Mussel‐Inspired Self‐Assembly of PtO₄ Atomic Catalysts for Interfacial Synergistic Hydrogen Evolution

Abstract

Molecular self‐assembly strategy for Pt‐O electrocatalysts is presented, achieved by facile solution processing of dopamine monomers. This mussel‐inspired strategy ensures a highly defined PtO₄ atomic structure with uniform monolayer deposition across various 2D nanomaterials. By leveraging the intimate interplay between PtO₄ sites and 2D substrates, while maintaining a consistent coordination environment, underlying mechanism for enhanced kinetics in interfacial synergistic hydrogen evolution reaction is systematically elucidated. Notably, the self‐assembled PtO₄ site exhibits up to a 30‐fold enhancement of mass activity compared to commercial Pt/C catalysts, contingent upon the choice of substrate material. Theoretical investigation illustrates facilitated electron transfer, optimized energy barriers, and additional reaction pathways resulting from the synergistic interplay between PtO₄ and Ti₃C₂O₂ MXene. This straightforward, energy‐efficient, and highly reliable scheme for atomic‐level catalysts prospects a valuable platform toward sub‐molecular level engineering of tailored electronic structures and properties.

Mussel‐inspired strategy for self‐assembling PtO₄ atomic catalysts on different 2D substrates is presented. Notably, the self‐assembled PtO₄ site exhibits up to a 30‐fold enhancement of mass activity compared to commercial Pt/C catalysts, contingent upon the choice of substrate material. This work highlights the critical role of non‐covalent interfacial engineering in catalytic performance.

1. Introduction

The hydrogen evolution reaction (HER) is a vital element for the green energy ecosystem with high energy density as well as inherent zero‐emission characteristics.^{[ 1} , ^{2 ]} Platinum (Pt)‐based catalysts have been principally utilized to activate this reaction scheme, taking advantage of the optimal characteristics for hydrogen binding.^{[ 3} , ^{4 ]} Unfortunately, the intrinsic high cost arising from the scarcity of Pt has been a longstanding bottleneck for the practical utilization. Since the discovery of Fe‐N₄ and other single‐atom catalysts (SACs),^{[ 5} , ⁶ , ⁷ , ⁸ , ^{9 ]} Pt‐based atomic catalytic units have attracted a great deal of research attentions, particularly owing to the maximal atomic utilization and favorable electronic structure.^{[ 10} , ¹¹ , ^{12 ]}

Apart from the delicate metal‐support electronic interactions that modulate the d‐band occupation near the Fermi level,^{[ 13} , ^{14 ]} various strategies have been developed thus far to promote the catalytic activity of Pt atomic sites for HER.^{[ 15} , ^{16 ]} For instance, recent studies have discovered the excellent catalytic activity of Pt─O bonding in acidic HER, even surpassing that of Pt metals.^{[ 17} , ¹⁸ , ^{19 ]} The oxygen (O) atom not only serves as an additional active site with an optimal proton binding energy but also kinetically facilitates the hydrogen desorption.^{[ 20 ]} In addition, Pt─O─Pt atomic clusters stabilized on isolated cobalt single atoms (Pt‐ACs/CoNC) were developed as an efficient HER electrocatalyst, in which the effective interaction at the Pt─O─Co interface induces favorable charge density redistribution at both Pt and O atoms, kinetically facilitating the hydrogen evolution.^{[ 21 ]} These studies underscore the significance of electron redistribution at Pt─O active sites for maximizing catalytic activity, while also suggesting an increased interest in developing interfacial strategies to facilitate and optimize the electron density more effectively. To further advance this promising aspect, more precise and uniform model structures that enable in‐depth mechanistic studies are highly demanded. Unfortunately, the uncontrolled reactivity of O atoms at high‐energy conditions, involving high‐temperature fabrication or electrochemical deposition, poses significant challenges in precisely modelling the Pt─O atomic units.

In this work, inspired by the excellent adhesive nature and facile chelation reaction of dopamine,^{[ 22} , ^{23 ]} we successfully assembled well‐defined monodisperse PtO₄ atomic catalytic sites on various 2D substrates, including Graphene, MXene, and Transition metal dichalcogenides (TMDs), via an ambient random solution mixing process. Dopamine is a natural substance, derived from mussel adhesive proteins, whose low energy barrier for chelation reaction not only effectively prevents the aggregation of metal atoms,^{[ 24} , ²⁵ , ^{26 ]} but also ensures the uniform PtO₄ atomic configuration. The opposite charge states at 2D substrates lead to the electrostatic self‐assembly, uniformly distributing the monolayered PtO₄ atomic units on 2D surfaces. Importantly, this mussel‐inspired approach offers an idealized model system for investigating the PtO₄ active sites, while integrating the different types of substrate materials. This facile yet highly effective interfacial interaction enables the precise modulation of electron redistribution, thereby enhancing the HER efficiency. Remarkably, PtO₄ self‐assembled at MXene surfaces (PtO₄@MX) exhibits 63 times greater mass activity than the counterpart without MXene (PtO₄‐PDA_NP), highlighting the profound impact of interfacial synergistic effects with underlying substrates. Density functional theory (DFT) revealed the detailed electrocatalytic mechanism, further elucidating the role of interfacial synergistic effects in terms of facile electron redistribution. Notably, the calculated energy barrier for the rate‐determining step in PtO₄@MX is 0.26 eV, significantly lower than 0.66 eV of Pt (111). This study proposes an idealized design principle for catalytic systems with subtle synergistic cooperation between atomic configuration and supporting substrate materials.

2. Results and Discussion

2.1. Preparation and Characterization of PtO₄@MX

Based on the straightforward room‐temperature solution processing of adhesive dopamine precursors, uniform PtO₄ atomic sites are successfully incorporated on various 2D surfaces (Figure S1, Supporting Information). The typical protocol for PtO₄@MX consists of two distinct stages: atomically precise synthesis of PtO₄ active sites via dopamine chelation and subsequent electrostatic stabilization on 2D nanomaterials (Figure 1a). In our proposed mechanism of dopamine self‐oxidation illustrated in Figure S2 (Supporting Information), ammonium cations can effectively suppress the covalent polymerization via cation‐π interaction, while allowing the cyclization of amine chains (See Figure S2, Supporting Information for synthetic details).^{[ 27 ]} Attributed to the reductive environment,^{[ 28 ]} the Pt‐catechol coordination bonding can be efficiently stabilized into Pt─O covalent bonds, yielding the Pt‐chelated dopamine (PtO₄‐DA) in low‐energy states. By contrast, the synthesis of PtO₄‐DA under alkaline conditions readily leads to the formation of polydopamine nanoparticles (PtO₄‐PDA_NP) as shown in Figures S3 and S4 (Supporting Information).

