grokkingstuff Home Blog Projects Wiki Calculators About

Energy Dispersive Spectroscopy

tags:meche:

Select specimen elements from the Pre-defined Elements tab. The display then shows peak labels or X-ray maps for those elements only.

#+DOWNLOADED: file:///tmp/Spectacle.cSNtjD/Screenshot_20250725_225343.png @ 2025-07-25 22:56:24 [image: 2025-07-25_22-56-24_Screenshot_20250725_225343.png]

Scan image

[image: EDS_AZTEC_ScanImage.png]
FIG. 01

Scanning an image lets AZtec take over the SEM to acquire it.

[image: EDS_AZTEC_ScanImage_AcquisitionToolbar.png]
FIG. 02

The image mode is set to BSE. You should see the microstructure appear on your screen.

Change the acquisition parameters using the Settings cog:

Image scan size sets the pixel count per axis. Higher values produce better quality but take longer. Options are 64, 128, 512, 1024, 2048, 4096, or 8192. Our SEM images are about 700 pixels wide, so I recommend 1024 unless you have a higher-quality picture.

The beam acquires the image through raster scanning. It dwells on each pixel before moving on. Higher dwell times produce better quality but take longer. Options are 1 us, 5 us, 10 us, 35 us, 65 us, or 400 us.

This selects signals from the microscope detector. Use BSE mode since we examine flat polished metal samples and care about phases.

Other options exist but we do not cover them here.

Influence of Image Scan Size and Dwell Time on Image acquisition time

Image Scan SizeDwell Time (us)Time taken (seconds)
102411.048576
102455.24288
10241010.48576
10243536.70016
10246568.15744
204814.194304
2048520.97152
20481041.94304
204835146.80064
204865272.62976

You should also see the image appear in your data tree: [image: EDS_AZTEC_datatree.png]

Acquire Spectra

[image: EDS_AZTEC_AcquireSpectra.png]
FIG. 03

This step acquires the EDS spectra from the current electron image.

Acquiring EDS spectra requires starting with maps. Maps are easier to understand and often needed for publication. Industry professionals focus on point and line scans because they are faster and provide the data needed for real analysis.

[image: EDS_AZTEC_AcquireSpectra_AcquisitionToolbar.png]
FIG. 04

Adjust the acquisition parameters using the Settings cog:

Use 20 keV as the standard energy range for spectra.

Number of Channels

The number of channels is the number of bins in the EDS spectrum. AZtec offers 1024, 2048, or 4096 channels in a drop-down list.

Accelerating VoltageNumber of channelsEnergy resolution
2000040964.9
2000020489.8
20000102419.5

Energy resolutions of 10 eV are acceptable. Too many channels hurt performance. Total X-ray counts distribute among all channels. More channels spread these counts more thinly. Each channel receives fewer counts, though the overall peak may look smoother. Low counts per channel make statistical analysis harder. You only need to distinguish peaks, so fewer channels perform better, especially in quantitative analysis.

I use as many channels as possible for point scans. For map scans, reducing the channel count to 2048 is worthwhile.

Total counts in spectrum

The displayed default controls spectrum acquisition before the beam calibration step. Enter your preferred value.

The default total count is 600,000, from a cobalt spectrum acquired at 20 kV. It gives approximately 1 percent precision in the beam current value over plus or minus 0.5 percent.

Process Time

Select the process time from the drop-down list of values 1 to 6.

Process time reduces noise in the X-ray signal from the EDS detector. Longer process times produce lower noise and narrower peaks. Lower noise improves peak resolution and separates peaks from nearby signals.

Selecting Default lets the software set the process time automatically. Process time trades off against data acquisition speed.

EDS Checklist

Core performance

EDS analysis quality depends on spectral resolution and total count. Practical constraints deliver robust results faster.

Spectral resolution measures the ability to separate overlapping characteristic peaks. Peaks from sulfur K-alpha, molybdenum L-alpha, and lead M-alpha overlap. Characteristic X-rays have discrete energies. Detection broadens these energies into Gaussian-shaped peaks. Broadened peaks can hide other peaks. Spectral resolution uses the full width at half maximum (FWHM) in electron-volts (eV) to measure peak broadening. Lower values indicate better resolution. Longer process times reduce noise and produce narrower peaks with lower FWHM values. This improvement costs time.

Total counts is the sum of all X-ray events across all energy channels. This sum measures statistical quality. Higher counts produce a better signal-to-noise ratio. This ratio makes peak identification and analysis easier.

The SDD anode collects charge. That charge feeds into a pulse processor. The processor converts each X-ray charge packet into a measurable voltage step on a rising ramp. Each step represents one X-ray detection. The step size depends on the X-ray energy.

[image: eds_pulse_processor.jpg]
FIG. 05 - Typical output voltage ramp showing events induced by Mn K-alpha X-rays. Taken from Oxford Instruments blog post by Haithem Mansour, at https://www.oxinst.com/blogs/what-dead-time-you-should-choose-for-your-eds/-edx-analysis

The pulse processor measures each voltage step, removes noise, and averages the signal before measuring energy. X-ray energies range from 50 eV to 40 keV. The averaging duration is the process time (Tp). Longer Tp reduces noise and improves peak resolution but increases acquisition time.

The output count rate shows how many X-ray counts the processor handles per second. The input count rate shows how many X-ray counts hit the detector. Maximizing throughput collects statistically robust data in shorter times.

[image: eds_pulse_processor_averaging_voltage_step.jpg]
FIG. 06 - Measurement of steps on a voltage ramp with short Tp and long Tp. Taken from Oxford Instruments blog post by Haithem Mansour, at https://www.oxinst.com/blogs/what-dead-time-you-should-choose-for-your-eds/-edx-analysis

The processor handles only a limited number of X-rays per second, which depends on Tp. Longer Tp means fewer processed X-rays per second and lower throughput. The incoming X-ray count rate stays the same.

Core Performance Metrics: Spectral Resolution (FWHM), Dead Time, and Throughput

EDS analysis quality depends on spectral resolution and total count. Practical constraints deliver robust results faster.

Electronic settings control spectral resolution and total counts. The main setting is the process time.

Throughput (output count rate) is the rate at which X-ray counts are successfully processed and added to the spectrum. Maximizing throughput collects statistically robust data in shorter times.

Dead time is a misnomer. It is a ratio, not a time, of the output count rate to the input count rate. Some events may not be measured when the process time is short.

