I'll keep it short. Toying with me or not here is how its done in a nutshell:
1.) Mass with microgram balance of your choice, a Sartorius SE2 is a prime example of a scale good to 0.0000001 g. Take triplicate mass, avg. and deviation.
2.) dissolution in a PTFE vessel using spectral grade acids (as in trace metals, usually $100/L or more from Baker). The dissolution vessel has been leached repeatedly with acid and alkali and rinsed repeatedly with 18,2 MΩ water.
3.) Take it to a calibrated ICP-OES, remove a sample with a calibrated syringe (should also be massed) that has also been pre-leached with acid/alkali and rinsed. The syringe must have no metal components.
Much of it comes down to sample size.
And since these errors are additive and tend to
propagate any mistake can quickly throw off the results.
You should also remember this for the fire assay technique as well--it has the same limitations with respect to error causation, if not more. Thinking of a few off the top of my head I would say:
- vapour pressure of values (even solid Pt has a vapour pressure that must be accounted for in analytical work)
mechanical losses during transfer from cupel to balance
impurities in inquarts, litharge, cupels etc.
impurities in reagents during any dissolution
inaccuracies in mass measurement
incomplete cupellation
deposition due to capillary 'creep' (failed agglomeration in button)
etc.
As you can see, there are many possible sources of error in fire assay and also in ICP.
I have nothing against fire assay--for what it costs and what it provides it is a good technique, but I would be highly hesitant to say it is better than ICP technologies or vice versa.
[/i]