Enhanced Electrocatalytic Activity and Stability toward the Oxygen Reduction Reaction with Unprotected Pt Nanoclusters

The Pt particles within diameters of 1–3 nm known as Pt nanoclusters (NCs) are widely considered to be satisfactory oxygen reduction reaction (ORR) catalysts due to higher electrocatalytic performance and cost effectiveness. However, the utilization of such smaller Pt NCs is always limited by the synthesis strategies, stability and methanol tolerance of Pt. Herein, unprotected Pt NCs (~2.2 nm) dispersed on carbon nanotubes (CNTs) were prepared via a modified top-down approach using liquid Li as a solvent to break down the bulk Pt. Compared with the commercial Pt/C, the resultant Pt NCs/CNTs catalyst (Pt loading: 10 wt.%) exhibited more desirable ORR catalytic performance in 0.1 M HClO4. The specific activity (SA) and mass activity (MA) at 0.9 V for ORR over Pt NCs/CNTs were 2.5 and 3.2 times higher than those over the commercial Pt/C (Pt loading: 20 wt.%). Meanwhile, the Pt NCs/CNTs catalyst demonstrated more satisfactory stability and methanol tolerance. Compared with the obvious loss (~69%) of commercial Pt/C, only a slight current decrease (~10%) was observed for Pt NCs/CNTs after the chronoamperometric measurement for 2 × 104 s. Hence, the as-prepared Pt NCs/CNTs material displays great potential as a practical ORR catalyst.


Introduction
The oxygen reduction reaction (ORR) is a vital component in fuel cells.
Currently, Pt-based nanomaterials are the most effective catalysts to improve the ORR performance [1]. However, they suffer from a series of drawbacks including high costs, unsatisfactory catalytic activity, poor stability and lower methanol tolerance [2,3]. Hence, a large number of research articles have been focusing on figuring out these problems [4][5][6]. Among a wide variety of strategies, reducing the size of Pt into smaller size less than 3 nm referred as nanoclusters (NCs) has been frequently highlighted in that Pt NCs present more satisfactory ORR performance compared with larger Pt nanoparticles (NPs) because of the smaller size, narrower size distribution and more active sites [7][8][9]. More importantly, the smaller size endows Pt NCs with a large proportion of low-coordinated surface atoms, which is helpful for the adsorption and activation of O 2 on the catalyst surface [10]. As a consequence, Pt NC catalysts adopted in ORR can both dramatically enhance the electrocatalytic activity and lower the cost of Pt. However, Pt NCs usually require to be protected by the stabilizing agents to prevent the aggregation, which would hinder the mass transfer process and electron transportation and thus give

Electrocatalytic Measurements
Cyclic voltammograms (CV) and rotating disk electrode (RDE, Pine Instrument, Hong Kong, China) measurements were conducted on CHI660E to evaluate the ORR performance. All data were recorded at room temperature in a three-electrode cell in 0.1 M HClO 4 . Pt foil and Ag/AgCl electrodes are used as counter electrode and reference electrode, respectively. All potentials in this work are reported versus reversible hydrogen electrode (RHE) via converting the measured potentials using Equation (1): Prior to the CV and RDE measurements, the glassy carbon electrode was polished with 0.3 µm alumina slurries (Buehler, IL, USA) to a mirror finish and then washed and ultrasonically cleaned in ultrapure water for 3 min. The 0.1 M HClO 4 electrolyte was saturated with O 2 (99.999%), N 2 (99.999%) or Ar (99.999%). The working electrode was prepared as follows: Suspend 6 mg Pt NCs/CNTs-10% in the mixture of Nafion (25 µL, 5 wt.%) and C 2 H 5 OH (475 µL) and then sonicated for 30 min to obtain a homogeneous suspension. 10 µL of the suspension was deposited on the Φ5 mm glassy carbon disk and dried in air. The Pt loading on the glassy carbon was ∼0.06 mg cm −2 . CV data were measured between 0 and 1.2 V at 50 mV s −1 . RDE data were obtained at 10 mV s -1 . The stability and the methanol tolerance of Pt NCs/CNTs-10% were investigated by the chronoamperometric measurement.

