Final Report Summary - NANOPEC (Nanostructured Photoelectrodes for Energy Conversion)
Executive Summary:
Hydrogen has the potential to meet global requirements for carbon-neutral energy carriers if produced from carbon-free primary sources such as water using carbon-free energy such as solar energy. Sustainable, cost-efficient large-scale production of hydrogen can, in principle, be established by solar photoelectrochemical (PEC) water splitting, where semiconductor electrodes absorb sunlight to drive water electrolysis: 2 H2O → 2 H2 + O2. Despite four decades of worldwide intensive research on PEC water splitting, this challenge has not been met yet, largely because of material limitations. Photoelectrode materials for solar water splitting must meet multiple and severe criteria such as efficient absorption of sunlight, separation and collection of photogenerated carriers, catalysis of water splitting reactions, chemical stability and durability for a long period, and abundance. Unfortunately, no single material is likely to meet all of the criteria needed to achieve competitive solar hydrogen production by water splitting.
NanoPEC (http://nanopec.epfl.ch/) employs innovative concepts and new methods, enabled by nanotechnology, to design nanocomposite photoelectrodes for solar water splitting, where each component performs specialized functions to overcome intrinsic limitations of single phase materials. Advanced nanostructures that maximize performance of photoanodes and photocathodes are designed, fabricated, characterized, and optimized. Newly-developed p- and n-type semiconducting oxides and oxynitrides are explored and nanostructured into innovative structures and device architectures designed to achieve high solar-to-hydrogen (STH) conversion efficiency, balancing the tradeoff between light harvesting, charge separation and collection, stability and durability. In parallel, fundamental studies and detailed investigations of model systems are carried out to improve quantitative understanding of the effect of material properties, defects and interfaces on PEC processes taking place in water photoelectrolysis. These investigations provide guidelines for rationale optimization of materials and nanocomposite electrode structures designed for high efficiency, and the performance of these structures is evaluated in operative conditions using advanced testing and analytical setups.
Significant progress in PEC water splitting has been achieved during the course of the project. In particular, a new photocurrent record of 7.6 mA cm-2, measured at the reversible hydrogen evolution potential under simulated terrestrial (AM1.5G) solar radiation, was achieved with small-size Cu2O photocathodes protected against photodegradation using a novel overlayer structure. The champion photocathode can potentially reach a STH conversion efficiency of 10% in an ideal tandem configuration, exceeding the NanoPEC target milestone. Scaled-up large (63 cm2) Cu2O photocathodes displayed a photocurrent of 4 mA cm-2, reaching 87% of our target for large modules. On the photoanode front, small-size Fe2O3 nanostructured electrodes reached a photocurrent of 3.3 mA cm-2 at the reversible oxygen evolution potential under simulated solar radiation. This could yield a STH conversion efficiency of 5% in an ideal tandem configuration, breaking the highest record ever reported for stable photoanodes for solar-induced water photo-oxidation. Although the efficiency of large Fe2O3 photoanodes was not as high as their small-size counterparts, largely due to processing difficulties in scaling-up from small to large size electrodes, they did not show any signs of degradation for at least three days.
Based on the progress achieved in this project we expect that a STH conversion efficiency of 10% is within reach using ideal tandem cells. This is comparable to the best performance obtained with state-of-the-art electrolysers powered by state-of-the-art photovoltaic (PV) cells. Unlike the latter technology, which is inherently expensive because it cannot be integrated into a single device, the PEC-based tandem cell technology has the potential to become an affordable means for large scale hydrogen production from water using solar energy. The price of hydrogen produced with such tandem cells is estimated to be in the 7 to 14 € per kg ballpark. This price is quite close to the ambitious target (5 € kg-1) established by the European Hydrogen & Fuel Cell Technology Platform, and suggests that meeting this goal may be realistic with further technological advances assisted by mass production and the economy of scale.
Project Context and Objectives:
Hydrogen (H2) will play a central role in future European energy systems. According to the Strategic Research Agenda (European Hydrogen & Fuel Cell Technology Platform, 2005), hydrogen is expected to be widely available in industrial nations at competitive cost and also serve as a major transport fuel for vehicles with a share of up to 50% by 2050. Moreover, hydrogen can be used for CO2 capture and conversion into other liquid hydrocarbons using Fischer-Tropsch catalysis and also in proton exchange membrane fuel cells to generate electric and thermal energies efficiently while leaving water alone as a by-product. Efficient, cost-effective, and carbon neutral hydrogen production will therefore be essential to a sustainable energy infrastructure.
PEC production of hydrogen under solar illumination is one of the most promising renewable technologies in the future hydrogen economy owing to the huge capacity of the solar energy, which is roughly four orders of magnitude larger than the current global energy consumption and direct conversion from electricity to chemical fuels which is ideal for energy storage and transport. PEC systems use solar photons absorbed by semiconducting photoelectrodes to split water into hydrogen and oxygen (O2) gases directly. PEC systems can thus be relatively simple compared to other solar hydrogen production systems. Primary challenges for PEC systems are to develop materials which are characterized by high photon utilization efficiencies, long lifetimes (stability against corrosion and photodegradation), and low costs.
Semiconductor photoelectrodes must meet several criteria to accomplish efficient solar-induced PEC water splitting: (a) absorb a significant portion of the solar spectrum, (b) conduct charge efficiently, (c) catalyze water reduction or oxidation, and (d) exhibit long-term stability in an aqueous solution. No single material with acceptable performance, stability, and cost has been found, despite decades of investigation. For example, n-type hematite (α-Fe2O3) and p-type cuprous oxide (Cu2O) are inexpensive semiconductors that can be used for water oxidation and reduction under solar illumination, respectively. However, hematite does not utilize photogenerated carriers efficiently on account of a very short charge carrier diffusion length. As for cuprous oxide, the redox potentials for the reduction and oxidation of monovalent copper oxide lie within the band gap, making decomposition of cuprous oxide by photogenerated charges thermodynamically favored over reduction of water 3.
To address these and other significant challenges, we use new concepts and methods, enabled by nanotechnology, to design innovative composite nanostructures in which each component performs specialized functions. This approach reduces the number of criteria that any single component must meet, thus overcoming the basic materials limitations. Computational studies are used to assist analysis of experimental results and provide a solid theoretical background to our investigations, and a wide range of advanced analytical techniques is employed to improve the understanding of structure-composition-property relationships.
NanoPEC consortium consists of eight partners, each of which is assigned to distinct tasks to make best use of their unique expertise and facilities. We aimed to go beyond the state-of-the-art in three critical areas: (1) advanced nanostructures, (2) novel semiconductors, and (3) fundamental studies on model systems, which are the topic of the WPs 1, 2, and 3, respectively. In the fourth WP, we employ the most successful materials to construct and evaluate test devices. WP5 and WP6 are organized for project management and dissemination, respectively.
As a final objective, the NanoPEC project set ambitious targets that are well beyond the state-of-the-art, specifically a one square-centimeter test device that converts solar energy to hydrogen energy with a sustained 10% efficiency (6.6 mA cm-2 with an applied voltage of 1.23 V in a tandem configuration) and a maximum performance decay of 10% over the first 5000 hours of operation and a 100 square-centimeter test device with a sustained 7% efficiency (4.6 mA cm-2 with an applied voltage of 1.23 V in a tandem configuration) and similar stability. Such performance should give a significant impact in short-to-medium term studies which are necessary to achieve long-term environmental, political, and economic benefits as well as European leadership in a sustainable hydrogen economy.
Project Results:
There are several sustainable methods to produce hydrogen other than direct PEC water splitting, such as water electrolysis based on hydroelectric, geothermal, wind, or PV power generations, thermochemical water splitting using concentrated sunlight, and catalytic conversion of biomass. Since we are facing an extra powder demand of 10–15 TW by 2050, it is important to estimate potential availability of the resources first. The potential of hydroelectric generation is approximately 1 TW and one-third of that has already been installed. It is likely that the rest of the resource is rather distributed and available only on a relatively small scale, which requires an additional investment for development. The potential of geothermal power is probably less than 1 TW unless breakthrough in drilling technologies happens. The quantity of biomass can be as large as 5 TW, although the realistic amount is much lower due to water and soil fertility limitations and necessity to feed several billion people. Wind power generation is promising, accounting for 2 TW. Among the other resources, the amount of solar energy is outstanding, being estimated to be 50 TW, although significant scientific, technological and manufacturing breakthroughs are needed for efficient utilization of solar energy.
The price of hydrogen is expected to increase as the scale of hydrogen plants becomes smaller. Therefore, hydroelectric and geothermal energies, and biomass are not likely competitive in industrial and commercial applications in near future. Consequently, practical hydrogen production has been mostly associated with wind and solar PV power generations and thermochemical solar water splitting to date. Typical commercial electrolyser system efficiencies are 56–73% and requires 70.1–53.4 kWh of electricity to generate one kilogram of hydrogen4. The hydrogen cost via electrolysis is dependent on electricity prices. We estimate in consultation with a major car company that the costs of hydrogen via wind and solar PV power generations are approximately 2–3 and 30 € kg-1, respectively. The efficiency of thermochemical water splitting peaks at a certain temperature depending on the degree of sunlight concentration due to a balance between the light absorption efficiency of a reactor and the efficiency of the reaction. The efficiency peak increases with the degree of sunlight concentration from 50% at 100 sun to 80% at 10000 sun5, while increasing the capital cost greatly. In practice, the cost of hydrogen produced by thermochemical solar water splitting is estimated to be approximately 16 € kg-1 by a major car company.
PEC hydrogen production
Solar hydrogen production through photocatalytically assisted water splitting has attracted a great deal of attention since its first discovery almost 40 years ago. The investigations into the use of titanium dioxide, TiO2, photoanodes have continued throughout the years, together with the search of alternative semiconductor materials. A photoanode may be combined in a photo-electrochemical, PEC, cell either with a p-type semiconductor photocathode or with a metallic cathode. The latter configuration was used in the photoelectrolysis cell which first demonstrated the feasibility of PEC water splitting, where an oxygen-evolving TiO2 photoanode was associated with a hydrogen-evolving platinum cathode. Since the band-gap energy of TiO2, (Eg= 3-3.2 eV) precludes any significant solar light absorption, a large number of further studies were directed towards the search of photoelectrode materials that might be able to capture a substantial part of the visible spectrum. Those early investigations, including large variety of inorganic semiconductors, showed that the only materials able actually to photo-oxidize water whilst avoiding photocorrosion were some metal oxides.
The choice of a suitable photoanode material for water splitting imposes an inherent tradeoff between its band-gap energy, which determines its ability to absorb the visible part of the solar spectrum, and its band energetics, which should allow water splitting to proceed with a minimum applied voltage. In fact, only few large band-gap semiconducting oxides, absorbing exclusively UV light, fulfill the requirement of having the potential of the conduction band, CB, edge more negative than the hydrogen evolution potential.
On the other hand, most of the semiconducting oxides, including those able to absorb visible light, exhibit valence band, VB, edge potentials positive enough to allow oxygen evolution. Consequently, the amount of the external voltage bias, required to drive effectively the photoelectrolyser employing a semiconducting oxide photoanode, will primarily depend on its CB edge potential and on the extent of charge recombination within the semiconductor . This determines the photocurrent onset potential and the potential at which the photocurrent attains its maximum (saturation) value. As is discussed below, the minimum voltage that must be applied to reach the saturation photocurrent under sunlight illumination will directly affect the conversion efficiency of the photoelectrolyser.
Titanium dioxide, which served to demonstrate the feasibility of PEC water splitting for the first time, has subsequently been the subject of a large number of studies into the reduction of its band-gap energy to sensitize the oxide to visible portions of the solar spectrum. Those early investigations involving transition metal dopants, such as Cr(III), demonstrated indeed some extension of the photoresponse towards visible wavelengths but also showed that it was accompanied, in all cases, by a significant degradation of the conversion efficiencies within the fundamental absorption range of TiO2 due to enhanced electron–hole (e- - h+) recombination10. More recently, a large number of reports on the “TiO2 band gap narrowing” by doping with non-metallic elements, in particular nitrogen, carbon and sulfur have been published. Among the studies in which such materials were tested as photoanodes for water splitting, those involving carbon, C–modified TiO2, films raised vivid interest. In particular, an initial report of 8.35 % energy conversion efficiency for PEC water splitting obtained using C-modified rutile TiO2 films formed by oxidation of titanium metal in a natural gas flame11, prompted a number of further studies and provoked a lively discussion. Although these films exhibited an optical absorption threshold shifted to 535 nm, corresponding to a band-gap energy of ca. 2.3 eV, the conditions under which the water oxidation photocurrents were measured have been questioned. In fact, the incident photon-to-current efficiency, IPCE, spectra obtained later using similar C-modified TiO2 films15, showed only a moderate shift of the photoresponse towards visible wavelengths consistent with a water splitting photocurrent of ca. 0.6 mA cm-2 determined at a potential of 1.23 V vs. RHE (reversible hydrogen electrode in the same solution) under simulated solar AM 1.5 illumination. In another study, carbon was incorporated into anodically formed TiO2 nanotubes by annealing in a carbon monoxide atmosphere up to 700°C resulting in significant photocurrents, reaching 0.8 mA cm-2 measured under visible light irradiation of 100 mW cm2. The latter work is representative of numerous efforts to improve the visible-light photoresponse of TiO2 by combining doping with structuring the film surface through formation of nanotubes, nanorods or nanowires. In contrast with the aforementioned examples, where the amount of carbon included in TiO2 films exceeded 10 at.% C/Ti ratio, the doping levels in various synthesized nitrogen, N-modified TiO2 samples were of the order of 1 at.% only. In most of those studies the N-modified TiO2 was employed as powder suspensions to promote photodegradation of organic contaminants under visible-light. An attempt to use N-modified TiO2 films formed by reactive magnetron sputtering as photoanodes to oxidize water, resulted in low photocurrents, ca. 0.2 mA cm-2 observed under 100 mW cm-2 visible-light illumination. The poor performance of nitrogen-modified TiO2 materials for water splitting, which is in contrast with significant visible-light absorption extending above 500 nm, is indicative of enhanced e- - h+ recombination facilitated by the high density of electron trap states distributed below the CB.
Beyond the work oriented towards sensitization of TiO2 to visible light, significant efforts were directed over the last decade into improving the PEC performance of other semiconducting oxides allowing efficient capturing of visible light. In this regard there appears to be a broad agreement that an optimal semiconductor to be used for the cleavage of water in a PEC cell should have a band-gap energy close to 2 eV 7,8. This would account, in addition to the thermodynamic water decomposition voltage of 1.23 V, for overpotential (polarization) and Ohmic (resistance) losses. Therefore, considering the band-gap energy alone, ferric oxide, hematite α-Fe2O3, is an ideal candidate material as it has a band gap of 2–2.2 eV, thereby allowing visible light absorption up to 550–600 nm. This leads to a respectable maximum theoretical efficiency for water splitting of 18 %. However, because of the combined effects of a CB edge potential that is too positive and large recombination losses occurring in this material, the photocurrent onset potentials for oxygen evolution at α-Fe2O3 photoanodes are in the best cases (i.e. for doped films) located at rather positive potentials, ca. 0.8 V vs. RHE. The small hole diffusion length in α-Fe2O3, typically of the order of 4 nm, and the poor electric conductivity affect also the shape of the photocurrent-voltage curve to reach 1.23 V vs. RHE, which is normally set as an upper potential limit for the PEC water splitting. To increase the conductivity, doping α-Fe2O3 with Ti (IV) was early proposed, and allows indeed a substantial improvement of IPCEs for the films formed by spray pyrolysis on conductive glass (fluorine-doped SnO2, FTO) substrates, which attained 25% at 400 nm and 1.42 V vs. RHE 20. A major improvement in the PEC performance of thin film α-Fe2O3 photoanodes was brought about by combining atmospheric-pressure chemical vapor deposition (APCVD) with silicon doping21. Those films, which included a thin SiO2 barrier layer pre-formed on the FTO substrate and exhibited pronounced dendritic nanostructures, reached in fact at 1.23 V vs. RHE under AM 1.5 illumination the water oxidation photocurrent of 2.2 mA cm-2 and even larger after adsorption at the Fe2O3 surface of a monolayer of Co2+ ions. The effect of silicon is apparently closely related to the film deposition method. For example, in the case of the α-Fe2O3 photoanodes obtained either by spray pyrolysis or by reactive magnetron sputtering, doping with titanium was more effective at increasing the photocurrents than the doping with silicon. It has been suggested, that besides improving the conductivity of α-Fe2O3 films, the Ti or Si dopants may act by decreasing the extent of charge recombination in the film through passivation of the grain boundaries.
Another semiconducting oxide capable of absorbing a sizeable portion of the visible spectrum is tungsten trioxide, WO3. Although the band-gap energy of WO3 (in the monoclinic form) is larger than that of hematite α-Fe2O3, 2.5 eV compared to 2.2 eV, the photocurrent onset potential of WO3 is less positive, ca. 0.4 V vs. RHE. This reflects the almost generally applicable rule for semiconducting oxides whereby the decrease in band gap is accompanied by a down-shift in the CB edge.
The water oxidation photocurrents reported for both the single-crystal, and polycrystalline film WO3 photoanodes, were adversely affected by the small absorption coefficients for photons with energies close to the band gap. In fact, the optical pathways in WO3 for the visible wavelengths are at least an order of magnitude longer than the hole diffusion length in this material which is of about 0.15 microns22. The charge collection efficiency was significantly improved by the introduction of porous, nanocrystalline film photoanodes consisting of a network of WO3 particles with sizes in the range of tens of nanometers, i.e. smaller than the hole diffusion length. The employed sol-gel synthesis method involving various organic additives allows formation of highly crystalline mesoporous WO3 films which can be largely permeated by the electrolyte. This largely reduces the electron-hole recombination and permitted the optimized WO3 film photoanodes to deliver water splitting photocurrents larger than 2.5 mA cm-2, at 1.23 V vs. RHE under simulated 1.5 AM irradiation. Prolonged photoelectrolysis tests performed in various acidic electrolytes revealed formation of often important amounts of peroxide species causing reversible deactivation of the WO3 photoanodes. It has been shown that the deactivation of the WO3 surface can be avoided by the addition of even small amounts of Cl-, Br- or Co2+ ions to the acidic electrolyte.
Nanostructuring has also been proved to be effective in improving the PEC performance of BiVO4 photoanodes and permitted to reach significant (over 40%) IPCEs for oxygen evolution in the visible spectral range.
NanoPEC: developing new materials and nanostructured electrodes towards efficient water photoelectrolysis
First generation materials
Systematic studies, aimed at optimizing the nanostructure of WO3 photoanodes, involved use of various organic additives to the sol-gel precursor and different annealing protocols. As shown by impedance spectroscopy measurements, high porosity is essential to activate the entire film, providing homogeneous photocurrent distribution across the film. This is particularly important when relatively thick (several micrometers in thickness) films, required to improve visible-light absorption in WO3, are involved. The control of porosity in the course of sequential film deposition allowed, in particular, successful preparation of large scale (10 x 10 cm2) WO3 electrodes for the final NanoPEC tests performed in December 2011. Despite non-optimized PEC cell configuration, such an electrode (90 cm2) exhibited remarkable photocurrents of ca. 1.3 mA cm-2 at 1.23 V vs. RHE under AM 1.5 irradiation. This was also favored by the identification of methane sulphonic acid as the most suitable electrolyte for the PEC water spitting at WO3.
