Corrosion of silver alloys in sulphide environments: a multianalytical approach for surface characterisation

Abstract

Sterling silver samples, prepared to simulate cultural heritage surfaces, were subjected to accelerated ageing tests through exposure to sulphide containing environments. Data obtained by X-ray diffraction, ultraviolet-visible spectrophotometry, contact angle goniometry, ellipsometry and scanning electron microscopy revealed, contrarily to what has been suggested, that the colour of corroded surfaces is related to the thickness of the corrosion layer and to the multi-layer structure of various corrosion products. At the early stages of corrosion, Cu prevails over the Ag compounds. In subsequent stages, AgCuS complexes were also detected. Ag₂S is the prevailing corrosion product after longer periods of time.

---

1. Introduction

In addition to the influence of relative humidity and temperature, the gaseous pollutants present in the atmosphere, such as H₂S, NO_x, SO₂ and O₃, act to modify the characteristics, the functionality, and the aesthetics of silver surfaces.¹ When exposed to high sulphide containing atmospheres, silver surfaces become tarnished and present a dull appearance with a colour variation that goes from yellow to dark grey. This colour range is usually ascribed to the corrosion layer thickness, which is said to be mainly composed of Ag₂S.²

The atmospheric sulphidation of pure silver has been extensively investigated as well as the associated corrosion mechanisms.^3–12 An increase of the corrosion rate under the influence of CO₂, UV radiation, and oxidizing agents, like ozone, has also been reported.^13–15

In spite of the impact of tarnishing on items made from silver and its alloys – used in semiconductors and electronic components, catalysis, jewellery, silverware, etc. – only the corrosion mechanisms of pure Ag have so far been accurately approached. It has been suggested that the corrosion of Ag–Cu alloys in H₂S environments displays two-stage corrosion kinetics: initially similar to the corrosion of pure Cu followed by an abrupt corrosion rate decrease.¹⁶ Cu can also easily react with O and S to form oxides and sulphides on Ag–Cu–Zn alloys surfaces, lowering their resistance to further tarnishing.¹⁷ In NaOH solutions, the corrosion resistance of Ag–Cu alloys is higher than that of pure Cu and pure Ag.¹⁸

In the field of cultural heritage, the majority of “silver” objects are in fact made from silver alloys usually obtained by addition of different amounts of Cu. The morphology of the corrosion products, the corroded surfaces colour, and the alloying elements role^19–21 have hardly been identified or even explored.^22,23 In this field, most publications on silver tarnishing problems focus on pure Ag and Cu.^24,25 Only a few works report the identification of corrosion products of one particular object or object type,^26–29 or on the characteristics of the binary and ternary precious metals alloys suffering from tarnishing.^30–33

The effect of minor alloying elements on the corrosion is still currently unknown.^34–36 In order to implement, for cultural heritage items, the most suitable and preventive conservation procedure and to define the best conservation treatments, each corrosion stage, in atmospheric conditions, must be fully understood and described.

This work focuses on the early stages of the process of atmospheric corrosion of Ag–Cu alloys. For this first approach, sterling silver (92.5 wt% Ag–7.5 wt% Cu) samples were exposed to high sulphide containing atmospheres for different ageing times. For a specified artificial ageing period, the colour, the morphology, and the structural composition of the corroded surfaces were characterised.

The analytical strategy for the identification of the corrosion products was mainly based on microscopy and spectroscopy techniques. The corrosion morphology was studied by scanning electron microscopy with a field emission gun (SEM-FEG). The optical characterisation of the surfaces was achieved by ultraviolet-visible-infrared (UV-Vis-IR) spectrophotometry. Energy dispersive X-ray spectrometry (EDS) was used to assess the surface elemental compositions, and X-ray diffraction (XRD) allowed the crystalline corrosion products to be identified. The wettability of the surfaces was characterised by contact angle goniometry. The corrosion layer thickness, at several different stages, was determined by ellipsometric measurements.

2. Material and methods

2.1. Silver alloy samples

The samples used in this work were made from standard sterling silver, an alloy of 92.5 wt% Ag and 7.5 wt% Cu, as coupons of 1.5 cm² area and 1 mm thickness, acquired from a local jewellery supplier (Lima & Teixeira, Lisbon). This alloy is commonly used in jewellery and silverware fabrication. The samples were grinded with emery paper of different grits up to 1000 mesh. The objective of this surface finishing was to obtain a rough surface that approaches that normal on culture heritage objects. It was not intended to replicate a perfect grinded surface. Before the ageing experiments the samples were rinsed with deionised water in an ultrasonic bath to remove grinding residues.

