Академический Документы
Профессиональный Документы
Культура Документы
ABSTRACT
Spacecraft solar array engineers now have cell
choices running from inexpensive and less efficient
silicon (Si) cells, to gallium arsenide on germanium
(GaAdGe) cells, to more expensive and efficient multijunction (MJ) cells. This paper finds that the more array
weight can be reduced by using more efficient cells,
even when they are a very expensive option in terms of
the array alone, and put into the spacecraft payload
i. e. the scientific instruments or in the case of
commercial spacecraft, the communications equipment,
the more cost effective the spacecraft array system.
This is true for a wide variety of spacecraft. This is
because of the very high price of launching a spacecraft
payload and supporting it with a spacecraft.
Table I
Blanket Array Attributes Populated With Different Cells
Attribute
INTRODUCTION
Nominal Efficiency
This paper reports the cost effectiveness of using
Si, G W G e and MJ solar cells on varying types of
spacecraft and solar arrays. The spacecraft selected
are mainly differentiated by their weight. The largest
weighs approximately 4,700 Kg, the next 3,500 Kg and
the lightest 240 Kg. All of the spacecraft are tow earth
orbiters.
In all of the cases studied, the more efficient solar
cells offer substantial performance advantages to the
spacecraft. Namely, more spacecraft payload can be
launched and used. If any reasonable cost estimate is
assigned to launching the payload and supporting it in
orbit, the more efficient cells offer an enormous price
advantage to less efficient cells.
Si
Thin
Si
GaAs
14.8
14.0
18.5
7.8
11.8
16
52.9
35.0
25.8
EOL Efficiency
7
59.0
Array Area (d)
Cell Wt. (Kg)
36.7
Blanket Wt. (Kg)
102.1
Blkt. Hsing Wt. (Kg) 72.8
Mast Assy. Wt.(Kg) 33.6
Other Weight (Kg)
24.6
Total Weight (Kg)
233.2
Cost (M$)
28.0
MJ
24
74.5
32.3
23.8
73.8
72
53.6
70.4
61.5
56.2
32.1
26.8
23.6
23.5
199.9
19.9
180.2
18.1
151.5
29.5
38.0
33.3
LARGE SPACECRAFT
The large spacecraft studied is a typical low earth
orbiting spacecraft. It will have a life of 5 years, a weight
of 4,690 Kg, an altitude of 700 km, an inclination of 98
degrees and a single side whose normal is parallelto the
spacecraft's velocity and another side whose normal
always points to the nadir. The spacecraft will be
equipped with a flexible blanket array that is sun
tracking.
The cost effectiveness of .14 m thick multijunction, .14 mm thick gallium arsenide, .062 mn thick
293
CF = . 9
(12)
PF
.451
(13)
SF
.863
(14)
difference in weight in
The solution to these equations is:
ADATA
4.61
(15)
2.55
(16)
dELEC =
1.87
(17)
APWR =
2.71
(18)
0.83
(19)
Total = 12.57
(20)
"A
&SA =
load
Not all of the 28.7 Kg can
because as the payload
the
spacecraft subsystem
to support it. From
1163.3
(1)
DATA =
369.7
(2)
COMM =
204.1
(3)
ELEC =
294.8
(4)
PWR =
427.6
(5)
SA
151.5
(6)
294
/rem
ells by Type
Additional
Instrument Wf.
Over Si Array (Kg)
Cost Per Unit
Wejght to "BUY
Science with High
Perf. Cells ( M g )
Si
Item
552
531
519
Instrument Weight
633
664
678
31
45
152
127
Additional Instrument
Wf. Over Si Array (Kg)
Cost Per Unit Weight to
"Buy"Science with High
Perf. Cells (UKg)
NIA
Table IV
Cost Effectiveness of Solar Cells by Type
For a Small Space aft with i qigid Ari r
Item
Cost Per Unit Wt.
to Launch & Support
Science (K$/Kg)
Instrument Weight
Additional Instrument
Wt Over Si Array (Kg)
Cost Per Unit Weight to
'Buy"Science with High
Perf. Cells (YKg)
MEDIUM SPACECRAFT
MJ
Si
GaAs
680
641
617
58.2
62.4
64.9
4.2
6.7
NIA
189
164
295
It is particularly so in the
size spacecraft. In these
296
1)
In order to counteract the Ge fragility for cell manufacturing and array assembly a minimum thickness of the
base material has to be provided (here: > 145 pm = 57
mil). The higher specific weight follows a total mass coverage, including 100 pm cover glasses of at least l ,2
kg/m2 compared to 0,9kg/mZ for 200 pm Si and twice as
much as achievable with thin Si cells. Obviously the panel substrates must take into account the higher loads by
added mass.
1.4)
1.1)
Cell Efficiency
357
D.C.
3)
2)
Coupon T h ~ r m a~ ~y ~ ~ i R
The implementation of the Ga-As cells for mass production is definitely different compared to the development
and pre-production activities. The significantly higher brittleness has resulted in "softer"manufacturing approaches,
especially in the welding and bonding processes. Another
improvement was introduced by better matching of the
thermal expansion coefficients and consequently the inplane forces between Ge-substrate and cover glass. For
the low temperatures (down to - 185 "C) the differences
below about -100 "C,i.e. where the glass adhesive is frozen, are lowest when using CMZ type glass.
During the cell -> p contact welding -> calm glassing process it was detected, that repeated measurements
in order to gain a feeling about measurement reproducibility - exhibited permanent and consecutive degradation
(see Fig. 2). The degradation was caused by mechanically
contacting the cell edges in the magazines and with jigs
and fixtures. Application of a "friendly" coating cured the
effect.
Cracks:
* The majority of the defects appear after few thermal
necessary, repair.
The rearside welding exhibited a so far unknown phenomenon: generation of "hairlineeffects" (HL's). The investigations let assume, that in fact the focal and temporal heat
weave generated during the welding (Ag rearside to Ag interconnector, 123 pm thick) causes some softening of the
Ge substrate and by this opening a crystal gap which closes again after cool-down. The defects are very narrow,
hard to detect and, a s shown in the following, obviously
"contained".They generally start at the welding locations,
5)
Manufacturing Results
O The knowledge of the reverse cell behaviour, its stability and longtime behaviour is currently insufficient and
should be improved by carefully observing the on orbit
performance.
O
Not directly comparable cycling tests for LEO missions indicate potential degradation problems. Results
will be published, when available.
New HL's have been detected (or first seen?) after the
N cycles
O Basically the distribution of the defects looks well randomly distributed over the panel areas, i.e. no biasing
due to built-in substrate discontinuities as inserts, doublers
The number of new and total cracks (they were already repaired before the bonding process) is significantly
higher than of the HL's.
...
mm
V.l,.,.
(VI
"cm
1.1
nor
RMRC CUUCII~
Specific Cost
tv no1
z
' 0
@e
vat,.,.
(VI
V.I.,.
(VI
Si, BSR
17
2 from 4(5)" 0 wafer
200
100
123
20.000
'Ow/
1Array:
EOM Specific Power
EOM Sp. Mass
IVD
Ga-AdGe
Basics:
AM0 (28 "C)
Cell Size
Cell Thickness
Breakage
Typ. cost / m2
bl8a..
f ; ;[-*;J
EKRX
Cell Type
Conclusion
'4 L o o
25
143
17,7
1,s
117
18,3
43x43"
145
hi, 0 190
120.000
168
17,8
-2-
# of Actions
Magazin
coated
+avg X lop
Magazin
-4%
uncoated
-5 %
o
l
o
R
M
F
Thermal
Cycles
I;
100
I F
\
10
5
a
40
accum.
0.1
--E
-5
30
Y
F 20
2
0.01
Hairlines
accum.
0,001
0.0001
1000
10
100
Thermal
Cycles
10
0
1
10
100
1000
Normalised post N D e f e c t s
.;3
.-a
i3
i3
23
+ 2
0
1
Current
9 1 0 1 1 1 2
Class
Current
9 1 0 1 1 1 2
Class
360
ABSTRACT
The UV stability of Si solar cells passivated by lowtemperature remote PECVD silicon nitride films is tested.
Perfect stability of the front surface passivation and the
rear surface passivation of both p-n junction as well as
MIS-IL Si solar cells is observed. Using the microwavedetected photoconductance decay (MW-PCD) method, a
very small and slow degradation of the differential
effective surface recombination velocity S e f d is observed
at silicon nitride-passivated p-Si surfaces corresponding
to the non-metallized rear surface regions of bifacial cells.
However, the degradation is too small to have any impact
on the long-term stability of encapsulated 17-18% rear
efficient bifacial cells. Thin-silicon-oxide/silicon-nitride
double layers incorporating Cs as used at the front
surface of MIS-IL solar cells provide perfectly stable and
excellently low differential s & d values of 23 cmls on
1.5-Qcm wafers. Applied to the rear surface of bifacial Si
solar cells, this double-layer scheme gives the potential of
stable rear efficiencies of even 20%.
INTRODUCTION
Many investigations in the past have shown that lowtemperature (< 500C) plasma-enhanced chemical vapor
deposition (PECVD) of silicon nitride (SiN,:H) is well
suited to provide excellent surface passivation of silicon
solar cells [I-51. There are two possible applications of
SiN,:H films: (a) passivation of diffused surfaces, e.g. at
the front surface of diffused p-n junction silicon solar
cells, (b) passivation of the pure silicon bulk material, e.g.
at the front surface of metal-insulator-semiconductor
inversion layer (MIS-IL) solar cells or at the rear surface
of both p-n junction as well as MIS-IL solar cells.
In case (a) UV stability is not a critical problem since
the doping profile reduces the surface recombination
velocity (SRV) at the SiN,:H/Si interface to a much lower
effective SRV at the edge of the space charge region [6].
The UV stability is much more important, however, if
SiN,:H is used in case (b) to passivate the non-diffused
surfaces of high-efficiency Si cells. In these applications
the silicon nitride must directly pravide excellent Seff
values, which have to be stable under UV exposure since
an increase directly affects the surface passivation. As an
example, for 18% rear efficiency of bifacial silicon solar
cells, Seffmust be better than about 200 cm/s [15]. This
problem is already known from the Stanford concentrator
Si solar cells, where a front surface passivation by a
417
0-7803-3166-4/96/$5.00 0 1996 EEE
1 1 1 1
1 1 1 1
1 1 1 1
1 1 1 1
a)
- 2018
b)
uv
16
5 700
650
p0 600
30
0
J
v)
a)
- 19
b)
E 17
p-SI
15
80
75
70
650
-600
I l l 1
U_
r TuvT t
a
-
p 550
40
35
30
$
v)
a)
- 1917
Fig. 1 Spectrum of the 400-W metal halide UV lamp in comparison to the AMI .5Gspectrum. The intensity of the UV lamp was
measured at the test plane (24 cm from the bulb).
b)
1 iuvl 1
15
650
- 600
a-
550
-409
35
30
0
2
3
p-si
Fig. 4 (a) Measured cell parameters of standard and truncatedpyramid MIS-IL silicon solar cells during UV exposure. The solid
lines are first-order regressions. (b) Schematic of the illuminated
front surface of the standard MIS-IL cell.
418
film thickness
80
50
100
A. Experimental details
All PCD samples are (100)-oriented boron-doped
i wafers with a
single-crystalline float-zone (FZ) S
resistivity of 1-2 Qcm and a thickness of 300 pm. The
wafers exhibit excellent minority-carrier lifetimes above
1.7 ms. Prior to the Si nitride depositions onto both
surfaces, the samples receive a standard RCA clean.
Samples with double-layer passivation schemes are
additionally oxidized at 800C and dipped into a Cs
solution. The MW-PCD measurements are performed
under a bias light intensity of about 50 mW/cm2 at room
temperature. Differences in the measured effective
minority-carrier lifetime zeff are exclusively related to
surface passivation effects, since we observe no variation
of the bulk minority-carrier lifetime during all processing
steps. From zeff we calculate a differenfial SRV Seffd [I81
and use its increase to characterize the passivation
stability. It should be noted that the acfuaf SRV at silicon
nitride-passivated p-Si surfaces is less than a factor of 2
higher than the differential SRV Seffd [19].
B. Film thickness dependence
To obtain clear effects within reasonable times, a
more extreme UV test is performed using harder UV
radiation with h,,=320
nm. Fig. 5 shows the influence of
419
C. Wavelength dependence1
Fig. 6 shows the waveleingth dependence of the UV
degradation. A set of identical PCD samples was
illuminated using different 1JV cut-off filters. To obtain
clear effects within reasonable times, samples with thin
SiN,:H films (35 nm) were iinvestigated. With increasing
UV cut-off wavelength the stability improves enormously.
For wavelengths greater than 400 nm no degradation is
detectable. To verify this result, an additional test was
performed using a 250-W halogen lamp. The films are
perfectly stable under the halogen illumination (3.7 suns,
400 hours).
80
50
100
-3
30
y$
20
0)
10
50
100
150
cm -*I
420
Efficiency evaluation of a-Si and c-Si solar cells for outdoor use
M. Kameda, S. Sakai, M. Isomura, K. Sayama, Y. Hishikawa, S. Matsumi, H. Haku,
K. Wakisaka, M. Tanaka, S. Kiyama, S. Tsuda and S. Nakano
New Materials Research Center, Sanyo Electric Co., Ltd.
1-1, Dainichi Higashimachi, Moriguchi, Osaka 570,Japan
ABSTRACT
We have achieved the world's highest
stabilized efficiency of 8.9% for an a-Si single-junction
solar cell (lcm2) and 10.6% for an a-Si/a-SiGe tandem
solar cell (I cm2).
To apply these results to practical outdoor
power use, the annual output power of a-Si singlejunction (a-Si) and tandem (a-Si/a-Si) solar cells and a
crystalline silicon (c-Si) solar cell was calculated based
on annual meteorological data and the solar spectrum in
Tokyo, Japan, including the effect of the output power
dependence on temperature, incident irradiance and
solar spectrum. As a result, this simulation revealed that
the annual change of efficiency for the c-Si solar cell is
most affected by the solar spectrum among the three
types of solar cells, and the annual fluctuations of the aSi and a-Si/a-Sisolar cells are mostly caused by recovery
by the annealing effect.
INTRODUCTION
We have been investigating improvements in
light
v-r7
5sun(AM-l.5), 25"C,
160min;open circuit
------------
Stabilized,t'
Cell area : 1cm2
0.5
1.o
0.5
1.5
Voltage (V)
Voltage (V)
Fig. 1 Illuminated I-V characteristics of a-Si and a-Si/a-SiGe tandem solar cells.
1049
0-7803-3166-4/96/$5.000 1996 IEEE
20
0 Single a-Si
A Tandem a-Si (Si/Si)
10
20
30
40
50
60
Temperature('C )
Fig. 2 Temperature dependence of a-Si single-junction
and c-Si
solar cell (O),a-Si/a-Si tandem solar cell (A)
solar cell (U).
Table 1 Temperature coefficients.
~~
Single a-Si
Tandem a-Si(Si/Si)
c-Si
DEP
CE
"
-0.2
0.2
0.4
0.6 0.8
1
Voltage (V)
Fig.3 Dependence of I (V) on the illumination intensity.
Symbols indicate experimental data.
SP
N@
where,
lE2(V): current-voltage characteristics at E, mW/cm2,
I&/):
current-voltage characteristics at OmW/cm2
(dark),
lloo(V): current-voltage characteristics at 100 mW/cm2,
V:
applied voltage.
6 25
20
2 15
10
0 0.5 1
1050
CALCULATION
RESULTS
In this simulation, the following equations are
used to calculate the output power of each solar cell,
including the output power dependence on temperature,
irradiance and solar spectrum as mentioned above.
Annual total output power Ptotal is calculated as a sum of
instant output power P(t) through a year by Eq. (2),and
instant output power P(t) can be expressed by a product
of Pstd and a coefficient of temperature CTemp(t),a
coefficient of irradiation Clr(t), and a coefficient of
spectrum CSpec(t)by Eq. (3),and here Pstd is the output
power measured under the standard conditions of AM1.5,
100mW/cm2, 25C. CTemp(t),Clr(t), and CSpec(t)are
calculated by normalizing each parameter dependent
output power PTemp(t),Plr(t), and PSpec(t)by the standard
output power, respectively, by Eqs. (4)-(6). The efficiency
of the total output power is expressed by a ratio of each
total output power to each total irradiance by Eq. (7).
Each parameter dependent output power is calculated
by equations (8)-(IO).
where
P(t) = P,,d.XC,e,p(t)XClr(t)XCSpec(t)
(3)
(4)
(7)
where
1.20
x c-Si
a-Si
A a-Si/a-Si
(9)
0.8%
'1
where
(x : temperature coefficient (measured)
S,,,( A ,t) : spectrum at irradiance 100mW/cm2
'
'
'
'
6 ' 7
Month
'
'
1051
solar cells and the c-Si cell comes mainly from material
properties, such as bandgap, and the difference between
the a-Si and a-Si/a-Si solar cells comes mainly from cell
design. The a-Si/a-Si solar cell is less affected by the
annealing effect than the a-Si solar cell, and is less
affected by spectrum dependence than c-Si, but it is
inversely affected through a year compared with c-Si
because of an imbalance in photocurrent between the top
1.50
c-Si
0
6
C
a,
.0
Simulated
Measured (outdoortest)
U 1.00 -
.m
1.5
a-Si single
0
0
%
a,
.0
Simulated
Measured (outdoor test)
U 1.0
.m
(1)
A technique for simulating the annual output
power of solar cells has been developed using only the
indoor data of solar cells, including output power
dependence on temperature, irradiance and spectrum,
using meteorological data and the solar spectrum in
Tokyo, Japan.
(2)
The annual behavior of the actual output
efficiency of the c-Si solar cell has been well expressed
using this simulation technique, and It is confirmed that
this simulation technique can reasonably express the
annual behavior of the output power of solar cells
including only the dependence on temperature, irradiance
and spectrum.
(3)
As a result, it has turned out that the annual
behaviors of the efficiency of a-Si and a-Si/a-Si solar cells
are more stable against both temperature and spectrum
than that of c-Si by successfully including the spectrum
dependence in this simulation, and it is confirmed that the
annual fluctuation of the efficiency of a-Si and a-Si/a-Si
solar cells is mostly caused by recovery by the annealing
effect.
(4)
This simulation can be used to estimate the
annealing effect, which is a main factor of the annual
fluctuation of a-Si solar cells, and can also be used to
optimize cell design, including the practical temperature,
irradiance, and, especially, solar spectra as mentioned
above.
This work is supported by NEDQ as a part of the
New Sunshine Program under the Ministry of
InternationalTrade and Industry.
0.5
9 1 0 1 1 1 2
Month
Fig. 7 Normalized monthly average efficiency of a-Si
solar cell by simulation and outdoor measurement.
1.5c
a-Si/a-Sitandem
0
ox
C
a,
.0
Simulated
Measured (outdoor test)
p 1.oc
N
._
-
0.50
C
9 1 0 1 1 1 2
Month
Fig. 8 Normalized monthly average efficiency of a-Si/a-Si
solar cell by simulation and outdoor measurement.
REFERENCES
[ l ] Y. Hishikawa et al., "Approaches for stable multijunction a-Si solar cells", First WCPEC, 1994, pp. 386393.
[2] M. Kobayashi and T. Tani, "The characteristics of PV
modules output considered the environmental factors", J.
Japan. Solar Energy Soc., 19, 1993, pp. 39-46.
[3] Y. Hishikawa and S. Qkamoto, "Dependence of the IV characteristics of amorphous silicon solar cells on
illumination intensity and temperature", Solar Energy
Mate. and Solar Cells, 33, 1994, pp. 157-168.
[4] M. Habu et al., "Measurement of solar spectral
irradiance at Tanashi, Tokyo (Ill)", Res. Nectrotech. Lab.,
830, 1983, pp. 35-175.
[5] D. L. Morel, R. D. Wieting and K. W. Mitchell, "Design
and performance of thin film si1icon:hydrogen tandem
modules", Technical digest of Ist PVSEC, 1984, pp. 567570.
1052
ABSTRACT
We report on fabrication, properties and stability of aSi:H and a-(Si,Ge):H solar cells made using remote low
pressure ECR deposition. We have fabricated both
substrate and superstrate type solar cells. We can make
solar cells with high fill factors in both geometries, but
the voltages are higher with substrate-type solar cells than
with superstrate type cells. Special problems related to
diffusion of B have to be solved in superstrate cells
because the deposition is done at higher temperatures(
350-375 C). Several novel player grading schemes and
buffer layers which allow us to fabricate these types of
cells are described. The substrate cells were made with
both H-ECR and He-ECR discharges We fmd that while
the cells prepared with He discharge have lower H
concentration, and lower H content, they are less stable
than cells prepared using H, discharges. The stability of
cells was measured using ELH and xenon lamps, and
compared with the stability of cells made using standard
glow discharge techniques. We find that the cells
prepared using H,-ECR discharges are more stable than
standard glow discharge cells with comparable fill
factors, voltages and thicknesses of i layers. We also
report on a new type of graded gap a-(Si,Ge):H cell,
which appears to show improved stability.
Introduction
Improving the stability of a-Si:H solar cells remains a
problem of interest and fervent research. Several new
deposition techniques, among them remote ECR
growth[ 1,2],hot wire growth[3] and W-annealing[4] have
been used to improve the stability of a-Si:H solar cells.
While reports on film data are plentiful, there is little
work on making and studying a-Si;H solar cells made
using these new techniques. In this papr,we describe the
fabrication of a-Si:H solar cells on both stainless and tin
oxide substrates using remote, low pressure ECR growth
techniques. We show that good cells can be made in both
geometries. We also report on the special problems one
faces when making these cells, and how to solve them.
Growth Technique
The ECR( Electron-Cyclotron-Resonance) plasma-CVD
growth technique we use has been described
previously.[ 11. To recapitulate, it uses microwaves and
magnetic fields to create a highly ionized beam of either
hydrogen or helium atoms, which stream towards the
substrate. The streaming beam of hydrogen reacts with
silane to create primarily SiH, species, which leads to the
growth of the material. During growth, the material is
subjected to controlled ion bombardment and etching by
the H ions and radicals, thus achieving the condition of
etchingduring-growth.Thisetchingduring-growthshould
lead to better localized microstructure and hence, better
stability.
The previous work on films showed that indeed, the film
properties were comparable to the best glow-discharge aSi:H, and that the degradation in the material properties
upon light exposure was significantly less than
comparable glowdischarge films[11. In particular,we
found that the increase in defect density upon light
soaking was only about a factor of 2.5-3, and that the
addition of minute levels of B compensationreduced both
the initial and final defect state densities.
Device Fabrication
We have fabricated two types of p-i-n devices: superstrate
devices on tin oxide substrates, and substrate devices on
bright-polished stainless steel. Note that no special
reflectors were provided in either case. The stainless
substrate, in particular, was not electro-polished, nor
provided with a textures Ag reflector or any other
reflection enhancement means.
Asuperstrate devices. These were made on low haze
Asahi tin oxide, and were of standard p-i-n types, with a
1069
0-7803-3166-4/96/$5.00 0 1996 IEEE
0.17
strate cell wi
1.0
0.8
evice 1:
The I(V) curve for a solar cell produced using Hdischarge is shown in Fig. 2. The corresponding QE
curve is shown in Fig.3, and the QE ratio curve ( Ratio
of QE at OV/QE at + O S V) in Fig. 4. From these
figures, it is apparent that one can produce good cells
using H-ECR. The voltage is 0.786 V, which translates
tion, significantlybetter
metries. The fill factor
d the QE ratio shows no collection
a similar result when
stead of Hydrogen.
efficiencies of the two
and other using He, we
0.
f
1070
1.8
Stability of ceUs
1.6$
A) Superstrate cells:
To measure the stability of superstrate cells, we
compared two cells, one produced using glow discharge,
and one using ECR. Both had the same thickness of i
layer(within lo%),and very similar voltages(0.78V) and
fill factors(0.67) initially. The QE curves also were very
similar. Then, we proceeded to light soak these cells
using 2 x sun xenon illumination from an Oriel simulator.
In Fig.5, we show the results of the degradation of these
cells as a function of time. It is clear from Fig. 5 that the
ECR-produced cell is more stable than its glow
discharge-produced cousin. We measured the QE and
QE ratio for the two cells before and after degradation.
These results are shown in Fig. 6, and we see that the
QE ratio after degradation for the glowdischarge cell is
much higher in the red region of the spectrum than for
the ECR cel1,implying a worse problem with hole
collection, i.e. greater i layer degradation for the glowdischarge cell. Thus, we have shown that the ECR cell is
more stable because the i layer itself is more stable.
.A
1*2f+.t.
1.0
.;0
I
4 0 0 4 5 0 5 0 0 5 5 0 600 650 7 0 0 7 5 0 8 0
Wavelength (nm)
ECR
B) Substrate cells
We again selected three cells which were as near identical
as we could make them, one deposited using H-ECR, one
with He-ECR, and one, glow discharge. All three had
similar ss substrates, top Cr contacts, and i layer
thicknesses, and initial fill factors in the range of 0.67.
We subjected the three cells to light soaking from an
ELH lamp which was focussed down to give incident
intensities of about 6x sun, which is about 1.9 sun into
the cel!. The cells were maintained at temperatures of 20
C using a cooled mounting block.
1.5
The results for the degradation of the fill factor for the
three cells, which is the primary indicator of material
quality degradation, are shown in Fig.7. From this
figure, we see that the H-ECR cell degrads much less
than either the glow discharge or He-ECR cells, whose
relative degradations are very similar. This is a striking
result. What it says is that H content by itself is not an
indicator of material stability. Even though the
He-ECR film had less bonded H, it degraded just as
badly as the glowdischarge device. We therefore,
conclude that only by changing the localized
microstructure, and therefore H bonding, can we
change stability. H-etching during growth does change
'
1-t
503
803
6co
Wavelength (nm)
it
more
1.0
%
Graded gap a-(Si,Ge):H Cells
s
II
Y
e --.
\ -
H-ECR
.- _- -__
H+ECR
GD
2a
0.5
+!
25
50
75
Time ( Hours )
1.0
References
0.9
5
1. V. L. Dalal,E. Ping, S. Kaushal, M. K. Bhan and M.
Leonard, Appl. Phys. Lett. 64,1862(1994)
2. V. L. Dalal et al. J.NonCryst.Sol(1996)To be publ.
3. A. H. Mahan et al, J. Appl. Phys. 69,6728(1991)
4. H. Shirai, J. Hanna and I. Shimizu, Proc. AIP,
268,388( 199 1)
0.5
25
50
75
Fig. 9 Degradation in f
a factor of H-ECR a(Si,Ge):H cell with grading in the middle.
1072
ABSTRACTS
N-Contact(front) Si02Layer
NfLayer
\
1
ARC
/
P-Substrate
INTRODUCTION
In 1995, some kinds of high efficiency silicon (HES)
solar cells designated as NRS/LBSF (Non-Reflective
Surface / Local Back Surface Field) cells, NRS/BSF (NonReflective Surface / Back Surface Field) cells and
NRS/BSF cells with IBF (Integrated Bypass Function)
have been developed and commercialized[l][2][3]. These
HES cells were categorized as the first generation of high
efficiency silicon solar cells and had 12.2 % to 12.5%
efficiencies at EOL condition (1MeV electron fluence of
lX105 e/cm2). Totally about 1.5 million HES cells have
been manufactured and supplied not only for NASDA
satellites but also for extensive satellites in the world
(totally over 50 satellites).
In 1998, the development project of the second
generation of high efficiency silicon solar cells designated
as BJ and AHES cells were started to meet the customer
requests for higher EOL efficiencies, This project was
executed under the cooperative development between
NASDA and SHARP in the same manner as the first
generation of HES cells. Preliminary data of BJ and
AHES-1 cells were presented in the previous papers[2][4].
This paper summarizes the main features and
qualification status of BJ, AHES-1 and AHES-2 cells
generated in this project.
1114
N-Contact
c'
N+ Layer
Contact Win dow
P-Contact
P+ Layer
.&
(Rear View a t Left-Top Corner7
0.7
$
QI
o.6
0.5
E
0.4
U)
-E
[r 0.3
ti, 0.2
*a
0.1
0
300
400
500
600
700
800
1000
1100
1200
-8
E
5
60
a,
50
.E
40
'm
g
d
30
20
5:
10
n
" l
-~
300 500 700 900 1100 1300 1500 1700 1900 2100 2300 2500
Wavelength(nm)
1115
900
Wavelength (nm)
\
/
7
/
/.
'
,'
,,
/
'
BJ solar cell-
P-Substrate
P' Layer
Fig. 5. Newly developed interconnector
-/ I,
Si02 Layer
P-Contact
'
-
Fig. 7. EnlargedTextures
I
Type
HES (bare cell)
AHES-1 (bare cell)
BJ (with opaque substrate)
BJ (with transparent kapton substrate)
v(%) I
@28"C
12.5
13.1
13.5
13.5
I v(%) I EfficiencylClCWeighqEfficiency/CICWeigh4
I
as
0.86
0.84
0.84
0.80
Top("C)
51.8
48.9
48.5
43.8
1116
Ratio
Ratio
@Top Ratio
11.2
100
100
100
11.8
105.4
78.1
135
12.2
108.9
70.1
155
111.6
12.5
70.1
159
Measurement Condition : AM0 135.3mW/cm2,28C
Top:Operating Temperature Solar Cell Size:2x2cm2
0.7
9
a 0.6
0.5
3 0.4
E2 0.3
'
& 0.2
0. I
0 '
300
400
500
600
700
800
900
1000
1100
I200
Wavelength (nm)
CONCLUSION
The BJ structure improvedthe EOL efficiency of silicon
cells moreover. These cells have p-n junctions at both
front and rear surface and showed less radiation
degradation and better remaining factor than previous
silicon cells. Comparing with HES, the BJ cells improved
EOL performanceat operating temperature by about 8.9%
and the cell weight was reduced by about 0.7 times. So
the power weight ratio of the BJ cells is greater than HES
cells by about 1.5 times.
Development of BJ cells were accomplished and
qualified as space solar cells in accordance with the
NASDAs specification. The detailed design and process
of the cells were settled for their mass production.
The radiation resistance of silicon space solar cells has
been improved by the thinning substrate and enlarging
surface texture size. The decreased substrate thickness
and enlarged textures give the effects of reduced
recombination centers and carrier generation closer to the
front p-n junction. The AHES-1 cells have been developed
and commercialized. Their EOL efficiency has reached
13.1%.
The AHES-2 cells also have been developed. Current
performance data of AHES-2 cells were presented.
REFERENCES
Test
Humidity
Thermal shock
-\-
1117
Chanae rate(%\
ABSTRACT
Lock-in thermography has been applied for shuntanalysis on epitaxial siiiwn thin-film solar cells. The solar
cell material was made by epitaxial deposition of the base
layer on highly doped monwrystaliine (Czochralski) and
multicrystalline silicon substrates in an APCVD-system.
Solar cells were prepared in a laboratory-type and an
industrial-type process.
Characterization of the solar cells by infrared Lock-in
thermography and microscopy revealed a clear correlation between shunts and epitaxial defects in case of the
lab-type solar cells. Furthermore an increased concenlration of shunts located under the emitter grid lines of the
industrial-type solar cells compared to the lab-type solar
cells was observed. The analysis by thermography thus
gave insight Into the quality of the epitaxial layers and into
problems concerning a transfer of the solar cell process
from laboratory to industrialscale manufacturing.
INTRODUCTION
Within the high-temperature route for crystalline silicon solar cells the epitaxial deposition of silicon layers
plays an important role in most solar cell concepts. Starting from a recrystallized silicon seeding layer or from a
low-grade silicon substrate the base layer Is usually deposited by epitaxy. The performance of the thin-film solar
cell strongly depends on the crystal quality of the deposited base layer. Research activities in the field of silicon
deposition for an application in silicon thin-film solar cells
therefore concentrate on the development of inexpensive
deposition methods which are capable of producing layers
of high crystal quality.
Another important topic conceming crystalline silicon
thin-film solar cells is the transfer of solar cell processing
from laboratory to industrial-scale manufacturing. Epitaxial
thin-film solar cells represent a wafer equivalent which
can be directly introduced into standard solar cell fabrication.
In this paper we report the manufacturingof epitaxial
thin-film solar cells by applying a laboratory-type and an
industrial-type solar cell process. The solar cells were
characterized by I-V measurements, microscopy and lockin thermography. Lock-in thermography has proved to be
a powerful tool for an analysis of leakage currents
(shunts) affecting the solar cell performance. However up
to now this technique has mainly been applied for the
0-7803-7471-1/021$17.0002002 IEEE
111, PI.
I-V CHARACTERISTICS
A comparison between the I-V characteristics of the
iab-type and the industrial-type solar cells reveals that the
latter suffer from extremely low fill factors (<30%) as well
as low open-circuit voltages. The reference cells have
equally low fill factors but high open-circuit voltages of
600 mV. In Table 1 the best solar cell results obtained
with both solar cell processes are listed.
In Fig. 1 two typical dark current density curves of
epitaxial thin-film solar cells are shown: one cell was
manufactured via lab-type the other via industrial-type
solar cell process. An analysis of the dark current characteristics revealed that in case of the industrial-type prccess all solar cells, including references, suffer from high
series resistance in the range of > l o Rcm. An Increased
thickness of the SIN. antireneclion coating combined with
1335
FF
lsc
RP
PA]
[mv] [mn/cm']
FA]
Lab. I 6 1 9
30.0
73.9
79.8
630
33.7
Ind. 521
15.4
28.9
604
20.8
29.2
'References: Cz material
Solar ceN area: 21.16 cm' (lab.).
13.8 2.6~10'
17.0 7.8xld
*I
I
'1
49
65x18
2.3
3.7
RS
ram?
0.54
0.54
32.5
20.3
Table 1. Best solar cell results obtained for both solar cell
processes. Thin-fiim cells based on Cz substrate.
. , . , .
10
.
1~
.( .. ,.
I
' :..
.
..
Fig. 2. Lock-in thermogram (0" image) of an epitaxial thinfilm solar cell on monocrystalline substrate at 0.6 V forward bias.
Voltage M
Fig. 1. Dark current density curves of WO epitaxial thinfilm solar cells on monocrystalline silicon substrate.
LOCK-IN THERMOGRAPHY
Lock-in thermography was carried out for an analysis
of shunts affecting the performance of the manufactured
solar cells. A detailed description of the measurement
principle can be found in [Z].
An investigation of the cell surface by microscopy revealed that three types of surface irregularities are present: pure silver droplets, epitaxial spikes (mostly coated
with silver) and epitaxial stacking faults. The spikes are
the most frequent defect type followed by the stacking
faults. A microscopical analysis (SEM. Nomarski microscopy) revealed that most spikes are grown on top of a
stacking fault.
1336
m.
ner) the corresponding -90" image was studied. This image indicated that regions which appear black in the
0' image actually are affected by shunts. The crystal
structure in these areas is strongly disturbed due to a
comparatively high density of spikes. A similar relation
between shunts and regions of poor crystal quality or regions with a high density of grain boundaries is reported
in [6].
0-7803-7471-1/02/$17.0002002 IEEE
,
1337
ACKNOWLEDGEMENTS
This work was supported by the European Commission in
contract no. ERK6-CT-1999-00014 and by the Federal
Ministry of Economics and Technology under contract
no. 0329836 (TEKSI). D.M. Huljie is financially Supported
by the scholarship program of the Deutsche Bundesstiftung Umwelt. Germany and a scholarship of the
Albert-Ludwigs-Universityof Freiburg, Germany.
REFERENCES
[ l ] 0. Breitenstein, M. Langenkamp, 0. Lang,
A. Schirrmacher, Solar Energy Materials and Solar Cells
65, 2001, pp. 55-62.
[5]J. Rentsch, D.M. Huljit, S. Reber. R. Preu, R. Ludemann, "Progress in screen printed front side metallization
schemes for CSiTF solar cells", this conference.
1338
Introduction
282
Experimental methods
3.1
growth rate
4
dark conductivity
103
2
105
d, cm1
growth rate, /s
107
0
0
2000
4000
6000
pressure, mtorr
(220)
160
(111)
(311)
2000
Si-O
1500
80
, cm1
Topt 20C
120
Topt
Topt + 140C
1000
Topt + 20C
500
40
Topt + 20C
0
20
30
40
Topt
Topt 20C
50
500
60
1500
2000
2500
wavenumber, cm1
2, deg
1000
Fig. 3
We have measured Ld in an actual solar cell conguration, using a CV technique at a frequency of 100 Hz. At
such a frequency, all states in the intrinsic layer of the cell up
to B0.5 eV below the conduction band should respond to
the capacitance signal. Figure 4 shows the CV curve and
the slopes of the curves give values of doping density
2.1 1015 cm3 near the p layer, and 7.6 1015 cm3 at a
distance 0.7 mm from the p layer. Knowing the depletion
width from CV measurements, Ld can be estimated from
the reverse bias quantum efciency (QE); this involves
deposition of ITO on the n+ layer, and the measurement of
QE from a 500 nm light beam at different bias. At this
wavelength, as most of the photons are absorbed within
o0.1 mm, then the holes are generated essentially at the
back surface of the intrinsic layer. By varying the bias
voltage, the depletion width is changed, and therefore, the
eld-free distance that the hole diffuses before it is swept
away by the depletion eld, also changes. By matching this
curve with a simple exp(x/Ld) relationship for carrier
collection, where x is the thickness of the eld free region,
we estimated the diffusion length Ld to be B1.2 mm.
10
8
6
voltage, V
0
0.1
0.1
0.3
0.5
0.7
0.9
0.5
0.3
0.1
0.1
0.3
0.5
0.7
0.9
reverse voltage, V
J, mA/cm2
5
0
3.2
10
nc-p/a-i/a-n
nc-p/nc-i/a-n
nc-p/a-i(2 nm)/nc-i/a-n
15
reflectance, %
101
103
a-Si:H
60
(ii)
(i)
40
105
(iii)
20
107
process B
process A
200
109
300
400
500
600
wavelength, nm
0
500
1000
1500
2000
2500
thickness,
nc-Si:H
80
10
Topt + 140C
1.0
Topt
Topt 20C
0.8
relative QE
0.6
0.4
0.2
0
400
600
800
1000
wavelength, nm
voltage, V
0
0.1
0.1
0.3
0.5
0.7
3
1.0
25
Topt
12
0.9
Topt + 20C
Topt + 140C
15
0.8
FF
20
Jsc, mA/cm2
J, mA/cm2
pulsed PECVD
15
Topt
10
0.7
18
5
100
200
0.75
0.70
0.65
0.60
0.55
100
100
200
TH- T opt, C
2.0
1.8
1.6
1.4
1.2
1.0
0.5 1.0 1.5 2.0 2.5
i-layer thickness, m
6
efficiency ,%
FF
TH- T opt, C
log (J)
Jsc , mA/cm2
Voc, mV
100
25
20
15
10
5
100
0.6
2.5
2.0
2
550
500
450
400
350
1.5
i-layer thickness, m
1.0
100
TH- T opt, C
200
8
7
6
5
4
3
100
0.84 m
1.4 m
100
200
TH- T opt, C
0.8 m
2.1 m
8
0.4
0.2
0.0
0.2
0.4
0.6
voltage, V
Fig. 13
Conclusions
The pulsed PECVD technique, of necessity, leads to a nonequilibrium plasma condition and appears to suppress the
nucleation centres that subsequently lead to dust, which in
turn could reduce structural defects in the lm. This
reduction in particulate count in the plasma has manufacturing yield implications. In this work, an existing pulsed
PECVD technique was modied for the growth of nc-Si:H
lms and solar cells. The technique has been demonstrated
to reduce/eliminate the amorphous incubation phase at the
p/i interface (which has a deleterious effect on nc-Si:H
devices). An efciency of B7.5% (FF of 0.69) at an i-layer
286
Acknowledgments
References
1 Yang, J., and Guha, S.: Amorphous silicon alloy materials and solar
cells near the threshold of microcrystallinity, Mater. Res. Soc. Symp.
Proc., 1999, 557, pp. 239250
2 Nasuno, Y., Kondo, M., and Matsuda, A.: Microcrystalline silicon
thin-lm solar cells prepared at low temperature using RF-PECVD.
Proc. 28th IEEE Photovoltaic Spec. Conf., Anchorage, USA,
September 2000, pp. 142145
3 Roschek, T., Repmann, T., Kluth, O., M.uller, J., Rech, B., and
Wagner, H.: High rate deposition of microcrystalline silicon solar cells
using 13.56 MHz PECVD prerequisites and limiting factors, Mater.
Res. Soc. Symp. Proc., 2002, 664, pp. A4.3.1A4.3.12
4 Shah, A., Meier, J., Torres, P., Kroll, U., Fischer., D., Beck, N.,
Wyrsch, N., and Keppner, H.: Recent progress on microcrystalline
solar cells. Proc. 26th IEEE Photovoltaic Spec. Conf., Anaheim,
USA, October 1997, pp. 569574
5 Jones, S.J., Crucet, R., and Izu, M.: Use of a gas jet deposition
technique to prepare microcrystalline Si solar cells. Proc. 28th IEEE
Photovoltaic Spec. Conf., Anchorage, USA, September 2000,
pp. 134137
6 Klein, S., Finger, F., Carius, R., Rech, B., Houben, L., Luysberg, M.,
and Stutzmann, M.: High efciency thin lm solar cells with intrinsic
microcrystalline silicon prepared by hot wire CVD, Mater. Res. Soc.
Symp. Proc., 2002, 664, pp. A4.3.1A4.3.12
7 Meier, J., Vallat-Sauvain, E., Dubail, S., Kroll, U., Dubail, J., Golay,
S., Feitknecht, L., Torres, P., Fay, S., Fischer, D., and Shah, A.:
Microcrystalline/micromorph silicon thin lm solar cells prepared by
VHF-GD techinque, Sol. Energy Mater. Sol. Cells, 2001, 66,
pp. 7384
8 Yamamoto, K., Yoshimi, M., Suzuki, T., Nakata, T., Sawada, T.,
Nakajima, A., and Hayashi, K.: Large-area and high efciency a-Si/
poly-Si stacked solar cell module. Proc. 28th IEEE Photovoltaic Spec.
Conf., Anchorage, USA, September 2000, pp. 14281432
9 Sansonnens, L., Howling, A.A., and Hollenstein, C.H.: Large area
deposition of amorphous and microcrystalline silicon by very high
frequency plasma, Mater. Res. Soc. Symp. Proc., 1998, 507,
pp. 541546
10 Kirimura, H., Maeda, H., Murakami, H., Nakahigashi, T., Ohtani, S.,
Tabata, T., Hayashi, T., Kobayashi, M., Mitsuda, Y., Nakamura, N.,
Kuwahara, H., and Doi, A.: Study of deposition process in
modulated silane plasma, Jpn. J. Appl. Phys., 1994, 33, pp. 43894393
11 Morrison, S., XI, J., and Madan, A.: High deposition rate amorphous
silicon solar cells and thin lm transistors using the pulsed plasma
PECVD technique, Mater. Res. Soc. Symp. Proc., 1998, 507,
pp. 559564
12 Morrison, S., and Madan, A.: Deposition of amorphous silicon solar
cells via the pulsed PECVD technique. Proc. 28th IEEE Photovoltaic
Spec. Conf., Anchorage, USA, September 2002, pp. 928931
13 Morimoto, A., Matsumoto, M., Yoshita, M., and Kumeda, M.:
Doping effects of oxygen or nitogen impurity in hydrogenated silicon
lms, Appl. Phys. Lett., 1991, 59, pp. 21302132
14 Kamei, T., Wada, T., and Matsuda, A.: Effects of oxygen impurity
on microcrystalline silicon lms. Proc. 28th IEEE Photovoltaic Spec.
Conf., Anchorage, USA, September 2002, pp. 784787
15 Nasuno, Y., Kondo, M., and Matsuda, A.: Passivation of oxygenrelated donors in microcrystalline silicon by low temperature
deposition, Appl. Phys. Lett., 2001, 78, pp. 23302332
16 Chern, H.N., Lee, C.L., and Lei, T.F.: The effects of H2O2 plasma
treatment on the charcteristics of polysilicon thin lm transistors,
IEEE Trans. Electron Devices, 1993, 40, p. 2301
17 Kocka, J., Fejfar, A., Vorlicek, V., Stuchlikova, H., and Stuchlik, J.:
Microcrystalline silicon relation of transport properties and
microstructure, Mater. Res. Soc. Symp. Proc., 1999, 557, pp. 483493
18 Kitagawa, T., Kondo, M., and Matsuda, A.: Control of crystallinity
and orientation of microcrystalline silicon using in situ RHEED
observation. Proc. 28th IEEE Photovoltaic Spec. Conf., Anchorage,
USA, September 2000, pp. 780783
Introduction
316
ECR zone
H2 in
microwaves
magnets
vacuum
silane in
substrate
Device structure
ITO
p+ a-(Si,C):H layer
buffer layer
buffer layer
1.0
n+
0.8
QE
0.6
0.4
0.2
0
400
thin a-Si
The eld does not extend across the entire base layer
600
700
wavelength, nm
104
QE
101
QE
800
103
102
102
103
10
104
1.0
1.2
1.4
1.6
1.8
absorption coeff. of Si
500
1
2.0
E, eV
1.2
FF = 0.7
1.0
I, mA
0.8
0.6
0.4
0.2
0
0
0.1
0.2
0.3
0.4
0.5
V, V
Fig. 4 I(V) curve for microcrystalline Si:H cell with base layer
thickness of B0.8 mm
reproduced over time, showing that there is no ambientinduced oxidation and degradation in these devices. It is
believed that this result may be a consequence of the fact
that the lms and devices are subjected to B1015 eV ion
bombardment during growth [21], thereby producing dense
materials resistant to oxygen percolation along grain
boundaries.
The inuence of changing the hydrogen/silane ratio was
studied, and Fig. 7 shows that as the silane/hydrogen ratio
increased, the open-circuit voltage increased. Good microcrystalline cell characteristics were obtained, even at silane/
hydrogen ratios of 1:8, in contrast to the results of other
groups where for higher silane content beyond about 1:16
ratio, the material became primarily amorphous [11]. As the
voltage increased, the QE at 800 nm decreased, indicating
the more amorphous-like nature of the basic n material
(Fig. 7).
We also studied the inuence of varying the thickness of
the interfacial buffer layer for a xed composition
microcrystalline n base layer. Figure 8 shows the affect of
increasing the thickness of the a-(Si,C):H buffer layer from
15 nm to 30 nm on the IV curve. Clearly, the voltage has
IEE Proc.-Circuits Devices Syst., Vol. 150, No. 4, August 2003
0.20
0.6
0.15
Voc, V
0.4
0.3
0.10
QE at 800 nm
0.5
0.2
0.05
0.1
0
0.05
0
0.15
0.10
silane/hydrogen ratio
Fig. 9 Proposed band diagram at pn interface showing development of a notch as bandgap of buffer layer is increased
The notch traps holes, and a thick notch prevents tunneling of photogenerated holes into the p layer when electric eld is reduced in the
power quadrant
0.8
0.6
1.0
0.4
0.8
0.2
0.6
I, mA
I, mA
FF = 0.67
0.4
0
0
0.2
0.4
0.6
0.2
V, V
0
0
0.2
0.4
0.6
V, V
Fig. 10 I(V) curve of cell made with base layer fabricated with
helium dilution in addition to hydrogen dilution
12
10
C 2 (1016), F 2
4
8
6
4
2
0
1.0
0.5
0.5
1.0
reverse voltage, V
103
104
105
106
I, A
v v0 expEc Et =kT
14
107
108
109
10
1010
0
0.2
0.4
0.6
0.8
1.0
capacitance, nF
V, V
0
10
102
103
104
frequency, Hz
104
exp(qV /kT )
105
I, A
106
107
exp(qV /2kT )
108
109
1010
0
0.1
0.2
0.3
0.4
0.5
0.6
V, V
Conclusions
Acknowledgments
References
1 Yamamoto, K., Yoshimi, M., Tawada, Y., Fukuda, S., Sawada, T.,
Meguro, T., Takata, H., Suezaki, T., Koi, Y., and Hayashi, K.: Large
area thin lm Si module, Sol. Energy Mater. Sol. Cells, 2002, 74,
pp. 449455
2 Rech, B., Kluth, O., Repmann, T., Roschek, T., Springer, J.,
M.uller, J., Finger, F., Stiebig, H., and Wagner, H.: New materials
and deposition techniques for highly efcient silicon thin lm solar
cells, Sol. Energy Mater. Sol. Cells, 2002, 74, pp. 439447
3 Mulato, M., Chen, Y., Wagner, S., and Zanatta, A.R.: Microcrystalline Si with high electron eld-effect mobility deposited at 2301C,
J. Non-Cryst. Solids, 2000, 266269, pp. 12601264
4 Chen, Y., Pangal, K., Sturm, J.C., and Wagner, S.: p channel thin
lm transistor and complementary metal-oxide-silicon inverter made
of microcrystalline silicon directly deposited at 3201C J. Non-Cryst.
Solids, 2000, 266269, pp. 12741278
5 Vetterl, O., Gro, A., Jana, T., Ray, S., Lambertz, A., Carius, R., and
Finger, F.: Changes in electric and optical properties of intrinsic
microcrystalline silicon upon variation of the structural composition,
J. Non-Cryst. Solids, 2002, 299302, pp. 772777
6 Unold, T., Br.uggemann, R., Kleider, J.P., and Longeaud, C.:
Anisotropy in the transport of microcrystalline silicon, J. Non-Cryst.
Solids, 2000, 266269, pp. 325330
7 Kocka, J., Stuchl!kov!a, H., Stuchl!k, J., Rezek, B., Mates, T., Vrk, V.,
Fojt!k, P., Pelant, I., and Fejfar, A.: Model of transport in
microcrystalline silicon, J. Non-Cryst. Solids, 2002, 299302, pp.
355360
8 Vanacek, M., Poruba, A., Remes, Z., Rosa, J., Kamba, S., Vorlicek,
V., Meier, J., and Shah, A.: Electron spin resonance and optical
characterization of defects in microcrystalline Si, J. Non-Cryst. Solids,
2000, 266269, pp. 519523
9 Nasuno, Y., Kondo, M., and Matsuda, A.: Microcrystalline silicon
thin-lm solar cells prepared at low temperature using PECVD, Sol.
Energy Mater. Sol. Cells, 2002, 74, pp. 497503
10 Vetterl, O., Finger, F., Carius, R., Hapke, P., Houben, L., Kluth, O.,
Lambertz, A., M.uck, A., Rech, B., and Wagner, H.: Intrinsic
microcrystalline silicon: a new material for photovoltaics, Sol. Energy
Mater. Sol. Cells, 2000, 62, pp. 97108
11 Meier, J., Dubail, S., Golay, S., Kroll, U., Fa.y, S., Vallat-Sauvain, E.,
Feitknecht, L., Dubail, J., and Shah, A.: Microcrystalline silicon and
the impact on micromorph tandem solar cells, Sol. Energy Mater. Sol.
Cells, 2002, 74, pp. 457467
12 Klein, S., Finger, F., Carius, R., Rech, B., Houben, L., Luysberg, M.,
and Stutzmann, M.: High efciency thin lm solar cells with intrinsic
microcrystalline Si prepared by hot wire CVD, Mater. Res. Soc.
Symp. Proc., 2002, 715, p. A26.2
13 van Veenendaal, P.A.T.T., van der Werf, C.M.H., Rath, J.K., and
Schropp, R.E.I.: Inuence of grain environment on open circuit
voltage of hot-wire chemical vapour deposited Si:H solar cells, J. NonCryst. Solids, 2002, 299302, pp. 11841188
14 Dalal, V.L., Kaushal, S., Ping, E.X., Xu, J., Knox, R., and Han, K.:
Microcrystalline and mixed-phase Si:H: Preparation, properties and
potential for devices, Mater. Res. Soc. Symp. Proc., 1995, 377, p. 137
15 Erickson, K., and Dalal, V.L.: Growth of microcrystalline Si:H
and (Si,Ge):H lms on polyimide substrates using ECR deposition
techniques, Mater. Res. Soc. Symp. Proc., 1997, 467, p. 409
16 Erickson, K., and Dalal, V.L.: Preparation and properties of
microcrystalline (Si,Ge) on plastic substrates, Mater. Res. Soc. Symp.
Proc., 1998, 507, p. 987
17 Dalal, V.L., and Erickson, K.: Microcrystalline Si solar cells, Mater.
Res. Soc. Symp. Proc., 2000, 609
18 Dalal, V.L., and Erickson, K.: Microcrystalline Si and (Si,Ge) solar
cells on plastic substrates. Proc. 28th IEEE Photovoltaic Specialists
Conf., 2000, pp. 792795
19 Pontoh, M., Dalal, V., and Gandhi, N.: Characterization of ECR
plasma, Mater. Res. Soc. Symp. Proc., 2002, 715
20 Das, U., Morrison, S., Centurioni, E., and Madan, A.: Thin lm
silicon materials and solar cells grown by pulsed PECVD technique,
IEE Proc., Circuits Devices Syst., 2003, 150, pp. 282286
21 Kaushal, S., Dalal, V.L., and Xu, J.: Growth of high quality a(Si,Ge):H lms using low pressure remote ECR discharge, J. NonCryst. Solids, 1996, 198200, p. 563
22 Dalal, V.L., and Baldwin, G.: Design and fabrication of graded
bandgap solar cells in a-Si and alloys, Mater. Res. Soc. Symp. Proc.,
1993, 297, p. 833
23 Street, R.A.: Hydrogenated amorphous Si (Cambridge University
Press, 1993)
321
model [2] for fabrication of high efficiency solar cells. However, research of these solar cells was interrupted in the beginning of 1980s because of the degradation of Cu2S layer. In
spite of too many efforts amorphous (a) Si based thin film
solar cells performed degradation as well. The physical
mechanism of this is Stabler-Wronsky effect (interruption of
Si=H bond under illumination).
Then, R & D in this field was continurd on CulnSe2(ClS)
and CdTe films based solar cells. From 1980s until present
their efficiency has been increased significantly from 5-6 %
to 18-19 % for CIS and to 16-17% for CdTe cells. While,
single junction c-Si and GaAs solar cells are approaching
their upper limits in terms of the theoretical maximum efficiency.
Current status of the PV market and recent results on different main materials based solar cells: c- and pc-and a-Si,
Ill-V, II-VI and I-IIl-V12 and their alloys are reviewed in this
report.
1. 1NTRODLICTlON
II. PV MARKET
- 291 -
--
OWorld
0
AJapan
m
.
Company
Kyocera
Sharp
BP Solar
Siemens Solar
Astropower
OUSA
RWE
Sanyo
lsofoton
Mitsubishi
Photowan
0
A
~~
13.5
20
7
4
10.0
4.7
6.5
13.0
6.1
5.7
9.5
10.0
14.0
30.3
32.5
22.2
12.0
50.4
41.9
28
18.
14.0
17
9.S
12.0
14.0
75.0
54.2
39.0
26.0
23.0
19.0
18.0
14.0
i4~n
Technology
,c+,
7-3
,.I %
.SI
!,cl
2-3
,me
c-Si
v..
pc-si
a-Si
14.40
4.0
c-Si
concentrators
Ribbon4
CdTe
CulnSq
Si on low cost subs.
Si on Cz slicc
Tola1
II
5.4
0..33
]I
100.32
I ::;
I .2
]I
18.0
171.22
86.38
2.62
1
111. SILICON
SOLAR
CELLS
- 298 -
3 y O d
I Dcvice
Pc-Si wafers can be fabricated on large area. Plasma processing of lower cost pc-Si is used to form a highly transmissive surface and to increase the light absorption. This known
as reactive ion ekhing allowed about 40 % relative increase
in the absorption. Importance of texturing pc-Si to exhibit
full potential was considered in [9, IO]. 19.8% efficiency
textured pc-Si solar cells was fabricated [ 1 I].Bulk hydrogenation and nitride passivation of cell surface have given
good results. Despite a number of advantages of pc- Si there
is no significant difference between costs of c- and pc-Si solar cells. Commercial cells have efficiency of 12-15 % for
pc-Si wafers.
