Research Article

Engineering bunched Pt-Ni alloy nanocages for efficient oxygen reduction in practical fuel cells

See allHide authors and affiliations

Science  15 Nov 2019:
Vol. 366, Issue 6467, pp. 850-856
DOI: 10.1126/science.aaw7493

Nanocage-chain fuel cell catalysts

The expense and scarcity of platinum has driven efforts to improve oxygen-reduction catalysts in proton-exchange membrane fuel cells. Tian et al. synthesized chains of platinum-nickel alloy nanospheres connected by necking regions. These structures can be etched to form nanocages with platinum-rich surfaces that are highly active for oxygen reduction. In fuel cells running on air and hydrogen, these catalysts operated for at least 180 hours.

Science, this issue p. 850

Abstract

Development of efficient and robust electrocatalysts is critical for practical fuel cells. We report one-dimensional bunched platinum-nickel (Pt-Ni) alloy nanocages with a Pt-skin structure for the oxygen reduction reaction that display high mass activity (3.52 amperes per milligram platinum) and specific activity (5.16 milliamperes per square centimeter platinum), or nearly 17 and 14 times higher as compared with a commercial platinum on carbon (Pt/C) catalyst. The catalyst exhibits high stability with negligible activity decay after 50,000 cycles. Both the experimental results and theoretical calculations reveal the existence of fewer strongly bonded platinum-oxygen (Pt-O) sites induced by the strain and ligand effects. Moreover, the fuel cell assembled by this catalyst delivers a current density of 1.5 amperes per square centimeter at 0.6 volts and can operate steadily for at least 180 hours.

Platinum (Pt) is the most active electrocatalyst for the oxygen reduction reaction (ORR) in fuel cells and metal-air batteries with promising stability (13). Nevertheless, the state-of-the-art Pt catalysts still lack activity and stability with respect to the cost and availability for large-scale commercial implementation (4, 5). Engineering the near-surface composition of nanostructured Pt alloys represents one promising approach to enhance the electrocatalytic performance of Pt-based electrocatalysts, in which the exposure of highly active sites with optimum performance can be maximized (6, 7). Adding other transition metals can enhance the catalytic performance via ligand and strain effects through modifying the binding strength of Pt-oxygen intermediates (810).

The introduction of open nanostructures, including hollow and porous nanoparticles such as nanocages (NCs) and nanoframes, may help in achieving this goal and also enhance mass transfer (11). Porous metal structures usually exist as nanoparticles and typically do not display long-term stability because they agglomerate or detach from the (usually carbon) support (12, 13). By contrast, one-dimensional (1D) nanostructures and their assemblies can exhibit enhanced stability because of their inherent anisotropic, higher flexibility and higher conductivity (14). The anisotropy of 1D nanostructures can lead to greater surface contact with the usual carbon support, thus resulting in high stability (15). Recently, 1D Pt-based alloy nanostructures have been explored for their promising ORR activity and stability (16, 17). However, these solid 1D nanostructures contain a substantial proportion of noble metals in the bulk versus at the surface, which limits the noble-metal utilization (18). Moreover, although desirable ORR performance has been established in model studies with, for example, rotating disk electrodes (RDEs), relatively few works have studied nanostructured Pt electrocatalysts in a complete cell configuration (19). The limitations in comparing performance at the RDE level with those at the cell level are known (20). To this end, it follows that optimum ORR catalysts based on Pt should be designed by having a high Pt utilization efficiency (porous nanostructures), a favorable nanoscale chemical environment for ORR (chemical alloying) with an optimized structure in terms of the exposure of active sites, and stability derived from both the nanostructure itself as well as its interface with the support.

