Reduction of Oxygen on Dispersed Nanocrystalline CoS2 - The

Oct 31, 2012 - Food Chem. ..... It was possible to crush the larger spherical meso-structures by ball ... that was left standing for 1 day after the u...
4 downloads 0 Views 2MB Size
Article pubs.acs.org/JPCC

Reduction of Oxygen on Dispersed Nanocrystalline CoS2 Jakub S. Jirkovský,*,† Alexander Björling, and Elisabet Ahlberg* Department of Chemistry and Molecular Biology, University of Gothenburg, SE-412 96 Gothenburg, Sweden S Supporting Information *

ABSTRACT: The electrocatalytic properties of nanocrystalline CoS2 have been investigated for the oxygen reduction reaction (ORR) in 0.1 M HClO4. CoS2 with pyrite structure was prepared by hydrothermal synthesis and attached to a glassy carbon electrode from solution with a mixture of carbon and Nafion. The prepared CoS2 electrode layers showed high activity toward the ORR and very good stability under oxygen reducing conditions. Selectivity of the ORR toward H2O2 was determined by rotating (ring) disk electrode measurements, and relatively high selectivity was obtained with up to 80% H2O2 formation around 0.4 V (vs Ag/AgCl), but this dropped to zero for potentials below 0.0 V. The amount of H2O2 produced between 0.6 and 0.0 V was dependent on the quality of the CoS2 dispersion within the electrode layer, and decreasing CoS2 particle size resulted in significant improvement in the ORR electrocatalytic activity, both by increasing the turnover frequency and through decreasing the selectivity toward H2O2 production.



known to show quite high activity for the ORR.7−10 However, ruthenium and selenium do not represent the best solution from the economical point of view due to relatively high price.11 A significant step forward was the replacement of Ru by Co and CoSe2 materials are promising.12−14 However, it appears that stability of the selenides and in particular CoSe2 might be an issue and a longstanding durability of the CoSe2 materials investigated so far in literature is questionable. In addition to selenides, sulfides represent another important group of the chalcogenide materials family.15 With respect to selenium, employing sulfur would represent further significant reduction of cost of these materials. Sulfur with iron form iron disulfide, commonly known as pyrite.15 Pyrite (FeS2) is a very stable material and can be found as a natural mineral and is stable even in acid media.16 Although pure FeS2 is a semiconductor and as such is not suitable for application in electrocatalysis, successful electrochemical measurements on ORR were performed with FeS2 mineral single crystal electrodes.16,17 This was possible due to impurities in the FeS2 structure that cause significant improvement in conductivity. CoS2 forms with the pyrite structure and in addition shows metallic conductivity. Recently, a couple of studies with

INTRODUCTION Electrocatalysis of the oxygen reduction reaction (ORR) remains a major challenge in polymer electrolyte fuel cell (PEFC) applications due to a high overpotential.1 To date, only materials based on Pt have shown sufficient activity and stability for reliable proton exchange membrane fuel cell (PEMFC) operation.2 The need for Pt, however, poses problems due to the high cost and limited resources of Pt; thus, mass production of Pt-based PEMFCs might be restricted.3 Extensive research efforts have been devoted to reducing the amount of Pt required in electrocatalysts that would show sufficient stability in acid, which is to date the only relevant medium for PEFCs due to the unavailability of alkaline membranes.1 Various strategies have been employed in order to lower the Pt loading required in fuel cells.1−5 The most efficient approach would be, however, to replace the Pt-based materials by alternative catalysts. Although it was believed that materials without Pt cannot achieve as good performance in oxygen reduction, the work by Lefévre et al.6 on so-called Fe/N/C catalysts proved that the electrocatalytic activity toward ORR of nonprecious metal catalyst can match that of Pt. The Fe/N/C catalysts, however, do not show sufficient stability to withstand the aggressive conditions in PEMFCs.1,6 Chalcogenides represent another possible alternative to the Pt-based catalyst.7 Ruthenium selenides and their analogues are © 2012 American Chemical Society

Received: August 2, 2012 Revised: October 30, 2012 Published: October 31, 2012 24436

