Alloyed Pt3M (M = Co, Ni) nanoparticles supported on S- and N-doped carbon nanotubes for the oxygen reduction reaction

  1. 1,2 ,
  2. 2 ORCID Logo ,
  3. 3 ORCID Logo ,
  4. 2 ORCID Logo ,
  5. 1 ,
  6. 2 and
  7. 1 ORCID Logo
1LCC-CNRS, Université de Toulouse, CNRS, INPT, Toulouse, France
2Université Grenoble Alpes, CEA-LITEN/DEHT, 17 rue des martyrs 38000 Grenoble, France
3CEMES-CNRS, 29 rue Jeanne Marvig, 31055 Toulouse Cedex 4, France
  1. Corresponding author email
Guest Editor: B. Etzold
Beilstein J. Nanotechnol. 2019, 10, 1251–1269. https://doi.org/10.3762/bjnano.10.125
Received 01 Feb 2019, Accepted 21 May 2019, Published 21 Jun 2019
Full Research Paper
cc by logo

Abstract

Sulfur- (S-CNT) and nitrogen-doped (N-CNT) carbon nanotubes have been produced by catalytic chemical vapor deposition (c-CVD) and were subject to an annealing treatment. These CNTs were used as supports for small (≈2 nm) Pt3M (M = Co or Ni) alloyed nanoparticles that have a very homogeneous size distribution (in spite of the high metal loading of ≈40 wt % Pt), using an ionic liquid as a stabilizer. The electrochemical surface area, the activity for the oxygen reduction reaction and the amount of H2O2 generated during the oxygen reduction reaction (ORR) have been evaluated in a rotating ring disk electrode experiment. The Pt3M/N-CNT catalysts revealed excellent electrochemical properties compared to a commercial Pt3Co/Vulcan XC-72 catalyst. The nature of the carbon support plays a key role in determining the properties of the metal nanoparticles, on the preparation of the catalytic layer, and on the electrocatalytic performance in the ORR. On N-CNT supports, the specific activity followed the expected order Pt3Co > Pt3Ni, whereas on the annealed N-CNT support, the order was reversed.

Introduction

Proton exchange membrane fuel cells (PEMFCs) convert chemical energy from the hydrogen oxidation reaction (HOR) and the oxygen reduction reaction (ORR) into electrical energy. PEMFCs are one of the most promising technologies in the field of renewable energy (and especially for transport applications), but the cost and lifetime are factors still to be improved in order to achieve widespread dissemination of this technology [1,2]. Due to the sluggish reaction kinetics for the ORR, the cathode active layer contains generally four times more catalyst than the anodic layer, which explains why most of the research is focused on the optimization of the cathodic catalytic layer. Platinum nanoparticles (NPs) supported on carbon black (CB), especially Vulcan XC-72 [3,4], are usually used as the catalyst. To meet the performance and durability requirements for transport applications, a metal loading of 0.4 mgPt·cm−2 for the cathode side is commonly used, which explains the high cost of these systems [5]. One lever to reduce the cost of this technology is the reduction of the cathode Pt loading, but this must be done without compromise to the catalyst layer performance and durability.

It is known that catalyst degradation via platinum dissolution and carbon corrosion plays an important role in the voltage degradation of PEMFCs [6-8]. CB, which is widely used, particularly for its low cost, suffers from thermochemical instability and corrosion in fuel cell applications. In the cathodic layer, the oxidizing, wet and acidic environment, the high electrochemical potential, and the high platinum loading all lead to the oxidation of the carbon surface, and occasionally to the formation of CO2 [9]. This carbon corrosion modifies the mass transport properties of the active layer, especially for the water management, and accelerates the degradation of the Pt NPs [10,11].

One way to reduce the Pt content is to use more active, tailored NPs [12,13], for example, bimetallic NPs with a core–shell structure [14,15]: a Pt shell can be deposited on a low-cost transition metal such as Co [16-18], Ni [19,20] or Cu [21] or their nitrides [22]. Kristian et al. have described a redox–transmetalation method for the synthesis of Cocore–Ptshell particles with a high activity for the ORR [23]. Platinum-based alloys can also been used [24,25]. Therefore, it seems important to develop nanostructured catalysts supported on a material with electronic conductivity and surface area close to the common CB but with more resistance towards corrosion. Interestingly, it has recently been shown that the introduction of small amounts of ionic liquids (ILs), which are known as NP stabilizers, including on carbon supports [26], into Pt-based catalysts can further improve the ORR performance. This is likely due to the high O2 solubility in the IL phase [27]. It was also demonstrated that the choice of the carbon support, in combination with ILs, is also important to achieve high Pt dispersion, and functionalized carbons should be preferred, presumably because of their stronger interaction with the IL [28].

Carbon nanotubes (CNTs) are well known for their remarkable chemical and physical properties and appear to be an interesting alternative to replace CB in fuel cell applications [29,30]. It has been described that CNTs could be used as resistant material to support nanostructured PtNi hollow particles, but it appears that the structure of the used CNT might be responsible for the large external diameter of the deposited particles, which is close to 25 nm [31]. Previous works have shown the possibility to dope CNTs with nitrogen (N-CNT) or sulfur (S-CNT) in order to modify properties such as electronic conduction and surface chemistry [32-34]. This strategy contributes to improve the metal dispersion and to increase the performance of the catalyst for the ORR due to the structural and electronic properties of the doped CNT [34,35]. Additionally, the amount of heteroatom in the doped structure has an effect on the hydrophobicity of the material, which could provide a solution to facilitate the water management in the active layer. In fact, water management in the cathodic layer is a key challenge: the electrolytic membrane has to be hydrated enough to favor the proton conduction, but an excess of water in the cathodic layer will decrease the oxygen accessibility to the active sites. Thermal treatment can be used to improve the carbon corrosion resistance of the CNT. In fact, annealing at high temperature (above 1000 °C) is used to remove structural defects from the CNT in order to obtained more stable [36] and more conductive [37] structures. In the case of the N-CNTs, the thermal annealing can also modify the ratio of the different nitrogen groups at the surface [38], and consequently, the metal–support interaction.

This work proposes: i) to reduce the amount of Pt on the catalyst by using Pt alloyed compounds, and ii) to increase the support corrosion resistance using heat-treated carbon materials. Several CNT materials have been synthetized, fully characterized (structural and surface properties) and used as catalyst supports. CNT-supported bimetallic Pt3Co or Pt3Ni NPs were synthetized. To evaluate catalyst performance, electrochemical characterization was performed using a rotating ring disk electrode (RRDE) experiment to determine the active surface and the activity for the ORR of each catalyst. Finally, membrane electrode assemblies (MEAs) of 25 cm2 active area, integrating the synthesized catalysts, have been prepared and tested. The one giving the best beginning-of-life performance has been aged following a recommended accelerated stress test (AST) cycle for catalyst support corrosion.

Results and Discussion

Synthesis and characterization of the CNTs and Pt3M/CNT

Three kinds of CNTs have been produced by catalytic chemical vapor deposition (c-CVD): undoped (CNTs), N-doped (N-CNTs) and S-doped CNT (S-CNTs). To further increase the corrosion resistance and the electrical conductivity of the N-CNTs, they were annealed at 1000 °C to produce N-CNTHT.

The introduction of nitrogen or sulfur into the CNT structure has an effect on the structural properties of the prepared materials. High resolution transmission electron microscopy (HRTEM) analysis shows a remarkable difference between the carbon structures synthetized (Figure 1).

[2190-4286-10-125-1]

Figure 1: HRTEM micrographs of: a) CNTs; b) N-CNTs; c) S-CNTs, and d) N-CNTsHT. (Scale bar = 10 nm).

Very regular structures were obtained for the CNT sample (Figure 1a), while the N-CNT sample presented a “bamboo-like” structure typically found in N-doped CNTs (Figure 1b) [39], and the structure of the S-doped CNTs presents some alterations (bulbous segments, Figure 1c), which are different than those observed for the N-CNTs (bamboo structure). The N-CNTsHT show similar structure to the HRTEM observations (Figure 1d). Low magnification TEM micrographs of the carbon supports are given in Supporting Information File 1, Figure S1. The CNTs, N-CNTs and S-CNTs have mean external diameters of 15 ± 5, 18 ± 8, and 15 ± 7 nm, respectively. The specific surface area (SSA) of these materials ranged between 150 m2/g (CNTs) and 190 m2/g (S-CNTs).

Structural characterization was performed using Raman spectroscopy and X-ray diffraction (XRD, see Table 1 and Supporting Information File 1, Figure S2 for Raman spectra). In Raman spectroscopy, a useful parameter for carbon nanotubes is the ratio between the D band (ID) at ≈1380 cm−1, attributed to the defects of the CNT structure, and the G band (IG) at ≈1580 cm−1, the first-order Raman band of all sp2 carbon materials. The presence of disorder in CNTs can also impact: i) the intensity of other bands, such as the G’ band at ≈2700 cm−1, and ii) the position and shape of the peaks [40]. The G’ band is indicative of long-range order in a sample. Finally, another parameter, measurable by Raman spectroscopy that is relevant to catalyst preparation, is the LD: LD is a typical inter-defect distance that we have measured as described in [41]. A lower ID/IG (and higher LD) is obtained for the CNT sample and a higher ID/IG (and lower LD) for the N-CNT sample, which is in accordance with the TEM observations. The N-CNTHT sample shows a decrease in the number of defects compared to N-CNT, as expected after the high temperature treatment. S-CNTs constitute an intermediate situation. The d002 inter-planar spacing results obtained with XRD are presented in Table 1.

Table 1: Chemical, textural, and crystalline properties of the carbon supports.

Supports XPS analysis Textural properties Elemental analysis
  C (%) O (%) N (%) S (%) BET surface area (m2·g−1) Pore volume (cm3·g−1) C (%) N (%) S (%)
CNT 98.4 1.6 151 2.6 92.3
N-CNT 91.2 4.9 3.9 0.2 182 2.7 92.3 2.9
N-CNTHT 94.1 3.5 2.4 168 2.4 95.6 1.6
S-CNT 95.6 2.8 1.3 190 1.1 78.6 5.0
  Raman analysis Crystallite properties      
  ID/IG ID/IG ID/IG ID/IG d002 (nm) d002 (nm)      
CNT 0.87 0.87 0.87 0.87 0.3444 0.3444      
N-CNT 1.02 1.02 1.02 1.02 0.3436 0.3436      
N-CNTHT 0.95 0.95 0.95 0.95 0.3379 0.3379      
S-CNT 0.99 0.99 0.99 0.99 0.3422 0.3422      

All the values are larger than that of graphite (3.334 Å), and the smallest value is obtained for N-CNTHT, indicating a higher level of graphitization for this sample.

