Diffusive exchange of trace elements between alkaline melts: implications for element fractionation and timescale estimations during magma mixing

The diffusive exchange of 30 trace elements during the interaction of natural mafic and silicic alkaline melts was experimentally studied at conditions relevant to shallow magmatic systems. In detail, a set of 12 diffusion couple experiments have been performed between natural shoshonitic and rhyolitic melts from the Vulcano Island (Aeolian archipelago, Italy) at a temperature of 1200 {\deg}C, pressures from 50 to 500 MPa, and water contents ranging from nominally dry to ca. 2 wt. %. Concentration-distance profiles, measured by Laser Ablation ICP-MS, highlight different behaviours, and trace elements were divided into two groups: (1) elements with normal diffusion profiles (13 elements, mainly low field strength and transition elements), and (2) elements showing uphill diffusion (17 elements including Y, Zr, Nb, Pb and rare earth elements, except Eu). For the elements showing normal diffusion profiles, chemical diffusion coefficients were estimated using a concentration-dependent evaluation method, and values are given at four intermediate compositions (SiO2 equal to 58, 62, 66 and 70 wt. %, respectively). A general coupling of diffusion coefficients to silica diffusivity is observed, and variations in systematics are observed between mafic and silicic compositions. Results show that water plays a decisive role on diffusive rates in the studied conditions, producing an enhancement between 0.4 and 0.7 log units per 1 wt.% of added H2O. Particularly notable is the behaviour of the trivalent-only REEs (La to Nd and Gd to Lu), with strong uphill diffusion minima, diminishing from light to heavy REEs. Modelling of REE profiles by a modified effective binary diffusion model indicates that activity gradients induced by the SiO2 concentration contrast are responsible for their development, inducing a transient partitioning of REEs towards the shoshonitic melt.

💡 Research Summary

The authors investigated the diffusive exchange of thirty trace elements between natural mafic (shoshonitic) and silicic (rhyolitic) alkaline melts under conditions relevant to shallow magmatic systems. Twelve diffusion‑couple experiments were performed at 1200 °C, pressures ranging from 50 to 500 MPa, and water contents from nominally dry to approximately 2 wt % H₂O. Concentration‑distance profiles were obtained by laser‑ablation ICP‑MS. The trace elements segregated into two distinct behavioural groups. The first group (13 elements, mainly low‑field‑strength and transition metals such as Na, K, Ca, Ti, Mn, Fe, Co, Ni, Cu, Zn, Mo, Sn, Sb) displayed normal, Fickian diffusion profiles. For these, concentration‑dependent diffusion coefficients were calculated at four intermediate melt compositions (SiO₂ = 58, 62, 66, and 70 wt %). A clear coupling between trace‑element diffusivity and silica diffusivity was observed, with diffusivities decreasing as SiO₂ content increased, reflecting the increasing polymerisation of the melt network.

The second group (17 elements, including Y, Zr, Nb, Pb and virtually all rare‑earth elements except Eu) exhibited uphill diffusion, i.e., migration against the concentration gradient. The trivalent‑only REEs (La‑Nd and Gd‑Lu) showed pronounced minima in their profiles, with the intensity of the minima decreasing from light to heavy REEs. The authors attribute this behaviour to activity‑gradient driven fluxes generated by the strong SiO₂ concentration contrast between the two melts. A modified effective binary diffusion (EBD) model that incorporates activity gradients successfully reproduces the REE profiles, confirming that transient partitioning of REEs toward the shoshonitic melt is driven by silica‑induced chemical potential differences.

Water proved to be a decisive factor: each additional 1 wt % H₂O enhanced diffusion coefficients by 0.4–0.7 log units across the entire suite of elements. This enhancement is consistent with the well‑known role of water in depolymerising silicate melts and increasing ionic mobility. Pressure variations within the experimental range had a comparatively minor effect on diffusivity, indicating that under the studied temperature regime water content dominates the diffusion kinetics.

The study provides several key implications for magmatic processes. First, the coupling of trace‑element diffusivity to silica diffusivity and water content means that element fractionation during magma mixing cannot be interpreted using constant diffusion coefficients; instead, composition‑dependent diffusivities must be employed. Second, the occurrence of uphill diffusion for REEs and other high‑field‑strength elements highlights the importance of non‑Fickian fluxes driven by activity gradients, especially in systems with large compositional contrasts. Third, the quantitative relationship between water content and diffusion enhancement offers a potential proxy for estimating the water budget of natural magmas from observed diffusion‑controlled zoning. Finally, the successful application of the modified EBD model provides a framework for back‑calculating mixing durations: by fitting observed concentration profiles, one can constrain the time elapsed since the onset of mixing, the initial compositional contrast, and the water content of the interacting melts.

Overall, this work advances our understanding of trace‑element transport in alkaline magmas, demonstrates the pivotal role of water and silica polymerisation in controlling diffusion rates, and introduces a robust modelling approach for interpreting complex diffusion signatures in natural volcanic rocks.