A longer process time improves resolution and increases dead time. The processor spends more time on each event. It also filters more electronic noise. This change reduces overall acquisition speed.

Dead time is the percentage of total acquisition time that the processor spends measuring an X-ray event. The processor cannot process any other incoming photons during that time.

Optimizing acquisition parameters usually aims to maximize throughput: collecting statistically robust data in the shortest time.

Operating procedures recommend adjusting the beam current to achieve a dead time between 30 percent and 70 percent. Operating below 30 percent means a short process time. This change gives poor spectral resolution and causes peak overlap. Operating above 70 percent means the system rejects a large fraction of incoming X-rays inefficiently. high count rates and dead times also worsen sum-peak artifacts. Correction algorithms become overwhelmed at those levels. The 30 to 70 percent range balances acquisition efficiency, data quality, and artifact management for most routine analyses.

Deconstructing the EDS Spectrum: Peak Identification and Spectral Artifacts

An acquired EDS spectrum contains genuine characteristic peaks and predictable spectral artifacts.

Peak identification is the main task in qualitative analysis. Match the energy positions of characteristic peaks to known X-ray emission lines. This match identifies the elements present. Modern software such as AZtec automates this process. The software overlays theoretical line markers (e.g., Fe K-alpha, Cr K-alpha) on the spectrum to confirm element presence.

A sum peak (or pulse pile-up peak) occurs when two X-ray photons strike the detector simultaneously. This artifact is more likely at high X-ray count rates and dead times above about 60 percent. The pulse processor cannot distinguish the sum of two X-ray photons from a single X-ray with energy equal to their sum. It registers the sum as a single event. For a calcium sample, the Ca K-alpha peak at 3.8 keV may create a sum peak at 7.6 keV.

[image: EDS_spectral_artifacts_sum_peaks.png]
FIG. 07 - The energies of two incoming X-rays add together and a sum peak records in the spectrum. From Myscope at https://myscope.training/EDS_EDS_spectral_artefacts

Silicon escape peaks occur when an incoming X-ray excites a silicon atom in the detector. The detector emits a Si K-alpha X-ray at 1.74 keV. If this photon escapes, the energy is lost and the detector records the original event at its true energy minus 1.74 keV. A Ca K-alpha peak at 3.8 keV produces an escape peak at 2.1 keV.

[image: EDS_spectral_artifacts_escape_peaks.png]
FIG. 08 - The energy of the incoming X-ray reduces by 1.7 keV, the Si K-alpha energy, and an escape peak records in the spectrum. From Myscope at https://myscope.training/EDS_EDS_spectral_artefacts

Modern EDS software detects and labels these artifacts automatically. Pulse pile-up correction algorithms subtract sum-peak counts and reassign them to the parent peaks. This process ensures accuracy at high count rates.

Best Practices for High-Fidelity Data Acquisition

Acquiring reliable EDS data requires optimized instrument parameters and careful sample preparation. The quality of the final quantitative result or elemental map depends on choices made before you start acquisition. No post-processing can rescue data acquired under suboptimal conditions.

2.1. Optimizing Critical Instrument Parameters: Accelerating Voltage, Probe Current, and Working Distance

Three SEM parameters control EDS data quality directly: accelerating voltage, probe current, and working distance.

Accelerating Voltage (kV)

The accelerating voltage sets the kinetic energy of electrons striking the sample. Incident electron energy must exceed the critical excitation energy of an X-ray line to generate that X-ray efficiently. Set the accelerating voltage to 1.5 to 2.5 times the highest-energy X-ray line of interest. This ratio is the overvoltage. To analyze copper (Cu K-alpha at 8.04 keV), use at least 15 kV and preferably 20 kV. Low voltage produces weak or no excitation of higher-energy lines.

Higher accelerating voltage increases the interaction volume size and depth. This change degrades spatial resolution. For sub-micron features, use 5 to 10 kV to confine the interaction volume. These settings sacrifice excitation of high-energy lines. The kV choice balances excitation efficiency against spatial resolution. For general analysis of an unknown sample, 15 to 20 kV is a common starting point.

Probe Current

Probe current is the total current in the focused electron beam hitting the sample. You control the X-ray generation rate with probe current. Higher probe current delivers more electrons per second. This change generates more X-rays and increases the detector count rate. Adjust the probe current during setup to reach the desired dead time (typically 30 to 70 percent) on the EDS software ratemeter. Select a larger spot size or aperture in the SEM column to increase probe current. Higher probe current degrades SEM image resolution. You often trade optimal imaging conditions for optimal EDS conditions.

Working Distance (WD) and Detector Geometry

The working distance is the distance from the final SEM lens to the sample surface. EDS requires a short and consistent working distance, typically 7 to 10 mm. The detector sits at a fixed location relative to the sample stage. A shorter WD places the sample closer to the detector. This change increases the solid angle of X-ray collection. A larger solid angle captures a higher fraction of emitted X-rays. The change produces a higher count rate for a given beam condition.

The detector elevation angle (above horizontal) and azimuthal angle (around the sample) are also critical, especially for absorption correction. The sample should sit at 0 degrees tilt. This setting ensures a consistent take-off angle for X-rays toward the detector and minimizes shadowing artifacts.

2.2. The Crucial Role of Sample Preparation: Polishing, Cleaning, and Conductive Coating

Sample preparation quality determines quantitative EDS accuracy. Preparation errors are fundamental and cannot be corrected by software.

Surface Topography

Quantitative analysis requires a flat, mirror-polished surface. Roughness, scratches, or porosity create significant artifacts. Parts of the sample can shadow from the detector. X-rays generated beneath a rough surface travel through varying path lengths to escape. Low-energy X-rays absorb more easily than high-energy X-rays. This effect skews the spectrum to under-represent light elements and over-represent heavy elements.

Conductive Coating

Ceramics, polymers, and geological samples are often electrically non-conductive. The electron beam charges the surface with negative charge. This charge deflects the incoming beam. The charge also distorts the image and prevents stable analysis. Non-conducting specimens require a thin, uniform conductive coating.

Gold or gold-palladium is common for secondary electron imaging due to its high yield. For microanalysis, carbon is the universally preferred coating. Its low atomic number minimizes X-ray absorption. Carbon produces only a single, low-energy C K-alpha peak that is easy to account for. The coating should be as thin as practicable, typically 5 to 30 nm.

Coating Correction

The EDS software must know the coating material and thickness. The software applies corrections for two effects. First, the slight energy loss of the primary electron beam as it passes through the coating layer. Second, the absorption of emerging sample X-rays as they pass back through the coating. Uncorrected coating effects cause significant errors in quantitative results, especially for low-energy X-ray lines.