The elemental composition as well as the surface chemistry is also affected by heteroatom doping. Elemental and X-ray photoelectron spectroscopy (XPS) analysis results are shown in Table 1. The elemental analysis confirmed the efficiency of the doping, and showed that S-CNT contains a significant amount of residual catalyst (iron, encapsulated in the structure of the tubes). An effect of heat treatment on the N-CNTs (besides the reduction of disorder) is to decrease the amount of nitrogen, which decreases from 2.9 to 1.6%. XPS analysis confirmed the bulk analyses and showed that the S-CNT sample also contains a significant amount of surface oxygen groups. The introduction of oxygen may correspond to the oxidation of the sulfur species introduced during the doping during the purification step in H2SO4. As the S 2p peaks are typically presented in spin–orbit doublets of S 2p3/2 and S 2p1/2 (splitting magnitude ≈1.18 eV), four S 2p3/2 peaks representing sulfur bonding of FeS2 (≈162.5 eV), H–S–C (≈163.5 eV), R–S–C (≈164.5 eV), and S–O (≈168.0 eV) were observed (Supporting Information File 1, Figure S3) [42,43]. The intense peak at 163.5 eV indicated the doping of CNTs with mainly thiol surface groups (Table 2). The peak at 164.5 eV could arise from the presence of sulfur in the carbon matrix, while the oxidation of surface thiols should produce S–O bonds (peak at 168 eV). The presence of pyrite could arise from the significant amount of remaining iron catalyst in this sample. Different nitrogen groups are present in N-CNT, and the proportion of these groups was determined by deconvolution of the main N 1s peak: pyridinic nitrogen, pyrrolic nitrogen, quaternary nitrogen and nitrogen-oxidized species were identified (Table 2, and Supporting Information File 1, Figure S4). XPS analysis also showed the presence of sulfonic acid groups on the surface of N-doped CNTs (Supporting Information File 1, Figure S4b), and the S content increases with the nitrogen content. This functionalization occurs during the purification step with sulfuric acid. In the N-CNTHT sample, we measured a clear decrease of the nitrogen-oxidized species.

Table 2: Contribution of species detected by deconvolution of XPS spectra for several N-CNT samples.

Sample Surface groups
S-doped SFeS2 (%) Sthiol (%) Ssulphide/thioether (%) Sox (%)
S-CNT 5.0 64.9 22.8 7.3
N-doped Npyridinic (%) Nquaternary (%) Npyrrolic (%) Noxidized (%)
N-CNT 23.9 23.4 9.1 43.6
N-CNTHT 30.9 54.3 10.3 4.5

Electrocatalyst support materials are crucial to both the performance and durability of PEM fuel cells [44,45]. These materials should combine some key characteristics such as: i) an adapted surface chemistry to allow high dispersion of the metallic phase at very high metal loading, ii) a good balance between hydrophobicity and hydrophilicity to allow water management and interaction with the electrolyte, iii) a good dispersibility in the ink to limit mass transfer, and iv) structural features allowing high conductivity and chemical stability. As some of these characteristics are not compatible (e.g., a high metal dispersion should be favored on defective carbon supports but this should be detrimental to the stability and electronic conduction), some compromises have to be made.

Figure 2 shows the evolution of the ID/IG ratio from Raman spectroscopy and the percent of surface heteroatoms from XPS in the investigated supports. The ID/IG ratio reflects the concentration of defects in these supports, where a high ratio favors metal dispersion. A high percentage of surface heteroatoms should favor metal dispersion and interaction with the electrolyte but may have negative impacts on the electronic conductivity, the stability, and can modify the metal/support interactions and thus the electrocatalytic performance. From this figure, we can see that the CNTs and S-CNTs constitute the extremes and that the compromise could lie in the N-doped nanotubes, particularly the N-CNTHT.

[2190-4286-10-125-2]

Figure 2: Evolution of the ID/IG ratio (from Raman spectroscopy) and the percent of surface heteroatoms (from XPS) in the investigated CNT supports.

These CNTs have been used to support Pt3M (M = Co, Ni) NPs. The bimetallic NPs were prepared using the transmetalation method described by Kristian et al. [23]. The redox couple used for the transmetalation is Co2+(II)/Co(0) (E° = −0.77 V/RHE) and PtCl42−/Pt(0) (E° = 0.67 V/RHE) for Pt3Co NPs. In our procedure, the ammonium salt used as a stabilizer has been replaced by an ionic liquid in order to obtain a better distribution of NPs at the surface of the CNTs [46]. Indeed, the use of hexadecyl trimetyl ammonium bromide leads to the formation of small NPs (1.90 ± 0.77 nm) that are not well-dispersed on the support (see Supporting Information File 1, Figure S5a for the Pt3Co/CNT sample). In order to improve the NP distribution on the support, [bmim][Tf2N] was chosen as stabilizer. It is known that imidazolium salts show good interaction with CNTs due to π–π interactions and could also be used as a stabilization agent for NPs [46-48]. The TEM images of sample Pt3Co/CNT (Supporting Information File 1, Figure S5b) show that the use of ILs does not result in well-dispersed NPs on this support. The presence of heteroatoms in the CNTs significantly affects the distribution of the NPs on the CNT surface (Figure 3a–c). N-doped or S-doped structures are known to be more reactive due to the presence of different functional groups at the surface of the CNTs, while un-doped CNTs are relatively inert. With nitrogen or sulfur groups acting as nucleation centers, the distribution of NPs on the CNT walls is better on N-CNTs and S-CNTs than on CNTs. A decrease of the nitrogen content was observed with annealing of the N-CNTs. This could explain why the NP distribution is better for Pt3Co/N-CNT compared to Pt3Co/N-CNTHT. Non-annealed CNTs seem to have better interaction with the Co precursor and the as-obtained catalysts show a better NP distribution. Nevertheless, the mean size of the NPs is around 2 nm for all the prepared catalysts (Table 3 and Supporting Information File 1, Figure S6).

[2190-4286-10-125-3]

Figure 3: TEM micrographs of: a) Pt3Co/N-CNT; b) Pt3Co/N-CNTHT; c) Pt3Co/S-CNT; d) Pt3Ni/N-CNT; e) Pt3Ni/N-CNTHT, and f) Pt3Ni/S-CNT. Scale bar = 100 nm.

Table 3: Mean nanoparticle size and metal loading for PtM/CNT catalysts.

Sample Particle size (nm) M (wt %) Pt (wt %)
Pt3Co/CNT 2.28 ± 0.82 12.5 33.5
Pt3Co/N-CNT 1.71 ± 0.73 15.4 41.5
Pt3Co/N-CNTHT 2.08 ± 0.90 7.9 40.2
Pt3Co/S-CNT 2.00 ± 0.71 15.4 33.0
Pt3Ni/N-CNT 2.42 ± 1.24 15.3 40.3
Pt3Ni/N-CNTHT 1.87 ± 0.86 18.3 42.3
Pt3Ni/S-CNT 2.30 ± 0.80 14.8 23.3

Pt3Ni/N-CNT catalysts were prepared using the same procedure. The redox couple used for the transmetalation is Ni2+(II)/Ni(0) (E° = −0.257 V/RHE) and PtCl42−/Pt(0) (E° = 0.67 V/RHE). A poor metal distribution was obtained on the CNTs, thus Pt3Ni NPs were further studied only on N-CNTs and S-CNTs. For the Pt3Ni catalysts, the mean size of the NPs is around 2 nm as observed for Pt3Co (Figure 3d–f and Supporting Information File 1, Figure S6). From these syntheses, we can estimate the influence of the metal precursor on the NP distribution. Regardless of the carbon support, much better distribution on the CNT walls was obtained with the Ni precursor. The affinity between the metal precursor and the carbon surface seems to play a role on the active phase distribution. The higher affinity of Ni compared to Co for graphene [49,50] and N-doped carbon [51,52] surfaces has already been reported in the literature and could hint at the origin of our results. The annealing of the N-CNTs has also an impact, since the Pt3Ni/N-CNTHT catalyst shows a better distribution than the Pt3Ni/N-CNT one. We suspect that the annealing of the N-CNTs and the as-obtained modifications of the chemical surface causes an important change in the adsorption and diffusion of the metal. It has been shown that the binding energy between a transition metal and a carbon support depends on the nature of the metal but also on the carbon material composition [53].

Particularly, Kattel showed that the binding energy of several transition metals could change if they are bound to two carbon atoms (M–C2), one carbon atom and one nitrogen atom (M–CN) or two nitrogen atoms (M–N2) [52]. Good metal distribution was also obtained on the S-CNT sample; in this case, however, the Pt loading is rather low (≈23%) compared to the N-CNTs. A recent study has shown that good metal distribution can be obtained on S-CNTs in the case of Pt catalysts (Pt loading = 20%, Pt NP size = 2.4 nm) [32,33]. The mean particle size for the Pt3Ni series is also around 2 nm, highlighting the effectiveness of the synthetic procedure followed in this work.

Moreover, these binding energies are also dependent on the nature of the metal. These observations could explain the strong differences between Pt3Co and Pt3Ni catalysts. In both cases, the theoretical Pt loading (50 wt %) is never reached, while the Co (or Ni) loading is higher than expected (theoretical content ≈7 wt %), strongly suggesting the presence of residual Co (or Ni) in these catalysts.

RRDE measurements of electrochemical surface area (ECSA) and ORR activity and selectivity

In order to proceed to a first screening of the as-prepared catalysts, their electrochemical properties were evaluated by RRDE measurements. It is worth noting that electrocatalyst investigations are usually performed with a rotating ring-disk electrode (RRDE) in acidic or alkaline media. Previous works have shown that the ORR activities of Pt catalysts are strongly dependent on the electrolyte [54]. According to these studies, activities were found to increase from H2SO4 to HClO4 due to the specific effect of the adsorbed anion on different Pt(hkl) sites. Furthermore, the thin film RRDE method, with a low Nafion amount, is recommended to avoid diffusion resistance into the deposited active layer [55]. Therefore, low catalyst loadings are known to give higher activities, but these conditions are not always representative of the true working of the PEMFC [56]. Furthermore, the difficulty of dispersing CNTs in a highly diluted ink without using a dispersing agent is a real issue, which is why we have preferred to use a higher catalyst loading (100 µgPt·cm−2).

For Pt3Ni and Pt3Co catalysts, the typical specific activities (expressed as kinetic current densities) for ORR at 25 °C are in the range 2.5–4.5 mA·cm−2 in HClO4 [57-59] and around 0.3–0.7 mA·cm−2 in H2SO4 [59-61]. Interestingly, it was also reported that for Pt3Ni and Pt3Co catalysts in H2SO4, the activity increases in the order Pt3Ni > Pt3Co, and in HClO4 the order of activity at 25 °C was PtCo > PtNi [59]. However, these tendencies can also be particle size dependent. Thus, it was demonstrated that in H2SO4, this order (Pt3Ni > Pt3Co) is respected for particles >6 nm, whereas the opposite order prevails for particle sizes smaller than 6 nm [60]. In our work, the electrochemical properties were evaluated in H2SO4 and compared with a commercial Pt3Co/Vulcan XC-72 (Pt3Co/CB) catalyst. This commercial reference has shown excellent activity in RRDE and during single cell tests [5]. Figure 4a,b shows the cyclic voltammetry (CV) curves obtained in N2-purged 0.5 M H2SO4 for Pt3Co/N-CNT, Pt3Co/N-CNTHT, Pt3Co/S-CNT, Pt3Ni/N-CNT, Pt3Ni/N-CNTHT, Pt3Ni/S-CNT and the commercial Pt3Co/CB samples. For all the catalysts characterized, two small peaks at 0.14 and 0.24 V/RHE were observed, corresponding to the adsorption of H+ on the (110) and the (100) crystallographic faces of platinum, respectively [62]. Figure 4c,d represents the current–potential curves obtained by RRDE experiments in O2-saturated electrolyte at 900 rpm. These voltammograms are used to measure the ORR specific current density at 0.9 V/RHE. All the RRDE results are shown in Table 4.

