The Use and Deterioration of Intumescent Fire-Retardant Paint on Louise Nevelson’s Erol Beker Chapel of the Good Shepherd

Abstract

Louise Nevelson’s Erol Beker Chapel of the Good Shepherd (1977) is a sculptural environment consisting of wooden sculptures painted a monochromatic white color. The paints show signs of degradation including cracking, chipping, peeling, and the formation of blisters and powdery efflorescence. A significant amount of pentaerythritol (PER) detected during a former analysis was concluded to originate from an alkyd paint. We show that the PER originates from the PVAc paint on the sculptures, which we have determined to be an intumescent, fire-retardant (IFR) coating. IFR paints and coatings are functional materials designed specifically to delay the combustion of their substrate. At least one other sculpture by Louise Nevelson is known to have been painted with an IFR coating. Our analyses by Fourier transform infrared spectroscopy (FTIR), scanning electron microscopy with energy dispersive X-ray spectroscopy (SEM-EDX), pyrolysis–gas chromatography/mass spectrometry (Py-GCMS), and cross-section microscopy show the presence and distribution of common IFR additives including PER, dicyandiamide, melamine, inositol, ethylenediamine, and phosphates. These are present throughout the PVAc paint and are enriched in the powdery efflorescence. In addition, the degradation behavior of the paint is typical for IFR coating systems that have been exposed to uncontrolled environmental conditions and especially high humidity events.

1. Introduction

1.1. The Nevelson Chapel—Composition and Condition

Erol Beker Chapel of the Good Shepherd, also known as the Nevelson Chapel, is a permanent sculptural environment completed by Louise Nevelson in 1977 (Figure 1A). The Chapel is part of Saint Peter’s Church in a midtown New York City office building and consists of five wooden sculptural elements attached to the walls and one group of hanging sculptures. The elements are made from irregular shapes cut from different types of wood and plywood, which were originally painted in a homogenous white color. Visual examination and the staining of the paint by components of the wood substrates suggest that the sculptures were not primed. According to previous analyses and conservation reports, the sculptures bear the original, artist-applied paint and several layers of overpaint that were applied by a restorer in the 1980s [1,2,3]. This overpainting was likely carried out to disguise losses, degradation, and dust. In the 2000s, the sculptures were examined by several professional conservators who identified condition issues including lifting and flaking paint, paint losses, and dirt and grime. Powdery efflorescence and brown streaks could also be observed on the sculptures (Figure 1B–D). All of these conditions, combined with the variations in color and texture of the various layers of paints not only made the work unstable but also disrupted its intended aesthetic unity.

A 2012 survey postulated that the primary cause of the deterioration was a lack of environmental control leading to adhesion failure between the paint and the wood substrate. Scientific analyses to determine the composition of the original and overpaints were carried out with the intention of identifying an appropriate conservation treatment. The findings from these studies, which include analyses by gas chromatography/mass spectrometry, Fourier transform infrared spectroscopy, X-ray fluorescence spectroscopy, and solid-state and unilateral nuclear magnetic resonance spectroscopy (NMR), suggested that the original paint was an oil-based alkyd with titanium dioxide pigment and calcium carbonate filler [1,2,3]. The overpaint was determined to be a modern, synthetic poly(vinyl acetate) paint (PVAc), assumed to be a water-based emulsion. The researchers reported the detection of a significant amount of pentaerythritol (PER), a polyol used in the synthesis of alkyd paints. This compound was detected in the paint samples (presumed to be both original and restoration), at the interface between them, and in the form of a powdery efflorescence on the surface of the sculptures. The repeated periods of high humidity in the chapel environment over thirty years, as well as the water-based emulsion overpaint, were postulated to have caused a hydrolysis reaction in the alkyd paint, resulting in the accumulation and migration of free PER both within and on the surface of the paint layers [3]. However, data from solid state ¹³C NMR did not show degradation products associated with alkyd paints, and carboxylates due to de-esterification of the polymer chains were not detected [3]. The authors hypothesized that PER may have been added in excess in the original alkyd formulation.

Following this analysis and conservation testing, the areas of lifting paint were consolidated with funori. Once the paint was sufficiently stable, a hydrogel-based treatment was designed for the removal of the overpaint. This was carried out on three of the sculptures—Trinity Columns, Grapes and Wheat Lintel, and The Cross of the Resurrection. Alkaline poly(vinyl alcohol)-borax gels (pH 9–10) [4,5] were employed in the treatment to soften and remove the PVAc overpaint. This was followed by clearance with propan-2-ol. While this removed a considerable amount of the overpaint, it was not completely removed from the sculptures thus treated. As of the beginning of 2020, all of the paint was stabilized, and the overpaint was reduced to some extent on some sculptures while remaining intact on others.

For over three decades, the Chapel shared a heating, ventilation, and air conditioning (HVAC) system with the attached office building. That system was designed to maintain a comfortable indoor environment for its occupants rather than to provide an ideal environment for the preservation of artwork. In 2018, the Chapel was equipped with a separate HVAC system to ensure environmental control in line with the current recommendations for museums and galleries. However, the new HVAC remained partially reliant on the office building’s system. During the COVID-19 pandemic in 2020, the office building had limited environmental controls due to the reduced on-site occupancy, and this lack of regulation contributed to the failure of the Chapel’s HVAC system. In addition, the church building had been unoccupied due to the pandemic and thus for two weeks, a prolonged rise in humidity (between 70–90% RH), temperatures of up to 21 °C, and excessive airflow went unnoticed. These circumstances caused a marked degradation of the sculptures’ paint layers, much more severe than in the past, in particular with respect to lifting paint and efflorescence. This prompted additional analysis and treatment of the sculptures.