dx.doi.org/10.1021/jp307669k | J. Phys. Chem. C 2012, 116, 24436−24444

The Journal of Physical Chemistry C

Article

thin layers formed of Co, Fe,18−20 and Ni20 disulfides with pyrite structure have shown promising performance of these materials for the ORR.20 Importantly, the CoS2 layer showed a good stability in acid media under oxygen reducing conditions, advantageous over the other disulfides. For “real” fuel cell applications, however, an electrocatalytic material has to be dispersed within a matrix such as, e.g., polymer electrolyte membrane so that a good electrical connectivity is achieved.1,21,22 For instance, CoSe2 nanoparticles were successfully employed in oxygen reduction while supported on carbon black and stabilized within a thin Nafion layer.13 For precious metal nanoparticles supported on carbon a robust methodology on how to create defined electrocatalytic layers already exists.21−23 For nonprecious metal catalysts, however, such methodology is lacking. Nonprecious materials such as chalcogenides are powders with morphologies very different from that of size defined metal nanoparticles supported on carbon. As a result, properties of the electrocatalytic layers formed with chalcogenides can differ significantly depending on morphology of the employed material, and special attention needs to be paid to watch the quality of these layers. This is in particular important since the assessment of the layer properties is essential for determination of the intrinsic kinetic parameters for ORR on studied materials, and the kinetic parameters can be only extracted if the contributions from transport processes, layer connectivity, and catalyst utilization can be neglected or controlled. The aim of this work is to (1) transpose experience gained in studies on ORR on carbon supported metals 21−23 to chalcogenide powders, specifically for nanocrystalline CoS2, and establish a methodology for studying these nonprecious catalysts in environment similar to that in PEMFCs; (2) to examine the performance of nanocrystalline CoS2 toward the ORR and to assess its stability; and (3) to evaluate to what extent the measured ORR performance is influenced by the quality of the CoS2 electrocatalytic layer (i.e., the degree of CoS2 dispersion or connectivity within the layer). In this paper, nanocrystalline CoS2 powder with pyrite structure was prepared using hydrothermal methods.24 Methodology for preparation of electrocatalytic layers from CoS2 with Nafion and testing of the quality of the obtained layers are presented. The electrocatalytic layers were investigated by rotating (ring) disk electrode (R(R)DE) measurements for the ORR. Kinetics of the ORR on nanocrystalline CoS2 is discussed with respect to the electrocatalytic layer quality and the degree of CoS2 dispersion within the electrocatalytic layer.

= 1 cm). Repeated sequence of milling of suspension of CoS2 in water at 400 rpm (rotations per minute) for 20 min and pauses of 5 min in between was employed for 12 h. Material Characterization. The powder diffraction patterns were obtained with a Siemens D5000 diffractometer (Cu Kα = 1.5418 Å radiation). The average crystallite size was estimated using the Scherrer formula: Dh =

λ βh cos θ h

(1)

Dh is the domain size of the diffraction line, λ is the wavelength of the Co source used (1.788 965 Å), βh is the width in radians of the diffraction peak measured at half-maximum intensity (fwhm) corrected for instrumental broadening, and θh is the angle of the particular hkl reflection. Microstructural analysis was performed with a Leo Ultra 55 field emission scanning electron microscope (SEM), equipped with an Oxford Inca energy-dispersive X-ray (EDX) spectroscopy system. EDX data were collected with an acceleration voltage of 15 kV. Dynamic light scattering (DLS) measurements were performed at a fixed angle of 173° with a Zetasizer Nano ZS (ZEN 3600, Malvern, UK) equipped with a He−Ne laser with a wavelength of 633 nm to determine a particle size distribution in liquid solutions. Electrochemical Characterization. For the ring−disk electrode (RRDE) experiments, a homemade glassy carbon (GC)/Pt disk−ring electrode of 0.189 cm2 disk area was employed using a homemade rotator and an ES-Elektronik (Sweden) speed controller. The ring potential was set at 1.0 V for the quantitative detection of the peroxide produced at the disk. A collection efficiency of 0.4 was determined from measurements of hexacyanoferrate(III) reduction. The electrochemical experiments were performed in a single-compartment cell with controlled atmosphere and temperature (293 K for all the experiments). The applied potential was controlled with an Autolab potentiostat (The Netherlands), and the measurements were carried out in a three-electrode cell. KCl saturated Ag/AgCl electrode was used as the reference electrode (−200 mV against the standard hydrogen electrode) and a gold plate as counter electrode. The electrolyte in all the measurements was aqueous 0.1 M HClO4 saturated with N2 or O2. The sweep rate in all ORR experiments was 10 mV s−1. Electrocatalytic Layer Preparation. CoS2 and an appropriate amount of carbon black was ground mechanically on an agate mortar. Weight ratios between CoS2 and C used for preparations of mixtures studied in this paper were 0.5, 1, and 2. The resulting powder (typically 20 μg) was dispersed in a mixture of water (typically 2 mL) and Nafion solution (5%) (typically 225 μL) and sonicated for 1 h. An appropriate amount of this suspension was transferred to the glassy carbon (GC) disk with an area of 0.189 cm2. The GC disk was polished to a mirror finish prior to each experiment using increasingly finer diamond suspensions (1.0, 0.3, and 0.05 μm, Struers, Sweden) followed by repeated sonication in water. The electrodes were prepared by placing the required amounts of the catalyst suspension on the glassy carbon disk and then left to evaporate overnight. The thickness of each catalytic layer was calculated from the density of the catalyst and of Nafion.23 10 μL of suspension dropped on the electrode was expected to form a layer of a thickness of 1 ± 0.1 μm.22 Thicker layers were prepared by increasing the amount of the suspension to be placed on the electrode. The full coverage of the glassy carbon support was checked by observation with an optical microscope. O2 reduction measurements carried out on



