Photocatalytic transfer hydrogenation of nitriles to primary amines over a Pd nanocube-modified poly(heptazine imide) catalyst

Abstract

Hydrogenation reactions are integral to the fine chemical, pharmaceutical, and petrochemical industries. However, achieving efficient and selective hydrogenation under mild conditions without stoichiometric reductants or high-pressure hydrogen remains a significant challenge. While photocatalytic transfer hydrogenation offers a sustainable alternative, its application has been largely restricted to the reduction of C=C, C≡C, and C=O bonds. In this work, we expand the scope of this strategy to nitrile groups (C≡≡N) by developing a novel catalyst (C–Pd–PHI) based on palladium cubic nanoparticles supported on poly(heptazine imide). The experimental results indicate that the Pd nanocubes primarily expose the (100) crystal facets and exhibit a strong interaction with the PHI support, which facilitates the formation of Pd–H species through protonation. Concurrently, the relatively weak binding affinity of the Pd nanocubes toward hydrogen intermediates promotes the desorption and release of hydrogen gas, which effectively suppresses over-hydrogenation reactions. Given the aforementioned properties, with 2-(3-chloro-5-(trifluoromethyl)pyridin-2-yl)acetonitrile employed as the model substrate, the C–Pd–PHI exhibits a substrate conversion efficiency of 73% and a target amine selectivity of 79% under visible light irradiation. Notably, after undergoing five consecutive catalytic cycles, the selectivity of the target amine remains stable at 75%. This work underscores the critical role of nanocrystal facet engineering in photocatalyst design and provides new ideas for sustainable transfer hydrogenation reactions.

---

1. Introduction

Hydrogenation reactions constitute a fundamental methodology in modern synthetic chemistry, facilitating the reduction of unsaturated functional groups with broad applications spanning from pharmaceutical synthesis to energy-related processes.¹ These transformations are broadly classified into two categories based on the hydrogen source: direct hydrogenation and transfer hydrogenation.² Direct hydrogenation typically employs molecular hydrogen (H₂) as the reductant but often requires elevated temperatures and pressures, specialized reactor designs, and stringent safety measures to mitigate explosion risks.³ In contrast, transfer hydrogenation utilizes alternative hydrogen donors (e.g., alcohols, hydrazine, acids, and water), which allow operation under milder conditions, offering enhanced practicality and scalability.^4–6 Consequently, the development of efficient and tunable catalytic systems, whether derived from noble metals, earth-abundant elements, or metal-free frameworks, is crucial for advancing the selectivity, sustainability, and applicability of hydrogenation technologies.

Photocatalytic systems, operating under ambient temperature and pressure conditions, present a uniquely compatible platform for transfer hydrogenation reactions.⁷ This distinctive advantage stems from the capacity of photocatalysis to circumvent conventional harsh thermal and high-pressure hydrogenation environments, thereby preserving sensitive functional groups while enhancing reaction selectivity.⁸ The synergistic interplay between photogenerated charge carriers (electrons and holes) and alternative hydrogen donors enables precise regulation of hydrogenation pathways under energetically favorable and sustainable conditions. However, current photocatalytic transfer hydrogenation systems remain constrained by their reliance on noble metal catalysts (e.g., Pd, Au, Pt), which present challenges for sustainable implementation.^9–11 This limitation has intensified research focus on the rational design of supported catalyst architectures, where strategic engineering of metal-support interactions, particle size distribution, and spatial confinement becomes crucial for optimizing catalytic performance, stability, and selectivity in these light-driven transformations.

Polymeric carbon nitride (PCN), a class of visible-light-responsive semiconductor materials, has garnered extensive attention in photocatalytic applications due to its non-toxic nature, low cost, and remarkable chemical stability.^12–14 The presence of rich surface functional groups, especially different nitrogen species, provides multiple coordination sites that facilitate effective metal immobilization, rendering PCN an ideal platform for designing supported metal catalysts.¹⁵ Recent advances demonstrate that carbon nitride matrices can effectively host metallic species across multiple scales, including nanoparticles, single atoms, and dual atoms.^16–18 The resulting composite photocatalysts exhibit exceptional performance in transfer hydrogenation reactions, achieving enhanced activity and selectivity through synergistic interactions between the metal and support.^19,20 For instance, Chen et al. constructed a diatomic palladium-decorated carbon nitride catalyst via a precursor-preselected wet deposition method, which enabled a superior 92% yield for the photocatalytic transfer hydrogenation of 4-vinylphenol to 4-ethylphenol with water as the hydrogen donor.²¹ Moreover, Zhao et al. developed a single-atom nickel-modified carbon nitride material, in which the atomically dispersed Ni²⁺–N₄ catalytic centers promote the visible-light-driven activation of water and enable the selective semi-hydrogenation of alkynes with remarkable precision.²² While these advances have significantly broadened the scope of carbon nitride-mediated photocatalytic transfer hydrogenation, the substrate range has remained largely confined to conventional functional groups such as C=C, C=O, and N=O bonds.^23–26 The transfer hydrogenation of more challenging unsaturated bonds using carbon nitride-based photocatalysts remains a significant challenge.

