Author Topic: Using the blank correction for major elements  (Read 186 times)

Probeman

  • Emeritus
  • *****
  • Posts: 1883
  • Never sleeps...
    • John Donovan
Using the blank correction for major elements
« on: April 05, 2019, 10:34:58 am »
This is going to sound strange, but let's discuss how the blank correction feature in Probe for EPMA *might* be utilized in certain cases to improve the accuracy of major elements. Yes, the blank correction was originally introduced to correct for continuum artifacts in order to improve trace element accuracy, both for off-peak measurements as discussed here:

https://probesoftware.com/smf/index.php?topic=29.msg387#msg387

Full paper here:

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

and also for MAN background acquisitions as described here:

https://probesoftware.com/smf/index.php?topic=307.msg5427#msg5427

Full paper here:

https://epmalab.uoregon.edu/publ/A%20new%20EPMA%20method%20for%20fast%20trace%20element%20analysis%20in%20simple%20matrices.pdf

I first had the idea of using the blank correction for major elements when analyzing oxygen to determine water in hydrous glasses (from Withers) using the method first proposed by Barbara Nash where one quantitatively measures oxygen and also all the cations, then one calculates the oxygen expected from these cations and then subtracts them from the measured oxygen, to obtain the actual excess or deficit oxygen.

The difficulty in this case being the accuracy of the oxygen measurement. To test this method I used MgO as my oxygen standard, and then applied Area Peak Factors (APFs) from Bastin, empirically determined mass absorption corrections and also TDI curves. Read the white paper here for details:

https://pages.uoregon.edu/epmalab/reports/Withers%20hydrous%20glass.pdf

But the bottom line was when testing the oxygen accuracy on the NIST mineral glasses, I found that I was still off by some 1% absolute or about 2% relative, which gave somewhat reasonable results on the Wither's glasses, but just wasn't good enough. Hence the application of the blank correction to improve the accuracy of a major element, in this case oxygen, by applying the NIST glass as a blank sample which also brings me to the strange subject of this topic: using the blank correction for major elements (yes, one could have simply utilized the NIST mineral glass a the primary standard for oxygen (and TDI curves), but what would be the fun in that?).   :)

It turns out that the blank correction equation as shown here:



is designed to handle not only blank samples that are "true" blanks, that is they contain a zero concentration of the element in question, but also blank samples that contain non-zero concentrations of the element of interest. E.g., my synthetic SiO2 which contains 1.4 PPM Ti.

So in principle the blank sample could be any concentration of the element of interest, so long as it is an *accurate* concentration (note the blank sample will usually be a standard, but it must be measured as an unknown to be applied as a blank, but that's another discussion).

So let's look at another example I recently ran into in the lab.  In this case we were analyzing a Mg-Li alloy, where Li (of course) had to be measured by difference. The samples were beautifully polished even though they were very soft (approaching pure Mg), but I cannot say the same for our standard block which contained pure Mg metal, but even after being freshly polished, it had a bit of an "orange-peel" surface. Most likely because it was polished along with all the other standard materials causing a bit of cross contamination (e.g., the infamous NIST Au-Cu-Ag sample polishing issues). The standard block was also carbon coated, while the unknown sample was not carbon coated. However we did specify a carbon coat correction for the standards, but not for the unknowns, as described here:

https://probesoftware.com/smf/index.php?topic=23.0

Anyway, after all that we obtained the following analysis for the Mg-Li alloy as seen here:

Element Li is Calculated by Difference from 100%

No Sample Coating and/or No Sample Coating Correction

Un    2 Mg-Li, 630C, Results in Elemental Weight Percents
 
ELEM:       Mg      Mn       O      Mo      Si      Li
TYPE:     ANAL    ANAL    ANAL    ANAL    ANAL    DIFF
BGDS:      EXP     LIN     EXP     LIN     LIN
TIME:    25.00   25.00   25.00   25.00   25.00     ---
BEAM:    27.54   27.54   27.54   27.54   27.54     ---

