Written and maintained by CASRAI Editorial Board
Last updated
Polyatomic interference is the single most common source of biased results in ICP-MS, and it’s the one that’s easiest to miss — a polyatomic ion sitting on the same mass as your analyte doesn’t produce a flag or a failed QC check by itself, it just adds counts. The result looks like a clean, plausible number until it’s checked against a certified reference material or a second isotope and comes back high. Fixing it isn’t one technique — it’s a decision among four genuinely different approaches (collision cell, reaction cell, mathematical correction, and mass resolution), each with a real failure mode. This guide works through how to recognize a polyatomic interference, how the four correction approaches actually work, and which one to pick for a given matrix and detection-limit requirement — including where each one breaks down.
What a Polyatomic Interference Actually Is
ICP-MS separates ions purely by mass-to-charge ratio (m/z). It has no way to distinguish an analyte ion from any other ion that happens to land on the same nominal mass — and the plasma, the sample matrix, and entrained atmospheric gases constantly generate polyatomic ions (two or more atoms bonded together, retaining a single charge) that do exactly that. Because a quadrupole ICP-MS resolves to roughly unit mass, a polyatomic combination whose summed mass matches the analyte’s isotope reads as part of that analyte’s signal, biasing the result high.
The interferences that show up most often trace back to a small set of sources: the argon plasma itself, entrained air (nitrogen, oxygen, carbon), water and dissolved carbon in the sample, and whatever acid was used for digestion (chloride from HCl, sulfate from H₂SO₄, phosphate from H₃PO₄). Some of the classic, well-documented overlaps:
- Arsenic-75 (monoisotopic, no correction-by-isotope-ratio option) — 40Ar35Cl+ from argon plus chloride (from HCl digestion or a chloride-bearing matrix like seawater or blood) lands directly on mass 75.
- Selenium-80 (major isotope) — 40Ar2+ (argon dimer) overlaps the most abundant Se isotope; analysts often move to Se-77 or Se-78 instead, which carries its own, smaller interferences.
- Vanadium-51 — 35Cl16O+ from chloride-containing matrices.
- Iron-56 — 40Ar16O+, one of the most persistent interferences in ICP-MS because argon-oxide forms readily in any aqueous sample.
- Calcium-40 — direct isobaric overlap with 40Ar+ itself (the plasma gas), which is why Ca is often run on a minor isotope instead.
- Cadmium and other mid-mass elements — molybdenum, zirconium and other oxide-forming elements can generate MO⁺ species that land on Cd and other analytes a full atomic-mass-unit-plus-oxygen away from their parent.
These are the interferences EPA Method 200.8 and 6020 both build interference-check standards around (ICS-A for the interferents alone, ICS-AB for interferents plus analytes) — if you need the full method-comparison context for which EPA method requires what, see CASRAI’s EPA Method 200.8, 200.7, and 6020 comparison.
Recognizing a Polyatomic Interference Before It Reaches a Report
A biased-high result from a polyatomic interference does not announce itself. Three checks catch it before it goes out:
- Run an interference-check standard. A solution containing only the interfering elements (no analyte) that reads a false-positive signal on the analyte mass confirms an active interference and lets you quantify its magnitude under your instrument’s current conditions.
- Compare two isotopes of the same element when one exists. If an element has more than one measurable isotope and they disagree by more than method precision allows, one of them is very likely interfered — use the isotope with the smaller polyatomic burden, even if it’s less abundant.
- Check spike recovery in the real matrix, not just in a clean calibration matrix. A method that recovers cleanly in reagent water but runs high in a chloride- or sulfate-rich sample matrix is telling you the interference is matrix-driven, not an instrument problem — which points straight at which correction approach will actually fix it (see the decision table below).
The Four Correction Approaches
1. Collision Cell (Kinetic Energy Discrimination)
A collision cell sits between the plasma interface and the mass analyzer and fills with an inert gas — almost always helium. Polyatomic ions have a larger collisional cross-section than the monatomic analyte ion at the same mass, because they’re physically bigger (more bonds, more atoms), so they lose more kinetic energy per collision as they pass through the pressurized cell. An energy barrier (kinetic energy discrimination, KED) at the cell exit then rejects the lower-energy polyatomic ions while passing the higher-energy monatomic analyte ions through to the detector.
Where it works well: it’s a single, non-specific setting that knocks down essentially the whole range of common polyatomic interferences at once, without needing a different reaction gas or method development for each analyte. That makes He/KED the default first choice for multi-element survey methods (EPA 200.8-style panels, environmental and food-safety screening) where dozens of elements run in one method and per-analyte optimization isn’t practical.
Where it fails: KED reduces total ion transmission for every ion passing through the cell, analyte included, so it costs sensitivity — typically a factor of a few. For ultra-trace work where the analyte signal is already marginal, that sensitivity loss can push the detection limit above what the method requires. It also can’t help with a true isobaric interference where the interfering species is itself monatomic and the same mass as the analyte (Ar-40 on Ca-40, for instance) — if the interferent isn’t polyatomic, KED has nothing to discriminate against.
2. Reaction Cell (Dynamic Reaction Cell / Chemistry-Based)
A reaction cell also sits in the ion path but uses a chemically reactive gas — hydrogen, ammonia, or oxygen are the common choices — selected to react with a specific interfering species (converting it to a different mass and removing it from the analyte’s channel) or, less commonly, to react with the analyte itself and shift it to an interference-free mass. Because the chemistry is analyte- and interferent-specific, reaction-cell methods are typically developed and validated per analyte or per small analyte group rather than run generically across a whole panel.
Where it works well: for a specific, well-characterized interference that a collision cell can’t fully remove at the required detection limit — it’s more selective and can achieve a bigger reduction in the target interference than KED, with less sensitivity penalty on the analyte, precisely because the removal mechanism is chemistry, not just energy loss.
