A different question to ask the plasma
We already met the inductively coupled plasma — that Sun-hot torch that tears a sample into glowing atoms. In ICP-OES we looked at the light those atoms emit. But the plasma does something else, almost for free: it is so hot that it knocks an electron off most atoms, turning them into electrically charged particles called ions. This stripping is ionization, and it opens a completely different way to measure.
Once the atoms are charged, we can grab them with electric and magnetic fields, sort them by weight, and *count them one by one*. That is ICP-MS — inductively coupled plasma mass spectrometry. Instead of asking "what color do you glow?" we ask "how much do you weigh?" Both answers reveal the element, but counting individual atoms turns out to be the more sensitive question by far.
Sorting ions by weight
How do you sort charged atoms by weight? Imagine flicking marbles of different masses with the same push: the light ones fly off fast and curve sharply in a breeze, the heavy ones lumber along. Electric and magnetic fields act on ions in just this way. A light ion responds nimbly; a heavy ion barely budges. By tuning the fields, an instrument can let through only ions of one chosen weight at a time.
What the fields actually respond to is not weight alone but the mass-to-charge ratio — the ion's mass divided by how many charges it carries. Since the plasma usually strips just one electron, most ions carry a single charge, and the mass-to-charge ratio is essentially just the atomic weight. So a reading at, say, ratio 208 points straight to lead, and ratio 63 to copper.
The part that does this sorting is the mass analyzer. The most common one in ICP-MS sweeps quickly across the whole range of masses, pausing at each to count how many ions arrive. The result is a tidy list: at every atomic weight, how many ions per second. Tall counts mean lots of that element; the position along the scale tells you which element it is.
Why it reaches so low
ICP-MS is the champion of trace analysis for metals, routinely reaching parts per trillion — a thousand times lower than ICP-OES, and far below what any flame can do. A part per trillion is roughly one second in thirty thousand years. The reason it goes so low is the nature of counting: against a near-black background, even a few stray ions stand out, so detecting tiny amounts is much easier than spotting a faint glow buried in the light of a bright plasma.
A bonus: seeing isotopes
Because ICP-MS sorts purely by weight, it sees something the light-based methods cannot: isotopes. Most elements come in a few varieties that are chemically identical but differ slightly in weight — like otherwise-identical coins that weigh a hair more or less. ICP-MS reports each isotope as its own peak, in fixed proportions characteristic of the element. That set of proportions is the isotope pattern.
This is more than a curiosity. Matching the measured pattern to the known one confirms you really found the element you think you did. And it powers a remarkably precise calibration trick: spike the sample with an unusual isotope of the same element and compare ratios — a powerful kind of internal standard that geologists use to date rocks and food scientists use to trace where a product came from.
When to reach for ICP-MS
Walk the whole atomic ladder and a clear order of power emerges, from gentlest to mightiest:
- Flame photometry / flame AAS: simple and cheap, good down to parts per million.
- ICP-OES: many elements at once, good down to parts per billion.
- ICP-MS: the most sensitive routine choice, parts per trillion, plus isotope information — but costly and demanding to keep clean.