Visual Characterization of Anti-Reflective Coating on Solar Module Glass

Abstract

Anti-reflection coatings (ARCs) are widely used on PV module glass to increase light transmission. The PV community is increasingly concerned with how long these coatings last in the field and would benefit from a simple method for quantifying performance on fielded modules. In this work, we demonstrate how a straightforward visual inspection of the color of reflected light can identify the presence of an interference-based ARC. By tracking the color shift over time a qualitative measurement of ARC degradation can be made. This method is applicable in full-sun outdoor conditions and only requires a flashlight and a standard RGB camera. We demonstrate how the physics of thin-film coating interference and color theory accurately predict the color the reflected light. This technique could gain widespread use for inspecting PV modules in the field because it is easy to perform and requires no specialized equipment.

Full text

Visual Characterization of Anti-Reflective Coating
on Solar Module Glass
Todd Karin
Lawrence Berkeley National Laboratory
Berkeley, CA, USA
toddkarin@lbl.gov
Anubhav Jain
Lawrence Berkeley National Laboratory
Berkeley, CA, USA
ajain@lbl.gov
Abstract—Anti-reflection coatings (ARCs) are widely used on
photovoltaic (PV) module glass to increase light transmission.
The PV community is increasingly concerned with how long these
coatings last in the field and would benefit from a simple method
for quantifying performance on fielded modules. In this work, we
demonstrate how a straightforward visual inspection of the color
of reflected light can identify the presence of an interference-
based ARC. By tracking the color shift over time a qualitative
measurement of ARC degradation can be made. This method
is applicable in full-sun outdoor conditions and only requires a
flashlight and a standard RGB camera. We demonstrate how
the physics of thin-film coating interference and color theory
accurately predict the color the reflected light. This technique
could gain widespread use for inspecting PV modules in the
field because it is easy to perform and requires no specialized
equipment.
I. INTRODUCTION
ANTI-REFLECTION coatings (ARCs) on the air-glass
interface of photovoltaic (PV) modules are widely used
to improve module performance. The most common variety of
ARC used on PV glass is a thin layer of porous silica deposited
by the sol-gel method with a thickness around ∼120 nm, and
can improve power output by up to 3% [1]. These coatings
reduce reflection both because they have an index of refraction
between that of glass and air and because the thickness is
tuned to provide destructive interference of reflected light at
the wavelengths most important for the solar absorber.
Recently, there has been increasing concern as to how
long ARCs last in the field [2]–[4]. Visual inspection is a
widely-used and powerful technique for assessing PV module
health [5]. Presently, it is not recognized that visual inspection
can also be used to assess the presence and quality of a
∼100 nm thick ARC. In fact, reflected light from interference-
based ARCs have a characteristic color shift that varies with
the angle of incidence (AOI) and is observable to the trained
eye. Similarly, an inexpensive RGB color camera can detect
these color shifts.
In this paper, we describe how to characterize ARCs on
outdoor modules using simple equipment: a low-power flash-
light, white card and either the human eye or a camera. By
Funding was primarily provided as part of the Durable Modules Consortium
(DuraMAT), an Energy Materials Network Consortium funded by the U.S.
Department of Energy, Office of Energy Efficiency & Renewable Energy,
Solar Energy Technologies Office. Lawrence Berkeley National Laboratory is
funded by the DOE under award DE-AC02-05CH11231.
Fig. 1. Dependence of spectral reflectivity on the physical thickness of a
porous silica ARC (effective index of refraction 1.23) on BK7 glass at 8
degrees angle of incidence. Also shown is a spectrum of the relative irradiance
from a white LED source (iPhone X LED). Units of white LED spectrum are
proportional to W/cm2/nm. Color of plot is a saturated version of the color
observed if a white LED were reflected from the sample.
comparing to broadband spectral reflectivity measurements,
we demonstrate that these simple methods can accurately
qualify interference-based ARCs without specialized equip-
ment. Further we provide a detailed description of the optical
physics that creates this effect and explore opportunities for
quantitative ARC characterization using similar methods.
II. INTERFERENCE-BA SE D ARCS
An interference-based ARC is a thin film that causes de-
structive interference of light reflected from the top and bottom
surface of the film. The film optical thickness is chosen to
be one-quarter of the wavelength where minimum reflection
is desired. At each location, the reflection properties can
be calculated using the complex-matrix form of the Fresnel
equation [6].
As an example, Fig. 1 shows the dependence of the re-
flection spectrum on coating thickness for a typical ARC
on a silicon solar cell. This simulation considers a layer of
porous silica with an effective index of refraction of 1.23 on

