Solar cell technologies are improving in several key ways:
1) Creating materials that better exploit the photovoltaic effect and having higher efficiencies.
2) Using multiple junctions with different bandgaps to capture more of the solar spectrum.
3) Decreasing costs through larger scale production and reductions in material thickness.
These improvements are driving down the costs of solar electricity and enabling new applications. Further advances could make solar a major electricity source.
Solar Cell Performance and Cost Improvements Over Time
1. A/Prof Jeffrey Funk
Division of Engineering and Technology Management
National University of Singapore
For information on other technologies, see http://www.slideshare.net/Funk98/presentations
2. What are the important dimensions of
performance for solar cells and their higher
level systems?
What are the rates of improvement?
What drives these rapid rates of
improvement?
Will these improvements continue?
What kinds of new systems will likely
emerge from the improvements in solar
cells?
What does this tell us about the future?
3. Session Technology
1 Objectives and overview of course
2 Two types of improvements: 1) Creating materials that
better exploit physical phenomena; 2) Geometrical scaling
4 Semiconductors, ICs, electronic systems
5 MEMS and Bio-electronic ICs
6 Nanotechnology and DNA sequencing
7 Superconductivity and solar cells
8 Lighting and Displays
9 Human-computer interfaces (also roll-to roll printing)
10 Telecommunications and Internet
11 3D printing and energy storage
This is Part of the Seventh Session of MT5009
4. Creating materials (and their associated
processes) that better exploit physical
phenomenon
Geometrical scaling
◦ Increases in scale
◦ Reductions in scale
Some technologies directly experience
improvements while others indirectly
experience them through improvements in
“components”
A summary of these ideas can be found in
1) forthcoming paper in California Management Review, What
Drives Exponential Improvements?
2) book from Stanford University Press, Technology Change and
the Rise of New Industries
5. Creating materials (and their associated processes)
that better exploit physical phenomenon
◦ Create materials that better exploit phenomenon of
photovoltaic
◦ Create processes that enable these materials to better exploit
phenomenon of photovoltaic
Geometrical scaling
◦ Increases in scale: larger substrates and production
equipment lead to lower cost in much the same way that
cost of LCDs and other displays have fallen (see other
sessions for more details)
Some technologies directly experience
improvements while others indirectly experience
them through improvements in “components”
◦ Better solar cells lead to better modules and new ways of
organizing electricity production
9. Coal-fired power plants $2.10 a watt
Large hydroelectric systems can be cheaper
◦ Three Gorges Dam was supposedly about $1 a watt
◦ But actual costs are widely believed to be much higher
Natural gas-fired peaking power plants -$6 a watt
Large wind turbines -$2 a watt – but low capacity
utilization (27%)
Solar panels currently selling for as low as US$0.70
a watt in large quantities
◦ But installation costs are $2-4$ per watt
◦ And capacity utilization is very low (18%)
Except for peaking plants (see next slide), we need
much lower module and installation costs
http://en.wikipedia.org/wiki/Price_per_watt; http://thebreakthrough.org/index.php/programs/energy-and-
