The Impact of Cracked Solar Cells on Solar Panel

Energy Delivery
Andrew M. Gabor Eric J. Schneller Hubert Seigneur Michael W. Rowell
BrightSpot Automation LLC Florida Solar Energy Center, Florida Solar Energy Center, D2Solar
Westford, MA, USA University of Central Florida University of Central Florida San Jose, CA, USA
[email protected] Cocoa FL, USA (now at Silfab, Cocoa FL, USA [email protected]
Bellingham, WA) [email protected]
[email protected]

Dylan Colvin Michael Hopwood Kristopher O. Davis

Florida Solar Energy Center, Florida Solar Energy Center, Florida Solar Energy Center,
University of Central Florida University of Central Florida University of Central Florida
Cocoa FL, USA Cocoa FL, USA Cocoa FL, USA
[email protected] [email protected] [email protected]

Abstract — Solar panel degradation is usually assessed by the these Pmax values, and the passing of module certification tests
change in power at standard testing conditions (STC). However, and warranty violations are based on this Pmax degradation.
some degradation mechanisms have shunting or recombination In contrast, the economics of system revenues depend on
characteristics which have the potential to reduce performance at
low irradiances significantly more than at 1-Sun conditions. We energy delivery over the course of years. Much of the time that
present data at both the single cell coupon level and at the module a system is feeding electricity into the grid occurs at irradiances
level that demonstrate this effect with cracked cells, where the significantly lower than 1-Sun, and as PV penetration grows,
effect scales with the total length of the cracks. The effect is the value of the electrons generated at the lower irradiances
present even for modules with tightly closed cells where the grows as well. Recognizing this, solar panel specification sheets
metallization is continuous across the cracks and no dark areas are and PAN files show the performance of new panels at lower
seen in the electroluminescence (EL) images. Depending on the
system geographic location, mounting angles, the time of year, and
irradiances. The problem we are addressing in this paper is that
the clipping characteristics, the daily energy delivery of a system degraded panels may degrade quite differently at low
can depend quite strongly on the module performance at low irradiances than at 1-Sun, and that this effect is largely ignored
irradiances. We show through simulations that energy delivery in the literature and can have large economic impacts.
may degrade significantly more than Pmax with damage from Some degradation modes have shunt-like characteristics or
cracked cells. Since electricity generated at lower irradiances low diode quality characteristics that change the slope of the I-
often has more value than electricity generated at high V curve in the direction of ISC toward Pmax. As is seen in Fig. 1,
irradiances, the impact on system revenue may be even larger. We
conclude that the degradation from accelerated testing and field a particular level of shunting at a low irradiance can cause a
exposure should be assessed not just at an irradiance of 1-Sun but much higher relative percentage degradation in Pmax than if that
also at lower values as well. same level of shunting occurs at a high irradiance.
Keyword — Electroluminescence, Photovoltaic cells, Power
system stability, Solar Panels, Stress, Energy Delivery

I. INTRODUCTION 16% degradation

The PV industry is highly sensitive to the performance of

solar panels at Standard Testing Conditions (STC). These
conditions correspond to the performance at a temperature of
25°C and an irradiance of 1-Sun (1000 W/m2), and the term Pmax
generally refers to the maximum power point at STC. Any
particular measurement will occur at different temperatures and
irradiances, but well-developed equations have been developed 5% degradation
to correct each point on the I-V curve back exactly to STC
conditions [1]. Indoor measurements are generally quite close
to the STC conditions, while outdoor measurements are often
performed at quite different conditions with larger corrections Fig. 1. I-V curve simulations of a cell at 1-Sun and 0.3-Sun
needed. The selling prices of solar panels are determined by irradiances for a cell with no shunting and a cell with shunting. The
shunting causes a higher relative degradation in Pmax at low irradiance.

978-1-7281-6115-0/20/$31.00 ©2020 IEEE 0810

Authorized licensed use limited to: Escuela Superior de Ingeneria Mecanica. Downloaded on January 31,2023 at 01:22:16 UTC from IEEE Xplore. Restrictions apply.
 Fig. 3. EL images of 3 coupons showing Good (left), Cracked
(middle) and Badly Cracked (right) cells.

