**2. Materials and Methods**

Scientific analysis was undertaken to better understand the pigments and binders used in *Man-Eater* and the associated maquette. Both invasive and non-invasive techniques were employed, with the latter involving both scrapings and cross sections. Initial analysis with portable X-Ray Fluorescence (p-XRF) yielded information about possible pigments and extenders as well as the composition of the metals. However, in the case of *Man-Eater*, the extensive overpainting, coupled with the lack of historical narrative about said campaigns necessitated micro-invasive sampling for more fingerprint spectroscopic techniques. Samples for analysis by micro-Fourier transform infra-red (μ-FTIR) and Raman spectroscopies were taken from areas that could be representative of different layers. While μ-FTIR also gave some indication of the binder used in the paint formulation, the pigments in each layer of the cross section were conclusively identified by Raman spectroscopy in addition to scanning electron microscopy (SEM), coupled with electron dispersive spectroscopy (EDS). Cross sections from the maquette showed a single campaign of colored paint, which makes the results more definitive than those taken from *Man-Eater*, especially as it relates to binder analysis by μ-FTIR.

The complementary nature of these methods made for a rich visualization and interpretation of the original colors of Man-Eater. These results helped determine the final colors to be used in the restoration campaign. The following is a description of the techniques used.

XRF analysis was performed with a Bruker Tracer III-SDD handheld XRF instrument with a Rh excitation source and silicon drift detector, with a 5 mm diameter approximate spot size. The instrument was operated at 40 kV and 3 μA, and spectra were acquired for 120 s. Additional XRF was performed on the maquette using a Bruker Tracer 5i at 40 kV and 4.5 μA and spectra were acquired for 120 s. All the spectra were examined with the Bruker Artax 8.0 software.

Micro-μ-FTIR (μ-FTIR) analysis was carried out in transmission mode using a Nicolet iS50-μ-FTIR coupled with a Thermo Nicolet Continuum infrared microscope equipped with an MCT detector. Spectra were collected in the 4000–600 cm−<sup>1</sup> range with a 4 cm−<sup>1</sup> resolution and 128 scans and using the Thermo Scientific OMNIC 9.12 software package. Spectra were examined using the Spectral Search and Multicomponent Search tools available in the Thermo Scientific OMNIC Specta 2.0 software and IRUG spectral databases [10].

Raman spectra were collected using a Renishaw In-via Raman system, equipped with a 785 nm diode laser at powers between 0.3 to 3 mW, a 1200 lines/mm grating, and a Leica confocal microscope with a 50× LWD or 100× objective. Final spectra represent an average of five acquisitions of 10 s. Raman spectra were evaluated Spectral Search and Multicomponent Search tools available in the Thermo Scientific OMNIC Specta 2.0 software and SOPRANO [11] and UCL [12]. spectral databases.

SEM-EDS was carried out under low vacuum and using a Hitachi TM3000 microscope (Tokyo, Japan) fitted with a Bruker Xflash MIN SVE and Quantax 70 software. Images were acquired using analysis mode at 15 kV and a four-segment backscatter electron (BSE) detector. The cross sections were not coated as imaging was carried out under low vacuum; remaining gas molecules in the chamber ionize the negative buildup up on the surface of uncoated, organic samples.

Optical Microscopy was carried out using a Leica DM IRM microscope using 5× and 10× objectives.

Cross sections were embedded in BioPlastic ®, [Aldon Corp., Avon, NY, USA] a blend of polyester and methacrylate monomers in a styrene solvent, trimmed with a jeweler's saw, and dry polished with Micro-Mesh® [Micro-surface Finishing Products, Wilton, IA, USA] silicon carbide or aluminum oxide abrasives.

