Leveraging an Artificial Intelligence (AI) Based Approach for Semiquantitative Analysis Using ICP-OES for Sample Preparation and Screening

Metals Analyses
Oral Presentation

Presented by M. Allin
Prepared by S. Sengupta1, D. Kutscher2, T. Stichel3
1 - Thermo Fisher Scientific, Hanna-Kunath Strabe 11 , Bremen, 28199, Germany
2 - Thermo Fisher Scientific, , ,
3 - Thermo Fisher Scientific, Im Steingrund 4, Dreieich, 63303, Germany


Contact Information: [email protected] ; +494215493528

ABSTRACT

Modern inductively coupled plasma optical emission spectroscopy (ICP-OES) systems have seen significant advances versus prior generations. Current state systems can cover the complete wavelength range across the UV and the visible range of the spectrum in a single analysis. This allows for almost the entire periodic table of elements to be analyzed with high accuracy, low detection limits, and short analysis times.
However, for optimum results, some understanding of the sample composition is highly beneficial to find the right instrument setup (i.e. sample inlet configuration) or define calibration ranges that ensure all analytes are covered at the concentration level in which they are present in a sample. In environmental analysis, this however is not always the case, and laboratories may often face samples with large differences in matrix composition or analyte concentrations. To facilitate effective analysis, an estimate of the presence of elements and their concentration ranges, with reasonable uncertainty, might be sufficient for the laboratory to proceed with next steps in their workflow.
An innovative AI/ML based approach for semiquantitative analysis can speed this up and enable scientists to obtain multielement data of unknown samples even without the need for running any calibration standards. This study will compare the results of fully quantitative versus semiquantitative analysis for different sample types commonly found in environmental laboratories, such as brackish or sea waters, leachates, rock digests etc. using ICP-OES to demonstrate the usability of such analyses in various geological applications.