We demonstrate Pt-Ni bunched NCs (BNCs) in 1D form that exhibit superior ORR activity and durability compared with conventional Pt electrocatalysts. We first prepared 1D Pt-Ni bunched nanospheres (BNSs) by reducing Pt and Ni precursors with varying ratios in oleylamine by a one-pot solvothermal method. Treatment under acidic conditions selectively removed Ni species to leave 1D Pt-Ni BNCs with ultrathin walls composed of a Pt skin and a residual Pt-Ni alloy below this skin (Fig. 1A). The nanostructures with a Pt1.5Ni starting composition exhibited the highest mass and specific activities of 3.52 A mgPt−1 and 5.16 mA cmPt−2, respectively. These values are more than one order of magnitude higher than those of a commercial Pt/C catalyst [60 weight % (wt %), Johnson Matthey (JM)] and comparable to those of very recently reported Pt nanostructure catalysts (table S1). The catalyst also exhibited superb durability with a negligible activity decay (less than 1.5%) after 50,000 potential-scanning cycles. In addition, the Pt-Ni BNC electrocatalyst showed improved performance in a fuel cell test compared with commercial Pt/C, both in a hydrogen-air proton exchange membrane fuel cell (PEMFC) and an air-breathing PEMFC setup at room temperature and ambient pressure. In situ x-ray absorption fine structure (XAFS) and theoretical calculations revealed that the Pt-Ni alloy nanostructures had optimal adsorption energies of oxygen intermediates with respect to Pt.

Fig. 1 Structural and compositional characterizations of Pt1.5Ni-BNSs.

(A) Schematic illustration of the preparation of PtNi-BNCs. (B and C) TEM images of PtNi-BNSs. (D) Enlarged TEM image of the area indicated in (C) and the corresponding HRTEM images of the areas marked by yellow squares. The white dots and arrows indicate the lattice spacing. (E and F) Atomic-resolution HRTEM images. Orange bars indicate the edges of the lattice planes, yellow arrows indicate the surface steps, and yellow circles indicate the outermost atoms. (G) High-angle annular dark-field (HAADF)–STEM image and the corresponding EDS elemental mapping of PtNi-BNSs. (H) EDS profile of PtNi-BNSs (top) and STEM-EDS line-scanning profile (bottom) of a single nanoparticle. The inset shows the studied nanoparticle and the line-scanning analysis along the yellow arrow. a.u., arbitrary units.

Bunched nanospheres

The Pt-Ni BNSs were synthesized in nearly 100% yield from platinum(II) acetylacetonate [Pt(acac)2], nickel(II) acetylacetonate [Ni(acac)2], cetyltrimethylammonium bromide (CTAB), and oleylamine mixtures that were ultrasonicated and then heated at 180°C for several hours (see supplementary materials and fig. S1). Composition studies included Pt3Ni, Pt2Ni, Pt1.5Ni, Pt1Ni1, and PtNi2 samples, and the transmission electron microscopy–energy-dispersive spectroscopy (TEM-EDS) profiles, together with inductively coupled plasma optical emission spectrometry (ICP-OES) results, suggest complete reduction of the precursors because the Pt/Ni ratios of the final products were nearly the same as the ratios used in the synthesis mixture (fig. S2). We further studied the effect of the concentration of the various reagents, including CTAB and glucose, and the Pt/Ni ratio in the synthesis solution on the formation of the structures (figs. S3 to S6). The x-ray diffraction (XRD) patterns of Pt-Ni BNSs with varying Pt-Ni composition (PtxNiy-BNSs) confirmed the efficient alloy formation (fig. S5), indicating that the composition can be adjusted by judicious choice of the amount of Pt and Ni precursors (21). Only when no Ni was present during the synthesis were Pt nanowires (NWs) with a smooth surface obtained. The surfaces became rougher with the increasing Ni/Pt feeding ratio (fig. S6). We used TEM-EDS to monitor time-dependent reaction progress, which revealed the formation of Pt NWs before Ni inclusion. This sequence is expected, given the higher redox potential of Pt (figs. S7 and S8). Finally, stable Pt-Ni alloys were formed under solvothermal conditions (22).

