English

• 0.   Introduction

  Carbon dioxide reduction has received tremendous attention for its potential in solving the energy and environmental crises^[1-3]. The reverse water gas shift (RWGS) reaction is an effective route for syngas production via CO₂ hydrogenation^[4-6]. Numerous experimental studies have investigated the RWGS reaction, using Au/MoO_x^[7], Ru/Al₂O₃^[8], Cu/FeO_x^[9], and other monometallic^[10-12], bimetallic^[13], and metal oxides^[14] as catalysts. The reaction pathway can be summarized into two theories: the direct path and the indirect path. The direct path involves the direct CO₂ dissociation into CO_ad and O_ad, and then the CO_ad is desorbed as CO^[15]. The indirect path involves the hydrogenation of CO₂ via formate or carbonate species as intermediates, and the CO₂ further decomposes into CO as the main product^{[3-4, 16-17]}. The metal support interface also plays a vital role in the RWGS reaction^{[6-7, 18]} and affects the selectivity^[9]. To date, the reaction mechanism, including the active interface structure, is still under debate.

  Iron oxide-supported palladium catalysts have been extensively studied for the RWGS reaction^[19], water gas shift (WGS) reaction^[20], and CO oxidation^[21-23]. The Pd - FeO _x interface has been found to be crucial in a variety of reactions, as it promotes oxygen vacancies and oxygen transportation^[24-26]. The higher reducibility of the FeO_x - supported Pd species, which feature active interface structures, has also been extensively reported^{[20, 26-28]}. However, these works focus solely on the metal-support interaction from a static point of view, determined via ex situ characterization^[5]. The dynamic evolution of the Pd-FeO_x interface during the reaction has not yet been clarified owing to the lack of in situ techniques.

  In this study, Pd/FeO_x, Pd/SiO₂, and Pd - Fe/SiO₂ catalysts were systematically investigated for the RWGS reaction. The catalytic performance for the RWGS reaction was evaluated, and a promoting role of the Pd-FeO_x interface was observed. Using a combination of ex situ and in situ characterizations, we identified a dynamic evolution of the Pd-FeO_x interface during reaction, in which interfacial Pd^δ+-Fe²⁺ species were formed as the active sites. The underlying reaction mechanism and the role of the Pd-FeO_x interface were further investigated using temperature-programmed desorption/reduction/oxidation (TPD/TPR/TPO) and Fourier-transform infrared (FTIR) spectroscopy.

  1.   Experimental

  1.1   Catalysts preparation

  The FeO_x sample was prepared by the precipitation of Fe(NO₃)₃·9H₂O in a NaOH solution. The Pd/ FeO_x catalyst was prepared by the impregnation method, with 28.8 mg Pd(acac)₂ as the precursor and 200 mg FeO_x and 60 mL toluene as the solvent in a flask^[29]. Moreover, Pd/SiO₂ was prepared by a wet impregnation method, using 148.5 mg Pd(NH₃)₄Cl₂·H₂O and 1 g SiO₂. A GEMSTAR-6TM Benchtop atomic layer deposition (ALD) system was used to synthesize Pd-Fe/ SiO₂^[30]. Nitrogen (99.999%) was the carrier gas at 200 mL·min^-1, and FeO_x ALD was performed at 150 ℃ by exposing the as - prepared Pd/SiO₂ to 20~25 cycles of ferrocene and oxygen (For more details, see in Supplementary Information).

  1.2   Catalytic testing

  Catalytic testing was performed in a continuous - flow fixed-bed reactor with 40 mg of catalyst under atmospheric pressure, with a feed gas of 24% CO₂ and 72% H₂ (volume fraction, unless otherwise specified, all other gas contents in the paper are volume fractions), balanced with Ar. The total flow was 24 mL· min^-1, which resulted in a weight hourly space velocity of 36 000 mL·h^-1·g_cat^-1. After the gas phase was allowed to stabilize for 30 min to reach the steady-state for each reaction temperature, the outlet gas composition was analyzed by an on - line gas chromatograph (Agilent 7820A) equipped with a thermal conductivity detector and a flame ionization detector, and He was the carrier gas.

  CO₂ conversion was calculated as follows:

  $ {X_{{\rm{C}}{{\rm{O}}_{\rm{2}}}}} = \frac{{{n_0} - {n_1}}}{{{n_0}}} \times 100\% $

  (1)

  Where n₀ and n₁ stand for the amounts of CO₂ in feed gas and outlet gas.

  CO selectivity was calculated as follows:

  $ {S_{{\rm{CO}}}} = \frac{{{n_{{\rm{CO}}}}}}{{{n_0} - {n_1}}} \times 100\% $

  (2)

  Where n_CO stands for the amount of CO in outlet gas.

  1.3   Catalysts characterization

  X -ray diffraction (XRD, PAN analytical, X' Pert PRO X) was performed with a Cu Kα radiation source (λ =0.154 06 nm) in reflection mode. X - ray tube was operated at 40 kV and 40 mA. XRD diffraction range was set to 15° to 85°.

  The catalysts were characterized via highresolution transmission electron microscopy (TEM), using a JEOL JEM - 2100 microscope at 200 kV. The sample was dispersed by ethanol under ultrasonication. The samples were also analyzed via energy-dispersive X-ray spectroscopy (EDS), using the Aztec X-Max 80T EDS system (Oxford instrument, UK).

