Zhuoran
Gan
a,
Zheng
Lu
b,
Muntaseer
Bunian
a,
Larissa B.
Lagria
a,
Christopher L.
Marshall
b,
R. Michael
Banish
a,
Sungsik
Lee
c and
Yu
Lei
*a
aDepartment of Chemical and Materials Engineering, University of Alabama in Huntsville, Huntsville, AL 35899, USA. E-mail: yu.lei@uah.edu; Tel: +1-256-824-6527
bChemical Sciences and Engineering Division, Argonne National Laboratory, Lemont, IL 60439, USA
cX-ray Science Division, Advanced Photon Source, Argonne National Laboratory, Lemont, IL 60439, USA
First published on 1st February 2023
Pt3Zn1 and Pt1Zn1 intermetallic nanoparticles supported on SiO2 were synthesized by combining atomic layer deposition (ALD) of ZnO, incipient wetness impregnation (IWI) of Pt, and appropriate hydrogen reduction. The formation of Pt1Zn1 and Pt3Zn1 intermetallic nanoparticles was observed by both X-ray diffraction (XRD) and synchrotron X-ray absorption spectroscopy (XAS). STEM images showed that the 2–3 nm Pt-based intermetallic nanoparticles were uniformly dispersed on a SiO2 support. The relationships between Pt–Zn intermetallic phases and synthesis conditions were established. In situ XAS measurements at Pt L3 and Zn K edges during hydrogen reduction provided a detailed image of surface species evolution. Owing to a combined electronic and geometric effect, Pt1Zn1 exhibited much higher reactivity and stability than Pt3Zn1 and Pt in both the direct dehydrogenation and oxidative dehydrogenation of ethane to ethylene reactions.
Pt–Zn intermetallic nanoparticles have exhibited excellent catalytic performance in electrooxidation,9 hydrogenation,10 and dehydrogenation.11 Catalytic dehydrogenation of light alkanes, such as ethane and propane, is an on-purpose production method that exclusively yields the desired alkene of polymer-quality purity rather than a mixture of products.12 Cybulskis and coworkers have shown that the intermetallic Pt1Zn1 nanoparticles are more active, selective, and stable than the un-promoted Pt nanoparticles in the dehydrogenation of ethane.7a They ascribed the improved performance to both electronic and geometric effects. Nevertheless, how exactly these two factors affect the ethane dehydrogenation activity, selectivity, and stability were not well established with the limited examples of Pt1Zn1 and metallic Pt. Also, as their work and many recent studies on alkane dehydrogenation using PtZn nanocatalysts11,13 focused on more characterization on the Pt side, the information on the changes of Zn species was lacking, this might overlook some important aspects for the successful synthesis of active phases. The Pt1Zn1 catalyst they prepared utilized incipient wetness impregnation, a relatively rapid method widely used in catalyst synthesis. However, this method is known to produce nanoparticles with broad size distribution. A controlled synthesis method may offer a well-defined catalyst system to establish precise structure and performance relationships.
Chemical methods used to synthesize intermetallic alloy nanoparticles include impregnation, colloidal synthesis, and chemical vapor deposition. Atomic layer deposition (ALD) has emerged as an important technique for thin film deposition.14 It produces conformal coatings of materials via self-limiting chemical reactions between the precursors. By controlling ALD parameters such as dose time, the number of cycles, and deposition temperature, we can quantitatively design catalysts with the desired thickness, loading, and microstructures. ZnO ALD, as one of the well-developed ALD processes,15 can potentially aid in the design of mono-dispersed nanoscale Pt–Zn intermetallic alloys, such as Pt3Zn1, Pt1Zn1, Pt3Zn10, Pt1Zn8, etc.,16 which will provide more information regarding Pt site isolation and electronic states with the addition of Zn.
