TMCnet News

Effect of Structural Characteristics of Binary Palladium-Cobalt Fuel Cell Catalysts on the Activity for Oxygen Reduction [ChemPlusChem]
[August 25, 2014]

Effect of Structural Characteristics of Binary Palladium-Cobalt Fuel Cell Catalysts on the Activity for Oxygen Reduction [ChemPlusChem]


(ChemPlusChem Via Acquire Media NewsEdge) In view of possible use as cathode materials in polymer electrolyte membrane fuel cells, the electrocatalytic activity of palladium-cobalt catalysts for oxygen reduction has been investigated in acid medium. In this minireview, the effect of structural characteristics, such as degree of alloying, particle size and palladium segregation on the alloy surface to form a core-shell system, on the electrocatalytic activity of palladium-cobalt catalysts for oxygen reduction is discussed.



Keywords : cobalt · electrochemistry · nanoparticles · oxygen reduction reaction · palladium 1. Introduction Carbon-supported platinum or platinum alloys are commonly used as anode and cathode electrocatalysts in polymer electro- lyte membrane fuel cells (PEMFCs). The cost of platinum and a limited world supply are significant barriers to the extensive use of these types of fuel cells. To reduce the platinum content in the cell, the partial or total substitution of platinum with palladium seems a promising way.[1] Indeed, platinum and pal- ladium have very similar properties (same group of the Period- ic Table, same crystal structure, similar atomic size), and the cost of palladium is lower than that of platinum. Moreover, pal- ladium is at least fifty times more abundant on earth than plat- inum. Palladium presents electrocatalytic activity for the oxygen reduction reaction (ORR), but it is considerably lower than that of platinum. Indeed, by current density and activa- tion enthalpy studies, it was found that palladium was signifi- cantly less active for the ORR than platinum.[2] As in the case of platinum,[3] the addition of cobalt increases the ORR activity of palladium. It is well known that the alloying of palladium cata- lysts with transition metals (M = Fe, Co, Ni, etc.) improves their ORR activity, owing to changes in the PdPd bond length, modification of the electron configuration, and/or alteration of surface species and compositions.[4, 5] Based on a thermodynam- ic model, Fernndez et al. demonstrated that the main func- tion of the incorporation of active elements, such as cobalt, could facilitate the dissociation of OO bonds and thereby the produced CoOads species might transfer to the palladium site in which the electrochemical reduction reaction could take place immediately.[4] Unlike previous reviews, which are more general and deal with the catalytic activity of palladium and various palladium- based catalysts for different reactions (hydrogen and alcohol oxidation, and oxygen reduction) in different media (acid and alkaline),[1, 6, 7] in this minireview, only a part of this wide topic has been analyzed in depth. The catalytic activity and relation- ship of the activity to the structural characteristics of PdCo catalysts for oxygen reduction in acid medium have been eval- uated. Whether cobalt has to be present in the metallic form alloyed to palladium or in the oxide form is discussed, along with whether the most suitable structure of the PdCo cata- lysts is in the alloyed or core-shell form.

2. Structure of PdCo Catalysts First-group transition metals and palladium form a continuous series of face-centered cubic (fcc) solid solutions, and certain ordered phases may be observed with compositions of Pd3M, PdM, or PdM3 (M = Fe, Co, Ni), depending on the preparation and heat treatment of alloys.[8] In the system Pd+ Co, only two types of ordered phases of intermetallic compounds can be formed.[9] One is the Ll2 type (Cu3Au type), existing over the range of composition of palladium between 60 and 90 at %. The other has a structure of Ll0 type (CuAuI type), existing within a narrow region of compositions around CoPd. This phase is accompanied by two-phase regions over fairly wide ranges at both sides.[9] X-ray photoelectron spectroscopy (XPS) was used to elucidate the variation of electronic properties of palladium and cobalt upon alloying.[10, 11] Shifts of the XPS Pd 3 d peak and the valence band towards higher binding ener- gies (BEs) and a narrowing of the Pd 3 d peaks in comparison to that of bulk Pd were observed in alloyed PdCo samples. The shift to higher BE in the d band of palladium is due to d- band hybridization upon alloying with cobalt. This shift indi- cates a decrease in the density of states (DOS) at the Fermi level. On the other hand, the Co 2 p peak shifted to higher BE than metallic cobalt and the shift was attributed to either al- loyed cobalt[12] or cobalt oxide forms.[13] One of the most im- portant questions to be addressed when characterizing sup- ported bimetallic catalysts is whether alloy particles are formed or not. The formation of well-mixed PdCo alloys was ob- served in silica-supported 10 wt % PdCo catalysts, in the range of 25-75 at % Co, prepared by incipient wetness impreg- nation of SiO2 by catalyst precursors, followed by reduction at 380 8C.[14,15] Formation of PdCo solid solutions during reduc- tion treatment at 500 8C of graphite-supported, cobalt-rich Pd Co catalysts (Pd/Co atomic ratio 33 :67 and 16 :84) was revealed by magnetic measurements and X-ray diffraction (XRD) analy- sis. Palladium and cobalt K-edge extended X-ray absorption fine structure (EXAFS) analysis confirmed that bimetallic parti- cles with a palladium- or cobalt-rich phase were formed.[16, 17] Small particles were palladium-enriched because the larger ones showed a higher cobalt content.[16] Generally, in carbon- supported PdCo catalysts, with a Pd/Co atomic ratio of 1, prepared by low-temperature methods, the lattice parameter of the fcc crystalline structure decreased with increasing Co content in the catalyst (Figure 1) ;[11, 16-20] this indicated the for- mation of PdCo substitution solid solutions, where some atoms of Pd are replaced by Co atoms.


