Programming ORR Activity of Ni/NiO x @Pd Electrocatalysts via Controlling Depth of Surface-Decorated Atomic Pt Clusters.

Carbon nanotube supported ternary metallic nanocatalysts (NCs) comprising Nicore-Pdshell structure and Pt atomic scale clusters in shell (namely, Ni@Pd/Pt) are synthesized by using wet chemical reduction method with reaction time control. Effects of Pt4+ adsorption time and Pt/Pd composition ratios on atomic structure with respect to electrochemical performances of experimental NCs are systematically investigated. By cross-referencing results of high-resolution transmission electron microscopy, X-ray diffraction, X-ray absorption, density functional theoretical calculations, and electrochemical analysis, we demonstrate that oxygen reduction reaction (ORR) activity is dominated by depth and distribution of Pt clusters in a Ni@Pd/Pt NC. For the optimum case (Pt4+ adsorption time = 2 h), specific activity of Ni@Pd/Pt is 0.732 mA cm-2 in ORR. Such a value is 2.8-fold higher as compared to that of commercial J.M.-Pt/C at 0.85 V (vs reversible hydrogen electrode). Such improvement is attributed to the protection of defect sites from oxide reaction in the presence of Pt clusters in NC surface. When adsorption time is 10 s, Pt clusters tends to adsorb in the Ni@Pd surface. A substantially increased galvanic replacement between Pt4+ ion and Pd/Ni metal is found to result in the formation of Ni@Pd shell with Pt cluster in the interface when adsorption time is 24 h. Both structures increase the surface defect density and delocalize charge density around Pt clusters, thereby suppressing the ORR activity of Ni@Pd/Pt NCs.

Introduction

Mitigation of global pollution level relies on the development of clean and sustainable energy sources. To meet the growing energy demands by new applications in modern society, high-efficiency energy-conversion technologies are urgent issues to be developed. With the consideration mentioned above, fuel cell is one of the most promising solution among the existing energy-conversion means. Oxygen reduction reaction (ORR) on cathodic nanocatalysts (NCs) incurs the highest energy barrier (∼0.3–0.4 V) among all components and is a pivotal performance-determining factor in a fuel cell.[1−3] Considering performances and durability of the materials, carbon-supported platinum-based (Pt/C) NCs with high Pt loading are commonly employed as cathodic catalysts in fuel cells.[4] In addition to material cost, scarcity, reactivity, and severe selectivity are long-standing issues dragging down the reaction kinetics of Pt catalysts in ORR. Therefore, development of low Pt loading catalysts is one the most important issues in the fuel cell device. Intensive researches have been done on less expensive and more abundant materials in the form of alloys,[5−8] core–shell structures,[9−11] nitrides,[12] nanoframes,[13] nanowires,[14,15] porous structures,[16,17] sandwich structures,[18] nanotubes,[19] etc. for ORR NCs. However, efficiency and stability of state-of-the-art advanced cathodic catalysts remain far from the commercial standards.

Among the existing nanostructure configurations, Mcore–Ptshell (M stands for transition metals with high oxygen affinity, e.g., Ni, Co, Pd) structured NCs with proper heteroatomic intermix and shell thickness (low Pt loading) are considered the most efficient design in terms of cost reduction, ORR activity facilitation, and stability improvement. From heterogeneous catalysis theory, redox kinetics of NCs are dominated by three fundamental effects including bifunctional mechanism, ligand effect, and lattice strain. Bifunctional mechanism arises when dissimilar surface atoms work together to interact with adsorption species. For adsorption in heteroatom sites, formation of incoherent bond vibration modes and presence of intermediate species in local coordination environment generate inequality in chemical stress at adsorption bonds and thus facilitate redox kinetics of NCs. Ligand effects are caused by the atomic vicinity of two dissimilar surface metal atoms. With difference in electronegativity, ligand effect induces charge transfer between two neighboring atoms and thus affects the band strength of the adsorbed molecule. Lattice strain is caused by differences in atomic arrangement between surface atoms and crystal structure underneath. It can turn compressive or expansive form in surface layer, therefore, localize (delocalize) electron in shell crystal to enhance (suppress) the ORR activity of NCs. In core–shell NCs, an active shell crystal with a compressive strain provides direct access for oxygen reduction and high oxygen affinity core provides low-energy pathways for the allocation and recombination of radicals (i.e., O*, OH*, and H*) into H2O. Both the two manners reduce the desorption barrier of intermediate species, thereby facilitating the ORR activity of NCs.

With proper composition and configuration, core–shell-structured NC seems a perfect design in ORR application. However, further optimization on ORR performance is limited due to the lack of room for strain or ligand effect engineering bound to a critical condition of the core crystal capping with a monolayer active material (noble metal in most cases) shell. Such a situation not only hinders the performance improvement but also the cost and durability of electrocatalysts. To further improve the utilization and reduce the loading of noble metal, decoration of active materials in the form of atomic clusters in core–shell NC surface is a possible strategy.[20−24] Due to the electronegative difference and lattice strain in interfaces, those atomic clusters localize electrons from neighboring atoms and thus weaken oxygen adsorption in NC surface. Meanwhile, formation of heterogeneously adjacent nanocomponents with different redox activity in intermediate steps offers facile pathways to boost ORR kinetics (activity) of NCs.

Surface oxide on noble metal inhibits the subsequent adsorption of oxygen molecules and is a known obstacle in ORR kinetics. In this work, a facile wet chemical reduction method with reaction time control is developed for the synthesis of ternary catalysts comprising Ni/NiO core and Pd shell with atomic Pt cluster decoration on the surface (namely, Ni@Pd–Pt). The end products show tunable surface oxophilicity and ORR activity in an alkaline environment (0.1 M KOH). We demonstrated that the ORR activity of experimental NCs is determined by the atomic rearrangement of Pt clusters in Ni@Pd crystal. By changing the adsorption time of Pt4+ ions before the reduction time, the oxophilicity of experimental NCs is controlled with respect to the Pt atoms diffusion in Pd shell region. A worthy finding is that the specific surface activity (SA) is 0.732 mA cm–2 (which is 2.8-fold as compared to that of “0.261 mA cm–2” for commercial Pt catalyst) when the adsorption time for Pt4+ ions in Ni@Pd surface is 2 h. The coverage effect of Pt clusters is revealed by changing Pt loading in the crystal growth recipe with optimized Pt4+ ion adsorption time (2 h) in Ni@Pd surface.