Herein, we synthesized silica-supported Pt3Zn1 and Pt1Zn1 intermetallic nanocatalysts by sequential deposition of ZnO via atomic layer deposition (ALD) and Pt via incipient wetness impregnation (IWI) on SiO2 with H2 reduction. The synthesis method was examined using standard characterization techniques as well as synchrotron X-ray absorption spectroscopy (XAS) under hydrogen reduction conditions. We revealed that ZnO reduction is a key step in the synthesis of the Pt1Zn1 phase by providing a detailed picture of Pt and Zn speciation kinetics under H2 reduction via in situ XAS. Silica-supported Pt3Zn1, Pt1Zn1, and Pt monometallic nanocatalysts were tested in both dehydrogenation of ethane (DHE) and oxidative dehydrogenation of ethane (ODHE), and the electronic and geometric effects on the catalytic performance of PtZn intermetallic nanocatalysts are discussed thoroughly. The Pt1Zn1 intermetallic nanocatalysts exhibited improved selectivity and a production rate one order of magnitude greater than those of supported Pt and Pt3Zn1 with similar Pt loading and particle sizes for the dehydrogenation of ethane. Enhanced catalyst stability was also observed.
X-ray absorption spectroscopy (XAS) measurements were conducted at 12-BM of the Advanced Photon Source (APS) at the Argonne National Laboratory. Transmission mode spectra were taken at the Pt L3 edge (∼11564 eV) and the Zn K edge (∼9659 eV). For the ex situ measurements, Pt/SiO2, Pt1Zn1/SiO2, and Pt3Zn1/SiO2 samples were first reduced in 3.5% H2 with balance He at 600 °C for 30 min to remove any surface oxides which formed due to sample transfer. The samples were then cooled and measured in helium at RT. For the in situ XAS study, the Pt/ZnO/SiO2 precatalyst was reduced by 3.5% H2/He at the designated temperature for 15 min and then measured at RT under helium protection. The XANES linear combination fittings were processed from Athena in the normalized μ(E) space, with Pt metal foil and PtO2 used as references. The fittings were done in the range of 10 eV below and 40 eV above the Pt L3 edge. EXAFS regime data fittings were performed using Artemis in the IFEFFIT software package.
The Pt–Zn intermetallic phases were generated by reducing Pt/ZnO/SiO2 in hydrogen. The optimal reduction temperature was determined by varying the temperature between 600 °C and 800 °C, while the optimal reduction time was determined by varying the length of time at which Pt/ZnO/SiO2 was reduced, from 15 min to 10 h, as shown in Fig. S1 (ESI†). The best preparation conditions to generate Pt1Zn1 and Pt3Zn1 phases were determined to be 6 h reduction at 600 °C and 8 h reduction at 800 °C, respectively. The stoichiometry of the as-prepared Pt1Zn1 and Pt3Zn1 intermetallics were confirmed using EDX (Fig. S2, ESI†) and ICP-OES. Fig. 2(a) shows that the shape of the XRD pattern of Pt3Zn1 (JCPDS no. 65-3257) resembles that of the monometallic Pt phase (JCPDS no. 65-2868) but the peaks are positioned at higher diffraction angles, as shown in the detailed scans in Fig. 2(b) and (c). The shifts in peak location in the Pt3Zn1 XRD pattern are evidence that Zn has been incorporated into Pt's face-centered cubic (FCC) structure since Zn has a smaller atomic radius.20 Based on the XRD patterns, the lattice parameters are calculated to be 3.92 Å for FCC Pt and 3.89 Å for the L12 Pt3Zn1, respectively, and 4.03 Å and 3.49 Å for the a and c unit vectors, respectively, of the AuCu L10 tetragonal structured Pt1Zn1 (JCPDS no. 06-0604). The Pt–Pt bond distance for Pt/SiO2, Pt3Zn1/SiO2, and Pt1Zn1/SiO2 nanoparticles are calculated to be 2.77 Å, 2.75 Å, and 2.85 Å (Table S1, ESI†), respectively, which are similar to literature values.7a,9c,21
Fig. 2 (a) XRD patterns of Pt/SiO2, pre-catalyst Pt/ZnO/SiO2, Pt3Zn1/SiO2 and Pt1Zn1/SiO2. Detailed scans of XRD patterns of the four samples between (b) 35–50° and (c) 75–90°. |
