Study of corrosion in archaeological gilded irons by Raman imaging and a coupled scanning electron microscope–Raman system

Abstract

In this work, analytical and chemical imaging tools have been applied to the study of a gilded spur found in the medieval necropolis of Erenozar (Bizkaia, Spain). As a first step, a lot of portable equipment has been used to study the object in a non-invasive way. The hand-held energy-dispersive X-ray fluorescence equipment allowed us to characterize the artefact as a rare example of an iron matrix item decorated by means of a fire gilding technique. On the other hand, the use of a portable Raman system helped us to detect the main degradation compounds affecting the spur. Afterwards, further information was acquired in the laboratory by analysing detached fragments. The molecular images obtained using confocal Raman microscopy permitted us to characterize the stratigraphic succession of iron corrosions. Furthermore, the combined use of this technique with a scanning electron microscope (SEM) was achieved owing to the use of a structural and chemical analyser interface. In this way, the molecular characterization, enhanced by the magnification feature of the SEM, allowed us to identify several micrometric degradation compounds. Finally, the effectiveness of one of the most used desalination baths (NaOH) was evaluated by comparing its effects with those provided by a reference bath (MilliQ). The comparison proved that basic treatment avoided any side effects on the spur decorated by fire gilding, compensating for the lack of bibliographic documentation in this field.

This article is part of the themed issue ‘Raman spectroscopy in art and archaeology’.

1. Introduction

In recent years, technological progress has allowed the development of new analytical tools that enhance the effectiveness and the application field of scientific research. In the field of cultural heritage conservation, the practical experience of restorers is increasingly supported by the information provided by analytical studies, which allows restorers to identify the characteristics of objects [1], to describe the degradation problems affecting them [2], and above all to quantify the magnitude of any damage [3].

In this sense, one of the most useful improvements in recent decades consists in the development and the widespread use of portable analytical instruments [4,5]. In fact, nowadays, many instruments are applied to the in situ study of archaeometallurgic heritage. For example, metallic objects need to be conserved in a controlled environment with constant values of temperature and humidity because they are extremely sensitive to changes in climatic conditions. Transport of such objects to research laboratories would result in a significant disruption of this environmental balance that can involve the reactivation of degradation phenomena [6,7]. For this reason, the development of portable instrumentation allows in situ analysis to be carried out, avoiding any physical, chemical or mechanical stresses that are related to the transportation of the items.

For elemental analysis, a key role is played by portable and hand-held energy-dispersive X-ray fluorescence (EDXRF) equipment [8,9], which enables non-destructive analysis to be carried out, providing information about the atomic composition of the items. Depending on the requirements, EDXRF analysis can also be supplemented or replaced by other multi-elemental techniques, such as laser-induced breakdown spectroscopy (LIBS) [10,11] and particle-induced X-ray emission (PIXE) [12].

On the other hand, Raman spectroscopy, owing to its sensitivity, versatility, portability and non-destructive features, is becoming the reference analytical technique for molecular characterization of cultural heritage items [13–15]. Depending on the requirements, Raman spectroscopy can also be replaced or accompanied by additional molecular instrumentation, such as Fourier transform infrared spectroscopy (FTIR) [16,17] and fibre-optic reflectance spectroscopy (FORS) [18,19]. Even if the in situ analysis provides essential information to identify the degradation processes affecting the items, sometimes these results need to be supported by laboratory analysis in order to understand, for example, their stratigraphic and surface distribution.

To understand the distribution in depth of all corrosion products covering archaeological irons is extremely important because each of them has a different influence on the conservation of the object; corrosion phases such as magnetite and goethite create compact and stable layers that help to preserve iron-based artefacts [20], whereas lepidocrocite and akaganeite are highly reactive compounds that accelerate their degradation [6,21]. For this reason, it is important to identify the distribution of the most dangerous degradation products in order to have a deeper knowledge about the real conservation state of the items.

For this reason, several research studies are now focused on the interpretation of both the molecular and elemental maps of sample cross-sections.

Regarding elemental images, the scanning electron microscope (SEM) is extensively used for cultural heritage research purposes [22,23]. Its magnification feature and its ability to identify most of the elements helps researchers to thoroughly understand the composition of the alloys used for the specimen as well as to understand the presence of layers or heterogeneities.

