I would urge extreme caution in using MgF2 as a standard for analysis of F in apatite. In this case the F Ka absorption correction is astronomically large, and hence the accuracy of the correction is not likely to be good. For instance, using PAP in conjunction with MAC values suggested by Pouchou and Pichoir (in Electron Probe Quantitation) and using 15 kV potential and 40 deg. takeoff angle, f(chi) for F Ka in MgF2 is 0.6090, while in Ca5(PO4)3F it is 0.1355. This produces an absorption correction factor of 4.494, i.e. a correction of ~350%. Any small error in MACs or in the phi(rho*z) model could affect calculated wt% F significantly. Not only is F Ka strongly absorbed by oxygen, but it is roughly equally strongly absorbed by Ca, as it is energetic enough to ionize Ca L1,2,3. For instance, comparison of F Ka f(chi) in MgF2 and CaF2 gives 0.6090 versus 0.2180. Other problems notwithstanding (TDI, for instance), if there ever were a case in which a “matrix match” were needed between standard and unknown, the analysis of F in apatite is it.
Also, when analyzing for F using LDE1, are you taking into account the overlap from P Ka(3)? If you are using pulse amplitude discrimination, have you verified that you have completely eliminated the contribution from P Ka? For instance, have you done a wavelength scan for F Ka in differential mode on a nominally F-free material such as Ca2P2O7 (calcium pyrophosphate), which has roughly similar wt% CaO and P2O5 as apatite?
Hi Brian,
I think you make some good points.
On the standard for fluorine I suspect that a robust standard may be more import than the matrix match. Here are some comparisons between the Heinrich, Henke and FFAST MACs for these matrices:
MAC value for F ka in O = 12439.63 (LINEMU Henke (LBL, 1985) < 10KeV / CITZMU > 10KeV)
MAC value for F ka in O = 12390.00 (CITZMU Heinrich (1966) and Henke and Ebisu (1974))
MAC value for F ka in O = 11927.75 (MAC30 Heinrich (Fit to Goldstein tables, 1987))
MAC value for F ka in O = 11863.62 (FFAST Chantler (NIST v 2.1, 2005))
MAC value for F ka in Ca = 12415.20 (LINEMU Henke (LBL, 1985) < 10KeV / CITZMU > 10KeV)
MAC value for F ka in Ca = 12370.00 (CITZMU Heinrich (1966) and Henke and Ebisu (1974))
MAC value for F ka in Ca = 12623.93 (MAC30 Heinrich (Fit to Goldstein tables, 1987))
MAC value for F ka in Ca = 12132.31 (FFAST Chantler (NIST v 2.1, 2005))
MAC value for F ka in P = 5550.00 (LINEMU Henke (LBL, 1985) < 10KeV / CITZMU > 10KeV)
MAC value for F ka in P = 5526.00 (CITZMU Heinrich (1966) and Henke and Ebisu (1974))
MAC value for F ka in P = 5219.60 (MAC30 Heinrich (Fit to Goldstein tables, 1987))
MAC value for F ka in P = 4905.57 (FFAST Chantler (NIST v 2.1, 2005))
The MAC for F ka by P is fairly small and the others are larger but in relative agreement, but I do think having a matrix match is ideal as you suggest.
I wonder if we should talk to Marc Schrier about the possibility of synthesizing an end member fluor-apatite? Wouldn't that be nice to have for everyone?I already have a synthetic chlor-apatite (just a few grains left, but hydrothermally grown from the University of Nice, see Argiolas and Baumer, Can. Min., v. 16, pp 285-290, 1978), which is wonderful, though a little beam sensitive. Fortunately the Cl intensity doesn't seem to change much over time with mild beam conditions. I should also mention to Marc him synthesizing this in 100 - 200 gram quantities as well...
By the way, I received about 200 grams of RbTiOPO4 and after we analyze it for traces I will be sending it to Marc Schrier for distribution, so I'll let you know when that is ready.
Back to fluor-apatite though- the question isn't so much the size of the matrix correction relative to the standard, though it is worth considering, but the relative accuracy error on that correction is important I agree. Here is a comparison of two different matrix corrections on a fluor-apatite material (all using the same MACs) done by a student at the USGS so please ignore the accuracy! I also cut out most of the elements because she analyzed 36 elements in this standard!
