the opto-electronic physics that just broke the efficiency ... - … · solar energy mini-series...
TRANSCRIPT
Owen D. Miller & Eli Yablonovitch
UC Berkeley
Electrical Engineering & Computer Sciences Dept.
http://arxiv.org/abs/1106.1603
Solar Energy Mini-Series
Jen-Hsun Huang Engineering Center
Stanford, California
Sept. 26, 2011
The Opto-Electronic Physics That Just Broke
the Efficiency Record in Solar Cells
2010
Cost per peak Watt for Solar Panels
Date
$10
$1
$0.30
X
X
1981 2011 2020
40GW installed cumulative
1TW installed
cumulative
subsidies will no
longer be needed
Year
From Allen Barnett
Univ. of Delaware
Value of High Photovoltaic Efficiency
Efficiency and Cost of Electricity (COE) inversely related:
OperatingFinanceEfficiencyInsolation
)Cost(CostCOE BOSmodule
Factors that impact COE:
Location: decides the available
input energy
Efficiency: decides the portion
that can be converted to
electricity
5% 8% 12%
18%
27%
40%
0
5
10
15
20
25
30
35
40
0 100 200 300 400 500 600 700 800 900 1000
Cost of Photovoltaics ($/m2)
Co
st
of
Ele
tric
ity
(¢
/kW
h)
9
$
CdTe 11% Efficiency
$
c-Si 23% Efficiency
1. Why the pn junction is merely optional in solar
cells?
2. What determines the voltage of a solar cell?
3. What is the statistical mechanical approach to
optics that is needed in solar cells?
4. Are photonic crystals of any help toward solar
cells?
5. What are the top competing technologies?
6. What is the status of the industry today?
What is a Solar Cell?
A Solar Cell does not require a p-n junction!
E
EFn
EFp
V
h
e-
e-
e- e
-
e-
e-
h+
h+
h+
h+ h
+ h+ h
+ Wider Bandgap
top, bottom
and sides
double hetero-structure Selective
electron
contact
Selective
hole
contact
k
+ -
-
+
e-
e- e-
h+
h+
h+
e-
Diffusion potential to contacts is typically <1mVolt.
What is a Selective Contact?
It passes one type of carrier but not the other.
Most Ohmic electrical contacts barely work on even one type of carrier.
Therefore they are almost all selective.
The ideal type of selective contact is a hetero-contact:
e-
EC
EV
EFn
E
EF
semiconductor metal
EFp
EC
EV
EFn
E
EF
semiconductor metal
EFp
electron contact: hole contact:
Solar Cell as a Bucket feeding a Water Wheel
rainfall incident sunlight
sidewall leakage
surface recombination
bottom leakage
bulk recombination
pressure head voltage
water flow current
water wheel output solar cell power
sidewall leakage
surface recombination
power flow pressure
There are three things to know about Solar Cells:
1.Short Circuit Current
2.Open Circuit Voltage
3.Fill Factor
Open valve Low-pressure, high-flow
Closed valve High-pressure, low-flow
Partially restricted valve Optimal pressureflow
Three possible valve conditions
(Water wheel barely turns) (Max water wheel output) (Water wheel doesn’t turn)
Fill-Factor =
ratio of
these
two
rectangles
Fill-Factor is more a thermodynamic
property, rather than a circuit property.
Ideal Fill-Factor ~ 89%
What is the current to expect?
Absorb all the incoming light, in as thin a layer as
possible.
A direct bandgap uses less material than an indirect
bandgap
h
h
Solar Cell:
light trapping:
path length increased by 4n2=50
where n = refractive index
semi-
conductor
semiconductor
single pass only
h
h
Solar Cell:
light trapping:
path length increased by 4n2=50
where n = refractive index
semi-
conductor
semiconductor
single pass only
h
h
h
Light Emitting Diode:
semi-
conductor
semiconductor
only 1/4n2 = 1/50 = 2%
of the light escapes.
all the light eventually escapes.
