Managing Spectral Intensity Variations With In-Line Fiber-Optic High-Volume Noncontact Raman Probes: A Strategy Exploration

Sammendrag

Noncontact fiber-optic Raman probes are increasingly deployed in industrial environments, where variations in laser power andworking distance can compromise quantitative performance. We investigated the stability of partial least squares regressionunder combined changes in laser power (200–450 mW) and working distance (5–30 cm), comparing intensity-correction strate-gies that preserve absolute intensity related to absolute concentration with approaches that suppress it. The effect was discussedin terms of absolute fat concentration measurements and compared with the relative concentration measurement of fatty acid(FA) 18:1 expressed as % of fat. Overall, both correction strategies yielded comparable slope and variance errors, with bias emerg-ing as the primary distinction. Laser power fluctuations were effectively corrected using sapphire-based normalization or othermethods suppressing intensity variations (bias fat ∈ [ − 0. 2 % , 0. 0 % ] and bias FA ∈ [0. 0 % , 0. 4 % ]). Working distance effects werenonlinear and required dedicated compensation. We demonstrated the potential of using nitrogen signals from the air for work-ing distance corrections in bulk fat measurements (bias fat ∈ [ − 0. 2 % , − 0. 1 % ]). For FA 18:1, accurate quantification dependedon explicit removal of instrument and air background, and only an extended multiplicative scatter correction (EMSC) includingboth background components as separate interferents achieved acceptable bias, providing bias FA ∈ [ − 0. 2 % , 0. 3 % ], comparedwith bias FA ∈ [0. 0 % , 7. 5 % ] for MSC. These results outline practical strategies for maintaining quantitative robustness in non-contact Raman measurements under realistic industrial variability.

Les publikasjoner her:

DOI : doi.org/10.1002/jrs.70160
NVA : hdl.handle.net/11250/5512182