climate/how-fast-are-the-costs-of-solar-really-coming-down/
11. How do solar cells work?
Improvements in cost of electricity
from solar cells
Improvements in Efficiency
Multi-junction cells
Improvements in cost per area
Increases in scale
Reductions in thicknesses
Conclusions
15. 1839: First recognized by the French
physicist Alexandre-Edmond Becquerel
1883: First solar cell constructed from
selenium by Charles Fritts
1946: modern junction semiconductor solar
cell was first patented in 1946 by Russell Ohl
1954: first silicon solar cell was constructed
by Calvin Fuller, Daryl Chapin, and Gerald
Pearson
Subsequently other materials have been
found some as recent as 10-15 years ago
16. Semiconductor materials were found to exhibit the
photovoltaic effect in the 1950s, 60s and 70s
◦ Silicon, including single crystalline, polycrystalline, and
amorphous silicon
◦ Cadmium Telluride (CdTe)
◦ Copper indium gallium selenide (CIGS)
◦ Gallium arsenide
But also non-semiconductor materials, which were
more recently found
◦ Photo-sensitive dyes (titanium oxide)
◦ Some organic materials
◦ Some of these materials can be used to make quantum dots
or have Perovskite crystals
17. Incoming solar radiation creates “electron-hole” pairs in material
These electrons and holes create electricity when they reach
opposite terminals of device
Only photons whose energy exceeds band-gap of material create
electron-hole pairs
◦ Other photons do not create electron-hole pairs
◦ Energy greater than this band gap is lost
There has been a search for materials that
◦ exhibit photovoltaic effect
◦ have appropriate band gap
◦ have little recombination of electrons and holes
◦ are inexpensive to acquire and process
Within a type of solar cell, there has been a search for the
appropriate combination of material and process specifications
Conduction
band
Band gap
Valence
band
18. Materials with higher band gaps increase the
amount of energy from each absorbed photon
◦ but reduce the percentage of incoming radiation that can
be transformed into electrons and holes.
Thus, there is a tradeoff between low and high
band gaps
Given the distribution of the solar spectrum
◦ the optimal band gap in terms of efficiency can be
calculated
◦ the maximum theoretical efficiencies can be calculated
(about 30%)
◦ Many materials have a maximum theoretical efficiency of
about 30%
19.
20. Examples of Efficiency
Losses
Reflection of photons by glass.
Absorption of photons by glass
(heat)
Recombination of electron hole
pairs before reaching terminals
(crystalline materials have less
recombination)
Photons pass through material
without generating electron-
hole pairs
But even if the best band-gap is used,
there will be losses
21. One way to overcome limitations of individual
materials is to use multiple junctions
◦ Each has band gap that is appropriate for different part of
solar spectrum
These solar cells can have
◦ much higher efficiencies than single junction ones
◦ but they also have higher costs as multiple layers must be
deposited, patterned and etched
One way to reduce costs is to
◦ focus sunlight onto multi-junction cells using concentrators
◦ but concentrating mirrors require mechanical and electronic
controls, gears and other potentially unreliable components
More on this later
22. How do solar cells work?
Improvements in cost of electricity
from solar cells
Improvements in Efficiency
Multi-junction cells
Improvements in cost per area
Increases in scale
Reductions in thicknesses
Conclusions
23. Depends on a lot of factors but to simplify….
Cost of electricity depends on
◦ amount of incoming solar radiation
◦ cost per “Peak Watt” of a module (based on a
predetermined amount of incoming solar
radiation)
◦ cost of installation including cost of capital, land,
labor, etc.: becomes more important as cost of
“modules” drop
Cost per “Peak Watt” depends on
◦ cost per area of solar cells
◦ efficiency of solar cells
25. world electricity demand
(18,000 TWh/y)
can be produced from
300 x 300 km²
=0.23% of all deserts
distributed over “10 000” sites
3000 km
Sources: Gerhard Knies, CSP 2008 Barcelona
and Vinod Khosla 25
deserts as solar farms
27. Source: Physica C: Superconductivity Volume 484, 15 January 2013, Pages 1–5. Proceedings of the 24th International
Symposium on Superconductivity (ISS2011). CIGRÉ SC D1 WG38 Workshop on High Temperature Superconductors (HTS)
for Utility Applications Beijing, China, 26 April 2013
Can these improvements make solar
economical for Europe?
