Inductively Coupled Plasma Spectrometry

A plasma at 6,000–10,000 K can turn almost any element on the periodic table into a countable signal. This is the working principle behind inductively coupled plasma spectrometry. This instrument replaced the humble flame in serious trace-element laboratories over the last four decades. Flame Photometry and Atomic Emission tops out around 2,000–3,000 K. It also struggles with refractory oxides. The argon ICP shrugs off chemical matrices that would choke a flame. It atomizes calcium phosphates, titanium oxides, and rare-earth complexes with equal indifference. The stakes are concrete: an environmental lab certifying lead in drinking water at 15 ppb. A geochemist dates rocks from uranium/lead ratios at the femtogram level. A pharmaceutical QC chemist screens for elemental impurities under ICH Q3D limits. All of them lean on the same annular plasma. It is sustained by a radio-frequency magnetic field. This plasma does work no chemical flame could manage. The technique comes in two closely related flavors. They share the same torch and sample-introduction hardware but diverge at the exit. ICP-OES (optical emission spectrometry) reads photons, while ICP-MS (mass spectrometry) reads ions. Both build directly on ideas from Interferences in Atomic Absorption — spectral overlap, ionization suppression, matrix effects. But the physics of the source changes the entire interference landscape. The big picture is that a plasma solves the flame's fundamental limitation. That limitation is insufficient thermal energy and residence time. Flames cannot fully atomize and excite refractory species. Plasma solves this by using electromagnetic induction rather than chemical combustion. This sustains a much hotter, much longer-lived fireball of ionized argon. This single design choice cascades into higher sensitivity and wider linear dynamic range. It also brings multi-element simultaneous capability. And it introduces a whole new set of interferences. These include argon-based polyatomic ions, doubly charged species, and matrix suppression. These must be understood and corrected. This article works through torch construction and RF power coupling. It covers the physical stages of desolvation-vaporization-atomization-ionization inside the plasma. It explains the Saha equation governing ionization balance. It also covers the optics of ICP-OES and the mass analyzers of ICP-MS. Finally, it addresses calibration strategies that make quantification trustworthy. These include external calibration, standard additions, and internal standardization. Three fully worked numerical examples anchor the theory in real numbers. These are a foundational ionization calculation, an applied ICP-OES trace-metal determination, and an exam-level ICP-MS isobaric interference correction.

Key takeaways

The ICP torch sustains a 6,000–10,000 K argon plasma by RF induction (27.12 or 40.68 MHz) coupled through a load coil, not by chemical combustion.
Sample aerosol threads through the injector into the plasma's central channel, where it undergoes desolvation, vaporization, atomization, and (partially) ionization within milliseconds.
The Saha equation predicts the ionization fraction of each element at plasma temperature; most metals are >90% ionized, which is exactly what ICP-MS needs and ICP-OES exploits via ion lines.
ICP-OES measures emission intensity at characteristic wavelengths with echelle or Czerny-Turner optics and CCD/CID detectors, giving simultaneous multi-element analysis over ppb–percent range.
ICP-MS couples the same plasma to a quadrupole, sector-field, or time-of-flight mass analyzer through a differentially pumped interface, reaching sub-ppt detection limits but facing isobaric and polyatomic interferences.
Matrix effects, ionization suppression, and spectral/spectral-overlap interferences are controlled with internal standards, collision/reaction cells, matrix matching, and standard addition.
Key idea: an RF-sustained argon plasma supplies enough thermal and electron density to atomize and ionize virtually any element completely, converting the chemistry of atomization (the flame's weak point) into a solved problem, so the remaining analytical challenge is purely optical or mass-spectral discrimination.

