Skip to Main content Skip to Navigation
Theses

Étude métrologique de solutions ioniques par spectrométrie Raman et analyses statistiques

Abstract : The objective of this research is to study the possibility of using conventional Raman spectroscopy for analysis, or in situ control of substances dissolved in water and to determine the limits of this technique. Several salt solutions were studied. Polyatomic ions such as nitrate (NO3-) give intense Raman lines around 1000 cm-1. Halide ions (F-, Cl-, Br-, I-) have no direct signature but indirectly affect the water spectrum. The methodology proposed in this work for identification and quantification of salt solutions is based on classical analysis of spectra using pre-treatment methods (normalization, baseline correction). The limits of detection of the studied ions correspond to the water potability limits. It is possible to go below the potability limits for nitrates and sulfates for example but it is impossible to reach them for some other substances (nitrite, ammonium). Then, statistical methods were used. Thus, the principal component analysis (PCA) allows to identify the nature of a salt solution (with direct or indirect signatures) and to acces to the molecular arrangements of ions with water. PLS regression quantifies the ions through predictive models. This method was used on pure solutions of nitrate and chloride with errors on the concentration predictions 4 to 5 times lower than those found by classical methods. Lower concentration ranges (10-100 mM) for the Cl- ion are reachable using these statistical methods.
Document type :
Theses
Complete list of metadata

https://hal-centralesupelec.archives-ouvertes.fr/tel-01498644
Contributor : Thomas Kauffmann <>
Submitted on : Thursday, March 30, 2017 - 1:45:59 PM
Last modification on : Thursday, November 22, 2018 - 8:44:04 AM

Identifiers

  • HAL Id : tel-01498644, version 1

Citation

T.H. Kauffmann. Étude métrologique de solutions ioniques par spectrométrie Raman et analyses statistiques. Physique [physics]. Université de Lorraine (Nancy), 2016. Français. ⟨tel-01498644v1⟩

Share

Metrics

Record views

40

Files downloads

68