You are on page 1of 60

Journal

of
the
Korean
Physical
Society,
Vol.
39,
No.
2,
August
2001,
pp.
369373

Development
of
High-Eciency
Silicon
Solar
Cells
for
Commercialization

S.
H.
Lee*
National
Research
Laboratory
for
Silicon
Photovoltaics,
Corporate
R&D
Center,
Samsung
SDI
Co.,
Ltd.,
Yongin
449-712

(Received
2
May
2001)

High-eciency
silicon
solar
cells
have
potential
applications
for
mobile
electronics
and
electrical
vehicles.
An
overview
of
the
high-eciency
crystalline
silicon
solar
cells
developed
at
Samsung
SDI
is
reported.
High-eciency
silicon
solar
cells
with
eciencies
over
20
%
are
too
expensive
to
be
commercialized.
The
National
Reaearch
Laboratory
for
Silicon
Photovoltaics
at
Samsung
SDI
aims
at
the
development
of
high-eciency
single-crystalline
silicon
solar
cells.
Several
structures
with
high
eciencies
have
been
developed
using
a
low-cost
processes
at
Samsung
SDI
for
commercialization.
The
PERC
(passivated
emitter,
rear
cell)
type
is
a
good
candidate
for
commercialization
while
the
PERL
(passivated
emitter,
rear
locally
diused
cell)
type
exhibits
the
highest
conversion
eciency.
Conversion
eciencies
of
19.0
%,
20.43
%,
and
20.5
%
for
the
PESC
(passivated
emitter
solar
cell),
PERC,
and
PERL
types,
respectively,
were
con
rmed
at
Fraunhofer
Institute
for
Solar
Energy
Systems.
A
mechanical
scriber
and
a
Ni/Cu
contact
system
have
been
used
to
fabricate
low-cost
PERCs
suitable
for
mass
production.

I.
INTRODUCTION
In
2000,
80
%
of
the
total
PV(photovoltaics)
market
was
due
to
crystalline
silicon
solar
cells.
It
is
expected
that
production
of
single-crystal
silicon
modules
will
reach
eciency
of
22
%
by
2010
[1].
The
eciency
of
crystalline
silicon
solar
cells
has
greatly
improved
over
the
last
few
years.
With
a
conversion
eciency
of
over
24
%
[2],
the
eciency
of
silicon
solar
cells
is
approaching
the
theoretical
limit
of
a
silicon
homo-junction.
However,
the
production
of
crystalline
silicon
solar
cells
with
such
a
high
eciency
is
very
small
compared
to
that
of
conventional
silicon
solar
cells
because
of
the
complicated
fabrication
processes
with
cost-intensive
process
steps
and
because
of
their
small
areas.
In
order
to
commercialize
high-eciency
solar
cells
for
consumer
electronics,
it
is
essential
to
produce
cells
with
low
cost
and
simple
fabrication
processes.
The
other
factors
for
solar
cells
to
be
used
as
power
sources
for
mobile
electronics
are
conversion
eciency
and
appearance.
Several
approaches
have
been
tried
to
achieve
these
goals,
but
eciency
was
too
low
to
access
the
market.

Although
crystalline
silicon
solar
cells
were
developed
almost
50
years
ago,
the
PESC
cell
was
the
rst
cell
to
exceed
an
eciency
of
20%
[3],
which
was
done
by
passivating
this
silicon
surface
using
silicon
dioxide.
In
rear
point
contact
solar
cell
[4]
and
the
PERC
cell
[5],
the
front
and
the
rear
surfaces
were
passivated
by
silicon

E-mail:
euikyu@samsung.co.kr

dioxide
to
improve
the
performance
of
the
silicon
solar
cells.
The
PERC
structure
is
a
well-known
cell
concept
that
combines
a
fairly
simple
process
sequence
with
high
eciencies.

In
this
work,
we
report
the
recent
results
achieved
by
our
group
for
single-crystalline
solar
cells.
Low-cost,
simpli
ed
processes
for
commercialization
of
high-eciency
solar
cells
will
be
presented.

II.
EXPERIMENTAL
The
sizes
of
the
cells
designed
for
high-eciency
crystalline
silicon
solar
cells
were
2
cm
×
2
cm
and
6.7
cm
×
6.7
cm
on
a
4-inch-diameter
wafer.
FZ
wafers
with
resistivities
of
0.5
and
2
cm
were
processed
to
fabricate
PESC,
PERC
and
PERL
type
cells.
We
used
0.5
cm
to
fabricate
the
PESC,
2
cm
for
the
PERL
and
both
resistivities
for
the
PERC.
Each
processing
sequence
is
described
in
Table
1.
High-eciency
silicon
solar
cells
with
PESC,
PERC
and
PERL
structures
were
fabricated
using
a
photolithography
process.
The
fronts
of
cells
were
passivated
using
silicon
dioxide
which
was
formed
on
silicon
at
1050
oC
in
Trans-LC
and
oxygen.
The
oxide
lm
acted
as
a
diusion
barrier
during
the
cell
process
and
as
an
ection
antire
coating
after
the
fabrication.

