Discussion
Several maritime archaeological items, made of bronze, wrought iron and grey cast iron, were studied by the FMM method. The primary objective was to compare the results obtained by this innovative NDT technique with conventional destructive metallography. As demonstrated in the results of this study, and shown below, these metallographic methods yield similar results.
The powder chamber (Fig.
1) was made of cast bronze. Although the bronze was well-preserved, different corrosion products were observed on its external surface prior to cleaning by abrasively brushing, as expected from bronze archaeological objects, after a long burial period in a marine environment. This corrosion phenomenon is occasionally referred to as ‘bronze disease’ [
17,
31]. LM and SEM examinations of the object, by both the FMM technique and by conventional metallography, revealed dendritic microstructure and interdendritic shrinkage porosity (Figs.
4 and
5a), typical of cast copper alloy objects [
8,
10,
16,
32]. Both metallographic techniques revealed a dual-phase microstructure of an orange major phase and a silver minor interdendritic phase, as well as interdendritic porosity (Figs.
4 and
5a).
EDS spectral analysis and elemental mapping of the chamber’s conventional metallographic sample revealed presence of Cu, Sn, Pb, Zn, As, Sb, Ag, and Ni (Figs.
5b and
6). The reliability of the minor element composition detected by SEM-EDS analysis was confirmed by comparison with the XRF and OES analysis results [
16, p. 5, table 2 and p. 6, table 3] and good agreement was achieved between the results. According to Iddan et al. [
16], the elemental composition of the abrasively brushed surface of the powder chamber based on XRF results was 71.9–85.9 wt.% Cu, 11.2–20.1 wt.% Sn, and 1.1–4.7 wt.% Pb, as well as small amounts of Si, S, Sb, As, Ag, Ni, Zn, and Fe. The OES analysis of the abrasively brushed surface revealed that the item was made of homogeneous bronze with a composition of 84.6–86.2 wt.% Cu, 10.5–11.8 wt.% Sn, 1.9–2.2 wt.% Pb, and up to 0.6 wt.% of Sb, As, Ag, Ni, and Zn. The SEM–EDS analysis of the conventional metallography specimen produced similar results, with a composition of 85.2–87.0 wt.% Cu, 9.5–10.8 wt.% Sn, 1.0–2.4 wt.% Pb, and up to 0.8 wt.% of Sb, As, Ag, Ni, and Zn [
16].
Three-phase texture was observed in the cast bronze powder chamber microstructure (Fig.
6): (1) an α-copper dendritic matrix as a main phase composed of a Cu-Sn solid solution (Fig.
6b), (2) an interdendritic Sn-rich α + δ phase scattered at the boundaries of the copper-rich phase (Fig.
6c), and (3) and a minor Pb-rich phase (Fig.
6e) [
16,
33, p. 10, Fig. 5,
34, pp. 11–12]. The similarity between the elemental distribution of the tin (Fig.
6c) and antimony (Fig.
6d) detected in the SEM–EDS elemental mapping analysis of the powder chamber may result from an overlap uncertainty between the Sn and Sb peaks in the X-ray spectra (Fig.
5b). However, since both Sn and Sb were also detected in the OES analysis, their mutual appearance may originate in their ore mineral composition. The bright areas, observed by SEM–EDS elemental mapping, are mainly composed of lead (Fig.
6e). However, the parallel elemental distribution between the lead and the arsenic images (Fig.
6e and f, respectively) may result from an overlap uncertainty between the Pb and As peaks [
35] (Fig. 5b). However, since both elements were also spotted in the OES analysis, their mutual appearance may be related to their ore composition.
According to Iddan et al. [
16], based on the examination of the conventional metallography samples, the concentration of copper was higher in the core of the powder chamber than at its external abrasively brushed surface; whereas, the concentrations of tin and lead are higher at the surface. This phenomenon probably results from the solidification process during casting, as well as according to various diffusion and corrosion mechanisms during the long burial period at sea [
16]. Lead was probably added to the molten copper alloy based on economic considerations, and/or to improve castability [
8,
16,
36]. Sulphur, silicon, and iron were detected only at the surface of the powder chamber, and are therefore related to the long-term exposure to the underwater environment [
16,
36].
The anchor chain links were made from wrought iron, although the studs themselves were of grey cast iron (Fig.
2c). The stud-link anchor chain was well-preserved (Fig.
2), although it was found covered with concretion. Beneath this concretion, different corrosion products were observed on the external surface before it was removed by sandblasting [
12], as expected of a ferrous archaeological object after a long burial period in the sea, LM and SEM examinations of typical links removed from the chain by both the FMM technique and by conventional metallography revealed preferentially oriented slag inclusions (dark areas, longitudinal plan) surrounded by iron ferrite grains (Fig.
