Engineering Ultrathin Alloy Shell in Au@AuPd Core‐Shell Nanoparticles for Efficient Plasmon‐Driven Photocatalysis

Bimetallic core‐shell nanoparticles (NPs) possessing a synergetic coupling of plasmonic and catalytic metals have emerged as promising prototypes for efficient plasmon‐driven photocatalysis. Here, Au@AuPd core‐shell NPs composed of Au core NP and ultrathin AuPd shell are introduced to acquire improved light utilization efficiency in photocatalytic selective oxidation. With systematical control of the composition of the ultrathin alloy shell, it reveals that the Au@AuPd core‐shell NPs with 10 at% Pd (in the single‐atom alloy regime) is an effective nanostructure capable of maximizing the quantum yield of plasmon‐induced light absorption and optimizing the surface electronic structure for catalytic reactions. This controlled system provides new insights into shell engineering for enhancing photocatalytic performance via the regulation of energy funneling processes in core‐shell nanocatalysts.


DOI: 10.1002/admi.202301070
[3][4] However, these strategies have been limited by their inherent lack of thermal stability.Utilizing solar energy for chemical reaction has attracted tremendous attention to overcome the limitation of the conventional thermal catalytic reaction by reducing the reaction temperature. [5]One of the efficient ways to harvest solar energy on metal nanoparticles is using surface plasmons phenomena generating the redistributed electromagnetic field, the excited hot carriers, and the local heating under visible-light irradiation. [6,7]So far, many recent studies have reported that plasmon-induced energetic hot electrons transfer to reactant molecules on the metal surface and activate the molecular orbitals of adsorbates or metal-adsorbate complexes, enhancing reactivity and selectivity. [8,9]However, inherently negligible catalytic properties of plasmonic noble metals have limited the improvement of the catalytic performance.12][13][14] Among several bimetallic geometries, core-shell (or antennareactor) nanostructures have been proposed as the efficient geometry for efficient plasmon-driven photocatalysis. [10,15]Typically, the core-shell nanostructures are designed with the core of plasmonic nanoparticles (NPs) exhibiting dominant optical extinctions throughout the visible range and the shell composed of catalytically active layers. [16,17]In these structures, light energy harvested by the plasmonic core NPs could be effectively transferred to the catalytic shell and activate the chemical reactions. [18]herefore, finding a way to maximize the usage of the light energy obtained from the plasmonic NPs and enhance the catalytic performance is a primary goal in this field.[21] When the thickness of the Pd shell of the Au@Pd core-shell nanorods was adjusted between 2 and 27 layers, the one with Pd shells of 14 atomic thick showed the highest catalytic styrene hydrogenation due to the maximal local photothermal heating and efficient transfer of plasmonic hot-electrons on the Pd shell under light illumination. [16]u@Pd core-shell nanorods also exhibited superior catalytic activity and selectivity under the hydrogenation of butadiene compared to alloyed Au-Pd nanorods.[22] Simultaneously, a thin Pt shell (≈1 nm) covering the Ag nanocube core was suggested as an effective light absorption channel confining energetic charges, amplifying the reaction rate and selectivity in preferential CO oxidation in the presence of excess H 2 .[17] However, most studies on plasmonic core-shell nanocatalysts have been biased in the case that the thickness of the shell is over one atomic layer.Therefore, there is a lack of systematic exploration for finding the optimum surface structure for enhanced catalytic performance when the shell is thinner than the monolayer.
A dilute nanoalloy provides an ideal structure with a monolayer alloy shell whose composition is systematically controllable. [23,24]These nanoalloys effectively retain the surface plasmon effect of the core while optimizing the electronic structure of the surface alloy, offering advantages for reactions.Besides, while a minor fraction of reactive metal functions as catalytic reaction sites, the predominant presence of less reactive host material contributes remarkably to the selectivity of the overall process, yielding exceptional catalytic performance and selectivity consequently. [23,24]In this study, we have selected the AuPd bimetallic system motivated by the remarkable selectivity and resistance to deactivation exhibited by Au, its miscibility to form homogeneous alloys with Pd. [25] Accordingly, we employ dilute Au/Pd alloy NPs, used for highly selective oxidation without light irradiation in our previous work, [26,27] for plasmon-mediated photocatalytic benzyl alcohol oxidation.The benzaldehyde produced from the selective oxidation of benzyl alcohol is widely used in various industries. [28,29]However, achieving high conversion and selectivity of benzaldehyde requires addressing the challenge of promoting the cleavage of the O─H bond and dehydrogenation processes facilitated by electron transfer in the catalyst and adsorption of reactant.Through precise modulation of the Pd ratio in the dilute Au/Pd alloy NPs, we develop Au@AuPd core-shell NPs, whose ultrathin shell consists of an AuPd alloy with the tuned atomic Pd fraction from 3 at% to 45 at% corresponding to the Pd composition of the shell from 7.2% to 205.4%.To investigate the effect of a synergistic combination of plasmon-driven hot carriers/phonon excitation and optimum electronic structure of alloy shell modified by contents ratio, we conducted selective oxidation of benzyl alcohol with a different compositions of ultrathin AuPd shell in Au@AuPd NPs.Consequently, we reveal that the yield to benzaldehyde presents a volcano activity plot as a function of the Pd composition of the shell, and Au@AuPd NPs with 10 at% Pd exhibit the highest yield of 70.13% under light illumination.Furthermore, the light enhancement is maximized at the identical composition, implying that the merits of both surface plasmons and the electronic structure of the alloy shell are optimized when the localized Pd composition of the shell is 20.6%.

