Imaging Applications

What is an image, and what is a map? Just for the sake of introducing the term map, the term image is a general term, whereas map is more specific. That is, a map is an image, but an image is not necessarily a map ... which is a small distinction, and most analysts generally refer to maps as images. However, to be more exact, many refer to maps as images of flat surfaces. For example, because all EPMA projects are with flat samples, we generally refer to spatial distributions of elements as elemental maps ... and while many will reference a backscatter electron image (BSE image), it is more appropriately an atomic number map, or a backscatter map. Because so many simply refer to maps as images, we will also do so here.

Elemental map of calcium in garnet
Types of images (or maps) The microprobe's detectors determine what types of images the instrument is capable of. For example, if the detector of choice is the backscattered electron (BSE) detector, then whatever the BSE detects will be associated with a location in the image. Modern microscopes (electron, EPMA, and optical) have all gone digital ... which is to mean, all images end up stored on a computer as a matrix of pixels. So when we refer to location in an image, or map, we generally refer to a pixel or group of pixels. For example, the figure below is the BSE signal taken from titanium oxide minerals set in magnetite ... and note that the BSE signal is really a grayscale value given to a pixel, somewhere in the image. The pixels are given their values while the BS detector acquires its signal, and while the computer associates the rastering beam location with the appropriate pixel location. The same acquisition process occurs for any selected detector, and many detectors can be selected at the same time.

Types of detectors

Imaging by detecting emitted electrons Many types of electrons are emitted from of the sample when it interacts with the incident beam. Emitted electrons are classified according to their energy.
 
Secondary electrons have very little energy (less than 50eV), and backscattered electrons have energies which approach that of the incident beam (thousands of eV). There are energies between, but by-in-large, the energy distribution for emitted electrons is strikingly bi-modal. Secondary electrons are useful for imaging surface topography, and are a subject for another webpage, scanning electron microscopy.
 
Secondary electron image of projector lamp filament
Backscattered (BS) electrons are useful in combination with EPMA because the information is compositional, ... i.e., a function of average atomic number. They can be used to show the distribution of different minerals in a given sample area, and to illustrate compositional zoning of single crystals of a mineral. For example, differences in Ca(At#=20) and Na (At#=11) concentrations, between the core and the rim of plagioclase grains that crystallized from melt, contribute to a significant difference in the average atomic number across individual grains with Ca-rich areas giving a brighter image. Magnesium and iron relationships are just as striking, and with care, an analyst can present different pixel values between olivines as close as FO79 and FO80.
 
BSE images are most useful because they can be acquired in very little time (~1 minute), but they can also be misleading if the analyst is not careful. For example, if you were informed the average atomic number of a mineral was 19, would you then be able to identify the mineral or determine its composition?
 
The answer would be "No" because many different elements could contribute to the same atomic average. Given the variety of mineral compositions and how elements contribute to the average atomic number, garnets can be indistinguishable from clinopyroxenes, Na-rich feldspar from quartz ... and confirmation needs to come from another detector (see below). Still, once confirmed, feel free to acquire many BSE images.
 
Electron Microprobe
The images on the left and below compare optical microscopy with backscattered electrons. The specimen provides an example of metamorphic textures in a high-pressure metagabbro. Direct comparisons can made by rolling your mouse over the images.
 
Immediately apparent is the color offered by optical, and crossed pols further adds information important to mineral identification. However, what should be equally apparent, is the visual confusion caused by the size of many of the mineral grains being on the same scale as the thickness of the thinsection. Contrast this confusion with the BSE image, which simplifies the assemblage of minerals allowing their relationships to become readily apparent.
 
 
At this scale, BSE imaging with the scanning electron microscope is the logical extension of the optical microscope. Chemical micro-analysis is also an extension, however notice many characteristics of composition are already apparent, and there is no need to map the composition.
 
Brief acquisition of x-ray spectra quickly identifies the dark-gray blades of spinel (Sp) and light-gray garnet (Gt). A slightly longer acquisition, and the presence of potassium identifies the darkest gray as amphibole (Am), which also separates it from the clinopyroxene, which is apparently zoned relative to Ca (lighter) versus Na (darker).