Where this lives today

ICP-OES and ICP-MS are now the backbone of elemental analysis across an enormous range of industries. Environmental laboratories run EPA Methods 200.7 (ICP-OES) and 200.8 (ICP-MS) routinely for drinking water, wastewater, and soil digests, quantifying dozens of regulated metals — Pb, As, Cd, Cr, Hg (via cold-vapor generation coupled to ICP), and Se — in a single injection. Pharmaceutical quality control uses ICP-MS to satisfy ICH Q3D elemental impurity guidelines, screening drug substances and excipients for catalyst residues (Pd, Pt, Ir from synthetic catalysis) and toxic elements at low-ppb limits that no wet-chemical method could reach reliably. Geochemistry and cosmochemistry rely on multi-collector ICP-MS (MC-ICP-MS) for isotope-ratio measurements — U-Pb geochronology, Sr/Nd/Hf isotope systematics for petrogenesis, and even meteorite dating — exploiting the plasma's ability to fully ionize refractory elements that resist other ionization sources. Clinical and forensic toxicology laboratories use ICP-MS for whole-blood and urine trace-metal panels (lead poisoning screening, arsenic speciation when hyphenated with HPLC), while semiconductor and advanced-materials manufacturers use high-resolution ICP-MS to certify ultra-high-purity process chemicals and wafer-etching reagents at part-per-trillion contamination levels. Laser ablation ICP-MS (LA-ICP-MS) extends the technique to solid samples directly — geological thin sections, archaeological glass and ceramics, forensic paint chips — bypassing solution digestion entirely by vaporizing a micron-scale spot with a focused laser and sweeping the ablated aerosol into the same torch described here. In every one of these applications, the detector technology on the back end — photomultiplier arrays, CCDs, CID arrays for OES; electron multipliers and Faraday cups for MS — is itself a rich topic, one that connects directly to Detectors for Atomic Spectroscopy.

Ask the tutor: How do I choose the best internal standard element for a given ICP-MS method, and what criteria (mass, ionization energy, chemistry) should guide that choice?

The physical picture

An ICP source begins with an ordinary bottle of argon gas and a coil of copper tubing carrying a radio-frequency current. The torch itself is three concentric quartz tubes. The outermost carries the plasma gas (typically 12–17 L/min Ar), which both sustains the discharge and thermally insulates the outer quartz wall. The middle tube carries the auxiliary gas (0.5–1.5 L/min Ar), used to lift the plasma base off the injector tip and adjust the base geometry. The innermost tube is the injector, which threads the sample aerosol (carried in ~1 L/min Ar) as a narrow, cool channel straight through the middle of the hot plasma — this channel is why sample material gets processed rather than merely skirting the fireball. A two- or three-turn water-cooled copper load coil wraps around the top of the outer tube and is driven by an RF generator, most commonly at 27.12 MHz (a few instruments use 40.68 MHz) with forward power of 700–1,500 W. The oscillating magnetic field induces azimuthal eddy currents in a seed population of electrons and ions (produced by a Tesla-coil spark at ignition) inside the tube. These electrons are accelerated and collide with argon atoms, ionizing them further in a self-sustaining chain — this is why the discharge is "inductively coupled": energy is transferred to the plasma the same way a transformer's secondary is energized, with no electrodes touching the gas at all. The result is a toroidal (donut-shaped) region of maximum current density and maximum temperature, roughly 8,000–10,000 K, sitting a few millimeters above the coil. The sample-carrying central channel threads through the hole of the donut, which is somewhat cooler (6,000–8,000 K) but still far hotter than any flame. As the aerosol droplets travel up this channel over roughly 2 milliseconds of residence time, they pass through four sequential physical stages: desolvation (solvent evaporates, leaving a solid particle), vaporization (particle melts and vaporizes to a gas-phase molecule), atomization (molecule dissociates to free atoms), and excitation/ionization (collisions with energetic electrons and argon metastables promote atoms to excited states or strip an electron entirely).

Figure

Cross-section of the ICP torch showing gas flows, load coil, and plasma zones

Once atoms and ions in the plasma are excited, they relax by emitting photons at characteristic wavelengths — the basis of ICP-OES. Alternatively, ions can be extracted through a sampling interface into vacuum and separated by mass-to-charge ratio — the basis of ICP-MS. The two techniques share everything up to the point of extraction; they diverge entirely at the exit.

Figure

ICP-OES vs ICP-MS: shared torch, divergent detection

The core ideas

The central quantitative relationship governing the plasma's internal state is the Saha ionization equation, which describes the equilibrium between neutral atoms, ions, and free electrons at a given temperature — exactly the equilibrium that decides how much of an analyte reaches the ICP-MS detector as M⁺ versus how much remains as neutral M (invisible to a mass spectrometer that only extracts ions):

\frac{n_i n_e}{n_a} = \left(\frac{2 g_i}{g_a}\right)\left(\frac{2\pi m_e k_B T}{h^2}\right)^{3/2} \exp\!\left(-\frac{E_i}{k_B T}\right)

where:

$n_i, n_a$ = number densities of ions and neutral atoms of the element (m⁻³)
$n_e$ = electron number density in the plasma (m⁻³), typically $10^{20}$ – $10^{21}$ m⁻³ for an argon ICP
$g_i, g_a$ = statistical weights (degeneracies) of the ion ground state and atom ground state
$m_e$ = electron mass, $9.109 \times 10^{-31}$ kg
$k_B$ = Boltzmann constant, $1.381 \times 10^{-23}$ J/K
$T$ = plasma excitation temperature (K)
$h$ = Planck constant, $6.626\times10^{-34}$ J·s
$E_i$ = first ionization energy of the element (J, convert from eV via $1\text{ eV}=1.602\times10^{-19}$ J)

