Introduction

Glass has a long history of development in China that can be traced back thousands of years [1]. China’s ancient bronze smelting and refining technology originated in the Shang Dynasty (16th century BC–11th century BC) [2], which increased the development of primitive porcelain technology, when the firing temperatures of proto-porcelain generally reached 1100–1200 ℃. In the Western Zhou Dynasty and the early Eastern Zhou Dynasty (the Spring and Autumn), the furnace temperatures could have been higher. When the firing temperature was too high, the proto-porcelain glaze flowed to form glassy transparent glaze droplets of vitreous material, which may be the earliest ancient glass in China [3]. The excavation sites in China containing lead-barium silicate glass often coincide with the distribution sites of lead ore [4]. The translucent eye beads made of glass, which were excavated in Bozhou, Anhui Province in the Yangtze River Basin, were from the late Spring and Autumn Period and the early Warring States Period (6th–5th century BC) [5], which are the earliest lead-barium silicate glasses found thus far, and lead-barium silicate glass products from the middle and late Warring States Period have been found in the southern and southwestern regions of China [6, 7].

The ancient glass was buried in different environments, which generated a variety of corrosion forms during the burial process; currently, it is generally accepted that the corrosion mechanisms of glass include ion exchange (H+/alkali and alkaline earth metal ions) [8, 9]; decomposition and pitting of the Si-O-Si glass skeletal structure [10,11,12,13]; hydrolysis of metal ions that make up the various glass networks as well as PbO that makes up the glass intermediates, and one possible hydrolysis reaction is shown in Eq. (1) [14, 15]:

$${{\text{S}}_{\text{i}}}{\text{O}} - {\text{C}}{{\text{a}}^{{\text{2 + }}}} - {\text{O}}{{\text{S}}_i} + {\text{2}}{{\text{H}}_{\text{2}}}{\text{O}} \to {\text{2}}{{\text{S}}_i} - {\text{OH}} + {\text{Ca(OH}}{{\text{)}}_{\text{2}}}$$
(1)

Regarding microbial corrosion of glass, it has been found that ancient glazed tiles were covered by a biological patina composed of microalgae, cyanobacteria, bacteria and some lichenized fungi, some of which penetrated cracks and pores to produce biodegradation [16, 17]. The main effects of biodegradation on historical glass include etching, pitting, leaching, discolouration, glass degradation, gel layer formation and corrosion of the original glass surface [18,19,20,21]. The research objects showing microbial corrosion of glass relics are primarily weathered samples, and previous analyses of excavated glass relics have been focused on analysing the chemical compositions of corrosion products, mainly by studying the chemical-physical corrosion processes of cultural relics, but few papers have linked microbial corrosion to the corrosion mechanisms of unearthed artefacts.

Yaozhuang Cemetery located in Liuhe District, Nanjing city, China, was excavated in November 2021, and it is a rectangular vertical pit tomb with a straight wall and a flat bottom. The tomb entrance is 2.54 m long, 1.48 m wide and 2.48 m below the surface. The soil was very wet when the cultural relics were unearthed. Fifteen relics of the Warring States period were cleared, including 12 tombs, 1 chariot pit, 1 pottery figurine pit and 1 utensil pit. The excavated artefacts were mainly pottery combinations such as tripods, beans, Gui, Fang, and cups. In place of the head of the tomb owner in the burial coffin M342 (Fig. 1a, b), a piece of a round artefact with a broken and crispy white surface was found, which was tentatively identified as aquamarine lead-barium glass from the fracture surface. The soil water content in the excavated area is 24.2%, the pH ranges between 7 and 8, and the electrical conductivity is 47.36 µs/m. Although there was a single artefact sample, the corrosion conditions were diverse and have great research value. Optical microscopy (OM), X-ray fluorescence spectrometry (XRF), scanning electron microscopy equipped with energy dispersive X-ray spectroscopy (SEM-EDX), X-ray diffraction analysis (XRD), fluorescent inverted microscopy (FIM), and attenuated total reflectance-Fourier transform infrared spectroscopy (ATR-FTIR) were utilized to determine the composition and structure of the corroded material on the surfaces of lead-barium glass artefacts, to explore the biological corrosion mechanisms of ancient glass in buried environments, to elaborate the corrosion processes of ancient glass artefacts and to establish a relevant model. The results of the study deepen our understanding of the corrosion mechanisms of ancient lead-barium glass, provide new insight into research on the corrosion mechanisms of excavated glass relics and are also of great significance for the protection of lead-barium glass.

