2-Photon Water Splitting

This presentation can be found at
https://dcwww.fysik.dtu.dk/~brse/EMRS.pdf
2-Photon Water Splitting
Brian Seger
Center for Individual Nanoparticle Functionality (CINF)
Department of Physics, DTU
In collaboration with:
Bastian
Mei
Dowon
Bae
Thomas Ivano
Pedersen Castelli
Karsten
Jacobsen
Peter
Ole
Ib
Vesborg Hansen Chorkendorff
Why Photoelectrolysis
• Solar irradiation produces 69,000TW just on land.
• For a sustainable future we need an energy source that give us
~14 TW of power (28TW by 2050).
• To make 28TW of photoabsorbers, we can only use earth
abundant materials.
-Vesborg and Jaramillo,
RSC Advances, 2012
Looking at Efficiency
• Theoretically you need 1.23 eV for water splitting.
• However there are losses:
•
•
•
•
Expensive
H2O/H2 overpotential (in base) = 50mV (PtNi)
H2O/O2 overpotential (in base) = 350mV (RuO2)
Ohmic losses = 100mV
Semiconductor = 500mV (per semiconductor)
Cheap
100mV (NiMo)
350mV (NiFeOx)
• Realistically we need a voltage of
1.8 V to split water.
• This would correspond to needing
a 2.3 eV band gap material.
• This drastically limits our
absorption of the solar spectrum.
Photoelectrolysis vs. PV + Electrolyzers
% of Solar Cell Cost
• For solar cells, the cost is primarily due to balance of plant (BOP)
100
issues, not the photoabsorber.
Solar Cell Module
– BOP issues favors high efficiency
solar cells.
– Photoelectrolysis will only have 1
balance of plant cost, while PV+Elec.
will have 2.
– Thus a 2-photoabsorber may become
economically viable.
Other Costs
80
60
40
20
0
2008
2009
2010
2011
2012
2013
Year
Rethinking Energy, IRENA, 2014
and greentech (for 2008 data)
• An optimal 2-photon solar cell produces nearly the optimal voltage for
water splitting.
2-Photon Type
Voperating
Band gap #1
Band gap #2
Solar Cell
2.0 V
1.9 eV
1.0 eV
Photoelectrolysis
1.8 V
1.6-1.8 eV
0.9-1.0 eV
-Marti et al., Sol. Ene. Mat., 1996 and Hu et. al, E&ES, 2012
Physical
Design
Optical Absorption
Properties
Energetic
Properties
Energy Levels of Our Processes
•
•
•
The photoanode will oxidize water to oxygen while the
photocathode will reduce the protons to hydrogen.
The Fermi levels of both photoabsorbers must equilibrate.
Catalysts will be needed to improve reaction kinetics.
Walter et al. , Chem Review, 2010
Photo-Catalysis
•
•
•
Using the cubane catalysts increased the H2 evolution reaction
efficiency substantially compared to pure Si.
With this composite we past the break even point, but still have
much further to go.
These composite dies if the oxygen concentration is above
~100ppb.
Hou Y. D., et al., Nature materials, 10, 434-438 (2011).
Photochemistry Issues
•
The highly reductive conduction band is Si actually hurts
performance.
- We have a guaranteed loss of at least 500 mV with Si.
Band bending issues prevent us from maximizing voltage.
e-0.5
-0.5
0.0
h+
eH+/H2
H+/H2
h+
Si
V vs. NHE
0.7
V vs. NHE
•
Si
Isolating the Photovoltage
• Adding a high doped n+ layer to the Si does 2 things:
– Creates optimal band bending independent of electrolyte.
– Allows tunneling at the semiconductor-electrolyte interface.
• This doping in effect allows the band positions to effectively
float.
-0.5
e-
Tunneling
through Si
H+/H2
0.7
V vs. RHE
h+
Doping
p
n+
Si
(Boettcher et. al., JACS, 2011)
Using TiO2 as a Protective Layer
• The TiO2 layer actually allows for a slight improvement
compared to Pt on Si.
• The anti-reflectivity properties of TiO2 greatly help in light
absorption.