#### **3. Results and Discussion**

#### *3.1. Metals*

Based on XRF analysis, *Man-Eater* is made of a steel of containing manganese. Areas denuded of paint were severely rusted. In most cross sections, there are two metallic layers, a reddish layer followed by a silvery one, on top of which the first colored paint layer has been applied. SEM-EDS analysis confirmed what could be concluded visually from microscopy: a layer of rusted steel rich in iron coated with a silvery, flaky layer of aluminum. Aluminum paint functions as an anti-corrosion coat on top of a steel substrate. Due to its rapid oxidation when exposed to air, an aluminum flake paint can confer a high degree of corrosion protection on a steel substrate by forming a thin, transparent layer of aluminum oxide film. This impervious layer is self-adherent and reaches a maximum thickness of 100 Å [13]. The effect has been understood since the early 1900s and was popular in the 1920s and 1930s for automotive and structural parts. In line with Calder's practice, the original color layer is applied directly atop the protective coating of aluminum [14]. The steel rust penetrated upwards through both the aluminum paint and the first colored paint layers in many cases, suggesting that the first repaint campaign was necessitated by surface paint loss due to rusting of the steel substrate.

The maquette's metal structure is unusual because the colored flags are a different metal than the remaining black elements [15]. XRF analysis showed the colored pennants atop the vertical rods to be made of an aluminum alloy containing lead, iron, silicon, manganese, and zinc. This could suggest an effort to lighten the load on the vertical rods. The lower black pennants are an iron-based alloy whose weight serves to keep the rods vertical. The XRF spectrum of the rods showed an intense peak for copper in addition to iron. A closer examination of the dull brown patches on the black-painted rods shows that they appear to be copper-clad steel, which has application in the electric and automotive industries, for example [16,17]. This metal has been observed on rare occasions in other sculptures by Calder [15]. The rivets connecting the black rods to the colored aluminum pennants are made of a brass alloy of copper, zinc, arsenic, and possibly titanium.

#### *3.2. Yellow Paint*

The color of the sole yellow pennant in *Man-Eater* had bleached considerably, rendering it remarkably light in comparison with underlying layers visible in chips on the edges. The surface was also visibly scratched and scuffed but had the brush marks of hand application. Calder is known for hand painting many of his early painted outdoor works, and retaining that quality would have been paramount to any repainting campaign.

In the yellow cross section (Figure 3), nine individual layers of paint are clearly delineated, varying in shades from intense, to pale, to greenish. Chrome yellow (P.Y. 34; C.I. 77600), chemically a lead chromate (PbCrO4), was observed in the Raman spectra of the first eight layers. This was also confirmed by SEM-EDS (Figure 3), where Pb and Cr were recorded across those layers. Chrome yellow pigment was first synthesized in 1804 but only came into prominence when more abundant sources of chrome minerals were available [18]. Chemically, chrome yellow is available as a pure PbCrO4 or as solid solutions, with PbCrO4 and lead sulfate (PbCr1−xSxO4), in shades that range from yellow to orange (*x* < 0.1) to lemon-yellow (0.2 ≤ *x* ≤ 0.4) and pale yellow (*x* > 0.5) with increasing sulfate concentration (C.I. 77603 when coprecipitated with PbSO4) [19]. Partial replacement of the chromate in solid solutions brings about a reduction in tinctorial strength with increasing sulfate concentration but allows for the manufacture of yellows with a greenish hue. In terms of crystallography, PbCrO4 presents as monoclinic and PbSO4 as orthorhombic, and substitution of chromate ions by smaller sulfate ions leads to a compression of the monoclinic unit cell at low sulfate concentration and a change from monoclinic to orthorhombic when *x* exceeds 0.4 in a solid solution.

**Figure 3.** Cross section imaging of a sample taken from the yellow pennant in *Man-Eater*. (**a**) BSE image of the cross section and (**b**) associated EDS mapping of relevant chemical elements, of particular note is the Cd and Ti in Layer 9 and Pb and Cr in the remaining layers. (**c**) Light microscope image at 20× magnification showing all 9 layers of paint and aluminum anticorrosive paint, indicated by an arrow. © 2021 MoMA, New York, NY, USA.