Considerable efforts were also directed towards the search of the most adequate plasmonic metallic nanostructures to be implemented within the mesoporous WO3 films to enhance the PEC reactions. An important enhancement of the water splitting photocurrents was indeed made possible by incorporation plasmonic silver nanoparticles (Ag NPs) within WO3 thin film photoanodes. The choice of the embedded configuration allowed to preserve Ag NPs, with sizes of 10-30 nm deposited on FTO from the contact with the electrolyte, allowing repetitive PEC experiments in acidic electrolytes. For a 1 μm thick WO3 photoanode featuring an Ag NP under-layer of a nominal thickness of 4 nm, a saturation water oxidation photocurrent as large as 2.1 mA cm-2 was observed under AM 1.5 irradiation through the substrate (uncorrected for the absorption losses within the substrate). This corresponds to more than 60% increase (by 0.8 mA cm-2) in comparison to the WO3 photoanode of a similar thickness deposited directly on FTO. Most importantly, the photocurrent at the composite Ag/WO3 photoanode attains already 2 mA cm-2 at 0.8 V vs. RHE to reach a plateau at a potential of only 0.9 V vs. RHE. This makes such composite photoanodes particularly promising in view of the application in tandem cell configuration.
Hydrothermal post-treatment of metal oxides turned out to be a promising processing step towards more efficient water electrolysis. Starting with a hematite Np synthesis route, we obtained solid precipitates and supernatant liquid. The solid precipitates formed to nano particles upon thermal treatment. The supernatant liquid was used for dip coating on FTO substrates and yielded nanostructured pristine α-Fe2O3 thin films after annealing at 500°C with a photo current of 150 µA cm-2.
In a further hydrothermal processing step, we could refine the microstructure of the pristine hematite films to an extent that the photocurrent was doubled. The microstructure of the pristine hematite films is subject to dramatic microstructural changes upon hydrothermal processing. In particular, we could form stellate structures that grow from a turf like seed structure. The nano-flowers have a regular 3-dimensional structure with inclined wings reminiscent of the wind rose (NATO symbol) with inclined wings when viewed in the transversal plane, which we demonstrate by direct overlapping in third image. From the functional point of view, this nano structuring increases the photocurrent from 150 to 250 µA cm-2.
Extending the aforementioned self-assembly based nanostructuring, we explored the structuring of electrode architectures on the micron scale with electric field-assisted and dewetting-assisted pattern formation.
before NanoPEC the state-of-the-art photoanode was hematite doped with silicon, which was prepared using a particle-assisted APCVD system resulting in 500 nm thick films with dendritic cauliflower-like structure. The small features size of those mesoporous film was on the order of the hole diffusion length (ca. 5 nm), enabling efficient hole collection at the surface. Additionally, efficient electron transport to the back contact was allowed owing to its high dopant concentration.
The NanoPEC consortium improved the performance of the state-of-the-art hematite photoanodes further by controlling the APCVD process parameters and using a highly-active water oxidation catalyst. Improved alignment of the hematite grains facilitated more efficient electron transport, improving the electron conductivity through the film. This enabled thicker films (700 nm) to be effectively utilized than in the previous work (500 nm), improving light absorption, increasing the surface area and facilitating more effectively charge carrier extraction. The kinetics of the water oxidation reaction was improved by incorporating IrO2 NPs 2 nm in size at the hematite/electrolyte interface. IrO2 NPs were successfully deposited by passing a current of 40 mA cm-2 between a hematite anode and a Pt mesh cathode which were separated by approximately 1 mm. It is noteworthy that the incident light was not screened by IrO2 NPs owing to the small particle size. Consequently, a new photocurrent record of 3.3 mA cm-2 was achieved, surpassing the old record by more than 50%. The new record corresponds to a STH conversion efficiency of 5% in an ideal tandem configuration. However, the thick stems of the cauliflower structure still prevents a large proportion of the absorb light to participate in the water splitting reaction. This necessitates alternative nanostructuring strategies such as host-guest nanocomposite structures, new synthetic routes of nanostructures with added functionality, plasmon-enhanced light absorption, semiconductors with graded compositions, and overlayer and underlayer modifications, in addition to the development of new materials and catalysts, in order to optimize the performance of the 2nd generation materials and nanostructures.
Second generation materials
In addition to the development of various approaches to enhance the performance of first-generation materials such as Fe2O3 and WO3 (nanostructuring, catalysis, light management), the development of novel photoanode and photocathode materials was an important part of the consortium’s efforts. A total of 12 new materials have been synthesized and subjected to initial performance and stability screening: Cu2O, CuAlO2, CuCrO2, SrTi1-xFexO3-delta, BiFeO3 (photocathodes) and BiVO4, TaON, LaTiO2N, Ta3N5, and MNbOxNy with M=La, Nd, Sm (photocathodes). An overview of the results achieved is shown in Table 1. Four of these materials have met NanoPEC’s stringent requirements for continued efforts, and have been designated as “second generation” materials: Cu2O, BiVO4, and the (oxy)nitrides LaTiO2N and Ta3N5.
Cu2O
Cuprous oxide is a p-type semiconductor that seems an ideal candidate for water splitting applications: it has a band gap of 2.0 eV and it is one of the very few oxides that has high enough CB edge to reduce water to hydrogen. Earlier studies on this material have met with limited success due to excessive photodegradation, in which the Cu+ ions are reduced to their metallic state by the photo-excited electrons: Cu2O(s) + H2O(l) + 2e- -> 2Cu0(s) + 2 OH-(aq). NanoPEC partners have solved this issue by depositing 20 nm thick protective overlayers of Al-doped ZnO and TiO2. In addition, Pt NPs were deposited at the surface to catalyze the hydrogen evolution reaction. This has resulted in photocurrents of -7.6 mA cm-2 at 0 V vs. RHE under illumination with simulated sunlight (AM1.5) which is the highest value ever recorded for any metal oxide photoelectrode. The stability was improved further by optimizing the deposition parameters of the TiO2 overlayer and by choosing a phosphate buffer solution as the electrolyte.
BiVO4
With a band gap of 2.4 eV, BiVO4 can theoretically convert up to ~9% of the incident solar energy into hydrogen. NanoPEC researchers have used spray pyrolysis as a low-cost and easily scalable technique to deposit thin dense films of this material onto FTO-coated substrates. Initial studies revealed that poor electron transport and slow water oxidation kinetics limit the performance of this material38. To address these issues, three improvement strategies were pursued. First, the spray deposition process was optimized in order to obtain a favorable film morphology. Second, a cobalt-phosphate oxygen evolution catalyst was deposited onto the BiVO4 films by electrodeposition. Finally, 0.5% W (tungsten) was incorporated in the film as a dopant to enhance the electronic conductivity. These combined measures increased the AM1.5 photocurrent from 0.2 to 2.2 mA cm-2, which is currently the record for Co-Pi-catalyzed BiVO4. Moreover, the absorbed-photon-to-current efficiency (APCE) is nearly 100% at low light intensities, indicating that defect-induced recombination in the bulk of the BiVO4 is virtually absent. This is unusual for an oxide semiconductor synthesized with a low-cost, low-temperature process and suggests that higher efficiencies are within reach.
(Oxy)nitrides: LaTiO2N and Ta3N5
Oxynitrides are promising photoelectrode candidates because their band gap is generally smaller than of corresponding oxides, while still being reasonably stable. LaTiO2N has a nearly ideal band gap of 2.0 eV, but the performance is limited by the presence of defects – presumably Ti3+ centers. NanoPEC researchers have developed a successful strategy for suppressing the formation of Ti3+ in LaTiO2N by substitutional doping of Ca2+ on La3+ sites. This was achieved by treating A-site deficient La1-xTiOy with Ca(NO3)2 before the ammonolysis step. As a result of this so-called “back-filling” strategy, the oxygen evolution efficiency of the resulting powders is ~2 times higher than that of the previous generation of LaTiO2N powders.
Efforts to synthesize phase-pure beta-TaON have resulted in a new optical monitoring technique that allowed NanoPEC partners to follow the oxide-to-nitride conversion in-situ, at temperatures of up to 800°C and with a time resolution of ~1 sec. These experiments revealed that conversion of Ta2O5 to Ta3N5 proceeded via the N-doped Ta2O5 phase, without any indications of the formation of -TaON. The resulting Ta3N5 films showed AM1.5 photocurrents of up to 1.2 mA cm-2 when catalyzed with IrO2. Moreover, resonant light trapping in these films resulted in significant improvement of the quantum efficiency of these films. The stability of these films should, however, be improved.
Fundamental understanding of underlying processes and novel investigation technique
Fundamental investigations aiming at understanding the physical and electrochemical processes underlying the operation of selected photoelectrode materials were carried in order to identify the performance limiting factors and suggest new routes to overcome these limitations in order to improve performance. Detailed investigations were carried out on selected materials that we identified as promising photoelectrode candidates, namely -Fe2O3, WO3, Cu2O, BiVO4, TaON and La(Ti,Nb)(O,N)3. Complementary experimental techniques were combined with advanced analysis methods, detailed theoretical modeling and ab-initio calculations in order to reveal the effect of microstructure, interfaces and grain boundaries, chemical composition and defects on the electronic structure and PEC properties of these electrodes.
NanoPEC partners developed a powerful analysis method for resolving the different loss mechanisms in the water photoelectrolysis reaction and quantifying their contributions to the overall performance. The water photoelectrolysis reaction is a cascade of three processes: (1) light absorption and carrier (electron-hole pair) generation; (2) carrier separation and collection at the counter interfaces; and (3) selective carrier injection at the respective interfaces. The overall STH conversion efficiency is the product of the efficiencies of these processes. While the light absorption efficiency can be readily quantified using optical measurements, the other two processes (charge separation and injection) are difficult to deconvolute and their respective contributions to the overall efficiency has been unknown. To rectify this we developed a new analysis method that is simple, robust and very effective and informative. Using H2O2 as hole scavenger that sets the injection efficiency of holes to the electrolyte to 100%, we compare the photocurrent obtained for the same photoanode under the same conditions in aqueous solutions with and without H2O2. The quotient of the photocurrent measured without H2O2 divided by the photocurrent measured with H2O2 in the electrolyte gives the injection efficiency, which is the probability of holes reaching the photoanode/electrolyte interface to drive the water oxidation reaction rather than being lost for surface recombination. Thus, the quotient of the overall efficiency divided by the optical and injection efficiencies yields the charge separation and collection efficiency. We applied this method to quantify and analyze the different loss mechanisms in state-of-the-art nanostructured α-Fe2O3 photoanodes prepared by APCVD and ultrasonic spray pyrolysis, as well as BiVO4 photoanodes prepared by spray deposition. In the former case we found that at sufficiently high potentials the injection efficiency of holes is high, exceeding 90%, while the separation and collection efficiency remains low (< 20%), indicating that bulk recombination is the dominant loss mechanism of these electrodes. By analyzing the potential dependence of the charge separation and collection efficiency we found that most of the holes arriving at the surface to oxidize water are collected from the space charge (depletion) region, in agreement with other studies reporting extremely short diffusion length of holes in α-Fe2O3. In case of the BiVO4 photoanodes we found that the charge separation efficiency can reach values close to 100%. The injection efficiency, however, was less than 40%, which shows that hole transfer across the semiconductor/electrolyte interface is the limiting process. This was solved by using a cobalt phosphate catalyst, which increased the injection efficiency to more than 90%. In order to overcome the tradeoff between light harvesting and charge separation and collection efficiency, NanoPEC partners developed an innovative approach for light trapping in ultrathin films. This approach enables trapping most of the light in films as thin as ~20 nm, for which the charge separation and collection efficiency is intrinsically high. Theoretical calculations for Fe2O3 photoanodes comprising ultrathin films on reflective substrates suggest that photocurrent densities approaching 8 mA cm-2 can be reached with an optimal cell design. This innovative technology is being patented and efforts to achieve high performance are undergoing. To understand the microstructure – properties correlation in nanostructured photoelectrodes for water splitting, we carefully analyzed the microstructure and chemical composition of our electrodes using a wide array of characterization techniques scaling from the macroscopic specimen dimensions down to the microscopic nanometer level, which is the scale of interfaces. These multiscale characterizations have led to new insights on the effect of microstructure on PEC properties of nanostructured electrodes. In particular, we found a remarkable anti-correlation between the photocurrent and missorientation angles between domains of highly oriented nanograins in state-of-the-art nanocrystalline α-Fe2O3 photoanodes. Changing the deposition rate from 2 to 6 L min-1 and the substrate temperature from 415 to 425 C led to a twofold increase in the photocurrent density, reaching more than 4 mA cm-2 at 1.4 V vs. RHE. Standard investigations of the microstructure of these electrodes using conventional scanning and transmission electron microscopy methods revealed no clues as for the origin of the remarkable improvement in performance; the two electrodes look the same. The mystery was resolved by careful analysis of the missorientation angles of nanosized grains and sub-micron domains in these nanocrystalline mosaics. Towards this end we developed a multiscale orientation mapping method that involves composite imaging of dark field transmission electron microscopy micrographs obtained at different zone axes. The orientation maps of the champion electrode display mosaic structures (“cauliflowers”) of unmixed orientations, for the most part while its sister electrode displays mixed orientations in most of the cauliflowers. The anti-correlation between the photocurrent and missorientation angles of these electrodes indicates that high-angle grain boundaries are detrimental to the performance of nanocrystalline α-Fe2O3 photoanodes.
Beside the grain boundary effect we also found interesting correlations between the electrode microstructure and the charge transfer processes at the electrode/electrolyte interface. These effects are particularly important in mesoporous electrodes of high surface area, such as our WO3 photoanodes. To understand these effects we combined the conventional I-V measurements together with careful impedance spectroscopy analysis in electrolytes of different conductivities. The photocurrent plateau was found to increase with increasing electrolyte conductivities, up to a certain limit at an electrolyte conductivity of about 200 mS cm-1. Using impedance spectroscopy analysis to resolve the different contributions to the overall impedance, the dominant contribution was found to the charge transfer resistance at the electrode/electrolyte interface. The latter increases with decreasing electrolyte conductivities, which is the primary reason for the degraded performance of mesoporous WO3 photoanodes in non-concentrated electrolytes (0.5 M in the case of sulphonic acid electrolytes). This result is important for getting the most out these electrodes.
Taking another step further beyond the effect of microstructure, the effect of atomic defects such as vacancies on the electronic structure and the effect of electronic defects such as trapped charge states induced by illumination or electrochemical pretreatment on the PEC properties of WO3 and α-Fe2O3 photoanodes were investigated using near-edge X-ray absorption fine structure (NEXAFS) spectroscopy.
Distinct surface and sub-surface states are formed on hematite upon anodic polarization. Anodic polarization at 600 mV vs. Ag/AgCl reference electrode (i.e. anode PEC operation) gave rise to a new electronic transition that is the signature of the upper Hubbard band of the oxygen NEXAFS spectrum. This is associated with a high oxidation state of Fe species at the surface. Anodic polarization at 200 mV vs. Ag/AgCl is not sufficient to induce this transition. However, anodic polarization at 0, 200 and 600 mV vs. Ag/AgCl caused the Fe-resonant X-ray photoelectron spectra (XPS) to shift systematically towards the Fermi level, indicative of the hole doping upon anodic polarization.
Transient surface states were observed in a spectacular experiment by taking advantage of recent progress in synchrotron radiation instrumentation which enabled us to investigate a working PEC system operando with soft x-ray spectroscopy under electrochemical polarization and illumination with visible light. 30 nm thin α-Fe2O3 films were deposited on Si3N4 substrates fitted with a 100 nm thick window that allows soft X-rays to access the electrode/electrolyte interface. These tiny micro-cells were assembled on an electrochemical micro-cell specifically designed for in-situ investigations at the Advanced Light Source in Berkeley USA.
X-ray spectra were recorded at the oxygen K-edge, which is very sensitive to the local bonding and symmetry properties of the oxygen ions. The O 1s spectra of α-Fe2O3 have a well-developed doublet in the pre-edge region at 528 to 530 eV, originating from transitions of hybridized Fe(3d)-O(2p) states with t2g and eg orbital symmetry. We notice clearly two new spectral features when the film is illuminated and biased between 300 to 700 mV with respect to the counter electrode. The energy position of the two new peaks relative to the absorption edge reveals that the lower energy peak and the higher energy peak at 525.5 and 527 eV are electron hole transitions with t1u and a1g symmetry into the charge transfer band and into the upper Hubbard band, respectively.
The variation of the relative spectral weight of these new transitions in comparison with the photocurrent versus bias shows that both types of holes are active for the water photoelectrolysis reaction. Thus, we have directly identified two different electron hole states in the VB, which, contrary to published literature, are both active for water photoelectrolysis.
In addition to advanced experimental techniques we also used detailed physical modeling and ab-initio calculations to achieve deeper and broader understanding of the underlying physics and to reveal the atomistic origin of the electronic structure of the photoelectrodes and their PEC properties.
Numerical calculations based on finite element method were used to simulate the electric field intensity around Au/SiO2 core/shell NPs of interest for plasmonic effects. Simulations for 100/17 nm Au/SiO2 core/shell NPs assembled in the form of an ordered array on top of a 50 nm thick α-Fe2O3 film on a 350 nm thick FTO layer on a glass substrate reveal the important role of the interparticle spacing on the distribution of the electric field intensity around the NP. At a wavelength of 610 nm, which is close to the absorption edge of α-Fe2O3, the electric field is scattered into the substrate or the α-Fe2O3 film for interparticle spacing of 250 or 150 nm, respectively. Therefore, it is important to design not only the size of the plasmonic NPs but also their arrangement.
Ab-initio density functional theory (DFT) calculations were carried to understand the electronic structure of selected photoelectrodes materials (α-Fe2O3, WO3, BiVO4, SrTi0.5Fe0.5O3 and TaON) and to model the electronic interaction between α-Fe2O3 and the adsorbed water layer at the surface. A new calculation method was developed to achieve more accurate calculations that the available methods by taking into account different populations of electrons and holes in different electronic states in the VB and CB. The band gap energy obtained by this method (2.28 eV) is very close to the experimental values reported for α-Fe2O2 (2.1 – 2.2 eV). Calculations for a slab of α-Fe2O3 covered with two monolayers of water reveal the presence of a potential barrier for hole transfer from the sub-surface region to the outermost atomic layer where charge transfer to the electrolyte takes place. The barrier increases upon surface oxidation, reaching about 900 mV wherein 300 mV falls at the sub-surface region and the rest (600 mV) is the injection barrier from surface to the electrolyte. These results are consistent with the typical overpotential values of α-Fe2O3 photoanodes. The origin of the potential barrier can now be traced to the electronic structure at the α-Fe2O3/water interface. Our calculations show that the oxidized surface of α-Fe2O3 is enriched with holes that are trapped at the outermost oxygen sites in contact with water. These self-trapped holes mitigate hole transfer to the electrolyte, stabilizing the adsorbed water and impeding the water oxidation reaction. This is illustrated in the spin density maps of an isolated Fe(HO)3 complex, that is a complex of an hydrated Fe(III) with two water molecules, shown in Figure 16 before (right) and after (left) oxidation. When this complex is oxidized to FeO(HO)2 by trapping a hole and releasing a proton, the trapped hole is stabilized at the O-1 surface oxygen site, mitigating its transfer to the coordinated water layer.
