download PDF - International Center for Materials Research

Semiconductor Materials and Tandem Cells
for Photoelectrochemical Hydrogen
Production: GaP 1-x N x and Cu(In,Ga)Se,S)
ICMR Symposium on Materials Issues in Hydrogen Production
and Storage
Santa Barbara, CA
August 24, 2006
John A. Turner
John_Turner@nrel.gov 303-275-4270

Sustainable Paths to Hydrogen
Solar Energy
Heat
Biomass
Mechanical Energy
Electricity
Conversion
Thermolysis
Electrolysis
Photolysis
Hydrogen

Direct Conversion Systems
Visible light has sufficient energy to split water
(H 2 O) into Hydrogen and Oxygen
Combination of a Light Harvesting System and a Water Splitting System
Semiconductor photoelectrolysis
Photobiological Systems
Homogeneous water splitting
Heterogeneous water splitting
Thermal cycles
(Sunlight and Water to Hydrogen
with No External Electron Flow)

Photoelectrochemical-Based Direct
Conversion Systems
• Combines a photovoltaic system
(light harvesting) and an
electrolyzer (water splitting) into a
single monolithic device.
– Electrolysis area approximates
that of the solar cell - the current
density is reduced.
– Efficiency can be 30% higher than
separated system.
• Balance of system costs reduced.
– Capital cost of electrolyzer
eliminated
• Semiconductor processing
reduced.

Current Density vs. Voltage for 2 Pt Electrodes of
Equal Area.
V,I
H 2
O 2
Water
Current Density, mA/cm 2
120
100
80
60
40
20
2M KOH
AM1.5 Solar Cell
current density
1.45V means ~85%
LHV electrolysis
efficiency
0
1.30 1.32 1.34 1.36 1.38 1.40 1.42 1.44 1.46
Voltage, V
O. Khaselev, A. Bansal and J. A. Turner, International Journal of
Hydrogen Energy, 26, p 127-132 (2001).

Chemical Reactions at a Semiconductor
Electrolyte Interphase
Electron energy
E c
Acceptor
Light (photons)
Donor
E v + e -
+
Solid (semiconductor)
Liquid (electrolyte)
P169-A302101

Band Edges of p- and n-Type
Semiconductors Immersed in Aqueous
Electrolytes to Form Liquid Junctions
E conduction
O 2
p-type
E valence
2H 2
O + 2e – = 2OH – + H
2
E conduction
H 2
n-type
H 2
O + 2h + = 2H + + 1/2 O 2
P158-A199107
E valence

Electron
Energy
Technical Challenges (the big three)
Material Characteristics for Photoelectrochemical
Hydrogen Production
H 2 O/H 2
‣Efficiency – the band gap
(E g ) must be at least 1.6-1.7
eV, but not over 2.2 eV, high
photon to electron
conversion.
E g
1.23 eV
1.6-1.7 eV
Counter
Electrode
‣Material Durability –
semiconductor must be
stable in aqueous solution
p-type
Semiconductor
H 2 O/O 2
‣Energetics – the band
edges must straddle H 2 O
redox potentials (Grand
Challenge)
i
All must be satisfied
simultaneously

Bandedge Energetic Considerations
T. Bak, J. Nowotny, M. Rekas, C.C. Sorrell, International Journal of Hydrogen Energy 27 (2002) 991– 1022

Gallium Indium Phosphide/Electrolyte System
Used to gain a fundamental understanding of
semiconductor/electrolyte junctions
(-)
E CB
Eg = 1.83 eV
OK
Energy
(+)
E VB
p-GaInP 2
E F
H 2
O/H 2
Band edges are
too negative
Band edges are
pH sensitive
H 2
O/O 2

Bandedge Energetic Considerations
GaInP
T. Bak, J. Nowotny, M. Rekas, C.C. Sorrell, International Journal of Hydrogen Energy 27 (2002) 991– 1022

Dual Photoelectrode Approach for
Bandedge Mismatch and Stability Issues
a
e
b
α -Fe
O
2 3
µA
p-GaInP
2
-2.0
-2.0
-1.0
e
CB
-1.0
hv
0.0
1.0
CB
e
E 0 (H /H O)
2 2
E 0 (H O/O )
2 2
h
VB
0.0
1.0
2.0
VB
2.0
h
Electrolyte
Lower cost Fe 2 O 3 electrode is a possibility, but while the system
splits water, the efficiency is very low.