This phenomenon is also observed by Raman spectroscopy (Figure 4), where shifts to higher wavenumbers of some chromate bands indicate the presence of lead sulfate [19]. The ν1(CrO4 <sup>2</sup>−) symmetric stretching mode shifts to higher energy with increasing substitution of sulfate ions into the lattice. Discrete shifts between the spectra of the yellow layers point to at least two different yellows used in the overpainting. This is evident in Layers 1, 2, and 3 in the cross section, where ν1(CrO4 <sup>2</sup>−) is at 841 cm−<sup>1</sup> in comparison with 839 cm−<sup>1</sup> in the remaining layers. The ν4/ν2(CrO4 <sup>2</sup>−) bending multiplet is also influenced by sulfate substitution and cell compression; ν4(CrO4 <sup>2</sup>−) modes for pure chrome yellow are located at 400, 376, and 357 cm−1, while those at 336 and 323 cm−<sup>1</sup> are attributable to ν2(CrO4 <sup>2</sup>−) modes. Further pointing to the presence of sulfate in the first three layers, the mode at 400 cm−<sup>1</sup> is shifted to 403 cm−<sup>1</sup> and that at 357 cm−<sup>1</sup> to 360 cm−1. A band at 970 cm−<sup>1</sup> attributed to a ν1(SO4 <sup>2</sup>−) mode [19]. The pair of Layers 2 and 3 cements the presence of a solid solution of chromates and sulfates, perhaps one that is still monoclinic with few sulfates. While this peak was not seen in layer 1 due to noise, it can be said with some certainty that a paint containing a PbCr1−xSxO4 pigment was the original color of the yellow pennant, one that was possibly lemony in hue at one time. Similarly, Layer 5 exhibits the presence of ν1(SO4 <sup>2</sup>−) from lead, barium, and calcium sulfates, based on

Raman analysis (Figure 3). Layer 5 was also rich in silicates but contained less pigment material, as indicated by the elemental distribution observed in SEM-EDS. Layer 4 consists only of PbCrO4 and BaSO4, Similarly, the color and spectra of Layers 6, 7, and 8 consist of PbCrO4 and BaSO4 based on Raman analysis (Figure 3). They could have been applied successively in one repainting campaign with drying time between coats.

**Figure 4.** Raman spectra of the first 8 layers in the cross section of a sample taken from the yellow pennant in *Man-Eater*. The slight shifts to higher wavenumbers in Layers 1, 2, 3, and 5 in both the ν1(CrO4 <sup>2</sup>−) stretching band and ν4/ν2(CrO4 <sup>2</sup>−) bending multiplet suggests the presence of a solid solution of lead chromate and lead sulfate (PbCr1−xSxO4). This is confirmed by the ν1(SO4 <sup>2</sup>−) stretching mode observed between 973 and 978 cm−<sup>1</sup> in the inset.

Curiously, the Raman spectrum of Layer 9 in the cross section showed no signs of chrome yellow, only that of the tetragonal rutile form of titanium white (TiO2) with bands at 144 (B1g), 445 (Eg), and 610 (A1g) cm−<sup>1</sup> [20]. The Raman spectrum exhibited a recently characterized luminescence emission pattern (1222, 1306, 1385, 1497, 1600, and 1686 cm−1; see Figure 5 and Figure 7) attributed to neodymium (Nd3+) ions substituting into the orthorhombic alkaline earth sulfates of titanium dioxide pigments, made through co-precipitation with barium sulfate (BaSO4) or calcium sulfate (CaSO4), where Nd3+ occurs naturally in ilmenite (FeTiO3), the source ore for Ti [21]. Observing this pattern can help with dating issues, as it was only detected in works dating from 1945–1977. These co-precipitated pigments of lower tinting strength were more prevalent in industrial paints, in particular oils and alkyds. However, the presence of both BaSO4 (988 cm<sup>−</sup>1) and CaSO4 (1017 cm<sup>−</sup>1) complicates the identification of the type of co-precipitate, where either could have been added mechanically to the pigment mixture. In turn, this makes it more difficult to establish the date of the final repainting campaign since BaSO4 and CaSO4 coprecipitates were phased out in the late 1940s and 1970s, respectively. Internal records show that, after the *Salute to Calder* exhibition closed in 1970, a repaint was considered but never executed [22]. The Nd3+ luminesce pattern in the Raman spectrum places a last repainting within the accepted range of 1945–1977.

