Available online at www.sciencedirect.

com

ScienceDirect
Solar Energy 127 (2016) 206–215
www.elsevier.com/locate/solener

Simulation and optimization of n-type interdigitated back

contact silicon heterojunction (IBC-SiHJ) solar cell structure
using Silvaco Tcad Atlas
M. Belarbi a,⇑, M. Beghdad b, A. Mekemeche b
a
Laboratory ‘‘Physico-Mecanical and Metallurgical Elaboration and Control of Materials” and Department of Technics Sciences,
Faculty of Engineering Sciences, University Abdelhamid Ibn Badis, Mostaganem, Algeria
b
Laboratory ‘‘Signal-Systems” and Department of Physics, Faculty of Exact Sciences and informatics, University Abdelhamid Ibn Badis,
Mostaganem, Algeria

Received 30 November 2015; received in revised form 3 January 2016; accepted 12 January 2016

Communicated by: Associate Editor Takhir M. Razykov

Abstract

Simulation models of interdigitated back contact silicon heterojunction (IBC-SiHJ) solar cells, not only help in understanding the cell
behavior in line with the experimental results but also help further in predicting the cell performance, adding to the cost effectiveness in
the cell processing.
IBC-SiHJ solar cells that combine the hydrogenated amorphous silicon/crystalline silicon (a-Si:H/c-Si) heterojunction and interdig-
itated back contact (IBC) concepts are very promising in order to reach the highest one-junction efficiency (g). In this paper, we have
studied these solar cells by two dimensional modeling using Silvaco Tcad Atlas software which has recently extended its capability to
simulate these devices and given accurate bulk and interface complex defect models and allowed special specification of transport physics
for the hetero-interface. The study has been done on the IBC-SiHJ structure based on n-type crystalline silicon (c-Si) by introducing a
very thin intrinsic hydrogenated amorphous silicon (i-a-Si:H) layer between the c-Si base and the doped a-Si:H layers and varying the
values of the following parameters: c-Si substrate and back-surface field (BSF) doping concentration, thickness of i-a-Si:H layer (Thi-a-Si)
and rear side geometry. The impact of these parameters has been tested by generating the current–voltage characteristics under illumi-
nation. It is shown that the open circuit voltage (VOC) and g of IBC-SiHJ solar cells increase with decreasing i-a-Si:H thickness. The g
improves further with the increase of p-type emitter width (2Wp), the decrease of n-type BSF width (2Wn) and gap width (Wg) which are
explained by the simulation. The S-shaped J–V curves with low fill factor (FF) observed previously in experiments are confirmed by sim-
ulation. To improve FF, Thi-a-Si and Wg should decrease. Results indicate that to achieve high g, c-Si substrate and BSF doping concen-
tration must be optimized. The Wg (spacing between BSF and the emitter) must be kept as small as possible to avoid recombination of
minority carriers in the base. The optimum geometry corresponds to a minimum size BSF region and a maximum size emitter region.
With these optimizations, an enhanced g 23.20% is demonstrated by the simulation.
Ó 2016 Elsevier Ltd. All rights reserved.

Keywords: Silicon solar cells; Interdigitated back contact; Amorphous-silicon; IBC-SiHJ; Atlas software

⇑ Corresponding author.
E-mail address: wbelarbi@yahoo.fr (M. Belarbi).