Metal Oxides – A materials assessment
PEC devices must have the same internal photon-toelectron
conversion efficiency as PV devices.
• Transitions which impart color to the oxides typically
involve d and f levels and are generally forbidden
transitions = low absorption coefficients.
– The low absorption coefficients result in the incident light
penetrating deep into the material = 100’s um of material
• Carrier mobilities in oxides are very low
– Recombination of photogenerated carriers occurs before
they can reach the semiconductor/electrolyte interface.
• Even good single crystals of metal oxides have very
low energy conversion efficiencies due to inherently
short diffusion lengths.
However a success in this area, e.g. an oxide with good
to excellent conversion properties could revolutionize
the PV industry as well as make PEC near-term.

Approach: High Efficiency Materials and
Low-Cost Manufacturing
PEC devices must have the same internal photon-to-electron conversion
efficiency as commercial PV devices.
• III-V materials have the highest solar conversion efficiency of any
semiconductor material
– Large range of available bandgaps
– ….but
• Stability an issue – nitrides show promise for increased lifetime
• Band-edge mismatch with known materials – tandems an answer
• I-III-VI materials offer high photon conversion efficiency and
possible low-cost manufacturing
– Synthesis procedures for desired bandgap unknown
– ….but
• Stability in aqueous solution?
• Band-edge mismatch?
• Other thin-film materials with good characteristics
– SiC: low cost synthesis, stability
– SiN: emerging material

Approach: Materials Summary
The primary task is to synthesize the semiconducting material or the
semiconductor structure with the necessary properties. This involves
material research issues (material discovery), multi-layer design and
fabrication, and surface chemistry. Activities are divided into the task areas
below – focus areas in Green:
GaPN - NREL (high efficiency, stability)
CuInGa(Se,S) 2 - UNAM (Mexico), NREL (Low cost)
Silicon Nitride – NREL (protective coating and new material)
GaInP 2 - NREL (fundamental materials understanding)
Energetics
Band edge control
Catalysis
Surface studies

Gallium Indium Phosphide/Electrolyte System
Used to gain a fundamental understanding of
semiconductor/electrolyte junctions
(-)
E CB
Eg = 1.83 eV
OK
Energy
(+)
E VB
p-GaInP 2
E F
H 2
O/H 2
Band edges are
too negative
Band edges are
pH sensitive
H 2
O/O 2

Review: Comparison of p-GaInP 2 and PEC/PV device
0
Current Density, mA/cm 2
-40
-80
-120
3 M H 2
SO 4
, ~ 11.6 suns
1: p-GaInP 2
(Pt)/TJ/GaAs
2 : p-GaInP 2
(Pt)
2 1
-160
-1.5 -1.0 -0.5 0.0 0.5
Voltage versus Pt
The GaAs integrated PV cell compensates
for the energetic mismatch – but lifetime…

Photocurrent time profile for PEC/PV Water-Splitting device,
showing current decay due to corrosion.
O. Khaselev and J. A. Turner, Science, 280, pg 425 (1998).
Current Density, mA/cm 2
-40
-60
-80
-100
CD, mA/cm 2
-50
-100
0 5 10 15 20
Time, hours
We are looking for
materials with
inherently greater
stability.
-120
-140
Efficiency(%) = (120mA/cm 2 2
*1.23v/1190 mW/cm )*100
= 12.4%
0 500 1000 1500
Time, sec

Elemental Abundances in Earth’s Crust
Element
Concentration
(mg/kg)
Element
Concentration
(mg/kg)
Gallium
19
P
1050
Lead
14
Silicon
2.8 x 10 5
Boron
10
Indium
0.25
Lithium
20
Silver
0.075
Tungsten
1.25
Tin
2.3
Cobalt
25
Uranium
2.7
Gallium is found in bauxite (aluminum ore). There are 2
million kg of Ga in Arkansas bauxite deposits alone. Also
found in coal ash (up to 1.5%).

