Report

Interfacial Effects in Iron-Nickel Hydroxide–Platinum Nanoparticles Enhance Catalytic Oxidation

See allHide authors and affiliations

Science  02 May 2014:
Vol. 344, Issue 6183, pp. 495-499
DOI: 10.1126/science.1252553

Improving Reactions at Interfaces

Alloying precious metals such as platinum with more abundant transition metals, such as iron and nickel, can both improve their catalytic reactivity and lower catalyst cost. Chen et al. (p. 495) explored using coatings of iron oxide–hydroxide layers on supported platinum nanoparticles for CO oxidation. The presence of this layer allowed the reaction to run rapidly at room temperature by bringing together different reaction sites on the two metals. The addition of nickel improved catalyst lifetime, and an oxidative transformation created a more complex nanoparticle morphology that increased platinum utilization.

Abstract

Hybrid metal nanoparticles can allow separate reaction steps to occur in close proximity at different metal sites and accelerate catalysis. We synthesized iron-nickel hydroxide–platinum (transition metal-OH-Pt) nanoparticles with diameters below 5 nanometers and showed that they are highly efficient for carbon monoxide (CO) oxidation catalysis at room temperature. We characterized the composition and structure of the transition metal–OH-Pt interface and showed that Ni2+ plays a key role in stabilizing the interface against dehydration. Density functional theory and isotope-labeling experiments revealed that the OH groups at the Fe3+-OH-Pt interfaces readily react with CO adsorbed nearby to directly yield carbon dioxide (CO2) and simultaneously produce coordinatively unsaturated Fe sites for O2 activation. The oxide-supported PtFeNi nanocatalyst rapidly and fully removed CO from humid air without decay in activity for 1 month.

The fabrication of multicomponent active sites (115), particularly metal-metal hydroxide and oxide interfaces (111), to facilitate the activation of reagents has been emerging as an effective method for preparing heterogeneous catalysts with improved activities. For example, Pt/FeOx interfaces exhibit excellent performance in CO oxidation and CO preferential oxidation (PROX) (1, 2), Au/CeOx and Au/TiOx interfaces improve the activity of the water-gas shift reaction (57), and Pt/M(OH)2 (where M is metal) interfaces enhance the performance of H2 evolution reaction (10, 11). The structure and reactivity of such catalysts can be modeled with well-defined synthetic nanoparticles (NPs) (1618). For instance, rational design of nanoscaled metal–metal oxide interfaces in nanostructured catalysts has been successfully achieved by controlled assembly of metal and metal oxide nanocrystals (NCs) (8). Monodisperse Pd NCs were deposited on ceria to investigate the Pd-CeO2 interfacial effect on catalysis (19). However, assembling metal NPs with oxide NPs or depositing metal NPs on oxide supports introduced rather limited metal-metal (hydr)oxide interfaces.

We report on a wet-chemical method for fabricating M-OH-Pt interfaces (M-OH, where M is Fe and Ni) by partially covering the surface of monodisperse Pt NCs (which are <5 nm in diameter) with atomically thick M-OH layers. The Fe3+-OH-Pt interfaces increased room-temperature catalytic CO oxidation by generating coordinatively unsaturated iron sites for activating O2, and the Ni2+ incorporation dramatically enhanced the long-term catalyst stability by stabilizing the Fe3+-OH-Pt interfaces during catalysis. Based on this understanding, an alloy-assisted strategy was developed to produce a practical Pt nanocatalyst with catalytic M-OH-Pt sites over 50% of total Pt atoms in the catalyst.

