Syngas Production Through H2O/CO2 Thermochemical Splitting Over Doped Ceria-Zirconia Materials

This study investigates the catalytic properties of K+ and Cu2 + /Fe3 + co-doped ceria-zirconia (CeZr) toward water and carbon dioxide co-splitting. These materials can convert separate feeds of CO2 and H2O into CO and H2. In co-splitting tests, water reacts faster on the K-Cu-CeZr catalyst with negligible CO production. The reduction of the K-Fe-CeZr catalyst occurs over two broad temperature ranges: at low temperature, only H2 is produced; whereas CO is the most abundant product at high temperature. A kinetic model was developed to get insights into the reasons of the observed selectivity toward H2 at low temperature and CO at a higher temperature. The different reaction orders in the sites fraction were evaluated for CO2 and H2O reactions, highlighting that H2 production requires a larger number of adjacent reduced sites than CO production. Three regimes were identified through the model: Regime I- H2O driven regime @T ≤ 650°C; Regime II- mixed regime @ 560 < T < 700°C and Regime III: CO2 driven regime @ T > 700°C. These results indicate the appropriate conditions for tuning H2/CO selectivity, depending on the feed composition.


INTRODUCTION
The transformation of solar energy into synthetic fuels holds huge promise for sustainable approaches to harvesting renewable energy (Nguyen and Blum, 2015;Chuayboon and Abanades, 2020;Mao et al., 2020). Thermochemical splitting cycles have been proposed as a promising sustainable option, as this approach uses concentrated solar energy to convert H 2 O and CO 2 into H 2 and CO (Costa Oliveira et al., 2018;Bhosale et al., 2019;Takalkar et al., 2019). These building blocks can be further reacted into gaseous and liquid fuels.
Among thermochemical cycles, two-step high temperature processes are the most promising (Rao and Dey, 2015), because they are less complex and can achieve a higher and more efficient theoretical solar-to-fuel energy conversion (η solar−to−fuel ) (Rao and Dey, 2015). The thermochemical splitting catalyst ceria is recognized as an important development  due to its thermal stability, high oxygen storage capacity (OSC), without any structural changes, and the fast kinetics of reduction and splitting reactions (Costa Oliveira et al., 2018;Bhosale et al., 2019;Chen et al., 2019;Takalkar et al., 2019). The thermochemical cycle can be schematically defined as follows: CeO (2−δ) +δCO 2 → CeO 2 +δCO CeO (2−δ) +δH 2 O → CeO 2 +δH 2 Reaction 1 corresponds to high temperature self-reduction under concentrated solar power, whereas reactions 2 and 3 are the splitting reactions, which restore the oxidized state of the materials and producing fuels.
The improved splitting activity of ceria-zirconia solid solutions has been related to the formation of a Zr 2 ON 2 likephase at the nanoscale level during thermal treatment in N 2 atmosphere (Pappacena et al., 2016(Pappacena et al., , 2017, which is associated with a decrease of surface Ce/Zr ratio . Good splitting performance has been related also to the formation of both bulk and surface oxygen vacancies, enhancing surface reactivity (oxygen evolution during self-reduction and splitting reactivity during oxidation) and oxygen diffusivity from the bulk to the surface and vice-versa .
While a huge amount of literature has been devoted to the thermochemical splitting of bare H 2 O or CO 2 , little is known about their simultaneous splitting (Furler et al., 2012;Lorentzou et al., 2014Lorentzou et al., , 2017Falter and Pitz-Paal, 2018;de la Calle and Bayon, 2019;Tou et al., 2019;Hao et al., 2020). It is worth noting that co-splitting has been generally defined as concurrent or combined splitting, but it should be better defined as competitive splitting; as a matter of fact, CO 2 and H 2 O compete for the same reduced sites, whereas the overall fuel production depends on the reduction extent of the material.
Recently, we proved that potassium cation doping improved Ce 0 . 75 Zr 0 . 25 O 2 splitting performance toward CO 2 or H 2 O (Landi, 2019). Alkali metal ions doping as well as transition metals cations co-doping, such as Cu(II) and Fe(III), provided the formation of both bulk and surface oxygen vacancies, significantly enhancing the splitting performance. Interestingly, K-addition boosted the oxidation kinetics, in agreement with the results obtained in the three step cycles (Charvin et al., 2009). Recently, Takalkar et al. showed a positive effect of Li + cation on the CO production during CO 2 splitting cycles (Takalkar et al., 2020b).
In this work, K-M-doped ceria-zirconia solid solutions (M = Cu 2+ and Fe 3+ ), containing 5% mol/mol of both K and M cations, were tested as catalysts for carbon dioxide and water co-splitting. Furthermore, a kinetic model was developed and used to get insights into the active sites of the catalysts and their activity toward H 2 O and CO 2 splitting, to simulate the splitting performance as well as the composition of produced gases at different oxidation temperatures.

