3.1 Chemical compositions of the pristine glaze
The corrosion study of the glaze was based on an analysis of the chemical composition of the pristine glaze of the samples. Table 1 shows the SEM-EDS analysis results for the chemical composition of the glaze in an uncorroded area of a sample cross-section. Based on the chemical composition, the green glaze of all five pieces is identified as a silica-aluminum oxide-lead oxide (SiO2-Al2O3-PbO) low-temperature glaze, with Pb2+ as the main flux and divalent copper ions (Cu2+) as the main colorant.
Different structures can cause the Pb glaze to react differently under an external medium.A Pb glaze contains two different structural forms of Pb: Si-O-Pb bonds (where O denotes oxygen) that act as network modifiers, and Pb-O-Pb bonds that act as network formers. At low Pb concentrations (PbO<40 mol%), the structure of PbO-SiO2 glass is close to that of an alkali silicate glass, and Pb acts as a network modifier. At high Pb concentrations (PbO>60 mol%), Pb exists in the form of a PbOn (n=3 and 4) polymer chain that forms a network. Irrespective of whether Pb acts a network former or modifier, Pb remains in a positive divalent redox state.[21][22][23] As the samples have a PbO content of 31-33 mol%, Pb acts as a network modifier in the glaze, which contains typical Si-O-Pb bonds. In these samples, CuO is the main colorant at a content of 3-5 mol%, and the iron oxide (Fe2O3) content is low at 0.5-0.6 mol%. In the Pb glaze, Cu2+ acts as a modifying ion and exists in an octahedral configuration in a glassy matrix. Compared to tetrahedral coordination (Fe), octahedral coordination has a longer bond length a larger effective volume, a more open cage structure, and places less restriction on the movement and exchange of ions in terms of geometry and energy (that is, the electron cloud density and activation energy are lower). [24][25][26]Therefore, Pb is more likely to dissolve from a green Pb glaze with Cu as the coloring element than from a yellow Pb glaze with Fe as the coloring element.
Table1 Pristine glaze chemical composition (Wt%)
Sample No.
|
Na2O
|
MgO
|
Al2O3
|
SiO2
|
K2O
|
CaO
|
Fe2O3
|
CuO
|
PbO
|
N1
|
0.4
|
0.2
|
4.4
|
28.2
|
0.4
|
0.2
|
0.7
|
2.4
|
63.1
|
N2
|
0.3
|
0.1
|
5.2
|
30.7
|
0.3
|
0.3
|
0.7
|
2.2
|
60.2
|
N3
|
0.4
|
0.1
|
4.3
|
29.6
|
0.6
|
0.3
|
0.8
|
2.9
|
61.0
|
N4
|
0.4
|
0.1
|
4.3
|
28.9
|
0.5
|
0.3
|
0.8
|
3.8
|
60.9
|
N5
|
0.5
|
0.2
|
6.4
|
30.0
|
0.3
|
0.1
|
0.5
|
2.8
|
59.2
|
3.2 Analysis of glaze corrosion morphology
Microscopic observation and compositional analysis show few differences in the chemical composition and similar corrosion morphologies of the glazes of the five samples. In this study, Sample N4 was selected as a representative sample, and its corrosion phenomenon was analyzed in detail. The N4 corroded glaze is mainly golden yellow with local iridescence and black.
3.2.1 Corrosion morphology in golden yellow area of glaze
The following results were obtained by optical microscopy (Fig.2a) and SEM-EDS (Fig.2b-e). The corrosion products in the golden crust area are distributed in layers: in the outermost golden crust, the Fe2O3 content is 49 wt%, the sulfur trioxide (SO3) content is relatively small (Table2-EDX8), and a Raman spectrum analysis indicates the presence of hematite (Fig.3b). In golden-yellow crusts, spherical substances are observed with high sulfur (S) and Fe contents, which are identified as authigenic microspheroid pyrites (Fig.3b,Table 2-EDX1). There are white and yellowish corrosion products beneath the golden crust that are mainly short columnar lead sulfate (PbSO4) (Table 2-EDX5) and hexagonal plate lead carbonate (PbCO3), where the PbCO3 crystals in local areas may also be infiltrated by surface Fe (Table 2-EDX3,EDX4). The corrosion products in the golden crust area were scraped off and found to consist mainly of PbCO3 and PbSO4 by XRD analysis (Fig.3a). The inner corrosion layer appears translucent white and was identified by SEM-EDS analysis to consist of granular aggregates with a high Si content (Fig.2e, Table 2-EDX10). This layer is referred to as a Si-rich (gel) layer in this paper and is discussed below.
