Proton exchange membrane fuel cells (PEMFCs) have attracted considerable attention as distributed power sources for automotive and residential applications. In PEMFCs, platinum (Pt) nanoparticles supported on carbon materials are used as electrocatalysts for oxygen reduction reactions. However, improved durability of the electrocatalysts is urgently required for commercialization purposes. We describe an ion implantation technique for the preparation of Pt nanoparticles with superior catalytic properties, suitable for fuel cell applications. The Pt nanoparticles were obtained by implanting a glassy carbon (GC) substrate with 100 keV Pt-ions, followed by electrochemical etching and a heat treatment. Electrochemical measurements of the Pt nanoparticles prepared by the proposed implantation technique demonstrated superior durability when compared to those prepared by the conventional magnetron sputtering method. We suggested that the defective structure of the GC substrate, caused by the Pt-ion implantation, has led to the improved electrochemical stability of the GC substrate and Pt-carbon support interactions, thereby enhancing the durability of our Pt nanoparticles.
The ion implantation is used as a powerful technique for the addition of impurities to host materials,1 thereby modifying their properties, for e.g., those of carbon materials through physical and chemical effects.2 The electrochemical properties, including wear resistance and wettability, of glassy carbon (GC) are improved by the implantation of Li-, N-, O-, F-, Na-, K-, and Zn-ions at energies of 25–150 keV.3–5 Such modifications are considered to be affected by the formation of lattice defects caused by atomic collisions, resulting in the formation of new functional groups on the surfaces of the GC substrate.
Ion implantation has also been studied extensively for the fabrication of nanoparticles in ceramic oxide substrates such as SiO2 and Al2O3, used in optical applications and chemical catalysis. Previous research studied the dispersion of metal and/or semiconductor particles in optical materials, and investigated their resulting luminescence and nonlinear optical behaviors.6–11 Only a few studies have focused on the preparation of nanoparticles on the surface of substrates for catalytic applications; most of these involved the growth of carbon nanotubes.12,13 If the incident energy of the ions is too high, the nanoparticles form inside the target material due to a high ion-stopping range of more than a few tens of nanometers. A process such as chemical etching is therefore required to expose the implanted atoms and/or particles at the surface of the substrate. Adhikari et al. synthesized nickel-nanoparticles on SiO2 surface layers via nickel-ion implantation, thermal annealing, and chemical etching in a hydrogen fluoride aqueous solution.12 Choi et al. achieved the formation of iron nanoparticles on SiO2 surfaces through iron-ion implantation and subsequent annealing in a H2 flow.13
In our preparation of nanoparticles for electrocatalytic applications we made use of a conductive carbon substrate as the target material. The use of a conductive substrate differentiates our approach from those described in the previous studies.12,13 Liang et al. fabricated platinum (Pt) nanoparticles on the surface of Pt+-implanted indium tin oxide substrates for application in electrochemical devices, and reported that the prepared Pt nanoparticles exhibited high activity and durability in methanol oxidation reactions.14 Our research specifically aimed at the preparation of electrocatalysts for use in proton exchange membrane fuel cells (PEMFCs). PEMFCs have a low environmental impact and are highly efficient as distributed power sources for automotive and residential applications. However, their commercialization is impeded by the high cost of the Pt catalysts that are used in the oxygen reduction reactions (ORRs) of PEMFCs. To curb these costs, the Pt loading must be reduced, while improving the catalytic activity and durability of the Pt nanoparticles.15
During PEMFC operation, Pt nanoparticles are subjected to continuous potential fluctuations, with their surfaces alternating between metallic and oxidized states, promoting the dissolution of Pt and the corrosion of the carbon support. The durability of the Pt nanoparticles for ORR is affected by the nature of the carbon support, which is determined by the electronic interactions at the interface.16 Several computational studies have suggested that the strong electronic interactions between Pt and a defective carbon support enhance the electrochemical properties of Pt catalysts.17,18 We have recently reported Pt-carbon support interactions in GC substrates irradiated with Ar-ions.19,20 This has motivated our current research, in which we investigated how modifications of the carbon substrate surface by high-energy ion implantation and the resulting interface interactions enhance the durability of Pt nanoparticles.