Figure 1.

Preparation of PtO₄ atomic catalysts. a) Schematic illustration of the two‐step synthesis strategy for mussel‐inspired PtO₄ atomic sites self‐assembled on various 2D nanomaterials with distinct electron redistributions. b–d) HAADF‐STEM images of PtO₄@MX, PtO₄ @NrGO, and PtO₄ @MoS₂ with magnified inset images and measured lattice fringes. e,i,m) Corresponding FFT patterns with specific lattice planes. f–h) Corresponding elemental mapping of Ti, O, and Pt in the PtO₄@MX. j,k,l) Corresponding elemental mapping of C, O, and Pt in the PtO₄@NrGO. n–p) Corresponding elemental mapping of Mo, O, and Pt in the PtO₄@MoS₂. All the scale bars are 500 nm.

The electrostatic substrate deposition was also conducted in an acidic condition for the uniform dispersion of PtO₄‐DA on Ti₃C₂T_x MXene surfaces. Owing to the opposite surface charge between the as‐assembled PtO₄‐DA and MXene in the acidic pH condition (Figure 2a), the electrostatic attraction effectively facilitates the self‐assembly of PtO₄@MX, whereas the repulsive force between PtO₄‐DA prevents aggregation. The importance of an acidic environment is further verified by the differences in Brunauer–Emmett–Teller (BET) surface areas of each sample derived from their nitrogen adsorption/desorption isotherms (Figure S5, Supporting Information). MXene exhibits a BET surface area of 5.90 m² g⁻¹, whereas PtO₄@MX demonstrates a significantly higher value of 28.2 m² g⁻¹, surpassing the 24.7 m² g⁻¹ of PtO₄‐DA (Figure 2b). This enhancement is attributed to the well‐dispersed PtO₄‐DA on the MXene surface. Consequently, PtO₄‐DA can achieve an exceptional atomic‐scale dispersion on the substrates at pH 3.5 (Figure S6, Supporting Information) with a negligible change in BET surface area, whereas PtO₄‐DA strongly aggregates into clusters at pH 8.5 (Figure S7, Supporting Information). This underscores the critical role of pH conditions as a key parameter for our electrostatic self‐assembly.^{[ 29} , ³⁰ , ^{31 ]}

Figure 2.

Structure characterizations of PtO₄ atomic catalysts. a) Zeta potential measurements of PtO₄‐DA, NrGO, MXene, and MoS₂. b) BET surface areas for pristine MXene, PtO₄‐DA, PtO₄‐PDA_NP@MX, and PtO₄@MX with the schematic illustration of electrostatic self‐assembly. c) XRD θ‐2θ spectra of pristine MXene and PtO₄@MX. d) XANES spectra at the Pt‐L₃ edge of PtO₄@MX, PtO₄@NrGO, PtO₄@MoS₂, PtO₂, and reference Pt foil with the calculated oxidation states. e) Corresponding FT‐EXAFS curves and fitting curves with 4 Pt‐O pathways. f) Corresponding WT‐EXAFS plots.

The structure evolution of PtO₄‐DA is studied by X‐ray photoelectron spectroscopy (XPS) and Fourier transform Infrared (FT‐IR) spectroscopy. The XPS O 1s spectrum of PtO₄‐DA only shows the negative shift (−0.4 eV) in the deconvoluted C─O peak, attributed to the formation of C─O─Pt bonds from the initial C─O─H bonds, while inducing a higher electron density (Figure S8, Supporting Information). The intensity of the C═O deconvoluted peak also decreased with the Pt─O bond formation. No significant difference is observed except for the minor shift of the deconvoluted C─O peak in XPS C 1s and N 1s spectra, supporting the selective chelation reaction between catechol groups and Pt ions (Figure S9, Supporting Information). Chloride (Cl⁻) from the initial PtCl₆ ²⁻ precursor was completely removed upon the nucleophilic reaction with catechol groups, as confirmed by the absence of XPS Cl 2p signal (Figure S10, Supporting Information). In FT‐IR spectra, the C─O─H conformation in 1502 shifted to 1467 cm⁻¹, indicating the presence of heavier Pt atoms instead of hydrogen atoms (Figure S11, Supporting Information). The detailed electronic state and coordination environment of Pt atoms are verified by X‐ray absorption near‐edge structure (XANES), and Fourier transform extended X‐ray fine structure (FT‐EXAFS) of PtO₄‐DA, PtO₂, and reference Pt foil (Figure S12a, Supporting Information). The valence state of Pt can be quantitatively investigated by calculating the highest derivative points at the pre‐edge region of each sample with the relative contribution based on a reference Pt foil and PtO₂ (Figure S12b, Supporting Information).^{[ 13} , ²¹ , ^{32 ]} The calculated oxidation state of PtO₄‐DA is + 3.00, noticeably higher than Pt foil (0) but lower than PtO₂ (+4). The white‐line intensity also followed the same tendency, indicating the high charge loss of the Pt single atom. FT‐EXAFS results of Pt L₃‐edge in PtO₄‐DA further confirmed the single atomic structure with the dominant Pt─O peak at 1.65 Å, which shares a similar peak position to the representative Pt‐O coordination environment in PtO₂ (Figure S12c, Supporting Information). The absence of a typical broad Pt‐Pt peak appearing at 2.4 Å and Pt─O─Pt lattice peak at 2.7 Å further illustrates the unique configuration of PtO₄‐DA. Furthermore, the first‐shell EXAFS fitting precisely confirms the high yield of uniform PtO₄ atomic sites (Figure S12d, Supporting Information).