Regarding molecular mapping, most of the modern works are based on the use of Raman spectroscopy [24]. Among all mapping methods developed for this technique, the ‘point by point’ one can be considered the most common. It works by focusing the laser on a sample spot. Then, each point of the selected region is sequentially analysed by means of a motorized stage that moves the sample under the laser. After analysing all the points in the region selected by the operator one by one, dedicated software is used to study the distribution of each detected compound along the point grid in order to create molecular distribution images.

In recent years, the acquisition of molecular maps has been facilitated by the development of Raman imaging methods [25]. This analysis offers several advantages compared with the point-by-point method, such as the considerable reduction in time and the possibility to use higher laser powers without damaging the sample.

One of the last frontiers of scientific research is the combination of Raman and SEM techniques by means of an interface that allows the molecular analysis of samples to be carried out inside the vacuum chamber of the SEM system [26]. Because of the optical system connecting both instruments, it is possible to take advantage of the magnification feature of SEM systems to collect the molecular vibration spectra of compounds that cannot be detected through the use of conventional techniques.

Fortunately, the role of analytical chemistry is not limited to the description of metallic alloys and their degradation compounds. In fact, it must be considered that each archaeological object has specific degradation problems and a customized restoration protocol has to be designed to properly conserve it. For this reason, a further contribution of analytical research is to devise experiments that help to predict the consequences of applying restoration treatments to a specific object.

Several scientific articles have focused on the study of the effect of desalination baths applied to iron-based archaeological items. Based on the literature, techniques such as ion chromatography (IC) are often used to study the consequences of desalination treatments on artefacts with conservation problems related to the infiltration of soluble salts [27].

In summary, from the discovery of an archaeological object until its cataloguing in a museum, this research follows the conservation work of a unique iron artefact by using some of the most innovative analytical tools. In this paper, we present the multi-analytical study of a gilded spur from the twelfth to the fifteenth centuries by applying bench-top (Raman imaging and SEM-EDS-SCA) and portable (Raman and XRF) instrumentation, before and after restoration works. The benefits of the interdisciplinary collaboration between both fields are also highlighted.

2. Experimental section

(a). Artefact description

The archaeological excavations that have been carried out since 2009 at the top of the Ereñozar Mountain (Bizkaia, Spain) have proved the presence of remains belonging to the Middle Ages. Among them was the discovery of a necropolis with over 70 burials, dated between the thirteenth and the sixteenth centuries. The excavations done in this archaeological site enabled the discovery of several metallic artefacts, such as spurs, ring and buckles.

Among all of the artefacts, a gilded spur warranted specific study. This object (discovered within one of the previously mentioned burials and dated between the thirteenth and the fifteenth centuries) was fragmented and presented thick corrosion layers almost completely covering the gilded surface (electronic supplementary material, figure S1).

As will be explained below, the technology used to manufacture the item as well as the characteristics of the degradation processes that affected it required the use of a three-step analytical approach. The data provided by the analytical research, combined with the skilful conservation work applied by restorers, prevented further degradation of the spur and enabled its original aesthetic value to be determined. Currently, the artefact is preserved in perfect condition and represents one of the most important pieces in the Archaeological Museum of Bizkaia (electronic supplementary material, figure S1b).

(b). Instrumental set-up

(i). In situ analysis

The first step of the analytical study was devoted to the characterization of the spur alloys as well as any degradation within it. The analysis was carried out in situ in the museum by using portable and non-invasive techniques able to provide both elemental and molecular data.

A hand-held X-MET5100 EDXRF spectrometer (Oxford Instruments, UK) was mounted on a bench-top stand to analyse the alloys of the artefact's matrix and decoration. This instrument has a measurement spot size of 9 mm and uses an X-ray tube excitation source composed of a rhodium anode. Its high-resolution silicon drift detector (SDD) allows energy differences of the order of 200 eV to be distinguished and enables the detection of light elements such as Mg, Al and Si. For the semi-quantitative analysis, a 50 s measurement time and an analysis method specific for metal alloys were set in the integrated personal digital assistant.

