Distribution (SPD) curves were plotted as graphs

of irradiance versus wavelength. Measurements of

Sunlight, UV,
sunlight were taken at solar noon with the sensor kept
at normal incidence to the sun with a solar tracking
equatorial drive system. The sensor viewed the whole

& Accelerated
sky, so that complete global sunlight was measured.
Measurements of the accelerated light sources were
taken with the sensor in the same position as an

Weathering
ordinary test sample, so that the irradiance measured
would be the same as a sample receives.
All measurements were taken with the same
instrument to ensure that the various SPD curves
By Patrick Brennan & C. Fedor1
would be strictly comparable. Attempting to com-
The Q-Panel Company pare SPD curves generated by different spectro-
radiometers is a frequent source of error in this field,
26200 First Street, Cleveland, Ohio 441 45
due to different types of input optics, different wave-
length bandpasses, or simply due to limitations in the
INTRODUCTION state of the art in spectral irradiance calibrations of
Sunlight is an important cause of damage to diffuse UV sources like the sky or fluorescent lights.
plastics, textiles, paints, and other organic materials.
The instrument used was an International Light
Short wavelength ultraviolet light has long been
IL 700 Spectroradiometer, No. 504, consisting of the
recognized as being responsible for most of this
following components as shown below:
damage2.Spectroradiometric measurements were
made to quantify the wide variations in UV content of Input Optics: IL-2WE Double Wide Eye, quartz
sunlight under the following conditions: direct summer double lens cosine diffuser for wide viewing angle
sunlight, winter sunlight, sunlight filtered through
Monochromatec Kratos GM-200, double grating
ordinary window glass and sunlight filtered through
monochromatorwith calibrated 1.0 nanometer band
automotive glass.
pass, stray light less than 1 part per million.
Accelerated weathering testers are widely used
Detectoc PM 270C Photo-multiplier, with an S-5
for research and development, quality control, and
response, operated from an IL 760 power supply.
material certification. They employ a variety of
light sources to simulate sunlight and the damage Picoammetec IL 700A Radiometer.
caused by sunlight. Comparative spectroradiometric
Calibration: Microwatts per square centimeter per
measurements of various types of accelerated
nanometer, traceable to the National Bureau of Stan-
testers also showed a wide variety of UV spectra.
dards. Calibration Certificate: 404045901.
These measurements help highlight the advantages
and disadvantages of the various accelerated light
sources: enclosed carbon arc, sunshine carbon arc,
xenon arc and fluorescent UV. The measurements
suggest recommendations for the use of different
IPhotomultiplier pic~ammeter I
light sources for different applications. Two new types
of fluorescent UV lamps show promise for improved
correlation with natural exposures. Spectroradiometer
Cosine Schematic
Diffuser
EXPERIMENTAL
A spectroradiometer was used to measure the
spectral irradiance received by test samples exposed
to sunlight or accelerated testers. "Irradiance" is
the rate at which light energy falls on a unit area of
surface. "Spectral irradiance" is the distribution of
irradiance with respect to wavelength. In this case
the irradiance was measured at each 1 nanometer
(nm) wavelength band throughout the wavelength
region of interest. The Resulting Spectral Power

1. Presented at the Society of Plastics Engineers Automotive RETEC. Nov. 1987.

2. Searle. N. and His. R. "UV SPD of Sunlight,'J. Optical Soc. America, Vol. 55. NI
SUNLIGHT These measurements are in essential agreement
with data reported by other investigators5.
The electromagnetic energy from sunlight is
normally divided into ultraviolet light, visible light, and Because the sun is lower in the sky during the
infrared energy. Figure 2 shows the spectral power winter months, it is filtered through a greater air mass.
distribution (SPD) of noon midsummer sunlight, This creates two important differences between
measured in Cleveland Ohio, June, 1986. Infrared summer and winter sunlight: changes in the intensity
energy (not shown) consists of wavelengths longer of the light and in the spectrum. Most important, the
than the visible red wavelengths and starts above shorter, more damaging UV wavelengths are filtered
about 760 nanometers (nm). Visible light is defined as out during winter. For example, the intensity of UV at
radiation between 400 and 760 nm. Ultraviolet light 320 nm changes about 8 to 1 from summer to winter.
consists of radiation below 400 nm. This is especially significant for polymeric materials
such as PVC. In addition, the short wavelength solar
The International Commission on Illumination
cut-off shifts from about 295 nm in summer to about
(CIE) further subdivides the UV portion of the spec-
310 nm in winter. Consequently, materials sensitive
trum into UV-A, UV-B and UV-C as shown below3.
to UV below 310 nm would degrade only slightly, if at
all, during the winter months. The sunlight spectrum
at the March 21 equinox falls between the June and
December curves.

