Validation of the reciprocity law for coating photodegradation.

Accelerating the photodegradation of commercial polymeric materials has great practical importance in the weathering community. However, questions exist as to whether high radiant flux exposure results can be extrapolated to in-service exposure levels. Based on the reciprocity law, the photoresponse of a material is dependent only on the total energy to which the specimen is exposed, and is independent of the exposure time and the intensity of the radiation taken separately. An experiment to validate the applicability of the reciprocity law for polymeric coatings has been carried out using the NIST integrating sphere-based ultraviolet (UV) weathering device. A nonpigmented, non-UV stabilized acrylicmelamine coating was exposed to six different UV radiation intensities ranging from 36 W/[m.sup.2] to 322 W/[m.sup.2], and in the spectral region between 290 nm and 400 nm. Chemical changes in the coating due to UV exposure were measured with transmission Fourier transform infrared (FTIR) spectroscopy and UV-visible spectroscopy. Using two dose-damage models, the reciprocity law photoresponse for this polymeric system was verified for different photodegradation mechanisms, including chain scission, oxidation, and mass loss.

Keywords: Reciprocity law, Schwarzschild law, acrylic-melamine, photodegradation, ultraviolet radiation, FTIR, ATR, photochemistry, photodegradation, acrylics, accelerated testing, durability, service life prediction, weatherability, melamine, UV spectroscopy

**********

Polymeric materials used in an outdoor environment are subjected to a wide variety of stress factors, including solar ultraviolet (UV) radiation, temperature, and moisture. The physical and chemical degradation resulting from outdoor exposure is known in the materials community as weathering. For commercially viable materials, weathering is usually a slow and lengthy process, often taking five years or longer before a critical performance property is deemed to have failed. These long testing periods are costly to the manufacturer and are also a barrier to innovation and the timely introduction of new products into the market. A need exists for developing exposure strategies that can accelerate weathering and yet permit valid extrapolations or predictions of service life from accelerated levels to in-service exposure conditions.

Although temperature and moisture acting alone are important stress factors in the degradation of many materials, UV radiation, in conjunction with temperature and moisture, plays a dominant role in the photodegradation of polymeric materials. It is well known that polymer photodegradation can be accelerated by increasing the radiant flux to which the material is exposed. The law of reciprocity states that the photoresponse of a material is dependent only on the total energy involved and is independent of the intensity (or irradiance) of the radiation and the exposure time taken separately. (1) Experiments in which the photoresponse of a material is measured as a function of irradiance levels are commonly called reciprocity experiments.

Over the last 100 years, the use of high irradiance exposures has been successfully employed by the biological, medical, and materials weathering communities. In materials weathering, for example, high solar irradiance field exposure devices have been recognized as a valid method for accelerating weathering since the early 1960s, (2,3) where claims of acceleration ranging from three to eight times greater than normal are not uncommon. Although accelerating outdoor weathering results obtained via high irradiance exposures are accepted by the materials community, laboratory-based high UV irradiance exposure devices have not gained the same level of acceptance. The objectives of this study were to (1) ascertain the relationship between the level of applied UV irradiance in a laboratory device and the photoresponse of a model acrylicmelamine polymer, and (2) to determine whether the reciprocity law is valid for this particular polymeric system.

OVERVIEW OF THE RECIPROCITY LAW

From the 1880s through the early part of the 20th century, the majority of reciprocity law experiments were performed on photographic materials. At the turn of the 20th century, the medical profession independently validated the reciprocity law for applications including erythema (reddening of the skin) and phototherapy (the use of high ultraviolet radiant fluxes to cure tuberculosis and rickets). In the 1930s, these medical studies were extended to biological applications, including the purification of water and the determination of the biocidal efficiencies of different spectral wavelengths and different light sources in inactivating viruses, bacteria, fungi, and mold. In comparison to the other fields of study, the use of high UV irradiance in the accelerated weathering of polymeric materials has been limited, largely due to the belief that operating at high radiant fluxes may introduce unnatural failure mechanisms into the weathering process. (4)

[FIGURE 1 OMITTED]

Bunsen and Roscoe (1) are credited with deriving and conducting the first reciprocity law experiments on photographic materials. They concluded from their experiments that all photochemical reaction kinetics depend only on the total absorbed energy and are independent of the two factors that determine total energy, i.e., the intensity or irradiance, I, and the exposure time, t. This hypothesis later became known as the law of reciprocity because, in photography, the quality of a series of photographic or radiographic films will be uniformly constant if the exposure times to which the films are subjected vary reciprocally with the intensities of the exposing radiation. (5) For a given photoresponse, the reciprocity law states that

[FIGURE 2 OMITTED]

It = constant (1)

Deviations from the reciprocity law are called reciprocity law failure.

