Sie sind auf Seite 1von 20

So~r ~ Mn~Uas

and Solar Cells

ELSEVIER Solar Energy Materials and Solar Cells 44 (1996) 79-98

Numerical analysis of field-effect surface


passivation for solar cells
H. Ohtsuka *, T. Uematsu, T. Warabisako
Central Research Laborato~, Hitachi Ltd., 1-280 Higashi-koigakubo, Kokubunji-shi, Tokyo, 185, Japan
Received 11 October 1995; revised 12 January 1996

Abstract

In order to investigate the influence of surface potential on the electric characteristics of solar
cells, the characteristics of conventional cells and back-contact type high-efficiency silicon cells
were analyzed using 2-dimensional numerical simulation, varying the surface electrical potential.
The locations where surface electrical potential is controlled are the rear side in conventional cells
and the front side in back-contact cells. As a result of the calculations, it was found that
field-effect surface passivation yields cell characteristics equivalent to those of a cell with
effective surface recombination velocity of 0 c m / s , even if the cell has a poor S i / S i O 2 interface
(i.e., Dit > 1.0 )< 1011 cm -2 e V - 1). It was also found that both the use of a higher resistivity
wafer and - - especially in p-type substrates - - the formation of inversion layers causes the
field-effect surface passivation to work the fullest effect. In addition, a computer simulation based
on physical-parameter measurements taken from actual materials forecasts that a back-contact cell
would realistically be able to exceed 25% efficiency under AM1.5 global, one-sun illumination.

Keywords: Electron-hole recombination; Passivation; Semiconductor device models; Silicon; Solar cells

1. Introduction

To develop a high energy conversion efficiency silicon solar cell, it is extremely


important to reduce the surface recombination rate. The general trend has been to
passivate both front and rear semiconductor surfaces with a high-quality oxide, because
a high-quality oxidized surface makes the surface recombination velocity low and brings
high efficiency. Two wellknown cells are listed as being more than 22% efficient. One
is the PERL (Passivated Emitter and Rear Locally diffused) cell which was developed
by Green et al. at the University of New South Wales (UNSW) and which has

* Corresponding author. Email: ohtuka@crl.hitachi.co.jp

0927-0248/96/$15.00 Copyright © 1996 Elsevier Science B.V. All rights reserved.


PII S 0 9 2 7 - 0 2 4 8 ( 9 6 ) 0 0 0 3 5 - 9
80 H. Ohtsuka et al. / Solar Energy Materials and Solar Cells 44 (1996) 79-98

demonstrated the highest efficiency of any silicon solar cell so far [1]. The other is the
back-point-contact cell which was made by Swanson et al. at Stanford University [2].
These cells do not resemble each other in their electrode geometry, but the surfaces of
both cells are similar in that they are oxidized thermally with a little TCA (trichloro-
ethane) added to the oxygen. However, even if a semiconductor surface is passivated
with good quality oxide layer, effective surface recombination velocity varies with the
oxidized surface potential. We are researching processes to lessen the number of
recombination centers and simultaneously searching for how to make them inert. As one
method, we are trying to apply field effect passivation to solar cell technology.
Since the late 70s, some attempts have been made to control surface recombination of
the solar cells by means of field effect passivation (for details see Ref. [3]). The
principal recent research on the use of the field effect passivation has been performed by
Aberle et al. [3,4]. They theoretically proved that the surface recombination velocity
depends on the excess-carrier concentration, using a numerical calculator algorithm
developed by Girisch et al. [5] for determining the recombination parameters of the
Si/SiO 2 interface. Furthermore, Aberle's group explained that the asymmetry in the
dependence of effective surface recombination velocity on applied gate voltage could be
attributed to the difference between the capture cross-sections of electrons and holes and
the difference in work functions between the aluminum gate metal and the bulk silicon.
Experimentally, Warabisako et al. demonstrated an effective decrease of surface recom-
bination on thermally oxidized n-type wafers on which they deposited an oxide with a
positive fixed charge by photochemical CVD [6]. Yasutake et al. succeeded in the
fabrication of a plasma CVD oxide layer which had a very low surface recombination
velocity [7] which they concluded was due to a high positive fixed oxide charge density
contained within the layer [8].
In this way, the mechanism of the reduction of effective surface recombination
velocity by the electric field effect has become clear and the principles of a control
system employing it has been discussed. The effectiveness of the field-effect surface
passivation method also has been experimentally confirmed. However, there have been
few studies in which the influence of surface potential on the electrical characteristics of
solar cells was elucidated, since multi-dimension computer simulations have been
necessary to predict those characteristics. Therefore, we chose to analyze the character-
istics of high-efficiency crystalline silicon conventional cells and back-contact cells as
functions of surface electrical potential, using a 2D numerical simulation based on
physical parameters measured from actual materials. In this article, we describe the
influence of the variety of surface electrical potential on solar cell electric characteristics
and the effectiveness of a field-effect surface passivation method.