2.3. Statistical Foundations for Reliable Analysis: Achieving Sufficient X-ray Counts

EDS is a photon-counting technique. Reliability depends on counting statistics. Acquiring sufficient X-ray counts is essential for distinguishing elemental peaks from background noise and for valid quantitative results.

Signal-to-Noise and Poisson Statistics

X-ray detection follows Poisson statistics. The standard deviation of a measurement equals the square root of the number of counts (N). The relative error is 1 over the square root of N. To halve the relative error, you must quadruple the number of counts. This principle shows why you need high counts, especially for minor and trace elements with low peak-to-background ratios.

Practical Count Recommendations

The required counts depend entirely on the analytical objective. A qualitative map needs far fewer counts than a precise quantitative analysis. The following guidelines synthesize general recommendations into a practical reference.

Table 2.1: Recommended X-ray Counts for Reliable EDS Analysis

Analysis Type | Metric | Minimum Counts | Rationale and Application ---------------------------+-----------------------+-----------------+------------------------------------------------------------ Point Spectrum (Quant.) | Total counts | 10,000-50,000+ | Ensures sufficient counts in minor/trace element peaks. Higher totals (e.g., 500,000) needed for high-precision analysis. Elemental Mapping (Qual.) | Counts Per Pixel (cpp)| 200 cpp | Sufficient to distinguish elemental distributions from background noise and generate a clear map. Quantitative Mapping | Counts Per Pixel (cpp)| 500 cpp | Required for per-pixel matrix correction algorithms to operate on statistically meaningful data and produce reliable quantification. Statistical Analyses | Counts Per Pixel (cpp)| 50 cpp | A lower threshold works for multivariate statistical methods (e.g., PCA). These methods analyze the entire spectrum shape and correlations. They do not analyze just individual peak heights.

This framework links analytical goals to acquisition time. High-quality quantitative maps require significantly more beam time than simple qualitative mapping. Decide on the required data quality before starting acquisition. Then allocate the appropriate time.

2.4. Managing Detector Electronics: Pulse Pile-Up Correction and Acquisition Modes

The final setup steps configure the software settings that control data collection and real-time processing.

Pulse Pile-Up Correction

This software feature mathematically corrects for sum-peak artifacts. Enable it for any analysis where compositional accuracy matters. It works best at single points or homogeneous regions.

Acquisition Termination Modes

EDS software offers several modes to define acquisition stop criteria:

Live Time: The acquisition runs for a preset active processing time. The software extends the real clock time to compensate for dead time. An acquisition set for 100 live seconds with 50 percent dead time takes 200 real seconds. This is the standard mode for quantitative analysis. This mode ensures different measurements receive the same effective X-ray dose. Their intensities are directly comparable.

Real Time: The acquisition runs for a fixed wall-clock duration. This mode is less reproducible for quantitative work. Varying dead time between measurements causes different total counts.

Counts: The acquisition stops once the system collects a user-defined total count in the spectrum. This approach ensures a specific level of statistical quality for every measurement.

Auto Mode: The software monitors the spectrum and stops when it collects enough counts for reliable standardless quantification. This option works well for routine analysis.

The parameter choices for an EDS experiment are not about finding a single set of correct numbers. You navigate a fundamental trade-off triangle between Speed (acquisition time), Spatial Resolution (level of detail), and Analytical Sensitivity (statistical quality). You can optimize for any two vertices only at the expense of the third. High spatial resolution (low kV and low probe current) combined with high analytical sensitivity (high counts) requires long acquisition times. Acquiring data quickly requires either high probe current (sacrificing spatial resolution) or low counts (sacrificing sensitivity). This framework manages expectations and guides experiment design for a specific scientific purpose.

The Practice of Quantitative Analysis

The most common EDS application determines elemental composition in weight percent (wt%) or atomic percent (at%). Methods using certified standards measured on the same instrument provide the highest accuracy. Most routine EDS work uses standardless quantitative analysis. This approach offers convenience. It requires a deep understanding of its principles and limitations.

3.1. The Standardless Quantification Paradigm: Principles, Algorithms, and Matrix Corrections (ZAF)

Standardless quantification provides compositional analysis without certified pure-element standards. Modern EDS systems such as the Oxford AZtec platform include pre-installed standardization databases. Detector performance is characterized at the factory using primary standards. The system provides quantitative results out of the box.

The process identifies elements present and measures the characteristic peak intensity after subtracting the Bremsstrahlung background. The software uses theoretical models and factory-calibrated detector efficiency to convert raw intensities into concentrations. Matrix correction accounts for inter-element X-ray generation and absorption within the interaction volume. The ZAF correction applies three multiplicative factors to the raw intensity ratios:

Z (Atomic Number) Correction: This factor accounts for electron behavior differences between the sample and a pure-element standard. It has two components. The backscattering correction accounts for primary electrons that scatter out of the sample without generating X-rays. The stopping power correction accounts for energy loss of primary electrons as they penetrate the material. Both effects depend on the average atomic number of the material.

A (Absorption) Correction: This is often the largest and most critical factor. It accounts for X-ray absorption as X-rays travel from their generation point within the interaction volume to the detector. Absorption depends on X-ray energy (low-energy X-rays absorb more strongly), the matrix composition, and the path length to the surface. The path length depends on sample tilt and detector take-off angle. A flat, polished sample and accurate system geometry are essential for this correction to be valid.

F (Fluorescence) Correction: This factor corrects for secondary fluorescence. A characteristic X-ray from a heavier element (e.g., Fe K-alpha) can be absorbed by a lighter element (e.g., Cr) within the interaction volume. This absorption causes the lighter element to emit its own characteristic X-ray (Cr K-alpha). The effect artificially enhances the lighter element peak intensity. The F correction calculates the magnitude and subtracts it.

Most standardless routines normalize the resulting concentrations to sum to 100 percent.

3.2. Can You Trust Your Data? Factors Influencing the Accuracy of Standardless Quantification

Some practitioners consider standardless analysis semi-quantitative or a black-box tool that produces plausible but incorrect results. The accuracy is not an inherent property of the software. Instead, it is a direct function of the analyst's adherence to analytical conditions. Trustworthy data requires:

Ideal Sample Conditions: The sample must be flat, highly polished, and homogeneous at the scale of the interaction volume. It must be conductive or properly coated to prevent charging.

