The cleanroom's constant‑temperature, constant‑humidity environment completely isolates external seasonal shifts and day‑night cycles, leaving only the low hum of instrument operation and the tireless flicker of indicator lights, forming a near‑eternal, cold background noise. Xiuxiu stood beside the massive EUV lithography prototype, her gaze penetrating the thick observation window, settling on the dual‑stage system inside—the system that consumed the team's countless efforts, just achieving preliminary stable operation. That pair of precise "dancers" now lay silent, as if gathering strength, awaiting command to dance with the next more severe challenge. Yet Xiuxiu deeply knew that conquering the peak of motion precision was merely one station on the long journey. Now, looming before them was another, more stubborn and physically fundamental barrier—light itself, or rather, the mirror that bears and guides light.
EUV lithography differs generationally in principle from the DUV (deep ultraviolet) lithography they were familiar with. DUV uses 193‑nm wavelength deep‑ultraviolet light; this light can be refracted and focused by special lens materials (like fused silica), through complex lens assemblies, finally forming reduced circuit patterns on silicon wafers. But entering the EUV era, the light‑source wavelength drops sharply to 13.5 nm, in the extreme‑ultraviolet band.
This is a natural obstacle set by physical laws. **Almost all known substances exhibit strong absorption, not transparency, to 13.5‑nm‑wavelength extreme‑ultraviolet light.** This means traditional refraction‑based lens assemblies completely fail in the EUV band. Light passing through any lens material would be drastically absorbed, unable to transmit and focus effectively.
The solution is abandoning refraction, returning to reflection. EUV lithography machines adopt all‑reflective optical systems. And the core constituting this reflective system is no longer ordinary mirrors but a special structure called **Bragg Reflector**.
Xiuxiu walked to a display sample reflective mirror; it looked like an extremely smooth disk with peculiar metallic sheen. But its secret hides on a nanoscale invisible to the naked eye.
"What we face is not a simple mirror," she explained to several newly‑joined young engineers beside her, voice especially clear in the silent cleanroom. "It is an **artificially synthesized periodic nanostructure**."
She elaborated the principles:
"Imagine we don't use a single layer of material to reflect light, but two materials with different optical properties, alternately stacked, forming hundreds or even thousands of pairs of precisely controlled thin‑film layers. Each pair's film thickness is calculated and controlled with extreme precision, making it exactly equal to **half the wavelength of 13.5‑nm‑wavelength light in that material**."
She gestured in air, simulating the stacked structure.
"When extreme‑ultraviolet light irradiates this multilayer‑film structure, weak reflection occurs at each interface. Because each film layer's thickness satisfies the 'half‑wavelength' condition, all these weak reflected lights from different interfaces, upon leaving the mirror surface, have their peaks align perfectly with peaks, troughs with troughs, thus producing **constructive interference**. Thousands of such constructive‑interference layers stack up ultimately achieve high reflectivity for a specific wavelength (13.5 nm) light at a particular angle (typically near‑normal incidence)."
This is the Bragg‑reflection principle, named after physicist father‑son Bragg who discovered it. Every reflective mirror in EUV lithography machines—from the collector mirror gathering light‑source emission, to the intermediate optical integrator, mask‑illumination system, to the projection objective (containing over ten mirrors) shrinking the mask pattern—all consist of such Bragg‑reflection mirrors formed by alternately depositing molybdenum (Mo) and silicon (Si). The entire optical path is light constantly reflecting, transmitting between these mirrors, finally reaching the silicon wafer.
"Our current achievable reflectivity, theoretical upper limit is around 70%," Xiuxiu's tone carried a hint of gravity, "while DUV lens transmittance easily exceeds 99%. This means EUV light loses over 30% energy each reflection. After a dozen reflections throughout the optical system, light energy ultimately reaching the silicon wafer might be less than 2% of source‑emission energy."
This leads to another core challenge of EUV lithography: **source power**. To obtain sufficient light intensity on silicon wafers, inducing effective chemical reactions in photoresist (achieving exposure), the source must possess extremely high power at the starting point. Industry consensus: to achieve economically viable mass production, EUV source power must reach **250 watts** or above. Xiuxiu's team's current target is first to conquer the entry‑level **100‑watt** barrier.
