I want to report the magnetite data comparison for off-peak and MAN that I ran last night. I'll be posting further results regarding sensitivity in the MAN Precision and Accuracy topic here:
http://smf.probesoftware.com/index.php?topic=307.msg3190#msg3190
once I finish coding Jared Singer's MAN regression sensitivity equations, but for now let's just look at the magnetite sample Ti and Mn standard deviations for the calculated concentrations starting with Ti Ka calculated using off-peak backgrounds:
ELEM: Ti Ti Ti Ti Ti Fe Mn Al Si Mg O SUM
XRAY: (ka) (ka) (ka) (ka) (ka) () () () () () ()
137 .031 .015 .027 .019 .017 72.080 .054 .201 .000 .072 27.803 100.318
138 .038 .028 .034 .031 .061 72.080 .054 .201 .000 .072 27.803 100.401
139 .022 .012 .022 .022 .005 72.080 .054 .201 .000 .072 27.803 100.293
140 .071 .066 .065 .068 .063 72.080 .054 .201 .000 .072 27.803 100.544
141 .077 .078 .076 .078 .095 72.080 .054 .201 .000 .072 27.803 100.614
142 .020 .010 .014 .021 .016 72.080 .054 .201 .000 .072 27.803 100.292
143 .027 .016 .026 .021 .013 72.080 .054 .201 .000 .072 27.803 100.314
144 .015 .016 .024 .005 .022 72.080 .054 .201 .000 .072 27.803 100.293
AVER: .038 .030 .036 .033 .037 72.080 .054 .201 .000 .072 27.803 100.384
SDEV: .024 .027 .022 .026 .032 .000 .000 .000 .000 .000 .000 .127
For now we will just focus on the variances since that is the best indication of data precision and consider accuracy later (I don't have an Fe3O4 blank standard sample in the lab- yet!). In rede above we can see the standard deviations for Ti off-peak measurements and below the same data but calculated for MAN backgrounds instead:
ELEM: Ti Ti Ti Ti Ti Fe Mn Al Si Mg O SUM
XRAY: (ka) (ka) (ka) (ka) (ka) () () () () () ()
137 .027 .020 .024 .020 .009 72.080 .054 .201 .000 .072 27.803 100.310
138 .032 .031 .033 .032 .050 72.080 .054 .201 .000 .072 27.803 100.388
139 .015 .018 .020 .016 .013 72.080 .054 .201 .000 .072 27.803 100.291
140 .066 .069 .062 .066 .065 72.080 .054 .201 .000 .072 27.803 100.538
141 .075 .080 .075 .073 .078 72.080 .054 .201 .000 .072 27.803 100.591
142 .015 .016 .015 .020 .017 72.080 .054 .201 .000 .072 27.803 100.293
143 .025 .019 .025 .018 .011 72.080 .054 .201 .000 .072 27.803 100.309
144 .015 .020 .023 .011 .021 72.080 .054 .201 .000 .072 27.803 100.300
AVER: .034 .034 .035 .032 .033 72.080 .054 .201 .000 .072 27.803 100.377
SDEV: .024 .026 .022 .024 .027 .000 .000 .000 .000 .000 .000 .120
As one can see the MAN background corrected Ti concentration variances are similar or smaller than the off-peak calculated concentrations. We will investigate how to calculate the sensitivity in the MAN Precision and Accuracy topic later, but now let's look at the Mn Ka data. It is worth noting however that off-peak measurements for Mn in magnetite are interfered with by the tail of the Fe Ka peak since it is so large, therefore we must utilize same side off-peaks or another method that avoids the tail of the Fe peak on the low side of the Mn ka position, as seen here:
(https://smf.probesoftware.com/oldpics/i60.tinypic.com/o6bhfr.jpg)
And here for Mn Ka using off-peak backgrounds:
ELEM: Mn Mn Mn Mn Mn Fe Al Si Ti Mg O SUM
XRAY: (ka) (ka) (ka) (ka) (ka) () () () () () ()
129 .038 .061 .028 -.001 .045 72.080 .201 .000 .012 .072 27.803 100.339
130 .104 .073 .056 .052 .069 72.080 .201 .000 .012 .072 27.803 100.521
131 .063 .049 .038 .024 .011 72.080 .201 .000 .012 .072 27.803 100.353
