**4. Discussion**

The five examples of application above show a small set of the vast range of possibilities for modern AQM systems, like Mineralogic, where visualization of the data and the generation of chemical, mineralogical and morphological data are combined. Traditionally, mineral compositions in metamorphic rocks are quantified with optical microscopy or microprobe analyses, while metamorphic textures are investigated with optical microscopy and often unquantified or, mainly in case of deformed samples, quantified by electron back-scatted diffraction. By applying AQM, visualization and quantification can be combined in one tool and executed on the same minerals. The AQM software is creating a false-coloured mineral map, while simultaneously measuring grain morphology and chemistry. The false-coloured map resembles the optical microscope mineral display of metamorphic

textures, but also forms the basis to gain quantitative data for the sample. Compared to the microprobe, AQM o ffers mineral maps and not only element maps. This is combined with the higher analytical speed for EDX analyses compared to the wavelength dispersive spectrometry analyses on the microprobe. The diversity and flexibility in the chemical, textural and mineralogic visualization is demonstrated in this paper as a means of contextualization of what can initially be complex data sets.

AQM is often regarded as a slow analytical tool. However, depending on the information required, analytical speed can be adapted to the optimal combination of mineralochemical precision, map-quality and time. An area of 1.5 × 2 cm can be analysed in 83 min at a 40 μm step size, and provide almost identical mineral concentrations, when compared to runs with smaller step sizes and longer analysis times (Table 3; Figures 3 and 5). A time-saving approach could be to first run a large (here 20 μm) step size line-scan or Mineral map (here 40 μm step size) on the selected sample, then build a solid mineral list and run selected area(s) with a smaller step size to obtain detailed information on specific features e.g., reaction rims. Further, these investigations indirectly showed that the general reproducibility of the data, even while generated at di fferent step-sizes (pixel sizes), is good.

Within the Mineralogic software platform, it is also possible to analyze samples with di fferent acceleration voltages, without changing the mineral list. Samples 521106 and 521111 originate from the same area, were analyzed at 20 and 15 kV respectively, and were recalculated with the same mineral list without any adaptions (Table 1).

The garne<sup>t</sup> and ruby examples outline unique capabilities the Mineralogic software is able to access when compared to other AQM platforms. Because Mineralogic's EDX software is matrix-correcting and quantifying every spectrum for every pixel in the mineral maps and for each spot or feature scan for every epoxy-embedded particle, the exact chemistry is known with the same precision as for EDX point analyses. Elements can be easily detected with concentrations down to ca. 0.5 wt%, and major and minor elements can be analysed with an accuracy of 1–2 wt%. This is a large improvement compared to spectrum matching AQM platforms, which are not able to detect elements in concentrations of less than 5% [22]. Also, by obtaining the chemistry for every pixel makes the addition of new minerals to the mineral library more certain, and easily adaptable to changes in composition due to slightly di fferent metamorphic conditions (e.g., Mg/Fe ratio dependence on pressure and especially temperature during metamorphism). The ability to measure detailed chemistry of the minerals for every pixel of the mineral maps has been applied to make element concentration maps (Figure 7c–e) and element ratio maps (Figure 10c). But the chemical data can also be exported after analysis, e.g., to obtain a bulk geochemical assay or to obtain raw data for a garne<sup>t</sup> XMg–XFeMn–XCa ternary diagram (Figure 8). A comparison with whole rock geochemical data and EDS generated data with Mineralogic on a homogeneous sample (Table 2) shows a good agreemen<sup>t</sup> in the results between the two methods. The tested sample had slightly higher Ca and lower Si in the analysed thin section than the bulk rock sample and comparable results for the other elements., The discrepancy in Ca and Si probably occurred due to the presence of a calcite vein in the thin section.