Physicochemical Characterizations
XRD patterns are conducted to characterize the crystal structure of Pt NCs/CNTs-10%. As depicted in Figure 1, a typical fcc crystal phase is found for Pt NCs/CNTs-10%. The as-prepared Pt NCs/CNTs-10% exhibits the same diffraction patterns as the standard Pt card (JCPDS: 04-0802). The XRD pattern of CNT support displays two peaks at~25.7 • and~42.5 • assigned to the (002) and (100) reflections, respectively [21]. The (002) peak is also clearly observed in the XRD pattern of Pt NCs/CNTs-10%. The (100) peak is weak and overlapped by the diffraction at~46.2 • attributed to the Pt (200) reflection. The mean particle size of Pt NCs is calculated to be 2.4 nm from the Pt (111) diffraction using the Scherrer equation. The geometric surface area of Pt NCs is determined to be 116.8 m 2 g −1 from the XRD analysis using the Equation (2): where ρ is the density of Pt (21.4 g cm -3 ) and d/nm is the average particle size determined by the XRD pattern.   Figure 1 shows the HRTEM image of the Pt NCs. The Pt (111) surface orientation is determined through investigating the distance between two columns of Pt atoms. The corresponding elemental mapping (Figure 2c,d) and selected-area EDS ( Figure 2e) analyses further clearly demonstrate the uniform distribution of Pt NCs on the CNTs. According to the ICP-OES analysis, the Pt mass loading is about 8.7%, which slightly dropped from the theoretical value of 10% probably caused by the limited surface area of CNTs, which is determined to be 253 m 2 g -1 according to the N2 absorption/desorption isotherms ( Figure 3). The Brunauer-Emmett-Teller (BET) surface area of Pt NCs/CNTs-10% and Pt/C-20% are also carried out, which are determined to be 217 m 2 g −1 and 411 m 2 g -1 , respectively. Obviously, the BET surface area of Pt NCs/CNTs-10% is smaller than that of CNTs, which can be ascribed to the framework shrinkage during the process of preparing the catalyst.    Figure 1 shows the HRTEM image of the Pt NCs. The Pt (111) surface orientation is determined through investigating the distance between two columns of Pt atoms. The corresponding elemental mapping (Figure 2c,d) and selected-area EDS (Figure 2e) analyses further clearly demonstrate the uniform distribution of Pt NCs on the CNTs. According to the ICP-OES analysis, the Pt mass loading is about 8.7%, which slightly dropped from the theoretical value of 10% probably caused by the limited surface area of CNTs, which is determined to be 253 m 2 g -1 according to the N 2 absorption/desorption isotherms ( Figure 3). The Brunauer-Emmett-Teller (BET) surface area of Pt NCs/CNTs-10% and Pt/C-20% are also carried out, which are determined to be 217 m 2 g −1 and 411 m 2 g -1 , respectively. Obviously, the BET surface area of Pt NCs/CNTs-10% is smaller than that of CNTs, which can be ascribed to the framework shrinkage during the process of preparing the catalyst.  According to the ICP-OES analysis, the Pt mass loading is about 8.7%, which slightly dropped from the theoretical value of 10% probably caused by the limited surface area of CNTs, which is determined to be 253 m 2 g -1 according to the N2 absorption/desorption isotherms ( Figure 3). The Brunauer-Emmett-Teller (BET) surface area of Pt NCs/CNTs-10% and Pt/C-20% are also carried out, which are determined to be 217 m 2 g −1 and 411 m 2 g -1 , respectively. Obviously, the BET surface area of Pt NCs/CNTs-10% is smaller than that of CNTs, which can be ascribed to the framework shrinkage during the process of preparing the catalyst.   