Results and Discussion
The Au@AuPd core-shell NPs were synthesized via the seedmediated method reported in previous work [26,27] with precise control of Pd/Au atomic ratios (Pd:Au = 3:97, 10:90, 20:80, 30:70, 45:55) (refer to Supporting Information for detailed information).Spherical Au NPs of 5.5 ± 0.6 nm were used as the plasmonic core because not only Au allows strong photon absorption via interband (d-to-s) transitions at visible frequencies, but also light scattering can be negligible when the diameter of NP is smaller than 20 nm. [18,22,30]Besides, the atomic loading of Pd is kept below 45 at% to ensure that the thickness of the AuPd shell remains ultrathin, with fewer than two atomic layers.The structures of the Au@AuPd NPs denoted Pd 3 Au 97 , Pd 10 Au 90 , Pd 20 Au 80 , Pd 30 Au 70 , and Pd 45 Au 55 NPs were characterized by transmission electron microscopy (TEM), high-resolution TEM (HRTEM), high-angle annular dark-field scanning TEM (HAADF-STEM), and STEMenergy dispersive X-ray spectroscopy (STEM-EDS).TEM images and size distributions illustrate that the spherical Au@AuPd NPs with different stoichiometric ratios of Pd to Au are formed uniformly with the narrow particle size distribution (Figure 1; Figure S1, Supporting Information).As the Pd ratio increases, the average diameter of the Au@AuPd NPs slightly increases from 5.6 ± 0.6 nm for Pd 3 Au 97 NPs to 7.0 ± 0.3 nm for Pd 45 Au 55 NPs, indicating atomic Pd growth on the surface of Au NPs.In addition, d-spacing determined based on the HRTEM images of all Au@AuPd NPs and Au NPs are the same as 2.33 Å, which is the value for the Au (111) plane, [31][32][33] suggesting that the crystallinity of the Au NPs is well-maintained regardless of the addition of Pd on the surface of the core (Figure S2, Supporting Information).This result also implies that the Pd atoms are grown on the surface of Au NPs, not to the interior, as confirmed by extended X-ray absorption fine structure (EXAFS) in our previous work. [26]For Pd NPs, it is verified that NPs of comparable size to Au NPs are formed, exhibiting a d-spacing of 2.23 Å, which corresponds to the Pd (111) plane (Figure S2g, Supporting Information). [34,35]We verified the core-shell design of NPs again by the elemental distributions of the Au@AuPd NPs acquired from the STEM-EDS maps in the case of Pd 10 Au 90 , Pd 20 Au 80 , and Pd 30 Au 70 NPs (Figure 1g-i; Figure S3, Supporting Information).In addition, the accurate chemical compositions of the NPs were obtained from STEM-EDS and inductively coupled plasma mass spectrometry (ICP-MS), and the results well matched the targeted atomic ratio of Pd and Au (Table S1, Supporting Information).Based on STEM-EDS analysis, we verified that Pd atoms gradually cover the surface of the Au NPs, and form an AuPd or Pd shell on the Au NPs within the ultrathin regime.Depending on the amount of Pd loading, the ultrathin shell exists as an AuPd alloy at lower Pd concentration (i.e., Pd 3 Au 97 , Pd 10 Au 90 , Pd 20 Au 80 NPs) and as a Pd layer at Pd 30 Au 70 and Pd 45 Au 55 NPs.
UV-vis spectra measured at a concentration of 0.05 mg ml −1 for the metal NPs show that the absorbance maximum of the localized surface plasmon resonance (LSPR) of Au NPs decreases, with an increase in Pd loading, accompanied by peak broadening (Figure 2a).The monometallic Au NPs exhibit a peak maximum centered at 518 nm, typically ascribed to the LSPR of Au NPs with comparable size. [26,36,37]Despite an increase in the Pd loading, the maximum peak wavelength ( peak ) of Au@AuPd NPs remained nearly identical to the value of the monometallic Au NPs regardless of peak broadening.(Figure 2a).The apparent broadening of the LSPR peak with increasing Pd loading implies modified electronic structures and plasmon-decay pathway in the Au@AuPd NPs. [16,17,36,38]The damped LSPR characteristic in Au@AuPd NPs is obviously demonstrated in Figure 2b, showing the peak height difference of NPs in absorbance value at 518 nm between each NPs and Pd 45 Au 55 NPs.The difference indicating the relative intensity of the plasmonic response decreased by 90% with a value of 0.05 in the Pd 30 Au 70 NPs compared to the Au NPs with a value of 0.49.It is because the LSPR generated on the surface of Au NPs is damped by Pd atoms owing to a large imaginary part of the dielectric constant of Pd, leading to prompt rapid plasmon dephasing. [16]We also found the almost nonexistent characteristic LSPR peak of Au in the Pd 45 Au 55 NPs as the amount of Pd reached 45 at%.The results suggest that the Pd atoms gradually cover the surface of Au NPs as the amount of Pd loading increases in the Au@AuPd NPs, which is consistent with the observation in TEM images.Therefore, the diminished LSPR peak in Au@AuPd NPs implies the formation of the core-shell NPs where the core consists of Au NP and the shell is composed of ultrathin AuPd alloy or Pd layer.The damped SPR intensity of Au@AuPd NPs was also confirmed by finite-difference timedomain (FDTD) simulations, showing a slightly decreasing trend of max and average intensity of the electric field as the Pd contents increase (Figure S4, Supporting Information).