To investigate the detailed structure evolution of PtO₄@MX, Pt nanoparticles on MXene (Pt@MX) catalyst were prepared as a reference, by using a simple wet‐impregnation method with reducing agents (Figure S13, Supporting Information). As shown in the high‐resolution XPS Pt 4f spectra (Figure S14, Supporting Information), Pt@MX exhibits dominant metallic Pt peaks, as Pt atoms tend to agglomerate into metal nanoclusters under reductive environments. Conversely, PtO₄@MX only shows major 2+ oxidation state peaks at 75.5 and 72.2 eV for the isolated Pt atoms. DFT simulations are conducted to compare the binding energies for Pt─O covalent bonding and Pt─Pt metallic bonding (Figure S15, Supporting Information). The calculated energy for Pt‐O chelation is significantly lower than that for Pt metal nucleation, even at high Pt concentrations, well‐supporting the spontaneous formation of PtO₄ atomic sites. X‐ray diffraction (XRD) analysis verifies the absence of the Pt (111) metal peak in PtO₄@MX (Figure S16, Supporting Information). Furthermore, no structural deformation is observed between PtO₄@MX and pristine MXene, suggesting a negligible effect from PtO₄‐DA species in the structure of MXene. The slight downfield shift of (002) from 7.44° to 7.17° can be explained by the increase in MXene d‐spacing along with the atomic deposition of PtO₄‐DA (Figure 2c), which is only 0.44 Å for the uniform monolayer coating at the surfaces.^{[ 33 ]}

2.2. Identical PtO₄ Atomic Sites on Various 2D Substrates

This mussel‐inspired strategy can be employed to various substrates such as Ti₃C₂T_x MXene, N‐doped reduced Graphene Oxide (NrGO), and Molybdenum Disulfide (MoS₂) owing to the generic adhesion feature of dopamine.^{[ 22 ]} Those pristine 2D nanomaterials were prepared via typical top‐down liquid exfoliation methods (See Experimental Section for synthetic details & Figures S17–S19, Supporting Information for characterizations).^{[ 34} , ^{35 ]} Under the identical preparation conditions with PtO₄@MX, we successfully self‐assembled PtO₄@NrGO and PtO₄@MoS₂ (Figure S20, Supporting Information). The electrostatic attraction derived from the opposite charge states between dopamine and 2D nanomaterials efficiently prevented the aggregation of PtO₄‐DA, resulting in the uniform atomic‐scale stabilization on various surfaces (Figure 2a). The zeta‐potential measurements revealed that the 2D nanomaterials became neutralized after the electrostatic self‐assembly with the oppositely charged PtO₄‐DA (Figure S21, Supporting Information). The XRD patterns of PtO₄@NrGO and PtO₄@MoS₂ verified the absence of the Pt (111) peak and showed negligible structural changes compared to pristine substrates. No evidence of the broad amorphous peak typically associated with PDA_NP was detected (Figure S22, Supporting Information).

DFT simulations are conducted to calculate the adsorption energy of PtO₄‐DA on each substrate to clarify the primary mechanism for atomic dispersion (Figure S23, Supporting Information). The obtained high adsorption energies corroborate the strong driving force for electrostatic self‐assembly, well‐supporting the uniform dispersion of Pt single atoms across different substrates, including Ti₃C₂O₂, pristine graphene, and 2H‐MoS₂. The as‐prepared PtO₄@MX, PtO₄@NrGO, and PtO₄@MoS₂ were characterized using aberration‐corrected high‐angle annular dark‐field scanning transmission electron microscopy (HAADF‐STEM). High‐magnified HAADF‐STEM images in Figure 1b–d reveal distinct bright spots corresponding to high atomic number elements, unequivocally confirming the successful incorporation of uniformly distributed Pt single atoms across the entire substrates. The magnified inset images and corresponding fast Fourier transform (FFT) patterns in Figure 1e,i,m further delineate the detailed substrate structures. The MXene substrates exhibit distinct (100) crystallographic planes with the lattice fringes of 0.26 nm,^{[ 36 ]} while MoS₂ predominantly comprises the 1T phase structure characterized by the (100) basal planes with the lattice fringes of 0.27 nm.^{[ 37 ]} Notably, no amorphous rings indicative of dopamine cluster formation are observed. Elemental mapping images in Figure 1f,j,n highlight the core elements of each substrate, including Titanium (Ti), Carbon (C), and Molybdenum (Mo). Concurrently, Figure 1g,k,o confirms the presence of O, attributed to the surface functional groups on the substrates and PtO₄‐DA. Moreover, Figure 1h,l,p provides compelling evidence for the presence of Pt atoms, further substantiating the successful synthesis of PtO₄ atomic sites.

To investigate the detailed valence states and atomic environments of PtO₄@MX, PtO₄@NrGO, and PtO₄@MoS₂, XANES and FT‐EXAFS were performed. As shown in Figure 2d, the XANES spectra reveal the distinct features with varying intensities of white‐line peaks corresponding to the transition of core electrons from Pt 2p orbitals to 5d states, serving as an indicator of the Pt 5d orbital occupancy. Hence, the lower peak intensity of PtO₄@MX implies higher 5d‐band occupancy compared to PtO₄@NrGO and PtO₄@MoS₂, indicating stronger interfacial interaction between PtO₄ atomic sites and MXene substrates. The oxidation states quantitatively calculated from the pre‐edge line confirm the higher electron density of Pt in PtO₄@MX (+2.40) compared to PtO₄@NrGO (+3.20) and PtO₄@MoS₂ (+3.07), which follows the same tendency with the white‐line area intensities (Figure S24, Supporting Information). The XPS Pt 4f spectra in Figure S25 (Supporting Information) also show the negative shift of PtO₄@MX, consistent with the XANES results indicating a higher electron density in Pt. The positive shift of XPS Ti 2p spectra and negative shift of XPS N 1s spectra in PtO₄@MX further support these tendencies and suggest the facile electron transfer from MXene to PtO₄‐DA (Figure S26, Supporting Information). Interestingly, this change is not observed in other substrates, as demonstrated by the XPS Mo 3d and N 1s spectra of MoS₂ in Figure S27 (Supporting Information). Raman spectroscopy, a complementary surface‐sensitive analytical technique, revealed that PtO₄@MX was the only substrate exhibiting a redshift, indicative of reduced electron density at the Ti sites resulting from surface‐mediated electron transfer (Figure S28, Supporting Information).

