1. 1
High Voltage Photovoltaic Cells
By
Roni Pozner
Carried out under the supervision of
Prof. Yossi Rosenwaks
Dep. of Physical Electronics
Tel-Aviv University

2. 1. CPV operating conditions:
1. Solar concentrators in CPV systems.
2. Current mismatches in PV modules.
3. Electrical and thermal coupling in CPV systems.
2. Series resistance losses in CPV systems:
1. Effects of series resistance on cell efficiency.
2. Series resistance components in MJ cells.
3. VJ’s series resistance.
3. Vertical Multi Junction (VMJ)- Basic Concepts
4. VJ Vs. horizontal cells in CPV systems
5. Theory of Solar Cells
6. Advantages & Disadvantages
7. Simulation
8. Fabrication
Outline

3. 15 x15 km
Israel consumes today
10 GW of electricity,
thus it needs only 200
km2 of solar panels to
supply most of its
electricity needs !
With Concentrated
PV Only 10x10 km2
are required

4. Theory Of Solar Cells
Illumination Absorption
Separation
Collection
    )
(
)
(
min 





 collection
separation
absorption
ation
illu
N
q
I 




5. Photovoltaic Cells – Basic
Concepts
 
sh
s
L
nKT
IR
V
q
ph
L
R
R
I
V
e
I
I
I
s














1
0
C
I
I ph
sc 

IN
M
M
IN
MAX
P
I
V
P
P



oc
sc
m
m
V
I
I
V
FF 

6. Vertical Junction (VJ) Cell
Vertical junction
No front grid, minimal inactive area
Decouple optics vs. electronics
High Voltage cell
Reduce series resistance effect
Parallel connection- reduce mismatch effect

7. Advantages & Disadvantages

8. Advantages - Cell
• Active Area Fraction - Less shading
“Back Contact” “Standard”

9. Decoupling Effects
Orthogonal carrier generation (optics) & carrier collection (electronics)
Collection efficiency independent of wavelength
High doping of N+,P+ layer, hence larger Voc

10. Record Devices-2008

11. • Lower resistance due to:
– Control over length and doping of N+,P+ layers.
– Photo conductivity phenomena.
Advantages - Concentration (cont.)

12. Solar concentrators in CPV systems
Fresnel Lenses Parabolic reflectors

13. Limitations of CPV technology
• Series resistance limits concentration
• Cell thickness: optics vs. electronics vs. mechanics
• Front grid, busbar: shading loss
• Illumination non-uniformity: mismatch loss

14. Series resistance
under light concentration
• Our cell efficiency still rises under 5000 suns. While ordinary Multijunction cells as
the highest efficiency under 500 suns and begin too decries under higher
concentrations. silicon solar cells as the highest efficiency under 200 suns.
• The reason for this phenomenon can be due to photoconductivity effects as well as
the low currents in the cell.
• This is because the mobility and the concentration of carriers are varying with
location and light concentration conditions.
• So in order to calculate the resistivity of the cell under concentration conditions we
need to take the mobility and the concentration of carriers as a function of their
location from the Sentaurus simulations.

15. MJ cell- Series Resistance
Components1
1. Electrode Resistance (RSE)
2. Contact Resistance
3. Lateral Resistance (RSL)
4. Layer (InGaP) Resistance
5. Tunnel Resistance (RT1)
6. Layer (InGaAs) Resistance
7. Tunnel Resistance (RT2)
8. Ge Substrate Resistance
Main Resistance Components:
RSE, RSL, RT1, RT2.
1K. Nishioka et al. / Solar Energy Materials & Solar Cells 90 (2006) 1308–1321

16. Vertical Diode- Series Resistance
Components
1. Electrode Resistance (RSE)
2. Contact Resistance (RSC)
3. p+ Layer Resistance (RSP)
4. Bulk Resistance (RSB)
5. n+ Layer Resistance (RSN)
Main Resistance Component: RSB.
p+ p n+
RSB

17. Series Resistance Losses
Neglecting the shunt resistance-
s
L
L
ph
R
I
I
I
I
q
nKT
V 


 )
1
ln(
*
0
s
L
L
ph
L
L R
I
I
I
I
q
nKT
I
P
2
0
)
1
ln(
* 



Rs
Rsh
Isc
Id
IL
V




2
L
Rs I
P 

18. Photoconductivity Effects
     
 
n
p n
n
p
p
e
y
x 

 




 0
0
,
 
C
f
RSB 
     
C
y
x
y
x
y
x ph ,
,
,
, 0 

 

Horizontal cells:
Vertical Diodes:
   
