The Influence of Archaeometallurgical Copper Alloy Castings Microstructure Towards Corrosion Evolution in Various Corrosive Medi

corrosion and materials degradation

Article The Inﬂuence of Archaeometallurgical Copper Alloy Castings Microstructure towards Corrosion Evolution in Various Corrosive Media

Olga Papadopoulou * and Panayota Vassiliou

School of Chemical Engineering, National Technical University of Athens, 9 Iroon Polytechneiou str., 15780 Zografou, Greece; [email protected] * Correspondence: [email protected]

Abstract: The local patterns at the interfaces of corrosion stratiﬁcation, developed on two archaeometallurgical bronzes (a Cu-Sn-Pb and a Cu-Zn-Sn-Pb alloy), in the as-cast condition, were assessed by OM and SEM-EDS systematic elemental chemical analyses. Previously, the alloys—whose metallurgical features and electrochemical behaviour were already well studied—have been subjected to laboratory corrosion experiments. The corrosion procedures involved electrochemical anodic polarization experiments in various chloride media: 0.1 mol/L NaCl, 0.6 mol/L NaCl and two other synthetic chloride-containing solutions, representing electrolytes present in marine urban atmosphere and in the soil of coastal sites. The characterization of the Cu-Sn-Pb alloy electrochemical patinas after anodic sweep (OCP+ 0.6 V) revealed that the metal in all electrolytes undergoes extensive chloride attack and selective dissolution of copper which initiates from the dendritic areas acting as anodic sites. The most abundant corrosion products identiﬁed by FTIR in all electrochemical patinas were

Cu2(OH)3Cl), Cu2(OH)2CO3 and amorphous Cu and Sn oxides. The characterization of the Cu-Sn- Citation: Papadopoulou, O.; Pb alloy electrochemical patina after slow anodic sweep (OCP+ 1.5 V) in 0.1 mol/L NaCl reveals Vassiliou, P. The Influence of Archaeometallurgical Copper Alloy selective oxidation of dendrites and higher decuprification rate in these areas. Corrosion products of Castings Microstructure towards Sn-rich interdendritic areas are dominated by oxygen species (oxides, hydroxides, hydroxyoxides) Corrosion Evolution in Various and Cu-rich dendrites by chlorides. In the case of Cu-Zn-Sn-Pb, Zn in dendritic areas is preferentially Corrosive Media. Corros. Mater. attacked. The alloy undergoes simultaneous dezincification and decuprification, with the former Degrad. 2021, 2, 227–247. https:// progressing faster, especially in dendritic areas. The two processes at the alloy/patina interface leave doi.org/10.3390/cmd2020013 behind a metal surface where α-dendrites are enriched in Sn compared to the alloy matrix. The results of this study highlight the dynamic profile of corrosion layer build-up in bronze and brass. Academic Editor: Guang-Ling Song Moreover, the perception of the dealloying mechanisms progression on casting features, at mid-term corrosion stages, is extended. Received: 13 April 2021 Accepted: 14 May 2021 Keywords: archaeometallurgy; bronze; brass; microstructure; accelerated corrosion; chlorides; patina; Published: 19 May 2021 decuprification; dezincification; FTIR spectroscopy

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional afﬁl- 1. Introduction iations. Archaeometallurgical and historical bronzes and brasses are either cast or worked. In most cases, a combination of manufacturing and shaping techniques have been employed for their production. Therefore, the microstructural features of the ﬁnal products are

Copyright: © 2021 by the authors. the result of alloying element concentration, presence of impurities, cooling rate and Licensee MDPI, Basel, Switzerland. subsequent thermal or mechanical treatment [1]. Numerous modern archaeological and This article is an open access article archaeometric studies have been dedicated to the investigation and reconstruction of distributed under the terms and manufacturing procedures and available bronze and brass metallurgical technology in conditions of the Creative Commons many eras and geographical sites where ancient civilizations of the Old and New World Attribution (CC BY) license (https:// have thrived [2]. A great number of the antiquity bronze artefacts, tools and coinage, and creativecommons.org/licenses/by/ also historic and contemporary copper-based sculptures, are produced by casting. The 4.0/).

Corros. Mater. Degrad. 2021, 2, 227–247. https://doi.org/10.3390/cmd2020013 https://www.mdpi.com/journal/cmd Corros. Mater. Degrad. 2021, 2 228

corrosion phenomena on these unique categories of alloys inevitably take place on surfaces that, in some cases, are characterized by complex and heterogeneous topography. During the last few decades, research on corrosion kinetics governing long-term exposure in various contexts—such as soil, marine and atmospheric environments (outdoor and indoor), etc.—has provided the field for important interdisciplinary studies [3–6]. Both case studies [7–11] and laboratory experiments [12–15] have investigated the composition of corrosion layer strata formed after long-term natural exposure and the role of numerous parameters affecting the electrochemical behaviour at the initial corrosion stages, respectively. The chemical composition and topography of the alloy substrate and the physicochemical properties of the corrosive medium are equally important for the assessment of corrosion process development and the nature of corrosion products. The well-established dealloying models in bronzes and brasses—i.e., decuprification [4] and dezincification [16]—are the necessary background for the interpretation of all types of degradation mechanisms in copper-based cultural heritage objects. The addition of tin as an alloying element in copper alloys can be responsible for either a positive or negative impact in corrosion rate, depending on the pH and the aggressive corrosive species concentration of the electrolyte, as well as the moisture levels that can promote the dissolution of water-soluble species. The role of Sn corrosion products in (i) electrochemical reactions related to ‘bronze disease’ mechanism [17], (ii) redox reactions involved in deposition and dissolution processes among chloride-containing corrosion layers in unsheltered atmospheric conditions [18] and (iii) dezincification degree during long-term soil corrosion of ancient brass [19] has been the subject of recent studies. Two challenging intercorrelated research fields for cor- rosionists are the understanding of local galvanic cell action in dealloying processes and the assessment of local chemical reactions between distinct solid phases in the presence of aggressive corrosive agents. New emerging corrosion aspects and relevant physical models can facilitate a more accurate perception and validation of the corrosion kinetics of ancient bronze and brass in a wide range of corrosion environments. Despite the progress and the increasing literature sources in the field of bronze and brass cultural heritage degradation and conservation, the experimental evidence focusing on the local corrosion patterns related to micro-segregation structures—present in archaeometallurgical castings—is rather limited [17,20–23]. Masi et al. [22] have investigated the corrosion patterns related to the dendritic microstructures of as-cast historical bronzes. Characterization of patinas, after artificial ageing tests simulating atmospheric corrosion, has pointed out the emergence of local galvanic cells due to different alloy phase chemical composition. Dendrite cores—α-phase with the lowest Sn content—act as anodic sites, while interdendritic areas—eutectoid phase and δ-phase—act as local cathodes and are corroded to a lesser extent. Apart from the degree of electrochemical attack, the nature of corrosion products also varies between the dendrite core and external boundaries, and even more so between dendrite structures and interdendritic areas [22,23]. Filling the gap of knowledge between the early stages of electrochemical corrosion and the complex stratification after long-term corrosion phenomena in cultural heritage metals is an on-going field of research, and many questions still remain to be answered. The aim of this work is to contribute to the establishment of a universal framework for the interpretation of localized corrosion phenomena in segregated archaeometallurgical copper-based castings, with particular interest in the deeper understanding of severe electrochemical dissolution and chemical leaching mechanisms in various corrosive media. The study of the localized corrosion patterns encountered in archaeological and historical metals is also of great importance and would allow the design of suitable cleaning procedures and protection systems that would not affect or destroy the original metallurgical microstructure or the corrosion stratification. This information is valuable in archaeometric studies and should be treated as part of the archaeological record of cultural heritage metals. Two reference archaeometallurgical copper alloys—representing a typical ternary bronze whose production can be traced from Early Bronze Age in many locations of the Mediterranean and a quaternary brass of the Roman Period—were originally produced for Corros. Mater. Degrad. 2021, 2 229

the purposes of the EFESTUS project [24]. The two castings were used for electrochemical potentiodynamic tests in various synthetic solutions representing the chloride-containing electrolytes involved in marine, soil or atmospheric corrosion processes. The local corrosion patterns of the produced electrochemical patinas—related to dendritic segregation of the reference cast alloys—were studied by combined microscopic and analytical methods. A comprehensive demonstration of the alloying elements and corrosive species distribution across the corrosion stratiﬁcation as a result of accelerated dealloying is attempted.

2. Materials and Methods Specimens of two reference copper alloys, a Tin-bronze (TB): 92.3 Cu, 7.5 Sn and 0.2 Pb and a Zinc-bronze (ZB): 82.5 Cu, 14.0 Zn, 3.0 Sn and 0.5 Pb (% weight composition), produced by a casting technique similar to ancient metallurgy practices [25], were used for laboratory corrosion tests. The bronze specimens were cut from rod castings (cylinder disks of 26 mm diameter and 3 mm thickness) and were used in the as-cast condition without further metallurgical treatment. As a result, the alloy surfaces are highly inhomogeneous when examined from the center towards the disc edge and maintain all the typical characteristics of castings before annealing: dendritic microstructure and defects, such as pores and impurities. The bronze coupons were soldered on metallic bases and then mounted in epoxy resin in order to manufacture reusable working electrodes for the electrochemical tests. Before each potentiodynamic sweep, the electrode surface was ground from 400 up to 1500 grit; afterwards they were rinsed with deionised water, degreased with ethanol and finally dried in an air stream. Before the corrosion experiments, metallographic studies on the alloy surfaces was performed by colour etching with Klemm II reagent [26], after suitable grinding and polishing procedures. Macro-photographic documentation and metallographic observations were conducted by optical microscopy (OM) using a Leitz Aristomet metallographic microscope. Four aqueous solutions—representing chloride-containing electrolytes available at wet conditions in various corrosion environments of the Mediterranean basin (such as seawater, soil, atmosphere)—were employed in accelerated electrochemical corrosion: 0.1 and 0.6 mol/L NaCl (average seawater salinity in Mediterranean basin), a filtrate extracted by Piraeus soil (PRSF) and a aqueous mixture of typical atmospheric pollutants in the city of Athens (AAPM), which can be regarded as an industrial-marine atmospheric environment. The PRSF electrolyte was produced by adding 100 g soil to 400 mL deionised water, heating at 60–70 ◦C for 30 min under vigorous stirring and by subsequent filtration and dilution to a final volume of 500 mL. The soil was collected from an excavation site near the Archaeological Museum of Piraeus, where many archaeological bronze objects have been recovered during the last few decades. The chemical composition has been analysed according to ISO11464 method (Table1).

Table 1. Piraeus soil analysis according to ISO11464.

