Introduction

The surface of an outdoor bronze sculpture deteriorates with time, developing a greenish/blackish patina mainly constituted of corrosion products. The formation of these corrosion products implies complex chemical, electrochemical, and physical processes that strongly depend on the constituents of the surrounding environment (Leygraf et al. 2000). In a simplified model, the corrosion of an outdoor bronze sculpture starts with the formation of a reddish-brown layer of cuprite (Cu2O), sometime mixed with cassiterite (SnO2), over the surface of the metal due to the interaction between the alloy components and the atmospheric oxygen. In urban environment, the presence of pollutant gases such as sulfur dioxide (SO2) leads to the formation of green basic copper hydroxysulfates (i.e., brochantite and antlerite) while in chloride-rich environment, basic hydroxychlorides (i.e., atacamite and paratacamite) usually prevail. The above-mentioned copper hydroxysalts are generally considered harmful for the bronze sculpture since they produce a modification of the original bronze surface and trigger irreversible mechanisms of alloy deterioration (Scott 2002).

In order to protect the surface of outdoor sculptures from corrosion, a series of organic coatings that act as a physical barrier against the aggressive atmosphere has been developed. The mostly used coatings belong to the families of acrylic resins and microcrystalline waxes. Usually, when wax is applied on top of a layer of acrylic resin (Scott 2002; Marabelli and Napolitano 1991), a durable protection can be achieved.

The identification of corrosion products and evaluation of the durability of coatings are critical issues among metal conservators-restorers. In fact, the proper characterization of the surface constituents of outdoor bronze sculptures helps in better understanding the degradation processes and planning the proper conservation interventions. Micro destructive techniques are well-established methods for the characterization of alteration patinas. In particular, scanning electron microscope coupled with energy-dispersive x-ray detection (SEM-EDX), Raman spectroscopy, x-ray diffraction (XRD), and x-ray photoelectron spectroscopy (XPS) have been used even though they often required complex and costly instrumentations or specific sample pretreatments (Cicileo et al. 2004; Bertolotti et al. 2012; Squarcialupi et al. 2002). Raman spectroscopy, in particular, has revealed to be particularly suited for the analysis of corrosion products (Martens et al. 2003a, b) due to the wide number of inorganic compounds that can be identified in the available spectral range. Successful examples also exist on the application of Raman microscopy to the study of either natural (Hayez et al. 2004) or artificial patinas (Hayez et al. 2005). Electrochemical characterization by means of cyclic voltammetry (CV) and impedance spectroscopy (EIS) has also been proposed to investigate the electrochemical properties of metal patinas (Serghini-Idrissi et al. 2005; Chiavari et al. 2007). All these analytical methods are principally able to provide information on the presence of inorganic components, but they are not suitable for gathering information on the chemical nature of organic coatings. On the other hand, chromatographic techniques allow to characterize the organic fraction, not providing details on inorganic salts (Chiavari et al. 1991; Pitthard et al. 2011). Infrared spectroscopy (400–4000 cm−1), in either transmission or attenuated total reflection (ATR) mode, has been presented as a valid method to identify both organic and inorganic patina components (Matteini et al. 1984; Mazzeo and Joseph 2007). Nevertheless, the transmission mode, especially obtained by dispersing a small quantity of patina sample in KBr salt, proved to be destructive while the attenuated total reflection mode usually requires efficient contact between the ATR crystal and the surface, sometimes difficult to be achieved due to the roughness of the patina.

Infrared reflectance spectroscopy appears as an interesting alternative to identify organic and inorganic constituents of bronze patinas by exploiting both mid-infrared (MIR, 400–4000 cm−1) and near-infrared (NIR, 4000–12500 cm−1) spectral ranges. Reflectance spectroscopy analysis is non-invasive, thus no samples are required. Infrared fiber optic reflectance spectroscopy (FORS), in particular, has been widely applied for the analysis of inorganic and organic compounds in cultural heritage objects (Bacci et al. 1992; Miliani et al. 2007a, b, 2012), although no studies are reported on bronze patinas. Well-known drawbacks of the technique refer to the distortion of bands in the MIR range induced by specular and/or diffuse reflection. Specular reflection is ruled by Fresnel’s law and contributes in the resulting reflectance spectrum by modifying the bands of inorganic compounds into inverted (retststrahlen) bands and those of organic into derivative-shaped signals (Miliani et al. 2012; Griffiths and De Haseth 2007). Conversely, diffuse reflection is governed by Kubelka-Munk’s law and produces modifications in respect of relative band intensity (Miliani et al. 2012), broadening of the normal absorption bands and enhancement of weak signals (Miliani et al. 2012; Rosi et al. 2016).