7), typical of wrought iron products [
7,
11,
37‐
39]. The absence of cementite morphology in the link’s microstructure indicates that it was made of wrought iron produced by the indirect smelting process of refining pig iron (the ‘puddling’ method) [
7,
11,
37]. SEM–EDS elemental mapping of a typical link revealed presence of Fe, O, Si, and P (Fig.
8), typical of wrought iron products [
7,
11]. According to Iddan et al. [
16], the elemental composition of the links based on XRF analysis of the ground surface revealed that it was made of iron, with presence of only 0.3 ± 0.2 wt.% Si and 0.2 ± 0.2 wt.% P. Similar XRF analysis results were also observed in the rod, with a composition of 99.6 wt.% Fe and 0.4 wt.% Si. Since silicon does not reduce in either the direct or indirect smelting techniques, its presence in the link represents the distribution of the slag inclusion in the wrought iron matrix [
7,
11,
12,
37].
According to the OES analysis results, which is considered as a reliable and accurate method for the determination of carbon concentration in iron alloys [
40], the carbon content detected in the ground surface of a typical link was 0.0648 ± 0.0003 wt.% C, which is a typical value of ferritic iron [
12]. The SEM–EDS analysis of the link’s ground surface revealed similar results, with a composition of 95.6–99.5 wt.% Fe, up to 2.9 wt.% O, 0.1–0.5 wt.% Si, and up to 0.4 wt.% of Mn, Pb, Ca, and Mg [
12].
LM and SEM examinations of a typical stud (Fig.
2c) from the stud-link anchor chain, by both the FMM technique and by conventional metallography, revealed a dendritic pattern with black graphite flakes and graphite rosettes (Fig.
9), typical of type B according to ASTM A247-67 [
41] grey cast iron products. SEM–EDS elemental mapping of a grey cast iron stud by conventional metallography revealed presence of Fe, P, C, S, and Mn (Fig.
10).
Based on Iddan et al. [
12] XRF analysis of a typical stud demonstrated that this item was mainly composed of iron, with presence of 1.9 ± 0.4 wt.% Si, and 0.2 ± 0.1 wt.% Mn, as well as up to 0.4 wt.% of P, S, and Ti. The carbon content in the stud material, according to Iddan et al. [
12] OES analysis, was 1.9333 ± 0.0455 wt.% C. The manganese was perhaps added deliberately to reduce gas porosity [
5,
42] and to form manganese sulphide with the sulphur impurities in the metal. The presence of silicon, which serves as a graphite stabilizer material, is a typical element of grey cast iron.
LM and SEM examinations of the winch drum (Fig.
3a) by both the FMM technique and by conventional metallography revealed a dendritic microstructure with black graphite flakes and graphite rosettes surrounded by ferrite grains and iron phosphide eutectic (Fig.
11), typical of type B [
40] grey cast iron products. A qualitative elemental mapping of the winch drum is presented in Fig.
11f, showing the distribution of the graphite flakes (dark blue flakes), surrounded by ferrite phase (bright green matrix).
SEM–EDS elemental mapping of the drum revealed presence of Fe, C, S, P, Ti and Mn (Fig.
11f). According to Iddan et al. [
17], the elemental composition of the drum surface based on XRF analysis was 95.1–96.3 wt.% Fe, 2.6–3.8 wt.% Si, as well as less than 1.0 wt.% of P, Mn, and Ti. The SEM-EDS analysis of the drum’s conventional metallographic samples revealed a composition of 95.0–96.0 wt.% Fe, 2.6–2.8 wt.% Si, 0.9–1.1 wt.% Mn, 0.2–1.0 wt.% P, and up to 0.2 wt.% of Ti, S, and Cu; whereas, SEM-EDS analysis of black flakes revealed they were composed of 100 wt.% C [
17]. Iddan et al. [
17] found that the iron phosphide (Fe
3P) eutectic morphology was composed of 83.5–83.7 wt.% Fe 13.7–13.8 wt.% P, and 2.0–2.3 wt.% Mn, as well as presence of 2.0–2.3 wt.% Mn, and 0.1 wt.% of Ti and Cr. Manganese sulphide (MnS) precipitations were also observed on the drum surface by SEM-EDS analysis, with a composition of 59.2 wt.% Mn and 36.9 wt.% S [
17].
LM and SEM examinations of the winch shaft (Fig.
3a) by both the FMM technique and the conventional metallography revealed preferentially oriented slag inclusions (dark areas) surrounded by iron ferrite grains (Fig.
12), typical of wrought iron products produced by refining pig iron by the indirect smelting process [
7,
11,
37]. SEM–EDS qualitative elemental mapping of the wrought iron shaft revealed that the preferentially oriented slag inclusions were composed mainly of silicon oxide (SiO
2, pink-red areas, Fig.
11f) surrounded by ferrite phase (Fig.