Hydrogen Evolution Current-Time Profile for p-GaN
Current density (A/cm 2 )
8.0x10 -5
7.0x10 -5
6.0x10 -5
5.0x10 -5
4.0x10 -5
3.0x10 -5
2.0x10 -5
9.0x10 -5 1 M KOH
pH 7 buffer
3 M H 2
SO 4
1.0x10 -5
0 1000 2000 3000 4000
Time (s)
S. Kocha, M. Peterson, D. J. Arent, J. M. Redwing, M. A. Tischler and J. A. Turner, “Electrochemical Investigation
of the Gallium Nitride-Aqueous Electrolyte Interface”, Journal of The Electrochemical Society, 1995,142, pg. L238-
L240.

Goal: Stable III-V nitride material
• GaN
– Capable of water-splitting and stable
– Band gap direct but too wide (~3.4 eV)
• GaP
– Band edges almost aligned
– Band gap indirect and too wide (~2.3
eV), but better
• GaP 1-x N x
– Addition of small amounts of N causes
GaP band gap to narrow (bowing) and
transition to become direct
– Nitrogen enhances stability
III-V Nitrides for PEC Water Splitting
Systems: GaP 1-x N x
Normalized Photocurrent 2 (AU)
GaP 1-x N x Epilayer Direct Band Gap
0.06% N
0.11% N
0.15% N
0.20% N
1.10% N
1.60% N
0
2 2.1 2.2 2.3 2.4
Photon Energy (eV)
Dr. Todd Deutsch

Goal: Stable III-V nitride material
• Control band gap energy
by varying nitrogen
composition,
• Direct band gap for all
nitrogen concentrations,
even at GaP .9994 N .0006
• The crossover from
indirect to direct occurs
significantly below the
theoretically predicted
0.26%
(F. Benkabou, J. P. Becker, M.
Certier, H. Aourag, Calculation of
Electronic and Optical Properties
of Zinc Blende GaP1-xNx.
Superlattices and Microstructures,
1998. 23(2): p. 453-465.)
Dr. Todd Deutsch
GaP 1-x N x Bandgap as a Function of
Nitrogen Concentration
Direct Band Gap (eV)
2.25
2.2
2.15
2.1
2.05
2
1.95
1.9
1.85
This Study
Yu et. al.
0 1 2 3 4
Percent Nitrogen
K.M. Yu, W. Shan, J. Wu, Nature of the fundamental band gap in
GaN x P 1-x alloys. Applied Physics Letters, 2000. 76(22): p. 3251-3253

Goal: Stable III-V nitride material
Corrosion
experiments on
GaP substrate
clearly show the
enhanced
corrosion
resistance from
nitrogen addition.
Dr. Todd Deutsch

Goal: Stable III-V nitride material
1.2
1.11
GaPN Stability Measured by Profilometry
and by ICP-MS of Solution
Average Etch Depth (µm)
1
0.8
0.6
0.4
0.2
0
0.56
0.36
0.07
GaInP2 GaP no Pt GaP GaPN
0.2%N
Epilayer Composition
0.25
GaPN
2.1%N
ppb Ga
3000
2500
2000
1500
1000
500
0
5.0
Dark
5mA/cm^2
130
AM 1.5
5mA/cm^2
540
AM 1.5
Open Circuit
2500
AM 1.5
5mA/cm^2
No Surfactant
2mm stylus scan
Conditions
5mm
Exposed/active
surface
Masked/native
surface
GaInP 2 ~ 1 µm
Nitrides ~ 0.1 µm
• Photocathode durability testing
– 5mA/cm 2 , AM 1.5, 24 hours, 3M H 2
SO 4
,
pulsed Pt treatment
– Ga content of solutions by ICP-MS
– Etching by profilometry
4mm
• Represents ~300 hrs of stability for 1 um layer
Dr. Todd Deutsch

Goal: Stable III-V nitride material
Band Edge Energetics
The band edges of these materials are too negative for
spontaneous water-splitting.
Potential (V vs. Ag/AgCl)
-1.5
-1
-0.5
0
0.5
1
GaP
0.06%N
0.11%N
0.10%N
0.15%N
0.17%N
0.18%N
0.19%N
1.1%N
1.6%N
GaAsPN
GaAsP
Hydrogen
Oxygen
Potential (V vs. Ag/AgCl)
0.2
0.24
0.28
0.32
0.36
0.4
0.44
0.48
GaP
0.06%N
0.11%N
0.19%N
1.6%N
GaAsPN
GaAsP
1.5
0 5 10 15
pH
0.52 1.5 2.5 3.5 4.5
pH

