Chemische Charakterisierung und Spurenanalytik
Filtern
Erscheinungsjahr
- 2021 (5) (entfernen)
Dokumenttyp
- Vortrag (2)
- Posterpräsentation (2)
- Zeitschriftenartikel (1)
Sprache
- Englisch (5)
Schlagworte
- Machine learning (5) (entfernen)
Organisationseinheit der BAM
- 1 Analytische Chemie; Referenzmaterialien (3)
- 1.1 Anorganische Spurenanalytik (3)
- 1.3 Instrumentelle Analytik (3)
- 1.6 Anorganische Referenzmaterialien (3)
- P Präsident (3)
- P.0 Präsident und andere (3)
- S Qualitätsinfrastruktur (3)
- S.1 Qualität im Prüfwesen (3)
- 6 Materialchemie (2)
- 6.3 Strukturanalytik (2)
Eingeladener Vortrag
- nein (2)
ML@BAMLINE
(2021)
In this talk I’ll describe the use of artificial neural networks (ANN) for quantifying X-ray fluorescence (XRF) measurements. The main idea of this talk is to give an overview of the process needed to generate a model that can then be applied to a specific problem.
In XRF, a sample is excited with X-rays and the resulting characteristic radiation is detected to determine elements quantitatively and qualitatively. This is traditionally done in several time-consuming steps. I’ll show the possibilities and problems of using a neural network to realise a "one-click" quantification. This includes generating training data using Monte Carlo simulation and augmenting the existing data set with an ANN to generate more data. The search for the optimal hyperparameters, manually and automatically, is also described. For the case presented, we were able to train a network with a mean absolute error of 0.1% by weight for the synthetic data and 0.7% by weight for a set of experimental data obtained with certified reference materials.
In this poster for the SCIX Conference, we apply an analytical procedure based on the monitoring of the lithium isotope through the partially resolved isotope shift in the electronic transition 22P<-22S around wavelength 670.80 nm using high-resolution continuum source atomic absorption spectrometer (HR-CS-AAS) in combination with machine learning (ML) for the determination of Li Isotope ratio analysis.
An alternative method for lithium isotope amount ratio analysis based on a combination of high-resolution atomic absorption spectrometry and spectral data analysis by machine learning (ML) is proposed herein. It is based on the well-known isotope shift of approximately 15 pm for the electronic transition 22P←22S at around the wavelength of 670.8 nm, which can be measured by the state-of-the-art high-resolution continuum source graphite furnace atomic absorption spectrometry. For isotope amount ratio analysis, a scalable tree boosting ML algorithm (XGBoost) was employed and calibrated using a set of samples with 6Li isotope amount fractions, ranging from 0.06 to 0.99 mol mol–1, previously determined by a multicollector inductively coupled plasma mass spectrometer (MC-ICP-MS). The calibration ML model was validated with two certified reference materials (LSVEC and IRMM-016). The procedure was applied toward the isotope amount ratio determination of a set of stock chemicals (Li2CO3, LiNO3, LiCl, and LiOH) and a BAM candidate reference material NMC111 (LiNi1/3Mn1/3Co1/3O2), a Li-battery cathode material. The results of these determinations were compared with those obtained by MC-ICP-MS and found to be metrologically comparable and compatible. The residual bias was −1.8‰, and the precision obtained ranged from 1.9 to 6.2‰. This precision was sufficient to resolve naturally occurring variations, as demonstrated for samples ranging from approximately −3 to +15‰. To assess its suitability to technical applications, the NMC111 cathode candidate reference material was analyzed using high-resolution continuum source atomic absorption spectrometry with and without matrix purification. The results obtained were metrologically compatible with each other.