English

• Alkaline water splitting has been one of the most powerful approaches for large-scale hydrogen production [1-3], which requires efficient hydrogen evolution reaction (HER) electrocatalysts to overcome the high overpotential and reduce energy consumption [4-6]. Supported catalysts have attracted substantial research focus owing to the distinguished interfacial electronic interaction between the support and catalyst, which could regulate the electron distribution at the interface and thus influence the electronic structure of metal sites [7-9]. According to d-band center theory, the center of the d-band of a transition metal site serves as a key electronic structure descriptor in electrocatalysis, which is associated with the ability to adsorb reaction intermediates [10,11]. Therefore, the physicochemical properties of the support, including the structure, conductivity, exposed surface with functional groups/atoms, and chemical stability should be taken into consideration when evaluating the support effect on electrocatalysis. To meet the above requirements, various kinds of supports have been developed, such as metal hydroxide/oxide, sulfide, nitride, carbide, carbons, and transition metal substrates (Ni, Fe, Cu, Ti, etc.) [12-17]. It is still of great urgency to develop suitable support that could induce optimal interfacial electronic interaction with the electroactive sites toward d-band center tuning.

  TiO₂ has emerged as one of the most promising supports for electrocatalysis owing to its chemical inertness but rich surface hydroxyl groups for anchoring electroactive species [18,19]. Previous research works have demonstrated the employment of various TiO₂ nanostructures as support in electrocatalysis, including nanowires, nanorods, and nanoplates [20]. For example, Liu’s group reported on the P-doped defective TiO₂-supported Ru cluster for HER, with rich surface oxygen vacancies on rutile-TiO₂ facilitating the adsorption and dissociation of water [21]. Our recent work highlights that the universal utilization of nanostepped TiO₂ as support for nanoparticles or single-atom catalyst (SAC) anchoring, benefiting from the unsaturated edge configuration on the steps that influences the coordination environment of Co atoms [22,23]. It is worthwhile to note that most of these works focus on the direct interaction between TiO₂ support and the catalyst, taking great effort into tuning the intrinsic structure of TiO₂ itself, which aims to achieve a desired interfacial interaction. Commonly used strategies such as atom-doping, vacancy/defect construction, or morphology engineering might be arduous to sustain desired regulation, even at the cost of damaging the surface structure [24-26]. To this end, a feasible but universal strategy to modify the surface chemistry of TiO₂ support with controllable interfacial interaction is highly desired but challenging.

  Herein, we report on a heterostructured support by covalently linking carbon nitride quantum dots (CNQDs) on stepped TiO₂ for anchoring CoP clusters (TiO₂—CNQD/CoP). Varying the number of conjugated N-heterocycles in CNQDs leads to tunable π-electron delocalization, which manipulates the interfacial electronic interaction between the heterostructured support (TiO₂—CNQD) and CoP. Furthermore, the shift of the Co d-band center (Δd) has been proposed as a descriptor to predict the adsorption strength of intermediates, exhibiting a volcano-shaped correlation with the HER performance. TiO₂—CNQD/CoP-C2 with moderate size of CNQD shows the optimal HER performance, with 98 mV overpotential to achieve a current density of 100 mA/cm².

  Fig. 1a schematically illustrates the synthetic procedure of TiO₂—CNQD/CoP, including acid-etching formed stepped TiO₂, covalently linking of CNQD, and Co²⁺ impregnation and phosphidation conversion. Nanostepped TiO₂ nanowire arrays supported on Ti mesh were prepared according to previous protocol [27], in which the hydrothermally grown TiO₂ nanowire arrays were etched with dilute hydrochloric acid, generating abundant surface steps. CNQDs were synthesized by the hydrothermal oxidation and exfoliation of bulk graphitic carbon nitride (g-C₃N₄), introducing –COOH and –OH functional groups on the edge. HRTEM images show a typical 0.32 nm lattice fringe corresponding to the (002) plane of g-C₃N₄ (Fig. S1 in Supporting information) [28]. XRD pattern of CNQDs revealed the characteristic peak of g-C₃N₄, demonstrating the crystal structure. XPS analysis shows the presence of –COOH, –OH, and –NH₂ groups on conjugated N-heterocycles in CNQDs for further functionalization (Fig. S2 in Supporting information). Covalent grafting of CNQDs onto stepped TiO₂ can be accomplished by esterification reaction between the –COOH on CNQDs and –OH on TiO₂, which leads to the formation of heterostructured TiO₂—CNQD support. HRTEM image shows a uniform decoration of CNQDs on TiO₂ (Fig. S3 in Supporting information). Varying the oxidation and exfoliation period could control the size of CNQDs to be 1.67 nm (CNQD-1), 2.96 nm (CNQD-2), and 3.79 nm (CNQD-3) in diameters, respectively (Fig. S4 in Supporting information).

  Figure 1

  

  Figure 1.  (a) Schematic illustration of the synthetic procedure. (b) SEM image, (c) TEM image, (d, e) HRTEM images, (f) HAADF-STEM and elemental mapping of TiO₂—CNQD/CoP. (g) XRD patterns of TiO₂—CNQD/CoP, TiO₂—CNQD/CoP-NC and TiO₂/CoP.