Representative scanning electron microscopy (SEM) and TEM images emphasize the uniform 1D morphology of the as-synthesized samples for the Pt1.5Ni composition (labeled as PtNi-BNSs, and this one, after etching, leads to the best ORR activity) with a length of hundreds of nanometers consisting of BNSs (Fig. 1, B and C). High-resolution TEM (HRTEM) images reveal lattice fringes of 0.227 nm at the stems and 0.216 nm at the shells (Fig. 1D), corresponding to (111) planes of Pt and Pt-Ni alloys (23), respectively, suggesting that the alloyed Pt-Ni nanospheres would be connected by Pt-rich NWs. Scanning transmission electron microscopy (STEM)–EDS line scanning of the stems of different regions showed the existence of Ni in the stems and an average Pt/Ni ratio of roughly 5/1 (fig. S9). It seems that the nanospheres were Ni-rich compared with the stems, because the scans may not reveal the overall compositions, whereas Pt-rich alloys can be confirmed at the stems. Atomic-resolution HRTEM images depict the uneven surface of the PtNi-BNSs (i.e., Pt1.5Ni composition) and the occurrence of high-index (211) and (311) facets (Fig. 1, E and F). The presence of abundant high-index facets is favorable for ORR, because they exhibit distinct binding energies with oxygen-containing surface intermediates in the ORR mechanism (24). The presence of alloyed features is verified by STEM-EDS elemental mapping analysis, emphasizing the homogeneous distribution of Pt and Ni across the analyzed zone, which contains a Ni-rich core (Fig. 1G). The line-scanning analysis of PtNi-BNSs demonstrates that both Pt and Ni elements are dispersed uniformly across the whole sphere, with an atomic Pt/Ni ratio near 60/40, according to TEM-EDS profile (Fig. 1H). This result is consistent with the bulk elemental ICP-OES analysis that informed a Pt/Ni ratio of 62/38.

Bunched nanocages

We obtained Pt-enriched BNCs by selectively removing the more reactive Ni species through a simple acid-etching method. TEM images of the PtNi-BNCs (this sample is from the PtNi-BNSs after the etching) showed that the products were still highly dispersed without structural collapse after the etching treatment (Fig. 2A). The presence of internal voids was deduced from the observation of the darker walls of the NCs (Fig. 2B). The average wall thickness was ~2.2 nm (Fig. 2C), roughly corresponding to 11 atomic Pt layers. The lattice fringe of 0.221 nm was consistent with the (111) plane of a Pt-Ni alloy (23) and was slightly larger than that of PtNi-BNSs because of the decreased Ni content after etching. In addition, the high density of steps and edges was well preserved during the etching step (Fig. 2D). STEM-EDS elemental mapping analysis indicated that the skin structure of the wall in the Pt-Ni NCs was Pt-rich (Fig. 2E). Figure 2F shows the EDS-line scanning of a single sphere in PtNi-BNCs, further confirming a well-defined Pt-skin structure of the NCs formed after the Ni etching. The thickness of the outer shell was ~0.63 nm, corresponding to roughly two to three atomic Pt layers (25).

Fig. 2 Structural and compositional characterizations of PtNi-BNCs derived from etching of Pt1.5Ni-BNSs.

(A and B) TEM images of PtNi-BNCs. (C) Enlarged TEM image and the corresponding HRTEM images of the areas marked by yellow squares. (D) Atomic-resolution HRTEM image. Orange dotted lines indicate the atoms along the plane, and yellow arrows indicate the surface steps. (E) HAADF-STEM images. (F) EDS profile of PtNi-BNCs (top) and STEM-EDS line-scanning profile (bottom) of a single NC. The inset shows the studied NC and the line-scanning analysis along the yellow arrow. The red dashed-dotted lines and arrows indicate the thickness of outermost Pt layers.

On the basis of the above analysis, we propose that the walls of NCs are heterogeneous in composition, with Pt-rich skins on both the outer and inner surfaces and Pt-Ni alloy in the inner region of the wall. The precise Pt/Ni atomic ratio of PtNi-BNCs was revealed by the TEM-EDS profile as 80.7/19.3 (fig. S10, A and B), consistent with the ICP-OES result (81/19). The increased Pt/Ni ratio was likely caused by selective Ni leaching. The XRD diffraction peaks of PtNi-BNCs showed a small shift to lower angles compared with PtNi-BNSs (fig. S10C), which is consistent with the lower overall Ni content of the former. X-ray photoelectron spectroscopy (XPS) of Pt 4f revealed a shift of the Pt core levels to lower binding energies compared with Pt/C, likely because of electron donation from Ni to Pt (fig. S10D). The ligand effect would downshift the Pt d-band center, lowering the binding affinity between Pt and oxygen intermediates and thus enhancing the ORR activity (5).

