Large-Area Quantitative Phase Mapping
Materials and Methods Samples for analysis . All samples for analysis were mounted in a bakelite mount, polished with successively fi ner grit paper, followed by polishing with 9 µm, 3 µm, and then 1 µm diamond slurries. T e samples were coated with 100 nm of carbon to prevent charging. The first sample was a crushed rock from a mine. T e rock was pulverized and sieved through a mesh with 250 µm openings. T e goal of this analysis was to determine the quantity and distri- bution of the specifi c minerals within the rock and the locked and liberated fraction of each mineral aſt er pulver- izing. T e second sample was a polished section of a complex ceramic composite. T e end-user application required determination of the size, areal fraction, and distribution of the phases present. Instrumentation . The results presented here were obtained on a JEOL JSM-7001F field emission SEM. The electron microscope was operated at a 15 kV accelerating voltage and 12 nA probe current. The EDS data were collected with a pair of 30 mm 2 T ermo Scientifi c TM UltraDry EDS detectors mounted at a 35-degree take-off angle and 55 mm from the sample. T ese detectors were mounted 120 degrees relative to each other. The geometry was based on available high-angle ports. Collected X-ray events were channeled from each X-ray detector into independent pulse processors that discriminated
Figure 2 : Detailed analysis of a single grain extracted from a boxed region like that shown in Figure 1. (a) SEM image with brighter regions indicating higher atomic number in grayscale. Image width = 76 µm. (b) Threshold designations on the image histogram of (a) showing the numbers of pixels at each gray level. The four increasing brightness levels show four distinct phases: 1 (red), 2 (yellow), 3 (blue), and 4 (violet). (c) Color map of BSE intensities formed by thresh- olding the gray levels in (b). (d) Higher magnifi cation showing the four “phases” selected for EDS analysis.
legitimate X-ray events from background and determined the energy of each X-ray event. T ese digitized energy events were then collected by a “headless” computer along with their position coordinates from the electron beam scan controller to create a spectral imaging (SI) data set. T is SI data set was then transferred to the computer that operates the T ermo Scientifi c Pathfi nder soſt ware with which the user performs EDS X-ray acquisition and data analysis. Principal component analysis . The critical technology that enables large-area, quantitative phase mapping is principal component analysis (PCA). T e commercial implementation of PCA as provided by Thermo Fisher Scientific is called COMPASS TM . Principal component analysis is predicated on the assertion that a small number of dominant components can very nearly, although not perfectly, describe an arbitrarily complex system. An example of this is the classic red-green- blue color set that creates the full range of colors displayed in a digital image or video. While one may perceive thousands or millions of colors in an image, every pixel in that image is actually some combination of the original three-color
38
components. T e technical background to PCA is described in detail elsewhere as a general technique [ 6 ], and it applies to EDS spectral imaging analysis [ 4 , 5 ]. T ere are several benefi ts to this technique as applied to EDS phase mapping. T e fi rst is that very complex data sets can be described by a relatively small number of dominant components (that is, 3 to 20) that are more easily handled for interpretation. Moreover, because it is customary to identify each component, this method is eff ective at locating minority phases or phases that are diff erentiated only by the presence of a trace element. T e statistical nature of the algorithm prohibits user intervention, rendering the technique free of human bias. T e fi nal and possibly most important advantage to large-area phase mapping is that the dominant components can be extracted even when there are relatively few X-ray counts in the individual spectrum for each pixel. As a result, the dominant component maps are extracted with somewhere between 25 to 150 spectrum X-rays per pixel. Because the required X-ray collection is now reduced by a factor of roughly 50–100 times
www.microscopy-today.com • 2017 March
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72 |
Page 73 |
Page 74 |
Page 75 |
Page 76