Even though mathematical function exists to correct the reflectance effects, in most cases, specular or diffuse reflections, contribute both, in different amounts, to the resulting spectrum (Rosi et al. 2016). When the diffuse reflection component is dominant, a significant increment of weak bands, namely overtones and combination bands, is usually observed. These bands are not distorted and appear well recognizable in reflectance mode in both NIR and MIR spectral ranges. For this reason, overtone and combination bands have been considered as markers for the identification of both organic and inorganic compounds. In particular, the potentialities of these bands have been proposed for the identification of artist’s pigments (Miliani et al. 2012; Rosi et al. 2010; Buti et al. 2013) and binders (Vagnini et al. 2009), but they have not been sufficiently explored for the study of corrosion products. To the authors’ knowledge, only a single paper reported the analysis of bronze corrosion patinas by means of portable FTIR spectrometer operating in total reflectance mode just in the MIR (Letardi et al. 2016) in which the signals have been submitted to the Kramers-Kronig mathematical correction.

Considering the interesting possibilities recently explored in infrared microscopy thanks to the extended range of analysis offered by MCT detectors (675–8000 cm−1) (Sciutto et al. 2014), the present research is aimed at evaluating reflectance microscopy technique in the MIR-NIR range (675–7500 cm−1) for the study of bronze historical surfaces.

Thus, we carried out a systematic study to identify the peculiar features showed by total reflection spectra of corrosion products usually found in outdoor bronze patinas. Different corrosion products namely brochantite, antlerite, atacamite, and paratacamite, were synthesized and characterized by means of FTIR microscopy in the NIR-MIR range. We focused the attention on the most characteristic bands of corrosion products with particular attention to NIR features of hydroxychlorides, which are almost unexplored. In fact, while FTIR bands of brochantite and antlerite in the near-infrared range have been previously reported (Rama Subba Reddy et al. 2002; Reddy et al. 1987), no information is available on the spectral features of atacamite or paratacamite (Gaydon et al. 2009). In addition, common organic coatings for bronzes were also included in this research. Primary results on standard samples have been organized in a database, which has been used for the characterization of the composition of micro-samples collected from the Neptune Fountain (sixteenth century), the renaissance bronze sculptures located in Bologna (Fig. 1). This remarkable case study represented a complex system due to the presence of the corrosion products (typical of urban environment) together with other degradation products referred to the deposition of salts contained in the water of the fountain (Marabelli et al. 1991), as well as protective coatings applied on the occasion of the 1989–1990 restoration campaign.

Fig. 1
figure 1

The Neptune Fountain in Bologna

Full size image

Materials and methods

Reference materials in this research

Different corrosion products and organic bronze protective coatings have been considered as reference materials in the research. Copper hydroxysulfates (brochantite and antlerite) and copper hydroxychlorides (atacamite and paratacamite) were synthesized in laboratory using reagent-grade materials and distilled water. Whatman™ Grade 4 (very fast filtering) 55-mm-diameter qualitative filter papers were used in all the syntheses to isolate the obtained compounds after reaction.

Synthesis of brochantite

Brochantite was obtained by addition of sodium hydroxide (200 mg, 100 mmol) to a 50-mL solution of copper sulfate pentahydrate (3.3 g, 264 mmol). The solution was stirred for 1 h at room temperature. The precipitate was collected by filtration and washed (3 × 50 mL) (Marabelli et al. 1991).

Synthesis of antlerite

Antlerite was synthesized by adding dropwise a 10-mL solution of sodium bicarbonate (550 mg, 187 mmol) to a 25-mL boiling solution of copper sulfate pentahydrate (4.15 g, 475 mmol). The solution was stirred for 15 h at 100 °C. The precipitate was collected by filtration and washed with water (3 × 50 mL) and ethanol (2 × 10 mL) (Strandberg 1998).

