Under the surface: silvering in Adrian Athenaeum

Introduction

A commercial excavation in Piazza Madonna di Loreto, Rome, uncovered the Adrian Athenaeum. The Athenaeum itself, almost completely obliterated by the latter uses of the space, was the extension of the Trajan Forum commissioned by the emperor Hadrian. Its initial use as a school room and its rich marble decorations were followed, during the sixth century, by a transformation of the space into a metallurgical workshop, where part of the materials present in the Athenaeum was reused for production purposes. Several furnaces with different shapes and different functions were set in place (Figs. 1 and 2) between the former rows of marble benches.

OPEN IN VIEWER

Figure 1

Reconstruction of transformation of Athenaeum into metallurgical workshop with systematic reuse of decorative material

OPEN IN VIEWER

Figure 2

Image of excavated area of workshop showing evidence of furnaces (their density indicates intense level of activity at site)

A small portion of the material recovered at the site was selected for further archaeometric studies, and at least nine metallurgical processes were identified.1 This paper aims to investigate the silvering processes related to copper and copper alloy artefacts, how these processes are visible in the artefacts, and to understand which analytical method is more suitable for obtaining the appropriate information for interpreting the material. The level of corrosion of the artefacts and the discontinuous layer of silver makes it hard to gather information on their surface, and hence, a series of analytical techniques were used to investigate the problem and propose a technological approach that could be used when cutting a cross-section is impossible.

Experimental

Of the 39 samples analysed during the initial archaeometallurgical characterisation, five (Fig. 3) were identified as showing indications of silvering processes. These are two samples of metal debris (samples 13 and 16), one silvered metal fragment (sample 14) and two silvered metal objects.

OPEN IN VIEWER

Figure 3

a macroscopic image of sample 13, b macroscopic image of sample 16, c macroscopic image of sample 14 and d macroscopic image of ingot and cross-section of planchet

The metal fragments (samples 13, 14 and 16) seem to be linked to the same process: the silvering of leaded bronze; on the other hand, the silvered metal objects (one planchet and one ingot) are related to the silvering of copper, and present different characteristics, possibly indicating two distinct production processes.2

The samples were analysed using a field emission gun scanning electron microscope (SEM) to investigate the corrosion layer and the distribution of silver and chlorine. The planchet was then chosen to investigate the potential use of time of flight secondary ion mass spectrometry (TOF-SIMS) and energy dispersive X-ray fluorescence (XRF) for routine analyses on silvered objects, as well as to determine the possible presence of mercury on the surface and the consequent identification of the use of mercury paste for the silvering process.

The samples could be cut and mounted in resin for the analyses with the SEM (Supra Zeiss). The mounted section was polished with carbide paper down to 8 μm and then further polished with diamond paste down to a quarter of a micrometre. The polished sections were then coated with a thin gold layer to increase their conductivity and improve the imaging and chemical analysis.

The TOF-SIMS is a very sensitive surface technique, whose range of action is limited to the first few nanometres of the surface. Considering the level of corrosion of the artefacts under consideration in this paper, this technique could only provide information on the top surface of the corrosion layer. In this case, the technique was used to investigate the cross-section and, in particular, the border between the corrosion layer and the actual surface of the artefacts, where the porosity observed with the SEM is located and where the ‘islands’ of silver were detected. The technique was used to inspect this boundary in order to verify the presence of a very thin continuous layer of silver that may have not been visible using the SEM, as well as the possible presence of mercury.

In order to perform the TOF-SIMS analysis, a 2 mm slice of the planchet was cut from the residual fraction after the mounting of the sample in resin; this section was freshly polished with carbon paste and then cleaned with acetone in order to reduce the level of contamination of the surface to be analysed. The instrument used was a Kore Technology Ltd time of flight secondary ion mass spectrometer, and the instrumental conditions used for the analyses were as follows: 25 keV indium primary ion source (FEI liquid metal ion gun) operating at 1 μA current. Secondary ions were analysed in a reflectron mass spectrometer and detected with a dual microchannel plate assembly. Flight times were recorded with a 0·5 ns time to digital converter. Spectra were taken from an area of approximately 150 μm². Calibration of the mass spectra was established by defining a common series of CxHy peaks with a known mass using the mass calibration function in the instrument's software.

