Dealloying of Nitrogen-Introduced Pt-Co Alloy Nanoparticles: Preferential Core-Shell Formation with Enhanced Activity for Oxygen Reduction Reaction.

Voltammetric dealloying is a typical method to synthesize Pt-shell/less-noble metal (M) alloy core nanoparticles (NPs) toward the oxygen reduction reaction (ORR). The pristine nanostructures of the Pt-M alloy NPs should determine the ORR activity of the dealloyed NPs. In this study, we investigated the voltammetric dealloying behavior of the Pt-Co and nitrogen-introduced Pt-Co alloy NPs generated by synchronous arc-plasma deposition of Pt and Co. The results showed that the dealloying behavior is sensitive to cobalt nitride in the pristine NPs, leading to the preferential generation of a Pt-rich shell/Pt-Co alloy core architecture having enhanced ORR activity.

Introduction

Proton-exchange-membrane fuel cells (PEMFCs) are promising energy conversion systems, particularly in the case of fuel cell vehicle applications. However, the slow kinetics of the oxygen reduction reaction (ORR) that proceeds at the cathode requires a large amount of Pt as an electrocatalyst, which represents a major obstacle to the widespread adoption of PEMFCs. Pt-based alloy nanoparticles (NPs) containing transition metals (Pt–M, where M = Co, Fe, Ni, Pd, etc.) have attracted particular attention, primarily because of their superior ORR activity per unit mass of Pt, that is, their Pt mass activity toward the ORR.[1] The mass activity enhancement of the Pt–M alloy catalysts is largely attributable to the lattice strain (geometric) and ligand (electronic) effects,[2,3] which depend upon the surface morphologies and near-surface Pt/M alloy compositions. Numerous studies published to date clearly show that optimum nanostructural control of the Pt–M NPs is essential to maximize the enhancement in the activity toward the ORR.

Dealloying is an effective method to synthesize highly ORR-active-structured n class="Chemical">NPs such as alloy nanoframes.[4] Dealloying can be achieved through either of two mechanisms: selective less-noble metal surface dissolution and Pt-rich shell layer construction. Therefore, the nanostructures generated using the dealloying process should be strongly influenced by the structures of the pristine Pt–M alloy NPs used as starting materials, for example, their surface roughness,[5] alloy composition,[6−8] and particle size.[9,10] Strasser et al. have demonstrated that core–shell-structured NPs with a Pt-rich shell surrounding a Pt–M alloy core formed through voltammetric dealloying and exhibit very high ORR activity.[6] Although core–shell-structured NPs have been extensively studied, the critical problem of insufficient stability of these materials under PEMFC operating conditions, that is, the durability of the core–shell NPs, remains unsettled. To stabilize the core–shell structures, several groups have focused on nitridation of the Pt–M alloy core.[11−13] For example, Adzic and co-workers reported that Pt–M alloy cores could be stabilized through nitrogen bonding of M.[14] Because dealloying behaviors of the Pt–M NPs should be determined by the less-noble M dissolution accompanied by the surface diffusion of Pt atoms, the nitridation of the Pt–M alloy core should strongly influence the core dissolution behavior. However, at present, the role of nitrogen in the Pt–M NP systems remains unclear.

The arc-plasma deposition (APD) method is one of the physical vapor deposition techniques used to synthesize monodispersed metal or alloy NPs without introducing any organic impurities.[15] Because the amounts of deposited materials can be easily controlled via the arc voltage and through pulse repetitions, the APD-fabricated NPs are expected to be suitable for evaluating the relationship between the nanostructures and the electrocatalytic behaviors. Indeed, we have previously reported the electrochemical (EC) properties of APD-fabricated Pt-based alloy catalysts, where we investigated the relationship between the ORR activities and the nanostructures of Pt–Ni NP-stacking thin films[7] and Au-modified Pt NPs.[16,17] Furthermore, because APD conducted under a reactive gas atmosphere can induce chemical reactions between a transition metal and the activated reactive gas, the EC properties of the Pt–M NPs can easily be modified through precise control of the reactive gas pressure during APD. Actually, Ohnishi and co-workers demonstrated through the APD of niobium (Nb) under controlled partial pressures of N2 and O2 that finely dispersed Nb oxynitride-like species can be synthesized on the graphite substrate.[18] In the aforementioned context, we here investigated the voltammetric dealloying behaviors of Pt–Co alloy NPs generated in the presence (PtCo with N2) and absence (PtCo w/o N2) of N2 by synchronous APD of Pt and Co.