The spread of data in Figure 1 is related to the different amounts of cobalt alloyed. The decrease of the lattice parame- ter by thermal treatment, caused by the incorporation of cobalt into the palladium fcc structure, reveals a transformation of the cobalt-rich hexagonal close packed (hcp) PdCo phase into the fcc phase, both present in as-prepared PdCo cata- lysts, according to the PdCo phase diagram of Ishida and Nishizawa.[23] The amount of cobalt in the fcc palladium lattice of PdCo catalysts prepared by a solution-based reduction procedure and annealed at 350 and 500 8C was lower than the nominal cobalt content and increased with increasing anneal- ing temperature and, for a given annealing temperature, line- arly increased with increasing nominal cobalt content.[24] Palladium segregation on the PdCo catalyst surface was re- ported.[14,25-27] Palladium surface enrichment in 10 wt % PdCo/ SiO2 catalysts (range 25-75 at % Co) pretreated in oxygen at 300 8C was observed by Juszczyk et al.[14,25] In the same way, Krawczyk et al.[26]found that Pd segregates to surfaces of the Co30Pd70 and Co50Pd50 alloys at temperatures above 300 8C. Kim et al. reported that, at a heat-treatment temperature of 300 8C in a reducing atmosphere of carbon-supported PdCo cata- lysts (range 30-70 at % Co), a phase-segregated PdCo core- shell structure (Pd shell on a Co core, Co@Pd) is formed, whereas, at higher temperatures (above 500 8C), a PdCo alloy phase is formed and the core-shell structure is destroyed.[27] Wei et al. reported that CeO2 could act as a surface modifier of PdCo/C catalysts : carbon-supported PdCo alloy core/Pd shell catalysts could be obtained by the addition of an appropriate amount of CeO2.[28] The addition of cobalt to palladium also affects the particle size of the catalyst. The average particle size of MgO-support- ed Pd and Pd40Co60 catalysts, prepared by controlled decompo- sition of metal acetylacetonate (acac) precursors on MgO, was approximately 6 nm for Pd and approximately 4 nm for Pd40Co60.[29] Likewise, as seen in Figure 2, the particle size of carbon-supported PdCo catalysts decreases with increasing cobalt content, regardless of the synthetic method. The parti- cle size was larger for samples prepared by an impregnation method.[22] By thermal treatment in the range of 300-700 8C, the particle size of a Pd67Co33/C catalyst linearly increased with increasing thermal treatment.[30] Liu and Manthiram found that the crystallite size of PdCo catalysts for a given annealing temperature decreased rapidly with increasing nominal Co content up to 30 at % Co and then remained almost constant for nominal Co contents of 30-50 at %.[24] Tominaka et al. ob- served that the morphology of electrodeposited PdCo de- pended on the applied current density. PdCo deposited at 10 mA cm2 was smooth and formed by smaller particles (ca. 10 nm), whereas that at 200 mA cm2 was rough and formed by larger particles (ca. 50 nm).[31] 3. Evaluation of the ORR Activity of Non-Plati- num Bimetallic Catalysts 3.1. Combinatorial Methods One approach for discovering effective electrocatalysts is to rapidly evaluate large libraries of potential candidates.[32, 33] Promising materials identified during this preliminary screening step can then be subjected to more extensive and quantitative testing. Several techniques have been reported for rapid elec- trocatalyst screening, such as scanning electrochemical micros- copy (SECM)[4, 34] and bipolar electrochemistry.[35] These meth- ods were used to evaluate the ORR activity of non-platinum MCo and PdM catalysts. For both series of bimetallic cata- lysts, PdCo showed the highest ORR activity, which was near to that of platinum.

