The Tarnishing of Silver in Museum Collections: A Study at the National Archaeological Museum (Spain)

Abstract

Silver tarnishing in museum environments depends on multiple, interacting factors that are not often studied in situ. With the aim of addressing the problem in real-world scenarios, this study presents a one-year assessment at the National Archaeological Museum of Spain, in Madrid, a museum that houses a significant collection of silver objects. Pure Ag coupons were placed in four display cases—two designs with different airtightness—and in an adjacent gallery. Tarnishing was quantified by colorimetry, gravimetry, and galvanostatic reduction, and analyzed in relation to environmental parameters (T/RH) and gaseous pollutants (H₂S, SO₂, HF, HCl, formic and acetic acids), measured with passive samplers. Coupons showed different degrees of tarnish, with annual corrosion rates ranging from IC1 (very low) to IC2 (low), without a straightforward relation to hydrogen sulfide concentrations. Electrochemical profiles and XPS on representative coupons identified Ag₂S as the dominant product, with AgCl and minor Ag₂SO₄ in the coupons exposed outside the airtight cases, indicating different contributions inside and outside the cases. Findings highlight that sulfide concentration is not the sole driver; case airtightness, internal materials, cleaning products used on adjacent areas, and, possibly, other aspects influence silver tarnishing.

1. Introduction

Preserving silver cultural artifacts in museum collections remains challenging. Despite being a noble metal, silver is highly sensitive to sulfides, one of the most frequent airborne pollutants found in indoor environments, such as museum galleries and display cases. Silver corrosion caused by reaction with sulfides results in surface tarnishing, which, while not endangering the physical integrity of the objects, has a significant visual impact. Silver objects were created to be shiny; therefore, to preserve them as originally conceived, conservator-restorers are compelled to carry out periodic cleanings, which inevitably cause wear of the original material [1].

Although silver sulfidation has been studied for a long time [2,3], there is no consensus regarding the influence of the different factors involved, since most studies address only partial aspects. Moreover, the vast majority of studies are conducted under laboratory conditions that do not reflect the complexity of real museum environments, where multiple variables converge, such as the materials used in displays, differing environmental conditions, and different types of objects. As far as we have been able to ascertain, only the British Museum has conducted systematic research on the impact of environmental conditions on silver sulfidation [4,5]. In the first of these studies, in which pure silver test samples were exposed in various display cases and galleries, no correlation was observed between the degree of tarnishing and the measured sulfide concentration, nor with the relative humidity (RH) and temperature (T) values. This indicates that the problem is much more complex and suggests that other factors may play a relevant role in sulfidation processes, such as airflow within display cases and exhibition spaces, the nature of display materials, or the characteristics of the exposed metal surface.

The National Archaeological Museum in Madrid has more than 13,000 objects on display across 40 galleries, including a wide variety of pieces—ceramics, sculptures, tools, weapons, jewelry, and coins —from Antiquity to recent times. The museum’s extensive metal collection is particularly noteworthy, with silver objects predominating in some collections. The sulfidation problems encountered, particularly in certain display cases, prompted the present study to understand the factors affecting the tarnishing of silver in a real museum environment. In order to define and delimit the most relevant aspects, an initial assessment was conducted at five representative locations, quantifying the corrosion experienced by a set of silver coupons over 12 months, together with environmental data (temperature, relative humidity, and pollutant concentrations). To advance the understanding of other factors that may influence silver alteration, the results were analyzed in relation to the characteristics and surroundings of the display cases, seeking to identify sources of emissions responsible for sulfidation and other variables that may affect the rate or extent of its occurrence. The ultimate goal is to provide a general basis for designing more effective preventive conservation strategies for silver objects, tailored to real-world contexts.

2. Materials and Methods

2.1. Selection of Display Cases

The coupons were placed in four display cases from two manufacturers—designated as type 1 and type 2—and in the gallery where the cases are located. Both types are constructed from the same kinds of materials, as shown in

Table 1. Nomenclature and information on the display cases selected for the study.

