Preparation of highly active and stable polyaniline-cobalt-carbon nanotube electrocatalyst for oxygen reduction reaction in polymer electrolyte membrane fuel cell

Electrochimica Acta 119 (2014) 144–154

Contents lists available at ScienceDirect

Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta

Preparation of highly active and stable polyaniline-cobalt-carbon nanotube electrocatalyst for oxygen reduction reaction in polymer electrolyte membrane fuel cell Zhong-shu Yin a , Tian-hang Hu a , Jian-long Wang a , Cheng Wang a , Zhi-xiang Liu b , Jian-wei Guo a,∗ a b

Institute of Nuclear and New Energy Technology, Tsinghua University, Beijing 100084, PR China School of Electrical Engineering, Southwest Jiaotong University, Chengdu 610031, PR China

a r t i c l e

i n f o

Article history: Received 6 September 2013 Received in revised form 13 December 2013 Accepted 13 December 2013 Available online 27 December 2013 Keywords: Oxygen reduction reaction Non-noble-metal electrocatalyst Carbon nanotube Citrate acid In-situ synthesis

a b s t r a c t This paper established an in-situ synthesis strategy that the mixing solution of aniline, CNTs and CoCl2 was directly reduced to prepare polyaniline-cobalt-carbon nanotube (PANI-Co-CNT) electrocatalyst. Furthermore, this strategy was effectively modified by pretreating CoCl2 precursor with citric acid (CA), forming 2-4 nm cobalt nanoparticles uniformly distributed on PANI-CNT support with porous structure. The control experiments revealed various PANI states in the growth stage, further proposing the self-assembly mechanisms in these two routes with and without CA pretreatment. These two PANI-Co-CNT electrocatalysts were also checked by oxygen reduction reaction (ORR) in acid environment, to corroborate their basically 4-electron processes. Inspiringly, the large activity and stability for the pretreated route could be comparable with those of the advanced electrocatalysts. All these progresses lay a bottom-up approach for future electrocatalysts. Crown Copyright © 2013 Published by Elsevier Ltd. All rights reserved.

1. Introduction The polymer electrolyte membrane fuel cells (PEMFCs) are highly regarded as energy converting systems for power fields in stationary and mobile devices [1–3], but their low kinetics in oxygen reduction reaction (ORR) is a major limiting factor especially in acid media [4]. Up to now, platinum (Pt)-based catalysts show remarkable improvements in activity and stability [5–9] though these progresses cannot yet meet future targets. Moreover, Pt is scarce in reserve and its cost is too expensive to meet large applications, so it is still urgent to prepare non-noble-metal electrocatalysts with high activity and stability towards ORR. As a promising candidate, cobalt has long been focused on its macrocycles, oxides, chalcogenides and carbon-based structures [10–14], whose activity and stability cannot be achieved simultaneously. It is expected these drawbacks can be improved by loading these structures on supports, which is a general strategy for most electrocatalysts. As an ideal support, carbon nanotube (CNT) is notable for its excellent properties [15,16] in high

∗ Corresponding author. Institute of Nuclear and New Energy Technology, Beijing 100084, PR China. Tel.: +86 1080194009; fax: +86 1080194009. E-mail address: [email protected] (J.-w. Guo).

conductivity, large surface area and high stability [17]. If CNTs are wrapped with conducting polymers such as polyaniline (PANI), the composite support of PANI-CNT not only keeps most properties of CNTs, but also composes of the basis for metal-nitrogen-carbon (MN-C) electrocatalyst [10]. The electrocatalysis activities are usually restricted by many structural factors. Generally, the miniaturization of nanoparticles is beneficial to enhance activity due to large surface areas and quantum effects. For example, the activity of platinum nanoparticles is effective when their size lies in the range of 2-5 nm [18], and it will be enhanced significantly when their size is less than 1 nm [19,20]. However, to obtain small nanoparticles should deliberately control their nucleation-growth paths in most aqueous synthesis. Additionally, the stable core-shell structure has been designed to improve the electrocatalysis activity. For example, Wei et al. found that the activity and durability of PANI-Pt-C electrocatalyst with core-shell structure are sensitive to the thickness of the PANI shell [6]. Thus, to pursue PANI-Co-CNT electrocatalysts is deduced to decrease nanosizes and stable composite structure, which are expected to enhance electrocatalysis remarkably. It is worthy mentioned that the modification of CNT usually depends on covalent methods whose chemical treatments inevitably destroy the CNT surface. As a non-covalent method, ␲-␲ stacking based on two aromatic molecules interaction provides non-destructive route for aniline adsorption on CNT surface

0013-4686/$ – see front matter. Crown Copyright © 2013 Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.electacta.2013.12.072

