Strong electronic metal-support interactions for enhanced hydroformylation activity and stability over Rh single-atom catalysts through phosphorus doping

1.   Introduction

Hydroformylation, an important industrial process that converts H₂, CO, and alkenes to aldehydes, underpins an annual production of approximately 24 million tons of aldehydes and their derivatives, including alcohols, esters, carboxylic acids, and amines^[1]. These products serve as essential precursors and building blocks and have extensive applications in detergents, plasticizers, cosmetics, pharmaceuticals, and other fine chemical industries^{[2, 3]}. The hydroformylation of alkenes is currently conducted in a homogeneous way in industry, using molecular phosphine-modified Rh(I) complexes as catalysts, such as HRh(CO)(PPh₃)₂, owing to their high activity and selectivity^[4]. However, such homogeneous hydroformylation processes face challenges in separating molecular catalysts from products and produce a large amount of phosphorus-containing waste^[5–7]. Therefore, there is an urgent need to develop a greener, more sustainable, highly efficient catalytic process.

The development of an efficient solid hydroformylation catalyst to transform industrial homogeneous processes into heterogeneous processes has garnered much attention over the past decade. This heterogeneous process greatly simplifies catalyst–product separation and largely reduces phosphorus-containing waste^{[8, 9]}. Metallic Rh nanoparticles supported on materials such as carbon, TiO₂, SiO₂, and ZIF-8^[10–13] have been investigated as potential candidates for heterogeneous hydroformylation. However, their catalytic performance, particularly in terms of activity, remains inferior to that of homogeneous catalysts in general. More critically, they are often accompanied by prominent metal Rh leaching, resulting in notable deactivation during recycling tests^[14].

Given the high cost of Rh metal, improving the metal utilization efficiency is crucial. In this context, the development of single-atom catalysts (SACs), which maximize precious metal utilization, has recently emerged as a particularly promising strategy for the heterogeneous hydroformylation of alkenes^{[15, 16]}. For example, Qiao et al. demonstrated that the Rh₁/ZnO-nw SAC could achieve an activity of 3166 h⁻¹ in styrene hydroformylation, largely surpassing that of the homogeneous Wilkinson’s catalyst (RhCl(PPh₃)₃, 1583 h⁻¹)^[17]. This study demonstrates the promising potential of SACs in heterogeneous olefin hydroformylation. However, the catalyst stability was not investigated in this work. For SACs, the local chemical coordination environments of metal atoms are critical in defining metal–support interactions and the electronic properties of metal centers^[18], thereby governing their catalytic activity and stability^[19–21]. Precise regulation of the local coordination structure, including the coordination number (CN) and composition^{[22, 23]}, has emerged as a powerful strategy for optimizing the performance of SACs^{[24, 25]}. Ma et al. demonstrated that the calcination of Rh₁/CeO₂ SACs at high temperatures facilitated the formation of low-coordination Rh active species with a lower oxidation state of Rh via the generation of abundant oxygen vacancies. The resulting Rh₁/CeO₂ SAC exhibited a remarkably high catalytic activity of ~5000 h⁻¹ with 100% aldehyde selectivity and high stability in the hydroformylation of propylene^[26]. Christopher et al. also reported the use of atomically dispersed WO_x to modulate the coordination environment of Rh to form Rh₁–W₁O_x paired-site catalysts for ethylene hydroformylation. In this system, the adjacent Rh₁–W₁O_x species not only tuned the chemical state of the Rh single atoms but also enhanced ethylene adsorption, which led to an 18-fold increase in catalytic activity compared with that of isolated Rh single atoms^[27]. Ding et al. reported that anchoring Rh single atoms onto a porous organic polymer (POP) support containing PPh₃ organic phosphine ligands resulted in the formation of the Rh₁/POP SAC, where the Rh atom was tightly coordinated to three P atoms. This innovative catalyst exhibited remarkable performance in olefin hydroformylation and impressive durability, maintaining stability over 1000 hours of continuous operation with ethylene, propene, and long-chain olefins^{[28, 29]}. Although remarkable progress has been made in the design of high-performance SACs, their activities are considerably lower than those of industrial homogeneous catalysts in general. Moreover, in most cases, their stability against sintering and metal leaching is still a key concern. Therefore, the development of Rh SACs with both higher activity and enhanced stability is highly desirable.