XPS measurement is employed to analyze the chemical states of Pt NCs. The C 1s peak at 284.6 eV is adopted as a reference for the correction of the binding energy positions [22]. As shown in Figure 4a, both C and Pt are detected at their specific binding energy positions. As displayed in Figure  4b, high-resolution XPS spectrum of Pt 4f presents the characteristic peaks of Pt 4f7/2 and Pt 4f5/2 at ~71.4 and ~74.7 eV, respectively, which can be divided into two pairs of doubles. The peaks appeared at ~71.35 and ~74.73 eV are characteristic for metallic Pt 0 [23]. The peaks located at 72.12 and 75.68 eV are originated from Pt 2+ [24]. Clearly, Pt 0 is predominant. Based the deconvoluted XPS, it can be determined that the Pt NCs contain 69% metallic Pt and 31% Pt oxide. In addition, there is a slight shift of the Pt 0 peak to higher binding energies, which may be attributed to the strong interactions between Pt NCs and CNTs [25], which are beneficial for improving the stability of Pt NCs/CNTs-10%.  Figure 5 shows the CVs of Pt NCs/CNTs-10% and Pt/C-20% in Ar-saturated 0.1 M HClO4 solution. A typical hydrogen adsorption/desorption behavior on Pt is clearly observed on both Pt NCs/CNTs-10% and Pt/C-20%. The electrochemical surface area (ECSA, m 2 g -1 ) are calculated using Equation (3):

Electrochemical Evaluation
where QH (mC) is the charge due to the hydrogen adsorption in the hydrogen region of the CV. 0.21 (mC cm -2 ) is the electrical charge associated with monolayer adsorption of hydrogen on Pt. mPt is the Pt loading on working electrode. Results indicate that the ECSA of Pt NCs/CNTs-10% and Pt/C-20% are determined to be 48.23 m 2 g -1 and 38.79 m 2 g -1 , respectively. The higher ECSA of Pt NCs/CNTs-10% can be attributed to the small size and uniform distribution of Pt NCs on the CNTs. XPS measurement is employed to analyze the chemical states of Pt NCs. The C 1s peak at 284.6 eV is adopted as a reference for the correction of the binding energy positions [22]. As shown in Figure 4a, both C and Pt are detected at their specific binding energy positions. As displayed in Figure 4b, high-resolution XPS spectrum of Pt 4f presents the characteristic peaks of Pt 4f 7/2 and Pt 4f 5/2 at 71.4 and~74.7 eV, respectively, which can be divided into two pairs of doubles. The peaks appeared at~71.35 and~74.73 eV are characteristic for metallic Pt 0 [23]. The peaks located at 72.12 and 75.68 eV are originated from Pt 2+ [24]. Clearly, Pt 0 is predominant. Based the deconvoluted XPS, it can be determined that the Pt NCs contain 69% metallic Pt and 31% Pt oxide. In addition, there is a slight shift of the Pt 0 peak to higher binding energies, which may be attributed to the strong interactions between Pt NCs and CNTs [25], which are beneficial for improving the stability of Pt NCs/CNTs-10%.  XPS measurement is employed to analyze the chemical states of Pt NCs. The C 1s peak at 284.6 eV is adopted as a reference for the correction of the binding energy positions [22]. As shown in Figure 4a, both C and Pt are detected at their specific binding energy positions. As displayed in Figure  4b, high-resolution XPS spectrum of Pt 4f presents the characteristic peaks of Pt 4f7/2 and Pt 4f5/2 at ~71.4 and ~74.7 eV, respectively, which can be divided into two pairs of doubles. The peaks appeared at ~71.35 and ~74.73 eV are characteristic for metallic Pt 0 [23]. The peaks located at 72.12 and 75.68 eV are originated from Pt 2+ [24]. Clearly, Pt 0 is predominant. Based the deconvoluted XPS, it can be determined that the Pt NCs contain 69% metallic Pt and 31% Pt oxide. In addition, there is a slight shift of the Pt 0 peak to higher binding energies, which may be attributed to the strong interactions between Pt NCs and CNTs [25], which are beneficial for improving the stability of Pt NCs/CNTs-10%.  Figure 5 shows the CVs of Pt NCs/CNTs-10% and Pt/C-20% in Ar-saturated 0.1 M HClO4 solution. A typical hydrogen adsorption/desorption behavior on Pt is clearly observed on both Pt NCs/CNTs-10% and Pt/C-20%. The electrochemical surface area (ECSA, m 2 g -1 ) are calculated using Equation (3):

