Calcareous materials, like marble used in connection with cultural heritage objects such as statues and pedestals, or as wall facings on buildings, often show a brownish staining owing to contact with iron metal or iron-containing minerals in the stone. The discolouration alters the appearance of the stone, which is undesirable from an aesthetic point of view. Despite rust staining being a conspicuous phenomenon and numerous works that have dealt with the problem of removing rust stains, a simple and non-toxic method has so far been missing. This paper describes a highly efficient method for cleaning rust stains from marble by introducing the chelating amino acid cysteine in a Laponite poultice in combination with the strong reducing agent sodium dithionite.

Results


Cleaning experiments were performed on artificially discoloured samples of various types of Carrara Bianco marble and on naturally rust stained marble. To begin with, solutions of cysteine in combination with sodium dithionite and ammonium carbonate were tested by immersion of samples into the different solutions. Secondly, solutions of cysteine and sodium dithionite with and without buffering were used in a poultice consisting of Laponite? RD, Arbocel? BC1000 and CMC. The poultice was applied on three different marble types: Carrara Fabricotti, Carrara Vagli and Carrara La Piana. Thirdly, the optimized method was tested on original rust stained material of luxury marble, which has been used as wall facing, and finally in situ in Copenhagen on a larger area of The Marble Church showing rust stains due to pyrite oxidation. The cleaning results were evaluated by visual observations, cross sections, and etching of the surface by testing on high gloss marble.


Conclusion


Cleaning of iron-discoloured marble surfaces has been investigated and a new method for removal of rust stained marble has been developed. A solution of 0.1 M cysteine and 0.1 M sodium dithionite in a poultice consisting of Laponite? RD/Arbocel? BC1000/CMC = 10:10:1 has shown to be a fast, simple, cheap, and non-toxic, do-it-yourself method.


Since ancient times, white marble has been used as a popular material for sculptural artefacts such as statues, busts, and friezes as well as an architectural building material with numerous applications from flooring, wall facings, and pedestals, to columns and fountains. Although marble is a relatively stable material, the desired white surface is unfortunately prone to tarnishing when used in outdoor environments [1]. One of the major sources of tarnishing is iron. In addition to the oxidation of internal iron compounds present in stone like pyrite (FeS2) and siderite (FeCO3) [1, 2], contact with iron-rich ground water when Full Body Marble is used in, for example, garden fountains, results in severe and unsightly discolouration [3]. Another cause is the proximity to iron metal, which is oxidized by air in the presence of rain. The solubilized ions are then transported by rain onto the marble surface, resulting in rust formation [4].

The detailed mechanism for rust formation is highly complex; depending on the pH value, different species, all characterized by a brownish colour, are formed. The atmospheric corrosion of iron, regardless of the pH value of the reaction may, however, be summarized by the overall stoichiometric reaction (1) where the product FeOOH represents the generic formula for rust [5].


The general name rust consists of a variety of iron(III) oxyhydroxides or hydrated oxides of high stability and low solubility. The actual species formed depend as mentioned on the pH value and the presence of different anions [6–8]. The thermodynamic parameters and solubility products have been estimated for many of the rust species, such as ferrihydrite and α-, β- and γ-FeOOH (goethite, akaganeite and lepidocrocite). These investigations have shown that goethite defines a thermodynamic minimum of the rust system [7, 9] and the solubility product of goethite (Ksp = 10?41) is the lowest among the different rust species [7]. This means, from a thermodynamic point of view, that rust can be examined as goethite, and thus the cleaning of rust can be considered as removal of goethite.


Rust discolouration of marble is characterized by areas or stains having an orange to brownish colour, which alters the appearance of the stone. From an aesthetic point of view, the discolouration is undesirable and stone conservators and conservation scientists have therefore worked for several decades with various cleaning methods in attempts to remove rust stains from marble and calcareous stone materials [3, 10–12].

Due to the nature of the discoloration and the possibility of damaging the stone, the stain can only be removed by chemical cleaning. The current method for rust cleaning involves application of different ligands and reducing agents mixed in a poultice and placed onto the stone surface. One of the ligands most widely used is the citrate ion [10, 11, 13], though salts of other carboxylic acids, such as oxalic and tartaric acid, have also been used [10]. Other methods involve the use of fluoride [10] or EDTA [12]. A relatively new method is the use of the hexadentate ligand tpen, which, in contrast to EDTA, has a high affinity towards iron and a low affinity towards calcium [3]. This ligand has shown excellent results when tested on a discoloured marble fountain, however this method is rather expensive. The ligands are used either alone or in combination with reducing agents like thiosulfate, dithionite or polythiophene [3, 10]. Thioglycolic acid and ammonium thioglycolate have been applied in several conservation treatments of calcareous stone [12]. Thioglycolate is presumably the most efficient ligand for cleaning rust stained marble [12, 13]. However, thioglycolic acid is a toxic chemical, and is thus difficult to acquire for private stone conservators without access to a laboratory. In addition to this, a slightly violet colour may appear on the marble when cleaning with thioglycolic acid, which demands a second cleaning [12].