C
y
x
y
x ph ,
,
,
0 
 
σph Changes significantly with the
illumination
Const
RS 

19. Simulation Results
 
 



L
d
SB
dy
C
y
x
dx
w
C
R
0
0
,
,
1

R
SB
[Ω]
100
101
102
103
104
0
0.2
0.4
0.6
0.8
1
1.2
1.4
Concentration

20. Vertical Junction Under
Concentration
2Slade et al. / Proceedings of the SPIE, Volume 5942, pp. 236-243
100
101
102
103
104
15
20
25
30
35
Concentration
Efficiency
[%]
Vertical Diode
Best Si cell2

21. Grid Pitch Dependence of the Series
Resistance and Shadow Loss1.
Grid Pitch
1K. Nishioka et al. / Solar Energy Materials & Solar Cells 90 (2006) 1308–1321

22. Efficiency for Concentration
(various cells)
Spectrolab- C1MJ2
2Kinsey et al. / Prog. Photovolt: Res. Appl. 2008; 16:503–508.
Best Silicon cells
4Mulligan et al. / Proceedings of the 28th IEEE Photovoltaics Specialists Conference.
3Slade et al. / Proceedings of the SPIE, Volume 5942, pp. 236-243 (2005).
10
1
10
2
22
23
24
25
26
27
Concentration
Efficiency
[%]
Best Si Cell3
(1.35 cm2
)
SunPow er Chipsize4
(0.0529 cm2
)

23. Nishioka et al., Solar Energy Mat. Solar Cells, 2006
Limit for High Concentration
High concentration
→ High current
→ High Rs loss
Reduce grid pitch
→ High shading loss
Spectrolab C1MJ
Kinsey et al., Prog. Photovolt: Res. Appl. 2008 0 100 200 300 400
23
24
25
26
27
28
Concentration
Efficiency
[%]
Measured
Model
SunPower Si cell
Mulligan et al., Proc. 28th IEEE PV Specialists Conf.

24. High Output Voltage Module
N cells
Series Connection of N Vertical Cells:

25. N Junctions
VMJ Cell
Junctions in series = Cell
• High voltage, low current
• External contacts at edges
Easy to convert to back contacts (external wrap-around)
Insulator Back contact

26. • Easy MIM connection (Monolithic process)
• High output voltage of each module.
• Small area of each module leads to uniform illumination on
each one of them.
Wire
Vertical Cell
Horizontal Cell
Advantages - Module (cont.)

27. Due to the relatively high output voltage of each module, it
can be connected in parallel to the other modules (instead
of series connection like in horizontal cells) this offers the
following advantages:
• Voltage coupled instead of current coupled response Better response to partly shading
conditions
• Less sensitive to non-uniform flux avoiding the use of homogenizer in concentrated PV
(Gideon’s Lecture…)
Advantages - Module (cont.)

28. VJ module under non-
homogenous illumination
V1
V2
VN
Vm
+ If the dimensions of the VJ module
are small comparing to the change
in illumination:
N
ph
ph
ph I
I
I ,
2
,
1
, .. 


4cm
4cm
Y
X
contact
P
N
contact
40 um
dX
1
2
N
1cm

29. Illumination input:
Average concentration=444
Maximum concentration=1148
1.66% spillage

30. VMJ Mismatch Loss
0 50 100 150 200
22
23
24
25
26
Number of junctions in cell (N)
Efficiency
[%]
VMJ cell array
Uniform illumination: 27.72%
Non-uniform illumination: 25.86% (N = 40)
→
Series Mismatch Loss (relative)= 6.7%
Conventional cell array
Series Mismatch Loss > 80%
No homogenizer !

31. Efficiency for concentration, Width=43.6μm, Depth=60
surface recombination =100
10
1
10
2
10
3
10
4
20
25
30
35
Concentration
Efficiency
[%]
Parallel illumination
Multidirectional Illumination
Multidirectional Illumination, no surface tex turing

32. • High output voltage of each module enable the use of a more
efficient DC/DC inverter.
Advantages - Module (cont.)

33. Disadvantages
• Fabrication difficulties
– The depth of the cell is relatively large (H)
– Metal thickness need to be as thin as possible (X)
(Karmiel’s Lecture…)
H
X
X

34. Semiconductor has low resistivity, hence the two cells above modeled as
connected in parallel by conducting wires.
Hole current from top to bottom reduces the voltage of 1 close to the voltage of 2
(circular currents)
Voc of the device will be close to the voltage at the deepest point.
Open Circuit Voltage
(1)
(2)

35. As H getting bigger, Voc decrease and Isc increases => so while increasing H
result in absorbing more light, it also reduces Voc, hence, the efficiency will
begin to drop at some point.
H
• Voc
Voc is proportional to )
1
log(
H
Disadvantages (cont.)