3.1.1. Screening of MCo Catalysts Fernndez et al. proposed guidelines for the design of im- proved bimetallic electrocatalysts for the ORR in acid media.[4] This guide is based on simple thermodynamic principles by as- suming a simple mechanism in which one metal breaks the oxygen-oxygen bond of molecular O2 and the other metal acts to reduce the resulting adsorbed atomic oxygen. Analysis of the Gibbs free energies of these two reactions guides the selection of combinations of metals that can produce alloy sur- faces with enhanced activity for the ORR when compared with the constituent metals. Selected systems were tested by fabri- cating arrays of metallic catalysts consisting of various binary and ternary combinations of palladium, gold, silver, and cobalt deposited on glassy carbon (GC) substrates. SECM, in a new rapid-imaging mode, was used to rapidly screen arrays cover- ing a wide range of catalyst compositions for their ORR activity in acidic medium. The addition of cobalt to palladium, gold, and silver decreased the ORR overpotential, in agreement with the proposed model. Catalyst spots that exhibited enhanced electrocatalytic activity in the SECM screening technique were then examined through classical rotating disk electrode (RDE) experiments. The ORR activities obtained by both RDE and the SECM screening technique showed excellent agreement. Pd Co/GC electrodes (10-30 % Co) showed significant activity for the ORR, close to that of Pt/C.

3.1.2. Screening of PdM Catalysts Palladium-based bimetallic catalysts with eight different metals were computationally evaluated for the ORR and tested in acidic media by using combinatorial methods.[5] APdCo alloy showed the closest oxygen adsorption energy to platinum in simple theoretical model calculations, which suggested the highest ORR activity. This prediction was confirmed experimen- tally and indicated that the single parameter of oxygen adsorp- tion energy could be a useful guide to developing non-plati- num ORR catalysts.

A rapid screening of bimetallic electrocatalyst candidates for the ORR was performed by means of bipolar electrochemis- try.[36] The bipolar electrode (BPE) screening devices consist of indium tin oxide (ITO) electrodes with chromium microbands deposited at their anodic poles. Bimetallic catalysts are dis- pensed onto the cathodic poles. During a screening experi- ment, the BPE arrays are immersed in an acidic electrolyte so- lution and a potential bias is applied between two driving electrodes positioned at either side of the array. The most ef- fective catalyst candidates result in electrodissolution of the largest number of chromium microbands. By using this ap- proach, three potential bimetallic ORR electrocatalysts, PdAu, PdCo, and PdW, were evaluated. The histograms in Figure 3 show the average number of chromium microbands removed for each composition of the three bimetallic electrocatalyst candidates. The PdCo materials exhibit the most positive shifts of the ORR onset potential (Figure 3 b). Data indicate that materials containing 30-90 % palladium exhibit much more positive ORR onset potential values than that of pure palladium.

