ASML repurposed a bankrupt rival's multibeam technology for defect inspection
Mapper Lithography tried to replace EUV lithography with massive arrays of weak electron beams; after burning through investor patience it went bankrupt, and ASML bought the technology not to make lithography tools but to do wafer defect inspection.
The argument · tap a timestamp to hear it
Inspection tools inherited from a failed competitor
ASML's new-generation multibeam inspection tool, the eScan 2200, actually uses multibeam technology from a bankrupt Dutch company, Mapper Lithography. Mapper once tried to replace ASML's own EUV lithography tools with massive arrays of electron beams, but lost that bet and went bankrupt in 2018. But its technology didn't disappear—ASML bought it and put it to work in an entirely different application: wafer defect inspection. This episode traces that technological line through twenty years of a failed company's history.
SCALPEL failed because electrons repelled each other
Before Mapper, the SCALPEL scheme led by Bell Labs tried to adapt lithography's projection approach to electron beams: using a scattering mask magnified 4× to control the electron beam, then an aperture to filter away strongly scattered electrons. The problem was at the aperture—the mandatory crossover point—where all electrons had to squeeze through the same narrow opening, repelling each other and creating random interference that ordinary lenses couldn't correct. Bell Labs spent over a decade unable to solve it; Applied Materials and Nikon followed up with similar approaches, both failed. This became a pitfall Kruit deliberately set out to avoid when designing the Mapper approach later.
The solution: millions of faint electron beams
Kruit's approach was the opposite: instead of squeezing all electrons through a bottleneck, create millions of separate, low-power electron ‘beamlets’—low power means no mutual interference. His initial paper envisioned one million to 100 million beams. Two graduate students, Bert Kampherbeek and Marco Wieland, built the concept, naming it ‘Multiple Aperture Pixel by Pixel Enhancement of Resolution’—the acronym was MAPPER. The company was founded in 2000; one early investor was Arthur Del Prado, founder of ASM International and ‘half-father’ to ASML.
Misunderstanding the physics cost two wasted years
Mapper's initial design wanted to pair each beamlet with an independent electron source: use an array of light spots to illuminate a conversion plate that would turn light into an array of electron beams. The team first assumed the conversion plate worked like a photocathode—electrons knocked out directly by light, stable current, switchable. Later they discovered the error: it was actually a photodiode driving a field-emission tip, electrons came from an electron pool pulled out by the electric field, light just replenished that pool. Result: they couldn't precisely control electron energy and position, nor switch quickly. A year or two later they had to start over, switching to splitting a single electron beam into multiple beamlets.
TSMC invested in Mapper to hedge its bets
At the time, the industry had no consensus on next-generation lithography technology; some people even argued for abandoning the search for new tech and just using multiplex exposure indefinitely. TSMC wasn't a company to put all its bets on one technology—while funding ASML's EUV work, it also invested in Mapper in 2008, treating it as an EUV backup plan in case EUV didn't pan out. It sent engineers to buy Mapper tools for testing and feedback. A single Mapper tool targeted only 10 wafers per hour throughput, far below the hundred-wafer-per-hour class of EUV or multiplex exposure. Both sides' countermeasure was to cluster multiple machines together, trading quantity for throughput.
EUV accelerated faster than Mapper could follow
While Mapper was still struggling to reach 10 wafers per hour, ASML's EUV light source began reshaping tin droplets using ‘pre-pulse’ technology, sending power and throughput into rapid growth. Tight funding meant Mapper's iteration speed consistently lagged behind—the show mentions that if they'd been able to get even a prototype machine out faster to grab market attention, the result might have been different. Shortage of cash even forced Mapper in 2012 to accept a funding round led by Russian state-owned investment fund RUSNANO, and later to open a MEMS factory in Moscow producing lens elements.
One investor's death eliminated the company's last lifeline
The real turning point came when ASM International founder Arthur Del Prado died in 2016. He was Mapper's most reliable backer; his estate trustees said they wouldn't keep funding unless a new lead investor was found. But Mapper had no compelling story left for new investors: competing with EUV needed too much money and EUV was already too far ahead; pivoting to small-volume trial production or secure custom chips would need hundreds of millions of dollars to commercialize. The company finally filed for bankruptcy in December 2018; ASML stepped in to keep the technology and assets in the Netherlands, but after acquiring them didn't continue the lithography tool product line.
The technology didn't die; it became a defect-inspection tool
In 2016 ASML spent heavily to acquire Taiwan's Hermes Microvision (HMI) for electron-beam inspection. The problem inspection had to solve: in the EUV era, defects get smaller and optical inspection misses them, and in 2.5D/3D stacked structures many defects aren't even on the surface—electron beams can do ‘voltage contrast inspection’ that optical can't. But electron-beam inspection had the old problem: too slow, couldn't keep pace with manufacturing lines. Combining HMI's electron-beam inspection with Mapper's multibeam technology solved the throughput problem exactly. In 2022 ASML released eScan 1000 and 1100; now there's the eScan 2200, with beam counts steadily climbing.
In their own words · checked verbatim
So what is the catch? The technique's Achilles Heel is its throughput. EBDW is slow as molasses.
Kruit got two of his graduate students Bert Kampherbeek and Marco Wieland to develop the concept, which was named "Multiple Aperture Pixel by Pixel Enhancement of Resolution", or MAPPER.
Turns out that understanding was wrong. The plate wasn't actually a photocathode, but rather a photodiode feeding a field emitter.
I am somewhat reminded of water flow going through a shower head.
This achievement is a significant breakthrough. This proves the multi-beam resolution capability of Mapper's low voltage approach.
Burn Lin15:34
Had the company managed to get something out faster - even if that something was half usable - then perhaps it could have generated more momentum.
Mapper Lithography's technology failed to compete with ASML's, but its revitalization as part of this larger strategy is a nice reminder that in the semiconductor industry, a failure in one era can be a success in another.
Figures
| ASML acquisition of HMI price | $3.1 billion | 21:47 |
| ASML acquisition of Mapper IP and assets price | €75 million | 20:47 |
| Mapper employee count at acquisition | 240–270 | 20:47 |
| RUSNANO-led Mapper funding round | €80 million | 17:41 |
| eScan 1000 / 1100 beam count | 9 beams (3×3 array) / 25 beams (5×5 array) | 22:48 |
| FLX-1200 beam count and target | 1,300 beams, needs to scale to 13,000 | 18:45 |
| Contemporary EUV light source power and throughput | 80–100 watts, ~60 wafers/hour | 18:45 |
| Mapper Lithography founding year | 2000 | 7:13 |
Glossary
- EBDW
- Electron-beam direct writing: using a focused electron beam to draw chip patterns directly on photoresist without a photomask.
- SCALPEL
- Scattering-Angle Limited Projection Electron Lithography: Bell Labs' electron-beam projection approach that failed due to electron repulsion-induced blur.
- crossover
- The narrow aperture (electron convergence point) where electron beams must pass; high-density electrons repel each other here, creating blur.
- Voltage Contrast inspection
- Using an electron beam to strike the surface and measuring back-scattered electrons to assess deep subsurface electrical contact quality.
- holistic lithography
- ASML's approach of integrating exposure, computational lithography, and inspection/metrology into one closed-loop optimization system.
How to listen
Semiconductor equipment engineers, investors following ASML/EUV history, and people interested in how failed technology roadmaps get resurrected.
The opening two minutes (0:00–2:00) on EBDW fundamentals and classification; heavy on technical background, can skip if short on time.