In this study, we have aimed to investigate and develop a new method for rust cleaning of simm marble. The focus has been on the use of cheap and commercially available chemicals. Another target was reduction of Fe(III) to Fe(II) while cleaning. Efficient removal of a slightly soluble material requires a ligand having an overall stability constant comparable to the reciprocal value of the solubility product in order to achieve a favourable equilibrium constant. Based on the solubility product of goethite, efficient removal of rust in Fe(III) stage requires a ligand having a stability constant approaching 1041, whereas removal of Fe(OH)2 only requires a stability constant of 1014. Additionally, the ligand should possess low affinity towards Ca(II) to prevent dissolution of calcite.

Introducing new chemistry for rust cleaning


In the search for an efficient method for rust cleaning, the focus has been both on a ligand showing strong complex formation with iron and weak binding to the major constituent ions in marble i.e. Ca(II) and Mg(II), as well as on the identification of a fast reducing agent able to reduce Fe(III) to Fe(II). Among the reducing chemicals, sodium dithionite (SD), Na2S2O4, has been successfully used in combination with different ligands as a dissolving agent for goethite in soil analyses [14, 15] and for removal of rust from paper [16]. Furthermore, the use of dithionite in conservation science in general is well described [17].


The standard reduction potential, e°, of dithionite in the basic solution given in Eq. (2) has been determined to ?1.12 V (vs. NHE) [15, 17] and is thereby one of the strongest reducing agents among the simple, cheap, commercial reagents. The reducing power decreases with lower pH values and using pKa2 = 7 for hydrogen sulphite the potential can be calculated to e°′ = ?0.29 V at pH = 7.


In aqueous solution dithionite partly dissociates, forming the highly reactive monomeric sulphur dioxide radical anion with the dissociation equilibrium constant K = 10?9 [18].


Even though the amount of the radical anion is relatively small and can be estimated to 10?5 M in a 0.1 M dithionite solution, the anion has shown to be the dominant reducing species in the reduction and dissolution of iron oxides [14, 15]. From biochemical experiments, the standard reduction potential of the radical anion has been determined to ?1.39 V (vs. NHE) in basic solution [18, 19], giving a calculated value e°′ = ?0.56 V at pH = 7 in accordance with experimentally determined values [18].

The reduction potential for reduction and dissolution of synthetic goethite has been calculated to e°′ = ?0.14 V (vs. NHE) at pH = 7 [20]. Using this value and either dithionite or the sulfur dioxide radical anion in the reduction and dissolution of goethite to Fe(II), the reactions can be written as in Eqs. (5), (6) with the electrochemical potentials of E°′ = +0.15 V or E°′ = +0.42 V.


Both reactions are spontaneous processes with relatively large equilibrium constants, which can be calculated to K = 105 or K = 107, respectively. From a thermodynamic point of view, dissolution of rust could be achieved by SD solutions only. However, the presence of a ligand for removal of the Fe(II) ions is preferable in order to avoid re-precipitation caused by oxidation from oxygen.

In search of a ligand useful for rust removal, a sulphide-containing species similar to thioglycolate were examined. The amino acid cysteine (cys), commonly found in natural proteins as the L-isomer, is commercially available and affordable. Cysteine forms complexes with Fe(III) and Fe(II) with high stability constants and only very weak complexes with Ca(II) and Mg(II) [21]. At the same time cysteine reacts as a reducing agent in the iron(III)-cysteine complexes with formation of colourless Fe(II)-cysteine complexes [22]. The intense violet colour known for Fe(III) complexes with ligands containing thiol groups like cys and thioglycolate [12, 22] may therefore be avoided. In addition to this, cys is also able to perform reductive dissolution of iron(III) oxyhydroxides, thereby independently having a solubilizing effect of rust [23].


Table 1 shows the stability constants of the marble constituents Ca(II), Mg(II), Fe(II) and Fe(III), with the commonly used ligands for rust cleaning i.e. citrate [24], oxalate [24], tartrate [24], edta [25], tpen [26, 27] and thioglycolate [24, 28], together with cys [21, 28]. The solubility products of CaCO3 [29], MgCO3 [29], Fe(OH)2 [29], and FeOOH [7] are also given. As seen from the constants, only edta shows affinity towards Mg(II) and Ca(II) in an order resulting in serious dissolution of MgCO3 and CaCO3, whereas the remaining ligands display relatively weak binding constants, causing little dissolution of marble itself. The stability constants of cys are similar to the values of thioglycolate, and cys possess very high affinity towards iron(III), which is even higher than for edta. Towards iron(II) the overall stability constant is of an order of magnitude close to the value for tpen, thus making cys an ideal candidate for cleaning of rust stained marble.