To confirm the chemical and electronic states of Au and Pd in Au@AuPd NPs, we analyzed X-ray photoelectron spectroscopy (XPS) spectra calibrated based on the C1s (284.8 eV) peak (Figure 3a,b; Figure S5 and Table S2, Supporting Information).The Au 4f spectra in Figure 3a shows typical doublets of peak that are the characteristic peaks of metallic Au 4f 5/2 (≈87 eV) and Au 4f 7/2 (≈83 eV) caused by spin-orbit coupling of the electrons in f orbital.All Au@AuPd NPs exhibit well-separated spin-orbit components (Δ = 3.7 eV) with an asymmetric peak shape, indicating that the Au core maintains the metallic state regardless of the loading amount of Pd.[41] This phenomenon also indicates that the surface of Au@AuPd NPs exists in the form of an AuPd alloy with a strong interaction between two metals as elucidated in previous studies. [38,41]In the case of Pd 3d spectra (Figure 3b), it shows a typical doublets feature composed of Pd 3d 3/2 (≈339 eV) and Pd 3d 5/2 (≈334 eV), where Pd 3d 5/2 peak was overlapped by Au 4d 5/2 peak (≈335 eV). [40,41]Therefore, we deconvolute the Pd 3d spectra considering three phases; metallic Au, metallic Pd, and oxidized Pd.As shown in Figure 3b and Table S2 (Supporting Information), as Pd composition increases, the peak of Au 4d is suppressed, and the values of the binding energy of Pd 3d shift gradually to higher energies in Au@AuPd NPs.A shift to lower binding energies of Au 4f and toward higher binding energies of Pd 3d evinces again the formation of the AuPd alloy phase on the surface of Au@AuPd NPs. [41]Notably, a minor peak assigned to Pd-O coordinations is first shown in Pd 10 Au 90 NPs and slightly increases with higher Pd loadings; from 2.97% for Pd 10 Au 90 NPs to 9.13% for Pd 45 Au 55 NPs (Table S2, Supporting Information).This observation indicates that a fraction of the Pd on the surface of AuPd NPs underwent natural oxidation, resulting in the formation of PdO. [40,26,42]The natural PdO formation is evident when the Pd content is higher (i.e., Pd 45 Au 55 NPs) due to the less interaction between Pd and Au.Therefore, a decrease in the reaction activity could be associated with a reduction of active sites in Pd 45 Au 55 NPs, while the impact of PdO remains marginal with the lower fraction of PdO in Pd 10 Au 90 , Pd 20 Au 70 , and Pd 30 Au 70 NPs. [42]Here, Pd 45 Au 55 NPs assumed to form a Pd shell on the surface of the Au NP exhibit an abrupt increase of the Pd-O coordination ratio, demonstrating that the AuPd shell is prone to be oxidized as the Pd composition increases and forms a Pd shell.Based on STEM-EDS, UV-vis, and XPS results, we confirm that Au@AuPd NPs synthesized in this study have a core-shell structure where the ultrathin shell structure changes from dilute AuPd alloy to Pd layer depending on the Pd/Au ratio.
Figure 3c shows the Pd composition of Au@AuPd NPs calculated based on the normalized area of Pd 3d 3/2 and Au 4f 5/2 peaks considering the sensitivity factors of the two metals.The Pd composition systematically increases from 3.1 at% to 50.3 at%, which is well-matched with the targeted value and the results estimated from the STEM-EDS and ICP-MS measurements (Table S1, Supporting Information).Because all Pd atoms are confined to the outer monolayer surface of NPs, the composition of Pd atoms in the assumed monolayer shell can be calculated by the following equation: where  = Pd composition of shell, V c = 4R core 3 /3 is the volume of the Au core NP, and V s ≅ 8r Pd (R core +r Pd ) 2 is the monolayer Pd shell volume. [36]The radius of the core (2R core = 5.5 ± 0.6 nm) acquired from the TEM images of Au NPs and the radius of the metal atoms (r Au = 1.74 Å, r Pd = 1.69 Å) are used for the calculation.As shown in Figure 3d, as the targeted ratio of Pd in the NPs increased from 3 at% to 45 at%, the Pd composition of shell has expanded from 7.2% to 205.4%.At low Pd compositions, Pd incorporates into the surface of the Au core, forming an AuPd alloy. [26]This indicates the presence of an ultrathin AuPd shell on the surface of Au NPs when the Pd content is less than 30 at%, while the shell is composed of an approximate Pd layer of two-atom thick in Pd 45 Au 55 NPs.Since the coordination number (CN) of Pd-Pd is expected to zero when the concentration of Pd is below 10 at%, [26,41] it suggests that single-atom alloys have been For further understanding of the electronic structure of the Au@AuPd NPs, the valence band (VB) spectra of Au, Pd, and Au@AuPd NPs were measured by XPS.All the VB spectra were normalized in the same region (−4 to 10 eV).As shown in Figure 4a, the VB of Au@AuPd NPs showing the sp-d rehybridization of Au and Pd is gradually broadened and shifted toward the Fermi level (E F , normalized to the binding energy of 0 eV) as the fraction of Pd increases.The deformation of VB with adding Pd is valid because the d-band level of Pd is much closer to the E F than the Au generating the strong LSPR at visible frequencies, [43] as confirmed in Au 4f and Pd 3d spectra.This trend is also clearly observed in the positions of the d-band center of the NPs relative to the E F (Figure 4b), which is calculated according to the following equation; where E d is the energy assigned to the d-band center, representing the average energy of the d-electrons on the metal surface.