Sulfur Peak Shifts

[Probeman]

This topic is to discuss sulfur peak shift issues in glasses and minerals.

I'll start by posting a short technical note on using different sulfur standards as the standard for VG2 glass. The summary is here:

Analysis of sulfur in VG-2 glass

The following analyses demonstrate the effect of trace sulfur analyses using different primary standards. So long as the sulfur standard was peaked for each standard (to account for peak shift) the analysis of the VG-2 glass was essentially consistent.

The only problem observed was that the composition for the Pyrrhotite standard was originally entered nominally as FeS. Once the actual Fe-S ratio was entered into the standard database, this standard gave consistent analyses with the other standards.

Conclusion: It doesn't matter what standard one uses for sulfur analyses provided that the analyst is measuring the peak intensity AT THE ACTUAL PEAK POSITIONS for both the unknown and standard.

VG-2 glass S analysis (each standard peaked individually)- NOTE, this sulfur in glass measurement is not intended to be accurate, only as a basis for understanding the change in composition as a function of peak position and standard.

Standard    S (weight percent, analyses at 15 kev, 10 sec)
Anhydrite:    .14452
Pyrite        .14470
Pyrrhotite   .13285 (assumed 50/50 stoichiometry)
Pyrrhotite   .14968 (Fe(1-x)S, where x=0.17) (from web database)
Pyrrhotite   .14472 (Fe(1-x)S, where x=0.13) (from EPMA Fe-S analysis of actual standard)

The full document is attached below.

[Spratt NHM]

I am currently looking to publish work on mineral that is a thiosulphate and have problems in that I constantly get low values for S. This is compounded by the fact that Pb and Mo are also present. In the case of thiosulphate the sulphur is present in two oxidation states. I have taken a large number of measurements on standards also of differing oxidation states and have noted the differing peak positions for S ka.  There is also the possibility that the Mo is substituting in the thiosulphate for the S6+. In the case of thiosulphate does this produce two peaks (one from each valence state of S) which would produce a wide peak of lower intensity or a single peak of a hybridised energy with a value of X? I have not found much in the way of suitable thiosulphate standards (Sodium thiosulphate - way too much water for vacuum, Pb thiosulphate - too fine grained in supplied material, and Ba thiosulphate which I have not managed too source yet but could be explosive!)

[jon_wade]

Just out of interest how do you asses the stoichometry with respect to oxygen and the possibility of multiple S valences?

Are totals low by a lot?

[Probeman]

Just out of interest how do you asses the stoichometry with respect to oxygen and the possibility of multiple S valences?

Are totals low by a lot?

Jon: did you just call us all "asses"?   

Seriously I just noticed this question from last year, but I do not know how to answer your question.  Maybe you can provide an example?

john

[Doug_Meier]

From a first approximation, I propose that one can make the following estimates of sulfur oxidation state in inorganic compounds:

sulfide: mostly ionic and by itself. S 2- The S Ka (2p-->1s) transitions are a single peak (at least within our resolution). Filled 3(s,p) increases shielding of 2p, shifts Ka to (slightly) higher energy.

sulfur: forms a covalently bound 8-member ring in nature. Still S 0, and a single peak.

sulfate: covalently bound [SO4]2- ion. Since oxygen's electronegativity is a bit higher than sulfur's, the molecular orbitals won't be distributed evenly across all elements in the ion. Thus one expects O >1- (ranging from 1.1 to 1.3 depending upon how one sets up the calculation and what else is in the local environment) and S >2+ (again, in a range from 2.3 to 3.3 depending upon how one sets up the calculation) as computed from a Mulliken population analysis. The sulfur Ka transition should still be a single peak, shifted from sulfide or sulfur to lower energy due to the reduced shielding of the 2p electrons by 3(s,p) electrons that are now in more remote molecular orbitals.

thiosulfate: covalently bound [SO3S]2- ion. Typed that way to highlight that it is structurally similar to the sulfate ion, but one sulfur replaces one oxygen in the structure. So now there is one sulfur atom at essentially S 1- (I haven't done this calculation; perhaps it is even a little less), the oxygen are still at O >1-, and the central sulfur is at S>2+ (though I suspect not as high as the sulfate S in an otherwise equivalent environment). Since the 2p position in the S energy diagram for each atom is shielded differently by their different local molecular orbital environments, I would expect to see a broader S Ka line that, in a perfect world, could be deconvolved into two separate lines that fall between the endpoints established by sulfide and sulfate. I've not tried this yet.