2. Analysis

2.1. Simulation method

We use the software CADDETH, a general purpose 2 and 3 dimensional semiconduc-


tor-device simulation program developed by HITACHI Ltd. [9]. In this work, we added
14. Ohtsuka et al. / Solar Energy Materials and Solar Cells 44 (1996) 79-98 81

a sub-program to it which calculates the photo-generation rate in a device to make it


possible to analyze solar cell characteristics [10]. Additionally, we applied Eq. (2),
obtained by an integration over the bandgap at the interface to the surface recombination
model [11] to calculate the recombination rate as accurately as possible. We did this
because the usual surface recombination model (Eq. (1)) assumes a single surface state
at midgap independent of the trap energy:
vth ~n O'pUsT( Psn, -- n~ )
Rsurf = O'p{ps q- n i e x p ( - ( E T - E i ) / k T ) } + o'.(n~ + n~exp(( E T - E i ) / k T ) } ' (1)

R = uthf, ec
Ev
on(ET) O'p(ET) Dit ( E T ) ( p~n~ - n~)
×
% ( E T ) {Ps + niexp( -- ( ET -- Ei) / k T ) } + o'n( E T ) { n , + niexp(( ET -- Ei) / k T ) }
×dE T. (2)
In these equations vth is the thermal velocity, % and (rp are the energy dependent
capture cross-section of electrons and holes, respectively, and NsT is the number of
interface trapping centers per unit area. Dit is the energy-dependent interface trap
density, n s and Ps are the carrier concentration of electrons and holes at the surface,
respectively, i n is the intrinsic carrier concentration, E T is the trap energy level, E i is
the intrinsic Fermi level, k is the Boltzmann constant, and T is the temperature. As
shown in Fig. 1 and Fig. 2, Gaussian distributions are assumed both for the interface trap
density and the capture cross-sections of the electrons and holes. In our revised
simulator, multi-spectrum and light trapping are taken into consideration, but surface
and inner reflectance must be input by hand because the program cannot handle true

-,'" I 0 ~4

~. ::--: 6~5e10
oE 1 0 ~3 ..... 6.5ell
sz,
O~
e= 1 012

1011
0
(@

Q)
1 01° = , , I , , * 1 = , = 1 , , , 1 , , , I I

0 0.2 0.4 0.6 0.8 1


Trap Energy (eV)
Fig. 1. Assumed density of interface states in the energy gap used for the calculation. Gaussian distribution
was assumed for the curves instead of U-shaped distribution. The minimum interface trap densities for the
solid and the dashed lines are at 6.5 × 101° cm -2 eV- l and 6.5 × 10jl cm -e eV- 1, respectively.
82 H. Ohtsuka et al. / Solar Energy Materials and Solar Cells 44 (1996) 79-98

10 -14
, electron \
/ \

~ 10 -~s
/
~ 10 16
r tl \
~ 10 -17
0
1048

~. 10 -19
0
1 0 -20 , I , , , I

0 0.2 0.4 0.6 0.8 1


Trap Energy (eV)
Fig. 2. Assumed capture cross-sections of the electrons and the holes in the energy gap used for the
calculation. Gaussian distributions are assumed. The maximum cross-sections were set from our experimental
data. The solid line represents the hole capture cross-section and the dashed line the electron one. The tr,/trp
ratio is much larger than that of reported by UNSW [4].

textured surfaces. Bandbending induced by a fixed oxide charge (1.0 X 101° cm - 2 ) is


taken into account the original C A D D E T H program, but we do not consider bandbend-
ing induced by the difference in the work functions of the silicon and electrode material.
One point that we must note is: If the capture cross-section for an electron is equal to
that for a hole, the surface recombination rate is m a x i m u m when the electron concentra-
tion at surface is equal to the hole concentration at surface. In this work, however, the
surface recombination rate reaches its m a x i m u m when the hole concentration at the
surface is about 10 ~ 1000 times as large as the electron concentration, because the ratio
of the m a x i m u m capture cross-section of an electron to that of a hole is 1000 [3].
The program ran on a supercomputer, H I T A C S-820, using the Coupled Newton
method to obtain the numerical solution. A typical run required about 10,000 mesh
points.

2.2. D e v i c e modelling and simulation p a r a m e t e r s

Cross-sections of the unit cell used in the calculations are shown in Fig. 3. W e
simulated the characteristics of a conventional type cell to which we added a metal-oxide
semiconductor capacitor with gate electrodes positioned at the rear side, and of a
back-contact cell to which we added an electrode at the front side. The reason for
choosing these structure is that surface recombination at the rear side in the conventional
cells and at the front side in the back-contact cells is the dominant factor of the cell
characteristics: in a p-type substrate with a long lifetime, the surface recombination rate
is very high at the interface of the oxide layer and a non-diffused base region owing to
fact that the capture cross-sections of electrons at the midgap are much broader than
those of holes. The oxide layers of our MOS capacitors are 200 nm and 100 nm thick in
the conventional cells and the back-contact cells, respectively.
H. Ohtsuka et al. / Solar Energy Materials and Solar Cells 44 (1996) 79-98 83

SiO2Layer hv

t
/../. -//////t/////////,
4
P coo .
n+ /~ ~'~'~

0.25--~
~ r/tlllllllll/lltllll~/ll~llll/lllllll-i
Contact ~ ~_0~8
\225 ~ 250
unit: I.tm
(a) Back-contact cell (b) Biracial cell
Fig. 3. Schematic diagram of the silicon solar-cell structures investigated in this work. The diagram (a) is the
conventional cell, and (b) is the back-contact cell. The conventional cell has the MOS structures for
field-effect passivation at the rear side, while the back-contact cell has them at the front side.

