Metallographic Investigation of the Bronze Sword from Vértesszőlős

The sample taken from the blade fragment of the bronze sword has a dendritic structure (Fig. 3). The sword was made by casting. The secondary dendrite arm spacing is ~ 25 µm. The secondary arm spacing and the whole microstructure is uniform in the examined section of the sample (Fig. 3b). Etching reveals a significant micro segregation in the brightfield images (Fig. 3c). The traces of micro-segregation and dendrite arm spacing indicate a relatively fast cooling. Comparing this with the geometry and dimensions of the tip of the sword, it is an excepted result.

High magnification optical microscopy images reveal a significant number of small-sized inclusions. The diameter range of these inclusion is from 1 to 2 µm to 15 µm. At least three types of inclusions can be distinguished based on their colour and morphology in the optical microscopic images: a) larger (5–10 µm) dark grey, mostly compact inclusions are copper sulphide inclusions (1 in Fig. 3d); b) spherical light grey inclusions in the small size range are lead droplets (2 in Fig. 3d); c) larger, more polyhedral light grey inclusions are intermetallic compounds with a high tin content (3 in Fig. 3d).

SEM–EDS study was performed on the same sample that was examined with an optical microscope to investigate these phases. The preparation of the sample was not supplemented by an extra additional step. The average composition of the blade fragment was estimated based on the EDS spectrum measured in different area (Fig. 4a, b). The results are summarized in Table 1. The average composition was measured in typical fields on the sample. The results show that the blade fragment is bronze with medium tin content (~ 5 wt.%). In addition, a relatively high nickel content can be measured. In addition to nickel, the concentration of arsenic, silver and antimony suggests a nickel-containing version of the tetrahedrite group as a source of ore. Among the alloys of artifacts found in the Carpathian Basin, this is a characteristic type. Archaeological analyzes link the source of this type of alloy, specifically the copper base of it to the Alps [4, 11]. The lead content is extremely inhomogeneous, which is not surprising to see in the copper-lead alloys. High lead content was determined (~ 4.5 wt.%). The addition of this amount of lead helps the mold filling during casting as there is liquid phase in the alloy at lower temperatures during the solidification. Such a high lead content could be beneficial for proper manufacturing, conversely it is modified the mechanical properties of course. The literature investigated the effect of lead content on an as cast copper alloy [40, 43] containing 6 wt.% tin. The hardness of the alloy was measured at ~ 70HV1. 2 wt.% added lead increased hardness by 19% and 10 wt.% lead reduced it by 15%. The measured 4.5w% represents nearly the same hardness as the hardness of the alloy without any added lead. However, it does not reduce the mechanical properties to such an extent that the sword is not suitable for its weapon function.

Analysis points 4 and 6 show extremely high lead content (Fig. 4c and d). Copper does not dissolve lead in their alloy either in a melt or solid state. For this reason, lead is found in the microstructure in an almost pure state as solidified droplets. The sulfur content of point 5 is extremely high (Fig. 4d), which indicates that the inclusion is copper sulfide. The content of tin, antimony and lead in the inclusion is probably a consequence of the excitation of the metal matrix. An extremely high content of tin is measured in the composition of point 3, close to the composition of the β compound phase of copper-tin alloys. This suggests that the intermetallic phase formed because of micro segregation. These SEM–EDS studies have revealed these 3 types of inclusions and intermetallic phases.

Two samples were taken from the broken grip plate of the bronze sword. One is from the edge of the grip. The sample was prepared for examination using the same method as for the blade fragment. Optical microscopic images are shown in Fig. 5. A duplex, recrystallized grain structure can be seen (Fig. 5a and b) where the difference between the diameters of large and small grains is large. In addition to large (~ 150 µm) grains, small (~ 25 µm) grains are also visible in the microstructure of the sample (Fig. 5b). The nature of the particles and the thermal twins in the grains show that these formed during a recrystallization process due to annealing after a metal forming process (Fig. 5c). The duplex grain structure suggests a slight plastic deformation (less than 50% in reduction) of the examined volume at elevated temperature. The nature of the sword suggests casting instead of forging. This small extent of plastic deformation could be made either when the sword was taken out of the mould or during the forming of the final shape of the grip. If plastic deformation was intended, it was carried out while the sword was still warm, at a temperature where recrystallization takes place even after a slight deformation (> = 400 °C) [42, 43].

The prepared section is heavily affected by corrosion. Thinner-thicker corrosion products are situated on the grain boundaries (denoted by 1–2 on Fig. 5d). However, the mostly spherical, dark grey sulphide inclusions in the optical microscopic images (Figs. 5e and f) are clearly separated from the corrosion products. To verify this and determine the composition, SEM–EDS tests were carried out. The images are shown in Fig. 6, and the results of the composition measurement are shown in Table 2.