Z.-s. Yin et al. / Electrochimica Acta 119 (2014) 144–154

[21–23]. This route also helps PANI wrap CNT surfaces completely under the condition of aniline polymerization. Moreover, the functional group of NH2 in aniline can be coordinated with cobalt precursor and play capping role for cobalt nanoparticles [24], thus controlling cobalt nanosize effectively. As a bridge between CNT and cobalt precursor, the aniline is deduced to open an in-situ route for composite PANI-Co-CNT structure. We also noticed there are many capping agents such as citric acid (CA) [25–28], whose extension for in-situ route is expected. Herein, we disclosed that the mixing solution of aniline, CNTs and cobalt chloride (CoCl2 ) precursor can be in-situ reduced to prepare polyaniline-cobalt-carbon nanotube (PANI-Co-CNT) electrocatalyst. We found 2-4 nm cobalt nanoparticles could be mounted in the porous PANI-CNT structure, but the CoCl2 precursor pretreated by CA improves nanoparticles distribution significantly. Our control experiments revealed the differences for CoCl2 precursor with and without pretreatment, thus proposing their preparation mechanisms as the self-assembly manner. We also related the basically 4-electron processes in ORR behaviors to their porous structures, determining the large activity and stability for pretreated PANI-Co-CNT sample. All these progresses help to make a complete breakthrough for future electrocatalysts. 2. Experiment 2.1. Materials and sample synthesis Multiwall carbon nanotubes (MWCNTs) without any surface groups were purchased from Scientech Corporation. Aniline, cobalt chloride hexahydrade (CoCl2 ·6H2 O), sodium citrate tribasic dehydrate (C6 H5 Na3 O7 ·2H2 O), ethanol, sodium borohydride (NaBH4 ) were all bought from Sigma-Aldrich and 5 wt% Nafion were procured from DuPont. All chemicals were analytical grade reagents and used as received. Synthesis of PANI-Co-CNT: First, 0.05 g of CNTs and 0.03 g aniline were mixed in 12.5 ml ethanol and 12.5 ml DI water, dispersed via ultrasonication for 3 hours. The amount of aniline was excess for complete PANI coverage on CNT surface. Then, 5 ml CoCl2 (0.1 mol L−1 ) was added into the above solution and further ultrasonicated for 2 hours. Next, 1.5 g NaBH4 reducer was poured into the mixture solution and allow the reaction to last for at least 2 hours until no bubbles appeared. Finally, the solution was filtered to obtain suspended solids, which was dried at 60 ◦ C to achieve prepared samples. To extend synthesis, the above procedures were same except that CoCl2 precursor was pretreated by CA, which was carried out to mix 5 ml CoCl2 (0.1 mol L−1 ) and 0.15 g C6 H5 Na3 O7 ·2H2 O into 25 ml water and ultrasonicated for at least 5 hours. Thus we obtained two PANI-Co-CNT samples, namely Co + CA and Co-CA, corresponding to the pretreated and no-pretreated with CA species. To probe the growth manners, we made control experiments whose procedures were the same as the above synthesis except that no CNTs were added, so we acquired another two samples, named as PANI-Co (+CA) and PANI-Co (-CA) respectively.

145

ion contents were detected by ICP-OES. (2) The morphologies were observed by transmission electron microscope (TEM) and scanning transmission electron microscope (STEM) on Tecnai G2 Transmission Electron Microscope (FEI). (3) The X-ray diffraction (XRD) patterns were acquired on a Bruker D8-advance X-ray diffractometer. (4) X-ray photoelectron spectroscopy (XPS) equipped with PHI Quantera Scanning X-ray Microprobe (ULVAC-PHI, Inc) was used to determine element content and the valence state of cobalt species. (5) Surface areas and pore size distributions were measured by nitrogen adsorption at 77 K on a NOVA 3200e Analyzer (Quantachrome). 2.3. Electrochemical characterizations A three-electrode system was composed of glassy carbon electrode (GCE) with an area of 0.1256 cm2 as the working electrode, Pt wire as the counter electrode and saturate calomel electrode (SCE) as the reference electrode. The catalyst ink was composed of 5 mg prepared sample and 950 ␮L ethanol containing 50 ␮L of 5 wt% Nafion, and ultrasonicated for at least an hour. The GCE surface was initially polished with alumina suspension, then pipetted with 20 ␮L catalyst ink and finally covered with a drop of 0.1 wt% Nafion, with the catalyst loading at 0.8 mgcatalyst cm−2 . Cyclic voltammetry (CV) and rotating disk electrode (RDE) measurements were carried out on potentiostat (EG&G Princeton, Model 263). CV experiments were conducted at room temperature in 0.5 M H2 SO4 solution saturated with nitrogen or oxygen. Before each experiment, the working electrode was cycled at least 10 times with 50 mV s−1 from 1.19 V to 0 V vs. RHE until data stabilization. Then the ORR was made at a scan rate of 5 mV s−1 , from 1.19 V to 0 V vs. RHE, accompanied with RDE at the rotating speed from 400-3600 rpm controlled by Pine instrument. Normalized currents were labeled as (mA cm−2 ) in this paper. To evaluate durability, we made some accelerated degeneration tests which were carried out by CV in the range of 1.19 V-0 V vs. RHE at scan rate of 150 mV s−1 in the O2 saturated 0.5 mol L−1 H2 SO4 solution at 0 rpm. After every 90 cycles, the ORR was made at the scan rate of 5 mV s−1 and the rotation speed at 1600 rpm. The final retention ratio for current was defined as the ratio of the last current to the initial current under various potentials. 3. Results and discussion 3.1. Physicochemical characterizations Fig. 1 displays the TGA curves in the air for both Co-CA and Co + CA samples. A typical three-step loss for PANI has been

2.2. Physical characterizations The UV-vis spectroscopies were made on a UV-2101 (Shimadzu) to evaluate solution properties in control experiments. Some techniques were used to realize the prepared PANI-Co-CNT samples: (1) Both Thermogravimetric analysis (TGA) on STA 490 PC (Netzsch) and ICP-OES measurements on IRIS Intrepid II XSP (ThermoFisher) were coordinately conducted to determine cobalt amount. TGA was carried out at 5 ◦ C min−1 in air from room temperature to 1000 ◦ C, then the residues were dissolved in hydrochloric acid whose cobalt

Fig. 1. TGA curves for PANI-Co-CNT (+CA, -CA) samples in the air from 25 ◦ C to 1000 ◦ C.

146

Z.-s. Yin et al. / Electrochimica Acta 119 (2014) 144–154

Fig. 2. TEM and STEM images of the samples: (a), (b), (c) TEM images for CNT and PANI-Co-CNT (+CA, -CA); (d), (e) STEM images for PANI-Co-CNT (+CA, -CA).