[46] Since the spcharacter of late-transition metals such as Au and Pd is generally considered insignificant near E F , it is possible to determine the d-band center by integrating the VB spectra in the range from −2 eV up to about 9 eV, including contributions of both sp-band and d-band. [44,47]he calculated d-band center of NPs is clearly shifted toward the E F by the inclusion of Pd, from −3.46 eV of the Au NPs to −1.68 eV of the Pd 45 Au 55 NPs. Figure 4 shows that the incorporation of Pd atoms on the Au core increases the electron density at the Fermi level and shifted the position of the d-band center toward the E F , confirming the formation of the AuPd alloy phase on the surface of Au@AuPd NPs by incorporating Pd atoms on the surface of Au NPs (Figure 4b).The changes in electronic properties of the Au@AuPd NPs are caused by the ligand effect that occurs on the surface of the AuPd alloy shell due to direct charge exchange or orbital rehybridization. [19,36]When Pd atoms interact with the intimate Au atoms, Pd atoms lose spelectrons and achieve d-electrons, while Au receives sp-electrons with partial loss of d-electrons. [41]The mixing of the atomic orbitals through sp-d orbital rehybridization in the AuPd alloy leads to a broadening of the valence band, positioning the d-band center between the values of Au and Pd to maintain the equivalent level of d-band filling.The Pd atoms on Au NPs also experience tensile strain due to a mismatch in the lattice parameters of Au (4.08 Å) and Pd (3.89 Å), which cause a reduction in the degree of d-orbital overlap and a narrowing of the d-band width. [48]In As the adsorbate molecule interacts with the metal surface, the distinct bonding and antibonding molecular orbitals (MOs) are formed and the occupancy of the antibonding MOs is determined by the location of the d-band center. [49]For that reason, the precise modulation of the composition of ultrathin AuPd alloy shell in Au@AuPd NPs enables control over the d-band center, which in turn allows for effective manipulation of the adsorption, activation, and dissociation energy of reactants. [50]o investigate the synergistic effects of the ultrathin AuPd alloy shell and plasmonic functions of the Au@AuPd NPs, we conducted benzyl alcohol oxidation experiments under dark and light conditions at atmospheric pressure.The 0.4 wt% of NPs (Au, Pd, and Au@AuPd NPs) were loaded on -Al 2 O 3 , which served as a support that is not only photo-catalytically inactive but also thermally and chemically stable.By utilizing -Al 2 O 3 as the support, we also intended to avoid the additional effects of hotspots between NPs as well, solely observing the photocatalytic performance of NPs themselves. [51]The catalysts were employed in the reaction mixture consisting of 0.2 mmol of benzyl alcohol and 2 ml of trifluorotoluene, which was subsequently heated to 55 °C for 8 h.Trifluorotoluene is chosen as the solvent due to its inertness to oxidation, high solubility for molecular oxygen. [52]o illuminate the reaction system, a 100 W halogen lamp was utilized as a light source, and to ensure selective transmission of light within the visible wavelength of 325 -700 nm, both a band-pass filter and water filter were employed.The intensity of light measured in the reactor was 1.07 mW cm −2 with filters.Details of reaction experiments are described in Supporting Information.All catalytic reaction results presented in this study were acquired with high reliability, based on 2 to 3 independent experimental replicates conducted under identical conditions.Moreover, it was confirmed that no reaction transpired in the absence of metal NPs on -Al 2 O 3 , regardless of light exposure, indicating the successful isolation of the reaction system from the undesired effects of incident light in the ultraviolet and infrared spectra.The structural stability of the catalysts was also confirmed based on TEM measurements before and after the catalytic reac-tion, revealing well-maintained size and shape of the NPs without any significant aggregation (Figures S6 and S7, Supporting Information).Furthermore, XPS analysis revealed that the chemical state of Au, Pd 3 Au 97 , Pd 10 Au 90 , Pd 20 Au 80 , and Pd 30 Au 70 NPs remained unchanged after the benzyl alcohol oxidation (Figure S8 and Table S3, Supporting Information).However, in Pd 45 Au 55 and bare Pd NPs, a slight increase in the amount of PdO was observed, attributed to the weakened interaction of Pd with Au. [40,42] Figure 5a demonstrates the results of the catalytic conversion of Au@AuPd NPs with different Pd compositions of the shell during the benzyl alcohol oxidation under atmospheric conditions, both in the presence and absence of light.In all catalytic reactions, we found that benzaldehyde is the sole product, indicating 100% selectivity.The results show that Au@AuPd NPs manifest higher catalytic efficiency than individual Au and Pd NPs, under dark conditions and visible-light irradiation (Figure 5a; Figure S9, Supporting Information).][55][56][57] In particular, Au@AuPd NPs with 10 at% of Pd (i.e., Pd 10 Au 90 NPs) exhibit a significant increase in the conversion from benzyl alcohol to benzaldehyde compared to the other catalysts under visible-light irradiation.In the absence of light, the catalytic performance of varying Pd compositions of the shell exhibits a similar volcano-like trend as observed under light irradiation with markedly diminished product yields.To further clarify the photo-induced chemical reaction, the light-enhanced conversion rate is quantified as the ratio of the difference in conversion values under dark conditions and light irradiation to the conversion value under dark conditions.59] Therefore, the electron transfer from the surface active site to the reactant emerges as a crucial determinant facilitating the efficient cleavage of the O─H bond and dehydrogenation, thereby promoting the catalytic alcohol oxidation reactions.In this regard, as previously mentioned, the interaction between Pd atoms and adjacent Au atoms on the surface of Au@AuPd NPs results in Pd atoms acquiring d-electrons from the Au