T h e c o n v e n t i o n a l - t y p e structures in this w o r k are in a c c o r d a n c e w i t h the P E R L cell


m a d e b y G r e e n et al.: the b a c k - c o n t a c t cell s t r u c t u r e m a t c h e s w i t h the b a c k - c o n t a c t cell
m a d e b y S w a n s o n , et al. H o w e v e r , to p e r f o r m 2 D s i m u l a t i o n s , w e a s s u m e t h a t b o t h
t y p e s o f cells h a d l i n e a r d i f f u s i o n r e g i o n s a n d l i n e a r c o n t a c t s o n the rear side a n d that
the d i f f u s i o n r e g i o n s a n d the c o n t a c t s o f this s i m u l a t i o n are the s a m e area as t h o s e o f the
actual p o i n t - c o n t a c t structures. T h i s is b e c a u s e r e c o m b i n a t i o n at the rear surface is
m a i n l y d o m i n a t e d b y the areas at the c o n t a c t s a n d p + regions. A l t h o u g h the d i f f u s i o n
r e g i o n s a n d c o n t a c t s are too n a r r o w to a c t u a l l y fabricate, u n d e r the s a m e - a r e a a s s u m p -
tion the results o f o u r l i n e - c o n t a c t - m o d e l 2 D s i m u l a t i o n s h o w g o o d a g r e e m e n t w i t h

Table 1
Parameters of the solar cells under investigation
Parameters Conventional cell Back-contact cell
Resistivity 2.0 D-cm *--
SRH lifetime T =
~n0
1+ NA /NA0
(%0 = 1.0 ms, NAO= 7.1X l015 cm -3)
Auger recombination coefficient C, = 2.8X 10 -31 cm6/s
Cp = 1.07X I0 -31 cm6/s
n + Layer surface doping concentration 1.0X 1019 cm -3 1.Ox 10 2o cm -s
n-- + Layer surface doping concentration 2.0X 1020 cm -3
p+ Layer surface doping concentration 5.0X 1020 cm 3 1.0 × 1020 cm- 3
Doping distribution Complementary error function
Surface reflection ratio 0.05 0.03
Inner reflection ratio (front side) 0.92
Inner reflection ratio (rear side) 0.97
Temperature 298K (25°C)
Intrinsic carrier concentration 1.01X 101° cm -3
Sunlight GLOBAL AM1.5, 1 k W / m 2
84 H. Ohtsuka et al. / Solar Energy Materials and Solar Cells 44 (1996) 79-98

those of the actual design (a point contact model) by full 3D simulation in the above
assumption [10].
Parameters other than those indicated in Fig. 3, including those of device structure,
bias condition, and diffusion distribution, are listed in Table 1. As shown in the table,
the reflectance of the conventional cells is 3% higher than that of the back-contact cells
due to shadowing loss caused by the front electrode. Table 1 also shows the empirical
doping-dependent SRH lifetime equation which is usually used in silicon device
simulations. For example, the SRH lifetime of the p-type 2 f l . cm silicon substrate used
in this work is estimated at about 500 txs. The maximum interface trap density and the
capture cross-sections of electrons and holes refer to experimental data obtained by our
group. Here, it is conspicuous that the maximum capture cross-section of electrons is
about 1000 times as large as that of holes.
In the following, we use usual units in place of S.I. units for convenience sake.

3. Results and discussion

3.1. D e p e n d e n c i e s on applied voltage

We will first investigate the base line model of a typical conventional cell and
back-contact cell. Fig. 4. shows the calculated electrical AM1.5 characteristics, effi-
ciency, open-circuit voltage, and short-circuit current of baseline cells as a function of
the voltage applied to the MOS structure. In the top figure, the points marked " S = 0"
indicate the results calculated with a surface recombination velocity, S 0, of 0 in Eq. (1),
not changing the surface potential. As shown in Fig. 4, the efficiency of conventional
cells and back-contact cells saturate at about 2.0 V and 1.0 V, respectively, as the
positive gate voltage increases. If one induces a fixed oxide charge instead of using
MOS structure for field-effect surface passivation, these voltages are equivalent to a
positive fixed oxide charge of about 2.0 X 1011 cm -2. The reasons for the improvements
in the efficiency up to these levels are explained as follows: when a positive bias is
applied to the MOS structure in the ceils, the band edge bends downward, and an
inversion layer is formed. This reduces the surface recombination rate, due to the
dramatic decrease in the hole concentration at the surface. As a result, the efficiency
increases up to the point equivalent to a surface recombination velocity of 0 c m / s . Also,
when the increasing positive bias is gradually applied to the MOS structure in the p-type
substrate, a depletion layer is formed before the inversion layer is, so the electron
concentration must be equal to the hole concentration at the surface. However, the
surface recombination rate do not actually reach its maximum, since we are assuming
that the maximum capture cross-section of electrons is 1000 times as large as that of

Fig. 4. Base-line cell characteristics calculated as a function of applied gate voltage. The characteristics of the
conventional cell are shownby solid lines, while the dashed lines show the results for the back-contactcell. In
the efficiencyfigure, the symbols marked "S = 0" show the result of a surface-recombinationvelocityof 0
cm/s calculated by the usual surface recombinationequation. (see Eq. (1)).
11. Ohtsuka et al. / Solar Energy Materials and Solar Cells 44 (1996) 79-98 85

...................Conventional Cell I

25.5
Back-c0ntact Cell
I
I ' ' ' ' 1 ' ' ' ' 1 .... I ' ' ' ' 1 ' ' ' ' 1 ' ' ' ' 1 ' ' ' ' -

25.0 S = 0 []

24.5 -. //

i
,_
24.0

23.5

23.0

2 2 . 5 -,
..........