Fig. 1
figure 1

a Location of the burial site; b location of the sample when it was excavated

Full size image

Experimental

Ancient glass sample

Ancient glass fragments excavated from Yaozhuang Cemetery in the Liuhe District of Nanjing were selected as the objects of study. The surface condition of the specific aquamarine corroded glass is shown in Fig. 2, and the surface was covered with a thick corrosion layer. In this paper, the causes of the sample corrosion layer are analysed in detail.

Fig. 2
figure 2

Photographs of sample morphology, a front side with severe corrosion; b reverse side

Full size image

Experimental methods and detection techniques

Optical microscopic analysis (OM)

A 4 K digital microscope (Tianzhu Shanghai optical instrument factory, XTZ-4KHD) was used to observe the morphology of the corrosion layer on the sample surface, and observations were made at magnification levels of 0.7 and 3.5.

X-ray fluorescence spectroscopy (XRF)

A German Bruker M4 Tornado plus microzone X-ray fluorescence imaging spectrometer was used to observe the corrosion layer on the sample surface and test its composition and distribution. XRF is a nondestructive test that only needs a plane without sample preparation and uses an X-ray tube; the target material is Rh, the FP method and type-calibration method are used, and a national standard ore sample is used to establish a quantitative method. The accelerating voltage and beam current were 50 kV and 600 µA, respectively.

Scanning electron microscopy equipped with energy dispersive X-ray spectroscopy (SEM–EDX)

The FEI Magellan 400 field emission scanning electron microscope/energy spectrum was used to observe the microscopic morphologies of the corroded areas and cross sections of the sample fragments after chromium plating with ion sputtering and to perform compositional tests. Because the sample was fragile, the corrosion layer and cross section of the sample surface were not polished. The voltage was set to 10 kV, and the working distance varied from 4.0 to 6.4 mm under vacuum conditions. The test results included secondary electron images and backscattered electron images.

X-ray diffraction analyses (XRD)

A Bruker D8 Advance X-ray diffractometer (Germany) was used to study the sample corrosion layer. The maximum voltage of the instrument was 60 kV, the maximum tube current was 80 mA, the optical tube power was 2.2 kW (Cu target), and the angle reproducibility was ± 0.0001°. A Vantec 500 two-dimensional surface detector with a test spot diameter of 0.5 mm and an integration time of 300 s was used.

Fluorescent inverted microscope (FIM)

A Japanese Keyence BZ-X800E inverted fluorescence phase contrast microscope configured with three different filters, DAPI-V, GFP and TRITC, was used to test the corrosion layer of the sample. The transmittable light source power was 3.7 W, and the fluorescent fall-out light source power was 40 W.

Attenuated total reflectance-fourier transform infrared spectroscopy (ATR-FTIR)

An American Thermo Scientific Nicolet iS50 Fourier infrared spectrometer was used to test the corrosion layer of the sample, analyse the functional groups of the corrosion material, and determine the presence of microbial corrosion in the sample [22, 23]. Infrared absorption spectroscopy was used to directly test the corrosion area on the surface of the sample. The test range was 500–4000 cm− 1 with a spectral resolution better than 0.09 cm− 1, and the wavenumber accuracy was better than 0.01 cm− 1.

Results and discussion

Optical morphology of the corrosion surface

Figure 3 shows the OM images of the sample surface and cross-section. The glass cross-section in Fig. 3a shows that the sample substrate is blue glass, and the corrosion layer extended from the sample surface to the sample interior, up to approximately 1 mm deep inside the sample. Figure 3b, c shows that a large area of the sample surface was covered with a white corrosion layer, and the corrosion layer was distributed with pits of different sizes and shapes. A yellow corrosion material was generated inside the pits, and the yellow corrosion was in the form of loose particles, as shown in Fig. 3d.

