Revealing the detrimental CO₂ reduction effect of La_0.6Sr_0.4FeO_3-δ-derived heterostructure in solid oxide electrolysis cells

Summary

Solid oxide electrolysis cells hold unique Faraday efficiency and favored thermodynamic/kinetics for CO₂ reduction to CO. Perovskite oxide–based composite materials are promising alternatives to Ni-based cermet electrodes in SOECs. However, contrary results of the electrocatalytic activity over single-phase perovskite oxide exist and the rationale of the negative effect is not well revealed. In this work, two-phase perovskite materials with various complementary properties and unique interfaces are self-assembled, which was realized by “subtractive” defect-driven phase separation. The obtained heterostructure electrodes showed reduced performance over that of single-phase materials although the cyclic stability was improved. The main reasons for the performance degradation are the decrease of electrical conductivity, oxygen vacancy concentration while increasing the average valence state of B-site Fe cations, and electrode surface Sr aggregation. This work highlights the self-assembly method and insight into the rational design and synthesis of active electrodes/catalysts for CO₂ conversion in solid oxide cells.

Graphical abstract

Highlights

• •Synthesis of two-phase perovskite composite by the self-assembly method
• •The negative effect of heterogeneous structures on CO₂ electrolysis
• •Reduced oxygen vacancy and conductivity, increased Fe valence and Sr aggregation
• •Kinetics CO₂ reduction on heterogeneous perovskite oxide in SOECs

Chemistry; Electrochemistry

Introduction

Solid oxide electrolysis cell (SOEC), a device that converts electrical energy and thermal energy into chemical energy, has the advantages of high efficiency and environmental protection.^1,2 In particular, the use of SOEC for CO₂ electrolysis to produce sustainable fuels/chemicals can not only stores renewable energy but also is a perfect solution to complete the task of carbon neutrality and to alleviate the greenhouse effect.^3,4 However, the electrolytic efficiency of SOEC is limited by the catalytic activity of the cathode due to the linear nature of CO₂ molecules and high C=O binding energy, so over the years, researchers have conducted many studies on the development of highly active electrode materials for efficient CO₂ activation and reduction.^5,6,7 In recent years, Ni-based cermet cathodes like Ni-YSZ (nickel-yttrium stabilized zirconia) were first reported because of their high electrical conductivity and superb catalytic activity for CO₂ reduction reaction (CO₂-RR).^8,9 However, it still suffers from a series of problems, such as carbon deposition and grain aggregation, which makes the long-term application of Ni-based cermet materials in SOEC problematic.^10,11,12

To date, perovskite materials with mixed ionic-electronic conducting properties, excellent resistance to carbon deposition, and good cycle stability have received constant attention as alternatives to Ni-cermet electrodes for CO₂ reduction in SOECs.^13,14 Representative materials such as La_0.2Sr_0.8TiO_3+δ (LST),¹⁵ Sr₂Fe_1.5Mo_0.5O_6-δ (SFM),^16,17,18 La_0.75Sr_0.25Cr_0.5Mn_0.5O_3-δ (LSCM),^19,20 and (La, Sr)FeO_3-δ (LSF)^21,22 have been well explored. Besides, many methods have been proposed to increase the catalytic activity, such as A-site,^20,23 B-site doping,^24,25 and anion element doping,^21,26 as well as cation deficiency projects.²⁷ However, there is always a limit to the modification of single-phase materials and the performance gap remains compared with the Ni-based electrode. Alternative electrodes with improved CO₂ chemical adsorption, activation, and facilitated ionic/charge transfer capability are still needed.

