Identification des processus géochimiques entre deux signatures

The Quick Detection module, Process Identification tab, determines which geochemical or industrial processes are most likely to explain the δ gap between a known source signature and a measured final signature. It relies on the ML engine and the Nexus fractionation database.

The Problem This Tab Solves

Two measurements are available: the isotopic signature of a stibnite ore (δ¹²³Sb/¹²¹Sb = +0.43‰) and the signature of a sediment sampled 3 km downstream from the mine (δ = +0.81‰). The gap is +0.38‰. But which process produced this isotopic enrichment? Adsorption onto ferrihydrite? Partial Sb(III) → Sb(V) oxidation? A combination of both? And in what order?

Without rigorous identification of the dominant process, the Nexus workflow cannot be correctly parameterised for provenance searching or traceability. This tab serves precisely to identify candidate processes before building a complete workflow.

How the Engine Identifies Processes

The engine compares the total δ gap (Δδ = δ_final − δ_source) against the IsoFind fractionation database, which catalogues experimentally measured ε values for each process type, each element, each mineral phase. It uses two analysis layers:

• ML Layer 1 (speciation): predicts the distribution of chemical species under the provided conditions (pH, Eh, T), allowing adjustment of expected ε values according to the dominant redox state.
• ML Layer 2 (adsorption): identifies mineral phases likely to adsorb the element under the provided conditions and their associated fractionation coefficients.

For each candidate process, the engine calculates the sigma distance between its characteristic ε and the observed Δδ, weighted by analytical uncertainties and uncertainties on ε parameters. Processes are ranked by decreasing composite confidence.

Accessing the Tab

Step-by-Step Workflow

1. Enter the Initial Signature (Source) Enter the δ value of the known source (raw ore, geological reference sample, input signature of a process). Use the From Database button to load a registered sample directly.
2. Enter the Final Signature (Measured Product) Enter the measured δ value in the transformed sample (water, sediment, soil, industrial product). IsoFind automatically calculates Δδ = δ_final − δ_source, which is the gap to be explained.
3. Filter Process Types The multi-selection list allows restricting the search to process types consistent with the study context. In a purely environmental context, uncheck industrial processes (smelting, electrolysis) to avoid false positives.
4. Run the Identification Click Identify Processes. The results panel displays candidate process chains ranked by sigma fit, with ε detail per step and associated bibliographic references.

Reading the Results

Each result corresponds to a process or combination of processes whose total ε is compatible with the observed Δδ. The result card shows:

Practical Case: Identifying the Process in the Oruro Basin

Observed Δδ: δ_source (MIN-OR-001) = +0.430‰, δ_final (SED-OR-001) = +0.551‰. Δδ = +0.121‰.

With the Dissolution, Adsorption, Precipitation, Redox, Biological filters enabled:

• Adsorption onto ferrihydrite: ε = +0.15‰, distance 0.4σ (highly compatible)
• Adsorption onto goethite: ε = +0.18‰, distance 0.8σ (compatible)
• Sb(III)→Sb(V) oxidation: ε = +0.30‰, distance 2.1σ (possible but less likely)

Conclusion: adsorption onto ferrihydrite is the process most compatible with the observed gap for this sediment sampled in an acid mine drainage context, consistent with the documented presence of ferrihydrite in AMD sediments. This process becomes the primary candidate to insert into the Nexus workflow for provenance searching.

Link to the Full Nexus Module

When identification returns several plausible candidate processes (sigma distances close between multiple types), best practice is to build competing workflows in the Nexus and compare their predicted signatures. The Open in Nexus link transfers the current parameters to the canvas to facilitate this step.

For situations where the Δδ is too large to be explained by a single process, the Nexus multi-step mode allows chaining several successive processes. The ML Layer 1 and 2 engine adjusts coefficients according to the geochemical conditions at each step in the chain. See the Workflow page for building a complete multi-step workflow.