Molybdenum‐doped ordered L10‐PdZn nanosheets for enhanced oxygen reduction electrocatalysis

Ultrathin Pd‐based two‐dimensional (2D) nanosheets (NSs) with tunable physicochemical properties have emerged as promising candidate for oxygen reduction reaction (ORR). Unfortunately, structurally ordered Pd‐based NSs can be hardly prepared as high temperature annealing (>600°C) is necessary for disorder to order phase transition, making it a considerable challenge for morphology control. Herein, a new class of ultrathin structurally ordered Mo‐doped L10‐PdZn NSs with curved geometry and abundant defects/lattice distortions is reported as an efficient oxygen reduction electrocatalyst in alkaline solution. It is found that Mo(CO)6 serves as reducing agent and Mo source to generate the unique ordered 2D morphology, which leads to the significantly modified electronic structure. The developed L10‐Mo‐PdZn NSs exhibit excellent ORR mass activity of 2.6 A mgPd−1 at 0.9 V versus reversible hydrogen electrode, 31.5 and 17.6 times higher than those of Pd/C and Pt/C, respectively, outperforming most of the reported Pd‐based ORR electrocatalsyts. Impressively, L10‐Mo‐PdZn NSs is extremely stable for ORR, with only 2.3% activity loss after 10 000 potential cycles. Density functional theory study suggests that ordered L10 structure and Mo doping can raise the vacancy formation energy of Pd atom and thus promote the ORR stability.


INTRODUCTION
The sluggish kinetics of oxygen reduction reaction (ORR) is the key bottleneck limiting the performance of fuel cells or metal-air batteries. [1][2][3] To enhance the performance and advance the practical applications of these energy conversion devices, numerous efforts have been devoted to developing efficient ORR electrocatalysts. To date, platinum (Pt) and its alloy (Pt-M, M = Fe, Co, Ni, etc.) nanocrystals (NCs) are the most active ORR electrocatalysts. [4][5][6] Yet, the limit reserve, high cost, and unsatisfactory stability of Pt and Pt-M NCs arise as other concerns for widespread commercializations. [7][8][9] It is necessary to develop alternative electrocatalysts with respectable ORR activity and stability.
Owing to the similar electronic structure with Pt, and relatively higher abundance of Palladium (Pd) on earth, Pd-M NCs have recently attracted much research interest as ORR electrocatalysts. [10][11][12][13] Pd-based two-dimensional (2D) structures, with the advantages of high atomic utilization efficiency and larger interfacial areas contacting with support compared to 0D nanoparticles (NPs), [14][15][16] have become promising candidate electrocatalysts for ORR. For instance, PdMo metallenes, 14 and porous Pd metallenes, 17 have displayed much enhanced ORR activity in both halfcell and Zn-air battery tests. Nevertheless, another obstacle is the durability issue of these Pd-M NCs, which can be ascribed to the less stabilized M in disordered Pd-M alloy, and the M atoms will be etched away during electrochemical tests. 18 By contrast, structurally ordered intermetallic PtM and PdM NCs, due to the high cohesive energy and negative structural formation energy (E f ), display much improve electrochemical stability and thus have attracted a broad range of research interest. 19,20 In principle, the more negative E f would lead to the higher vacancy formation energy (E vac ). For instance, our recent work demonstrates that the E vac of M atom in L1 0 -ordered structure, that is, Pt/M = 1/1, is much increased compared to that in disordered A1 counterpart. 21 Therefore, designing ultrathin 2D Pd-based nanosheets (NSs) together with ordered L1 0 structure would be of significance to improve the ORR activity and stability simultaneously. Unfortunately, the preparation of L1 0 structure usually requires high temperature annealing (>600 • C), which will lead to agglomeration of the NCs and make it difficult for morphology control. As a result, the reports on Pt-or Pd-based 2D NSs with ordered L1 0 structure are much limited. Meanwhile, it has proven that doping Pt-or Pd-based NCs with a third metal can not only improve the ORR activity by tuning the local strain and electronic structure but also stabilize the structure and enhance the stability. For example, Modoped PtNi NCs show impressive ORR activity and stability because of the stabilization of Pt/Ni atoms against dissolution. 22,23 Besides, the cohesive energy of Mo is as large as 6.82 eV/atom, much higher than Pd (3.89 eV/atom) and Zn (1.35 eV/atom), which may improve the stability from the perspective of thermodynamics.