The formation of Pt–Zn intermetallic alloys was further investigated using XAS measurements at Pt L3 (11.564 keV) and Zn K (9.659 keV) edges. Prior to carrying out XAS measurements, the Pt-containing samples were reduced by 3.5% hydrogen at 600 °C for 30 min to remove any surface oxides that formed due to sample transfer. As compared with the Pt L3 XANES spectrum of Pt/SiO2, the whiteline intensities (2p → 5d transition) of Pt3Zn1/SiO2 and Pt1Zn1/SiO2 remain nearly the same but their edge positions shift to higher photon energies by 1.5 and 2.4 eV, respectively. These positive shifts indicate changes in hybridization and lower 5d electronic density (Fig. 3(a)).7a,21,22 At the Zn K-edge, whitelines of Pt3Zn1/SiO2 and Pt1Zn1/SiO2 do not show obvious shifts as compared with metallic Zn, but their intensities are higher (Fig. 3(c)). This feature, corresponding to the 1s → 4p transition, is sensitive to the change of coordination of Zn and bond distance due to the formation of the Pt–Zn intermetallic phase.23
The Pt L3 edge EXAFS spectrum of Pt/ZnO/SiO2 in Fig. 3(b) shows a Pt–O peak at ∼ 1.7 Å and three peaks between 2 Å and 3 Å, indicating the Pt was partially oxidized. The EXAFS spectrum of Pt1Zn1 shows a Pt–Zn single peak at ∼2.4 Å (uncorrected) while the Pt3Zn1 spectrum contains a cluster of three peaks between 2 Å and 3 Å resembling that of Pt owing to their similar crystalline structure. Data fitting in Table S2 and Fig. S3 (ESI†) shows that each Pt atom on Pt/SiO2 has an average of 10.5 nearest neighbors at a bond distance of 2.76 Å. With the incorporation of Zn, the number of nearest Pt neighbors surrounding one Pt atom in Pt3Zn1/SiO2 decreases to 6.1 and the Pt–Pt bond distance is shortened to 2.73 Å. In Pt1Zn1 nanoparticles, the Pt–Pt bond distance is elongated to 2.85 Å and each Pt atom has an average of 3.1 nearest Pt neighbors, indicating that these Pt atoms were structurally more isolated. The bond distances determined by EXAFS data fittings are comparable to those obtained from XRD patterns. In the Zn K edge EXAFS spectra (Fig. 3(d)) and data fitting (Table S3 and Fig. S4, ESI†), ZnO and Zn have single peaks at ∼1.5 Å (Zn–O) and 2.3 Å (Zn–Zn), respectively. In the bulk structure of metallic Zn, each Zn atom has 6 Zn neighbors, with a bond distance of 2.64 Å between atoms. The Zn–Pt bond distances in Pt3Zn1 and Pt1Zn1 are 2.73 Å and 2.65 Å, respectively, consistent with the bond distances derived from Pt L3 EXAFS data fitting (Table S2, ESI†) and XRD patterns (Table S1, ESI†).
Fig. 4 shows the STEM images and particle size distribution histograms of Pt/SiO2, Pt3Zn1/SiO2, and Pt1Zn1/SiO2. The nanoparticles were well dispersed on the SiO2 support. The particle size distribution, calculated for more than 500 nanoparticles, demonstrates that the average diameter of Pt/SiO2, Pt1Zn1/SiO2, and Pt3Zn1/SiO2 nanoparticles are all about 2.2 nm. In the EDX mapping, all Pt and Zn co-locate at the same spots, suggesting a homogeneous formation of Pt–Zn alloys.
Fig. 4 (a) and (b) HAADF-STEM images and particle size distribution histograms of (a) Pt/SiO2, (b) Pt3Zn1/SiO2, (c) Pt1Zn1/SiO2, and (d)–(g) EDX mapping of Pt1Zn1/SiO2. |
In situ XAS spectra at Pt L3 and Zn K edge were measured during hydrogen reduction to determine the surface species evolution. The measurements were obtained by quickly switching between Pt L3 and Zn K edges (<1 min), and thus the data on both edges were taken almost simultaneously. Fig. 5 shows the XANES and EXAFS spectra at Pt L3 and Zn K edges during reduction of Pt/ZnO/SiO2 from 25 °C to 800 °C. The in situ XAS spectra clearly show the transition from the initial oxidation state to the final intermetallic phase. In Fig. 5(a) and (b), the Pt L3 edge whiteline gradually shifts to a higher energy, accompanied by the shift of Zn K edge to a lower energy. In the EXAFS spectra in Fig. 5(c) and (d), Pt–Zn bonds arise with the elimination of Pt–O and Zn–O bonds. The quantitative changes to an atom's nearest neighbors and the associated bond distance with respect to Pt and Zn are summarized in Tables S4 and S5 (ESI†), and the fitting quality is shown in Fig. S5–S8 (ESI†).
