H.
Liu
and
A.
Manthiram
*
Electrochemical Energy Laboratory & Materials Science and Engineering Program, The University of Texas at Austin, Austin, TX 78712, USA. E-mail: rmanth@mail.utexas.edu; Fax: +1 512 4717681; Tel: +1 512 4711791
First published on 14th October 2008
Carbon-supported Pd4Co nanoalloy samples (denoted as Pd4Co/C) have been synthesized by a modified polyol reduction process, involving a mixture of ethylene glycol (a polyol) and a small amount of sodium borohydride as reducing agents, and evaluated as electrocatalysts for the oxygen reduction reaction (ORR) in fuel cells. The synthesis process offers Pd4Co nanoparticles with an average particle size of 4.6 nm and a narrow size distribution, which increases to 7.3 and 11.0 nm on annealing, respectively, at 350 and 500 °C in 10% H2–90% Ar. Although the as-prepared Pd4Co/C sample exhibits instability during ORR, annealing at 350 and 500 °C enhances the stability significantly due to an increasing degree of alloying between Pd and Co and increasing particle size. While both Pd/C and Pt/C exhibit a drastic decline in catalytic activity on annealing at 350 and 500 °C due to a significant increase in particle size, the Pd4Co/C sample maintains high catalytic activity even after annealing at 350 °C due to an increase in the degree of alloying. The Pd4Co/C catalyst offers an added advantage of excellent tolerance to methanol poisoning and lower cost compared to the conventional Pt/C catalyst.
Broader contextRapid depletion of fossil fuels and growing environmental concerns pose serious scientific and technological challenges to address the increasing global demand for energy. Fuel cells are appealing in this regard, but the high cost of the currently used platinum electrocatalysts is an impediment to the commercialization of the fuel cell technology. Palladium-based alloy catalysts have recently been found to exhibit high catalytic activity for the oxygen reduction reaction in fuel cells, but with the multiple metals involved, synthesis and processing play a significant role on the degree of alloying, chemical homogeneity, particle size and distribution, and catalytic activity. To address this issue, we present here a novel modified polyol reduction method to obtain carbon-supported Pd4Co nanocrystals with a high degree of alloying and dispersion on the carbon support while keeping the particle size small. The Pd4Co catalysts thus obtained exhibit high catalytic activity for the oxygen reduction reaction with excellent tolerance to methanol in direct methanol fuel cells. |
Recently, we reported that palladium-based alloy electrocatalysts such as Pd–Co–Au, Pd–Ti, and Pd–Co–Mo20–22 show catalytic activity comparable to that of Pt for ORR. However, for structure and surface sensitive reactions such as the ORR, particle size as well as dispersion and compositional homogeneity of the alloy clusters on the carbon support are important factors to obtain good catalytic activity in practical applications. Generally, the Pd-based alloys supported on carbon have been prepared by a conventional reduction of the metal ions with alkali metal borohydrides or by a reduction of the metal ions within microemulsions with borohydrides, followed by heat treatment at 500–900 °C in a reducing atmosphere or by simple thermal treatment of the constituents with H2 to promote alloy formation.20–23 Unfortunately, the faster reduction with the strongly reducing borohydrides and the high heating temperatures lead to undesired particle growth, which results in a decrease in surface area and catalytic activity. Development of novel, controlled synthesis approaches that can provide enhanced alloy formation at the lowest temperatures possible while keeping the particle size small at the nanoscale is critical to realize the full potential of the Pd-based alloy electrocatalysts.