Results and Discussion

The surface morphologies of as-prepared NCs were analyzed by using high-resolution transmission electron microscopy (HRTEM). Figure compares the HRTEM images of experimental NCs (NPP-10 s, NPP-2 h, and NPP-24 h). The corresponding inverse Fourier transformed (IFT) images and line histogram of selected fringes are compared in insets. Variations in NCs shape are mainly attributed to the factors including differences in surface free energy, surface oxidation, and heteroatomic intermix in different facets during crystal growth. As shown in Figure a, the NCs are grown nearly in isotropic spheres with preferential orientation in (111) facet (d(111) 2.25 Å) in NPP-10 s. Truncated surface reveals the presence of terrace defects (denoted by yellow arrow) by heterogeneous nucleation and crystal growth in the presence of moderate lattice mismatch upon accommodation of Pt atoms in the topmost layer to form a multifaceted twin NC. Such a hypothesis is consistently revealed by the random displacement of interatomic distances in the NC surface (R1, R2, and R3 shown in line histogram). Hereby, with a short period of Pt4+ adsorption time (10 s), the majority of Pt atoms is posited in the surface defect sites of the Pd shell. Low-magnification image in Figure d and histogram show that NPP-10 s is grown in a broad size distribution ranging from 2 to 7 nm with an average particle size of 3.93 nm. Agglomeration of NC is attributed to the absence of stabilizers upon crystal growth. For NPP-2 h (Figure b), the average particle size is 4.7 nm. As compared to that of NPP-10 s, the surface defect is reduced (denoted by yellow arrows), as is consistently observed in the local structure regime in the X-ray absorption spectroscopy (XAS) analysis (in later section). Again, such a structure results from the galvanic replacement between Pt4+ ions and Pd in the crystal growth stage. In this event, substantial surface restructuring appears through strong homoatomic clustering between the Pt atoms and the resulting atomic scaled Pt cluster in the Pd shell region. Severe galvanic replacement between Pt4+ and Pd shell with atomic structure rearrangement is evidenced by the formation of a well crystal shell (denoted by yellow arrows) accompanied by slightly disordered core region (denoted by red arrow). Intercalation of Pt clusters in the shell region reduces the surface free energy and extended adsorption time consumes the residual Pt4+ ion in the reaction system before the addition of the reducing agent. Both phenomena reduce atomic diffusion in the heterogeneous interface (solid “NC”–liquid “solution” interface and solid NC–solid NC interface), leading to less agglomeration between the contacted NCs in the carbon nanotube (CNT) surface (Figure e). Formation of Pt clusters “inside” the Pd shell with reduced structure complexity is also revealed by line histogram of particle size (Figure e) and lattice constant (R1, R2, and R3 shown in line histogram). Further increasing the Pt4+ adsorption time to 24 h (Figure c) increases the d(111) of NPP-24 h to 2.282 Å (Figure c). As compared to that of NPP-2 h, even a narrowed particle size distribution and a reduced average particle size (4.54 nm) are found (see the low-magnification image in Figure f) in NPP-24 h. Both characteristics can be explained by the formation of the Pt cluster in the near-core region. Re-deposition of Pd and Ni atoms increases the heteroatomic intermix, therefore, reduces surface energy of the shell crystal. In this event, as compared to that of monometallic Pt nanoparticle, this size of experimental NCs can be stabilized in a smaller size ranging from 3 to 7 nm (Figure f). The HRTEM images with Fourier transform patterns provide certain proofs for the determination of the atomic structure of the experimental NiPdPt nanoparticles for density functional theory (DFT) calculations. Shown in Figure a, symmetrically aligned arc patterns of a–a* and b–b* pairs (red dashed double arrows) suggest the preferential arrangement of atoms in the (111) facet in NPP-10 s. The presence of bright spots c–c* pair with a longer radius (Rc–o) than Ra–o (Rb–o) indicates an ordered atomic structure in the (200) facet. The coexistence of two fast Fourier transform (FFT) symmetric spots with intensity difference implies a slight atomic restructuring due to insufficient reaction time for galvanic replacement between Pt4+ and Ni/Pd atoms in NPP-10 s. In the case of NPP-2 h, the presence of symmetrical FFT patterns (a–a* and b–b*) with an intersect angle of 90° across the center axis o indicate the preferential atomic arrangement at the (111) facet. By increasing the Pt4+ ion adsorption time to 24 h, the presence of a ring pattern indicates the formation of a polydispersed atomic arrangement and a locally disordered structure in a NPP-24 h. These structural differences show a consistent result to that expected by experimental design and X-ray absorption analysis in this study. The FFT patterns of the selected regions for NPP-10 s, NPP-2 h, and NPP-24 h NCs have been depicted in Figure S1 with a detailed description. High- and low-magnification images of Ni@Pd/Pt NCs for changing Pt/Pd ratio (NPP-0.1 and NPP-0.2) with a Pt4+ adsorption time of 2 h are given in Figure S2 as comprehensive discussions.

Figure 1

HRTEM images of Ni@Pd/Pt NCs with varying Pt4+ adsorption times: (a) 10 s, (b) 2 h, and (c) 24 h. The d-spacing values (denoted by white lines) of the as-prepared NCs are calculated by using inverse Fourier transformed (IFT) images and their corresponding line histograms (insets). Fourier transformation patterns of selected areas in HRTEM images are shown in an upper row. Low-resolution HRTEM images and particle size distribution histogram (insets) have been shown in parts (d)–(f).

Figure compares the X-ray diffraction (XRD) patterns of experimental NCs, and the corresponding structural parameters are summarized in Table . In presented patterns, peaks A (19.789°) and B (22.917°) are diffraction signals from the (111) and (200) facets of metallic Pt nanocrystals in Pt-CNT (gray dashed line). For XRD pattern of Pd-CNT, the two peaks (green line) are slightly shifted to the right, suggesting its relatively smaller lattice constant as compared to Pt-CNT.

Figure 2

Comparative XRD patterns of CNT-supported NPP NCs with different Pt4+ adsorption times and control samples (Pd-CNT and Pt-CNT). The Pt/Pd ratio is 13 atom %. *Line “A” and “B” correspond to the diffraction signals from the (111) and (200) facets of NCs.