First the JTA matrix correction (the default in my software):
ELEM: W Ta Fe Ti Nb Mn Na F SUM
100 .000 .000 .079 .000 .011 .037 .157 4.342 101.390
101 .000 .000 .000 .306 .000 .007 .130 3.655 100.569
102 .000 .000 .000 .000 .020 .038 .139 4.484 101.617
AVER: .000 .000 .026 .102 .011 .028 .142 4.160 101.192
SDEV: .000 .000 .045 .176 .010 .017 .014 .444 .551
SERR: .000 .000 .026 .102 .006 .010 .008 .256
%RSD: .16 .19 173.20 173.21 96.92 63.20 9.61 10.66
PUBL: n.a. n.a. n.a. n.a. n.a. n.a. .178 3.700 99.662
%VAR: --- --- --- --- --- --- -20.44 12.44
DIFF: --- --- --- --- --- --- -.036 .460
STDS: 468 467 7852 7840 442 7845 7815 8811
STKF: 1.0000 1.0000 .4985 .5548 1.0000 .1418 .0505 .1537
STCT: 602.94 622.84 44.65 43.13 135.92 13.23 52.31 16.58
UNKF: .0000 .0000 .0002 .0008 .0001 .0002 .0007 .0083
UNCT: -1.53 -.86 .01 .05 .00 .02 .68 .89
UNBG: 4.20 3.11 .29 .03 .27 .29 .80 .15
ZCOR: 1.1917 1.2114 1.1812 1.2234 1.2559 1.2066 2.1646 5.0168
KRAW: -.0025 -.0014 .0002 .0011 .0000 .0016 .0129 .0540
PKBG: .64 .72 1.04 .52 .99 1.08 1.85 7.21
INT%: 5.62 .30 ---- ---- ---- ---- ---- -18.21
And here is the original PAP (XPP) correction:
ELEM: W Ta Fe Ti Nb Mn Na F SUM
100 .000 .000 .079 .000 .012 .037 .153 4.386 101.453
101 .000 .000 .000 .307 .000 .007 .127 3.715 100.627
102 .000 .000 .000 .000 .022 .038 .136 4.549 101.685
AVER: .000 .000 .026 .102 .012 .028 .139 4.217 101.255
SDEV: .000 .000 .046 .177 .011 .017 .013 .442 .556
SERR: .000 .000 .026 .102 .006 .010 .008 .255
%RSD: .13 .15 173.20 173.21 96.94 63.20 9.43 10.48
PUBL: n.a. n.a. n.a. n.a. n.a. n.a. .178 3.700 99.662
%VAR: --- --- --- --- --- --- -22.07 13.96
DIFF: --- --- --- --- --- --- -.039 .517
STDS: 468 467 7852 7840 442 7845 7815 8811
STKF: 1.0000 1.0000 .4892 .5435 1.0000 .1383 .0485 .1498
STCT: 602.94 622.84 44.65 43.13 135.92 13.23 52.31 16.58
UNKF: .0000 .0000 .0002 .0008 .0001 .0002 .0006 .0081
UNCT: -1.53 -.86 .01 .05 .00 .02 .68 .90
UNBG: 4.20 3.11 .29 .03 .27 .29 .80 .15
ZCOR: 1.3693 1.3918 1.2122 1.2560 1.3782 1.2383 2.2085 5.2191
KRAW: -.0025 -.0014 .0002 .0011 .0000 .0016 .0129 .0540
PKBG: .64 .72 1.04 .52 .99 1.08 1.85 7.21
INT%: 5.62 .30 ---- ---- ---- ---- ---- -18.19
The matrix correction for F Ka goes from 5.0168 to 5.2191 so about 4% which is less than the relative standard deviation.
Also, the interference from P Ka III is important and I observe that on my Sx100, and it should be a slightly larger overlap on a JEOL instrument. For that of course we'd like to have a material with known P and no F, for instance the chlor-apatite standard I mentioned above. By the way, the above analyses are with the P III interference on F Ka corrected (see the line labeled INT% which is the interference correction percent which is ~18% so that is quite important as you mentioned.
Lots more to discuss here...