Water Splitting by MnO_x/Na₂CO₃ Reversible Redox Reactions

Abstract

Thermal water splitting by redox reactants could contribute to a hydrogen-based energy economy. The authors previously assessed and classified these thermo-chemical water splitting redox reactions. The Mn₃O₄/MnO/NaMnO₂ multi-step redox cycles were demonstrated to have high potential. The present research experimentally investigated the MnO_x/Na₂CO₃ redox water splitting system both in an electric furnace and in a concentrated solar furnace at 775 and 825 °C, respectively, using 10 to 250 g of redox reactants. The characteristics of all reactants were determined by particle size distribution, porosity, XRD and SEM. With milled particle and grain sizes below 1 µm, the reactants offer a large surface area for the heterogeneous gas/solid reaction. Up to 10 complete cycles (oxidation/reduction) were assessed in the electric furnace. After 10 cycles, an equilibrium yield appeared to be reached. The milled Mn₃O₄/Na₂CO₃ cycle showed an efficiency of 78% at 825 °C. After 10 redox cycles, the efficiency was still close to 60%. At 775 °C, the milled MnO/Na₂CO₃ cycles showed an 80% conversion during cycle 1, which decreased to 77% after cycle 10. Other reactant compounds achieved a significantly lower conversion yield. In the solar furnace, the highest conversion (>95%) was obtained with the Mn₃O₄/Na₂CO₃ system at 775 °C. A final assessment of the process economics revealed that at least 30 to 40 cycles would be needed to produce H₂ at the price of 4 €/kg H₂. To meet competitive prices below 2 €/kg H₂, over 80 cycles should be achieved. The experimental and economic results stress the importance of improving the reverse cycles of the redox system.

1. Introduction

1.1. The Need to Develop H₂ Production

The global energy production is influenced by the growing shortage of traditional fossil fuels and the threat of disturbing the ecological balance due to climate change. Conferences like COP21 and COP26 set goals to reduce greenhouse gases such as CO₂. These COPs proclaimed the leading role of renewable energy in solving climate problems. The wider use of hydrogen was advocated [1,2] through, e.g., hydrogen-enriched natural gas [3], the direct reduction of iron ore [4], the use of hydrogen in the minerals’ processing [5], and the production of CO₂-based synthetic natural gas [6], among others. It is important to assess the ways that hydrogen production can help to meet the renewable energy goals.

Since there are no significant resources of hydrogen on earth, hydrogen is referred to as a secondary energy carrier, made from a primary energy source (nowadays, ~95% from fossil fuels) [1,3]. The global transition to hydrogen as an energy carrier accompanies the rise in required “green” energy.

In 2020, the annual production of hydrocarbons reached 14 billion tons of oil-equivalent [7]. To affect the global CO₂ emissions, at least 10% of hydrocarbons should be replaced by renewable energy resources and/or by hydrogen. If considering hydrogen as a major potential renewable energy carrier, roughly 1 billion tons of H₂ produced from non-fossil sources are required to accomplish this. In 2020, only 87 million tons of hydrogen were produced worldwide [8], mainly used in ammonia production and oil refining. This is more than 10 times below the required quantity to replace 10% of hydrocarbons.

The main method currently considered in producing green hydrogen is electrolysis using electricity from renewable sources (wind turbines, photovoltaics). It theoretically takes about 50 kWh to produce electrolysis H₂ at an overestimated efficiency of 80% [1], without taking other losses into account. The annual global energy production from renewable sources (excluding hydro-energy) was estimated to be 2800 TWh in 2019, which would translate into 56 million tons of H₂ annually, if these sources were solely used to produce H₂. This is lower than the current hydrogen production. In addition, the production cost of the renewable-based H₂ is two to four times higher than when using traditional petrochemical pathways, and it is not deemed practical to expand these renewable technologies to meet the goal of 1 billion tons of H₂ [7].

The global electricity production of hydropower in 2019 was around 38 Exajoules (10⁴ GWh) [9], which allows an annual production of 200 million tons of H₂ if no electricity is sent to the power grid, which is only three times the current production. Since hydropower is already realized for over 20% of its potential, it is therefore unable to solve the problem [9]. The same reason applies to nuclear energy, where the energy production is only 25 Exajoules (even less than hydropower). With the often negative public attitude towards nuclear energy and the limited reserves of Uranium, nuclear power will not sustain a future for hydrogen production.