Synthesis of atacamite

Calcium carbonate (111 mg) was added to a 50-mL solution of copper chloride dehydrate (874 mg, 103 mmol). The solution was left unstirred for 1 day at room temperature. The precipitate was collected by filtration and washed with water (3 × 30 mL) (Tennent and Antonio 1981).

Synthesis of paratacamite

Cuprous chloride (0.99 g, 50 mmol) was added to a 200-mL solution of copper chloride dehydrate (852 mg, 250 mmol) and stirred for 7 days at room temperature. The precipitate was collected by filtration and washed with water (3 × 50 mL) (Sharkey and Lewin 1971).

Incralac, a solvent-based coating formulated from the acrylic resin Acryloid B-44 (copolymer of methyl and ethyl methacrylate) dissolved in xylene/toluene was purchased from Phase Restauro (Florence, Italy). Microcrystalline wax, a mixture of saturated aliphatic hydrocarbons, soluble in aromatic solvents, was purchased from Bresciani s.r.l. (Milan, Italy).

Samples from the bronze sculptures of the Neptune Fountain

Samples were collected in powder form from the different bronze decorative elements of the Neptune Fountain (see Table 1 and Table SM1). Samples were scraped within an area of few millimeters until the characteristic color of the alloy appeared. Micro-samples were selected for their characteristic color and analyzed.

Table 1 List of the samples collected from the bronzes of the Neptune Fountain
Full size table

Reflection infrared microscopy

A Thermo Nicolet iN10MX infrared microscope fitted with a mercury-cadmium-telluride (MCT) detector cooled by liquid nitrogen was used in the research. An integrated camera operating in bright field (BF) mode was used to select the area for the analysis. Spectra were collected in the NIR-MIR range (675–7500 cm−1) using a Cassegrain objective. The spectral resolution was 4 cm−1 and 128 accumulations per spectrum were used. The aperture was set at 150 × 150 μm for analyzing the standard materials and at 120 × 120 μm for the microparticles selected from the samples collected from the bronze sculptures. In order to evaluate the most suitable preparation methods for spectroscopic measurements, reflectance spectra from the synthesized corrosion products were collected either directly on samples in their powder form or on the surface of pressed pellets of pure material, applying a pressure of 1 and 6 tons for 1 min. The highest contribution of diffuse reflection was achieved by working directly on powder, which lead to less distorted spectra (Fig. SM1). Indeed, the diffuse reflectance allows to increase the intensity of combination and overtones not distorted bands, as described by the Kubelka-Munk’s theory. Incralac and microcrystalline wax were dissolved in acetone and white spirit respectively and applied on glass slides. After solvent evaporation, small chips were collected and analyzed. All the spectra collected on standard materials showed good reproducibility.

Microparticles, collected from the bronze sculptures of the Neptune Fountain, were analyzed without any preparation procedure. A good signal was obtained on few grains of patina powder. Spectra were processed by OMIC PICTA Thermo Scientific software.

Transmission infrared spectroscopy

Transmission spectra were recorded using a Thermo Nicolet Avatar 350 equipped with a deuterated triglicine sulfate (DTGS) detector. Spectra were collected by dispersing the sample in KBr (Sigma Aldrich) medium (ratio 1:100) and preparing transparent pellets by applying a pressure of 5 tons for 1 min. The spectra were recorded in the MIR range (400–4000 cm−1) with a spectral resolution of 4 cm−1 and 64 accumulations per spectrum. Transmission spectra in diamond cell were collected by means of the Thermo Nicolet iN10MX infrared microscope presented in the previous section. Each spectrum was obtained using a spectral resolution of 4 cm−1, 64 accumulations, and an aperture of 100 × 100 μm. Spectra were processed by OMNIC and OMNIC PICTA Thermo Scientific software.

X-ray diffraction

X-ray diffraction (XRD) analyses were performed using a Rigaku MiniFlex XRD equipped with Cu Kα radiation. The XRD data were acquired in the range 4–64 2θ° with scan speed of 2°/min.