EXPERIMENTAL SECTION Chemicals. CoCl2·5H2O, Na2S2O3, and redistilled HClO4 (99.999%) 70%, CS2, and Nafion (5%) solution were supplied by Aldrich. Oxygen (99.95%), nitrogen (99.997%), and 10% H2/ 90%Ar mixture were obtained from Aga (Sweden). Milli-Q water (Millipore Inc.) was used in all experiments. XC-72R carbon black (Cabot Corp.) was used as electrocatalyst support. Material Preparation. CoS2 powders were prepared by hydrothermal synthesis.24,25 CoCl2·5H2O and Na2S2O3 were dissolved in ca. 35 mL of water in a 40 mL PEEK-lined stainless steel autoclave. The autoclave was placed in an oven (Memmert) at 150 °C for 12 h. After cooling the autoclave, a black powder was filtered and washed by CS2, ethanol, and water. The powder was dried in the air at 60 °C. The dispersed CoS2 material was obtained by ball milling of the as prepared CoS2 in water in a homemade Teflon container with two zirconium balls (diameter 24437

dx.doi.org/10.1021/jp307669k | J. Phys. Chem. C 2012, 116, 24436−24444

The Journal of Physical Chemistry C

Article

different electrode surfaces prepared with the same material showed good reproducibility.



RESULTS AND DISCUSSION Material Characterization. A typical X-ray diffractogram of the prepared CoS2 is shown in Figure 1. The pattern corresponds

Figure 1. X-ray diffraction pattern of the as-prepared CoS2.

to a pyrite structure (space group Pa3) with a lattice parameter of 5.533 Å, and no other phases were detected. The crystallite size determined using the Scherrer formula was ∼30 nm. The SEM imaging shows that the synthesis procedure leads to a formation of spherical structures (1−2 μm) that are composed of smaller particles ( 0.0 V). The reduced material is then more active in ORR during the backward sweep to more positive potentials. This is an interesting observation implying that the surface properties of the CoS2 can be tuned to enhance the ORR rate. Closer investigation of the surface properties before and after the proposed oxidation would be needed to elucidate this phenomenon. This study was especially focused on assessing the quality of the electrocatalytic layers prepared with CoS2 to find out whether it is possible to determine an intrinsic electrocatalytic activity of CoS2 and compare its performance with conventional electrocatalysts for the ORR. Besides the quality of dispersion of CoS2

Figure 4. Cyclic voltammograms measured on an electrode prepared with CoS2 in N2-saturated 0.1 M HClO4. The first potential sweep cycle measured with fresh electrode marked as the 1st cycle (solid line), 2nd (dashed line), and 10th (dotted line) are indicated in the figure. 24439

dx.doi.org/10.1021/jp307669k | J. Phys. Chem. C 2012, 116, 24436−24444

The Journal of Physical Chemistry C

Article

within the electrocatalytic layer that is likely closely related to the agglomeration properties in electrode suspension discussed above, another important factor could be electrical connectivity within the electrocatalytic layer as well as good accessibility for O2. These parameters could significantly differ between the CoS2-based layers and layers prepared with conventional M/C catalysts and affect the assessment of electrocatalytic performance. In the case of M/C catalysts, the only parameter that needs to be optimized is the catalyst/ionomer ratio. In the case of a catalytic powder that is mechanically mixed with carbon, however, the textural properties of the layer and subsequently the connectivity will depend on the CoS2/C ratio as well. A methodology to determine the quality of electrocatalytic layers has been developed by Behm et al. for Pt/C,21 which can be indeed generalized for any type of M/C catalyst. This methodology comprises the measurement of current−potential dependence on RDE rotation rate and subsequent analysis. This approach has been extended to the CoS2/C mixture used here. The resulting polarization curves measured on electrocatalytic layers prepared with different ratios of the as prepared CoS2 and C are shown in Figure 6. The current−potential dependence on rotation rate can be described by the Koutecky−Levich (K−L) equation:26 1 1 1 1 1 = + =− − b 2/3 −1/6 b 1/2 j jk jd nFk O2cO2 0.62nFDO2 ν cO2ω (2)

