A Detailed Study On Loss Processes in Solar Cells

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Energy 144 (2018) 490e500

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

A detailed study on loss processes in solar cells


Ao Wang, Yimin Xuan*
School of Energy and Power Engineering, Nanjing University of Science and Technology, No. 200 Xiaolingwei Road, 210094 Nanjing, China

a r t i c l e i n f o a b s t r a c t

Article history: Only a small part of the incident solar energy converts to the electrical power in photovoltaic devices.
Available online 29 December 2017 The majority of the energy loss contributes to the heat generation in devices and thus leads to a
temperature rise, causing an inevitable impact on the performance of photovoltaic devices. Hence, loss
Keywords: processes in solar cells play very important roles in solar-electric conversion process. This paper
Solar cell systematically studies both the intrinsic and extrinsic losses in solar cells. Energy distributions of solar
Loss process
cells with different kinds of parameters are presented to characterize the different kinds of loss
Sensitivity
processes in detail. The sensitivities of loss processes to the structural and operating parameters of
Efficiency
solar cells such as external radiative efficiency, solid angle of absorption and operating temperature are
discussed, for the parameters have significant impact on the loss processes. The external radiative
efficiency, solid angle of absorption (e.g., the concentrator photovoltaic system), series resistance and
operating temperature are demonstrated to greatly affect the loss processes. Furthermore, based on
the calculated thermal equilibrium states, the temperature coefficients of solar cells versus the
bandgap Eg are plotted.
© 2017 Elsevier Ltd. All rights reserved.

1. Introduction highest efficiency of photovoltaic devices theoretically [5]. Hirst


and Ekins-Daukes studied the mechanisms of five intrinsic loss
Solar energy photovoltaic technology has developed rapidly processes quantitatively and provided a mathematical and
for the past years and researchers over the world have been graphical demonstration [10]. The majority of papers discussed
working hard on improving the efficiency and reducing the cost the fundamental (intrinsic) loss processes of photovoltaic devices.
of photovoltaic devices. However, the efficiency of photovoltaic In recent years, more and more groups have been studying the full
devices grows slowly in recent years. Many groups have been loss processes, including the intrinsic and extrinsic losses, for the
studying the limiting efficiency and the factors that limit the extrinsic loss also plays an important role in photovoltaic energy
efficiency of photovoltaic devices, trying to find effective ways conversion [11e15]. And as well known, the heat generated in
to reduce the energy loss in the process of photovoltaic energy solar cells will lead a temperature rise, which unavoidably causes
conversion and increase the final output efficiency [1e9]. an efficiency drop [16e20]. Thus, when studying the loss pro-
Shockley and Queisser developed the detailed balance cesses and output parameters of photovoltaic devices, the impact
approach with generalized Planck function to derive the of heat generation must be taken into consideration. Dupre  et al.
maximum power point and give the elementary detailed balance presented a thermal model for photovoltaic devices, involving
limit of efficiency, which only involved the fundamental loss intrinsic loss, extrinsic loss and heat generation [11,12,14]. The
processes [1]. Henry published a graphical method, which clearly temperature of solar cells was derived based on the calculation of
demonstrated the distribution of various intrinsic losses for both heat generation and a given global heat transfer coefficient [12].
single and multiple energy gap ideal solar cells and gave the As discussed above, loss processes are very important in the
maximum work done by per absorbed photon [2]. Wurfel pub- process of solar-electric conversion, and so far a number of
lished the thermodynamic limitations to solar energy conversion research efforts have been devoted to them. However, in a real
based on the second principle of thermodynamics, which is the solar cell, these loss processes are dependent on the different
kinds of parameters, including the structure parameters and
operating parameters of the cell. While little work has been
concentrated on the control ability of parameters on different loss
* Corresponding author.
E-mail address: ymxuan@njust.edu.cn (Y. Xuan). processes in solar cells.

https://doi.org/10.1016/j.energy.2017.12.058
0360-5442/© 2017 Elsevier Ltd. All rights reserved.
A. Wang, Y. Xuan / Energy 144 (2018) 490e500 491