On the other hand, the XRF analyses were performed on the surface of the planchet. Different energies of the gun were used to investigate if the different degree of penetration of the X-ray could obtain information on the subsurface layer, the one that was observed with the SEM to contain the silver. The analyses were performed on the surface of the sample without further treatment, and a ZnTi pellet standard was used to cross-check the data. The ZnTi is a National Institute of Standards and Technology standard used to check the position of the peaks in the XRF; the standard is pure and does not contain any silver. The XRF instrument used (Oxford ED 2000) has a silver source. A thick aluminium filter is available in the instrument to avoid the detection of silver radiation coming from the tube. At high energy, the function of the filter is limited and part of the radiation reaches the detector, invalidating the detection of silver from the sample since part of the signal comes, in effect, from the tube itself. A series of analyses were performed on the ZnTi to investigate the energy at which only the signal from the sample is visible (no peaks of silver were detected from the standard).

Results

The results obtained from the investigation of the corrosion layer using SEM indicate that:

OPEN IN VIEWER

Figure 4

Polyhedral leaded bronze (sample 16) with areas of coalescing lead (on left close to surface in white) and deep corrosion layer (in darker grey on left)

OPEN IN VIEWER

Figure 5

Image of surface of copper ingot showing high level of porosity and small inclusions attesting presence of silver (in white)

OPEN IN VIEWER

Figure 6

Image of surface of planchet showing clear-cut fracture parallel to surface indicating presence of chlorine and inclusions attesting presence of silver (in white)

OPEN IN VIEWER

Figure 7

Detail of Fig. 4 showing area of association of silver and chlorine on surface

These observations suggested that if the silver was not looked for, it could ‘escape’ the analyses and mislead the interpretation to bronze or copper artefacts. To investigate this hypothesis, further analyses of the planchet were carried out using TOF-SIMS and XRF.

An initial mass spectrum was measured in the two areas analysed by TOF-SIMS, showing the main copper composition of the planchet (just below 63 and at 65; Figs. 8, 12 and 13), a small peak for silver (at 107; Fig. 9) and mainly the absence of mercury (the main peaks of which should be at 200, 198 and 202; Figs. 10 and 11).

OPEN IN VIEWER

Figure 8

Portion of TOF-SIMS spectrum related to copper

OPEN IN VIEWER

Figure 9

Portion of TOF-SIMS spectrum related to silver (at 107)

OPEN IN VIEWER

Figure 10

Portion of TOF-SIMS spectrum related to mercury (at 200, 198 and 202)

OPEN IN VIEWER

Figure 11

Detail of spectrum in Fig. 10 showing absence of peaks between 196 and 204, indicating absence of mercury in sample

OPEN IN VIEWER

Figure 12

Total map showing corrosion products (upper dark grey layer) and body of object (copper: grey layer)

OPEN IN VIEWER

Figure 13

Map of copper showing strongest intensity in body of artefact, as expected

Subsequently, chemical maps of the two areas were measured in order to attempt a description of the distribution of silver in the sample. The map (Fig. 14) shows that silver is detected from under the layer of corrosion towards the body of the sample.

OPEN IN VIEWER

Figure 14

Map of silver showing barely visible line at border between corrosion products and body of object

No clear silver layer is detected on the surface of the sample; instead, it looks as if the silver diffused towards the body of the sample (Fig. 14). This observation, in correlation with the observations made using the SEM, seems to indicate that the silver is concentrated on the surface of the object (under the corrosion layer). Larger concentrations can be observed in the aggregates detected by the SEM. The diffusion towards the centre of the object seems to be related to the smearing of the silver through polishing on the most superficial layer (the one visible with TOF-SIMS).

The measurements with XRF were carried out at 50, 45, 40, 35, 30 and 27 kV, using a thick aluminium filter on the zinc/titanium standard. At 50 kV, the intensity of the silver peak in the standard is ∼15 cps subtracting the background (Fig. 15); decreasing the energy of the beam, the intensity of the peak of silver decreases down to 2 cps at 27 kV (Fig. 16). The planchet was analysed through the corrosion layer at 50 and 27 kV (Figs. 17 and 18). At 50 kV, an intensity of 4 cps was detected for the silver K_α line (the same line observed in the zinc/titanium standard and described above). At 27 kV, there was no visible silver K_α line and the signal was completely lost. The data deriving from the zinc/titanium standard, however, lead to the conclusion that the silver visible at 50 kV is coming from the X-ray source, and the presence of silver in the sample could not be detected.