Correct Instrument Parameters: The accelerating voltage must provide sufficient overvoltage for all elements of interest. The probe current must generate statistically significant X-ray counts.

Accurate Geometry: The software must have the correct sample tilt and detector take-off angle for accurate absorption correction.

Complete Peak Identification: All significant elements must be correctly identified. Overlapping peaks (e.g., Pb M and S K) must deconvolute. Spectral artifacts (sum and escape peaks) must correct.

When all conditions are met rigorously, the accuracy of standardless quantification for major elements (greater than 10 wt%) is often within plus or minus 2 to 5 percent relative error. Accuracy degrades for minor elements. It degrades significantly for light elements (C, O, F), small particles, or rough surfaces.

The Normalization Trap

The final normalization step conceals many analytical errors by forcing results to sum to 100 percent.

A missed major element (e.g., oxygen from an oxide phase or carbon from a carbide) causes the software to quantify remaining elements and normalize them to 100 percent. The result is completely erroneous but may look plausible.

Incorrect background modeling or ignored peak overlaps produce wrong concentrations. Normalization masks the discrepancy.

Inaccurate ZAF correction from poor sample prep or incorrect geometry produces flawed results that still sum to 100 percent.

Experienced analysts prefer unnormalized totals as a quality signal.

3.3. Methodological Limitations and Strategies for Validating Results

You must acknowledge inherent limitations beyond operational factors.

Light Element Analysis

Quantifying light elements (beryllium to fluorine) is difficult with EDS. These elements have low X-ray fluorescence yields. They produce more Auger electrons than X-rays. Their low-energy X-rays absorb heavily within the sample and the detector window. Count rates are low. Absorption corrections are large and uncertain. Light element quantification is inherently less accurate than heavier elements.

Validation Strategies

Given the potential for error, validation is essential.

Check the Analytical Total: Some standardless routines calculate an un-normalized result when using a single standard calibration. If the analytical total is close to 100 percent (e.g., 98 to 102 percent), it confirms that all major elements are accounted for and matrix corrections are reasonably accurate. A total far from 100 percent flags a problem.

Analyze Known Standards: Periodically analyze a certified reference material matrix-matched to the unknown samples. If the system quantifies the CRM accurately, it validates results from unknowns.

Analyze Multiple Points/Areas: For non-homogeneous samples, acquiring data from multiple points or areas provides a representative bulk composition. It indicates material variability.

Correlate with Other Techniques: Cross-validate with complementary methods. Wavelength-Dispersive X-ray Spectroscopy (WDS) offers superior spectral resolution and lower detection limits. It is considered the gold standard for microanalysis. Bulk techniques such as X-ray Fluorescence (XRF) or Inductively Coupled Plasma (ICP) analysis verify overall composition.

Standardless quantification has a spectrum of reliability. Accuracy depends more on the analyst's meticulousness in sample preparation, parameter optimization, and critical evaluation than on the software. Used with expertise and caution, it is a powerful and efficient tool. Used carelessly, it is dangerously misleading.

Visualizing Composition: The Art and Science of X-ray Mapping

Point analysis provides composition at a specific location. X-ray mapping offers a visual representation of element distribution across a microstructure. This technique transforms EDS from a simple compositional tool into a key for connecting chemistry to morphology. Mapping techniques form a ladder of analytical abstraction, from raw photon counts to fully interpreted phase distributions.

4.1. Conventional Elemental Mapping: From Raw Counts to Qualitative Insights

X-ray mapping acquires a full EDS spectrum at every pixel in a scanned sample area. This process generates a three-dimensional dataset, also called a spectral image, hyperspectral dataset, or SmartMap in AZtec. The X and Y dimensions are pixel position. The Z dimension is the energy-dispersed X-ray spectrum. Two-dimensional maps extract the intensity of a specific element's characteristic X-rays at each pixel.

Elemental mapping links SEM image features (especially Backscattered Electron, BSE, images) to responsible elements. Overlaying color-coded maps on the electron image shows, for example, that a bright BSE phase is tungsten-rich and the darker matrix is cobalt-rich.

Acquisition Parameters for Mapping

Mapping involves different compromises than point analysis:

Magnification: Maps typically use 50x to 1000x magnification. Below 50x, geometric distortions from the flat scan projection become significant. Above 1000x, oversampling is likely. Pixel size becomes much smaller than the interaction volume. Adjacent pixels collect non-independent information.

Pixel Dwell Time: The beam dwells on each pixel for a few milliseconds to one second. This is much shorter than point analysis. Mapping visualizes major and minor element distributions well. Trace element distributions are generally not revealed. Not enough counts collect at each pixel.

Map Resolution and Acquisition Time: Total map acquisition time equals map resolution times dwell time per pixel. Time ranges from 30 minutes to over 12 hours for high-quality maps. Long acquisitions require excellent beam and stage stability. Modern software includes drift correction for minor shifts.

4.2. Advanced Mapping Routines: Quantitative Mapping, TruMap Corrections, and Large-Area Montaging

Raw elemental count maps give a qualitative picture. They are subject to significant artifacts. Advanced processing extracts more accurate information.

Quantitative Element Mapping: This technique applies full ZAF matrix correction to the spectrum at every pixel. The intensity represents weight percent or atomic percent, not just raw counts. It requires statistically robust spectra (typically greater than 500 counts per pixel) above qualitative mapping needs.

TruMap (Overlap and Background Correction): Background influence and overlapping peaks cause errors in simple elemental maps. The TruMap algorithm corrects pixel by pixel. First, it models and subtracts the continuous background. Second, it uses a deconvolution algorithm (often least-squares fitting) to separate overlapping peak intensities. Barium L-alpha at 4.47 keV and Titanium K-alpha at 4.51 keV overlap in a raw counts map. TruMap separates their contributions into true maps of each element.

Large-Area Mapping (Montaging): Features such as coating-substrate interfaces or large inclusions often exceed the maximum SEM field of view at the desired magnification. Large-Area Mapping automates acquisition of maps from adjacent fields. It stitches them into a seamless, high-resolution dataset. The system moves the stage between fields, acquires data, and uses image recognition to align tiles, correcting for drift.

4.3. Unveiling Microstructure: Principles and Application of Phase Mapping

Phase mapping identifies and maps distinct chemical phases (carbides, intermetallics, mineral compounds). It moves beyond single-element peaks.