According to the arguments above, fossil hydrocarbons are deemed to remain a valid source for hydrogen production in the near future. However, the production of hydrogen with Steam Methane Reforming (SMR) results in the emission of 10 kg CO₂/kg H₂, and such hydrogen is classified as “gray hydrogen” according to ecological standards [1]. It is environmentally unattractive and does not contribute to the reduction of CO₂ emissions. An additional process to make this technology “cleaner” is carbon capture and storage (CCS). The major disadvantage of this process is that the total cost rises because the process requires more energy. CCS would not be sufficient to have an environmental impact. With the rise of the hydrogen cost to almost the double, it is not a viable solution [10].

It is hence important to develop novel green H₂ production methods, as reviewed and studied by the authors [1,3,11]. Among these methods, thermo-chemical water splitting by redox systems [12] could be considered to have significant potential and thus is the subject of the present research.

1.2. Hydrogen Production by Water Splitting in Thermal Redox Systems

An earlier paper by the authors [12] assessed and classified different thermo-chemical water splitting redox reactions. A redox system ranking was developed by applying multiple assessment criteria, including energy efficiency, H₂ yield, conversion, cyclic operation, production cost, and environmental and safety impacts.

Several very high temperature reactions (≥1000 °C) were not taken into consideration, e.g., involving. metal–metal oxides/hydroxides, perovskites, or doped ceria. At such high temperatures, concentrated solar energy would be the potential heat source, but at the expense of having to use extravagant construction materials for the solar receiver-reactor. Selected systems should operate at lower temperatures, preferably below 1000 °C for the reactor wall temperature. Deng et al. [12] finally selected 4 out of 24 redox reactions that meet the systems’ selection priorities. These systems included Mn₃O₄/MnO/NaMnO₂, MnFe₂O₄, U₃O₈/UO₂CO₃ and ZnO/Fe₃O₄/ZnFe₂O₄ redox reactions. The U₃O₈ system was discarded for nuclear hazard reasons, and the ZnO system scored 30% lower than both other redox reactions and was hence eliminated from a priority investigation. Both the MnFe₂O₄ and Mn₃O₄/MnO/NaMnO₂ redox processes are of a multi-step nature, where water is decomposed into H₂ and O₂ via a medium-temperature two-step (MnFe₂O₄) or a medium-temperature four-step process (Mn₃O₄), respectively, that forms a closed cycle [12]. This process was previously fostered as having a high potential [13] and involves four reaction steps, as shown below.

1.3. Literature Findings on MnO_x/Na₂CO₃ Cycle

Xu et al. [13] comprehensively studied the Mn₃O₄/MnO/Na₂CO₃ system. Theoretical considerations were validated by experiments on a small scale (200 mg) in a thermogravimetric analyzer. The Mn-based reduction and oxidation reactions required an operating temperature in excess of 850 °C. This four-step redox cycle is given by the equations below.

$$3{\mathrm{Na}}_2{\mathrm{CO}}_3 \left(\mathrm{s}\right)+2{\mathrm{Mn}}_3{\mathrm{O}}_4\left(\mathrm{s}\right)\to 4{\mathrm{NaMnO}}_2\left(\mathrm{s}\right)+2{\mathrm{CO}}_2\left(\mathrm{g}\right)+2\mathrm{MnO}\left(\mathrm{s}\right)+{\mathrm{Na}}_2{\mathrm{CO}}_3 \left(\mathrm{s}\right)$$

(1)

$$2\mathrm{MnO}\left(\mathrm{s}\right)+{\mathrm{Na}}_2{\mathrm{CO}}_3 \left(\mathrm{s}\right)+{\mathrm{H}}_2\mathrm{O}\left(\mathrm{v}\right)\to {\mathrm{H}}_2\left(\mathrm{g}\right)+{\mathrm{CO}}_2\left(\mathrm{g}\right)+2{\mathrm{NaMnO}}_2\left(\mathrm{s}\right)$$

(2)

$$6{\mathrm{NaMnO}}_2\left(\mathrm{s}\right)+{\mathrm{ayH}}_2\mathrm{O}\left(\mathrm{v}\right)+\left(3+\mathrm{b}\right){\mathrm{CO}}_2\left(\mathrm{g}\right)\to 3{\mathrm{Na}}_2{\mathrm{CO}}_3 \left(\mathrm{s}\right)+{\mathrm{aH}}_{\mathrm{x}}{\mathrm{MnO}}_2·{\mathrm{yH}}_2\mathrm{O}\left(\mathrm{s}\right)+{\mathrm{bMnCO}}_3 \left(\mathrm{s}\right)+{\mathrm{cMn}}_3{\mathrm{O}}_4\left(\mathrm{s}\right)$$

