Introduction

Ethylene, one of the basic building blocks to produce plastic and key chemicals, is predominantly manufactured from steam cracking of hydrocarbons thus usually concomitant with small amount of co-produced acetylene. The co-existed acetylene can severely poison the downstream catalyst for ethylene transformation, thereby has to be diminished to an acceptable level (often <5 ppm)1,2. Diverse methods have been developed to eliminate the acetylene impurity among which the electrocatalysis has been proven a green chemistry approach3,4. For example, Shi et al. reported a room-temperature electrochemical reduction strategy of acetylene over a layered double hydroxide (LDH)-derived Cu catalyst, which manifested high catalytic performance but suffers from unaddressed issues for large-scale applications, such as the low cell energy efficiency3. On the other hand, thermocatalytic semi-hydrogenation of acetylene into ethylene seems more efficient, and has been extensively applied in industry for decades. Among various catalysts explored, supported Pd catalysts have attracted most attention on account of their superior intrinsic activity. Unfortunately, the low selectivity, especially at the full conversion of acetylene, has long been a serious concern. Several strategies based on the “active site isolation” concept, such as selective poisoning/covering special Pd sites (Lindlar catalysts5) or forming Pd-M alloy/intermetallic compounds (industrially used Ag-Pd/Al2O3 catalysts)6,7,8,9,10,11 to weaken the adsorption of ethylene have been frequently used to improve the selectivity12,13,14,15,16,17,18,19,20,21, which are, nevertheless, often at the cost of activity loss due to the presence of substantial inaccessible Pd sites.

Single-atom catalysts (SACs) have attracted rapid growing interests as a new frontier in heterogeneous catalysis field22,23. In SACs, isolated metal atoms are spatially separated and uniformly distributed on the surface of the support, perfectly meeting the “active-site isolation” concept while simultaneously maximizing the metal utilization efficiency. Hence, SACs have been regarded as an ideal candidate for semi-hydrogenation of alkyne and have shown promising catalytic performance24,25,26,27,28,29. Unfortunately, SACs are generally less effective for H2 activation, giving rise to a depressed hydrogenation activity30. Moreover, to maintain the isolated dispersion and good stability of SACs, a very low metal loading is often used. This is particularly true for Pd-based SACs31. All these render the semi-hydrogenation of acetylene on Pd SACs working currently at elevated temperatures. Therefore, to meet industrial application, it is necessary to develop stable and efficient Pd SACs meanwhile lowering the working temperature.

Strong metal–support interaction (SMSI), a topic being extensively studied for more than 40 years in heterogeneous catalysis area32,33, has sparked renewed interests due to their potential in modifying catalyst performance, and especially in stabilizing catalysts34,35,36,37,38,39,40. Recently, we found that isolated Pt atoms supported on TiO2 can manifest classical SMSI34 but at a much higher reduction temperature. A feature of this finding is that the co-existed nanoparticles (NPs) can be selectively encapsulated while single atoms keep exposed through reduction at suitable temperatures. This finding might be extended to TiO2 supported other metal catalysts thus providing a new strategy to construct stable SACs. On the other hand, Photo-thermo catalysis is an emerging sub-discipline that involves the integration of thermo- and photocatalytic processes, which is distinct from the traditional thermo-catalysis because photogenerated carriers can directly transfer into the orbitals of adsorbed molecules to promote their desorption, dissociation, or activation thus trigger the chemical reaction, giving rise to a totally different reaction pathway41,42,43,44. Recent pioneering studies have demonstrated that the coupling of thermo- and photocatalytic processes overcomes the low activity in photocatalysis and high reaction barrier in thermocatalysis, thus offering a promising strategy to promote the activity and/or selectivity for various meaningful reactions, such as hydrogenation, oxidation, CO2 reduction, Fischer-Tropsch synthesis, water–gas shift reaction45,46,47,48,49,50,51,52. Despite of these great progress, whether photo-thermocatalysis is possible to boost semi-hydrogenation of acetylene still remains inconclusive. Few studies related to photocatalysis of selective hydrogenation for nitrobenzene53, benzaldehyde54, and alkynyl group55,56,57,58,59 were reported, but the related works for photo-thermocatalytic acetylene semi-hydrogenation are limited. Swearer et al. firstly used Pd NPs and aluminum nanocrystals (AlNC) to construct a heterometallic antenna-reactor complexes photocatalyst for semi-hydrogenation of acetylene but with a low product yield60. The other one is working at relatively high temperature by converting photo into heat rather than an integration of photo-thermo catalysis process at lower temperature28.

Herein, we report a simple yet general strategy to improve the selectivity of TiO2 supported Pd catalysts prepared by a variety of methods via selectively encapsulating the co-existed small amount of Pd nanoclusters/nanoparticles (NPs) due to their different SMSI occurrence temperatures. In addition, on account of the superior photocatalysis of TiO2 support, a much-improved catalytic activity was obtained by integrating photo-thermo catalysis and a dramatically decreased working temperature of as low as 70 °C was realized. Detailed studies reveal that photo-induced electrons transferred from TiO2 to the adjacent Pd atoms facilitate the activation of acetylene and thus benefit the photo-thermo catalytic semi-hydrogenation reaction.