Electrochemical Evaluation
where QH (mC) is the charge due to the hydrogen adsorption in the hydrogen region of the CV. 0.21 (mC cm -2 ) is the electrical charge associated with monolayer adsorption of hydrogen on Pt. mPt is the Pt loading on working electrode. Results indicate that the ECSA of Pt NCs/CNTs-10% and Pt/C-20% are determined to be 48.23 m 2 g -1 and 38.79 m 2 g -1 , respectively. The higher ECSA of Pt NCs/CNTs-10% can be attributed to the small size and uniform distribution of Pt NCs on the CNTs.  Figure 5 shows the CVs of Pt NCs/CNTs-10% and Pt/C-20% in Ar-saturated 0.1 M HClO 4 solution. A typical hydrogen adsorption/desorption behavior on Pt is clearly observed on both Pt NCs/CNTs-10% and Pt/C-20%. The electrochemical surface area (ECSA, m 2 g -1 ) are calculated using Equation (3):

Electrochemical Evaluation
where Q H (mC) is the charge due to the hydrogen adsorption in the hydrogen region of the CV. 0.21 (mC cm -2 ) is the electrical charge associated with monolayer adsorption of hydrogen on Pt. m Pt is the Pt loading on working electrode. Results indicate that the ECSA of Pt NCs/CNTs-10% and Pt/C-20% are determined to be 48.23 m 2 g -1 and 38.79 m 2 g -1 , respectively. The higher ECSA of Pt NCs/CNTs-10% can be attributed to the small size and uniform distribution of Pt NCs on the CNTs. The RDE measurements for ORR are carried out in O2-saturated 0.1 M HClO4 at a potential scan rate of 10 mV s -1 . Figure 6a exhibits the liner sweep voltammetry (LSV) curves of CNTs, Pt NCs/CNTs-10% and Pt/C-20% at the rotation rate of 1600 rpm. Obviously, it can be seen that the OMC is inactive for ORR. The diffusion and diffusion-kinetic controlled regions are found for Pt NCs/CNTs-10% and Pt/C-20% in the potentials below 0.65 V and 0.65-0.95 V, respectively. It is worth mentioning that Pt NCs/CNTs-10% possess a higher onset potential (Eonset = 0.87 V) and half-wave potential (E1/2 = 0.83 V) than that of Pt/C-20% (Eonset = 0.86 V and E1/2 = 0.81 V), indicating the higher ORR activity of Pt NCs/CNTs-10%. Furthermore, the ORR diffusion limited current (JL) of Pt NCs/CNTs-10% is higher compared with that measured for Pt/C-20%. Figure 6b presents the tafel plots determined on the basis of the RDE data at 1600 rpm. The slope value for Pt NCs/CNTs-10% is 81 mV decade −1 , which is smaller than that of Pt/C-20% (87 mV decade −1 ), further indicating the more efficient ORR process on Pt NCs/CNTs-10%. The LSV curves of Pt NCs/CNTs-10% at different rotation speeds are further performed to clarify the ORR pathway. As displayed in Figure 7a, the LSV curves of Pt NCs/CNTs-10% exhibit rotatingspeed-dependent current densities. Potential in the diffusion and diffusion-kinetic regions, varying from 0.6 to 0.7 V, was chosen to calculate the values of electron transfer number (n) by Koutecky-Levich (K-L) Equation (4): The RDE measurements for ORR are carried out in O 2 -saturated 0.1 M HClO 4 at a potential scan rate of 10 mV s -1 . Figure 6a exhibits the liner sweep voltammetry (LSV) curves of CNTs, Pt NCs/CNTs-10% and Pt/C-20% at the rotation rate of 1600 rpm. Obviously, it can be seen that the OMC is inactive for ORR. The diffusion and diffusion-kinetic controlled regions are found