Shedding Light on Daylight Fluorescent Artists’ Pigments, Part 2: Spectral Properties and Light Stability

ABSTRACT Daylight fluorescent pigments are complex artists’ colorants made of multiple admixed dyes and additives infused in a polymer resin. Their unique photochemistry creates interesting optical effects that make them appealing for a range of applications. The wavelength dependence of the dyes’ emissive properties has also led to their use in works meant to be displayed under nontraditional high energy light sources, such as blacklights. The intended display methods and the photochemistry of the constituent dyes can lead to chemical instability and also pose unique challenges for conservation and exhibition. In Part 1 of this research, we reported on the chemical constituents of colorants from two major manufacturers. In this paper, we provide a comprehensive report of the spectral properties and lightfastness of paints prepared with the pigments. The optical properties and chemical stability of these colorants are correlated with the composition. Variation in composition can lead to unique markers that could be useful for conservation treatment and exhibition considerations. Changes in the dyes used in the pigment formulation over time are considered in the comparison of two objects: a Stephen Sprouse silvered leather motorcycle jacket featuring fluorescent art by Stefano Castronovo and selections from the Day-Glo Designer’s Guide from 1969.


Introduction
Daylight fluorescent pigments have unique appearances and show interesting photochemistry as has been showcased in a range of optical studies (Connors-Rowe, Hinde et al. 2013;Metcalf 2014;Beckett, Holden, and Smith 2019). The pigments in these paints rely on different combinations of fluorescent dyes and optical brighteners to provide a mechanism for energy transfer following irradiation that yields a wavelength-dependent appearance of the paint colors. Unfortunately, this can also lead to conservation problems as the materials age and components degrade, often at different rates. Specifically the original hues tend to shift with exhibition driven aging, primarily due to light exposure, and the enhanced emission decays away reducing the "punch" of the perceived color and depth unique to the original material (Aach 1970;De Winter 2018).
In Part 1 of this two-part study, we reported on the chemical formulations of pigments from the major American and European manufacturers of daylight fluorescent colorants, Day-Glo and Radiant Corporations. The pigments are made of dyes, optical brighteners, and additives such as UV-absorbers, mixed into a polymer resin. The components are not covalently bonded to the polymer and organic additives can be easily extracted from the matrix using organic solvents (Sobeck, Chen, and Smith 2021). A summary of the composition of the extracts, detailed in Part 1, is included in the Appendices for reference. Variations in the compositions were found that can aid in identifying these colorants in artworks and possibly the pigment's manufacturer, but these compositional differences also impact their light stability. Specifically, it was found that the majority of pigments from Radiant Corporation contain a fluorescent brightener, whereas only the blue color from Day-Glo contains this type of additive. The main classes of dyes are similar between the two companies, as well as dye color combinations used to achieve a given color. Orange and red tones are comprised of mixtures of a yellow coumarin dye, Solvent Yellow 172, and red and violet rhodamines with the precise rhodamine derivatives varying by manufacturer. The blue and green tones are achieved using phthalocyanine blue pigments along with a fluorescent brightener to enhance the color intensity, plus the same yellow dye for green (Sobeck, Chen, and Smith 2021).