  DownLoad: Full-Size Img PowerPoint

  The integration of such a heterostructured support enlarges the surface area with oxygen-containing groups on CNQDs as an anchoring site for Co²⁺. Phosphidation topologically converts the Co²⁺···O into CoP clusters bridged on CNQD-TiO₂ (TiO₂—CNQD/CoP) by C-P-Co (Fig. 1a). TiO₂—CNQD/CoP preserves the nanowire arrays morphology with clearly defined nanosteps (Figs. 1b-d). HRTEM image exhibits lattice fringes of 0.35 nm, 0.32 nm, and 0.20 nm (Fig. 1e), which can be indexed to the (101) plane of TiO₂, (002) of CNQDs, and (210) of CoP, respectively, evidencing well-resolved hetero-interfaces in TiO₂—CNQD/CoP. STEM-EDX elemental mapping images show the homogeneous distribution of Ti and O signals on the entire nanowire, while a more concentrated surface location of Co, P, C, and N elements (Fig. 1f). A noncovalent heterostructured TiO₂—CNQD support that is driven by electrostatic adsorption was also fabricated and employed as a counterpart (TiO₂—CNQD/CoP-NC). Fig. S5 (Supporting information) shows that CoP aggregates into large particles, which might originate from the random distribution of CNQDs on TiO₂. Both counterparts exhibit uniform distribution of CoP on the TiO₂—CNQD support (Fig. S6 in Supporting information).

  XRD patterns of TiO₂—CNQD/CoP, TiO₂—CNQD/CoP-NC, and TiO₂/CoP exhibit sharp reflections at 31.6°, 36.3°, 46.2°, and 48.1° that correspond to the (011), (111), (112), and (211) planes of cubic CoP (JCPDS No. 29–0497) [29]. For TiO₂—CNQD/CoP and TiO₂—CNQD/CoP-NC, a broad feature centered between 20° and 30° reflects the disordered (002) plane of CNQDs [30]. Peaks at 25.5, 35.1, 38.3, 40.1 and 74.1 were ascribed to the TiO₂ or Ti substrate [31]. Similarly, TiO₂—CNQD/CoP-C1 and TiO₂—CNQD/CoP-C3 were synthesized using CNQDs of ~1.67 nm and ~3.79 nm, respectively.

  Fourier transform infrared (FTIR) spectra of TiO₂—CNQD, TiO₂—CNQD/NC, TiO₂ and CNQDs were shown in Fig. 2a. It reveals C = O stretching at 1720 cm⁻¹ for free CNQDs and the noncovalent composite [32,33], while this band emerges a positive shift to 1750 cm⁻¹ in TiO₂—CNQD. Meanwhile, the broad peak associated with –OH around 3200 cm⁻¹ in TiO₂ shifts to lower wavenumber with decreased intensity [34]. Both of the above are indicative of ester bond formation between CNQDs and TiO₂. The solid-state ¹³C NMR spectroscopy was further collected for the TiO₂–CNQD (Fig. 2b), which displays a characteristic resonance at 174.2 ppm that can be assigned to an ester carbonyl (O–C=O) environment [35]. Zeta-potential (ζ) of TiO₂ and CNQDs is measured to be +19.6 mV and −24.7 mV, respectively (Fig. 2c). Both TiO₂—CNQD/NC and TiO₂—CNQD converge to near-neutral ζ values (+3.3 mV and +4.7 mV, respectively). Such sharp changes in ζ are relative to the strong interfacial electronic interactions, which would promote interfacial electron transfer between TiO₂ and CNQDs.

  Figure 2

  

  Figure 2.  (a) FTIR spectra of TiO₂—CNQD, TiO₂—CNQD/NC, CNQDs and S-TiO₂. (b) ¹³C solid-state NMR spectra of TiO₂–CNQD. (c) Zeta potential measurements of TiO₂—CNQD, TiO₂—CNQD/NC, CNQDs and S-TiO₂. (d) Full XPS spectra of TiO₂—CNQD/CoP, TiO₂—CNQD/CoP-NC and TiO₂/CoP. The high-resolution XPS of (e) C 1s and (f) N 1s in TiO₂—CNQD/CoP, TiO₂—CNQD/CoP-NC and TiO₂/CoP, (g) Ti 2p, (h) P 2p and (i) Co 2p in TiO₂—CNQD/CoP, TiO₂—CNQD/CoP-NC and CoP/TiO₂.

  DownLoad: Full-Size Img PowerPoint

  XPS survey spectra detect Ti, O, Co, and P in all catalysts, while C and N signals were shown in CNQD-containing samples (Fig. 2c). High-resolution C 1s spectra resolve signals from C–C (284.8 eV), C–P (285.35 eV), C–O (286.3 eV), N–C=N (287.95 eV), and O–C=O (289.0 eV) in Fig. 2d [36]. The peak at 285.35 eV could be assigned to the C–P bond [37,38], which reflects the interfacial bridges between CoP and CNQDs. N 1s spectra reveal contributions from sp² N (398.85 eV), pyridinic N (399.7 eV), and N–H (400.8 eV) for TiO₂—CNQD/CoP (Fig. 2e) [39]. Compared to CNQDs, the N–C=N peak shifts positively by 0.5 eV in TiO₂—CNQD/CoP, suggesting electron transfer between CNQDs with CoP and TiO₂ [40]. O 1s spectra display characteristic C=O (531.6 eV) and C–O (532.4 eV) features that originate from oxygen-containing groups in CNQDs (Fig. S7 in Supporting information) [41]. Ti 2p spectra show Ti⁴⁺ at 458.92 and 464.67 eV, alongside Ti³⁺ signals at 457.4 and 462.45 eV due to acid-etching–induced surface oxygen vacancies in TiO₂/CoP, (Fig. 2g) [42]. After covalently linking CNQDs, the Ti⁴⁺ peaks shift positively by 0.32 eV in TiO₂—CNQD/CoP, suggesting CNQD-induced electron transfer from the TiO₂ support. Further insights arise from the Co 2p and P 2p regions (Figs. 2h and i). In TiO₂/CoP, Co 2p shows Co–P (778.15, 796.05 eV) and Co–O (781.5, 797.8 eV) doublets with associated satellites, while P 2p reveals P–Co (127.85, 128.8 eV) and P–O (133.35 eV) components [43]. Co–P and P–Co binding energies shift negatively by ~0.42 eV and ~0.25 eV in TiO₂—CNQD/CoP sample, indicating CNQDs induce more electron transfer from the support to CoP. A new P–C peak at 132.65 eV further supports heterojunction formation (Fig. 2i) [44,45]. TiO₂—CNQD/CoP-C1 and TiO₂—CNQD/CoP-C3 show a similar trend in the binding energy shift of Co 2p and P 2p with that of TiO₂—CNQD/CoP-C2 (Fig. S8 in Supporting information). This size-dependent binding energy shift is consistent with the value of Δd, revealing that CNQDs of moderate size (C2) induce an optimal π-electron delocalization. In contrast, too small (C1) or too large (C3) CNQDs result in less favorable electronic perturbations. It is found that Ti 2p and O 1s spectra share similar binding energy positions in the TiO₂—CNQD support and TiO₂—CNQD/CoP, while the N 1s spectrum in TiO₂—CNQD exhibits binding energies located between that of CNQDs and TiO₂—CNQD/CoP (Fig. S9 in Supporting information), which suggests the binding energies shifts of TiO₂—CNQD/CoP are caused by the electronic interactions between CoP and the TiO₂—CNQD support. Taken together, the positive shifts of N and Ti signals and negative shifts of Co and P could suggest directional electron transfer from TiO₂ to CNQDs and to CoP, finally. Such an interfacial electronic interaction is essential to tuning the Co d-band center and optimizing hydrogen adsorption energetics, toward enhanced HER performance.

  The HER activity of TiO₂—CNQD/CoP was evaluated in 1.0 mol/L KOH using a standard three-electrode configuration at room temperature. Linear sweep voltammetry (LSV) curves demonstrate that covalent linkage of CNQDs markedly enhances the catalytic activity of TiO₂/CoP. Among all samples, CNQDs exhibit negligible catalytic activity for HER in the tested potential window, while TiO₂—CNQD/CoP-C2 exhibits the lowest overpotential of 98 mV at 100 mA/cm², outperforming both TiO₂—CNQD/CoP-C1 (116 mV) and TiO₂—CNQD/CoP-C3 (128 mV) counterparts. The noncovalently linked TiO₂—CNQD/CoP-NC and unmodified TiO₂/CoP show significantly inferior performances, requiring 148 mV and 252 mV, respectively, to reach the same current density. The corresponding Tafel slopes further differentiate their intrinsic HER kinetics (Fig. 3b), in which TiO₂—CNQD/CoP-C2 delivers the smallest value (65.3 mV/dec) following a faster Volmer–Heyrovsky mechanism. Similar electrochemically active surface area (ECSA) of TiO₂—CNQD/CoP indicates that the incorporation of conjugated N-heterocycles of different sizes in TiO₂/CoP did not increase the number of active sites (Fig. S10 in Supporting information) [46]. Electrochemically active surface area (ECSA)-normalized LSVs further confirm the superior intrinsic activity and efficient CoP utilization in TiO₂—CNQD/CoP-C2 (Fig. 3c). The mass loading of CoP in these TiO₂—CNQD/CoP catalysts was determined to be comparable (ca. 1.2 mg/cm², Table S1 in Supporting information). TiO₂—CNQD/CoP-C2 delivers a higher mass activity at η₁₀₀ than TiO₂—CNQD/CoP-C1 and C3 (Fig. S11 in Supporting information), suggesting that the superior HER activity of TiO₂—CNQD/CoP-C2 may stem from the improved intrinsic activity induced by CNQDs. At an overpotential of 200 mV, the turnover frequency (TOF) of TiO₂—CNQD/CoP-C2 reaches 11.4 s⁻¹, significantly higher than that of bare TiO₂/CoP (Fig. S12 in Supporting information). The radar diagram suggests that TiO₂—CNQD/CoP-C2 performs as one of the best alkaline HER electrocatalysts compared with several other advanced Co-based HER catalysts (Fig. 3d) [22,47-49]. Furthermore, the stability of TiO₂—CNQD/CoP-C2 was evaluated by chronoamperometry and multistep chronopotentiometry, with a continuous stability over 100 h of electrocatalysis, outperforming that of TiO₂—CNQD/CoP-NC (Fig. 3e and Fig. S13 in Supporting information). Besides, the charge transfer amount (Q) of TiO₂—CNQD/CoP-C2 (36 × 10³ C/cm²) during operating time is one to two orders of magnitude higher than those of reported TMP-based electrocatalysts (Fig. S14 and Table S2 in Supporting information). Such observation also confirms a rapid surface reaction kinetics in the heterostructured TiO₂—CNQD/CoP, which ensures efficient transfer and consumption of electrons.

