HydroGEN Seedling: Novel Chalcopyrites For Advanced …

 

 

HydroGEN Seedling: Novel Chalcopyrites for Advanced

Photoelectrochemical Water Splitting

Nicolas Gaillard

University of Hawaii / Hawaii Natural Energy Institute

2440 Campus Road, Box 368

Honolulu, HI 96822

Phone: 808-956-2342

Email: ngaillar@hawaii.edu

DOE Manager: Katie Randolph

Phone: 720-356-1759

Email: Katie.Randolph@ee.doe.gov

Contract No: DE-EE0008085

Subcontractors:

• University of Nevada, Las Vegas, Las Vegas, NV

• Stanford University, Stanford, CA

HydroGEN Energy Materials Network nodes:
• Lawrence Livermore National Laboratory, Livermore,

CA

• National Renewable Energy Laboratory, Golden, CO

• Lawrence Berkeley National Laboratory, Berkeley, CA

Project Start Date: October 1, 2017

Project End Date: September 30, 2020 (subject to
annual Go/NoGo decision)

Overall Objectives

The overarching goal of this project is to create a
chalcopyrite-based, semi-monolithic, tandem
hybrid photoelectrode device prototype that can
operate for at least 1,000 hours with solar-to-
hydrogen STH efficiency >10%. This effort is

supported by advanced characterization and

theoretical modeling to accelerate the development

of materials and interfaces. Specifically, our

program aims to:

• Develop high-throughput synthesis techniques

to create efficient copper chalcopyrite-based

materials with ideal optoelectronic properties

for photoelectrochemical (PEC) water splitting

•

Identify appropriate surface treatments to

prevent photocorrosion, improve surface

energetics, and facilitate the hydrogen

evolution reaction

• Create a new method to integrate temperature-

incompatible materials into a semi-monolithic

PEC device structure.

Fiscal Year (FY) 2019 Objectives

• Synthesize new wide-bandgap chalcopyrite

made of Cu, In, Al, and Se using printable

molecular inks.

•

Identify mechanisms responsible for improved
durability and energetics in surface-modified

chalcopyrites.

• Further develop the concept of

transparent/conductive binders to serve as a

binding material in the semi-monolithic

integration scheme.

Technical Barriers

This project addresses the following technical

barriers from the Fuel Cell Technologies Office

Multi-Year Research, Development, and

Demonstration Plan1:

• Materials Efficiency (AE)

• Materials Durability (AF)

•

Integrated device configuration (AG)

• Synthesis and Manufacturing (AJ).

Technical Targets

In Task 1, we further evaluated the effect of alkali

doping on point defects in CuGa3Se5 and

developed wide-bandgap Cu(In,Al,B)Se2 and
developed a baseline “printing” process. In Task 2,
we integrated non-precious catalytic-protecting

layers and Cd-free buffers to enhance
chalcopyrites’ durability and charge separation

efficiency, respectively. Finally, in Task 3, we

tested the transparent conductive epoxy/particle
composites developed in FY 2018 for semi-

monolithic PEC device integration. The status of

this project’s technical targets is documented in

Table 1.

1 https://www.energy.gov/eere/fuelcells/downloads/fuel-cell-technologies-office-multi-year-research-development-and-22

FY 2019 Annual Progress Report

1

DOE Hydrogen and Fuel Cells Program

Gaillard – University of Hawaii

Hydrogen Fuel R&D / Photoelectrochemical Production

FY 2019 Accomplishments

• Modeled various chemical defects in

chalcopyrites and assessed their impact on
optical and electrical properties.

• Evaluated Si doping on the efficiency and

durability of CuGa3Se5 photocathodes.

• Analyzed printed chalcopyrites via advanced

spectroscopy techniques, showing a reduction

in surface contamination after post-annealing

in selenium atmosphere.

• Performed continuous operation of a WO3-

coated CuGa3Se5 photocathode for over 6

weeks.

• Successfully fabricated highly conductive and

transparent binders made of epoxy and silver
particles.

INTRODUCTION
Our multidisciplinary program combines advanced synthesis techniques, unique characterization, and
theoretical approaches to improve the efficiency and durability of chalcopyrite-based hybrid photoelectrode

devices, with the final goal of producing a chalcopyrite-based, semi-monolithic device with at least 10% STH

efficiency.