For the study of the corrosion phases constituting the thick degradation layer covering the spur surface, an InnoRamTM 785S Raman system (B&WTek, US) coupled to a 20× magnification objective was used. This instrument, equipped with a 785 nm excitation laser reaching a maximum power of 300 mW, works in a range between 175 and 3200 cm⁻¹ with a mean spectral resolution of 4.5 cm⁻¹. The electronic control of the laser allowed us to set energies below 30% of its total power in order to avoid thermal transformations of the materials. Acquisition of the spectra was performed with the BWSpecTM 3.26 software using a number of accumulations from 5 to 40 and integration times between 0.5 and 10 s depending on the signal-to-noise ratio.

(ii). Laboratory analysis

The overall characterization by the in situ measurements was complemented by the laboratory analysis of four rust samples naturally detached from the spur. These analyses were mainly centred on the acquisition of elemental and molecular images by SEM-EDS and Raman imaging, respectively.

The surfaces of some fragments were analysed in order to better understand the characteristics of the decorative layer (M01) as well as the surface degradations (M02). Also, M03 and M04 fragments were prepared for cross-section analysis with an acrylic resin (following the methodology described by Jiménez et al. [28]) and polished to highlight the stratigraphic distribution of the corrosion phases.

For the elemental analysis, measurements were performed using an EVO40 SEM (Carl Zeiss STS, Germany) coupled to an X-Max energy-dispersive X-ray spectrometer (Oxford Instruments). EDS analysis was carried out using an I Probe at 180 pA, an acceleration potential of 30 kV and the number of scans between 6 and 10.

Molecular analysis was performed by means of an inVia Renishaw confocal micro-Raman spectrometer (Renishaw, UK) coupled to a 50–100× magnification microscope (DMLM Leica). The 785 nm excitation laser was set at energies below 10% of its total power in order to avoid thermal photodecomposition. Molecular images were generated characterizing the main peaks of each spectrum that was sequentially acquired from selected sample areas. The measurement parameters were set at 10 s and one accumulation in a spectral range of 150–1200 cm⁻¹. Raman imaging enabled us to characterize the stratigraphic distribution of the molecular phases in a reduced time compared with the Raman point-by-point mapping.

Finally, the confocal micro-Raman spectrometer and scanning electron microscopy were combined by means of a structural and chemical analyser (SCA; Renishaw). The SCA interface, described by Aramendia and co-workers [26], allowed us to perform punctual Raman analysis of the degradation products present at the micro-scale inside the vacuum chamber of the SEM-EDS system.

(iii). Treatment assessment

The third step of this paper was focused on the study of the advantages and disadvantages related to the treatment of iron-gilded artefacts with basic desalination baths. There is a significant lack of bibliographic documentation regarding this issue. The conclusions were obtained by comparing the effects of the most used treatment in the restoration of archaeological irons (NaOH bath 0.5 M) with a neutral pH treatment (MilliQ water) reference.

The experiment was carried out by treating samples M01 and M03 with MilliQ and NaOH baths, respectively. Both the NaOH treatment and MilliQ reference were applied with a constant temperature of 25°C for 60 days. During and after treatment, the desalination efficiency was evaluated by quantifying the chlorides extracted from the samples and dissolved in the baths. The analysis was performed using an Agilent Capillary Electrophoresis System (Agilent CE, Agilent Technologies, Spain). Agilent G1600-60211 fused silica capillaries coated with polyimide were used: 50 µm internal diameter × 375 µm external diameter and a 40 cm effective length. The calibration was performed by injecting 10 standards at 400 mbar from a 500 µg l⁻¹ mixed anion standard solution in water. Data acquisition was performed by HP Chemstation software (Agilent CE, Agilent Technologies, Spain).

The characterization of possible leached Fe, Ag, Au and Hg elements in both solutions was carried out in order to control any damage potentially produced by the desalination process in the gilded layer. The quantitative analyses of ⁵⁶Fe, ¹⁰⁷Ag, ¹⁰⁹Ag, ¹⁹⁷Au and ²⁰²Hg isotopes were performed by inductively coupled plasma mass spectrometry (ICP-MS; NexION 300; Perkin Elmer) under the following experimental conditions: nebulizer flow of 0.9–1.0 l min⁻¹, plasma flow of 18 l min⁻¹ and radio frequency power of 1400 W. Taking into account that the solutions analysed by ICP-MS must contain an acid concentration below 1% HNO₃, the samples were first subjected to a dilution process.

