Nanoscale Energy Conversion in the
Quantum Well Solar Cell
Keith Barnham, Ian Ballard, Amanda Chatten, Dan Farrell,
Markus Fuhrer, Andreas Ioannides, David Johnson,
Marianne Lynch, Massimo Mazzer, Tom Tibbits
Experimental Solid State Physics, Imperial College London, London
SW7 2BW, UK
k.barnham@ic.ac.uk
http://www.sc.ic.ac.uk/~q_pv
Rob Airey, Geoff Hill, John Roberts, Cath Calder,
EPSRC National Centre for III-V Technology, Sheffield S1 3JD, UK
Solarstructure , Permasteelisa, FULLSPECTRUM EU Framework VI,
Outline
 First practical nanoscale photovoltaic cell
 Enhanced spectral range of the strain-balanced
quantum well solar cell (SB-QWSC)
 Efficiency enhancement by photon recycling
 Evidence for hot electron effects in the QW
Cell efficiency cell versus l or Eg
 GaAs cells - highest effic. single
junction cells, Eg too high
 lower Eg => higher efficiency
 Can grow InyGa1-yAs bulk cells
on virtual substrates but never
dislocation free
 Maximum at 1.1 mm ~ 1.1 eV
 Multi-junction cells need 4th
band-gap ~ 1.1 mm ~ 1.1 eV
Enhancing GaAs Cell Efficiency
 From 30x – 1000x AM1.5
optimum single junction
efficiency band-gap ~ 1.1 eV
 Multi-junction approaches
going for GaInNAs cell
 No ternary alloy with lower Eg
than GaAs lattice matched to
GaAs/Ge
GaAs1-yPy (y ~ 0.1) + InxGa1-xAs, (x~ 0.1 – 0.2)
strain-balanced to GaAs/Ge => novel PV material
GaAsP/InGaAs Strain-Balanced QWSC
Balance stress between layers to
match lattice parameter of the
substrate
Advantages:
 Can vary absorption bandedge and absorb wider
spectral range without
strain-relaxation
 no dislocations > 65 wells
 single junction with wide
spectral range
 ability to vary Eg gives
higher tandem effic.
SB-QWSC – Ideal Dark-Currents
at High Concentration
 Dark current of 50 well QWSC
 Low current fits one parameter
Shockley-Read-Hall model
 High (concentrator) current slope
changes
ideal Shockley current
+ radiative recombination in QW
 Minimum recombination radiative
at concentrator current levels
Investigation of Photon Cavity Effects
 50 well SB- QWSC
 In0.1Ga0.9As wells
 GaAs0.91P0.09 barriers
 Control and distributed
Bragg reflector (DBR)
devices grown
side-by-side
 Processed as concentrator,
fully metalised, and photodiode
devices
 11 finger concentrator mask,
3.6% shading
Ta2O5 / SiNX
Ta2O5 / SiNX
BSF
BSF
DBR
100
100
100
100
80
80
80
80
60
60
60
60
40
40
40
40
20
20
00
400
400
20
20
500
500
600
600
700
700
800
800
900
900
 Increase photon
absorption
 Increase photocurrent
 No series resistance
 In-situ growth
Wavelength
Wavelength (nm)
(nm)
JSC (mA/cm2)
Device
AM1.5d 1000W/m2 AOD 913W/m2
NonDBR
28.0
DBR
28.6
100
00
1000
1000
26.3
26.9
Reflectivity (%)
DBR IQE
Non-DBR IQE
DBR reflectivity
Reflectivity (%)
Internal quantum efficiency (%)
Distributed Bragg Reflectors
80
60
40
20
[3]
0 D.C. Johnson et al. Solar Energy
0
20
40
60
80
Materials and Solar Cells, 2005
Incident angle (°)
Concentrator Measurements
 27% efficiency at 328x
low-AOD spectrum
 Single junction record is
(27.6 +/-1)% at 255x
26
Non-DBR
DBR
Efficiency (%)
25
24
D.Johnson et al. WCPEC4, Hawaii May 06
23
AM1.5d 1000W/m
2
22
10
100
Concentration (suns)
 Efficiency increase
higher than expect from
double pass in QWs
 Enhanced Voc
[3] Vernon S.M., et al. “High-efficiency concentrator cells
from GaAs on Si”, 22nd IEEE PVSC 1991 pp53–35
Why the Efficiency Enhancement?
MQW
DBR
Aim of DBR was to absorb
photons on second pass
 Some photons from radiative
recombination at high bias
trapped in the device
 Photons reabsorbed in the
QWs reduce dark current
 Generalised Plank model
for EL shows reduction
consistent with dark
current suppression
 Photon recycling could
take cell to 30% efficiency
MQW
DBR
Single QW Electroluminescence low bias
1
Bulk
Luminescence (a.u.)
