3D-printed photoacoustic spectroscopy for industrial process control

Student thesis: Doctoral Thesis


Photoacoustic spectroscopy (PAS) is a well-established technique for trace gas sensing reaching above parts-per-quadrillion levels. This technique consists of the generation and detection of an acoustic signal from a gas sample using sensitive microphones, by the interaction of the gas with a modulated laser beam at an optical wavelength coinciding with a spectral feature of the target gas molecule. Over the past decades, many PAS techniques and sensors have been developed and deployed for applications such as: environmental monitoring, medical applications and life science applications. However, PAS applications in industrial uses was dwarfed by other gas sensing techniques, due to its requirement for frequent re-calibration for changing environmental conditions and input light power fluctuations limiting its deployment for on-site, long-term remote applications and limitations in simultaneous multi-gas measurement. The work presented here aims to present 3D-printed longitudinal resonant PAS cells that can be constructed in 4 hours with a robust, miniaturised form factor (volume < 20,000 mm3) which can address these two main drawbacks of PAS technique: calibration and multi-gas measurement. To address the calibration problem, the development of theory and application of a calibration-free wavelength modulation photoacoustic Spectroscopy (CF-WM-PAS) technique using quantum cascade lasers operating at 8.65 μm wavelength is presented. The CF-WM-PAS technique was mainly developed for measurement of SO2 gas in industrial desulphurisation process. This method uses 2f/1f calibration technique, where the second harmonic (R2f ) component which is dominated by laser-gas interaction and optical intensity, is normalised by the first harmonic (R1f ) component dominated by a newly discovered DC offset in PAS called cell dependent absorption signal (CDAS) found to be originating from laser-resin interaction, in order to isolate the output from changes in the gas matrix, optical intensity and electrical gain. The normalisation technique is confirmed for changes in laser modulation frequency, gas concentration and attenuation in input light intensity. A normalized noise equivalent absorption (NNEA) of 1.37 × 10−8 Wcm−1Hz−1/2 for calibration-free R2f/R1f measurements is demonstrated, reaching sensitivity of σ = 232 ppb for SO2 gas. Using the same PAS sensor and setup, the effect of different types of acoustic sensors, (an electret microphone (ECM) and a micro-electromechanical system (MEMS) microphone) on the PAS signal was also investigated, and it was concluded that ECM can result in higher sensitivities owing to their broader variant types allowing a simpler assembly process. Using calculated NNEA values for both microphones, 70 % superior performance was found with the ECM compared to the MEMS microphone. To address PAS multi-gas measurement, a fiber optic based, 3D-printed, miniaturised PAS sensor capable of measuring two gases simultaneously, operating at telecommunications wavelength range, is also developed as a part of this work. The multi-gas PAS gas sensor employs a double resonator for measuring CO and CO2 in near-infrared range was developed using two DFB lasers, as an alternative to gas chromatography technique. Using the resonators operating at 10.25 kHz and 13.8 kHz resonant frequencies, the simultaneous gas measurement was demonstrated using erbium doped fibre amplifiers (EDFA), resulting in CO2 sensitivities of 2,032 ppm at 26 s and 13,008 ppm at 81 s for resonators 1 and 2 respectively. The corresponding NNEA values are 1.52 × 10−7 Wcm−1Hz−1/2 and 2.89 × 10−6 Wcm−1Hz−1/2 for resonators 1 and 2 respectively, demonstrating the possibility for multi-gas sensing measurement that would have normally required two sensors, which results in overall reduction of cost, compactness and complexity.
Date of Award4 Aug 2022
Original languageEnglish
Awarding Institution
  • University Of Strathclyde
SponsorsUniversity of Strathclyde
SupervisorMichael Lengden (Supervisor) & Walter Johnstone (Supervisor)

Cite this