http://dx.doi.org/10.1016/j.solener.2016.01.020
0038-092X/Ó 2016 Elsevier Ltd. All rights reserved.
 M. Belarbi et al. / Solar Energy 127 (2016) 206–215
1. Introduction
Physically-based simulation for solar cells has become
very important for the reasons that, it is almost always
much quicker and cheaper than performing experiments
and it provides information that is difficult or impossible
to measure.
The IBC-SiHJ solar cell using thin layers of a-Si:H
deposited at low temperature on a c-Si substrate is one
of the most interesting technological solutions for the
photovoltaic market, basically due to the excellent per-
formance and the simple low-temperature process
(Taira et al., 2007). Recently, Panasonic Corporation
has achieved a world conversion efficiency of 25.6% on
a commercial sized (143.7 cm2) monocrystalline-based
‘HIT’ solar cell (Masuko et al., 2014). The reduction
improvements in recombination loss, optical loss
and resistance loss were contributors to the efficiency
record.
The IBC-SiHJ solar cell combines the advantages of the
IBC which has all the contacts at the back of the cell elim-
inating contact shading, leading to a higher short-circuit
current (JSC) and silicon heterojunction solar cells with
high VOC due to the better surface passivation of the depos-
ited i-a-Si:H layer (Lammert and Schwartz, 1977). As
result of this, IBC-SiHJ has the potential of higher VOC
and JSC (Smith et al., 2010). However, it is found experi-
mentally that, this deposited i-a-Si:H layer leads to a low
FF, and ‘‘S” shape J–V curve is observed (Lu, 2008; Lu
et al., 2007). But experimental results show that lowering
its band gap improves the fill factor (Lu et al., 2007). To
confirm these experimental results, there were several pub-
lications about 2-d modeling approaches using numerical
simulator such as: Sentaurus and Atlas devices. Their
simulations studies were performed to optimize the cell
parameters and help further in predicting its performance.
They reported high performance (>20%) for small area
cells demonstrating the potential of IBC-SiHJ approach
(Diouf et al., 2009, 2011; Lu et al., 2009). Even though they
have achieved some improvements, IBC-SiHJ cell is still
far from its expected efficiency potential. The main bottle-
neck for higher efficiency is up to now mainly limited by
low FF values and high series resistance (Rs), compared
with the conventional Si solar cell with diffused emitter
homojunction and metal–Si direct contacts. The FF is
low due to the résistance of i-a-Si:H layer, rear emitter
(Diouf et al., 2009; Tucci et al., 2008) and contact (Lu
et al., 2007; Ji et al., 2012). Improving the efficiency further
requires the following:
(1) Optimized intrinsic buffer layer for surface passiva-
tion to provide lowest surface recombination.
(2) The geometry of back side (pitch, p-type emitter,
n-type BSF and gap width) should be optimized to
minimize series resistance. In this work, methods to
optimize IBC-SiHJ solar cell with improved FF, less
resistive losses and passivation quality are discussed
207
and guided by Atlas device simulation software oper-
ating in two dimensions and including a wide variety
semiconductor physics models for drift-diffusion
transport, SRH recombination, Auger recombina-
tion, surface recombination, carrier generation,
Fermi–Dirac statistics, doping effects, band gap nar-
rowing, tunneling, etc. We first present the geometri-
cal structure of the solar cell, then explore different
ways of the simulation by specifying the different
physical models, and finally examine the simulation
results in order to determine the important
parameters to solve these problems and reach high
g values.
2. Device structure
2.1. Geometry of the structure
Due to the structure periodicity of the IBC-SiHJ solar
cell, an elementary cell is used, that will serve as a basis
for optimizing the performance of this type of cell and its
schematic illustration is shown in Fig. 1. The width of this
elementary structure (pitch) is equal to half the distance
between two electrodes with the same polarity. The geo-
metrical and material parameters of the simulated device
were chosen in agreement with a realistic fabrication
process.
The substrate is a 250 lm thick and 1180 lm wide n-
type Cz-Silicon wafer which contains fewer defects than
float-zone (FZ) or multicrystalline material and therefore
higher efficiencies are obtained. The sheet resistance is
2.25 X cm (Basore et al., 2014).
The base is chosen n-type leading to distinctly higher
VOC and hence higher g compared to n-type a-Si:H emit-
ters on p-type c-Si substrates (Sawada et al., 1994; Jensen
et al., 2002).
For the reduction in the interfacial recombination state
density to a minimum, it is necessary to introduce a very
thin i-a-Si:H layer between the c-Si base and the heavily
doped a-Si:H layers (Taguchi et al., 2000). On the front
of the c-Si substrate an optical layer is placed that plays
the role of anti-reflective coating layer like silicon nitride
(SiNx). This front surface is subject to the illumination.
On the rear side, the contacts of the p-type emitter and
n-type BSF have an interdigitated pattern (that are
formed using multiple lithography/masking, alignment,
etching and/or deposition steps) of which the pitch was
determined by optimization toward so called pseudo sha-
dow losses. The thickness of Al contacts is 0.2 lm (Lu
et al., 2009). The emitter (p-strip) and the BSF (n-strip)
are completely covered by aluminum (Al) metal contacts,
their depth is 0.02 lm (Lu et al., 2009), theirs half default
widths are respectively 950 and 180 lm and are interdig-
itated as mentioned above, and the gap between them is
set to be 50 lm. Some dimensions will be tuned in the
simulation.
208 M. Belarbi et al. / Solar Energy 127 (2016) 206–215
Fig. 1. Schematic picture of IBC-SHJ solar cell: (a) cross sectional view and (b) bottom view.
2.2. Physical models
Modeling of IBC-SiHJ solar cell is a matter of some dif-
ficulty caused by the necessity of treating 2-dimensional
currents, the presence of amorphous materials with
ambiguous parameters some of which are difficult to obtain
from experiment, complex transport mechanisms, modified
light generation profile and so on. Neither of conventional
programs used in photovoltaic (e.g. PC1D, AFORS-HET,
PC2D, . . .) are suitable. However, Silvaco ATLAS
supports all these requirements. In this paper, the work
presented is a step toward understanding the IBC-SiHJ
cells with simulation performed using Silvaco ATLAS,
which has recently extended its capability to support
heterojunctions, midgap and interface traps for these types
of solar cells (Atlas, 2013). The paper focuses on the
behavioral trends in the cell electrical parameters viz.
JSC, VOC, FF and g, with variations in IBC-SiHJ
geometrical parameters and c-Si substrate and back-surface
field (BSF) doping concentration.
Atlas is a physically-based two and three dimensional
device simulator that predicts the electrical behavior of
semiconductor devices at specified bias conditions. The
physical structures simulated with Athena are used as input
by Atlas. The combination of Athena and Atlas makes pos-
sible to determine the impact of process parameters on
device characteristics.
The simulation is based on the solution of the three gov-
erning semiconductor equations: Poisson’s equation, elec-
trons and holes continuity equations. Fermi statistic was
used for carriers with drift-diffusion combined with Bohm
Quantum Potential for quantum correction. Fermi model
and recombination models (Shockley–Read–Hall (SRH),
Auger and surface recombination) were also included into
the simulation. The radiative recombination is not
included, because in silicon, band to band generation/
recombination is insignificant. The carrier mobility was
taken dependent on the doping concentration. Recombina-
tion of carriers through SRH and Auger mechanisms were
also modeled as a function of doping concentration (Lu
et al., 2009).
2.3. Recombination models applied to c-Si
2.3.1. SRH recombination
The SRH recombination is modeled as follows:
pn  n2ie
RSRH ¼ h D Ei h D Ei
sp n þ nie exp Ekt E i
þ s n p þ n ie exp Ei Et
B T k B T
ð1Þ
sn and sp are the electron and hole lifetimes that depend on
the c-Si doping concentration (Fossum and Lee, 1982) and
nie is the effective intrinsic carrier concentration. Et and Ei
are the trap energy level and the intrinsic Fermi level
respectively, T is the lattice temperature in degrees Kelvin.
n and p are the electron and hole concentrations. This
model only presumes one trap level in which, Et = Ei and
it corresponds to the most efficient recombination center.
The lifetime associated to SRH recombination is given
by Law et al. (1991):
 M. Belarbi et al. / Solar Energy 127 (2016) 206–215 209

