Sie sind auf Seite 1von 7

Evaluation of methods to extract parameters from currentvoltage

characteristics of solar cells


Yuan Li
a
, Wenxiao Huang
a
, Huihui Huang
b
, Corey Hewitt
a
, Yonghua Chen
a
,
Guojia Fang
b
, David L. Carroll
a,
a
Center for Nanotechnology and Molecular Materials, Department of Physics, Wake Forest University, Winston-Salem, NC 27109, USA
b
Department of Electronic Science & Technology, School of Physics & Technology, Wuhan University, Wuhan, Hubei 430072, PR China
Received 7 September 2012; received in revised form 28 November 2012; accepted 5 December 2012
Available online 4 February 2013
Communicated by: Associate Editor Sam-Shajin Sun
Abstract
An evaluation of methods is represented to extract parameters from the currentvoltage (IV) characteristics of solar cells. This eval-
uation includes ve aspects: applicability, convergence, stability, calculation speed, and error on various types of IV data. Six current
popular extraction methods are discussed and evaluated in this work. We nd that the widely adopted method using the reciprocals of
slope at two points is the fastest and simplest approach to estimating resistances, but it is an incomplete method to obtain the other
parameters. Of the ve complete methods, the IshibashiKimura method shows the best accuracy (4.5% average error) while having
bad applicability. Generally speaking, the Lambert W-function method is the best comprehensive method with good accuracy, applica-
bility, and convergence, though the calculation speed is relatively lower than the other four methods.
2012 Elsevier Ltd. All rights reserved.
Keywords: Solar cell; Currentvoltage; Extract parameters; Equivalent circuit; Evaluation
1. Introduction
In recent years, the solar cell has drawn much research
interest from many institutes and companies all over the
world. The currentvoltage (IV) of illuminated solar cells
is generally characterized using the equivalent circuit
(Fig. 1) (Li et al., 2011), which is the most important exper-
imental measurement for showing the performance of a
solar cell. The IV curve is not only useful for cell arrays
and system simulation, but also as an analysis tool to gain
an understanding of the internal physical mechanisms of
the solar cell (Li et al., 2012, 2009; Brabec et al., 2001; Fra-
as et al., 1991; Nie et al., 2012; Mingjun et al., 2011). In
Fig. 1 the equivalent circuit of a planar cell is described
by a current source and a diode, with two resistances, in
series and parallel. The resistance R
p
in parallel with the
diode represents the shunt resistance that can occur in real
solar cells across the surfaces, at pin holes in the pn junc-
tion, or at grain boundaries. The series resistance R
s
accounts for all voltage drops across the transport resis-
tances of the solar cell and its connections to a load (Wu rfel
and Wu rfel, 2009). For conventional solar cells, the shunt
resistance R
p
comes from recombination of charge carriers
near the dissociation site (e.g. donor/acceptor interface)
and it may also include recombination farther away from
the dissociation site (e.g. near the electrode). The series
resistance considers the conductivity, i.e. mobility, of spe-
cic charge carriers in the respective transport medium,
where the mobility is aected by space charges and traps
or other barriers (hopping) (Jain and Kapoor, 2005a). Its
IV characteristics could be expressed by the generalized
Shockley equation (Bube and Fahrenbruch, 1981).
0038-092X/$ - see front matter 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.solener.2012.12.005

Corresponding author. Tel.: +1 (336) 727 1806.


