Role of lattice oxygen and oxygen vacancy sites in platinum group metal catalysts supported on Sr₃Fe₂O_7−δ for NO-selective reduction

Abstract

This study demonstrates that NO-selective reduction with C₃H₆ and CO proceeds over platinum group metal (PGM; Pd, Rh, or Pt) catalysts supported on Sr₃Fe₂O_7−δ with a layered perovskite-type oxide. Among the examined catalysts, Pd-loaded Sr₃Fe₂O_7−δ having high oxygen storage capacity shows the highest catalytic activity at stoichiometric oxygen concentration because the Pd catalyst can release the lattice oxygen of Sr₃Fe₂O_7−δ at lower temperature compared to the Rh- or Pt-loaded catalyst. When the NO-selective reduction at 773 K is carried out at various oxygen concentrations, PGM/Sr₃Fe₂O_7−δ shows superior catalytic activity over a wide range of oxygen concentrations to PGM/Al₂O₃, which does not have oxygen storage capacity. The oxygen vacancy sites in Sr₃Fe₂O_7−δ, which are generated by the oxidation of C₃H₆ and CO over PGM/Sr₃Fe₂O_7−δ, are revealed to receive oxygen ions formed by NO reduction on the PGM species; as a result, the PGM species maintain their metal state and act as active sites for NO reduction. Namely, Sr₃Fe₂O_7−δ plays the role of an “oxygen buffer” in inhibiting the PGM-to-oxide transformation. Pt-loaded Sr₃Fe₂O_7−δ can utilize the oxygen vacancy sites more effectively for the catalytic reaction than the Pd and Rh catalysts. To the best of our knowledge, this is the first report on the effective application of perovskite materials with oxygen storage capacity as catalyst supports for NO-selective reduction. The study results are expected to provide valuable information for designing a novel catalyst based on oxygen storage materials.

---

Introduction

CeO₂ and CeO₂-based oxides having excellent redox properties can release and store oxygen molecules depending on the reductive or oxidative atmosphere.¹ Based on this unique redox function, CeO₂-based oxides have been applied as useful materials in various fields such as organic synthesis,^2,3 oxide ion conductors,^4,5 soot oxidation,^6,7 and automotive catalysts.^8–11 The redox properties of CeO₂ have been improved by heterogeneous metal doping,^12–14 morphology control,^6,15etc. Zr-doped CeO₂, i.e., CeO₂–ZrO₂ solid solution, is a representative material because it has high oxygen storage capacity^12,16,17 and is widely used as a catalyst support for the purification of automotive exhaust gas.^8,18 Furthermore, since CeO₂-based materials are non-stoichiometric oxygen compounds, many studies on oxygen vacancy sites have been reported.^3,8,11,18 For example, the oxygen vacancy sites over CeO₂ have been found to be effective for the formation of N₂O or N₂ in the NO–CO reaction.^8,11,18

Recently, because of their structural diversity, perovskite-type oxides and perovskite-derived oxides showing oxygen storage performance have gained much attention as alternatives to CeO₂-based materials.^19–23 Perovskite-derived materials such as Ca_0.8Sr_0.2MnO₃ (ref. 23) and BaYMn₂O_5+δ (ref. 20 and 21) have been reported to deliver high performance as oxygen storage materials. However, these materials have hardly been applied as supports for automotive catalysts.

We have also reported that Sr₃Fe₂O_7−δ having a layered perovskite structure shows high oxygen storage capacity and high structural stability under severe reductive conditions.²⁴ Pd loading on Sr₃Fe₂O_7−δ has been found to drastically lower the operating temperature for oxygen storage and improve the oxygen release rate.²⁵ It is noteworthy that Sr₃Fe₂O_7−δ can release and store oxygen through topotactic transition, which is interpreted as a transition with retention of cation ordering.^26–28 When considering the phase transition of simple transition metal oxides such as α-Fe₂O₃, the oxygen ions in the lattice of FeO, obtained by the reductive treatment of α-Fe₂O₃, assume a close-packed structure by rearrangement of the cation and oxygen ions in α-Fe₂O₃.^29,30 Therefore, it is difficult to obtain oxygen vacancy sites in simple transition metal oxides. On the other hand, Sr₃Fe₂O_7−δ, which releases lattice oxygen via topotactic transition, can form a large number of oxygen vacancy sites in the layered perovskite structure. The oxygen storage performance and oxygen vacancy sites in Sr₃Fe₂O_7−δ are expected to play an effective role in various catalytic reactions.

With this background, the present study demonstrates NO-selective reduction over platinum group metal (PGM = Rh, Pd, Pt) catalysts supported on Sr₃Fe₂O_7−δ. The contribution of the lattice oxygen and oxygen vacancy sites of Sr₃Fe₂O_7−δ to NO-selective reduction is investigated. To the best of our knowledge, this is the first example of the purification of automotive exhaust gas utilizing lattice oxygen and oxygen vacancy sites in a perovskite-type oxide having oxygen storage capacity.