Phase mapping algorithms analyze the entire spectrum from each pixel. Multivariate statistical methods group pixels with statistically similar spectra. A group of pixels represents one phase and receives a unique color. Statistical engines use techniques like Principal Component Analysis (PCA) or Vertex Component Analysis (VCA). These methods find fundamental spectral shapes (endmembers) and determine the proportion of each endmember at every pixel.

AZtec includes AutoPhaseMap. It automatically identifies phases, maps their spatial distribution, and calculates average spectrum, composition, and area fraction for each phase. This tool discovers minor or hidden phases missed by visual inspection of elemental maps.

4.4. Identifying and Mitigating Common Artifacts in X-ray Maps

X-ray maps can show artifacts that lead to severe misinterpretation. The most common artifacts are:

Topography Artifacts (Shadowing): Surface relief on unpolished samples blocks X-ray paths to the detector. This creates false dark regions in maps that correspond to topography, not composition. Flat, polished samples prevent this.

Background (Z-Contrast) Artifacts: In raw counts maps, high-average-atomic-number phases generate more Bremsstrahlung background across all energies. A high-Z phase may appear weakly in the map of an element not actually present there. Background subtraction routines like TruMap remove this artifact.

Peak Overlap Artifacts: Uncorrected maps show identical distributions when two elements have characteristic peaks at nearly the same energy. The system cannot distinguish photon sources. Peak deconvolution (a core component of TruMap) resolves this issue.

The progression from a simple elemental map to a corrected TruMap, a quantitative map, and finally a phase map increases analytical power. It moves from "Where is the iron?" to "What is the distribution, composition, and area fraction of the Fe3C phase versus the Fe-Cr solid solution matrix?" This level of interpretation solves real-world materials problems.

Application Showcase: Solving Materials Problems with EDS

EDS solves specific problems in materials science and engineering. The principles in this note combine into powerful workflows for characterizing material structure, performance, and failure. The case studies below characterize alloys, coatings, and corrosion. They represent three pillars of industrial materials characterization: understanding bulk properties, evaluating surface engineering, and conducting in-service failure analysis.

5.1. Case Study: Phase Analysis in Multicomponent Alloys (e.g., Stellite)

Cobalt-based superalloys like Stellite derive their wear and corrosion resistance from hard carbide phases in a tough cobalt-rich solid-solution matrix. Characterizing phase composition, morphology, and distribution is critical for quality control and alloy development.

Point and ID: Identifying Carbides and Matrix

Point analysis determines the composition of constituent phases. Focus the electron beam on a large matrix area and then on several large carbide particles. Acquire high-quality spectra for quantitative analysis. In a Stellite 6 alloy, point analysis reveals a Co-rich matrix with Cr and W in solution and separate Cr-rich and W-rich carbide phases. Select features significantly larger than the interaction volume. Signal from the surrounding matrix would lead to inaccurate composition. Analyze both cross-sections of bulk material and top-down views of exposed surfaces.

Linescan Analysis Across Carbide-Matrix Interfaces

A linescan visualizes compositional changes across phase boundaries. Draw a line across a carbide-matrix interface. The software plots intensity or concentration of key elements (Co, Cr, W, C) along that line. The profile shows a sharp decrease in Co and an increase in Cr and W from matrix to carbide. Use advanced linescan modes like AZtec's TruLine or QuantLine. These modes apply background and overlap corrections along the entire line. They prevent artifacts from skewing profiles. Plot all elemental profiles on a single graph with a common distance axis. Mark the interface at x equals zero for direct comparison.

Quantifying Phase Area Fractions

The carbide area fraction correlates directly with bulk hardness and wear resistance. Quantify it using two methods:

Phase Mapping: Acquire a spectral image (SmartMap) of a representative area. Run the AutoPhaseMap routine. The software segments the image into constituent phases (Co-matrix, Cr-carbide, W-carbide) and provides the area percentage for each. This method uses the full chemical signature of each phase. It is effective even when BSE contrast is ambiguous.

Image Analysis (ImageJ): If phases show strong, unambiguous contrast in a BSE image, use ImageJ software. Export the BSE image. Use thresholding to create a binary image isolating the phase of interest. Use "Analyze Particles" to calculate the area percentage. This method is faster than phase mapping. It is reliable only for simple, high-contrast systems. It requires careful user decisions on thresholding. Quantitative EDS mapping methodologies support this area fraction approach.

5.2. Case Study: Characterization of Coatings and Interfaces

Coatings enhance surface properties: wear resistance, corrosion resistance, and thermal insulation. The interface integrity between coating and substrate is paramount. EDS is the primary tool for evaluating interfaces.

High-Resolution Linescans

The linescan draws a line perpendicular to the coating-substrate interface on a polished cross-section. It generates concentration profiles for major elements of both coating and substrate. In a Stellite 6 coating on a steel substrate, the coating shows high Co, Cr, and W. The substrate shows high Fe. The transition zone shape and width provide critical information:

Diffusion Zone: Measure the width of the inter-diffusion region.

Dilution: Quantify how much substrate material (e.g., Fe) mixes into the first coating layers.

Interfacial Phases: Sharp plateaus or peaks in the transition zone can indicate brittle intermetallic phase formation.

Elemental Mapping of the Interface

A linescan gives detailed information along a single path. Elemental mapping gives a two-dimensional overview. It identifies defects that a single line might miss: pores, cracks, and delamination. Using Large-Area Mapping is often necessary to capture the entire interface at sufficient resolution. The maps reveal coating thickness uniformity and diffusion zone consistency.

5.3. Case Study: Investigating Surface Degradation and Corrosion

EDS diagnoses failure analysis, especially corrosion. It identifies elemental species in the corrosion process and maps their distribution. These details reveal the underlying mechanism.

Comparative Mapping

The core strategy for corrosion studies is comparative mapping. Acquire elemental maps under identical instrument conditions from two samples: a pristine control and a corroded sample exposed to a corrosive environment (e.g., seawater, industrial chemicals).

Identifying Corrosive Species

Comparing corroded and control maps visualizes the ingress of aggressive environmental species. A steel sample exposed to seawater shows Chlorine (Cl) and Sodium (Na) on the corroded surface. These elements are absent on the pristine surface. Oxygen (O) co-locates with corrosion products. Chlorine concentrated at the bottom of a pit indicates chloride-induced pitting corrosion.