Absorbed current Before I leave BS images, I should mention backscattered electrons can be considered simplistically to be incident electrons which have "bounced" out of the sample, and the remainder of the incident electrons having been absorbed by the sample. These absorbed electrons can also be detected and collected as pixel values ... which, as you might expect, would ultimately become an image related to the BS image but opposite (i.e., dark where the BS image was bright). BS images are easier to acquire and have better spatial resolution, but keep absorbed current detection in your imaging toolbox.

Characteristic x-rays Most useful is to actually visualize the spatial distribution of the elements of interest. The map at the top of this page presents to the viewer how calcium is distributed in a particular metamorphic garnet. The compositional map on the left is a similar variation, however the elemental content is more interestingly pseudo-colored ... with the reference on the right indicating which colors present nil versus lots of calcium.
 
Ca-zoning in garnet ( ~1cm in diameter )
Compositional, or elemental, maps are time consuming, especially compared to imaging with backscattered electrons. A medium resolution map (e.g., 512x512 pixels) can take 8-12 hours to acquire. It's all dependent on how much of the element of interest is present, but for each pixel a small amount of time is required to count x-rays. Even if the dwell time is limited to a tenth of a second, be prepared for an 8 hour wait. These times can be reduced if EDX detection of the x-rays are utilized ... and many more elements can be simultaneously acquired. However, EDX background noise is a significant contribution, and elements at minor concentrations are difficult to distinguish and map with clarity.

Image Analysis

Image analysis involves the post-acquisition evaluation and processing of a wide variety of information from a digital image. Image processing allows for both field-specific measurements (e.g. number of objects, area percent) and feature-specific measurements (e.g. size, shape, grey level of individual objects). As stated previously, EPMA acquired images should be considered to provide qualitative information. That is, the value of each pixel should not be considered to be anything other than some relative amount. However, and as the following examples should imply, pixels and their neighboring pixels are informative, and therefore can be classified, analyzed, counted, and otherwise yield quantitative spatial information.
 
An example of a field-specific analysis would be modal analysis. For example, consider the question how much of the field-of-view is olivine relative to pyroxene? Elemental maps are required for the analysis, and for these 2 minerals, we'd need acquire only Si, Al, Mg (or Fe). That is, the analysis requires each pixel to represent a silicate, and the presence of aluminum to separate PX from OL, and the presence (or relative amount) of Mg,Fe to separate OL & PX from other minerals (e.g., feldspar). Other suspected minerals would dictate which additional elements are needed. For example, if amphibole is present a map of potassium would be useful. Of course, a complete modal analysis would also quantify the other areas of the image with respect to the other minerals present (see the The concepts of modal analysis with digital images case study).
 
An example of feature-specific analysis would be to analyze the shape of many minerals of the same composition. For example, consider a BS image of a sandstone that presents all quartz grains as a distinct gray and other minerals, including the cement, as different grays. Consider the questions ... How round are the quartz grains? ... Is there any elongation? ... Is the elongation oriented to a specific direction? Image analysis softwares, intended for quantifying, are all capable of providing quantitative values for roundness, elongation (implying an axis), axis rotations, and a number of other features. Therefore image analysis softwares may not provide you with an image at all, but rather interface with a spreadsheet for creating graphs (e.g., number of grains, a function of grain size) and other informational processing for ultimately deriving genetic conclusions regarding the original image.
 
Image analysis has a wide variety of applications in earth, biological and materials sciences and engineering. The microprobe facility at Memorial has been contracted in the past by the Iron Ore Company of Canada for better understanding its milling process. A quantitative mineral liberation study of the textural variability of the iron ores provided insight into the milling process and led to a better understanding and methods of enhancing ore recovery.
 
Other Applications: The microprobe has a wide range of applications in many disciplines. Recent applications outside the earth sciences in our laboratory include analysis of metallurgical materials, and studies of the corrosion products of archeological artifacts.