2.2. Accelerate ageing conditions and test

In an open container several samples were immersed in a 0.1 M Na₂S aqueous solution for different times (1 to 7 min, and then 10, 15, 30, 60, 120, 240, 480 and 1020 min). The replicating conditions were, for all the experiments, close to those corresponding to a normal atmosphere,⁷ i.e., solutions kept in equilibrium with the normal atmosphere (no forced convection) at room temperature (22 ± 2 °C). Under these conditions the oxygen concentration in the oxidising medium is about 2.9 × 10⁻⁴ M.³⁷ The local oxygen concentration is maintained only by the diffusion of the dissolved gas through the electrolytic medium. Several replicate ageing tests were done under the same conditions to check whether the surface changes for each immersion time were visually similar. After immersion, samples were removed, washed with deionised water and dried in a N₂ stream. The corroded samples characterisation was performed ex situ.

2.3. Surface characterisation

2.3.1. μXRF. The samples used in this study were cut from the same sterling silver sheet. Before artificial ageing, elemental analysis was carried out to check the surface homogeneity and to confirm the elemental composition. μXRF provide results from the bulk of the sample, therefore this technique is not suitable to characterise thin corrosion layers and was not used to characterise the corroded surfaces.³⁸

The XRF equipment was a M4 Tornado from Bruker comprising an Oxford Instruments Eclipse IV X-ray source with an anode of Rh and a 250 μm thick Be window (50 kV, 300 μA, 150 s) with a poly-capillary lens, offering a spot size down to a 25 μm at a working distance of 10 mm, coupled to a XFlash® silicon drift detector (SDD) technology, with a 30 mm² sensitive area and an energy resolution < 145 eV. The elemental quantification was performed using ESPRIT software. Analysis were carried out in twelve different spots and the obtained composition is homogeneous (93.1 ± 0.1 wt% Ag; 6.9 ± 0.1 wt% Cu). The Ag content is 0.5 wt% higher than indicated. This fact is justified by the strict requirements to obtain the legal hallmark which states that sterling silver alloy has to have at least 925 parts of silver. Consequently, the jewellery suppliers slightly increase the Ag content to assure that the final silver alloy has the required 92.5 wt% of silver.

After the artificial ageing experiments, several analytical techniques were used to fully characterise the sample surfaces and the tarnished layers.

2.3.2. UV-Vis-IR. The optical characterisation of the corroded surfaces was carried out by UV-Vis-IR spectrophotometry using a Shimadzu UV-2450PC spectrometer. Two analyses were made for each sample. The diffuse reflectance spectra (DRS) were recorded with a detecting wavelength ranging from 220 to 1400 nm, using sterling silver as reference. Absorption data (α/Λ) were calculated from reflectance data using Kubelka–Munk equations: α/Λ = (1 − R)²/(2R), where R is the reflectance and α and Λ are the absorption and scattering coefficients, respectively.

2.3.3. SEM-FEG-EDS. The surface morphology was observed with a SEM-EDS Philips XL 30 FEG model operated with acceleration voltage from 10 to 15 kV. Semi-quantitative elemental composition was obtained with an EDS (EDAX) system equipped with a Si(Li) detector with a 3 μm super ultra-thin window (SUTW), allowing detection of light elements. The X-ray spectra were collected in spot mode analysis for 300 s acquisition time. EDS analyses were performed on the top layer of each sample for two different sites with similar corrosion morphology and microstructure. For the sample immersed during 1020 min and due to surface irregularities, five analyses were made.

2.3.4. Goniometry. The wettability of the corroded surfaces was measured with a contact angle goniometer using the Krüss DSA30 instrument. 2 μL of deionised water was dropped onto the corroded surface at room temperature. A photograph of the drop was taken a few seconds later to avoid any problems related to the drop drying. With DSA3 software and using the drop image, the contact angle (CA) was measured at the intersection point of the three-phases (liquid–vapour, solid–liquid and solid–vapour) between the drop contour and the projection of the surface (baseline). Six measurements were made for each sample.

2.3.5. Ellipsometry. Ex situ spectroscopic ellipsometric measurements were performed using a Sentech SE400 Ellipsometer operating in PSA mode at a constant wavelength of 632.8 nm. The ellipsometric parameters, azimuthal angle (Ψ) and phase shift, (Δ) were measured at the incident angle of 70°. A single measurement was performed for each sample. For samples exposed to the Na₂S solution up to 3 min, the ellipsometric values were acquired at three distinct angles of incidence (ϕ = 60, 65 and 70°) in order to perform a multiangle analysis to retrieve directly the optical parameters and thickness of the corrosion layers.