Characterizing Corrosion Products

EDS provides elemental composition, not chemical state or compound identification. It cannot distinguish hematite (Fe2O3) from magnetite (Fe3O4). It can provide semi-quantitative elemental ratios in the corrosion scale. This information, combined with SEM morphology (uniform scale, localized pits, intergranular attack), provides strong evidence for the corrosion mechanism. EDS is the essential first step in corrosion investigation. It guides the analyst toward X-ray Diffraction (XRD) or X-ray Photoelectron Spectroscopy (XPS) for definitive compound identification.

Beyond the Microscope: Advanced Data Processing and Analysis

Vendor-supplied software handles data acquisition and routine analysis well. Research often requires capabilities beyond the vendor software. Exporting data to open-source platforms like Hyperspy unlocks advanced algorithms, custom scripting, and transparent, reproducible workflows. This section guides you in liberating data from the proprietary ecosystem.

6.1. Data Export Strategies: From Proprietary Formats to Open Standards

Exporting data into an environment with complete analyst control enables advanced processing not available in vendor software. It allows integration of EDS data with other characterization results. Scripting automates repetitive tasks. Archiving in an open-standard format ensures future accessibility. Oxford AZtec exports simple images (PNG, TIF), reports (Word, Excel), and raw spectral data files.

6.2. The EMSA/MAS Standard File Format for Spectral Data Portability

The EMSA/MAS standard format is the most important open format for single spectra. The Electron Microscopy Society of America and the Microanalysis Society developed it. It is a human-readable ASCII text file designed for cross-software and cross-laboratory portability.

The file has a header section and spectral data. Each header line begins with a hash character. Spectral data follows as a comma-separated list of X-ray counts per channel. The header contains critical parameters for reconstructing and re-quantifying the spectrum.

Table 6.1: Essential EMSA/MAS File Format Keywords

Keyword | Description | Example | Importance for Analysis --------------+--------------------------------------------------+-----------------------------+----------------------------------------------------------- #FORMAT | File format and version | EMSA/MAS Spectral Data File | Identifies the file type for parsing. #TITLE | User-defined title | Stellite 6 - Carbide Phase | Basic sample identification. #BEAMKV | Accelerating Voltage (kV) | 20.0 | Critical. Input for Z (atomic number) and F (fluorescence) corrections in the ZAF model. #PROBECUR | Probe Current (nA) | 1.5 | Needed for standard-based quantification; less critical for standardless ZAF. #LIVETIME | Detector live time (s) | 60.0 | Critical. Defines the acquisition dose and is essential for comparing intensities between spectra. #REALTIME | Total clock time (s) | 62.5 | Used with LIVETIME to calculate average dead time. #XPERCHAN | Energy per channel (eV) | 10.0 | Critical. Defines the energy scale (dispersion) of the horizontal axis. #OFFSET | Energy of the first channel (eV) | 0.0 | Critical. Defines the energy origin. Used with #XPERCHAN to calibrate the energy axis. #ELEVANGLE | Detector elevation angle (deg) | 35.0 | Critical. Defines the take-off angle, a primary input for the A (absorption) correction in the ZAF model. #AZIMANGLE | Detector azimuthal angle (deg) | 0.0 | Affects absorption calculations, especially on tilted or non-flat samples.

Parameters like BEAMKV and ELEVANGLE are not mere notations. They are fundamental physical inputs required by the ZAF correction models. Without this metadata, quantitative analysis in an external program is flawed and impossible to reproduce.

6.3. A Practical Guide to Exporting Data from Oxford Instruments AZtec

Exporting data from AZtec for different purposes requires different procedures:

Exporting Spectra: Right-click the spectrum plot in AZtec. Select the export option. Save with an .msa extension for the EMSA/MAS format.

Exporting Images and Maps: Export electron images and elemental maps as standard image files. Use TIF for highest quality or PNG for reports. Control resolution, annotations, and scale bars for presentations or publications.

Exporting for Hyperspy - The Ripple Format: To export a full spectral image (SmartMap) or linescan, right-click the data in the AZtec data tree. Choose the Ripple format export. This creates two files: a metadata file (.rpl) and a binary data file (.raw). Both files must sit in the same folder for correct reading. This is the gateway to unlocking data from the proprietary .oip project file.

6.4. Introduction to Hyperspy for Advanced EDS Data Interrogation

Hyperspy is a powerful, open-source Python library for multi-dimensional dataset analysis. It emphasizes electron microscopy data (EDS, EELS, EBSD). It offers several advantages over standard vendor software:

Advanced Algorithms: Hyperspy provides signal processing algorithms (filtering, background subtraction), multivariate statistical analysis (PCA, NMF, VCA for phase separation), and advanced curve fitting. Many exceed the built-in software capabilities.

Customization and Automation: As a Python library, Hyperspy enables custom scripts for novel analyses, batch automation, and tailored data visualization.

Reproducibility and Transparency: A scripted Hyperspy workflow is transparent, self-documenting, and reproducible. Another researcher can run the same script on the same data and obtain the identical result. This is a cornerstone of modern scientific practice.

Cost-Free and Open-Source: The software is free, and its source code is open. Community development and method validation are encouraged.

Mastering Hyperspy transitions the analyst from software operator to data scientist. The analyst designs and implements custom analytical solutions instead of being limited by GUI buttons.

6.5. A Step-by-Step Workflow: Processing AZtec Ripple (.rpl/.raw) Data in Hyperspy

Combining AZtec acquisition with Hyperspy analysis represents state-of-the-art research. The following steps outline the process for an AZtec SmartMap in Hyperspy.

Export and Organize Data: Export the SmartMap as a Ripple file (.rpl/.raw) from AZtec. Export one representative single spectrum as an .msa file. Export the corresponding electron image as a .tif file. Place all three files in the same working folder.

Load the Spectral Image: Use hs.load() in a Python environment with Hyperspy installed. Point it to the .rpl file. Hyperspy automatically finds and reads the .raw file.

Python import hyperspy.api as hs s_map = hs.load('my_dataset.rpl')

Calibrate the Dataset: The .rpl file often lacks necessary calibration metadata. Add it manually from the other exported files.

Python s_cal = hs.load('calibration_spectrum.msa') s_map.set_signal_type('EDS_SEM') s_map.axes_manager.signal_axes.offset = s_cal.axes_manager.signal_axes.offset s_map.axes_manager.signal_axes.scale = s_cal.axes_manager.signal_axes.scale s_map.axes_manager.signal_axes.units = 'keV'

Python img_cal = hs.load('electron_image.tif') s_map.axes_manager.navigation_axes.scale = img_cal.axes_manager.navigation_axes.scale s_map.axes_manager.navigation_axes.units = 'µm'

Perform Analysis: The full power of Hyperspy is available once the data is calibrated.

