A Novel Approach to Thermal Design of Solar Modules: Selective-

Spectral and Radiative Cooling

Xingshu Sun†, Rajiv Dubey††, Shashwata Chattopadhyay††, Mohammad Ryyan Khan†, Raghu Vamsi
Chavali†, Timothy J. Silverman†††, Anil Kottantharayil††, Juzer Vasi††, and Muhammad Ashraful Alam†
†Purdue University, West Lafayette, IN, 47907, USA
†† Indian Institute of Technology Bombay, Mumbai, 400076, India
††† National Renewable Energy Laboratory, Golden, Colorado, 80401, USA

Abstract - For commercial solar modules, up to 80% of the panels already heated by the sunlight. However, a scheme
incoming sunlight may be dissipated as heat, potentially raising designed to ‘prevent’ or suppress self-heating could be far more
the temperature 20-30°C higher than the ambient. In the long run, effective fundamental physics of self-heating of PV may yield
extreme self-heating may erode efficiency and shorten lifetime,
thereby, dramatically reducing the total energy output by almost a simpler, yet more efficient cooling for modules, which has not
~10%. Therefore, it is critically important to develop effective and been fully explored.
practical cooling methods to combat PV self-heating. In this paper,
we explore two fundamental sources of PV self-heating, namely, 2.5 100

Pmax degradation rate

sub-bandgap absorption and imperfect thermal radiation. The 2 80
analysis suggests that we redesign the optical and thermal

Tmax (C)
(%/year)
properties of the solar module to eliminate the parasitic absorption 1.5 60
(selective-spectral cooling) and enhance the thermal emission to the 1 40
cold cosmos (radiative cooling). The proposed technique should
0.5 20
cool the module by ~10°C, to be reflected in significant long-term
energy gain (~ 3% to 8% over 25 years) for PV systems under 0 0
different climatic conditions. Hot zone w/ Hot zone w/o Non-hot zone Non-hot zone
hot spots hot spots w/ hot spots w/o hot spots
Pmax degradation rate Tmax
I. INTRODUCTION
Fig. 1 Impact of hot spots on output power decline of PV modules
A typical solar module converts ~20% of the incoming in hot and non-hot climates in India (unpublished 2015 field survey).
sunlight into electricity. Therefore, up to ~80% of the sunlight
may turn into heat dissipated in the module, causing undesired In this paper, the physical origin of PV self-heating is
self-heating as well as performance/reliability degradation [1]. explored, where part of heating is attributed to parasitic sub-
Depending on the environment, the average temperature of a bandgap (sub-BG) absorption and imperfect thermal radiation
PV module can be 20 − 30° C higher than the ambient. In through the atmospheric window. Therefore, we propose to
addition, aged cells with delaminated encapsulation forms local implement the sub-BG optical filter (selective-spectral cooling)
hot spots with temperature approaching 80 ° C, leading to to eliminate the parasitic absorption, and photonic crystal
performance degradation as well as potential fire hazards [2]. thermal emitter (radiative cooling) to enhance thermal emission
The temperature increase in PV modules can create both to the cold cosmos [3]. The cooling design is validated by our
short-term and long-term power losses. In the short term, the electro-thermal coupled simulation, and significant temperature
efficiencies of different PV technologies decreases with reduction is also found for different climates.
temperature, e.g., the efficiency drops by ~ 0.45% for every 1
° C increase for crystalline silicon. In the long term, the
II. PHYSICAL ORIGINS OF SELF-HEATING
reliability of modules suffer from thermally activated
degradations, such as contact corrosion and moisture ingress. A. Parasitic sub-BG absorption
For example, Fig. 1 shows the annual power degradation in In general, only photons with energy above the bandgap can
several climates in India. Note that in both hot and cool excite electron-hole pairs to produce electricity. For Si or CIGS
climates, only a 10 °C temperature rise hot spots reduces the based solar cells (Eg ≈ 1.1 eV), the above-bandgap spectrum
power output by 20%-40%. As a result, it is important to accounts for ~81% of the incident solar irradiance. A 20% cell
develop effective cooling schemes for PV modules to improve converts only one-fourth of the above-bandgap solar energy
both the short-term and the long-term energy yields. into electricity, the rest is converted to heat through carrier
There are several cooling schemes already in use. These recombination, thermalization, and entropy generation [4]. One
include forced airflow, active fluid cooling, liquid submerged way to lower the heat generation from above-bandgap photons
PV, heat pipe-based system, and so on. These methods cool the (~61% of AM 1.5) is to increase the intrinsic solar cell