j is the measured current density (A cm−2), and jk and jd are the kinetic and diffusion-limited current densities, respectively. n and kO2 are the potential-dependent number of electrons transferred per single O2 molecule reduced and the electrochemical rate constant (cms−1), respectively, F is the Faraday constant (96 485 C mol−1), ω is the rotation rate (rad s−1), cbO2 is the bulk concentration of O2, which for a saturated solution of 0.1 M HClO4 is 1.26 × 10−6 mol cm−3,26,27 DO2 is the diffusion coefficient of O2 (1.93 × 10−5 cm2 s−1),27 and ν is the kinematic viscosity of the solution (0.01 cm2 s−1).28 A good criterion to judge the quality of electrocatalytic layers according to Behm et al.21 is the intercept of K−L plots, which in case of improperly connected or too thick layer gives unrealistically high values (low values of kO2). This method was used here to optimize the ratio between CoS2 and carbon used for electrode preparation. Electrodes with CoS2/C weight ratios of 0.5, 1, and 2 were prepared. As shown in Figures 6A−C, a considerable difference between polarization curves measured for these layers in O2-saturated solution can be observed. The current attributable to ORR is in general highest for the CoS2/C ratio of 1. To estimate the connectivity within layers with different CoS2/C ratio, intercept values from the K−L plots were determined as shown in Figure 6D. Indeed, the intercept in Figure 6 D shows the lowest value for the ratio of 1. The optimal CoS2/C weight ratio appears rather high if compared to metal/ carbon (M/C) loadings used for precious-metal-based catalysts, where the M/C is typically 0.2. This result reflects, however, different function of carbon in the films prepared with the CoS2 powder. Here, carbon acts as a binder that improves mechanical stability and connectivity of the layer, and the resulting conducting “framework” of the layer will depend on both the quality of the carbon and CoS2, as opposed to the M/C case where the framework is formed by the carbon only. The lower or higher CoS2/C weight ratios likely lead to a formation of

Figure 6. Polarization curves measured in O2-saturated 0.1 M HClO4 on electrodes prepared with different CoS2/C weight ratio of (A) 0.5, (B) 1, and (C) 2. Number (1) indicates background measured in N2-saturated solutions. Rotation rates: (2) 200, (3) 400, (4) 800, (5) 1000, and (6) 1500 rpm. Sweep rate = 10 mV s−1. (D) Koutecky−Levich plots constructed from data in (A−C) at −0.2 V with the CoS2/C indicated in the figure.

inhomogeneity within the electrocatalytic layer. This demonstrates the high importance of the presented layer quality evaluation, since nonoptimized catalytic layer will give poor results irrespectively of the “real” electrocatalytic activity in the ORR of any material. ORR Selectivity on the Optimized CoS2 Electrode. An important factor for evaluating the electrocatalytic activity of the ORR is the selectivity toward H2O2 production. As reported elsewhere, the ORR can proceed via a two-electron or fourelectron pathway to form hydrogen peroxide or water, respectively. The information on the number of electrons transferred per single O2 reduced, n, can be extracted by Koutecky−Levich analysis according to eq 2. Figure 7 shows an example of the applicability of this equation to the ORR and the corresponding potential dependence of n, shown in Figure 7B. The values of n closely follow the ORR selectivity toward H2O2 production according to the equation22 24440

dx.doi.org/10.1021/jp307669k | J. Phys. Chem. C 2012, 116, 24436−24444

The Journal of Physical Chemistry C

Article

Figure 7. (A) Koutecky−Levich (K−L) plots constructed from data in Figure 6B at different potentials as indicated in the figure legend. (B) Potential dependence of values of n calculated from the K−L analysis.

SH2O2 =

j(O ,2e−) 2

j(O ,4e−) + j(O ,2e−) 2

2

=

4−n 2

Figure 8. (A) Polarization curve measured in O2-saturated 0.1 M HClO4 on electrodes prepared with CoS2 (CoS2/C = 1) at 200 rpm (circles, left y-axis) and simultaneously recorded ring current (ER = 1 V) (squares, right y-axis). (B) H2O2 selectivity calculated from the ring H2O2 detection (squares) and from the K−L analysis (circles).

(3)