Finally, considering that chloride extraction may cause the transformation of Cl-containing phases into more stable compounds, an inVia Renishaw Raman spectrometer was used for the molecular characterization of several selected points of the samples before, during and after treatments.

3. Results

(a). In situ analysis

The analytical study started with the identification of the metallic alloys used for the manufacture of the gilded spur. Considering the fragility characterizing the spur fragments after excavation, the in situ elemental analysis was achieved by mounting the hand-held EDXRF on a bench-top stand, which ensured contact between the instrument and the spur, thus avoiding any pressure or mechanical stress. For the characterization, 60 points of the artefact were analysed, with three replicates for each point, ensuring reliable results. According to the analysis, iron was the only element detected on the spur core, whereas the ends also showed small quantities of lead and copper. However, considering the limit of detection of the portable instrument, the presence of elements with an atomic number of less than 12 (Mg) cannot be excluded.

The gilded layer was composed of a mixture of gold, silver and mercury. The presence of Hg suggested the use of mercury gilding as the decoration technique. Also called fire gilded, this method is based on the use of an Au–Hg alloy that, once applied over the item's surface, is heated until the mercury evaporates. As shown in the literature [29], this method presupposed a higher technological level than the more usual techniques based on the use of gold leaf and gold foil.

The elemental composition of the spur is really interesting considering the lack of examples of fire gilding applied on iron matrix items. Furthermore, compared with other documented cases, this spur can be considered extremely rare because its decoration stands out because of the presence of high percentages of Ag [30,31]. The presence of silver is unusual in fire gilding because this method permits production costs to be reduced without resorting to the mix of gold with other metals.

Switching to the molecular analysis, more than 80 Raman spectra were collected, allowing the identification of four iron oxide phases. The blackish corrosion spots were composed of magnetite (Fe₃O₄, Raman peaks at 542 and 669 cm⁻¹), a stable iron oxide whose formation is favoured by anoxic environments [32]. In the orange-brownish corrosion layers, three iron oxyhydroxides where found: goethite, lepidocrocite and akaganeite. Goethite (α-FeOOH, Raman peaks at 247, 301, 388, 482, 552 and 546 cm⁻¹) is a stable compound forming a protective layer on the spur core. On the other hand, lepidocrocite (γ-FeOOH, Raman peaks at 250, 379, 525 and 650 cm⁻¹) and akaganeite (FeO_0.883(OH)_1.167Cl_0.167, Raman peaks at 311, 390, 537 and 722 cm⁻¹) are reactive phases that can be considered as degradation accelerators [6,21]. In fact, akaganeite is the most dangerous one because the presence of chloride ions in its molecular structure increases the porosity of the corrosion layer, facilitating cracks and swellings [33,34].

(b). Laboratory analysis

(i). Surface analysis

As explained above, laboratory analyses were carried out on M01 and M02 fragments to better understand the characteristic of the decorative layer and the surface degradation.

SEM-EDS maps of sample M01 allowed us to study in depth the nature of the gilded layer. The elemental image (figure 1c) obtained by use of INCA software showed a degraded silver layer. A possible reason for this degradation was found in the spatial distribution of elemental Cl that perfectly overlaps the Ag distribution, suggesting the presence of chlorinated corrosion phases. Considering that silver degradation compounds are not easily detectable by means of regular Raman spectrometers, definitive evidence was provided by the use of the coupled SEM–Raman system. The spectra obtained by the analysis of several selected points of the silver surface showed the characteristic peaks of silver chloride (97, 143 and 233 cm⁻¹; figure 1b) [35], a degradation compound often found in Ag-containing artefacts from archaeological sites near the coast.

Figure 1.

Optical image (a) of sample M01. The silver degradation product (b) was detected on the gilded layer surface through the use of the coupled SEM–Raman system. Raman analysis spots were selected by taking advantage of the magnification feature of the scanning electron microscope (SEM) (c). (Online version in colour.)