Spectral properties and lightfastness
Connors-Rowe, Morris, and Whitmore (2005) reported on a comprehensive investigation of the spectral characteristics and light stability of one brand of daylight fluorescent artists' watercolors under different lighting conditions. Spectral analysis of the paint appearance via reflectance measurements indicated that high correlated color temperature (CCT) or simulated daylight is necessary to gain the full effect of the emissive properties of the material. Blacklights can also produce an emissive effect, which can have a startlingly different visual appearance if reflected visible light is not present. Comparison of fading with and without UV-filtering revealed that most of the watercolors had significantly increased lightfastness when UV was omitted, but a direct correlation of the stabilization with the hue was not apparent. The majority of paints studied showed similar increases in fading using daylight that included UV wavelengths or with blacklight exposure, though the absolute amount of fading and the change in the perceived color varied across the set of watercolors. Changes in the reflectance spectra were attributed to different factors, adding to the complexity of this media. Color loss was sometimes due to degradation of a single dye in the color, as was observed in the yellow tones, but sometimes degradation of one dye of an energy transfer pair resulted in a shift in the emission and the perceived color. As an example of the latter process, many of the fluorescent pigments contain FRET (Förster resonance energy transfer) pairs where the emission of one dye, the "donor," overlaps with the absorbance of the second, the "acceptor." The energy transfer process results in an enhancement of the emission from the acceptor dye relative to its emission in the absence of the donor. The perceived color though is impacted if the individual dyes in the pair have different stabilities, and the energy transfer process is disrupted resulting in either a perceived emission from the donor or loss of the enhancement of the acceptor's emission. Blacklight exposure caused observable hue shifts for several of the red and gold colors thought to contain mixtures of red and violet dyes, indicating that such a disruption of the FRET pair may be occurring .
Whereas the previously mentioned study used a light aging chamber and periodic static reflectance measurements to determine colorimetric changes, dynamic microfade testing (MFT) has also been applied to daylight fluorescent colors. MFT combines accelerated visible light aging exposure by focused optics on a sample surface with continuous visible color monitoring through 45 reflection spectrophotometry using the illumination source (Whitmore, Pan, and Bailie 1999). CIE L*a*b* color coordinates and color difference calculations (ΔE76) are otherwise handled identically and record the appearance impact of both reflected and emitted light.
Studies using MFT on paintings and screenprints featuring daylight fluorescent pigments demonstrated the fleeting nature of the colorants (Metcalf 2014;Beckett, Holden, and Smith 2019). These studies show that daylight fluorescent pigments can be "superfaders," showing color change much greater than the most fugitive reference swatch, Blue Wool Standard 1 (BWS1), when irradiated under the same conditions. In a study of Corita Kent screenprints, repeat fading trials with intermittent light showed some recovery of color following a dark period, which the authors attribute to local heating from the instrument light source (Metcalf 2014). This led those authors to conclude that the daylight fluorescent materials fade on the order of BWS2-3, which is more stable than previous studies indicated Metcalf 2014). The MFT analysis of a fluorescent Mona Lisa painting by Castronovo on a leather motorcycle jacket designed by Stephen Sprouse at the Indianapolis Museum of Art (IMA) at Newfields showed differences in the light stability of the red and yellow paints. The yellow and yellow-ish green fluorescent paints were found to be more stable, BWS1-2, than the red paint, which faded significantly faster than BWS1 (Beckett, Holden, and Smith 2019).
The organic dyes and additives identified previously by the authors in the pigments studied here include rhodamines, coumarins, optical brighteners, and a UVabsorber, as shown in Figure 1 (Sobeck, Chen, and Smith 2021). Fundamental studies of the photochemistry of rhodamine dyes, a xanthene derivative, have indicated that their stability is sensitive to the medium and that incorporation of the dye into polymer matrixes, as is the case in fluorescent pigments, can significantly increase their lightfastness (Hinckley, Seybold, and Borris 1986;Reisfeld et al. 1988). The earliest rhodamine dye commercially used was rhodamine B (RhB). Ester derivatives of RhB, like rhodamine 6G (R6G) or 3B (R3B), were explored later for their increased quantum efficiency of fluorescence. This is attributed to the absence of a lactonic form in the ester derivatives, which competes with the emissive zwitterionic state (Hinckley, Seybold, and Borris 1986). Our studies found that both ester derivatives are major dye components in the 2019 dated pigments, whereas a 1969 Day-Glo sample contains RhB and R6G (Sobeck, Chen, and Smith 2021). In addition to being a less efficient fluorophore, RhB is a suspected carcinogen and is falling out of commercial favor (Ismael, Schwander, and Hendrix 2013). The combination of a plastic resin medium for these pigments and the ester forms of rhodamine used today work to increase the quantum efficiency of the fluorophore and its light stability in modern fluorescent pigments; nevertheless, the rhodamine dyes used in these modern pigments are still prone to rapid fading.
The mechanism of photodegradation is not fully understood and has been the subject of many fundamental photochemical studies. There is significant interest in photocatalyzed oxidative degradation of commercial dyes like rhodamines for wastewater treatment. A study of degradation of RhB with UV radiation and a TiO 2 catalyst showed effective degradation via oxidative processes (Natarajan et al. 2011). The rate of UV-degradation of RhB is impacted by the TiO 2 catalyst, pH, and H 2 O 2 indicating the degradation can occur via oxidative and hydroxyl-catalyzed reactions. The impact of oxygen though is complex for rhodamines. Photobleaching experiments at high light intensities illustrate competing effects of oxygen: reaction with the excited state to form photo-oxidation byproducts and efficient deactivation of the excited dye molecule to return to its stable ground state (Zondervan et al. 2004). Investigations of the photobleaching of R6G illustrate how experimental conditions impact the competing reactive pathways. At low light intensities, anoxia reduces photobleaching, thereby increasing the lightfastness of the dye, whereas at high light intensities, the opposite effect was observed. Based upon the complexities of the degradation of the rhodamines, the impact of oxygen on the stability of the fluorescent pigments is not readily apparent and may depend upon experimental conditions including the intensity of the light source (both in accelerated light aging experiments and in realistic gallery exhibition conditions) and the dye concentration in the resin.
The yellow dye in current pigment formulations is a coumarin dye, Solvent Yellow 172 (Sobeck, Chen, and Smith 2021). Coumarins were developed in the late 1960s. They are considered relatively light stable compared to other fluorescent dyes (Sokołowska, Czajkowski, and Podsiadły 2001;Ismael, Schwander, and Hendrix 2013). A study of the impact of air versus nitrogen on the photo-degradation of several coumarins showed increased degradation of the dyes under nitrogen (Sokołowska, Czajkowski, and Podsiadły 2001). Those authors suggest that the inhibition effect of air is indicative of a photo-reductive mechanism for the degradation, potentially at the lactone moiety. Earlier studies of the photophysics of coumarin dyes illustrated the complexity of oxygen in its excited state reactivity. Coumarin dyes are sensitive to the formation of singlet oxygen due to the quenching of its triplet excited state by molecular oxygen, but singlet oxygen does not appear to impact the overall emissive properties of the dye. The singlet oxygen participates in some side reactions, but not the dominant photo-degradation pathways (von Trebra and Koch 1986). In general, coumarin dyes are quite stable amongst the fluorescent dyes used in the modern daylight fluorescent pigments and show limited sensitivity to the ambient environment.