This paper studies the loss processes in photovoltaic devices across the bandgap and finally their energy is transferred to
depending on different kinds of parameters, such as external phonons in the device, contributing to heat generation instead of
radiative efficiency (ERE), solid angle of absorption, resistances electricity.
and operating temperature. Energy distributions of a crystalline
silicon (c-Si) solar cell and a CH3NH3PbI3 perovskite (C-P) solar ZEg
cell are presented to characterize the intrinsic and extrinsic losses PBelow ¼ n ð1  Rc  Tc ÞPFDðEÞ$EdE (2)
in detail, calculated by a thermal model based on the model 0
proposed by Dupre  et al. [11,12,14]. Energy distributions of solar
cells with different concentration ratios and external radiative The thermalization loss originates from the fraction of the
efficiencies (EREs) are also plotted versus the bandgap Eg. Among photons transmit into the cells that have the energy larger than the
the losses, below Eg loss, thermalization loss, angle mismatch loss bandgap. The free carriers excited by these photons release their
and non-radiative recombination loss are demonstrated playing energy in excess of the bandgap (E-Eg) to the phonons, contributing
the leading roles in energy loss processes. Considering that the to heat generation [12].
parameters of the cells greatly affect the loss processes in
Z∞
photovoltaic devices, the sensitivities of loss processes to struc-  
PThermalization ¼ n ð1  Rc  Tc ÞPFDðEÞ$ E  Eg dE (3)
ture parameters (e.g., external radiative efficiency, solid angle of
absorption, resistances, etc.) and operating parameters (e.g., Eg
operating temperature) are studied. Furthermore, different kinds
The emission loss originates from the photon emission of the
of parameters are demonstrated to have different control abilities
cells as a result of radiative recombination.
on different loss processes in photovoltaic devices. Meanwhile,
heat generation resulting from the loss processes will cause a Z∞
considerable temperature rise. Taking it into consideration, the PEmission ¼ Eg nðE; Ta ; qVc ; Ue ÞdE (4)
efficiency of a cell drops remarkably with the decrease of
Eg
the photocurrent and voltage, especially for the concentrator
photovoltaic (CPV) system if the heat generation is not well
where Ta is the temperature of the cell (absorber), Vc is the voltage
dissipated. Suppressing the heat generation to reduce the oper-
across the cell and Ue is the solid angle of emission. Here the total
ating temperature is very important to the optimization of output
photon emission flux n(E, T, m, U) is introduced [2,10]:
efficiency. Based on the calculated thermal equilibrium states, the
temperature coefficients of solar cells versus the bandgap Eg are 2U E2
plotted. This paper is structured as follows. Firstly, the general nðE; T; m; UÞ ¼ Em
(5)
c h eð kT Þ  1
2 3
thermal model for solar cells is introduced in Section 2, including
the intrinsic loss and extrinsic loss. In Section 3, the detailed an-
where T is the temperature of the emitter, m is the chemical
alyses of intrinsic and extrinsic loss processes are conducted,
potential, c is the speed of light, h is Planck's constant and k is
respectively. Effects of different parameters on the performance of
Boltzmann's constant. Thus Eq. (4) is expressed as:
a cell are further studied in Section 4. Finally, the conclusions are
drawn and some suggestions are presented to help researchers Z∞ Z∞
Eg
design high-efficiency solar cells. PEmission ¼ Eg nðE; Ta ; qV; Ue ÞdE ¼ $q nðE; Ta ; qV; Ue ÞdE
q
Eg Eg
2. General model for solar cells
Eg  
z $Jemit Ue ; Ta ; Eg ; V
2.1. Intrinsic losses in solar cells q
(6)
Loss processes in solar cells consist of two parts: intrinsic
losses (fundamental losses) and extrinsic losses. Intrinsic losses where Jemit(U, T, E, V) refers to the photocurrent loss caused by
are unavoidable in single bandgap solar cells, even if in the radiative recombination, defined as:
idealized solar cells [10]. In this paper, intrinsic losses are divided  
into six processes: the optical loss, the below Eg loss, the ther- 2Uq  2  qV  E
Jemit ðU; T; E; VÞ ¼ E kT þ 2Ek2 2
T þ 2k3 3
T exp
malization loss, the emission loss, the Carnot loss and the angle c2 h3 kT
mismatch loss, and the model in this paper is based on the model (7)
presented by Dupre  et al. [11,12,14].
The optical loss corresponds to the fraction of incident energy The Carnot loss corresponds to the energy that needs to be
sacrificed for doing work between the hot reservoir (the Sun) and
of the Sun's radiation that is reflected or transmits through the
cells. the cold reservoir (a solar cell) [10]. This process leads to a voltage
loss DVCarnot, and the corresponding heat generation is:
Z∞
POptical ¼ n ðRc þ Tc ÞPFDðEÞ$EdE (1) PCarnot ¼ J DVCarnot (8)
0
where J is the output photocurrent of the cell.
The angle mismatch loss corresponds to the energy loss caused
where n is the concentration ratio of the cells (n suns), Rc and Tc
by the mismatch between the solid angles of absorption and
are the reflectance and transmittance of the cells, respectively.
emission [10], leading to a voltage loss DVAngle, and the corre-
PFD(E) is the AM1.5 photon flux density as a function of photon
sponding heat generation is:
energy E [12].
The below Eg loss corresponds to the fraction of the photons PAngle ¼ J DVAngle (9)
transmit into the cells that have the energy below the bandgap Eg.
These photons don't have enough energy to excite free electrons
492 A. Wang, Y. Xuan / Energy 144 (2018) 490e500

2.2. Extrinsic losses in solar cells Taking all the loss processes mentioned above into consider-
ation, the output photocurrent density of a cell can be written as:
Besides the intrinsic losses, extrinsic losses, such as non-
radiative recombination (NRR) loss, series resistance (Rse) loss,
shunt resistance (Rsh) loss and parasitic absorption loss [12,15], J ¼ Jph  ðJemission þ Jnonemission Þ  Jshunt
also play a very important role in loss processes in single bandgap Z∞
solar cells. Different from intrinsic losses, they are avoidable [10]. 1 V þ J$Rse
¼q ð1  Rc  Tc ÞPFDðEÞdE  $J 
For simplicity, parasitic absorption is ignored in the following ERE emission Rsh
Eg
calculation. Fig. 1 shows the equivalent circuit diagram of a solar
cell. (10)
To study the extrinsic loss processes quantitatively, the concept
of external radiative efficiency (ERE) is defined to describe the where Jph is the maximum photocurrent density, Jshunt is the
photocurrent loss caused by recombinations in a solar cell [21,22]. photocurrent loss caused by shunt resistance and V is the output
The term ERE refers to the fraction of recombinations that leads to voltage.
photon emission from a cell [22]. The rest fraction of the re- Non-radiative recombination (NRR) losses consist of photocur-
combinations (1-ERE) refers to the non-radiative recombinations rent loss (NRR-J) and voltage loss (NRR-V), originating from
taking place inside a solar cell, and thus ERE ¼ Jemission/ different mechanisms: Shockley-Read-Hall (SRH) recombination,
(Jemission þ Jnon-emission), where Jemission and Jnon-emission refer to Auger recombination, inner imperfection, etc [12]. The heat gen-
the photocurrent losses caused by radiative recombination eration corresponding to the NRR-J loss is [14]:
and non-radiative recombination, respectively. In practical
solar cells, structural defects or impurities are very common, and  
JNRRJ Eg Eg 1
they generally result in the recombinations of the electron- PNRRJ ¼ Eg $ ¼ $Jnonemission ¼ $  1 Jemission
q q q ERE
hole pairs, especially the non-radiative recombinations (e.g.,
Shockley-Read-Hall recombination), reducing the carriers (11)
lifetime or suppressing the cell efficiency [23]. For the random-
Here Jemission ¼ Jemit (Ue, Ta, Eg, V þ J$Rse), thus:
ness and unpredictability of the defects, it is difficult to analyze
the effects of structural defects quantitatively and accurately. To
study the loss processes in solar cells systematically, in this paper,  
Eg 1 2Ue q
the concept of external radiative efficiency is used to quantita- PNRRJ ¼ $ 1 $ 2 3
q ERE c h
tively analyze the recombination processes in solar cells. The ERE   
qðV þ J$Rse Þ  Eg
of a solar cell is similar to the concept of external quantum Eg 2 kTa þ2Eg k2 Ta 2 þ2k3 Ta 3 exp
efficiency (EQE) in a light-emitting diode [22]. With this defini- kTa
tion, the effects of radiative recombinations and non-radiative (12)
combinations in a cell can be analyzed quantitatively and
To derive the heat generation corresponding to the voltage
comprehensively, including the effects of structural defects on
loss, DVNRR is needed. Using eq. (10), the maximum power point
recombination processes.
of the cell can be found. Based on the research of Dupre et al.
[11,14], VMPP-Angle and VMPP-NRR can be calculated respectively
with the specific parameters, which will be discussed subse-
quently. Thus, DVMPP-NRR ¼ VMPP-Angle - VMPP-NRR, and the heat
generation corresponding to the NRR-V loss at the maximum
power point is:

