Page 119 - Inorganic Mass Spectrometry - Fundamentals and Applications

Inductively Coupled Plasma Mass Spectrometry 109

[l
the center axis through collisions with a low-molecular-weight neutral gas 171.
Second, efficient chemical reactions that reduce or nearly eliminate reactive
polyatomic ions can be attained. Attempts to use collisions, without reaction, to
fragment polyatomic ions in ICP-MS are of limited value because the scattering
cross section is typically larger than the fragmentation cross section [ 1 1 l]. There-
fore, losses of analyte ions due to scattering are large compared to the additional
fragmentation of polyatomic ions.
Results from initial ion trap ICP-MS experiments indicated that signals due
to argon ions and many polyatomic ions were much smaller than expected
[ 1331.
Reactions between k+ and H, result in formation of low-mass ions such as H+
and H3+ and Ar atoms [148,149]. Ar-containing polyatomic ions, such as kO+,
ArOH+, kCl+, h,+, and Arc+, can also be removed by reaction with H, or water
vapor in a reaction cell [ 1 15,148,1491. Other gases, such as oxygen, may be useful
reagents to remove other molecular ions.
Reaction cells appear to be a much better way to reduce signals due to Ar-
containing molecular ions and k+ itself than the use of “cold” plasma condi-
tions. Because normal plasma conditions are used, elements with high ionization
energies, such as Se and As, do not suffer from sensitivity losses, unlike “cold”
plasma conditions. The severe chemical matrix effects that are typical of “cold”
plasma conditions are prevented. The first commercial ICP-MS instrument to use
this concept was introduced by Micromass UK Ltd. However, as noted, reaction
product ions must be controlled or removed to prevent other (new) spectral
overlaps.

at~e~atica~ Correction for Spectral Overlaps
Mathematical correction procedures can be used to remove the contribution of a
spectral overlap from a measured signal. However, if the signal due to the spectral
overlap is much larger than the analyte signal, the signal-to-noise ratio of the
corrected signal may be poor. Furthermore, it may not be easy to predict and
account for quantitatively all of the potential sources of spectral overlap, partic-
ularly those due to polyatomic ions. For isobaric overlaps (Table 3.2), for which
the relative isotopic abundances are predictable, mathematical corrections are
straightforward. Instrument software often has built-in correction equations for
this case.
Contributions from molecular ions are typically more difficult to correct
quantitatively because there may be many molecular ions important over a short
mass range; the molecular ion intensity varies, depending on the sample matrix;
and the intensities may vary over time more dramatically than elemental ion
signals. Multivariant methods including multiple linear regression [ 150,15 l],
principal component analysis [ 1521, and multicomponent analysis [ l531 have
been used. Improvements in detection limits by up to two orders oE magnitude