• Up to 300nm the TiO2
thickness has no effect on
the CV onset or slope.
• Vacuum annealing at
400°C for 90 minutes
has no effect on onset
or slope.
(Seger et. al., JACS, 2013)
Long Term Stability
•
TiO2 protected a Si photocathode (Design 1) for 30 days with no
noticeable degradation.
•
Performance decrease after 20 days was due to catalyst detachment
or contaminants in the electrolyte.
•
Redeposition of catalyst after 30 days brought back original
performance.
4
-4
Photocurrent (mA/cm2)
Photocurrent (mA/cm2)
0
ALD 100nm TiO2 /5nm Ti/n+p Si
(Vacuum annealed at 400°C for 1 hour)
Ran at +300mV vs. NHE
-8
-12
-16
-20
-24
0
5
10
15
20
25
30
0
-4
-8
-12
100nm TiO2 /5nm Ti/n+p Si
(Vacuum annealed at 400 °C for 1 hour)
Initially
After 1 Day
After 30 Days
Replatinize
(after 30 days)
-16
-20
-24
0.1
0.2
Time (days)
(Seger et. al., RSC Advances, 2013)
0.3
0.4
Voltage (V vs. RHE)
0.5
• Fundamentally the large bandgap (LBG) must come before the
small bandgap (SBG) photoabsorber.
Optical Absorption
Properties
• The question is what side is the photoanode and what side is
the photocathode?
(Seger et. al., E&ES, 2014)
Using Computational Screening
• High throughput screening of 2,400 candidates (all of which have
been synthesized) were used to look for potential candidates.
• Looking for stable
candidates in either acidic
or basic conditions.
• We only had 4 conditions:
–
–
–
–
Correct band gap
Correct band positions
Stability
Earth abundant
Earth Abundant
Bandgap
Design Electrode
(eV)
Photo0.9-1.5
Design cathode
1
Photo1.5-2.1
anode
Photo- 1.5-2.1
Design cathode
2
Photo- 0.9-1.5
anode
Candidates Candidates
in Basic
in Acidic
Conditions Conditions
2
1
FeSbS,
NaTiCuS3
BaFeMoO6
0
1
Ca3(CoO3)2
0
0
0
0
(Seger et. al., E&ES, 2014)
Big Problems
• Why is finding the right photoabsorbers so hard?
– Answer: Photoabsorber stability.
• Can we eliminate the stability problem?
– Yes, with corrosion resistant protection layers.
• Potential Materials- Metals, Semiconductors, Insulators
Metal Protection Layers
• Issue: Metals absorb a lot of light.
– Solution: Metals can only be used on small bandgap side.
• Issue: They interfere with the photovoltage/bandbending.
– Solution: Create a p-n junction.
• Issue: Many metals convert to oxides at their surface.
– Solution: Make this oxide works as a catalyst (favors O2 evolution
catalysts). (Our use of Ir on Si for O2 evolution.)
Metal won’t work
Metal may work
Metal won’t work
Metal may work
Oxidized metal
may work
Semiconductor Protection Layers
• Issue: They are not very conductive.
– Solution: Protection layers can be ~50nm or less.
• Issue: They may absorb some light.
– Solution: Use large bandgap semiconductors (bandgap > 3.0 eV).
• Issue: Bandbending may prevent electron transfer.
– Solution: Align the bands properly. This takes a little work.
Semiconductor Protection Layers
• At the Photocathode:
– The semiconductor transfers electrons through the conduction band (n-type).
– The conduction band needs to be near the H2 evolution potential.
– We have tested Ti, TiO2, and MoS2 for photocathodic protection.
• At the Photoanode:
– The semiconductor transfers holes through the valence band (p-type).
– The valence band needs to be near the O2 evolution potential.
– We have tested NiO for photoanodic protection.
Work-around for Protection Layers
• Some semiconductors form ohmic contacts with metals (i.e.
catalysts).
• If your protection layer sees minimal electrolyte, band bending
and hence band position becomes irrelevant.