  Hydrogen temperature-programmed reduction (TPR) was performed on a Micromeritics AutoChem Ⅱ chemisorption analyzer. First, 50 mg catalyst was loaded into a U-shape quartz reactor and purged with helium at 120 ℃ for 2 h to remove physically adsorbed water and surface carbonates. After the catalyst was cooled to -50 ℃, the gas was switched to a flow of 10% (V/V) H₂ balanced with Ar, and the catalyst was heated to 500 ℃ at a ramping rate of 10 ℃·min^-1. The amount of H₂ consumption was calculated with the H₂ peak area and calibration curve of the 10% (V/V) H₂ balanced with Ar standard gas.

  Furthermore, CO₂ temperature-programmed desorption (TPD) experiments were performed on the AutoChem Ⅱ 2920. The samples were placed in a quartz tube. Before the TPD experiment, all of the samples were pre-reduced by 10% (V/V) H₂/Ar at 300 ℃ for 30 min and then cooled down to room temperature (RT). Then the gas was switched to 10% (V/V) CO₂/Ar at 50 mL·min^-1 for 1 h for the saturated adsorption of CO₂. After N2 was purged for 30 min, the sample was heated from 50 to 800 ℃ at a ramping rate of 10 ℃ · min^-1, during which TPD data were obtained.

  Cyclic H₂-TPR and CO₂ temperature-programmed oxidation (TPO) experiments were performed on an OmniSTAR gas analysis system. First, 20 mg catalyst was loaded into a U-shape quartz reactor and purged with helium at 150 ℃ for 2 h to remove physically adsorbed water and surface carbonates. After the temperature was cooled to room temperature, the gas was switched to a flow of 5% (V/V) H₂ balanced with He, and the catalyst was heated to 400 ℃ at a ramping rate of 10 ℃· min^-1. Before the CO₂-TPO experiment, the sample was purged again with He at 150 ℃ for 2 h. After the temperature was cooled to room temperature, the gas was switched to a flow of 5% (V/V) CO₂ balanced with He, and the catalyst was heated to 400 ℃ at a ramping rate of 10 ℃·min^-1.

  The FTIR spectra of CO adsorption were recorded in transmission mode on a Bruker Tensor 27 model. CO was introduced for 30 min at room temperature to reach saturated adsorption, followed by purging with N₂ to remove gaseous CO.

  X-ray photoelectron spectroscopy (XPS) experiments were performed using ThermoFisher ESCALAB 250Xi. A semi-in situ experiment was conducted by transferring the catalyst directly for XPS measurements after a certain treatment without exposing the catalyst to air, as reported elsewhere^[31]. The first set of experiments consisted of a reaction gas (24% CO₂+72% H₂+ 4% Ar, V/V) treatment for 30 min at room temperature and 400 ℃. The second set of experiments included three steps for each catalyst. First, the as-prepared catalyst was purged with N₂ at room temperature. Then, the sample was heated to 400 ℃ under 10% (V/V) H₂/ Ar for 30 min. Finally, the gas was switched to 10% (V/V) CO₂/Ar for 30 min. The sample was transferred back to the analysis chamber for XPS measurements in between each treatment.

  2.   Results and discussion

  2.1   Catalysts characterization

  The TEM images of as-prepared Pd-Fe/SiO₂ and the Pd/FeO_x catalysts are displayed in Fig. 1a and 1b. The Pd-Fe/SiO₂ showed uniformly dispersed Pd nanoparticles (NPs) of ~3 nm on amorphous SiO₂. In contrast, Pd/FeO_x displayed a highly polycrystalline FeO_x, with no significant Pd particle present (see EDS mapping in Fig. S1). The EDS spectra identified the presence of Pd with a similar loading (Table S1) mea- sured by inductively coupled plasma-optical emission etry (Table S2). The results indicate that a system of highly dispersed Pd species, with a size likely below 2 nm, in as- prepared 3.6% (mass fraction) Pd/ FeO_x catalysts was formed^[32-33].

  图 1

  

  Figure 1.  TEM images of Pd-Fe/SiO₂ (a) and Pd/FeO_x (b), inserted scale bar is 10 nm; (c) XRD patterns of the fresh and spent catalysts; (d) H₂-TPR results of Pd/FeO_x, blank FeO_x, Pd/SiO₂, and Pd-Fe/SiO₂ catalysts

  下载: 全尺寸图片 幻灯片

  Fresh (as-prepared) and spent catalysts were examined via XRD, and the results are presented in Fig. 1c. The fresh FeO_x and the 3.6% (mass fraction) Pd/ FeO_x catalysts before reaction showed a typical XRD pattern of Fe₂O₃, which was reduced to Fe₃O₄ completely after the RWGS reaction. The absence of Pd and PdO XRD patterns in both the fresh and spent Pd/FeO_x catalyst indicates the presence of small Pd/PdO_x nanoparticles (＜2 nm) or amorphous nature of Pd species in accordance with the TEM results^[34]. In the XRD patterns of the Pd/SiO₂ and Pd-Fe/SiO₂, a broad band existed at 21.9°, which can be assigned to amorphous SiO₂. The additional diffraction peaks at 40.4°, 46.7°, and 68.4° are ascribed to Pd NPs. Moreover, no FeO_x or Fe diffraction peaks were observed in Pd - Fe/SiO₂, suggesting a very thin layer of Fe species on the ALD- synthesized Pd NPs.