  Figure 3

  

  Figure 3.  (a) LSV curves of CNQDs, TiO₂/CoP, and TiO₂—CNQD/CoP samples. (b) Tafel slopes, (c) LSV curves normalized by ECSA of TiO₂—CNQD/CoP-C1, TiO₂—CNQD/CoP-C2, TiO₂—CNQD/CoP-C3, TiO₂—CNQD/CoP-NC and TiO₂/CoP. (d) Comparison of the alkaline HER performance in terms of overpotential (η₁₀₀), Tafel slope, TOF, C_dl, specific activity, and mass activity at η₁₀₀. (e) Chronoamperometry measurements of TiO₂—CNQD/CoP-C2 TiO₂—CNQD/CoP-NC and at the current densities of 0.1 A/cm². Nyquist plots of (f) TiO₂/CoP and (g) TiO₂—CNQD/CoP-C2 at various voltages. (h) Schematic of electrode structure and the equivalent circuit. Bode phase plots of (i) TiO₂/CoP and (j) TiO₂—CNQD/CoP-C2 at various voltages. (k) The H/D KIE values of TiO₂/CoP and TiO₂—CNQD/CoP-C2 at different potentials.

  DownLoad: Full-Size Img PowerPoint

  The post-TiO₂—CNQD/CoP catalyst after continuous electrocatalysis maintains the nanoparticles on nanowire morphology (Figs. S15a and b in Supporting information). HRTEM image further reveals the presence of characteristic lattice fringes corresponding to CNQDs and CoP (Fig. S15c in Supporting information), respectively, which indicates the structural robustness of the TiO₂—CNQD/CoP. Survey XPS spectrum confirms the presence of N and C signals in the post-HER catalyst with nearly consistent elemental composition to those in the fresh catalyst (Fig. S15d in Supporting information). Furthermore, C 1s and N 1s spectra exhibit unchanged binding energies, demonstrating that CNQDs were well preserved during long-term electrocatalysis (Figs. S15e and f in Supporting information). No new phase in CoP was observed (Figs. S15g–j in Supporting information), further evidencing the structural stability.

  The charge transfer kinetics and the adsorption and desorption kinetics of reaction intermediates on active sites during the HER is further investigated using in-situ electrochemical impedance spectroscopy (EIS) measured at different applied potentials. For the heterostructured TiO₂—CNQD/CoP, in-situ EIS data were fitted with a two-time-constant equivalent circuit (Fig. 3h). The chosen model consists of a R_s (solution resistance) in series with two parallel R–CPE sub-circuits. The fitting reveals two characteristic charge-transfer resistance components at the outer electrode/electrolyte interface (R₁) and the internal heterointerface between CoP and the TiO₂—CNQD support (R₂). The two-time-constant EIS analysis indicates that the incorporation of CNQDs markedly decreases the inner-layer charge transfer resistance (Table S3 in Supporting information), confirming that the heterostructured support improves electronic conduction [50]. Nyquist plots of TiO₂—CNQD/CoP-C2 exhibited a smaller semicircular arc compared to that of TiO₂/CoP in the potential window from −0.02 V to −0.2 V vs. RHE (Figs. 3i and j), which suggests that TiO₂—CNQD/CoP-C2 surface was favorable for reactant adsorption and charge transfer in HER [51]. It is well accepted that peaks in the Bode plots located at low and high-frequency regions can be ascribed to the charge transfer reaction that occurs at either the electrolyte-catalyst interface or the catalyst inner-layer/interface, respectively [52,53]. For TiO₂—CNQD/CoP-C2, the high phase angle at low frequencies and the very low phase angle at high frequencies indicate that the electron transfer resistance in HER is primarily due to the electrolyte-catalyst interface (Figs. 3i and j, and Fig. S16 in Supporting information), while TiO₂/CoP experiences both kinds of charge transfer resistances [54], suggesting that the introduction of covalently linked CNQD at the hetero-interface promoted charge transfer at the catalyst inner-layer/interface [55].

  Furthermore, the kinetic isotope effects (KIE) involving deuterium and hydrogen for HER were studied (Fig. 3k and Fig. S17 in Supporting information). Both TiO₂—CNQD/CoP-C2 and TiO₂/CoP exhibit a primary KIE, which matches well the previous Tafel slope results. A significant decrease in the KIE values for TiO₂—CNQD/CoP-C2 compared to TiO₂/CoP occurs, indicating the accelerated water dissociation kinetics for TiO₂—CNQD/CoP-C2 [56]. All these observations suggest that covalently formed TiO₂—CNQDs heterostructured support can significantly optimize the water adsorption, dissociation, coupled with interfacial charge transfer in TiO₂—CNQD/CoP during the HER process [57].

