Metal Oxides Applied to Thermochemical Water-Splitting for Hydrogen Production Using Concentrated Solar Energy

Abstract

Solar thermochemical processes have the potential to efficiently convert high-temperature solar heat into storable and transportable chemical fuels such as hydrogen. In such processes, the thermal energy required for the endothermic reaction is supplied by concentrated solar energy and the hydrogen production routes differ as a function of the feedstock resource. While hydrogen production should still rely on carbonaceous feedstocks in a transition period, thermochemical water-splitting using metal oxide redox reactions is considered to date as one of the most attractive methods in the long-term to produce renewable H₂ for direct use in fuel cells or further conversion to synthetic liquid hydrocarbon fuels. The two-step redox cycles generally consist of the endothermic solar thermal reduction of a metal oxide releasing oxygen with concentrated solar energy used as the high-temperature heat source for providing reaction enthalpy; and the exothermic oxidation of the reduced oxide with H₂O to generate H₂. This approach requires the development of redox-active and thermally-stable oxide materials able to split water with both high fuel productivities and chemical conversion rates. The main relevant two-step metal oxide systems are commonly based on volatile (ZnO/Zn, SnO₂/SnO) and non-volatile redox pairs (Fe₃O₄/FeO, ferrites, CeO₂/CeO_2−δ, perovskites). These promising hydrogen production cycles are described by providing an overview of the best performing redox systems, with special focus on their capabilities to produce solar hydrogen with high yields, rapid reaction rates, and thermochemical performance stability, and on the solar reactor technologies developed to operate the solid–gas reaction systems.

1. Introduction

Solar thermochemical processes are efficient routes for converting high temperature solar heat into valuable and sustainable chemical energy carriers (solar fuels). Thermochemical water-splitting cycles consist of the thermal conversion of water into separate streams of hydrogen and oxygen via a series of endothermic and exothermic chemical reactions. It is particularly attractive due to the abundance, availability and low cost of the water feedstock. Endothermic reactions can be driven by high-temperature concentrated solar thermal energy [1,2,3,4]. A thermochemical water-splitting cycle operates at much lower working temperatures than direct water thermolysis while resulting in the same global water decomposition reaction.

The interest in thermochemical water-splitting cycles boomed in the late 70s and early 80s with the oil crisis [5,6,7,8,9], and most of the cycles were proposed for being combined with a primary nuclear energy source, thereby imposing constraints on the operating temperature that should remain below 900 °C. Although several hundred cycles have been proposed, only a few has been subjected to substantial research, and the demonstration of technical feasibility and potential for high efficiency has been barely carried out. In previous studies dealing with screening and evaluation of thermochemical cycles, the maximum cycle temperature level corresponded to the deemed optimum for an advanced high-temperature Generation IV nuclear reactor (e.g. high-temperature gas-cooled reactor HTGR, very-high-temperature reactor VHTR), i.e. about 850 °C [10]. Processes involving higher temperatures were eliminated. A large research effort has mainly been focused on UT-3 and Iodine-Sulphur (I-S) cycles [10,11,12,13,14,15,16], in which the primary energy input was the high-temperature heat released from an advanced nuclear reactor (4th generation power station). These cycles have also been proposed in combination with a solar energy source [17,18,19,20,21,22]. The medium temperature I-S and hybrid sulphur (Westinghouse) cycles that may use either solar or nuclear heat have been examined (e.g., EU-projects HYTHEC (FP6), HycycleS (FP7) and SOL2HY2 (FCH JU)). In particular, the solar decomposition of H₂SO₄ above 850 °C involved in both cycles was studied [17]. Regarding the hybrid sulphur cycle (Ispra Mark 11), sulphuric acid is decomposed in the first reaction at a high temperature to generate sulphur dioxide and oxygen (that can be separated and valorized as byproduct). Sulphur dioxide is then electrolyzed with water at about 80 °C in the second reaction forming hydrogen and fresh sulphuric acid to be recycled back to the first reaction. This electrolysis step requires only about one tenth of the electrical power needed for conventional water electrolysis, thereby reducing the global process energy demand for hydrogen production, which is crucial for future industrial technology commercialization. However, such complex cycles operating below 1000 °C usually involve incomplete reactions or electrochemical steps (hybrid cycles), additional separation steps, hazardous or corrosive reactants and/or products, which results in materials issues for reactor construction and may compromise viable commercial process implementation.

Significant progress has recently been achieved regarding solar energy collectors and concentrators while enabling to reach solar powers of several dozens of megawatts. In this context, more efficient cycles with a lower number of steps should now be suited for operation in a broader range of temperature from 1000 °C to 2000 °C. Short two-step solar-driven cycles are particularly attractive owed to their operational simplicity favoring easier process scale implementation, thus offering the potential for safe, low-cost, and large-scale production of H₂ as a clean energy carrier [23,24]. The cycles based on non-toxic metal oxides [25,26,27,28,29] using process heat at temperature above 1000 °C are the simplest and most promising routes for solar H₂ production.

The solar thermochemistry governing two-step water-splitting cycles involves metal oxide redox reactions. In the first step driven by concentrated solar energy, the reduction of the metal oxide leads to both oxygen release and material activation. In the second water-splitting step, oxidation of the redox-active material (reduced valence state of the metal oxide) with water steam produces hydrogen (Figure 1). A few redox cycles have been investigated for solar hydrogen production from water-splitting. The generic two-step process that makes use of metal oxides redox pairs is based on the following reactions:

High temperature reduction step (endothermic):  M_xO_y → M_xO_y−1 + ½ O₂

(1)

Hydrolysis step (exothermic):  M_xO_y−1 + H₂O → M_xO_y + H₂

(2)

Such a cycling process presents noticeable advantages: (i) the upper cycle temperature (generally in the range of 1200–1600 °C) is compatible with renewable concentrated solar energy; (ii) water and heat are the only process inputs, hydrogen and oxygen the only chemical outputs; (iii) the produced H₂ and O₂ streams are inherently separated by the different reactions; (iv) the other chemicals and reactants are continually recycled in the closed cycle; (v) the produced H₂ is pure for being directly processed, for instance using a polymer electrolyte membrane fuel cell (PEMFC).

Conventional electrolysis, representing the state-of-the-art technology, also generates pure H₂, but this approach is limited by thermodynamic inefficiencies and has low overall energy conversion efficiency as it requires first the generation of electricity (global energy conversion efficiency below 20% when taking into account the electricity production efficiency). Thermochemical water-splitting cycles are expected to reach higher efficiency than water electrolysis since their energy efficiency is not impaired by the intermediate conversion of heat to electricity. Benchmark is the alkaline water electrolysis using solar electricity generated by either photovoltaics (PV) or concentrating solar power (CSP). With currently available technologies, the overall solar-to-hydrogen energy conversion efficiency reaches 10–14%, when assuming 70% efficiency for electrolysis and 15% (PV) to 20% (CSP) annual efficiency for solar-produced electricity [30]. On the other hand, direct thermal splitting of water would require temperatures exceeding 2000–2500 °C to yield significant amounts of hydrogen. Therefore, thermochemical cycles are proposed to lower the maximum process temperature to a technical manageable level.