The synthesis of primary amines via the transfer hydrogenation of nitriles is a highly atom-efficient strategy.²⁷ However, this transformation is hindered by the high dissociation energy of the C≡N bond and the thermodynamic instability of reactive imine intermediates, necessitating harsh conditions and posing difficulties in selectivity control.²⁸ Additionally, photocatalytic transfer hydrogenation of nitriles is still in its initial and underdeveloped stage. A few reported photocatalytic systems rely on high-energy ultraviolet light sources, and the substrates are limited to aromatic nitriles.²⁹ Notably, the application of carbon nitride-based materials, despite their promising photoredox properties, remains completely unreported for this challenging transformation. These limitations underscore the urgent need to develop and design advanced carbon nitride-based photocatalysts to enable efficient and selective transformations under mild conditions.

In this study, palladium nanocubes terminated by (100) facets were prepared through a chemical reduction route and subsequently loaded onto poly(heptazine imine) (PHI) via an electrostatic self-assembly strategy. The formation of a Schottky heterojunction between Pd nanocubes and PHI results in an increased electron density on Pd atoms, facilitating the binding of protons to form Pd–H species and thereby promoting the efficiency of transfer hydrogenation reactions. In comparison to PHI-supported Pd nanorods with exposed (111) facets (R–Pd–PHI) and amorphous Pd nanoparticles (A–Pd–PHI), the synthesized Pd nanocube/PHI composite (C–Pd–PHI) exhibits a moderate adsorption capacity for hydrogen species, thereby effectively suppressing hydrogenation side reactions. Taking the transfer hydrogenation of the pharmaceutical intermediate 2-(3-chloro-5-(trifluoromethyl)pyridin-2-yl)acetonitrile as a representative example, the catalyst C–Pd–PHI achieves 73% substrate conversion with 79% selectivity toward the target amine product under visible light irradiation. Our findings broaden the application scope of photocatalysis in hydrogen transfer reactions and underscore the fundamental significance of crystal facet engineering in the rational design of metal–semiconductor hybrid catalysts.

2. Results and discussion

2.1. Synthesis, and structural and morphological characterization

The preparation process of Pd nanocubes and nanorods is shown in Fig. 1a. The detailed preparation methods are outlined in the SI. Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) were used to characterize the morphology of the prepared Pd nanoparticles. As shown in Fig. 1b and c, the as-prepared nanoparticles are in the form of cubes and rods, respectively. TEM images (Fig. 1d–f) confirm that the Pd nanocubes are terminated by (100) facets, exhibiting an interplanar spacing of 0.197 nm. In contrast, the Pd nanorods are predominantly terminated by (111) facets, exhibiting a lattice spacing of 0.228 nm (Fig. 1g–i).³⁰ These results confirm the successful synthesis of two distinct types of Pd nanoparticles with well-defined morphological characteristics. Without the addition of cetyltrimethylammonium bromide (CTAB), the synthesized Pd nanoparticles exhibit spherical and irregular morphologies, underscoring the critical role of surfactants in directing crystal nucleation and growth (Fig. S1). For this study we selected poly(heptazine imine) (PHI), a two-dimensional, highly crystalline carbon nitride derivative, which is recognized as an excellent support material for metal species (Fig. S2).³¹ Zeta potential measurements reveal that PHI dispersions in water exhibit a negative surface charge, while both Pd nanocubes and nanorods display positive surface potentials (Fig. S3). This significant charge difference indicates strong electrostatic interactions between the Pd nanoparticles and PHI, which facilitate the formation of composite materials via self-assembly.

Fig. 1 (a) The preparation process of Pd nanocubes and Pd nanorods (CTAB – cetyl trimethyl ammonium bromide). SEM images of (b) Pd nanocubes and (c) Pd nanorods. TEM images of (d–f) Pd nanocubes and (g–i) Pd nanorods.

To verify the successful assembly between Pd nanoparticles and the carbon nitride support, TEM characterization was performed on the composite (C–Pd–PHI) formed by Pd nanocubes and PHI as a representative case. As shown in Fig. 2a, PHI exhibits a characteristic lamellar stacking morphology, with uniformly dispersed Pd nanocubes approximately 22 nm in size distributed across its surface. High-resolution TEM imaging confirms that these nanocubes expose the (100) facets, displaying an interplanar spacing of 0.201 nm (Fig. 2b). The selected-area electron diffraction (SAED) pattern reveals distinct diffraction rings assignable to the (200), (111), and (311) facets of palladium, confirming the face-centered cubic (FCC) crystal structure of the nanocubes (Fig. 2c).³² Furthermore, energy-dispersive X-ray spectroscopy (EDS) elemental mapping demonstrates homogeneous spatial distribution of carbon, nitrogen, and palladium throughout the examined area (Fig. 2d–g), collectively verifying the successful construction of the composite catalyst.

