Asking for help with WDS (having negative weight percent of oxygen)

[wlsdnrud12]

Happy New Year, everyone. I am Wookyung Jin, a PhD candidate at Lehigh University.  Currently, I am working on detecting the oxygen weight percent for my oxide sample.

To run this, I made the settings as follows: spec 1 (LDEN) for O, spec 3 (PET) for Ta, spec (PET) for W, spec 4 (PET) for W, spec 3 (LIF) for Cr, and spec 4 for V. However, when I ran WDS, the total weight percentage was around 80%, which indicates that something was wrong. So, I stepped back to test the standard. I tested the standard as an unknown (detecting Cr standard as unknown to check the setting).

In principle, it should give 100 wt% for every unknown sample. However, every standard detected a negative wt % of oxygen. In more detail: Cr gave Cr (98.1 wt%) with O (-1.7 wt%), V gave V (92.4 wt %) with O (-6.1 wt %), Ta gave Ta (88.2 wt%) with O (-13.4 wt %), W gave W (82.7 wt %) with O (-15.5 wt%).

When I add up their O and each wt %, they roughly add up to 100 wt %. I think this may relate to the issue with my lower wt % for my actual unknown sample.

I hope this is not a hard thing to fix. Is there any suggestion to fix this problem? Thanks for the reading my post.

[John Donovan]

In principle, it should give 100 wt% for every unknown sample. However, every standard detected a negative wt % of oxygen. In more detail: Cr gave Cr (98.1 wt%) with O (-1.7 wt%), V gave V (92.4 wt %) with O (-6.1 wt %), Ta gave Ta (88.2 wt%) with O (-13.4 wt %), W gave W (82.7 wt %) with O (-15.5 wt%).

When I add up their O and each wt %, they roughly add up to 100 wt %. I think this may relate to the issue with my lower wt % for my actual unknown sample.

I hope this is not a hard thing to fix. Is there any suggestion to fix this problem? Thanks for the reading my post.

Negative net intensities indicates that your off-peak backgrounds are not set properly.  Did you perform wavescans on your standards and unknowns to see where the high and low backgrounds should be placed? See:

https://smf.probesoftware.com/index.php?topic=68.0

Also, oxygen is usually a highly curved background, so you will likely need to utilize an exponential or polynomial background fit:

https://smf.probesoftware.com/index.php?topic=68.msg7072#msg7072

For oxygen to be fully quantitative you will also need to make use of empirical mass absorption coefficients (MACs):

https://smf.probesoftware.com/index.php?topic=1340.0

and most likely also area peak factors (APFs) for correction of peak shape/shift from different bonding characteristics between your standard and your unknown:

https://smf.probesoftware.com/index.php?topic=536.0

Obtaining quantitative oxygen requires significant effort on the part of the analyst.  Let us know if you have further questions.

By the way, what is the rough composition of your actual unknown?

[John Donovan]

There's also a nice video tutorial here on setting ones background positions from wavescans:

https://www.youtube.com/watch?v=qpSFcr-J5L4&;

About 2/3 of the way through, though the whole video is worth watching for those new to EPMA.