2.3.6. XRD. Corrosion products were identified by XRD using a Bruker-AXS D8 Discover diffractometer in the θ–2θ configuration. The Cu K_α1 line was collimated with a Göbel mirror and an asymmetric two-bounce Ge (2 2 0) monochromator. The secondary beam passes through an anti-scattering (0.2 mm) slit and is detected with a scintillation detector. The data were collected with a step size of 0.02° (or 0.01° in grazing configuration) and an acquisition time of 600 s deg⁻¹ (or 1200 s deg⁻¹). Two diffractograms per sample were acquired. Powder diffraction files were obtained from the Pearson's Crystal Database (PCD).

3. Results and discussion

3.1. The colour of the corroded surface

On silver alloys, the surface colour modification is the main visual hint of atmospheric corrosion caused by high sulphide concentrations. With the immersion time in a 0.1 M Na₂S aqueous solution, the sterling silver samples colour varies as shown in Fig. 1. The artificially aged surfaces, when immersed for 1, 3, 5, 7, 30 and 60 min, changed from light yellow (1–2 min) to orange (3 min), red (5 min), and then to violet (7 min), blue-grey (15 min), blue (30 min) and grey (60 min). From 60 minutes onwards the surface maintains the grey colour becoming darker with the immersion time increase. These colours are similar to those published for pure silver sulphidation³⁹ suggesting that the Cu presence in the alloy does not alter the visual aspect of the corroded layer: however, as we show below, this assumption leads to the erroneous conclusion that the corrosion products formed on pure silver and Ag–Cu alloys are similar.

Fig. 1 Colour evolution of the corroded sterling silver samples for different immersion times (1, 3, 5, 7, 15, 30, 60, 120, 240, 480 and 1020 min) in a 0.1 M Na₂S aqueous solution.

Fig. 2a shows the UV-Vis-IR absorption spectra for the samples obtained after immersion times of 1, 3, 5 and 7 min. The data indicate that the colour variation is associated with the corrosion development. It is possible to establish several groups depending on the immersion times used. For all the samples an increase on the absorption bands starting at about 400 nm is observed. The samples immersed during 1 and 3 min, which had developed a yellow to yellow-orange corroded layer, are characterised by an absorption band with a maximum absorbance around 342 nm, for the sample immersed 1 min, and 500 nm, for the sample immersed during 3 min. These two UV-Vis-IR spectra profiles suggest the formation of identical species on the surface of these samples. The red shift of the 342 nm band for the sample immersed 3 min can suggest an increase in size of the formed species.^40,41

Fig. 2 UV-Vis-IR absorption spectra for the sterling silver samples artificially aged: (a) 1, 3, 5 and 7 min; (b) 15, 30 and 60 min; (c) 120, 240, 480 and 1020 min.

The UV-Vis-IR spectra obtained for the samples immersed for 5 and 7 min, which display a dark red and a light blue surface colour, show two absorption bands at 430 nm and at 800 nm for the 5 min sample, and at 430 nm and 860 nm for the 7 min sample. The presence of these two bands, which are distinct from those observed for the 1 and 3 min samples, could be caused by a difference on the Ag/Cu ratio.⁴¹ This may be explained by the formation of distinct Cu- and Ag-based corrosion products.

The UV-Vis-IR spectra obtained for the samples immersed for 30 and 60 min, shown in Fig. 2b, present one absorption band at 642 nm for the sample immersed 30 min and at 584 nm for the sample immersed 60 min. These absorption bands shift towards shorter wavelengths getting closer to the ones measured for the samples immersed 1 and 3 min. This phenomenon may be caused by the formation of new corrosion products that are similar to those observed for the first corrosion stage.

The UV-Vis-IR spectrum obtained for the sample immersed for 15 min, which has a dark blue grey colour, shows an absorption band with a maximum around 560 nm.

Fig. 2c presents the UV-Vis-IR spectra obtained for the samples immersed 120, 240, 480 and 1020 min. The 120 min sample shows an absorbance increase at 330 nm with a maximum around 720 nm. This band broadens with the immersion time increase, which can be related to the corrosion layer thickness.

The sample with the highest immersion time, 1020 min, absorbs radiation in the 400–1180 nm range.