Exhibition and preservation issues
An important outcome of the fading studies of thin washes of fluorescent watercolors was that different paints fade to the same color over time, so visual identification of the specific artists material is not possible and the intended experience is lost (Connors-Rowe, Morris, and Whitmore 2005). Microscopy coupled with spectroscopic analysis of paint samples can reveal original composition from undegraded regions, but there remains the problem of matching or restoring the aged materials. Different methods have been used to color match faded fluorescent colors, including preaging paints and using mixtures of fluorescent and nonfluorescent colorants to color match the original (Beckett, Holden, and Smith 2019;Sharp 2019). Additionally, many of the original pieces were intended to be displayed or worn under blacklights, so the ideal treatment must match the original appearance under both visible and UVA (315-400 nm) blacklights (Winter 2010).
Conservator Kamila Korbela has used the method of pre-aging fluorescent paints in her treatment of Frank Stella's Bampur, dating from 1965, owned by the Los Angeles County Museum of Art (LACMA). Her treatment, coupled with scientific analysis carried out at the LACMA conservation science lab, focused largely on the yellow paint that showed a greater loss of intensity compared to the pink and blue regions of the painting (Sharp 2019). By pre-exposing the restoration paints to intense lighting for a short period of time, a colorant consistent with the faded original paint's appearance can be generated, hopefully generating a seamless color match (De Winter 2018; "Bampur | LACMA Collections" n.d.). Stella's work predates the formulation change in the yellow colorant to coumarin dyes used in current Day-Glo formulations, so the stability of the yellow paint is not anticipated to match those of the modern samples (Sobeck, Chen, and Smith 2021).
A contrast to this conservation approach is the analysis and treatment of similar yellow hues on a Castronovo painted Stephen Sprouse jacket from the 1980s treated at the Indianapolis Museum of Art (IMA) at Newfields (Beckett, Holden, and Smith 2019). The composition of the yellow paint on this jacket is consistent with the dye composition of the modern Day-Glo samples, indicating the presence of a coumarin-based yellow (Sobeck, Chen, and Smith 2021). MFT studies indicated the yellow regions were less fleeting than the pink regions, a contrast to the low relative lightfastness observed for yellow in the Stella piece. Comparison of the relative stability of the colors on the jacket is consistent with that of rhodamines versus coumarins, with the yellow being more stable. Inpainting was carried out on Castronovo's work by using modern fluorescent pigments blended with non-fluorescent colorants of similar hues to color match the original work under both visible and long-wavelength UVA lighting. The inpainting regions appeared different than the original material under short-wavelength UVC light, due to the presence of fluorescent brighteners in the restoration colors, and could potentially allow conservation documentation of the treatment areas, as was reported in the treatment of the jacket (Beckett, Holden, and Smith 2019;Sobeck, Chen, and Smith 2021).
Different approaches have been applied for the preservation and exhibition of works featuring daylight fluorescent materials. One nettlesome consideration about lighting is the artist's intended display environment: outdoor daylight, indoor blacklight, or full-spectrum visible. Both visible and UV regions of the spectrum can induce fading, but there is particular concern regarding damaging high energy UVA radiation, which is typically excluded from museum lighting. One way to address exposure is to use a visitor activated UV lighting system that only illuminates the piece when an observer is viewing it (Winter 2010). A more conservative method is to have a still or digital image of the illuminated piece, though this can be problematic as photography cannot capture the actual enhanced emission intensity and visual effect. Protective casework or coatings can also be considered such as use of a UV filtering layer or anoxic framing to avoid photo-oxidative fading (Winter 2010;Beckett, Holden, and Smith 2019). Our earlier chemical analysis of daylight fluorescent colorants shows the presence of UV absorbers incorporated into the pigment from Radiant Corporation products, and Day-Glo Corporation recommends use of an overcoat containing UV absorber(s) to enhance the fluorescent lightfastness of paints made with their pigments ("Day-Glo Color Corp" n.d.). The impact of the UV absorber in the Radiant pigments, as well as potential for anoxia to stabilize the colorants are considered in the current study.

Present study
In Part 2 of our investigation of daylight fluorescent colors, we report on the spectral properties of the sets of pigments from the major manufacturers Day-Glo and Radiant Corporations, as well as the artist supplier Kremerknown to source their pigments from Radiant. Accelerated fading tests were carried out on mock-up acrylic paint samples using MFT measurements in open air, as well as in enclosed environments in the presence and absence of oxygen. Fading curves were measured for color guides from current and historic mid-century publications to assess how light stability has changed with product formulation over the past decades. Variations in the fading curves are considered in light of the chemical formulation differences between the manufacturers as demonstrated in Part 1 of our study (Sobeck, Chen, and Smith 2021). Further insight into the impact of FRET pairs, between a fluorescent brightener donor and a fluorescent dye acceptor, on lightfastness are probed using a solution phase model. Overall, this set of studies on chemical composition (Part 1) and light stability (Part 2) provide molecular-level insight into the observed visual properties and exhibition considerations of daylight fluorescent pigments. This establishes a basis for analysis and preservation of artworks that utilize this modern class of colorants. Paint sample mock-ups were prepared based upon previous literature (Beckett, Holden, and Smith 2019). Briefly, 0.1 g of pigment was blended into 0.5 g of an n-butyl methacrylate solution acrylic medium (Golden MSA Gel) and 0.5 g of mineral spirits. Drawdowns of these paints on 5 mil Mylar sheets were made by dragging the medium down the sheet between Scotch® magic tape runners (0.0625 mm) using a razor blade. The samples were then allowed to dry in the dark. Samples were made from each of the Day-Glo, Radiant, and Kremer pigments. The Infrared spectrum of the acrylic medium can be found in spectral databases, such as that of the Infrared and Raman Users Group (IRUG) (Price, Pretzel and Lomax 2007).