sn0 sp0 has been introduced at the hetero-interface a very thin

sn ¼ sp ¼
1 þ N =N SRHN 1 þ N =N SRHP defective layer of c-Si (Diouf et al., 2011). An AM1.5G
here N is the doping concentration of c-Si, solar spectrum is used for the optical generation to simu-
NSRHN = NSRHP = 5  1016 cm3. late the J–V curve under standard one-sun illumination
conditions at an intensity of 0.1 W cm2. A Sopra database
is used for a-Si index of refraction (Atlas, 2013).
2.3.2. Auger recombination
Auger recombination is commonly modeled using the
2.4. Meshing of the structure
expression:
RAuger ¼ AUGN ðpn2  n0 n2ie Þ þ AUGp ðnp2  p0 n2ie Þ ð2Þ The correct specification of a mesh is critical in process
simulation. The number of nodes in the mesh has a direct
where Auger recombination parameters are taken as:
influence on simulation accuracy and time. A mesh as thin
AUGN = 8.3  1032 cm6/s and AUGP = 1.8  1031
as possible applied to the whole structure ensures good
cm6/s.
accuracy of calculations but requires greater computation
n (p) the concentration of electrons (holes), n0(p0) the
time to simulate the behavior of this structure. It is there-
corresponding values at equilibrium and nie is the effective
fore necessary to find a compromise between computa-
intrinsic concentration.
tional time and accuracy of the calculation. A finer mesh
For low-injection conditions, the Auger lifetime for
should exist in those areas of the simulation structure
n-type c-Si is:
where ion implantation will occur, where p-n junction will
1 be formed, or where optical illumination will change pho-
sAug ¼
1:8  1024 N 1:65
D
toactive component concentration and a coarse mesh in
areas where these quantities are quasi static. In the middle
of the c-Si substrate, a coarse mesh is used as the physical
2.3.3. Surface recombination quantities do not vary significantly. Fig. 3 represents the
The standard method is to model surface recombination mesh used to simulate IBC-SiHJ solar cell.
in a similar manner as the bulk generation–recombination
rate. The calculation of this rate is an extension of the SRH 3. Results and discussion
theory by introducing the surface recombination velocities
for electrons and holes respectively (Sn0 or Sp0). The recom- The efficiency is a major parameter in order to reduce
bination rate is calculated as follows: the overall costs of a given PV silicon technology. This
pn  n2i means that every effort to improve the efficiency at the cell
Rsurf ¼ h D Ei h D Ei and module level has a direct impact on the whole system
seff
n n þ ni exp Ekt E
BT
i
þ seff
p p þ ni exp Eki E
BT
t
cost. On IBC-SiHJ structure on n-type crystalline silicon
ð3Þ substrate, previous simulations studies (Lu et al., 2007,
2009; Diouf et al., 2009, 2011) reported limiting factor in
Here: the improvement of JSC, VOC, FF and g. Our work pre-
1 1 di 1 1 di sented is a step toward improving further this factors spe-
¼ i þ S n0 and ¼ i þ S p0 cially the g.
seff
n s n Ai seff
p s p Ai
The simulation of IBC-SiHJ solar cell was performed,
sin is the bulk lifetime calculated at node i along the inter- and Figs. 4 and 5 show the current density–voltage (J–V)
face. The di and Ai parameters are the length and area of curves for the variation of parameters i-a-Si thicknesses
the interface for node i. and n-BSF dopings respectively.
For a-Si layers, critical parameters like band gap, dop- The influence of the i-a-Si:H thickness variation is illus-
ing and defect distribution are defined in the input deck. trated in Fig. 4, where the J–V characteristic is depicted. It
The critical parameters for accurate simulation are energy helps us determine the electrical performance of an
distribution of the exponential band tails, and the Gaussian IBC-SiHJ solar cell. Increasing the i-a-Si:H thickness from
distribution of the mid-gap trap states. They were chosen 2 to 15 nm has no influence on JSC and on VOC, also the FF
according to reference (Lu et al., 2009; Munos et al., is not affected, but beyond the value 15 nm the ‘‘S”
2011) and shown in Fig. 2. The input parameters used in shape with a low FF is demonstrated and it confirmed
simulation and the parameters characterizing the mid-gap the experiment results given in Lu et al. (2009). The varia-
and band tail defect densities are shown in Table 1 and sim- tion of n-BSF doping (we do not consider the possible
ilar to Lu et al. (2009) and Munos et al. (2011). increase in interface recombination velocity with increased
For c-Si/a-Si interfaces at the back surface we have used doping level of a-Si:H (n) (Jeyakumar et al., 2014)) and its
a thermionic emission model in which the distribution influence on the J–V characteristic is illustrated in Fig. 5.
function of the interface defect is modeled by two capture As it can be seen from the J–V curves, the S-shape is
cross-sections, one for the holes and one for the electrons. totally removed when n-BSF doping is P5  1018 cm3.
To have realistic modeling of this interface defect states, it Increasing the n-BSF doping from 5  1016 to
210 M. Belarbi et al. / Solar Energy 127 (2016) 206–215