Results and discussion

Copper hydroxychlorides

Copper chlorides represent the class of the most dangerous components for the stability and conservation of archeological or outdoor bronzes. Cuprous chloride (CuCl) usually grows adjacent to the bronze surface and, in most cases, generates a cycling and self-feeding irreversible corrosion process known as bronze disease (Scott 2002) which attacks and internally damages the bronze alloy. Moisture and oxygen induces an expansion in volume of this compound and a subsequent transformation into copper hydroxychlorides. Differently from cuprous chloride (which grows underneath the cuprite layer), copper hydroxychlorides may be present internally in the corrosion patina up to the outermost layers.

A clear symptom that the bronze sculpture is suffering from chloride corrosion is the detection of copper hydroxychlorides in the corrosion patina. Copper hydroxychlorides are, moreover, soluble in acidic conditions. They can thus be dissolved by thin layers of acidic water that usually form on the surfaces of bronze sculptures (Mazzeo 2005) or leached by acid precipitation, causing patina instability. For these reasons, the identification copper hydroxychlorides is of outmost importance in conservation practice.

There are four isomers of copper hydroxychlorides: (i) clinoatacamite, (ii) atacamite, (iii) paratacamite, and (iv) botallackite. Among them, atacamite and paratacamite (Cu2Cl(OH)3) are probably the most common hydroxychlorides identified in chemical investigations of bronzes. These two compounds present an orthorhombic and rhombohedral structure, respectively (Scott 2002).

Both atacamite and paratacamite were synthesized and characterized by XRD and infrared transmission spectroscopy in the mid-infrared range to ascertain their purity. XRD spectra of synthesized products are in good agreement with those present in XRD mineralogy database, confirming that both the two isomers were successfully obtained (Figure SM2a, b).

According to this outcome, the IR transmission spectrum of atacamite showed the characteristic OH stretching modes at 3341 and 3442 cm−1 and a series of weak bands due to the OH bending modes between 800 and 1000 cm−1 (Fig. SM2c). The mid-infrared transmission spectrum of paratacamite (Fig. SM2d), showed similar features to atacamite, although differences in the number of bands and position were present: the bands corresponding to the OH stretching lied at 3313, 3359, and 3447 cm−1, while the OH bending modes present a characteristic pattern between 820 and 1000 cm−1 (Liu et al. 2011; Martens et al. 2003b).

NIR and MIR spectra were then acquired in total reflection directly on the standard powders. As pointed out in the introduction, reference spectra in the NIR region of atacamite and paratacamite were never clearly reported in previous studies thus allowing, for the first time, the presentation of the characteristic features of these compounds in the considered spectral region.

The NIR reflectance spectra of both atacamite (Fig. 2a, left part of the spectrum) and paratacamite (Fig. 2b, left part of the spectrum) showed similar characteristic features. A strong and sharp band at around 4050 cm−1 and a series of bands at approximately 4200, 4600, 5000 and 5380 cm−1 were identified. Additional bands appeared for both compounds in the region 6500–6900 cm−1. Nevertheless, NIR spectra revealed also interesting differences. In particular, in the region within 4150–4300 cm−1, atacamite showed a band at 4270 cm−1 while paratacamite a doublet at 4191 and 4254 cm−1. Further, a diagnostic band, present in both compounds at 4628 cm−1 for atacamite and 4618 cm−1 for paratacamite, may be used to discriminate between the two hydroxychlorides. It is worth to say that, the correct assignments of bands is difficult due to the absence of specific spectroscopic studies in this region for these compounds. Spectroscopic studies undertaken on hydroxyl-containing minerals may, however, allow to tentatively assign the bands between 6500 and 7000 cm−1 to the first overtone of the OH stretching mode while the bands between 4100 and 4300 cm−1 to the combination bands of OH stretching and bending (Frost et al. 2007; Hunt 1970; Sreeramulu et al. 1990). Indeed, here, for the first time, NIR spectral profiles of atacamite and paratacamite are compared and discussed.