3.3 Polycrysfalline Thin Film Si Solar Cells
Thin film Si solar cells have the following important advantages in comparison with crystalline cells: I.Thickness of
Si material can be drastically reduced to more than 50pm; 2.
Thin films can be deposited on low cost substrates; 3. Thin
films can be fabricated on module-sized substrates and in
integrally-interconnected structures. According to calculations of [5] thickness of Si films can be reduced up to 1 pm.
Despite different technologies of Si films were developed
in microelectronics many years ago, a long period (around 30
years), has been required to fabricate thin film pc-Si solar
cells with reasonable efficiency. Thin film Si solar cells were
fabricated on ceramic, carbon and glass substrates. Pc-Si thin
films for solar cells with a few pm thickness were deposited
by plasma CVD, hot-wire CVD, rapid thermal CVD, zone melting recrystallization and liquid phase epitaxy. 16.6%
efficiency for small area cells and 11.8% efficiency for large
area module of 238.7~111~havebeen achieved [12-14]. The
cell with a thickness of 2 . 0 p has performed efficiency of
10.7% [15].
I Eff. Yo
I Kef.
c-Si (wafer)
pc-Si (wafer)
24.1
PI
19.8
[Ill
16..6
[12. 131
Cl71
13.7
- 299 -
BP Solar, 5/00
Fin1 Solar, 6/96
Matsushita, 6/00
BP Solar, 4/00
Golden Photon
United Solar, 6/97
BP Solar, 9/96
Kaneka. 9/00
38.0
a-Si triple
35.7
44.3
59.0
53.9
31.0
70.8
56.0
stab.
._/
1.6
10.0
stab.
0.01
5.4
3.7
10.2
8.2
4. I
.COS
a, P . C d T B
,
n-ClnTa*
P-CulnTe,
0.02
junction
CdS/CIGS
4.0
TABLE 5.THERI~ODYXALIIC
EFFICIE~~CY
UNCOHCEUTRATED
SUNLIGHT
LIMITS,
I[:
I
2
30%
I.3
42%
I.9-I .O
2.3-1.4-0.8
2.6-1 .8-1.2-0.8
149%
Infinity
53%
68%
- 300 -
121.
131.
141.
151.
161.
171.
1x1.
191.
1131.
_. .
[IS].
1191.
. .
1201.
1211.
122).
1231.
1241.
1251.
I261.
1271.
1281.
1291.
[301.
1311.
WI.
1331.
PI.
3P-B5-I 1
ABSTRACT
High radiation tolerance of ClGS solar cells is
demonstrated in space for the first timc by monitoring the
performance of Cu(ln,Ga)Scz thin-film solar cells on
MDS-I since February 2002. Short-circuit current of the
ClGS cells did not degrade and open-circuit voltage of the
cells degraded only about 1 %. In contrast, the
performance of other solar cells on the satellite, including
Si and GaAs space solar cclls, degraded I ycar after launch.
The high recovery of radiation damage of CIGS solar cells
due to thermal annealing was found from ground test. We
prcdicted the degradation of the ClGS solar cells in space
using the relative degradation coefficient, the annealing
rates of Voc and Isc for protons irradiating the cells and the
radiation response of the cells without thermal annealing.
The results are in good agreement with flight data of the
CIGS solar cells on the satellite. We were thus able to
demonstrate high radiation tolerance of ClGS solar cells in
space for the first time.
1. INTRODUCTION
A copper indium gallium di-selenide (CIGS) thin-film
solar cell is a promising candidate for future thin-film
space solar cells since the cells havc demonstrated
conversion efficiencies exceeding 18%, which is
significantly higher than other thin-film solar cells [I].
Excellent radiation tolerance of CIGS thin-film cells has
also been reported [2]. In addition, CIGS thin-film solar
cells have better performance than conventional Si and
GaAs space solar cells, such as low cost and light weight,
because the cells can be formed on polyimide [ 3 ] or
stainless steel sheet [4] substrates.
Radiation damage studies for CIGS thin-film solar
cells have revealed that (I) electrical properties of the cells
are not degraded by high-energy electron irradiation.
However, cell performance is degraded by high-energy
proton irradiation similar to other types of solar cells [2],
and (2) radiation damage of the cells due to proton
irradiation gradually recovers when irradiated cells are
kept at room temperaturc [SI. The recovery rate of the cells
depends on the temperature [6]. Since ClGS solar cells
recover well from radiation damage, a prediction of
on-orbit performance of ClGS solar cells must fake into
2. EXPERIMENT
ClGS thin-film solar cell samples (effective area of
0.6 cm') were separated from a 30cmx30cm CIGS
thin-film integrated mini-module. The circuit was
fabricated by a selenizationisulfurization method [ 7 ] , and
the stmcture of the cells is ZnO/Zn(O,S,OH)x/Cu(ln,Ga)Se, iMoiGlass [XI. The solar cells have an
average efficiency of 10.0%. open-circuit voltage (Voc) of
570 mV, short-circuit current density (Jsc) of 40 mAIcm',
and fill factor (FF) of 60%, which were measured under
AMO, I-sun conditions. Thc cclls have lower FF than
typical ClGS thin-film solar cells reported because of their
relatively high series resistance due to their contact
StNCture [ I ] .
Poster
693
B = Bd exp(-At)
(1)
2.5
3.5
1OOOfl (K-)
IO
IO
(a) Short-CircuitCurrent
IO
-m21
::10
>
I , ,
YI
25
35
IOOOiT (K-)
3.3 I n t r i n s i c D e g r a d a t i o n on CIGS S o l a r C e l l s by
P r o t o n s a t Low Temperature
0. I
I
10
Proton Energy (MeV)
Fig. 2 Relative damage coefficient of the ClGS solar
cells for protons derived from Fig. 1 .
0.01
~~
3.2 A n n e a l i n g
694
Poster
Isc
Voc
$0
1x10"
1x10'2
0.15
1.1
(cm.7
0.03
0
L.
I 1
10
100
IO00
MET (days afler hunch)
10000
D=D,-Clog
I+-
4 3
IO
IO0
1000
10000
Poster
695
Osoka.Japan
4. CONCLUSION
We investigated the on-orbit performance of ClGS
solar cells with thermal annealing. The ClGS solar cells
were irradiated with high-energy protons for the radiation
response. We measured the RDC of proton, the thermal
annealing rate of Isc and VOCof the proton irradiated cells
and the intrinsic degradation coeffiicicnts ofthe cells for 10
MeV proton. The thermal annealing rate of the cells
exceeds that of other types of solar cells.
A validation of thermal annealing on orbit is provided
through accurate prediction of data from the cxpcriments
of the MDS-I satellite. Consequently, the performance is
not degraded by thermal annealing. This result is in good
agreement with the flight data collected from cells on the
MDS-I satellite. Sincc the ClGS solar cells continuously
recover due to thermal annealing in space, we assume that
radiation degradation and recovery of ClGS solar cells
occur simultaneously in space. We predicted the
performance of ClGS solar cells in LEO and GEO using
the results and found that cell. Performance was not
dcgradcd at all.
We plan to demonstrate the cell performance of CIGS
solar cells without cover-glasses to validate of the model
by a small satellite in LEO.
5. ACKNOWLEDGEMETS
The authors wish to thank Dr. T. Negami of
Matsushita Electric Industrial CO., LTD., Dr. S. Niki of
AIST, Professor T. Wada of Ryukoku University, and
Professor A. Yoshida of Toyohashi University of
696
Poste1
1% and VOCwith a m a h s
.p.95
oc without a m a
0.9
0.85
,
0.8
, ,.
10
100
1000
10000
0.2
.,...
....
. , , . , ,I
10
100
1000
MET (days afler bunch)
(b) CiEO
I
10000
REFERENCES
[ I ] M.A. Contreras, B. Egaas, K. Ramanathan, J. Hiltner,
A. Swartzlander, E Hasoon, R. Noufi: Prog. Photovolt.
7 (1999) 31 I.
[2] 7. Aburaya , T. Hisamatsu and S. Matsuda, Second
WCPEC, 1998, pp.3568-3571
[3] S. Wideman, M. E. Beck, R. Butchcr, 1. Repins, N.
Gomer, B. Joshi, R. G. Wcndt and 1. S. Britt,
Twenty-ninth IEEE PVSC, 2002, pp.575-578
[4] J. R. Tuttle, A. Sralaj and J. Keane, Twenty-eighth
IEEE PVSC, 2000, pp. 1042-1045.
[ 5 ] A. Bohen, D. Briunig, J. Klaser, F. H. Kang, B.
Hosselharth and G. La Roche, Twenty-eighth lEEE
PVSC, 2000, pp.1038-1041
[6] S. Kawakita, M. Imaimmi, M. Yamaguchi, K. Kushiya,
T. Ohshima, H. Ito, and S. Matsuda, lpn. J. Appl. Phys.,
41 (2002), p. L797
[7] K. Kushiya, M. Tachiyuki, T. Kase, Y. Nagoya, I .
Sugiyama, 0. Yamase and H. Takeshita: Sol. Energy.
Mater. Sol. Cells 49(1997) 277.
[SI K. Kushiya, M. Tachiyuki, Y. Nagoya, A. Fujimaki, B.
Sang, D. Okumura, M. Satoh and 0. Yamase: Sol.
Energy Mater. Sol. Cells 61 (2001) 11
[9] S. Matsuda, M. Imaizumi, S. Kawakita, S. Taishi and K.
Aoyama, this proceeding.
[IO] http://nssde.gsfc.nasa.gov/space/model/
3P-B5-I 1
ABSTRACT
High radiation tolerance of ClGS solar cells is
demonstrated in space for the first timc by monitoring the
performance of Cu(ln,Ga)Scz thin-film solar cells on
MDS-I since February 2002. Short-circuit current of the
ClGS cells did not degrade and open-circuit voltage of the
cells degraded only about 1 %. In contrast, the
performance of other solar cells on the satellite, including
Si and GaAs space solar cclls, degraded I ycar after launch.
The high recovery of radiation damage of CIGS solar cells
due to thermal annealing was found from ground test. We
prcdicted the degradation of the ClGS solar cells in space
using the relative degradation coefficient, the annealing
rates of Voc and Isc for protons irradiating the cells and the
radiation response of the cells without thermal annealing.
The results are in good agreement with flight data of the
CIGS solar cells on the satellite. We were thus able to
demonstrate high radiation tolerance of ClGS solar cells in
space for the first time.
1. INTRODUCTION
A copper indium gallium di-selenide (CIGS) thin-film
solar cell is a promising candidate for future thin-film
space solar cells since the cells havc demonstrated
conversion efficiencies exceeding 18%, which is
significantly higher than other thin-film solar cells [I].
Excellent radiation tolerance of CIGS thin-film cells has
also been reported [2]. In addition, CIGS thin-film solar
cells have better performance than conventional Si and
GaAs space solar cells, such as low cost and light weight,
because the cells can be formed on polyimide [ 3 ] or
stainless steel sheet [4] substrates.
Radiation damage studies for CIGS thin-film solar
cells have revealed that (I) electrical properties of the cells
are not degraded by high-energy electron irradiation.
However, cell performance is degraded by high-energy
proton irradiation similar to other types of solar cells [2],
and (2) radiation damage of the cells due to proton
irradiation gradually recovers when irradiated cells are
kept at room temperaturc [SI. The recovery rate of the cells
depends on the temperature [6]. Since ClGS solar cells
recover well from radiation damage, a prediction of
on-orbit performance of ClGS solar cells must fake into
2. EXPERIMENT
ClGS thin-film solar cell samples (effective area of
0.6 cm') were separated from a 30cmx30cm CIGS
thin-film integrated mini-module. The circuit was
fabricated by a selenizationisulfurization method [ 7 ] , and
the stmcture of the cells is ZnO/Zn(O,S,OH)x/Cu(ln,Ga)Se, iMoiGlass [XI. The solar cells have an
average efficiency of 10.0%. open-circuit voltage (Voc) of
570 mV, short-circuit current density (Jsc) of 40 mAIcm',
and fill factor (FF) of 60%, which were measured under
AMO, I-sun conditions. Thc cclls have lower FF than
typical ClGS thin-film solar cells reported because of their
relatively high series resistance due to their contact
StNCture [ I ] .
Poster
693
B = Bd exp(-At)
(1)
2.5
3.5
1OOOfl (K-)
IO
IO
(a) Short-CircuitCurrent
IO
-m21
::10
>
I , ,
YI
25
35
IOOOiT (K-)
3.3 I n t r i n s i c D e g r a d a t i o n on CIGS S o l a r C e l l s by
P r o t o n s a t Low Temperature
0. I
I
10
Proton Energy (MeV)
Fig. 2 Relative damage coefficient of the ClGS solar
cells for protons derived from Fig. 1 .
0.01
~~
3.2 A n n e a l i n g
694
Poster
Isc
Voc
$0
1x10"
1x10'2
0.15
1.1
(cm.7
0.03
0
L.
I 1
10
100
IO00
MET (days afler hunch)
10000
D=D,-Clog
I+-
4 3
IO
IO0
1000
10000
Poster
695
Osoka.Japan
4. CONCLUSION
We investigated the on-orbit performance of ClGS
solar cells with thermal annealing. The ClGS solar cells
were irradiated with high-energy protons for the radiation
response. We measured the RDC of proton, the thermal
annealing rate of Isc and VOCof the proton irradiated cells
and the intrinsic degradation coeffiicicnts ofthe cells for 10
MeV proton. The thermal annealing rate of the cells
exceeds that of other types of solar cells.
A validation of thermal annealing on orbit is provided
through accurate prediction of data from the cxpcriments
of the MDS-I satellite. Consequently, the performance is
not degraded by thermal annealing. This result is in good
agreement with the flight data collected from cells on the
MDS-I satellite. Sincc the ClGS solar cells continuously
recover due to thermal annealing in space, we assume that
radiation degradation and recovery of ClGS solar cells
occur simultaneously in space. We predicted the
performance of ClGS solar cells in LEO and GEO using
the results and found that cell. Performance was not
dcgradcd at all.
We plan to demonstrate the cell performance of CIGS
solar cells without cover-glasses to validate of the model
by a small satellite in LEO.
5. ACKNOWLEDGEMETS
The authors wish to thank Dr. T. Negami of
Matsushita Electric Industrial CO., LTD., Dr. S. Niki of
AIST, Professor T. Wada of Ryukoku University, and
Professor A. Yoshida of Toyohashi University of
696
Poste1
1% and VOCwith a m a h s
.p.95
oc without a m a
0.9
0.85
,
0.8
, ,.
10
100
1000
10000
0.2
.,...
....
. , , . , ,I
10
100
1000
MET (days afler bunch)
(b) CiEO
I
10000
REFERENCES
[ I ] M.A. Contreras, B. Egaas, K. Ramanathan, J. Hiltner,
A. Swartzlander, E Hasoon, R. Noufi: Prog. Photovolt.
7 (1999) 31 I.
[2] 7. Aburaya , T. Hisamatsu and S. Matsuda, Second
WCPEC, 1998, pp.3568-3571
[3] S. Wideman, M. E. Beck, R. Butchcr, 1. Repins, N.
Gomer, B. Joshi, R. G. Wcndt and 1. S. Britt,
Twenty-ninth IEEE PVSC, 2002, pp.575-578
[4] J. R. Tuttle, A. Sralaj and J. Keane, Twenty-eighth
IEEE PVSC, 2000, pp. 1042-1045.
[ 5 ] A. Bohen, D. Briunig, J. Klaser, F. H. Kang, B.
Hosselharth and G. La Roche, Twenty-eighth lEEE
PVSC, 2000, pp.1038-1041
[6] S. Kawakita, M. Imaimmi, M. Yamaguchi, K. Kushiya,
T. Ohshima, H. Ito, and S. Matsuda, lpn. J. Appl. Phys.,
41 (2002), p. L797
[7] K. Kushiya, M. Tachiyuki, T. Kase, Y. Nagoya, I .
Sugiyama, 0. Yamase and H. Takeshita: Sol. Energy.
Mater. Sol. Cells 49(1997) 277.
[SI K. Kushiya, M. Tachiyuki, Y. Nagoya, A. Fujimaki, B.
Sang, D. Okumura, M. Satoh and 0. Yamase: Sol.
Energy Mater. Sol. Cells 61 (2001) 11
[9] S. Matsuda, M. Imaizumi, S. Kawakita, S. Taishi and K.
Aoyama, this proceeding.
[IO] http://nssde.gsfc.nasa.gov/space/model/
2186
IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 51, NO. 12, DECEMBER 2004
( )
I. INTRODUCTION
ACKSIDE contact solar cells have shown outstanding performance under both one-sun and concentrated illumination. Perhaps the most notable backside contact solar cells are
point contact solar cells [1], which have reached conversion efficiencies of 28.4% at 200-suns [2] and 22.3% at one-sun [3]
conditions. The point contact solar cell is similar to the interdigitated backside contact solar cell in that both electrical contacts
are on the back side [4][6]; however, there is a difference since
its contact metal only touches the silicon in an array of points.
The major advantage of restricting the contact coverage to small
points is that it significantly increases the cell output voltage.
These advantages, combined with the use of a high-resistivity
float-zone silicon substrate, and high-quality SiO passivation
on all surfaces, contribute to the high efficiency of the point contact solar cell. However, the rear contact side of point contact
solar cells often requires four to six mask levels to fabricate [7].
The advantages of the backside contact design include no
contact grid shading on the sunward side and potentially low
Manuscript received April 21, 2004; revised September 15, 2004. This work
was supported by the Australian Research Council, through The Center of Excellence for Advanced Silicon Photovoltaics and Photonics, University of New
South Wales. The review of this paper was arranged by P. Panayotatos.
The authors are with the Center of Excellence for Advanced Silicon Photovoltaics and Photonics, University of New South Wales, Sydney, NSW 2052,
Australia (e-mail: jiun-hua.guo@student.unsw.edu.au; j.cotter@unsw.edu.au).
Digital Object Identifier 10.1109/TED.2004.839870
wafer cost due to the use of a thin wafer, although high diffusion length and low front-surface recombination are required
to maintain high charge carrier collection needed for high efficiency. The contact side can be optimized for series resistance
whereas the sunward side can be optimized for light collection without the need to compromise. Moreover, the module assembly costs are potentially lower as both contacts are on the
rear, and the solar cells can be placed closer together in the
module because there is no need for a space between the cells
for interconnect tabs [8].
Significant cost reduction while maintaining the above-mentioned advantages have been the focus of research on the backside contact solar cells. The Pegasus cell, a simplified backside
point contact solar cell, achieved 21.9% AM1.5G front illumination efficiency with a reduced number of photolithography
masking steps from six to three [9], [10]. The trench-mesa type
of backside contact solar cell demonstrated 21.9% efficiency
with only two photolithographic steps [11].
Several methods have been utilized to fabricate interdigitated
backside contact solar cells commercially. Robert et al. used
the screen-printing technology and found that the cell performance was hindered by the limitations on the contact metallization geometry imposed by the screen-printing process [12].
EBARA Solar used a screen-printing process that incorporated
self-doping contacts, demonstrating an efficiency of 10.4% on
dendritic web silicon substrates [13]. The Back-OECO cell presented by ISFH has achieved a preliminary result of more than
18% efficiency [14]. More recently, SunPower demonstrated an
outstanding backside contact solar cell result with 20.4% efficiency by utilizing low-cost processing without photolithography [15].
Three designs of the backside contact cell have been developed for the implementation on lower lifetime substrates: 1)
emitter wrap-through (EWT) solar cell [16], [17]; 2) metallization wrap through (MWT) solar cell [18], [19]; and 3) metallization wrap around solar cell (MWA) [20]. The EWT cell uses
laser-drilled holes to wrap the emitter on the front surface to the
interdigitated contacts on the back surface. The MWT cell has
front collecting junction and front contact grid with the busbar
brought to the backside through laser-drilled vias in each front
contact finger, and the front grid of the MWA cell is connected
to the backside contact pads via two opposite edges of the cell.
A 41 cm screen-printed EWT cell with 15.3% efficiency under
AM 1.5G illumination has been demonstrated [21]. Efficiency
of more than 17% has been achieved by both the 25 cm MWT
and the MWT solar cells using solar grade CZ wafers [19].
2187
TABLE I
PROCESSES FOR IBBC SOLAR CELLS
Fig. 1. Schematic diagram of (a) IBBC and (b) DSBC solar cells (not to scale).
IBBC structure. Longer groove diffusion drive-in time is preferable because it not only pushes the metallurgical junction far
away from groove walls, preventing nickel contact metal spiking
through the groove diffusion, but also increases the width of
the potential energy barrier to prevent junction tunneling [25].
2188
IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 51, NO. 12, DECEMBER 2004
TABLE II
TYPICAL IMPLIED V OF DSBC AND IBBC SOLAR CELLS
GROOVE FORMATION, AND
AFTER OXIDATION, n
GROOVE FORMATION FABRICATION STEPS
p
Fig. 2. Schematic diagram of the rear electrode pattern showing the two
interdigitated electrodes (not to scale).
2189
TABLE III
RESULTS OF A SPLIT BATCH OF DSBC AND IBBC OLAR CELLS
3 Solar cells are 8 cm and have been cleaved from their host wafers.
Fig. 4. Effect of the wafer thickness on the performance of IBBC solar cells.
of the DSBC
As shown in Table III, front illumination
solar cells is about 310 mV lower, and
is about 0.6 mA/cm
smaller compared to those of IBBC solar cells. It is found that
in DSBC solar cells mainly results from higher
smaller
front-surface recombination velocity of the boron diffused selective emitter. PC-1D simulation using the surface recombination velocities extracted in the previous section confirms this
as well. The DSBC solar cells exhibit
of
difference in
32.2 mA/cm due to contact shading loss (approximate 2.5%)
compared to the value obtained from the simulation of the IBBC
solar cells under front illumination (33 mA/cm ). The main reathan front illumination
sons for smaller rear illumination
in DSBC solar cells are: 1) the collecting junction is located
on the front surface and 2) the contact shading of the rear base
contacts is larger. The electroless nickel-plating rate is higher in
the phosphorus-doped grooves, and the subsequent electroless
copper plating is always thicker, leading to a naturally higher
2190
IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 51, NO. 12, DECEMBER 2004
TABLE IV
CELL PARAMETERS USED FOR PC-1D SIMULATION
Fig. 5. Effect of the bulk lifetime on the performance of IBBC and DSBC solar
cells. Modeling parameters are listed in Table IV.
metal contact shading on the rear side contacts. Fill Factor in the
DSBC and IBBC devices is relatively low due to both shunting
and series resistance. The detailed cause of the parasitic effects
is currently under investigation.
High bulk lifetime and low front surface recombination velocity are essential for high efficiency in IBBC and DSBC solar
cells. While these conditions can be maintained in a laboratory,
it is much more challenging to do so in a commercial environment. To examine the impact of the bulk lifetime and the front
surface recombination on the performance of the IBBC and the
DSBC solar cells, we extended our experiment using the PC-1D
simulation based on measured and extracted device parameters.
Fig. 5 shows the influence of the wafer bulk lifetime on the
performance of the IBBC and the DSBC solar cells. Modeling
parameters are listed in Table IV. Here, it is clear that the backside contact design relies on high quality wafers to obtain the
benefit of no grid shading. In the case of low wafer lifetime, the
DSBC device is superior, as expected, owing to its front-side
collecting junction. Alternatively, a very thin wafer can be used
in order to maintain a high ratio of minority carrier diffusion
length to wafer thickness, although thin wafers tend to introduce higher mechanical yield losses.
Fig. 6 shows the effect of the front surface recombination velocity on the IBBC and the DSBC solar cells. Modeling parameters are listed in Table IV. It is expected that the decrease in
of the IBBC design with increasing front surface recombination is much more significant than that of the DSBC design.
However, one must keep in mind that the IBBC design has a
lower inherent front surface recombination velocity because it
uses a phosphorus diffused front surface field, which results in
a substantially lower effective surface recombination velocity
compared to the boron diffused emitter on the front surface of
the DSBC design [35], [36]. The overlaying circles on the
curves indicate the extracted effective surface recombination velocities of the experimental devices made in a split batch, and
this highlights the importance of the inherent differences in the
front surface recombination on the performance of the IBBC
and the DSBC designs.
Two important observations can be made from the simulation
results presented in Figs. 5 and 6. Firstly, the DSBC design is superior to the IBBC design in that it is more tolerant to variations
in wafer lifetime and in front surface recombination velocity,
which is a significant consideration for commercial production.
2191
>18%
2192
IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 51, NO. 12, DECEMBER 2004
' University ofKassel, Wilhelmshoeher Allee 73, D-34121, Kassel, Germany, nhenrel~uni-kassel.de
' DaimlerChrysler AG, Wilhelm-Runge-Strasse 11, D-8908 I Ulm, pascal.hofmann@daimlerchrysler.com
' Institute for Solar Energy Supply Technology, Koenigstor 59, D-34119, Kassel, Germany, cbendel@iset.uni-kassel.de
Abstract - This paper describes the application of photovoltaic (PV) solar cells in planar antenna stmctures. The
radiating patch element of a planar antenna is replaced by a
solar cell. Furthermore radiating slots are built due to the
cell spacing in a solar cell array. The original feature of a
solar cell (DC current generation) remains, but additionally
the solar cell is now able to receive and transmit electromagnetic waves. Both single solar cells as well as solar cell
arrays can be used as antennas. At first some basics regarding the RF properties of solar cell patches are discussed.
The efficiency of solar cell antennas is measured by means
of the Wheeler Cap method, which is investigated by means
of numerical simulations. The lossy substrate of the feed
lines has the strongest influence on the antenna efficiency.
Finally two applications are presented: a solar cell GPS antenna for vehicular applications and a solar cell slot antenna
for mobile communications (GSM). The aim is to integrate
these antennas into vehicular glass roofs which are covered
with photovoltaic solar cells in order to deliver the electric
power for the indoor ventilation of the car. The GPS antenna
provides circular polarisation and a main lobe in zenith direction whereas the GSM antenna is vertically polarized and
has a monopole-like radiation pattern. Both antennas are
built up with commonly used solar cells. The comparison of
measured and simulated antenna properties shows a good
agreement.
Keywords - planar antenna, solar cells, simulation, mobile
communication, vehicular application, Wheeler Cap
I. lNTRODUCTION
Wireless communication systems require electric energy for
their operation and their use. Stand alone applications like
environmental monitoring systems, vehicular communication systems or satellite systems need a net-independent
power supply which is preferably realisable by photovoltaics, an advanced technology distinguished by reliability,
longevity and eco-friendliness. Besides that antennas are
needed in order to receive or transmit electromagnetic
waves.
At present, photovoltaic generator and antenna are two separate devices. They compete for the available space on mobile and stand alone systems, which are generally limited in
size. Furthermore they may be bulky and expensive and they
NA
NA
622
, , .,..I
....,....I
RRAi~
I=
RR4D + RLOSS
1,0E+9
1.5E+9
2.E-S
Z.5PO
W E 4
FmuencylHZ
L-Me.llbpl.te,
Re,Z)-Solarse,,,
SIIWC =
IrniZ)J
B. Ef$ciency considerations
Efficiency measurements are performed by means of the
Wheeler Cap method. The method described in [4] and [ 5 ]
was expanded to measurements of aperture coupled patch
antennas. The test equipment is shown in Figure 3.
RLOSS
- R.s
RLOSS + Rs
(3)
(4)
As can be seen from (2) and (3) the used S-parameter in (4)
are real. That means the measurements must be made at
resonance. In this study the efficiency of aperture coupled
patch antennas is measured. The aperture is excited with a
microstrip line of length 1. Due to the transformation properties of microstrip lines the real feed point impedance Ge,at
resonance (which is the impedance at the aperture location)
is transformed to the complex port impedance I&, according to
where
I/&
In order to verify the Wheeler Cap measurement the complete test setup was simulated by means of numerical simu-
623
lations with Microwave Studioa. The antenna substrate (FR4) was modelled with a loss tangent of tan&O.OIJ and a
permittivity of ~ , . = 4 . 3 .
1
0.8
0.6
0.4
!i .,
5
-0.2
p -0.4
5 -0.6
48
13W
1400
l500
1200
1700
16W
la00
Frequency IMHz
~--ReallSl1). Sim. -1rnaglS11).
Sim.-Resl(Sl1/.
2 -0.4
e -0.6
-0.8
1
lZoo l 3 O o
1400
1700
l6O0
1500
l6O0
Frequency IMHZ
~-Rsal(Sl1).Lm.-lmag(Sll).S~rn.
-Real(Sll).Mear.
-Imag(Sli).Mess.l
Slips
Sllwc q
Direct calculation
Copper patch
-0.2
-0.18
-0.75 79%
-0.76 80%
--
-0.14 -0.67
74%
--
78%
Simulated and measured efficiencies show a good agreement. The direct calculation is the total efficiency that
624
Similar to the GPS antenna in the previous section this antenna is also designed for the integration into vehicular glass
roofs in order to provide a dual use of the solar cells as
electric power generator and antenna respectively. The aim
was to design antennas both for the GSM 900 and GSM
1800 band. The fixed size of the solar cells complicates the
construction of a dual mode antenna therefore a single hand
antenna for the GSM 1800 band is presented here. This integrated planar antenna device should substitute conventional
rod antennas on the car roof. Thus the requirements of the
antenna are vertical polarisation in the horizontal plane and
a monopole-like gain and radiation characteristic.
The concept of the present antenna ties up to flush disc antennas as presented in [7].Basically it is a rectangular slot
antenna as can be seen in Figure 8 and Figure 9. Since standard solar cells are used with an edge length of IOOmm the
dimension of this patch and the arising slot around the patch
doesnt correspond to a operating frequency of 1800 MHz.
V. CONCLUSIONS
These studies have shown that a fully integration of planar
antennas and solar cells in only one device is possible. The
investigated antennas are thin and suitable for vehicular applications. However solar cell antennas can be utilized in
many other areas. The Solar Planar Antenna - SOLPLANP
625
Gain
Gain
27V
Maaweman. Max :dBilF 3.m
REFERENCES
N. Henze, C . Bendel, J. Kirchbof and H. Friichting:
Application of Photovoltaic Solar Cells in Planar Antenna Structures, 12. International Conference on Antennas and Propagation ICAP 2003, Exeter UK, 3 Ird
March - 3 April 2003, pp.731-734.
[2] C. Bendel, N. Henze and 1. Kirchhof: Solar Planar Antenna - SOLPLANT, 17th European Photovoltaic Energy Conference, Munich 22-26 Oct. 2001.
[3] C. Bendel, N.Henze and J. Kirchhof: Apparatus for
converting solar e n e r a into electrical energy andfor
radiating and/or receiving highfrequency electromagnetic waves,United States Patent 6,395,971, Aug. 2000
(US), Deutsches Patent- und Markenamt, Patentschrift
DE 199 38 199C1, Aug. 1999 (DE). .
[41 D. Agahi and W. Domino: Efficiency Measurement of
Portable Handset Antennas using the Wheeler Cap, AppliedMicrowave & Wireless, Vol 12, No. 6, pp. 34-42,
June 2000.
[ 5 ] H. Wheeler: The Radiansphere around a Small AIlenna, Proceedings o f t h e l m , Vol. 47 pp. 1325-1331,
August 1959.
[ 6 ] C. Bendel, J. Kircbhof and N.Henze: ,,Application of
Photovoltaic Solar Cells in Planar Antenna Structures,
3. World Conference on Photovoltaic Energv, Osaka,
12.-16. Mai 2003.
[7] N. Henze, A. Giere, H. Friichting and P. Hofmann:
GPS Patch Antenna with Photovoltaic Solar Cells for
Vehicular Applications, 58th IEEE Vehicular Technologv Conference Fall 2003, Orlando, US, pp. October 69,2003.
[8] Ziircher, J.-F. and Gardiol, F. E.: BroadbandPatch Antennas, Artech House, Boston-London, 1995.
[9] K. Kaneta, T. Kondo, M. Ando and N. Goto: A Flushmounted Antenna for Mobile Communications, IEEE
AP-S. Syracuse, N.Y., pp. 1323-1326, June 1988.
[I]
1W
Sirnukhan, Mar IdBC =-0.91 M B B S U ~ ~6 aTx~rail
~ . =-0.74
90
-042
300
210
AKNOWLEDGEMENTS
This work has been carried out in cooperation with DaimIerChrysler AG, Ulm, Germany. Special thanks to Pascal
Hofmann for his valuable support of this work.
626
1. INTRODUCTlON
I25
Glass substrate
Substrate
Holder
I26
-.-EE
2.0
0.0
v)
-2.0
-0.4
-0.2 0
0.2
0.4
0.6
0.8
1.0
Voltage [ V ]
Fig.3 I-V characteristics of the solar cell in the dark
(a) and under AMI .5G 100mWlcm2 (b).The
conversion efficiency : 2.35% , short-circuit current :
5.06mA , open xircuit voltage : 0.821V , Fill factor :
0.566 (c) Spin coating cell I-V curve , Efficiency 2.3%
MDMO-PPVIPCBM
blend byer
10 urn
Blend laver
(a)
. . Scren printing
10 Bm
127
Initial
0000000000
a
Squeezing
around 20 micron
Acknowledgements
We are grateful to Y. Kim and J. R. Durrant from Imperial
College for the fundamental instruction of MDMOPPV/PCBM solar ceIls.
Leveling
a
Solvent Evaporation
References
[I]
N.S.Sariciftci, LSmilowitz, A.J.Heeger, and F.Wudl,
'Photoinduced electron-transfer from a conducting
polymer to buckminsterfullerenre",
Science 258, 1474 (1992)
121 J.J.M.Halls, C.A.Walsh, N.C.Greenham. E.A.Maeseglia.
R.H.Friend, S.C.Moratti and A.B.Homes, "Efficient
photodiodes from interpenetrating polymer networks",
Nature 376, 498(1995)
[3] C.J.Brabec, N.S.Sariciftci and J.C.Hummelen,
"Plastic solar cells",
Adv. Func. Mater.11, 15, (2001)
141 S.E.Shaheen, R.Radspinner, N. Peyghambarian,
G.E.Jabbour. "Fabrication of bulk heterojunction plastic
solar cells by screen printing".
Appl. Phys. Lelt.79,2996 (2001)
[5] T.Aemouts. P.Vanlaeke, J.Poortmans and P.
Heremans , "Polymer solar cells: screen printing as a
novel deposition technique" , 2004 MRS fall Meeting
[6] A. Zakhidov, K. lnoue, S. Lee, Mike Sampson,
CPallavi and R. Ulbrich ,"Temperature and Time
Dependence of Heat Treatment on P3HTIPCBM
Polymer 3- Hexalthiophine soluble C60 Solar Cells"
2004 MRS fall Meeting
~ ~ u n d
\
128
ANALYSIS OF THE RADIATION HARDNESS OF TRIPLE- AND QUINTUPLEJUNCTION SPACE SOLAR CELLS
C. Baur, M. Meusel, F. Dimroth, A.W. Bett, M. Neil', G. Strobl, S. Taylor3, C. Signorid
Fraunhofer Institute for Solar Energy Systems, D-79110 Freiburg, Heidenhofstr. 2, Germany
RWE Space Solar Power GmbH, 74072 Heilbronn, Germany
European Space Research & Technology Centre. Keplerlaan 1, NL-2200 AG Noordwijk, The Netherlands
email: carsten.baur@ise.fraunhofer.de
1
ABSTRACT
The Fraunhofer ISE has developed a characterization
RESULTS
35 cells
It is well known that the GalnP and Ge layers show a
superior radiation hardness compared to the GalnAs
middle cell [5,6]. This can also be seen in Fig. 1 (EQE)
and Fig. 2.
1001,
.
,
, .
,
, .
. , .I
1
80
-E
INTRODUCTION
60
0-7803-8707-4/05/$20.00
02005 EEE.
40
20
Wavelength [nm]
Fig. 1: External quantum efficiency (EQE) of top and
middle cell of the 35 cell 15852-9 before (BOL) and after
irradiation (EOL) with 1 MeV electrons at a fluence of
1-105cmS. Only the top and the middle cell EQEs are
shown since the bottom cell has a lot of excess current
and is known to be very radiation hard.
In Fig, 1 the external quantum efficiency (EQE) of the
top and the middle cell of a 35 cell before (BOL) and after
irradiation with 1 MeV electrons at a fluence of l.1015cmh2
(EOL) are shown. White the EQE of the top cell remains
almost constant the EQE of the middle cell shows a high
degradation. The remaining factors of the photocurrentsof
top and middle cell calculated from spectral responsesare
0.97 and 0.78, respectively. In Fig. 2 the spectrometric
characterization of the same cell is shown. tn the course
of a spectrometric characterization several I-V
measurements are performed under different spectral
548
JZ
JL
where
and
are the photocurrents of the top
and middle cells and E is the spectral irradiance,
According to Eq. (1) the reference point at 1.0
corresponds to an I-V measurement under AM0
conditions. Moving from left to the right on the x-axis the
photocurrent of the top cell is increased while the
photocurrent of the middle cell is decreased at the same
time thereby always satisfying Eq. (9).
= =
0.95
1.00
-=
0.90
J"tE(h.))
1.05
1.10
J*"F,tU)
with XCM being the x value where the cells are current
matched. Thus, the middle cell generates 7.3 % more
photocurrent at AM0 compared to the top cell. This is in
perfect agreement with the results obtained from EQE
549
5J cells
The 55 cell was developed to improve the radiation
hardness. The diffusion length in the tayers is the
parameter mainly affected by particle irradiation. The
radiation hardness in a 55 cell should be improved by
using thinner absorber layers. Thus, the diffusion lengths
may be lower without losses in photocurrent.
T38
5 %
E34
32
7-
(,J = 7.3 w v c m 2 )
( FF = 86.3 % )
( eta = 24.0 % )
U)
U
c.
1
E2.641%E=+u=q9p%5
-n-
'
2.62
' V
0.90
Jpm
c
0.95
;
1.00
AMO, 136NV/m2.T=25"C
0
,
1.05
(EIW JpmlOPIE,(M
= 5.196 V
V,
.-
Voltage
1.10
.''r.
2nd
100
-O--m-
400
600
700
Wavelength [nm]
500
800
900
E
> 2.50
0.90
0.95
1.00
1.05
1.10
Jpm""(E(m I JW""(E,,(U
550
3J
5J
Imv]
[mq
2638
5196
2516
4937
ACKNOWLEDGEMENTS
This work was partry supported by the European
Space Agency (ESA) through the funding of the project
"Development of next generation GaAs based multi
junction solar cells" (Ref. N.: A0/3-1f053/04/NUFM).
[%I
95.4
95.0
REFERENCES
[ I ] M. A. Stan, D. Aiken, P. R. Sharps et al., "The
Development of >28% Efficient Tnple-Junction Space
Solar Cells at Emcore Photovoltaics", froc. of 3rd Wudd
Conference on Photovofiaic Energy Conversion,2003, pp.
662-665.
i
400
I
I
I
.
6M3
8M)
3-"3
F]M.
SUMMARY
55 1
PERFORMANCE OF HIGH-EFFICIENCY ADVANCED TRIPLEJUNCTION SOLAR PANELS FOR THE LILT MISSION
DAWN
Navid S. Fatemi, Surya Shana, Oscar Buitrago, John Crisman & Paul R. Sharps
Emcore Photovottaics, 10420 Research Rd., SE, Albuquerque, NM 87123, USA
Ron Blok, Martin Kroon 8r Cees Jatink
Dutch Space, P.O. Box 32070,2303 DB, Leiden, The Netherlands
Robin Harris
Orbital Sciences Corporation, 21839 Atlantic Blvd., Dulles, VA 20166, USA
Paul Stella & Sal Distefano
Jet Propulsion Laboratory (JPL), 4800 Oak Grove Dr., Pasadena, CA 91 109, USA
ABSTRACT
Lavncl?mn8@"wl
0-7803-8707-4/05/$20.0002005 IEEE.
pr
618
I":-h
>l"Cetk
n.0
W.0
A0
-ny
(W
619
Sun Distance
(AU1
'1.oo
Intensity
(MO)
'
Required
Fowl3 (W)
Calculated
POWW (W)
1.000
3,600
f0,3f7
0.319
3,650
3,804
0.197
2,350
0.148
0.111
1,770
1.300
1,781
1.323
620
2.00
I
9 1.50
c
10.50
U
0.00
WMO)
t .oo
0.277
0.20
0.125
5,568
1,057
622
5,852
1.551
SUMMARY
1,082
663
based on analysis.
150.0
200.0
VOLTAGE (VOLTS)
1.495
50.0
0 .o
Predictad Power -
1.00
Intensity
2.50
[5] V.G. Weirer 8 J.D. Broder, 'On the cause of the flatspot phenomenon observed in silicon solar cells at low
temperatures and low intensities," J. Appl. fhys. Vol. 53
(8),Aug. 1982, pp. 5926-5930.
Mass measurements
Protection diode functionality lnfra-red (IR) test
Insulation resistance
Functionality of bleed resistors, blocking
diodes & thermaf sensors
62 1
ABSTRACT
Ttis paper briefly introduces silicon based thin film solar
cells: amorphous (a-Si:H). microcrystalline (pc-Si:H)
single junction and a-Si:H/pc-Si:H tandem solar cells. The
major difference of a-Si:H and pGSi:H cells comes from
electro-optical properties of intrinsic Si-films (active layer)
that absorb incident photon and generate electron-hole
pairs. The a-Si:H film has energy band-gap (Eg) of 1.71.8eV and solar cells incorporating this wide Eg a-Si:H
material as active layer commonly give high voltage and
low current, when illuminated, compared to pc-Si:H solar
cells that employ low Eg (l.leV) material. This Eg
difference of two materials make possible tandem
configuration in order to effectively use incident photon
energy. The a-Si:H/paSi:H tandem solar cells, therefore,
have a great potential for low cost photovoltaic device by
its various advantages such as low material cost by thinfilm structure on low cost substrate instead of expensive
c-Si wafer and high conversion efficiency by tandem
structure. In this paper, the structure, process and
operation properties of Si-based thin-film solar cells will
be discussed.
1. INTRODUCTION
Silicon based thin-film solar cells (Si-TFSC) have been
promising candidate for low cost and high efficiency
photovoltaic devices. Although wafer-based crystalline
silicon (c-Si) solar cells have possessed more than 90%
of world PV market over the fast few decades['], high
material cost and feedstock shortage of c-Si substrates
have been still obstacles for wide spread use of c-Si solar
cells. On the contrary to c-Si celts, Si-TFSC has a great
potential for reducing material cost since it use very thinfilms on low cost substrate such as glass, stainless steel
sheet, plastic, etc, instead of using expensive thick (250350 pm) c-Si wafer. Regardless of this low cost potential
of Si-TFSC, the market possession is still low. The
amorphous silicon (a-Si:H) thin-film solar cells has been
introduced early in the 1970s and market share has been
more than 25% in 1990 but drastically decreased to about
6% in 2003[']. The main reason for this reduced market
possession of a-Si:H solar cells might be resulted from
low module efficiency and stability problem (StaeblerWronski Effect)121,
by which the production cost ($Iwp) of
a-Si:H solar cells has not been competitive with c-Si solar
cells. Although a-Si:H thin-film modules have similar or
slightly less production cost over the c-Si cells, the
installation area should be more than double in order to
0-7803-8707-4/05/$20.00
02005 IEEE,
L:!.i
I552
a-SiC:H, pc-Si:H single layer and pc-Si:H/a-SiC:H bylayer are employed. The effectof different p-type window
layers on a-Si:H solar cells will be discussed in the
following section. The tandem solar cells composes of
series connection of pin a-Si:H top-cell and pc-Si:H
bottom-cell. The ZnO buffer between top and bottom cells
reportedly gives higher current to topcell by intemal
reflection of incident light at ZnQ buffer, by which
degradation (SW-effect) of a-Si:!+ top cell can be reduced
by making a-Si:H intrinsic layer thin. The Ag and ZnO/Ag
are used as a back contact.
'"
1553
U,
20
absorbing layer.
Fig. 4 shows the illuminated I-V curves and spectral
response of pin pc-Si:H solar cells with different light
trapping scheme by changing front ZnO roughness and
structure of back reflector. The figure (a) through (c) show
the effect of front ZnO films, (c) and (d) represent the
effect of back reflector. The solar cell without any light
trapping. flat ZnO and Ag back contact, show low jsc
(14mAlcm2) and VOC(490mV). On the other hand, solar
cell performances linearly improve by just enhancing light
scattering at a surface of ZnO films. The increase of jsc
with surface texturing of ZnO films resulted from the
longer travel length of incident light within the i-layer.
However, the improvement of VOCand FF has not we[l
understood until now. The results by Nasuno el all4] is
opposed to our experimental results. They have observed
the reduction of VOCand FF in solar cells deposited on
textured ZnO films, which is caused by increased
crystalline defects within the poSi:H films deposited on
textured ZnO and Asahi-U glass. The insertion of ZnO
thin-layer behnreen n a-Si:H and Ag back metat is very
effective for improving jsc of solar cell as shown in figure
(c) and (d). It should be noted that high jsc (22mAlcm2)
and efficiency (8.5%) can be achieved in a solar cell with
thickness of only lpm i-layer by enhancing light trapping
scheme. The 4mAlcm' increase of jsc with optimizing
.
.
.
3(D
54m
600 700
900
900
1OOO1IM
WaVeh$th(ma)
1554
and A. Nakajima, Technical Digesf of the fntemafional PVSEC12, Jeju, Korea. 547,2001
[g] J. Meier et al. Solar Energy Mafe!fa/s & Solar Cells 49, 35,
1997
[9] Y. Tawada. H. Yamagishi, K. Yamamoto. Solar Energy
Materisls and Sdar Cells 78, 647-662,2003
[IO] Franc Smote, Marko Topic, Jore furlan. J. of Non-Crystalline
Sdids 194,312-318,1996
[ll] M. Kubon. E.Boehmer, F. Siebke, B. Rech, C. Beneking and
H. Wagner, Sdar Energy Materials and Solar Cells 41-42, 485492,1996
[ 121 Corinne Dror, Ph. D. Thesis, Universityde Neuchatel, 2003
[I31 H. Stiebig. T. Brammer, J. Zimmer, 0. Vetted. H. Wagner, J.
ofNon-CrysfaI/imSdaS 266-269, 1104-1 108,2000
[I41 Y. Nasuno, M. Kondo and A. Matsuda, Mat. Res. Symp.
P~oc.W. 664, 2001, A15.5.1-A15.5.11.
1A
13 12
- - - - -I- - - - - __ _ _ -
A-)-A
,
-.-
E :
Augbss (tandem)
1.1
-19
-+&ON
$11:
03:
tV. CONCLUSION
-.1
OJ-
a7
M
.,
4
RETERENCES
[l] Paul Maycock, Renewable Energy World, July-Augusf 2003,
pp. 84-101.
[2] D.L. Staebler and C. R. Wronski, Appl. Phys. Lett. 31, 292294,1997
[3] .I. Meier, S. Dubail, R. Fluckiger, D. Fischer, H. Keppner, A.
Shah, I" WCPEC, Dec. 59,409412, 1994
[4] 0.Velterl, F. Finger, R. Carius, P. Hakpe, L. Houben, 0.
Kluth, A. tambertz, A. Muck, 6.Rech, H. Wagner, Solar Energy
Materia/$&Solar Cells 62, 97-108, 2000
[5] N. Wyrsch, C. DOE, L. Feitknecht, M. Goerlitrer. U. Krolf, J.
Meier, P. Tortes, E. Vallats-Sauvain, A. Shah, M. Vanecek, MRS
Sping Meeting 2000, Vol. 609
[6] R. W. Collins, A. S. Ferlauto, G. M. Ferreira, Chi Chen,
Joohyun Koh, R. J. Koval, Yeeheng Lee, J. M. Pearce, C. R.
Wronski. Solar Energy Materials & Solar Cells 78, 143-180,2003
[7] K. Yamamoto. M. Yoshimi, Y. Tawada, S.Fukuda, T. Sawada.
T. Msguro, H. Takata, T.Suezaki. Y . Koi, K. Hayashi, T. Suzuki
Fig. 5. IV parameters under AM 1.5 illumination of aSi:H/pc-Si tandem solar cells as a funcfion of
a-Si:H top-cell thickness.
1555
Institute of Physics, Academy of Sciences of the Czech Republic, Cukrovarnicka 10, 162 53 Prague 6, Czech Republic
2
Solartec s.r.o., Televizni 2618, 756 61 Roznov pod Radhostem, Czech Republic
ABSTRACT
Fourier-transform Photocurrent Spectroscopy (FTPS)
was introduced four years ago [1] as a method for fast and
very sensitive evaluation of the spectral dependence of
the optical absorption coefficient of photoconductive thin
films and recently also for the quality assessment of thin
film silicon solar cells [2]. In this contribution we refer
about the FTPS characterization of different thin film solar
cells (amorphous silicon, microcrystalline silicon,
micromorph tandem, polymer solar cells) with the main
target of a fast quantum efficiency (QE) measurement.
Further, we study the possibility of the FTPS-QE
measurement of single cell incorporated in the solar
module when only the module terminals can be used for
contacting. The FTPS measurement has proved to be very
fast (time scale of seconds), enabling a quick verification
of quantum efficiency and subgap absorption
measurement of absorber of different thin film solar cells.
sapphire window
mirror
FTIR
spectrometer
sample
(thin film cell)
INTRODUCTION
Quantum efficiency measurement belongs to the group
of solar cell characterization providing complex information
about its quality. It is used both in the research area of a
new cell structure testing as well as for the reproducibility
verification in production of solar cells and modules.
Relatively long measurement of QE curve by standard
QE set-up (light sources, monochromator, chopper, lock-in
amplifier) is disadvantage of this method. In this paper we
refer about the new possible QE set-up which affords to
obtain the data within few seconds and thus it gives the
potential for its integration into the fast and easy-to-use
method for global cell characterization in the R&D area or
production process.
Basic idea of the new method utilizes time-variables
spectrum of the illumination performed by FTIR
interferometer. Response of the measured solar cell is
then compared with the background spectrum from the
absolute detector or calibrated solar cell. A scheme of
the utilization of the Fourier transform infrared (FTIR)
spectrometer in the FTPS set-up is shown in Fig.1. FTIR
spectrometer equipped for the near infrared and visible
region (quartz beamsplitter and white light source) has
possibility of an external light beam output and an
Current
preamplifier
A/D convertor
and interface
Fig.1. Scheme of the Fourier Transform Photocurrent
Spectroscopy (FTPS) set-up used for the QE
measurement.
EXPERIMENTAL DETAILS
We have used the FTPS method for the QE
measurement to characterize different thin film solar cells
(amorphous, microcrystalline and micromorph silicon and
organic solar cells). First of all, QE-FTPS results were
carefully compared with the independent measurement
obtained by standard QE set-up. This comparison was
done for the p-i-n structure of a small area microcrystalline
silicon solar cell having conversion efficiency of 8.3 %.
1525
f F = 2vs
100
80
60
40
20
400
1000
1200
0.8
800
wavelength (nm)
EXPERIMENTAL RESULTS
For the all FTPS measurements in the range 400
1200 nm the set of various optical filters has been used to
optimize the light intensity in the whole spectrum and by
this way to limit the necessity to measure over the high
dynamic range. The baseline measurement has also been
taken with the same filters, but due to the sensitivity of the
c-Si reference detector (contrary to the thin film solar cells)
also in the near IR region, special care had to be applied
for the blue/violet region of a low light intensity.
Figure 2 shows the single beam FTPS signal of a
microcrystaline silicon p-i-n cell together with the FTPS
result of the calibrated c-Si detector. As it was mentioned
above the spectra were scanned with the same blue
filters reducing the red light intensity and thus having the
relatively flat response for the measured thin film cells.
600
0.6
noise from
He-Ne laser
0.4
0.2
0.0
400
600
800
1000
wavelength (nm)
Fig.3 Quantum efficiency results evaluated from standard
QE set-up and using fast FTPS measurement.
Final curve is compared with the independent QE
result measured by standard QE set-up and very good
agreement can be seen except the slightly noisy result
close to the UV range due to the extremely low light
intensity from halogen lamp inside the spectrometer.
1526
90
60
50
40
20
10
0
400
500
100
800
10
50
40
20
700
60
30
600
wavelength (nm)
70
70
30
relative QE
80
relative QE
80
QE results measured
in 6 different spots
of the a-Si:H module
90
a-Si:H mini-module
(5 sub-cells)
"partially"
microcrystalline
region
0.1
10
0
400
500
600
700
0.01
800
wavelength (nm)
Fig.4 Relative quantum efficiency results of the measured
spot of one a-Si:H solar cell incorporated into a solar
module. Heavy line represents result when directly
measured cell has been contacted, dotted line shows
result where the module terminals and white light bias for
non-active sub-cells were used.
The experiment described above was used for
homogeneity study of the large area a-Si:H solar module
(30 x 30 cm) grown close to the amorphousmicrocrystalline conditions consisting of 27 series
connected single junction cells. Non-homogeneity in
thickness of the active layer or of the top TCO is usually
well visible by a naked eye due to the light interference
effect. Here we investigated influence of nonhomogeneities of three sub-cells measured in two different
spots on the local quantum efficiency. Fig.5 shows these
results in the linear scale, while Fig. 6 shows the same
data plotted in logarithmic scale of QE from which
differences in the material properties are better visible.
1E-3
400
"fully" amorphous
region
500
600
700
800
wavelength (nm)
Fig.6 Relative QE results of a-Si:H module (data from
Fig.5) in logarithmic scale grown close to the amorphousmicrocrystalline regime measured in 6 different places
over the module area.
1527
120
-6.4 V
80
relative QE (a.u.)
relative QE (-)
100
60
a-Si:H cell
40
20
-4.8 V
80
40
20
0
600
800
-1.6 V
mc-Si:H cell
400
-3.2 V
60
1000
1200
400
500
600
700
800
wavelength (nm)
wavelength (nm)
Acknowledgement
This work was financially supported by the Czech Ministry
of Environment (project VaV-SN-3-172-05), EU projects
LPAMS (INCO-CT-2004-509178) and ATHLET (019670),
BIL Cooperation Flanders - Czech Republic BOF04B03
and the Institutional Research Plan AVOZ10100521.
REFERENCES
[1] M. Vanecek, A. Poruba, Appl. Phys. Lett. 80 (2002)
719
[2] A. Poruba, J. Springer, L. Mullerova, M. Vanecek, T.
Repmann, B. Rech, J. Kuendig, N. Wyrsch, A. Shah, Proc.
3rd World Conference on Photovoltaic Energy Conversion,
Osaka, Japan, 2003
[3] J. Meier, S. Dubail, J. Cuperus, U. Kroll, R. Platz, P.
Torres, J.A. Anna Selvan, P. Pernet, N. Beck, N. Pellaton
Vaucher, Ch. Hof, D. Fischer, H. Kepner and A. Shah, J.
Non-Crystal. Solids 227-230 (1998) 1250
[4] N. S. Sariciftci, L. Smilowitz, A. J. Heeger, and F. Wudl,
Science 258, 1474 (1992)
[5] F. Padinger, R. Rittberger, and N. S. Sariciftci, Adv.
Funct. Mater. 13, 85 (2003)
[6] C. J. Brabec, Sol. Energy Mater. Sol. Cells 83, 273
(2004)
[7] L. Goris, A. Poruba, L. Hodakova, and M. Vanecek, K.
Haenen, M. Nesladek, P. Wagner, D. Vanderzande, L. De
Schepper, and J. V. Manca , Appl. Phys. Letters 88,
052113 (2006)
[8] W.D. Mihailetchi, L.J.A. Koster, J.C. Hummelen, and
P.W.M. Blom, Phys. Rev. Letters 93, 21 (2004)
1528
Fasooniehchil, M. Taherbaneh23
ModeL"i[3].