Table 1

XRD Determined Structural Parameters of Pt-Decorated Ni@Pd NCs (Pt/Pd = 0.3)

sample	d(111)	d(111)Ni	d(200)	D(111) (nm)	D(200) (nm)	D(111)Ni (nm)	H(111)/H(200)a	H(111)/H(111)Nia
Pt-CNT	2.259		1.961	6.28	4.68		2.81
Pd-CNT	2.253		1.953	8.23	5.11		3.34
NPP-10 s	2.257	2.162	1.971	3.95	2.66	2.67	2.76	4.05
NPP-2 h	2.253	2.101	1.961	4.25	3.26	2.31	3.42	11.62
NPP-24 h	2.261	2.147	1.971	4.16	2.65	2.64	3.07	7.53

Intensity ratios of diffraction lines for the (111) and (200) peaks (H(111)/H(200)) and Pd(111)/Ni(111) peaks (H(111)/H(111)Ni).

With the same position of diffraction peaks, experimental NCs are indexed as metallic face-centered cubic phase and show different extent of preferential growth in the (111) facet. As compared to patterns of Pd-CNT and Pt-CNT, a substantially increased diffraction background at two sides of the diffraction peaks indicates the lower content of Pd scattering matters combining with the (111) diffraction signal from the sublayer Ni metal (denoted by fitting curves in the Supporting Information (SI)) in NPP NCs. As compared to the peak positions of Pt-CNT, an upshift of A and B peaks indicates the decreasing lattice constant and the increasing Pt–Pd alloy with increasing Pt4+ adsorption time from 10 s to 2 h. As compared to the peaks of NPP-2 h, downshift of the (111) peak indicates that an increased lattice constant possibly attributed to the formation of Pt-enriched core and Pd–Pt alloy shell in NPP-24 h. Average coherent length at the (111) facet (D(111)) (D(200) for (200) facet) is 3.95 (2.66) nm for NPP-10 s, 4.25 (3.65) nm for NPP-2 h, and 4.16 (2.65) nm for NPP-24 h. By cross-referencing the XRD and HRTEM results, the effects of Pt4+ adsorption time on the crystal structure configuration provide certain physical prospects on the ORR activity of NPP NCs. At a low Pt4+ adsorption time (NPP-10 s), high defect density appears by positing Pt atoms in the NC surface. As denoted by truncated surfaces, those Pt atoms would form a slight intermix and lattice strain (XRD) and thus result in semicoherent interfaces (twin boundaries) in the interfaceted edges of NC. By increasing the Pt4+ adsorption time to 2 h, Pt clusters are mostly intercalated in the NC surface and generated a largest local compressive strain in the Pd shell crystal. In diffraction pattern of NPP-2 h, the highest ratio (3.42) of peak intensity for the (111) and (200) facets indicates its largest extent of preferential growth in the (111) facet among the experimental NCs. Increasing Pt4+ adsorption time (i.e., 24 h) leads to the formation of the high extent of residual Pd ions in the intermediate steps and thus end up with the formation of a Pt-rich core capped with a Ni/Pd/Pt shell structure. In such a structure, the local strain is relaxed and the surface active sites is reduced in NPP-24 h. For comparison analysis, NPP NCs with Pt/Pd ratios of 0.1 and 0.2 and identical adsorption times (10 s, 2 h, and 24 h) are synthesized. Details of the structural interpretation of the NCs by the XRD mentioned above are given in the Supporting Information (SI) and the XRD patterns (Figure S3), and the calculated structural parameters are summarized in Table S3.

X-ray absorption spectroscopy (XAS) analysis complementarily proves the effects of Pt4+ ion adsorption time on the NPP NCs structures in atomic and electronic regimes. Figure compares the normalized Pt L3-edge X-ray absorption near-edge spectra (XANES) and Fourier transform extended X-ray absorption fine structure (EXAFS) spectra of the experimental NCs. In L3-edge spectrum, the position of the inflection point (arrow X) refers to the threshold energy (E0) for 2p-to-5d electron transition and is linearly proportional to the oxidation state of the target atom (particularly for transition metals). Intensity (HA) and width (WA) of the near-edge absorption peak (white line) elucidate the relative extent of empty states and splitting of 5d5/2 orbital. Width (WB) and intensity (HB) of oscillation hump in the postedge region explain the extent of structure ordering around the target atom. As shown, an identical position of an inflection point in XANES region suggests that all the NCs possess a similar metallic characteristic. For XANES of NPP-10 s, the highest HA indicates the largest extent of oxygen adsorption in the Pt atom among the presented NCs, again rationalizing the positioning of the Pt clusters in the NC surface as proposed by HRTEM. For NPP-2 h, the lowest white line intensity (HA) among experimental NCs reveals an inhibition of Pt oxidation and an offset of X (inset of Figure a) to the low-energy side reveals a charge relocation to Pt from neighboring atoms. Both characteristics rationalize the steric effects by intercalation of Pt clusters in Pd shell and are consistently explained by results of quantitative structure analysis by EXAFS fitting (Table ). Such an intercalation results in a local disorder structure, as revealed by a significantly suppressed backscattering intensity in the postedge region (HB) and the formation of disordered core region, as shown by HRTEM (Figure b). For NPP-24 h, shown by an increased postedge hump, the atomic structure of NC is rearranged into the ordering phase due to sufficient galvanic replacement time and subsequent redeposition of the PdPtNi structure in the shell region by a reduction agent.

Figure 3

(a) XANES and (b) Fourier transform EXAFS spectra of control sample (Pt-CNT) and experimental NPP NCs prepared by different Pt4+ adsorption times.

