Hierarchically structured catalyst layer for the oxygen reduction reaction fabricated by electrodeposition of platinum on carbon nanotube coated carbon fiber

Abstract

The cost-efficient fabrication of cathode catalyst layers with low Pt-loadings and high oxygen reduction reaction (ORR) performances is of prime importance towards the commercialization of low temperature fuel cells. Here, an attempt has been made to fabricate a hierarchically structured cathode catalyst layer consisting of Pt-nanoparticle clusters supported on defective carbon nanotube (CNT) coated carbon fiber (CNTCF). CNTs were grown on carbon fiber (CF) using chemical vapor deposition (CVD), while electrodeposition was employed to deposit Pt-nanoparticle clusters onto the CNTCF. The effect of Pt-loading on the oxygen reduction reaction (ORR) performance of the Pt-coated CNTCF (Pt-CNTCF) electrocatalysts was studied by varying the electrodeposition time (t_ed) between 5 and 30 min, which resulted in a variation of the Pt weight fraction in Pt-CNTCF from almost zero to ∼0.3. Linear sweep voltammetry using a Pt-CNTCF modified glassy carbon (GC) rotating disc electrode (RDE) was employed to study the ORR performance of the samples. The CNTCF support, due to the defective structure of the CNTs, itself exhibits significant ORR activity with an electron transfer number (n) of ∼2.6, which leads to a synergistic enhancement of the overall electrocatalytic performance of Pt-CNTCF. For low Pt-loading on the GC (∼5 μg cm⁻²), the contribution of the CNTCF support dominated, while the Pt-nanoclusters governed the ORR performance at higher Pt-loading (>10 μg cm⁻²), with n > 3.5. Hence, a very low Pt-loading (∼5 μg cm⁻²) may not be suitable for polymer electrolyte membrane fuel cells as the production of substantial amounts of H₂O₂ on the catalyst supports may accelerate membrane degradation.

---

1. Introduction

Heterogeneous catalysis of the oxygen reduction reaction (ORR) is of utmost importance in various electrochemical devices, such as low-temperature fuel cells (FCs) and metal–air batteries. Conventionally, the reaction is catalyzed using precious metal (Pt group metals and their alloys)-based catalysts, where a typical cathode catalyst layer consists of metallic particles supported on high surface area carbons. The requirement for large amounts of precious catalyst imposes serious limitations on the large-scale applications of these electrochemical devices.¹ Apart from the high cost, corrosion of the catalyst support and catalyst particle agglomeration impose serious durability issues on the Pt-based catalyst layers.² Hence, the fabrication of cost-effective catalyst layers with low metal loading and high durability is of significant importance.^3–6

Efforts towards the realization of such catalyst layers have been focused on either the reduction of Pt-loading or Pt-free catalysts.^7,8 To synthesize highly active catalyst layers with low-Pt loading, Pt-particles with an optimized size are dispersed uniformly on a highly corrosion resistant support material. Among others, carbon nanomaterials such as carbon nanotubes (CNTs) and graphene, due to their high strengths, large surface areas,^9–11 high corrosion resistances¹² and excellent thermal as well as electronic conductivities,^13,14 have been studied extensively as catalyst supports.^15–17 Their performance is affected significantly by factors such as the electronic connectivity between individual catalyst particles and the catalyst support,¹⁸ contact resistance at the catalyst layer/gas diffusion layer interface,¹⁹ agglomeration of the catalyst support and/or catalyst particles,²⁰ catalyst layer thickness,²¹ catalyst particle size,²² etc. These factors can be controlled by employing various structural modifications in the catalyst support and/or by using different routes to deposit the catalyst particles.^23–28 Catalyst layers having various CNT arrangements such as randomly orientated, aligned or 3-dimensional (3D) networked CNTs have been studied.^18,23,24,29 Among these, the 3D network formed by the secondary growth of CNTs on a primary aligned CNT array has been reported to provide a considerably high surface area and high internal connectivity.²⁹ Superior performance can also be attained by reducing the catalyst layer thickness through the high loading of Pt-nanoparticles with a narrow size distribution.²⁵ Composites consisting of CNT coated carbon cloth/carbon paper (CC/CP) synthesized by the direct growth of CNTs on CC/CP have been employed as catalyst supports for highly durable catalyst layers.^30–33 However, the direct coating of CNTs onto the CC/CP only partially covers their fibers and forms a layer of CNTs on the surface.

Apart from their applications as catalyst supports, defective carbon nanomaterials, particularly nitrogen containing CNTs and graphene, are being studied as Pt-free electrocatalysts for the ORR.³⁴ More recently, apart from Pt-free catalysts, N-doped CNTs/graphene have also been employed as catalyst supports for Pt-nanoparticles to fabricate catalyst layers with enhanced ORR activities compared to those of pristine CNT/graphene supported Pt.^35–37 The enhancement may be attributed to the modified structural as well as electronic configurations.³⁸ Chetty et al. have attributed the enhanced activity to the anchoring effect of the pyridinic/pyrrolic-N sites during metal deposition.³⁹ Similarly, a better dispersion of Pt-nanoparticles on N-doped CNTs compared to that of CNTs has been reported.³⁶ More recently, Liu et al. have shown that the presence of defects in the outer CNT walls improves their ORR activity significantly.⁴⁰ However, the contribution of such a defective CNT-based support on the ORR performance of a catalyst layer has not been investigated well.