Fig. 3
figure 3

OM images of the sample, a cross-section of the glass sample; b and c surface corrosion layer; d corrosion layer yellow pits

Full size image

Element distribution on the corrosion surface

XRF mapping was used to test the corrosion layer and determine the compositional distribution of the sample, as shown in the red box in Fig. 4a. The greater the brightness in the figure is, the higher the element content. The distribution of Si indicated in Table 1; Fig. 4b shows that the framework of the structural unit connected at the same angle as the [SiO4] tetrahedra of the lead-barium glass was corroded, with obvious corrosion pits on the surface. The contents of Pb, Si and Cu in the yellow area inside the corrosion pits had decreased significantly; however, the accumulated and aggregated Al, Fe, K and Ti showed the same morphology as the corrosion pit. Moreover, the Ba and S contents in the yellow area at the edge of the corrosion pit shown in the yellow circle of 4a were higher, indicating that the accumulation and aggregation of these two elements were consistent in the corrosion pit.

Table 1 Average contents of the chemical elements in the red box area of the sample surface in Fig. 4a (wt%)
Full size table
Fig. 4
figure 4

a XRF analysis area; b element distribution in the analysed area

Full size image

Microstructure and composition of the corrosion layer

To analyse the differences in micromorphologies and compositions of the different areas, the surface and cross-section of the lead-barium glass sample were studied with SEM-EDX, and the results are shown in Fig. 5; Table 2. The micrographs for different areas on the surface of the sample in Fig. 5 showed that there were curly layers in the areas with mild corrosion (Fig. 5a), and the contents of Fe and C were relatively high at P1; compared with the severely corroded area with obvious stratification and circular stripes (Fig. 5b, c), the content of Ba was higher at P2 on the relatively flat surface and approached 10.0%, while the contents of Pb and Ba were lower and the content of Si was higher at P3 in the adjacent striated stratification. Furthermore, in Fig. 5d, there were many cracks with widths of 3 μm on the corrosion surface, in which a large number of white particles with obvious contrast were distributed. At P4, the contents of Pb, P and Cl were 71.60%, 4.16% and 2.04%, respectively, and the Pb and P ions were dissolved and obviously dispersed. The microarea XRD results indicated that PbCO3 was formed on the surface (Fig. 6a).

Fig. 5
figure 5

SEM images of the corrosion surface in the sample, a curled layered corrosion; b circular striated corrosion; c layered corrosion; d cracking

Full size image
Table 2 Elemental compositions at different locations in Fig. 5 (wt%)
Full size table
Fig. 6
figure 6

a XRD spectrum of the white corrosion layer on the surface; b XRD spectrum recorded for the yellow corrosion layer on the surface

Full size image

SEM-EDX was used to observe the cross-sectional morphology of the sample, and the results are shown in Fig. 7; Table 3. The corrosion degree of the sample decreased gradually from the surface to the interior. The corrosion of the lead-barium glass began with a change in the lamellar curly layer on the surface (Fig. 7a). The contents of metal oxides such as PbO, Na2O and MgO gradually decreased, while the contents of Ba and S increased to match the highest contents of Ba and S in the pit, which were 37.13% and 3.75%, respectively (P10). A comparison of the microstructure of the corrosion pit surface and its interior showed that the corrosion surface formed a circular stripe with obvious stratification (Fig. 7b, c); the pits were semicircular, and the internal particles were obviously separated (Fig. 7d, e). The BaS2 crystalline particles formed in the pits were analysed by XRD (Fig. 6b).

The morphologies of the corrosion pits formed in the cross section of the sample were observed (Fig. 7d). The contents of Na, K, Al, Si and other elements in the corrosion pits were low, but the contents of S and Ba were high. The source of S in the newly generated BaS2 is worthy of further research.

