adapt to different shapes of spaceships.”
“The essence of the electromagnetic rail is a space gun, which could be seen as a potential kinetic weapon launch platform, causing geopolitical turmoil—is this irresponsible behavior?
This Chinese friend with IP in Japan, isn’t your question a bit too professional?”