Inspired by the widespread use of phosphorus (P) in homogeneous catalysts for olefin hydroformylation, we doped P atoms into the g-C₃N₄ support (denoted as PCN) to modulate the coordination environment of the Rh atoms. We then deposited Rh atoms onto g-C₃N₄ and PCN supports with different P contents via atomic layer deposition (ALD) to fabricate Rh SACs (denoted as Rh₁/g-C₃N₄ and Rh₁/PCN). X-ray photoelectron spectroscopy (XPS) revealed that P doping effectively enhances electronic transfer between single Rh atoms and the support, thereby strengthening electronic metal–support interactions (EMSIs). We showed that a high-valence Rh₁/PCN SAC optimized with a moderate level of P doping exhibited remarkable activity, approximately 5.8- and 3.3-fold higher than that of the low-valence Rh₁/g-C₃N₄ SAC and the industrial homogeneous catalyst HRh(CO)(PPh₃)₃, respectively, in the reaction of styrene hydroformylation. Additionally, this Rh₁/PCN SAC demonstrated largely improved recyclability, with no observable metal agglomerations after the recycling test.

2.   Materials and methods

2.1   Chemicals

Rhodium(III) 2,4-pentanedionate (Rh(acac)₃), styrene and n-decane were purchased from Shanghai Aladdin Bio-Chem Technology. Urea (99%), (NH₄)₂HPO₄ (99%) and toluene were purchased from Sinopharm Chemical Reagent. All the chemicals were used as received without further purification.

Ultrahigh-purity N₂ (99.999%), O₂ (99.999%), H₂ (99.999%) and Ar (99.999%), as well as gas mixtures including synthesis gas (50% CO/H₂) and 10% H₂ in Ar, were obtained from Nanjing Special Gases.

2.2   Catalyst Preparation

Synthesis of the g-C₃N₄ support. The urea was placed in a crucible and heated to 600 °C for 4 h in static air at a ramp rate of 5 °C min⁻¹. The resulting yellow material was milled into a fine powder in a mortar. A light yellow powder of the g-C₃N₄ support was finally obtained.

Synthesis of the PCN support. The as-prepared g-C₃N₄ was thoroughly mixed with (NH₄)₂HPO₄ by grinding for 30 min and then heated at 300 °C for 2 h (at a heating rate of 2 °C min⁻¹) under an Ar atmosphere to obtain PCN. Specifically, 1 g of g-C₃N₄ was separately mixed with 30 mg, 50 mg, and 80 mg of (NH₄)₂HPO₄, ground into a homogeneous mixture, and then calcined. The resulting PCN materials were denoted as 3PCN, 5PCN, and 8PCN, respectively.

Synthesis of Rh₁/g-C₃N₄ and Rh₁/PCN SACs. Rh ALD was carried out on a viscous-flow stainless steel tube reactor system (ACME (Beijing) Technology) at 300 °C using Rh(acac)₃ and oxygen as the precursors. Ultrahigh-purity N₂ (99.999%) was used as a carrier gas at a flow rate of 60 ml min⁻¹. The Rh(acac)₃ precursor container was heated to 150 °C to reach a sufficient vapor pressure. The inlet manifold was held at 180 °C to avoid precursor condensation. The timing sequence was 600, 300, 600 and 300 s for Rh(acac)₃ exposure, N₂ purge, O₂ exposure and N₂ purge, respectively. One cycle of Rh ALD was performed on g-C₃N₄ to obtain Rh₁/g-C₃N₄ catalysts. The same method was employed to deposit Rh single atoms on 3PCN, 5PCN and 8PCN, which were denoted as Rh₁/3PCN, Rh₁/5PCN and Rh₁/8PCN, respectively.