atoms, leading to the surface charge heterogeneity at the Au-Pd interface (i.e., positively charged Au and electron-rich Pd sites). [40,56,57]Consequently, the electron-rich Pd sites in the ultrathin AuPd shell of Au@AuPd NPs are beneficial to facilitate the cleavage of O─H bond and dehydrogenation in the process of benzyl alcohol oxidation, resulting in a higher conversion of benzyl alcohol to benzaldehyde compared to bare Au and Pd NPs. [48]The amount of d-electron gain at Pd sites and the surface charge heterogeneity can be maximized at the optimal composition of the shell, exerting a direct influence on the catalytic reaction process. [40]Therefore, the highest catalytic activity observed in Pd 10 Au 90 NPs, even under dark conditions, can be attributed to their optimal singleatom alloy structure, where Au atoms serve to isolate Pd atoms, thus maximizing the number of active single-atom Pd sites and the surface charge heterogeneity. [26,56,57]In contrast, the catalytic activity is reduced at Pd concentrations below and above 10 at%, primarily due to the diminished surface charge heterogeneity at the Au-Pd interface.Specifically, a limited number of singleatom Pd sites could result in the case of Pd 3 Au 97 NPs.On the other hand, the formation of Pd clusters and layers on the surface could account for the lower activity of Pd 20 Au 80 , Pd 30 Au 70 and Pd 45 Au 55 NPs, respectively.When the catalytic system is irradiated with visible light, the plasmonic Au core functions as an efficient antenna, harvesting solar energy through the excitation of localized plasmon resonance.[65] For further validation, we explored the correlation between benzaldehyde conversion and light intensity, revealing a linear dependence on light intensity.This finding provides compelling evidence for the existence of non-thermal pathways.[68][69] Additionally, the reactions were conducted using well-separated NPs on the support to prevent heat accumulation, with precise temperature control and continuous stirring.Since we also employed the Au NPs of 5.5 ± 0.6 nm, the local heating effect is expected to be negligible, [61,[70][71][72] and the majority of the excited plasmon energy can be effectively utilized through light absorption (i.e., sp-sp intraband transition), primarily decaying via nonradiative electron-hole pair generation. [18,22,30]herefore, under visible light exposure, the pure Au NPs display slightly enhanced catalytic activity as the hot electrons generated on the surface of the NPs by surface plasmon resonance are transferred to the molecular adsorbates (Figure 5a).In contrast, the catalytic activity of pure Pd NPs, which lack the ability to absorb visible light, remains unchanged (Figure S9, Supporting Information).The photo-induced enhancement of the catalytic activity is much more considerable in the Au@AuPd NPs, where atomic Pd sites are present on the surface of the Au NPs.The rapid dephasing of the Au, facilitated by robust coupling to the electrons in the d-band of Pd at the Fermi level, and the transfer of hot carriers from the Au core to Pd sites in the AuPd shell significantly influence the activation of the reactant. [36]Consequently, the photo-induced enhancement of catalytic activity observed in the antenna-reactor Au@AuPd NPs can be attributed to the presence of available hot carriers at the ultrathin AuPd alloy shell and the electromagnetic field associated with the plasmon of the Au core. [65,73]As previously stated, the presence of atomic Pd sites on the surface of the Au NPs increases the surface charge heterogeneity in the Au@AuPd NPs.Besides, due to Pd's higher imaginary dielectric function and d-electron density at the Fermi level in comparison to the Au, the plasmoninduced hot carriers within Au core are efficiently funneled to the Pd sites in ultrathin AuPd shell through electron−electron scattering, [17,74] thereby augmenting the surface charge heterogeneity and promoting the electron transfer to the lowest unoccupied molecular orbital (LUMO) of benzyl alcohol molecules at the surface of Au@AuPd NPs (Figure 5c,d).When an energetic electron accumulates in the LUMO of reactant, it promotes bonding weakening or elongation, leading to the cleavage of the O─H bond and dehydrogenation. [12,57]The hot hole combines with the electron, which has lost energy after activating the reactant, completing the cycle of energy transfer within the catalytic system. [12,75]Additionally, it has been reported that the plasmon decay channel through the shell can also be affected by the electric field at the surface of NPs.][78] Based on these facts, it is demonstrated that the notable light-enhanced catalytic performance of Au@AuPd NPs with 10 at% of Pd arises from the synergistic effect of two key factors: first, the maximization of the surface charge heterogeneity achieved through an increasing number of single-atom Pd sites (i.e., Au-Pd interfaces); and second, the efficient funneling of hot electrons from Au to Pd amplified by the surface plasmon effect.Notably, the performance of the Au@AuPd catalyst was highly comparable to other AuPd photocatalysts, even under air atmosphere and at relatively lower temperatures and metal loading.Additionally, it offers the advantage of effective charge distribution by adopting a structure with a dilute alloy shell instead of the conventional thick shell thickness typically found in core-shell structures (Table S4, Supporting Information).As a result, our findings suggest that the introduction of an ultrathin alloy shell in the core-shell nanostructure is a promising strategy to maximize the catalytic activity and minimize optical losses, thereby enhancing photon-to-chemical energy conversion.

