Quantum Simulations for Extreme Ultraviolet Photolithography

TL;DR

This paper develops quantum simulation algorithms to accurately model high-energy photon interactions in EUV lithography, aiming to improve the resolution limits in semiconductor manufacturing.

Contribution

It introduces novel quantum algorithms for computing photoabsorption and photoelectron spectra at 92 eV, enabling more precise modeling of EUV lithography processes.

Findings

Quantum algorithms resolve photoabsorption cross-section at 92 eV.

Resource estimates show feasibility with ~200 qubits and 10^9 gates.

Modeling continuum states requires significantly higher computational resources.

Abstract

Extreme Ultraviolet (EUV) lithography is the state-of-the-art process in semiconductor fabrication, yet its spatial resolution is fundamentally limited by the ``blur'' originating from absorption of photons at 92 eV, which induce physical and chemical changes in the photoresist via excited state processes and electron cascades. Accurate modeling of these phenomena requires precise ab initio data for high-energy decay channels, specifically photoabsorption and photoelectron emission. These are computationally difficult for classical methods due to prohibitive scaling in simulating electron dynamics, or due to the inability to resolve the ionization continuum in an efficient manner. In this work, we present quantum simulation algorithms to compute these key observables. First, we introduce a coherent time-domain spectroscopy algorithm optimized to resolve the photoabsorption cross-section at the 92 eV operating frequency. Second, we develop a first-quantized plane-wave simulation to compute the photoelectron kinetic energy spectrum, utilizing real-time dynamics and energy windowing to treat bound and delocalized scattering states on equal footing. Additionally, we provide logical resource estimation for a model photoresist monomer, 4-iodo-2-methylphenol (IMePh), and demonstrate that 92 eV absorption sensitivity can be resolved using roughly $200$ logical qubits and $10^{9}$ total non-Clifford gates per circuit with approximately $10^3$ shots for the smallest instance. The more sophisticated photoemission algorithm that models the continuum explicitly, incurs gate costs of $\geq 10^{13}$ total non-Clifford gates per circuit, $10^4$ shots, and requires a few thousand logical qubits. These results establish high-fidelity quantum simulations as a key component to parameterize the multi-scale macroscopic models required to overcome the electron blur bottleneck in semiconductor miniaturization.