I’ve pressed into service the Cs
2Co
2Al(PO
4)
3 that I (accidentally) grew as a provisional standard for analyzing the natural pollucite that I had previously been using as a Cs standard. I’ve assumed the synthetic compound to be stoichiometric. Based on a combination of microprobe analyses, stoichiometric constraints, and solution ICP-MS (to determine Li content, which turns out to be negligible), my previous best guess as to the composition of the pollucite was as follows (in wt%):
SiO
2: 43.0
Al
2O
3: 16.3
FeO: <0.01
CaO: <0.01
Li
2O: <0.01
Na
2O: 1.4
K
2O: 0.01
Rb
2O: 0.1
Cs
2O: 37.6
H
2O: 1.6
At beam energy = 15 keV, I’ve found that the Cs
2Co
2Al(PO
4)
3 shows no evidence of change in Cs Lα count rate over a period of ten minutes using a 10 nA current with the beam defocused to 10 microns, and so I used these conditions to analyze the pollucite (counting for 10 s peak and 10 s bkg on Cs Lα). (Normally I defocus the beam anyway when analyzing pollucite.)
Using natural sanbornite (BaSi
2O
5, Si), natural celsian (Al), natural albite (Na), natural adularia (K), RbTiOPO
4 (Rb), and Cs
2Co
2Al(PO
4)
3 as standards and assuming 1.4 wt% H
2O, I get the following results:
PAP/MAC30:
The usual constraints on the pollucite-analcime solid solution for the six-oxygen formula unit (excluding H
2O) are as follows (see Beger, Z. Kristallogr. 129:280-302; Černý, Can. Min. 12:334-341):
n
Si + n
Al = 3
n
Cs + n
H2O = 1
sum large cations = n
AlFor this case (average of 52 analyses),
n
Si + n
Al = 3.000
n
Cs + n
H2O = 1.000 (obviously I had to choose a value of H
2O)
sum large cations = 0.911, n
Al = 0.912
Of course I can’t prove that the result is accurate, but it sure looks pretty reasonable. Also, note that the matrix correction factors for Cs Lα are close to unity, and so the choice of matrix correction model doesn't have much effect on wt% Cs
2O. The Armstrong model gives slightly lower wt% Cs
2O due to the smaller atomic number correction.
Armstrong/FFAST: