Research ArticleChemistry

A sulfur-tethering synthesis strategy toward high-loading atomically dispersed noble metal catalysts

See allHide authors and affiliations

Science Advances  25 Oct 2019:
Vol. 5, no. 10, eaax6322
DOI: 10.1126/sciadv.aax6322


Metals often exhibit robust catalytic activity and specific selectivity when downsized into subnanoscale clusters and even atomic dispersion owing to the high atom utilization and unique electronic properties. However, loading of atomically dispersed metal on solid supports with high metal contents for practical catalytic applications remains a synthetic bottleneck. Here, we report the use of mesoporous sulfur-doped carbons as supports to achieve high-loading atomically dispersed noble metal catalysts. The high sulfur content and large surface area endow the supports with high-density anchor sites for fixing metal atoms via the strong chemical metal-sulfur interactions. By the sulfur-tethering strategy, we synthesize atomically dispersed Ru, Rh, Pd, Ir, and Pt catalysts with high metal loading up to 10 wt %. The prepared Pt and Ir catalysts show 30- and 20-fold higher activity than the commercial Pt/C and Ir/C catalysts for catalyzing formic acid oxidation and quinoline hydrogenation, respectively.


Maximizing the atom utilization efficiency of noble metal–based heterogeneous catalysts by downsizing them into nanoscale and even atomic scale is of great significance for catalysis, as only the exposed metal atoms are involved directly in the catalytic process. The limitation of increasing the metal utilization efficiency is the atomic dispersion of metals on porous solid supports, which presents a hot research topic in the past few years (15). A wide variety of atomically dispersed metal catalysts with unprecedented catalytic activities have been reported recently for catalyzing CO oxidation (6), C─H activation (7), hydrogenation (5, 8), methane oxidation (9), electrocatalytic hydrogen evolution (10), and oxygen reduction (11). On the other hand, the electronic properties of the atomically dispersed metals are quite different from their bulk counterparts, endowing the metals with unexpected catalytic selectivity (12) and steady poison-resistant properties (13) in various reactions. In addition, the atomically dispersed catalysts bridge the gap between heterogeneous and homogeneous catalysis and offer a fundamental platform for understanding the catalytic mechanism of heterogeneous catalysts at the molecular level (5, 14).

The synthetic challenge of atomically dispersed catalysts lies in the difficult-to-stabilize ligand-free separated atoms on solid supports, since metal atoms tend to aggregate into larger particles owing to the sharply increased surface free energy with the decrease of particle size. Some emerging synthetic strategies have been proposed to obtain such a distinctive class of catalysts, such as a mass-selected soft-landing technique (15), a precursor-preselected method (16), a photochemical strategy (5), an iced photochemical reduction (17), an atomic layer deposition method (10), and a thermal transformation of metal nanoparticles (4, 18). These methods, however, still face a few challenging problems for practical applications, such as the complicated equipment, high-cost metal precursors used for synthesis, and very low metal loading. Another efficient strategy toward atomically dispersed catalysts is to use high–surface area supports that can supply numerous anchor sites for fixing large-content metal atoms, typically including metal oxide nanoparticles (4, 5), zeolite (19), mesoporous graphitic carbon nitride (20), covalent triazine frameworks (21), and surface-functionalized porous carbons (22). In particular, Zeng and co-workers achieved a very high loading of 7.5 wt % for atomically dispersed Pt supported on monolayered MoS2 (23); Kim and co-workers reported a 5 wt % atomically dispersed Pt on a microporous sulfur-doped carbon (micro_S-C) support that was prepared by templating zeolite and using toxic H2S gas as sulfur source (12). Although these great progresses have been made, fabricating stable atomically dispersed catalysts with metal loading higher than 2.0 wt% through a general approach remains a formidable challenge.

In this work, we demonstrate a general and scalable synthesis approach to a family of atomically dispersed metal catalysts with high metal loading of up to 10 wt %, including Ru, Rh, Pd, Ir, and Pt, by using mesoporous sulfur-doped carbons (meso_S-C) as supports (Fig. 1). The meso_S-C supports with high sulfur content (>10 wt %) and large surface area (>1000 m2 g−1) can effectively prevent metal atoms from aggregating when removing the ligands from metal precursors by the thermal reduction process at 200° to 500°C owing to the strong chemical metal-sulfur interactions. The prepared atomically dispersed catalysts display much enhanced catalytic performance for formic acid oxidation reaction (FAOR) and quinoline hydrogenation compared to commercial catalysts.