− 2− − Cl SO4 HCO3 TOC Mg Fe Ca (mg/g) (mg/g) (mg/g) (mg/g) (mg/g) (mg/g) (mg/g) 11.5 25.0 0.5 1.2 32.0 4.9 98.0

The AAPM electrolyte, of a total salt concentration 0.97% w/v, contains the following ionic species: sulphates, nitrates, chlorides and carbonates. The synthesis of the aqueous solution was based on reported data concerning the detected atmospheric aerosol and gaseous pollutants in Athens city during the 1990s [27]. It can be assumed that it approxi- mates the chemical composition of a precipitated electrolyte ﬁlm that could form on bronze sculpture exposed in unsheltered outdoor conditions. The particular electrolyte must be regarded as less corrosive than acid rain. The molar concentration of all ionic species is given in Table2. Corros. Mater. Degrad. 2021, 2 230

Table 2. Molar concentrations of the ionic species in the synthesised AAPM electrolyte.

2− − − − SO4 (mol/L) NO3 (mol/L) Cl (mol/L) HCO3 (mol/L) 0.050 0.025 0.008 5 ·10−5

A GAMRY CMS 100 potentiostat and software connected to a three-electrode cell was employed for the electrochemical potentiodynamic tests. The archaeometallurgical coupon was the static working electrode (WE), a saturated calomel electrode (SCE) was used as reference and a Pt wire was used as a counter electrode. The active surface area of the WE was 5 cm2. Anodic polarization sweeps (OCP+ 0.6 V) at a scan rate of 1 mV/s were performed in 0.6 mol/L NaCl, PRSF and AAPM, using TB reference electrodes in an electrochemical cell of 500 mL volume. Broader anodic polarization sweeps (OCP+ 1.5 V) at a very slow scan rate of 0.25 mV/s were conducted in 0.1 mol/L NaCl, using TB and ZB in an electrochemical cell of 2 L. The two scan rates for the anodic polarization were selected in order to achieve two different dealloying conditions that would represent slow and accelerated dissolution. The very slow scan rate of 0.25 mV/s better approaches the progression of natural electrochemical actions and leaching in dense electrolytes, where mass transport takes control over charge transfer. The fast scan rate (1 mV/s) is a typical rate used in most of the wet electrochemistry potentiodynamic techniques aimed at corrosion rate determination. The electrochemical potentiodynamic curves of TB in 0.6 mol/L NaCl and in PRSF and AAPM electrolytes, as well as those of ZB in 0.1 mol/L NaCl, have been presented in previous works [28,29], where all the details about the electrochemical experiments methodology, setup and the resulting electrochemical behaviour of the alloys are discussed. In this work, the authors continue the results evaluation with a systematic ex situ characterization of all patinas. Scanning Electron Microscopy (SEM) coupled with Energy Dispersive Spectrometry (EDS) was employed for the morphological and elemental chemical analysis, using a FEI Quanta 200 scanning electron microscope equipped with a tungsten ﬁlament coupled with an EDAX analyser. The accelerating voltage of the incident electron beam was set to 20 kV. OM observations of the corroded surfaces, under polarized light, were carried out in order to study the corroded electrode surface after each test. • (OCP+ 0.6 V, scan rate 1 mV/s): OM observations on the corroded electrode surfaces were conducted in some representative areas of TB electrochemical patinas. Patina fragments were collected from all three electrodes to perform morphological examination by SEM and chemical analyses by FTIR and EDS. Patina fractions were detached using a carbon adhesive tape to examine both internal and external surfaces by SEM- EDS. The EDS data (elemental analyses) were processed to serve the investigation of the alloy/patina and patina/electrolyte interface reactions. Similar analyses were also conducted on the metal substrate (areas under detached patinas). Finally, powder micro-samples were mechanically scraped from the three corroded electrode surfaces and were homogenized in order to produce KBr pellets. The FTIR measurements were conducted by an JASCO FT/IR-4200 spectrometer (JASCO International Co. Ltd., Tokyo, Japan) with a TGS detector. All spectra were recorded at a scan range of 4000–400 cm−1, with accumulation set to 32 and resolution to 4.0 cm−1, and un- derwent baseline correction, smoothing and CO2 peak reduction through Spectra Manager software. • (OCP+ 1.5 V, scan rate 0.25 mV/s): The same methodology was applied for the characterization of TB and ZB electrochemical patinas after the end of anodic polarization sweeps in 0.1 mol/L NaCl. These systematic chemical and morphological analyses were employed in order to study the elemental distribution as a result of the alloy dissolution and the precipitation of corrosion compounds at the metal/patina and patina/electrolyte interfaces. In the case of ZB patina, a brief characterization of corrosion patterns by OM and SEM-EDS has been published by the authors in a previous Corros. Mater. Degrad. 2021, 2, FOR PEER REVIEW 5

sweeps in 0.1 mol/L NaCl. These systematic chemical and morphological analyses were employed in order to study the elemental distribution as a result of the alloy dissolution and the precipitation of corrosion compounds at the metal/patina and patina/electrolyte interfaces. In the case of ZB patina, a brief characterization of cor‐ Corros. Mater. Degrad.rosion2021 patterns, 2 by OM and SEM‐EDS has been published by the authors in a previous 231 work [29], together with time‐lapse photographic documentation of the electrochem‐ ical reactions during the sweep and comments on the electrochemical curves. work [29], together with time-lapse photographic documentation of the electrochemi- 3. Results cal reactions during the sweep and comments on the electrochemical curves.

3.1. Metallographic Characterization3. Results of Reference Archaeometallurgical Alloys Colour etching with3.1. Metallographic Klemm II reagent Characterization revealed of the Reference microstructure Archaeometallurgical of the as Alloys‐cast elec‐ trodes. The surfaces macroColour‐photographic etching with documentation Klemm II reagent reveals revealed the the macro microstructure‐segregation of the as-cast features. A great variationelectrodes. of the The grain surfaces size macro-photographic and shape distribution documentation along the reveals electrode the macro-segregation disc radius is observed, features.as a result A great of the variation ingot of cooling the grain process. size and These shape distributioncharacteristics along are the electrode demonstrated in Figuredisc 1 radius for the is observed, ZB alloy. as Small a result‐sized of the grains ingot cooling (skin area) process. are These the characteristicsfirst to are demonstrated in Figure1 for the ZB alloy. Small-sized grains (skin area) are the ﬁrst to solidify as a result ofsolidify the contact as a result of the of melt the contact with ofthe the mould. melt with Moving the mould. towards Moving the towards centre, the centre, the growth of columnarthe growthgrains of(basaltic columnar zone) grains is (basalticobserved, zone) and is observed,at the ingot and centre at the ingot a cluster centre a cluster of equiaxed grains isof the equiaxed last to grains solidify. is the Inside last to the solidify. grains, Inside the the dendritic grains, the segregation dendritic segregation is is clearly depicted (Figureclearly 1b), depicted especially (Figure for the1b), ZB especially alloy, where for the the ZB alloy,dendrites where are the larger dendrites and are larger well‐shaped. and well-shaped.

(a) (b)

FigureFigure 1. 1.Macro-photographic Macro‐photographic documentation documentation of ZB referenceof ZB reference alloy after alloy colour after etching colour with etching Klemm with II reagent: Klemm ( a) the wholeII reagent: specimen (a) surface; the whole (b) detail specimen of previous surface; depicting (b) thedetail dendritic of previous structure depicting inside the grains.the dendritic The dotted structure lines highlight theinside boundaries the grains. between The the dotted distinct lines zones highlight of the solidiﬁcation the boundaries process between (where the S: skin, distinct B: basaltic zones area of the and solid E: equiaxed‐ grainsification area). process (where S: skin, B: basaltic area and E: equiaxed grains area). According to the nominal composition of the archaeometallurgical alloys and the According to thephase nominal diagrams composition [1] corresponding of the toarchaeometallurgical non-equilibrium conditions alloys that and apply the for usual phase diagrams [1] correspondingcasting procedures, to non three‐equilibrium phases are conditions present in thethat TB apply alloy—an for usualα-(Cu-Sn) cast‐ dendritic ing procedures, threenetwork, phases are an ( αpresent+ δ)-eutectoid in the TB phase alloy—an and an immiscible α‐(Cu‐Sn) Pb dendritic phase—and network, two phases in the an (α + δ)‐eutectoid phaseZB alloy—an and anα immiscible-(Cu-Zn-Sn) phasePb phase—and and an immiscible two phases Pb phase. in the Optical ZB microscopealloy— images from the etched surfaces, indicative of the micro-segregation features, are presented in an α‐(Cu‐Zn‐Sn) phase and an immiscible Pb phase. Optical microscope images from the Figure2. TB exhibits a cored dendritic structure (Figure2a). The Cu-depletion gradient of etched surfaces, indicativealpha dendrites of the micro is visible‐segregation by the colour features, variation are from presented dark brown in Figure in the centre 2. TB of the alpha exhibits a cored dendriticdendrites structure to pink (Figure borders. The2a). dendritesThe Cu‐depletion have rounded gradient branches of and alpha irregular den‐ shape. drites is visible by the colourThe interdendritic variation from area correspondsdark brown to inα +theδ eutectoid centre of solid the solution, alpha den which‐ is visible drites to pink borders.in The dark dendrites red with silver-grey have rounded islets of branchesδ phase (Cu and31Sn irregular8). The grain shape. boundaries of this alloy are not easily distinguished. Usually, the Sn-rich δ phase tends to precipitate between the grains (Figure2a), because it solidiﬁes at a lower temperature. ZB also exhibits a dendritic segregation, but no coring was evidenced. The grain boundaries of large grains are visible after etching along with porosity and inclusions related to the casting process (Figure2b).

Corros. Mater. Degrad. 2021,Corros. 2, FOR Mater. PEER Degrad. REVIEW2021, 2 6 232

(a) (b)