Last but not least, the stability of semiconductor photoelectrodes in aqueous solution – an import issue that is often neglected by the PEC community – was carefully analyzed using electrochemical equilibria considerations. The Pourbaix diagrams52 were found to be an effective and easy to use tool for evaluating the stability of binary oxides such as TiO2, Fe2O3, WO3 and Cu2O. This approach can be extended to evaluate the stability of complex compositions (ternary and quaternary oxides and oxynitrides) using thermodynamic databases.
Novel nanostructuring approaches
Host-guest mesostructure approaches employ extremely-thin semiconductor films deposited on a nanostructured collector with a large specific surface area, in analogy to dye sensitized solar cells (DSCs). In this architecture, photoexcited carriers are generated within the range of charge transport lengths and most of them can access the surface. Although incident light could not be absorbed by a single ultrathin film, sufficient photoresponse should be obtainable by coating ultrathin films conformally on a suitable conductor with a large specific surface area. We found that this approach actually increased the photocurrent and APCE of ultrathin hematite photoanodes using a nanostructured scaffold made from WO3.
WO3 host layers and Si-doped hematite guest absorbers were deposited by APCVD successively. WO3 host layers created a porous structure with increased roughness over the underlying substrate. The WO3 scaffold was completely covered with ca. 10 nm particles of silicon doped hematite after the deposition of the guest absorbers. A significant increase in the photocurrent plateau was observed for the WO3/Fe2O3 photoanode over the control electrode prepared without the WO3 scaffold as shown in Figure 18e. We confirmed that the APCE was increased over all wavelengths below the absorption edge of Fe2O3. The enhancement was particularly significant near the absorption edge, because photoexcitation occurred near the semiconductor-liquid junction (SCLJ) at the surface even for the large photon penetration lengths at these wavelengths.
The benefit of the guest-host nanocomposite design is most important for ultrathin absorber layers. However, extremely-thin hematite films suffer from an inactive “dead” layer near the interface with the substrate, likely due to deleterious electronic interactions (interface recombination) and structural defects (uncoated regions). To improve interfacial properties, ultrathin interfacial oxides (< 3 nm typically) were developed as a buffer layer. We found that a buffer layer of SiOx, Ga2O3, Nb2O5, or TiO2 improved the performance of ultrathin hematite films in PEC water oxidation. The function of the buffer layers is attributable to enhancement of crystallinity and conformal coating of the ultrathin hematite layers and improvement in the photovoltage and charge separation as a result of surface acid-base interaction, crystallographic template effect, and/or suppression of electron back injection from the FTO to the hematite film 57, depending on the kinds of buffer layers. The ultrathin hematite electrodes modified with Nb2O5 underlayers demonstrated the highest APCE ever reported for ultrathin hematite films. Additionally, a Ga2O3 buffer layer was successfully combined with a rough conducting substrate made of antimony-doped SnO2 (ATO) NPs, which improved the photocurrent by a factor of 1.4 while maintaining the positive effect of the Ga2O3 buffer layer on the photocurrent onset. These results clearly demonstrate the potential of buffer layers to boost the performance of guest-host nanocomposite electrodes.
Functional overlayers were developed for ultrathin and cauliflower-type hematite photoanodes in order to further improve the performance of guest-host electrode designs. Al2O3, Ga2O3, and In2O3 overlayers deposited on ultrathin hematite films improved the photocurrent onset by up to 200 mV. Similar improvement in the photocurrent onset was observed for cauliflower-type hematite films modified with an Al2O3 overlayer. It was concluded that corundum-type overlayers released lattice strain of ultrathin hematite films and/or decreased a density of surface states. Importantly, stoichiometric water splitting for longer than one day was confirmed on Ga2O3-deposited ultrathin hematite using a newly-developed gas detection system, attesting the excellent stability of the system.
Porous colloid-based films with feature diameters of ca. 20 nm are attractive morphologies owing to their potential to achieve efficient light absorption and carrier collection. High temperature annealing was usually necessary to make colloid-based hematite films photoactive, which inevitably led to sintering of NPs and deterioration of the original mesoporosity. To prevent coarsening, a silica layer was coated on the hematite films before the thermal treatment As a result, the feature size of the hematite particle was unchanged upon annealing at 800°C when a SiO2 scaffold was applied. After removal of the silica scaffold, the films demonstrated an increased photocurrent plateau, but also an anodic shift in the onset potential most likely due to surface traps generated during the silica scaffold removal. It was found that the photocurrent was enhanced by applying an Al2O3 overlayer, yielding a photocurrent close to 4 mA cm-2 with a sufficient external voltage. Although the photocurrent onset has yet to be improved by healing surface traps, this result highlights the potential of mesoporous hematite films for PEC water splitting.
Mesoporosity was also introduced to WO3 photoanodes using a sol-gel process which involved poly(ethylene glycol) (PEG) as a structuring agent. The amount of carbon played the dominant role in film nanostructuring. The best films was obtained by combining polymers with the lowest molecular weight and an ultrasonic precursor pretreatment, improving the photocurrents exceeding 3 mA cm-2 at 1.23 V vs. RHE. Such films were characterized by porosity of ca. 50% and individual particle sizes of 30-50 nm, which was smaller than the hole diffusion length in WO3. An additional increase in the current density was obtained with samples containing boron, with an optimal content of 10 at .%.
PEC cells and measurement protocols. An experimental, computer controlled test bench equipped with a solar simulator (AM 1.5 G, 1000 W m-2) was assembled, allowing photocurrent-voltage, electrochemical impedance spectroscopy and stability characterization of small (~ 1 cm2) and large (~ 100 cm2) PEC devices. To evaluate the performance of large PEC devices, another test bench with a larger solar simulator equipped with a horizontal light beam was designed.
The efficiency calculations for large electrodes in the reports were reviewed and shown to be accurate. Concerning the above-mentioned new photoreactor, the authors assume that the reviewers are referencing the newly designed PEC cell presented during the last meeting of the project by partner University of Porto (Adelio Mendes). However, the prototype design presented is not part of the final NanoPEC report and was only presented to show how the project’s results would evolve in the near future. Since this presentation, the new prototype design was much improved and is only now being fabricated. After being fabricated, it will be assessed (including efficiency calculations) and, most probably, improved again and only after being patented it will be diclosed and disseminated. Photoelectrochemical cell setup - Small Prototype
The first characterization experiments of small photoanode samples (up to 1 cm2) performed in the framework of NanoPEC project were conducted in a Cappuccino cell provided by EPFL60. However, a new PEC cell was designed at FEUP allowing the performance evaluation of different photoelectrode samples with an active area from 1 to 9 cm2. This new testing device was called Portocell and it was used as the small prototype for PEC cells. The Portocell PEC cell has two removable windows (front and back) made of black acrylic both screwed to a transparent acrylic part, which fixes the synthetic quartz window by pressing them against an O-ring using five screws. These windows may be changed according to the illumination area desired. It is also possible to close the PEC cell by screwing an acrylic cap with only open spaces for O2 and H2 evolution.
In this configuration the photoelectrode is kept vertically aligned and facing perpendicularly to the light beam. A standard three-electrode configuration is used with a 99.9 % pure platinum wire (Alfa Aeasar®, Germany) as a counter electrode and an Ag/AgCl/Sat. KCl as a reference electrode (Metrohm, Switzerland). The electrodes should be immersed in the appropriate electrolyte solution, depending on the semiconductor material.
Photoelectrochemical cell setup - Large Prototype
In order to test large photoelectrodes samples with an active area up to 10 x 10 cm2, the Portocell PEC cell was redesiged. The large Portocell PEC cell works as a single and compact vessel made of transparent acrylic. This cell has two removable windows (front and back) made of black acrylic, allowing an illumination area of 100 cm2, both screwed to the vessel. Moreover, a stainless steel part was used to fix the synthetic quartz window to the vessel, which is pressed against an O-ring by means of twelve screws. This cell allows using a Teflon diaphragm between both electrodes, avoiding hydrogen and oxygen gas bubble to mix. After assembling all the previously mentioned parts, two different caps are used to cover the vessel. The internal cap is used to fix the electrical contacts allowing to establish a correct connection between the electrodes and the potentiostat; the external cap closes the cell allowing to collect oxygen and hydrogen separately.
Performance and stability results with best materials
All the investigations performed to develop new materials for both photoanode and photocathode culminated in the construction of a stable device architecture that achieves the high targets of conversion efficiency proposed. Two prototypes were developed: a small device (up to 1 cm2) achieving a 7 mA cm-2 photocurrent under 1 sun illumination, corresponding to a 10 % STH efficiency; and a large device (~ 100 cm2) that achieves 7 % efficiency. α-Fe2O3, WO3 and Cu2O photoelectrodes were the most promising materials developed during NanoPEC project and these were the ones selected to build the prototypes.
The current-voltage (I-V) characteristic curves of the selected materials were obtained by applying an external bias potential to the PEC cell and measuring the generated photocurrent using a ZENNIUM workstation (Zahner Elektrik) - Figure 22. The measurements were performed in the dark and under simulated sunlight and a standard three-electrode configuration was used.
The small hematite photoanode is a titanium doped structure prepared by the solution-based colloidal method. Its water oxidation photocurrent was measured in a 1 M NaOH (pH 13.6) aqueous solution. A photocurrent value of 1.86 mA cm-2 at 1.43 VRHE and a maximum photocurrent of 2.34 mA cm-2 before the onset of the dark current were reached. The maximum photocurrent presented here for small devices represents the highest photocurrent ever reported for a solution-processed hematite film under standard solar illumination conditions and is furthermore near the values reported with the previously-reported silicon doped films prepared by APCVD. The hematite photoanode scaled up to 100 cm2 was prepared by spray pyrolysis combined with a SiOx buffer layer and it reached a photocurrent value of 0.38 mA cm-2 at 1.43 VRHE.
Concerning the WO3 photoanodes, a series of organic additives to the sol-gel precursor, acting as structure directing agents, were tested to optimise the photocurrent-voltage behaviour. In particular, various ratios of different molecular weight polyethylene glycols (PEGs 200, 300, 600 and 1000) were added to the freshly formed aqueous polytungstic acid solution and thermally decomposed in one or two steps. This allowed to largely modify the porosity and, subsequently, the current distribution within the films. The measurements of tungsten trioxide samples (small and large prototype) were conducted in a 3 M CH3HSO3 (pH 0.26) electrolyte solution. Figure 26 presents the I-V characteristic curves for WO3 small photoanode (~1 cm2) deposited on FTO glass substrate and for WO3 photoanode with ~100 cm2 of active area deposited on a tungsten substrate by the sol-gel method.
WO3 small size photoanode reached a photocurrent of 2.92 mA cm 2 at approximately 1.23 V vs. RHE. After the scale up of these photoanodes, a photocurrent density of 1.22 mA cm-2 was obtained at the same potential. The observed shape of the I-V plot is affected by ohmic drops within pores and unfavorable current distribution over the electrode surface.
Finally, in what concerns Cu2O photocathodes modified with the ultrathin oxide overlayers and Pt catalyst, the current–voltage curves were measured in a 1.0 M Na2SO4 aqueous solution buffered at pH 5 with potassium phosphate (0.1 M) - Figure 27(b). A photocurrent density of 7.6 mA cm-2 for the small devices was achieved, corresponding to 10 % STH efficiency. This efficiency value meets one of the project targets. The enhancement of the photocurrent was attributed to formation of a p-n heterojunction between Cu2O and Al:ZnO layers which could assist charge separation. For an active area of 63 cm2 the photocurrent onset at approximately 0.4 V vs. RHE is similar to the small electrode. However, the increase of the photocurrent during the potential sweep toward negative potentials was not steep, which was indicative of a higher resistance of the electrode due to the thicker TiO2 overlayer. Still, a current density of ~ 4.0 mA cm-2 at 0 V vs. RHE was obtained, corresponding to a STH efficiency of 6 %. This value is slightly below the project target of 7 % STH efficiency. Moreover, an optimized photoelectrode sample of Cu2O may allow to produce solar water splitting devices with a very high efficiency. The proposed 100 cm2 photoelectrode active area was not obtained due to equipment size limitations, though the scale up of this photoelectrode is direct and simple.
Stability tests
In PEC cells stability issues is one of the major problems to be solved. Usually, when a semiconductor electrode is brought into contact with an electrolyte solution some reactions may occur, for instance ionic oxidation or reduction of the semiconductor with simultaneous reduction or oxidation of a component. The electrolytic reduction of a semiconductor is often associated with the electrons in the VB, while the electrolytic oxidation reaction is related to holes in the CB as electronic reactants. Taking this into account, the stability/corrosion tests were performed to identify possible degradation pathways in the semiconductors materials. The overall stability is controlled by two factors, namely physical and chemical factors. Physical stability is related to electrolyte evaporation, which heats up under the solar simulator illumination. The chemical stability is associated to irreversible electrochemical and thermal degradation of the semiconductor and electrolyte64. The corrosion phenomena in the electrolyte were investigated by two techniques: inductively couple plasma (ICP) and atomic absorption. Modifications on the photoelectrodes surfaces induced by the stability tests were analyzed by high resolution scanning electron microscopy. In order to reach medium- and long-term stability targets, a set of stability measurements was performed.
Medium-term stability testing
In what concerns medium-term stability tests, the measurements were performed over 72 hours, under constant irradiation conditions (AM 1.5 G, ~1000 W•m-2, 25 °C). Ti-doped α-Fe2O3 photoanode prepared by the solution-based colloidal method was very stable, exhibiting a constant photocurrent density response of approximately 0.82 mA cm-2, at a constant potential of 1.45 VRHE, over the entire 72 h. The hematite film was characterized in 1.0 M NaOH electrolyte (pH 13.6). After the stability test, the photoanode did not exhibit any signs of degradation, suggesting that no major corrosion phenomena took place during the stability test. SEM images before and after the stability test show no significant differences between the samples analyzed before and after stability tests. Additionally, in what concerns hematite particles, both exhibited similar morphological structure; the sample after stability showed larger particles (30–70 nm). Moreover, the presence of iron in the electrolyte solutions was analyzed by ICP technique to evaluate the possible dissolution of iron. The electrolyte volume tested was ca. 50 cm3. The ICP results revealed only trace amounts of this compound in the electrolyte used during the stability tests, with values lower than 100 µg L-1, proving the stability of this photoanode.
A WO3 film with a thickness of ca. 2 µm deposited on a conducting glass substrate using a PEG 300 precursor was also submitted to stability tests. The WO3 photoanode was immersed in 1 M MSA (pH 0.15).
This sample was submitted to two periods of 72 hours at 1.20 VRHE and between the two testing periods the cell was filled with fresh electrolyte solution. The photocurrent density decreases over time for a period of about 10 hours, starting to increase afterwards. At the end of the first stability test period, WO3 photoanode exhibited a photocurrent density of ca. 0.59 mA cm-2 and at the end of the second period exhibited 0.47 mA cm-2. Here, it is visible that this sample lost just 5 % during the first 72 hours. Analyzing SEM images it can be concluded that WO3 sample maintains mostly the initial morphological structure, although more compact.
The presence of tungsten in the electrolyte solution was obtained by ICP and reveals trace amounts of this compound after the stability test (two periods). The electrolyte volume tested was ca. 50 cm3 and it showed a concentration of 0.150 mg L-1. The electrolyte solution revealed also the presence of sodium, potassium and manganese.
Long-term stability testing
In order to evaluate the major drawbacks of PEC devices when applied to long-term stability tests, the test bench assembled was adapted. This study allows achieving a more complete understanding of the device performance under real applications. To perform high-temperature measurements up to 70 °C, the Portocell was also adapted by replacing the back-side of the cell with a silicone rubber heater . The electrolyte temperature was controlled using a water bath and, inside the Portocell, with a thermocouple. The electrolyte was then continuously pumped in and out of the PEC cell by a recirculation system. The pH of the electrolyte was controlled using a pH meter.
This way, long-term stability tests were performed by monitoring aging tests effects at 60°C. A Ti-doped α-Fe2O3 film (prepared by solution-based colloidal method) immersed in 1.0 M NaOH was used to perform the stability analysis at an applied potential of 1.45 VRHE for 72 hours under 1 sun constant illumination. At this potential the fresh sample exhibited a photocurrent density of ca. 0.54 mA cm-2 at 60 °C. A significant drop in the current density was observed in the first hours, subsequently reaching a photocurrent density of approximately 0.13 mA cm-2. Then, in the following time the photocurrent density remains more or less constant. After the stability test, the sample exhibited a rougher surface with a less compact structure. It is likely that the alkaline electrolyte solution could etch the dopant of hematite photoanodes at high temperatures and deteriorate the performance. For iron oxide, we performed accelerated tests for 72 hours, which, with an acceleration factor of about 10, corresponds to 720 hours of stability testing under constant illumination, or the equivalent of 90 eight-hour days. After a decrease in photocurrent during the first few hours, the current was quite stable for the remainder of the accelerated test. Unfortunately, the photocurrents for the hematite electrodes did not scale well on our standard substrates (FTO), and further work on the hematite was devoted to improving charge collection.
As for the Cu2O electrodes, we observed stability issues on the order of one day under non-accelerated conditions, and thus focused on improving the stability rather than measuring unstable electrodes for long periods of time. In fact, since the completion of the project, a deeper understanding of the mechanism of instability has been attained, involving erosion of the catalyst particles from the surface of the photoelectrode (re-platinization recovers the photocurrent). To date, there is no evidence of degradation of the underlying Cu2O material.
As for the lower active area, this is simply an artifact of the size of our ALD machine. Thus, we constructed the mosaic electrode composed of nine smaller Cu2O electrodes. Due to necessary masking of conducting surfaces, this reduced the active area. Cuprous oxide – Attaining the highest STH conversion efficiency
The instability problem of Cu2O photocathodes were solved by covering the photoactive semiconductor with suitable protective layers by atomic layer deposition as described above. The resultant photocathodes exhibited a photocurrent of –7.6 mA cm-2 at 0 V vs. RHE, which successfully fulfilled the photocurrent target of NanoPEC3. However, the photocurrent decreased to 33% of the initial value in 20 min. It was clarified that electrons were trapped in an amorphous TiO2 overlayer. Such reduced Ti species could prevent photoexcitation of Cu2O. In addition, an amorphous TiO2 layer was found to be not stable in certain electrolytes. Therefore, our effort was focused on improvement of the quality of a TiO2 overlayer.
It was found that a heat treatment in air at 200°C for 45 minutes improved the photocurrent stability of Au/Cu2O/Al:ZnO/TiO2 although it also reduces the initial photocurrent to –5.7 mA cm-2. This result suggests that thermal treatments could improve the quality of the overlayers and thus the stability of the electrode. In fact, the TiO2 overlayer exhibited significantly-improved stability when being deposited at relatively higher temperatures. Additionally, TiO2 was found to be stable in a phosphate buffer solution while not in an acetate buffer solution. Consequently, Cu2O electrodes were stabilized by refining deposition conditions of the overlayers and choosing an appropriate electrolyte. The photocurrent was stable over 60 min at least. It was further confirmed that a sample did not show significant deactivation for 12 hours.
Performance comparison with other water splitting and H2 production routes
The production of hydrogen via water splitting using solar energy was first described by Fujishima and Honda in the early 1970s. Currently, water splitting can be accomplished via three general types of devices: i) composite devices - PV cells connected with an electrolyser and or PEC cells; ii) single devices - SCLJ PEC cells; and iii) thermochemical cycles. For the present report only type i) and ii) are considered.