[2190-4286-10-125-4]

Figure 4: CVs from an RRDE experiment for the catalysts in 0.5 M H2SO4 at 5 mV·s−1 under N2 at 25 °C for a) the Pt3Co/CB and Pt3Co/CNT series and b) the Pt3Co/CB and Pt3Ni/CNT series. Background of linear sweep voltammetry (LSV) with RRDE experiment for the catalysts in 0.5 M H2SO4 at 5 mV·s−1 at 900 rpm under O2 at 25 °C for the c) Pt3Co/CNT and d) Pt3Ni/CNT series.

Table 4: Electrochemical surface area (ECSA), current density and iK at 0.9 V/RHE for Pt3M catalysts using RRDE experiments.

Sample Pt loading (µgPt cm−2) ECSA (m2·gPt−1) Current density at 0.9 V/RHE (mA·cm−2) Kinetic current at 0.9 V/RHE (mA·cm−2) Specific activity at 0.9 V/RHE (A/m2Pt) Mass activity at 0.9 V/RHE (A·gPt−1)
Pt3Co/CB 100 22.3 1.22 1.96 0.88 19.6
Pt3Co/N-CNT 106 55.5 2.12 5.33 0.91 50.3
Pt3Co/N-CNTHT 100 34.9 1.29 1.97 0.56 19.7
Pt3Co/S-CNT 99 23.1 1.30 2.47 1.08 25.0
Pt3Ni/N-CNT 106 34.7 1.85 3.49 0.95 32.9
Pt3Ni/N-CNTHT 106 46.4 1.66 3.09 0.63 29.1
Pt3Ni/S-CNT 94 5.6 0.58 0.74 1.41 7.9

The electrochemical surface area (ECSA) of the Pt3Co/N-CNT and Pt3Ni/N-CNT catalysts were higher than that of Pt3Co/CB (NP size ≈5 nm), probably due to the smaller size of the NPs (NP size ≈2 nm). These values (23–55 m2·g−1) are consistent with values reported in the literature for PtCo and PtNi catalysts on CNT [31,63-66], but showing larger particle size. These values are also lower than that obtained for Pt3Co particles with a slightly larger diameter (3.2–4.2 nm) (but on other carbon supports) where the ECSA values between 50 and 74 m2·g−1 were reported [67-70]. We think that remaining traces of ionic liquid adsorbed on the NP surface could be at the origin of this phenomenon. Indeed, it is not uncommon to observe an ECSA loss for Pt/C catalysts after IL modification [27].

Furthermore, we recently showed that when CNTs are used as catalyst supports, they show a tendency to form aggregates, making the accessibility of the electrolyte difficult, which could lead to a Pt utilization of 40%. For Pt3Co/S-CNT catalysts (and particularly the Pt3Ni/S-CNT catalyst), a much lower ECSA was obtained.

The specific activity has been calculated for each catalyst at 0.9 V/RHE. It appears that this value is mainly driven by the ECSA. The catalyst with the lower ESCA presents a higher specific activity and the catalyst with the higher ECSA has the lowest specific activity. This effect has been previously reported in the literature [71]. Nevertheless, we need to mention that the relatively high thickness of our prepared electrodes enhances the O2 diffusion resistance, which might enhance the surface activity for a material with lower ECSA.

This highlights the influence of the carbon support on the performance obtained in RRDE measurement. With 25.0 A·gPt−1, the activity measured for PtCo/S-CNT is slightly higher than that of the commercial reference. The value obtained for PtNi/S-CNT is significantly lower (7.9 A·gPt−1). A possible explanation could be linked to the Pt3Ni NP–support interaction. As Pt3Ni catalysts on S-doped carbon materials have already shown good performance for the ORR [72,73], we suspect that the explanation arises from an excessive amount of carbon support during the measurement. Indeed, the measurements were made with an equivalent Pt loading on the electrode; however, the Pt loading in PtNi/S-CNT is rather low, which should have led to a thicker active layer during RRDE measurements. This could explain the low activity of this catalyst.

The two highest ECSAs were measured for Pt3Co/N-CNT and Pt3Ni/N-CNTHT at 55.5 m2·gPt−1 and 46 m2·gPt−1, respectively. For the RRDE characterization, a thin film of the catalyst was formed at the surface of the working electrode. For each carbon support used in this study, there was a modification of the properties of the catalytic layer and certainly a strong change of the mass transfer limitation. The proton transport should be better at the surface of the N-CNT due to the nitrogen doping and the hydrophilic behavior of these carbon supports [74]. For both metals, the catalysts with the best NP distribution on the CNT walls presented the higher ECSA. In these cases, we can assume that the interactions between Nafion®, NPs and N-CNTs are optimum to allow for high activity for the ORR. It should be noted that all the catalysts synthesized on N-doped CNTs outperformed the commercial Pt3Co/CB (22.3 m2·gPt−1 and 19.6 A·gPt−1) material. The highest mass densities were calculated for Pt3Co/N-CNT, Pt3Ni/N-CNT and Pt3Ni/N-CNTHT, with 50.3 A·gPt−1, 32.9 A·gPt−1 and 29.1 A·gPt−1, respectively. Different hypotheses could explain the excellent results for these two catalysts. First, the well-distributed NPs are the result of the good interaction between the transition metal and the nitrogen-doped CNT. It is known that high binding energies between the transition metal and the carbon support modify the electronic properties of the NPs and can facilitate the adsorption of the O2. However, XPS data (vide infra) do not support this hypothesis, since the binding energy of Pt 4f electrons is consistent with Pt(0) in these two catalysts. Another hypothesis could be the particle proximity effect. Speder et al. show the influence of the distance of the neighboring particles on the ORR activity [75]. In fact, catalysts with small inter-particles distance with no agglomeration display excellent activity. Computational investigations have shown that decreasing the inter-particle distances causes an overlap of the electrochemical double layer. This overlap could reduce the oxide coverage of the NPs and thus increase the activity of the catalyst. Moreover, the synergetic effect of a Pt3CO catalyst supported on Co containing N-doped carbon material has recently been demonstrated [76].

It is worth noting that the effect of support annealing is not the same for the PtCo and PtNi catalysts. On the non-heat-treated support N-CNT the activity order follows that expected from the literature for 2 nm NPs in H2SO4 electrolyte: Pt3Co > Pt3Ni [58]. However, this order is reversed for the N-CNTHT support since the specific activity (SA) is 1.66 and 1.29 mA·cm−2 for Pt3Ni and Pt3Co catalysts, respectively. On Pt3Co catalysts, the support annealing induces a pronounced decrease of both the ECSA and SA, whereas on Pt3Ni catalysts, the annealing induces an increase of the ECSA and a slight decrease of the SA. It is thus demonstrated that the specific activity of Pt3Ni and Pt3Co electrocatalysts for the ORR is electrolyte- and particle-size-dependent, as is already known, but also support-dependent, at least for small (2 nm) nanoparticles. For such smaller NPs, it is plausible that a modification of the d-band center occurs upon modification of the support [77].

The ORR selectivity is also a critical issue for a PEMFC catalyst. A well-known phenomenon of catalyst degradation is due to the formation of hydrogen peroxide near the electrolyte membrane [78]. Moreover, it was shown that in aqueous KNO3 solutions, nitrogen-doped carbon structures were active for ORR with lower H2O2 selectivity than Pt/C [79]. During the CV measurements in O2-saturated electrolyte, the amount of hydrogen peroxide produced during the ORR was monitored to compare its production for each catalyst. Figure 5 presents the hydrogen peroxide yield during the ORR for all the catalysts shown before.

[2190-4286-10-125-5]

Figure 5: Percentage of H2O2 formation during O2 reduction of the a) Pt3Co and b) Pt3Ni catalysts.

First, between 1 and 0.7 V/RHE the fractions of H2O2 produced are below 1% for all catalysts. In this voltage region, the ORR occurs only through the four electron process. The H2O2 production is significantly higher at low potential, between 0.05 and 0.4 V/RHE, because of the oxygen reduction on carbon [80]. The amount of H2O2 produced by the reference Pt3Co/CB is higher than for all the Pt3Co samples and most of the Pt3Ni catalysts prepared in this study. Thus, the H2O2 production is not an obstacle for using these catalysts for PEMFC applications.

Structural characterization of the best catalysts

For a better understanding of the structure of these catalysts, additional characterization was carried out on the Pt3Co/N-CNT and Pt3Ni/N-CNTHT samples, which presented the best performance in the ORR. We first used HRTEM to analyze the product resulting from the first step of the catalyst preparation, i.e., the reduction of the cobalt and nickel salts. The HRTEM images of Co/N-CNT samples are depicted in Figure 6a,b.

[2190-4286-10-125-6]

Figure 6: HRTEM images of a) Co/N-CNT and c) Ni/N-CNTHT; and STEM-HAADF images of b) Co/N-CNT and d) Ni/N-CNTHT.

First, no residual IL can be observed. Surprisingly, despite a high cobalt loading (52% w/w), Co NPs were also not observed on the surface of the N-CNTs. Instead, Co atoms (Figure 6a,b) and a few non-crystallized Co islands (Supporting Information File 1, Figure S7a,b) were identified using the scanning transmission electron microscopy/high-angle annular dark-field imaging (STEM-HAADF) technique. Energy-dispersive X-ray spectroscopy (EDX) analysis on the N-CNT surface or in the Co aggregates confirmed the presence of Co. The same analysis was made on the sample Ni/N-CNTHT. There also, despite the high Ni loading (48% w/w), no Ni NPs were observed (Figure 6c,d). STEM-HAADF analysis shed light into the presence of non-crystallized Ni at the surface of the carbon support. EDX analysis confirmed the presence of Ni at the surface of the N-CNTs.