I , , , , I , , ~ , 1
\
\

....
S =0 0

I ....
/

,//

t ....
/'/ ........

I .... I,,,

-5 -4 -3 -2 -1 0 1 2
Applied Voltage (V)

,-,0.73

8,0.72
'~ 0.71
>

•5 0.70 f

0.69

.0.68
0
0-67.5
-4 -3 -2 -1 0 1 2
Applied Voltage (V)

E 43.0
o
42.5
E
42.0 /
.=
I,,
41.5
o 41.0 \
,l,=l
°D
f ....................
.................... -- . . . . . . f
o= 40.5 - ~ ............. /z
I.=

40.0
~
o 39.5
-5 -4 -3 -2 -1 0 1 2 3
Applied Voltage (V)
86 H. Ohtsuka et al. / Solar Energy Materials and Solar Cells 44 (1996) 79-98

Front Surface
-0.1;
under 1 sun (AM1.5)
>
---0.5V I
:\ - -- -2.0V
'~ -0.2
o
0
\
_o
\ Inversion
w
e -0.a
o
T \'\ \
w
Depletion
t

-0.4 . . . . I . . . . I . . . . I . . . . I . . . .
0 0.1 0.2 0.3 0.4 0.5
Distance from Surface (pro)
Fig. 5, Plots of electrostatic potential from the front surface of a conventional cell with a 2.0 1~. cm substrate
at short-circuit ( = 0 V) condition. The solid line represents an applied gate voltage of - 0.5 zV and the dashed
line of 2.0 V, They correspond to a depletion layer and an inversion layer, respectively.

holes. As a consequence, the potential applied to the Si/SiO 2 interface effectively


decreases the surface recombination velocity and improves the efficiency. The back-con-
tact cell, especially, which is 24.1% efficient with no gatebias can rise to 25.1%
efficiency if we change the surface potential.
On the other hand, when applying negative gate voltage, the efficiencies of conven-
tional cells and back-contact cells fall to minima of about 0 V and 1.0 V, respectively,
before they gradually begin to increase again. The reason for the efficiencies growing
worse is as follows: When the negative bias is applied to the MOS structure in the cells,
the band edge bends upward and an accumulation layer is formed. In our p-type, 2
. cm substrate, when the accumulation layer starts to be formed, the ratio of electron
concentration to hole concentration becomes just that value which maximized the
surface recombination rate. In this way, the applied voltage with the minimum efficiency
is very dependent on the ratio of the capture cross-sections of electrons to holes. In case
where no field-effect surface-passivation is applied, there is a risk that the short-circuit
current in the solar cells may drop rapidly and that the efficiency may grow worse when
the surface potential is not perfectly stable.
When applying negative voltage beyond that point, both cell efficiencies gradually
increase. This is because the electron concentration at the surface declines due to the
band edge bending up, which causes the surface recombination rate to decrease.
Fig. 5 and Fig. 6 clarify these mechanisms. The electrostatic potential distribution of
a conventional cell is shown in Fig. 5. In this figure, the electrostatic potential
H. Ohtsuka et al. / Solar Energy Materials and Solar Cells 44 (1996) 79-98 87

1 023
. . . . i q ,, ' I . . . . I . . . . ~ . . . .

Gate Bias =
E 1 022 ~\ s v | .~- ....
hole
f
C
O
\ /
°m
1 021
, /2.0V
C

'; \2.0 V
o= 1 020 /
O
!
electron
•" 1 019
/........ -0.5 V
0

1 018 i b i i I , ~ i , I i , i , I i i , i I , ~ i ,

0 0.1 0.2 0.3 0.4 0.5


Distance from S u r f a c e (t~m)
Fig. 6. Plots of cartier concentration from the front surface of a conventional cell with a 2.0 f~. cm substrate at
short-circuit ( = 0 V) condition. The solid lines represent the concentrations of the holes and electrons of the
applied gate voltage of - 0.5 V and the dashed line represent them at an applied gate voltage of 2.0 V. It is
clear that the electron and the hole concentration crossover is at about 800 nm depth at a gate bias 2.0 V.

distributions for the short-circuit condition (i.e., 0 V) in the range from the front surface
to 500 nm depth are plotted at applied gate voltages of each - 0 . 5 and 2.0 V. The
electrostatic potential distribution curves are equivalent to those of the valence or
conduction band-diagram.
If one applies - 0.5 V to the rear surface of a conventional cell with a 200 nm thick
oxide layer, an accumulation layer is formed. In the short-circuit condition, the hole
concentration at the surface is over I000 times as high as the electron concentration.
This condition, that is this ratio of electron concentration to hole concentration,
maximizes the surface recombination rate. Applying 1.0 V, the electrostatics potential
edge near the front surface bends up by about 0.15 V, so that the hole and electron
concentration are inverted at 80 nm from the surface. The electron concentration at
surface is then about 10 4 times the hole concentration. In this state, the surface
recombination rate is much smaller.
The characteristics show asymmetry in their dependence on applied gate voltage as
shown in Fig. 4. In substrates used in this work, it is much more effective to form an
inversion layer than it is to form an accumulation layer to reduce the surface recombina-
tion rate. That is to say, we do not need to apply as much gate voltage to form the
inversion layer as we would to form the accumulation layer. This is also because the
maximum capture cross-section of electron is 1000 times as large as that of hole. As
shown Fig. 5, when the inversion layer is formed, the characteristics of the solar cell
saturates when the surface concentration of electrons is about 1000 times as large as that
of holes. Considering the difference in capture cross-sections between electrons and
88 H. Ohtsuka et al. / Solar Energy Materials and Solar Cells 44 (1996) 79-98

24
A _ 6.5 x 109

>, 2 2 6.5 x 10 l° /
J
t~
e..