Despite the pronounced differences in electronic states, the first‐shell EXAFS fitting (Figure 2e) precisely verified the identical PtO₄ atomic environments in PtO₄@MX, PtO₄@NrGO, and PtO₄@MoS₂, inferring that the interfacial charge transfer does not affect the atomic structure of the PtO₄ active site itself. The fitting results confirm that all samples share the same Pt─O coordination pathway of PtO₄‐DA with the same coordination number (CN) of 4.0, irrespective of the substrate type (Figure S29, Supporting Information). The fitting process employed a DFT‐optimized structure of planar PtO₄ atomic sites, and the low R‐factor values further validate the accuracy of the fitting results (Table S1, Supporting Information). The wavelet transform (WT), a powerful technique capable of simultaneously reflecting structural information in both k and R spaces, is employed to further substantiate the experimental results. In the WT contour plot, PtO₄@MX, PtO₄@NrGO, and PtO₄@MoS₂ exhibit the intensity maximum at the identical values in both R and k spaces, verifying a consistent Pt coordination environment across all the samples (Figure 2f). This high uniformity starkly contrasts with the WT contour plots of reference Pt foil and PtO₂ (Figure S30, Supporting Information), which distinctly highlight the presence of Pt─Pt and Pt─O─Pt configurations. Taken together, compared to other SAC studies, the PtO₄ atomic catalytic sites fabricated through molecular self‐assembly without external energy input exhibit an exceptionally high precision in its single atomic structure across different types of 2D substrates, free from random bonding or unexpected deformation. Furthermore, it is noteworthy that the electronic state of the Pt atomic sites exhibited variations depending on the species of self‐assembled substrates.

2.3. HER Performance of PtO₄@MX

The HER catalytic activity of PtO₄@MX was evaluated using a three‐electrode system in 0.5 m H₂SO₄ solution at ambient temperature. All the synthesized catalysts were uniformly drop‐cast onto carbon paper (CP) under identical conditions (Detailed procedures in the Experimental Section).^{[ 10 ]} As depicted in Figure 3a, the HER polarization curves of PtO₄ active sites exhibited substantial variations depending on the nature of substrates. Intriguingly, PtO₄@MX displayed an exceptionally low Tafel slope of 27.2 mV dec⁻¹ and the minimal overpotential of 33.1 mV to achieve a current density of 10 mA cm⁻², comparable to that of the 40 wt.% commercial Pt/C catalyst (Figure 3b). Such a high performance is even more pronounced when considering mass activity, an essential metric for objectively assessing HER performance. Specifically, PtO₄@MX achieved a remarkably high mass activity of 8.85 A mg⁻¹, exceeding those of PtO₄@NrGO, PtO₄@MoS₂, and PtO₄‐PDA_NP by factors of 10, 18, and 63, respectively (Figure 3c). The precise Pt loading in each sample was measured by inductively coupled plasma‐optical emission spectroscopy (ICP‐OES) (Table S2, Supporting Information).

Figure 3.

HER performance. a) HER polarization curves of PtO₄@MX, PtO₄@NrGO, PtO₄@MoS₂, PtO₄‐PDA_NP, and commercial Pt/C catalysts in 0.5 m H₂SO₄ solution. b) The Tafel slope originated from the LSV curves. (c) Calculated mass activity of the catalysts at the overpotential of 70 mV. d) Schematic illustration of interfacial synergistic hydrogen evolution reaction in PtO₄@MX. e) EIS Nyquist plots of PtO₄@MX, PtO₄@NrGO, PtO₄@MoS₂, and PtO₄‐PDA_NP with the equivalent circuit. f) Comparison plot of Tafel slope and overpotential at 10 mA cm⁻² for various Pt‐based catalysts on 2D nanomaterials. g) Chronoamperometry curve of PtO₄@MX obtained at the overpotential of −40 mV for 80 h. h) LSV curves of PtO₄@MX before and after 5000 LSV cycles at a scan rate of 5 mV s⁻¹. All the measurements were conducted in the identical conditions.

The substantial enhancement in mass activity underscores the synergistic effect at the hetero‐interface between the MXene substrate and PtO₄ active sites, which facilitates the Volmer–Tafel mechanism for HER under acidic conditions.^{[ 21} , ^{38 ]} Importantly, the observed catalytic enhancement is not attributed to the intrinsic activity of 2D substrates. Pristine MXene exhibited sluggish Tafel reaction and significantly higher overpotentials compared to NrGO and MoS₂, primarily due to its excessively high proton binding energy (Figure S31, Supporting Information). Hence, the superior performance is predominantly governed by the PtO₄ active sites, with their efficacy being highly dependent on their interaction with the underlying substrates. The markedly lower electrocatalytic activity of PtO₄‐PDA_NP and PtO₄‐PDA_NP@MX in Figure S32 (Supporting Information)can be attributed to the less efficient charge transport through the low‐conductive PDA_NP. The inferior catalytic performance of Pt@MX further indicates that Pt‐O active sites, rather than conventional Pt─Pt active sites, are more favorable for acidic HER. As illustrated in the comparison plot of Figure 3f, our PtO₄@MX catalyst exhibits competitive overpotential and Tafel slope values relative to the most of the previously reported Pt catalysts derived from similar 2D platforms, validating the potential of this self‐assembled catalytic system (Table S3, Supporting Information).