E-mail address: carroldl@wfu.edu (D.L. Carroll).
www.elsevier.com/locate/solener
Available online at www.sciencedirect.com
Solar Energy 90 (2013) 5157
I I
ph
I
s
exp
qV R
s
I
n
D
k
B
T
_ _
1
_ _

V R
s
I
R
p
1
where n
D
, k
B
and T are the diode ideality factor, Boltz-
mann constant, and temperature, respectively. I
s
is the re-
verse saturation current of the diode (Fahrenbruch and
Aranovich, 1979; Rand et al., 2007). I
ph
is the photocurrent
from the illumination of the sun, which could be calculated
by a transfer matrix method and nite element method (Li
et al., 2012; Koster et al., 2005).
Many methods have been proposed to extract these
parameters from IV data (Jain and Kapoor, 2005a;
Ortiz-Conde et al., 2006; Nehaoua et al., 2009; Ouennoughi
and Chegaar, 1999; Li et al., 2005; Fred A, 1966; Ishibashi
et al., 2008; Jain and Kapoor, 2005b; Bouzidi et al., 2007),
with various levels of success of the calculation. In this
work, an evaluation will be given in to review these meth-
ods of extracting parameters. Six popular methods are
evaluated in terms of their applicability, convergence, sta-
bility, calculation speed, and error.
2. Parameter extraction methods
The fastest estimation method for series resistance and
parallel resistance is to use the reciprocals of slope of the
output curve under dark conditions at 0 V and 2 V (a shunt
resistance point of 2 times the open circuit voltage (V
oc
)
may be used as well) to nd the shunt and series resistances,
respectively, as shown in Eqs. (2) and (3) (Li et al., 2005;
Fred A, 1966). It is based on the assumption that the series
resistance is small and the shunt resistance is large, i.e. the
solar cell has reasonably good properties. Furthermore, the
calculation of R
s
is only valid if a good voltage point is
chosen for the slope calculation. In the study of organic
solar cells, especially with new materials and varying solar
cell architectures, the assumptions of low R
s
and high R
p
are not strictly valid (Williams, 2013).
R
p

dV
dI

V0
2
R
s

dV
dI

V2V oc or 2V
3
To further extract I
s
, n
D
, and J
ph
, some complicated
methods were proposed in the past years. A simple manner
in which to extract full parameters is through the use of a
nonlinear least squared error t to Eq. (1). Unfortunately,
since Eq. (1) is a transcendental equation, it is very dicult
to nd the numerical solutions. In order to circumvent this
problem, it is convenient to use the Lambert W function
described in Jain and Kapoor (2005a, 2005b). Eq. (1) is
then written as:
I
V
R
s
R
p

n
D
V
th
R
s
Lambert W
R
s
I
s
R
p
RpVRsIsRsI
ph

n
D
V
th
R
s
R
p

_ _
n
D
V
th
R
s
R
p

_
_
_
_

R
p
I
s
I
ph

R
s
R
p
4
where the Lambert W function assumes I
ph
is approxi-
mately equal to short current I
sc
, and V
th
= k
B
T/q is the
thermal voltage. Based on Eq. (4), a simple and accurate
(without approximations) method using the Lambert W
function is introduced by Jain and Kapoor (2005b).
Assuming R
p
as the slope in Eq. (2), the unknown param-
eters R
s
, n
D
and I
s
are found by an iterative search method
(described in the Supplementary Material). Unfortunately,
this method is computationally intensive and susceptible to
divergence problems as well as local minima convergence
issues. Furthermore, the requirement of initial guesses in-
fers that the researcher has some knowledge regarding
the device characteristics prior to analysis, which may not
always be true (Williams, 2013).
In 2008, the IshibashiKimura method gave another
iterative method to calculate the IV array in organic photo-
voltaics (OPV) (Ishibashi et al., 2008). It connects R
s
and
n
D
with all other parameters, and R
s
and n
D
are found by
tting the data in terms of Eq. (5). However, since Eq.
(5) includes dV/dI, adjacent current data points cannot
be the same (this will lead to innity); therefore, this
method is not suited to some IV data with very high ll
factors (FF). In addition, some bad data with high S shape
curve (namely low FF) is hard to converge to a stable solu-
tion. Additionally, Bouzidi and Chegaar reported a
method by collecting the non-exponential current terms
that can deal with bad data including electrical noise
or random errors. The solution is obtained based on the
Newton functional iteration procedure (Bouzidi et al.,
2007).