Wavelength (nanometers)

Figure 2 -The Sunlight Spectrum

The effects of the various UV wavelength regions

can be summarized as shown in the following table4.
260 280 300 320 340 360 380 400

WAVELENGTH REGIONS OF THE UV Wavelength (nanometers)

UV-A Causes polymer damage Figure 3 - Seasonal Variation of Sunlight UV

400 to 315 nm
Includes the shortest
wavelengths found at the
ACCELERATED LIGHT SOURCES
UV-B earth's surface; responsible COMPARED TO SUNLIGHT
315 to 280 nm for severe polymer damage; The following discussion of light sources will con-
absorbed by window glass. fine itself to the question of UV spectrum. It will not
Found only in outer space; address problems of light stability, the effects of mois-
uv-C
filtered out by earth's ture and humidity, the effects of cycles, or the repro-
280 to loo nm
atmosphere; germicidal ducibility of results.
For simulations of direct sunlight, artificial light
VARIABILITY OF SUNLIGHT sources should always be compared to what we will
Because UV is easily filtered by air mass, cloud call the Solar Maximum condition: global, noon sun-
cover, pollution, etc., the amount and spectrum of light, on the summer solstice, at normal incidence.
natural UV exposure is extremely variable. Figure 3 'The Solar Maximum is the most severe condition
shows a comparison of the UV regions of sunlight, met in outdoor service, and as such it controls
measured at Cleveland at noon on: which materials will fail. It is misleading to compare
The summer solstice (longest day of the year), light sources against so-called "average optimum
sunlight", which is simply an average of the much
The winter solstice (shortest day of the year),
less damaging March 21 and September 21 equinox
The spring equinox. readings. Graphs labeled "sunlight" in this paper refer
to the Solar Maximum - noon, global, midsummer
3. ClE Standard No. 20.
4. Grossman, D. " Know Your Enemy: The Weather." J. Vinyl Techn.. Vol. 3. No. 1.
5. Zerlaut, G. "Accelerated Weathering & UV Measurements", ASTM STP 781, pp
 our measurements show surprisingly little variation The UV output of the enclosed carbon arc primarily
in the Solar Maximum at different locations. Figure 4 consists of two very large spikes of energy, with a
shows noon summer solstice measurements at three very small amount of output below 350 nm. Figure
widely varied locations. 6 shows the same SPD comparison graphed on a
different vertical scale to include all of the output from
1
- Kilt Peak 6/86 the spikes. Since the shortest UV wavelengths are
Cleveland 6/86 the most damaging, the enclosed carbon arc gives
"O - Miami 6/87 very slow tests on most materials and poor correlation
.E on materials sensitive to short wavelength UV.
s
N

1
$
C
0.6 - Enclosed Carbon Arc
.-'0m
E 0.4 -
-
L

0.2 -
0 . O I . I l . I . I . I . I
260 280 300 320 340 360 380 400
Wavelength (nanometers)