ELEM:       Mg      Mn       O      Mo      Si      Li   SUM 
...
   340  99.375    .023    .435   -.007    .160    .014 100.000
   341  99.542    .013    .515    .000    .102   -.173 100.000
   342  99.758    .021    .499    .009    .147   -.433 100.000
   343  99.582    .017    .495   -.009    .106   -.192 100.000
   344  99.825    .014    .513    .009    .138   -.499 100.000
   345  99.800    .020    .536   -.042    .149   -.463 100.000
   346  99.934    .018    .491   -.003    .084   -.524 100.000
   347  99.682    .016    .362    .009    .049   -.118 100.000
   348  99.413    .028    .610    .015    .166   -.230 100.000
   349  99.907    .043    .536    .012    .117   -.615 100.000
   350  99.849    .028    .594   -.023    .191   -.640 100.000
   351  99.532    .015    .580   -.003    .146   -.269 100.000
   352  99.641    .023    .545   -.042    .183   -.350 100.000

AVER:   98.054    .018    .634   -.003    .142   1.154 100.000
SDEV:    1.945    .010    .346    .016    .054   1.724    .000
SERR:     .106    .001    .019    .001    .003    .094
%RSD:     1.98   53.41   54.50 -611.61   37.77  149.42
STDS:      512     525     913     542     514     ---

STKF:    .9984  1.0000   .2509   .9910  1.0000     ---
STCT:  44026.0 20271.5  8110.8  8847.3  4129.0     ---

UNKF:    .9776   .0002   .0029   .0000   .0008     ---
UNCT:  44134.6     3.2   103.1     -.2     3.3     ---
UNBG:     35.7    17.9    56.5     5.3      .8     ---

ZCOR:   1.0030  1.1951  2.1997  1.4395  1.8420     ---
KRAW:   1.0024   .0002   .0127   .0000   .0008     ---
PKBG:  1242.53    1.18    2.80    1.00    6.17     ---
INT%:     ----    ----    -.01    ----    ----     ---

The averages don't look right because I deleted most of the (300+) analysis lines to save space.

It's not an awful analysis, but the Li by difference numbers do go slightly negative, and I suspected that this might be due to the Mg standard being surface contaminated during polishing by the other materials in the standard mount.  Let's continue in a bit in the next post.
« Last Edit: April 05, 2019, 11:50:02 am by Probeman »
The only stupid question is the one not asked!

Probeman

  • Emeritus
  • *****
  • Posts: 1883
  • Never sleeps...
    • John Donovan
Re: Using the blank correction for major elements
« Reply #1 on: April 05, 2019, 11:49:36 am »
So to continue with this strange idea of possibly using the blank correction for improving the accuracy of major elements, in certain cases where it *might* be warranted, here is an analysis of the edge of the Mg-Li alloy which is ostensibly pure Mg:

Element Li is Calculated by Difference from 100%

No Sample Coating and/or No Sample Coating Correction

Un    3 edge of Mg-Li, Results in Elemental Weight Percents
 
ELEM:       Mg      Mn       O      Mo      Si      Li
TYPE:     ANAL    ANAL    ANAL    ANAL    ANAL    DIFF
BGDS:      EXP     LIN     EXP     LIN     LIN
TIME:    25.00   25.00   25.00   25.00   25.00     ---
BEAM:    27.48   27.48   27.48   27.48   27.48     ---

ELEM:       Mg      Mn       O      Mo      Si      Li   SUM 
   353 100.819    .027    .442   -.001    .067  -1.353 100.000
   354 100.573    .002    .512   -.019    .053  -1.121 100.000
   355 100.593    .008    .465   -.015    .080  -1.132 100.000
   356 100.229    .008    .487    .000    .118   -.843 100.000
   357 100.239    .014    .539    .007    .093   -.892 100.000

AVER:  100.491    .012    .489   -.006    .082  -1.068 100.000
SDEV:     .253    .009    .038    .011    .025    .206    .000
SERR:     .113    .004    .017    .005    .011    .092
%RSD:      .25   78.41    7.84 -191.39   30.45  -19.29
STDS:      512     525     913     542     514     ---

STKF:    .9984  1.0000   .2509   .9910  1.0000     ---
STCT:  44253.5 20243.7  8099.2  8825.2  4078.1     ---

UNKF:   1.0004   .0001   .0022   .0000   .0004     ---
UNCT:  45392.4     2.1    78.1     -.3     1.8     ---
UNBG:     35.8    18.5    55.1     5.6      .7     ---

ZCOR:   1.0045  1.1960  2.2249  1.4494  1.8642     ---
KRAW:   1.0257   .0001   .0096   .0000   .0004     ---
PKBG:  1274.62    1.12    2.42     .95    4.17     ---
INT%:     ----    ----     .00    ----    ----     ---

Here the Li by difference values are significantly negative leading me to believe that the Mg primary standard is more oxidized, and/or contaminated with other elements from polishing a soft material with other materials even though the standard block was recently polished and immediately carbon coated.  Edit: and in looking at the Mg standard I am seeing about 0.9 wt% oxygen, so it is apparently more oxidized, probably from the polishing.

