Page 28 - Inorganic Mass Spectrometry - Fundamentals and Applications

18 S~it~

nature; care must be exercised in these cases (although there is a NIST certified
isotopic standard for lead).
Certified standards or no, the same procedure is followed. The value of an
isotopic calibration ratio is measured for the reference material and compared to
the certified (or accepted) value. The correction factor necessary to adjust the
measured value to the certified is calculated; it is then the bias correction for that
ratio. Most often, a bias correction per mass is used. Even though the actual
variation of bias is not strictly linear, the limited mass range swept for a single
element makes it a good approximation; any deviation fkom linear is insignificant
in comparison to measurement uncertainties. Typical biases are a few tenths of a
percent per mass when a single-fil~ent configuration is used and somewhat less
for multifilament. Multifilament aqalyses are in! gener sceptible to varia-
tions in bias correction than single-filament, but they no means immune,
Bias corrections determined from analysis of s are applidd to the
samples under test. Use of such an average bias correction can be viewed only as
an approximation to the truth; so many factors contribute to bias that it is impos-
sible to control them all. For example, as previously stated, the work
rhenium filament is determined by which crystal face is involved:
loading samples on filaments is through use of single resin beads
beads are 100-1200 pm in diameter, which is about the size of rhenium c~stal~ites
in a polycrystalline filament [17]. Clearly the work function applicable to the
analysis in question may or may not be that operative when ins~ment calibration
was carried out. Another parameter difficult to control in real-world conditions is
sample purity, which also affects bias. It is impossible to purify all samples to the
same degree, and contaminants adversely affect ionization efficiency; low effi-
ciency means higher filament temperatures, which in turn mean a different bias
correction. These are only two of sundry variables that can affect ionization
efficiency.
In practice, the analyst monitors the bias correction through analysis of a
reference standard on a routine, often daily, basis. This value comes to be known
very well and makes insignificant con~butions to overall precision. Even though
it may not be truly applicable to the sample being analyzed, using it is far better
tban applying no correction; it is the best that can be done in an imperfect world. A
model of thermal fractionation on mass spectrometer filaments has been devel-
oped by Habfast 1581.
Fractionation is such a vexing problem that other means of addressing it
have been devised. One is total e~~a~stiu~, in which the entire sample is con-
sumed; it has been successfully applied to uranium [59] and the rare earths 1601.
The idea here is that, if all ions emitted from the sample are collected, they will be
representative of the sample itself, and no bias correction will be required.
Because signal intensity varies rapidly, running to exhaustion can only be accom-
plished using a multicollector mass spectrometer. It also requires a reasonable