APPROACH
We aim to advance the performance of previously identified wide-bandgap chalcopyrite materials through
alkali doping, as well as develop and test the water-splitting viability of the next generation of chalcopyrites.
We also see the unrealized potential to improve the PEC/electrolyte interface energetics and stability, which is

addressed by investigating alternative buffer materials and protective layers. Finally, to avoid the heat-stress
issues facing all-chalcopyrite monolithic tandem devices, we investigate a semi-monolithic structure, which
will utilize a transparent conductive bonding polymer and exfoliation technique to avoid exposing the bottom

photovoltaic (PV) driver in these devices to the high temperatures required for wide-bandgap chalcopyrite

synthesis.

RESULTS

Modeling

Task 1: Modeling, Synthesis, and Characterization of Chalcopyrite Photocathodes

The theoretical studies over the past year have aimed at improving the understanding of fundamental and
defect-related properties of the absorbers and interfaces to correlate with the observed performance in the
synthesized material and resulting devices. We have extensively considered the role of point defects in the

parent chalcopyrite compounds of desirable top-cell absorber alloys (e.g., Cu(In,Ga)S2) to investigate how

synthesis conditions influence defect populations. This knowledge is critical to both targeting desired electrical

properties and mitigating problematic traps and recombination centers; it also intimately depends on the

chemical potentials of the atomic species during growth and post-processing steps. In Figure 1(a), we include

the phase diagram of one parent compound, CuGaS2, as calculated with advanced hybrid functionals and
showing the phase boundaries with competing phases as a function of the Ga and Cu chemical potentials. We

translate this information into the calculated formation energies of native defects in Figure 1(b), where we find
significant regions in the phase diagram that yield a strong competition between Cu vacancies (VCu) acceptors,

interstitial donors (Cui), and disorder where Cu incorporates as a deep acceptor antisite on the Ga sites (CuGa).

While high VCu concentrations are desirable for highly p-type absorbers, significant incorporation of defects

exhibiting deep states within the bandgap like CuGa may lead to unwanted sub-gap absorption, which is

detrimental for absorbers in tandem devices. We continue to actively investigate the influence of the dominant

defects (both intrinsic and extrinsic) on the optical and electrical properties as a function of growth and

annealing conditions to identify mitigation strategies (e.g., passivation) or targeted synthesis conditions.

FY 2019 Annual Progress Report

2

DOE Hydrogen and Fuel Cells Program

Gaillard – University of Hawaii

Hydrogen Fuel R&D / Photoelectrochemical Production

Figure 1. (a) Calculated phase diagram for CuGaS2 as a function of Cu and Ga chemical potentials, showing the phase

boundaries with competing phases. Formation energy diagrams of native defects shown as a function of the Fermi level are

included in (b) corresponding to various points in the phase diagram. While more Cu-poor conditions favor the most p-type
material (e.g. point F) owing to Cu vacancies (VCu), nominally more Cu-rich conditions can still yield considerable VCu and p-
type conductivity, but also CuGa antisites that exhibit localized states within the bandgap may detrimentally influence the
optical transmission and impact performance in tandem cells.

Ordered vacancy compounds

The prospect of using Si doping to improve the PEC performance of CuGa3Se5 photocathodes was

investigated. Evaporating Si onto the absorbers in situ led to the formation of SiSe2, which immediately

reacted to form SiO2 and H2Se when exposed to air (SiSe2 + 2 H2O  SiO2 + 2 H2Se; as evidenced by the

H2Se gas that was detected when the co-evaporation chamber was vented). It is therefore likely that the
CuGa3Se5 polycrystal surfaces are coated with thin SiO2 layers, in agreement with a former report that found
an accumulation of Si dopant at Cu(In,Ga)Se2 grain boundaries. Chopped light linear sweep voltammetry
(CLIV) characteristics of bare photocathodes (glass/Mo/CuGa3Se5) in 0.5 M H2SO4 under 1 Sun illumination

(Figure 2) indicated that the addition of Si passivates the films. Si doping increased the onset potential by ~0.1

V and led to better photocurrents at 0 V vs. RHE, relative to the best baseline photocathode. It is unlikely that

Si dopes grain interiors because the SiCu and SiGa substitution defects should donate electrons, whereas the
absorbers were all p-type. Intrinsic CuGa3Se5 is p-type, and the Si-doped solar cells (CuGa3Se5/CdS p-n
junctions) performed similarly to the baseline (not shown). The improvement in initial PEC performance is

promising, as SiO2 may also act as a protective layer that can improve the durability of the films due to its acid

resistance. More characterization is underway to confirm the hypotheses that Si leads to electronic and

chemical passivation.

Figure 2. CLIV characteristics of standard photocathodes (glass/Mo/CuGa3Se5): the best baseline (black), 3.2 nm Si (blue)
and 6.2 nm Si (purple).