Since the reciprocity law depends only on total absorbed energy, validation of the reciprocity law for a material can have many experimental manifestations, as shown in Figure 1. Each manifestation is equivalent to the others as long as the integrated total absorbed energy is the same for each exposure regime. (6) Thus, assuming that the reciprocity law is obeyed, the same photoresponse should be observed when exposures are made: (1) at a high radiant flux for a short period of time (e.g., Figure 1a); (2) at a low radiant flux for a long period of time (e.g., Figure 1b); (3) by repeatedly cycling a light source on and off and controlling both the on/off frequency of the light and the length of time that the light remains in the on and the off state (e.g., Figure 1c). Experiments in which the light is turned on and off at an extremely high frequency are called flash photolysis experiments, while experiments in which the light is turned on and off at a low frequency are called intermittency experiments; or (4) by ramping the radiant flux to a given level, holding at the flux level for a specified period of time, and then ramping back down to a lower level of flux, or any other variant of these cycles (e.g., Figure 1d).

Shortly after Bunsen and Roscoe published their results, papers began to appear challenging the validity of the reciprocity law. (7-11) Reciprocity law failures were most commonly observed for experiments conducted at either very low or very high irradiance levels. To account for these failures, Schwarzschild, an astronomer, proposed a modification of the reciprocity law that fit his low intensity stellar data. (12) This model later became known as Schwarzschild's law and is given by

I[t.sup.p] = constant (2)

where p is the Schwarzschild coefficient, hereafter referred to as the p-coefficient. Note that when p = 1, Schwarzschild's law and the reciprocity law are identical and hence, Schwarzchild's law is a generalization of the reciprocity law. Schwarzschild originally postulated that p was a constant having a value of 0.85; however it quickly became apparent that p varies from material to material and in some cases, within a material, depending on the irradiance range over which the material was exposed.

[FIGURE 3 OMITTED]

[FIGURE 4 OMITTED]

Although the reciprocity and Schwarzschild's laws are the most common models used in describing the photoresponse of a material as a function of radiant intensity, many other models and graphical techniques have been proposed. (13,14) Few of these techniques have gained widespread acceptance either within or outside the photographic field. One technique is the graphical technique popularized by Halm (15) that is based on Kron's catenary equation, (16) having the form

log It = constant + log [(I/[I.sub.0])[.sup.a] + (I/[I.sub.0])[.sup.-a]] (3)

where [I.sub.0] and a are constants having values that vary from emulsion to emulsion. At low intensities, equation (3) reduces to:

log It = constant - alogI (4)

where a is related to the p-coefficient via:

p = (1 [+ or -] a)[.sup.-1] (5)

Equation (4) is used as the basis for the plot shown in Figure 2, where the log dose (log It) necessary to achieve a given photoresponse is plotted against log irradiance (log I). (13) When the log It versus log I plot is linear and parallel to the abscissa, then the reciprocity law is assumed to be obeyed. Correspondingly, when the log It versus log I plot is nonlinear (that is, it has a catenary shape), then reciprocity law failure is said to have occurred.

Another method of representing reciprocity data was introduced by Arens and Eggert in 1928, (17) in which log I is plotted versus log t, where t is the time necessary to achieve a given photoresponse (see Figure 3). If reciprocity is obeyed, the plotted curves are straight lines with slope = -1. If reciprocity failure is observed, then the slope of these curves at any point represents the value of the Schwarzschild p-coefficient.