1.2. Alkyd Resins, PVAc Paints, and Intumescent Fire-Resistant Coatings

In the 1940s, both alkyd resins and polymeric emulsion coatings based on poly(vinyl acetate) were introduced on the market as house and industrial paints [6,7,8,9]. These materials were quickly adopted by artists, including the use of alkyd house paints most famously by Jackson Pollock [10,11]. Alkyd paints have been identified on individual sculptures (but not sculptural environments) by Louise Nevelson [12]. Alkyd resins are synthetic polyester materials synthesized from polyols, aromatic polybasic acids, and a fatty acid source such as vegetable oil. Pentaerythritol and glycerol are two of the most common polyols used in this class of materials. Poly(vinyl acetate) emulsion paints were also quickly adopted by modern artists including Bridget Riley and Sidney Nolan [6]. A recent review of the composition, characterization, and degradation behavior of PVAc emulsion paints shows that in addition to the primary polymer binder, these paint systems contain many additives, including surfactants, biocides, antifoaming agents, plasticizers, and other compounds [9].

Pentaerythritol (PER) was not noted as a common additive in PVAc emulsion paints. However, PER is a major component in intumescent, fire-retardant (IFR) coating systems [13]. Fire-retardant paints and coatings are functional materials designed specifically to delay the combustion of their substrate. IFR coatings are formulated with four component classes: a solvent carrier (they can be water- or solvent-borne), functional additives for the fire-retardant process (Figure 2), a binder, and fillers (including pigments).

The term ‘intumescent’ denotes the method of fire protection: when exposed to heat, these coatings swell up to 100 times their original thickness, forming a multicellular, foam-like charred layer, which acts as an insulator. The components in an IFR coating have very specific functions during heat exposure, and the combined concentration of the fire-retardant additives (Figure 2) can amount to as much as half the concentration, by weight, of the entire emulsion system [14]. The additives in these systems include an inorganic acid or acid salt such as melamine polyphosphate (MPP) or ammonium polyphosphate (APP), a blowing agent (dicyandiamide (DICY), melamine (MEL), and chlorinated paraffins), and a char-forming polyol compound (pentaerythritol, PER). Some compounds can serve dual functionalities—for example, MPP can also serve as a blowing agent due to the melamine moiety. A synthetic polymer is used as a binder, with poly(vinyl acetate) being a common choice. A number of fire-retardant fillers, including kaolinite clay, metal hydroxides (e.g., aluminum and magnesium), calcium and magnesium carbonates, and titanium dioxide, also contribute to the functionality of the coating [15].

During exposure to very high temperatures, the intumescence process begins with the decomposition of the inorganic acid (APP/MPP), which then reacts with the carbonific compounds (binder, char-formers—PVAc, PER) converting them into carbonaceous char through a dehydration reaction. This char is expanded into a foam through the decomposition of the blowing agent (DICY, MEL), which produces a large amount of non-flammable gas when exposed to high temperatures.

The first report in the literature of an IFR coating mentions a 1938 patent, while the first comprehensive review of these systems was published in 1971 [16,17], indicating that such paint systems were available on the market when the Chapel of the Good Shepherd was created by Nevelson, as well as during the repainting campaigns carried out starting in the mid-1980s. Though a recent publication by Soares et al. proposes the potential future use of IFR coatings for the protection of cultural heritage [18], we believe that our study is the first to report that such materials have already been used by artists like Nevelson. Analysis of the paints and efflorescence carried out using FTIR, SEM-EDX, and Py-GCMS shows the presence of common IFR additives, including PER, DICY, MEL, inositol, ethylenediamine, phosphates, and a chlorinated agent likely to be chlorinated paraffin, throughout the PVAc-based paints on the sculptures. In addition, the degradation behavior of the paint is typical for IFR coating systems that have been exposed to uncontrolled environmental conditions, especially high humidity events.

2. Materials and Methods

Microsamples (3–6 µm) for analysis were collected from all six sculptures in the Chapel of the Good Shepherd (

Table 1. Analysis techniques in this study.

Sculpture	Description of Samples Analysed	Methods of Analysis
Sky Vestment	Restoration paint, soft exudate, original, and overpaints	FTIR, SEM-EDX, Py-GCMS, optical microscopy
Cross of the Good Shepherd	Powdery exudate, blister of underbound powdery overpaint, intact overpaint	FTIR, Py-GCMS, optical microscopy
Frieze of the Apostles	Blister of overpaint, stained area of paint and wood, intact overpaint	FTIR, optical microscopy
Trinity Columns	Paint with staining, powdery efflorescence, intact overpaint	FTIR, optical microscopy
Grapes and Wheat Lintel	Surface paint, efflorescence	FTIR, SEM-EDX, Py-GCMS, optical microscopy
Cross of the Resurrection	Powdery efflorescence	FTIR

). Sampling for FTIR analysis was carried out on what were perceived as top or lower layers of paint or targeting the efflorescence, rather than the full stratigraphy of the paint. The samples collected from Bicentennial Dawn were obtained from an area of the sculpture that was protected from its most recent restoration repainting campaign, which is reported to have employed Sherwin Williams A100 Exterior Oil Primer and Sherwin Williams interior latex Super Paint. The sample is associated with paint applied either originally by the artist or by conservator Susan Schussler in the 1980s, see Section 4.1.