2.3   Catalyst characterization

Spherical aberration-corrected high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) measurements were taken on a Themis Z instrument (University of Science and Technology of China) at 300 keV. Energy-dispersive X-ray spectroscopy (EDS) mapping was performed on field emission high-resolution transmission electron microscopy (JEM-F200) equipment. Transmission electron microscopy (TEM) measurements were performed on another JEOL-2100F instrument operated at 200 keV. The Ni and Cu loadings of the catalysts were analyzed via an inductively coupled plasma atomic emission spectrometer (ICP‒AES). X-ray diffraction (XRD) measurements were performed on a Rigaku SmartLab instrument, with a 2θ range from 10° to 80° and a scanning speed of 10°/min. X-ray photoelectron spectra were recorded on a Thermo ESCALAB250Xi spectrometer with an excitation source of monochromatized Al Ka (hv = 1486.6 eV) and a pass energy of 30 eV. The values of the binding energies were calibrated with the C1s peak of the contaminant carbon at 284.80 eV. In situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) of CO chemisorption was performed on a Nicolet iS10 spectrometer equipped with a mercury-cadmium-telluride detector and a low-temperature reaction cell (Praying Mantis Harrick). After a sample was loaded into the cell, it was reduced under 10% H₂ in Ar at 250 °C for 2 h. After the sample was cooled to room temperature in Ar, a background spectrum was collected. The sample was subsequently exposed to 10% CO in Ar for approximately 15 min until saturation. Next, the sample was purged with Ar for another 30 min to remove the gas-phase CO, after which the DRIFT spectrum was collected with 256 scans at a resolution of 4 cm⁻¹.

2.4   Catalytic activity measurement

The olefin hydroformylation reaction was carried out in a high-pressure stainless-steel autoclave with a Teflon lining. Prior to the reaction tests, the catalyst was reduced under a 10% H₂/Ar atmosphere at 250 °C for 2 hours. Specifically, a catalyst containing 0.4 µmol of Rh, 10 mL of toluene, 2 mmol of styrene and 0.5 mmol of internal standard n-decane was sequentially added to a stainless-steel autoclave, which was then sealed. The autoclave was connected to a syngas cylinder, and the syngas was purged through the system five times to expel any residual air. Subsequently, syngas (50% CO/H₂) was introduced at a pressure of 3 MPa. The sealing of all valves was checked, and the autoclave was heated to 100 °C with stirring initiated. The reaction was allowed to proceed for a specified time. Upon completion, the autoclave was placed in an ice–water bath to cool, and the pressure was gradually released once the temperature reached room temperature. A sample of the reaction mixture was filtered, diluted with anhydrous ethanol, and analyzed via gas chromatography (Shimadzu GC-2014). After analysis, the catalyst and products were separated via vacuum filtration via a Büchner funnel. The catalyst was washed repeatedly with anhydrous ethanol and subsequently placed in a Petri dish to air dry for later use in cyclic stability tests. The reaction solutions from five cycles were combined to determine the total amount of Rh leached via ICP‒AES.

Here, styrene conversion and phenylpropyl aldehyde selectivity were calculated via the following equations:

For the kinetic measurements, we reduced the amount of catalyst to ensure that the styrene conversion was less than 20%. The turnover frequencies (TOFs) of the catalysts were calculated via the following equation:

Here, n_styrene is the number of moles of the converted styrene within a specified time, D is the dispersion of metal NPs or clusters/atoms, and D =1 for SACs.

3.   Results and discussion

3.1   Synthesis and structural characterization of PCN

The PCN support was synthesized through a two-step process involving the thermal polymerization of urea to form g-C₃N₄, followed by P doping with (NH₄)₂HPO₄ under controlled inert conditions^[30]. The PCN support was synthesized and labeled 3PCN, 5PCN, and 8PCN on the basis of the weight ratio of (NH₄)₂HPO₄ to g-C₃N₄ during synthesis. The XRD curves of the g-C₃N₄ and PCN composites with different P contents are shown in Fig. S1. For g-C₃N₄, there were two peaks located at approximately 12.9° and 27.9°, which are attributed to the (100) and (002) diffraction planes^[31]. Upon P doping, the intensity of the (100) and (002) peaks decreased slightly with the appearance of new peaks, suggesting a disruption of the long-range order and successful incorporation of P into the g-C₃N₄ framework, which is in line with the literature^{[32, 33]}.

