Sunteți pe pagina 1din 10

1.

INTRODUCTION
In this paper, we shall study the sandwiched SnO
2
(n)/SiO
2
-Si(n) solar
cells (Fig.1), in which thin film SnO
2
has a high electrical conductivity and
transparency at least in the visible region of solar spectrum. The interfacial
203
J. of Active and Passive Electronic Devices, Vol. 1, pp. 203-212 2006 Old City Publishing, Inc.
Reprints available directly from the publisher Published by license under the OCP Science imprint,
Photocopying permitted by license only a member of the Old City Publishing Group
Analytical Modelling and Simulation
of SnO
2
/SiO
2
-Si (n) Solar Cell
A. CHEKNANE
1
, B. BENYOUCEF
1
, J.-P. CHARLES
2
,
N.E. CHABANE-SARI
1
, A.CHAKER
3
AND F. DUJARDIN
4
1
Unit de Recherche Matriaux et Energies Renouvelables,
Universit Abou Bekr Belkaid Tlemcen Algrie
2
LMOPS-CNRS UMR 7132, SUPELEC, 2 rue Edouard Belin, 57070 Metz, France
3
Laboratoire de Physique Energtique, Universit Mentouri de Constantine, Algrie
4
Laboratoire de Physique, Metz, France
Determination of solar cell parameters is a key issue for obtaining a good
design of the device. This paper presents the results of an analytical model-
ling study regarding the effects of interfacial oxide layer thickness (), inter-
face state density
0
(the level above the valence band to which surface
states are filled in isolated semiconductor) on the open circuit voltage, the
series resistance, and the efficiency of SnO
2
/SiO
2
/Si(n) solar cell. For dif-
ferent (), the physical parameters affect directly the solar cell conversion
efficiency (). The obtained results by simulation using Matlab packages
(Simulink and Neural network) are presented and discussed. However, the
learning neural network we proceed is the supervised learning, where
the error minimization between the desired and computed unit values is
done by the least mean square (LMS) convergence.
Keywords: Series resistance, Efficiency, Simulation, Solar Cell, Neural network
*Corresponding Author: E-mail: a.cheknane@mail.lagh-univ.dz, cheknanali@yahoo.com;
Fax: 00213 29 93 26 98
layer in a Schottky barrier solar cell plays an important role in reducing the
dark current, which in turn improves the open-circuit voltage and the efficiency
of the solar cell [1].
The rutile SnO
2
occurs mostly as n-type by oxygen extraction implying the
presence of Sn
2+
with a change of colour; however, because of its lone pair
cation 4s
2
the understoichiometry is low. Sn
2+
does not enjoy a polyhedra co-
ordination without deformation; the fact that SnO
2-
crystallises in the same
structure as the parent oxide with no significant change in the lattice constant
indicate that the Sn-5s
2
is delocalised and the introduced electrons go to the
conduction band. This was confirmed by the absence of EPR signal and
Mossbauer spectra showed one signal ascribed to Sn
4+
. Its refractive index lies
between 1.9 and 2.0 and hence can be used as a low resistance top contact to
the junction and an anti-reflection coating of the active substrate. In addition,
the low cost Sn and Si are by far the most attractive with regard to toxicity
concerns in the environmental issue. Since SnO
2
is highly transparent, it acts as
a window for the transmission of solar radiation falling directly on the active
substrate (Unpublished results).
In the present work, we present the results of an analytical modelling and
simulation of SnO
2
/SiO
2
-Si(n) type solar cells. The effect of interfacial oxide
thickness and interface state density has been studied. The interfacial thickness
has been optimised to 21 for maximum efficiency.
In addition to this, the CELLO technique is used for a precise measurement
of the physical parameters.
2- THEORY
In Schottky junctions when used in solar cells, the space charge electric
field used to separate the hole-electron pairs is generated by the work function
difference between the metal and the semiconductor. Schottky carrier current
versus voltage characteristics are primarily due to majority carrier transport [2].
A subclass of the Schottky barrier type of solar cell is the MOS or metal-
oxide-semiconductor solar cell. Atypical example, utilizing n-type silicon and
SnO
2
as the metal, is shown in Fig.1. The thin film oxide layer (SiO
2
) is
processed carefully to insure that the metal and semiconductor are nowhere in
contact. This processing involves an elaborate annealing sequence to prevent
microcracking of the oxide [3].
The most useful effect of the oxide interfacial layer is the effective increase
in the barrier voltage.
204 CHEKNANE, et al.
The photovoltage (V) developed leads to a net current density (J) if the
output terminals are closed through a certain load. The current density (J)
consists of:
(i) hole tunnelling current from the semiconductor to the metal (J
TH
),
(ii) the electrons tunnelling current from the semiconductor to the metal (J
TE
).
FIGURE 1
Energy band diagram for SnO
2
-SiO
2
-Si(n) solar cell under illumination.
The steady state equivalent circuit of a solar cell under illumination,
coupled to an external load R
L
, is provided in Figure2.
FIGURE 2
The electrical equivalent circuit.
ANALYTICAL MODELLING AND SIMULATION 205
h
h
E
C
E
F
E
V
SnO
2
Si(n)
SiO
2