Results and Discussion

The diffraction patterns for 2θ angles from 38° to 46° are presented in Figure A (whole 2θ patterns ranging from 20° to 80° are presented in Figure S1). A diffraction peak for both samples present around 41.7° is assigned to the (111) peak of face-centered cubic (fcc) PtCo (ICCD #03-065-8970), indicating that the prepared samples are mostly composed of a solid solution of PtCo alloy with an fcc crystal structure. The diffraction peak of PtCo with N2 emerged at a lower angle compared to that of PtCo w/o N2: the peak shift is probably caused by the expansion of a Pt–Co fcc lattice by interstitial nitrogen atoms in the alloy NPs (as Table S1).

Figure 1

(A) XRD patterns for the samples of PtCo w/o N2 (black) and with N2 (pink). The blue vertical line indicates the (111) peak positions of Pt. XPS spectra of (B) N 1s, (C) Co 2p, and (D) Pt 4f regions for the PtCo w/o N2 (black) and with N2 (pink) NPs. The peaks correspond to N1: pyridinic-N, N2: Co-N, N3: pyrrolic-N, N4: graphitic-N, and N5: nitric oxide for the N 1s region; Co1: Co(0), Co2: CoO, Co3: Co-N, Co4: satellites for the Co 2p region; and Pt1: Pt(0) and Pt2: Pt(II) for the Pt 4f region.

The X-ray photoelectron spectroscopy (XPS) spectra of N 1s (Figure B: 405–395 eV), Co 2p (Figure C: 802–774 eV), and Pt 4f (Figure D: 78–70 eV) regions for the as-prepared PtCo w/o N2 (black) and PtCo with N2 (pink) are summarized. The N 1s band in the spectrum of PtCo with N2 can be deconvoluted into five components: pyridinic-N (398.0 eV), Co-N (399.9 eV), pyrrolic-N (400.4 eV), graphic-N (401.0 eV), and nitric oxide (404.0 eV).[19−21] The Co-N peak provides strong evidence for Co–N bond generation; the other nitrogen-related peaks are attributed to C–N bonds formed through nitridation of the highly oriented pyrolytic graphite (HOPG) substrate surface. Ohnishi et al. deduced that these incidental products (C–N bonds) synthesized through the reactive-APD of Nb on the HOPG substrate contribute little to the ORR.[18] Thus, we are not concerned with the influence of the C–N incidental products. In the Co 2p region, deconvolution revealed peaks at 778.5, 780.1, and 781.5 eV; these peaks are assigned to Co(0), cobalt oxide, and Co-N, respectively.[22] By contrast, the peak positions of the Pt 4f doublet for both PtCo samples were almost identical. Specifically, each doublet was deconvoluted into two chemically different diverse valence states, that is, metallic Pt(0) and oxidized Pt(II). The integrated intensity ratio between the Pt(II) and Pt(0) peaks increased after nitridation of the Co atoms in the PtCo alloy NPs, implying that the Pt electronic structure was modified by the introduction of nitrogen into the Pt–Co alloy system. In any event, the X-ray diffraction (XRD) (Figure A) and XPS (Figure B–D) results confirm that Co nitridation was achieved by synchronous APD of Pt and Co under an N2 partial pressure of 0.1 Pa.