1E21 1E21
Acceptor tail state density
Acceptor tail state density
Donor tail state density
Donor tail state density
1E20 Acceptor mid-gap defects 1E20 Acceptor mid-gap defects
Donor mid-gap defects
Donor mid-gap defects
-1
)

)
1E19

-1
1E19
Defect levels (cm -3 .eV

Defect levels (cm-3 .eV

1E18 1E18

1E17 1E17

1E16 1E16

1E15 1E15
0,0 0,2 0,4 0,6 0,8 1,0 1,2 1,4 1,6 1,8 0,0 0,2 0,4 0,6 0,8 1,0 1,2 1,4 1,6 1,8
Energy (eV) Energy (eV)
Defect density levels in p-type amorphous layer Defect density levels in n-type amorphous layer

1E18 Acceptor tail state density 1E17

Donor tail state density
1E17 Acceptor mid-gap defects
Donor mid-gap defects
Defect levels (cm -3.eV -1)

1E16 1E16
Defect levels (cm- 3.eV - 1 )

1E15

1E14 1E15

1E13 Acceptor mid-gap defects

Donor-midgap defects
1E12 1E14

1E11

1E10 1E13
0,0 0,2 0,4 0,6 0,8 1,0 1,2 1,4 1,6 1,8 0,0 0,2 0,4 0,6 0,8 1,0
Energy (eV) Energy ( eV )

Defect density levels in intrinsic amorphous layer Defect density levels in c -Si defective layer

Fig. 2. Density of states (DOS) in a-Si:H p-type, a-Si:H n-type, i-a-Si:H and defect region layer respectively.

5  1018 cm3 has a significant influence on the JSC, but For doping concentrations below 1016 cm3 (which cor-
not on the VOC. responds to NSRH), the SRH recombination mechanism is
the predominant one. The lifetime is mainly determined by
3.1. Sensitivity on c-Si substrate doping the SRH process as seen in Fig. 6. The Auger recombina-
tion becomes significant when concentrations are above
A set of simulation is performed in order to estimate the 1016 cm3 and even becomes the dominant recombination
sensitivity of cell performance to different bulk dopings, process for Nd > 5  1016 cm3.
which is varied within the range 2  1016–3.5 1016 cm3. When the doping concentration increases, the JSC
This sensitivity was evaluated while keeping the total decreases as we observe in Fig. 7. This is due to the lifetime
device parameters fixed, with front surface recombination of carriers and their mobility which decrease with higher
velocity (SRV) 10 cm/s derived from lifetime measurements doping concentration and the Auger recombinations
(Lu et al., 2009), which gives good passivation at the front becomes stronger. The impact of c-Si substrate doping on
and rear surfaces and low recombining a-Si:H/c-Si inter- VOC is negligible.
faces. The bulk lifetime is defined as: By increasing the c-Si doping concentration, the FF is
1=sbulk ¼ 1=sSRH þ 1=sAug improved, see Fig. 7. This improvement is related to the
decrease of the resistivity of the c-Si substrate.
It is determined by SRH and Auger recombination pro- The parameters JSC and FF do not evolve in the same
cess and these processes depend on the doping concentra- direction when the doping concentration varies. These var-
tions. sbulk is depicted in Fig. 6. ious changes result in the existence of an optimum c-Si
 M. Belarbi et al. / Solar Energy 127 (2016) 206–215 211