Recently, heterostructure catalysts generally composing of active material and various functional additives demonstrated improved electrocatalytic performance toward CO₂-RR.^27,28,29 Researchers tried to develop a composite electrode material with a heterogeneous interface based on the above materials.^30,31 For example, the heterogeneous interface between perovskite and metal constructed by in situ exsolution possesses both good ion conductivity of perovskite and good electrical conductivity of metals.^32,33 Previous works in our group also demonstrated that the constructed heterogeneous interface composed of carbonate and perovskite though in situ A-site deficiency and surface carbonate deposition induced by the addition of low valence alkali metal ions with high basicity and low electronegativity enhanced CO₂ adsorption, dissociation process, and CO desorption primitive steps.^20,23 Furthermore, many works also adopted the method of self-assembly to construct heterogeneous interfaces to combine the advantages of each phase. Demont et al.³⁴ doped large radius W or Mo into Ba_0.5Sr_0.5Co_0.7Fe_0.3(BSCF) to form B-site cation ordered double perovskite and disordered single perovskite oxide phases. Song et al.³⁵ synthesized a nanocomposite Sr_0.9Ce_0.1Fe_0.8Ni_0.2O_3–δ, which was self-assembled to a single perovskite phase and a Ruddlesden–Popper phase, accompanied by surface-decorated NiO and CeO₂ minor phases, both showed improved catalytic activity toward the targeted reactions^36,37 However, the process of modification of the composite structure may bring negative effects. Zhu et al.¹⁰ studied that the Fe and Ni cations in the Sr₂Fe_1.5-xNi_xMo_0.5O_6-δ (x = 0, 0.05, 0.1, 0.2) that are exsolved and aggregated at the surface of perovskite oxide and reduced to metal nanoparticles did not improve the CO₂ reduction performance as expected because of the severe SrCO₃ formation on the surface of the perovskite. Luo et al.³⁸ also mentioned that the formation of B-site vacancies in perovskite scaffolds because of metal nanoparticle exsolution was the major contributor to the degradation of perovskite activity for CO₂ splitting, while the B-site supplement endows appealing electrochemical performance of both stability and activity for CO₂ reduction.

Thus, self-assembly has been verified as an effective method for synthesizing perovskite oxide-based composite nanocrystals.^38,39,40 However, the performance of resulting heterogeneous catalysts is often difficult to be predicted. The originality of the negative results should be carefully discussed and more general cases should be reported. In this work, we synthesized a heterostructural phase material containing single-phase (SP) perovskite and Ruddlesden–Popper (RP) phase perovskite from the precursor of La_0.6Sr_0.4FeO_3-δ perovskite oxide by intentionally adjusting B-site defects/stoichiometry. Different from the previous strategy of “doing addition” (Adding different elements), we use the strategy of “doing subtraction” in situ to construct such a heterostructure. In this way, we exclude the influence of conventional in situ exsolved metal particles and focus on the role of SP-RP oxide/oxide heterostructures. It is found that the resultant SP-RP heterostructures did not bring a positive effect on CO₂ reduction performance. On the contrary, a detrimental effect was revealed with the increased B-site non-stoichiometry even though the operational stability is improved. The rationale for the reduced CO₂ splitting performance on the oxide-oxide heterostructural interface is carefully analyzed and discussed based on extensive material physical and chemical properties investigation.

Results and discussion

Powders’ XRD and their refinement results are shown in Figure 1. The as-synthesized La_0.6Sr_0.4FeO_3-δ (LSF) powders exhibit a pure single perovskite phase with an R-3c:H space group (Figure 1C), while the samples of La_0.6Sr_0.4Fe_0.9O_3-δ (LSF9) and La_0.6Sr_0.4Fe_0.8O_3-δ (LSF8) exhibit one additional phase besides LSF, a Ruddlesden-Popper perovskite phase (RP, PDF card No. 71–1774) appeared with the increased B-site defects (Figures 1D and 1E). The mass ratios of the RP phase in LSF9 and LSF8 are 10.56 wt. % and 27.87 wt. %, respectively (Figures 1C–1E and Table S1). The RP phase content of the samples, which were synthesized by self-assembly, increases gradually with the decrease of the B-site element non-stoichiometry. Due to excessive B-site defects, the cubic perovskite molecule is changed into a layered RP phase during the phase formation process (Figure 1B). According to the detected phase in XRD results, the self-assembly processes can be expressed by the following Equation 1:

$${La}_{0.6}{Sr}_{0.4}{Fe}_{1-x}{O}_3\to {yLa}_{0.6}{Sr}_{0.4}{FeO}_{3-\delta }+{zLa}_{1.2}{Sr}_{0.8}{FeO}_{4+\delta }$$ (Equation 1)

Figure 1.

Phase structures and lattice texture of prepared powders samples

(A) XRD, (B) schemetic illustration model of the phase structural transformation through self-assembly process.

(C–E) the corresponding XRD refinement results, and (F) high-resolution TEM image and the selected area lattice texture.

