Dear colleagues,
i've done ROI x-ray mapping on powders containing TiO2 and Carbon, observed by SEM.
My goal is to have an idea of local composition of observed aggregates and to make statistics about composition/size.
For now i can't make automatical semi-quantitative x-ray mapping but I'm very interested in doing it.
What corrections can I apply to ROI images to "correct" each pixel grey level for each element images, in order to be able to correlate intensity detected for an element to the concentration (%w) ?
In advance, thanks for your answers.
Hello Enrique,
As a big picture, your issue is in strong relation with the multispectral/hyperspectral analysis problematic and the plethora of techniques associated with. What really differs here is the preprocesssing stage that has to be thought of as to specifically target SEM-EDS X-ray analysis.
To address your issue, I may suggest that you check out the following references:
- Friel et al., " X-ray Mapping in Electron-Beam Instruments ", 2006 - http://www.geology.wisc.edu/~johnf/g777/MM/Friel-xraymap.pdf
- Zhu et al., " Reducing Poisson noise and baseline drift in X-ray spectral images with bootstrap Poisson regression and robust nonparametric regression ", 2013 - https://arxiv.org/pdf/1302.0492.pdf
- K. Schladitz, "Quantitative micro-CT ", 2011 -
Article Quantitative micro-CT
- Ofner et al., " Image-Based Chemical Structure Determination ", 2017 - https://www.nature.com/articles/s41598-017-07041-x.pdf
- Wuhrer et al., " Characterisation of Minerals through X-Ray Mapping ", 2006 - http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.585.8770&rep=rep1&type=pdf
- Kotula et al., " Automated Analysis of SEM X-Ray Spectral Images: A Powerful New Microanalysis Tool ", 2003 - http://booksc.org/book/29352707/e8fdca
- [Connex paper of interest about image denoising] Zhao et al., " Detail-Preserving Image Denoising via Adaptive Clustering and Progressive PCA Thresholding ", 2017 - https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=8169017
Let me know if these references are helpful to you.
Regards
Hello Enrique,
As a big picture, your issue is in strong relation with the multispectral/hyperspectral analysis problematic and the plethora of techniques associated with. What really differs here is the preprocesssing stage that has to be thought of as to specifically target SEM-EDS X-ray analysis.
To address your issue, I may suggest that you check out the following references:
- Friel et al., " X-ray Mapping in Electron-Beam Instruments ", 2006 - http://www.geology.wisc.edu/~johnf/g777/MM/Friel-xraymap.pdf
- Zhu et al., " Reducing Poisson noise and baseline drift in X-ray spectral images with bootstrap Poisson regression and robust nonparametric regression ", 2013 - https://arxiv.org/pdf/1302.0492.pdf
- K. Schladitz, "Quantitative micro-CT ", 2011 -
Article Quantitative micro-CT
- Ofner et al., " Image-Based Chemical Structure Determination ", 2017 - https://www.nature.com/articles/s41598-017-07041-x.pdf
- Wuhrer et al., " Characterisation of Minerals through X-Ray Mapping ", 2006 - http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.585.8770&rep=rep1&type=pdf
- Kotula et al., " Automated Analysis of SEM X-Ray Spectral Images: A Powerful New Microanalysis Tool ", 2003 - http://booksc.org/book/29352707/e8fdca
- [Connex paper of interest about image denoising] Zhao et al., " Detail-Preserving Image Denoising via Adaptive Clustering and Progressive PCA Thresholding ", 2017 - https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=8169017
Let me know if these references are helpful to you.
Regards
In addition, here is a related issue about multispectral image analysis on ResearchGate that might be useful to you:
https://www.researchgate.net/post/How_to_do_subpixel_mapping
What is the size of your particles? If it is in a few microns range (or less), then the method you want to use is prone to artifacts and should not be used for quantification.
Hello Stéphane,
Thank you for your answer and all these references.
- Friel et al., " X-ray Mapping in Electron-Beam Instruments ", 2006 - http://www.geology.wisc.edu/~johnf/g777/MM/Friel-xraymap.pdf
=> was really helpful, especially the optimization part called X-ray map acquisition (spatial, magnification, max collected x-ray counts, detection limits) and Artifacts part.
- Wuhrer et al., " Characterisation of Minerals through X-Ray Mapping ", 2006 - http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.585.8770&rep=rep1&type=pdf
=> This paper was also useful to me because it compares informations provided by ROI mapping and Quantitative X-Ray mapping (QXRM).
I'll try to have more représentatives results : first by adjusting my image acquisition settings and then correcting all intensities collected for each element on each pixel by ZAF method.
Regards
Hello Vladimir,
my particles are in micron and submicron ranges (~ 0,3 - 10 µm). You are right, i realized that there should be artifacts using this method for my particles. On the other hand i found this paper about ZAF correction for semi-quantitative analysis of particles in a broad micrometer range (0,1-100 µm) :
Article A standardless ZAF correction for semi‐quantitative electron...
Please let me know what you think about this method,
Regards
Hello Enrique,
Thanks for your feedback. In view of your latest information, here are 2 additional references that may be useful:
- Trincavelli et al., " Standardless quantification methods in electron probe microanalysis ", 2014 - http://www.famaf.unc.edu.ar/~trincavelli/papers/SAB14Stdless.pdf
- Gauvin et al., " Win X-ray: A New Monte Carlo Program that Computes X-ray Spectra Obtained with a Scanning Electron Microscope ", 2006 -
Article Win X-ray: A New Monte Carlo Program that Computes X-ray Spe...
Regards
Dear Enrique,
The article you found is more than quarter century old. There were quite a few other papers on the subject, but with almost no of further developments. It looks like suggested methods work only for the authors (means they require deep knowledge of a subject and good command of author's software and its limitations). Manufacturers are eager to supply additional software for all kinds of measurements, at least as an option. If they are not giving it now, it means there are no reliable method yet.
And please, do not use standardless method. You task is already very difficult (to state it mildly), utilizing standardless analysis for C and O will make it impossible.
- Badges
- Science method