Experimental

Preparation of PGM catalysts

Sr₃Fe₂O_7−δ was synthesized by a polymerized complex method, as mentioned in our previous report.²⁵ PGM catalysts were synthesized by an impregnation method. The following PGM precursors were used as source materials: Pt(NH₃)₂(NO₂)₂ (HNO₃ solution) (4.64 wt%, FURUYA METAL Co., Ltd.), Rh acetylacetonate (97%, Sigma-Aldrich), and Pd acetate (99.9%, Sigma-Aldrich). Rh acetylacetonate or Pd acetate was dissolved in ethyl acetate (99.5%, Wako Pure Chemicals), and Pt(NH₃)₂(NO₂)₂ in water. Then, the support (Sr₃Fe₂O_7−δ or γ-Al₂O₃ (JRC-ALO-7)) was added, and the solution was evaporated at 353 K. The powder obtained after drying the solution containing the PGM precursor and Sr₃Fe₂O_7−δ was calcined at 1073 K for 5 h. The loading amount of the PGM species was 1.0 wt% on a metal basis.

NO-selective reduction

The catalytic reaction was carried out using a fixed bed flow reactor at atmospheric pressure. In a tubular reactor, 200 mg of the catalyst sample (25/50 mesh) was placed. The total gas flow rate was 100 mL min⁻¹. The catalyst was pretreated by heating to 773 K under He flow (30 mL min⁻¹) and held for 1 h. The gas concentration was analysed using a gas chromatograph (Shimadzu GC8A) with MS-5A and Porapak-Q columns. The temperature-programmed reaction was carried out by increasing the temperature from 373 K in a stepwise manner (50 K intervals) under stoichiometric conditions: NO, 1000 ppm; CO, 1000 ppm; C₃H₆, 250 ppm; O₂, 1125 ppm. The catalytic activity for various oxygen concentrations was evaluated at 773 K under the following reaction atmosphere: NO, 1000 ppm; CO, 1000 ppm; C₃H₆, 250 ppm; O₂, various concentrations (675–1462.5 ppm); He as balance gas. The λ value was defined as follows: the number of oxygen atoms in the reaction system ([NO] + [CO] + [O₂] × 2) was divided by the number of oxygen atoms under stoichiometric conditions ([NO] + [CO] + [O₂] × 2 = 4250 ppm). The catalytic reaction was commenced under lean conditions (O₂, 1462.5 ppm) and the oxygen concentration was changed to 625 ppm (rich conditions) in a stepwise manner while maintaining the NO and CO concentrations. Subsequently, the oxygen concentration was varied to achieve lean conditions. The reaction gas was held for 30 min at each oxygen concentration. The lean–rich cycle repeatability test for NO-selective reduction was carried out with switching the atmosphere between lean (NO, 1000 ppm; CO, 1000 ppm; C₃H₆, 250 ppm; O₂, 1462.5 ppm) and rich (NO, 1000 ppm; CO, 1000 ppm; C₃H₆, 250 ppm; O₂, 675 ppm) conditions. The catalyst was heated to 773 K under lean conditions and held for 30 min. Then, the reaction conditions were changed between lean and rich conditions every 1 h. The outlet gases were detected by using a micro gas chromatograph (Agilent 490 Micro GC) with MS-5A and PoraPLOT Q columns.

H₂-TPR

Temperature-programmed reduction with H₂ (H₂-TPR) was carried out on a flow-type reactor, as outlined in a previous report.^24,25 H₂ (2 vol% in Ar; 30 ml min⁻¹) was passed through the reactor charged with a sample (0.05 g) under atmospheric pressure. The reactor was heated to 1223 K in an electric furnace at a heating rate of 5 K min⁻¹, and the amount of H₂ consumed was monitored with a thermal conductivity detector (TCD) (Shimadzu GC8A).

NO pulse

NO pulse experiments were carried out with a flow-type reactor (MicrotracBEL BELCAT-B). The samples (25 mg) were pretreated at 773 K by switching the atmosphere between H₂ and 1 vol% O₂/He (30 mL min⁻¹) every 20 min, and finally under a H₂ atmosphere. Then, the samples were heated to 773 K under He flow (30 mL min⁻¹). The NO pulse (0.997 cm³, 1 vol% in He = 0.434 μmol_NO per pulse) was injected into the reactor. The individual injection of pulse was purged with He. The outlet gas was analyzed by means of a TCD-gas chromatograph with an MS-5A or a Porapak Q column. The number of NO pulses was 40.

