27 Novembre – Thesis defense - Julian Guézénoc
14 h30 Amphi G - Building A29
AGROLIBS Laser-induced plasma spectroscopy for the sustainable management of agricultural resources
In a context of global climate change, optimizing the management of agricultural resources, and in particular the mineral elements in soils, has become a major challenge. Mineral elements are at the same time indispensable resources for plant growth and carbon storage, potential pollutants in certain environmental compartments, and potentially toxic contaminants that can be transferred from plants to humans. Current methods for the determination of these elements are very costly in terms of time and money, and from an operational point of view, it is difficult in these conditions to carry out a frequent diagnosis in the field of a deficiency or contamination situation, or to characterize the spatio-temporal variability of the characteristics of an agricultural land.
The recent emergence of portable sensors and systems has made it possible to develop new methods for measuring soil and plant properties and to complement or replace conventional laboratory techniques. This is the case of LIBS, a method of atomic emission spectroscopy from a laser-generated plasma, which is fast and green as it does not rely on the use of hazardous chemical reagents. The objective of this study is to evaluate the performance of the LIBS method, and more precisely of a portable instrument, applied to the quantitative multi-elemental analysis of plant and agricultural soil samples. The evaluation of this method takes into account the sample preparation and processing steps of LIBS spectra. It was implemented according to three axes: i. The characterization of plant samples of various species under ideal conditions, ii. The characterization of soil samples representative of the major French agricultural crops (wheat, corn, sunflower) always under ideal conditions and iii. The study of the impact of factors degrading the measurement conditions in order to evaluate the possibility of implementing the LIBS technique outside the laboratory.
Our study, based on the use of the SciAps Z300 LIBS, allowed the detection of the elements C, Ca, Fe, K, Mg, Na, Si and P, concentrated in the g/kg range in soil and plant samples, both in laboratory conditions and in deliberately degraded conditions. On the other hand, neither nitrogen nor the elements Cd, Cu, Mn and Zn could be detected in these samples. For the analysis of plant samples, the elements Ca, Fe, Mg, Na, and P were quantified using univariate regression models coupled with an adapted normalization strategy. Quantitative soil analysis required the use of PLS models to account for matrix effects. As regards the analyses in degraded conditions, we established, using an experimental design, that the moisture content, higher than 40% in plants and 20% in soils was one of the main obstacles to LIBS field analyses. We also showed that the drop in LIBS signal observed when the moisture content increases could be corrected by applying a normalization step.
The results of our study are partially satisfactory in terms of the quantitative performance of the regression models, even under ideal laboratory conditions. The diversity of physico-chemical conditions encountered in the field means that direct LIBS analyses are clearly not realistic. The principle of a field laboratory, with simplified sample preparation, could however be envisaged, in order to place the LIBS technique at the heart of new sampling strategies in the context of precision agriculture.