Ref.	Case ID	Materials	Objects	Dimensions
MAN 1	14.8	Galvanized steel, anodized aluminum, and glass. Plinths and modules: Foamed PVC (Simopor®)	Painted ceramic, clay, silver, bronze, gold, and iron	450 × 190 × 160 cm (13.68 m3)
MAN 2	13.4		Painted ceramic, silver, and bronze	364 × 190 × 80 cm (5.53 m3)
MAN 3	11.9		Silver and bronze	270 × 190 × 80 cm (4.1 m3)
MAN 4	10.1	Galvanized steel, baked enamel aluminum sheet, and glass. Plinths and modules: Foamed PVC (Simopor®)	Painted ceramic, gold, silver, bronze, iron, lead, and stone	590 × 230 × 100 cm (13.57 m3)
MAN 5	Gallery 11	Marble floors and plaster walls	-	337 m2

: anodized or lacquered metal, glass, and foamed PVC (Simopor^®, Simona Group, Kirn, Germany ) in the interior modules. The main difference between the two types is their level of airtightness: type 1 display cases (MAN 1, MAN 2, and MAN 3) are more airtight than type 2 (MAN 4). This is relevant because it governs the entry of external pollutants and the accumulation of those generated internally, where present. With regard to the exhibited objects, in all cases, they are inorganic materials—ceramics, clay, stone, and various metals—with cases 13.4 and 11.9 containing the largest number of silver objects.

2.2. Preparation and Exposure of Silver Coupons

Six coupons made of pure silver (99.95%, Goodfellow, Cambridge, UK) measuring 50 × 10 × 0.2 mm with a 4.5 mm diameter hole drilled in the top were prepared for each of the five locations (30 in total) to assess the corrosiveness of the different locations. Coupons were sanded with 1200-grit sandpaper and degreased with ethanol in an ultrasonic bath for one minute. Then, they were hung on a fiberglass rod on a polyethylene terephthalate (PET) frame (Veralite^®, IPB, Waregem, Belgium) with 1 cm of separation and displayed in the museum’s display cases in accordance with ISO 11844-2 (Corrosion of metals and alloys—Classification of low corrosivity of indoor atmospheres. Part 2: Determination of corrosion attack in indoor atmospheres) [6] as shown in Figure 1. Each series of coupons with its location has been labeled with a reference number from 1 to 5 (MAN 1, MAN 2, MAN 3, MAN 4, MAN 5); the correspondence between the label and the location is explained in Table 1. Two removals of three coupons each were planned at 6 and 12 months (24 and 48 weeks).

2.3. Monitoring of Environmental Conditions

Temperature and relative humidity were recorded throughout the entire exposure period of the coupons using a Lascar EL-USB-2 thermohygrometer at each location, at 1-h intervals. Pollutant concentrations were measured by the IVL, Swedish Environmental Research Institute Laboratory, using passive samplers exposed alongside the coupons for 4 weeks. One device was placed for H₂S (concentration range: 0.1–200 µg/m³) and another for SO₂ (0.1–100 µg/m³), HF (0.2–40 µg/m³), HCl (0.3–100 µg/m³), HCOOH (1.5–200 µg/m³), and CH₃COOH (1.5–250 µg/m³).

2.4. Color Measurements

The visual impact of tarnishing was evaluated by color measurements of the coupons using a Konica Minolta CM-700-d spectrophotometer with standard illuminant D65 and a 10° observer, with a 6 mm diameter mask. Six measurements per coupon were taken of the L*, a*, and b* coordinates in the CIELab 1976 color space, SCI, and the average value of the six measurements was calculated.

2.5. Determination of the Corrosion Rate

To quantify coupon corrosion, gravimetric and electrochemical methods were used, based on ISO 11844-2:2020 standard. First, mass change was determined by weighing the coupons before and after exposure with an M5 microbalance (Mettler Toledo, Barcelona, Spain) with a sensitivity of 0.001 mg. This quantifies the mass gain due to the formation of corrosion products in mg/m² year.

After weighing, the coupons were electrochemically reduced in a three-electrode cell designed ad hoc for this type of coupon [7], applying a constant current density of 125 µA/cm² using a Gamry Reference 600 potentiostat, with an Ag/AgCl (3M) reference electrode and a graphite counter electrode. The electrolyte was a 0.1 M KCl solution, deaerated for 60 min by nitrogen bubbling to avoid interference from dissolved oxygen.

From the reduction time, the amount of reduced silver was calculated using Faraday’s equation:

$$m = {\displaystyle \frac{Q·M}{n·F}}$$

(1)

where the following is true:

Q is the charge consumed (I·t), where I is the current in amperes and t is the time required for complete reduction of the corrosion products, in seconds;

M is the molar mass, in g/mol, i.e., 107.9 for silver;

n is the oxidation state of the metal, +1 for silver;

F is Faraday’s constant, 96,485 C/mol.