Samples selected for cross-section analysis, featuring the full stratigraphy of the paint layers and, in some cases, the wood support, were mounted in Extec^® polyester resin (Excel Technologies, Inc., Enfield, CT, USA) and cured overnight. Cross-sections were polished with Micro-Mesh silicon carbide (Adaptas Scientific Instrument Services, Palmer, MA, USA) in successive steps to 12,000 grit. Visible and ultraviolet light microscopy was performed with a Ni-U microscope (Nikon Instruments, Melville, NY, USA) with 10× oculars, 4×, 10×, and 20× Plan Fluor objectives, and a 2.5× F-mount adapter for a total magnification of 100×, 250×, and 500×. Samples were illuminated by a SOLA SM II light engine (Lumencor) in visible or blue/violet illumination (395 nm). Images were acquired with a DS-Ri2 16.25 MP scientific CMOS camera (Nikon Instruments) and NIS-Elements F4.30 (Nikon Instruments).

FTIR spectroscopy was carried out on 27 paint samples obtained from the Chapel of the Good Shepherd using a Thermo Scientific Nicolet iN-10 FTIR microscope (Waltham, MA USA) with a liquid nitrogen-cooled MCT-A detector and controlled by OMNIC Picta software (V2.02). Prior to mounting on a diamond half-cell support for transmission mode analysis, the samples were pre-rolled on a glass slide to ensure separation between paint and alteration layers where present. A total of 128 scans were collected over a spectral range of 4000–650 cm⁻¹ with a spectral resolution of 4 cm⁻¹. Two samples from Bicentennial Dawn were analyzed using a Thermo Scientific Nicolet iS5 spectrometer with an iD7 ATR accessory. A total of 3 scans from each side of the paint samples were collected over a spectral range of 4000–500 cm⁻¹ with a spectral resolution of 4 cm⁻¹. Sample spectral processing for all data was performed with OMNIC Specta software V9.9.549. Data are displayed from 4000 to 675 cm⁻¹ due to noise at lower wavenumbers. Interpretation was carried out in the same software using reference libraries including the Infrared Users Group (IRUG), the Spectral Library of the Gettens collection, the HR Hummel Polymer and Additives library, and the HR Coatings Technology reference library.

Secondary electron and backscattered electron images were obtained using an FEI Quanta 650 FEG Environmental SEM. Energy dispersive X-ray spectra and elemental maps were collected using an accelerating voltage of 15 kV and processed by twin Bruker XFlash silicon drift detectors (SDD) using Bruker ESPRIT software. Samples were carbon coated to reduce surface charging.

Pyrolysis–gas chromatography/mass spectrometry was carried out on a paint sample from Sky Vestment, and on samples of powdery efflorescence from Grapes and Wheat Lintel and Cross of the Good Shepherd. The samples were derivatized in situ with 2 µL tetramethyl ammonium hydroxide solution (25% w/w TMAH in methanol) in a deactivated stainless-steel cup and placed into a helium-purged Frontier Labs PY-3030D (Koriyama, Japan) furnace at 600 °C. TMAH is a strong methylating reagent and will methylate carboxylic acids, phenolic alcohols, and amines [19]. The resulting pyrolysates enter a Shimadzu GCMS-QP2010 SE (Columbia, MD, USA) equipped with a split/splitless injector held at 320 °C with a 50:1 split and a Restek capillary column (Rtx-5 ms; 30 m × 0.25 µm i.d.; 0.25 µm film thickness) with a helium flow of 1.2 mL/min controlled by linear velocity. The temperature program starts at 40 °C for 5 min and ramps at 20 °C/min to 300 °C for 5 min. The quadrupole mass spectrometer has a solvent cut time of 2 min, an interface temperature of 300 °C, and the ion source is maintained at 200 °C, with ionization set at 70 V over a scan range of 45–600 amu at 3 scans/s. Data were analyzed using the Shimadzu GCMSsolution (version 4.54) software and NIST 2023 spectral libraries.

3. Results

3.1. Optical Microscopy

Cross-sections from five of the six sculptures (Table 1) were examined under visible and UV light (395 nm). The samples show differences in color, texture, layering, and fluorescence, as well as the concentration and size of particulate materials dispersed in the binder. In a sample collected from Sky Vestment, a two-layer system can be observed, with a thick, bottom layer of cream-colored paint, followed by a thinner layer of cooler white paint (Figure 3A,B). The lower layer exhibits a foam-like structure that is not typical for alkyd paints. The surface bright-white layer contains a higher proportion of large, transparent particulate inclusions. The cream-colored bottom layer shows a yellow fluorescence when viewed under UV light that is also present in the thinner top layer, though less obvious due to the higher volume of particulates. A cross-section from Grapes and Wheat Lintel exhibits cooler, blue-green fluorescence and a more compact character to the binder (Figure 3C,D). Blue-white, fluorescent particles can be observed in this cross-section. The distinct separation between layers is not always present in the samples. The roughness of the surface layer demonstrates the powdering degradation phenomenon observed in some areas of the sculptures. The cross-section analysis confirms visual observations—the degradation effects due to adverse environmental conditions are not uniform throughout the sculptural environment.