Z.-s. Yin et al. / Electrochimica Acta 119 (2014) 144–154

reported as that: initially the water entrapped in the PANI matrix and low molecular oligomers lost between 50-140 ◦ C, then the eliminations of acid dopant and micromolecular organics occurred in 150-300 ◦ C, next the PANI backbone chains decomposed about 350 ◦ C [29]. In consistent with above results for the Co-CA sample, the Co + CA sample demonstrated uniform loss before 350 ◦ C, reflecting its regular micro-structure. When the temperature increased to ca. 600 ◦ C, all CNTs began to burn up and this process lasted until 1000 ◦ C [30], in which the mass percentages of cobalt oxides were determined as 28.76% (Co + CA), 35.50% (Co-CA). To evaluate further, the remnants after TGA were collected and dissolved in hydrochloric acid to obtain their ionic states, whose cobalt contents were determined as 17.26% (Co + CA) and 19.93% (Co-CA) in the ICP-OES measurements, indicating small differences for cobalt contents. Fig. 2a-2c exhibit the TEM images for CNT, Co + CA and CoCA samples, showing the average diameter of CNT at ca. 6.85 nm and clear interface between PANI and CNTs. These images have difficulties in determining small cobalt nanoparticles, due to the TEM technique mainly depending on phase-contrast. Therefore, the STEM which uses electrons passing through sample to form bright field and scattering electrons to establish dark field was adopted as high contrast imaging. Fig. 2d, 2e illustrate a lot of brighter points in the perpendicular direction which were identified as cobalt species by EDS (not shown). Their comparisons demonstrate that the cobalt nanoparticles agglomerate and PANI connect CNT arrays for Co-CA sample (Fig. 2e), whereas 2-4 nm nanoparticles distribute uniformly and PANI separate CNT arrays for Co + CA sample (Fig. 2d), verifying the beneficial role for CA pretreatment. Fig. 3 demonstrates their XRD patterns, whose peaks at 2␪ =20.2◦ and 25.8◦ were caused by the periodicity parallel and perpendicular

147

Fig. 3. XRD patterns of two samples PANI-Co-CNT (+CA, -CA).

of the PANI chains [29,31]. The characteristic peaks of (002) and (100) for CNT were found in the ca. 26.5◦ and ca. 43.0◦ [32], while strong cobalt peaks at 34.0◦ and 42.5◦ were also suggested. In Fig. 3, though there are some overlaps for peaks at ca. 26.0◦ , ca. 43.0◦ , it is still definite that the Co + CA sample has regular crystalline structure, in contrast with the amorphous structure for Co-CA sample. Due to the cobalt nanoparticles mounted in the PANI network, we speculated that the composite structures would easily produce pore size. Fig. 4a and 4b show their adsorptiondesorption isotherms, exhibiting hysteresis loops due to capillary condensation-evaporation at high pressure, which belong to Type IV according to the IUPAC classification [33]. Moreover, the BET

Fig. 4. (a), (b) N2 adsorption-desorption isotherms at 77 K of PANI-Co-CNT (+CA, -CA); (c), (d) Pore size distributions of PANI-Co-CNT (+CA, -CA).

148

Z.-s. Yin et al. / Electrochimica Acta 119 (2014) 144–154

Fig. 5. (a), (b) XPS spectrum of two samples PANI-Co-CNT (+CA, -CA); (c), (d) Valence states for cobalt species in PANI-Co-CNT (+CA, -CA) after separating and fitting the cobalt peaks.

are achieved as 349.1 m2 g−1 (Co + CA) and 322.9 m2 g−1 (Co-CA), verifying their large surface areas. The BJH method is used to determine their pore size distribution in Fig. 4c and 4d, indicating most pores belonging to mesopore ranges (2-50 nm) which is expected to benefit for mass transfer. With X-ray photoelectron spectroscopies, Fig. 5a and 5b show their element distributions whose contents are listed in Table 1. It is interesting to find the cobalt contents at ca. 17% are attainable to the results of ICP-OES, implying the photoelectrons in XPS collecting whole real surface areas. This similarity is deduced as under the background of lighter carbon elements of PANI and CNT, the electron depth may attain 10 nm which can penetrate through whole composite structure, resulting in the mounted nanoparticle to be detected. To realize the valence states for cobalt species, the regional Spectra of Co2p illustrate doublet peaks at low energy bands (Co2p3/2 ) and high energy bands (Co2p1/2 ), exhibiting 780.8 eV and 796.7 eV for the Co + CA sample while 780.7 eV and 796.2 eV for the Co-CA sample. After separating and fitting, Fig. 5c and 5d show that CoIII 3/2 configurations correspond to 780.3 eV (Co + CA)

Table 1 Mass content of every element for PANI-Co-CNT (+CA, -CA) samples according to the XPS results. Element (%)

C

N

O

Co

Co0 /Co

Co + CA Co-CA

63.81 61.31

0.28 0.50

18.43 20.87

17.47 17.32

20.94 17.77

and 780.0 eV (Co-CA), while those of CoII 3/2 are related to the peaks around 782.3 eV (Co + CA) and 781.6 eV (Co-CA), besides Co0 3/2 peaks are ca. 777.4 eV (Co + CA) and 777.8 eV (Co-CA). All these results are consistent with those reported in literatures [34,35], suggesting a core-shell structure of Co0 -Co2+ -Co3+ for both samples. 3.2. Preparation mechanism To realize the growth manners, we made control experiments whose processes were the same as synthesis except that no CNTs ware added. In the initial aniline stage, we found both samples of PANI-Co (+CA) and PANI-Co (-CA) had slight violet colors whose UV-Vis exhibited absorbance at ca. 500-550 nm. It should be realized there are three kinds of PANI: totally reduced leucoemeraldine (LB) with benzene unit, fully oxidized pernigraniline (PB) with quinoid units, exactly half-oxidized emeraldine (EB) with both benzene and quinoid units. In UV-Vis, the band at 300-400 nm is ascribed to the ␲-␲* transition in benzene units (B-band) while the band at 600-700 nm is associated with exciton-like transitions in quinoid units (Q-band) [31]. Thus various PANI can be identified as that the LB presents at B-band and the EB presents both B-band and Q-band, and the PB occurs at ca. 535 nm due to some B-band impact [36]. These proofs support that the initial complex processes of ANI-Co2+ help to form PANI oligomer in PB state. In the next NaBH4 stage, the brown and black colors present for PANI-Co (+CA) and PANI-Co (-CA) samples respectively (inset in Fig. 6). In our UV-Vis measurements, only the regular curves