The present study reports a hierarchically structured and 3D networked FC cathode/anode catalyst layer consisting of Pt-nanoparticle clusters supported on CNT coated carbon fiber (CNTCF). Here, hierarchically structured CNTCF was synthesized by the direct growth of CNTs on the carbon fiber (CF) substrate by chemical vapor deposition (CVD). The direct growth of CNTs on CF provides good electronic connectivity between the CNTs and CF surface.^11,41 The catalyst coating on the CNTCF was performed using an electrodeposition technique. Furthermore, the effects of increasing Pt-loading on the catalytic activity were studied by varying the electrodeposition time (t_ed). Finally, the ORR activity of the Pt-coated CNT samples was compared with that of the CNTCF support.

2. Experimental

2.1. Synthesis of Pt-CNTCF

The hierarchical catalyst layer was synthesized using polyacrylonitrile (PAN)-based CF as the starting material, which served as the substrate on which to grow the CNTs. The CNTs, as well as the Pt-nanoparticles, were synthesized directly on their respective substrates, namely the CF and CNTCF. Initially, the CF was heat treated at 450 °C for 10 min to remove the polymeric sizing agent. The Ni catalyst coating onto the heat treated CF was performed using an electroless process according to a procedure mentioned elsewhere.⁴¹ In brief, the heat treated CF was dipped for 10 min in an alkaline bath held at 80 °C, where the coating bath consisted of nickel sulphate (30 g l⁻¹), sodium hypophosphite (12 g l⁻¹), ammonium chloride (50 g l⁻¹), trisodium citrate (40 g l⁻¹) and liquor ammonia (30% NH₃ to adjust the pH to 8). Before further processing, the catalyst coated CF was washed several times with deionized (DI) water and dried at ∼60 °C for 2 h. Subsequently, CNTs were grown on the catalyst-coated CF by the catalytic decomposition of C₂H₂ in a N₂ atmosphere using CVD. The C₂H₂ flow rate, reaction temperature and growth duration during the CVD process were taken as 90 ml min⁻¹, 800 °C and 15 min, respectively. Then, the as-synthesized CNTCF was treated successively with 5 M HNO₃ and 0.5 M H₂SO₄ for 12 h each to remove residual Ni particles from the CNT tips.⁴² Finally, to synthesize the Pt coated CNTCF (Pt-CNTCF), Pt nanoparticles were deposited onto the acid-treated CNTCF through an electrodeposition method with a two-electrode setup using CNTCF and Pt wire as the working and counter electrodes, respectively. A coating bath containing H₂PtCl₆·6H₂O (1 g l⁻¹) in 0.5 M H₂SO₄ aqueous solution was used as the electrolyte. Electrodeposition was performed at an applied potential of −2.0 V at t_ed values of 5, 10, 15, 20, 25 and 30 min to synthesize the Pt-CNT-05, Pt-CNT-10, Pt-CNT-15, Pt-CNT-20, Pt-CNT-25 and Pt-CNT-30 samples, respectively. The parameters for electrodeposition were selected so that the Pt-loading varied gradually with time. For example, large platinum particles were obtained for electrodeposition at −3 V (Fig. S1†), which possibly may not be interesting due to the high Pt-loading.

2.2. Characterization of Pt-CNTCF

Scanning electron microscopy (SEM, EVO MA 15, Zeiss) and field emission SEM (FESEM, JEOL JSM-7100F) were employed to study the Pt-cluster size distributions and surface morphologies of the Pt-CNTCF samples. Elemental analysis was performed using energy dispersive X-ray (EDX) spectroscopy on a FEI Quanta 200 SEM. The evolution of the crystalline size was studied using X-ray diffraction (XRD; Rigaku Miniflex 600 with Cu K_α (λ = 1.5418 Å) radiation) studies. Additionally, Raman spectroscopy was performed using a Horiba Jobin Yvon LabRAM HR Raman microscope equipped with a laser excitation source of 632.7 nm.

Cyclic voltammetry (CV) measurements were performed to study the catalytic performance in terms of the ORR activity and electrochemical surface area (ESA) using a three-electrode setup attached to an Agilent 5500 AFM. A Pt-CNTCF modified glassy carbon (GC) electrode (φ = 3 mm) was used as the working electrode, while Ag/AgCl (1 M KCl) and Pt-wire electrodes served as reference and counter electrodes, respectively. To prepare the working electrode, 10 mg of Pt-CNTCF was dispersed in 2 ml of C₂H₅OH via ultrasonication for 1 h and 5 μl of the dispersion was drop-casted onto a cleaned GC electrode and dried for 1 h. The electrocatalytic activity towards the ORR was measured by CV in O₂ saturated 0.1 M KOH with a potential scan ranging from −1.0 to 0.2 V at a scan rate of 50 mV s⁻¹. Similarly, the ESA was calculated from the hydrogen desorption peak of the CV curve in 1.0 M H₂SO₄ with a potential scan ranging from −0.1 to 0.8 V at a scan rate of 50 mV s⁻¹. Similarly, linear sweep voltammetry (LSV) was performed using a home-made rotation disk electrode (RDE) setup.^43,44 The amount of Pt loaded on the Pt-CNTCF modified working electrode was estimated by thermogravimetric analysis (TGA) using a Perkin-Elmer Diamond TG/DTA analyzer. TGA of the Pt-CNTCF dispersion (250 μl) used to prepare the modified GC electrode was performed in an O₂ environment (flow rate: 200 ml min⁻¹) from 40 to 750 °C at a heating rate of 10 °C min⁻¹.

