Citation

Yu Mulin, Liu Shuo, Tang Yufeng, Lu Guoqiang, Liu Linbo, Sui Pengfei, Fu Xianzhu, Liu Subiao, Sun Yifei, Luo Jingli. Mediating electron delocalization of surface palladium atoms by diethylamine ligand for efficient CO₂ electroreduction[J]. Chin. Chem. Lett., 2026, 37(3): 111958  

Mulin Yu^a, Shuo Liu^a, Yufeng Tang^a, Guoqiang Lu^a, Linbo Liu^a, Pengfei Sui^b, Xianzhu Fu^c, Subiao Liu^a,*, Yifei Sun^d, Jingli Luo^b,c     

The sharply growing CO₂ concentration in the atmosphere has drawn particular attention worldwide due to the severe climate issues caused by the global warming, which makes an appeal for reducing CO₂ emission from the source, and ultimately reaching the carbon-neutral synthesis from CO₂. When powered with intermittent renewable energy sources, electrochemical CO₂ reduction reaction (CO₂RR) to produce C₁ products (e.g., CO and HCOOH) stands out from various CO₂ conversion routes [1–3], especially on palladium (Pd)-based electrocatalysts. Different from other noble metals (e.g., Au, Ag, Pt, Ru), Pd nanostructures normally display a comparable current density (j) at a desirable overpotential (η), and a tunable selectivity toward CO or HCOOH formation [4,5]. However, the unavoidable CO poisoning caused by the high adsorption energy CO on Pd(111) surface would result in the accumulation of *CO and hindrance of further C–O bond breakage [6,7]. Moreover, the difficult CO desorption hinders CO₂ adsorption on the surface because of the lack of adsorption sites, and the prolonged CO coverage induces surface reconstruction and changes in electronic configuration of Pd-based electrocatalysts, ultimately leading to electrocatalyst deactivation and stability decay [8,9]. In addition, due to the severe agglomeration of Pd nanoparticles (Pd NPs) during long-term operation, and the formation of PdO_x caused by the reaction between the oxygen species in the electrolyte and the surface Pd atoms [10,11]. Pd-based electrocatalysts are also prone to deactivation and structure reconstruction during CO₂RR, which, as a result, lead to a poor stability. To this end, various design strategies, e.g., crystal facet control [12,13], assembling heterostructures [14], alloys [15], metal-based compound nanohybrids [16], surface modification [5], a different anoded-coupled reaction systems [17–19] have been proposed to improve CO₂RR performance. However, a concomitant issue inevitably occurs that the long-chain alkyl compounds/polymers (e.g., polyvinylpyrrolidone (PVP), oleylamine (OAm) and cetyltrimethyl ammonium bromide (CTAB)) only act as an end-blocking agent for the synthesis of colloids or nanoparticles with a controllable size and/or shape, which are difficult to be completely removed, and sometimes even block the exposure of electroactive sites on material surface. Upon acting in contravention, retaining the residue on the surface or exchanging it with certain ligands holds the potential to tune the physiochemical properties of material surface at molecular level, thereby affecting the electronic states and the freedom degrees of active centers, which ultimately regulate the reaction pathways for various target products toward CO₂RR [20,21].

In fact, a consensus on the ligand effect has been reached that the surface ligands tune the CO₂RR performance mainly from two angles, i.e., the steric effect caused by different branch chain number, chain length and molecule spacing, and the electronic effect induced by various electron-donating and electron-withdrawing groups. To be specific, the ligands can not only spatially restrict a particular conformation of adsorbed intermediates and/or prevent reactants from contacting the material surface, but also modulate the electron density via the Lewis acid-base interaction to tune the material electronic structure. Previous studies have reported using macromolecule or long-chain organic ligands to promote the interplay between intermediates and metal active sites by constructing rigid organic-modified layers with an intermolecular van der Waals interaction. However, the severe coverage of ligands on metal surface blocks CO₂ accessibility to active sites. In the meantime, due to the lack of strong intermolecular interactions between various small organic ligands, building rigid channels on metal surface only partially but insufficiently regulates H₂O and dissolved CO₂ diffusions, and intermediate adsorption and/or desorption behaviors toward efficient CO₂RR.