Figure 3e presents the Nyquist plots of PtO₄@MX, PtO₄@NrGO, PtO₄@MoS₂, and PtO₄‐PDA_NP that display significant differences in the charge transfer resistance (R_ct), which is a critical parameter for evaluating the underlying synergy mechanism.^{[ 39 ]} Given the identical electrochemical setups and conditions, the reduced semicircle diameter of PtO₄@MX indicates the synergistic interactions between MXene and PtO₄ with the rapid charge transfer and facilitated hydrogen evolution kinetics at the active sites (Table S4, Supporting Information). To further elucidate the underlying catalytic mechanisms, the electrochemical double‐layer capacitance (C_dl) was measured (Figure S33, Supporting Information) to estimate the electrochemical surface area (ECSA). The cyclic voltammograms (CV) at scan rates ranging from 10 to 70 mV s⁻¹ in a non‐faradaic region confirmed the high C_dl value of PtO₄@MX compared to other substrates, following the same tendency in Nyquist plots (Figure S34, Supporting Information).^{[ 40 ]} However, the discrepancy in mass activity far exceeds that of ECSA, indicating that ECSA is not the predominant factor dictating the overall HER performance. Along with the electrochemical impedance spectroscopy (EIS) results as well as the reduced valence state of Pt, the enhanced catalytic performance of PtO₄@MX should be attributed to the accelerated electron transfer from MXene to PtO₄ atomic sites. The resulting electric field from the rapid charge transfer can modulate the Fermi energy, further stabilizing the Pt 5d orbital. Specifically, the resultant high electron density in the Pt 5d orbital reduces the electron density of the O 2p orbitals, thereby weakening the binding energy between oxygen and protons. This facilitates more efficient proton transfer from O to the Pt active sites. Overall, the decreased proton desorption energy not only significantly enhances the overall Tafel process but also ensures a more efficient and sustained supply of protons to the catalytic sites, further optimizing the energy barrier (Figure 3d).

PtO₄@MX presented a good long‐term stability, attributed to the strong atomic‐level adhesion feature of dopamine. In a chronoamperometry test conducted at −40 mV, the catalyst sustained a stable current density of 20 mA cm⁻² for over 80 h without any noticeable degradation (Figure 3g). By contrast, the commercial Pt/C catalyst exhibited an inferior stability, with its current density beginning to decline rapidly after 40 h (Figure S35a, Supporting Information). The repeated cycling tests further validated the superior robustness of PtO₄@MX. After 5000 cycles, the overpotential of PtO₄@MX increased by only 3 mV (Figure 3h), whereas the commercial Pt/C catalyst exhibited a ≈12 mV increase, indicating a substantial degradation (Figure S35b, Supporting Information). HADDF‐STEM images of PtO₄@MX acquired after 5000 LSV cycle tests demonstrated that the Pt single atom configuration remained uniform and well‐dispersed, with no evidence of aggregation (Figure S36, Supporting Information). Further structure characterizations using XANES and FT‐EXAFS verified that the Pt atoms retained their Pt─O coordination even after extensive electrochemical cycling (Figure S37, Supporting Information).

2.4. Theoretical Investigations

DFT calculations were performed for the atomistic‐level catalytic mechanism of interfacial synergistic HER. Planar PtO₄ active sites were constructed on Ti₃C₂O₂ MXene, pristine Graphene, and 2H‐MoS₂ in order to optimize the atomic structure (See Experimental Section for computational details). As a primary factor, the electron transfer is highly dependent on the electron density at the interface. The large Fermi level difference between Ti₃C₂O₂ and PtO₄‐DA (Table S5, Supporting Information), along with the exceptionally high electron density at the Fermi level of Ti₃C₂O₂, strongly facilitates the electron transfer to the hetero‐interface.^{[ 41} , ^{42 ]} The Bader charge analysis clearly reveals electron accumulation at the hetero‐interface of PtO₄@Ti₃C₃O₂ (Table S6, Supporting Information). This electron accumulation induces a strong localized electric field at the hetero‐interface, which in turn effectively redistributes the electron density of PtO₄ (Figure 4a).^{[ 43 ]} While graphene also possesses a high electrical conductivity, its origin lies in the high charge carrier mobility via delocalized 2p orbitals. Graphene cannot effectively drive rapid electron transfer to the hetero‐interface due to its low charge carrier density, resulting in no significant electric field effect (Figure 4c). The interfacial electric field in PtO₄@MX increases the electron density of Pt (as demonstrated in the experimental characterizations and DFT calculations), while decreasing the electron density in O, making both more favorable for the Volmer–Tafel process in acidic HER. Although O is generally recognized as an excellent proton‐absorbing site, its binding energy is often excessively strong, impeding the proton desorption. However, the reduced electron density in the O 2p orbital effectively weakens the proton binding energy, allowing an easier proton transfer to the Pt active sites, and thereby significantly reduces the energy required for the Tafel step.

Figure 4.

Theoretical investigations. a,c) Charge difference plots with computational models and localized electric field distribution of PtO₄@MX, and PtO₄@Graphene. b,d) Electron density distribution at the VBM and CBM of PtO₄@MX, and PtO₄@Graphene in real space. e) Calculated PDOS of PtO₄@MX, and PtO₄‐DA with aligned Fermi level. f) Free energy diagrams of HER pathways for the Pt─O₄ active site with and without Ti₃C₂O₂ MXene. g,h) Comparison of the H₂ evolution kinetics with calculated energy barriers in PtO₄@MX and Pt (111).

Figure 4b provides the visualization of the electron density corresponding to the valence band maximum (VBM) and conduction band minimum (CBM) of PtO₄@MX. In our self‐assembled model system, the Fermi levels for the substrates and active sites are effectively separated. The presence of VBM states mainly on the PtO₄‐DA enables the upper active sites to be optimized for proton adsorption and a subsequent reduction by donating the electrons from the VBM, while the presence of CBM on the underlying substrate is tailored for an efficient electron transfer from the electrode to the catalyst. The intrinsic counterbalancing tendency is well recognized between the Fermi energy required for rapid electron transport via delocalized orbitals and the optimal hydrogen binding energy in localized orbitals.^{[ 44 ]} Nonetheless, the well‐defined functional separation of the two key features in our PtO₄@MX enables the unprecedented realization of high electrical conductivity and optimal proton adsorption sites simultaneously. When using graphene as a substrate, its CBM hinders the efficient electron transfer, while the VBM of PtO₄ active sites remains suboptimal for accelerating HER kinetics (Figure 4d). A similar result is reproduced in the DFT simulation of PtO₄@MoS₂, emphasizing the unique optimal feature of PtO₄@MX for an interfacial synergistic system (Figure S38, Supporting Information).