Composite devices
PV cells can power conventional electrolysers to achieve hydrogen production from water using solar energy as the only source of energy for driving the endothermic water splitting reaction. The most developed technology is the solar-powered electrolysis with PV/electrolyser arrangement. Most PV cells are based on silicon, with typical solar to electrical conversion efficiency of 15 %. The electrical to chemical (hydrogen) conversion efficiency of the best electrolysers reach up to 75 % . Thus, combining a commercial a-Si multijunction PV module with an efficiency of 12 % with a water electrolysis unit operating with an energy conversion efficiency of 65 % (input voltage of 1.9 V) results in a STH conversion efficiency of 7.8 % or lower (taking into account resistance and other system losses). In order to obtain higher efficiencies optimized PV and electrolyser technologies should be used. Recently, a STH efficiency of 12.4% was reported for the champion PV and electrolyser system.
Although PV powered electrolysis systems can reach respectively high STH conversion efficiencies in excess of 10%, this technology is expensive because it combines two separate devices that cannot be integrated together, complicating the system and increasing the cost, and high efficiency electrolysers are very expensive because they employ some of the rarest materials on Earth, Pt and Ir. Moreover, combination of PV and electrolysers is less efficient because the photocurrent must be collected, which leads to significant ohmic losses. Furthermore, in standard solar radiation conditions (AM1.5G) four Si-based PV cells connected in series are necessary to generate the required voltage for water splitting. Unfavorable conditions such as partial shading, haze or cloudiness degrade the power output of these cells below the power input requirement of the electrolyser, which may interrupt the electrolysis process unless sufficient margins are taken into account, reducing the average efficiency and increasing system costs.
Single devices
Most photoelectrodes require extra bias from an external power source to split water. Concerning biased systems, there are chemically biased photoelectrolysis cells and electrically biased photoelectrolysis cells in tandem configurations. In the former case, the chemical bias is achieved using two different electrolytes (e.g. acid and basic electrolytes) placed in two separated half-cells with a salt bridge for the mobile ions. This configuration is not self-sufficient, relying not only on solar energy but also on additional chemical input to stabilize the electrolyte solutions66,68. In the tandem configuration, the PV cell is normally characterized by layered stacked or hybrid structures involving several different semiconductor films placed on top of each other in order to provide the necessary voltage for water splitting (typically about 1.8 V). There are important advantages in this configuration over PV powered electrolysis systems. First and foremost, PV powered PEC cells for water splitting can be made more affordable by combining the two systems, PV and PEC, into one monolithic cell. Second, resistance (Ohmic) losses can be reduced, taking advantage of the low current densities of the PEC cells, relative to electrolysers, and the fact that the interconnection between the PV and PEC components can be made very short (on the order of micrometers rather than meters). Third, the PV + PEC combination is a multijunction cell that can, in principle, surpass the efficiency of single junction PV cells. Fourth, a part of the voltage that is required to split water is provided by the photovoltage generated in the photoelectrode of the PEC cell, therefore the voltage provided by the PV cells is smaller compared to the PV + electrolyser configuration. For instance, a single DSC in tandem with a PEC cell can split water. DSC are also less sensitive to non-optimal illumination conditions such as shade, haze and clouds and therefore DSC + PEC tandem cells could operate also in non-optimal conditions, extending the overall operation time of the system and reducing the extra margins needed for the Si + electrolyser arrangement.
Structures of tandem devices can be subdivided into the following configurations: i) PV/PEC; ii) PV/PV68 and; iii) PEC/PEC. The use of PV/PEC systems has an advantage over PV/PV systems because the PEC face (layer) can replace the face conductor grids that partially obscure the PV layer. Consequently, PEC panels are able to replace costly components and improve photon capture of the PV layer. Recently a new multiphoton combination of a PEC cell in tandem with two DSC was proposed. The authors found that the “trivial” tandem architecture (hematite/squaraine dye/black dye) yields the highest water splitting rate. The maximum STH conversion efficiency is estimated to be 1.36 %.
PEC devices with no additional bias represent a prospective pathway for overcome the complexity of biased systems. Unbiased PEC devices comprise single and multiple photo-system arrangements. The possible arrangements of single photo-system are: i) n-type semiconductor photoelectrode and a metal counter-electrode; ii) p-type semiconductor photoelectrode and a metal counter-electrode; iii) monolithic-bipolar system and a layered metal counter-electrode. PEC cells based on a single semiconductor electrolyte junction employ two photons to produce one hydrogen molecule. Materials having a band gap over 3 eV whose CB and VB straddle the redox potentials required for water reduction and oxidation, such as SrTiO3 and TiO2 (anantase), are able in principle to accomplish the water photolysis without assistance from a bias. However, even in this case, a bias is required to reach the plateau of the photocurrent which remains low due to the lack of visible light response. Hence overall STH conversion efficiencies remain well below 1 %.
Multiple photo-systems may result in a high efficiency for water splitting without an additional bias. The possible arrangements of multi photo-systems are: i) n- and p-type semiconductors (acting as photoanodes and photocathodes, respectively), wired or linked by an ohmic contact; and ii) hybrid systems with layered structures involving several different semiconductor films stacked on top of each other. All strategies share the feature of having two semiconductors with different band gaps. This provides a mechanism by which a single electron is photoexcited twice and, correspondingly, generates a larger bias from light. It has been calculated74 that this type of system could realistically achieve a STH conversion efficiency of 21.6 %, well above NanoPEC’s targeted efficiency of 10%.
Various combinations of n-type and p-type semiconductors such as n-TiO2/p-GaP, n-SrTiO3/p-GaP, n-Fe2O3/p -Fe2O3 have been used to eliminate the bias need for water splitting. Because of the low performance of the individual electrodes in these dual-photoelectrode devices, the resulting overall efficiency is low. NanoPEC’s present work on Fe2O3, WO3 and Cu2O as well as other semiconductors suggest the possibility for an efficient and inexpensive embodiment of a dual photoelectrode device. With photocurrents approaching 4 mA cm-2 with these inexpensive electrodes (corresponding to a 6 % efficiency), the prospect of unbiased water splitting becomes realistic. In our present system, good performance is still achieved with the Cu2O electrode held at 0.4 VRHE and with the Fe2O3 and WO3 electrodes held at 1.0 VRHE68. Therefore only a bias of 0.6 V is needed and the bias will be decreased as better catalysts are identified.
Potential Impact:
Significant progress in PEC water splitting was achieved by NanoPEC by the end of the project (December 2011). In particular, small-size Cu2O photoelectrodes (< 1 cm2) exhibited a photocurrent as high as 7.6 mA cm-2 at the reversible hydrogen evolution potential (0 V vs. RHE). This enables a STH conversion efficiency of 10% in an ideal tandem configuration, exceeding the key technology target of our project. Importantly, the inherent instability of Cu2O was overcome by newly-developed protective overlayers. A large electrode of Cu2O (63 cm2) exhibited a photocurrent of 4 mA cm-2, which corresponded to 87% of the target. On the other hand, small-size Fe2O3 showed a photocurrent of 3.3 mA cm-2 at the reversible oxygen evolution potential (1.23 V vs. RHE). This enables a STH efficiency of 5% in an ideal tandem configuration, the highest record ever reported for stable oxide photoanodes. The Fe2O3 photoanodes did not show any signs of performance degradation for at least three days. It is noted, though, that due to difficulties in scaling-up from small to large area Fe2O3 photoanodes the performance of the later was significantly lower than the champion small-size Fe2O3 photoanode.
The photoelectrodes developed by NanoPEC were used for PEC water splitting in different configurations: one using an external bias to assist water photoelectrolysis; and the other with a new multiphoton combination of a PEC cell in tandem with two DSC. The DSC provide the excess potential, on top of the photovoltage produced at the photoelectrode, necessary to split water and produce H2 and O2 using solar energy as the only energy source. An outdoor demonstrator comprising a PEC reactor, two reservoirs and a Ritter flow meter in the backside of the PEC cells was also developed for practical demonstration of our achievements. The PEC cell contains a photoelectrode, a platinized titanium grid counter electrode (Davis-K, USA) and a Nafion membrane for separating the electrodes (Nafion 424). All these components were immersed in an aqueous electrolyte. The outdoor demonstrator worked continuously for three consecutive months, during which its operation was constantly monitored.
The PEC + PV tandem approach for solar water splitting is a viable way for energy storage through the conversion of solar energy to chemical products. A techno-economic analysis based on published data is reported in the following to give a frame of comparison for the estimated PEC present cost. The present scenario for H2 production presents two primary methods: “thermal reforming of natural gas” and “chemical electrolysis”. Steam methane reforming (SMR), used for decades as the main production way, produces hydrogen rich gas that must be subsequently purified. Moreover, this process has to be combined with CO2 capture and sequestration. The cost with existing technologies is 2.5 $/kg and will range between 1.8-2.8 €/kg in 2030-2050 time frame [Fuel Cells & Hydrogen – SETIS.ec.europa.eu 2011]. The use of fossil fuels does not include external costs related to their environmental footprint. Electrolysis is a well-established technology, however on large-scale plants, efficiency is rather limited (35-40%). Mainly small electrolysers are used in distributed generation. In recent years materials improvements resulted in improved efficiency (80-85%). The cost of H2 produced by direct electrolysis is still high (around 10-20 €/kg) but it is expected to get lower over time. A McKinsey report (2011) suggests a European target of 9.9 €/kg for 2015 [Renewable Energy Focus, ITM Power, UK]. The current estimated cost of hydrogen production via electrolysis in Japan is 9.3 €/kg, while it is expected to go down to 7.5 €/kg for 2015-2020 [Fuel Cells & Hydrogen – SETIS.ec.europa.eu 2011]. The European Hydrogen Roadmap HYWAYS (2008), a study developed under the 6th Framework Program and mainly devoted to hydrogen in automotive applications, evaluated the target costs combining socio-techno-economic analyses to evaluate scenarios for future sustainable hydrogen systems. These studies took into account also storage and transformation problems. The results in terms of hydrogen production were, respectively 4 €/kg in 2020 and 3 €/kg in 2030. Including fuel cells results in additional costs of the order of 100 and 50 €/kW, in 2020 and 2030 respectively. Including storage adds another 10 €/kWh in 2020 and 5 €/kWh in 2030. The Independent Review Panel Report (2009) by NREL-USA, titled: “Hydrogen Production Cost Estimate Using Water Electrolysis”, concluded that in USA the electrolysis H2 production costs 3.3 $/kg, plus1.9 $/kg for compression, storage and dispensing, so that the base estimated cost is 5.2 $/kg. This value depends on the electricity cost in the USA (0.053 $/kWh). A detailed analysis for PEC hydrogen production was carried out by a DOE Final Report (2009) titled Technoeconomic Analysis of PEC H2 Production, by B.D. James, G.N. Baum, J. Perez, K.N. Baum. This document illustrates different production systems both made by colloidal suspensions and by PEC arrays, the latter being either planar or concentrated. The system we intend to refer to is Type 3 – PEC fixed planar array tilted towards the sun, using multi-junction cells immersed in an electrolyte reservoir. The following bar chart reports the evaluated production costs in relation to the efficiency, cell cost and lifetime. Actually, this technology refers to high-tech deposition technologies for planar semiconductor material, which is an especially costly way to produce photo-electrodes. Therefore, the following data have to be considered as the present most expensive solutions. This techno-economic analysis shows that there is still an uncertainty bar in the estimation of the solar hydrogen cost as it is a technology in continuous progress. Therefore, in the following, estimated costs for a real PEC solar-hydrogen-generator are reported, by considering the production capability (kg h-1), the lifetime and the material cost for a photo-electrode of 1 m2. Note that the solar generator is active for 8 h per day and it is supposed to last for 20 years.
PEC Solar H2 Generator
H2 production: 6 x10 -4 kg/h à x8x365x20 = 35 kg/20 years
Unitary panel cost: 200-600 €/m2
Production cost: 6-17 €/kg
This value does not include the capture and storage costs.
The productivity of these integrated solar systems based on PEC technology could be devoted to feed small fuel cells (portable and small stationary systems) provided that a capture and storage as an intermediate system will be designed. Other exploitable results
Apart from the achievements directly intended for PEC solar water splitting, other exploitable results and potential new applications in different fields such as solar cells, (photo)electrochemistry, and material science may emerge from the new experimental techniques and nanostructuring approaches developed in this project:
* advanced host-buffer-guest mesostructure to maximize charge collection and minimize interfacial recombination
* isomorphic template thin films for crystallinity improvement
* silica coating to preserve nanostructures during thermal treatments
* plasmonic light harvest to enhance light absorption
* protective overlayers to prevent corrosion
* novel preparation techniques of high quality compact semiconductor films applicable for solar cells and photocatalysis
* in-situ optical transmission monitoring of semiconductors during oxidation/nitridation
* low-light intensity quantum efficiency measurements as a recombination probe for semiconductors
* resonant light trapping method to control light absorption by ultrathin films
* analytical methods for (photo)electrochemistry by the use of redox couples and sacrificial electron/hole scavengers.
* new PEC module configurations for harvesting the sun light
* unsteady state phenomenological modeling to analyze and simulate PEC phenomena. The innovative approach for light trapping in ultrathin films has been mentioned elsewhere in the report and is presently in the patentin process. On top of this result, the use of overlayers deposited by ALD to form p/n junctions is most probably patentable. At the time of the paper published in Nature Materials (available online since May 8th, 2011), the p/n nature of the overlayer was unclear and unstated in that manuscript. A publication was recently submitted (Apr 28th, 2012) in which the p/n nature of the overlayer is definitive. Once this manuscript is published, the right to patent in Europe will be lost. However, in the United States, there is one year to patent after the publication. The only issue is whether or not the Nature Materials paper constitutes a disclosure of the invention, although as statde above, no mention of a p/n junction is made therein. At the time of the writing of the present report, no other patenting procedures on the above results has been initiated yet by the respective consortium entities. The cost of patenting a result such as the ones listed above ranges from 10k€ to 100 k€, depending on the number of countries in which the patent is be valid or on the related translation requirements, for example. A quantitatively precise cost-benefit analysis is rather difficult to provide, since the expected financial return depends on the conditions in which a company would intend to translate the technology onto the market. The contacts between consortium members and potentially interested companies have not yet gone far enough, at the time of the writing of the present report, to provide a definitive cost-benefit assessment.
Diffusion of project results, dissemination and impact
Results of the NanoPEC project have been disseminated at various levels, namely towards the scientific community, industrial companies and the broad public.
NanoPEC partners have published a large number of articles in scientific peer-reviewed publications. The list and references of all published papers is available on the project website. Whenever possible, NanoPEC publications have been posted on the Open Access system at http://infoscience.epfl.ch.
While a detailed evaluation of the impact of the dissemination actions undertaken by the NanoPEC consortium is very difficult to establish, the latter believes that it has been able to bring considerable attention to PEC-based hydrogen production, by presenting the amazing progress accomplished in terms of overall efficiency, along with the advantage of working with widely available, affordable and non-toxic materials.
Scientific community
Over three years, Nano PEC partners has submitted and/or published 39 international peer-reviewed papers with NanoPEC acknowledgement and 19 proceedings papers. Three book chapters were also published based on NanoPEC results.
Moreover, a total of 40 presentations in conferences and workshops were given by NanoPEC consortium partners from 2009 to 2011.
NanoPEC consortium organized two international workshops in order to increase dissemination of the obtained results and to promote exchange with the scientific community:
* ESF-FWF-LFUI Conference on Nanotechnology for Sustainable Energy, held in Obergurgl (Austria), from July 4th to 9th, 2010. Prof. Michael Grätzel (EPFL) was Conference Chair. Several NanoPEC partners have participated and presented their results. 19 invited speakers, 26 short talks, and 63 accepted posters made up the program.
* EMRS-MRS 2011 Bilateral Energy Conference took place in Nice, France from May 9th to 12th, 2011. The workshop was organized as a symposium within the Conference (Symposium T), named “Materials for solar hydrogen via photo-electrochemical production”. The chairmen were NanoPEC consortium members. The main aim of Symposium T was the dissemination of NanoPEC results obtained after two years and a half of activity. Indeed, NanoPEC partners have prepared 18 oral presentations and 4 posters. Another aim of the Symposium T was to collect new ideas from the scientific international community. As a result, a total of 48 oral presentations and 22 posters were made part of the symposium program. The proceedings of Symposium T will be published in a special issue of “Energy Procedia” edited by the well-known scientific publisher Elsevier. 18 papers have been collected. A peer-review process was put in place with the contribution of NanoPEC consortium researchers and members of the scientific committee. The proceedings are expected to be published in the spring of 2012.
Finally, ranging from 2009 to 2011, some researchers of the NanoPEC consortium have actively participated in several meetings of the IEA-HIA Annex 26 program in the USA, as well as in the U.S. DoE PEC Working Group meetings.
Industrial companies
NanoPEC consortium has established an Industrial Review Panel (IRP), which convened one entire morning during the last General Assembly, on December 16th, 2011 on Eni S.p.A. premises in Novara (Instituto Donegani). The goal of the IRP is in particular to provide potential direct outlets of our scientific findings to industry while also providing a feedback mechanism to ensure that we are utilizing synthetic techniques that have good prospects for industrial scale-up.
The companies stemming from Switzerland, France and Italy, that agreed to participate were: Fiat, PSA Peugeot-Citroën, Total, Solaronix, Solvay, Belenos (Swatch Group), Granit Technologies, HySyTech, Hydro2Power and, of course, Eni S.p.A. A representative of the Swiss electric utilities also attended. The discussions of the IRP focused on:
1) an evaluation of PEC technology towards hydrogen production;
2) NanoPEC innovative potential and IP situation;
3) hydrogen future applications and problems.
A demonstration of hydrogen production with NanoPEC-developed devices has been presented to the attendees.
Broad public
ENI Educational Department prepared a brief new movie describing a water splitting experiment carried out in the laboratory of Istituto ENI-Donegani in Novara (Italy). While the movie is in Italian, written comments in English as movie subtitles have been added during summer 2011. The movie will be made available on NanoPEC website and is presently available at http://www.eniscuola.it.
Another important event in terms of the outreach activities took place in the frame of the collaboration established with the Museum of Science and Technology, in Milano (Italy). During two weekends in March and April 2011, the Museum staff organized four disclosing meetings for visitors on solar energy, nanotechnologies and water splitting for solar hydrogen. At these events, two researchers from ENI-Istituto Donegani were available in the Museum to present a demonstrative apparatus for water splitting and to speak about NanoPEC objectives. These initiatives had a large success, with about 50 attendants per day (4 times). An advertisement poster was produced specifically, while a picture gallery is available at http://nanopec.epfl.ch/events/.
Finally, NanoPEC consortium members have been interviewed in various broad-public magazines, as well as on national radio programs. Recently, a large article on NanoPEC appeared in « Projects Magazine », one of Europe's leading science and technology research magazine, covering the latest innovations, research projects and breakthroughs from across Europe and broadly distributed in institutional and R & D circles.
List of Websites:
The project website is available at http://nanopec.epfl.ch. It not only contains general information on hydrogen production by way of photo-electrochemical water splitting, but also:
1) Presentation of all the consortium partners, with links to their respective websites.
2) List of all project-related publications, with authors, abstracts and references. Papers for which open access has been granted can be downloaded in pdf;
3) Contact addresses.