Reading: the Saha equation is the Boltzmann distribution's cousin for ionization equilibria — it says that even elements with substantial ionization energies (5–8 eV) become almost completely ionized once $k_BT$ is a sizeable fraction of $E_i$ , because the huge phase-space prefactor $(2\pi m_e k_B T/h^2)^{3/2}$ compensates for the exponential penalty. At flame temperatures (2,500 K, $k_BT \approx 0.22$ eV) this prefactor is far too small to matter and most metals stay neutral; at ICP temperatures (7,500 K, $k_BT \approx 0.65$ eV) the same equation predicts >90% ionization for elements with $E_i$ below about 8 eV — essentially all metals and many nonmetals. The second governing relationship connects measured signal to concentration — the calibration equation used identically in OES (intensity) and MS (ion counts), and the reason internal standardization works:

S = k \, C + S_{\text{blank}}, \qquad k = \Phi(T, n_e)\,\varepsilon\,\eta

where:

$S$ = measured signal (emission intensity in counts/s, or ion counts/s)
$C$ = analyte concentration in the introduced solution (typically µg/L or mg/L)
$k$ = overall sensitivity (calibration slope), lumping transport efficiency, atomization/ionization efficiency $\Phi$ , and detection efficiency $\eta$ (optical or ion-transmission), with $\varepsilon$ a species-specific spectroscopic/ionization constant
$S_{\text{blank}}$ = background/blank signal, subtracted to give net signal

Reading: because $k$ depends on plasma conditions ( $T$ , $n_e$ ) that drift with aerosol loading, gas flow, and matrix composition, a single external calibration curve is only as good as the day-to-day (or even sample-to-sample) stability of $k$ — which is precisely why internal standards (an added element whose $k_{IS}$ tracks the analyte's $k$ ) and matrix-matched standards are the workhorses of quantitative ICP work, not an optional refinement.

Derivation AIonization fraction from the Saha equation for a trace metal in argon plasma

Goal. Derive the fractional ionization $\alpha = n_i/(n_i+n_a)$ of a trace element in an ICP at temperature $T$ and electron density $n_e$ , and evaluate it for sodium ( $E_i = 5.139$ eV) at $T = 7500$ K, $n_e = 1\times10^{21}$ m⁻³.

Step 1 — Define the Saha ratio. Start from the equation above and define the right-hand side as a single quantity, the Saha constant for this element/temperature combination:

\frac{n_i n_e}{n_a} \equiv K_{\text{Saha}}(T)

so that $K_{\text{Saha}}$ absorbs the statistical weights, the phase-space prefactor, and the exponential.

Step 2 — Compute the phase-space prefactor. At $T = 7500$ K:

\left(\frac{2\pi m_e k_B T}{h^2}\right)^{3/2} = \left(\frac{2\pi (9.109\times10^{-31})(1.381\times10^{-23})(7500)}{(6.626\times10^{-34})^2}\right)^{3/2}

The bracket evaluates to $2\pi \times 9.109\times10^{-31} \times 1.381\times10^{-23} \times 7500 = 5.927\times10^{-50}$ (numerator), divided by $h^2 = 4.390\times10^{-67}$ , giving $1.350\times10^{17}$ m⁻². Raising to the 3/2 power:

(1.350\times10^{17})^{3/2} = 1.568\times10^{25}\ \text{m}^{-3}

Step 3 — Compute the exponential factor. For sodium, $E_i = 5.139$ eV $= 8.233\times10^{-19}$ J. With $k_B T = (1.381\times10^{-23})(7500) = 1.036\times10^{-19}$ J:

\frac{E_i}{k_B T} = \frac{8.233\times10^{-19}}{1.036\times10^{-19}} = 7.947

\exp(-7.947) = 3.55\times10^{-4}

Step 4 — Assemble $K_{\text{Saha}}$ . Sodium's ion (Na⁺, closed shell) has $g_i = 1$ ; the neutral ground state ${}^2S_{1/2}$ has $g_a = 2$ , so $2g_i/g_a = 1$ :

K_{\text{Saha}} = (1)(1.568\times10^{25}\ \text{m}^{-3})(3.55\times10^{-4}) = 5.57\times10^{21}\ \text{m}^{-3}

