Accuracy of a reference-free Monte Carlo approach for SEM-EDS thickness assessment of TiN coatings onto diverse substrates. (March 2022)

• Record Type: Journal Article
• Authors: Cruz, Juan P.N.; Garzón, Carlos M.; Noreña, Johan K.; Pineda, Fabian; Cachaya, Juan. S.; Recco, Abel A.C.
• Is Part Of: Micron. Volume 154(2022)
• Publication Date: 2022-03
• Subjects: Energy dispersive spectroscopy -- Scanning electron microscopy -- Coating thickness -- Monte Carlo simulation -- TiN nitride; Microscopy -- Periodicals; Electron Probe Microanalysis -- Periodicals; Microscopy -- Periodicals; Microscopie -- Périodiques; Microscopy; Periodicals; 502.82
• DOI: 10.1016/j.micron.2021.103204
• Languages: English
• ISSNs: 0968-4328

Abstract

Graphical abstract: Highlights: EDS experiments and Monte Carlo modeling allowed to assess the coating thickness. Thicknesses were assessed from a single EDS spectrum using no reference materials. Accuracy of thicknesses assessed is sensitive to the microscope operational conditions. For V0 higher than 20 kV, 90 % of predicted thicknesses showed accuracy better than 85 %. Abstract: Energy dispersive spectroscopy (EDS) experiments were carried out in the scanning electron microscope (SEM) on top of either stainless steel (SS), silicon or glass substrates covered with TiN nitride coatings. The nominal coating thicknesses were 0.43, 1.1, 2.1 and 3.0 μm. From the EDS spectra obtained, the ratio between the strongest peak coming from the substrate and the Tik peak, from the coating, was assessed at an increasing microscope accelerating voltage (V0 ) between 10 and 30 kV. This ratio is afterward called Intensity-ratio (IR ). A methodology was developed for assessing the coating thickness from the experimental IR and the V0 by using Monte Carlo simulations of the dispersion of electrons inside the covered samples and the concomitant emission of characteristic X-ray photons. Three freeware Monte Carlo graphical user interfaces were used, and the predicted thicknesses were compared between them, namely Casino, MC-Xray and DTSA. At fixed V0, numerical simulations predict a mono-valued and smooth relationship between IR and coating thickness. Monte Carlo predictions on the coating thickness were compared with reference thicknesses independently assessed by cross-sectional micrographs obtained in the SEM. Absolute value of percentage discrepancy between the predicted and the reference thicknesses was lower than 30 %, wherein the lower the V0, the less accurate the predicted thickness. In general, none of the three pieces of software compared outperformed the others, although predicted thicknesses significantly varied between them. For V0 higher than 20 kV, 90 % of predicted thicknesses deviated less than 15 % from the reference value. Predicted thicknesses for TiN coatings onto Si substrates presented the largest discrepancy regarding reference values.