where j(O2,2e−) and j(O2,4e−) are the fluxes of O2 per unit area leading to H2O2 and H2O, respectively. The K−L analysis is depicted in Figure 7, and the K−L plots obtained from the RDE results in Figure 6B are shown here. According to eq 2, the dependence of values of n on potential can be determined. The results in Figure 7 show that, as previously also observed for gold supported on carbon,22 the reduction selectivity for H2O2 decreases as the potential is made more negative, until it reaches values close to zero where almost all O2 is being converted to water at potentials below 0.0 V. The H2O2 production at the electrodes was also followed by independent detection at the Pt ring, since the K−L analysis may not provide reliable results above 0.3 V. The polarization curve measured in O2-saturated solution followed by the simultaneous ring H2O2 detection is shown in Figure 8A. From a simple relation between disk and ring currents the H2O2 selectivity can be also calculated according to SH2O2 = (2 × IR/N)/(IR/N − ID), where IR and ID are the currents measured on the ring and disk electrodes, respectively. The comparison between the selectivity calculated from the K−L analysis (Figure 7) and the ring detection (Figure 8A) is shown in Figure 8B. A very good agreement between results obtained by these independent methods was observed similarly to investigations carried on Au/ C.22 This indicates that the H2O2 produced during the ORR is being removed efficiently from the electrocatalytic layer. The reason for this can be that the further reduction of H2O2 on CoS2 is very slow at potentials above 0.0 V and/or that the ORR proceeds mainly close to the surface of the electrocatalytic layer. Comparison between values of n obtained during potential sweeps toward more negative potentials and back to more positive potentials shows differences that can be ascribed to a lower selectivity of CoS2 toward H2O2 production during the backward sweep to more positive potentials. The decreased

H2O2 selectivity during the backward sweep is only observed when the potential limit of −0.2 V is crossed during the potential sweep cycle as illustrated in Figure 7. Hence, the ORR enhancement is caused by a reduction process that occurs below −0.2 V and reflects a property difference between the reduced and oxidized CoS2 surface. Electrocatalytic Activity of CoS2 in ORR. In addition to n, another important parameter that describes the electrocatalytic performance of CoS2 is the kinetic constant kO2 in eq 2. As reported previously, an apparent kinetic constant (kapp) can be extracted by a simplified kinetic analysis of ORR performance on electrocatalytic layers prepared with different loadings of electrocatalyst. This approach can be also used to determine whether the ORR is controlled by reaction kinetics or by mass transport effects,22 since the kinetic control implies linear increase of kapp with CoS2 loading on the electrode. The control by reaction kinetics is desirable since it allows extraction of intrinsic kinetic parameters that can be used for further comparisons between different electrocatalytic materials. The CoS2/C ratio was fixed at 1 according to results presented above and electrocatalytic layers with thicknesses of 1 ± 0.1, 1.5 ± 0.2, 2.0 ± 0.2, and 3.0 ± 0.3 μm were prepared. The polarization curves recorded on these layers are shown in Figure 9. Nonlinear regression was applied according to eq 4 to extract apparent kinetic parameters:22 ⎛ exp(38.39α(E − E 0)) j = 1/⎜⎜ − Fn(E)kappcOb2 ⎝ +

24441

⎞ ⎟ 0.62n(E)FDO2 2/3ν−1/6cOb2ω1/2 ⎟⎠ 1

(4)

dx.doi.org/10.1021/jp307669k | J. Phys. Chem. C 2012, 116, 24436−24444

The Journal of Physical Chemistry C

Article

Figure 9. Polarization curve measured in O2-saturated 0.1 M HClO4 on electrodes prepared with different amounts of CoS2 (CoS2/C = 1) at 200 rpm. The calculated electrocatalytic layer thickness is indicated in the figure.

A nonlinear curve fit to eq 4 was performed with only one overall value of the standard rate constant, kapp but with a potential dependent value of n(E) to take into account the reduction pathway of peroxide to water. This simplified approach has been shown to lead to similar results compared with a more rigorous approach considering the full oxygen reduction reaction sequence22 and was found sufficient for the initial tests carried out for CoS2 layers in this paper. To extract reaction kinetics information, the dependence of n on potential must be explicitly included in the analysis. From the K−L results shown in Figure 6, n(E) was approximated by a combination of two functions: n(E) = aE + b for n > 2; else n = 2 (a, b = constants derived from the K−L analysis). The nonlinear regression facility in the Origin software33 was used for these calculations. The current density j employed in the regression analysis was calculated using the geometric electrode area, and therefore, kapp is the overall rate constant for oxygen reduction calculated per unit geometric electrode area. The reduced chisquare for all the fits was less than 5 × 10−9, and the values reported are averages of the fitting parameters for different rotation rates obtained for each j−E data set. Examples of the fitted curves with some fitting details are shown in Supporting Information Figure S1. The apparent kinetic constant, kapp, was found to be approximately constant for all considered layer thicknesses of (7.4 ± 0.4) × 10−4 cm s−1. This indicates that the current difference observed in Figure 9 only reflects difference in the selectivity toward H2O2 production and, importantly, that not all CoS2 present within the electrocatalytic layer is utilized in the ORR. The ORR likely proceeds mainly close to the electrode surface, and the utilization of a material from within the layer is limited by mass transport effects. It should be noted that the K−L plots intercept values were similar for all the considered layer thicknesses; thus, the interlayer electrical connectivity remains unaffected by the film thickness, which agrees with a very good conductivity of CoS2. The obtained results are different from those obtained with Au/C supported catalyst and reflect different transport conditions within CoS2/C composite layers. Improved Oxygen Reduction Rate on the Dispersed CoS2. To achieve better utilization of CoS2 within the electrocatalytic layer, electrodes were prepared with the dispersed CoS2. Polarization curves measured on electrocatalytic layers prepared with the dispersed CoS2 are shown in Figure 10. A clear improvement in the ORR activity compared to the layer prepared with the as-prepared CoS2 is observed. The potential