The garne<sup>t</sup> example shows how the zonation into core-inner rim-outer rim could be observed by the garne<sup>t</sup> elements Fe, Ca, and Mg. These three zones in garne<sup>t</sup> were each associated with their own set of inclusion minerals, visualized in the mineral map of the sample (Figures 6 and 7a), revealing three stages of the metamorphic history of the garnet. Similarly, the Si/Al element ratio map show how silica-concentration is increasing in the reaction rims around corundum in the sample, causing the formation of sapphirine, cordierite and anorthite, each with a higher Si/Al ratio. Sapphirine, cordierite and anorthite are clearly associated with the ruby, as is visible in the mineral association map for these three minerals (Figures 9 and 10). Mineral mapping of the reaction rim showed that these are a prograde/peak metamorphic feature and not part of the retrograde reaction path [41].

The chromite-bearing gabbro example and the biotite example show the power of morphochemical classifications of the minerals (Figure 11). The ability to precisely select certain minerals and to cluster and colour-code them by morphological features is a very powerful feature in AQM software products. For the chromite-bearing leucogabbro sample it was demonstrated that the Mineralogic software can evaluate and colour-code the grain size distribution of grains in a rock. Image processing tools help to automatically separate near-touching grains (though not the entirely clustered grains). The strong preference of chromite for the hornblende layers is quantified by the mineral association data that AQM software is able to produce.

The biotite sample's Feret analysis is an example of a monomineralic display used to illustrate a morphological criterion. Here, the most important mineral is mapped out only, and colour-coded by its mineral orientation (Figure 12). However, the results are not only displayed, but the data for the analysed particles from the mineral map formed the basis for association data, Feret angle orientation and Feret elongation, grain size (area) as well as modal mineralogy (Figure 13 and Table 4), showing how intensive isoclinal foliation causes mineral layering in the formation of a schistosity.

Even though most scientific investigations cannot exist without the availability of numerical data and the quantification of processes. However, the visualization of results and qualitative observations still remain very important in order to fully understand the data, and often this visualization is the key to gaining new insights through conceptualization of the data.

Here, we show that AQM software in general and especially Mineralogic software is able to generate many di fferent ways of displaying the same sample and thereby highlighting several aspects of same sample. In the examples above we apply BSE and CL images, mineral maps, element concentration maps, element ratio maps, single mineral maps, mineral association maps on paired mineral groups, and morphology maps highlighting grain shape and orientation. Still these range of maps are only a small selection of the range ways the same sample can be displayed. As most humans are visually oriented, this displaying of results makes it easier for us to gain new insights. Visualization of the data adds context and meaning to the data tables. For example: the zonation of the garnets would not be visible without the Fe, Ca, and Mg concentration maps (Figure 7) and the orientation of biotite with respect to the local and main orientation of the foliation was more easily visible with colour coded oriented minerals (Figure 12).

The flexibility of having platform independent, acceleration voltage independent mineral classification, precise chemical measurements allowing for element concentrations and element ratio maps supplement and support the quantified data analysis, and interpretations, and on these topics ZEISS Mineralogic has proven its worth for metamorphic texture investigations, as well as a large number of other areas of investigation.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2075-163X/10/1/47/s1, Supplementary 1: Outcrop Images and Mineral Composition; Supplementary 2: Mineralogical Method (Figure S1: Overview of the SEM recipe (left) and the image navigation tool (right); Figure S2: Holder recipe; Figure S3: Calibration recipe; Figure S4: Image processing recipe; Figure S5: Morphology recipe; Figure S6: Mineral recipe set-up; Figure S7: Mining recipe); Table S1: EDX point analyses on garne<sup>t</sup> standard.

**Author Contributions:** Conceptualization, N.K., S.G. and S.N.M.; methodology, N.K., S.G. and S.N.M.; validation, N.K., and S.N.M.; formal analysis, N.K., S.N.M.; investigation, N.K., S.N.M.; resources, N.K.; writing—original draft preparation, N.K. and S.N.M.; writing—review and editing, N.K., S.G. and S.N.M.; visualization, N.K. and S.N.M.; supervision, N.K.; project administration, N.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** The authors wish to thank R. Kahn and S. Lode for discussion and suggestions on the use of the Mineralogic software. J.C. Schumacher is thanked for his work on the metamorphic petrology in the Fiskenæsset area. Two anonymous reviewers and the editor provided valuable comments to an earlier version of this publication.

**Conflicts of Interest:** The authors declare no conflict of interest.