FigureFigure 2. Optical 2. Optical microscope microscope images (× images100) of: (×100) (a) TB (the of: arrows(a) TB indicate(the arrows the islets indicate of δ phase the islets inside of the δ αphase+ δ eutectoid); in‐ (b) ZBside archaeometallurgical the α + δ eutectoid); alloy ( afterb) ZB colour archaeometallurgical etching by Klemm II alloy reagent, after revealing colour theetching dendritic by Klemm microstructure II rea‐ of the as-castgent, surfaces. revealing the dendritic microstructure of the as‐cast surfaces. 3.2. Accelerated Corrosion in Total Immersion Conditions/Anodic Polarization in The interdendriticVarious area Electrolytes corresponds to α + δ eutectoid solid solution, which is visible in dark red with silver3.2.1.‐ 1stgrey Experimental islets of δ Section—Characterizationphase (Cu31Sn8). The grain of Electrochemical boundaries of Patinas this alloy in Three are not easily distinguished.Synthetic Electrolytes Usually, the Sn‐rich δ phase tends to precipitate between the grains (Figure 2a), becauseThe electrochemical it solidifies at behaviour a lower temperature. of TB casting in ZB three also solutions exhibits representing a dendritic the elec- segregation, but notrolytes coring available was evidenced. in different The marine grain environments boundaries (urban of large atmosphere grains are near visible coast, soil in coastal site, seawater) and the most representative corrosion features of the distinct solidi- after etching alongfication with porosity zones, created and inclusions by macro-segregation, related to havethe casting been reported process in [(Figure28]. In all 2b). corrosive media, the corrosion attack starts from the electrode disc edge and proceeds towards its 3.2. Accelerated Corrosioncentre, following in Total Immersion the direction Conditions/Anodic of the solidification Polarization front. More specifically, in Various EDS elemental Electrolytes analyses on TB corroded surface after anodic dissolution in 0.6 mol/L NaCl has pointed out higher tin and oxygen levels near the electrode edge compared to the electrode centre. 3.2.1. 1st Experimental Section—Characterization of Electrochemical Patinas in Three This finding indicates a faster progress of copper dissolution at the periphery. In this work, Synthetic Electrolytesthe investigation continues further with validation of these findings by more elaborate The electrochemicalchemical behaviour characterization of TB of casting electrochemical in three patinas solutions and with representing studies on micro-segregation the elec‐ trolytes available influencein different in corrosion marine evolution.environments The anodic (urban polarization atmosphere curves near in 0.6 coast, mol/L soil NaCl, in PRSF Corros. Mater. Degrad. 2021, 2and, FOR AAPMPEER REVIEW electrolytes [28] are presented again in Figure3 to give an overview7 of the coastal site, seawater) and the most representative corrosion features of the distinct solid‐ corrosion systems and to facilitate discussion. ification zones, created by macro‐segregation, have been reported in [28]. In all corrosive media, the corrosion attack starts from the electrode disc edge and proceeds towards its centre, following the direction of the solidification front. More specifically, EDS elemental analyses on TB corroded surface after anodic dissolution in 0.6 mol/L NaCl has pointed out higher tin and oxygen levels near the electrode edge compared to the electrode centre. This finding indicates a faster progress of copper dissolution at the periphery. In this work, the investigation continues further with validation of these findings by more elab‐ orate chemical characterization of electrochemical patinas and with studies on micro‐seg‐ regation influence in corrosion evolution. The anodic polarization curves in 0.6 mol/L NaCl, PRSF and AAPM electrolytes [28] are presented again in Figure 3 to give an over‐ view of the corrosion systems and to facilitate discussion.

Figure 3. AnodicFigure polarization3. Anodic polarization curves curves (OCP+ (OCP+ 0.6 0.6 V) V) acquired acquired in the the three three corrosive corrosive media media at a typical at a scan typical rate of scan 1 mV/s, rate of 1 mV/s, after 1 h at OCPafter 1 conditions. h at OCP conditions. Data ﬁrst Data presented first presented in [in28 [28].].

OM images of electrochemical patina grown on a TB‐0.6 mol/L NaCl system are pre‐ sented in Figures 4 and 5. The local patterns from electrode edge (Figure 4) and centre (Figure 5) reveal a variety of precipitated dissolution products which follow the cored dendritic structure of the alloy substrate.

(a) (b) Figure 4. OM images (×100) of the corrosion patterns observed at the segregated surface of TB castings after the end of anodic sweeps in 0.6 mol/L NaCl: (a) corroded dendritic structure and (b) Cu(II) corrosion products formation inside voids —These features are encountered near the elec‐ trode edge.

The centres of dendrites bodies and branches correspond to the metallic Cu or oxi‐ dized Cu(I) state, while dendrite boundaries (shown in pink colour) are a mixture of Cu(I) and Sn(IV) compounds. The interdendritic areas are covered by white Sn(IV) compounds, which contribute to the stabilization behaviour of the alloy at high anodic potentials. In Figure 4b, the evolution of green Cu(II) hydroxyl‐chlorides inside the voids is observed.

Corros. Mater. Degrad. 2021, 2, FOR PEER REVIEW 7

Corros. Mater. Degrad. 2021, 2 233 Figure 3. Anodic polarization curves (OCP+ 0.6 V) acquired in the three corrosive media at a typical scan rate of 1 mV/s, after 1 h at OCP conditions. Data first presented in [28].

OM images of electrochemicalOM images of patina electrochemical grown on patina a TB‐0.6 grown mol/L on NaCl a TB-0.6 system mol/L are NaCl pre‐ system are sented in Figures presented4 and 5. The in Figures local 4patterns and5. The from local electrode patterns from edge electrode (Figure edge 4) and (Figure centre4) and centre (Figure 5) reveal a(Figure variety5) revealof precipitated a variety of dissolution precipitated products dissolution which products follow which the followcored the cored dendritic structuredendritic of the alloy structure substrate. of the alloy substrate.

(a) (b)

FigureFigure 4. OM 4. images OM images (×100) (×100) of the corrosionof the corrosion patterns patterns observed observed at the segregated at the segregated surface of TB surface castings of afterTB the end of Corros. Mater. Degrad. 2021, 2, FORanodic PEERcastings sweeps REVIEW inafter 0.6 the mol/L end NaCl: of anodic (a) corroded sweeps dendritic in 0.6 mol/L structure NaCl: and ( ab)) corroded Cu(II) corrosion dendritic products structure formation and inside (b) 8 voids

—TheseCu(II) features corrosion are encountered products near formation the electrode inside edge. voids —These features are encountered near the elec‐ trode edge.

(a) (b)

FigureFigure 5. OM 5. pictures OM pictures (×100) of(×100) the corrosion of the corrosion patterns observedpatterns atobserved the segregated at the segregated surface of TB surface castings of after TB the end of anodiccastings sweeps after in 0.6 the mol/L end NaCl: of anodic (a) and sweeps (b) corresponding in 0.6 mol/L toNaCl: two different(a) and ( areasb) corresponding where the cored to two dendritic differ structure‐ exhibitsent local areas dealloying where the patterns cored —These dendritic features structure are encountered exhibits local near dealloying the electrode patterns centre. —These features are encountered near the electrode centre. The centres of dendrites bodies and branches correspond to the metallic Cu or oxidized The relevant OMCu(I) images state, while from dendrite the corroded boundaries TB surfaces (shown inin pinkPRSF colour) and AAPM are a mixture electro of‐ Cu(I) and lytes were dominatedSn(IV) by compounds. light green powdery The interdendritic Cu(II) compounds. areas are covered A thicker by white deposition Sn(IV) of compounds, these compounds whichwas observed contribute on to the the α‐ stabilizationdendrite network behaviour in of PRSF the alloy while, at high in the anodic case potentials.of In Figure4b, the evolution of green Cu(II) hydroxyl-chlorides inside the voids is observed. AAPM, the green electrochemical patina was uniform [28]. The corrosion patterns on the The relevant OM images from the corroded TB surfaces in PRSF and AAPM elec- internal surface (interfacetrolytes were with dominated alloy) of by the light two green latter powdery patinas Cu(II) are compounds.depicted by A SEM thicker in deposition Figure 6. of these compounds was observed on the α-dendrite network in PRSF while, in the case of AAPM, the green electrochemical patina was uniform [28]. The corrosion patterns on the internal surface (interface with alloy) of the two latter patinas are depicted by SEM in Figure6.

(a) (b) Figure 6. SEM backscattered electron images (×300) of TB electrochemical patinas samples (inter‐ nal surface in contact with metal substrate) formed after anodic polarization (OCP+ 0.6 V) in (a) AAPM and (b) PRSF electrolyte (D: dendrite, I: interdendritic area, L: lead islet).

The chemical characterization of AAPM‐produced electrochemical patina is dis‐ cussed in detail. EDS data from the distinct dendritic, interdendritic areas and lead glob‐ ules were processed to assess the alloying elements and corrosive species distribution (Ta‐ ble 3). The corrosion products that evolved on dendrites contain higher Cu concentration compared to the interdendritic areas, in accordance with the micro‐segregation terrain of the bronze substrate. The general trend of Cu and Sn alloying elements distribution in the corroded micro‐segregated structures of TB casting after the anodic sweep in AAPM are in accordance with the corroded structures found in a quaternary Cu ‐Sn‐Zn‐Pb alloy cast‐ ing after artificial ageing tests (dropping test, wet and dry cycles) with artificial rain solu‐ tion [23], which also contains sulphates, chlorides and nitrate species. However, both pat‐ ina areas are Sn‐enriched in respect to the metal matrix atomic concentration (4% at Sn). The dendrite corrosion features exhibit an average 13% Sn content, and the interdendritic ones exhibit 16% Sn. The lead‐rich corrosion islets have an average 20% Pb and 18% Sn.

Corros. Mater. Degrad. 2021, 2, FOR PEER REVIEW 8

(a) (b) Figure 5. OM pictures (×100) of the corrosion patterns observed at the segregated surface of TB castings after the end of anodic sweeps in 0.6 mol/L NaCl: (a) and (b) corresponding to two differ‐ ent areas where the cored dendritic structure exhibits local dealloying patterns —These features are encountered near the electrode centre.

The relevant OM images from the corroded TB surfaces in PRSF and AAPM electro‐ lytes were dominated by light green powdery Cu(II) compounds. A thicker deposition of these compounds was observed on the α‐dendrite network in PRSF while, in the case of Corros. Mater. Degrad. 2021, 2 AAPM, the green electrochemical patina was uniform [28]. The corrosion patterns234 on the internal surface (interface with alloy) of the two latter patinas are depicted by SEM in Figure 6.

(a) (b)

FigureFigure 6. SEM 6. SEM backscattered backscattered electron images images (× (×300)300) of of TB TB electrochemical electrochemical patinas patinas samples samples (internal (inter‐ nalsurface surface in in contact contact with with metal metal substrate) substrate) formed formed after anodic after anodic polarization polarization (OCP+ 0.6 (OCP+ V) in (0.6a) AAPM V) in (a) AAPMand ( andb) PRSF (b) electrolytePRSF electrolyte (D: dendrite, (D: dendrite, I: interdendritic I: interdendritic area, L: lead area, islet). L: lead islet).

TheThe chemical chemical characterizationcharacterization of AAPM-producedof AAPM‐produced electrochemical electrochemical patina is discussedpatina is dis‐ cussedin detail. in detail. EDS data EDS from data the from distinct the dendritic,distinct dendritic, interdendritic interdendritic areas and lead areas globules and lead were glob‐ processed to assess the alloying elements and corrosive species distribution (Table3 ). The ules were processed to assess the alloying elements and corrosive species distribution (Ta‐ corrosion products that evolved on dendrites contain higher Cu concentration compared bleto 3). the The interdendritic corrosion products areas, in accordance that evolved with on the dendrites micro-segregation contain higher terrain ofCu the concentration bronze comparedsubstrate. to The the general interdendritic trend of Cuareas, and in Sn accordance alloying elements with the distribution micro‐segregation in the corroded terrain of themicro-segregated bronze substrate. structures The general of TB casting trend afterof Cu the and anodic Sn alloying sweep in elements AAPM are distribution in accordance in the corrodedwith the micro corroded‐segregated structures structures found in a quaternary of TB casting Cu -Sn-Zn-Pb after the alloy anodic casting sweep after in artiﬁcial AAPM are in accordanceageing tests with (dropping the corroded test, wet structures and dry cycles) found with in a artiﬁcial quaternary rain Cu solution ‐Sn‐Zn [23‐],Pb which alloy cast‐ ingalso after contains artificial sulphates, ageing tests chlorides (dropping and nitrate test, wet species. and However,dry cycles) both with patina artificial areas rain are solu‐ tionSn-enriched [23], which in also respect contains to the metalsulphates, matrix chlorides atomic concentration and nitrate (4%species. at Sn). However, The dendrite both pat‐ inacorrosion areas are features Sn‐enriched exhibit in an respect average to 13% the Sn metal content, matrix and theatomic interdendritic concentration ones exhibit (4% at Sn). 16% Sn. The lead-rich corrosion islets have an average 20% Pb and 18% Sn. The dendrite corrosion features exhibit an average 13% Sn content, and the interdendritic onesTable exhibit 3. Normalized 16% Sn. % The atomic lead elemental‐rich corrosion distribution islets of O, have Cl, S and an Naverage (corresponding 20% Pb to and O2− ,18% Cl−, Sn. 2− − SO 4 and NO3 anionic species), detected by EDS on the internal surface of TB electrochemical patina, produced by anodic polarization in AAPM. Average values of indicative spot analyses conducted in dendritic, interdendritic areas and Pb islets.

