Exploring weak spots in cutting-edge lithography for microchip manufacturing

Per “Moore’s Law,” the miniaturization of integrated circuit (IC) patterns drives increases in data processing capacities with system-level chip integrations. But as circuit feature sizes approach several nanometers and the number of microchip transistors approaches the trillions, continued scaling is now seriously challenged by physical and economic limitations, which makes enhancing efficiencies an important and challenging objective.

IC fabrication mainly repeats three steps: thin layer deposition, lithography for forming fine resist patterns, and etching away layers above the patterns. While deposition and etching are achieved with atomic precision, the second step – cutting-edge lithography using extreme ultraviolet light – uses several photons or reactions to expose a 1-nm square area of resist material. Because photon numbers inevitably fluctuate, various pattern anomalies occur, including transistor feature size variations, edge misalignment, and even occasional pattern failures.

While they’ve been investigated in depth separately, Hiroshi Fukuda examined the anomalies as different aspects of the same phenomena – stochastic “hot-spots” of photon and electron activity correlating to molecular clustering.

“They are fluctuations in the dissolution of resist molecule clusters due to stochastic processes of photon absorptions, photoelectron scatterings, and chemical reactions, characterized by amplitude and spatial size,” said Fukuda. “Both circuit patterns and their anomalies are expressed by probabilities depending on locations in the patterns and materials.”

Normally, examining stochastic processes requires Monte Carlo methods, which involve a large number of time-consuming numerical experiments that are difficult to apply to big IC data. In contrast, Fukuda’s study suggested that considering them as part of a deep neural network is more appropriate for addressing lithographic inefficiencies in IC technology.

“I also hope it is helpful for better materials and designs,” said Fukuda.

Source: “Stochastic hot-spots in extreme ultra-violet exposed nano-patterns as correlated molecular sub-cluster formation probabilities,” by Hiroshi Fukuda, Journal of Applied Physics (2023). The article can be accessed at http://doi.org/10.1063/5.0150936 .