Approach: Two configurations
One for material study and one for possible high-efficiency
tandem cell.
• GaPN/tj/ p-GaP or
n-silicon substrate
– Undoped
– No low energy
transition expected
• GaPN/tj/ p/n silicon
substrate (tandem)
– Undoped
– Low energy
transition expected
1 µm
i-GaP.9806N.0194
1 µm p-GaP.9818N.0182
0.04 µm
GaP
n-Si substrate
0.04 µm GaP
n-Si
p-Si substrate
Ti/Pd/Al/Pd/AuOhmic contact
MF097
Al Ohmic contact
MF098

Goal: III-V Stable nitride III-V tandem nitride material water splitting cell
-
E
EF
VB
+
Ohmic Contact
p
Si
n
E
Tunnel Junction
GaP .98 N .02
E g ~2eV
e -
H 2
CB
O 2
e -
H + hν
h +
hν

Goal: Stable III-V nitride material
Light-Driven Water Splitting
MF771-4 2-electrode J-V in acid 1 sun illumination
0.0005
0
-0.2 -0.1 0 0.1 0.2 0.3 0.4
Current Density (A/cm^2)
-0.0005
-0.001
-0.0015
Summary:
• GaPN/Si tandem can
photoelectrochemically split
water but very low efficiency.
• GaPN has enhanced corrosion
resistance
Dr. Todd Deutsch
-0.002
Applied Potential (V)

Goal: Stable III-V nitride material
GaInPN:Si – Two Tandem Cell
Configurations
1 µm
GaInPN
5.0E-02
4.5E-02
Ga .96
In .04
N .024
P .976
4.0E-02
Silicon
npc 2 (AU)
3.5E-02
3.0E-02
2.5E-02
2.0E-02
1.5E-02
1.0E-02
5.0E-03
E g
= 2.0 eV
1 µm
0.1 µm
GaInPN
0.0E+00
1.5 1.7 1.9 2.1 2.3 2.5 2.7
photon energy (eV)
Silicon
Zn-doped GaInPN
Jeff Head, Paul Vallett

I vs. V Plot
Ga .96 In .04 N .024 P .976
0.2
Characterization
PC Onset
-1.2 -1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8
0
-0.2
J (µA/cm 2 )
-0.4
-0.6
Dark
Light
Jeff Head, Paul Vallett
-0.8
-1
Potential (V vs. Ag/AgCl)
Cathodic PhotoCurrent
H 2 Evolution
p-type

Goal: Stable III-V nitride material
Stability by Profilometry
Ga .96 In .04 N .024 P .976 :
Etched: 0.13µm/24hrs
Ga .95 In .05 N .025 P .975 :
Etched: 0.14µm/24hrs
GaInP 2 :1-2µm/24hrs
Jeff Head, Paul Vallett

Goal: Stable III-V nitride material
Water Splitting
• Photon to chemical energy
conversion efficiency
– GaInPN: ~30%
–GaInP 2 : ~90%
• Zn-doped layer improves
efficiency
• Nitrogen decreases electronic
properties of material
Solar to Hydrogen Efficiency (%) .
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
Zn -
GaInPN/Si
GaInPN/Si
Jeff Head, Paul Vallett

Goal: Evaluation of high efficiency thin-film CIS-based material
Electrodeposited CIGSS
• CIS material system: CuInSe 2
–CuGaS 2
– E g
range: 1.0 – 2.5eV
• CuGaSe 2
: E g
= ~1.7
• High PV efficiencies in this system
• Potential low cost, low energy deposition
• Two configurations:
– Single materials
– As a layer in a multijunction system
Relationship of bandgap to alloy composition. Gray
area represents compositions that produce
bandgaps in the range 1.7 – 2.1eV
Dr. Jennifer E. Leisch
Goal: Evaluation of high efficiency thin-film
material and thin-film based tandem cell

Goal: Thin-film based PEC tandem cell
Tandem Cell Configuration
CGS Grown by Thermal Evaporation
Maximum theoretical
water-splitting
efficiency = 28%
CGS: 1.3µm
ITO: 150nm
Dr. Jennifer E. Leisch

Goal: Evaluation of high efficiency thin-film CIS-based material
Results – Band Positions
Jennifer E. Leisch
Illuminated Open Circuit Potential Measurements
High intensity DC W lamp illumination

