You are on page 1of 6

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/3520492

Weathering degradation of EVA encapsulant and the effect of its yellowing on


solar cell efficiency

Conference Paper · November 1991


DOI: 10.1109/PVSC.1991.169275 · Source: IEEE Xplore

CITATIONS READS

40 1,940

4 authors, including:

Keith A. Emery
National Renewable Energy Laboratory
310 PUBLICATIONS   27,975 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Daido Steel View project

All content following this page was uploaded by Keith A. Emery on 04 June 2014.

The user has requested enhancement of the downloaded file.


WEATHERING DEGRADATION OF EVA ENCAPSULANT AND THE EFFECT
OF ITS YELLOWING ON SOLAR CELL EFFICIENCY

FJ. Pern, A.W. Czanderna, K.A. Emery, and R.G. Dhere

National Renewable Energy Laboratory


(formerly the Solar Energy Research Institute)
Golden, Colorado

ABSTRACT The encapsulant materials provide optical coupling,


mechanical support, electrical isolation, physical isolatiod
After five or more years of weathering, the degradation of protection, and thermal conduction for the solar cell assembly
ethylene-vinyl acetate (EVA) encapsulant in photovoltaic (PV) (43). EVA copolymer (33% vinyl acetate) is extensively used
modules resulted in a yellow to dark brown color. Degraded for crystalline Si-based PV module encapsulation. EVA is
EVA shows a substantial increase in the gel content and a formulated and processed to provide the desired mechanical
large to complete loss of the ultraviolet (UV) absorber, strength and stability (43). The processed EVA contains a
Cyasorb UV 531. The EVA discoloration is caused by the 65% to 70% gel content (degree of cross-linking), 0.30 wt %
formation of polyconjugated (C=C), double bonds of various Cyasorb W 531 (a UV absorber identified as "Cyasorb" here),
lengths. Acetic acid and other volatile organic components are and two antioxidants(4-6). Upon weathering degradation, the
also produced from the photothermal decomposition of the EVA between the cover glass plate and solar cells in PV
EVA. The solar cell efficiency was reduced by -9% by a light modules may become discolored. In previous papers, we
yellow brown EVA and -50% by a dark brown EVA. Wea- studied the structuraleffects and relationship among the extent
thered PV modules with dark brown EVA also show a -50% of degradation,gel content, and Cyasorb concentration in EVA
decrease in efficiency. (6,7). In this paper, we emphasize more the effect of EVA
yellowing on the solar cell efficiency.

INTRODUCTION EXPERIMENTAL

In the past two decades, significant improvements in the EVA Materials and Analvtical Procedure
efficiency and reliability of new semiconductor materials for
PV devices and modules have resulted in PV becoming an The EVA materials we studied, our analytical proce-
increasingly viable energy alternative. While efficiency im- dures, and the analytical instruments we used are essentially
provements in large-area CuInSq, CdTe, and a-Si-based thin- the same as those reported previously (6,7). The laminated
film PV modules are increasingly promising, single-crystal and cured EVA in PV modules that are stored in the dark for
(and polycrystalline) Si-based PV modules have been the most six years remains clear. Clear and discolored (degraded) EVA
dominant products widely used. Typical efficiencies of these samples were cut from PV modules weathered outdoors for
PV modules range from about 11% to 14% (1). A long-term more than five years. The Cyasorb concentration (wt %) and
stability of over 20-30 years for all types of PV modules is gel content (%) are the two crucial measurements of EVA
one of the requirements to be cost-competitive. degradation (6).

However, some Si-based modules deployed outdoors for Measurement of the Effect of EVA Yellowing on Solar
five years or more develop a light yellow, to yellow-brown, to Cells Efficiencv
dark brown color, depending on their locations, use tempera-
ture, and configuration. A common factor in these discolored The effect of EVA discoloration on solar cell efficiency
modules is their use of EVA films as the encapsulant. The was evaluated as before (6), using a calibrated SpectrolabX-25
consequence of the EVA weathering degradation can be solar simulator. Quantum efficiency (spectral response) was
significant. For example, Gay and Berman reported that EVA measured on a computerized system with periodic (440 Hz)
browning resulted in a -30% loss in the annual energy output monochromatic light directed through one of an array of 10-
from the six-megawatt PV systems at the Carrisa, California, nm band-pass interferencefilters. The lower wavelength limit
power plant (2). A detailed study on the module performance was 290 nm. In both cases, the virgin or degraded EVA films
of the Canisa PV systems revealed that the EVA degradation were pressed above the reference solar cells with or without
is highly nonuniform from module to module, and that the adding a cover glass above the film. Similar measurements
aveme module power output is 35.9%below that of the initial were also performed for a solar cell specimen (26 cm2) taken
rating (3). The mismatch between neighboring modules, from a broken PV module. The I-V performance of three
caused by nonuniform EVA degradation, contributed an large PV modules was measured on a Spire Model 240A solar
additional power loss of 11.1%. simulator.