During the analysis of sample M02, both optical and SEM images acquired from the outer surface of this fragment (which was in contact with the soil during burial) showed the presence of white filamentous incrustations. The elemental maps (figure 2a) proved that these filaments were mainly composed of calcium. Raman analysis carried out inside the SEM chamber allowed us to identify the molecular composition of these incrustations (figure 2b) as calcium carbonate (CaCO₃, main Raman peak at 1085 cm⁻¹). The particular structure of such degradation suggests a calcification process promoted by biological activity. In fact, manifold varieties of organisms able to mineralize CO₂ to carbonate compounds such as CaCO₃ are usually present in soils [36].

Figure 2.

SEM image and element maps of sample M02, showing the Ca-based filamentous incrustations covering the iron corrosion surface (a). Molecular analysis carried out by means of the coupled SEM–Raman system proved the presence of calcium carbonate (b). (Online version in colour.)

(ii). Cross-section analysis

The instruments used for the characterization of the surfaces of samples M01 and M02 were also applied to study the cross-sections of samples M03 and M04.

M03 fragment was first analysed to better understand the gilded layer composition. As can be observed in figure 3, several elemental maps proved the presence of two decorative layers. The inner one was composed of pure silver and was overlapped by a gold-based coat applied by fire gilding. As described in several Middle Ages manuscripts, the accessories belonging to both squires and knights contained a rigorous symbolism: the squires could wear only silver-plated decorations because gilded items were for the exclusive use of the knights [37]. According to this information, it can be assumed that the spur was decorated in two phases. At first, the squire applied a silvered decoration, and, then, the spur was coated by a gilded layer after his promotion to knight.

Figure 3.

SEM images of a cross-section of sample M03 focusing on the decorative layers (a) and (d). (b) and (e) show the silver layer overlapped by a gold coat (c) and (f) applied by the fire gilding technique.

Coming back to the study of the distribution of the iron oxide phases, the elemental image (figure 4) obtained from the M03 SEM-EDS map showed the presence of two corrosion crusts (red colour identifies the Fe element) separated by the decorative layer (blue colour identifies the Au element). In the same image, the violet colour assigned to the Cl element demonstrates that Cl ions reached the inner corrosion layers by penetrating through the cracks and vacuums.

Figure 4.

SEM image of sample M03 proving the presence of chlorine infiltration problems. The stratigraphic distribution of goethite (G), lepidocrocite (L), akaganeite (A) and magnetite (M) in the selected areas was carried out by means of the Raman imaging system. (Online version in colour.)

The elemental characterization of the sample helped to select the most interesting areas for Raman molecular imaging. Although, in comparison with the in situ data, no new degradation compounds were detected, molecular images provided new important information regarding their stratigraphic distribution. In fact, as can be seen in figure 4, the outer corrosion crust on sample M03 was predominantly composed of goethite (blue image), while magnetite (red image) was the main compound of the inner layer.

This stratigraphic composition is quite common in Fe-based archaeological finds. In fact, iron oxides tend to be located in the inner corrosion layers (in this case, magnetite was probably generated in a second step as a result of the decrease of oxygen levels due to the formation of new outer rust layers). On the contrary, the corrosion surface, being more exposed to the external environment, tends to be mainly composed of iron oxyhydroxides [32].

As shown in figure 4, Raman imaging also helped to demonstrate that Cl-affected areas were characterized by the presence of reactive phases such as akaganeite (yellow image) and lepidocrocite (light blue image).

The analytical procedure applied to the study of sample M03 was repeated on other samples at our disposal. Both elemental and molecular data obtained from all samples highlighted a similar composition to sample M03. In fact, all cross-sections show an inner part almost entirely composed of magnetite and covered by external iron hydroxide layers. Furthermore, the correspondence between Cl-affected areas and the development of reactive iron oxyhydroxides phases (akaganeite and lepidocrocite) was also repeated.

(c). Treatment assessment

As mentioned above, the third part of this analytical study was focused on the study of the effects caused by NaOH desalination baths (pH ≈ 13.3) on fire gilded iron artefacts. Considering the gilded area preserved on it, sample M01 was selected for the application of the NaOH treatment. The results were then compared with those obtained by treating sample M03 with an ultrapure water bath (pH ≈ 6.8).

The main data obtained from both experiments are summarized in figure 5. Starting from the study of the dechlorination effect of the treatments, the overall amount of chloride extracted from both samples (figure 5a) described a similar curve that reflects the experimental results carried out by Guilminot et al. [27]. Most of the chlorides were extracted in the first week, then a progressive decrease in the efficiency of the treatment was detected.