Fig. 2 (a and b) TEM images of C–Pd–PHI. (c) The selected-area electron diffraction pattern of Pd nanocubes on C–Pd–PHI. (d–g) EDS elemental mapping of C–Pd–PHI.

X-ray diffraction (XRD) was first used to analyze the crystal structure of samples. As shown in Fig. 3a, the PHI support exhibits characteristic diffraction peaks at 2θ = 8.2° and 28.3°, corresponding to the (100) and (002) crystal planes, respectively.³³ In the C–Pd–PHI and R–Pd–PHI composites, distinct diffraction peaks emerge at 40.1°, 46.6°, and 68.1°, which can be assigned to the (111), (200), and (220) planes of face-centered cubic Pd, which is consistent with the TEM results.³⁴ Notably, the A–Pd–PHI sample prepared via photoreduction exhibits no discernible diffraction peaks attributable to Pd, which can be ascribed to either the amorphous nature or the ultrafine particle size. This observation is further corroborated by TEM analysis, which confirms the presence of amorphous Pd nanoparticles dispersed on the PHI surface (Fig. S4). Inductively coupled plasma-optical emission spectrometer (ICP-OES) analysis was employed to determine the actual Pd content in the composite materials. While the nominal Pd loading during synthesis is set at 3 wt%, the measured Pd contents in A–Pd–PHI, C–Pd–PHI, and R–Pd–PHI are 1.5, 2.2, and 2.4 wt%, respectively (Table S1). These results demonstrate that the electrostatic self-assembly method enables significantly more efficient loading of morphology-controlled Pd nanoparticles compared to the photoreduction approach. In addition, N₂ adsorption–desorption measurements were employed to characterize the specific surface area and pore structure of the synthesized samples. As shown in Fig. S5, the Pd-loaded sample exhibits a slight decrease in the Brunauer–Emmett–Teller (BET) specific surface area, which may be attributed to the partial coverage of the catalyst surface with anchored metal nanoparticles. Notably, the pore structure remains mainly unchanged after metal deposition, suggesting that the incorporation of Pd species does not alter the support morphology (Table S2).

Fig. 3 (a) The powder XRD patterns of the samples. The XPS data of the samples: (b) C 1s, (c) N 1s, and (d) Pd 3d.

X-ray photoelectron spectroscopy (XPS) was subsequently employed to investigate the chemical states of the samples' surface. As shown in Fig. 3b, the C 1s spectra can be deconvoluted into three peaks. The peak observed at approximately 288.4 eV is assigned to sp²-hybridized carbon (N=C–N), while the signal at around 286.5 eV corresponds to carbon at the edges (C–NH_X).³⁵ Notably, in comparison to PHI and A–Pd–PHI, the relative intensity of the C–C/C=C peak (located at 284.8 eV) is significantly enhanced in both C–Pd–PHI and R–Pd–PHI. This enhancement may be attributed to residual surfactants on the surfaces of the Pd nanoparticles introduced during the wet-chemical synthesis process. The N 1s spectra can be resolved into three characteristic peaks (Fig. 3c), corresponding to the sp² hybridized pyridine nitrogen (C=N–C) in the heptazine ring (399.9 eV), the ternary nitrogen (–N(C)₃) (400.1 eV), and the amino group (–NH_X) (401.0 eV).³⁶ As shown in Fig. 3d, spectra of all three composite materials exhibit signals corresponding to both Pd⁰ and Pd²⁺ species.³⁷ Notably, the A–Pd–PHI sample prepared via photoreduction demonstrates a higher proportion of Pd in the oxidized state. In contrast, Pd is present in its zero-valent state in both C–Pd–PHI and R–Pd–PHI. Moreover, as all types of Pd precursors contain acidic species, protonation of the PHI support is induced, thereby leading to a marked decrease in the potassium content within the structure (Fig. S6). This phenomenon is further corroborated by the Fourier transform infrared spectroscopy result (Fig. S7).

2.2. Photophysical properties characterization

UV-vis diffuse reflectance spectra (DRS) were employed to investigate the light absorption properties of the samples. The incorporation of Pd nanoparticles consequently leads to a significant enhancement in light absorption at longer wavelengths across all three composites (Fig. 4a). It is noteworthy that the absorption edge of the composites displays a significant blue shift compared to that of pristine PHI. These observations can be attributed to the protonation of the PHI support during the composite preparation process, a phenomenon also documented in related studies.^38,39 In Fig. S8, the absorption band edge of the acid-treated PHI (H–PHI) exhibits a clear blue shift. Based on the Tauc plots, the band gap energies for PHI, A–Pd–PHI, C–Pd–PHI, and R–Pd–PHI are estimated to be 2.72, 2.74, 2.77, and 2.82 eV, respectively (Fig. S9). Photoluminescence (PL) measurements were conducted to investigate the influence of Pd nanoparticle loading on the separation of photogenerated electron–hole pairs. As shown in Fig. 4b, all three composites exhibit a significant reduction in PL intensity compared to pristine PHI. The decrease in the PL intensity of the composite material can be attributed to the modified Pd species, which act as an “electron reservoir” to capture photogenerated electrons from PHI, thereby suppressing the recombination of electron–hole pairs. Among them, C–Pd–PHI demonstrates the weakest PL emission, suggesting that the Pd nanocubes facilitate the most efficient inhibition of charge carrier recombination.⁴⁰ Furthermore, the fact that protonation does not lead to a reduction in PL intensity confirms that the observed PL quenching in the composites is attributable to the presence of Pd, rather than the protonation effect (Fig. S10).