The STEM image of 3PCN and its corresponding EDS mappings (Fig. 1a) further demonstrated the uniform distribution of P, C, and N, confirming the successful uniform incorporation of P into the g-C₃N₄ framework without any noticeable P agglomerations. XPS was performed to investigate the electronic structure of g-C₃N₄ and the chemical state of P. As shown in Fig. 1b, for g-C₃N₄, there were two peaks located at 284.8 and 288.2 eV in the C 1 s region, which are attributed to the C–C bonds arising from sp²-bonded carbon and the N–C=N bonds in the aromatic ring, respectively^[34]. With increasing P doping levels, the intensity of the C 1 s peak gradually decreased, indicating modifications to the carbon coordination environment. In the N 1 s region (Fig. 1c), we found that the N 1 s peak for g-C₃N₄ can be deconvoluted into three peaks located at 398.7, 400.0, and 401.2 eV, which are assigned to the sp² hybridized aromatic nitrogen atoms bonded to carbon atoms (C–N=C), tertiary nitrogen N–(C)₃ groups linking structural motifs or amino group atoms bonded three carbon atoms in the aromatic cycles^[35]. However, no obvious shifts to higher binding energies were observed in all three PCN samples in the C 1 s and N 1 s regions compared with those of g-C₃N₄. The P 2p spectra of 3PCN, 5PCN, and 8PCN displayed two peaks at 133.5 and 134.0 eV (Fig. 1d), corresponding to the P 2p_3/2 and P 2p_1/2 states, respectively. The P 2p_3/2 peak at 133.5 eV is consistent with P–N coordination, indicating a +5 oxidation state of phosphorus in PCN^[36]. In addition, elemental analyses revealed that the C/N ratio for g-C₃N₄ was 0.98, which was obviously higher than that of 8PCN (0.84) (Table S1). Therefore, it might be speculated that partial C atoms were probably replaced by P atoms, as illustrated in the schematic structural model (Fig. 1e). The substitution of C atoms by P (indicated by red circles) within the graphitic carbon nitride framework aligns well with previous reports^{[37, 38]}, further confirming successful P doping.

Figure  1.  Structural characterization of PCN. (a) A STEM image of 3PCN and corresponding EDS maps of P (rose red), C (cyan), and N (green). XPS spectra of g-C₃N₄, 3PCN, 5PCN and 8PCN in the C 1 s (b), N 1 s (c) and P 2p (d) regions. (e) Structure diagram of P-doped g-C₃N₄. The gray, blue, and carmine balls denote carbon, nitrogen and phosphorus atoms, respectively.

DownLoad: Full-Size Img PowerPoint

3.2   Synthesis and structural characterization of Rh SACs

Rh ALD was carried out on these supports at 300 °C for one ALD cycle to synthesize Rh₁/g-C₃N₄, Rh₁/3PCN, Rh₁/5PCN and Rh₁/8PCN SACs (Fig. S2). According to ICP‒AES, the Rh contents in these four catalysts were all approximately 0.3 wt. % (Table S2). For Rh₁/g-C₃N₄, no Rh nanoparticles were observed in the low-magnification aberration-corrected HAADF-STEM image (Fig. 2a). At high magnifications (Fig. 2b), the atomic dispersion of Rh atoms was clearly observed, as highlighted by yellow circles. For Rh₁/5PCN, the aberration-corrected HAADF-STEM image at low magnification revealed the absence of Rh nanoparticles (Fig. 2c). Corresponding elemental mapping further revealed a homogeneous distribution of P, N, C, and Rh within the 5PCN support (Fig. 2d). Aberration-corrected HAADF-STEM images at high magnification (Fig. 2e) show a higher density and uniform distribution of Rh single atoms, also marked with yellow circles. Similarly, there were also no detectable Rh nanoparticles in Rh₁/3PCN or Rh₁/8PCN (Figs. S3–S5).

Figure  2.  Structural characterization. Aberration-corrected HAADF-STEM images of Rh₁/g-C₃N₄ at low (a) and high (b) magnifications, showing that Rh single atoms are present (marked by yellow circles). Aberration-corrected HAADF-STEM image of Rh₁/5PCN at low (c) magnification. (d) Corresponding EDS maps of P (rose red), C (cyan), N (green) and Rh (yellow) for Rh₁/5PCN. Aberration-corrected HAADF-STEM image of Rh₁/5PCN at high (e) magnification. (f) In situ DRIFT spectra of CO chemisorption on the Rh₁/g-C₃N₄, Rh₁/3PCN, Rh₁/5PCN and Rh₁/8PCN catalysts. XPS spectra of Rh₁/g-C₃N₄, Rh₁/3PCN, Rh₁/5PCN and Rh₁/8PCN in the Rh 3d (g), N 1 s (h) and P 2p (i) regions.