In order to confirm the presence of Ag₂S on the samples surfaces, which should be the main corrosion product according to the literature, the band gap energy of this compound was calculated for all the UV-Vis-IR spectra. According to Chattopadhyay & Roy,⁴² depending on the particle size, the band gap energy for Ag₂S ranges from 0.72 to 1.1 eV. For the 7 min sample, the energy band gap was estimated to 0.9 eV, close to the published value for Ag₂S thin films (0.96 eV).⁴² However, it must be noted that this value is very close to the band gap energy reported for AgCuS.⁴³ Contrary to what was expected,⁴⁰ it was not possible, from the band gap energy, to confirm the presence of Ag₂S in the samples immersed during 1 and 3 min. However, by comparison with published UV-Vis spectra, the two absorption bands (318 nm and 506 nm) of the 3 min sample can be attributed to the formation of Cu₂O.⁴⁴

Although it was not therefore possible to identify the corrosion products by UV-Vis-IR spectrophotometry, the results point to a relation between the observed colours and the compounds formed during the different corrosion stages.

The contact angle (CA), formed by a liquid at an interface, is sensitive to the surface molecular structure, to the local surface roughness, and to the surface chemical composition;⁴⁵ it can be used to gather information on a corrosion film characteristics.

CA measurements were carried out on the corroded samples using the sessile drop technique in static mode. As shown in Table 1, all the samples have CAs higher than 90°, values that are characteristic of hydrophobic surfaces,⁴⁶ but a decrease of the CA values was observed for long immersion times. Sometimes, the measured angles are not enough to confirm the surface hydrophobicity once the values can be influenced by the surface roughness.⁴⁷ In fact, the slight decrease of the CA values with the immersion time increase can be related to the reduction of the surface roughness. This is in accordance with the formation of a thicker and less rough corrosion layer.

Table 1 Contact angles measured (CA) and standard deviations (SD) on the bare surface and on 1 to 1020 min artificially aged sterling silver samples

Sample ID	CA (°)	SD
0 min	102.4	5.4
1 min	104.7	3.2
2 min	104.3	3.1
3 min	105.5	4.8
4 min	108.4	8.0
5 min	103.9	4.8
7 min	114.9	6.9
15 min	111.3	10.0
30 min	94.2	4.0
60 min	110.5	12.1
120 min	97.5	11.0
240 min	96.7	4.1
480 min	91.4	6.7
1020 min	98.0	8.1

---

3.2. Thickness of the corrosion layers

Some structure-dependent material properties such as the optical constants or coating thicknesses can be assessed by ellipsometry. In order to evaluate the corrosion layer thickness and to obtain information on the dielectric constants changes across the film, the artificially aged sterling silver samples were analysed by ex situ spectroscopic ellipsometry. The highly absorber corrosion layer disabled the analyse of the samples corroded with immersion times higher than 60 min since the reflected radiation intensity was too low to be detected.

Fig. 3 presents the experimental values obtained for the azimuthal angle (Ψ) and phase shift (Δ) of the samples, related to the ratio of the Fresnel reflection coefficients of p- and s-polarised light. In order to estimate the formed film thickness, a regression analysis was applied. The experimental data were fitted to simulate Δ vs. Ψ values generated for a theoretical two-layer model that approaches the sample structure. The complex index of refraction (ñ = n − ik, n being the real part of the refractive index and k the extinction coefficient) of the corrosion product, as well as its thickness, are computed based on the best fit of the theoretical and experimental data.⁴⁸

Fig. 3 Experimental values of Δ vs. Ψ collected by ellipsometry for sterling silver samples corroded until 60 min immersion time. The angle of incidence was fixed at 70°.

The two-layer model (one homogeneous layer coating on a semi-infinite homogeneous substrate) employed to describe the samples was found to be appropriate to evaluate n, k and the thickness of the samples exposed to the corroding medium for up to 3 min. It is evident from Fig. 4 that there is a change of the “snail-shape” behaviour of the Δ vs. Ψ evolution characteristic of a film thickening, after the 3 min sample. This observation must correspond to the development of a new phase with distinct properties. Specimens with higher immersion times in Na₂S solution will require more complex physical models to be employed, but the correct definition of such a model cannot be deduced from discrete ex situ measurements and requires in situ ellipsometric monitoring of the corrosion film formation. Indeed, the observed differences of the corrosion layers suggest that they may consist of combinations of several chemical compounds, whose proportions may change throughout the film thickness, and also from one sample to another.

Fig. 4 SEM-secondary electrons (SE) images of the samples prepared by immersion in a 0.1 M Na₂S aqueous solution at different times: (A) 0 min, bare silver, (B) 1 min, (C) 3 min, (D) 5 min, (E) 7 min and (F) 15 min (scale bar is 5 μm).