Table 1
List of modeling parameters and defect distributions used in the simulations.
Material Interface c-Si n-Type a-Si:H Buffer a-Si:H p-Type a-Si:H
defects
Bulk lifetime (s) 2  103 2  1012 2  1012 2  1012
Band gap (eV) 1.17 1.17 1.70 1.70 1.65
Electron affinity (eV) 4.05 3.9 3.9 3.9
Effective conduction band DOS (cm3) 2.89  1019 2.5  1020 2.5  1020 2.5  1020
Effective valence band DOS (cm3) 3.14  1019 2.5  1020 2.5  1020 2.5  1020
Conduction tail states Nctail A (cm3) 0 n/a 1021 1018 1021
Ectail A (eV) 0.07 n/a 0.12 0.09 0.07
Valence tail states NVtail D 0 n/a 1021 1018 1021
(cm3)
EVtail D (eV) 0.12 n/a 0.12 0.09 0.12
3
Acceptor-like (A) dangling bond states db
N A (cm ) 2.4  10 16
n/a 1019
10 16
1019
Edb A (eV) 0.5 n/a 0.7 1.1 1.3
rdb A (eV) 0.2 n/a 0.2 0.15 0.2
Donor-like (D) dangling bond states Ndb D (cm3) 2.4  1016 n/a 1019 1016 1019
Edb D (eV) 0.5 n/a 0.45 0.9 1.1
rdb D (eV) 0.2 n/a 0.2 0.15 0.2
Electron mobility (cm2 V/s) 1417 1417 2.03 2.03 2.03
Hole mobility (cm2 V/s) 470 470

40

30
Current density (mA.cm-2 )

0.002 μm
20
0.006 μm
0.010 μm
0.015 μm
0.020 μm
10

0
0,0 0,1 0,2 0,3 0,4 0,5 0,6 0,7
Fig. 3. Meshing of elementary IBC-SiHJ solar cell structure. Voltage ( V )

Fig. 4. J–V curves as function of i-a-Si:H thicknesses.

doping concentration corresponding to the maximum g of
IBC-SiHJ solar cells, as shown in Fig. 7. This optimal dop-
ing concentration is around ND = 2.4  1016 cm3
(q = 2.25 O cm) for s0,SRH = 1 ms and tends to decrease
due to the degradation of the carrier lifetime and mobility.
Then, it is necessary to use c-Si substrates resistive
(q = 2.25 O cm) for better performance.
3.2. Sensitivity on n-BSF doping
The a-Si:H applied to the rear side of IBC-SiHJ devices
should have an enhanced rd no matter the rph, since there is
no light-activated conduction at the back side of the solar
cell. Besides, low defect state density is also required to
enhance passivation properties. Then, since dark conduc-
tivity of good quality (low defective) intrinsic layers is very
low, the only way to reduce series resistance is by doping
them, which thus also introduces defects.
The effect of n-BSF doping was evaluated while keeping
the total device parameters fixed as above. The JSC
increases slightly when the doping concentration increases
from 4.5  1018 to 5.4  1018 cm3, beyond this value,
the JSC becomes saturated as we observe in Fig. 8. When
increasing active doping concentration of (n) a-Si:H, band
bending at the a-Si:H/c-Si hetero interface increases, which
212 M. Belarbi et al. / Solar Energy 127 (2016) 206–215

40 the maximum g of IBC-SiHJ solar cells, as shown in

Fig. 8. This optimal doping concentration is around
ND = 4.8  1018 cm3 which corresponds to the maximum
of g.
30
Current density (mA.cm-2 )

3.3. Effect of IBC-SiHJ solar cell device dimensions

The results of the data of experiments (DoE) are shown

20
in Table 2 below.

15 -3
3.3.1. Effect of Thi-a-Si
5x10 ( cm ) The aim of including an intrinsic a-Si:H layer is to pas-
10 16 -3
5x10 ( cm ) sivate the dangling bonds on the c-Si surface. As a result,
17 -3
5x10 ( cm ) the a-Si:H/c-Si interface defect-state density is significantly
18 -3
1x10 ( cm ) reduced. Applying an emitter and BSF buffer layer, 726 mV
18 -3
0 5x10 ( cm ) is reached (Table 2) due to the relatively low recombination
0,0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 rate at the a-Si:H/c-Si interface.
Voltage ( V ) Table 2 shows a slight increase of VOC from 720 to
726 mV when the Thi-a-Si increases from 3 to 17 nm. This
Fig. 5. J–V curves as function of n-BSF doping. enhancement of the VOC is due to low defective i-a-Si:H
layers. However, the FF reduces drastically from 83.44%
to 74.60% with increasing i-layer thickness. The impact of
Thi-a-Si on JSC is negligible, as we observe in Table 2. These
predicted trends are qualitatively consistent with the results
of other authors e.g. in references Lu et al. (2009) and
Diouf et al. (2011). We found an optimum Thi-a-Si = 6 nm,
which leads to a highest g.