Step 5 — Solve for ionization fraction. With $n_e = 1\times10^{21}$ m⁻³ (a representative mid-plasma value), the Saha equation $n_in_e/n_a = K_{\text{Saha}}$ rearranges to

\frac{n_i}{n_a} = \frac{K_{\text{Saha}}}{n_e} = \frac{5.57\times10^{21}}{1\times10^{21}} = 5.57

so the ionization fraction is

\alpha = \frac{n_i}{n_i + n_a} = \frac{5.57}{6.57} = 0.848 \approx 85\%

Limiting case check. As $T \to 0$ , $k_BT \to 0$ , the exponential $\exp(-E_i/k_BT) \to 0$ faster than the prefactor $T^{3/2}$ grows, so $\alpha \to 0$ — no ionization at absolute zero, as physically required. As $T \to \infty$ , the exponential saturates to 1 and the $T^{3/2}$ prefactor dominates, driving $\alpha \to 1$ : essentially complete ionization, consistent with the ICP's design goal of ionizing almost every element it touches (excepting a few high- $E_i$ nonmetals such as He, Ne, F, and Ar itself, which is why Ar-based interferences persist rather than washing out).

Go deeper · Derivation B · 2 minNet signal and internal-standard ratio for drift correction

Derivation BNet signal and internal-standard ratio for drift correction

Goal. Show explicitly why dividing the analyte signal by an internal-standard signal cancels first-order plasma drift, and quantify the residual error for a asymmetric two-element case.

Step 1 — Write the raw signal model. For analyte A and internal standard IS, introduced together in solution at concentrations $C_A$ and $C_{IS}$ :

S_A = k_A(T, n_e)\,C_A, \qquad S_{IS} = k_{IS}(T,n_e)\,C_{IS}

Both $k_A$ and $k_{IS}$ depend on the same instantaneous plasma state $(T, n_e)$ , because both species pass through the identical torch, injector flow, and RF power at the same moment.

Step 2 — Model a drift event. Suppose plasma robustness falls (e.g., from a viscous matrix suppressing nebulization efficiency) such that both sensitivities scale by a common multiplicative factor $(1+\delta)$ , to first order, because both are governed by the same $T$ and $n_e$ :

k_A' = k_A(1+\delta), \qquad k_{IS}' = k_{IS}(1+\delta)

This is an approximation — real elements respond with slightly different sensitivity to drift depending on ionization energy and excited-state energy — but it holds well when A and IS have similar first ionization energies and emission/mass characteristics.

Step 3 — Form the ratio. Define the response ratio $R = S_A/S_{IS}$ :

R = \frac{k_A' C_A}{k_{IS}' C_{IS}} = \frac{k_A(1+\delta) C_A}{k_{IS}(1+\delta) C_{IS}} = \frac{k_A C_A}{k_{IS} C_{IS}}

The $(1+\delta)$ factors cancel exactly — the ratio $R$ is independent of $\delta$ to first order, so a calibration curve built from $R$ versus $C_A$ (at fixed, known $C_{IS}$ ) is immune to common-mode plasma drift.

Step 4 — Quantify residual error for unequal drift sensitivity. In practice $k_A$ and $k_{IS}$ do not respond identically; write $k_A' = k_A(1+\delta_A)$ and $k_{IS}' = k_{IS}(1+\delta_{IS})$ with $\delta_A \neq \delta_{IS}$ . Then

R' = R\cdot\frac{1+\delta_A}{1+\delta_{IS}} \approx R\left[1 + (\delta_A - \delta_{IS})\right]

using a first-order binomial expansion for small $\delta$ . Take a realistic asymmetric case: $\delta_A = -0.12$ (12% suppression of analyte, e.g. As at 5% HCl matrix loading) and $\delta_{IS} = -0.09$ (9% suppression of internal standard, e.g. Ge). Then the fractional error in $R$ is

\delta_A - \delta_{IS} = -0.12-(-0.09) = -0.03

a residual 3% bias — much smaller than the uncorrected 12% bias in $S_A$ alone, but not zero. This is exactly why internal standards must be chosen with ionization energy and mass (for ICP-MS) or excitation energy (for ICP-OES) close to the analyte: matching reduces $|\delta_A - \delta_{IS}|$ .

Limiting case check. If $\delta_A = \delta_{IS}$ exactly (perfectly matched internal standard), the residual bias vanishes identically, recovering the ideal cancellation of Step 3. If instead $\delta_{IS} = 0$ (an internal standard immune to drift, unrealistic but a useful bound), the residual bias equals the full analyte drift $\delta_A$ — showing an internal standard only helps to the extent it actually tracks the analyte's behavior.