The samples were electrochemically dealloyed by being cycled between 0.05 and 1.05 V for 300 cycles at 500 mV s–1 in N2-purged 0.1 M HClO4 using a triangular wave potential. Figure presents scanning tunneling microscopy (STM) images and the corresponding particle-size distributions of PtCo w/o N2 (top) and with N2 (bottom) before (left) and after (right) voltammetric dealloying. The STM images show well-dispersed NPs with almost the same mean particle size (approximately 5 nm) for both as-prepared (before dealloying) samples. The estimated mean particle size corresponds well to the crystallite size estimated from the XRD patterns (Table S1). The results show that there is little influence of Co nitridation on the average diameters of the samples. By contrast, the STM image and particle-size distribution of the dealloyed PtCo w/o N2 differed substantially from those of the dealloyed PtCo with N2; the reduction in the particle size of PtCo with N2 (5.3–4.5 nm) was suppressed in comparison with that of PtCo w/o N2 (4.8–3.5 nm) using the voltammetric dealloying process.

Figure 2

STM images (100 × 100 nm2) and particle-size distributions of PtCo w/o N2 (top) and with N2 (bottom) before (left) and after (right) voltammetric dealloying.

To determine the alloy compositions of the PtCo NPs, we used high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) equipped with an energy-dispersive X-ray (EDS) apparatus. In the EDS line profile of the dealloyed PtCo w/o N2 (Figure , left top), the Co profile (green) is significantly less intense than the Pt profile (blue), indicating that the dealloying of Co proceeds deeper into the particle bulk. By contrast, as shown in the EDS line profile of the dealloyed PtCo with N2 (Figure , right top), the EDS intensity of Co stands in comparison with that of Pt. A HAADF-STEM image of PtCo with N2 (right bottom) shows a bright Pt-enriched shell on a relatively darker PtCo alloy core (the region inside of the yellow-dotted circle). On the basis of the STEM image of PtCo with N2, the thickness of the Pt-enriched layer can be estimated to be approximately 0.9–1.2 nm (3–4 atomic layers). On the basis of the EDS spectra for several particles in the HAADF-STEM images of PtCo w/o N2 and with N2 (Figure S2), the average Pt/Co ratios of the dealloyed PtCo w/o N2 and with N2 were estimated to be approximately 80:20 and 42:58, respectively, as listed in Table . Thus, the results of STM (Figure ) and HAADF-STEM (Figure ) observation indicate that the nitridation of Co in the PtCo alloy NP system might suppress Co dissolution through the dealloying process, resulting in different dealloyed NP structures. The voltammetric dealloying mechanism will be further discussed in the following section.

Figure 3

Typical STEM–EDS line profiles (top) and HAADF-STEM images (8 × 8 nm2: bottom) of PtCo w/o N2 (left) and with N2 (right) after voltammetric dealloying. The insets show the corresponding HAADF-STEM images, where the red arrows indicate the line-scan direction.

Table 1

Estimated Alloy Compositions of the As-Prepared (Charge Ratios) and Dealloyed Samples Estimated using QCM and EDS

Pt:Co	as-prepared (QCM)	dealloyed (EDS)
PtCo w/o N2	1:2	80:20
PtCo with N2	1:2	42:58

As shown in Figure A, a cyclic voltammogram (CV) of the dealloyed PtCo with N2 enlarged when compared with that of the dealloyed PtCo w/o N2: shapes of the CVs correspond well to the results reported for PtCo[9] and PtCoN[13] in the literature. Furthermore, the linear sweep voltammograms collected at 1600 rpm (Figure B) show that the half-wave potential of PtCo with N2 shifts positively when compared with that of PtCo w/o N2, indicating the enhanced ORR activity for the former sample. We used Koutecky–Levich plots to estimate the Pt mass activities on the basis of the kinetic controlled currents at 0.9 V (vs RHE); the results are summarized in Figure C. The Pt mass activities of the dealloyed PtCo w/o N2 (black) and with N2 (pink) are 1.1 and 1.8 A mg–1 Pt, respectively: the ORR enhancement factors for the dealloyed samples are higher than the average Pt mass activity for a typical commercial Pt/C catalyst (gray).[12,14,23] The EC surface area (ECSA) values for the dealloyed PtCo w/o N2 and with N2 were estimated to be 0.070 and 0.109 cm2, corresponding to 108 and 169 m2 g–1 Pt, respectively. Thus, respective area-specific ORR activities are 1.0 and 1.1 mA cm–2 ECSA, respectively. At any rate, the dealloyed PtCo with N2 sample is more ORR-active than the dealloyed PtCo w/o N2. XPS spectra of the Co 2p bands for both dealloyed samples are presented in Figure D: the Co 2p bands are almost identical, irrespective of Co nitridation. After the dealloying process, the existence of Co nitride in the core of PtCo with N2 NPs is excluded; that is, the XPS results for the dealloyed samples (Figure D) suggest the dissolution of the Co–N bonds and/or the detachment of the introduced N atoms in the core during the voltammetric dealloying process. Furthermore, the Pt/Co atomic ratio estimated from the XPS spectrum for the dealloyed NPs of PtCo with N2 increases from 1.5 to 8 through the dealloying process, indicating the enrichment of Pt at near-surface regions. Consequently, the results obtained in this study show that the Pt mass activity enhancement of PtCo with N2 could be caused by the enlargement of the specific surface area and the preferential formation of the core–shell architecture through the consumption of not only Co atoms but also cobalt nitride in the pristine core during the voltammetric dealloying process.