From the mass of reduced silver, the corrosion rate is expressed in mg/m² year.

For the calculation of the corrosion rate, the nominal area of the coupons was used. In the gravimetric case, this is calculated as the area of each face minus the area of the drill hole (multiplied by 2 faces), plus the area of the edges. For the electrochemical reductions, the reduction is performed on the lower 4 cm. The resulting values are 992 mm² and 820 mm², respectively.

2.6. Identification of Corrosion Products

Identification of corrosion products was performed from the reduction potentials of the coupons and by X-ray photoelectron spectroscopy (XPS). For XPS, a SPECS GmbH (Berlin, Germany) instrument with a UHV system, a PHOIBOS 150 9MCD energy analyzer, and a monochromatic X-ray source (Al anode) was used.

Depth-profiling XPS analysis was carried out by argon-ion sputtering in order to confirm the distribution of corrosion compounds on silver after exposure. Given that corrosion product layers formed under museum atmospheric exposure are expected to be very thin, two sputtering conditions were used to obtain the concentration profiles. The first 10 min were performed at 3000 eV; the next 10 min at the same energy; and the final 10 min at 5000 eV. Therefore, the total sputtering time was 30 min.

3. Results

3.1. Environmental Conditions

The temperature and humidity values were adequate and stable in all locations during the recorded period (see

Table 2. Environmental conditions and pollutant concentrations measured at different locations. Detailed RH and T measurements are available at https://doi.org/10.20350/digitalCSIC/17838.

Location	Pollutant Level					Environmental Conditions
	SO2 µg/m3	HCOOH µg/m3	CH3COOH µg/m3	HCl µg/m3	H2S µg/m3	T (°C) avg.	T (°C) max	T (°C) min	RH avg.	RH max	RH min	Time RH ≥ 35%
MAN 1	0.20	44	<2.0 *	<0.2 *	>7.5 **	22.0	24.5	19	35.5	39.5	31	67%
MAN 2	0.32	36	<2.0 *	<0.2 *	>7.5 **	21.6	23.5	19	35.6	39.5	33	77%
MAN 3	1.5	33	<2.0 *	<0.2 *	>7.5 **	21.7	24	19	35.4	39	33	69%
MAN 4	0.45	12	<2.0 *	<0.2 *	3.3	22.1	24	20.0	35.4	42	21.5	62%
MAN 5	0.42	7.0	<2.0 *	<0.2 *	3.3	21.9	24	19.5	34.9	44	20.5	58%

* Below detection limit; ** above saturation limit.

). The average relative humidity (RH) is 35%, which is considered safe for silver [8], although it is slightly higher around two-thirds of the time, especially in type 1 display cases. RH did not once reach 50%, the value at which the thickness of the adsorbed water layer exceeds a monolayer and can act as an electrolyte, dissolving and transporting gaseous contaminants [3]. Despite these values, some of the exposed coupons have visibly tarnished.

With regard to pollutants, the level of sulfides is high at all locations, reaching saturation of the passive samplers in three of the cases studied and far exceeding the recommended limits, which for H₂S are generally set at 1 µg/ m³ year (0.7 ppb), but for particularly sensitive materials such as silver are reduced to 0.10 µg/m³ year, (for an RH of 50–60% and a temperature between 20–30 °C) [9]. These values, according to the standards of the Canadian Conservation Institute (CCI) and the American Society of Heating, Refrigerating and Air-Conditioning Engineers (AHSRAE), refer to the minimum quantity that produces an observable change, the LOAED (Lowest Observable Adverse Effect Dose) for H₂S. From this concept, it is possible to extrapolate the exposure time at a given pollutant concentration required for appreciable damage to appear on a material (such as a color change). The lower the pollutant concentration under given environmental conditions, the longer it will take before the damage becomes observable [10].

Formic acid concentrations are also appreciable, although they do not exceed the most recent limits proposed by AHSRAE in general terms, and they do not pose a problem for silver [10]. The rest of the pollutants do not show significant concentrations, except for a slight excess of SO₂ in the MAN 3 showcase, with no apparent effects.