Table 2

Atomic Structure Parameters of Experimental Pt-Decorated Ni@Pd NCs and Control Samples (Pt-CNT)

Pt4+ adsorption time	NCs	bond pair	CN	R (Å)	χa
10 s	Pt-CNT	Pt–Pt	6.45 ± 0.4	2.752 ± 0.02	100
	NPP-10 s	Pt–Pt	3.4 ± 0.8	2.678 ± 0.02	62.7
		Pt–Pd	1.1 ± 0.3	2.634 ± 0.02	20.3
		Pt–Ni	0.92 ± 0.2	2.615 ± 0.02	17.0
2 h	NPP-2 h	Pt–Pt	4.49 ± 0.4	2.706 ± 0.01	65.1
		Pt–Pd	1.08 ± 0.2	2.635 ± 0.01	15.7
		Pt–Ni	1.33 ± 0.3	2.582 ± 0.02	19.3
24 h	NPP-24 h	Pt–Pt	3.65 ± 0.5	2.701 ± 0.02	60.1
		Pt–Pd	1.09 ± 0.3	2.606 ± 0.02	18.0
		Pt–Ni	1.33 ± 0.3	2.582 ± 0.01	21.9

χ stands for the extent of heteroatomic intermix for M atoms around Pt in a bond pair of Pt–M (i.e., χ for Pt–Pd refers to the extent of Pd atoms among the total coordination numbers around Pt atom. For optimization of structure parameters, σ2 is determined to be 0.004 Å2 by fitting the spectrum of Pt foil and is adopted for fitting all experimental spectra.

Figure b compares the Fourier transform Pt L3-edge EXAFS spectra (i.e., radial structure functions, RSFs) of the experimental NPP NCs and Pt-CNT. Radial peaks (A) across 1.7–3.2 Å account for X-ray interference with metallic Pt–Pt bond in Pt-CNT. Quantitative structure parameters are obtained by fitting (Figure S4) the corresponding spectra with proper atomic models, and the results are summarized in Table . Compared to that of Pt-CNT, the offset of peak A position indicates a reduction of interatomic bond length by ∼0.2 Å around Pt atoms in all NPP NCs. Compared to that of NPP-10 s and NPP-24 h, a substantial increment in peak B intensity refers to the largest extent of in-phase outgoing X-ray interferences between Pt–Pt bond pairs (i.e., the highest coordination number of Pt–Pt bond pairs) for NPP-2 h. Quantitative structural analysis exposes the same results, indicating the largest coordination number of 4.49 at the Pt–Pt shell (CNPt–Pt) for NPP-2 h among experimental NCs (Table ). The structure of embedded Pt cluster in Pd shell is consistently revealed by the presence of the ordered lattice domains in the HRTEM image (Figure b). For NCs prepared by a Pt4+ adsorption time longer than 2 h, trend on formation of homoatomic cluster is revealed by offset of RSF peak closing to that of metallic bond pair (peak A). Meanwhile, decreases in the peak intensity corresponding to a CNPt–Pt of 3.65 also explains the increasing heteroatomic intermix around Pt atoms in Pd shell. To further prove our findings, NPPs with Pt/Pd ratios of 0.1 and 0.2 were prepared and found to have similar structural characteristics (Figure S5), and the corresponding electrochemical properties are given in the Supporting Information (ESI).

To get an insight into the approximate chemical composition of the uppermost layers (1–2 nm from surface) and the binding energy of experimental NCs, we performed X-ray photoemission spectroscopy (XPS) analysis for the photoelectrons of Pt 4f and Pd 3d levels. Figure S6 shows the typical fitted XPS spectra in the Pt 4f and Pd 3d region for the experimental NCs. In a Pt 4f spectrum, doublet peaks about 71 and 74 eV, respectively, refer to photoelectron emission from Pt 4f7/2 and Pt 4f5/2 orbitals. However, for Pd 3d spectrum, doublet peaks around 336 and 341 eV are responses to photoelectron emission from Pd 3d5/2 and Pd 3d3/2 orbitals, respectively. The peaks are further deconvoluted to separate the signals from different oxidation states, and the corresponding results are listed in Table .

Table 3

XPS Determined Composition Ratios and Binding Energy of Experimental NCs (Pt/Pd = 0.3)

	surface composition (%)			binding energy (eV)
NCs	Pt	Pd	Ni	Pt	Pt(OH)2	PtO2	Pd	PdO	PdO2	NiOx
NPP-10 s	49.72	31.15	19.13	71.1	72.21	73.36	335.8	337.2	338	68.58
NPP-2 h	32.65	42.57	24.78	69.92	71.27	74.28	335.85	337.25	338.05	68.72
NPP-24 h	28.02	43.91	28.07	69.68	71.33	72.43	336	337.25	338	68.17

Through the analysis of the XPS patterns of experimental NCs, it can be found that the electronic structure of the ternary NCs strongly depends on the Pt4+ reduction time on the Ni@Pd surface. The intensity of emission peaks in the XPS spectrum is positively related to the electron density of the target atoms. A higher intensity refers to a higher electron density. Surface chemical composition and binding energy reveal a nanostructure and trends of electron relocation between atomic Pt clusters and shell crystal, respectively. Therefore, the highest intensity found in the XPS spectra of NPP-10 s is attributed to the most abundant 4f electrons of the Pt atoms. We also observe that the intensity gradually decreases with increasing Pt4+ reduction time series, which indicates the lower density of the 4f electrons. This is consistent with our XRD, HRTEM, and XAS findings. According to Table , the binding energy of the Pt decreases significantly for NPP-2 h and NPP-24 h. The difference in the binding energy of Pt is due to electron relocation from the neighboring atoms to the Pt clusters, which itself is a result of electronegativity gap and lattice expansion in Pt crystals. An evidence of electron relocation is presented in the Pd XPS spectrum, which exhibits a lower intensity and thus an increasing 4d electron vacancy as the Pt4+ reduction time increases.