Non-Ergodic Shapes:
h sphere
h
rectangle
h parallel ogram
Most of the internal light
can never escape
Ergodic Shapes:
h
h
ellipse
trapezoid
All the internal light
eventually escapes
3
2
c
8π
Density of
Optical Modes
per Unit Volume
in air
3
22
c
n8π
Density of
Optical Modes
per Unit Volume
in semiconductor
What is the ratio? …… n2
There is a factor 2 from double pass
and
There is a factor 2 from cos averaging
The net benefit is 4n2
Light Trapping Requirement:
1. Either the high index material has to be textured,
or the photon recycling should be very efficient.
The Benefit:
a. 4n2 ~50 times thinner layer can be sufficient to absorb the
sunlight saving semiconductor cost.
b. In a thinner cell, there is less series resistance, and a
higher Fill Factor
c. The operating point voltage, (not just Voc), increases by
kT ln{4n2} ~ 0.1Volts.
Total Internal Reflection is completely compatible
with an Anti-Reflection Coating.
semiconductor layer
A randomly rough surface does approach
the statistical mechanical limit,
regardless what angle the light is coming in on.
h h
h
h
h
Can a photonic crystal patterned surface do better
than a randomly rough surface?
What is the operating voltage?
e-
e-
e- e
-
e-
e-
h+
h+
h+
h+ h
+ h+ h
+
+ -
To extract current, voltage at contacts must be slightly lower than Voc
But, operating voltage linked directly to Voc
kT
qV
q
kTVV OC
OCOP ln
We only need to understand the open-circuit voltage
What is the ideal voltage Voc to expect, i.e. the Quasi-Fermi
Level separation, chemical potential, or Free Energy?
In molecules and quantum dots:
qVoc=Free energy = kT ln
In semiconductors with mobile electrons & holes:
Free energy = EFc-EFv=2kT ln
darktheinpopulationstateexcited
lighttheinpopulationstateexcited
darktheindensityelectron
lighttheindensityelectron
darktheinpopulationstateexcited
lighttheinpopulationstateexcited
kT
EnergyFreeexp
Boltzmann Factor
What is the voltage to expect, i.e. the Quasi-Fermi
Level separation, chemical potential, or Free Energy?
Shockley-Queisser Limit (1961):
qVoc = kT ln
But in quasi-equilibrium:
qVoc = kT ln
darktheinemissionbandtoband
sunlightincoming
darktheinemissionbandtoband
emissionnt Luminesceexternal
qV =
Eg – kT{ln(/s) + ln(4n2) + ln(qVop/kT) –ln()-ln }
operating
point
Photon Free Energy = h - TS
Photon Free Energy = h - kT lnW
Yes photons have entropy, S
where s is the solid angle subtended by the sun
Entropy due
to loss of
directivity
information
Entropy
due to
incomplete
light
trapping
Free energy
loss due to
power-point
optimization
-0.28eV -0.1eV -0.1eV -0.0-0.3eV
Free energy
loss due to
poor
Quantum
Efficiency
nicest treatment:
R.T.Ross "Some Thermodynamics of PhotoChemical Systems", J. Chem. Phys. 46, 44590 (1967)
s
2
s
g
T
T
kT
E
+0.02eV
optical
mode
density
correction
qV =1.1eV – 0.8eV = 0.3eV operating
point
Small bandgaps are particularly vulnerable to entropy:
After you subtract off all the entropy terms, you don't
have much Free Energy left.
A lousy 0.3eV from all those big photons
In general we cannot afford to compromise with regard
to quantum efficiency.
no
rmal
so
lar
cel
ls
new
p
hys
ics
efficiency %
high performing cells
~25%
33.5% Shockley-Queisser limit (single-junction)
0%
Solar
radiation
Outgoing
luminescence
In an ideal solar cell, at open-circuit,
the outgoing luminescence will equal the incoming solar radiation
What did Shockley-Queisser say?
Any absorber also has to be an emitter!
At the open-circuit condition the electrons & holes
have no place to go.
They build up, and ideally they are lost only to luminescence.
h
h
semi-
conductor
only 1/4n2 = 1/50 = 2%
of the light escapes.
What if the material is not ideal, and the electrons and
holes are lost to heat before they can luminesce?
qVoc = qVoc-ideal – kT|ln{ext}|
The external
fluorescence yield ext is what matters!