The cycles working in the temperature range of 900–2000 °C were not considered in previous screening studies [10], because they were not practically compatible with the operating temperature of advanced nuclear reactors. In contrast, they can be plainly considered for a coupling with concentrating solar thermal as primary energy source. Therefore, a new screening, selection, and evaluation of the most promising solar-driven thermochemical cycles for H₂ production was conducted [28]. A database including over 280 referenced thermochemical cycles was constructed at CNRS-PROMES laboratory (France) from a detailed bibliographic survey to sort out the cycles previously proposed in the literature. The developed database incorporates the detailed information on the processing conditions, involved chemical reaction steps, bibliographic references data describing or referring to the cycles. The numerous listed cycles can be classified into a restricted number of defined subcategories (sulphur, iodine, chlorine, bromine, carbon, miscellaneous, hybrid). Objective screening criteria were defined and applied to decrease the number of investigated cycles to a manageable number and to identify potentially applicable solar-driven cycles.

Roughly 30 interesting cycles were selected for further investigations by using a set of defined criteria: (1) upper temperature of the cycle for being compatible with solar concentrating systems, (2) number of reactions and separation steps, (3) number of chemical elements, (4) nature of the cycle (either purely thermal or hybrid thermo-electrochemical), (5) thermodynamics, (6) technical operational feasibility of the cycle accounting for reaction kinetics and chemical conversion yields, (7) expected cycle exergy efficiency, (8) availability and cost of process chemicals and reactants, (9) corrosiveness, (10) presence of non-stationary solid reactants, (11) environmental safety and health issues, (12) type/availability/cost and ease for implementing the separation steps. Chiefly, two and three-step cycles were retained due to their ease of process implementation and integration that should imply favorable process economics. In addition, these cycles basically involve low number of reactions and separation steps (thereby lowering penalties associated with heat transfer limitations and products separation), available, low-cost, and safe (non-toxic) materials, and a heat input temperature (900–2000 °C) compatible with concentrated solar energy.

The cycles that were eliminated did not respond to the above criteria, as it was the case for those requiring very high temperatures (higher than 2200 °C) (MoO₂/Mo, SiO₂/SiO, WO₃/W), hybrid cycles involving an electrochemical step (Cu-Cl cycle [31,32]), cycles based on toxic (Cd, Hg, or bromide) or corrosive compounds such as potassium hydroxide, cycles that necessitate advanced gas separation techniques (such as membrane technologies), or cycles for which thermodynamics is not favorable such as most cycles based on sulphates [33,34] or chlorine [35,36,37,38]. In addition, a method of exergy analysis was developed to determine the exergy requirements and the global exergy efficiency of each selected thermochemical cycle [28]. This method was used to compare the cycles and to identify the main operations responsible for irreversibilities, which is a thermodynamic tool for future process optimization. Energy efficiency analyses related to process implementation in large-scale solar tower chemical plants were conducted [29] to estimate the global solar-to-hydrogen energy conversion efficiency that was about 20% for iron and zinc oxide cycles using currently available data (including the different heat losses taking place in the main process sub-systems comprising solar concentrating system for solar energy collection, solar receiver and reactor, and chemical cycling process).

The research investigations regarding high-temperature cycles were mainly focused on CdO/Cd [39], ZnO/Zn [40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76], SnO₂/SnO [76,77,78,79,80,81,82], Mn₂O₃/MnO [83], CeO₂/Ce₂O₃ [84,85,86], and Fe₃O₄/FeO [87,88,89,90,91,92,93,94,95]. The utilization of mixed metal oxides (e.g. Zn-ferrite, Mn-ferrite, or Ni-ferrite cycles [96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127]) decreases the temperature of the solar activation step (O₂ release) while the reduced oxide still remains active for the H₂ generation step. Due to their promising growing interest, increasing research activities devoted to the investigation of solar-driven two-step cycles (ZnO/Zn, SnO₂/SnO, Mn₂O₃/MnO, Fe₃O₄/FeO, ferrites, CeO₂/CeO_2−δ, perovskites) have been conducted mainly in Switzerland (ETHZ, PSI [40]), Germany (DLR [122]), France (CNRS-PROMES [28]), USA (DOE projects [53,54]) and Japan (Tokyo Institute of Technology [97] and Niigata University [114]).

The current studies are mainly related to thermochemical systems based on metal oxide redox pairs as chemical intermediates. The two-step cycles commonly rely on volatile (ZnO/Zn, SnO₂/SnO) or non-volatile redox pairs (Fe₃O₄/FeO, CeO₂/CeO_2−δ, and mixed oxides). The products of the high-temperature reduction reaction are in the gaseous state for the volatile oxide cycle category, whereas redox reactions proceed in the condensed state for the non-volatile oxides.

2. Volatile Metal Oxide Cycles

2.1. ZnO/Zn Cycle

The ZnO/Zn cycle is classified as a volatile oxide cycle because the reduced zinc metal species produced during the decomposition reaction (ZnO(s) → Zn(g) + ½O₂) is released from the reactor as a gaseous product. This system has been studied extensively and it is considered as one of the most attractive candidate cycles for being coupled with a solar energy source at high temperature. ZnO can be dissociated near 1800 °C in a solar reactor [58] and Zn is recovered at the reactor exit after rapid quenching the gaseous products. The reported activation energies of the ZnO dissociation varied in the range 312–376 kJ/mol [43,44]. The products recombination between Zn and O₂ represents a parasitic reverse reaction that limits the Zn yield from the solar step [49]. Above the Zn condensation temperature, Zn recombination is controlled by diffusion of Zn(g) and O₂ to the reactor walls [62]. The condensation of Zn vapor in the presence of O₂ was studied by fractional crystallization in a temperature-gradient tube furnace, which revealed that Zn oxidation is predominantly a heterogeneous process and, in the absence of nucleation sites, Zn(g) and O₂ can coexist in a metastable state [42]. Otherwise, quenching the products is necessary in order to avoid Zn re-oxidation, which brings process irreversibility (energy losses) and may further introduce complexity issues during large-scale application. The gas quenching aiming to alleviate Zn recombination issue (e.g., via the addition of large amounts of inert gas) constitutes the main challenge of this redox cycle. A fast products quenching to cool the gas species down to below 900 °C (that corresponds to the Zn(g) condensation temperature) is advocated [41,61]. The quenching efficiency is sensitive to the dilution ratio of Zn(g) in a flow of inert gas [41]. Alternatively, an effective technique for in-situ Zn/O₂ separation directly at high temperatures may also be used to avoid recombination (electrothermal separation methods [128,129]).

The solar thermochemical reactor represents the main key component of solar processes involving water-splitting cycles. A specific research effort must be thus dedicated to suitable reactor design and optimization. The materials used for both the window and the cavity receiver must be carefully chosen as they represent the most sensitive components of solar reactors that are commonly based on either direct or indirect heating of the reactants using concentrated solar energy. Direct heating provides efficient and rapid heat transfer directly to the reacting matter but such reactor technologies must be designed so that to avoid particle deposition on the optical window (case of entrained particles and particle-laden flow). Alternatively, indirect heating via a heat transfer wall requires using high-temperature resistant refractory materials and may suffer from additional heat losses due to indirect heat transfer. Conventional solar reactor designs usually make use of insulated cavity-type blackbody receivers that allow obtaining almost isothermal conditions in the cavity volume and high solar energy absorption efficiencies [57,58,59,60,61]. High-temperature resistant ceramic materials (refractory) are usually employed for lining the inner reactor volume (including both cavity walls and insulation materials).

