2. Methods
Neutron studies were carried out at the IR-8 reactor. Most of the measurements were carried out at the “DRAKON” facility [
5] ((NRC) “Kurchatov Institute”, Moscow, Russia), whereas some measurements were conducted on a neutron tomograph [
6]. At the DRAKON facility, the neutron beam was formed by a double monochromator with pyrolytic graphite crystals in 002 reflection; the neutron wavelength was 0.175–0.2 nm. Neutron projections were recorded by a position-sensitive detector consisting of a scintillation screen based on a 200 μm thick mixture of ZnS(Ag) and
6LiF, a mirror, an objective lens, and a charge-coupled device (CCD) matrix with dimensions of 2048 × 2048. The beam size was 7 × 7 cm. The exposure time was 60–150 s. The spatial resolution of the obtained images was about 200 µm with a pixel size of 65 × 65 µm. When studying objects exceeding the size of the beam, the object was scanned with subsequent stitching of full projections or stitching of tomographic arrays. At the facility [
6], the neutron beam was formed by a monochromator with a copper crystal in 111 reflection, the neutron wavelength was 0.152 nm, and a Soller slit was present with a divergence of 20°. Neutron projections were recorded using the same type of position-sensitive detector with a 100 μm thick screen. The beam size was 10 × 10 cm, and the exposure time was 480 s. The spatial resolution was about 160 µm, with a pixel size of 55 × 55 µm.
Synchrotron studies were carried out at the tomographic station on the K6.3 beamline of the Kurchatov synchrotron radiation source “KSRS-Kurchatov”. The spectrum of the synchrotron radiation (SR) from a bending magnet was formed by a copper filter 1.5 mm thick, which gave the SR beam a spectrum maximum of about 56 keV and a spread of about 20 keV. The projections of the object were recorded using a position-sensitive detector consisting of a CsI(Tl) scintillation screen, a mirror, an objective lens, and a 2048 × 2048 CCD matrix. The beam size was 50(H) × 3(V) mm, and the spatial resolution was 75 μm with a pixel size of 25 × 25 μm. The exposure time was 0.5–2 s. Since the SR beam height is small, the projections were compiled from a series of images taken with a shift in height at a step of 1 mm.
ImageJ software package [
7] (National Institute of Health, Bethesda, MD, USA) was used for primary image processing, projection assembly, and analysis, whereas Octopus Reconstruction [
8] (Inside Matters, Ghent, Belgium) was used for tomographic reconstruction and rendering. Thermal neutron diffraction phase analysis and X-ray fluorescence elemental analysis were used as additional methods.
3. Results and Discussion
Corrosion is the main process that threatens the safety of metal archeological objects. As examples of the detection of corrosion, we refer the reader to tomographic studies of the reliquary cross pendants of the 11th–13th centuries, made by casting from tin–lead bronze [
9,
10]. These works were mainly aimed at studying the contents of the reliquaries, but the evaluation of the state of the metal was also part of the research task. Various cases were observed; as a rule, it was more or less homogeneous surface corrosion or inhomogeneous corrosion where volumetric areas of damage of the metal took place. An example of the first case is the cross from the rural site of Fedosyino-1 of Suzdal Land (
Figure 1). Corrosion products are clearly visible in the tomographic slices as regions of high attenuation due to their content of hydrogen, which has a high thermal neutron attenuation cross-section.
Inhomogeneous corrosion was observed in the cross from the rural site of Soroguzhino-2, Suzdal Land (
Figure 2). It can be seen that the regions of corrosion are 1–5 mm in size and are more or less evenly distributed over the product (
Figure 2b). A comparison of neutron and SR images shows that these regions have the opposite contrast; for SR, they are more transparent than metal, due to the fact that the density of the corrosion products is less than the density of the metal (
Figure 2c,d). In the stacks of neutron projections of this cross, the diffraction effect of “flickering” is observed. The effect arises due to the strong attenuation of the neutron beam passing through small parts of the volume of the material in a narrow angular range of rotation of the object. This attenuation is due to the fact that, at a certain angular position, a single crystal grain of the metal enters the reflecting position, and a significant part of the beam intensity is removed from the image formation. This diffraction contrast was used, in a more complicated way, for the reconstruction of grain morphology in the polycrysrtalline materials [
11]. This effect can also be detected in the following way: considering a sequence of projections in a certain range of angles (up to 20°), and using the ImageJ program function, one can construct a map of the minimum brightness values of these projections. Thus, the regions are distinguished for which the diffraction attenuation of brightness occurs in the given angular range (
Figure 3a). These areas are the images of crystallites. The diffractive nature of the contrast in these images is confirmed by the map of maximum values for the same sequence of projections; diffraction reflections are observed in an empty field (
Figure 3b). The reflections occur during the sample rotation angle of about 1.5° (full width at half maximum), as shown in
Figure 3c. Thus, in the bronze cross with large crystallites of 1–1.5 mm in size, nonuniform internal corrosion is observed, whereas, in the crosses with small crystallites (less than the spatial resolution), surface corrosion is observed. Whether these phenomena are related is not yet clear.