Goal: Thin-film based PEC tandem cell
Results – Band Energy Diagram
Energetics say:
H 2 O splitting, but
very low currents
Why?
- Poor ITO TJ
- Kinetics for H 2
- Recombination
- Band shifting
- Corrosion
Dr. Jennifer E. Leisch

Results – Water Splitting
Efficiency
Possible band edge alteration on anodic scan. . .
Jennifer E. Leisch
J sc (Cathodic Scan)
pH 6 RuHex, 1sun -5.7µA
Acid, ~ 20 Suns -3.4µA
Base, ~ 20 Suns -7µA
.01%

2mA/cm 2 , 3M H 2 SO 4 , 2hrs.
100mW/cm 2 W Illumination
.03mL/min-cm 2 H 2 Evolution
Results – Stability
The high temperature CGS
deposition process damages
the ITO layer.
Corroded
Electrodeposition may be a
better approach.
As-deposited
As-deposited
Dr. Jennifer E. Leisch

IV – Goal: Low-Temp CuGaSe 2 Deposition
CuGaSe 2 Electrodeposition
2+
−
0
Cu + 2e
⇔ Cu E = + 0. 342V
3+
−
0
Ga + 3e
⇔ Ga E = −0.
549V
HSeO e H Se HO E V
0
2 3
+ 4 − + 4 + ⇔ + 3
2
=+ 0.739
Lincot, D., et al., Chalcopyrite thin film solar cells by electrodeposition. Solar Energy, 2004. 77(6): p. 725-737.
Dr. Jennifer E. Leisch

IV – Goal: Low-Temp CuGaSe 2
Deposition
Film Microstructure
Dr. Jennifer E. Leisch
XRD
No Ga-containing
phases observed

IV – Goal: Low-Temp CuGaSe 2
Deposition
Film Composition by ICP-MS
(inductively coupled plasma-mass spectrometery)
• Films close to the proper
stoichiometry (1:1:2) were
electrodeposited (ambient T
and P)
• Gallium deposition can be
obtained via codeposition
mechanism
• Ga must be amorphous or of
very small structure
• Excess Se inhibits Ga
deposition
• Next step: Annealing
Dr. Jennifer E. Leisch

Gallium Indium Phosphide/Electrolyte System
Used to gain a fundamental understanding of
semiconductor/electrolyte junctions
(-)
E CB
Eg = 1.83 eV
OK
Energy
(+)
E VB
p-GaInP 2
E F
H 2
O/H 2
Band edges
are too
negative
Band edges
are pH
sensitive
H 2
O/O 2

Band Edge Engineering
The goal is to shift the band edges by surface
modification to provide the proper energetic overlap.
-
CB
Modification
Negative charges move the
bandedges negative.
E
+
VB
CB
VB
H 2 O/H 2
H 2 O/O 2
Positive charges move the
bandedges positive.
x

Metallo-Porphyrins
x
x
Octaethyl Porphyrin
~OEP~
Tetra(N-Methyl-4-Pyridyl)porphyrin
~TMPyP(4)~
Insoluble porphyrin dissolved in methlene chloride and electrode coated
by drop evaporation. Water soluble porphyrin applied by soaking the
electrode in a solution of the porphyrin.

Band Edge Engineering - GaInP 2
V FB
vs. pH for RuCl 3
+ RuOEP Treated GaInP 2
Electrode
VFB
-0.200
0.000
0.200
0.400
0.600
0.800
1.000
1.200
1.400
RuCl3 & RuOEP
Untreated
-2 0 2 4 6 8 10 12
The band edges here should
be able to split water…..but no
water splitting observed.
pH
H 2
O Oxidation
Potential
S. Warren and J. Turner, USDOE Journal of
Undergraduate Research, p 75 (2002)

Photoelectrochemical Water Splitting
Current Staff
Heli Wang
Todd Deutsch, Postdoc
Mark Reimann, Colorado School of Mines (PhD
Student)
Recent Past
Jennifer Leisch , Colorado School of Mines (PhD
2006 – now at Stanford)
Jeff Head, Colorado State University (SS- now at
ASU)
Paul Vallett, University of Vermont (SS)
Scott Warren, Whitman College (SS - now at
Cornell)
380 ppm
Others
Vladimir Aroutiounian, University of Yerevan,
Armenia (CRDF, ISTC)
Arturo Fernández, Centro de Investigación en
Energía-UNAM, Mexico
Funding: DOE Hydrogen Program