Figure 4

(A) CVs in N2-purged 0.1 M HClO4, (B) linear sweep voltammograms in O2-saturated 0.1 M HClO4, (C) Pt mass activity, and (D) Co 2p XPS spectra for the dealloyed PtCo w/o N2 (black) and with N2 (pink) NPs. (E) Schematic of the dealloying behaviors at near-surface regions of the PtCo w/o N2 (top) and with N2 (bottom) NPs during anodic (left) and cathodic (right) scans. Asterisk (*) indicates typical activity for the commercial Pt/C catalysts.[12,14,23]

Given the potential window used in our experiments (0.05–1.05 V), the oxygenated adsorption/desorption behaviors should modify the voltammetric dealloying process.[24] Thus, we here discuss the structural dependence of the pristine nitrogen-introduced Pt–Co alloy core on the oxygen-related adsorbates. The XRD results (Figure A) provide strong evidence that Co nitride and/or doped N atoms in the core exist as interstitial solutes, thereby expanding the PtCo lattices. Such a lattice expansion should relax the compressive strain of the Pt-rich shell in the initial stage of dealloying. The compressive strain on the Pt-rich shell surface is well-known to increase with decreasing amounts of the oxygen-related species adsorbed onto the surface of the Pt atoms.[2] The desorption of the oxygen-related adsorbates by cathodic sweep induces the replacement of the surface Pt atoms to eliminate the difference in the surface energy.[26] Indeed, Sasaki et al. showed that “place exchange” at the surface of Pt NPs should occur at relatively low potential such as 0.91 V.[25] Therefore, the surface Pt atoms might be replaced through place exchange under voltammetric dealloying conditions. Consequently, the nitrogen in the core should lead to less compressive strain of the Pt-rich shell, accompanied by the activation of Pt surface diffusion, during the voltammetric dealloying process. Furthermore, the introduced nitrogen is expected to affect the EC stability of Co in the core. According to the literature, the electrode oxidation (dissolution) current for the amorphous CoN electrode begins at much more positive potentials than that for Co metal.[27] Furthermore, Zhong et al. carried out density functional theory calculations that showed that the oxygenated adsorption of Co4N was weaker than that of the Co metal.[20] Because the adsorption of oxygen-related species onto the Pt-based alloy surface tends to induce the surface segregation of the less-noble metals,[28] Co nitridation might contribute to the preferential formation of the core–shell architecture. On the basis of the previous discussions, we illustrated schematics of the voltammetric dealloying behaviors of the PtCo w/o N2 (left) and with N2 (right) NPs during anodic (left) and cathodic (right) scans (Figure E). Published studies clearly reveal that the extent of the dealloying of the Pt-based alloy NPs is largely controlled by the ratio (R) between the Pt surface diffusion and the less-noble metal dissolution rates.[10,29] Therefore, the introduced nitrogen atoms in the core might increase the value of R, resulting in the preferential multilayered Pt-rich shell/Pt–Co alloy core formation for PtCo with N2 through the consumption of the introduced nitrogen atoms in the core.