Solution-phase spectroscopy
Absorbance spectra of dye solutions and pigment extracts in methanol were collected at room temperature on an Agilent Cary 60 Bio Spectrophotometer. Extracts were made by stirring a small amount of pigment (ca. 0.2 g) in methanol (ca. 5 mL) and allowing the undissolved solid to settle. The extraction liquid was then diluted with methanol, if necessary, so the peak absorbance fell within the linear spectral range and around 0.2 absorbance units or less for fluorescence measurements. The instrument was blanked with pure methanol in a 1 cm quartz cuvette, and the absorbance was measured from 250 to 800 nm with a scan rate of 600 nm/min in the dual beam mode.
Fluorescence spectra were collected on a Varian Cary Eclipse Spectrofluorimeter. Excitation and emission spectra were taken at the peak excitation or emission wavelengths for each dye and pigment extract sample. The emission with UV-excitation was taken for each pigment extract sample using the excitation wavelength of 350 nm. The PMT setting was adjusted to achieve sufficient signal to noise. Fluorescence spectra were recorded with a 120 nm/min scan rate, 5 nm slit width, and 1 nm step size.

Microfade testing
A Newport Oriel type microfade tester similar to that first described by Whitmore et al. was used for microfade testing (Whitmore, Pan, and Bailie 1999). The 75 W xenon arc lamp source was filtered to emit only visible light in the wavelength range of 400-700 nm. The luminous flux was measured at 600-1000 millilumens using an ILT1700 radiometer with SPD024Y probe (International Light Technologies). A digital exposure controller adjusted the lamp output to maintain constant illuminance over the course of the experiments.
Spectral reflectance data were acquired from an approximately 400 μm spot every 10 s over 5-10 min exposures and converted into CIE L*a*b* color coordinates. Colorimetry conditions included D65 illuminant, 45°geometry, and 2°observer. Color difference was calculated using the CIE L*a*b* color difference equation from 1976 (ΔE'76). Fading data were analyzed and plotted using SpectralViewer software provided by the Getty Conservation Institute.
Fading measurements were carried out on each paint samples in triplicate, moving the focused illumination to a new spot location for each measurement, and averaged. The paint samples were mounted on plexiglass for stability during MFT measurements. To test the influence of anoxia, identical sets of paint samples were placed in two ESCAL pouches prior to the measurement. One set was purged with nitrogen and then sealed with oxygen absorber (Mitsubishi RPK) and oxygen indicators (Ageless Eye) and allowed to sit in the dark until the indicator showed anoxic conditions were achieved. The oxygen indicators turn pink once oxygen levels fall below 0.1%. A second set was left in an open ESCAL bag to make comparable measurements of the paints with the light filtered by the bag, but with oxygen present. Color swatches from the 1969 DayGlo Designer's Guide were analyzed in a similar manner, with three measurements taken per color analyzed for each image or guide. Blue wool standards 1, 2, and 3 (BWS1-3) were measured in triplicate in room atmosphere prior to each set of experiments for reference, and BWS 1-2 were remeasured in duplicate at the start and end of each day to ensure consistency of their fading over the course of measurements. Each set of MFT measurements (paint samples, anoxia, color swatches) was typically conducted over the course of 2 days.

Light chamber irradiation
2.4.1. Paint mock-up aging Swatches of paints were placed in a Luzchem LZC-ORG photoreactor with a cooling fan and 10 LZC-LWW warm white LED 3000 K (8 Watt) lamps ("Luzchem Research Inc." n.d.). Half of each sample was covered with aluminum foil, and the sample sets were placed in ESCAL pouches. One pouch was sealed with oxygen scavengers and indicators as for the MFT anoxia trials, while the other set was left open, and both sets were exposed in the light chamber for 18 days. The intensity of the visible light at the sample was monitored using a SPER Scientific Advanced Light Meter 840022 throughout the trial (36.7 ± 0.8 klx).