Different solar reactor configurations designed for ZnO thermal dissociation have been explored in the literature. The first simple type is based on the direct irradiation of a ZnO rod [44]. Such a reactor involves a packed bed or a moving rod of ZnO placed behind a window at the focus of a solar concentrating system. A number of drawbacks arise while operating such a process. First, only the ZnO irradiated surface is solar heated, whereas the underlying ZnO remains at relatively low temperatures, thus inducing radiative heat losses at the surface and temperature gradient through the layer. The reactant surface area exposed to solar flux is weak when compared with a dispersed particulate system, which results in relatively low reaction rates. High re-radiation losses also occur because the reaction does not proceed at fast enough rate to absorb most of the incident solar energy. This in turn reduces the process energy efficiency. Second, the necessary presence of a window requires using a screening stream of inert gas at its surface, in order to both cool it and protect it from products deposition. Any material or dust deposition on the window may result in opaque absorbing regions and subsequent localized overheating and thermal stress, with a possible window melting. Another developed reactor configuration consists of the rotating conical cavity receiver (“ROCA”). A 10 kW prototype and 100 kW pilot were constructed and tested in a high-flux solar furnace [57,58,59,60,61]. The design was based on a windowed cavity receiver that rotates at a high angular velocity. The cavity walls were made of a refractory material (either ZrO₂, HfO₂, or ZnO itself). ZnO was fed in batch pulses by a screw feeder, and the cavity rotation evenly dispersed the ZnO particles throughout the reactor on the cavity walls. Incident solar radiation was absorbed through the window to heat the ZnO particles, and products were swept by inert carrier gas into a cooling zone for rapid quenching. This reactor concept is promising, but may face difficulties and barriers during scale-up. The limitations imposed by the window that was protected by both water-cooling and inert gas purging, could result in a significant decrease of process efficiency due to both conductive and convective heat losses, and energy requirements for inert gas recycling. The cavity rotation further increases mechanical complexity to the high-temperature chemical process, which presents significant engineering challenges and capital costs issues upon scaling-up. Similar to other systems in which a layer of ZnO is directly heated, the lack of particle separation, mixing, and agitation could hinder the kinetics due to heat and mass transfer limitations. In light of the challenges to be faced by these still promising reactor prototype technologies, the use of rapid reaction aerosol flow reactors was considered to obtain rapid ZnO decomposition kinetics while still being scalable with existing chemical engineering technology [53,54,55,56]. The small size of particles yields greater surface area for enhancing heat and mass transfer rates, and the increase of this available area enhances the overall rate of surface processes. Such particles dispersed in a gas flow have been shown to reach extremely high heating rates, on the order of 10⁵ K s⁻¹, when exposed to a nearby high temperature hot emitting surface because of their ability to efficiently absorb the radiative heat flux from the surrounding hot walls. The increased heat and mass transfer rates in aerosol systems fasten solid–gas reaction kinetics, lower the global residence time required for converting the particles, and finally make such a solar receiver more efficient. Indirect solar heating of the particles stream by infrared radiation emitted from the surrounding tube walls further eliminates the need for using an optical window. However, low ZnO conversions were obtained (maximum 17%) and particulate products deposition occurred on the walls [54].

The solid metallic Zn powder produced in the solar step may be used in either a fuel cell or battery, or in the exothermal water-splitting reaction to produce pure H₂ and regenerate ZnO that can be recycled back to the solar step. The produced Zn is a solid that can be stored and fed to the hydrolysis reactor during night time or cloudy periods. The H₂ production step of the ZnO/Zn cycle involves a surface hydrolysis reaction between solid Zn particles (or liquid molten Zn) and steam. The Zn hydrolysis was performed and studied with steam bubbling through molten Zn at 500 °C [63], but the reaction rate and Zn conversion were limited by the formation of a ZnO(s) layer around the steam bubbles, thus preventing efficient contact between fresh reactants. The kinetic rates of such a surface reaction are expected to be limited by mass transfer through either the gas boundary layer or the developing ZnO film at the surface. To maximize reaction rates during the off-sun step of the cycle, the use of particles with a high specific surface area is beneficial, thereby reducing the resistance to mass transfer within the particle and through the developing ZnO film.

Most of previous studies focusing on Zn particles hydrolysis reaction have been carried out either using commercial Zn powders [71,72] or Zn produced from solar carbo-thermal ZnO reduction [73]. The Zn conversion achieved in an aerosol flow reactor fed with both water vapor and commercial Zn particles (average particle size of 158 nm) was about 24% at 540 °C with a gas residence time of about 0.6 s [67]. The reaction rate may become negligible because of a low-permeable ZnO layer growing at the particles surface [71]. Non-isothermal thermogravimetric analysis (TGA) with Zn powder dispersed on quartz wool showed that complete Zn conversion could be reached with longer residence times. Regarding Zn obtained from solar carbo-thermal reduction, increasing the temperature of injected steam flow from 200 to 550 °C increased the Zn conversion from 24% to 81% during the fast reaction stage (observed from 400 °C) [73]. Besides, other studies on hydrolysis using an aerosol flow reactor consisted of the intermediate Zn vaporization before steam quenching the Zn vapors to co-produce both H₂ and Zn/ZnO nanoparticles [64,65,66,67,68,69], but this process suffered from weak particle yield at the outlet and significant particle deposition at the wall. The advantages of hydrolyzing Zn nanoparticles are three-fold: (1) their high specific surface area inherently increases heat and mass transfer, and reaction kinetics (2) their large surface-to-volume ratio favors their complete bulk oxidation, and (3) their possible entrainment by a gas flow is suitable for allowing a simple, continuous, and controllable feeding of Zn reactants and removal of ZnO products. The in-situ formation and steam hydrolysis of entrained Zn nanoparticles in an aerosol (size: 70–100 nm) freshly produced by vapo-condensation allowed reaching 70% H₂ yield [64]. Ernst et al. [66] obtained up to 90% H₂ yield with reactor temperatures in the range 627–1000 °C, however at the expense of weak particle yield downstream.

Finally, steam hydrolysis of Zn-rich solar-produced nanopowder (synthesized in a high temperature solar chemical reactor) results in complete Zn particle conversion at relatively low temperatures (360–500 °C) [74]. Figure 2 shows the time course of Zn conversion during steam hydrolysis. This conversion is obtained from the ratio between the amount of converted Zn and the initial amount of Zn in the powder. The initial presence of ZnO clusters in the particles promotes the Zn oxidation reaction. The complete Zn particle conversion during hydrolysis is due to both the synthesis method and the involved mechanisms during the solar step, which allows the formation of Zn-rich nanoparticles containing ZnO clusters dispersed in the bulk. The ZnO serves as nucleation sites promoting further oxidation during the water-splitting step. Therefore, the solar-produced Zn nanoparticles containing a fraction of recombined ZnO show better reaction extent than pure Zn particles.

Assuming an endothermic ZnO decomposition reaction performed at 2000 °C, the energy conversion efficiency of ZnO/Zn cycle reaches about 45% (equal to HHV_H2/ΔH_{ZnO(25 °C)→Zn(g)+0.5O2(2000 °C)}) and the maximum exergy efficiency is 29% without including any heat recovery [40] (defined as the ratio of chemical energy stored in the form of H₂ at ambient temperature to the solar power input, η_exergy = −ΔG^°_{H2+0.5O2→H2O}/Q_solar). Hence, the ZnO/Zn redox system presumably represents one of the most favorable candidate cycles given its potential for achieving both high energy and exergy efficiencies, although significant technical challenges are still remaining and the development of alternative cycles is thus necessary.