Contacts :
1) Prof. Michael Grätzel – Head, Laboratory for Photonics and Interfaces
Ecole Polytechnique Fédérale de Lausanne
Station 6
1015 Lausanne
Switzerland
michael.graetzel@epfl.ch
2) Prof. Hans Björn Püttgen – Director, Energy Center
Ecole Polytechnique Fédérale de Lausanne
Station 5 – Château En Bassenges
1015 Lausanne
Switzerland
hans.puttgen@epfl.ch
Hydrogen has the potential to meet global requirements for carbon-neutral energy carriers if produced from carbon-free primary sources such as water using carbon-free energy such as solar energy. Sustainable, cost-efficient large-scale production of hydrogen can, in principle, be established by solar photoelectrochemical (PEC) water splitting, where semiconductor electrodes absorb sunlight to drive water electrolysis: 2 H2O → 2 H2 + O2. Despite four decades of worldwide intensive research on PEC water splitting, this challenge has not been met yet, largely because of material limitations. Photoelectrode materials for solar water splitting must meet multiple and severe criteria such as efficient absorption of sunlight, separation and collection of photogenerated carriers, catalysis of water splitting reactions, chemical stability and durability for a long period, and abundance. Unfortunately, no single material is likely to meet all of the criteria needed to achieve competitive solar hydrogen production by water splitting.
NanoPEC (http://nanopec.epfl.ch/) employs innovative concepts and new methods, enabled by nanotechnology, to design nanocomposite photoelectrodes for solar water splitting, where each component performs specialized functions to overcome intrinsic limitations of single phase materials. Advanced nanostructures that maximize performance of photoanodes and photocathodes are designed, fabricated, characterized, and optimized. Newly-developed p- and n-type semiconducting oxides and oxynitrides are explored and nanostructured into innovative structures and device architectures designed to achieve high solar-to-hydrogen (STH) conversion efficiency, balancing the tradeoff between light harvesting, charge separation and collection, stability and durability. In parallel, fundamental studies and detailed investigations of model systems are carried out to improve quantitative understanding of the effect of material properties, defects and interfaces on PEC processes taking place in water photoelectrolysis. These investigations provide guidelines for rationale optimization of materials and nanocomposite electrode structures designed for high efficiency, and the performance of these structures is evaluated in operative conditions using advanced testing and analytical setups.
Significant progress in PEC water splitting has been achieved during the course of the project. In particular, a new photocurrent record of 7.6 mA cm-2, measured at the reversible hydrogen evolution potential under simulated terrestrial (AM1.5G) solar radiation, was achieved with small-size Cu2O photocathodes protected against photodegradation using a novel overlayer structure. The champion photocathode can potentially reach a STH conversion efficiency of 10% in an ideal tandem configuration, exceeding the NanoPEC target milestone. Scaled-up large (63 cm2) Cu2O photocathodes displayed a photocurrent of 4 mA cm-2, reaching 87% of our target for large modules. On the photoanode front, small-size Fe2O3 nanostructured electrodes reached a photocurrent of 3.3 mA cm-2 at the reversible oxygen evolution potential under simulated solar radiation. This could yield a STH conversion efficiency of 5% in an ideal tandem configuration, breaking the highest record ever reported for stable photoanodes for solar-induced water photo-oxidation. Although the efficiency of large Fe2O3 photoanodes was not as high as their small-size counterparts, largely due to processing difficulties in scaling-up from small to large size electrodes, they did not show any signs of degradation for at least three days.
Based on the progress achieved in this project we expect that a STH conversion efficiency of 10% is within reach using ideal tandem cells. This is comparable to the best performance obtained with state-of-the-art electrolysers powered by state-of-the-art photovoltaic (PV) cells. Unlike the latter technology, which is inherently expensive because it cannot be integrated into a single device, the PEC-based tandem cell technology has the potential to become an affordable means for large scale hydrogen production from water using solar energy. The price of hydrogen produced with such tandem cells is estimated to be in the 7 to 14 € per kg ballpark. This price is quite close to the ambitious target (5 € kg-1) established by the European Hydrogen & Fuel Cell Technology Platform, and suggests that meeting this goal may be realistic with further technological advances assisted by mass production and the economy of scale.
Project Context and Objectives:
Hydrogen (H2) will play a central role in future European energy systems. According to the Strategic Research Agenda (European Hydrogen & Fuel Cell Technology Platform, 2005), hydrogen is expected to be widely available in industrial nations at competitive cost and also serve as a major transport fuel for vehicles with a share of up to 50% by 2050. Moreover, hydrogen can be used for CO2 capture and conversion into other liquid hydrocarbons using Fischer-Tropsch catalysis and also in proton exchange membrane fuel cells to generate electric and thermal energies efficiently while leaving water alone as a by-product. Efficient, cost-effective, and carbon neutral hydrogen production will therefore be essential to a sustainable energy infrastructure.
PEC production of hydrogen under solar illumination is one of the most promising renewable technologies in the future hydrogen economy owing to the huge capacity of the solar energy, which is roughly four orders of magnitude larger than the current global energy consumption and direct conversion from electricity to chemical fuels which is ideal for energy storage and transport. PEC systems use solar photons absorbed by semiconducting photoelectrodes to split water into hydrogen and oxygen (O2) gases directly. PEC systems can thus be relatively simple compared to other solar hydrogen production systems. Primary challenges for PEC systems are to develop materials which are characterized by high photon utilization efficiencies, long lifetimes (stability against corrosion and photodegradation), and low costs.
Semiconductor photoelectrodes must meet several criteria to accomplish efficient solar-induced PEC water splitting: (a) absorb a significant portion of the solar spectrum, (b) conduct charge efficiently, (c) catalyze water reduction or oxidation, and (d) exhibit long-term stability in an aqueous solution. No single material with acceptable performance, stability, and cost has been found, despite decades of investigation. For example, n-type hematite (α-Fe2O3) and p-type cuprous oxide (Cu2O) are inexpensive semiconductors that can be used for water oxidation and reduction under solar illumination, respectively. However, hematite does not utilize photogenerated carriers efficiently on account of a very short charge carrier diffusion length. As for cuprous oxide, the redox potentials for the reduction and oxidation of monovalent copper oxide lie within the band gap, making decomposition of cuprous oxide by photogenerated charges thermodynamically favored over reduction of water 3.
To address these and other significant challenges, we use new concepts and methods, enabled by nanotechnology, to design innovative composite nanostructures in which each component performs specialized functions. This approach reduces the number of criteria that any single component must meet, thus overcoming the basic materials limitations. Computational studies are used to assist analysis of experimental results and provide a solid theoretical background to our investigations, and a wide range of advanced analytical techniques is employed to improve the understanding of structure-composition-property relationships.
NanoPEC consortium consists of eight partners, each of which is assigned to distinct tasks to make best use of their unique expertise and facilities. We aimed to go beyond the state-of-the-art in three critical areas: (1) advanced nanostructures, (2) novel semiconductors, and (3) fundamental studies on model systems, which are the topic of the WPs 1, 2, and 3, respectively. In the fourth WP, we employ the most successful materials to construct and evaluate test devices. WP5 and WP6 are organized for project management and dissemination, respectively.
As a final objective, the NanoPEC project set ambitious targets that are well beyond the state-of-the-art, specifically a one square-centimeter test device that converts solar energy to hydrogen energy with a sustained 10% efficiency (6.6 mA cm-2 with an applied voltage of 1.23 V in a tandem configuration) and a maximum performance decay of 10% over the first 5000 hours of operation and a 100 square-centimeter test device with a sustained 7% efficiency (4.6 mA cm-2 with an applied voltage of 1.23 V in a tandem configuration) and similar stability. Such performance should give a significant impact in short-to-medium term studies which are necessary to achieve long-term environmental, political, and economic benefits as well as European leadership in a sustainable hydrogen economy.
Project Results:
There are several sustainable methods to produce hydrogen other than direct PEC water splitting, such as water electrolysis based on hydroelectric, geothermal, wind, or PV power generations, thermochemical water splitting using concentrated sunlight, and catalytic conversion of biomass. Since we are facing an extra powder demand of 10–15 TW by 2050, it is important to estimate potential availability of the resources first. The potential of hydroelectric generation is approximately 1 TW and one-third of that has already been installed. It is likely that the rest of the resource is rather distributed and available only on a relatively small scale, which requires an additional investment for development. The potential of geothermal power is probably less than 1 TW unless breakthrough in drilling technologies happens. The quantity of biomass can be as large as 5 TW, although the realistic amount is much lower due to water and soil fertility limitations and necessity to feed several billion people. Wind power generation is promising, accounting for 2 TW. Among the other resources, the amount of solar energy is outstanding, being estimated to be 50 TW, although significant scientific, technological and manufacturing breakthroughs are needed for efficient utilization of solar energy.
The price of hydrogen is expected to increase as the scale of hydrogen plants becomes smaller. Therefore, hydroelectric and geothermal energies, and biomass are not likely competitive in industrial and commercial applications in near future. Consequently, practical hydrogen production has been mostly associated with wind and solar PV power generations and thermochemical solar water splitting to date. Typical commercial electrolyser system efficiencies are 56–73% and requires 70.1–53.4 kWh of electricity to generate one kilogram of hydrogen4. The hydrogen cost via electrolysis is dependent on electricity prices. We estimate in consultation with a major car company that the costs of hydrogen via wind and solar PV power generations are approximately 2–3 and 30 € kg-1, respectively. The efficiency of thermochemical water splitting peaks at a certain temperature depending on the degree of sunlight concentration due to a balance between the light absorption efficiency of a reactor and the efficiency of the reaction. The efficiency peak increases with the degree of sunlight concentration from 50% at 100 sun to 80% at 10000 sun5, while increasing the capital cost greatly. In practice, the cost of hydrogen produced by thermochemical solar water splitting is estimated to be approximately 16 € kg-1 by a major car company.
PEC hydrogen production
Solar hydrogen production through photocatalytically assisted water splitting has attracted a great deal of attention since its first discovery almost 40 years ago. The investigations into the use of titanium dioxide, TiO2, photoanodes have continued throughout the years, together with the search of alternative semiconductor materials. A photoanode may be combined in a photo-electrochemical, PEC, cell either with a p-type semiconductor photocathode or with a metallic cathode. The latter configuration was used in the photoelectrolysis cell which first demonstrated the feasibility of PEC water splitting, where an oxygen-evolving TiO2 photoanode was associated with a hydrogen-evolving platinum cathode. Since the band-gap energy of TiO2, (Eg= 3-3.2 eV) precludes any significant solar light absorption, a large number of further studies were directed towards the search of photoelectrode materials that might be able to capture a substantial part of the visible spectrum. Those early investigations, including large variety of inorganic semiconductors, showed that the only materials able actually to photo-oxidize water whilst avoiding photocorrosion were some metal oxides.
The choice of a suitable photoanode material for water splitting imposes an inherent tradeoff between its band-gap energy, which determines its ability to absorb the visible part of the solar spectrum, and its band energetics, which should allow water splitting to proceed with a minimum applied voltage. In fact, only few large band-gap semiconducting oxides, absorbing exclusively UV light, fulfill the requirement of having the potential of the conduction band, CB, edge more negative than the hydrogen evolution potential.
On the other hand, most of the semiconducting oxides, including those able to absorb visible light, exhibit valence band, VB, edge potentials positive enough to allow oxygen evolution. Consequently, the amount of the external voltage bias, required to drive effectively the photoelectrolyser employing a semiconducting oxide photoanode, will primarily depend on its CB edge potential and on the extent of charge recombination within the semiconductor . This determines the photocurrent onset potential and the potential at which the photocurrent attains its maximum (saturation) value. As is discussed below, the minimum voltage that must be applied to reach the saturation photocurrent under sunlight illumination will directly affect the conversion efficiency of the photoelectrolyser.
Titanium dioxide, which served to demonstrate the feasibility of PEC water splitting for the first time, has subsequently been the subject of a large number of studies into the reduction of its band-gap energy to sensitize the oxide to visible portions of the solar spectrum. Those early investigations involving transition metal dopants, such as Cr(III), demonstrated indeed some extension of the photoresponse towards visible wavelengths but also showed that it was accompanied, in all cases, by a significant degradation of the conversion efficiencies within the fundamental absorption range of TiO2 due to enhanced electron–hole (e- - h+) recombination10. More recently, a large number of reports on the “TiO2 band gap narrowing” by doping with non-metallic elements, in particular nitrogen, carbon and sulfur have been published. Among the studies in which such materials were tested as photoanodes for water splitting, those involving carbon, C–modified TiO2, films raised vivid interest. In particular, an initial report of 8.35 % energy conversion efficiency for PEC water splitting obtained using C-modified rutile TiO2 films formed by oxidation of titanium metal in a natural gas flame11, prompted a number of further studies and provoked a lively discussion. Although these films exhibited an optical absorption threshold shifted to 535 nm, corresponding to a band-gap energy of ca. 2.3 eV, the conditions under which the water oxidation photocurrents were measured have been questioned. In fact, the incident photon-to-current efficiency, IPCE, spectra obtained later using similar C-modified TiO2 films15, showed only a moderate shift of the photoresponse towards visible wavelengths consistent with a water splitting photocurrent of ca. 0.6 mA cm-2 determined at a potential of 1.23 V vs. RHE (reversible hydrogen electrode in the same solution) under simulated solar AM 1.5 illumination. In another study, carbon was incorporated into anodically formed TiO2 nanotubes by annealing in a carbon monoxide atmosphere up to 700°C resulting in significant photocurrents, reaching 0.8 mA cm-2 measured under visible light irradiation of 100 mW cm2. The latter work is representative of numerous efforts to improve the visible-light photoresponse of TiO2 by combining doping with structuring the film surface through formation of nanotubes, nanorods or nanowires. In contrast with the aforementioned examples, where the amount of carbon included in TiO2 films exceeded 10 at.% C/Ti ratio, the doping levels in various synthesized nitrogen, N-modified TiO2 samples were of the order of 1 at.% only. In most of those studies the N-modified TiO2 was employed as powder suspensions to promote photodegradation of organic contaminants under visible-light. An attempt to use N-modified TiO2 films formed by reactive magnetron sputtering as photoanodes to oxidize water, resulted in low photocurrents, ca. 0.2 mA cm-2 observed under 100 mW cm-2 visible-light illumination. The poor performance of nitrogen-modified TiO2 materials for water splitting, which is in contrast with significant visible-light absorption extending above 500 nm, is indicative of enhanced e- - h+ recombination facilitated by the high density of electron trap states distributed below the CB.
Beyond the work oriented towards sensitization of TiO2 to visible light, significant efforts were directed over the last decade into improving the PEC performance of other semiconducting oxides allowing efficient capturing of visible light. In this regard there appears to be a broad agreement that an optimal semiconductor to be used for the cleavage of water in a PEC cell should have a band-gap energy close to 2 eV 7,8. This would account, in addition to the thermodynamic water decomposition voltage of 1.23 V, for overpotential (polarization) and Ohmic (resistance) losses. Therefore, considering the band-gap energy alone, ferric oxide, hematite α-Fe2O3, is an ideal candidate material as it has a band gap of 2–2.2 eV, thereby allowing visible light absorption up to 550–600 nm. This leads to a respectable maximum theoretical efficiency for water splitting of 18 %. However, because of the combined effects of a CB edge potential that is too positive and large recombination losses occurring in this material, the photocurrent onset potentials for oxygen evolution at α-Fe2O3 photoanodes are in the best cases (i.e. for doped films) located at rather positive potentials, ca. 0.8 V vs. RHE. The small hole diffusion length in α-Fe2O3, typically of the order of 4 nm, and the poor electric conductivity affect also the shape of the photocurrent-voltage curve to reach 1.23 V vs. RHE, which is normally set as an upper potential limit for the PEC water splitting. To increase the conductivity, doping α-Fe2O3 with Ti (IV) was early proposed, and allows indeed a substantial improvement of IPCEs for the films formed by spray pyrolysis on conductive glass (fluorine-doped SnO2, FTO) substrates, which attained 25% at 400 nm and 1.42 V vs. RHE 20. A major improvement in the PEC performance of thin film α-Fe2O3 photoanodes was brought about by combining atmospheric-pressure chemical vapor deposition (APCVD) with silicon doping21. Those films, which included a thin SiO2 barrier layer pre-formed on the FTO substrate and exhibited pronounced dendritic nanostructures, reached in fact at 1.23 V vs. RHE under AM 1.5 illumination the water oxidation photocurrent of 2.2 mA cm-2 and even larger after adsorption at the Fe2O3 surface of a monolayer of Co2+ ions. The effect of silicon is apparently closely related to the film deposition method. For example, in the case of the α-Fe2O3 photoanodes obtained either by spray pyrolysis or by reactive magnetron sputtering, doping with titanium was more effective at increasing the photocurrents than the doping with silicon. It has been suggested, that besides improving the conductivity of α-Fe2O3 films, the Ti or Si dopants may act by decreasing the extent of charge recombination in the film through passivation of the grain boundaries.
Another semiconducting oxide capable of absorbing a sizeable portion of the visible spectrum is tungsten trioxide, WO3. Although the band-gap energy of WO3 (in the monoclinic form) is larger than that of hematite α-Fe2O3, 2.5 eV compared to 2.2 eV, the photocurrent onset potential of WO3 is less positive, ca. 0.4 V vs. RHE. This reflects the almost generally applicable rule for semiconducting oxides whereby the decrease in band gap is accompanied by a down-shift in the CB edge.
The water oxidation photocurrents reported for both the single-crystal, and polycrystalline film WO3 photoanodes, were adversely affected by the small absorption coefficients for photons with energies close to the band gap. In fact, the optical pathways in WO3 for the visible wavelengths are at least an order of magnitude longer than the hole diffusion length in this material which is of about 0.15 microns22. The charge collection efficiency was significantly improved by the introduction of porous, nanocrystalline film photoanodes consisting of a network of WO3 particles with sizes in the range of tens of nanometers, i.e. smaller than the hole diffusion length. The employed sol-gel synthesis method involving various organic additives allows formation of highly crystalline mesoporous WO3 films which can be largely permeated by the electrolyte. This largely reduces the electron-hole recombination and permitted the optimized WO3 film photoanodes to deliver water splitting photocurrents larger than 2.5 mA cm-2, at 1.23 V vs. RHE under simulated 1.5 AM irradiation. Prolonged photoelectrolysis tests performed in various acidic electrolytes revealed formation of often important amounts of peroxide species causing reversible deactivation of the WO3 photoanodes. It has been shown that the deactivation of the WO3 surface can be avoided by the addition of even small amounts of Cl-, Br- or Co2+ ions to the acidic electrolyte.
Nanostructuring has also been proved to be effective in improving the PEC performance of BiVO4 photoanodes and permitted to reach significant (over 40%) IPCEs for oxygen evolution in the visible spectral range.
NanoPEC: developing new materials and nanostructured electrodes towards efficient water photoelectrolysis
First generation materials
Systematic studies, aimed at optimizing the nanostructure of WO3 photoanodes, involved use of various organic additives to the sol-gel precursor and different annealing protocols. As shown by impedance spectroscopy measurements, high porosity is essential to activate the entire film, providing homogeneous photocurrent distribution across the film. This is particularly important when relatively thick (several micrometers in thickness) films, required to improve visible-light absorption in WO3, are involved. The control of porosity in the course of sequential film deposition allowed, in particular, successful preparation of large scale (10 x 10 cm2) WO3 electrodes for the final NanoPEC tests performed in December 2011. Despite non-optimized PEC cell configuration, such an electrode (90 cm2) exhibited remarkable photocurrents of ca. 1.3 mA cm-2 at 1.23 V vs. RHE under AM 1.5 irradiation. This was also favored by the identification of methane sulphonic acid as the most suitable electrolyte for the PEC water spitting at WO3.