Worked examples

Example 1 (Foundational): Ionization fraction of calcium in the ICP, and why ICP-OES can use both atom and ion lines. Calcium has first ionization energy $E_i = 6.113$ eV. Estimate its ionization fraction at $T = 7000$ K and $n_e = 1.2\times10^{21}$ m⁻³, using $2g_i/g_a = 2$ (Ca⁺ ground state $^2S_{1/2}$ , $g_i=2$ ; Ca atom ground state $^1S_0$ , $g_a=1$ ). Convert energies: $E_i = 6.113 \times 1.602\times10^{-19} = 9.793\times10^{-19}$ J. At $T=7000$ K, $k_BT = 1.381\times10^{-23}\times7000 = 9.667\times10^{-20}$ J, so

\frac{E_i}{k_BT} = \frac{9.793\times10^{-19}}{9.667\times10^{-20}} = 10.13

\exp(-10.13) = 3.98\times10^{-5}

The phase-space prefactor at $T=7000$ K scales as $T^{3/2}$ relative to the 7500 K value computed earlier:

\left(\frac{7000}{7500}\right)^{3/2} = 0.9037

(1.568\times10^{25})(0.9037) = 1.417\times10^{25}\ \text{m}^{-3}

Assemble:

K_{\text{Saha}} = (2)(1.417\times10^{25})(3.98\times10^{-5}) = 1.128\times10^{21}\ \text{m}^{-3}

\frac{n_i}{n_a} = \frac{1.128\times10^{21}}{1.2\times10^{21}} = 0.940

\alpha = \frac{0.940}{1.940} = 0.485 \approx 48\%

So calcium is roughly half ionized, half neutral, under these representative conditions — which is exactly why calcium (and many other elements with $E_i$ in the 6–8 eV range) shows both strong atom lines (Ca I 422.7 nm) and strong ion lines (Ca II 393.4, 396.8 nm) in ICP-OES spectra, unlike sodium (85% ionized, atom line dominant) or elements with $E_i < 6$ eV (nearly all-ion emission). This mixed-population behavior is a direct, quantitative consequence of the Saha balance and has no counterpart in flame emission, where essentially everything stays neutral.

Example 2 (Applied): ICP-OES determination of manganese in a groundwater sample with internal standardization. A groundwater sample is analyzed for Mn by ICP-OES using Sc (45 mg/L standard internal-standard stock) at 50.0 µg/L final concentration in every solution. Calibration standards (0.00, 5.00, 10.0, 20.0, 40.0 µg/L Mn, all spiked with 50.0 µg/L Sc) give the following net intensity ratios $R = I_{\text{Mn }257.6}/I_{\text{Sc }361.4}$ :

$C_{\text{Mn}}$ (µg/L)	0.00	5.00	10.0	20.0	40.0
$R$	0.0021	0.1046	0.2058	0.4114	0.8189

A linear least-squares fit (five points) gives slope and intercept. Compute the sums needed: $\bar C = 15.0$ µg/L, $\bar R = 0.30856$ .

S_{CC} = \sum (C_i - \bar C)^2 = (-15)^2+(-10)^2+(-5)^2+(5)^2+(25)^2 = 225+100+25+25+625 = 1000\ (\mu\text{g/L})^2

S_{CR} = \sum (C_i-\bar C)(R_i-\bar R)

Computing term by term with $R_i - \bar R$ : $-0.3065, -0.2040, -0.1028, 0.1028, 0.5103$ (for $C_i = 0,5,10,20,40$ ) and $C_i - \bar C = -15,-10,-5,5,25$ :

S_{CR} = (-15)(-0.3065)+(-10)(-0.2040)+(-5)(-0.1028)+(5)(0.1028)+(25)(0.5103)

= 4.598+2.040+0.514+0.514+12.758 = 20.42

\text{slope } m = \frac{S_{CR}}{S_{CC}} = \frac{20.42}{1000} = 0.02042\ \text{(µg/L)}^{-1}

\text{intercept } b = \bar R - m\bar C = 0.30856 - (0.02042)(15.0) = 0.30856 - 0.3063 = 0.00226

So the calibration line is $R = 0.02042\,C_{\text{Mn}} + 0.00226$ , with $R^2 > 0.9995$ (a quick residual check: at $C=20$ , predicted $R = 0.02042(20)+0.00226 = 0.4107$ , versus observed $0.4114$ — residual $0.0007$ , well within typical ICP-OES noise). The unknown groundwater sample (diluted identically, spiked with the same 50.0 µg/L Sc) gives $R_{\text{sample}} = 0.1560$ . Solve for concentration:

C_{\text{Mn}} = \frac{R_{\text{sample}} - b}{m} = \frac{0.1560 - 0.00226}{0.02042} = \frac{0.1537}{0.02042} = 7.53\ \mu\text{g/L}

If the original groundwater was diluted 1:1 with 2% HNO₃ prior to analysis (dilution factor 2), the reported concentration is

C_{\text{original}} = 7.53 \times 2 = 15.1\ \mu\text{g/L Mn}

comfortably below the EPA secondary standard of 50 µg/L for manganese in drinking water, reported to three significant figures given the quality of the calibration fit.