36. 0.5um
Constant distance (between
junction and contact)
• Front Surface Recombination
Varying distance (between
junction and contact)
Most of the minority carriers in the vertical cell have a longer path compared
with the horizontal cell-more affected by surface recombination.
Most of the light is absorbed in the top 5 microns of the cell
back surface recombination is neglected in both the horizontal and
vertical cells due to very small carriers density generated there.
Vertical Cell
Horizontal Cell
Disadvantages (cont.)

37. • Contact Surface Recombination (Point Contacts):
Disadvantages (cont.)
Vertical Cell
Horizontal Cell
Disadvantages (cont.)

38. Simulations

39. Analysis
0
)
,
(
)
,
(
)
,
(
)
,
(
2
2
2
2






z
x
n
x
G
dz
z
x
n
d
D
dx
z
x
n
d
D n
n
Continuity equation:
)
)
(
exp(
)
(
)
(
)
,
( x
N
x
G 




 

Generation equation:
x
z
Light Intensity
Absorption Coefficient

40. Matlab Simulation
A Matlab simulation of varying cell depths:
The result gave us an initial intuition regarding the optimization that should be
done.
L
H

41. Sentarus Simulation

42. Optimization Factors
Bulk Material – N / P
Lifetime – SC / MC
Length
Depth
SR – Front / Bottom
SR - Contacts
Front Pyramids
Front Anti Reflective Coating
Back Reflective Coating
Back Diffused Mirror
Bulk Doping
N+,P+ Doping
N+,P+ Length
Metal Contact (Shading)
Temperature
Sun Angle
Flux Non-Uniformity

43. • The thickness of the thin film (ARC and RC) is determined by:
Phase Changes between the mediums has great
significance .
• Si3N4 is a good ARC due to two reasons:
– Si3N4 has n~2 and the best ARC need refractive index of
– Si3N4 is also very good for the passivation.
• SiO2 is not the optimal material for RC but it is built in the SOI substrate so we can
use it for our needs
Optimization of ARC ,
RC and texture
2 2
1
( 0.5)
2 4
m
m
d
n n
 


 
1.87
ARC air si
n n n
 

44. Optimization of ARC ,
RC and texture
Figure 18 - Absorption
depth for silicon [3]
Figure 17 - Spectral Radiation at
the Earth's Surface [3]
 The optimal width of ARC is for the
wavelength of 0.6um.
 The optimal width of RC is for the
wavelength of 0.9um .
This is due to:
- Sunlight intensity .
- Absorption probability.
Silicon-
Handle wafer
light
T1
T2
R1 T3
R2
SiO2
Figure 15

45. Optimization of ARC ,
RC and texture
 The thickness of the thin film (ARC and RC) is determined by:
2 2
1
( 0.5)
(1)
2 4
m
m
d
n n
 


 
Figure 19 - Anti-reflective coating
 The difference between
ARC and RC is the number of
phase Changes between the
mediums.
Texturing
 Reduced reflection.
 Better "light trapping”

47. Optimization of ARC ,RC and texture
Reflectivity without ARC layer is R=40% .
Reflectivity ARC (Si3N4 ) is R=15%.
Reflectivity with ARC (Si3N4 ) and RC is R=18%.
Transmitted light without RC is T=10%
Transmitted light with RC is T=6%. Relation between the refraction
index and the wavelength for a
silicon surface
Reflectivity for 40um depth cell with ARC & RC
coatings
0
0.2
0.4
0.6
0.8
1
0.3 0.5 0.7 0.9 1.1
Wavelength [um]
R(%)
Cell with ARC (Si3N4)
Cell with ARC (Si3N4) & RC
Cell without coatings
The high end of the
spectrum the graph is
not consistent due to
the reflectivity from the
second interface of air
and silicon at the
bottom of the cell.