[FIGURE 5 OMITTED]

Reciprocity experiments have been carried out on a small number of polymeric systems, with limited data reported in the literature. Much of this published data has been compiled in an extensive review by Martin et al. (18) For most of the polymeric systems studied, Schwarzschild's law appears to adequately model their response to reciprocity, with the p-coefficient generally falling between 0.5 and 1.0. Chen (19) measured the molecular mass decrease in polystyrene following exposure to various intensities of UV radiation from a mercury lamp and found that the rate of polystyrene degradation was proportional to the first power of the incident UV intensity. The depolymerization of poly-[alpha]-methylstyrene studied by Stokes and Fox, (20) and photooxidation of acrylonitrile butadiene-styrene (ABS) terpolymers by Davis and Gordon, (21) also showed a first order dependence on the incident irradiance. Okada (22) also showed that the gel content in polyethylene was a function of the total X-ray dose delivered, and not of the irradiance levels used. It was documented by Schafer (23) that the UV irradiation time necessary to achieve a 50% decrease in impact strength in poly(vinyl chloride) was linear with respect to irradiance level. In contrast, reciprocity failure was observed in a study of the solar absorptance of space coatings by Olson et al. (24)

[FIGURE 6 OMITTED]

More recently, Pickett (25,26) documented that the relative rate of yellowing for modified poly(phenylene oxide) (PPO) and ABS was not a linear function of light intensity, and concluded that the yellowing response of polycarbonate, poly(butylene terephthalate), poly(styrene-acrylonitrile), and a polycarbonate/poly(butylene terephthalate) blend did appear to obey the reciprocity law. Using the UV concentrator at the National Renewable Energy Laboratory (NREL), Jorgensen et al. (27) reported that in the photooxidation of acrylic coatings, Schwarzschild's law was obeyed with p-coefficients ranging from 0.667 to 0.706. In the same study, poly(vinyl chloride) demonstrated a p-coefficient of 0.669, and polycarbonate was shown to obey the reciprocity law with a p-coefficient of 1.093.

EXPERIMENTAL*

Specimens

A model, nonpigmented, non-UV stabilized, acrylicmelamine coating was selected for this study. This particular thermoset acrylic-melamine has been extensively studied with and without UV stabilizers, (28,29) and a mechanism for photodegradation has been proposed. (30) Based on extensive FTIR analysis, the proposed structure of the crosslinked coating is shown in Figure 4. (31) The coating was composed of a hydroxy-terminated acrylic polymer and a partially-alkylated amine crosslinking agent combined in a mass ratio of 70:30. The acrylic polymer contained 68 mass % normal butylmethacrylate, 30 mass % hydroxyl ethylacrylate, and 2 mass % acrylic acid; and was supplied as a solution of 70.2 mass % polymer and 29.8 mass % heptanone solvent. The amine crosslinking agent was Cymel 325 (Cytec Industries), a solution of 80 mass % melamine-formaldehyde resin in isobutanol.

Calcium fluoride (Ca[F.sub.2]) windows with a diameter of 19 mm and a thickness of 4 mm were used as substrates for the exposure studies. Ca[F.sub.2] is transparent in the spectral range from 0.13 [micro]m to 11.5 [micro]m, making it suitable for both infrared and UV-visible spectroscopy measurements. Solutions of acrylic polymer and melamine-formaldehyde crosslinking agent were hand-mixed in the required ratio, degassed in a vacuum oven for 30 min at room temperature, and spin-coated onto the Ca[F.sub.2]. Coatings were then cured at 130[degrees]C for 20 min. The average thickness of the cured coatings was 11.4 [micro]m [+ or -] 0.5 [micro]m, as measured by X-ray reflection; the glass transition temperature as determined by differential scanning caloriometry was 45[degrees]C [+ or -] 2[degrees]C.

High Radiant Flux Exposure

The acrylic-melamine coated (Ca[F.sub.2]) specimens were inserted into holders consisting of aluminum disks with slots machined out to hold the specimens. The specimen holder configuration, shown in Figure 5a, contains 17 symmetrically arranged specimen windows. An identical holder was used to hold the neutral density filters, which have varying transmittance values and are used to control the UV irradiance incident on the specimens. The specimen and filter holders were aligned in the layered assembly shown in Figure 5b, so that each specimen was paired with a neutral density filter positioned between itself and the UV light source. Each filter holder can hold multiple stacked filters in each slot, so that additional filters (such as bandpass filters) could also be incorporated into the experiment if desired. A quartz plate in the filter holder protects the filters from the potentially degradative effects of moisture in the ambient environment.