Dye solution irradiation
Sealed cuvettes (Starna Cell fluorimeter quartz cuvette with septum top, 3-Q-10-GL14-S) of a methanolic solution of SY172, prepared to an initial visible absorbance maximum of 1 absorbance unit at 450 nm, were irradiated under two separate exposure regimes, UV and visible, and monitored throughout the irradiation using absorbance and fluorescence spectroscopy. In addition, for each exposure a concentrated solution of fluorescent brightener FB184 in methanol was added dropwise to a separate SY172 cuvette to achieve an absorbance maximum at 375 nm of 0.2 absorbance units while maintaining the 450 nm absorbance maximum of ca. 1. This mimicked the ratio of fluorescent brightener to SY172 absorbance peaks observed in the pigment extract from Radiant Yellow (PC30). Emission spectra were recorded throughout the light exposure experiment on a Cary Eclipse Fluorimeter monitoring the emission profile following visible (450 nm) and UV (350 nm) excitation. The PMT was set at 450 and 500 V for the visible and UV excitations, respectively. The exposure experiment with UV irradiation was carried out using a Rayonet mini-reactor (RMR-600) equipped with a merry-go-round sample holder, a cooling fan (run temp ca. 28°C), and 8 UVA lamps at 350 nm (RMR-3500A). Visible light irradiation was carried out using the same LED light aging chamber described above for the light aging of the paint samples. The fluence of the UV light chamber was measured at the sample using a SPER Scientific UVA/B Light Meter 850009 throughout the experiment (3.0 ± 0.8 mW/cm 2 ). The intensity of the visible light chamber at the sample was monitored using a SPER Scientific Advanced Light Meter 840022 throughout the trial (23 ± 2 klx).

Spectral analysis
Absorbance and emission measurements were compared for the dyes and pigment extracts to understand the color contributions of each pigment's formulation. The absorbance and emission spectra for the pure dye standards are shown in Figure 2. The colors of the dyes are evident from the absorbance maximum, with the colorless fluorescent brighteners absorbing in the UV followed by yellow, red, and violet hues in order of increasing wavelength. The emission spectra appear as mirror images of the absorbance shifted to longer wavelength and the wavelength is indicative of the color. The fluorescent brighteners emit in the blue region of the spectrum, the coumarin in the yellow, and rhodamines on the red edge of visible region. The spectral curves also show distinctive profiles for different chemical classes of additives. For the brighteners, there is a triplet of peaks for FB184, a bis-benzoxazolyl derivative, versus a single Gaussian curve for FB61, a coumarin. The coumarin dye SY172 also has a single broad absorbance and emission peak. The rhodamine dyes have a small shoulder on the short wavelength edge of the absorbance and long wavelength edge of emission. Table 1 summarizes the measured peak absorbance and emission wavelengths for the dye and fluorescent brightener standards.
The methanol extracts of the pigments were assessed using spectroscopy to assess the presence of constituent dyes and brighteners. The full set of excitation and emission spectra are found in the appendices, along with a summary of the extracts' peak absorbance and emission wavelengths. To compare emission from all components and simulate the effect of materials under UV irradiation, emission spectra with UV-excitation are shown in Figure 3. This would correspond to the enhanced color appearance of the materials when viewed under blacklights or UVA radiation typically centered around 350 nm. While the corresponding colors between the Day-Glo and Radiant lines have similar emission at wavelengths greater than 500 nm, their spectral profiles are distinct in the region from 400 to 500 nm. The spectra of the Kremer pigment extracts (not shown) are like those of the Radiant colorants from which they are sourced. The distinctive triplet emission in the blue region of the spectrum for most of the Radiant colorants is due to the presence of FB184, a chemical marker for the Radiant product line (Sobeck, Chen, and Smith 2021). The blue pigments from both manufacturers have a singlet peak in the region found for FB61 in Figure 2. The yellow and green colors have enhanced emission at 500 nm, corresponding to SY172. The orange, red, and pink tones have emission indicative of mixtures of the red and violet rhodamines. Importantly, it has been shown previously that the blue and green pigments get much of their color from a non-fluorescent, insoluble phthalocyanine blue pigment (Fremout and Saverwyns 2014; Boscacci et al. 2020;Sobeck, Chen, and Smith 2021). The fluorescent appearance of these pigments is consequently due to the incorporation of the brightener FB61, as well as the fluorescent SY172 in the green pigments.
In addition to providing insight into formulations and the presence of additives like fluorescent brighteners, the spectral analysis also highlights FRET pairs in the different colors. For example, the fluorescent Figure 2. Normalized absorbance (upper) and emission (lower) for the dye standards. Emission spectra were taken with excitation at the absorbance peak for each compound. Labels correspond to those in Figure 1. Note: Labels correspond to those in Figure 1. brighteners emit around 450 nm, which is the region of maximum absorbance for SY172, making these two components an excellent donor/acceptor pair. This combination is found in the Radiant pigments. Similarly, the basic red rhodamines, R6G and R6GD, emit near the absorbance maximum of the basic violet rhodamines, R3B and RhB. This rhodamine FRET combination is common for orange and red hues from both manufacturers.

