ICP-MS: Inductively Coupled Plasma Mass Spectrometry

What ICP-MS does

ICP-MS answers one question with terrifying precision: how many atoms of a given element are in this sample? Not what molecule they were in, not what oxidation state — the plasma vaporizes all of that. It counts atoms of iron, of lead, of uranium, of arsenic, and it will find them even when there is one part in a trillion of the total. That sensitivity is the whole reason it exists.

The instrument is a relay race handing the sample down five stations, each at a wildly different pressure and temperature:

1. Nebulizer + spray chamber. The liquid sample is pumped (typically ~0.2–1 mL/min) into a nebulizer that shears it into a fine aerosol with a stream of argon. Only the finest 1–2% of droplets survive the spray chamber; the rest drain to waste. This selects a reproducible, uniform droplet population.
2. The plasma torch. The aerosol is swept into an inductively coupled argon plasma — a doughnut of ionized argon sustained at 6,000–10,000 K by a radiofrequency coil. Here the droplets desolvate, the residue vaporizes, the molecules atomize, and the atoms are ionized to M⁺.
3. The interface (cones). The plasma sits at atmospheric pressure; the mass analyzer needs high vacuum. A pair of water-cooled metal cones — the sampler and the skimmer, each with a ~1 mm orifice — extract a slice of the plasma into a vacuum stage, dropping the pressure by a factor of roughly a billion across two cones.
4. Ion optics. Charged lenses focus the M⁺ ions into a tight beam and steer them off-axis so that neutral atoms and photons from the plasma glare hit a stop instead of the detector.
5. Mass analyzer + detector. A quadrupole (or magnetic sector, or time-of-flight) filters the beam so that only one mass-to-charge ratio reaches the detector at a time. An electron multiplier counts the ions — literally, individual ions become measurable current pulses.

The output is a spectrum of counts-per-second versus mass-to-charge (m/z). Because most elements form singly charged M⁺ ions, m/z is effectively the isotope mass. Height at each mass = how much of that isotope is present.

The ionization mechanism, step by step

The heart of ICP-MS is what the plasma does to a single droplet in the ~2 milliseconds it spends in the torch. Follow one atom of, say, lead through it:

  droplet ─(desolvation)→ dry salt particle ─(vaporization)→ gas-phase molecules
        ─(atomization)→ neutral atoms  ─(ionization)→  M⁺  +  e⁻

1. Desolvation. As the aerosol droplet enters the 6,000 K outer plasma, the water flashes off, leaving a microscopic dry particle of the dissolved salts.
2. Vaporization. The particle sublimes into gas-phase molecules (oxides, chlorides, whatever the salt was).
3. Atomization. Those molecules are torn into free neutral atoms — every chemical bond is broken. This is why ICP-MS is element-specific but molecule-blind: it cannot tell you Cr(III) from Cr(VI); both arrive as bare Cr atoms.
4. Ionization. This is the key electron-transfer step. Argon has an ionization energy of 15.76 eV, and the plasma is thick with energetic Ar⁺ and free electrons. Any atom whose first ionization energy is below ~8 eV — which is nearly every metal — loses one electron to the plasma:

       M  +  (plasma energy)  →  M⁺  +  e⁻

   The fraction ionized follows the Saha equation, which relates the ion/atom ratio to temperature and ionization energy. At plasma conditions, elements with IE below ~8 eV are >90% ionized; even arsenic (9.79 eV) and selenium (9.75 eV) are a few tens of percent ionized, which is enough. Elements with IE above the argon line — fluorine (17.4 eV), neon, helium — cannot be ionized and are invisible to ICP-MS.

The electron flow is the essential logic: the plasma is an electron reservoir held at a chemical potential set by argon. Any atom that binds its outermost electron more loosely than argon (lower IE) will surrender it to the plasma and emerge as a cation. The whole technique is one giant, temperature-driven electron-transfer equilibrium, frozen in place the instant the ions are yanked into vacuum.

Hardware, gases, and real conditions

• Plasma gas. Argon, ~15 L/min total (outer/coolant ~14, auxiliary ~1, nebulizer ~1 L/min). RF power 1,200–1,600 W at 27 or 40 MHz coupled through a load coil.
• Cones. Sampler and skimmer are nickel for routine work, platinum for corrosive matrices (concentrated acids, high chloride) or when nickel background at m/z 58/60 must be avoided. Orifice ~0.8–1.1 mm (sampler), ~0.4–0.9 mm (skimmer).
• Vacuum. Interface region ~1–3 mbar (roughing pump); analyzer region ~1×10⁻⁵ to 1×10⁻⁶ mbar (turbomolecular pump). The pressure drop across the interface is what defines the whole engineering challenge.
• Collision/reaction cell. A pressurized multipole (usually an octopole or quadrupole) between the optics and the analyzer, fed with He (1–5 mL/min) for kinetic energy discrimination or H₂/NH₃/O₂ for chemical resolution. This is now standard on almost every commercial instrument.
• Internal standards. Elements not in the sample — commonly Sc, Ge, Rh, In, Tb, Bi — are spiked in at a fixed level to correct for drift and matrix suppression; each analyte is referenced to an internal standard of similar mass and ionization energy.
• Sample prep. Solids are digested in trace-metal-grade concentrated HNO₃ (± HF for silicates, ± HCl or H₂O₂), usually in a sealed microwave vessel at 180–220 °C, then diluted to keep total dissolved solids below ~0.1–0.2%.