Reduction of Fe(III) to Fe(II) by cys is accomplished by oxidation to cystine, which is insoluble in water, causing unwanted precipitation. However, the presence of SD together with cys prevents precipitation of cystine due to the ability of dithionite to re-reduce cystine formed. The reduction potential of cys is estimated to approximately e′ = ?0.25 V at pH = 7 [22] which is higher than the potential of dithionite. In Fig. 1, the reduction reaction from cystine to cys (zwitterion form) is shown together with the acid dissociation of the thiol group, forming a cysteinate species. This anion may react as a bidentate ligand towards metal ions via the sulphur and oxygen donor atoms [22], but other coordination involving O, N and O, N, S donor atoms are also possible. The iron-cysteinate complexes are complicated and not straightforward due to redox reactions similar to those observed for the iron-thioglycolate system [22, 28, 30–32].


The pKa values of three functional groups i.e. carboxylic, thiol and protonated amino group are 1.88, 8.15 and 10.29, respectively [23]. Using the values of the first two pKa constants, pH in solution of the cys zwitterion can be estimated to pH 5. In general this pH value is too low for cleaning marble, due to acid dissolution of CaCO3 [12, 13]. The pH value can be adjusted by the addition of a base such as ammonia (NH3) or ammonia carbonate ((NH4)2CO3), and in some cases when the cleaning mixture is used in a poultice, the poultice itself can act as a buffering agent. Laponite, for example, releases OH? below its point of zero charge, which is obtained around pH = 11 and an aqueous suspension of Laponite is alkaline [33] (measurement shows pH = 9.3). Since the dissolution of goethite consumes H+ (Eqs. 5 and 6), the pH is also raised during the reaction. Considering that the oxidation of iron(II) and cys is eased with increasing pH favouring precipitation of both iron(III) oxyhydroxides and cystine, a reaction value around pH = 7 may be preferred, although pH = 9?10 is desired with respect to the solubility of calcite [3, 12].


Introducing a new poultice for rust cleaning


The chemicals used for cleaning of stained marble are commonly applied in a poultice and a wide range of poultice material has been tested and applied in stone conservation. Clay materials, such as bentonite, attapulgite and sepiolite, are widely used either alone or in combination with cellulose fibres [4, 10, 34]. Other methods use cellulose fibres alone [35, 36], MC (methyl cellulose) [37], CMC (carboxymethyl cellulose) [38], cotton pads [10, 38], and gels like glycerine [10], agar [39], agarose [40], or xanthan gum [3]. One of the newer materials used for poultices is the synthetic magnesium silicate clay Laponite? RD [41–44]. When dispersed in water, Laponite produces a colourless thixotropic gel that is easy to apply on specific areas and on vertical surfaces. The high purity of Laponite and thereby the absence of natural iron impurities means that discolouration of the marble surface from the poultices itself is avoided. In this study, Laponite? RD is mixed with cellulose fibres (Arbocel? BC1000) with dimensions of 700 × 20 μm (lenght and thickness) in order to increase the porosity, the absorbing properties and the water retention of the poultice. In addition to this, a small amount of sodium CMC (carboxymethyl cellulose, sodium salt) was also added. This resulted in better mechanical properties, increasing both the adherence and the cohesion of the poultice, making it easy to apply and remove in large pieces without crumbling. Another advantage of this poultice composition was its shrinkage properties: when drying it shrank practically only in the direction of thickness, leaving the area dimension intact. Hence a uniform cleaning from the centre to the edge of the poultice was obtained.


Three different types of white Carrara marble (Carrara Bianco): Carrara Fabricotti, Carrara Vagli and Carrara La Piana from the Carrara quarry in Italy were received. Prior to the study and the artificial discolouration, the marble samples were characterised by the European Standards for water absorption, DS/EN 13755:2008 and water absorption coefficient by capillarity, DS/EN 1925:1999. Original samples of naturally rust stained Greenlandic marble from 1937 were retrieved from the government building of The Public Guardian in Copenhagen, Denmark in connection with restoration of the building. The marble plates were used as wall facing and, when dismounted, a heavy iron discolouration was present on the backside of the plates. A high gloss polished marble of the type Carrara Bianco, Lorano was used for etching experiments.