The element P11 in the crack was tested, and there was a high content of P in the crack. The microarea XRD diffraction pattern showed that there was brushite [CaHPO4(H2O)2] in the corrosion layer (Fig. 6a). Additionally, there was Cl in the sample; since Cl− is an active anion with a small radius and strong penetration, it was likely to cause crack corrosion [24]. The XRD diffraction pattern revealed the presence of Pb(OH)Cl in the corrosion layer (Fig. 6a), indicating that the reaction of Cl− with Pb2+ generated Pb(OH)Cl and other products, which existed in the glass corrosion and cracks.

Fig. 7
figure 7

a SEM image of the cross section of the sample; b, c corrosion in the form of layered circular stripes in the surface layer; d and e phase separation of the corrosion pits inside the sample and particle phase separation inside the pits

Full size image
Table 3 Chemical compositions at the marked positions in Fig. 7 (wt%)
Full size table

Fluorescence-generating reactions and infrared spectra of the corroded surfaces

The source of S in the corrosion product BaS2 is unknown, the buried soil environment of the sample was complex and changeable, and there were microorganisms. It is speculated that the S came from microbial metabolites. To verify the existence of microbial corrosion, FIM was used to look for fluorescence from the surface of the corroded area. The results showed that the uncorroded area did not generate a fluorescence image upon excitation at 395 nm, 470 and 545 nm, while the corroded area showed obvious blue, green and red fluorescence with the same excitation light; the green and red fluorescence was significant in some areas with severe corrosion (Fig. 8c, d). The materials in the corrosion area exhibited self-excitation.

Fig. 8
figure 8

Fluorescence from the corroded areas of the sample, a image of the corroded area without an excitation light source; b fluorescence from the sample with excitation at 395 nm; c fluorescence from the sample with excitation at 470 nm; d fluorescence from the sample with excitation at 545 nm

Full size image

Further studies of the micromorphology and fluorescence distribution in the cross section of the corroded glass showed that there were tadpole-shaped corrosion pits on the inner surface of the sample (Fig. 9), and this area exhibited obvious self-excited fluorescence (Fig. 9b, c), which was presumed to be related to microorganisms with caudal flagella.

Fig. 9
figure 9

a Cross-sectional SEM-EDX image of the sample; b and c tadpole-shaped fluorescence

Full size image

ATR-FTIR was used to test the corrosion area, and the resulting IR spectra are shown in Fig. 10. The ATR-FTIR results indicated the presence of a biofilm and the presence of S-H bonds on the corroded surfaces of the lead-barium glass. The spectral band located between 1178 and 1050 cm− 1 arose from C–O–C skeletal vibrations or S–O stretching vibrations (Fig. 10a); the peak at 1646 cm− 1 was the amide I band, which results from a C=O stretching vibration, and the peak at 1542 cm− 1 was the amide II band, which results from an N-H bending vibration and C–N stretching vibration (the amide I and II bands are characteristic spectral bands of proteins) (Fig. 10b); S–H stretching vibration peaks were at 1950–2260 cm− 1 (Fig. 10c); −CH3 asymmetric and symmetric stretching vibration peaks were at 2960 cm− 1 and 2875 cm− 1, and −CH2 asymmetric and symmetric stretching vibration peaks were at 2920 cm− 1 and 2850 cm− 1 (Fig. 10d).

Fig. 10
figure 10

Infrared spectra of the corrosion layer, a C–O–C skeleton vibrational absorption band or S–O stretching peak; b amide I and amide II stretching vibration peak; c S–H stretching vibration peak; d −CH3 asymmetric and symmetric stretching vibration peak and −CH2 asymmetric and symmetric stretching vibration peak