  Density functional theory (DFT) calculations were conducted to investigate the CNQD boosted HER mechanism. Based on the experimental results, four models were constructed: TiO₂—CNQD/CoP-C1-C3 containing different size of CNQDs, and TiO₂/CoP without CNQD as counterpart (Fig. 4a). CNQDs with different number of triazine rings were linked to the (101) plane of TiO₂ via ester bond, and decorated with CoP through C-P-Co bond in TiO₂—CNQD/CoP-C1-C3 series. Charge density difference maps show the electron transfer direction from TiO₂ to CNQDs and to CoP (Fig. 4a and Fig. S18 in Supporting information), leading electron accumulation near CoP at the CoP/CNQDs interface, which is consistent with the XPS analysis. As revealed in the density of states (DOS) plots (Fig. 4b), covalently grafting of CNQDs induced more electrons to appear around the Fermi level, which might suggest an improved electronic conductivity to facilitate electron transfer during HER [58].

  Figure 4

  

  Figure 4.  (a) DFT-optimized models with the insert charge density difference (yellow: electron accumulation, and cyan: electron depletion). (b) The density of states (DOS) of TiO₂/CoP, TiO₂—CNQD/CoP-C1, TiO₂—CNQD/CoP-C2 and TiO₂—CNQD/CoP-C3. (c) Volcano plot of the electronic descriptor Δd vs. overpotential at 100 mA/cm². (d) Gibbs free energy profiles of TiO₂/CoP and TiO₂—CNQD/CoP-C1 for alkaline HER. (e) Gibbs free energies of H (ΔG_H*) on TiO₂/CoP and TiO₂—CNQD/CoP-C1. (f) Mechanism illustration of the covalently grafted CNQDs in the heterostructured support enhanced HER.

  DownLoad: Full-Size Img PowerPoint

  The projected DOS (PDOS) results show that the d-band center of Co in TiO₂—CNQD/CoP-C1-C3 series experiences different shifts relative to that of the TiO₂/CoP. To quantify the effect of CNQDs size on the d-band center of Co, a heterostructured support based descriptor, denoted as Δd was introduced. This descriptor is defined as the shift in the Co d-band center relative to the unmodified TiO₂/CoP. A volcano-shaped relationship was observed between Δd and the HER performance (Fig. 4c), in which CoP/CNQD-S/TiO₂—C2 located at the apex with optimal configuration for balanced H*/OH* adsorption and desorption. It is observed that TiO₂—CNQD/CoP-C3 exhibits an opposite direction of Δd with that of TiO₂—CNQD/CoP-C1 and TiO₂—CNQD/CoP-C2, toward a much deeper d-band center (–1.17 eV). Such observation suggests that CNQDs with moderate size can modulate the d-band center of Co to achieve optimal HER kinetics. Such d-band center modulation could be rationalized to that the conjugated N-heterocycles in CNQDs induced tunable π-electron delocalization depends on the size of CNQDs, which influences the interface electron distribution in TiO₂—CNQD/CoP and thus the d-band center of Co.

  Considering that all TiO₂—CNQD/CoP series exhibit superior electrocatalytic HER performance to TiO₂/CoP, the simplest TiO₂—CNQD/CoP-C1 was employed as-proof-concept for demonstration. The energy barrier for water dissociation at TiO₂—CNQD/CoP-C1 is only 0.9 eV (Fig. 4d), which is much smaller than that of TiO₂/CoP (−1.94 eV), indicating a significant acceleration of the Volmer step on the heterostructured interface [59]. TiO₂—CNQD/CoP-C1 exhibits favorable OH* and H* adsorption behaviors along the alkaline HER pathway, with a close to zero free energy of hydrogen adsorption (∆G_H*) (Fig. 4e). All these observations confirm that the covalently grafted CNQDs to form the heterostructured support modulate the electronic d-band center of Co (Fig. 4f), optimize water dissociation and intermediate H*/OH* adsorption, toward facilitated HER kinetics and boosted electrocatalytic performance.

  Considering the excellent hydrogen evolution performance of TiO₂—CNQD/CoP-C2 in alkaline media, a flow-type AEM electrolyzer was constructed to further investigate its feasibility under actual industrial conditions. The schematic diagram for a typical AEM electrolyzer is shown in Fig. 5a, in TiO₂—CNQD/CoP-C2 and our previous work NiFe-LDH/NGQDs serve as the cathode and anode, respectively. Fig. 5b presents the polarization curves measured at 60 ℃ (without iR correction), where the TiO₂—CNQD/CoP-C2||NiFe-LDH/NGQDs has a better electrochemical performance than 20% Pt/C||RuO₂. The cell voltage increased from 1.88 V to 1.90 V after 100 h at a current density of 1 A/cm² with a voltage degradation rate of 0.22 mV/h (Fig. 5c). This TiO₂—CNQD/CoP-C2||NiFe-LDH/NGQDs AEMWE device demonstrates comparable operational stability (Table S4 in Supporting information), underscoring the promising practical potential in high-efficiency water electrolysis.

  Figure 5

  

  Figure 5.  (a) Schematic illustration of a typical anion-exchange membrane (AEM) electrolyzer. (b) Polarization curves of TiO₂—CNQD/CoP-C2||NiFe-LDH/NGQDs and 20% Pt/C||RuO₂ electrolyzers (without iR correction). (c) Chronopotentiometry curve of the AEM electrolyzer at 1 A/cm².