Fig. 1 Schematic illustration of the preparation and model structure of the atomically dispersed noble metal catalysts.

Five atomically dispersed noble metal (Ru, Rh, Pd, Ir, and Pt) catalysts were prepared on the meso_S-C supports with high metal loading of up to 10 wt %.


Synthesis and characterization of the catalysts

The meso_S-C supports were prepared by a cobalt-assisted carbonization of molecular precursors with SiO2 nanoparticles as hard templates at 800°C, followed by NaOH and H2SO4 etching steps to remove the SiO2 template and cobalt, respectively (see the details in Materials and Methods). The as-prepared meso_S-C support had a large Brunauer-Emmett-Teller (BET) surface area of >1000 m2 g−1 and a pore volume of >2.0 cm3 g−1, as well as a high sulfur content of >10 wt % (see table S1). Inductively coupled plasma atomic emission spectrometry measurement showed a negligible cobalt species (lower than 0.2 wt %) left in the meso_S-C supports.

To demonstrate the capability of meso_S-C for supporting atomically dispersed metals, we first synthesized the xPt/meso_S-C catalysts (x represents the percentage content of Pt) via the wet impregnation of a H2PtCl6 precursor and thermal reduction by 5 vol % H2/Ar at 250°C for 2 hours (Fig. 1). The reduction ability of H2 for H2PtCl6 on meso_S-C at 250°C was confirmed by H2 temperature-programmed reduction (H2-TPR) analysis (fig. S1A). After H2 reduction, no Cl signal could be detected in the Pt/meso_S-C catalyst by x-ray photoelectron spectroscopy (XPS) measurement (fig. S1F), suggesting the complete removal of the Cl ligand.

Atomic-resolution high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) was then performed to analyze the morphologies of Pt species at the atomic scale for the catalysts with gradually increased Pt loading from 5 to 30 wt %. For the catalysts with Pt loading of less than 20 wt %, we observed numerous separated Pt atoms along with some loose Pt aggregations consisting of a few randomly dispersed Pt atoms (Fig. 2, A to C, and fig. S2, A to F). No Pt crystalline lattice or larger particles could be found throughout the meso_S-C support for these catalysts. With the increase of Pt loading to 30 wt %, clear Pt crystal lattice appeared in the HAADF-STEM image of the 30Pt/meso_S-C catalyst (Fig. 2D and fig. S2, G and H). X-ray diffraction (XRD) pattern analyses verified the emergence of a crystalline Pt phase for 30Pt/meso_S-C (Fig. 2E). XPS was carried out to study the electronic state change of the Pt species during the catalysis synthesis (fig. S1K). The binding energy (BE) of Pt negatively shifted from 74.1 eV for H2PtCl6 to 72.4 eV for H2PtCl6/meso_S-C, indicating that the Pt (IV) component in H2PtCl6 was already reduced to Pt (II) during the wet impregnation and drying process. A similar phenomenon was also reported on other carbon supports (24). After H2 reduction, the BE of Pt species in the 5Pt/meso_S-C catalyst exhibited a further negative shift of 0.2 eV but was still higher than that of bulk Pt (BE = ~71.0 eV).

Fig. 2 Characterizations of Pt/meso_S-C.

(A to D) HAADF-STEM images of (A) 5Pt/meso_S-C, (B) 10Pt/meso_S-C, (C) 20Pt/meso_S-C, and (D) 30Pt/meso_S-C. (E) XRD patterns of the Pt/meso_S-C catalysts. The standard peaks of Pt (JCPDS no. 04-0802) were shown in red lines. a.u., arbitrary units. (F) Normalized XANES spectra of 10Pt/meso_S-C, 30Pt/meso_S-C, H2PtCl6/meso_S-C, and Pt foil at the Pt L3-edge. (G) EXAFS spectra of 10Pt/meso_S-C, 30Pt/meso_S-C, Pt foil, and PtO2.