3.3.2. Effect of the rear side geometry

The geometry of the rear side is also one of the research
areas to optimize the performance of IBC-SiHJ solar cells.
The Wp, Wn and Wg, may have an impact on IBC-SiHJ
solar cells. Note that the varied widths in the simulation,
Wp and Wn, correspond to half of the entire emitter and
BSF stripe width respectively.

3.3.2.1. Effect of Wp. The study of the Wp influence is

Fig. 6. Effect of SRH and Auger recombination mechanisms on the bulk
lifetime sbulk, depending on the doping concentrations ND.
is favorable for better charge carrier collection. The doping
reduces series resistance, but also introduces defects which
thus limit the JSC.
As it can be seen from the curve VOC vs n-BSF doping
(Fig. 8), the VOC decreases slightly when n-BSF doping
increases, because defect density (number of gap states in
the a-Si:H layers and interface states) is increased. By
increasing the n-BSF doping concentration, the FF is
improved, see Fig. 8. This improvement is related to
enhanced conductivity and to the strong electrical field
which collects carriers and provokes enough band bending.
The parameters VOC and JSC or FF do not evolve in the
same direction when the doping concentration varies.
These various changes result in the existence of an
optimum n-BSF doping concentration corresponding to
important to optimize the geometry of the rear side. We
therefore varied Wp by fixing Wn and Wg and maintainig
the pitch constant. Increasing Wp, from 400 to 1050 lm,
results in a large increase of JSC from 35 to 38.5 mA cm2,
also increasing efficiencies at first, followed by a decrease,
as it can be seen in Table 2. Minority carriers in the c-Si
generated over the emitter region do not need to travel lat-
erally, resulting in higher efficiencies for wider emitters.
Wider emitters also increase the length that the majority
carriers have to travel to reach the BSF increasing the series
resistance. This explains the decrease of the FF observed in
Table 2 when Wp increases. Once series resistance is large
enough a drop in efficiency occurs.
The different evolution of JSC, VOC, FF and g leads to an
optimum of the Wp of 950 lm.
3.3.2.2. Effect of Wn and Wg. Results show that large Wn
and Wg result in decreased efficiencies. This is due to
increased average lateral distance that minority carriers
have to travel to reach the emitter, which increases their
 M. Belarbi et al. / Solar Energy 127 (2016) 206–215 213

38,5 0,724 84,2

23,20

38,4 84,0
23,18
Jsc
0,723
83,8
38,3 23,16

Voc ( V )
Jsc (mA.cm -2)

η (%)
FF (%)
83,6 η 23,14
38,2 Voc
0,722
FF
83,4 23,12
38,1

83,2 23,10
38,0 0,721
2,0E16 2,4E16 2,8E16 3,2E16 2,0E16 2,4E16 2,8E16 3,2E16

c-Si doping concentration (cm ) -3

c-Si doping concentration (cm-3)

Fig. 7. IBC-SiHJ solar cell key figure of merits as a function of c-Si doping concentration.

38,378
83,65 23,20

0,724
83,60
23,19
Jsc (mA.cm )
-2

Jsc 83,55
FF (%)

23,18

η(%)
38,377
Voc (V)

η
83,50
0,722 FF 23,17
Voc 83,45

23,16
83,40

38,376 0,720 83,35 23,15

4,8E18 5,2E18 5,6E18 4,8E18 5,2E18 5,6E18

n-BSF doping concentration ( cm )

-3
n-BSF doping concentration ( cm -3 )
Fig. 8. IBC-SiHJ solar cell key figure of merits as a function of n-BSF doping concentration.