PNRRV ¼ JMPP $DVMPPNRR (13)


Series loss corresponds to the energy loss that caused by
the series resistance in solar cells. This series resistance can also
include the contact resistance, and leads to the heat generation
corresponding to the voltage loss (DVse ¼ JRse) in the form of Joule
heating [11]:

Pseries ¼ J 2 Rse (14)


Shunt loss corresponds to the energy loss that caused by
the shunt resistance in solar cells. Its loss process can be assimi-
lated to recombination process, leading to the heat generation
corresponding to the photocurrent loss [11]:

JShunt Eg V þ J$Rse
PShunt ¼ Eg $ ¼ $ (15)
q q Rsh

where JShunt is the current of the shunt path.


Fig. 1. Equivalent circuit diagram of a solar cell showing the load, series resistance Based on the discussion above, the thermal equilibrium
(Rse), shunt resistance (Rsh) and the voltage across the cell (Vc). equation at the maximum power point is:
A. Wang, Y. Xuan / Energy 144 (2018) 490e500 493

3.1.1. Intrinsic loss processes of a crystalline silicon solar cell


PIncident ¼ POptical þ PBelow þ PThermalization þ PEmission þ PCarnot The intrinsic loss processes of a crystalline silicon (c-Si) solar
þ PAngle þ PNRRJ þ PNRRV þ PSeries þ PShunt þ POutput cell at different concentration ratios (n ¼ 1 and 5) with the
bandgap of 1.1246 eV at 298.15 K (25  C) are presented in Table 1,
(16)
under an AM1.5 solar illumination (PIncident ¼ 1000.37 W/m2,
calculated by the integral of PFD(E)). The errors D in the table
where POutput ¼ VMPP·JMPP is the output electrical power at the
equal to the differences between the standard values (Jph, Vmax and
maximum power point and PIncident is the incident power density of
PIncident, where Vmax ¼ Eg/q is the maximum voltage can be built in
solar illumination. The total heat generation in a cell is:
the cell) and the sums of the values of different processes, which
PHeat ¼ PBelow þ PThermalization þ PCarnot þ PAngle þ PNRRJ prove the model of high accuracy. The result indicates that the
dominant intrinsic losses are below Eg loss, thermalization loss,
þ PNRRV þ PSeries þ PShunt (17) and angle mismatch loss. They take up the majority of the incident
energy, and the angle mismatch loss decreases as the solid angle
The output efficiency of the cell at room temperature (298.15 K)
of absorption increases, improving the output efficiency (from
is h ¼ POutput/PIncident.
33.52% of n ¼ 1e35.27% of n ¼ 5), which agrees with the detailed
Heat generation due to the losses results in a temperature rise.
balance theory. When the concentration ratio increases, the
Considering that the thermal radiative heat exchange between the
output current density and voltage both increase. Table 1 also
cell and the environment is negligible [11,12,14], the cell's heat
indicates that a large amount of solar energy is converted into
equilibrium between the heat generation and the heat lost via
heat generation.
conduction and convection between the cell and the environment
can be approximately written as [12]:
3.1.2. Intrinsic loss processes of a CH3NH3PbI3 perovskite solar cell
PHeat ¼ hðTa  Tenv Þ (18) The intrinsic loss processes of a CH3NH3PbI3 perovskite (C-P)
solar cell with the bandgap of 1.639 eV at 298.15 K (25  C) are
where h is a global heat transfer coefficient and Tenv is the presented in Table 2. Compared with the c-Si solar cell in Table 1,
temperature of the environment. PHeat is a function of Ta and the C-P solar cell has larger bandgap Eg, and thus, it has lower
thus the final operating state of the cell needs to be calculated photocurrent density, higher voltage, larger below Eg loss and
iteratively. With the final heat generation PHeat-f and operating smaller thermalization loss.
temperature Ta-f, the output efficiency of the cell operating at the Fig. 2 shows the energy distributions of a c-Si solar cell and a
final thermal equilibrium state is: hf ¼ POutput-f/PIncident. C-P solar cell. Fig. 3 illustrates the efficiencies of ideal solar cells
varying with the bandgap Eg only considering the intrinsic
losses at different concentration ratios. In Fig. 3, the three lines
3. Detailed analysis of loss processes in solar cells distinguish with each other obviously at the bandgaps lower than
about 2.0 eV, which is because that for the intrinsic losses, the
3.1. Analysis of intrinsic loss processes concentration ratio mainly has the impact on the angle mismatch
loss, and as discussed later, for the bandgaps larger than about
To discuss the intrinsic losses of solar cells quantitatively, the 2.0 eV, the angle mismatch loss is becoming negligible gradually.
reflectance Rc and the transmittance Tc of the cell are set to zero, And that's why the line of higher concentration ratio is higher
ERE is set to 1, and resistances of the cell (Rse and Rsh) are ignored. than that of lower concentration ratio.
Then eq. (10) is written as: For ideal solar cells considering only the intrinsic losses (Fig. 3),
the relative error is d ¼ (|POutput þ PHeat þ PEmission - PIncident|)/
Z∞ Z∞ PIncident. For calculation, the sum of the energy distributions
 
J¼q PFDðEÞdE  Jemission ¼ q PFDðEÞdE  Jemit Ue ; Ta ; Eg ; V expressed in the bracket agrees well with the incident solar
Eg Eg energy, and the relative errors are all below 107.

