Garrett Gay

University of Rochester, Department of Earth and Environmental Sciences

1. Introduction

Garnets are a truly amazing mineral. Aside from being incredible examples of nature’s capacity for creating wonderful shapes, the garnet takes on a cubic or more interestingly a rhombic dodecahedral shape when it is euhedral. They are incredibly important to our experimental ability to understand our Earth.

Table 1: Garnets exist naturally on two solid solutions series: pyrope-almandine-spessartine and uvarovite-grossular-andradite. The above table shows these series and associated chemical compositions.[1]

Garnets are one of the geologists most valuable tools and tell us a lot about the geological history of an area. A number of methods have been developed to date garnets and thus, the rock units they reside in. Not only is this important to building a complete history of an area it also can tell us about the nature of the unit and its potential value for mining.

Three samples were selected to be analyzed for this project. The three samples were collected from, Hooper mine, Adirondack region, New York; Coast near Scourie, Scotland; Unknown locality in Western Scotland. Garnets vary widely in their composition as a function of the miscibility of their endmembers. Thus, leading to garnet compositions generally being expressed in terms of percent of pure endmembers. A possible example being: Pyr63Alm30Grs7.

A number of techniques were utilized to try to understand the differences in garnets caused by composition. Scattered electron imaging (SE2) was used to gain a better understanding of the surface and near surface of the crystal, backscatter detection (BSD) for basic compositional differences, interaction volume modeling to understand how compositional differences would affect electron flight, X-ray spectrums and mapping to resolve inter-crystal spatial compositional differences if any, and colorization to highlight interesting features in SEM images.

2. Sample Preparation

To prepare the specimens for the SEM, crystals were knocked off of the rocks and ensconced in epoxy. After the epoxy hardened the sample puck was sanded and then polished down. Afterwards, the sample had carbon tape and carbon paint applied to allow for charge dissipation. Finally, the sample was placed in the gold sputter coater and had 1.5um of gold applied to its surface.

Figure 1: The fully prepared sample, showing inlayed samples, carbon tape and gold coating. Gold coating was applied using the Denton Vacuum Sputter coater.

3. Scanning Electron Microscopy

SEM imaging is predicated on interactions that electrons experience when they come in contact with a surface. The first set of images collected were done using the SE2 lens. This lens detects electrons that have inelastically collided with the specimen matter and because they are generally emitted from the uppermost layer of the interaction volume, and reflect the topographical information of the specimen. The BSD was also used, which functions through the use of elastically scattered electrons. BSD also allows for preliminary compositional information to be gathered, as areas with a lighter color contain heavier elements. All images were collected on a Zeiss Auriga CrossBeam SEM-FIB.

All three garnet samples were imaged using the SE2 and BS detectors.

Figure 2-7: Comparison of SE2 images (left three) and BSD images (right three). The first set of images being from the unknown Scottish sample, the next two from the Hooper mine sample (the image itself known colloquially as the "disappointed Brian") and the last set from the Scourie shore sample.

The differences between SE2 and BSD are fairly clear in these images. The SE2 images reveal topographical information about the crystals, while the BSD shows some possible compositional differences. The missing chunks in the first set of images is most likely grains that were ripped from the crystal during sanding. It is interesting however to note that in BSD this appears as a compositional difference when it is very likely to be a function of the depth gradient. Overall, the BSD images show small differences, although there is some evidence of inter-crystalline grains. This is shown in the first set of images and the last. The last set of images also exhibits possible contamination which seems to be of heavier elements than the surrounding crystal.

Figure 8: A high magnification image taken of the the gap observed in the unknown Scottish sample. SEM data: 5kV, 10mm, 3.4Kx, SE2.

On first glance the garnets exhibit a few details that differentiate them, the removed grains feature was only observed in the Unknown Scottish sample and some kind of compositional differences were only seen in the previously mentioned sample and the Scourie shore sample.