  DownLoad: Full-Size Img PowerPoint

  In summary, we have developed a heterostructured support based on covalently grafting of CNQDs on TiO₂ to anchor CoP clusters as HER electrocatalyst. Benefitting from the tunable size of CNQDs induced controllable degree of π-electron delocalization at the interface, the Co d-band center could be modulated. A descriptor demonstrating the shift of d-band center relative to that of TiO₂/CoP (Δd) was proposed and exhibits a volcano-shaped relationship with the HER performance. TiO₂—CNQD/CoP-C2 with a moderate CNQD size exhibits an optimal d-band center that favorable for water dissociation, H*/OH* adsorption and HER kinetics, delivering superior HER activity with only 98 mV overpotential to achieve 100 mA/cm². This work not only provides mechanistic insight into size-dependent interfacial regulation of d-band center via nanosized CNQDs, but also establishes a general strategy for rationally engineering support-catalyst interfaces for heterogeneous electrocatalysis.

  Declaration of competing interest

  The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

  CRediT authorship contribution statement

  Sheng Qian: Writing – original draft, Methodology, Investigation. Jingying Wei: Validation, Methodology, Data curation. Junhua Wang: Methodology, Formal analysis. Yi Zhang: Visualization, Validation. Huaiguo Xue: Supervision, Project administration. Tengfei Jiang: Software, Funding acquisition. Jingqi Tian: Writing – review & editing, Resources, Funding acquisition, Conceptualization.

  Acknowledgments

  This work was supported by the Natural Science Foundation of Jiangsu Province (No. BK20211602), the Qing Lan Project of Yangzhou University.

  Supplementary materials

  Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2025.112106.

  
  
• 1. [1]

     Y. Jiang, Z. Liang, H. Fu, et al., J. Am. Chem. Soc. 146 (2024) 9012–9025. doi: 10.1021/jacs.3c13367

  2. [2]

     Y. Zhang, M. Zhao, J. Wu, et al., ACS Catal. 14 (2024) 7867–7876. doi: 10.1021/acscatal.4c01133

  3. [3]

     J. Liu, J. Ren, Y. Du, et al., Adv. Funct. Mater. 34 (2024) 2315773. doi: 10.1002/adfm.202315773

  4. [4]

     Y. Lu, C. Pei, W. Li, et al., Chin. Chem. Lett. 36 (2025) 111646. doi: 10.1016/j.cclet.2025.111646

  5. [5]

     M. Li, H. Wang, Z. Yang, et al., Chin. Chem. Lett. 36 (2025) 110199. doi: 10.1016/j.cclet.2024.110199

  6. [6]

     X. Liu, H. Pang, X. Kou, et al., Chin. Chem. Lett. 37 (2026) 111567. doi: 10.1016/j.cclet.2025.111567

  7. [7]

     W. Tom, H. Carlos, P. Krijn, Nat. Catal. 2 (2019) 955–970. doi: 10.1038/s41929-019-0364-x

  8. [8]

     H. Li, C. Chen, D. Yan, et al., J. Mater. Chem. A 7 (2019) 23432–23450. doi: 10.1039/c9ta04888j

  9. [9]

     T. Liu, R. Lin, Y. Liu, et al., Chin. Chem. Lett. 37 (2026) 111312. doi: 10.1016/j.cclet.2025.111312

  10. [10]

      S. Sun, X. Zhou, B. Cong, et al., ACS Catal. 10 (2020) 9086–9097. doi: 10.1021/acscatal.0c01273

  11. [11]

      L. Zhang, K. Chen, L. Li, et al., Chin. Chem. Lett. 37 (2026) 111002. doi: 10.1016/j.cclet.2025.111002

  12. [12]

      Y. Li, H. Zhang, M. Jiang, et al., Adv. Funct. Mater. 27 (2017) 1702513. doi: 10.1002/adfm.201702513

  13. [13]

      G. Malta, S.A. Kondrat, S.J. Freakley, et al., Science 355 (2017) 1399–1403. doi: 10.1126/science.aal3439

  14. [14]

      S. Qian, F. Xu, Y. Fan, et al., Nat. Commun. 15 (2024) 2774. doi: 10.1038/s41467-024-47061-6

  15. [15]

      Y. Gao, S. Qian, H. Wang, et al., App. Catal. B: Environ. 320 (2023) 122014. doi: 10.1016/j.apcatb.2022.122014

  16. [16]

      G. Zhao, K. Rui, S. Dou, et al., Adv. Funct. Mater. 28 (2018) 1803291. doi: 10.1002/adfm.201803291

  17. [17]

      Y. Du, H. Lu, J. Wu, et al., Angew. Chem. Int. Ed. 64 (2025) e202420903. doi: 10.1002/anie.202420903

  18. [18]

      Y. Zhang, J. Zhao, H. Wang, et al., Nat. Commun. 13 (2022) 58.

  19. [19]

      Z. Jiang, X. Xu, Y. Ma, et al., Nature 586 (2020) 549–554. doi: 10.1038/s41586-020-2738-2

  20. [20]

      Z. Wang, W. Xu, G. Tan, et al., Sci. Bull. 68 (2023) 613–621. doi: 10.1016/j.scib.2023.03.003

  21. [21]