Element Dendritic Network Interdendritic Areas Leaded Areas O 76.0 71.5 73.2 Cl 22.8 27.0 14.0 S 1.2 1.5 8.0 N - - 4.8

The identiﬁcation of the majority of patina constituents (from all corrosion strata and not just the superﬁcial layer) was possible by spectroscopic analyses on pulverized and homogenized corrosion depositions. Here, the vibrational patterns of the FTIR spectra of all three electrochemical produced patinas are presented (Figure7). Corros. Mater. Degrad. 2021, 2, FOR PEER REVIEW 9

Corros. Mater. Degrad. 2021, 2 235

0.25 3355 & 3339 & 3327 FTIR // 4000-3000 bands 3447

0.2

3355 & 3340

0.15 3450 PRSF 3352 & 3347 &3327 AAPM 3447

Absorbance 0.1 0.6 M NaCl 3522

0.05

0 4000 3900 3800 3700 3600 3500 3400 3300 3200 3100 3000 -1 Wavenumber (cm ) (a)

0.08

1385 0.07

0.06

PRSF 0.05 1558 1653 AAPM 0.04 1635 1541 0.6M NaCl 1457 1507 1409 Absorbance 0.03 1472 1521

0.02

0.01

0 1700 1650 1600 1550 1500 1450 1400 1350 1300 Wavenumber (cm-1)

(b)

0.08 897 PRSF 0.07 849 AAPM 0.6M NaCl 0.06 834 988 919 823

0.05 926 1124 952 1065 0.04 917

Absorbance 924 0.03

0.02 1113 1165 0.01 922

0 1220 1170 1120 1070 1020 970 920 870 820 770 -1 Wavenumber (cm ) (c)

Figure 7. Cont.

Corros. Mater. Degrad. 2021, 2, FOR PEER REVIEW 10 Corros. Mater. Degrad. 2021, 2 236

0.09 PRSF 0.08 AAPM 0.6M NaCl 0.07

0.06 419 518 0.05 459 600 474 420 0.04 506 444 526 482 Absorbance 0.03 493 450

0.02 515 489 473 435 0.01 508 458 474 0 434 650 600 550 500 450 400

Wavenumber (cm-1)

(d)

Figure 7. FTIRFigure spectra 7. FTIR of spectracorrosion of corrosionproducts productscollected collectedfrom TB fromelectrode TB electrode surfaces surfacesat the end at theof end of anodic anodic sweepssweeps (OCP+ (OCP+ 0.6 0.6V) V)in 0.6 in 0.6mol/ mol/LL NaCl, NaCl, AAPM AAPM and andPRSF PRSF electrolytes: electrolytes: (a) Bands (a) Bands 4000–3000 4000–3000 cm−1, cm−1, (b) bands 1700–1300 cm−1, (c) bands 1270–770 cm−1 and (d) bands 650–400 cm−1. (b) bands 1700–1300 cm−1,(c) bands 1270–770 cm−1 and (d) bands 650–400 cm−1.

The two powderThe two samples powder (AAPM samples and (AAPM PRSF and patinas) PRSF were patinas) light were green, light and green, the third and the third (0.6 M NaCl(0.6 Mpatina) NaCl was patina) whitish. was whitish. The most The important most important vibrational vibrational patterns patterns are summa- are summarized rized in Tablein Table 4. 4.

Table 4. ObservedTable 4. Observed bands in bands FTIR inspectra FTIR spectraof homogenized of homogenized patina patina powder powder samples samples after after anodic anodic polarization polarization in PRSF, in PRSF, AAPM AAPM andand 0.6 0.6 mol/L mol/L NaCl. NaCl.

CorrosiveCorrosive Me- Medium Type of Vibration Bands (cm−1) Attribution Reference Type of Vibration Bands (cm−1) Attribution Reference dium 3522 (w), 3447 (s), 3355, 3339, PRSF 3522 (w), 3447 (s), 3355,3327 (triple broad peak) Cu (OH) Cl—intermediate of 3522 (w), 3450 (s), 3355, 3340 2 3 PRSF AAPM OH3339, stretching 3327 (triple broad botallackite and [30–32] (triple broad peak) peak) atacamite polymorphs 3522 (w), 3447Cu2(OH) (s), 3352,3Cl—intermediate 3347, 0.6 mol/L NaCl 3522 (w), 3450 (s), 3355, AAPM OH stretching 3327 (tripleof botallackite broad peak) and ataca- [30–32] 3340 (triple broad peak)many weak bands in-plane OH deformation mite polymorphs all electrolytes 3522 (w), 3447 (s), 3352,within 1620–1660 Sn oxyhydroxides [33] of H2O 0.6 mol/L NaCl 3347, 3327 (triple broad range weak overtones of Sn-O-Sn 1558, 1541, 1521, 1507, 1472, all electrolytes peak) Sn oxyhydroxides [34] and Sn-O-H 1457, 1409 many weak bands within PRSFin-plane OH defor- 1385 (strong) all electrolytes 1620–1660 Sn oxyhydroxides [33] AAPMmation of H2O 1385 (weak) Cu (OH) CO —malachite [35,36] range 2 2 3 0.6 mol/L NaCl 1385 (weak) weak overtones of 1558, 1541, 1521,2− 1507, Cu4(OH)6SO4—brochantite or all electrolytes AAPMSn-O-Sn and Sn-O-internal vibration of SO4 1124Sn oxyhydroxides [34] 1472, 1457, 1409 posnjakite or Sn oxyhydroxides H [34,35] weak overtones of Sn-O-Sn 0.6 mol/L NaCl 1113 Sn oxyhydroxides PRSF and Sn-O-H 1385 (strong)

AAPM PRSF 1385 (weak) 988, 952, 926,Cu2 919,(OH) 897,2CO 8493—malachite Cu2(OH) 3Cl—atacamite[35,36] [30–32,35]

0.6 mol/L NaCl 1385 (weak) Cu2(OH)3Cl—atacamite AAPM 988 *, 952, 924, 917, 897, 849 * also Cu (OH) SO —brochantite [30–32,35] Cu4(OH)6SO4—brochantite4 6 4 internal vibrationOH deformation (Cu-OH bending) AAPM 1124 or posnjakite or Sn oxyhy- 0.6 mol/L NaClof SO42− 988, 952, 922, 897, Cu (OH) Cl—atacamite [30–32] droxides 2 3 All electrolytes 834 (sh), 823 (sh) Cu (OH) CO —malachite[34,35] [35] weak overtones of 2 2 3 nanocrystalline or amorphous SnO 0.6 mol/L NaClall electrolytes Sn-O-Sn and Sn-O- Sn-O or Cu-O vibrations1113 600 (broad)Sn oxyhydroxides 2 [36] or Cu2O H

Corros. Mater. Degrad. 2021, 2 237

Table 4. Cont.

Corrosive Medium Type of Vibration Bands (cm−1) Attribution Reference PRSF 526 (sh), 518, 506 intermediate of atacamite and AAPMCu-O and Cu-OH 526, 515, 508 [31,35] botallackite, malachite 0.6 mol/L NaCl 526 (sh), 518, 508

PRSF 492, 482 Cu2.5(OH)2CO3—malachite and Cu-O stretching Cu (OH) SO —brochantite [35] AAPM 489, 482 4 6 4 PRSF 474

AAPMCu-O stretching 473 Cu2(OH)3Cl—atacamite [31] 0.6 mol/L NaCl 474 PRSF 459, 450, 444, 435 AAPMCu-O stretching 458, 444 Cu2(OH)3Cl—botallackite [30,31] and atacamite 0.6 mol/L NaCl 458, 444, 434

PRSF 419 Cu2(OH)3Cl—botallackite

Cu-O stretching or internal Cu2(OH)3Cl—botallackite or AAPM2− 420 [31,35] vibration of SO4 Cu4(OH)6SO4—brochantite

0.6 mol/L NaCl 419 Cu2(OH)3Cl—botallackite

The main detections in all cases are the Cu(II) hydroxychlorides with the general formula Cu2(OH)3Cl. In fact, the peaks observed are a better match with synthetic compounds—produced by precipitation experiments [30]—rather than natural minerals [31,32] and correspond to intermediate products between two Cu2(OH)3Cl polymorphs, botallackite and atacamite, and their copper chloride precursors. This mixture is poorly crystallized, as indicated by the strong wide band in the range 3360–3310 cm−1. This band consists of a triplet of convoluted peaks, merely resolved. The weak peak at 3522 cm−1 and the strong peak at 419–429 cm−1 are an exact match with the botallackite reference spectrum. Malachite (Cu2(OH)2CO3) was also identified among the corrosion products in all three patinas. In the case of PRSF patina, the higher malachite content is in accordance with the high calcium carbonate concentration dissolved in the soil filtrate. In AAPM patina, the presence of Cu2(OH)2CO3 originates from the dissolved carbonate salts of the solution, and of course in all corrosive media, dissolved atmospheric CO2 should be considered as a source of carbonates. Some weak characteristic bands (1124 and 482 cm−1), related to basic copper sulphates—a closer match to Cu4(OH)6SO4 (brochantite)—are also observed in the AAPM patina and, to a lesser extent, in the PRSF. The particular peaks in many cases overlap with Cu2(OH)3Cl bands and were difficult to distinguish. Despite the nitrogen detection by EDS elemental analysis locally in the AAPM patina, no matches to basic copper(II) nitrates were found in the AAPM IR spectrum. The FTIR spectra of some of the weak bands around 1620–1660 cm−1 could be at- tributed to deformation of H2O. Broad peaks in this range have been detected in corrosion product samples after anodic polarization in 0.1 mol/L NaCl [33] and in corrosion crusts on buried archaeological bronzes [36] by Robbiola et al. The identification of peaks observed within the range 1570–1020 cm−1 was incomplete, although it could be possible that the overtones of Sn-O-Sn and Sn-O-H vibrations are included. Some individual peaks −1 appearing at 526, 518, 506 and 473, 917–922 cm , could be attributable to CuSnO3 and to abhurite (Sn21O6Cl16(OH)14), respectively. According to the Pourbaix themodynamic diagrams and the relevant literature [33,36,37], the presence of mixed Cu-Sn oxides was considered unlikely, and the formation of Sn(II) chloride-containing compounds is not favoured at that particular electrolyte pH and potential domain. Especially at the end of the polarization sweep (highly anodic potential and strongly alkaline pH), compounds such as abhurite—whose formation and stability is favoured at acidic pH—could not be possibly sustained within the patina without further oxidation. In any case, the above-mentioned Corros. Mater. Degrad. 2021, 2 238

−1 bands were matched to Cu2(OH)3Cl compounds. Within the range 400–720 cm , characteristic vibrations of M-O and M-OH type (where M: Cu or Sn) are observed.