Herein, 2D Mo-doped L1 0 -ordered PdZn (L1 0 -Mo-PdZn) NSs with ultrathin thickness (∼2.1 nm) are developed as ORR electrocatalysts. Compared to Fe and Cu ions, which may catalyze Fenton reaction to generate reactive oxygenate species, 24,25 Zn is an antioxidative element and can restrain Fenton reaction. 21 Meanwhile, the E f of L1 0 -PdZn intermetallics is as large as −0.575 eV/atom (compared to −0.066 eV/atom for L1 0 -PdFe and −0.124 eV/atom for B2-PdCu), which may render much improved stability. The developed L1 0 -Mo-PdZn NSs demonstrate promising ORR activity and stability in alkaline electrolyte, with a high mass activity (MA) of 2.6 A mg Pd −1 at 0.9 V versus reversible hydrogen electrode (RHE), nearly 17.6 times higher than that of commercial Pt/C. Meanwhile, negligible activity loss and structural change are observed after 10 000 potential cycles. The enhanced ORR activity is ascribed to the modified electronic structure of Pd and high electrochemical active surface area (ECSA), while the stability is attributed to the increased E vac of Pd as a result of L1 0 -ordering and Mo doping.

RESULTS AND DISCUSSION
Mo-doped L1 0 -PdZn (denoted as L1 0 -Mo-PdZn) NSs were prepared by a wet-chemical co-reduction approach (for details see Supporting Information, SI). Briefly, the L1 0 -Mo-PdZn NSs were produced in oleylamine (OAm) at 300 • C in the presence of Mo(CO) 6 . Mo(CO) 6 can decompose into carbon monoxide (CO) gas and Mo atoms, where CO serves as a capping agent for the controlled growth of 2D morphology, and Mo atoms can dope into PdZn lattice. Under high temperature (300 • C), Pd and Zn atoms can diffuse and rearrange into L1 0 intermetallic structure. As a comparison, disordered A1-PdZn NSs and ordered L1 0 -PdZn NSs were prepared at 280 and 300 • C in the presence of CO gas, respectively. The as-prepared NSs were loaded onto carbon (Vulcan XC-72, denoted as A1-PdZn/C, L1 0 -PdZn/C, and L1 0 -Mo-PdZn/C NSs) and washed with acetate acid to remove the residual OAm for further use. The morphology of L1 0 -Mo-PdZn NSs was characterized by transmission electron microscopy (TEM) and highangle annular dark-field-scanning TEM (HAADF-STEM). 2D sheets with lengths up to several hundred nanometers are the dominant product ( Figure 1A and Figure S1). Some winkles can be clearly observed within L1 0 -Mo-PdZn NSs ( Figure 1B), indicating the high flexibility similar to graphene, which may generate strain as a result of curved geometry ( Figure S2). 14,16,17 Besides, the contrast of  Figure 1C), with only 7-10 atomiclayer. Meanwhile, the lateral size is estimated in the range of 45-200 nm. Atomic force microscopy was employed to further measure the thickness of L1 0 -Mo-PdZn NSs. As shown in Figure 1D, the height profile is about 2.7 nm, slight higher than that measured in STEM results, which may be ascribed to the residual capping agent on the NSs. 26,27 Figure 1E shows the high resolution TEM (HRTEM) image of the lateral region of L1 0 -Mo-PdZn NSs. The spacing lattice fringe is measured to be 0.219 nm, corresponding to (111) facet of L1 0 -PdZn. Meanwhile, several grain boundaries in L1 0 -Mo-PdZn NSs can be clearly observed ( Figure 1E). The corresponding fast Fourier transform (FFT) pattern also confirms the ploy-crystalline nature of  Figure 1F), where the distortion of lattice fringes along the curved edge is clearly observed. Noteworthy, these grain boundaries and lattice distortions can provide plenty of defect sites, which have been proven to benefit the catalytic activity. 