Figure 4 - Solar Maximum, 3 Locations

0.0
260 280 300 320 340 360 380 400
IMPORTANCE OF SHORT
Wavelength (nanometers)
WAVELENGTH CUT-OFF
Photochemical degradation is caused by photons Figure 5 - Enclosed Carbon Arc and Sunlight
of light breaking chemical bonds. For each type of
chemical bond there is a critical threshold wavelength
of light with enough energy to cause a reaction.
Light of any wavelength shorter than the threshold
can break the bond, but longer wavelengths of light
cannot break it-regardless of their intensity (bright-
ness). Therefore, the short wavelength cut-off of a
light source is of critical importance. For example,
-
-
-
Enclosed Carbon Arc -
if a particular polymer is only sensitive to UV light -
below 295 nm (the solar cut-off point), it will never
experience photochemical deterioration outdoors. -
If the same polymer is exposed to a laboratory light
source that has a spectral cut-off of 280 nm, it will 1 ' 1 ' 1 ' 1 ' 1 ' 1 '
4

deteriorate. Although light sources that produce

shorter wavelengths produce faster tests, there's Wavelength (nanometers)
a possibility of anomalous results if a tester has a
wavelength cut-off too far below that of the material's Figure 6 - Enclosed Carbon Arc and Sunlight
end use environment.
Sunshine Carbon Arc (ASTM G-23 Open Flame
ARC TYPE LIGHT SOURCES Carbon Arc). The introduction of the sunshine carbon
arc in 1933 was an advance over the enclosed carbon
Enclosed CarbonArc (ASTM G-23). 'The enclosed arc. Figure 7 shows the UV SPD of summer sunlight
carbon arc has been used as a solar simulator in compared to the SPD of a sunshine carbon arc
accelerated weathering and lightfastness testers (with Corex D filters). While the match with sunlight
since 1918. Many ASTM and Federal Test Methods is superior to the enclosed carbon arc, there is still
still specify its use. When the light output of this a very large spike of energy, much greater than
apparatus is compared to sunlight, some deficiencies sunlight, at about 390 nm.
become evident. Figure 5 shows the UV spectral
power distribution (SPD) of summer sunlight (Solar
 A more serious problem with the spectrum filters. Figure 9 shows a xenon arc with quartaboro
of the sunshine carbon arc is found in the short filters compared to Solar Maximum. This filter combi-
wavelengths. To illustrate this, a change of scale nation allows severe, unrealistic short wavelength UV
is necessary to expand the low end of the graph. as low as 270 nm.
Figure 8 shows Solar Maximum compared to
The most common filter combination is borosilicate
sunshine carbon arc between 260 nm and 320 nm.
inner and outer filters (borolboro). Figure 10 shows
The carbon arc emits a great deal of energy in the
the SPD of summer sunlight compared to a xenon
UV-C portion of the spectrum, well below the normal
arc with borolboro filters. 'This is a better simulation of
solar cut-off point of 295 nm. Radiation of this type
sunlight UV than the quartzlboro filters, because the
is realistic for outer space, but is never found at the
cutoff wavelength of the borolboro is approximately
earth's surface. These short wavelengths can cause
280 nm, somewhat closer to the sunlight cutoff of
unrealistic degradation when compared to natural
295 nm.
exposures.
Effect of lrradiance Setting: Recent xenon arc
models have a light monitoring system to compensate
for the inevitable light output decay due to lamp ag-
4 - Sunshine Carbon Arc
E ing. The most common irradiance settings are .35 or
5 - .55 watts per meter2at 340 nm. Figure 11 shows how
these two settings (with borolboro) filters compare
with Solar Maximum. While .55 compares better with
$ -
5 summer sunlight, .35 is more like winter sunlight.
.--0 2-
However, for practical operational reasons, .35 is the
-E -
L
setting most often used.
1 -

260 280 300 320 340 360 380 400

Wavelength (nanometers)

Figure 7 - Sunshine Carbon Arc and Sunlight

Wavelength (nanometers)

Figure 9 -Xenon with Quart&/Boro Filters

260 280 300 320

Wavelength (nanometers)