Conclusions

We investigated the voltammetric dealloying of PtCo with N2 and w/o N2 samples fabricated by the synchronous APD of Pt and Co. The XRD and XPS results provided strong evidence for Co nitridation in the core for the synchronous APD-PtCo NPs under an N2 partial pressure of 0.1 Pa. The morphologies and nanostructures of the dealloyed PtCo samples were observed using STM and STEM, indicating that the N atoms introduced into the pristine core (PtCo with N2) contribute to the preferential multilayered Pt-rich shell/Pt–Co core formation through voltammetric dealloying. The ORR activity of the dealloyed PtCo with N2 was 1.6 times greater than that of the dealloyed PtCo w/o N2. The results obtained in this study demonstrate that the introduction of nitrogen into the pristine Pt–M alloy NPs is a key factor in controlling the core–shell nanostructures via voltammetric dealloying.

Experimental Details

Sample Fabrication

Synchronous APD using two APD sources (ULVAC-RIKO ARL-300) was used to synthesize the Pt–Co alloy NPs on the HOPG (Optigraph) substrates (10 × 10 mm2; thickness: 1 mm) under ultrahigh vacuum (∼10–8 Pa) for the Pt–Co NPs (denoted as PtCo w/o N2) or under an N2 partial pressure of 0.1 Pa (denoted as PtCo with N2). The substrate temperature during APD was maintained at 773 K to control the size distributions and morphologies of the NPs. Using a quartz-crystal microbalance (QCM) installed in the APD chamber, we estimated the amounts of the deposited Pt and Co normalized to the HOPG geometric area to be approximately 0.57 μg cm–2 HOPG (Pt) and 0.34 μg cm–2 HOPG (Co).

EC Studies

The APD-fabricated samples were transferred from the APD chamber to an EC measurement setup in an N2-purged glove box without air exposure. A potentiostat with a rotating disk electrode (RDE) system (Hokuto Denko HZ-5000) was used for the EC measurements. The voltammetric dealloying and EC measurements were taken using a glass cell that included a Pt counter electrode and a reversible hydrogen electrode (RHE) under flowing H2 gas. The electrolyte solution was prepared from perchloric acid (HClO4; Sigma-Aldrich, Ultrapure) and Milli-Q water. The samples were electrochemically dealloyed by being cycled between 0.05 and 1.05 V for 300 cycles at 500 mV s–1 in N2-purged 0.1 M HClO4 using a triangular wave potential. The EC properties for the dealloyed PtCo NPs were evaluated by cyclic voltammetry in N2-purged 0.1 M HClO4 at a scan rate of 50 mV s–1 and by linear sweep voltammetry performed using the RDE in O2-saturated 0.1 M HClO4 at a scan rate of 10 mV s–1. The ECSA was estimated from the hydrogen adsorption values in the potential region from 0.08 to 0.38 V, and the charge for the monolayer hydrogen adsorption was assumed to be 210 μC cm–2. We used Koutecky–Levich plots to estimate the Pt mass activities on the basis of the kinetic controlled currents at 0.9 V (vs RHE).

Characterization

XRD patterns for the corresponding as-prepared samples were obtained on a X-ray diffractometer (Rigaku SmartLab) operating at 9 kW (45 kV, 200 mA) using Cu Kα radiation (λ = 0.1542 nm). The samples for XRD measurements were prepared on an amorphous carbon-coated glass plate; the deposition amounts were approximately 16 times greater than those of the samples used for other structural observations. XPS was performed using a VG Microtech Multilab XPS system (Thermo Electron, Thermo VG). The APD-fabricated samples were transferred from the APD chamber to the XPS analysis chamber under an ambient atmosphere. The sample morphologies of the Pt–Co alloy NPs on the substrate before and after the cycles were observed using STM (Bruker multimode V) in air. The NP distributions were calculated from the corresponding STM images using the ImageJ analysis software. The Pt–Co alloy NP microstructures prepared on the amorphous carbon thin films were examined using a HAADF-STEM (JEOL JEM-ARM200F) equipped with an EDS apparatus.