HRTEM images of the bimetallic catalysts Pt3Co/N-CNT and Pt3Ni/N-CNTHT do not evidence the formation of core–shell NPs (Figure 7). The interplanar distance of 0.22 nm was found for Pt3Co/N-CNT NPs and Pt3Ni/N-CNTHT NPs (Figure 7), which is slightly smaller than the common distance found for d111 in Pt (0.23 nm) [81]. The slight contraction of the crystalline structure is probably due to the presence of Co (or Ni) atoms in the Pt structure. EDX spectra (Supporting Information File 1, Figure S8) reveal the presence of Co and Pt in the individual NPs (Figure 7b, 001 selected area) displaying a Pt3Co composition. The absence of Co NPs in the Co/N-CNT sample, and the composition and structure of the Pt3Co/N-CNTs, indicate that the Pt3Co/N-CNT is more likely an alloy than a core–shell structure. The presence of residual cobalt atoms or clusters on the CNT surface was also evidenced by STEM-HAADF images of Pt3Co/N-CNT (Figure 7b, 001 selected area). In the same way, the Pt3Ni composition was determine by EDX analysis for the sample Pt3Ni/N-CNTHT (Supporting Information File 1, Figure S9). This sample also displays some residual nickel atoms or clusters.

[2190-4286-10-125-7]

Figure 7: HRTEM images of a) Pt3Co/N-CNT and c) Pt3Ni/N-CNTHT, and STEM-HAADF images of b) Pt3Co/N-CNT and d) Pt3Ni/N-CNTHT.

Interestingly, Co single atoms on nitrogen-doped carbon have been reported to be active for the ORR in acidic media [82-85], and Ni single atoms on nitrogen-doped carbon are known to be active for some electro-reduction reactions [86,87]. Thus, the involvement of these species in ORR cannot be discarded. Wide angle X-ray scattering (WAXS) analysis was performed on Pt3Co/N-CNT and Pt3Ni/N-CNTHT (Figure 8). After corrections and taking a Fourier transform of the scattering data, the related pair distribution function (PDF) is well defined, with a low structural disorder. For Pt3Co/N-CNT, the coherence length is close to 2.3 nm, which gives a measurement of the average size of crystalline domains.

[2190-4286-10-125-8]

Figure 8: WAXS analysis – red: experimental PDF from a) Pt3Co/N-CNT and b) Pt3Ni/N-CNTHT, green: simulation from a spherical pure Pt model.

To accurately evaluate the actual cell parameter, thus the average composition, a simulation was performed from a model derived from pure Pt (spherical NPs, 2 nm in diameter).

To obtain a good agreement with the experimental data, a contraction factor of 1.4% was required, leading to a metallic bond length of 0.2721 nm, significantly shorter than in pure Pt (0.2760 nm – JCPDS 04–0802) but very close to the value in Pt3Co also crystallized in the Fm-3m system (0.2725 nm – JCPDS 29-0499) [88]. These results indeed point to an alloyed structure since in similar studies of core–shell NPs, the bond length obtained from the PDF was clearly related to the nature of the compact core [89]. The relatively poor agreement in amplitude for longer distances indicates that most crystalline domains are close to 2 nm but that some size and/or shape dispersion compared to the model is likely. The same conclusions can be made for the sample Pt3Ni/N-CNTHT. To obtain a good correlation with experimental data, a contraction factor of 2% was required. Concerning the NP size, most crystalline domains are close to 2 nm, which is consistent with the TEM result (1.86 nm).

XPS analysis was performed on the Pt3M/N-CNT catalysts in order to study the electronic structure of Pt, Co and Ni (Table 5 and below in Table 6).

Table 5: Binding energies and ratio of Pt and Co species obtained from XPS spectra.

Pt3Co/N-CNT Species Binding energy (eV) Ratio (%)
Pt 4f7/2 Pt 4f5/2
Pt Pt(0) 71.1 74.3 64.3
  PtO 72.1 75.2 12.8
  Pt(OH)2 73.1 76.6 22.9
    Co 2p3/2 Co 2p1/2  
Co Co(0) 778.0 792.0 3.9
  CoO 779.8 785.0 60.4
  Co(OH)2 781.4 796.9 35.7

The Pt 4f region of the XPS spectrum is shown in Supporting Information File 1, Figure S10a. The spectrum was deconvoluted in three pairs of doublets (corresponding to Pt 4f7/2 and Pt 4f5/2). The first doublet at 71.08/74.3 eV corresponds to metallic Pt(0) [90]. The Pt surface is mainly in the metallic state (64.3 wt %). The doublets at 72.1/75.2 eV and 73.1/76.6 eV are assigned respectively to Pt(II) and Pt(IV) oxidation state species. The presence of Pt oxide and hydroxide is common for ultrafine Pt NPs [90-92]. The Co 2p region of the XPS spectrum (Supporting Information File 1, Figure S10b) reveals the presence of Co on the surface of the catalyst. The Co is mainly observed under oxidized state. In this area, peaks corresponding to CoO and Co(OH)2 can be observed at 779.8/785.0 eV and 781.4/796.9 eV, respectively. Considering the easy oxidation of cobalt in air, cobalt is always observed under these forms when PtCo alloys are studied [60,93,94]. As far as Co single atoms on nitrogen-doped carbon materials, it has been shown that nitrogen doping of the carbon provides sites for Co incorporation. On such supports, cobalt is usually found in the ionic state in a CoN4 environment. XPS characterization of Co single atoms on nitrogen-doped carbon has shown two peaks for Co at a binding energy of 781.1 and 796.2 eV [95]. However, the CoN4 single atoms are usually prepared by high temperature pyrolysis, and in our case, the synthesis is conducted at room temperature. It is very unlikely that CoN4 species will be formed under these conditions.

The results of the XPS analysis performed on Pt3Ni/N-CNTHT revealed the species detailed in Table 6.

Table 6: Binding energies and ratio of Pt and Ni species obtained from XPS spectra.

Pt3Ni/N-CNTHT Species Binding energy (eV) Ratio (%)
Pt 4f7/2 Pt 4f5/2
Pt Pt(0) 71.2 74.4 60.1
  PtO 72.1 75.2 19.2
  Pt(OH)2 73.1 76.3 20.7
    Ni 2p3/2 Ni 2p1/2  
Ni Ni(0) 852.8 869.9 4.4
  NiO 854.3 873.3 16.4
  Ni(OH)2 856.0 874.3 51.6
  NiOOH 857.6 874.6 27.6

The Pt 4f region of the XPS spectrum is shown in Supporting Information File 1, Figure S11a. The deconvolution shows three pairs of doublets (corresponding to Pt 4f7/2 and Pt 4f5/2). The first doublet at 71.2/74.4 eV corresponds to metallic Pt(0). The metallic state represents 60.1 wt % of the sample. The doublets at 72.1/75.2 eV and 73.1/76.6 eV are assigned to PtO and Pt(OH)2 oxidation state species, respectively. The Ni 2p region of the XPS spectrum is presented in Supporting Information File 1, Figure S11b. The deconvoluted spectrum indicates that most of the Ni is present under oxidized states. In this area, doublets corresponding to NiO, Ni(OH)2 and Ni(OOH) can be observed at 854.3/873.3 eV, 856.0/874.3 eV, and 857.6/874.6 eV, respectively [96]. The important concentration of oxidized species is generally observed when PtNi alloys are studied [96,97]. As far as Ni single atoms on nitrogen-doped carbon as concerned, oxidized species have also been reported [98].

Taking into account all these results, we propose the following mechanism for the synthesis of Pt3M NPs supported on N-CNTs. During the first step of the synthesis, the transition metal, M, will anchor the carbon surface in atomic form after reduction of the metal salt. Cobalt and nickel atoms are well-dispersed due to the use of the ILs. The addition of water for the hydrolysis of the NaBH4 may explain the partial oxidation of the transition metal and the presence of residual metal in the final product, as oxidized species cannot undergo the galvanic displacement. During the galvanic replacement, the M atoms in metallic form can react with the platinum salt. After washing the catalyst and removing of the IL, Pt3M NPs supported on N-CNTs are obtained with unreacted M atoms (and clusters) at the surface of the carbon support.

Catalyst treatment: EDTA washing

The previous characterization revealed a large amount of unalloyed cobalt or nickel species on the support surface, which can be easily dissolved in acidic media, which then poisons the protonic group of the ionomer and the proton exchange membrane [5]. The impact of dissolved Co or Ni will be much more important in MEA configuration than in RRDE testing, specifically because the poisoned sulfonic groups could not be washed by liquid electrolyte in MEA, as it could be in RRDE setup. In order to minimize the impact of the treatment on both the catalyst and its support, we avoid the classically used acid leaching method [99], but preferred using and internally developed method, based on washing the material with a solution of ethylenediaminetetraacetic acid (EDTA). This consists of dispersing the Pt3Co/N-CNT or Pt3Ni/N-CNTHT in a 0.1 M EDTA solution and using ultrasonication. For the Pt3Co/N-CNT catalyst, the solution become purple in a few seconds, indicating the fast and easy dissolution of the Co; while for Pt3Ni/N-CNTHT, the solution become blue after a few hours. One can suppose there is a better interaction between the Ni and the N-CNTHT than the Co and the N-CNT. The catalysts were then filtered, washed with deionized water and dried for 24 hours at 80 °C.

The same structural characterization was performed on the washed Pt3Co/N-CNT catalyst. The results of the XPS analysis are reported in Supporting Information File 1, Table S1. The data do not show significant variations of the Pt(0)/PtO/Pt(OH)2 and Co(0)/CoO/Co(OH)2-CoN4 values, except a low decrease of the metallic content for both Pt and Co after the EDTA washing. TEM and STEM-HAADF images are reported in Figure 9. The same interplanar distance of 0.22 nm for the NPs as before washing has been found by HRTEM (Figure 9a). The STEM-HAADF image in Figure 9b shows an important decrease of Co atoms clusters on the surface of the N-CNT, and the EDX analysis shows an important decrease of the Co amount after the washing (Supporting Information File 1, Figure S12).

[2190-4286-10-125-9]

Figure 9: a) HRTEM images of Pt3Co/N-CNT and b) STEM-HAADF image of Pt3Co/N-CNT after washing with EDTA.

WAXS analysis was also performed. The superimposition of the PDF after and before washing is shown in Supporting Information File 1, Figure S13. The results show that the nanoparticles are metallic and well crystalized but with a mean diameter slightly larger after washing (2.6 nm) than before washing (2.3 nm). In that case, a correction factor of 1.5% should be applied to obtain a good agreement between the experimental results and a simulation performed from a model derived from pure Pt (spherical NP, 2 nm in diameter). From the TEM images, the mean diameter was calculated to be 2.4 ± 1 nm. The inductively coupled plasma optical emission spectrometry (ICP-OES) analysis give a weight ratio of 3.8% for Co and 45.5% for Pt. These results show that this new washing procedure is efficient to removed unalloyed non noble metals. Its impact on the structure of the catalyst particles is limited and, in the case of Co, it takes just a few minutes. One can assume that the conditions are mild enough to avoid damaging the catalyst support. Nevertheless, after this treatment, the catalyst composition seems to be closer to Pt4Co than Pt3Co.