In recent years, amine-functionalized metal-organic frameworks (MOFs) and molecular complexes have demonstrated considerable potential in CO₂RR through various methods (e.g., tailoring central metal node [22], changing functional group [22,23], adjusting local electronic distribution [24] and forming hydrogen bonding networks [25]) to control the adsorption behaviors of intermediates. Inspired by these molecular-level design principles, our work extends the concept to nanostructured catalysts by grafting alkylamine ligands onto the surface of Pd NPs. Notably, amino [26], pyridinyl [27], and imidazole [28] could selectively construct metal-ligand interfaces, which further interact with reactants and/or intermediates with the aid of C–N bonds and/or H-bonds. Moreover, due to the electronegativity difference between the central heteroatoms in the ligand and the adjacent atoms in the surface metal, the electronic distribution on metal surface would be rearranged, thus leading to the shift in the d-band center of surface metal. Inspired by this, the diethylamine (DEA), as a small amine molecule, has been inferred effective to promote CO₂RR by strongly adsorbing on the metal surface through forming covalent bonds and/or coordinated covalent bonds with surface metal atoms induced by the lone pair electron of its N atom. More importantly, DEA, also acting as an intermediate ligand, is widely employed to exchange or replace other macromolecular long chain cap ligands (e.g., PVP, OAm, and CTAB) that are difficult to be removed from metal surface [29]. It is worth noting that the difference in spatial configuration between DEA and OAm could change the adsorption configurations of CO₂ and its intermediates on surface Pd atoms. Moreover, the long alkyl chain of OAm limits its electron donation ability, whereas DEA with two short branched-chains is easier to transfer more electrons to surface Pd atoms, thus weakening the adsorption of *CO and improving the selectivity of CO [30,31]. Thus, the "one stone killing two birds" postulation addresses the residual issue of end-blocking agents on metal surface for specific nanostructure synthesis in organic media, and synchronously introduces the surface molecular functionalization, i.e., ligand effect, to induce electron redistribution on N-mediated surface metal atoms and their associated electronic structure tuning for efficient CO₂RR.

To validate it, DEA was assembled on the surface of ultrasmall Pd NPs (DEA-Pd NPs) by exchanging the residue OAm ligands on Pd NPs (OAm-Pd NPs). The DEA-Pd NPs achieved a CO selectivity of close to 100% in a wide potential window of −1.3 ~ −0.7 V. The density functional theory (DFT) calculations revealed that the donating effect of N in DEA delocalized the electrons on the surface Pd atoms to the lower energy orbital, resulting in a negative shift of the d-band center, which weakened the adsorption of *COOH and *CO intermediates. Moreover, the H-bond formed between the N-H in DEA and the *COOH increased the adsorption energy of *COOH. This targets in tuning the binding energies of various intermediates during CO₂RR, which fundamentally addresses the issues of difficult *COOH adsorption and *CO desorption, and breaks the unfavorable linear scaling relationship of their simultaneous increase or decrease.

The introduced ligands on surface metal atoms inevitably influence the gas-liquid (CO₂–H₂O) balance adjacent to the active site [32], which regulates the associated CO₂ mass transfer and the final CO₂RR kinetics. To explore the effects of DEA on CO₂ and H₂O diffusion behaviors, MD simulations were first conducted to obtain the time sequence of typical snapshots of H₂O and CO₂ diffusion (Figs. S1a–c in Supporting information). Clearly, different from the random transportation behaviors on Bare-Pd NPs surface, H₂O and CO₂ diffusions were gradually limited when DEA was uniformly grafted on Pd NPs, and were completed prevented at 400 ps (Fig. 1a–c). This was further quantitatively determined by the diffusion coefficient (DC), as judged from the slope in Fig. 1d. The DC(CO₂) of DEA-Pd NPs (0.272 × 10⁻² Ų/ps) was smaller than that of Bare-Pd NPs (9.475 × 10⁻² Ų/ps), while the DC(H₂O) of Bare-Pd NPs (0.163 × 10⁻² Ų/ps) was larger than that of DEA-Pd NPs (0.143 × 10⁻² Ų/ps). This indicated that DEA limited both CO₂ and H₂O diffusions on Pd NPs surface, with a larger impact on CO₂ due to the larger diameter (d = 3.3 Å) of CO₂ than that (d = 2.8 Å) of H₂O. Notably, although CO₂ exhibited a stronger hindrance effect for diffusion on the surface of DEA-Pd NPs due to the larger molecule size than H₂O, it was not the sole determinant for diffusion behavior. Based on this, the radial distribution functions (RDFs) of CO₂ and H₂O were obtained to further reflect the density of CO₂ and H₂O near Pd NPs surface, i.e., the accessibility to Pd NPs surface. Apparently, the density of CO₂ on Bare-Pd NPs and DEA-Pd NPs, reflected by g(r) values in Fig. 1e, were higher than that of H₂O. Nevertheless, the g(r) values of CO₂ and H₂O on DEA-Pd NPs were lower than those on Bare-Pd NPs, indicating that DEA reduced the probability of CO₂ and H₂O reaching the surface of Pd NPs.