X-ray absorption near-edge structure (XANES) and extended x-ray absorption fine structure (EXAFS) were measured to further characterize the electronic and coordination structures of Pt species in the Pt/meso_S-C catalysts. The intensity of the white line of Pt species in Pt/meso_S-C was much lower than that of H2PtCl6 precursor but higher than that of metallic Pt foil (Fig. 2F and fig. S2I), indicating that the Pt species in Pt/meso_S-C were partially oxidized, which was well consistent with the XPS results. The Fourier-transformed k2-weighted EXAFS spectra of the Pt/meso_S-C catalysts are presented in Fig. 2G and fig. S2J. The Pt/meso_S-C catalysts with the Pt loading of 5 to 20 wt % presented a prominent peak at ca. 1.9 Å in EXAFS, which could be ascribed to the Pt-S coordination (note that Pt-O coordination was located at a much lower R value in the PtO2 sample) (12, 22, 25), and no peak of Pt-Pt coordination could be observed at the position of around 2.5 to 2.8 Å. Differently, Pt-Pt coordination appeared in addition to the Pt-S coordination when the Pt loading increased to 30 wt %.

Fitting of EXAFS was then carried out to obtain the quantitative chemical configurations of Pt species, and the results are summarized in Table 1. The coordination number (CN) of Pt-S in 5Pt/meso_S-C was equal to that in 10Pt/meso_S-C (ca. 3.6), and no Pt-Pt contribution was observed, suggesting that the Pt species in these two catalysts were in the same isolated environment. Considering that the sulfur species are mainly in the form of C-S-C as revealed by high-resolution XPS (fig. S2K), we therefore presume that, in these atomically dispersed catalysts, four thiophenic S moieties coordinated with one Pt atom to form the Pt-S4 mononuclear complex-like structure (fig. S2L), which typically endowed the heterogeneous Pt species with homogeneous catalyst properties (12). Similar single-atom metal-N4 structures were frequently reported in the metal-N-C catalyst system (26, 27). When the Pt loading reached 20 wt %, the component of Pt-S decreased with the average CN down to 2.8 and the Pt-Pt contribution emerged. With further increase of Pt loading to 30 wt %, we found much lower Pt-S CN (ca. 2.2) and higher Pt-Pt CN (ca. 4.4). Note that the Pt-Pt distance (ca. 2.717 Å) in 20Pt/meso_S-C was much shorter than that in Pt foil (ca. 2.763 Å). The contraction of the Pt-Pt bond might come from the decreasing electron density in the disordered ultrasmall nanocluster (25). Increased Pt-Pt distance (ca. 2.734 Å) was measured in 30Pt/meso_S-C catalyst compared to 20Pt/meso_S-C, suggesting that the Pt atoms were aggregating from disorder to order phase with the increase of Pt loading.

Table 1 XAFS fitting parameters of Pt species in Pt/meso_S-C and reference materials.

N, coordination number for the absorber-backscatterer pair; R, the average absorber-backscatterer distance; σ2, the Debye-Waller factor; ΔE0 is the inner potential correction; r factor shows the goodness of the fit. The k range for Fourier transformation was from 2.5 to 12.4 Å−1 and the fitted R range was from 1.1 to 3.2 Å.

View this table:

Overall, the absence of a Pt-Pt contribution in 5Pt/meso_S-C and 10Pt/meso_S-C catalysts revealed by EXAFS analyses definitely proved the atomic dispersion of Pt on the meso_S-C supports with high Pt loading of up to 10%, which exceeds the loading of recently reported ligand-free noble metal catalysts with atomic dispersion (12, 23). The randomly packed Pt atom aggregations observed by HAADF-STEM in 5Pt/meso_S-C and 10 Pt/meso_S-C should come from the overlapping of Pt atoms in 3D meso_S-C structures or the loose packing with a monolayer of Pt atoms on carbon surface. Moreover, the strong chemical interaction between metal and the meso_S-C supports was also strongly verified by the EXAFS analyses.