OSC measurement

Oxygen storage capacity (OSC) measurement was carried out with a thermogravimeter (Rigaku Thermoplus), as mentioned in our previous report.²⁴ The catalysts (100 mg) were heated to 773 K in 5% O₂/Ar and held until constant weights were achieved. Then, the weight changes of the samples upon switching the atmosphere every 20 min between 5% O₂/Ar and 5% H₂/Ar were recorded at 773 K.

Characterization

X-ray diffraction (XRD) patterns were recorded on a Rigaku Ultima IV apparatus using CuKα radiation. X-ray photoelectron spectroscopy (XPS) measurement was conducted on a Shimadzu ESCA-3400 using MgKα radiation. Each spectrum was calibrated by C1s at 284.6 eV. The specific surface area was evaluated from the N₂ adsorption isotherm using the Brunauer–Emmett–Teller method. N₂ adsorption was carried out on a MicrotracBEL BELSORP-mini II at 77 K.

Results and discussion

Fig. 1 shows the results of NO-selective reduction at stoichiometric oxygen concentration over PGM catalysts supported on Sr₃Fe₂O_7−δ. The oxidation activity over Sr₃Fe₂O_7−δ was observed at 623 K and it reached 60% at 773 K. NO reduction activity was not observed even at 773 K. These catalytic activities were drastically enhanced by PGM loading, although Sr₃Fe₂O_7−δ had a very low surface area (<10 m² g⁻¹) as a catalyst support. Pd/Sr₃Fe₂O_7−δ showed higher catalytic activity than Pt/Sr₃Fe₂O_7−δ and Rh/Sr₃Fe₂O_7−δ; CO and C₃H₆ oxidation commenced at 473 K and NO reduction proceeded from 523 K. Before NO conversion to N₂ reached 100%, a small amount of N₂O as a by-product was observed in all catalysts (Fig. S1†), indicating that N₂O species might be an intermediate species in NO-selective reduction over PGM catalysts. The oxidation activity of Pt/Sr₃Fe₂O_7−δ became higher than that of Rh/Sr₃Fe₂O_7−δ in the high temperature region. The CO and C₃H₆ conversions over the PGM catalysts are displayed in Fig. S1.† Pd and Pt species enhanced the oxidation of both CO and C₃H₆. However, Rh species promoted the CO oxidation well, but moderately enhanced the C₃H₆ oxidation. Then, the activity inversion between Pt- and Rh-loaded catalysts is attributed to the different oxidation characteristics of PGM species for the oxidation reaction. In addition, the reduction of NO proceeded after the oxidation of C₃H₆ and CO in all catalysts.

Fig. 1 (A) NO conversion to N₂ and (B) CO and C₃H₆ conversion to CO₂ (λ = 1) over PGM catalysts supported on Sr₃Fe₂O_7−δ.

The XRD patterns of PGM/Sr₃Fe₂O_7−δ are displayed in Fig. 2. The XRD patterns of PGM/Sr₃Fe₂O_7−δ were similar to that of the bare catalyst. Furthermore, peaks due to the PGM species were not observed in the XRD patterns of PGM/Sr₃Fe₂O_7−δ. Previously, the Pd species in Pd/Sr₃Fe₂O_7−δ has been revealed to be highly dispersed as Pd²⁺, whose coordination state is different from that in the PdO structure.²⁵ Similar to Pd/Sr₃Fe₂O_7−δ, a highly dispersed PGM species must be stabilized on Sr₃Fe₂O_7−δ in Rh- and Pt-loaded catalysts.

Fig. 2 XRD patterns of PGM catalysts supported on Sr₃Fe₂O_7−δ. a, Pd; b, Pt; c, Rh; d, bare.

The peaks of Pt/Sr₃Fe₂O_7−δ were broader than those of the other catalysts. This might be due to the formation of a hydrate phase by water used as an impregnation solvent. Since Sr₃Fe₂O_7−δ has been reported to be hydrated under the existence of excess water,³¹ the crystal structure of Sr₃Fe₂O_7−δ must be changed into the hydrate phase during the preparation of Pt/Sr₃Fe₂O_7−δ. The hydrate phase was mostly recovered from the crystal structure of Sr₃Fe₂O_7−δ by calcination at 1073 K. Although the relative intensity of the peaks in the XRD pattern of Pt/Sr₃Fe₂O_7−δ was slightly different from that of Sr₃Fe₂O_7−δ, the peak positions in the XRD pattern of Pt/Sr₃Fe₂O_7−δ were almost the same as those of Sr₃Fe₂O_7−δ. Therefore, the crystal structure of Sr₃Fe₂O_7−δ in the Pt catalyst is essentially identical to those in the other catalysts. In addition, the specific surface area (6 m² g⁻¹) of Pt/Sr₃Fe₂O_7−δ was comparable to those (4–5 m² g⁻¹) of Rh/Sr₃Fe₂O_7−δ and Pd/Sr₃Fe₂O_7−δ. From these results, the peak broadening of the Pt catalyst was not thought to affect the results of the reaction.