Figure 1

	Download: JPG larger image
Fig. 1. Typical snapshots of H2O and CO2 diffusion on DEA-Pd NPs at (a) 0 ps, (b) 200 ps and (c) 400 ps. (d) Fitted MSD versus time for H2O and CO2 diffusion. (e) RDFs of H2O and CO2. (f) Contact angles of Bare-Pd NPs and DEA-Pd NPs.

By comparing the reduction degree of Pd-O (CO₂) and Pd-O (H₂O) (i.e., the value of Δg(r)), it was found that Δg(r)CO₂ was significantly smaller than Δg(r)H₂O, proving that the hindrance degree of DEA toward CO₂ to Pd NPs surface was smaller than that toward H₂O. It is also worth noting that the V-shaped H₂O structure holds the potential to form strong H-bonds, making it more susceptible to be trapped by DEA within the H-bond network, leading to a slower diffusion [33,34]. In contrast, the nonpolar CO₂ molecule does not engage in H-bonding with DEA, which causes a less diffusion hindrance tread in reality, and an easier CO₂ diffusion to the DEA-Pd NP surface. Moreover, the –NH– and –C₂H₅ groups in DEA could induce a stable adsorption layer on the Pd surface and create a hydrophobic microenvironment [35,36], which reduced the affinity of Pd NPs for polar molecules, thus limiting the diffusion of H₂O molecules. In addition, the diffusion of H₂O molecule on DEA-Pd NPs was restricted by the solvation effect, since the polar H₂O molecule could form H-bonds with surrounding H₂O molecules or other polar molecules, and the H-bond network traps H₂O molecule, and restricts its diffusion. In contrast, nonpolar CO₂ exhibits a weaker solvation effect, and diffuses more freely in solution [37,38]. It was further verified by the larger contact angle of 106.1° on DEA-Pd NPs relative to that of 89.9° on Bare-Pd NPs (Fig. 1f, Figs. S2 and S3 in Supporting information). Notably, CO₂ primarily diffused in its dissolved form (HCO₃⁻ or CO₂ hydrate) on a hydrophilic surface, since the hydrophilic surface tended to form a thick water layer, which prolonged the diffusion pathway of the dissolved CO₂ and slowed its mass transfer rate. However, it could diffuse in the form of gas molecules on a hydrophobic surface, because the hydrophobic surface could reduce the thickness of water layer, allowing gaseous CO₂ to approach to the catalyst surface more easily, and consequently improving its diffusion efficiency [39–41]. In addition, instead of rapidly diffusing away from the surface, the DEA-functionalized Pd surface prolonged the residence time of CO₂ on Pd atoms since the "molecular network" formed by DEA can effectively confine CO₂ within it, thereby accelerating the subsequent CO₂RR.