To create the iron (hydr)oxide interface, monodisperse Pt NCs capped by organic amine (fig. S1) were synthesized by thermally reducing Pt(acac)2 (where acac is acetylacetonate) in the presence of CO. A small amount of Fe(acac)3 was then introduced and thermally decomposed to deposit submonolayer Fe(OH)x onto the surface of the premade Pt NCs, yielding the Pt/Fe(OH)x composite (20). Aberration-corrected scanning transmission electron microscopy (STEM) images (Fig. 1, B and C, and fig. S2) revealed the deposition of a less electron-dense layer on the Pt NCs; energy-dispersive spectroscopy (EDS) analysis showed that this layer contained iron (fig. S3). Although the surface of the Pt NCs was covered by iron species, lattice fringes with interplanar distances of 0.23 nm, corresponding to Pt(111) planes, were observed. The Fe x-ray photoelectron spectroscopy (XPS) spectra suggested an Fe3+ oxidation state (Fig. 1C and fig. S4). The O1s XPS spectra are best fitted by the presence of Fe-OH and bridging oxo groups (21) in a 4.2:1 ratio (Fig. 1D). X-ray absorption near-edge structure (XANES) analysis (fig. S5) at the pre-edge of the Fe K-edge at 7.114 keV showed that the Fe-O coordination sphere is a slightly distorted octahedron (22). Extended x-ray absorption fine structure (EXAFS) revealed that the bond distances of Fe-O and Fe-Fe in the overgrown Fe(OH)x layer were 1.97 ± 0.02 Å and 3.01 ± 0.02 Å, respectively (fig. S5 and table S1), and an Fe-O coordination of 6 ± 1. Low-energy ion scattering spectroscopy (LEISS) (fig. S6) showed that the as-obtained Pt/Fe(OH)x NPs still had open Pt sites at their edges and corners.

Fig. 1 Structural characterizations and catalytic performance of the Pt/Fe(OH)x nanoparticles.

(A) Representative high-angle annular dark-field (HAADF) STEM image of Pt/Fe(OH)x NPs. (B) High-resolution HAADF-STEM image projected along [110] axis and the corresponding bright-field image of an individual Pt/Fe(OH)x NP. (C and D) XPS core-level spectra of Fe2p and O1s, respectively. (E) CO conversions as a function of reaction temperature over TiO2-supported Pt/Fe(OH)x NPs and Pt NCs. Reaction conditions: 1% CO; 16% O2; N2 balance; SV = 400 L g−1 Pt h−1; pressure = 0.1M Pa.

Given the strong binding of CO to Pt (1), we studied CO oxidation to evaluate the Fe3+-OH-Pt interfacial effect. TiO2-supported Pt NCs displayed negligible CO oxidation activity at room temperature (Fig. 1E); 100% CO conversion was achieved only at temperatures above 373 K. However, TiO2-supported Pt/Fe(OH)x NPs (fig. S7) readily achieved 100% CO conversion at room temperature, even with a space velocity (SV) as high as ~400 L g−1 Pt h−1 (Fig. 1E). The enhanced activity of Pt/Fe(OH)x NPs was also observed in a humid stream with less oxidizing reactant (reducing the O2/CO ratio from 16 to 4) (fig. S8). The activity was lost when the Pt/Fe(OH)x catalyst was simply treated with nitric acid (fig. S9). A decline in activity from 100 to 27% in 70 min occurred when the reaction was switched from humid air to dry air (fig. S10). Even in a humid reaction stream with a relative humidity of 50%, the Pt/Fe(OH)x catalyst began to degrade after 8 hours at room temperature. We propose that the activity decrease was caused by instability of the interfacial Fe3+-OH-Pt sites with respect to dehydration. Because CO oxidation is exothermic, we subjected the Pt/Fe(OH)x catalyst to thermal treatment at 453 K in dry air for 2 hours. The OH:O ratio from the O1s XPS was reduced to only 2.7 (fig. S11), suggesting the loss of water. The Fe(OH)x submonolayers dehydrated into three-dimensional iron oxide-hydroxide or iron oxide.

To prevent the dehydration-induced loss of Fe3+-OH-Pt sites, Ni2+, which forms a stable, layered structure of Ni(OH)2 with nearly perfect octahedral coordination, was incorporated into the Fe(OH)x submonolayer. The ready formation of hydrotalcite-like Ni/Fe hydroxides (23, 24) was expected to stabilize catalytically active Fe3+-OH-Pt interfacial sites through the strong interaction between Ni(OH)x and Fe(OH)x. Moreover, Ni(OH)2 species are active in the dissociation of adsorbed water molecules and in proton transport (11, 25). The addition of a Ni(acac)2 precursor during the thermal overgrowth process deposited FeNi(OH)x hybrid submonolayers on Pt NCs with a boundary gap of 3 Å (Fig. 2A). STEM-EDS measurements showed that the Fe(OH)x and Ni(OH)x were mixed on the surface (Fig. 2B). As identified by LEISS (fig. S12), open Pt sites were present with Ni and Fe on the outermost surface of the Pt/FeNi(OH)x NPs. The XANES and EXAFS analyses of Pt/FeNi(OH)x (fig. S13) indicated that Ni(OH)2 was present in a nearly perfect octahedral coordination environment with Ni-O distances of 2.05 ± 0.01 Å (table S2), and iron was still present as Fe3+. The O1s XPS spectrum indicated an OH:O ratio of 7.5 (fig. S14A).