Specimens and filters were randomly assigned between four specimen holders, with eight replicates for each irradiance level. In each specimen holder, the center specimen was not paired with any neutral density filters, but instead was blocked completely from UV exposure. This specimen served as the "dark" control to ensure that changes in the coating chemistry that were not due to photolysis (e.g., hydrolysis) could be accounted for.

Target transmittance values of 10, 20, 40, 60, 80, and 100% (full UV) exposure were selected for the thin film neutral density filters used in this study. The actual measured transmittance values in the spectral region between 290 nm and 400 nm were nominally 9.7, 20.5, 41.8, 58.0, 76.0, and 86.4%. Hereafter, the filters will be referred to by the target transmittance values. UV-visible transmission spectra for the filters are shown in Figure 6.

High UV irradiance accelerated exposures were carried out in the NIST 2 m integrating sphere-based weathering device, referred to as SPHERE (Simulated Photodegradation via High Energy Radiant Exposure). The high UV irradiance output of the SPHERE makes it possible for reciprocity experiments to be carried out over a wide irradiance range and in a reasonable test period. The mercury arc lamp system used with SPHERE produces a collimated and highly uniform UV flux of approximately 480 W/[m.sup.2] at each of 32 exit ports; the UV flux incident on the specimens ranged from 36 W/[m.sup.2] to 322 W/[m.sup.2]. The irradiance level corresponding to each of the neutral density filters used is reported in Table 1. The use of a borosilicate glass window between the lamp system and the integrating sphere eliminated all UV wavelengths < 290 nm, while the use of a dichroic reflector in the lamps removed wavelengths > 450 nm. Additional details on the construction and properties of SPHERE have been published elsewhere. (32) Exposure of the four specimen holders was carried out simultaneously under ambient temperature and humidity conditions (nominally 22[degrees]C and 30% RH).

[FIGURE 7 OMITTED]

Measurements

Damage in the acrylic-melamine specimens was assessed via changes in the peak height of specific infrared absorbance bands, using a Perkin Elmer 1760X Fourier transform infrared (FTIR) spectrometer equipped with a mercury-cadmium-telluride (MCT) detector. Dry air was used as the purge gas. FTIR spectra were recorded between 4000 [cm.sup.-1] and 800 [cm.sup.-1] at a resolution of 4 [cm.sup.-1], and were averaged over 128 scans, using an autosampling device previously described. (31) This autosampler allowed spectra to be recorded at approximately the same location on the specimens during each inspection and thus minimized error due to differences in sampling location. The uncertainty associated with the FTIR absorbance measurements is typically [+ or -] 2%.

The FTIR spectra of the bare, uncoated Ca[F.sub.2] disks were recorded and subtracted from the spectra of the coated disks. Changes in the absorbance intensities of the bands of interest were obtained by subtracting the FTIR spectrum of the unexposed specimen from the spectrum of the UV-exposed specimen taken at time t. Changes in band absorbance were measured on the resulting difference spectrum, and are hereafter referred to as "damage." These changes in the FTIR absorbance bands were plotted versus time as well as dosage.

An important parameter in a reciprocity experiment is the dosage, defined as the radiation that is incident upon and absorbed by the specimen, which is calculated via:

Dosage = It (1-[e.sup.-A([lambda])])

where I is the source irradiance, t is time, and (1-[e.sup.-A([lambda])]) is the spectral absorption of the specimen. For the purpose of calculating dosage, the spectral irradiance of the light source; the spectral transmittance of the filters, quartz plate, and coated Ca[F.sub.2] specimens; and the UV-visible absorbance spectrum of the coated Ca[F.sub.2] specimens were recorded with a HP 8452a UV-visible spectrophotometer between 190 nm and 820 nm at a resolution of 2 nm. UV-visible measurements were taken each time that an FTIR measurement was taken on the specimens. Table 2 summarizes the measurement protocol and schedule for the initial measurements as well as for measurements taken throughout the study.

