X-ray Fluorescence (XRF) in Soil and Rock Analysis

When a high-energy X-ray photon strikes an atom, it can eject an inner-shell electron, leaving a vacancy. An outer-shell electron drops down to fill that vacancy, releasing the energy difference as a photon. Because the electron shells in each element sit at fixed energies governed by the nuclear charge, the emitted photon has a characteristic wavelength (or energy) specific to that element. Measure the energy, identify the element. Measure the intensity, quantify its concentration. This is the entire physical basis of XRF.

The lines of interest in geological analysis are named by the shell transitions. K-alpha lines, the most intense, arise from the L-to-K transition and are used for elements from sodium to barium in most laboratory instruments. L-lines from heavier elements fill in the higher atomic numbers. The instrument detects each line separately, building a spectrum that lists every element present above the detection threshold.

Both instrument types measure the same fluorescence, but they sort the photons differently. WD-XRF passes the emitted beam through a diffraction crystal (lithium fluoride, pentaerythritol, or others depending on the energy range), which spreads wavelengths spatially so that only one wavelength reaches the detector at a time. ED-XRF skips the crystal: a semiconductor detector (typically silicon drift) produces a voltage pulse proportional to each incoming photon's energy, and software bins the pulses into a spectrum.

For forensic purposes, WD-XRF run in a certified laboratory with pressed-pellet or fused-bead preparation is the gold standard for court submission. Portable ED-XRF is the workhorse for field triage: an investigator can press a small patch of soil from a boot sole against the instrument window, obtain a profile in under a minute, and decide whether the location warrants full sample collection. The two tiers complement each other rather than compete.

XRF reads the top few micrometres of a sample. If the surface is rough, heterogeneous, or contains large particles, the result reflects the local chemistry of that surface patch rather than the bulk soil. Rigorous preparation is therefore essential: it determines whether the result reflects bulk soil chemistry rather than the local composition of the surface patch under the beam.

1. Drying and sieving
   Soil is dried at 105 degrees Celsius to drive off hygroscopic water, then sieved to less than 75 micrometres. This removes coarse mineral grains whose individual compositions may dominate a rough surface and reduces particle-size matrix effects.
2. Pressed pellet
   The dried powder is mixed with a wax or cellulose binder and pressed under 15-30 tonnes into a flat disc. The result is mechanically stable, has a smooth surface, and can be measured within an hour of collection. Pressed pellets work well for trace-element analysis; they retain some particle-size effect for major elements.
3. Fused bead
   About 0.5 g of ignited soil is mixed with 5 g of lithium tetraborate flux and fused in a platinum crucible at 1050-1100 degrees Celsius. The melt is poured onto a flat mould to cool as a glass disc. The dilution and glass matrix destroy all mineralogical heterogeneity, eliminating particle-size and inter-element absorption effects. Fused beads give the best accuracy for major-element oxide analysis (SiO2, Al2O3, Fe2O3, CaO, MgO) but dilute trace elements, reducing sensitivity.

A typical soil XRF report lists ten to fifteen major and minor element oxides (SiO2 typically 40-80%, Al2O3 5-20%, Fe2O3 1-10%, and so on) plus twenty or more trace elements in parts per million. The ratios and absolute values together encode the geology of the parent rock, the weathering history, and any anthropogenic additions such as slag, ash, or pesticides.

Geochemists use a range of discrimination plots to visualise where an unknown soil plots relative to reference populations. Classic bivariate plots (Zr vs. TiO2, for example) can separate soils derived from different rock types. Ternary plots of alkali-silica-iron discriminate volcanic from sedimentary and metamorphic parents. In forensic work, a simpler but powerful approach is to compare the full multi-element profile of the trace soil against reference samples from candidate locations, computing a similarity measure or applying multivariate statistics.

Trace elements such as barium, strontium, rubidium, zirconium, niobium, and the rare earth elements are especially useful forensically because they vary strongly at the centimetre-to-metre scale in heterogeneous soils. Two adjacent fields on different parent materials can differ by an order of magnitude in zirconium. For geochemically immobile elements such as zirconium, rubidium, niobium, and the rare earth elements, these variations are stable over years and are not easily altered by weathering on human timescales. Mobile trace elements such as zinc and barium can, however, be elevated by agricultural inputs including phosphate fertilisers and manure.

Modern handheld XRF analysers weigh under 2 kg and return a 30-element spectrum in 30 to 120 seconds. In a forensic context they have two main field roles. The first is rapid exclusion: if a soil on a suspect's clothing matches none of the reference soils from candidate scenes, resources can be redirected before any laboratory time is spent. The second is scene characterisation: an investigator can map elemental variation across a burial site, construction area, or flood plain, identifying anomalous zones worth sampling.

• Preparation shortcut and its cost: pXRF is often used on minimally prepared surfaces. Results carry larger uncertainties than laboratory measurements, particularly for elements below atomic number 22 (titanium). Certified reference materials should be measured before and after each session to anchor the data.
• Moisture effects: water in the sample absorbs low-energy X-rays, suppressing the apparent concentration of light elements. Wet soils measured in the field can differ noticeably from the same soil dried and re-measured. Documenting moisture content at the time of measurement is good practice.
• Heterogeneous surfaces: the beam footprint is typically 10-20 mm in diameter. Coarse grit, pebbles, or organic fragments within that footprint bias the result. Taking three or more measurements from different spots and averaging is standard for pXRF work.

Precision describes how closely repeated measurements of the same sample agree with each other. For pressed-pellet major elements on a WD-XRF instrument with routine calibration, relative precision is typically 0.5-1.5% relative. Accuracy describes how close the reported value is to the true value, which requires calibration against certified reference materials such as USGS or NIST geological standards.

Matrix effects are the most important systematic error source. If the soil is rich in iron, iron absorbs the K-alpha lines of elements near it in atomic number, causing their apparent concentrations to drop. Mathematical correction approaches include the influence-coefficient method (empirical, calibrated on a specific soil type) and the fundamental-parameters method (physics-based, more general). Fused-bead preparation reduces matrix effects so thoroughly that the same calibration curve works for most geological materials.

XRF soil comparison in forensic work follows a logic that mirrors any trace-evidence comparison: characterise the questioned material, characterise the reference material, and assess whether the profiles are consistent with a common origin or clearly different. The elemental profile is one layer of evidence; it is typically combined with colour (Munsell notation), particle size, and mineralogy for a stronger composite discrimination.

One documented application type is footwear soil in criminal investigations, where soil from the outsole of a seized boot is compared against soil from a suspect location (a field boundary, a burial site, a restricted industrial area). If the questioned soil and reference soil share major-element ratios and trace-element concentrations within measurement uncertainty, and other candidate soils from nearby areas fall outside that range, the comparison supports an association. The strength of the association depends on how distinctive the profile is geographically.