To elucidate ORR performances of experimental NCs, electrochemical impedance spectroscopy (EIS), cyclic voltammetry (CV), and linear sweep voltammetry (LSV) analyses are employed. Figure compares the Nyquist plots of the NPP NCs prepared by varying Pt4+ adsorption times (Figure a) and Pt/Pd ratios (Figure b). The corresponding electrochemical parameters are determined by matching the features of EIS spectra with an equivalent electrical circuit containing appropriate elements using EC-Lab V10.40 software (shown in Figure S7). The Nyquist plot of all the experimental NCs shows a Warburg-type straight line in the low-frequency region and two overlapping semicircles in the high-frequency region and medium-frequency region. In an EIS spectrum, intercept at the Z′real axis is attributed to the internal resistance arising from the electrolyte (i.e., series resistance (RS) of the electrolyte). The diameter of the high-frequency semicircle resembles the ion diffusion resistance of the electrolyte (RSEI), and the semicircle in the medium-frequency region is responsible for the charge transfer resistance (RCT) at the bulk surface of active material. The Warburg-type straight line in the low-frequency region represents the diffusion of ions into the electrode related to Warburg impedance (i.e., the ion intercalation kinetics in the channels of the active material).[25] According to Figure a and Table S4, all the NCs possess RS at the same level (21.56 Ω for NPP-10 s, 18.93 Ω for NPP-2 h, and 19.83 Ω for NPP-24 h) in bulk electrolyte transportation resistance between working and reference electrodes. Such a characteristic is an index which proving that all experimental spectra are measured without artificial error. An even closer look reveals that NPP-2 h shows the lowest series resistance (18.93 Ω) among experimental NCs, which results in lower electrolyte resistance for NPP-2 h and fast surface kinetics. The lowest RSEI value of NPP-2 h (RSEI is 177.8 Ω for NPP-2 h, 192.7 Ω for NPP-10 s, and 190.8 Ω for NPP-24 h) suggests its highest ion diffusion kinetics among experimental NCs. Meanwhile, charge transfer resistance (RCT) of NCs under inspection follows the trend of NPP-2 h (572.1 Ω) < NPP-24 h (617.8 Ω) < NPP-10 s (775.1 Ω), which consistently proves the highest surface reaction kinetics (i.e., SA) of NPP-2 h among all NCs. Comprehensive results are illustrated by its highest exchange current density (J) which is inversely proportional to this RCT by the equationwhere R, T, n, and F are the universal constant, absolute temperature, number of electrons involved, and Faraday’s constant, respectively. The lowest RCT value leads to the highest exchange current density; therefore (with the improved ion diffusion at the solid–liquid interface), the highest electrochemical performance of NPP-2 h is achieved among experimental NCs. Straight lines in the low-frequency region of the Nyquist plot are mainly attributed to the diffusion behavior of the ions. A steep slope of the straight line for NPP-2 h (line “a”) indicates a relative faster ion diffusion in the solutions in comparison to others. These are archetypal indications of kinetics controls on the electrochemical properties and consistently reveal the highest SA of NPP-2 h in ORR. To get further insight into the proposed statements, EIS characterizations are conducted on NPP NCs with Pt/Pd ratios of 5% (NPP-2 h-5), 9% (NPP-2 h-9), and 14% (NPP-2 h-14) (Figure b). Among all the samples, NPP-2 h-14 shows the lowest RS and RSEI values (Table S5), along with a steep slope of Warburg-type straight line, indicating its highest ORR activity.

Figure 4

Electrochemical impedance spectroscopy (EIS) of (a) NPP-10 s, NPP-2 h, and NPP-24 h and (b) comparative EIS plots of NPP NCs with variable Pt loading and a Pt4+ adsorption time of 2 h.

In heterogeneous catalysts, dissociation of chemisorbed oxygen plays a major role in the ORR steps. In these steps, the oxygen adsorption strength could determine the activity of NCs. A reduction in the oxygen adsorption energy (corresponding to Oa peaks) reduces the applied energy for initiating ORR at the reaction sites, which increases the stability of the NCs in the redox reactions. By cross-referencing the results of HRTEM, XRD, and XAS analysis, we notice that different configurations of atomic Pt clusters in experimental NCs change with increase in Pt4+ adsorption time. These characteristics are consistently elucidated by electrochemical analysis. Cyclic voltammetry (CV) analysis is used to get information of the surface chemical composition and electrochemically active surface area (ECSA) of NCs under investigation. In a CV curve, three distinctive potential regions are found, including underpotential deposition of hydrogen (UPD-H) region between 0 < E < 0.4 V, the double-layer region between, and the chemisorption of oxygen species over 0.6 V versus reversible hydrogen electrode (RHE) because of hydrogen adsorption/desorption, OH– ligand chemisorption, and reduction of chemisorbed oxygen (forward scan), as well as reduction of Pt or Pd oxides (backward scan) respectively. Through integration and double-layer correction, Coulombic charge (0.405 mC cm–2) is acquired for the reduction of Pt or Pd oxides and then converted into ECSA. These characteristics clearly appear in the electrochemical responses of the standard samples of commercial J.M.-Pt/C EC and control samples (Pt-CNT and Pd-CNT) in Figure a. For J.M.-Pt/C, position of two characteristic peaks (a* and e*) in the forward scan denotes the potential to be applied for the dissociation of H+ from close packed (111) and opened (200) facets and corresponding current, respectively. However, peaks “a” and “e”, respectively, refer to the current responses of H+ adsorption in the (111) and (200) facets. Compared to the CV profile of J.M.-Pt/C, upshift of peaks b* and d* in the forward scan and downshift of peaks b and d in the backward scan refers to increased energy barrier for redox desorption/adsorption of H+ in Pt-CNT. As consistently revealed by XRD observation, a substantially higher intensity of peak b* than that of peak d* (i.e., weakened H+ interaction on (200) facets) reveals a preferential crystal growth at (111) facets in Pt-CNT. At the same time, a higher potential of oxygen adsorption peak Ob as compared to Oa indicating a lower oxygen affinity of Pt-CNT as compared to J.M.-Pt. The CV profile of Pd-CNT is compared as a reference.

Figure 5

Comparative CV curves of experimental control samples (Pt-CNT, Pd-CNT, and Ni-CNT) with (a) commercial J.M.-Pt/C, (b) Ni@Pd, (c) Ni@Pd compared with NPP-0.3 with varies Pt4+ adsorption time, and (d) Ni@Pd with NPP NCs with different Pt/Pd ratios and Pt4+ adsorption time of 2 h. The CV curves are recorded in N2-purged 0.1 M KOH (pH = 13) electrolyte solution at room temperature, with an applied potential of 20 mV s–1.