[FIGURE 8 OMITTED]

To calculate dosage, the spectral irradiance of the SPHERE UV source was multiplied by the spectral transmittance of the neutral density filters and the quartz window to yield the dose incident upon the specimen, which is then convoluted with the UV-visible absorbance spectrum of the specimen to yield the dosage (for a more detailed explanation of this calculation, see references 33-35). Hereafter, the term "dosage" is used to refer to this quantity, which is the energy actually absorbed by the specimen. This term is not to be confused with the dose, defined as the energy incident upon the specimen. A custom software program has been developed at NIST to catalogue and analyze the IR and UV-visible spectra, and to calculate dosage and damage from the measurements previously described. (36)

In order to convert the UV-visible spectrometer signal into irradiance units, the HP 8452a UV-visible spectrometer was calibrated in the Spectral Irradiance and Radiance Calibrations with Uniform Sources (SIRCUS) facility in the NIST Physics Laboratory. The responsivity of the UV-visible spectrometer to a collimated, tunable laser was compared to the known spectral power responsivity of a reference detector (which, in turn, is calibrated against a cryogenic radiometer). These measurements yield a calibration curve that allows the UV-visible signal measured by the spectrometer to be converted into an absolute power measurement in units of W/[m.sup.2].

[FIGURE 9 OMITTED]

RESULTS AND DISCUSSION

A representative FTIR transmission spectrum of the unexposed acrylic-melamine coating, superimposed on the spectrum of the same coating following 1000 hr of UV exposure in the SPHERE under ambient conditions, is shown in Figure 7a. The difference spectrum resulting from the subtraction of the non-UV exposed spectrum from the UV exposed spectrum is shown in Figure 7b. Following 1000 hr of UV exposure, the bands at 1085 [cm.sup.-1], 1730 [cm.sup.-1], and 2960 [cm.sup.-1], attributed to ether C-O stretching in the crosslinks, ester C=O stretching in the acrylic side chains, and aliphatic C-H stretching, respectively, (34) had decreased in intensity. The peak at 1670 [cm.sup.-1], attributed to amide C=O stretching, had increased in intensity. The loss of the C-O crosslink and the formation of amide moieties is consistent with the photooxidation mechanism proposed by Lemaire. (30) These four bands are the primary infrared bands of interest for quantifying damage in the acrylic-melamine coating. The 1085 [cm.sup.-1] band was used to follow chain scission, and the band at 1670 [cm.sup.-1] was used to follow photooxidation. Changes in coating mass loss due to degradation were followed by monitoring the intensity changes in the C-H stretching band at 2960 [cm.sup.-1] and the acrylic ester C=O stretching band at 1730 [cm.sup.-1]. No changes were observed in the FTIR spectrum of the dark specimen (no UV) during the 1080 hr of exposure in the SPHERE, indicating that essentially no competing reactions (e.g., hydrolysis, thermal degradation) had occurred.

The UV-visible spectra of the acrylic-melamine coating before and after 1080 hr of UV exposure are shown in Figure 8. Prior to UV exposure, no significant absorbance was observed above 280 nm. Following UV exposure, minor peak broadening was observed in the region below 300 nm.

Changes in the intensities of the aforementioned infrared bands as a function of UV irradiance level are shown in Figures 9a-9d, where each irradiance level is identified by the corresponding neutral density filter. The reproducibility of the data is indicated by one standard deviation as shown for the 100% filter data. For each irradiance level, the data could be fitted by smooth monotonic curves that are well separated. This data was used to develop the damage versus dosage curves, and will also be discussed in the context of the log irradiance versus log time plots later in this section.

[FIGURE 10 OMITTED]

When damage is graphed as a function of dosage, the individual curves for each irradiance level superimpose onto a single curve, as shown in Figures 10a-d. This observation provides an initial confirmation that reciprocity is obeyed for this coating system. The superposition of data from all irradiance levels in the damage-dosage plots is indicative that any point on the master curve represents an equivalent level of damage for all specimens, independent of the UV irradiance, to which individual specimens were exposed. Thus, it can be inferred that the dosage needed to cause a given level of damage is not a function of the incident intensity.