DEA-Pd NPs were synthesized by a ligand exchange method [29], as schemed in Fig. 2a. Specifically, palladium(Ⅱ) acetylacetonate was reduced to Pd NPs in OAm solution, followed by the addition of an appropriate amount of DEA solution dispersed in isopropyl alcohol to exchange the residue OAm on Pd NPs. Notably, DEA was a branched amine with stronger Lewis's basicity, and the lone pair electrons of its N atom can more effectively coordinate with the empty orbitals of Pd, forming a more stable Pd–N bond, thereby enabling the original end-capping ligands OAm to be effectively exchanged and removed. The transmission electron microscopy (TEM) image (Fig. 2b and Fig. S4 in Supporting information) confirmed that DEA-Pd NPs, OAm-Pd NPs and Bare-Pd NPs displayed a highly monodispersed spherical morphology on carbon black, with an average size of 5.16 ± 0.46 nm (Fig. 2c and Fig. S5 in Supporting information), revealing that the ligand introduction had no impact on the size of Pd NPs, thus eliminating the size effect interference on CO₂RR performance. The high-resolution TEM (HRTEM) image (Fig. 2d) and X-ray diffraction (XRD) pattern (Fig. S6 in Supporting information) confirmed the crystallinity nature of DEA-Pd NPs, and the lattice fringe spacing of 0.224 nm well corresponded to the face-centered cubic Pd(111) (JCPDS No. 01-1201). Besides, the energy-dispersive X-ray (EDX) elemental mappings presented a uniform distribution of the characteristic N element in DEA (Fig. 2e) and OAm (Fig. S7 in Supporting information) on Pd NPs. To further confirm the existing form of N in DEA with surface Pd atoms, FTIR was employed, as shown in Fig. 2f and Fig. S8 (Supporting information). Clearly, the as-synthesized OAm-Pd NPs before ligand exchange showed prominent vibrational peaks at 2923 and 2842 cm⁻¹, corresponding to the symmetric stretching C–H and asymmetric C-H of OAm, respectively. After ligand exchanging with DEA, the two vibrational peak intensities of DEA-Pd NPs were weakened, and well-resolved peaks belonging to –NH– group at 668 cm⁻¹ were remarkedly strengthened, implying the successful ligand exchange of long-chain OAm by DEA [42,43].

Figure 2

	Download: JPG larger image
Fig. 2. (a) Scheme for synthesizing DEA-Pd NPs. (b) TEM image, (c) size statistics, (d) HRTEM image, and (e) EDS mappings of DEA-Pd NPs. (f) FTIR spectra, XPS spectra of (g) Pd 3d and (h) N 1s of Bare-Pd NPs and DEA-Pd NPs, respectively.

X-ray photoelectron spectroscopy (XPS) was then performed (Fig. S9 in Supporting information), and the presence of Pd²⁺ peak in the XPS spectrum of Pd 3d (Fig. 2g and Fig. S10 in Supporting information) was attributed to the partial surface oxidation after exposure to air, leading to the formation of PdO or Pd(OH)₂ species on the surface [44,45]. Notably, the Pd 3d spectrum was negatively shifted by about 0.30 eV on DEA-Pd NPs as compared to that on Bare-Pd NPs. This is because the strong Pd–N interaction caused more electrons to be transferred from the amino ligand to the surface Pd atom, thus increasing the electron density of the surface Pd atoms [46–49]. Notably, a slight negative shift (~0.15 eV) in the Pd 3d binding energy was observed after ligand exchange, confirming an increased electron density on Pd from stronger electron donation by DEA. Again, the unique N 1s signal of DEA-Pd NPs and OAm-Pd NPs were observed in Fig. 2h and Fig. S11 (Supporting information), and the peak of the –NH– typically appeared in the binding energy range of the N 1s orbital, but its exact position was mainly depended on the chemical environment. Apparently, the peak at 400.4 eV in N 1s was attributed to the Pd-N bond, while the peak at 398.4 eV corresponded to the C–N bond [50–52], verifying that –NH– in DEA formed covalent bonds with surface Pd atoms. It is worth noting that the area of C–N bond in OAm-Pd NPs was significantly larger than that of Pd–N bond, while the opposite was true for DEA-Pd NPs, which indicated that DEA had successfully exchanged OAm. In addition, we further analyzed the C 1s spectrum (Fig. S12 in Supporting information). It was found that the proportion of C–C bond in OAm-Pd NPs (60.52%) was higher than that in DEA-Pd NPs (44.48%), while the proportion of C–N bond was higher in DEA-Pd NPs, since the long chain OAm possessed more C atoms. Obviously, the abovementioned results indicated that a large amount of OAm has been successfully exchanged by DEA.