The technical route adopted is LPP (laser‑produced plasma): using high‑power carbon‑dioxide lasers to precisely bombard micron‑scale tin droplets falling at tens‑of‑thousands per second, generating high‑temperature plasma; plasma during cooling radiates extreme‑ultraviolet light including 13.5‑nm wavelength.
Through arduous, determined effort, they achieved a series of breakthroughs in laser stability, tin‑droplet generation and control, plasma confinement, etc. Finally, in latest experiments, measurement‑device readings trembled yet indeed touched the **100‑watt** threshold!
The control room briefly erupted in cheers, but soon frozen by new, more severe reality.
As source power increased, a previously faintly visible problem abruptly magnified into a fatal obstacle—**mirror thermal deformation**.
Extreme‑ultraviolet light absorbed by reflective mirrors (though single reflection absorbs ~30%, after multiple reflections and source‑power increase itself, total absorbed energy becomes extremely significant) converts into thermal energy. This heat deposits in the mirror's multilayer‑film structure.
"Molybdenum and silicon, these two materials have different thermal‑expansion coefficients," Xiuxiu pointed at the thermal‑imaging screen showing uneven temperature‑distribution reflective‑mirror model, brows furrowed. "When heated, they expand differently. More importantly, these hundreds of film pairs are deposited on a specific substrate material (typically ultra‑low‑expansion glass or silicon‑carbide compound). The substrate itself also has thermal‑expansion coefficient; moreover heat conduction, distribution in multilayer film and substrate isn't uniform."
Consequences are catastrophic. Uneven thermal load causes **microscopic yet sufficient‑to‑destroy‑imaging‑quality deformation** of the mirror surface.
"This deformation may be only a few picometers to a few nanometers," Xiuxiu's voice lowered, "but for EUV lithography, required wavefront error needs controlling below **50 picometers**! This means the mirror surface's shape accuracy must be maintained within approximately **one atomic diameter** scale!"
Thermal deformation, under power increase, instantly turns this atomic‑scale precision requirement into foam. The mirror no longer is ideal plane or curved surface; it may warp like a heated potato chip—tiny yet enough to divert optical paths. This directly causes light‑spot distortion projected onto silicon wafers, image blur, overlay‑accuracy loss. Simply put: light exists, but the "bowl" bearing light deforms under energy impact, unable to precisely "scoop" light where needed.
Experimental data mercilessly confirms this. When light‑source power barely maintains 100‑watt output, subsequent optical‑system monitoring shows wavefront error deteriorating drastically, far beyond tolerance limits. The overlay‑precision advantage brought by dual‑stage conquest becomes meaningless before this optical‑system‑fundamental distortion.
Immense frustration, like the cleanroom's cold air, permeates every pore. Team members' excited glow rapidly dims, replaced by exhaustion and bewilderment. They see a seemingly summit‑near peak, at final stretch, transforms into smooth, mirror‑like,anywhere‑foothold cliff.
Xiuxiu feels unprecedented helplessness. Dual‑stage challenges, though difficult, ultimately belonged to mechanical‑control, servo‑algorithm, material‑mechanics domains—always countless parameters to adjust,thinking to try. But current thermal‑deformation problem points directly at material's physical limits under extreme conditions. Not solvable via more ingenious algorithms or more extreme processing. It involves complex thermal‑mechanical‑optical multi‑physics coupling, involves material behavior's unpredictability at nanoscale—this is almost a hard battle at engineering's existing cognitive boundary.
Late night, she stays alone in rest area outside lab, lights off, only distant city lights outside window providing faint source. Immense pressure and confused path make her first time so clearly feel personal tininess. In the macro, state‑will and industry‑demand‑driven technology long march, individual effort and wisdom sometimes seem so insignificance. She recalls Mozi's "black swan" experience—that feeling struck by force beyond own cognitive scope—now she empathizes.
Unconsciously, she takes out private communicator, without thought, dials Mozi's number. She doesn't know if he's busy, nor what to say. Just under this immense loneliness and pressure, instinctively wanting to hear an understanding voice.
Phone quickly connects; Mozi's background quiet,seems also working.
"Xiuxiu?" His voice transmits, carrying a hint of surprise, but more concern. "So late not resting? Something wrong?"