3.2.2. 2nd Experimental Section—Characterization of Electrochemical Patinas in 0.1 mol/L NaCl The anodic polarization curves of TB and ZB in 0.1 mol/L NaCl at a slow scan rate are presented in Figure8. The same basic domains are observed in both curves (Tafel region, intense anodic peak and a stable plateau at high current density levels. A second, weak anodic peak, followed by a dissolution ramp, are observed in brass casting (ZB). The steady state condition, reached above +400 mV/SCE, leads to the formation of compact corrosion Corros. Mater. Degrad. 2021, 2, FOR PEER REVIEW 13 layers. A further investigation is necessary in order to determine whether the mechanism of this stage fulﬁls the criteria of electrochemical passivation.

FigureFigure 8. 8. AnodicAnodic polarization polarization curves curves of TB of TBand and ZB ZB(OCP+ (OCP+ 1.5 V) 1.5 acquired V) acquired in 0.1 in mol/L 0.1 mol/L NaCl NaCl at a slow at a scan slow rate scan of rate 0.25 of mV/s,0.25 mV/s, after 30 after min 30 at minOCP at conditions. OCP conditions. ZB curve ZB was curve first was presented ﬁrst presented in [29]. in [29].

TheThe corrosion corrosion products products expected expected to to be be found found in in the the patinas patinas of of the the Cu Cu-Sn-Pb‐Sn‐Pb bronze bronze andand Cu Cu-Zn-Sn-Pb‐Zn‐Sn‐Pb brass, brass, after after the the sweeps, sweeps, were were copper copper oxides oxides and chlorides, and chlorides, tin and tin lead and oxideslead oxides and hydrated and hydrated tin hydroxy tin hydroxy-oxides.‐oxides. Due to the Due high to theanodic high dissolution anodic dissolution rate, the ma rate,‐ joritythe majorityof these products of these are products amorphous are amorphous or nano‐crystalline. or nano-crystalline. In our study, In very our few study, crystal very‐ linefew compounds crystalline were compounds identified were by XRD identified (nantokite by XRD and (nantokite cuprite) and and other cuprite) spectroscopic and other techniques,spectroscopic such techniques, as Raman suchanalyses as Raman (not presented analyses (notin this presented study), inalso this gave study), poor also results gave becausepoor results of the because highly ofimpure the highly nature impure and the nature distorted and the stoichiometry distorted stoichiometry (due to out (due‐of‐equi to out-‐ libriumof-equilibrium formation formation conditions) conditions) of these ofcorrosion these corrosion compounds. compounds. Similar Similarobservations observations about poorlyabout crystallized poorly crystallized compounds compounds are also are reported also reported by case by studies case studies dealing dealing with natural with natural and artificialand artificial atmospheric atmospheric corrosion corrosion processes—simulating processes—simulating unsheltered unsheltered urban outdoor urban outdoor condi‐ conditions—where leaching takes place up to some degree [22]. Following the same tions—where leaching takes place up to some degree [22]. Following the same methodol‐ methodology as in the previous experimental section, we focused on the elemental mapping ogy as in the previous experimental section, we focused on the elemental mapping by EDS by EDS analyses that can provide valuable information about the corrosion products analyses that can provide valuable information about the corrosion products stoichiome‐ stoichiometry and the dissolution profiles of distinct patina features. The elemental analyses try and the dissolution profiles of distinct patina features. The elemental analyses graphs graphs correspond to the average values of small groups of spot or fullframe analyses. The correspond to the average values of small groups of spot or fullframe analyses. The data data can be utilized for qualitative assessment of local chemical composition and dealloying can be utilized for qualitative assessment of local chemical composition and dealloying trends. More extensive analytical work would be required in order to report accurate trends. More extensive analytical work would be required in order to report accurate quantitative results and support statistical analysis. Some indicative SEM images of the quantitative results and support statistical analysis. Some indicative SEM images of the internal surface of TB detached patina (Figure 9) and of the TB and ZB alloy (Figure 10) show the typical morphological characteristics of corrosion products epitaxial growth on dendritic structure as a result of patina/alloy interface electrochemical processes.

Corros. Mater. Degrad. 2021, 2 239

internal surface of TB detached patina (Figure9) and of the TB and ZB alloy (Figure 10) Corros. Mater. Degrad. 2021, 2, FOR PEERshow REVIEW the typical morphological characteristics of corrosion products epitaxial growth on14

Corros. Mater. Degrad. 2021, 2, FOR PEER REVIEWdendritic structure as a result of patina/alloy interface electrochemical processes. 14

Figure 9. SEM backscattered electron image (×200) of TB detached patina (internal surface in con‐ Figure 9. SEM backscattered electron image (×200) of TB detached patina (internal surface in contact tact with metal) after anodic polarization (OCP+ 1.5 V) in 0.1 mol/L NaCl solution. Figurewith 9. SEM metal) backscattered after anodic electron polarization image (OCP+ (×200) 1.5 of V) TB in detached 0.1 mol/L patina NaCl (internal solution. surface in con‐ tact with metal) after anodic polarization (OCP+ 1.5 V) in 0.1 mol/L NaCl solution.

(a) (b) Figure 10. SEM backscattered(a) electron images of alloy substrates (in(b contact) with patina) after an‐ odic polarization (OCP+ 1.5 V) in 0.1 mol/L NaCl solution: (a) of TB alloy (×100); (b) ZB (x50) (D: FigureFigure 10. SEM 10. SEMbackscattered backscattered electron electron images images of alloy of alloy substrates substrates (in contact (in contact with with patina) patina) after after an‐ anodic odic polarizationdendrite, I: (OCP+interdendritic 1.5 V) in area). 0.1 mol/L NaCl solution: (a) of TB alloy (×100); (b) ZB (x50) (D: polarization (OCP+ 1.5 V) in 0.1 mol/L NaCl solution: (a) of TB alloy (×100); (b) ZB (x50) (D: dendrite, dendrite, I: interdendritic area). I: interdendritic TB Casting area).  •TB CastingTBThe Casting alloying element concentration profile from the metal matrix towards pat‐ Theina/electrolyte alloying element interface concentration testify to the copperprofile depletionfrom the (decuprification),metal matrix towards which patreaches‐ a The alloying element concentration profile from the metal matrix towards patina/electrolyte ina/electrolyteminimum interface at the internal testify topatina the copper surface. depletion This tendency (decuprification), is partially whichreversed reaches towards a ex‐ interface testify to the copper depletion (decuprification), which reaches a minimum at the minimumternal at patina the internal surfaces patina (Figure surface. 11). At This high tendency anodic potentials,is partially a reversed steady state towards is reached ex‐ on internal patina surface. This tendency is partially reversed towards external patina surfaces both bronze and brass in NaCl solution (Figure 8), so the formation of a thick and compact ternal(Figure patina 11 surfaces). At high (Figure anodic 11). potentials, At high anodic a steady potentials, state is reached a steady on state both is bronze reached and on brass both bronzefilm is andpossible. brass inAt NaCl the solutionsame time, (Figure the 8),electrolyte so the formation enrichment of a thickwith anddissolved compact species in NaCl− solution2− (Figure8), so the formation of a thick and compact film is possible. At film is(CuCl possible.2 , CuCl At3 the etc.) same and thetime, local the pH electrolyte increase atenrichment the patina/electrolyte with dissolved interface− species 2allows− a the same time, the electrolyte enrichment with dissolved species (CuCl2 , CuCl3 etc.) partial− redeposition2− of amorphous or nanocrystalline Cu products. The internal patina (CuCland2 , theCuCl local3 etc.) pH increaseand the local at the pH patina/electrolyte increase at the patina/electrolyte interface allows a partialinterface redeposition allows a of surface is characterized by an intense Sn enrichment. The outer Sn‐deficient corrosion lay‐ partialamorphous redeposition or nanocrystalline of amorphous Cuor products.nanocrystalline The internal Cu products. patina surfaceThe internal is characterized patina ers are considered as a mixture of partially dissolved corrosion layers (formed at the early surfaceby anis characterized intense Sn enrichment. by an intense The Sn outer enrichment. Sn-deficient The outer corrosion Sn‐deficient layers are corrosion considered lay‐ as a polarization stages) and precipitated Cu compounds from the solution. The findings of ers aremixture considered of partially as a mixture dissolved of partially corrosion dissolved layers (formed corrosion at layers the early (formed polarization at the early stages) polarizationandthis precipitated study stages) are inand Cu good compoundsprecipitated agreement from Cu with compounds the the solution. general from The framework findingsthe solution. ofof thiscorrosion The study findings aremechanisms in of good this agreementstudyreported are in with [36],good the where agreement general a very framework with similar the electrochemicalgeneral of corrosion framework mechanisms setup of atcorrosion a reportedslow scan mechanisms in rate [36 ],was where used reportedafor very thein similar [36], study where electrochemicalof anodic a very dissolution similar setup electrochemical of at a a homogeneous slow scan setup rate annealed wasat a usedslow α‐ forscanbronze the rate study (10%was of usedwt. anodic Sn) in for thedissolution0.1 study mol/L of NaCl. anodic of a homogeneous A dissolution further examination of annealed a homogeneous ofα -bronzethe EDS annealed (10%data allows wt. α‐ Sn)bronze for in the 0.1 (10% correlation mol/L wt. Sn) NaCl. inof den A ‐ 0.1 mol/Lfurtherdritic NaCl. segregation examination A further structure of examination the EDS with data the of allowsthedistribution EDS for data the of allows correlation Cu and for Sn the ofin correlation dendriticTB corrosion segregationof denstratifica‐ ‐ driticstructure tion.segregation On withthe basestructure the metal distribution with in contact the of distribution Cu with and patina, Sn of in Cu TB andin corrosion dendritic Sn in TB stratification. areas corrosion is preferentially stratifica On the‐ base oxi‐ tion.metal Ondized. the in Therefore,base contact metal with thein patina, contactevolved Cuwith corrosion in patina, dendritic products Cu areasin dendritic on is preferentiallythe dendriticareas is preferentially oxidized.network are Therefore, oxienriched‐ dized.in Therefore, Cu, and the the interdendritic evolved corrosion areas areproducts enriched on inthe Sn dendritic (Figures network11 and 12). are enriched in Cu, and the interdendritic areas are enriched in Sn (Figures 11 and 12).

Corros. Mater. Degrad. 2021, 2 240

Corros. Mater. Degrad. 2021, 2, FOR PEER REVIEW 15 Corros. Mater. Degrad. 2021, 2, FOR PEERthe evolvedREVIEW corrosion products on the dendritic network are enriched in Cu, and the 15

interdendritic areas are enriched in Sn (Figures 11 and 12).