Conclusion
We have investigated the optical and electronic properties as well as the photocatalytic activity of the Au@AuPd core-shell bimetallic nanoparticles.It was found that the Au@AuPd NPs demonstrated superior catalytic activity toward benzyl alcohol oxidation compared to pure Au or Pd NPs, with an optimum Pd composition of the shell is 20.6% in the ultrathin AuPd shell, highlighting the effectiveness of the single-atom alloy regime.Volcano-like tendencies of oxidation results under visible light irradiation and in the dark confirmed the synergistic interactions between Au and Pd when Pd atoms gradually cover the core Au NPs, forming AuPd alloys at the surface.Precisely tailoring the composition of alloy shells on Au@AuPd bimetallic NPs within a ultrathin regime changes d-band center and surface charge heterogeneity, along with the corresponding changes in adsorption energy and the interactions between adsorbates and surface metal atoms.Moreover, by introducing core-shell morphologies consisting of plasmonic core and catalytic monolayer shell in bimetallic alloy design, plasmon-driven hot electrons of NPs could significantly boost catalytic conversion rates while simultaneously promoting optical effects and electronic interactions.Our findings suggest that the rational design of ultrathin alloy shells in core-shell nanoparticles can tailor optical and electronic properties, thereby optimizing visible-light utilization and catalytic activity toward alcohol oxidation reactions.

Experimental Section
Synthesis of Au Nanoparticles (NPs): Au NPs were synthesized using a Brust method with a slight modification. [26,79]To prepare a reducing agent, 88.6 mg of TBAB (Borane tert-butylamine complex 97%, Aldrich) was first dissolved in the mixture of 2 ml OLAM (Oleylamine 70%, Aldrich) and 2 ml tetralin (1,2,3,4-Tetrahydronaphthalene 98%, Acros) through ultrasonication for 20 min.After mixing 20 ml OLAM and 20 ml tetralin through magnetic stirring at 25 °C, HAuCl 4 •3H 2 O 200 mg was prepared as a gold precursor.Right after the injection of the reducing solution, the prepared HAuCl 4 •3H 2 O was added to the mixture of 20 ml OLAM and 20 ml tetralin.As soon as the gold precursor was dissolved in the solution, the color of the solution changed to dark purple instantaneously.For the growth of Au nanoparticles, the solution was stirred at 300 rpm for 1 h at 25 °C.The Au nanoparticles were recovered by precipitation using 25 ml ethanol and 5 ml isopropyl alcohol followed by centrifugation at 8000 rpm for 5 min.The clear supernatant was discarded, and the precipitated particles were dispersed in 5 ml hexanes.The precipitation process was repeated twice, and the particles were finally dispersed in 10 ml hexanes.
Synthesis of Au@AuPd Core-Shell NPs: The Au@AuPd core-shell NPs were prepared using a seed-mediated process following the previous work with a minor modification. [27,26]Before the synthesis, the amount of Pd was calculated to form the targeted Pd composition on the surface of the Au core NP.After preparing a solution of Pd(NO 3 ) 2 •2H 2 O (40% Pd, Aldrich) precursor dissolved in OLAM (1 mg ml −1 ), 15 mg Au NPs in hexanes and the desired amount of Pd(NO 3 ) 2 •2H 2 O solutions were added to 15 ml OLAM, 0.95 ml OLAC (Oleic acid 90%, Aldrich) in a 3-neck flask.The solution was mixed using magnetic stirring at 300 rpm and degassed at 50 °C for 30 min, followed by N 2 purging.The mixture was heated at 140 °C for 30 min with magnetic stirring and cooled to 25 °C.The nanoparticles were precipitated using 25 ml ethanol and 5 ml isopropyl alcohol followed by centrifugation at 8000 rpm for 5 min.The particles were redispersed in hexanes, and the precipitation process was repeated twice more.In each step, the clear supernatant was discarded, and the final particles were dispersed in 10 ml hexanes.
