Rocks are an agglomeration of various minerals or chemical compounds. The shape, type and distribution of minerals in rocks are determined by the formation mechanism and environment.
Gaining insights into the spatial distribution and chemical compositions of minerals in agglomerates is crucial to identify the mineral and textural development and history.
This article discusses the ability of Energy Dispersive Spectroscopy (EDS) to rapidly detect and determine sub-millimeter scale mineral phases in a sample.
The experiment used a thin section of contact metamorphosed Leadville Limestone. After sectioning and polishing the sample, it was placed in a glass slide for final polishing to transparency and analyzed in a FESEM without carbon coating by means of the Thermo Scientific™ NORAN™ System 7 (Figure 1) with seamless integration of Thermo Scientific™ UltraDry™ EDS detector (Figure 2).
Figure 1. NORAN System 7 X-ray microanalysis system
Figure 2. UltraDry silicon drift X-ray detector
The following table lists the analytical conditions:
||Contact Metamorphosed Calc-Silicate
||NORAN System 7
||UltraDry 10 mm2 silicon drift detector
|Magnification (for spectral imaging)
||450× and 4300×
||512 × 384
Secondary electrons were used to collect the images in the SEM. This was followed by performing EDS Spectral Imaging (SI) mapping acquisitions to acquire the distribution of all of the elements present in the map analysis region.
The SI technique acquires every X-ray from each pixel point in the map analysis region and stores them into a spectrum for every pixel.
By this way, a 3D histogram of X-ray intensities is generated for the dimensions (x, y, Energy). Spectra can be extracted from this 3D histogram for selected regions in the x-y image plane.
It is possible to extract elemental X-ray maps for any elemental lines along the energy axis. SI data acquisitions enable performing Multivariate Statistical Analyses (MSA) on the entire 3D histogram.
The 3D data set is used as an input by MSA to analyze all of the contained spectra in a self-consistent manner. The spectra at each pixel are compared with the spectra at every other pixel to determine similarities and differences.
Pixels having a statistically similar shape (elemental ratios) are grouped into a common map. Pixels are stored in separate maps in the case of spectra having statistically different shapes.
This results in a chain of spectrum-map pairs of the unique chemical materials in the analysis region. For these analyses, skilled analysts are not required to choose any data inputs, as elements of interest but the X-ray intensities within the 3D SI histogram are required to determine common spectral shapes.
Since each mineral has a characteristic spectral shape, it can be identified at its acquisition point in the analysis area. These map and spectral results help determining complete mineralogical phase distributions using the Thermo Scientific™ COMPASS™ software option for the NORAN System 7.
Spectral Imaging Acquisition
Figure 3 shows the secondary electron image of the sample area, revealing many different phases with interesting phase boundary contrasts. However, the secondary electron emission of the sample determines the intensity and contrast in the electron image, which are indirectly related to the composition of the region.
Two regions with similar secondary electron intensities can significantly differ in compositions. As a result, it may not be possible to differentiate all of the minerals present in the electron image from other neighboring minerals. However, EDS chemical analyses are capable of providing unique mineral distributions.
The cumulative spectrum from all of the scanned pixels in the SI acquisition is the main starting data result for most of the analyses. The peak identification of this spectrum reveals that many different elements are there to be analyzed beginning with the expected O, Mg, Al, Si, K, and Ca, and also the slightly less likely C (probably from beam contamination), S, Fe, and Ba (Figure 4).
It should be noted that the peak intensities in this spectrum rely on the composition of the minerals studied as well as on the area fraction of the mineral phases. Large area regions such as the mid-gray dominating the electron image will yield an over abundance of X-ray intensity, whereas a tiny mineral particle, whose contribution to the spectrum is only from a few pixels, may not contribute adequate elemental intensity in a peak to be detected.
Traditional Elemental Mapping
The "Extract Map Images" toolbar button is pressed to extract elemental X-ray count maps from the peak identification of this spectrum for display and analysis. The extracted elemental maps reveal contrast indicative of the presence of various minerals in the sample (Figure 5).
However, all of the intensity in these count maps is not solely contributed by the characteristic X-rays of the desired elements but background X-rays in the spectrum have also some contribution.
As a result, analyzing these maps without removing the background intensity could lead to erroneous interpretation. Quantitative elemental mapping is the technique applied to display the correct elemental distributions in maps.
Quantitative Elemental Mapping
Quantitative elemental mapping translates X-ray counts into meaningful composition values using the same procedures as quantitative spectral processing for composition measurements. Peak deconvolution (for overlapped peaks), background intensity removal, and matrix corrections for significant density or X-ray absorption edge effects are the required corrections.
Quantitative elemental mapping is carried out by choosing the option on the Processing tab in the lower left pane of NORAN System 7 software. Selecting the type of quantification output, i.e., Atomic %, is the first option. It is possible to automatically or manually set the kernel size to a value useful for the intensity of X-ray data in the 3D data histogram (here 1 x 1).