The
schematic
diagrams
of
the
PESC,
the
PERC,
and
the
PERL
type
cells
are
compared
in
Fig.
1.
The
PESC
cell
adapts
a
back
surface
eld
by
sintering
aluminum
at
high
temperature.
The
rear
surface
of
the
PERC
cell
is

-369
-370-Journal
of
the
Korean
Physical
Society,
Vol.
39,
No.
2,
August
2001
Table
1.
Processing
sequences
of
the
PESC,
PERC,
and
PERL
cells.

Steps
for
PESC
Steps
for
PERC
Steps
for
PERL
Wafer
selection
Wafer
selection
Wafer
selection
Oxide
growth
Oxide
growth
Oxide
growth
Photolithography
Photolithography
Photolithography
Anisotropic
etching
Anisotropic
etching
Anisotropic
etching
Oxide
growth
Oxide
growth
Oxide
growth
Photolithography
Photolithography
Photolithography
P
heavy
diusion
P
heavy
diusion
Boron
diusion
P
light
diusion
P
light
diusion
Oxide
growth
Oxide
growth
Oxide
growth
Photolithography
Al
evaporation
Photolithography(front)
P
heavy
diusion
Al
alloy
Photolithography(rear)
P
light
diusion
Photolithography
Top
contact
evap.
Oxide
growth
Top
contact
evap.
Rear
contact
evap.
Photolithography(front)
Ag
plating
Ag
plating
Photolithography(rear)
Annealing
Annealing
Top
contact
evap.
Rear
contact
evap.
Ag
plating
H2
Annealing

passivated
by
silicon
dioxide,
except
for
the
metal
contact
which
is
formed
by
aluminum
after
opening
the
silicon
dioxide.
In
the
PERL
cells,
boron
is
diused
under
the
rear
metal
contact
in
order
to
reduce
the
recombination
rate
of
minority
carriers
at
the
metal-silicon
interface.

We
developed
low-cost
processes
without
signi
cant
loss
of
eciency
by
decreasing
the
number
of
photo
masks,
simplifying
the
production
process,
and
applying
a
low-cost
metallization
process.
A
mechanical
scribing
process
replaced
the
costly
photolithography
process
to
pattern
the
rear
contact
structure
of
the
PERC
cells.
Ni/Cu
metallization
was
applied
to
the
front
of
PERC
cells,
instead
of
expensive
Ti/Pd/Ag
metallization.
The
performances
of
the
solar
cells
were
con
rmed
independently
by
the
Fraunhofer
Institute
for
Solar
Energy
Systems.

III.
RESULTS
AND
DISCUSSION
The
PESC
cell
using
a
0.5-
cm
FZ
wafer
showed
an
eciency
of
19
%,
which
was
con
rmed
by
the
Fraunhofer
Institute
for
Solar
Energy
Systems.
The
area
of
the
cells
used
in
this
study
was
6.7
×
6.7
cm2,
which
is
the
maximum
size
for
a
4-inch
FZ
wafer.
The
resistivity

is
low
in
the
PESC
structure
because
the
recombination
velocity
of
the
aluminum-sintered
rear
surface
is
high.
The
I-V
characteristics
of
PESC
cells
are
shown
in
Fig.

2.
A
large
shading
loss
due
to
the
front
metal
grid
is
responsible
for
the
low
short-circuit
current
density
of
the
solar
cell
with
a
large
area.
The
shading
loss
due
to
the
front
metal
grid
with
a
65-m
width
was
calculated
to
be
11.5
%
while
that
for
a
grid
with
a
20-m
width
is
3.5
%.
All
the
cells
fabricated
for
this
study
had
a
front
metal
nger
with
a
65-m
width.
The
PESC
cell
is
expected
to
have
an
eciency
of
over
20
%
when
the
area
of
the
cell
is
decreased
to
2
cm
×
2
cm
due
to
the
increased
short-circuit
current
density.
Silicon
dioxide
with
a
110-nm
thickness
on
the
front
surface
acts
as
an
ection
antire
layer
in
addition
to
a
passivation
layer.
The
aluminum
sintering
generates
a
highly
defective
layer
at
the
rear
surface
of
the
PESC
cell.
This
limits
the
open-circuit
voltage
to
below
660
mV
for
PESC
cells
on
a
0.5-
cm
substrate.
Except
for
the
rear
contact
structure,
PERC
cells
are
the
same
as
PESC
cell.
The
aluminum-sintered
layer
was
replaced
by
a
passivating
silicon
dioxide
layer.
Table
2
shows
the
cell
parameters
of
the
PERC
cell
for
various
substrate
resistivities.
The
passivation
by
silicon
dioxide
increased
the
open-
circuit
voltage
up
to
661.9
mV.
A
resistivity
higher
than

