AI's end is the lithography machine, and the end of the lithography machine is the lens

--- Why are lithography machine lenses difficult?
EUV and high-end DUV optics are the ultimate combination of ultra-precision industrial systems. They rely on materials, coating, metrology, assembly, thermal control, vibration control, algorithms, error modeling, and long-term experience accumulation. The real limit to expansion is often not a single component, but the entire "precision closed loop."
The core of this loop is: you cannot manufacture something more precise than your measurement capability.
The 13.5nm wavelength of EUV is almost absorbed by all materials, so EUV cannot use traditional lenses, only multilayer mirrors. Zeiss's EUV mirror surface is essentially an atomic-level reflection system. Surface error requirements are usually in the tens of picometers.
1 pm=10−12 meters
Atomic diameter is about 100pm, meaning many permissible errors on EUV mirror surfaces are already close to half an atom layer.
What makes it even more difficult than making such a mirror is how to stabilize the measurement of these errors. How to perform measurements under thermal drift, air disturbances, and vibrations. How to maintain consistency on large-sized mirrors. How to achieve long-term stable industrial repeatability. Because at this point, what is being measured is no longer just length, but the phase of the light wave itself.
EUV measurement systems are themselves ultra-high-end industrial chains. They include laser interferometers, ultra-stable laser sources, reference optics, ultra-low thermal expansion materials, active vibration isolation systems, ultra-precision displacement stages, wavefront sensors, vacuum systems, and long-term drift compensation algorithms. Many core suppliers, globally, may only number 1-3.
And these metrology systems also require higher-grade measurement systems to manufacture. This creates a recursive (vicious cycle): advanced measurement equipment manufacturing requires even more advanced measurement equipment.
Taking one of the bottlenecks, reference optics, as an example.
Reference optics are not ordinary mirrors. They are essentially the "primordial" in the optical world, the apex of the entire industrial precision hierarchy. Because to measure EUV mirrors, you must first have a reference mirror more precise than the EUV mirror itself. This creates a terrifying question: who will manufacture the most precise mirrors in the world?
Manufacturing reference optics is essentially an infinite approach to approaching a perfect surface. It first relies on ultra-low thermal expansion materials, such as SCHOTT's Zerodur or ULE-type materials. These materials require not only extremely low thermal expansion but also internal uniformity, very low internal stress, long-term stability, and large-size consistency. Many materials need months of annealing.
Then it enters the ultra-precision shaping stage. This is no longer ordinary polishing but MRF (Magnetorheological Finishing), CCOS (Computer-Controlled Optical Surfacing), Ion Beam Figuring. Ion beam figuring is especially critical because mechanical polishing is no longer sufficient; atomic-level material removal is required. The real difficulty is that removing even a tiny amount of material causes the entire surface shape to change. So the entire manufacturing process becomes: measurement → correction → re-measurement → re-correction, possibly looping hundreds of times.
Finally, the most difficult part: how to know whether the error comes from the mirror or from the measurement system itself? Industry uses three-mirror methods, multi-mirror mutual measurement, cross-calibration, and national laboratory benchmarks. Often, there is no absolute correctness—only the continuous reduction of uncertainty.
When precision reaches tens of picometers, the entire environment begins to become an enemy. Microearthquakes, building vibrations, airflow, temperature changes, human footsteps—all can affect the results. Therefore, many top metrology laboratories control temperature to 0.001℃, use active vibration isolation, deep foundations, vacuum environments, and sometimes only measure at night, because daytime vibrations are greater.
Thus, the real difficulty of EUV and high-end DUV is never a single component but the entire ultra-precision industrial civilization's collaborative capability. Zeiss's true irreplaceability is not just the lens itself but the decades of optical design, error compensation, system-level algorithms, reference optics, assembly experience, ultra-precision metrology, process databases, and talent systems formed over time. These elements together constitute the "precision infrastructure" of modern lithography industry.