28,29 The atomic-scale structure of L1 0 -Mo-PdZn NSs was further characterized by spherical aberration (Cs)-corrected HAADF-STEM ( Figure 2). Figure 2A shows the atomic arrangement of L1 0 -Mo-PdZn NSs viewed along (11-2) zone axis. The ordered structure is clearly indicated by the bright (Pd) and dark (Zn) atoms and their periodic arrangement along < 1-10 > direction. The atomic arrangements are in good agreement with the standard L1 0 structure along (11-2) projection ( Figure 2B). Interestingly, the lattice distortion, with a twisted angle of ca. 4.4 • , can be identified as a result of curved geometry and is in accordance with HRTEM results ( Figure 1F). The inset of Figure 2A is the corresponding FFT pattern of this area, and the presence of (110) superlattice spots is observed. The spacings of lattice fringes are measured to be 0.309 and 0.220 nm, cor-responding to (110) and (111) facets, respectively. Besides, the periodic L1 0 structure is further confirmed by HAADF line-scan profile ( Figure 2C). The alternating intensity profile of line 2 reveals the ordered arrangement of Pd and Zn atoms, while line 1 shows the presence of Pd atom only. These results confirm the presence of L1 0 ordered structure with some lattice distortion. HAADF-STEM energydispersive X-ray spectroscopy (EDX) elemental mappings of L1 0 -Mo-PdZn NSs demonstrate the uniform distribution of Pd/Zn/Mo elements throughout the nanosheet ( Figure 2D). The molar ratio of Pd/Zn/Mo in L1 0 -Mo-PdZn NSs is measured to be 52/42/6 by the TEM-EDX analysis ( Figure S3), in line with the result obtained by the inductively coupled plasma optical emission spectrometer (ICP-OES, 57/41/2). Figures S4 and S5 show the morphology of A1-PdZn NSs and L1 0 -PdZn NS. Unlike L1 0 -Mo-PdZn NSs, the thickness of A1-PdZn NSs and L1 0 -PdZn NSs becomes thicker, and some aggregates are observed, which can be ascribed to the low CO concentration in the OAm solution. Therefore, the in situ generated CO molecule via the decomposition of Mo(CO) 6 is vital to prepare ultrathin 2D NSs. The Pd/Zn  Figures S6-S8).
X-ray diffraction (XRD) was performed to characterize the crystal structure of the obtained NSs. As shown in Figure 3A, after reacting at 280 • C, the A1-PdZn/C NSs show similar diffraction peaks with standard face-centered cubic Pd (PDF#65-2867), which can be attributed to the insufficient reduction and diffusion of Zn element at 280 • C. The diffraction peaks reveal a slight right-shift as a result of the incorporation of smaller Zn atom. By contrast, when reacted at 300 • C, the diffraction patterns become different, where the peak at around 68 • disappears, while other peaks at 63 • and 73 • appear, suggesting the transition from disordered A1 structure to ordered L1 0 structure. Indeed, the diffraction peaks of L1 0 -Mo-PdZn/C and L1 0 -PdZn/C can be well indexed to the L1 0 -PdZn structure (PDF#65-9523). Importantly, the ultrathin 2D morphology of Pd-Zn NSs is well maintained after phase transition ( Figure 1A), which makes it possible to take the advantages of ultrathin 2D architecture and ordered structure for electrocatalysis.