Materials Preparation
Reagents were purchased from Sigma-Aldrich and used as received. The material synthesis was carried out by employing the following reagents: cerium (  . Bare ceria-zirconia, as well as doped and co-doped materials, were prepared according to the co-precipitation method (Luciani et al., 2018;Landi, 2019). Stoichiometric amounts of precursor salts were dissolved in 75 ml of bi-distilled water and stirred for 3 h. Then, solutions were heated in an MW oven (CEM SAM-155) until a homogeneous gel was obtained. The calcination in air at 1,100 • C for 4 h was then carried out to decompose nitrates and to obtain the oxides. Table 1 summarizes the composition of produced materials as well as the label used.

H 2 O/CO 2 Splitting and Co-splitting Tests
The experimental equipment and experimental details described above were used to assess the splitting ability of the samples under investigation (Luciani et al., 2018Portarapillo et al., 2019). Temperature programmed reduction (TPR) and oxidation (TPO) were carried out on powdered samples (500 mg; 170-300 µm), placed in a tubular quartz reactor, and inserted into an electric tubular furnace (Lenton). The catalyst temperature was measured by a K-type thermocouple placed inside a tube co-axial with the reactor.
Oxygen, hydrogen, and carbon monoxide amounts (mol/g), n O2 , n H2 , and n CO respectively, were calculated as follows: where m cat (g) is the sample mass, P (atm) and T (K) are standard pressure and temperature, R (atm·l·mol −1 ·K −1 ) is the ideal gas constant, F red and F ox (l(STP)/s) are the flow rates of permanent gases during H 2 treatment and splitting test, respectively, n H2_cons,i and n H2_prod,i are consumed and produced hydrogen moles during the H 2 reduction treatment and splitting test, respectively. The area H2_cons,i , and area H2_prod,i (%·s) are the calculated areas of the reduction and oxidation H 2 profiles, respectively. n CO_prod,i is the amount of carbon monoxide produced during the splitting test, and area CO_prod,i (%·s) is the calculated area of the oxidation CO profile. The n O2 does not represent a real molecular oxygen evolution but oxygen subtracted from the materials.
The reduction degree (x red,i ) after the current reduction or oxidation step (i) was calculated as follows: and as in the following for oxidation steps: where n O2,max is the maximum O 2 amount (mol/g) that could be evolved, if reducible cations were reduced to their lowest oxidation state (Ce 4+ to Ce 3+ , Fe 3+ to Fe 2+ , Cu 2+ to Cu 0 ; corresponding values are reported in Table 1). x red,i−1 is the reduction degree of the previous (oxidation or reduction) step. Therefore, x red,i does not simply quantify the current fraction of Ce 3+ ions on the overall content of Ce atoms in the system. Oxidation yield (α i ) is calculated through produced CO (mol/g) during the current re-oxidation step and the corresponding O 2 evolved amount (mol/g) during the previous reduction step: Reduction yield (β i ) is calculated through the overall O 2 amount (mol/g) produced during the current reduction step and CO evolved amount (mol/g) during the previous oxidation step:

Kinetic Model
A kinetic model was developed to get insights into the role of the superficial reaction on the selectivity in co-feed splitting.
The model is based on the Polanyi-Wigner equation under the Redhead approximation, which states that catalytic sites do not interact. The reactor is modeled as a continuous stirred tank reactor (CSTR).
The reaction mechanism is a redox pathway involving active sites σ: σ red as the reduced form and σ ox as the oxidized form.
According to these hypotheses, the reaction steps are: Sites fractions are defined by the following equations: The reaction rates of steps (10-11) are: where θ red is the reduced sites fraction, n i are the reaction orders, Y i are the reactants molar fractions, and k i are the kinetic constants which are evaluated as follows: where E i are the activation energies, and k 0 i are the frequency factors.
The unsteady balance equations on reduced sites fraction for CO 2 splitting, H 2 O splitting, and co-splitting are respectively: The unsteady balance equations on gaseous molar fractions are: where τ is the residence time equal to the experimental one. The initial conditions are: Differential equations were numerically solved using the Eulero explicit method.
To analyze the fitting quality, the differences between the experimental data and the model were evaluated through the root mean square error (SRMSE), normalized for the maximum value of the curve of the flow test for each step: Model and experimental curves were considered in good agreements for SRMSE values ≤0.045.