Here, we describe our ion implantation technique, and report on the evaluation of the chemical and catalytic properties of the prepared Pt nanoparticles. Our Pt nanoparticles demonstrated model catalyst properties using the GC substrate as support. In a previous study we found that the implanted Pt atoms remained in an amorphous state and did not aggregate in the Pt-ion-implanted GC substrate, leading us to speculate that the electron transfer from Pt to C was a result of lattice defects in the GC substrate, introduced during Pt-ion implantation.21 We therefore used electrochemical etching and a heat treatment to produce our Pt nanoparticles from the Pt+-implanted layer in the GC substrate. We evaluated the defect structure of the GC in the prepared Pt nanoparticle samples using Raman spectroscopy, and assessed the chemical and catalytic properties of the Pt nanoparticles with x-ray photoelectron spectroscopy (XPS) and electrochemical measurements. The possible reasons for the high durability of the prepared Pt nanoparticles are also discussed.
II. PREPARATION PROCESS
Figure 1 shows a schematic of the preparation of nanoparticles by (i) ion implantation, (ii) electrochemical etching, and (iii) heat treatment.
Pt-ions were implanted into polished GC (10 × 10 × 1 mm3) [Fig. 1(i)]. The GC substrate was obtained from Tokai Carbon Co., Ltd., Japan. Pt+-implantation was done at the Takasaki Ion Accelerators for Advanced Radiation Application (TIARA) facility at the Takasaki Advanced Radiation Research Institute, QST, Japan. The ion implanter provided low-charge DC ion beams of Pt. The fluence of the Pt-implantation was set to a range of 1.0 × 1016–2.0 × 1016 ions/cm2, based on the findings of a previous study.21 Implantation was performed at an energy of 100 keV. The current density was measured every ∼30 min (0.035–0.12 μA/cm2), and the total irradiation time was ∼12 h. A Pt disk (99.95%, RARE METALLIC CO., LTD.) was used as the ion source. The base pressure of the irradiation chamber was ∼6 × 10−5 Pa.
Due to the chemical stability of GC, etching agents such as hydrogen fluoride aqueous solutions used for SiO2-substrates are not effective for the etching of GC substrates. The Pt+-implanted layer in the GC substrate was therefore exposed by electrochemical anodization [Fig. 1(ii)].22,23 This electrochemical etching technique allows for the control of the etching process and etching depths by monitoring the current densities. Electrochemical etching was particularly suitable for the Pt+-implanted layer, as it was performed at only several tens of nanometers below the surface. In this process, the GC layer on the sample surface was etched by approximately 40 nm to expose the Pt+-implanted layer.
Pt nanoparticles were produced by a heat treatment of the exposed implanted layer [Fig. 1(iii)]. The conditions of heat treatment were specified by in situ transmission electron microscopy (TEM) observation for the etched samples.
A. Electrochemical etching
The cross-section of the Pt+-implanted GC substrate was examined using TEM. The cross-sectional specimens were prepared using a focused ion beam device (FB2100, Hitachi HighTechnologies Corporation, Japan). Amorphous carbon and tungsten films were deposited on the samples for surface protection.
Raman spectroscopic analysis was performed over the wavenumber range of 800–2400 cm−1, using a LabRAM HR Evolution Raman spectrometer (HORIBA, Ltd., Japan). The incident radiation at 514 nm was generated using an argon laser. The laser intensity at the sample surface was set to 0.3 mW using an ND filter of 10%. Raman spectra were obtained with a back-scattering configuration by accumulating the signal during a 1 min exposure, and analyzed by fitting all peaks to a Lorentzian profile after background subtraction.
XPS measurements were performed using a PHI 5000 Versa Probe (ULVAC-PHI, Inc., Japan) with an Al Kα (1486.6 eV) x-ray light source at pass energy of 23.5 eV. Charge correction was not required as GC has a high electrical conductivity and charge-neutralizer system was used during the measurement. A 2 keV Ar+ sputtering gun was used for surface cleaning, set at a sputtering duration of 5 s, corresponding to a sputter depth of less than 1 nm.