Python s_map.set_elements() s_map.set_lines()

Python elemental_maps = s_map.get_lines_intensity() elemental_maps.plot()

Python s_map.decomposition(algorithm='pca', output_dimension=3) s_map.plot_decomposition_results()

This workflow bridges proprietary acquisition and open-source analysis. It enables analytical flexibility and rigor essential for cutting-edge materials research.

Conclusion

Energy-Dispersive X-ray Spectroscopy extends scanning electron microscope imaging into chemical analysis. Its successful application is not simple point-and-shoot work. It demands understanding of X-ray generation and detection physics. It requires meticulous sample preparation and instrument setup. It demands critical interpretation of spectra and maps laden with potential artifacts.

The journey from raw X-ray photon to reliable quantitative result or meaningful phase map involves navigating critical trade-offs between acquisition speed, spatial resolution, and analytical sensitivity. Standardless quantification accuracy depends on the analyst's diligence in meeting strict prerequisites for sample quality and instrument parameters. Artifact-free, interpretable elemental and phase maps require advanced background and peak-overlap correction routines.

Researchers pushing materials science boundaries should not limit their workflow to the manufacturer's software. Exporting data to open-source platforms like Hyperspy grants unprecedented power to customize analyses, apply advanced machine learning algorithms, and ensure work is transparent, scriptable, and reproducible. Mastering the full pipeline from electron-solid interaction fundamentals to hyperspectral dataset scripting transforms EDS from a routine tool into a sophisticated engine for scientific discovery.

Appendix A: Expanded Table of Essential EMSA/MAS Keywords

Keyword | Description | Example | Importance for Analysis --------------+----------------------------------------------+-----------------------------+----------------------------------------------------------- #FORMAT | File format and version | EMSA/MAS Spectral Data File | Identifies the file type for parsing. #VERSION | Version of the file format | 1.0 | Specifies the standard version. #TITLE | User-defined title | Stellite 6 - Carbide Phase | Basic sample identification. #DATE | Date of acquisition | DD-MMM-YYYY | Records when the data was collected. #TIME | Time of acquisition | HH:MM:SS | Records when the data was collected. #OWNER | User/owner of the data | John Doe, Materials Lab | Records the analyst or group. #NPOINTS | Number of channels in the spectrum | 2048 | Defines the length of the data array. #NCOLUMNS | Number of data columns | 1 | Typically 1 for a single spectrum. #XUNITS | Units for the X-axis | eV | Specifies the units of the energy axis. #YUNITS | Units for the Y-axis | COUNTS | Specifies the units of the intensity axis. #DATATYPE | Type of data on Y-axis | Y | Indicates intensity data. #XPERCHAN | Energy per channel (eV) | 10.0 | Critical. Defines the energy scale of the horizontal axis. #OFFSET | Energy of the first channel (eV) | 0.0 | Critical. Defines the energy origin of the horizontal axis. #SIGNALTYPE | Type of signal | EDS | Identifies the spectroscopy type. #BEAMKV | Accelerating Voltage (kV) | 20.0 | Critical. Input for Z and F corrections in the ZAF model. #PROBECUR | Probe Current (nA) | 1.5 | Needed for standard-based quantification. #LIVETIME | Detector live time (s) | 60.0 | Critical. Defines the acquisition dose. #REALTIME | Total clock time (s) | 62.5 | Used with LIVETIME to calculate dead time. #ELEVANGLE | Detector elevation angle (deg) | 35.0 | Critical. Defines the take-off angle for A (absorption) correction. #AZIMANGLE | Detector azimuthal angle (deg) | 0.0 | Affects absorption calculations. #SOLIDANGLE | Collection solid angle (sR) | 0.13 | Needed for absolute quantification; less critical for standardless. #SPECTRUM | Start of data block | Spectrum data follows: | Header keyword indicating the start of numerical data.

Appendix B: Table of Recommended Acquisition Parameters for Common Analyses

Application | Primary Goal | Recommended kV | Target Counts / Dwell Time | Key Considerations ------------------------------------------------------+-----------------------------------------------------+----------------+-----------------------------------------+------------------------------------------------------------------- Quantitative Point Analysis of Stellite Carbides | Accurate wt% of Co, Cr, W, C in distinct phases. | 20 kV | >50,000 total counts; ~60s live time | Ensure beam spot on a feature >2-3 um. Use ZAF with light element (C) correction. Linescan across Steel Coating Interface | Profile elemental diffusion (Fe, Cr, Co) across interface. | 15-20 kV | >500 counts per pixel; ~20-50 ms/px | Use TruLine/QuantLine for background and overlap correction. Ensure line is perpendicular. Qualitative Mapping of Corroded Steel Surface | Identify presence and location of contaminants (Cl, O). | 15 kV | >200 cpp; ~5-10 ms/pixel dwell time | Compare directly to a clean control sample map. Look for O and Cl co-localization. Phase Mapping of Eutectic Alloy | Determine area fraction and composition of lamellar phases. | 15 kV | >200-500 cpp; ~10-20 ms/pixel dwell | Requires high beam stability for long map times. AutoPhaseMap is ideal. High-Resolution Mapping of Nanoparticles | Map elemental distribution within features <100 nm. | 5-10 kV | >500 cpp; >100 ms/pixel dwell | Use low kV to minimize interaction volume. Accept long map times. Sample must be flat.

Recommended Counts for a Reliable EDS Spectrum

Generating enough X-ray counts produces distinct characteristic peaks above the background noise. A general guideline is at least 10,000 to 50,000 counts for a point spectrum.

The statistics threshold depends on the data analysis technique. The metric is total X-ray counts per pixel spectrum (cpp).

EMSA File Format

EDS data exports to the human-readable EMSA file format. This format stores spectrum data and metadata about the instrument and data acquisition.

More information is available at https://the-mas.org/wp-content/uploads/2018/11/emmff_ascii.txt

For most analysis, use 20.0 kV.

Magnification in x or times. Camera length in mm.

EDS spectral resolution

Characteristic X-rays have discrete energies. The detection process spreads discrete lines into Gaussian peaks. The spectral resolution of an ED spectrum is the peak width in eV at half the peak height. This is the full width at half maximum (FWHM) resolution. Resolution depends on the detector and its electronics.