4. X-Ray Maps and Spectra

X-ray maps and spectra allow for a more detailed look at compositional variations within a sample. While BSD allows for relative comparisons based on mass of an element, energy dispersive spectroscopy, identifies actual energy peaks and is able to assign them to a specific element. Here, this is used to investigate inter-crystalline and inter-granular relationships. The spectra give a compositional analysis while the maps show composition in space.

Figure 9: The unknown Scottish sample shows fairly predictable peaks.The main peaks at Si and O representing the constant building blocks for any silicate mineral and peaks at Ca, Fe and Mg being indicative of the garnet endmember compositions present, Al is present in all the pyralspite group and grossular garnets. Au is from the gold coating and C is from contamination during prep work.

Figures 10-14 & 3: The maps above show in order going from left to right and then starting on the right-hand side again: Al, Ca, Fe, Mg and Si (all K spectrum energies). For the most part the spectra taken shown in Figures 12-15 seem to have a detection gradient based off of the decreased X-rays from the deeper pits in the sample. These pits would lessen the number of X-rays that make it back to the detector. Figures 12-15 can be compared to Figure 3 to reaffirm this fact. It is interesting however that Figure 11 does not seem to follow this trend. Unfortunately, the possible grain differences seen in figure 3 cannot be resolved as a function of composition using these X-ray maps.

Figure 15: The Hooper mine sample shows a similar spectrum to that of the unknown Scottish sample. Here though we see increased Mg and Ca and decreased Al, alluding to an increased presence of the pyrope and grossular garnet endmembers.

Figures 16-19 & 5: In the same order as before: Al, Ca, Fe and Mg. The maps shown here exhibit no compositional differentiation. This makes sense as shown in figure 5, where the image seems to be all one color (indicative of no compositional differences) without any obvious grain boundaries.

Figure 20: Once again the Scourie shore spectrum shows similar peaks to the other two spectra collected. Here there seems to be a decrease in Mg for an increase in Ca.

Figures 21-23 & 7: Same order: Ca, Fe and Mg. Here we see two distinct sites. The smaller site is enriched in Ca and depeleted in Mg and Fe. The other larger areas show equal abundances of Mg, Fe and Ca. The Ca enriched area is most likely a purer grossular grain, as even though the pyralspite and ugrandite series are relatively miscible, some differentiation will still occur.

5. Interaction Modeling

Using a program called WinXRay (an electron flight simulator), the interaction volumes of each of the most common garnet sub-varieties and a unique synthesized garnet commonly known as YIG (Yttrium Iron Garnet), Y3Fe2(FeO4)3, was tested. Each simulation was run at 10kV, 0-degree tilt and 1E-7 amp current. The scale bars on each image represent 49nm in the Y-direction and 29nm in the X-direction.

Figure 24: Interaction volume of almandine.

Figure 25: Interaction volume for pyrope.

Figure 26: Interaction volume for grossular.

Figure 27: Interaction volume for spessartine.

Figure 28: Interaction volume for the YIG.

Pyrope, grossular, almandine and spessartine all exhibit relatively deep and laterally expansive electron paths. Unsurprisingly, the more dense and electronically potent YIG shows a more condensed set of electron paths.

6. Colorization

Adobe Photoshop was used to colorize an image of possible contamination on a the Scourie shore garnet.

Figure 29: This photo shows a colorized image of contamination on a garnet surface. SEM data: 5kV, 10mm, 808x, SE2.

7. Conclusions

The results showed that there were inter-crystalline compositional differences. These results were identified using BSD and then shown definitively through the use of X-ray maps and spectra. The most dramatic example can be seen with the images taken from the Scourie shore sample which shows what can be tentatively assumed to be a purer grossular grain in a more evenly mixed grossular-almandine-pyrope matrix. Overall, differences could be seen and this combination of techniques has proved itself to be a useful first step into more rigorous analysis.


I would like to thank Brian McIntyre for being a fantastic teacher and giving me a tangible skill. SEM work is not only useful but something I have found myself truly enjoying more and more. I would also like to thank my TA Caleb, who was a mentor and friend throughout this course.

Please enter any comments, criticisms, questions, etc. below.

Your name:

Email address:


[1]Dietrich, R. (2014, August 27). Garnet. Retrieved from