Trace elements behaviour in direct- and indirect iron metallurgy : the case of Pays de Bray (France)
Résumé
This study aims to understand the behaviour of trace elements in direct- and indirect-iron metallurgy in order recognize the provenance region of iron objects and the metallurgical processes involved. Major elements, such as P and Mn, help in discriminating some ores (minette de Lorraine) but in general fail in recognizing the region of provenance making necessary the use of trace elements. To address these goals we have chosen the Pays de Bray region where the succession of the two techniques and the diffusion of the iron products are well-documented. Ores, slags, cast iron and final products have been studied and analysed in the following sites: Glinet (indirect, 15th -16th century), le Prés de Montadet (direct, from the 1st century BC to the 1st century AD), le Chemin des Flots (direct, 14th century). Some experimental reductions have been made using geological ores directly collected near to the archaeological sites. Since the Pays de Bray has been an important supplier for building irons of Rouen and Amiens during the Middle Age we plan to analyse the building irons of the Rouen and Amiens cathedrals and of the church of Saint Ouen. Compositional studies have been carried out on a hundred of bulk ores, slags, cast iron and final products. We also studied major and trace element composition of trapped slag inclusions in the final iron objects in both archaeological and experimental materials.Bulk major element composition has been measured by Energy Dispersive Spectrometry coupled with a Scanning Electron Microscope. Bulk trace element composition has been determined by Neutron Activation and ICP-MS (Inductively Coupled Plasma-Mass Spectrometry). Trapped slag inclusions have been analysed by microbeam laser ablation coupled with ICP-MS.During the indirect process only Co, Ni and Cu completely move to cast iron. All other elements partition between fluid-slag (laitier) and cast iron. By a consequence the inclusions formed after cast iron refining are strongly depleted in trace elements which have affinity with the silicate phases (La, Ce, Hf …). Such inclusions therefore forget the ore chemical signature.On the contrary, in the direct process, a large spectrum of trace elements strongly partition from ore to slag. They are hosted in trapped slag inclusions in the final iron. Therefore trapped slag inclusions allow discriminating between the direct and indirect processes based on their relative trace element enrichment. We focused on non-volatile, not sensitive to alteration and negligible trace elements in the hearth construction materials and in the charcoal ash. By using their concentration ratios, we demonstrate that is possible to trace the chemical signature from ores to final products in the case of direct process.Based on a multi-elemental statistic approach we show that the chemical signature of the Pays de Bray is well separated among other French reference sites (after PalSid data base) giving a powerful tool to discriminate the provenance of the irons as for instance those of the cathedrals of Rouen and Amiens.