Similar superimposed curves (not shown) were obtained when the damage was plotted versus "dose," indicating that the UV absorbance of the acrylic-melamine polymer does not change significantly with exposure time in the spectral range of the SPHERE UV source, in good agreement with the UV-visible data presented in Figure 8. Thus, in this instance, dose and dosage can be used interchangeably.

To further confirm that reciprocity is obeyed for this coating system, Kron's constant density curves (log It versus log I) (16) were constructed from the damage-dosage data. Working with the plots of damage versus dosage, a horizontal line was drawn on each plot through the last data point on the 10% neutral density filter curve. The logarithm of the dosage value corresponding to the intersection of the horizontal line and the curve for each neutral density filter was plotted against the logarithm of the integrated spectral irradiance transmitted for that neutral density filter (the spectral irradiance is integrated between 300 nm and 400 nm for each neutral density filter). Plots of log dosage (log It) versus log irradiance (log I) are shown in Figures 11a-d for the 1085 [cm.sup.-1], 1670 [cm.sup.-1], 1730 [cm.sup.-1], and 2960 [cm.sup.-1] bands. Taking the standard deviations into account, the linear regression lines plotted through the data points in these figures were generally observed to be parallel to the abscissa, indicating that the reciprocity law is obeyed for these modes of damage.

[FIGURE 11 OMITTED]

The reciprocity law was also verified for this coating system by plotting log irradiance (log I) versus log time (log t) via the method of Arens and Eggert. (17) "Time" is defined as the period of exposure needed to achieve a given photoresponse, or in this case, a specified change in infrared absorbance. Working with the damage versus time plots displayed in Figures 9a-d, an FTIR absorbance was chosen such that a horizontal line drawn on the plot at that absorbance level would cross over the plotted data from all six UV irradiance levels. The logarithm of the time value corresponding to the intersection of the horizontal line and the curve for each neutral density filter was plotted against the logarithm of the integrated spectral irradiance transmitted for that neutral density filter. If reciprocity is obeyed, log I versus log t plots should yield a straight line with a slope of -1. If reciprocity is not obeyed, then the slopes of the lines are equal to -p, where p is the Schwarzschild coefficient.

In the log I versus log t plots shown in Figures 12a-12d, linear regression lines could be fitted to the data with correlation coefficients of 0.98 or better. The slopes of these lines, or the p-coefficients, are shown in Table 3. The p-coefficients for the damage modes corresponding to the 1085 [cm.sup.-1], 1730 [cm.sup.-1], and 2960 [cm.sup.-1] bands are close to 1, indicating that reciprocity is obeyed for all modes of photodegradation for this coating system. A slightly higher pcoefficient for the 1670 [cm.sup.-1] band (attributed to amide formation) suggests that this particular degradation mode is more complex than the chain scission or mass loss reactions, and may involve competition between amide formation and depletion.

It should be noted that the reciprocity law has been demonstrated for a model acrylic-melamine coating exposed to ambient temperature and relative humidity, nominally 22[degrees]C and 30% RH. Work is in progress in our laboratory to validate the reciprocity law for other coating systems (including pigmented and UV-stabilized) under a wider range of temperature and relative humidity conditions.

[FIGURE 12 OMITTED]

SUMMARY AND CONCLUSIONS

Accelerating the photodegradation of polymeric materials is of great interest to producers of commercial polymeric materials as well as to end-users of polymeric materials in outdoor applications. Using the high irradiance UV source in the NIST SPHERE, an experiment has been carried out to validate the reciprocity law using a nonpigmented, non-UV stabilized acrylic-melamine coating exposed to six different irradiance levels. The UV irradiance ranged from 36 W/[m.sup.2] to 322 W/[m.sup.2] and was controlled using neutral density filters.

Chemical degradation, including chain scission, oxidation, and mass loss, in the acrylic-melamine coating, as a function of time and dosage, was measured with transmission Fourier transform infrared (FTIR) and UV-visible spectroscopies. The results were analyzed via various models of the reciprocity law. Based on the results, the following conclusions are made:

(1) When damage is expressed as a function of time, both the rate and magnitude of the degradation increase with increasing irradiance level and exposure time.