Considerable efforts were also directed towards the search of the most adequate plasmonic metallic nanostructures to be implemented within the mesoporous WO3 films to enhance the PEC reactions. An important enhancement of the water splitting photocurrents was indeed made possible by incorporation plasmonic silver nanoparticles (Ag NPs) within WO3 thin film photoanodes. The choice of the embedded configuration allowed to preserve Ag NPs, with sizes of 10-30 nm deposited on FTO from the contact with the electrolyte, allowing repetitive PEC experiments in acidic electrolytes. For a 1 μm thick WO3 photoanode featuring an Ag NP under-layer of a nominal thickness of 4 nm, a saturation water oxidation photocurrent as large as 2.1 mA cm-2 was observed under AM 1.5 irradiation through the substrate (uncorrected for the absorption losses within the substrate). This corresponds to more than 60% increase (by 0.8 mA cm-2) in comparison to the WO3 photoanode of a similar thickness deposited directly on FTO. Most importantly, the photocurrent at the composite Ag/WO3 photoanode attains already 2 mA cm-2 at 0.8 V vs. RHE to reach a plateau at a potential of only 0.9 V vs. RHE. This makes such composite photoanodes particularly promising in view of the application in tandem cell configuration.
Hydrothermal post-treatment of metal oxides turned out to be a promising processing step towards more efficient water electrolysis. Starting with a hematite Np synthesis route, we obtained solid precipitates and supernatant liquid. The solid precipitates formed to nano particles upon thermal treatment. The supernatant liquid was used for dip coating on FTO substrates and yielded nanostructured pristine α-Fe2O3 thin films after annealing at 500°C with a photo current of 150 µA cm-2.
In a further hydrothermal processing step, we could refine the microstructure of the pristine hematite films to an extent that the photocurrent was doubled. The microstructure of the pristine hematite films is subject to dramatic microstructural changes upon hydrothermal processing. In particular, we could form stellate structures that grow from a turf like seed structure. The nano-flowers have a regular 3-dimensional structure with inclined wings reminiscent of the wind rose (NATO symbol) with inclined wings when viewed in the transversal plane, which we demonstrate by direct overlapping in third image. From the functional point of view, this nano structuring increases the photocurrent from 150 to 250 µA cm-2.
Extending the aforementioned self-assembly based nanostructuring, we explored the structuring of electrode architectures on the micron scale with electric field-assisted and dewetting-assisted pattern formation.
before NanoPEC the state-of-the-art photoanode was hematite doped with silicon, which was prepared using a particle-assisted APCVD system resulting in 500 nm thick films with dendritic cauliflower-like structure. The small features size of those mesoporous film was on the order of the hole diffusion length (ca. 5 nm), enabling efficient hole collection at the surface. Additionally, efficient electron transport to the back contact was allowed owing to its high dopant concentration.
The NanoPEC consortium improved the performance of the state-of-the-art hematite photoanodes further by controlling the APCVD process parameters and using a highly-active water oxidation catalyst. Improved alignment of the hematite grains facilitated more efficient electron transport, improving the electron conductivity through the film. This enabled thicker films (700 nm) to be effectively utilized than in the previous work (500 nm), improving light absorption, increasing the surface area and facilitating more effectively charge carrier extraction. The kinetics of the water oxidation reaction was improved by incorporating IrO2 NPs 2 nm in size at the hematite/electrolyte interface. IrO2 NPs were successfully deposited by passing a current of 40 mA cm-2 between a hematite anode and a Pt mesh cathode which were separated by approximately 1 mm. It is noteworthy that the incident light was not screened by IrO2 NPs owing to the small particle size. Consequently, a new photocurrent record of 3.3 mA cm-2 was achieved, surpassing the old record by more than 50%. The new record corresponds to a STH conversion efficiency of 5% in an ideal tandem configuration. However, the thick stems of the cauliflower structure still prevents a large proportion of the absorb light to participate in the water splitting reaction. This necessitates alternative nanostructuring strategies such as host-guest nanocomposite structures, new synthetic routes of nanostructures with added functionality, plasmon-enhanced light absorption, semiconductors with graded compositions, and overlayer and underlayer modifications, in addition to the development of new materials and catalysts, in order to optimize the performance of the 2nd generation materials and nanostructures.
Second generation materials
In addition to the development of various approaches to enhance the performance of first-generation materials such as Fe2O3 and WO3 (nanostructuring, catalysis, light management), the development of novel photoanode and photocathode materials was an important part of the consortium’s efforts. A total of 12 new materials have been synthesized and subjected to initial performance and stability screening: Cu2O, CuAlO2, CuCrO2, SrTi1-xFexO3-delta, BiFeO3 (photocathodes) and BiVO4, TaON, LaTiO2N, Ta3N5, and MNbOxNy with M=La, Nd, Sm (photocathodes). An overview of the results achieved is shown in Table 1. Four of these materials have met NanoPEC’s stringent requirements for continued efforts, and have been designated as “second generation” materials: Cu2O, BiVO4, and the (oxy)nitrides LaTiO2N and Ta3N5.
Cu2O
Cuprous oxide is a p-type semiconductor that seems an ideal candidate for water splitting applications: it has a band gap of 2.0 eV and it is one of the very few oxides that has high enough CB edge to reduce water to hydrogen. Earlier studies on this material have met with limited success due to excessive photodegradation, in which the Cu+ ions are reduced to their metallic state by the photo-excited electrons: Cu2O(s) + H2O(l) + 2e- -> 2Cu0(s) + 2 OH-(aq). NanoPEC partners have solved this issue by depositing 20 nm thick protective overlayers of Al-doped ZnO and TiO2. In addition, Pt NPs were deposited at the surface to catalyze the hydrogen evolution reaction. This has resulted in photocurrents of -7.6 mA cm-2 at 0 V vs. RHE under illumination with simulated sunlight (AM1.5) which is the highest value ever recorded for any metal oxide photoelectrode. The stability was improved further by optimizing the deposition parameters of the TiO2 overlayer and by choosing a phosphate buffer solution as the electrolyte.
BiVO4
With a band gap of 2.4 eV, BiVO4 can theoretically convert up to ~9% of the incident solar energy into hydrogen. NanoPEC researchers have used spray pyrolysis as a low-cost and easily scalable technique to deposit thin dense films of this material onto FTO-coated substrates. Initial studies revealed that poor electron transport and slow water oxidation kinetics limit the performance of this material38. To address these issues, three improvement strategies were pursued. First, the spray deposition process was optimized in order to obtain a favorable film morphology. Second, a cobalt-phosphate oxygen evolution catalyst was deposited onto the BiVO4 films by electrodeposition. Finally, 0.5% W (tungsten) was incorporated in the film as a dopant to enhance the electronic conductivity. These combined measures increased the AM1.5 photocurrent from 0.2 to 2.2 mA cm-2, which is currently the record for Co-Pi-catalyzed BiVO4. Moreover, the absorbed-photon-to-current efficiency (APCE) is nearly 100% at low light intensities, indicating that defect-induced recombination in the bulk of the BiVO4 is virtually absent. This is unusual for an oxide semiconductor synthesized with a low-cost, low-temperature process and suggests that higher efficiencies are within reach.
(Oxy)nitrides: LaTiO2N and Ta3N5
Oxynitrides are promising photoelectrode candidates because their band gap is generally smaller than of corresponding oxides, while still being reasonably stable. LaTiO2N has a nearly ideal band gap of 2.0 eV, but the performance is limited by the presence of defects – presumably Ti3+ centers. NanoPEC researchers have developed a successful strategy for suppressing the formation of Ti3+ in LaTiO2N by substitutional doping of Ca2+ on La3+ sites. This was achieved by treating A-site deficient La1-xTiOy with Ca(NO3)2 before the ammonolysis step. As a result of this so-called “back-filling” strategy, the oxygen evolution efficiency of the resulting powders is ~2 times higher than that of the previous generation of LaTiO2N powders.
Efforts to synthesize phase-pure beta-TaON have resulted in a new optical monitoring technique that allowed NanoPEC partners to follow the oxide-to-nitride conversion in-situ, at temperatures of up to 800°C and with a time resolution of ~1 sec. These experiments revealed that conversion of Ta2O5 to Ta3N5 proceeded via the N-doped Ta2O5 phase, without any indications of the formation of -TaON. The resulting Ta3N5 films showed AM1.5 photocurrents of up to 1.2 mA cm-2 when catalyzed with IrO2. Moreover, resonant light trapping in these films resulted in significant improvement of the quantum efficiency of these films. The stability of these films should, however, be improved.
Fundamental understanding of underlying processes and novel investigation technique
Fundamental investigations aiming at understanding the physical and electrochemical processes underlying the operation of selected photoelectrode materials were carried in order to identify the performance limiting factors and suggest new routes to overcome these limitations in order to improve performance. Detailed investigations were carried out on selected materials that we identified as promising photoelectrode candidates, namely -Fe2O3, WO3, Cu2O, BiVO4, TaON and La(Ti,Nb)(O,N)3. Complementary experimental techniques were combined with advanced analysis methods, detailed theoretical modeling and ab-initio calculations in order to reveal the effect of microstructure, interfaces and grain boundaries, chemical composition and defects on the electronic structure and PEC properties of these electrodes.
NanoPEC partners developed a powerful analysis method for resolving the different loss mechanisms in the water photoelectrolysis reaction and quantifying their contributions to the overall performance. The water photoelectrolysis reaction is a cascade of three processes: (1) light absorption and carrier (electron-hole pair) generation; (2) carrier separation and collection at the counter interfaces; and (3) selective carrier injection at the respective interfaces. The overall STH conversion efficiency is the product of the efficiencies of these processes. While the light absorption efficiency can be readily quantified using optical measurements, the other two processes (charge separation and injection) are difficult to deconvolute and their respective contributions to the overall efficiency has been unknown. To rectify this we developed a new analysis method that is simple, robust and very effective and informative. Using H2O2 as hole scavenger that sets the injection efficiency of holes to the electrolyte to 100%, we compare the photocurrent obtained for the same photoanode under the same conditions in aqueous solutions with and without H2O2. The quotient of the photocurrent measured without H2O2 divided by the photocurrent measured with H2O2 in the electrolyte gives the injection efficiency, which is the probability of holes reaching the photoanode/electrolyte interface to drive the water oxidation reaction rather than being lost for surface recombination. Thus, the quotient of the overall efficiency divided by the optical and injection efficiencies yields the charge separation and collection efficiency. We applied this method to quantify and analyze the different loss mechanisms in state-of-the-art nanostructured α-Fe2O3 photoanodes prepared by APCVD and ultrasonic spray pyrolysis, as well as BiVO4 photoanodes prepared by spray deposition. In the former case we found that at sufficiently high potentials the injection efficiency of holes is high, exceeding 90%, while the separation and collection efficiency remains low (< 20%), indicating that bulk recombination is the dominant loss mechanism of these electrodes. By analyzing the potential dependence of the charge separation and collection efficiency we found that most of the holes arriving at the surface to oxidize water are collected from the space charge (depletion) region, in agreement with other studies reporting extremely short diffusion length of holes in α-Fe2O3. In case of the BiVO4 photoanodes we found that the charge separation efficiency can reach values close to 100%. The injection efficiency, however, was less than 40%, which shows that hole transfer across the semiconductor/electrolyte interface is the limiting process. This was solved by using a cobalt phosphate catalyst, which increased the injection efficiency to more than 90%. In order to overcome the tradeoff between light harvesting and charge separation and collection efficiency, NanoPEC partners developed an innovative approach for light trapping in ultrathin films. This approach enables trapping most of the light in films as thin as ~20 nm, for which the charge separation and collection efficiency is intrinsically high. Theoretical calculations for Fe2O3 photoanodes comprising ultrathin films on reflective substrates suggest that photocurrent densities approaching 8 mA cm-2 can be reached with an optimal cell design. This innovative technology is being patented and efforts to achieve high performance are undergoing. To understand the microstructure – properties correlation in nanostructured photoelectrodes for water splitting, we carefully analyzed the microstructure and chemical composition of our electrodes using a wide array of characterization techniques scaling from the macroscopic specimen dimensions down to the microscopic nanometer level, which is the scale of interfaces. These multiscale characterizations have led to new insights on the effect of microstructure on PEC properties of nanostructured electrodes. In particular, we found a remarkable anti-correlation between the photocurrent and missorientation angles between domains of highly oriented nanograins in state-of-the-art nanocrystalline α-Fe2O3 photoanodes. Changing the deposition rate from 2 to 6 L min-1 and the substrate temperature from 415 to 425 C led to a twofold increase in the photocurrent density, reaching more than 4 mA cm-2 at 1.4 V vs. RHE. Standard investigations of the microstructure of these electrodes using conventional scanning and transmission electron microscopy methods revealed no clues as for the origin of the remarkable improvement in performance; the two electrodes look the same. The mystery was resolved by careful analysis of the missorientation angles of nanosized grains and sub-micron domains in these nanocrystalline mosaics. Towards this end we developed a multiscale orientation mapping method that involves composite imaging of dark field transmission electron microscopy micrographs obtained at different zone axes. The orientation maps of the champion electrode display mosaic structures (“cauliflowers”) of unmixed orientations, for the most part while its sister electrode displays mixed orientations in most of the cauliflowers. The anti-correlation between the photocurrent and missorientation angles of these electrodes indicates that high-angle grain boundaries are detrimental to the performance of nanocrystalline α-Fe2O3 photoanodes.
Beside the grain boundary effect we also found interesting correlations between the electrode microstructure and the charge transfer processes at the electrode/electrolyte interface. These effects are particularly important in mesoporous electrodes of high surface area, such as our WO3 photoanodes. To understand these effects we combined the conventional I-V measurements together with careful impedance spectroscopy analysis in electrolytes of different conductivities. The photocurrent plateau was found to increase with increasing electrolyte conductivities, up to a certain limit at an electrolyte conductivity of about 200 mS cm-1. Using impedance spectroscopy analysis to resolve the different contributions to the overall impedance, the dominant contribution was found to the charge transfer resistance at the electrode/electrolyte interface. The latter increases with decreasing electrolyte conductivities, which is the primary reason for the degraded performance of mesoporous WO3 photoanodes in non-concentrated electrolytes (0.5 M in the case of sulphonic acid electrolytes). This result is important for getting the most out these electrodes.
Taking another step further beyond the effect of microstructure, the effect of atomic defects such as vacancies on the electronic structure and the effect of electronic defects such as trapped charge states induced by illumination or electrochemical pretreatment on the PEC properties of WO3 and α-Fe2O3 photoanodes were investigated using near-edge X-ray absorption fine structure (NEXAFS) spectroscopy.
Distinct surface and sub-surface states are formed on hematite upon anodic polarization. Anodic polarization at 600 mV vs. Ag/AgCl reference electrode (i.e. anode PEC operation) gave rise to a new electronic transition that is the signature of the upper Hubbard band of the oxygen NEXAFS spectrum. This is associated with a high oxidation state of Fe species at the surface. Anodic polarization at 200 mV vs. Ag/AgCl is not sufficient to induce this transition. However, anodic polarization at 0, 200 and 600 mV vs. Ag/AgCl caused the Fe-resonant X-ray photoelectron spectra (XPS) to shift systematically towards the Fermi level, indicative of the hole doping upon anodic polarization.
Transient surface states were observed in a spectacular experiment by taking advantage of recent progress in synchrotron radiation instrumentation which enabled us to investigate a working PEC system operando with soft x-ray spectroscopy under electrochemical polarization and illumination with visible light. 30 nm thin α-Fe2O3 films were deposited on Si3N4 substrates fitted with a 100 nm thick window that allows soft X-rays to access the electrode/electrolyte interface. These tiny micro-cells were assembled on an electrochemical micro-cell specifically designed for in-situ investigations at the Advanced Light Source in Berkeley USA.
X-ray spectra were recorded at the oxygen K-edge, which is very sensitive to the local bonding and symmetry properties of the oxygen ions. The O 1s spectra of α-Fe2O3 have a well-developed doublet in the pre-edge region at 528 to 530 eV, originating from transitions of hybridized Fe(3d)-O(2p) states with t2g and eg orbital symmetry. We notice clearly two new spectral features when the film is illuminated and biased between 300 to 700 mV with respect to the counter electrode. The energy position of the two new peaks relative to the absorption edge reveals that the lower energy peak and the higher energy peak at 525.5 and 527 eV are electron hole transitions with t1u and a1g symmetry into the charge transfer band and into the upper Hubbard band, respectively.
The variation of the relative spectral weight of these new transitions in comparison with the photocurrent versus bias shows that both types of holes are active for the water photoelectrolysis reaction. Thus, we have directly identified two different electron hole states in the VB, which, contrary to published literature, are both active for water photoelectrolysis.
In addition to advanced experimental techniques we also used detailed physical modeling and ab-initio calculations to achieve deeper and broader understanding of the underlying physics and to reveal the atomistic origin of the electronic structure of the photoelectrodes and their PEC properties.
Numerical calculations based on finite element method were used to simulate the electric field intensity around Au/SiO2 core/shell NPs of interest for plasmonic effects. Simulations for 100/17 nm Au/SiO2 core/shell NPs assembled in the form of an ordered array on top of a 50 nm thick α-Fe2O3 film on a 350 nm thick FTO layer on a glass substrate reveal the important role of the interparticle spacing on the distribution of the electric field intensity around the NP. At a wavelength of 610 nm, which is close to the absorption edge of α-Fe2O3, the electric field is scattered into the substrate or the α-Fe2O3 film for interparticle spacing of 250 or 150 nm, respectively. Therefore, it is important to design not only the size of the plasmonic NPs but also their arrangement.
Ab-initio density functional theory (DFT) calculations were carried to understand the electronic structure of selected photoelectrodes materials (α-Fe2O3, WO3, BiVO4, SrTi0.5Fe0.5O3 and TaON) and to model the electronic interaction between α-Fe2O3 and the adsorbed water layer at the surface. A new calculation method was developed to achieve more accurate calculations that the available methods by taking into account different populations of electrons and holes in different electronic states in the VB and CB. The band gap energy obtained by this method (2.28 eV) is very close to the experimental values reported for α-Fe2O2 (2.1 – 2.2 eV). Calculations for a slab of α-Fe2O3 covered with two monolayers of water reveal the presence of a potential barrier for hole transfer from the sub-surface region to the outermost atomic layer where charge transfer to the electrolyte takes place. The barrier increases upon surface oxidation, reaching about 900 mV wherein 300 mV falls at the sub-surface region and the rest (600 mV) is the injection barrier from surface to the electrolyte. These results are consistent with the typical overpotential values of α-Fe2O3 photoanodes. The origin of the potential barrier can now be traced to the electronic structure at the α-Fe2O3/water interface. Our calculations show that the oxidized surface of α-Fe2O3 is enriched with holes that are trapped at the outermost oxygen sites in contact with water. These self-trapped holes mitigate hole transfer to the electrolyte, stabilizing the adsorbed water and impeding the water oxidation reaction. This is illustrated in the spin density maps of an isolated Fe(HO)3 complex, that is a complex of an hydrated Fe(III) with two water molecules, shown in Figure 16 before (right) and after (left) oxidation. When this complex is oxidized to FeO(HO)2 by trapping a hole and releasing a proton, the trapped hole is stabilized at the O-1 surface oxygen site, mitigating its transfer to the coordinated water layer.
Last but not least, the stability of semiconductor photoelectrodes in aqueous solution – an import issue that is often neglected by the PEC community – was carefully analyzed using electrochemical equilibria considerations. The Pourbaix diagrams52 were found to be an effective and easy to use tool for evaluating the stability of binary oxides such as TiO2, Fe2O3, WO3 and Cu2O. This approach can be extended to evaluate the stability of complex compositions (ternary and quaternary oxides and oxynitrides) using thermodynamic databases.