48. Sentarus Simulation
12.5053 18.6185
18.2978 18.9897 21.3228
Bulk Material – N / P
Lifetime – SC / MC
Length
Depth
SR – Front / Bottom
SR - Contacts
Front Pyramids
Front Anti Reflective Coating
Back Reflective Coating
Back Diffused Mirror
Bulk Doping
N+,P+ Doping
N+,P+ Length
Metal Contact (Shading)
Temperature
Sun Angle
Flux Non-Uniformity

49. Sentarus Simulation
0
5
10
15
20
25
1.00E-02 1.00E-03 1.00E-04 1.00E-05 1.00E-06 1.00E-07
tau (Sec)
Efficency
(%)
L
H
Bulk Material – N / P
Lifetime – SC / MC
Length
Depth
SR – Front / Bottom
SR - Contacts
Front Pyramids
Front Anti Reflective Coating
Back Reflective Coating
Back Diffused Mirror
Bulk Doping
N+,P+ Doping
N+,P+ Length
Metal Contact (Shading)
Temperature
Sun Angle
Flux Non-Uniformity

50. Cell Optimization
12
14
16
16
18
18
20
20
20
22
22
2
2
22
22
22
2
2
24
24
2
4
2
4
2
4
2
5
Width [um]
Depth
[um]
50 100 150 200 250
20
40
60
80
100
120
140
Efficiency for Various cell sizes (1 sun)

51. Cell Optimization
30
30.5
30.5
31
31
31
31.5
31.5
31.5 31.5
32
3
2
32
32
3
2
.
5
32.5
3
2
.
5
33
3
3
Width [um]
Depth
[um]
30 40 50 60 70
40
60
80
100
120
140
Efficiency for Various cell sizes (1000
suns)

52. Sentarus Simulation
L
H
Bulk Material – N / P
Lifetime – SC / MC
Length
Depth
SR – Front / Bottom
SR - Contacts
Front Pyramids
Front Anti Reflective Coating
Back Reflective Coating
Back Diffused Mirror
Bulk Doping
N+,P+ Doping
N+,P+ Length
Metal Contact (Shading)
Temperature
Sun Angle
Flux Non-Uniformity

53. Sentarus Simulation
L
H
Bulk Material – N / P
Lifetime – SC / MC
Length
Depth
SR – Front / Bottom
SR - Contacts
Front Pyramids
Front Anti Reflective Coating
Back Reflective Coating
Back Diffused Mirror
Bulk Doping
N+,P+ Doping
N+,P+ Length
Metal Contact (Shading)
Temperature
Sun Angle
Flux Non-Uniformity

54. Sentarus Simulation
0
2
4
6
8
10
12
14
16
18
20
1
.
0
0
E
+
1
3
1
.
0
0
E
+
1
4
1
.
0
0
E
+
1
5
5
.
0
0
E
+
1
5
1
.
0
0
E
+
1
6
5
.
0
0
E
+
1
6
1
.
0
0
E
+
1
7
1
.
0
0
E
+
1
8
Na(Bulk) [cm-3]
Efficency
L
H
Na
Bulk Material – N / P
Lifetime – SC / MC
Length
Depth
SR – Front / Bottom
SR - Contacts
Front Pyramids
Front Anti Reflective Coating
Back Reflective Coating
Back Diffused Mirror
Bulk Doping
N+,P+ Doping
N+,P+ Length
Metal Contact (Shading)
Temperature
Sun Angle
Flux Non-Uniformity

55. Sentarus Simulation
17.3
17.4
17.5
17.6
17.7
17.8
17.9
18
18.1
18.2
18.3
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Nd+,Na+ [um]
Efficency
L
H
Bulk Material – N / P
Lifetime – SC / MC
Length
Depth
SR – Front / Bottom
SR - Contacts
Front Pyramids
Front Anti Reflective Coating
Back Reflective Coating
Back Diffused Mirror
Bulk Doping
N+,P+ Doping
N+,P+ Length
Metal Contact (Shading)
Temperature
Sun Angle
Flux Non-Uniformity

56. 5
10
15
20
25
30
35
0.5 1.0 1.5 2.0 2.5
Absorber Energy Gap (eV)
Solar
Cell
Efficiency
(%)
Ge
Si
CISe
CIS
CIGSe
CIGSSe
CdTe
GaAs
InP
CGSe
Cu2S
CdS
Theoretical limit for
ground-based solar
cells.
Comparison of record performances with theoretical efficiency limits.
Single Junction
Solar Cells
a-Si
Si-based
III-V based
CuIn-chalcogenide
Si and GaAs are
near theoretical
limits already.
Single-crystal devices
III-V Multijunction Cell record: 32%
Photovoltaic Opportunities

57. Junction Optimization
Optimize:
Junction geometry
Depth, width
N+, P+ width
Metal contact width
Material properties
Doping
Hole/electron lifetime
Surface treatments
Passivation
Front AR
Back reflector
Front texturing 40m
P 1016
P+ 1018
Double AR
Coating
Pyramids
Reflective
Coating
Process
dependent
1ms
h e
 
 
0.5 m
25–50
m
N+ 1019
0.5 m
Current analysis for Si

58. 0 0.2 0.4 0.6 0.8 1
0
0.5
1
1.5
2
x 10
-5
V [Volt]
I
[Amp] Auger coeff=6.7*10
-32
; Efficiency= 29.14%
Auger coeff=1.6*10
-30
; Efficiency= 29.00%
Auger coeff=1*10
-28
; Efficiency= 25.99%
43.6um*60um cell under x 1000 concentration for several Auger
coefficients. The line in blue is for the default Auger coefficient in
Sentaurus.