As an example of the destructive effect of fire, consider the study of the “Dancing Cupid” statuette (
Figure 4a) from the collection of the A.S. Pushkin State Museum of Fine Arts, Moscow (Pushkin Museum) [
12]. The statuette entered the Pushkin Museum in 1946 among looted art. It was previously stored in Berlin and suffered a fire in May 1945 [
13]. Due to significant damage, it was kept in the storerooms of the Moscow museum for a long time, presenting a difficult problem from the point of view of restoration. The “Dancing Cupid” statuette is cast from tin–lead bronze. It has a height of 26.5 cm. The left arm and wing are practically lost, while the right ones are damaged. The color is mainly black, and no traces of patina are observed. In some places, the surface has bulges, and the surface is also covered with spots of various colors. External examination of the figurine raised concerns that the damage affected a significant part of its material. The figurine was examined using neutron tomography. In the neutron projections of the “Dancing Cupid” statuette, the diffraction effect of “flickering” was observed throughout its entire volume (
Figure 4b,c). This phenomenon led to the conclusion that almost all the material of the figurine consists of metal, and the fear of oxidation of a significant part of the material was not confirmed. The average grain size was about 500 μm, although individual grains larger than 1 mm were observed. The appearance of large crystallites of copper oxide was not possible due to the higher melting point in comparison with the metal. Neutron tomography of the “Dancing Cupid” statuette shows significant internal damages caused by the fire. Basically, they are expressed in the detachment of a part of the material, and the number of such defects increases in the direction from the head to the feet. The shin of the figurine’s right leg suffered most; cracking occurred not only near the surface, but also in the volume of the metal (
Figure 4d).
As an example of the protective effect of fire, the reader can consider the study of an iron arrowhead from the Chernaya Mogila (Black Grave) burial mound [
14]. The mound is a 10th century pagan cremation burial, and most of the items found in it bear traces of fire. The mound is located in the city of Chernigov (present-day Ukraine), it was excavated in 1872–1873, and a significant proportion of the finds was transferred in the beginning of the 20th century to the Historical Museum in Moscow (now the State Historical Museum). The object of study was a two-horned arrowhead with a shank. Some fragments of the object were lost, and traces of restoration work are present (
Figure 5a). SR and neutron tomography of the arrowhead generally gave similar results; however, the contrast ratios of different image details differed (
Figure 5b,c). The tomography showed the presence of a dense outer part in the object and a large number of voids of various shapes inside it. It can be concluded from SR images that the densest material is concentrated in a shell 0.3–0.5 mm thick. The voids occupy about 6%, while the shell makes up about 22% of the total volume of the object. According to neutron images, the shell material has a relatively low attenuation coefficient and, thus, contains less hydrogen. In contrast, the outer layer of an object on top of the shell and the material inside the dense shell attenuate neutrons more than the dense shell. The thermal neutron diffraction experiment showed that the material of the object consists of three phases: goethite α-FeO(OH), magnetite Fe
3O
4, and a small amount of hematite α-Fe
2O
3. According to the diffraction data, the object does not contain metallic iron. Aggregation of the data suggests that the shell material of the object is magnetite, which is most clearly visible in synchrotron images, while the outer and inner parts of the object are goethite and, possibly, amorphous products of iron corrosion.
From the observed picture of the structure of the object and the diffraction data, it can be concluded that a dense layer of mill scale formed on the surface of the arrowhead in the funeral pyre. At temperatures above 575 °C, which corresponds to the temperature of the flame of a large bonfire, iron is oxidized to wustite FeO, which is further oxidized upon cooling, forming magnetite, and, on the outer surface, there is a small amount of hematite [
15]. It is the layer of magnetite and hematite that has protective properties. Formation of a layer of scale on the arrowhead ensured its preservation until the moment of excavation, since all the metal was destroyed by corrosion, which led, in particular, to the formation of an internal cavity. Due to such preservation of the arrowhead, it was possible to find and to reconstruct the original appearance of the ornament (
Figure 5d,e) [
14].
Various aspects of the study of preservation state can be considered from the example of two other artifacts from the Chernaya Mogila burial mound [
16]. The objects are oblong plates with dimensions 103 × 36 × 8 mm and 160 × 45 × 7 mm of unclear purpose. With the help of SR and neutron imaging, an ornament on them was revealed, which, together with data on other items of the collection, makes it possible to suggest that the studied artifacts are fragments of a scepter, a symbol of the power of its owner. According to the results of elemental analysis, the objects are almost entirely composed of iron corrosion products. SR imaging revealed a thin layer of mill scale near the surface of the objects, which could have contributed to the preservation of the objects similarly to the arrowhead (
Figure 6). The artifacts also exhibit both transverse cracks 0.05–0.1 mm wide and delamination in the plane of the object, the width of which reaches 1 mm. Neutron tomography showed the presence of “flickering” regions due to the effect of diffraction of monochromatic neutrons on single crystals about 1 mm in size, which indicates that the metal of the objects is partially intact. Metal regions are located in the bulk of objects and have a longitudinal dimension of about 5 mm and a thickness of about 1–2 mm. Identification of one of these regions as a preserved metal was carried out using thermal neutron diffraction (
Figure 7). A neutron beam of 0.167 nm wavelength, shaped by a 2(H) × 10(V) mm slit was directed to a section of the object in which the effect of neutron diffraction by single crystals was observed, as well as to a section where the effect was not observed. Accordingly, for the first section, diffraction lines of α-iron were recorded, and, for the second one, there were no diffraction lines of α-iron, while mainly magnetite and goethite were detected. No sample rotation was used; hence, the intensity ratio of reflections may be irregular.