ICP-MS and ICP-OES: Trace-Element Fingerprinting

An inductively coupled plasma is formed by passing argon gas through a quartz torch surrounded by a radio-frequency coil. The coil induces a fluctuating magnetic field that sustains a plasma discharge at 6000-10000 K. Any sample introduced as an aerosol is instantly desolvated, vaporised, atomised, and ionised. The efficiency of ionisation depends on the element's ionisation energy: most elements with ionisation potentials below about 15 eV (which covers almost everything geochemically interesting) are ionised to 90-100% efficiency. Only a few problematic exceptions (fluorine, oxygen, nitrogen, noble gases) are not ionised well.

For ICP-OES the plasma serves as both the atomisation and excitation source. Excited atoms emit characteristic wavelengths of light as they return to ground state. A polychromator or echelle spectrometer disperses the emission and a detector array measures the intensities simultaneously. For ICP-MS the plasma serves as the ion source only: the ions are extracted through a pair of metal cones (sampler and skimmer) into a high-vacuum mass spectrometer where they are separated by mass-to-charge ratio and counted by a detector such as a Faraday cup or an electron multiplier.

Solution-mode ICP-MS requires the sample to be fully dissolved. For geological materials the standard approach is acid digestion. The most common procedure uses aqua regia (three parts hydrochloric acid to one part nitric acid by volume), which dissolves most silicate minerals, many sulphides, and organic matter. Resistant phases such as zircon, chromite, and cassiterite require hydrofluoric acid, either in an open-vessel hot-plate digest or in a sealed microwave vessel that drives the digestion at elevated temperature and pressure to completion.

1. Weighing and drying
   Typically 0.1-0.5 g of fine-ground, 105-degree-dried soil is weighed into a vessel. Smaller masses reduce reagent blanks and are used when material is scarce, as in forensic trace soil.
2. Acid addition and digestion
   Aqua regia or HF-HNO3 mixtures are added. Microwave digestion at 180-220 degrees Celsius in sealed Teflon vessels takes 20-45 minutes and gives complete dissolution for most geological materials, minimising volatile element loss (arsenic, selenium, mercury).
3. Evaporation and reconstitution
   After digestion, excess HF (which would damage the quartz torch) is evaporated by heating with perchloric or nitric acid. The residue is reconstituted in dilute nitric acid and made up to a known volume, typically 50-100 mL.
4. Internal standard addition
   An element not naturally present at significant levels in the sample (indium, rhenium, or bismuth are common choices) is added at a known concentration to every solution. Monitoring its signal corrects for instrument drift and matrix-suppression effects during the run.

Laser ablation couples a pulsed UV laser (typically Nd:YAG at 213 nm or excimer at 193 nm) to the ICP-MS via a sealed ablation cell. The laser is focused to a spot 20-200 micrometres in diameter on the polished surface of the sample, held under helium. A brief pulse ablates 10-50 nanograms of material per shot. The ablated vapour is swept by helium carrier gas into the plasma and analysed as a transient pulse. Repeating the shot at the same spot drills down; scanning the beam across the surface produces a trace across the grain.

This matters for forensic geology because many trace-soil samples contain only a handful of diagnostic grains. A single zircon crystal from a questioned soil, ten to hundreds of micrometres across, can be analysed for its full lanthanide pattern and hafnium isotope ratio without destroying the grain. The result is an in-situ geochemical signature that places the grain in a geological province even when the bulk sample is too small for conventional digestion.

The rare earth elements (REE) sit in the first row of the f-block, atomic numbers 57 (lanthanum) to 71 (lutetium), plus yttrium at 39. They are geochemically coherent: all trivalent in most geological settings, all with similar ionic radii that decrease smoothly from La to Lu. That smooth size change is why REEs fractionate systematically during magmatic and sedimentary processes. Heavy REEs concentrate in small, dense accessory minerals (zircon, garnet); light REEs prefer feldspars and carbonates. The specific fractionation pattern is a function of the source rock composition and the temperature and pressure at which crystallisation happened.

To make REE patterns comparable across samples, concentrations are divided by a reference material. Chondrite meteorites represent the primitive (undifferentiated) solar abundance; the Post-Archean Australian Shale (PAAS) is a widely used average crustal sedimentary reference. After normalisation, patterns are plotted as concentration ratio vs. atomic number (La to Lu on the x-axis, normalised concentration on a log y-axis). The shape of this curve, its slope from light to heavy REEs, and specific anomalies (cerium and europium have characteristic anomalies that reflect oxidation conditions and feldspar crystallisation) encode the geological history of the material.

A solution ICP-MS run on a soil can return concentrations for 50-70 elements. Using all of them simultaneously requires multivariate methods. Principal component analysis (PCA) condenses the data into a small number of orthogonal dimensions that capture most of the variance. Plotting the first two or three principal components separates soil populations that would overlap on any single bivariate plot. Discriminant function analysis (DFA) goes further: trained on a reference database of soils from known locations, it assigns an unknown sample to the most probable source population with a quantified posterior probability.

For gem and mineral provenance work, where the question is often whether a stone is from mine A, mine B, or another source entirely, a reference database of confirmed-origin stones is built up over years. The 'geochemical fingerprint' of each mine differs in its trace-element ratios (Cr, V, Fe in emeralds; Ti, Fe, Cr, V in rubies; REEs and isotopes in diamonds). A questioned stone is measured, projected onto the PCA space, and compared against the reference clusters.

Major elements present at percent levels (silicon, aluminium, iron, calcium, magnesium, sodium, potassium, titanium, phosphorus) saturate the electron multiplier detector in ICP-MS, producing nonlinear responses that require heavy dilution of the solution and loss of detection limit for true trace elements. ICP-OES handles these elements well: the optical detector has a far larger linear dynamic range, so major and minor elements at parts-per-million to percent levels are measured accurately without dilution.

A complete geochemical characterisation of a forensic soil therefore typically uses ICP-OES for the ten major-element oxides (cross-checked against fused-bead WD-XRF), ICP-MS for the forty or more trace and ultra-trace elements, and LA-ICP-MS when individual grains need to be characterised. This three-technique combination covers the entire periodic table of geochemically relevant elements and provides the most defensible dataset for provenance comparison.

The sensitivity of ICP-MS is both its strength and its quality-assurance challenge. At parts-per-trillion concentrations, contamination from reagents, vessels, and laboratory air can contribute measurable signals. Certified reference materials (CRMs), such as NIST SRM 2780 (hard rock mine waste), USGS BHVO-2 (basalt), and BCR-2, are measured at the start and end of every analytical batch. Agreement within the certified uncertainty range of the CRM is the primary quality indicator.

• Proficiency schemes: organisations such as GEOPT (Geochemical Earth Reference Panel) circulate unknown samples to participating laboratories worldwide. Published z-scores show how each laboratory compares to the consensus. Forensic laboratories are expected to demonstrate ongoing proficiency by participating in these or equivalent schemes.
• Method validation: before a new soil matrix type is analysed in a forensic context, accuracy and precision should be validated using a CRM with a similar matrix. Spike recovery experiments (adding a known amount of analyte to a digested soil blank) check for losses or additions in the digestion step.
• Uncertainty budgeting: a forensic ICP-MS report must state measurement uncertainties for each element. Without them, the comparison of two soils cannot be framed probabilistically, and the analyst cannot explain to a court whether two values that differ by 3% represent a real difference or measurement noise.