To identify the pivotal structural parameters of meso_S-C for the synthesis of high-loading atomically dispersed metal catalysts, we carried out several control syntheses. First, meso_S-C with low sulfur content (LS-meso_S-C) was prepared by heat treatment of pristine meso_S-C under flowing 5 vol % H2/Ar at 900°C for 2 hours and also used as support for loading Pt by the same process. The sulfur content markedly decreased from 14.3 to 1.2 wt % upon the second heat treatment in the reducing atmosphere, while the BET surface area remained nearly unchanged (see table S1). Many crystalline Pt particles were clearly present in HAADF-STEM images of the LS-meso_S-C–supported 5 wt % Pt (fig. S3A). In addition, we also synthesized a microporous S-doped carbon (micro_S-C) support by carbonizing a thiophene-containing conjugated microporous polymer to study the influence of a porous structure for loading Pt atoms. The sulfur content (13.8 wt %) of the micro_S-C support was almost the same as that of meso_S-C, while the BET surface area (666 m2 g−1) and pore volume (0.39 cm3 g−1) were much less than those of meso_S-C. HAADF-STEM observations revealed that individual Pt atoms, clusters, and particles bigger than 2.5 nm coexisted in the micro_S-C–supported 5 wt % Pt (fig. S3B). Similarly, the maximum loading for atomically dispersed Pt on zeolite-templated micro_S-C that has a very high BET surface area (2400 to 2800 m2 g−1) and a high sulfur content (17 wt % S) was also limited to 5 wt % (12). We therefore conclude from these results that the high sulfur content, large surface area, and mesoporous structure would enable the meso_S-C supports with abundant and easily accessible sulfur-containing sites for strongly anchoring Pt atoms with high mass percentage.

For further comparison, several commercially available and home-made carbons were also examined as supports for loading Pt, including Vulcan XC-72R (XC-72R), carbon nanotubes (CNTs), reduced graphene oxide (RGO), mesoporous nitrogen-doped carbon (meso_N-C), and mesoporous oxygen-doped carbon (meso_O-C). For all these carbon supports, we failed to prepare atomically dispersed Pt catalysts for the loading of 5 wt %, as indicated by the HAADF-STEM results (fig. S3, C to H). Note that although meso_N-C and meso_O-C also had a mesoporous structure and a large BET surface area (table S1), they could not effectively suppress the Pt atom agglomeration when removing the ligands by thermal reduction with H2. The heteroatoms in meso_N-C (4.7 wt % N) and meso_O-C (23.4 wt % O) and some defects in the commercial carbon black may also provide anchor sites for binding Pt atoms (2830), but these kinds of interactions seem to be much weaker than those in meso_S-C. With the strong affinity of sulfur to noble metals as a result of the soft acid–soft base interaction (25, 31), aggregation of Pt atoms could be notably suppressed on the meso_S-C support.

To extend the applicable scope of our synthetic strategy, we additionally prepared another four atomically dispersed noble metal catalysts, including Ir, Pd, Ru, and Rh. H2-TPR profiles confirmed that the metal precursors of IrCl3, PdCl2, RuCl3, and RhCl3 could be reduced by H2 at 250°C (fig. S1, B to E). XPS results indicated that the Cl ligand could be thoroughly removed from Pd, Ru, and Rh precursors on meso_S-C under 5 vol % H2/Ar at 250°C (figs. S1, H to J). In the case of Ir/meso_S-C, the thermal reduction in pure Ar at 500°C was proven to be more efficient to remove the Cl ligands without Ir atom aggregation than the H2 reduction at 250°C (fig. S1G). The BE of all the metals showed a negative shift compared to corresponding precursors, indicating a decreased oxidation state after thermal reduction at an evaluated temperature (figs. S1, L to O). We did not find obvious Ir atom agglomeration by HAADF-STEM for the Ir/meso_S-C catalysts when the Ir loading was below 10 wt % (Fig. 3A and fig. S4A). EXAFS analyses confirmed the exclusive presence of an Ir-S bond with CN of ca. 3.8 for both 5Ir/meso_S-C and 10Ir/meso_S-C samples (Fig. 3E, fig. S4C, and table S2). When the Ir loading was further increased to 15 wt %, a large number of Ir particles appeared in the HAADF-STEM image, and an Ir-Ir bond was observed in the EXAFS spectra (fig. S4, B and C). By similar loading-dependent synthesis experiments, we identified that the maximum metal loading for atomically dispersed Pd, Rh, and Ru on the meso_S-C supports was 5, 5, and 3 wt %, respectively (Fig. 3, B to D and F to H, and fig. S5).

Fig. 3 Characterization of atomically dispersed noble metal catalysts.

(A to D) HAADF-STEM images of (A) 5Ir/meso_S-C, (B) 5Rh/meso_S-C, (C) 5Pd/meso_S-C, and (D) 3Ru/meso_S-C. (E to H) Corresponding EXAFS spectra of (E) 5Ir/meso_S-C, (F) 5Rh/meso_S-C, (G) 5Pd/meso_S-C, and (H) 3Ru/meso_S-C at R space. EXAFS spectra of metal oxide and foil were also shown for comparison.