We published this
general principle in
2012. (See fig. 5)
Ohmic
contact
Won’t Work
Will Work
We are in process of
publishing this in more
detail
Work-around for Protection Layers
• Some semiconductors form ohmic contacts with metals (i.e.
catalysts).
• If your protection layer sees minimal electrolyte, band bending
and hence band position becomes irrelevant.
Great paper
explaining this
concept
Won’t Work
May
work
Will
Work
Insulating Protection Layers
• Issue: Insulators aren’t conductive
– Solution: Electronically tunnel using very thin layers (~2 nm)
• Issue: Can a film 2nm thin be pinhole free?
– Maybe.
• Issue: Is a 2nm thick layer mechanically durable enough?
– Maybe.
Protection Layers Summary
• Metals
– Great for protecting the small bandgap.
• Semiconductors:
– TiO2 based candidates for cathodic protection layers.
– Thin, pure NiO for anodic protection layers.
• Insulators- Works, but a gamble with long term stability.
Protected Photoabsorbers Candidates
• If we use protection layers, stability is not an issue.
• If we create a p-n junction, band position is not an issue.
• Thus we are left with 2 parameters
– Bandgap
– Earth Abundance
Design
Screening
# of
Parameters Candidates
0.9 ≤ EG ≤ 1.5
51
LBG 1.5 ≤ EG ≤ 2.1
50
SBG
BaAs2, BaCaSn, Ba2Cu(PO4)2, Ba2FeMoO6, Ba3(Si2P3)2, BaLaI4, Ba3P4,
CaBaSi, Ca3(CoO3)2, Ca2Si, Ca3SiO, CoAsS, CuCl2, CuP2, FeS2, FeSbS,
K2Mo6S6, KNbS2, KPb, KSnAs, KZnAs, LaAs2, LaZnAsO, LaZnPO,
LaS2, MgP4, MnP4, Na4FeO3, Na4FeO4, NaNbS2, NaNiO2, Na3Sb, NaSnP,
NaTiCuS3, NaTiS2, NaZnP, NbFeSb, NbI3, Si, SnS, Sr2As2, Sr3As4, Sr3SbN,
SrCaSi, SrCaSn, SrLaI4, Sr(ZnP)2, V(S2)2, Zn2Cu(AsO4)2, ZrBr3, ZrCl3
B, BP, BaCu2SnS4, Ba(MgSb)2, BaP3, Ba4Sb2O, Ba2ZnN2, Ca3AlAs3,
Ca(BC)2, Ca3(BN2)N, Ca(MgSb)2, Ca Na10Sn12, Ca3VN3, Ca(ZnP)2,
CoBr2, CuSbS2, Cu2O, Cu3VS4, FeBr2, FeSO4, Fe(SiP)4, I2, K3As, K2Ni3S4,
K4P6, K3Na2SnAs3, K2NiAs2, KSb, KV(CuS2)2, KZnP, KCuZrS3, MgAs4,
NaCuO2, NaNbN2, NaP, NaSbS2, Nb6F15, NbI5, SnZrS3, SrP, Sr3P4, SrPbO3,
TiBrN, TiI4, TiNCl, Sn2TiO4, WBr6, ZnSiAs2, ZrCl2, Zr2SN2,
Conclusions
• The 2-Photon approach to water splitting has certain advantages over
using a solar cell + electrolyzer.
• Finding the perfect photocatalyst is extremely hard.
• Using a protection layer isolates the stability issue, thus making
finding a good photoabsorber much easier
• We have very promising candidates for protection layers.
• We also need to realize that we aren’t constrained by having a large
band gap photoanode and a small band gap photocathode.
Acknowledgements
• The following people made this work possible:
Bastian
Mei
Dowon
Bae
Thomas Ivano
Pedersen Castelli
Karsten
Jacobsen
Peter
Ole
Ib
Vesborg Hansen Chorkendorff
(Section Leader)
• We most importantly must thank the CINF and CAMD
grants from Danmarks Grundforskningfond for
funding this research.
This presentation can be found at
https://dcwww.fysik.dtu.dk/~brse/EMRS.pdf