Fig. 2. Angle dependence for a porous silica ARC (effective index of
refraction 1.23) on BK7 glass with 120 nm thickness. Also shown is a
spectrum of the relative irradiance from a white LED source (iPhone X LED).
Units of white LED spectrum are proportional to W/cm2/nm. Color of plot is
a saturated version of the color observed if a white LED were reflected from
the sample. Color variation is similar if sunlight is used instead of a white
LED.
a thick slab of BK7 glass. The coating provides broadband
anti-reflection where the wavelength of minimum reflectivity
is related to the coating thickness. The ARC is most effective
for normally incident light.
The spectral shift at different coating thicknesses or angles
causes an observable color shift in the reflected light. Perceived
color is a complex subject, but can be calculated based on
the CIE color functions and translated into an RGB value.
In Fig. 1, we also show the relative irradiance from a white
LED. This spectrum was taken by calibrating a spectrometer
(Ocean Insight HDX) using a 2500 K blackbody spectrum
(Ocean Insight HDX) and finding the relative irradiance from
a white LED (iphone X). Fig. 1 displays each spectrum
with the color a human would observe from the reflection
spectrum. To do this, we calculate the CIE hsv color value
from the reflection spectrum and the white LED light source
[7], increase the saturation to 1 and value to 0.8 (in order
to show the more saturated color). The resulting plots show
that a significant color shift should be observable for different
coating thicknesses. The color shift is similar if 6000 K
blackbody radiation is used instead.
Similarly a coating color shift from blue to magenta to
yellow is observed as the angle-of-incidence increases from
0 (normal) to 60 degrees, see Fig. 2.
III. VIS UAL CHARACTERIZATION OF ARC
Several simple methods are available for detecting the
presence of an ARC by the color shift of white light reflected
from the glass surface. In these methods, shown schematically
in the inset of Fig. 3, a broadband white light source (flashlight
or sunlight) is reflected from the PV glass and the reflected
Light scattered
from paper
Light reflected
from glass
Flashlight
Observe
Light
θAOI
Fig. 3. Demonstration of simple identification of ARC on outdoor PV module.
A flashlight is reflected from the module glass, a significant blue shift indicates
that an ARC is present. Image is taken on a PV module with an ARC
(Panasonic n330) outdoors with overcast conditions. The flashlight spreads
light over a large range of angles, leading to diffuse reflection observed from
the paper and specular reflection observed from the glass, and s specular
reflection observed from the glass.
light is collected. The reflected light can be observed using an
RGB camera, a white card or directly by eye if a low-intensity
light source is used.
First, we demonstrate ARC identification by visually in-
specting the color of a flashlight as it reflected from the
glass. Fig. 3 shows that light reflected from the glass is
much more blue than the light scattered from a sheet of
white paper. This agrees with how a typical interference-based
ARC at low AOI reflects more blue light than green or red,
see Fig. 1, thus making the reflection appear blue. This is
a remarkable demonstration that a simple measurement can
detect the presence of a 100-nm thick coating.
The blue reflection for a full-thickness coating is also easily
visible by eye. When directly observing the reflected light
by eye, it is important to use a flashlight with the ability
to give a very low output intensity (∼1-10 lumens). If the
flashlight is too bright then the human observer can no longer
distinguish color and the measurement could pose a safety
hazard. Because the reflection from the glass is predominantly
specular, the measurement can be performed in full sunlight.
It is only important to align the flashlight and observer so
that the the specular reflection from the sun does not overlap
with the flashlight reflection. Incidentally, if a camera is used,
it is possible to use the sun as a light source instead of the
flashlight.
Next, we inspect in more detail the color of the reflected
light and how it depends on angle, see Fig. 4(a). Images were
taken using a consumer RGB camera (Canon 5DS R, Canon
100 mm f/2.8L lens) on a PV module with an interference-
based ARC (Panasonic n330) outdoors (overcast conditions)
and a white LED keychain flashlight (Nitecore TIP, lowest
intensity setting). The camera settings were set to a white
balance color temperature of 6000 K, aperture f/32, ISO 100