(2) When damage is expressed as a function of the dosage, damage-dosage curves obtained at different irradiance levels superimpose onto a single curve.

(3) All photodegradation processes of the nonpigmented, non-UV stabilized acrylic-melamine coating, including chain scission, oxidation, and mass loss, obey the reciprocity law, with Schwarzschild p-coefficients of approximately 1.

The ability to weather polymeric materials at high UV irradiance levels and confidently extrapolate the results to in-service levels has tremendous implications in accelerated weathering test methods. If information on the photoresponse of a polymer to elevated UV irradiance levels can be obtained, then accelerated aging results can be extrapolated to in-service conditions with a high degree of confidence. Carrying out predictive tests by exposing materials to high UV irradiance for short periods of time will greatly enhance the polymer industry's ability to bring new products to market and minimize liability due to warranty failures.

Table 1 -- Irradiance Levels Corresponding to Neutral Density Filters

                                          Irradiance
Neutral Density Filter Transmittance (%)  (W/[m.sup.2])

 9.7                                       35.6
20.5                                       74.7
41.8                                      154.1
58.0                                      219.9
76.0                                      278.9
86.4                                      321.9

Table 2 -- Measurement Protocol and Schedule

                         Time = 0
                         (Initial Measurements)  Time = t > 0
                         UV-Visible   Infrared    UV-Visible  Infrared

ND filters               # (a)
Quartz windows           #         #                 #
Ca[F.sub.2] windows             #       #            #           #
(uncoated)
Ca[F.sub.2] windows             #       #            #           #
(with acrylic-melamine)
Port irradiance--               #                    #
with filters
Port irradiance--               #                    #
without filters

(a) # indiciates that a measurement was taken.

Table 3 -- P-coefficients Calculated from Log Irradiance versus Log Time
Plots

FTIR Band ([cm.sup.-1])  p-Coefficient [+ or -] One Standard Deviation

1085                     1.09 [+ or -] 0.18
1670                     1.22 [+ or -] 0.10
1730                     1.05 [+ or -] 0.18
2960                     0.97 [+ or -] 0.10

ACKNOWLEDGMENTS

The support of the NIST Coatings Service Life Prediction Consortium, including Dow Chemical, Atofina, Akzo Nobel, Sherwin Williams, Atlas Material Testing Technologies LLC, the Air Force Research Laboratory, the Federal Highway Administration, and the Forest Products Laboratory is gratefully acknowledged. The authors also thank Dr. Keith Lykke and Dr. Steve Brown of the NIST Physics Laboratory for carrying out the irradiance calibration of the UV-visible spectrometer used for dosage measurements.

Presented at the 82nd Annual Meeting of the Federation of Societies for Coatings Technology, October 27-29, 2004, in Chicago, IL.

* Certain trade names and company products are mentioned in the text or identified in an illustration in order to adequately specify the experimental procedure and equipment used. In no case does such an identification imply recommendation or endorsement by the National Institute of Standards and Technology, nor does it imply that the products are necessarily the best available for the purpose.

References

(1) Bunsen, R.W. and Roscoe, H.E., Ann. Phys., 108, 193 (1859).

(2) Caryl, C.R. and Rheineck, A.E., "Outdoor and Accelerated Weathering of Paints," Official DIGEST, 34, No. 452, 1017 (1962).

(3) Davis, A. and Sims, D., Weathering of Polymers, Applied Science Publishers, New York, NY, 1983.

(4) Gerlock, J.L., Kucherov, A.V., and Smith, C.A., Atlas Sunspots, 31, 1 (2001).

(5) Morgan, R.H., Radiology, 42, 471 (1944).

(6) Forbes, P.D., Davies, R.E., and Urbach, F., J. Invest. Dermatol., 73, 131 (1979).

(7) Scheiner, J., Bulletin du Comite Permanent International pour I'Execution Photographique de lu Carte due Ciel, 1, 227 (1889).

(8) Abney, W.D.W., The Photographic Journal, 18, 56 (1893).

(9) Abney, W.D.W., Proc. Royal Society of London, 54, 143 (1893).

(10) Abney, W.D.W., The Photographic Journal, 18, 302 (1893).

