The direction and magnitude of magnesium (Mg) isotope fractionation attendant to the formation of clay minerals is fundamental to the use of Mg isotopes to decipher the biogeochemical cycling of Mg in the critical zone and for the oceanic Mg budget. This study provides experimental data on the Mg fractionation factor for two smectite- group minerals (stevensite and saponite) at temperatures relevant for Earth surface processes. The resultant solids were characterised by X-ray di↵raction (XRD) and Fourier-transform infrared spectroscopy (FT-IR) to confirm the mineralogy and crystallinity of the product. A series of experiments were performed to asses the impact of temperature and pH on isotope fractionation. Bulk solid samples were treated with ammonium chloride to remove exchangeable Mg in order to distinguish the Mg isotopic fractionation between these sites and octahedral sites. All bulk and residual solids were enriched in 24Mg compared to the initial solution and 26Mg values of the exchangeable pool were lower than, or within error of, the initial solution. Final solutions were either within error of, or enriched in, 26Mg compared to the initial solution, depending on the fraction of Mg removed from solution (f Mg) For experiments with similar f Mg, increasing the pH resulted in a higher reaction rate and reduced fractionation from the initial solution. This could point to a kinetic effect, but we note composition of the residual solid (Li/Mg ratio) was also dependent on pH. The change in the Li/Mg ratio was reflected in the wavenumber of the Mg3- OH stretch in FT-IR data, which is a proxy for bond strength, and suggests an equilibrium control. An equilibrium control is further supported by the observation of reduced fractionation compared to the initial solution with increasing temperature. Rayleigh and batch fractionation models were fitted to the data giving fractionation factors of 0.9991 and 0.9990 respectively.

We compare our results with existing field and experimental data and suggest that the apparent contradictions surrounding the direction of Mg isotope fractionation into phyllosilicate minerals could be due to the similarity of Mg bond lengths between clay octahedral sites and dissolved Mg. Thus small changes in mineral structure or initial solution conditions may result in a change in bond length suffcient to alter the direction of fractionation, implying that the magnitude and direction of Mg isotope fractionation into clay minerals could be dependent on local field conditions. Alternatively, if the precipitation of secondary clay minerals in the field preferentially incorporates light Mg, as observed in this experimental study, this implies the contribution of carbonate weathering to dissolved Mg fluxes has been underestimated, with major implications for the global biogeochemical cycle of Mg.