Figure 5.

NaOH chlorine extraction on spur samples and its comparison with the MilliQ bath reference (a). SEM maps of the M01 fragment before and after NaOH treatment exclude any damage of the gilded layer (b). The study carried out by Raman analysis (c) shows no transformation of akaganeite after dechlorination. (Online version in colour.)

The most relevant conclusions obtained from the comparison of the treatments arises from the analysis performed by ICP-MS. The quantification of the metals dissolved by the NaOH solution during the M01 sample treatment showed that the dissolution of Ag, Au and Hg was below the quantification limit of the instrument. Considering that the sensitivity of the ICP-MS reaches a level of parts per billion, it can therefore be confirmed that this treatment did not involve any solubilization of gilding metals.

This result was corroborated by the monitoring study, as both SEM (figure 5b) and Raman (figure 5c) analysis carried out on the sample before, during and after the desalination bath excluded any change in the elemental and molecular composition of the sample, excluding a decrease in the Cl concentration.

Finally, it must be highlighted that the amount of Fe solubilized from the sample treated with NaOH was extremely low with respect to the MilliQ reference bath, providing further evidence in support of the use of alkaline treatments for iron artefact desalination.

4. Conclusion

The aim of the work carried out on the Ereño gilded spur was to provide the scientific community with a practical example of the potential offered by the application of Raman imaging and coupled SEM–Raman systems in the study of iron-based archaeometallurgical artefacts.

In the first step of this work, it was demonstrated that, by combining the use of molecular and elemental portable systems, it is possible to achieve an overall characterization of the item. In fact, the spur alloy was successfully characterized by in situ elemental analysis, while the portable molecular system effectively detected the main degradation product threatening its conservation.

However, data provided by the use of portable instrument need to be supported by laboratory analysis in order to better characterize the composition of the items and estimate the magnitude of their deterioration phenomena. In fact, owing to the elemental mapping of sample cross-sections by SEM-EDS, the presence of a double decorative layer was observed, providing more information about the history and the evolution of the spur. Furthermore, the molecular images obtained from both the M03 and M04 sample cross-sections using the Raman imaging technique allowed us to evidence the stratigraphic distribution of the iron corrosion phases detected by in situ analysis.

The most innovative part of this work is the use of the coupled SEM–Raman system, which enables the characterization of new degradation products. For example, Raman analysis of silver degradation crusts, carried out by taking advantage of the magnification feature of the SEM system, allowed us to identify the presence of silver chloride. Even if this compound is easy to find in Ag-containing items, its molecular characterization is hard to determine by means of conventional spectroscopic techniques. In this sense, this work presents the first documented case of AgCl Raman characterization on archaeological artefacts.

Finally, the treatment comparison described above showed how analytical chemistry can be seen as an indispensable tool for restorers, because it provides crucial information to choose the most appropriate conservation treatment. Crucially, the experiments carried out showed that the use of basic treatments (NaOH) caused no side effects on the iron artefacts decorated by fire gilding. At the same time, the comparison with the MilliQ bath reference proved that the use of alkaline baths also helps to inhibit the dissolution of the iron matrix and limits the reactivation of corrosive processes. Finally, the monitoring carried out before, during and after sample treatments showed that the NaOH bath does not cause any transformation of akaganeite into a more stable compound. However, owing to the elimination of uncomplexed chlorides from the artefact core, the chance of reactivation of this degradation process dramatically decreases.

Authors' contributions

M.V. collected field data, carried out the Raman spectroscopic analyses, participated in the design of the study and drafted the manuscript; I.C. collected field data; S.F.-O.V. collected field data; L.G. carried out the restoration works and helped with the interpretation of data; I.G. helped with the interpretation of data; K.C. coordinated the study and helped draft the manuscript; A.A. provided historical background and helped with the interpretation of data; and J.M.M. provided important contributions to interpret the acquired field data, participated in the design of the study and helped draft the manuscript. All authors gave final approval for publication.

Competing interests

The authors declare that there are no competing interests.

Funding

This project has been funded by the UFI ‘Global Change and Heritage’ project (ref. UFI 11–26 UPV-EHU), the DISILICA-1930 project (ref. BIA2014–59124) and the European Regional Development Fund (FEDER).