(3)

$${\mathrm{aH}}_{\mathrm{x}}{\mathrm{MnO}}_2·{\mathrm{yH}}_2\mathrm{O}\left(\mathrm{s}\right)+{\mathrm{bMnCO}}_3 \left(\mathrm{s}\right)\to \left(2-\mathrm{c}\right){\mathrm{Mn}}_3{\mathrm{O}}_4\left(\mathrm{s}\right)+{\mathrm{ayH}}_2\mathrm{O}\left(\mathrm{v}\right)+{\mathrm{bCO}}_2\left(\mathrm{g}\right)+0.5{\mathrm{O}}_2\left(\mathrm{g}\right)$$

(4)

The results of Xu et al. [13] exhibited a more than 90% yield for both the H₂ and O₂ evolution for five successive redox cycles. The key recyclability feature of the Mn-based system was demonstrated to be the complete in/out shuttling of Na⁺ of the manganese oxides, where the reactions to form α-NaMnO₂ are thermodynamically favorable. It was demonstrated that the Na⁺ extraction is enhanced by the mobility of Na⁺ in the layered structure when intercalated by water and is promoted by CO₂, thus driving the reaction equilibrium to a mixture of Mn₃O₄, and MnCO₃. After thermal reduction at 850 °C, Mn₃O₄ is formed, and this closes the thermo-chemical cycle.

The cycles consist of multiple stages, operating from 50 °C to 850 °C [13,14]. Dissolved Na₂CO₃ must be recovered from the aqueous solution before the needed re-heating to 850 °C, and it will impose a considerable energy cost. The wide range of the temperatures in the different steps requires adequate heat management and makes the process more complex.

The possible loss of volatile Na species at high temperature can be responsible for the progressive loss of process cyclability [15]. The hydrolysis reaction, however, drives the formation of Na-birnessite. Bayón et al. demonstrated that the surface area of Mn₃O₄ is the main factor, while the crystalline domain size has a secondary influence [16]. Xu et al. [17] performed the multistep reactions on 200 mg Mn₃O₄/Na₂CO₃ below 850 °C and demonstrated that the rates of H₂ releasing and Na⁺ extraction depend on the redox properties of metals in ferrite-based oxides and the facility of intercalating alkali cations. Bayón et al. [14] also studied the feasibility of the MnO/Na₂CO₃ cycle, with a maximum conversion of 47% and a productivity of 20.1 μmol H₂ min⁻¹ g⁻¹. Alonso et al. [18] studied the Mn₂O₃/Mn₃O₄/MnO cycles at >1835 K using solar-driven thermogravimetry and derived the overall kinetic rate equations.

1.4. Objectives of the Research

Studies related to thermo-chemical water splitting have largely been limited to individual steps of the cycles and their kinetics analysis. Experiments were mostly performed on gram scale. This kind of research typically shows limitations in cyclability assessment, realistic energy conversion, and operation complexity due to the small size of the experiments used. The present study aims to demonstrate the water splitting processes on a larger scale, using both an electrical and a concentrated solar reactor to examine the proposed systems under realistic operation conditions. The selected redox pair was the MnO_x/Na₂CO₃ cycle, based on previous studies [12]. It is expected that the result will give evidence of the thermochemical hydrogen-releasing approaches by water splitting, without complex microstructures and reactor design. In an attempt to reduce the reaction temperatures, the operating conditions were reduced from 850 °C [13] to between 775 °C and 825 °C, while avoiding the 50 °C condensation step [13] of reaction (3), and step (3) was operated at 140 °C, as suggested by Rao et al. [19]. These reactions were moreover performed at a 10 to 250 g scale, instead of using small-scale (200 mg) thermogravimetry as applied by Xu et al. [13]. It will also clarify the efforts required to transform the small-scale test results into industrial implementation in a renewable energy field.

2. Materials and Methods

The priority redox systems under scrutiny involve the Mn₃O₄/MnO/Na₂CO₃ cycles, as described before.

2.1. Precursor Materials and Characteristics

The characteristics of all precursor materials used are given in

Table 1. Reactants used.