As illustrated in Figure 4e, the detailed changes in O 2p and Pt 5d orbitals were analyzed through the calculated projected density of state (PDOS) of PtO₄@MX and PtO₄‐DA, with their Fermi energy levels aligned. The detailed results exhibit the left shift of PDOS for both O 2p orbital and Pt 5d orbital, indicating the weaker binding strength of the proton intermediate (H^*) at both atomic sites. The effect of electron redistribution is even more distinct in the free energy diagram (Figure 4f), which clearly displays the detailed HER pathway in PtO₄ active sites. Without the Ti₃C₂O₂ MXene substrate, the calculated energies for 2H^* adsorption and generation of hydrogen intermediate (H₂ ^*) are significantly high, attributed to the strong binding energy of O─H^*. Thus, the markedly reduced free energy due to the facile electron redistribution clearly supports the synergistic effect of hetero‐interfacial interaction between the Ti₃C₂O₂ substrate and PtO₄ active sites. Interestingly, the hetero‐interfacial interaction also influences the electrocatalytic property of the MXene substrate, while allowing an additional reaction pathway (Sur‐O‐O₂, where Sur‐O refers to the Ti₃C₂O₂ surface as defined in Figure 4f) for HER. This unique characteristic mainly arises from the diminished proton adsorption energy at the Sur‐O (from −0.52 to −0.14 eV) and the reduced proton coverage on the Ti₃C₂O₂ surface, induced by the pre‐adsorption of PtO₄‐DA, which further optimizes the proton binding energy (Figure S39, Supporting Information). It can facilitate proton transfer from MXene to the PtO₄ active sites and establish abundant pathways for H^* supply. As a result of these synergistic effects, PtO₄@MX achieves a remarkably low energy barrier of 0.26 eV in the rate‐determining step of HER, as calculated in Figure 4g. This offers superior HER kinetics compared to Pt (111) with a higher energy barrier of 0.66 eV for the same rate‐determining step (Figure 4h). In conjunction with the experimental results, DFT simulation successfully confirms the fluent electron transfer, additional reaction pathway, and optimized energy barrier originating from the ideal interfacial HER, as illustrated in Figure 3d.

3. Conclusion

We have demonstrated an innovative molecular self‐assembly approach from dopamine precursors as an adhesive module to construct structurally precise and stable atomic catalytic sites. This intriguing mechanism facilitates the controlled and uniform monolayer deposition of monodisperse PtO₄ atomic sites across various substrate surfaces. Taking advantage of these unique features, we systematically investigated the beneficial mechanisms of hetero‐interfacial interactions in enhancing the HER kinetics, as solely driven by the physical nature of substrate materials. This mechanistic study is uniquely feasible in our self‐assembled system without any specific binding between the substrate and the active site, yet well‐securing their intimate interplay. Notably, a remarkable enhancement was observed in the mass activity of PtO₄ sites, up to 30‐fold increase from the commercial Pt/C reference. Specifically, the hetero‐interface between the electron‐rich MXene and the PtO₄ atomic sites readily facilitates the efficient electron transfer and builds up a strong localized electric field. This further increases the electron density in Pt 5d orbital, while decreasing the hydrogen binding energy in the O 2p orbital, and thereby optimizes the hydrogen adsorption/desorption at the PtO₄ sites. These findings highlight the critical role of noncovalent environmental interaction upon electrocatalytic activity, which has been difficult to systematically explore thus far, and also open‐up a valuable pathway toward superior catalytic performance by judiciously pairing atomic catalytic sites with the substrates of distinct synergistic features.

4. Experimental Section

Synthesis of Delaminated Ti₃C₂T_x MXene

Ti₃AlC₂ MAX powder (particle size: 40 µm) was acquired from Carbon‐Ukraine Ltd (Ukraine). Hydrofluoric acid (HF, 48%) was purchased from Alfa Aesar (USA). Hydrochloric acid (HCl, 37%) and Lithium chloride (LiCl, 99%) were supplied by Sigma–Aldrich (USA). Briefly, 3.0 g of MAX powder was added to a 60 mL of 6:3:1(volume ratio) mixture of 12 m HCl, deionized (DI) water, and 48 wt.% HF for the etching process. After fully eliminating aluminium layers by stirring with etchant at 400 rpm for 24 h at 36 °C, the etched multilayer MXene was washed 4–5 times with DI water to remove the HF and remaining metal ions. The resulting sediment was then dispersed in 150 mL of 0.5 m LiCl solution and stirred at 400 rpm for 4 h at room‐temperature. After the delamination process, the MXene/LiCl mixture was repeatedly centrifuged with DI water (6000–10,000 rpm) until the pH of the supernatant became 5–6. In the final washing cycle, 1 or 2‐layer MXene solution can be collected as a supernatant after 20 min of centrifugation at 4000 rpm.

Preparation of NrGO

NRGO was supplied by Graphene All (Republic of Korea). In order to maximize electrical conductivity while preserving excellent dispersibility, Graphene Oxide (GO) was reduced via ammonia solution reduction rather than conventional thermal methods or plasma reduction. Furthermore, to selectively obtain highly dispersible flakes, the NrGO dispersion was left undisturbed for 1 h, allowing the sedimented particles to settle, after which only the supernatant was carefully collected for further experiments.

Synthesis of Delaminated MoS₂

Bulk MoS₂ and n‐Butyllithium were purchased from Sigma–Aldrich (USA). 0.3 g of bulk MoS₂ was added to 6 mL of 2.5 m n‐Butyllithium in hexane solution, and stirred at 300 rpm for 48 h at 40 °C. The fully intercalated MoS₂ suspension was filtered over a Millipore membrane and repeatedly washed with high‐purity hexane to remove the residual n‐Butyllithium. The aforementioned procedures were carried out in an N₂‐filled glove box. Then, the sediments were centrifuged several times with DI water at 17 000 rpm in order to fully remove the hexanes. In the final washing cycle, 1 or 2‐layer MoS₂ aqueous solution can be collected as a supernatant after 30 min of centrifugation at 3000 rpm.