Where 1-x = y + z based on the Fe element stoichiometry, together with the phase with “O₄” represents the RP perovskite phase characteristics in a layered feature along the c-axis (Figure 1B). The theoretical stoichiometric ratio of La and Sr in Equation 1 is 6:4, but in the actual self-assembly process, their stoichiometric ratio may change to diversify the products. The above results indicated that the composite structural materials were successfully synthesized by intentional B-site stoichiometry adjusting, “doing subtraction” instead of classic “doing addition”. Although the composition and proportions of each phase vary, their microstructures are very similar (Figure S1), which will help exclude the influence of microstructure on CO₂ electrolysis.

To further verify the complete heterostructural characteristics of LSF8, the texture structure of LSF8 was visually characterized by TEM (Figure 1F). Two different phases can be seen in the image; the lattice spacing of 0.351 nm represents the (111) planes of single-phase LSF (SP-LSF) and 0.251 nm represents the (112) planes of Ruddlesden-Popper phase (RP-LSF) (more information of lattice spacing parameters of SP and RP phase can be referred to Figure S2), which is consistent with the XRD refinement results (Figures 1C–1E). Between the SP phase and RP phase in the TEM image, an apparent consecutive heterogeneous interface can be found. Those further illustrate the successful synthesis of composite materials and the presence of heterogeneous interfaces with connected channels between different phases.

We continue to evaluate the catalytic activity of the various electrodes for CO₂ reduction in SOECs based on LSGM electrolyte-supported configuration with LSCF-SDC anode and LSFx cathodes under a 100% CO₂ atmosphere. The electrochemical performances are shown in Figures 2A–2E. As can be seen from Figure 2A, with the reduction of the Fe stoichiometric ratio at the B-site, the CO₂ electrolysis performance of single cells deteriorates. The exact CO₂ electrolysis current densities of single cells with the single phase LSF, LSF9, and LSF8 electrodes were 1.22, 0.94, and 0.77 A cm⁻² at 1.5 V and 800°C, respectively. The corresponding current densities at other temperatures (750 and 850^°C) shared a similar tendency (Figure S3). Moreover, the EISs of single cells with different cathode materials were also investigated. Generally, the ohmic resistance (Ro) is the spectra intercept with the real axis at high frequency, while the total electrode polarization resistance (Rp) is calculated from the distance between the two intercepts of the depressed arcs with the real axis.¹⁶ As shown in Figure 2B and Figures S4 and S5, the Ro of LSF, LSF9, and LSF8 are 0.4, 0.52, and 0.61 Ω cm² respectively, and the Rp is 0.13, 0.16, and 0.21 Ω cm², respectively. Both the ohmic resistance and polarization resistance of single cells increase as the proportion of the RP phase increases. Generally, ohmic resistance is mainly composed of the resistance values of the electrolyte and is also related to the electronic conductivity of the electrode.⁴ Since electrolytes' thickness and cells' manufactory processes are the same, the increase in ohmic resistance should be due to the insufficient electrical conductivity of the applied cathodes. For verifying such a hypothesis, the electrical conductivities of LSF, LSF8, and RP-LSF samples in air (oxidizing atmosphere) and 95: 5 CO₂-CO (reducing atmosphere) were measured. Prior to this, the phase structure of different powders at calcination temperature (1300°C) are also verified in Figure S6 to ensure the reliability of the conductivity test. Compare with fresh samples the powders calcined at 1300°C just exhibited slightly different crystallinity and had no effect on the conductivity test.⁴¹ The electrical conductivity of LSF and LSF8 samples in an oxidizing atmosphere and reducing atmosphere both reveal a transition from semiconductor-like to metal-like behavior^42,43,44 (Figures 2C and S7), i.e., the electrical conductivity first increases with increasing temperature until it reaches a maximum, then decreases with temperature subsequently. In addition, the electrical conductivity of RP-LSF is almost negligible compared to LSF. Those results confirm that the decrease in LSF8 conductivity is precisely due to the introduction of RP-LSF. As expected, the electrical conductivity of LSF8 is lower than LSF in both atmospheres, leading to the increased Ro in the CO₂ electrolysis condition. Such a phenomenon is consistent with the conclusions from other articles.⁴⁵ In addition, the electrical conductivity of RP-SLF, in both oxidizing and reducing atmospheres increases with the increase in temperature (Figure S7), showing a simple semiconductor-like behavior. Perhaps it is the absence of metal-like properties that leads to the extremely low conductivity of RP-LSF.