59. Scalloped sidewalls Simulations

60. Process Flow

61. 61
• How to fabricate the deep vertical PN junction?
• How to fabricate high aspect ratio trenches with
steep(?) side walls?
• How to deposit the contacts in these high aspect
ratio trenches?
The main Challenges:

62. Other VMJ Cells
Multi-wafer process, Si
40 junctions/cell
25.5 V
19.2% at ×2,500
Sater & Sater, 29 IEEE PV Specialists Conference, 2002
Green Field Solar
Our approach
Monolithic production of junctions in a single wafer
Optimal junction width < wafer thickness
Higher efficiency, better use of materials
Cost

63. Other VMJ Cells-Sliver
Multi-wafer process, Si
40 junctions/cell
25.5 V
19.2% at ×2,500
Sater & Sater, 29 IEEE PV Specialists Conference, 2002
Green Field Solar
Our approach
Monolithic production of junctions in a single wafer
Optimal junction width < wafer thickness
Higher efficiency, better use of materials
Cost

64. Another High Voltage Cell
Monolithic Inline Module (MIM)
Horizontal junctions, GaAs
25 junctions/cell
25.4 V
22% at ×200 (peak)
Loeckenhoff et al., IEEE 4th World Conf. PV Energy Conversion, 2006
Fraunhofer ISE
• Series resistance
• Large gaps- inactive area
• Front grid- shading
• Complex manufacturing
Our approach
Monolithic+ Vertical Junctions

65. Proposed basic Process Flow
SiO2
Handle
wafer
1. SOI Substrate
2. Trench fabrication
using DRIE process
3. Ion implantation of the
n+ side of the trench
4. Ion implantation of the
p+ side of the trench
5. Contact fabrication
using deposition
~40[um]
~0.5[um
]
~0.5[um
]
~40[um]
Previous
Cell
Next Cell
Generic concept
Can be applied to III-V material.
We will send you soon alternative process steps

66. 66
Process B (alternative)
SiO2
Active
surface
Handle
wafer
•
SOI
Substrate
•
trenches manufacture in
DRIE process
•
Doping by
diffusion p+ of
the left trench.
•
Doping by diffusion
n+ of the right
trench.
•
removal of the
barrier by DRIE
process
•
contact
manufacture by
deposition
Suggestions to solve the
difficulties in the PV junction
fabrication

67. In our original cell design there are several difficulties :
• The light that encounter the contacts was reflected and was lost.
• There is no proper way to overcome the surface recombination at the
contacts.
My suggestion:
Neighboring
cell
Bulk P
Contacts
ARC
p+ side n+ side
Neighboring
cell
SiO2-RC
silicon
The cell efficiency was
improved by 1% in comparison
to the original cell design with
2um metal contacts( without
surface recombination).
Proposed alternative
process step
localized contacts
Trade in between :
Higher passivated light absorption area (Green color)
And higher series resistance. Optimization will be done.

68. The advantages of this design over the original design are:
• Passivation layer on the sidewalls of the trench (ARC) to reduce recombination.
• Localized contacts at the bottom of the trench.
• Light can enter the cell from the trench sidewalls.
• Improved light trapping than the original cell design.
The disadvantages of this design are:
• The fabrication process is more complicated.
• The series resistance becomes higher than the original cell design and the portion that
becomes higher is the part that isn't effected with high light concentrations.
Sidewalls angle + Metals Simulations

69. Fabrication

70. The fabrication process
and the cell structure
• The new vertical cell design is a very
ambitious design. Almost all of the
fabrication steps are at the edge of the
technology abilities today.
p+ region 0.5um doping 1019
Contacts 1-3um
n+ region 0.5um doping 1019
p bulk region 39um doping 1016
Handle wafer
SiO2
Figure 5 – The vertical cell on a SOI substrate