Table 2 presents the complex refractive index and the estimated thickness of the corrosion films for the samples immersed from 1 to 3 min. It is worth noting the feasibility of the multiangle analysis approach to compute the ellipsometric data assuming the formation of a uniform deposit. This is only possible for isotropic layers, which is another characteristic of the very thin corrosion films studied in this work (at least for up to 3 min immersion time). The values obtained are within the range of those reported for silver alloys corrosion layers containing Ag₂S.^49,50 It should be emphasised that the small thickness of the corrosion layer grown at the early stages of artificial ageing already presents an accentuated colour alteration. Contrary to what has been reported,⁵⁰ this work indicates that a 100 nm thick film of corroded silver is not black; the sample aged for 3 min is orange and the corroded layer is 166 nm thick.

Table 2 Complex refractive index (ñ) of the corroded sterling silver alloy samples after different immersion times and calculated coating thickness (d)

Sample ID	Complex refractive index ñ	d nm
1 min	1.8 − 0i	34
2 min	1.8 − 0.5i	41
3 min	0.6 − 0.8i	166

---

3.3. Corroded surfaces composition

All the samples were studied by SEM-EDS to check whether the colour evolution can be associated to morphological changes of the species (size of the particles) formed over time or to surface composition changes (modification of the species formed).

It must be noted that sterling silver samples were abraded only to a level corresponding to a used-wear jewellery. This explains the sharp tool marks on the surface even under low magnification (Fig. 4A). No additional grinding was carried out in order to keep the samples at conditions as similar as possible to ancient cultural heritage items. For this reason, few morphological differences could be observed for each sample.

In spite of the tool marks on the surface of the samples artificially aged for 1 to 5 min, a few remaining materials from the grinding process could still be found, particularly in rougher areas.

When observed with adequate magnification it is possible to distinguish, on the rough surface close to the tool marks, the nucleation of the corrosion products (Fig. 4B–D). Moreover, the nucleation of corrosion products inside some surface scratches has already been reported for the formation of AgCl on silver wires, where it is stated that the scratched bottom facilitates heterogeneous nucleation.⁵¹ Fig. 4A–E shows a particles size increase on the samples until de seventh minute of immersion. After this period one porous film is observed on the surfaces (Fig. 4F). This data could suggest that for the early stages the corrosion mechanism is distinct.⁵²

The micrographs of the corroded surfaces immersed for 60 minutes onwards (Fig. 5A–E) show overlapping corrosion layers with different morphologies.

Fig. 5 SEM − SE images of the samples prepared by immersion in a 0.1 M Na₂S aqueous solution at: (A) 60 min, (B) 120 min, (C) 240 min, (D) 480 min and (E) 1020 min (scale bar is 5 μm).

The EDS analyses of the surface crystals (Table 3) show an increase of Cu, S and O when compared to the results obtained on the surface of samples exposed to 1 to 7 min immersion times. The ratio Ag/S provided by EDS analysis in at% is consistent with the presence of Ag₂S.

Table 3 Elemental composition by EDS (wt%), normalised to 100%, of the set of artificially aged sterling silver samples

Sample ID	Composition in wt%
	Ag	Cu	S	O
	Lα	Kα	Kα	Kα
1 min	90.1	8.5	0.5	0.9
2 min	88.5	8.0	1.8	1.7
3 min	85.6	12.6	0.7	1.1
4 min	89.5	8.1	1.3	1.2
5 min	87.6	9.0	1.9	1.4
7 min	92.1	5.3	1.2	1.4
15 min	78.4	12.6	5.8	3.3
30 min	68.5	20.4	8.7	2.4
60 min	76.3	10.2	11.0	1.4
120 min	74.7	14.5	9.7	1.1
240 min	79.0	6.7	12.8	1.6
480 min	78.9	6.7	12.8	1.7
1020 min	85.4	0.5	12.9	1.3

---

For Ag–Cu alloys, the formation of both Ag–Cu–S complexes and Ag- and Cu-based species is expected. The continuous formation of a film, which grows with time and that has a layer-by-layer morphology, is observed particularly for higher immersion times (Fig. 5C–E). This layer microstructure, previously reported for silver chloride films¹⁹ and silver sulphide films,⁴⁹ is also suggested by the UV-Vis results, which indicate the formation with time of several different corrosion products. It should be emphasised that, although the layer-by-layer morphology is observed from 120 min onwards, there is a size increase of the corrosion products crystals. This increase occurs on the crystals of the top layer but also on the intermediary corrosion layers, suggesting that there is a thickening of the intermediary layers with time.

An attempt to determine the nature of the corrosion products formed on the surface of the sterling silver alloys was made using XRD. Fig. 6 shows the diffractograms obtained for selected samples and Table 4 presents the compounds that could be so far identified.

Fig. 6 X-ray diffractograms obtained for the 2, 4, 5, 15 and 60 min corroded Ag–Cu alloy. ● metallic silver (PCD file no. 00-004-0783); ★ metallic copper (PCD file no. 00 002 1225) diffraction peaks.

