Concerted Oxygen Diffusion Across Heterogeneous Oxide Interfaces For Intensified Propane Dehydrogenation

Propane dehydrogenation (PDH) is an industrial technology for direct propylene production that has received extensive attention in recent years. Nevertheless, existing non-oxidative dehydrogenation technologies still suffer from thermodynamic equilibrium limitations and severe choking. Here, we develop the intensified propane dehydrogenation to propylene by the chemical looping engineering on nanoscale core-shell redox catalysts. The core-shell redox catalyst combines dehydrogenation catalyst and solid oxygen carrier at one particle, preferably composed of two to three atomic layer-type vanadia coating ceria nanodomains. The highest 93.5% propylene selectivity is obtained, sustaining 43.6% propylene yield under 300 long-term dehydrogenation-oxidation cycles, which outperforms an analog of industrially relevant K-CrOx/Al2O3 catalysts and exhibits 45% energy savings in the scale-up of chemical looping scheme. Combining in situ spectroscopies, kinetics, and theoretical calculation, an intrinsically dynamic lattice oxygen “donator acceptor” process is proposed that O2- generated from the ceria oxygen carrier is boosted to diffuse and transfer to vanadia dehydrogenation sites via a concerted hopping pathway at the interface, stabilizing surface vanadia with moderate oxygen coverage at pseudo steady state for selective dehydrogenation without significant overoxidation or cracking.

Click Here To Get More Info About Cistanche Anti-Oxidation Function

Propane dehydrogenation (PDH) is an industrially important alternative to oil-based cracking processes1,2. However, the commercial non-oxidative propane dehydrogenation containing CrOx or Pt-based catalysts is endothermic and equilibrium-limited, necessitating much heat to achieve viable propylene yield3,4. Although the oxidative dehydrogenation of propane (ODH) has the potential to improve conversion for favorable thermodynamics, propylene selectivity is hampered by overoxidation to CO2 5,6. A similar challenge is faced in selective oxidation reactions in the chemical industry7,8.

Chemical looping engineering offers exciting new opportunities for the challenges through the physical or temporal separation of dehydrogenation and oxidation by solid oxygen carrier mediums9,10. Unlike traditional catalysts, the oxygen carriers react with alkanes and undergo reversible changes by donating and replenishing oxygen to close the loop in the reducer and oxidizer reactors. Most oxygen carriers involve the metal centers or oxide composites to modulate lattice oxygen reactivity, using bulk doping11, surface modification12, or con- refinement in supports13. Recently, vanadia/ceria catalysts have attracted increased attention in the oxidative dehydrogenation of propane with O2 co-feeding. The electronic effects and redox properties were investigated at the molecular level13–16. Nevertheless, direct experimental and theoretical insights into the lattice oxygen diffusion and the surface dynamics have not been reported yet for the anaerobic oxidative dehydrogenation via chemical looping engineering. 

In this work, to unravel the oxygen diffusion and reaction dynamics regarding the active sites, a nanoscale core-shell redox catalyst combining dehydrogenation catalyst and oxygen carrier at one particle is designed. The core-shell redox catalyst is preferably composed of two to three atomic layer-type vanadia coating ceria nanodomains to achieve the synergetic modulation of lattice oxygen bulk diffusion and surface reaction. In the dehydrogenation step (reducer), ceria-vanadia redox catalysts donate lattice oxygen for the dehydrogenation of propane to produce propylene, H2O, and H2, affording a reduced valence state that can be reoxidized in the reoxidation step (oxidizer) by air to close the loop (Fig. 1a). Combining in situ spectroscopies, kinetics, and theoretical calculation, an intrinsically dynamic lattice oxygen “donator-acceptor” process is proposed, which accounts for the synergetic modulation of bulk diffusion and surface reaction in the core-shell redox catalyst. O2− generated from ceria oxygen carrier is boosted to diffuse and transfer to vanadia dehydrogenation sites via a concerted hopping pathway at the interface, stabilizing surface vanadia with moderate oxygen coverage without signifificant overoxidation or cracking.