FigureFigure 11. 11.% % Cu Cu atomic atomic concentration concentration proﬁle profile of the of corrosionthe corrosion layer layer stratiﬁcation stratification in dendritic in dendritic and and interdendriticinterdendritic areas areas after after TB TB anodic anodic polarization polarization in 0.1in 0.1 mol/L mol/L NaCl—Based NaCl—Based on on EDS EDS elemental elemental chemicalchemical analyses analyses data data (fullframe (fullframe and and spot spot analyses). analyses).

Figure 12. % Sn atomic concentration profile of the corrosion layer stratification in dendritic and FigureFigure 12. 12.% % Sn Sn atomic atomic concentration concentration proﬁle profile of the of corrosionthe corrosion layer layer stratiﬁcation stratification in dendritic in dendritic and and interdendritic areas after TB anodic polarization in 0.1 mol/L NaCl—Based on EDS elemental interdendriticinterdendritic areas areas after after TB TB anodic anodic polarization polarization in 0.1in 0.1 mol/L mol/L NaCl—Based NaCl—Based on on EDS EDS elemental elemental chemical analyses data (fullframe and spot analyses). chemicalchemical analyses analyses data data (fullframe (fullframe and and spot spot analyses). analyses).

AtAt an an intermediate intermediate depth depth of corrosion of corrosion layers, layers, the decupriﬁcation the decuprification process process develops develops fasterfaster on on products products epitaxially epitaxially grown grown on dendrites. on dendrites. At the externalAt the external patina surface, patina the surface, distri- the dis‐ faster on products epitaxially grown on dendrites.− At the external patina surface, the dis‐ bution of both Cu and Sn is balanced. Very high Cl concentration,− reaching approximately tribution of both Cu and Sn is balanced. Very high Cl− concentration, reaching approxi‐ 41% at the external and 27% at the internal patina surface, and low levels (14.5%) of oxygen mately 41% at the external and 27% at the internal patina surface, and low levels (14.5%) speciesof oxygen (i.e., species oxides, hydroxides(i.e., oxides, and hydroxides oxyhydroxides) and oxyhydroxides) are observed. The are chloride observed. diffusion The chloride gradientof oxygen towards species the (i.e., alloy oxides, matrix hydroxides is depicted in and Figure oxyhydroxides) 13. are observed. The chloride diffusion gradient towards the alloy matrix is depicted in Figure 13. Focusing on the relative distribution of oxygen and chloride species at the internal patina surface, it can be deduced that the corrosion compounds on Cu‐enriched are char‐ acterized by a higher Cl/O atomic ratio—equal to 1.33—(Table 5), while the corrosion com‐ pounds on Sn‐enriched interdendritic areas are dominated by oxygen species—exhibiting a Cl/O ratio of 0.75 (Table 5).

Corros.Corros. Mater. Mater. Degrad. Degrad. 20212021,, 2, FOR PEER REVIEW 241 16

40 Cl_average Cl_dendrites 35 Cl_interdendritic 30

% Atomic Concentration 10

0 1_metal 1_patina (inner) 2_interm thickness 3_patina(outer)

Level in Corrosion Stratification

FigureFigure 13. 13.% % Cl Cl atomic atomic concentration concentration proﬁle profile of the of the corrosion corrosion layer layer stratiﬁcation stratification in dendritic in dendritic and and interdendriticinterdendritic areas areas after after TB TB anodic anodic polarization polarization in in 0.1 0.1 mol/L mol/L NaCl—Based NaCl—Based on on EDS EDS elemental elemental chemicalchemical analyses analyses data data (fullframe (fullframe and and spot spot analyses). analyses).

TableFocusing 5. Normalized on the % relative atomic distribution elemental distribution of oxygen andof O chloride and Cl (corresponding species at the internal to O2−, Cl− ani‐ patinaonic species) surface, detected it can be by deduced EDS in thatthe electrochemical the corrosion compounds patinas of onTB Cu-enrichedand ZB after areanodic charac- polariza‐ terizedtion in 0.1 by amol/L higher NaCl. Cl/O Average atomic values ratio—equal of indicative to 1.33—(Table analyses5 conducted), while the on corrosion both the com- internal sur‐ poundsface in contact on Sn-enriched with the interdendriticalloy and the outer areas surface are dominated of the patina by oxygen (interface species—exhibiting with electrolyte). a Cl/O ratio of 0.75 (Table5). TB ZB 2− − Table 5. Normalized % atomic elemental distributionInterface of O with and Cl Alloy (corresponding to O , Cl anionic species) detected by Element Interface with Interface with Interface with EDS in the electrochemical patinas of TB andDendritic ZB after anodic Interdendritic polarization in 0.1 mol/L NaCl. Average values of indicative analyses conducted on both the internal surface in contact with the alloy and theElectrolyte outer surface of theAlloy patina (interfaceElectrolyte with electrolyte). Network Areas O 43.0 57.1 26.3 38.6 13.1 Cl TB57.0 42.9 73.7 ZB61.4 86.9 Element InterfaceCl/O with Alloy 1.33 0.75Interface with 2.80Interface with 1.59Interface with 6.64 Dendritic Network Interdendritic Areas Electrolyte Alloy Electrolyte O 43.0More specifically, 57.1 local EDS 26.3spot analyses confirm 38.6 that Sn‐enriched 13.1 interdendritic Cl 57.0areas contain three 42.9 times higher oxygen 73.7 and significantly 61.4 lower chloride 86.9 levels (26%) com‐ Cl/O 1.33 0.75 2.80 1.59 6.64 pared to the average atomic concentration acquired from low magnification fullframe analyses. More specifically, local EDS spot analyses confirm that Sn-enriched interdendritic areas  containZB three Casting times higher oxygen and significantly lower chloride levels (26%) compared to theThe average determination atomic concentration of ZB bulk acquired patina fromchemical low magnification composition fullframe and the preliminary analyses. char‐ •acterizationZB Casting of local corrosion features (induced by macro‐ and micro‐segregation) by OM have already been presented in [29]. In this work, a more detailed analysis of the dealloy‐ The determination of ZB bulk patina chemical composition and the preliminary characterizationing mechanisms of on local dendritic corrosion microstructure features (induced is undertaken, by macro- and including micro-segregation) data re‐evaluation bywhere OM havenecessary. already The been dealloying presented process in [29]. Inof thisthe work,ZB alloy a more is initiated detailed analysisby a fast of selective thedissolution dealloying of mechanismsZn (dezincification) on dendritic towards microstructure the patina/electrolyte is undertaken, interface. including At data the internal re-evaluationpatina surface, where Zn necessary. is completely The dealloying eliminated, process while of theat the ZB alloyexternal is initiated patina by surface a fast a very selectivelow concentration dissolution ofof Zn Zn (dezincification) has been left undissolved towards the patina/electrolyte (Figure 14). interface. At the internal patina surface, Zn is completely eliminated, while at the external patina surface a very low concentration of Zn has been left undissolved (Figure 14).

Corros. Mater. Degrad. 2021, 2, FOR PEER REVIEW 17

Corros.Corros. Mater. Mater. Degrad. Degrad. 20212021, ,22, FOR PEER REVIEW 242 17

16 16 14 Zn_average 14 12 Zn_averageZn_dendrites Zn_interdendritic 12 Zn_dendrites 10 Zn_interdendritic 10 8 8 6 6 4 % Atomic Concentration 4

% Atomic Concentration 2 2 0 0 0_Alloy matrix 1_metal 1_patina (inner) 2_patina(outer) 0_Alloy matrix 1_metal 1_patina (inner) 2_patina(outer) Level in Corrosion Stratification Level in Corrosion Stratification Figure 14. % Zn atomic concentration profile of the corrosion layer stratification in dendritic and Figureinterdendritic 14. 14.% % Zn Zn areas atomic atomic after concentration concentration ZB anodic profile polarization profile of the of the corrosion in corrosion0.1 mol/L layer layerNaCl—Based stratification stratification in on dendritic EDS in dendritic elemental and and interdendriticchemical analyses areas areas data after after (fullframe ZB ZB anodic anodic and polarization polarization spot analyses). in in 0.1 0.1 mol/L mol/L NaCl—Based NaCl—Based on on EDS EDS elemental elemental chemical analyses analyses data data (fullframe (fullframe and and spot spot analyses). analyses). At the same time, Cu is also being dissolved at a slower rate. The general decuprifi‐ At the same time, Cu is also being dissolved at a slower rate. The general decuprifica- cationAt profile the same of thetime, corrosion Cu is also stratification being dissolved reveals at aa slowergradual rate. Cu Thedissolution general fromdecuprifi alloy‐ tion profile of the corrosion stratification reveals a gradual Cu dissolution from alloy matrix cation profile of the corrosion stratification reveals a gradual Cu dissolution from alloy towardsmatrix towards internal patina internal surface, patina followed surface, by followed a small increase by a small at the increase external patinaat the surfaceexternal patina (Figurematrixsurface towards15 (Figure). As a consequence,internal15). As apatina consequence, an surface, extensive followedan enrichment extensive by ina enrichmentsmall Sn is observedincrease in atSn thethe is internal observedexternal patina at the patinasurfaceinternal surface (Figure patina (Figure surface15). As 16 ).(Figurea Asconsequence, with 16). the As TB with an alloy, extensive the the TB increase alloy, enrichment in the Cu increase concentration in Sn in is Cu observed atconcentration the at the internalexternalat the external patina patina surface patinasurface could surface (Figure be justified could 16). As be by with justified the deceleration the TBby alloy,the ofdeceleration passivethe increase layer of dissolutionin passive Cu concentration layer at disso‐ athighlution the anodic external at high potentials anodicpatina and surfacepotentials the redeposition could and be the justified ofredeposition dissolved by the species. ofdeceleration dissolved species. of passive layer disso‐ lution at high anodic potentials and the redeposition of dissolved species.

100

10090 Cu_average Cu_averageCu_dendrites 9080 Cu_dendritesCu_interdendritic 8070 Cu_interdendritic 7060

6050 5040

4030 3020 % Atomic Concentration 20 % Atomic Concentration 10

100 0 0_Alloy matrix 1_metal 1_patina (inner) 2_patina(outer) 0_Alloy matrix 1_metal 1_patina (inner) 2_patina(outer) Level in Corrosion Stratification Level in Corrosion Stratification FigureFigure 15. 15.% % Cu Cu atomic atomic concentration concentration proﬁle profile of the of corrosionthe corrosion layer layer stratiﬁcation stratification in dendritic in dendritic and and Figureinterdendriticinterdendritic 15. % Cu areas areas atomic after after concentration ZB ZB anodic anodic polarization polarization profile of in the 0.1in corrosion0.1 mol/L mol/L NaCl—Based layerNaCl—Based stratification on on EDS EDS in elemental dendritic elemental and interdendriticchemicalchemical analyses analyses areas data data after (fullframe (fullframe ZB anodic and and spot polarization spot analyses). analyses). in 0.1 mol/L NaCl—Based on EDS elemental chemical analyses data (fullframe and spot analyses). At the metal/patina interface, a faster propagation of Zn dissolution in the dendritic networkAt the is observed.metal/patina Copper, interface, on the a faster other propagationhand, appears of toZn be dissolution dissolved inat the samedendritic rate networkin both anodic is observed. and cathodic Copper, sites.on the The other overall hand, Cu appears distribution to be dissolved gradient inat thedendritic same rateand ininterdendritic both anodic areasand cathodic of the alloy sites. (in The contact overall with Cu patina) distribution remains gradient unchanged. in dendritic Dezincifica and‐ interdendritiction and decuprification areas of the leave alloy behind(in contact α‐dendrites with patina) enriched remains in Sn unchanged. compared Dezincifica to the alloy‐ tionmatrix. and The decuprification internal patina leave surface behind appears α‐dendrites chemically enriched uniform in Sn with compared no local to corrosionthe alloy matrix. The internal patina surface appears chemically uniform with no local corrosion

Corros. Mater. Degrad. 2021, 2, FOR PEER REVIEW 18

Corros. Mater. Degrad. 2021, 2 243 patterns, indicating that, at high potential anodic dissolution of Zn and Cu through the corrosion film, steady‐state conditions are reached.

