Application Notes
Author: Oxford Instruments
Published: 05 Mar 2021 · Last updated: 05 Mar 2021
Tags: EDS
Detailed surface analysis of samples which are extremely large in size and contain a distribution of object sizes from micro- to nano-scale can pose a challenge when faced with a limited analysis time. Obtaining elemental maps at low magnifications by conventional analysis at relatively low pixel density can misrepresent the chemical and structural composition. This is particularly true in areas such as archaeology, geology, and materials science where micro- and nano-structure can have important implications, and samples are often of unknown morphology and chemistry. In this application note we describe how AZtecEnergy Large Area Mapping can be used as a research tool to maximise the level of surface characterisation, in terms of phase distributions and chemistry, while saving time and effort.
At the British Museum in London, in collaboration with the Early Mines Research Group, metallurgists have investigated the early Bronze Age smelting strategies used around the British Isles by employing SEM-EDS analysis (Craddock et al 2007). Early Copper and Bronze Age mines discovered in Ireland (O'Brien 2004) and Wales (Chapman 1997) show few metallurgical smelting waste products, in particular slag, that are often associated with ancient copper smelting. Oxidised copper ores were probably the most easily smelted but there are large deposits of exploited sulphidic copper ore, chalcopyrite, suggesting this was also used (Timberlake 2003).
To replicate possible ancient processes to extract copper metal, smelting experiments were carried out on samples of this ore. The aim of this study was to locate and understand the distribution of the small copper metal prills formed from the ore smelting process and to accurately determine their purity. A high concentration of copper metal would signify a successful smelting experiment while the absence of silicaceous slag would show the success of the non-slagging experimental ore reduction. Furthermore, accurate elemental maps and PhaseMaps of the sulphur and iron regions could be used to highlight the underlying process surrounding the copper metal formation. With AZtec® Large Area Mapping, all this information can be collected simultaneously in one automated run on a large area of a polished sample, with enough information to continue an in-depth analysis offline.
To cover the entire sample (5×3 mm), approximately 900 individual electron image and X-ray spectral map datacubes were collected in a grid pattern with AZtecEnergy Large Area Mapping software and an X-MaxN 80 mm² SDD detector. This allows the overall structure of the sample to be viewed while still maintaining the high resolution of each individual field. AZtec automatically aligns and 'montages' the 900 fields together to produce one individual area which contains all the electron image and X-ray spectral data. These montaged areas were used to investigate the sample at a micro- and nano-scale to locate and study the distribution of the copper and other minerals, as shown in Figure 1. The red regions correspond to copper-rich minerals and metal; it became apparent that the metal was extending out into the porosity of the bulk structure as it cooled from the smelting process, forming copper metal 'prills' (Figure 1 inset). In addition, the original chalcopyrite ore was fairly homogenous and pure, so the large intrusive silica (sand) grains seen (in light brown below) within the matrix proves the reaction product was in the molten state and the sand became fused in-situ.
Fig. 1. X-ray map produced by combining 900 individually collected fields. The multi-coloured ‘Layered Image’ shows large silica grains (brown) and porosity (black) and Cu (Red). The area analysed is 5x3 mm with a pixel density of 10,000 by 6,000.
Converting the X-ray maps into Phase maps using AZtec AutoPhaseMap shows precisely the constituent elements of the mineral phases, and how the phases are distributed over the entire specimen (Figure 2A). The Phase maps highlighted that there were two specific regions which formed after the smelting process: copper-iron sulphide with enriched copper, and iron sulphide with little copper (Figure 2A).
Interestingly, the original fairly homogenous chalcopyrite ore mineral appears to have partly reduced and revealed that the copper metal 'prills' were only forming at the interfacial regions containing copper sulphide – essentially the 'feeder' reaction zone on the copper-rich mineral phase, and not in areas with iron sulphide.
The chemical reactions producing elemental copper would have been in the molten matte, and the intrusive entrapped silica grains show that it was certainly molten. The microstructure of the copper prills is indicative of the copper cooling slowly from the reaction products giving them time to form on the inner surfaces of gas porosity regions of the solidifying matte.
However, the copper prills are not in a dendritic or globular form as would be the case if they had solidified from molten metal (i.e. from above 1085°C), but appear to be formed by a process of nucleation and growth – probably in the solid state producing the linear surface features (Figure 2B inset). The temperature to account for the features that are found would probably have been in the region of 950–1000°C (Dutrizac 1976).
Fig. 2. AutoPhaseMap generated from the montaged 900 datacube maps. The phases of SiO2 (yellow), CuFeS (green), FeS (blue) and Cu (red) were automatically identified and separated. Analysis shows that throughout the entire sample the copper ‘prills’ preferentially formed away from the Iron rich regions. Figure 2B (inset), fractured sample of the reaction products showing copper prills growing from the internal surface of a gas pore, but with no ‘roots’ within the matrix. (SEM micrograph Nigel Meeks).
To investigate the purity of the copper metal regions it was possible to reconstruct spectra from any point within the montaged image. Although the original data was collected at short process time and high count rate (200,000 cps) suitable for fast X-ray mapping, the data quality was not affected (Figure 3 and Table 1). Any artefacts due to high count rate such as pulse pile-up were corrected, allowing accurate quantitative composition results to be obtained. These quantification analyses identified that the copper metal regions had an average composition of Cu 99% with Fe 1%, which agrees with previous data (Craddock et al 2007, Craddock and Meeks 1987). This confirmed that the copper metal was fully reduced and its relative purity supports the microstructural observations that the prills probably did not form from molten copper – as it would most likely have dissolved more impurities (Percy 1861). It is also apparent that the FeSCu phase is nearer to a bornite composition than chalcopyrite. The copper/sulphur ratio is 1:1 rather than 0.5:1 of the chalcopyrite ore, therefore, this phase is an intermediate reaction product that has been enriched in copper, and lost some iron and sulphur to the adjacent FeS phase. This is consistent with the position of the copper prills, which are only found attached to the copper enriched mineral phase.
Fig. 3. Magnified area of montaged image showing the copper prills (red), iron sulphide reaction product (blue), and chalcopyrite ore (green) with reconstructed spectra and atomic composition quantification (Full Quant results in Table 1 on the next page). Pulse pile-up has been automatically accounted for to give reliable results, despite the high count rate used for data acquisition.
Table 1. Accurate quantitative results reconstructed from the stored montage image showing depleted copper mineral FeS, the copper metal deposits and copper-rich ore. At 99% Cu with only 1% Fe, it suggests the copper underwent a full reduction process during smelting.
Using AZtec Large Area Mapping to analyse the chalcopyrite smelting products was an effective means to cope with the large physical size of the sample while maintaining the high resolution required to investigate large (mm) and small (µm) scale distributions. By collecting the data in one automated run and performing the in-depth analysis off-line, valuable microscope time was preserved. AZtec AutoPhaseMap results identified trends in the formation of the copper which only formed in regions rich in CuFeS rather than FeS; this was evidence to indicate the temperature used during the smelting process (Dutrizac 1976). Reanalysed spectra from multiple regions revealed that the copper metal had a very high purity (~99%) which meant it underwent a full reduction procedure during smelting and was not co-deposited with any other metals. These results highlight that AZtec Large Area Mapping can be a very effective research tool for the investigation of large complex samples.
The extended application of the advanced instrumentation techniques used in this study was only possible by the provision of experimental samples from the dedicated team from the Early Mines Research Group and interpretation of the metallurgy by Paul Craddock.
Nigel Meeks at the British Museum is gratefully acknowledged for his significant contribution to this paper. Images on pages 1 and 3 courtesy of British Museum.
© Oxford Instruments 2026