      S. Zhou, H. Jang, Q. Qin, et al., Angew. Chem. Int. Ed. 61 (2022) e202212196. doi: 10.1002/anie.202212196

  22. [22]

      S. Qian, T. Jiang, J. Wang, et al., ACS Catal. 14 (2024) 18690–18700. doi: 10.1021/acscatal.4c05624

  23. [23]

      Z. Wei, H. Wang, C. Zhang, et al., Angew. Chem. Int. Ed. 60 (2021) 16622–16627. doi: 10.1002/anie.202104856

  24. [24]

      L. Zhang, W. Shang, S. Qiao, et al., ACS Catal. 14 (2024) 15827–15836. doi: 10.1021/acscatal.4c03006

  25. [25]

      J. Matsubu, S. Zhang, L. DeRita, et al., Nat. Chem. 9 (2016) 120–127.

  26. [26]

      X. Chang, S. Thind, A. Chen, ACS Catal. 4 (2014) 2616–2622. doi: 10.1021/cs500487a

  27. [27]

      J. Wang, S. Qian, D. Jia, et al., Inorg. Chem. 64 (2025) 1532–1540. doi: 10.1021/acs.inorgchem.4c04811

  28. [28]

      M. Albolkany, Y. Wang, W. Li, et al., Chem. Int. Ed. 59 (2020) 21499. doi: 10.1002/anie.202009710

  29. [29]

      Y. Fan, J. Wang, S. Qian, et al., J. Colloid Interface Sci. 667 (2024) 128–135. doi: 10.1016/j.jcis.2024.04.066

  30. [30]

      T. Huo, Q. Deng, F. Yu, et al., ACS Appl. Mater. Interfaces 14 (2022) 13419–13430. doi: 10.1021/acsami.2c01522

  31. [31]

      X. Wang, P. Wu, Z. Wang, et al., ACS Sustainable Chem. Eng. 9 (2021) 3083–3094. doi: 10.1021/acssuschemeng.0c07292

  32. [32]

      G. Nagaraju, T. Ravishankar, K. Manjunatha, et al., Mater. Lett. 109 (2013) 27–30. doi: 10.1016/j.matlet.2013.07.031

  33. [33]

      S. Barman, M. Sadhukhan, J. Mater. Chem. 22 (2012) 21832–21837. doi: 10.1039/c2jm35501a

  34. [34]

      X. Wu, F. Chen, X. Wang, et al., Appl. Surf. Sci. 427 (2018) 645–653. doi: 10.3727/096504017x15031557924141

  35. [35]

      Y. Liu, W. Fang, Z. Yue, et al., Nat. Water 3 (2025) 430–438. doi: 10.1038/s44221-025-00419-6

  36. [36]

      H. Liu, X. Lv, J. Qian, et al., ACS Nano 14 (2020) 13304. doi: 10.1021/acsnano.0c05143

  37. [37]

      X. Huang, X. Xu, C. Li, et al., Adv. Energy Mater. 9 (2019) 1803970. doi: 10.1002/aenm.201803970

  38. [38]

      H. Li, S. Xu, H. Yan, et al., Small 14 (2018) 1800367. doi: 10.1002/smll.201800367

  39. [39]

      B. Liu, J. Du, G. Ke, et al., Adv. Funct. Mater. 32 (2021) 2111125.

  40. [40]

      L. Tian, Z. Li, P. Wang, et al., J. Energy Chem. 55 (2021) 279–294. doi: 10.1016/j.jechem.2020.06.057

  41. [41]

      I. Bertóti, M. Mohai, K. László, Carbon 84 (2015) 185–196. doi: 10.1016/j.carbon.2014.11.056

  42. [42]

      V. Veettil, A. Vijayakumar, D. Purayil, et al., ChemElectroChem 7 (2020) 3319–3323. doi: 10.1002/celc.202000680

  43. [43]

      W. Yuan, T. Jiang, X. Fang, et al., Chem. Eng. J. 439 (2022) 135743. doi: 10.1016/j.cej.2022.135743

  44. [44]

      X. Yu, L. Feng, H.S. Park, J. Power Sources 390 (2018) 93–99. doi: 10.1016/j.jpowsour.2018.04.032

  45. [45]

      X. Hu, M. Fan, Y. Zhu, et al., Green Chem. 21 (2019) 5274–5283. doi: 10.1039/c9gc01910c

  46. [46]

      C. Xu, H. Yu, H. Huang, et al., Angew. Chem. Int. Ed. 64 (2025) e202504667. doi: 10.1002/anie.202504667

  47. [47]

      Y. Li, S. Zhang, B. Li, et al., ACS Nano 19 (2025) 7401–7416. doi: 10.1021/acsnano.5c00318

  48. [48]

      Y. Liu, H. You, C. Bai, et al., Adv. Energy Mater. 15 (2025) e03695. doi: 10.1002/aenm.202503695

  49. [49]

      Z. Zhou, Y. Su, H. Tan, et al., J. Am. Chem. Soc. 147 (2025) 3994–4004. doi: 10.1021/jacs.4c11788

  50. [50]

      S. Zhang, L. Yin, S. Wang, et al., ACS Nano 17 (2023) 23103–23114. doi: 10.1021/acsnano.3c08921

  51. [51]