Where it fails: the gas and cell conditions that work for one analyte’s interference can introduce a new interference for a different analyte measured in the same run, so a reaction-cell method built for a multi-element panel often needs multiple gas modes switched during the run (adding run time) or accepts a compromise condition that isn’t optimal for every element. It also requires real method-development time per analyte rather than a single blanket setting, which is a cost a KED-only method doesn’t carry.
3. Mathematical (Equation-Based) Correction
Rather than removing the interference physically, mathematical correction measures a secondary isotope known to come from the same interferent (or a proxy element that tracks it) and uses an empirically determined coefficient — established by running interference-check standards on the specific instrument, under the specific operating conditions — to calculate and subtract the interference’s contribution from the raw analyte signal. EPA Method 200.8 documents correction equations of exactly this form for several analyte/interferent pairs.
Where it works well: on instruments without a collision or reaction cell (older quadrupole systems, or when cell gas isn’t available), and for interferences that are well-characterized and reasonably stable in magnitude relative to the analyte signal.
Where it fails: the correction coefficient has to be re-established for the specific instrument and tune conditions in use, and it degrades badly as the interference grows relative to the analyte — when the correction term becomes a large fraction of the total signal, the subtraction amplifies noise from both measurements, and precision on the corrected result suffers even if the mean is approximately right. It’s also a poor fit for samples with unpredictable, highly variable interferent loads, since the coefficient assumes a consistent relationship that a genuinely variable matrix won’t hold.
4. Mass Resolution (High-Resolution / Sector-Field ICP-MS)
A magnetic-sector (high-resolution) ICP-MS resolves mass differences far finer than a quadrupole’s unit-mass resolution — fine enough, at sufficiently high resolving power, to physically separate an analyte peak from a polyatomic interference peak that a quadrupole cannot distinguish, because the two species’ exact masses differ slightly even though their nominal (rounded) masses match.
Where it works well: interferences that no collision or reaction cell chemistry handles well, or where the analyte/interferent mass difference is large enough to resolve at practical resolving power without an unacceptable sensitivity loss. It’s the standard route for isotope-ratio work and ultra-trace applications (geochronology, semiconductor-grade purity, certain environmental ultra-trace methods) where neither cell approach reaches the required combination of accuracy and detection limit.
Where it fails: resolving power and sensitivity trade off directly — running at higher resolution to separate a close mass difference costs signal, sometimes substantially, and the instrument itself is a much larger capital and operating-cost commitment than a quadrupole with a collision/reaction cell. For routine multi-element panels where a cell-based correction already meets the required detection limit, sector-field resolution is solving a problem the method doesn’t actually have.
The Decision, in Practice
| Situation | Best-fit approach | Why |
|---|---|---|
| Multi-element survey/screening panel, moderate detection limits | Collision cell (He/KED) | One setting handles most common polyatomics without per-analyte development |
| One or two analytes with a well-characterized, stubborn interference a cell alone can’t clear | Reaction cell, analyte-specific gas mode | Chemistry removes the specific interferent with less sensitivity cost than KED |
| Older/simpler quadrupole with no cell, interference is stable and well-characterized | Mathematical correction | No hardware requirement; degrades badly only when the interferent load is large or variable |
| Ultra-trace or isotope-ratio work where cell chemistry can’t fully resolve the overlap | Mass resolution (sector-field) | Physically separates masses a quadrupole cell cannot distinguish |
| Highly variable matrix, interferent load unpredictable sample-to-sample | Collision or reaction cell, not mathematical correction | Physical removal doesn’t depend on a fixed coefficient the matrix keeps violating |
In practice, most routine multi-element ICP-MS methods run collision-cell KED as the default and reserve reaction-cell chemistry or mathematical correction for the specific one or two analytes KED alone doesn’t clear to the required detection limit — rather than choosing one approach for the whole method. Whichever approach is selected, it should be validated the same way any analytical procedure change is: against a certified reference material or a documented interference-check standard, not assumed correct from a manufacturer’s default setting. For a broader look at how method validation is structured and documented, see CASRAI’s ICH Q2(R2) analytical procedure validation guide and the ICP-MS sample preparation and acid digestion guide for how digestion acid choice itself introduces some of the chloride- and sulfate-based interferences discussed above. For the instrument-selection question one level up — ICP-MS versus ICP-OES — see the ICP-MS vs ICP-OES comparison.
Frequently Asked Questions
Can I just always run helium collision cell mode and skip the other approaches?
For most routine multi-element methods, yes as a starting point — but check the specific analytes against your required detection limits first. KED’s sensitivity cost can push a marginal analyte (one already near its detection limit before any correction) above the method’s reporting limit, and KED does nothing for a true isobaric (monatomic) interference. Confirm performance against an interference-check standard for each analyte that matters, rather than assuming a blanket collision-cell setting clears everything.
Why does the same element sometimes need a different correction approach in a different sample type?
Because the interference itself is matrix-driven. Chloride-based interferences (on As, V) are severe in seawater, blood, or HCl-digested samples but largely absent in a low-chloride matrix; sulfate- and phosphate-based interferences behave the same way relative to their source acids. A correction approach validated in one matrix isn’t automatically valid in another — re-check interference magnitude whenever the matrix changes meaningfully.
Is mathematical correction ever preferable to a collision or reaction cell, even when both are available?
Occasionally, for a well-characterized interference where the correction is small relative to the analyte signal and the goal is throughput on an instrument where switching cell gas modes mid-run adds unwanted time. It’s rarely the first choice on a modern instrument with an operating collision/reaction cell, precisely because it degrades as interferent load grows, where physical removal doesn’t have that same failure mode.