DownLoad: Full-Size Img PowerPoint

In situ DRIFT spectra of CO chemisorption were measured for the Rh₁/g-C₃N₄, Rh₁/3PCN, Rh₁/5PCN, and Rh₁/8PCN catalysts after hydrogen reduction at 250 °C to investigate the electronic environment and morphology of the Rh active sites (Fig. 2f). For Rh₁/g-C₃N₄, CO adsorption peaks were observed at 2088 cm⁻¹ and 2024 cm⁻¹, corresponding to the symmetric and asymmetric stretches of Rh(CO)₂ gem-dicarbonyl species on atomically dispersed, positively charged Rh atoms^[17]. No peaks associated with CO adsorption on Rh nanoparticles (~2060 cm⁻¹ for linear and ~1860 cm⁻¹ for bridged geometries) were detected, confirming the atomic dispersion of Rh atoms, which is consistent with the STEM observations (Fig. 2b). P doping induced significant blueshifts in the CO adsorption peaks, indicating increased electron deficiency at the Rh centers. The peaks shifted to 2096 cm⁻¹ and 2025 cm⁻¹ for Rh₁/3PCN, 2098 cm⁻¹ and 2026 cm⁻¹ for Rh₁/5PCN, and 2102 cm⁻¹ and 2031 cm⁻¹ for Rh₁/8PCN. These shifts were attributed to strong electronic interactions between Rh and the PCN support, effectively modulating the electronic structure of the Rh sites and weakening Rh–CO interactions. Similarly, for all the Rh₁/PCN catalysts, no peaks corresponding to the linear or bridged adsorption modes of CO on the Rh nanoparticles were observed. This indicated that the Rh species in all the catalysts were highly dispersed, which is consistent with the conclusions drawn from our TEM analysis (Fig. 2e and Figs. S3–S5).

XPS analysis was conducted to elucidate the electronic structures of Rh₁/g-C₃N₄ and Rh₁/xPCN. In the Rh 3d spectra (Fig. 2g), Rh₁/g-C₃N₄ exhibited two prominent peaks corresponding to the Rh 3d_5/2 and Rh 3d_3/2 states. The Rh 3d_5/2 peak at 308.8 eV is attributed to Rh³⁺ species, with no detectable Rh species in the zero-valent state (307.6 eV)^{[3, 39]}, further confirming the atomic dispersion of Rh in Rh₁/g-C₃N₄. This finding is consistent with the STEM observations (Fig. 2a, b). For Rh₁/xPCN (x = 3, 5, 8), the Rh 3d_5/2 binding energy systematically shifted to higher values (from 309.0 eV to 309.4 eV) as the P content increased. This positive shift indicates that the Rh atoms Rh₁/xPCN became more electron deficient, reflecting the increase in the EMSIs induced by P doping. This observation aligned with the blueshift in wavenumbers detected in the DRIFT spectra of CO chemisorption (Fig. 2f). In the N 1 s spectra (Fig. 2h), Rh₁/g-C₃N₄ displayed three peaks located at 398.7, 400.0, and 401.2 eV, corresponding to C=N–C, N–(C)₃, and C–N–H bonds, respectively, without any noticeable shift compared with the g-C₃N₄ support (Fig. 1c). Upon P doping, the binding energy of the C=N–C moiety gradually decreased as the P content increased. In the P 2p region (Fig. 2i), two peaks corresponding to the P 2p_3/2 and P 2p_1/2 states are observed for the Rh₁/xPCN samples; as the P content increases, the binding energy of the P 2p_3/2 peak gradually decreases from 133.4 to 133.2 eV. Notably, compared with that of xPCN (Fig. 1d), where the binding energy was observed at 133.5 eV, the valence state of P was considerably reduced after Rh deposition. In brief, although P is less electronegative (2.19) than are carbon (2.55) and nitrogen (3.04)^{[40, 41]}, the doping of high-valence P species into the g-C₃N₄ framework enables considerable enhancement of charge transfer from Rh atoms to the support with both N and high-valence P atoms as electron acceptors, thus increasing the EMSI for high stability.