Table I. Cell coupon efficiency vs irradiance and cracking level

Fig. 2. Module efficiency vs irradiance for undegraded and PID

affected modules, taken from [2].
Table I shows the cell efficiency as measured on the I-V
One degradation mode that has shunt-like characteristics is tester at irradiances of 1-Sun, 0.4-Suns, and 0.2-Suns. The
Potential Induced Degradation (PID). Schneller et. al. [2] higher the level of cell cracking, the worse the 1-Sun
showed the irradiance-dependent behavior of PID affected performance, but notably, the performance at lower irradiances
modules as is shown in Fig. 2. At 1-Sun the PID affected drops off much more steeply for the cracked cells than is the
module has degraded by ~14.5% relative, but at 0.2-Sun it has case for the Good cell. At 0.2-Sun, the Good cell is only down
degraded by ~32% relative. Projecting future energy delivery ~ 4% relative, while the Badly Cracked cell is down ~30% from
of a PID-affected system based on 1-Sun values could an already low level at 1-Sun. The degradation at low
significantly overestimate future revenues. The plot of module irradiances appears to be strongly correlated to the total length
efficiency vs irradiance is particularly useful for studying these of cracks in the cell.
irradiance dependent effects, and such data is generated
automatically using the FMT-500 flash I-V tester from Sinton
Instruments. III. DEGRADATION OF MODULES
In this paper we use EL imaging and Sinton Instruments I-V In order to study the effect of cracking vs irradiance on
testing to study such irradiance dependent degradation for modules, we took EL and I-V measurements both before and
different types solar cell cracking. after loading on the LoadSpot mechanical load tester from
BrightSpot Automation, as we have described previously [3-6].
II. DEGRADATION OF SINGLE CELL COUPONS Fig. 4 shows EL images captured at the -ISC point for a 4-busbar
multicrystalline module. The image prior to loading shows no
Prior to studying entire modules, we first look at the effects cracks, while the image after frontside loading to 5400 Pa
of cracking on individual cells. We soldered interconnect wires shows extensive cell cracking. After applying 200 standard
on the front and back sides of 156mm monocrystalline solar cycles of +/-1000 Pa, several of the cracks have opened up,
cells and then encapsulated them within the structure 3.2mm- leading to darker isolated regions.
glass/EVA/cell/EVA/polymer backsheet to form individual cell
coupons. We then applied a front side load to several of the
coupons as we describe elsewhere [3] until the cells fractured.
Fig. 3 shows the EL images of 3 of the coupons biased near the
-ISC point. One cell is uncracked (Good), another shows a
moderate to high level of cracking (Cracked), and the last
shows very high level of cracking (Badly Cracked). Note that
most of these cracks are of the “closed” variety determined by
the metallization being continuous across the cracks. The
degradation is largely linked to recombination and shunting
along the cracks rather than due to lost active area.

Fig. 4. EL images of a module before loading (left), after loading to

5400 Pa (middle), and after 200 cycles of +/-1000 Pa (right).

978-1-7281-6115-0/20/$31.00 ©2020 IEEE 0811

Authorized licensed use limited to: Escuela Superior de Ingeneria Mecanica. Downloaded on January 31,2023 at 01:22:16 UTC from IEEE Xplore. Restrictions apply.
 and the remaining inputs were left default. A module that
exhibits a strong falloff in efficiency at low irradiances will
perform worse in less sunny locations, and thus we chose to
explore both a sunny location (Phoenix, AZ) as well as a
northern cloudy location (Seattle, WA).
Efficiencies below 0.2-Sun are expected to be dramatically
lower than at 0.2-Sun or above for modules with substantial
cracks, as shown by the trends in the efficiency vs irradiance
curves in Figs. 2 and 5. The nonlinear trend of the efficiency
suggests an even more severe efficiency reduction at below 0.2-
Sun. However, in our irradiance-dependent analysis, we pulled
discrete points off the graph in Fig. 5 (1.0, 0.8, 0.6, 0.4, 0.2
Suns). For irradiances < 0.2 Sun, the model used the 0.2 Sun
Fig. 5. Efficiency vs irradiance for the module states shown in Fig. data point, and thus overestimated energy delivery at the lowest
4. irradiances.
For each location, the analysis was performed at 1) a constant
Fig. 5 shows the module efficiency vs irradiance for the 3 efficiency (1-Sun peak) and 2) using the intensity-dependent
cases. As with the single cell coupons, after loading and crack efficiency as discussed above. For the static loaded module,
formation, the module has a much stronger falloff in efficiency Fig. 6 shows that there is a 1.4% error between the 1-Sun and
as irradiance is decreased than it did before loading. The intensity-dependent performance for the sunny condition. In
undamaged module falls off 3.9% at 0.2-Suns, while the loaded contrast, there is a more severe 2.6% error between the 1-Sun
module falls off 9.2%. Interestingly, after cyclic loading, the 1- and intensity-dependent performance for the cloudy location.
Sun efficiency drops still further, but the falloff is less strong at For the baseline condition (before loading), these percent errors
lower irradiances such that the module actually has a higher are much smaller at 0.4% and 0.9% for the sunny and cloudy
efficiency below 0.2-Suns than before cycling, with a falloff of locations, respectively.
only 5.6% at 0.2-Suns. This might be explained in that cyclic
loading opened up some cracks such that some regions of the
cells that had internal cracks were effectively removed from the
circuit. With a lower total length of cracks remaining in active
areas of the cells, the shunting/recombination is lower, leading
to a less sharp falloff with decreasing irradiance, although the
reduced active area reduces the efficiency at higher irradiances.