3.2. SEM-EDX

The elemental compositions of two cross-sections from Grapes and Wheat Lintel and five cross-sections from Sky Vestment were analyzed using SEM-EDX. Phosphorus, in significant quantities, is detected in all samples (Figure 4, Figure 5 and Figure 6); its near absolute correlation with oxygen supports the hypothesis that this element is present in the form of phosphate (Figure 5 and Figure 6, image F show the result of signal subtraction of oxygen from phosphorous). The elemental maps of phosphorus can also be used to determine the presence and location of the polyphosphate compounds in the intumescent, fire-retardant paint on the sculptures. In samples analyzed by SEM-EDX that contain part of the wood support, we observed the presence of phosphorus inside the cells of the wood, suggesting that in the studied areas, the fire-retardant coating was either applied directly to the wood support or that the polyphosphate additives and/or their degradation products are mobile and have migrated into the wood substrate (Figure 4).

Alongside the phosphorus, an unusually high level of chlorine, correlated with the location of the binder, was detected, indicating the presence of a chlorinated compound in the paint (Figure 5I and Figure 6I). Chlorinated paraffin is a secondary blowing agent used in intumescent, fire-retardant coatings and may be the origin of the chlorine content in these samples. Chlorinated paraffin is also added to synthetic paint formulations to prevent cracking. The SEM-EDX analysis also confirms the presence of fillers containing silicon, calcium, and magnesium (Figure 5 and Figure 6). Calcium carbonate, dolomite, kaolinite clay, and pyrophyllite were identified by FTIR spectroscopy (Section 3.3).

It is evident from the SEM-EDX data in Figure 5 and Figure 6 that the primary pigment in the paint from this sample is titanium dioxide. Titanium dioxide is also used as a filler in IFR coatings, serving to reduce the average diameter of the cells in the char [13]. The SEM data indicate that the primary filler in Grapes and Wheat Lintel is calcium-based—calcium carbonate was also confirmed by FTIR—with a bimodal particle size distribution. Note the presence of acicular structures that emanate from the top and bottom of the paint layer in Figure 6. These are rich in phosphorus, indicating that the soft and friable surface of the paint layer and the material efflorescing from the paint layer are dominated by the polyphosphate component of the paint.

Samples obtained from the blister-like formations in Grapes and Wheat Lintel exhibit a triple-layer structure, with more compact, brighter white layers at the top and bottom, and a thicker, cream-colored layer in between. The color difference can be attributed to the distribution of the titanium dioxide pigment in the samples, which can be seen in both Figure 5C and Figure 6C. The magnesium and silicon-based fillers are also more concentrated in these outer layers, while the calcium-based filler (likely calcium carbonate) remains distributed throughout the sample. To understand the correlation between calcium, magnesium, and silicon seen in some particles, further research using for example X-ray diffraction is necessary.

3.3. FTIR Spectroscopy

Samples from all six sculptures were analyzed by FTIR spectroscopy. The composition of the paint binder, efflorescence, substrate and wood glues were identified by this technique. A summary of the results can be seen in

Table 2. Summary of results from FTIR spectroscopic analyses on samples from the Chapel of the Good Shepherd.

Sculpture	Components
Sky Vestment-Trinity	Paint: poly(vinyl acetate), pentaerythritol, titanium dioxide, polyphosphate salt 1, kaolinite clay
Cross of the Good Shepherd	Paint: poly(vinyl acetate), pentaerythritol, dicyandiamide, polyphosphate salt
	Wood: cellulose, synthetic wood glue (a formaldehyde-based resin), an epoxy curing agent with similar signals to dicyandiamide
Frieze of the Apostles	Paint: poly(vinyl acetate), pentaerythritol, titanium dioxide, polyphosphate salt
	Wood: cellulose, natural resin varnish (shellac), wood glue (formaldehyde-based resin), an epoxy curing agent with similar signals to dicyandiamide
Trinity Columns	Poly(vinyl acetate), pentaerythritol, polyphosphate salt, dicyandiamide, natural resin (terpenoid fraction)
Cross of the Resurrection	Poly(vinyl acetate), pentaerythritol, dolomite (calcium magnesium carbonate), pyrophyllite (aluminum silicate commonly associated with quartz or kaolinite clay), unidentified melamine salt 2, unidentified inorganic filler 3
Grapes and Wheat Lintel	Poly(vinyl acetate), pentaerythritol, titanium dioxide, dicyandiamide, polyphosphate salt, calcium carbonate (chalk), unidentified melamine salt, unidentified inorganic filler
	Wood: cellulose, natural resin varnish (shellac), synthetic wood glue (formaldehyde-based resin), an epoxy curing agent with similar signals to dicyandiamide.

¹ The nature of the polyphosphate compound (e.g., MPP, APP, or other) cannot be determined by FTIR in the paint and efflorescence samples. ² Bands associated with amine compounds are present in the spectra; however, due to the complex mixture of materials present, we are unable to identify their structure by FTIR. Py-GCMS analysis (see Section 3.4) indicates the presence of melamine. ³ See Figure S2 in the Supplementary Information.

and additional spectra are available in Figures S1–S6 in the Supplementary Information File. Alkyd resin was not detected in any of the 27 samples analyzed in this study.