Now
2 years
4 years
28. What About Cost Per Peak Watt of Modules?
Depends on cost per area and Efficiency
Cost per area
($/m2) 50 100 200
Cost
per
Area
200
100
50
0 2 4 6 8 10 12 14 16 18 20 22
Cost($)perPeakWatt
1.5
1
.5
0
Efficiency
29. Improvements in either Cost per area or Efficiency can lead to a
Lower Cost per Peak Watt
Cost per area ($ per square meter)
50 100 200
Cost
per
Area
200
100
50
0 2 4 6 8 10 12 14 16 18 20 22
Cost($)perPeakWatt
1.5
1
.5
0
Efficiency
Improvements in Improvements in
efficiency or cost per area
lead to lower costs per peak watt
30. Rapidly falling costs and prices
Sometimes faster than expected, other times
slower than expected
Subsidies distort prices
◦ Not just subsidies for installations (U.S., Germany,
Japan, Spain)
◦ But also alleged subsidies for producers (China)
Alleged subsidies have led to
◦ Large exports of solar cells from China
◦ Trade dispute
◦ Smaller value added for solar cell producers both in
absolute and percentage terms
34. Hard to separate long term and short term trends
Likely that Chinese subsidies for Chinese producers
have caused short term fall in prices
Thus prices may go up when subsidies are gone
◦ This may be why Suntech went bankrupt in March 2013
Rising use of natural gas in U.S. has also reduced
demand for solar panels and thus prices in the U.S.
◦ Firms must reduce prices or have unused capacity, so most
firms will sell below costs
Subsidies also make it hard to understand which
technology might be the cheapest in the future
Sources: http://sync.democraticunderground.com/112739038;
http://www.nytimes.com/2013/03/21/business/energy-environment/chinese-solar-
companys-operating-unit-declares-bankruptcy.html?_r=0
35. How do solar cells work?
Improvements in cost of electricity
from solar cells
Improvements in Efficiency
Multi-junction cells
Improvements in cost per area
Increases in scale
Reductions in thicknesses
Conclusions
36. Maximum Theoretical Efficiency
◦ Similar for a wide range of materials
◦ Black body limit
Best Laboratory Efficiencies
◦ Best efficiencies for cells produced in a
laboratory
Best production efficiencies
◦ Best efficiencies for cells (or modules)
produced in a factory
37. Technology Production
Facilities
Laboratories Theoretical
Limits
Crystalline Silicon 18% 25% 29%
Micro-crystalline silicon 14% 20% 29%
Cadmium-Indium Gallium Selenide
(CIGS)
11% 20% 29%
Cadmium Telluride (CdTe) 11% 17% (20.4%) 29%
Amorphous Silicon 8% 13% 20%
Organic Cells 2% 8% (11.1%) 31%
Dye-Sensitized Cells 12% 31%
Best Solar Cell Efficiencies and Theoretical limits
(for single materials in 2010)
Sources: U.S. DOE, 2010; Wang Qing and Palani Balaya (personal communication)
38. Crystalline materials have lower recombination of
holes and electrons
◦ Also high efficiencies for other crystalline materials
(e.g., GaAs)
More research on silicon
◦ Longer history of silicon research than other materials
◦ Silicon’s current dominance (cheaper equipment and
demand-based subsidies) reinforces this perspective
Do the other materials have more potential for
improvements? And if so, how much?