Results Formation of ceria-vanadia core-shell redox catalysts The core-shell redox catalysts were prepared using a two-step incipient wetness impregnation method. The ceria-vanadia samples were named xV/yCeAl, where x(y) is the percent weight of V(Ce). The vanadia and ceria catalysts were obtained by VOx and CeO2 supported on γ-Al2O3, respectively. At vanadia surface density of 4.3 V/nm2 (Supplementary Table 1)17–19 (6 V/30CeAl), atom-resolved high-angle annular dark-field scanning transmission electron microscope (HAADF-STEM) images identified vanadia mainly existed as monolayers and bilayers along ceria surface (Fig. 1b–d). Electron energy loss spectra (EELS) mappings of V L2,3 and Ce M4,5 edges affirmed vanadia sites anchored on widespread ceria nanodomains (Fig. 1e–h). The well-defined core-shell structure was further validated by line-scanning EELS that crossed an individual particle, wherein the outer shell was ~1 nm, corresponding to roughly two to three atomic vanadia layers (Fig. 1i). Judged from variations of V L2,3 and O K edges, vanadia presented dominantly as a mixture of V5+ and V4+, while relative intensity ratios of Ce M4,5 edges (1.11–1.19)20 indicated the presence of Ce3+ and Ce4+ ((1), (2), (3) in Fig. 1b) inside one particle (Fig. 1j, Supplementary Fig. 2, and Supplementary Table 2)21.

Chemical looping oxidative dehydrogenation performance Application of core-shell redox catalysts was proven in a continuous chemical looping oxidative dehydrogenation scheme (Supplementary Fig. 3). Ceria-vanadia redox catalysts exhibited traceable CO2 (<3%) with high propylene selectivity of 93.5% and formation rate of 42.5 mmol C3H6/gcat/h (at 5th min in one cycle), implying excessive overoxidation or cracking were inhibited. At 600 o C and GHSV of 2500 h−1, an average 90% propylene selectivity at propane conversion of 49% was obtained within 60 mins (Fig. 2b and Supplementary Fig. 4), superior to that of ceria (30CeAl) (78.3%), vanadia (6 V/Al) (71.6%), and state-of-the-art catalysts (Fig. 2c). Industrially relevant K-CrOx/Al2O3 4,22 was compared under identical reaction conditions. The propylene space-time yield (STY) of ceria-vanadia redox catalysts was 10.3 mmol C3H6/gcat/h, comparable to that of K-CrOx/Al2O3 (10.6 mmol C3H6/gcat/h) (Supplementary Fig. 3g). However, considering different reaction sites in two catalysts, propylene STY normalized by moles of V (13.3 mol C3H6/molV/h) was about fifive times higher than that normalized by moles of Cr (2.8 mol C3H6/molCr/h). The deactivation rate constant (kd) using a first-order deactivation model was used to determine its life in the dehydrogenation step. Ceria-vanadia redox catalysts exhibited smaller kd (0.04 h−1 ) than K-CrOx/Al2O3 (0.99 h−1 ) (Supplementary Figs. 5, 6). When the temperature was increased to 650 o C, propylene selectivity remained at 80%. However, pure vanadia showed quick deactivation (kd = 1.4 h−1 ) and propylene selectivity decreased to 34%.

The reversible charge-discharge of lattice oxygen in ceria-vanadia redox catalysts was verified by in situ XRD. In the dehydrogenation step at 600 o C, diffract peaks of CeO2 shifted to lower diffraction angles, e.g., the (111) diffract peak shifted from 28.4 o to 28.0 o due to the formation of larger Ce3+ ions. Oxidation with air then recovered its position (Fig. 2a and Supplementary Fig. 7). During 300 long-term chemical looping cycles, structure durability and robust performance with an average 43.6% C3H6 yield and space-time yield of 9.9 mmol C3H6/gcat/h was achieved (Fig. 2d and Supplementary Table 5). When altering either shell or core components in core-shell redox catalysts, comparable C3H6 formation rates were obtained (Supplementary Fig. 8). For chemical looping oxidative dehydrogenation of ethane, the ceria-vanadia redox catalysts also presented 92% ethylene selectivity with 31% ethane conversion at 600 o C (Supplementary Fig. 9), validating its potential application in the dehydrogenation of light alkanes. Compared with the commercialized Oleflflex scheme (Supplementary Fig. 10 and Supplementary Tables 6–10), 45% of energy savings can be anticipated from the chemical looping oxidative dehydrogenation system (Fig. 2e), with separation being the main driver for energy consumption.