dV
dI

n
D
k
B
T=q
I
sc
I fV R
s0
I
sc
I n
D
k
B
T=qg=R
p
R
s
5
I
ph
R
s
R
p
I
s V
I
Fig. 1. Equivalent circuit of a conventional planar solar cell. I
ph
, I
s
, R
s
,
and R
p
are photocurrent source, reverse saturation current, series
resistance and parallel resistance, respectively. I and V are the output
current and voltage of the cell, respectively.
52 Y. Li et al. / Solar Energy 90 (2013) 5157
Apart from these iterative methods, there are some fast
calculation methods that can directly extract parameters
from IV arrays by numerical tting. For example, Neh-
aoua proposed a simplied equation based on the assump-
tions that R
p
R
s
and V R
s
I k
B
T that can solve the
series resistance and diode ideality factor of organic solar
cells, which involves a linear regression on the set of data
derived from subsets of the initial IV output array (Neh-
aoua et al., 2009). Another method, developed by Ortiz-
Conde et al. (2006), introduces an auxiliary function to cal-
culate the parameters just from the coecients from tting
the auxiliary function. (These six methods are introduced
in detail in the Supplementary Material.)
3. Evaluation
A good parameters extraction method should be appli-
cable to various IV curves. As mentioned above, some
methods are not applicable to certain data. For instance,
method 2 cannot calculate the discrete date having high
FF, i.e. dV/dI ! Innity at V = 0. We chose six materials
IV curves from Ghozati et al. (1998); NREL (2011);
Todorov et al. (2010); Nie et al. (2011); Riede et al.
(2011); Roensch et al. (2011) to evaluate these extraction
methods. These materials include silicon (Ghozati et al.,
1998), GaInP/GaInAs/Ge (Roensch et al., 2011), CIGS
(NREL, 2011), CZTS (Todorov et al., 2010), polymer
(Nie et al., 2011), and small molecule (Riede et al., 2011).
Some IV curves showed standard diode characteristics,
such as Si, but some have a lower FF, such as the polymer,
which are used to check the applicability of the six extrac-
tion methods.
Apart from applicability, there are another four aspects
used to evaluate these methods, as shown in Fig. 2; the
speed of computation on parameters extraction, conver-
gence of extraction process, the comparison between origi-
nal IV curve and resulting IV from extracted parameters,
and the comparison between the extracted parameters and
the extracted parameters of the next iteration. The key step
in the evaluation is to generate IV arrays in terms of
extracted parameters. It could be obtained directly through
solving Eq. (1) after substituting the values of these param-
eters, at a given certain V. However, this method is very
slow and it is not feasible for all cases. Therefore, the Lam-
bert W function (Jain and Kapoor, 2005a, 2005b) is used
instead to produce IV arrays faster and easier. As shown
below, set I is generated from Eq. (5) as a function of set V.
Two comparisons are mentioned in Fig. 2. The parame-
ter error is the dierence between two extracted parameters
that are calculated from successive rounds, which is to indi-
cate the stability of the results. It is represented by a nor-
malized error (e
norm
) calculated from 3 parameter errors:
I
sc
error (e
jsc
), V
oc
error (e
voc
), and FF error (e

), as shown
in Eq. (6). e
norm
does not include the error of eciency since
eciency is the product of I
sc
, V
oc
and FF. Another com-
parison, curve error e
area
, is used to measure the accuracy
of extraction methods. As shown in Fig. 3, it is dened
as the area dierence (between the original IV curve and
the reproduced IV curve from extracted parameters) over
the area of the original IV array in the fourth quadrant.
e
norm
e
2
Isc
e
2
voc
e
2
FF