To clarify the reduction behaviour of PGM/Sr₃Fe₂O_7−δ, H₂-TPR profiles were investigated (Fig. 3). Each PGM catalyst showed one reduction peak. The reduction peaks of PGM/Sr₃Fe₂O_7−δ were drastically shifted to lower temperature compared to those of Sr₃Fe₂O_7−δ. The oxygen release amounts up to 773 K were 657 μmol_O2 g_cat⁻¹ and 693 μmol_O2 g_cat⁻¹ in the Pt and Rh catalysts, respectively. We have reported that the oxygen release amounts of the Pd and bare catalysts are 665 μmol_O2 g_cat⁻¹ and 618 μmol_O2 g_cat⁻¹, respectively, and that the reduction peak of Pd/Sr₃Fe₂O_7−δ is attributed to the reduction from Pd²⁺ and Fe⁴⁺ to Pd⁰ and Fe³⁺.^24,25 Since the oxygen release amounts of Pt/Sr₃Fe₂O_7−δ and Rh/Sr₃Fe₂O_7−δ were comparable to that of Pd/Sr₃Fe₂O_7−δ, each reduction peak of PGM/Sr₃Fe₂O_7−δ was also attributed to the reduction of Fe and PGM species. This result indicates that the oxygen release properties of Sr₃Fe₂O_7−δ are promoted by Pt and Rh loading, as in the case of Pd loading on Sr₃Fe₂O_7−δ. The order of the effect for lowering the oxygen release temperature was Pd > Rh > Pt, corresponding to the order of the light-off temperature for CO and C₃H₆ oxidation, especially CO oxidation, in each PGM catalyst.

Fig. 3 H₂-TPR profiles of PGM catalysts supported on Sr₃Fe₂O_7−δ. a, Pd; b, Pt; c, Rh; d, bare.

It is known that the oxidation reaction over metal oxide catalysts proceeds by the Mars–van Krevelen mechanism, which is interpreted as follows: a reactant such as CO or C₃H₆ is oxidized by utilizing the lattice oxygen of a catalyst, and the lattice oxygen is recovered by the oxidant in a reaction system.^32–36 Therefore, the activity in an oxidation reaction via the Mars–van Krevelen mechanism is known to correlate with the mobility of the lattice oxygen, which can be evaluated by H₂-TPR analysis.^34–36 For example, Scirè et al. reported that the oxidation activity for the combustion of volatile organic compounds (VOCs) over Au/CeO₂ was well correlated with the mobility of the lattice oxygen in CeO₂.³⁴ Therefore, the Au species loaded over CeO₂ enhances the release of lattice oxygen, thus improving the oxidation activity via the Mars–van Krevelen mechanism. From the previous reports^32–36 and the results of H₂-TPR, highly dispersed PGM species enhance the lattice oxygen release of Sr₃Fe₂O_7−δ, resulting in improvement of CO and C₃H₆ oxidation via the Mars–van Krevelen mechanism. Generally, oxygen vacancy sites, which are generated by the release of lattice oxygen during the oxidation reaction, are filled by molecular oxygen as in the case of the combustion of VOCs. In NO-selective reduction, NO or molecular oxygen can be utilized as an oxidant. That is, NO would fill the oxygen vacancy sites to form N₂ molecules. We will describe the details of NO reduction later.

PGM/Sr₃Fe₂O_7−δ having oxygen storage capacity may be able to adjust the oxygen concentration in NO-selective reduction. NO-selective reduction at various oxygen concentrations over PGM/Sr₃Fe₂O_7−δ was carried out. PGM/Al₂O₃, which does not have oxygen storage capacity, was used for comparison. CO and C₃H₆ oxidation activity exhibited a similar trend in all catalysts (Fig. S2†): high oxidation activity was observed at high oxygen concentrations (lean conditions, λ > 1), and the activity decreased with decreasing oxygen concentration. On the other hand, the catalysts showed different NO reduction behavior from one another (Fig. 4). In the case of all the catalysts, when the reaction conditions were changed from lean to stoichiometric (λ = 1), the NO reduction activity improved with decreasing oxygen concentration, reaching 100% near the stoichiometric conditions (λ = 1). Under rich conditions, the NO reduction activities over Pd/Sr₃Fe₂O_7−δ and Pt/Sr₃Fe₂O_7−δ decreased slightly but Rh/Sr₃Fe₂O_7−δ maintained high NO reduction activity. When the reaction conditions were changed from rich to stoichiometric, the NO reduction activity improved with increasing oxygen concentration, reaching 100% around the stoichiometric conditions in Pd/Sr₃Fe₂O_7−δ and Pt/Sr₃Fe₂O_7−δ. Interestingly, the high NO reduction activity over PGM/Sr₃Fe₂O_7−δ was maintained even when the reaction conditions were changed from stoichiometric to lean. The order of maintenance of the high NO reduction activity under the lean conditions was Pt/Sr₃Fe₂O_7−δ ≈ Rh/Sr₃Fe₂O_7−δ ≫ Pd/Sr₃Fe₂O_7−δ. PGM/Al₂O₃ did not show any improvement in NO reduction activity under the lean conditions. These results suggest that Sr₃Fe₂O_7−δ contributes to the enhancement of NO reduction activity over a wide range of oxygen concentrations. When the repeatability test of Pt/Sr₃Fe₂O_7−δ, which showed the highest NO reduction activity under slightly lean conditions in the NO-selective reduction, was carried out, the catalyst was used repeatedly at least three times without deactivation during lean–rich cycles (Fig. S4†).