Fig. 4 (a) The UV-vis diffuse reflectance absorption spectra. (b) PL spectra recorded upon sample excitation at 325 nm. (c) Nyquist plots of the samples (R₁: series resistance, R₂: charge transfer resistance, CPE₁: constant phase element). (d) The photoelectrochemical responses.

To assess the charge transport characteristics, transient photocurrent and electrochemical impedance spectroscopy (EIS) measurements were conducted. After loading Pd, the samples exhibit a smaller arc radius in the Nyquist plots, which corresponds to a lower charge transfer resistance (Fig. 4c).⁴¹ A pronounced enhancement in the photocurrent response is observed in the sequence of PHI < A–Pd–PHI < R–Pd–PHI < C–Pd–PHI (Fig. 4d). The superior photocurrent intensity of C–Pd–PHI directly correlates with its excellent efficiency in promoting the separation of photogenerated charge carriers.⁴² While it is challenging to definitively state that the Pd (100) facet is universally superior, the above findings clearly demonstrate that, within our specific PHI-based system, the Pd nanocubes enclosed by (100) facets achieve remarkable enhancements in both charge separation and transport efficiency.

2.3. The performance of photocatalysts

The selective transfer hydrogenation of 2-(3-chloro-5-(trifluoromethyl)pyridin-2-yl)acetonitrile (1) was employed to evaluate the photocatalytic performance of the sample. The target hydrogenation product, 2-(3-chloro-5-(trifluoromethyl)pyridin-2-yl)ethan-1-amine (2), serves as a key synthetic intermediate for fluopicolide (the synthetic route is shown in Fig. S11), a novel benzamide fungicide.⁴³ Furthermore, both the Laplacian bond order and Mayer bond order of the C≡N bond in 2-(3-chloro-5-(trifluoromethyl)pyridin-2-yl)acetonitrile are higher than those in benzonitrile and acetonitrile, indicating that the transfer hydrogenation of this compound presents greater challenges (Table S3). As shown in Fig. 5a, no target product was formed when the pristine PHI was used as the photocatalyst. In contrast, after the deposition of Pd species, both the conversion and yield enhanced significantly. Among these catalysts, C–Pd–PHI exhibits the best catalytic performance, achieving a substrate conversion of 73%, and a yield of 57% for the target product 2. When the reaction time is extended to 24 hours, the conversion of the substrate significantly increases; however, the yield decreases to 54%. This decline is attributed to the formation of an imine via the reaction of compound 2 with acetone (the dehydrogenation product of isopropanol), which reduces the selectivity toward the desired product 2 (Fig. S12). Despite achieving complete substrate conversion with A-Pd-PHI, the yield of the target product is severely limited by its poor selectivity, reaching only 33%. Given the complexity of the hydrogenation products of nitriles, it is challenging to perform a precise quantitative analysis of all product types. Nevertheless, dechlorinated products as well as hydrogenated pyridine derivatives can be identified from the mass spectrum of the reaction mixture (Fig. S13). The R–Pd–PHI catalyst, incorporating Pd nanorods, exhibits a lower conversion of 33% and a selectivity of 70%. Similarly, the PHI modified with surfactant-free synthesized Pd nanoparticles also shows unsatisfactory performance in the transfer hydrogenation reaction (Fig. S14). Collectively, these findings indicate that the reaction activity is highly dependent on the predominant crystallographic facets of the Pd catalyst.

Fig. 5 Photocatalytic transfer hydrogenation control experiments using (a) different catalysts, (b) different solvents (with C–Pd–PHI as a photocatalyst), and (c) different additives (with C–Pd–PHI as a photocatalyst). (d) Photocatalytic cycle test (with C–Pd–PHI as a photocatalyst).