Novel nanostructuring approaches
Host-guest mesostructure approaches employ extremely-thin semiconductor films deposited on a nanostructured collector with a large specific surface area, in analogy to dye sensitized solar cells (DSCs). In this architecture, photoexcited carriers are generated within the range of charge transport lengths and most of them can access the surface. Although incident light could not be absorbed by a single ultrathin film, sufficient photoresponse should be obtainable by coating ultrathin films conformally on a suitable conductor with a large specific surface area. We found that this approach actually increased the photocurrent and APCE of ultrathin hematite photoanodes using a nanostructured scaffold made from WO3.
WO3 host layers and Si-doped hematite guest absorbers were deposited by APCVD successively. WO3 host layers created a porous structure with increased roughness over the underlying substrate. The WO3 scaffold was completely covered with ca. 10 nm particles of silicon doped hematite after the deposition of the guest absorbers. A significant increase in the photocurrent plateau was observed for the WO3/Fe2O3 photoanode over the control electrode prepared without the WO3 scaffold as shown in Figure 18e. We confirmed that the APCE was increased over all wavelengths below the absorption edge of Fe2O3. The enhancement was particularly significant near the absorption edge, because photoexcitation occurred near the semiconductor-liquid junction (SCLJ) at the surface even for the large photon penetration lengths at these wavelengths.
The benefit of the guest-host nanocomposite design is most important for ultrathin absorber layers. However, extremely-thin hematite films suffer from an inactive “dead” layer near the interface with the substrate, likely due to deleterious electronic interactions (interface recombination) and structural defects (uncoated regions). To improve interfacial properties, ultrathin interfacial oxides (< 3 nm typically) were developed as a buffer layer. We found that a buffer layer of SiOx, Ga2O3, Nb2O5, or TiO2 improved the performance of ultrathin hematite films in PEC water oxidation. The function of the buffer layers is attributable to enhancement of crystallinity and conformal coating of the ultrathin hematite layers and improvement in the photovoltage and charge separation as a result of surface acid-base interaction, crystallographic template effect, and/or suppression of electron back injection from the FTO to the hematite film 57, depending on the kinds of buffer layers. The ultrathin hematite electrodes modified with Nb2O5 underlayers demonstrated the highest APCE ever reported for ultrathin hematite films. Additionally, a Ga2O3 buffer layer was successfully combined with a rough conducting substrate made of antimony-doped SnO2 (ATO) NPs, which improved the photocurrent by a factor of 1.4 while maintaining the positive effect of the Ga2O3 buffer layer on the photocurrent onset. These results clearly demonstrate the potential of buffer layers to boost the performance of guest-host nanocomposite electrodes.
Functional overlayers were developed for ultrathin and cauliflower-type hematite photoanodes in order to further improve the performance of guest-host electrode designs. Al2O3, Ga2O3, and In2O3 overlayers deposited on ultrathin hematite films improved the photocurrent onset by up to 200 mV. Similar improvement in the photocurrent onset was observed for cauliflower-type hematite films modified with an Al2O3 overlayer. It was concluded that corundum-type overlayers released lattice strain of ultrathin hematite films and/or decreased a density of surface states. Importantly, stoichiometric water splitting for longer than one day was confirmed on Ga2O3-deposited ultrathin hematite using a newly-developed gas detection system, attesting the excellent stability of the system.
Porous colloid-based films with feature diameters of ca. 20 nm are attractive morphologies owing to their potential to achieve efficient light absorption and carrier collection. High temperature annealing was usually necessary to make colloid-based hematite films photoactive, which inevitably led to sintering of NPs and deterioration of the original mesoporosity. To prevent coarsening, a silica layer was coated on the hematite films before the thermal treatment As a result, the feature size of the hematite particle was unchanged upon annealing at 800°C when a SiO2 scaffold was applied. After removal of the silica scaffold, the films demonstrated an increased photocurrent plateau, but also an anodic shift in the onset potential most likely due to surface traps generated during the silica scaffold removal. It was found that the photocurrent was enhanced by applying an Al2O3 overlayer, yielding a photocurrent close to 4 mA cm-2 with a sufficient external voltage. Although the photocurrent onset has yet to be improved by healing surface traps, this result highlights the potential of mesoporous hematite films for PEC water splitting.
Mesoporosity was also introduced to WO3 photoanodes using a sol-gel process which involved poly(ethylene glycol) (PEG) as a structuring agent. The amount of carbon played the dominant role in film nanostructuring. The best films was obtained by combining polymers with the lowest molecular weight and an ultrasonic precursor pretreatment, improving the photocurrents exceeding 3 mA cm-2 at 1.23 V vs. RHE. Such films were characterized by porosity of ca. 50% and individual particle sizes of 30-50 nm, which was smaller than the hole diffusion length in WO3. An additional increase in the current density was obtained with samples containing boron, with an optimal content of 10 at .%.
PEC cells and measurement protocols. An experimental, computer controlled test bench equipped with a solar simulator (AM 1.5 G, 1000 W m-2) was assembled, allowing photocurrent-voltage, electrochemical impedance spectroscopy and stability characterization of small (~ 1 cm2) and large (~ 100 cm2) PEC devices. To evaluate the performance of large PEC devices, another test bench with a larger solar simulator equipped with a horizontal light beam was designed.
The efficiency calculations for large electrodes in the reports were reviewed and shown to be accurate. Concerning the above-mentioned new photoreactor, the authors assume that the reviewers are referencing the newly designed PEC cell presented during the last meeting of the project by partner University of Porto (Adelio Mendes). However, the prototype design presented is not part of the final NanoPEC report and was only presented to show how the project’s results would evolve in the near future. Since this presentation, the new prototype design was much improved and is only now being fabricated. After being fabricated, it will be assessed (including efficiency calculations) and, most probably, improved again and only after being patented it will be diclosed and disseminated. Photoelectrochemical cell setup - Small Prototype
The first characterization experiments of small photoanode samples (up to 1 cm2) performed in the framework of NanoPEC project were conducted in a Cappuccino cell provided by EPFL60. However, a new PEC cell was designed at FEUP allowing the performance evaluation of different photoelectrode samples with an active area from 1 to 9 cm2. This new testing device was called Portocell and it was used as the small prototype for PEC cells. The Portocell PEC cell has two removable windows (front and back) made of black acrylic both screwed to a transparent acrylic part, which fixes the synthetic quartz window by pressing them against an O-ring using five screws. These windows may be changed according to the illumination area desired. It is also possible to close the PEC cell by screwing an acrylic cap with only open spaces for O2 and H2 evolution.
In this configuration the photoelectrode is kept vertically aligned and facing perpendicularly to the light beam. A standard three-electrode configuration is used with a 99.9 % pure platinum wire (Alfa Aeasar®, Germany) as a counter electrode and an Ag/AgCl/Sat. KCl as a reference electrode (Metrohm, Switzerland). The electrodes should be immersed in the appropriate electrolyte solution, depending on the semiconductor material.
Photoelectrochemical cell setup - Large Prototype
In order to test large photoelectrodes samples with an active area up to 10 x 10 cm2, the Portocell PEC cell was redesiged. The large Portocell PEC cell works as a single and compact vessel made of transparent acrylic. This cell has two removable windows (front and back) made of black acrylic, allowing an illumination area of 100 cm2, both screwed to the vessel. Moreover, a stainless steel part was used to fix the synthetic quartz window to the vessel, which is pressed against an O-ring by means of twelve screws. This cell allows using a Teflon diaphragm between both electrodes, avoiding hydrogen and oxygen gas bubble to mix. After assembling all the previously mentioned parts, two different caps are used to cover the vessel. The internal cap is used to fix the electrical contacts allowing to establish a correct connection between the electrodes and the potentiostat; the external cap closes the cell allowing to collect oxygen and hydrogen separately.
Performance and stability results with best materials
All the investigations performed to develop new materials for both photoanode and photocathode culminated in the construction of a stable device architecture that achieves the high targets of conversion efficiency proposed. Two prototypes were developed: a small device (up to 1 cm2) achieving a 7 mA cm-2 photocurrent under 1 sun illumination, corresponding to a 10 % STH efficiency; and a large device (~ 100 cm2) that achieves 7 % efficiency. α-Fe2O3, WO3 and Cu2O photoelectrodes were the most promising materials developed during NanoPEC project and these were the ones selected to build the prototypes.
The current-voltage (I-V) characteristic curves of the selected materials were obtained by applying an external bias potential to the PEC cell and measuring the generated photocurrent using a ZENNIUM workstation (Zahner Elektrik) - Figure 22. The measurements were performed in the dark and under simulated sunlight and a standard three-electrode configuration was used.
The small hematite photoanode is a titanium doped structure prepared by the solution-based colloidal method. Its water oxidation photocurrent was measured in a 1 M NaOH (pH 13.6) aqueous solution. A photocurrent value of 1.86 mA cm-2 at 1.43 VRHE and a maximum photocurrent of 2.34 mA cm-2 before the onset of the dark current were reached. The maximum photocurrent presented here for small devices represents the highest photocurrent ever reported for a solution-processed hematite film under standard solar illumination conditions and is furthermore near the values reported with the previously-reported silicon doped films prepared by APCVD. The hematite photoanode scaled up to 100 cm2 was prepared by spray pyrolysis combined with a SiOx buffer layer and it reached a photocurrent value of 0.38 mA cm-2 at 1.43 VRHE.
Concerning the WO3 photoanodes, a series of organic additives to the sol-gel precursor, acting as structure directing agents, were tested to optimise the photocurrent-voltage behaviour. In particular, various ratios of different molecular weight polyethylene glycols (PEGs 200, 300, 600 and 1000) were added to the freshly formed aqueous polytungstic acid solution and thermally decomposed in one or two steps. This allowed to largely modify the porosity and, subsequently, the current distribution within the films. The measurements of tungsten trioxide samples (small and large prototype) were conducted in a 3 M CH3HSO3 (pH 0.26) electrolyte solution. Figure 26 presents the I-V characteristic curves for WO3 small photoanode (~1 cm2) deposited on FTO glass substrate and for WO3 photoanode with ~100 cm2 of active area deposited on a tungsten substrate by the sol-gel method.
WO3 small size photoanode reached a photocurrent of 2.92 mA cm 2 at approximately 1.23 V vs. RHE. After the scale up of these photoanodes, a photocurrent density of 1.22 mA cm-2 was obtained at the same potential. The observed shape of the I-V plot is affected by ohmic drops within pores and unfavorable current distribution over the electrode surface.
Finally, in what concerns Cu2O photocathodes modified with the ultrathin oxide overlayers and Pt catalyst, the current–voltage curves were measured in a 1.0 M Na2SO4 aqueous solution buffered at pH 5 with potassium phosphate (0.1 M) - Figure 27(b). A photocurrent density of 7.6 mA cm-2 for the small devices was achieved, corresponding to 10 % STH efficiency. This efficiency value meets one of the project targets. The enhancement of the photocurrent was attributed to formation of a p-n heterojunction between Cu2O and Al:ZnO layers which could assist charge separation. For an active area of 63 cm2 the photocurrent onset at approximately 0.4 V vs. RHE is similar to the small electrode. However, the increase of the photocurrent during the potential sweep toward negative potentials was not steep, which was indicative of a higher resistance of the electrode due to the thicker TiO2 overlayer. Still, a current density of ~ 4.0 mA cm-2 at 0 V vs. RHE was obtained, corresponding to a STH efficiency of 6 %. This value is slightly below the project target of 7 % STH efficiency. Moreover, an optimized photoelectrode sample of Cu2O may allow to produce solar water splitting devices with a very high efficiency. The proposed 100 cm2 photoelectrode active area was not obtained due to equipment size limitations, though the scale up of this photoelectrode is direct and simple.
Stability tests
In PEC cells stability issues is one of the major problems to be solved. Usually, when a semiconductor electrode is brought into contact with an electrolyte solution some reactions may occur, for instance ionic oxidation or reduction of the semiconductor with simultaneous reduction or oxidation of a component. The electrolytic reduction of a semiconductor is often associated with the electrons in the VB, while the electrolytic oxidation reaction is related to holes in the CB as electronic reactants. Taking this into account, the stability/corrosion tests were performed to identify possible degradation pathways in the semiconductors materials. The overall stability is controlled by two factors, namely physical and chemical factors. Physical stability is related to electrolyte evaporation, which heats up under the solar simulator illumination. The chemical stability is associated to irreversible electrochemical and thermal degradation of the semiconductor and electrolyte64. The corrosion phenomena in the electrolyte were investigated by two techniques: inductively couple plasma (ICP) and atomic absorption. Modifications on the photoelectrodes surfaces induced by the stability tests were analyzed by high resolution scanning electron microscopy. In order to reach medium- and long-term stability targets, a set of stability measurements was performed.
Medium-term stability testing
In what concerns medium-term stability tests, the measurements were performed over 72 hours, under constant irradiation conditions (AM 1.5 G, ~1000 W•m-2, 25 °C). Ti-doped α-Fe2O3 photoanode prepared by the solution-based colloidal method was very stable, exhibiting a constant photocurrent density response of approximately 0.82 mA cm-2, at a constant potential of 1.45 VRHE, over the entire 72 h. The hematite film was characterized in 1.0 M NaOH electrolyte (pH 13.6). After the stability test, the photoanode did not exhibit any signs of degradation, suggesting that no major corrosion phenomena took place during the stability test. SEM images before and after the stability test show no significant differences between the samples analyzed before and after stability tests. Additionally, in what concerns hematite particles, both exhibited similar morphological structure; the sample after stability showed larger particles (30–70 nm). Moreover, the presence of iron in the electrolyte solutions was analyzed by ICP technique to evaluate the possible dissolution of iron. The electrolyte volume tested was ca. 50 cm3. The ICP results revealed only trace amounts of this compound in the electrolyte used during the stability tests, with values lower than 100 µg L-1, proving the stability of this photoanode.
A WO3 film with a thickness of ca. 2 µm deposited on a conducting glass substrate using a PEG 300 precursor was also submitted to stability tests. The WO3 photoanode was immersed in 1 M MSA (pH 0.15).
This sample was submitted to two periods of 72 hours at 1.20 VRHE and between the two testing periods the cell was filled with fresh electrolyte solution. The photocurrent density decreases over time for a period of about 10 hours, starting to increase afterwards. At the end of the first stability test period, WO3 photoanode exhibited a photocurrent density of ca. 0.59 mA cm-2 and at the end of the second period exhibited 0.47 mA cm-2. Here, it is visible that this sample lost just 5 % during the first 72 hours. Analyzing SEM images it can be concluded that WO3 sample maintains mostly the initial morphological structure, although more compact.
The presence of tungsten in the electrolyte solution was obtained by ICP and reveals trace amounts of this compound after the stability test (two periods). The electrolyte volume tested was ca. 50 cm3 and it showed a concentration of 0.150 mg L-1. The electrolyte solution revealed also the presence of sodium, potassium and manganese.
Long-term stability testing
In order to evaluate the major drawbacks of PEC devices when applied to long-term stability tests, the test bench assembled was adapted. This study allows achieving a more complete understanding of the device performance under real applications. To perform high-temperature measurements up to 70 °C, the Portocell was also adapted by replacing the back-side of the cell with a silicone rubber heater . The electrolyte temperature was controlled using a water bath and, inside the Portocell, with a thermocouple. The electrolyte was then continuously pumped in and out of the PEC cell by a recirculation system. The pH of the electrolyte was controlled using a pH meter.
This way, long-term stability tests were performed by monitoring aging tests effects at 60°C. A Ti-doped α-Fe2O3 film (prepared by solution-based colloidal method) immersed in 1.0 M NaOH was used to perform the stability analysis at an applied potential of 1.45 VRHE for 72 hours under 1 sun constant illumination. At this potential the fresh sample exhibited a photocurrent density of ca. 0.54 mA cm-2 at 60 °C. A significant drop in the current density was observed in the first hours, subsequently reaching a photocurrent density of approximately 0.13 mA cm-2. Then, in the following time the photocurrent density remains more or less constant. After the stability test, the sample exhibited a rougher surface with a less compact structure. It is likely that the alkaline electrolyte solution could etch the dopant of hematite photoanodes at high temperatures and deteriorate the performance. For iron oxide, we performed accelerated tests for 72 hours, which, with an acceleration factor of about 10, corresponds to 720 hours of stability testing under constant illumination, or the equivalent of 90 eight-hour days. After a decrease in photocurrent during the first few hours, the current was quite stable for the remainder of the accelerated test. Unfortunately, the photocurrents for the hematite electrodes did not scale well on our standard substrates (FTO), and further work on the hematite was devoted to improving charge collection.
As for the Cu2O electrodes, we observed stability issues on the order of one day under non-accelerated conditions, and thus focused on improving the stability rather than measuring unstable electrodes for long periods of time. In fact, since the completion of the project, a deeper understanding of the mechanism of instability has been attained, involving erosion of the catalyst particles from the surface of the photoelectrode (re-platinization recovers the photocurrent). To date, there is no evidence of degradation of the underlying Cu2O material.
As for the lower active area, this is simply an artifact of the size of our ALD machine. Thus, we constructed the mosaic electrode composed of nine smaller Cu2O electrodes. Due to necessary masking of conducting surfaces, this reduced the active area. Cuprous oxide – Attaining the highest STH conversion efficiency
The instability problem of Cu2O photocathodes were solved by covering the photoactive semiconductor with suitable protective layers by atomic layer deposition as described above. The resultant photocathodes exhibited a photocurrent of –7.6 mA cm-2 at 0 V vs. RHE, which successfully fulfilled the photocurrent target of NanoPEC3. However, the photocurrent decreased to 33% of the initial value in 20 min. It was clarified that electrons were trapped in an amorphous TiO2 overlayer. Such reduced Ti species could prevent photoexcitation of Cu2O. In addition, an amorphous TiO2 layer was found to be not stable in certain electrolytes. Therefore, our effort was focused on improvement of the quality of a TiO2 overlayer.
It was found that a heat treatment in air at 200°C for 45 minutes improved the photocurrent stability of Au/Cu2O/Al:ZnO/TiO2 although it also reduces the initial photocurrent to –5.7 mA cm-2. This result suggests that thermal treatments could improve the quality of the overlayers and thus the stability of the electrode. In fact, the TiO2 overlayer exhibited significantly-improved stability when being deposited at relatively higher temperatures. Additionally, TiO2 was found to be stable in a phosphate buffer solution while not in an acetate buffer solution. Consequently, Cu2O electrodes were stabilized by refining deposition conditions of the overlayers and choosing an appropriate electrolyte. The photocurrent was stable over 60 min at least. It was further confirmed that a sample did not show significant deactivation for 12 hours.
Performance comparison with other water splitting and H2 production routes
The production of hydrogen via water splitting using solar energy was first described by Fujishima and Honda in the early 1970s. Currently, water splitting can be accomplished via three general types of devices: i) composite devices - PV cells connected with an electrolyser and or PEC cells; ii) single devices - SCLJ PEC cells; and iii) thermochemical cycles. For the present report only type i) and ii) are considered.