Cyclic voltammetry was performed in an N2-saturated 0.1M HClO4 solution at 25 °C using an HZ-5000 Potentiostat (Hokuto Denko Corp., Japan), a three-electrode cell with a KCl-saturated Ag/AgCl reference electrode, and a Pt-wire counter electrode. For the electrochemical characterization of the samples, CVs were recorded in the potential range of 0.05–1.25 V (vs RHE) at a potential scan rate of 50 mV/s. The electrochemical surface area (ECSA) of the samples was determined from the hydrogen adsorption region using a conversion factor of 210 μC/cm2.24 The ORR activity of the samples was evaluated by the rotating disk electrode (RDE) setup in an O2-saturated 0.1M HClO4 solution, at a scan rate of 10 mV/s and a rotation speed of 400 rpm. An accelerated durability test was performed in the potential range of 1.0-1.5 V (vs RHE) at a potential scan rate of 500 mV/s for 52 000 cycles.25 This potential range was applied when evaluating the resistance to oxidative corrosion of the carbon support for the Pt catalyst. The CV was measured after a durability test scan of several cycles, and the durability of the sample was evaluated based on the decrease in the ECSA. The potentials (V vs Ag/AgCl) were converted to the RHE standard.
A. Aggregation by heat treatment
The optimal conditions for the heat treatment were established by observing the morphological changes of Pt atoms upon heat treatment with in situ TEM. A GC substrate sample, implanted with 100 keV Pt+ at a fluence of 1.5 × 1016 ions/cm2 and electrochemically etched at 1.9 V for 30 s, was used for this purpose. The etched surfaces were covered with amorphous carbon and tungsten films. The heat treatment was performed in the vacuum chamber of the TEM equipment while observing. Only the implanted layer was observed at room temperature, as previously reported,21 and this remained so up to 350 °C. The TEM images were taken when the particle changes had settled down, i.e., 10 min after each of the observation temperatures 375, 400, and 420 °C had been reached. After a gradual increase of temperature from RT to 375 °C, the particles started to agglomerate [Fig. 3(a)]. Particles of approximately 10 nm were observed on the upper and lower sides of the implanted layer. Larger particles formed at a temperature of 400 °C [Fig. 3(b)], while the particles grew to a size of approximately 40 nm at a temperature of 420 °C [Fig. 3(c)]. Larger particles also formed on the sides of the deposited amorphous carbon film. While particle agglomeration started at 375 °C, most of the implanted atoms remained static.
The particle sizes observed at 425 °C were deemed too large, and we thus selected 400 °C as optimal temperature for the agglomeration and preparation of our Pt nanoparticles. We therefore heat-treated our etched sample at 400 °C for 1 h in an N2 atmosphere to convert the Pt atoms to Pt nanoparticles, after which the sample was allowed to cool naturally. Figures 4(a) and 4(b) show cross-sectional TEM images of the prepared nanoparticle samples. The black circles indicate the Pt nanoparticles. When heat-treatment was performed under the same conditions without etching, no agglomeration of Pt particles was observed at the implanted layer. Figure 4(a) shows that the larger particles migrated into the interior of the GC substrate, and that only small particles formed near the surface of the GC substrate. As shown in Fig. 4(b), the Pt nanoparticles with diameters ranging from a few nanometers to 10 nm were found to be uniformly dispersed near the surface. The electron diffraction image corresponding to the  plane of the face-centered cubic (fcc) lattice was obtained from one of the Pt nanoparticles, indicating that it has an fcc structure [inset in Fig. 4(b)]. Figures 3(c) and 3(d) show the scanning electron microscope (SEM) images of the sample surface. The Pt nanoparticles appear brighter in the SEM images. In Fig. 3(c), holes of approximately 50 nm are observed in addition to larger particles of approximately 20–50 nm. In the cross-sectional TEM image in Fig. 3(a), a tunnel-like trace is observed near the large particles at an approximate depth of 50–150 nm, suggesting that the particles formed on the sample surface and migrated into the substrate interior. The migration of the particles into the substrate may be enabled by corrosion of the GC substrate, presumably caused by residual oxygen resulting from catalysis by the Pt nanoparticles. The magnified image [Fig. 3(d)] also shows the formation of many smaller particles with diameters ranging from a few nanometers to 10 nm. SEM observations confirm that the small particles were exposed on the sample surface.