CO₂RR activities on DEA-Pd NPs, OAm-Pd NPs, Bare-Pd NPs, OAm/C, DEA/C and C were evaluated, and products were quantified by using online GC for CO and H₂, and IC for HCOO⁻ (Fig. S13 in Supporting information). All reported potentials are referred to the reversible hydrogen electrode (RHE) unless otherwise noted. The linear sweep voltammetry (LSV) curves (scan rate = 20 mV/s) were depicted in Fig. 3a and Fig. S14 (Supporting information), which could roughly examine their potential electrocatalytic activity because j included the contribution from CO₂RR and HER. Clearly, DEA-Pd NPs showed an enhanced j of 36.5 mA/cm² at −1.4 V, whereas Bare-Pd NPs only reached 20.9 mA/cm². Following this, the potential-dependent product distribution on DEA-Pd NPs was carried out, and the results illustrated that DEA-Pd NPs exhibited an extremely high CO selectivity with a nearly 100% FE_CO over a wide potential window from −0.7 V to −1.3 V. Even at an ultralow potential of −0.6 V, DEA-Pd NPs still reached a FE_CO of around 90% (Fig. 3b), whereas OAm-Pd NPs and Bare-Pd NPs only achieved a FE_CO of 74.7% and 68.6%, respectively (Fig. S15 in Supporting information). In contrast with the Bare-Pd NPs, the amino ligand on OAm-Pd NPs also worked for CO₂RR. However, due to the steric effect, the long chain of OAm restricted the proximity of CO₂ to the surface-active site to some extent, which well explained the limited performance improvement of OAm-Pd NPs relative to DEA-Pd NPs. To confirm that the enhanced CO₂RR performance intrinsically originates from the chemically-bonded DEA on surface Pd atoms, i.e., Pd–N bond, rather than the residual OAm or DEA adsorbed on carbon black, CO₂RR activity on OAm/C and DEA/C was further examined. As expected, the CO FE on DEA/C remained below 20% across the entire potential range, while OAm/C performed almost no CO₂RR activity. Importantly, the turnover frequency (TOF) results indicated that DEA-Pd NPs possessed a TOF value higher than DEA/C (Fig. S16 in Supporting information), and the same phenomenon also occurred on OAm-Pd NPs and OAm/C, implying that DEA and OAm could maximize the activity of Pd NPs only when chemically bonded with Pd NPs, and its promotion effect became more pronounced at more negative potentials. Besides, through integrating the j obtained at various potentials (Fig. S17 in Supporting information) and FE_CO, the CO partial j (j_CO) was plotted in Fig. 3c and Fig. S18 (Supporting information), which displayed the exact j contribution from CO₂RR. Clearly, the j_CO of DEA-Pd NPs was much higher than those of OAm-Pd NPs and Bare-Pd NPs, confirming its remarkable CO₂RR performance.

Figure 3

	Download: JPG larger image
Fig. 3. (a) Cathodic LSV curves, (b) FECO, (c) jCO, and (d) EEs of DEA-Pd NPs and Bare-Pd NPs.

Although typical electrochemical indexes, such as j and FE, have confirmed the outstanding CO₂RR performance on DEA-Pd NPs, energy efficiency (EE) is currently receiving increasing attention as a leading indicator to assess its industrial feasibility and economic viability. Therefore, we calculated the EEs of CO₂RR and HER (Fig. S19 in Supporting information), respectively. It was found that the EE toward CO₂RR on DEA-Pd NPs at −0.7 V was 69.2%, much higher than that of Bare-Pd NPs (53.8%), indicating that DEA-Pd NPs could reduce power input and production cost upon achieving the same output (Fig. 3d). Furthermore, the electrochemical double layer capacitance (C_dl) was measured to determine the electrochemical active surface areas (ECSAs) in pursuit of normalizing the activity and kinetics (Fig. 4a and Fig. S20 in Supporting information). Surprisingly, C_dl of DEA-Pd NPs was 19.32 mF/cm², relatively smaller than that of Bare-Pd NPs (22.36 mF/cm²), probably due to the partial coverage of surface Pd atoms by DEA. However, the activity of a catalyst is normally governed by two principal factors, i.e., the number and the intrinsic activity of active site. As expected, DEA-Pd NPs possessed a TOF value of 2.46 s⁻¹ at −0.7 V, larger than Bare-Pd NPs (Fig. 4b), implying that the enhanced CO₂RR activity on DEA-Pd NPs was derived from the superior intrinsic activity of active sites induced by -NH- in DEA. This was further verified by the kinetics insights based on the ECSA-corrected Tafel plot of log[j_(CO)] versus η in Fig. 4c, where a Tafel slope of 111.1 mV/dec was achieved on DEA-Pd NPs, considerably smaller than that of 134.2 mV/dec on Bare-Pd NPs, which was also closer to the theoretical 118 mV/dec governed by the rate-determining step of CO₂ + e⁻ → CO₂^•− [12,53]. In addition, during the stability test of 50 h at −0.7 V (Fig. 4d), DEA-Pd NPs maintained the FE_CO above 90% and j at 5.8 mA/cm², whereas the FE_CO of Bare-Pd NPs decayed to 63%, confirming the impressive stability of DEA-Pd NPs toward CO₂RR. To verify the structural stability of DEA-Pd NPs after long-term electrolysis, TEM image (Fig. S21 in Supporting information), EDX mappings (Fig. S22 in Supporting information), size distribution (Fig. S23 in Supporting information), FTIR tests (Fig. S24 in Supporting information) of DEA-Pd NPs were obtained. The results showed negligible changes in morphology, elemental distribution and particle size. Also, the characteristic absorption peaks of DEA were observed in DEA-Pd NPs, whereas they were not detected in KHCO₃ solution. These results confirmed that the excellent catalytic stability primarily originated from its robust structural stability. For a better holistic view, a radar map presenting stability reflected by FE_CO retention ratio, potential numbers with FE_CO over 90%, FE, EE and j_CO was plotted in Fig. 4e. Apparently, all indicators of DEA-Pd NPs were far better than those of Bare-Pd NPs, highlighting the excellence of DEA in promoting CO₂RR. Furthermore, compared with other Pd-based catalysts (Table S1 in Supporting information), DEA-Pd NPs demonstrated a more outstanding CO₂RR performance, which further highlighted the crucial role of DEA in enhancing the CO₂RR performance of Pd NPs.