Composite devices
PV cells can power conventional electrolysers to achieve hydrogen production from water using solar energy as the only source of energy for driving the endothermic water splitting reaction. The most developed technology is the solar-powered electrolysis with PV/electrolyser arrangement. Most PV cells are based on silicon, with typical solar to electrical conversion efficiency of 15 %. The electrical to chemical (hydrogen) conversion efficiency of the best electrolysers reach up to 75 % . Thus, combining a commercial a-Si multijunction PV module with an efficiency of 12 % with a water electrolysis unit operating with an energy conversion efficiency of 65 % (input voltage of 1.9 V) results in a STH conversion efficiency of 7.8 % or lower (taking into account resistance and other system losses). In order to obtain higher efficiencies optimized PV and electrolyser technologies should be used. Recently, a STH efficiency of 12.4% was reported for the champion PV and electrolyser system.
Although PV powered electrolysis systems can reach respectively high STH conversion efficiencies in excess of 10%, this technology is expensive because it combines two separate devices that cannot be integrated together, complicating the system and increasing the cost, and high efficiency electrolysers are very expensive because they employ some of the rarest materials on Earth, Pt and Ir. Moreover, combination of PV and electrolysers is less efficient because the photocurrent must be collected, which leads to significant ohmic losses. Furthermore, in standard solar radiation conditions (AM1.5G) four Si-based PV cells connected in series are necessary to generate the required voltage for water splitting. Unfavorable conditions such as partial shading, haze or cloudiness degrade the power output of these cells below the power input requirement of the electrolyser, which may interrupt the electrolysis process unless sufficient margins are taken into account, reducing the average efficiency and increasing system costs.
Single devices
Most photoelectrodes require extra bias from an external power source to split water. Concerning biased systems, there are chemically biased photoelectrolysis cells and electrically biased photoelectrolysis cells in tandem configurations. In the former case, the chemical bias is achieved using two different electrolytes (e.g. acid and basic electrolytes) placed in two separated half-cells with a salt bridge for the mobile ions. This configuration is not self-sufficient, relying not only on solar energy but also on additional chemical input to stabilize the electrolyte solutions66,68. In the tandem configuration, the PV cell is normally characterized by layered stacked or hybrid structures involving several different semiconductor films placed on top of each other in order to provide the necessary voltage for water splitting (typically about 1.8 V). There are important advantages in this configuration over PV powered electrolysis systems. First and foremost, PV powered PEC cells for water splitting can be made more affordable by combining the two systems, PV and PEC, into one monolithic cell. Second, resistance (Ohmic) losses can be reduced, taking advantage of the low current densities of the PEC cells, relative to electrolysers, and the fact that the interconnection between the PV and PEC components can be made very short (on the order of micrometers rather than meters). Third, the PV + PEC combination is a multijunction cell that can, in principle, surpass the efficiency of single junction PV cells. Fourth, a part of the voltage that is required to split water is provided by the photovoltage generated in the photoelectrode of the PEC cell, therefore the voltage provided by the PV cells is smaller compared to the PV + electrolyser configuration. For instance, a single DSC in tandem with a PEC cell can split water. DSC are also less sensitive to non-optimal illumination conditions such as shade, haze and clouds and therefore DSC + PEC tandem cells could operate also in non-optimal conditions, extending the overall operation time of the system and reducing the extra margins needed for the Si + electrolyser arrangement.
Structures of tandem devices can be subdivided into the following configurations: i) PV/PEC; ii) PV/PV68 and; iii) PEC/PEC. The use of PV/PEC systems has an advantage over PV/PV systems because the PEC face (layer) can replace the face conductor grids that partially obscure the PV layer. Consequently, PEC panels are able to replace costly components and improve photon capture of the PV layer. Recently a new multiphoton combination of a PEC cell in tandem with two DSC was proposed. The authors found that the “trivial” tandem architecture (hematite/squaraine dye/black dye) yields the highest water splitting rate. The maximum STH conversion efficiency is estimated to be 1.36 %.
PEC devices with no additional bias represent a prospective pathway for overcome the complexity of biased systems. Unbiased PEC devices comprise single and multiple photo-system arrangements. The possible arrangements of single photo-system are: i) n-type semiconductor photoelectrode and a metal counter-electrode; ii) p-type semiconductor photoelectrode and a metal counter-electrode; iii) monolithic-bipolar system and a layered metal counter-electrode. PEC cells based on a single semiconductor electrolyte junction employ two photons to produce one hydrogen molecule. Materials having a band gap over 3 eV whose CB and VB straddle the redox potentials required for water reduction and oxidation, such as SrTiO3 and TiO2 (anantase), are able in principle to accomplish the water photolysis without assistance from a bias. However, even in this case, a bias is required to reach the plateau of the photocurrent which remains low due to the lack of visible light response. Hence overall STH conversion efficiencies remain well below 1 %.
Multiple photo-systems may result in a high efficiency for water splitting without an additional bias. The possible arrangements of multi photo-systems are: i) n- and p-type semiconductors (acting as photoanodes and photocathodes, respectively), wired or linked by an ohmic contact; and ii) hybrid systems with layered structures involving several different semiconductor films stacked on top of each other. All strategies share the feature of having two semiconductors with different band gaps. This provides a mechanism by which a single electron is photoexcited twice and, correspondingly, generates a larger bias from light. It has been calculated74 that this type of system could realistically achieve a STH conversion efficiency of 21.6 %, well above NanoPEC’s targeted efficiency of 10%.
Various combinations of n-type and p-type semiconductors such as n-TiO2/p-GaP, n-SrTiO3/p-GaP, n-Fe2O3/p -Fe2O3 have been used to eliminate the bias need for water splitting. Because of the low performance of the individual electrodes in these dual-photoelectrode devices, the resulting overall efficiency is low. NanoPEC’s present work on Fe2O3, WO3 and Cu2O as well as other semiconductors suggest the possibility for an efficient and inexpensive embodiment of a dual photoelectrode device. With photocurrents approaching 4 mA cm-2 with these inexpensive electrodes (corresponding to a 6 % efficiency), the prospect of unbiased water splitting becomes realistic. In our present system, good performance is still achieved with the Cu2O electrode held at 0.4 VRHE and with the Fe2O3 and WO3 electrodes held at 1.0 VRHE68. Therefore only a bias of 0.6 V is needed and the bias will be decreased as better catalysts are identified.
Potential Impact:
Significant progress in PEC water splitting was achieved by NanoPEC by the end of the project (December 2011). In particular, small-size Cu2O photoelectrodes (< 1 cm2) exhibited a photocurrent as high as 7.6 mA cm-2 at the reversible hydrogen evolution potential (0 V vs. RHE). This enables a STH conversion efficiency of 10% in an ideal tandem configuration, exceeding the key technology target of our project. Importantly, the inherent instability of Cu2O was overcome by newly-developed protective overlayers. A large electrode of Cu2O (63 cm2) exhibited a photocurrent of 4 mA cm-2, which corresponded to 87% of the target. On the other hand, small-size Fe2O3 showed a photocurrent of 3.3 mA cm-2 at the reversible oxygen evolution potential (1.23 V vs. RHE). This enables a STH efficiency of 5% in an ideal tandem configuration, the highest record ever reported for stable oxide photoanodes. The Fe2O3 photoanodes did not show any signs of performance degradation for at least three days. It is noted, though, that due to difficulties in scaling-up from small to large area Fe2O3 photoanodes the performance of the later was significantly lower than the champion small-size Fe2O3 photoanode.
The photoelectrodes developed by NanoPEC were used for PEC water splitting in different configurations: one using an external bias to assist water photoelectrolysis; and the other with a new multiphoton combination of a PEC cell in tandem with two DSC. The DSC provide the excess potential, on top of the photovoltage produced at the photoelectrode, necessary to split water and produce H2 and O2 using solar energy as the only energy source. An outdoor demonstrator comprising a PEC reactor, two reservoirs and a Ritter flow meter in the backside of the PEC cells was also developed for practical demonstration of our achievements. The PEC cell contains a photoelectrode, a platinized titanium grid counter electrode (Davis-K, USA) and a Nafion membrane for separating the electrodes (Nafion 424). All these components were immersed in an aqueous electrolyte. The outdoor demonstrator worked continuously for three consecutive months, during which its operation was constantly monitored.
The PEC + PV tandem approach for solar water splitting is a viable way for energy storage through the conversion of solar energy to chemical products. A techno-economic analysis based on published data is reported in the following to give a frame of comparison for the estimated PEC present cost. The present scenario for H2 production presents two primary methods: “thermal reforming of natural gas” and “chemical electrolysis”. Steam methane reforming (SMR), used for decades as the main production way, produces hydrogen rich gas that must be subsequently purified. Moreover, this process has to be combined with CO2 capture and sequestration. The cost with existing technologies is 2.5 $/kg and will range between 1.8-2.8 €/kg in 2030-2050 time frame [Fuel Cells & Hydrogen – SETIS.ec.europa.eu 2011]. The use of fossil fuels does not include external costs related to their environmental footprint. Electrolysis is a well-established technology, however on large-scale plants, efficiency is rather limited (35-40%). Mainly small electrolysers are used in distributed generation. In recent years materials improvements resulted in improved efficiency (80-85%). The cost of H2 produced by direct electrolysis is still high (around 10-20 €/kg) but it is expected to get lower over time. A McKinsey report (2011) suggests a European target of 9.9 €/kg for 2015 [Renewable Energy Focus, ITM Power, UK]. The current estimated cost of hydrogen production via electrolysis in Japan is 9.3 €/kg, while it is expected to go down to 7.5 €/kg for 2015-2020 [Fuel Cells & Hydrogen – SETIS.ec.europa.eu 2011]. The European Hydrogen Roadmap HYWAYS (2008), a study developed under the 6th Framework Program and mainly devoted to hydrogen in automotive applications, evaluated the target costs combining socio-techno-economic analyses to evaluate scenarios for future sustainable hydrogen systems. These studies took into account also storage and transformation problems. The results in terms of hydrogen production were, respectively 4 €/kg in 2020 and 3 €/kg in 2030. Including fuel cells results in additional costs of the order of 100 and 50 €/kW, in 2020 and 2030 respectively. Including storage adds another 10 €/kWh in 2020 and 5 €/kWh in 2030. The Independent Review Panel Report (2009) by NREL-USA, titled: “Hydrogen Production Cost Estimate Using Water Electrolysis”, concluded that in USA the electrolysis H2 production costs 3.3 $/kg, plus1.9 $/kg for compression, storage and dispensing, so that the base estimated cost is 5.2 $/kg. This value depends on the electricity cost in the USA (0.053 $/kWh). A detailed analysis for PEC hydrogen production was carried out by a DOE Final Report (2009) titled Technoeconomic Analysis of PEC H2 Production, by B.D. James, G.N. Baum, J. Perez, K.N. Baum. This document illustrates different production systems both made by colloidal suspensions and by PEC arrays, the latter being either planar or concentrated. The system we intend to refer to is Type 3 – PEC fixed planar array tilted towards the sun, using multi-junction cells immersed in an electrolyte reservoir. The following bar chart reports the evaluated production costs in relation to the efficiency, cell cost and lifetime. Actually, this technology refers to high-tech deposition technologies for planar semiconductor material, which is an especially costly way to produce photo-electrodes. Therefore, the following data have to be considered as the present most expensive solutions. This techno-economic analysis shows that there is still an uncertainty bar in the estimation of the solar hydrogen cost as it is a technology in continuous progress. Therefore, in the following, estimated costs for a real PEC solar-hydrogen-generator are reported, by considering the production capability (kg h-1), the lifetime and the material cost for a photo-electrode of 1 m2. Note that the solar generator is active for 8 h per day and it is supposed to last for 20 years.
PEC Solar H2 Generator
H2 production: 6 x10 -4 kg/h à x8x365x20 = 35 kg/20 years
Unitary panel cost: 200-600 €/m2
Production cost: 6-17 €/kg
This value does not include the capture and storage costs.
The productivity of these integrated solar systems based on PEC technology could be devoted to feed small fuel cells (portable and small stationary systems) provided that a capture and storage as an intermediate system will be designed. Other exploitable results
Apart from the achievements directly intended for PEC solar water splitting, other exploitable results and potential new applications in different fields such as solar cells, (photo)electrochemistry, and material science may emerge from the new experimental techniques and nanostructuring approaches developed in this project:
* advanced host-buffer-guest mesostructure to maximize charge collection and minimize interfacial recombination
* isomorphic template thin films for crystallinity improvement
* silica coating to preserve nanostructures during thermal treatments
* plasmonic light harvest to enhance light absorption
* protective overlayers to prevent corrosion
* novel preparation techniques of high quality compact semiconductor films applicable for solar cells and photocatalysis
* in-situ optical transmission monitoring of semiconductors during oxidation/nitridation
* low-light intensity quantum efficiency measurements as a recombination probe for semiconductors
* resonant light trapping method to control light absorption by ultrathin films
* analytical methods for (photo)electrochemistry by the use of redox couples and sacrificial electron/hole scavengers.
* new PEC module configurations for harvesting the sun light
* unsteady state phenomenological modeling to analyze and simulate PEC phenomena. The innovative approach for light trapping in ultrathin films has been mentioned elsewhere in the report and is presently in the patentin process. On top of this result, the use of overlayers deposited by ALD to form p/n junctions is most probably patentable. At the time of the paper published in Nature Materials (available online since May 8th, 2011), the p/n nature of the overlayer was unclear and unstated in that manuscript. A publication was recently submitted (Apr 28th, 2012) in which the p/n nature of the overlayer is definitive. Once this manuscript is published, the right to patent in Europe will be lost. However, in the United States, there is one year to patent after the publication. The only issue is whether or not the Nature Materials paper constitutes a disclosure of the invention, although as statde above, no mention of a p/n junction is made therein. At the time of the writing of the present report, no other patenting procedures on the above results has been initiated yet by the respective consortium entities. The cost of patenting a result such as the ones listed above ranges from 10k€ to 100 k€, depending on the number of countries in which the patent is be valid or on the related translation requirements, for example. A quantitatively precise cost-benefit analysis is rather difficult to provide, since the expected financial return depends on the conditions in which a company would intend to translate the technology onto the market. The contacts between consortium members and potentially interested companies have not yet gone far enough, at the time of the writing of the present report, to provide a definitive cost-benefit assessment.
Diffusion of project results, dissemination and impact
Results of the NanoPEC project have been disseminated at various levels, namely towards the scientific community, industrial companies and the broad public.
NanoPEC partners have published a large number of articles in scientific peer-reviewed publications. The list and references of all published papers is available on the project website. Whenever possible, NanoPEC publications have been posted on the Open Access system at http://infoscience.epfl.ch.
While a detailed evaluation of the impact of the dissemination actions undertaken by the NanoPEC consortium is very difficult to establish, the latter believes that it has been able to bring considerable attention to PEC-based hydrogen production, by presenting the amazing progress accomplished in terms of overall efficiency, along with the advantage of working with widely available, affordable and non-toxic materials.
Scientific community
Over three years, Nano PEC partners has submitted and/or published 39 international peer-reviewed papers with NanoPEC acknowledgement and 19 proceedings papers. Three book chapters were also published based on NanoPEC results.
Moreover, a total of 40 presentations in conferences and workshops were given by NanoPEC consortium partners from 2009 to 2011.
NanoPEC consortium organized two international workshops in order to increase dissemination of the obtained results and to promote exchange with the scientific community:
* ESF-FWF-LFUI Conference on Nanotechnology for Sustainable Energy, held in Obergurgl (Austria), from July 4th to 9th, 2010. Prof. Michael Grätzel (EPFL) was Conference Chair. Several NanoPEC partners have participated and presented their results. 19 invited speakers, 26 short talks, and 63 accepted posters made up the program.
* EMRS-MRS 2011 Bilateral Energy Conference took place in Nice, France from May 9th to 12th, 2011. The workshop was organized as a symposium within the Conference (Symposium T), named “Materials for solar hydrogen via photo-electrochemical production”. The chairmen were NanoPEC consortium members. The main aim of Symposium T was the dissemination of NanoPEC results obtained after two years and a half of activity. Indeed, NanoPEC partners have prepared 18 oral presentations and 4 posters. Another aim of the Symposium T was to collect new ideas from the scientific international community. As a result, a total of 48 oral presentations and 22 posters were made part of the symposium program. The proceedings of Symposium T will be published in a special issue of “Energy Procedia” edited by the well-known scientific publisher Elsevier. 18 papers have been collected. A peer-review process was put in place with the contribution of NanoPEC consortium researchers and members of the scientific committee. The proceedings are expected to be published in the spring of 2012.
Finally, ranging from 2009 to 2011, some researchers of the NanoPEC consortium have actively participated in several meetings of the IEA-HIA Annex 26 program in the USA, as well as in the U.S. DoE PEC Working Group meetings.
Industrial companies
NanoPEC consortium has established an Industrial Review Panel (IRP), which convened one entire morning during the last General Assembly, on December 16th, 2011 on Eni S.p.A. premises in Novara (Instituto Donegani). The goal of the IRP is in particular to provide potential direct outlets of our scientific findings to industry while also providing a feedback mechanism to ensure that we are utilizing synthetic techniques that have good prospects for industrial scale-up.
The companies stemming from Switzerland, France and Italy, that agreed to participate were: Fiat, PSA Peugeot-Citroën, Total, Solaronix, Solvay, Belenos (Swatch Group), Granit Technologies, HySyTech, Hydro2Power and, of course, Eni S.p.A. A representative of the Swiss electric utilities also attended. The discussions of the IRP focused on:
1) an evaluation of PEC technology towards hydrogen production;
2) NanoPEC innovative potential and IP situation;
3) hydrogen future applications and problems.
A demonstration of hydrogen production with NanoPEC-developed devices has been presented to the attendees.
Broad public
ENI Educational Department prepared a brief new movie describing a water splitting experiment carried out in the laboratory of Istituto ENI-Donegani in Novara (Italy). While the movie is in Italian, written comments in English as movie subtitles have been added during summer 2011. The movie will be made available on NanoPEC website and is presently available at http://www.eniscuola.it.
Another important event in terms of the outreach activities took place in the frame of the collaboration established with the Museum of Science and Technology, in Milano (Italy). During two weekends in March and April 2011, the Museum staff organized four disclosing meetings for visitors on solar energy, nanotechnologies and water splitting for solar hydrogen. At these events, two researchers from ENI-Istituto Donegani were available in the Museum to present a demonstrative apparatus for water splitting and to speak about NanoPEC objectives. These initiatives had a large success, with about 50 attendants per day (4 times). An advertisement poster was produced specifically, while a picture gallery is available at http://nanopec.epfl.ch/events/.
Finally, NanoPEC consortium members have been interviewed in various broad-public magazines, as well as on national radio programs. Recently, a large article on NanoPEC appeared in « Projects Magazine », one of Europe's leading science and technology research magazine, covering the latest innovations, research projects and breakthroughs from across Europe and broadly distributed in institutional and R & D circles.
List of Websites:
The project website is available at http://nanopec.epfl.ch. It not only contains general information on hydrogen production by way of photo-electrochemical water splitting, but also:
1) Presentation of all the consortium partners, with links to their respective websites.
2) List of all project-related publications, with authors, abstracts and references. Papers for which open access has been granted can be downloaded in pdf;
3) Contact addresses.
Contacts :
1) Prof. Michael Grätzel – Head, Laboratory for Photonics and Interfaces
Ecole Polytechnique Fédérale de Lausanne
Station 6
1015 Lausanne
Switzerland
michael.graetzel@epfl.ch
2) Prof. Hans Björn Püttgen – Director, Energy Center
Ecole Polytechnique Fédérale de Lausanne
Station 5 – Château En Bassenges
1015 Lausanne
Switzerland
hans.puttgen@epfl.ch