3. Results and discussion

3.1. Structural hierarchy of Pt-CNTCF

As shown in Fig. 1, the multiscale hierarchical structure of the catalyst layer consists of a mat of CNTCF over which, Pt-nanoparticles were electrodeposited to form Pt coated CNTCF (Pt-CNTCF). The structure possesses a three-level hierarchy at various length scales. At the microscale, a single CF strand consists of thousands of microsized fibers. Similarly, at the submicroscale, each fiber is coated with Pt-coated CNTs, while at the nanoscale, Pt-nanoparticle clusters are coated onto the CNTs. The unique structure may provide improved interactions between the CNT–CF as well as CNT–Pt interfaces, and hence better electronic connectivities between the catalyst particles within the catalyst layer. Thus, electrons transported from/to the catalyst particles can find easy paths through the microsized CF.

Fig. 1 Schematic of the hierarchically structured catalyst layer consisting of Pt-CNTCF at different length scales.

3.2. Structural characterization

Fig. 2a shows the SEM image of the electroless Ni-coated CF, revealing the deposition of Ni in the form of particulate clusters with diameters of ∼120 nm. The XRD pattern of the Ni coated CF shown in Fig. 2b exhibits diffraction peaks at ∼26, ∼45 and 52°, which correspond to graphitic C(002) (JCPDS 41-1487), Ni(111) and Ni(200) (JCPDS-87-0712), respectively. Elemental analysis of the catalyst sample by EDX (Fig. 2c) shows the presence of Ni (∼15 at%) and P (∼1 at%), suggesting the presence of nickel phosphides (NiP, NiP₂, etc.) in small fractions along with Ni. The SEM image of the as-synthesized CNTCF (Fig. 2d) shows a dense CNT coating on the CF surface with no uncovered CF surface being observable. Ni particles are visible as bright spots on the CNT tips, as shown in the inset of Fig. 2d, which are removed during acid-treatment. Hence, the acid-treated CNTCF consists of a CF surface covered by catalyst-free CNTs with diameters ranging between ∼150 to 200 nm (Fig. 2e). The surfaces of the acid-treated CNTs exhibit rough structures due to the presence of groove-like structures that are possibly formed due to the partial corrosion of the outer CNT walls during the acid-treatment. The removal of the acid-unstable parts of the CNTs before Pt coating is favorable for achieving a catalyst support with high durability. Moreover, the rough corroded surface may also provide active sites for Pt electrodeposition. The Raman spectra of as-synthesized and acid-treated CNTs (Fig. 2f) exhibit a graphitic nature with the presence of intense D-bands, attributed to the highly defective nature of the CNTs. Additionally, the Raman D-band to G-band intensity ratio (I_D/I_G) decreases from 1.53 for the as-synthesized CNTs to 1.17 for the acid-treated CNTs, which is contrary to the fact that the number of defects in the CNTs increases with acid-treatment.⁴⁵ This may be attributed to the decreased number of defects due to the removal of highly defective/amorphous carbon during acid-treatment, which dominates over the concurrent increase due to the degradation of the outer CNT walls.

Fig. 2 (a) SEM image, (b) XRD pattern and (c) EDX spectrum of the catalyst-coated CF. (d) SEM image of CNTCF showing a dense CNT forest grown on the CF. The inset of (d) shows Ni nanoparticles on the CNT tips. (e) Higher magnification SEM image of the acid-treated CNTCF. (f) Raman spectra of the as-synthesized and acid-treated CNTs. Typical Voigt fits to the Raman D- and G-bands of the as-synthesized CNTs are shown by the dotted curves.

3.3. Pt electrodeposition on the CNTCF

SEM and FESEM images of Pt-CNTCF are shown in Fig. 3. Individual CFs are covered with a thick (∼10 μm) CNT forest, while Pt-nanoparticle clusters (∼100 nm) are coated uniformly onto the CNTs. Higher magnification FESEM image of the Pt coated CNTs shown in the inset of Fig. 3b reveals the presence of individual Pt-nanoparticles of ∼10 nm average diameter in the cluster. The deposition of such Pt-clusters onto CNTs is possibly due to the presence of the large diameter (∼100 nm; of the order of CNT diameter) active sites on the CNT surface formed during acid-treatment, where the nucleation of multiple particles takes place. The presence of Pt is confirmed by EDX analysis, which reveals the increasing weight fraction of Pt from ∼0 to ∼0.3 with increasing t_ed from 5 to 30 min (ESI; Fig. S2†).

Fig. 3 SEM images of Pt-CNTCF for a t_ed of 20 min showing (a) a single CF covered with a thick forest of Pt-coated CNTs and (b) a coating of Pt-nanoparticle clusters on the CNTs. The inset of (b) shows the FESEM image of a Pt cluster revealing the individual Pt-nanoparticles.

The Pt-cluster size distributions of Pt-CNTCF for varying t_ed are shown in Fig. S3 (ESI†). To obtain the Pt-cluster size distribution of a typical Pt-CNTCF sample, the areas of individual Pt-clusters visible in the SEM image were approximated by the areas of ellipses encircling them. Significant increments in the cluster density (number of clusters/area) were observed with increasing t_ed, suggesting the continuous formation of new clusters as well as the growth of the individual particles within a cluster during the electrodeposition process.

