Author Topic: Re: Trace Element Strategies  (Read 1177 times)

Nicole

  • Student
  • *
  • Posts: 2
Re: Trace Element Strategies
« on: October 19, 2017, 08:30:28 am »
Hi all,

I want to ask if there are people working on sulphides on a JEOL microprobe with the Probe for EPMA software. I am measuring sulphides (mainly pyrites and arsenopyrites associated with gold) and I want to know how to manage the raw data and the counting times.

Regards,

Nicole
« Last Edit: October 19, 2017, 09:45:13 am by John Donovan »

Probeman

  • Emeritus
  • *****
  • Posts: 1672
  • Never sleeps...
    • John Donovan
Re: Re: Trace Element Strategies
« Reply #1 on: October 19, 2017, 10:50:58 am »
I want to ask if there are people working on sulphides on a JEOL microprobe with the Probe for EPMA software. I am measuring sulphides (mainly pyrites and arsenopyrites associated with gold) and I want to know how to manage the raw data and the counting times.

Hi Nicole,
Can you provide more details on what you mean by "how to manage the raw data and the counting times"?  Are you asking about analytical strategies for measuring trace elements in sulfides?  That would partly depend on what your instrument's spectrometer/crystal configuration is.

Are you currently using Probe Software on a JEOL instrument?  Which model JEOL do you have?
john
The only stupid question is the one not asked!

Nicole

  • Student
  • *
  • Posts: 2
Re: Re: Trace Element Strategies
« Reply #2 on: October 20, 2017, 01:17:02 pm »
Hi John,

Yes I am asking for analytical strategies for measuring trace elements in sulfides with a JEOL JXA-8530F microprobe. The configuration now used counting time on the peak 300 seconds for Au, 180 seconds for As, Se, Mg, Al, and 60 seconds for the rest of the elements. Background count times are 30 seconds at high and low positions for all elements. Are these conditions common for sulfides?

I also want to ask how to use the multiple backgrounds. Is there an automatic procedure to extract the backgrounds and to use the data, without having to the corrections afterwards?

Nicole

Probeman

  • Emeritus
  • *****
  • Posts: 1672
  • Never sleeps...
    • John Donovan
Re: Re: Trace Element Strategies
« Reply #3 on: October 20, 2017, 06:58:40 pm »
Hi John,

Yes I am asking for analytical strategies for measuring trace elements in sulfides with a JEOL JXA-8530F microprobe. The configuration now used counting time on the peak 300 seconds for Au, 180 seconds for As, Se, Mg, Al, and 60 seconds for the rest of the elements. Background count times are 30 seconds at high and low positions for all elements. Are these conditions common for sulfides?

I also want to ask how to use the multiple backgrounds. Is there an automatic procedure to extract the backgrounds and to use the data, without having to the corrections afterwards?

Nicole

Hi Nicole,
It is difficult to be specific without more information on your spectrometer/crystal configuration because the choice of emission lines will affect the sensitivities, and this depends on which crystals are available.

But in general one should increase counting time and/or beam current (and beam energy) to improve sensitivity.  Accuracy at the trace level is also important, hence the need for a "blank" sample. That is a sulfide with zero or known trace elements from another technique.  I have a very pure natural iron sulfide (pyrite) that I have had ICP-MS performed on, so I can measure that as an unknown along with my other unknowns (using the same conditions) to check trace accuracy. The question is: can one accurately measure zero?  See here for more info:

http://probesoftware.com/smf/index.php?topic=928.msg5958#msg5958

A paper I wrote on this topic looks at SiO2 but the strategies are applicable to any material:

http://epmalab.uoregon.edu/pdfs/3631Donovan.pdf

Back to improving sensitivity, you should realize that at zero concentration the background measurement is statistically as important as the on-peak measurement. That is, when measuring concentrations close to zero (say, below 500 PPM), we generally count as long off peak as we do on-peak.   So for example, 300 seconds on-peak, 150 second on the high off-peak and 150 seconds on the low off-peak. This is because the variances add in quadrature when calculating the net intensities.  As the expected concentrations increase one can decrease the amount of time integrating the off-peak intensities.