557

CH2953-8/91/0000-0557$1.00 0 1991 IEEE

Authorized licensed use limited to: National Renewable Energy Laboratory. Downloaded on February 10, 2010 at 15:17 from IEEE Xplore. Restrictions apply.
RESULTS AND DISCUSSION Virgin cured EVA Exposed EVA
& t + t
unexposed, clear + clear + light-yellow- yellow-brown
EVA Degradation and Discoloration (acetic acid+)
I I
- 100
Figure 1 shows the results obtained from analyzing a
large number of non-degraded and field-degraded EVA
samples. The gel content increased substantially from 65%-
70% in virgin EVA to 85%-88%in clear degraded EVA, to - 90
90%-92% in light yellow EVA, to 96%-97% in yellow-brown
and dark brown EVA, while the Cyasorb concentration de-
creased concomitantly during the degradation process. No - 80
Cyasorb was observed in the extensively degraded EVA.
I \
When the Cyasorb concentration dropped below about 0.21 wt , \

%, the EVA films became discolored (6). As the extent of ,I Clear Yellow
,
degradation increased, the color darkened from light yellow, to K/ -I \
\ - 70
yellow, to yellow-brown, and to dark brown. The EVA
degradation across each solar cell unit on weathered PV mod-
ules was not uniform. The Cyasorb concentration was much
lower and the gel content was higher in EVA from over the
central region than from over the cell edges (6,7). No notice-
able change in either the gel content or the Cyasorb concentra- Figure 1. The Cyasorb UV 531 concentration, gel content,
tion was observed in EVA samples taken from unweathered and EVA color at different extents of degradation, summar-
modules from similar locations. The EVA mostly remained ized from the results of analyzing a large number of unde-
clear around the cell edges and in the areas between neighbor- graded and field-degraded EVA samples. The actual degra-
ing solar cells. In addition to the discoloration,acetic acid and dation is not known except for the virgin EVA.
volatile organics were present in the degraded EVA.

Figure 2 compares the transmission spectra for virgin and


degraded (to various degrees) EVA samples. As the color on
the EVA darkens, more UV and visible light is absorbed. The
yellow-browning of degraded EVA is caused by the formation
of polyconjugated carbon-carbon double bonds (polyenes)
(6,7). The discoloredEVA luminesces strongly upon illumina-
tion by UV and visible light. In fact, the extent of EVA de-
gradation can be detected directly using fluorescence analysis
from the shift in the emission peak position (from shorter to
longer wavelengths in the 500-650 nm region) and the increase
in peak intensity (6,7). Intense acetic acid odors and other
volatile organic components were detected when the weathered
PV modules were cut open. The presence of acetic acid is sig-
nificant. In various simulated degradation experiments, acetic
acid was produced from EVA exposed to UV light at 45°C or
heated in the dark at 130°C. Acetic acid has been found to
catalyze EVA yellow-browning from 85" to 130°C. A detailed
discussion will be given elsewhere (7). Wavelength (nm)