While no yellow pigment was detected via Raman spectroscopy in Layer 9, SEM-EDS (Figure 3) showed the top layer to contain cadmium unlike the remaining layers of paint, indicating the presence of the semiconductor pigment cadmium sulfide, known as cadmium yellow (P.Y. 37; C.I. 77199). Cd was also seen in the XRF spectra taken of the yellow pennant. SEM-EDS also showed this layer to be particularly rich in magnesium silicates and silicates that are used as fillers.

**Figure 5.** Microscopy images of three cross sections (**a**–**c**) taken from the Mobile with 14 Flags, or the maquette, as referred to in the text. The Raman spectrum of each of these colors (**d**) shows the presence of ultramarine in the blue, chrome yellow in the yellow, and P.R. 4 and molybdate orange (†) in the red. The white priming layers observed in (**a**,**c**) contain titanium white in rutile form deposited on CaSO4 (\*), which again shows a luminescence pattern (\*) from a Nd3+ impurity. © 2021 MoMA, New York, NY, USA.

The presence of cadmium yellow could explain the pale appearance of the top yellow layer, as cadmium yellow is known to blanch with exposure to light, humidity, and environmental acid—a given for any outdoor sculpture. This exposure leads to the formation of cadmium sulfate (CdSO4), which can react further with carbon dioxide (CO2) to form cadmium carbonate (CdCO3) [23]. While these moieties were not identified directly, this drastic fading is characteristic of cadmium yellows, even under gallery conditions [24]. It was also found that CdS degrades most when illuminated with blue light, which is fully absorbed and generates the highest photocurrent electrochemically [25]. Consequently, the abundance of energetic ultraviolet (UV) and blue light from solar radiation can promote more rapid decay of CdS. Oddly, the reverse is also true: cadmium yellow has been shown to darken considerably when embedded in an alkyd resin [26,27]. While that was not seen here, it points to the photoactivity of cadmium sulfide. Additionally, TiO2 has a photocatalytic effect on some pigments when exposed to UV radiation [28], as when *Man-Eater* was installed outdoors. TiO2 also exhibits photocatalytic chalking, or degradation of the paint film that exposes pigment particles, and could have further hastened the blanching of Layer 9 [29]. Conversely, the dark color observed visually in Layer 1 can perhaps be attributed to the photo-induced reduction in chromate ions to Cr (III) compounds, which is driven by both visible and UV light [23] and can markedly affect those chromate yellows of the rhombic varieties [18]. Sulfur-rich orthorhombic yellows are more prone to browning, possibly due to the increased solubility of PbCrO4 and PbCr1−xSxO4 in this phase, making more chromate ions available for redox reactions [13].

In contrast, the paint of the yellow pennant (Figure 5) in the maquette is still vibrant yellow. A cross section from the maquette showed only two layers, a white priming layer followed by a yellow one. The yellow was similarly identified as a chrome yellow, one that is probably a solid solution of lead chromate and lead sulfate (PbCr1−xSxO4). As in Layers 1 through 3 in the cross section from *Man-Eater*, the symmetric ν1(CrO4 <sup>2</sup>−) stretch

is broadened and shifted to 845 cm−1, and the ν4(CrO4 <sup>2</sup>−) bending modes to 404 and 364 cm−1, all of which are results of crystal compression. This paint is also characterized by a strong ν1(C–O) symmetric stretch at 1090 cm−<sup>1</sup> and a weak in-plane bend ν4(C–O) at 717 cm−1, both characteristic of calcium carbonate (CaCO3). This confirms that the maquette and *Man-Eater* have two different yellow paints. Interestingly, the white priming layer also exhibited the same luminescence emission pattern attributed to Nd3+ ions in titanium-based whites. Both CaSO4 (1020 cm<sup>−</sup>1) and CaCO3 (1090 cm<sup>−</sup>1) are identified in this white, suggesting a different rutile-based paint pigment than in *Man-Eater*. The absence of BaSO4 is more diagnostic for dating and places the painting sometime between 1959, when it left the museum and 1969, when it came back, confirming internal registrar records.