FigureFigure 16. 16.% % Sn Sn atomic atomic concentration concentration proﬁle profile of the of the corrosion corrosion layer layer stratiﬁcation stratification in dendritic in dendritic and and interdendriticinterdendritic areas areas after after ZB ZB anodic anodic polarization polarization in in 0.1 0.1 mol/L mol/L NaCl—Based NaCl—Based on on EDS EDS elemental elemental chemicalchemical analyses analyses data data (fullframe (fullframe and and spot spot analyses). analyses).

AtThe the relative metal/patina fractions interface, of oxygen a faster and propagation chloride ofspecies Zn dissolution at the two in thesides dendritic of the electro‐ networkchemical is patina observed. are Copper, given onin Table the other 5. hand,Although appears the to absolute be dissolved chloride at the concentration same rate de‐ intected both on anodic the external and cathodic surface sites. of The the overallelectrochemical Cu distribution TB and gradient ZB patinas in dendritic was roughly and equal interdendritic areas of the alloy (in contact with patina) remains unchanged. Dezincification (40.8% for TB versus 43.4%for ZB), the Cl/O atomic ratio observed in ZB (6.64) is nearly and decuprification leave behind α-dendrites enriched in Sn compared to the alloy matrix. The2.5 times internal higher patina compared surface appears to the chemically TB alloy (see uniform Table with 5). no local corrosion patterns, indicating that, at high potential anodic dissolution of Zn and Cu through the corrosion film,4. Discussion steady-state conditions are reached. TheIn previous relative fractions publications, of oxygen the and electrochemical chloride species behaviour at the two of sides a ternary of the electro-Cu‐Sn‐Pb (TB) chemicaland a quaternary patina are givenCu‐Zn in‐ TableSn‐Pb5. (ZB) Although alloy the in absolutevarious chloride chloride concentration‐containing detectedcorrosive media on the external surface of the electrochemical TB and ZB patinas was roughly equal (40.8% have been reported [28,29]. The current research article is dedicated to the evaluation of for TB versus 43.4%for ZB), the Cl/O atomic ratio observed in ZB (6.64) is nearly 2.5 times highersurface compared and interface to the TBchemical alloy (see analyses Table5). indicative of corrosion topography and stratigra‐ phy after these potentiodynamic tests. This systematic approach enables, at a basic level, 4.the Discussion assessment of macro‐ and especially micro‐segregation influence in corrosion mecha‐ nismsIn of previous archaeometallurgical publications, the copper electrochemical alloy castings. behaviour of a ternary Cu-Sn-Pb (TB) and aThe quaternary chemical Cu-Zn-Sn-Pb characterization (ZB) alloy of in the various TB electrochemical chloride-containing patinas corrosive in the media first experi‐ havemental been section reported showed [28,29 ].that The the current overall research chemical article composition is dedicated toof thethe evaluation produced of corrosion surfaceproducts and mixture interface in chemical all synthetic analyses corrosive indicative media of corrosion has many topography common and stratigraphyconstituents, as can after these potentiodynamic tests. This systematic approach enables, at a basic level, the be deduced by the interpretation of the three FTIR spectra. Apparently, chlorides domi‐ assessment of macro- and especially micro-segregation influence in corrosion mechanisms ofnate archaeometallurgical the corrosion reactions copper alloyin all castings. media, even in AAPM electrolyte—which is a mixture 2−) of corrosiveThe chemical agents, characterization with sulphates of the (SO TB4 electrochemical being the primary patinas ionic in the species. first experi- Knowing the mentaldifferences section in showedphysicochemical that the overall properties chemical of compositionthe three chloride of the produced‐containing corrosion media (similar productspH but different mixture in ionic all synthetic conductivities corrosive and media concentrations has many common of aggressive constituents, Cl as− and can other be ionic deducedspecies) by reported the interpretation in [28], this of the observation three FTIR spectra.seems contradictory. Apparently, chlorides This dominatefact offers the a motiva‐ corrosiontion to evaluate reactions the in all benefits media, even and in limitations AAPM electrolyte—which of an experimental is a mixture methodology of corrosive based on 2−) agents,anodic with polarization sulphates (SOtests.4 Inbeing the first the primary experimental ionic species. section, Knowing one type the differencesof metal substrate in is physicochemicalemployed and propertiesthe exposed of the surface three chloride-containing is uniformly wetted media by (similar an electrolyte pH but different layer of infinite ionic conductivities and concentrations of aggressive Cl− and other ionic species) reported thickness. The most important kinetic parameters are being controlled (application of the in [28], this observation seems contradictory. This fact offers a motivation to evaluate the benefitssame potential and limitations scan rate of that an experimental imposes equal methodology charge transfer based onrate). anodic The polarization polarization occurs at roughly the same potential domain. Consequently, it is reasonable that the fundamental electrochemical reactions promoted in these three systems are very similar. However, the

Corros. Mater. Degrad. 2021, 2 244

tests. In the first experimental section, one type of metal substrate is employed and the exposed surface is uniformly wetted by an electrolyte layer of infinite thickness. The most important kinetic parameters are being controlled (application of the same potential scan rate that imposes equal charge transfer rate). The polarization occurs at roughly the same potential domain. Consequently, it is reasonable that the fundamental electrochemical reactions promoted in these three systems are very similar. However, the corrosion layer stratification and the local topography is unique for each corrosion system (as has been demonstrated by OM and EDS analyses at the alloy-patina interface). The results of the second experiment section point out that slow potentiodynamic scans proved to be the most suitable approach to study the electrochemical dissolution and leaching processes of bronze in total immersion conditions. On the subject of experimental methodology design and evaluation, it can be con- cluded that the electrochemical potentiodynamic tests in soil filtrate and synthetic atmospheric electrolytes cannot reproduce or simulate the natural corrosion processes in these media. The contribution of an electrochemical setup of this type could be in the investigation of (i) the general trends of dealloying mechanisms in the presence of various corrosive species, (ii) the function of local galvanic cells in a well-studied metal topography, and (iii) the redox reactions at the interfaces of metal/corrosion layer/electrolyte. The most important findings pointed out by the accelerated corrosion through electrochemical potentiodynamic tests in chloride-media (1st and 2nd experimental section) can be summarized as follows: • Macro-segregation affects the oxidation and dealloying rates at a local scale, according to the geometry of the solidification front. Disc-shaped specimens cut from rod castings exhibit lower Sn concentration and smaller grain sizes at the periphery and higher Sn and larger grains at the specimen centre. The corrosion attack initiates from the Cu-enriched edge and develops towards the centre. • The differential chemical composition of dendritic microstructure inside the grains is responsible for the action of local micro-galvanic cells where the formation of different corrosion compounds is favoured and variant dealloying rates occur. The establishment of anodic and cathodic sites means that different electrochemical actions occur locally. As a result, the nature of corrosion products is also site-specific. • In the studied α-bronze and α-brass corrosion systems, the electrochemical attack initiates from the dendrites which act as anodes, while the interdendritic areas are the cathodic sites. All initial corrosion products are epitaxially grown on the given microstructure and tend to dissolve faster. • The TB casting—exhibiting a cored dendritic microstructure—has undergone more severe corrosion in the atmospheric pollutants mixture compared to the soil filtrate and the 0.6 mol/L NaCl, after anodic polarization (OCP+ 0.6 V) in these electrolytes. The lead phases were selectively attacked by sulphates and nitrates. • The main corrosion products identified by FTIR in the electrochemical patinas (PRSF, AAPM, 0.6 mol/L NaCl) of TB casting were basic Cu(II) hydroxychlorides, basic Cu(II) hydroxycarbonates and amorphous Cu and Sn oxides. In AAPM, some additional bands corresponding to Cu(II) hydroxysulphate compounds were detected. • In the first experimental section, it was evidenced that, at moderately anodic potential ranges (OCP+ 0.6 V), the external patina surface still exhibits local corrosion features evolved on the pre-existing dendritic structure chemical inhomogeneity. However, the electrochemical and chemical reactions at the patina/electrolyte interface tend to eliminate or counter-balance the inhomogeneous distribution of alloying elements. Thus, at high anodic potentials (OCP+ 1.5 V), the external surface of TB and ZB electrochemical patinas exhibits uniform chemical composition. This was confirmed by the characterization of patinas, presented in the second experimental section. • The stratification of electrochemical patinas produced on TB and ZB castings by slow anodic polarization in 0.1 mol/L NaCl reveals a decuprification profile for TB and synchronous dezincification and decuprification for ZB. Corros. Mater. Degrad. 2021, 2 245

• The ternary bronze casting (TB) exhibits selective oxidation of the Cu-enriched dendrites and higher decuprification rate in these areas. The corroded interdendritic areas in ternary bronze remain Sn-enriched in respect to the average of the particular surface, and the corrosion compounds exhibit a Cl/O atomic ratio of 0.75, lower than the value calculated for the dendrite network (1.33), which is mainly attacked by chlorides. The Cl/O ratio of the external patina surface is determined as 2.80. • The quaternary brass casting (ZB) exhibits faster dealloying processes under the same conditions. Selective attack of Zn initiates from the dendritic network. Dezincification progresses faster at these areas, as indicated by Zn elimination from the internal corrosion layers, and is further assisted by the higher solubility of the Zn chloride species. Decuprification rate is uniform in all segregated areas. The two processes at the alloy/patina interface leave behind a metal surface, where α-dendrites are enriched in Sn compared to the alloy matrix. Both patina interfaces (with metal and electrolyte) exhibit a higher Cl/O atomic ratio compared to the Cu-Sn-Pb alloy—the calculated values are 1.59 and 6.64, respectively. • On the external patina surfaces of both TB and ZB castings, after slow anodic polarization (OCP+ 1.5 V) in 0.1 mol/L NaCl, the uneven elemental distribution is balanced by dissolution redox reactions and redeposition of Cu complexes. At this point, it is necessary to clarify the use of the terms ‘decuprification’ and ‘dezincification’. These terms, when used for the interpretation of accelerated corrosion processes (induced by wet electrochemistry potentiodynamic techniques), clearly refer to the selective anodic dissolution of these elements. The morphology of dealloyed surfaces should not be compared with porous metal structures. The latter structures are formed by gradual solid-state diffusion processes that govern natural long-term corrosion phenomena. A particularly interesting future work would be to extend the investigation of the impact caused by several microstructural features (encountered in archaeometallurgical bronze and brass) in their corrosion mechanisms. It is necessary that data produced by accelerated corrosion protocols will be compared with data from multiple natural corrosion environments for further validation.