978-1-5090-2724-8/16/$31.00 ©2016 IEEE 3584

efficiency (by multi-junction design, bulk defect passivation, for solar modules. Each solar cell within the module is divided
etc.), which is not the topic in this paper. On the other hand, into individual sub-cells represented by equivalent circuits and
~19% of the sun light is photons with energy below the connected both electrically and thermally [1], with the
bandgap. Ideally, the sub-BG photons will not be absorbed by thermally balanced boundary conditions.
solar cells, rather it should be reflected back by the back metal To validate the model, we compare the experimental data
sheet. Our measurement of commercial cells, however, against simulation results in Fig. 3. Given a source of localized
indicates strong sub-BG absorbance for both Si and CIGS solar self-heating due to delamination-induced current mismatch [2],
cells, as shown in Fig. 2(a). The absorption may be attributed the simulation results (Fig. 3 (b)) show considerable intra-cell
to imperfect metal reflection as well as free carrier absorption heat transfer through the Si absorber, which helps equalize the
at the highly-doped back surface field in Si and fine grain layer temperature differences within a single cell. This result is in
between CIGS and Mo layers. Consequently, around 15% of excellent agreement with the measured IR image, Fig. 3 (a). In
the sunlight, which consists of the sub-bandgap photons, now contrast, the inter-cell heat transfer through the glass and
heats the solar module, see Fig. 2(b). Hence, in Sec. IV, we backsheet is very limited. Thus, most of the hot cells are
propose to redesign the solar module so that the sub-BG thermally isolated from their “cold” neighbors in both
photons are reflected by a filter before entering the absorber. experimental and simulated results. The inter-cell thermal
Eg ≈ 1.1 eV Energy Budget: isolation prevents hot cell cooling through lateral thermal
AM 1.5 (100 mW/cm2) conduction, prompting the need of novel cooling techniques of
100
~20 % (electricity) these hot spots.
Absorbance (%)

Above Eg
(~ 81%) T (C)
Si ~61 % (heat) Heat
M1 M2 80
50
CIGS (~76 %) 70
Below Eg ~15% (heat)
M1 M2 60
AM 1.5
AM 1.5 (~19%) Hot Cell
~4% (reflected)
0 50
500 1000 1500 2000 2500
wavelength (nm) 40
(a) (b) 30
Heat
Fig. 2 (a) Measured absorbance of crystalline silicon and CIGS
solar cells (no encapsulation layers) vs. photon wavelength. The Si and (a) (b)
CIGS cells are measurement at Indian Institute of Technology Bombay Fig. 3 (a) IR image (facing the backsheet) of a two years old solar
and National Renewable Energy Laboratory, respectively. (b) Energy module installed in India showing the distribution of hot cells. (b)
budget of incident solar irradiance for a Si solar cell with 20% Electro-thermal coupled simulation results of the temperature profile
efficiency (100% transmission through the top encapsulation layers is for cells M1 and M2 in (a).
assumed).

B. Imperfect thermal radiation IV. SELECTIVE-SPECTRAL AND RADIATIVE COOLING

It is well known that the wavelengths of thermal radiation A. Selective-spectral Cooling
from many terrestrial objects exactly match the “transparent” In Fig. 2(a), we have experimentally demonstrated that there
window (between 8 um and 13 um) of the earth atmosphere [5], is significant sub-BG absorption in Si and CIGS solar cells.
see Fig. 4(b). Unfortunately, crystalline silicon does not emits These photons do not contribute to electricity, but cause
much at these wavelength, however, the silica glass (used to significant self-heating. Therefore, we propose to place a low-
encapsulate the c-Si cells) do: its emissivity may approach 75% pass filter between the solar absorber and the top glass layer, as
( ̅ = 0.75 ). The emissivity can be further improved using shown Fig. 4(a). The main function of this filter will be to
appropriately designed photonic crystal, as discussed later in prevent the sub-BG photons from being absorbed by the solar
Sec. IV. This results are consistent with field survey in India, modules, while allowing the photons with higher energy pass
which shows that the module temperature is noticeably higher through. By selectively reflecting the sub-BG photons, one can
in the humid regions [6]. We interpret the difference in cool the solar module as illustrated here.
temperature to the humidity- and CO2-dependent transparency B. Radiative Cooling
of the sky. The presence of water vapor and CO2 reduces the As demonstrated in [5], photonic crystal (PhC) structure
thermal transmittance between 8 – 13 um of the atmosphere, based on silica acts as very good thermal emitter with nearly
directly suppressing thermal radiation from the glass unity emissivity between the atmospheric window, compared
encapsulation to the outer space [3]. to ̅ = 0.75 for planar glass. Such PhC with enhanced thermal
radiation through the atmosphere to the cold universe will lower
the module temperature. Note that the gain through radiative
III. ELECTRO-THERMAL PANEL SIMULATION cooling is suppressed in humid condition and patterning glass
To investigate the degree and distribution of PV self-heating could also rise soiling issues (dust accumulation).
and the benefits from the cooling method, we have developed a
SPICE-based electro-thermal coupled simulation framework

978-1-5090-2724-8/16/$31.00 ©2016 IEEE 3585

 price and improving bankability for photovoltaics are
substantial.
Electricty Improvement Over
reflected: emitted: 8 ~ 13 um photonic crystal

Cumulative Output
25 Years
λ > 1.1 um thermal emitter

increase (%)
10
8 7.8 %
Low pass filter 6
4 3%
2
Solar Cell 0
Hot&Humid Hot&Dry
(a)
Low pass filter Photonic crystal
Emissivity/Absorptivity

(a) (b)
1 Fig. 5 (a) The temperature of a solar module with/without the
cooling design vs. convective heat transfer coefficient, h, for different
desired
climates in India. The wind velocity was related to h by Sturrock model
and ambient temperature = 35 °C is assumed. (b) The total increase of
0.5 Atmospheric electricity output over 25 years by implementing the cooling technique
Transmittance (h = 15 W/m2/K is used in the calculation).