Fig. 4 NO conversion to N₂ over each PGM catalyst supported on Sr₃Fe₂O_7−δ and Al₂O₃.

XRD analysis of the catalyst recovered at each point during the reaction was performed to evaluate the lattice oxygen content in PGM/Sr₃Fe₂O_7−δ. The as-synthesized sample, sample treated under the rich conditions (λ = 0.79), and sample after the reaction (i.e., the samples were exposed to rich-to-lean conditions) were compared. In addition, the sample reduced at 773 K under a H₂ atmosphere was investigated. The XRD pattern of each catalyst in a wide range is displayed in Fig. S3.† All the patterns were essentially similar to those of Sr₃Fe₂O_6.75 or Sr₃Fe₂O₆ without additional peaks due to by-products, indicating that PGM/Sr₃Fe₂O_7−δ maintained its crystal structure during the reaction. A magnified view of the XRD patterns of PGM/Sr₃Fe₂O_7−δ during the reaction is shown in Fig. 5. The XRD peaks of the samples treated under the rich conditions shifted to lower angles as compared to those of the as-synthesized samples. The peak positions of the samples treated under the rich conditions corresponded to those of the samples reduced at 773 K under a H₂ atmosphere, indicating that the amounts of lattice oxygen released from both samples were similar. Considering the fact that the Fe⁴⁺ species in Sr₃Fe₂O_6.75 is completely reduced to Fe³⁺ under a H₂ atmosphere at 773 K, PGM/Sr₃Fe₂O_6.75 is reduced to PGM/Sr₃Fe₂O₆ under the rich conditions. This result implies that lattice oxygen is used for the oxidation of CO or C₃H₆ according to the Mars–van Krevelen mechanism. The peaks for each catalyst after the reaction did not return to those for the as-synthesized catalysts but shifted to a higher angle than those for the samples treated under the rich conditions. This phenomenon indicated that the lattice oxygen is partly recovered by NO or O₂ after the lean/rich/lean transition.

Fig. 5 Magnified view of the XRD patterns of PGM/Sr₃Fe₂O_7−δ. Black line, as-synthesized samples; red line, the samples reduced at 773 K under a H₂ atmosphere; blue line, the samples treated under rich conditions; green line, the samples after the reaction.

The PGM/Sr₃Fe₂O_7−δ catalysts, which showed the same oxygen release amount under the rich conditions, had different properties for maintaining the NO reduction activity under the lean conditions. To clarify the factors affecting the NO reduction activity of PGM/Sr₃Fe₂O_7−δ, we conducted NO pulse experiments (Fig. 6(A)). No N₂ formation was observed over Sr₃Fe₂O_7−δ reduced at 773 K in a H₂ atmosphere, and PGM loading drastically improved the N₂ formation over Sr₃Fe₂O_7−δ (Fig. 6(B)). PGM/Sr₃Fe₂O_7−δ showed higher N₂ formation than PGM/Al₂O₃ (Fig. 6(B) and S5†). When the formation amount of N₂ decreased, N₂O and NO were initially detected in the NO pulse experiment (Fig. S6†). The material balance for nitrogen species estimated from the total amount of N₂, N₂O and NO was almost 100%, indicating that NO species is hardly adsorbed or stored during the NO pulse experiments on Pt/Sr₃Fe₂O_7−δ. Interestingly, O₂ formation was not observed in all catalysts which exhibit N₂ formation for the NO pulse experiment (Fig. S7†); that is, O species formed by NO dissociation must oxidize the PGM species or support. Then, the ideal N₂ formation amount over a 1 wt% PGM catalyst is calculated from the following equations representing the reaction between PGM and NO:

Pt + 2NO → N₂ + PtO₂ 51.3 μmol_N2 g_cat⁻¹

Pd + NO → 1/2N₂ + PdO 47.0 μmol_N2 g_cat⁻¹

Rh + 3/2NO → 3/4N₂ + 1/2Rh₂O₃ 72.9 μmol_N2 g_cat⁻¹

Fig. 6 (A) Results of NO pulse over PGM/Sr₃Fe₂O_7−δ reduced at 773 K under a H₂ atmosphere. () Rh, () Pt, () Pd and (○) Pd without reduction treatment. (B) The amount of N₂ production during the NO pulse.