Poly(vinyl acetate) is the primary polymeric binder identified in the paint samples. The diagnostic signals used in this characterization include C=O stretching band ca. 1740 cm⁻¹, C-H asymmetric deformation ca. 1435 and 1375 cm⁻¹ (CH₂ and CH₃ deformations, respectively), C-H in-plane bending at 1240 cm⁻¹, C-O stretching 1125 cm⁻¹, CH₂ twisting ca. 1020 cm⁻¹, and CH₃ wagging ca. 945 cm⁻¹ [20]. Pentaerythritol is detected in every sculpture based on its characteristic signals, including a strong and sharp OH stretch at ca. 3330 cm⁻¹, C-H stretching ca. 2955 and 2885 cm⁻¹, C-O stretching ca. 1030 cm⁻¹, C-H rocking ca. 874 cm⁻¹, and O-H torsion ca. 670 cm⁻¹ [21]. Dicyandiamide was also detected in every sculpture with the following absorption signals used for its characterization: N-C≡N stretching ca. 2207 and 2164 cm⁻¹, -NH₂ bending 1663, 1643, and 1634 cm⁻¹, C-N asymmetric stretch and C-N-H bending ca. 1575 cm⁻¹, N-C=N asymmetric stretching at 1507 cm⁻¹, N-C≡N symmetric stretching at 1255 cm⁻¹, and N-C≡N symmetric stretching at 930 cm⁻¹ [22,23]. DICY and PER were often detected together in the efflorescence samples (Figure 7).

Signals associated with polyphosphate salts and, possibly, their degradation products are also detected in several samples: for APP, these include the asymmetric stretching vibration of N-H in NH₄⁺ at ca. 3250 cm⁻¹, asymmetric deformation vibration of N-H in NH₄⁺ ca. 1450 cm⁻¹, symmetric deformation vibration of N-H in NH₄⁺ ca. 1260 cm⁻¹, asymmetric stretching vibration of P=O in PO₄³⁻ ca. 1095 cm⁻¹, symmetric stretching vibration of P=O in PO₄³⁻ ca. 908 cm⁻¹ [24]. We have attributed the signal ca. 1680 cm⁻¹ to the N-H bending signal of MPP [25], though basing the characterization on a single band in this complex mixture is difficult. A strong and broad deformation in the baseline ca. 2360 cm⁻¹ was detected in many samples and is associated with the P-H stretching mode (Figure 8, Figures S2 and S6 in the Supplementary Information) [26]. Based on the FTIR spectra, we are unable to make an absolute determination of the polyphosphate salt(s) that has been used in the IFR coating. The complex mixtures of materials and potential shifts in band wavenumber due to their complexation or degradation merit further research. However, the combination of characteristic signals listed above, the large quantity of phosphate detected in the SEM-EDX and Py-GCMS data confirms the presence of a polyphosphate compound. APP and MPP may both be present, along with ethylenediamine phosphate (see Section 3.4) further supporting the hypothesis that different IFR coatings have been applied to the sculptures over time (e.g., originally and during the restoration campaigns).

Though not ubiquitous, kaolinite clay, calcium and magnesium carbonates, and titanium dioxide can be noted throughout the samples. Though chlorine noted in the SEM-EDX results is assumed to be associated with chlorinated paraffin; the sharp C-H stretching vibration bands associated with wax were not noted in the FTIR spectra from these samples, indicating either a low proportion of this additive in the paint samples or chlorine’s association with other salts. The wooden substrates in the sculptures contain a formaldehyde-based resin wood glue. Two epoxy curing agents were detected in the samples—one based on poly(styrene) resin and the other with absorption signals similar to dicyandiamide (N-C≡N stretching ca 2210 and 2160 cm⁻¹). We detected a natural resin varnish (shellac is suggested by its pale orange fluorescence) on samples from Frieze of the Apostles and Grapes and Wheat Lintel, in agreement with previous observations [27]. In addition, samples obtained from the stained areas of the sculptures showed absorption signals associated with terpenic resins (polymers of low molecular weight hydrocarbons such as beta-pinene) [28]. Data associated with these findings can be found in Figures S3 and S4 in the Supplementary Information.

The following bands remain unassigned: ca. 1700 cm⁻¹ C=O stretching typical of carboxylic acids, aliphatic ketones, conjugated acids, or aldehydes; ca. 1600 cm⁻¹ possible N-H bend (primary amine) or C=C stretching in conjugated or cyclic alkenes; the broad band ca 1506 cm⁻¹ in Figure 8B may be associated with N-O asymmetric stretching of nitro compounds or N-H stretching; the signal at 1450 cm⁻¹ may be also associated with the C-N stretch of an amide. In Cross of the Resurrection and Grapes and Wheat Lintel, we also note several signals in the O-H stretch region that remain unidentified—3540 and ca. 3490 cm⁻¹ (Figure S2 in the Supplementary Information). Signals in this region are associated with aluminum hydrate species such as bayerite and gibbsite, though we would also expect sharp absorption bands ca. >3600 cm⁻¹ for these materials. Phosphate salts may also show similar bands. Further research is needed to understand the nature of all compounds present in the complex mixtures found in the samples.