0.5
cm
gives
a
higher
eciency
in
a
cell
with
a
well-
Table
2.
Performances
of
PERC
cells
fabricated
on
dierent
substrates
(AM1.5
spectrum,
100
mW/cm2,
25
oC,
45
cm2
cell
area).

Resistivity
of
substrate
Voc
(mV)
Jsc
(mA/cm2)
FF
(%)
Eeff
(%)
2
cm
589
38.3
76.9
17.4
0.5
cm
661.9
36.2
80.6
19.3
Development
of
High-Eciency
Silicon
Solar
Cells
for
Commercialization
S.
H.
Lee
-371

Fig.
1.
Schematic
diagrams
of
high
eciency
cells:
(a)
PESC,
(b)
PERC,
and
(c)
PERL.

passivated
surface
as
long
as
the
metal
contact
is
ohmic.
However,
the
aluminum
contact
to
a
substrate
with
a
resistivity
higher
than
0.5
cm
forms
a
Schottky
barrier
and
reduces
the
open-circuit
voltage
and
the
ll
factor
since
the
depletion
width
of
a
Schottky
junction
is
nar-
Fig.2.Current-voltagecharacteristicsofPESCcellsmea-
suredattheFraunhoferInstituteforSolarEnergySystems(AM1.5spectrum,100mW/cm2,25oC,
44.96cm2cellarea).

Fig.3.Current-voltagecharacteristicsoftheSiO2-coatedPERLcellmeasuredattheFraunho
ferInstituteofSolarEn-
ergySystems(AM1.5spectrum,100mW/cm2,25oC,45.75

cm
2
cell
area).

row
[6].
The
formation
of
a
Schottky
junction
at
the
rear
contact
of
the
substrate
with
a
1.5-
cm
resitivity
is
responsible
for
low
open-circuit
voltage
in
Table
2.
The
rear
structure
of
the
PERC
cell
ects
re
light
eectively
while
the
Al-alloyed
rear
contact
of
the
PESC
cell
has
a
limitations
on
increased
ectivity
re
[7].
As
the
thickness
of
the
cell
becomes
thinner,
the
advantages
of
enhanced
passivation
and
ectivity
re
of
the
rear
structure
of
the
PERC
cell
become
signi
cant.
The
low
open-circuit
voltage
of
the
higher
eciency
cell
in
a
Table
2
is
attributed
to
its
non-optimized
rear
contact
design.
The
spacing
between
contact
holes
needs
to
be
longer
than
the
diusion

Table
3.
Output
parameters
of
PERC
cells
with
three
types
of
rear
contact
patterns
on
0.5-
cm,
p-type
FZ
wafers
(AM1.5
spectrum,
100
mW/cm2,
25
oC,
4
cm2
cell
area).

Rear
pattern
type
Rear
contact
area
Series
resistance
Jsc
Voc
FF
Eeff
fraction
(%)
(
)
(mA/cm2)
(mV)
(%)
(%)

Dot
pattern
with
photolithography
0.8
0.125
37.2
666.3
80.7
19.98
Dashed
line
pattern
with
mechanical
scriber
0.9
0.213
37.8
646
79.5
19.42
Line
pattern
with
mechanical
scriber
2
0.346
39
655
71
18.27
-372-Journal
of
the
Korean
Physical
Society,
Vol.
39,
No.
2,
August
2001

Fig.
4.
Schematic
diagram
of
the
MS-PERC
cell
with
a
rear
structure
patterned
using
a
mechanical
scriber.

length
of
the
minority
carrier
to
reduce
the
recombination
of
light-generated
minority
carriers.
However,
the
contact
spacing
was
designed
to
be
250
m
with
a
contact
window
width
of
30
m.
An
optimum
rear
contact
design
should
improve
the
cell
eciency
up
to
20
%
for
a
cell
with
an
area
of
45
cm2
.

