Quantification of Synthetic Nonmetallic Inclusion Multiphase Mixtures from a CaO–Al2O3–MgO–CaS System Using Raman Spectroscopy. Issue 1 (24th July 2020)

• Record Type: Journal Article
• Authors: Gyakwaa, Francis; Aula, Matti; Alatarvas, Tuomas; Shu, Qifeng; Huttula, Marko; Fabritius, Timo
• Is Part Of: Steel research international. Volume 92:Issue 1(2021)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2020-07-24
• Subjects: calcium aluminate -- CaS -- MgO·Al2O3 spinel -- nonmetallic inclusions -- Raman spectroscopy; Steel -- Periodicals; Steel -- Metallurgy -- Periodicals; 669.142
• DOI: 10.1002/srin.202000322
• Languages: English
• ISSNs: 1611-3683

Abstract

Abstract : Raman spectroscopy has features such as its relatively easy sample preparation. Therefore, its application plays a role in nonmetallic inclusion studies. Herein, use of Raman spectroscopy to quantify multiphase synthetic inclusion mixtures consisting of C12A7, CA, C3A, CaS, and MgO.Al2 O3 is assessed. Partial least squares (PLS) regression is used to obtain a calibration model, and enhancement of the model performance is done using standard normal variate (SNV). From the calibration model, the root mean standard error (RMSE) in cross‐validation (RMSECV) values range between 3.82 and 6.13 wt%, the root mean square error of prediction (RMSEP) is within 2.0–4.04 wt%, and the R 2 value is between 0.93 and 0.99 for estimating the phases based on SNV Raman data. The raw Raman spectra data have RMSECV values between 9.83 and 16.46 wt%, the RMSEP ranges between 9.41 and 15.33 wt%, and R 2 is estimated within 0.71–0.95. The PLS regression has a satisfactory prediction performance with a high range error ratio (RER) and ratio of prediction‐to‐deviation (RPD) values for SNV Raman data. Herein, the use of Raman spectroscopy and a calibration model to quantify the specific phase in a multiphase synthetic inclusion mixture in CaO–Al2 O3 –MgO–CaS system are demonstrated. Abstract : Herein, the application of Raman spectroscopy for quantifying mixtures of the CaO–Al2 O3 –MgO–CaS system is illustrated. Partial least squares regression model and the enhanced feature of standard normal variate are used for estimating the phase composition present in the sample mixtures.