Finally, I would expect that no matter what the chemical state of the sulfur is, the yields would be affected only marginally, since none of the above discussion changes anything about the participating orbitals that results in a violation of selection rules.

[Probeman]

Hi Doug,
From the scans that I have run on pyrite and anhydrite I don't see any peak shape changes, just peak shift changes, but they could be subtle.

I've attached some scans I've run below (remember to login to see attachments!), looking at moving the stage vs. a stationary stage (to see the effect of sitting on one spot and possibly oxidizing the sulfur in basaltic glass).  Also comparing sulfur Ka and Kb peak shifts in various materials.

[Probeman]

Here's some sulfur Ka peak shifts:



and this one on a different spectrometer (both are PET crystals on my Sx50):

[Doug_Meier]

I'd only expect to notice a shape change in the Kb, since its participating orbitals are changing hybridization with chemistry, while the Ka orbitals only adjust their potential energy in response to shielding changes. This matches my observations, though admittedly I haven't looked at the Kb very closely. Quite possibly the change I expect is not resolvable.

The sulfate shift to higher energy versus sulfide is the opposite of my first guess, though, indicating that the core 1s changes more with shielding change than does the 2p. The difference is pretty subtle, though; roughly a volt or so?

[Probeman]

Hi Doug,
Here is the same plot but in keV space:



Looks like 2 eV or so.

[jon_wade]

Just out of interest how do you asses the stoichometry with respect to oxygen and the possibility of multiple S valences?

Are totals low by a lot?

Jon: did you just call us all "asses"?   

Seriously I just noticed this question from last year, but I do not know how to answer your question.  Maybe you can provide an example?

john

oops! being British, you can tell I didn't mean it - if I had it'd have been 'arses!'

In glasses made under reducing conditions (low oxygen fugacity), S can exist as S2- substituting for Oxygen.  Hence one might over estimate the amount of oxygen.  S in glasses is also  prone to beam induced shifts in apparent oxidation state, which I guess we all know, but this is something we also see on the beam line.

[Mike Spilde]

Here's couple of papers on the subject of sulfur valence measurements by EPMA. The papers are dated, but the method does work. We've used it successfully to measure sulfur valence in volcanic glasses.

[Probeman]

Hi Mike,
In fact I was very involved with the work in the AM79 paper, at UC Berkeley when Paul was a grad student with Ian and I ran the probe.

I hadn't seen the Brown paper.  Thanks for attaching it.
john

[Probeman]

I was curious if sodium thiosulfate would show a peak that was wider than pyrite or anhydrite due to its dual valence states of sulfur, so a student here at UofO, Alan Lerner, was able to mount up a sample and polish it fairly well (avoid water and even ethanol!).

We ran relatively gentle conditions (15 keV, 30 nA and a 10 um beam) and we stepped the stage 2 microns every 60 seconds during the scan.



At first I thought that the thiosulfate peak was a little wider than the anhydrite peak, but after I plotted the peaks normalized to each other, I don't think so:

[Probeman]

Here's a possibly interesting method for determining sulfur concentrations *and* the approximate oxidation states *at the same time* on a microprobe.

As many of you know, depending on the oxidation state of sulfur, the emission peak on a WDS spectrometer varies enough to make it necessary to accurately position the spectrometer on the actual emission peak position.  Basically there is roughly a 10% change in intensity between the pyrite and anhydrite sulfur peak positions. 

Determining the degree of peak shift in a basaltic glass when the sulfur concentrations are around 1000 PPM is difficult, partly because a high enough precision wavescan requires considerable time, causing sample damage, and also possible oxidation of the sulfur in the glass, which causes further peak shifting. Specifically we've found it necessary to acquire sulfur wavescans on our basaltic glasses for around an hour each, and even then we've found it necessary to have PFE increment the stage position a few microns every minute or so to avoid further oxidizing the sulfur.

