Carbon mapping in steel using 12C(d,pγ)13C in external beam

Carbon concentration and distribution has a dramatic impact on mechanical properties of steel. The manufacture technology of ancient steel is an important question in archaeometallurgy, especially during medieval times when the production process evolves between indirect and direct modes. Steel is an iron alloy incorporating up to a few % of carbon in weight, heterogeneously clustered in various carbide phases that are usually imaged by metallographic microscopy after acid etching. We have investigated the capability to obtain in a direct, non-destructive way and quantitative maps of carbon in steel samples using the 12C(d,pγ)13C reaction with the simultaneous detection of the emission of the 3089 keV γ-ray (d-PIGE) and of the p0 proton group (d-NRA). Carbon spot measurements and mappings were carried out using an external deuteron microbeam of 1.5 MeV at the New AGLAE facility of the Centre de Recherche et de Restauration des Musées de France. The beam diameter on the target was around 50 µm and its mean intensity 4nA, monitored by using the PIXE signal (Fe Kα line) emitted by the sample. Spectra were fitted using the SIMNRA program and the calculated composition checked against standard reference steel targets. Carbon distribution maps were extracted from the collected γ-ray and proton spectra datacubes. Spot measurements acquired in 2 min yielded a limit of detection of 200 ppm carbon weight for d-PIGE and 10 ppm for d-NRA. Proton detection appeared more effective than γ-ray detection owing to better statistics, mainly due to higher detection efficiency and better S/N ratio while γ-rays detection is affected by the spurious emission from nitrogen contained in the exit foil. Carbon concentration was mapped on a medieval archaeological iron alloy sample (140 × 32 points, 7 mm × 1.6 mm area) using a 1.6 sec dwell time, yielding a carbon content of 0.395 % with an uncertainty of 10 %. Heterogeneities could be evidenced in the maps, and incidentally the presence of oxygen. A simple methodology was developed to derive the depth distribution maps of carbon in the first microns of the sample.