The N₂ formation amounts over PGM/Al₂O₃ were smaller than the ideal amounts, suggesting that all PGM species on Al₂O₃ could not be attributed to NO reduction. On the other hand, PGM/Sr₃Fe₂O_7−δ generated a larger amount of N₂ than the ideal amounts. These results suggest that a synergetic effect between the PGM species and Sr₃Fe₂O_7−δ contributes to NO reduction.

Since N₂ formation was not detected in the case of PGM/Sr₃Fe₂O_7−δ without H₂ treatment, the oxygen vacancy sites in Sr₃Fe₂O_7−δ generated by the pretreatment are considered to participate in N₂ formation. If the oxygen vacancy sites receive oxygen ions generated from NO dissociation on PGM species, the amount of N₂ must agree well with the amount of oxygen vacancy sites. The Pt catalyst, which showed the highest N₂ formation amount among the examined catalysts, stored 111 μmol_O2 g_cat⁻¹, while the value was much lower than the amount (657 μmol_O2 g_cat⁻¹) of oxygen vacancy sites in Pt/Sr₃Fe₂O₆ obtained by reduction in the H₂ atmosphere. Furthermore, the ideal number of oxygen vacancy sites (22–33 μmol_O2 g_cat⁻¹), calculated from the surface area of PGM/Sr₃Fe₂O_7−δ and the area of the (001) facet of Sr₃Fe₂O₆ (ref. 24) obtained by reduction under H₂ flow, is much smaller than the amount of the oxygen species (72–111 μmol_O2 g_cat⁻¹) estimated from the NO pulse experiment. Therefore, we conclude that PGM/Sr₃Fe₂O_7−δ generate N₂ by utilizing the oxygen vacancy sites at the surface and subsurface of the support.

Summarizing the results of NO pulse, the interpretations are as follows. NO reduction must proceed over the metal state of PGM species in PGM/Sr₃Fe₂O_7−δ, as the metal state of PGM species has been well known to be the active site for NO reduction. Actually, when the valence state of Pt in Pt/Sr₃Fe₂O_7−δ during the NO pulse experiment was evaluated by XPS analysis (Fig. S8†), the valence state of Pt remained mainly a metal state³⁷ in Pt/Sr₃Fe₂O_7−δ before decreasing the N₂ production amount. However, the Pt species was oxidized by NO at the point where N₂ formation was not observed. Since the Pt species maintaining its metal state for a long period can dissociate NO molecules, Pt/Sr₃Fe₂O_7−δ retains a high N₂ formation amount. On the other hand, O₂ formed by NO dissociation was not detected. Considering the relationship between the N₂ formation amount and oxygen storage amount, the dissociated oxygen species must be preferentially incorporated into the oxygen vacancy sites around the surface and subsurface of Sr₃Fe₂O_7−δ, instead of oxidation of PGM species occurring. In other words, the vacancy site in Sr₃Fe₂O_7−δ acts as an “oxygen buffer” for inhibiting the oxidation of the PGM species. In addition, NO reduction to N₂ needed PGM loading and the reduction treatment of the catalyst; that is, the oxygen vacancy site around the PGM species plays an important role in NO reduction.

Although the Pd catalyst showed the lowest N₂ production in the NO pulse experiment, this catalyst exhibited the highest activity for NO conversion among the PGM catalysts in NO-selective reduction (Fig. 1). In NO-selective reduction, the reductants for the catalysts are CO and C₃H₆, and the catalyst having high CO and C₃H₆ oxidation activity must be easily reduced during the reaction. Pd/Sr₃Fe₂O_7−δ had the highest oxidation property and was reduced at the lowest temperature among the examined catalysts in H₂-TPR analysis. Therefore, Pd/Sr₃Fe₂O_7−δ, which could be easily reduced by CO and C₃H₆, showed the highest NO reduction activity in the NO-selective reduction (λ = 1).