Fig. 4a shows the thermal drift corrected TGA of the Pt-CNTCF dispersion (250 μl), performed in an O₂ atmosphere at a heating rate of 10 °C min⁻¹. A correction for thermal drift was performed using the TGA of an empty sample pan under similar conditions and the weight drift during heating was recorded. The final thermal drift calculated by fitting the data with a polynomial of order three was used to correct the TGA data (inset; Fig. 4a). The residual weights of the drift corrected TGA plots, which exhibit an increasing trend with t_ed, are considered as the net weight of Pt-nanoparticles in a 250 μl dispersion. The variations in the Pt-cluster size (D_Pt), % area coverage and Pt loading on the GC/electrode area (w_Pt), with t_ed are shown in Fig. 4b. Increasing t_ed from 10 to 30 min, caused a large increment in the value of D_Pt from ∼70 to ∼170 nm. However, for t_ed values below 20 min, D_Pt increased marginally due to the combined effects of the growth of existing clusters and the formation of new ones. Again, on increasing t_ed from 05 to 30 min, the Pt-loading increased from 3.8 to 15.7 μg cm⁻², while the fractional area coverage changed from ∼0.5 to ∼15%. For higher t_ed values, the area coverage neared saturation due to the overlap of individual clusters.

Fig. 4 (a) TGA of the Pt-CNTCF dispersion (250 μl) corrected for thermal drift using the empty pan TGA data shown in the inset. (b) Variations in average Pt-cluster size, % area coverage and Pt loading on the GC/electrode area (w_Pt), with deposition time t_ed.

The variations in Pt-loading and area coverage with t_ed exhibit complex behavior, revealing the rate of deposition to be highest between a t_ed of 10 to 15 min, followed by saturation. This may be attributed to the combined effects of the nucleation and growth of the Pt-nanoparticles and the decreasing Pt-ion concentration in the coating bath with increasing t_ed. Initially, the rate of deposition remained low during the incubation period, while it increased with t_ed due to particle growth, until the Pt-ion concentration was depleted. Similarly, for t_ed values below 20 min, D_Pt increased marginally due to the combined effects of the growth of existing clusters and the formation of new ones. For higher t_ed values, the D_Pt exhibited fast growth due to the overlap of individual clusters.

A typical XRD pattern of Pt-CNTCF for a t_ed of 30 min is shown in Fig. 5. The presence of a diffraction peak at a 2θ value of ∼26° corresponding to the (002) plane of graphite (JCPDS 41-1487) reveals the graphitic structure of the CNTCF substrate. Again, the presence of diffraction peaks at 2θ values of ∼40, ∼47, ∼68 and ∼81° corresponding respectively to the Pt(111), Pt(200), Pt(220) and Pt(311) planes (JCPDS 87-0642) reveal the coating of polycrystalline Pt-nanoparticles. The XRD patterns in the inset of Fig. 5 show the variation in the intensity of the Pt(111) and Pt(200) peaks with t_ed. With increasing t_ed, the intensities of the individual diffraction peaks increase due to the larger number of Pt-crystallites. However, the area ratios of the Pt(111) and Pt(200) peaks exhibit no significant variation, revealing similar crystallographic orientation for all the Pt-CNTCF samples. Similarly, the Pt-crystalline size also remains constant at ∼10 nm, which suggests a continuous nucleation of the Pt-particles during electrodeposition.

Fig. 5 XRD pattern of Pt-CNTCF for a t_ed of 30 min. The inset shows the variation in the Pt(111) and Pt(200) peaks with t_ed from 5 to 30 min.

3.4. Electrocatalytic activity of Pt-CNTCF

The electrocatalytic performances of the Pt-CNTCF samples synthesized at varying t_ed values were studied in terms of the specific electrochemical surface area (ESA) and ORR activity in alkaline media. Both hydrogen underpotential deposition (H upd) and copper upd (Cu upd) were performed to study the ESA variation. However, as the Cu stripping peak of Pt coincides with that of the CNTCF support (Fig. S4†), the ESA was calculated by H upd using CV in Ar saturated 1.0 M H₂SO₄ at a scan rate of 50 mV s⁻¹. Fig. 6a shows the evolution of the hydrogen desorption (H_des) peaks of the Pt-CNTCF modified GC electrodes with t_ed. Since the integrated area of the H_des peak corresponds to the total charge associated with H⁺ desorption (Q_H), assuming the adsorption of a H⁺ monolayer on Pt with a one to one H–Pt correspondence, the ESA is given by: ESA [cm² g⁻¹] = Q_H [μC]/(210 × w_Pt [g]),⁴⁶ where 210 μC cm⁻² is the charge per cm² of hydrogen monolayer adsorbed onto the polycrystalline Pt,⁴⁷ while w_Pt is the Pt-loading on the electrode, determined from the TGA data. Again, Q_H is calculated from the H_des peak of the CV curve using the relation: Q_H = (1/scan rate)∫idV, where ∫idV is the area of the hydrogen desorption peak assumed to be in between 0.05 to −0.1 V (vs. Ag/AgCl). The variation in the ESA with t_ed is shown in Fig. 6b. The low ESA for a t_ed of 5 min may be attributed to the lower Pt-loading, at which a large fraction of the Pt-nanoparticles may remain obscured by the CNTs when the CNT dispersion is drop-casted onto the electrode. Again, the ESA increases monotonously for t_ed values of 10–25 min due to the increased number of Pt-nanoparticle clusters, while it decreases slightly at a higher t_ed due to the overlap of the clusters, which reduces the effective surface area. The maximum value for the ESA was observed to be ∼82 m² g⁻¹ for a t_ed of 25 min.