Catalytic performance for FAOR

We first studied the catalytic performance of atomically Pt/meso_S-C catalysts for FAOR, which were tested in an aqueous solution of 0.5 M H2SO4 and 0.5 M HCOOH under room temperature by the rotating disk electrode technique. A commercial Pt/C catalyst (20 wt %; Johnson Matthey) was also tested under the same conditions. We observed two obvious peaks in the anodic sweep for the FAOR on Pt/C (Fig. 4A). The first peak at ~0.6 V comes from the direct oxidation route that involves a dehydrogenation process, while the secondary peak at ~1.0 V arises from the electrochemical oxidation of adsorbed CO molecules (COads) through the indirect pathway (32, 33). The FAOR on Pt/C mainly goes through the indirect route, in which the Pt surface is easily poisoned by the reaction intermediate COads. Differently, the FAOR anodic sweep voltammograms of the atomically Pt/meso_S-C catalysts show a prominent peak at ~0.85 V, indicating that the oxidation of formic acid is dominant in a direct pathway owing to the isolated Pt sites in 10Pt/meso_S-C (33). An additional peak at ~1.0 V gradually rose with the increase of the Pt loading on meso_S-C from 10 to 30 wt %, which is probably associated with the emergence of Pt clusters in 20Pt/meso_S-C and 30Pt/meso_S-C.

Fig. 4 Catalytic performance of atomically dispersed noble metal catalysts.

(A) FAOR anodic sweep voltammograms of the atomically Pt/meso_S-C and reference catalysts, as normalized by the Pt mass. The inset shows the magnified curve of Pt/C. Test conditions: sweep rate, 10 mV s−1; 0.5 M H2SO4; 0.5 M HCOOH; and Pt loading, 10.2 μg cm−2 for all samples. (B) Comparison of the mass activity of the catalysts for FAOR at 0.6 V. (C) Catalytic performance of quinoline hydrogenation over 5Ir/meso_S-C and reference catalysts. (D) Recycle of 10Ir/meso_S-C for catalytic hydrogenation of quinoline.

We then compared the catalytic activity of the catalysts by normalizing the current at ~0.6 V versus reversible hydrogen electrode. As shown in Fig. 4B, the atomically dispersed 10Pt/meso_S-C catalyst exhibited a high FAOR activity of 2.38 A mgPt1, which is 31 times higher than that of Pt/C (0.074A mgPt1). Although the activity of 20Pt/meso_S-C and 30Pt/meso_S-C decreased with the increase of Pt loading, the activity of these two catalysts was still much higher than that of commercial Pt/C because of the high dispersion of Pt atoms on meso_S-C. We attribute the much enhanced FAOR activity on Pt/meso_S-C to the improved intrinsic activity induced by the isolated single-atom Pt sites and high Pt atom utilization (33). The stability of the Pt/meso_S-C and Pt/C catalysts was then investigated by chronoamperometry (CA) at 0.4 V (fig. S6). The current decay was observed with the testing time for all catalysts as reported in the literatures (32, 34), but the stable current for the FAOR on 10Pt/meso_S-C was still significantly higher than that on Pt/C after 7200 s.