Spectral resolution depends on the time constant or process time used during collection. The process time is the pulse processor duration for averaging the incoming X-ray signal. Longer process times produce better resolution. In the diagram above, P1 is the shortest process time and P6 is the longest. Longer processing means longer dead time. The system counts no incoming X-rays during dead time. The total acquisition time increases.

EDS spectral artefacts

Sum peaks and escape peaks become significant at high X-ray count rates or when dead time exceeds 60 percent. Commercial EDS software packages mark the positions of escape peaks and sum peaks related to major characteristic X-ray lines. You can identify them readily in ED spectra.

Quantitative ED microanalysis requires correction for spectral artefacts. The software must subtract escape peaks and sum peaks from the spectrum. It must return counts to the peaks where they should have been recorded. This procedure is pulse pile-up correction.

Mapping

X-ray mapping produces images of elemental distributions in a sample. Users assimilate X-ray map information readily. Knowing the distribution of a particular element can solve a problem without quantitative point analysis. BSE images show compositional variations. X-ray maps show which elements cause those variations.

Maps collect at magnifications of 50x to 1000x. Below 50x, geometrical errors are possible. Above 1000x, oversampling is likely.

Pixel dwell time ranges from a millisecond to a second or more. These times are much less than point analysis (30 to 60 seconds). Mapping is ideal for major elements. It does not reveal minor or trace element distributions because not enough X-rays collect.

Total acquisition time depends on the map resolution (number of pixels) and the dwell time per pixel. Time ranges from about 0.5 hours to over 12 hours. A stable beam is required for long acquisitions. Newer software packages adjust for beam drift.

https://myscope.training/EDS_Parameters_for_X_ray_mapping https://myscope.training/EDS_Artefacts_in_X_ray_mapping https://www.thermofisher.com/blog/mining/4-eds-mapping-routines-that-can-be-used-on-your-mineralogical-sample/ https://nano.oxinst.com/library/past-blogss/acquiring-high-res-eds-maps https://assets.thermofisher.com/TFS-Assets/CAD/Vector-Information/WS52731-Optimum-Settings-Mapping.pdf

Separating phases in SEM with EDS

https://link.springer.com/article/10.1007/s13632-023-01020-7#citeas

EDS Training book

https://www.mse.ucr.edu/sites/default/files/2023-09/UCR%20MSE%20EDS%20Training%20Notebook%20rev%201.3.pdf https://utw10193.utweb.utexas.edu/InstrumentManuals/Oxford%20EDS%20AZtec%20User%20Manual.pdf https://cst.temple.edu/sites/cst/files/documents/EDS%20Oxford-Quantitative-Guide.pdf

Element mapping - Exporting maps as csv files https://www.youtube.com/watch?v=rdlpac7yIeI

Element mapping - Montaging element maps https://www.youtube.com/watch?v=-KFABI3Sj48

Element mapping - TruMap to remove background and overlaps https://www.youtube.com/watch?v=7vFUKRWpJF8

Point and ID - Exporting mineral chemical data https://www.youtube.com/watch?v=76gtV9n4cJw

https://www.youtube.com/watch?v=CD3UyZdq58Y&t=19s

Exporting images in AZTech https://www.oxinst.com/blogs/exporting-data-from-aztec-effectively https://www.oxinst.com/blogs/exporting-data-from-aztec-effectively-part-2

https://groups.google.com/g/hyperspy-users/c/l163WxA3pIk?pli=1

Hyperspy https://hyperspy.org/hyperspy-doc/v1.4/user_guide/eds.html https://github.com/hyperspy/exspy

An excellent tutorial for Ni superalloy https://zenodo.org/records/6566768

A YouTube tutorial https://www.youtube.com/watch?v=zfLjByjm2N0

Advanced EDS https://www.youtube.com/watch?app=desktop&v=kjPSOeu8cg0&t=678s

PHASE MAPPING: An Open Evaluation of Hyperspectral Unmixing Strategies for EDS Analysis https://pmc.ncbi.nlm.nih.gov/articles/PMC7540739/

A presentation (template worth borrowing) https://pages.nist.gov/2019-06_CCEM_presentation/#/34

Exporting from AZtec for python

https://groups.google.com/g/hyperspy-users/c/l163WxA3pIk?pli=1

In the right-hand menu of AZtec, right-click each map or line spectra. Export as a raw/RPL file using the Ripple format. This exports the dataset into two files with .rpl and .raw extensions. Both files must sit in the same folder. Feed the .rpl file to Hyperspy. There is no way to read the .oip or .dat files directly.

Things to do with EDS

Point analysis of carbides and matrix (crosssection)

https://www.researchgate.net/publication/341497126/figure/fig4/AS:893081754800130@1589938691098/Specific-EDS-microstructure-analysis-a-Stellite-6-and-b-Stellite-6-with-added-Fe.png

Point analysis of carbides and matrix (top view)

Examine the exposed surface to determine if it is matrix or carbide. This approach is not always feasible for deep regions.

https://www.researchgate.net/publication/341497126/figure/fig4/AS:893081754800130@1589938691098/Specific-EDS-microstructure-analysis-a-Stellite-6-and-b-Stellite-6-with-added-Fe.png

EDS line mapping across substrate/coating interface

Measure elemental distribution (e.g., Fe, Cr, Co, W) along the material change such as a thermal spray coating on a substrate. Always plot all elements on the same X axis with the zero point at the surface.

https://www.researchgate.net/profile/Zidong-Lin/publication/344413531/figure/fig3/AS:941024809611275@1601369206586/Elemental-distribution-measured-using-EDS-a-condition-A-Stellite-6-S355-b.png https://www.researchgate.net/profile/Qiran-Cheng/publication/340797802/figure/fig8/AS:887504467132420@1588608962246/Stellite-12-coating-EDS-linear-scan-results-shows-the-EDS-line-scan-results-of-stellite12.ppm

For even more street cred: https://www.researchgate.net/publication/348863477/figure/fig4/AS:985179535466497@1611896513414/Multi-track-cladding-Stellite-6-on-mild-steel-top-and-EDS-imagery-near-interface.png

EDS line mapping across carbide matrix boundary

Area fractions of phases

Small regions may be ignored. Hyperspy may be worthwhile for this analysis.

The chemical compositions of the phases identified from the microstructure were examined using EDS. The BSE mode of SEM imaging shows clear brightness contrast between individual phases. Examples include Cr-rich carbides, W-rich carbides, and Co-rich matrix. Using ImageJ software, compute the area fraction of each phase from the SEM image.

EDS mapping of top view regions

Collect maps from:

Related