Help with some Bruker tests on a CAMECA running PeakSight

[klcrispin]

On my new SX-5, I have a Bruker EDS detector.  This detector has limited independent operational capability.  In other words, it is designed primarily to run through PeakSight software.  A few months ago, I discovered a very, very bad problem while doing intensity line scans.  WDS scans are fine but EDS scans THROUGH PeakSight are not smooth values, even running on absolutely pure metal standards.  Basically from point to point the EDS intensity can vary by as much as 200%!!!!  Quantitative line scans do not "appear" to show the same problem (i.e. the values are close to nominal within error).  So, it is only the intensity that seems to have this wild fluctuation.  However, knowing that the intensities have this fluctuation, I can no longer trust the quantitative values.

Attached to this message is a powerpoint that I put together to show the problem to both Bruker and CAMECA. 

Bruker finally visited campus Monday and ran some tests and tried some settings and a new SVE box, but the problem still persists.  At this point neither CAMECA nor Bruker are sure whether it is on their side or the other.  I did learn a couple of things in the past two days through more testing.  The problem does not appear when running quick line scans (i.e. dwell times of <2000 ms).  Also, the problem is greatly reduced if you run the EDS detector at an extraordinary dead time (>80%), using a 20 kcps card instead of 130 kcps card.

What I am hoping that some of you out there can help me with is a test to determine whether this is a widespread problem or whether it is limited to my instrument alone.  To run the test, please use the following settings:  (see screenshots attached for setup used)
1.  Use the 130 kcps max throughput setting on the Bruker software
2.  Use an aluminum stub or foil or standard to run this test
3.  Collect Al signal on both TAP and EDS through PeakSight
4.  15 keV, 20nA, 0 spot size
5.  Dwell time of 20000 ms (20 sec), line spacing of 2-5 um for 70+ points

Should take 30-45 minutes.

Please report back what version of Bruker software, what version of PeakSight, and what CAMECA model.  If you can attach either a screenshot to compare them or send me a text file of the data, I would appreciate it.

I appreciate any help to narrow down this frustrating problem.

[Probeman]

Bruker finally visited campus Monday and ran some tests and tried some settings and a new SVE box, but the problem still persists.  At this point neither CAMECA nor Bruker are sure whether it is on their side or the other.  I did learn a couple of things in the past two days through more testing.  The problem does not appear when running quick line scans (i.e. dwell times of <2000 ms).  Also, the problem is greatly reduced if you run the EDS detector at an extraordinary dead time (>80%), using a 20 kcps card instead of 130 kcps card.

Hi Katherine,
So is the problem just the fast pulse processing mode not being stable?   But then why would quantifying these scans fix the problem?  Is this standardless quant and you are normalizing the sum to 1.0?
john

[klcrispin]

It may be the pulse processing mode.  Not sure. 

And, I didn't quantify these particular scans.  The graphs just show a normalization to the average intensity.  It is easier to plot a direct comparison of WDS to EDS that way.

What I meant by the quantification, is that if I go into the Quant mode after standardizing on both WDS and EDS, then run a series of points in quant mode, I get the values I expect.  i.e. Aluminum metal standard comes out to 100% on WDS and the EDS values range from 97-102%.  Not perfect, but ok.

In both cases, the count time per spot is 20 seconds so, no, it makes no sense why intensity varies in the profile mode, but quant mode seems ok.

Either way, I can't trust my EDS right now.

[Probeman]

In both cases, the count time per spot is 20 seconds so, no, it makes no sense why intensity varies in the profile mode, but quant mode seems ok.

Either way, I can't trust my EDS right now.

Exactly. That the intensities are unstable, but quant with standards fixes it, is what is stumping me also!

Maybe Bruker quant normalizes to the zero eV peak? 

I assume the beam is stable?

[klcrispin]

Yep, beam is stable.

Bruker is having a meeting right now as we speak to try to figure this out.  I'm just hoping to get some data from other systems to see if I am the only one who sees this problem.