Experimental Results
Previous research has reported on H 2 /H 2 O cycling, outlining that it can be used to define the splitting properties of materials (Landi, 2019). Results of thermogravimetric (TG) measurements (reduction under inert Ar atmosphere; oxidation under 40 vol.% CO 2 /Ar) and H 2 O splitting tests on H 2 -reduced samples were successfully compared (Al-Shankiti et al., 2013;Pappacena et al., 2016Pappacena et al., , 2017Luciani et al., 2018).
Results by Landi (2019) showed that K-addition to bare ceria-zirconia and materials doped with transition metals (K-Fe-CeZr and K-Cu-CeZr) significantly enhance the evolved oxygen amount and lower the reduction onset temperature. These reduction profiles obtained for the co-splitting reaction are consistent with those reported by other studies (Landi, 2019) and will, therefore, not be further discussed.
Figure 1 (bottom) shows H 2 profiles measured in the H 2 O splitting tests. The profiles refer to the stable performance obtained after the multiple reduction/oxidation cycle (Landi, 2019). One wide peak was detected over the K-Cu-CeZr sample, while two peaks were found over K-Fe-CeZr catalysts.
The low temperature oxidation peak could be addressed to the contribution of surface and sub-surface layers and/or less ordered bulk structures to splitting reactions, while the high temperature peak could be related to the oxidation of bulk ceria-zirconia. Figure 1 shows H 2 and CO concentration profiles as obtained in CO 2 /H 2 O co-splitting tests and separate H 2 O and CO 2 splitting experiments over co-doped materials. Figure 2 also shows the fuel amount produced over K-Fe-CeZr and K-Cu-CeZr catalysts.
K-Fe-CeZr sample shows two oxidation peaks independently from the oxidizing stream. In CO 2 splitting, the former peak (450 • C) occurs at a higher temperature than that found in H 2 O Frontiers in Energy Research | www.frontiersin.org splitting. This result suggests that the H 2 O splitting reaction is faster than the CO 2 reaction.
In H 2 O/CO 2 co-splitting tests, the H 2 O was reduced first. However, H 2 O started reducing at 200 • C reaching the peak at 330 • C. This peak temperature is higher than bare H 2 O splitting conditions. This behavior can be addressed to the CO 2 adsorption on the surface, inhibiting H 2 O adsorption and further reaction. Actually, in our previous papers, we showed that CO 2 desorption on ceria-based materials occurs at about 200 • C (Di Benedetto et al., 2013;Barbato et al., 2016).
Moreover, CO 2 feed results in a lower produced H 2 amount ( Figure 2C). The second oxidation phenomenon occurs 2 | Amount (in µmol/g) of released oxygen (as O 2 ) and produced H 2 during H 2 O splitting tests, reduction degrees after each step (x red ), oxidation yields (α), and reduction yields (β ). at the same temperature independently from the oxidizing stream (Figures 1, 2). As reported above, this peak is related to bulk oxidation. This reaction is controlled by the bulk-to-bulk oxygen diffusion and is thus not influenced by the chemical nature of the oxidant. However, during H 2 O/CO 2 co-splitting only CO production is (unexpectedly) detected, suggesting a key role of gaseous molecules on the reaction pathway. CO 2 splitting on K-Cu-CeZr (Figure 1) features one CO production peak at about 600 • C, confirming lower activity toward CO 2 than the K-Fe-CeZr sample, which was also detected by TG analysis (Landi, 2019). In the H 2 O/CO 2 co-splitting tests, the presence of CO 2 shifts H 2 production to higher temperatures (Figures 1, 2D) and reduces H 2 amounts (Figures 2B,D), as reported for the K-Fe-CeZr sample. Furthermore, CO production is negligible, probably because the material is fully oxidized at a temperature lower than that of CO 2 activation.

Kinetic Model of K-Fe-CeZr
The results of co-splitting tests (Figure 1) show that the H 2 O splitting reaction is faster than CO 2 . However, at high temperatures, the H 2 O splitting reaction is prevented and CO 2 splitting is favored. To understand the observed behavior, we FIGURE 2 | Quantitative analysis of splitting tests under different oxidants. Partial and overall amounts of H 2 + CO on K-Fe-CeZr (A) and K-Cu-CeZr (B); H 2 (blue symbols) and CO (red symbols) amounts during H 2 O splitting ( ) co-splitting ( ), and CO 2 splitting ( ) on K-Fe-CeZr (C) and K-Cu-CeZr (D).