FY 2019 Annual Progress Report

3

DOE Hydrogen and Fuel Cells Program

Gaillard – University of Hawaii

Hydrogen Fuel R&D / Photoelectrochemical Production

Photophysical characterization

The optical properties of CuInGaS2 thin film samples were investigated to determine their bandgap and sub-
bandgap absorption (defect absorption). The ability to tune the optical bandgap of these materials enables

optimization of the balance between light-harvesting efficiency and maximum photovoltage for driving
hydrogen evolution from water splitting. We investigated the optical properties of the materials by

photothermal deflection spectroscopy, a sensitive type of absorption spectroscopy (Figure 3). Here we

observed a clear band onset at 1.91 eV with tailing from defect absorption. However, the maximum

photovoltage can be limited by intrinsic defects in the material, which can vary for a given composition and
growth/processing strategy. Therefore, a careful understanding of band position and related defects is critical

to further the application of CuInGaS2 materials for hydrogen production. To access intrinsic defects, we

investigated the light emission properties of the films as a function of temperature down to 11 K. Typically,
defect luminescence efficiency at room temperature is low yet at cryogenic temperature de-trapping by thermal

excitation and non-radiative recombination can be minimized. From these measurements, two luminescence

peaks were observed at 1.42 eV and 1.57 eV where can be related to deep trap states located at 340 meV and

490 meV below the band edges, respectively. Further work may seek to identify the origin of these defects,
mitigation, and their relationship to function PEC performance.

Figure 3. Photothermal deflection spectrum (black), taken at room temperature, and photoluminescence spectrum (green)

of CuInGaS2 on fluorinated tin oxide (FTO) glass, taken at 13 K. Also show are the two Gaussian peaks used to fit the PL
spectrum centered at 1.42 and 1.57 eV.

Task 2: Interface Engineering for Enhanced Efficiency and Durability

X-ray photoelectron spectroscopy

In the second year of the project, University of Las Vegas (UNLV) continued to focus on characterizing
HNEI-supplied samples using X-ray Photoelectron Spectroscopy (XPS, at UNLV) and X-ray Emission

Spectroscopy (XES, at the Advanced Light Source, Lawrence Berkeley National Lab). Two sample sets,

consisting of CuInS2 and CuIn(S,Se)2, were analyzed by the UNLV team. The CuInS2 samples are precursors
to the selenized CuIn(S,Se)2 samples and are made by depositing a solution-based mixture, consisting of CuCl,

InCl3, and thiourea (CH4N2S), onto a molybdenum-coated soda-lime glass substrate. One sample set was

synthesized in air and spin-coated in air (labeled as “Air”). A similar procedure was used to prepare the other

sample set (labeled as “GB HPLC CuInS2”); in contrast, however, the synthesis was done in a glove-box (GB),
while the spin-coating was still performed in air. The “HPLC” abbreviation refers to the HPLC-grade methanol

used to prepare these two samples. The XPS surveys of four samples from the two sets are shown in Figure 4 .
For “Air” synthesis, the carbon signal is stronger; this is not unexpected, as the synthesis environment

generally impacts the surface chemical environment found on exposed sample surfaces.

A similar trend can be seen at the near-surface bulk using XES (Figure 4). The oxygen and sodium signals in
the XPS survey spectra also increase after selenization, and the Na signal is most pronounced for the sample

FY 2019 Annual Progress Report

4

DOE Hydrogen and Fuel Cells Program

Gaillard – University of Hawaii

Hydrogen Fuel R&D / Photoelectrochemical Production

set synthesized in the glovebox. In contrast, the selenization process removes nitrogen and chlorine signals,
likely due to the removal of excess starting materials. The reduction in nitrogen and chlorine after selenization

can also be seen using XES (Figure 4), which gives insight into the chemical structure in the near-surface bulk.
Furthermore, we find lateral inhomogeneities in the chemical composition (in particular the presence of Ca,
Mo, and S signals) that can be correlated with the morphology of the different films under study. Figure 4
shows the XES spectra of three samples from the same two sample sets, each measured at two different spots.

Using excitation energy of 380 eV, C K, Ca L2,3, S L2,3, Cl L2,3, and Mo M4,5 can be detected. Due to higher

harmonics/orders from the undulator/beamline, N K emission can also be observed.

Figure 4. (left) XPS survey spectra of four samples: CuIn(S,Se)2 and CuInS2, with precursors synthesized in air (“Air”) and in

a glovebox (“GB HPLC”); both sample sets were then spin-coated in air, and Se-containing samples were further selenized.