With ARC With ARC With ARC 1 cm
(a) 10◦(b) 45◦(c) 60◦
Fig. 4. Dependence of reflected LED (nitecore TIP) light on angle of incidence for a PV module with ARC (Panasonic n330). As the angle of incidence
increases, reflected light goes from blue to more violet to yellow. The grid lines are lightly visible.
and 1 second exposure. The small aperture setting was chosen
in order to increase the depth of field so the full reflected spot
is in focus even at higher angles-of-incidence.
The 0.1 mm scale roughness in Fig. 4(a) is visible as a non-
uniform surface glitter. A similar phenomenon of ocean glitter
has been studied extensively [8], finding that the spatial extent
and highlight distribution is related to the angular distribution
of surface normals and solid angle of the light source. These
glitter images could be analyzed in detail to determine the
local surface normal at each point in the image. Since the
distribution of surface normals affects the spectral reflectivity
and local color, a more-precise understanding of the surface
normal distribution can lead to a better prediction of how
averaged spectral reflectivity relates to coating performance.
By inspecting Fig. 4(a) more closely, we note that there
is some minimal color variability in the highlights. This is
explained by comparing with Fig. 2, where there is only
minimal perceived color variation as AOI ranges from 0 to
20 degrees. In contrast, Fig. 4(b) shows the reflected spot for
an image taken at 45◦AOI. In this image, some blue, violet,
magenta and orange colors are seen in the glitter highlights,
agreeing with the predicted colors for higher AOI.
At 60◦AOI, see Fig. 4(c), the overall reflected spot becomes
wholly more yellow, while individual highlights are mostly
violet and yellow, in agreement with the simulations. This
overall shift toward yellow of the reflected light at higher AOI
is also clearly visible by eye. We note that some of the color
variation could be due to non-uniform coating thickness (this
possibility will be explored in detail in the final manuscript).
In contrast, Fig. 5 shows reflected light from a module
without an ARC. The reflected light in these images does not
show any color compared with a white card, or a color shift
with AOI.
IV. COMPARISON WITH SPECTRAL MEASUREMENTS
A broadband spectral reflection measurement can confirm
the ARC properties observed using the simple methods. In
No ARC No ARC No ARC 1 cm
(a) 10◦(b) 45◦(c) 60◦
Fig. 5. Reflected light from a PV module without an ARC (Megsun
B07PV6HBQ8) is white, regardless of angle. Lightsource: Nitecore TIP.
this measurement, a fiber-coupled (400 µm multimode, 0.22
NA) tungsten-halogen light source (Ocean Insight HL-2000) is
directed onto a PV module at a 45 degree AOI (Thorlabs RPH-
SMA). Light is analyzed on a compact CCD spectrometer
(Ocean Insight HDX).
The reflection spectrum is calculated by comparing the
reflected light from the sample of interest to a back-blackened
BK7 sample (Filmetrics REF-BK7). The module with ARC
(Panasonic n330) shows a broadband dip centered around
600 nm, see Fig. 6. On the other hand, a module without
ARC (Megsun B07PV6HBQ8) shows a flat response with
a reflection similar to that of glass. The offset between the
glass reflection and the reflection from the module without
ARC is due to the roughness of the glass, which reduces the
light collected. These spectral measurements confirm which
modules have ARCs.
V. C ONCLUSION AND OUT LO OK
PV module ARCs are widely used for increasing perfor-
mance. We have described an exceedingly simple method for
measuring these ARCs in the field. As the module ages and
the ARC thickness is reduced or scratched away, we predict
the reflected color will shift in a quantifiable way. This method
could easily find widespread use in field inspection of solar
modules.

Fig. 6. Reflection spectrum from PV module glass with and without ARC.
The module with ARC shows a pronounced dip centered at 600 nm, while
the module without ARC has a flat reflectance. The solar weighted reflection
(SWR) is calculated using the standard AM1.5 spectrum, the uncertainty
quoted in the legend is the standard deviation of the mean SWR for
50 measurements at slightly different locations on the same module. This
averages the variable reflection due to the glass roughness.
REFERENCES
[1] B. M. Freiburger, C. S. Thompson, R. A. Fleming, D. Hutchings, and
S. C. Pop, “High efficiency anti-reflective coating for pv module glass,” in
2017 IEEE 44th Photovoltaic Specialist Conference (PVSC), June 2017,
pp. 1869–1872.
[2] N. Ferretti, K. Ilse, A. Sonmez, C. Hagendorf, and J. Berghold, “High
efficiency anti-reflective coating for pv module glass,” in 32nd European
Photovoltaic Solar Energy Conference and Exhibition, 2017, pp. 1697–
1700.
[3] K. Ilse, P.-T. Miclea, V. Naumann, and C. Hagendorf, “Cleaning
resistance of glass coatings,” Fraunhofer Center for Silicon-Photovoltaics
CSP, Tech. Rep., 2018. [Online]. Available: https://www.fsolar.de/
Test-report-V403 2018- Cleaning-resistance- of-glass- coatings.pdf
[4] D. C. Miller, A. Einhorn, C. L. Lanaghan, J. M. Newkirk, B. To,
D. Holsapple, J. Morse, P. F. Ndione, H. R. Moutinho, A. Alnuaimi, J. J.
John, L. J. Simpson, and C. Engtrakul, “The abrasion of photovoltaic
glass: A comparison of the effects of natural and artificial aging,” IEEE
Journal of Photovoltaics, vol. 10, no. 1, pp. 173–180, Jan 2020.
[5] W. M¨
uhleisen, C. Hirschl, G. Brantegger, L. Neumaier, M. Spielberger,
H. Sonnleitner, B. Kubicek, G. Ujvari, R. Ebner, M. Schwark,
G. Eder, Y. Voronko, K. Kn ¨
obl, and L. Stoicescu, “Scientific
and economic comparison of outdoor characterisation methods for
photovoltaic power plants,” Renewable Energy, vol. 134, pp. 321 – 329,
2019. [Online]. Available: http://www.sciencedirect.com/science/article/
pii/S0960148118313545
[6] S. J. Byrnes, “Multilayer optical calculations,” 2016.
[7] G. Taylor. python-colormath. [Online]. Available: https://
python-colormath.readthedocs.io/en/latest/
[8] C. Cox and W. Munk, “Measurement of the roughness of the
sea surface from photographs of the sun’s glitter,” J. Opt. Soc.
Am., vol. 44, no. 11, pp. 838–850, Nov 1954. [Online]. Available:
http://www.osapublishing.org/abstract.cfm?URI=josa-44-11-838