3.2. Tarnishing of Silver Coupons After Exposure

After exposure of the coupons at the different locations in the museum, some differences are observed. Silver remains practically intact in the MAN 2 and MAN 3 cases, becomes slightly sulfidized in the MAN 4 case, and is markedly tarnished in the MAN 1 case and in gallery MAN 5 (Figure 2). In those locations where sulfidation of the coupons occurs, the effect is already noticeable after 6 months of exposure.

The color variation experienced by silver can be seen in

Table 3. Color variation of the coupons exposed for 6 and 12 months relative to a silver reference coupon.

	ΔL*		Δa*		Δb*		ΔE
	6 m	12 m	6 m	12 m	6 m	12 m	6 m	12 m
MAN 1	−28.71	−52.8	19.8	23.44	41.02	8.57	53.84	58.40
MAN 2	−1.73	−4.06	0.08	0.69	1.4	1.87	2.23	4.52
MAN 3	−3.72	−4.27	0.14	0.47	1.38	5.14	3.97	6.70
MAN 4	−5.6	−7.37	0.03	0.29	4.41	16.42	7.13	18.00
MAN 5	−7.91	−17.63	0.83	5.31	7.49	27.03	10.93	32.71

. The MAN 1 and MAN 5 coupons show a significant variation relative to the reference, with a decrease in lightness (L*) and a shift of the a* component toward red and b* toward yellow. In addition, MAN 1, which exhibits the greatest alteration (Figure 2), begins to trend toward a more bluish component as the exposure time increases to 12 months compared to 6 months.

These color differences are reflected in the mass gain (

Table 4. Mass gain of the coupons exposed at each location for 6 and 12 months (coupon MAN1-6m-3 was discarded for yielding an inconsistent result).

Coupon	Δm (6 m)/mg	Coupon	Δm (12 m)/mg
MAN1-6m-1	0.026	MAN1-12m-1	0.041
MAN1-6m-2	0.020	MAN1-12m-2	0.036
MAN1-6m-3	-	MAN1-12m-3	-
MAN2-6m-1	0.002	MAN2-12m-1	0.002
MAN2-6m-2	0.004	MAN2-12m-2	0.002
MAN2-6m-3	0.005	MAN2-12m-3	0.000
MAN3-6m-1	−0.001	MAN3-12m-1	0.000
MAN3-6m-2	−0.001	MAN3-12m-2	0.002
MAN3-6m-3	0.002	MAN3-12m-3	0.000
MAN4-6m-1	0.008	MAN4-12m-1	0.014
MAN4-6m-2	0.006	MAN4-12m-2	0.013
MAN4-6m-3	0.008	MAN4-12m-3	0.010
MAN5-6m-1	0.016	MAN5-12m-1	0.033
MAN5-6m-2	0.021	MAN5-12m-2	0.028
MAN5-6m-3	0.011	MAN5-12m-3	0.020

); moreover, comparing the 6 m and 12 m values, sulfidation can be considered linear, since the mass increase is approximately double for most coupons. Mass change in the MAN 2 and MAN 3 coupons is negligible, slight in MAN 4, and much higher in MAN 1 and MAN 5.

The same results are obtained after galvanostatic reduction (Figure 3) with respect to the degree of tarnishing; however, differences are observed in terms of composition. In the profile of the reduction curves, a plateau is observed around −0.7 V vs. Ag/AgCl, corresponding to the reduction of silver sulfides; this plateau is evident in the MAN 4 and MAN 1 cases and in gallery MAN 5, minimal in the MAN 3 case at 12 months, and practically non-existent in the rest. In the curves of the coupons from the MAN 4 case and from gallery MAN 5, another small plateau appears slightly above 0 V vs. Ag/AgCl which, according to the reduction potentials (

Table 5. Standard reduction potentials for some silver salts [12,13].

Reaction	Eº/V	Ag/AgCl (3M KCl)
Ag2SO4 + 2e− ⇆ 2 Ag + SO42−	0.654	0.446
2 AgO + H2O + 2e− ⇆ Ag2O + 2OH−	0.607	0.399
Ag2O + H2O + 2e− ⇆ 2Ag + 2OH−	0.342	0.134
AgCl + e− ⇆ Ag + Cl−	0.22233	0.014
Ag2S + 2e− ⇆ 2 Ag + S2−	−0.691	−0.899

), could be related to the presence of AgCl or Ag₂O. Considering the characteristic reactivity of silver, it is more logical to assume that this is silver chloride [11], as confirmed by the XPS analysis (see the following section).

