Abstract

We report on the application of neural-network processing to pulsed photoacoustics for improving the detection limit by subtracting the window-heating-associated background. This technique was applied to the measurement of ethylene traces excited by a TEA (transverse electrical discharge in gas at atmospheric pressure) ${\text{CO}}_{\text{2}}$ laser. The signal contains a term that shows absorption saturation, characteristic of the absorbing gas, and another, generated by window heating, linearly dependent on laser energy. At low concentrations, normalization to laser energy is not possible owing to the different absorption mechanisms. To overcome this problem we relied on a neural-network filter, trained with experimentally obtained patterns, that subtracts the background and returns the sample concentration. This way, we reduced the detection limit to 20% of the previous limit obtained by reading the main resonance peak amplitude.

© 2007 Optical Society of America

Photoacoustic spectroscopy on trace gases with continuously tunable CO₂ laser

Philippe Repond and Markus W. Sigrist
Appl. Opt. 35(21) 4065-4085 (1996)

Quantitative signal analysis in pulsed resonant photoacoustics

Stefan Schäfer, András Miklós, and Peter Hess
Appl. Opt. 36(15) 3202-3211 (1997)

Pulsed-laser excitation of acoustic modes in open high-Q photoacoustic resonators for trace gas monitoring: results for C₂H₄

Christian Brand, Andreas Winkler, Peter Hess, András Miklós, Zoltán Bozóki, and János Sneider
Appl. Opt. 34(18) 3257-3266 (1995)

Highly sensitive electronically modulated photoacoustic spectrometer for ozone detection

M. A. Gondal and Z. H. Yamani
Appl. Opt. 46(29) 7083-7090 (2007)

Multi-MEMS-microphone schemes in a miniature photoacoustic cell for acetylene trace gas measurement

Qing He, Weirun Zhu, Haifei Lv, Xiaoyan Wen, Zhexuan Zheng, Jiafu Wang, and Min Li
Appl. Opt. 62(6) 1647-1653 (2023)