71. Basic fabrication process
Active surface
SiO2
Silicon- Handle wafer
Active surface
SiO2
Silicon- Handle wafer
SiO2
Active surface
SiO2
Silicon- Handle wafer
SiO2
Photoresist
Active surface
SiO2
Silicon- Handle wafer
SiO2
Photoresist
Active surface
SiO2
Silicon- Handle wafer
SiO2
SiO2
Silicon- Handle wafer
SiO2
Silicon- Handle wafer
SiO2
Silicon- Handle wafer
A – SOI Substrate
B – Oxide growth
C – Covering with Photoresist
D – Exposure to light
E – Photoresist developing
F – DRIE (Deep Reactive Ion Etching)
G – Photoresist removal
SiO2
Silicon- Handle wafer
H - Implantations
J – Oxide removal
SiO2
Silicon- Handle wafer
I – Contacts fabrication
SiO2
SiO2
SiO2
SiO2
Mask
SiO2
SiO2
Figure 6 – The Basic fabrication process

72. 74
Why High aspect ratio?

73. High AR (Aspect Ratio) trench
fabrication methods
 Our requirements are high AR (Aspect Ratio) trench >20:1 (not standard
process)
 literature review results:
- Laser grooving.
- mechanical grooving.
- DRIE (Deep Reactive Ion Etching) methods (Bosch and Cryo processes).
Suitable for
our cell
Suitable for
contacts
Step
sidewalls
Aspect
ratio
depth
Technique
X
√
√
About 2:1
About 50um
Laser grooving
X
√
√
About 2:1
>60um
Mechanical grooving
√
√
√
>20:1
>60um
Bosch process
√
√
√
>20:1
>60um
Cryogenic process
Table 1- Abilities of the different techniques for the fabrication of the trenches

74. • There are other methods for DRIE ,but these methods are not published in the
literature because of confidentiality reasons.
• The DRIE process in TOWER:
High AR (Aspect Ratio) trench
fabrication methods
Mag x 4000 Mag x 15000 Mag x 100000
Mag x 100000
Mag x 300000
Mag x 5000
Mag x 28000
Stage
pecimen
SEM
Electron beam
Figure 7- Test
utline

75. • The Bosch process is the only commercial method for making high aspect ratio trenches, it
is necessary to check the effects of this process on the cell efficiency.
• Scalloped sidewalls - The Bosch process as two working cycles etching and passivating.
Scalloped are formed an the sidewalls of the trench.
• The changes in the cell efficiency from a cell without scalloping were decrease of less than
0.3%
High AR (Aspect Ratio) trench
fabrication methods
characterization
Figure 1 – characteristic
scalloped [7]
scalloping dimensions [9]
structure in the simulation

76. • Leg effect – the leg effect caused when the etching process reaches the SiO2 but
the etching process continue. This usually occur when we have trenches with
different widths on the same wafer.
• The changes in
the cell efficiency
from a cell without
the leg effect were
decrease of
less than 0.25%.
• The leg dimensions are 2x2 um The leg effect on a trench from tower
the structure in the simulation
High AR (Aspect Ratio) trench
fabrication methods
characterization

77. The most suitable methods for metal
deposition are:
• Advanced PVD Sputtering methods like
Highly Ionized Sputter Process.
• Atomic layer deposition (ALD).
High cost, rare academic use.
Deposition methods
for high AR trenches
Ta layer deposited with HIS
on the bottom and sidewall
scallops of a trench with AR
of 30:1
[
10
.]

78. • Sidewalls angel – are a byproduct of the Bosch process that can be caused by
wrong calibration of the machine.
• The angels for the left side of the figure are 92.66 deg and for the right side 87.44
deg. Thus angle where chosen because they are more than the average deviation
of the machine which is ± 1 deg.
• The cell efficiency improved from a cell with 90 deg sidewalls in about 0.1%
High AR (Aspect Ratio) trench
fabrication methods
characterization

79. Our Mask

80. 82
Work Environment
Testing and simulations were carried out using the
Synopsys TCAD Sentaurus Tool Suite:
1. Process Simulator (real physical models)
The creation of PN junction will be done using Ion Implantation:
Figure 11. Ion Implantation
Implantation Models
Monte Carlo
(atomistic)
Analytic
Based on point
response
distributions

81. 83
Work Environment
(a,b)



gas
ds
s
b
s
a
y
x
F
N
y
x
C d ))
(
),
(
,
,
(
)
,
(
Analytic Model:
 An ion beam incident at point (a,b) is assumed to generate
a distribution function F(x,y,a,b).
 In order to calculate the concentration C(x,y) at point (x,y),
the superposition of all distribution functions F(x,y,a,b) of all
possible points of incidence needs to be computed:
Figure 12. Ion Implantation Model Sentaurus Process [14]
Where Nd- total dose per
exposed area