X-ray photoelectron spectroscopy (XPS) was carried out to analyze the valence state and electronic structures of Pd in the studied samples. All the binding energies were calibrated with C 1s at 284.8 eV as a reference. Figure 3B,C show the high resolution Pd 3d XPS spectra, where the binding energies at around 336 eV and 341 eV can be assigned to 3d 5/2 and 3d 3/2 orbitals of metallic Pd. Meanwhile, the peaks at around 338 eV and 343 eV are associated with Pd (II). These results indicate the coexistence of metallic Pd (0) and oxidative Pd (II) in Pd/C, A1-PdZn/C, L1 0 -PdZn/C, and L1 0 -Mo-PdZn/C NSs. Specifically, the ratio of Pd (0) to Pd (II) in L1 0 -Mo-PdZn/C NSs is calculated to be 76.5/23.5, much higher than those in Pd/C (43/57), A1-PdZn/C NSs (60.5/39.5), and L1 0 -PdZn/C NSs (67/33) ( Figure 3D). Such results demonstrate that phase transition from A1 to L1 0 as well as Mo doping can improve the antioxidant capability of L1 0 -Mo-PdZn/C NSs. In addition, the Pd 3d 5/2 binding energies of A1-PdZn/C (336.2 eV), L1 0 -PdZn/C (336.1 eV), and L1 0 -Mo-PdZn/C (336.2 eV) NSs are higher than that in Pd/C (335.7 eV), which indicates a downshift of the d-band center after the incorporation of Zn. 14,30 Meanwhile, the Mo doping has little influence on the electronic structure (d-band center) compared to A1-PdZn/C and L1 0 -PdZn/C, which may be ascribed to the low Mo content and similar electron negativity of Mo (1.8) and Zn (1.6). The downshift of the d-band center of Pd can weaken the adsorption of oxygenated species, thus enhancing the ORR activity. 31 Figure S9 shows the high resolution Mo 3d XPS spectrum of L1 0 -Mo-PdZn/C NSs. Both of metallic state (Mo (0)) and oxidative state (Mo (VI)) are observed in XPS spectrum, in consistence with previous  Figure S10 shows the TEM image of Pd/C. The Pd nanoparticles distribute uniformly on the carbon support, with an average diameter of only 1.9 ± 0.2 nm. Pd NSs can also be prepared in the presence of CO gas. TEM image shows that Pd NSs is the major product, while some aggregates are observed as well ( Figure S11), similar to A1-PdZn NSs. The studied samples were first subjected to 50-100 cyclic voltammetry (CV) cycles between 0.0 and 1.0 V (vs. RHE) in N 2 -saturated 0.1 M KOH for activation. Figure 4A and Figure S12 show the stable CV curves of the studied electrocatalysts. The Pd-containing electrocatalysts display similar electrochemical behavior in 0.1 M KOH solution ( Figure 4A), where the redox peaks at the potential range of 0-0.4 V are associated with the underpotential adsorption and desorption of hydrogen (H upd ). At the potential range of 0.65-0.75 V, cathodic peaks associated with Pd reduction (reduction of Pd-OH ad ) can be clearly observed. Notably, the Pd-OH ad reduction peak of L1 0 -Mo-PdZn/C NSs positively shifts by 70 mV compared to Pd/C, which indicates a weak oxygen affinity and may be beneficial to ORR by reducing the adsorption of oxygenated species. [31][32][33] By contrast, Pt/C reveals different electrochemical behaviors with more symmetric redox peaks of H upd and Pt-OH ( Figure S12).