Reactant	Chemical Formula	Supplier	Purity (%)
Sodium carbonate	Na2CO3	Lushi Co., Ltd. (Guangzhou, China)	≥99.9
Manganese (II) carbonate	MnCO3	Sigma-Aldrich Chemie GmbH (St. Louis, MO, USA)	≥99.9
Manganese (II–III) oxide	Mn3O4	Sigma-Aldrich Chemie GmbH	≥99
Manganese (II) oxide	MnO	Sigma-Aldrich Chemie GmbH	≥99

. Solid mixed particles were prepared.

The α-Al₂O₃ and olivine were used as inert filler materials to form a porous fixed bed in the electric furnace experiment or a fluidizable bed (in the solar rig) and to enhance the heat transfer. The list of facilities, including parts of the experimental set-up, is summarized in

Table 2. Experimental and analytical facilities.

Names	Models	Manufacturer
Syringe water pump	LSP02-1B	LongerPump (Hebei, China)
Electronic balance	ME403/02	Mettler Toledo (Shanghai, China)
Multifunctional crusher	800Y	Yongkang Boou Hardware Products Co. (Zhejiang, China)
Tubular furnace reactor	Customized	ZSHIELD Inc. (Cupertino, CA, USA)
Particle size distribution	Mastersizer 2000	Malvern panalytical (Malvern, UK)
GC-MS	HAS-301-1474	DECRA (Berlin, Germany)
XRD	RINT2000	RIGAKU (Tokyo, Japan)
SEM	JSM-7800F	JEOL (Tokyo, Japan)
BET	ASAP 2020	Micromeritics (Norcross, GA, USA)

.

Particle sizes were measured by Malvern laser diffraction and confirmed by SEM-imaging. Particle sizes of precursors and inert fillers are given in

Table 3. Average volume diameter (d_V) and surface/volume diameter (d_SV) of the feedstock particles used.

Chemicals	dv (μm)	dsv (μm)	ψ
Mn3O4	15.12	12.70	0.84
Na2CO3	394.45	331.34	0.84
MnO	10.94	9.19	0.84
α-Al2O3	60.4	56.95	0.85
Olivine (100–150 mesh)	167.85	142.67	0.84

. The olivine particles used in the experiments were crushed spherical magnesium iron silicates with corresponding values of Waddell sphericity factor ψ of 0.8–0.9 [20]. When dealing with powders, a fundamental knowledge of their physical parameters is indispensable, with different methods and approaches proposed in the literature. The results obtained differ widely, and it is important to define the standards to be applied, both towards the methods of investigation and the interpretation of experimental results [21]. First, describing the size of a particle is not as straightforward as one might suspect. Factors of non-sphericity and size distributions make it impossible to put “size” in just one number. Whereas sieving can be used for coarser particles of a size in excess of about 50 μm, instrumental techniques span a wide size range. For fine particles, the occurrence of cohesive forces needs to be overcome, and dispersants and sample mixing need to be applied. Secondly, the shape of the particles is examined. By defining sphericity, irregularly shaped particles are described. Finally, the density of particles and of particle assemblies and their voidage (volume fraction of voids) need to be defined. The particle size distributions of the feedstock commercial materials are mostly Gaussian, with a fairly narrow particle size distribution, as illustrated in Figure 1.

The final ψ value was chosen between 0.84 and 0.85. While Mn₃O₄, MnO and Na₂CO₃ are common salts, the corresponding ψ value is 0.84. The particle size and distribution are given by the Malvern particle size test, where d_V is the volume average particle size (μm). The surface-volume particle diameter, ds_V, is calculated as

$${\mathrm{d}}_{\mathrm{sv}}={\mathsf{\Psi } \mathrm{d}}_{\mathrm{v}}$$

(5)

For the pulverized mixtures, the grain size is a lot smaller and is better determined from SEM imaging, as shown in Figure 2. No obvious differences in the microscopic morphology of crushed Mn₃O₄ and MnO were detected. The particle size of Na₂CO₃ also becomes smaller after pulverization.

Clearly, by pulverization, the particle size is reduced to far below 2 μm, with smaller particles around 150 nm.

2.2. Synthesis of the Redox Reactants

Ball-milling [22] was used to prepare the reactants. In the present study, the synthesized MnO_x/Na₂CO₃ mix was calcined at 700 °C in a fixed bed reactor for 1 h and in a N₂ atmosphere. This calcination under N₂ is required to “activate” the reactants before the first thermal reduction step. After calcination, the reactants were collected and XRD-analyzed toward their phase composition and their stability under N₂ atmosphere. After the XRD analysis, reactants were used in the thermochemical water splitting reaction.