Table 2 EDS analysis results of the positions marked in Fig.2 (Wt%)
Test position
|
Na2O
|
MgO
|
Al2O3
|
SiO2
|
SO3
|
Cl
|
K2O
|
CaO
|
Fe2O3
|
CuO
|
PbO
|
Possible compounds
|
EDX1
|
1.4
|
0.3
|
0.6
|
3.0
|
42.9
|
0.6
|
-
|
-
|
37.9
|
4.7
|
7.2
|
FeS2
|
EDX2
|
-
|
-
|
-
|
-
|
22.6
|
-
|
-
|
-
|
1.9
|
-
|
75.4
|
PbSO4
|
EDX3
|
2.3
|
0.1
|
2.9
|
12.1
|
-
|
-
|
-
|
0.8
|
7.5
|
0.5
|
73.8
|
PbCO3,Fe2O3
|
EDX4
|
-
|
-
|
-
|
-
|
-
|
-
|
-
|
1.2
|
-
|
-
|
98.8
|
PbCO3
|
EDX5
|
0.9
|
|
1.4
|
4.6
|
16.8
|
|
|
|
2.9
|
1.7
|
71.7
|
PbSO4
|
EDX6
|
1.1
|
0.3
|
6.5
|
34.6
|
-
|
1.2
|
1.2
|
0.9
|
9
|
4.5
|
41.9
|
Si-rich layer
|
EDX7
|
4.4
|
0.5
|
6.8
|
48
|
-
|
|
1.9
|
0.3
|
2.6
|
2.9
|
32.6
|
Si-rich layer
|
EDX8
|
0.9
|
-
|
1.5
|
8.8
|
4.4
|
2.1
|
0.7
|
0.9
|
48.8
|
0.8
|
31.0
|
Fe2O3
|
EDX9
|
0.4
|
-
|
1.1
|
5.2
|
21.7
|
1.5
|
0.5
|
-
|
1.9
|
1.8
|
65.9
|
PbSO4
|
EDX10
|
3.0
|
0.1
|
7.5
|
48.9
|
-
|
-
|
1.5
|
-
|
0.4
|
3.5
|
35.1
|
Si-rich layer
|
The elemental line distribution for a cross-section of the golden crust shows wave-like opposing distributions for the Si and Pb contents and that the sodium (Na) and potassium (K) contents are highest in the innermost corrosion layer (Fig.4c). The Fe content decreases linearly from the exterior to the interior, which presumably results from the diffusion of substances from the external environment into the corroded glaze layers. The Fe content is considerably higher in the EDX1 and EDX4 layers(Fig.4b,Table 3) than in the other layers. Thus, EDX1 and EDX4 are the Fe-rich layers of the corrosion layers with n(Fe)/n(Si)=1.4 and n(Fe)/n(Si)=1.5, respectively, and have a well-defined Si content. The micrographs show many pore structures in the Fe-rich layer (Fig.4b), presumably as a permeable deposit of Fe-rich minerals in the Si-rich gel layer. The presence of an interaction between the gel layer and Fe-rich minerals could not be decisively determined in this study. Spherical particles are attached to the surface of the corrosion layer that consists mainly of elemental S and Fe (n(S):n(Fe)≈2:1) and are authigenic microspheroid pyrites (Table 3-EDX3). The golden yellow color of the sample surface results from the deposition of hematite and pyrite.