Figure 8 - Sunshine Carbon Arc and Sunlight

Xenon Arc (ASTM G-26). The xenon arc was adapt-

ed for accelerated weathering in Germany in 1954.
Understanding xenon arc spectra is complicated by
two variables: the effect of filters and the effect of ir-
radiance settings.
Effect of Xenon Filters: Xenon arcs require a com-
bination of filters to reduce unwanted radiation. Auto- Wavelength (nanometers)
motive test methods that call for xenon arc exposure
Figure 10 -Xenon with BoroIBoro Filters
usually specify quartz inner and borosilicate outer
 260 280 300 320 340 360 380 400
270 290 310 330 350 370 390
Wavelength (nanometers)
Wavelength (nanometers)
Figure 11 - Effect of lrradiance Setting
Figure 12 - UVB-313 and FS-40

FLUORESCENT UV LAMPS UVA-340 Lamp (ASTM G-53). The UVA-340 was

Over the past 10 years, Fluorescent UV and introduced in 1987 to enhance correlation in the G-53
Condensation testers have come into wide use. devices. Most of this lamp's emission is in the UV-A
There are now different types of fluorescent lamps, region, with a small amount in the UV-B. The UVA-
with different spectrums, for different exposure appli- 340 has been tested on both plastics and coatings
cations. The fluorescent UV testers use a different and greatly improves the correlation possible with
approach than the arc testers. They do not attempt the Fluorescent UV and Condensation devices8.
to reproduce sunlight itself, just the damaging effects The Society of the Plastics Industry VSI and VWDl
of sunlight. This approach is effective because short are currently using this lamp in extended outdoor
wavelength UV causes almost all of the damage to correlation studies on rigid vinyl.
durable materials exposed outdoors. Consequently, Figure 13 shows the UVA-340 compared to Solar
fluorescent UV testers confine their primary emission Maximum. This new lamp is an excellent simulation of
to the UV portion of the spectrum. sunlight from about 370 nm, down to the solar cut-off
FS-40 Lamp (F40-UVB) (ASTM G-53). In the early of 295 nm.
1970's the FS-40 became the first fluorescent UV
lamp to achieve wide use. This lamp is currently spec-
ified in some automotive specifications, particularly
for coatings. Most of the FS4O's output is in the UV-B
portion of the UV spectrum, along with some UV-A.
This lamp has demonstrated good correlation to out-
door exposures for the gloss retention on coatings6
and for the material integrity of plastics. However the
short wavelength output below the solar cut-off can
occasionally cause anomalous results, especially for
color retention of plastics and textile materials7.
UVB-313 Lamp (ASTM G-53). Introduced in 1984,
the UVB-313 is essentially a second generation 260 280 300 320 340 360 380 400
FS-40. It has the same SPD as the FS-40, but with Wavelength (nanometers)
higher, more stable output. Figure 12 shows the SPD
of sunlight compared to the UVB-313 and the FS- Figure 13 - UVA-340 and Sunlight
40. Because of its higher output the UVB-313 gives
significantly greater acceleration over the FS-40 for
most materials. With the exception of the automotive
industry, the UVB-313 is the most widely used light
source for the ASTM G-53 devices.

6. Grossman. G. "Correlation of Weathering." J. Coatings Technology, Vol. 49. No 33. 1977

7. Dick, J., et al. "Weatherability of Pigmented Plastics." SAE Technical Paper. No 50350,1985
8. Fischer, R. "Accelerated Test w. Fluorescent UV-Condensation," SAE Tech. Pal r No. 841022.1984
FILTERING EFFECT OF GLASS
ON SUNLIGHT
Common Window Glass. Glass of any type acts as
a filter on the sunlight spectrum. The shorter, more
damaging wavelengths are the most greatly affected.
Figure 14 shows direct summer sunlight compared
to sunlight filtered through ordinary, single strength,
untinted, 0.125 inch thick window glass. As the figure
shows, ordinary glass is essentially transparent to
light above about 370 nm. However, the filtering effect
becomes more pronounced with decreasing wave-
length. The most damaging wavelengths below about 260 280 300 320 340 360 380 400
310 nm are completely filtered out. Wavelength (nanometers)
Automotive Glass. Automotive glass is thicker than
Figure 15 - Sunlight 'Through Windshield Glass
window glass. The thicker glass acts as a more ef-
ficient filter. In addition, auto glass windshields are
often tinted and usually contain a layer of plastic for
safety enhancement. Each of these factors adds to
the filtering efficiency. Figure 15 shows direct sum-
mer sunlight compared to sunlight filtered through
tinted automotive windshield glass. Almost all of the
most damaging ultraviolet light has been filtered out.
Figure 16 shows that various other types of auto win-
dow glass filter sunlight less than windshield glass
but more than ordinary window glass. Further data on
sunlight through automotive glass compared to labo-
ratory light sources has been reported el~ewhere.~.