Fig. 5 Pt L3 edge (a) XANES, (b) EXAFS Fourier transform of Pt/ZnO/SiO2 during in situ hydrogen reduction at elevated temperatures and their Zn K edge XANES (e) and (d) EXAFS counterparts. |
XANES linear combination fittings (LCF, Fig. 6) are consistent with the EXAFS fittings summarized in Tables S4 and S5 (ESI†). The as-prepared Pt/ZnO/SiO2 sample was initially partially oxidized, containing 69% metallic Pt and 31% PtO2, while all the zinc was in the form of ZnO. PtO2 was completely reduced to the metallic state at 200 °C, coinciding with the formation of the Pt–Zn phase and conforming to previous studies on hydrogen reduction of Pt/ZnO/Al2O3 and Pd/ZnO.24 The hydrogen reduction was likely catalyzed by platinum, which was able to reduce Zn at such a low temperature. Upon increasing temperature from 200 °C to 600 °C, more Pt–Zn phase formed as more ZnO was reduced. The formation of Pt1Zn1 seems to be limited by the catalytic reduction rate of ZnO instead of Zn atom diffusion into Pt. At 600 °C, all platinum existed in the form of Pt1Zn1 phase while excess ZnO remained. The excess ZnO could be removed by long reduction or vaporization at higher temperature, i.e., 800 °C. Note that it was still in the Pt1Zn1 phase as it took 8 hours at 800 °C to form the Pt3Zn1 phase (Fig. S1b, ESI†).
Fig. 6 (a) Scheme of surface species evolution, and XANES linear combination fitting (LCF) of (b) Pt L3 and (c) Zn K edges. |
There are three main concerns regarding the reduction of the Pt/ZnO/SiO2 sample to Pt3Zn1/SiO2 at 800 °C in hydrogen. The first concern is the possible collapse of the porous SiO2 support, which will cause a close encapsulation of platinum and prevent reactants from accessing the catalyst's active sites. The second concern is the occurrence of sintering of platinum nanoparticles due to diffusion-coalescence at high temperatures and a hydrogen environment, again leading to the loss of active sites for catalytic reactions. The last concern is the potential formation of the Pt–Si phase induced by platinum-catalyzed SiO2 reduction. The latter two concerns were observed in an in situ TEM study of Pt/SiO2 reduction at 800 °C in hydrogen.25 The diameter of Pt nanoparticles increased from 2 nm to 4.8 nm, which is associated with the formation of Pt3Si after being reduced in hydrogen at 800 °C. In our study, these three concerns were ruled out as evidenced by combining BET surface area analysis, XRD patterns, and EXAFS spectra. As shown in Fig. S9 and Table S6 (ESI†), the surface area, pore diameter, and pore volume are consistent between Pt/ZnO/SiO2, Pt1Zn1/SiO2, and Pt3Zn1/SiO2, indicating that the structural integrity of SiO2 support was maintained. Neither any Pt–Si phase was observed in XRD patterns, nor was Pt–Si present in the EXAFS spectra. As the SiO2 support was uniformly covered by ZnO deposited by ALD, it may have prevented the contact between Pt and SiO2 and the possible subsequent formation of Pt3Si.
The dehydrogenation of ethane reaction at 550 °C with a WHSV of 7.2 h−1 was used to evaluate the catalytic performance of the Pt–Zn intermetallic nanoparticles. Catalytic performance results are shown in Fig. 7. Based on the Pt dispersion (Table S6, ESI†), the initial ethylene turnover frequency (TOF), normalized per surface Pt was 0.56 s−1 for Pt1Zn1/SiO2, much higher than the Pt-rich Pt3Zn1/SiO2 (0.48 s−1) and the monometallic Pt/SiO2 (0.04 s−1). Although all three catalysts had close to 100% C2H4 selectivity, the Pt1Zn1 catalyst exhibited significantly improved stability. The Pt1Zn1 catalyst maintained at a higher conversion (∼24.5%) during the reaction period as compared to the Pt3Zn1 whose conversion rapidly decreased from 7% to 1.7%.