0
0 AM 1.5 5 10 15 IV. CONCLUSION
wavelength (nm)
To summarize, we find that a significant amount of self-
(b)
heating in PV can directly be attributed to the sub-BG
Fig. 4 (a) Schematic of the proposed design for cooling solar absorption and inadequate thermal radiation through the
modules. (b) Desired emissivity/absorptivity spectra of the proposed ‘transparent window’. These results are consistent with
structure. *The atmospheric transmittance was obtained for New experimental data taken from the Indian Survey. To address
Delhi, India in spring from [7].
these issues, we have proposed an idealized thermal design of
C. Results solar modules consisting of a selective-spectral filter to
eliminate parasitic photon absorption and photonic crystal top
Integrated with the sub-bandgap filter and thermal emitter on
layer to enhance thermal radiation to the cold universe.
top of the solar absorber, the proposed module structure with
Substantial temperature reduction in various climate conditions
the cooling effects is in Fig. 4(a), which should be engineered
has been demonstrated based on our coupled electro-thermal
to produce the ideal thermal spectrum shown in Fig. 4(b). To
simulation. To be cost competitive, the filter must be
evaluate the performance of the cooling techniques, we
inexpensive, with negligible absorption loss – it would require
calculate the module temperature for two different climatic
careful design and manufacture. Nonetheless, the simplicity
conditions for which extensive field data are available, namely,
and effectiveness of the cooling method brings new potentials
(i) hot and humid and (ii) hot and dry. The calculation is done
to improve reliability and performance of solar farm installed
using our self-consistent, coupled electro-thermal model
in harsh environments. This work is made possible through
discussed in Sec. II. The results are summarized in Fig. 5.
financial support from the US-India Partnership to Advance
First, simulated results show that the temperature with a
Clean Energy-Research (PACE-R) for the Solar Energy
conventional module is much higher in humid environment due
Research Institute for India and the United States (SERIIUS).
to the suppressed thermal radiation from glass encapsulation.
With the selective-spectral and radiative cooling methods, the [1] T. J. Silverman, M. G. Deceglie, X. Sun, R. L. Garris, M. A. Alam, C. Deline,
module temperature is significantly lowered (∆ ≈ 10 °C at and S. Kurtz, “Thermal and Electrical Effects of Partial Shade in Monolithic
zero airflow) for both humid and dry ambiences, see Fig. 5(a). Thin-Film Photovoltaic Modules,” IEEE J. Photovoltaics, pp. 1–6, 2015.
[2] S. Chattopadhyay, R. Dubey, V. Kuthanazhi, J. J. John, C. S. Solanki, A.
As the weather gets windier (larger h), the most of the heat is Kottantharayil, B. M. Arora, K. L. Narasimhan, V. Kuber, J. Vasi, A. Kumar,
taken away by convection even in the conventional module. and O. S. Sastry, “Visual Degradation in Field-Aged Crystalline Silicon PV
Modules in India and Correlation With Electrical Degradation,” Ieee J.
Here we have used the Sturrock model which relates ℎ linearly Photovoltaics, vol. 4, no. 6, pp. 1470–1476, 2014.
to wind speed (10 W/m2/K for 0 m/s and 40 W/m2/K for 8 m/s). [3] C. G. Granqvist, “Radiative cooling to low temperatures: General
considerations and application to selectively emitting SiO films,” J. Appl.
Therefore, we expect lower gain in cooling using our new Phys., vol. 52, no. 6, p. 4205, 1981.
structure; nonetheless, some degree of cooling is anticipated at [4] L. C. Hirst and N. J. Ekins-daukes, “Fundamental losses in solar cells,” Prog.
PHOTOVOLTAICS Res. Appl., vol. 19, no. August 2010, pp. 286–293, 2011.
wind flow conditions. Given the temperature coefficient of Si [5] L. Zhu, A. P. Raman, and S. Fan, “Radiative cooling of solar absorbers using a
solar cell and the empirical equation for preformation visibly transparent photonic crystal thermal blackbody,” Proc. Natl. Acad. Sci.,
degradation rate [9], simulation shows significant improvement p. 201509453, Sep. 2015.
[6] “All-India India Survey of Photovoltaic Module Degradation : 2013.”
(~ 3 to 8 %) in cumulative power output over 25 years by the [7] “ATRAN,” 1992. [Online]. Available: https://atran.sofia.usra.edu/cgi-
cooling method for different climates, see Fig. 5(b). The bin/atran/atran.cgi.
implications of such large electricity increment in reducing the

978-1-5090-2724-8/16/$31.00 ©2016 IEEE 3586