The basic idea being that once we determine the actual emission peak position for a given glass sample, we can set the spectrometer to that position and collect our sulfur intensities at the peak position for accurate quantitative analysis, regardless of the sulfur primary standard. 

But earlier this week two graduate students, Dan Rasmussen and Michele Muth, came up with an idea for determining the sulfur peak position by essentially creating a "multi-collector" microprobe for characterizing reduced to oxidized sulfur "species", by tuning each of our 5 PET spectrometers to cover the range of sulfur oxidation peak positions.

So we first tuned all the spectrometers to our pyrite primary standard (-1 valance),  spectrometer offset equals 0, and knowing that the pyrrhotite sulfur peak (-2 valence) is shifted to the right +4 units (in Cameca units), and anhydrite is shifted to the left by 30 units, we adjusted our 5 WDS spectrometers as seen here:

    +4    0    -10    -20    -30

Spectrometer 1 being the spectrometer tuned to the pyrrhotite sulfur peak, spectrometer 2 tuned to the pyrite peak and spectrometer 5 tuned to the anhydrite sulfur peak position, with spectrometers 3 and 4 tuned to intermediate peaks positions between pyrite and anhydrite.

The idea being that the spectrometer that produces the highest sulfur concentration will therefore probably be the correct concentration, and also give us some information on what the oxidation state of the sulfur is, depending on which spectrometer it is.  Furthermore, instead of counting an hour or so for a high precision wavescan, we only need to count for a few hundred second to get excellent sensitivity for our ~1000 PPM sulfur measurements!

When we then analyzed our ND70 SIMS glass standard for all 5 spectrometers, which reportedly has a sulfur concentration of 900 PPM, we obtained the following results:

Un   24 ND70, Results in Elemental Weight Percents

SPEC:       Si      Al      Fe      Mg      Ca      Na       K      Ti       P      Mn       O       F
TYPE:     SPEC    SPEC    SPEC    SPEC    SPEC    SPEC    SPEC    SPEC    SPEC    SPEC    SPEC    SPEC

AVER:     .000    .000    .000    .000    .000    .000    .000    .000    .000    .000    .000    .000
SDEV:     .000    .000    .000    .000    .000    .000    .000    .000    .000    .000    .000    .000
 
ELEM:        S       S       S       S       S
BGDS:      LIN     LIN     LIN     LIN     LIN
TIME:   160.00  160.00  160.00  160.00  160.00
BEAM:    49.62   49.62   49.62   49.62   49.62

ELEM:        S       S       S       S       S   SUM 
XRAY:     (ka)    (ka)    (ka)    (ka)    (ka)
   492    .072    .073    .069    .077    .073    .364
   493    .069    .075    .073    .070    .074    .362
   494    .071    .076    .068    .068    .064    .347
   495    .072    .072    .070    .062    .071    .347
   496    .070    .074    .070    .069    .072    .354
   497    .071    .072    .070    .064    .066    .344
   498    .073    .070    .069    .065    .074    .352
   499    .073    .076    .072    .069    .069    .358
   500    .073    .069    .070    .067    .062    .341
   501    .071    .071    .072    .074    .073    .360

AVER:     .071    .073    .070    .068    .070    .353
SDEV:     .001    .002    .002    .004    .004    .008
SERR:     .000    .001    .001    .001    .001
%RSD:     2.00    3.28    2.40    6.43    6.12
STDS:      730     730     730     730     730

STKF:    .5044   .5044   .5044   .5044   .5044
STCT:   152.32  451.70  499.97  127.50  166.44

UNKF:    .0007   .0007   .0007   .0007   .0007
UNCT:      .22     .65     .70     .17     .23
UNBG:      .10     .18     .23     .08     .11

ZCOR:   1.0000  1.0000  1.0000  1.0000  1.0000
KRAW:    .0014   .0014   .0014   .0014   .0014
PKBG:     3.23    4.58    4.08    3.24    3.09

Remember spectrometers 1, 3, 4, and 5 were detuned from the pyrite peak position to accomodate the range of sulfur oxidation states from pyrrhotite to anhydrite.