3.3   Catalytic Performance in Styrene Hydroformylation

The catalytic performances of the Rh₁/g-C₃N₄, Rh₁/3PCN, Rh₁/5PCN, and Rh₁/8PCN catalysts for styrene hydroformylation were evaluated at 100 °C and 3 MPa. For comparison, the industrial homogeneous catalyst HRh(CO)(PPh₃)₃ was also tested under the same conditions. We found that the HRh(CO)(PPh₃)₃ homogeneous catalyst exhibited a notable TOF of 2254 h⁻¹ (Fig. 3a). In comparison, the Rh₁/g-C₃N₄ catalyst exhibited a lower TOF of 1300 h⁻¹, which was related to the Rh content. Remarkably, P doping markedly enhanced the catalytic activity. For Rh₁/3PCN, the TOF increased dramatically to 5649 h⁻¹. Rh₁/5PCN displayed the highest TOF of 7528 h⁻¹, which was 5.8 and 3.3 times greater than those of Rh₁/g-C₃N₄ and the homogeneous industrial catalyst HRh(CO)(PPh₃)₃, respectively. However, further increasing the P content of Rh₁/8PCN resulted in a decrease in the TOF to 4042 h⁻¹. The products for the hydroformylation of styrene included a linear aldehyde (3-phenylpropanal), a branched aldehyde (2-phenylpropanal), and a small fraction of ethylbenzene (Fig. 3b). Notably, all the catalysts exhibited excellent selectivity (>95%) for the aldehyde product, with only trace amounts of ethylbenzene detected. The linear-to-branched (l/b) ratio remained consistent, with values of approximately 1.1, regardless of the Rh catalyst used. This indicated that the active Rh SACs on inorganic supports exhibited low steric hindrance, similar to what has been previously reported for ZnO- and MgO-supported Rh SACs^[14]. Kinetic studies further revealed the apparent activation energies for Rh₁/g-C₃N₄, Rh₁/3PCN, Rh₁/5PCN, Rh₁/8PCN, and HRh(CO)(PPh₃)₃ to be 227.8, 115.5, 109.7, 121.3, and 126.3 kJ/mol, respectively (Fig. 3c). The exceptionally low apparent activation energy observed for Rh₁/5PCN further confirmed its remarkably high intrinsic activity.

Figure  3.  Catalytic performance in styrene hydroformylation. Comparing the TOF (a) and selectivity (b) for Rh₁/g-C₃N₄, Rh₁/3PCN, Rh₁/5PCN, Rh₁/8PCN and HRh(CO)(PPh₃)₃ for the hydroformylation of styrene. The l/b ratio represents the molar ratio of 3-phenylpropanal to 2-phenylpropanal in the product mixture. Reaction conditions: 2 mmol of styrene, 10 mL of toluene, 30 bar of syngas, 100 °C. The TOF of each catalyst was calculated on the basis of the yield of aldehydes at a conversion level of < 20% using each catalyst. (c) Arrhenius plots and calculated apparent activation energies in the kinetic reaction regions. (d) Kinetic reaction order with respect to CO and styrene during the hydroformylation of styrene. Recyclability of Rh₁/g-C₃N₄ (e) and Rh₁/5PCN (f) for styrene hydroformylation.

DownLoad: Full-Size Img PowerPoint

To investigate the enhanced catalytic performance of the Rh₁/xPCN catalysts, kinetic reaction orders with respect to CO and styrene were determined (Fig. 3d). For the homogeneous catalyst HRh(CO)(PPh₃)₃, the reaction order with respect to CO was -1.41, reflecting the inhibitory effect of CO partial pressure due to the formation of inactive rhodium species such as dicarbonyl [(RCO)Rh(CO)₂(PPh₃)] and tricarbonyl [(RCO)Rh(CO)₃(PPh₃)]^[42]. In contrast, Rh₁/g-C₃N₄ exhibited a positive CO reaction order of 1.47, indicating weakened CO adsorption. P doping further modulated the CO reaction order, increasing to 1.83 for Rh₁/3PCN and 1.93 for Rh₁/5PCN before decreasing to 1.63 for Rh₁/8PCN (Fig. S6a). This trend suggested that electron-deficient Rh species on PCN support weakened CO adsorption through reduced π back-donation^[43], which is consistent with the CO chemisorption DRIFT spectra (Fig. 2f). Conversely, olefins, such as styrene, possess π-electrons that interact more favorably with electron-deficient Rh sites^[44]. For HRh(CO)(PPh₃)₃, the styrene reaction order was 0.27, whereas that of Rh₁/g-C₃N₄ was 0.91. P doping decreased the styrene reaction order to 0.41 for Rh₁/3PCN and 0.27 for Rh₁/5PCN and then increased it to 0.60 for Rh₁/8PCN (Fig. S6b). Therefore, the reduced CO adsorption but enhanced styrene adsorption on Rh₁/5PCN with a high valance state likely accounts for the large improvement in catalytic activity.