Prominent peaks are highlighted to emphasize changes between the different sample sets. (right) XES spectra of three

samples, each measured at two different spots (A1, A2, etc.). C K, S L2,3, Cl L2,3, and Ca L2,3 spectral features are labeled in
the main graph. The exploded graph shows Mo M4,5 and N K detail spectra (194–205 eV). For clarity, the intensity of the

Mo M4,5 peaks of the “GB HPLC CuInS2 (B2)” and “GB HPLC CuInS2 (B1)” was divided by 80 and 2.5, respectively. The

label “2nd” indicates peaks observed in the second order of the spectrometer grating.

Development of tunable “buffers” via combinatorial methods

Due to the efforts in FY 2019, we now have a better understanding of the factors leading to the increased open-

circuit voltage up to 925 mV for Zn1-xMgxO-coated and Cd2+ surface-treated (Cd PE) CuGa3Se5 absorber
devices. The effect of Cd PE on the CuGa3Se5 surface was investigated with XPS, UPS, and Kelvin probe at

NREL. Ultraviolet to visible spectroscopy and Kelvin probe measurements were performed on Zn1-xMgxO:Ga

films to find the band positions. We found that NH4OH treatment reduced surface oxidation. Cd PE treatment

introduced Cd on the absorber surface. A reduction in Cu concentration suggested that Cd may be occupying
Cu sites. Cd PE treatment moved the fermi level position upward in the absorber, indicating a change in

surface conductivity type towards becoming intrinsic. The n-type counterpart for the p-type CuGa3Se5 absorber
is possibly doped ZnO. Change of absorber surface conductivity type to either intrinsic or n-type may create a
better p-i-n junction. The other notable outcome is the expected conduction band maxima upward shift in Zn1-
xMgxO with increasing Mg composition. This reduces the conduction band offset between CuGa3Se5 absorber
and Zn1-xMgxO. We believe that, together, these two aforementioned beneficial factors are the reason for the

improved VOC of these devices.

FY 2019 Annual Progress Report

5

DOE Hydrogen and Fuel Cells Program

Gaillard – University of Hawaii

Hydrogen Fuel R&D / Photoelectrochemical Production

Figure 5. (left) CBM shift for Zn1-xMgxO with Mg composition, and energy band positions for CuGa3Se5 with different surface
treatments. (right) three-electrode chopped light linear sweep voltammograms (LSV) measurement data for CuGa3Se5/Zn1-
xMgxO devices.

Durability

During FY 2019, our team worked to develop protective coatings for PEC devices in acidic electrolyte
conditions. This work focused on improving the durability of CuGa3Se5 photocathodes that were coated by
atomic layer deposition with WO3, which is predicted to have stability in acid. Figure 6 shows PEC
measurements of two photocathodes: a bare CuGa3Se5 absorber and a CuGa3Se5 absorber that has been coated
with an approximately 4 nm thick WO3 layer and nanoparticulate Pt catalyst (CuGa3Se5|WO3|Pt). The linear
sweep voltammograms (LSV) in Figure 6a show that both electrodes demonstrate similar photocurrent onset at
+0.3 V vs. RHE and reach a saturation photocurrent density (j) of about -8 mA/cm2 under 1 Sun illumination.
Figure 6b shows the constant illumination chronoamperometric (CA) durability testing of these same devices
along with a replicate sample of each type, with the potential being held in the light-limited region for each.
Both of the WO3-coated CGSe devices achieved new durability records for any non-Si photocathode,
bypassing 19,510°C cm-2 and 21,490°C cm-2 respectively over the course of the durability experiments,
compared to 17,200°C cm-2 reported by our team on bare CuGa3Se5.[1]

Figure 6. (a) LSV of a CuGa3Se5/WO3/Pt photocathode (green) and a bare CuGa3Se5photocathode (maroon) under
hydrogen evolution reaction conditions, with their respective dark currents shown in dashed lines (overlapping the j=0 axis)
(b) CA of these same devices held at constant potential, along with the CAs for a replicate sample of each type (light green
and light maroon); all experiments were conducted under continuous 1 Sun illumination in 0.5 M sulfuric acid electrolyte
with a Hg/HgSO4 reference electrode and an Ir/IrOx counter electrode, with hydrogen gas bubbling through the solution.