Single cell testing and accelerated stress tests

First, the washed catalysts Pt3Co/N-CNT, Pt3Co/N-CNTHT and Pt3Ni/N-CNTHT were integrated into the MEA with a platinum loading of 0.26, 0.24 and 0.3 mgPt/cm2, respectively. A reference MEA, integrating the commercial reference catalyst Pt3Co/CB with a loading of 0.3 mgPt·cm−2, has been prepared and used as a reference. The polarization curves registered under air are reported in Figure 10. Focusing on the activation part of the curve, and as expected thanks to the RDE screening, Pt3Co/N-CNT is more active than Pt3Co/N-CNTHT and Pt3Ni/N-CNTHT. The performance obtained under air was shown to achieve 574 mW·cm−2 for Pt3Co/N-CNT, 278 mW·cm−2 for Pt3Co/N-CNTHT and 494 mW·cm−2 for Pt3Ni/N-CNTHT. When the Pt3Co/N-CNT is tested under pure O2 the power density obtained at 1.5 A·cm−2 is 996 mW·cm−2, which is higher than the results previously reported on an MEA integrating catalysts supported on CNTs [100,101].

[2190-4286-10-125-10]

Figure 10: Polarization of Pt3Co/CB (red); Pt3Co/N-CNT (dark blue); Pt3Co/N-CNTHT (light blue) and Pt3Ni/N-CNTHT (green) recorded under air, Pinlet = 2.5 bar, T = 80 °C, StH2 = 1.2; StAir = 3.5, RHanode = 50%; RHcathode = 30%.

The polarization curves recorded under O2 are shown in Supporting Information File 1, Figure S14. Even this performance is still lower than the performance of the reference MEA, 863 mW·cm−2 under air and 1118 mW·cm−2 under O2 at 1.5 A/cm2.

To evidence the added value of the EDTA washing protocol, an MEA integrating the catalysts not treated with a solution of EDTA are shown in Figure 11. One can see that, even at very low current, the beneficial impact of the developed washing protocol is clear. We suspect that for the unwashed catalyst the non-alloyed Co or Ni could be leached during the ink preparation and then trapped by the ionomer.

[2190-4286-10-125-11]

Figure 11: Polarization of Pt3Co/CB (red), unwashed Pt3Co/N-CNT (dark blue) and unwashed Pt3Ni/N-CNTHT (green) recorded under air, Pinlet = 2.5 bar, T = 80 °C, StH2 = 1.2; StAir = 3.5, RHanode = 50%; RHcathode = 30%.

Based on the quantitative analysis of the catalyst composition, it can be calculated that for the unwashed Pt3Co/N-CNT, 1.5 µmol of Co2+ is released into the ink, for each square centimeter of prepared MEA. As the cathode contain 20% weigh ratio of Nafion (equivalent weight: 1000), each square centimeter of cathodic active layer contains 0.18 µmol of sulfonic acid site. The anode is composed of 0.2 mgPt·cm−2 and a Nafion content of 25%, which leads to 0.13 µmol of sulfonic acid site per square centimeter of MEA. According to the membrane thickness (20 µm) the Nafion density (1.02 g·mL−1), and disregarding the impact of the reinforcement, it can be considered that 1 cm2 of HP membrane contains around 2 µmol of sulfonic acid site. Therefore, the total amount of sulfonic acid group in the MEA is around 2.31 µmol·cm−2geo, which is less than two times the theoretical amount of released Co. As one Co2+ cation can neutralize two sulfonic acid groups, the proton transport in the MEA integrating unwashed catalyst is almost impossible, which explains such low performance, even at low current.

Electrochemical impedance spectra (EIS) were collected on single cell containing an MEA based on Pt3Co/N-CNT and Pt3Co/CB at 0.1 A/cm2, under air and in the same conditions as the polarization experiments. The high frequency resistance was 4.6 and 3.2 mΩ, respectively (see Supporting Information File 1, Figure S15). The higher value for the MEA integrating the catalyst supported on CNTs might be due to worse dispersion of the ionomer into the active layer and worse interface of the active layer with the membrane than with the reference catalyst. Indeed, it is known that CNTs tend to form aggregates and worsen dispersion than Vulcan-like carbon black. This was also observed in SEM images of the prepared MEA (Figure 12).

[2190-4286-10-125-12]

Figure 12: SEM micrographs of MEA sections prepared with Pt3Co/CB at 0.3 mgPt·cm−2 and Pt3Co/CNT-N at 0.25 mgPt·cm−2.

Next, the catalyst support stability was evaluated using an AST for this purpose [102]. The evolution of the polarization registered during the AST is shown is Figure 13 for both the reference MEA and the MEAs integrating the Pt3Co/N-CNT catalysts.

[2190-4286-10-125-13]

Figure 13: Polarization curves performed on MEAs integrating: a) Pt3Co/CB; and b) Pt3Co/N-CNT. Dark blue: after conditioning, red: after 10 cycles, light blue: after 100 cycles, yellow: after 200 cycles, green: after 500 cycles and black: after 1000 cycles.

It can be observed that the degradation rate for the catalyst supported on N-CNTs is slower than the reference commercial catalyst. Indeed, even if the beginning-of-life performance of the reference MEA is better, after only 200 cycles, the performance of the Pt3Co/N-CNT MEA is then better. This validates the robustness of the synthesized catalyst.

Conclusion

In summary, a series of undoped, S- and N-doped CNTs have been synthetized by c-CVD. These carbon materials have been used together with an ionic liquid as a support structure for bimetallic Pt3Co and Pt3Ni NPs by a transmetalation method. The nature of the support significantly produced the NP distribution on the supports. For Pt3Co catalysts, a high metal loading (>45% w/w) and small NP size (<2 nm) was obtained on the support presenting the higher nitrogen content. Surprisingly, for Pt3Ni catalysts, a higher loading and smaller particle size was obtained on N-doped CNTs that were submitted to a high temperature treatment, which decreased the nitrogen content. HRTEM, WAXS and XPS analyses of the Pt3Co and Pt3Ni samples revealed that bimetallic Pt3M NP alloys were obtained with this method. Residual cobalt or nickel atoms were also present on the N-CNT surface. The experimental results showed enhanced catalytic performance for catalysts prepared with N-CNTs compared to the S-CNTs and commercial Pt3Co/CB catalysts. The best results were obtained for Pt3Co that used the more hydrophilic supports (N-CNT), and for Pt3Ni with the heat-treated N-CNTHT supports. Interestingly, our study shows that the catalytic performance is support-dependent. For the N-CNT support, the SA followed the trend Pt3Co > Pt3Ni, in accordance with literature reports; whereas for N-CNTHT, the SA followed the opposite order.

The use of Pt3M/N-CNT catalysts results in the reduction in the quantity of H2O2 produced during the ORR compared to the commercial Pt3Co/CB catalyst. After ex situ validation of the catalyst, the treatment with EDTA solution to remove unalloyed non noble metals (Co or Ni) was employed. The electrochemical characterization of the MEA containing washed and unwashed catalysts validated this new protocol. The ageing tests (AST), characterizing the catalyst support degradation, showed better resistance toward degradation of the Pt3Co/N-CNT than the reference Pt3Co/CB catalyst. Beyond these results, our future works will focus on the end-of-life analysis on the aged MEA and on the integration of catalysts supported on CNTs in the active layer to increase the beginning-of-life performance of the MEA.

Experimental

The metal precursors used for the syntheses were purchased from Strem Chemicals Inc., the ionic liquid was purchased from Solvionic, and the other chemicals were purchased from Sigma-Aldrich. All operations were carried out under argon atmosphere using standard Schlenk techniques or in an MBraun glovebox. For comparison purpose, the commercial catalyst Pt3Co/CB consists of 6 wt % Co, 46.7 wt % Pt on carbon black from Tanaka Kikinzoku Kogyo (reference TEC36V52) with NPs of around 5 nm.

CNT synthesis and functionalization

CNTs were grown using a catalytic-CVD process: ethylene as a carbon source, acetonitrile as a nitrogen/carbon source and thiophene as a sulfur/carbon source were decomposed at 650 °C on a Fe/Al2O3 catalyst in a vertical oven to produce CNTs and N-CNTs. First, the catalyst was reduced under argon/hydrogen (Ar/H2 (1.5/1): 375 mL·min−1) during 30 min at 650 °C. Undoped structures (called CNTs) were prepared from ethylene/H2 (375 mL·min−1 (1.5/1)) mixtures, N-doped structures (called N-CNTs) from acetonitrile/Ar/H2 (375 mL·min−1, Ar/H2 (1.5/1) bubbling through acetonitrile at 35 °C, 0.19 bar vapor pressure), and S-doped structures (called S-CNTs) according to a published procedure [103]. All CNTs were purified with a refluxing mixture of H2O/H2SO4 (50/50 v/v) for 3 h to remove the catalyst. The annealing of the N-CNTs was carried out at 1000 °C. 0.5 g of sample was placed in the chamber of a horizontal oven under Ar (200 mL·min−1). The annealing temperature was reached with a rate of 10 °C·min−1 and held for 2 h. Finally, the sample was cooled down to room temperature under Ar to obtain N-CNTHT.

Catalyst preparation

The Pt3Co/CNT catalysts were prepared following a modified procedure reported elsewhere [23]. Here, the ionic liquid, 1-butyl-3-methylimidazolium bis(trifluoromethanesulfonyl)imide ([bmim][Tf2N]) was used as a stabilizer in order to increase the interaction between the carbon support and the metallic salt and favor the formation of small nanoparticles. 2.04 mmol of CoCl2·6H2O were dissolved in 60 mL of ethanol and added in a 30 mL ethanol solution containing 0.2 g of CNTs and 0.48 mmol of [bmim][Tf2N]. The reaction mixture was sonicated for 20 min and then stirred vigorously. A freshly prepared NaBH4 solution in ethanol (0.15 mol·L−1) was added to the reaction mixture and allowed to react for 30 min. Afterwards, 100 mL of deionized water were added and the suspension was stirred for 3 h. Next, 1.27 mmol of K2PtCl4 was dissolved in 60 mL of deionized water added to the solution. After stirring overnight, the solution was filtered; the product was washed with ethanol and deionized water, and finally dried at 80 °C. The Pt3Ni/CNT catalysts were prepared using the same procedure using NiCl2·6H2O as the metal precursor.

Characterization

The samples were characterized using transmission electron microscopy (TEM, JEOL, JEM-1011 at 100 kV) and high-resolution TEM (HRTEM, JEM-ARM200F Cold FEG, STEM-EDX CENTURIO-X, GATAN Gif quantum ER), chemical analysis (CHN Perkin Elmer elemental analyzer), Raman spectroscopy at 633 nm (SmartsSPM-1000 AIST-NT) and thermal analysis under air (thermobalance Perkin Elmer Diamond TG). The textural characterization (BET surface area, SBET) of the material was evaluated by N2 adsorption–desorption analysis at −196 °C using a Quantachrome autosorb device. X-ray photoelectron spectroscopy (XPS) was performed on a ThermoScientific XPS K-alpha apparatus, which operated with an achromatized Mg K source (1253.6 eV). Pt and Co loadings were determined by inductively coupled plasma optical emission spectrometry (ICP-OES) analysis (Thermo Scientific, ICAP 6300 instrument). Wide angle X-ray scattering (WAXS) measurements were performed on a diffractometer dedicated to pair distribution function (PDF) analyses: graphite-monochromatized molybdenum radiation (0.07169 nm), solid state detection and low background setup. The samples were sealed in Lindemann glass capillaries (diameter 1.5 mm).