Table 4 Crystalline elements and compounds identified by XRD on the surface of the corroded samples

Sample ID	Chemical compounds
0 min	Ag	Cu
1 min	Ag	Cu	Cu2O	Ag2S/AgCuS
2 min	Ag	Cu	Cu2O	Ag2S/AgCuS
3 min	Ag	Cu	Cu2O	Ag2S/AgCuS
4 min	Ag	Cu	Cu2O	Ag2S/AgCuS
5 min	Ag	Cu	Cu2O	Ag2S/AgCuS
7 min	Ag	Cu	Cu2O	Ag2S/AgCuS
15 min	Ag	Cu	Cu2O	AgCuS
30 min	Ag	Cu	Cu2O
60 min	Ag	Cu	Ag3CuS2
120 min	Ag	Cu	Ag3CuS2
240 min	Ag	Cu	Ag3CuS2
480 min	Ag	Cu	Ag3CuS2	Ag2S
1020 min	Ag	Cu	Ag3CuS2	Ag2S

---

Crystalline Cu₂O was only identified for the samples immersed until 30 minutes together with Ag–S corrosion products, either Ag₂S or AgCuS. These two latter compounds show a similar crystallographic structure and their detection is in agreement with previous reports.⁵³ The low intensity of the diffracted signal (mainly related to the thin thickness of the corrosion and to the signal contribution of the substrate) prevents an accurate identification of these two compounds.

Only from 30 min onwards, it is possible to identify the presence of both Ag₂S and AgCuS as corrosion products. For longer ageing times, the presence of Ag–Cu–S complexes prevail. The Fig. 7 diffractogram includes the diffraction peaks of jalpaite (Ag₃CuS₂) and shows that this compound is the main corrosion product of the sample aged 60 min. EDS results are in agreement with this observation, as the at% of Ag, Cu and S are consistent with the Ag₃CuS₂ stoichiometry. Some small peaks in this diffractogram, like the weak peak at 16.62°, could not be identified. This peak completely disappears for the samples immersed for longer times (120 min onwards).

Fig. 7 X-ray diffractogram obtained for the 60 min corroded Ag–Cu alloy with the identification of the Ag, Cu and Ag₃CuS₂ patterns. ● Ag (PCD file no. 00-004-0783); ★ Cu (PCD file no. 00 002 1225), □ Ag₃CuS₂ (PCD file no. 00 012 0207) diffraction peaks.

It is interesting to point out that for long immersion times (higher than 30 min), and contrary to what is referred by several authors,¹⁷ it is AgCuS that is the most prevalent corrosion product instead of Ag₂S and copper sulphides (Cu₂S and CuS). Ag₂S was identified on samples immersed between 1 and 7 min and on samples immersed for 480 min and 1020 min. This compound was not detected on samples immersed for 15, 30, 60, 120 and 240 min. This could be justified by the presence of either non-crystalline species (concentrations below detection limits) or by a multilayer structure composed of several corrosion products, but not including Ag₂S.

The XRD results obtained for the analysed samples suggest the presence of different corrosion products, which nature depend on the time of immersion. However, as XRD cannot resolve information as a function of depth at this level, it was impossible to determine whether the layer-by-layer corrosion process, identified by SEM, corresponds to the formation of distinct corrosion products.

4. Conclusions

In spite of extensive investigation conducted on the atmospheric corrosion of pure silver, the corrosion mechanisms of Ag–Cu alloys in the field of cultural heritage are poorly studied. In this work, sterling silver, an alloy often used to fabricate jewellery and silverware, was corroded by immersion in a Na₂S solution, reproducing a high sulphide environment.

The main visual characteristic of the silver alloys sulphidation is the change of the surface colour, from yellow to violet, blue grey, and dark grey. The colour variations obtained for pure Ag and our Ag–Cu alloy are the same, which could suggest the formation of the same corrosion products. However, the analyses presented here using different techniques on the silver alloy samples showed the formation of different layer-by-layer corrosion films developing over time.

Data obtained by UV-Vis-IR spectrophotometry confirms that the colour variation is related to the corrosion process development. For the early stages of corrosion, similar species are formed as the film increases in thickness. This is followed by the formation of distinct corrosion products of apparently different composition. For higher corrosion times the data suggests the appearance of a further corrosion step with the formation of species similar to those formed during the early stages of corrosion. The calculated band gap energies are in agreement with the formation of Cu based compounds for the early stages of corrosion and either Ag₂S or AgCuS for longer corrosion times.