Fig. 2 Structural characterizations and catalytic performance of the Pt/FeNi(OH)x nanoparticles.

(A) Representative HAADF-STEM image of Pt/FeNi(OH)x NPs. (B) STEM-EDS elemental mapping of a single Pt/FeNi(OH)x NP. (C) Catalytic performances of TiO2-supported Pt/FeNi(OH)x NPs, Pt NCs, and Pt/Fe(OH)x NPs as a function of time-on-stream. Reaction conditions: 1% CO; 16% O2; N2 balance; T = 303 K; SV = 400 L g−1 Pt h−1; relative humidity = 50%; pressure = 0.1M Pa.

The Pt/FeNi(OH)x NPs exhibited greater stability in CO oxidation. When loaded onto supports (fig. S15), the Pt/FeNi(OH)x catalyst was stable in the reaction stream for more than 28 hours without any decrease in its activity at room temperature (Fig. 2C) and achieved 100% CO conversion (fig. S16). After a 2-hour heat treatment at 453 K under reaction atmosphere, the OH:O ratio in the Pt/FeNi(OH)x catalyst was maintained at 5.3 (fig. S14B). In comparison, NPs with Ni(OH)x submonolayer grown on Pt NCs, denoted as Pt/Ni(OH)x, performed poorly at room temperature relative to Pt/Fe(OH)x (figs. S17 and S18).

To understand how the Fe3+-OH-Pt interface promoted the activity of the catalysts, structural models of periodic Pt [n(111)×(111)] and [n(111)×(100)] stepped surfaces with the overgrowth of Fe(OH)x on (111) terraces (figs. S19 to S21 and table S3) were built to simulate the hydroxide overgrowth structures on Pt NCs and optimized by using Vienna Ab-initio Simulation Package (VASP) code. Density functional theory (DFT) calculations suggested that CO oxidation at the interfacial Fe3+-OH-Pt site could be triggered by the coupling reaction between the adsorbed CO and interfacial –OH moieties (Fig. 3A), leading to formation of CO2 either directly or via –COOH intermediate (figs. S22 to S24) with a low energy barrier of 0.69 eV (or 0.93 eV). On the [n(111)×(111)] stepped surface, the release of CO2 created a vacancy at the interface (i.e., Fe-Vac-Pt, where Vac is a vacancy) with an activation barrier and reaction energy of 0.16 and –0.66 eV (Fig. 3A), respectively. The proton on the reacted interfacial –OH was transferred to nearby –OH to form adsorbed H2O. On the interfacial Fe-Vac-Pt site, O2 adsorption was preferential to CO adsorption (–1.34 versus –0.87 eV) (fig. S21), suggesting that the created interfacial vacancy readily served as an effective adsorption site to convert O2 into highly reactive –OOH species (table S3). The produced –OOH species readily coupled with the second CO directly to form CO2, simultaneously recovering the Fe3+-OH-Pt interfaces and completing the catalytic cycle. Such H2O-mediated proton transfer often occurs on the surface of solid oxides (e.g., iron oxide) (26) and in our case facilitates the whole catalytic oxidation process (fig. S10).

Fig. 3 Mechanism on how the Fe3+-OH-Pt interface promotes CO oxidation.

(A) Energies of intermediates and transition states in the mechanism of CO oxidation at the Fe3+-OH-Pt[n(111)×(111)] interfaces from DFT calculation. (B and C) Time-dependent mass spectra of C16O2 and C16O18O species during the pulse experiments by pulsing CO into the Pt/Fe(OH)x catalyst with (B) and without (C) H218O pretreatment. (D and E) Time-dependent mass spectra of C16O2 and C16O18O species during the pulse experiments by reacting the Pt/Fe(OH)x catalyst with pulses containing a mixture of CO, O2, and H218O (D) or normal H2O (E) in N2.