A series of reaction conditions were subsequently screened. No target product was formed in the absence of either the catalyst or light irradiation, confirming that the transfer hydrogenation reaction necessitates both catalyst mediation and photoexcitation (Fig. S15). No target product 2 was formed when the reaction was conducted using pure water or 0.1 M HCl as the solvent (Fig. 5b). Furthermore, a significant decrease in selectivity and yield was observed when 0.1 M HCl was used in mixed solvents with methanol (MeOH) or ethanol (EtOH), underscoring the superior suitability of isopropanol (IPA) as the hydrogen donor and hole sacrificial agent. The transfer hydrogenation exhibited low selectivity when isopropanol was mixed with water or other acids/bases (Fig. 5c). This suggests that a sufficiently acidic environment may provide a large number of protons (H⁺) to react with photoelectrons (H⁺ + e⁻ → H^˙), thereby promoting the hydrogenation reaction. The addition of trace amounts of CTAB to the system did not enhance the yield of the transfer hydrogenation product, indicating that residual carbon species on the surface of the Pd nanoparticles (Pd nanocubes and Pd nanorods) have no discernible impact on the catalytic performance (Fig. S16).

The cycling performance of the C–Pd–PHI catalyst was evaluated to assess its stability and reusability. As illustrated in Fig. 5d, the catalyst retains 75% selectivity after five cycles of its reuse. The high consistency of the FT-IR spectral features before and after recycling (Fig. S17) indicates no significant chemical alteration of the PHI frameworks. Furthermore, XRD patterns confirm that the characteristic diffraction peaks of both PHI and Pd remain virtually unchanged, suggesting excellent structural stability of the composite (Fig. S18). These results collectively affirm the catalyst's outstanding durability and consistent performance.

2.4. Mechanism investigations

The proton transfer process is critical in the transfer hydrogenation reaction. Linear sweep voltammetry (LSV) measurements reveal a very small cathodic current for pristine PHI, indicating its poor capability for proton reduction (Fig. 6a). In contrast, the deposition of Pd species significantly enhanced the current density. Among the tested catalysts, C–Pd–PHI exhibits the highest current density, suggesting that the introduction of Pd nanocubes most effectively promotes the hydrogen evolution reaction (HER), which is also beneficial for the transfer hydrogenation process. Subsequently, electron paramagnetic resonance (EPR) spectroscopy was conducted using 5,5-dimethyl-1-pyrroline N-oxide (DMPO) as a spin-trapping agent in the reaction solvent without substrates. The results reveal a characteristic six-line signal, attributable to carbon-centered radicals, in all samples (Fig. 6b). This indicates that the oxidative dehydrogenation of isopropanol can occur directly at the valence band of PHI. Notably, C–Pd–PHI exhibits the most intense EPR signal, further evidencing its superior capability in promoting this radical generation process. These findings collectively demonstrate that the loading of Pd nanocubes simultaneously enhances both the proton reduction and the sacrificial agent oxidation reactions, accounting for the high efficiency of C–Pd–PHI in the transfer hydrogenation system. Furthermore, the introduction of radical scavengers (2,2,6,6-tetramethylpiperidin-1-oxyl (TEMPO) or N-tert-butyl-α-phenylnitrone (PBN)) into the reaction system led to a substantial decrease in product 2 yield (Fig. S19). The on-off light cycling experiment demonstrates that product 2 formation occurs exclusively under illumination and ceases immediately in the dark (Fig. S20). These results confirm that the reaction proceeds via a photo-driven radical mechanism.

Fig. 6 (a) Linear sweep voltammetry plots of the samples. (b) EPR spectra of DMPO capturing carbon radicals in IPA. The XPS data of C–Pd–PHI: (c) N 1s and (d) Pd 3d. (e) H₂-TPD profiles of the samples. (f) Gas-phase analysis after the reaction.

The interfacial electron transfer between Pd nanoparticles and the PHI substrate was characterized by ultraviolet photoelectron spectroscopy (UPS). As shown in Fig. S21, the work function of PHI is determined to be 4.08 eV. This value is considerably lower than the reported work functions of Pd (100) (5.3–5.6 eV) and Pd (111) (5.4–5.9 eV) facets.⁴⁴ The significant difference indicates that electron transfer occurs from PHI to the Pd nanoparticles upon their contact.⁴⁵ Then, in situ XPS analysis was employed to further investigate electron flow under photoexcitation. The C 1s spectra of C–Pd–PHI show no significant shift before and after illumination (Fig. S22). Notably, the N characteristic peaks shift to higher binding energies (Fig. 6c), while the Pd peaks shift to lower binding energies (Fig. 6d). This observation confirms a strong electronic interaction between the PHI and Pd nanocubes. Under light irradiation, electrons are transferred from carbon nitride to metallic Pd, resulting in an electron-enriched state on Pd. This electron-rich environment facilitates the binding of H⁺ to form Pd–H species, thereby promoting the transfer hydrogenation reaction.