The ellipsometry results support the formation of a film growing thicker for early stages of corrosion and then the formation of a surface with distinctly different characteristics. The film thickness for the early stages of corrosion was estimated to be about 166 nm.

With SEM-FEG was possible to observe clearly the formation of a layer-by-layer structure with the nucleation of small and slightly spherical particles that agglomerate over time to form a porous film. For longer corrosion times an increase of the corrosion products crystal size was observed.

The corrosion products identified by XRD confirmed the formation of distinct compounds at different corrosion stages. During the early stages of corrosion, Cu-based compounds prevail over the Ag-based ones. In the following stages, the predominance of Ag–Cu–S complexes is observed, instead of Ag₂S as suggested in the literature.

Moreover, the morphology of the layer-by-layer film can also be attributed to the overlapping of films of distinct compositions continuously growing at different rates. Further work, using in situ ellipsometric monitoring, is being carried out to gather detailed information on each corrosion layer and to determine its characteristics and thickness.

Acknowledgements

We thank Dr James Tate from National Museums Scotland and the reviewers for the useful comments and suggestions provided for improvement of the manuscript. This research was partially supported by the Fundação para a Ciência e Tecnologia (SFRH/BDE/51439/2011 and ID/MULTI/00612/2013).

Notes and references

1. J. Novakovic, P. Vassiliou and E. Georgiza, Int. J. Electrochem. Sci., 2013, 8, 7223.
2. H. Lin, G. S. Frankel and W. H. Abbott, J. Electrochem. Soc., 2013, 160, C345.
3. H. Gil, C. Buitrago and A. Echavarría, J. Solid State Electrochem., 2015, 19, 1817.
4. C. E. Lemon, PhD thesis, Ohio State University, 2012.
5. B. V. Salas, M. S. Wiener, G. L. Badilla, M. C. Beltran, R. Zlatev, M. Stoycheva, J. de Dios Ocampo Diaz, L. V. Osuna and J. T. Gaynor, in Corrosion of electronic components, ed. G. L. Badilla, B. Valdez and M. Schorr, Intech, Rijeka, Croatia, 2012, pp. 263–286.
6. S. Pretty, P. Keech, X. Zhang, J. Noël and J. Wren, Electrochim. Acta, 2011, 56, 2754.
7. C. Kleber, R. Wiesinger, J. Schnöller, U. Hilfrich, H. Hutter and M. Schreiner, Corros. Sci., 2008, 50, 1112.
8. M. Bernard, E. Dauvergne, M. Evesque, M. Keddam and H. Takenouti, Corros. Sci., 2005, 47, 663–679.
9. H. Kim, Mater. Corros., 2003, 54, 243–250.
10. L. Volpe and P. Peterson, Corros. Sci., 1989, 29, 1179.
11. J. Franey, G. Kammlott and T. Graedel, Corros. Sci., 1985, 25, 133.
12. T. Graedel, J. Franey, G. Gualtieri, G. Kammlott and D. Malm, Corros. Sci., 1985, 25, 1163.
13. R. Wiesinger, I. Martina, C. Kleber and M. Schreiner, Corros. Sci., 2013, 77, 69.
14. R. Wiesinger, M. Schreiner and C. Kleber, Appl. Surf. Sci., 2010, 256, 2735.
15. Z. Y. Chen, D. Liang, G. Ma, G. S. Frankel, H. C. Allen and R. G. Kelly, Corros. Eng., Sci. Technol., 2010, 45, 169.
16. P. Castello, Y. Niu and F. Gesmundo, Corros. Sci., 1999, 41, 919.
17. J. Hirunyagird, G. Lothongkum and E. Nisaratanaporn, Adv. Mater. Res., 2014, 894, 138.
18. F. H. Assaf, A. M. Zaky and S. S. A. El-Rehim, Appl. Surf. Sci., 2002, 187, 18.
19. S. Capelo, P. Homem, J. Cavalheiro and I. Fonseca, J. Solid State Electrochem., 2013, 17, 223.
20. P. Homem, PhD thesis, Faculdade de Letras, Universidade do Porto, 2013.