Accelerated aging and MFT studies
Acrylic paint mock-ups were prepared for all of the pigments and tested with MFT (Ford 2011). Overall, the Radiant and Kremer paints showed rapid fading, faster than BWS1 in the first minute of fading, with the exception of the violet, blue, and white colors. Artist materials that fade faster than BWS1 are colloquially known as "superfaders," and they are likely to be noticeably damaged in even short museum exhibitions. The paints prepared with Day-Glo pigments displayed slower fading curves, but some of the red and orange paints still exceeded the fading rate of BWS1. The majority of Day-Glo paints faded closer to BWS2 with the blue color fading the least (around BWS3). It is important to note that all of these blue wool equivalent fading rates would place these colors in the fugitive category (Whitmore, Pan, and Bailie 1999;Ford 2011). Figure 4 (upper panel) shows fading curves for a subset of colors from Radiant and Day-Glo. Red, yellow, and blue pigments are compared to illustrate fading for the different classes of colorants, the rhodamines, coumarin, and inorganic pigments, respectively. The MFT curves for all the pigments as well as a summary table of each sample's BWS fading equivalent can be found in the Appendices.
In general, this illustrates that the fluorescent pigments prepared with organic dyes (yellow, orange, red) are more fleeting than those based upon inorganic pigments (blue, green). The yellow and red colors are inverted in stability between the two manufacturers. The primary difference in the formulation of the yellow pigments, based upon our analysis, is the presence of FB184 in the Radiant pigments. The impact of this change is seen in Figure 5, which compares reflectance spectral changes for the yellow hues from the two manufacturers illustrating the rapid drop in the intensity seen in the first minute for Radiant (lower panel) relative to Day-Glo (upper panel). The overall spectral shape remains consistent, just decreasing in intensity over the course of the fading experiment. The red pigment from Radiant also contains FB184 and a UV-absorber (BP3) that are not in the Day-Glo materials (Sobeck, Chen, and Smith 2021).
Previous MFT studies of fluorescent colorants indicated that there may be a recovery of some of the color when the light source is turned off and that the initial rise in the MFT curves can be an artifact of local heating (Metcalf 2014). To test this observation on our paint samples a similar pulsed experiment was carried out where the light beam was turned off for one-minute periods during the course of an 8-minute trial. This was done for both the red and yellow paints from both manufacturers, and the resulting MFT curves are shown in Figure 4 (lower panel). No recovery, beyond the limit of error, was seen in the experiment, and the overall trajectory of the fading curves mirrors the initial, continuous irradiation trial with breaks at the points where the light was turned off. Heating does not appear to account for the rapid initial fading of these colorants.
Color guides from the 1969 Day-Glo Designer's Guide and the current Day-Glo product lines were examined for lightfastness. Two colors, Rocket Red (A13) and Saturn Yellow (A17) were examined since the earlier compositional analysis showed that these colors had changed formulations in the 50-year interim. This allows comparison of how changes in the rhodamines used to create the reds, and the switch to a coumarin-based yellow dye in modern formulations, impact lightfastness. The two lithographed images from the 1969 guide that were tested are shown in Figure 6. An image of the book cover and the included color chart from 1969 that was tested are shown in previous paper (Sobeck, Chen, and Smith 2021). Figure 7 shows a comparison of the MFT curves for the current materials relative to select images in the 1969 guide. The relative stability of the colors is consistent across images tested in the midcentury guide with yellow more fleeting than the red. This matches the relative fading observed visually in Stella's artwork from 1965 (Sharp 2019). Comparing the mid-century pieces to the modern Day-Glo color guide, there is an inversion in the relative stability of Rocket Red and Saturn Yellow. Our previous chemical analysis shows that both reds are rhodamine-based, but the yellow colorants differ between the modern and historic samples (Sobeck, Chen, and Smith 2021). Both formulations of Rocket Red contain a mixture of basic red and violet rhodamines, though rhodamine B has been replaced with rhodamine 3B derivatives in modern formulations. The modern yellow formulations contain Solvent Yellow 172, whereas the historical samples rely on a dye that was tentatively identified as Solvent Yellow 135.

Effects of anoxia
The impact of anoxia upon the lightfastness was compared for the samples by examination of the MFT curves for the paints in ESCAL pouches, which transmit all visible wavelengths but form a gas barrier allowing for anoxic conditions to be achieved using oxygen absorbers. Acrylic paint samples prepared with pigments from all three sources were tested in open and anoxic pouches to monitor changes in the relative fading rates. No appreciable or consistent stabilization was observed, and in fact many of the colors showed a slight increase in the fading rate under anoxic conditions. The results of these trials are summarized in the Appendices. One possible explanation for this result is the extremely high illuminance in MFT measurements (megalux levels), as past studies have shown that the rhodamine dyes have competing photo-degradation and recovery pathways influenced by oxygen (Zondervan et al. 2004). A slower fading experiment under air and anoxia was carried out using a LED light chamber simulating the type of lighting used in many museum galleries. The paint samples showed similar fading trends across the color range for the two sets of paints under each  (BWS1-3, squares). The lower panel shows the pulsed experiment for the red and yellow paints. The curves represent the average of triplicates with error bars showing 1σ. In the lower panels the light beam was turned off during the intervals from 2-3 and 5-6 min, and the data collected during the light-on periods is shown. environmental condition, and again exhibited a slight enhanced fading for the samples that were exposed to anoxic conditions.
The experiments run at high and low light levels provided consistent resultslittle benefit is realized using anoxic exhibition conditions with daylight fluorescent colors. Representative MFT traces and images of the faded samples are found in the Appendices. A comprehensive study of the impact of atmospheric conditions on colorant stability by Beltran, Druzik, and Maekawa showed similar variable effects of anoxia on fluorescent highlighters and overall, less influence of anoxia on lightfastness compared to other materials (2012).