B. Defect structure of GC
The structure of the GC substrate in the prepared samples was characterized by Raman spectroscopy. Raman spectroscopy has extensively been used to characterize carbon materials, including HOPG, with Pt nanoparticles deposited because the vibrational frequencies of optical phonons are sensitive to the structure of carbon materials and C–C bonding.26–29 Pt nanoparticles were deposited on the GC substrate by radio-frequency (rf) magnetron sputtering to prepare a reference sample for the characterization experiments (Fig. S1; supplementary material). A sample produced by rf magnetron sputtering is considered a suitable reference because the damage in the GC substrate during rf magnetron sputtering is smaller than that by ion implantation of 100 keV Pt+. We prepared Pt nanoparticles by rf magnetron sputtering in our previous studies.19,30,31 In the present study, the Pt nanoparticles deposited on the GC substrate were then annealed under the same conditions as those prepared by ion implantation [Fig. 1(iii)]. As particle size affects durability and catalytic performance of nanoparticles, the deposition conditions were set so that the particle size obtained by rf magnetron sputtering was approximately the same as that obtained by ion implantation. The sputtering time and plasma output were 24 s and 20 W, respectively. The size of the resulting particles was approximately 5–10 nm (Fig. S2; supplementary material). We labeled the Pt nanoparticles prepared by ion implantation the “implanted sample”, and those obtained by rf magnetron sputtering the “deposited sample.”
Figure 5 shows the Raman spectra of the implanted and deposited samples. Three bands (D, G, and D′) are observed in the Raman spectra. The position and full width at half maximum (fwhm) of the D- and G-bands obtained by multi-peak fitting the Raman spectra are listed in Table I. The peak at 1585 cm−1 is identified as the E2g mode of the graphitic structure, known as the G-band. The peak at 1350 cm−1, called the D-band, is attributed to the disordered mode of graphite (A1g mode). For the deposited sample, the G band is located at 1585 cm−1; the intensity ratio between the D- and G-peaks (ID/IG) is 1.86. This result is similar to that obtained for the Raman spectrum of pristine GC,19 showing that the structural changes of the GC caused by Pt nanoparticle deposition are slight. Conversely, the fwhm of the G-band of the implanted sample (65.7 cm−1) is wider than that of the deposited sample (61.5 cm−1), while the intensity ratio is smaller (ID/IG = 1.33), suggesting the formation of a more defective structure in the GC substrate32,33 during Pt+ implantation.
|Sample .||Implanted sample .||Deposited sample .|
|Position (cm−1) .||FWHM (cm−1) .||Position (cm−1) .||FWHM (cm−1) .|
|Sample .||Implanted sample .||Deposited sample .|
|Position (cm−1) .||FWHM (cm−1) .||Position (cm−1) .||FWHM (cm−1) .|
C. Chemical composition of the catalyst surface
The chemical states of the prepared nanoparticles were investigated by XPS. Figure 6 shows the Pt 4f XPS spectra for the (a) implanted sample, (b) deposited sample, and (c) the standard Pt foil. The spectra were subjected to background-correction, normalization of intensity, and alignment. The most intense pair of XPS peaks at binding energies of 71.0 and 74.3 eV for both samples is attributed to metallic Pt(0). It should be noted that the Pt 4f XPS doublets in spectra (a) and (b) are broader on the high-energy side than those in (c); the presence of chemically different species is evident in the difference spectra [Figs. 6(a-c) and 6(b-c)].
In the difference spectrum between the implanted sample and standard Pt foil [Fig. 6(a-c)], the doublet appears at the binding energies of 72.1 and 75.4 eV. The Pt 4f7/2 peak was also located at around 72.1 eV in the XPS spectrum of the Pt-ion-implanted layer before electrochemical etching and heat treatment, due to Pt–C interactions.21, Figure 6(a-c) suggests that the Pt–C interactions were still present at the interface between the Pt nanoparticles and the GC substrate after annealing. In addition, the difference spectrum suggested that states with energies higher than 72.1 eV also existed, indicating that Pt–O (73.3 eV)34 may form on the Pt surface during the heat treatment process. The difference spectrum between the deposited sample and the standard Pt foil [Fig. 6(b-c)] also shows a doublet around 72.1 and 75.4 eV, similar to the implanted sample. The difference spectra [Figs. 6(a-c) and 6(b-c)] suggests that the deposited sample has a larger fraction of Pt atoms involved in the Pt–C bonding compared with the implanted sample. The XPS results show that both the implanted and deposited samples have Pt–C bonds at the interface between the Pt particles and the GC substrate, with metallic Pt as the main component.