Well
0.1
0.92V < Vapp < 0.98V
820
840
860
880
900
920
Wavelength (nm)
940
960
980
Single QW EL at high bias
1
Well
Luminescence (a.u.)
Bulk
Vapp= 1.10V
Vapp= 1.00V
0.1
0.92V < Vapp < 0.98V
820
840
860
880
900
920
Wavelength (nm)
940
960
980
10 QW Electroluminescence low bias
1
Luminescence (a.u.)
Well
0.1
Bulk
0.84V < Vapp < 1.02
840 850 860 870 880 890 900 910 920 930 940 950 960 970 980 990
wavelength (nm)
10 QW EL at high bias
1
Luminescence (a.u.)
Well
Bulk
Vapp = 1.16V
Vapp= 1.04V
0.1
0.84V < Vapp < 1.02
840 850 860 870 880 890 900 910 920 930 940 950 960 970 980 990
wavelength (nm)

Model EL (radiative recombination)
 Detailed Balance leads to generalised Planck:1
2n LW a (E)E
L(E,F)dE 
dE
3 2
(E DE F ) kB T
hc e
1
2
where
2
a(E) = absorption coefficient
T = temperature of recombining carriers
DEF = quasi-Fermi level separation
 a(E) (use measured QE) and T determine shape
 DEF requires absolute calibration
J.Nelson et al., J.Appl.Phys., 82, 6240, (1997)
M.Fuhrer et at Proc. EU PVSEC Dresden,Sept 06
EL - model and experiment
data
model
1
T=300.0K
T=320.0K
T=340.0K
T=360.0K
T=380.0K
luminescence
Luminescence (a.u.)
(a.u.)
1
increasing V
Increasing T
0.1
920
0.1
930
940
950
960
wavelength (nm)
970
980
990
920
930
940
950
960
wavelength (nm)
970
980
EL - Bulk Peak
1.1
1
0.9
0.8
Fits T = 299 K
Luminescence (a.u.)
0.7
0.6
0.5
0.4
0.3
0.2
840
850
860
870
Wavelength (nm)
880
890
Conclusions
 SB-QWSC concentrator cells (near) highest efficiency and
widest spectral range of single junction cells
 Radiative recombination dominates at high current levels and
photon recycling observed with DBR
 EL reduction with DBR consistent with dark-current
 Evidence for hot carrier effects at high current levels in EL
shape consistent with generalised Planck
 These nanoscale properties occur at the high current levels to
be expected in terrestrial concentrator systems
Advantages of the SB-QWSC
 Approximately double the efficiency of current cells
 Widest spectral range in a single junction cell so
keeps high efficiency as sunlight spectrum varies
 Nano-scale effectss – photon cavity, hot electrons
 Small size ~ mm – optoelectronic fabrication.
 Need high concentration to bring price down
What application?
Building integrated concentrator photovoltaics (BICPV)
Novel Application - Building Integrated Concentrators
SB-QWSC - highest efficiency single junction cell, ~ 1mm size
UK – over 60% electricity used in buildings
over 7 x as much solar energy falls on those buildings
SMART WINDOWS
 No transmission of direct sunlight
 Reduce glare and a/c requirement
 Max diffuse sunlight - for illumination
 No need for lights when blinds working
 (2 – 3) x power from Silicon BIPV
 Electricity at time of peak demand
 Cell cooling in frame - hot water
Barnham, Mazzer, Clive, Nature Materials, 5, 161 (2006).
Calculated output : San Francisco
9.3%
180
5.0% 1.2%
Electricity
2.1%
170
Space Heating
kWh/m 2
160
Water Heating
17.9%
150
Cooking
23.1%
Average electricity generated by
1 m2 of façade over 1 year
140
130
Lighting
Cooling
Ventilation
120
110
Refrigeration
8.6%
Office Equipment
100
1.0
10.0
100.0
Larger side/Smaller side
Other
7.3%
25.6%
Savings
120%
100%
80%
Consumption = 145 kWh/m2
60%
Fraction of electricity consumption
provided by photovoltaic cells
40%
20%
0%
0
30
60
90
120 150 180 210 240 270 300 330 360
Day
L
6L
3L
Luminescent Concentrators for
Diffuse Component of Sunlight
Dye-doped luminescent concentrators (1977):
 Advantages
no tracking required
accept diffuse sunlight
stacks absorb different l
Eg ~ Eg, gives max. effic.
thermalisation in sheet
 Disadvantages
dyes degrade in sunlight
loss from overlap of
absorption/luminescence
narrow absorption band
A Goetzberger and W Greubel, Appl. Phys. 14, 1977, p123.