Electrocatalytic activity

We evaluated PtNi-BNCs as ORR catalysts, and Fig. 3A presents cyclic voltammetry (CV) results of Pt/C (60 wt %, JM), Pt NWs/C, PtNi-BNSs/C, and PtNi-BNCs/C in a N2-saturated 0.1 M HClO4 solution (fig. S11). The CV curves for carbon-supported PtNi-BNCs were investigated as a function of the Pt/Ni ratio (fig. S12). The electrochemically active surface area (ECSA) measured by hydrogen underpotential deposition (HUPD) of PtNi-BNCs/C was 68.2 m2 gPt−1, which is substantially higher than that of PtNi-BNSs/C (43.5 m2 gPt−1), confirming the advantage of a hollow structure created by acid leaching. Moreover, the ECSA values for Pt/C, Pt NWs/C, Pt3Ni-BNCs/C, Pt2Ni-BNCs/C, Pt1.5Ni-BNCs/C, Pt1Ni1-BNCs/C, and PtNi2-BNCs/C (for convenience, all x and y values in the names of BNCs were labeled the same values as in the starting BNSs) were 56.5, 46.3, 37.4, 59.1, 68.2, 70.7, and 80.3 m2 gPt−1, respectively, demonstrating the enhanced ECSA of Pt-Ni BNCs with increasing Ni content. TEM images of Pt-Ni BNCs with varying Ni contents demonstrate the increased porosity with increasing Ni/Pt ratio (fig. S13), in agreement with the larger ECSA values.

Fig. 3 Electrochemical and fuel-cell performance of various Pt-based samples.

(A to D) CVs (A), LSVs (B), corresponding Tafel plots (C), and mass and specific activities (D) of Pt/C, Pt NWs/C, PtNi-BNSs/C, and PtNi-BNCs/C at 0.9 V (versus RHE). J, current density. (E) H2-air fuel cell polarization plots with Pt/C, PtNi-BNSs/C, and PtNi-BNCs/C as the cathode catalysts. (F) Stability test of H2-air and air-breathing fuel cells at 0.6 V.

We also measured the ECSA from the electro-oxidation of CO (CO stripping), given the sensitive nature of the Pt-alloyed surface, because the ECSA derived from the HUPD method would be lower than the real values because of the weakened binding interaction caused by the alloying effects (the ECSA obtained from the CO stripping should be unaffected, fig. S14). The ECSA values of Pt/C and PtNi-BNCs/C obtained from the CO stripping are 56.7 and 101.5 m2 gPt−1, respectively; thus, the ECSACO/ECSAHupd ratio measured for Pt/C was equal to 1 but was 1.49 for PtNi-BNCs/C, verifying the formation of a Pt-skin structure of PtNi-BNCs (4, 26). Figure 3, B and C, shows the linear sweep voltammetry (LSV) curves and Tafel plots, respectively, of Pt NWs/C, PtNi-BNSs/C, PtNi-BNCs/C, and the commercial Pt/C reference. The smaller Tafel slope obtained for PtNi-BNCs/C (54 mV decade−1) as compared with that of the other samples demonstrates the enhanced kinetics for the ORR. The mass and specific activities of Pt NWs/C (1.02 A mgPt−1 and 2.20 mA cmPt−2) and PtNi-BNSs/C (1.89 A mgPt−1 and 4.34 mA cmPt−2) at 0.9 V [versus reversible hydrogen electrode (RHE)] were much higher than those of the Pt/C reference (0.21 A mgPt−1 and 0.36 mA cmPt−2). Furthermore, PtNi-BNCs/C shows mass and specific activities of 3.52 A mgPt−1 and 5.16 mA cmPt−2, which are respectively ~17 and ~14 times higher than those of the Pt/C reference. In general, the catalytic performance was enhanced with increasing Ni/Pt ratio, which can be ascribed to the higher amount of exposed active sites achieved by Ni leaching (fig. S12). The inferior ORR activities of Pt1Ni1-BNCs/C and PtNi2-BNCs/C were ascribed to the structural collapse upon excessive Ni dissolution (fig. S13).