Reduction of metal ions by polyols has been pursued by several groups for the synthesis of metal nanoparticles.24–30 The polyol process, which exploits high-boiling polyalcohol solvents as mild reducing agents at moderate temperatures, offers a facile route for obtaining multi-metallic alloy nanoparticles without requiring significant solid state diffusion over large atomic distances. Size and shape controlled nanocrystals with particle size as small as 2 nm has been obtained by the polyol process.28 However, the conventional polyol process often employs large amounts of stabilizers such as the poly(vinylpyrrolidone) (PVP) polymer to produce mono-dispersed, small particles, resulting in polymer-protected colloidal dispersions. It is difficult to remove the polymer completely from such colloidal dispersions, and the presence of polymer is undesirable for practical applications such as electrocatalysts in fuel cells. In addition, the large differences in the standard reduction potentials of ions like Pd2+ and Co2+ and the mild reducing power of polyols lead to the formation of hollow nanospheres of the easily reducible components like Pd with no Co.31 In other words, it is difficult to reduce ions like Co2+ with polyol due to their low reduction potential compared to that of Pd2+.
We present in this paper a modified polyol reduction process, involving a mixture of a polyol (ethylene glycol) and a small amount of sodium borohydride as reducing agents as well as much lower amounts of the protective polymer than that used in conventional polyol process. While the use of borohydride helps to reduce the Co2+ ions easily, the lower amount of polymer helps to obtain electrocatalysts nearly free from the protective polymer contamination. The reduction products formed are then annealed at a moderate temperature of 350 °C to obtain carbon-supported Pd4Co nanocrystals with a high degree of alloying and dispersion on the carbon support while keeping the particle size small. Characterization of the Pd4Co/C electrocatalysts by X-ray diffraction and transmission electron microscopy, an investigation of their electocatalytic activity for ORR in single cell PEMFC and DMFC and by rotating disk electrode (RDE) measurements, and an assessment of their stability in a three-electrode test cell in the presence and absence of methanol are presented.
The electrocatalytic activity for ORR was also assessed in single cell DMFC. In this case, the anode catalyst was a commercial 40 wt% PtRu/C (E-TEK) and the cathode catalysts were Pd/C-350, Pt/C-350, Pd4Co/C-350, and a commercial 20 wt% Pt/C. The electrodes were prepared by the same procedure described above for PEMFC. The metal or alloy loadings in the anode and cathode were, respectively, 0.8 and 0.25 mg cm−2. The electrochemical performances in DMFC of the MEAs fabricated with Nafion 112 membrane (DuPont) were evaluated by feeding a preheated 1 M methanol solution into the anode and humidified oxygen into the cathode. The temperatures of the preheated methanol solution and humidified oxygen were same as that of the cell temperature (65 °C). Further details of the PEMFC and DMFC single cell evaluations are described in ESI.†
Sample | Average crystallite size from XRDa (nm) | Average particle size from TEMb (nm) |
Pd![]() ![]() |
---|---|---|---|
a Obtained by using the (111) reflections; the numbers in parentheses give the standard deviation. b The numbers in parentheses give the standard deviation. | |||
Pd4Co/C-Ap | 4.4 (0.2) | 4.6 (2.1) | 83![]() ![]() |
Pd4Co/C-350 | 8.5 (0.3) | 7.3 (4.9) | 82![]() ![]() |
Pd4Co/C-500 | 10.8 (0.2) | 11.0 (4.2) | 81![]() ![]() |
Pd/C-Ap | 4.8 (0.2) | — | — |
Pd/C-350 | 8.7 (0.3) | — | — |
Pd/C-500 | 11.5 (0.3) | — | — |
Pt/C-Ap | 4.5 (0.2) | — | — |
Pt/C-350 | 12.1 (0.3) | — | — |
Pt/C-500 | 18.3 (0.2) | — | — |
![]() | ||
Fig. 1 XRD patterns of the Pd4Co/C and Pd/C catalysts before and after annealing in 10% H2–90% Ar atmosphere at various temperatures. The dashed vertical line indicates the standard 2θ value corresponding to the (111) reflection of Pd metal. |
Fig. 2 and 3 compare the TEM photographs and the particle size distributions of the Pd4Co/C samples before and after annealing at various temperatures. As seen, the nanosize particles have predominantly spherical morphology with a uniform distribution on the carbon support. The histograms and a Gaussian fit of the TEM data in Fig. 3 indicate a narrow particle size distribution, although the 350 °C sample has a slightly broader distribution than the as-prepared and 500 °C samples. The data indicate an average particle size value of 4.6, 7.3, and 11.0 nm, respectively, for the Pd4Co/C-Ap, Pd4Co/C-350, and Pd4Co/C-500 samples, which agree closely with the crystallite size values obtained from the XRD data (Table 1).