Z.-s. Yin et al. / Electrochimica Acta 119 (2014) 144–154

149

Fig. 6. (a), (b) UV-Vis spectra for PANI-Co (+CA, -CA); (c), (d) TEM images for PANI-Co (+CA, -CA), control experiments in the absence of CNTs.

depending on transmittance were achieved, in contrast with those erratic curves depending on absorbance. Previous studies revealed the transmittance is significantly increased under the polymer protection for nanoparticle [37–39], pointing out aniline polymerization dynamically. Thus, Fig. 6b suggests the LB state for Co-CA sample with its peak at ca. 360 nm. With its peaks at ca. 360 nm and ascending tail after ca. 600 nm, Fig. 6a indicates Co + CA sample have both EB and LB states which are doped with some Co2+ species [29]. The TEM images in Fig. 6c and 6d show their amorphous states. In the recent self-assembly progress, the precritical clusters in amorphous state are determined by their crystalline structures and molecular adsorptions, whose detachments help their independent growth. With successive detach-growth processes, the precritical clusters evolve to some crystalline structures gradually until nanoparticles formatted completely [40]. Thus the amorphous states in Fig. 6c and 6d still lie in incubation stage, revealing their further growths according to self-assembly manners. Compared with Fig. 2d and 2e, it is obvious some crystalline structures formatted. Though these differences are still unclear, it is safeguard to deduce that the ␲-␲ interaction between aniline and CNT dominates the heterogeneous nucleation, thus directing the selfassembly manners which help the PANI and cobalt nanoparticles detachment. Based on the above results, Scheme 1 proposed the preparation mechanisms for both routes. In the branch of Co-CA route, the aniline as bridge for CNTs and cobalt precursor helped establish initial nucleation. The Co2+ coordination with aniline was reported to follow two step redox processes, in which first Co2+ oxidized the

polymer unit and get reduced to Co, then Co again bind with imine nitrogen from quinoide ring and get oxidized to Co2+ , leading to the formation of radical cation segments [31]. These in-situ processes drive aniline to be oxidized as PB oligomer initially, but they will be accelerated towards reduction in the subsequent NaBH4 stage thus driving cobalt nanoparticles formation and PANI polymerization simultaneously. Because they have different growth laws, the PANI and cobalt nanoparticles should coordinate with each other to establish porous structure. Furthermore, the strong and excess NaBH4 reducer helps PANI to attain its totally reducing LB state. Due to the rigid PANI structure, some dissolved O2 in the solution would easily make cobalt nanoparticles partially oxidization in their surfaces, producing Co0 -Co2+ -Co3+ as core-shell structure. In the branch of Co + CA sample, the initial heterogeneous nucleation for cobalt is strengthened because the COO− from CA and NH2 from aniline can establish strong COO- -NH2 bonds. In the NaBH4 reduction, it is easily understood that the CA complex for Co2+ will experience detachment in the self-assembly process [41]. These isolated CA will drive aniline to form brain-like [42] or nanotube [43] structures of PANI, matching well with the CNT surface. Furthermore, the CA detachment also implies the Co + CA sample cannot be easily reduced, producing both the LB and EB states. Because the ␲-␲ interaction increased strongly especially through quinonoid rings in EB, the adhesion between PANI-CNT interfaces is increased to form some regular structure [44]. All these advantages suggest CA pretreatment helps improvement for cobalt nanoparticle distribution and PANI matrix, and separates PANI-CNT arrays effectively.

150

Z.-s. Yin et al. / Electrochimica Acta 119 (2014) 144–154

Scheme 1. The proposed preparation mechanisms for two routes of PANI-Co-CNT (+CA, -CA) samples.

Our in-situ preparations are different from most synthesis depending on oxidant to polymerize aniline. If aniline and Co2+ were mixed previously, the next oxidation process helps to form PANI-Co structure with diameter in 20-200 nm which is larger than ours [31,44]. If aniline experience initial polymerization, the next capping agent to control cobalt nanosize should be further removed by heat-treatment, inevitably destroying PANI matrix and weakening the anchor forces between PANI-Co [1,27]. With the advantage of ambient synthesis, our in-situ methods open doors towards jointcontrol of nanoparticles and aniline polymerization. We expect future improvement on pretreatment helps achieve layer-by-layer structures, which are predicted as high-efficient and durable electrocatalysts. 3.3. Kinetic Analysis To evaluate ORR behaviors, Fig. 7 shows CV curves in O2 - and N2 -saturated 0.5 M H2 SO4 , with the similar onset potentials at ca. 0.92 V vs. RHE for both Co + CA and Co-CA samples. At the potential of 0.45 V to 0.92 V, there are conspicuous redox peaks in O2 -saturated electrolyte and plateau shape appeared less than 0.45 V, indicative of typical ORR phenomena [45]. The peak currents reach highest at 0.62 V (Co + CA) and 0.60 V (Co-CA). In addition, the half-height width of Co + CA in the positive direction is ca. 100 mV (theoretical value: 96 mV), in accordance with one-electron process in surface species transition as rate-determining step [1]. We also checked the controlled samples of PANI-Co (+CA) and

PANI -Co (-CA), whose stable ORR behaviors cannot be obtained after successive cycles. These phenomena strongly implied that their incubation states are not stable structure, in contrast with the stable structure to be directed by CNT skeleton. The RDE measurements were made under the rotation speed from 400 to 3600 rpm. To compare with Fig. 8a and 8c, the current density of Co + CA is higher than that of Co-CA at the same rotation speed, verifying the advantage of the CA pretreatment. Because the measured currents are affected by kinetic current and diffusion-limited current, we divide three regions to character ORR processes [46]. For E > ca. 0.80 V vs. RHE, the current is mainly decided by the reaction kinetics. With overpotentials increasing in the range of 0.80 V to 0.45 V, the curves separation indicates the current is jointly controlled by kinetics and mass transport. When E
In Eq. (1), J is the measured current density, Jk and Jl stand for the kinetic current and diffusion current density respectively, and ω is the rotation speed of the working electrode. The K-L plots reveal the inverse of measured current density (J−1 ) as a function of the inverse of square root of the rotation speed (ω−1/2 ) at different potentials. B is determined by the slope of the fitting K-L plots on the Levich equation: 2 B = 0.62nFCo Do ⁄3 v−1/6

Fig. 7. CV curves of PANI-Co-CNT (+CA, -CA) at the scan rate of 10 mV s-1 at 0 rpm in N2 or O2 saturated 0.5 M H2SO4 solution at 25 ◦ C.