for Pt NCs/CNTs-10% and Pt/C-20% in the potentials below 0.65 V and 0.65-0.95 V, respectively. It is worth mentioning that Pt NCs/CNTs-10% possess a higher onset potential (E onset = 0.87 V) and half-wave potential (E 1/2 = 0.83 V) than that of Pt/C-20% (E onset = 0.86 V and E 1/2 = 0.81 V), indicating the higher ORR activity of Pt NCs/CNTs-10%. Furthermore, the ORR diffusion limited current (J L ) of Pt NCs/CNTs-10% is higher compared with that measured for Pt/C-20%. Figure 6b presents the tafel plots determined on the basis of the RDE data at 1600 rpm. The slope value for Pt NCs/CNTs-10% is 81 mV decade −1 , which is smaller than that of Pt/C-20% (87 mV decade −1 ), further indicating the more efficient ORR process on Pt NCs/CNTs-10%. The RDE measurements for ORR are carried out in O2-saturated 0.1 M HClO4 at a potential scan rate of 10 mV s -1 . Figure 6a exhibits the liner sweep voltammetry (LSV) curves of CNTs, Pt NCs/CNTs-10% and Pt/C-20% at the rotation rate of 1600 rpm. Obviously, it can be seen that the OMC is inactive for ORR. The diffusion and diffusion-kinetic controlled regions are found for Pt NCs/CNTs-10% and Pt/C-20% in the potentials below 0.65 V and 0.65-0.95 V, respectively. It is worth mentioning that Pt NCs/CNTs-10% possess a higher onset potential (Eonset = 0.87 V) and half-wave potential (E1/2 = 0.83 V) than that of Pt/C-20% (Eonset = 0.86 V and E1/2 = 0.81 V), indicating the higher ORR activity of Pt NCs/CNTs-10%. Furthermore, the ORR diffusion limited current (JL) of Pt NCs/CNTs-10% is higher compared with that measured for Pt/C-20%. Figure 6b presents the tafel plots determined on the basis of the RDE data at 1600 rpm. The slope value for Pt NCs/CNTs-10% is 81 mV decade −1 , which is smaller than that of Pt/C-20% (87 mV decade −1 ), further indicating the more efficient ORR process on Pt NCs/CNTs-10%. The LSV curves of Pt NCs/CNTs-10% at different rotation speeds are further performed to clarify the ORR pathway. As displayed in Figure 7a, the LSV curves of Pt NCs/CNTs-10% exhibit rotatingspeed-dependent current densities. Potential in the diffusion and diffusion-kinetic regions, varying from 0.6 to 0.7 V, was chosen to calculate the values of electron transfer number (n) by Koutecky-Levich (K-L) Equation (4): The LSV curves of Pt NCs/CNTs-10% at different rotation speeds are further performed to clarify the ORR pathway. As displayed in Figure 7a, the LSV curves of Pt NCs/CNTs-10% exhibit rotating-speed-dependent current densities. Potential in the diffusion and diffusion-kinetic regions, varying from 0.6 to 0.7 V, was chosen to calculate the values of electron transfer number (n) by Koutecky-Levich (K-L) Equation (4): where J, J L , and J K is the measured current, diffusion limited current, and kinetic current, respectively. ω is the electrode rotating rate. B is determined from the slope of the K-L plots. Faraday constant (F) is 96485 C. Diffusion coefficient (D) is 1.9 × 10 -5 cm 2 s -1 . Kinematic viscosity of the electrolyte (ν) is 0.01 cm 2 s -1 and bulk concentration of O 2 (C) is 1.2 × 10 -6 mol cm −3 . As shown in Figure 7b, the corresponding K-L plots show excellent linearity, indicating the the first-order reaction kinetics. Meanwhile, the slopes of the K-L plots are almost same, suggesting a similar values of n. Based on the K-L slopes, the value of n can be determined to be 3.80 (0.60 V), 3.79 (0.65 V), and 3.81 (0.70 V), respectively. It suggests that Pt NCs/CNTs-10% proceeds dominantly with a preferential 4e − pathway, which is beneficial to improve the efficiency of fuel cells. where J, JL, and JK is the measured current, diffusion