Fig. 6 (a) CV scans of Pt-CNTCF in Ar saturated 1.0 M H₂SO₄ at a scan rate of 50 mV s⁻¹. (b) The variation of the ESA with t_ed.

The ORR activity of Pt-CNTCF was studied in alkaline media through CV (stationary electrodes) and LSV (RDE; 1600 rpm) measurements in an O₂ saturated 0.1 M KOH electrolyte using a Pt-CNTCF modified GC working electrode. The CV scans of the Pt-CNTCF electrodes in the O₂ saturated electrolyte at a potential scan rate of 50 mV s⁻¹ are shown in Fig. 7, where a cathodic peak at ∼−0.4 V (vs. Ag/AgCl) is observed. The cathodic peak, which disappears in the Ar saturated electrolyte (ESI; Fig. S5†), corresponds to the ORR. The variation in the ORR performance of the Pt-CNTCF samples with varying t_ed can be analyzed qualitatively in terms of the intensity of the ORR peak. As can be inferred from Fig. 7, the ORR peak intensities remain high for a t_ed value of 10 to 20 min, which can be attributed to the increasing number and uniform dispersion of agglomeration-free Pt-nanoparticles. With further increases in the t_ed beyond 20 min, despite the increasing w_Pt, the ORR peak intensities reduce due to the reduction in the effective surface area from the overlapping Pt-clusters.

Fig. 7 CV scans of the Pt-CNTCF samples in O₂ saturated 0.1 M KOH at a scan rate of 50 mV s⁻¹.

The variation in the ORR activity with t_ed was further studied through LSV using a Pt-CNTCF modified GC working RDE (1600 rpm) in 0.1 M KOH at a potential scan rate of 10 mV s⁻¹ (Fig. 8a). The corresponding Tafel plots are shown in Fig. 8b. As defective CNTs are known to show catalytic activity in the ORR,⁴⁰ LSV corresponding to the CNTCF (without a Pt coating) has also been incorporated. The CNTCF modified GC electrode was prepared through a route similar to that adopted for the preparation of the Pt-CNTCF modified GC electrode. As can be seen from Fig. 8a, the CNTCF modified GC electrode exhibits significant ORR current and hence contributes to the overall ORR activity, due to the defective structure of the CNTs (Fig. 2f; Raman spectroscopy). The variation in the ORR activities of the Pt-CNTCF samples with the t_ed was evaluated in terms of the ORR onset potential (V_Onset), ORR current (|J_ORR|), and Tafel slope, with high ORR activity being characterized by a high |J_ORR|, more positive value of V_Onset, and lower Tafel slope.^43,48 The parameter V_Onset is estimated from the intersection point of the tangents to the LSV curve in the proximity of the ORR peak, while the |J_ORR| is taken as the current density at a potential of −0.3 V. The variation in the |J_ORR| shown in Fig. 8c suggests low ORR activity for a t_ed of 5 min, which increased with increasing t_ed to reach a saturation for t_ed values above 15 min. For example, the |J_ORR| exhibited a two-fold increment on increasing the t_ed from 5 to 15 min. Again, the variations in the V_Onset and Tafel slope with t_ed shown in Fig. 8d exhibit similar trends with respective shifts from ∼−190 to ∼−75 mV and from 190 to 120 mV per decade on increasing the t_ed from 5 to 20 min. Compared to Pt-CNTCF, the CNTCF exhibited a more negative V_Onset of ∼−200 mV and a larger Tafel slope of ∼240 mV per decade. The large variations in the V_Onset and Tafel slope with the t_ed may be attributed to the varying relative contributions of the CNT support and supported Pt-nanoparticles to the overall ORR current, where the ORR current from the CNT support dominates at lower t_ed. The variations in the V_Onset and Tafel slope parameters suggest a different nature of the ORR active sites and hence, different ORR mechanisms for the CNT support and the supported Pt.

Fig. 8 (a) LSV curves of the Pt-CNTCF modified GC RDEs (1600 rpm) in O₂ saturated 0.1 M KOH at a potential scan rate of 10 mV s⁻¹. (b) Tafel plots obtained from (a) with solid lines showing linear fits to data. (c) The variation in the ORR current at −0.3 V (|J_ORR|) with t_ed. (d) The variations in the ORR onset potentials (V_Onset) and Tafel slopes.

LSV curves at different rotation rates were obtained for the Pt-disk (2 mm diameter), Pt-CNTCF modified and CNT modified GC RDEs (Fig. 9a–c). Furthermore, to understand the nature of the ORR active sites, the measured disk current densities (j_d) were applied to the Koutecky–Levich (KL) equation (eqn (1)).⁴⁹

where j_k and j_diff are the kinetic and diffusion-limiting current densities, ω is the angular velocity of the RDE and B is a constant dependent on the concentration and diffusion coefficient of O₂, the kinematic viscosity of the electrolyte and the electron transfer number (n) through eqn (2):

B = 0.62nFC₀D₀^2/3v^−1/6 = nA, with A = 0.62FC₀D₀^2/3v^−1/6 (2)

Fig. 9 (a–c) RDE voltammograms at varying rotation rates and KL plots for the (a) Pt-disk RDE, (b) the CNT support GC RDE and (c) the Pt-CNTCF modified GC RDE. (d–e) The KL plots corresponding to the (d) Pt-disk, (e) CNT support and (f) Pt-CNTCF.