21. G. Ingo, E. Angelini, C. Riccucci, T. de Caro, A. Mezzi, F. Faraldi, D. Caschera, C. Giuliani and G. D. Carlo, Appl. Surf. Sci., 2015, 326, 222.
22. P. Homem, I. Fonseca and J. Cavalheiro, Corros. Prot. Mater., 2008, 27, 82.
23. M. Dowsett, A. Adriaens, M. Soares, H. Wouters, V. Palitsin, R. Gibbons and R. Morris, Nucl. Instrum. Methods Phys. Res., Sect. B, 2005, 239, 51.
24. R. Wiesinger, R. Grayburn, M. Dowsett, P. J. Sabbe, P. Thompson, A. Adriaens and M. Schreiner, J. Anal. At. Spectrom., 2015, 30, 694.
25. I. Martina, R. Wiesinger and M. Schreiner, J. Raman Spectrosc., 2013, 44, 770.
26. G. Marchand, E. Guilminot, S. Lemoine, L. Rossetti, M. Vieau and N. Stephant, Heritage Sci., 2014, 2, 5.
27. A. Mezzi, C. Riccucci, T. de Caro, E. Angelini, F. Faraldi, S. Grassini and V. K. Gouda, Surf. Interface Anal., 2014, 46, 801.
28. E. A. Marquis, Y. Chen, J. Kohanek, Y. Dong and S. A. Centeno, Corros. Sci., 2015, 94, 438.
29. E. Angelini, T. de Caro, A. Mezzi, C. Riccucci, F. Faraldi and S. Grassini, Surf. Interface Anal., 2012, 44, 947.
30. I. Tissot, L. Troalen, M. Manso, M. Ponting, M. Radtke, U. Reinholz, M. Barreiros, I. Shaw, M. Carvalho and M. F. Guerra, Spectrochim. Acta, Part B, 2015, 108, 75.
31. M. F. Guerra and I. Tissot, X-Ray Spectrom., 2016, 45, 5.
32. A.-M. Hacke, C. M. Carr and A. Brown, in Proceedings of the International Conference on Metals Conservation, METAL04, ed. J. Ashton and D. Hallam, Canberra, Australia, 2004, pp. 415–426.
33. M. B. McNeil and B. J. Little, J. Am. Inst. Conserv., 1999, 38, 186.
34. M. F. Guerra and I. Tissot, Nucl. Instrum. Methods Phys. Res., Sect. B, 2013, 306, 227.
35. T. Palomar, B. R. Barat, E. García and E. Cano, J. Cult. Herit., 2016, 17, 20.
36. P. Storme, O. Schalm and R. Wiesinger, Heritage Sci., 2015, 3, 2.
37. M. Huang, Y. Shao, X. Sun, H. Chen, B. Liu and S. Dong, Langmuir, 2005, 21, 323.
38. R. Linke and M. Schreiner, Microchim. Acta, 2000, 133, 165.
39. T. Baše, Z. Bastl, V. Havránek, K. Lang, J. Bould, M. G. S. Londesborough, J. Machaček and J. Plešek, Surf. Coat. Technol., 2010, 204, 2639.
40. S. Baset, H. Akbari, H. Zeynali and M. Shafie, Digest Journal of Nanomaterials and Biostructures, 2011, 6, 709.
41. M. Valodkar, S. Modi, A. Pal and S. Thakore, Mater. Res. Bull., 2011, 46, 384.
42. P. Chattopadhyay and S. Guha Roy, Indian J. Pure Appl. Phys., 2013, 51, 98.
43. S. N. Guin, J. Pan, A. Bhowmik, D. Sanyal, U. V. Waghmare and K. Biswas, J. Am. Chem. Soc., 2014, 136, 12712.
44. Y. Yu, L. Zhang, J. Wang, Z. Yang, M. Long, N. Hu and Y. Zhang, Nanoscale Res. Lett., 2012, 7, 347.
45. Y. Yuan and T. Lee, in Science Techniques, ed. G. Bracco and B. Holst, Springer, Berlin Heidelberg, 2013, pp. 3–34.
46. J. Drelich, E. Chibowski, D. D. Meng and K. Terpilowski, Soft Matter, 2011, 7, 9804.
47. N. Saleema, D. K. Sarkar, R. W. Paynter, D. Gallant and M. Eskandarian, Appl. Surf. Sci., 2012, 261, 742.
48. L. Abrantes, J. Correia and A. Melato, J. Electroanal. Chem., 2010, 646, 75.
49. R. van Langh, H. A. Ankersmit and I. Joosten, in Proceedings of the International Conference on Metals Conservation, METAL04, ed. J. Ashton and D. Hallam, Canberra, Australia, 2004, pp. 137–141.
50. L. Selwyn, Metals and corrosion. A handbook for the conservation professional, Canadian Conservation Institute, Ottawa, 2004.
51. H. Ha and J. Payer, Electrochim. Acta, 2011, 56, 2781.
52. C. Kebler, U. Hilfrich and M. Schreiner, Appl. Surf. Sci., 2007, 253, 3712.
53. Y. Asadov, Y. Alyev and K. Jafarov, Inorg. Mater., 2008, 44, 460.

---