132 .076 .050 .057 .023 .079 72.080 .201 .000 .012 .072 27.803 100.452
133 .051 .063 .054 .029 .040 72.080 .201 .000 .012 .072 27.803 100.404
134 .048 .036 .033 .013 .048 72.080 .201 .000 .012 .072 27.803 100.346
135 .039 .053 .035 .048 .022 72.080 .201 .000 .012 .072 27.803 100.365
136 .058 .048 .026 .050 .034 72.080 .201 .000 .012 .072 27.803 100.383
AVER: .060 .054 .041 .030 .043 72.080 .201 .000 .012 .072 27.803 100.395
SDEV: .022 .011 .013 .019 .022 .000 .000 .000 .000 .000 .000 .063
and the same raw data calculated using MAN backgrounds:
ELEM: Mn Mn Mn Mn Mn Fe Al Si Ti Mg O SUM
XRAY: (ka) (ka) (ka) (ka) (ka) () () () () () ()
129 .027 .042 .030 .014 .035 72.080 .201 .000 .012 .072 27.803 100.316
130 .069 .058 .054 .044 .060 72.080 .201 .000 .012 .072 27.803 100.453
131 .041 .030 .034 .022 .015 72.080 .201 .000 .012 .072 27.803 100.309
132 .064 .049 .054 .044 .061 72.080 .201 .000 .012 .072 27.803 100.440
133 .051 .044 .047 .030 .043 72.080 .201 .000 .012 .072 27.803 100.383
134 .040 .032 .032 .021 .041 72.080 .201 .000 .012 .072 27.803 100.334
135 .021 .033 .030 .034 .018 72.080 .201 .000 .012 .072 27.803 100.305
136 .044 .026 .027 .022 .034 72.080 .201 .000 .012 .072 27.803 100.322
AVER: .045 .039 .039 .029 .038 72.080 .201 .000 .012 .072 27.803 100.358
SDEV: .017 .011 .011 .011 .017 .000 .000 .000 .000 .000 .000 .060
For Mn, the precision is even better probably partly due to the extrapolated nature of the off-peak calculation necessitated by the Fe Ka tail interference on the low side Mn off-peak position.
And a quick mention concerning accuracy. You may have noticed above that while the off-peak and MAN background corrected data concentrations for Ti look very similar, there was a greater difference between the concentrations reported for Mn comparing the off-peak and MAN background corrections. In fact the Mn Ka MAN concentrations are consistently lower. Now why would that be?
Let's first look at an MAN background regression plot first for Ti ka and observe that it looks fine:
(https://smf.probesoftware.com/oldpics/i61.tinypic.com/11sfdqx.jpg)
Note that the average Z of magnetite is around Z-bar = 21 or so. Now the Mn MAN regression fit that was utilized in the above post:
(https://smf.probesoftware.com/oldpics/i61.tinypic.com/9hon6w.jpg)
It is clear that standards 27 (CoO) and 730 (FeS2 or pyrite) fall *above* the main line of the regression curve. This may be due to some trace Fe contamination in the CoO crystal and the Fe Ka tail in the FeS2 interfering now not only with the Mn Ka low off-peak position, but in this case even with the Mn Ka on-peak position!
In the off-peak situation we were able to deal with this off-peak and on-peak interferences by utilizing "same side" off-peak positions, but for the MAN regression we must simply avoid the on-peak interference by removing the CoO and FeS2 standards from the fit as seen here:
(https://smf.probesoftware.com/oldpics/i58.tinypic.com/28mpn4m.jpg)
Removing these problematic standards we move the MAN regression fit to a lower intercept. Remember, background is the lowest intensity we can measure generally, so by removing the standards with the highest positive variances above the regression trend, we will improve the *accuracy* of the MAN background correction for Mn Ka. Also note that the variances of the Mn concentrations remain *exactly* the same!