1=2
6
Fig. 2. Schematic diagram of evaluating these six parameter extraction methods. Method X means one of the six methods.
Y. Li et al. / Solar Energy 90 (2013) 5157 53
We use 1 and 0 to represent the extracting speed of
fast and slow, respectively. The fast method refers to
nishing the calculation in one second, and the slow
method usually takes more than 10 s or more typically,
ten minutes to obtain the result, which depends on the I
V array. Generally speaking, slow methods include some
iterative equations so that they need to spend more time
on nding a stable solution. However, the slow methods
usually have better accuracy, stability, and convergence.
Here, convergence is reached when the curve dierence is
equal to zero when n approaches innity (Eq. (7)).
lim
n!1
CurveArea
n
CurveArea
n1
0 7
Finally, an objective function (Eq. (8)) is given to com-
prehensively rank dierent methods, where the ranking
value R is ranged from 0 to 1 indicating worst to best.
Although all ve aspects mentioned above are very useful
to review an extraction method, they have dierent weights
on the study of solar cells. Since the calculation speed
depends on the CPU and they all could be nished in a
few minutes, it is not as important as the two evaluation
errors (e
norm
and e
area
) and application (N
applicable
). It is
therefore combined with the number of convergence (N
con-
vergence
) over two, as shown in the rst term in Eq. (8). Con-
sequently, the ideal value of R should be 1 if the speed is
fast, it is convergent and applicable to all data, and the
evaluation error is zero.
R V
speed
N
convergence
=2 N
applicable
expe
norm
=100 expe
area
=100=4 8
4. Result and discussion
The parameters extracted by the six methods and corre-
sponding evaluations are listed in Table 1. Methods 16
represent Reciprocals of slope at two points, Ishibashi
Kimura method, Lambert W-function method, Bouzidi
Chegaar method, Nehaoua method, and Ortiz-Conde
method, respectively. Method 1 cannot calculate the
parameters n
D
, I
s
, and I
ph
since this is not a complete
method. It therefore has no review of convergence and
errors. This method is used to rapidly and roughly estimate
the values of R
s
and R
p
, and to obtain the initial values for
some iterative methods, such as method 2. As introduced,
method 2 is incapable of dealing with data that has the
same current values at two neighboring voltages, which
mostly means high FF, since it will lead to dI/dV = Innity
and make the kernel Eq. (5) invalid. Although there is only
one IV curve (polymer) applicable to method 2, it shows
an excellent performance on all errors (lower than 4.5%)
and convergence among all six methods, which is consis-
tent with the report from the OPVAP system (Li, 2010).
Method 3 and 4 are another two iterative methods having
good accuracy and applicability, but still having slow cal-
culation speed due to iteration. Especially method 3, keep-
ing the second accuracy and applicability, shows a
comprehensive performance overwhelming method 4 and
method 5 for all aspects. Although methods 5 and 6 have