Figure 2.

Electrochemical performance of SOECs based on different cathode materials in pure CO₂ atmosphere and at 800°C

(A) I-V curves.

(B) EIS curves at 1.3V, (C) electrical conductivity in 95: 5 CO₂-CO atmosphere, (D) short-term stability, (E) long-term stability at 1.1V, 800°C, and (F) CO production and faradaic efficiency of LSF and LSF8 with different current densities at 800°C.

The electrochemical stability of the single cells was also verified. The short-term CO₂ electrolysis stability at 1.5, 1.4, 1.3, and 1.2 V on single cells with LSF, LSF8, and LSF9 fuel electrodes and each for 20 min is shown in Figure 2D. We can find that LSF showed higher resulting initial current density at voltages above 1.3 V over other samples, and its performance severely decreased with the applied time. When being operated at 1.2 V, the electrolytic current density of the LSF8 electrode under the same voltage exceeded the LSF. Among these, the single cells with an LSF fuel electrode exhibited exaggerated performance degradation, while the single cell with an LSF8 fuel electrode showed wonderful stability under the same circumstances. The I-V curves (Figure S8) after short-term CO₂ electrolysis stability can further verify this conclusion. To further compare the difference in stability, a comparison of long-term CO₂ electrolysis stability tests with LSF and LSF8 fuel electrodes was also performed and the results are shown in Figure 2E. The current density of the single cell with the LSF8 fuel electrode dropped from 0.37 A cm⁻² to 0.31 A cm⁻² at 1.1 V for 100 h, while that of the LSF fuel electrode dropped from 0.42 A cm⁻² to 0.19 A cm⁻² at the same condition. Furthermore, the performance of a single cell with an LSF electrode decayed below the LSF8 electrode within 1 h, which is consistent with short-term stability results. Apart from the operation stability, the production rate and faradaic efficiency are also considered for the practical application of CO₂ electrolysis. As shown in Figure 2F, the CO production rate of LSF8 (5.50, 7.31, and 7.42 mL min⁻¹ cm⁻²) has surpassed LSF (4.84, 5.65 and 6.35 mL min⁻¹ cm⁻²) at 0.25, 0.275 and 0.3 A. The faradaic efficiency of LSF8 is almost 100% in most circumstances³² while LSF is only about 80%. The reduced Faraday efficiency of CO₂ reduction on the LSF electrode surface may be ascribed to the additional reaction pathway, such as a CO₂ completely decomposed reaction: ${\text{CO}}_2\to \mathrm{C}+{\mathrm{O}}_2$. Anyway, despite the reduced catalytic activity, better stability, CO production rate, and faradaic efficiency suggested the superiority of composite structure materials in potential industrial applications.

In Figure 2B, we can see that the electrode polarization resistance (Rp) gradually increased with the applied B-site defects/non-stoichiometry. Although the different sources are ascribed to the cathode, polarization resistance from the anode complexed its deconvolution. To further evaluate the influence of phase transformation on the catalytic activity in terms of the CO₂ reduction reaction, the electrochemical impedance spectra (EIS) data using LSGM electrolyte-supported symmetrical cells were acquired. Symmetrical cells with LSF/LSF9/LSF8 electrodes were operated at open circuit voltages (OCV) conditions at different temperatures and atmospheres. As shown in Figure 3A, the R_p values of LSF, LSF9, and LSF8 electrodes are 0.33, 0.39, and 0.52 Ω cm⁻² at 800°C in 1: 1 CO₂: CO atmosphere respectively. Obviously, with the decrease of the stoichiometric ratio of the B-site element, the R_p values of heterostructural electrodes increased, which is consistent with the results obtained in single cells. DRT is a useful tool to analyze the complicated electrochemical process which contains different sub-steps.³ According to literature,^{17,21,24,32,42,46} the peaks of CO₂ reduction in SOECs can be divided into three peaks: high frequency (HF, 10⁴–10⁵ Hz) peaks which is linked to O²⁻ transportation at the interface between the electrode and electrolyte, intermediate frequency (IF, 10³–10⁴ Hz) peaks which are associated with the charge transfer or CO₂ activation process in electrode and low frequency (LF, 10⁰–10³ Hz) peak which indicates gas-related process including adsorption, activation, and dissociation process. As shown in the DRT of LSF at 800°C in 1: 1 CO₂: CO, the spectra show the standard three peaks (Figure 3D). When the stoichiometric ratio of the B-site Fe element in LSF is reduced to 0.9 and 0.8, all frequencies exhibit resistance to a higher value, and a brand-new low-frequency peak P4 gradually emerges (Figures 3D–3F). The reason for the increase of HF and LF in LSF9 and LSF8 is the decrease in their electrical conductivity (Figure 2C), increasing the resistance values of the HF and IF processes representing the O²⁻ and e⁻ transportation in the electrode and the interface between the electrode and electrolyte. The variation of LF can be attributed to the changes in CO₂ chemical adsorption and activation ability caused by the intervention of the RP phase and heterostructure. We can also find that the low frequency part still occupies the dominant position and is the rate-determining step of the whole CO₂-RR process.