Fig. 2
figure 2

Infrared spectra of atacamite and paratacamite. a NIR-MIR reflectance spectrum of atacamite. b NIR-MIR reflectance spectrum paratacamite

Full size image

MIR total reflection spectrum of atacamite showed a high contribution of the diffuse reflection (Fig. 2a, right part of the spectrum). Thus, bands of OH stretching (3350–3500 cm−1) and OH bending modes (850–1000 cm−1) appeared broader and less defined with respect to those present in the transmission spectrum. In addition, peculiar bands were observed at 1655, 1770, 1842, and 1950 cm−1, which may most likely be ascribable to combinations or overtones. In addition, a sharp band at 2343 cm−1 was identified. A previous research assigned such band to the CO2 entrapped into the crystalline system of natural ultramarine blue from Afghanistan (Miliani et al. 2008). Thus, further investigations are currently ongoing to clarify the origin of this signal. Also for paratacamite, bands related to OH stretching (3329, 3375, and 3464 cm−1) and OH bending modes (820–1000 cm−1) were discernible in the total reflection spectrum (Fig. 2b, right part of the spectrum). As expected, the paratacamite spectrum presented some differences compared to the spectrum of atacamite and some peaks are characterized by a derivative-like shape. Even if this type of deformation is usually referred to organic components, this behaviour was also already observed in weak bands of other minerals (Buti et al. 2013). Similarly to atacamite, paratacamite shows a pattern of bands in the region 1600–2000 cm−1 where the most characteristic are at 1657, 1736, and 1945 cm−1. Bands at 1399 and 1458 cm−1 were detected in atacamite, but they were not present in the spectrum of paratacamite. Table 2 lists the bands observed for the two compounds both in transmission and reflection mode in near and mid-infrared region with tentative assignments.

Table 2 Main absorption bands observed in atacamite and paratacamite spectra. The principal diagnostic bands are reported in italics
Full size table

Copper hydroxysulfates

Copper hydroxysulfates such as posnjakite (Cu4(SO4)(OH)6 • H2O), brochantite (Cu4SO4(OH)6), and antlerite (Cu3SO4(OH)4) are the main constituents of the green patina present on bronze sculptures in urban/industrialized environments, due to the presence of SO2 gas in the atmosphere and the acidity of rain (Scott 2002; Hayez et al. 2004). Posnjakite is usually formed in an initial stage of the patina and turns into the more stable brochantite after several years. Antlerite may also be formed after years and its formation and stability is favored at lower pH conditions than brochantite.

Brochantite and antlerite, which are characterized by a monoclinic and orthorhombic structure, respectively, were considered in this study (Scott 2002). XRD analysis was carried out on the two synthesized products, confirming their purity (Fig. SM3a, b). In FTIR transmission spectra, brochantite presented the OH stretching bands at 3394, 3564, and 3587 cm−1, while the OH bending modes can be observed in the region 650–1000 cm−1 (Fig. SM3c). The antisymmetric stretching of the sulfate group was observed between 1000 and 1150 cm−1 (Secco 1988). Also for antlerite the results of transmission analysis were in agreement to those already reported in the literature (Secco 1988). In particular, OH stretching absorptions were centered at 3486, 3572, and 3581 cm−1, while SO42− antisymmetric stretching bands were at 1082, 1107, and 1153 cm−1 (Fig. SM3d). The OH bending modes can be observed in the region 650–1000 cm−1.

In the near-infrared range, the synthesized brochantite showed a strong band at 4274 cm−1 that can be assigned to the combination of the OH stretching and bending modes (Fig. 3a, left part of the spectrum) and a weak signal at 7000 cm−1 likely related to the first OH overtone. In addition, other small bands were visible at 4026, 4070, 4416, 4689, and 5151 cm−1. Conversely, it is interesting to note that antlerite presented an increased number of well-defined bands in such spectral region, if compared with those referred to brochantite (Fig. 3b, left part of the spectrum). In particular, the most prominent bands are centered at 4246 and 4334 cm−1, where the latter might be ascribed to the combination of the OH stretching and bending. A sharp band at 6979 cm−1 was identified and likely referred to the first overtone of OH stretching modes (Rama Subba Reddy et al. 2002). Antlerite shows also bands at 4080, 4447, and 6775 cm−1 along with several weaker bands in the region between 4500 and 5500 cm−1 (Table 3).