(1)

(2)

In Eq. (2), n is the overall electron transfer number per O2 , F is the Faraday constant (F = 96485.3 C mol−1 ). Co is the bulk concentration of dissolved O2 in the electrolyte (1.1 × 10−6 mol cm−3 in 0.5 mol L−1 H2 SO4 ), Do is the diffusion coefficient of O2 (1.4 × 10−5 cm2 s−1 ) and ␯ is the kinematic viscosity (0.01 cm2 s−1 ). Based on Equations of (1) and (2), Fig. 8b and 8d present their K-L plots at E= 0.45 V, 0.40 V, 0.35 V, 0.30 V and 0.25 V. Furthermore, Table 2 tabulates that the number of n is 3.7-3.8 for Co + CA sample while it is 3.2-3.4 for Co-CA sample. Because the ORR proceeds either by 2-electron or by 4-electron process, these n values suggest both samples are dominated by 4-electron transfer pathways and Co + CA sample undergo 4-electron process preferentially.

Z.-s. Yin et al. / Electrochimica Acta 119 (2014) 144–154

151

Fig. 8. (a), (c) ORR polarization curves of PANI-Co-CNT (+CA, -CA) at different rotating speeds from 400 to 3600 rpm at scan rate of 5 mV s-1 in O2 saturated 0.5 M H2SO4 solution. (b), (d) K-L plots of the ORR at different potentials, 0.45 V, 0.40 V, 0.35 V, 0.30 V, 0.25 V vs. RHE from PANI-Co-CNT (+CA, -CA).

Table 2 Overall reaction electron numbers for PANI-Co-CNT (+CA, -CA) samples calculated from Eq. (1) and (2). E (V vs. RHE)

0.45

0.40

0.35

0.30

0.25

n

3.8 3.4

3.8 3.3

3.7 3.3

3.7 3.2

3.7 3.2

Co + CA Co-CA

We also made Tafel analysis through the Eq. (3), whose results are shown in Table 3. E = E 0 + (2.303RT/an˛ F) log(J) − (2.303RT/an˛ F) log(J0 )

(3)

where E and E0 represent the measured and standard electrode potential respectively, (E0 = 1.23 V vs. RHE), J0 is the exchange current density, ˛ is the electron transfer coefficient in the ratedetermining step, n˛ means the electron transfer number in the RDS, R is the universal gas constant (8.314 J mol−1 K−1 ) and T is the thermodynamic temperature (293.15 K) [46]. Fig. 9 shows that the Tafel plots increase sharply at low overpotential zones, but exhibit linear regions with the slope of ca. 126-132 mV dec−1 at high overpotential zones (Table 3). In the most planar electrodes, two linear regions with the slope of 60 mV dec−1 (Temkin isotherm, high coverage) at low overpotential zones Table 3 Kinetic parameters obtained from Eq. (3) for ORR on PANI-Co-CNT (+CA, -CA) samples. Catalyst

E (V vs. RHE)

Tafel slope (mV dec−1 )

˛n˛

J0 (mA cm−2 )

Co + CA Co-CA

0.85-0.64 0.86-0.68

-132.1 -126.6

0.439 0.412

5.7 × 10−5 5.0 × 10−5

and 120 mV dec−1 (Langmuir isotherm, low coverage) at high overpotential zones can be generally observed [49–52]. However, on thin porous Pt/C electrode, large capacitive currents will result in high deviation at low overpotential zones, which can be effectively corrected by the flooded agglomerate model [51]. Thus, with the overpotential increasing, our mesoporous structures will gradually utilize their mounted cobalt nanoparticles, driving the Langmuir adsorption as low coverage present. Furthermore, the complete utilization for cobalt nanoparticles implies the Tafel slope may exceed 120 mV dec−1 , and this situation is expected to be present especially at high overpotential zones [53]. It should be stressed that most cobalt electrocatalysts were checked their activities in alkali environment [3,7,16,52,54,55], where the active sites were generally regarded as surface Co3+ as donor-acceptor reduction sites. The OH− in the environment was also important for surface confined redox couples such as: CoOOH + OH- ↔ CoO2 + H2 O + e-

(4)

The increase of cobalt valence state in Eq. (4) not only makes the ORR easily occurring, but also maintains the surface stability dynamically [52]. However, in an acid medium, the H+ involvement will corrode surface oxides. Due to this difficulty, only a few studies on cobalt species have been proved their stable ORR catalysis in acid environment. For example, the cobalt-tripyridyl triazine (CoTPTZ) complexes after heat-treatment at 700 ◦ C would yield the best activity, whose active sites were deduced as Co (II)-N/C structure [45]. For PANI-Co-C catalyst, its heat-treatment at 400-1000 ◦ C would facilitate nitrogen of PANI to incorporate into the graphitized carbon matrix, possibly initiating Co-N-C as active sites [1]. These

152

Z.-s. Yin et al. / Electrochimica Acta 119 (2014) 144–154

Fig. 9. Tafel plots for ORR on PANI-Co-CNT (+CA, -CA) electrodes at 1600 rpm in O2 saturated. 0.5 M H2SO4 solution at 25 ◦ C.

Fig. 10. Currents density of PANI-Co-CNT (+CA, -CA) catalysts measured by CV at scan rate of 5 mV s-1 at 1600 rpm under 0.80 V, 0.70 V, 0.65 V and 0.60 V vs. RHE after cycling at scan rate of 150 mV s-1 in O2 saturated 0.5 M H2SO4 solution at 0 rpm.