82. 84
Work Environment
1. Process Simulator (real physical models)
2. Structure Editor (geometry based)
3. Device Simulator (electronic behavior)
4. Tecplot Viewer (result analyzer)
Important, saves money and time!
Structure
Editor
Process
Simulator
Device
Simulator

83. 85
Fabrication Issues
Figure 13. Effect of various implantation energies [Tecplot viewer]
N-Type Implant
~1018[cm-3]
Phosphorus
N-Type Implant
~1018[cm-3]
Phosphorus
P-Type Bulk
~1015[cm-3]
Boron
P-Type Bulk
~1015[cm-3]
Boron
P-Type Bulk
~1015[cm-3]
Boron
There are some major issues with conventional Doping
Techniques:
• Ion Implantation and Diffusion are isotropic processes,
this is bad for the profile. The deeper, the wider.
• Penetration depths reach a few microns at most at high
energy.

84. 86
There is no possible way to create a Vertical
PN Junction using conventional techniques!
Fabrication Issues

85. 87
According to the literature review that I conducted, the best
method for creating the Vertical PN Junction, is by Implantation
through a Trench Sidewall, apparently this method is being
used in the fabrication of both High Power and High Mbit DRAM
devices [6,7]:
Vertical PN Junction Fabrication Method
Figure 14. Implantation in Trench Sidewall [6]
Trench

86. 88
Trench
Trench
My Suggested Solution
[tecplot]
Figure 16. Ion Implantation through trench sidewall
Trench

87. 89
Process Flow - Simulation Results
[tecplot]
Previous
Cell
Next Cell

88. 90
Figure 18. Zoom in of Vertical Junction
0.5[um]
[tecplot]
Process Flow - Simulation Results

89. 91
[tecplot]
0.5[um]
0.5[um]
Process Flow - Simulation Results

90. 92
Verification
0.5[um]
Process
Simulator
Structure
Editor
Device
Simulator
The Process Simulator, which takes into account more
physical effects due to processing steps, verifies the
results yielded from the geometry based Structure Editor.
Device
Simulator
Same Results

91. 93
Verification
0.5[um]
SIMS Characterization (Solid State Institute, Technion):
Samples from Tower
Semiconductor LTD.

92. 94
Verification
0.5[um]
SIMS Characterization:
Micro – Cleaving from SELA
LTD.

93. 95
Verification
0.5[um]
SIMS Characterization Issue:
Implanted Sidewall
Bottom of trench

94. 96
• A new method for creating vertical PN junction cell arrays was
proposed.
• Fabrication Techniques required for the creation of a Vertical
PN Junction were introduced.
• A new simulation platform based on real physical models was
described.
• A novel solution for doping apposing trench sidewalls with
different dopants was tested for the very first time.
• This solution can be applied to other fields in the Micro
Electronic Industry, such as High MBIT DRAM devices and
lateral power devices.
Summary

95. 97
Substrate
Substrate will be SOI wafer with an active surface
equal to the thickness of the cell it self 25-50um.
For example, Icemostech company Inventory sample100 mm
Diameter wafer of SOI with Device Thickness of 35 + 1 um cost
about 200$
SiO2
Active
surface
Handle
wafer
35 + 1 um
0.5 -3 Ohm cm
400+5um
Ohm cm
1-100
10+0.5u
m
Icemostech:
Samples with trenches in the
active surface. Minimum
opening 3 microns. Aspect
ratio up to 1:40 cost ?
•
Formation of a texture
surface underneath the active
surface in order to increase
light trapping in the cell
(diffusive reflection).

96. 98
Initial Model-
Synopsys TCAD Sentaurus Process simulator

We used this concept to
implant both p+ and n+ doping
on each of the trench’s
apposing sidewalls, using the
Process Simulator.

Implantation Parameters:
- Tilt: +1 degree
- Dose: 1X10^17[cm-2],
- Energy: 120[KeV]
Implantation Model: The
concentration behaves like a
Gaussian function, Peak near
the surface, while reducing
exponentially deeper into the
substrate.

97. 99
P-Type Boron Concentration:
1X10^15[cm-3]
N+-Type Phosphorus Concentration:
~ 3X10^19[cm-3] (within 0.5[um] of surface)
~1.2[um]
~0.5[um]
~1.2[um]
~1.2[um]
~0.5[um]
Peak
0.7[um]
Implantation Parameters: Tilt: +1 degree,
Dose: 1X10^17[cm-2], Energy: 120[KeV]
Aluminu
m
Contact
Results:
P+-Type Phosphorus Concentration:
~ 3X10^19[cm-3] (within 0.5[um] of surface)

98. 100
Alternative Approach :
high voltage/high current
P- Type Substrate
P
P P
P- Type Substrate (thin and cheap)
Metal
Contact
Trench
Dopin
g
Trench
n+
n+
Doping the walls of the trenches by diffusion of one type doping

99. 101
Alternative Approach
(Cont.)