As shown in Fig. 8, the Pt1Zn1 catalysts also exhibited the best performance in terms of C2H4 TOFs, conversion, and selectivity in the ODHE. The time-on-stream catalytic tests demonstrate that Pt/SiO2 has an initial C2H4 TOF of 0.04 s−1 and a C2H4 selectivity of 70.2%, and the selectivity decays rather quickly during the first hour, with 39.3% left after 5 h. The Pt-rich Pt3Zn1/SiO2 exhibited better initial C2H4 TOF (0.66 s−1) and selectivity (77%) as compared to Pt/SiO2, but after 2 h the selectivity drops steeply and almost follows the decay path of Pt/SiO2. Pt1Zn1/SiO2 shows not only improved initial C2H4 TOF (2.79 s−1, 70 times higher than Pt/SiO2) and C2H4 selectivity (89.4%), but also more stable. Its selectivity did not show a drastic decrease and remained at 67.2% after 5 h. It is shown in Fig. 8d that the main by-products for ODHE are CO, CO2, and CH4, the CO plus CO2 initial selectivity for Pt/SiO2 is the highest, revealing that Pt monometallic is favorable for the combustion reaction path.
The improved performance may be caused by the combined electronic and geometry effects due to the presence of Zn in Pt.7a,24a Cybulskis observed that Pt1Zn1/SiO2 was more active and stable than Pt/SiO2 in dehydrogenation of ethane. By using density functional theory (DFT) calculations and synchrotron resonant inelastic X-ray spectroscopy (RIXS), they attributed the improvement to the electronic effect where the structure of Pt 5d valence orbitals changed due to the presence of Zn. Our studies by including the performance of an additional Pt3Zn1/SiO2 intermetallic catalysts support their conclusion. The improved performance of Pt1Zn1/SiO2 is more likely due to a combined electronic and geometric effect. In other words, the electronic states of Pt 5d are changed by adding the Zn element. As shown by XANES in Fig. 3(a), the Pt L3 edge position of Pt1Zn1 and Pt3Zn1 shifted to higher energy by 2.4 eV and 1.5 eV, respectively, compared with Pt monometallic. Those shifts in edge energy reflect an upward shift of unfilled Pt 5d bands, which also cause a downward shift of filled 5d bands. These energy changes are related to the bonding of Pt and the adsorbate, which in turn affects the catalytic turnover on the Pt active site. These results are consistent with the initial performance of the catalyst in the order of Pt1Zn1 > Pt3Zn1 > Pt/SiO2 in both DHE and ODHE reactions. Regarding the geometric effect, the Pt atoms are atomically more isolated in Pt1Zn1 (CNPt–Pt = 3.1, RPt–Pt = 2.85 Å) than in either Pt3Zn1 (CNPt–Pt = 6.1, RPt–Pt = 2.73 Å) or Pt (CNPt–Pt = 10.5, RPt–Pt = 2.76 Å). The C2H4 TOF for both DHE and ODHE are compared with those state-of-the-art Pt-based catalysts from the literature (Tables S7 and S8, ESI†), Pt1Zn1/SiO2 shows comparable or enhanced C2H4 TOF with other Pt–Zn or Pt–Sn nanocatalysts, considering our reaction temperature and space velocity. The more isolated Pt in Pt1Zn1 may be responsible for the improved catalyst stability as it slows the coke deposition. The use of intermetallic alloys could be a promising approach to synthesize well-defined, atomically dispersed catalysts.
Pt1Zn1/SiO2 exhibited an ethylene TOF much higher than those of Pt3Zn1/SiO2 and Pt/SiO2 in direct dehydrogenation and oxidative hydrogenation of ethane to ethylene. The Pt1Zn1/SiO2 catalysts also exhibited considerably enhanced stability. The presence of Zn in Pt–Zn intermetallic alloys leads to both electronic structure and geometric structure changes of Pt. The electronic structure of Pt 5d orbitals, which concerns and influences the nanoparticle's catalytic properties, is similar in both Pt1Zn1 and Pt3Zn1, as determined by the Pt L3 XAS spectra. This may explain the initial high TOF of both catalysts as compared to the Pt catalyst. The improved stability of Pt1Zn1 is likely associated with the geometric effect, as the Pt atoms in Pt1Zn1 are significantly more atomically isolated than the Pt atoms in Pt3Zn1 and Pt.
Footnote |
† Electronic supplementary information (ESI) available: XRD patterns, EDX spectrum, EXAFS fitting, and N2 absorption–desorption isotherms. See DOI: https://doi.org/10.1039/d2cp04173a |
This journal is © the Owner Societies 2023 |