Synthesis of Pd NPs: Pd NPs were synthesized using a previously reported procedure with a slight modification. [26]307.9 mg of Palladium 2,4pentanedionate (99%, Aldrich) was dissolved in the mixture of 19 ml TDE (1-Tetradecene, 92%, Aldrich) and 21 ml ODE (1-Octadecene, 90%, Thermofisher).0.84 ml OLAM and 1.6 ml OLAC were added to the mixture and magnetically stirred at 500 rpm.After degassing for 30 min, 1.16 ml Trioctylphosphine (90%, Thermofisher) was injected under a vacuum, and the mixture was heated at 50 °C for 30 min, followed by N 2 purging.Under static nitrogen, the mixture was heated at 25 °C min −1 to 280 °C; the temperature at which the mixture of TDE and ODE starts to reflux.The mixture was heated at 280 °C for 15 min with magnetic stirring and cooled to 25 °C.The nanoparticles were recovered by precipitation using 25 ml ethanol and 5 ml isopropyl alcohol followed by centrifugation at 8000 rpm for 5 min.In each step, the clear supernatant was discarded, and the particles were redispersed in hexanes and 10 μl OLAM.The precipitation process was repeated twice more, and the final particles were dispersed in 10 ml hexanes.
Preparation of Supported NPs: NPs (Au, Pd, and Au@AuPd) were loaded onto -Al 2 O 3 (97%, STREM) calcined in air at 700 °C for 24 h using a previously reported procedure with a slight modification. [26]After dispersing the 600 mg of -Al 2 O 3 powders in 30 ml hexane with vigorous magnetic stirring for 30 min, the appropriate volume of NPs solution (0.4 wt% metal loading) was added into the mixture and stirred for 30 min.Here, the concentration of nanoparticles dispersed in hexane was determined based on ICP-MS and TGA measurements.After the deposition of NPs on Al 2 O 3 powders in hexane, the catalysts (Au/-Al 2 O 3 , Pd/-Al 2 O 3 , and Au@AuPd/-Al 2 O 3 ) were recovered by centrifugation at 3000 rpm for 3 min and dried at room temperature overnight.Organic ligands adsorbed on the surface of catalysts were removed by rapid heat treatment in air at 700 °C for 30 s.
Catalyst Characterization: The morphologies of NPs were characterized by transmission electron microscopy (TEM) using a JEM-2100F (JEOL, Japan) operated at an accelerating voltage of 200 kV and a FEI Titan environmental transmission electron microscope (ETEM) G2 at an acceleration voltage of 300 kV.Samples were prepared by drop-casting onto ultrathin carbon films and lacey carbon films supported on Cu grids for NPs and NPs/-Al 2 O 3 powders, respectively.The particle size distributions and d-spacing of NPs were determined based on the images of TEM and HRTEM with at least 100 particles per sample.The elemental distribution and chemical compositions of NPs were characterized by STEMenergy dispersive X-ray spectroscopy (STEM-EDS) using a Titan cubed G2.The optical properties of the NPs were characterized by UV-vis spectroscopy.For the measurement of UV-vis spectra, NPs solution was diluted to the concentrations of 0.05 mg ml −1 .The absorbance spectra were measured on UV-3150, SHIMADZU in the range from 300 to 800 nm.X-ray photoelectron spectroscopy (XPS) was carried out on Nexsa G2 (Thermofisher Scientific).Spectra were acquired using a monochromated, lowpower Al Kɑ source operating with a spot size of 200 μm.Samples were prepared by drop-casting 0.2 mg of NPs onto a silicon wafer.To obtain quantitative compositional information on NPs, we fitted and deconvoluted the peaks of the XPS spectra using CasaXPS software version 2.3.25.The collected binding energies were referenced to the C1s binding energy of 284.8 eV.The final metal loading on -Al 2 O 3 was determined by inductively coupled plasma mass spectrometry (ICP-MS, 7700 Series, Agilent Technologies).
FDTD Simulation: Computational simulations with a finite-difference time-domain (FDTD) method from FDTD Solutions (Lumerical Solutions, Ansys) were performed to analyze the electric field distribution around the Au@AuPd NPs under the light source with a wavelength of 518 nm.For each simulation, six spherical Au@AuPd models having AuPd with different Pd contents on the shell of the Au core were created.The diameters of each Au@AuPd nanoparticle were selected from the average diameters measured in individual TEM results.Pd contents and Pd coverage of each nanoparticle were based on the values in the experimental results.The simulation domain of 620 × 620 × 800 nm 3 along the x-, y-, and z-axis was used for simulation and was calculated using perfectly matched layers with 3D boundary conditions.2D frequency domain field and power were monitored in the x-y plane with a monitor across the center of each nanoparticle.The source of the plane wave was injected into the nanoparticle with the axis perpendicular to the monitor, and the electric field vector was in the x-axis direction.
Catalytic Reaction: To conduct catalytic benzyl alcohol oxidation, 0.2 mmol of benzyl alcohol with 2 ml of trifluorotoluene and 160 mg of catalysts were placed in a 50 ml custom-made quartz beaker.The reaction mixture was heated at 55 °C for 8 h and magnetically stirred at 800 rpm.A 100 W Halogen lamp (FOK-100W, Fiber-Optic-Korea) was used as a light source, where a band-pass filter (460 nm CWL, 200 nm FWHM, Hoya B460, Edmund Optics) was used to screen the ultraviolet spectrum and pass only the light with a wavelength range of 325-700 nm.A water filter consisting of a 2 mm thick layer of water sandwiched between two quartz plates was placed on the top of the reactor to prevent heating due to an infrared incident.The intensity of light measured in the reactor was 3.41 mW cm −2 without filters and 1.07 mW cm −2 with filters.All reaction experiments were performed at the same temperature (55 °C) and time (8 h) in both dark and light conditions, while the temperature of the reaction system was carefully controlled using an oil bath with a K-type thermocouple.To ensure the reliability and reproducibility of the data, we repeated every experiment twice in a dark condition and three times in a light condition.The composition of the reaction products filtered through a Millipore filter (pore size 0.2 μm) was analyzed by a gas chromatograph (GC, YoungIn Chromass, Ar carrier gas).The products were identified by comparing each GC spectrum with those of standard, and every data was repeatedly acquired twice by GC.