"~ 20
6 . 5 x 10 ~ . . . . . . .
1.1,.I
18

16 .... i .... I .... ~ .... I .... L .... I .... i,,L

-5 -4 -3 -2 -1 0 1 2
Applied Voltage (V)

A
0.72 .... h .... t .... i .... I .... ~ .... ] .... I ....
>

0.70
= . . . . . . . . . . . . . . . . . - ~-=--=---=----=---~-= --

0.68 /"
o

0.66
o
.m
0.64
o I

o
Q.
0.62
O
0.60 , , , I , , , , L .... [ .... I,,,,I~,,I,,~,I

-5 -4 -3 -2 -1 0 1 2
Applied Voltage (V)

E 45 ,,,, .... i .... i .... , .... , .... , .... ,

E
40
\\. '
0
35

~5
0
30 ,,,L~,,,I,,L,I .... I .... I .... I .... I,,1
-5 -4 -3 -2 -1 0 1 2
Applied Voltage (V)
Fig. 7. Calculated cell efficiency, open-circuit voltage, a n d short-circuit current o f the conventional cell as a
function o f applied gate voltage a n d interface trap density at the rear S i / S i O 2 interface. O b s e r v e that the
calculated c u r v e o f " B i t = 6.5 X 109 c m - 2 e V - i ,, does not d e p e n d o n the gate voltage applied.
14. Ohtsuka et al. / Solar Energy Materials and Solar Cells 44 (1996) 79-98 89

holes, to form the accumulation layer and saturate the solar cell characteristics, the
surface concentration of hole must be about 10 9 times that of electron. So a large gate
voltage must be applied to bend the band near the surface in order to get such an
enormous electron-hole concentration gap. It is for this reason that the solar cells have
the asymmetric characteristics for zero bias as shown in Fig. 4.
The differences between the maximum and the minimum of both efficiency and
short-circuit current cells are larger for back-contact than they are for conventional cells.
This is partly because the density of photo-generation carders on front side, where their
concentration is higher, is controlled by the field effect in back-contact cells. We think
that such field-effect passivation of the front side is very effective for improving
efficiency.

3.2. Influence o f the interface trap density

Fig. 7 shows the simulation results for conventional cells as a function of applied gate
voltage and interface trap density parameters. Fig. 8 shows the simulation results for
back-contact cells. In both figures the dashed lines are for a Si/SiO 2 interface trap
density of 6.5 X 101° c m -2 e V -1, which is a value we obtained experimentally by MOS
C - V measurement. The other lines indicate the results calculated for cells which have
0. l and times the higher-interface trap density of the baseline cell.
As shown in the figures, the efficiency, open-circuit voltage, and short-circuit current
reach their minima at around 0 V in the conventional cells and at around - 0 . 5 V in the
back-contact cells. As can be seen, the higher the interface trap density is, the lower
these minima are. This tendency is remarkably distinct in the results for the back-contact
cells. As we mentioned above, this is because, in back-contact ceils, the density of the
front-side (high concentration) photo-generation carriers is controlled by bending the
band edge. However, all of the characteristics reach a maximum at more than 1.0 V.
This is to say, however high the interface trap density is, the surface recombination rate
is sure to decrease so long as the number of pairs of carder which can recombine
decreases.
In conclusion, for a p-type 2 12- cm wafer solar cell, applying a positive gate voltage
of more than 1.0 V, or inducing a positive charge of about 1.0 × 1011 c m -2 improves
the cell characteristics, and can make them equivalent to those for an effective surface
recombination velocity of 0 c m / s , even if the oxide layer has an interface trap density
on the order of l 0 n c m -2 e V -1. On the other hand, there would be a risk that minor
irregularities of surface potential might cause very low performance in such a solar cell,
especially a back-contact cell which did not have good passivated oxide layers (Dit > 1011
cm-2 e V - 1). Such susceptibility to fluctuations could be, however, controlled by means
of gate bias, etc.

3.3. Influence o f substrate resistivity

Fig. 9 shows the calculated results for surface carrier concentration of 2.0 f~. cm and
0.2 ~ . cm wafer conventional cells as a function of applied gate voltage. As shown
with arrows in the lower graph of Fig. 10, the points of gate bias for both minimum and
90 H. Ohtsuka et al. / Solar Energy Materials and Solar Cells 44 (1996) 79-98

26
.................... 6,.5 x.!O 9 / .......
," /
24
6.5 x 1010 ,

~,
¢,,}
22
c
(
"6 20 \\ i
6.5X 101~ ,'
UJ i
". i
18 \
\
i
!6.5 x 1012

'-J ! t
-5 -4 -3 -2 -1 0 1 2 3
Applied Voltage (V)

0.72 .... ~ .... ~ .... i .... I .... i .... k .... ~ ....