44. Many types of organic
materials, but they all
contain carbon
◦ Many substitutions are
tried
One substitution is
fullerenes
Placing them in the
right place is important
Synchrotron is used to
analyze the energy
levels and thus the
right places to place the
fullerenes
45. Worldwide OPV production forecast
Source: Solar&Energy, Recent Organic Solar Cell Technology and Market Forecast (2010 -
2015
46. 2000 2005 2010 2015
16
14
12
10
8
6
4
2
Rapid Improvements in Efficiency of Perovskite Solar Cells
Perovskite
Organic
Dye Sensitized
Amorphous Silicon
Perovskite-based solar cells, Hodes G, science 342, 317 (2013, Oct): 317-318
47. Perovskite cells are a hybrid of organic and
inorganic materials and they have a certain type
of crystal structure
◦ Thus may use materials classified as other types of
solar cells
◦ Key difference is in crystalline structure
Efficiencies similar to crystalline silicon are
possible due to its single crystalline structure
First two cells in 2009 and 2010 were liquid
junction cells that were not stable
Perovskite-based solar cells, Hodes G, science 342, 317 (2013, Oct): 317-318;
http://www.technologyreview.com/news/521491/a-new-solar-material-shows-its-potential/
48. Recent ones have high diffusion lengths and
long lifetimes for holes and electrons (i.e., low
recombination)
Researchers have shown that it is relatively easy
to modify the material so that it efficiently
converts different wavelengths of light into
electricity
May be possible to form a solar cell with
different layers, each designed for a specific
part of the solar spectrum (i.e., multi-junction
cell)
Perovskite-based solar cells, Hodes G, science 342, 317 (2013, Oct): 317-318;
http://www.technologyreview.com/news/521491/a-new-solar-material-shows-its-potential/
49. Low-temperature deposition methods
◦ typically solution-based spin coating as
compared to sputtering or vapor deposition
One-fifth the cost of current silicon-based
solar cells on an area basis, due to the simpler
manufacturing process
No rare materials or toxic (lead is worst
material)
If lifetime related problems are solved and if
lab efficiencies reach 20%, costs of $0.20 per
peak Watt are expected
Perovskite-based solar cells, Hodes G, science 342, 317 (2013, Oct): 317-318;
http://www.solarika.org/blog/-/blogs/new-hope-for-cheaper-solar-cells-using-perovskites
51. How do solar cells work?
Improvements in cost of electricity
from solar cells
Improvements in Efficiency
Multi-junction cells
Improvements in cost per area
Increases in scale
Reductions in thicknesses
Conclusions
52. One way to overcome limitations of individual
materials
◦ Each has band gap that is appropriate for different part of
solar spectrum
These solar cells can have
◦ much higher efficiencies than single junction ones
◦ but they also have higher costs as multiple layers must be
deposited, patterned and etched
One way to reduce costs is to
◦ focus sunlight onto multi-junction cells using concentrators
to reduce amount of photovoltaic material
◦ but concentrating mirrors require mechanical and electronic
controls, gears and other potentially unreliable components
and they can only be currently used in cloudless skies
55. Ideally as you increase the number of
band gaps the efficiency increases
56.
57. For the highest
efficiency solar
cells, $50,000 per
square meter
High costs come
from maintaining
crystalline structure
even with 20 layers
Grown as one large
crystal
58. Organic materials
can be roll printed
or sprayed on top
of each other
Organic tandem
cells have the
same efficiencies
as do single-
junction organic
cells
But this may
change
59. How do solar cells work?
Improvements in cost of electricity
from solar cells
Improvements in Efficiency
Multi-junction cells
Improvements in cost per area
Increases in scale
Reductions in thicknesses
Conclusions
60. Type of material
◦ Availability in earth’s crust, Processing requirements
Number of layers
◦ More layers means more processing steps
Temperature of processing
◦ Higher temperatures means higher costs (e.g.,
semiconducting materials, crystalline silicon)
◦ http://www.youtube.com/watch?v=F2KcZGwntgg (from 1:50)
◦ Organic materials can be roll printed
Thickness of materials
◦ More difficult to reduce thickness of epitaxial formed silicon
(crystalline silicon) than thin-film deposited materials (CIGS,
CdTe, amorphous silicon)