Figure 3.

Electrochemical performances of electrolyte-supported symmetrical cells with LSF, LSF9, and LSF8 electrodes at OCV

(A) EIS at 800°C.

(B) EIS of LSF8 under different applied gas atmospheres.

(C) EIS of LSF8 at different temperatures and (D–F) their corresponding DRT curves.

For further analysis, EIS and DRT data of LSF, LSF8, and LSF9 symmetrical cells under different applied gas atmospheres and voltages were also received (Figure S9). The EIS values of the three samples decrease with the increase of the temperature and CO₂ partial pressure. For LSF, all processes are influenced by temperature, while HF and IF processes, which signify the O²⁻ and e⁻ transportation, are almost independent of the CO₂ partial pressure (Figure S9). For LSF9 and LSF8, all of their processes are also influenced by temperature, but unlike LSF, the HF and IF are also affected by the CO₂ partial pressure (Figures 3 and S9). This result illustrates that the RP phase has the same reaction process as the SP phase, but due to the difference in catalytic activity, it changes the law of change of P1 and P2 peaks under different CO₂ partial pressures. However, the characteristic peak P4 of LSF8 composite material is not affected by the CO₂ partial pressure which is affected by the activation process in CO₂-RR and CO oxidation processes, so P4 may be summarized as a pure adsorption peak occurring on the RP phase and the interface between two phases.⁴⁷ It’s worth noting that, compared with the LSF8 sample, the variation of DRT of the LSF9 sample is more similar to LSF, which indicates that LSF8 has the complete characteristics of heterogeneous structure and composite materials. The content of the RP-LSF in LSF8 favors forming a continuous channel and may play an important role over that in LSF9. Furthermore, the P4 peak is greatly influenced by temperature (Figure 3F), which further proves that is related to the adsorption process.

To further explore the reasons for the worsening electrochemical performance of the heterostructure material electrodes in single cells and symmetrical cells, the physical and chemical properties of the surface and bulk phase of LSF, LSF9, and LSF8 are explored by XPS, TGA, and CO₂-TPD techniques. As shown in Figure 4A, the Fe 2p_3/2 spectrum can be deconvoluted into 709.6, 710.8, and 712.9 eV three peaks, each represents the Fe²⁺, Fe^3+, and Fe⁴⁺. The calculated average valence states of Fe in LSF, LSF9, and LSF8 are 2.562, 2.630, and 2.684, respectively (Table S2), which is reasonable since it helps to make the charge balance with the reduced Fe content in bulk materials. The electronic conduction in perovskite structure is realized via B-site (transition metal) lattice ions through overlapping B-O-B bond via Zerner double exchange process as shown in the Equation 2 ¹³:

$${\mathrm{B}}^{\mathrm{n}+}-{\mathrm{O}}^{2-}-{\mathrm{B}}^{\left(\mathrm{n}+1\right)+}\to {\mathrm{B}}^{\left(\mathrm{n}+1\right)+}-{\mathrm{O}}^{2-}-{\mathrm{B}}^{\left(\mathrm{n}+1\right)+}\to {\mathrm{B}}^{\left(\mathrm{n}+1\right)+}-{\mathrm{O}}^{2-}-{\mathrm{B}}^{\mathrm{n}+}$$ (Equation 2)

Figure 4.

Physicochemical properties of LSF, LSF9, and LSF8 powders

(A–C) XPS of Fe 2p_3/2, O 1s, and Sr 3d spectra.

(D) TG curves.