Synthesis of PtO₄‐DA

Chloroplatinic acid (H₂PtCl₆) and dopamine hydrochloride (DA) were obtained from Sigma–Aldrich (USA). To prepare PtO₄‐DA, 89.2 mg of H₂PtCl₆ and 200 mg of DA were added to 50 mL deoxygenated DI water solution, and stirred at 400 rpm for 4 h in acidic conditions. It is important to note that the commonly used Tris buffer solution should not be used in order to minimize the possibility of rapid dopamine polymerization during the chelation reaction. Then, 1 mL of 1 wt.% NH₄OH was added dropwise to slowly initiate the self‐oxidation of dopamine, simultaneously facilitating the reduction of chelated platinum ions. All the procedures were conducted in room‐temperature with ambient solution conditions. After 1 h stirring and 10 min sonication, the resulting solution was washed several times at 17 000 rpm with ethanol and DI water to separate the fabricated PtO₄‐DA from excessive dopamine monomers and ion impurities. In the final washing cycle, atomically dispersed PtO₄‐DA can be collected as a supernatant after 10 min of centrifugation at 3000 rpm.

Synthesis of PtO₄@MX, PtO₄@NrGO and PtO₄@MoS₂

To prepare PtO₄@MX, a predetermined amount of PtO₄‐DA was added to the MXene aqueous solution with the concentration of 0.1 mg mL⁻¹. The pH condition of MXene was controlled to 3.5 by using 0.5 m HCl to induce the opposite zeta potential between MXene and PtO₄‐DA. Since the mixture was not a buffered solution, its pH could fluctuate within the range of 3.0–3.5 over time. The minus‐charged 2D MXene flakes could be readily self‐assembled with plus‐charged PtO₄‐DA during the 2 h of stirring at 400 rpm. After three times of repeated centrifugation and decanting, PtO₄@MX was obtained as a resulting sediment. The equal procedures and conditions were required for the fabrication of PtO₄@NrGO and PtO₄@MoS₂ as a predetermined amount of different 2D nanomaterials could electrostatically self‐assemble with PtO₄‐DA. Facile pH control was essential for the uniform dispersion of PtO₄‐DA on substrates. All samples were freeze‐dried to obtain the powder and stored in a vacuum desiccator to minimize oxidation.

Synthesis of Pt@MX

A predetermined amount of H₂PtCl₆ was added to the MXene aqueous solution, stirred at 400 rpm for 4 h. Then, an excessive amount of 1 mg mL⁻¹ NaBH₄ (strong reducing agent) solution was added dropwise to fully reduce the platinum precursors. After repeated centrifugation at 10 000 rpm and decanting, Pt@MX was obtained.

Synthesis of PtO₄‐PDA_NP and PDA_NP

89.2 mg of H₂PtCl₆ and 200 mg of DA were added to 50 mL deoxygenated DI water solution, and stirred at 400 rpm for 4 h in acidic conditions. Then, 50 mL of 0.02 m Tris buffer solution was added, and the pH was adjusted to 8.5 using a 0.5 m NaOH solution. The mixture was stirred at 400 rpm for 24 h under ambient conditions, allowing the formation of PDA nanoparticles. After repeated centrifugation at 10 000 rpm and decantation, PtO₄‐PDA_NP was obtained as the sediment. PDA_NP was synthesized following the same procedure, with the exception that H₂PtCl₆ precursors were not used.

Characterizations

Zeta potential measurements were conducted in aqueous solution with the same concentration of 0.01 mg mL⁻¹ using ELSZ‐2000 (Otsukael). All the pH values were measured accurately using AB33PH (OHAUS) after calibration. The XRD spectra were obtained using a SmartLab (RIGAKU) diffractometer with a scanning step of 0.01° and a scan rate of 5° min⁻¹. High‐resolution Raman was performed using LabRAM HR Evolution Vis–NIR (HORIBA) with a wavelength laser of 514 nm. XPS spectra were acquired using K‐alpha equipment (Thermo VG Scientific) with an aluminum X‐ray source (1486.7 eV). All the binding energy values were compared after applying carbon correction based on the C 1s spectrum to compensate for peak shifts caused by charging effects. FT‐IR was conducted using Nicolet iS50 (Thermo Fisher) without pellet preparation. The thickness of 2D nanomaterials was measured by INNOVA‐LABRAM HR800 (Horiba Jobin Yvon), and the Pt content of the catalysts was measured using ICP‐OES 720 (Agilent). The morphological characterizations and EDS analysis were investigated by Spectra Ultra (Thermo Fisher), which was a High‐Resolution Double Cs corrected TEM, at 300 kV accelerating voltage. BET isotherms were obtained through N₂ adsorption–desorption measurements using a 3Flex (Micromeritics). The BET‐specific surface area was calculated based on adsorption data within a selected relative pressure range of P/P₀ = 0.05–0.3. In order to verify the atomic structures of the PtO₄ catalytic sites, X‐ray absorption fine structure (XAFS) measurements at Pt L₃‐edge were performed using the 8C beamline of Pohang Accelerator Laboratory (PLS‐II, 3.0GeV0, South Korea). Pt foil was used as a reference sample for the precise energy calibration. The obtained XANES and EXAFS data were analysed using ATHENA software, following standard analytical procedures. The fitting results were obtained from the k³‐weighted EXAFS oscillation using the ARTEMIS module, incorporating Feff calculations for the optimized PtO₄ structure.

Electrochemical Measurements

All electrochemical analyses were carried out on an electrochemical workstation (VSP, Bio‐logic) at room‐temperature using a three‐electrode system with a Pt wire as the counter electrode, a 3.0 m Ag/AgCl electrode as the reference electrode, and carbon paper as the working electrode. 0.5 m H₂SO₄ was used as an electrolyte, and the reference electrode was calibrated to reversible hydrogen electrode (RHE), E _RHE  =  E _Ag/AgCl + 0.059  ×  pH + 0.210 V. The electrode ink was prepared by mixing 5 mg of catalyst with 1 mL of solvent (1:1 v/v water/ethanol) and 10 µL of Nafion (10 wt.%), followed by grinding and sonication. Subsequently, 200 µL of the prepared ink was precisely deposited onto the 1 cm² carbon paper (Thermo Scientific) and subjected to vacuum drying. All the catalytic inks, including the 40 wt.% commercial Pt/C, were fabricated using the identical preparation procedures and drop‐cast in equal amounts to ensure consistency. The linear sweep voltammetry (LSV) curves were measured at a scan rate of 5 mV s⁻¹, with all data calibrated by 95% iR‐compensation within the electrochemical cell. The EIS spectra were performed at an overpotential of 10 mV, in the frequency range from 100 kHz to 10 mHz. Any possibility of Pt contamination was eliminated by conducting identical experiments using a carbon rod counter electrode as a control.