3.2. ORR Activity, Methanol Tolerance, and Stability of Large-Scale PdM Catalysts Generally, large-scale bimetallic PdM(M= first-row transition metals) catalysts showed a higher ORR electrocatalytic activity than palladium, but lower than that of platinum. Conflicting re- sults regarding the order of ORR activity of various PdM(M= Co, Cr, Fe, Ni) catalysts obtained by voltammetric measure- ments have been reported.[37-40] Savadogo et al. observed that the ORR activity of PdCo was higher than that of PdCr if the amount of non-precious metal was about 30 at %.[37] The oppo- site result was reported by Lee et al. for a M content of about 60 at %: the ORR activity decreased in the order of PdCr > PdCo> PdNi.[38] In the presence of methanol, instead, the order of ORR activity was PdNi> PdCo > PdCr, that is, the methanol tolerance of PdNi was higher than that of PdCo and PdCr. Pires and Villullas studied the ORR on carbon-sup- ported PdM(M= Co, Fe, Ni) catalysts prepared by a polyol method with various reflux times : all materials with nominal composition of Pd/M 70 :30 with a reflux time of 30 min were the most active, following the order of PdFe > PdCo >Pd Ni> Pd. Conversely, the stability was in the order of PdNi> PdFe > PdCo> Pd.[39] Finally, Alexeyeva et al. observed that the ORR specific activity (SA) of PdCo/C was higher/lower than that of PdFe/C, depending on the preparation method.[40] Generally, the ORR activity of PdM catalysts de- pends on PdPd bond length, which, in turn, depends on the amount and the type of M alloyed. The enhanced catalytic ac- tivity of palladium alloy catalysts in oxygen reduction and the "volcano" relationship between activity and degree of alloying were studied by experiments and DFT calculations.[41] Catalytic activity correlates with the adsorption energy of oxygen, which, in turn, depends on lattice strain owing to alloying. Ac- cording to Lee et al. ,[38] filling of the palladium d band by alloy- ing decreases the DOS at the Fermi level. The decrease in the DOS inhibits the formation of palladium oxide on the surface of the electrocatalyst and contributes to the improvement of the ORR activity of palladium by alloying. The differences in the order of ORR activity of PdM catalysts should depend es- sentially on the different alloying degree and also on the parti- cle size of various PdM catalysts.

4. ORR on PdCo Catalysts 4.1. Screening of PdCo Catalysts A SECM-based technique allowed a preliminary ORR activity test of PdCo catalysts over a range of possible compositions to be performed.[4,42] Figure 4 a identifies alloys of PdCo with 10-20 at % cobalt that could efficiently catalyze the ORR at overpotentials for which no activity was observed for pure pal- ladium. PdCo/C catalysts with these cobalt contents and a metal loading of 20 wt % were evaluated in RDE experiments (Figure 4 b) and showed performances near to that of plati- num. Successively, methanol tolerance during the ORR of Pd Co with various cobalt contents and pure platinum electrocata- lysts was evaluated by using the TG-SC mode of SECM.[43] From the results of cyclic voltammetry and SECM images, the Pd80Co20 electrocatalyst showed higher methanol tolerance than that of platinum during the ORR, especially in the pres- ence of a high methanol concentration. The degree of alloying of different PdCo samples, however, was not evaluated.

4.2. Alloyed/Partially Alloyed PdCo Catalysts The fcc palladium-rich PdCo solid solution is the active phase for the ORR.[4, 42] As reported in Section 2, the amount of cobalt alloyed in the fcc PdCo solid solution in as-prepared (not ther- mally treated) catalysts is lower than the nominal cobalt con- tent. The ORR on PdCo catalysts essentially depends on the amount of cobalt alloyed, that is, on geometric (PdPd distan- ces) and/or electronic (Pd d-band vacancies) effects by alloying and on the particle size. Both of these parameters depend, in turn, on the amount of cobalt in the catalyst (see Section 2). The amount of cobalt alloyed influences the SA, whereas the particle size affects the electrochemically active surface area (ECSA): these parameters govern the mass activity (MA), ac- cording to Equation (1): (ProQuest: ... denotes formula omitted.) To evaluate the effect of cobalt content on the ORR activity on polycrystalline, electrodeposited PdxCo electrocatalysts, peak currents for the cathodic voltammetric sweep were com- pared for compositions in the range of 17-50 at % Co.[20] As seen in Figure 5, in which the ORR peak current of PdCo cata- lysts is plotted as a function of nominal cobalt content, the highest ORR activity was observed for an atomic composition of 25 % Co. Considering that the particle size decreases monot- onously with increasing cobalt content in the catalyst (Figure 2), resulting in an increase of ECSA with increasing cobalt content, the presence of a maximum in Figure 5 has to be ascribed to alloying effects on SA. It seems that the PdPd distance presents an optimal value for oxygen adsorption and reduction : indeed, the compression of the Pd lattice, corre- sponding to a reduction in the PdPd bond length induced by alloying of smaller atoms, such as Co, into the Pd lattice, can weaken the interaction between the catalyst and adsorbate (such as O2), resulting in a change in the surface activity of Pd sites. It is hard, however, to evaluate the net effect of the Pd Pd distance on the ORR activity of PdCo at fixed particle sizes because both of these parameters depend on the Co content in the catalysts and thermal treatment. From data reported by Liu and Manthiram,[24,44] the dependence of the SA, expressed as charge transferred or cathodic peak area divided by the sweep rate, on PdPd distance at fixed particle sizes of ap- proximately 8, 9, and 10.5 nm is shown in Figure 6. As seen from data in Figure 6, at a fixed particle size, for a PdPd dis- tance of < 0.272 nm, SA increases with increasing PdPd dis- tance, whereas for a PdPd distance of > 0.273 nm SA decreas- es with increasing PdPd distance ; thus, extrapolating these results, the optimum PdPd distance should fall between 0.272 and 0.273 nm. Also, in the case of platinum-based alloy catalysts, the enhancement of SA for the ORR is related to changes in PtPt distance, and a suitable PtPt distance for adsorption and reduction of oxygen is between 0.271 and 0.275 nm,[45] which is very close to the optimum PdPd dis- tance.