I. INRODUCTION
A. Space Environment
In this paper, standard conditions are assumed Air Mass Zero
Solar cells, in space, are mainly affected from irradiance
level, temperature and particles radiation [1,2]. Fig. 1 shows or AMO, with irradiance and temperature values of 1367 w/m2
the electrical characteristics of solar cells that we have used to and 28 C [2,4]. Variation of space environmental parameters
in two orbits is determined as follow:
describe cell behavior in space applications.
Cyclic profile for irradiance in LEO and GEO is used,
because of daylight and eclipse times [1,2]. Ambient
I
MZIXifllU]fl
temperature is calculated using irradiance variation, earth
_
LV cumte XorXot X
X
=albedo and IR emissions. Earth albedo and IR emissions have
no significant effect on GEO, so they are only considered for
p onsiant
Volag
X~ ~~~Cl 11|(V)
1~~~~~
Soa
celIVcaatrsi.uveadmxmmotu
Fig.
oe
768
solar flare protons and outer zone electrons for GEO are the
most common particles that affect cell performance.
G
The radiation input ((O) of the cell model, is the "equivalent
D
Ta
1P
1MeV electron fluence" [4,5,6]. The equivalent fluence is
determined by applying Relative Damage Coefficients (RDCs)
of the specified cells to the omni directional proton and
electron fluxes in the specified orbit, which are output data of
AE8 and AP8 models [4]. The radiation data is also verified
Fig. 2. New solar cell electrical model with radiation input
with the certified European Space Agency's web-based
software, "Space Environment Information System D.
and Temperature
D. Equationsfor Irradiance and
Irradiance has utiize
a cyclic pattern, depending on the orbit [2,8].
(SPENVIS)",
(SEVS1 which utilizes AE8, AP8 and EQFLUX models
[4,12].
The values of eclipse time and orbital period are given by (1),
(2).
B. Solar Cell Modeling
T T -(I/180o) -cos1 [(I-(RE )2)1/2 /COSfi
Radiation, as a new input ( O), is added to the general "one
(1)
E~~~~~~
a
diode model" of solar cells, to form a new cell model as shown
in Fig. 2. To avoid additional complexity of mathematical
T=2f afu
S
modeling, the simplest solar cell model is used [9,10]. Since
(2)
the results are used for a comparative evaluation of cells,
Where RE=6372 km, a=RE+Altitude, and ,u =3.986*10'
possible inaccuracy of the model is not a very critical subject.
Complete models with more accurate results will be presented kM3/S2. / is the angle between sun vector and orbital plane.
in future works.
Equation (1) gives eclipse time, only if ,6 < 6 , where
AE8,lc andmperatureel
whichon
Table 11 shows the specifications of selected si solar cells [24]. Selected cells include: n-type-base silicon cell, polycrystalline silicon cell, and High Efficiency Silicon (HES) solar
cell.
With cover glass thickness of approximately 100 ,um, only
protons with the energies higher than 4MeV can penetrate into
the cell base region [4].
There are essential differences between selected terristerial
and space-qualified solar cells.
Differences include: p-type vs n-type base , textured front
surface, thin cells, and Back Surface Reflectors (BSRs) in
space cells, especially H-ES cell. Space cells also have an
additional layer over the cover glass that deceeases glass
darkening and increases absorptance ofthe cells [3].
PARTICLEs ENERGY
Electrons Electrons Trapped
Protons
(LEO)
(GEO)
(LEO) PaGEicl) Eey (LEO)
0 - 100
-7 0 Ener1 00.
Pt(MeV)
Type of Particle
Solar
Flare
Protons
(GEO)
1 -200
TABLE II
SPECIFICATIONS OF SELECTED SILICON SOLAR CELL TYPES
silicon solar cell types (CG 100 urn)
Efficiency 7 (%)
(1,/7) dl
(/O
C)
Absorptance
Thickness( ,Umn)
(3)
E. Cell Equations
TABLE I
Parameter
n-type-base
poly-Si
13.2
14.8
HES
(2 Q - cm)
16.6
-0.55
-0. 46
-0. 35
0.75
150
0.8
100
0.85
50, 100
|Parameter
DEnsity
769
30 5(0%)
Earth JR
(W/m2
(4
I
'SC
:-
'ph
VD VOCnVTln(1+ISC/ISAT)
.
'SC
'SAT
(5)r
loo
FF(
' FFO)
Eg (T)
KT
ph
x10
(13)
/nVT)1)
(14)
(4
BOL
(15)
*(T-T,)]*G/G,
(16)
Vm (G, T)
sVc
-
ds
'ph
Vo(G, T)
currents.
(9)
1 1
nVT
1)]
1
(17)
RS I,, (G, T)
(10)
(18)
0.0004.73xT2
Time (Seconds)
(8)
Rs (G, T)= Rs
LEO
-200
0o
n/-
Fig. 3. Irradiance and Cell Temperature Profiles for LEO and GEO
(VmJm)A(VOCJIS() RS=0-
JPH(G,T)I[JsCr
Timc (Scconds)
n/- /n-f=
---
f-
(6)
R = Voc I(JAsc
14:.
5100
6EO
Gn lOf i --n- n
()1500
a(T T) + 3(G
-
Gr )
(11)
(12)
can now
(19)
V0c (f' C,G Tr) is calculated from (15) Using (20) and (21).
770
~~~~~.....................................................
~~~~~~~(~ ~ ~ ~ ~ ~ ~ ~ ~ o~ ~ ~ ~ ~ C~ ,T)canbecalculated
(2 )from
.~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~. . . . . . . . . . . . . . . . . . . . . . . . . . .
~~~~~~~~~~~
dJ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~*(T~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~Tr)]*C/Cr
~ ~dT~ FOL
(2 )~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~4~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~...................................................
~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~. . . . . . . . . . . . . . . . . . . . . . . . . . .
Similarrelationcanbeusedfor calculation
~of~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ m(...............o,C,T). . . . . . . . . . . . . . . .
AndV0~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ (~ ~ ~ ~ ~ ~ ~ ~ o(~2~3)~.~C,~ ~T~c)~anbe~ca~lc~ul~fromat~ed~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ . . . . . . . . . . .
CelL
odel ain.........................................
dV
nV
Fig.
MATLAB/Simulink.~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
inmodel
4. Cell
ofpresented
inpredictimodel
ng~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ . . . . . . . . . . . . . . . . . . .
solar...cell..performance,....and
G1nscJG
G(G,p2C,T)..the..effects...of.space...environment....on.
second; to com pare various teristerial and space solar cel s in~ ~ ~ ~ ~ ~ ~ ~ ~ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
space ap lications.
tLhEeO spaacned apEOlicationo,rbits,
5-year-period~ ~ ~ ~ ~ ~ ~ ~ ~ ~. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Then~~ ~ ~ ~ proper
~ ~ ~ ~ ~ ~Syp,Aform
TCT,i)caslfromculated of (6). in-orbit eration at Cells
p~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~is~ ~ ~c~on~si~de~re~d.~~ ~~ ~~ ~~ ~~ ~~ ~~~~~~~~~~~~.~.~.~.~.~. ~. ~. ~. ~. ~. ~. ~. .~ .~ .~ .~ .~ .~ .~ .~.~.~.~.~.~.~.~.~.~. ~ ~ ~ ~ ~ ~ ~. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
AndfinallyVm(~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~C,o, T)and~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ (to,
maRemaining
x C,Ta)recalculated Factor(PRF)ofthecel s.~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
from(24)and(25).~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ . . . . . . . . . . . . . . . . . . . . . . . . . .
Th~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~. . . . . . . . . . . . . . . . . . . . . . . . . .
Vm(~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ fl,G,=T)
EZD
G~~~~~~~~~~~~~~~~~~~~~~~................................................
Th(C)
(Fi,G,T)
J~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~..(ID..........................................
V0~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~(q9,G,T) (24)~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~.....................................................
-1))
D
C Fi(~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~.............................,...2).......HHH
R~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ JmQ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~~~~
~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~~~
~ ~ ~ ~ ~ ~ fl,G~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~.............................................
,T)~ ~ ~
Pmax(G,T>~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ Vm(G,T) Jm(G,T) (25)~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ . . . . . . . . . . . . . . . . . . . . . . . .
AloadresistanceRisconnectedtotheoutputportofFig.2.~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ . . . . . . . . . . . . . . . . . . . . . . . . .
fordrawingI-Vcharacteristiccurveofthecell. Socellcur ent~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~. . . . . . . . . . . . . . . . . . . . . . . . . . . .
andvoltage(I,V)arerelatedtothecellmodelparametersas~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~. . . . . . . . . . . . . . . . . . . . . . . . . . . .
statedin(26).WhereV~~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~. . . . . . . . . . . .RI. . . . . . . . . . . . . .
J(p,G,T) ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ J5~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~@p,G,T) -~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~. . . . . . . . . . . . . . . . . . . . . . . . . .
'SAT (~~~~~~~~~o, T)(((
P~~~~~~~~~~~~~~~ C, T)
R5 J(tp, C, T)
(26)
FV(eI~ ~ FiP~~~~~~~~~~~~~~~~~~....................................................
~ ~(1). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
FiP(ek~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~n2)
ThecellmodelisimplementedinSimulinkashownFign.~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~. . . . . . . . . . . . . . . . . . . . . . . . . . . .
includes
4.Themodel
some in~ ~ p~ar~am~et~e~rdse~pe~nd~specifications.
nd. . . . . . . . . . . .internal
inJogSTnceerlnasl y. . . .defined. . . .constants. . . . aMain
VP~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ In,..........................i,G,T)G. . . . .D. . . . . . .
proces eas reshowniTheFnig.5.
model
is capable
of~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ . . . . . . . . . . . . . . . . . . . . . . . . . . . .
applyingvariousvaluesotemperature,
fir adiance,
and~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~. . . . . . . . . . . . . . . . . . . . . . . . . . . .
radiation.Simulationfollowstwomaingoals.~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~. . . . . . . . . . . . . . . . . . . . . . . . . . .
RI~~~~~~~~~~~~~~~~~~~~~~~...................................................
7 1~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ . . . . . . . . . . . . . . . . . . . . . . . . .
1
0.9
BSR2ohm-cm2mil
er.8uR
PolySi
0.7
N
n-type base Si
0.6
0.5
0.4
10
12
10
10
13
10
14
10
10
IV.REEsULTS
E 0.018-
C.)l_
' 0.016
S 0.014 .......
E
0.012-
0.01-
v 0.008
l_
l_
l_
l_
l11
0
10 12
1013
l104
Equivalent IMevElectronFluence (e/cm2)
0.006
l0 5
0.02
|
0.015
TABLE IV
SIMULATION DATA SUMMAvRY FOR LEO
LEO
Solar Cell
Type
BOL
BOLat
EOL
(mW/cm2)
(mW/cm2)
2Q-cm BSR2mil
10.5
6.5
2Q-cm BSR4mil
9.2
5.3
Pmax
PR_BSR
mil
0.62
at
LEO/
0.58
0.005
1010
1011
1012
1013
1014
1015
Fig. 7. Comparison of Radiation Effects in AMO, GEO (up), and LEO (down).
GEO
L_______
Pmax at
BOLT
(mW/cm2)
13.5
12.5
Pmax at
Pmax
EOL,
(mW/cm2) |____
9.2
0.68
8.1
0.65
772
b8
s-
\
>
100
<
2
~
~~
50
-1e12
10
ss T 6 -\ | /1
+
-IeI4
+- | ---
-----lel5
<---jl
-lel6
-:.
~ ~ ~ ~~~~~~0
<
0
-3
Temperature(C)
I
J
I
\REFERENCES
[1] J.R. Wertz and W.J. Larson, Space Mission Analysis and Design, 3rd ed.,
--leI3 KLUWER Academic Publications, 2005, pp.413-417.
[2] M.R. Patel, Spacecraft Power Systems, CRC Press, 2005, pp.135-180.
[3] A. Luque, S. Hegedus, Handbook of Photo voltaic Science and
Engineering, John Wiley, 2003, pp.413-446.
V.L. Pisacane, Fundamental of Space Systems, 2nd edition, Oxford
[4]< 1>500
1000
Irradiance(W/m2)
(b) LEO
V.CONCLUSION
In the paper we presented a new model for space solar cells.
[12] L.B. Rainey, Space Modeling and Simulation Roles and Applications
Throughout the System Life Cycle, AIAA,2004, pp.534-541.
[13] 17- N.S. Fatemi, H.E. Pollard, H.Q. Hou, and P.R. Sharps, "Solar Array
Trades Between Very High-Efficiency Multi-Junction and Si Space Solar
Cells", presented at the 28th IEEE PVSC, Anchorage, Alaska, September
The model
model uses
uses cummulative
cummulative radiation
radiation fluence
The
fluence as
as aa new*input
new input
2001
(with irradiance and temperature). The presented solar cell [14] T.V. Torchynska and G.Polupan, "High efficiency solar cells for space
model is capable of predicting the behavior of various solar
applications", Superficies y Vacio 17(3), pp.21-25, September 2004.
cells in space applications. The model requires a complete test
data of the manufacturer (data sheet ofthe cell).
773
706
AbstractThis paper identifies absorbers for multiple transition solar cells that are implemented with nanostructured heterojunctions [e.g., quantum well solar cells with quasi-Fermi-level
variations and quantum dot (QD) intermediate-band solar cells].
In the radiative limit, the solar cells implemented with these absorbers are capable of achieving a conversion efficiency 50%
with a geometric solar concentration of at least 1000. The technical approach enumerates a set of quantitative design rules and
applies the rules to the technologically important IIIV semiconductors and their ternary alloys. A novel design rule mandates a negligible valence band discontinuity between the barrier material and confined materials. Another key design rule
stipulates that the substrate have a lattice constant in between
that of the barrier material and that of the quantum-confined
material, which permits strain compensation. Strain compensation, in turn, allows a large number of QD layers to be incorporated into the solar cell because each layer is free of defects.
Four candidate materials systems (confined/barrier/substrate)
are identified: InP0 .8 5 Sb0 .1 5 /GaAs/InP, InAs0 .4 0 P0 .6 0 /GaAs/InP,
InAs/GaAs0 .8 8 Sb0 .1 2 /InP, and InP/GaAs0 .7 0 P0 .3 0 /GaAs. Resulting from the design features, the candidate systems may also find
use in other optoelectronic applications.
Index TermsHeterojunction, intermediate band, quantum dot,
quantum well, solar cell.
I. INTRODUCTION
HYSICAL phenomena that may increase the efficiency of
a single-junction solar cell above the ShockleyQueisser
limit [1] include proposals such as multiple transitions [2], multiple electronhole pair generation [3], and hot carriers [4].
Identifying materials that may be used to implement these
phenomena is of significant interest. For example, multiple
electronhole pair generation is demonstrated in several materials [5], [6]. Several approaches are proposed to implement
multiple transition solar cells, all of which require the introduction of valid quantum states deep into the previously forbidden
bandgap of a host semiconductor [2]. Two of these, which rely
on nanostructured heterojunctions, are the quantum well solar
Manuscript received May 22, 2007; revised September 25, 2007. This
work was supported by the U.S. National Renewable Energy Laboratory under Subcontract XAT-5-44277-01, administered by M. Symko-Davies and R.
McConnell. The work of M. Y. Levy was supported by the U.S. National Science
Foundation under a Ph.D. Dissertation Enhancement Award OISE-0509001,
administered by J. Hudson. The review of this paper was arranged by Editor
P. Panayotatos.
M. Y. Levy is with the University of Delaware, Newark, DE 19716 USA.
He is also with the Georgia Institute of Technology, Atlanta, GA 30332 USA
(e-mail: mlevy@ece.gatech.edu; qb@UDel.edu; mail2MYL@gmail.com).
C. Honsberg is with the University of Delaware, Newark, DE 19716 USA
(e-mail: honsberg@ece.UDel.edu).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TED.2007.914829
LEVY AND HONSBERG: NANOSTRUCTURED ABSORBERS FOR MULTIPLE TRANSITION SOLAR CELLS
intermediate states and the conduction and valence band continua, respectively.
Previous work on the QD IBSC revolves around the
InAs/GaAs (confined/barrier) [11], [19][21] and InGaAs/GaAs
[22] material systems, both grown on GaAs substrates. These
choices derive from the systems being relatively well understood. In addition, the bandgap of a GaAs barrier is near to
the optimum conduction-to-valence bandgap associated with
the upper limiting efficiency of an IBSC [23]. These materials
systems have two characteristics limiting their usage for development of an QD IBSC: they have valence band discontinuities
and they accumulate strain. The first characteristic (type I heterojunction) enables the luminescent spectra of QD IBSC samples [11], [19] to illustrate narrowly separated and overlapping
peaks, rather than the luminescent signature of an intermediate
band material. The most rudimentary signature is the appearance of three [24] (or four [25]) strong luminescence peaks
separated across a wide energy range; each of the three peaks
corresponds to one of the three allowed interband electronic
transitions (whereas the fourth peak corresponds to an intraband
transition [25]). The second characteristic (strain accumulation)
causes the open-circuit voltage to degrade as the number of QD
multilayers increases [22]. This is a problem to the design of a
high-efficiency QD IBSC, for a large number of multilayers is
essential for a high absorptivity of subbandgap photons [26]
the high absorptivity of subbandgap photons may be viewed
experimentally by an increase in their external quantum efficiencies with an increase in the number of QD layers [22]. This
papers design rules identify candidate material systems that
may overcome these two characteristics.
The candidate material systems identified here yield heterojunctions at the transition between type I and type II heterojunc-
707
708
valence band edge of the barrier material (see Fig. 1). The absence of a valence band discontinuity eliminates the possibility of introducing intermediate states resulting from quantumconfined holes. There are two reasons why the valence band
discontinuity is set to a negligible value rather than the conduction band discontinuity. First, in the technologically realizable
interval of QD radii, for even a small valence band discontinuity,
many hole states appear; however, fewer electron states appear
with a similar magnitude conduction band discontinuity. This
result stems from the fact that the technologically important
IIIV semiconductors have larger hole effective masses than
that of electron effective masses [45]. Thus, within technologically realizable QD radii, it is possible to provide a single electron state within the confining potential, whereas, at those same
radii, many hole states would exist. Second, in general, for the
technologically important IIIV semiconductors, the hole mobilities are less than the electron mobilities. In an actual device,
the higher the mobility of the carriers, the higher is the diffusion
length for a given carrier lifetime. Thus, the mobility, through the
diffusion length, enhances the collection of generated electron
hole pairs, and hence, the performance of the device.
B. Substrates Lattice Constant Between Those of the Confined
Material and the Barrier Material
The second design rule specifies inequalities relating the lattice constants of the confined, barrier, and substrate materials.
Specifically, the lattice constant of the substrate must be between the lattice constants of the barrier and the confined materials. Material systems complying with this rule may allow the
incorporation of tensile strain to compensate for compressive
strain accumulated during the growth of the multilayer quantum
well [46], [47] and QD structures [34]. Strain compensation reduces the density of defects in multilayer QD structures [48].
Thus, strain compensation is relevant to the number of QD
layers, which has a direct impact upon the absorptivity of subbandgap photons [26], that can be incorporated into multilayer
structures before relaxation occurs. Further, the reduced density of defects in multilayer QD structures [48] abates large
open-circuit voltage losses in experimental samples [19], [22].
C. Barriers Bandgap Between 1.25 and 3.13 eV
The third design rule specifies the required interval of the
conduction-to-valence bandgap CV . In a previous examination
of the QD IBSC [8], a bandgap of 1.95 eV is considered to be
necessary for candidate barrier materials. This consideration is
derived from detailed balance calculations [23] where the solar
irradiance is concentrated at 46 000, which is the physical
limit of geometrical concentration (i.e., the reciprocal of the
geometrical dilution factor [1]). The bandgap of 1.95 eV is associated with the apex of the efficiency curve found there. Here,
the authors consider a more practical geometrical concentration
factor of 1000. In addition, unlike previous calculations [23],
when obtaining the maximum-power-detailed balance efficiencies, the authors include the following as parameters in the
search space: the energetic width of intermediate states [49] and
the spectral extent of the absorption overlap [50]. Similarly to
Fig. 2. Maximum power isoefficiency contour plot of an IBSC under a geometric solar concentration of 1000 (shown here are the 50% and 55% contours)
as a function of the energetic gaps IV and CI (see Fig. 1). Contours obtained
using the rapid and precise computation method of [51].
LEVY AND HONSBERG: NANOSTRUCTURED ABSORBERS FOR MULTIPLE TRANSITION SOLAR CELLS
709
TABLE I
MATERIAL SYSTEMS WITH WHICH TO IMPLEMENT NANOSTRUCTURED
ABSORBERS FOR MULTIPLE TRANSITION SOLAR CELLS
710
REFERENCES
[1] W. Shockley and H. J. Queisser, Efficiency of p-n junction solar cells,
J. Appl. Phys., vol. 32, pp. 510519, 1961.
[2] M. Wolf, Limitations and possibilities for improvement of photovoltaic
solar energy converters. Part I: Considerations for Earths surface operation, in Proc. Inst. Radio Eng., vol. 48, no. 7, pp. 12461263,
1960.
[3] P. Landsberg, H. Nussbaumer, and G. Willeke, Bandband impact ionization and solar cell efficiency, J. Appl. Phys., vol. 74, no. 2, pp. 14511452,
1993.
[4] R. T. Ross and A. J. Nozik, Efficiency of hot-carrier solar energy converters, J. Appl. Phys., vol. 53, no. 5, pp. 38133818, 1982.
[5] R. Schaller and V. Klimov, High efficiency carrier multiplication in PbSe
nanocrystals: Implications for solar energy conversion, Phys. Rev. Lett.,
vol. 92, no. 18, pp. 186601-1186601-4, 2004.
[6] M. C. Beard, K. P. Knutsen, P. Yu, J. M. Luther, Q. Song, W. K. Metzger,
R. J. Ellingson, and A. J. Nozik, Multiple exciton generation in colloidal
silicon nanocrystals, Nano Lett., vol. 7, no. 8, pp. 25062512, 2007.
[7] S. P. Bremner, R. Corkish, and C. B. Honsberg, Detailed balance efficiency limits with quasi-Fermi level variations, IEEE Trans. Electron
Devices, vol. 46, no. 10, pp. 19321939, Oct. 1999.
[8] A. Mart, L. Cuadra, and A. Luque, Quantum dot intermediate band solar
cell, in Conf. Rec. 28th IEEE Photovoltaic Spec. Conf., Anchorage, AK,
2000, pp. 940943.
[9] A. Luque, A. Mart, and L. Cuadra, Thermodynamic consistency of subbandgap absorbing solar cell proposals, IEEE Trans. Electron Devices,
vol. 48, no. 9, pp. 21182124, Sep. 2001.
[10] J. Nelson, J. Barnes, N. Ekins-Daukes, B. Kluftinger, E. Tsui, K. Barnham,
C. T. Foxon, T. Cheng, and J. S. Roberts, Observation of suppressed
radiative recombination in single quantum well p-i-n photodiodes, J.
Appl. Phys., vol. 82, no. 12, pp. 62406246, 1997.
[11] A. Luque, A. Mart, N. Lopez, E. Antolin, E. Canovas, C. Stanley,
C. Farmer, L. J. Caballero, L. Cuadra, and J. L. Balenzategui, Experimental analysis of the quasi-Fermi level split in quantum dot intermediate-band
solar cells, Appl. Phys. Lett., vol. 87, no. 8, pp. 083505-1083505-3,
2005.
[12] A. Mart, L. Cuadra, and A. Luque, Partial filling of a quantum dot
intermediate band for solar cells, IEEE Trans. Electron Devices, vol. 48,
no. 10, pp. 23942399, Oct. 2001.
[13] A. Mart, L. Cuadra, and A. Luque, Intermediate band solar cells, in
Next Generations Photovoltaics: High Efficiency Through Full Spectrum
Utilization, A. Mart and A. Luque, Eds. Bristol, U.K.: Inst. of Physics
Publishing, 2004, ch. 7, pp. 150160.
[14] E. F. Schubert, Cambridge studies in semiconductor physics and microelectronic engineering1, in Doping in IIIV Semiconductors,
H. Ahmed, M. Pepper, and A. Broers, Eds. Cambridge, U.K.: Cambridge
Univ. Press, 1993, ch. 1, pp. 4144.
[15] O. Madelung, Introduction to Solid-State Theory. Heidelberg, Germany:
Springer-Verlag, 1978, pp. 287, 382, 435447.
[16] O. L. Lazarenkova and A. A. Balandin, Miniband formation in a quantum
dot crystal, J. Appl. Phys., vol. 89, pp. 55095515, 2001.
[17] A. Mart, L. Cuadra, and A. Luque, Design constraints of the quantumdot intermediate band solar cell, Physica E, vol. 14, no. 1/2, pp. 150157,
2002.
[18] M. Y. Levy, A. Mart, C. Honsberg, and A. Luque, Calculation of the band
properties of quantum dot intermediate band solar cells with hydrogenic
impurities, in Proc. 19th Eur. Photovoltaic Energy Conf., Paris, France,
2004, pp. 332335.
[19] A. Luque, A. Mart, C. Stanley, N. Lopez, L. Cuadra, D. Zhou, J. L.
Pearson, and A. McKee, General equivalent circuit for intermediate band
devices: Potentials, currents and electroluminescence, J. Appl. Phys.,
vol. 96, no. 1, pp. 903909, 2004.
[20] A. Luque, A. Mart, N. Lopez, E. Antolin, E. Canovas, C. Stanley,
C. Farmer, and P. Diaz, Operation of the intermediate band solar cell
under nonideal space charge region conditions and half filling of the intermediate band, J. Appl. Phys., vol. 99, no. 9, pp. 94503-194503-9,
2006.
[21] A. Mart, E. Antolin, C. Stanley, C. Farmer, N. Lopez, P. P. Daz,
E. Canovas, P. Linares, and A. Luque, Production of photocurrent due
to intermediate-to-conduction-band transitions: A demonstration of a key
operating principle of the intermediate-band solar cell, Phys. Rev. Lett.,
vol. 97, no. 24, pp. 247701-1247701-4, 2006.
[22] A. G. Norman, M. C. Hanna, P. Dippo, D. H. Levi, R. C. Reedy, J. S. Ward,
and M. M. Al-Jassim, InGaAs/GaAs QD superlattices: MOVPE growth,
structural and optical characterization, and application in intermediate-
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
[35]
[36]
[37]
[38]
[39]
[40]
[41]
[42]
[43]
band solar cells, in Conf. Rec. 31st IEEE Photovoltaic Spec. Conf., Orlando, FL, 2005, pp. 4348.
A. Luque and A. Mart, Increasing the efficiency of ideal solar cells
by photon induced transitions at intermediate levels, Phys. Rev. Lett.,
vol. 78, pp. 50145017, 1997.
N. Ekins-Daukes, C. Honsberg, and M. Yamaguchi, Signature of intermediate band materials from luminescence measurements, in Conf. Rec.
31st IEEE Photovoltaic Spec. Conf. (IEEE Cat. No. 05CH37608), Lake
Buena Vista, FL, 2005, pp. 4954.
M. Y. Levy, N. Ekins-Daukes, and C. Honsberg, Photouminescent signature of absorbing medium with deep levels based upon absorption coefficicents derived from parabolic band approximation, presented at the
22nd Eur. Photovoltaic Energy Conf., Milan, Italy, 2007.
B. Kochman, A. D. Stiff-Roberts, S. Chakrabarti, J. D. Phillips, S. Krishna,
J. Singh, and P. Bhattacharya, Absorption, carrier lifetime, and gain
in InAs-GaAs quantum-dot infrared photodetectors, IEEE J. Quantum
Electron., vol. 39, no. 3, pp. 459467, Mar. 2003.
A. Sibille, Miniband transport, in Semiconductor Superlattices: Growth
and Electronic Properties, H. T. Grahn, Ed. Singapore: World Scientific,
1995, ch. 2, p. 37.
M. Levy, C. Honsberg, A. Mart, and A. Luque, Quantum dot intermediate band solar cell material systems with negligible valence band
offsets, in Conf. Rec. 31st IEEE Photovoltaic Spec. Conf. (IEEE Cat. No.
05CH37608), Lake Buena Vista, FL, 2005, pp. 9093.
Y. Sugiyama, Y. Nakata, K. Imamura, S. Muto, and N. Yokoyama,
Stacked InAs self-assembled quantum dots on (001) GaAs grown by
molecular beam epitaxy, Jpn. J. Appl. Phys. 1, Reg. Papers, Short Notes,
vol. 35, no. 2B, pp. 13201324, 1996.
K. Shiramine, Y. Horisaki, D. Suzuki, S. Itoh, Y. Ebiko, S. Muto, Y. Nakata,
and N. Yokoyama, TEM observation of threading dislocations in InAs
self-assembled quantum dot structure, J. Crystal Growth, vol. 205, no. 4,
pp. 461466, 1999.
C. H. Roh, Y. J. Park, K. M. Kim, Y. M. Park, F. K. Kim, and K. B. Shim,
Defect generation in multi-stacked InAs quantum dot/GaAs structures,
J. Crystal Growth, vol. 226, no. 1, pp. 17, 2001.
K. Stewart, M. Buda, J. Wong Leung, L. Fu, C. Jagadish, A. Stiff-Roberts,
and P. Bhattacharya, Influence of rapid thermal annealing on a 30 stack
InAs/GaAs quantum dot infrared photodetector, J. Appl. Phys., vol. 94,
no. 8, pp. 52835289, 2003.
M. Gutierrez, M. Hopkinson, H. Y. Liu, J. S. Ng, M. Herrera, D. Gonzalez,
R. Garcia, and R. Beanland, Strain interactions and defect formation
in stacked InGaAs quantum dot and dot-in-well structures, Phys.
E: Low-Dimensional Syst. Nanostruct., vol. 26, no. 14, pp. 245251,
2005.
K. Akahane, N. Ohtani, Y. Okada, and M. Kawabe, Fabrication of ultrahigh density InAs-stacked quantum dots by strain-controlled growth on
InP(311)B substrate, J. Crystal Growth, vol. 245, pp. 3136, 2002.
P. Lever, H. H. Tan, and C. Jagadish, InGaAs quantum dots grown with
GaP strain compensation layers, J. Appl. Phys., vol. 95, no. 10, pp. 5710
5714, 2004.
Z. H. Zhang, C. F. Xu, K. C. Hsieh, and K. Y. Cheng, Defect-free 50-layer
strain-balanced InAs quantum dots grown on InAlGaAs/InP for infrared
photodetector applications, J. Crystal Growth, vol. 278, no. 14, pp. 61
66, 2005.
K. Akahane, N. Yamamoto, and N. Ohtani, Long-wavelength light emission from InAs quantum dots covered by GaAsSb grown on GaAs substrates, Physica E, vol. 21, pp. 295299, 2004.
Z. H. Zhang, K. Y. Cheng, C. F. Xu, and K. C. Hsieh, Defect-free 100layer strain-balanced InAs quantum dot structure grown on InP substrate,
Appl. Phys. Lett., vol. 89, no. 6, pp. 63115-163115-3, 2006.
A. S. Brown and M. A. Green, Intermediate band solar cell with many
bands: Ideal performance, J. Appl. Phys., vol. 94, pp. 61506158,
2003.
T. Trupke, M. A. Green, and P. Wurfel, Improving solar cell efficiencies
by up-conversion of sub-band-gap light, J. Appl. Phys., vol. 92, pp. 4117
4122, 2002.
T. Trupke, M. A. Green, and P. Wurfel, Improving solar cell efficiencies of high-energy photons, J. Appl. Phys., vol. 92, pp. 16681674,
2002.
U. Laheld, F. Pedersen, and P. Hemmer, Excitons in type-II quantum dots:
Finite offsets, Phys. Rev. B, Condens. Matter, vol. 52, no. 4, pp. 2697
2703, 1995.
C.-K. Sun, A. Wang, J. Bowers, B. Brar, H.-R. Blank, H. Kroemer,
and M. Pilkuhn, Optical investigations of the dynamic behavior of
GaSb/GaAs quantum dots, Appl. Phys. Lett., vol. 68, no. 11, pp. 1543
1545, 1996.
LEVY AND HONSBERG: NANOSTRUCTURED ABSORBERS FOR MULTIPLE TRANSITION SOLAR CELLS
[44] H. Munekata, T. Smith III, F. Fang, L. Esaki, and L. Chang, Electrons and
holes in InAs-Ga(Al)Sb(As) quantum wells, J. Phys. Colloq., vol. 48,
no. C-5, pp. 151154, 1987.
[45] I. Vurgaftman, J. R. Meyer, and L. R. Ram-Mohan, Band parameters for
IIIV compound semiconductors and their alloys, J. Appl. Phys., vol. 89,
no. 11, pp. 58155875, 2001.
[46] N. Ekins-Daukes, K. Barnham, J. Connolly, J. Roberts, J. Clark, G. Hill,
and M. Mazzer, Strain-balanced GaAsP/InGaAs quantum well solar
cells, Appl. Phys. Lett., vol. 75, no. 26, pp. 41954197, 1999.
[47] N. Ekins-Daukes, K. Kawaguchi, and J. Zhang, Strain-balanced criteria
for multiple quantum well structures and its signature in X-ray rocking
curves, Crystal Growth Des., vol. 2, no. 4, pp. 287292, 2002.
[48] N. Nuntawong, S. Birudavolu, C. Hains, S. Huang, H. Xu, and D. Huffaker,
Effect of strain-compensation in stacked 1.3 m InAs/GaAs quantum dot
active regions grown by metalorganic chemical vapor deposition, Appl.
Phys. Lett., vol. 85, no. 15, pp. 30503052, 2004.
[49] A. S. Brown, M. A. Green, and R. P. Corkish, Limiting efficiency for a
multi-band solar cell containing three and four bands, Physica E, vol. 14,
pp. 121125, 2002.
[50] L. Cuadra, A. Mart, and A. Luque, Influence of the overlap between
the absorption coefficients on the efficiency of the intermediate band solar
cell, IEEE Trans. Electron Devices, vol. 51, no. 6, pp. 10021007, Jun.
2004.
[51] M. Y. Levy and C. Honsberg, Rapid and precise calculations of energy
and particle flux for detailed-balance photovoltaic applications, SolidState Electron., vol. 50, pp. 14001405, 2006.
[52] E. Ribeiro, R. Maltez, W. Carvalho, D. Ugarte, and G. Medeiros-Ribeiro,
Optical and structural properties of InAsP ternary self-assembled quantum dots embedded in GaAs, Appl. Phys. Lett., vol. 81, no. 16, pp. 2953
2955, 2002.
[53] R. L. Maltez, E. Ribeiro, W. Carvalho, Jr., D. Ugarte, and G. MedeirosRibeiro, Controlling alloy composition of InAsP self-assembled quantum
dots embedded in GaAs, J. Appl. Phys., vol. 94, no. 5, pp. 30513056,
2003.
[54] H. Y. Liu, M. J. Steer, T. J. Badcock, D. J. Mowbray, M. S. Skolnick,
P. Navaretti, K. M. Groom, M. Hopkinson, and R. A. Hogg, Longwavelength light emission and lasing from InAs/GaAs quantum dots covered by a GaAsSb strain-reducing layer, Appl. Phys. Lett., vol. 86,
pp. 143108-1143108-3, 2005.
[55] H. Y. Liu, M. J. Steer, T. J. Badcock, D. J. Mowbray, M. S. Skolnick,
F. Suarez, J. Ng, M. Hopkinson, and J. P. R. David, Room-temperature
1.6 m light emission from InAs/GaAs quantum dots with a thin GaAsSb
cap layer, J. Appl. Phys., vol. 99, pp. 46104-146104-3, 2006.
[56] M. Copel, M. C. Reuter, E. Kaxiras, and R. M. Tromp, Surfactants in
epitaxial growth, Phys. Rev. Lett., vol. 63, no. 6, pp. 632635, 1989.
711
Christiana Honsberg (S87M91SM03) received the Ph.D. degree from the University of
Delaware, Newark, in 1992.
From 1993 to 2000, she was a faculty member at
the University of New South Wales, Sydney, Australia, where she worked on high-efficiency commercial silicon solar cells, advanced concept photovoltaics, and developing the worlds first undergraduate degree program in photovoltaic engineering. In
2001, she joined the School of Electrical and Computer Engineering, Georgia Institute of Technology,
where she was involved in ultrahigh-efficiency solar cells capable of reaching the thermodynamic limits of conversion of solar radiation. In 2004, she
joined the University of Delaware, where she was engaged in developing the
university program on high-performance photovoltaic devices, encompassing
high-efficiency silicon, new materials for tandem solar cells, and new concepts in photovoltaic energy conversion. Dr. Honsberg was awarded a National
Science Foundation Integrative Graduate Education and Research Traineeship
(NSF-IGERT) in 2006 to develop a graduate education and research program in
solar hydrogen.
Jinyu Wen
AbstractTraditional
illumination
lights
are
energyconsumption and not economical in the long term. While the
solar lights are safe, easy to maintenance, independent of
conventional energy, no pollution to environment. With the cost
of the solar cells decreasing, the solar lights have been applied
widely. Meanwhile photovoltaics are one of the fastest growing
solar energy technologies. Photovoltaic devices, commonly called
solar cells or modules, use semiconductor material to directly
convert sunlight into electricity. In this paper, firstly the
development of solar light in Wuhan city is introduced. A
comprehensive analysis on the principle and the structure of the
solar light is conducted, and the mathematical model of solar
cells is proposed. Through tests on the solar cells, the I-V and PV characteristic curves are obtained which verify the theoretical
curves.
Keywords- solar lights; Photovoltaic cell; I-V curve; energy
I.
INTRODUCTION
A. Design principle
The general design principle of the solar light system is as
follows [4]:
B. System structure
The solar light illumination system, in essence, is a system
of converting light from sun to electricity, which is called as
silicon photovoltaic module. Silicon panel receives the sunlight
(3)
U +I R
q
I RL = I s I so exp
(U RL + I RL Rs ) 1 RL RL s (4)
Rsh
AkT
C =
Cdis
100%
Cch
(1)
Q =
Qdis
100%
Qch
(2)
Parameters
Electricity
Performance
Wp
Peak Power Output (W)
Vmp
Maximum Power Voltage (V)
Imp
Maximum Power Current (A)
Isc
Short Circuit CurrentA
Voc
Open Circuit Voltage(V)
Insulation Resistance( M )
Maximum System Voltage
MBF125
125
17.5
7.15
7.87
21.5
50
800 V
DC
G
(10)
1 + Vmpp (T Ts )
Gs
Where parameters I scs , Vocs , I mpps and Vmpps are defined at
Vmpp ( G , T ) = Vmpps
dI I
+ =0
(11)
dV V
As for the model in Fig.1, the value of dI / dV at the peak
power point can be calculated. By comparing the dI / dV with
I mpps / Vmpps , the values of A and Rs in the model which are
critical can be obtained by computer software. The desirable
estimation of A and Rs is achieved when the absolute value of
dI
dV
V =Vmpp
I mpp is minimized.
Vmpp
IV.
CONCLUSION
(5)
I I
V = C2 Voc log 1 + sc
(6)
I sc C1
C2 =
Vmpp / Voc 1
ACKNOWLEDGMENT
log (1 I mpp / I sc )
I mpp ( G , T ) = I mpps
G
1 + I mpp (T Ts )
Gs
(8)
(9)
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
:r""
.;l~
.... : I'
.L~,~~~------Y
~ 60
w
Q
40
20
o
200
......
400
......--..
600
1000
800
Wavelength
....
1200
1400
(nm)
'(]4
0.2
'."'"
0.2
0.4
Stt'p
Diffusion
Oxidation
LPCVD
Diffusion
Oxidation
Diffusion
FGA 1
FGA2
Dt'tail
Emitter formation
Grow thin passivation oxide
Deposit AR coating
n++ formation
To fonn diffilsion barrier
p+ fonnation
Forming gas annealing to promote
hydrogenation
Forming gas annealing to promote
hydrogenation
..
0.6
08
.
Voc(mV)
Process
step
:c- 200.0
150.0
.. 100.0
.s:
50.0
0.0
1-+- 1.006-151
L__~__~__~__~__~::~~~~25.6
Fig. 6 Completed Si solar cells before dicing
Process
Step
DE'taii
iUIl
... tlIoQ~1-
11111
1.lID
Ul!
D..UI
n.sm o.~
11m
Si Etch
RCA Clean
HF:HNO etch
To clean organic/inorganic contaminant
Oxidation
Photolithography
Laser Scribe
Diffusion
Oxidation
Emitter formation
Grow thin passivation oxide
LPCVD
RIE
Deposit AR coating
Make opening for n++ diffilsion
Diffusion
n++ fonnation
RIE
Diffusion
p+ fonnatioll
Metallization
Sintering
Dicing
Testing
I-V testing
IV. TESTING
After dicing, individual cells were wired with electrodes
on both sides of n and p regions. Cells were tested with
current voltage flash tester under 1 sun illumination
condition. Of many cells, the best performing cells are
selected for further characterisation. Cells tested are
observed to have high shunt resistance but significant
series resistance, as shown in figure 7.
978-1-4244-1641-7/08/$25.00
2008 IEEE
oQ.~
............................'
.".
.~"
.~"
Voe
(mV)
Ce116wfr2
Ce117wfr2
Ce118wfr2
Ce119wfr2
612
619
619
610
Ise
(mA)
4.25
4.10
4.34
4.00
FF
9%)
75.9
27.4
66.2
54.9
Erf
(%)
16.5
5.79
14.8
11.2
V. CONCLUSIONS
PICMET 2009 Proceedings, August 2-6, Portland, Oregon USA 2009 PICMET
Using Patent Data to Analyze the Development of the ext Generation of Solar Cells
Fang-Mei Tseng, Yi-Wei Chu, Ya-Ni Peng
Yuan Ze University, International Business Dept., Tao-Yuan, Taiwan, R.O.C
Abstract--With the shortage of raw materials for the
production of crystalline silicon solar cells, the next generation
of solar cells has reached the perfect stage for development.
While in the past researchers have laid emphasis on the
development of crystalline silicon solar cells, the application of
the next generation of crystalline silicon solar cells is different.
Among the thin film solar cells, the most widely-recognized is
the a-Si thin film solar cell with the greatest potential to be
developed and many manufacturers have already invested in
R&D. In this study, we employ the patent portfolio proposed by
Ernst to embark on the analysis of technology development
regarding the a-Si thin film solar cell and find that the major
technology field has reached the mature stage in the technology
life cycle; moreover, four patent strategic clusters are obtained.
The results regarding the company level and technology level of
the clusters are integrated with the profile data and the
development focus to realize the patent performance, technology
capacity and R&D background concerning the technological
fields of the a-Si thin film solar cell and to propose patent
strategies to companies.
I. INTRODUCTION
In recent years, the energy industries throughout the world
have been affected by the dramatic rise in the price of fossil
fuel energy, the short supply of energy, and the crisis brought
about by environmental pollution. This has resulted in the
promotion of a significant change in the industrial framework.
With the sharp increase in world population and economic
growth, the need for energy has been stimulated, and this in
turn has given rise to a sustained increase in the international
price of oil. In August 2006, the international crude oil price
increased to US$78.65 a barrel, and in April 2008 the price
broke through the US$100 a barrel landmark, a feat never
accomplished before. However, as the fossil fuel resources
will eventually become depleted, in the long-run a rise in
prices will surely be an inevitable trend.
In addition, humans have overused the traditional fossil
fuels and this has resulted in the increased density of carbon
dioxide in the atmosphere, which has worsened the global
warming problem, and caused serious damage to the natural
environment. Therefore, the Kyoto Protocol that was signed
by countries worldwide officially went into effect on
February 16th, 2005. The Kyoto Protocol set standards to
control greenhouse gas emissions by its members.
Of all the green energies, solar energy has experienced the
most vigorous growth, and its energy consumption continues
to rise. Due to the rapid growth of the PV market, the demand
for solar cells has increased significantly. The upstream
silicon, found in crystalline silicon solar cells, started to be in
short supply in 2004 and its prime cost increased. The next
generation of solar cells consists of various types, including
the second generation thin film types (amorphous (a-Si), cSi, etc.), compound types, and the third generation organic
types (dye-sensitized solar cell (DSSC), etc.). At present,
within the Si thin film solar cells, the a-Si thin film solar cell
is the main product of interest and has the most
developmental potential in terms of market share. Currently,
many enterprises have already entered the field of a-Si thin
film solar cells and are intensely competing with each other,
as obtaining the patent for this will be the major objective of
their R&D results. Furthermore, through patent analysis,
firms can acquire valuable information on the technological
trends and their competitors respective positions, identify
important technological developments, or engage in patent
portfolio management[1,3,16]. Moreover, further decisions in
regard to the direction of the company patent portfolio will
greatly benefit the corporate strategic planning in the future
[8].
This study uses a-Si thin film solar cells as an example to
modify the patent portfolio analysis proposed by Ernst [8],
through which to further understand the R&D of a-Si thin
film solar cells in the following ways:
1. Using patent information to analyze the developmental
trend of the a-Si thin film solar cell technology and
preparing a report analyzing the patent portfolios among
the main competitive corporations, thereby providing a
valuable source of reference for domestic enterprises and
academia;
2. Using the CHI Research patent indicators to categorize
corporations according to their technical competency in
order to understand their patent technology, and using the
methods and concepts of the patent portfolios as sources
of reference to list which corporations are the technology
leaders and the potential competitors, and which ones are
falling behind;
3. Exploring the patent strategy, technical competency, and
R&D focus on the a-Si thin film solar cell field, which can
be used as a source of reference to draft a patent strategy
and plan R&D directions in the future.
II. LITERATURE REVIEW
A. Patent Analysis
Patents can safeguard the inventions of enterprises and
bring about economic benefits. Through patent data, we also
can see the industry tendency as well as the competitive
power of enterprises or countries [4,19]. Patent analysis is a
method used to transfer these data to useful patent
information. It is a good tool for planning R&D and
managing intellectual property, and it is often used to analyze
competition in relation to techniques and trends [7,12]. The
3064
PICMET 2009 Proceedings, August 2-6, Portland, Oregon USA 2009 PICMET
main process of patent analysis is to use statistical analysis,
multivariate analysis or other quantitative models to analyze
and interpret each field of patents (such as the application
date, assignee name, assignee country, international
classification,etc.) to obtain patent information.
We consider summarized four aspects of patent analysis
from previous papers.
1. Discussing R&D inputs and
outputs: Trajtenberg [18] discussed the relationship between
patents and profit or R&D expenditures; Griliches [11]
pointed out that patent data can be an alternative indicator of
invention-related inputs and outputs when R&D data is
lacking. 2. Measuring a companys technical abilities: Jaffe
[15] used patent analysis to characterize the technological
position of U.S. manufacturing firms; Choung et al. [6]
investigated the evolution of the technological capabilities of
Korean semiconductor firms. 3. Planning a companys
strategies: Ernst [8] used the patent data of mechanical
engineering companies to evaluate the companies
technologicaly ability and demonstrated the application of
patent portfolios for strategic R&D planning purposes. 4.
Discussing market value and financial performance.
However, the above researchers elaborated mostly on the
application of patent information or used statistical methods
to carry on perform an empirical analysis of the companies
patent information. They did not further investigate the
technicalque position or patent strategy of companies. For
this reason, we aim at to remedy this deficiency to by
providing a thorough discussion. We revise the patent
portfolio analysis proposed by Ernst to measure the
technologicaly ability of the sample companies in the nextgeneration solar cell industry and Chen use the patent
portfolio analysis to assess technological ability of solar cell
vendors [5,8,9]. We also use multivariate analysis to develop
a set of suitable patent indicators, determine the patents
position, and induct select groups to discuss the patent
characteristics and the specialty of each group. Through these
analyses, we want seek to provide the more objective
information of regarding the firms R&D strategy.
B. Patent Indicators
Many research organizations and scholars develop
different patent indicators by themselves. The following is an
introduction to different patent indicators, which we use to
design a set of suitable and easy-to-estimate indicators to help
us to carry out the patent portfolio analysis, which objectively
evaluating the firms technological ability and R&D strategies.
CHI patent indicators are brought up by CHI Research
(now is ipIQ) to analyze corporate patent and technological
strength. CHI patent indicators include three standard patent
indicators, four citation indicators and two science linkage
indicators. Narin [17] showed how various technology
indicators (including NP, CII, TCT) can be used to
characterize the position of the EU in 13 technologies
compared to the North American region and to Japan, and be
used by corporate management in competitive analysis.
Huang et al. [14] thought that those indicators were also a
3065
PICMET 2009 Proceedings, August 2-6, Portland, Oregon USA 2009 PICMET
patent quality. As a result, we adopt the 9 patent indicators
drawn up by CHI Research, in addition to another type of
patent citation indicator. The simplicity involved in using
these 10 patent indicators for measurement facilitates the
execution of the patent portfolio analysis methodology.
We adopt factor analysis to reveal the 10 patent indicators
potential characteristics, condensing them into a few factors
and using these extracted factors to measure the corporations
overall patent performance instead of using the traditional
approach of analyzing only the patent quality and quantity
performances. In the next step, we employ a two-step cluster
analysis to categorize the corporations patent competency for
confirming the patent position of each corporation and look
further into the factor differences amongst them. This helps
us to understand the characteristics of each category and to
acquire crucial information. The following list describes the
10 indexes used in this study:
1. Number of Patents (NP): A count of a companys issued
patents in the a-Si thin film solar cell field. By tracking
the number of patents, we can observe the trend of
companies technological development;
2. Patent Growth Percent in Area (PGPA): The change in the
number of patents from one time period to another,
expressed as a percentage. A higher growth rate means
that the technologies are more emphasized;
3. Percent of Company Patents in Area (PCPA): The number
of patents in the a-Si thin film solar cell field divided by
the total number of patents for that company, expressed as
a percentage. This is used to measure the R&D situation
of a company and how the company emphasizes the
technology;
4. Cites Per Patent (or Citation Index, CI): A count of the
citations received by a companys patents from
subsequent patents divided by the total number of patents
for that company. It is used to evaluate the technological
impact of patents;
5. Current Impact Index (CII): The number of times a
companys previous five years of patents are cited in the
current year, relative to all patents in the U.S. patent
system. It can measure the influence of a company in the
most recent five years;
6. Technology Strength (TS): The number of patents
multiplied by the current impact index. It can measure the
scale of the impact of a company on a technological field;
7. Technology Cycle Time (TCT): This is the median age in
years of the U.S. patent references cited on the front page
of the companys patents and is used to measure the speed
of innovation or how fast the technology is turning over;
8. Science Linkage (SL): The average number of science
references cited on the front page of the companys
patents. It mainly evaluates the relationship between the
patenting technique and the scientific research of a
company;
9. Science Strength (SS): The number of patents multiplied
by the science linkage. It can measure the scientific
development level of a company;
3066
PICMET 2009 Proceedings, August 2-6, Portland, Oregon USA 2009 PICMET
IV. ANALYSIS AND DISCUSSION
A. Retrieval and Selection of Patent Data
USPTO is retrieved via the Patent Guider developed by
Learning Tech Corp. in this study. Moreover, the key words
of the patent related to the a-Si thin film solar cell as the
fundamental of data retrieval are collated by a literature
review and expert interviews. As a result, the key words are
(photovoltaic or solar cell) AND thin film AND
(amorphous silicon or a-Si or amorphous Si). As the
conditions for retrieval are decided, the cases applied,
approved and announced during 1/1/1976 to 11/27/2007 are
inquired of based on the fields of the abstract and title in the
USPTO. In the end, 806 patents and 263 assignees are
obtained.The following steps are processed for the further
realization of the assignees and the efficiency of the patent
analysis and for narrowing the extent of patent retrieval to
match the demands of this study: unification in accordance
with companies and corporate M&A, and deletion of the
personal patents and the patents with the citation number 0.
Through manual reading and confirmation article by article,
618 patents and 67 assignees are obtained. Then we list the
leading 20 companies with comprehensive R&D capacities in
the Appendix A by Patent Guider software.
B. Patent Portfolio Analysis
1. The Company Level
To inquire regarding the patent performance of the above
list of 20 companies and research centers, 9 patent indicators
proposed by CHI Research are employed with another patent
citation indicator for the factor analysis of the patent
indicators. Two patents of Teijin Limited are for the same
year, so it is out of the question to calculate the patent growth;
accordingly, the indicator of patent growth is removed.
According to the Kaiser method, the components with
eigenvalues larger than 1 will be reserved for factors
numbering not more than 6 and accounting for 70% or more
of the cumulative percentage of variance. The KMO and
Bartle test are used with the individual MSA variables, all of
which exceed 0.5. The MSA for 2 indicators, the technology
cycle time and science linkage, are both less than 0.5, and so
they are deleted; subsequently, the factor analysis proceeds
for the other 7 indicators following the same steps. Although
3 factors are extracted, the third factor contains only one
variable, namely, the citation index, and therefore CI is
isolated. The factor analysis proceeds with the 6 remaining
patent indicators. In the end, 2 factors are extracted; The
cumulative percentage of variance is 91.796%, indicating that
the factors extracted are able to represent the raw-data
structure. Factor 1 is referred to as Patent Activity. Factor 2 is
3067
PICMET 2009 Proceedings, August 2-6, Portland, Oregon USA 2009 PICMET
RPP
TF1
TF3
TF2
RDGR
Figure 1: The Patent Portfolio for the Technology Level (RGR)
RPP
TF1
TF3
TF2
RDGR
Figure 2: The Patent Portfolio for the Technology Level (RDGR)
of TF3, but RDGR of TF2 is higher than that of both TF1 and
TF3. Compared to the other 2 technological fields, RDGR of
TF2 is relatively higher, meaning that perhaps TF2 is worth
being developed by the companies merely investing their
research resources in TF1 or TF3.
(3) TF3 (Semiconductor device manufacturing: process)
Among all the technological fields, RGR of TF3 is the
highest, but RDGR of TF3 is lower than that of TF2,
indicating that development is worthwhile despite there being
less potential to grow than in the case of TF2. The companies
investing in TF3 may stop investing excessive R&D
3068
PICMET 2009 Proceedings, August 2-6, Portland, Oregon USA 2009 PICMET
resources and search for another technological field in
addition at the same time when developing TF3.
With the cumulative patent number and time of the three
TFs, we use Peral curve to draw a TLC figure as Figure 3.
Figure 3 demonstrates that the development trend of the 3
major fields matches S-curve pattern in the mature period.
That is, the progress of technology slows down in the mature
period. As companies are familiar with the technology
applied to products, the rate of patent licensing will decline.
In addition, the competition among companies is intense. In
other words, R&D staff members in this industry have
accumulated enough experience and knowledge regarding the
technology, so that the commercialized technology is
frequently improved. Since most customers have bought
products developed using such technology, the related
knowledge has been made known and learned publicly.
Therefore, companies should formulate patent strategies such
as authorization, avoidance of designs with patent protection
and improved development, proper patent strategies are to be
chosen for the beneficial usage of patents.
3069
PICMET 2009 Proceedings, August 2-6, Portland, Oregon USA 2009 PICMET
cell field from the silicon cell. When both of them are
compared, the conversion efficiency of the a-Si thin film
solar cell is relatively low and the module efficiency is about
6% to 10%. Sharp changed the tandem of its a-Si thin film
solar cell from two tiers to three in order to improve the
conversion efficiency from the original 8.6% to 10%. With 19
patents related to the a-Si thin film solar cell, Sharp has put
the most R&D effort into TF1 and TF2 and has maintained a
competitive patent position in the three mainstreams. Sharps
R&D strategy, however, focuses more on Patent Activity
rather than Patent Quality. Therefore, its technology is
relatively less competitive in relation to the a-Si thin film
solar cell. Although Sharp is ranked third among the a-Si thin
film solar cell manufacturers in terms of market share, it is
not as dominant as in the patent. The indicator expressed by
the Percent of Company Patents in Area has shown that Sharp
is mostly devoted to the silicon cell instead of the thin film
solar cell. We suggest that only if there is a key thin film solar
cell patent can Sharp remain in a leading position.
As for Kaneka, the citation index ranks it among the
bottom four, which has shown that its technology and market
are significantly different. The main product produced by
Kaneka is the tandem on the glass of a single a-Si solar cell
and module, where the module efficiency can reach up to 8%.
