I wanted to show the interference corrections in CalcImage so I picked an Fe and FeCu sulfide and so tried to measure Mo in it. As you know, there is a large interference of S ka on Mo La. With a quant analysis of the two minerals we get (15 keV, 100nA, 40 seconds):
Un 5 Reed sulfide, Results in Elemental Weight Percents
ELEM: Fe S Cu Mo Zn Ag Ca SUM
132 45.536 53.359 .039 1.862 .019 .001 .002 100.817
133 45.839 53.387 .082 1.867 .021 .007 .007 101.210
134 45.604 53.253 .091 1.887 .022 .018 .007 100.881
135 45.501 53.360 .034 1.850 .005 -.003 .008 100.754
136 45.604 53.277 .055 1.868 .028 .000 -.019 100.814
AVER: 45.617 53.327 .060 1.867 .019 .004 .001 100.895
SDEV: .132 .059 .025 .013 .009 .008 .011 .182
Un 6 Reed sulfide (CuFe), Results in Elemental Weight Percents
ELEM: Fe S Cu Mo Zn Ag Ca SUM
137 29.300 34.383 33.779 1.231 .036 .048 .003 98.779
138 29.184 34.471 33.914 1.244 .004 .068 -.025 98.861
139 29.176 34.335 33.706 1.241 .030 .073 .002 98.563
140 29.306 34.437 33.654 1.230 .029 .051 .015 98.722
141 29.327 34.501 33.770 1.242 .021 .063 -.005 98.919
AVER: 29.259 34.425 33.764 1.238 .024 .061 -.002 98.769
SDEV: .073 .067 .098 .007 .012 .011 .015 .137
And here are the two minerals again, but this time *with* the interference correction of S on Mo (and also Cu on Zn though it is very small):
Un 5 Reed sulfide, Results in Elemental Weight Percents
ELEM: Fe S Cu Mo Zn Ag Ca
BGDS: LIN LIN LIN LIN EXP LIN LIN
TIME: 40.00 80.00 40.00 40.00 80.00 40.00 40.00
BEAM: 99.98 99.98 99.98 99.98 99.98 99.98 99.98
ELEM: Fe S Cu Mo Zn Ag Ca SUM
132 45.661 53.479 .039 -.032 .019 .001 .002 99.169
133 45.964 53.507 .083 -.027 .021 .007 .007 99.561
134 45.729 53.373 .091 -.002 .022 .017 .007 99.237
135 45.626 53.480 .034 -.042 .005 -.003 .008 99.107
136 45.728 53.397 .056 -.020 .028 .000 -.019 99.170
AVER: 45.742 53.447 .061 -.025 .019 .004 .001 99.249
SDEV: .132 .059 .026 .015 .009 .008 .011 .181
Un 6 Reed sulfide (CuFe), Results in Elemental Weight Percents
ELEM: Fe S Cu Mo Zn Ag Ca
BGDS: LIN LIN LIN LIN EXP LIN LIN
TIME: 40.00 80.00 40.00 40.00 80.00 40.00 40.00
BEAM: 100.02 100.02 100.02 100.02 100.02 100.02 100.02
ELEM: Fe S Cu Mo Zn Ag Ca SUM
137 29.318 34.451 33.847 .012 .020 .048 .003 97.699
138 29.202 34.539 33.983 .023 -.013 .068 -.025 97.778
139 29.194 34.404 33.774 .025 .013 .073 .002 97.485
140 29.325 34.506 33.722 .010 .013 .051 .015 97.642
141 29.346 34.570 33.838 .020 .004 .063 -.005 97.837
AVER: 29.277 34.494 33.833 .018 .007 .061 -.002 97.688
SDEV: .073 .067 .098 .007 .013 .011 .015 .136
I don't like that the interference corrected Mo value on the FeS2 is -0.025 (negative 250 PPM), but it is within less than 2 standard deviations of zero so maybe it's just bad luck... anyway we'll never see that in the x-ray maps.