You ask about "multiple backgrounds" but it sounds to me that you are using the JEOL software, correct?  I think that only Probe Software's Probe for EPMA software can acquire and regress multiple off-peak backgrounds as described here:

http://probesoftware.com/smf/index.php?topic=701.msg4283#msg4283

Of course one could perform these multiple off-peak background corrections manually (as many labs used to do, until they obtained Probe for EPMA for their JEOL instruments), but it's painful and slow.

I suggest you start choosing emission lines that provide the best intensity (for sensitivity), then increase the on and off-peak counting times, and increase the beam energy to increase to interaction volume and run at least 100 or 200 or more nA of beam current and run some test measurements on some "blank" sulfides and see what sort of detection limits you can obtain.

Maybe someone reading this has a JEOL probe run measuring traces in sulfides that they can share with us... I have a Cameca instrument which is a different animal when looking at trace elements, e.g., the Cameca is better at low energy element sensitivities, while the JEOL is better at high energy emission lines due to its Xe detectors.  Of course this may all change in the next few years as the instrument vendors utilize SDD detectors in place of WDS gas detectors.
john

PS Do not forget about spectral interferences (another nice thing is that Probe for EPMA corrects these easily):

http://probesoftware.com/smf/index.php?topic=803.0

And to add to your woes(!), be careful measuring trace elements near phases with high concentrations of those elements, due to secondary fluorescence from boundary phases:

http://probesoftware.com/smf/index.php?topic=58.0

By the way, if you input standard and unknown count rates into this dialog in CalcZAF (Run | Model Detection Limits menu), you can obtain estimates of your sensitivity as shown here:

http://probesoftware.com/smf/index.php?topic=121.0

john
« Last Edit: October 21, 2017, 09:01:52 am by Probeman »
The only stupid question is the one not asked!

Karsten Goemann

  • Global Moderator
  • Professor
  • *****
  • Posts: 168
Re: Trace Element Strategies
« Reply #4 on: October 22, 2017, 04:05:20 pm »
Hi Nicole,

We've been doing a lot of complex sulphide analyses (including traces) on our Cameca SX100 and now have a JEOL 8530F Plus microprobe ourselves that we're using for similar work.

I agree with John that for trace element analysis background correction is absolutely crucial and "blank standards" such as a pure pyrite in your case are important for method validation and/or blank correction. Like John says for trace elements the total background counting time should be equal to the peak counting time.

I can't really say that the Cameca is better than the JEOL at low x-ray energies but that would depend a lot on your specific instrument configuration.

In addition to John's suggestions you could also run the trace elements at a "second condition" at higher beam current than the major element.

Any specific reason to analyse Mg and Al in those sulphides? Are you concerned about inclusions of silicate or carbonate minerals?

If you send us more details about your method (accelerating voltage, beam current, instrument type, monochromator crystal configuration, distribution of elements across the spectrometers) we can comment more specifically.

Cheers, Karsten

Probeman

  • Emeritus
  • *****
  • Posts: 1672
  • Never sleeps...
    • John Donovan
Re: Trace Element Strategies
« Reply #5 on: November 02, 2017, 09:52:05 pm »
I can't really say that the Cameca is better than the JEOL at low x-ray energies but that would depend a lot on your specific instrument configuration.

Hi Karsten,

I should have explained myself better.

Both JEOL and Cameca instrument WDS detectors start with similar performance.  But I have observed that sealed Xe detectors can get rather noisy over time.  This is primarily due to out gassing inside the detector (contamination), and subsequent loss of Xe gas to the spectrometer/column vacuum. Colin MacRae tells me that he replaces his Xe sealed detectors every 3 years or so for this reason.   

The nice thing about flow detectors is that they stay relatively clean and at their rated pressure for many years.  Even after 10 years, all 5 of my SX100 detectors are still good for trace element analysis, though I will say I'm envious of the JEOL Xe detectors' high energy sensitivity.

One day when we all have solid state WDS detectors this will be a moot point!
john
The only stupid question is the one not asked!