Table 3 EDS analysis results of the positions marked in Fig.4 (Wt%)
Test position
|
Na2O
|
MgO
|
Al2O3
|
SiO2
|
SO3
|
K2O
|
CaO
|
Fe2O3
|
CuO
|
SnO2
|
PbO
|
Possible compounds
|
EDX1
|
1.4
|
1.3
|
5.4
|
18.2
|
-
|
0.3
|
0.8
|
33.2
|
3.9
|
1.4
|
34.1
|
Fe2O3,Si-rich layer
|
EDX2
|
1.0
|
4.9
|
4.7
|
8.0
|
35.6
|
0.2
|
0.2
|
16.2
|
1.6
|
12.6
|
15.0
|
FeS2,SnO2
|
EDX3
|
-
|
-
|
1.2
|
4.0
|
63.2
|
-
|
-
|
31.6
|
-
|
-
|
-
|
FeS2
|
EDX4
|
0.9
|
0.4
|
4.8
|
20.5
|
-
|
0.2
|
0.8
|
41.3
|
1.1
|
0.4
|
29.6
|
Fe2O3,Si-rich layer
|
EDX5
|
0.3
|
0.6
|
2.6
|
7.9
|
22.4
|
-
|
-
|
10.1
|
1.6
|
2.2
|
52.3
|
PbSO4
|
EDX6
|
1.1
|
0.3
|
6.5
|
34.6
|
-
|
1.2
|
0.9
|
9.0
|
4.5
|
-
|
41.9
|
Si-rich layer
|
EDX7
|
4.4
|
0.5
|
6.8
|
48.0
|
-
|
1.9
|
0.3
|
2.6
|
2.9
|
-
|
32.6
|
Si-rich layer
|
3.2.2 Corrosion morphology in iridescence area of glaze
The local area under the golden-yellow crust on the glaze surface has various colorations. Optical microscopy reveals a colored stripe on a white base and various single colors, including dark blue, yellow, and silver black (Fig.5a). In the SEM micrographs, colored fringes appear on the Si-rich layer (Fig.5b,Table 4-EDX1), and the Pb-rich layer (Fig.5b,Table 4-EDX2) has a single color and is identified as PbSO4 by Raman spectroscopy(Fig.5d). Observation of a cross-section reveals that the Si-rich gel layer and Pb-rich layer are 200-900 nm in thickness (Fig.4b), covering the wavelength range of visible light. Irradiation of the film with white light produces a thin-film interference effect. The specific interference color is controlled by the optical path difference of the coherent light, which is related to the refractive index and thickness of the film. Colored stripes appear for a film with an uneven thickness. A film with a uniform thickness of ca.200 nm (which is half the wavelength of blue light) appears blue. In the silver black area, tiny cubic crystals of galena can be observed embedded in the pores of the Si-rich layer (Fig.5c). The galena is characterized by a lead-gray color with gray-black streaks, a metallic luster, and opaqueness. It is speculated that the silver-black area primarily contains a chemical colorant, which obscures the structural color produced by the thinness of the layer.
Table 4 EDS analysis results of the positions marked in Fig.5 (Wt%)
Test position
|
Na2O
|
MgO
|
Al2O3
|
SiO2
|
SO3
|
Cl
|
K2O
|
CaO
|
Fe2O3
|
CuO
|
PbO
|
Possible compounds
|
EDX1
|
3.1
|
0.8
|
7.4
|
46.8
|
-
|
-
|
0.6
|
-
|
0.8
|
4.4
|
36.2
|
Si-rich layer
|
EDX2
|
0.6
|
0.3
|
1.9
|
8.7
|
13.2
|
2.5
|
0.1
|
0.2
|
9.4
|
2.4
|
60.7
|
PbSO4
|
EDX3
|
-
|
-
|
-
|
4.3
|
-
|
-
|
-
|
-
|
5.1
|
2.0
|
88.6
|
PbCO3
|
EDX4
|
-
|
-
|
1.4
|
10.1
|
16.9
|
-
|
-
|
-
|
1.6
|
2.2
|
67.8
|
PbS
|
EDX5
|
-
|
-
|
1.9
|
18.9
|
12.1
|
-
|
0.2
|
-
|
1.2
|
2.2
|
63.5
|
PbS,Si-rich layer
|
3.2.3 Corrosion morphology in black area of glaze