2.2. SnO₂/SnO Cycle

More recently, the new SnO₂/SnO redox pair was proposed as an efficient system for two-step water-splitting cycle [77,78] which shares many similar characteristics with the ZnO/Zn system, including: (1) the formation of vapor phase products (volatile oxide) in the solar step, (2) the similar temperature range required for thermal reduction (about 1600–1800 °C at atmospheric pressure), (3) the need for dilution with neutral gas and quenching of the products in the solar step to avoid reverse re-oxidation, and (4) the solar-driven synthesis of reduced SnO particles with nanometric size scale by vapor condensation if a fast enough quenching is applied.

The SnO₂/SnO two-step cycle thus belongs to the class of volatile metal oxide cycles, similarly to ZnO/Zn. The only known cycle based on tin oxides was previously referenced as the “Sn-Souriau” three-step cycle: (i) SnO₂ → SnO + ½O₂; (ii) 2SnO → Sn + SnO₂; (iii) Sn + 2H₂O → SnO₂ + H₂. Nevertheless, the concept of this cycle patented in 1972 [130] was not subjected to any experimental validation to confirm its operational feasibility. Recently, reaction (ii) was proven to reach completion at 600 °C after heating for 10 min under inert gas atmosphere to avoid SnO oxidation. Unfortunately, metallic Sn cannot be separated from SnO₂ by liquefying Sn as advocated in the original “Sn Souriau” cycle, because the molten liquid phase of Sn does not form in the blend. The reaction (iii) generating H₂ from the Sn/SnO₂ mixture issued from disproportionation reaction (ii) is slow and only partial at 600 °C (Sn conversion yield of 45% after 30 min). For the above mentioned reasons, this three-step cycle was not selected as a promising candidate. Instead, the following innovative SnO₂/SnO two-step cycle was proposed and assessed:

Solar reduction step:  SnO₂(s) → SnO(g) + ½ O₂   (endothermic, 1600 °C)

(3)

Hydrolysis step:  SnO(s) + H₂O(g) → SnO₂(s) + H₂  (exothermic, 550 °C)

(4)

The first endothermic solid–gas reaction (ΔH = 557 kJ/mol SnO₂ at 1600 °C) consists of the high temperature thermal reduction of tin(IV) oxide (stannic oxide) into tin(II) oxide (stannous oxide). The required temperature for reaction (3) is suitable for the coupling with a concentrating solar thermal energy source. The SnO product (T_m = 1042 °C, T_b = 1527 °C) is gaseous at the reaction temperature, and SnO(g) is released along with O₂. When the gas temperature drops at the reactor outlet, the gaseous SnO vapors are condensed as nanoparticles via formation of nuclei and particle growth. As for ZnO decomposition, this reaction is also favored by an oxygen partial pressure decrease (via vacuum pumping or inert gas dilution) to limit the reverse recombination reaction with O₂. The beneficial effect of high dilution ratio or reduced pressure operation was demonstrated [49]. A detailed kinetic analysis of the recombination reaction with O₂ was performed. The kinetic parameters were determined for both SnO and Zn recombination (global reaction order of about 1.5 corresponding to the reaction stoichiometry, and activation energies of 42 kJ/mol for SnO and 35 kJ/mol for Zn [49]).

The second solid–gas reaction generating H₂ consists in hydrolyzing SnO nanoparticles by steam water. This moderately exothermic reaction (ΔH = −49 kJ/mol at 500 °C) occurs above 450 °C at atmospheric pressure, with both a satisfactory SnO oxidation rate and a final H₂ yield over 90%. The recovered stannic oxide, SnO₂, is then recycled into the first reaction (solar reduction step), which closes the cycle. The cycle H₂ productivity assuming a complete reactant conversion is 166.3 mL_H2/g_SnO at normal conditions, which equates to about 14.8 mg_H2/g_SnO (i.e., the H₂ mass storage capacity of SnO is 1.48 %). The potential applications of such metal oxide systems are related to the stationary or portable generation of H₂. SnO (or Zn) can be long-term stored and transported more easily and surely in the form of solids than H₂, which represents a potential H₂ storage tank. In addition, the reduced species are stable in air and the reactivity of SnO (or Zn) with steam water is not altered even after a long storage period in ambient air. Therefore, on-demand H₂ can be generated at the delivery site when required.

Initially, the SnO hydrolysis was considered as an intermediate step of metallic Sn hydrolysis [131]. Subsequently, the kinetics for hydrolysis of solar-produced SnO particles (and solar Zn for comparison), consisting of nanoparticles synthesized from solar thermal SnO₂ reduction, was studied in details [79,80,81]. Hydrolysis experiments were carried out to investigate the hydrolysis reaction in a packed-bed reactor and to establish kinetic rate laws. Kinetic parameters encompassing activation energies and reaction orders were determined (122 ± 13 kJ/mol and 2.0 ± 0.3 for SnO, and 87 ± 7 kJ/mol and 3.5 ± 0.5 for Zn, respectively). The reduced SnO species react efficiently in the water-splitting step to generate H₂ in the range of 450–600 °C. The hydrolysis reaction showed a fast initial step at the time of steam injection before the chemical conversion gradually leveled off, which is typical for solid–gas reactions exhibiting a passivating oxide layer growing at the particle surface. The suggested reaction mechanism is composed of a fast reaction-controlled regime during the initial step, followed by a diffusion-controlled regime which increasingly strengthens while the oxide layer is formed at the particle surface. The hydrolysis of solar SnO nanoparticles nearly approached completion (Figure 3), while Zn hydrolysis was fast and reached complete H₂ yield thanks to the nano-sized powders [79]. Further thermogravimetric investigations confirmed SnO conversion approaching 90% at 600 °C [80], but competing reactions such as SnO disproportionation into Sn and SnO₂ occurred concomitantly, as evidenced by Mössbauer spectroscopy [81].

The global flow diagram of the thermochemical process integrating the two chemical cycle steps (both solar reactor and hydrolyser) and the H₂ utilization via fuel cell is shown in Figure 4. The produced H₂ is converted in an ideal fuel cell generating both electrical power (W_FC = −237 kJ/mol at 25 °C) and heat (Q_FC = −49 kJ/mol). The intrinsic energy conversion efficiency of the SnO₂/SnO cycle reaches about 42% accounting for the high heating value of H₂ (HHV_H2 = 286 kJ/mol). The maximum solar absorption efficiency of a perfectly insulated black-body solar receiver is about 86% at T=1600 °C (assuming a concentration ratio C = 5000 and a direct normal irradiation I = 1 kW/m²) [132]. Thus, the amount of solar energy input (Q_solar) necessary to carry out the endothermic SnO₂ reduction reaction is about 795 kJ/mol, which then allows assessing the global process efficiencies of SnO₂/SnO cycle. The resulting global exergy process efficiency reaches about 29.8% at 1600 °C, which is similar to ZnO/Zn cycle at 2000 °C.