decent computation speed and they show fair performance
on some special IV arrays (Ortiz-Conde et al., 2006; Neh-
aoua et al., 2009), their errors are not acceptable for the I
V characteristics used in this study.
Generally speaking, method 2 is the best choice to
extract the parameters if the IV array is valid, i.e. dI/
dV 0 for all points. Usually, the IV data of organic
solar cells is appropriate for it since their FFs are not
too high to show an ideal diode curve (dI/dV = 0 before
threshold voltage). For some crystal solar cells with a good
diode curve, method 3 is able to nd an acceptable and
convergent result. As a fast and simple calculation, method
1 might be used to estimate R
p
and R
s
for some comparison
study. Furthermore, using the objective function in Eq. (8),
ve completed methods are ranked in the last column in
Table 1. The method 3 (Lambert W-function method),
method 2 (IshibashiKimura method), and method 6
(Ortiz-Conde method) get the top 3 scores of 0.75, 0.64,
and 0.57, respectively.
As discussed above, each parameter in the equivalent
circuit is an aggregate involving many intrinsic physical
Fig. 3. Denition of curve error.
54 Y. Li et al. / Solar Energy 90 (2013) 5157
Table 1
Extraction results and evaluation comparison. Method 16 represent reciprocals of slope at two points, IshibashiKimura method, Lambert W-function method, BouzidiChegaar method, Nehaoua
method and Ortiz-Conde method, respectively. Inf, ID and NA means Innity, Indeterminate, and Not Applicable respectively. Number in IV column indicates the IV curve from the publication
(Ghozati et al., 1998; NREL, 2011; Todorov et al., 2010; Nie et al., 2011; Riede et al., 2011; Roensch et al., 2011).
Method IV R
p
(X) R
s
(X) n
D
I
0
(A) I
ph
(A) Speed value N
Convergence
(%) Avere
norm
Avere
area
(%) N
Applicable
Rank
1 1 Inf 1.5 NA NA ID 1 0 NA NA 0 NA
2 36.5 5.1 NA NA ID
3 63 14.3 NA NA ID
4 296.7 13.3 NA NA ID
5 32.3 1.9 NA NA ID
6 88.7 51.9 NA NA ID
2 1 Inf ID ID ID ID 0 1/1 4.5 4.4 1/6 0.64
2 36.5 ID ID ID ID
3 63 ID ID ID ID
4 296.7 3.2 5.6 2.49E-05 0.013
5 32.3 ID ID ID ID
6 88.7 ID ID ID ID
3 1 Inf 10 2 1.00E-10 0.031 0 5/5 21.6 15.2 5/6 0.75
2 1.42E + 14 1.8 1.6 7.17E-08 0.028
3 155.9 9 4.8 1.03E-11 0.017
4 341.6 9.8 3.6 6.43E-07 0.014
5 139.6 0.5 2.4 8.02E-07 0.032
6 1967.4 18.7 6.5 6.71E-07 0.012
4 1 NA NA NA NA NA 0 1/5 55.3 40.6 5/6 0.54
2 1.42E + 14 14.7 0.3 2.64E-38 0.028
3 128.8 27 32.4 8.21E-04 0.021
4 334.5 7.1 9.8 3.91E-04 0.014
5 188.9 80 14.2 3.84E-03 0.024
6 1956.8 12 11.8 5.57E-05 0.012
5 1 8.35E + 14 NA NA NA 0.031 1 1/4 1118.1 161.6 2/3 0.37
2 5.39E + 14 NA NA NA 0.028
3 741.8 2953.2 201.5 4.22E-06 0.017
4 315.8 44.5 3.9 1.92E-03 0.014
5 58.7 618.6 96.7 1.18E-04 0.032
6 5016.9 37747 1463.5 4.07E-07 0.012
6 1 7355.5 1.33E-03 1.1 4.43E-14 0.031 1 6/6 253.1 169.1 6/6 0.57
2 254.1 0.2 1.9 5.35E-11 0.029
3 804315.7 9.11E-08 1.7 2.30E-48 0.017
4 368.8 6.7 2 8.71E-18 0.014
5 9168 1.59E-03 0.4 1.70E-13 0.032
6 1670.7 1.03E + 00 23.7 1.43E-29 0.012
Y
.
L
i
e
t
a
l
.
/
S
o
l
a
r
E
n
e
r
g
y
9
0
(
2
0
1
3
)
5
1

