Statistical analysis of low-voltage EDS spectrum images Page: 1 of 3
This report is part of the collection entitled: Office of Scientific & Technical Information Technical Reports and was provided to Digital Library by the UNT Libraries Government Documents Department.
The following text was automatically extracted from the image on this page using optical character recognition software:
Statistical analysis of low-voltage EDS spectrum images
Ian M. Anderson MAY 0
Oak Ridge National Laboratory, Bldg. 5500, MS-6376, PO Box 2008, Oak Ridge, TN 37831G7$Sl j
The benefits of using low ( 5 kV) operating voltages for energy-dispersive X-ray
spectrometry (EDS) of bulk specimens have been explored only during the last few years.'"4 This
paper couples low-voltage EDS with two other emerging areas of characterization: spectrum imaging
and multivariate statistical analysis (MSA). The specimen analyzed for this study was a cross section
of a computer chip manufactured by a major semiconductor company. Data acquisition was performed
with a Philips XL30-FEG SEM operated at 4 kV and equipped with an Oxford super-ATW detector
and XP3 pulse processor. The specimen was normal to the electron beam and the take-off angle for
acquisition was 35. The microscope was operated with a 150 tm diameter final aperture at spot size
3, which yielded an X-ray count rate of ~2000 s . EDS spectrum images were acquired as Adobe
Photoshop files with the 4pi plug-in module. (The spectrum images could also be stored as NIH
Image files, but the raw data are automatically rescaled as maximum-contrast (0-255) 8-bit TIFF
images - even at 16-bit resolution - which poses an inconvenience for quantitative analysis.) The 4pi
plug-in module is designed for EDS X-ray mapping and allows simultaneous acquisition of maps from
48 elements plus an SEM image. The spectrum image was acquired by re-defining the energy intervals
of 48 elements to form a series of contiguous 20 eV windows from 1.25 kV to 2.19 kV. A spectrum
image of 450 x 344 pixels was acquired from the specimen with a sampling density of 50 nm / pixel
and a dwell time of 0.25 live seconds per pixel, for a total acquisition time of -14 h. The binary data
files were imported into Mathematica for analysis with software developed by the author at Oak Ridge
National Laboratory.' A 400 x 300 pixel section of the original image was analyzed. MSA required
185 Mbytes of memory and -18 h of CPU time on a 300 MHz Power Macintosh 9600.
The results of this study are shown in Fig. 1. Image (a) is a secondary electron (SE)
image of the analyzed area of the device. Images (b-d) are component images I0-I2 of the statistical
analysis and spectra (e,f) are the MSA spectral components El and E2 corresponding to images Il and
12. The plot in (g) gives the amplitude variations along the line trace shown in (d), averaging 10 pixels
(0.5 m) in the vertical direction. Image 10 gives the average X-ray intensity levels in the spectrum
image; it is the image that would be acquired with a non-dispersive detector that collected all X-rays in
the -1 kV band of energies analyzed. Images II and 12 are chemically sensitive MSA components. The
image contrast in I1 arises mainly from spectral differences between Al and Si, as indicated by El, and
the contrast in 12 is mainly due to the W plugs, as indicated by E2. In Ii, the Al lines appear dark, the
Si substrate (at bottom of image) appears brightest, and the Si02 dielectric is a medium grey. The W
plugs assume a grey level that almost matches that of the Si02, and are almost invisible. However, this
feature is dominant in 12. Note that excellent image contrast is obtained, even though the EDS detector
cannot resolve the Si-K and W-Ma lines, which are only -34 eV apart, as shown in E2. The next
generation of EDS detectors will be able to resolve these X-ray lines,6 which at the present -100 eV
resolution give rise to the "first derivative" feature in E2. A line trace (vertical average of 10 pixels)
across the 600-nm-wide W plugs shown in (g) illustrates the excellent spatial resolution of the EDS
spectrum image. The vertical averaging improves the noise at the expense of resolution. The resolution
of individual line traces, defined as 90% of the total change in intensity, is -4 pixels or -200 nm.7
1. Nockolds C.E., Microbeam Analysis 3 (1994) 185; Proc. Microscopy & Microanalysis 1996, 476.
2. Boyes E., Proc. 13th ICEM: Electron Microscopy 1994 1 (1994) 51.
3. Johnson M.T. et al., Proc. Microscopy & Microanalysis 1996, 478.
4. Newbury D.E., Proc. Microscopy & Microanalysis 1997, 881.
5. Anderson I.M. and Bentley J., ibid., 931; Mater. Res. Soc. Symp. Proc. 458 (1997) 81.
6. Wollman D.A. et al., J. Microscopy 188 (1997) 196.
7. Research at the Oak Ridge National Laboratory (ORNL) SHaRE User Facility was sponsored by the Division of
Materials Sciences, U.S. Department of Energy, under contract DE-AC05-960R22464 with Lockheed Martin Energy
Research Corporation. The author thanks Dr. Jim Bentley of ORNL for helpful discussions about the data acquisition.
'The submitted manuscript has been authored by a
~ contractor of the U. S. Government under contract No.
R 7 , r DE-ACOS.96OR22464. Accordingly, the U.S.
DISTRIBUTIQN OF This D .C EN IS UNUMhED M S E =: = ..s
Ji~ G IINO TSD'i3o I NITI QN Government retains a nonexclusive, royalty-free license
to publish or reproduce the published form of this
contribution, or allow others to do so, for U.S.
Here’s what’s next.
This report can be searched. Note: Results may vary based on the legibility of text within the document.
Tools / Downloads
Get a copy of this page or view the extracted text.
Citing and Sharing
Basic information for referencing this web page. We also provide extended guidance on usage rights, references, copying or embedding.
Reference the current page of this Report.
Anderson, I.M. Statistical analysis of low-voltage EDS spectrum images, report, March 1, 1998; Tennessee. (https://digital.library.unt.edu/ark:/67531/metadc712519/m1/1/: accessed April 18, 2019), University of North Texas Libraries, Digital Library, https://digital.library.unt.edu; crediting UNT Libraries Government Documents Department.