Example 3 (Exam/competition level): Correcting an ICP-MS arsenic determination for the ⁴⁰Ar³⁵Cl⁺ polyatomic interference at m/z = 75. Arsenic is monoisotopic at mass 75 (¹⁰⁰% natural abundance), which is analytically convenient (no isobaric overlap from another element's isotope) but suffers from a severe polyatomic interference: ⁴⁰Ar³⁵Cl⁺ also appears at nominal m/z 75, and chloride is ubiquitous in biological and environmental matrices (e.g., digested blood, seawater, saline extracts). A quadrupole ICP-MS (unit mass resolution, cannot resolve As at 74.9216 Da from ArCl at 74.9312 Da — a mass difference of only 0.0096 Da) reports raw counts at m/z 75 of 48,200 counts/s in a serum digest containing 3,200 mg/L chloride. To correct, the analyst uses the known, matrix-independent isotope ratio of chlorine (³⁵Cl : ³⁷Cl = 75.77 : 24.23) together with the ArCl signal appearing at m/z 77 (⁴⁰Ar³⁷Cl⁺), corrected in turn for the ⁷⁷Se isobaric overlap (selenium is not of interest here and assumed negligible in this matrix, a fact confirmed by running a Se-free matrix blank). Measured signal at m/z 77 is 2,150 counts/s.

Step 1. Since ⁴⁰Ar³⁵Cl⁺ and ⁴⁰Ar³⁷Cl⁺ come from the same ArCl population, their count ratio must mirror the ³⁵Cl:³⁷Cl natural abundance ratio (both chlorine isotopes react with Ar⁺ at essentially the same rate in the plasma/interface region):

\frac{I(\text{m/z }75, \text{ArCl})}{I(\text{m/z }77, \text{ArCl})} = \frac{75.77}{24.23} = 3.127 $質$ **Step 2.** Estimate the ArCl contribution at mass 75 from the measured mass-77 signal:

I_{\text{ArCl at }75} = 3.127 \times 2{,}150 = 6{,}723\ \text{counts/s}

**Step 3.** Subtract from the raw mass-75 signal to recover the true arsenic ion signal:

I_{\text{As}} = 48{,}200 - 6{,}723 = 41{,}477\ \text{counts/s} \approx 4.15\times10^4\ \text{counts/s}

**Step 4.** Convert to concentration using a matrix-matched external calibration where a 10.0 µg/L As standard (also containing 3,200 mg/L Cl to replicate the interference and any suppression) gave a background-corrected signal of 41,300 counts/s at m/z 75 (after the identical ArCl correction procedure was applied to the standard, since the standard itself also generates ArCl):

C_{\text{As, sample}} = \left(\frac{41{,}477}{41{,}300}\right)\times 10.0\ \mu\text{g/L} = 10.04\ \mu\text{g/L}

Without the correction, the raw signal (48,200 counts/s) would have been read against the same calibration and given a falsely elevated

C_{\text{uncorrected}} = \left(\frac{48{,}200}{41{,}300}\right)\times 10.0 = 11.67\ \mu\text{g/L}

a 16% positive bias — large enough to move a borderline clinical or regulatory sample across a compliance threshold. This is precisely the scenario that motivates collision-cell ICP-MS (using He collision gas with kinetic energy discrimination to preferentially attenuate the larger-cross-section polyatomic ArCl⁺ relative to the compact atomic As⁺), which removes the need for this mathematical correction but relies on the same underlying isotope-ratio logic to validate that the correction (mathematical or collisional) worked.