The MEA cross-sections were prepared by first cutting MEA samples (8 × 8 mm2) and embedding them in epoxy resin. Then, the MEA cross-sections were prepared by mechanical polishing until a mirror-like surface was achieved and were observed using a Zeiss FEG-SEM LEO1530.

Electrochemical measurements

RRDE measurements

The electrochemical properties of the prepared catalysts were investigated in a three-electrode system in 0.5 M H2SO4 solution at room temperature using a RRDE. A saturated mercury sulfate electrode (MSE, Bioanalytical system Inc., RE-2C) was used as the reference electrode and a platinum wire as the counter electrode (CE). All the potentials are presented to the reversible hydrogen electrode (RHE). The measurements were carried out using a BIOLOGIC VSP potentiostat. The used RRDE is the model AFE7R9GCPT from PINE research, and the disk is glassy carbon with an area of 0.2475 cm2. The ring is in Pt with a collection factor of 37%. The electrode is polished with 1 µm diamond paste and 0.05 µm alumina paste before use. The RRDE electrodes were prepared from a suspension of PtCo/CNT catalyst in 4 mL of isopropanol/DI water/Nafion® dispersion (type D-2020 from Dupont Fluoroproduct, 20% Nafion® dissolved in aliphatic alcohol) (80/19.5/0.5), and sonicated for 30 min. 30 µL of the prepared ink was deposited three times onto the polished glassy carbon disk electrode. A thin catalytic layer with a Pt loading of 100 µgPt/cm2 was obtained after evaporation of the solvent under air at room temperature. The electroactive surface area (ECSA) was calculated from the second cyclic voltammetry (CV) curves using an electrolyte saturated with N2 at 5 mV·s−1, from 0.04 to 1.08 V/RHE. The ECSA was estimated by integrating the current in the hydrogen desorption region between 0.04 and 0.4 V/RHE on the positive-going potential scan, corrected by the double layer current at 0.4 V/RHE and assuming 210 µC·cm−2(Pt) (Table 5). The activity was measured on the cyclic voltamograms using an electrolyte saturated with O2 at 5 mV·s−1 from 0.04 to 1.08 V/RHE fixing the rotation speed to 900 rpm The rotation is controlled by a PINE Research Instrumentation/model AFMSRCE device. The current density is normalized to the geometric surface area of the glassy carbon (0.247 cm2). The kinetic current is calculated following the equation:

[Graphic 1]

where i is the measured current at 0.9 V/RHE under O2 and corrected by a reference measurement made under N2 at the same potential and id is the measured current at 0.4 V/RHE under O2 and corrected by a reference measurement made under N2 at the same potential. There is no internal resistance correction.

The mass activity is finally obtained using the kinetic current density and the mass of the Pt loaded on the electrode. While the ORR occurs on the disk, the ring current was recorded, corresponding to H2O2 oxidation. This method was already reported elsewhere [104].

Single cell testing and accelerated stress tests

The single cell tests were performed using a Green Light (GL-40) test station. A graphite monopolar plate with a single serpentine flow field and 1 mm channel and landing dimensions were used. The active areas were 25 cm2. The polarization curves were recorded under current control, from open-circuit voltage (OCV) to high current with a ramping of 2 A·min−1. The polarization curves were recorded at 80 °C, with a pressure inlet of 2.5 bar on both sides, a stoichiometry of 1.2 at the anode side and 3.5 at the cathode side when air was used, or 5 when pure O2 was used. The relative humidity was managed by boilers and the values are 50% at the anode and 30% at the cathode. The conditioning is performed under air by maintaining the cell voltage at 0.5 V for 8 h, the operating conditions are similar but the relative humidity fixed at 100% on both sides.

For the ASTs, the cell was maintained at 80 °C, and the inlet pressure was 2.5 bar. The anode and cathode were fed with H2 and N2, respectively, both with 100% relative humidity. The cell voltage was controlled by a Gamry Instruments reference 3000 potentiostat. The cell voltage was cycled between 1.0 V and 1.5 V with a scan rate of 500 mV·s−1. The polarization curves were registered under air after 10, 100, 200, 500 and 1000 cycles. The MEAs were prepared by hot pressing gas diffusion electrodes on a reinforced HP Nafion® membrane. The cathodes were prepared by manually spraying the cathodic active layer on the gas diffusion layer (GDL) (SGL-sigracet® 24 BC). The used ionomer was Nafion® D2020, its dry extract in the active layer is 20% for the synthesized catalysts and 25% for the commercial catalyst. The anode catalyst is Pt/CB 50% weight ratio from TKK, the loading of the anode is 0.2 mgPt/cm2 and the gas diffusion electrode was prepared by screen printing.

Supporting Information

Supporting Information File 1: Additional experimental data.
Format: PDF Size: 2.3 MB Download

Acknowledgements

The authors wish to thank the CEA LITEN, the CNRS and Région OCCITANIE for funding the thesis in which most the results have been obtained. This work has been partially supported by the European Union’s Seventh Framework Program (FP7/2007-20013) for Fuel Cell and Hydrogen Joint Technology Initiative under grants no 325239 (NanoCAT).