Haneka actively develops Hybrid (a-Si/Poly-Si, tandem on
glass) in order to improve product efficiency. Consequently,
the citation index and current impact indicator of Kaneka are
both quite low, which results in a technology lag cluster.
V. CONCLUSIONS
The research targets of this study are mostly large-scale
resource-integrated enterprises, such as Sharp, Canon, Fuji
Electric and Mitsubishi, which have diversified their
activities into the semiconductor or TFT-LCD fields. The
technologies, materials, equipment and manpower of these
sectors are similar to those of the a-Si thin film solar cell
industry. Besides, silicon is a defective item and is a waste
material in semiconductor manufacturing. Therefore, the
inferior silicon has become the material used to produce the
solar cell. For these two reasons, for those firms to enter the
solar cell field may be regarded as an extension based on
their original technologies. Moreover, half the firms have
developed the first and second generations of solar cells. On
the one hand, the cost structure has made it hard for silicon
cell manufacturers to reduce costs so that they search for
another development. On the other hand, some of the
equipment and technologies are common to both generations,
such as PECVD, and hence many silicon cell manufacturers
will enter the thin film field to engage in parallel
development. The key findings of this study are as follows:
First, the numbers of patents can indicate the situation and
trend for a particular field. As for the solar cell industry,
manufacturers have turned to the a-Si thin film solar cell due
to the shortage of silicon. Besides, the a-Si thin film solar cell
technology and its patents have grown rapidly in recent years
for the reason that the technology can reduce costs. The
conversion efficiency, however, is less effective than that of
the silicon cell so that not all competitors have the same
manufacturing process. We find out that Canon is in the
leading technology cluster, which has dominant performance.
Evergreen Solar is in the emerging potential cluster and has
kept growing. The remaining companies need to put more
effort into key patents even though they currently have high
market share. They may lose their competitive advantage if
their technologies fall behind, if their conversion efficiency
remains inferior, or if they have to rely on technology turnkey services.
Second, there is one circumstance where firms may
engage in R&D but without manufacturing. Take Canon for
example. Canon has not regarded the solar cell as its main
product. This is also the case with the firms that provide
technology turn-key services. Ulvac is an early developer for
thin film solar equipment and is the major supplier to wellknown Japanese thin film solar cell manufacturers, such as
Sharp and Sanyo. Given that fact that Ulvac has provided
mature equipment and the thin film solar industry is growing
rapidly, Ulvac has started to provide turn-key services. This
will lead to further growth as R&D firms file patents and then
transfer technology to others. Firms may then choose to
specialize in either R&D or manufacturing instead of both.
Third, Campell [3] pointed out that the patent indicator is
an important tool when exploring technological development.
However, the messages revealed by patents are always less
evident than those related to technological development,
because the period of the patent review is too long. It usually
takes more than one year when patents are issued. Besides,
some firms do not file patents or file the wrong ones for
strategic purposes to confuse competitors. All of the aspects
will affect the patents quality. Nevertheless, the technology
for the a-Si thin film has already reached the mature stage.
Through R&D it is possible to have a breakthrough to
improve the products conversion efficiency by up to 10% to
12%. Therefore, every firm needs to put more effort into
R&D.
REFERENCES
[1]
[2]
[3]
[4]
[5]
[6]
[7]
3070
PICMET 2009 Proceedings, August 2-6, Portland, Oregon USA 2009 PICMET
[8]
[9]
[10]
[11]
[12]
[13]
[14]
Patent
number
253
40
33
36
32
10
19
19
9
14
12
9
4
9
7
5
2
3
6
4
3071
Self citation
457
26
32
10
5
0
5
4
3
4
0
2
0
0
0
6
0
0
1
1
Other
citation
80
123
37
24
20
31
14
16
23
12
16
14
13
5
8
2
8
7
0
1
Cluster
A
C
C
D
D
C
D
D
C
D
D
D
A
D
D
B
C
C
D
D
PICMET 2009 Proceedings, August 2-6, Portland, Oregon USA 2009 PICMET
APPENDIX B: THE PERFORMANCE INTEGRATION OF THE PATENT PORTFOLIO
Cluster
Leading
Technology
Cluster
Emerging
Potential
Cluster
Technology
Quality
Orientation
Cluster
Company
Canon
Patent Position
TF1
TF2
H
H
TF3
H
Importance
TF1
TF2
H
H
Group
Generation
TF3
H
1,2
H
H
H
L
L
H
H
H
L
L
O
O
O
-
2
1,2
1,2
1,2
1,2
L
H
H
H
L
L
H
L
L
H
H
L
L
H
O
O
O
O
O
O
X
1,2
2
1,2
2
2
1,2
1,2
L
L
L
L
O
O
2
2
L
L
H
H
L
L
H
H
O
O
X
-
1,2
2
1,2
1,2,
Evergreen Solar
3072
Fig. 1. Our R&D scenario for realizing super high- efficiency of 45% and low cost of less than 7 Japanese
Yen/kWh with concentrator MJ solar cells.
002332
002333
have been used to solve problems for low minoritycarrier lifetime of 1-10 ps in the films. Figure 5
shows efficiency of homo junction GaAsN solar
cells as a function of minority carrier lifetime. It is
very important to understand origins of defects in
(In)GaAsN films and to reduce defects density in
order to increase minority-carrier lifetime and to
realize high efficiency (In)GaAsN single-junction
and 4-junction solar cells.
In this study, majority and minority traps in ptype GaAsN films grown by CBE were measured by
the DLTS method and a one electron trap with high
capture cross section was observed by using double
carrier pulse DLTS technique.
Although 6 electrons traps EC1 to EC6 at 0.228,
0.266, 0.292, 0.303, 0.349, and 0.660 eV form the
conduction band minimum Ec, respectively, were
detected, densities of 2 traps EC1 and EC2 are found
to be dependent on N concentration in GaAsN film
as shown in Fig. 6. Because a decrease in EC1 peak
height using a double carrier pulse DLTS technique
was observed with the injection of majority carrier,
the electron trap EC1 is thought to act as a
recombination center.
Fig. 4. I-V curve of a p-GaAs/i-n InGaAsN heterojunction cell with intrinsic layer wideth of 600nm.
002334
Summary
Outline of the Japanese Innovative Photovoltaics
R&D Program was presented, and epitaxial grown
of GaAsN films by using CBE and characterization
of the films by Hall effect and DLTS measurements
were also reported. We have demonstrated high
quality CBE-grown GaAsN films with the
minimum XRD-FWHM of 35.4 arcsec and the
higher electron mobility (2000cm2/Vs) compared
to those with GaAsN films grown by the other
methods. CBE is thought to offer a new means for
growing InGaAsN films that are necessary for use
as the third junction in super high efficiency 4junction AlInGaP/GaAs/InGaAsN/Ge solar cells.
However, one of major problems in (In)GaAsN is
lower minority carrier lifetime because only 2% and
11.3% efficiencies have been obtained with GaAsN
p-n and InGaAsN p-i-n single-junction cells. We
have characterized defects in CBE-grown GaAsN
by using DLTS measurements. The electron trap
EC1 in CBE-grown p-type GaAsN is found to act as
a recombination center. Possible origin of defect
level EC1 is also discussed.
Acknowledgments
This work is partially supported by the New
Energy and Industrial Technology Development
Organization (NEDO) under the Ministry of
Economy, Trade and Industry (METI).
References
1) T. Takamoto, M. Kaneiwa, M. Imaizumi and M.
Yamaguchi, Progress in Photovoltaics 13, 495
(2005).
2) S. R. Kurtz, D. Myers and J.M. Olson,
Proceedings of the 26th IEEE Photovoltaic
Specialists Conference, (1997) p. 875.
002335
ABSTRACT
Sithinfilm
I
I
I
Ge thm film
Botton contact
Figure 1. One-sun AM 1.5 solar spectra (left), spectral
absorption coefficients for Si, Ge, and Ino.s3Gao.47As
substrates (center), and proposed solar cell configuration
(right).
000608
000609
- --
---,
'E l e+9
u
le+8
.~ le+7
c: le+6
Experimental
OJrve fitting (our data)
l e+5
Vi le+4
le+3
0.01
0.1
10
100
1000
10000
--------1
3. Modeling
Photod et ecto r _ _
Foe us in g len
Glas 5 SU b strate
Gefilm
C o ll im at in glen 5
C h op per
C o lo r fi lt er w h e e l
Colle ct in g lens
Xe n on A r c li ght
(a)
(c)
(b)
(d)
000610
6. REFERENCES
____ 1 . 4 . - - - - - - - - - - - - - - - - - - - - ,
;J
--- Reference
< 1 2 -___
~ 1.0
--
- -0 - -0 - -0 - -0 - -0 - -0
'5 0.8
'"
=
~
l-
Eo-<
"0
0'6 ~
0.4
. V~.==:::-~.~
.t:::l 02
';
5
lo
0-0
0.0
1050
A_
1200
1350
1500
1650
1800
A(nm)
Figure 8. Optical transmission measurements from Si/Ge
films in 1-1.6-llm wavelength range.
____ 1 . 4 , - - - - - - - - - - - - - - - - - - ,
;J
--- Reference
<
__
___ 1.2
~ 1.0
.s
'5
'"
g
- -0 - -0 - -0 - - 0 - -0 - - 0
0.8
0.6
i:: =::::.--\ _=
5
l-
O.O-t-~-..-~-r-~--r-~-,--~-i
1050
1200
1350
1500
1650
1800
A(nm)
Figure 9. Optical transmission measurements from Ge
films in t -t .s-um wavelength range.
smoother Ge film that has no defects and no scattering
(Fig. 4, left) in contrast with highly rough (Fig. 2, left) Ge
films that enhance scattering and thus lead to reduced
normal-incident transmission.
5. SUMMARY AND FUTURE WORK
7. ACKNOWLEDGEMENTS
Partial funding for this work was provided by WPAFB and
the University Kebangsaan Malaysia.
000611
I. INTRODUCTION
2
The photocurrent IL produces a voltage drop across the
resistive load, which forward biases the pn junction. The
forward-bias voltage produces a forward bias current IF as
indicated in the figure. The net pn junction current, in the
reverse-bias direction is:
I = IL IF = IL IS [exp
-1]
(2.0)
(b)
The net current is zero and the voltage produced is the opencircuit voltage. The photocurrent is just balanced by the
forward-biased junction current so we have:
I = 0 = IL Is [exp
) (2.1)
x 100% =
x 100%
(2.2)
3
Financial:
There are no financial constraints due to the fact that the
project will be very inexpensive. The only purchase that will
be needed is a laser diode to shine light on the solar cells and
the amorphous silicon cells. Figure 2.3 shows two laser
diodes found on ebay.com, the left ones starting price was
$9.99 and the right ones starting price was $14.99. The
amorphous silicon cells are also readily available and
inexpensive. The rest of the equipment although very
expensive will be provided by Professor Leon Shterengas in
the lab. Proving that cost for this project will be minimal.
(b)
(b)
(a)
Figure 2.5 (a) Laser hat (b) Solar cells in parallel and series
4
The laser hat shown in figure 2.5 (a) will be used to hold the
laser diode as well as control the lasers voltage and
temperature.
Figure 2.5 (b) is an example of solar cells used for testing.
The multimeter shown in figure 2.6 will be used to measure
voltage and current for analysis. The multimeter can be
controlled by the computer using LABview through the
connector shown in figure 2.6 (a).
(a)
Figure 2.7 (a) Connector (b) Laser Diode Controller
(b)
Figure 2.8
Figure 2.9
Figure 2.8 and 2.9 Examples of LabView which will be used
to control data in our lab to our discretion.
We decided to use blue laser diodes to shine light onto the
solar cells due to the fact that they are reasonably price and
easy to control
C. Final Design
(b)
5
apparatus will have a dye sensitized solar cell under a laser
diode controlled by lab view as well. Due to the spectral
response of the dye and amorphous silicon, we have chosen a
laser diode that falls within the range of 400-500nm.
We will most likely be using a laser diode with the parameters
of 405nm (Figure 2.3). The amorphous silicon cell is readily
available to us in Professor Leons lab. The DSSC will be
made in a lab that we have access to on campus so this does
not pose any great issues.
Building the dye sensitized solar cell and silicon based solar
cell:
Figure 2.10 Circuit used to measure I-V curve on oscilloscope
The dye sensitized solar cell will be made in a lab. The silicon
based solar cell will be made available to us without any
design from our part.
In order to measure the efficiency of the solar cells we will be
looking at the I-V curves, an example shown in figure 2.13 of
both cells under a blue laser diode. We will measure the IV
curve using two different methods. LAB VIEW will be used
when necessary and applicable.
The first method is by using an oscilloscope to get a solid
curve. First we will build the circuit show in figure 2.10. R1
should be 1K, the frequency of the sin wave should be 1kHz
and the voltage across the secondary winding of the
transformer should be in the range of 5V to 10V. The resistor
R1 not only serves as a means of limiting the current but can
also be used to monitor current because the voltage across the
resistor corresponds to the current across the solar cell. Due
the fact that the solar cell and the resistor R1 are in series the
current across both of them is equal. Equation 2.4 shows how
the voltage across the resistor corresponds to the current
across the solar cell in mA.
(2.4)
A transformer is used to provide DC isolation of the circuit
ground from the signal generator ground. DC isolation is
important because we need to simultaneously display the
voltage of the solar cell and the voltage of R1 on the
oscilloscope. A floating signal source is created by
grounding the AC signal source and the transformer. This
floating signal source allows us to place a ground were the
solar cell the resistor meet. The voltages of the solar cell and
the resistor are then measured from the points shown on figure
2.10 and the ground between them. The top connection in
figure 2.10 is connected to the X-input of the oscilloscope
(obtaining the solar cells voltage signal) and the bottom
connection in figure 2.10 is connected to the Y-input of the
oscilloscope (obtaining the resistors voltage signal). Display
both of the voltages on the oscilloscope reverse the voltage
signal from the resistor (Y) to obtain set the oscilloscope to the
X-Y regime and the I-V characteristic curve should be
displayed for the solar cell.
Figure 2.11 Circuit used to measure I-V curve using point-topoint measurements
6
III. SYSTEM IMPLEMENTATION AND TESTING
IV Measurements:
There are a few ways the IV curve will be measured for this
experiment.
How will measurements be made?
Primarily, plot the simulated diode I-V characteristics in linear
scale, i.e. I versus V. Identify the voltage range in which this
dependence is linear. From the slope of the linear dependence
of I on V in this voltage range find the diode series resistance.
Secondly, plot the same simulated diode I-V it in semi-log
scale, i.e. ln(I) versus V. Identify the voltage range in which
the dependence of ln(I) on V is linear. Determine the diode
non-ideality factor from the slope of the I-V characteristic in
semi-log scale.
Figure 2.14
The final design was chosen based on the constraints of the
silicon and dye sensitized solar cells constraints. As explain
earlier in the paper the laser diode used to shine light on the
solar cells must be in the wavelength of the peak of the solar
spectrum of both solar cells. The peak for both solar cells is
around 400nm -500nm. The laser diodes shown in figure 2.3
are 405nm which fits into that constraint.
7
noting. Specifically, mining and nuclear waste are both
hazardous to people in terms of direct physical harm as well as
the direct physical harm of the environment in which these
facilities exist. Not only are the people who are near the
plants negatively affected, but the long-term effects of acid
rain and global climate change hurt ecosystems that we
depend on for our survival.
Our current energy use is unethical because it harms life
around the globe. Outputs from coal plants cause toxic
chemicals to accumulate in the bodies of plants and animals.
Habitat destruction causes extinction and displacement of
species. For example, hydropower has severely threatened the
lives of salmon in the Pacific Northwest--species have become
regionally extinct in many parts. Solar power provides a
natural way to sustain life in all parts of the world without
causing harm to any species.
C. IMPACT ON SOCIETY
Given the current energy crisis our study of DSSC will only
impact society and environment in a positive way. Providing
a less expensive method of collecting solar energy will
positively impact society and the environment because it will
aid to decrease its dependence on fossil fuels. In parts of the
world that are deprived of any means to electricity, solar cells
can perhaps help if a way is found to make them less
expensive. Refrigeration and cooling, two necessary
ingredients for a comfortable life are being denied to humans
in hundreds of cities abroad.
Solar energy is better to our environment than any known
traditional forms of energy especially against fossil fuel and
coal. It has many uses such as for electricity production or for
heating of water through using photovoltaic. Solar energy can
be used to power heated swimming pools; it can be used to
power cars or attic fans, calculators and other appliances. Due
to high prices of the electric bill using conventional sources,
the demand of solar energy is increasing. The worldwide
demand for solar energy is greater than its supply.
V. SUMMARY AND CONCLUSIONS
Our senior design project deal with two types of solar cell
technologies: industry standard Si-based and the emerging one
based on novel dye sensitized materials. Dye sensitized solar
cells will be fabricated in a laboratory setting. FTIR and
Raman microscopy will be used to study the surface make-up
of the novel solar cells. A thin film flexible photovoltaic cell
will be characterized.
The efficiency of the dye sensitized solar cells will be
compared to that of thin film Si-based solar cells. We will
design the experiments and perform measurements using
LabView controlled data acquisition. The prospects to
improve the efficiency of the novel solar cell technology will
be discussed.
ACKNOWLEDGMENT
The authors would like to sincerely thank their advisor,
2010 International Conference on Design & Technology of Integrated Systems in Nanoscale Era
semiconductor alloy that is substantially latticematched to the silicon substrate. The polarity of the
active p-n junction cell is either p-on-n or n-on-p.
Emitter layer
Base layer
Bottom active
substrate
I.
INTRODUCTION:
Top active
-1-
2010 International Conference on Design & Technology of Integrated Systems in Nanoscale Era
III.
EXPERIMENTS DETAILS
(1)
(2)
Where:
(3)
(4)
10
GaAs
BInGaAs epilayer
10
Intensity (a u)
10
BInGaAs
10
10
10
10
10
10
31,5
32,0
32,5
33,0
/2
33,5
34,0
34,5
-2-
35,0
2010 International Conference on Design & Technology of Integrated Systems in Nanoscale Era
10
BInGaAs QW
10
intensity (a u)
10
10
10
10
10
-1
10
30
31
32
33
34
35
36
/2
Figure.2.b. High resolution X-ray diffraction /2 spectra of
BxInyGa1-x-yAs/GaAs quantum well.
BInGaAs QW
T=300 K
2
Pexe=80W/cm
PL Intensity (a u)
Epl = 1,19 eV
IV.
1.1
1.2
1.3
CONCLUSION:
1.4
Energy (eV)
-3-
2010 International Conference on Design & Technology of Integrated Systems in Nanoscale Era
REFERENCES
[1] G. Leibiger, C. Krahmer, J. Bauer, H. Hernberger, V.
Gottschalch, J. Crystal Growth 272, 732-738 (2004)
[2] V. Gottschalch, G. Leibiger, G. Bendorf, J. Crystal Growth
248, 468 (2003)
[3] R. Asomoza, V. A. Elyukhin, and R. Pena-Sierra Appl. Phys.
Lett 78, 2494 (2001)
[4] J .F. Geisz, D.J.Friedman, J.M.Olson, S.R.Reedy,
A.B.Swartzlander, B.M. Keyes, A.G. Norman, Appl. Phys. Lett
76, 1443 (2000)
[5] Tom MarkvartLuis and Castafier , Solar Cells:
Materials,Manufacture and Operation, Elsevier Ltd, ISBN13:978-1-85617-457-1 (2006)
[6] H. Dumont, D. Rutzinger, C. Vincent, J. Dazord, Y. Monteil,
F. Alexander, and J.L.Gontner, Appl. Phys. Lett 82, 1830 (2003)
[7] L. Auvray, H. Dumont, J. Dazord, Y. Monteil, J,Bouix, C.
Bru Chevallier, J. Crystal Growth 221, 475 (2007)
-4-
000090
1
15
2
ITO/PEDOT/MWNTs+P3HT:PCBM/TiOx/Al
ITO/PEDOT/MWNTs+P3HT:PCBM/TiO2/Al
ITO/PEDOT/P3HT:PCBM/TiOx/Al
ITO/PEDOT/P3HT:PCBM/Al
10
(c)
(b)
ITO/PEDOT/P3HT:PCBM/TiO2/Al
(a)
ITO/PEDOT/P3HT:PCBM/Al
0
-5
-10
-15
-1
-0.5
0.5
Voltage (V)
(b)
000091
1
10
2
8
6
Al/PCBM:P3HT/PEDOT:PSS/ITO(a)
Al/C60/PCBM:P3HT/MoO3/ITO(c)
4
2
Al/PCBM:P3HT/MoO3/ITO (b)
0
-2
-4
-6
-8
-0.5
0.5
Voltage (V)
Figure 3 Energy band diagram of the fabricated device
with incorporation of MWNTs and TiOx layer as electron
selective layer presenting possible charge transportation
path in the device.
Possible
charge
transportation
with
MWNTs
incorporation and TiOx as electron selective layer is
explained in terms of energy band diagram as shown in
figure 3. The Work function of MWNTs is in the range of
4.5~5.1 eV, which is close to HOMO level of P3HT; it
signifies possible hole transportation through MWNTs. It is
expected that surface chemical groups, due to
functionalization increase the work function of the MWNTs.
With increase work function of MWNTs it will be much
closer to HOMO level of P3HT and provide efficient hole
transportation. Whereas TiOx layer is expected to work as
an efficient hole blocking layer due to deep HOMO level
and suitable LUMO level close to Fermi energy of Al as
shown in the energy band diagram.
In a different set of experiments, organic solar cells were
fabricated with MoO3 as a hole selective interlayer at the
anode instead of PEDOT:PSS. It has been previously
demonstrated that PEDOT:PSS is aqueous and naturally
acidic, which would cause interface instability in organic
solar cells. Hence, PEDOT:PSS facilitate degradation of
000092
1
Conversion efficiency(%)
0.4
(Al/PCBM:P3HT/MoO3/ITO)
0.2
(Al/C60/PCBM:P3HT/PEDOT:PSS/ITO)
0
0
10
12
Time (h)
Figure 6 Device performance as a function of time with
device configuration (a) ITO/PEDOT:PSS/P3HT:PCBM/Al
(b) ITO/MoO3/P3HT:PCBM/Al
(c) ITO/MoO3/P3HT:PCBM/C60/Al under illumination (AM
2
1.5 100 mW/cm ).
From the stability measurement, it shows that with MoO3
as hole collecting layer instead of PEDOT:PSS layer is
much more stable. It has been demonstrated that in an
organic solar cells the hygroscopic PEDOT:PSS layer can
adsorption water in air, thereby affecting the active layer.
Measurements of the charge mobility and hole injection
after air exposure showed that the degradation increases
the resistance of the PEDOT:PSS/blend layer interface.[]
On the other hand, It has been observed that PEDOT:PSS
layer accelerate oxidation of the aluminum electrode at the
aluminum/photovoltaic layer interface. Whereas, device
fabricated with MoO3 interlayer can prevent water
adsorption in air and providing a stable interface with the
000093
1
000094
1
(b)
002204
Layout Design
The active (or junction) area designs proposed below are
based on combinations of grid shapes. The goal is to
grow different absorber thicknesses to vary the area of the
silicon/dielectric interface (Figure 2). On the other hand,
different geometries can exhibit a dependence relationship
on Voc. Higher quality growths are possible for specific
dielectrics and the crystal defect density can be optimized.
Other characteristics of the epitaxial growth are:
coalescence and substrate crystal orientation vs. design
orientation.
Achieving coalescence increases the crystal defect
density which can then be reduced with a clever via
design. In general, the geometrical design of the active
area is a tool to determine contributions and losses from
the epitaxial absorber back surface regarding open circuit
voltage.
With our microelectronic approach, these
advanced grid designs are compared to their planar
equivalents.
SOLAR CELL FABRICATION
The microelectronic approach starts with the growth of a
dry oxide on the conductive Silicon substrate. The
substrate is (100) p-type crystalline silicon of
approximately 0.01ohm-cm bulk resistivity.
The dry
oxidation is done at 900C in a tube furnace. After the
oxide
growth,
active
areas
are
defined
by
photolithography.
The solar cell is grown by high
temperature atmospheric pressure chemical vapor
deposition (APCVD) with a high growth rate.
002205
002206
002207
(b)
002204
Layout Design
The active (or junction) area designs proposed below are
based on combinations of grid shapes. The goal is to
grow different absorber thicknesses to vary the area of the
silicon/dielectric interface (Figure 2). On the other hand,
different geometries can exhibit a dependence relationship
on Voc. Higher quality growths are possible for specific
dielectrics and the crystal defect density can be optimized.
Other characteristics of the epitaxial growth are:
coalescence and substrate crystal orientation vs. design
orientation.
Achieving coalescence increases the crystal defect
density which can then be reduced with a clever via
design. In general, the geometrical design of the active
area is a tool to determine contributions and losses from
the epitaxial absorber back surface regarding open circuit
voltage.
With our microelectronic approach, these
advanced grid designs are compared to their planar
equivalents.
SOLAR CELL FABRICATION
The microelectronic approach starts with the growth of a
dry oxide on the conductive Silicon substrate. The
substrate is (100) p-type crystalline silicon of
approximately 0.01ohm-cm bulk resistivity.
The dry
oxidation is done at 900C in a tube furnace. After the
oxide
growth,
active
areas
are
defined
by
photolithography.
The solar cell is grown by high
temperature atmospheric pressure chemical vapor
deposition (APCVD) with a high growth rate.
002205
002206
002207
2010 International Conference on Computer Application and System Modeling (ICCASM 2010)
I.
INTRODUCTION
978-1-4244-7237-6/$26.00
II.
2010 IEEE
Figure 1.
V9-26
2010 International Conference on Computer Application and System Modeling (ICCASM 2010)
(a)
(b)
Figure 2. (a) Proposed SRAM bitcell and (b) its transfer characteristics.
C. Standby Mode
In the standby mode, the supply voltage of SRAM bitcell
is cut off by the NMOS transistor N1/N2 to reduce the
leakage power. However, the supply voltage cannot be
reduced arbitrarily as bitcells will not be able to hold the
contents of the bitcells. So the size of N1/N2 transistors
should be set carefully. The comparison of hold SNM at low
supply voltages is shown in Fig. 3. As SRAM bitcells of [2],
[3] and [4] adopt the conventional inverter pair based
architecture, their transfer characteristics are therefore
similar with 6T in hold mode.
It can be clearly observed from the Fig. 3 proposed
bitcell exhibits superior hold margin. In IBM 0.13 m
process technology, the hold SNM of 6T and ST bitcells is
166 mV and 193 mV, while proposed bitcell exhibits 238
mV hold SNM at 400 mV, which is significantly better. Its
hold margin is 30.2% and 18.9% greater than 6T and ST (at
400 mV).
Since the cut-off transistors N1 and N2 neither introduce
extra load on the wordline nor require an auxiliary circuit to
switch the bitcell into a stand-by mode, N1 and N2 do not
have any influence on dynamic power dissipation and access
time. Proposed SRAM bitcell reduces leakage power without
increasing dynamic power dissipation in the standby mode.
B. Write Operation
Maintaining a sufficient write margin is challenging in
subthreshold/nearthreshold SRAM design due to the small
V9-27
2010 International Conference on Computer Application and System Modeling (ICCASM 2010)
(a)
(b)
III.
(c)
Figure 3.
Hold margin of (a) 6T, (b) ST, and (c) proposed design (at 400
mV).
D. Read Operation
Read failure is the most critical problem in realizing
subthreshold memory as it imposes the smallest margin in
comparison with the hold and write margins.
EXPERIMENTAL RESULTS
9P9
5HDG610
P9
GFP9
9P9
(a)
(b)
IV.
(c)
Figure 4. The read margin of (a) 6T, (b) ST, and (c) proposed design (at
400 mV).
CONCLUSION
V9-28
2010 International Conference on Computer Application and System Modeling (ICCASM 2010)
[3]
[4]
[5]
[6]
REFERENCES
[1]
[2]
[7]
[8]
V9-29
(b)
(a)
Figure 1: (a) Laboratory characterization of a MJSC under continuous concentrated solar radiation. (b)
University of Ottawa's solar tracker station, in which the performance of concentrated photovoltaic systems are
measured over long periods of time under real-world conditions.
209
1042
Divergence of Technologies
Case Studies
Japan
Sharp
USA
SunPower
First Solar
Nanosolar
Solyndra
China
Suntech
Theories
Dynamics of
Innovation
The S-curve
Germany
Q-Cells
Background
Paradigm Shift
in Solar Cell Competition
solar cell in 2007 and 2008 [3]. Q-Cells focused on the c-Si
and poly-Si solar cell process by the automation to reduce the
cost. However, Q-Cells got into red in 2008 and fell out of
first place by First Solar. Therefore, Q-Cells restructured the
solar cell business and expand from c-Si and poly-Si cell
process to solar cell module and CIGS solar cell [3].
E. Summary of Global Solar Cell Industries
Based on the above case studies of global solar cell
industry, the diverse technologies of solar cell are classified
according to the S-curve [6], [7], as shown in Fig. 4. c-Si and
poly-Si solar cells are mature technologies. A-Si thin film
solar is near maturity. Cupper gallium diselenide (CGS),
CIGS and CdTe solar cells are in growth stage. The
embryonic technologies are organic, spherical Si, quantum
dot and many embryonic solar cells, which many venture
companies are researching.
New
Invention
Period
Technology
Improvement
Period
Maturity
poly-Si
a-Si TF SC
Growth
CGS, CIGS
Aging
c-Si
MatureTechnology
Period
CdTe SC
Embryonic
Quantum dot
Spherical Si SC
Organic SC
Embryonic SC
Time
Japan
USA
CdTe SC
1st Group
(Sharp, Kyocera,
Sanyo,
Mit bi hi)
Solar Island
Kyusyu
SolarPower
Nanosolar
Solyndra
First Solar
Venture
China
Suntech
Other Compalies
Germany Q-cells
Production Level
R&D Level
Semiconductor
1000
100
poly-Si
Solr Cell
10
a-Si
Solar Cell
LCD
1
1
10
100
Fig. 5. Process days and unit area value of Sola Cell, LCD
and semiconductor devices. (Author made)
C. Conditions of divers technologies
Utterback researched the dynamics of innovation from the
case studies of plate glass, assemble products, film
photography, ice etc [4]. However, as described above, in
case of the solar cell industry, the innovation dynamics theory
of Utterback can not apply to the innovation.
What are the conditions to accept the diverse technologies
such as the solar cell industries?
Therefore, the conditions of diverse technologies in
business are extracted from the results of the paradigm shift
of solar cell competition, case studies of global solar cell
industries.
First condition of diverse technologies is simple
technology, which means short process day, because it is easy
to enter the business by low input resources and time. In other
word, simple process accepts the diverged technologies to
enter the business. In the contrast, complex process needs to
integrate many technologies, not one technology, in case of
long process day, such as semiconductor. The industries with
complex processes not accept diverse technologies to simplify
Acknowledgments
I express my appreciation for Professor Richard B. Dasher
of U.S.-Asia Technology Management Center of Stanford
University to give me the opportunity to research at Stanford
University. I thank Japan Society for the Promotion of
Science for the financial support.
References
1. New Energy and Industrial Technology Development
Organization (NEDO), "Why did Japan achieve the
world's No. 1 share in solar cells? (Japanese),"
Nedobooks, 2007.
2. PV News, May and April 2008.
3. Outlook of Solar Cell Industry 2011, Sangyo Times
2010.
4. James M. Utterback, Mastering the Dynamics of
Innovation, Harvard Business School Press 1994.
5. J.C. Fisher and R. H. Pry Simple Substitution Model of
Technical Change. Technological Forecasting and
Social Change, vol. 3, pp. 75 88.
6. Tarek Khalil, Chapter 5 Technology Life Cycles,
Management of Technology, McGraw-Hill Higher
Education, (2000), pp80-94.
7. Press Announcement of Sharp Corporation, Sharp to
Develop Manufacturing Complex for the 21st Century,
http://sharp-world.com/corporate/report/plant/index.html,
31 July 2007.
8. Paul Wormser, Thin-Film Companies Heading for
Terawatt-hour Module Manufacturing: Micromorph,
presented at PHOTONs PV Thin Film Conference,
February 2, 2010, San Francisco.
9. Sharp Corporation, Utility-Scale Thin Film
Photovoltaics, exhibited at Solar Power International 09
Exhibition, Oct 27 29, 2009, Anaheim, California, USA.
10. Sharp Corporation, Operation Starts at Solar Cell Plant
in GREEN FRONT SAKAI, press-released in March 29,
2010.
11. SunPower, [Online]. Available:
http://us.sunpowercorp.com/business/
12. SunPower, PG&E Signs 800 MW Photovoltaic Solar
Power Agreements, [Online]. Available:
http://investors.sunpowercorp.com/releasedetail.cfm?Rel
easeID=328221
1042
Divergence of Technologies
Case Studies
Japan
Sharp
USA
SunPower
First Solar
Nanosolar
Solyndra
China
Suntech
Theories
Dynamics of
Innovation
The S-curve
Germany
Q-Cells
Background
Paradigm Shift
in Solar Cell Competition
solar cell in 2007 and 2008 [3]. Q-Cells focused on the c-Si
and poly-Si solar cell process by the automation to reduce the
cost. However, Q-Cells got into red in 2008 and fell out of
first place by First Solar. Therefore, Q-Cells restructured the
solar cell business and expand from c-Si and poly-Si cell
process to solar cell module and CIGS solar cell [3].
E. Summary of Global Solar Cell Industries
Based on the above case studies of global solar cell
industry, the diverse technologies of solar cell are classified
according to the S-curve [6], [7], as shown in Fig. 4. c-Si and
poly-Si solar cells are mature technologies. A-Si thin film
solar is near maturity. Cupper gallium diselenide (CGS),
CIGS and CdTe solar cells are in growth stage. The
embryonic technologies are organic, spherical Si, quantum
dot and many embryonic solar cells, which many venture
companies are researching.
New
Invention
Period
Technology
Improvement
Period
Maturity
poly-Si
a-Si TF SC
Growth
CGS, CIGS
Aging
c-Si
MatureTechnology
Period
CdTe SC
Embryonic
Quantum dot
Spherical Si SC
Organic SC
Embryonic SC
Time
Japan
USA
CdTe SC
1st Group
(Sharp, Kyocera,
Sanyo,
Mit bi hi)
Solar Island
Kyusyu
SolarPower
Nanosolar
Solyndra
First Solar
Venture
China
Suntech
Other Compalies
Germany Q-cells
Production Level
R&D Level
Semiconductor
1000
100
poly-Si
Solr Cell
10
a-Si
Solar Cell
LCD
1
1
10
100
Fig. 5. Process days and unit area value of Sola Cell, LCD
and semiconductor devices. (Author made)
C. Conditions of divers technologies
Utterback researched the dynamics of innovation from the
case studies of plate glass, assemble products, film
photography, ice etc [4]. However, as described above, in
case of the solar cell industry, the innovation dynamics theory
of Utterback can not apply to the innovation.
What are the conditions to accept the diverse technologies
such as the solar cell industries?
Therefore, the conditions of diverse technologies in
business are extracted from the results of the paradigm shift
of solar cell competition, case studies of global solar cell
industries.
First condition of diverse technologies is simple
technology, which means short process day, because it is easy
to enter the business by low input resources and time. In other
word, simple process accepts the diverged technologies to
enter the business. In the contrast, complex process needs to
integrate many technologies, not one technology, in case of
long process day, such as semiconductor. The industries with
complex processes not accept diverse technologies to simplify
Acknowledgments
I express my appreciation for Professor Richard B. Dasher
of U.S.-Asia Technology Management Center of Stanford
University to give me the opportunity to research at Stanford
University. I thank Japan Society for the Promotion of
Science for the financial support.
References
1. New Energy and Industrial Technology Development
Organization (NEDO), "Why did Japan achieve the
world's No. 1 share in solar cells? (Japanese),"
Nedobooks, 2007.
2. PV News, May and April 2008.
3. Outlook of Solar Cell Industry 2011, Sangyo Times
2010.
4. James M. Utterback, Mastering the Dynamics of
Innovation, Harvard Business School Press 1994.
5. J.C. Fisher and R. H. Pry Simple Substitution Model of
Technical Change. Technological Forecasting and
Social Change, vol. 3, pp. 75 88.
6. Tarek Khalil, Chapter 5 Technology Life Cycles,
Management of Technology, McGraw-Hill Higher
Education, (2000), pp80-94.
7. Press Announcement of Sharp Corporation, Sharp to
Develop Manufacturing Complex for the 21st Century,
http://sharp-world.com/corporate/report/plant/index.html,
31 July 2007.
8. Paul Wormser, Thin-Film Companies Heading for
Terawatt-hour Module Manufacturing: Micromorph,
presented at PHOTONs PV Thin Film Conference,
February 2, 2010, San Francisco.
9. Sharp Corporation, Utility-Scale Thin Film
Photovoltaics, exhibited at Solar Power International 09
Exhibition, Oct 27 29, 2009, Anaheim, California, USA.
10. Sharp Corporation, Operation Starts at Solar Cell Plant
in GREEN FRONT SAKAI, press-released in March 29,
2010.
11. SunPower, [Online]. Available:
http://us.sunpowercorp.com/business/
12. SunPower, PG&E Signs 800 MW Photovoltaic Solar
Power Agreements, [Online]. Available:
http://investors.sunpowercorp.com/releasedetail.cfm?Rel
easeID=328221
D.T. Barki
Email: hasitakaja@rocketmail.com
Email: dtbarki@usa.net,
dtbarki@solarnest.net ; Web:
www.solarnest.net
Abstract:
Achieving energy security through sustainable
technologies is the most challenging objective that
needs to be fulfilled within the context of rising
energy demand in a country like India. Although
nuclear energy is expected to meet a part of Indias
energy demand by 2030, this would still be non
substantial and more importantly an unsafe way of
producing power as demonstrated by a series of
negative events that have marred its reputation as
a sustainable technology, particularly in the wake
of recent nuclear disaster at the Fukushima power
plant in Japan. And with increasing pressure on
every country to switch over from fossil-fuel based
energy to clean and renewable energies, India has
rightly announced its ambitious National Solar
Mission (NSM) to install 20 GW of solar power by
2022 in Nov 2009. This paper discusses how India
can achieve this target by adopting various solar
PV technologies and processes. Until not so far
back in time, poly-silicon (Si) production primarily
catered to the semiconductor markets; however,
the world now realizes that much silicon is
required for the fast growing solar industry as well.
The only Si production process that has dominated
the world market has been the Siemens process,
which is highly capital and energy intensive.
Further this process is not carbon-neutral, since it
uses coke and woodchips as reducing agents,
causing considerable deforestation. Scientists and
technologists across the world realize that spending
I.
INTRODUCTION
II.
IV.
V.
SOLSILC
The Direct Metallurgical route produces silicon from ultrahigh purity raw materials. Silicon production according to the
SOLSILC technology has resulted in silicon of very high
purity and has already been running at a pilot scale of 50-100
ton/yr. This carbothermic production route can, in principle,
reduce the cost of feedstock to below 15/kg. India should
consider working on this technology.
The idea behind the SOLSILC Technology was to develop a
new two-step high temperature plasma-process for solar
grade silicon (SoG-Si) production, based on carbothermic
reduction. The SOLSILC Technology relied on expertise of
the metallurgical production methods for Silicon and SiC,
and PV cell processing.
The work involves with the development of a low-purity Siprocess facility (say, 6N purity instead of 10N or 11N) in
order to control the production process before a clean
environment is established. In order to control the purity of
the silicon, the selection and supply of raw materials is
crucial. Specially selected Carbon Black and quartz are used
in the process.
The bottleneck, as of now, that all producers of SoG-Si based
on carbothermic reduction of quartz face, is to remove carbon
and metals impurities like iron, aluminium etc from the
silicon. The source of the impurities is mainly the quartz. In
order to comply with the objective of low cost, this has to be
done on large-scale in a cost effective and reliable manner. A
large number of experiments were carried out first in small
scale.
Kazakhstan carbothermic route
A research group at the Kazakh National Technical
University in Kazakhstan is also working on a carbothermic
reduction process for solar grade silicon. They are using very
clean and high purity quartz from deposits in Kazakhstan.
Silicon by electrolysis
Electrolysis has been the way to aluminium for more than
100 years. Silicon can also be produced by electrolysis. A
process was patented by Stubergh in 1994. The process is
based on the dissolution of quartz in a fluoride-containing
bath at 1000C, and decomposition of the quartz into silicon
and oxygen. Silicon precipitates at the electrode. Afterwards,
the silicon is crushed and cleaned with acid. In order to
obtain a clean product, the silicon is melted and thereafter
crystallized into ingots for subsequent sawing into wafers.
The slag that comes with the silicon from the electrolyte has
a high solubility for impurities and can be separated from the
silicon. Today, Stubergh has scaled up his tests to a larger
laboratory scale. The work is partly done in cooperation with
SINTEF and IFE. He has established a company Norwegian
Silicon
Refinery and has made plans for a full scale test facility.
SINTEF is also doing research on production of silicon by
electrolysis. RSI Silicon, USA, has also adopted similar
electrolytic process that consumes only 1/15th energy as
compared to Siemens process. And the process is almost
carbon-neutral.
SUMMARY AND
RECOMMENDATIONS
VI.
VII.
REFERENCES
1.
2.
3.
4.
5.
214
I. INTRODUCTION
ECAUSE of band gap limitations, a single-junction solar
cell cannot achieve efficiency higher than 30% under one
sun illumination. However, higher efficiency can be reached
with tandem solar cells [1]. Over the past three years, the efficiency reported of the two-terminal triple-junction solar cells
has increased systematically from 41.1% [2] to 41.6% [3] and
to 42.3% [4]. The highest reported efficiency so far for a twoterminal triple-junction solar cell system from Solar Junction
has the efficiency of 43.5 2.5% [5]. One way to further improve the efficiency of a multijunction solar cell system is to
increase the number of junctions beyond 3 [1], [6]. Nell et al.
have developed a spectral pn junction model for tandem solar
cell design [6]. In their paper, they demonstrated that theoretical
efficiencies are 28.3% for a single-junction solar cell and 39.1%,
Manuscript received August 16, 2011; revised November 15, 2011; accepted
December 18, 2011. Date of publication February 23, 2012; date of current
version March 16, 2012.
X. Lu was with the Department of Electrical and Computer Engineering, University of Delaware, Newark, DE 19716 USA. He is now with Applied Materials
Company, Santa Clara, CA 95054 USA (e-mail: luxuesong001@gmail.com).
S. Huang was with the Department of Materials Science and Engineering, University of Delaware, Newark, DE 19716 USA. She is now with the
IBM Thomas J. Watson Research Center, Yorktown Heights, NY 10598 USA
(e-mail: rongxue.huang@gmail.com).
M. B. Diaz and N. Kotulak are with the Department of Electrical and Computer Engineering, University of Delaware, Newark, DE 19716 USA (e-mail:
mdiaz@udel.edu; nkotulak@udel.edu).
R. Hao was with the Department of Electrical and Computer Engineering,
University of Delaware, Newark, DE 19716 USA. She is now with Crystal Solar
Company, Santa Clara, CA 95054 USA (e-mail: ryhao@udel.edu).
R. Opila is with the Department of Materials Science and Engineering, University of Delaware, Newark, DE 19716 USA (e-mail: opila@udel.edu).
A. Barnett was with the Department of Electrical and Computer Engineering,
University of Delaware, Newark, DE 19716 USA. He is now with the School
of Photovoltaic and Renewable Energy Engineering, University of New South
Wales, Sydney 2052, Australia (e-mail: allen.barnett@unsw.edu.au).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/JPHOTOV.2011.2182180
Fig. 1.
44.5%, and 57.3% for two-, three-, and ten-junction solar cells
under AM 1.5G, respectively.
A five-junction solar cell system has been developed for a
separated electrical contact system with lateral spectrum splitting design, which could theoretically achieve over 50% efficiency [7][9]. The lateral spectrum splitting system design
is described in [9]. It provides several benefits, including increased theoretical conversion efficiency, reduced current mismatch losses, reduced lattice mismatch problems, increased flexibility in material choices, and device designs that can more
closely approach the ideal performance limits of existing solar
cell technology. In addition, the multijunction solar cell system
based on a silicon platform has the unique advantages of meeting both efficiency and cost targets. Fig. 1 shows the structure
of this five-junction solar cell system with optics [9]. Light first
passes through the wide band gap solar cell. Then, the dichroic
mirror is used to split the light spectrum.
A first principle calculation described elsewhere [10] shows
that this five-junction solar cell system can achieve the efficiency
of 50% when the top junction solar cell has a band gap between
2.1 and 2.7 eV (the band gaps of the bottom four solar cells
are shown in Fig. 1). This leaves a broad range for material
selection.
In this study, we choose GaP as the material for the top
junction solar cell. The advantages of this include proper band
gap (2.26 eV), high carrier mobility [11] allowing high collection probability, high crystal quality, and well-developed
215
Fig. 2. Measured JV curve of the best GaP solar cell in this study under AM
1.5G one sun illumination (the fourth-generation GaP solar cell).
the base region, and the depletion region. Then, the total IQE of
the solar cell is the sum of the IQEs of the three regions [16]:
IQEtotal () = IQEe () + IQED () + IQEB ()
(1)
where IQEe , IQED , and IQEb , are the IQEs of the emitter,
depletion, and base regions, respectively. The IQE of the emitter
region is as in (2), shown at the bottom of the page, where is the
absorption coefficient of GaP; Le and De are the diffusion length
and the diffusivity of the minority carriers in the emitter region,
respectively; Sf is the front surface recombination velocity; and
We is the thickness of the emitter (without the depletion region).
The IQE of the depletion region is
IQED = exp(We )[1 exp(Wd )]
(3)
(Sf Le /De + Le ) e W e [(Sf Le /De ) cosh We /Le + sinh(We /Le )]
Le
W e
exp
e
2 L2e 1
(Sf Le /De ) sinh(We /Le ) + cosh(We /Le )
(2)
IQEb =
Lb
exp[(We + WD )]
2 L2b 1
Lb (Sr Lb /Db ) [cosh W b/Lb exp(Wb )] + sinh(Wb /Lb ) + Lb exp(Wb )
216
IQE curve fitting gives the diffusion lengths in the emitter and
base region of 0.55 and 0.04 m, respectively. The front surface
recombination velocity is approximately 1 6 cm/s.
The analysis of the JV and IQE curves indicates that the
first-generation GaP solar cell has three problems: high value of
J02 that indicates a high space charge region recombination rate;
a low-diffusion length in the n-type base region; and a high front
surface recombination velocity. Therefore, we developed a new
solar cell structure in which the open circuit voltage of the GaP
solar cell was improved by decreasing the high recombination
rate in the space charge region.
C. Second-Generation GaP Solar Cell
One way to decrease the space charge region recombination
rate is to use the relatively high-quality p-type epitaxial layer as
the lower doped base region. It is known that the width of the
space charge region in each side is determined by the dopant
concentrations. If one side of a pn junction is relatively lower
doped, the space charge region in that side will be relatively
wider. For the first-generation GaP solar cell, the n-type base
region has a much lower diffusion length than the p-type emitter
region, which can be seen from the IQE curve-fitting results of
the first-generation solar cell. Most of the space charge region
is located in the low-quality n-type base region because it was
relatively lower doped. Thus, the recombination rate in that
space charge region was high. However, if we use the higher
quality p-type region as the lower doped base, most of the space
charge region will be located in this high-quality p-type region
so that the recombination in the space charge region will be
greatly decreased. As a result, J02 will decrease, and Vo c of the
solar cell will increase.
Table I lists the calculated Vo c values for different J02 values
assuming that the other parameters remain unchanged. Based
on our simulation, if J02 decreases to 1 1016 A/cm2 , then
Vo c of the solar cell will increase to 1.50 V. However, the further
decrease of J02 will not increase Vo c significantly, because J01
will dominate the performance of the solar cell.
Fig. 4 shows the structure of the second-generation GaP solar
cell with a GaP pn junction grown on a p-type substrate. The
only change from the first generation to the second generation
217
TABLE I
CALCULATED Vo c VALUES FOR DIFFERENT J0 2 VALUES
Fig. 4.
Fig. 5.
218
TABLE II
PARAMETERS OF THE FOURTH-GENERATION OF GAP SOLAR CELLS
Fig. 7. Simulated IQE curves with different parameters. (a) Changed Le and
Sf and (b) changed Lb . The unit for Le and Lb is micrometers and for Sf is
centimeters per second.
219
IV. CONCLUSION
Four generations of GaP solar cells have been developed, fabricated, and analyzed. Their performance has been tested using
two primary methods: JV and QE. A predictive QE model has
been developed to further analyze the solar cells performance
and to improve the solar cell design. Continuous improvements
in design with each generation of solar cell led us to a GaP
solar cell with the best reported efficiency grown by LPE. This
GaP solar cell has the Vo c of 1.55 V, the Jsc of 1.97 mA/cm2 ,
and the FF of 79.4%. Its efficiency is 2.42%. Further increasing
the diffusion length in the emitter and the base and further decreasing the front surface recombination velocity will allow us
to substantially increase the efficiency.
Fig. 9. Simulated IQE curves of GaP solar cells based on the fourth-generation
structures with different parameters. The unit for Le and Lb is micrometers and
for Sf is centimeters per second.
TABLE III
CALCULATED VALUES OF Jsc , J0 , Vo c , FF, AND EFFICIENCY
FOR A FOURTH-GENERATION GAP SOLAR CELL WITH THE PARAMETERS
LISTED IN THE FIRST THREE COLUMNS
ACKNOWLEDGMENT
The authors would like to thank N. Faleev for his help with the
XRD measurements, J. Mutitu for his help in depositing the AR
coating, and J. Allen from the Institute of Energy Conversion
for his help with the QE measurements.
REFERENCES
[1] A. De Vos, Detailed balance limit of the efficiency of tandem solar cells,
J. Phys. D, Appl. Phys., vol. 13, pp. 839846, 1980.
[2] F. Dimroth et al., Metamorphic GaInP/GaInAs/Ge triple-junction solar
cell with >41% efficiency, in 34th IEEE Photovolt. Spec. Conf., Philadelphia, PA, 2009, pp. 19331937.
[3] R. King et al., Band-gap-engineered architectures for high-efficiency
multi-junction concentrator solar cells, in Proc. 24th Eur. Photovolt.
Solar Energy Conf. Exhib., Hamburg, Germany, 2009, pp. 5561.
[4] S. Wojtczuk et al., InGaP/GaAs/InGaAs concentrators using Bi-facial
epi-growth, in 35th IEEE Photovolt. Spec. Conf., Honolulu, HI, 2010,
pp. 12591264.
[5] (2011). [Online]. Available: http://cleantechnica.com/2011/04/19/solarjunction-breaks-concentrated-solar-world-record-with-43-5-efficiency/.
[6] M. Nell and A. Barnett, The spectral pn junction model for tandem
solar cell design, IEEE Trans. Electron Devices, vol. ED-34, no. 2,
pp. 257266, Feb. 1987.
[7] J. Gray et al., Efficiency of multijunction photovoltaics systems, in Proc.
33rd IEEE Photovolt. Spec. Conf., San Diego, CA, 2008, pp. 16.
[8] X. Lu et al., Gallium phosphide solar cells for multijunction systems, in
Proc. 34th IEEE Photovolt. Spec. Conf., Philadelphia, PA, 2009, pp. 1113
1116.
[9] A. Barnett et al., Very high efficiency solar cell modules, Prog. Photovolt.: Res. Appl.., vol. 17, pp. 7583, 2009.
[10] X. Lu et al., A high performance multi-junction solar cells fabricated by
liquid phase epitaxy, in Proc. 24th Eur. Photovolt. Solar Energy Conf.
Exhib., Hamburg, Germany, 2009, pp. 299304.
[11] I. Fritz et al., High temperature electron mobilities in LPE-grown gallium
phosphide, J. Electron. Mater., vol. 14, pp. 7384, 1985.
[12] I. Tan et al., Wafer bonding of 75 mm diameter GaP to AlGaInPGaP
light-emitting diode wafers, J. Electron. Mater., vol. 29, pp. 188194,
2000.
[13] S. Bittner et al., GaP solar cells with InGaP quantum well for high
temperature application, presented at the Proc. 37th IEEE Photovolt.
Spec. Conf., Seattle, WA, 2011.
[14] X. Lu et al., Design, fabrication, characterization and analysis of wide
band gap gallium Phosphide solar cells, in Proc. 25th Eur. Photovolt.
Solar Energy Conf. Exhib., Valencia, Spain, 2010, pp. 212217.
[15] C. Allen et al., Results of a gallium phosphide photovoltaic junction with
an AR coating under concentration of natural sunlight, Solar Energy
Mater. Solar Cells, vol. 95, pp. 26552658, 2011.
[16] W. Yang et al., Internal quantum efficiency for solar cells, Solar Energy,
vol. 82, pp. 106110, 2008.
[17] G. Araujo, E. Sanchez, and M. Martil, Determination of the twoexponential solar cell equation parameters from empirical data, Solar
Cells, vol. 5, p. 199, 1982.
220
Martin B. Diaz was born in Midland, TX. He received the B.S.E.E degree from the University of
Nebraska-Lincoln in 2009. He received the Ph.D. degree in electrical engineering from the University of
Delaware, Newark, in 2011.
His research interest includes high-performance,
next-generation photovoltaic solar cells. In 2009, he
joined the Solar Power Program with the University
of Delaware. His current research is focused on liquid
phase epitaxy growth of IIIV solar cells on silicon.
Ruiying Hao received the M.S. degree in materials science and engineering from Sichuan University,
China, in 2006, where she worked on CdTe solar cells
and received the Ph.D. degree in electrical and computer engineering from the University of Delaware,
Newark, in 2011 for research on single crystalline
thin Si solar cells based on epitaxial absorber.
She is currently with Crystal Solar Company,
Santa Clara, CA, working on Si solar cells. Her research interest includes photovoltaics.
Lu Wang , Yi Wang , Andrew Gerger , Anthony Lochtefeld , Kevin Shreve , Allen Barnett
Solar Power Program, Electrical and Computer Engineering, University of Delaware, Newark, DE 19716, USA
2
AmberWave, Inc, 13 Garabedian Drive, Salem, NH 03079, USA
ABSTRACT
INTRODUCTION
Si:Ge solar cell grown on Si can theoretically contribute
an absolute 5.7% efficiency below Si to a high
performance multi-junction solar cell system at 50 suns
as shown in Figure 1 [1 2]. In this system, the optics
needs to be designed carefully for the back side of
transparent Si solar cell to optimize the long wavelength
light transmission. Photon losses cannot be eliminated
in this way. In this paper a 3-terminal Si-Si:Ge tandem
solar cell is designed and fabricated by adding a Si solar
cell on Si substrate of Si:Ge solar cell. It can replace
stand-alone Si solar cell and Si:Ge solar cell in the
multi-junction solar cell system as shown in Figure 1 [2].
1
978-1-4244-9965-6/11/$26.00 2011 IEEE
000286
Si solar cell
Jsc (mA/cm ) Voc (mV)
FF
Outdoor 1 sun
22.9
598.5
77.3%
Table 1 Summary of a Si solar cell performance in
group one
When testing the Si:Ge solar cells IV curves, no good IV
curve was obtained, all curves behaved like that a Si PN
junction and a Si:Ge PN junction were connected in
series. The back side N type silicon diffused layer was
not etched away completely before P and N type Si:Ge
epitaxial layers were grown on it, so there were two PN
junctions on the back side. A second design was
developed to eliminate this problem based on an
epitaxial silicon emitter.
Figure 3 Schematic vertical cross section of a 3terminal Si-Si:Ge tandem solar cell
The current flow of the 3-terminal solar cells follows. All
generated holes will flow towards the common base,
and be collected by the base Al contact. Electrons
generated by silicon cell will flow towards the upper (Si)
emitter, and electrons generated in the Si:Ge cell will
flow towards the bottom cell (Si:Ge) emitter. The
equivalent circuit of ideal solar cells consists of two PN
junctions sharing common base (Figure 4).