A lab-scale solar reactor prototype (Figure 5) was designed, operated, and simulated for application to the solar thermal reduction of volatile metal oxides involved in the first step of thermochemical cycles. This reactor features a windowed rotating cavity receiver, in which dispersed solid reacting particles are continuously injected by a feeding system [46,47]. The volatile generated products are subsequently recovered at the outlet as nanoparticles agglomerates in a downstream ceramic filter for the subsequent hydrolysis step. The oxide particles injected and dispersed in the directly-irradiated cavity under a controlled inert atmosphere serve simultaneously as both reacting species in the solid–gas decomposition reaction and as radiative absorbers enabling homogeneous heating. The direct particles irradiation provides efficient and rapid heat transfer to the reaction site, thus bypassing the limitations imposed by indirect transfer of high-temperature heat via intermediate absorbing opaque walls. The solar ZnO dissociation was successfully performed as a function of the cavity pressure and inert carrier gas flow-rate, and the obtained Zn product yield in the collected powder reached above 90%. The reaction extent (ZnO conversion) was highly dependent on the particle temperature and the particle conversion increased markedly when the particle temperature slightly increased. Likewise, the conversion increased when decreasing the initial diameter of fed particles according to computational fluid dynamics (CFD) simulations of the reactive gas-particle flow. As a result, complete reaction was predicted when the particles temperature exceeds 1927 °C for initial particle diameter of 1 μm.

Another optimised solar reactor, shown in Figure 6, was then designed enabling the continuous measurement and monitoring of several operating parameters (evolution of temperatures, O₂ concentration at the reactor outlet, Figure 7), and the synthesis of significant amounts of solar Zn and SnO nanopowders (active materials) for the exothermal hydrolysis step [48]. Scanning electron microscopy (SEM) characterizations (Figure 8) show that these powders are composed of agglomerates of nanoparticles (10–50 nm) with a high specific surface area (20–60 m²/g). This vertical-axis cavity-type solar reactor (1 kW of thermal power absorbed) effects the reaction above 1600 °C at reduced pressure (about 20 kPa) with the compressed oxide reactant injected at the cavity base by the means of an ascending screw piston. This reactor was also used to determine the kinetics of the ZnO and SnO₂ dissociation reactions by fitting reaction models with experimental data [50] thanks to an inverse method involving the online diagnosis of outlet gas combined with a reactor model coupling heat and mass transport phenomena, radiation, and chemical reaction. The activation energy of the oxide dissociation reaction was determined to be 313 ± 31 kJ/mol for ZnO and 353 ± 18 kJ/mol for SnO₂ [50].

3. Non-Volatile Metal Oxide Cycles

The iron oxide-based (Fe₃O₄/FeO) two-step cycle is particularly attractive because it features simple reactants and chemical steps (in comparison with nuclear-based cycles) and available safe reactants (resulting in lower irreversibility inducing potentially superior cycle energy efficiency). Furthermore, it also involves non-corrosive solid materials involved in solid–gas reactions, and it avoids the drawback of potential reverse recombination reaction with O₂ during gas quenching commonly encountered with volatile metal oxides such as zinc, tin or cadmium oxides [62,88]. A key benefit relies on the fact that non-volatile iron oxide systems proceed with the continuous release and removal of the evolved gaseous O₂ from the condensed reacting phase during the solar reduction step, thus promoting high reduction extents. The Fe₃O₄/FeO redox pair was originally proposed by Nakamura [87]. This two-step cycle proceeds as follows:

Solar reduction step:  Fe₃O₄(l-s) → 3FeO(l-s) + 1/2O₂

(5)

Hydrolysis step:  3FeO(s) + H₂O(g) → Fe₃O₄(s) + H₂

(6)

The first high-temperature solar step (thermal reduction of Fe₃O₄) is highly endothermic (ΔH° = 319.5 kJ/mol), whereas the second low-temperature step (FeO hydrolysis) is slightly exothermic (ΔH°= −33.6 kJ/mol). The amount of solar energy theoretically needed to produce 1 mol of H₂ corresponds to the summation of the sensible energy required to heat 1 mol of Fe₃O₄ from 600 °C to 2100 °C (446.51 kJ), the endothermic reduction enthalpy (242.84 kJ), and the energy required to heat and vaporize water from 25°C to 600 °C (64.9 kJ) [126]. On the basis of these temperature levels and the HHV of H₂ (286 kJ/mol), the theoretical energy conversion efficiency of Fe₃O₄/FeO cycle is 37.1%.

The high-temperature solar thermal reduction of magnetite to wüstite theoretically proceeds at temperatures above 2200 °C under 1 bar (according to Gibbs free energy variation) [133]. In addition to stoichiometric FeO, thermodynamics predicts the formation of non-stoichiometric wüstite phases (Fe_1−yO such as Fe_0.947O) (Figure 9). The thermal reduction of magnetite has been experimentally studied [88,89,90,91,92,93,94] and demonstrated by using a 2 kW (thermal power) parabolic solar concentrator. The reduction of magnetite was only noticeable at temperatures above the Fe₃O₄ melting point (1597 °C). Tofighi et al. [89,90] showed that a 0.8 g sample reached 80% conversion after 5 min at 2000 °C in an Ar atmosphere (the conversion in air only reaches 40%). Because of the high temperature thermal treatment, the Fe₃O₄ reduction occurs concomitantly with a slight material vaporization. Weidenkaff et al. [92] reported that the exothermic hydrolysis reaction of wüstite with steam water is influenced by the non-stoichiometry, structure and morphology of the parent wüstite phases and by the reaction temperature. The water-splitting hydrolysis reaction using FeO is favorable only below 800 °C (when ΔG° < 0) and the theoretical FeO conversion decreases with increasing temperature according to thermodynamics (Figure 10).

Besides, the complete thermal reduction of hematite (Fe₂O₃) into wüstite was achieved in a high-temperature solar reactor operated at 1600 °C and 0.1 bar under inert gas atmosphere [93]. The oxygen-releasing reaction proceeds via the formation of the molten oxide and it requires about 2 min to reach completion, which confirms previously reported results. Further TGA experiments confirmed that wüstite formation starts near the Fe₃O₄ melting point (Figure 11). The reduced oxide is thus recovered as a hardened structure that cannot react with water efficiently and the milling of the solar-reduced material into a fine powder is then necessary to carry out the subsequent hydrolysis step.

A quantitative gas analysis of the amount of H₂ evolved was performed in a continuous way to determine the reaction kinetics and chemical conversion of wüstite hydrolysis. A conversion of 83% was measured regarding the hydrolysis reaction at 575 °C of non-stoichiometric solar-produced wüstite (Fe_(1−y)O) milled into a fine powder (particle size of 30–50 μm) [93]. The hydrolysis kinetics is dependent on the oxide material physical properties, reaction temperature, and particle size (Figure 12). The reaction rate decreases with time because of the formation of a low permeable oxide layer (Fe₃O₄) at the particles surface, which then grows with time and hinders the reaction. The hydrolysis reaction of particles is governed by two successive controlling phenomena: the fast steam oxidation of the external surface of particle followed by steam diffusion inside the particle pores. The maximum initial hydrolysis reaction rates were achieved for solar-produced wüstite, enabling a bulk hydrogen production at the beginning of the reaction with steam corresponding to the rapid oxidation of the surface area. The higher reaction rates obtained with solar FeO-rich powders are due to the formation of non-stoichiometric wüstite phases (Fe_1−δO with δ close to 0.05) during the solar synthesis step. Indeed, solar thermal processing of materials is commonly characterized by high heating and quenching rates favoring formation of defects and cation vacancies in the lattice structure, in turn yielding materials with high contents of non-stoichiometric phases, as validated by thermodynamics predicting the formation of different non-stoichiometric wüstite phases at high temperature. In summary, wüstite hydrolysis is hindered by the formation of a Fe₃O₄ diffusion barrier and the use of fine particles, obtained for instance from an intermediate mechanical particle grinding step, is needed to improve the particle conversion extent. In contrast, the two-step iron oxide cycle (Fe₃O₄/FeO) involves low cost and abundant chemical compounds and could potentially become an attractive option for large-scale H₂ production in future solar processes, thus requiring the design, demonstration and scale-up of efficient and reliable solar reactor technologies.