To further corroborate the ORR performance obtained at the RDE level, practical H2-air and air-breathing PEMFC tests were conducted (fig. S15) with fuel cells containing PtNi-BNSs/C and PtNi-BNCs/C at a loading of 0.15 mg cm−2 as a cathode catalyst material. The H2-air fuel cell of PtNi-BNCs/C delivers a current density of 1.5 A cm−2 at 0.6 V and achieves a peak power density of 920 mW cm−2 (Fig. 3E), outperforming the PtNi-BNSs/C-based (1.0 A cm−2 and 770 mW cm−2) and Pt/C-based (0.8 A cm−2 and 600 mW cm−2) systems. Considering comparable Pt loadings, this performance is among the best reported performance for Pt-based catalysts (table S1). In addition, the air-breathing PEMFC operated at room temperature and ambient pressure exhibited a current density of 170 mA cm−2 at 0.6 V, compared with 100 mA cm−2 for Pt/C, resulting in an improvement of 70% (fig. S15E). The H2-air and air-breathing PEMFCs were stably operated at a constant working voltage of 0.6 V for at least 180 hours with a negligible decay (<3%) of the output current densities (Fig. 3F).

Structural changes under electrochemical operation

To evaluate the stability under better defined conditions, Pt/C, PtNi-BNSs/C, and PtNi-BNCs/C were subjected to continuous cycling for 20,000 cycles between 0.6 and 1.1 V in an O2-saturated 0.1 M HClO4 solution. The ORR performance variations recorded every 4000 cycles show that the mass activity and ECSA of Pt/C declined by 61.9 and 59.4%, respectively, and the half-wave potential shifted by 54 mV to more negative values compared with that of the fresh sample (Fig. 4, A and B). By contrast, the high durability of PtNi-BNSs/C was evidenced by the much smaller loss in mass activity and ECSA by 13.7 and 9.4%, respectively, along with a 6-mV negative-shifted half-wave potential (Fig. 4, C and D). The PtNi-BNSs/C sample became hollow and porous after the stability test through electrochemical leaching (dealloying) (fig. S16, A to C), whereas the bunched architecture was destroyed to some degree compared with PtNi-BNCs/C. These results suggest a strong chemical difference between electrochemical dealloying and chemical etching, because structural collapse and Ni leaching caused substantial performance loss of PtNi-BNSs/C (fig. S16D). Comparatively, the performance loss of PtNi-BNCs/C is negligible, even after the prolonged durability test for 50,000 cycles (Fig. 4, E and F). The mass activity and ECSA only dropped by 1.3 and 1.1%, respectively, after the durability test compared with that of the fresh one, demonstrating the robustness of PtNi-BNCs/C as an ORR electrocatalyst. The composition and structure of the PtNi-BNCs/C catalysts after the durability test were further investigated. The hollow and core-shell (Pt-rich surface) structures of PtNi-BNCs/C were retained during the durability test (fig. S17), and TEM-EDS results revealed that Ni loss was negligible (fig. S18). The Pt-skin structure may have protected the electrocatalysts against further Ni leaching from the inner region of the wall.

Fig. 4 Durability performance of various catalysts.

(A, C, and E) LSV evolutions. The insets show the CV variations. (B, D, and F) Mass and specific activity evolutions for Pt/C [(A) and (B)], PtNi-BNCs/C [(C) and (D)], and PtNi-BNCs/C [(E) and (F)] before and after the durability test for various potential-scanning cycles.