Limiting cases, scope & common traps

Regime / case	Behavior	Consequence
$T \to 0$ (cold plasma / flame comparison)	Saha exponential dominates, $\alpha \to 0$	Flames leave most refractory elements as neutral, undissociated oxides — the flame's core weakness that ICP fixes
$T \to \infty$	$\alpha \to 1$ for virtually all elements with $E_i \lesssim 10$ eV	Full ionization means ICP-MS "sees" nearly every metal; only high- $E_i$ nonmetals (F, Cl, N, O, noble gases) resist
High dissolved solids (>0.2–2% TDS)	Nebulizer clogging, torch instability, signal drift	Requires dilution, matrix matching, or robust (high-power, low-flow) plasma conditions
Easily ionized elements (Na, K, Ca at high concentration)	Shift plasma electron density, suppressing analyte ionization	"Matrix suppression" — corrected via internal standard or matrix-matched calibration
Isobaric atomic overlap (e.g. ⁴⁰Ar⁺ vs ⁴⁰Ca⁺)	Same nominal mass, quadrupole cannot resolve	Requires high-resolution sector-field MS, alternate isotope, or reaction-cell chemistry
Doubly charged ions (e.g. Ba²⁺ at m/z 68.5, rounds to 69, overlapping Ga)	Second ionization energy low enough that M²⁺ forms in the plasma	Monitor M²⁺/M⁺ ratio with a barium oxide/doubly-charged tune solution; correct or avoid affected masses
Equal-sensitivity limiting case (Step 3 of second derivation)	$\delta_A = \delta_{IS}$	Internal-standard ratio is drift-free to first order — the ideal, rarely fully realized

The chemist behind it

The modern ICP source traces its lineage to work in the 1960s by Stanley Greenfield in the United Kingdom and, independently and more influentially for the technique's analytical development, Velmer A. Fassel at Iowa State University and the Ames Laboratory in the United States. Fassel, working with graduate students and collaborators through the late 1960s and 1970s, systematically characterized the torch geometry, RF coupling, and spectroscopic behavior that turned a laboratory curiosity into a routine analytical tool. His group demonstrated that the toroidal, RF-induced argon plasma could atomize and excite refractory elements — rare earths, borates, uranium compounds — that flames simply could not touch, and that simultaneous multi-element spectrometry was achievable with the right optical design. Fassel's sustained, methodical optimization of plasma parameters (gas flows, power, observation height) established the operating windows still used, with only incremental refinement, in commercial instruments today.

Origins · 1 minVelmer A. Fassel and how this was discovered

Velmer A. Fassel

1919–2010 · Iowa State University and Ames Laboratory, developer of practical ICP-OES

Fassel's career exemplified translational analytical chemistry: he took a physically striking but temperamental plasma discharge and, through painstaking parametric studies of gas flow, forward power, and viewing geometry, converted it into a reproducible commercial technology. His collaborations with instrument manufacturers in the 1970s directly enabled the first commercial ICP-OES instruments, and the design principles he established — the three-tube torch, the induction-coupled toroidal discharge, simultaneous multi-element optical readout — remain recognizable in every ICP instrument sold today.

History

- 1961 — Stanley Greenfield (UK) describes RF-induced atmospheric argon plasmas as spectroscopic excitation sources for solutions.

- 1964–1965 — Velmer Fassel and coworkers at Iowa State begin systematic torch-design and plasma-diagnostic studies at Ames Laboratory.

- 1974–1975 — Commercial ICP-OES instruments become available, incorporating Fassel-type torches and simultaneous polychromator optics.

- 1980 — Alan Gray and Alan Date couple an ICP source to a quadrupole mass spectrometer, demonstrating practical ICP-MS.

- 1983 — First commercial ICP-MS instruments (Sciex ELAN, VG PlasmaQuad) reach the market, opening sub-ppt elemental and isotopic analysis.

Why is argon used as the plasma gas instead of air or nitrogen?

Argon is chemically inert, so it does not form stable, spectrally interfering molecular species (like NO or OH bands) as readily as nitrogen or oxygen would, and it has a relatively low ionization energy (15.76 eV) for a noble gas, letting the RF field sustain a self-consistent discharge efficiently. Argon's low thermal conductivity relative to helium also helps keep the plasma hot and stable at manageable RF power (typically 700–1500 W). Nitrogen-based or nitrogen-mixed plasmas exist for cost reasons in some industrial ICP-OES applications, but they are markedly harder to ignite and sustain, and they introduce their own strong molecular emission bands (N₂, NO) that clutter the optical spectrum, so pure argon remains the standard choice for both ICP-OES and ICP-MS.

How does ICP-MS achieve such dramatically lower detection limits than ICP-OES for most elements?