limited current, and kinetic current, respectively. ω is the electrode rotating rate. B is determined from the slope of the K-L plots. Faraday constant (F) is 96485 C. Diffusion coefficient (D) is 1.9 × 10 -5 cm 2 s -1 . Kinematic viscosity of the electrolyte (ν) is 0.01 cm 2 s -1 and bulk concentration of O2 (C) is 1.2 × 10 -6 mol cm −3 . As shown in Figure 7b, the corresponding K-L plots show excellent linearity, indicating the the first-order reaction kinetics. Meanwhile, the slopes of the K-L plots are almost same, suggesting a similar values of n. Based on the K-L slopes, the value of n can be determined to be 3.80 (0.60 V), 3.79 (0.65 V), and 3.81 (0.70 V), respectively. It suggests that Pt NCs/CNTs-10% proceeds dominantly with a preferential 4e − pathway, which is beneficial to improve the efficiency of fuel cells. The SA and MA are important parameters for investigating intrinsic activities of catalysts. Here, the SA and MA of Pt NCs/CNTs-10% and Pt/C-20% are calculated from JK and normalization with the Pt surface area and loading, respectively. The JK is calculated from the mass-transport correction of RDE using Equation (5): As shown in Figure 8, at 0.85 V, the SA over Pt NCs/CNTs-10% (0.482 mA cm -2 ) is 2.6 times larger than that of Pt/C-20% (0.183 mA cm -2 ) and the MA (358 mA mg -1 ) is 4.1 times higher than that of Pt/C-20% (87 mA mg -1 Pt). These values over Pt NCs/CNTs-10% at 0.9 V are calculated to be 0.194 mA cm -2 and 167 mA mg −1 respectively, which are 2.5 and 3.2 times higher than those of Pt/C-20% (0.079 mA cm -2 and 52 mA mg -1 ). The values of Pt NCs/CNTs-10% are not only superior to that of Pt/C-20%, but also that of most reported Pt-based catalysts as presented in Table 1. The SA and MA are important parameters for investigating intrinsic activities of catalysts. Here, the SA and MA of Pt NCs/CNTs-10% and Pt/C-20% are calculated from J K and normalization with the Pt surface area and loading, respectively. The J K is calculated from the mass-transport correction of RDE using Equation (5): As shown in Figure 8, at 0.85 V, the SA over Pt NCs/CNTs-10% (0.482 mA cm -2 ) is 2.6 times larger than that of Pt/C-20% (0.183 mA cm -2 ) and the MA (358 mA mg -1 ) is 4.1 times higher than that of Pt/C-20% (87 mA mg -1 Pt). These values over Pt NCs/CNTs-10% at 0.9 V are calculated to be 0.194 mA cm -2 and 167 mA mg −1 respectively, which are 2.5 and 3.2 times higher than those of Pt/C-20% (0.079 mA cm -2 and 52 mA mg -1 ). The values of Pt NCs/CNTs-10% are not only superior to that of Pt/C-20%, but also that of most reported Pt-based catalysts as presented in Table 1. where J, JL, and JK is the measured current, diffusion limited current, and kinetic current, respectively. ω is the electrode rotating rate. B is determined from the slope of the K-L plots. Faraday constant (F) is 96485 C. Diffusion coefficient (D) is 1.9 × 10 -5 cm 2 s -1 . Kinematic viscosity of the electrolyte (ν) is 0.01 cm 2 s -1 and bulk concentration of O2 (C) is 1.2 × 10 -6 mol cm −3 . As shown in Figure 7b, the corresponding K-L plots show excellent linearity, indicating the the first-order reaction kinetics. Meanwhile, the slopes of the K-L plots are almost same, suggesting a similar values of n. Based on the K-L slopes, the value of n can be determined to be 3.80 (0.60 V), 3.79 (0.65 V), and 3.81 (0.70 V), respectively. It suggests that Pt NCs/CNTs-10% proceeds dominantly with a preferential 4e − pathway, which is beneficial to improve the efficiency of fuel cells. The SA and MA are important parameters for investigating intrinsic activities of catalysts. Here, the SA and MA of Pt