There were 271 valid samples of non-ferrous metal specimens analysed in LTS, and the number of gilt-gold copper ornaments is about 72%. The statistical result has shown that the type of them are mainly coffin rings, bubble nails, buckles and decoration sheets (see Table 1). Considering the surface corrosion and decuprification proceeding affection, we set the weight% at 5% of Pb and 1% of Sn, As and other alloy elements as a line to distinguish the copper and bronze artefacts with the p-XRF data and 1% of all the alloy elements with the SEM-EDS data. The result was also shown in Table 1. The data distribution shows a similar pattern of non-ferrous specimens unearthed from the Shiguo Tomb and Longhai Tomb areas. The same kind of copper and bronze artefacts barely coexisted in the same tomb, and the bronze ones were absent in LH but abundant in SG. The p-XRF data show that the chemical composition of bronze is mostly copper alloy containing arsenic, tin, and lead. Most of the specimens contain trace amounts of bismuth and antimony. And the copper ornaments were relatively pure, and only lead and silver could be widely detected in the bulk material. Based on the p-XRF detection data of LTS specimens, we conducted a statistical pattern on the bronze and copper relics. The data of zinc and iron were also involved as an indicator of the corrosion condition of the specimens (Fig. 2 upper left). Since these two elements were not detected in the metal bulk of specimens by SEM-EDS, this pattern was set as a baseline for comparing with other data clusters to estimate the relationship base on the element composition and shown in the background of each coordinate system translucently in Fig. 2.
Since LTS is considered a royal cemetery, we selected some gilt artefacts from the urban capital region of the kingdom as references to the ones that belong to the elite class. Such as Bohai-style gilt bronze Buddha statues and gilt copper ornaments unearthed from the SJ Palace and Royal Garden for the p-XRF testFootnote 1. The ‘numerical distribution’ of the composition data of the two types of relics in each element component has shown a somewhat similar pattern to the artefacts of LTS. Only the tin portion of the bronze Buddha was concentrated at a relatively high level (Fig. 2 upper right). Involving the commoners’ tombs from a border perspective on all the Bohai sites, we selected the bubble nail and buckles as targets, the two most widespread kinds of unearthed ornamentsFootnote 2. The alloy composition distribution of the bronze bubble nail agreed with the bronze data of LTS well, and there is no batch of copper bubble nails found at these commoners’ sites in the Bohai Kingdom (Fig. 2 lower left). Buckles were also made from bronze, and the tin and lead content were higher than the bronze baseline of the LTS artefact (Fig. 2 lower right). All the data with a zinc or iron ratio higher than one was abandoned, considering the rusty surface would affect the p-XRF data too much in this research.
The metallographic phase observation and SEM-EDS investigation of interventional sampling of LTS show that the process of making copper and bronze artefact was quite different. A hot forging process mainly produces the copper relics unearthed at M10 and M13 with pure copper. The metallographs showed the residual twin crystalline phase and elongated inclusion in the bulk material, which referred to the large deformation and heating treatment during the production. Similar techniques are found in the copper bubble nails unearthed at M1C of SG and other copper decoration sheets in SJ (upper Fig. 3). All the bronze artefacts of LTS were produced by casting. The (α + δ) eutectoid structure formed dendritic segregation shown in the bronze specimens’ metallographic figures. And the segregated structure has no deformation with relatively rare copper sulfide inclusions (lower Fig. 3). Among the 60 samples for interventional testing. However, the raw material of the artefacts was different, and their gold gilt layer shared a resemblance in both the structure and content. The thickness of the gilt layer of the small copper and bronze ornaments was around 1–3 micrometres (Fig. 3), except for two crown decorations, which were covered by over 6-micrometre-thick golden layers. The content of the gold layer was a mixture of gold, 1–5% silver, and 10–23% mercury, which are the typical constitution of the gold amalgam for fire gilding. The SEM-EDS detected the copper specimens containing slight lead and a trace amount of silver. The bronze specimens mainly bore insoluble lead within the tin/arsenic eutectoid segregation with a bit of antimony bismuth conformed to the p-XRF data (Additional file 1).
The mechanical characteristics of copper and bronze are somewhat different and directly affect the processing of the artefact production, even for the same type of ornament like bubble nails. The copper nail was forged into a hat-like formation and fixed on the wooden coffin surface with two tiny copper nails on the rim (upper Fig. 3). The bronze bubble nails were just like the large pushpin. They were nailed to the coffin (lower Fig. 3). Both kinds of bubbles were plain surfaces with 2–4 centimetres diameters and used to fix the fabric onto the surface of coffins for decoration. The different techniques and materials combinations of the same product could cause by distinct technology traditions. The ultra-depth microscope observation provided some clues by the details of the emblazonry. The themes, types, and lines of the patterns on the gilt copper coffin ring (M13:27) were carved on the surface, which was quite similar to the silver-back of the bronze mirror (M13:168) and the golden crown (M14:1) unearthed in M14 of the LH [8]. The fluency lines and roe-like pattern carved by hollow chisel constructed the basic elements of the emblazon background. Similar carving techniques could be found only on the copper artefact, such as the leaf-shaped gilt-bronze crown ornament (M1C: 12) unearthed in M1C of SG and other decoration sheets. On the contrary, the surface treatments of bronze artefacts were quite different. The gilt bronze coffin ring (M2A:75) unearthed in the M2A of SG with entangled floral patterns was combined with little carved dens by a needle-like chisel. The techniques of the bronze process may be related to the difference in the mechanical properties of the material (Fig. 4). The diameters of the two kinds of coffin ring heads were 11–12 centimetres, and their thickness was around 2–3 millimetres. Though the size and type were rather similar, the carving techniques of copper ones were much more complicated and sophisticated. And the production details of the artefact would need more processing relics for further study.