Figure
3
shows
the
performance
of
the
PERL
cell
in
which
boron
was
heavily
doped
under
the
rear
metal
contact.
The
rear
contact
design
was
the
same
as
that
of
the
PERC
cell.
The
sheet
resistance
of
the
layer
diused
with
boron
ranged
from
10
to
20
W/2.
Heavy
doping
under
the
rear
metal
reduces
the
recombination
velocity
by
blocking
minority
carrier
diusion
and
the
contact
resistance,
which
enhances
open-circuit
voltage.

With
110-nm-thick
SiO2
as
an
ection
antire
coating,
an
eciency
of
20.5
%
was
demonstrated
at
the
Fraunhofer
Institute
for
Solar
Energy
Systems
for
a
cell
with
an
area
of
45
cm2
.
With
an
ection
antire
coating
of
ZnS/MgF2,
an
eciency
of
over
21
%
becomes
feasible
since
a
3
%
increase
in
the
short-circuit
current
can
be
expected.
If
the
front
metallization
and
the
double
antire
ection
coating
are
optimized,
the
ll
factor
and
the
short-circuit
current
density
can
be
increased
up
to
80
%
and
40
mA/cm2,
respectively.
Even
though
the
eciency
of
the
PERL
cell
is
higher
than
PERC
cell,
the
boron
diffusion
process
is
used
for
rear
metal
contact
formation
to
reduce
recombination
under
the
rear
metal
contact.

For
commercialization
of
high-eciency
solar
cells,
the
Fig.
5.
Schematic
diagrams
of
the
conventional
Ti/Pd/Ag
contact
system
and
the
new
Ni/Cu
contact
system
for
the
PERC
cell.

PERC
cell
appears
to
have
a
more
competitive
structure
because
it
can
provide
a
high
eciency
of
over
20
%
without
the
boron
process.
Moreover,
the
rear
contact
design
can
accept
various
opening
methods
other
than
the
expensive
photolithography
process
because
the
spacing
and
the
feature
size
are
larger
than
those
of
the
PERL
cell.

The
new
method
for
patterning
the
rear
passivation
layers
of
the
PERC
with
a
mechanical
scriber
was
adapted
at
Samsung
SDI
to
replace
the
photolithography
process.
The
mechanical
scriber
has
three
scribing
tips
that
enables
it
to
scribe
3
wafers
simultaneously.
The
scribing
tip
was
made
of
arti
cial
diamond,
and
the
minimum
scribing
width
was
10
µ
in
our
study.
Figure
4
shows
the
MS-PERC
(mechanically
scribed
PERC)
structure
with
the
rear
pattern
opened
by
using
the
mechanical
scriber
developed
at
Samsung
SDI.
Various
kinds
of
rear
patterns
can
be
easily
scribed,
and
three
types
of
rear
contact
patterns
were
processed
in
this
study.
The
results
for
the
MS-PERC
and
the
conventional
PERC
cell
are
shown
in
Table
3.
The
eciency
of
the
MS-PERC
cell
is
comparable
to
that
of
the
conventional
PERC
cell.
The
lower
open-circuit
voltage
of
the
MS-PERC
cell
is
attributed
to
homogeneous
emitter
diffusion
which
eliminated
one
photolighography
step
and
one
diusion
step
from
the
process
for
a
conventional
PERC
with
a
selective
emitter.

An
evaporated
Ti/Pd/Ag
contact
system
is
most
widely
used
to
make
high-eciency
silicon
solar
cells.
However,
the
system
is
not
cost
eective
due
to
expensive
materials
and
vacuum
techniques.
Ni/Cu
metalliza-
Table
4.
Output
parameters
of
high-eciency
PERC
cells
with
Ni/Cu
and
Ti/Pd/Ag
contact
systems
(AM1.5
spectrum,
100
mW/cm2,
25
oC,
4
cm2
cell
area).
Front
metal
contact
Jsc
(mA/cm2)
Voc
(mV)
FF
(%)
Eeff
(%)
Rseries
(
)
Rshunt
(k
)
Electroless
plated
Ni/Cu
38.1
664.4
0.798
20.19
0.163
547
Vacuum
evaporated
Ti/Pd/Ag
37.7
662.9
0.805
20.13
0.153
1790
Development
of
High-Eciency
Silicon
Solar
Cells
for
Commercialization
S.
H.
Lee
-373