Si solar cell
Jsc (mA/cm )
Voc (mV)
FF
Outdoor 1 sun
17.5
583.3
81.8%
Table 2 Summary of a Si solar cell performance in
group two
2
978-1-4244-9965-6/11/$26.00 2011 IEEE
000287
9.7 suns
79.08
282
44.9%
Table 5 Summary of group three Si:Ge solar cell
performance
2
Si solar cell
Jsc (mA/cm ) Voc (mV)
FF
Outdoor 1 sun
16.0
553.2
78.9%
Table 6 Summary of group three Si solar cell
performance
Figure 5 Actual pictures of Si solar cell (left) and
Si:Ge solar cell (right)
Ge
Jsc
Voc
FF
Eff
2
(%)
(mA/cm ) (mV) (%)
(%)
1 sun
88
3.41
263
68
0.61
102.3
30 suns
88
350
66
0.79
Table 4 Summery of the best Si:Ge solar cell
performance [1]
ACKNOWLEDGEMENT
This research was, in part, funded by the U.S.
Government Defense Advanced Research Projects
Agency under
Agreement
No.:
HR0011-07-90005. The views, opinions, and/or findings contained
in this article/presentation are those of the authors and
should not be interpreted as representing the official
views or policies, either expressed or implied, of the
Defense Advanced Research Projects Agency or the
Department of Defense.
The author would also like to thank China Scholarship
Council for providing stipends during his Ph.D.
education.
Jsc (mA/cm )
8.16
3.26
Voc (mV)
215
189
REFERENCES
[1] Yi Wang, et al. Design, Fabrication, Characterization
and Improvement of Si:Ge Solar Cell below Si Solar
Cell In A Multi-junction Solar Cell System, this
conference.
[2] Allen Barnett, et.al. Very High Efficiency Solar Cell
Modules, Prog. Photovolt: Res. Appl. 17 (2009) 7583.
FF
56.7%
No data
3
978-1-4244-9965-6/11/$26.00 2011 IEEE
000288
4
978-1-4244-9965-6/11/$26.00 2011 IEEE
000289
INTRODUCTION
ABSTRACT
Monolithic multi-junction solar cells are becoming prevalent in concentrator photovoltaic (CPV) systems due to
their high demonstrated conversion efficiencies. However, these devices often operate at sub-optimal levels
due to current mismatch losses arising between each
p-n junction as a result of natural variation in the solar
spectrum at the earths surface. The use of quantum
wells in solar cell design affords improved control over the
spectral response of the cells via band-gap engineering.
This makes it possible to tailor the spectral response of
a cell for optimum performance under a given annual
spectral resource. Establishing the optimum spectral
response for a given location is a major challenge of this
approach to cell design, and relies heavily on averaged
and modelled data based on combinations of satellite and
sparse ground measurements. In addition, there is only
limited field experience of such technology to date.
The purpose of this work is to investigate through
experiment the effect of incorporating multi-quantumwell (MQW) structures into photovoltaic cells to respond
to a specific range of annual irradiation spectra and
to compare the results obtained with those predicted
through modelling. A number of triple-junction cells of
different design have been placed side-by-side on an
accurate solar tracker, and current-voltage characteristics
of each taken at regular intervals over a few months.
The outputs are then compared to those predicted by
a model of cell performance that includes simulated
spectra, generated with the SMARTS program, specific
to that location and period of time.
Results obtained from outdoor testing indicate that the
cells designed to provide improved current matching
under the spectral conditions in which they have
been tested have performed more consistently than
conventional cell designs. This is evident when the
short-circuit current is normalised to a fixed direct
normal irradiation and temperature, and plotted against
atmospheric depth. Detailed analysis of the spectral
resource is expected to reveal additional information on
the performance of the cells.
High-efficiency solar cells are finding promising terrestrial applications within concentrator photovoltaic (CPV)
systems that offer to deliver higher overall conversion
efficiencies than conventional flat panel photovoltaic
installations. Most high-efficiency cells are designed as
monolithic devices with two or three junctions in order
to exploit a broad range of the solar spectrum. Whilst
the solar spectrum is constant outside of the atmosphere,
terrestrial deployment of these cells can be complicated
by the varying content of the solar spectrum due to
atmospheric attenuation processes. The spectral content
of sunlight over a year varies with location on earth, with
the result that a similar cell will not necessarily deliver the
same energy yield from place to place, even if the total
annual irradiation levels are similar.
Thanks to advances in nanotechnology, quantum well
solar cells can now be designed and fabricated to
deliver optimum performance under the specific spectral
resource of a given location on earth. This is done
by careful design of materials with band-gaps suited
to the particular spectra available. Compared with
a conventional multi-junction cell design, a tuned cell
should then exhibit a more stable power output over a
range of atmospheric conditions and depths.
Cells have now been designed with a variety of different
spectral responses [1, 2]. The designs have been based
on the results of analyses of the spectral resource at
a number of locations. These analyses make use of
a combination of real and modelled data based on
satellite and ground-based measurements. Whilst this
approach offers convenience and avoids the cost of
long-term site measurements, it is also likely to introduce
a wide margin of error. Consequently, it is desirable to
obtain real performance measurements of a range of cell
designs under varying environmental conditions to both
determine the relative advantages of tuning cells, and to
correlate the results with the models used to design the
cells.
000534
APPROACH
1000
Experimental apparatus
900
800
Data analysis
To more clearly examine the relationship between
atmospheric depth and cell output, the data are filtered
to remove days of poor atmospheric clarity. This is done
by using the DNI as an indicator of clearness, k , based
on a method described by Merten and Andreu [3, 4]. In
this approach, DNI is plotted against atmospheric depth,
and then a curve is fitted to describe the upper limit
of the recorded DNI at each atmospheric depth, as in
Figure 1. In effect, the fitted curve describes a very clear
atmosphere (k approaches 1). The clearness value of
each data point can then be calculated as a function of the
DNI, I , the top-of-atmosphere DNI, I0 , and atmospheric
depth, M , as in equations 1 and 2 below.
I(M ) =
I
I(M )
(1)
I0 e0.2M
(2)
Modelling
Hourly solar spectra for the test location in Nicosia,
Cyprus were calculated using the SMARTS model. Inputs
to the SMARTS model included Aerosol Optical Depth
data taken from the CUT-TEPAK NASA AErosol RObotic
NETwork site in Southern Cyprus, and precipitable
water calculated from on-site measurements of ambient
Data points
Fitted curve
DNI [Wm2 ]
700
600
500
400
300
200
100
0
10
15
20
25
30
35
000535
1.1
Relative Isc
0.9
0.8
0.7
1.5
2.5
3
3.5
4
Atmospheric Depth [AM]
4.5
2.5
3
3.5
4
Atmospheric Depth [AM]
4.5
1
0.9
1.1
0.8
0.7
1.5
2
2.5
3
3.5
4
Atmospheric Depth [AM]
4.5
Relative Isc
1.1
1
0.9
0.8
0.7
1.5
Figure 3 Scatter plots of side-by-side current measurements of two different cell designs, filtered to
show clear sky behaviour. (Above: a conventional
cell output; below: a MQW cell output)
fill factor at higher atmospheric depths, consistent with a
current mismatch between junctions.
An examination of the power output of the two cells has
so far not been able to indicate a clear difference in
performance characteristics of the two designs. This is
because whilst the Isc of the conventional cell dropped
with increased atmospheric depth, the corresponding
increase in fill factor maintained the power output of the
cell for the range of spectral conditions tested so far.
CONCLUSIONS
The depth of atmosphere has been shown to be a
useful approximate indicator of spectral content when
measured data is filtered to reveal clear-sky conditions.
The particular analyses conducted show that a k value
calculated as described from the atmospheric depth and
DNI alone is a convenient parameter to use in filtering
data for clear sky conditions.
The results obtained so far indicate that there is a
more consistent output of the tuned quantum well solar
cells over a range of atmospheric depths. Since the
atmospheric depth directly affects the spectrum of the
light incident on the cells, this can be interpreted as an
indication that the cells are better suited to the range of
spectra that they have been exposed to. Further work
000536
1
0.95
0.85
0.8
0.75
0.7
1.5
0.9
2.5
3
3.5
4
Atmospheric Depth [AM]
4.5
1
0.95
0.9
0.85
0.8
0.75
0.7
1.5
2.5
3
3.5
4
Atmospheric Depth [AM]
4.5
000537
Study on the effect of reverse current blocking layer for concentrator solar cells
1
Myoung-Gyun Suh* , Ji-Eun Chang , Dong-Ho Kim , Kyu-Sang Kim , and Yun-Gi Kim
Energy Lab, Samsung Advanced Institute of Technology, Yongin-si, Gyeonggi-do, Republic of Korea
2
R&D CAE Group, Samsung LED, Suwon-si, Republic of Korea
ABSTRACT
In high concentration photovoltaic (HCPV) systems, III-V
multijunction solar cells work under high concentration
2
condition at a small size, normally less than 1cm . In this
case, the metal coverage ratio and the current density
become larger than those in the case of flat-plate solar
cells. Due to the high current density, combined with the
sheet resistance of solar cells, most of currents reverse to
photocurrents flow vertically under front metal grids. This
non-uniformity is critical when the metal coverage ratio is
large as the case of concentrator solar cells.
In this paper, the effects of metal coverage ratio and nonuniform reverse current density (i.e. forward bias current
density) on the performance of InGaP/InGaAs/Ge triplejunction solar cells under high concentration were
analyzed in detail. Moreover, by introducing reverse
current blocking layer between metal busbar and AlInP
window layer, we have successfully reduced the unwanted
reverse current and achieved significant improvement in
cell efficiency under high concentration condition.
Introduction
(a)
(b)
000556
Experimental
0.7
0.6
0.010
0.4
0.008
Current (A)
Current (A)
0.5
0.3
0.006
0.004
0.002
0.2
0.000
0.0
0.5
0.1
0.0
0.0
1.0
1.5
2.0
2.5
3.0
Voltage (V)
0.5
1.0
1.5
2.0
2.5
3.0
Voltage (V)
(b)
0
2.5
2.6
2.7
2.8
2.9
3.0
Voltage (V)
Figure 2. Schematic device structure of solar cells w/
and w/o reverse current blocking layer (RCBL)
000557
w/ RCBL
0.067
28.45 %
30.08 %
30.90 %
29.43 %
26.95 %
w/o RCBL
0.229
28.30 %
29.54 %
30.35 %
27.73 %
26.21 %
(a)
15
10
w/ RCBL
w/o RCBL
Voc (V)
2
Jsc (mA/cm )
FF (%)
Efficiency (%)
0
0.0
0.5
w/ RCBL
2.383
13.73
86.95
28.45
1.0
w/o RCBL
2.379
13.70
86.83
28.30
1.5
2.0
2.5
Voltage (V)
(b)
2.0
1.8
Current (A)
1.6
454 suns
1.4
1.2
1.0
0.6
0.2
0.0
0.0
Conclusion
299 suns
0.8
0.4
75 suns
38 suns
0.5
1.0
1.5
2.0
2.5
3.0
3.5
Voltage (V)
000558
ACKNOWLEDGEMENT
The authors gratefully acknowledge the technical support
by Jin-Soo Moon and Tak-Kyun Ro in the module
processing. The authors also thank to previous team
members, Joo-Sung Kim, Sang-Moon Lee, and Taek Kim.
REFERENCES
[1] R.R. King, et al., Band-Gap-Engineered Architectures
for High-Efficiency Multijunction Solar Cells, Proceedings
th
24 European Photovoltaic Solar Energy Conference and
st
th
Exhibition, Hamburg, Germany, 21 -25 September 2009.
[2] J.Geisz, D.Friedman, J.Ward, et al., 40.8% efficient
inverted triple-junction solar cell with two independently
metamorphic junctions, Appl. Phys. Lett. 93 (12),
p.123505, 2008
[3] W.Guter, J.Schne, S.Philipps, M.Steiner, G.Siefer,
A.Wekkeli, E.Welser, E.Oliva, A.Bett, and F.Dimroth,
Current-matched triple-junction solar cell reaching 41.1%
conversion efficiency under concentrated sunlight, Appl.
Phys. Lett. 94, p.223504, 2009
[4] K.Nishioka, T.Takamoto, et al., Evaluation of
InGaP/InGaAs/Ge
triple-junction
solar
cell
and
optimization of solar cells structure focusing on series
resistance for high-efficiency concentrator photovoltaic
systems, Solar Energy Materials and Solar Cells, 90 (9),
pp. 1308-1321.
[5] Omer Korech, Baruch Hirsch, Eugene A. Katz, and
Jeffrey M. Gordon, High-flux characterization of ultrasmall
multijunction concentrator solar cells, Appl. Phys. Lett. 91,
p.064101, 2007
[6] G.Duggan and I.M.Ballard, Evaluation of competing
triple junction concentrator cells by electroluminescence,
th
34 IEEE PVSC, pp. 655-659, 2009
[7] B.Galiana, C. Algora, and I. Rey-Stolle, Explanation
For the Dark I-V Curve of III-V Concentrator Solar Cells,
Prog. Photovolt: Res. Appl. 16, pp. 331-338, 2008
000559
C. Paola Murcia , Ruiying Hao , Christopher Leitz , Anthony Lochtefeld , Allen Barnett
Electrical and Computer Engineering, University of Delaware, Newark, DE, United States
2
AmberWave Inc., Salem, NH, United States
ABSTRACT
Thin-film crystalline silicon (c-Si) solar cells have the
potential for very high efficiency through high open circuit
voltage (Voc) [1]. The best performance reported in thinfilm crystalline silicon solar cells with absorber thickness
below 20 microns is 16.9% efficiency for a solar cell grown
epitaxially on a crystalline silicon conductive substrate [2].
The efficiency potential of thin c-Si solar cells is above
20%. In previously reported thin c-Si solar cells, layers of
porous silicon have been included as a Bragg reflector in
the light trapping design or as a separation layer in the
mechanical design [3, 4]. The Bragg reflector uniformity
and reliability has been analyzed [5]. The challenges of
this Bragg reflector are uniformity and the maximum
achievable optical gain.
This work presents a systematic approach to the design,
fabrication, testing and analysis of thin c-Si solar cells.
Solar cell designs presented include a thin c-Si solar cell
with no light trapping structures and a thin c-Si solar cell
with an optical design comprised of a metallic back
reflector on the back surface and a chemical texture on
the front surface. These baseline designs enable us to
separately measure and localize voltage and current
losses. The metallic back reflector has low electrical
losses, good mechanical stability, and uniform optical
properties. To optimize the baseline designs we measure
open circuit voltage (Voc) and external quantum efficiency
(EQE). Voc shows high material quality in the absorber
layer. EQE shows an evidence of light trapping with a
current gain of 3.6%.
BACKGROUND
Despite the indirect band gap that leads to lower
absorption, silicon is the most used material for
commercial solar cells and modules. From material
properties, silicon has a low absorption for most of the
energy available in the solar spectrum. The relatively low
band gap makes silicon unsuitable for very high efficiency
solar cells especially when used in a single junction
structure.
Part of this is due to the limitations of
conventional thick solar cell designs. The very low energy
conversion in the infrared requires either optimal light
trapping structures or a material thickness of a few
hundred microns that effectively collects the generated
power.
The high performance potential of thin crystalline silicon
solar cells is based on the smaller active volume that can
000661
000662
Jsc
FF
Eff.
mV
mA/cm2
S19
631.4
21.49
75.01
10.18
14.05
S27
620.5
22.04
81.32
11.12
15.35
S27
618.8
20.50
81.76
10.37
14.31
Cell
Eff. w/AR
%
Figure 5. EQE measurement of a baseline solar cell, and
a baseline solar cell with a metallic back reflector.
CONCLUSIONS
000663
[2]
Nieuwenhuysen, K.V., et al. Epitaxial thin film
silicon solar cells with efficiencies up to 16.9% by
combining advanced light trapping methods and CVD
emitters. in 24th European Photovoltaic Solar Energy
Conference. 2009. Hamburg, Germany.
[3]
Duerinckx, F., et al., Simulation and
implementation of porous silicon reflector for epitaxial
silicon solar cells. Progress in Photovoltaics: Research
and Applications, 2008. 16: p. 399-407.
[4]
Haase, F., et al. Back contact monocrystalline
thin-film silicon solar cells from the porous silicon process.
in 34th IEEE Photovoltaic Specialists Conference. 2009.
[5]
Hoeymissen, J.V., et al., Thin-film epitaxial solar
cells on low cost Si substrates: closing the efficiency gap
with bulk Si cells using advanced photonic structures and
emitters, in 23rd European Photovoltaic Solar Energy
Conference2008: Valencia, Spain. p. 2037-2041.
[6]
Schmich, E., et al., n-Type emitter epitaxy for
crystalline silicon thin-film solar cells. Progress in
Photovoltaics: Research and Applications, 2008. 16: p.
159-170.
[7]
Tsujino, K. and M. Matsumura, Morphology of
nanoholes formed in silicon by wet etching in solutions
containing HF and H2O2 at different concentrations using
silver nanoparticles as catalysts. Electrochemica Acta,
2007. 53: p. 28-34.
[8]
Reuter, M., et al., 50um thin solar cells with
17.0% efficiency. Solar Energy Materials and Solar Cells,
2009. 93: p. 704-706.
000664
T. Potlog , N. Spalatu , V. Fedorov , N. Maticiuc , C. Antoniuc , V. Botnariuc and J. Hiie , T. Raadik , V. Valdna
1
Physics Department, Moldova State University, A. Mateevici str. 60, MD 2009, Chisinau, Moldova
2
Department of Materials Science, Tallinn University of Technology, Ehitajate tee 5, 19086, Tallinn, Estonia
ABSTRACT
This paper focuses on the photovoltaic parameters of
ZnSe/CdTe, CdS/CdTe, and ZnTe/CdTe thin film
heterojunction solar cells. ZnSe/CdTe, CdS/CdTe, and
ZnTe/CdTe thin film heterojunction solar cells were
fabricated by Close Space Sublimation (CSS) on TCOcoated glass substrates. All types of solar cells were
fabricated in a superstrate configuration. The thickness of
ZnSe and ZnTe layers was varied in order to adjust the
solar cell performance. A similar cadmium chloride
solution for the treatment of a CdTe layer with an elevated
temperature air annealing of the completed devices before
the back contact deposition was applied to ZnSe/CdTe
and CdS/CdTe thin film heterojunctions solar cells with
exception of ZnTe/CdTe. All cells were characterized
through light and dark current density-voltage (J-V)
measurements
and
quantum
efficiency
(QE)
measurements. The saturation current, ideality factor and
photovoltaic parameters for all thin film heterojunction
solar cells are presented. The investigation at the room
2
temperature under illumination of 100 mW/cm through the
wide gap components of ZnSe/CdTe, CdS/CdTe, and
ZnTe/CdTe heterojunctions showed a value of conversion
efficiency () of solar energy to electric energy about
4.7%, 9.9%, and 1.3%, respectively. The incorporation of
Zn at the ZnSe and CdTe interface doubles the short
circuit current density and improves the performance of
ZnSe/CdTe thin film heterojunction solar cells.
INTRODUCTION
The primary aim of this research paper is to develop low
cost photovoltaic systems. Routes to decrease cost per
Watt are either to reduce the cost of the technology or to
increase its efficiency. The first of these is addressed by
the use of thin film technology for low cost deposition at
low temperatures and with low material usage. The
second is addressed by using multiple threshold devices
to push efficiencies beyond the Schockley-Queisser limit
for single band-gap PV cells [1]. Our studies will focus on
the development of new electronic devices for PV
applications. Materials of interest for PV systems include
II-VI compound thin film semiconductors, especially CdTe
and transparent conducting oxides required for the
fabrication of solar cells. The main goals of this paper are
to develop polycrystalline growth techniques for
ZnSe/CdTe and ZnTe/CdTe thin film heterojunction solar
cells on glass/TCO superstrate configurations and
compare them with 10% efficiency CdS/CdTe solar cells
001365
2.5x10
ZnSe- 4 min
ZnSe- 8 min
ZnSe-11 min
14
2.0x10
( h )2, (eV/cm)2
14
1.5x10
14
1.0x10
13
5.0x10
0.0
2.5
100
2.6
2.7
2.8
2.9
3.0
h , eV
80
T, %
60
40
glass/ZnSe_ 1
glass/ZnSe_ 2
glass/ZnSe_ 3
glass/ZnTe_ 1
glass/ZnTe_ 2
glass/ZnTe_ 3
20
0
300
600
900
1200
, nm
the
d=
12
2(1n2 2 n1 )
(1)
ZnSe
deposition time
d, m
Eg, eV
4 min
1.4
2.684
8 min
2.7
2.675
3
ZnTe
1
2
3
11 min
deposition time
4 min
8 min
11 min
4.1
d, m
1.8
3.9
5.2
2. 670
Eg, eV
2.247
2.245
2.242
001366
eU ,
I = I O exp
nkT
( h )2, (eV/cm)2
4x10
Glass/ZnTe-1
Glass/ZnTe-2
Glass/ZnTe-3
3x10
(3)
14
2x10
12
1x10
C1_n1
C1_n2
Z3_n2
Z3_n1
ZT2_n
8
2.25
2.30
2.35
2.40
h, eV
0
2.20
10
6
2
4
2
300
320
340
360
380
400
T,K
001367
-1.0
-0.8
-0.6
-0.4
-0.2
0
0.0
0.2
0.4
0.6
0.8
1.0
U, V
-5
-10
Z3
Z2
Z1
-15
-20
J, mA/cm
1.1
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
-0.1
-0.2
Z3
Z1
Z2
1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0 3.2
h (eV)
J, mA/cm
ZT1
ZT2
ZT3
-1.0
6
4
2
-0.5
0
0.0
U, V
0.5
1.0
-2
-4
-6
-8
001368
Table 2 The photovoltaic and electric parameters of the ZnSe/CdTe, CdS/CdTe, and ZnTe/CdTe thin film
heterojunction solar cells at room temperature
Sample
Jsc,
2
mA/cm
Ucd, V
FF, %
, %
Rs, cm
Z3 with Zn at the
interface
Z2
Z1
ZT1
ZT2
ZT3
C1
C2
C3
18.48
0.71
35.67
4.68
10.70
4.58
2.90
4.94
3.37
23.27
17.30
22.96
0.75
0.72
0.63
0.69
0.43
0.79
0.78
0.70
41.36
30.80
27.39
37.04
28.90
54.00
55.10
31.60
3.33
1.01
0.50
1.27
0.42
9.90
7.40
5.20
Rsh,
2
cm
Jo,
2
mA/cm
21.00
128.13
1.28E-2
2.89
34.70
110.37
234.52
53.67
90.91
11.66
10.67
86.42
187.83
218.75
321.64
332.15
188.92
1054.54
487.80
101.38
1.60E-5
1.11E-5
1.30E-4
1.86E-6
2.78E-4
1.07E-4
5.86E-3
3.15E-2
2.46
2.81
5.69
2.94
8.62
12.93
7.73
10.90
J, mA/cm
C1
C2
C3
1.0
ZT1
0.8
-1.0
-0.5
20
10
0
0.0
ZT2
0.5
1.0
U, V
ZT3
-10
0.6
-20
0.4
-30
0.2
0.0
1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0 3.2
h (eV)
q (U + JRs
J = J f J 0 [exp
nkT
(U + JRs ) ,
1]
Rsh
(6)
001369
1.0
0.8
0.6
C3
C2
C1
0.4
0.2
0.0
1.0
1.5
2.0
2.5
h, eV
REFERENCES
[1] F. Meillaud, A. Shah, C. Droz, E. Vallat-Sauvain and C.
Miazza, Efficiency limits for single-junction and tandem
solar cells Solar Energy Materials and Solar Cells. V. 90,
Iss. 18-19, 23, 2006, pp. 2952-2959.
[2] T. Potlog G. Khrypunov. M. Kaelin, H. Zogg, and A. N.
Tiwari. Thin-Film Compound Semiconductor, Book, V.
1012, Photovoltaics-2007, edited by T. Gessert, K.
Durose, C. Heske, S. Marsillac, and . Wada, 2007, pp.
181-188. www.mrs.org/meetings/.
[3] X. Wu,High-efficiency polycrystalline CdTe thin-film
solar cells, Solar Energy 77, 2004, pp. 803-814.
[4] J. Gutowski, N. Presser, G. Kudlek Optical Properties
of ZnSe Epilayers and Films, Physica Status Solidi (a)
V. 120, N.1, , 1990, pp. 1159.
[5] [7] Huanyong Li, Wanqi Jie Growth and
characterizations of bulk ZnSe single crystal by chemical
vapor transport, Journal of Crystal Growth, 257, 2003, pp.
110115.
[6] Swanepoel R. Determination of the thickness and
optical constants of amorphous silicon, J. Phys. E: Sci.
Instrum. 16, 1983, pp. 1214-1219.
[7] M. Orita, T. Narushima, and H. Yanagita Transparent
Conductive Cu-doped ZnSe Film Deposited at Room
Temperature Using Compound Sources Followed by
Laser Annealing, Jpn. J. Appl. Phys. 46, 2007, pp. 976978.
[8] D. T. F. Marple Refractive Index of ZnSe, ZnTe, and
CdTe, J. of Appl. Physics, V.35 , N.3,1964, pp.539-542.
[9] Donnelly J.P., Milnes A.G The photovoltaic response
of nGe-nSi heterodiodes, Sol. St. Electronics, V.9, 1966,
pp.174-178.
[10] H. Bayhan and A. S. KavasoluTunnelling enhanced
recombination
in
polycrystalline
CdS/CdTe
and
CdS/Cu(In,Ga)Se2 heterojunction solar cells, Solid-State
Electronics, V. 49, N. 6, 2005, pp. 991-996.
[11] S. S. Ou and O. M. Stafsudd, B. M. Basol Current
transport mechanisms of electrochemically deposited
CdS/CdTe
heterojunction,
Solid-State
Electronics
V. 27, Nr. 1,1984, pp. 21-25.
[12] C. Ercelebi, A.W. Brinkman, T.S. Furlong and J.
Woods Current transport mechanisms in epitaxial
CdS/CdTe heterojunctions, J. of Crystal Growth
V. 101, Nr. 1-4, 1990, pp. 162-166.
ACKNOWLEDGEMENTS
th
001370
How to further enhance the efficiencies of the silicon solar cells which can be applied
into commercial production for the next generation is a most important issue. On the way to
achieving higher efficiencies, back contact solar cells and heterojunction solar cells seem to
be very promising. Therefore, we proposed a amorphous/crystal hetero-junction with metalwrap structure (HJ-MWT) which combines the advantages of different high-efficiency
concepts such as back contacts, and great heterojunction passivation. The advantages of
MWT are less shadowing loss and lower module electrical power losses. It will increase
1.5-3 % module conversion efficiencies compared to conventional silicon solar cells.
Furthermore, the other advantage of MWT isnt fully limited by the bulk lifetime and carrier
diffusion lengths contrast to IBC solar cells. It doesnt need very-high-quality silicon
wafer(minority lifetime> 1ms).
We proposed amorphous silicon or insulation materials were designed as electrical
insulation layers within the via(wrap-through holes) to avoid shunt current occurring. The
fabricating process flow that realizes HJ-MWT solar cells, with only 2-3 additional
processes steps compared to conventional HIT solar cells. Its suitable to the mass
production associated the cost consideration in the future. Before experiments, we used the
software tool to do the two-dimension simulation with HJ-MWT structures. From theoretical
calculation, we found HJ-MWT solar cell efficiencies increased up to 0.6% compared with
conventional HIT solar cells. We have also done the further experimental investigation on
new proposed HJ-MWT structures and corresponding data will be published in this
conference. It shows the HJ-MWT solar cell has the potential to further increase the
efficiencies of solar cell and solar panels in the future.
ITO
i,p a-Si
TCOa-Si(i,p)
n-Si
a-Si(i n)
Ag
n-Si
n-Si
Ag
n-Si
Ag
a-Si(i,p)
Ag
Ag
Insulation
Ag
001515
001516
TCOa-Si(i,p)
n-Si
a-Si(i n)
n-Si
Ag
Ag
Ag
(a) a-Si(i,p)
ITO
i,p a-Si
n-Si
n-Si
Ag
Ag
Insulation
Ag
(b)
001517
Fig. 4: Simulation results chart of HJ-MWT. For simplifying the simulation, no surface texturing structure was used in this
study. We used the software tool to do the two-dimension simulation with HJ-MWT structures. Two proposed devices are
theoretically verified and no shunt current occurs from the simulation results. The summarized efficiency table showed HJMWT solar cell efficiencies increased up to 0.6% compared with conventional HIT solar cells. It shows the HJ-MWT solar
cell has the potential to further increase the efficiencies of solar cell in the future. We have tried to implemented the
experimental results and verified the utilities of HJ-MWT, and the data will be disclosed in PVSC conference.
[1] RM. Swanson, Proceedings of the 31th IEEE PVSC,
Lake Buena Vista, 889 (2005).
th
[2] M. Taguchi et al., Proceedings of the 24 EU-PVSEC,
Hamburg, 2CV.2.78 (2009).
[3]S. D. Wolf et al. Appl. Phys. 93, 032101 (2008).
[4]T. F. Schulze et al. 25th EU-PVSEC, Hamburg, (2010).
[5]L. Zhao et al., Phys. Stat. sol. (a) 205, 1215 (2008).
[6]D. C. Wu et al., 35th IEEE PVSC, Honolulu USA, (2010).
th
[7] D. Munoz et al., 35 IEEE PVSC, Honolulu USA,
(2010).
th
[8] D. Lachenal et al. 25 EU-PVSEC, Hamburg, (2010).
[9]R. Street, Hydrogenated Amorphous Silicon, Cambridge.
University, Cambridge, (1991).
[10]M. Schmidt et al. Thin Solid Film, 515, 7475 (2007).
[11]F. Clement et al. Sol. Energy Mater. Sol. Cells, 93,
1051 (2009).
001518
ABSTRACT
We have proposed the first demonstration of hybrid
thermodynamic approach between multiple junction solar
cells and the intermediate band solar cells. Because of
increasing number of transitions inside each multiple
junction from intermediate band, reduced number of
stacks in this hybrid solar cell can be possible to increase
efficiency compared to more than four stacks of
conventional multijunction solar cells. The maximum
efficiency of two stacks hybrid solar cells has been
enhanced and its value is larger than over 4 stacks of
multijunction solar cells. We have also found optimum
bandgap of each stack and studied optimum transition of
each stack both from valence band to intermediate band
and intermediate band to conduction band.
.
INTRODUCTION
Since, Shockley and Queisser (SQ) have been developed
detailed balance model for single junction solar cell and
calculated its maximum conversion efficiency blackbody
radiation [1]. But, 69% is still the loss of energy conversion
And, its dominant factor is the thermalization loss while
excited electrons have been released on the conduction
band (CB) edge. One way for of reducing loss is splitting
the broad solar spectrum into different energy range and
converting each range with a cell of a well matched
bandgap [2,3]. The other way for enhancing performance
is added another absorption path like isolated energy level
to interact between valence band and conduction band [4].
It is the intermediate band solar cell (IBSC) and its benefit
is that its characteristics shows as tandem solar cells. And,
its maximum efficiency from detailed balance has been
enhanced [4]. If IBSC can be included into each stack of
multijunction solar cell, the performance of this hybrid
multijunction solar cell will be improved because of
increasing multi-transitions from IB effect. For instance, 2
stacks of tandem solar cell would be behaved like more
than 4 stacks of tandem cells. Because of increasing
number of transitions, the reduced number of stacks for
hybrid solar cell can be possible as well as its
performance will be enhanced. For this detailed balance
calculations, we have been devised a new model for this
hybrid system and been investigate the optimum IB level
of each stacks for high conversion efficiency.
THEORY
002082
N(E1,E2 ,T, )
2
h 3c 2
E2
e
E2
(E )/kT
E1
1
dE
(1)
(2)
(3)
(4)
(5)
where Eg1,low and Eg1,high are start and end photon energy
in the first stack, Eg2,low and Eg2,high are beginning and
nd
ending photon energy point at 2 stack. Eg1,opt and Eg2,opt
are the optimum energy points
(a) Spectrum splitting of multijunction solar cells
002083
70
One Sun
Maximum Concentration
60
Efficiency (%)
50
40
30
20
10
0
0.5
1.5
2.5
3
Eg (eV)
3.5
4.5
Figure 3 Spectrum
positions.
splitting
and
its
bandgaps
RESULTS
In this section, we introduce the results of 2 stacks of
independently connected IBT solar cells. To analyze the
results, we have also compared the efficiencies of
conventional independent connection of multijunction (or
tandem) solar cells and intermediate band solar cells. To
evaluate the performance of IBT, we have considered the
position of IB to maximize the efficiencies. For this
consideration, we have changed IB positions to calculate
local maximum power point of each bandgaps (Eg or ECV).
ECI (eV)
EIV (eV)
ECV (eV)
Efficiency
(%)
One
0.92
1.49
2.41
46.89
Max
0.72
1.25
1.97
63.17
Table 2 Optimum bandgap positions and its maximum
efficiencies under one sun and maximum
concentrated sunlight.
To compare the results for IBT solar cell, we have
summarized the optimum bandgaps and its maximum
efficiencies up to 6 stacks of independent connection of
tandem solar cells in Figure 3 and Table 3 and 4 [2].
N
2
Con
Eg1
Eg2
Eg3
Eg4
Eg5
Eg6
One
0.98
1.87
Max
0.77
1.70
3
One
0.82
1.44
2.26
Max
0.62
1.26
2.10
4
One
0.72
1.21
1.77
2.55
Max
0.52
1.03
1.61
2.41
5
One
0.66
1.07
1.50
2.03
2.79
Max
0.45
0.88
1.34
1.88
2.66
6
One
0.61
0.96
1.33
1.74
2.26
3.00
Max
0.40
0.78
1.17
1.60
2.12
2.87
where N is a number of stack, Con is concentration of
sunlight, One is one sun and Max is maximum
concentration. unit of Egi (i=1,2,3,4,5,6,) is eV [2,3].
Table 3 Optimum bandgaps of each stacks for
independent connection of conventional tandem solar
cells.
002084
N
2
Con
Efficiency (%)
One
42.9
Max
55.9
3
One
49.3
Max
63.8
4
One
53.3
Max
68.8
5
One
56.0
Max
72.0
6
One
58.0
Max
74.4
Table 4 Maximum efficiency for independent
connection of conventional tandem solar cells.
75
One Sun
C=1000
C=10000
Maximum Concentration
70
Efficiency (%)
65
60
55
50
45
0.5
1.5
Eg1 (eV)
2.5
EIV2
Eg2
2.13
3.7
1.89
3.22
1.79
3.02
1.79
3.02
and Max is
002085
002086
meas
bulk
diff =
diff
1
+ surf
(1)
d2
2 .D n , p
(2)
d
2.S
(3)
surf =
002940
17.2
1.2
1.05
16.9
1
629
36
35
627
34
33
32
Voc (mV)
17
1.1
37
Jsc (mA/cm^2)
1.15
631
38
17.1
Efficiency (%)
1.25
625
31
0.95
16.8
Sample 1
Sample 2
Sample 3
30
623
1.136
Sample
1.06
1.206
002941
002942
122
I. INTRODUCTION
OLAR power has become an attractive alternative of electricity energy due to growing environmental concerns and
global oil shortage. Solar cells that convert the photons from
the sun to electricity are mostly based on crystalline silicon in
the current market because it can generate good performance in
usable lifespan and conversion efficiency among the currently
available techniques. Multicrystalline solar cells are more popular than monocrystalline solar cells. They dominate the production volume in the photovoltaic industry due to lower manufacturing costs.
The surface of a multicrystalline solar wafer shows multiple
crystal grains of random shapes and sizes in random positions
Manuscript received September 02, 2011; revised October 12, 2011 and February 06, 2012; accepted May 19, 2012. Date of publication July 20, 2012; date
of current version December 19, 2012. Paper no. TII-11-504.
The authors are with Yuan-Ze University, Chungli 320, Tao-Yuan, Taiwan
(e-mail: iedmtsai@saturn.yzu.edu.tw)
Digital Object Identifier 10.1109/TII.2012.2209663
Fig. 1. CCD-captured images of (a) multicrystalline solar wafer; (b) solar cell;
(c) solar module composed of 36 solar cells.
and orientations. It results in a heterogeneous texture in the surface. No two solar wafer surfaces contain completely identical
patterns. A solar wafer is a thin slice of a silicon ingot. It is
further processed and fabricated into a solar cell that forms the
basic unit of a solar power system. A set of solar cells are assembled and interconnected into a solar module to offer a large
amount of electricity power for a variety of commercial applications.
Fig. 1(a) shows the image of a defect-free multicrystalline
solar wafer. It contains random crystal grains on the surface.
Fig. 1(b) is a solar cell image taken by a typical charge-coupleddevice (CCD) camera, where the vertical thin metal strips are
finger electrodes that supply current to the two horizontal thick
bus bars. The bus bars are used for interconnecting solar cells
to make a solar module. Fig. 1(c) demonstrates a CCD-captured
solar module image. It is composed of a matrix of 36 (6 6)
solar cells.
Image processing techniques have become an important tool
for biometric recognition [1], [2], robotic automation [3], [4]
and industrial inspection [5][7]. The quality inspection of
a solar module in manufacturing is very important to ensure
the expected conversion efficiency and durable lifespan. A
few automatic defect detection algorithms in the literature
have been focused on the detection of defects in solar wafers
and solar cells. The inspection in the early manufacturing
stages can prevent delivering defective materials to the subsequent process. However, some of critical defects including
micro-cracks, breaks and finger interruptions may inevitably
occur in solar modules during the transportation, handling and
assembly. Fuyuki and Kitiyanan [8] have also pointed out that
the point-pressure cracks due to the failure in the metal wire
soldering could occur in the solar module assembly process.
The defects of a solar module may reduce the current or even
consume the power stored in the battery bank. They may
also cause thermal effect that deteriorates the materials and
eventually shorten the lifespan of the solar module. This paper
proposes an efficient machine vision scheme for automatic
defect detection in multicrystalline solar modules. It can detect
subtle defects in solar modules with inhomogeneous surfaces,
and is computationally very fast to implement. It requires no
complicated feature design and feature extraction process for
distinguishing local defects from a wide variety of crystal grain
patterns.
Due to the random crystal grain surface of a solar cell or
interior defects that do not visually appear on the cell surface,
automatic defect detection in solar modules with the typical
CCD camera and lighting system cannot be effectively realized.
In order to highlight the deficiencies that degrade the conversion efficiency of a solar module, the electroluminescence
(EL) imaging technique [8][10] has been proposed in recent
years. In the EL imaging system, current is sent through a
solar module in a darkened room, and then a cooled Si-CCD
or InGaAs camera is used to capture the infrared light emitting
from the excited solar module. Areas of crystal silicon with
higher conversion efficiency exhibit brighter luminescence in
the sensed image. Process deficiencies such as micro-cracks,
breaks and finger interruptions will appear as dark regions
because they are inactive and hardly emit light. The dislocation
and grain boundary of silicon wafers also create dark regions in
the background of the EL image. Because the crystal grain pattern is random and unique for each individual multicrystalline
solar wafer, the crystal grain backgrounds of all solar cells in
a solar module are also randomly present in the EL image.
This causes automatic defect detection of solar modules in EL
images extremely difficult.
Fig. 2(a) and (b) demonstrates the EL images of a defect-free
and a defective solar module, respectively. As seen in Figs. 1(c)
and 2(b), the solar module captured in the CCD image cannot
detect the inner micro-cracks and some subtle local defects,
while the EL image can well present various defects embedded
in the solar module. Fig. 3(a)(c) further shows three enlarged
solar cell subimages in a solar module, which contain respectively micro-crack, break and finger interruption. As seen in
Fig. 2(a), each solar cell in the solar module presents a random
background pattern. The defective solar cells in a solar module
can be visually present in the EL image, as shown in Fig. 2(b).
Because a typical solar module is composed of multiple solar
cells in series, the image size of a solar module is generally
very large. However, the local defects of a small solar cell in
the module image could be very small. The large image size
with a low resolution of defects demands a very efficient and
yet very effective automatic visual inspection scheme for defect
detection of multicrystalline solar modules in EL images.
As aforementioned, most of the machine vision algorithms
for defect detection in the photovoltaic industry are mainly focused on the process levels of solar wafers and solar cells. Fu
et al. [11] presented a machine vision method to detect cracks
123
Fig. 2. Demonstrative EL images of (a) defect-free solar module, and (b) defective solar module.
Fig. 3. Defect types of solar cells displayed in the EL image: (a) micro-crack;
(b) break; (c) finger interruption.
124
125
Once the structuring element -SE is chosen, the gray-level dilation (i.e., local maximum gray level) is applied to the solar cell
subimage with the structuring element in direction . That is
If -SE contains all defect points, the dilated value will be still
small for a dark region. Otherwise, the dilated value will be large
for a bright region in the background. In this study, we do not
use a horizontal structuring element (i.e., 0 -SE) because we
would like to remove the two horizontal thick bus bars in the
solar cell subimage. Fig. 5(a1)(d1) shows, respectively, a defect-free and three defective solar cell subimages in the EL images. Fig. 5(a2)(d2) presents the corresponding morphological
smoothing results of Fig. 5(a1)(d1). The filtering results show
that the dark regions in the background are better smoothed
while all the defects of micro-crack, break and finger interruption are well preserved. The morphological smoothing process
allows a better extraction of representative basis images in the
ICA learning stage and a better image reconstruction in the detection stage.
B. ICA Model and Basis Images
In the ICA model [17], [24], the observed mixture signals
can be expressed as
(1)
where is an unknown mixing matrix; represents the latent
source signals, meaning that they cannot be directly observed
from the mixture signals . The ICA model describes how the
observed mixture signals are generated by a process that uses
the matrix to mix the latent source signals . The source signals are assumed to be mutually statistically independent. Based
on this assumption, the ICA solution is obtained by finding a
demixing matrix . It is used to linearly transform the observed
mixture signals to yield the estimated independent signals
with
(2)
The separated components in , called independent components (ICs), are required to be as mutually independent as possible. The matrix is an estimate of the latent source signals .
There exist many algorithms performing ICA [25][29]. FastICA [26], [30] is one of the most widely used techniques to
solve for the ICA model due to its computational efficiency
and effectiveness for source separation. The objective of an
126
(3)
where
is a Gaussian random vector of the same covariance matrix as .
is the entropy of a random vector .
The negentropy is always non-negative and is zero if and only
if the random variable has a Gaussian distribution. It is well justified as an estimator of the non-Gaussianity of the ICs. Since
the exact probability density of the random variable is unknown,
the exact negentropy computation using (3) above is prohibited.
An approximation of the negentropy is proposed as follows:
(4)
where
ance.
is a Gaussian variable of zero mean and unit variis a non-quadratic function, which can be given by
, as suggested by Hyvarinen et al. [17].
The demixing matrix
with
that maximizes the
negentropy can be efficiently calculated with the fixed-point
search algorithm [26], [27] in FastICA.
To find a set of representative basis images, a collection of
partitioned solar cells randomly chosen from the defect-free
where
is the coefficient vector of the linear
combination. It can be used as the feature vector of the test
sample . The coefficient vector can be obtained by
(7)
is the pseudo-inverse of , and is given by
.
To determine the presence or absence of defects in the test
sample , the cosine distance is used to evaluate the similarity
between the test subimage with feature vector and every
sample
with feature vector
in the training data set .
Hence, the cosine distance
is defined as
where
(8)
where the feature vector
of training sample
is given by
(9)
127
A cosine distance of zero is obtained if the two compared subimages are identical. The final cosine distance of the test sample
is given by the one with the minimum value among all the
training samples in the data set , i.e.,
(10)
is larger than some preIf the minimum cosine distance
determined threshold, the test sample is classified as a defective one. Otherwise, it is claimed to be defect-free.
Since the feature extraction approach is based on the distance
measure of the test sample with respect to every training sample
in the data set, the choice of defect-free solar cell subimages
used in the training data set becomes very critical. We therefore
also propose a self-referential image reconstruction process to
detect the presence/absence of defects in a test image. Any test
sample can be reconstructed from the basis images by
(11)
where is the coefficient vector as defined in (7). The reconstruction error of the test sample is defined as
(12)
where the constant is used as a regularization, and is given by
(13)
If the test sample contains defects, we expect that it cannot be
well reconstructed from the defect-free basis images and the resulting reconstruction error should be distinctly large. If the reconstruction error is larger than some pre-determined threshold,
the test sample is identified as a defective one. Otherwise, it
is classified as a defect-free solar cell subimage.
For self-referential image reconstruction with ICA, there is a
special finding that is worth noting here. Because the demixing
matrix
given by the FastICA algorithm is orthogonalized, it
indicates that
. The reconstruction of a test sample
can be thus given by
computational complexity is
for the coefficient vector
and
for the reconstruction image . Therefore, the
overall computational complexity is
. The
computation can be easily and efficiently executed with matrix
operations.
III. EXPERIMENTAL RESULTS
This section presents the experimental results from a number
of multicrystalline solar modules in EL images to evaluate the
performance of the proposed method. All proposed algorithms
and experiments were implemented on a personal computer
with a Pentium Core 2 Duo 3.00 GHz processor. The whole
image of a solar module is 1250 1250 pixels wide with 8-bit
gray levels. The solar module under test comprises a matrix of
36 solar cells, each of image size 208 208 pixels. The proposed ICA-based defect detection method is highly dependent
on the defect-free training samples for extracting representative
basis images. Especially, the image reconstruction approach
evaluates the reconstruction error directly from the original
training samples. Different defect-free samples used to generate
the representative basis images may produce different detection
results. Therefore, we conducted 10 training sets of randomly
chosen defect-free samples for the experiments. Each training
set contains 30 random defect-free solar cell subimages and,
thus, each data set creates a different combination of 30 basis
images.
The performance of the proposed method under different
scenarios (e.g., input images with and without morphological
smoothing, and feature extraction approach vs. image-reconstruction approach) is evaluated by the recognition rate. It is
defined as
%
where
In order to identify defective solar cells with large reconstruc, we use the simple statistical control limit to set
tion errors
the threshold value. It is given by
(14)
In the equation above,
is the coefficient vector for the
linear combination of the original data matrix . Therefore, the
image reconstruction approach for detecting the presence/absence of defects in a test sample does not require the ICA
optimization search, i.e., the learning stage can be omitted. The
original data set itself forms the basis images and the same
reconstruction error can be obtained either from the ICA basis
images or from the original training images .
The proposed method needs to calculate the coefficient vector
using (7) and then reconstruct the inspection image from the
basis images using (11). Given an image of size
and
basis images, the reshaped 1-D image is of size
with
and the matrix
is of size
, the
where
and
are the mean and standard deviation of
from a set of defect-free test samples. The parameter
is a control constant. It is given by 3 in this study to follow
the commonly-used 3-sigma rule in statistical process control.
Likewise, the threshold for cosine distance
is also determined by the statistical control limit. In the experiments, there
are a total of 80 solar cell subimages used as the test samples, of
which 28 are defect-free and 52 are defective. The test samples
used in the experiments were provided by a local solar cell manufacturer. The quality engineers care more about the detection
capability for various defect types that may present randomly in
any locations in solar modules.
128
TABLE I
RECOGNITION RATES OF ICA-BASED FEATURE EXTRACTION AND IMAGE
RECONSTRUCTION APPROACHES
The first experiment was conducted to evaluate the performance between the feature-extraction
and the image-reconstruction
. The morphological preprocessing is applied to both ICA approaches, and the number of basis images
is 30. Table I summarizes the statistics of recognition rates R%
of the two ICA methods for the 10 training data sets. It shows
that the average recognition rate of the image reconstruction approach is 93.4% with a maximum of 98.7%, while the average
recognition rate of the feature extraction approach is only 78.4%
with a maximum of 82.5%. The Std. dev. in the table means
the standard deviation of recognition rate % obtained from
the 10 training sets. Table I shows that most of the 10 training
sets can yield a good recognition rate larger than 92% and the
standard deviation is only 3.2% for the image reconstruction approach. The variation is due to the different basis images derived from individual training samples. If the training samples
contain good representations of various background patterns of
defect-free solar cells, the derived basis images can then sufficiently represent all possible background patterns of the test
samples. Since the training process can be performed off-line,
we can always carry out multiple training sets that contain different defect-free samples in a preliminary experiment and then
select the one that results in the maximum recognition rate for
on-line inspection in manufacturing.
To further compare the performance of these two approaches, the receiver operating characteristic (ROC) curves
are also given in Fig. 6 for the 9th training data set that gives
the best recognition rate. In the ROC plot, the true positive (TP)
rate is shown in the vertical axis and the false positive (FP)
rate is shown in the horizontal axis. The results show that the
ROC curve of the image reconstruction approach rises swiftly
upward and far outperforms the feature extraction approach.
The image reconstruction method uses the whole image (
data points of an
image) to calculate the reconstruction
error. It performs the self-comparison between the input image
and the template image pixel by pixel, where the template
image is generated from the input image itself. However, the
feature-extraction method uses only a small number of data
points, where
is the number of basis images, to calculate
the difference of feature vectors between the inspection image
and each individual defect-free image used in the training set.
Fig. 6. Receiver operating characteristic (ROC) curves of the ICA-based feature extraction and image reconstruction approaches.
TABLE II
RECOGNITION RATES OF IMAGE RECONSTRUCTION APPROACH WITH AND
WITHOUT MORPHOLOGICAL SMOOTHING
TABLE III
EFFECT OF THE NUMBER OF BASIS IMAGES ON THE RECOGNITION RATE
129
130
Fig. 7. Comparison of the proposed method with adaptive thresholding/morphology approaches: (a1)(d1) EL images of one defect-free and 3 defective solar
cells for testing; (a2)(d2) respective detection results from adaptive thresholding and binary morphology; (a3)(d3) respective detection results from gray-level
morphology.
REFERENCES
[1] Y. Si, J. Mei, and H. Gao, Novel approaches to improve robustness,
accuracy and rapidity of iris recognition systems, IEEE Trans. Ind.
Inf., vol. 8, pp. 110117, Feb. 2012.
[2] S. Jin, D. Kim, T. T. Nguyen, D. Kim, M. Kim, and J. W. Jeon, Design and implementation of a pipelined datapath for high-speed face
detection using FPGA, IEEE Trans. Ind. Inf., vol. 8, pp. 158167,
Feb. 2012.
[3] R. Gerndt, S. Michalik, and S. Krupop, Embedded vision system
for robotics and industrial automation, in IEEE Int. Conf. Ind. Inf.
(INDIN), Caparica, Lisbon, 2011, pp. 895899.
[4] S. Livatino, F. Banno, and G. Muscato, 3-D integration of robot vision and laser data with semiautomatic calibration in augmented reality
stereoscopic visual interface, IEEE Trans. Ind. Inf., vol. 8, pp. 6977,
Feb. 2012.
[5] C.-H. Chan and G. K. H. Pang, Fabric defect detection by Fourier
analysis, IEEE Trans. Ind. Appl., vol. 36, pp. 12671276, 2000.
[6] A. Picon, O. Ghita, P. F. Whelan, and P. M. Iriondo, Fuzzy spectral
and spatial feature integration for classification of nonferrous materials
in hyperspectral data, IEEE Trans. Ind. Inf., vol. 5, pp. 483494, 2009.
[7] D.-M. Tsai, I.-Y. Chiang, and Y.-H. Tsai, A shift-tolerant dissimilarity
measure for surface defect detection, IEEE Trans. Ind. Inf., vol. 8, pp.
128137, Feb. 2012.
[8] T. Fuyuki and A. Kitiyanan, Photographic diagnosis of crystalline silicon solar cells utilizing electroluminescence, Appl. Phys. A, vol. 96,
pp. 189196, July 2009.
[9] T. Fuyuki, H. Kondo, T. Yamazaki, Y. Takahashi, and Y. Uraoka, Photographic surveying of minority carrier diffusion length in polycrystalline silicon solar cells by electroluminescence, Appl. Phys. Lett.,
vol. 86, p. 262108, Jun. 2005.
[10] Y. Takahashi, Y. Kaji, A. Ogane, Y. Uraoka, and T. Fuyuki, LuminoscopyNovel tool for the diagnosis of crystalline silicon solar
cells and modules utilizing electroluminescence, in Proc. 4th World
Conf. Photovoltaic Energy Conv., Hawaii, 2006, pp. 924927.
[11] Z. Fu, Y. Zhao, Y. Liu, Q. Cao, M. Chen, J. Zhang, and J. Lee, Solar
cell crack inspection by image processing, in Proc. Int. Conf. Business Electronic Product Reliability Liability, Shanghai, China, 2004,
pp. 7780.
[12] M. A. Ordaz and G. B. Lush, Machine vision for solar cell characterization, in Proc. SPIE, San Jose, CA, 2000, pp. 238248.
131
[26] A. Hyvarinen and E. Oja, A fast fixed-point algorithm for independent component analysis, Neural Comput., vol. 9, pp. 14831492, Oct.
1997.
[27] A. Hyvarinen, Fast and robust fixed-point algorithms for independent
component analysis, IEEE Trans. Neural Netw., vol. 10, pp. 626634,
May 1999.
[28] R. Boscolo, H. Pang, and V. P. Roychowdhury, Independent component analysis based on nonparametric density estimation, IEEE Trans.
Neural Netw., vol. 15, pp. 5565, Jan. 2004.
[29] T. Blaschke and L. Wiskott, CuBICA: Independent component
analysis by simultaneous third- and fourth-order cumulant diagonalization, IEEE Trans. Signal Process., vol. 52, pp. 12501256, May
2004.
[30] J. Hurri, H. Gavert, J. Sarela, and A. Hyvarinen, FastICA Package [Online]. Available: http://www.cis.hut.fi/projects/ica/fastica/
Du-Ming Tsai received the B.S. degree in industrial engineering from the
Tunghai University, Taiwan, in 1981, and the M.S. and Ph.D. degrees in
industrial engineering from Iowa State University, Ames, in 1984 and 1987,
respectively.
From 1988 to 1990, he was a Principal Engineer of Digital Equipment Corporation, Taiwan, where his work focused on process and automation research and
development. Currently he is a Professor of Industrial Engineering and Management at the Yuan-Ze University, Taiwan. His research interests include automated visual inspection, object recognition and texture analysis.
Wei-Yao Chiu received the B.S. degree in industrial management from the
Chung-Hua University, Taiwan, in 2005 and the M.S. degrees in industrial engineering and management from the Yuan-Ze University, Taiwan, in 2007, where
he is currently pursuing the Ph.D. degree in industrial engineering and management.
His research interests include automated visual inspection and video surveillance.
S3-1
Invited
High efficiency amorphous and nanocrystalline silicon thin film solar cells on flexible
substrates
Baojie Yan, Guozhen Yue, Jeffrey Yang, Subhendu Guha
United Solar Ovonic LLC
1100 West Maple Road, Troy, Michigan 48084, USA
Phone: +1-248-370-5410
Fax: +1-248-362-4442
E-mail: byan@uni-solar.com
We review the progresses and issues towards manufacturing hydrogenated amorphous silicon (a-Si:H) and nanocrystalline silicon (nc-Si:H)
based thin film multi-junction solar cells using a roll-to-roll process on flexible substrates. United Solar has been heavily involved in the
research and development of high efficiency a-Si:H and nc-Si:H multi-junction solar cells since 2001. We have resolved several critical issues
limiting nc-Si:H solar cell performance, such as nanocrystalline evolution, impurities, and porosity/ambient degradation. We have developed
new cell designs, including the proper optimization of the seeding layer for nc-Si:H growth and proper n/i and i/p buffers for improvement of
cell efficiency. We have optimized Ag/ZnO back reflectors for nc-Si:H cell performance. Combining all of the efforts in the improvement of
material quality and optimization of device structure, we have advanced thin film silicon solar cell efficiency. We reported 16.3% initial
active-area efficiency using an a-Si:H/a-SiGe:H/nc-Si:H triple-junction solar cell. Furthermore, we attained 12.5% stable total area (0.27 cm2)
efficiency using a-Si:H/nc-Si:H/nc-Si:H triple-junction solar cells and 11.3% stable aperture area (800 cm2) efficiency using the same cell
structure, where the efficiencies were measured by NREL and were the records for thin film silicon photovoltaic technology.