5
7
5
5
concepts of thin lms. For instance, R
s
accounts for all
voltage drops across the transport resistances of the solar
cell and its connections to a load, and even relates to mor-
phology. Therefore, there is no exact value for each param-
eter that can fully cover these many micro-phenomenon
just from a simplied model of solar cells, which can
explain why the same parameter calculated by the six meth-
ods (Table 1) varies so much. Moreover, the precision of
the measurement is another factor eecting the calculation,
especially for the methods based on the component of dV/
dI. Consequently, it should be more meaningful to use an
identical method to make the comparison of parameters
of dierent solar cells measured at the same conditions.
5. Summary
IV characteristics are described by the equivalent cir-
cuit consisting of several parameters, and each parameter
is a united value quantifying many internal physical mech-
anisms, such as the series resistance R
s
that is related to the
charge carrier mobility and interface energy barrier. This
work rst reviewed six current popular approaches to
extract parameters from IV characteristics, and then gave
an evaluation to test their applicability, convergence, sta-
bility, calculation speed, and error. It turns out that
although reciprocals of slope at two points is an incomplete
method, it is a very fast and simple way to calculate the val-
ues of R
s
and R
p
, which are widely used to roughly estimate
the resistances. Generally speaking, the IshibashiKimura
method shows the best accuracy (4.5% average error) but
has bad applicability (1/6), which is only applied to some
IV curves with relatively lower FF, such as organic solar
cells. The Lambert W-function method is the best compre-
hensive method having good accuracy, applicability and
convergence, though the calculation speed is slower than
the other three methods.
It should be emphasized that there are still many other
extraction methods having various features that are not
mentioned in this paper. However, through the evaluation
given above, these un-reviewed methods could be further
tested and compared with each other in the later work.
Appendix A. Supplementary material
Supplementary data associated with this article can be
found, in the online version, at http://dx.doi.org/10.1016/
j.solener.2012.12.005.
References
Bouzidi, K., Chegaar, M., Bouhemadou, A., 2007. Solar cells parameters
evaluation considering the series and shunt resistance. Solar Energy
Materials and Solar Cells 91, 16471651.
Brabec, C.J., Cravino, A., Meissner, D., Sariciftci, N.S., Fromherz, T.,
Rispens, M.T., Sanchez, L., Hummelen, J.C., 2001. Origin of the open
circuit voltage of plastic solar cells. Advanced Functional Materials 11,
374380.
Bube, R.H., Fahrenbruch, A.L., 1981. Advances in Electronics and
Electron Physics. Academic, New York, p. 163.
Fahrenbruch, A.L., Aranovich, J., 1979. Solar Energy Conversion.
Springer-Verlag, New York, pp. 257.
Fraas, L.M., Avery, J.E., Gruenbaum, P.E., Ballantyne, R.J., Malocsay,
E., 1991. Tandem gallium solar-cell voltage-matched circuit perfor-
mance projections. Solar Cells 30, 355361.
Fred A, S., 1966. The history, design, fabrication and performance of
CdS thin lm solar cells. Advanced Energy Conversion 6, 201
221.
Ghozati, S.B., Ebong, A.U., Honsberg, C.B., Wenham, S.R., 1998.
Improved ll-factor for the double-sided buried-contact bifacial
silicon solar cell. Solar Energy Materials and Solar Cells 51, 121
128.
Ishibashi, K.I., Kimura, Y., Niwano, M., 2008. An extensively valid and
stable method for derivation of all parameters of a solar cell from a
single currentvoltage characteristic. Journal of Applied Physics 103.
Jain, A., Kapoor, A., 2005a. A new approach to study organic solar cell
using Lambert W-function. Solar Energy Materials and Solar Cells 86,