Probeman

  • Emeritus
  • *****
  • Posts: 1672
  • Never sleeps...
    • John Donovan
Re: Trace Element Strategies
« Reply #6 on: November 07, 2017, 03:05:14 pm »
I always find something interesting when teaching the EPMA class here in Oregon.  This year the students and I ran some tests of measuring trace Au, Cu and Ag in pyrite and we found a number of interesting stories to tell.

The classic one that many have noted is the "hole" in the continuum for Au La on an LiF crystal as reported by Self et al. (Self PG, Norrish K, Milnes AR, Graham J, Robinson B. Holes in the background in XRS. X-Ray Spectrom. 1990;19:59–61) as seen here:



Since instruments have improved slightly since then, we have the following scan on pyrite using a 60 sec per point wavescan:



One might suspect that this could be a sample absorption edge (since Au La is at 9.7 keV and the Fe absorption edge is 7.1 keV though that isn't close enough, but if we perform a similar scan on say Ti metal, we see the same "hole" in the continuum as seen here:



So the "hole" in the continnum is another Bragg crystal secondary diffraction position, similar to the hole in the continuum underneath the Ti Ka peak position on PET Bragg crystals.

So what gets interesting is this: what is that other peak in the scan on Ti metal?  If I plot all the element KLM markers I get this plot:



Could it actually be Pt?  I don't think so, but maybe I should check.  What else could be a trace element in Ti metal at this position?
The only stupid question is the one not asked!

Probeman

  • Emeritus
  • *****
  • Posts: 1672
  • Never sleeps...
    • John Donovan
Re: Trace Element Strategies
« Reply #7 on: November 07, 2017, 03:22:46 pm »
I beginning to think that all my Bragg crystals are getting strange artifacts...

Here's another weird story from the EPMA class run looking at Au, Cu and Ag in pyrite. After peaking the LiF spectrometer on pure copper, we ran a quick WDS scan to check our bgds for interferences. This wavescan on our pyrite standard (with 70-100 PPM Cu from AA and XRF) was 3 sec per point:



So we thought : OK, looks like we have no off-peak interferences. So we ran quant on the pyrite and we got this:

Un    8 pyrite #730
TakeOff = 40.0  KiloVolt = 20.0  Beam Current = 100.  Beam Size =    0

Un    8 pyrite #730, Results in Elemental Weight Percents
 
ELEM:       Au      Au      Cu      Ag      Fe       S      Ni      Co      As      Zn      Pb      Cr      Ti
TYPE:     ANAL    ANAL    ANAL    ANAL    SPEC    SPEC    SPEC    SPEC    SPEC    SPEC    SPEC    SPEC    SPEC
BGDS:      LIN     LIN     LIN     LIN
TIME:   400.00  400.00  400.00  400.00     ---     ---     ---     ---     ---     ---     ---     ---     ---
BEAM:   100.19  100.19  100.19  100.19     ---     ---     ---     ---     ---     ---     ---     ---     ---

ELEM:       Au      Au      Cu      Ag      Fe       S      Ni      Co      As      Zn      Pb      Cr      Ti   SUM 
XRAY:     (la)    (ma)    (ka)    (la)      ()      ()      ()      ()      ()      ()      ()      ()      ()
   389   -.001    .005   -.007    .004  46.550  53.450    .000    .000    .009    .009    .010    .000    .058 100.087
   390    .007    .007   -.007    .001  46.550  53.450    .000    .000    .009    .009    .010    .000    .058 100.094
   391    .006    .006   -.006    .013  46.550  53.450    .000    .000    .009    .009    .010    .000    .058 100.105
   392    .012    .004   -.009    .011  46.550  53.450    .000    .000    .009    .009    .010    .000    .058 100.104
   393   -.006    .008   -.012    .004  46.550  53.450    .000    .000    .009    .009    .010    .000    .058 100.079
   394   -.003    .008   -.008    .007  46.550  53.450    .000    .000    .009    .009    .010    .000    .058 100.089