Fig. 3
figure 3

Infrared spectra of brochantite and antlerite. a NIR-MIR reflectance spectrum of brochantite. b NIR-MIR reflectance spectrum of antlerite

Full size image
Table 3 Absorption bands observed in Brochantite and Antlerite spectra
Full size table

The NIR spectral features of brochantite and antlerite have been already described in few previous publications, referred to the analysis of natural minerals (Rama Subba Reddy et al. 2002; Reddy et al. 1987). Zaffino et al. (2015) published the NIR spectrum of synthetic brochantite in mixture with arabic gum or egg white. Here, we presented the spectra of pure brochantite and antlerite synthesized in our laboratory, discussing a more detailed overview on characteristic spectral features. In fact, although the most intense bands are in accordance with those published in literature (Rama Subba Reddy et al. 2002; Reddy et al. 1987; Zaffino et al. 2015), a series of additional well-defined bands which has not been presented before were observed for the two compounds and reported in Table 3.

The MIR total reflection spectrum of powdered brochantite (Fig. 3a, right part of the spectrum) is characterized by a group of bands between 1300 and 2100 cm−1, likely ascribable to overtone and combination bands. The presence of these bands demonstrated that a consistent contribution of diffuse component is present. The most prominent sulfate features at 1114 and 1083 cm−1 related to the antisymmetric SO42− stretching (Secco 1988) appear here inverted, due to the reststrahlen effect. Diffuse reflection was also dominant in the MIR reflectance spectrum of powdered antlerite (Fig. 3b, right part of the spectrum) and here, too, a characteristic group of bands between 1400 and 2300 cm−1 appeared. In the antlerite spectrum, the characteristic features of the antisymmetric SO42− stretching lie in the region 1000–1170 cm−1 while a peak at 988 cm−1 may be assigned to the symmetric stretching of SO42− group (Secco 1988).

For both compounds, the OH stretching bands at 3276, 3406, and 3588 cm−1 for brochantite and 3494 and 3585 cm−1 for antlerite broadened compared to those in transmission while the characteristic bands in the 750–950 cm−1 region became hardly recognizable.

Coatings

Color changes observed on the surfaces of outdoor sculptures and induced by the formation of the corrosion patina do not only imply an aesthetical problem of disfigurement of the sculpture but also can be referred to a more serious damage of the alloy. For this reason, different organic protectives and corrosion inhibitors have been found particularly suited for short-term protection of outdoor bronze sculptures (Scott 2002). In 1991, Marabelli and Napolitano had proposed a more durable protective system compared to those previously used, which could last for several years (Marabelli and Napolitano 1991). This system, known as “double layer system,” implies the application of a first layer of acrylic resin on top of which a second layer of microcrystalline wax, acting as sacrificial layer, is applied. Such protective system is nowadays probably the most widely used in outdoor bronze sculptures in south and central Europe. Thus, acrylic resins such as Incralac and microcrystalline waxes may often be found on metal surfaces intrinsically mixed with the different corrosion products. On this base, the total reflection spectral features of these two organic protectives were investigated and Table 4 reports the bands and tentative assignment for both compounds.

Table 4 Absorption bands observed in Incralac and microcrystalline wax spectra. The principal diagnostic bands for total reflection spectra are reported in italics
Full size table

Incralac and microcrystalline wax presented peculiar spectral profiles in the near-infrared region, which may be used as a diagnostic tool for their identification in complex mixtures. The NIR spectrum of Incralac, which reflects the composition of its main compound, a methyl methacrylate/ethyl acrylate copolymer, is presented in Fig. 4a, left part of the spectrum. The near-infrared region showed several bands with a strong absorption at 4432 cm−1 that can be assigned to the combination of CH stretching and bending motion, while a second combination band due to stretching (CH + C = O) appeared at 4675 cm−1. In addition, the first overtone of the CH was at 5947 cm−1 (Rosi et al. 2016). NIR region also proved to be efficient for the discrimination between the two organic treatments investigated, allowing their clear discrimination and detection even in a complex mixture. In fact, wax was characterized by two peaks appearing at 4252 and 4322 cm−1 that can be attributed to the combination of CH stretching and bending modes. The first overtone of CH stretching lies at 5660 and 5771 cm−1 (Fig. 4b, left part of the spectrum) (Chaffin and Griffiths 1998; Poli et al. 2011).