Introduction
The photovoltaic (PV) solar energy industry has been growing
very fast in the last a few years. Crystal silicon (c-Si) PV panel
solar cells imply higher stable solar cell and module efficiencies
than a-Si:H and a-SiGe:H based multi-junction solar cells. With the
(a-SiGe:H) based thin film silicon have been used as the absorber
layers in multi-junction solar cells. The high defect density and low
lower efficiency than c-Si solar cells. Although 14.6% initial and
1.
1)
8.2%. The lower efficiency in the product than the lab devices is
uniformity are still the major tasks for the thin film silicon PV
United Solar,
2)
2.
Experimental Details
We have investigated three methods for nc-Si:H solar cell
AM-FPD '12
67
deposit nc-Si:H at very high rates such as >3 nm/s, but the material
pressure, and total gas flow rate may also be used, but no
Eff
Sample
Jsc
Voc
No.
(mA/cm2)
(V)
A-C
21.48
0.482
0.632
6.54
CHD
A-P
25.15
0.502
0.663
8.37
PHD
1200 nm. For large area modules, metal grid wires are bonded on
B-C
26.61
0.504
0.665
8.92
CHD
B-P
26.73
0.502
0.685
9.12
PHD
(%)
Comment
68
AM-FPD '12
Fig. 1. QE spectra of nc-Si:H solar cells with (a) n-type impurity and (b)
p-type impurity contaminations.
the best nc-Si:H solar cell efficiency.5) Fig. 3 shows the J-V and QE
the weakly n-type doped nc-Si:H is used in an n-i-p solar cell, the
region near the n-i interface becomes n-type and the electric field in
this region are not easily collected by the external circuit without
reverse bias, but a large collection loss causes a poor FF and low
Due to the same reason, when nc-Si:H solar cells have p-type
quality.
5
-5
J (mA/cm 2)
(a)
Jsc=30.47 mA/cm2
Voc=0.488 V
FF=0.646
Rs=2.33 W.cm2
Eff=9.61%
-10
-15
-20
-25
reduction in fill factor (FF). Both the n-type and p-type impurities
-30
-35
-0.1 0.0
0.1
0.2
0.3
0.4
0.5
V (V)
(b)
level down to less than 51018 at./cm3 and B level down to less
0.8
0.6
0V
QE
0.4
-5 V
0.2
0.0
300
500
700
900
Wavelength (nm)
1100
Fig. 2. (a) J-V characteristics and (b) QE curves of a nc-Si:H solar cell. The
between the nc-Si:H i-layer and the p-layer to reduce the shunt
current and improve Voc.
Light trapping
Because of the indirect bandgap in the crystalline phase,
nc-Si:H solar cells need much more thicker i-layers than a-Si:H
Multi-junction structures
In order to use the solar spectrum effectively and reduce the
light-induced degradation, multi-junction solar cells are commonly
used for high efficiency. We have investigated various
substrates such as Ag/ZnO BRs are used for highly effective light
efficiency
one cannot use very textured BRs in nc-Si:H solar cells for high
is
lower
than
triple-junction
structures.
nc-SiOx:H layer between the a-SiGe:H middle cell and the nc-Si:H
bottom cell, we achieved an initial active-area efficiency of
16.3%.6) The J-V characteristics and QE curves of the high
AM-FPD '12
69
limit the solar cell efficiency, we have improved thin film silicon
still very limited. The plasma process can be further optimized and
achieved using this structure is not the highest, the light induced
production.
degradation is low and the deposition rates can be high for all three
J (mA/cm2)
-2
11.3%
Voc=2.242 V
FF=0.771
J sc=9.43 mA/cm2
Eff=16.3%
-4
an
a-Si:H/nc-Si:H/nc-Si:H
-6
-8
0.0
0.5
1.0
Voltage (V)
1.5
2.0
2.5
1.0
28.6 mA/cm2
0.9
4.
Summary
We reviewed the issues and solutions of nc-Si:H solar cells.
0.8
0.7
0.6
are the major areas for improving nc-Si:H component cells. Using
0.5
9.43
0.4
9.44
0.3
9.73
0.2
the issues and to improve thin film silicon solar cell efficiency. We
0.1
believe that thin film silicon solar cell efficiency will be improved
0.0
300
500
700
900
1100
Wavelength (nm)
manufacturing.
Acknowledgments
The work is supported by the US DOE SAI under Sub-Contract
17
16
using
-10
QE
respectively,
15
References
14
1) J. Yang, A. Banerjee, and S. Guha, Appl. Phys. Lett. 70, 2975 (1997).
13
12
860 (1994).
11
2002
2004
2006
2008
2010
2012
Year
cell is very high, the a-SiGe:H middle cell has to be thick enough
70
AM-FPD '12
6) B. Yan, G. Yue, L. Sivec, J. Yang, S. Guha, and C.-S. Jiang, Appl. Phys.
17
I. INTRODUCTION
IGH-EFFICIENCY solar cells require absorption of photons of the full solar spectrum followed by effective generation and collection of charge carriers. The high bandgap of
amorphous silicon of 1.8 eV implies that the material is transparent for sub bandgap, near infrared (NIR) light, constituting a
high photon loss. Upconversion (UC) may enhance the response
of solar cells in the infrared [1]. UC is a luminescence process,
whereby two or more low-energy photons are converted to one
higher energy photon. When a layer containing UC species is
attached to the rear of a solar cell, the sub bandgap photons are
absorbed and higher energy photons are emitted; these can subsequently be directed to the solar cell using an optical reflector
behind the cell, where they can be absorbed in the active layer.
An increasing research effort is undertaken on crystalline silicon solar cells as well as thin-film silicon solar cells using
Manuscript received April 25, 2012; revised July 27, 2012; accepted August
9, 2012. Date of publication September 18, 2012; date of current version December 19, 2012.
J. de Wild, T. F. Duindam, J. K. Rath, and R. E. I. Schropp are
with the Faculty of Science, Debye Institute for Nanomaterials Science, NanophotonicsPhysics of Devices, Utrecht University, Utrecht
3508, The Netherlands (e-mail: j.dewild@uu.nl; T.F.Duindam@students.uu.nl;
j.k.rath@uu.nl; r.e.i.schropp@uu.nl).
A. Meijerink is with the Faculty of Science, Debye Institute for Nanomaterials Science, Condensed Matter and Interfaces, Utrecht University, Utrecht
3508, The Netherlands (e-mail: A.meijerink@uu.nl).
W. G. J. H. M. van Sark is with the Copernicus Institute, Utrecht 3584, The
Netherlands (e-mail: W.G.J.H.M.vanSark@uu.nl).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/JPHOTOV.2012.2213799
(1)
where n is the number of photons needed to excite the upconverted state. Nn is the nth excited state in the Er3+ ion
and Ns the excited state of the sensitizer ion Yb3+ . When a
18
20
1000
cm-1
4I
9/2
4I
11/2
2F
5/2
980
nm
9/2
4F
7/2
2H
11/2
4S
3/2
4F
9/2
15
10
2H
410
nm
2F
7/2
660
nm
1540
nm
Yb3+
525
550 nm
nm
Er3+
4I
13/2
4I
15/2
Fig. 1. UC in the (Yb3 + , Er3 + ) couple. The dashed lines represent energy
transfer, the full lines the radiative decay, and the curly lines indicate multiphonon relaxation processes. The main route is a two-step energy transfer that
leads to excitation to the 4 F7 / 2 state of the Er3 + ion. After relaxation from this,
state emission is observed from the 2 H1 1 / 2 , the 4 S3 / 2 level (green), and the
4F
9 / 2 level (red).
(2)
UC
the photogenwhere Pin is the incoming light intensity, and ISC
erated short-circuit current in the solar cell. Thus, for current
increase due to UC we expect a quadratic power dependence
on the concentration factor because the upconverted emission
happens after absorption of two photons.
Fig. 2. Transmission of light for flat and textured solar cells and the collection efficiency when light is coming though the n-layer, which is the case for
upconverted light. Most emission of the upconverted light is at 550 nm, where
the quantum efficiency for the textured and flat solar cells are the same.
DE WILD et al.: INCREASED UPCONVERSION RESPONSE IN a-Si:H SOLAR CELLS WITH BROAD-BAND LIGHT
19
B. Upconverter
The absorption and emission spectra are shown in Fig. 3.
The absorption is highest around 950 nm. The upconverter was
excited with a xenon lamp at 950 10 nm and 980 10 nm.
The 4 F7/2 state at 2.52 eV is reached after two times energy
transfer from Yb to Er. The upconverter was already shown to
be very efficient at low light intensities. Saturation was measured
under light intensities of less than 1 W/cm2 [12]. Although the
absorption at 950 (1.31 eV) nm is higher, excitation at 980
(1.26 eV) nm leads to a two times higher upconverted emission
intensity. This may be attributed to the perfectly resonant energy
transfer step of 980 nm (1.26 eV), since the 4 F7/2 state is at
2.52 eV.
The upconverter was for further experiments excited at
980 nm with the Opolette laser. Because UC is a two-photon
process, the efficiency should be quadratically dependent on the
excitation power density. The intensity of the laser light was varied with neutral density filters. UC spectra were recorded in the
range of 400 to 850 nm under identical conditions with varying
excitation powers. Varying the intensity shows that for low light
intensities the red part is less than 6% of the total emission, see
Figs. 3 and 4. Only when the emission from the green emitting
states becomes saturated the red emission becomes more significant and even blue emission from 2 H9/2 state is measured, see
Fig. 4. By comparing the emission intensities, it becomes clear
that the emission intensity is not increasing quadratically with
the excitation power density. Instead emission from higher and
lower energy states are visible. The inset shows the integrated
emission peaks for the green and total emission. With this insert,
it is shown that at very high laser intensities the total emission
is saturated.
C. a-Si:H Solar Cells
The subbandgap response in the NIR due to the band
tails of a-Si:H solar cells is not neglectable [4]. To distin-
Fig. 5. Dark and light IV curves of textured solar cells with upconverter.
The solar cell was illuminated by long wavelength light (>900 nm), which was
concentrated 25 times.
20
Total current
Flat Upconverter, slope 1.4
Textured Upconverter, slope 1.2
Flat Reference, slope 0.9
Textured Reference, slope 1.0
0.25
-1
10
10
15
20
25
30
Current (mA/cm )
0.30
Current (mA/cm )
10
0.35
Current (mA/cm )
0.20
0.15
0.10
0.05
0.00
900
-1
10
1100
1300
-2
10
10
15
20
25
30
Solar Concentration
Fig. 6. Current measured in the solar cells under illumination of subbandgap
light. In the upper figure, the total current of the reference and UC cells are
plotted as a function of the concentration factor; in the lower graph, the current
generated by the upconverter is shown. The slope for subbandgap response is 1
for flat and textured solar cells. The contribution of the upconverter increases
the slope slightly; when corrected for the subbandgap response, the slope is 1.5
for the textured and 1.8 for the flat solar cells.
DE WILD et al.: INCREASED UPCONVERSION RESPONSE IN a-Si:H SOLAR CELLS WITH BROAD-BAND LIGHT
21
of
appropriate
materials
for
4-
or
5-junction
solar
cell
_ [PV
to
alloy
scattering
and
non-homogeneity
of
N.
Present
Co urt esy
: D:iido
(FM-CBE)
method
in
order
to
increase
To solve above
iNDEX TERMS
and
M.Yam
A.
11.
__
"""
(NEOO
;;:::
::
Fra"nhofE[.
-CPV ""
BSQ,
,!I!..J!I.J.:.ft
_ .. , _
I.INTRODUCTION
space
and
terrestrial
applications
because they
have
high
Japanese
Innovative
Photovoltaic
R&D
program
since
978-1-4673-0066-7/12/$26.00 2011
IEEE
000831
[NJ
'm
.....
"b
0.85%, O.4Ilm/h
As-grown
b\ ni
and p-type GaAsN films were carried out from 2SK to 3S0K by
0.92
/3
.....
...
located
Improved!!
100000
50000
21d(cm')
150000
temperature,
the thermal ionization of H2, whose density affects in great part the
30
AR
T300
&200
:c
100
-<111-<111.
4.
I\fBE
[1][2]
IVOCVD[1]
-<III
o
0.0
0.5
E20
(J
'"
015
.....
[110
<?
z'r.
-<III
CBE Ga&As-step
25
1.0
composition rIo]
5 '"
",-
()
1 -8
-,
-12
AM1.5G
Cell area: O.2686cnt
Jsc= 15.04 rnlVcnt
1
"
1:>'
"
1t:J.D. 9
C,LlOI
,
/II
1/1
1/1
,I ..{]
/
-16!:,.--x-..
!--.-'---;h,.--'--,
...
---;h;:
.. 0.2....0.4
...
....
.4 ...-'-- -0.2....
0.0
0.6
0.8
1.0
v [V]
Fig. 4. Dark and light JV curves of GaAsN solar cell with anti
'"
'"
'"
'"
"
Voc= 0.662 V
FF = 0.650
Eff=7.18%
'"
'"
'"
'"
OL-------
0.1
0.2
0.3
0.4
0.5
[N] (%)
o'------+-----------+-----
'"
","'.
'"
'"
'O"[],
"
10
1.40
to
1.45
1000fT(111<1
1.50
0
/
DJd
0/6
r-----
o 400'C
.'oc
420'C
o
5
[N.H] (x10"cm"l
.30C
10
reflection coating.
III. DLTS ANALYIS
978-1-4673-0066-7/12/$26.00 2011
IEEE
000832
"I""
03
><
89
,
200
Temperature (K)
estimation
of
minority
carrier
lifetime
using
the
978-1-4673-0066-7/12/$26.00 2011
IEEE
000833
2.0r------I1'---,
3)
1.5
S'
C> 1.0
GaAS
'(ij
(/)
>
00.5
'"
W
>.
30
35
150
Temperature(K)
Q)
I:
W
C
o
300
-3 V, a
24.4
S-I,
r-------------------...,
--20k
--30k
--40k
--50k
CBM
1.29
I.
0.890.84-
:.;:::;
I1l
>
:;::::
CJ
GaAs
E3
0.59'
(Ni
0.48 .
E2 EL2 type
H5
Double
<t
III
II.
0.15 -
...
H2
0.05
- HO
O
I-
Fig.
12.
VBM
--.......-
Development
Organization
under
the
Ministry
of
Fig.
II.
1.33
Photon Energy (eV)
promising results:
I)
1997) p.875.
978-1-4673-0066-7/12/$26.00 2011
been found to also act as acceptor state and correlate with N-H
2)
I)
IEEE
000834
Green Energy Research Institute, Hyundai Heavy Industries, Co., Ltd., Yongin, 446-716, KOREA
2
Green Energy Business Division, Hyundai Heavy Industries, Co., Ltd., Eumseong, KOREA
Abstract
Si:H/c-Si
structure
is
one
of
the
candidates
for
high
efficiency solar cells. Despite SHJ solar cells show more than
(6
inch,
mono-crystalline,
n-type,
0.53.0
were
deposition
industrial
application.
Dimensions
of
back
contact
deposited
by
plasma-enhanced
(PECVD)
on
back
chemical
side.
vapor
Deposition
layers
were fabricated.
formed
Index Terms
hetero-junction
were
on
deposited.
Finally,
both
and
p-
metal
n-type
contacts
were
electrodes.
J-V
I. INTRODUCTION
Texturing process
(Voe),
low
Front AR coating
Back a-Si:H(i)
back a-Si:H(i)
Back a-Si:H{n)
Back electrode
Fig. I.
Back contact pattern formation processes for heterojunction back contact (HBC) solar cells
In HBC solar cell structure, the ratio of p-type emitter
to n-type BSF region on the back side was important to
interdigitated
silicon
heterojunction
(a-Si:H/c-Si)
and
fabricated.
Especially
deposition
conditions
of
III. RESULTS
To obtain high efficiency HBC solar cells, the intrinsic a
Si layer with good interface passivation properties on back
side should be deposited after texturing process. Low
001023
25
i
'120
lS
10
]2D'lm",.tI
Measured
Voc(mv)
power B. 240'C
Fig. 3.
Comparison with measured
simulation result of SHJ solar cell.
power C. 240'C
I-V
data
and
2D
I k M ..,...----.....,
N type c-Si
20
Hz/SiH.
I
I
A. 220C
power B. 200'C
f-
+-
.-Si:H
(P)
Fig. 2.
Variations of implied Voc as a function of deposition
condition for the intrinsic a-Si layers.
Parameters
Fig. 4.
Schematic structure of HBC solar cell used for
ATLAS 2D simulation.
a-Si(i)
Thickness (urn)
180
10
Dielectric constant
11.8
11.8
11.8
11.8
1.12
1.74
1.74
1.72
2.8xl0'9
1.04x102O
0
lx102
0
lxl02
0
lxl02
20
Ix10
0
lx102
0
lxl02
1040
412
lxlO'7
lxlO'7
lxl0'7
lxlO'7
lxl0'7
lxlO'7
lxl0'7
lxl0'7
0
lx102
lxl0'6
20
Ix10
001024
24 r----,
23.5
..
'u
23
::
w
22.5
22
==
9.
p
10
U
..
to
l
0
. a__________
aoo
900
1000
1100
1200
Fig. 7.
cell.
Pitch (um)
Fig. 5.
Variations of efficiency for HBC solar cells as a
function of the pitch distance (The ranges of the area ratio of p
type emitter to n-type BSF are 0.60.8, gap 10Ilm & 100Ilm).
=
IV. CONCLUSION
The deposition conditions of amorphous silicon were
optimized to fabricate HBC solar cells. Also the ATLAS
2D simulation was used to design back contact patterns.
Patterning processes were performed by using etching
paste and polymer resist, respectively. 5 x 5 cm2 size
HBC solar cells were manufactured successfully. I-V
characteristic data of HBC samples will be shown in the
poster presentation. Optimization of pattern design and
development of processes are ongoing project in our
group.
REFERENCES
[I] T. Mishima et aI., Solar Energy Mater. Sol. Cells 95 (2010)
18
[2] H. Fujiwara and M. Kondo, Appl. Phys. Lett. 90, 013503
(2007).
[3] T. H. Wang,E. Iwaniczko,M. R. Page,D. H. Levi,Y. Yan,
H. M. Branz, and Q. Wang, Thin Solid Films 501, 284
(2006).
[4] M. Lu, U. Das, S. Bowden,S. Hegedus, R. Birkmire, Prog.
Photovolt: Res. Appl. 19 326 (2011)
[5] N. Mingirulli, J. Haschke, R. Gogolin, R. Ferre, T.F.
Schulze, J. Dusterhoft, N. Harder, L. Korte, R. Brendel, B.
Rech,Phys. Status Solidi RRL 5,No. 4,159 (2011)
[6] M. Tucci,L. Serenelli,S.D. Iulliis,M. Izzi, G.D. Cesare,D.
Caputo,Phys. Status Solidi C 8,No. 3,932 (2011)
[7] M. Lu, S. Bowden, U. Das, R. Birkmire, Appl. Phys. Lett.
91,063507 (2007)
Fig. 6.
Patterning processes of amorphous silicon by using
etching paste (left) and polymer resist (right)
001025
INTRODUCTION
EXPERIMENTAL
001158
001159
001160
Cell
Reference
PERC
JSC
(mA/cm2)
VOC
(mV)
FF
(%)
Eta
(%)
36.7
37.9
639
646
79.4
79.2
18.64
19.36
Reflectivity
Front (%)
Reflectivity
Rear (%)
BSRV
(cm/s)
Reference
93
73
300
PERC
93
93
53
Cell
001161
001162
2
0epartment of Electrical Engineering, Korea Advanced Institute of Science and Technology,
Yuseong-gu, Oaejeon, 305-701 Korea
KIER-UNIST Advanced Center for Energy, Korea Institute of Energy Research, Yuseong-gu,
Daejeon, 305-343 Korea
Abstract
enables
ultrathin
film
absorber
application
and
single
electrode,
which
might
perform
high
light
node
separation.
We
found
the
signature
of
would
provide
promising
route
to
high
Index Terms
cells,
photovoltaic
photon
management,
Staebler
I. INTRODUCTION
In thin film Si solar cells, it is widely accepted that an
innovative light trapping technology is crucial for further
efficiency improvement [1]. The importance of light
trapping in thin film Si solar cells originates from
inherent light induced degradation property of amorphous
Si (a-Si) [2], because we cannot increase stabilized
efficiency simply by increasing the absorber layer
thickness. In addition, micromorph double junction
scheme [3] is regarded as a mandatory technology for a
large scale commercialization because of low power
conversion efficiency. However, low throughput of
microcrystalline Si deposition process [4] still limits cost
reduction opportunities.
II. EXPERIMENTAL
001182
FrontTCO
FrontTCO
(a)
(b)
TABLE I
VARIATION OF STEP COVERAGE WITH
DIFFERENT PROCESS CONDITIONS IN PECVD
DEPOSITED A-Si.
PECVD Condition
Sidewall Coverage
Normal Condition
29%
Higher Pressure
38%
Higher Flow
34%
Front Teo
Front TeO
(c)
(d)
Back TCO
4-....
4- ....
0
4-....
o
nip
nip
pin
4- ....
0
pin
(b)
Front TCO
(e)
Fig. 1.
(a)
(e)
Fabrication
sequence
of
proposed
200
400
600
800
1000
Wavelength (nm)
1200
gTCO.
001183
12 ---,-----,--.
N
E
10
(a)
u
(/)
--,
100
planar TCO
gTCO
200
300
iE
0.8
(b )..::.. ..
,
E 04
\WI
400
Sidewall
solar cell
\:;:.-+
___ gTCO
0.0
400
600
....
..::.
Wavelength (nm)
800
001184
Top cell
Sidewall cell
Corner cell
Teo
12 .---------,--,_--__.
NE
'iii
c
(])
"0
::J
1a
r---=-
(a)
ACKNOWLEDGEMENT
4
2
OL-________L-________
0.0
0.4
0.8
O -------.--.--.
1nm(gTCO)
- 2nm(gTCO)
- 3nm(gTCO)
- 1nm(planar TCO)
Voltage (V)
-5
hv
_
-
C)
.2
-10
-1
(b)
5 u------
0.0
0.7
-0.7
Voltage (V)
Fig. 5 (a) Photo J-Y characteristics of solar cell on gTCO and
planar ZnO. (b) Dark I-V characteristics of solar cells on gTCO
with various p-a-SiC thicknesses and solar cell on planar ZnO.
a-Si:H, Ilc-Si:H
and
'micromorph
tandem
solar cells", J
microcrystalline
silicon:
new
material
for
001185
Kroll,
with very high efficiency potential" , Appl. Phys. Lett., vol. 98,
2011
163503,2011
001186
Energy and Environmental Applications Center (E2TAC), College ofNanoscale Science and Engineering,
University at Albany, State University ofNew York, Albany, NY 12203, USA
2
Magnolia Solar Inc., 251 Fuller Road, CESTM-B250, Albany, NY 12203, USA
Abstract
layers for inexpensive thin film solar cells that are fabricated on
thin
film
solar
cells,
they
don't
compare
to
the
due
to
the
recombination
of
minority
carriers
at
grain
aligned
silicon
nanowire
(SiNW)
solar
cells
have
Therefore, a device
reduce the cost of the final PV cell and help increase market
penetration. Vertically aligned silicon nanowire (SiNW) solar
A number
high
performance
and
scale-up
utilizing
low
cost
The
nanowires
substrates.
were
fabricated
on
gold
coated
p-type
Si
Other methods,
deposition (PLD)[10],
12],
(VSS)[13]13,
SiNWs.
Index Terms
vapor-solid-solid
Usually
SiNWs
grown
[18]
by
14]
using
and
VLS
electroless
or
VSS
I. INTRODUCTION
semiconductor
material.
Recently
there
has
been
an
001911
SiNWs, which would trap electrons and holes in Si, and then
Throttle Valve
Heating
Precise
nanomaterials
remains
doping
a
of
challenge.
semiconductor
Recently,
(SLS)
method,
relatively
CNSE
'--
Vacuum Gauge
straightfordward
for
devices
make
their
incorporation
extremely
Fig. 1.
Schematic of catalytic CVD system that employs
vapor-liquid-solid mechanism to grow SiNWs.
of the reaction zone on the outer wall of the quartz tube, was
quick access door that was installed on the one end of the
quartz tube. On the same side of the tube, two gas delivery
been
suggested,
however,
its
feasibility
of
high
II. EXPERIMENTAL
A high volume and large sample area catalytic chemical
WO=
5mm
stainless
chemical
steel
vapor
substrates
deposition
using
system.
the
in-built
Figure
Fig. 2.
Typical SEM image of catalytic nanoparticels formed
after pre-treatment.
catalytic
shows
Si source and high purity argon (Ar) gas was used for
excess gas were removed from the other end of the reactor.
001912
Catalytic
n+
Al
emitter
i-SiNW
pSi
Diameter: -100nm
AI
approximately 5
substrate
on the substrate.
Height: 5-lOum
ionized water.
The
(SEM),
Transmission
Electron
Microscope
Fig. 4.
SEM micrographs of Si nanowires grown by catalytic CVD at various process temperatures using Au as a catalyst. (a) 450 C;
(b) 550 C; (c) 650 C; (d) 550 C and textured Si substrate.
001913
Fig. 5.
SEM micrograph of silicon nanowires grown by catalytic CVD at various process pressures. (a) 150 torr; (b) 170 torr; (c) 250
torr; (d) high resolution image of a nanowire in figure (c),shows the presence of gold catalyst on top.
type Si substrate.
i) light I-V curves under standard test conditions, ii) dark I-V
curves.
3) at room
Phosphorofilm from
efficiencies.
The annealed
cleaning
process.
The
cell
was
completed
by
EDS
compositional
analysis
indicated
that
the
observed
nm .
In order to control
001914
150, 175 and 250 torr pressure, 550C for 60 minutes using
nanowire is gold.
resolution
image
of
the
nanowire
shows
layer by layer
Fig. 6.
SEM micrograph and EDS elemental mapping analysis of a silicon nanowire. (a). Micrograph showing the nanowires. (b) Si
Ka spectral image; (c) Au Ma spectral image.
(a)
0.2 m
Fig.7.
TEM micrographs of silicon nanowires grown at 550 C using gold catalyst. (a). Micrograph showing presence of defects in the
nanowires. (b) High resolution micrograph shows nanotubes and gold catalyst cap are in single crystalline structure.
At
region. The gold signal from the substrate layer (catalytic film)
001915
Catalytic
CVD
process
parameter.
Further
process
IV. SUMMARY
A high volume and large sample area catalytic CVD system
was built and silicon nanowires were successfully grown on Si
and stainless steel substrates.
ACKNOWLEDGEMENT
This work is supported by National Science Foundation.
[14] H.F. Yan, Y.1. Xing, Q.L. Hang, D.P. Yu, Y.P. Wang, J. Xu,
Z.H. Xi, and S.Q. Feng, Chern. Phys. Lett. vo1.323, p.224,
2000.
[15] K.Q. Peng, Y.1. Yan, S.P. Gao, J. Zhyu, Adv. Mater. vol. 14,
pp.II64-1167,2002.
[16] K.Q. Peng,Y.1. Yan, S.P. Gao, J. Zhyu,Adv. Funct. Mater. 13,
2003, pp. 127-132.
[17] K.Q. Peng, 1.1. Hun, Y.1. Yan, et. al. Adv. Funct. Mater. 16,
2006, pp.387-394.
[18] Y. Wang, V. Schmidt, S. Senz, and U. Gosele, Nature
Nanotechnology, vol. I,p. 186,2006.
[19] N. Elfstrom, 1. Linnros, Appl. Phys. Lett., vo1.91, p. 103502,
2007.
[20] K.Q. Peng, X. Wang, S.T. Lee, Appl. Phys. Lett., vol.92,
p. 163103,2008.
[21] A.1. Hochbaum, R. K. Chen, R.D. Delgado, et.al. Nature
vo1. 451,pp. 163-168,2008.
[22] A. Asenov, A.R. Brown, 1.H. Davies, et al. IEEE Trans Elec
Dev., vo1.50,pp. 1837-1852,2003.
[23] 1.H. Lee, P.H. Rogers, M.A. Carpenter, E.T. Eisenbraum, Y.
Xue, and R.E. Geer, Proc. 81h IEEE conference on
Nanotechnology, (IEEE NANO'08) (Aug.,2008).
[24] T. Vo, A.J. Williamson, G. Galli, Phy. Rev. B, vo1.74, p.
045116,2006.
[25] L. Tsakalakos, J. Balch, J. Fronheiser,B.A. Korevaar, Appl.
Phys. Lett., vo1. 91,p.233117,2007.
[26] A.V. Nguyen, H. Efstathiadis, Z. Zhouying, A. Fioramonti, D.
Morrissey,P. Haldar, Proc. 341h IEEE PVSC, 2009.
REFERENCES
[I]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
001916
I. INTRODUCTION
An important goal of the PV industry is to reduce the PV
electricity cost in order to achieve the grid parity. This
objective is influenced by many factors as the introduction of
evolutionary technologies in order to increase the efficiency
and/or reduce the production cost. The first step to introduce
an evolutionary approach is the development in the laboratory,
followed by the pilot scale and mass production [1]. In lab.
scale, 4 cm2 silicon solar cells are usually developed and when
the area is enlarged technical problems may be overcome.
The industry of silicon solar cells is based on the aluminum
back surface field (Al-BSF) formed by the conventional
screen printing metallization, using an aluminum paste
deposited on Czochralski-grown (Cz-Si) p-type substrate. The
average efficiency of industrial solar cells is about 16.5 % [1].
The Al-BSF rear passivation technology limits the efficiency
mainly in thin substrates [2]. Moreover, the screen printed Al
paste produces bowing of the wafer, despite of the evolution
of low-bow pastes. To overcome these challenges, the p+ layer
can be produced with boron doping.
In p-type Si substrates, the boron diffusion is used to
produce the back surface field (B-BSF) and the challenge is to
obtain the surface concentration that fits in well the standard
screen printing metallization and without contamination of the
silicon substrate during the high temperature diffusion process
[2]. Boron BSF silicon solar cells have been developed with
dielectric passivation and the efficiency of 20 % was achieved
in 4 cm2 cells and with screen printing metallization [3].
The boron diffusion in n-type silicon material forms the
emitter (B-emitter). Beyond of the above mentioned
challenges, the Al/Ag pastes currently used on p+ regions
present higher resistivity than the Ag paste to form the front
metal grid. Nevertheless, since the last decade n-type Cz-Si
solar cells are being investigated due to the potential to
produce high efficiency solar cells. Substrate doped with
phosphorus is more stable and have higher minority carrier
lifetimes than p-type silicon [4].
Sun Power developed the interdigitated contact solar cell
and the average efficiency in industrial production in float
zone substrate (FZ-Si) was above 22 % [1]. With other
structure and front boron emitter, Mihailetchi et al [5]
developed large area solar cells with efficiency of 17.4%. In
multi-crystalline silicon wafers (MC-Si), solar cells with an
area of 156.25 cm2 were processed with boron and
phosphorus simultaneous diffusion and the efficiency of 16.4
% was obtained [6]. The influence of the substrate in boron
doped emitter cells was experimentally studied and the
efficiency of 14.7 %, 15.9 % e 17.1 % was achieved with MCSi, Cz-Si and FZ-Si, respectively [7]. Solar cells were also
processed with boron doped p+ layer using spin-on dopant
instead the conventional BBr3 dopant. In FZ-Si n-type
material, the efficiency of 15.9 % was obtained with
SiO2+SiNx:H passivation.
The aim of this paper is to present the development and
comparison of small and large area screen printed silicon solar
cells, processed by using spin-on dopant to form the boron
emitter in n-type Cz-Si substrates and B-BSF in p-type
wafers.
II. SOLAR CELL PROCESS
The baseline process used to develop the solar cells was the
following: texture etching, RCA cleaning, boron spin-on
deposition, boron diffusion in a quartz tube furnace, resist
deposition and oxide etching, RCA cleaning, phosphorus
diffusion, borosilicate glass etching, TiO2 antireflection
coating deposition, screen-printing metallization in both sides
and edge isolation. Solar cells were developed in 1 .cm 20
.cm n-type and p-type solar grade Cz-Si wafers with
thickness of 200 m. The boron dopant was spun onto one
002284
side of the wafer and the diffusion was carried out in a quartz
tube furnace at 1000 C. The n+ layer was produced by
phosphorus diffusion using POCl3. The oxide to protect the p+
layer of the phosphorus diffusion was grown in the same
thermal step of boron diffusion. A specific passivation was
not implemented.
The boron and phosphorus diffusions were independently
and experimentally optimized for n-type and p-type material
as well as the metal grid firing process. The metal grid was
formed with two busbars in large area cell and with only one
busbar in small area devices, in order to compare the typical
lab. cells with an area of 4 cm2 with the large area (61.58 cm2)
silicon solar cells.
All solar cells were characterized under standard conditions
(100 mW/cm2, AM1.5G and 25C) in a solar simulator
calibrated with a small (4 cm2) and a large (61.58 cm2) area
silicon solar cell previously measured at CalLab - FhG-ISE
(Fraunhofer-Institut fr Solare Energiesysteme), Germany.
Ag/Al paste used to form the metal grid on the frontal face.
With the same metal grid, but using an Ag paste on the frontal
surface, fill factors of 0.78 were obtained in n+pp+ solar cells
doped with phosphorus and aluminum.
TABLE I
VOC, JSC, FF AND EFFICIENCY OF THE 4 CM2 B-EMITTER (P+NN+) SILICON
SOLAR CELLS PROCESSED IN THE WAFER 3.
JSC
FF
(%)
Cell
VOC (V)
(mA/cm2)
3-A
598.4
35.1
0.72
15.1
3-B
586.3
34.7
0.74
15.0
3-C
591.5
35.4
0.73
15.2
3-E
587.5
35.1
0.74
15.2
3-F
597.5
35.9
0.72
15.4
3-G
595.3
35.5
0.70
14.8
3-H
596.9
36.0
0.70
15.1
Aver.
5935
35.40.5
0.720.02
15.10.2
TABLE II
ELECTRICAL CHARACTERISTICS OF THE 4 CM2 B-EMITTER (P+NN+)
SILICON SOLAR CELLS PROCESSED IN THE WAFER 7.
JSC
VOC
FF
(%)
Cell
(mV)
(mA/cm2)
7-A
584.2
35.0
0.74
15.2
7-B
576.5
34.6
0.73
14.7
7-C
585.5
35.2
0.74
15.3
7-D
577.7
34.6
0.74
14.8
7-F
577.9
34.7
0.73
14.7
7-G
588.7
35.5
0.74
15.5
7-H
580.2
34.9
0.72
14.5
7-I
589.7
35.2
0.73
15.3
Aver.
5825
35.00.3
0.740.08
15.00.4
1E+21
1E+20
1E+19
1E+18
1E+17
1E+16
1E+15
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Depth (um)
+
002285
in the frontal face. In large and small area cells, 9.4 % and 6.0
% of the area is covered by the metal grid, respectively. The
p+nn+ cells were simulated by using PC-1D device modeling
program and the 3.4% higher metal grid coverage in large area
cells implied in a JSC reduction of only 3.9%. For instance, it
is worth to comment that the difference between JSC of small
area and large area cells was also observed in n+pp+ cells
doped with phosphorus/aluminum, produced in the PUCRS
and measured at CalLab - FhG-ISE. In this case, the
difference was of around 7%.
1E+21
TABLE IV
ELECTRICAL CHARACTERISTICS OF THE 4 CM2 B-BSF (N+PP+) SILICON
SOLAR CELLS PROCESSED IN THE WAFER 8.
JSC
FF
(%)
Cell
VOC (V)
(mA/cm2)
8-F
580.6
33.3
0.74
14.3
8-G
580.5
32.7
0.73
13.8
8-I
580.9
33.6
0.73
14.3
Aver.
580.80.3
33.20.5
0.730.02
14.10.8
1E+20
1E+19
1E+18
1E+17
1E+16
1E+15
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Depth (um)
Fig. 2.
TABLE III
VOC, JSC, FF AND EFFICIENCY OF THE LARGE AREA B-EMITTER (P+NN+)
SILICON SOLAR CELLS AND AVERAGE VALUES.
JSC
Cell
VOC (V)
FF
(%)
(mA/cm2)
B1-A
596.2
31.0
0.73
13.5
B1-B
590.4
31.4
0.73
13.5
B2-A
596.1
31.6
0.70
13.2
B2-B
599.4
31.6
0.71
13.4
Aver.
595.51.7
31.40.5
0.720.4
13.40.2
In large area cells, fill factor was also lower than expected.
High metal grid resistivity could explain the low FF obtained.
To investigate the influence of the Ag/Al paste resistivity,
Ag/Sn/Cu ribbons were soldered on the front busbars and
cells were measured at standard conditions. Fill factor rose
from 0.70 to 0.75 causing an increase of 13.2 % to 14.9 % in
the efficiency. PV modules with 36 cells connected in series
were fabricated and FF reached values ranging from 0.74 to
0.76.
IV. BORON-BSF SOLAR CELLS
The process was also implemented to obtain n+pp+ (B-BSF)
cells. The sheet resistivity of the phosphorus emitter and the
B-BSF were 30 / , with the relative standard deviation of
20 % and 9%, respectively. Cells with different areas were
processed in the same batch.
TABLE V
MEASURED ONE-SUN PARAMETERS OF 4 CM2 B-BSF (N+PP+) SILICON
SOLAR CELLS PROCESSED IN THE WAFER 14.
JSC
Cell
VOC (mV)
FF
(%)
(mA/cm2)
14-A
596.1
36.6
0.71
15.4
14-B
594.8
37.3
0.60
13.4
14-C
591.6
36.4
0.71
15.3
14-D
591.8
35.4
0.73
15.4
14-E
594.5
36.7
0.71
15.4
Aver.
593.72.0
36.50.7
0,690.05
15.00.9
TABLE VI
VOC, JSC, FF AND EFFICIENCY OF THE LARGE AREA B-BSF (N+PP+)
SILICON SOLAR CELLS AND AVERAGE VALUES.
JSC
Cell
VOC (V)
FF
(%)
(mA/cm2)
A
601.7
30.8
0.72
13.4
B
599.8
30.9
0.69
12.9
C
602.3
31.4
0.72
13.7
Aver.
601.31.3
31.00.3
0.710.02
13.30.4
002286
1E+21
1E+20
1E+19
TABLE VII
AVERAGE VALUES OF VOC, JSC, FF AND OF B-EMITTER AND B-BSF
SILICON SOLAR CELLS WITH DIFFERENT AREAS.
1E+18
1E+17
B-emitter B-emitter
B-BSF
B-BSF
small cells large cells small cells large cells
1E+16
1E+15
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Depth (um)
+
5935
JSC
(mA/cm2)
35.40.5
31.40.5
36.50.7
FF
0.720.02
0.720.4
0.690.05 0.710.02
(%)
15.10.2
13.40.2
15.00.9
1E+21
13.30.4
1E+20
40
1E+19
1E+18
1E+17
1E+16
1E+15
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Depth (um)
Fig. 4.
VOC (mV)
Fig. 3.
30
20
10
& & & & & & & & & & & & & & & & & &
&
&
,, , , , , , , , , , , , , , , , , ,
&
,,
, &
Voc = 596.2 mV
,,
,, &
,,
Jsc = 31.0 mA/cm2
,,&
Voc = 588.7 mV
,,&
FF = 0.729
,,
2
Jsc = 35.5 mA/cm
,,&
= 13.4 %
,,
FF = 0.740
,,&
2
,,
61.58 cm
= 15.5 %
,&,
,,
,,
2
4 cm
&,
,,
,
&,,
,,
& ,,
0.1
0.2
0.3
0.4
0.5
p+nn+
0.6
0.7
Voltage (V)
30
20
10
n+pp+
&& & & & & & & & & & & & & & &
&&&&&
&&&
&&
&&
,, , , , , , , , , , , , , , , , ,
, , &&&
,,, &
,, &&
Voc = 602.3 mV
,, &
,, &
,,&
Jsc = 31.4 mA/cm2 Voc = 596.1 mV
,,&
,,&
FF = 0.724
2
Jsc = 36.6 mA/cm
,,&
= 13.7 %
,&,
FF = 0.707
,&,
2
&,
61.58 cm
= 15.4 %
&,,
&,
&,,
2
4 cm
&,,
&,,
&,
&,,
,
&&,
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Voltage (V)
002287
VI. CONCLUSIONS
+
[2]
+
[3]
[4]
[5]
[6]
[7]
[8]
002288
I. INTRODUCTION
Recent progress in III-V multijunction space solar cell
technology has led to production GaInP/GaAs/Ge triplejunction space cells with an average 1-sun efficiency close to
30% (AM0, 28C, 1-sun).
The latest lattice-matched
multijunction solar cell, XTJ, typical-size, 26.62-cm2, and
LEONE, 59.65 cm2, shown in Fig. 1 passed qualification per
AIAA S-111-2005 and AIAA S-112-2005 for CIC assembly
level qualification and manufacturing change readiness review.
Jsc[mA/cm2]
Voc [V]
Vload
Jload@2.31V [mA/cm2]
Jload @ 2.27V [mA/cm2]
Jmp [mA/cm2]
Vmp [V]
Eff @Pmp [%]
Eff@2.31V [%]
Eff@2.27V [%]
Fill Factor
Absorptance [Bare]
EOL]
Absorptance [CIC, 5 mil
AR]
Pm @1 MeV electron, 5E14
Pm @1 MeV electron, 1E15
(26.62cm2~32.3 cm2)
(53.3cm2~59.65cm2)
17.76
2.633
2.310
17.14
N/A
17.02
2.348
29.54%
29.26%
N/A
0.85
0.88
0.87
17.76
2.633
2.270
N/A
17.14
17.02
2.300
28.91%
N/A
28.76%
0.84
0.88
0.87
17.76
2.633
2.270
N/A
17.14
17.02
2.300
28.91%
N/A
28.76%
0.84
0.88
0.87
0.90
0.89
0.85
0.90
0.89
0.85
0.90
0.89
0.85
003146
gr
ow Ga
th As
su
bs
tra
te
gro
wt GaA
hs s
ub
str
ate
metal
gridline
metal
gridline
transparent buffer
transparent buffer
0.35
0.30
Current (A)
Area = 26 cm2
0.25
0.20
0.15
0.10
0.05
0.00
0.0
0.5
1.0
1.5
2.0
Voltage (V)
2.5
3.0
3.5
Fig. 5. Illuminated current-voltage characteristic of a largearea IMM3J cell tested under 1-sun, AM0 simulator
illumination.
Fig. 7 plots the illuminated current-voltage characteristic of
recent Spectrolab IMM 4-junction cells. One of the IMM4J
cells measured an VOC of 3.42 V, a JSC of 15.8 mA/cm2, a fill
factor of 0.82, or a 1-sun, AM0 efficiency of 32.7%. Again,
the solar simulator was calibrated using LEAR Jet-flown
003147
50
40
30
100
20
90
10
80
0
350
60
50
550
750
950
1150
1350
Wavelength (nm)
1550
1750
40
30
20
IV. CONCLUSION
10
0
350
550
750
950
1150
1350
Wavelength (nm)
EQE (%)
70
0.016
Spectrolab continues to develop high-efficiency latticematched and inverted metamorphic solar cell technology for
space and near-space applications. Large-area IMM3J space
solar cells with 1-sun, AM0 efficiency of 32% have been
demonstrated. In addition, IMM4J cells (1-cm2) have reached
a 1-sun, AM0 conversion efficiency of 33%. Improvements in
crystal growth, as well as more optimal subcell current density
balance among subcells, will further raise the 1-sun, AM0
efficiency of the IMM3J as well as the IMM4J cells to
efficiencies of 33% and 35%, respectively.
0.014
0.012
ACKNOWLEDGEMENT
0.010
0.008
0.006
0.004
0.002
0.000
0
Voltage(V)
Fig. 7. Illuminated current-voltage characteristic of IMM4J
cells (1-cm2) tested under 1-sun, AM0 simulator illumination
calibrated to LEAR-jet flown standards.
REFERENCES
[1] M. W. Wanlass, S. P. Ahrenkiel, R. K. Ahrenkiel, D. S. Albin,
J. J. Carapella, A. Duda, J. F. Geisz, S. Kurtz, T. Moriarty, R. J.
Wehrer, and B.Wernsman, Proceedings of the 31st IEEE
Photovoltaic Specialists Conference, p. 530 (2005).
[2] R. R. King, D. C. Law, C. M. Fetzer, R. A. Sherif, K. M.
Edmondson, S. Kurtz, G. S. Kinsey, H. L. Cotal, D. D. Krut, J.
H. Ermer, and N. H. Karam, Proc. 20th European Photovoltaic
Solar Energy Conference, p. 118 (2005).
[3] Yoon, Hojun; Haddad, Moran; Mesropian, Shoghig; Yen,
Jason; Edmondson, Kenneth; Law, Daniel; King, Richard R.;
Bhusari, Dhananjay; Boca, Andreea; Karam, Nasser H.,
003148
003149
340
AbstractA multijunction solar cell can extract higher solar energy compared to a single junction cell by splitting the solar spectrum. Although extensive research on solar cell efficiency enhancement is in place, limited research materials are available to identify the optimum interconnection of multijunction solar subcells
using power electronic circuits. Multijunction solar cells could be
grouped into two main categories: vertical multijunction (VMJ)
solar cells and lateral multijunction (LMJ) solar cells. In this paper,
a detailed study to identify the optimum interconnection method
for various multijunction solar cells has been conducted. The authors believe that the conducted research in this area is very limited, and an effective power electronic circuit could substantially
improve the efficiency and utilization of a photovoltaic (PV) power
system constructed from multijunction solar cells. A multiple input
dc-to-dc boost converter has been used to demonstrate the advantage of the proposed interconnection technique. In order to ensure
maximum power point (MPP) operation, a particle swarm optimization (PSO) algorithm has been applied needing only one MPP
control for multiple solar modules resulting in cost and complexity
reduction. The PSO algorithm has the potential to track the global
maxima of the system even under complex illumination situations.
A complete functional system with the implementation of the proposed algorithm has been presented in this paper with relevant experimental results.
Index TermsInterconnection, maximum power point tracking
(MPPT), multijunction solar cell.
I. INTRODUCTION
Manuscript received October 26, 2011; revised August 05, 2012; accepted
September 29, 2012. Date of publication November 16, 2012; date of current
version March 18, 2013.
The authors are with the University of Utah, Salt Lake City, UT
84112 USA (e-mail: khorshed.alam@utah.edu; faisal.khan@utah.edu;
as.imtiaz@utah.edu).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TSTE.2012.2223493
Fig. 1. (a) VMJ solar cell. (b) LMJ solar cell structure.
341
Fig. 2. (a) characteristics and (b) power curve of a GaAs subcell. (c)
characteristics and (d) power curve of a Si subcell. (e) Photograph of the actual
multijunction cell assembly (courtesy of Prof. R. Menon and his research group
at University of Utah).
342
343
TABLE I
SUBCELL MATERIAL, BAND GAP ENERGY, AND SUMMERY OF THE PROPOSED INTERCONNECTION
Fig. 5. (a) String and module formation; (b) interconnection of modules for
higher power output.
344
Fig. 6. Proposed interconnection: each module is connected to a separate input of the multiple input dc-to-dc converter.
(2)
where
345
Fig. 8. Simulation results: (a) variation of output power versus time; (b) variation of duty ratio of the dcdc converters versus time.
TABLE III
CONSTANTS FOR PSO ALGORITHM
346
Fig. 10. Experimental results: (a) variation of output power versus time;
(b) variation of duty ratio of the dcdc converters versus time.
Fig. 9. (a) DCDC converters with control circuits; (b) experimental setup.
delivered by four solar modules. The efficiency of the MPPT algorithm has not been considered here, and the reason has been
explained in the later part of this section. Moreover, the simulation results shown in Fig. 8 incorporate the MJ solar cell system
summarized in Table I, and the experimental results shown in
Fig. 10 incorporate the proof of concept modules built using
the commercially available crystalline solar cells. These results
have been summarized in Table IV. The measured efficiency
of the converter with this loading was 84.6% at
steady-state.
The multiple input dc-to-dc converter shown in Fig. 6 consists
of four single input dc-to-dc boost converters, and the output
terminals of these single input boost converters are connected in
parallel. The efficiency of the multiple input dc-to-dc converter
depends on the operating points of the individual single input
dc-to-dc converters. The overall efficiency can be written as
(3)
is the efficiency of the multiple input dc-to-dc
where
converter, and is the efficiency of the th single input dc-to-dc
boost converter. and
are the input power of the th single
input dc-to-dc converter and summation of the input power of
the individual single input dc-to-dc converters, respectively.
is the output power of the multi-input dc-to-dc converter.
is the number of single input dc-to-dc converters with paralleled output.
It is possible to design a dc-to-dc boost converter so that it
operates near its peak efficiency at the MPP of the connected
TABLE IV
MEASURED VOLTAGE AND CURRENT OF THE SOLAR MODULES
DURING THE EXPERIMENT
347
TABLE V
SUMMARY OF SERIES INTERCONNECTION OF ALL SUBCELLS
SUMMARY
Fig. 11. Efficiency versus load current curve of individual single input dc-to-dc
boost converters.
OF
TABLE VI
OPEN CIRCUIT VOLTAGE MATCHED INTERCONNECTION
OF ALL SUBCELLS
energy gain per year for a 1-kW unit. In the second phase, the
subcells constructed from the same material were connected in
such a way that the open circuit voltages of four modules match
the best [15], [25]. The resulted voltage and current ratings of
these modules are shown in Table VI, and these modules were
connected in parallel. The maximum power output of the system
from PSIM simulation was found to be 15.78 W, and therefore,
the proposed interconnection still provides 10.71% power gain.
This amount of efficiency enhancement is really significant in
solar power systems in critical applications such as in a battlefield of scientific expedition. The above comparisons are performed using PSIM simulations considering ideal dc-to-dc converters. In practice, the efficiency of the converter will affect the
power gain magnitudes.
IX. EFFECT OF PARTIAL SHADING
The effect of partial shading in an MJ solar cell system
might be different from multistring solar systems constructed
from single junction solar cells. The subcells of a mechanically stacked MJ solar cell or LMJ solar cell share the same
incident solar beam. Therefore, any partial shading will affect
all the subcells of the MJ cell under partial shading. As the
subcells are part of different modules, all the modules will be
affected. This phenomenon is illustrated in Fig. 12. Therefore,
partial shading will have more power degrading impact on the
MJ solar cell system discussed here compared to multistring
system of single-junction solar cells. However, most of the
concentrated PV (CPV) MJ solar cells system comes with a
concentrator system with homogenizer. The homogenizer is
used to improve flux uniformity, and it compensates for the
misalignment, reduces chromatic aberration, and compensates
for focusing errors [43]. The layout of a typical homogenizer
system has been shown in Fig. 13 which can potentially solve
the partial shading problem.
X. CONCLUSION
The efficiency of crystalline multijunction solar cells is much
higher than the single junction solar cells, and the use of MJ
348
ACKNOWLEDGMENT
The authors would like to acknowledge Prof. R. Menon of
University of Utah for his valuable suggestions regarding MJ
solar cells.
REFERENCES
Fig. 12. Effect of partial shading on proposed MJ solar cell system. Partial
shading will affect all the modules.
Fig. 13. MJ solar cell system with homogenizer in order to mitigate the effect
of partial shading.
solar cells is gradually increasing due to new material combinations, advancement in concentrator materials and optical splitting techniques, and progress in fabrication processes at the cost
of reasonably increased price. In this paper, a subcell interconnection technique for different crystalline MJ solar cells has
been presented. Due to higher $/watt, the application of these
MJ solar cells is limited to aerospace application and concentrated PV (CPV) systems [39], [40]. The ongoing research on
MJ solar cells is in place to enhance the efficiency at the cell
level. To the knowledge of the authors, there is no significant
work done on the optimal subcell interconnection of MJ solar
cells with efficient power electronic solution. The identification
of the optimum subcell interconnection for multijunction solar
cells using suitable power electronic circuit is the main consideration of this work. It has been shown that the proposed interconnection will result in achieving a completely matched subcell
interconnection system for different multijunction solar cells.
The possible partial shading scenarios have also been discussed
with tentative solutions.
[18] M. A. Green and A. Ho-Baillie, Forty three per cent composite splitspectrum concentrator solar cell efficiency, Prog. Photovoltaics: Res.
Applicat., vol. 18, pp. 4247, 2010.
[19] L. Zhao, High Efficiency Mechanically Stacked Multi-Junction
Solar Cells for Concentrator Photovoltaics, Ph. D. dissertation,
Katholieke Universiteit LeuvenFaculteit Ingenieurswetenschappen
Arenbergkasteel, Belgium, Mar. 2011.
[20] L. L. Kazmerski, Solar photovoltaics R&D at the tipping point: A
2005 technology overview, J. Electron Spectroscopy and Related Phenomena, vol. 150, pp. 105135, 2006.
[21] National Renewable Energy Laboratory News Release NR-1907 [Online]. Available: http://www.nrel.gov/news/press/2007/524.html
[22] J. D. M. Cambridge, M. A. Steiner, B. L. Unger, K. A. Emery, E. L.
Christensen, and M. W. Wanlass et al., Compact spectrum splitting
photovoltaic module with high efficiency, Prog. Photovoltaics: Res.
Applicat., vol. 19, no. 3, pp. 352360, May 2011.
[23] X. Wang and A. Barnett, One lateral spectrum splitting concentrator
photovoltaic architecture: Measurements of current assemblies and
analysis of pathways to 40% efficient modules, in Proc. 35th IEEE
Photovoltaic Specialists Conf. (PVSC), 2010, pp. 27452750.
[24] T. Esram and P. L. Chapman, Comparison of photovoltaic array maximum power point tracking techniques, IEEE Trans. Energy Convers.,
vol. 22, no. 2, pp. 439449, Jun. 2007.
[25] A. L. Lentine, G. N. Nielson, M. Okandan, W. C. Sweatt, J. L. CruzCampa, and V. Gupta, Optimal cell connections for improved shading,
reliability, and spectral performance of microsystem enabled photovoltaic (MEPV) modules, in Proc. 35th IEEE Photovoltaic Specialists
Conf. (PVSC), 2010, pp. 30483054.
[26] L. Gao, R. A. Dougal, S. Liu, and A. P. Iotova, Parallel-connected
solar PV system to address partial and rapidly fluctuating shadow conditions, IEEE Trans. Ind. Electron., vol. 56, no. 5, pp. 15481556,
May 2009.
[27] M. Calais, J. Myrzik, T. Spooner, and V. G. Agelidis, Inverters for
single-phase grid connected photovoltaic systemsAn overview, in
Proc. Power Electronics Specialists Conf., 2002, pp. 19952000.
[28] J. M. A. Myrzik and M. Calais, String and module integrated inverters
for single-phase grid connected photovoltaic systemsA review, in
Proc. Power Tech Conf., 2003, vol. 2, p. 8.
[29] M. Meinhardt, G. Cramer, B. Burger, and P. Zacharias, Multi-stringconverter with reduced specific costs and enhanced functionality,
Solar Energy, vol. 69 (Suppl.), no. 16, pp. 217227, 2000.
[30] S. A. Khajehoddin, A. Bakhshai, and P. Jain, A novel topology and
control strategy for maximum power point trackers and multi-string
grid-connected PV inverters, in Proc. APEC 2008, pp. 173178.
[31] M. Meinhardt and G. Cramer, Past, present and future of grid connected photovoltaic- and hybrid-power-systems, in Proc. IEEE-PES
Summer Meeting, 2000, vol. 2, pp. 12831288.
[32] P. A. Lynn, Electricity From Sunlight an Introduction to Photovoltaics,
1st ed. Hoboken, NJ: Wiley, 2010.