![]() | ||
Fig. 2 TEM micrographs of (a) Pd4Co/C-Ap, (b) Pd4Co/C-350, and (c) Pd4Co/C-500. |
![]() | ||
Fig. 3 Particle size distribution histograms of (a) Pd4Co/C-Ap, (b) Pd4Co/C-350, and (c) Pd4Co/C-500. |
Fig. 4 shows the high resolution TEM (HRTEM) photographs of the Pd4Co/C-Ap, Pd4Co/C-350, and Pd4Co/C-500 samples. The data indicate the presence of predominantly twinned or polycrystalline nanocrystals. For each sample, the spacing between the (111) planes are also indicated in Fig. 4. The decrease in the (111) spacing from 0.2259 to 0.2222 nm with increasing annealing temperature confirm the substitution of smaller Co atoms for larger Pd atoms and an increase in the degree of alloying, which is consistent with the XRD data in Fig. 1. Moreover, EDS elemental analysis of the nanocrystals indicate a Pd:
Co atom ratio of around 82
:
18 (Table 1), which is close to the nominal composition of 80
:
20.
![]() | ||
Fig. 4 High resolution TEM micrographs of the Pd4Co/C samples: (a) Pd4Co/C-Ap, (b) Pd4Co/C-350, and (c) Pd4Co/C-500. |
The TEM and XRD data thus indicate that the addition of a small amount of sodium borohydride to the polyol reduction process helps to reduce nearly all the Co2+ ions into Co, which is difficult to achieve by the conventional polyol reduction process employing polyol alone as a reducing agent. On the other hand, the controlled, slow reduction within the polymer-protected colloidal dispersions with the predominantly present mild reducing agent, polyol, offers small particle size with a narrow size distribution. The short diffusion distance within the small (∼5 nm), as-prepared nanoparticles helps to achieve a high degree of alloying at low annealing temperatures of 350–500 °C. The modified polyol method employing a mixture of a polyol and borohydride as reducing agents offers an effective approach to obtain Pd–Co alloys with a high degree of alloying while keeping the particle size small. In addition, FT-IR measurements revealed the absence of PVP in the samples.
![]() | ||
Fig. 5 Cyclic voltammograms (CV) of the Pd4Co/C and Pd/C samples in 0.5 M H2SO4 solution at a sweep rate of 20 mV s−1 at room temperature: cycling data of (a) Pd4Co/C-Ap, (b) Pd4Co/C-350, (c) Pd4Co/C-500, (d) Pd/C-Ap, and (e) Pd/C-350 and (f) the 15th cycle data of Pd4Co/C-Ap, Pd4Co/C-350 and Pd4Co/C-500. The inset in (f) compares the 15th cycle electrochemical surface area of the three samples. |
The decline in electrochemical surface area with cycling becomes less pronounced with increasing annealing temperature, and the Pd4Co/C-500 sample exhibits little decline in electrochemical surface area on going from 2nd to 15th cycle. The faster decline in the electrochemical surface area of the Pd4Co/C-Ap sample on cycling in the acidic environment could be due to both the leaching out of the poorly alloyed Co and the smaller particle size (high particle surface area). The former is consistent with the increased degree of alloying between Pd and Co and improved homogeneity on annealing at 350 and 500 °C as indicated earlier by the XRD and TEM data as well as the recent theoretical calculations33 indicating that Co in Pd–Co alloy is likely to be stripped at lower potentials in acidic solutions. The latter is consistent with the faster decline in the electrochemical surface area of the Pd/C-Ap sample (with no Co) on cycling (Fig. 5(d)) due to smaller particle size while the annealed Pd/C-350 sample exhibits improved stability on cycling (Fig. 5(e)) due to larger particle size. However, the decline in electrochemical surface area on cycling is smaller for the Pd/C-Ap sample compared to that for the Pd4Co/C-Ap sample as both the leaching out of Co and the smaller particle size contribute to the instability. The decline in electrochemical surface area with smaller particle size (high particle surface area) could be due to either sintering of the smaller particles on potential cycling or easier dissolution in acid. Although one could differentiate these possibilities by a TEM evaluation of the catalyst after CV cycling, the use of Nafion in the microelectrode preparation makes it difficult to establish this unambiguously at the present time. Fig. 5(f) compares the CV scans recorded at the 15th cycle for Pd4Co/C-Ap, Pd4Co/C-350 and Pd4Co/C-500. The inset in Fig. 5(f) compares the peak area for the surface oxide reduction peak at the 15th cycle for the three samples. Although the onset potential does not vary significantly, the electrochemical surface area decreases with increasing annealing temperature due to an increase in particle size and a decrease in particle surface area.