References

  1. Wilberforce, T.; Alaswad, A.; Palumbo, A.; Dassisti, M.; Olabi, A. G. Int. J. Hydrogen Energy 2016, 41, 16509–16522. doi:10.1016/j.ijhydene.2016.02.057
    Return to citation in text: [1]
  2. Gallo, A. B.; Simões-Moreira, J. R.; Costa, H. K. M.; Santos, M. M.; Moutinho dos Santos, E. Renewable Sustainable Energy Rev. 2016, 65, 800–822. doi:10.1016/j.rser.2016.07.028
    Return to citation in text: [1]
  3. Antolini, E. J. Mater. Sci. 2003, 38, 2995–3005. doi:10.1023/a:1024771618027
    Return to citation in text: [1]
  4. Marković, N. M.; Ross, P. N. Surf. Sci. Rep. 2002, 45, 117–229. doi:10.1016/s0167-5729(01)00022-x
    Return to citation in text: [1]
  5. Gasteiger, H. A.; Kocha, S. S.; Sompalli, B.; Wagner, F. T. Appl. Catal., B 2005, 56, 9–35. doi:10.1016/j.apcatb.2004.06.021
    Return to citation in text: [1] [2] [3]
  6. Pavlišič, A.; Jovanovič, P.; Šelih, V. S.; Šala, M.; Hodnik, N.; Gaberšček, M. J. Electrochem. Soc. 2018, 165, F3161–F3165. doi:10.1149/2.0191806jes
    Return to citation in text: [1]
  7. Bodner, M.; Senn, J.; Hacker, V. Chapter 7 - Degradation Mechanisms and Their Lifetime. In Fuel Cells and Hydrogen; Hacker, V.; Mitsushima, S., Eds.; Elsevier, 2018; pp 139–154. doi:10.1016/b978-0-12-811459-9.00007-4
    Return to citation in text: [1]
  8. Meier, J. C.; Galeano, C.; Katsounaros, I.; Topalov, A. A.; Kostka, A.; Schüth, F.; Mayrhofer, K. J. J. ACS Catal. 2012, 2, 832–843. doi:10.1021/cs300024h
    Return to citation in text: [1]
  9. de Bruijn, F. A.; Dam, V. A. T.; Janssen, G. J. M. Fuel Cells 2008, 8, 3–22. doi:10.1002/fuce.200700053
    Return to citation in text: [1]
  10. Macauley, N.; Papadias, D. D.; Fairweather, J.; Spernjak, D.; Langlois, D.; Ahluwalia, R.; More, K. L.; Mukundan, R.; Borup, R. L. J. Electrochem. Soc. 2018, 165, F3148–F3160. doi:10.1149/2.0061806jes
    Return to citation in text: [1]
  11. Roen, L. M.; Paik, C. H.; Jarvi, T. D. Electrochem. Solid-State Lett. 2004, 7, A19–A22. doi:10.1149/1.1630412
    Return to citation in text: [1]
  12. Lv, H.; Li, D.; Strmcnik, D.; Paulikas, A. P.; Markovic, N. M.; Stamenkovic, V. R. Nano Energy 2016, 29, 149–165. doi:10.1016/j.nanoen.2016.04.008
    Return to citation in text: [1]
  13. Kulkarni, A.; Siahrostami, S.; Patel, A.; Nørskov, J. K. Chem. Rev. 2018, 118, 2302–2312. doi:10.1021/acs.chemrev.7b00488
    Return to citation in text: [1]
  14. Shao, M.; Chang, Q.; Dodelet, J.-P.; Chenitz, R. Chem. Rev. 2016, 116, 3594–3657. doi:10.1021/acs.chemrev.5b00462
    Return to citation in text: [1]
  15. Hwang, S. J.; Yoo, S. J.; Shin, J.; Cho, Y.-H.; Jang, J. H.; Cho, E.; Sung, Y.-E.; Nam, S. W.; Lim, T.-H.; Lee, S.-C.; Kim, S.-K. Sci. Rep. 2013, 3, 1309. doi:10.1038/srep01309
    Return to citation in text: [1]
  16. Wang, L.; Gao, W.; Liu, Z.; Zeng, Z.; Liu, Y.; Giroux, M.; Chi, M.; Wang, G.; Greeley, J.; Pan, X.; Wang, C. ACS Catal. 2018, 8, 35–42. doi:10.1021/acscatal.7b02501
    Return to citation in text: [1]
  17. Wang, D.; Xin, H. L.; Hovden, R.; Wang, H.; Yu, Y.; Muller, D. A.; DiSalvo, F. J.; Abruña, H. D. Nat. Mater. 2013, 12, 81–87. doi:10.1038/nmat3458
    Return to citation in text: [1]
  18. Lin, R.; Cao, C.; Zhao, T.; Huang, Z.; Li, B.; Wieckowski, A.; Ma, J. J. Power Sources 2013, 223, 190–198. doi:10.1016/j.jpowsour.2012.09.073
    Return to citation in text: [1]
  19. Chen, Y.; Liang, Z.; Yang, F.; Liu, Y.; Chen, S. J. Phys. Chem. C 2011, 115, 24073–24079. doi:10.1021/jp207828n
    Return to citation in text: [1]
  20. Zhao, X.; Takao, S.; Higashi, K.; Kaneko, T.; Samjeskè, G.; Sekizawa, O.; Sakata, T.; Yoshida, Y.; Uruga, T.; Iwasawa, Y. ACS Catal. 2017, 7, 4642–4654. doi:10.1021/acscatal.7b00964
    Return to citation in text: [1]
  21. Zhu, H.; Li, X.; Wang, F. Int. J. Hydrogen Energy 2011, 36, 9151–9154. doi:10.1016/j.ijhydene.2011.04.224
    Return to citation in text: [1]
  22. Kuttiyiel, K. A.; Sasaki, K.; Choi, Y.; Su, D.; Liu, P.; Adzic, R. R. Nano Lett. 2012, 12, 6266–6271. doi:10.1021/nl303362s
    Return to citation in text: [1]
  23. Kristian, N.; Yu, Y.; Lee, J.-M.; Liu, X.; Wang, X. Electrochim. Acta 2010, 56, 1000–1007. doi:10.1016/j.electacta.2010.09.073
    Return to citation in text: [1] [2] [3]
  24. Zhang, C.; Shen, X.; Pan, Y.; Peng, Z. Front. Energy 2017, 11, 268–285. doi:10.1007/s11708-017-0466-6
    Return to citation in text: [1]
  25. Bing, Y.; Liu, H.; Zhang, L.; Ghosh, D.; Zhang, J. Chem. Soc. Rev. 2010, 39, 2184–2202. doi:10.1039/b912552c
    Return to citation in text: [1]
  26. Tunckol, M.; Durand, J.; Serp, P. Carbon 2012, 50, 4303–4334. doi:10.1016/j.carbon.2012.05.017
    Return to citation in text: [1]
  27. Zhang, G.-R.; Wolker, T.; Sandbeck, D. J. S.; Munoz, M.; Mayrhofer, K. J. J.; Cherevko, S.; Etzold, B. J. M. ACS Catal. 2018, 8, 8244–8254. doi:10.1021/acscatal.8b02018
    Return to citation in text: [1] [2]
  28. Guo, S.; Dong, S.; Wang, E. Adv. Mater. (Weinheim, Ger.) 2010, 22, 1269–1272. doi:10.1002/adma.200903379
    Return to citation in text: [1]
  29. Luo, C.; Xie, H.; Wang, Q.; Luo, G.; Liu, C. J. Nanomater. 2015, 560392. doi:10.1155/2015/560392
    Return to citation in text: [1]
  30. Akbari, E.; Buntat, Z. Int. J. Energy Res. 2017, 41, 92–102. doi:10.1002/er.3600
    Return to citation in text: [1]
  31. Asset, T.; Job, N.; Busby, Y.; Crisci, A.; Martin, V.; Stergiopoulos, V.; Bonnaud, C.; Serov, A.; Atanassov, P.; Chattot, R.; Dubau, L.; Maillard, F. ACS Catal. 2018, 8, 893–903. doi:10.1021/acscatal.7b03539
    Return to citation in text: [1] [2]
  32. Ayala, P.; Arenal, R.; Rümmeli, M.; Rubio, A.; Pichler, T. Carbon 2010, 48, 575–586. doi:10.1016/j.carbon.2009.10.009
    Return to citation in text: [1] [2]
  33. Kiciński, W.; Szala, M.; Bystrzejewski, M. Carbon 2014, 68, 1–32. doi:10.1016/j.carbon.2013.11.004
    Return to citation in text: [1] [2]
  34. Fan, J.-J.; Fan, Y.-J.; Wang, R.-X.; Xiang, S.; Tang, H.-G.; Sun, S.-G. J. Mater. Chem. A 2017, 5, 19467–19475. doi:10.1039/c7ta05102f
    Return to citation in text: [1] [2]
  35. Chen, Y.; Wang, J.; Liu, H.; Banis, M. N.; Li, R.; Sun, X.; Sham, T.-K.; Ye, S.; Knights, S. J. Phys. Chem. C 2011, 115, 3769–3776. doi:10.1021/jp108864y
    Return to citation in text: [1]
  36. Chen, J.; Shan, J. Y.; Tsukada, T.; Munekane, F.; Kuno, A.; Matsuo, M.; Hayashi, T.; Kim, Y. A.; Endo, M. Carbon 2007, 45, 274–280. doi:10.1016/j.carbon.2006.09.028
    Return to citation in text: [1]
  37. Fujisawa, K.; Tojo, T.; Muramatsu, H.; Elías, A. L.; Vega-Díaz, S. M.; Tristán-López, F.; Kim, J. H.; Hayashi, T.; Kim, Y. A.; Endo, M.; Terrones, M. Nanoscale 2011, 3, 4359–4364. doi:10.1039/c1nr10717h
    Return to citation in text: [1]
  38. Liu, H.; Zhang, Y.; Li, R.; Sun, X.; Abou-Rachid, H. J. Nanopart. Res. 2012, 14, 1016. doi:10.1007/s11051-012-1016-0
    Return to citation in text: [1]
  39. Podyacheva, O. Y.; Cherepanova, S. V.; Romanenko, A. I.; Kibis, L. S.; Svintsitskiy, D. A.; Boronin, A. I.; Stonkus, O. A.; Suboch, A. N.; Puzynin, A. V.; Ismagilov, Z. R. Carbon 2017, 122, 475–483. doi:10.1016/j.carbon.2017.06.094
    Return to citation in text: [1]
  40. Cançado, L. G.; Jorio, A.; Martins Ferreira, E. H.; Stavale, F.; Achete, C. A.; Capaz, R. B.; Moutinho, M. V. O.; Lombardo, A.; Kulmala, T. S.; Ferrari, A. C. Nano Lett. 2011, 11, 3190–3196. doi:10.1021/nl201432g
    Return to citation in text: [1]
  41. DiLeo, R. A.; Landi, B. J.; Raffaelle, R. P. J. Appl. Phys. 2007, 101, 064307. doi:10.1063/1.2712152
    Return to citation in text: [1]
  42. Serp, P.; Machado, B. CHAPTER 4 Surface Chemistry of Nanostructured Carbon Materials and Preparation of Nanocarbon Supported Catalysts. Nanostructured Carbon Materials for Catalysis; The Royal Society of Chemistry, 2015; pp 163–222. doi:10.1039/9781782622567-00163
    Return to citation in text: [1]
  43. Mattila, S.; Leiro, J. A.; Laajalehto, K. Appl. Surf. Sci. 2003, 212–213, 97–100. doi:10.1016/s0169-4332(03)00413-6
    Return to citation in text: [1]
  44. Shahgaldi, S.; Hamelin, J. Carbon 2015, 94, 705–728. doi:10.1016/j.carbon.2015.07.055
    Return to citation in text: [1]
  45. Du, L.; Shao, Y.; Sun, J.; Yin, G.; Liu, J.; Wang, Y. Nano Energy 2016, 29, 314–322. doi:10.1016/j.nanoen.2016.03.016
    Return to citation in text: [1]
  46. Wu, B.; Hu, D.; Yu, Y.; Kuang, Y.; Zhang, X.; Chen, J. Chem. Commun. 2010, 46, 7954–7956. doi:10.1039/c0cc02956d
    Return to citation in text: [1] [2]
  47. Wang, J.; Chu, H.; Li, Y. ACS Nano 2008, 2, 2540–2546. doi:10.1021/nn800510g
    Return to citation in text: [1]
  48. Janiak, C. Z. Naturforsch., B: J. Chem. Sci. 2013, 68, 1059–1087. doi:10.5560/znb.2013-3140
    Return to citation in text: [1]
  49. Cao, C.; Wu, M.; Jiang, J.; Cheng, H.-P. Phys. Rev. B 2010, 81, 205424. doi:10.1103/physrevb.81.205424
    Return to citation in text: [1]
  50. Sessi, V.; Stepanow, S.; Rudenko, A. N.; Krotzky, S.; Kern, K.; Hiebel, F.; Mallet, P.; Veuillen, J.-Y.; Šipr, O.; Honolka, J.; Brookes, N. B. New J. Phys. 2014, 16, 062001. doi:10.1088/1367-2630/16/6/062001
    Return to citation in text: [1]
  51. Makaremi, M.; Mortazavi, B.; Singh, C. V. J. Phys. Chem. C 2017, 121, 18575–18583. doi:10.1021/acs.jpcc.7b04511
    Return to citation in text: [1]
  52. Kattel, S. RSC Adv. 2013, 3, 21110–21117. doi:10.1039/c3ra43810d
    Return to citation in text: [1] [2]
  53. Hu, L.; Hu, X.; Wu, X.; Du, C.; Dai, Y.; Deng, J. Phys. B (Amsterdam, Neth.) 2010, 405, 3337–3341. doi:10.1016/j.physb.2010.05.001