12d, light green matrix).
According to Iddan et al. [
17], the XRF analysis of the shaft revealed that the wrought iron was composed of iron 99.9 wt.% Fe and 0.1 wt.% Ti; whereas, the shaft’s composition according to the SEM-EDS analysis was 99.3–99.8 wt.% Fe, 0.2–0.4 wt.% P, up to 0.2 wt.% Si, and up to 0.1 wt.% S [
17].
VT inspection of the heart-shaped shackle after sandblasting exposed a fibrous texture of parallel fine lines (Fig.
3b), typical of wrought iron rods [
12]. LM examination by both the FMM technique and the conventional metallography revealed preferentially oriented slag inclusions, as well as small circular inclusions surrounded by iron ferrite grains (Fig.
13), typical of wrought iron products. According to Iddan et al., the XRF analysis of the heart-shaped shackle revealed that it was made of a wrought iron, with a composition of 98.6–99.8 wt.% Fe, and up to 0.8 w% of Si, P, S, Mn, and Ti [
17]. The SEM-EDS analysis of the link’s wrought iron (conventional metallographic samples) revealed a composition of 96.3–99.2 wt.% Fe, 0.2–2.3 wt.% O, 0.5–0.6 wt.% Si, and up to 0.6 wt.% P, as well as up to 0.2 of S and Mn. The SEM-EDS analysis of the shackle revealed it was composed of 97.0 wt.% Fe, 1.8 wt.% O, 0.5 wt.% Si, and less than 0.5 wt.% of P, S, and Mn [
17].
LM examination of the deadeye strap with futtock plate (Fig.
3c) by both the FMM technique and conventional metallography revealed deformed (elongated) iron ferrite grains, as well as preferentially oriented slag inclusions and small circular inclusions (Fig.
14), typical of hot-forged wrought iron products. According to Iddan et al. [
17], the XRF analysis of the deadeye strap with a futtock plate showed that it was made of wrought iron, with a composition of 98.9–99.5 wt.% Fe and up to 0.7 wt % of Si, P, S, and Ti. The SEM-EDS analysis of this item (conventional metallographic samples) revealed a composition of 96.2–99.5 wt.% Fe, up to 2.8 wt.% O, and less than 0.5 wt.% Si, P, S, Mn, and Cu [
17].
LM and SEM examinations of the stud-link chain controller (Fig.
3d) by both the FMM technique and conventional metallography revealed microstructure of graphite flakes surrounded by ferrite matrix, perlite, and white iron phosphide eutectic (Fig.
15), typical of type B grey cast iron products [
41,
43]. The XRF analysis of the chain controller revealed a composition of 96.0–97.6 wt.% Fe, 1.2–2.3 wt.% Si, and 0.4–1.5 wt.% P, as well as less than 0.5 wt % of Mn and Ti [
17]. The SEM-EDS analysis of this item (conventional metallography samples) after omitting carbon peaks revealed a composition of 95.7–96.9 wt.% Fe, 1.5–2.2 wt.% Si, and less than 1.0 wt.% of O, P, S, Mn, Ti, and V. The composition of a typical Fe
3P eutectic morphology according to SEM-EDS analysis was 88.9 wt.% Fe, 0.5 wt.% Si, 10.2 wt.% P, and 0.2 wt.% of Cr and V [
17].
This study was aimed at assessing the feasibility of the FMM method as an advanced non-destructive technique for examining the microstructure of both ferrous and non-ferrous archaeological artefacts retrieved from marine environments. The advantages of the FMM method and conventional metallography are summarised in Table
2. The examination of the ferrous and non-ferrous archaeological objects by using the novel FMM technique produced metallographic images equivalent in their quality to the images achieved by conventional destructive metallography. Thus, results obtained in both metallographic methods agree well for all examined bronze, cast iron and wrought iron objects (Figs.
4,
5a,
7–
9,
11,
12a‒c,
13–
15).
Table 2
FMM and conventional destructive metallography methods and their main advantages and disadvantages
FMM | Minimally destructive | Expensive |
Enables examination of crystalline grains, different solid-state phases and microstructural defects (with multi-focal LM observation) | Surface preparation is essential, skill needed to properly prepare surface |
Includes advanced microscopy with auto-focus and multi-focal system, with large depth of field with large working distance | Large objects cannot be inserted into a conventional scanning electron microscope |
Real-time in situ analysis | Skills required to prepare surface of small objects |
Possible to observe the surface of a convex object. | Near surface area is observed only after grinding, polishing, and etching |
Conventional metallography | Inexpensive | Destructive |
The core of the material can be observed after grinding, polishing, and etching | Possible to observe only flat surfaces |
Enables examination of crystalline grains, different solid-state phases and microstructural defects (with LM and SEM observation) | Surface preparation is essential |
Skill needed to properly prepare specimens |