0.70
/
0.68
O
> /

0.66
P 0.64
0
o 0.62
0
0.60 L k , l t , l l k t l I .... I .... i .... I ....

-5 -4 -3 -2 -1 0 1 2 3
Applied Voltage (V)
E 45 T~-is ]~ ~ , , i r ~ ~ Jl , , , , IE , , , I ~ ,',, II,T~ I 'E ~

E
40
.=
t.

i
O

35 \ f
o
O i '
\

O
p.
30 i i J [ l l h ~ l ~ l l l

-5 -4 -3 -2 -1 0 1 2 3
Applied Voltage (V)
Fig. 8. Calculated cell efficiency, open-circuit voltage, and short-circuit current of the back-contact cell as a
function of the applied gate voltage and the interface trap density at the rear S i / S i O 2 interface, For all
characteristics the dependence on Dit is much stronger for back-contact cells than it is for conventional cells.
14. Ohtsuka et al. / Solar Energy Materials and Solar Cells 44 (1996) 79-98 91

~ " 1 018
, I' '~ '1' ~ ~' I ~ ~ ' ~ ,, ~ ~ t , , ~ I , ~, ~
E under 1 sun (AM1.5)
e- 1 017 .... 2 ~cm / electron

1 016 " ' \ acm / ~ectto

tO
=01501 ~",,,,,,, -" / " " " 0.2 ~cm ,/hol:
"E 1 0 TM

° 1 013
_/
- - -
\,"
0
,~ ~2 ~cm / hole
1 012 I,, I .... I .... I .... "i ,,, I I I I~

-' 0 1 2 3 4 5
Applied Voltage (V)
Fig. 9. Surface carrier concentration plot of a conventional cell at short-circuit condition as a function of
applied gate voltage. The solid lines represent the hole and electron surface concentrations of a 2.0 l ) . c m
substrate and the dashed lines represent those of a 0.2 l ) . c m substrate. Observe that less gate voltage is
required to form an inversion layer in the 2.0 ~ . c m substrate cell than in the 0.2 ~ . c m substrate.

saturated short-circuit current move as substrate resistivity becomes higher. These


phenomena are similar to the behaviour of MOS-diode threshold voltage, which is
higher as wafer resistivity becomes low. In fact, as shown in Fig. 9, for a 2.0 11 • cm
wafer the applied voltage at which the surface concentration of electrons exceeds that of
holes is about 0.6 V whereas the crossover point for a 0.2 l ] . cm wafer is about 3.0 V.
In fact, it is undesirable to apply high voltage to oxide layers, because that can cause
degradation of the layer and result in short circuits. Also, there is a limit on the density
of the fixed oxide charge that can be induced. Therefore, in applying surface field-effect
passivation to solar cells, it is desirable to use higher resistivity wafers because of their
lower required gate voltage.
Fig. 10 shows the calculated results for conventional cells as a function of applied
gate voltage and wafer resistivity parameters. Fig. 11 shows the results for back-contact
cells. In spite of the above statement, however, the simulation results for conventional
cells with several resistivity indicate that the efficiency of 10 12. cm wafer cells stays
lowest for any applied gate voltage although the open-circuit voltage and the short-cir-
cuit current are higher. This is because the fill factor of higher-resistivity cells is low,
due to current crowding around the rear contacts. This problem will be mitigated by
narrowing the rear contact pitch.
Contrary to the conclusion in Section 3.1, the performance of a back-contact cell
constructed on a 0.2 1). cm substrate cannot be elevated to that of "S = 0 c m / s " for
any positive or negative applied gate voltage. The quantum efficiency of a back-contact
cell is strongly dependent on the carrier diffusion length. The worst is due to the fact
92 H. Ohtsuka et al. // Solar Energy Materials and Solar Cells 44 (1996) 79-98

24 '''I'''"I''''I''''I''''I''''I''''I'''

0.5 acm /7 -~-----........


23

o -0.2- acr ........ :-:: . . . . . - --

._~
•~- 22 ..... 10 f~cm ,/ . . . . . . .
W

21 ,,,,l,,~,l,,,,l,~t,l ,,I,,,,I,,,,I,ii
-5 -4 -3 -2 -1 0 1 2
Applied Voltage (V)

0.705 '''1 .... I''"'1 .... I''''1''''1''''1 r


>

0.700
o
>
j/-¸

=,! O.695
.--'_
f
.o

,L 0.690
o
Q.
0
0.685 ~,,I,,.,I,.,,I,,,,I,,,,I,,.,I,,,,I,
-5 -4 -3 -2 -1 0 l 2
Applied Voltage (V)

I=
42

E 41

i
=1 40
o
0~

39
oT
1::
0 38
-5 -4 -3 -2 -1 0 1 2 3
Applied Voltage (V)
Fig. 10. Calculated cell efficiency, open-circuit voltage, and short-circuit current of a conventional cell as a
function of applied gate voltage and substrate resistivity. The arrows show the directions in which the points of
gate bias for both minimum and saturated short-circuit current move as the resistivity becomes higher.
H. Ohtsuka et al. / Solar Energy Materials and Solar Cells 44 (1996) 79-98 93

26
."'''l''~rl''''l''''l''''l'~''l''''l'''r t

25 10 ~ c m . / ~ .....