3.4. Pyrolysis GCMS

The chromatograms of a paint sample from Sky Vestment and powdery efflorescence from Grapes and Wheat Lintel and Cross of the Good Shepherd can be seen in Figure 9; the elution times for each major compound can be found in

Table 3. Pyrolysis products from paint and efflorescence samples. Where elution times of different compounds overlap, the sculptural origin of the sample has been noted.

Peak	Retention Time (min)	Compound
1 (SV) 1	2.26	Benzene
1 (GWL 2, CGS 3)	2.27	N,N,N′-trimethyl-1,2-ethanediamine
2	7.76	Trimethylphosphate
3	8.615	1-ethenyl-2-methylbenzene
4	9.875	Pentaerythrotetramethyl ether
5	10.445	3-methoxy-2,2-bis(methoxymethyl)-1-propanol
6	12.455	1,2,3,4,5,6-hexa-o-methyl-myo-inositol
7 (CGS)	13.61	N,N,N′,N′-tetramethyl-1,3,5-triazine-2,4,6-triamine
7 (SV)	13.61	2-N-diethylmelamine
8 (GWL, CGS)	13.865	Altretamine
8 (SV)	13.88	Pentamethylmelamine

¹ Sky Vestment. ² Grapes and Wheat Lintel. ³ Cross of the Good Shepherd.

. Mass spectra for the main compounds eluted from each sample can be found in Figures S7–S17 in the Supplementary Information. The primary components in all three spectra are phosphate (derivatized from: trimethylphosphate), pentaerythritol (derivatized forms: pentaerythrotetramethyl ether and 3-methoxy-2,2-bis(methoxymethyl)-1-propanol), and melamine (derivatized forms: N,N,N′,N′-tetramethyl-1,3,5-triazine-2,4,6-triamine, 2-N-diethylmelamine, altretamine, pentamethylmelamine). The last may have been added to the paints as neat melamine (MEL) or as the polyphosphate salt, MPP, or both.

The primary components in the powdery efflorescence from Grapes and Wheat Lintel are phosphate (74%) and pentaerythritol (14%). The efflorescence from Cross of the Good Shepherd is dominated by phosphate (37%), pentaerythritol (36%), and melamine (16%). These same three compounds also dominate the chromatogram of the paint sample from Sky Vestment—phosphate (45%), pentaerythritol (11%), and melamine (32%). The eluted benzene and 1-ethenyl-2-methylbenzene in the samples support the identification of PVAc binder via FTIR spectroscopy. Benzene, along with acetic acid, are considered markers for PVAc in py-GCMS. Acetic acid is formed through an elimination reaction during the pyrolysis of PVAc; following this step, benzene is produced through a rearrangement reaction of the polyene backbone [29]. 1-Ethenyl-2-methylbenzene forms through further recombination and condensation reactions of the original benzene molecules formed during the pyrolysis of PVAc [30]. The base peak for methyl acetate (the derivatized product of acetic acid), 43 m/z, falls below the mass spectral collection range employed in the current analysis (45–600 amu). Present in the efflorescence samples is also ethylene diamine (derivatized form: N,N,N′-trimethyl-1,2-ethanediamine), likely derived from ethylene diamine phosphate, another phosphate salt used in IFRs [16,31]. All three samples also show a significant amount of inositol (derivatized form: 1,2,3,4,5,6-hexa-o-methyl-myo-inositol), an oligomer of erythritol and a char-forming polyol used in IFR coatings [32].

4. Discussion

4.1. IFR Coatings in the Works of Louise Nevelson

The combined microscopic and spectroscopic information from the analysis of all six sculptures in the Chapel of the Good Shepherd indicate the consistent use of IFR paints in this Nevelson artwork. Although alkyd resin was not detected in this analysis, we cannot rule out the possibility that this type of coating had been applied by Nevelson originally. However, the majority of the current paint on the surfaces contains significant amounts of IFR additives, including PER, DICY, MEL, phosphates, inositol, ethylenediamine, chlorinated compounds, titanium dioxide, and inorganic fillers used for their fire-retardant capacity.

In 2007, another Louise Nevelson sculpture from the same period (1976) underwent conservation treatment. Bicentennial Dawn is an installation of 26 white multi-faceted geometric columns located in the Byrne Federal Building and U.S. Courthouse in Philadelphia. Similar to the Chapel sculptures, the paint on the sculptural surfaces showed severe delamination from the wood. The treatment report from this campaign includes a history of previous paints applied to the sculpture, including the following:

The soft and powdery characteristics of the paint, too, seem to be what [conservator Susan Schussler] reported also in 1983. (…) [t]he sculpture was originally painted with the fireretardant (sic) Everseal products (and deteriorated), and (…) as stated, she repainted portions with the same product and (…) in the early 1990’s it again was painted with the same or similar fire retardant paint (…).