Regarding other metal-oxide redox systems, most of them are not feasible practically because they require a too much elevated temperature during the solar reduction step (ΔG° < 0 for T > 2200 °C in the case of MoO₂/Mo, SnO₂/Sn, TiO₂/TiO_2−x, MgO/Mg, or CaO/Ca redox pairs). In contrast, the non-volatile systems such as Co₃O₄/CoO and Mn₃O₄/MnO redox pairs exhibit a low reduction temperature (902 °C and 1537 °C in air, respectively), but the hydrolysis step is not thermodynamically feasible (thermodynamic calculations predict H₂ yields below 1% for two-step cycles involving these oxides [35,133]).

In order to lower the maximum temperature of the solar reduction step, ferrite cycles were alternatively proposed [97,98]. Oxygen deficient with cation excess Ni-Mn spinel ferrites (Ni_0.5Mn_0.5Fe₂O_4−δ with δ indicating the oxygen deficiency of the spinel) were proven to be capable for both water-splitting below 800 °C and regeneration (oxygen release) above 800 °C [98]. However, the low water-splitting capability of the (Ni,Mn)-ferrite system was arising from their small content in activated oxygen deficiency (oxygen vacancies) reached in the solid [96]. Strategies for improving the amount of oxygen vacancies in solid compounds have thus been investigated.

The partial substitution of iron by nickel [115], manganese [95], cobalt [114], aluminum-copper [113] or zinc [103,104,105,106,107,108,109,110,111] in the Fe₃O₄ spinel structure forms mixed iron oxides of general formula (Fe_1−xM_x)₃O₄. This ferrite can be reduced at much lower temperatures than those required for Fe₃O₄ thermal reduction (thus avoiding melting), while the reduced phase (Fe_1−xM_x)_1−yO is still capable to perform water-splitting [95]. Allendorf et al. [127] reported detailed thermodynamic analysis applied to ferrites (MFe₂O₄, with M = Co, Ni, or Zn). Their results point out that the use of ferrites is beneficial because the thermal reduction step is thermodynamically more favorable in comparison with pure iron oxide (Fe₃O₄). Thermodynamics also indicates that Ni-ferrite shows the most favorable combination of both thermal reduction and hydrolysis reaction.

Among many feasible cycles, the Zn/Fe₃O₄ system was able to react with H₂O to form H₂ and ZnFe₂O₄ (Zn-ferrite) at 600 °C [103,104]. Nevertheless, to form metallic Zn during the O₂-releasing step, either lowering the O₂ partial pressure or cooling the gas mixture was necessary to avoid metallic Zn oxidation, as for the ZnO/Zn cycle [110]. It was further shown that the H₂ production reaction can be carried out by using both ZnO and Fe₃O₄ to split water, which is thus more attractive since the O₂-releasing step does not require any quenching step and can simply proceed in air at 1500 °C [105].

3 ZnO + 2 Fe₃O₄ + H₂O → 3 ZnFe₂O₄ + H₂   T = 600–8000 °C

(7)

3 ZnFe₂O₄ → 3 ZnO + 2 Fe₃O₄ + ½ O₂   T = 1500 °C

(8)

As previously observed for the Fe₃O₄/FeO system, mixed iron oxides may similarly undergo rapid deactivation during thermochemical cycling due to coarsening and sintering of the iron oxide particles. The use of refractory ceramics as stable supports can help to alleviate material sintering, and enhance H₂ productivity without deactivation through cyclic reaction. Hence, the ferrite particles were supported on monoclinic ZrO₂ fine particles (m-ZrO₂) that provide a good resistance to agglomeration (densification) and sintering at the working temperatures of 1000–1400 °C, because m-ZrO₂ features a higher melting point than ferrite while being chemically inert to ferrite material at high temperatures. As a result, repeatable and stable thermochemical performance during two-step water-splitting cycles was demonstrated by using active ZrO₂-supported ferrite particles [114,115,116,117].

In addition, partially-stabilized tetragonal zirconia (t-PSZ) and yttria-stabilized cubic zirconia (c-YSZ) can also be used as an inert support [112,118]. Thermochemical cycling involving Fe-YSZ particles is suitable to alleviate the iron oxides sintering or melting at high temperatures thanks to the incorporation of active Fe ions into the YSZ crystal lattice, thus resulting in enhanced and reproducible thermochemical performance of the cyclic reaction. The suggested mechanism was based on the fact that the YSZ support can react with particles of ferrite at temperatures above 1400 °C in an inert atmosphere, in turn forming a solid solution. For Fe₃O₄/c-YSZ (ZrO₂ doped with over 8 mol% of Y₂O₃), the following mechanism was proposed [118]:

(1) Formation of Fe²⁺-YSZ at 1400 °C (under N₂):

x/3Fe₃O₄ + Y_yZr_1−yO_2−y/2 → Fe²⁺_xY_yZr_1−yO_2−y/2+x + x/6O₂

(9)

(2) Water-splitting at 1000 °C (60 min.) and thermal reduction at 1400 °C (30 min.):

Fe²⁺_xY_yZr_1−yO_2−y/2+x + x/2H₂O → Fe³⁺_xY_yZr_1−yO_2−y/2+3x/2 + x/2H₂

(10)

Fe³⁺_xY_yZr_1−yO_2−y/2+3x/2 → Fe²⁺_xY_yZr_1−yO_2−y/2+x + x/4O₂

(11)

In parallel to the fundamental research on the abovementioned redox working materials, a number of solar chemical reactor concepts have been designed and developed. These include reactors incorporating multi-channel honeycomb structures [122,123,124] and ceramic foams coated with ferrite [117,118], rotary-type reactors [120,126], and fluidized beds [119].