Figure 10. (A) Polarization curve measured in O2-saturated 0.1 M HClO4 on electrodes prepared with the as-prepared (squares) and the dispersed (circles) CoS2 (CoS2/C = 1) at 200 rpm. Sweep rate = 10 mV s−1. (B) Values of n as a function of potential calculated form the K−L analysis.

dependence of values of n obtained from K−L analysis is shown in Figure 10B. As evident from these results, the higher electrocatalytic activity of the dispersed CoS2 in the ORR can partly be attributed to the lower selectivity toward H2O2 formation (see Figure 10B). The nonlinear curve fit according to eq 4, however, gives the apparent kinetic constant of (2.2 ± 0.2) × 10−3 cm s−1, which is ∼3-fold higher than in the case of the as-prepared CoS2 (see also Supporting Information Figure S1). Since the lower H2O2 selectivity (SH2O2) is already taken into account in the nonlinear regression analysis, it shows clearly that O2 reduction kinetics are also enhanced at the layer with the dispersed CoS2. This demonstrates that the mass transport of O2 plays an important role, and the transport of reactant and products can be enhanced by the improved dispersion of CoS2 within the electrocatalytic layer. Indeed, the lower H2O2 production with dispersed CoS2 then likely reflects a longer residence time of H2O2 within the electrocatalytic layer and higher probability of its reduction due to increased reduction site availability.22 CoS2 Stability during the ORR. The main issue in the ORR catalysis in acid media on non-precious-metal-based materials such as CoS2 is the stability against the rather aggressive environment of a PEMFC. The results shown above demonstrate that the CoS2 studied in this paper shows sufficient stability in such an environment (Nafion) in a time scale of given experiments (∼1 h), since good reproducibility of the measured polarization curves within this time range was observed. The operation times required in a real fuel cell, however, are much longer. To at least approximate a long-term operation in a fuel cell, the electrocatalytic layer prepared with the ball-milled CoS2 was tested for long time durability by a simple experiment. The potential was fixed at a potential where the ORR proceeds at a high rate (0.0 V) and the electrode rotation rate was set at 1500 rpm (current density of ∼5 mA cm−2) for 15 h. Results of this experiment are shown in Figure 11. As can be seen in this figure, no significant current difference has been observed during the 24442

dx.doi.org/10.1021/jp307669k | J. Phys. Chem. C 2012, 116, 24436−24444

The Journal of Physical Chemistry C

Article

the chalcogenide group, such as RuxSey or CoSe2. Importantly, an electrocatalytic layer composed of nanocrystalline CoS2, carbon, and Nafion was prepared, and the CoS2 material stability was probed within the environment compatible with the proton exchange membrane fuel cell technology. CoS2 showed a surprisingly high stability in Nafion, and the CoS2 electrodes did not show any sign of deterioration after a 15 h long durability test under oxygen reducing conditions (∼5 mA cm−2, 0.0 V vs Ag/AgCl). Oxygen reduction reaction kinetics analysis was performed to study properties of prepared layers with CoS2. It showed that special attention has to be paid to the degree of dispersion of CoS2 material within the electrocatalytic layer. The resulting layer quality influences the final performance in terms of both selectivity toward H2O2 production and the electrocatalytic activity of studied electrodes, irrespective of the “real” intrinsic catalytic activity of the material employed. The asprepared CoS2 electrodes have shown high selectivity toward H2O2 up to 80% at potentials around 0.4 V. Irrespective of the CoS2 film quality, the H2O2 production rate dropped to zero below 0.0 V, and the oxygen reduction proceeded along the O2 reduction pathway to yield H2O. By employing a finer CoS2 material, it was possible to decrease the O2 reduction selectivity to H2O2 by improving dispersion of CoS2 within the electrocatalytic layers. This led to significant enhancement of the electrocatalytic activity of CoS2 toward ORR, which was achieved by both suppressing the O2 reduction channel toward H2O2 production and enhancing the O2 reduction rate by improving the layer accessibility to O2.