Anyway, none of the spectrometers gave us a concentration close to 900 PPM, but of course we analyzed sulfur on all 5 spectrometers so 99%+ of the matrix is missing (that's why the ZCOR is 1.0000 because the software thinks this is a pure sulfur sample, with a total of .353 weight percent!

So let's specify a nominal basaltic glass composition, so the matrix correction can perform its physics magic, and now we obtain the following results:

Un   24 ND70, Results in Elemental Weight Percents

SPEC:       Si      Al      Fe      Mg      Ca      Na       K      Ti       P      Mn       O       F
TYPE:     SPEC    SPEC    SPEC    SPEC    SPEC    SPEC    SPEC    SPEC    SPEC    SPEC    SPEC    SPEC

AVER:   23.750   7.441   9.203   4.046   7.947   1.944    .158   1.109    .087    .170  43.964    .045
SDEV:     .000    .000    .000    .000    .000    .000    .000    .000    .000    .000    .000    .000
 
ELEM:        S       S       S       S       S
BGDS:      LIN     LIN     LIN     LIN     LIN
TIME:   160.00  160.00  160.00  160.00  160.00
BEAM:    49.62   49.62   49.62   49.62   49.62

ELEM:        S       S       S       S       S   SUM 
XRAY:     (ka)    (ka)    (ka)    (ka)    (ka)
   492    .089    .090    .085    .095    .091 100.314
   493    .085    .093    .091    .087    .091 100.311
   494    .088    .093    .083    .084    .079 100.294
   495    .088    .089    .087    .077    .088 100.294
   496    .086    .091    .087    .085    .088 100.302
   497    .088    .089    .087    .079    .082 100.290
   498    .091    .086    .085    .080    .092 100.300
   499    .090    .094    .089    .085    .085 100.307
   500    .090    .086    .086    .082    .077 100.286
   501    .088    .087    .089    .091    .090 100.310

AVER:     .088    .090    .087    .084    .086 100.301
SDEV:     .002    .003    .002    .005    .005    .010
SERR:     .001    .001    .001    .002    .002
%RSD:     2.01    3.28    2.40    6.43    6.12
STDS:      730     730     730     730     730

STKF:    .5044   .5044   .5044   .5044   .5044
STCT:   152.32  451.70  499.97  127.50  166.44

UNKF:    .0007   .0007   .0007   .0007   .0007
UNCT:      .22     .65     .70     .17     .23
UNBG:      .10     .18     .23     .08     .11

ZCOR:   1.2340  1.2340  1.2340  1.2340  1.2340
KRAW:    .0014   .0014   .0014   .0014   .0014
PKBG:     3.23    4.58    4.08    3.24    3.09

Notice two things, first our ZCOR values for sulfur Ka are now 1.2340 (as opposed to 1.0000), and spectrometer 2, which was tuned to the pyrite peak position (spectrometer offset equals 0), now gives us 900 PPM *and* this is the highest concentration measured, so we might assume that this is the correct spectrometer position for this sample's oxidation state.

Does this mean that SIMS standard glass ND70 is quite reduced with a valence similar to pyrite?  I do not know (has anyone out there done EXAFS or XANES work on this glass standard?), but I will say these results are intriguing and maybe there is some value is setting up one's microprobe as a sulfur "multi-collector" using all 5 spectrometers.   

[Probeman]

Just to follow up on the sulfur 5 WDS spectrometer "multi-collector" idea for determining trace sulfur concentrations and approximate charge states, the Cameca spectrometer offsets when spectrometer 2 is peaked on pyrite are:

    SP1         SP2          SP3           SP4            SP5   (all PET or LPET crystals)
    +4            0            -10            -20            -30

Which corresponds to these charge states:

     -2           -1              ?               ?              +6

Or these minerals:

  pyrrhotite  pyrite          ?               ?         anhydrite

The thing I don't understand is why don't we see two convolved peaks when looking at partially oxidized sulfur in a glass?  Shouldn't there be a mixture of reduced and oxidized sulfur atoms if the emission peak shows up between pyrite and anhydrite? Or is it because it's a glass and therefore the charge states are averaged over many atoms to yield a single intermediate charge state which shows as a single emission peak?