Quantum Dot Concentrator
QDs replace dyes in luminescent concentrators:
 QDs degrade less in sunlight
 core/shell dots high QE
 absorption edge tuned by dot size
 absorption continuous to short l
 red-shift tuned by spread in dot size
 spread fixed by growth conditions
(K.Barnham et al. App. Phys.Lett.,75,4195,(2000))
Thermodynamic Model for QDC
 The brightness, B(n), of a radiation field that is in equilibrium
with electronic degrees of freedom of the absorbing species:
n = refractive index
8n 2n 2
1
Bn ) 
= 1/kT
c 2 ehn  m )  1
m = chemical potential
 Applying the principle of detailed balance within the slab:
Ns e n )
F m )   dnNs e n )I C n )   dn
Bn )  0
Qe
 IC = concentrated radiation field, Qe = quantum efficiency,
se = absorption cross section

Extend to 3-D fluxes + boundary conditions
I1(n)
x
Wc
y
A.J.Chatten et al, 3rd WCPEC, Osaka, 2003
E Yablonovitch, J. Opt. Soc. Am. 70, 1362, 1980.
z=0
Wc
z
W2
z=D
Characterisation of ZnS/CdSe QDs in
Acrylic with Thermodynamic Model
SD387 Red
SD396 yellow
 Thermodynamic model fits PL shape and red-shift
of Nanoco QDs assuming only absorption cross section
Fitting current measured at cell on edge gives
Qe(SD387) = 0.56 (c.f. Nanoco 0.4 – 0.6)
Thermodynamic Model confirms
unexpected luminescent stack result
Layer
Experimental Jsc
(mA/m2)
Predicted
Jsc (mA/m2)
Top
10.2 ± 2.0
9.1 ± 2.1
Bottom
35.1 ± 2.0
37.9 ± 1.3
Incident light
Total output = 45.3 (mA/m2)
Incident light
Layer
Experimental Jsc
(mA/m2)
Predicted
Jsc (mA/m2)
top
47.5 ± 2.0
46.9 ± 2.1
Bottom
4.8 ± 2.0
3.8 ± 1.3
Total output = 52.3 (mA/m2)
EL Modeling Confirms Recycling
 50 QW dark current show 33% reduction of J01
 Model EL by detailed balance ~ 30% reduction
 Supports efficiency increase results from
photon recycling
0.1
10
DBR
Non-DBR
Non-DBR Ideality n=1
DBR Ideality n=1
Current (A)
Normalised emission (a.u.)
Calculated
Measured
0.01
Ideality n = 1 reduction
1
1.32
1.34
1.36
1.38
1.40
Energy (eV)
1.42
1.44
1.04 1.06 1.08 1.10 1.12 1.14 1.16 1.18
Bias (V)
Compare SB-QWSC with Tandem in Smart Windows
A tandem cell 13%
more efficient than a
SB-QWSC harvests
only 3% more
electrical energy
P.Tandem
P.Single
1.2
Power/(kWh/m 2)
London –
Vertical South - East
Facing Wall
1.0
0.8
0.6
0.4
0.2
0.0
0
28 56 84 112 140 168 196 224 252 280 308 336 364
Day
Series current constraint means tandem optimised
for one spectral condition (and one temperature)
Single Molecule Precursor ZnS/CdSe Core-Shell QDs
Currently part of
“FULLSPECTRUM”
Framework VI Integrated
Project
300
absorption and luminescence of Nanoco OMN29 QDs
1.2
Experimental absorption with a linear background subtracted
Gaussian used to fit absorption threshold
250
Absorption fit used in predicting the luminescence
1
Normalised predicted luminescence
Normalised experimental luminescence
200
0.8
150
Absorption and emission
data from Sarah
Gallagher
0.6
100
0.06% by mass QDs in
chloroform
0.4
Pathlength 1cm
0.2
50
0
1.90
2.10
2.30
2.50
2.70
2.90
3.10
3.30
0
3.50
E/eV
(T.Trindade et al. Chemistry of Materials, 9, 523, (1997))
(A.J.Chatten et al, Proc. 3rd WCPEC, Osaka, 2003)
Luminescence/a.u.
Absorption/a.u.
 Core shell ZnS/CdSe
dots by thermolysis at
270 °C of singlemolecule precursors
in PLMA using with
TOPO cap
 Luminescence fit
is two-flux
thermodynamic model.
BICPV – Smart Windows
 Transparent Fresnel Lenses
 (300 – 500)x concentration
 1.5 or 2-axis tracking
 Novel secondaries
Lenses
 ~ 1 mm solar cells
Direct
Sunlight
 Cell efficiency ~ 30%
 Adds ~ 20% to façade cost
Diffuse
Daylight
Solar
Cells
Diffuse
Daylight
Front
Glass
Heat
Electricity