This is followed by defining the quality and precision balanced against computation time. "High" Detail and "Normal Precision" Fit Type are the preferred methods for most sample analyses. The "Extract Map Images" toolbar button is pressed to execute the quantification processing (Figure 6).
The true elemental distributions can be observed after applying these corrections to the maps on a pixel-by-pixel basis. Significant variations in the contrast of many elemental maps may be there for most samples. Slightly modifying the background intensities of the maps is the primary change for this sample.
Analysts may show interest on the distribution of the minerals in the sample instead of elements. If each mineral has a single unique element, it is possible to perform the identification and localization of the corresponding mineral. However, most elements present in more than mineral and so it is impossible to isolate unique minerals by single elements.
Manual Phase Mapping
Selecting and overlapping a set of three elemental maps to look for unique color formation is the simplistic method of mineral identification. However, each elemental map color has to be changed to one of the primary colors in the color spectrum (Red, Green, or Blue) (Figure 7).
In this case, the analyst has to choose three major elements from the list of nine potential elements, leading to 84 potential combinations (Figure 8).
Each unique color represents a potential unique mineral. There are unique colors for at least the red, green, and blue colors depending on these three elements. However, some of the blue and green regions may have the combinations of other minerals consisting of other elements, which cannot be incorporated into this simple model. Moreover, pixels with no color are also there. This means that other elemental combinations are needed to conclude the final solution.
Hence, high skill set and more time are required to select the best elemental maps for each combination owing to the possibility of performing many combinations. This technique may be useful for simple samples with only a few elements. It is recommended to use a better method for complex geological analyses with at least six elements.
Automated Phase Mapping
Spectrum acquisition at every pixel in the sample using Spectral Imaging offers a platform for a multivariate statistical analysis, which identifies the distribution of the chemical phases throughout the sample. This proven procedure is licensed to Thermo Fisher Scientific from Sandia National labs (Sandia, NM USA) and is sold as the COMPASS option for the NORAN System 7.
Pressing the "View Compass Data" toolbar button and then "Extract Map Images" toolbar button is the first step in the analysis, displaying the resultant map-spectrum pairs within a minute (Figure 9). A more visually appealing result is to transform these results into binary phase maps, which are then applied for extraction of spectra from the 3D SI histogram.
The use of the Match option of the NORAN System 7 software is the preliminary step to automatically assign a mineral. It is necessary to populate the Match user-defined database with a spectrum for each phase that is anticipated in the sample and enable the Match option for the phase analysis.
The spectra are included by pressing the Match Database "..." button on the Analysis Setup tab (Figure 10) and loading each spectrum in the Database Manager dialog (Figure 11). Creating a special database for the current project is generally helpful.
To enable the Match capabilities, it is necessary to select the Auto Match option on the Processing tab (Figure 12). The Auto button on the Phase toolbar is pressed to generate phase results. Phase maps will be generated that are binary colored and labeled with the title from the match spectrum that better fits the phase spectrum.
Automatic overlaying of all of the phase maps on to the electron image facilitates interpretation. The phase maps (Figure 13) acquired at 450× magnification shows six unique chemical phases: pyrite (FeS2), barite (BaSO4), diopside (CaMgSi2O6), phlogopite (K2Mg6 [Si6Al2O20](OH) 4), quartz (SiO2) and hematite (Fe2O3).
Figure 14 shows the individual mineral phase maps and corresponding spectra. Phlogopite and diopside form the calc-silicate matrix of the sample. The barite exists as an addition within a large phlogo-pite grain. The barite grain also consists of an inclusion of pyrite with what looks like a reaction texture between the pyrite and barite composed of hematite.
A detailed acquisition at a magnification of 4300x concentrating on the pyrite inclusion and barite interface (Figure 15) shows the reaction texture more clearly. For this dataset, the total acquisition and analysis time was below 4 minute.
Phase Mapping During Acquisition
It is possible to perform this method of phase mapping during data collection. The settings are adjusted as discussed above before beginning the SI acquisition, followed by the acquisition of the electron image and clicking of the Start button to acquire the SI data set.
The COMPASS and match routines are carried out during acquisition, displaying phase maps and spectra in place of the conventional elemental maps. This approach allows performing data analysis and interpretation during data collection, thus saving considerable time. This match routine took below 1min for data collection and clearly reveals the existence of these six phases.
The results clearly demonstrate the applicability of the NORAN System 7 microanalysis system equipped with UltraDry EDS detector and COMPASS and Match software for rapid determination of the distribution and compositions of mineral phases in a geological sample. The results are helpful to understand the mineralogical and textural development of this calc-silicate rock during contact metamorphism.
This information has been sourced, reviewed and adapted from materials provided by Thermo Scientific – Surface Analysis and Microanalysis.
For more information on this source, please visit Thermo Scientific – Surface Analysis and Microanalysis.