  In the H₂ - TPR results (Fig. 1d), the blank FeO_x support showed two reduction peaks: at 280 and 390 ℃, which are characteristic of the reduction of Fe₂O₃ to Fe₃O₄^[20]. In the case of the Pd/FeO_x catalyst, the first peak (0 ℃) is ascribed to the reduction of PdO, and the latter two (150 and 250 ℃) are associated with the reduction of FeO_x, which shift to lower temperatures. The presence of Pd thus facilitates the reduction of FeO_x, likely via H₂ spillover effect^{[20, 24]}. Compared with the H₂-TPR results of Pd/SiO₂ and Pd-Fe/SiO₂, the significantly enhanced H₂ consumption of Pd/FeO_x at 0 ℃ indicates a promoted reduction of the Pd - FeO_x interface, since the H₂ consumption was far more than the stoichiometric PdO amount (Table S3 and supplementary discussion). In addition, Pd-Fe/SiO₂ also showed a small reduction peak at 320 ℃, which corre-sponds to the reduction of ALD - synthesized FeO_x on Pd NPs.

  2.2   Catalytic tests

  The catalytic test was performed in a fixed - bed reactor under atmospheric pressure, with a feed gas of 24% CO₂ and 72% H₂ (V/V), balanced with Ar. As shown in Fig. 2a and 2b, the Pd/FeO_x catalyst exhibited a remarkably high activity for the RWGS reaction, with a CO₂ conversion of 10% at 300 ℃ and 29% at 400 ℃, and a CO selectivity of over 98% at all reaction temperatures. In contrast, both the blank FeO_x and Pd/SiO₂ catalysts were inactive, with CO₂ conversion below 5% at temperatures up to 400 ℃ (see carbon balance data in Table S4). This strongly indicates that the Pd-FeO_x interface is highly required for the RWGS reaction. To further prove this point, a thin layer of FeO_x (mass fraction of 1.1%) was added onto the inactive Pd/SiO₂ using ALD (the formed product was denoted as Pd-Fe/ SiO₂ catalyst). The Pd-Fe/SiO₂ catalyst showed a dramatically enhanced CO₂ conversion, from 5% to 20%, at 400 ℃ (Fig. 2c). Only a slight deactivation (＜4%) was observed, which was likely due to coking or mild sintering (Fig. S2 and S3). The enhanced activities of Pd/FeO_x and Pd-Fe/SiO₂ compared with those of their counterparts (FeO_x and Pd/SiO₂) reveal the promoting role of the Pd-FeO_x interface as highly active sites for catalytic CO₂ hydrogenation. The long-term stability test of the Pd/FeO_x catalyst was performed at 400 ℃ with time on stream up to 30 h, as shown in Fig. 2d. The CO₂ conversion and CO selectivity remain stable at ~29% and ~98% over the entire testing, suggesting an excellent stability of the catalyst. The water formation during reaction also shows a minor effect on the catalytic performance of Pd/FeO_x catalyst, with no significant deactivation in both CO₂ conversion and CO selectivity (Fig.S4).

  图 2

  

  Figure 2.  Catalytic performance of all of the catalysts tested for RWGS: (a) CO₂ conversion; (b) CO selectivity; (c) Comparison of Pd/FeO_x, Pd /SiO₂, and Pd-Fe/SiO₂ at 400 ℃; (d) Long term stability of Pd/FeO_x catalyst at 400 ℃ with time on stream up to 30 h

  下载: 全尺寸图片 幻灯片

  Furthermore, the Pd-FeO_x interface also promoted CO selectivity. As shown in Fig. 2b, the CO selectivity on the Pd/SiO₂ catalyst declined with increasing reaction temperature. The production of methane thus suggests a deep hydrogenation of CO on Pd/SiO₂ at elevated temperatures. In comparison, both Pd/FeO_x and Pd-Fe/SiO₂ catalysts maintained a CO selectivity of over 98% at all temperatures, suggesting that the addition of FeO_x could prevent the further hydrogenation of CO to methane. Table 1 summarizes the catalytic performance of other literature-reported RWGS catalysts. The 3.6% (mass fraction) Pd/FeO_x and 4.4% (mass fraction) Pd-Fe/SiO₂ catalysts in this work features a higher activity and CO selectivity than 1.5% (mass fraction) Rh/SrTiO₃^[35], 1.67% (mass fraction) Pt/TiO₂^[10], 3% (mass fraction) Pd/Fe₃O₄^[19], and other metal carbide catalysts^{[13, 36]}.

  表 1

  

  Table 1.  Summary of the conditions and reactivities of various catalysts in this work and the literature

  下载: 导出CSV

  Catalysta	VH2/VCO2	T/℃	Weight hourly space velocity/(L·h-1·g-1)	XCO2/ %	SCO/ %
  3.6% Pd/FeOx (this work)	3	300	36	11	100
  4.4% Pd-Fe/SiO2 (this work)	3	300	36	6	100
  3.6% Pd/FeOx (this work)	3	400	36	29	100
  4.4% Pd-Fe/SiO2 (this work)	3	400	36	20	100
  Mo2C[36]	3	300	36	8.7	93.9
  7.5% Co/Mo2C[36]	3	300	36	9.5	99
  1.67% Pt/TiO2[10]	1	300	351	4.5	99.1
  1.67% Pt/SiO2[10]	1	300	72.1	3.3	100
  1.1% Rh/SrTiO3[35]	1	300	120	7.9	95.4
  5% Co/MCF-17[13]	3	200~300	60	~5	90
  5% Pt50Co50/MCF-17[13]	3	200~300	60	~5	99
  1% Pt/TiO2[11]	3	600	12	~58	~95
  3% Pd/Fe3O4 (dopPPh2)[19]	4	400	60	8.8	94.6
  a All the percents are mass fraction.