The composition of the sword is significantly different from the blade fragment. The tin content is high (~ 8 wt.%) but lead content is not detectable. Although the base copper alloy seems similar in its nature, antimony in the sword material could not be detected by SEM–EDS analysis. In addition, the content of nickel, arsenic and silver is also lower than in case of blade fragment. Compared to the difference in tin content, the quality of the basic copper alloy is close to each other except for the antimony content. Previous classification of other artifacts by raw material revealed that the existence of the antimony is important. The basic composition of the inclusions was checked using local analyses. The copper and tin content within the grains was also measured as the basic values. Indeed, the dark grey inclusions of optical microscopy are copper sulphide inclusions. The element map of the dendritic inclusion shown in the optical microscopic image of Fig. 5f is shown in Fig. 7. These types of inclusions were enclosed in the microstructure of the metal during the smelting of the copper. The tiny light phase incorporated into the inclusion is lead. The lead inclusions generally occur mostly in association with sulphide inclusions [44, 45]. The number and the size of lead phases explains why the average composition analysis does not show this low lead content.

Figure 6 shows SEM images taken from the sample. The noted area and points show the places of chemical analysis. The inside of the grain is the inclusion and the phase on the grain boundary are also analyzed at several points. The phase at the grain boundaries is a corrosion product. This is suggested by local analysis showing a significant increase in oxygen content compared to measurement inside the grains. With a small sulphur content, phosphorus and iron also appear. Arsenic and significant tin enrichment can be measured in the corrosion product. The enrichment of tin in the corrosion process is a known phenomenon that should be expected. However, here its extent is also significant in comparison with the high tin content of the alloy. Although the local analysis did not detect the silicon and aluminium content, the chemical element maps of Fig. 8 show higher silicon and aluminium concentration. So, the aluminium content can also be linked to the corrosion product.

It can be seen on the chemical element maps that the silicon content is not uniform in the corrosion product. There are areas where a higher silicon amount is detected. This explains why it was not detected in local analysis. It can be clearly seen that the distribution of phosphorus content is correlated with the silicon distribution, so it comes from the same corrosion process as silicon. The sulphur and tin content seem uniformly higher than in the metal matrix. To investigate this, a tin and copper distribution function was measured nearly perpendicular to the grain boundary to the corrosion product, which is shown in Fig. 9. The transition from the metal matrix is diffuse, however, the tin content rises sharply in the corrosion product. What is interesting from the diagram is that the tin content also varies significantly in the bronze matrix in addition to the corrosion product.

The corrosion product contains a significant amount of tin compounds in addition to copper compounds. It also contains small amounts of iron, phosphorus, sulphur, and silicon certainly in the form of compounds. It is necessary to take this into account, the phosphorus, silicon, and aluminium content of the measured average composition on area of interest by SEM is due to the corrosion product.

The previous sample differs from the blade fragment both in its alloy and in its microstructure. For this reason, the grip plate was sampled from another location. Figure 10. shows the microstructure of the sample. This sample also has a dendritic microstructure, which suggests casting as a manufacturing technique (Fig. 10a and b). The section shows that the entire cross-section is affected by corrosion (Fig. 10b), which should be considered when the results are evaluated. The dendrite arm spacing is ~ 25 µm, as in the case of the blade fragment (Fig. 10d). The dendrites are not so strongly articulated, but the grain structure is more visible in polarized light than in the case of the blade fragment (Fig. 10c).

The average grain size is ~ 120 µm. Considering these data, the rate of crystallization, as well as the rate of cooling was slower in this part. This is explained by the geometry and the greater metal mass, compared to the blade fragment. Inclusions can be clearly distinguished from the corrosion product in the high magnification micrographs (Fig. 10e and f). At least 2 types of inclusions can be identified on the photos. Dark gray inclusions are copper sulphide inclusions. It was found with two types of morphology, dendritic and spherical. Next to these, lead inclusions can be seen in a light grey colour. This sample contains far fewer lead inclusions than the blade fragment. The larger phases visible between the dendrite arms and following the contour of the dendrite arms are also a corrosion product in this sample. SEM–EDS analysis was performed in areas to determine the average composition (Fig. 11). The results of the EDS analysis are shown in Table 3.