Pd nanoparticles can function as both active sites for transfer hydrogenation and cocatalysts for hydrogen evolution (2H⁺ + 2e⁻ → H₂). To elucidate the origin of different hydrogenation selectivities among the various Pd species, H₂ temperature-programmed desorption (H₂-TPD) experiments were conducted, with subsequent analysis of the gaseous products. As shown in Fig. 6e, the desorption peak observed at temperatures above 600 °C corresponds to catalyst decomposition, while the peak between 400 and 530 °C is assigned to hydrogen desorption from metallic Pd species. A clear shift of the hydrogen desorption peaks to lower temperatures is observed in the 150–300 °C range for the composite catalysts. This phenomenon suggests that the introduction of Pd nanoparticles reduces the adsorption energy of weakly chemisorbed hydrogen. Compared to other catalysts, C–Pd–PHI exhibits the smallest hydrogen desorption peak at a lower temperature, indicating that Pd nanocubes with (100) facets possess a weaker binding affinity for hydrogen and facilitate its release. This property directly correlates with the gas-phase analysis results, where C–Pd–PHI catalyzes the highest yield of H₂ (Fig. 6f). The integration of LSV and H₂-TPD analyses reveals that C–Pd–PHI not only demonstrates the highest proton reduction activity but also moderates the adsorption of reactive hydrogen intermediates. As a result, C–Pd–PHI can simultaneously promote selective transfer hydrogenation and hydrogen production. This synergy prevents excessive hydrogen accumulation on the active sites, thereby minimizing undesired over-hydrogenation. As reported in the literature, the Pd (111) facet exhibits a more negative free energy for atomic hydrogen adsorption (ΔG_H), indicating a stronger binding affinity for H* (chemisorbed hydrogen atoms on the catalyst surface).⁴⁶ While this property enhances hydrogenation activity, it may also promote over-hydrogenation side reactions. This may offer a plausible explanation for the lower selectivity observed with R–Pd–PHI (exposing (111) facets) and A–Pd–PHI (with mixed/amorphous surface structures).

Based on the aforementioned findings, we propose a possible mechanism (Fig. 7). Under visible light excitation, electron–hole pairs generated in PHI separate and migrate to the surface. Isopropanol is oxidized by photogenerated holes to acetone, releasing protons. Pristine PHI exhibits poor proton-reducing capability, which prevents the effective generation of active hydrogen intermediates and thus fails to catalyze the transfer hydrogenation reaction. In contrast, when PHI is coupled with Pd nanocubes, the photogenerated electrons are efficiently captured by the nanocubes and react with protons in solution to form the key Pd–H intermediate, thereby successfully driving the transfer hydrogenation process.

Fig. 7 The possible mechanism of photocatalytic transfer hydrogenation.

3. Conclusion

In summary, a composite material made of Pd nanocubes and poly(heptazine imide) has been successfully constructed via electrostatic self-assembly. The introduction of Pd nanocubes effectively enhances electron transport and promotes proton reduction. Moreover, the relatively weak adsorption of reactive hydrogen species on the Pd nanocubes, while leading to increased hydrogen evolution as a side reaction, suppresses over-hydrogenation of the substrate. As a result, C–Pd–PHI catalyzes the transfer hydrogenation of 2-(3-chloro-5-(trifluoromethyl)pyridin-2-yl)acetonitrile to 2-(3-chloro-5-(trifluoromethyl)pyridin-2-yl)ethan-1-amine with a high selectivity of 79%. Our findings demonstrate an expansion of the application scope for visible-light-driven transfer hydrogenation and provide valuable insights into the relationship between nanostructure and chemical activity.

Author contributions

Chong Wang: conceptualization, formal analysis, investigation, methodology, visualization, writing–original draft, writing–review & editing. Wenwen Tian: investigation, methodology, visualization. Yichun Lu: supervision, resources, writing–original draft. Hongyu Chen: investigation. Zhu Yin: data curation, software. Jingru Zhuang: investigation. Huali Zhang: investigation. Liuyong Chen: resources. Oleksandr Savateev: supervision, conceptualization, funding acquisition, writing–original draft, writing–review & editing. Jiajia Cheng: supervision, conceptualization, project administration, funding acquisition, project administration.

Conflicts of interest

The patents EP 3 700 910 B1 and US 11,813,596 B2 have been issued, in which O.S. is listed as the inventor.

Data availability

The data supporting this article have been included as part of the supplementary information (SI). Supplementary information is available. See DOI: https://doi.org/10.1039/d5ta10314b.

Acknowledgements

This work received financial support from the National Natural Science Foundation of China (22072020), the Science Foundation (2022HZ027004 and 2022L3082), and the Young and Middle-aged Teacher Education Research Project (JAT231024) of Fujian Province. O.S. acknowledges the Vice-Chancellor start-up funding (4933629) and matching support from the Faculty of Science of the Chinese University of Hong Kong (5501776). C.W. is thankful for the RGC Junior Research Fellow Scheme (JRFS) for partial support of this work (JRFS2526-4S07).