These XRD results are discussed in Section 3.2.

2.3. Experimental Set-Up

Isothermal experiments of the redox system were carried out in both an 8 kW electrically heated furnace (Figure 3) and in a 20 kWth solar-heated reactor (Figure 4). The H₂ and CO₂ productions were monitored. The operating temperature was set by a thermocouple inserted in the redox reactant’s bed.

Solar H₂ production was performed in the solar thermal furnace of Figure 4, with both a heliostat and a parabolic mirror to concentrate the solar rays into the irradiation focus (0.3 m diameter) of the 0.5 × 0.5 × 0.5 m cavity. The maximum solar heat captured in the cavity was ~20 kWth, starting from about 30 kWth collected by the heliostat. Since cohesive sub-micron particles needed to be treated, a vibrating fluidized bed was installed inside the cavity. The solar rig during operations is shown in Figure 5. A 0.05 m internal vibrated fluidized bed reactor was used. The advantages of using solar concentrated as a heat source were recently highlighted in number of publications with respect of solar-to-power uses [23,24,25], solar-driven thermo-chemical reactions [4,25] and solar-driven physical applications [26,27,28].

Water/steam and N₂ carrier gas were preheated in a coil heat exchanger within the cavity as illustrated at the right of the reactor in Figure 4. A 0.25 m deep bed was formed. About 250 g of the Mn₃O₄/Na₂CO₃ reactant particles were mixed with the same amount of olivine. A multi-orifice tubular distributor provided a uniform gas distribution at the bottom of the fluidized bed [29].

Table 4. Experimental conditions.

Bed Height	Fluidizing Carrier Gas	Reaction T	Superficial Gas Velocity (cm/s)	Vibration Frequency	Vibration Amplitude	Vibration Direction
0.25 m	N2	775–825 °C	<2	<50 Hz	0.6 mm	Vertical

lists the geometrical and operating parameters.

The accuracy of the measurements was 1 g for the weight, 0.05 cm/s for the superficial gas velocity, 0.1 °C for the temperature, and 0.1 Hz for the vibration frequency. The vibration frequency was adjusted by varying the motor inverter frequency, and up to 47.7 Hz could be reached. A fixed amplitude of 0.6 mm was used. The bed pressure drop from distributor to the top of the bed was continuously measured and recorded by a differential pressure transmitter (Rosemount Inc., Shakopee, MN, USA). Nitrogen was supplied using a separate mass flow controller. After the redox reactions and subsequent cooling, the products were collected for analysis.

2.4. Mn₃O₄/Na₂CO₃ Cycle

The redox reactants were prepared by milling and mixing of Na₂CO₃ and Mn₃O₄. Olivine or α-Al₂O₃ (100–150 mesh) was added to avoid sintering, and its dosage was half of the total mass of Mn₃O₄ and Na₂CO₃.

• In a first step, the furnace was slowly heated to the required testing temperature (775 or 825 °C) under N₂ flow. The outlet gas passed through saturated clarified lime water to capture evolved CO₂. When a constant temperature was achieved in the testing section of the reactor, and when no CO₂ exhaust was detected, water was injected by syringe pump and heated until reaching the steam temperature in the preheating section of the furnace and reactor. The high temperature steam reacted with the reactants, and the produced H₂ was measured at the outlet by GC-MS. The column used was the TDX-01 packed column produced in Lanzhou Zhongke Antai Analysis Technology Co., Ltd. (Shenzhen, China). The outlet gas flow rate was measured by an electric flowmeter, and the real-time record was transmitted to the computer. A hydrogen alarm was placed at the outlet. The step was assumed to be nearly completed when the hydrogen concentration was lower than 0.1%.
• The reduction step consisted of 2 subsequent reactions: firstly, a cooling to 140 °C (for about 5 h), followed by the reduction step using pure CO₂ at 825 °C (step 4 of the reaction scheme). It was observed that steps 1 and 2 of the reaction scheme were started simultaneously, although the quantity of produced H₂ remained very low (<0.1%).

3. Results and Discussion

3.1. Hydrogen Yield in the Electric Furnace

The behavior of the MnO_x/Na₂CO₃ system was assessed for different reactant preparations, as illustrated in

Table 5. Experimental conditions.