    Return to citation in text: [1]
  54. Nesselberger, M.; Ashton, S.; Meier, J. C.; Katsounaros, I.; Mayrhofer, K. J. J.; Arenz, M. J. Am. Chem. Soc. 2011, 133, 17428–17433. doi:10.1021/ja207016u
    Return to citation in text: [1]
  55. Paulus, U. A.; Schmidt, T. J.; Gasteiger, H. A.; Behm, R. J. J. Electroanal. Chem. 2001, 495, 134–145. doi:10.1016/s0022-0728(00)00407-1
    Return to citation in text: [1]
  56. Shih, Y.-H.; Sagar, G. V.; Lin, S. D. J. Phys. Chem. C 2008, 112, 123–130. doi:10.1021/jp071807h
    Return to citation in text: [1]
  57. Stamenkovic, V. R.; Mun, B. S.; Mayrhofer, K. J. J.; Ross, P. N.; Markovic, N. M. J. Am. Chem. Soc. 2006, 128, 8813–8819. doi:10.1021/ja0600476
    Return to citation in text: [1]
  58. Wang, C.; van der Vliet, D.; Chang, K.-C.; You, H.; Strmcnik, D.; Schlueter, J. A.; Markovic, N. M.; Stamenkovic, V. R. J. Phys. Chem. C 2009, 113, 19365–19368. doi:10.1021/jp908203p
    Return to citation in text: [1] [2]
  59. Stamenković, V.; Schmidt, T. J.; Ross, P. N.; Marković, N. M. J. Phys. Chem. B 2002, 106, 11970–11979. doi:10.1021/jp021182h
    Return to citation in text: [1] [2] [3]
  60. Loukrakpam, R.; Luo, J.; He, T.; Chen, Y.; Xu, Z.; Njoki, P. N.; Wanjala, B. N.; Fang, B.; Mott, D.; Yin, J.; Klar, J.; Powell, B.; Zhong, C.-J. J. Phys. Chem. C 2011, 115, 1682–1694. doi:10.1021/jp109630n
    Return to citation in text: [1] [2] [3]
  61. Xu, Y.; Hou, S.; Liu, Y.; Zhang, Y.; Wang, H.; Zhang, B. Chem. Commun. 2012, 48, 2665–2667. doi:10.1039/c2cc16798k
    Return to citation in text: [1]
  62. Yamamoto, K.; Kolb, D. M.; Kötz, R.; Lehmpfuhl, G. J. Electroanal. Chem. Interfacial Electrochem. 1979, 96, 233–239. doi:10.1016/s0022-0728(79)80380-0
    Return to citation in text: [1]
  63. Vinayan, B. P.; Ramaprabhu, S. Nanoscale 2013, 5, 5109–5118. doi:10.1039/c3nr00585b
    Return to citation in text: [1]
  64. Du, S.; Lu, Y.; Malladi, S. K.; Xu, Q.; Steinberger-Wilckens, R. J. Mater. Chem. A 2014, 2, 692–698. doi:10.1039/c3ta13608f
    Return to citation in text: [1]
  65. Rosado, G.; Verde, Y.; Valenzuela-Muñiz, A. M.; Barbosa, R.; Miki Yoshida, M.; Escobar, B. Int. J. Hydrogen Energy 2016, 41, 23260–23271. doi:10.1016/j.ijhydene.2016.07.098
    Return to citation in text: [1]
  66. Huang, H.; Fan, Y.; Wang, X. Electrochim. Acta 2012, 80, 118–125. doi:10.1016/j.electacta.2012.06.129
    Return to citation in text: [1]
  67. Choi, D. S.; Robertson, A. W.; Warner, J. H.; Kim, S. O.; Kim, H. Adv. Mater. (Weinheim, Ger.) 2016, 28, 7115–7122. doi:10.1002/adma.201600469
    Return to citation in text: [1]
  68. Oezaslan, M.; Hasché, F.; Strasser, P. J. Electrochem. Soc. 2012, 159, B394–B405. doi:10.1149/2.075204jes
    Return to citation in text: [1]
  69. Vinayan, B. P.; Imran Jafri, R.; Nagar, R.; Rajalakshmi, N.; Sethupathi, K.; Ramaprabhu, S. Int. J. Hydrogen Energy 2012, 37, 412–421. doi:10.1016/j.ijhydene.2011.09.069
    Return to citation in text: [1]
  70. Rao, C. V.; Reddy, A. L. M.; Ishikawa, Y.; Ajayan, P. M. Carbon 2011, 49, 931–936. doi:10.1016/j.carbon.2010.10.056
    Return to citation in text: [1]
  71. Jayasayee, K.; Rob Van Veen, J. A.; Manivasagam, T. G.; Celebi, S.; Hensen, E. J. M.; de Bruijn, F. A. Appl. Catal., B 2012, 111–112, 515–526. doi:10.1016/j.apcatb.2011.11.003
    Return to citation in text: [1]
  72. Xu, C.; Hoque, M. A.; Chiu, G.; Sung, T.; Chen, Z. RSC Adv. 2016, 6, 112226–112231. doi:10.1039/c6ra19924k
    Return to citation in text: [1]
  73. Higgins, D.; Hoque, M. A.; Seo, M. H.; Wang, R.; Hassan, F.; Choi, J.-Y.; Pritzker, M.; Yu, A.; Zhang, J.; Chen, Z. Adv. Funct. Mater. 2014, 24, 4325–4336. doi:10.1002/adfm.201400161
    Return to citation in text: [1]
  74. Martin-Martinez, M.; Machado, B. F.; Serp, P.; Morales-Torres, S.; Silva, A. M. T.; Figueiredo, J. L.; Faria, J. L.; Gomes, H. T. Catal. Today 2019, in press. doi:10.1016/j.cattod.2019.03.014
    Return to citation in text: [1]
  75. Speder, J.; Altmann, L.; Bäumer, M.; Kirkensgaard, J. J. K.; Mortensen, K.; Arenz, M. RSC Adv. 2014, 4, 14971–14978. doi:10.1039/c4ra00261j
    Return to citation in text: [1]
  76. Chong, L.; Wen, J.; Kubal, J.; Sen, F. G.; Zou, J.; Greeley, J.; Chan, M.; Barkholtz, H.; Ding, W.; Liu, D.-J. Science 2018, 362, 1276–1281. doi:10.1126/science.aau0630
    Return to citation in text: [1]
  77. Hammer, B.; Nørskov, J. K. Theoretical surface science and catalysis—calculations and concepts. Advances in Catalysis; Academic Press, 2000; Vol. 45, pp 71–129. doi:10.1016/s0360-0564(02)45013-4
    Return to citation in text: [1]
  78. Goellner, V.; Armel, V.; Zitolo, A.; Fonda, E.; Jaouen, F. J. Electrochem. Soc. 2015, 162, H403–H414. doi:10.1149/2.1091506jes
    Return to citation in text: [1]
  79. Maldonado, S.; Stevenson, K. J. J. Phys. Chem. B 2005, 109, 4707–4716. doi:10.1021/jp044442z
    Return to citation in text: [1]
  80. Hasché, F.; Oezaslan, M.; Strasser, P.; Fellinger, T.-P. J. Energy Chem. 2016, 25, 251–257. doi:10.1016/j.jechem.2016.01.024
    Return to citation in text: [1]
  81. Mukerjee, S.; Srinivasan, S.; Soriaga, M. P.; McBreen, J. J. Electrochem. Soc. 1995, 142, 1409–1422. doi:10.1149/1.2048590
    Return to citation in text: [1]
  82. Wang, X. X.; Cullen, D. A.; Pan, Y.-T.; Hwang, S.; Wang, M.; Feng, Z.; Wang, J.; Engelhard, M. H.; Zhang, H.; He, Y.; Shao, Y.; Su, D.; More, K. L.; Spendelow, J. S.; Wu, G. Adv. Mater. (Weinheim, Ger.) 2018, 30, 1706758. doi:10.1002/adma.201706758
    Return to citation in text: [1]
  83. Cheng, Q.; Yang, L.; Zou, L.; Zou, Z.; Chen, C.; Hu, Z.; Yang, H. ACS Catal. 2017, 7, 6864–6871. doi:10.1021/acscatal.7b02326
    Return to citation in text: [1]
  84. Dilpazir, S.; He, H.; Li, Z.; Wang, M.; Lu, P.; Liu, R.; Xie, Z.; Gao, D.; Zhang, G. ACS Appl. Energy Mater. 2018, 1, 3283–3291. doi:10.1021/acsaem.8b00490
    Return to citation in text: [1]
  85. Han, Y.; Wang, Y.-G.; Chen, W.; Xu, R.; Zheng, L.; Zhang, J.; Luo, J.; Shen, R.-A.; Zhu, Y.; Cheong, W.-C.; Chen, C.; Peng, Q.; Wang, D.; Li, Y. J. Am. Chem. Soc. 2017, 139, 17269–17272. doi:10.1021/jacs.7b10194
    Return to citation in text: [1]
  86. Su, P.; Iwase, K.; Nakanishi, S.; Hashimoto, K.; Kamiya, K. Small 2016, 12, 6083–6089. doi:10.1002/smll.201602158
    Return to citation in text: [1]
  87. Zhao, C.; Dai, X.; Yao, T.; Chen, W.; Wang, X.; Wang, J.; Yang, J.; Wei, S.; Wu, Y.; Li, Y. J. Am. Chem. Soc. 2017, 139, 8078–8081. doi:10.1021/jacs.7b02736
    Return to citation in text: [1]
  88. Jen, S. U. J. Alloys Compd. 1996, 234, 231–234. doi:10.1016/0925-8388(95)02046-2
    Return to citation in text: [1]
  89. Bergounhou, C.; Blandy, C.; Choukroun, R.; Lecante, P.; Lorber, C.; Pellegatta, J.-L. New J. Chem. 2007, 31, 218–223. doi:10.1039/b614910c
    Return to citation in text: [1]
  90. Kim, K. S.; Winograd, N.; Davis, R. E. J. Am. Chem. Soc. 1971, 93, 6296–6297. doi:10.1021/ja00752a065
    Return to citation in text: [1] [2]
  91. Anumol, E. A.; Kundu, P.; Deshpande, P. A.; Madras, G.; Ravishankar, N. ACS Nano 2011, 5, 8049–8061. doi:10.1021/nn202639f
    Return to citation in text: [1]
  92. Tian, Z. Q.; Jiang, S. P.; Liang, Y. M.; Shen, P. K. J. Phys. Chem. B 2006, 110, 5343–5350. doi:10.1021/jp056401o
    Return to citation in text: [1]
  93. Li, D.; Wang, H.; Shi, C.; Xu, X. RSC Adv. 2016, 6, 35340–35347. doi:10.1039/c6ra04818h
    Return to citation in text: [1]
  94. Duong, H. T.; Rigsby, M. A.; Zhou, W.-P.; Wieckowski, A. J. Phys. Chem. C 2007, 111, 13460–13465. doi:10.1021/jp072586i
    Return to citation in text: [1]
  95. Fei, H.; Dong, J.; Arellano-Jiménez, M. J.; Ye, G.; Dong Kim, N.; Samuel, E. L. G.; Peng, Z.; Zhu, Z.; Qin, F.; Bao, J.; Yacaman, M. J.; Ajayan, P. M.; Chen, D.; Tour, J. M. Nat. Commun. 2015, 6, 8668. doi:10.1038/ncomms9668
    Return to citation in text: [1]
  96. Luo, B.; Xu, S.; Yan, X.; Xue, Q. J. Electrochem. Soc. 2013, 160, F262–F268. doi:10.1149/2.056303jes
    Return to citation in text: [1] [2]
  97. Zignani, S. C.; Baglio, V.; Sebastián, D.; Saccà, A.; Gatto, I.; Aricò, A. S. Materials 2017, 10, 317. doi:10.3390/ma10030317
    Return to citation in text: [1]
  98. Yang, H. B.; Hung, S.-F.; Liu, S.; Yuan, K.; Miao, S.; Zhang, L.; Huang, X.; Wang, H.-Y.; Cai, W.; Chen, R.; Gao, J.; Yang, X.; Chen, W.; Huang, Y.; Chen, H. M.; Li, C. M.; Zhang, T.; Liu, B. Nat. Energy 2018, 3, 140–147. doi:10.1038/s41560-017-0078-8
    Return to citation in text: [1]
  99. Mani, P.; Srivastava, R.; Strasser, P. J. Phys. Chem. C 2008, 112, 2770–2778. doi:10.1021/jp0776412
    Return to citation in text: [1]
  100. Berber, M. R.; Hafez, I. H.; Fujigaya, T.; Nakashima, N. Sci. Rep. 2015, 5, 16711. doi:10.1038/srep16711
    Return to citation in text: [1]
  101. Higgins, D. C.; Meza, D.; Chen, Z. J. Phys. Chem. C 2010, 114, 21982–21988. doi:10.1021/jp106814j
    Return to citation in text: [1]
  102. Park, Y.-C.; Kakinuma, K.; Uchida, M.; Uchida, H.; Watanabe, M. Electrochim. Acta 2014, 123, 84–92. doi:10.1016/j.electacta.2013.12.120
    Return to citation in text: [1]
  103. Louisia, S.; Contreras, R. C.; Heitzmann, M.; Rosa Axet, M.; Jacques, P.-A.; Serp, P. Catal. Commun. 2018, 109, 65–70. doi:10.1016/j.catcom.2018.02.024
    Return to citation in text: [1]
  104. Sethuraman, V. A.; Weidner, J. W.; Haug, A. T.; Motupally, S.; Protsailo, L. V. J. Electrochem. Soc. 2008, 155, B50–B57. doi:10.1149/1.2801980
    Return to citation in text: [1]
Other Beilstein-Institut Open Science Activities