Worked example: arsenic in drinking water

The US EPA limit for arsenic in drinking water is 10 µg/L (10 ppb). Suppose a lab measures a well-water sample and needs to know if it passes.

  Analyte:      ⁷⁵As  (arsenic is monoisotopic — only mass 75 exists)
  Interference: ⁴⁰Ar³⁵Cl⁺  lands at m/z 75  ← the classic arsenic problem
  Fix:          He collision cell (KED)  OR  react with O₂ → measure ⁷⁵As¹⁶O⁺ at m/z 91
  Internal std: ⁷²Ge  (similar mass, corrects for drift/suppression)

• Calibrate. Run blank + standards at 1, 5, 10, 25, 50 µg/L As, each with a fixed 10 µg/L Ge internal standard. Plot (As counts ÷ Ge counts) vs concentration — the ratio cancels instrument drift. The line is linear over the whole range.
• The chloride trap. Well water is full of chloride. Without a collision cell, ⁴⁰Ar³⁵Cl⁺ at m/z 75 is indistinguishable from ⁷⁵As⁺ and would report a false-high arsenic result — potentially condemning safe water or masking a real exceedance. The helium cell knocks the bulkier ArCl⁺ down by kinetic energy discrimination while As⁺ passes.
• Measure. The sample gives 12,400 counts/s for As, 41,000 for Ge. Interpolating the ratio on the calibration line returns 14 µg/L As.
• Verdict. 14 µg/L > 10 µg/L limit → the water fails. A method blank at <0.05 µg/L and a spike recovery of 98% confirm the number is real, not contamination.

Note that ICP-MS reports total arsenic. It cannot distinguish highly toxic inorganic As(III)/As(V) from relatively harmless organic arsenobetaine — for that you couple an HPLC column in front (LC-ICP-MS speciation), because the plasma destroys the very molecular information you might care about.

ICP-MS vs other elemental methods

	ICP-MS	ICP-OES	Flame AAS / GF-AAS
What it measures	Ion mass (m/z)	Emitted light	Absorbed light
Typical detection limit	0.1–1 ppt (ng/L)	0.1–10 ppb (µg/L)	ppb (GF) to ppm (flame)
Multi-element?	Yes — full periodic table per run	Yes — simultaneous	No — one element at a time (flame)
Isotope information	Yes — its unique strength	No	No
Linear dynamic range	8–9 orders of magnitude	5–6 orders	2–3 orders
Salt/matrix tolerance	Low (<0.2% TDS; needs dilution)	High	Moderate
Main interference class	Polyatomic + isobaric	Spectral line overlap	Chemical + ionization
Capital + running cost	Highest (argon-hungry, vacuum)	High	Low
Best at	Ultratrace + isotope ratios	Routine ppb multi-element	Single-element budget work

Interferences: the hard part

ICP-MS is limited less by sensitivity than by selectivity — telling your analyte apart from something at the same nominal mass. A unit-resolution quadrupole only separates integer masses, so anything landing on the same integer collides:

• Polyatomic interferences. The plasma, the argon, the solvent, and the matrix combine into molecular ions. The infamous ones: ⁴⁰Ar¹⁶O⁺ on ⁵⁶Fe (m/z 56), ⁴⁰Ar³⁵Cl⁺ on ⁷⁵As (75), ⁴⁰Ar⁴⁰Ar⁺ on ⁸⁰Se (80), ⁴⁰Ar¹²C⁺ on ⁵²Cr (52), ⁴⁰Ca¹⁶O⁺ on ⁵⁶Fe (56). These are why the collision/reaction cell became mandatory.
• Isobaric overlaps. Two different elements share an isotope mass — ¹¹⁴Cd and ¹¹⁴Sn, ⁴⁰Ar and ⁴⁰Ca, ⁸⁷Rb and ⁸⁷Sr. You correct mathematically by measuring an interference-free isotope of the offending element and subtracting its known abundance ratio.
• Oxide and doubly-charged ions. A few percent of some elements form MO⁺ (e.g. ¹⁴⁰Ce¹⁶O⁺ at m/z 156 overlaps ¹⁵⁶Gd) or M²⁺ (which appears at m/z = mass/2). Instruments are tuned to keep the CeO⁺/Ce⁺ ratio below ~2–3% as a health check.
• Matrix suppression. A high concentration of an easily-ionized element (Na, K, Ca) shifts the plasma's ionization balance and space-charge in the ion optics, suppressing the analyte signal. Internal standards and dilution manage it.