The recyclability of the Rh₁/g-C₃N₄ and Rh₁/xPCN catalysts in styrene hydroformylation was further evaluated under identical reaction conditions. For Rh₁/g-C₃N₄ (Fig. 3e), a sharp decrease in styrene conversion was observed after only three cycles, indicating considerable catalyst deactivation. In contrast, Rh₁/5PCN (Fig. 3f) demonstrated largely improved stability, retaining ~40% styrene conversion and 96% aldehyde selectivity over five consecutive cycles. However, both Rh₁/3PCN and Rh₁/8PCN also exhibited poor catalytic stability, as evidenced by a considerable decline in activity after four consecutive reaction cycles (Fig. S7). To gain further insight into the origins of the stability differences among the catalysts, we characterized the morphology and metal leaching of the samples after the stability tests. For Rh₁/g-C₃N₄ after the stability tests, TEM (Fig. 4a-c) revealed substantial Rh aggregation into large Rh nanoparticles. ICP‒OES analysis (Table S3) further revealed severe Rh leaching, with more than 60% Rh lost after three cycles of recycling. These results indicate poor stability due to a weaker EMSI. Notably, for Rh₁/5PCN after the stability tests, aberration-corrected HAADF-STEM analysis (Fig. 4d-f) revealed that the Rh species remained atomically dispersed after extended reaction cycles. The moderate decrease in activity was attributed primarily to the loss of Rh metal (43%) and P (46%) during recycling, which was much lower than that observed for Rh₁/g-C₃N₄ (Table S3). The Rh₁/5PCN catalyst, with a moderate P doping level, achieved an optimal balance between activity and structural stability, effectively strengthening EMSIs to stabilize single Rh atoms under catalytic conditions. In contrast, postreaction characterization of the Rh₁/3PCN catalyst also revealed considerable Rh aggregation and metal loss (58%) (Fig. S8, Table S3). However, there was no considerable Rh aggregation for Rh₁/8PCN after the stability tests (Fig. S9, Table S3). Therefore, proper P doping allows enhanced EMSIs, thus effectively stabilizing Rh atoms against metal aggregation. Nonetheless, the development of a method to completely prevent metal leaching from the support under reaction conditions still requires further study.

Figure  4.  Morphology of the Rh₁/g-C₃N₄ and Rh₁/5PCN samples after styrene hydroformylation stability testing. TEM images at low (a) and high (b, c) magnifications of the Rh₁/g-C₃N₄ used after 4 cycles of the styrene hydroformylation test. The aggregated Rh nanoparticles in a-c are highlighted by red circles. Aberration-corrected HAADF-STEM images at low (d) and high (e, f) magnifications of Rh₁/5PCN after 5 cycles of the styrene hydroformylation test; the Rh single atoms in e and f are highlighted by yellow circles.

DownLoad: Full-Size Img PowerPoint

4.   Conclusions

In summary, we have demonstrated that incorporating P atoms into g-C₃N₄ frameworks effectively modulates the local electronic structure of metal anchor sites, reducing the electron density of Rh SACs and facilitating charge transfer from single Rh atoms to the PCN support. This significantly enhances EMSIs, leading to markedly improved catalytic activity and stability in styrene hydroformylation. Notably, the optimized Rh₁/5PCN SAC demonstrated exceptional performance, with hydroformylation activities approximately 5.8- and 3.3-fold greater than those of Rh₁/g-C₃N₄ and the industrial homogeneous catalyst HRh(CO)(PPh₃)₃, respectively. Furthermore, Rh₁/5PCN exhibited significantly enhanced multicycle stability, with no detectable metal agglomeration after repeated reactions. Kinetic studies revealed a lower activation energy for Rh₁/5PCN, confirming its superior intrinsic activity. The reaction order data further suggested that this performance enhancement was driven by electron-deficient Rh species, which reduced CO adsorption while promoting alkene adsorption via strengthened EMSIs. These findings highlight P doping as an effective strategy for tailoring SACs, providing a promising approach for the development of highly active and stable catalysts for a broad range of catalytic reactions.

Funding Information

This work was supported by the Petrochemical Research Institute Foundation (21-CB-09-01), the National Natural Science Foundation of China (22302186 and 22025205), the China Postdoctoral Science Foundation (2022M713030 and 2023T160618) and the Fundamental Research Funds for the Central Universities (WK2060000058 and WK2060000038). The authors wish to acknowledge the support of the Xiaomi Young Talents program.