Evidence of oxygen diffusion and surface reaction 

When exposed to propane for 120 mins, peaks of B2 and C in the Ce L3- edge shifted to lower energy (Δ2.1 eV). The B0 white line located at 5726 eV was then dominated, a characteristic of Ce3+ (Fig. 3a), indicating the reduction of ceria (Ce4+→Ce3+) in the ceria-vanadia catalysts, in contrast to the negligible formation of Ce3+ in pure ceria (Ce L3-edge shift of Δ0.7 eV). V K pre-edge close to 5467 eV featuring V4+ oxidation state kept nearly unmoved within 30 mins. After that, a decrease of pre-edge peak intensity and shift of edge position to lower energy (~Δ1.2 eV) occurred as the reduction of CeO2 stopped (Fig. 3b and Supplementary Fig. 11). This implied that in the ceria-vanadia redox catalysts, vanadia tended to be reduced to lower valence states when oxygen was not timely supplied from ceria. For pure vanadia without CeO2 supporting, V K pre-edge featuring V5+ was easily and quickly reduced to V3+ (V K-edge shift of 2.7 eV)11. Together, the changes of Ce L3-edge and V K-edge indicated that ceria in ceria-vanadia redox catalysts acted as an “oxygen reservoir” that could supply the lattice oxygen to stabilize the surface vanadia, which accords with the previous research that ceria helped to oxidize the reduced vanadia via ceria lattice oxygen23–25.

We further evidenced the dynamic evolution of lattice oxygen in ceria-vanadia redox catalysts. Raman spectra of CeO2 were dominated by the strong F2g mode of the fluorite phase at 464 cm−1 with weak bands at 598 cm−1 due to defect-induced (D) mode. With vanadia coating, in addition to V = O and V-O-V stretching, additional bands of V-O-Ce (859 and 720 cm−1 ) emerged (Supplementary Fig. 1j, k), affirming the construction of vanadia-ceria interface26,27. Upon propane exposure, in situ, Raman spectra verified the continuous reduction of CeO2 in ceriavanadia redox catalysts that the intensity of F2g mode dramatically decreased with the time on stream. It is noted that the band of V-O-Ce kept relatively stable. In contrast, intensity ratios of V-O-Ce band and F2g mode in terms of IV-O-Ce/IF2g increased, validating that Ce-O species in ceria were gradually consumed to supplement and stabilize interfacial and surface V-O species (Fig. 3c and Supplementary Fig. 12)24,25. As much oxygen was depleted after 30 min, the time was also shown in situ XANES spectra. The D1 band and G band corresponding with coke deposits were then observed, implying that the cracking and coking of propane occurred on reduced vanadia sites. Comparatively, pure vanadia was readily reduced to V3+, which leads to more coke deposition in the characteristic of the more dominated intensity of D1 and G band27–29.

Fig. 1 | Identifification of vanadia layers that coat ceria nanodomains. a Diagram of core-shell redox catalysts in propane dehydrogenation by the chemical looping engineering: dehydrogenation and oxidation in fuel reactor (reducer) and air reactor (oxidizer), respectively. b–d HAADF-STEM images and e–h EELS mappings of core-shell ceria-vanadia redox catalysts (6 V/30CeAl): (f): V; (g): Ce; (h): V + Ce. i Line-scanning EELS. j EELS of the domains ((1), (2), (3)) in (b).

Fig. 2 | Chemical looping oxidative dehydrogenation performance. a In situ XRD patterns of ceria-vanadia redox catalysts (6 V/30CeAl). b Comparison of ceria (30CeAl), vanadia (6 V/Al), and ceria-vanadia redox catalysts (6 V/30CeAl). Conditions: 600 oC, GHSV = 2500 h−1, C3H8/N2 = 0.25. c Comparing ceria-vanadia redox catalysts (6 V/30CeAl) with established oxide-based and Pt-containing catalysts (see Supplementary Tables 3, 4). Motifs of triangle, rhombus, and sphere represent ODH, PDH, and CL-ODH, respectively. d Cyclic performance over ceria-vanadia redox catalysts (6 V/30CeAl). Dehydrogenation step: 600 oC, GHSV = 2500 h−1, C3H8/N2 = 0.25 for 30 min; Inert purge: 600 oC, N2 = 40 mL/min for 5 min; Oxidation step: 600 oC, 20 vol.% O2/N2 = 20 mL/min for 15 min. e Comparison of energy consumption and CO2 emission of traditional Oleflflex technology and chemical looping scheme (see Methods and Supplementary Tables 6–8).