The reported mechanisms for CO oxidation over Pt-FeOx catalysts begin with O2 activation, either at coordinatively unsaturated Fe2+ sites on Pt (1) or at an oxygen vacancy on the hematite surface (2). These coordinatively unsaturated O2 activation sites were mainly generated by thermally prereducing the Pt-FeOx catalysts in H2 and were unstable in air. However, our Fe3+-OH-Pt mechanism suggests that these Fe2+ sites are not a prerequisite for CO oxidation and that coordinatively saturated Fe3+ species are more stable under ambient conditions. We verified the above mechanism by using isotope-labeling experiments (20). In the absence of O2, 55% of the total CO2 yielded in the first five pulses was determined by mass spectroscopy to be singly 18O-labeled CO2 (C16O18O) when C16O was pulsed into the H218O pretreated Pt/Fe(OH)x catalyst (Fig. 3B). The production of C16O18O declined with the increased number of CO pulses because of the continuous consumption of the interfacial –18OH groups. In comparison, the reaction of a C16O pulse with the untreated Pt/Fe(OH)x catalyst yielded negligible C16O18O (Fig. 3C). While confirming that CO could directly react with the interfacial Fe3+-OH-Pt sites to yield CO2, this result also suggested that 18O-labeled interfacial Fe3+-18OH-Pt sites were readily obtained by treating the Pt/Fe(OH)x catalyst with H218O. Thus, it was not surprising that the reaction of pulses containing CO, O2, and H218O in N2 with the Pt/Fe(OH)x catalyst steadily yielded much more C16O18O than those pulses containing CO, O2, and non-18O–labeled H2O (Fig. 3, D and E). In all of our isotope-labeling experiments, no detectable isotope exchange between H2O and O2 was observed (fig. S25).

Our Fe3+-OH-Pt mechanism was also confirmed by kinetic studies. Over the catalysts incorporated with Fe(OH)x, the activation energies of CO oxidation were reduced to 16 kJ/mol compared with 51 kJ/mol for TiO2-supported Pt NCs (fig. S26). The incorporation of Ni2+ did not reduce the activation energy further. Over TiO2-supported Pt/Fe(OH)x, the introduction of humidity reduced the activation energy to 12 kJ/mol. Moreover, CO oxidation was determined to be close to zero order for CO and first order for O2 (fig. S27), which is expected from the Fe3+-OH-Pt mechanism. These results are consistent with CO oxidation triggered by CO oxidative coupling with interfacial OH groups at Fe3+-OH-Pt that also creates coordinatively unsaturated Fe sites for O2 activation.

In terms of Pt use, the core-shell overgrowth structure described above is not an ideal structure because most Pt atoms are not located on the surface (Fig. 4A) (1, 27). To maximize Pt use and Fe/Ni-OH-Pt interfaces as well, we developed an alloy-assisted strategy to maximize the use of Pt. In this strategy, ternary alloy PtFeNi NCs were prepared by thermally reducing metal precursors [i.e., Pt(acac)2, Fe(acac)3, and Ni(acac)2] under 1 atm CO atmosphere and then aged in the air for 3 days to yield the highly active catalyst (denoted as PtFeNi) (Fig. 4A). The mean diameter of the PtFeNi NPs was 4.9 ± 0.5 nm. When the thermal reduction time at 513 K was 30 min, STEM analyses revealed that the PtFeNi NPs were mainly core-shell structures with Pt-enriched cores (Fig. 4, B and C, and fig. S28). However, Pt-enriched cores were not present in the PtFeNi NPs when the thermal reduction time at 513 K was extended to 1 hour (fig. S29). Each individual NP apparently had a single-crystalline face-centered cubic structure. However, more careful STEM-EDS investigations revealed an interwoven, highly irregular structure to the surface of the air-aged PtFeNi NPs (Fig. 4D and fig. S30) that improved the Pt use compared with those from both Pt NCs and Pt/Fe(OH)x or Pt/FeNi(OH)x NPs with the core-shell overgrowth structure.

Fig. 4 Synthetic strategy, characterizations, and catalysis of a practical PtFeNi catalyst.