E
C
E
FM
J
TH
e
-J
TE
I
PH
I
D1
I
D2
I
SH
I
R
SH
R
S
R
L
The solar cell current-voltage characteristic developed under
illumination is given by J.P. Singh and R.S. Srivastava [4] as:
I=I
PH
-I
D1
-I
D2
-I
SH
(1)
where
I
PH
is the photogenerated current,
I
D1
is the current due to majority carriers,
I
D2
is the current due to the minority carriers,
I
SH
is the current through the shunt resistance R
SH
.
The currents I
D1
, I
D2
, and I
SH
may be expressed by the following
equations:
(2)
(3)
(4)
with V
S
=V-R
S
I (5)
R
S
is the series resistance.
(6)
E
g
the semiconductor (Si(n)) energy gap,

Bn
=Schottky barrier height for n-type device (eV).

h
the minority carrier (hole) quasi-Fermi-level from the valence band et
the surface (
h
0.085 eV for N
D
=10
18
cm
-3
)
I
S
and I
0
are saturation and diffusion (holes) currents respectively.
h Bn g
E

SH
S
SH
R
V
I

]
]
]


}
`

.
| +
1
kT
V
exp I I
S
0 2 D

]
]
]


}
`

.
|
1
nkT
V
exp I I
S
S 1 D

206 CHEKNANE, et al.
2. MATLAB/SIMULINK SIMULATION
A Matlab/Simulink program is used for simulating a based SnO
2
/SiO
2
-
Si(n) solar cell by calculation of the different cell parameters at different
values of interfacial oxide layer thickness ().(Fig.3)[5].
FIGURE 3
Matlab/Simulink block.
3. THE CELLO TECHNIQUE
A simplified schematic diagram of CELLO is depicted in Fig.4. The
solar cell is illuminated homogeneously by a set of halogen lamps with
near to AM1.5 intensity. Additionally, an intensity modulated infrared
laser beam focussed to about 100 m diameter is scanning the sample
using piezo-controled mirrors allowing for high resolution maps of the
local linear responds. A home-made, computer controlled, low noise
pot ent i ost at / gal vanost at i s used for vol t age/ current cont rol and
current/voltage measurement with high S/N ratio. A lock-in amplifier,
synchronized to the laser beam modulation signal, is used to measure the
solar cell response to the laser beam perturbation. The CELLO technique
allows constructing the complete current-voltage curve.
The si mul at i on resul t s al l ow t he vari at i ons represent at i on of
efficiency, open circuit voltage, and series resistance of SnO
2
/SiO
2
/Si(n)
solar cell with interfacial oxide layer thickness () are shown in Fig.5,6,7,
respectively, for specific values of interface state density (=0.1 eV) and
for
0
=E
g
/3.
ANALYTICAL MODELLING AND SIMULATION 207
SIMULINK
Block Varying
thickness
V
oc
I
sc
R
s
R
sh
(%)
Current-voltage
characteristic
(CELLO
Technique)
FIGURE 4
Block scheme of the CELLO.
Figure 5
Variations of efficiency versus the interfacial oxide layer thickness ().
208 CHEKNANE, et al.
PC
Laser-control
bias-illum.ctrl
temp.control
Lock-in
laser
Potentiostat-galvanostat
X-Y mirror ctrl
0 5 10 15 20 25
6
8
10
12
14
16
18

(
%
)
(angstrom)

0
=Eg/3

0
=0.1 eV
FIGURE 6
Variations of the open circuit voltage versus the interfacial oxide layer thickness ().
FIGURE 7
Variations of series resistance versus the interfacial oxide layer thickness ().
The efficiency increases with at its initial stages. Beyond a certain ,
the efficiency decreases due to an increase of the series resistance (Fig.7).
ANALYTICAL MODELLING AND SIMULATION 209
R
s
m
(