(E) CO₂-TPD curves.

(F) Raman of LSF and LSF8 surface after long-term stability test.

Therefore, the charge density around the B-site atoms and the overlap of B-O will determine the electrical conductivity of perovskite oxide. So, the average valence of Fe can provide a strong basis for the variation of electrical conductivity of different samples. To maintain charge equilibrium, as the stoichiometric ratio of the Fe element in the B-site decreases, the average valence of Fe continues to rise which means a decrease in the density of the surrounding electrons. This could also explain the reason for the decrease in conductivity in Figure 2C and reduced CO₂ reduction activity.^17,41,42 Figure 4B exhibits the XPS spectra of O 1s in as-synthesized LSF, LSF9, and LSF8 samples, which can be deconvoluted into four peaks (Figure 4B), containing surface lattice oxygen (∼528.5 eV, O_lat), adsorbed oxygen (∼530.3 eV, ∼531.4 eV, O_ads) and hydroxides and carbonates adsorbed in the molecular (∼532.8 eV, OH⁻/CO₃²⁻).^20,48 The ratio of adsorbed oxygen in the total O species in LSF8 and LSF9 is less than LSF sample (Table S3), generated by the increase of the average valence states of Fe, which we think is another reason for the degradation of the electrochemical properties of composite materials. In addition, Figure 4C shows the Sr 3d spectrum of three samples, the lattice one indicates Sr in the perovskite oxide lattice, whereas the surface one includes Sr present on the surface state like SrCO₃. The ratio of Sr_lattice/Sr_surface decreased from 6.79 in LSF to 3.00 in LSF8 (Table S4), The non-conductive SrCO₃ enriched on the surface will occupy the adsorption site of CO₂ and also affect the electrical conductivity of the samples as well as charge transportation process. Therefore, the analysis of the Sr 3d spectrum can also explain the reason for the reduction of electrical conductivity and the possible reduced CO₂-RR electrochemical performance. The results of XPS indicate that the decrease in catalytic activity of the sample is the result of the increase of the average valence state of Fe, the decrease of surface oxygen vacancy, and the increase of surface SrCO₃ segregation.

Figure 4D shows the TG analysis results of three samples. The mass loss of LSF is far more than LSF9 and LSF8 at the temperature of 850°C. In other words, the intentional B-site defects lead to the reduction of oxygen vacancy content that is supposed to the active site for CO₂ chemical adsorption. Consequently, CO₂-TPD measurements were performed and the results are shown in Figure 4E. The peaks in the low temperature range can be summarized as a physical desorption signal, while peaks occurring in the high temperature range can be considered signals of chemical desorption.²¹ All physical and chemical desorption signals of LSF8 are weaker than LSF. Specifically, chemical desorption decreases from 472 μmol g⁻¹ for LSF to 170 μmol g⁻¹ for LSF8 (Figure S10). This decrease in CO₂ adsorption capacity directly explains the reason for the large increase in low-frequency peak area in DRT figures which is related to gas adsorption and activation processes.

Although LSF-based materials have excellent catalytic activity in CO₂ reduction, they have always had stability problems in CO₂ electrolysis due to unstable phase transitions in redox atmospheres. Therefore, the XRD results of LSF and LSF8 powders after heat treatment in a 5:95 CO-CO₂ atmosphere for 2 h (respectively named LSF-C and LSF8-C, Figure S11) and the Raman test after long-term stability examination (Figure 4F) were also analyzed. According to XRD results, after treatment in a reducing atmosphere, SrCO₃ was observed in both LSF and LSF8 samples. Due to the excessive stoichiometric ratio of A-site elements in LSF8, the content of SrCO₃ segregated in LSF8 is even significantly higher than that in LSF, which can be judged according to the peak intensity of SrCO₃ (Figure S11). Interestingly, the RP phase originally present in LSF8 cannot detected by XRD after the treatment with a gas mixture. Sr elements in the rock salt layer in the RP phase have a greater dynamic tendency to segregate than that in the SP phase. In the latter phase, Sr is only located in the A-site of perovskite which also enabled the structural integrity of the SP phase under the reducing atmosphere, and consequently resulting in better stability of composite material in the process of CO₂ reduction. Nevertheless, there are no C-C bond signals (the band at 1588 and 1385 cm⁻¹ can be due to the G-band and D-band of carbonaceous species) after a long time of operation on the LSF8 cathode.⁴⁸ Therefore, heterostructure materials show potential for practical applications in CO₂ utilization using SOEC technology.