Figure 1 .
Figure 1.TEM images of a) Au NPs, b) Pd 3 Au 97 NPs, c) Pd 10 Au 90 NPs, d) Pd 20 Au 80 NPs, e) Pd 30 Au 70 NPs, and f) Pd 45 Au 55 NPs.The insets of (a-f) show the TEM images of nanoparticles with a 5 nm scale.The STEM-EDS maps of g) Pd 10 Au 90 NPs, h) Pd 20 Au 80 NPs, and i) Pd 30 Au 70 NPs.

Figure 2 .
Figure 2. Optical properties of NPs.a) Normalized UV-vis absorbance spectra for Au and Au@AuPd NPs with various Pd compositions.b) The calculated peak height difference between the absorbance values of Au and Au@AuPd NPs at 518 nm and the absorbance value of Pd 45 Au 55 NPs at 518 nm in the normalized UV-vis spectra.

Figure 3 .
Figure 3. XPS spectra of a) Au 4f and b) Pd 3d regions for the Au@AuPd NPs with different Pd compositions.c) Actual Pd compositions in the Au@AuPd NPs analyzed from ICP-MS and XPS measurements compared to the target composition.d) Calculated Pd composition of the shell on the surface of the Au core NP based on XPS data.

Figure 4 .
Figure 4. a) Normalized valence band (VB) spectra of Au, Pd, and Au@AuPd NPs with various Pd compositions obtained by XPS.b) Calculated d-band center energy relative to the Fermi level estimated from VB spectra.
order to keep the number of d-electrons constant, the d-states move up toward the Fermi level with a concomitant upshift of the d-band center.Therefore, the modified VB and d-band center clearly confirm the formation of the AuPd alloy phase on the surface of Au@AuPd NPs.Besides, the fact that the d-band center of the Pd 45 Au 55 NPs is comparable to the value of the bare Pd NPs, −1.57eV, demonstrates that the shell of the Pd 45 Au 55 NPs consists of Pd layer on the surface Au NPs.In contrast, the shell is composed of ultrathin AuPd alloy in the NPs with Pd content below 30 at% (i.e., Pd 3 Au 97 , Pd 10 Au 90 , Pd 20 Au 80 , and Pd 30 Au 70 NPs).

Figure 5 .
Figure 5. a) Conversion for the oxidation of benzyl alcohol to benzaldehyde with the Au@AuPd/-Al 2 O 3 catalysts of various Pd compositions of the shell under visible-light irradiation (red bar) and in the dark (gray bar).Reaction conditions: 0.2 mmol of benzyl alcohol, 160 mg of catalysts in trifluorotoluene solvent at 55 °C for 8 h.b) The light-enhanced conversion rate of Au@AuPd NPs determined by the ratio of the difference in conversion values under dark condition and light irradiation to the conversion value under dark condition.c) Schematic showing the mechanisms for the atmospheric benzyl alcohol oxidation to benzaldehyde on Au@AuPd NP. d) Schematic illustration of the charge transfer from Au to Pd, inducing surface charge heterogeneity in the dark condition.Subsequently, in the presence of visible-light irradiation, a further increase in surface charge heterogeneity occurs due to the generation of hot electrons and holes, followed by charge transfer to the lowest unoccupied molecular orbital (LUMO) of adsorbates.
, Pd 10 Au 90 NPs show a remarkable photo-induced activity enhancement of 186%, surpassing those of the Au NPs and Au@AuPd NPs with other Pd compositions (i.e., Pd 3 Au 97 , Pd 20 Au 80 , Pd 30 Au 70 NPs and Pd 45 Au 55 NPs) by 1.7 to 6.0 times.In contrast, for bare Pd NPs, no activity enhancement was observed upon light irradiation (Figure S9, Supporting Information).These results suggest the utilization of the photo-induced plasmon energy from the Au core NPs in the benzyl alcohol oxidation process and highlight the superior effectiveness of the ultrathin AuPd shell configuration for the solar-tochemical energy conversion.In the conventional atmospheric benzyl alcohol oxidation process in the presence of O 2 , the conversion of benzyl alcohol to benzaldehyde involves a series of three essential steps: 1) adsorption of benzyl alcohol to the surface active sites, 2) cleavage of the O─H bond in the alcohol molecule, leading to the formation of an alkoxy-intermediate (R─CH 2 ─OH ad + O* ad → R─CH 2 ─O* ad + OH* ad ), 3) hydrogen abstraction from the alkoxy-intermediate for the subsequent formation and desorption of the benzaldehyde (R─CH 2 ─O* ad + OH* ad → R─CH═O + H 2 O) (Figure