We probed the dynamic changes in oxidation state and local coordination environment at ORR-relevant potentials by in situ XAFS. Figure 5A shows the potential-dependent Pt-L3 edge x-ray absorption near-edge spectra (XANES) of PtNi-BNCs/C, emphasizing the metallic state of Pt at applied potentials of 0.54, 0.7, and 0.9 V (versus RHE). A 5-nm Pt/C (TEC10E50E-HT) reference was used for comparison to reduce the influence of the size effect. For comparison, the Pt-L3 edge XANES of PtNi-BNSs/C is depicted in fig. S19. The normalized white-line intensities (μnorm) at the Pt-L3 edge increased with increasing applied potentials because of the chemisorption of surface oxygenated species (Fig. 5B) (27, 28). The trends in μnorm with potential differed between the catalysts and were used to understand the mechanistic origins of ORR activity (29). At 0.54 V, both PtNi-BNCs/C and PtNi-BNSs/C showed a higher μnorm than Pt/C, probably because of the chemisorption of *OH and/or *O species in the double layer region (30). With an increase of the potential to 0.7 V, the increase of μnorm [i.e., Δμnorm (0.7_0.54 V) = μnorm (0.7 V) − μnorm (0.54 V)] was larger for PtNi-BNSs/C and smaller for PtNi-BNCs/C in comparison to the change for Pt/C. When the potential was increased to 0.9 V, a negligible Δμnorm (0.9_0.7 V) was seen for PtNi-BNSs/C and PtNi-BNCs/C relative to Pt/C (Fig. 5, B and C). Together, PtNi-BNSs/C showed an initially higher Δμnorm from 0.54 to 0.7 V and subsequently an inhibited Δμnorm from 0.7 to 0.9 V relative to that of Pt/C, similar with previous reports (28, 30). However, PtNi-BNCs/C only showed a higher μnorm at 0.54 V and subsequently the suppressed Δμnorm over the whole range of 0.54 to 0.9 V compared with that of PtNi-BNSs/C and Pt/C. Thus, the in situ XANES characteristic of PtNi-BNCs/C may partly account for its particularly high ORR activity and durability through the inhibition of strongly bonded oxygenated species from 0.54 to 0.9 V relative to Pt/C and PtNi-BNSs/C.

Fig. 5 In situ XAFS and XANES analysis.

(A) In situ Pt-L3 edge XANES spectra for PtNi-BNCs, with an enlarged view of the area marked by the green square. (B) In situ potential-dependent normalized white-line peak intensities [μ(E)norm]. (C) Δμ-XANES spectra (0.9_0.7 V) of PtNi-BNSs and PtNi-BNCs. (D) First-shell EXAFS fitting in R space for spectral data at 0.54 V. FT[k2χ(k)], Fourier-transformed k2-weighted EXAFS. Exp., experimental.

The fitted extended x-ray absorption fine structure (EXAFS) spectra conducted at 0.54 V (Fig. 5D, figs. S19 to S21, and tables S2 to S5), a potential in the double-layer region with the minimum interference from the adsorption of *OH and/or *O species, showed bond lengths of Pt-Pt (BLPt-Pt) and coordination numbers of Pt-Ni (CNPt-Ni) in the PtNi-BNCs/C (BLPt-Pt = 2.69 Å; CNPt-Ni = 0.6), PtNi-BNSs/C (BLPt-Pt = 2.71 Å; CNPt-Ni = 1.0), and Pt/C (BLPt-Pt = 2.76 Å). The evidently shorter BLPt-Pt for both PtNi-BNCs/C and PtNi-BNSs/C than that for Pt/C could be well ascribed to the smaller size of Ni atoms compared with Pt thus inducing the global stress for PtNi-BNCs and PtNi-BNSs. However, compared with PtNi-BNSs/C, PtNi-BNCs/C exhibited a slightly smaller CNPt-Ni but a slightly shorter BLPt-Pt, which indicated an additional source of strain, presumably induced by structural defects, as demonstrated in a recent report (7). Additionally, the fitting results of EXAFS at 0.9 V revealed the absence of a definite Pt-O path for both PtNi-BNCs/C and PtNi-BNSs/C (tables S3 and S5), despite a clear electrochemical adsorption of oxygenated species on their surface. This lack of a Pt-O scattering peak in the EXAFS spectra for both PtNi-BNCs/C and PtNi-BNSs/C agrees with their XANES results and may be correlated with the surface disorder produced by the defects (steps and grain boundaries; see TEM analysis) and the corresponding disturbance to interfacial water structures (28, 30).