[33] L. Gao, R. A. Dougal, S. Liu, and A. P. Iotova, Parallel-connected
solar PV system to address partial and rapidly fluctuating shadow conditions, IEEE Trans. Ind. Electron., vol. 56, no. 5, pp. 1548156, May
2009.
[34] L. Gao, R. A. Dougal, S. Liu, and A. Jotova, Portable solar systems
using a step-up power converter with a fast-speed MPPT and a parallelconfigured solar panel to address rapidly changing illumination, in
Proc. APEC 2007, pp. 520523.
[35] N. E. Ball, Parallel and Virtual Parallel Interconnection of Solar Cells
in Solar Panel, U.S. Patent 7521630, Apr. 21, 2009.
[36] M. Miyatake, M. Veerachary, F. Toriumi, N. Fujii, and H. Ko, Maximum power point tracking of multiple photovoltaic arrays: A PSO
approach, IEEE Trans. Aerosp. Electron. Syst., vol. 47, no. 1, pp.
367380, Jan. 2011.
[37] B. Mitchell, G. Peharz, G. Siefer, and M. Peters et al., Four-junction spectral beam-splitting photovoltaic receiver with high optical efficiency, Prog. Photovoltaics: Res. Applicat., vol. 19, no. 1, pp. 6172,
2011.
[38] N. E. Ball, Parallel and Virtual Parallel Interconnection of Solar Cells
in Solar Panels, U.S. Patent 7,521,630 B2, Apr. 2009.
[39] D. Crisp, A. Pathareb, and R. C. Ewell, The performance of gallium
arsenide/germanium solar cells at the Martian surface, Acta Astronautica, vol. 54, no. 2, pp. 83101, Jan. 2004.
349
548
I. INTRODUCTION
URRENTLY, crystalline Si solar cells dominate the photovoltaic market. However, high material and processing
costs are still the most significant drawbacks of these systems.
To reduce the cost and improve the overall efficiency of the
cell, new-design-based thinner silicon wafers with improved
light-trapping structures should be used. Because Si is an indirect bandgap material, when the wafer thickness is less than
200 m, Si wafers are not capable of absorbing the entire solar
spectrum. To overcome this problem and improve the efficiency
of the cell, absorption throughout the cell must be improved by
increasing the path length of the light and the number of scattering events of the incident photons by trapping. Surface texturing
and antireflection (AR) coatings have been traditionally used for
such purposes [1], [2].
In recent years, nanostructures such as Si nanowires have attracted considerable attention due to their unique properties and
functionalities. Si nanowires can be vertically aligned and arranged in integrated arrays on the solar cell surface, which leads
to remarkable light-trapping strength with reflectivity values
of less than 1% around 500 nm [3]. The light-trapping ability of Si nanowires is particularly superior in the blueviolet
part of the spectrum compared with conventional pyramidal
textures [3]. Several different methods for the synthesis of Si
nanowire arrays have been employed, such as vaporliquid
solid [4], laser ablation [5], direct reactive ion etching [6], and
metal-assisted etching (MAE) [7][9]. Among these, MAE is
the most appealing approach, because it can be conducted at
room temperature and does not require expensive production
equipment. Thus, the nanowire formation via MAE contributes
little to the overall processing costs. Moreover, contrary to the
conventional nanowire fabrication methods, doping is not an issue in the MAE method. It is because the nanowires are etched
down from bulk Si, where the doping level and growth direction
of the fabricated Si nanowires would be the same as the starting wafer. Thus, cost-effective Si nanowire fabrication methods
that enable large-scale and large-area production in a controlled
manner could be adapted to the industrial-scale fabrication of solar cells. Some reports focusing on the photovoltaic properties
of Si nanowire solar cells have already been published [10]
[13]. However, realization of a full-scale solar cell has not been
demonstrated yet. In addition to the light-trapping applications,
Si nanowires can be applied to several different types of innovative devices, including radial p-n junctions [14]. In a recent
work, we have already demonstrated a remarkable improvement
in the photovoltaic conversion efficiency in the heterojunction
solar cells employing nanowires on the surface [15].
In this study, we fabricated industrial-size (156 mm
156 mm, with an area of 239 cm2 ) monocrystalline Si solar
cells with vertically aligned nanowire arrays formed on their
surface by MAE. To the best of our knowledge, this is the first
demonstration of the utilization of nanowires in a solar cell of
commercial size in this scale. The effect of the Si nanowire
length on the solar cell characteristics was worked out in a systematic way. The light-trapping and antireflective properties of
the Si solar cells were investigated and compared with that of the
conventional texturing technique. External quantum efficiency
(EQE) measurements were also conducted for comparison. The
technical procedures presented herein are fully compatible with
the present monocrystalline Si solar cell production lines.
KULAKCI et al.: APPLICATION OF SI NANOWIRES FABRICATED BY METAL-ASSISTED ETCHING TO CRYSTALLINE SI SOLAR CELLS
549
Fig. 1. Cross-sectional SEM images of Si nanowires on mono-c cells fabricated at an etching time of (a) 10 min, (b) 20 min, (c) 30 min, (d) 40 min, and
(e) 50 min. (f) Nearly linear relationship was observed between the etching time
and nanowire length.
550
Fig. 3. Images of fully processed solar cells. A significant color change was
observed between the conventionally textured cells (a) and the nanowiretextured cells etched for 50 min (b) due to enhanced absorption at wavelengths
less than 450 nm.
and after Si3 N4 deposition are shown in Fig. 2(a) and (b),
respectively.
As the nanowire length increased, the reflectivity of the wafers
decreased accordingly. When an AR coating was not present on
the surface, all of the nanowire-coated wafers exhibited lower
optical reflection than the standard pyramid-textured wafer, except for the wafer fabricated at an etching time of 10 min.
However, following AR deposition, the pyramid-textured wafer
revealed lowest reflectance for the wavelengths greater than
450 nm. In contrast, nanowire-textured wafers did not show
significant differences following AR deposition, except for the
nanowires fabricated with an etching time of 10 min. Therefore, the use of Si3 N4 layer on Si nanowires can have a passivation effect without any significant AR property. For the
wavelengths lower than 450 nm, the samples with nanowires
exhibited low reflectivity values and excellent light-trapping
properties. This feature was evidenced in the color of the fully
processed monocrystalline Si solar cells, as shown in Fig. 3. The
surface of cells was black due to the presence of nanowires.
C. Solar Cell Device Performance
The sample with the shortest nanowire arrays (formed with
an etching time of 10 min) revealed similar performance with
the pyramid-textured device. Although further optimization is
Fig. 4. IV characteristics of different monocrystalline samples under standard testing conditions (AM1.5G, 100 W/cm2 , 25 C).
needed to improve the performance, our results prove the successful implementation of Si nanowires to standard solar cells
at an industrial scale.
Currentvoltage (IV) characteristics of the solar cells are
provided in Fig. 4. The solar cell device parameters, such as the
photovoltaic conversion efficiency, fill factor (FF), short-circuit
current density Jsc , series resistance, and open-circuit voltage
Vo c extracted from the IV curves, are shown in Fig. 5.
Jsc and Vo c decreased with an increase in the nanowire length,
leading to a reduction in the overall power conversion efficiency.
These results can be attributed to an increase in surface recombination along the nanowire surfaces. Due to MAE etching, broken/dangling Si bonds were generated on the nanowire surfaces,
which act as trap energy levels for photogenerated carriers. The
presence of broken Si bonds becomes more of an issue for longer
nanowires. It is because the number of recombination sites increases with the nanowire length with an overall increase in the
total surface area.
In order to get more insight into the current mechanisms, we
have measured dark IV characteristics of the solar cells, not
shown here. We have determined the ideality factors and the
saturation currents and seen that the ideality factor of the cells
is relatively good for the samples with pyramid texture (ideality
factor n of 2.03 and saturation current Jo of 7 107 A/cm2 )
KULAKCI et al.: APPLICATION OF SI NANOWIRES FABRICATED BY METAL-ASSISTED ETCHING TO CRYSTALLINE SI SOLAR CELLS
551
Fig. 5. Effect of nanowire length on various solar cell parameters. The pyramid textured solar cell parameters are inserted as a reference (red horizontal
lines) on the efficiency and Jsc graphs.
552
[1] M. A. Green, Z. Jianhua, A. Wang, and S. R. Wenham, Very high efficiency silicon solar cells-science and technology, IEEE Trans. Electron
Devices, vol. 46, no. 10, pp. 19401947, Oct. 1999.
[2] J. Zhao, A. Wang, and M. A Green, 245% efficiency silicon PERT cells
on MCZ substrates and 247% efficiency PERL cells on FZ substrates,
Progr. Photovoltaics: Res. Appl., vol. 7, pp. 471474, 1999.
[3] B. Ozdemir, M. Kulakci, R. Turan, and H. E. Unalan, Effect of electroless etching parameters on the growth and reflection properties of silicon
nanowires, Nanotechnol., vol. 22, pp. 155606-1155606-7, 2011.
KULAKCI et al.: APPLICATION OF SI NANOWIRES FABRICATED BY METAL-ASSISTED ETCHING TO CRYSTALLINE SI SOLAR CELLS
553
J. A. Rodrguez, M. Vetter
I. I NTRODUCTION
In recent years, the interest of hydrogenated amorphous
silicon (aSi:H) tandem solar cells (aSi:H/aSi:H) has increased since these devices present reduced the Staebler
Wronski Effect (SWE) regarding to the aSi:H singlejunction
(SJ) device [1]. Tandem solar cells provide better stability and
increased efficiency than aSi:H SJ solar cell [2] [3]. The
tunneljunction, which acts as joint in the tandem structure
between the top and the bottom aSi:H cells in tandem
structure, is an important design factor of tandem solar cells
because of its impact on the solar cell efficiency. Different
materials have been investigated in literature with the objective
to improve recombination rates in the the tunnel junction and
to reduce the light absorption in the doped layers [4] [5] [6].
The structure of the tandem solar cell analysed is basically
composed by a stack of two pin junction solar cells based
on aSi:H. The structure of this typical aSi:H solar cell with
a pin superstate configuration is depicted in Fig. 1. It is a
consecutive deposition of a p type layer (player), an intrinsic
type layer (ilayer) and n type layer (nlayer), on a float glass
with a transparent conductive oxide (TCO). The TCO layer
Fig. 1. Typically structure of a pin aSi:H SJ solar cell with flat interfaces.
acts as a front contact and the Aldoped zinc oxide (AZO) and
aluminium (Al) metal layers, which are deposited on the pin
structure, form the back contact of the solar cell. In this work,
we will study the current densityvoltage (JV) characteristic
curve of aSi:H/aSi:H tandem solar cell with a heavy doped
tunnelling junction of hydrogenated microcrystalline silicon
(c-Si:H) n+ /p+ layers.
373
II. P HYSICAL
B. Simulation
The simulated structure of the DJ solar cell analysed in
this paper is a stack of two pin junction aSi:H solar cells
with flat interfaces such as is shown in Fig. 4. Between the
top and the bottom solar cells a tunneljunction is included
with heavy doped cSi:H p+ /n+ layers. Table I shows the
baseline input parameters of these two pin top and bottom
cells which form the structure of aSi:H/aSi:H tandem solar
cell [11]. The behaviour of the JV characteristic curves
for the tandem structure and also for the top and bottom
cells was obtained from a coupled solution of Poisson and
continuity equations. The performed simulation considers that
the photocurrent of the cells is only generated by the active
layers and it assumes that photons which are absorbed only
generate only one electronhole pair. The optical model to
simulate the the illumination of the solar cell is based on the
transfer matrix method (TMM) using a wavelength range from
300nm to 800nm under the standard AM 1.5 spectrum [12]. A
nonlocal bandtoband tunneling model is used to simulate
the tunneljunction in DJ solar cell.
374
TABLE I
I NPUT PARAMETERS USED IN THE SIMULATION FOR P I N TOP AND
BOTTOM CELLS WHICH FORM THE DJ SOLAR CELL .
player
ilayer
nlayer
3 1018
1015
8 1018
Relative permittivity
7.2
11.9
11.9
3.90
4.00
3.99
1.95
1.78
1.80
20
20
20
Bulk material
Doping (cm3 )
20
10
20
10
1020
1020
1020
1020
0.144
0.144
0.144
+/0
18
8 10
16
5 10
2 1019
-0.70
-0.89
-1.40
0.20
0.20
0.20
CB mobility edge
2 1021
8 1021
1021
VB mobility edge
1021
4 1021
2 1021
0.180
0.032
0.070
0.090
0.047
0.160
Tail states
DOS (cm3 /eV) at:
electron/hole in CB tail
1017 /1015
electron/hole in VB tail
1015 /1017
III. R ESULTS
The aim of our computer simulation is to obtain the behaviour of the JV characteristic curve, under the standard
AM1.5 sunlight sprectrum, for the aSi:H/aSi:H DJ solar cell
presented in Fig. 4.
The schematic of the band diagram of this tandem solar cell
under the AM 1.5 spectrum generated by the simulation tool
when the applied voltage is 0V, including the cSi:H tunnel
junction simulated with a heavy doped p+ /n+ layers is shown
in Fig. 5. In Fig. 6 the JV characteristic curve for the top
and bottom cells is presented. The Jsc and Voc values for the
top cell are higher than the bottom cell since in this example
the layer properties are not the optimized ones. The electronic
properties of the bottom solar cell play in this example the
critical role in the global structure of the DJ solar cell because
e.g. the Jsc of the tandem solar cell is currentlimited by this
bottom cell. In Fig. 7 the JV characteristic curve of DJ solar
cell is depicted. This figure shows that the Jsc value of the
tandem solar cell is limited by the bottom cell. Nevertheless,
Fig. 5. Band diagram of tandem solar cell under the AM 1.5 spectrum when
the applied voltage is 0V.
375
Fig. 7.
the Voc of the DJ solar cell is nearly the sum of the Voc value
of the top and bottom solar cells.
IV. C ONCLUSION
In this paper the behaviour of the JV characteristic curves
of the aSi:H/aSi:H DJ solar cell and also the top and bottom
cells of the stack individually are analysed. The top cell of the
here analysed layer stack with 5080nm thickness of the top
cell and 200300nm thickness of the bottom cell presents a
higher value of Jsc than the bottom cell value. For this reason,
the Jsc value of the bottom cell works as a currentlimited
value of the tandem solar cell. The missmatching in the Jsc
value between top and bottom solar cells may be improved
changing the relation between the thickness of the active layers
of the tandem solar cell. The Voc value of the DJ solar cell is
nearly the sum of the Voc value of the top and bottom cells
376
Deen Md Abdullah
and Wali Md Abdullah
AbstractThis research paper carries out a realistic experimental approach to enhance the solar output power to a
significant level with the application of solar concentrator and
tracking system and also proposes a fuzzy based control system
for the utilization of the duration for which solar power is
available. By using only a solar panel, the theoretical value is
supposed to be around 40% more. The method is arranged
using Double Sun Technologyand Solar Tracker. In addition
to Double Sun Technology, four flat mirrors have been used
to increase the output power. A remarkable enhancement in the
output power is achieved with this method. Rather than using
only a solar panel, the output power increases by 58.32% by
using a photovoltaic (PV) panel with four flat mirrors with solar
tracker system. The concentrators are designed to collect the
incident light from the sun and concentrate them on the PV panel
so that the solar irradiance is more. The experiment is conducted
in four different setups. In the first setup, the output power is
calculated without the use of Double Sun and solar tracker. In the
second setup, power is measured with Double Sun but without
solar tracker. Both Double Sun and solar tracker are used in
the third setup. Then four flat mirrors and solar tracker are
considered. Output power has been calculated using open circuit
voltage and short circuit current. Finally, the proposed system
improves the total output by maximum utilization of available
photovoltaic energy and maximum utilization of the duration for
which solar power is available based on fuzzy logic.
Index TermsPhotovoltaic (PV) panel, passive device, Double
Sun Technology, solar tracker, concentrator, fuzzy logic.
I. I NTRODUCTION
Power generation is going to take a complete new shape in
twenty first century. Fossil fuels like coal, gas, petroleum are
not going to be plenty for electricity generation. Renewable
energy presents a better alternative to fossil fuels with greener
impact on environment. Though hydroelectricity offers a potential amount of renewable energy to the world, yet it is not
available to all the places of the globe. Solar energy on the
other hand has all the opportunity to take over the existing
power generation system [1]. Solar energy can be used for
small industries and in a very small scale for large industries;
like garment industries where the solar energy can be used as
secondary source for lighting and ventilation, reducing power
consumption from the main power line. But the output power
of photovoltaic (PV) panel is considerably low [2]. Extensive
research shows that output power can be increased in either of
two ways - by fabrication or by using passive devices [3], [4],
[5], [6], [7]. As fabrication is costly and hard, passive devices
are widely used to increase the efficiency as well as fill factor.
Concentrators with two flat mirrors and solar tracker are used
to enhance solar power. Concentration with two flat mirrors is
known as Double Sun Technology.
Fuzzy control is a methodology for constructing control
system in which the mapping between real-valued input and
output parameters is represented by fuzzy rules. Fuzzy control
has been very successful in commercial products such as
automatic transmission, video cameras etc.
In this paper, experimental results have been given which
shows the enhancement of output power by using concentrator
and solar tracker with a solar panel. The experiment is
conducted in four different setups. In the first setup, the output
power is calculated without the use of Double Sun and solar
tracker. In the second setup, power is measured with Double
Sun but without solar tracker. Both Double Sun and solar
tracker are used in the third setup. And finally four flat mirrors
and solar tracker are considered. An increase of 33.50% solar
power is found with the Double Sun and solar tracker, the
theoretical value of which is 40% [8]. 58.32% increase in
solar energy is observed by applying four fat mirrors with
solar tracker. Then we have proposed a scheme as [9], [10]
based on the use of conventional DC-DC converter, where
464
Fig. 1.
C=
Aef f
Amirrors + Amodule
Amirrors
=
=1+
Amodule
Amodule
Amodule
Fig. 2.
Here, Aef f is the effective area (the area seen by the sun),
Amirrors is the area of the mirrors that is seen by the sun and
Amodule is the total area of the modules. In this experiment,
for Double Sun Technology
465
Fig. 3.
Fig. 4.
Sun and with solar tracker. Open circuit voltages and short
circuit currents are measured for seven hour period starting
from 9 : 00 am to 4 : 00 pm with half hour interval and
power is calculated from the data. Maximum power is achieved
when the sun ray is at right angle to the solar panel surface.
By implementing fuzzy control system we can ensure that the
sun ray will be at right angle to the solar panel surface during
our defined period.
A. Without Concentrator and without Tracker
The flat solar panel has been exposed to the sun. No solar
tracker and no mirror is used in this stage. The solar panel is
inclined at a fixed position for the entire time period. Fig. 4
shows the variation of the experimentally measured power with
time. Data are collected for three different days of similar
weather condition.
B. With Double-Sun but without Solar Tracker
In this setup, data are taken with the Double sun Technique,
i.e. using two side mirrors, but no tracking has been done.
The power versus time graphs for the same days are shown in
Fig. 5.
C. With Double-Sun and with Solar Tracker
In third setup, open circuit voltages and currents are measured, with the Double Sun Technique and using solar tracker.
The power versus time graphs are shown in Fig. 6.
D. With Four Sun and with Solar Tracker
V. DATA A NALYSIS
466
Fig. 5. Solar output power with Double Sun concentrator but without solar
tracker.
Fig. 6.
tracker.
Solar output power with Double Sun concentrator and with solar
Fig. 7. Solar output power with Four Sun technology and with solar tracker.
Fig. 8.
VI. C ONCLUSION
The authors would gratefully like to acknowledge the endless support by Military Institute of Science and Technology
(MIST), Dhaka, Bangladesh for supporting this research-work.
This paper has shown the experimental results of the performance of the Double Sun Concentration and Four Sun Concentration PV systems with geometric concentration factors
of 1.8x and 3.98x, respectively, using conventional silicon
modules and no active cooling. The result for the Double
Sun Technology with solar tracker is increased by 33.50%.
It also demonstrates that the use of four flat mirrors and
ACKNOWLEDGMENT
R EFERENCES
[1] F. Kreith and D. Y. Goswami, Handbook of Energy Efficiency and
Renewable Energy. CRC Press, Taylor & Francis Group, 2007.
[2] A. McEvoy, T. Markvart, and L. Castaner, Practical Handbook of
Photovoltaic: Fundamentals and Applications. Academic Press, 2011.
467
TABLE I
COMPARISON OF THE FOUR EXPERIMENTAL SETUPS
Time
(Hour)
9:00 AM
9:30 AM
10:00 AM
10:30 AM
11:00 AM
11:30 AM
12:00 PM
12:30 PM
1:00 PM
1:30 PM
2:00 PM
2:30 PM
3:00 PM
3:30 PM
4:00 PM
Pavg
Percent
Increase
2.14%
33.50%
[3] M. Brogren, Optical efficiency of low-concentrating solar energy systems with parabolic reflectors, Doctoral thesis, 2004.
[4] R. M. Swanson, The promise of concentrators, Progress in Photovoltaics: Research and Applications, vol. 8, no. 1, pp. 93111, 2000.
[5] A. Luque and S. Hegedus, Handbook of Photovoltaic Science and
Engineering. John Wiley & Sons, 2011.
[6] R. B. Pettit and E. P. Roth, Solar Mirror Materials: Their Properties
and Uses in Solar Concentrating Collectors. Solar Materials Science,
1980.
[7] A. Luque, G. Sala, and I. L. Heredia, Photovoltaic concentration at
the onset of its commercial deployment, Progress in Photovoltaics:
Research and Applications, vol. 14, no. 5, pp. 413428, 2006.
[8] F. Reis, V. Corregidor, M. C. Brito, R. Rodrigues, J. Wemans, and
G. Sorasio, Power generation and energy yield using doublesunr
photovoltaic solar concentration, EUPVSEC, pp. 803806, 2009.
[9] S. A. Khan and M. I. Hossain, Design and implementation of microcontroller based fuzzy logic control for maximum power point tracking
of a photovoltaic system, 6th ICECE, Dhaka, Bangladesh, 2010.
[10] M. S. Hussain, D. M. Abdullah, W. M. Abdullah, and L. Ali, Development of a fuzzy logic based smart solar system for irrigation, 9th
ICME, Dhaka, Bangladesh, pp. RT050, 2011.
[11] S. P. Sukhatme and J. K. Nayak, Solar Energy: Principles of Thermal
Collection and Storage. Tata McGraw-Hill Education, 2008.
[12] F. Reis, Development of solar concentrators for large photovoltaic
plants, M. Sc. thesis, IST, Personal communication, 2008.
58.32%
468
Prasad Motepalli
Professor, ECE-Department,
BVRIT,
Narsapur, Medak (Dist), A.P, India
E-mail: sanjay.dubey@bvrit.ac.in
maximum
power
to
load
from
PV
panel
MPPT
is
for
designing
Solar
DC
pump
of
required
rated
and
DC
pump
motor
are
demonstrated
using
MATLAB/Simulink model.
I.
INTRODUCTION
429
(1000kg/m3*25m3/day)*9.8m/s2*16.008m.
1089.43 watt-hour/day.
Step 4: Solar radiation data
6hours/day. (Peak
=1000wattim2).
Step 5: Size and number of solar PV modules required are ...
Total wattage of PV panel = Total hydraulic energy/no.
of
hours.
= 1089.43/6 =18l.57 watt.
System losses = Total PV panel wattage/ (pump efficiency *
mismatch factor).
=181.57 watt/ (0.3*0.85) = 712.03
Considering operating factor = Total PV panel wattage after
losses/operating factor.
PV panel capacity = 712.03/0.75 =949.38watt.
Number of PV panels required = 949.38/240 4.
1 HP DC motor can be run with a solar capacity of 960Wp
with the available 240Wp model solar panel.
( )
Rs = _
dV
d1
nKT I q
KT
lo eQ(v+1Rs)ln
-1]
(6)
\ I
III.
Q(v+fRs)lnKT
1= Isc- lo e
o IT = 1SC ITref
Isc IT = IsQTref
[1
-1]
IGo
( J
Tref
:
150
(1)
100
50
31
n
e
+ a T - Tref
+IRs )lnKT
1= Isc_1o eQ(v
(qEg)(
(2)
(3)
-1]- V
(4)
Rp
Quality factor, N:
(5)
60
80
8.42
1000
W/m'2
+ IRs
40
100
)]
20
1.5
Fig 5.parameters
430
Simulation
of
Boost
converter
in
Multisim
environment
C.
IV.
CONCLUSIONS
Geoff Walker
"Evaluating
MPPT
Converter
topologies
using
[3]
1.\t 1k"". ,,,
.J.
:- EA2ICII I R.IA'ItL-.rJCf'..! ",n.n,,"
.
http://www.solarsemiconductor.com/asialssi-m6.pdf
[5]
N.Mohan,
TM.Undeland,
and
WP.Robbins,
"Power
Electronics
Converters, Applications and Design," John Wiley & Sons, Inc., NY,
D.
nd
2003
[7]
Reddy,"MPPT
Photovoltaic
module
Algorithm
using
Implementation
Microcontroller",
India
for
Solar
Conference
Mukesh Kumar
Gupta, Rohit
Jain, "MPPT
Simulation with DC
431
1298
I. INTRODUCTION
OT-CARRIER [1], intermediate-band [2], impactionization [3], and multiple-exciton-generation solar cells
[4] are novel solar cells proposed to exceed the efficiency limit
of single-gap solar cells [5]. Their fundamental models have
been reviewed several times and, each time, our understanding
of their operation improves or their practical implementation
becomes more feasible [6][12]. However, a unified treatment
of all these cells at a fundamental level is lacking, and this paper
is intended to fill this gap. Electrochemical potentials (or quasiFermi levels) will serve as the basis for this unified treatment.
A single-gap solar cell [see Fig. 1(a)] absorbs photons with
energy higher than the semiconductor bandgap EG . Its efficiency is limited, on one hand, by the fact that photons with
energy lower than the bandgap are not absorbed. On the other
hand, photogenerated electrons quickly relax to the bottom of
the bands by giving away their energy to phonons. The output voltage of the single-gap solar cell V, equals the electron
and hole quasi-Fermi level split (e,C e,V ) divided by the
Manuscript received May 2, 2013; revised June 14, 2013; accepted July 18,
2013. Date of publication August 6, 2013; date of current version September
18, 2013. This work was supported by the European Commission through the
NGCPV Project under Grant 283798.
The authors are with the Instituto de Energa Solar, ETSI Telecomunicacion, Universidad Politecnica de Madrid, Madrid 28040, Spain (e-mail:
amarti@etsit.upm.es; a.luque@upm.es).
Digital Object Identifier 10.1109/JPHOTOV.2013.2274381
electron charge. The cell emits photons whose chemical potential also equals this split. Throughout this paper, eV will be the
electron charge multiplied by the output voltage of the cell.
To fight the energy losses of the single-gap solar cell, the hotcarrier solar cell (HCSC) proposes that photogenerated carriers
are extracted from the cell before they recombine [see Fig. 1(b)].
This leads to a model in which, contrary to the intuition gained
from the operation of single-gap solar cells, there is no electron
and hole quasi-Fermi level split in the material that absorbs the
light at the time the cell is still capable of producing output
voltage.
Aiming also to optimize the use of the energy provided by the
absorption of high-energy photons, the impact-ionization solar
cell (IISC), and also the so-called multiple-excitation-generation
solar cell (MEGSC), pursues that the absorption of a high-energy
photon creates more than one electronhole pair [see Fig. 1(c)].
MARTI AND LUQUE: ELECTROCHEMICAL POTENTIALS (QUASI-FERMI LEVELS) AND THE OPERATION
1299
(a)
(b)
(c)
(d)
(e)
Fig. 2. Illustration of different electronelectron interactions. Two electrons
interacting within (a) the CB and (b) the VB. (c) Auger recombination and
(d) generation (also called impact ionization). (e) Auger thermalization.
(1)
(2)
where i represents the electrochemical potential or quasiFermi level of an electron at energy state i that can be degenerated without loss of generality in our argumentation. Energy
conservation implies that
1 + 2 = 3 + 4 .
(3)
(4)
1300
fFD =
exp
1
i a i b
kTc
=
+1
exp
1
i e
kTh
(5)
+1
where
Th = Tc /(1 a)
(6)
e = b/(1 a).
(7)
These equations where first deduced by Wurfel [6] and imply that our electron distribution can be visualized, instead of
as a distribution at temperature Tc and an energy-dependent
electrochemical potential, as a distribution at an equivalent
(hot) temperature Th and an energy-independent electrochemical potential e . Notice that, under this constant-hot-electrontemperature description, the difference in the electrochemical
potential between electrons in the CB and VB would be zero
because e is the same for electrons in both bands. The basis
of this conclusion is the fact that electrons in the CB have been
allowed to interact with electrons in the VB. This interaction
can consist of not only Auger generation recombination [see
Fig. 2(c) and (d)] but also Auger thermalization [see Fig. 2(e)].
To make a solar cell, our core material must interact with photons. This interaction takes place through radiative generationrecombination processes and implies that a photon can be absorbed by pumping an electron from the VB to the CB (generation) and also that an electron can recombine with a hole by
emitting a photon (luminescent photon). Because this photon
can be reabsorbed again, luminescent photons and electron
hole pairs stay in equilibrium through the reaction
ei,VB + hv ei,CB
(8)
(9)
(10)
where, given the low value of the phonon energy, ei and ej both
represent, for example, electrons in the CB. We arrive at
aC i + bC + phonon = aC j + bC aC = 0
(11)
which implies [based on (6)] that electrons are not hot anymore
but at the temperature Tc , which can now be interpreted as the
lattice temperature or room temperature. Similar equations
would hold for electrons in the VB and leading to aV = 0. For
Fig. 3.
F
=
F = 1
h e . (13)
kTh
kTa
Th
Th
The output voltage of the cell is then given by
1
1 +
Ta
F
=
V =
(
e h ) .
1
F
e
e
Th
(14)
+
F F is not equal to the difference of the electrons and hole
electrochemical potentials in the core material, which is zero.
Elaborating on this model, it is possible to demonstrate that
the efficiency limit for solar energy conversion for this device
is 86.5% (assuming that the sun is at 6000 K and the cell at
300 K) and is obtained for a bandgap EG = 0 [6], [15]. This
MARTI AND LUQUE: ELECTROCHEMICAL POTENTIALS (QUASI-FERMI LEVELS) AND THE OPERATION
(15)
i,V = aV i + bV
(16)
(17)
Th,C = Tc /(1 aC ).
(18)
Similarly, electrons in the VB can be described by an energyindependent electrochemical potential and hot-carrier temperature given by
e,V = bV /(1 aV )
(19)
Th,V = Tc /(1 aV ).
(20)
1301
(21)
1302
(a)
(b)
(22)
e,V = bV .
(23)
(25)
This result was obtained in [3] and [18] using other thermodynamic arguments. Obtaining the result in (25) from (24)
provides new insight because (24) is conceptually different from
MARTI AND LUQUE: ELECTROCHEMICAL POTENTIALS (QUASI-FERMI LEVELS) AND THE OPERATION
(a)
1303
(b)
Fig. 6. Summary of the novel cells discussed in this study and how electron,
hole, and photon electrochemical potentials are distributed: IBSC, S&Q model,
IISC/MEGSC, and HCSC.
imposing different restrictions on this electronelectron interaction, assuming energy-selective contacts or electron-selective
contacts and turning ON and OFF different electronphonon
and electronphoton interactions. Throughout this paper, we
have explained the mechanisms by which electrons in each band
can be described by a single electrochemical potential, whether
electrons can be considered hot and whether the photon electrochemical potential equals eV. The discussion presented in this
paper is summarized in the form of a tree in Fig. 6. The trunk
is the core photovoltaic material and the branches are various
novel solar cells that stem from the hypotheses made.
When selective energy contacts are attached to the core
material and electronacoustic phonon interaction is neglected,
we obtain the HCSC. Still, there are two models for this cell:
the impact-ionization model [6] and the particle conservation
model [7]. In the first, electrons and holes share the same electrochemical potential; in the second, they do not. The chemical
potential of the photons emitted by the cell is equal to the electron and hole electrochemical potential difference; therefore,
this chemical potential is equal to zero in the impact-ionization
model and, in general, different from zero according to the
1304
[6] P. Wurfel, Solar energy conversion with hot electrons from impact ionisation, Solar Energy Mater. Solar Cells, vol. 46, pp. 4352, Apr. 1997.
[7] P. Wurfel, A. S. Brown, T. E. Humphrey, and M. A. Green, Particle conservation in the hot-carrier solar cell, Progr. Photovoltaics: Res. Appl.,
vol. 13, pp. 277285, 2005.
[8] G. J. Conibeer, D. Konig, M. A. Green, and J. F. Guillemoles, Slowing
of carrier cooling in hot carrier solar cells, Thin Solid Films, vol. 516,
pp. 69486953, 2008.
[9] A. Luque and A. Mart, A metallic intermediate band high efficiency
solar cell, Progr. Photovoltaics: Res. Appl., vol. 9, pp. 7386, 2001.
[10] R. Strandberg and T. W. Reenaas, Limiting efficiency of intermediate
band solar cells with spectrally selective reflectors, Appl. Phys. Lett.,
vol. 97, pp. 031910-1031910-3, 2010.
[11] R. J. Ellingson, M. C. Beard, J. C. Johnson, P. R. Yu, O. I. Micic,
A. J. Nozik, A. Shabaev, and A. L. Efros, Highly efficient multiple exciton generation in colloidal PbSe and PbS quantum dots, Nano Lett.,
vol. 5, pp. 865871, May 2005.
[12] O. E. Semonin, J. M. Luther, S. Choi, H.-Y. Chen, J. Gao, A. J. Nozik, and
M. C. Beard, Peak external photocurrent quantum efficiency exceeding
100% via MEG in a quantum dot solar cell, Science, vol. 334, pp. 1530
1533, Dec. 16, 2011.
[13] D. Kondepudi and I. Prigogine, Modern Thermodynamics. Chichester,
U.K.: Wiley, 1998, p. 87.
[14] M. l. Bellac, F. Mortessagne, and G. G. Batrouni, Equilibrium and
Non-Equilibrium Thermodynamics. Cambridge, U.K.: Cambridge Univ.
Press, 2004, p. 336.
[15] A. Luque and A. Mart, Theoretical limits of photovoltaic energy conversion and new-generation solar cells, in Handbook of Photovoltaic
Science and Engineering, A. Luque and S. Hegedus, Eds. Chichester,
U.K.: Wiley, 2011.
[16] G. J. Conibeer, J. F. Guillemoles, and M. A. Green, Phononic band gap
engineering for hot carrier solar cell absorbers, in Proc. 20th Eur. Photovoltaic Solar Energy Conf., 2005, pp. 3538.
[17] R. Brendel, J. H. Werner, and H. J. Queisser, Thermodynamic efficiency
limits for semiconductor solar cells with carrier multiplication, Solar
Energy Mater. Solar Cells, vol. 4142, pp. 419425, Jun. 1996.
[18] A. Luque and A. Mart, Entropy production in photovoltaic conversion,
Phys. Rev. B, vol. 55, pp. 69946999, 1997.
[19] G. L. Araujo and A. Marti, Absolute limiting efficiencies for photovoltaic
energy conversion, Solar Energy Mater. Solar Cells, vol. 33, pp. 213240,
1994.
[20] A. Devos and H. Pauwels, On the thermodynamic limit of photo-voltaic
energy-conversion, Appl. Phys., vol. 25, pp. 119125, 1981.
[21] A. Mart and G. L. Araujo, Limiting efficiencies for photovoltaic energy
conversion in multigap systems, Solar Energy Mater. Solar cells, vol. 43,
pp. 203222, 1996.
[22] A. Mart, L. Cuadra, and A. Luque, Quantum dot intermediate band solar
cell, in Proc. 28th IEEE Photovoltaic Spec. Conf., 2000, pp. 940943.
[23] A. Luque, P. G. Linares, E. Antoln, E. Canovas, C. D. Farmer,
C. R. Stanley, and A. Mart, Multiple levels in intermediate band solar cells, Appl. Phys. Lett., vol. 96, pp. 013501-1013501-3, 2010.
2-2
Invited
Fax: +81-3-5734-2897
An overview of the recent status of photovoltaic (PV) power generation is first presented. Next, the Japanese roadmap for the research and
development (R&D) of PV power generation and the progress in the development of various solar cells are explained. In addition, the
present status and future prospects of Si thin-film solar cells, which are expected to enter the stage of full-scale practical application in the near
future, are described. Thus far, a-Si/c-Si tandem solar cell modules with conversion efficiency exceeding 13% have been reported. In
addition, triple-junction solar cells are introduced, as well as innovative thin-film full-spectrum solar cells and Si nano-wire solar cells, whose
target year of realization is 2030-2050.
1.
Introduction
In 2011, the cumulative capacity of installed PV systems in
for business use (14 yen/kWh) by 2017 and to that feasible for
industrial use (7 yen/kWh) by 2025. To realize these electricity
Japan aims to
PV Roadmap
Figure 1 shows PV2030+ presented by NEDO. The milestones
Among the
AM-FPD '13
21
temperature.
The total
ZnO films with a textured surface for confininng light are used as
the TCO layer.
22
AM-FPD '13
Table 1 Confirmed terrestrial celll efficiencies measured under the global AM1.5 spectrum (10000W/m2) at 25.
completely suppressed.
t
solar cells include
The development targets of triple-junction
in the tunnel junction between the top and botttom cells, although
TCO interlayer is to reflect part of the incident ligght towards the top
thin-film solar cells are 14% for 2017 and 188% for 2025. To
attain these goals, conventional a-Si/c-Si doubble-junction tandem
solar cells must be superseded by triple-junnction solar cells.
Theoretically, it is very difficult to attain the above goals using
conventional double-junction solar cells.
However, it will
AM-FPD '13
23
solar cells. One is the thin-film full spectrum solar cells project
possible material.
from 1.5 eV band gap which is logically the most suitable band
exceeds four billion yen over the entire periodd. The themes of
wisdom(Fig.3).
open and the most advanced reseaarch and development base in the
the connection method for each cell and the opptical coupling are
from Japan and overseas will gathher at the research facility that will
not explained in the figure, not only material devvelopment but also
conversion efficiency.
Si-based thin films, the development targets of the first cell are Si
thin films with a band gap of 2 eV or higher,, such as a-Si1-xOx.
Amorphous Si has strong potential as the mateerial for the second
7. Conclusions
Today, the strongest rival to thhin-film solar cells is bulk Si solar
cell. For the third cell, with a band gap of appproximately 1.4 eV,
the fourth cell, c-Si is suitable. For the fifth cell as the bottom
thin-film solar cells to compete wiith bulk-Si solar cells in the future.
Si thin-film solar cells are exppected to achieve a conversion
efficiency of 20% by exploiting new materials developed in the
future and by applying a multijuncction structure. Now is the time
to bring together the experiencee and know-how of researchers
working in the area of photovoltaaics to speed up the development
Research
plan for develooping high efficiency Si NW/c-Si
ofFig.3
materials
and devices.
tandem solar cells for FY2012-20016.
References
24
AM-FPD '13
20:606614
I. NTRODUCTION
InGaP/InGaAs/Ge 3-juction solar cells have proved
to the most effective solar cells for space applications [16]. However, these cells are not ideally suited to Air mass
Zero solar spectrum. The top layer in these cells is current
rich whereas the middle layer is current poor. So in order
to compensate this current, the top layer is made thin, allowing unabsorbed sunlight to penetrate to the lower junction. This increases the current in the middle layer and
hence contributing to the overall current in the multijunction cell. A preferable, but more challenging approach
is to find a higher band-gap material to replace the InGaP
material, thereby increasing the current in the lower junction and increasing the top-cell open-circuit voltage [2].
Recently, AlInGaP was proposed as a top layer, which has
the future prospects of super high efficient solar cells.
Al
Composition
Bi Doped
InGaP
0%
No
AlInGaP
4%
No
Bi doped InGaP
0%
Yes
Bi doped AlInGaP
4%
Yes
However, the efficiency of the AlInGaP based multijunction solar cells that are under investigation is not up to
the expected levels. In order to understand the reason behind this low efficiency, we have explored role of Al content on native defects in the performance of AlGaInP solar
cells. In addition, we have also explored the interaction of
Bi on p-type InGaP and AlInGaP solar cell structures. The
Deep Level Transient Spectroscopy (DLTS) was used to
investigate the different characteristics of these observed
defects and their possible role in low lifetime/efficiency of
the solar cells. In view of depth analysis it has been concluded that the deep levels induced by Bi and Al play an
important role in governing the efficiency of Bi doped
InGaP and Bi doped AlInGaP solar cells.
II. EXPERIMENTAL
All samples investigated during the course of this
study were obtained from the EMCORE Corporation as a
part of collaboration with the University of South Alabama. Four samples of solar cells, as shown in table-1
with two different compositions of aluminum were studied
using the DLS-83D spectrometer.
2770
2771
C.
Hole Levels
The doping of Al in Bi doped InGaP solar cell structure affected the deep levels as shown in the Fig. 4. The
addition of Al noticeably enhanced the concentration of
the peak (E3), with a minor shift in the general position of
the peak. The effect of Al in Bi doped InGaP is found to
be similar to the one discussed in section A.
The trap concentration of electron peak E4 increased
from 4.6 x 1015 cm-3 to 1.7 x 1016 cm-3, with the addition
of Al. Hence, it can be concluded that addition of Al is
introducing the new defects closely related to native defects, which could be the cause of lowering the efficiency
of Bi doped AlInGaP solar cell compared to the Bi doped
InGaP.
Addition of Al, did not affect the majority carrier
traps observed in Bi doped InGaP solar cell structure. As
shown in Fig 4, trap H1 reappeared as H2, with an activation energy of (Ev+0.79 eV) above the valence band with
concentration of 2.6 x 1016 cm-3.
E.
Bi-InGaP
2772
VI. SUMMARY
In this study the influence of Al and Bi content on deep
levels in p-type InGaP has been reported. DLTS temperature scans were performed with the aim to investigate these deep levels. Four electron traps E1, E2, E3 and E4 were
observed in four different structures of solar cells.
Two hole traps H1and H2 were observed in Bi doped
InGaP and Bi doped AlInGaP solar cell structures. The
activation energies of the traps E1, E2, E3 and E4 were
calculated to be 0.22 eV, 0.27 eV, 0.17 eV, and 0.19 eV,
respectively. The activation energies of majority carrier
hole traps H1 and h2 were calculated to be 0.77 eV and
0.79 eV, respectively.
The DLTS temperature spectra of traps E1 and E2 were
compared to analyze the impact of Al on the solar cells.
Based on this comparison, it was observed that addition of
Al is noticeably enhancing the intrinsic traps in p-type
InGaP solar cell structure and thus limiting the AlInGaP
solar cell from attaining its maximum efficiency.
In addition to the impact of Al on the solar cell, traps E1
and E3 were compared to evaluate the impact of Bi on the
solar cell. Based on this comparison, it was observed that
Bi introduced additional deep levels which were found
closer to the emitter-base junction. These results suggested that the trap level H1 in Bi doped InGaP solar cell is
acting as a recombination center and is thus limiting the Bi
doped InGaP device from attaining its maximum efficiency.
The DLTS temperature spectra of traps E3 and E4
were compared to observe the impact of aluminum on the
on Bi doped InGaP solar cell solar cell. Based on this
comparison, it was observed that Al enhanced the native
defect concentration in Bi doped InGaP and thus the dominant peak E4 is seems responsible for lower efficiency in
Bi doped AlInGaP solar cell as well.
In addition to the above results, the shallow level
concentrations of the solar cell samples were found to
decrease one order of magnitude with the addition of Al
and two orders of magnitudes with the addition of Bi to
the p-type InGaP solar cell structures. This indicates that
the hole trap H1 may also act as a compensating center
and plays a major role in the carrier removal effect, which
can significantly affect the solar cell efficiency.
REFERENCES
[1] F. Dimroth, C. Baur, M Meusel, S. van Riesen, and A.W.
Bett. 5-junction III-V solar cells for space applications, 3rd
World Conference and Exhibition on Photovoltaic Energy Conversion, 2003.
[2] T. Takamoto et al., Multi-junction solar cell technologies high efficiency, radiation resistance and concentrator applications, 3rd World Conference and Exhibition on Photo voltaic
Energy Conversion, 2003.
[3] Chiang P. K., Kurt D. D., Cavicchi B. T., Bertness K. A.,
Kurtz S. R., and Olson J. M., Large-area InGaP/GaAs/Ge
multijunction solar cells for space applications, proc. 1st World
Conf. Photovoltaic Energy Conversion, pp. 2120-2123 (1994).
[4] A. Khan, S. Marupaduga, S. Anandkrishnan S. Alam M. et
al, Radiation response analysis of wide-gap P-AlInGaP for
super high-efficiency space photovoltaics, Applied physics
letters, Vol. 85, pp. 5218-5220 (2004).
[5] A. Khan., Jacques C. Bourgoin., and Tatsuya Takamoto.,
Thermal annealing study of 1 MeV electron-irradiation induced
defects in n+/p InGaP diodes and solar cells, Journal of Applied
Physics, Vol. 91, pp.2391-2397(2002).
ACKNOWLEDGEMENT
Special note of recognition goes to Mr. Aurthur B.
Cornfeld (EMCORE) for providing the samples and solar
cell data.
2773
I. INTRODUCTION
Solar arrays populated with high-efficiency III-V based
multi-junction solar cells have been powering the payloads of
earth orbiting satellites and interplanetary spacecraft for well
over a decade. Due to their higher performance compared to
silicon cells, multi-junction cells have enabled the realization
of higher power and larger capacity satellites.
The conversion efficiency of the InGaP/InGaAs/Ge triplejunction (3J) solar cells, under Air-Mass Zero (AM0)
illumination at 28C, has been steadily improved from about
23% to approximately 30% at beginning-of-life (BOL) since
the year 2000 [1-2]. The vast majority of the scientific,
military, exploration and telecommunication spacecraft now
use the highest efficiency 3J solar cells available. The power
requirements for these spacecraft range from ~1 kW to as
much as 25 kW at end-of-life (EOL).
28.0
28.5
29.0
29.5
30.0
30.5
AM0 Efficiency
2793
2794
AMOS-6
The AMOS-6 solar array consists of 2 wings of 6 panels per
wing. The ZTJ solar cell assemblies are laid down on
conventional kapton-covered aluminum honeycomb core
substrates with CFRP Facesheet. Space qualified, proven
processes are applied to the panels to mitigate and protect the
array against ESD (electrostatic discharge) events in the GEO
environment. More than 6,000 ZTJ cells (cell area ~60 cm2)
were laid down on 12 panels to meet the >10 kW EOL power
requirement for AMOS-6s 15-year mission.
NuStar
The NuStar solar panels that were delivered to ATK were
populated with 2-per-wafer ZTJ solar cells. The solar array
consisted of 1 wing of 5 panels. A photograph of the stowed
wing inside the Pegasus rocket is shown in Fig. 5.
Number of Cycles
Temperature Extremes
(APTC)
Estimated
Completion
GEO
8 TVAC
2,000
-175C to +80C
-175C to +80C
GEO
22,000
-165C to +125C
Completed
GEO
4,000
-165C to +130C
Completed
Completed
GEO
2,500
-170C to +145C
Completed
LEO
25,500
-100C to +125C
Completed
LEO
1,500
-106C to +88C
Completed
LEO
5,500
-75C to +135C
Completed
TABLE II
ZTJ 1-PER-WAFER LIFE-CYCLE COUPONS
Orbit
Number of Cycles
Temperature Extremes
(APTC)
Estimated
Completion
GEO
8 TVAC
4,000
-175C to +120C
-165C to +105C
Completed
GEO
2,800
-175C to +143C
Completed
GEO
2,000
-170C to +145C
June 2013
GEO
12 TVAC
2,000
7,500
5,000
-180C to +130C
-180C to +105C
-165C to +105C
-140C to +105C
2014
LEO
10 TVAC
2,400
-128C to +128C
-128C to +128C
Completed
LEO
14 TVAC
32,000
-70C to +130C
-70C to +130C
2014
LEO
9 TVAC
3 TVAC
700
600
-110C to +70C
-150C to +130C
-110C to +70C
-150C to +0C
Completed
TABLE III
ZTJ FLIGHT PANEL ACCEPTANCE TESTING
Orbit
LEO
LEO
LEO
LEO
LEO
LEO
LEO
LEO
LEO
GEO
Number of
Cycles (APTC)
Temperature Extremes
Vac. Bake-Out
8 TVAC
Vac. Bake-Out
8 TVAC
Vac. Bake-Out
8 TVAC
Vac. Bake-Out
14 APTC
Vac. Bake-Out
12 TVAC
Vac. Bake-Out
8 TVAC
Vac. Bake-Out
5 TVAC
Amb. Bake-Out
14 APTC
8 APTC
Vac. Bake-Out
10 TVAC
48 Hrs. +125C
-90C to +90C
24 Hrs. +100C
-101C to +117C
96 Hrs. +110C
-110C to +110C
48 Hrs. +125C
-103C to +103C
96 Hrs. +135C
-100C to +125C
24 Hrs. +135C
-100C to +130C
72 Hrs. +120C
-95C to +120C
24 Hrs. +125C
-97C to +95C
-90C to +90C
2 Hrs. +135C
-179C to +88C
Estimated
Completion
Luanch Date
Completed
June 2012*
Completed
2013
Completed
2014
Completed
2014
Completed
2014
2014
2015
2013
2015
2014
2015
Completed
2014
2013
2014
*Launched
2795
2796
I. INTRODUCTION
InGaP/InGaAs/Ge triple-junction solar cells with over
30% efficiency, which consist of lattice-matching
epitaxial layers grown on a Ge substrate, have already
been put to practical use in power generators for space
satellites. However, this type of solar cell is not perfectly
matched to the solar spectrum [1, 2]. For the high
efficiency performance of monolithically fabricated III-V
multijunction solar cells, high quality epitaxial growth of
subcells with high internal quantum efficiency is a
requirement as well as an optimized combination of band
gap materials for efficient conversion of the solar
spectrum [1, 2]. The first approach is the invertedmetamorphic (IMM) structure [3], that is, the GaInP top
cell and the GaAs middle cell are grown lattice matched
and therefore virtually defect free on a GaAs substrate.
Then the lattice constant is changed by a metamorphic
2832
250
200
Current (mA)
150
100
50
0
-50
Fig. 2.
1000
2000
Vbias (V)
3000
Contact layer
100
Step-graded buffer
80
Current (mA)
60
40
Top
Middle
Bottom
20
0
400
600
800
1000
1200
Wavelength (nm)
Fig. 3.
2833
15
IV. CONCLUSION
Current (mA)
10
5
0
-5
-10
0
Fig. 4.
1000
2000
3000
Vbias (V)
50
4 layer + 1
experimental data
40
simulation data
R(%)
20
10
0
400
600
800
1000
1200
1400
Wavelength(nm)
1600
30
200
1800
2834
I. I NTRODUCTION
LIFE TESTS
EA
L(T ) = Ce kT
(1)
(2)
(3)
AF =
L(T2 )
k
T1
T2
Therefore, once the parameters of the life distribution and
life-stress relationship have been estimated, the same information that is available from real time tests can be obtained with
the ALT.
III. E XPERIMENTAL APPROACH
The methodology followed in the temperature accelerated life test which is being carried out on commercial
GaInP/GaInAs/Ge triple-junction solar cells (lattice matched)
presented in this work, is an adaptation from the procedure
already carried out satisfactorily on GaAs concentrator solar
cells [6]. As we pointed out above, the parameter used to
accelerate the aging of the solar cells is the temperature
and at least three temperatures are necessary to calculate the
acceleration factor. Therefore, we have placed three groups
of twenty solar cells inside three climatic chambers at three
different temperatures. The working conditions are emulated
by injecting in darkness the equivalent current that the solar
cell would photo-generate in illumination at the nominal
concentration level. The dark I-V curve of the solar cells
is periodically monitored during the test inside the climatic
TABLE I
E QUIVALENT CONCENTRATION NEEDED TO PHOTO - GENERATE THE SAME
CURRENT DENSITY THAN THE CURRENT DENSITY WHICH FLOWS IN
DARKNESS DUE TO THE INJECTION OF 3.2 A.
Subcell
TC
MC
BC
Busbar
450 X
440 X
430 X
Fingers
470-820 X
320 X
315 X
Active Area
250-340 X
330 X
320 X
11.5 A/cm2 and zero or negative in the bus bar and fingers.
Ideally this current density distribution should be emulated
in the ALT by forward biasing the solar cell in darkness
but in darkness the main of the current flows beneath the
busbar and fingers. We have used our 3D distributed model
to simulate the distribution of the current density through the
subcells when different levels of current were injected into
the solar cell. We have been increasing the current injected
into the solar cell so that the current density distribution
in darkness was as close as possible to the current density
distribution at 820 X, but without exceeding in darkness the
current density photo-generated at 820 X (11.5 A/cm2 ) in
any point on the solar cell. Therefore, the criterion followed
0 X, 3.2 A
Fig. 3. False color maps representing the current density (A/cm2 ) through the top cell (top figures), middle cell (medium figures) and bottom cell (bottom
figures) for an illumination of 820 X at the maximum power point (left figures) and injecting 3.2 A in darkness (right figures). The vertical scale is A/cm2.
C. Data analysis
Once all the solar cells have failed, the analysis of
accelerated life test data consists of: 1) finding a life
distribution model which describes the solar cells failures
at different temperatures, 2) finding a life-stress model
that quantifies the manner in which the life distribution
changes across different temperatures, 3) to combine the life
distributions and the life-stress model to obtain reliability
Fig. 4. Sketch of the thermal cycles of the solar cells inside the climatic
chamber.
f (t) =
1
t
t
e ( )
Fig. 5. Weibull plot of the failure probability versus time (hours) for the
solar cells at 164 o C.
(4)
Fig. 6. Probability density function (pdf) versus time (hours) for the solar
cells at 164 o C.
(5)
Fig. 8. Photograph of the metal deteriorated of the solar cells which were
injected 3.2 A in the ALT at 164 o C (left figure) and photograph of the metal
intact of the reference solar cells (right figure).
International Conference on Electrical Engineering and Information & Communication Technology (ICEEICT) 2014
I.
fabrication;
cell
efficiency;
cost
INTRODUCTION
energy comes from solar power [2]. Hence, this means the sun
shine can produce about 35000 times more power on the earth
than the daily power generation using solar energy. In
Bangladesh, the average sunshine hours are 6.69, 6.16 and
4.81 in winter, summer and monsoon, respectively.
Photovoltaic (PV) method which converts solar radiation
directly into electricity using solar cells is a simple and unique
method to use solar radiation as it is noiseless, reliable and has
a long life. Solar cells are made from different semiconductor
materials which affect the cost and efficiency. Solar cells can
be made of crystalline silicon, called Generation-I which
includes mono-crystalline silicon, polycrystalline silicon,
amorphous cadmium telluride, and copper indium
selenide/sulfide [7]. There are also thin film solar cells, called
Generation-II which includes cadmium telluride solar cell,
copper indium gallium selenide, gallium arsenide junction,
light-absorbing dyes (DSSC), quantum dot solar cells
(QDSCs), organic/polymer solar cells, silicon thin films [8].
The paper aims in analyzing the manufacturing processes that
is practiced in the laboratory with an aim to developing costeffective crystalline solar cell. Sect. II deals with the principle
how solar cell works.
II.
A. Existing Technology
Solar cells, which has become a rapidly growing and
Fig. 3: Flow chart of solar cell fabrication (Frond end process) processing
steps adopted in SCFL
Cell type
Doping
Thickness
Silicon
wafer
Monocrystalline
P type
150150 mm2
200m
Once the samples are loaded in the tube, the furnace is heated
up to the preferred temperature before any processing can take
place. During the initial stage or pre-deposition stage of the
diffusion process, carrier gas nitrogen is passed to feed liquid
POCl3 into the process chamber.
(1)
IV.
EFFICIENCY CALCULATION
Voc I sc FF .. .(2)
=
Pin
V.