Catalytic hydrogenation of quinoline

We then studied the partial hydrogenation of quinoline on the prepared 5Ir/meso_S-C catalyst to further demonstrate the utility of this class of atomically dispersed catalysts. The tetrahydroquinoline ring system plays a significant role in pharmaceutical synthesis and has been studied extensively in recent years (35). Ir complexes with various ligands were proved to be effective for quinoline hydrogenation (36), yet the homogeneous catalysis processes always require extremely complicated separation processes and sometimes need harsh added I2. With 0.1 mol% of 5Ir/meso_S-C catalyst, full conversion of the substrate was achieved in ethanol at 100°C under 2 MPa of hydrogen pressure for 2 hours. No by-products, such as 5,6,7,8-tetrahydroquinoline and decahydroquinoline, were detected. For comparison, Ir/XC-72R, Ir/CNT, and Ir/RGO catalysts with 5 wt % Ir loading were also prepared on the corresponding supports and tested for catalyzing quinolone hydrogenation under the same conditions. Less than 20% conversions of quinolone were obtained for all these reference materials (Fig. 4C). To further evaluate the activity of the atomically dispersed Ir catalysts, we calculated turnover frequencies (TOFs) under the quinoline conversion in 15 to 25% of all cases to ensure that the reaction remained under kinetic control. TOF for quinoline hydrogenation over 5Ir/meso_S-C and 10Ir/meso_S-C was as high as 1292 and 1149 hour−1, respectively, giving an around 20-fold enhancement compared to the commercial Ir/C catalyst (57 hour−1). The catalytic performance of the atomically dispersed Ir/meso_S-C catalysts was better than the single Ru site catalyst (37), Cu@Ru nanocrystals (38), Rh particles (39), and even the homogeneous iridium complexes catalyst (36), revealing the superior advantage of the atomically dispersed metal catalyst. When single Ir atoms and Ir clusters coexisted in 15Ir/meso_S-C (see fig. S4B), the TOF value sharply decreased to 604 hour−1 under the same condition. Besides the high Ir atom utilization, the electronic interaction between Ir and S may also positively affect the quinoline hydrogenation. XANES spectra revealed that the Ir in 5Ir/meso_S-C and 10Ir/meso_S-C had a higher oxidation state than bulk Ir (fig. S4D), implying an electron transfer process between Ir and meso_S-C. The electron-deficient Ir sites may tend to the preferential adsorption of N-containing rings, leading to the regioselective hydrogenation of the heterocyclic ring (8). Moreover, H2 might undergo a heterolytic dissociation route during the H2 activation process owing to the non-availability of M-M pairs in the atomically dispersed metal catalysts for homolytic dissociation of H2 (5). Then, the Hδ+/Hδ− pair preferentially transfers to the unsaturated bond, which in turn enhances the catalytic performance (40).

In addition, the 10Ir/meso_S-C catalyst sustained its initial activity and selectivity after five recycling tests (Fig. 4D). No agglomeration of Ir atoms was observed by HAADF-STEM after recycling (fig. S7), indicating the high immobility of Ir atoms on meso_S-C when catalyzing the quinoline hydrogenation. We also explored the catalytic hydrogenation reaction on other quinoline derivatives with different functional groups over the 5Ir/meso_S-C catalyst. As listed in Table 2, the hydrogenation of quinolines exhibited nearly 100% regioselectivity to hydrogenate the nitrogen-containing ring with the electron-donating groups (−CH3, −OCH3, −OH) or the electron-withdrawing halogen substituents at various positions, although some substrates might suffer from much harsher reaction conditions.

Table 2 Hydrogenation of various quinoline derivatives on the 5Ir/meso_S-C catalyst.

Reaction condition: T = 100°C, P = 2 MPa, 2 hours, Ir/substrate = 0.1 at %, 0.5 mmol substrate, 1 ml ethanol as solvent, o-xylene as standard.

Embedded Image
View this table:


To summarize, we have demonstrated a sulfur-tethering strategy for the scalable synthesis of a family of atomically dispersed catalysts with high noble metal loading by using meso_S-C as supports. The meso_S-C supports with high porosity and sulfur content offered numerous accessible sulfur sites for immobilizing noble metal atoms based on the strong chemical metal-sulfur interactions. The prepared atomically dispersed catalysts exhibited much enhanced performance for catalyzing FAOR and quinoline hydrogenation compared to the commercial carbon–supported nanoparticle catalysts. Our work may open a new gate for fabricating high-loading atomically dispersed metal catalysts by an easy scale-up route for practical industrial use.



SiO2 fumed powder (7 nm, 99%; S5130) was purchased from Sigma-Aldrich. 2,2′-Bithiophene (98%) was obtained from J&K Scientific Ltd. Commercial Pt/C was purchased from Johnson Matthey Chemicals Ltd. All other chemicals were purchased from Sinopharm Chemical Reagent Co. Ltd., China. All the chemicals were used as received without further purification.

Synthesis of the meso_S-C support

Meso_S-C was prepared through cobalt-assisted carbonization of molecular precursor according to our previous works (41, 42). In a typical synthesis, 2 g of 2,2-bithiophene, 2.0 g of SiO2, and 1.0 g of Co(NO3)2·6H2O were added into 150 ml of tetrahydrofuran and stirred for 6 hours under room temperature to form a homogeneous mixture. The solvent was then removed by rotary evaporation. The dried powder was ground in an agate mortar and then carbonized under flowing N2 for 2 hours at 800°C with a heating rate of 5°C min−1. Afterward, the carbonized product underwent alkaline etching with 2.0 M NaOH for 3 days and acidic etching with 0.5 M H2SO4 for 6 hours at 90°C successively. After drying the product overnight at 85°C, the meso_S-C support was finally yielded. The meso_N-C and meso_O-C supports were prepared by the same process with 1,10-phenanthroline and glucose as precursor, respectively. Meso_S-C supports with low sulfur content were obtained by thermal treatment of the pristine meso_S-C supports under N2 at 900°C for 2 hours.