The two great weapons are (1) the collision/reaction cell — He kinetic-energy discrimination removes polyatomics by exploiting their larger collision cross-section, while reactive gases chemically convert either the interference or the analyte to a new mass — and (2) high-resolution sector instruments, which physically resolve ⁴⁰Ar¹⁶O⁺ (55.957 u) from ⁵⁶Fe⁺ (55.935 u) because their exact masses differ by 0.022 u.

Real-world applications

• Environmental monitoring. Lead, arsenic, cadmium, and mercury in drinking water, wastewater, and soil against EPA Method 200.8 / 6020. ICP-MS can confirm a lead level below the 15 ppb action limit — and below the 1 ppb aspirational goal — in a single fast run.
• Semiconductor purity. Silicon wafer and process-chemical manufacturers specify contaminants at sub-ppt because a few metal atoms can kill a transistor. ICP-MS is the workhorse for ultrapure-water and high-purity-acid certification.
• Clinical and toxicology. Blood lead, urine arsenic speciation, whole-blood mercury, and trace elements (Zn, Cu, Se) in serum. The technique underpins heavy-metal poisoning diagnostics.
• Geochemistry and dating. ²⁰⁶Pb/²⁰⁷Pb/²⁰⁸Pb and U–Pb isotope ratios for dating zircons; rare-earth-element patterns for provenance. Laser-ablation ICP-MS reads solids directly, spot by spot.
• Nuclear forensics and safeguards. ²³⁵U/²³⁸U enrichment measurement and detection of trace plutonium — isotope ratios ICP-MS delivers that optical methods cannot.
• Food and pharma. USP <232>/<233> elemental-impurity limits for drugs (Pb, As, Cd, Hg and catalyst metals like Pd, Pt) and heavy metals in rice, seafood, and infant formula.

Limitations and pitfalls

• Molecule-blind. The plasma destroys speciation. If you need As(III) vs As(V), or methylmercury vs inorganic Hg, you must hyphenate a chromatograph (LC/GC-ICP-MS) in front.
• Salt intolerance. Above ~0.1–0.2% dissolved solids, cones clog and signal drifts. Seawater (3.5% salt) and biological fluids need heavy dilution or matrix separation.
• Blind spots. Elements with ionization energy above argon's plasma line — H, C, N, O, F, the noble gases — are poorly ionized or invisible. And the argon spectrum itself blocks easy measurement of Ar (40), Ca (40 isobaric), and background-limited masses.
• Contamination is the enemy. When you are counting parts per trillion, a fingerprint, a dusty pipette tip, or ordinary reagent-grade acid ruins the blank. Ultratrace work demands clean rooms, PFA labware, and sub-boiling-distilled acids.
• Cost and argon appetite. ~15 L/min of argon runs continuously; a busy lab consumes a large liquid-argon dewar every day or two. Capital cost runs from six figures for a quadrupole to much more for a sector instrument.
• Mass discrimination. Heavier ions transmit slightly more efficiently than light ones, biasing raw isotope ratios; this is corrected with a known-ratio standard bracketing the samples.

Who invented it, and when

The inductively coupled plasma as a light source for optical emission was developed independently by Stanley Greenfield in the UK and Velmer Fassel at Iowa State in the mid-1960s. The leap to mass spectrometry came from coupling that atmospheric-pressure plasma to a vacuum mass analyzer — an unlikely marriage, because bridging a billion-fold pressure gap without destroying the vacuum was thought impractical.

Robert S. Houk, working with Velmer Fassel at the Ames Laboratory (Iowa State), published the foundational demonstration of ICP-MS in 1980, showing that ions extracted from the plasma through a sampling orifice could be mass-analyzed. In parallel, Alan Gray and Alan Date in the UK developed a related system. The first commercial instrument, the VG PlasmaQuad, launched in 1983. The collision/reaction cell — the innovation that tamed the argon interferences and made everyday sub-ppb iron, arsenic, and selenium routine — arrived commercially around the turn of the millennium (late 1990s to early 2000s). Today ICP-MS is the reference method for elemental trace analysis across the world's regulatory and research labs.

Safety and lab notes

• RF and hot plasma. The torch runs at >6,000 K behind interlocked shielding; the RF generator carries lethal voltages. Interlocks must never be defeated.
• Argon asphyxiation. Large volumes of argon are heavier than air and can pool in low, poorly ventilated spaces. Argon dewars require oxygen monitors and ventilation.
• Acid digestion. Sample prep uses hot concentrated HNO₃ and sometimes HF. HF causes deep, delayed burns and requires calcium gluconate on hand; digestions run in fume hoods or sealed microwave vessels.
• Waste. Analyzing heavy metals means concentrating heavy-metal waste — spent solutions and cones need hazardous-waste disposal.