Co-N-C sites are expected to be increased in our in-situ approaches, laying basis to compare activities further. As shown in Table 3, the values of exchange current (J0 ) are 5.7 × 10−5 mA cm−2 (Co + CA) and 5.0 × 10−5 mA cm−2 (CoCA), which can be converted to 7.1 × 10−5 mA mg−1 (Co + CA) and 6.3 × 10−5 mA mg−1 (Co-CA) based on our 0.8 mg cm−2 loading. Compared with others under acid condition, the J0 in the optimum of Co-TPTZ lied at ca. 10−7 mA cm−2 , which can be transformed to 2 × 10−5 mA mg−1 with its 5 ␮g cm−2 loading [45]. The J0 in the optimum PANI-Co-C sample was 5 × 10−5 mA cm−2 , which can be converted to 8.3 × 10−5 mA mg−1 with its 0.6 mg cm−2 loading [1]. These comparisons clearly indicate that the activity in our Co + CA sample can be comparable to that of the advanced PANI-Co-C catalyst, even only considering total loading. These results further verify that our large porous structure and the uniform distribution of cobalt nanoparticles are mainly responsible for activity increasing. The durability of electrocatalyst can be checked by accelerated degradation test (ADT), in which the electrochemically forced aging method was largely used. This was usually made under a threeelectrode system with initial cycling and then ORR measurement [56]. To evaluate ORR activity after every 90 cycles, Fig. 10 illustrates the current density corresponding to 0.60-0.80 V vs. RHE after successive cycling, and the activity for retention ratios are also denoted in their tails. It can be found that after 1500 cycles, the activity for retention ratios at 0.80 V are 93.1% (Co + CA), 92.7% (Co-CA), indicating their stable electrocatalysis in kinetic regions. But in the range of 0.70-0.60 V, the activity for retention ratios can only be kept at 74.9%, 73.8% and 76.2% for Co + CA sample and 74.3%, 72.8%,

and 70.5% for Co-CA sample, reflecting the large impact from mass transfer. These values are similiar to those on advanced GrapheneCo/CoO catalysts [3], whose alkaline application further supports our samples as highly stable electrocatalysts. We attribute our good durability to two reasons. In physical reason, our samples have rigid CNT and wrapped PANI matrix to resist acid corrosion, and the mounted cobalt nanoparticles were protected somehow. In electrochemical reason, we have found the intermediate species of H2 O2 in ORR can establish Fenton’s reaction [57], which induces active species to attack electrocatalysts thus leading to complete degeneration. Fig. 10 demonstrates that the CoCA sample decays more quickly than Co + CA sample, in accordance with its more paths for 2-eletron to produce H2 O2 . Therefore, the H2 O2 production should be avoided to the best, and the O2 adsorption manners should be investigated. Both of them are expected to improve durability significantly. 4. Conclusion We disclosed the mixing solution of CNT, aniline and CoCl2 precursor at room temperature can be directly reduced to achieve PANI-Co-CNT structure with an in-situ approach, which can also be obtained if CoCl2 precursor was pretreated by citric acid. The physical characterizations found both routes achieve porous structures and small cobalt nanoparticles (2-4 nm), but the pretreated route exhibits some advantages in nanoparticles distribution and regular structure. With the help of control experiments, we realized their PANI states and the large impact of ␲-␲ stacking, proposing