The alloy is almost completely identical to the previous sample. Its tin content is high (~ 7.5 wt.%). In addition, the lead distribution is inhomogeneous and lead inclusions are found in little amount in the section. The nickel and arsenic content must be highlighted, which can also be detected in the other sample. No silver content was detected in the studied areas. Silver was measured in very small quantities in Sample 1. The different state of the corrosion may be the cause of this deviation. Antimony was not detected in this section either. The appearance of silicon, aluminium, phosphorus, and iron is due to the existence of corrosion product, hence the corrosion processes. Sulphur is also found in sulphide inclusions and in the corrosion product according to previous test results.

This sample also suggests that the blade fragment is not made from the same alloy as the sword. The tin content differs significantly from the blade fragment. Although the blade fragment has a similar base copper alloy, it has a significantly lower tin content and a higher lead content than the alloy of the sword. Additionally, no antimony could be detected in the sword material. In addition, this sample prove the premise that the sword itself was made by casting. However, based on Sample 1, either the last manufacturing steps were carried out while the sword was still warm, or where the previous sample was taken, the material of the sword was slightly deformed when it was taken out of the mould.

The grip plate of the sword and the place of the cross-guard are designed in such a way that the grip and the guard element can be fixed there with rivets. The holes in the grip plate and the bronze rivets show this. Nothing of the grip part remains to reconstruct it. But one of the rivets could be removed for inspection. The rivet was examined in its entire cross-section. Optical microscopy images are shown in Fig. 12. The section is significantly affected by the effects of corrosion, although the grain structure is clearly visible. The micrographs show recrystallized structure with a uniform medium grain size (~ 50 µm) (Figs. 12a and b). Dark grey copper sulphide inclusions are mainly at the grain boundaries. In addition to inclusions, a corrosion product is also found in this sample at the grain boundaries (Fig. 12c and d).

SEM–EDS test was performed to analyse the composition (Fig. 13). In terms of the composition of the rivet, it is bronze with a medium tin content (4.8 wt.%) (Table 4). In addition to the tin content, the rivet also contains a significant amount of arsenic. A minimum amount of silver can be detected, but no antimony content is measured. Nickel is contained in a larger amount by the corrosion product on the grain boundary however the alloy itself does not contain it. This suggests that the alloy is not identical to the material of the sword, which is caused not only by the difference in the tin content. Due to the smaller tin content, the estimated strength of the material of the rivet is lower, it is more malleable [42, 43]. It corresponds to the application as a rivet. When analysing the corrosion product, tin is significantly enriched in it. In addition, the titanium, manganese, iron and in this case nickel content can also be associated with a corrosion product. The average composition analysis detected a minimum amount of lead content. Lead is found in the corrosion product as an independent phase.

The bronze dagger is corroded to such an extent that the sample contains only a small number of metallic parts (Fig. 14a). The metallic area visible in the section is unsuitable for determining the manufacturing technique (Fig. 14b). The local composition of the metallic part can be investigated by SEM–EDS analysis. During the evaluation the effect of corrosion must be taken into account, for example the enrichment of elements. In addition, the corrosion product can also be excited during the test which could modify the results (Fig. 14c). With these the tin content of metallic grains is high at 11.8 wt.%. In addition, a significant (2.2 wt.%) arsenic content can be measured. Silver and antimony were not detectable. Nickel in relatively high quantities: 2.6 wt.% can be measured, although in the case of the rivet it is not clear whether it is contained in the metal or in the corrosion product. Sulphur was detected in the amount of 1.5 wt.%. Between the metallic particles and in some places in the particle itself, copper sulphide inclusions can be seen (Fig. 14d). The measured sulphur content can be attributed to these inclusions. No lead could be detected. In addition to copper sulphide inclusions, small sized lead inclusions are also found in this sample (Fig. 14d). The size and quantity of lead particles are so small that these really do not manifest themselves in the test results of the metallic grains. The remaining detected elements (Al, Si, P, Cr, Fe) are a consequence of the excitation of the corrosion product.

In terms of tin content, even if the tin is enriched in metallic grains, the dagger is made of an alloy with a high tin content. In terms of arsenic and nickel, it can also be considered to be a similar alloy to sword. Unfortunately, due to strong corrosion, a more detailed comparison cannot be made between the two objects. However, it appears that the sword and dagger are made of a higher hardness, higher strength high-tin alloy, which is supposed based on their use. For each object have an as cast or annealed state. Tin content is one of the determinants in terms of strength based on the literature data. In the case of blade fragment and rivet can be estimated a tensile strength of ~ 350 N/mm2 in terms of tin content of 5 wt.%. In the case of sword and dagger it is more than ~ 410 N/mm2. Lead content does not significantly alter the strength of the blade fragment. The strength is also modified by other constituents of course. For example, the amount of arsenic in the case of dagger is further increases the strength of the alloy.