Only external
Luminescence can balance
the incoming radiation.
no
rmal
so
lar
cel
ls
new
p
hys
ics
efficiency %
high performing cells
~25%
33.5% Shockley-Queisser limit (single-junction)
0%
(a)
e-
h+
h h h
(b) h h h hg hg
hg
e-
h+
h
h
semi-
conductor
only 1/4n2 = 1/50 = 2%
of the light escapes.
You may need an internal efficiency of int=99%
just to get an external efficiency of ext=50%
h
h
semi-
conductor
only 1/4n2 = 1/50 = 2%
of the light escapes.
Another way to look at this,
1. the recycled photons are not lost,
2. the carrier lifetime increases,
3. increasing carrier density
4. Increasing Voc
h
h
semi-
conductor
only 1/4n2 = 1/50 = 2%
of the light escapes.
But this is really hard to do:
You may need an internal efficiency of int=99%
just to get an external efficiency of ext=50%
0 0.2 0.4 0.6 0.8 1 24
27
33
Cel
l Eff
icie
ncy
(%
)
0 0.2 0.4 0.6 0.8 1
1.15
Vo
c (V
) 1.05
0.95
0.85
31.2%
33.2%
30.6%
1.061 1.081
1.145
Internal Fluorescence Yield
Internal Fluorescence Yield
30
32.5mA/cm2
0 0.2 0.4 0.6 0.8 1
Internal Fluorescence Yield
Jsc
( m
A/c
m2)
Efficiency vs. Internal Fluorescence Yield, GaAs 3m
90% Internal
Fluorescence Yield Is Not
Enough!
0 0.2 0.4 0.6 0.8 1 30.5
Reflectivity
Cel
l Eff
icie
ncy
(%
)
0 0.2 0.4 0.6 0.8 1
1.16
Reflectivity
Vo
c (V
olt
s)
1.14
1.12
1.10
1.08
1.06
Reflectivity 0 0.2 0.4 0.6 0.8 1
Jsc
(mA
/cm
2)
32
32.2
32.2%
33.2%
31.9%
1.104 1.115
1.145
32.43 32.46
32.50
31.5
32.5
33.5
32.4
32.6
Efficiency vs. Rear Reflectivity, GaAs 3m
90% Rear
Reflectivity Is Not
Enough!
Solar Cell Efficiency Limits for int < 1
Bandgap (eV) 0 0.5 1 1.5 2
5
10
15
20
25
30
35 So
lar
Cel
l Eff
icie
ncy
33.4% 31.0%
4%
int=100%
int=80% int=90%
Bandgap (eV)
0 0.5 1 1.5 2
Op
en-c
ircu
it V
olt
age
(V)
0.4
0.8
1.2
1.6 70 mV 20 mV
Textured good mirror Untextured good mirror Untextured bad mirror
Thickness 500nm 1 m 10m 500nm 1m 10m 500nm 1m 10m
Voc (volts) 1.14 1.13 1.12 1.16 1.15 1.14 1.08 1.08 1.07
Jsc (mA) 32.3 32.7 33.5 29.5 31.6 32.8 25.2 29.5 32.6
Fill Factor 0.89 0.89 0.89 0.89 0.89 0.89 0.89 0.89 0.89
efficiency % 32.8 33.1 33.4 30.6 32.6 33.3 24.3 28.3 30.9
Textured front
surface,
perfectly
reflecting mirror
Planar front
surface,
perfectly
reflecting mirror
Planar front
surface,
absorbing mirror
24
26
28
30
32
34
0.8 1 1.2 1.4 1.6
GaAs
33.5%
26.4%
Theory
Previous Record (2010) Voc=1.03V
GaAs, theory & expt.
25.1% record (1990- 2007)
~30%
Alta Devices 28.3% Voc=1.11V (2011)
32.1mA
Incoming light
External emission 0.04mA
Auger 710-6mA
Internal luminescence 5.40mA
Reabsorption 5.36mA Extracted
current 31.3mA
Planar, good mirror efficiency = 33.2%
efficiency = 30.6%
Planar, bad mirror
Mirror Loss 0.79mA
32.5mA
Incoming light
External emission 0.79mA
Auger 410-4mA
Extracted current 31.7mA
Mirror Loss 0mA
In both cases, you lose 0.76 mA. In the first case, that is luminescence out the top. In the second case, it is lost in the mirror. But the difference in external luminescence, and hence voltage, is significant.