Based on the gravimetric and electrochemical data, the average corrosion rate of silver and the atmospheric corrosivity classification of each location were calculated in accordance with ISO 11844-1:2020 [14]. The corrosion rate (Vcorr) is presented in

Table 6. Annual corrosion rate and atmospheric corrosivity category for different locations.

	Reduction (mg Ag)		Gravimetry (Mass Gain, mg)
Location	Vcorr (mg/m2 Year)	Category	Vcorr (mg/m2 Year)	Category
MAN1 (14.8)	274.73	IC 2	46.40	IC 2
MAN2 (13.4)	9.60	IC 1	4.91	IC 1
MAN3 (11.10)	11.82	IC 1	1.00	IC 1
MAN4 (10.1)	57.36	IC 1	14.83	IC 1
MAN5 (Gallery)	76.35	IC 1	29.04	IC 1

, gravimetrically as mass gain per m² and year, and electrochemically as mass of silver corroded per m² and year, as described in the methodology. The corrosivity values correspond to category IC 1 (very low; Vcorr ≤ 170 mg/m²·year) for all locations except display case MAN 1, classified as IC 2 (low; Vcorr ≤ 670 mg/m²·year).

3.3. Identification of Corrosion Products

To confirm the composition of the corrosion layer, an in-depth XPS analysis was performed on one of the MAN5 coupons, which exhibited a more complex reduction curve. Survey spectrumof the fresh sample indicates the presence of O, C, Ag, Cl, and S. Figure 4 shows XPS spectra obtained for Ag and S both in the initial sample (fresh sample) and after each Ar sputtering time (10 min at 3000 eV + 10 min at 3000 eV+ 10 min at 5000 eV). The atomic percentage can be calculated after fitting of the spectra and it is shown in

Table 7. Calculated atomic percentage of the elements identified in the fresh sample and after each Ar sputtering time (AIB).

	% at Element
Sample	O	C	Ag	Cl	S
Fresh	11.7	35.17	39.4	5.4	8.4
10 min AIB (3000 eV)	6.3	13.2	63.2	7.3	10.1
20 min AIB (+10 min 3000 eV)	4.6	9.4	69.4	5.5	11.1
30 min AIB (+10 min 5000 eV)	1.1	2.4	96.5	nd	nd

.

The high-resolution Ag 3d spectrum in Figure 4 displays two bands at about 368 eV and 374 eV, which could be assigned to Ag 3d_5/2 and Ag 3d_3/2 binding energies, respectively [15]. Upon further analysis, the two bands were fitted into four distinct bands at 367.9 eV, 368.5 eV, 373.9 eV, and 374.5 eV. The bands at 368.5 eV and 374.5 eV were ascribed to the metallic Ag⁰, Ag₂S y AgCl and the peaks at 367.9 eV and 373.9 eV were indicative of Ag⁺ derived from Ag₂SO₄. No peak of O 1s owing to Ag₂O was detected. The existence of both species of sulfur in the sample can be confirmed by the presence of S²⁻ and SO4²⁻ peaks at 161.69 eV and 168.52 eV, respectively, after fitting of the S2p spectra depicted in Figure 4, central column.

S2p spectra showed the presence of multicomponent peaks, consisting of multiple doublets S2p3/2-Sp2p1/2 (binding energy difference constrained to 1.2 eV together with the intensity ratio 2:1). The presence of three doublets due to sulfides (S2p3/2 at 161.55 eV), polysulfides (S2p3/2 at 162.3 eV) and sulfates (S2p3/2 at 168.4 eV) can also be observed [16].

After the first 10 min of sputtering, the C and O due to adventitious carbon was lost, and the amount of Cl and S increased, indicating that they are in a layer below the carbon layer and they are associated to Ag. The next 10 min of sputtering had a lesser effect, with a similar result. The last 10 min of sputtering at a higher energy (5000 eV) almost completely destroyed the corrosion layer, only showing the presence of Ag⁰.

In order to determine the compositional depth profiling of silver, chloride, and sulfur, the atomic percentage of the different species was calculated and is shown in

Table 8. Atomic percentage of the different species of silver, chlorides, and sulfur after sputtering.