1.2- 260 280 300 320 340 360 380 400

Wavelength (nanometers)
E 1.0-
E Figure 16 - Sunlight Through Auto Glasses
N
E 0.8-
5 Figure 16, Sunlight Through Various Types of Auto
$ 0.6- Direct Window Glass
C Sunlight
.-um A = 0.1 28 inch thick, Clear
2 0.4-
-L B = 0.228 inch thick, Clear
0.2- C = 0.159 inch thick, Lightly Tinted
D = 0.194 inch thick, Tinted

260 280 300 320 340 360 380 400

Wavelength (nanometers)

Figure 14 - Sunlight Through Window Glass

I
9. Brennan,"Lab Spectra and Sunlight Through Auto Glass", IFAl Auto Fibers & Fabrics Seminar, 1987.
ACCELERATED LIGHT SOURCES AND too severe. Light sources that eliminate wavelengths
SUNLIGHT THROUGH GLASS below the solar cut-off of 295 nm will give better,
more accurate results, but the price for increased
UVA-351 Fluorescent Lamp (ASTM G-53). Figure correlation is reduced acceleration. The user must
17 shows the UVA-351 compared to sunlight through educate himself to make this choice. In addition we
ordinary window glass. This fluorescent lamp is should point out that, despite the current interest in
an excellent simulation of sunlight for most interior

-
light energy, the spectrum of a test device is only one
applications. part of the picture. With any accelerated tester, there
1.2 , I are a number of parameters that can be programmed:
Sunlight through Window Glass UV spectrum, moisture, humidity, temperature and
test cycle. Furthermore, the parameters that one
chooses are, to a certain extent, arbitrary. No single
test cycle or device can reproduce all the variables
found outdoors in different climates, altitudes and
latitudes. Consequently, even the most elaborate
tester is really just a screening device. The real
usefulness of accelerated testers is that they can give
reliable, relative indications of which material performs
best under a specific set of conditions.

260 280 300 320 340 360 380 400

Wavelength (nanometers) ACKNOWLEDGMENTS

Figure 17 - UVA-351 & Sunlight Through Glass The authors are grateful for the cooperation of:
K l l T PEAK NATIONAL OBSERVATORY,
Xenon Arc (ASTM G-26). The current xenon arc
automotive test method for textiles and soft trim Kitt Peak, AZ
specifies the quartzlboro filter combination OHIO SPECTROGRAPHIC SERVICE,
(at .55 wlm2). Figure 18 shows this xenon method
compared to sunlight through ordinary glass. Parma, OH

1.2-

E 1.0-

E 0.8-

$ 0.6 - QuartzIBoro
c Xenon Arc

260 280 300 320 340 360 380 400

Wavelength (nanometers)

Figure 18 -Xenon & Sunlight Through Glass

CONCLUSIONS AND CAVEATS

Correlation between laboratory and natural
exposure test results will probably always be
controversial. As Fischer has showns,test speed and
accuracy tend toward opposition. Accelerated light
sources with short wavelength UV give fast tests, but
may not always be accurate. Usually where they are
wrong, they are wrong on the safe side-they are
 European Branch:
Q-Panel Lab Products Q-Panel Lab Products
26200 First Street Express Trading Estate
Cleveland, OH 44145 USA Farnworth, Bolton
Phone: (216) 835-8700 England BL49TP
Fax: (216) 835-8738 Phone: (01204) 861616 0 1994 Q-Panel Lab Products.
Fax: (01204) 861617 All Rights Resewed.