  2.3   Dynamic formation of Pd^δ+ - Fe²⁺ interface as determined by semi-in situ XPS

  To identify the dynamic evolution of surface active sites, semi-in situ XPS experiments were performed, which enabled transferring the catalyst directly for XPS measurements after certain treatment without exposing it to air (for details, see in supplementary methods). The Pd/FeO_x, Pd/SiO₂, and Pd-Fe/SiO₂ catalysts were stepwise heated in 10% H₂ and 10% CO₂ at 400 ℃, and the Pd3d core-level evolution spectra are displayed in Fig. 3a. Three characteristic chemical states of Pd species could be identified from the evolu- tion spectra: 337.2 eV^[37], for oxidized Pd2+ state; 335.7 eV^[38], for partially oxidized Pd^δ+ state; and 334.8 eV^[38], for metallic Pd0 state (Fig.S5~S8). For Pd/SiO₂, the pristine catalyst showed a Pd3d_5/2 binding energy (BE) of 355.6 eV, corresponding to a Pd^δ+ state, which likely originated from the native oxide on the Pd NPs. In previous studies, a similar Pd^δ+ state was revealed by ambient - pressure XPS as a surface oxide and further assigned to Pd₅O₄ stoichiometry^[39-41]. Reduction in H₂ at 400 ℃ led to a prominent metallic Pd⁰ of 334.8 eV, which re-oxidized to a Pd^δ+ state in the subsequent CO₂ treatment (400 ℃). Considering the chemical inertness of the SiO₂ support, we can conclude that Pd⁰ was oxidized to Pd^δ+ by CO₂ at 400 ℃, strongly indicating a direct dissociation of CO₂ to CO on Pd NPs. The pristine Pd-Fe/SiO₂ and Pd/FeO_x catalysts showed a higher Pd3d_5/2 BE of 336.8 and 337.2 eV respectively, characteristic of PdO oxides^[37]. The similar redox properties of Pd - Fe/SiO₂ and Pd/FeO_x under H₂ and CO₂ treatments indicate the same reaction path of direct CO₂ dissociation for the RWGS reaction. Moreover, the addition of FeO_x significantly promoted the enrichment of Pd^δ+ species in the reduced Pd/SiO₂, Pd - Fe/SiO₂, and Pd/FeO_x catalysts after H₂ treatment, as indicated by the gradual upshift of Pd3d5/2 BE from metallic Pd⁰ to Pd^δ+ (334.8, 335.0, and 335.7 eV for Pd/SiO₂, Pd-Fe/SiO₂, and Pd/FeO_x, respectively) with increasing FeO_x content. The formation of Pd^δ+ species during the RWGS reaction was further verified by the treatment of Pd/FeO_x, Pd/SiO₂, and Pd - Fe/SiO₂ catalysts under the reaction gas of 24% CO₂+72% H₂+4% Ar (V/V) at 400 ℃. As shown in Fig. 3b, Fig.S9, and Table S5, the Pd/FeO_x, Pd/SiO₂, and Pd-Fe/SiO₂ catalysts after reaction exhibited a prominent Pd^δ++ state with a Pd3d_5/2 peak of 335.7 eV, which further indicates the stability of the Pd^δ+ surface oxides during the RWGS reaction.

  图 3

  

  Figure 3.  Semi-in situ XPS spectra: Pd3d core-level spectra of Pd/FeO_x, Pd/SiO₂, and Pd-Fe/SiO₂ catalysts during stepwise H₂ and CO₂ treatments (a) and after 400 ℃ treatment in the reaction gas (b); (c) Fe2p core-level spectra of Pd/FeO_x, Pd/SiO₂, and Pd-Fe/SiO₂ catalysts during stepwise H₂ and CO₂ treatments; (d) Oscillation of Fe²⁺ concentration and Pd valence state, deduced from Fe2p and Pd3^d XPS results

  下载: 全尺寸图片 幻灯片

  The Fe^2p XPS core-level spectra are displayed in Fig. 3c and Fig.S10~S11. In the stepwise H₂/CO₂ heating experiments (Fig. 3c), an evolution of the chemical state of iron oxides in Pd-Fe/SiO₂ occurred, resulting in a downshift of Fe²⁺p_3/2 BE from 711.0 eV (Fe³⁺ state) to 710.3 eV (a reduced Fe²⁺ state) after H₂ reduction and an upshift back to 711.0 eV after CO₂ treatment. The dynamic formation of Fe²⁺ species was thus facilitated by H₂ reduction, and the Fe²⁺+ species were further re-oxidized to Fe3+ state by reacting with CO₂. A similar oscillation of the Fe²⁺p_3/2 core level was also observed on Pd/FeO_x, but less prominent owing to the higher background signal from the bulk Fe²⁺O₃ support in the Pd/FeO_x catalyst. This further indicates the surfaceenrichment of Fe²⁺ species on the Pd/FeO_x catalyst, formed via hydrogen spillover to the periphery of the Pd NPs^[42-43]. After reaction gas treatment (Fig.S12), the Fe²⁺p3/2 of both Pd/FeO_x and Pd - Fe/SiO₂ underwent a downshift from 710.9 to 710.4 eV, and the characteristic satellite peak for Fe²⁺O3 (718 eV) was absent^[44], suggesting a reduction in iron oxide support, which agrees with our XRD results. Further reduction of the surface/ bulk Fe species may require higher temperature or extended reaction time.