)
V
o
c
(
V
)
The open-circuit voltage is terminated by the photogenerated current I
PH
and the reverse current as given below:
(7)
With an increase in interfacial layer thickness, resulting in an increased
V
OC
. Later on, the open-circuit voltage becomes saturated, because for
higher values of the photogenerated current also decreases [1].
4. PARAMETERS PREDICTION USING THE NEURALNETWORK
In order to predict the parameters cell values ((%), V
OC
, R
S
), a
perceptron multilayer neural network circuit is used for simulating the
device. The inputs to the model are the interfacial oxide layer thickness
() and
0
(level above valence band to which surface states are filled in
isolated semiconductor (eV).
Step 1: the supervised learning which incorporates an external
teacher, so that each output unit is told what its desired response to input
signals ought to be. During the learning process global information may
be required. Paradigms of supervised learning include error-correction
l earni ng, rei nforcement l earni ng and st ochast i c l earni ng.
An important issue concerning supervised learning is the problem of
error convergence, ie the minimisation of error between the desired and
computed unit values. The aim is to determine a set of weights which
minimises the error. One well-known method, which is common to many
learning paradigms, is the least mean square (LMS) convergence.
Step 2: parameters prediction.
Once the supervised learning is done, the simulation is proceeding as
usual. Thus the omitted data was predicted and the difference between
experimental value and predicted value was determined.
Fig.8 illustrates the predicted values of the open circuit voltage using the
neural network:
210 CHEKNANE, et al.
FIGURE 8
Calculation and prediction of the open circuit voltage versus the interfacial oxide layer thickness ().
In order to compare our method with others (for instance P. Singh & R.S.
Srivastava ) we summarises in table 1 some measured values parameters
and their predicted values using our modelling method.
In this work, the predictive errors were measured by MEF (mean error
function) as
(8)
where e
i
is the experimental value of sample i and p
i
is the predicted value of
sample i, n is the number of the whole samples, e
max
, e
min
are the maximum
and minimum experimental values of whole samples, respectively.
( )
% 100
1
1 min max

n
i
i i
e e
e p
n
MEF
ANALYTICAL MODELLING AND SIMULATION 211
V
o
c
(
V
)
TABLE 1
Predicted and measured values parameters.
5. CONCLUSION
In this work, we have developed an analytical modelling study regarding
the effects of interfacial oxide layer thickness (), interface state density
0
(the level above the valence band to which surface states are filled in
isolated semiconductor) on the open circuit voltage, the series resistance,
and the efficiency of SnO
2
/SiO
2
/Si(n) solar cell; the complete model of the
system has been implemented on Matlab packages (Simulink and Neural
network). Satisfactory agreement between experiment and theory has been
demonstrated.
Using the neural network modelling, we could successfully perform our
calculation; moreover, this method can give some guidance to the
technologist, on how to predict physical parameters of MIS solar cells
development work and in production control, and may help the theoretician
as well in relating theoretical results with experimental.
REFERENCES
[1] N.K. Swami, S. Srivastava and H.M. Ghule, The role of the interfacial layer in Schottky
barrier solar cells, J.Phys.D: Appl.Phys. Vol. 12, 1979.
[1] J.P. Singh & R.S. Srivastava, Efficiency of SnO
2
/Si Solar cells, Indian Journal of Pure &
Applied Physics, Vol. 20. (1982) pp.104-110.
[2] A.M. Cowley and S.M. Sze, in Journal of Applied Physics, Vol. 9, 1966, p.1035.
[3] Ricaud & C. Neville, Solar Energy Conversion: The Solar cell, Elsevier edition (1995).
[4] J. Carstensen, G. Popkirov, J. Bahr, H.F. FLL, CELLO: An advanced LBIC measurement
technique for solar cell local characterization, Photovoltaic and Photoactive Materials-
Properties, Technology and Applications, eds. (2002).
212 CHEKNANE, et al.
p
Predicted values Parameter
Our method Other methods [1]
Measured values
V
OC
(V) (=16,
0
=0.3 eV) 0,477 0.498 0.470
R
S
(m)(=21,
0
=0.3 eV) 719.05 717.98 721.3
(%)(=0,
0
=0.3 eV) 6.75% 6.3% 7%
(%)(=24,
0
=0.3 eV) 17.13% 16.94 % 17.16 %

S-ar putea să vă placă și