(19)
3.2. Analysis of intrinsic and extrinsic loss processes
The solid angles of emission (Ue) and absorption (Ua) of the cell
from the earth to the sun without concentration in this paper are: By taking all loss processes into consideration, eq. (10) is used
Ue ¼ p and Ua ¼ 6.8  105 sr, respectively. to derive the necessary parameters. Firstly, VMPP and JMPP are

Table 1
Energy distribution and corresponding parameters of a c-Si solar cell with different concentration ratios at room temperature.

Crystalline silicon solar cell, Eg ¼ 1.1246 eV

Concentration ratio: n ¼ 1 Concentration ratio: n ¼ 5

Current Voltage Power /PIncident Current Voltage Power /PIncident

A/m2 V W/m2 % A/m2 V W/m2 %

Jph/Vmax/PIncident 436.25 1.1246 1000.37 100 2181.28 1.1246 5001.85 100


PBelow 194.60 19.45 973.01 19.45
PThermalization 315.16 31.51 1575.78 31.50
PEmission 13.68 15.39 1.54 65.17 73.29 1.47
PCarnot 0.0623 26.33 2.63 0.0623 131.83 2.64
PAngle 0.2687 113.54 11.35 0.2286 483.74 9.67
POutput 422.57 0.7936 335.35 33.52(h) 2116.11 0.8337 1764.20 35.27(h)
D 0 0 0 0 0 0 0 0
PHeat 649.63 64.94 3164.36 63.26
494 A. Wang, Y. Xuan / Energy 144 (2018) 490e500

Table 2
Energy distribution and corresponding parameters of a C-P solar cell with different concentration ratios at room temperature.

CH3NH3PbI3 perovskite solar cell, Eg ¼ 1.639 eV

Concentration ratio: n ¼ 1 Concentration ratio: n ¼ 5

Current Voltage Power /PIncident Current Voltage Power /PIncident

A/m2 V W/m2 % A/m2 V W/m2 %

Jph/Vmax/PIncident 244.33 A 1.639 1000.37 100 1221.67 1.639 5001.85 W 100


PBelow 456.38 45.62 2281.88 45.62
PThermalization 143.53 14.35 717.65 14.35
PEmission 4.87 7.99 0.80 23.58 38.65 0.77
PCarnot 0.106 25.26 2.52 0.1055 126.40 2.53
PAngle 0.271 64.89 6.49 0.2305 276.16 5.52
POutput 239.46 1.262 302.32 30.22(h) 1198.09 1.303 1561.11 31.21(h)
D 0 0 0 0 0 0 0 0
PHeat 690.06 3402.09

calculated finding the maximum power point (POut-


by  
put ¼ VMPP·JMPP) with the chosen parameters: Rse ¼ 1 U cm2 and J ¼ Jph  Jemission ¼ Jph  Jemit Ue ; Ta ; Eg ; V (21)
Rsh ¼ 250 U cm2. The resistance values are chosen to clearly illus-
trate the impact of resistances on loss processes of a solar cell, and Then we can derive the voltage (VAngle-MPP) at the maximum
cases of different resistance values are calculated subsequently to power point, and thus the voltage loss due to angle mismatch loss is
further discuss the impact. Noting that, in some recent studies, the DVAngle-MPP ¼ VCarnot-MPP -VAngle-MPP.
solar cells have EREs as high as about 102 or even higher [22e25], To calculate the non-radiative recombination loss, the voltage
thus in this paper, cases of different EREs (ERE ¼ 1, 102 and 105) loss DVNRR ¼ VAngle eVNRR is needed [11]. The same conditions as in
are calculated to illustrate the impact of external radiative effi- angle mismatch loss above are used except that the impact of EREs
ciency on loss processes. For simplicity, the reflectance Rc is set to (cases of ERE ¼ 1, 102 and 105) is considered, and eq. (10) is
0.05 and the transmittance Tc is set to zero. rewritten as:
To calculate the Carnot loss, the voltage loss DVCarnot ¼ Eg
1 1  
eVCarnot is needed [11]. Rse ¼ 0, Rsh ¼ ∞, Ua as the solid angle of J ¼ Jph  $J ¼ Jph  $J Ue ; Ta ; Eg ; V (22)
emission and in the radiative limit (ERE ¼ 1), eq. (10) is rewritten ERE emission ERE emit
as: Then we can derive the voltage (VNRR-MPP) at the maximum
power point, and thus the voltage loss due to NRR loss is DVNRR-
  MPP ¼ VAngle-MPP eVNRR-MPP.
J ¼ Jph  Jemission ¼ Jph  Jemit Ua ; Ta ; Eg ; V (20) Cases of c-Si solar cell with different EREs (1, 102 and 105) at
room temperature are presented in Tables 3 and 4. Compared with
Then we can derive the voltage (VCarnot-MPP) at the maximum
the energy distribution in the radiative limit (ERE ¼ 1), the output
power point, and thus the voltage loss due to Carnot loss process is
power decreases significantly with the decrease of ERE, down to
DVCarnot-MPP ¼ Eg eVCarnot-MPP. 23.84% (ERE ¼ 102) and 17.58% (ERE ¼ 105) from 27.92%
To calculate the angle mismatch loss, the voltage loss DVAn-
(ERE ¼ 1). The ERE has little impact on the loss processes except
gle ¼ VCarnot -VAngle is needed [11]. The same conditions as in Carnot
the NRR loss, and almost all the increased power loss (i.e., NRR
loss above are used except that the solid angle of emission is now
loss) is converted into heat generation, which will further increase
set to Ue, and eq. (10) is rewritten as:
the temperature of the cell, reducing the output efficiency as a