So here, as a reminder, are the Mn MAN background corrected concentrations as calculated with CoO and FeS2 in the MAN regression fit:
ELEM: Mn Mn Mn Mn Mn Fe Al Si Ti Mg O SUM
XRAY: (ka) (ka) (ka) (ka) (ka) () () () () () ()
129 .027 .042 .030 .014 .035 72.080 .201 .000 .012 .072 27.803 100.316
130 .069 .058 .054 .044 .060 72.080 .201 .000 .012 .072 27.803 100.453
131 .041 .030 .034 .022 .015 72.080 .201 .000 .012 .072 27.803 100.309
132 .064 .049 .054 .044 .061 72.080 .201 .000 .012 .072 27.803 100.440
133 .051 .044 .047 .030 .043 72.080 .201 .000 .012 .072 27.803 100.383
134 .040 .032 .032 .021 .041 72.080 .201 .000 .012 .072 27.803 100.334
135 .021 .033 .030 .034 .018 72.080 .201 .000 .012 .072 27.803 100.305
136 .044 .026 .027 .022 .034 72.080 .201 .000 .012 .072 27.803 100.322
AVER: .045 .039 .039 .029 .038 72.080 .201 .000 .012 .072 27.803 100.358
SDEV: .017 .011 .011 .011 .017 .000 .000 .000 .000 .000 .000 .060
And here are the Mn MAN concentrations again, but this time with the CoO and FeS2 (pyrite) standards removed from the MAN fit:
ELEM: Mn Mn Mn Mn Mn Fe Al Si Ti Mg O SUM
XRAY: (ka) (ka) (ka) (ka) (ka) () () () () () ()
129 .044 .054 .041 .026 .045 72.080 .201 .000 .012 .072 27.803 100.377
130 .086 .070 .064 .057 .070 72.080 .201 .000 .012 .072 27.803 100.515
131 .057 .042 .044 .034 .025 72.080 .201 .000 .012 .072 27.803 100.371
132 .081 .061 .064 .056 .071 72.080 .201 .000 .012 .072 27.803 100.501
133 .067 .056 .058 .042 .053 72.080 .201 .000 .012 .072 27.803 100.444
134 .056 .044 .042 .033 .051 72.080 .201 .000 .012 .072 27.803 100.395
135 .037 .045 .041 .047 .029 72.080 .201 .000 .012 .072 27.803 100.367
136 .061 .038 .038 .035 .045 72.080 .201 .000 .012 .072 27.803 100.384
AVER: .061 .051 .049 .041 .049 72.080 .201 .000 .012 .072 27.803 100.419
SDEV: .017 .011 .011 .011 .017 .000 .000 .000 .000 .000 .000 .060
Remember the lowest intensities on can measure, must be the background- by definition!
Now these Mn MAN background corrected concentrations seem to agree much closer to the off-peak corrected concentrations shown here again for comparison:
ELEM: Mn Mn Mn Mn Mn Fe Al Si Ti Mg O SUM
XRAY: (ka) (ka) (ka) (ka) (ka) () () () () () ()
129 .038 .061 .028 -.001 .045 72.080 .201 .000 .012 .072 27.803 100.339
130 .104 .073 .056 .052 .069 72.080 .201 .000 .012 .072 27.803 100.521
131 .063 .049 .038 .024 .011 72.080 .201 .000 .012 .072 27.803 100.353
132 .076 .050 .057 .023 .079 72.080 .201 .000 .012 .072 27.803 100.452
133 .051 .063 .054 .029 .040 72.080 .201 .000 .012 .072 27.803 100.404
134 .048 .036 .033 .013 .048 72.080 .201 .000 .012 .072 27.803 100.346
135 .039 .053 .035 .048 .022 72.080 .201 .000 .012 .072 27.803 100.365
136 .058 .048 .026 .050 .034 72.080 .201 .000 .012 .072 27.803 100.383
AVER: .060 .054 .041 .030 .043 72.080 .201 .000 .012 .072 27.803 100.395
SDEV: .022 .011 .013 .019 .022 .000 .000 .000 .000 .000 .000 .063
But we won't know for sure which background correction method is more accurate, until we have a pure Fe3O4 standard to utilize as a "blank"!