NCs/CNTs-10% and Pt/C-20% are calculated from JK and normalization with the Pt surface area and loading, respectively. The JK is calculated from the mass-transport correction of RDE using Equation (5): As shown in Figure 8, at 0.85 V, the SA over Pt NCs/CNTs-10% (0.482 mA cm -2 ) is 2.6 times larger than that of Pt/C-20% (0.183 mA cm -2 ) and the MA (358 mA mg -1 ) is 4.1 times higher than that of Pt/C-20% (87 mA mg -1 Pt). These values over Pt NCs/CNTs-10% at 0.9 V are calculated to be 0.194 mA cm -2 and 167 mA mg −1 respectively, which are 2.5 and 3.2 times higher than those of Pt/C-20% (0.079 mA cm -2 and 52 mA mg -1 ). The values of Pt NCs/CNTs-10% are not only superior to that of Pt/C-20%, but also that of most reported Pt-based catalysts as presented in Table 1.  The remarkable ORR activity of Pt NCs/CNTs-10% catalyst can be attributed to the three factors: First, the reduced coordination of Pt NCs as well as the decreased electrophilicity give rise to a corresponding reduction in the activation energy of the dissociative chemisorption of O 2 , facilitating the 4e − transfer process [36]. Second, the unprotected Pt NCs with smaller size and narrower size distribution can provide more active sites and promote the charge transfer during the ORR process. Third, CNTs with graphitic structure exhibit higher electron transfer rate [37]. All these together make the as-prepared Pt NCs/CNTs-10% exhibit more desirable ORR performance.

Electrochemical Stability and Methanol Tolerance
Besides the ORR activity, the stability of catalysts are also important for the practical application of fuel cells. Here, the stability of Pt NCs/CNTs-10% is performed by chronoamperometric measurement, which is an effective approach to investigate the stability of catalysts [38]. For comparison, Pt/C-20% is also tested as a baseline catalyst. As shown in Figure 9, after chronoamperometric measurement for 2 × 10 4 s, only~10% current decrease is observed on Pt NCs/CNTs-10%. However, the value for Pt/C-20% is~69%, much larger than that of Pt NCs/CNTs-10%, demonstrating a more satisfactory stability of Pt NCs/CNTs-10%. As reported in literatures, the poor stability of Pt/C-20% is mainly derived from the agglomeration of Pt particles [39]. Hence, TEM image of Pt NCs/CNTs-10% after the chronoamperometric measurement is conducted. As shown in Figure 10, no obvious aggregation is observed and the average particle size is determined to be~2.35 ± 0.41 nm. According to the aforementioned results, we can conclude that the excellent stability of Pt NCs/CNTs-10% may be attributed to the trapping the Pt NCs in the CNTs nano-network as well as the interactions between Pt NCs and CNTs, suppressing the migration, aggregation, and growth in size of the small Pt NCs.    The crossover of methanol from the anode to the cathode can lead to a reduction in cell voltage by ~200-300 mV. There is a competitive reaction between ORR and methanol oxidation on Pt-based cathodes. Therefore, it is highly desirable to develop ORR electrocatalysts with a high methanol tolerance for the application of fuel cells. Figure 11 displays the methanol tolerance of Pt NCs/CNTs-10% and Pt/C-20% evaluated by adding methanol into the electrolyte during the chronoamperometric measurement. It can be seen that the current of Pt/C-20% decreases to 63% after methanol injection. In contrary, only a slight oscillation in the current is observed for Pt NCs/CNTs-10%, suggesting a great methanol tolerance of Pt NCs/CNTs-10%. The methanol oxidation in 0.1 M N2-saturated HClO4 solution containing 0.5 M CH3OH is further performed to analyze the origin of