Fig. 2. Energy distributions of a c-Si solar cell: (a) and (b), and a C-P solar cell: (c) and Fig. 3. Efficiency of ideal solar cells only considering the intrinsic losses at different
(d). The concentration ratios in (a) and (c) are n ¼ 1 and in (b) and (d) are n ¼ 5. concentration ratios (n ¼ 1,5 and 10).
A. Wang, Y. Xuan / Energy 144 (2018) 490e500 495

Table 3
Energy distribution and corresponding parameters of a c-Si solar cell with EREs as 1 and 102 at room temperature.

result. Besides the optical loss process, in a real solar cell (i.e., Rsh (Eq. (15)). Thus double Rsh (from 250 U cm2 to 500 U cm2) will
EREs1), the photocurrent loss is mainly caused by NRR loss reduce JShunt by half from 27.04 A/m2 to 13.55 A/m2, increasing the
process and shunt loss process. NRR loss process also contributes output current JMMP significantly and having little impact on
to an important component of voltage loss. Thus, it is very output voltage VMMP. In general, to achieve higher output current
important to improve the ERE of a solar cell to achieve higher density and output voltage, it is necessary to increase the shunt
output current density and output voltage. resistance and reduce the series resistance, respectively.
To further study the impact of resistances on loss processes, To study loss processes in a concentrated photovoltaic (CPV)
cases of different resistances are presented in Table 5 to be system, cases of different concentration ratios (n ¼ 2 and 5) are
compared with that in Table 3(ERE ¼ 102). The series loss is presented in Table 6 to be compared with the case of n ¼ 1 in
proportional to the series resistance Rse according to Eq. (14). Thus Table 3, and the transmittance of the concentrator is set to 0.95 in
reducing Rse by half (from 1 U cm2 to 0.5 U cm2) will also reduce this calculation. For angle mismatch loss process, increasing the
the series loss by half from 13.88 W/m2 to 6.97 W/m2 (i.e., reduce concentration ratio means increasing solid angle of absorption
the voltage loss due to series resistance by half, from 0.0373 V to (Ua) of the cell, which reduces the voltage loss DVAngle (from
0.0187 V), having little impact on output current JMPP. The photo- 0.2687 V of n ¼ 1 to 0.2514 V of n ¼ 2 and 0.2285 V of n ¼ 5), also
current loss JShunt is inversely proportional to the shunt resistance demonstrated by generalized Planck equation and Boltzmann

Table 4
Energy distribution and corresponding parameters of a c-Si solar cell with ERE as 105 at room temperature.
496 A. Wang, Y. Xuan / Energy 144 (2018) 490e500

Table 5
Energy distribution and corresponding parameters of a c-Si solar cell with different resistances at room temperature.

approximation: DVAngle ¼ Tak/q·ln(Ue/Ua) [8]. For a high concen- reached. Thus, we can get the final operating parameters (Ta-f, PHeat-
tration ratio, though the angle mismatch loss PAngle decreases f, POutput-f, and hf) of the cell at the thermal equilibrium state. The
(from 10.01% of n ¼ 1e8.29% of n ¼ 5), the high photocurrent final thermal equilibrium states of c-Si solar cell with different
density leads to the increase of some other losses (NRR-J loss and concentration ratios (n ¼ 1 and 2) are presented in Table 7 to be
series resistance loss, which are related to the photocurrent compared with the states at room temperature in Table 3(n ¼ 1)
density J), leading to a larger heat generation (i.e., a higher tem- and Table 6(n ¼ 2).
perature rise of the cell), which will further reduce the output Heat generation causes a considerable temperature rise, near
efficiency. Thus, reducing the angle mismatch loss via increasing to 100 K in the case of non-CPV c-Si solar cell (considering the
Ua is not effective if the heat generation is not well dissipated. global heat transfer coefficient as h ¼ 10 W m2 K1 in this paper).
Besides, according to Eq. (14), the enlarged photocurrent density J And for lower ERE, there will be a higher Ta-f. In calculation, the
will increase the voltage loss DVse due to series resistance (from impact of the cell's temperature on bandgap Eg is considered
0.0373 V of n ¼ 1 to 0.1814V of n ¼ 5), reducing the output voltage approximately with dEg/dT ¼ 0.27 meV/K for c-Si around 300 K
(from 0.6401 V of n ¼ 1 to 0.5507 V of n ¼ 5). A CPV system is [14]. When temperature increases, Eg decreases, and thus the
designed to achieve much more energy from solar irradiation in maximum photocurrent density increases. However, at the same
unit area of solar cells, with the ultimate purpose of saving money. time, the photocurrent loss caused by emission and NRR is
However, concentrated solar irradiation results in a large amount enlarged. When the temperature rise is not significant (e.g., 75 K
of heat generation, and if not dissipated timely, it will cause a of n ¼ 1), the increase of the photocurrent loss is smaller than that
dramatic temperature rise, reducing the output efficiency greatly of maximum photocurrent density. However, when the temper-
and finally making the ultimate purpose impossible. ature rise is significantly high (e.g., 153 K of n ¼ 2, or higher tem-
All the study above doesn't involve the impact of heat genera- perature rise of higher n), the increase of the photocurrent loss
tion (i.e., temperature rise). However, temperature is a crucial will be bigger than that of the maximum photocurrent density,
limiting factor to increasing the output efficiency of solar cells. which means the output photocurrent density (JMPP-f) at thermal
Substitute the calculated parameters above into eq. (17) and (18) equilibrium state will be lower than that at room temperature,
iteratively, and the final thermal equilibrium state will be especially for the CPV solar cells. For the case of higher

Table 6
Energy distribution and corresponding parameters of a c-Si solar cell with different concentration ratios at room temperature.
A. Wang, Y. Xuan / Energy 144 (2018) 490e500 497

Table 7
Energy distribution and corresponding parameters of a c-Si solar cell with different concentration ratios at thermal equilibrium state.