	Reactants	T (°C)	N2 Flow Rate (L/min, 20 °C)
1	Mn3O4 + Na2CO3	825	0.5
2	Milled Mn3O4 + Na2CO3	825	0.5
3	Mn3O4 + Na2CO3	775	0.5
4	Milled Mn3O4 + Na2CO3	775	0.5
5	Milled MnO + Na2CO3	775	0.5

below.

The time-dependent H₂ production is illustrated in Figure 6 for milled MnO + Na₂CO₃. Other reactants showed a similar profile. The hydrogen generation progresses inversely proportional to the oxidation of the reactant. The yield is fast for the first 100 min, then slows down, to reach a final reactant-depletion yield after 300 min.

It should be remembered that the maximum H₂ production is 0.5 mol/mol MnO. For all experiments, cumulative H₂ production results were calculated and are presented in

Table 6. Experimental results for hydrogen yields obtained in the electric furnace (average value with standard deviation of ±0.1%).

	Reactants	T (°C)	N2 Flow Rate (L/min, 20 °C)	H2 yield
				Cycle 1	Cycle 2	Cycle 5	Cycle 10
1	Mn3O4 + Na2CO3	825	0.5	72.3%	67.9%	68.2%	54.0%
2	Milled Mn3O4 + Na2CO3	825	0.5	78.3%	-	71.6%	60.4%
3	Mn3O4 + Na2CO3	775	0.5	Very low	Not further studied
4	Milled Mn3O4 + Na2CO3	775	0.5	56.4%	Not further studied
5	Milled MnO + Na2CO3	775	0.5	80.2%	-	77.4%	61.9%

. It is clear that most of the H₂ is produced within 250 min. The extra subsequent production is negligible.

Clearly, the particle size is very important, and milled reactants achieve a higher H₂ yield than the non-milled commercial reactant mixes. After milling, the Mn₃O₄/Na₂CO₃ mix has an average particle size of 0.2 to 4 μm, whereas the MnO/Na₂CO₃ is more uniform around 1 μm. This difference is also reflected in the external surface area (BET), as measured by Micromeritics and Micrometrics at 297.0 and 498.9 m²/g, respectively. Both particle size and BET differences favor the MnO/Na₂CO₃ system. Since Na₂CO₃ needs to be recovered at 140 °C before reheating to 850 °C, this will impose a considerable energy cost. Due to the wide difference between the temperatures of the different steps, an adequate heat management needs to be applied, and this adds complexity to the process.

The cycle performance of commercial Mn₃O₄ and Na₂CO₃ needs to be further improved. For the milled MnO and Na₂CO₃ reacting at 775 °C, the H₂ yield reaches 80.2%. The use of finer size reactants reduces the required reaction temperature and hence the cost of hydrogen production.

3.2. Morphology of the Mn₃O₄/MnO/Na₂CO₃ Reactants

Typical SEM images are presented in Figure 7.

Particles remain in the μm to sub-μm size range. Through regeneration (reduction), they appear to become more spherical. XRD analysis is generally used to identify the crystalline phase and crystal structures of minerals and their mixes. To index the XRD patterns of each component, the JCPDS numbers were used as follows: MnO (JCPDS 78-0424), NaMnO₂ (JCPDS 25-0845), MnCO₃ (JCPDS 85-1109), Na₂CO₃ (JCPDS 37-0451) and Mn₃O₄ (JCPDS 80-0382). The diffraction peaks of olivine were assigned to the Mg₂SiO₄ (JCPDS 80-0944) and Fe₂SiO₄ (JCPDS 87-0317) crystal planes, which are the main constituents of olivine. Only the corresponding olivine peaks are marked as yellow stars in Figure 8b–f, in order to distinguish olivine, reactants and products. Figure 8 illustrates the various XRD patterns and confirms that the reactants had conversions at varying degrees, except for Mn₃O₄ + Na₂CO₃, which did not complete the reaction at 775 °C. However, the crushed MnO + Na₂CO₃ can almost completely react at the same T, indicating that the small particle size and high BET reactants are more likely to react, thus reducing the reaction temperature, the heat loss, and the thermal cost of hydrogen production.