Z.-s. Yin et al. / Electrochimica Acta 119 (2014) 144–154

preparation mechanisms as self-assembly manners which can be effectively improved by CA participation. The pretreated PANI-CoCNT sample is more likely to be dominated by 4-electron path in ORR processes, and its large activity and durability in acid condition can be comparable with those of advanced electrocatalysts. Therefore, our in-situ approaches open a door to achieve non-noble-metal electrocatalysts towards ORR application, which are expected to drive research further. Acknowledgments This work is supported by the National Natural Science Foundation of China (51273103, 50873050 and 21106079) and National Basic Research Program of China (2012CB215500). References [1] G. Wu, K.L. More, C.M. Johnston, P. Zelenay, High-performance electrocatalysts for oxygen reduction derived from polyaniline, iron, and cobalt, Science 332 (2011) 443–447. [2] K. Jayasayee, J.A.R. Van Veen, T.G. Manivasagam, S. Celebi, E.J.M. Hensen, F.A. de Bruijn, Oxygen reduction reaction (ORR) activity and durability of carbon supported PtM (Co, Ni, Cu) alloys: Influence of particle size and non-noble metals, Appl. Catal. B: Environ. 111 (2012) 515–526. [3] S. Guo, S. Zhang, L. Wu, S. Sun, Co/CoO nanoparticles assembled on fraphene for electrochemical reduction of oxygen, Angew. Chem. Int. Ed. 124 (2012) 11940–11943. [4] J. Zhang, M.B. Vukmirovic, K. Sasaki, A.U. Nilekar, M. Mavrikakis, R.R. Adzic, Mixed-Metal Pt monolayer electrocatalysts for enhanced oxygen reduction kinetics, J. Am. Chem. Soc. 127 (2005) 12480–12481. [5] B.R. Camacho, C. Morais, M.A. Valenzuela, N. Alonso-Vante, Enhancing oxygen reduction reaction activity and stability of platinum via oxide-carbon composites, Catal. Today 202 (2013) 36–43. [6] S. Chen, Z. Wei, X.Q. Qi, L. Dong, Y.G. Guo, L. Wan, Z. Shao, L. Li, Nanostructured polyaniline-decorated Pt/C@PANI core–shell catalyst with enhanced durability and activity, J. Am. Chem. Soc. 134 (2012) 13252–13255. [7] B. Lim, M. Jiang, P.H.C. Camargo, E.C. Cho, J. Tao, X. Lu, Y. Zhu, Y. Xia, Pd-Pt bimetallic nanodendrites with high activity for oxygen reduction, Science 324 (2009) 1302–1305. [8] Y. Li, Y. Li, E. Zhu, T. McLouth, C.Y. Chiu, X. Huang, Y. Huang, Stabilization of high-performance oxygen reduction reaction Pt electrocatalyst supported on reduced graphene oxide/carbon black composite, J. Am. Chem. Soc. 134 (2012) 12326–12329. [9] H. Yang, Platinum-based electrocatalysts with core–shell nanostructures, Angew. Chem. Int. Ed. 50 (2011) 2674–2676. [10] Z. Chen, D. Higgins, A. Yu, L. Zhang, J. Zhang, A review on non-precious metal electrocatalysts for PEM fuel cells, Energy Environ. Sci. 4 (2011) 3167–3192. [11] A. Morozan, B. Jousselme, S. Palacin, Low-platinum and platinum-free catalysts for the oxygen reduction reaction at fuel cell cathodes, Energy Environ. Sci. 4 (2011) 1238–1254. [12] D.S. Su, G. Sun, Nonprecious-metal catalysts for low-cost fuel cells, Angew. Chem. Int. Ed. 50 (2011) 11570–11572. [13] F. Jaouen, E. Proietti, M. Lefevre, R. Chenitz, J.P. Dodelet, G. Wu, H.T. Chung, C.M. Johnston, P. Zelenay, Recent advances in non-precious metal catalysis for oxygen-reduction reaction in polymer electrolyte fuel cells, Energy Environ. Sci. 4 (2011) 114–130. [14] H.R. Byon, J. Suntivich, Yang Shao-Horn, Graphene-based non-noble-metal catalysts for oxygen reduction reaction in acid, Chem. Mater. 23 (2011) 3421–3428. [15] S. Park, Y. Shao, R. Kou, V.V. Viswanathan, S.A. Towne, P.C. Rieke, J. Liu, Y. Lin, Y. Wang, Polarization losses under accelerated stress test using multiwalled carbon nanotube supported Pt catalyst in PEM fuel cells, J. Electrochem. Soc. 158 (2011) B297–B302. [16] Y. Liang, H. Wang, P. Diao, W. Chang, G. Hong, Y. Li, M. Gong, L. Xie, J. Zhou, J. Wang, T.Z. Regier, F. Wei, H. Dai, Oxygen reduction electrocatalyst based on strongly coupled cobalt oxide nanocrystals and carbon nanotubes, J. Am. Chem. Soc. 134 (2012) 15849–15857. [17] G. Lota, K. Fic, E. Frackowiak, Carbon nanotubes and their composites in electrochemical applications, Energy Environ. Sci. 4 (2011) 1592–1605. [18] A. Kloke, F. von Stetten, R. Zengerle, S. Kerzenmacher, Strategies for the fabrication of porous platinum electrodes, Adv. Mater. 23 (2011) 4976–5008. [19] C. Bock, C. Paquet, M. Couillard, G.A. Botton, B.R. MacDougall, Size-selected synthesis of PtRu nano-catalysts: reaction and size control mechanism, J. Am. Chem. Soc. 126 (2004) 8028–8037. [20] E.J. Yoo, T. Okata, T. Akita, M. Kohyama, J. Nakamura, I. Honma, Nano Lett. 9 (2009) 2255–2259. [21] P.M. Ajayan, J.M. Tour, Materials science: nanotube composites, Nature 447 (2007) 1066–1068. [22] D. He, C. Zeng, C. Xu, N. Cheng, H. Li, S. Mu, M. Pan, Polyanilinefunctionalized carbon nanotube supported platinum catalysts, Langmuir 27 (2011) 5582–5588.