Solution phase analysis of FRET pairs
One notable difference in the MFT fading curves is the relative rate of red versus yellow degradation between the two manufacturers. In general, the paints prepared with Radiant pigments show a sharper rise in the MFT fading curves than those of the corresponding Day-Glo materials. The main difference in the chemical composition of the two lines is the presence of FB184 in Radiant's formulations. It is most pronounced in the yellow pigments, where this is the only formulation difference in the dyes and additives identified between the two manufacturers. As seen in Figure 2, FB184 and SY172 are an excellent FRET pair, and so the effect of this pairing on fluorescence intensity was explored using a solution phase model. Controlled studies were carried out to monitor the emission of SY172 solutions in the presence and absence of FB184, and their relative photostability. Aging studies with both UVA radiation and white light LED illumination were performed. The UVA light source more directly excites FB184, whereas the LEDs mimic museum exhibition conditions, allowing the direct excitation of SY172. Figure 8 shows the absorbance and emission spectra for the solutions throughout the time course of the UV and LED irradiation experiments.
The presence of FB184 is observed in the absorbance spectrum as a shoulder in the spectral region of 350-400 nm on the high energy side of the dominant SY172 peak absorbance at 450 nm, consistent with the profiles found in Figure 2. The fluorescence emission profiles with UV excitation, near the peak of the FB184 absorbance, shows greatly enhanced fluorescence in solutions containing the fluorescent brightener: the  emission intensity is nearly 7 times stronger from the FRET pair under these conditions. In contrast, emission profiles with visible excitation at the peak absorbance of SY172 show negligible differences between the solutions with versus without FB184; reflecting the fact that the brightener is not excited at lower energy visible wavelengths.
White light (visible) aging of the solution causes no observed changes in the absorbance or emission of either dye solution. However, UV-irradiation shows a dramatic decrease in the emission of the dye solution with optical brightener (SY172 + FB184), with the emission dropping to the level of the plain dye (SY172) solution over the course of 1 day. The absorbance spectra indicate the SY172 concentration remains unchanged during the irradiation, but the shoulder from the FB184 absorbance dissipates during this time. The dramatic loss of emission intensity can be fully attributed to the loss of FB184, the FRET donor, and the remaining fluorescence is due to the direct excitation of the yellow dye. The strong ability of FRET partner to enhance the emission intensity, and the sensitivity of FB184 to UVlight exposure, provides a reasonable cause for the pronounced degradation observed in the Radiant and Kremer pigments that take advantage of this FRET pairing in their pigment formulations.