The micro_S-C support was prepared by the carbonization of thiophene-containing conjugated microporous polymer, which was synthesized via a FeCl3-catalyzing polymerization process (43). Briefly, 166 mg of 2,2-bithiophene and 1.30 g of FeCl3 were dissolved in 30 and 20 ml of anhydrous CHCl3, respectively. Then, the 2,2-bithiophene solution was dropwise added into the FeCl3 solution under nitrogen atmosphere, then stirred for 24 hours, and followed by adding 100 ml of methanol. The obtained precipitate was collected by filtration and washed with methanol and 2 M HCl solution, then extracted in a Soxhlet extractor with tetrahydrofuran for 24 hours, and dried in vacuum at 80°C for 24 hours to get the polymer. The as-prepared polymer was further carbonized at 800°C under flowing N2 for 2 hours to get the micro_S-C.

Synthesis of the atomically dispersed noble metal catalysts

For 5Pt/meso_S-C synthesis, 95 mg of meso_S-C was firstly dispersed in 20 ml of deionized water before a certain amount of H2PtCl6 aqueous solution (1.5 mg ml−1 Pt) was added. After ultrasonic mixing for 1 hour, the suspension was stirred vigorously for 12 hours, and the water was removed by rotary evaporation. Last, the precursor was treated with 5 vol % H2/Ar for 2 hours at 250°C with a heating rate of 5°C min−1. The preparation process of 5Rh/meso_S-C, 5Pd/meso_S-C, and 3Ru/meso_S-C was similar to that of Pt/meso_S-C. For the 5Ir/meso_S-C synthesis, the final reduction step was performed in flowing Ar at 500°C. Ir/CNT, Ir/RGO, and Ir/XC-72R were prepared under the same conditions.

Physicochemical characterizations

HAADF-STEM images were performed on JEM-ARM 200F with an accelerating voltage of 200 kV. XRD analysis was performed on a Philips X’Pert PRO SUPER X-ray diffractometer with Cu Kα radiation (λ = 1.54056 Å). The surface chemical state of the supported metal was analyzed by XPS conducted on an Emeso_S-CAL MKII with an excitation source of Mg Kα radiation (1253.6 eV). The BET surface areas and pore size distribution of the supports were measured by N2 sorption isotherms (77 K) and the Barrett-Joyner-Halenda model with ASAP 2020. The H2-TPR was performed on Autosorb-iQ (Quantachrome Instruments, USA) in a U-type quartz tube. First, the sample was heat-treated in He flow (50 ml min−1) for 1 hour at 300°C with a heating rate of 10°C min−1 and then the sample was cooled to 25°C. Subsequently, the sample was heated from 25° to 800°C with a heating rate of 5°C min−1 under 10% H2/Ar flow (50 ml min−1) after reaching the stable baseline (ca. 0.5 hours). The desorbed gas was detected by a thermal conductivity detector.

The XAFS spectra at the Pt L-edge, Ir L-edge, and Ru K-edge were measured in transmission or fluorescence mode at the beamline 1W1B station of the Beijing Synchrotron Radiation Facility operated at 2.5 GeV and 250 mA. Pd and Rh K-edge were tested in transmission mode at the BL14W1 station of Shanghai Synchrotron Radiation Facility (SSRF) operated at 3.5 GeV and 220 mA. The x-ray energy was tuned by a fix-exit Si (111) double-crystal monochromator and calibrated using Pt/Rh/Pd foil or Ir/Ru metal powder before the experiments. All samples were pelletized as disks with a diameter of 8 mm and a thickness of 1 mm. The XAFS data were processed using the IFEFFIT package (44). The raw data were background-subtracted, normalized, and Fourier-transformed by the ATHENA program. Least-squares curve fitting analysis of the EXAFS data was carried out using the ARTEMIS program. The many-body amplitude-reduction factor S02 was determined from the fitting of the standard Pt foil and then set for all other Pt samples. The k range for Fourier transformation was from 2.5 to 12.4 Å−1, and the fitted R range was from 1.1 to 3.2 Å. The fitting methods of multiple k-weight and multiple datasets were exploited to get better and reasonable parameters, such as setting the same energy shifts for all the Pt-S shell and all the Pt-Pt shell, respectively.