24

o 23
e- /
/

22
iii ..... -- ~ 0 . 2 f 2 c m /
21 ~---. ,

20 , , , n . . . . J . . . . n. . . . n. . . . J , , , I,,,,I,,,
-5 -4 -3 -2 -1 0 I 2 3
Applied Voltage IV)

,.., 0 . 7 0 5 ~''1''''1''''1''''1''''1''''1''''1''''
>

~ 0.700 / fl-

.--.. ~ //

/
,~- 0 . 6 9 5
- .... j//
P
~5
~: 0 . 6 9 0
Q.
0
,,,,I,,,,I,,,,I,,,,I,,,,I,,,,I,,,,I,,,
0.685
-5 -4 -3 -2 -1 0 1 2
Applied Voltage (V)
A
44 '''1 .... I .... I .... I''''1''''1''''1'''
<
f:42_-

~ 40 /

/i /
38

P
~ 36

~ 34
~ -5 -4 -3 -2 -1 0 1 2
Applied Voltage (V)
Fig. 11. Calculated cell efficiency, open-circuit voltage, and short-circuit current of the back-contact cell as a
function of applied gate voltage and substrate resistivity. The worst performance of the 0.2 ~ .cm substrate is
because it has the shortest lifetime of the three. In the back-contact cells lifetime influences cell performance
strongly since all contacts are arranged on the rear surface.
94 H. Ohtsuka et al. / Solar Energy Materials and Solar Cells 44 (1996) 79-98

that it had the lowest lifetime (cf. Table 1). In this case, the reduction of effective
surface recombination velocity induced by field-effect surface passivation is not com-
pletely able to compensate for the cell performance, because of the high recombination
rate in the bulk.

3.4. Influence of substrate thickness

Fig. 12 shows the simulation result for conventional cells as a function of applied
gate voltage and wafer thickness. Fig. 13 shows the simulation result for back-contact
cells. It is assumed that the front surface reflectance is 0% and the light-trapping rate is
100% (i.e., that the light incident into a cell reflects completely at the inner front and
rear surfaces). The reason for this assumption is that we want the total number of
absorbed photons to be constant and independent of cell thickness.
In the results for conventional cells, a dependency on cell thickness is observed for
all the characteristics, especially open-circuit voltage. The short-circuit current saturates
at about 44.5 m A / c m 2 for all thicknesses but the open-circuit voltage is seen to be
continue to increase as the cell becomes thinner. The reason for the saturation of the
short-circuit current is that it is approaching the total photo-generation current, as the
surface recombination rate decreases in a high lifetime substrate. On the other hand, the
non-saturating increase of the open-circuit voltage is due to the much lower surface
recombination rates at both front and rear surfaces. If we passivate by the field effect on
the rear surface, both the front and rear surfaces of a conventional cell have a much
lower recombination rate, because the front surface is originally "passivated" anyway
because of n-type layer. In this case, the thinner the substrate, the lower the diode
saturation-current density (the higher the open-circuit voltage).
In the results for back-contact cells, a dependency on applied voltage was observed
for all the characteristics. However, open-circuit voltage does not show as strong a
dependency on the cell thickness as it does in a conventional cell. In the case of negative
applied gate voltage, the efficiency and the short-circuit current decrease as the cell
become thicker, whereas they saturate and have no dependence on cell thickness in the
case of applying a positive applied gate voltage. When positive gate voltage is applied,
the surface recombination rate is very low. Therefore, in a high-lifetime substrate the
short-circuit current cannot increase with reduced thickness, because it nears its maxi-
mum value anyway. In the case where a negative gate voltage is applied, the surface
recombination rate is higher, so the short-circuit current depends on the cell thickness,
because surface cartier concentration as a function of cell thickness varies with the
surface recombination rate. Owing to the assumption of 100% light trapping, surface
carrier concentration shifts to the high injection condition as the cell becomes thinner, so
the effective surface recombination velocity becomes low [11]. As a result, short-circuit

Fig. 12. Calculated cell efficiency, open-circuit voltage, and short-circuit current of a conventional cell as a
function of applied gate voltage and cell thickness. 100% light trapping was assumed. Note especially that the
dependence of the open-circuit voltage on cell thickness is remarkable. The thinner the cell, the higher the cell
performance,
H. Ohtsuka et al. / Solar Energy Materials and Solar Cells 44 (1996) 79-98 95

27.0 ,,~,t,,,,i,,, ,] ,~,,E,,,,I,,~,I,,.,I_, L,


/
/
26.5 ; f

" - - 25.0
--._ 50 ~ //i
i

e-
/ J ............._.-
25.5 _ _ f - -

_._ 250 t.tm // ........ -


uJ
25.0 ------ j" ,'
-. , .350 Nm ...... ,,
. . . . . . . . . . _. _ . . . . . • ,"

24.5 ,,I .... J .... ~.... ~.... ~.... i .... ~....

-5 -4 -3 -2 -1 0 1 2 3
Applied Voltage (V)
0.73

0.72

t
t
"~ 0.71 J

= 0.70
.o
o
~. 0 . 6 9
........................... - .... .---'" ............
0
0.68 ,,I .... i .... i .... i .... i .... ~.... i,,,

-5.0-4.0-3.0-2.0-1.0 0.0 1.0 2.0 3.0


Applied Voltage (V)
E 45.0

E
46448. . . . . . . . . . . . . L
. .... . ..... ... . ... . ... . ... . . .. . ..

_~ ~ , , ~ ~ /// .

o ........ --.'<. ,~/ ............ _-

°-- 44.4 ~ ....... ~_,,_ <--.t/,,'


P
c,, 44.2 "'-, " - v - ' ,,

o
e,,, i .' . ..,... . \ /,'
44.0 ,,,' .... ' . . . . "r','i-,-i'l",, p. . . . I,,,
-5 -4 -3 -2 -1 0 1 2 3
Applied Voltage (V)
96 H. Ohtsuka et al. / Solar Energy Materials and Solar Cells 44 (1996) 79-98

27.0
'-iii
i
i, .... , .... , .... , .... , ....