Tafel Slope, Mass Activity, and Double Layer Capacitance Calculations

The Tafel slopes were determined by performing linear fitting on the plot of logarithmic current density versus overpotential. Tafel equation:

$$\eta =\mathrm{b}\log \left(\mathrm{j}\right)+\mathrm{c}$$ (1)

where b is the Tafel slope, η is the overpotential (mV), j is the measured current density (mA cm⁻²), and c is the intercept. The mass activity of the catalysts was calculated using the following equation:

$$\mathrm{mass}\mathrm{activity}=\mathrm{j}/{\mathrm{m}}_{\mathrm{Pt}}$$ (2)

where j is the measured current density (mA cm⁻²), and m_Pt is the platinum loading (mg cm⁻²). Cyclic voltammograms (CV) were performed in the potential range of 0.2–0.3 V (vs RHE) with the scan rates from 10 to 70. mV s⁻¹ to determine the double layer capacitance (C_dl) of the catalysts. C_dl was calculated using the following equation:

$${\mathrm{C}}_{\mathrm{dl}}=\frac{\mathrm{\Delta }\mathrm{j}/2}{\mathrm{\Delta }\nu }$$ (3)

where Δj is the difference of the anodic and cathodic currents (j_a–j_c) at 0.25 V versus RHE and Δν is the scan rate difference.

Computational Details

A self‐consistent approach within the density functional theory (DFT) framework, as implemented in the Vienna Ab initio Simulation Package (VASP),^{[ 45 ]} was employed to model and optimize a monolayer of Ti₃C₂O₂ along with a PtO₄‐DA adsorbed Ti₃C₂O₂ monolayer. The electronic wave functions were described using the projector‐augmented wave (PAW) pseudo‐potential formalism,^{[ 46 ]} while generalized gradient approximations (GGA) were utilized to approximate the electron‐electron correlation functionals.^{[ 47} , ^{48 ]} The electronic optimizations were performed with a convergence criterion of 10⁻⁶ eV, and ionic relaxations were iteratively executed until the absolute forces between all atoms were reduced to less than 0.01 eV Å⁻¹. The monolayer model of Ti₃C₂O₂ was constructed to be periodic along the x‐ and y‐directions, with a 20 Å vacuum in the z‐direction to prevent interactions between periodic images. The cell parameters of the monolayer models were not optimized, but atomic positions were fully relaxed based on the aforementioned criteria. For structural relaxation, the Brillouin zone was sampled with a 16 × 16 × 1 Γ‐centered k‐point mesh, while a 3 × 3 × 1 k‐point grid was used for the 5 × 5 × 1 supercell. The density of states (DOS) calculations were performed via non‐self‐consistent calculations, utilizing charge density and wavefunctions obtained from self‐consistent calculations.

Gibb's free energy calculation: To evaluate the HER catalytic activity of various active sites on the PtO₄‐DA adsorbed Ti₃C₂O₂ monolayer, the computational hydrogen electrode (CHE) model, developed by Nørskov et al.^{[ 49 ]}was employed. The Gibbs free energy change (ΔG) for hydrogen adsorption was calculated using the expression:

$$\mathrm{\Delta }{\mathrm{G}}_{{\mathrm{H}}^{\ast }}=\mathrm{\Delta }{E}_{{\mathrm{H}}^{\ast }}+\mathrm{\Delta }{E}_{\mathrm{ZPE}}-T\mathrm{\Delta }{S}_{\mathrm{H}}$$ (4)

where ΔE, T, ΔS, and ZPE denote the change in enthalpy upon hydrogen adsorption (computed from the DFT total energy of the system with and without an adsorbed H atom), the absolute temperature, the entropy changes of the adsorbed hydrogen, and the zero‐point energy correction, respectively. Standard entropy values for molecular H₂ were sourced from the NIST database, while those for adsorbed hydrogen atoms were derived from their vibrational frequencies.

Binding Energy Calculation: To compare the binding energies of covalent bonding and Pt─Pt metallic bonding, the Pt─Pt bond strength was systematically varied by adjusting the Pt concentration, ranging from a Pt‐poor condition, where a low Pt concentration leads to the formation of Pt₂ dimers, to a Pt‐rich condition, where a high Pt concentration enables the formation of bulk Pt particles. The binding energy (BE) is determined using the following equation:

$$\mathrm{BE}=E\left(\mathrm{Str}+\mathrm{Pt}\right)-E\left(\mathrm{Str}\right)-\mu \left(\mathrm{Pt}\right)$$ (5)

where E (Str+Pt) represents the total energy of the structure with a Pt atom, E (Str) denotes the total energy of the pristine structure (without Pt), and µ (Pt) corresponds to the chemical potential of a Pt atom. Here, the chemical potential of Pt in the Pt‐poor condition (Pt dimer state) was set to 0 eV.

Conflict of Interest

The authors declare no conflict of interest.

Author Contributions

Y.H.Y. led the project and mainly carried out the experiments. K.J. carried out the DFT calculations and commented on the mechanistic understanding. G.S.L. contributed to the structure characterizations and commented on the experiments. J.B. and S.P.S. contributed to the electrochemical measurements. C.W.L., C.C., and J.B.K. contributed to the material synthesis. H.L. commented to the interpretation of the dopamine mechanism. S.O.K. supervised and wrote the paper.

Supporting information

Supporting Information

Yoon Y. H., Jeyakumar K., Lee G. S., et al. “Mussel‐Inspired Self‐Assembly of PtO₄ Atomic Catalysts for Interfacial Synergistic Hydrogen Evolution.” Adv. Sci. 12, no. 37 (2025): 12, e07807. 10.1002/advs.202507807

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.