The black crust area appears black overall under the optical microscope (Fig.6a), has shed material in a scaly pattern, and the innermost local area is silvery white, and the corresponding SEM photos are shown in Fig.6b. The corrosion products in the black crust area were scraped and found to mainly contain galena, cerussite, and anglesite by XRD analysis (Fig.3a). The SEM micrograph reveals that the corrosion products in the surface layer (Fig.6c-d) are arranged in a disorderly manner: there are small cubic particles mainly containing Pb and S, with n(Pb):n(S)≈1:1, corresponding to galena; biconical dodecahedrons formed by the superposition of hexagonal thin plates mainly containing Cu and S, with n(Cu):n(S)≈1:1, that are identified as covellite from the Raman spectrum (Fig.7a); and columnar aggregates mainly containing Pb, S, with n(Pb):n (S) ≈ 1:1, that are identified as anglesite from the Raman spectrum (Fig.7b). Small particles, such as anorthite, albite, mica, and quartz, are also scattered through the surface layer. Under the surface corrosion layer (Fig.6e-h), flaky aggregates of cerussite and columnar aggregates of anglesite grow vertically on the glaze surface, and many cubic of galena are scattered in the dark Si-rich matrix. The numerous crystals grown vertically on the glaze destroy the layered structure, and the large galena content causes the corrosion glaze to appear black overall.
The inner corrosion layer (Fig.6i-o) is itself layered and appears bright white under the optical microscope. (Fig.6a) The SEM micrographs show the layered material is mainly composed of a bright layer and a dark layer. The dark layer has a relatively high content of SiO2 and is porous due to the agglomeration of granular substances with particle sizes of 50-150 nm. The pores between these particles are large, up to 500 nm in size (Fig.6l,n). The bright layer has a high PbO content, and many well-developed cubic crystal particles can be observed with sides of 1-2 μm lying flat on the porous dark Si-rich layer (Fig.6k). The plane structure and the surfaces of many crystals can strongly reflect light, manifesting as the bright silver color. Tetragonal and tetrahedral crystals with high contents of S, Fe, and Cu are also found, which are identified as chalcopyrite from a Raman spectral analysis (Fig.6o, Fig.7a).
Table5 EDS analysis results of the positions marked in Fig.6 (Wt%)
Test position
|
Na2O
|
MgO
|
Al2O3
|
SiO2
|
SO3
|
K2O
|
CaO
|
Fe2O3
|
CuO
|
PbO
|
Possible compounds
|
EDX1
|
-
|
0.6
|
25.9
|
34.1
|
-
|
-
|
21.4
|
10.9
|
2.3
|
4.6
|
anorthite
|
EDX2
|
13.3
|
0.4
|
20.3
|
61.2
|
-
|
-
|
-
|
0.2
|
0.9
|
3.7
|
albite
|
EDX3
|
0.9
|
2.5
|
33.1
|
42.9
|
-
|
7.6
|
-
|
1.8
|
1.4
|
9.5
|
mica
|
EDX4
|
-
|
0.5
|
0.9
|
2.4
|
21.7
|
-
|
-
|
0.8
|
1.2
|
72.5
|
PbSO4
|
EDX5
|
-
|
-
|
1.6
|
97.9
|
|
-
|
-
|
-
|
-
|
-
|
SiO2
|
EDX6
|
-
|
-
|
-
|
-
|
62.2
|
-
|
-
|
34.3
|
0.8
|
2.3
|
FeS2
|
EDX7
|
-
|
-
|
1.3
|
2.6
|
20.7
|
-
|
-
|
-
|
-
|
75.4
|
PbS
|
EDX8
|
-
|
-
|
1.0
|
1.9
|
42.4
|
-
|
-
|
2.2
|
40.9
|
11.8
|
CuS
|
EDX9
|
1.0
|
-
|
0.8
|
3.8
|
-
|
-
|
-
|
0.9
|
1.1
|
92.5
|
PbCO3
|
EDX10
|
2.5
|
0.2
|
3.9
|
38.8
|
-
|
0.4
|
-
|
0.6
|
2.3
|
51.3
|
Si-rich layer
|
EDX11
|
-
|
-
|
0.4
|
0.9
|
44.9
|
-
|
-
|
24.4
|
24.5
|
4.2
|
CuFeS2
|
The cross-section of the black area (Fig.8) has a similar corrosion morphology to that of the surface: the inner layer consists of parallel layers of PbS crystals to the Si-rich layer. Flaky PbCO3 crystals and columnar PbSO4 crystals are irregularly distributed in the outer layer, disturbing the parallel structure of the corrosion layer. Under a polarized light microscope (Fig.9), galena, anglesite, and cerussite have different colors of white, blue-green, and gray-white, respectively, which facilitates observation of the distribution of corrosive minerals.