Excess O₂ in NO-selective reduction up to λ = 1.08 was not detected at all over PGM/Sr₃Fe₂O_7−δ after a lean/rich transition, but was observed in PGM/Al₂O₃. That is, excess O₂ must be stored into the vacancy site in the Sr₃Fe₂O_7−δ support with maintaining the high NO conversion. Therefore, the oxygen vacancy sites receive excess O₂ molecules as well as oxygen species formed by NO dissociation. Considering the oxygen storage capacities of PGM/Sr₃Fe₂O_7−δ estimated from H₂–O₂ titration at 773 K (Fig. S9†), the oxygen storage amount of Sr₃Fe₂O_7−δ itself is essentially identical to those of PGM/Sr₃Fe₂O_7−δ. These results suggest that O₂ molecules penetrated into particles inside irrespective of the PGM species. However, the NO reduction property evaluated from the NO pulse experiment is dramatically enhanced by PGM loading, indicating that the oxygen vacancy site around the PGM species mainly receives the oxygen species formed by NO dissociation over the PGM species. Since these results reveal that the amount of available oxygen vacancy sites for O₂ molecules is much larger than that for NO molecules, NO-selective reduction over PGM/Sr₃Fe₂O_7−δ seems to proceed even under slightly lean conditions.

Based on all the results, a possible mechanism for NO-selective reduction over PGM/Sr₃Fe₂O_7−δ is proposed (Fig. 7), when the reaction conditions were changed from rich to lean conditions. First, the oxidation of CO and C₃H₆ proceeds with the aid of lattice oxygen in Sr₃Fe₂O_7−δ under rich conditions. The oxygen vacancy sites formed by the oxidation inhibit the oxidation of the PGM species even under slightly lean conditions; that is, the PGM species, which are the active sites for NO reduction, maintain their metal state by removing the excess oxygen species to the oxygen vacancy sites. Finally, the oxygen vacancy sites are recovered with the NO or O₂ molecule.

Fig. 7 Proposed mechanism of NO-selective reduction over PGM/Sr₃Fe₂O_7−δ.

Conclusions

In conclusion, this paper clarifies the following ambiguous facts concerning NO-selective reduction over PGM/Sr₃Fe₂O_7−δ. PGM loading on Sr₃Fe₂O_7−δ drastically improves CO and C₃H₆ oxidation activity and induces NO reduction. The Pd catalyst shows the highest activity among the examined samples. H₂-TPR experiments show that the Pd catalyst can release lattice oxygen at lower temperature than that for the other catalysts. These results suggest that the oxygen release properties of Sr₃Fe₂O_7−δ contribute to the catalytic activity. On the other hand, the Pt and Rh catalysts show higher NO reduction activity over a wider range of oxygen concentrations than the Pd catalyst. The NO pulse experiments reveal that oxygen ions generated by NO reduction over PGM on Sr₃Fe₂O_7−δ flow into the oxygen vacancy sites of Sr₃Fe₂O_7−δ, causing the PGM species to maintain their metal state as active species even for NO-selective reduction under slightly lean conditions. Pt/Sr₃Fe₂O_7−δ utilizes the oxygen vacancy sites more efficiently than the Rh- and Pd-loaded catalysts. These results are first reported for NO-selective reduction using lattice oxygen and oxygen vacancy sites in perovskite-type oxides showing high oxygen storage performance. Moreover, the findings of our study serve as a useful guideline for the development of novel automotive catalysts loaded on a perovskite-type oxide support with oxygen storage capacity.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This study was supported by the Program for Elements Strategy Initiative for Catalysts & Batteries (ESICB).