Figure 4

	Download: JPG larger image
Fig. 4. (a) Plots of charging difference versus scan rate. (b) TOFs at different potentials. (c) Tafel plots. (d) Long-term stabilities of Bare-Pd NPs and DEA-Pd NPs, and (e) their radar map involving FECO retention ratio (i.e., FECO(Final)/FECO(Initial)), FECO at −0.7 V, EE, jCO and potential number (FECO > 90%).

To uncover the underlying reaction pathway of DEA-Pd NPs toward CO₂RR, in situ FTIR measurements were conducted, and the results were shown in Figs. 5a and b, where the peaks at ~2106, ~1976 and ~1896 cm⁻¹ were attributed to a CO linear configuration intermediate (*CO_L), a CO bridge adsorption configuration intermediate (*CO_B), and a triple-bound adsorption configuration intermediate (*CO_T), and, respectively [54]. Clearly, a transformation between *CO_L, *CO_B and *CO_T at different potentials was observed for both DEA-Pd NPs and Bare-Pd NPs. As the potential increased, the *CO_L peak for Bare-Pd NPs became more prominent, while DEA-Pd NPs maintained a higher proportion of *CO_L throughout the entire potential range, and the peak was gradually shifted to 2116 cm⁻¹ due to the decrease in resonance frequency in the negative electric field caused by the Stark effect [55,56]. It was concluded that DEA on surface Pd atoms limited the formation of *CO_B and *CO_T adsorption configurations, but favored *CO_L adsorption. This was because the increased hydrophobicity of DEA-Pd NPs influenced the local solvation environment, which could promote a more facile CO desorption, since the thinner water layer on the hydrophobic surface contributed to a better CO diffusion into the electrolyte, thereby leading to a higher proportion of *CO_L. Meanwhile, the presence of DEA could tune the electronic configuration of Pd, which weakened the surface adsorption of CO by providing more electrons to the Pd-CO antibonding orbital, which preferentially stabilized the *CO_L. Moreover, the *CO_L only required a single Pd site, but the *CO_B and *CO_T involved multiple Pd atoms, which might be sterically hindered by the DEA on Pd surface, thus limiting the formation of *CO_B and *CO_T adsorption configurations and favoring the *CO_L adsorption, thereby alleviating the CO poisoning on surface Pd atoms [57,58].

Figure 5

	Download: JPG larger image
Fig. 5. Potential-dependent in situ FTIR patterns from −0.6 V to −1.3 V of (a) DEA-Pd NPs and (b) Bare-Pd NPs. (c) Differential charge density, (d) PDOS, (e) adsorption energy of *COOH and *CO, and (f) ΔG diagrams on DEA-Pd NPs and Bare-Pd NPs.