[33]

According to this report, not only was Bicentennial Dawn repainted twice with intumescent fire-retardant paints (in 1983 and in the 1990s), but the original paint used by Nevelson was also an ‘Everseal’ fire-retardant paint. The original 1983 conservation report by Susan Schussler confirms the summary above [34]. According to this report, the original paint used on Bicentennial Dawn was Everseal LD3 matte, a fire-retardant paint. After removing the overpaint, Schussler primed the sculpture with Everseal AT-15 primer (recommended for use with Everseal LD3) and then repainted it with Everseal LD3 [34]. FTIR spectroscopic analysis of a sample from this sculpture confirmed the presence of an IFR coating nearly identical to that found on Cross of the Resurrection, including a poly(vinyl acetate) binder, pentaerythritol, titanium dioxide, and likely, an amine blowing agent and a polyphosphate salt (Figures S5 and S6 in the Supplementary Information). A UV tracer called UVITEX OB (2,2′-(2, 5-thiophene diyl)bis(5-tert-butylbenzoxazole), also known as 2,5-Bis(5-tert-butyl-2-benzoxazolyl)thiophene, obtained from Ciba-Geigy Corporation, Trinity Skyline Drive, Hawthorne, NY, USA) was included in the paint to allow for the future identification of the restoration paint. The paint was thinned using mineral spirits and/or naphtha VM&P, confirming that this was a solvent-borne, rather than emulsion, intumescent paint, appropriate for the period.

Numerous patents and publications from the 1970s and 1980s can be found for solvent-based intumescent coatings containing DICY, PER, and amine polyphosphates such as APP, MPP, ethylene diamine phosphate, inositol, “Chlorowax”—chlorinated paraffin, and a poly(vinyl acetate) binder. Although the company that produced Everseal products (Everseal Manufacturing Co., Inc., Ridgefield, NJ, USA) is no longer in business, their products are referred to in other patents from this period. For example, a 1978 patent CA1109607A filed for the development of a “Semi-durable, water repellent, fire retardant intumescent composition” states that such preservative coatings were sold under the trademark Canvaseal SE-102 from Everseal Manufacturing Co., Inc., confirming that such materials were likely to have been available to Nevelson during the creation of the Chapel of the Good Shepherd as well as during previous conservation campaigns.

Optical microscopy and SEM-EDX analysis of a cross-section from a blister suggest a dual process of degradation is occurring in the intumescent, fire-retardant paint in these areas. Segregation of some components in the coating, such as titanium dioxide (see Figure 5C and Figure 6C), is accompanied by migration of the efflorescence-forming materials (dicyandiamide, pentaerythritol, phosphates). Alternatively, we may interpret the bilayer appearance of some samples as evidence of consecutive repainting campaigns on the sculptures. Cross-sectional analysis of IFR coating samples in the published literature is generally carried out after exposure to heat and fire to observe changes in the expanded char foam structure due to various aging and degradation mechanisms that the coatings have been exposed to [35]. The foamy appearance of the paint in Figure 3A may be consistent with intumescent coating prior to heat exposure. For example, Jimenez et al. have used SEM-EDX to demonstrate changes in IFR coatings as a result of immersion in saltwater (meant to simulate undersea environments) [36,37]. Their results show increased porosity in the coating, which the authors attribute to migration and loss of MEL [36]. It is possible that a similar mechanism has occurred in the Chapel, though the conditions are far less extreme. Further research on the mechanisms of degradation of such coatings in heritage environments is merited.

FTIR spectroscopic analysis of the paint samples identified several different photoluminescent materials, which are likely to be the cause of the yellow and blue-green fluorescence observed in the cross-sections. These include a terpenic resin from the wood as well as aromatic components of the coating such as melamine. It is also possible, though less likely, that a fluorescent optical brightener was used during the previous restoration campaigns, in the same manner as with Bicentennial Dawn. The variable visual characteristics in the samples (e.g., foamy or compact texture, large or fine particulates, yellow or blue fluorescence) observed in cross-sections from different sculptures can be attributed to the number of restoration repainting campaigns, and the likelihood that different types of IFR paints have been employed.

4.2. Environmental Degradation in IFR Coatings

Intumescent, fire-retardant coatings are especially sensitive to weathering because the high polarity of the components causes poor compatibility with the polymer matrix of the binder [38,39,40,41]. In humid or wet environments, the hydrophilic additives close to the surface can migrate out of the coating; as this process continues, further diffusion pathways can be created. During water-exposure tests, Bruckner et al. demonstrated that unreacted DICY, which is used as a curing agent, can be leached out of epoxy adhesives [42]. Wang and Yang have demonstrated that PER and APP can be extracted from IFR coatings when immersed in water [39]. Migration of water molecules into the polymer matrix can result in the breaking of intermolecular hydrogen bonding between the APP chains, which can then more easily migrate through the hole-free volume of the binder [40]. Water can also hydrolyze phosphate bonds in the main chain of APP to produce polyphosphoric acid, which, upon further hydrolysis, yields orthophosphates [40]. Furthermore, Jimenez et al. have demonstrated that sodium and chlorine ions migrate into IFR coatings very rapidly, generating far more soluble sodium polyphosphate and melamine chloride [36,37]. This variety of phosphate-based moieties makes their absolute identification in the current FTIR analysis difficult.

Alongside the loss of additives, changes in the surface condition of IFR coatings have also been studied in depth. Daus et al. carried out rain exposure simulation testing on samples of intumescent, fire-retardant coatings containing different binders, as well as poly(vinyl acetate), titanium dioxide, PER, APP, and MPP [38]. On removing the test samples from the aging chambers, the authors detected a strong smell of ammonia, associated with the thermal decomposition of APP, concluding that the blisters formed in the paint contain this volatile alkaline agent. PVAc is known to undergo basic hydrolysis to form poly(vinyl alcohol), and such degradation may also have occurred in the paints studied here. Blistering and delamination in the films result from a buildup of high osmotic pressures due to the permeation of water molecules into the films [43]. The buildup of leached hydrophilic molecules on the surface of the films also increases the overall surface hydrophilicity, accelerating the diffusion of water and the blister process. The leaching of additives also creates voids in the paint film where water molecules can be trapped furthering the degradation cycle [39,40]. These empty voids could form interconnected pores and may be the cause of the foam-like structure of the paint cross-sections observed in this study [40].