Results

Synthesis and structural characterization of Pd/TiO2

The Pd/TiO2 catalyst was firstly synthesized by ball milling based on the so called “precursor-dilution” strategy61,62,63,64 to obtain better dispersion but we will propose later that much more practical methods such as strong electrostatic adsorption (SEA) and even impregnation methods also work well. The obtained Pd/TiO2 was reduced at 200 °C and 600 °C, denoted as Pd/TiO2-200H, and Pd/TiO2-600H, respectively. For comparison, the Pd/TiO2-600H catalyst was re-oxidized by 10 vol% O2 at 300 °C, denoted as Pd/TiO2-600H-O300. In addition, pure rutile supported catalysts were also prepared and tested in similar procedures.

The BET specific surface area was measured to be about 70 m2 g−1 by N2 physical adsorption–desorption process, and the incorporation of Pd did not change the surface area much, Supplementary Fig. 1. As shown in Supplementary Fig. 2, the X-ray diffraction (XRD) spectrum of the synthesized TiO2 support displays typical patterns of both anatase and rutile, suggesting a mixture structure. After loading of Pd, and even after reduction at different temperatures and re-oxidation, there is no obvious structure change of the TiO2 support as evidenced by the similar diffraction patterns of various catalysts to that of TiO2 support. In addition, no any diffraction pattern associated with Pd species is observed, suggesting either the Pd is highly dispersed or the Pd loading is too low to be detected. The high dispersion of Pd was further examined by aberration-corrected scanning transmission electron microscopy (AC-STEM). High-magnification high-angle annular dark-field (HAADF) STEM images reveal the presence of relatively high density of Pd single atoms on all catalysts, Fig. 1a–d. Meanwhile some other HAADF-STEM images indicate the presence of small portion of Pd NPs, Supplementary Fig. 3. It stressed the great difficulty in fabricating “absolute” Pd SACs (presence of only isolated single atoms without any clusters/NPs) even with such an effective ball-milling method61,62,63,64. To our knowledge, so far the Pd SACs with relatively high metal loading on non-carbon supports have been rarely reported65,66,67.

Fig. 1: Structural characterization of Pd/TiO2 serial catalysts.
figure 1

ad AC-HAADF-STEM images of a Pd/TiO2, b Pd/TiO2-200H, c Pd/TiO2-600H, and d Pd/TiO2-600H-O300; Pd single atoms are highlighted in yellow circles. e, f DRIFT spectra of CO adsorption on Pd/TiO2-200H, Pd/TiO2-600H, and Pd/TiO2-600H-O300 e at CO saturation adsorption and f upon He purging for 2.5 min at room temperature. g XANES spectra of different Pt/TiO2 catalysts at Pd K-edge absorption edge.

Very recently, we discovered that Pt single atoms on TiO2 can manifest classical SMSI upon reduction but at a much higher reduction temperature compared with Pt NPs. The most meaningful feature of this discovery is that the NP active sites can be selectively encapsulated upon reduction at certain temperatures, therefore the catalytic performance can be finely tuned34,35,36,37,38,39,40. We believe this scenario is general and may be extended to TiO2 supported Pd catalysts to distinctly refine their catalytic performance and we will prove this in the following.

Diffuse reflectance infrared Fourier transform (DRIFT) spectra of CO adsorption were first employed to study the SMSI state of our Pd/TiO2 sample and the results are presented in Fig. 1e, f and Supplementary Fig. 4. For CO saturation adsorption on Pd/TiO2-200H, Fig. 1e, two peaks centered at 2106 and 2091 cm−1 and a broad band existed in the range of 1840–1990 cm−1 were observed in addition to the two gas phase CO bands. The former two are ascribed to linear CO adsorption on Pd single atoms and Pd NPs1 and Eq. 2:

$${{{{{\rm{Conversion}}}}}}\,=\,\frac{{{{{{\rm{C2H2}}}}}}({{{{{\rm{feed}}}}}})\,-\,{{{{{\rm{C2H2}}}}}}}{{{{{{\rm{C2H2}}}}}}({{{{{\rm{feed}}}}}})}\,\times\, 100\%$$
(1)
$${{{{{\rm{Selectivity}}}}}}\,=\,\left(1\,-\,\frac{{{{{{\rm{C2H6}}}}}}\,-\,{{{{{\rm{C2H6}}}}}}({{{{{\rm{feed}}}}}})}{{{{{{\rm{C2H2}}}}}}({{{{{\rm{feed}}}}}})\,-\,{{{{{\rm{C2H2}}}}}}}\right)\,\times\, 100\%$$
(2)