chances to recombine. Increasing the Wn from 80 to Increasing Wg results in a decrease of JSC, FF and g, but
450 lm results in a slightly decrease of JSC as we can see the VOC remains constant. The Wg will be taken as short
in Table 2. Wn variation has no influence on VOC. The as possible with good surface passivation to avoid recombi-
BSF should be taken as narrow as possible to reach best nation of minority carriers in the bulk c-Si and damage of
JSC values, but narrow n-strip can cause electrical shading the cell performance. To remove this gap region would be
effects. From the other side wide n-strip can increase series an attractive solution but this would lead to internal cell
resistance to majority carriers current. Thus Wn should be short circuit, thus making this solution is unsuited. Smaller
optimized. The optimum value which corresponds to the gap (between emitter and BSF region) was effective to
highest g is 180 lm. increase g and thus, we assumed a suitable value of
In Table 2, the smaller the Wg, the higher the efficiency 50 lm in order to eliminate the possibility of shunting.
is, because the recombination of minority carriers on the With these optimizations, a conversion g 23.20% is
surface between emitter and BSF is smaller. As for the reached.
BSF region, increased Wg corresponds to an additional lat-
eral distance to travel for minority carriers photogenerated 4. Conclusion
in c-Si above the BSF region. This additional distance
increases their chances to recombine before reaching the The simulation has provided an initiative to help
emitter. In addition, the interface with the c-Si substrate understand and obtain a detailed analysis of IBC-SiHJ cell
and the gap region can also be recombining. The impact characteristics. 2D numerical simulations for IBC-SiHJ solar
of the Wg on JSC, on FF and g is shown in Table 2. cells have been studied by using the software SILVACO
214 M. Belarbi et al. / Solar Energy 127 (2016) 206–215

Table 2
Summary of device performance of IBC-SiHJ solar cell.
Thi-a-Si (lm) JSC (mA cm2) VOC (mV) FF (%) g (%) Wp (lm) JSC (mA cm2) VOC (mV) FF (%) g (%)
0.003 38.4646 720 83.44 23.14 400 35.09 722 84.05 21.34
0.004 38.4647 721 83.40 23.17 500 36.08 722 84.02 21.94
0.006 38.4646 723 83.25 23.20 600 36.83 722 83.97 22.38
0.008 38.4645 724 83.06 23.16 700 37.48 722 83.91 22.76
0.010 38.4643 725 82.81 23.12 800 37.93 723 83.81 23.01
0.012 38.4641 725 82.47 23.05 900 38.27 723 83.64 23.16
0.015 38.4636 726 81.75 22.87 950 38.38 723 83.50 23.20
0.017 38.4632 726 80.60 22.54 1000 38.46 723 83.23 23.17
0.020 38.4625 724 74.60 20.80 1050 38.52 723 82.69 23.06
Wn (lm) JSC (mA cm2) VOC (mV) FF (%) g (%) Wg (lm) JSC (mA cm2) VOC (mV) FF (%) g (%)
080 38.52 723 82.71 23.07 30 38.41 723 83.52 23.22
110 38.48 723 83.12 23.16 40 38.39 723 83.51 23.21
130 38.46 723 83.26 23.18 50 38.37 723 83.50 23.20
180 38.37 723 83.51 23.20 60 38.35 723 83.49 23.18
250 38.21 723 83.68 23.14 70 38.33 723 83.48 23.16
300 38.03 723 83.77 23.06 80 38.31 723 83.47 23.15
350 37.84 723 83.83 22.96 90 38.29 723 83.46 23.13
400 37.62 723 83.88 22.83 100 38.27 723 83.45 23.12
450 37.32 723 83.92 22.67 110 35.25 723 83.44 23.11
Values in bold correspond to the maximum efficiency.