	% at Component
Sample	Ag(0)/Ag2S/AgCl	Ag+	Cl−	S2−	Sn−	SO42−
Fresh	33.89	5.35	5.4	5.62	0.9	1.88
10 min AIB (3000 eV)	57.4	5.32	7.3	7.06	1.16	1.85
20 min AIB (+10 min 3000 eV)	63.95	4.91	5.5	8.57	1.4	1.13
30 min AIB (+10 min 5000 eV)	96.3	0	nd	nd	nd	nd

. At the beginning of the sputtering, an increase in the concentration of all species was noticed as an effect of the removal of the outermost carbon layer. The corrosion products present included AgCl, Ag₂S, silver polysulfides, and Ag₂SO₄. After 20 min of sputtering, a small decrease in sulfates was recorded, as can be seen in the atomic percentage of the species Ag⁺ and SO₄²⁻. A decrease in AgCl was also observed. The increase in Ag(0)/Ag2S/AgCl signal with sputtering time was likely due to the contribution to the signal response of metallic Ag. No signal of corrosion products was recorded after 30 min of sputtering.

4. Discussion

From the results obtained, two questions arise: which factors have influenced the tarnishing of silver, and what is the origin of the contaminants that caused it.

RH remains consistently below 50%. Within the narrow ranges observed in the MAN galleries and display cases, relative humidity does not appear to influence tarnishing. In fact, the coupons exposed in the gallery (MAN 5), where RH remained below 35% for nearly half of the exposure time, are the second most sulfidized.

By contrast, pollutant concentrations are high. Comparing the results in Table 1 shows that the type 1 (airtight) display cases have similar contaminant levels to one another, while the values for the type 2 case, MAN 4, coincide with those of the gallery (MAN 5). We therefore have two different situations: airtight cases, where the origin of the pollutants is internal (MAN 1, MAN 2, and MAN 3), and a low-airtightness case (MAN 4) that shares its environment with the gallery where it is located (MAN 5) and in which the presence of pollutants can be attributed mainly to external sources.

Despite these clearly differentiated environments, the behavior of the coupons does not follow this division. The results show low levels of tarnishing in two of the cases (MAN 2, MAN 3), slight in one (MAN 4), and high in another case (MAN 1) and in the coupons exposed in the gallery (MAN 5). From a conservation standpoint, we could consider three different outcomes, which do not match the classification proposed by ISO 11844, according to which corrosion would be “low” for location MAN 5 and “very low” for the rest (Table 6).

Interpreting the tarnishing results from the environmental data is not straightforward. Despite the similarity of the environmental conditions measured, the corrosion suffered by the samples inside the non-airtight display case (MAN 4) was less than half of that observed in the gallery. Of the airtight display cases, two have very low sulfidation (MAN 2 and MAN 3), and one has high sulfidation (MAN 1). Because the passive samplers were saturated (Table 1), it is not possible to compare the actual sulfide concentration between them; strictly speaking, we cannot tell whether significant differences exist that would justify this result. What is indisputable, however, is that in all three cases the values are considered “very high” [17], and it would be reasonable to expect silver to be affected in all of them. It is possible, however, to compare with MAN4 and MAN5, which despite having at least half the sulfides, show much greater corrosion than MAN2 and MAN3. One possible explanation is that MAN 2 and MAN 3 contain a large number of silver objects, so the sulfides present are distributed over a relatively large surface area, whereas in the remaining cases the attack is concentrated on a small amount of silver.

For the differences observed between the MAN 4 case and the MAN 5 gallery, the relationship is not as direct. The MAN 4 case contains silver (and copper) objects, but not in quantities as large as in cases 2 and 3, so although this may be an influencing factor, it is not as evident as in the previous situation. Nor is RH, which is in fact slightly higher overall in the case than in the gallery. One hypothesis is that particle deposition—not measured in this work, but in principle expected to be higher—favored corrosion on the gallery coupons; this is an aspect to be assessed in future studies.

Another difference between the two groups of coupons, according to their environments, is the composition of the corrosion layer, as seen in the different profiles of the reduction curves. In the coupons located in the type 2 case (low airtightness) and in the gallery—MAN 4 and MAN 5, respectively—the corrosion layer consists of silver sulfide with small amounts of silver chloride and silver sulfate (identified by XPS in MAN 5). In the coupons located inside the type 1 cases (high airtightness), the corrosion layer is composed solely of silver sulfide, which can be related to the high H₂S concentration detected inside these cases. In this situation, the airtightness clearly favors the build-up of contaminants originating from internal sources; formic acid—typically emitted by materials present inside cases—is also higher than in the gallery and the MAN 4 case. We therefore find a source of sulfides both inside and outside the cases, and a source of chlorine outside.