FAOR catalytic performance

Formic acid electrochemical oxidation measurements were performed using a CHI-760E electrochemical workstation (CH Instruments Inc., Shanghai, China) by using a three-electrode electrochemical cell at room temperature. A glassy carbon disk (5 mm in diameter), a platinum wire, and a saturated calomel electrode were used as working, counter, and reference electrodes, respectively. For the preparation of catalyst ink, 2 mg of the 10Pt/meso_S-C was homogenously dispersed in 2 ml of the isopropanol/NAFION solution, followed by stirring and ultrasonication. A certain amount of the catalyst ink was placed uniformly on the glassy carbon disk working electrode with a Pt loading of 10.2 μg cm−2 for all catalysts. The electrocatalytic performance of Pt/meso_S-C for the FAOR was estimated by cyclic voltammetry (CV) and CA in a mixture of 0.5 M HCOOH and 0.5 M H2SO4 solution. All CV profiles were obtained in the range of +0.05 to +1.2 V with a sweep rate of 50 mV s−1.

Hydrogenation of quinoline

In a typical catalysis experiment, 58.8 μl of quinoline (0.5 mmol), a certain amount of 5Ir/meso_S-C catalyst, and 1 ml of ethanol were added into a glass reaction vessel. Next, the reaction vessel was placed into an autoclave reactor, subsequently purged with H2 five times at 1 MPa, and pressurized to 2 MPa H2. The reactor was preheated at 100°C and then stirred at 1000 rpm for 2 hours. After the reaction was completed, the autoclave was placed in ice water. Orthoxylene (70 μl) was then added in the reaction system as an internal standard, and the solution was diluted with 10 mL of ethyl acetate. Last, the reaction products were analyzed using a Shimadzu gas chromatograph with a flame ionization detector and with high-purity nitrogen as the carrier gas. The TOF value was calculated by keeping the quinoline conversion about 20% based on the total Ir loading in the catalyst. For the recycling experiments, 1 mmol of quinoline and 1.9 mg of 10Ir/meso_S-C and 1 mmol of ethanol were added into a glass reaction vessel and reacted for 1 hour in accordance with the above steps; then, the catalyst was separated by centrifugation, washed with ethanol and ethyl acetate for several times, and then dried under vacuum condition at 60°C overnight.


Supplementary material for this article is available at

Fig. S1. TPR and XPS characterizations.

Fig. S2. Structure characterizations and models of Pt/meso_S-C.

Fig. S3. HAADF-STEM images of 5 wt % Pt on reference supports.

Fig. S4. Structure characterizations of Ir/meso_S-C.

Fig. S5. HAADF-STEM images of additional Pd, Rh, and Ru catalysts.

Fig. S6. Stability of the catalysts for FAOR.

Fig. S7. HAADF-STEM images of the 10Ir/meso_S-C catalyst after five cycles.

Table S1. Physicochemical properties of meso_S-C and other reference carbon supports.

Table S2. XAFS fitting parameters of Ir species in Ir/meso_S-C and Ir powder.

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.


Acknowledgments: Funding: This work was supported by the National Natural Science Foundation of China (grant 21671184, 11874334, and 11404314), the National Key R&D Program of China (grant 2017YFA0403400), the Fundamental Research Funds for the Central Universities (Grant WK2340000076), and the Joint Funds from the Hefei National Synchrotron Radiation Laboratory (Grant KY2060000107), USTC start-up funding, the Recruitment Program of Thousand Youth Talents, and the Natural Science Foundation of Anhui Province (grant 1708085MA06). Author contributions: H.-W.L. and L.W. conceived and designed the experiments. L.W., Q.-Q.Y., and S.-L.X. synthesized the catalysts. Y.L. performed the aberration-corrected STEM characterization. S.-Q.C. and M.-X.C. performed the x-ray absorption spectra measurements and analyzed the data. P.C. performed the XPS characterization. L.W., H.-W.L., Y.L., and S.-Q.C. co-wrote the manuscript. All authors discussed the results and commented on the manuscript. H.-W.L. supervised the project. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.
View Abstract

Navigate This Article