ASML’s Lithography Edge: 40 Years to Maskless

How ASML Built Unmatched Lithography Leadership

Early foundations: from a joint venture to TWINSCAN

ASML’s lead wasn’t a single breakthrough. It was built over 40 years of systems engineering, tight supplier coordination, and big bets on new lithography generations. ASML was founded in 1984 as a Philips–ASM joint venture. Early modular steppers helped it stand out.

ASML then rode the shift from step-and-repeat to step-and-scan. In the late 1990s, TWINSCAN became a turning point. It used two wafer stages. One stage measured and aligned while the other exposed. Throughput rose without sacrificing overlay.

Immersion lithography and a supplier moat

Immersion lithography locked in ASML’s early advantage. DUV scanners use 193 nm ArF light. Putting ultra-pure water between the final lens and the wafer raises numerical aperture. Higher NA means better resolution.

ASML commercialized immersion quickly. Computational lithography, OPC, and multiple patterning extended 193 nm far beyond expected limits. In 2013, ASML acquired Cymer and secured critical laser sources. Long-term partnership with Zeiss built an ecosystem of extreme precision. That ecosystem is hard to copy at scale for semiconductor manufacturing.

EUV: the most complex industrial tool

EUV widened the moat further. It uses 13.5 nm light made by firing high-power lasers at tiny tin droplets. EUV is absorbed by air and glass, so the system runs in a vacuum. It also relies on reflective multilayer mirrors, not lenses.

Photon losses compound at every mirror. Teams had to solve many problems together: contamination, stochastic resist effects, mask defects, source power, overlay, and uptime. ASML coordinated thousands of engineers and billions of euros across customers and suppliers. EUV finally became viable for high-volume production.

High-NA EUV: pushing the next node

EUV scanners are full systems, not just optics. They integrate light generation, nanometer-precision mechatronics, thermal control, vacuum engineering, metrology, and software correction across the full process.

High-NA EUV raises NA from 0.33 to 0.55. It targets roughly 8 nm resolution. It can remove many multi-patterning steps. Customers buy yield, throughput, service data, and a roadmap aligned with leading fabs.

The coming maskless lithography disruption

Why masks are a bottleneck

Projection lithography is productive once a mask exists. But masks are costly and slow to qualify. EUV masks need correction, inspection, repair, pellicles, and long cycles.

That cost hits low-volume work hardest. Examples include custom chips, rapid revisions, heterogeneous integration, advanced packaging, photonics, and research.

How maskless works—and the core tradeoff

Maskless lithography writes patterns from digital data. It can use e-beams, ion beams, scanned lasers, or programmable micromirrors. The tradeoff is throughput. Serial writing is precise but slow.

Scaling requires parallelism. Multi-beam e-beam systems split work across hundreds of thousands of beamlets, and eventually millions. They also need fast computing for proximity correction, dose control, stitching, and defect compensation.

Evolution first, not immediate EUV replacement

Near-term impact is evolutionary. Multi-beam tools already help make EUV masks faster. Direct write fits niches where flexibility matters more than wafers per hour.

Likely uses include mask repair, prototypes, custom layers, chiplets, interposers, and low-volume AI accelerators. Digital exposure also enables per-die variation and distortion correction.

The transition will be driven by cost per good layer. Lead time, yield learning, energy, and volume mix will matter as much as headline resolution.

Final verdict: hybrid lithography

For leading-edge, high-volume logic, ASML’s optical replication advantage is still strong. Any maskless challenger must match resolution, overlay, defectivity, resist needs, and hundreds of wafers per hour. It must also process huge pattern data streams in real time.

Maskless is unlikely to replace ASML outright. A hybrid future is more plausible. EUV will print dense, repeated layers at scale. Maskless tools will print variable layers and serve cases where mask costs dominate.

ASML won by industrializing hard physics. The next challenge is knowing when computing and parallel direct write make masks optional for more layers.