Still I don't like it and will need to think about what the problem might be... the run was very stable. Standards were automatically run before the unknowns, after the unknowns (and before the x-ray maps) and again, after the x-ray maps as seen here:
Drift array standard intensities (cps/30nA) (background corrected):
ELMXRY: fe ka s ka cu ka mo la zn ka ag la ca ka
MOTCRY: 5 LIF 4 PET 5 LIF 2 LPET 3 LLIF 2 LPET 1 PET
STDASS: 730 730 529 542 530 547 358
STDVIR: 0 0 0 0 0 0 0
2824.4 3845.3 4795.7 8575.2 14586.0 8639.9 1056.7
2822.9 3837.4 4793.2 8587.0 14596.7 8629.6 1059.3
2826.8 3845.7 4792.8 8591.5 14554.8 8643.7 1057.0
I suspect because the interference standard for S on Mo (my standard FeS2), was run at 30 nA, but the unknown was run at 100 nA, there could be an effect from sample damage.
More unfortunately, I decided to run a beam scan rather than a stage scan, and even at 2900x, there is significant Bragg defocussing in the major elements. Also, I forgot and ran the maps at 30nA- the same as my standards, though I meant to run the maps at 100 nA. So I'll have to try again next weekend.
In the meantime here are the Mo and S maps *without* an interference correction:
(https://smf.probesoftware.com/gallery/395_11_04_16_2_25_43.jpeg)
Note the large Bragg defocussing on the sulfur map. This causes a problem with the quantitative interference correction for Mo, though it does try hard as seen here in the interference corrected maps:
(https://smf.probesoftware.com/gallery/395_11_04_16_2_26_12.jpeg)
The interference corrected Mo concentrations are quite close to zero (yellow-green) for much of the pyrite and Cu-Pyrite grains, except in the lower left corner where most of the Bragg defocusing occurs. I'll try again with a stage scan next weekend at 100nA. :D
Here's an interesting aspect to the quantification of maps with Bragg defocussing.
In the *total* map, the Bragg defocussing from the various spectrometers almost, kinda averages out and forms a high spot of 100% or so, in the middle of the map.
(https://smf.probesoftware.com/gallery/395_11_04_16_2_58_57.jpeg)
Exactly as one might expect!
I made an ImageJ stack of the Zn Ka corrected and uncorrected for the Cu Ka/Kb interference and it is barely visible as seen here in the attachment below. Click the attachment to see the animation.
As expected, the FeCu sulfide shows more Zn from the Cu interference than the Fe sulfide. When I run again at 100 nA the difference in the interference correction should be more visible.
The stage scan is running now... hopefully I will have interference corrected quant map results posted here early next week. :)
In the meantime, please note that this quantitative spectral interference correction is a quick and easy, two or three mouse click operation, once the data is acquired as seen here:
(https://smf.probesoftware.com/gallery/395_16_04_16_8_25_11.png)
Simply click the interfering element (S), the standard used for the S Ka interference correction (pyrite) and OK. The Calculate Interferences button is an optional mouse click, which is used only to see if the suspected interference is nominally possible.
OK, the stage scan finished and here are the results. Again standards (FeS2 for sulfur, diopside for Ca, pure metals for the other elements), were run at 15 keV, 30 nA and the stage scan at 15 keV, 100nA, 1 um pixels. All data was automatically corrected for deadtime, beam drift, standard drift, background, and matrix corrected.
First without any interference correction:
(https://smf.probesoftware.com/gallery/395_17_04_16_7_53_47.jpeg)
Without an interference correction the pyrite shows roughly 1.8 wt% Mo and the Cu-pyrite shows roughly 1.2 wt% Mo. Now with the quant interference correction as described in the above post:
(https://smf.probesoftware.com/gallery/395_17_04_16_7_56_12.jpeg)
We are now seeing almost no Mo in either phase, however, to the eye, there does appear to be some small non-zero amount of Mo in the Cu pyrite phase on the right. Let's do a polygon extraction using CalcImage, first the Fe pyrite phase:
(https://smf.probesoftware.com/gallery/395_17_04_16_7_57_51.jpeg)
We get 0 +/- 0.09 wt.% Mo. Can't do much better than that if the concentration is zero! In the Cu pyrite we get this:
(https://smf.probesoftware.com/gallery/395_17_04_16_7_59_39.jpeg)
The Cu pyrite phase shows 0.04 +/- 0.08 which is statistically zero, but the eye confirms that there does appear to be some very small concentration of Mo- which is confirmed by the point analyses in the above posts which gave ~200 PPM Mo in the Cu pyrite phase.