References

1. L. Zhang, M. Zhou, A. Wang and T. Zhang, Chem. Rev., 2020, 120, 683–733.
2. Y. Wang, Z. Huang, X. Leng, H. Zhu, G. Liu and Z. Huang, J. Am. Chem. Soc., 2018, 140, 4417–4429.
3. D. Wu, W. Baaziz, B. Gu, M. Marinova, W. Y. Hernández, W. Zhou, E. I. Vovk, O. Ersen, O. V. Safonova, A. Addad, N. Nuns, A. Y. Khodakov and V. V. Ordomsky, Nat. Catal., 2021, 4, 595–606.
4. F. Arcudi, L. Đorđević, N. Schweitzer, S. I. Stupp and E. A. Weiss, Nat. Chem., 2022, 14, 1007–1012.
5. W. Miao, W. He, Z. Fang, K. Guo and Z. Yang, Mater. Today Energy, 2023, 37, 101408.
6. R. Nie, Y. Tao, Y. Nie, T. Lu, J. Wang, Y. Zhang, X. Lu and C. C. Xu, ACS Catal., 2021, 11, 1071–1095.
7. E. Zhao, W. Zhang, L. Dong, R. Zbořil and Z. Chen, ACS Catal., 2023, 13, 7557–7567.
8. L. Marzo, S. K. Pagire, O. Reiser and B. König, Angew. Chem., Int. Ed., 2018, 57, 10034–10072.
9. E. Zhao, M. Li, B. Xu, X.-L. Wang, Y. Jing, D. Ma, S. Mitchell, J. Pérez-Ramírez and Z. Chen, Angew. Chem., Int. Ed., 2022, 61, e202207410.
10. E. M. Zahran, N. M. Bedford, M. A. Nguyen, Y.-J. Chang, B. S. Guiton, R. R. Naik, L. G. Bachas and M. R. Knecht, J. Am. Chem. Soc., 2014, 136, 32–35.
11. Y. Shi, H. Wang, Z. Wang, T. Wu, Y. Song, B. Guo and L. Wu, J. Mater. Chem. A, 2020, 8, 18755–18766.
12. F. K. Kessler, Y. Zheng, D. Schwarz, C. Merschjann, W. Schnick, X. Wang and M. J. Bojdys, Nat. Rev. Mater., 2017, 2, 17030.
13. C. Wang, J. Zhuang and O. Savateev, Acc. Chem. Res., 2026, 59, 246–257.
14. C. Wang, N. Shi, Y. Zhou, Y. Lu, J. Zhuang, H. Liao, C. Ye, H. Lei, X. Lin, J. Zheng, T. X. Liu and Z. Yuan, Adv. Sci., 2025, 12, e10084.
15. Z. Teng, Q. Zhang, H. Yang, K. Kato, W. Yang, Y.-R. Lu, S. Liu, C. Wang, A. Yamakata, C. Su, B. Liu and T. Ohno, Nat. Catal., 2021, 4, 374–384.
16. Z. Teng, H. Yang, Q. Zhang, W. Cai, Y.-R. Lu, K. Kato, Z. Zhang, J. Ding, H. Sun, S. Liu, C. Wang, P. Chen, A. Yamakata, T.-S. Chan, C. Su, T. Ohno and B. Liu, Nat. Chem., 2024, 16, 1250–1260.
17. G. A. A. Diab, I. F. Reis, C. Wang, D. Barreto, O. Savateev, I. F. Teixeira and P. Jiménez-Calvo, Adv. Funct. Mater., 2025, 35, 2501393.
18. G. F. S. R. Rocha, M. A. R. da Silva, A. Rogolino, G. A. A. Diab, L. F. G. Noleto, M. Antonietti and I. F. Teixeira, Chem. Soc. Rev., 2023, 52, 4878–4932.
19. C. Han, L. Du, M. Konarova, D.-C. Qi, D. L. Phillips and J. Xu, ACS Catal., 2020, 10, 9227–9235.
20. W. Qiao, X. Fan, W. Liu, F. N. Khan, D. Zhang, F. Han, H. Yue, Y. Li, N. Dimitratos, S. Albonetti, X. Wen, Y. Yang, F. Besenbacher, Y. Li, H. Niemantsverdriet, H. Lin and R. Su, J. Am. Chem. Soc., 2023, 145, 5353–5362.
21. E. Zhao, J. Morales-Vidal, Y. Yang, S. Mitchell, Y. Zhu, Z. Hu, J.-M. Chen, S.-C. Haw, T.-S. Chan, Z. Fan, Z.-J. Wang, N. López, J. Pérez-Ramírez and Z. Chen, J. Am. Chem. Soc., 2025, 147, 2029–2036.
22. T. Jia, D. Meng, R. Duan, H. Ji, H. Sheng, C. Chen, J. Li, W. Song and J. Zhao, Angew. Chem., Int. Ed., 2023, 62, e202216511.
23. P. Sharma and Y. Sasson, Green Chem., 2019, 21, 261–268.