Table 6 EDS analysis results of the positions marked in Fig.8 (Wt%)
Test position
|
Na2O
|
MgO
|
Al2O3
|
SiO2
|
SO3
|
K2O
|
CaO
|
Fe2O3
|
CuO
|
PbO
|
Possible compounds
|
EDX1
|
0.9
|
0.3
|
4.7
|
17.4
|
13.3
|
0.8
|
0.5
|
3.8
|
1.5
|
56.8
|
PbSO4
|
EDX2
|
1.0
|
0.4
|
6.3
|
23.5
|
6.1
|
0.7
|
0.8
|
21.0
|
3.1
|
37.1
|
Si-rich layer
|
EDX3
|
0.4
|
0
|
3.6
|
12.7
|
-
|
0.7
|
0.4
|
1.2
|
0.7
|
80.3
|
PbCO3
|
EDX4
|
1.0
|
0.6
|
3.2
|
11.5
|
16.5
|
0.2
|
0.2
|
1.5
|
2.1
|
63.2
|
PbS
|
EDX5
|
2.3
|
0.3
|
5.9
|
49.1
|
-
|
1.5
|
0.3
|
1
|
2.1
|
37.5
|
Si-rich layer
|
3.3 Analysis of Si-rich gel layer
3.3.1 Chemical composition of Si-rich gel
The corrosion layer in contact with the pristine glaze is a Si-rich layer, which is the corrosion reaction front. Thus, analysis of the Si-rich layer is important for understanding the corrosion mechanism. The glaze has multiple Si-rich layers. ,Compared with the composition of the pristine glaze in Figure 10, the Si-rich layer has significantly lower PbO content and slightly lower CuO content, and higher contents of Na2O, MgO, Al2O3, SiO2, K2O, CaO, and Fe2O3. The outer and inner (nearing the pristine glaze) Si-rich layer have quite different chemical compositions. Compared the composition of the pristine glaze, the inner Si-rich layer has approximately 20% higher SiO2 content, approximately 25% lower PbO content, approximately 1% lower CuO content, approximately 3% higher sodium oxide (Na2O) content, approximately 1% higher potassium oxide (K2O) content, and slightly higher contents of magnesium oxide (MgO), Al2O3, calcium oxide (CaO), and Fe2O3. Compared with the inner Si-rich layer, the outer Si-rich layer has significantly lower contents of SiO2, Na2O, and K2O, significantly higher Fe2O3 content, and slightly higher contents of PbO, CaO, and CuO. Table 6-EDX2 also shows a 6 wt% SO3 content in the outer Si-rich layer. The difference in the compositions of the two Si-rich layers at different locations is presumed to be related to the infiltration and deposition of materials in the ocean through the pores in the outer Si-rich layers. The Si, Na, and K originally enriched in the outer Si-rich layer are diluted by mineral deposits containing Fe, Ca, and Cu in the ocean. Cerussite and anglesite grow and penetrate the outer Si-rich layer, increasing the Pb content. In summary, the composition of the Si-rich layer is determined by the hydrolysis and polycondensation of the pristine glaze and by material penetration from the external marine environment.