Further advances on ferrite-based systems have been proposed in the collaborative EU-project HYDROSOL (coordinated by APTL/CERTH from Greece) aiming to operate the complete redox cycle in a single solar receiver/reactor [124]. This reactor features multi-channeled ceramic (siliconized and recrystallized SiC) honeycomb structures coated with active redox materials (iron oxide-based) capable for both water-splitting at 800 °C and regeneration at 1300 °C (temperature-swing cycle). This operational technique using a fixed reactant integrated inside the reactor avoids the need of continuous particle feeding and collection at the process outlet (no solid flow). The monolithic ceramic structure acts as an efficient volumetric solar absorber. The whole cycling process can be achieved in a single solar energy receiver/converter and it approaches a configuration similar to that encountered in automobile exhaust catalytic converters. The water steam flowing inside the ceramic structure oxidizes the reduced activated material producing pure hydrogen. Then, the material must be reduced during the regeneration step by increasing the solar receiver temperature, allowing a cyclical operation in a single reactor. Various compositions of M-Zn-doped iron oxides (M=Mn or Ni) with a spinel or wüstite phase structure were synthesized via aerosol spray pyrolysis (ASP) and combustion self-propagating high-temperature synthesis (SHS) [124,125], and then used as coatings on the monolith [122]. A pilot-scale (100 kW) solar reactor equipped with two parallel ceramic monolithic supports operating in opposite modes was then developed for quasi-continuous H₂ production [123]. The favored redox systems are zinc- and nickel/zinc-based ferrites. The reaction scheme in the case of zinc can be written as follows:

ZnFe²⁺_yFe³⁺_2−yO_4−y/2 + xH₂O → ZnFe²⁺_y−x/2Fe³⁺_2−y+x/2O_4−y/2+x + xH₂

(12)

ZnFe²⁺_y−x/2Fe³⁺_2−y+x/2O_4−y/2+x → ZnFe²⁺_yFe³⁺_2−yO_4−y/2 + x/2O₂

(13)

Gokon et al. [117,118] proposed and tested ceramic foam materials made of MgO-partially stabilized Zirconia (MPSZ). The porous foam was coated with both c-YSZ and Fe₃O₄ particles (Fe₃O₄/YSZ/MPSZ foam device) and it effectively absorbs solar radiation due to the large specific surface area. The H₂ generation from Fe₃O₄/YSZ/MPSZ foam devices is associated with the redox transition of Fe²⁺-Fe³⁺ ions in the YSZ lattice. The deactivation of Fe₃O₄ due to high-temperature sintering can be alleviated because the active Fe²⁺ ions remain in the YSZ lattice. The same ceramic foams were used with coatings made of NiFe₂O₄/m-ZrO₂ powders that exhibit the highest reactivity among the ferrite/zirconia systems.

Diver et al. [126] developed a concept of counter-rotating ring receiver/reactor/recuperator (CR5) enabling efficient sensible heat recovery during cyclic reactions to reach acceptable thermal efficiencies. The solar-driven heat engine is composed of a stack of counter-rotating rings to which are attached fins made of an active metal oxide ceramic. While the rings rotate, the metal oxide material moves alternately through a high temperature solar reduction zone and a lower temperature water-splitting zone. Since the fins counter-rotate, sensible heat can be internally recovered, thus reducing the overall energy input requirements for the aim of improving efficiencies.

Kaneko et al. [120] also developed a similar concept of rotary-type solar reactor integrating Ni-Mn ferrite loaded on a YSZ support.

Gokon et al. [119] proposed and tested a solar reactor for two-step water-splitting using m-ZrO₂ supported NiFe₂O₄ particles processed in an internally circulating fluidized bed. The ferrite particle conversion of approximately 45% was reached upon 1 kW of input simulated power supplied by Xe-beam irradiation. The proposed solar reactor concept was considered to be suited for being associated with beam-down optics [134].

Finally, the two-step cycle based on cerium oxides (CeO₂/Ce₂O₃) was first demonstrated in 2006 by Abanades and Flamant [84], consisting of two chemical steps: (1) solar reduction, 2CeO₂ → Ce₂O₃ + 1/2O₂; (2) hydrolysis, Ce₂O₃ + H₂O → 2CeO₂ + H₂. CeO₂ was first reduced into Ce₂O₃ during material melting over 2000 °C under inert atmosphere at reduced pressure (10–20 kPa), which induced also a partial sublimation of CeO₂. The reduced material was pounded into powder (50–150 μm) for the hydrolysis tests. Then, an efficient reaction of Ce₂O₃ with water producing H₂ was observed with a rapid and complete oxidation of Ce(III) oxide below 500 °C. The high reactivity of Ce₂O₃ with steam (3 mmol_H2/gCe₂O₃) was shown to be the main interest of the CeO₂/Ce₂O₃ cycle. This observation opened the door to a new research area related to the mixed oxide cycles containing cerium oxides [135,136]. Similar to ferrite cycles, the targeted result is the significant decrease of the reduction temperature of Ce(IV) into Ce(III) species, while taking advantage of the high reactivity of Ce(III) species with water (and/or CO₂). Early studies on ceria-based materials synthesis and physicochemical characterization at CNRS-PROMES laboratory (France) showed that binary systems such as CeO₂-ZrO₂ solid solutions can be promising candidates for two-step water-splitting.

The thermodynamic and kinetic properties of ceria can be altered by doping its fluorite structure with transition metals and rare earth metal oxides [135,136,137,138,139,140,141,142,143,144,145,146,147,148,149,150,151,152,153,154]. Indeed, ceria is an active compound used in many catalytic systems, and it is attractive due to its thermal stability (good resistance to sintering) and ability to exchange oxygen via storing and releasing oxygen reversibly. This oxygen mobility is mainly restricted to the material surface unless the ceria is modified by adding another oxide, such as typically zirconia, which favors the material bulk reduction thanks to enhanced oxygen diffusion rates. However, oxidation thermodynamics of Zr-substituted systems are not as favorable as for pure ceria. The addition of dopants in ceria fluorite structure (M_xCe_1−xO₂, where the added metal M is commonly a trivalent cation like Y³⁺, La³⁺, Gd³⁺, Sm³⁺, Pr³⁺, Sc³⁺ or tetravalent cation like Zr⁴⁺ or Hf⁴⁺) shows a favorable impact on the thermodynamics of ceria reduction, and this approach has been extensively investigated for improving the ionic conductivity of solid oxide fuel cells, which is directly linked to the extent of oxygen vacancies. The addition of dopants has been shown to enhance the reduction extent at low oxygen partial pressure when compared with un-doped ceria (at the expense however of lower oxidation capability with water). Hence, the introduction of dopants such as Zr [145,146,147,148,149,150,151,152,153,154], Sm [139] and transition metals oxides (MO_x with M = Mn, Ni, Fe, Cu) [135,140,141] has been extensively explored for applications in solar-driven thermochemical redox cycles, as an efficient means to enhance the thermodynamic driving force of the O₂-releasing reduction reaction at lower temperatures. However, it should be noted that trivalent dopants do not improve significantly the thermochemical performance of ceria because they do not present any meaningful increase of the fuel production or improvement of the thermal stability.

The deviation from oxygen stoichiometry (δ in CeO_2−δ), otherwise known as the oxygen storage capacity, is directly related to the maximum amount of H₂ capable of being generated during oxidation. While the oxygen exchange capability of ceria is lower compared to iron oxide-based cycles, sintering is less problematic because the melting point is considerably higher. Different research works have evidenced the high promise of ceria for application to solar fuel production according to the redox cycle represented in Figure 13, using various ceria microstructures and morphologies including dispersed powders [140,141,142,143,144,145,146,147,148,149], three-dimensionally ordered macroporous (3DOM) powders featuring templated macro-scale porosity [154,155], porous structured monoliths [137,138,139], porous felts or fibers [156,157], or reticulated porous foams [158,159,160]. The morphologies enabling both enhanced available geometric surface area and bulk interconnected pore system that facilitate the mass transport of reacting species to and from the oxidation sites, are the most suitable for rapid fuel production. Optimized materials shaping is thus required to provide a good solar absorptivity favoring homogeneous heating, along with a high specific surface area favoring solid–gas reactions. A long-term thermal stability is also required for the considered ceria structures regardless of the involved shaping method. However, the material shaping process represents an additional step required for the preparation of tailored materials and for their incorporation in suitable solar reactors, which could have an impact on the process economics.