OPEN IN VIEWER

Figure 15

K_α and K_β lines of silver measured in zinc/titanium standard using 50 kV beam and thick aluminium filter

OPEN IN VIEWER

Figure 16

Portion of spectrum of zinc/titanium measured at 27 kV showing small K_α peak of silver

OPEN IN VIEWER

Figure 17

Complete XRF spectrum of planchet measured at 50 kV with thick aluminium filter (K_α and K_β lines of silver are visible at 22 and 25 keV)

OPEN IN VIEWER

Figure 18

Detail of spectrum of planchet collected at 27 kV that shows no sign of silver

The area around 10 keV of the spectrum obtained at 50 kV was investigated to ascertain the presence of mercury. No peaks were visible in the area (Fig. 19), confirming the data deriving from the TOF-SIMS and the SEM regarding the absence of this element.

OPEN IN VIEWER

Figure 19

Detail of XRF spectrum of planchet collected at 50 kV showing no signs of mercury (at 10 keV)

Discussion and conclusion

The exploration of the materials and the investigation of the techniques reported above allow us to conclude the following.

As mentioned in the section on ‘Introduction’, the fragments presenting surface silvering are of two types: leaded bronze and copper. In all the objects, chlorine was present on the surface in high concentrations, indicating that this element was involved in the process of silvering. Vlachou et al.3 advance the hypothesis of the use of mercury paste in association with chlorides to treat the surface. The analyses of the planchet indicate that this artefact was treated with an alternative process (since mercury was not detected), suggesting the hypothesis of a treatment with hot cerargyrite baths.2, 4. It can be hypothesised that all the artefacts at the site received a similar treatment, but only further analyses of the leaded bronze fragments similar to the ones carried out for the planchet could provide conclusive answers. The bronze alloy presenting silvering treatment analysed here matches very well with the alloy presented by Vlachou et al.3 as the ‘best’ alloy for silvering with mercury paste. Nevertheless, the observations of the surface of the leaded bronze fragments do not present a continuous layer of silver, but a scattered rare presence of silver aggregates of small dimensions, even though Vlachou et al.3 show a different scenario in the case study they present. In any case, as already stated above, the data on the leaded bronze fragments are not yet conclusive, and further work is needed to reach final conclusions.

The main aim of this paper is anyway not to highlight the processes of silvering performed at Piazza Madonna di Loreto but to investigate the possibility of loss of data regarding the presence of silver in case of surface analyses through routine laboratory techniques. The possibility that silver is not present in a continuous layer, as in this case, and that it is not present in the body of the metal, but only in discrete particles scattered under the corrosion layer, decreases, first of all, its overall concentration in the object and makes its detection harder. In these conditions, already in cross-section, only a thorough investigation of the boundary between the corrosion layer and the body of the artefacts allowed the detection of a couple of silver aggregates per sample using SEM. A much more sensitive technique like the TOF-SIMS detects a more constant layer of silver (admittedly not even in a convincing layer shape), but again, this observation can only be made on a cross-section and would be impossible as a surface analysis since the TOF-SIMS examines only the first couple of nanometres of the surface.

The analysis of the surface carried out by XRF demonstrates the possibility of ‘missing’ the silver altogether. XRF (in its portable or laboratory form) is the principal technique that would allow routine non-destructive in situ or laboratory analysis. In this case, if the Sopraintendenza ai Beni Culturali di Roma would not have given permission for the sectioning of the planchet, no signs of the silvering process could have been observed and an important part of the archaeometallurgical information would have been lost.

The question that the author wants to conclude with is the following: how many of these types of data have been lost? How many copper and bronze fragments were actually silvered?

Careful planning and rethinking of the methodological approach to archaeological materials are a necessity that has to be continuously mediated with the availability of techniques, the possibility of invasive analysis and the possibility to move the materials from their environments. More importantly, the limitations of the data obtained by our instruments need always to be carefully taken into consideration and cross-checked.