(11) Englisch, E., Photograhische Korrespondenz, 36, 109 (1899).

(12) Schwarzschild, K., Astrophys. J., 11, 89 (1900).

(13) Mees, C.E.K., The Theory of the Photographic Process, MacMillan, New York, 1942.

(14) Mees, C.E.K. and James, T.H., The Theory of the Photographic Process, MacMillan, New York, 1966.

(15) Halm, J., Monthly No. R.A.S., 82, 473 (1922).

(16) Kron, E., Ann. Phys., 41, 751 (1913).

(17) Arens, H. and Eggert, J.Z., Phys. Chem., 131, 297 (1928).

(18) Martin, J.W., Chin, J.W., and Nguyen, T., Prog. Org. Coat., 47, 282 (2003).

(19) Chen, S.W., J. Phys. Chem., 53, 486 (1948).

(20) Stokes, S. and Fox, R.B., J. Polym. Sci., 56, 507 (1962).

(21) Davis, A. and Gordon, D., J. Appl. Polym. Sci., 18, 1159 (1974).

(22) Okada, Y., J. Appl. Polym. Sci., 7, 703 (1963).

(23) Schafer, V., Appl. Polym. Symp., 4, 111 (1967).

(24) Olson, R.L., McKellar, L.A., and Stewart, J.V., in Symposium on Thermal Radiation of Solids, Katzuff, S. (Ed.), National Aeronautics and Space Administration, 1965.

(25) Pickett, J.E., in Mechanisms of Polymer Degradation and Stability, Scott, G. (Ed.), 135-167, 1990.

(26) Pickett, J.E., ACS Polymer Preprints, 42, 424 (2001).

(27) Jorgensen, G., Bingham, C., King, D., Lewandowski, A., Netter, J., Terwilliger, K., and Adamsons, K., in Service Life Prediction: Methodology and Metrologies: ACS Symposium Series, Bauer, D.R. and Martin, J.W. (Eds.), 100-118, American Chemical Society, Washington D.C., 2002.

(28) Bauer, D.R., Gerlock, J.L., and Mielewski, D.R., Polym. Degrad. Stab., 36, 9 (1992).

(29) Bauer, D.R., Gerlock, J.L., Mielewski, D.R., Paputa Peck, M.C., and Carter, R.O., Polym. Degrad. Stab., 27, 271 (1990).

(30) Lemaire, J. and Siampiringue, N., in Service Life Prediction of Organic Coatings: A Systems Approach, Bauer, D.R. and Martin, J.W. (Eds.), 247-256, American Chemical Society, Washington D.C., 1999.

(31) Nguyen, T., Martin, J., Byrd, E., and Embree, N., "Relating Laboratory and Outdoor Exposure of Coatings: II. Effects of Relative Humidity on Photodegradation and the Apparent Quantum Yield of Acrylic-Melamine Coatings," JOURNAL OF COATINGS TECHNOLOGY 74, No. 932, 65 (2002).

(32) Chin, J.W., Byrd, E., Embree, N., Martin, J., and Tate, J.D., "Ultraviolet Chambers Based on Integrating Spheres for Use in Artificial Weathering," JOURNAL OF COATINGS TECHNOLOGY, 74, No. 929, 39 (2002).

(33) Martin, J.W., Prog. Org. Coat., 23, 49 (1993).

(34) Martin, J.W., Nguyen, T., Byrd, E., Dickens, B., and Embree, N., Polym. Degrad. Stab., 75, 193 (2002).

(35) Martin, J.W., Saunders, S.C., Floyd, F.L., and Wineberg, J.P., Methodologies for Predicting the Service Lives of Coating Systems, Federation of Societies for Coatings Technology, Blue Bell, 1996.

(36) Dickens, B. in Service Life Prediction Methodology and Metrologies, Martin, J.W. and Bauer, D.R. (Eds.), American Chemical Society, 2001.

Joannie Chin, Tinh Nguyen, Eric Byrd, and Jonathan Martin -- National Institute of Standards and Technology*

* Polymeric Materials Group, Building and Fire Research Laboratory, 100 Bureau Dr., Stop 8615, Gaithersburg, MD 20899.