tion
was
applied,
as
shown
in
Fig.
5,
to
the
front
contact
of
a
PERC
to
oer
a
relatively
inexpensive
electrical
contact.
The
advantages
of
Ni/Cu
over
the
standard
Ti/Pd/Ag
are
cheap
material
cost
and
easy
deposition
by
using
a
plating
method.
The
contact
resistivity
of
Ni/Si
was
3.5
×
10..5
cm2
,
which
is
even
lower
than
that
of
Ag/Pd/Ti/Si
system
(7.3
×
10..5
cm2).
A
contact
resistivity
less
than
1
×
10..3
cm2
was
found
to
give
suciently
low
power
loss
for
one-sun
applications
[8].
It
was,
therefore,
found
that
NiSi
was
suitable
for
high-
eciency
solar
cell
applications.
Nickel
was
deposited
by
electroless
plating
to
form
the
front
contact
for
the
PERC
cells
and
was
annealed
at
400
oC
for
30
min
in
forming
gas
(4
%
H2
+
96
%
Ar).
Copper
was
electroplated
on
top
of
the
nickel
contact.
The
performances
of
the
PERC
cells
with
Ni/Cu
contact
systems
were
investigated
and
are
summarized
in
Table
4.
The
Ni/Cu
contact
was
found
to
be
well
suited
for
high-eciency
solar
cells
and
was
successfully
formed
by
using
electroless
plating
and
electroplating,
which
are
more
cost
eective
than
vacuum
evaporation.

IV.
CONCLUSION
We
compared
the
performance
of
high-eciency
silicon
solar
cells
developed
recently
at
Samsung
SDI.
Using
a
high-eciency
process
scheme,
we
showed
that
solar
cells
could
be
produced
with
an
eciency
of
20.5
%
for
the
PERL
cell,
19
%
for
the
PESC
cell
and
19.3
%
for
the
PERC
cell
without
double
ection
antire
coating
and
optimization
of
the
front
metallization
in
PERL
structure.
It
is
thought
that
eciencies
up
to
22
%
can
be
achieved
on
a
cell
with
an
area
of
45
cm2
with
process
optimization.
However,
the
PERC
cell
appears
to
have
a
more
competitive
structure
for
commercialization
of
high-eciency
solar
cells
because
it
can
provide
a
high
eciency
over
20
%
without
the
boron
process
and
because
it
has
a
large
enough
rear
contact
size
to
adopt
various
cheap
opening
methods
for
the
rear
silicon
diox

ide.
If
high
eciency
is
achieved
in
a
PERC
structure,
low
resistivity
wafers
have
to
be
used,
and
the
spacing
between
metal
contacts
has
to
be
longer
than
the
diffusion
length
of
minority
carrier.
A
mechanical
scriber
was
used
to
replace
the
photolithography
process
to
open
the
rear
contact
pattern
of
the
PERC,
which
exhibited
an
eciency
comparable
to
that
of
the
standard
PERC
cell.
A
Ni/Cu
contact
system
was
adapted
to
the
PERC
cell
and
showed
promising
output
performances.
Formation
of
Ni/Cu
contact
by
plating
is
expected
to
provide
cost
eective
metallization
of
high-eciency
solar
cells
suitable
for
mass
production
without
eciency
loss.

ACKNOWLEDGMENTS

This
work
was
supported
by
the
National
Research
Laboratory
Project
of
Korean
Ministry
of
Science
and
Technology
under
Contract
No.
2000-N-NL-01-C-090.

REFERENCES

[1]
Paul
D.
Maycock,
Photovoltaic
Technology,
Performance,
and
Cost
1995-2010
(PV
Energy
Systems,
Warrenton,
January
2000),
p.
63.
[2]
J.
Zhao,
A.
Wang,
M.
Green
and
F.
Ferrazza,
Appl.
Phys.
Lett.
73,
1991
(1998).
[3]
M.
A.
Green,
A.
W.
Blaker,
S.
R.
Wenham,
S.
Narayanan,
M.
R.
Willison,
M.
Taouk
and
T.
Szpitalak,
18th
IEEE
PVSC,
61
(1985).
[4]
R.
M.
Swanson,
S.
K.
Beckwith,
R.
A.
Crane,
W.
D.
Eades,
Y.
H.
Kwark,
R.
A.
Sinton
and
S.
E.
Swirhun,
IEEE
Trans.
Electron
Dev.
ED-31,
661
(1984).
[5]
M.
A.
Green,
A.
W.
Blakers,
J.
Zhao,
A.
M.
Milne,
A.
Wang
and
X.Dai,
IEEE
Trans.
Electron
Dev.
ED-37,
331
(1990).
[6]
A.
B.
Sproul
and
M.
A.
Green,
J.
Appl.
Phys.
70,
846
(1991).
[7]
J.
M.
Gee,
20th
IEEE
PVSEC,
549
(1988).
[8]
Dieter
K.
Schroder
and
Daniel
L.
Meier,
IEEE
Trans.
Electron
Dev.
ED-31,
637
(1984).

You might also like