Fig. 4
figure 4

Infrared spectra of incralac and microcrystalline wax. a NIR-MIR reflectance spectrum of Incralac. b NIR-MIR reflectance spectrum of microcrystalline wax

Full size image

Concerning the MIR spectra of Incralac, the total reflection analysis revealed the presence of a strong derivative-like shape band in the range 1725–1745 cm−1 due to the distortion of the C=O peak (Fig. 4a, right part of the spectrum). Also, the CH bands in the range 1350–1480 cm−1 appeared with a derivative-like shape, but less intense. In the MIR wax spectrum (Fig. 4b, right part of the spectrum), the CH stretching modes in the spectral range between 2820 and 2940 cm−1 appeared distorted due to the specular reflection contribution. Moreover, the enhancement of a series of bands in the region 1700–2700 cm−1 (the most prominent at 1895 and 2019 cm−1) could be referred to the contribution of diffuse reflectance.

The bronzes of the Neptune Fountain

To support an ongoing conservation project aimed at preserving the Neptune Fountain and establishing its original function and majesty, an analytical diagnostic campaign was carried out to evaluate whether there was an evolution concerning the formation of corrosion products and to ascertain the presence of the double layer protective coating system applied during the 1989–1990 restoration. On that occasion, corrosion products (copper hydroxysulfates and copper hydroxychlorides) were identified by means of FTIR and XRD analytical technique and a double layer of acrylic resin and microcrystalline was applied as protective coating (Marabelli et al. 1991; Guida et al. 1994).

In the present research, a selection of samples collected from different areas, according to their color and/or their specific exposure respect to the spurts of the fountain (Table 1), were submitted to infrared micro spectroscopic total reflection measurements, without any pre-treatment.

Copper hydroxysulfate compounds were hardly identified by considering only the MIR range. An illustrative case was represented by sample A collected from the leg of the Neptune. In fact, the related total reflection MIR spectrum (Fig. 5a, right part of the spectrum) presented a broad and not well-resolved band in the OH region, while a peak is visible at around 1153 cm−1 (not attributed). On the other hand, the examination of the NIR spectral region (4000–5600 cm−1) of the same sample (Fig. 5a, left part of the spectrum) allowed the identification of brochantite thanks to the peculiar band at 4274 cm−1 excluding, at the same time, the contemporary presence of antlerite.

Fig. 5
figure 5

Infrared spectra of samples collected from the Neptune Fountain. a Sample A, Neptune leg. b Sample E, SE mermaid, chest. c Sample D, Wind mask, lip

Full size image

It is important to underline that NIR spectra have demonstrated in this case to be much more informative than the MIR spectra, not only in comparison with the total reflection deformed spectral features but also even with the transmission results. Indeed, also in the transmission spectrum collected on a single green particle with the diamond compression cell, the presence of brochantite cannot be observed due to the overlapping of its diagnostic bands with broad and not well-defined bands, sometimes accompanied by small shoulders, in the region between 3300 and 3600 cm−1 and 1000–1100 cm−1 (Fig. SM4).

Sample E revealed a complex spectral profile (Fig. 5b) and by exploring the NIR range, it was possible to recognize its specific composition. The band at 4050 cm−1 (along with the band at 1949 cm−1) suggested the presence of copper hydroxychlorides. However, the band at 4272 cm−1 (Fig. 5b, left part of the spectrum) may suggest the presence of atacamite. The absence of the double diagnostic peak at 4191 and 4254 cm−1 excludes the contribution of paratacamite. Copper hydroxychlorides are often found on bronzes that are part of monumental fountains. Their origin is linked to the presence of chloride salts in the water and/or to chlorinated chemicals intentionally added as disinfectant (Guida et al. 1994). Atacamite was also clearly identified in sample D collected from a wind mask, one of the decorative elements from which the water flowed. The reflectance spectrum was assigned to atacamite by exploiting both the information obtained from the near-infrared (Fig. 5c, left part of the spectrum) and the mid-infrared (Fig. 5c, right part of the spectrum) ranges. Bands in the region of OH bending modes were visible, even though distorted, together with the OH stretching at 3381 and at 3466 cm−1. In addition, the near-infrared bands undeniably attributed this spectrum to atacamite due to the characteristic pattern of copper hydroxychlorides and the absence of the paratacamite diagnostic band at 4191 cm−1.

