Why It Was Almost Impossible To Make Transistors Less Than 10 nm

Quick Overview

The creation of Extreme Ultraviolet (EUV) lithography machines, which enable feature sizes down to 8nm and beyond, was nearly impossible due to fundamental physics challenges, primarily the lack of sufficiently powerful EUV light sources and the inability to create mirrors smooth enough to reflect the necessary short-wavelength light, necessitating the development of complex, multi-layered mirrors (like those using Molybdenum and Silicon) and high-power laser-produced plasma sources (like those achieving 100W power) to continue shrinking transistors as predicted by Moore's Law.

Key Points: Moore's Law faced a critical challenge around the late 1990s/early 2000s because traditional light sources (like 193nm UV) could no longer print features as small as 90nm, leading to the industry shifting focus to EUV lithography. The key technical hurdle for EUV was finding a source powerful enough: early experimental sources produced only about 10W of EUV power, while commercial viability required over 100W (with some aiming for 150W). The early X-ray lithography work by Kinoshita and his team (1980s) proved the concept using multilayer mirrors made of Tungsten and Carbon, which reflected around 6% of the light at the required 13.5nm wavelength. ASML's high-NA EUV machines overcome the Rayleigh equation limits by using projection optics with numerical apertures (NA) exceeding 1.0 (up to 0.55 for low-NA, later higher) and employing complex mirror systems (like those from Zeiss) to precisely focus the light. The EUV light source (Laser Produced Plasma) requires extremely high energy pulses (e.g., 15 mJ) hitting a tin droplet traveling at high speed (50,000 droplets per second) to generate the necessary 13.5nm light, which is then collected by specialized mirrors. The extreme precision required for these systems is highlighted by the need for mirror surface roughness less than 0.1 nanometers, far smoother than a playing card, and the environment must be extremely clean (ASML fabs allow only 10 particles/m³ of 0.1-micron particles, compared to 10,000 in operating rooms). Despite early skepticism and funding setbacks (like the DOE ending funding for the National EUV Lithography Program in 1996), collaborative efforts, including EUV LLC, led to the development and commercialization of EUV lithography, exemplified by the ASML Twinscan NXE:3300B system.