Author Contributions: Conceptualization, O.P.; methodology, O.P.; formal analysis, O.P.; resources, O.P.; data curation, O.P.; writing—original draft preparation, O.P.; review and editing, P.V.; supervi- sion, P.V. All authors have read and agreed to the published version of the manuscript. Funding: This research received no external funding. Institutional Review Board Statement: Not applicable. Informed Consent Statement: Not applicable. Data Availability Statement: The data presented in this study are part of a PhD research and are available on request from the corresponding author, until the dissertation is made public. Afterwards, the data will be available in a publicly accessible repository. Conﬂicts of Interest: The authors declare no conﬂict of interest.

References 1. Scott, D.A. Metallography and Microstructure of Ancient and Historic Metals; The Getty Conservation Institute: Singapore, 1991; pp. 5–16, 25–29. 2. Scott, D.A.; Schwab, R. Metallography in Archaeology and Art; Springer: Berlin/Heidelberg, Germany, 2019; pp. 5–17. ISBN 978-3-030-11265-3. [CrossRef] 3. Nord, A.G.; Mattsson, E.; Tronner, K. Factors Inﬂuencing the Long-term Corrosion of Bronze Artefacts in Soil. Prot. Met. 2005, 41, 339–346. [CrossRef] 4. Robbiola, L.; Blengino, J.-M.; Fiaud, C. Morphology and mechanisms of formation of natural patinas on archaeological Cu–Sn alloys. Corros. Sci. 1998, 40, 2083–2111. [CrossRef] 5. MacLeod, I.D. Identiﬁcation of Corrosion Products on Non-Ferrous Metal Artifacts Recovered from Shipwrecks. Stud. Conserv. 1991, 36, 222–234. 6. Scott, D.A. Copper and Bronze in Art/Corrosion, Colorants, Conservation; Getty Conservation Institute: Los Angeles, CA, USA, 2002; pp. 43–65. ISBN 0-89236-638-9. Corros. Mater. Degrad. 2021, 2 246

7. Di Turo, F.; Coletti, F.; De Vito, C. Investigations on alloy-burial environment interaction of archaeological bronze coins. Microchem. J. 2020, 157, 104882. [CrossRef] 8. Ingo, G.M.; Riccucci, C.; Guida, G.; Pascicci, M.; Giuliani, G.; Messina, E.; Fierro, G. Micro-chemical investigation of corrosion products naturally grown on archaeological Cu-based artefacts retrieved from the Mediterranean Sea. Appl. Surf. Sci. 2019, 470, 695–706. [CrossRef] 9. Oudbashi, O. Multianalytical study of corrosion layers in some archaeological copper alloy artefacts. Surf. Interface Anal. 2015, 47, 1133–1147. [CrossRef] 10. Papadopoulou, O.; Vassiliou, P.; Grassini, S.; Angelini, E.; Gouda, V. Soil-induced corrosion of ancient Roman brass—A case study. Mater. Corros. 2016, 67, 160–169. [CrossRef] 11. Quaranta, M.; Catelli, E.; Prati, S.; Sciutto, G.; Mazzeo, R. Chinese archaeological artefacts: Microstructure and corrosion behaviour of high-leaded bronzes. J. Cult. Herit. 2014, 15, 283–291. [CrossRef] 12. Chiavari, C.; Colledan, A.; Frignani, A.; Brunoro, G. Corrosion evaluation of traditional and new bronzes. Mater. Chem. Phys. 2006, 95, 252–259. [CrossRef] 13. Hassairi, H.; Bousselmi, L.; Triki, E. Bronze degradation processes in simulating archaeological soil media. J. Solid State Electrochem. 2010, 14, 393–401. [CrossRef] 14. dos Santos, L.M.M.; Lemos Salta, M.M.; Fonseca, T.E. The electrochemical behaviour of bronze in synthetic seawater. J. Solid State Electrochem. 2007, 11, 259–266. [CrossRef] 15. Lins, A.; Power, T. Corrosion of Bronze Monuments in Urban Sites. In Ancient & Historic Metals—Conservation and Scientific Research; Proceedings of a Symposium organized by the J. Paul Getty Museum; Scott, D.A., Podany, J., Considine, B.B., Eds.; The Getty Conservation Institute: Singapore, 1991; pp. 119–151. ISBN 0-89236-231-6. 16. Zhou, P.; Ogle, K. The Corrosion of Copper and Copper Alloys. In Encyclopedia of Interfacial Chemistry; Wandelt, K., Ed.; Elsevier: Amsterdam, The Netherlands, 2018; pp. 478–489. [CrossRef] 17. Bozzini, B.; Alemán, B.; Amati, M.; Boniardi, M.; Caramia, V.; Giovannelli, G.; Gregoratti, L.; Kazemian Abyaneh, M. Novel insight into bronze disease gained by synchrotron-based photoelectron spectro-microscopy, in support of electrochemical treatment strategies. Stud. Conserv. 2017, 62, 465–473. [CrossRef] 18. Chang, T.; Maltseva, A.; Volovitch, P.; Wallinder, I.O.; Leygraf, C. A mechanistic study of stratified patina evolution on Sn-bronze in chloride-rich atmospheres. Corros. Sci. 2020, 166, 108477. [CrossRef] 19. Campanella, L.; Colacicchi Alessandri, O.; Ferretti, M.; Plattner, S.H. The effect of tin on dezincification of archaeological copper alloys. Corros. Sci. 2009, 51, 2183–2191. [CrossRef] 20. Brunoro, G.; Laguzzi, G.; Luvidi, L.; Chiavari, C. Corrosion evaluation of artificially aged 6 wt-% tin bronze. Br. Corros. J. 2001, 36, 227–232. [CrossRef] 21. Chiavari, C.; Rahmouni, K.; Takenouti, H.; Joiret, S.; Vermaut, P.; Robbiola, L. Composition and electrochemical properties of natural patinas of outdoor bronze monuments. Electrochim. Acta 2007, 52, 7760–7769. [CrossRef] 22. Ospitali, F.; Chiavari, C.; Martini, C.; Bernardi, E.; Passarini, F.; Robbiola, L. The characterization of Sn-based corrosion products in ancient bronzes: A Raman approach. J. Raman Spectrosc. 2012, 43, 1596–1603. [CrossRef] 23. Masi, G.; Esvan, J.; Josse, C.; Chiavari, C.; Bernardi, E.; Martini, C.; Bignozzi, M.C.; Gartner, N.; Kosec, T.; Robbiola, L. Characterization of typical patinas simulating bronze corrosion in outdoor conditions. Mater. Chem. Phys. 2017, 200, 308–321. [CrossRef] 24. EFESTUS-Tailored Strategies for the Conservation and Restoration of Archaeological Value Cu-Based Artefacts from Mediter- ranean Countries, Research Project Funded under FP5 Programme INCO 2, Grant Agreement ID: ICA3-CT-2002-10030, 2003–2005. Available online: https://cordis.europa.eu/project/id/ICA3-CT-2002-10030 (accessed on 20 April 2021). 25. Casaletto, M.P.; De Caro, T.; Ingo, G.M.; Riccucci, C. Production of reference “ancient” Cu-based alloys and their accelerated degradation methods. Appl. Phys. A 2006, 83, 617–622. [CrossRef] 26. Vander Voort, G.F. Color Etching. In ASM Metals Handbook; ASM International: Almere, The Netherlands, 2020; Volume 9, pp. 229–233. 27. Eleftheriadis, K.; Balis, D.; Ziomas, I.C.; Colbeck, I.; Manalis, N. Atmospheric aerosol and gaseous species in Athens, Greece. Atmos. Environ. 1998, 32, 2183–2191. [CrossRef] 28. Papadopoulou, O.; Vassiliou, P. Electrochemical behaviour of ancient-like ternary Cu alloys in aqueous chloride media. In Proceedings of the Joint European Corrosion Congress 2017, EUROCORR 2017 and 20th International Corrosion Congress and Process Safety Congress 2017, Prague, Czech Republic, 3–7 September 2017. 29. Papadopoulou, O.; Delagrammatikas, M.; Vassiliou, P.; Grassini, S.; Angelini, E.; Gouda, V. Surface and interface investigation of electrochemically induced corrosion on a quaternary bronze. Surf. Interface Anal. 2014, 46, 771–775. [CrossRef] 30. Stoilova, D.; Vassileva, V. Infrared spectroscopic study of solids in the Cu2(OH)3Cl (paratacamite)—Zn5(OH)8Cl2·H2O (si- monkolleite) series. Comptes Rendus Académie Bulg. Sci. 2002, 55, 51–54. 31. Braithwaite, R.S.; Mereiter, K.; Paar, W.H.; Clark, A.M. Herbertsmithite, Cu3Zn(OH)6Cl2, a new species, and the definition of paratacamite. Mineral. Mag. 2004, 68, 527–539. [CrossRef] 32. Martens, W.; Frost, R.L.; Williams, P.A. Raman and infrared spectroscopic study of the basic copper chloride minerals—implications for the study of the copper and brass corrosion and “bronze disease”. Neues Jahrb. Mineral. Abh. 2003, 178, 197–215. [CrossRef] Corros. Mater. Degrad. 2021, 2 247

33. Robbiola, L.; Tran, T.T.M.; Dubot, P.; Majerus, O.; Rahmouni, K. Characterisation of anodic layers on Cu–10Sn bronze (RDE) in aerated NaCl solution. Corros. Sci. 2008, 50, 2205–2215. [CrossRef] 34. Di Carlo, G.; Giuliania, C.; Riccucci, C.; Pascucci, M.; Messina, E.; Fierro, G.; Lavorgna, M.; Ingo, G.M. Artiﬁcial patina formation onto copper-based alloys: Chloride and sulphate induced corrosion processes. Appl. Surf. Sci. 2017, 421, 120–127. [CrossRef] 35. Malvault, J.Y.; Lopitaux, J.; Delahaye, D.; Lenglet, M. Cathodic reduction and infrared reﬂectance spectroscopy of basic copper(II) salts on copper substrate. J. Appl. Electrochem. 1995, 25, 841–845. [CrossRef] 36. Robbiola, L.; Hurtel, L.-P. Standard Nature of the Passive Layers of Buried Archaeological Bronze—The Example of Two Roman Half-length Portraits. In Proceedings of the METAL 95: International Conference on Metals Conservation, Semur-en-Auxois, France, 25–28 September 1995; pp. 109–117. Available online: https://hal.archives-ouvertes.fr/hal-00975718 (accessed on 1 March 2021). 37. Jouen, S.; Lefez, B.; Sougrati, T.; Hannoyer, B. Fourier transform infrared spectroscopic study of abhurite Sn21O6Cl16(OH)14. Mater. Chem. Phys. 2007, 105, 189–193. [CrossRef]