The amount of O₂ released during the reduction step (oxygen non-stoichiometry, δ) at 1400 °C and the associated reduction yield depend significantly on the presence and amount of Zr introduced in the materials [149]. The reduction yields of the materials (Ce³⁺/Ce_total) range from 6% for pure un-doped ceria to 28% for 50%-Zr in Zr-substituted ceria during the first cycle. By increasing the Zr amount in ceria, the reduction yield increases linearly. This result points out the beneficial effect of Zr addition in promoting Ce⁴⁺ reduction (Figure 14). In contrast, the presence of trivalent dopants has no noticeable effect on the material reducibility and hydrogen production yield.

Given that Zr⁴⁺ cation is not reducible, the gravimetric amount of O₂ released per unit mass of material (thus the potential H₂ production) evolves differently compared to the ceria reduction yield. The materials with Zr content over 25% are the most suitable compounds in term of oxygen availability and mobility. The Figure 15 represents the evolution of the O₂ amount released from the ceria materials during the reduction step as a function of the Zr content. In contrast to the evolution of the reduction yield (Figure 14), the gravimetric amount of O₂ released tends to reach a plateau above roughly 25% of Zr content. The quantity of oxygen released reaches a threshold because a balance settles: the reduction yield is indeed increased whereas the number of reducible cation (Ce⁴⁺) in ceria/zirconia solid solution is concomitantly decreased when increasing the Zr content.

The steam hydrolysis of the Zr-substituted ceria powders was also studied as a function of the reaction temperature (850 °C, 950 °C, and 1050 °C) for 25%-Zr content and a reduction temperature in the previous step of 1400 °C. The H₂/O₂ ratio increases with temperature and equals two at 1050 °C with a final amount of H₂ produced of 240 µmol/g (Figure 16). The kinetic parameters based on Arrhenius law have been identified from these isothermal experiments and the obtained activation energy was found to be 51 kJ/mol for the hydrolysis reaction [145].

More recently, perovskites of the form ABO_3−δ have emerged as a new attractive class of nonstoichiometric oxides that are widely unexplored for thermochemical cycles [161,162,163,164,165,166,167,168,169,170,171,172,173]. These oxygen-deficient materials exhibit interesting characteristics concerning oxygen storage capacities and transport properties through oxygen vacancies formation. Dopants can be substituted on both A and B cation sites, and thus the number of potential material configurations is substantially greater than for ceria-based systems. Elevated water-splitting activity has been demonstrated in La_1−xSr_xMnO_3−δ materials, so-called LSM (lanthanum strontium manganite) perovskites (Figure 17) [166], and H₂ fuel production performance comparing favorably to that of CeO₂ has been evidenced in Sr- and Mn-substituted LaAlO₃ [173], with H₂ productions (up to around 300 µmol/g with Sr_xLa_1−xMn_yAl_1−yO_3−δ for one cycle with thermal reduction at 1350 °C and oxidation at 1000 °C) that are only slightly lower than those measured for Zr-substituted ceria. The controlled introduction of Sr into LaMnO₃ allows for tuning the redox thermodynamics in the La_1−xSr_xMnO₃ compositional series. Decreasing the amount of substituted Sr in LSM enhanced the re-oxidation yield at the expense of a lowered final reduction extent (lower δ), thus decreasing the global amount of generated H₂. The rate of hydrogen production decreased when increasing Sr content, confirming that the intermediate materials formulations offer the most favorable combination of redox properties with trade-off between reduction and oxidation thermodynamics. The observed evolution of the Mn oxidation state during cycles in manganite perovskites implied that partial re-oxidation of Mn³⁺ into Mn⁴⁺ occurred, thus highlighting the activation of Mn⁴⁺/Mn³⁺ redox pair in the perovskites (whereas it is not active in Mn₂O₃/Mn₃O₄). Based on the possible number of doping schemes, series of even more attractive materials certainly remain to be discovered.

4. Conclusions

The hydrogen production from water-splitting using solar-driven thermochemical redox cycles based on metal oxide reactions is an attractive route for sustainable and carbon-neutral solar fuel generation. Such cycles involve endothermic reactions that make use of concentrated solar radiation as external heat source for supplying high-temperature process energy. Thermochemical water-splitting processes efficiently convert concentrated solar energy into storable and transportable clean fuels. Concentrating high-flux solar technologies including solar towers and parabolic dishes are already applied at large-scale for commercial electricity and power generation. They may be further coupled to chemical reactors for solar fuel production with potential solar-to-fuel energy conversion efficiencies exceeding 20–25%. Therefore, thermochemical redox cycles should represent an attractive pathway for solar fuels production at competitive costs while further outperforming benchmark water electrolysis process using solar electricity generated by photovoltaics (PV) or concentrating solar power (CSP).

In contrast to the direct thermolysis of H₂O above 2500 °C, thermochemical water-splitting cycles proceed at lower maximum operating temperatures and produce both H₂ and O₂ in separate steps, thereby avoiding their recombination and bypassing the need for high-temperature and costly downstream gas separation. Promising systems under investigation include two-step cycles based on metal oxide redox pairs. The feasibility of solar chemical reactor technologies for the thermal dissociation of volatile oxides (ZnO, SnO₂) was demonstrated and the necessity to operate at reduced pressure under low oxygen partial pressure or to implement gas quenching was especially highlighted to avoid recombination between reduced species and oxygen, and to reach significant oxide conversions. Non-volatile metal oxides cycles (Fe₃O₄/FeO, CeO₂/Ce₂O₃, Ce- and Fe-based mixed oxides, and perovskites) can be operated with continuous removal/separation of the evolved O₂ from the condensed reduced phase during the solar reduction step, so that high reduction extents/rates may be expected without requiring any quenching step. The reduction step proceeds with the material remaining in the condensed state and the evolved O₂ can be released and swept by a flow of inert gas, thus avoiding recombination issues. Non-stoichiometric oxides (CeO_2−δ, ABO_3−δ) have been considered for decreasing the temperature of the reduction step below 1400 °C, in turn alleviating material sintering and implying enhanced performance stability and cyclability. In contrast to the zinc and tin-based systems, ceria and perovskites remain in the solid state with their crystalline structure maintained stable throughout. Identification of materials formulations with improved redox activity, high oxygen transport and exchange properties, and long-term stability over extended cyclic operation under concentrated solar irradiation is required, as well as their relevant shaping as 3D porous structures capable to absorb concentrated solar radiation while offering large available geometrical surface area for the solid/gas reactions. Such materials can be integrated in monolithic solar reactors with enhanced efficiency and scalability for future application, in which O₂ and H₂ are produced by temperature-swing cyclic operation through sequential reduction and oxidation steps. Development of solar reactors incorporating heat recuperation between the reduction and oxidation steps is also challenging to help improving the overall energy conversion efficiency. The current scientific and technical challenges for developing clean and efficient H₂ production processes based on innovative two-step cycles are thus related to various research areas involving materials science and chemistry, solar reactors and process engineering.