      P. Yan, T. Yang, M. Lin, et al., Adv. Funct. Mater. 33 (2023) 2301343. doi: 10.1002/adfm.202301343

  52. [52]

      W. Chen, B. Wu, Y. Wang, et al., Energy. Environ. Sci. 14 (2021) 6428. doi: 10.1039/d1ee01395e

  53. [53]

      W. Gou, Z. Xia, X. Tan, et al., Nano. Energy 104 (2022) 107960. doi: 10.1016/j.nanoen.2022.107960

  54. [54]

      C. Xie, W. Chen, S. Du, et al., Nano. Energy 71 (2020) 104653. doi: 10.1016/j.nanoen.2020.104653

  55. [55]

      L. Zhou, T. Wan, Y. Zhong, et al., Small 20 (2024) 2406076. doi: 10.1002/smll.202406076

  56. [56]

      Q. Dai, L. Wang, K. Wang, et al., Adv. Funct. Mater. 32 (2022) 2109556. doi: 10.1002/adfm.202109556

  57. [57]

      K. Sun, X. Wu, Z. Zhuang, et al., Nat. Commun. 13 (2022) 6260. doi: 10.1038/s41467-022-33984-5

  58. [58]

      S. Li, M. Sun, K. Zhang, et al., Angew. Chem. Int. Ed. 64 (2025) e202508366. doi: 10.1002/anie.202508366

  59. [59]

      Q. Song, J. Li, S. Wang, et al., Small 15 (2019) 1903395. doi: 10.1002/smll.201903395

• 

  Figure 1  (a) Schematic illustration of the synthetic procedure. (b) SEM image, (c) TEM image, (d, e) HRTEM images, (f) HAADF-STEM and elemental mapping of TiO₂—CNQD/CoP. (g) XRD patterns of TiO₂—CNQD/CoP, TiO₂—CNQD/CoP-NC and TiO₂/CoP.

  下载: 全尺寸图片 幻灯片
  

  Figure 2  (a) FTIR spectra of TiO₂—CNQD, TiO₂—CNQD/NC, CNQDs and S-TiO₂. (b) ¹³C solid-state NMR spectra of TiO₂–CNQD. (c) Zeta potential measurements of TiO₂—CNQD, TiO₂—CNQD/NC, CNQDs and S-TiO₂. (d) Full XPS spectra of TiO₂—CNQD/CoP, TiO₂—CNQD/CoP-NC and TiO₂/CoP. The high-resolution XPS of (e) C 1s and (f) N 1s in TiO₂—CNQD/CoP, TiO₂—CNQD/CoP-NC and TiO₂/CoP, (g) Ti 2p, (h) P 2p and (i) Co 2p in TiO₂—CNQD/CoP, TiO₂—CNQD/CoP-NC and CoP/TiO₂.

  下载: 全尺寸图片 幻灯片
  

  Figure 3  (a) LSV curves of CNQDs, TiO₂/CoP, and TiO₂—CNQD/CoP samples. (b) Tafel slopes, (c) LSV curves normalized by ECSA of TiO₂—CNQD/CoP-C1, TiO₂—CNQD/CoP-C2, TiO₂—CNQD/CoP-C3, TiO₂—CNQD/CoP-NC and TiO₂/CoP. (d) Comparison of the alkaline HER performance in terms of overpotential (η₁₀₀), Tafel slope, TOF, C_dl, specific activity, and mass activity at η₁₀₀. (e) Chronoamperometry measurements of TiO₂—CNQD/CoP-C2 TiO₂—CNQD/CoP-NC and at the current densities of 0.1 A/cm². Nyquist plots of (f) TiO₂/CoP and (g) TiO₂—CNQD/CoP-C2 at various voltages. (h) Schematic of electrode structure and the equivalent circuit. Bode phase plots of (i) TiO₂/CoP and (j) TiO₂—CNQD/CoP-C2 at various voltages. (k) The H/D KIE values of TiO₂/CoP and TiO₂—CNQD/CoP-C2 at different potentials.

  下载: 全尺寸图片 幻灯片
  

  Figure 4  (a) DFT-optimized models with the insert charge density difference (yellow: electron accumulation, and cyan: electron depletion). (b) The density of states (DOS) of TiO₂/CoP, TiO₂—CNQD/CoP-C1, TiO₂—CNQD/CoP-C2 and TiO₂—CNQD/CoP-C3. (c) Volcano plot of the electronic descriptor Δd vs. overpotential at 100 mA/cm². (d) Gibbs free energy profiles of TiO₂/CoP and TiO₂—CNQD/CoP-C1 for alkaline HER. (e) Gibbs free energies of H (ΔG_H*) on TiO₂/CoP and TiO₂—CNQD/CoP-C1. (f) Mechanism illustration of the covalently grafted CNQDs in the heterostructured support enhanced HER.

  下载: 全尺寸图片 幻灯片
  

  Figure 5  (a) Schematic illustration of a typical anion-exchange membrane (AEM) electrolyzer. (b) Polarization curves of TiO₂—CNQD/CoP-C2||NiFe-LDH/NGQDs and 20% Pt/C||RuO₂ electrolyzers (without iR correction). (c) Chronopotentiometry curve of the AEM electrolyzer at 1 A/cm².

  下载: 全尺寸图片 幻灯片
•