![]() | ||
Fig. 6 Variations with annealing temperature of the catalytic activity for ORR and the surface area of the Pd4Co/C sample. The inset shows the relationship between the electrochemical activity and surface area. |
![]() | ||
Fig. 7 Comparison of the catalytic activities for ORR of the Pd4Co/C, Pd/C, and Pt/C samples before and after annealing at various temperatures: (a) comparison of the polarization curves recorded in single cell PEMFC of Pt/C-350, Pd/C-350 and Pd4Co/C-350 with the inset showing a comparison of the catalytic activity of the three samples at 0.73 V; (b) comparison of the linear polarization curves recorded with RDE of Pt/C-350, Pd/C-350 and Pd4Co/C-350 with the inset showing a comparison of the catalytic activity of the three samples at 0.73 V; (c) comparison of the polarization curves recorded in single cell PEMFC of Pt/C-Ap, Pd/C-Ap, and Pd4Co/C-Ap with the inset showing a comparison of the catalytic activities extracted from the single cell tests of the three samples at 0.73 V; and (d) comparison of the polarization curves recorded in single cell PEMFC of Pt/C-500, Pd/C-500, and Pd4Co/C-500 with the inset showing a comparison of the catalytic activities extracted from the single cell tests of the three samples at 0.73 V. |
As seen in Fig. 7(c), the as-prepared Pd/C-Ap and Pd4Co/C-Ap have lower catalytic activity for ORR than the as-prepared Pt/C-Ap although all of them have a similar particle size of ∼4.5 nm (Table 1). However, although all the three samples exhibit a decrease in catalytic activity on annealing at 350 °C (inset in Fig. 7(a)) and 500 °C (inset in Fig. 7(d)) due to an increase in crystallite size (Table 1), both the Pd/C and Pt/C samples exhibit a much faster decrease in catalytic activity compared to the Pd4Co/C sample. For example, despite a similar crystallite size of ∼8 nm after annealing at 350 °C (Table 1), the Pd4Co/C-350 sample exhibits higher catalytic activity than the Pd/C-350 sample and about 60% of the activity of Pt/C-350, indicating that the alloying of Pd with Co enhances the catalytic activity for ORR. This demonstrates that combining a metal like Co that cleaves the O–O bond easily due to the high negative free energy change for oxide formation with a metal like Pd that reduces the adsorbed oxygen easily due to the high positive standard oxidation potential offers an effective strategy to design low cost electrocatalysts.20,34 In addition, formation a core-shell structure for Pd-based alloys on annealing has been reported by Lima et al.35 due to the segregation of the noble metal Pd to the surface, which could also contribute to the variation in catalytic activity with annealing temperature. Besides, shifts in the Pd:4d band center and thus modifications in the Pd–O bond energy to a balanced level on alloying with Co so that the Pd–O bond is not too strong or too weak, as indicated by experimental and density functional theory (DFT) calculations,36 may make the ORR more facile on alloying. Thus, the slow decrease in the catalytic activity of the Pd4Co/C sample with annealing temperature compared to the Pd/C and Pt/C samples is due to a partial offset of the decrease in catalytic activity caused by the increasing particle size by the increasing degree of alloying between Pd and Co. Finally, the drastic, much faster decrease in the catalytic activity of Pt/C with annealing temperature is due to a rapid increase in crystallite size with annealing temperature (Table 1).