Task 3: Hybrid Photoelectrode Device Integration
The goal of this task is to solve process compatibility issues between material classes by bonding a
chalcopyrite photoelectrode onto a fully processed PV driver. In the last fiscal year, we focused on the
development of transparent conductive binders (TCBs) for semi-monolithic hybrid device integration. In this
approach, the TCB is made by dispersing conductive media into a liquid epoxy. The TCB is then applied
between the bottom and top cell, then cured to mechanically adhere to the two devices together, thereby

FY 2019 Annual Progress Report

6

DOE Hydrogen and Fuel Cells Program

Gaillard – University of Hawaii

Hydrogen Fuel R&D / Photoelectrochemical Production

fabricating a semi-monolithic tandem device. During FY 2019, we explored a new approach to integrating the

cells of a semi-monolithic device together. Where normally the liquid TCB would be cured after application,
we have instead discovered a means to cure the liquid TCB prior to application to form a transparent

conductive freestanding layer (TCFL), as shown in Figure 7. An advantage of using a TCFL as opposed to a

TCB is that it eliminates potential spillover of the liquid TCB, which can lead to electrically shorting in the

semi-monolithic device. The TCFL shown in Figure 7 was fabricated out of a TCB mixture of aluzine epoxy
embedded with Ag-coated PMMA spheres. The amount of Ag-coated PMMA spheres present in the TCFL is

such that the sum of the cross-sectional areas of the spheres equals 1% of the total cross-sectional area
(referred to as 1% loading). Even with only 1% loading, these TCFLs exhibit excellent optical and electrical

performance. Optically, the transmittance from 300–2,000 nm stays constant at approximately 90% for the

entire range, considerably outperforming that of a standard commercially available fluorinated tin oxide (FTO)

glass substrate, as seen in Figure 7. The TFCLs also exhibit excellent electrical conduction with an out-of-
plane series resistance of 0.27 Ω.cm2, as seen in Figure 7.

Figure 7. (Left) Image of a transparent conductive free-standing layer (Center) with high optical transmittance (T>90% in

the 300–2,000 nm range) and (Right) out-of-plane series resistance (0.27 Ω.cm2) outperforming that of commercial FTO

substrates.

CONCLUSIONS AND UPCOMING ACTIVITIES

• Further validate protection with WO3 film, targeting 1,000 hrs of continuous operation.

• With help from advanced spectroscopy, improve our solution processing technique, aiming for high-

efficiency wide-bandgap Cu(In,Al,B)Se2 absorbers.

• Continue the development of ZnMgO buffers, targeting over 1V open-circuit voltage.

• Further develop the concept of semi-monolithic PEC device with improves transparent conductive

binders.

FY 2019 PUBLICATIONS/PRESENTATIONS

1. N. Gaillard, “Emerging Chalcopyrite Photo-absorbers for Renewable Hydrogen Production,” The 236th

Electrochemical Society Meeting, Symposium I04, 1909, Atlanta (GA), 2019.

2. N. Gaillard, “Wide Bandgap Chalcopyrite-based Photoelectrodes for Renewable Hydrogen Production,”

The 2019 Spring Meeting of the European Materials Research Society (E-MRS), Symposium A “Latest

Advances in Solar Fuels III”, A.5.6, Nice (France), 2019.

3. N. Gaillard, “Wide Bandgap Chalcopyrites for Photoelectrochemical Water Splitting” The Materials

Research Society Spring Meeting, Symposium ES11, ES11.05.03, Phoenix (AZ), 2019.

4. D. Palm, C. Muzzillo, N. Gaillard, T. Jaramillo, “Atomic Layer Deposited Tungsten-Based Coatings for

Durable Solar Hydrogen Production,” The 236th Electrochemical Society Meeting, Symposium I04, 1899,
Atlanta (GA), 2019.

FY 2019 Annual Progress Report

7

DOE Hydrogen and Fuel Cells Program

Gaillard – University of Hawaii

Hydrogen Fuel R&D / Photoelectrochemical Production

5. J. Varley, A. Sharan, T. Ogitsu, A. Janotti, A. Deangelis, N. Gaillard, “First-Principles Simulations of

Stability, Optical and Electronic Properties of Competing Phases in Chalcopyrite-Based Photoelectrodes,”

the 235th Electrochemical Society Meeting, Symposium I03, 1627, Dallas (TX), 2019.

REFERENCES

1. Muzzillo, C. P.; Klein, W. E.; Li, Z.; Deangelis, A. D.; Horsley, K.; Zhu, K.; Gaillard, N. “Low-Cost,

Efficient, and Durable H2 Production by Photoelectrochemical Water Splitting with CuGa3Se5
Photocathodes,” ACS Appl. Mater. Interfaces, 10 (23) (2018):19573–19579.
https://doi.org/10.1021/acsami.8b01447.

FY 2019 Annual Progress Report

8

DOE Hydrogen and Fuel Cells Program