Regarding the silver sulfate detected by XPS, its origin is unclear; some previous studies have reported this product [4,18,19], mainly in outdoor environments and associated with elevated SO₂ concentrations. In our case, the measured SO₂ levels do not justify the formation of this product, which moreover does not appear in the case where its concentration is highest, as seen from the reduction profiles. In any case, the amount of AgSO₄ is of little relevance.

Externally, sources of sulfides are related to construction materials, furniture, and museum visitors [20]. In this area, there is scarcely any furniture beyond the cases, and the floors are marble, so the most important sulfide source is the presence of visitors. Inside the cases, the origin will be related to the case materials and/or interior supports, since no other organic materials are present in the cases studied. These materials include foamed PVC and some polyurethane and silicone adhesives and sealants, which could, a priori, emit organic acids, but not sulfides. Although the use of plasticized PVC board is widespread in museums and is theoretically a stable material, Samide and Smith [21], found that, in high-quality PVC-U boards that had passed the Oddy test, 2-ethylhexyl thioglycolate (2-EHTG)—a sulfur-containing by-product of a thermal stabilizer—can be released. The presence of additives is an issue that requires attention because it greatly complicates the selection of materials for museums; efforts to use stable, high-quality materials can be undermined by unknown additives that vary by manufacturer or can be changed without notice. Indeed, the American Institute for Conservation’s Oddy Test results database [22] includes some boards of this type that did not pass the test, and for one of the most commonly used, Forex^®, there are studies suggesting the possibility of corrosive emissions [23,24]. At the MAN, analyses of the materials used in display case construction did not detect this additive, although another sulfur-containing compound—ethoxycarbonyl isothiocyanate—was identified among the VOCs that may be released from Simopor ^® [25]. This does not exclude other types of emissions, since only VOCs—mainly those related to human health—were analyzed in this study, and other emissions such as H₂S, COS, HCOOH, CH₃COOH, which are those affecting metals, were not considered. To confirm or discard the possible responsibility of Simpor^® in the sulfidation of silver, an Oddy Test was carried out (Figure A1, Appendix A). Results of the test were negative, so it cannot be stated that this material is responsible for the sulfidation of silver. Further studies are needed to identify the source of sulfur in museum cases.

Finally, it is interesting to note that silver chlorides have been identified in some samples, given that HCl measurements (see Table 1) did not detect this contaminant at any of the measured locations. The origin of chloride-containing corrosion products in indoor atmospheres has typically been attributed to HCl emission from the thermal degradation of PVC, which, as noted, was used inside the cases. However, a recent study by Pok et al. has shown that HCl emission from PVC under museum conditions is too low to contribute significantly to environmental contamination and, consequently, to the deterioration of cultural heritage [26]. The non-detection of this pollutant—and, above all, the fact that chlorides are found mainly on the coupons located in the gallery and, to a lesser extent, on those in the low-airtightness case—indicates an external source. The main hypothesis is cleaning products used in the galleries or nearby areas of the museum, which prompted a study of the possible effects of cleaning products on silver [27]. The results of that study indicate that the use or handling of chlorinated cleaning products in washrooms and other nearby areas may be responsible.

5. Conclusions

The tarnishing of silver artifacts in museum collections is a complex problem dependent on numerous factors. The results of the one-year study of MAN display cases and a gallery show significant variability in the corrosivity of silver across different cases, despite the similarity of the materials used. In addition to sulfides, our work also identified chlorides, probably originating from products used in museum cleaning activities, as well as sulfates.

The high affinity of sulfide for silver and the strong visual impact it produces, even in very thin layers, makes it necessary to exercise great caution with the materials used in the construction of display cases and in mounting objects. However, identifying the sources of sulfur is not always easy, and very small emissions can become significant if concentrated by airtight display cases.

Our results show that sulfide concentration is not the only determinant of sulfidation, with other aspects contributing to making corrosion more or less severe, and low humidity levels measured at all locations being sufficient to keep silver safe from sulfide tarnishing. Therefore, it is necessary to continue studying additional factors and analyze them together to quantify their relative influence, with a view to designing preventive conservation strategies adapted to different situations.