Interestingly, TEM examinations after the ORR stability test showed a nearly hollow structure but irregular shape for PtNi-BNSs/C, whereas dimensionally uniform hollow structures were observed for PtNi-BNCs/C (figs. S16 and S17). The differences in hollow microstructures between PtNi-BNSs/C and PtNi-BNCs/C were likely caused by the different corrosion methods for the Ni dissolution and the accompanied restructuring, that is, a moderate acid corrosion for PtNi-BNCs/C versus an electrochemical corrosion for PtNi-BNSs/C. Expectedly, these microstructural differences induced the distinct density and distribution of defects and correlated microstrain on the surface of catalysts, causing different electrocatalytic in situ XANES and EXAFS behaviors between PtNi-BNSs/C and PtNi-BNCs/C. Specifically, the irregular hollow structures in PtNi-BNSs/C easily created local overstrained domains that undermined the electrochemical stability of PtNi-BNSs/C to some extent, whereas the dimensionally uniform hollow structures in PtNi-BNCs/C were observed to be stable under the conditions we investigated. Thus, our work suggests that the fine regulation of microstructures by tailored synthesis could produce electrochemically durable hollow-structured ORR catalysts, although discrete hollow particles are generally less stable because of their high free energy.

Theoretical studies

On the basis of these experimental data, we performed density functional theory (DFT) calculations to understand the high ORR performance of PtNi-BNCs, using Pt3Ni-skin and Pt4Ni-skin models (Fig. 6). It is generally accepted that the atomic O adsorption energy ΔEO* can be used to evaluate the ORR activity, and the optimal value of ΔEO* is ~0.2 eV weaker than the one for Pt(111) (5, 31). ΔEO* is modulated by the d-band center (labeled as εd) of Pt, whereas the downshift or upshift of εd is tuned by surface lattice strain (noted as χ) and coordination environment (coordination number, noted as CN) (3234). High-index crystal facets such as Pt(211) and Pt(311) exhibit a compressive lattice strain (χ) relative to Pt(111), which brings a downshift of εd of surface Pt atoms and weakening of ΔEO*. Herein, DFT calculation is focused on investigating the synergistic effect of χ and CN on εd, and finally on ΔEO*, and the correlations between them (Fig. 6, A and B). Figure 6C corroborates that the Pt4Ni-skin has higher ORR activity than the Pt3Ni-skin, but activities of both are substantially improved relative to that of Pt(111). ΔEO* values are 0.17 and 0.11 eV weaker on Pt4Ni- and Pt3Ni-skin, respectively, than on Pt(111), which are near the optimal ΔEO* value (table S6). Both skins have a 2.5% compressive lattice strain (fig. S22 and table S7). The bridge sites of Pt(211) and Pt(311) have a stronger O binding than those on Pt(111), whereas the hexagonal close-packed sites of them adsorb O slightly less strongly. For the ORR that happened at the electrode surface in aqueous solution, the bridge sites of high-index crystal planes are preferentially occupied by water molecules, and the other sites account for the observed high ORR activity (figs. S23 to S26). In addition, the determined potential energy profiles demonstrate that Pt4Ni-skin has a lower overpotential for ORR than Pt3Ni-skin and Pt(111) (fig. S27 and table S8). On the basis of the above analysis, we propose that the synergistic effects of strain and coordination environment, incorporation of Ni, and the appropriate Pt/Ni ratio provide a properly weakened Pt-O binding strength, thus leading to the superior ORR activity of PtNi-BNCs (Fig. 6D). Furthermore, DFT calculations were also conducted to gain insight into the high stability of PtNi-BNCs (fig. S28). The binding energy of the surface Pt atoms in Pt4Ni- or Pt3Ni-skin models is higher compared with that of other [Pt(111), Pt(211), and Pt(311)] models, indicating higher stability of the Pt-skin structures under the condition of lattice strain and coordination environment. Also, it was found that the required activation energy for the spillover of surface Pt atoms is substantial, indicating that dissolution of surface Pt atoms is kinetically prohibited for Pt-skin structures, even in the presence of oxygen adsorbates (table S9). The above results agreed well with the observed robust structure of our catalyst during the durability test.