ATLAS Device, and several geometrical parameters were
studied and their impact was shown on the IBC-SiHJ solar
cell output characteristics. The general results of 2D distri-
bution and current density–voltage curves were generated,
and the dependence of cell performance on the c-Si
substrate and n-BSF dopings and rear side geometry are
evaluated. The simulations suggest that c-Si substrate
and n-BSF dopings should be optimized with the values
2.4  1016 cm3 and 4.8  1018 cm3 respectively. The
optimum geometry corresponds to a minimum size Wn
equals to 180 lm and a maximum size Wp region equals
to 950 lm. The width of the gap region (spacing between
BSF and the emitter) must be kept as small as possible to
avoid recombination of minority carriers in the bulk c-Si
and its value is 50 lm. These specific parameters values
have increased the cell performance, and the g is enhanced
and permitted a conversion equals 23.20%, thus demon-
strating the great potential of these optimum parameters.
References
Atlas User’s Manual: Device Simulation from Silvaco International,
Version 5.19.20, 2 October 2013 (in <http://www.silvaco.com>).
Basore, P.A., Clugston, D.A., Cabanas-Holmen, K., 2014. Aid PC2D
V2.4 and PC1D V5.9 (in <http://pc2d.info>, on 10 April 2014).
Diouf, D., Kleider, J.P., Desrues, T., Ribeyron, P.J., 2009. Study of
interdigitated back contact silicon heterojunctions solar cells by two-
dimensional numerical simulations. Mater. Sci. Eng. B 159–160, 291–
294.
Diouf, D., Kleider, J.P., Longeaud, C., 2011. Two-Dimensional simula-
tions of interdigitated back contact silicon heterojunctions solar cells.
Book Physics and Technology of Amorphous-Crystalline Silicon
Heterostructure Solar Cells. Edit. Springer, pp. 483–519 (Chapter 15).
Fossum, J.G., Lee, D.S., 1982. A physical model for the dependence of
carrier lifetime on doping density in nondegenerate silicon. Solid State
Electron. 25, 741–747.
Jensen, N., Hausner, R.M., Bergmann, R.B., Werner, J.H., Rau, U.,
2002. Optimization and characterization of amorphous/crystalline
silicon heterojunction solar cells. Prog. Photovolt. Res. Appl. 10,
1–13.
Jeyakumar, R., Maiti, T.K., Verma, Amit, 2014. Two-dimensional
simulation studies on high-efficiency point contact back heterojunction
(a-Si:H/c-Si) solar cells. Sol. Energy 105, 109–115. http://dx.doi.org/
10.1016/j.solener.2014.03.030.
Ji, K., Syn, H., Choi, J., Lee, H.M., Kim, D., 2012. The emitter having
microcrystalline surface in silicon heterojunction interdigitated back
contact solar cells. Jpn. J. Appl. Phys. 51, 10NA05.
Lammert, M.D., Schwartz, R.J., 1977. The interdigitated back contact
solar cell: a silicon solar cell for use in concentrated sunlight. IEEE
Trans. Electron Devices 24, 337–342.
Law, M.E. et al., 1991. Self-consistent model of minority-carrier lifetime,
diffusion length, and mobility. IEEE Electron Device Lett. 12 (Nov 8).
Lu, Meijun, Das, U., Bowden, S., Birkmire, Robert, 2008. Rear surface
passivation of interdigitated back contact silicon heterojunction solar
cell and 2D simulation study. In: 33rd IEEE Photovoltaic Specialists
Conference, PVSC ’08, pp. 1–5. http://dx.doi.org/10.1109/PVSC.2008.
4922757.
Lu, M., Bowden, S., Das, U., Birkmire, R., 2007. Interdigitated back
contact silicon heterojunction solar cell and the effect of front surface
passivation. In: Proceedings of the 22nd European Photovoltaic Solar
Energy Conference, Milano, Italy, pp. 924-927, or Appl. Phys. Lett. 91
(2007) 063507. http://dx.doi.org/10.1063/1.2768635, URL: <http://
link.aip.org/link/?APL/91/063507/1>.
Lu, M. et al., 2009. Optimization of interdigitated back contact silicon
heterojunction solar cells by two-dimensional numerical simulation.
In: Proceedings of 34th IEEE PVSC, Philadelphia, USA.
Masuko, Keiichiro, Shigematsu, Masato, Hashiguchi, Taiki, Fujishima,
Daisuke, Kai, Motohide, Yoshimura, Naoki, Yamaguchi, Tsutomu,
Ichihashi, Yoshinari, Mishima, Takahiro, Matsubara, Naoteru,
Yamanishi, Tsutomu, Takahama, Tsuyoshi, Taguchi, Mikio, Mar-
uyama, Eiji, Okamoto, Shingo, 2014. Achievement of more than 25%
conversion efficiency with crystalline silicon heterojunction solar cell.
IEEE J. Photovolt. 4 (6).
Munos D. et al., 2011. Key features of highly efficient a-Si:Hc-Si
heterojunction solar cells. 7th Workshop on the Future Direction of
Photovoltaics (by JSPS 175th committee), March.
 M. Belarbi et al. / Solar Energy 127 (2016) 206–215
Sawada, T., Terada, N., Tsuge, S., Baba, T., Takahama, T., Wakisaka,
K., Tsuda, S., Nakano, S., 1994. High-efficiency a-Si/c-Si
heterojunction solar cell. In: Proceedings of 1994 IEEE 1st World
Conference on Photovoltaic Energy Conversion, Waikoloa, USA,
pp. 1219–1226.
Smith, D.D., Luan, H.C., Manning, J., Dennis, T.D., Waldhauser, A.,
Wilson, K.E., Harley, G., Mulligan, W.P., 2010. Generation 3,
improved performance at lower cost. In: Proceedings of 35th IEEE
PV SC, pp. 275–278.
Taguchi, M., Kawamoto, K., Tsuge, S., Baba, T., Sakata, H., Morizane,
M., Uchihashi, K., Nakamura, N., Kiyama, S., Oota, O., 2000.
215
HITTM cells-high efficiency crystalline Si cells with novel structure.
Prog. Photovolt. Res. Appl. 8, 503–505.
Taira, S., Yoshimine, Y., Baba, T., Taguchi, M., Kanno, H., Kinoshita,
T., Sakata, H., Maruyama, E., Tanaka, M., 2007. Our approaches for
achieving HIT solar cells with more than 23% efficiencies. In:
Proceedings of the 22nd EPVSEC, Milan, Italy, pp. 932–935
(ISBN:3-936338-22-1).
Tucci et al., 2008. Back enhanced heterostructure with interdigitated
contact behind solar cell. In: Proceedings of IEEE Conference on
Optoelectronic and Microelectronic Materials and Devices COM-
MAD, pp. 242–245.