24. Y. Kang, Y. Yang, L.-C. Yin, X. Kang, G. Liu and H.-M. Cheng, Adv. Mater., 2015, 27, 4572–4577.
25. E. Zhao, W. Kong, G. Zoppellaro, Y. Yang, B. Nan, L. Li, W. Zhang, Z. Chen, A. Bakandritsos, Z.-J. Wang, M. Beller, R. Zbořil and Z. Chen, Adv. Mater., 2025, 37, 2504108.
26. Y. Xu, C. Qiu, X. Fan, Y. Xiao, G. Zhang, K. Yu, H. Ju, X. Ling, Y. Zhu and C. Su, Appl. Catal., B, 2020, 268, 118457.
27. Z. Liu, F. Huang, M. Peng, Y. Chen, X. Cai, L. Wang, Z. Hu, X. Wen, N. Wang, D. Xiao, H. Jiang, H. Sun, H. Liu and D. Ma, Nat. Commun., 2021, 12, 6194.
28. X. Song, M. Liu, K. Xu and K. Xie, Int. J. Hydrogen Energy, 2025, 103, 268–287.
29. K. Imamura, T. Yoshikawa, K. Nakanishi, K. Hashimoto and H. Kominami, Chem. Commun., 2013, 49, 10911–10913.
30. W. Lei, S. Zhang, J. Shu, F. Li, Z. Deng, J. Liu, X. Guo, Y. Zhao and C. Shan, Small, 2025, 21, 2410326.
31. M. A. R. da Silva, N. V. Tarakina, J. B. G. Filho, C. S. Cunha, G. F. S. R. Rocha, G. A. A. Diab, R. A. Ando, O. Savateev, I. Agirrezabal-Telleria, I. F. Silva, S. Stolfi, P. Ghigna, M. Fagnoni, D. Ravelli, P. Torelli, L. Braglia and I. F. Teixeira, Adv. Mater., 2023, 35, 2304152.
32. W. Niu, Z.-Y. Li, L. Shi, X. Liu, H. Li, S. Han, J. Chen and G. Xu, Cryst. Growth Des., 2008, 8, 4440–4444.
33. C. Wang, Y. Hou, J. Cheng, M.-J. Lin and X. Wang, Appl. Catal., B, 2021, 294, 120259.
34. F. Xue, Q. Li, M. Lv, Y. Song, T. Yang, X. Wang, T. Li, Y. Ren, K. Ohara, Y. He, D. Li, Q. Li, X. Chen, K. Lin and X. Xing, J. Am. Chem. Soc., 2023, 145, 26728–26735.
35. C. Wang, S. Lin, Y. Lu, Y. Hou, O. Savateev and J. Cheng, ACS Catal., 2024, 14, 11308–11317.
36. C. Wang, H. Xiao, Y. Lu, J. Lv, Z. Yuan and J. Cheng, ChemSusChem, 2023, 16, e202300361.
37. X. Li, Y. Pan, H. Yi, J. Hu, D. Yang, F. Lv, W. Li, J. Zhou, X. Wu, A. Lei and L. Zhang, ACS Catal., 2019, 9, 4632–4641.
38. C. Wang, Q. Wan, J. Cheng, S. Lin, A. Savateev, M. Antonietti and X. Wang, J. Catal., 2021, 393, 116–125.
39. V. Shvalagin, N. Tarakina, B. Badamdorj, I.-M. Lahrsen, E. Bargiacchi, A. Bardow, Z. Deng, W. Wang, D. L. Phillips, Z. Guo, G. Zhang, J. Tang and O. Savateev, ACS Catal., 2024, 14, 14836–14854.
40. C. Wang, Y. Lu, Z. Wang, H. Liao, W. Zhou, Y. He, S. M. Osman, M. An, Y. Asakura, Y. Yamauchi, L. Wang and Z. Yuan, Appl. Catal., B, 2024, 350, 123902.
41. W. Wang, L. Du, R. Xia, R. Liang, T. Zhou, H. K. Lee, Z. Yan, H. Luo, C. Shang, D. L. Phillips and Z. Guo, Energy Environ. Sci., 2023, 16, 460–472.
42. H. Liao, C. Wang, J. Zheng, Z. Chen, Y. Zhou, C. Ye, X. Lin, M. An, Y. Yamauchi, Y. Asakura and Z. Yuan, Chem. Eng. J., 2025, 521, 166242.
43. C.-R. Yu, L.-H. Xu, S. Tu, Z.-N. Li and B. Li, J. Fluor. Chem., 2006, 127, 1540–1546.
44. G. N. Derry, M. E. Kern and E. H. Worth, J. Vac. Sci. Technol. A, 2015, 33, 060801.
45. Y.-Y. Cai, X.-H. Li, Y.-N. Zhang, X. Wei, K.-X. Wang and J.-S. Chen, Angew. Chem., Int. Ed., 2013, 52, 11822–11825.
46. S. Lu, B. Weng, A. Chen, X. Li, H. Huang, X. Sun, W. Feng, Y. Lei, Q. Qian and M.-Q. Yang, ACS Appl. Mater. Interfaces, 2021, 13, 13044–13054.

---