Fig. 6 Disentangling the correlation of lattice strain and ligand effects on the d-band center, and the atomic O adsorption energy from DFT calculations.

(A) DFT-determined correlation of d-band center and atomic O adsorption energy with the compressive lattice strain of Pt(111) facet. R2, coefficient of determination. (B) DFT-determined correlation between the d-band center and the generalized coordination number of surface Pt atoms of various Pt(hkl) planes. (C) DFT-determined correlation of atomic O adsorption energy with d-band center of surface sites. Pt(211)-B and Pt(311)-B denote the bridge sites of Pt(211) and Pt(311) planes, respectively. (D) Illustration of the synergistic effects derived from both the lattice strain and ligand effects in the catalysts.

Discussion

We prepared bunched Pt-Ni NCs and demonstrated that they are efficient and durable ORR electrocatalysts for fuel cells. The as-obtained bunched Pt-Ni alloy NCs showed high mass and specific activities of 3.52 A mgPt−1 and 5.16 mA cmPt−2, respectively, which are 16.8 times and 14.3 times higher than those of a commercial Pt/C catalyst. Our catalyst also exhibited robust stability with negligible activity decay even after 50,000 potential-scanning cycles. The H2-air fuel cell assembled by this Pt-Ni catalyst achieves a peak power density of 920 mW cm−2 and delivers a current density of 1.5 A cm−2 at a voltage of 0.6 V for at least 180 hours, demonstrating the great potential for practical application. In situ XAFS, theoretical calculation, and experimental results reveal that such excellent performance could be ascribed to the integration of a hollow structure and dimensional architecture in the bunched Pt-Ni alloy NCs. This work provides an effective strategy for the rational design of Pt alloy nanostructures and will help guide the future development of catalysts for their practical applications in energy conversion technologies and beyond.

Supplementary Materials

science.sciencemag.org/content/366/6467/850/suppl/DC1

Materials and Methods

Figs. S1 to S28

Tables S1 to S9

References (3554)

References and Notes

Acknowledgments: We thank S. Liao from South China University of Technology for support with the fuel cell test. We thank X. Guo from Huazhong University of Science and Technology for valuable discussions. We thank Y. Iwasawa for support with XAFS measurements and the New Energy and Industrial Technology Development Organization (NEDO), Ministry of Economy, Trade, and Industry (METI), Japan; XAFS measurements were performed with SPring-8 subject numbers 2018B7800 and 2019A7800. Supercomputing facilities were provided by Netherlands Organisation for Scientific Research (NWO). The authors also acknowledge the support of the Analytical and Testing Center of Huazhong University of Science and Technology for XRD, XPS, ICP-OES, SEM, and TEM measurements. Funding: This work is funded by the National Natural Science Foundation of China (21805104 and 21802048), the National 1000 Young Talents Program of China and the Fundamental Research Funds for the Central Universities (2018KFYXKJC044, 2018KFYYXJJ121, and 2017KFXKJC002), and the Start-up Research Foundation of Hainan University [KYQD(ZR)1908]. X.W.L. acknowledges funding support from the National Research Foundation (NRF) of Singapore via the NRF Investigatorship (NRF-NRFI2016-04). Author contributions: B.Y.X. and X.W.L. conceived the idea and designed the experiments. X.T. and L.W. carried out the sample synthesis, characterization, and measurements. X.Z. performed the in situ XAFS measurements. Y.-Q.S. and E.J.M.H. provided theoretical calculations. X.T. and D.D. performed the fuel cell measurements. B.C., H.W., and H.L. participated in the discussion of the experimental results. X.T., X.W.L., and B.Y.X. co-wrote and revised the manuscript. All the authors contributed to the overall scientific interpretation and edited the manuscript. X.T., X.Z., and Y.-Q.S. contributed equally to this work. Competing interests: The authors declare no competing financial interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper or the supplementary materials.

Stay Connected to Science

Navigate This Article