ICP-MS directly counts individual ions with essentially single-ion sensitivity at the detector (an electron multiplier or Faraday cup), and mass separation gives extremely low spectral background compared to optical emission, where continuum and line-rich argon/matrix backgrounds compete with weak analyte lines. Because the Saha equation predicts near-complete ionization for most metals, nearly all of the analyte atoms that survive transport into the plasma become ions available for MS extraction, so the technique is not "wasting" population on the neutral channel the way OES effectively does for its ion-line-dominant elements. Typical detection limits are sub-ppt to low-ppt for ICP-MS versus low-ppb to sub-ppb for ICP-OES, though ICP-OES retains advantages for high-concentration matrices, silicon/boron/sulfur (poor MS sensitivity or interference-prone), and simpler, lower-cost operation.

What causes "easily ionized element" (EIE) matrix suppression, and how is it corrected?

When a sample contains a high concentration of an easily ionized element (Na, K, Ca, often from seawater, biological fluids, or geological digests), that element contributes a large population of free electrons to the plasma via its own near-complete ionization, shifting the local electron density $n_e$ upward. Because the Saha equation shows $n_i/n_a$ scales inversely with $n_e$ at fixed temperature, this elevated electron density suppresses the ionization fraction of trace analytes present in the same plasma volume, lowering their ICP-MS signal (though ICP-OES ion lines are affected similarly, while atom lines can even be enhanced). The standard corrections are matrix matching (adding the same EIE background to standards and blanks), the method of standard additions, or robust/high-power plasma operating conditions that make $n_e$ less sensitive to matrix loading in the first place.

Why do ICP-MS instruments use a differentially pumped interface with sampler and skimmer cones rather than direct injection into the mass analyzer?

The plasma operates at roughly atmospheric pressure, while quadrupole and sector-field mass analyzers require high vacuum (around 10⁻⁶ torr) to allow ions to travel without colliding with background gas. The interface bridges this eight-to-nine-orders-of-magnitude pressure gap in stages: the sampler cone extracts a supersonic jet of plasma gas and ions into an intermediate vacuum region (roughly 1–3 torr), and the skimmer cone selects the central, coolest, most ion-rich portion of that jet into a second, lower-pressure region before ion-lens optics focus the beam into the analyzer's high vacuum. This staged approach minimizes both gas load on the vacuum pumps and space-charge-induced loss of light ions in the presence of the dominant Ar⁺ beam.

What is a collision/reaction cell, and why does it matter for ICP-MS accuracy?

A collision/reaction cell is a pressurized multipole (often a quadrupole or hexapole operated in RF-only "guiding" mode) placed between the ion source interface and the analyzing mass filter, filled with a light collision gas (typically helium) or a reactive gas (hydrogen, ammonia, oxygen). Polyatomic interferences (like ⁴⁰Ar³⁵Cl⁺ or ⁴⁰Ar¹⁶O⁺) have larger collision cross-sections than compact atomic analyte ions of the same nominal mass, so they undergo more frequent collisions with helium and lose more kinetic energy; an energy barrier at the cell exit (kinetic energy discrimination) then preferentially rejects these polyatomics while passing the analyte ion. This removes many polyatomic interferences (like the ArCl problem at mass 75 discussed in the worked examples) without requiring the mathematical isotope-ratio correction, though it comes at some cost in overall sensitivity.

Can ICP techniques analyze solid samples directly, without acid digestion?

Yes — laser ablation ICP-MS (LA-ICP-MS) focuses a pulsed laser (commonly Nd:YAG at 213 or 266 nm, or excimer lasers at 193 nm) onto a solid sample surface, vaporizing a micron-to-tens-of-micron spot into a fine particulate aerosol that is swept by a helium or argon carrier gas directly into the ICP torch. This bypasses acid digestion entirely, preserves spatial information (useful for elemental mapping across mineral zonation, tree rings, or archaeological artifacts), and avoids contamination or loss risks associated with wet digestion, though it introduces its own calibration challenges because matrix-matched solid reference materials are harder to obtain than aqueous standards, and ablation yield can vary with the sample's optical and mechanical properties.

How does ICP-OES achieve simultaneous multi-element analysis, and what limits how many elements can be measured at once?

Modern ICP-OES instruments use echelle-grating polychromators combined with two-dimensional CCD or CID (charge-injection device) detector arrays, which disperse light by both wavelength order and diffraction angle onto a two-dimensional detector surface, capturing the full useful wavelength range (roughly 165–800 nm) simultaneously rather than scanning sequentially. This lets an instrument record dozens of elements' emission lines from a single plasma observation in seconds, dramatically improving throughput over older sequential scanning monochromators. The practical limits are spectral line overlap in complex matrices (requiring careful line selection and interelement correction factors), detector dynamic range, and the fundamental Saha/Boltzmann-governed emission intensity of each chosen line at the plasma's excitation temperature.