Fig. 4. Energy distribution of solar cells in radiative limit at room temperature. The transmittance Tc and the reflectance Rc are both set to zero. The concentration ratios are n ¼ 1 in
(a), n ¼ 3 in (b) and n ¼ 100 in (c), respectively.

concentration ratio, the output photocurrent density decreases both the output photocurrent density and the output voltage.
significantly.
As for the voltage losses, the components due to Carnot loss,
4. Effects of cells' parameters on the loss processes
angle mismatch loss and NRR loss are all enlarged for they are
proportional to the temperature of the cell, and the component due
Energy distributions of solar cells with different concentration
to series resistance varies with output photocurrent density, for it is
ratios (n ¼ 1, 3, and 100) and EREs are presented in Fig. 4 (ERE ¼ 1)
proportional to J2MPP-f. In addition, Vmax ¼ Eg/q decreases with the
and Fig. 5 (ERE ¼ 102), respectively. The relative errors of the
rise of temperature. As a result of all the effects, the output voltage
calculations are all below 2% at different concentration ratios (n ¼ 1,
VMPP-f at thermal equilibrium state drops a lot than that at room
3, 100). For the case of ERE ¼ 1, the average relative errors over the
temperature.
bandgaps at different concentration ratios are 0.36% (n ¼ 1), 0.25%
In general, taking the temperature rise into consideration,
(n ¼ 3) and 0.62% (n ¼ 100), respectively. For the case of ERE ¼ 102,
output efficiency of a solar cell drops remarkably especially for the
the average relative errors are 0.34% (n ¼ 1), 0.27% (n ¼ 3) and 0.55%
CPV system if the heat generation is not well dissipated, reducing
(n ¼ 100), respectively. When considering the temperature rise of

Fig. 5. Energy distribution of solar cells considering the non-radiative recombination at room temperature. The transmittance Tc and the reflectance Rc are both set to zero. The
concentration ratios are n ¼ 1 in (a), n ¼ 3 in (b) and n ¼ 100 in (c), respectively.
498 A. Wang, Y. Xuan / Energy 144 (2018) 490e500

Fig. 6. Efficiency of solar cells at room temperature and at thermal equilibrium state. In calculation, n ¼ 1, Ta ¼ 298.15 K and Ta-f is the temperature at final thermal equilibrium state.

Fig. 7. Efficiency of solar cells with different concentration ratios at thermal equilibrium state (n ¼ 1, 2 and 3). Ta-f is the temperature at final thermal equilibrium state.

solar cells (Figs. 6 and 7), our calculations still agree well with the absorption angle Ua, which means the solar cell will collect much
incident solar energy, with all the relative errors below 2%. For the more solar energy in unit area and much more heat will be
case of n ¼ 1, the average relative errors over the bandgaps with generated, there will be a significant temperature rise in the cell if
different EREs are 0.35% (ERE ¼ 1), 0.32% (ERE ¼ 102) and 0.26% the cooling system is not well enhanced. In our calculation, taking
(ERE ¼ 105), respectively. For the case of n ¼ 2, the average relative the heat generation into consideration, PAngle in the case of n ¼ 2
errors are 0.28% (ERE ¼ 1) and 0.27% (ERE ¼ 102), respectively. For increases from 12.45% to 13.09%, rather than decreases, compared
the case of n ¼ 3, the average relative errors are 0.34% (ERE ¼ 1) and with the case of n ¼ 1, and what's even worth is that due to the
0.27% (ERE ¼ 102), respectively. temperature rise caused by the increased absorption angle, the
To design a photovoltaic device, the bandgap of material should output power POutput decreases from 19.28% to 13.61% for a c-Si
be the first parameter to be considered, for the majority of the solar cell. And in Figs. 6 and 7, the efficiency of solar cells drops
energy loss (below Eg loss and thermalization loss) depends on the remarkably with the significant temperature rise, even though the
bandgap. The below Eg loss and thermalization loss cover more concentration ratios are very low. So, reducing the angle mismatch
than half of the total incident power for cases of different con- loss via increasing the absorption angle is not effective unless the
centration ratios and EREs at all bandgaps. And these two loss problem of temperature rise is well solved, which is also the critical
processes only relate to the bandgap Eg. problem of a CPV system.
In some recent studies, increasing the absorption angle (e.g., in a In addition to the effect on angle mismatch loss process, in
CPV system) is proposed as a common approach to reduce the angle Figs. 4 and 5, concentration ratio is demonstrated to have effects on
mismatch loss [10,14]. A CPV system with a low concentration ratio series loss process, shunt loss process, emission loss process
(e.g., n ¼ 3 in Figs. 4 and 5) will not reduce the angle mismatch loss (ERE ¼ 1) and NRR loss process (EREs1). For a CPV system, the
remarkably. When the concentration ratio is high enough (e.g., photocurrent density is enhanced, and as a result, the series loss is
n ¼ 100 in Figs. 4 and 5), the angle mismatch loss is greatly sup- enhanced remarkably. Meanwhile, due to the enhanced photocur-
pressed. However, this theoretically feasible method will result in rent density, the voltage loss of the cell is enhanced, and according
another problem: the temperature rise. When multiplying the to Eq. (15), the shunt loss is thus reduced, negligible for high
A. Wang, Y. Xuan / Energy 144 (2018) 490e500 499