For the present RDE setup, the value of A was obtained from the averaged slope of the KL plots (j_d⁻¹ vs. ω^−0.5; Fig. 9d) at −0.45 and −0.50 V for the Pt RDE, assuming n = 4 for Pt. Similarly, the n values for Pt-CNTCF (t_ed = 20 min), as well as the CNT modified GC RDE, were obtained from the slopes of corresponding KL plots (Fig. 9e and f). The values of n for bare CNTs and Pt-CNTCF were calculated to be 2.57 and 3.53, respectively. This suggests the dominance of a two-electron pathway in the ORR on the CNT support, which reduces the overall n value for Pt-CNTCF.

Since the electrodeposition process depends not only on the t_ed but also on other parameters such as the temperature, applied potential, electrolyte composition and concentration, etc., w_Pt has been considered as the key parameter affecting the electrocatalytic activity of Pt-CNTCF, instead of t_ed.

Fig. 10 depicts the variations in the electrochemical activity of Pt-CNTCF, in terms of the parameters |J_ORR|, V_Onset and Tafel slope, with respect to Pt-loading and Pt-cluster size (D_Pt). Increasing the Pt-loading from ∼0.2 to 0.6 μg, increased the ORR activity of the Pt-CNTCF monotonously, while it remained practically constant for higher Pt-loadings (0.6–1.2 μg). Similarly, the variations in the parameters |J_ORR|, V_Onset and Tafel slope with D_Pt are exhibited in Fig. 10c and d. On increasing D_Pt beyond 100 nm, the ORR parameters exhibited constant values with no significant dependence on D_Pt. This is attributed to the reduction of the ESA with the increase in Pt-loading. The saturation of the ORR activities of the Pt-CNTCF samples with higher loadings of Pt is contrary to the known behavior of Pt-based ORR electrocatalysts, which have a lower overpotential at higher Pt-loading.⁵⁰ This can be understood by the fact that the ESA reduced from 82 to 64 m² g⁻¹ with an increase in the Pt-loading from ∼14 to ∼16 μg cm⁻² due to the increased overlap of the Pt-clusters. Hence, Pt-CNTCF with a high Pt-loading and agglomeration-free Pt-clusters can provide the highest ORR performance. Moreover, as the Pt-ion concentration of the coating bath decreases with time during electrodeposition, the growth of Pt-nanoparticles with different surface crystallographic orientations (hkl) could be possible. Such a preferred orientation may affect the ORR performance significantly.⁵¹ However, this possibility is discarded as the XRD analysis reveals no such preferred orientation, confirming that the variation in the ORR performance can be attributed solely to the variation in Pt-loading and the extent of Pt-cluster overlap.

Fig. 10 (a and b) The variations in the (a) |J_ORR| and (b) V_Onset and Tafel slope parameters with Pt-loading/area on the Pt-CNTCF modified GC electrode. (c and d) The correlations between the cluster size and the (a) |J_ORR| at −0.3 V and (b) V_Onset and Tafel slope ORR parameters.

Finally, the electrochemical stability of the Pt-CNTCF modified GC electrode (t_ed = 60 min) was investigated by potential cycling between −1.0 to 0.2 V in air saturated 0.1 M KOH electrolyte at a scan rate of 50 mV s⁻¹ and the area under the hydrogen desorption peak (forward scan) between −0.80 and 0.55 V was measured. A comparatively high Pt loading (t_ed = 60 min) was used for the stability study to obtain a measurable peak area. For the initial 100 cycles, the area shows a small (<2%) decrease. The Pt-CNTCF electrocatalyst exhibits a high stability as the peak area shows a reduction of <2% after the first 100 potential cycles (Fig. 11). Hence, Pt-nanoparticles supported on defective CNTs may be used to fabricate cathode catalyst layers with high stability, low Pt-loading and high ORR activity.

Fig. 11 Electrochemical stability of the Pt-CNTCF (t_ed = 60 min) modified GC electrode in air saturated 0.1 M KOH at a scan rate of 50 mV s⁻¹.

The hierarchically structured CNTCF supported Pt-based catalyst layer exhibits a high ORR performance in terms of both the ORR and ESA. Table 1 compares the ESA and ORR performance of the present catalyst with those of a few previously reported ones. The ORR performance of the various catalysts has been compared in terms of the ORR current (I_ORR; A g_Pt⁻¹) at −0.3 V obtained from RDE measurements under similar conditions (0.1 M KOH, Ag/AgCl reference). The present catalyst exhibits superior ORR performance (>2×) in terms of both the ESA as well as the I_ORR. Hence, catalyst layers with high ORR performance and low-Pt-loading can be fabricated using defective CNT coated CF as the catalyst support.

Table 1 Comparison of the I_ORR and ESA values with the literature

Parameter	Literature value	Present work
a IORR = current density at −0.3 V from RDE plots (A cm^−2)/Pt loading on the RDE (gPt cm^−2).
IORR at −0.3^a V (A gPt^−1)	Pt/C: 47 (ref. 52)	90–115
ESA (m^2 g^−1)	Pt/CNT/CP: 52 (ref. 30)	70–80
	Pt/CNx: 55.6; Pt/CNT: 40.9 (ref. 37)
	Pt/Pt–Ni alloy: 33–62 (ref. 53)
	PtxIry/MWCNT: 48–98 (ref. 54)
	Functionalized Pt/CNT: 71 (ref. 55)
	Pt/SWCNT/CNF: ∼39–43 (ref. 56)

---

However, the low electron transfer number (n = 2.57) of the ORR active CNT support suggests the production of H₂O₂ by reduction of O₂. As H₂O₂ is known to cause damage to perfluorosulfonated electrolyte membranes,⁵⁷ the use of such defective carbon supported, low Pt-loading catalyst layers in PMEFCs may decrease their durability. Additionally, as the facile synthesis of defect-free CNTs is difficult, the production of small amounts of H₂O₂ during the operation of fuel cells based on carbon supported cathode catalyst layers is inevitable. Hence, at low Pt-loadings, where the ORR current from the catalyst support is substantial, the defective nature of the CNTs and other carbonaceous catalyst supports may not be suitable, particularly for PEMFCs.