CO stripping voltammetry was also carried out to study the surface electronic structures. Compared to Pd/C, L1 0 -Mo-PdZn/C NSs display a more negative CO F I G U R E 5 Oxygen reduction reaction (ORR) polarization curves of (A) A1-PdZn/C nanosheets (NSs), (B) L1 0 -PdZn/C NSs, and (C) L1 0 -Mo-PdZn/C NSs before and after 10 000 potential cycles. (D) Changes in mass activity (MA) of the studied electrocatalysts before and after 10 000 potential cycles. (E) Vacancy formation energy of Pd in A1-PdZn, L1 0 -PdZn, and L1 0 -Mo-PdZn. Insets are the atomic model used for DFT calculations, where gray, dark blue, and purple spheres represent Pd, Zn, and Mo atoms, respectively stripping peak ( Figure 4B), which suggests a down-shifted d-band center of surface Pd atoms and the weakened Pd-O binding energy. 34 Meanwhile, the CO stripping peaks of A1-PdZn/C and L1 0 -PdZn/C NSs are similar with that of L1 0 -Mo-PdZn/C NSs, implying a similar electronic structure, in accordance with CV and XPS results. The electrochemical active surface areas (ECSAs) of Pd/C, Pd/C NSs, A1-PdZn/C NSs, L1 0 -PdZn/C NSs, and L1 0 -Mo-PdZn/C NSs, calculated by CO stripping voltammetry, are 80.9, 31.6, 27.5, 21.3, and 60.4 m 2 g Pd −1 , respectively. The higher ECSA of L1 0 -Mo-PdZn/C NSs is ascribed to the thinnest thickness compared to A1-PdZn/C and L1 0 -PdZn/C NSs. Figure 4C shows the positive-going ORR polarization curves of the studied electrocatalysts in O 2 -saturated 0.1 M KOH. L1 0 -Mo-PdZn/C NSs demonstrate the highest halfwave (E 1/2 ) of 0.894 V, much higher than those of Pd/C (0.837 V), Pt/C (0.84 V), Pd/C NSs (0.85 V), A1-PdZn/C NSs (0.876 V), and L1 0 -PdZn/C NSs (0.869 V). The ORR activ-ity of each electrocatalysts is further quantified by calculating the kinetic current at 0.9 V and normalizing with Pd loading and ECSA ( Figure 4D). Impressively, the L1 0 -Mo-PdZn/C NSs deliver MA and specific activity (SA) of 2.6 A mg Pd −1 and 4.31 mA cm −2 , respectively. The MA of L1 0 -Mo-PdZn/C NSs is about 31.5 and 17.6 times higher than those of Pd/C (0.08 A mg Pd −1 ) and Pt/C (0.14 A mg Pt −1 ), respectively, representing one of the most active Pd-based electrocatalysts in alkaline electrolyte (Table S1). In addition, the MA of L1 0 -Mo-PdZn/C NSs is much superior to those of Pd/C NSs (0.465 A mg Pt −1 ), A1-PdZn/C NSs (0.96 A mg Pt −1 ), and L1 0 -PdZn/C NSs (0.65 A mg Pt −1 ), while the difference in SA of these three electrocatalysts is not as noticeable as the difference in MA ( Figure 4D). In this regard, the higher MA of L1 0 -Mo-PdZn/C NSs may originate from the ultrathin 2D structure and higher ECSA. The ORR Tafel slopes of the studied electrocatalysts are calculated ( Figure 4E). The obtained Tafel slope of L1 0 -Mo-PdZn/C NSs is 58 mV dec −1 , closed to those of Pd/C (70 mV dec −1 ), A1-PdZn/C (67 mV dec −1 ), and L1 0 -PdZn/C (72 mV dec −1 ) NSs. The similar Tafel slopes of Pd-containing electrocatalysts suggest the similar reaction mechanism that the migration of reaction intermediates is the rate-determining step (RDS). 35,36 As for Pt/C, the Tafel slope is 92 mV dec −1 , which implies the different RDS for Pt/C and Pd-containing electrocatalysts. The electron transfer numbers as well as H 2 O 2 production rate of L1 0 -Mo-PdZn/C NSs during ORR process were evaluated by rotating ring-disk electrode, as shown in Figure 4F. L1 0 -Mo-PdZn/C NSs demonstrate an H 2 O 2 yield below 5% and an electron transfer number of 3.9 in the range of 0.4-0.9 V, verifying the high selectivity of L1 0 -Mo-PdZn/C NSs for ORR. Noteworthy, L1 0 -Mo-PdZn/C NSs also shows comparable ORR activity to commercial Pt/C with a similar loading in 0.1 M HClO 4 ( Figure S13). The enhanced ORR catalytic activity of L1 0 -Mo-PdZn/C NSs could be ascribed to the weakened Pd-O binding strength, ultrathin 2D architecture with abundant grain boundaries/defect sites.