  Fig. 3d displays the oscillation of Fe²⁺ concentra- tion and Pd valence state deduced from Fe²⁺p and Pd3d XPS results, as summarized in Tables S6 and S7. The enrichment of Pd^δ+ - Fe²⁺ on both Pd - Fe/SiO₂ and Pd/ FeO_x was identified under dynamic reduction by H₂, so that the Pd^δ++ - Fe²⁺+ further transformed into Pd²⁺ - Fe³⁺ after reacting with CO₂. Based on the higher activities of Pd-Fe/SiO₂ and Pd/FeO_x compared with that of Pd/ SiO₂, we can assign the dynamic Pd^δ+-Fe²⁺ interface as a highly active site for CO₂ dissociation during RWGS.

  2.4   Promoting role of Pd-FeO_x interface in adsorption/desorption process

  Fig. 4a displays the FTIR spectra of CO adsorption over spent catalysts. Two vibrational bands of CO stretching modes, at 2 086 and 1 965 cm^-1, were observed on Pd/SiO₂, which can be assigned to CO adsorption on multi-coordinated sites of Pd NPs^[45-46]. Moreover, the CO adsorption intensity dramatically diminished on Pd-Fe/SiO₂ and eventually became invisible on Pd/FeO_x. The absence of CO adsorption on Pd/ FeO_x can be ascribed to the higher oxidized Pd^δ+/Pd²⁺ state, which may also lead to encapsulation via strong metal-support interaction^[47]. The weakening of CO adsorption promoted by iron oxides facilitated the desorption of CO as a product and prevented the deep hydrogenation for methane formation.

  图 4

  

  Figure 4.  CO and CO₂ adsorption: (a) FTIR spectra of CO adsorption over spent Pd/SiO₂, Pd-Fe/SiO₂, and Pd/FeOx catalysts (All IR spectra were acquired at room temperature after evacuating gaseous CO); (b) CO₂-TPD spectra of reduced FeO_x, Pd/SiO₂, Pd/FeO_x, and Pd-Fe/SiO₂ at temperatures up to 750 ℃

  下载: 全尺寸图片 幻灯片

  Fig. 4b presents the CO₂ -TPD spectra of reduced FeO_x, Pd/SiO₂, Pd/FeO_x, and Pd - Fe/SiO₂ at temperatures up to 750 ℃. All of the tested catalysts were prereduced in 10% H₂ at 300 ℃. FeO_x presented only a subtle peak at 400 ℃, suggesting a very weak CO₂ adsorption on Fe²⁺O³. In contrast, Pd/SiO₂ showed a desorption peak at ~550 ℃, which can be ascribed to the molecular/reactive desorption of chemically adsorbed CO₂ on Pd NPs in a mixed Pd0/Pd^δ+ state, as identified by XPS. An additional desorption peak at higher temperatures (~670 ℃) was observed on both Pd/FeO_x and Pd - Fe/SiO₂, reflecting an enhanced CO₂ adsorption owing to the Pd^δ+-FeO_x interface, which will lead to the formation of carbonate species or other adsorbed intermediates^[48].

  2.5   Reaction mechanism

  The semi-in situ XPS results demonstrated a dynamic evolution of Pd and Fe chemical states. To further reveal the reaction pathway, we performed cyclic H₂ -TPR and CO₂ -TPO on Pd/SiO₂ and Pd/FeO_x catalysts. In the case of the Pd/SiO₂ catalyst (Fig. 5a), an indirect path was identified, as indicated by the pronounced CO production along the second H₂-TPR cycle, via the hydrogenation of either adsorbed CO₂ or carbonate species on the metallic Pd surfaces^{[17, 49]}. Meanwhile, the formation of a little amount of CO was also detected in the CO₂ - TPO cycle, suggesting the coexistence of direct CO₂ dissociation, likely on the remanence of Pd^δ+ species, as identified in the semi-in situ XPS experiments. Moreover, the Pd/FeO_x (Fig. 5b) showed a remarkably enhanced CO formation in the CO₂ - TPO cycle (direct path), while the hydrogenation path (indirect path) was significantly inhibited, as no CO signal was observed in the second H₂-TPR cycle.

  图 5

  

  Figure 5.  Cyclic H₂-TPR/CO₂-TPO experiments on Pd/SiO₂ (a) and Pd/FeO_x (b) catalysts (Y-axis is the ion intensity of m/z=28 for CO); Schematic reaction path of CO₂ hydrogenation on Pd/SiO₂ (c) and Pd/FeO_x (d) catalysts

  下载: 全尺寸图片 幻灯片

  Based on the dynamic evolution of the Pd/Fe chemical state based on the semi- in situ XPS results, we can propose the overall reaction pathways and underlying mechanism for the RWGS reaction over Pd-Fe catalysts (Fig. 5c and 5d).

  (1) On Pd/SiO₂, the majority of CO₂ is hydrogenated on the metallic Pd surface via an indirect path, leading to the formation of formate or bicarbonate intermediates, which are eventually decomposed to produce CO. Though less favorable, the direct path of CO₂ dissociation also coexists on Pd/SiO₂, which is likely facilitated by the remanence of Pd^δ+ species near the SiO₂ interface. The enhanced CO adsorption on metallic Pd⁰ favors the deep hydrogenation of CO, leading to undesired CH₄ as a byproduct^[50]. The low activity of Pd/SiO₂ is caused by the weak CO₂ adsorption, as indicated by the CO₂-TPD experiments.