4. Conclusions

In conclusion, hierarchically structured catalyst layers were synthesized using CNTCF as the substrate, on which Pt-nanoparticle clusters were deposited by an electrodeposition process. Clusters of Pt-nanoparticles with an average particle diameter of ∼10 nm and varying cluster size (∼80 to ∼170 nm) were grown on CNTCF by varying the deposition time. The formation of clusters was attributed to the nucleation of the Pt nanoparticles only at preferred active sites on the CNT surfaces. With increasing deposition time, although the Pt-loading on the CNTCF increased nearly linearly, the ORR performance of Pt-CNTCF attained saturation after the initial increase, due to the overlap of Pt-clusters. Furthermore, the CNTs used for the Pt-support also exhibited significant ORR activity due to their defective nature. The overall ORR activity of Pt-CNTCF was enhanced synergistically by the presence of Pt-nanoparticle clusters on the defective CNT surface. The synthesized hierarchical Pt-CNTCF can be used to fabricate PEMFC/AFC catalyst layers with high performance and low Pt-loading. However, a reduction of the Pt-loading below a certain limit may reduce the durability of polymer electrolytes due to the production of substantial amounts of H₂O₂ on the ORR active CNT supports (n = 2.57).

Acknowledgements

This work was financially supported by the Department of Science and Technology, Government of India, Mission on Nano Science and Technology (Nano Mission).