Scale of substrates and production equipment
◦ Same as with LCDs and semiconductor wafers
61. Fewer layers
Less materials
Lower temperature and simpler processes
◦ Organic materials, CIGS, and Perovskite can be roll printed
onto a substrate
Perhaps lower scale right now so greater potential
for increases in scale
◦ Many forms of thin film already use large scale production
equipment
◦ But large scale equipment has not been implemented for
some technologies, particularly roll printing
◦ Roll printing is applicable to some processes and many
processes for organic solar cells
63. Roll printing of organic solar cells
Notice the simplicity
Also notice the small size of the solar cells – still a
long way from reaching its optimum scale
Discussed more in two weeks
Organic solar cells, roll-to roll printing
16 May 2013
http://reneweconomy.com.au/2013/on-a-roll-csiro-
printing-australias-largest-solar-cells-58992
64. How do solar cells work?
Improvements in cost of electricity
from solar cells
Improvements in Efficiency
Multi-junction cells
Improvements in cost per area
Increases in scale
Reductions in thicknesses
Conclusions
66. Equipment costs per area of output fall as size
of equipment is increased, similar to chemical
plants
For chemical plants
◦ Cost is function of surface area (or radius squared)
◦ Output is function of volume (radius cubed)
◦ Thus, costs increase by 2/3 for each doubling of
equipment capacity
For LCD Substrates, IC Wafers, and Solar
Substrates
◦ Processing, transfer, and setup time (inverse of
output) fall as area of substrate increases since
entire area can be processed, transferred, and setup
together
67. Another Benefit from Large Panels is Smaller Edge E
Panel
Equipment
Effect Effects: the equipment must be much
wider than panel to achieve uniformity
Ratio of equipment to panel width falls as the
size of the panel is increased
68. Increases in LCD Substrate Size
Source: www.lcd-tv-reviews.com/pages/fabricating_tft_lcd.php
69.
70. Scale of photolithographic aligners (upper left),
sputtering equipment (top right), and mirrors for
aligners (lower left) for LCD equipment
Source: http://www.canon.com/technology/
canon_tech/explanation/fpd.html
74. Solar cells also benefit from increases in scale
of production equipment
Crystalline silicon solar cells are made in wafers,
just like semiconductor chips
◦ Their costs fall as wafers and production equipment
are made larger, but improvements are difficult
Thin-film solar is made on substrates, like LCDs
◦ Their costs fall as substrates and production
equipment are made larger
CIGS and organic solar cells can be roll printed
◦ Materials can be deposited and patterned using roll-to
roll printing
◦ Consider Self-Aligned Imprint Lithography (SAIL)
75. $0.0 $0.5 $1.0 $1.5 $2.0 $2.5 $3.0
Web preparation
Sputter Gate 1 Metal
Align and Expose
SiN, a-Si, N+ dep
Align and Expose
Si RIE & Resist Strip
Ultrasonic Clean
Align and Expose
Sputter Dep/ ITO
Align and Expose
Sputter Dep Interconnect
Align and Expose
Web cost
$0.0 $0.5 $1.0 $1.5 $2.0 $2.5
Condition web (de-hydro)
Gate metal deposition (Al)
PECVD oxide/nitride/Si/N+ deposition
SD metal deposition (Cr)
Imprint SAIL structure
Wet etch Cr
RIE etch n+&Si&SIN
RIE etch oxide
Plasma etch Al
Thin down 2P (clear gate-pad)
Pre-Cr-etch Cleaning
RIE etch n+&Si&SIN
Thin down 2P (clear gate-pad)
Wet etch Cr
RIE etch n+
Under-cut Al (1-3 um)
RIE etch oxide
Strip-off 2P
Web cost
costper
ft2
$0.00
$2.00
$4.00
$6.00
$8.00
$10.00
$12.00
$14.00
$16.00
$18.00
Photolithography SAIL
Cost of Patterning
Backplane materials costs for
R2R photolithography & SAIL
R2RSAILR2Rphotolith(AGI)
Multiple
photoresist applications
dominate photolithography
process materials costs
76. 2 3 4 5 6 7 8 9 10
10
-3
10
-2
10
-1
10
0
equipmentcost[M$]/throughput[cm2
/S]
generation
equipment cost scaling comparison: panel stepper vs R2R imprinter
100 mm R2R imprinter
330 mm R2R imprinter
R2R Imprinters are much cheaper than Panel Stepper
used in Photolithography
(and benefits from increases in scale)
Source: Roll-to-Roll Manufacturing of Flexible Displays, Hewlett Packard, Phicott
77. 1 2 3 4 5 6 7 8 9 10
10
-2
10
-1
10
0equipmentcost[M$]/throughput[cm2
/S]
generation
equipment cost scaling comparison: panel CVD vs R2R CVD
330 mm R2R PECVD
1 m R2R PECVD
Source: Roll-to-Roll Manufacturing of Flexible Displays, Hewlett Packard, Phicott
PECVD (plasma enhanced chemical vapor deposition)
is also cheaper when doing R2R Printing
(and benefits from increases in scale)
78. Installation costs are now more than
the module costs on a per Watt basis
Lower on a per Watt basis with
◦ large-scale than small-scale systems
◦ High-efficiency than low-efficiency
modules
Lower on a per-area basis with thin-
film or rolled materials than with thick
materials (crystalline silicon)