  (2) On Pd/FeO_x and Pd-Fe/SiO₂, the Pd-FeO_x interface significantly enhances the CO₂ adsorption, resulting in a remarkably high CO₂ conversion compared with that on Pd/SiO₂. The dynamic formation of Pd^δ+ - Fe²⁺ species promoted CO₂ dissociation, likely following the Mars-van Krevelen mechanism, via active oxygen vacancies on the interface^{[8, 18, 43]}:

  $ {\rm{P}}{{\rm{d}}^{{\rm{2 + }}}}-{\rm{O}}-{\rm{F}}{{\rm{e}}^{{\rm{3 + }}}}{\rm{ + }}{{\rm{H}}_{\rm{2}}} \to {\rm{P}}{{\rm{d}}^{\delta + }}-{{\rm{O}}_{\rm{v}}}{\rm{ + F}}{{\rm{e}}^{{\rm{2 + }}}}{\rm{ + }}{{\rm{H}}_{\rm{2}}}{\rm{O}} $

  (3)

  $ {\rm{P}}{{\rm{d}}^{\delta + }}-{{\rm{O}}_{\rm{v}}} + {\rm{F}}{{\rm{e}}^{2 + }} + {\rm{C}}{{\rm{O}}_2} \to {\rm{P}}{{\rm{d}}^{2 + }}-{\rm{O}} - {\rm{F}}{{\rm{e}}^{3 + }} + {\rm{CO}} $

  (4)

  The weakening of CO adsorption on the oxidized Pd^δ+/Pd²⁺ species enriched by iron oxide further enhances the CO selectivity (>98%) by preventing the deep hydrogenation for CH₄ formation.

  3.   Conclusions

  In this study, Pd/FeO_x, Pd-Fe/SiO₂, Pd/SiO₂, and FeO_x catalysts were systematically investigated for the RWGS reaction. The Pd/FeO_x and Pd-Fe/SiO₂ catalysts exhibited extraordinary activity for the RWGS reaction, with CO₂ conversions of 29% and 20% at 400 ℃, respectively, and remarkably high CO selectivity (over 98%) at all considered reaction temperatures. Thus, the Pd-FeO_x interface promoted the CO₂ hydrogenation to CO. Moreover, semi - in situ XPS and cyclic TPR/ TPO experiments identified the dynamic formation of Pd^δ+ -Fe²⁺ species during the RWGS reaction; the species acted as highly active sites for direct CO₂ dissociation via oxygen vacancies enriched on the interface. The formation of Pd^δ+ - FeO_x species was confirmed to play a significant role in the CO₂ adsorption and CO desorption, as they further enhanced the catalytic activity and selectivity of CO for the RWGS reaction. Our results provide new insights for the dynamic understanding of the metal- oxide interfaces during reactions and benefit the rational design of heterogeneous catalysts from a dynamic point of view.

  
  Supporting information is available at http://www.wjhxxb.cn
  
• 1. [1]

     Li W H, Wang H Z, Jiang X, Zhu J, Liu Z M, Guo X W, Song C S. RSC Adv., 2018, 8(14):7651-7669 doi: 10.1039/C7RA13546G

  2. [2]

     Dorner R W, Hardy D R, Williams F W, Willauer H D. Energy Environ. Sci., 2010, 3(7):884-890 doi: 10.1039/c001514h

  3. [3]

     Wang W, Wang S P, Ma X B, Gong J L. Chem. Soc. Rev., 2011, 40(7):3703-3727 doi: 10.1039/c1cs15008a

  4. [4]

     Daza Y A, Kuhn J N. RSC Adv., 2016, 6(55):49675-49691 doi: 10.1039/C6RA05414E

  5. [5]

     Parkinson G S. Surf. Sci. Rep., 2016, 71(1):272-365

  6. [6]

     Liu M, Yi Y H, Wang L, Guo H C, Bogaerts A. Catalysts, 2019, 9(3):275 doi: 10.3390/catal9030275

  7. [7]

     Carrasquillo-Flores R, Ro I, Kumbhalkar M D, Burt S, Carrero C A, Alba-Rubio A C, Miller J T, Hermans I, Huber G W, Dumesic J A. J. Am. Chem. Soc., 2015, 137(32):10317-10325 doi: 10.1021/jacs.5b05945

  8. [8]

     Yan Y, Wang Q J, Jiang C Y, Yao Y, Lu D, Zheng J W, Dai Y H, Wang H M, Yang Y H. J. Catal., 2018, 367:194-205 doi: 10.1016/j.jcat.2018.08.026

  9. [9]

     Halder A, Kilianová M, Yang B, Tyo E C, Seifert S, Prucek R, Panáček A, Suchomel P, Tomanec O, Gosztola D J, Milde D, Wang H H, Kvítek L, Zbořil R, Vajda S. Appl. Catal. B, 2018, 225:128-138 doi: 10.1016/j.apcatb.2017.11.047

  10. [10]

      Kattel S, Yan B H, Chen J G, Liu P. J. Catal., 2016, 343:115-126 doi: 10.1016/j.jcat.2015.12.019

  11. [11]

      Kim S S, Park K H, Hong S C. Fuel Process. Technol., 2013, 108:47-54 doi: 10.1016/j.fuproc.2012.04.003

  12. [12]

      Chen Y, Lin J, Li L, Qiao B T, Liu J Y, Su Y, Wang X D. ACS Catal., 2018, 8(2):859-868

  13. [13]

      Alayoglu S, Beaumont S K, Zheng F, Pushkarev V V, Zheng H M, Iablokov V, Liu Z, Guo J H, Kruse N, Somorjai G A. Top. Catal., 2011, 54(13/14/15):778-785

  14. [14]