In the UPD-H region, the broad CV peak in backward sweep of CV curve can be attributed to the strong affinity of H+ adsorption in Pd-CNT. Meanwhile, a profound and narrowed Oc peak shows a strong affinity with simple pathways for oxygen adsorption in the Pd surface. The CV curve of the control sample (Ni-CNT) is compared in Figure a. As shown, a “flat” CV sweeping curve indicates the formation of chemical inert Ni oxides in the ORR in KOH electrolyte. These characteristics prove that differences of the CV profiles of Pd-CNT could be attributed to the presence of Ni–Pd interfaces and bimetallic intermix in the Ni@Pd NC surface. Compared to the CV curves of the control samples, a downshift of peak “Ha*” and an upshift of peak “Ha” in the UPD-H region are observed in the CV curve of Ni@Pd. Such a peak offset refers to a decrease in the energy barrier for redox desorption/adsorption of H+. Similar to the CV profiles of Pd-CNT, a smeared H+ desorption peak (Ha*) reveals a strong H+ affinity in the Pd crystal and the Ni/NiO domain underneath. It is also the same reason that the Oa peak is shifted to left-hand side, indicating a stronger affinity for oxygen in Ni@Pd as compared to Pt-CNT. Compared to the CV curves of Pd-CNT and Pt-CNT, a downshift of the O reduction peak (Oa) by 0.03 V indicates a higher energy barrier for oxygen desorption and a higher double-layer thickness (ΔHd) refers to a higher extent of hydroxide formation in the Ni@Pd surface as compared to that in Pd-CNT and Pt-CNT. Figure c compares the CV curves of Ni@Pd and the experimental NPP NCs. As consistently revealed by the lowest HA in the XANES spectrum, the NPP NCs possess a lower oxidation affinity as denoted by their less intense oxide formation peaks (i.e., Oa*, Ob*, and Oc*) and oxygen reduction peak (Oa) as compared to control NCs. Those features further indicate the suppression of oxidation by Pt cluster intercalation in the Ni@Pd surface. For NPP-2 h, the lowest oxidation state and the highest ΔHd complementary reveal different reaction rates of reaction sites around the Pt clusters toward intermediate steps in the ORR. Such a phenomenon shares redox reaction steps between sites and thus facilitates the activity of Pt-decorated Ni@Pd NCs. For a comprehensive comparison, the CV curves of Ni@Pd/Pt NCs containing different Pt/Pd ratios and Pt4+ adsorption times are shown in Figure d, and the corresponding electrochemical performances are discussed in Figure b.

Figure 6

(a) Linear sweep voltammetry (LSV) curves of experimental NCs (Ni-CNT, Ni@Pd, NPP-10 s, NPP-2 h, and NPP-24 h) compared with those of commercial J.M.-Pt/C. (b) Comparative specific activity (SA) of NCs under investigation. The LSV curves are measured in an O2-saturated 0.1 M KOH solution (pH = 13) with rotation speeds of the working electrode fixed at 400–3600 rpm (only the spectra measured at 1600 rpm are shown in (a)).

Figure a compares the linear sweep voltammetry (LSV) curves of Ni-CNT, Ni@Pd, J.M.-Pt/C, and NPP NCs. Slopes of their linear fits at half current potential (normally at inflection point) are adopted to the Koutecky–Levich (K–L) equation (eqs and 3) to calculate the number of electrons transferred in the ORR (Figure S8b), and the electrochemical properties are summarized in Table where J is the measured current density, JK and JL are the kinetic and diffusion-limiting current densities, respectively, ω is the angular velocity, n is transferred electron number, F is the Faraday constant, Co is the bulk concentration of O2, v is the kinematic viscosity of the electrolyte, and K is the electron-transfer rate constant.

Table 4

Electrochemical Performances of NPP NCs Compared with Those of J.M.-Pt/C and Ni@Pd in CV and LSV Analysis

sample	N (0.5 V)	Voc (V vs RHE)	ECSA (cm2 mg–1)	SA0.85V (mA cm–2)
J.M.-Pt/C	4	0.91	257	0.261
Pd-CNT	4	0.895	555.1	0.075
Pt-CNT	4	0.956	582.4	0.368
Ni@Pd	3.9	0.881	206.4	0.142
NPP-10 s	4	0.893	315.9	0.281
NPP-2 h	4	0.898	165.5	0.732
NPP-24 h	4	0.903	285.8	0.372

In potential ranges of 0.95 to 0.8 V (vs RHE), ORR current is controlled by kinetics of mainly reduction reaction and partially mass diffusion. By reducing potential below 0.8 V (vs RHE), ORR current is dominated by mass diffusion kinetics. According to Table , the onset potential (Voc) of the experimental NCs follows the order Ni@Pd (0.881 V) < NPP-10 s (0.893 V) < NPP-2 h (0.898 V) < NPP-24 h (0.903 V) < J.M.-Pt/C (0.910 V). In a LSV curve, Voc refers to the threshold voltage for initiating ORR and is affected by trade-offs between the extent of surface oxidation (steric shielding effect), heteroatomic intermix (bifunctional mechanism), and configurational identity of heterogeneous clusters in NC. Surface oxide puts a negative impact on the ORR activity of NCs. It inhibits oxygen chemisorption and increase structural complexity around active sites (noble metal) to increases both internal resistance and steric diffusion barrier for the surface intermediate species, thereby reducing Voc of the NCs in the ORR. On the other hand, the presence of heteroatomic intermix provides low-energy pathways for intermediate steps and facilitates ORR activity. From these stand points, the highest Voc indicates the lowest initial barrier for ORR, possibly due to the lower surface defect of the Pt crystal surface in J.M.-Pt. Specific surface activity (SA) is an important index for the kinetic performance of active sites in the NC surface. As can be seen in Table , NPP-2 h possesses the highest SA among experimental NCs. It can be attributed to a proper control of local structure arrangements including atomic structure intermix and nanostructure configuration. Consistent structural configurations are proved by cross-referencing the results of HRTEM, XAS, and XRD. With a configuration of Pt clusters intercalated in the Pd shell crystal capping at Ni/NiO core, twin boundaries in the interfaceted edges are formed and reduce defect sites in the NC surface. In this event, oxygen chemisorption is suppressed (Figure b), facilitating subsequent reaction steps among Pt and neighboring atoms. Such a statement is consistently revealed by a less intense oxygen adsorption peak of NPP-2 h in the CV plot. For the rest of NPP NCs, high contents of defect sites increase oxygen adsorption in the surface and suppress SA. The electrochemical performances of NPP with different Pt/Pd ratios (Pt/Pd = 0.1 and 0.2) and identical Pt4+ adsorption time to the experimental samples (Pt/Pd = 0.3) are compared (Figure b). For a better understanding of the effects of Pt4+ adsorption time on the ORR activity, the electrochemical properties of Ni@Pd NC without Pt additive are compared, showing a substantial lower SA (0.142 mA cm–2) as compared to the samples with Pt decoration. Details of the Ni@Pd/Pt NCs series for variable Pt/Pd ratio and 2 h Pt4+ → Pt reduction time are given in the Supporting Information (Figure S8a) and calculated structural parameters are summarized in Table S6. The LSV curves of the experimental NCs at different rotation speeds from 400 to 3600 rpm have been depicted in Figure S9.