To analyze elemental migration between the pristine glaze and the Si-rich layer, SEM-EDS was used to obtain an elemental line distribution diagram (Fig.11). The following content variations can be observed from the diagram. The Si content reaches a maximum in the Si-rich layer and then decreases rapidly beyond this layer; the O content reaches a maximum in the Si-rich layer and gradually decreases to stabilize at approximately 700 nm beyond this layer; the Pb content gradually increases to stabilize at approximately 500 nm beyond the Si-rich layer; and the Na content gradually decreases to stabilize at approximately 500 nm beyond the Si-rich layer. The trend in the changes in the O, Pb, and Na contents indicates diffusion of these elements, which may be related to the ion exchange reaction between Pb2+ in the glaze and hydronium (H3O+), hydrogen (H+), and Na + ions in the solution and to the entry of water (H2O) molecules into the glaze matrix. The Si-rich layer is approximately 1.6 μm thick, and it is speculated that a diffusion layer of approximately 700 nm is present between this layer and the pristine glaze, where hydration and ion exchange reactions take place. There is a clear variation in the Si content at the Si-rich layer boundary, indicating that the Si-rich layer and the pristine glaze layer are not connected due to recondensation and deposition of the hydrolyzed Si-O framework.
3.3.2 Structure of Si-rich gel
Raman spectroscopy was used to analyze the Si-rich layer and the pristine glaze layer to determine the structure of the chemical bonds. In Raman spectra of the Si-rich layer and the pristine glaze of the N4 sample (Fig.12a), peaks appear at approximately 470 cm-1 and 950 cm-1, respectively, which originate from the bending vibration and stretching vibration modes of the orthosilicate (SiO4) tetrahedron, respectively. In highly connected structures, the larger the amplitude motion of oxygen perpendicular to the Si–O–Si link, the higher is the polarization change in this vibration and Raman intensity around 500 cm-1.In less connected structure, the larger the amplitude motion of oxygen by a stretching mode, the larger is the Raman intensity at ca. 1000 cm-1. [27]Therefore, the area ratio of the bending peak to the stretching peak (A470/A950) can be used to determine the degrees of polymerization of the Si-rich layer and the pristine glaze. For the N4 sample, A470/A950 = 0.18 in the pristine glaze, and A470/A950 = 0.56 in the Si-rich layer (Table 7), indicating a higher degree of polymerization (structural connectivity) in the Si-rich layer than in the pristine glaze. As the A470/A950 ratio of the pristine glaze is low (less than 0.3), it is speculated that a low firing temperature was used to produce the N4 sample glaze.[28]
Regarding the stretching vibration of Si–O in the SiO4 tetrahedron, Raman peaks at approximately 830 cm-1, 900 cm-1, 960 cm-1, 1000 cm-1, and 1150 cm-1 have been reported in the literature for no, one, two, three, and four bridging oxygen bonds, respectively (denoted by Q0 or isolated SiO4, Q1 or -SiO3, Q2 or =SiO2, Q3 or ≡SiO, and Q4 or SiO2, respectively). [29][30]The stretching vibration peaks at 950 cm-1 in the spectra of the pristine glaze and Si-rich layer were fitted and deconvoluted (Fig.12b-d), indicating the proportions of the Q1 and Q2 components in the spectrum of the pristine glaze are relatively high, and slightly lower in the spectrum of the Si-rich layer. The proportions of Q3 and Q4 in the spectrum of the Si-rich layer are higher than those in the spectrum of the pristine glaze, indicating there are more bridging O bonds in the Si-rich layer than in the pristine glaze. Therefore, the Si-rich layer is speculated to be composed of SiO2 nanocolloidal particles formed by polycondensation of silicic acid.
Table 7 The integral area ratios of the bending envelope to stretching envelope in the Raman spectra of the pristine glaze and the Si-rich layer (A470/A950) and the peak positions and integral area ratios of each component (Qn) in the stretching envelope (AQn/A950).
Regions
|
A470/A950
|
Parameters
|
Q0
|
Q1
|
Q2
|
Q3
|
Q4
|
Pristine glaze
|
0.18
|
Peak position
|
859
|
902
|
960
|
1001
|
-
|
AQn/A950
|
0.07
|
0.34
|
0.48
|
0.11
|
-
|
Si-rich layer
|
0.56
|
Peak position
|
855
|
900
|
955
|
995
|
1130
|
AQn/A950
|
0.07
|
0.23
|
0.26
|
0.26
|
0.20
|