The role of the crystallization kinetics on trace element partitioning between clinopyroxene and K-basaltic melts: investigation on a primitive composition from the Campi Flegrei Volcanic district (Italy)

The role of the crystallization kinetics on trace elements partitioning behaviour between clinopyroxene and alkaline silicate glass have been investigated experimentally. Time-series experiments were conducted on a hydrous (~1 - 4 wt% of H2O) K-basalt from Procida Island (Campi Flegrei Volcanic District, south Italy) at temperatures of 1080-1250 °C and at 0.8 GPa of pressure, conditions relevant for deep magmatic reservoirs. In general, crystallization products show that large ion lithophile elements (LILE) are incompatible (e.g., DSr ≤ 0.15, where D is partition coefficients between clinopyroxene and melt), light rare elements (LREE; e.g., DLa ≤ 0.20) are always more incompatible than heavy rare elements (HREE), which in some cases result to be compatible with clinopyroxene (e.g., DDy = 1.40); high field strength elements (HFSE) are generally incompatible (DHFSE ≤ 0.8), while transition elements (TE) range from slightly incompatible (e.g., DV = 0.6) to highly compatible (e.g., DCr = 63). Growth rate manifests a clear influence on the REE element partition coefficients as denoted by the highest DREE values calculated in the runs with the highest growth rate (~10-7 cm s-1). This is due to the less efficient rejection of incompatible elements during rapid crystal growth. Since growth rates are fast at the beginning of the crystallization process and then decrease with increasing time, Di values should decrease with increasing time. Therefore, the apparent increase in DREE values with time noticed in some runs is not ascribable to a change in time but rather to the dissimilar degrees of polymerization, expressed as the ratios NBO/T of these melts, strictly linked to a loss of Fe arisen during the experiments, and thus to a different melt viscosity. Indeed, the low NBO/T ratio calculated in the run with longer duration (9 hours) is associated to its higher loss of Fe with respect to the other run at 3 and 6 hours, pointing out the strong influence of melt structure and thus of melt viscosity on the partitioning behaviour of trace elements between clinopyroxene and basaltic melt. Finally, we used the calculated experimental clinopyroxene/melt partition coefficients to model the deepest step of the magmatic differentiation from a basaltic to a trachybasaltic composition in the Campi Flegrei Volcanic District, that results to be well reproduced by the fractionation of about 20-30% of a clinopyroxene >> olivine mineral assemblage from a basaltic parental magma.