These processes suggest that over many years, the humidity cycling in the Chapel, in addition to the HVAC failure of 2020, could have led to the degradation of the phosphate salts in the coatings and encouraged migration of small molecular compounds (PER, DICY) to the surface. While the use of an aqueous consolidant (funori) and of PVA-borax hydrogels may have accelerated this process in the already porous and degraded paint films, it should be noted that in 2020, the paint degradation was the worst in the sculpture that had not been treated with the hydrogels and that the paint on one sculpture that had been treated but was not in the Chapel during the HVAC, failure has remained stable.

The degradation processes described here alter the composition and behavior of IFR coatings, weakening their functionality. The loss and redistribution of key components in the system affects the uniformity of the char layer, leading to heterogeneity in the cells of the foam structure and decreasing the overall ability of the film to insulate the substrate during a fire [16]. These degradation processes raise significant challenges for conservation treatments.

5. Conclusions

This study has demonstrated the presence of historically applied intumescent, fire-retardant coatings to the sculptures in Louise Nevelson’s Erol Beker Chapel of the Good Shepherd. PVAc-based IFR coatings were the only type of paint detected on the sculptures in this study. In addition, conservation records from a contemporaneous Nevelson sculpture, Bicentennial Dawn, indicate the use of Everseal fire-retardant products by the artist and during later conservation campaigns. Significant quantities of IFR functional additives—pentaerythritol, inositol, melamine, dicyandiamide, ethylenediamine, titanium dioxide, phosphates, and inorganic fillers—were identified in cross-sections by SEM-EDX and in samples analyzed by FTIR spectroscopy and Py-GCMS. Due to the number of additives used in IFR coatings and the complexity of their degradation processes, further research is necessary for a complete characterization of the components in these paints, as well as their evolution and distribution throughout the paint stratigraphy. For a comprehensive characterization and determination of the nature of the fluorescent moieties, further analysis via Raman spectroscopy and µ-FTIR mapping spectroscopy, for example, is necessary but is beyond the scope of this study.

In this paper, we have also described how water, both in liquid and gaseous form, can impact the stability of the IFR paints, which are known to crack, peel, and blister during humidity and rain simulation testing. The hydrophilic additives in the IFR coatings, which are necessary for their fire-retardant properties, migrate to the surface creating a powdery efflorescence effect, or accumulate inside the blisters. Both of these phenomena were also observed on the Nevelson Chapel sculptures, though an exact distribution of the degradation products or IFR additives within the paint layers requires further study.

The sculptures within Nevelson’s Erol Beker Chapel of the Good Shepherd have undergone multiple exposures to humidity and/or liquid water. The HVAC failure of 2020 subjected all of the sculptures except Trinity Columns (which was not installed at the time) to high humidity and intense airflow. In addition, certain sculptures were treated with aqueous consolidants and water-based gels during a previous conservation campaign. Although the contact time with water is relatively short for consolidation and gel cleaning, it is possible that liquid water may have migrated into the IFR paint [44]. Studies assessing the suitability of PVA-borax hydrogels on porous, degraded IFR paint films are warranted given that borax forms the sodium salt of tetraborate in water.

However, the most important factor for the long-term preservation of the Erol Beker Chapel of the Good Shepherd and other artworks suspected of containing IFR coatings is a controlled environment. Following the HVAC incident in 2020, Nevelson Chapel established an Environmental Committee tasked with the ongoing responsibility of monitoring HVAC system performance. This committee operates proactively to identify and address any issues promptly, as well as develop strategies for mitigating future challenges. Whenever the system deviates from its optimal parameters, multiple committee members are immediately notified, ensuring that the HVAC system undergoes close monitoring and maintenance.

Researchers and conservators should be alert that the identification of large amounts of PER and/or DICY via FTIR spectroscopy in modern paints may indicate the presence of an IFR coating. Both of these additives have easily recognizable spectral components—strong and sharp OH stretch ca. 3330 cm⁻¹ and C-O stretch ca. 1030 cm⁻¹ for PER and a doublet related to N-C≡N stretching ca. 2207 and 2164 cm⁻¹ in the case of DICY. Furthermore, the polyphosphate salts used in large quantities in such coatings result in relatively easy to recognize FTIR spectral characteristics—a very broad and intense band in the region of 3500–2500 cm⁻¹ and a strong and broad deformation in the baseline ca. 2360 cm⁻¹. The polymer binders in IFR systems can be wide-ranging and, alongside PVAc, include vinyl acetate/dibutyl maleate copolymer, urea/formaldehyde, ethylene vinyl acetate, epoxy resins, polyamide, and cellulose. Therefore, it is the IFR additives that are more useful in determining whether modern paint belongs to this class of functional coatings. Visual observations of degradation such as blistering, large quantities of powdery efflorescence, delamination, and a foam-like texture to the paint can also point to the use of IFR coatings. Further systematic research around the identification, degradation, and appropriate conservation measures for these paints is warranted.