Different works on PdCo[8, 41, 46] and PdFe[47] catalysts re- ported that the optimal PdPd interatomic distance for the ad- sorption and reduction of oxygen was 0.273 nm : considering that an interatomic distance of 0.273 nm corresponds to a lat- tice parameter of 0.3860 nm and, according to the hypothesis that the fcc PdCo alloy obeys Vegard's law for the entire range of compositions (the lattice parameter of fcc Pd = 0.3890 nm[48] and the lattice parameter of fcc Co = 0.3548 nm),[49] it results in an optimal Co content in the PdCo alloy of approximately 9 at %. In summary, the results suggest that alloying palladium with a small amount of cobalt positively increases the ORR ac- tivity. Liu and Manthiram showed that the increase in the ORR activity of PdCo catalysts in the 0-30 at % nominal Co content is due to both a decrease in crystallite size and an increase in the degree of alloying.[24] As observed for the platinum cata- lyst, the ORR on PdCo/C catalysts also pass through the four- electron pathway, leading to water production.[13, 18, 19, 46] By ana- lyzing both Tafel and Arrhenius plots for the ORR on Pd3Co, the O2 bond cleavage step is the rate-determining step.[20] A new parameter for the evaluation of the ORR activity of PdCo catalysts, that is, the degree of surface oxidation (DSO) was proposed to illustrate their relationship between ORR ac- tivity and surface oxidation extent.[50, 51] APdCo (3 :1) catalyst was heat treated in air at temperatures in the range of 50- 350 8C. XRD analysis indicated that, as the oxidation tempera- ture increased from 50 to 150 8C, no formation of oxide phases occurred. However, on raising the temperature to 200-250 8C, reflections corresponding to the PdCo alloy vanished gradually and the characteristic peaks of PdO were observed. Moreover, a weak reflection corresponding to a CoO phase was also found. For the catalyst treated at 350 8C, the severe heat treat- ment resulted in carbon burning, sintering, oxidation, and phase separation of the alloy catalysts.

Figure 7 a shows the ORR activity of PdCo catalysts ob- tained by linear sweep voltammetry (LSV) measurements in an oxygen-saturated solution of HClO4. The mixed kinetic/diffu- sion-controlled region, reflecting the catalytic activity of alloy catalysts, ranges between 0.6 and 0.8 V versus a normal hydro- gen electrode (NHE). The synergistic effect of cobalt alloying and oxidation treatment gives rise to a progressive enhance- ment of the ORR activity of PdCo/C catalysts. Among the vari- ous thermal treatment temperatures, the PdCo catalyst treat- ed at 250 8C, for which the complete formation of PdO and highest DSO value (100 %) were attained, presented the high- est ORR activity. As seen in Figure 7 b, the enhancement of the ORR activity by oxidation treatment at 250 8C was more impor- tant for Pd3Co/C than that for Pd/C. Rahul et al. ,[52] however, showed that the enhancement of the ORR activity of PdCo catalysts following oxidative heat treatment (OHT) was not due to PdO formation. Heat treatment in an oxygen-containing at- mosphere at 400 8C resulted in the formation of palladium and cobalt oxides. However, PdO formed during the heat treatment was converted into reduced palladium in the potential range of 0.05-0.4 V versus NHE. Their electrochemical properties and ORR activity were compared with that of the corresponding as-synthesized catalyst. The ECSA and current owing to the for- mation of adsorbed oxygenated species of the heat-treated catalyst were higher than that of the non-heat-treated counter- part. Higher MA and SA values of the OHT catalysts than those of the corresponding as-prepared catalysts were observed. The PdCo catalyst with 30 wt % Co showed the best activity. Im- provement in the SA of OHT PdCo catalysts relative to the corresponding as-prepared catalysts suggests that the activity improvement is not simply due to an increase in ECSA. The in- crease in the activity upon heat treatment may be attributed to the change in the degree of alloying, segregation of alloy components, changes in the particle composition, formation of a chemically ordered alloy phase, and increase in the size of palladium nanoparticles.