Attached below (you'll need to log in to see them) are the quant and detection limit maps for Mo.
Note that although the 0.04 +/- 0.08 Mo wt. % in the polygon extraction of the Cu pyrite phase is below the Mo 3 sigma detection limit of 700 to 800 PPM, the eye picks up the subtle difference in the Mo quant map attached below:
To see fine details in the quantitative x-ray map interference correction I performed a cross section on the Mo quant pixel data using the CalcImage slice menu as seen here:
http://smf.probesoftware.com/index.php?topic=41.msg541#msg541
and I obtained this output *without* an interference correction of sulfur on Mo La:
(https://smf.probesoftware.com/gallery/395_16_08_16_10_06_39.jpeg)
and this output *with* an interference correction for sulfur on Mo La:
(https://smf.probesoftware.com/gallery/395_16_08_16_10_06_55.jpeg)
The pyrite (high sulfur) phase is on the left, the CuFe (lower sulfur) is on the right. The neat thing is, as described by Ben Buse here:
http://smf.probesoftware.com/index.php?topic=777.msg4823#msg4823
Is that the eye can (sort of) detect a (statistically insignificant) difference in the Mo content in the interference corrected map for the two phases even though the cross section plot does not (seemingly) show it!
How cool is that? 8)
By the way, in a previous set of point analyses (15 keV, 100 nA, 5 um, 40 sec on peak and 40 sec off peak, average of 5 points), as seen in the first post above, I got -0.025 +/- 0.015 Mo in the Fe sulfide phase (that is zero within 2 std devs), and in the CuFe sulfide I got 0.018 (180 PPM) +/- 0.007 (70 PPM) of Mo.
All analyses performed *without* a blank correction.
john
This is pretty cool. I wonder if averaging out over a line that is more than one pixel wide would show what the eye can see. I use the interference correction frequently for Ba and Ce, without it the maps certainly look very different ;D
Quote from: Gareth D Hatton on August 17, 2016, 12:48:05 AM
This is pretty cool. I wonder if averaging out over a line that is more than one pixel wide would show what the eye can see. I use the interference correction frequently for Ba and Ce, without it the maps certainly look very different ;D
Hi Gareth,
Yes, that is possible. I haven't tried that in Surfer yet. Can the "boundary" line in Surfer be specified to be more than one pixel wide?
john
Hello All,
Drew from Golden Software here. The profile tool and the Grid | Slice command will only slice the grids by one node; you cannot make this wider. I have added a request to our feature request database to have control over the profile slice where you would be able to specify a wider path than 1 grid node.
On a side note and technically speaking, the profile tool calculates the profile based on the grid nodes that the transect crosses and not pixels or grid cells. This might seem like the same thing, but there is actually a difference on where the value of the grids is located. In a Surfer grid, the values are located on the gird nodes and not the cell centroid as with a raster dataset. This is a good thing to conceptualize when working with both types of files.
I hope this helps!
Drew
I have added to this post one of the key uses of interfernce correction for us. I have only included the Ce map as it clearly shows the difference. Before correction we can see a phase, which is rich in Ba, but appears to contain around 20 wt%Ce. After correction you can clearly see that there is no Ce in this region.
I have also seen the classic interference of Ti on V doing point analysis on a TiAl6V4 alloy
| Al | Ti | V | Total |
| Before | 6.6 | 90.3 | 10.4 | 107.2 |
| After | 6.5 | 90.4 | 4 | 100.9 |
Edit by John: Remember to login to see attachments!
Quote from: Gareth D Hatton on August 24, 2016, 03:17:18 AM
I have also seen the classic interference of Ti on V doing point analysis on a TiAl6V4 alloy
| Al | Ti | V | Total |
| Before | 6.6 | 90.3 | 10.4 | 107.2 |
| After | 6.5 | 90.4 | 4 | 100.9 |
Very cool interference correction map!
On the TiAlV alloy you can also have a "cascade" interference when analyzing trace Cr in this matrix. That is Ti (Kb) interferes with V, and V (Kb) interferes with Cr! Nasty, but fortunately PFE handles this type of interference automatically. This is described on page 26 of my now ancient interference correction paper attached below. See table 1, NIST standard SRM 654b.
john