Stability tests were performed in O 2 -saturated 0.1 M KOH by cycling between 0.6-1.0 V versus RHE at a scan rate of 100 mV s −1 . All of Pd/C, Pt/C, and A1-PdZn/C NSs show noticeable activity loss after 10 000 potential cycles, with 12 mV, 27 mV, and 16 mV loss in E 1/2 , respectively ( Figure 5A and Figure S14). After A1 to L1 0 transition, the stability of L1 0 -PdZn/C NSs is much improved, and only 5 mV loss in E 1/2 is observed ( Figure 5B). Impressively, Mo doping can further enhance the stability, and negligible change in ORR polarization curves is observed after potential cycles ( Figure 5C). To be more quantified, we calculated the change in MA after potential cycles. L1 0 -Mo-PdZn/C NSs retain a MA of 2.54 A mg Pd −1 after stability test, while Pd/C NSs, A1-PdZn/C NSs, and Pt/C suffer from 26.7%, 58.4%, and 50% activity loss, respectively ( Figure 5D and Figure S15), further confirming the excellent stability of L1 0 -Mo-PdZn/C NSs. After potential cycles, L1 0 -Mo-PdZn/C NSs can maintain the 2D architecture, and the composition is nearly unchanged ( Figure S16). We performed Density functional theory (DFT) calculation to elucidate insights into the excellent stability of L1 0 -Mo-PdZn/C NSs. Vacancy formation energy (E vac ) of Pd atom is employed as stability descriptor, where a higher E vac suggests better stability. 37,38 As shown in Figure 5E, the E vac of Pd in A1-PdZn is 1.2 eV. After A1 to L1 0 transition, the E vac of Pd increases by 0.084 eV, suggesting improved antioxidative capability of Pd atoms due to the strong Pd-Zn interatomic interaction in L1 0 structure. Mo doping can further increase the E vac of Pd to 1.423 eV, rendering a much improve stability by inhibiting the oxidation and dissolution of Pd atoms. In addition, the ratio of Pd (0) to Pd (II), obtained by XPS results, is in good agreement with E vac of Pd in A1-PdZn, L1 0 -PdZn, and L1 0 -Mo-PdZn that L1 0 -Mo-PdZn shows the most antioxidative tendency. Besides, the 2D anisotropic architecture also helps improve the stability by reinforcing the interaction between PdZn NSs and carbon support. 26,39 Taking these factors together, the excellent stability of L1 0 -Mo-PdZn/C NSs is attributed to the unique L1 0 structure, Mo doping, and 2D architecture.

CONCLUSIONS
In summary, we have demonstrated the construction of Mo-doped, ultrathin, and defect-rich L1 0 -PdZn NSs as efficient ORR electrocatalysts in alkaline electrolyte. It is found that the in situ generated CO via the decomposition of Mo(CO) 6 is the key for the preparation of ultrathin 2D morphology. The developed L1 0 -Mo-PdZn NSs exhibit high ORR MA of 2.6 A mg Pd −1 at 0.9 V versus RHE, 31.5 and 17.6 times higher than those of Pd/C (0.08 A mg Pd −1 ) and Pt/C (0.14 A mg Pt −1 ), respectively. Notably, L1 0 -Mo-PdZn NSs display an excellent stability, and only 2.3% activity loss is observed after 10 000 potential cycles. The impressive ORR activity and stability make our L1 0 -Mo-PdZn NSs one of the best Pd-based ORR catalysts. According to XPS and CO stripping voltammetry, the outstanding catalytic activity can be attributed to the modified electronic structure of Pd atom that optimizes oxygen adsorption energy on Pd. DFT results demonstrate that the ordered L1 0 structure and Mo doping can simultaneously increase the vacancy formation energy of Pd atom, thereby inhibiting the oxiation and dissolution of Pd atoms and improving the stability. This work demonstrates a promising strategy for shape and structure controlled electrocatalysts with advanced activity and durability, which will be of great significane for energy conversion applications and beyond.

A C K N O W L E D G M E N T S
This work was financially supported by the National Nature Science Foundation of China (grant numbers: 22122202 and 21972051). The authors thank the Analytical and Testing Center of the Huazhong University of Science and Technology (HUST) for carrying out the transmission electron microscopy (TEM), X-ray photoelectron spectroscopy (XPS), and X-ray diffraction (XRD) measurements. Atomic-resolution high-angle annular darkfield (HAADF)-scanning TEM (STEM) was carried out on microscope Titan Themis G2 60-300 maintained by Southern University of Science and Technology Core Research facilities.

C O N F L I C T O F I N T E R E S T
The authors declare no conflict of interest.

D ATA AVA I L A B I L I T Y S TAT E M E N T
All data supporting the findings of this study are available within the article and the Supplementary Information file.