Concerning the characterization of the protective coatings applied on the occasion of the 1990 restoration intervention, the micro NIR spectroscopy resulted to be crucial for the correct identification of the treatment residues. Sample B, collected from the leg of a putto, revealed the presence of broad and weak bands in the MIR range (Fig. 6a, right part of the spectrum) related to the CH stretching modes in the region between 2800 and 3000 cm−1 and to the C=O stretching at 1738 cm−1 (maximum). The characteristic combination band of Incralac at 4437 cm−1 appeared well defined in the NIR region (Fig. 6a, left part of the spectrum). Traces of organic components were found in other analyzed samples (samples C and F). For sample C (Fig. 6b), it was possible to identify the presence of Incralac thanks to the diagnostic spectral features at 4432 and 4156 cm−1 in the NIR range and at 3442 and 1746 cm−1 (maximum) in the MIR range. Moreover, the NIR spectrum offered the advantages to disclose the presence of wax, due to the bands observed at 4250 and 4322 cm−1 (Fig. 6b, left part of the spectrum). The results were quite significant for the description of the state of conservation of the monument. Indeed, after a long time from the application of the “sacrificial” wax coating layer, wax was still present on the bronzes of the monument. Sample F can be considered as a good example of a complex chemical system present on a bronze surface (Fig. 6c). NIR range was crucial for the detection of wax (see the presence of bands at 4250 and 4323 cm−1) and Incralac (4431 cm−1) present in a certain amount. In addition, the band at 5171 cm−1 may be ascribable to gypsum, likely present as a deposition material. Atacamite and calcium carbonate were identified thanks to the presence of bands at 3383, 3483, 4050, and at 2514 cm−1, respectively. In addition, the strong band at 1736 cm−1 (maximum) is ascribable to the presence of Incralac. Table 5 summarizes the overall results achieved in the characterization of the analyzed samples collected from the Neptune Fountain.

Fig. 6
figure 6

Infrared spectra of samples collected from the Neptune Fountain. a Sample B, NE putto, leg. b Sample C, NE putto, forehead. d Sample F, Inscription

Full size image
Table 5 Summary of the main results obtained on the samples collected from the bronze sculptures of the Neptune Fountain
Full size table

The overview of the results obtained from the Neptune Fountain clearly depicts a worrying state of conservation of the bronzes of the entire monument. First, atacamite was detected in most samples. This carrion product is particularly dangerous because, in most cases, it is the result of an active corrosion mechanism and can easily be leached by acid precipitation. It is likely that atacamite was generated by the interaction between the chloride-rich water of the fountain and the bronze alloy in the areas in which the double protecting layer applied in the 1990 restoration was consumed.

Analysis of samples has shown a very uneven presence of the protective coatings. However, the detection of residues of the acrylic/wax coating or, of at least Incralac, does not guarantee an effective protection against the aggressive and polluted environments. Therefore, the presence on the coating of both macro and micro cracks as well as its consumption are likely to lead to further corrosion phenomena as its main hydrophobic property would be constantly reduced, allowing aggressive salt solutions to be in direct contact with the bronze alloy surface.

To protect the fountain from further corrosion, a restoration campaign was held in 2016–2017. Metal restorers decided to remove the patchy residues of the organic coatings applied on the occasion of the past 1990 restoration intervention and applied a new protective layer made of the same double acrylic/wax components.

Conclusion

In the present research. the potentialities of NIR spectral region were exploited for the characterization of complex patinas present on outdoor monumental bronzes. Common corrosion products and protective organic treatments for bronzes were investigated in the NIR-MIR range to provide a reference database. Particular attention was paid to the description of the characteristic bands observed in the NIR spectra of copper hydroxychlorides, which were almost unexplored. For these compounds a tentative assignment of the bands between 6500 and 7000 cm−1 to the first overtone of the OH stretching mode and of those between 4100 and 4300 cm−1 to the combination of the OH stretching and bending has also been proposed.

The integrated use of MIR and NIR total reflection spectra allowed characterizing bronze patina samples coming from the Neptune Fountain located in Bologna. The results were in agreement with those obtained with IR spectroscopy in transmission mode and with those obtained during the first diagnostic campaign in 1990 (Marabelli et al. 1991). In addition, the peculiar features observed in the NIR range allowed the clear discrimination of corrosion products and organic coatings even in complex mixtures, overcoming problems related to overlapping and distortion of diagnostic bands in the MIR range. The study can be considered as a proof of concept for the possible future application of the near-infrared reflectance spectroscopic technique for in situ diagnostic campaigns on bronze sculptures.