| |
Accurate peak identification plays a critical role in EDS analysis - not only for the purpose of knowing what elements are present in a sample, but for obtaining good quantitative results. However, there are a number of samples where the peaks for two or more elements are too close together to make a positive identification using normal qualitative methods.
EDAX pioneered the development of Halographic Peak Deconvolution (HPD), a visual method that allows an analyst to easily determine if all the elements in a spectrum have been identified at the push of a button. The HPD algorithm accounts for the detector parameters and efficiency, along with the experimental parameters used in operating the electron microscope during the acquisition of the unknown spectrum. These parameters are used to synthesize a spectrum model, which is then fitted to the unknown spectrum.
- Intelligence within HPD helps identify and resolve peak identities, especially in overlaps, so user is certain of constituents within sample
- Graphics assist in visualizing the results making the tool quick and easy to use
- Certainty of identifying peaks improves quantification
One common instance of overlapping peaks which can demonstrate the utility of HPD is the case of a small MnKα peak being obscurred by the presence of a larger CrKβ peak, as found in stainless steel. The HPD modeling is illustrated below in Figures 1 - 4.
 |
Figure 1. The HPD profile indicates a portion of this peak has not been accounted for in the model. |
 |
Figure 2. Subtracting the HPD profile reveals another peak. Notice the peak apex is shifted to the left from the original location, a common phenomenon with overlapping peaks. |
 |
Figure 3. The Mn peak markers line up nicely with the subtraction residual, so we add Mn to peak list. |
 |
Figure 4. Undo the substraction and recalculate HPD model. The peak is completely described. |
 |
Figure 5. Spectra from two different samples. The first example shows the correct identification of Si and Sr and the second example shows the incorrect identification of only Pb in a sample containing Pb and S. |
|
The Importance of HPD for Quantitative Analysis
HPD provides a high degree of certainty in accurately assessing the qualitative information, but its value extends to quantitative analysis, too. Standardless quant routines typically used for EDS work normalize the results to 100%. If an element is not identified and missing from the quantification routine, its weight percent value is distributed and among the other elements, incorrectly skewing the concentration values. This is why it is imperative that all the elements be identified when using a normalized quantification routine.
The following looks at the results for EDAX's true standardless analysis of a tool steel. The same Cr-Mn overlap discussed above occurs, but only a very small amount of Mn is present.
|
| Wt% Analysis of Commercial Grade Tool Steel with Peak Overlaps |
| |
Si-K |
W-M |
V-K |
Cr-K |
Mn-K |
Fe-K |
| EDAX ZAF |
0.45 |
3.32 |
0.19 |
1.27 |
0.22 |
94.38 |
| Given |
0.31 |
2.65 |
0.22 |
1.28 |
0.31 |
94.40 |
Making the analysis even more demanding was the presence of SiK and W-M, another peak overlap situation as seen in Figures 5-7.
 |
Figure 6. SiK, but no W, in peak list and peak not modeled well. |
 |
Figure 7. W M, but no Si, in peak list and peak not modeled well. |
 |
Figure 8. SiK and W M in peak list and peak completely modeled. |
|
|