153

[23] P. Jimenez, P. Castell, R. Sainz, A. Anson, M.T. Martinez, A.M. Benito, W.K. Maser, Carbon nanotube effect on polyaniline morphology in water dispersible composites, J. Phys. Chem. B 114 (2010) 1579–1585. [24] W.M. Millan, T.T. Thompson, L.G. Arriaga, M.A. Smit, Characterization of composite materials of electroconductive polymer and cobalt as electrocatalysts for the oxygen reduction reaction, Int. J. Hydrogen Energy 34 (2009) 694–702. [25] S. Ozkar, R.G. Finke, Nanocluster formation and stabilization fundamental studies: Ranking commonly employed anionic stabilizers via the development, then application, of five comparative criteria, J, Am. Chem. Soc. 124 (2002) 5796–5810. [26] J.W. Guo, T.S. Zhao, J. Prabhuram, R. Chen, C.W. Wong, Preparation and characterization of a PtRu/C nanocatalyst for direct methanol fuel cells, Electrochim. Acta 51 (2005) 754–763. [27] J.W. Guo, T.S. Zhao, J. Prabhuram, C.W. Wong, Preparation and the physical/electrochemical properties of a Pt/C nanocatalyst stabilized by citric acid for polymer electrolyte fuel cells, Electrochim. Acta 50 (2005) 1973–1983. [28] Y. Liu, X. Zhang, Effect of calcination temperature on the morphology and electrochemical properties of Co3 O4 for lithium-ion battery, Electrochim. Acta 54 (2009) 4180–4185. [29] S. Zhou, T. Wu, J. Kan, Effect of Co2+ , Ni2+ , Cu2+ , or Zn2+ on properties of polyaniline nanoparticles, J, Appl. Polym. Sci. 106 (2007) 652–658. [30] D. Born, R. Andrews, D. Jacques, J. Anthony, B. Chen, M.S. Meier, J.P. Selegue, Thermogravimetric analysis of the oxidation of multiwalled carbon nanotubes: Evidence for the role of defect sites in carbon nanotube chemistry, Nano Lett. 2 (2002) 615–619. [31] L. Tang, T. Wu, J. Kan, Synthesis and properties of polyaniline-cobalt coordination polymer, Synth. Met. 159 (2009) 1644–1648. [32] T. Belin, F. Epron, Characterization methods of carbon nanotubes: a review, Mat. Sci. Eng. B 119 (2005) 105–118. [33] M. Kruk, M. Jaroniec, Gas adsorption characterization of ordered organicinorganic nanocomposite materials, Chem. Mater. 13 (2001) 3169–3183. [34] L. Fu, Z. Liu, Y. Liu, B. Han, P. Hu, L. Cao, D. Zhu, Beaded cobalt oxide nanoparticles along carbon nanotubes: Towards more highly integrated electronic devices, Adv. Mater. 17 (2005) 217–221. [35] O. Metin, S. Ozkar, Hydrogen generation from the hydrolysis of ammoniaborane and sodium borohydride using water-soluble polymer-stabilized cobalt (0) nanoclusters catalyst, Energy Fuels 23 (2009) 3517–3526. [36] M. Blaha, M. Riesova, J. Zednik, A. Anzlovar, M. Zigon, J. Vohlidal, Polyaniline synthesis with iron (III) chloride-hydrogen peroxide catalyst system: Reaction course and polymer structure study, Syn. Met. 161 (2011) 1217–1225. [37] M. Yoshino, F. Yoshizawa, T. Matsumura, Control of optical transmittance by nano plastic forming, Int. J. Mater. Form. 1 (2008) 491–494. [38] S.A.A. Oloomi, A. Saboonchi, A. Sedaghat, Effects of thin film thickness on emittance, reflectance and transmittance of nano scale multilayers, Int. J. Phys. Sci. 5 (2008) 465–469. [39] D.H. Lee, Y.B. Pyun, K.S. Son, J.W. Choung, J.M. Lee, S.J. Son, W.I. Park, Controlled synthesis and optical transmission characteristics of silicon nanowires on colloid patterned glass substrates, J. Ceram. Process. Res. 10 (2008) 1–5. [40] D. Gebauer, H. Colfen, Prenucleation clusters and non-classical nucleation, Nano Today 6 (2011) 564–584. [41] R.Y. Jin, H.Q. Zhu, Quantum-sized gold nanoclusters: Bridging the gap between organometallics and nanocrystals, Chem. Eur. J. 17 (2011) 6584–6593. [42] Y. Zhu, J. Li, M. Wan, L. Jiang, Y. Wei, A new route for the preparation of brain-like nanostructured polyaniline, Macromol. Rapid. Commun. 28 (2007) 1339–1344. [43] L. Zhang, Y. Long, Z. Chen, M. Wan, The effect of hydrogen bond on selfassembled polyaniline nanostructures, Adv. Funct. Mater. 14 (2004) 693–698. [44] S. Giri, C.K. Das, S.S. Kalra, Preparation and characterization of the cobalt doped polyaniline/MWCNT nanocomposites for supercapacitor application, J. Mater. Sci. Res. 1 (2012) 10–18. [45] S. Li, L. Zhang, H. Liu, M. Pan, L. Zan, J. Zhang, Heat-treated cobalt-tripyridyl triazine (Co-TPTZ) electrocatalysts for oxygen reduction reaction in acidic medium, Electrochim. Acta 55 (2010) 4403–4411. [46] S.N.S. Goubert-Renaudin, X. Zhu, Carbon-supported cobalt oxide nanoparticles cobalt porphyrin for oxygen reduction in acids: Insights on reactivity, J. Electrochem. Soc. 159 (2012) B426–B429. [47] S. Wang, D. Yu, L. Dai, Polyelectrolyte functionalized carbon nanotubes as efficient metal-free electrocatalysts for oxygen reduction, J. Am. Chem. Soc. 133 (2011) 5182–5185. [48] S. Wang, D. Yu, L. Dai, D.W. Chang, J.B. Baek, Polyelectrolyte-functionalized graphene as metal-free electrocatalysts for oxygen reduction, ACS Nano 5 (2011) 6202–6209. [49] D.B. Sepa, M.V. Vojnovic, L.M. Vracar, A. Damjanovic, Apparent enthalpies of activation of electrodic oxygen reduction at platinum in different current density regions-I. Acid solution, Electrochim. Acta 31 (1986) 91–96. [50] D.B. Sepa, M.V. Vojnovic, A. Damjanovic, Reaction intermediates as a controlling factor in the kinetics and mechanism of oxygen reduction at platinum electrodes, Electrochim.Acta 26 (1981) 781–793. [51] J. Perez, E.R. Gonzalez, E.A. Ticianelli, Oxygen electrocatalysis on thin porous coating rotating platinum electrodes, Electrochim. Acta 44 (1998) 1329–1339. [52] J. Xu, P. Gao, T.S. Zhao, Non-precious Co3 O4 nano-rod electrocatalyst for oxygen reduction r. eaction in anion-exchange membrane fuel cells, Energy Environ. Sci. 5 (2012) 5333–5339. [53] A.L.N. Pinheiro, A. Oliveira-Neto, E.C. de Souza, J. Perez, V.A. Paganin, E.A. Ticianelli, E.R. Gonzalez, Electrocatalysis on noble metal and noble metal alloys dispersed on high surface area carbon, J. New. Mat. Electrochem. Systems 6 (2003) 1–8.

154

Z.-s. Yin et al. / Electrochimica Acta 119 (2014) 144–154

[54] J. Suntivich, H.A. Gasteiger, N. Yabuuchi, H. Nakanishi, J.B. Goodenough, Yang Shao-Horn, Design principles for oxygen-reduction activity on perovskite oxide catalysts for fuel cells and metal–air batteries, Nature Chem. 3 (2011) 546–550. [55] Y. Liang, H. Wang, J. Zhou, Y. Li, J. Wang, T. Regier, H. Dai, Covalent hybrid of spinel manganese-cobalt oxide and graphene as advanced oxygen reduction electrocatalysts, J. Am. Chem. Soc. 134 (2012) 3517–3523.

[56] Y. Shao, G. Yin, Y. Gao, Understanding and approaches for the durability issues of Pt-based catalysts for PEM fuel cell, J. Power Sources 171 (2007) 558–566. [57] J.W. Guo, X.F. Xie, J.H. Wang, Y.M. Shang, Effect of current collector corrosion made from printed circuit board (PCB) on the degradation of self-breathing direct methanol fuel cell stack, Electrochim. Acta 53 (2008) 3056–3064.