Full size image

Based on the fluorescence images, morphological characteristics and infrared functional group results described above, it is speculated that microbial corrosion products were present on the surface of the sample. The sample was buried in soil containing many microorganisms for a long time. The pH of the tested soil ranged between 7 and 8, which is the right pH for bacteria, and under anaerobic conditions, silicate bacteria corrode minerals and sulfate-reducing bacteria. Silicate bacteria are microorganisms that dissolve the insoluble potassium and phosphorus salts of silicate minerals, and their metabolic processes produce organic acids, amino acids, polysaccharides and other substances. Currently, the corrosion mechanisms suggested for silicate bacteria include contact between the bacteria and the ore as well as the production of special enzymes that destroy the crystalline structures of the ores or the metabolites of the bacterial production process to decompose the ore and the surfaces via the physical and chemical contact exchange [25,26,27]. Sulfate-reducing bacteria are the most important bacteria operating under anaerobic conditions; they are obligate anaerobic gram-negative bacteria, single celled, colourless and spore free, with single flagella movements, and a metabolism that produces high concentrations of hydrogen sulfide [28, 29]. Therefore, it is speculated that the microbial corrosion of the sample may have been caused by these two bacteria. The corrosion processes of specific microorganisms will be demonstrated with experimental simulations in the future.

Corrosion model for the lead-barium glass

Based on the experimental test results described above, it was determined that the corrosion mechanisms for the lead-barium glass included microbial corrosion, surface chemical corrosion and crack corrosion. Among them, the H2S and CO32− produced by microorganisms undergo chemical reactions with barium ions and lead ions, and chemical changes are included in microbial corrosion. The corrosion models were designed according to the different conditions.

First, the lead-barium glass sample showed surface corrosion and changes in the chemical structures, and the accumulation of the corrosion products was also selective and directional. The aqueous environment of the soil led to dissolution of the silica network, which underwent a hydrolysis reaction to generate -Si-OH, which decreased the pH of the sample surface and promoted the formation of corrosion products on the sample surface [30]. Additionally, silicate bacteria destroyed the network structure of the glass, leading to decomposition of the glass skeleton, rapid dissolution of metal oxide ions containing K, Fe, Al and Ti, and obvious circular corrosion pits, while the acidic H2S produced by the anaerobic silicate-reducing bacteria reacted with Ba ions to generate BaS2, which accumulated inside the corrosion pits (Fig. 11).

Fig. 11
figure 11

Formation of corrosion pits and BaS2

Full size image

On the sample surface, the weak acids provided by biocorrosion, such as carbonic acid, or the acids contained in the soil, especially under anaerobic conditions, played an essential role in corroding the lead-barium glass; the weak acids in the environment reacted with the Pb2 ions dissolved in the glass to form a slab of sparse PbCO3 (Fig. 12).

Fig. 12
figure 12

Formation of PbCO3 on the surface

Full size image

As for the cracks in the corroded surface of the sample, Cl− easily penetrated the glass, gathered in the cracks of the corroded layer and reacted with Pb 2+ dissolved from the glass to generate Pb(OH)Cl. Additionally, HPO42− from the soil reacted with the Ca2+ dissolved from the glass to generate CaHPO4(H2O)2. The corrosion products accumulated in the cracks, which resulted in crack corrosion of the sample (Fig. 13), and the cracks gradually expanded.

Fig. 13
figure 13

Crack corrosion process

Full size image

Conclusions

Based on the X-ray fluorescence spectroscopy, scanning electron microscopy-energy spectroscopy, microarea X-ray diffraction, fluorescence microscopy and Fourier infrared spectroscopic analyses, it was determined that the corrosion products on the lead-barium ancient glass excavated from Yaozhuang Cemetery in the Liuhe District, Nanjing, included newly generated PbCO3, Pb(OH)Cl, CaHPO4(H2O) 2 and BaS2. The forms of corrosion included microbial corrosion, surface chemical corrosion and crack corrosion.

Furthermore, three different corrosion models were built by identifying the corrosion products and types. First, it is speculated that the metabolites of the microorganisms in the buried soil accelerated the destruction of the glass network, formed corrosion pits, and produced H2S in the anaerobic environment that reacted with the Ba elements from the glass; the carbonic acid in the buried soil also reacted with the Pb2+ from the glass to produce white lead carbonate. Finally, in the corroded crack areas, Cl− and HPO42− from the buried environment accumulated and penetrated deep into the corrosion cracks to generate Pb(OH)Cl and CaHPO4(H2O)2, which accelerated the crack corrosion. This study provides vital reference value for research on excavated glass relics and serves as a preliminary basis for research on the conservation of glass artefacts.