References

1. S. Litster and G. McLean, J. Power Sources, 2004, 130, 61–76.
2. Y. Shao, G. Yin and Y. Gao, J. Power Sources, 2007, 171, 558–566.
3. K. Gong, F. Du, Z. Xia, M. Durstock and L. Dai, Science, 2009, 323, 760–764.
4. S. Pylypenko, S. Mukherjee, T. S. Olson and P. Atanassov, Electrochim. Acta, 2008, 53, 7875–7883.
5. H. Meng and P. K. Shen, Electrochem. Commun., 2006, 8, 588–594.
6. B. Li and S. H. Chan, Int. J. Hydrogen Energy, 2013, 38, 3338–3345.
7. Z. Qi and A. Kaufman, J. Power Sources, 2003, 113, 37–43.
8. A. Morozan, B. Jousselme and S. Palacin, Energy Environ. Sci., 2011, 4, 1238–1254.
9. R. R. Bacsa, C. Laurent, A. Peigney, W. S. Bacsa, T. Vaugien and A. Rousset, Chem. Phys. Lett., 2000, 323, 566–571.
10. E. A. D. Morais, G. Alvial, R. Longuinhos, J. M. A. Figueiredo, R. G. Lacerda, A. S. Ferlauto and L. O. Ladeira, Mater. Res., 2011, 14, 403–407.
11. J. Cherusseri and K. K. Kar, RSC Adv., 2015, 5, 34335–34341.
12. Y. Shao, G. Yin, J. Zhang and Y. Gao, Electrochim. Acta, 2006, 51, 5853–5857.
13. T. W. Ebbesen, H. J. Lezec, H. Hiura, J. W. Bennett, H. F. Ghaemi and T. Thio, Nature, 1996, 382, 54–56.
14. E. Pop, D. Mann, Q. Wang, K. Goodson and H. Dai, Nano Lett., 2005, 6, 96–100.
15. C. Hu, Y. Guo, Y. Cao, L. Yang, Z. Bai, K. Wang, P. Xu and J. Zhou, Mater. Sci. Eng., B, 2011, 176, 1467–1473.
16. M. Sahoo, B. P. Vinayan and S. Ramaprabhu, RSC Adv., 2014, 4, 26140–26148.
17. W. Yuan, S. Lu, Y. Xiang and S. P. Jiang, RSC Adv., 2014, 4, 46265–46284.
18. W. Zhang, P. Sherrell, A. I. Minett, J. M. Razal and J. Chen, Energy Environ. Sci., 2010, 3, 1286–1293.
19. I. Nitta, O. Himanen and M. Mikkola, Electrochem. Commun., 2008, 10, 47–51.
20. F. Maillard, S. Schreier, M. Hanzlik, E. R. Savinova, S. Weinkauf and U. Stimming, Phys. Chem. Chem. Phys., 2005, 7, 385–393.
21. S. Kamarajugadda and S. Mazumder, J. Power Sources, 2008, 183, 629–642.
22. H. Yano, J. Inukai, H. Uchida, M. Watanabe, P. K. Babu, T. Kobayashi, J. H. Chung, E. Oldfield and A. Wieckowski, Phys. Chem. Chem. Phys., 2006, 8, 4932–4939.
23. Z. He, J. Chen, D. Liu, H. Tang, W. Deng and Y. Kuang, Mater. Chem. Phys., 2004, 85, 396–401.
24. H. Tang, J. Chen, S. Yao, L. Nie, Y. Kuang, Z. Huang, D. Wang and Z. Ren, Mater. Chem. Phys., 2005, 92, 548–553.
25. S. Li, H. Liu, Y. Wang, W. Xu, J. Li, Y. Liu, X. Guo and Y. Song, RSC Adv., 2015, 5, 8787–8792.
26. L. Jiang, A. Hsu, D. Chu and R. Chen, J. Electrochem. Soc., 2009, 156, B643–B649.
27. G. Z. Hu, F. Nitze, X. Jia, T. Sharifi, H. R. Barzegar, E. Gracia-Espino and T. Wagberg, RSC Adv., 2014, 4, 676–682.
28. K. Yoshii, T. Tsuda, T. Arimura, A. Imanishi, T. Torimoto and S. Kuwabata, RSC Adv., 2012, 2, 8262–8264.
29. W. Xiong, F. Du, Y. Liu, A. Perez, M. Supp, T. S. Ramakrishnan, L. Dai and L. Jiang, J. Am. Chem. Soc., 2010, 132, 15839–15841.
30. S. Takenaka, H. Matsumori, H. Matsune, E. Tanabe and M. Kishida, J. Electrochem. Soc., 2008, 155, B929–B936.
31. C. H. Wang, H. Y. Du, Y. T. Tsai, C. P. Chen, C. J. Huang, L. C. Chen, K. H. Chen and H. C. Shih, J. Power Sources, 2007, 171, 55–62.
32. C.-T. Hsieh, W.-M. Hung and W.-Y. Chen, Int. J. Hydrogen Energy, 2010, 35, 8425–8432.
33. P. H. Maheshwari and R. B. Mathur, Electrochim. Acta, 2009, 54, 7476–7482.
34. K. K. Kar, S. Rana and J. K. Pandey, Handbook of polymer nanocomposites. Processing, performance and application, Springer, Berlin Heidelberg, 2015.
35. D. C. Higgins, F. M. Hassan, M. H. Seo, J. Y. Choi, M. A. Hoque, D. U. Lee and Z. Chen, J. Mater. Chem. A, 2015, 3, 6340–6350.
36. Y. Chen, J. Wang, H. Liu, M. N. Banis, R. Li, X. Sun, T.-K. Sham, S. Ye and S. Knights, J. Phys. Chem. C, 2011, 115, 3769–3776.
37. M. S. Saha, R. Li, X. Sun and S. Ye, Electrochem. Commun., 2009, 11, 438–441.
38. D. C. Higgins, D. Meza and Z. Chen, J. Phys. Chem. C, 2010, 114, 21982–21988.
39. R. Chetty, S. Kundu, W. Xia, M. Bron, W. Schuhmann, V. Chirila, W. Brandl, T. Reinecke and M. Muhler, Electrochim. Acta, 2009, 54, 4208–4215.
40. M. Liu and J. Li, Electrochim. Acta, 2015, 154, 177–183.
41. R. Sharma and K. K. Kar, Mater. Lett., 2014, 137, 150–152.
42. H. Hu, B. Zhao, M. E. Itkis and R. C. Haddon, J. Phys. Chem. B, 2003, 107, 13838–13842.
43. R. Sharma and K. K. Kar, Electrochim. Acta, 2015, 156, 199–206.
44. R. Sharma and K. K. Kar, J. Mater. Chem. A, 2015, 3, 11948–11959.
45. M. Saravanan, P. Sennu, M. Ganesan and S. Ambalavanan, J. Electrochem. Soc., 2013, 160, A70–A76.
46. B. Seger and P. V. Kamat, J. Phys. Chem. C, 2009, 113, 7990–7995.
47. S. Trasatti and O. Petrii, Pure Appl. Chem., 1991, 63, 711–734.
48. D. Geng, Y. Chen, Y. Chen, Y. Li, R. Li, X. Sun, S. Ye and S. Knights, Energy Environ. Sci., 2011, 4, 760–764.
49. C. Wang, H. Daimon, T. Onodera, T. Koda and S. Sun, Angew. Chem., Int. Ed., 2008, 47, 3588–3591.
50. H. A. Gasteiger, J. E. Panels and S. G. Yan, J. Power Sources, 2004, 127, 162–171.
51. N. M. Markovic, H. A. Gasteiger and P. N. Ross, J. Phys. Chem., 1996, 100, 6715–6721.
52. W.-J. Jiang, J.-S. Hu, X. Zhang, Y. Jiang, B.-B. Yu, Z.-D. Wei and L.-J. Wan, J. Mater. Chem. A, 2014, 2, 10154–10160.
53. J. Wu, J. Zhang, Z. Peng, S. Yang, F. T. Wagner and H. Yang, J. Am. Chem. Soc., 2010, 132, 4984–4985.
54. J. Yuan, B. He, L. Hong, J. Lu, J. Miao and L. Niu, J. Mater. Chem., 2012, 22, 19658–19665.
55. S. Zhang, Y. Shao, G. Yin and Y. Lin, J. Mater. Chem., 2010, 20, 2826–2830.
56. W. Zhu, D. Ku, J. P. Zheng, Z. Liang, B. Wang, C. Zhang, S. Walsh, G. Au and E. J. Plichta, Electrochim. Acta, 2010, 55, 2555–2560.
57. R. Borup, J. Meyers, B. Pivovar, Y. S. Kim, R. Mukundan, N. Garland, D. Myers, M. Wilson, F. Garzon and D. Wood, Chem. Rev., 2007, 107, 3904–3951.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra13866c

---