Thus, in this work, we synthesized heterostructural single-phase (SP) perovskite/Ruddlesden–Popper (RP) phase perovskite from precursors of La_0.6Sr_0.4Fe_1-xO_3-δ (x = 0.1 and 0.2) perovskite oxides by intentionally adjusting B-site stoichiometry. Compared with La_0.6Sr_0.4FeO_3-δ, SOECs based on the synthesized heterostructure electrode materials demonstrated a lower catalytic performance due to the low electrical conductivity of RP-LSF, the increase of the average valence state of Fe at B-site, more segregation of SrCO₃ as well as reduced oxygen vacancy and CO₂ adsorption capacity. This work provides a useful guide for the design of composite materials and offers a new perspective to study the electrochemical behavior of composite perovskite materials for CO₂-RR in SOEC. Future work will focus on investigating the effect of phase change under the typical in situ operational condition on the CO₂-RR electrochemical performance.

Limitations of the study

This study applies only to CO₂ electrolysis; the related method mentioned in the article might be used as a reference for other reactions in the future.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Chemicals, peptides, and recombinant proteins
Lanthanum nitrate hexahydrate	Shanghai Macklin Biochemical Technology Co., Ltd	E1722137; CAS:10277-43-7
Strontium nitrate	Aladdin Industrial Corporation	I1712036; CAS:10042-76-9
Ferric nitrate	Shanghai Macklin Biochemical Technology Co., Ltd	C15518746; CAS:7782-61-8
LSGM-8282	Wuxi City Kaistar Electro-Optic Materials Co., Ltd.	N/A

Resource availability

Lead contact

Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Liangdong Fan (fanld@szu.edu.cn).

Materials availability

The study did not generate new unique materials. The readers can buy the chemicals to remake the materials as mentioned in the text.

Data and code availability

• •Data: All data reported in this paper will be shared by the lead contact upon request.
• •Code: This paper does not report the original code.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Experimental model and subject details

There are no experimental models (animals, human subjects, plants, microbe strains, cell lines, primary cell cultures) used in the study.

Methods details

Material synthesis

La_0.6Sr_0.4FeO_3-δ (LSF), La_0.6Sr_0.4Fe_0.9O_3-δ (LSF9), and La_0.6Sr_0.4Fe_0.8O_3-δ (LSF8) powders were prepared by the classic sol-gel method as below. The starting materials, La(NO₃)₂·6H₂O (99%, Macklin), Sr(NO₃)₂ (≥99.0%, Aladdin), and Fe(NO₃)₂ (98.5%, Macklin), were dissolved in the beaker by deionized water with stoichiometric ratios. Then, citric acid monohydrate (CA, 99.5%, Aladdin) and ethylene diamine tetra-acetic acid (EDTA, 99.5%, Aladdin) were added as the complexing and combustion agents. The molar ratio of total metal ions, CA, and EDTA was set to 1:2:1. Eventually, ammonia was added to adjust the pH of the solution to 7-8 for absolutely dissolved salt nitrate. After 24 h of stirring, the precursor solution was transferred to evaporating dish and dried at 200 °C, until the organic matter burned out and the precursor powder was obtained. The precursor powder was calcined in air at 1000 °C for 5 h to gain the black powder with a perovskite structure.

The preparation method of anode powder, La_0.6Sr_0.4Co_0.2Fe_0.8O_3-δ, is the same as above but using the starting materials La(NO₃)₂·6H₂O (99%, Macklin), Sr(NO₃)₂ (≥99.0%, Aladdin), Fe(NO₃)₂ (98.5%, Macklin), Co(NO₃)₂·6H₂O (99%, Aladdin), CA (99.5%, Aladdin) and EDTA (99.5%, Aladdin), abd the calcination temperature and duration change to 900°C for 5 h. To improve the conductivity of oxygen ions, Sm_0.2Ce_0.8O_2−δ (SDC), a high oxygen ion conductor, was added. SDC was synthesized by glycine combustion method with the precursor calcined at 600°C for 2 h. The LSCF-SDC (50:50 wt. %) composite sample was obtained by ball milling for 24 h.