Fig.10 shows the LIV tester in SCFL. Solar cells are tested
under one-sun conditions using Xenon-arc lamps; a Xenon
spectrum is closest to the sunlight. Data acquisition based on
programmable current-voltage source power supplies capable
of handling currents up to 8A is used in conjunction with a
proprietary data acquisition system. Calibration of this LIV
measurements system is based on independently measured cSi solar cells at Sandia National Laboratories. The measured
LIV data are shown in Fig.11.
COST SCENARIO
CONCLUSION
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
www-inst.eecs.berkeley.edu/~ee143/fa10/lectu res/Lec_26.pdf
4. Rafique, S., Renewable Energy Scenario in Bangladesh: Estimation,
Expectation and Future Trend, presented and published as a Key-note
country paper in WREC (World Renewable Energy Congress) VIII,
Denver, Colorado, USA, Aug 29- Sept 3, 2004.
http://www.thedailystar.net/newDesign/news-details.php?nid=210387
Photovoltaic Education, http://www.pveducation.org
Mark Z. Jacobson (2009), Review of Solutions to Global Warming,
Air Pollution, and Energy Security, p. 4.
Wikipedia-Solar cell http://en.wikipedia.org/wiki/Solar_cell
Nigatu Teklu (2009) Nigatu Teklu 2009, Study the Effect of Ultra
Violate Radiation on Polycrystalline Silicon Solar Cell, Submitted in
Partial Fulfillment of the Requirements for the Degree of Master of
Science in Material Science, at ADDIS ABABA, ETHIOPIA March
(2010).
Andreas Bentzen, Phosphorus diffusion and gettering in Silicon solar
cells, University of Oslo.
Ferrada Martinez, Pablo, Diffusion through Oxide Barriers for Solar Cell
Applications, 2012-06-13,http://nbn resolving.de/urn:nbn:de:bsz:352196 190.
http://www.crystec.com/tridepe.htm
Spin-on Dopant Method, United States Patent by Bruce H. Justice,
Robert F. Aycock.
J. Lossen, L. C. Beneking. Making, Use of Silicon wafers with Low
Lifetimes by Adequate POCl3 Diffusion, 20th European Photovolatic
Solar Energy Conference and Exhibition, Barcelona, Spain, 6-10, 2005.
[15] www.festo.be/nl
[16] Martina A. Green, Crystalline Solar Cell, Photovoltaic Special
Research Centre, University of New South Wales, Sydney, N.S.W
Australia.
[17] Keith Emery, Measurement and Characterization of Solar Cell and
Module, National Renewable Energy Laboratory, Golden, CO, USA.
[18] Radiant Power, Bangladesh.
[19] Electro Solar Power Ltd, Bangladesh.www.electrosolarbd.com
[20] Rahimafrooz Renewable Energy Ltd (RRE), Bangladesh.
[21] AVA Renewable Energy Ltd. www.avasolarbd.com
[22] Greenfinity energy ltd. www.greenfinitybd.com
[23] How Its Made-Solar Panel by http://www.earth4energy.org on
www.youtube.com /watch?v=qYeynLy6pj8
[24] Wikipedia-Solar Panel. http://en.wikipedia.org/wiki/Solar_panel#Mod
ule_perfo rmance_and_lifetime.
[25] Energetics, Incorporated, Ed., Solar Electric Power: The U.S.
Photovoltaic Industry Roadmap.
[26] Md. Shafiqul Islam, Md. Rakibul Hasan, Fariba Mohammadi, Antara
Majumder and Ali Ahmed, Study of Cost-effective Crystalline Type
Solar Panels and Solar Cells Manufacturing Practices in Bangladesh,
Proceedings of the 6th Int. Mechanical Engineering Conf. & 14th
Annual Paper Meet (6IMEC&14APM), ET-12, Dhaka, Bangladesh, 2829 September (2012).
I. INTRODUCTION
During the last years a large number of new satellite
orders was placed by the need in replacement of old
platforms, whereas in terms of applications, the major focus
currently is on TV broad-casting, high definition TV and in
the near future, 3-D TV. New payloads shall provide
significantly larger capacity with simultaneously higher
transmitting signal quality so that the demand in electrical
power to be provided by solar generators increases
correspondingly. This demand extremely sharpens the need
for lightweight, high-performing and cost effective
components and sub-systems for spacecrafts. In case of solar
cells this particularly means i) increase in solar array power
(W/m2), ii) reduction of specific mass (kg/m2), iii) reduction
of manufacturing costs (/W) and iv) increased operational
lifetime (>15y).
II. CURRENT SPACE TRIPLE CELLS 3G28
The 28%-class triple junction cell 3G28 was developed
mainly for applications in GEO. From 2009 up to now it was
[mV]
[mA]
[mV]
[mA]
[mW]
[%]
BOL
2,667
506
2,371
487
1,155
28.0
5E+14
2,534
501
2,229
472
1,052
25.8
1E+15
2,480
486
2,205
458
1,010
24.7
3595
Material quality,
design of layers + grid
GaInP
tunnel diode
Higher transparency
GaInAs
tunnel diode
buffer
Fig. 1.
scheme
nucleation
Ge
Fig. 3.
B. Customized cells
For some ESA projects, special solar cells have been
derived from the basic 3G28 cell design. Based on
particular project specifications, various modifications in
cell dimensions, epitaxial structures, antireflective coatings,
grid designs, contact pads, as well as tunnel and bypass
diodes were realized. Examples are:
- Solar cell 3G28 with mesa groove and high density
grid and with a modified contact material system
allowing performance improvement with respect to
higher currents and higher temperature for near sun
missions such as BepiColombo and Sun Probe [3].
- 3G28 solar cell with LILT (low intensity, low
temperature) optimized design for deep space missions
such as Laplace demonstrating outstanding LILT
performance of about 35% with the absence of flat
spot effect on SCA and bare cell level [4].
For each special application a delta-qualification of the
modified cells has been carried out according to the very
special requirements of the individual mission.
Fig. 2.
scheme
Improvement of the
surface morphology
(substrate, nucleation,
buffer)
28
26
24
22
20
18
3G30-Advanced
1E14
1E15
-2
1E16
3596
TABLE II
ELECTRICAL BOL- AND EOL-DATA (1MEV EL) OF 3G30
2
CLASS SOLAR CELLS (28C, AM0 1,367 W/M )
Average Voc
Average Isc
Average Vmp
Average Imp
Average Pmp
Average AM0
[mV]
[mA]
[mV]
[mA]
[mW]
[%]
BOL
2,700
520
2,411
504
1,215
29.5
5E+14
2,564
514
2,290
501
1,147
28.1
1E+15
2,522
502
2,246
487
1,094
26.8
A. Lightweight Cells
AZUR has developed a special proprietary thinning
process, combining mechanical as well as chemical removal
of the main portion of the Ge substrate. As the Ge substrate
also constitutes the bottom sub-cell, a remainder of Ge has
to be saved. Using this process the thickness of the Ge
substrate could be reduced from nominally 145 m to 80
m. Feasibility studies have shown that the cells can also be
thinned further to only 50 or even 20m. These cells have to
be stabilized by a coverglass to avoid coiling up and to
facilitate handling. This way, up to 70% of the bare cell
weight has been saved (Table III). Additionally, low-weight
and flexible cells allow for a redesign of the complete PVA
structure permitting further reduction in mass for several
elements and therefore also in cost.
TABLE III
MASSES OF VARIOUS SCA-CONFIGURATIONS
Cell Mass
[mg/cm2]
84
84
49
49
33
33
17
17
17
SCA Mass
[mg/cm2]
114
108
79
74
63
50
60
47
34
3597
15
10
110
<28
22.0
28
29
Efficiency [%]
30
31
Pmp,nom[%]
150 m
80 m
20 m
105
100
95
07/2012 11/2012 03/2013 07/2013 11/2013
Date
B. Flight Data
Cells of the 3G30-Advanced type were flown on the
German technology satellite TET-1. TET-1 was launched on
July 22, 2012 from Baikonur with a Sojus Fregat Launch
vehicle into a circular sunsynchronous orbit at 506 km
altitude. The AZUR payload consisted of a CFC-panel with
6 advanced solar cells three of 3G30-Advanced type (Fig.
8). The cells had a size of 4cmx8cm each, were of different
thickness ranging from 150m, 80m to 20m and were
covered with a CMX100 coverglass from Qioptiq (UK). The
IV-characteristics of each cell as well as the substrate
temperature were measured regularly several times a week.
The first data transmission occurred on July 26, 2012 and
was continued until mission end on October 31, 2013. The
sun orientation was first at 30 due to thermal constraints
and was adjusted to 0 in May 2013. The transmitted IVdata were normalized to AM0 standard conditions at 28C.
This allowed comparison with the calibration measurements
prior to integration.
The accuracy of each measurement was 2%. By the high
number of measurement points the standard deviation could
be kept below 1%. In Figure 9 the evolution of maximum
power over operation time of the satellite is depicted. The
majority of the data lie well above the calibration points.
Thinning has hardly any impact on cell performance. In this
3598
30
lot [%]
100
80
20
EQE
60
10
40
BOL
-2
5E14 cm
-2
1E15 cm
20
400
600
800
1000 1200
[nm]
1400
1600
0
1800
30.0
30.5
31.0
31.5
eta [%]
Fig. 11. Efficiency distribution at AM0 (1367 W/m2, 28C) for
UMM 3J bare solar cells in standard thickness of 145m [11].
[mV]
[mA/cm2]
[%]
BOL
3370
14.9
0.87
32.0
EOL(5e14) EOL(1e15)
3200
3150
14.7
14.3
0.87
0.86
30.0
28.5
V. COMPLEMENTARY TECHNOLOGY
DEVELOPMENTS
In order to surpass the 32% efficiency for the next
generation solar cell further cell concepts are also
investigated by AZUR and partners [6]. Essentially all other
4J approaches target about the same bandgap combination
as the 4J UMM cell structure, but implement it with
different means.
Lattice-matched 4J cells incorporating InGaNAs still
require intensive material research on basic R&D level.
InGaNAs material can be grown either by MBE or
3599
[3] C.G.Zimmermann,
C.Nmayr,
W.Kstler,
A.Caon,
E.Fernandez, C.Baur and H.K.Fiebrich, Photovoltaic
Technology Development for the Bepi Colombo Mission,
Proc. of 9th European Space power Conference ESPC 2011,
Saint Raphael, France
[4] V.W.Khorenko, G.F.X.Strobl, R.Hoheisel, F.Dimroth,
R.Campesato, M.Casale and C.Baur, Triple-Junction Solar
Cell design for Low Intensity, Low Temperature Space
Applications, Proc. of 9th European Space power Conference
ESPC 2011, Saint Raphael, France
[5] W. Guter, L. Ebel, D. Fuhrmann, W. Kstler and M.Meusel,
Development of Space Solar Cells with 30% EOL
Efficiency, Space Power Workshop 2014, Manhattan Beach
(USA, CA), 2014.
[6] G.F.X.Strobl, D.Fuhrmann, W.Guter, V.Khorenko, W.Kstler
and M.Meusel, About AZURs 3G30-advanced Space Solar
Cell and next Generation Product, Proc. of the 27th
EUPVSEC, Frankfurt, p. 1BO10.3, 2012.
[7]
REFERENCES
[1] AZUR Final Summary Report HNR 0002274-01
Development of advanced GaAs-based triple-junction solar
cells under ESA Contract No. 15134/01/NL/FM
[2] AZUR document HNR 0002922-00, Executive Summary
Report Space Qualification of European Triple-Junction
Solar
Cell
AZUR-3G28
under
ESA-Contract
18118/04/NL/US
3600
expected
to
have
great
potential
of
space
and
terrestrial
direct
bonding
technologies
in
order
to
prevent
such
solar
cells.
The
paper
also
reviews
approaches
on
Index Terms
'.50
------,,---=TW::-:O:::-::T2::-:Eft:7:-::IN AL
, 25 ___-c-.lit=-'l:iI'1I"'_A_M_' _
.
:;o
"'
CJ
0.75
t,GO
THREE CEl.1.S
$eRIU'ONH(CTEO
..
': .-
,, ------c.:.....-
-n9'fff;,fT-----:-:::-
I.$O ____
;-
(a ) 2-junction cell
n = 0,
at
(2)
at
(3)
g.
d!1nldx = 0,
I. INTRODUCTION
( b) 3-junction cell
is dislocation density.
The resulting dislocation-limited minority-carrier lifetime is given by
IITd = ,,3NdDI4,(4)
0821
50
,-.
40
....
u
s::
30
:::...
.ilI
<.i
<=
'..
s::
.!:
'"
""
..
..
c
0
Coocentrator
Tandem
20
D=Do 1[1+DoK2t ]
Cells
(12)
Single-junction
f3 -dislocations
K2
used as follows:
10
v=9.86xlO-9(j'L6exp(-1.35e VlkT)
K =vlh=av
(14)
0
103
10
IO
10
10
(an -2 )
[emI5]
2
where V is dislocation velocity and, a is a constant, determined to
8
be 7.2xlO cm from experimental results for GaAs-on-Si [6]. The
stress as a driving force of dislocation motion is thought to be
derived from misfit stress at around annealing temperature. In this
case, stress value of 107 dyne/cm2 was used as the stress value.
Hetero-epitaxial Film
Prop'gation
Deflection
Combination
+Re-emlsslon
+Propgation
K
D-' a nnihilat ion
( 6)
Kz
D+D- Dz
(7)
(13)
where D,
constants.
The reaction equations for dislocations are given by
(8)
(9)
D(t = 0) =Do,
D2 (T = 0) = 0
10e
o
________
=5
________
________
10
Cycle number N
16
0822
3.2.
Ino lGao9As/GaAs SLS insertion and TCA [4,7]. The reduction of the
dislocation density is found to become more remarkable as the
108 r-------.
10 period
;:;'
0 InO.1GaO.9As/GaAs
(TEM)
(EPD)
TCA
(PO)
(900'C, N=4)
.....
lIS'grQwn
E
.3
10
Dista nee from Si
(.um)
101
1
10
tot
>-
t:
and
open
and
closed
circles
were
determined
by
density
decreases
as
the
film
thickness
increases.
Its
8 107
I-
,
"
b,
...... '"
<..>
-- '"
(/)
6
0 10
10 5
TEM
7 [10]. Closed
109
(/) 10
z
UJ
Q
GoAs/SI
Fig.
201
100
GoAs FILM THICKNESS
(I'm)
experiments,
respectively.
The
lines
represent
the
calculated values.
0823
Dislocation density
Nd(X)
C1 X
x < IO,um,
for
(15)
for
x>50,um,
(16)
2
7
C 5xlO-3 cm for these
1.9xl0 cm- ,
NAx)= NdOexP(-xlxo)
where constants NdO
experimental results.
Based on one-dislocation reaction such as annihilation and
deflection of dislocation, and two-dislocation reaction such as
coalescence of dislocations, the equation of dislocation density
GaAs film thickness x is given
dD(x) 1 dx
-aD - bD2
D for
(17)
where D,
constants.
Therefore,
dislocation
density
Nd
dependence
on film
thickness x is expressed by
(18)
NAx)11 bx
(19)
Nd(x)(alb)exp(-ax).
(20)
High quality GaAs film with an EPD of less than lx106 cm-2 on
Si substrates have been obtained by selective area growth [11, 12].
The selective area growth is thought to aid stress relief and thus
prevent dislocations from generating. The major component of the
9
large residual stress of about I.Sx10 dyne/cm2 in the as-grown GaAs
film on Si substrates is thermally induced stress after growth; stress
relief in the films fabricated by selective area growth has been
analyzed using the bi-metal model [13] as follows:
0"=
where
O"o[l-exp{-k(w- x)}].
a,
(21)
distance from the center of the patterned film, and kis the interfacial
compliance parameter related to the elastic constant, young's
modulus of film and substrate, and film thickness,
Numerical analysis [12] suggests that high quality III-V on Si
with a dislocation density less than 105 cm-2 will be obtained by
optimizing selective area growth and TCA, but selective area
effective for low dislocation density is thought to be limited with
selective area less than 20m..
3.5. Low Temperature Growth
0824
n+/n+---AGIGaAsaAs
G
aAs
n+
n+/n-i-lnGGaAaAss IIl1
SL
SLS
Si solar cell.
mAlcm2, FF=82.6 %) under 48.3 suns with InGaP/GaAs/Si 3junction solar cells with 0.OS4 cm2 area by using direct bonding
40
c:t
...
c:
G)
...
...
::s
SUN,AMO
T=28C
Pmax
lOr-
0
0
technique [24].
V. SPACE FLIGHT DEMONSTATE OF GAAS-ON-SI SOLAR
CELLS
In order to apply GaAs-on-Si solar cells in space, they have
undergone a space flight test [2S]. Forty-eight 2 cm x 2 cm GaAs-on
Si solar cells with an average AMO total-area efficiency of 16/9%
were evaluated in the Engineering Test Satellite No. 6 (ETS-VI),
launched on 28th August 1994. Figure 10 shows AMO efficiency
distribution of the solar cells with SO and 100 11m cover glass. The
beginning-of-life (BOL) specific power for the GaAs-on-Si cells
(with SOl1m thick) studied was evaluated to be 68SW/kg and was
found to be higher than lS8W/kg for the GaAs-on-GaAs cells and
SS8W/kg for the SOl1m thick Si cells.
30 r20-
Isc
Voe
FF
7]'0'01
1) active
i
33.2mA
0.940V
0.791
18.3%
19.2%
10
...
Q)
.0
::J
0.5
Voltage (V)
high
AMO
Average
AMO
Efficiency
16. 86%
4
2
0
, I
1.0
16.0
16.4
16.8
AMO Efficiency
17.2
(%)
17.6
max
0825
.s
0.9
0.8
0.7
0.6
0.5
E
Q)
0:: 0.4
GaAs on Si
(50um Cover
GaAs-on-Si
(1 OOum Cover
Cell
Glass)
e
1----I
--i -=---------------1
Q
t '""c:"'; "::'=:':
-'
-i ()U--5""'':"'"-":
---1
1----__
Cell
Glass)
0
----
2 OOum-thick Si Cell
...
I--------.-=---____;.;;;_='__l
( um Cover Glass)
.. 5
...
GaAs-on-GaAs
Bulk Cell
G '''=S)_--7100O''u=m-oCth' =ick''Cs:c-rn=ec'-'
ov=ec-"=
1-----"(5=O=um",C=
( 0um Cover Glass)
5
---1
--
--"---___--'---___---'
L-___-"---___
Various space solar cells evaluated using ETS-VI
for the
Fig. II. Remaining factor of maximum output power P
GaAs-on-Si solar cells after 94 days from launching in comparison
with those for liquid-phase epitaxy grown GaAs-on-GaAs cells and
50llm, 100llm and 200llm thick Si cells.
max
3xlO cm- .
ACKNOWLEDGEMENT
[1] J.C.C. Fan, B.Y. Tsaur and RJ. Palm, Proceedings of the 16th
IEEE Photovoltaic Specialists Conference, (IEEE, New York,
1982) p.692.
[2] M. Yamaguchi and C. Amano, J. Appl. Phys., 58, 3601 (1985).
[3] M. Yamaguchi, A. Yamamoto, M. Tachikawa, Y. Itoh and M.
Sugo, Appl. Phys. Lett., 53, 2293 (1988).
[4] M. Yamaguchi, T. Nishioka and M. Sugo, Appl. Phys. Lett. 54,
24 (1989).
[5] S.K. Choi, M. Mihara and T. Ninomiy Jpn. J. Appl. Phys. 16,
737 (1977).
[6] M. Yamaguchi, M. Tachikawa, Y. Itoh, M. Sugo and S. Kondo, J.
Appl. Phys. 68, 4518 (1990).
[7] M. Yamaguchi, J. Mater. Res., 6. 376 (1991).
[8] P. Sheldom, K.M. Jones and M.M. AI-Jassim, J. Appl. Phys. 63,
5609 (1988).
[9] H. Kroemer, T.Y. Liu and P.M. Petroff, J. Crystal Growth 65, 96
(1989).
[10] M. Tachikawa and M. Yamaguchi, Appl. Phys. Lett. 56, 484
(1990).
[II] S. Onozawa, T. Ued M. akiyama and M. Sakut Annual
Meeting of the Jpn. Soc. Appl. Phys., Toyam 7p-Y-7, 1988.
[12] M. Yamaguchi, M. Tachikaw M. Sugo, S. Kondo and Y. Itoh,
Appl. Phys. Lett. 56, 27 (1990).
[13] E. Suhir, J. Appl. Mech. 53, 657 (1986).
[14] M. Tachikawa and H. Mori, Appl. Phys. Lett. 56, 2225 (1990).
[15] K. Nozawa and Y. Horikoshi, J. Electron. Mater. 21, 641 (1992).
[16] Y. Okada, H. Shimomura and M. Kawabe, J. Appl. Phys. 73,
7376 (1993)..
[17] Y. Itoh, M. Yamaguchi, T. Nishioka, and A. Yamamoto, J. Appl.
Phys. 61, 762 (1987).
[18] Y. Kadot M. Yamaguchi, and Y. Ohmachi, Proceedings of the
4th International Photovoltaic Science and Engineering
Conference, Sydney, (The Institute of Radio Electronics and
Engineers Australi Edgecliff), 1989, pp. 873-880.
[19] M. Akiyama, Y. Kawarada, and K. Kaminishi, Jpn. J. Appl.
Phys. 23, L843 (1984)..
[20] H. Okamoto, Y. Watanabe, Y. Kadot and Y. Ohmachi, Jpn. J.
Appl. Phys. 26, L1950 (1987).
[21] Y. Watanabe, Y. Kadot H. Okamoto, M. Seki, and Y.
Ohmachi, 1. Cryst. Growth 93, 1988, pp.459.
[22] SA Ringel, R.M. Sieg, J.A. Carlin, S.M. Ting, EA Fitzgerald,
M. Bulsara, and B.M. Keyes, Proceedings of the 2nd World
Conference and Exhibition on Photovoltaic Solar Energy
Conversion, Vienn (European Communities, Ispra, Italy, 1998),
pp.3594.
[23] T. Jimbo et aI., Science and Technology of Advanced Materials,
6, 27 (2005).
[24] K. Derendorf et aI., IEEE Journal of Photovoltaics, 3, 1423
(2013).
[25] M. Yamaguchi, Y. Ohmachi, T. Oh'har Y. Kadota, M.
Imaizumi and S. Matsuda, Progress in Photovoltaics 9, 191
(2001).
0826
Abstract
Later
Index Terms
I.
INTRODUCTION
II.
1884
C. Testing
A. Hardware
device-under-test (DUT) and the ADCs are not shown. The DUT
enclosure also contains the probe station.
The housing for the DUT is provided with a white LED and a
optical diffuser made of three-layers of tissue paper.
B. Software
Fig. 2
and
fitting
are
optional
means
of
handling
the
The shielding for the DUT and for the ADC was adequately
effective. The performance of the two LNAs, compared over
a gain range from 1 to 10,000, was nearly identical. Spectra
were obtained using ADC windows of 1 and 10 V without
noticeable differences. Noise floor spectra were reproducible
to within a factor of 2. Spectra taken with our new digital
system compared well with spectra from a recently
recalibrated HP 3588 analog spectrum analyzer. The single
channel noise floor was competitively small at 3 x 10.17
V2/Hz. The dynamic range was measured to be over one
million. Sharp lines of unknown origin seen in some spectra
had widths of less than 0.04 Hz, so they are ignorable. The
hardware, software and results of testing the new digital
system will be described fully in another paper [9].
D. Features and Advantages
1885
Fig. 3.
correlated methods from the noise floor (bottom two) and a GaAs
solar cell (top two). Noise on the noise spectra is unimportant.
RadioShack
0.6-14
14
IXYS
2.4-80
80
TRONY
6-40
40
101
I,o"
10"t
:::
I
Fig. 5.
10
100
f'Dqurenc:y(
It
.. ... .. ..
"
ICM!
loot
1886
[3]
Fundamentals o/Solar
Cells: Photovoltaic Solar Energy Conversion, Ac. Press, 1983.
H. J. Hovel. Semiconductors and Semimetals. Volume II : Solar
[4]
[2]
IEEE Trans.
[9]
Fig. 7.
V.
CONCLUSION
1887
1505
Results from the High Efficiency Solar Panel Experiment Flown on CRRES
K.P. Ray and E.G. Mullen
Phillips Laboratory, PL/GPSP, 29 Randolph Rd.
Hanscom AFB, MA 01731-3010
T.M. Trumble
Wright Laboratory, WLIPOOC-2
Wright-Patterson AFB, OH 45433-6563
Abstract
This paper presents results from the High Efficiency Solar
Panel Experiment (HESP) flown on the Combined Release and
Radiation Effects Satellite (CRRES). The on-orbit solar cell
degradation is correlated with the proton and electron environments. Comparisons between gallium arsenide germanium
(GaAs/Ge) and silicon (Si) solar cells are presented, and
results from three different annealing methods of like GaAs
solar cells are compared.
I. INTRODUCTION
Damage to solar cells in space can reduce overall spacecraft power which can affect operations and curtail mission
lifetimes. In recent years, the standard silicon solar cell has
been replaced by thinner junction silicon cells and different
materials like GaAs and InP. Silicon solar cell performance in
space has been well documented [l-21.Recently, GaAs and
InP solar cell results have been published [3-41.
Much of the analysis done on the space performance of
solar cells is based on the common practice of reducing the
fluences produced by the NASA models, both AP8 (protons)
and AE8 (electrons), to a 1 MeV equivalent electron fluence
and then looking up a damage versus fluence curve in a solar
cell radiation handbook [5]. However, when discrepancies
arise, there is no way of knowing whether the estimations of
the space environment are in error or if the solar cell is not
degrading as expected.
The data from CRRES combines HESP solar cell degradation data from advanced technologies like GaAs and GaAs/Ge
with proton and electron differential fluence data in units of
particles/cm2/sr. From this data set, we have been able to
compare the performance of GaAs and silicon solar cells as
well as correlate the degradation of maximum power to
changes occurring in the space radiation environment. Finally,
we have compared three different annealing methods.
We do not attempt to reduce the radiation environment to
a 1 MeV equivalent electron fluence and compare the results
with the NASA models. Instead, our goal is to demonstrate
that the very dynamic nature of both the proton and electron
radiation belts and the changes brought about by solar flares
and geomagnetic storms directly affect the performance of
11. SPACECRAFT
& INSTRUMENTATION
The CRRES mission lasted 1067 orbits from July 25, 1990
to October 12, 1991. The satellite had an highly elliptical orbit
(350 km X 35,000km)and an inclination of 18.2". The nine
hour and fifty minute orbit took the satellite through the heart
of the outer (electron) and inner (mostly proton) radiation
belts.
HESP was designed to study the degradation of both
gallium arsenide germanium (GaAs/Ge) and silicon (Si) solar
cells in the space radiation environment and to study the
effectiveness of different types of annealing methods in GaAs
solar cells. Five different solar cell parameters were measured:
maximum power, short-circuit current, open-circuit voltage,
and voltage and current at maximum power. Other parameters
studied included solar cell efficiency, GaAs/Ge cells versus Si
cells, coverglass thickness, coverglass material, and soldered
versus welded interconnects. Two different solar panels were
built: an Ambient panel and an Annealing panel.
The Ambient panel was made up of a total of 12 cell
strings, 10 GaAs/Ge strings with four cells per string and 2
silicon strings with five cells per string. All solar cells were 2
cm X 2 cm in area. The two Si solar cell strings, both built by
Spectrolab, were the K4-314 solar cell with a solar cell
efficiency of 12.3 % and the thin K7-314solar cell with a solar
cell efficiency of 14.8%. The K4-314cell string served as the
reference string for the entire experiment. The K4-314 N/P
cell has a 10 n-cm base resistivity, 0.15 pm junction depth, a
dual layer Ti0,A1203 AR coating, PdTiAg contacts, no back
surface field, a back reflector, and is 200 pm thick. The thin
K7-3/4N/P cell has a 10 Q-cm base resistivity, a 0.15 pm
junction depth, a dual layer Ti0,A1203 AR coating, PdTiAg
contacts, a back surface field, a back reflector, and is 62 pm
thick [6].
The ten GaAs/Ge PIN cell strings were built by Applied
Solar Energy Corp (ASEC) using a Metal Organic Chemical
Deposition (MOCVD) process. The cells had a GaAs buffer
region of 6-9 pm thick, a 0.47 pm junction depth, a 0.06 pm
AlGaAs window, an AuZnAg front contact, a AuGeNiAg rear
contact, a Ti0,Al,03 AR coating, and were 200 pm thick. The
efficiencies of the ten GaAs/Ge cells ranged from 17.5 to
19.2%. Table I summarizes the 12 cell strings in the Ambient
1506
Coverglass &
Thickness (mils)
Cell Type
1 (18.0%)
Quartz 3.5
GaAslGe
2 (19.2%)
Quartz 6.5
GaAslGe
3 (18.5%)
Quartz 4.5
GaAslGe
4 (14.8 %)
Quartz 12
Thin Silicon
String # &
K7-314
5 (17.7%)
Quartz 6.5
GaAslGe
6 (17.5W )
Microsheet 4
GaAslGe
7 (18.0%)
Microsheet 12
GaAslGe
8 (18.2%)
Quartz 12
GaAslGe
9 (12.3%)
Microsheet 12
Si(Reference)
K4-314
10 (18.7 %)
Quartz 20
GaAslGe
11 (18.4%)
Quartz 20
GaAslGe
Cell Type
(Efficiency)
Coverglass &
Thickness (mils)
1 (16.1%)
Quartz 12
GaAs
2 (15.1 %)
Quartz 6
GaAs
3 (16.5 %)
Quartz 12
GaAs
4 (16.7%)
Quartz 6
GaAs
5 (15.4%)
Quartz 12
GaAs
6 (14.7%)
Quartz 6
GaAs
7 (16.4%)
Quartz 12
GaAs
8 (16.2%)
Microsheet 6
GaAs
111. DATARESULTS
Quartz 30
GaAslGe
1507
1.OEt08
>b 1 .OE+07
c
E
2
MeV
8 l.OEt06
z,
3
M ~ V
1.OEt05
200
400
600
800
1000
1200
ORMT NUMBER
ORBIT NUMBER
Figure 2. Total proton fluence versus orbit number for 6.8, 10.7, and
19.4 MeV protons. The 10.7 and 19.4 MeV proton fluences have
been multiplied by a factor of 10 to facilitate comparisons.
\
N
1.OEt08
>
1.0L+07
1.OEt06
1.OEt05
v
W
W
1.OEt04
1.OEt03
200
400
600
800
ORBIT NUMBER
1000
1200
Figure 3. Electron fluence per orbit versus orbit number for 1.1 and
7.5 MeV electrons.
1508
1 .OEt 14
O.OEtO0
1.OE+13
tl
1.OEt12
-1.OE+09
l.OE+l 1
5 l.OE+lO
0.08
-1.5E+09
0.075
-2.OE+09
1.OEt09
-I
3
1.OEt07
2
E
$
.,
0
1.OE+08
20.085
-
-5 .OEt08
(Y
3 0.09
I
.
.
1 0.07
-2.5Et09
9 0.065
-3.OE+O 9
;
2
1 .OEt06
0.06
1.OEt05
200
400
600
800
1000
1200
4
0
200
400
600
800
ORBIT NUMBER
1000
1;
-3.5Et09
IO
ORBIT hUMeER
Figure 4. Total electron fluence versus orbit number for 1 . 1 and 7.5
MeV electrons.
Figure 5 . Maximum power per cell versus orbit number for the thin
silicon K7-314 cell under 12 mils quartz coverglass and the GaAslGe
cell under 4 mils CMX coverglass. The total 10.7 MeV proton
fluence is inverted and plotted on the secondary y-axis.
1509
100
70
200
400
600
800
1000
1200
ORBIT NUMBER
90
a5
75
200
400
600
800
1000
1200
ORBIT NUMBER
The initial rapid drop in the thin silicon solar cell performance is due to the fact that it has a back surface field. The
back surface field is particularly susceptible to radiation
damage [ll] and usually becomes ineffective after a year in
1510
102
E
100
98
b
L
96
94
50
170
Refsrence 100
82
200
400
600
800
1000
1200
200
400
600
800
1000
1200
ORBIT NUMBER
ORBIT NWBER
Figure 10. Percentage of BOL maximum power per cell versus orbit
number for four annealing GaAs solar cells under 6 mils coverglass.
Figure 11. Percentage of BOL maximum power per cell versus orbit
number for four annealing GaAs solar cells under 12 mils Quartz
coverglass.
the pulse heater string, 77% BOL for the constant heater
string, and 75% BOL for the reference string. Figure 11
reveals the forward biased string has retained 88 % of its BOL
maximum power compared to 87%, 86%, and 84% BOL
maximum power for the pulse heater string, the constant
heater string, and the reference string respectively. Passing
current in the forward direction through the cells results in a
different voltage drop across each cell. Localized low resistance paths in a solar cell will result in very high current
densities at that point which will locally anneal out damage
sites. The forward bias and the pulse annealing experiment
were designed to operate at 190" C and 250" C, respectively,
for a half an hour each week. The pulse heater experiment
uses a heater under the cells to achieve annealing and thus is
not as effective in annealing out localized dislocations or
trapping centers.
A comparison between quartz and microsheet coverglass
shielding revealed essentially no difference between the two
materials in terms of degradation rates. Figure 12 shows the
percentage of BOL maximum power per cell versus orbit
number for four GaAs/Ge cells behind 4.5 mils (string #3) and
12 mils (string #8) of quartz and 4 mils (string #6) and 12
mils (string #7) of microsheet. The cells under 12 mils
essentially tracked one another with the CMX ending with
87 % of its BOL maximum power versus 85 % BOL maximum
power for the quartz. The cell under 4 mils CMX finished
with 77% of BOL maximum power versus 75% of BOL
maximum power for the 4.5 mils of quartz.
We also saw no difference between soldered and welded
electrical interconnects. During each eclipse, the temperature
in the ambient panel dropped approximately 140" C (90" C to
-50" C). There was no indication that the thermal stress
created during each eclipse over the 1067 orbit mission
degraded either the welded or soldered electrical interconnections.
1511
We found essentially no difference in the degradation of
maximum power per cell for GaAslGe solar cells under
similar thicknesses of quartz and microsheet coverglass. There
were also no differences observed between soldered and
welded interconnects.
pm 0.95
6
a 0.9
s
5
s
V. ACKNOWLEDGEMENT
0.85
0.8
0.75
VI. REFERENCES
0.7
200
400
600
800
1000
1200
ORBIT NUMBER
Figure 12. Percentage of BOL maximum power per cell versus orbit
number for four GaAs/Ge solar cells under 4.5 and 12 mils quartz
coverglass and 4 and 12 mils microsheet coverglass.
IV. DISCUSSION
& CONCLUSIONS
We have shown that the space radiation environment is
very dynamic, and that extreme care must be used in predicting solar cell degradation using constant particle fluences per
day derived from static models. Solar cell degradation showed
good correlation with changes in measured proton fluences.
Electron fluence levels did not appear to effect solar cell
performance, even after the multitude of geomagnetic storms
during orbits 600 to 900 increased the 1.1 to 7.5 MeV
electron fluence per orbit by several orders of magnitude.
However, it should be noted that during an extended mission,
these high energy electron fluences could eventually degrade
performance.
GaAslGe solar cells performed better than two different
types of silicon solar cells under equivalent coverglass
shielding. The thin silicon K7-3/4 solar cell degraded much
faster in the beginning of the mission due to its back surface
field. However, after fourteen and a half months, the K7-3/4
still had a higher maximum power per cell than the silicon K43/4 reference solar cell.
A comparison of like GaAs/Ge solar cells under different
quartz coverglass thicknesses showed that increasing the
coverglass thickness from 6.5 mils to 12 mils resulted in a
12 % increase (75 % to 87 %) in percentage of BOL maximum
power at the end of the mission. Increasing the quartz thickness from 12 mils to 30 mils resulted in only a 4 % increase
in percentage of BOL maximum power at the end of the
mission at almost triple the weight penalty.
Results from the Annealing panel showed that forward bias
annealing outperformed both constant heating and pulsed
heating biasing techniques for cells under 6 and 12 mils
coverglass. From an economic and practicality standpoint, the
forward biasing annealing method is much simpler and less
costly than methods using external heaters in providing high
annealing temperatures.
ABSTRACT
Due to its direct bandgap and high optical absorption
coefficient, GaAs offers the potential for ultrathin (< 10
micron) solar cells. Such cells have a very high output power
per unit mass, and applications both as single junction cells
and as top cells to mechanically stacked multijunction cells.
This paper reports the development of ultrathin GaAs solar
cells, with results up to 20% AM0 1 sun, based on
conventional coverglass and interconnect techniques.
Bevice Structure
The epitaxy (Figure 1) is similar to that for conventional
GaAs or GaAs/Ge cells, but includes a rear AlGaAs etch stop
layer, under a bifacial-style structure. A highly doped
(> 1E18,n-type) rear GaAs contact layer is grown under a thin - rear window/carrier reflector of AIGaAs, for which the
Aluminum mole fraction is chosen to give strong reflection
and good window properties, combined with good growth. A
conventional GaAs cell structure is grown thereafter, with a
2-4 micron, moderately doped (1E17<n< 1E18) base under
a thinner (0.5-1 micron), higher doped (2E18<p< 1E19)
emitter, a thin (optimally a quarter-wave thickness), high
Aluminum mole-fraction, AIGaAs window and a GaAs
contact cap layer.
INTRODUCTION
It has long been known that, for power- or area-limited
applicationsin space, GaAs solar cells are very attractive, due
to their high beginning of life efficiencies and superior
radiation resistance and temperature coefficients. GaAs cell
technologies have been developed for many years, both LPE
[refs 1-41, and the more advanced MOVPE [ref 5). However,
the advantages of early GaAs cells were somewhat offset by
their higher unit cost, increased handling requirements(due to
size limitations and high fragility compared to conventional
silicon cells), and higher optical absorptance - and hence
operating temperatures. One way of addressing these issues
has been growth of GaAs on a foreign substrate such as Ge which is now a commercially competitive reality [ref 61.
P+c.P
p N G a U Window
n GaAs 8.-
1399
&ice
F-
.
SOLAR CELL PHOTOVOLTAIC CURRENT-VOLTAGE PLOT
A W , 1 sun. 25 'C
Cell E2011 3 EEV 8 ym UIw.lhln G U S
01
02
03
04
05
00
07
08
OQ
ID
MLTS
10
(1
-5
-200
-100
100
200
300
400
Temperature IC
Bonding layer evaluation has been carried out using both LPE
and MOCVD grown GaAs solar cell material (ie AR coated
wafers with no metal contacts). No discernable difference in
bond characteristics has been observed between the two types
of structure. In order to bond to actual solar cells, grooves
have been provided within the coverglass, to accommodate the
collection grid pattern. Initial trials have used development
cells, and coverglasses (without any front surface ARC) with
recesses manufactured by diamond sawing; the resultant
coverglasses had wide grooves, with over I2 % of the cell area
in non-contact (compared to the 2-396 grid obscuration). First
results indicate that good bonding has been achieved, with
little degradation. Figure 6 shows an early result, and Table
1 shows the photovoltaic parameters before and after bonding
for this cell. The drop in current is attributable to: (i) the
relatively large (12%) non-contact area, which is very lossy
since these initial coverglasses have a rough, sawn finish, with
no light-bending; and (ii) non-optimised anti-reflective
coating. The initial measurement was made with a standard
SiN, coating which was stripped, before the DLAR applied
prior to bonding. The reflection plot shown in Figure 7
demonstrates that there is little optical change from bonding.
1401
__ -
a 2000
D i r e c t Glassed ( 3 2 0 C )
CONCLUSIONS.
High efficiency, ultrathin GaAs solar cells have been
demonstrated. Directly bonding of CMG coverglass to GaAs
solar cell materials, with good adhesion has been
demonstrated. Initial bonding trials of GaAs solar cells have
proved successful, and a first stage optimisation, to reduce
obscuration losses, is in progress. Directly glassed ultrathin
GaAs solar cells show promising potential for mass-critical
flights, and as components into mechanically stacked multijunction cells.
Interconnection
ACKNOWLEDGEMENTS
This work has heen carried out with the support of the
procurement executive, Ministry of Defence.
The development of CMG was carried out by PST in
conjunction with Pilkington Group Research, under contract
to DRA Farnborough.
REFERENCES
1.
\ /
COVEAGUSS
2.
Fig 4. Interconnection schematic for an Ultra-Thin Cell.
I402
3.
4.
5.
6.
7.
8.
9.
10.
1403
ABSTRACT
We have studied the light-induced degradation and
saturation behavior of component cells for double- and triplejunction a-Si alloy solar cells. Top a-Si alloy cells reach
saturation after 300 hours of one-sun light soaking with 1012% degradation. Both middle and bottom a-SiGe alloy cells
show signs of saturation at around 600 hours under simulated
light illumination appropriate for these cells; the degradation
after lo00 hours of exposure is 16-18 %. Based on this study,
it is estimated that double-junction cells will degrade by 1218 %. and triple-junction cells will degrade by 10-18 %.
Experimental results also reveal that the degradation of top
cells, particularly in open-circuit voltage, is determined by
the bulk properties instead of the interface between p and i
layers. Further improvement of bulk intrinsic layer quality of
top cells can reduce the degradation of multijunction solar
cells.
INTRODUCTION
Amorphous silicon (a-Si) alloy solar cells have
received a great deal of attention for their potential as lowcost energy source. It is well known that prolonged lightexposure lowers the efficiency of these cells and extensive
studies have been carried out [l] to investigate light-induced
degradation in both a-Si alloy materials and solar cells.
cells for
devica
97 1
devicu
TODcells
It is surprising to observe that the top cells,
although very thin, degrade by 10-12 %, of which the
degradation in V,w is about 5%. Note that the V,, of thin top
cells on the specular and textured substrates in this work are
0.98 and 0.94 V, respectively. Obviously, reducing the
degradation of WO, of top cells can improve multijunction
solar cell stability.
Table I. Characteristics of the component cells in the initial state. The top cells are measured under global
AM1.5 illumination, and the middle cell with global AM1.5/530 nm cut-on filter. The bottom cells for
double- and triple-junction cells are measured with global AM1.5/590 nm and global AM13610 nm cut-on
filters, respectively.
Sample
Designation
Pmax
(mW/cm*)
Jsc
(mNcm2)
voc
FF
(VI
5344 (Si)
top/double
6.84
10.3
0.94
0.71
5342 (Si)
top/triple
4.39
6.50
0.94
0.72
5343 (SiGe)
middldtriple
3.51
7.02.
0.77
0.65
5351 (Si)
middle/mple
4.06
6.44
0.90
0.70
5346 (SiGe)
I;ottom/double
4.16
9.80
0.68
0.63
5345 (SiGe)
bottom/triple
3.65
8.59
0.68
0.63
972
1.1
-*a:---.---
*.:-.
0
h
0.9
.C(
a
.
:
:
*
.
mm
ia
.r(
0.7
. = - 0 - 5 3 4 4/top
=
0.5
- - 0 - 5 3 4 4 /top
= - * = 5 3 4 6 /bottom
5 34 6 /bottom
fi
.0.5
1o-2
1 oo
1 o2
1 o4
1.1
10.7
-- 0-Q-
0.5
1 o-2
5344ltop
5 34 6 /bottom
0.5
1 oo
1 o2
1 o-2
1 o4
1 oo
1 o2
1 o4
Fig. 1. (a) Normalized efficiency , (b) short-circuit cwent density, (c) open-circuit voltage. and (d) fill factor of
the top and bottom cells for the double-junction configuration as a function of light-soaking time.
973
1.1
1.1
BN
?j
0.7
-*
--*-5342/top
--W=5343/middle
-I -5345/bottom
*-be 5 3 4 3 /middle
0.5
lo-*
-5345bottom
1
oo
o2
0.5
1 o4
1 oo
1 o2
1.1 7
' ,.dq-"l
a
.
g0.7
e
d
i?
0.5
:;: 5343/middle
53421top
- * -5345/bottom
--e5 34 2 4 0 p
5343/m idd Ie
--- --5345/bottom
H
1 o2
Fig. 2. (a) Normalized efficiency , (b) short-circuit current density, (c) open-circuit voltage, and (d) fill factor of
the top. middle and bottom cells for the triple-junction configuration as a function of light-soaking time.
974
Table 11. Characteristics of the top cells in both initial and saturated states. The top cells are measured under
global AM1.5 illumination.
Sample
5340 (Si)
5342 (Si)
SubstJate
specular
textured
State
Pmax
(mW/cm2)
Jsc
voc
(cm2)
(VI
FF
initial
4.92
6.67
0.98
0.76
saturated
4.37
6.62
0.93
0.7 1
degradauon(%) 11.2
0.8
5.1
6.6
initial
4.39
6.50
0.94
0.72
saturated
3.90
6.44
0.89
0.68
degradation(%) 11.2
0.9
5.3
5.6
Table 111. Characteristicsof a-Si and a-SiGe alloy middle cells in both initial and degraded stam. The middle
cells are degraded and measured under the illumination of global AM13530 nm cut-on filter.
Sample
5351 (Si)
5343 (SiGe)
State
Jsc
Pmax
(mW/cm2)
voc
FT
(mA/cm2)
cv)
4.02
6.44
0.90
0.70
3.29
6.22
0.87
0.61
degradation (%)
18.2
3.4
3.3
12.9
initial
3.51
7.02
0.77
0.65
2.89
6.85
0.74
0.57
degradation(%) 17.7
2.4
3.9
12.3
initial
975
ACKNOWLEDGEMENTS
Middle celh
We have also assessed the applicability of using aSi alloy for the intrinsic layer of the middle cell in the triplejunction configuration. The comparison between a-Si and aSiGe middle cells is listed in Table III for both initial and
degraded states. Although the a-Si middle cell reaches
saturation sooner than the a-SiGe cell, both cells show 1718% degradation after lo00 hours of light soaking. Another
difference between these two cells in the degraded state is that
a-Si cell has a higher overall efficiency of 3.29% than the aSiGe, 2.89%. However, the J,, of a-Si cell is only 6.22
mA/cmZ, which is much lower than that of the a-SiGe, 6.85
mA/cm2. Higher ,J for an a-Si middle cell requires a thicker
intrinsic layer which will cause more degradation.
From a stability study of another set of a-Si middle
cells with different thicknesses (3400 A - 4800 A), we found
that under no case could the middle cells produce more than 7
mA/cm2 of current density (under 530 nm cut-on filter) in the
degraded state after 1000 hours of one-sun light soaking.
Note, however, that this condition of light soaking is more
severe than that for the middle cells in the triple-junction
configuration, which may overestimate the degradation of aSi middle cells. A more systematic study of degradation of aSi cells with different thicknesses and preparation conditions
is required to further study this issue. We believe that 7
mA/cm2 of J, in the degraded state is necessary to obtain
stable efficiencies higher than 12% for triple-junction cells.
CONCLUSIONS
From a study of the stability of the component cells
of multijunction a-Si alloy solar cells. we show that it is
possible to reduce light-induced degradation to approach 1 1 %
by making the current top-cell limited. We also show that it
is necessary to incorporate germanium in the middle cell in
order to achieve a stabilized multijunction cell efficiency
above 12 %. Finally. we have demonstrated a stabilized
small-area efficiency of 11.2 % using a dual bandgap doublejunction structure.
REFERENCES
[l]See, for example, "Amorphous Silicon Materials and Solar
Cells", AIP COP$ Proc. No. 234, edited by B. Stafford
(American Institute of Physics, New York, 1991).
[2]J. Yang. R. Ross, R. Mohr, and J. Foumier. "Status of
fluorinated amorphous silicon alloy multi-junction solar
cells", Materials Research Society Symposia Proc. 70, p.475
(1986).
[3] J. Yang and. S. Guha, "Double-junction amorphous
silicon-based solar cells with 11% stable efficiency". Appl.
Phys. Lett. 61, 2917(1992).
[4] S. Guha, Final Report, SERInP-211-3918,
the National
Renewable Energy Laboratory, Golden. Colorado, 1990.
[5]S . Guha, J. Yang. A. Pawlikiewicz, T. Glatfelter. R. Ross,
and S. Ovshinsky, "Band gap profiling for improving the
efficiency of amorphous silicon alloy solar cells", A p p l .
Phys. Lett. 54, 2330 (1989).
[6]A. Banerjee, S. Guha. A. Pawlikiewicz. D. Wolf, and J.
Yang, "Light-induced degradation in a-Si:H alloy solar cells
at intense illumination", AIP Conf. Proc. No. 234. 186
(1 99 1).
[7] L. Yang. L. Chen, and A. Catalano, "Intensity and
temperature dependence of photodegradation of amorphous
silicon solar cells under intense illumination", Appl. Phys.
Lett. 59, 840 (1991).
[8]S. Wagner, X. Xu, X. R. Li, D.S. Shen, M. Isomura. M.
Bennett, A. E. Delahoy. X. Li. J. K. Arch, J.-L. Nicque, and S.
J. Fonash, "Performance and modeling of amorphous silicon
solar cells soaked at high light intensity". Proc. of 22nd IEEE
Phofovolfuic Specialists Conference, p.1307 (1991).
976
ABSTRACT
1-MeV electron irradiation effects on MBE-grown
InGaAs and AlGaAs solar cells have been examined in
comparison with our previous results for radiation
damage of InP and GaAs solar cells in order to clarify
radiation -resistance of advanced space solar cells.
Moreover, 1-MeV electron irradiation results of several
space solar cells such as InP, InGaP, InGaAsP, GaAs,
AIGaAs, InGaAs, Si, Ge, and CulnSe2 cells have also
been analyzed by considering their damage constants,
bandgap energies and optical absorption coefficients.
We believe that this study will provide a unified model for
radiation-resistance of advanced space solar cells.
INTRODUCTION
One of the key points for developing space solar cells
is to improve their radiation-resistance. Previously, the
InP and CulnSe2 solar cells were found to be superior
radiation-resistant than Si and GaAs solar cells [1,2].
These results show that new radiation-resistant materials
are expected to be found and advanced space solar cells
with hig h-efficiency and radiation-resistance are
expected to be developed. Material-oriented studies
should be carried out in order to design superior radiationresistant solar cells.
In this study, 1-MeV irradiation effects on advanced
space solar cells have been examined and their radiationresistance has been analyzed by a unified model in which
their damage constants, bandgap energies and optical
absorption coefficients are considered.
0.8
c
L
0.6
InGaAs
EX PERIMENTAL
The samples used in this study were heteroface p-on-n
AlxGal -xAs (x= 0.2-0.35) and homo-junction p-on-n
In G a l - y A s (y = 0.15) solar cells grown by MBE
(dlecular Beam Epitaxy) on n-GaAs substrates. Their
junction depths were about 0.3pm.
Typical AM1.5
efficiency 7 , short-circuit current density Jsc, opencircuit voltage Voc and fill factor FF of A10.3Ga0.7As
solar cells were 14.6%, 16.2mA/cm2, 1.28V and 0.707,
respectively.
Typical 7 , Jsc, Voc and FF of
ln0.1gGa0.85As solar cells were 13%, 23.55mAkm2,
1014
id5
ld6
2149
CH3365-4494/0000-2149 $4.00 0 1994 IEEE
semiconductor solar cells is explained by their directtransition band structures, a unified model should be
proposed in order to clarify variety of radiation-resistance
for solar cells.
A UNIFIED MODEL O F RADlATlO
RESISTANCE OF SPACE CELL5
I?
'
!2
6
n 10
Id4
, , , ,,,, 1
, ,
,,,,,I
1oI6
10
(11
(cm-2 )
- 1111x [ { ( ~ p ~ +d ~~ p- )~ x P ( -xj)
~p = [qF LP/{( 2 ~ p 2
((SpLp/Dp)COSh(Xj/Lp)+ sinh(xj/Lp))}/{(SpLdDp)
sinh(xj/Lp) + msh(xj/Lp)} - pLpexp(- xi)],
(2)
where suffixes p and n show carriers (hole and electron),
Xj is the junction depth, F the photon flux, (Y the
absorption coefficient, S the surface recombination
velocity and D the diffusion coefficient for the minority
carrier. P hotogenerated current density Jd, collected
from the depletion region, is
Jd = qF[1 - eXp(- W)]eXp(- Xj),
1d4
ldS
1-MeV electron fluence (cm-2 )
(3)
(4)
2150
Jx = f(Eg,
, L) Jp + Jd + Jn.
(5)
Bandgap energy e f f e c t s
0.8
0.6
0.4
1-MeV electron
0 = 1015c-2
Xj = 03ym
Calculated
0.2
1000
'
'
'
"
'"I
10000
'
'
'
'
".'I
100000
'
'
1000000
Calculated
I
2151
SUMMARY
1-MeV electron irradiation effects on MBE-grown
InGaAs and AlGaAs solar cells were examined in
comparison with our previous results for radiation
damage of InP and GaAs solar cells in order to clarify
radiation-resistance of advanced space solar cells.
Moreover, 1-MeV electron irradiation results of several
space solar cells such as InP, InGaP, InGaAsP, GaAs,
AIGaAs, InGaAs, Si, Ge, and CulnSe2 cells were also
analyzed by considering their damage constants,
bandgap energies and optical absorption coefficients. It
is found in this study that 1) Differences of radiationresistance of GaAs-based solar cells are explained by
considering bandgap energy effect on solar cell property
degradation; 2) Superior radiation-resistance of CulnSe2
cells is understood by higher optical absorption
coefficient of CulnSe2; 3) Better radiation-resistance of
InP-based solar cells is explained by lower damage
constants of InP-related materials.
We believe that this study will provide a unified model
for radiation-resistance of advanced space solar cells.
InP
0.8
1
0.4
'
10'~
10 IS
1Ol6
)
REFERENCES
x 10-8
100
0.2
0.6
InP Composition
0.4
0.8
Fig. 7. Changes in damage constant KL for minoritycarrier diffusion length for InP-related materials
and GaAs as a function of InP composition.
2152
ABSTRACT
INTRODUCTION
The Brazilian Complete Space Mission satellite is
the first space engine designed and constructed in Brazil. The
first data collection satellite (SCD1) is a real time repeater of
environmental data collected by ground platforms. The SCDl
is in orbit since 9 February, 1993 in a circular orbit 750 Km
high, with a period of 100 minutes, at an inclination angle of
250 with respect to the equation plane. In this orbit the
2232
E,[T]= E,[O]+XT'/(T+ p )
Figure 1. Electrical circuit of the solar cell
experiment.
THEORETICAL FUNDAMENTS
Two problems had to be solved to analyze the SCE
behavior: how to extract the current versus voltage
characteristics from the one orbit SCE output voltage versus
temperature curve and how to obtain all the electrical
parameters used to adjust the current versus voltage curve.
The characteristics current versus voltage of a solar
cell can be represented by the double exponential model [3,4],
see eq. (1).
I,,
= Cs,T3exp(--)
E,
kT
(2)
(3)
2233
(8)
DISCUSSION
a
m
m
TIME (niinute)
im
TIME (iiimute)
2234
REFERENCES
[I] D.H. Walker and R. L. Statler, "A satellite experiment to
study the effects of space radiation on solar cell power
generation." Solar Cells, 23, 1988, p. 245-250.
[2] C Huang et al., "Solar cell calibration experiment on
Chinese scientific satellite." Eighteenth IEEE PVSC, 1985,
pp. 634-637.
[3] H.J. Hovel, Solar Cell in: R. K. Willardson and A.C. Beer,
Semiconductor and Semimetals. Academic Press Inc., New
York, 11, 1976.
[4] H.S. Rauschenbach, Solar cell array design handbook,
Van Nostrand Reinhold Company, New York, 1980.
[8] N.Veissid and A.M. Andrade, "The I-V silicon solar cell
characteristic parameters temperature dependence: an
experimental study using the standar deviation method."
Tenth E.C. Photovoltaic Soalr EnergV Conference, 1991, pp.
43-47.
[9] H.Y. Tada et al. Solar Cell Radiation Handbook, Thrid
Edition, JPL Publication 82-69 (1982).
2235