Figure 11. Current measured at a CoS2 electrode in O2-saturated 0.1 M HClO4 held at constant potential of 0.0 V at 1500 rpm for 15 h. Polarization curves measured at same conditions before (red line) and after (black dashed line) this experiment are shown in the inset to the figure.

time scale of this experiment. To finalize the durability evaluation, a full RDE analysis as described above was carried out on the same electrode. For illustration, the polarization curve recorded on this electrode before and after this treatment is shown in the inset to Figure 11. No difference was observed, and the same parameters in terms of the n-values’ dependences on potential and the kinetic constant were obtained. Therefore, it can be concluded that no deterioration of the CoS2-containing electrode quality has been observed during the long-term ORR. It should be mentioned, however, that the electrode polarization seems also to be a requirement for the stabilization most likely by a form of cathodic protection known from corrosion science. Our results indicate that when the material is left in a wet environment (or electrode in the electrochemical cell at open circuit potential), the CoS2 slowly deteriorates and turns into soluble sulfate on the surface. The very good durability during ORR measurements represents an excellent result that suggests CoS2 material with pyrite structure as good candidate for a catalyst to be employed in low-cost PEMFC’s. Indeed, the possible operation potential of ∼0.3 V vs Ag/AgCl (∼0.5 V vs SHE) is not as high as for Pt/Cbased materials, and therefore the energetic efficiency of CoS2based PEMFCs would be below ∼30%.1 Employing these or analogous materials in, e.g., portable fuel cells,34 however, does not necessarily pose such high requirements upon efficiency. The results here show that the relatively high H2O2 production above 0.3 V can be suppressed considerably by increasing the degree of dispersion of CoS2 within the electrocatalytic layer. The achieved electrocatalytic activity of CoS2 toward ORR then appears to outperform or at least matches that of other comparable electrocatalytic materials such as RuxSey.10,35−37 On the other hand, the relatively high portions of H2O2 produced at potentials considerably higher than on Au22 or Au1−xPdx38 suggest another intriguing alternative: to tune the performance of this catalyst in the opposite direction, toward application in a fuel cell for H2O2 synthesis.38 Further investigations of the electrocatalysis on CoS2 would be helpful, to understand not only the tunable H2O2 selectivity and overall electrocatalytic activity in ORR but also the remarkable stability that seems to be connected with the pyritetype crystalline structure of CoS2.



ASSOCIATED CONTENT

S Supporting Information *

Figure S1. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected] (J.S.J.); [email protected] (E.A.). Present Address †

Materials Science Division, Argonne National Laboratory, Argonne, IL 60439.

Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS Financial support from the Swedish Research Council (Grant 621-2008-5536) and the European Commission through the FP7 Initial Training Network “ELCAT” (Grant Agreement No. 214936-2) is gratefully acknowledged. Special thanks to Patrick Steegstra for help with acquiring the SEM images, to Jenny Perez-Holmberg for help with the DLS experiments, and also to Dr. Nemanja Danilovic for critical review of the manuscript.



REFERENCES

(1) (a) Vielstich, W.; Lamm, A.; Gasteiger, H. A. Handbook of Fuel Cells, Fundamentals, Technology and Applications; Wiley: New York, 2003. (b) Hoogers, G. Catalysts for the Proton Exchange Membrane Fuel Cell; CRC: Boca Raton, FL, 2003. (2) Gasteiger, H. A.; Markovic, N. M. Science 2009, 324, 48−49. (3) Gasteiger, H. A.; Kocha, S. S.; Sompalli, B.; Wagner, F. T. Appl. Catal., B 2005, 56, 9−35. (4) Stamenkovic, V. R.; Fowler, B.; Mun, B. S.; Wang, G.; Ross, P. N.; Lucas, C. A.; Markovic, N. M. Science 2007, 315, 493−497.



CONCLUSIONS Nanocrystalline CoS2 with pyrite structure was synthesized using hydrothermal methods. The prepared CoS2 showed very good activity toward the oxygen reduction reaction, higher than or at least comparable to the performance of other active materials in 24443