In summary, weathering degradation of the EVA encap- Figure 2. Transmission spectra measured for (1) a clear
sulant in PV modules results in an increased gel content, a loss EVA film, (2) a light yellow-brown EVA film,(3) a darker
of the UV absorber, a yellow-browning of the EVA, and the yellow-brown EVA film, (4) a brown EVA film, and (4) a
production of acetic acid and volatile organic components (6). darker brown EVA film. Samples #1, #4, and #5 were lami-
nated in two glass plates and exposed to an RS4 sun lamp at
Effects of EVA Discoloration on Solar Cell Efficiency 90°C for 1600 h. Samples #2 and #3 were taken from wea-
thered PV modules.
The effect of yellow-browning on the light transmission
is shown in Figure 2 for samples that are clear to dark brown the measured open-circuit voltage (V,), short-circuit current
EVA. The effect of EVA yellow-browning on the electrical density (J%), maximum power (PmJ, and calculated loss (%)
performance and quantum efficiency (spectral response) was for two virgin clear and two degraded yellow-brown EVA
obtained using single-crystal Si reference cells with and films. When covered with a virgin clear EVA film, the Si cell
without a degraded EVA sample laid over them. The results lost about 6.7% in J, and 0.8% in P,. However, the loss
are given in Tables 1 and 2(A), respectively. Table 1 gives increased to 12%-14% in J, and 14%-16% in P, with the

558

Authorized licensed use limited to: National Renewable Energy Laboratory. Downloaded on February 10, 2010 at 15:17 from IEEE Xplore. Restrictions apply.
yellow-brown EVA films. The loss in V, was relatively are 67%-77% lower than the typical 13%-14% efficiency of
small. In effect, the Si cell's 11.6% efficiency changed to new 4"x4" mini-modules made by the same manufacturer. The
10.7% with the virgin EVA and 9.7% with the yellow-brown decrease is 17%-27% greater than the changes measured with
EVA; this corresponds to a net loss of 9.3% of its original dark brown EVA films alone (see Table 2A), suggesting that
efficiency when the EVA color changed from clear to yellow- other factors such as increased series resistance may have also
brown. These effects also appear clearly in the relative quan- contributed to the efficiency drop. This argument seems to be
tum efficiency ratios to the bare Si reference cell, as shown in supported by a f i i factor of about 0.50 as compared to a fill
Figure 3. The spectral responses for the two virgin cured factor of about 0.70 for new mini-modules. The relative spec-
EVA films are virtually identical, with a cutoff below 360 nm tral responses measured for the variously degraded EVA sam-
from the absorption of Cyasorb. For the two degraded yellow- ples are compared in Figure 4.
brown EVA films that differ slightly in color, the spectral
responses begin at about 290 nm because of the absence of Table 1. Effect of EVA Yellowing on the I-V Performance
Cyasorb. Because of light absorption by polyenes, the spectral of a Single-Crystal Si Reference Solar Cell'
response is lower between 360 and about 900 nm than that of
the virgin EVA (Figure 2, curves 3 and 4). Sample V, Jffi' PIU, Loss (%) in
EVA (VI (mw) vcc Jffi pm,
As the EVA color became dark brown, the change in
measured cell efficiency increased. Table 2(A) shows the re- No EVA 0.474 39.87 4.06 -_- --- --_
sults obtained for two dark brown EVA samples with or with-
out a 7059 glass plate pressed tightly onto the EVA over the A9918 0.470 37.21 3.74 0.84 6.67 0.79
reference Si cell. For comparison, a solar cell specimen taken
from a broken PV module with dark brown EVA was also 15295P 0.471 37.28 3.75 0.63 6.50 o.m
similarly analyzed. With the dark brown EVA, a nearly 50%
Yellow-1 0.466 35.00 3.48 1.69 12.21 1428
loss in PmUwas observed; this is -35% greater than that for
the yellow-brown EVA seen in Table 1. The large difference Yellow-2 0.465 34.51 3.41 1.90 13.44 16D1
is simply because the dark brown EVA absorbs more UV and
visible light (Figure 2) than does the yellow-brown EVA (6,7).
The presence of a cover glass plate reduced the P,, an * An aperture of 0.35 cm2was used over a single-crystal Si reference
additional 2%. For the solar cell specimen, the measured solar cell for the measurements. The EVA film was placed be-
efficiency for the three tape-defied areas ranged from 3.0% neath the aperture. No cover @ass was used on top of the cell or
to 4.4% (see Table 2B), further supporting our earlier conclu- EVA. The loss (%) would be slightly larger if a cover glass were
used. The P,, loss (%) would be about 7% less than the values
sion that the degradation of EVA is non-uniform across the
shown if they were corrected for the light scattering (see text).
cell surface (6). More importantly, the measured efficiencies Data for Yellow-1 and Yellow-2 are identified as samples #2 and
#3 in Figure 3 and Yellow-2 as sample #3 in Figure 4.
-1.oo 1

