Intelligent Agents for the Optimization of Atomic Layer Deposition

Area of Science:

• Materials Science
• Chemical Engineering
• Nanotechnology

Background:

• Atomic Layer Deposition (ALD) enables precise thin film synthesis for computing, energy, and separation technologies.
• The vast chemical and process diversity in ALD presents challenges in identifying optimal parameters for stable, uniform film growth and minimal precursor use.
• Current ALD process development relies on time-consuming trial-and-error, often hindered by experimental noise and complex surface chemistries.

Purpose of the Study:

• To develop and evaluate efficient, autonomous optimization strategies for ALD process development.
• To address the challenges of identifying optimal ALD processing parameters amidst experimental noise and complex surface chemistries.
• To accelerate the discovery of processing conditions for stable, uniform ALD films with minimal precursor consumption.

Main Methods:

• Development of three distinct optimization strategies for ALD process parameter identification.
• Simulation of four different ALD processes to test and compare the performance of the developed strategies.
• Investigation into the impact of experimental noise in growth per cycle (GPC) measurements on optimization convergence.

Main Results:

• Successful development of three novel optimization strategies for ALD.
• Comparative analysis demonstrating the performance of these strategies across simulated ALD processes.
• Quantification of the effect of GPC measurement noise on the convergence of optimization algorithms.

Conclusions:

• The developed optimization strategies offer a more efficient and autonomous approach to ALD process development.
• These methods can accelerate the identification of optimal ALD conditions, reducing experimental effort and precursor waste.
• Understanding the impact of noise is critical for robust ALD process optimization.