AVER:     .002    .006   -.008    .007  46.550  53.450    .000    .000    .009    .009    .010    .000    .058 100.093
SDEV:     .007    .002    .002    .005    .000    .000    .000    .000    .000    .000    .000    .000    .000    .010
SERR:     .003    .001    .001    .002    .000    .000    .000    .000    .000    .000    .000    .000    .000
%RSD:   292.87   25.87  -27.81   69.15     .00     .00     .00     .00     .00     .00     .00     .00     .00
STDS:      579     579     529     547     ---     ---     ---     ---     ---     ---     ---     ---     ---

STKF:   1.0001  1.0000   .9975   .9920     ---     ---     ---     ---     ---     ---     ---     ---     ---
STCT:  15410.4  5789.2 13183.3  3762.1     ---     ---     ---     ---     ---     ---     ---     ---     ---

UNKF:    .0000   .0001  -.0001   .0000     ---     ---     ---     ---     ---     ---     ---     ---     ---
UNCT:       .3      .3     -.9      .2     ---     ---     ---     ---     ---     ---     ---     ---     ---
UNBG:    166.4    12.3    36.2     8.1     ---     ---     ---     ---     ---     ---     ---     ---     ---

ZCOR:   1.4520  1.0365  1.1585  1.3372     ---     ---     ---     ---     ---     ---     ---     ---     ---
KRAW:    .0000   .0001  -.0001   .0001     ---     ---     ---     ---     ---     ---     ---     ---     ---
PKBG:     1.00    1.03     .97    1.02     ---     ---     ---     ---     ---     ---     ---     ---     ---

This was 400 seconds per point at 100 nA, 20 keV but note the statistically significant negative results for Cu (ignore the Au Ma results for a moment- I get to that story in a minute!).  Why are we getting 800 PPM negative for Cu?

So we ran a longer scan at 60 sec per point on our pyrite and saw this:



Look at the bgd interpolation- no wonder we are getting negative 800 PPM. But what the heck is this peak?  It's too wide for a normal emission peak I would think.  Any ideas?

Scanning wider so we can set the high side bgd properly we see this:



This gives us results consistent with AA and XRF as seen here:

Un   11 pyrite #730
TakeOff = 40.0  KiloVolt = 20.0  Beam Current = 100.  Beam Size =    0

Un   11 pyrite #730, Results in Elemental Weight Percents
 
ELEM:       Au      Au      Cu      Ag      Fe       S      Ni      Co      As      Zn      Pb      Cr      Ti
TYPE:     ANAL    ANAL    ANAL    ANAL    SPEC    SPEC    SPEC    SPEC    SPEC    SPEC    SPEC    SPEC    SPEC
BGDS:      LIN     LIN     LIN     LIN
TIME:   400.00  400.00  400.00  400.00     ---     ---     ---     ---     ---     ---     ---     ---     ---
BEAM:   100.49  100.49  100.49  100.49     ---     ---     ---     ---     ---     ---     ---     ---     ---

ELEM:       Au      Au      Cu      Ag      Fe       S      Ni      Co      As      Zn      Pb      Cr      Ti   SUM 
XRAY:     (la)    (ma)    (ka)    (la)      ()      ()      ()      ()      ()      ()      ()      ()      ()
   627   -.010    .016    .009    .009  46.550  53.450    .000    .000    .009    .009    .010    .000    .058 100.109
   628   -.009    .020    .014    .013  46.550  53.450    .000    .000    .009    .009    .010    .000    .058 100.124
   629    .001    .018    .012    .008  46.550  53.450    .000    .000    .009    .009    .010    .000    .058 100.125
   630    .004    .012    .009    .018  46.550  53.450    .000    .000    .009    .009    .010    .000    .058 100.130
   631    .005    .021    .009    .008  46.550  53.450    .000    .000    .009    .009    .010    .000    .058 100.130
   632    .003    .019    .010    .007  46.550  53.450    .000    .000    .009    .009    .010    .000    .058 100.125

AVER:    -.001    .017    .010    .011  46.550  53.450    .000    .000    .009    .009    .010    .000    .058 100.124
SDEV:     .007    .003    .002    .004    .000    .000    .000    .000    .000    .000    .000    .000    .000    .007
SERR:     .003    .001    .001    .002    .000    .000    .000    .000    .000    .000    .000    .000    .000
%RSD:  -796.37   19.68   17.11   41.90     .00     .00     .00     .00     .00     .00     .00     .00     .00
STDS:      579     579     529     547     ---     ---     ---     ---     ---     ---     ---     ---     ---