Fig. 3 | Experimental evidence of oxygen diffusion and surface reaction. an In situ XANES spectra of Ce L3-edge (CeO2 standards as the references) over ceria-vanadia (6 V/30CeAl) (top) and pure ceria (30CeAl) (bottom) and b V K-edge (V foil, V2O5, VO2, and V2O3 standards as the references) over ceria-vanadia (6 V/30CeAl) (top) and pure vanadia (6 V/Al) (bottom) at 600 °C under the fellow of 20% C3H8/N2 (20 mL/min). c In situ Raman spectra of ceria-vanadia (6 V/30CeAl) (top) and vanadia (6 V/Al) (bottom) at 600 °C under the fellow of 20% C3H8/N2 (20 mL/min). In situ DRIFTS spectra of temperature-programmed and isothermal propane dehydrogenation over ceria-vanadia (6 V/30CeAl) (d) and vanadia (6 V/Al) (e, f). g The calculated ratios of H2/H2O during the C3H8 transient pulses at 600 oC. h Experimental relaxation curve in the form of fractional weight change as a function of time at 600 oC under the fellow of 20% C3H8/He (10 mL/min). i Schematic representation of the concerted oxygen diffusion in the ceria-vanadia redox catalyst. The black, white, and red spheres represent C, H, and O atoms.

In situ diffuse reflectance infrared Fourier transforms spectroscopy (DRIFTS) upon propane exposure identified the co-existence of dehydrogenation and cracking of propane induced by this dynamic oxygen evolution. Peaks ascribed to asymmetric and symmetric CH3 stretching modes (2970 and 2875 cm−1 ) started at 100–150 o C (Fig. 3d)30. The presence of a band centering at 1645 cm−1 (ν(CH3CH=CH2)) implied that the propyl complex was oxidatively dehydrogenated to propenyl by heterolytically subtracting H to neighboring V-O sites, leading to the occurrence of vanadium hydroxyl band (V-OH, 3660 cm−1 ) 28. However, a peak of ν(C=O) (1680 cm−1 ) attributed to acetone, the intermediate of overoxidation of propane to COx, was dominated on the pure VOx catalysts when the temperature was higher than 150 o C, along with a signifificantly negative V = O band induced by the ready reduction of vanadia (V5+ → V3+) (Fig. 3e, f and Supplementary Fig. 11)11,28. This “over quick” oxygen removal would induce the transformation of oxidative to non-oxidative dehy-drogenation and the occurrence of propane cracking. At 250–600 o C, two peaks at 1545 and 1460 cm−1 attributed to the unsaturated or aromatic species, the precursors of coke deposits27,28 that lead to fast deactivation were observed, which were also evidenced by the more dominated D1 and G band in situ Raman spectra on pure vanadia catalysts during dehydrogenation step.

As evidenced by the in situ spectroscopies, the release of lattice oxygen would induce the existence and transformation of different reaction periods, including overoxidation, oxidative dehydrogenation, and non-oxidative dehydrogenation. Under differential reactor operation by controlling the propane conversion lower than 10%, the C3H6 formation rate showed a linear relationship with C3H8 pressure, while C3H8 conversion kept identical at different C3H8 pressures, indicating the rate of propene formation is typically related to propane partial pressure, i.e., a first-order reaction with respect to propane (Supplementary Fig. 13). To clarify the contribution of oxidative and nonoxidative dehydrogenation, the formation of H2O and H2 over ceria vanadia redox catalysts in their dehydrogenation tests were investigated. As shown in Supplementary Fig. 13, the initial ratio of H2O to H2 at the 5th min was 0.44; however, it decreased to about 0.05 after 60 mins. Therefore, oxidative dehydrogenation could be more dominant in the initial period (less than 5 mins) and it was transformed to non-oxidative dehydrogenation with time, accounting for the introduction of the reoxidation step to recover the lattice oxygen after the 30-min dehydrogenation test during the continuous dehydrogenation-reoxidation cycles with the ratios of H2O to H2 of ~0.21.

Ask for more:

Email:wallence.suen@wecistanche.com

Whatsapp/Tel:+86 15292862950