concentration ratio. For solar cells at high concentration ratio (e.g.,  


n ¼ 100), the output photocurrent density is very high, and thus hf  h
according to Eq. (11) and (12), the emission loss is greatly enhanced TCð=KÞ ¼   (23)
for solar cells in radiative limit (ERE ¼ 1), while for real solar cells Taf  298:15 h
(EREs1), the NRR-J loss is greatly enhanced and the NRR-V loss
Based on the preceding discussion, a conclusion is drawn that
becomes negligible. Because of the greatly enhanced losses, the
temperature rise is a very important limiting factor in the efficiency
output power of high-concentration-ratio photovoltaic system is
of a certain solar cell. For solar cells with bandgap Eg varying from
reduced significantly as presented in Figs. 4(c) and 5(c).
1 eV to 3 eV, we can see the main energy losses consist of the below
The effect of shunt resistance mainly reflects in PShunt. According
Eg loss, the thermalization loss and the angle mismatch loss. And all
to Eq. (15), PShunt is inversely proportional to Rsh, and thus for large
these three kinds of losses contribute to heat generation, causing a
Rsh, small JShunt will be achieved (i.e., large JOutput will be achieved).
significant temperature rise, which greatly limits the efficiency of
Here, we must emphasize the effect of series resistance on loss
solar cells. The concentration ratio (or Ua), series resistance and
processes. Series resistance plays a significant role in loss processes,
external radiative efficiency are also the key limiting factors in
mainly reflecting in PSeries, PEmission and PNRR-J. PSeries ¼ J2·Rse. So for
improving the efficiency of a solar cell. The series resistance will
the photovoltaic system having high photocurrent density (e.g.,
cause a significant energy loss when the photovoltaic system has a
high-concentration-ratio photovoltaic system), reducing the series
high photocurrent density. For the photovoltaic system working at
resistance is very necessary. According to Eq. (11) and (12), a high
a high temperature, the external radiative efficiency needs to be
series resistance will lead to a high Jemission, resulting in the high
enhanced to reduce the significant non-radiative recombination
emission loss (ERE ¼ 1) and NRR-J loss (EREs1), especially when
loss.
the cell operates at a high operating temperature. In our calcula-
tion, the NRR-J loss can reach up to 17% for the c-Si CPV system
considering the temperature variation (Rse ¼ 0.5 U cm2, 5. Conclusions
Rsh ¼ 500 U cm2 and n ¼ 5), and the loss will be much higher if Rse is
increased. Fig. 6 indicates that, for a lower ERE, the efficiency of This paper has systematically studied different loss processes of
solar cells drops a lot compared with cells in radiative limit solar cells considering the temperature rise, including intrinsic and
(ERE ¼ 1) at room temperature. Taking the heat generation into extrinsic losses. The heat generation due to the loss processes re-
consideration, the operating temperature of the cells with lower sults in a significant temperature rise about 100 K, which will be
ERE grows much higher than that of the cells in radiative limit, higher for CPV solar cells, further exacerbating the energy loss and
greatly reducing the output efficiency. Thus being infinitely close to reducing the efficiency of the cells by 5e15%. According to the
the radiative limit is what we should be working for. priority to increase the efficiency of a cell, loss processes are listed
The temperature coefficients (TC) of solar cells with different here to be addressed: (1) the below Eg loss; (2) the thermalization
external radiative efficiencies (ERE ¼ 1, 102 and 105) are calcu- loss; (3) the angle mismatch loss; (4) the non-radiative recombi-
lated by Eq. (23) [26], based on the final thermal equilibrium nation loss; (5) the series loss. Among the loss processes, the below
states, and plotted in Fig. 7. Solar cells in radiative limit (ERE ¼ 1) Eg loss and the thermalization loss play dominant roles in energy
have the biggest temperature coefficients, and solar cells with loss processes. These two kinds of loss processes are unavoidable in
lower ERE have smaller temperature coefficients. For a commer- traditional single bandgap solar cells for the mismatch between the
cial c-Si solar cell, ERE is about 105 [14], and in our calculation, broad incident solar spectrum and the single-bandgap absorption
the temperature coefficient of a c-Si solar cell with ERE ¼ 105 is of a cell [10,12]. To increase the number of available photons and
about 4.3‰ in Fig. 8, which agrees well with the temperature enhance the available energy of per absorbed photon, the effective
coefficient of commercial c-Si solar cells (4‰e5‰). In addi- way is to eliminate the mismatch between the broad incident solar
tion, solar cells of larger bandgap have bigger temperature co- spectrum and the single-bandgap absorption. The technology of
efficients, indicated in Fig. 8. multiple energy gap solar cells is considered as a promising solu-
tion. The angle mismatch loss is considered as the third important
loss process. Increasing the absorption angle is a commonly used
method to suppress this loss process. Non-radiative recombination
loss and series loss are extremely significant for the high-
concentration-ratio photovoltaic system, covering 15%e40% of the
total incident solar energy for the cells with bandgap below 2.0 eV
in the case of 100 suns. The series resistance needs to be decreased
to suppress the series loss. Considering that, the non-radiative
recombination process mainly consists of surface recombination,
Auger recombination and Shockley-Read-Hall (SRH) recombination
processes, the key to reduce the non-radiative recombination loss is
to suppress these three recombination processes. SRH recombina-
tion is negligible compared with the other two recombinations, and
it can be reduced by careful growth of crystal [23,27]. Auger
recombination is exacerbated with the increase of doping con-
centration in solar cells [28,29]. So it is an effective way to suppress
the Auger recombination by controlling the doping concentration.
As for the surface recombination, it is associated with the surface
area of a cell. For a cell with high surface area, it is particularly
significant. To suppress the surface recombination, surface passiv-
ation is a feasible method, e.g., the tetramethylammonium hy-
Fig. 8. Temperature coefficient of solar cells with different external radiative effi- droxide (TMAH) etch [30].
ciencies (ERE ¼ 1, 102 and 105). In addition to the loss processes, the effects of different
500 A. Wang, Y. Xuan / Energy 144 (2018) 490e500