the high methanol tolerance of Pt NCs/CNTs-10%. As depicted in Figure 11b, the methanol oxidation current densities on Pt NCs/CNTs-10% is 11.5 mA cm -2 , which is much smaller than that of Pt/C-20% (49.4 mA cm -2 ) implying that the methanol oxidation on Pt NCs/CNTs-10% is less active than that on Pt/C-20%, which can be attributed to the hindrance of chemisorption of methanol on the catalyst surface. Thus, the high methanol tolerance of the Pt NCs/CNTs-10% can be explained by the preferential adsorption of O2 over methanol on the Pt NCs site. The crossover of methanol from the anode to the cathode can lead to a reduction in cell voltage by~200-300 mV. There is a competitive reaction between ORR and methanol oxidation on Pt-based cathodes. Therefore, it is highly desirable to develop ORR electrocatalysts with a high methanol tolerance for the application of fuel cells. Figure 11 displays the methanol tolerance of Pt NCs/CNTs-10% and Pt/C-20% evaluated by adding methanol into the electrolyte during the chronoamperometric measurement. It can be seen that the current of Pt/C-20% decreases to 63% after methanol injection. In contrary, only a slight oscillation in the current is observed for Pt NCs/CNTs-10%, suggesting a great methanol tolerance of Pt NCs/CNTs-10%. The methanol oxidation in 0.1 M N 2 -saturated HClO 4 solution containing 0.5 M CH 3 OH is further performed to analyze the origin of the high methanol tolerance of Pt NCs/CNTs-10%. As depicted in Figure 11b, the methanol oxidation current densities on Pt NCs/CNTs-10% is 11.5 mA cm -2 , which is much smaller than that of Pt/C-20% (49.4 mA cm -2 ) implying that the methanol oxidation on Pt NCs/CNTs-10% is less active than that on Pt/C-20%, which can be attributed to the hindrance of chemisorption of methanol on the catalyst surface. Thus, the high methanol tolerance of the Pt NCs/CNTs-10% can be explained by the preferential adsorption of O 2 over methanol on the Pt NCs site.

Conclusions
In summary, highly dispersed Pt NCs (~2.2 nm) anchored on CNTs were successfully fabricated through a modified top-down strategy. The Pt NCs/CNTs-10% (Pt loading: 10 wt.%) exhibited superior catalytic activity, stability, and methanol tolerance compared with Pt/C-20%. The enhanced activity of Pt NCs/CNTs-10% is possibly ascribed to the reduced electrophilicity and the decreased coordination of the Pt NCs that lead to a decrease in the activation energy of O2, the smaller size and narrower size distribution of unprotected Pt NCs that provide more active sites, and the graphitic structure of CNTs that can improve the electron transfer rate. The Pt NCs/CNTs-10% would be a promising catalyst for fuel cells.

Conclusions
In summary, highly dispersed Pt NCs (~2.2 nm) anchored on CNTs were successfully fabricated through a modified top-down strategy. The Pt NCs/CNTs-10% (Pt loading: 10 wt.%) exhibited superior catalytic activity, stability, and methanol tolerance compared with Pt/C-20%. The enhanced activity of Pt NCs/CNTs-10% is possibly ascribed to the reduced electrophilicity and the decreased coordination of the Pt NCs that lead to a decrease in the activation energy of O 2 , the smaller size and narrower size distribution of unprotected Pt NCs that provide more active sites, and the graphitic structure of CNTs that can improve the electron transfer rate. The Pt NCs/CNTs-10% would be a promising catalyst for fuel cells.
Author Contributions: F.W. and T.X. conceived and designed the experiments; J.L. performed the major experiments; J.Y. and B.F. performed a part of the measurements; F.W. and J.L. analyzed the data and wrote the manuscript.