3.3. Solar H₂ Production Using the Mn₃O₄/MnO/Na₂CO₃ System

The H₂ production by the different MnO_x/Na₂CO₃ reactants was repeated in the solar reactor. Temperatures of the outer reactor wall were limited to below ~1000 °C (strength limitation of the Incoloy construction material). Due to the inertia of the heliostat focusing, temperatures in the bed varied between about 760 and 790 °C (average 775 °C) and between 815 and 835 °C (average 825 °C). Mostly, the first H₂ production cycle was investigated. For the cold mix Mn₃O₄/Na₂CO₃, the reactants after H₂ production were regenerated for 6 h at an average 825 °C using pure CO₂. Results are illustrated in Figure 9.

Again, it should be remembered that the maximum production yield is 0.5 mol H₂/mol Mn-reactant. It is however clear that (i) the milled MnO/Na₂CO₃ reactant has a high production yield even at 775 °C; (ii) other reactants need to be operated at higher temperatures to achieve fair conversions; (iii) the cyclability of Mn₃O₄/Na₂CO₃ needs to be improved.

3.4. Cycling Performance

Although only a maximum of 10 redox cycles was experimentally assessed, the results clearly demonstrated the decreasing but stabilizing H₂ production efficiency with the number of cycles. The four-step reverse reaction system of the Mn₃O₄/MnO system is complicated. If an overall H₂ production efficiency of 60% is accepted for the multi-cycle operation, the number of required cycles, N_c, to break even can be determined. Thermal costs are not taken into consideration since the system needs to operate on renewable electricity (PV, wind) or on concentrated solar power. This implies a favorable ratio of the costs of the main reactants (€/ton) [12] and the economic value of the H₂ generated, with a maximum target price of 4000 €/ton.

This is reflected in the following relationship:

$${\mathrm{N}}_{\mathrm{c}}=\frac{\cos \mathrm{t} \mathrm{of} \mathrm{main} \mathrm{reactant} (€)}{4000 (€)·{\mathrm{wt} \mathrm{H}}_2 \mathrm{generated} \mathrm{per} \mathrm{unit} \mathrm{amount} \mathrm{of} \mathrm{main} \mathrm{reactant}}$$

(6)

Since 0.5 mol of H₂ can maximally be produced per mole of main reactant, experiments determine the wt% of H₂ generated per unit amount of reactant. Na₂CO₃ is by far cheaper than Mn₃O₄/MnO and is not taken into consideration. This results in an H₂ generation at a 62% reversibility of around 0.31 wt%. With a main average reactant cost of 400 €/ton, the number of cycles to break even amounts to between 30 and 40 cycles. This needs be proven in subsequent experiments.

It should moreover be reminded that extra reactant and thermal operation costs should be added, but a good product and heat management should reduce these to a minimum. The loss of reactant from the fluidized bed was minimum due to the use of metal-fiber filters at the exit of the reactor [30]. CO₂ should be separated and used in the reverse reaction. Although zeolites and activated carbon could be used after cooling the reactor gases [31,32,33], this will again increase the heating costs. It is hence proposed to use carbonation/decarbonation reactions [32] operated in series and in parallel with the main oxidation (H₂ producing) and reduction (O₂ producing) steps.

It is moreover evident that the water splitting reactions should preferably be solar-driven or electrically heated from renewable electricity sources (wind, photovoltaics). These energy balance considerations are currently being assessed.

4. Conclusions and Recommendations

Thermal water splitting by redox reactants could contribute to a hydrogen-based energy economy. The previously selected redox water splitting cycle (Mn₃O₄/MnO/NaMnO₂) was applied in isothermal conversions (electric furnace and concentrate solar reactor). Both redox water splitting systems were experimentally investigated. XRD, SEM and size distribution measurement confirm that sub-micron nano-scale reactants have a higher H₂ yield than the micron-scale ones. Up to 10 complete cycles (oxidation/reduction) were assessed. The Mn₃O₄/MnO/Na₂CO₃ cycle achieved hydrogen production efficiencies of 78% in an electric furnace at 825 °C. The milled Mn₃O₄/Na₂CO₃ cycle achieved over 80% efficiency in the first cycle, and over 77% in the tenth cycle. In the solar furnace, the highest conversion (>95%) was obtained with the Mn₃O₄/Na₂CO₃ system at 775 °C. The economic assessments revealed that at least 100 cycles would be needed to achieve competitive H₂ prices below 2 €/kg for systems with a cheap energy supply, provided CO₂ could be separated from the reactant gas and reused in the reverse reaction. The experimental and economic results recommend a further improvement of the reverse cycles of the redox systems.