197205.
Jain, A., Kapoor, A., 2005b. A new method to determine the diode ideality
factor of real solar cell using Lambert W-function. Solar Energy
Materials and Solar Cells 85, 391396.
Koster, L.J.A., Smits, E.C.P., Mihailetchi, V.D., Blom, P.W.M., 2005.
Device model for the operation of polymer/fullerene bulk heterojunc-
tion solar cells. Physical Review B 72.
Li, G., Shrotriya, V., Huang, J.S., Yao, Y., Moriarty, T., Emery, K.,
Yang, Y., 2005. High-eciency solution processable polymer photo-
voltaic cells by self-organization of polymer blends. Natural Material
4, 864868.
Li, Y., Zhou, W., Xue, D., Liu, J.W., Peterson, E.D., Nie, W.Y., Carroll,
D.L., 2009. Origins of performance in ber-based organic photovol-
taics. Applied Physics Letters 95.
Li, Y., Wang, M., Huang, H., Nie, W., Li, Q., Peterson, E.D., Con, R.,
Fang, G., Carroll, D.L., 2011. Inuence on open-circuit voltage by
optical heterogeneity in three-dimensional organic photovoltaics.
Physical Review B 84, 085206.
Li, Y., Huang, H., Wang, M., Nie, W., Huang, W., Fang, G., Carroll,
D.L., 2012. Spectral response of ber-based organic photovoltaics.
Solar Energy Materials and Solar Cells 98, 273276.
Li, Yuan. Open Photovoltaics Analysis Platform (OPVAP). USA 2010.
<http://www.OPVAP.com>.
Mingjun, W., Yuan, L., Huihui, H., Eric, D.P., Wanyi, N., Wei, Z., Wei,
Z., Wenxiao, H., Guojia, F., Nanhai, S., Xingzhong, Z., David, L.C.,
2011. Thickness dependence of the MoO3 blocking layers on ZnO
nanorod-inverted organic photovoltaic devices. Applied Physics Let-
ters 98, 103305.
Nehaoua, N., Chergui, Y., Mekki, D.E., 2009. Determination of organic
solar cell parameters based on single or multiple pin structures.
Vacuum 84, 326329.
Nie, W., MacNeill, C.M., Li, Y., Noftle, R.E., Carroll, D.L., Con,
R.C., 2011. Macromolecular Rapid Communications 32 (15), 1163
1168.
Nie, W., Li, Y., Zhou, W., Liu, J., Carroll, D.L., 2012. Multi-layer
deposition of conformal, transparent, conducting oxide lms for device
applications. Thin Solid Films 520, 40084015.
NREL, 2011. Manufacturing-friendly advance seen in CIGS solar cell
Processing. NREL Highlights. NREL/FS-5200-51412.
Ortiz-Conde, A., Sanchez, F.J.G., Muci, J., 2006. New method to extract
the model parameters of solar cells from the explicit analytic solutions
of their illuminated IV characteristics. Solar Energy Materials and
Solar Cells 90, 352361.
Ouennoughi, Z., Chegaar, M., 1999. A simpler method for extracting solar
cell parameters using the conductance method. Solid State Electronics
43, 19851988.
Rand, B.P., Burk, D.P., Forrest, S.R., 2007. Oset energies at organic
semiconductor heterojunctions and their inuence on the open-circuit
voltage of thin-lm solar cells. Physical Review B 75, 115327.
56 Y. Li et al. / Solar Energy 90 (2013) 5157
Riede, M., Uhrich, C., Widmer, J., Timmreck, R., Wynands, D.,
Schwartz, G., Gnehr, W.M., Hildebrandt, D., Weiss, A., Hwang, J.,
Sundarraj, S., Erk, P., Pfeier, M., Leo, K., 2011. Ecient organic
tandem solar cells based on small molecules. Advanced Functional
Materials 21, 30193028.
Roensch, S., Hoheisel, R., Dimroth, F., Bett, A.W., 2011. Subcell I
V characteristic analysis of GaInP/GaInAs/Ge solar cells using
electroluminescence measurements. Applied Physics Letters 98,
251113.
Todorov, T.K., Reuter, K.B., Mitzi, D.B., 2010. High-eciency solar cell
with earth-abundant liquid-processed absorber. Advanced Materials
22, E156+.
Williams, G. Advances in Organic Photovoltaics and Methods for
Eective Solar Cell Parameter Extraction 2013.<http://www.eng.uwa-
terloo.ca/~g3willia/documents/ECE730_OSCReview.pdf>.
Wu rfel, P., Wu rfel, U. Physics of Solar Cells: from Basic Principles to
Advanced Concepts, 2nd, updated and expanded ed., Wiley-VCH,
Weinheim, 2009.
Y. Li et al. / Solar Energy 90 (2013) 5157 57

Das könnte Ihnen auch gefallen