Meanwhile looking at the above unknown Mg alloy sample analysis at the edge of the sample, there appears to be about 0.5 wt% of oxygen, undoubtedly from surface oxidation also.  But apparently less than our standard Mg metal.

By the way this surface oxidation issue is the reason I am still trying to find a vendor that can install an in situ cleaner inside the microprobe instrument that would allow the user to clean oxidation layers just prior to analysis, ideally just a small (~1 sq. mm) area or less to obtain better analyses of air sensitive materials as described here (especially at low beam energies!):

https://probesoftware.com/smf/index.php?topic=1073.0

Anyway, after subtracting 0.5% oxygen from 100%, if we now specify this supposedly pure Mg sample as a pure, 99.5 wt%, Mg sample and assign it as the blank correction sample in our traverse data we now obtain these results:

Element Li is Calculated by Difference from 100%

No Sample Coating and/or No Sample Coating Correction

Un    2 Mg-Li, 630C, Results in Elemental Weight Percents
 
ELEM:       Mg      Mn       O      Mo      Si      Li
TYPE:     ANAL    ANAL    ANAL    ANAL    ANAL    DIFF
BGDS:      EXP     LIN     EXP     LIN     LIN
TIME:    25.00   25.00   25.00   25.00   25.00     ---
BEAM:    27.54   27.54   27.54   27.54   27.54     ---

ELEM:       Mg      Mn       O      Mo      Si      Li   SUM 
   340  98.276    .023    .433   -.007    .159   1.115 100.000
   341  98.443    .013    .512    .000    .102    .929 100.000
   342  98.659    .021    .496    .009    .146    .668 100.000
   343  98.483    .017    .493   -.009    .106    .910 100.000
   344  98.727    .014    .510    .009    .137    .603 100.000
   345  98.701    .020    .533   -.042    .149    .639 100.000
   346  98.835    .018    .489   -.003    .083    .577 100.000
   347  98.586    .016    .360    .009    .049    .979 100.000
   348  98.314    .028    .606    .015    .165    .872 100.000
   349  98.809    .043    .534    .012    .116    .487 100.000
   350  98.751    .028    .591   -.023    .190    .463 100.000
   351  98.434    .015    .577   -.003    .145    .833 100.000
   352  98.542    .023    .542   -.042    .182    .752 100.000

AVER:   96.956    .018    .631   -.003    .142   2.255 100.000
SDEV:    1.944    .010    .344    .015    .053   1.724    .000
SERR:     .106    .001    .019    .001    .003    .094
%RSD:     2.01   53.41   54.48 -611.62   37.77   76.43
STDS:      512     525     913     542     514     ---

STKF:    .9984  1.0000   .2509   .9910  1.0000     ---
STCT:  44026.0 20271.5  8110.8  8847.3  4129.0     ---

UNKF:    .9680   .0002   .0029   .0000   .0008     ---
UNCT:  43698.5     3.2   103.1     -.2     3.3     ---
UNBG:     35.7    17.9    56.5     5.3      .8     ---

ZCOR:   1.0016  1.1946  2.1877  1.4347  1.8317     ---
KRAW:    .9925   .0002   .0127   .0000   .0008     ---
PKBG:  1230.26    1.18    2.80    1.00    6.17     ---
INT%:     ----    ----    -.01    ----    ----     ---
BLNK#:       3    ----    ----    ----    ----     ---
BLNKL: 99.5000    ----    ----    ----    ----     ---
BLNKV: 100.491    ----    ----    ----    ----     ---

So that seems more reasonable, but am I just kidding myself or what?  Please feel free to comment.
john
« Last Edit: April 05, 2019, 12:42:10 pm by Probeman »
The only stupid question is the one not asked!