dx.doi.org/10.1021/jp307669k | J. Phys. Chem. C 2012, 116, 24436−24444

The Journal of Physical Chemistry C

Article

(5) V Stamenkovic, V. R.; Mun, B. S.; Arenz, M.; Mayrhofer, K. J. J.; Lucas, C. A.; Wang, G.; Ross, P. N.; Markovic, N. M. Nat. Mater. 2007, 6, 241−247. (6) Lefevre, M.; Proietti, E.; Jaouen, F.; Dodelet, J. Science 2009, 324, 71−74. (7) Alonso-Vante, N.; Tributsch, H. Nature 1986, 323, 431−432. (8) Alonso-Vante, N.; Jaegermann, W.; Tributsch, H.; Honle, W.; Yvon, K. J. Am. Chem. Soc. 1987, 109, 3251−3257. (9) Solorza-Feria, O.; Ellmer, K.; Giersig, M.; Alonso-Vante, N. Electrochim. Acta 1994, 39, 1647−1653. (10) Malakhov, V.; Nikitenko, S. G.; Savinova, E. R.; Kochubey, D. I.; Alonso-Vante, N. J. Phys. Chem. B 2002, 106, 1670−1676. (11) http://www.platinum.matthey.com/prices/price_charts.html. (12) Feng, Y.; He, T.; Alonso-Vante, N. Chem. Mater. 2008, 20, 26−28. (13) Feng, Y. J.; He, T.; Alonso-Vante, N. Fuel Cells 2010, 10, 77−83. (14) Feng, Y.; He, T.; Alonso-Vante, N. Electrochim. Acta 2009, 54, 5252−5256. (15) Lowson, H. T. Chem. Rev. 1982, 82, 482−497. (16) Ahlberg, E.; Broo, A. E. J. Electrochem. Soc. 1997, 144, 1281−1286. (17) Ahlberg, E.; Broo, A. E. Int. J. Miner. Process. 1996, 46, 73−89. (18) Susac, D.; Sode, A.; Zhu, L.; Wong, P. C.; Teo, M.; Bizzotto, D.; Mitchell, K. A. R.; Parsons, R. R.; Campbell, S. A. J. Phys. Chem. B 2006, 110, 10762−10770. (19) Susac, D.; Zhu, L.; Teo, M.; Sode, A.; Wong, K. C.; Wong, P. C.; Parsons, R. R.; Bizzotto, D.; Mitchell, K. A. R.; Campbell, S. A. J. Phys. Chem. C 2007, 111, 18715−18723. (20) Zhu, L.; Susac, D.; Teo, M.; Wong, K. C.; Wong, P. C.; Parsons, R. R.; Bizzotto, D.; Mitchell, K. A. R.; Campbell, S. A. J. Catal. 2008, 258, 235−242. (21) Schmidt, T. J.; Gasteiger, H. A.; Stäb, G. D.; Urban, P. M.; Kolb, D. M.; Behm, R. J. J. Electrochem. Soc. 1998, 145, 2354−2358. (22) Jirkovsky, J. S.; Halasa, M.; Schiffrin, D. J. Phys. Chem. Chem. Phys. 2010, 12, 8042−8052. (23) Maillard, F.; Martin, M.; Gloaguen, F.; Leger, J. M. Electrochim. Acta 2002, 47, 3431−3440. (24) Chen, X.; Fan, R. Chem. Mater. 2001, 13, 802−805. (25) Bjorling, A. Synthesis, characterization and electrocatalytic properties of Co1‑xFexS2, Department of Chemistry, University of Gothenburg, 2009:10. Master Thesis, 2009. (26) Bard, A. J.; Faulkner, L. R. Electrochemical Methods − Fundamentals and Application, 2nd ed.; John Wiley & Sons: New York, 2001. (27) Stamenkovic, V.; Schmidt, T.; Ross, P. N.; Markovic, N. M. J. Electroanal. Chem. 2003, 554−555, 191−199. (28) CRC Handbook of Chemistry and Physics, 82nd ed.; Lide, D. R., Ed.; CRC Press: Boca Raton, FL, 2001. (29) Tammeveski, K.; Kontturi, K.; Nichols, R. J.; Potter, R. J.; Schiffrin, D. J. J. Electroanal. Chem. 2001, 515, 101−112. (30) Sarapuu, A.; Vaik, K.; Schiffrin, D. J.; Tammeveski, K. J. Electroanal. Chem. 2003, 541, 23−29. (31) Vaik, K.; Sarapuu, A.; Tammeveski, K.; Mirkhalaf, F.; Schiffrin, D. J. J. Electroanal. Chem. 2004, 564, 159−166. (32) Mirkhalaf, F.; Tammeveski, K.; Schiffrin, D. J. Phys. Chem. Chem. Phys. 2004, 6, 1321−1327. (33) OriginPro, Version 8.0, Origin Lab Corporation. (34) http://www.matthey.com ; http://www.fuelcelltoday.com/. (35) Lewera, A.; Timperman, L.; Roguska, A.; Alonso-Vante, N. J. Phys. Chem. C 2011, 115, 20153−20159. (36) Timperman, L.; Alonso-Vante, N. Electrocatalysis 2011, 2, 181− 191. (37) Kulesza, P. J.; Miecznikowski, K.; Baranowska, B.; Skunik, M.; Fiechter, S.; Bogdanoff, P.; Dorbandt, I. Electrochem. Commun. 2007, 8, 904−908. (38) Jirkovsky, J. S.; Panas, I.; Halasa, M.; Ahlberg, E.; Romani, S.; Schiffrin, D. J. J. Am. Chem. Soc. 2011, 133, 19432−19441.

24444

dx.doi.org/10.1021/jp307669k | J. Phys. Chem. C 2012, 116, 24436−24444