D. Electrochemical property
The change in the ECSA of Pt in the implanted sample, measured by the accelerated durability test, was compared with that in the deposited sample. Figure 7 shows the CVs of the (d) implanted and (e) deposited samples after 1000, 16 000, 32 000, and 48 000 cycles; the change in ECSA of both the samples is plotted against the number of cycles in Fig. 7(f). The vertical axis was normalized to 1.0, which is the ECSA in the initial state (0 cycle). The ECSA of both samples increased up to approximately 1000 cycles, owing to surface cleaning. For the deposited sample, the ECSA decreased monotonically, and after 52 000 scan cycles, the ECSA was 55% of the maximum. The decrease in ECSA corresponds to the decrease in the active area due to the desorption of Pt nanoparticles from the GC substrate. In contrast, the ECSA of the implanted sample was 87% of the maximum value after 52 000 scan cycles. By comparing the rate of decrease, the durability of the implanted sample was shown to be about 3.5 times higher than that of the deposited sample. It is noteworthy that Yang et al. evaluated the durability of Pt catalysts using the same protocol as was used in our study, and reported that the ECSA of commercial Pt-nanoparticles supported on carbon black (commercial CB/Pt) lost ∼50% of the initial value after 10 000 potential cycles between 1.0 and 1.5 V vs RHE.35 This indicated that the implanted sample had a significantly higher durability than the commercial catalyst.
We have investigated the probable reasons for the improved durability of our nanoparticles, based on a comparison of the cyclic voltammograms recorded for the implanted and deposited samples. As shown in Fig. 7(d), it can be seen that, the CV shape of the implanted sample does not change significantly until 48 000 cycles, except for 1000 cycles. Conversely, the CV shape of the deposited sample changes significantly between 1000 and 16 000 cycles and then continues to change [Fig. 7(e)]. A decline in durability is considered to be caused by an increase in Pt particle size due to agglomeration, dissolution of Pt particles, and desorption of Pt particles from the supporting carbon due to corrosion of the carbon support.36,37 Corrosion of the carbon support and desorption of the Pt particles would occur in the deposited sample, while in contrast no such degradation would occur in the implanted sample. A possible reason for the more stable voltammogram of the implanted sample compared to the deposited sample is that ion implantation improved the corrosion resistance of the GC substrate itself. Takahashi et al. reported that 150 keV Zn-, Cd-, Ti-, and Ar-ion implantations at fluences of 1.0 × 1016–1.0 × 1017 ions/cm2 stabilized the GC substrate electrochemically.38 The implanted sample had a more defective structure than the deposited sample (Fig. 5), and therefore it was expected that the GC substrate in the implanted sample would display a corrosion resistance similar to that reported by Takahashi et al. This suggests that the stability of the implanted sample resulted from an improvement of the electrochemical properties of the GC substrate itself. Furthermore, we suggest that the formation of Pt–C bonds may inhibit Pt desorption.
In this study, we prepared Pt nanoparticles by a novel process that combined Pt+ implantation into a GC substrate with surface layer removal by electrochemical etching, and agglomeration by heat treatment. The catalytic properties of the prepared Pt nanoparticles were evaluated using electrochemical measurements. The produced Pt nanoparticles demonstrated superior durability, considered to be attributable to the improvement of the electrochemical stability of the GC. The improved electrochemical stability, in turn, may be ascribed to the structural changes caused by ion implantation, resulting in a defective structure of the GC substrate. The defective structure could promote the formation of Pt–C bonds, thereby contributing to the improved durability of the Pt nanoparticles.
See the supplementary material for the schematic drawing of the preparation method for the deposited sample and the SEM image of the deposited sample.
This study was partially conducted under the NIMS-RIKEN-JAEA-QST Cooperative Research Program on Quantum Beam Science and Technology. This work was supported in part by JSPS KAKENHI (Grant Nos. 20760600, 22760678, 24561047, 18H01923, and 21H04669) from the Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan.
Conflict of Interest
The authors have no conflicts to disclose.
Tetsuya Kimata (木全哲也): Data curation (equal); Formal analysis (equal); Investigation (equal); Writing – original draft (equal). Sho Kato (加藤 翔): Data curation (equal); Formal analysis (equal); Investigation (equal). Tomohiro Kobayashi (小林 知洋): Formal analysis (equal). Shunya Yamamoto (山本 春也): Formal analysis (equal). Tetsuya Yamaki (八巻 徹也): Conceptualization (equal); Funding acquisition (equal); Investigation (equal); Project administration (equal); Writing – review & editing (equal). Takayuki Terai (寺井 隆幸): Supervision (equal).
The data that support the findings of this study are available within the article.