100. 102
Process Flow (Cont.)

101. 103
Connectivity Substrate
A
B E
D
C
A
B E
D
C
A. Top side, Pads for mating with Adhesive pads
B. Side View, Pads connected from top to bottom (vias)
C. Bottom Side, interconnection and Output pads

102. 104
Connective
Adhesive

103. VMJ

104. Vertical Multi Junction (VMJ)
Two terminal VMJ:
Eg1>Eg2>Eg3
Eg1
Eg2
Eg3
hν
Voltage boosting
junctions
U3
U3
U3
U2
U3
U2
U1
Equivalent
circuit
Um1=2Um2=4Um3
Voltage matching is possible!!!

105. Vertical Multi Junction (VMJ)
Extra junctions hν
contact
s
p+ doping
n+ doping
Eg1
Eg2
contact
s
direct band gap material
short diffusion length
Multi terminal VMJ: Each Layer Has separated contacts
Neither voltage nor current matching is required!!!

106. Vertical Multi Junction (VMJ)
Two terminal VMJ:
Eg1>Eg2>Eg3
Eg1
Eg2
Eg3
hν
Insulato
r
n+ doping p+ doping
Contact
s
Equivalent
circuit
U3 U2 U1
Um1=Um2=Um3 the output voltage will be restricted by U3

107. Vertical Multi Junction (VMJ)
Multi terminal VMJ: Each Layer Has separated contacts
Eg1 Cross Section:
contacts
n+
doping
p+ doping
insulation
n+ doping
Eg1 layer top view
n+ doping
& contact
p+ doping
& contact
hν
Current flow

108. Vertical Multi Junction (VMJ)
Multi terminal VMJ: Each Layer Has separated contacts
Neither voltage nor current matching is required!!!
Eg1 layer top view Eg2 layer top view Eg3 layer top view
n+ doping
&
contact
n+ doping
&
contact
p+ doping
&
contact
p+ doping
&
contact

109. Vertical Multi Junction (VMJ)
Eg1 layer top view Eg2 layer top view Eg2 layer top view
Multi terminal
VMJ top view:

110. Cost
Cell manufacturing: rough estimate
~ $1/cm2 (SOI technology)
1,000 suns, 20% system efficiency →
Additional cost savings
Reduce or eliminate homogenizer
Reduce or eliminate protection diodes (parallel connection)
Higher voltage to inverter- increased efficiency
$0.05/We

111. Cost Estimates S. Kurtz, Tech. report NREL, july 2008
Assuming the following parameters :
Tandem : 70 mW/cm2 @ Efficiency 40 % gives 0.028 W/cm2
Our design : 70 mW/cm2 @ Efficiency 30 % gives 0.021 W/cm2
Using :
This gives : Tandem – 0. 35 $/W + 0. 23 $/W = 58 cent/W
Our design - 0. 48 $/W + 0.05 $/W = 53 cent /W
($/W)
Cost
Cell
)
(W/cm
power
Cell
)
($/cm
cost
Area
($/W)
cost
Cell 2
2



112. Conclusion: VJ in CPV systems
• Advantages:
o High efficiency under high flux concentration.
o Less sensitive to flux non-uniformities.
o High output Voltage.
• Disadvantages:
o Low efficiency comparing to MJ cells.
o Difficult implementation of by-pass diodes.

113. Summary
VMJ Cells with high voltage
Low series resistance- higher efficiency under concentration
Reduced Mismatch loss under non-uniform illumination
Savings in other system components
Project status
Investigating Si
Performing simulations, cell design optimization
Defining manufacturing processes
Seeking opportunities to manufacture and test VMJ cells

114. Future Directions
• VJ operation under High concentration.
• Temperature effects on VJs
• Effects of non-normal illumination
• Setup a PV characterization system.
• Test prototypes.
Si
Build, test, optimize VMJ cells
Build, test, optimize arrays
III-V materials
Ideas for tandem VMJ cells

115. Acknowledgments
• Prof. Yossi Rosenwaks (supervisor)
• Prof. Abraham Kribus (supervisor)
• Dr. Rona Sarfaty
• Gideon Segev
• Michel Jurban
• Liav Grinberg
• Vlad Timofeev
Funding
Israel Ministry of National Infrastructures
IP
Provisional patent application, TAU

116. Thank
you!