A calibration/validation protocol for long/multi‐pathlength capillary waveguide spectrometers. (2nd October 2018)

• Record Type: Journal Article
• Authors: Cartisano, Carmen M.; Del Vecchio, Rossana; Blough, Neil V.
• Is Part Of: Limnology and oceanography, methods. Volume 16:Number 11(2018:Nov.)
• Page Start: 773
• Page End: 786
• Publication Date: 2018-10-02
• Subjects: Limnology -- Methodology -- Periodicals; Oceanography -- Methodology -- Periodicals; 551.48
• DOI: 10.1002/lom3.10282
• Languages: English
• ISSNs: 1541-5856

Abstract

Abstract: Long/multi‐pathlength spectrometers are being employed increasingly in ocean science, but detailed calibration/validation protocols for these instruments remain unavailable. The lack of such protocols has led to the collection of absorption data from different instruments that are not always comparable. The goal of this work was to develop procedures that improve inter‐laboratory data reproducibility and accuracy. A World Precision Instrument (WPI) multi‐pathlength liquid capillary waveguide spectrometer along with a Shimadzu dual‐beam spectrometer were employed to evaluate the proposed protocols, which address the following issues: (1) sample cross‐contamination due to improper cell cleaning, (2) nonstandardized sample preparation, (3) errors in effective pathlength determination, and (4) offsets due to refractive index effects (i.e., high‐salinity samples). Effective pathlength calculation is critical given that the measured and manufacturer provided pathlengths can differ significantly, with the provided pathlengths resulting in data with large variation (up to 21%) and percent error (up to 33%) when employing different pathlengths. When these proposed protocols were applied, we found that the WPI spectrometer's precision (measured as a coefficient of variation) was < 0.6% for intra‐day measurements and < 1.5% for inter‐day measurements, and there was < 2% error in absorbance measurements. These procedures allowed for the correction of the significant baseline offsets caused by differences in refractive index between reference and sample, by either subtracting a collected or a generated reference spectrum that matches the offset. Our results indicate that employing the procedures described herein will provide accurate, reproducible, and pathlength‐independent absorption data.