◦ Just unroll a roll of organic solar cells
79. How do solar cells work?
Improvements in cost of electricity
from solar cells
Improvements in Efficiency
Multi-junction cells
Improvements in cost per area
Increases in scale
Reductions in thicknesses
Conclusions
80. Can further decrease costs
It is easier to reduce thickness of thin-film
materials than crystalline silicon
◦ their thickness depends on deposition of materials
◦ thickness of crystalline silicon materials depends on
cutting silicon ingots into wafers
But it can also lead to lower efficiencies
◦ Less active material increases the chances that photons
will pass through the material before they create an
electron-hole pair
What if we can increase the amount of time that a
photon spends within the material for a given
thickness?
◦ Thus enabling reductions in thickness
81. Rationale: Light trapping increases the optical thickness
of a silicon cell by 10--‐50 times
Hence, theoretically it is possible to achieve
similar efficiencies with a thinner layer material
82. How do solar cells work?
Improvements in cost of electricity
from solar cells
Improvements in Efficiency
Multi-junction cells
Improvements in cost per area
Increases in scale
Reductions in thicknesses
Conclusions
83. Cost of electricity from solar cells is dropping rapidly
Silicon is most widely used material
◦ Will further cost reductions for silicon occur?
◦ Or have we reached the limits?
Large number of materials and processes suggests that
many improvements can still be achieved
◦ Rapid increases in efficiency are still occurring for some
materials: organic, quantum dot, perovskite, multi-junction
solar cells
◦ Scale up of substrates and equipment have not been done for
some materials
84. Other improvements are also occurring
◦ Reducing thickness of materials
Installation costs are becoming larger as a percentage of
total costs
◦ How can they be reduced?
◦ With roll printed solar cells, or with higher efficiency solar
cells?
98. Implications of learning curve (cost of producing
product falls as cumulative production increases)
Although learning curves don’t exclude non-factory
activities, linking cost reductions with production
◦ focuses policy, other analyses on production of final product
◦ implies that research done outside of factory is either
unimportant or being driven by production of final product
◦ for solar, has encouraged demand-based policies that subsidize
installation of solar cells ($130B just in Germany) as opposed to
more R&D, which has encouraged focus on existing
technologies
We should first understand direct drivers of cost
reduction, then develop good policy
100. Increases in best laboratory efficiencies of organic and
dye-sensitized solar cells
http://www.asiabiomass.jp/english/topics/1208_05.html
101. Worldwide OPV production forecast
Source: Solar&Energy, Recent Organic Solar Cell Technology and Market Forecast (2010 -
2015
Hinweis der Redaktion
How do we get all the red area?
Why are best production efficiencies lower than best laboratory efficiencies?
The size of LCD substrates were increased by about 4 times in change from 1st generation to 5th generation LCD substrate equipment. We are now at Generation 10
Production of generation 7.5 panels (with generation 7.5 equipment) was just starting in 2008. Firms are now implementing generation 10 panels and equipment.
Subsequent generations of equipment are very large. Notice the size of the humans in the pictures.
The capital cost per area output of LCDs falls as the size of the substrates (and production equipment) are increased. the area increased by 3.7 times while the capital costs only rose by 2.37 times as we moved from Gen 5 to Gen 7.5.
This data is for one type of LCD manufacturing equipment, called dry etch equipment. The productivity of the equipment (square meters per hour-$) rose about 8 times (2.7 to 23) as firms moved from Gen II to Gen VI.