![]() | ||
Fig. 8 Comparisons of the cyclic voltammogram of (a) Pd4Co/C-Ap, (b) Pd4Co/C-350, (c) Pd4Co/C-500, (d) Pt/C-Ap, (e) Pd/C-Ap, and (f) Pd/C-350 in 0.5 M H2SO4 and in 0.5 M H2SO4 + 1 M methanol solution at a sweep rate of 20 mV s−1 at room temperature. The insets compare the electrochemical surface area in the presence and absence of methanol. |
To understand the origin of the methanol tolerance of the Pd4Co catalyst, we also recorded the CV plots of Pd/C-Ap and Pd/C-350 in the presence and absence of methanol, and Fig. 8(e) and (f) display those CV plots at 15th cycle. The Pd catalyst shows similar behavior in the presence and absence of methanol, indicating that Pd possesses high tolerance to methanol. Thus, while Pt is highly active for the methanol oxidation reaction (MOR), the inactivity of Pd for MOR avoids the adsorption of intermediates that could form during MOR on the active sites of Pd–Co alloys, resulting in a superior tolerance of Pd4Co to methanol during ORR.
Fig. 9 compares the stability of the Pd4Co/C-350 sample in presence of methanol on subjecting to 15 cycles in the CV scan. The sample exhibits excellent stability in presence of methanol similar to that observed in the absence of methanol in Fig. 5.
![]() | ||
Fig. 9 Cyclic voltammograms of the Pd4Co/C-350 sample recorded in a 0.5 M H2SO4 + 1 M methanol solution at a sweep rate of 20 mV s−1 at room temperature, illustrating the stability on cycling. |
Fig. 10 compares the polarization curves recorded in single cell DMFC of Pd/C-350, Pt/C-350, Pd4Co/C-350, and a commercial 20 wt% Pt in carbon electrocatalysts for ORR. As seen, Pd4Co-350 shows higher open-circuit voltage (OCV) and better catalytic activity with lower polarization loss in the activation polarization region than both the Pd/C-350 and Pt/C-350 electrocatalysts prepared by the same modified polyol process. The performance of Pd4Co-350 is comparable to that of a state-of-the-art commercial Pt/C catalyst. We believe the better performance of Pd4Co-350 compared to the Pt/C sample prepared by us by the same modified polyol method is due to both a higher tolerance of the former to methanol poisoning in addition to smaller particle size (Table 1). The combination of high catalytic activity and excellent tolerance to methanol together with a lower cost makes Pd4Co/C an attractive cathode catalyst for DMFC.
![]() | ||
Fig. 10 Comparison of the polarization curves recorded in single cell DMFC of Pd/C-350, Pt/C-350, Pd4Co/C-350, and a commercial 20 wt% Pt in carbon electrocatalysts for ORR at 65 °C. The methanol concentration was 1 M. |
![]() | ||
Fig. 11 ORR Tafel plots of Pd4Co/C-Ap, Pd4Co/C-350, and Pd4Co/C-500 recorded in 0.5 M H2SO4 at a slow sweep rate of 5 mV s−1 at room temperature. |
Footnote |
† Electronic Supplementary Information (ESI) available: Experimental details describing synthesis and electrochemical characterization of the electrocatalysts. See DOI: 10.1039/b814708f |
This journal is © The Royal Society of Chemistry 2009 |