I
E
f

0.00U
250 450 650 850 1050 1250
Wavelength (nm)
250 450 650 850 1050 1250
Wavelength (nm)
Figure 3. Relative quantum efficiency measured for a
single-crystal Si reference solar cell covered with virgin Figure 4. Quantum efficiency measured for (1) a single-
clear (curves #1 and #2) and degraded yellow-brown (curves crystal Si reference solar cell, (2) the cell in (1) covered
#3 and #4) EVA films. The spectral responses are ratioed to with virgin cured EVA A9918, (3) the cell in (1) covered
that for the bare Si reference cell. The virgin EVA films are with yellow-brown EVA, and (4) a solar cell specimen with
EVA A9918 (curve #1) and EVA 15295P (curve #2). The dark brown EVA. Curves #1 and #4 are normalized to
two degraded EVA films differ slightly in color. 100%. Curves #2 and #3 are normalized to curve #1.

559

Authorized licensed use limited to: National Renewable Energy Laboratory. Downloaded on February 10, 2010 at 15:17 from IEEE Xplore. Restrictions apply.
Table 2. Effect of Dark Brown EVA on the I-V Perform- EVA changed from clear to yellow-brown and -42% when it
ance of (A) a Single-Crystal Si Reference Solar Cell' and changed to dark brown. While the integrated reflectance
(B) a Solar Cell Specimen measured for the solar cell specimen is about 35% lower than
that for a dark brown EVA with a smoother Si-side surface
(A) Two dark brown EVA films peeled from degraded PV modules (Figure 5b), this small difference is not sufficient to adversely
affect the interpretation of results in Table 2.
Sample V, Jw Pinax Loss (%) in
EVA (V) (mA/cm2) (mW) V, J, P, Effect of EVA Discoloration on PV Module Performance
Si Cell 0.398 22.54 3.54 ---- ---- ----
The effect of EVA discoloration on module I-V perfor-
Brown-1 0.347 13.15 1.80 12.81 41.66 49.15 mance was also evaluated for three PV modules of the same
+ slideb 0.341 12.64 1.70 14.32 43.92 5198 type (33 4"x4" cells). The EVA color ranged from light
yellow to dark brown. The shunt and series resistances were
Brown-2 0.345 13.21 1.79 13.32 41.39 49.44 determined from dark I-V measurements. Table 3 summarizes
+ slideb 0.343 12.76 1.72 13.82 43.39 51.41 the results. The original data on the I-V performance and
resistances are unavailable, which prevents an accurate inter-
pretation of the data. However, the results seem to suggest
(B) A solar cell specimen with dark brown EVA'
that an increase in the EVA yellow-browning is related to a
Efficiency decrease in the shunt resistance and an increase in the series
Spot Area V, J, P, Fill
No. (cm') (V) (mA/cm2) (mW) Factor (%)
resistance, which in turn results in a corresponding decrease in
the fill factor and hence the efficiency. The losses in fill fac-
1 0.170 0.330 26.62 0.746 0.50 4.4 tor and efficiency in the yellow-brown and dark brown EVA
modules are about 50% when compared to a typical fill-factor
2 0.190 0.325 21.62 0.665 0.50 3.5 value of about 0.70 and an efficiency of 210% for a new
module of the Same configuration. Our results on three mod-
3 0.163 0.308 20.16 0.489 0.48 3.0 ules are in good agreement with the field results obtained at
the Carrisa Plains power plant, where EVA yellow-browning
caused an average 36% loss in power output (2) and where
' The results shown in the table are not adjusted for the loss of light some modules with mirror-enhanced configuration lost nearly
transmission through the EVA film due to light scattering from the 70% more than average peak power (3).
come Si-side surface. See text for details. Data for Brown-1 are
identified as sample #3 in Figure 5(a).
A glass plate was pressed tightly on top of the EVA film. An ac-
tive aperture area of 0.967 cmz was used for the measurements.
The solar cell specimen (26 cm2) was taken from a broken PV
module with dark brown EVA and the superstrates and substrates
intact. The three areas were defmed by black tape. Contacts were
made by soldering wires to the front and back buslines.
0.20

t (a)