Sample preparation

The electrolyte-supported single cells with LSFx | LSGM | LSCF-SDC were used to demonstrate electrochemical performance. The LSGM electrolytes were formed by dry-pressing LSGM powder (Wuxi City Kaistar Electro-Optic Materials Co., Ltd.) which is mixed with 3 wt.% PVA (polyvinyl alcohol) by grinding for 2 h and then calcined at 1450 °C for 6 h to obtain the dense electrolyte (∼ 220 μm). The air electrode slurry was obtained by mixing LSCF-SDC powders and terpineol (containing 6 wt.% methyl-cellulose) for 1 h. The fuel electrode slurries, LSF / LSF9 / LSF8, were obtained in the same way. The slurries were printed onto LSGM electrolytes by screen print method and then co-fired at 1000 °C for 2 h in air. The symmetric cells were prepared in the same way while both sides were LSFx fuel electrodes. After sintering, silver paste (Guiyan Platinum Industry Co., LTD) was printed onto electrodes (0.2826 cm²) as the current collector.

LSF and LSF8 bar samples were prepared for the electrical conductivity test. 1 g LSF or LSF8 powders mixed with 3 wt.% PVA was ground in a mortar for 1 h, then the bars formed by the dry-press process. In the end, the bars were calcined at 1300°C in air for 6 h to achieve the required density for electrical conductivity testing.

Materials characterization

The phase structure of all materials was characterized by XRD (PANalytical, Empyrean) with Cu Kα radiation and TEM (JEOL, JEM-2100 & X-Max80). The crystal structure parameters and phase proportion of LSFx were fitted by GSAS refinement. The micromorphology of powders was observed by SEM (HIT, S-3400N). The electrical conductivity was measured in air and 95: 5 CO₂-CO by a four-paragraph method using a conductivity meter (KEITHLEY). The surface properties of powder samples were characterized by X-ray photoelectron spectroscopy (XPS, Thermo Fisher Scientific, K-Alpha+). The high-temperature physical and chemical properties of powder samples were detected by Thermo Gravimetric Analysis (TGA, Mettler, TGA\DSC 3+). Raman spectrometer (Renishaw) was used to determine the coke formation on the fuel electrode powders’ surface after long-term examination. CO₂-TPD was used to study the CO₂ adsorption capacity of LSF and LSF8 samples.

Electrochemical test

The single SOEC was sealed at the top of the Al₂O₃ tube using Cerama-bond 552-VFG sealant (Aremco). The fuel electrode was exposed to pure CO₂ while the air electrode was exposed directly to the atmosphere for the SOEC test. The flow rate of pure CO₂ was 50 ml min^-1. An electrochemical workstation (VersaSTAT 4) was used to evaluate the electrochemical performance; the applied voltage window is from open-circuit voltage to 1.6 V at a scan rate of 10 mV s^-1 for CO₂ electrolysis. The short-term CO₂ electrolysis stability test was operated at 1.2, 1.3, 1.4, and 1.5 V, each was stabilized for 20 min at 800°C. The long-term CO₂ electrolysis stability test was operated at a constant voltage of 1.1 V at 800°C. EIS measurement of a single SOEC was obtained in the frequency range of 10⁵ Hz to 0.1 Hz with an AC amplitude of 20 mV. The symmetric cell’s EIS measurement was obtained in the frequency range of 10⁵ Hz to 0.1 Hz with an AC amplitude of 10 mV, but it is under a 1: 1 CO–CO₂ atmosphere. EIS data of single SOEC and symmetric cells were collected at 750-850^oC.

The gas product was determined by the online gas chromatography (GC-2014, SHIMADZU) equipped with a thermal conductivity cell detector (TCD). The gas flow rate was determined by a flowmeter. The faradaic efficiency was calculated by the following Equation.^19,20,31,32

$${FE}_{CO}=\frac{n_{CO,measured}}{{I\times \left(n\times F\right)}^{-1}}\times 100\%$$

Where n_{CO, measured} (mol s^-1) is the measured carbon monoxide production rate in the exhaust gas from the electrochemical cell (GC data). I (A) is the current, n is 2 and F is Faraday’s constant (96485 C mol^-1).

Quantification and statistical analysis

No methods were used to determine whether the data met the assumptions of the statistical approach.

Additional resources

Our study has not generated or contributed to a new website/forum or has not been part of a clinical trial.

Published: March 29, 2024