      Zhang Y L, Fu D L, Liu X L, Zhang Z P, Zhang C, Shi B F, Xu J, Han Y F. ChemCatChem, 2018, 10(6):1272-1276 doi: 10.1002/cctc.201701779

  15. [15]

      Fujita S. J. Catal., 1992, 134(1):220-225

  16. [16]

      Goguet A, Meunier F C, Tibiletti D, Breen J P, Burch R. J. Phys. Chem. B, 2004, 108(52):20240-20246 doi: 10.1021/jp047242w

  17. [17]

      Boccuzzi F, Chiorino A, Manzoli M, Andreeva D, Tabakova T. J. Catal., 1999, 188(1):176-185

  18. [18]

      Bobadilla L F, Santos J L, Ivanova S, Odriozola J A, Urakawa A. ACS Catal., 2018, 8(8):7455-7467 doi: 10.1021/acscatal.8b02121

  19. [19]

      Caparrós F J, Soler L, Rossell M D, Angurell I, Piccolo L, Rossell O, Llorca J. ChemCatChem, 2018, 10(11):2365-2369 doi: 10.1002/cctc.201800362

  20. [20]

      Sun X C, Lin J, Zhou Y L, Li L, Su Y, Wang X D, Zhang T. AlChE J., 2017, 63(9):4022-4031 doi: 10.1002/aic.15759

  21. [21]

      Pavlova S N, Savchenko V I, Sadykov V A, Zaikovskii V I, Kalinkin A V. React. Kinet. Catal. Lett., 1996, 59(1):103-110 doi: 10.1007/BF02067998

  22. [22]

      Bi Y S, Dang G Y, Zhao X H, Meng X F, Lu H J, Jin J T. J. Hazard. Mater., 2012, 229-230:245-250 doi: 10.1016/j.jhazmat.2012.05.086

  23. [23]

      Liu L Q, Zhou F, Wang L G, Qi X J, Shi F, Deng Y Q. J. Catal., 2010, 274(1):1-10

  24. [24]

      Hensley A J R, Hong Y C, Zhang R Q, Zhang H, Sun J M, Wang Y, McEwen J S. ACS Catal., 2014, 4(10):3381-3392 doi: 10.1021/cs500565e

  25. [25]

      Cheng Z, Qin L, Guo M, Fan J A, Xu D, Fan L S. Phys. Chem. Chem. Phys., 2016, 18(24):16423-16435 doi: 10.1039/C6CP01287F

  26. [26]

      Liang J X, Yu Q, Yang X F, Zhang T, Li J. Nano Res., 2018, 11(3):1599-1611

  27. [27]

      Fu Q, Li W X, Yao Y, Liu H, Su H Y, Ma D, Gu X K, Chen L, Wang Z, Zhang H, Wang B, Bao X. Science, 2010, 328(5982):1141-1144 doi: 10.1126/science.1188267

  28. [28]

      Sun Y N, Giordano L, Goniakowski J, Lewandowski M, Qin Z H, Noguera C, Shaikhutdinov S, Pacchioni G, Freund H J. Angew. Chem. Int. Ed., 2010, 49(26):4418-4421 doi: 10.1002/anie.201000437

  29. [29]

      Xie Y H, Li B, Weng W Z, Zheng Y P, Zhu K T, Zhang N W, Huang C J, Wan H L. Appl. Catal. A, 2015, 504:179-186 doi: 10.1016/j.apcata.2014.12.008

  30. [30]

      Yi H, Xia Y J, Yan H, Lu J L. Chin. J. Catal., 2017, 38(9):1581-1587 doi: 10.1016/S1872-2067(17)62768-2

  31. [31]

      Hong Y C, Zhang H, Sun J M, Ayman K M, Hensley A J R, Gu M, Engelhard M H, McEwen J S, Wang Y. ACS Catal., 2014, 4(10):3335-3345 doi: 10.1021/cs500578g

  32. [32]

      Sun W Z, Li Q, Gao S A, Shang J K. Appl. Catal. B, 2012, 125:1-9 doi: 10.1016/j.apcatb.2012.05.014

  33. [33]

      Jiang T, Du S C, Jafari T, Zhong W, Sun Y, Song W Q, Luo Z, Hines W A, Suib S L. Appl. Catal. A, 2015, 502:105-113 doi: 10.1016/j.apcata.2015.05.013

  34. [34]

      Wang F G, Xu Y, Zhao K F, He D N. Nano-micro Lett., 2014, 6(3):233-241

  35. [35]

      Yan B H, Wu Q Y, Cen J J, Timoshenko J, Frenkel A I, Su D, Chen X Y, Parise J B, Stach E, Orlov A, Chen J G. Appl. Catal. B, 2018, 237:1003-1011 doi: 10.1016/j.apcatb.2018.06.074

  36. [36]

      Porosoff M D, Yang X F, Boscoboinik J A, Chen J G. Angew. Chem. Int. Ed., 2014, 53(26):6705-6709 doi: 10.1002/anie.201404109

  37. [37]

      Brun M, Berthet A, Bertolini J C. J. Electron. Spectrosc. Relat. Phenom., 1999, 104(1/2/3):55-60

  38. [38]

      Tura J M, Regull P, Victori L S, de Castellar M D. Surf. Interface Anal., 1988, 11(8):447-449 doi: 10.1002/sia.740110807

  39. [39]

      Klikovits J, Napetschnig E, Schmid M, Seriani N, Dubay O, Kresse G, Varga P. Phys. Rev. B, 2007, 76(4):045405