In addition, differential charge density was calculated, as shown in Fig. 5c and Fig. S25 (Supporting information), and the xyz table with coordinates of the optimized structures was demonstrated in Table S2 (Supporting information). Clearly, the charge density on DEA was depleted, whereas that at the interface of Pd and N was obviously accumulated. The Bader charge analysis indicated that Pd absorbed 0.21 e from DEA, suggesting that the N atom in DEA formed a strong Pd–N covalent bond with Pd, which donated electrons to Pd during CO₂RR. To further explore its electronic structure, the partial density of states (PDOS) was calculated, and the results in Fig. 5d showed that the electrons of Bare-Pd NPs were squeezed due to the electron donation effect induced by DEA, resulting in the whole shift of Pd d-band center of DEA-Pd NPs from −1.53 eV to −1.84 eV. This was confirmed by previous studies that a lower Pd d-band center could weaken the surface adsorption of CO by providing more electrons to the Pd-CO antibonding orbital [15,59]. In addition, we compared the adsorption energies of key intermediates (i.e., *CO, *COOH and *H) for DEA-Pd NPs and Bare-Pd NPs in Fig. 5e. It was found that the *COOH adsorption energy of −2.59 eV on DEA-Pd NPs was larger than that of −2.49 eV on Bare-Pd NPs, together with a lower *CO adsorption energy of −1.92 eV. This demonstrated that DEA-Pd NPs was beneficial to *COOH adsorption and *CO desorption, which precisely addressed the CO poisoning issue on Pd-based materials. With one step forward, the profiles of Gibbs free energy (i.e., ΔG) diagrams for CO₂RR to CO over DEA-Pd NPs and Bare-Pd NPs were displayed in Fig. 5f. Apparently, DEA-Pd NPs experienced a lower energy barrier of only 0.57 eV for the first proton-coupled electron transfer step (i.e., CO₂ + H⁺ + e⁻ → *COOH) than Bare-Pd NPs (0.67 eV). Similarly, the energy required to proceed the CO desorption step (i.e. *CO → * + CO) was 1.37 eV on DEA-Pd NPs, energetically more favorable than 1.52 eV on Bare-Pd NPs. All in all, the underlying mechanism was thus inferred that –NH– in the Pd–N bond formed H-bonds with *COOH in the first proton-coupled electron transfer step, which served as the driving force for DEA-Pd NPs to deliver a stronger *COOH adsorption ability. Meanwhile, the left shift of d-band center caused by the electron donating effect of DEA simultaneously inhibited the adsorption of *CO. It is thus concluded that through precisely tuning the electronic structure and distribution, the adsorption energy of *COOH and *CO could be individually tailored, indicating a deviation from the conventional linear scaling relationship, which typically assumed a synchronized increase or decrease in *COOH and *CO adsorption strength.

In summary, we developed a "one stone killing two birds" DEA-mediated strategy to functionalize surface Pd atoms by ligand exchanging the residual OAm on the ultrafine Pd NPs with DEA instead of deliberately removing it, and explored its crucial role toward CO₂RR. DFT calculations revealed that the electron donation effect of DEA could compress the electrons of Pd, which thus lowered the Pd d-band center, and weakened the CO surface adsorption by providing more electrons to the Pd-CO antibonding orbital. Moreover, the selective H-bond formation at the first proton-coupled electron transfer step accelerated the *COOH adsorption, and as such, the unfavorable linear scaling relationship of the isochronously increased or decreased adsorption energies of *COOH and *CO were broken. As a result, DEA-Pd NPs achieved a FE_CO close to 100% in an ultrawide potential window from −0.7 V to −1.3 V and a high CO selectivity exceeding 90% for over 50 h. Therefore, the smart reverse design will open up new possibilities to address the residual issue of ligands on different nanostructure during synthesis, and extend the ligand utilization toward various electrocatalytic systems.

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.

Mulin Yu: Writing – original draft, Investigation, Formal analysis, Data curation, Conceptualization. Shuo Liu: Data curation. Yufeng Tang: Methodology, Data curation. Guoqiang Lu: Data curation. Linbo Liu: Data curation. Pengfei Sui: Data curation. Xianzhu Fu: Data curation. Subiao Liu: Writing – review & editing, Visualization, Supervision, Project administration, Methodology, Funding acquisition, Conceptualization. Yifei Sun: Writing – review & editing, Supervision. Jingli Luo: Writing – review & editing, Supervision.

This work is supported by the Pilot Group Program of the Research Fund for International Senior Scientists (No. 22350710789) and the Start-up Funding of Central South University (No. 206030104). The authors are grateful for resources from the High-Performance Computing Center of Central South University.

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2025.111958.