Notes and references

1. T. Montini, M. Melchionna, M. Monai and P. Fornasiero, Chem. Rev., 2016, 116, 5987–6041.
2. H. Miura, K. Wada, S. Hosokawa, M. Sai, T. Kondo and M. Inoue, Chem. Commun., 2009, 4112–4114.
3. M. Tamura and K. Tomishige, Angew. Chem., Int. Ed., 2015, 54, 864–867.
4. H. Yahiro, K. Eguchi and H. Arai, Solid State Ionics, 1989, 36, 71–75.
5. J.-G. Li, T. Ikegami and T. Mori, Acta Mater., 2004, 52, 2221–2228.
6. E. Aneggi, D. Wiater, C. de Leitenburg, J. Llorca and A. Trovarelli, ACS Catal., 2014, 4, 172–181.
7. E. Aneggi, V. Rico-Perez, C. D. Leitenburg, S. Maschio, L. Soler, J. Llorca and A. Trovarelli, Angew. Chem., Int. Ed., 2015, 54, 14040–14043.
8. P. Fornasiero, G. R. Rao, J. Kašpar, F. L'Erario and M. Graziani, J. Catal., 1998, 175, 269–279.
9. J. Kašpar, P. Fornasiero and M. Graziani, Catal. Today, 1999, 50, 285–298.
10. M. Sugiura, M. Ozawa, A. Suda, T. Suzuki and T. Kanazawa, Bull. Chem. Soc. Jpn., 2005, 78, 752–767.
11. C. Deng, B. Li, L. Dong, F. Zhang, M. Fan, G. Jin, J. Gao, L. Gao, F. Zhang and X. Zhou, Phys. Chem. Chem. Phys., 2015, 17, 16092–16109.
12. M. Ozawa, M. Kimura and A. Isogai, J. Alloys Compd., 1993, 193, 73–75.
13. X. Wu, X. Wu, Q. Liang, J. Fan, D. Weng, Z. Xie and S. Wei, Solid State Sci., 2007, 9, 636–643.
14. K. Harada, T. Oishi, S. Hamamoto and T. Ishihara, Bull. Chem. Soc. Jpn., 2013, 86, 963–967.
15. S. K. Meher and G. R. Rao, J. Colloid Interface Sci., 2012, 373, 46–56.
16. P. Fornasiero, R. D. Monte, G. R. Rao, J. Kašpar, S. Meriani, A. Trovarelli and M. Graziani, J. Catal., 1995, 151, 168–177.
17. A. Suda, H. Sobukawa, T. Suzuki, T. Kandori, Y. Ukyo and M. Sugiura, J. Ceram. Soc. Jpn., 2001, 109, 177–180.
18. G. R. Rao, P. Fornasiero, R. D. Monte, J. Kašpar, G. Vlaic, G. Balducci, S. Meriani, G. Gubitosa, A. Cremona and M. Graziani, J. Catal., 1996, 162, 1–9.
19. T. Motohashi, Y. Hirano, Y. Masubuchi, K. Oshima, T. Setoyama and S. Kikkawa, Chem. Mater., 2013, 25, 372–377.
20. T. Motohashi, T. Ueda, Y. Masubuchi and S. Kikkawa, J. Phys. Chem. C, 2013, 117, 12560–12566.
21. T. Motohashi, M. Kimura, Y. Masubuchi, S. Kikkawa, J. George and R. Dronskowski, Chem. Mater., 2016, 28, 4409–4414.
22. J. Vieten, B. Bulfin, F. Call, M. Lange, M. Schmücker, A. Francke, M. Roeb and C. Sattler, J. Mater. Chem. A, 2016, 4, 13652–13659.
23. B. Bulfin, J. Vieten, D. E. Starr, A. Azarpira, C. Zachäus, M. Hävecker, K. Skorupska, M. Schmücker, M. Roeb and C. Sattler, J. Mater. Chem. A, 2017, 5, 7912–7919.
24. K. Beppu, S. Hosokawa, K. Teramura and T. Tanaka, J. Mater. Chem. A, 2015, 3, 13540–13545.
25. K. Beppu, S. Hosokawa, T. Shibano, A. Demizu, K. Kato, K. Wada, H. Asakura, K. Teramura and T. Tanaka, Phys. Chem. Chem. Phys., 2017, 19, 14107–14113.
26. Y. Tsujimoto, C. Tassel, N. Hayashi, T. Watanabe, H. Kageyama, K. Yoshimura, M. Takano, M. Ceretti, C. Ritter and W. Paulus, Nature, 2007, 450, 1062–1065.
27. H. Kageyama, T. Watanabe, Y. Tsujimoto, A. Kitada, Y. Sumida, K. Kanamori, K. Yoshimura, N. Hayashi, S. Muranaka, M. Takano, M. Ceretti, W. Paulus, C. Ritter and G. Andre, Angew. Chem., Int. Ed., 2008, 47, 5740–5745.
28. S. N. Achary, S. K. Sali, N. K. Kulkarni, P. S. R. Krishna, A. B. Shinde and A. K. Tyagi, Chem. Mater., 2009, 21, 5848–5859.
29. A. H. Hill, F. Jiao, P. G. Bruce, A. Harrison, W. Kockelmann and C. Ritter, Chem. Mater., 2008, 20, 4891–4899.
30. O. Crisan and A. D. Crisan, J. Alloys Compd., 2011, 509, 6522–6527.
31. M. Matvejeff, M. Lehtimaki, A. Hirasa, Y.-H. Huang, H. Yamauchi and M. Karppinen, Chem. Mater., 2005, 17, 2775.
32. P. Mars and D. W. van Krevelen, Chem. Eng. Sci., 1954, 3, 41–59.
33. M. Baldi, E. Finocchio, F. Milella and G. Busca, Appl. Catal., B, 1998, 16, 43–51.
34. S. Scirè, S. Minicò, C. Crisafulli, C. Satriano and A. Pistone, Appl. Catal., B, 2003, 40, 43–49.
35. M. Luo, J. Ma, J. Lu, Y. Song and Y. Wang, J. Catal., 2007, 246, 52–59.
36. S. Hosokawa, R. Tada, T. Shibano, S. Matsumoto, K. Teramura and T. Tanaka, Catal. Sci. Technol., 2016, 6, 7868–7874.
37. P. Bera, A. Gayen, M. S. Hegde, N. P. Lalla, L. Spadaro, F. Frusteri and F. Arena, J. Phys. Chem. B, 2003, 107, 6122–6130.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c7cy01861d

---