Discussion
Now that the chemical composition of these daylight fluorescent pigments is known (Part 1), the analysis of their spectral properties and light stabilities can be understood to provide insight into the origin of their observed fleeting nature, as well as guidance for their preservation and treatment. FRET pairs are important in enhancing the emission from dyes in the pigments, and the visual impact of the colors. Notably, fluorescence brighteners can serve as excellent FRET donors for yellow dyes, like SY172 in this case, and greatly enhance the fluorescent appearance under lights containing wavelengths less than 400 nm. A comparison of this effect can be made through visualization of the paint samples in Figure 9. The Day-Glo and Radiant paint samples are displayed under visible, UVA, and UVC lamps. Chemical analysis shows that Radiant paints contain fluorescent brighteners in all formulations, whereas Day-Glo only uses a brightener for the blue color. The UV-irradiation causes the Radiant paints to "pop," particularly in the higher energy UVC region, and show small variations between the different UV-wavelengths that is not observed in the absence of the UV-absorbing fluorescent brighteners. This is particularly evident in the tones that contain more of the yellow hues including the yellow, green, and orangish yellow colors.
This observed difference in the fluorescent appearance of the colorants containing fluorescent brighteners can provide a means to track conservation inpainting treatments and for possible assessment of the original pigment source. As was previously reported, inpainting on a Castronovo fluorescent Mona Lisa painting retouched with Kremer fluorescent yellow could be recognized under UVC irradiation even though the inpainting was engineered to match the original paint under UVA and visible illumination (Beckett, Holden, and Smith 2019). Subsequent re-analysis of the original fluorescent yellow paint in Part 1 of this study revealed that it matched the Day-Glo composition. This behavior of the original and retouching paints when viewed under UVC aligns with our chemical and spectral analysis and matches the behaviors shown in the multispectral images in Figure 9. The Kremer conservation pigments used in the inpainting are sourced from Radiant and contain fluorescent brighteners, whereas the original artist's material's formulation from Day-Glo do not (Sobeck, Chen, and Smith 2021). The optical brighteners have strong UV absorbance, and their subsequent emission in the long-wavelength UV to high energy visible region is distinct (Beckett, Holden, and Smith 2019;Springsteen 1999). This emission from the brightener further intensifies the fluorescence emission of the paint with UVC irradiation. The difference in appearance for regions with versus without the fluorescent brightener to serve as a FRET donor allows for documentation and location of conservator's retouching in this situation (Beckett, Holden, and Smith 2019;Hickey-Friedman 2002).
The use of pigments with the Radiant formulation may provide a viable way to achieve the proper color- match to Day-Glo paints when their use is confirmed in original artwork, but the presence of the fluorescent brightener can serve as a means of documenting the inpainting using a short-wavelength UV light source, so long as the brightener remains undegraded by exposure. The use of these high energy UV lamps for documentation of inpainting should be practiced with caution due to the sensitivity of the pigment's constituent dyes and additives to high-energy irradiation as demonstrated in these experiments. The object is more sensitive to UV exposure than to white light sources like LEDs, so it is advisable to minimize their exposure time to prevent further fading (Michalski 1987).
Unfortunately, the chemical instability of the fluorescent brighteners is apparent in the solution phase studies presented in Section 3.4. The Radiant pigments' formulation rely on these brighteners and show a rapid degradation in accelerated light aging experiments. As the MFT studies show, paints prepared using the Radiant pigments have an initially rapid loss of color relative to those prepared with Day-Glo materials. Following from the solution-phase studies this is attributed to the rapid degradation of the fluorescent brightener, which serves as a FRET donor for the yellow dye SY172 in these materials and greatly enhances the emission from it. Although the SY172 itself is relatively stable, the loss of the FRET pair causes a rapid decay in its perceived emission, making the pigment appear to darken. This is observed in the marked differences in the relative stability monitored by MFT for the two manufacturers' yellows and the particularly fugitive nature of the Radiant yellow pigment that gains much of its intensity from the FRET pair with FB184. The instability of the fluorescent brightener does appear, though, to be very sensitive to the wavelength of excitation, showing no degradation under LED lights that exclude UV wavelengths. The observed sensitivity of the fluorescence brightener in solution to UV-exposure, though simplified from the paint matrix, provides a likely explanation for the perceived fast fading observed in the MFT studies. Future studies are in process to more fully quantify and assess the photochemical properties of the dyes used in the daylight fluorescent materials.
Analysis of these materials in artwork when possible can inform display practices, and the use of lighting that has UV-components should be employed with caution when fluorescent brighteners are known to be in the artist's materials. Past studies show that simulated daylight exposure leads to loss of brightener content in photographic papers and moderate color changes around BWS3-4 (Connors-Rowe, Whitmore, and Morris 2005). The present study shows a more significant instability of the brighteners in the daylight fluorescent materials. In order to minimize exposure of the works to UV, an exhibition may include images of the piece under UV-lighting to illustrate the impact of light source on color or use visitor-activated lighting to provide the direct experience to the visitor in short, controllable episodes.
The light stability as assessed by MFT of pieces from the late 1960s in the Day-Glo Designer's Guide signals changes in formulation of the colorants over time.
Most notable is the change in the yellow dye from the early formulations of the 1960s to that used starting in the 1980s, and a corresponding enhancement in dye stability with this change. This is consistent with the observed instability of the Saturn Yellow color in the works of Frank Stella (Sharp 2019). Further work is needed to identify the precise colorant in the earlier formulation, but the present work indicates the pigments used in the paints after the 1980s will likely not have as significant of conservation issues as the precursors. The rhodamine dyes, responsible for the red and violet hues, tend to be less stable and warrant further studies to understand their degradation in the resin matrix of the pigments. The degradation of the red hues is most pronounced in the modern color palate, as was reported with the fast decays observed in MFT studies on the Sprouse Jacket (Beckett, Holden, and Smith 2019).
Overall, the research in Part 2 of this study is consistent with past reports of the fleeting nature of the daylight fluorescent pigments, particularly those based upon organic dyes (Connors-Rowe, Morris, and Whitmore 2005; Metcalf 2014; Beckett, Holden, and Smith 2019). This work indicates the impact of particular dye and additive combinations on the appearance and stability of this complex class of material. It does not appear that oxygen plays a critical role in the degradation processes for the pigments, and considering the costs of anoxic displays, this approach is not likely to be an effective method to decrease fading of exhibited fluorescent artworks. Light intensity and wavelength, though, do appear to have a great impact upon the preservation or relative stability of daylight fluorescent pigments. Depending upon the precise pigments used, and their formulation, inclusion of UV can significantly diminish the emission over time and will quickly affect the perceived experience of the viewer. Further work is warranted to examine this impact across all colors, particularly those that use combinations of red and yellow, which was not specifically studied here. Because of the UVA sensitivity shown in these experiments, a careful study of visible light sources of various correlated color temperatures is warranted; cooler light sources might accelerate degradation rates over warmer color temperature lights, although yellower illumination might lack the higher energy wavelengths necessary to fully activate FRET pairs based on fluorescent brighteners.

Conclusions
Our comprehensive examination of the chemical composition (Part 1) and spectral properties (Part 2) for two sets of daylight fluorescent pigments provides fundamental insight into their observed behavior as an artist's material. These investigations help to inform conservation and exhibition practices for objects, particularly highlighting the ability to identify the original materials' source through analysis. It is noted that the small difference in formulation, like the addition of fluorescent brighteners in the Radiant (and Kremer) pigments, can greatly impact the stability and appearance of a color under light sources that include UVwavelengths. Finally, changes in the types of dyes used in the formulations have occurred between the early daylight fluorescent materials of the 1960s and those since about 1980. Most notable the yellow dye has been changed, but the classes of rhodamines for the red and violets hues are also different; these changes have somewhat improved the light stability of the modern fluorescent palette. Future studies are planned to more fully assess the fundamental photochemistry of mixtures of dyes found in this complex modern artist's material and the insight it can bring to their preservation.