IV. ENERGY DELIVERY

The annual energy delivery of a PV system depends on many
factors, including the matrix of module power vs irradiance vs
temperature, the geographic location and weather, the mounting
angles, the local albedo, shading from neighboring modules or
objects, soiling, inverter type, whether the system is tracking or
static, whether storage is integrated, and whether the system is
clipped to a certain maximum system power. The system Fig. 6. Using NREL’s SAM [7], the energy yield for a module
before and after cracking is simulated 1) at constant efficiency (1-Sun
revenue is very strongly linked to the energy delivery, but with
peak) and 2) using irradiance-dependent efficiency. The simulations
high PV penetration on the grid, the value of the energy were run using meteorological data from Phoenix, AZ (sunny) and
produced during low irradiance times of the day may be higher Seattle, WA (cloudy).
than during the high irradiance periods. Overestimating the
performance at low irradiances by focusing on high irradiance
data could result in significant revenue shortfalls. VI. CONCLUSIONS
To explore the impact of cell cracking on energy delivery, we We have demonstrated how predicting a degraded PV system
took the "Initial" and "After static load" efficiency vs irradiance performance based on 1-Sun measurements of power can
curves from Fig. 5, and applied parametric and sensitivity overestimate system revenue due to the deterioration in module
analysis using NREL's System Advisor Model (SAM) [7]. The efficiency at lower irradiances in cases where the degradation
analysis was performed using module parameters from the has shunting/recombination characteristics. Specifically, we
specification sheet (namely, VMP, VOC, Pmax temperature have shown how a module with many cracked cells has a worse
correction coefficient, module area), assuming open rack performance at low irradiances, and through simulations we
installation for glass/cell/polymer sheet module configuration, have shown how the impact on system energy delivery is more

978-1-7281-6115-0/20/$31.00 ©2020 IEEE 0812

Authorized licensed use limited to: Escuela Superior de Ingeneria Mecanica. Downloaded on January 31,2023 at 01:22:16 UTC from IEEE Xplore. Restrictions apply.
significant in a less sunny northern location than a sunnier Renewable Energy (EERE) under Solar Energy Technologies
southern location. Cracks in solar cells are often tightly closed Office (SETO) Agreement Number DE-EE0008152.
with little degradation in STC measurements and with no dark
areas in EL images. This may give a false sense of security, as REFERENCES
the low-irradiance performance may have degraded [1] IEC 60891 Edition 2.0, 2009-12, “Photovoltaic devices – Procedures for
significantly more. temperature and irradiance corrections to measured I-V characteristics”
While PAN files containing a matrix of module power vs [2] E. J. Schneller et. al., ““Evaluating Module Performance and Reliability
Utilizing Multi-Irradiance I-V Measurements, ”NREL PV Reliability
irradiance and temperature are commonly used in energy
Workshop 2017.
delivery simulations, these PAN files are always generated [3] M.W. Rowell, S.G. Daroczi, D.W.J. Harwood, A.M. Gabor, “The Effect
using new module data. Degraded modules are usually of Laminate Construction and Temperature Cycling on the Fracture
measured only at STC for insight on their performance. For Strength and Performance of Encapsulated Solar Cells” 45th IEEE PVSC,
greater accuracy in energy delivery predictions, we encourage pp 3927-3931, 2018
the industry to start measuring the power of aged modules over [4] A. M. Gabor, R. Janoch, A. Anselmo, J. L. Lincoln, H. Seigneur, and C.
a range of irradiances and to generate PAN files from modules Honeker, "Mechanical load testing of solar panels - Beyond certification
testing," in IEEE 43rd Photovoltaic Specialists Conference (PVSC),
which have undergone a variety of typical degradation 2016, pp. 3574-3579.
mechanisms. A good place to start would be to perform these [5] E. J. Schneller, A. M. Gabor, J. L. Lincoln, R. Janoch, A. Anselmo, J.
tests at certification and testing labs after the modules have Walters, and H. Seigneur, “Evaluating Solar Cell Fracture as a Function
undergone the various IEC 61215 test legs. By using such data of Module Mechanical Loading Conditions” in IEEE 44th Photovoltaic
to simulate degradation in energy delivery at a few standardized Specialists Conference (PVSC), 2017.
worldwide installation locations as figures of merit, module [6] E. J. Schneller, H. Seigneur, J. Lincoln, A. M. Gabor, “The Impact of Cold
Temperature Exposure in Mechanical Durability Testing of PV Modules”
buyers can be empowered with more relevant and accurate data
in IEEE 46th Photovoltaic Specialists Conference (PVSC), 2019.
concerning future system energy delivery. This will encourage [7] N. Blair, N. DiOrio, J. Freeman, P. Gilman, S. Janzou, T. Neises, and M.
the adoption of module technologies with superior lifetime Wagner. 2018. System Advisor Model (SAM) General Description
energy deliveries. (Version 2017.9.5). Golden, CO: National Renewable Energy
Laboratory. NREL/ TP-6A20-70414.
https://www.nrel.gov/docs/fy18osti/70414.pdf

ACKNOWLEDGEMENTS
This material is based upon work supported by the U.S.
Department of Energy’s Office of Energy Efficiency and

978-1-7281-6115-0/20/$31.00 ©2020 IEEE 0813

Authorized licensed use limited to: Escuela Superior de Ingeneria Mecanica. Downloaded on January 31,2023 at 01:22:16 UTC from IEEE Xplore. Restrictions apply.