parameters on the performance of the cell are studied. Among the Energy Technol Int Soc Optics Photon 2010;7772:777211.
[8] Singh GK. Solar power generation by PV (photovoltaic) technology: a review.
parameters, operating temperature, series resistance and external
Energy 2013;53:1e13.
radiative efficiency play very important roles. The efficiency of [9] Da Y, Xuan Y, Li Q. From light trapping to solar energy utilization: a novel
high-concentration-ratio solar cells (e.g., 100suns) will be lower photovoltaicethermoelectric hybrid system to fully utilize solar spectrum.
than 10% if these parameters are not well optimized while the ef- Energy 2016;95:200e10.
[10] Hirst LC, Ekins-Daukes NJ. Fundamental losses in solar cells. Prog Photovolt
ficiency of non-CPV solar cells with the same parameters is higher 2011;19(3):286e93.
than twice of that. Without the effective cooling system, the effi- [11] Dupre  O. Physics of the thermal behavior of photovoltaic devices. 2015.
[12] Dupre  O, Vaillon R, Green MA. A full thermal model for photovoltaic devices.
ciency of a cell drops about 5% compared with the efficiency at
Sol Energy 2016;140:73e82.
room temperature. For the high-concentration-ratio solar cells, it [13] Kaushika N, Rai A. An investigation of mismatch losses in solar photovoltaic
drops even more. An effective cooling system must be applied to cell networks. Energy 2007;32(5):755e9.
reduce the operating temperature of the cell. A novel cooling [14] Dupre  O, Vaillon R, Green M. Thermal behavior of photovoltaic devices.
Springer International Publishing; 2017.
method, passive radiative cooling strategy, has been proved a [15] Shang A, Li X. Photovoltaic devices: opto-electro-thermal physics and
promising way to cool the solar cell and improve the efficiency modeling. Adv Mater 2017;29(8).
[31,32]. For a CPV system, the series loss due to series resistance is [16] Liu SH, Simburger EJ, Matsumoto J, Iii AG, Ross J, Nocerino J. Evaluation of
thin-film solar cell temperature coefficients for space applications. Prog
exacerbated remarkably. It dissipates about 10% of the incident Photovolt Res Appl 2005;13(2):149e56.
solar energy in the case of 100 suns. Thus, sufficient efforts must be [17] Pavanello D, Virtuani A, Friesen G. Overview of temperature coefficients of
concentrated on reducing the series resistance during the design different thin film photovoltaic technologies. In: Proc. 25th EU PVSEC; 2010.
p. 4248e52.
and fabrication of solar cells. Moreover, the external radiative effi-
[18] Koehl M, Heck M, Wiesmeier S, Wirth J. Modeling of the nominal operating
ciency is a very important parameter of a solar cell, for a high ERE cell temperature based on outdoor weathering. Sol Energy Mater Sol Cell
indicates a low internal recombination loss. Compared with a cell 2011;95(7):1638e46.
that has a low ERE (105), the efficiency of a cell in radiative limit [19] Kim N, Kim D, Kang H, Park Y-G. Improved heat dissipation in a crystalline
silicon PV module for better performance by using a highly thermal con-
(ERE ¼ 1) can be improved by 10e20%. One of the ways to improve ducting backsheet. Energy 2016;113:515e20.
the ERE is to suppress the non-radiative recombination, which has [20] Noro M, Lazzarin RM. Solar cooling between thermal and photovoltaic: an
been discussed above. Another way is to recycle the photons energy and economic comparative study in the Mediterranean conditions.
Energy 2014;73:453e64.
emitted by recombination in the cell. For example, textured surface [21] Rau U. Reciprocity relation between photovoltaic quantum efficiency and
design has been demonstrated an effective strategy for photon electroluminescent emission of solar cells. Phys Rev B 2007;76(8).
recycling [27,29]. [22] Green MA. Radiative efficiency of state-of-the-art photovoltaic cells. Prog
Photovolt Res Appl 2012;20(4):472e6.
We hope this study may help researchers design high-efficiency [23] Ciszek T. Controlled samples for silicon defect and impurity studies. Golden,
solar cells. And for the following researchers interested in loss CO (United States): National Renewable Energy Lab; 1995. p. 42e5.
processes of solar cells, we hope this paper could be a compre- [24] Warmann EC, Flowers C, Lloyd J, Eisler CN, Escarra MD, Atwater HA. Design of
photovoltaics for modules with 50% efficiency. Energy Sci Eng 2017;5(2):
hensive and detailed reference. 69e80.
[25] Geisz JF, Steiner MA, García I, Kurtz SR, Friedman DJ. Enhanced external
Acknowledgments radiative efficiency for 20.8% efficient single-junction GaInP solar cells. Appl
Phys Lett 2013;103(4):041118.
[26] Emery K, Burdick J, Caiyem Y, Dunlavy D, Field H, Kroposki B, et al. Temper-
This work has been supported by the National Natural Science ature dependence of photovoltaic cells, modules and systems. In: Photovoltaic
Foundation of China (Grant No. 51336003). Specialists Conference; 1996. p. 1275e8.
[27] Miller OD, Yablonovitch E, Kurtz SR. Strong internal and external lumines-
cence as solar cells approach the ShockleyeQueisser limit. IEEE J Photovolt
References 2012;2(3):303e11.
[28] Kerr MJ, Cuevas A. General parameterization of Auger recombination in
[1] Shockley W, Queisser HJ. Detailed balance limit of efficiency of p-n junction crystalline silicon. J Appl Phys 2002;91(4):2473e80.
solar cells. J Appl Phys 1961;32(3):510e9. [29] Karakasoglu I, Wang KX, Fan S. Optical-electronic analysis of the intrinsic
[2] Henry CH. Limiting efficiencies of ideal single and multiple energy gap behaviors of nanostructured ultrathin crystalline silicon solar cells. ACS
terrestrial solar cells. J Appl Phys 1980;51(8):4494e500. Photon 2015;2(7):883e9.
[3] Wurfel P, Ruppel W. Upper limit of thermophotovoltaic solar-energy con- [30] Oh J, Yuan H-C, Branz HM. An 18.2%-efficient black-silicon solar cell achieved
version. IEEE Trans Electron Dev 1980;27(4):745e50. through control of carrier recombination in nanostructures. Nat Nanotechnol
[4] Araújo GL, Martí A. Absolute limiting efficiencies for photovoltaic energy 2012;7(11):743e8.
conversion. Sol Energy Mater Sol Cell 1994;33(2):213e40. [31] Zhu L, Raman A, Wang KX, Anoma MA, Fan S. Radiative cooling of solar cells.
[5] Wurfel P. Thermodynamic limitations to solar energy conversion. Physica E Optica 2014;1(1):32e8.
2002;14(1e2):18e26. [32] Raman AP, Anoma MA, Zhu L, Rephaeli E, Fan S. Passive radiative cooling
[6] Markvart T. Thermodynamics of losses in photovoltaic conversion. Appl Phys below ambient air temperature under direct sunlight. Nature
Lett 2007;91(6):064102. 2014;515(7528):540e4.
[7] Tsakalakos L, Hirst LC, Ekins-Daukes NJ. Quantifying intrinsic loss mechanisms
in solar cells: why is power efficiency fundamentally limited? SPIE Solar

You might also like