Normalized for 1 cm2 area.
Internal luminescence 78.1mA
Reabsorption 77.3mA
For 3m thickness:
32.1mA
Incoming light
External emission 0.04mA
Auger 710-6mA
Internal luminescence 5.40mA
Reabsorption 5.36mA Extracted
current 31.3mA
Planar, good mirror efficiency = 33.2%
efficiency = 30.6%
Planar, bad mirror
Mirror Loss 0.79mA
32.5mA
Incoming light
External emission 0.79mA
Auger 410-4mA
Extracted current 31.7mA
Mirror Loss 0mA
In both cases, you lose 0.76 mA. In the first case, that is luminescence out the top. In the second case, it is lost in the mirror. But the difference in external luminescence, and hence voltage, is significant.
Normalized for 1 cm2 area.
Internal luminescence 78.1mA
Reabsorption 77.3mA
For 3m thickness:
(a)
e-
h+
h h h
(b) h h h hg hg
hg
e-
h+
Internal Fluorescence Yield int >> 90%
Rear reflectivity >> 90%
Both
needed for
good ext
Counter-Intuitively, to approach the
Shockley-Queisser Limit, you need to have
good external fluorescence yield ext !!
What is the voltage to expect, i.e. the Quasi-Fermi Level
separation, chemical potential, or Free Energy?
Shockley-Queisser Limit (1961):
qVoc = kT ln
But in quasi-equilibrium:
qVoc = kT ln
The external Luminescence intensity is a “Volt-meter”
into the solar cell.
darktheinemissionbandtoband
sunlightincoming
darktheinemissionbandtoband
emissionnt Luminesceexternal
For solar cells at 25%,
good electron-hole transport is already a given.
Further improvements of efficiency above 25%
are all about the photon management!
A good solar cell has to be a good LED!
Counter-intuitively, the solar cell performs best
when there is
maximum external fluorescence yield ext.
http://arxiv.org/abs/1106.1603
p-Al0.2Ga0.8As
n-GaAs Eg=1.4eV
n-Al0.5In0.5P
n+-Al0.5In0.5P Eg~2.35eV
p+-Al0.5In0.5P Eg~2.35eV
p-Ga0.5In0.5P Eg~1.9eV
n-Al0.5In0.5P Eg~2.35eV
n-Ga0.5In0.5P Eg~1.9eV
Tunnel
Contact
GaAs
VOC=1.1V
Solar Cell p-GaAs Eg=1.4eV
Ga0.5In0.5P
VOC=1.5V
Solar Cell
Tunnel
Contact
GaAs
VOC=1.1V
Solar Cell
Ga0.5In0.5P
VOC=1.5V
Solar Cell
Dual Junction Series-Connected Tandem Solar Cell
h h
All Lattice-Matched ~34% efficiency should be possible.
p-Al0.2Ga0.8As
n-GaAs Eg=1.4eV
n-Al0.5In0.5P
n+-Al0.5In0.5P Eg~2.35eV
p+-Al0.5In0.5P Eg~2.35eV
n-Al0.5In0.5P Eg~2.35eV
Tunnel
Contact
p-GaAs Eg=1.4eV
Wide
Bandgap
VOC>1.5V
Tunnel
Contact
Wide
Bandgap
VOC>1.5V
There is much latitude for creativity in the higher
bandgap—nano-particles; nano-rods, etc.
h h
p-Al0.5In0.5P Eg~2.35eV
GaAs
VOC=1.1V
Solar Cell
GaAs
VOC=1.1V
Solar Cell
>35% efficiency
should be possible.
GaAs solar cells are the preferred technology,
where cost is no objection: Space
The Epitaxial Liftoff Process:
5. What are the top competing technologies?
a. c-Si ~ 15%-23% in production
90% market share
b. poly-CuInGaSe2 ~ 13% promised
~100 Start-Ups
c. poly-CdTe ~ 11% in production
First Solar Corp.
d. flat-plate GaAs ~ 28.3% in R&D
Alta Devices Inc.
e. Concentrators ~ 43.5% in R&D
triple-junction III-V Solar Junction Inc.