Facilitation of SA (reaction kinetics) is not only a matter of composition (bifunctional mechanism) and composition complexity but most importantly the configuration-induced charge localization by Pt intercalation inside Ni@Pd NC. According to our discussions, combining the results of structural inspections (HRTEM, XRD, and XAS), one can notice that the size of the Pt cluster and its depth inside the Pd shell of the NPP NCs are controlled by Pt4+ adsorption time (i.e., extent of galvanic replacement between Pt4+ ion and Pd/Ni). With Pt4+ adsorption for 10 s, average CNPt–Pt is determined to be 3.40, suggesting that Pt atoms are mostly accumulated in the tetragonal bipyramid cluster (with an average CNPt–Pt of 3.60) located in the topmost layer of the Pd shell crystal in NPP-10 s. With increasing Pt4+ adsorption time to 2 h, CNPt–Pt is increased to 4.49, indicating Pt atoms are assembled in clusters larger than octahedral (average CNPt–Pt = 4) in NPP-2 h. Further increasing the Pt4+ adsorption time reduces CNPt–Pt to 3.6 in NPP-24 h. With such a long adsorption time, a severe galvanic replacement forms discrete Pt clusters in the Pd/Ni shell and deports substantial amount of Pd ions in the solution. In this event, subsequent addition of reduction agent simultaneously posits Pt and Pd atoms to grown Pd–Pt alloy in the shell region. Those parameters are adopted to atomic models in density functional theory (DFT) calculations to figure out the configuration effects on the ORR activity of the experimental NCs in the electronic structure regime. Figure compares the DFT-calculated electron structure distribution of the Pt cluster intercalated 4 × 4 Ni@Pd slab in the (111) facet. In these models, tetragonal bipyramidal and octahedral Pt clusters with different depth in the Pd shell are employed to mimick the atomic structure of NPP NCs. For NPP-10 s, tetragonal bipyramidal Pt clusters are located in the topmost layer of the Pd shell. A strong electron localization (blue region in electron contour) appears and could be attributed to local lattice strain around the Pt clusters. Presence of adsorbed Pt atom in surface layer results in severe charge depletion (yellow region in electron contour) and certain steric space for atomic O adsorption (Oads). Those features are consistent with the obtained structural parameters and explain the low reaction kinetics (SA) in NPP-10 s. For the cases of Pt clusters embedded in the Pd shell (NPP-2 h), regardless of steric configurations, all three models possess strong electron localization with negative charge appearing around the Pt clusters. Those negative-charge fields form a repulsive force to weaken the Oads bond and provide a complex combination of the reaction site identity. Both methods reduce the reaction barrier for Oads evolution in the reaction sites, thereby significantly improving the SA of NPP-2 h. For models of NPP-24 h, Pt cluster is formed in the Pd–Ni interface or the Ni core crystal. In this configuration, local lattice strain around the Pt cluster would couple with lattice strain in the Pd–Ni interface. Such a coupling effect triggers a strong electron relocation from the surface to the Pt cluster, thus depleting surface electron in this model. It reduces the identity of the reaction sites, delocalizes electron density, and consequently reduces reaction kinetics in NPP-24 h.

Figure 7

Density functional theory-determined electron density distribution for models of 4 × 4 Ni3L@Pd3L (111) slab (three layers of Pd capping atop three layers of Ni(111) facet) intercalated by tetragonal bipyramidal Pt cluster (NPP-10 s), octahedral Pt cluster in Pd3L (NPP-2 h), and octahedral Pt cluster in the Ni–Pd interface and Ni3L core.

Conclusions

A robust method in terms of time and ease of fabrication of the Pt cluster-decorated NiPdPt ternary NC with tunable electrochemical properties in the ORR is demonstrated. By cross-referencing the results of structural characterization, electrochemical analysis, and density functional theory calculations, we show that the atomic arrangement, surface oxophilicity, reaction site identity, and, consequently, charge density distribution of the ternary NCs can be engineered by changing the Pt4+ adsorption time and the Pt content on the Ni@Pd surface. Charge distribution in NCs is confined by local strain around the Pt clusters. It varies with their intercalation depth as changes with Pt4+ adsorption time. In optimum case with 14 atom % of Pt content, specific surface activity (SA) is 0.732 mA cm–2 when Pt4+ adsorption time is 2 h. This value is 2.8 times as compared to that of J.M.-Pt. Most importantly, the proposed scenarios are mostly compatible to be adopted for catalyst fabrication owing to simplicity and cost-effectiveness considerations in industrial sector, therefore, highlighting the possibility of the development of advanced alkaline fuel cell devices.