(A) Cartoon showing the structural difference between the core-shell overgrowth Pt/FeNi(OH)x catalyst and the interwoven PtFeNi catalyst, and the alloy-assisted strategy for the synthesis of the PtFeNi catalyst. (B) Representative HAADF-STEM image of PtFeNi NPs. (C and D) High-resolution HAADF-STEM image of PtFeNi NPs before (C) and after (D) aging in air, respectively. (E) STEM-EDS elemental mapping of a single PtFeNi NP.

As characterized by elemental mapping (Fig. 4E), the PtFeNi NPs exhibited compositional heterogeneities in the form of small interwoven domains (less than 3.5 Å) on their surfaces. The LEISS spectra (fig. S31) confirmed that Pt, Fe, and Ni were all exposed on the outermost surface. Both XANES and XPS spectra confirmed the presence of Fe3+ and Ni2+ oxidation states. EXAFS analyses (fig. S32) revealed that Fe and Ni in the PtFeNi catalyst were both six-coordinate with respect to oxygen, with average Fe-O and Ni-O bond distances of 1.99 ± 0.01 and 2.04 ± 0.01 Å, respectively (table S4). The O1s XPS spectrum indicated an OH:O ratio of 13 (fig. S33). These EXAFS and XPS studies demonstrated that PtFeNi NPs possessed the same interfacial sites discussed above for the Pt/FeNi(OH)x catalyst, which were active and stable for CO oxidation. As suggested by the temperature-programmed desorption measurements (fig. S34), besides the catalytically active sites that readily reacted with CO to yield CO2, no further CO chemical adsorption sites were available on the surface of the PtFeNi catalyst. CO titration was thus used to determine the percentage of catalytic sites over the total Pt sites in the catalyst. As illustrated in table S5 (20), the numbers of catalytic sites depended on the composition and also the thermal reduction conditions of the NPs. The PtFeNi catalysts with high percentages of Pt-OH-M sites (up to 54% over the total Pt atoms) were readily prepared by extending the thermal reduction time at 513 K. The high percentage of catalytic sites is also supported by the rough NP surfaces with highly dispersed Pt, and the high coordination number (up to 1.6) of O on Pt (table S4) that would allow surface Pt atoms to form more than one Pt-OH-M active center. As compared to the overgrown Pt/FeNi(OH)x catalyst with the same amount of Pt, the catalytic performance of TiO2-supported PtFeNi NPs (fig. S35) for CO oxidation was indeed enhanced by a factor of 1.4 to 1.8 (fig. S36). The PtFeNi catalyst exhibited the highest activity with a Fe/Ni molar ratio of ~0.5 (fig. S37). The PtFeNi catalyst exhibited very high stability, both in the reaction stream and during prolonged storage. No decay in the activity of the catalyst was experienced after more than 1 month in the reaction stream of CO with humid air (fig. S38).

The enhanced catalysis of the PtFeNi NPs over pure Pt NCs was also observed when γ-Al2O3 was used as the support (fig. S39). Even after being stored at room temperature in ambient air for 2 years, no deterioration of catalytic performance was found for the PtFeNi catalyst prepared by the alloy-assisted strategy. The excellent catalytic performance of the PtFeNi catalyst makes it a promising candidate for applications to remove CO in humid air or H2-rich stream and trace H2 in air (figs. S40 and S41). Creating submonolayer metal hydroxides on noble metal NCs is expected to provide an effective method to prepare efficient noble metal nanocatalysts for oxidative catalysis.

Supplementary Materials

www.sciencemag.org/content/344/6183/495/suppl/DC1

Materials and Methods

Figs. S1 to S41

Tables S1 to S5

References (2840)

References and Notes

  1. See supplementary materials on Science Online.
  2. Acknowledgments: We appreciate the financial support from the Ministry of Science and Technology of China (2011CB932403) and the National Nature Science Foundation of China (21131005, 20925103, 21373167, 21033006, and 21333008). P.Z. appreciates the funding from Natural Sciences and Engineering Research Council of Canada. We also thank L. S. Zheng, Z. Q. Tian, X. Lu, Y. Wang, and S. C. Lin for helpful discussions, and Shanghai Supercomputer Center (SSC) for providing computational resources. X-ray absorption spectroscopy studies were carried out at the beamline 17C1 of the National Synchrotron Radiation Research Center (NSRRC) in Taiwan.
View Abstract

Stay Connected to Science

Navigate This Article