  40. [40]

      Duan Z Y, Henkelman G. ACS Catal., 2014, 4(10):3435-3443

  41. [41]

      Toyoshima R, Yoshida M, Monya Y, Kousa Y, Suzuki K, Abe H, Mun B S, Mase K, Amemiya K, Kondoh H. J. Phys. Chem. C, 2012, 116(35):18691-18697 doi: 10.1021/jp301636u

  42. [42]

      Jung K D, Bell A T. J. Catal., 2000, 193(2):207-223

  43. [43]

      Yang B, Yu X, Halder A, Zhang X B, Zhou X, Mannie G J A, Tyo E, Pellin M J, Seifert S, Su D S, Vajda S. ACS Sustainable Chem. Eng., 2019, 7(17):14435-14442

  44. [44]

      Grosvenor A P, Kobe B A, Biesinger M C, McIntyre N S. Surf. Inter-face Anal., 2004, 36(12):1564-1574

  45. [45]

      Sun K, Wilson A R, Thompson S T, Lamb H H. ACS Catal., 2015, 5(3):1939-1948

  46. [46]

      Fan Q N, He S, Hao L, Liu X, Zhu Y, Xu S L, Zhang F Z. Sci. Rep., 2017, 7(1):42172

  47. [47]

      d'Alnoncourt R N, Friedrich M, Kunkes E, Rosenthal D, Girgsdies F, Zhang B S, Shao L D, Schuster M, Behrens M, Schlögl R. J. Catal., 2014, 317:220-228

  48. [48]

      Jacobs G, Davis B H. Appl. Catal. A, 2005, 284(1/2):31-38

  49. [49]

      Chen C S, Wu J H, Lai T W. J. Phys. Chem. C, 2010, 114(35):15021-15028

  50. [50]

      Park J N, Mcfarland E W. J. Catal., 2009, 266(1):92-97

• 

  Figure 1  TEM images of Pd-Fe/SiO₂ (a) and Pd/FeO_x (b), inserted scale bar is 10 nm; (c) XRD patterns of the fresh and spent catalysts; (d) H₂-TPR results of Pd/FeO_x, blank FeO_x, Pd/SiO₂, and Pd-Fe/SiO₂ catalysts

  下载: 全尺寸图片 幻灯片
  

  Figure 2  Catalytic performance of all of the catalysts tested for RWGS: (a) CO₂ conversion; (b) CO selectivity; (c) Comparison of Pd/FeO_x, Pd /SiO₂, and Pd-Fe/SiO₂ at 400 ℃; (d) Long term stability of Pd/FeO_x catalyst at 400 ℃ with time on stream up to 30 h

  下载: 全尺寸图片 幻灯片
  

  Figure 3  Semi-in situ XPS spectra: Pd3d core-level spectra of Pd/FeO_x, Pd/SiO₂, and Pd-Fe/SiO₂ catalysts during stepwise H₂ and CO₂ treatments (a) and after 400 ℃ treatment in the reaction gas (b); (c) Fe2p core-level spectra of Pd/FeO_x, Pd/SiO₂, and Pd-Fe/SiO₂ catalysts during stepwise H₂ and CO₂ treatments; (d) Oscillation of Fe²⁺ concentration and Pd valence state, deduced from Fe2p and Pd3^d XPS results

  下载: 全尺寸图片 幻灯片
  

  Figure 4  CO and CO₂ adsorption: (a) FTIR spectra of CO adsorption over spent Pd/SiO₂, Pd-Fe/SiO₂, and Pd/FeOx catalysts (All IR spectra were acquired at room temperature after evacuating gaseous CO); (b) CO₂-TPD spectra of reduced FeO_x, Pd/SiO₂, Pd/FeO_x, and Pd-Fe/SiO₂ at temperatures up to 750 ℃

  下载: 全尺寸图片 幻灯片
  

  Figure 5  Cyclic H₂-TPR/CO₂-TPO experiments on Pd/SiO₂ (a) and Pd/FeO_x (b) catalysts (Y-axis is the ion intensity of m/z=28 for CO); Schematic reaction path of CO₂ hydrogenation on Pd/SiO₂ (c) and Pd/FeO_x (d) catalysts

  下载: 全尺寸图片 幻灯片

  Table 1.  Summary of the conditions and reactivities of various catalysts in this work and the literature

  Catalysta	VH2/VCO2	T/℃	Weight hourly space velocity/(L·h-1·g-1)	XCO2/ %	SCO/ %
  3.6% Pd/FeOx (this work)	3	300	36	11	100
  4.4% Pd-Fe/SiO2 (this work)	3	300	36	6	100
  3.6% Pd/FeOx (this work)	3	400	36	29	100
  4.4% Pd-Fe/SiO2 (this work)	3	400	36	20	100
  Mo2C[36]	3	300	36	8.7	93.9
  7.5% Co/Mo2C[36]	3	300	36	9.5	99
  1.67% Pt/TiO2[10]	1	300	351	4.5	99.1
  1.67% Pt/SiO2[10]	1	300	72.1	3.3	100
  1.1% Rh/SrTiO3[35]	1	300	120	7.9	95.4
  5% Co/MCF-17[13]	3	200~300	60	~5	90
  5% Pt50Co50/MCF-17[13]	3	200~300	60	~5	99
  1% Pt/TiO2[11]	3	600	12	~58	~95
  3% Pd/Fe3O4 (dopPPh2)[19]	4	400	60	8.8	94.6
  a All the percents are mass fraction.

  下载: 导出CSV
•