FIGURE 3 | (A)
Experimental and model CO production (low temperature peak, bare CO 2 splitting) and (B) the corresponding surface site evolution calculated by the kinetic model. developed a kinetic model. As reported in section "Kinetic Model, " only the surface reactions have been modeled. The low temperature peaks of H 2 O and CO 2 splitting tests were used as experimental data to calculate the kinetic parameters. Figures 3, 4 show the experimental and model curves as obtained in H 2 O and CO 2 splitting tests, respectively. The results reported in Table 3, show a low value of the SRMSE, suggesting that the fitting is good.
From the fitting results, we conclude that the activation energy of the CO 2 splitting reaction is higher than that of H 2 O splitting. This enables the H 2 O splitting reaction to proceed at a lower temperature than CO 2 , as indicated by the results of this experiment.
We can also note that the H 2 O oxidation reaction is in the first order for active sites concentration (n θ _H2O = 1); whereas the CO 2 oxidation rate exhibits an order lower than 1 with respect to active sites concentration (n θ _CO2 = 0.75). This result suggests that CO 2 needs a lower number of nearby reduced sites, as expected due to the different structure of the CO 2 molecule.
To study the competition between the two reaction rates we defined the parameter R:  Three regimes can be identified: Regime I-H 2 O driven regime: @T ≤ 560 • C, the reaction rates do not intersect and R < 1 in the whole θ range. H 2 production prevails over CO production.
Regime II-mixed regime: @ 560 < T < 700 • C, r CO2 and r H2O intersect. At low θ values R > 1, at high θ values R < 1. Reaction rates are of the same order of magnitude, suggesting a co-production of carbon monoxide and hydrogen.
Regime III: CO 2 driven regime @ T > 700 • C reaction rates do not intersect and R > 1 in the whole θ range. CO production prevails over H 2 production.
Isothermal simulations were performed to compute the amount of H 2 and CO produced as a function of temperature. Figure 6 shows the ratio between CO and H 2 produced amounts FIGURE 5 | Surface reaction rates of H 2 O and CO 2 splitting reaction (left) and their ratio R (right) as a function of decreasing reduced sites fraction on K-Fe-CeZr at different temperatures as calculated by the kinetic model. (F, Eq. 32) as a function of temperature.
In this figure, the three regimes are also identified. It is worth noting that F = 1 at 670 • C and inversion occurs at higher temperatures. From these results, we conclude that by changing the temperature, the content of H 2 and CO in the output current can be modulated.

CONCLUSION
In this work, ceria-zirconia was co-doped with transition metal (Cu 2+ , Fe 3+ ) and potassium cations. We studied the co-doped ceria-zirconia to determine whether they act as catalysts for water and carbon dioxide co-splitting. All the investigated materials were able to convert separate feeds of CO 2 and H 2 O into CO and H 2 , respectively. A single oxidation peak was detected on the K-Cu-CeZr catalyst, independently from the oxidant. Furthermore, the water splitting reaction rate appeared faster. Two oxidation peaks were detected on the K-Fe-CeZr catalyst for both the oxidants; the water reaction rate appeared faster at low temperature, while the high temperature peak seemed independent from the oxidant, suggesting that the reaction rate is limited by the oxygen diffusion from the surface to the bulk.
Co-splitting tests were carried out as temperature programmed oxidations. On the K-Cu-CeZr catalyst, water reacted faster and a negligible CO production was detected. the H 2 production peak shifted to a higher temperature, probably due to CO 2 adsorption on the catalyst surface, thus blocking active sites. The K-Fe-CeZr catalyst behavior was more complex. Two oxidation peaks were detected at low and high temperatures. For the low temperature, only H 2 was produced in agreement with the faster H 2 O splitting kinetics measured during the separated H 2 O and CO 2 splitting. In contrast, almost 100% selectivity to CO 2 splitting was detected at high temperatures.
A kinetic model was developed to understand this behavior. This model revealed that the H 2 O splitting reaction featured a higher reaction order in terms of sites fraction compared to CO 2 . This suggests that H 2 production requires a larger number of adjacent reduced sites. By comparing the calculated surface reaction rates at different temperatures, three regimes were identified. At temperatures below 560 • C H 2 production is faster than CO production independently from the surface reduced sites fraction; accordingly, isothermal co-splitting simulations showed a preeminent hydrogen production. At temperatures between 560 and 700 • C, CO 2 and H 2 O reaction rates are comparable and the pre-eminence depends on both temperature and surface reduced sites fraction. In this temperature range, co-production of CO and H 2 occurs and their ratio can be tuned by an opportune choice of the co-splitting temperature. At temperatures higher than 700 • C CO production is faster than H 2 production independently from the surface reduced sites fraction, accordingly, isothermal co-splitting simulations showed a preeminent carbon monoxide production.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

AUTHOR CONTRIBUTIONS
GLa and GLu prepared the materials and performed the experiments. AD developed the model and performed the simulations. GLa, GLu, and AD contributed to the planning and interpretation of results, and the writing of the manuscript. All authors contributed to the article and approved the submitted version.