STKF:   1.0001  1.0000   .9975   .9920     ---     ---     ---     ---     ---     ---     ---     ---     ---
STCT:  15401.2  5782.1 13188.3  3764.1     ---     ---     ---     ---     ---     ---     ---     ---     ---

UNKF:    .0000   .0002   .0001   .0001     ---     ---     ---     ---     ---     ---     ---     ---     ---
UNCT:      -.1     1.0     1.2      .3     ---     ---     ---     ---     ---     ---     ---     ---     ---
UNBG:    166.9    11.7    34.2     8.1     ---     ---     ---     ---     ---     ---     ---     ---     ---

ZCOR:   1.4519  1.0365  1.1584  1.3372     ---     ---     ---     ---     ---     ---     ---     ---     ---
KRAW:    .0000   .0002   .0001   .0001     ---     ---     ---     ---     ---     ---     ---     ---     ---
PKBG:     1.00    1.08    1.03    1.04     ---     ---     ---     ---     ---     ---     ---     ---     ---

But what is this peak?  Does anyone else see it on their instrument?
The only stupid question is the one not asked!

Probeman

  • Emeritus
  • *****
  • Posts: 1672
  • Never sleeps...
    • John Donovan
Re: Trace Element Strategies
« Reply #8 on: November 07, 2017, 03:38:15 pm »
Now for the story of the elevated Au Ma result I mentioned above.

Because I also acquired a few other pure metals for an MAN bgd calibration I noticed this on the MAN calibration for the Au Ma on PET:



That is the thing about the MAN background calibration. Sometimes it tells you about trace contaminants in your standards, sometimes it tells you about subtle spectral interferences in your sample!  All useful stuff to know about. So then the class looked at our scan data for Au Ma on pyrite and noticed this:



This interference causes an additional 70 PPM of interference. We can't correct for this interference in this particular run because we didn't measure Fe, just the traces, but by simply removing the Fe containing standards from the MAN fit, we get a much nicer MAN curve for Au Ma as seen here:



Remember the "rule of thumb": because the background is generally defined as the lowest thing we can measure (aside from some nasty "holes" in the continuum!), standards that lie above the general MAN trend can be assumed to be either trace contamination or spectral interferences. So removing the high outliers is usually a safe bet.
The only stupid question is the one not asked!

BenjaminWade

  • Professor
  • ****
  • Posts: 148
Re: Trace Element Strategies
« Reply #9 on: November 07, 2017, 08:07:48 pm »
Hi John
That is very interesting...your xtals seem to be like swiss cheese! I have done some low level Au in pyrite work using La on LLIF, and likewise I definitely have the "hole" right next to the Au La peak which I see to varying levels no matter what standard I do the wavescan on. I have been getting around it by either putting the Lo background position further out and hoping that the peak position is just outside of the deep dark hole on the shoulder, and also by doing the MAN fit. Au La on LLIF has a very nice MAN fit with no Fe interference problems.

With regards to the peak in pyrite, is your standard a synthetic or natural pyrite? I agree the peak is quite wide and weird. Pyrite can contain a whole heap of elements at trace to 1000's ppm level (Zn, Tl, Cu, Pb, Ag, As, Mo, Ni, Co, Se, Bi...). If it was an element causing that peak I would be checking out Bi which can occur in quite high concentration in some pyrite. It usually coincides with an increase in Pb as well. I am fairly certain my Astimex marcasite "standard" contains over 1000ppm Bi which isn't reported! Not sure if that is enough to create the second order Bi interferences which are around that position though.

With regards to your peak in Ti metal, given the starting materials are usually ilmenite, rutile, leucoxene etc, you would be looking for elements that occur naturally in those minerals I guess. But looking at the "All Elements" selected there on your wavescan really the only elements that coincide with that peak that might make sense are perhaps Nb or Ta. It could be another one of those mystery peaks like on PC0...

Cheers