One concem in our measurements of efficiency and spec-


tral response is the accuracy offset by light scattering from the
textured surface of EVA films, and the extent of noncoupling
between the EVA and the reference Si cell. This concem is
particularly important for the dark brown EVA films (used for
measurements in Table 2A) that have a coarse surface side
after being forcefully removed from the micro-pyramidically
etched surface of the Si solar cells. The light-scattering effect
was assessed from the integrated reflectance measurements
from 250 to 1250 nm for a virgin clear EVA sample, a yellow-
brown EVA sample, two dark brown EVA samples (one of
them with a coarse Si-side surface as described above and the 0.00 I I I I I I I I I
250 450 650 850 1050 1250
other with a smoother Si-side surface), and the solar cell
specimen. The results are shown in Figure 5. Figure 5(a) Wavelength (nm)
shows that the integrated reflectances are nearly identical for
the virgin clear and yellow-brown EVA films (about 7%), and Figure 5. Integrated reflectance measured for (a) a virgin
double that for the dark brown EVA with a coarse Si-side EVA A9918 film (curve #1), a yellow-brown EVA film
surface. Therefore, the measured solar cell performance (curve #2), a dark brown EVA film (curve #3) with a coarse
parameters in Tables 1 and 2(A) should be corrected by 7% Si-side surface, and (b) a dark brown EVA with a smoother
and 14%-15% of their listed values, respectively. Accordingly, Si-side surface (curve #1) and a solar cell specimen
the corrected decrease in solar cell efficiency is -8.6% when (curve #2).

Authorized licensed use limited to: National Renewable Energy Laboratory. Downloaded on February 10, 2010 at 15:17 from IEEE Xplore. Restrictions apply.
Table 3. Electrical Performance Measured using a Spire to about 50%, depending on the extent of EVA discoloration
Simulatorand ResistancesDetermined from Dark I-V Data from the degradation.
for Three Weathered PV Modules'

Mod. EVA V, I, ,P Fill Eff. %h, %, ACKNOWLEDGMENT


ColoP (V) (A) (W) Factor (%) (kohm) (Ohm)
The authors thank R. DeBlasio for his support in this
A It ye1 6.755 7.217 34.0 0.70 9.34 1.35 0.109 work and S. Rummel and E. Beck for their technical contribut-
B yel-bm 6.733 6.428 16.35 0.38 4.49 0.577 0.508
ions. This work is supported by the Department of Energy
under Contract No. DE-AC02-83CH10093.
C drk bm 7.047 6.137 15.99 0.37 4.41 0.723 0.499
REFERENCES
' The three modules were of the same type, and each contains 33 4"x4"
square solar cells.
1. J. P. Thomton, R. DeBlasio, and K. Zweibel, Energv En-
It ye1 = light yellow, yel-bm = yellow-brown, drk bm = dark brown. pineerin& 87, 1990, pp. 63-79.
2. C. F. Gay and E. Berman, Chemtech, March 1990, pp.
182-186.
CONCLUSIONS 3. A. L. Rosenthal and C. G. Lane, Proc. PV Module Relia-
bilitv WorkshoD, Oct. 25-26, 1990, Lakewood, Colorado,
We have reported that weathering degradation of the SERI/CP-4079, pp. 217-229.
EVA encapsulant in PV modules results in an increase of 4. C.G. Gebelein, D.J. Williams, and R.D. Deanin, ed., polv-
copolymer cross-linking, a large decrease in W absorber con- mers in Solar Energy Utilization, ACS, Washington D.C.,
centration, the production of acetic acid, and a yellow to dark 1983, Ch. 22 and 23, pp. 353-385.
brown color. The discoloration results from the formation of 5. E. Cuddihy, C. Coulbert, A. Gupta and R. Liang, Flat-
polyenes of various lengths. The polyenes absorb UV and vis- Plate Solar Array Proiect Final ReDort. Vol. VII--Module
ible light and luminesce in the visible region; thus, they may Encapsulation, JPL publication 86-31, DOWJPL-1012-125,
partially compensate for the loss of solar cell efficiency due to 1986.
reduced light transmission. The light-absorbing nature of 6. F.J. Pem and A.W. Czandema, Solar Cells, in press.
polyenes reduces the net solar cell efficiency from about 9% 7. F.J. Pem and A.W. Czandema, to be published.

561

View publication stats


Authorized licensed use limited to: National Renewable Energy Laboratory. Downloaded on February 10, 2010 at 15:17 from IEEE Xplore. Restrictions apply.

You might also like