To seek the provenance of the raw materials of these artefacts, archaeological reconnaissance of LDG was conducted in 2014 and 2016 on LDG and GC sites, the essential mining and smelting sites discovered in the Bohai Kingdom. LDG had been excavated about 20 years ago, and the field investigation and sampling of the specimens were mainly based on the former archaeology records. The figure of LDG was made based on a 1970 s map of this zone because modern mining production has already changed the landscape. The records of GC can not be reached before the publication of the archaeology report, and the site has been backfilled after the excavation. Fortunately, we had been authorised to test some of the unearthed slag and ores as a reference.
According to the records of the excavation of LDG, there were more than twenty ancient copper mining ditches discovered in the mining zone of LDG (see the mining ditches in Fig. 5). The ditches were contacted by very narrow tunnels which could barely accommodate a single miner. Plenty of greenish ores could still be found near the ancient ditches now, which were all polymetallic copper ores (upper left of Fig. 7). Pyrite was the primary copper mineral of the skarn ores. Blending with the pyrite phases, molybdenite, bismuthinite, and stibnite could be found as the paragenetic minerals in the ores of LDG. More andradite and tetrahedrite minerals were discovered in the GC ores than in LDG; besides, the ores’ content from both sites was quite similar [5]. As the former excavation and latest reconnaissance records, the living relics like the Bohai Kingdom ceramic fragments were few in the ancient mining zone. Most of them were distributed along the valley where the dwelling and dock sites were excavated (green zone in Fig. 5). Unfortunately, it was impossible to estimate the scale of mining activities since part of the ancient mining and dwelling zones were buried by modern copper mill tailings. However, about 5–10 km to the south of the mining zone, there were still several piles of smelting slag well preserved in at least four sites.
The Gan Gouzi (means dry valley, abbreviated to GGZ), Xi Dadingzi, Jiaxin Gang, and Gan Fanguo (means rice pot, abbreviated to GFG) were believed as four major Bohai smelting sites at LDG. The sampling of smelting slag at GGZ and GFG, because these two sites were better well-preserved and located far from modern living zone. Tons of the smelting slag piles at GGZ were dumped down from the top of the hills and accumulated over one meter thick on the hillside (right of Fig. 5). The slag of GFG was also accumulated around the site and at the top of the hill. We could still find the excavated wreckage of the smelting furnace at GFG, which was built up with stones nearby and clay. This copper smelting furnace was similar to the typical iron blast furnace of the Bohai Kingdom [14], with a smaller diameter of around 5 m. Although there was no trace of smelting found at the site, the indentation on the slag around the furnace had been stamped from wood charcoal patterns. These patterns could be widely found on most slag piles identified as pines log charcoal, providing relatively high calorific value. And this kind of pines is still the primary local ligneous plant nowadays. According to the former study on smelting slag [9], with adequate high-quality fuel and proper furnace design, most slag with similar content could reach a reasonable softening, and melting temperature around 1200–1400 °C [15], Some abnormal composition of the slag might cause by the erosion of adhering furnace parts (Fig. 6).
The previous research on the little metal prill, including the slag matrix, suggested that the final production of the smelting was mainly high-grade matte and blister [9], which contains primarily arsenic, lead, and bismuth. And in a little of the matte and bronze prill, antimony, tin, and silver could also be found as trace compositions. All these elements could be found in the ore copper mineral collected at LDG and GC. The ore from GC was similar to the typical poly-metallic ones of LDG, but the slag matrix constitution was relatively dispersive. Since we do not have the excavation record of the GC slag, the smelting processing details could not be deduced accurately. What we can confirm from the relics of GC were the production and fuel. The vegetation of GC was quite the same as LDS, and a similar charcoal pattern could also be found on the slag (the left image of Fig. 5). The whole processing of the smelting activities, which was deduced from the evidence we obtained for now, of GC and LDG should be similar to each other. The distinction in the slag content could be relevant to the different minerals within the copper ores (Fig. 7). There were records about the content data of Bohai copper artefacts unearthed in Eastern Siberian in Russia, which also contains tin, arsenic, and antimony [16]. The bubble nails and buckles were also unearthed from the Bohai sites like Kraskino on the northern shore of Russia and the northeastern Korean Peninsula, which shared the same typological characteristic of bronze artefacts [17].