Experimental Section

Synthesis Methodology and Materials for Preparation of Ni@Pd/Pt NCs

Atomic Pt clusters-decorated Ni@Pd NCs were synthesized by using a sequential wet chemical reduction method in a controlled environment, and the corresponding steps for synthesis methodology are presented in Scheme . In the first step, 360 mg of catalyst support (multiwalled carbon nanotube in ethylene glycol, 5 wt % Cnano Technology Ltd.) is added to 1.28 g of aqueous solution containing 0.1 M of nickel(II) chloride hexahydrate (NiCl2·6H2O, Showa chemical Co. Ltd.) and stirred at 200 rpm at 25 °C for 6 h. The mixture (Ni2+-adsorbed CNT, CNT-Niads) contains 0.128 mmol (7.5 mg) of Ni metal ions in a metal loading of Ni/CNT = 30 wt %. After stirring, 5 mL of water solution containing 0.0386 g of NaBH4 (99%, Sigma-Aldrich Co.) was added to the sample prepared in the first step and the mixture stirred at 200 rpm for 10 s. In this step, metastable Ni metal nanoparticles were formed (Ni/CNT sample). In step 3, 1.28 g of Pd precursor solution containing 0.128 mmol of Pd metal ions (i.e., 0.1 M) was added to Ni/CNT sample to form a thin layer of Pd crystal on the surface of nickel nanoparticles (namely, Ni@Pd) with a Ni/Pd ratio of 1.0. The Pd precursor solution was prepared by dissolving the metal powder (Pd, 99%, Sigma-Aldrich Co.) in 1.0 M HCl(aq). After 20 min, 0.383 g of Pt precursor solution containing 0.0383 mmol of Pt ions (H2PtCl6·6H2O, 99%, Sigma-Aldrich Co.) was added to the Ni@Pd solution (step 4). In step 5, 5 mL of water solution containing 0.0116 g of NaBH4 was added to the mixture of Pt4+ ion and Ni@Pd, followed by stirring at 200 rpm for 2 h to complete Pt metal reduction. The end products are atomic Pt clusters-decorated Ni@Pd/Pt (NPP) NCs. In this step, distribution and depth of atomic Pt clusters in the Pd shell region are controlled through adsorption (i.e., galvanic replacement of Pd with Pt4+) time of Pt4+ in the Ni@Pd surface and the Pt/Pd ratio of 0.3. In this study, samples of Ni@Pd with Pt4+ adsorption times of 10 s, 2 h, and 24 h are, respectively, named NPP-10 s, NPP-2 h, and NPP-24 h. The resulted powder is washed several times with acetone, centrifuged, and then dried at 70° C, followed by natural cooling at room temperature. The synthesis of Ni@Pd followed similar processes as those used to prepare Ni@Pd/Pt, except that Pt metal precursor was absent in the reaction system. Control samples (Pt-CNT and Pd-CNT) were synthesized by immersing Pt (Pd) precursor solution in CNT at 25 °C for 6 h followed by adding reducing agent (0.023 mg of NaBH4 in 5.0 mL H2O). In the study, Pt (Pd)/CNT ratio is 30 wt %. For comprehensive discussions, electrochemical properties of NPPs in Pt/Pd ratios of 0.1 (namely NPP-01) and 0.2 (namely NPP-02) (synthesized with identical sequences and time to those of NPP-03) are compared in Figure . Details of their structure-to-electrochemical correlations are given in the Supporting Information (SI).

Scheme 1

Schematic Representation

Synthesis of atomic-scale Pt-decorated Ni@Pd (Ni@Pd/Pt) NCs with controls of Pt4+ ion adsorption time (i.e., step 5: Pt4+ + Ni@Pd → Ni@Pd/Pt) upon crystal growth.

Characterizations of Ni@Pd/Pt Ternary NCs

Surface morphological features of as-synthesized ternary NCs were analyzed by using high-resolution transmission electron microscopy (HRTEM) at an acceleration voltage of 200 kV in the Electron Microscopy Center at National Chiao Tung University. The average coherent length (Davg) of experimental NCs is calculated from the XRD peak broadening of the (111) facets using the Scherrer equation. For XRD analysis, the wavelength of incident X-ray is 0.7749 Å (16 keV) at a beamline of BL-01C2 at the National Synchrotron Radiation Research Center (NSRRC), Taiwan. The X-ray absorption near edge spectroscopy (XANES) of the experimental NCs was analyzed in the fluorescence mode at the BL-07A1 and BL-17C1 beamlines at the NSRRC. The X-ray photoelectron spectroscopy (XPS) spectroscopy of the experimental NCs was executed to identify the surface chemical states of the catalysts at BL-24A of the NSRRC (Hsinchu, Taiwan). The electrochemical impedance spectroscopy (EIS) was performed under open-circuit voltage conditions. During the measurements, the sinusoidal amplitude modulation was set at 10 mV in the frequency range from 100 mHz to 1000 kHz and the spectrum was recorded in an alkaline medium (0.1 M KOH) starting at high frequency and moving toward low frequency in a logarithmic scan. The electrochemical measurements were carried out at room temperature using a potentiostat (CH Instruments model 600B) equipped with a three-electrode system. The cyclic voltammetry (CV) and the linear sweep voltammetry (LSV) data were acquired at the voltage scan rate of 0.02 and 0.001 V s–1 and the potential range of 0.1–1.3 V (V vs RHE) and 0.4–1.1 V (V vs RHE), respectively, in an aqueous alkaline electrolyte solution of 0.1 M KOH (pH 13). The rotation rate of 1600 rpm was used for LSV. N2 and O2 atmospheres were used for CV and LSV, respectively. CNT-supported Ni@Pd/Pt catalyst, 5.0 mg, was mixed with isopropyl alcohol (1 mL) and 5 wt % Nafion (0.05 mL) to prepare slurry sample. The mixture was subjected to ultrasonication treatment for 30 min before ORR experiment. A glossy carbon rotating disk electrode (0.196 cm2 area) coated with 10.0 μL of the desired catalyst slurry sample, followed by air drying, is used as the working electrode. Hg/HgCl2 (the voltage was calibrated by 0.242 V, in alignment with that of RHE) electrode saturated in KCl aqueous solution and a platinum wire was used as a reference electrode and a counter electrode, respectively.

Computation Details for Density Functional Theory Calculations on Charge Density Distribution of Ni@Pd with Atomic Pt Cluster Decoration in Different Depths

To elucidate the experimental observations, we employed first-principles calculations based on the density functional theory (DFT) using the Vienna ab initio simulation package[26,27] with the projector augmented wave pseudopotentials.[28] A plane-wave basis set with a kinetic energy cutoff of 420 eV was used to expand the Kohn–Sham wave functions. The Pd–Ni alloy surface is simulated by a repeated slab with a 4 × 4 lateral periodicity in which three Pd top layers and three Ni bottom layers are included. All atoms in the simulation model are fully relaxed. The vacuum region between slabs is ∼15 Å. The Brillouin zone is sampled with a 5 × 5 × 1 k-point mesh, which produces well-converged results. The geometry is optimized until the total energy is converged to 10–4 eV. The total charge density difference is obtained by subtracting the total charge density of the Pt-contained alloy model from that of the Pt-free alloy model.