Machine learning-enabled prediction of optical emission spectra in scalable non-thermal Ar atmospheric pressure plasma multi jet

Abstract In this study, an Ar-multi-non-thermal atmospheric pressure plasma jet system was designed and fabricated to explore its scalability for large-volume processing. Optical emission spectroscopy measurements were conducted as a function of various operating parameters, including gas flow rate, voltage, duty cycle, and % air. The measurements revealed strong emissions from Ar I, originating from fine-structure transitions among the <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" overflow="scroll"> <mml:mrow> <mml:mn>3</mml:mn> <mml:msup> <mml:mi>p</mml:mi> <mml:mn>5</mml:mn> </mml:msup> <mml:mn>4</mml:mn> <mml:mi>p</mml:mi> <mml:mo stretchy="false">→</mml:mo> <mml:mn>3</mml:mn> <mml:msup> <mml:mi>p</mml:mi> <mml:mn>5</mml:mn> </mml:msup> <mml:mn>4</mml:mn> <mml:mi>s</mml:mi> </mml:mrow> </mml:math> levels within the 690–810 nm wavelength range. Additionally, significant emissions were observed from the OH rotational band ( <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" overflow="scroll"> <mml:mrow> <mml:msup> <mml:mi>A</mml:mi> <mml:mn>2</mml:mn> </mml:msup> <mml:msup> <mml:mi mathvariant="normal">Σ</mml:mi> <mml:mo>+</mml:mo> </mml:msup> <mml:mo stretchy="false">→</mml:mo> <mml:msup> <mml:mi>X</mml:mi> <mml:mn>2</mml:mn> </mml:msup> <mml:mi mathvariant="normal">Π</mml:mi> </mml:mrow> </mml:math> ) at 309 nm, the N 2 second positive system ( <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" overflow="scroll"> <mml:mrow> <mml:msup> <mml:mi>C</mml:mi> <mml:mn>3</mml:mn> </mml:msup> <mml:msub> <mml:mi mathvariant="normal">Π</mml:mi> <mml:mi>u</mml:mi> </mml:msub> <mml:mo stretchy="false">→</mml:mo> <mml:msup> <mml:mi>B</mml:mi> <mml:mn>3</mml:mn> </mml:msup> <mml:msub> <mml:mi mathvariant="normal">Π</mml:mi> <mml:mi>g</mml:mi> </mml:msub> </mml:mrow> </mml:math> ) between 330 nm and 380 nm, and atomic oxygen ( <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" overflow="scroll"> <mml:mrow> <mml:mn>2</mml:mn> <mml:msup> <mml:mi>s</mml:mi> <mml:mn>2</mml:mn> </mml:msup> <mml:mn>2</mml:mn> <mml:msup> <mml:mi>p</mml:mi> <mml:mn>3</mml:mn> </mml:msup> <mml:msup> <mml:mo stretchy="false">(</mml:mo> <mml:mn>4</mml:mn> </mml:msup> <mml:msup> <mml:mi>S</mml:mi> <mml:mn>0</mml:mn> </mml:msup> <mml:mo stretchy="false">)</mml:mo> <mml:mn>3</mml:mn> <mml:mi>p</mml:mi> <mml:msup> <mml:mstyle scriptlevel="0"/> <mml:mn>5</mml:mn> </mml:msup> <mml:mi>P</mml:mi> <mml:mo stretchy="false">→</mml:mo> </mml:mrow> </mml:math> <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" overflow="scroll"> <mml:mrow> <mml:mn>2</mml:mn> <mml:msup> <mml:mi>s</mml:mi> <mml:mn>2</mml:mn> </mml:msup> <mml:mn>2</mml:mn> <mml:msup> <mml:mi>p</mml:mi> <mml:mn>3</mml:mn> </mml:msup> <mml:msup> <mml:mo stretchy="false">(</mml:mo> <mml:mn>4</mml:mn> </mml:msup> <mml:msup> <mml:mi>S</mml:mi> <mml:mn>0</mml:mn> </mml:msup> <mml:mo stretchy="false">)</mml:mo> <mml:mn>3</mml:mn> <mml:mi>s</mml:mi> <mml:msup> <mml:mstyle scriptlevel="0"/> <mml:mn>5</mml:mn> </mml:msup> <mml:msup> <mml:mi>S</mml:mi> <mml:mn>0</mml:mn> </mml:msup> </mml:mrow> </mml:math> ) at 777.5 nm, providing valuable insights into reactive chemical species formation. To analyze the influence of operating conditions on these emissions and their impact on reactive species production, a data-driven machine learning (ML) approach was employed, leveraging a combination of dimensionality reduction and supervised learning. Principal component analysis was first applied to reduce data dimensionality, enhancing computational efficiency while preserving essential spectral features. Subsequently, two supervised ML models, Random Forest and deep neural networks, were trained to map input conditions to predicted emission intensities. To further exploit the predictive capability of these models, a differential evolution optimization strategy was implemented to estimate the optimal combinations of input parameters that maximize specific spectral features, particularly OH(A–X) emission. This study contributes to the development of more efficient and accelerated plasma diagnostic approaches, with the potential for real-time predictions.