26.5
-_ j5o~ S -

>, 26.0
F ' ,./
~-
- / f . -. - " - . \\ ~ ,-,.,. k'
/,.'
._~ r., ". ~,\"-- fl'
o 25.5 L'O0 pm ", . . . . - ~ 2 5 0 p,m

UJ

25.0
' ~00 gm
24.5
-5 -4 -3 -2 -1 0 1 2 3
Applied Voltage (V)

0.705 .... / .... I .... I .... I .... I .... I .... I ....


>
v

0
>
0.700 -'~ ~- ,/'

•,.. ~'~.._....J ,

0
0.695 ,,, I ...... i . . . . ~. . . . ~. . . . i . . . . i . . . .
-5 -4 -3 -2 -1 0 1 2 3
Applied Voltage (V)

E 45.0 .... I .... I .... I .... I .... I .... I .... I ....

E 44.5

..... "\
~ 44.0 -.......~\ \ \~ //,,; I

\ /:
43.5 \ /,
,,q
•, j

43.0
/

O
" 42.5
-5 -4 -3 -2 -1 0 1 2 3
Applied Voltage (V)
H. Ohtsuka et al. / Solar Energy Materials and Solar Cells 44 (1996) 79-98 97

current and efficiency show high values, since the ratio of surface recombination to total
recombination decreases as the cell becomes thinner.
From what we have said above, it can be seen that in the conventional cells the
combination of a thin substrate with the perfect light trapping and a complete field-effect
surface passivation on the oxidized rear surface leads high solar cell performance. On
the other hand, it was found that a back-contact cell passivated by the field effect have
the same performance as a thinner cell, even without thinning its substrate.

4. Conclusions
We have analyzed the characteristics of high-efficiency crystalline silicon conven-
tional cells and back-contact cells using 2D numerical simulations, in order to investi-
gate the influence of surface potential on solar cell electric characteristics and the
effectiveness of the field-effect surface passivation method. In the simulations, we paid
attention to the changes of the potential on the rear oxidized surfaces in the conventional
cell and on the front oxidized surfaces in the back-contact cell.
The simulation was based on physical parameters measurement from actual materials
and showed that field-effect surface passivation yields cell characteristics equivalent to
those of a cell with an effective surface recombination velocity of 0 c m / s . That is to
say, even if the cell has a poor S i / S i O 2 interface (Dit > 1011 cm -2 e V - 1), performance
can be recovered by changing the surface potential. On the other hand, in case where no
field-effect surface passivation is applied, there is a risk in a solar cell that the
short-circuit current may drop rapidly and the efficiency grow worse due to small
irregularity in the surface potential. The simulation also showed that the use of a higher
resistivity wafer and, especially in a p-type substrate, the formation of inversion layers
causes the field-effect surface passivation to work with the fullest effect.
As the result of super-low surface recombination, there is a realistic hope that a
back-contact cell may be able to achieve more than 25% efficiency under AM1.5 global,
one-sun illumination. Furthermore, in conventional cells, the combination of thin
substrates with perfect light trapping and complete field-effect surface passivation on the
oxidized surface promises higher solar cell performance.

Acknowledgements
This research was supported in part by the New Energy and Industrial Technology
Development Organization as a part of the Sunshine Project under the Ministry of
International Trade and Industry. The authors wish to thank Dr. Y. Ohkura of Hitachi
Central Research Laboratory for his guidance in using CADDETH.

Fig. 13. Calculated cell efficiency, open-circuit voltage, and short-circuit current of a back-contact cell as a
function of applied gate voltage and cell thickness. 100% light trapping was assumed. Note that all
characteristics are independent of cell thickness with a the positive gate bias. This suggests a need for
rear-surface field-effect passivation.
98 H. Ohtsuka et al. / Solar Energy Materials and Solar Cells 44 (1996) 79-98

References

[1] J. Zhao, A. Wang and M. A. Green, Prog. Photovoltaics 2 (1994) 227.


[2] R. A. Sinton, Y. Kwark, J. Y. Gan and R. M. Swanson, IEEE Electron. Dev. Lett. 7 (1986) 567.
[3] A. G. Aberle, S. Glunz and W. Warta, Sol. Energy Mater. Sol. Cells 29 (1993) 175.
[4] A. G. Abede, S. Glunz and W. Warta, J. Appl. Phys. 71 (1992) 4422.
[5] R. B. M. Girisch R. P. Mertens and R. F. De Keersmaecker, IEEE Trans. Electron Dev. 35 (1988) 203.
[6] T. Warabisako, T. Uematsu, K. Kanda and S. Iida, Proc. 5th Int. Photovoltaic Science and Engineering
Conf., Kyoto, 1990, p. 583.
[7] Z. Chen, K. Yasutake, A. Doolittle and A. Rohatgi, Appl. Phys. Lett. 63 (1993) 2117,
[8] K. Yasutake, Z. Chen, S. K. Pang and A. Rohatgi, J. Appl. Phys. 75 (1994) 2048.
[9] T. Toyabe, H. Masuda, Y. Aoki, H. Shukuri and T. Hagiwara, IEEE Trans. Computer-Aided Design 4
(1985) 482.
[10] H. Ohtsuka, Y. Ohkura, T. Uematsu and T. Warabisako, Prog. Photovoltaics 2 (1994) 275.
[11] W. D. Eades and R. M. Swanson, J. Appl. Phys. 58 (1985) 4267.

Das könnte Ihnen auch gefallen