4.3. Core-Shell PdxCo@Pd (0 x< 1) Catalysts Recently, platinum monolayer (PtML) electrocatalysts were pro- posed as cathode materials to resolve problems related to the slow ORR kinetics and high platinum loading ; these determine the efficiency and cost of low-temperature fuel cells. These cat- alysts consist of a monolayer of platinum on carbon-supported metal or metal-alloy nanoparticles and have the highest utiliza- tion of platinum because almost every platinum atom is pres- ent on the surface and participates in the electrocatalytic reac- tions.[53] The interaction between the PtML and the substrate material induces a synergistic effect for ORR kinetics. Indeed, the electrocatalytic activity of PtML depends on its interaction with the support. DFT calculations showed that the binding en- ergies and reactivities of small atom or molecule adsorption on strained surfaces and metal overlayers correlated well with the position of the d-band center of surface atoms.[54] Analogously, Shao et al. investigated the ORR on palladium monolayers on various metal surfaces to obtain a substitute for platinum and to elucidate the origin of their activity.[55] As for PtML, their ac- tivity was correlated with their d-band centers. They found a volcano-type dependence of activity on the energy of the d- band center of palladium monolayers, with Pd/Pt(111) at the top of the curve. To further reduce the content of other noble metals, which serve as substrates for palladium, metal alloys, such as PdCo alloys, were tested as supports for palladium monolayers.[27,55-58] First, core-shell Pd2Co/C catalysts were ob- tained by surface segregation of Pd at elevated tempera- tures.[55] The activity of these core-shell Pd2Co/C nanoparticles was comparable to that of commercial Pt/C catalysts. The ORR kinetics predominantly involved a four-electron step reduction, for which the first electron transfer was the rate-determining step. The downshift of the d-band center of the palladium "skin", which constituted the alloy surface owing to the strong surface segregation of palladium at elevated temperatures, de- termined its high ORR activity. Then, Wang et al. prepared carbon-supported PdCo@Pd core-shell nanoparticles by a two- step route.[56,57] First, palladium and cobalt precursors, pre-ad- sorbed on carbon, were simultaneously reduced. The as-pre- pared PdCo/C catalysts were then annealed at 500 8C under a flow of H2 to yield particles with a core-shell structure.

The core-shell structure and chemical distribution of PdCo@Pd/C nanoparticles were examined by using a scanning transmission electron microscope equipped with an electron energy loss spectrometer (EELS). In Figure 8 a and b, a pair of annular dark- (ADF) and bright-field (BF) scanning transmission electron microscopy (STEM) images show the internal crystal structure of the nanoparticle. Figure 8 c and d shows the cobalt and palladium projected distribution. The palladium (red) versus cobalt (green) composite image (Figure 8 e) dem- onstrates that a palladium-rich shell is formed on the surfaces. By comparing line profiles across the palladium-rich shell from the cobalt and palladium maps, as shown in Figure 8 f, the thickness of the shell is estimated to be approximately 1 nm. The dependence of the ORR MA of PdxCo@Pd catalysts with different cobalt contents on the lattice parameter of the palla- dium alloy exhibited a "volcano curve". The PdxCo@Pd (3 :1) catalyst showed the best activity, with an activity comparable to that of Pt/C. Kim et al. submitted PdxCo1x (x = 0.3, 0.5, 0.7, and 1) nanoparticles to heat treatments at 300, 500, and 700 8C under a flow of H2.[27] At a heat-treatment temperature of 300 8C, a phase-segregated PdCo core-shell structure is formed, whereas at higher temperatures a PdCo alloy phase is formed and the core-shell structure is destroyed.

The ORR activity of core-shell PdxCo1x catalysts was higher than that of the alloyed catalysts. All PdCo/C electrocatalysts with core-shell structures showed lower MAs than a commer- cial Pt/C, but the best core-shell Pd0.5Co0.5/C catalyst had a MA value of only approximately 17 % lower than that of that of the commercial samples. On the other hand, all PdCo/C elec- trocatalysts with core-shell structures showed higher SAs than that of commercial Pt/C. PdCo/C catalysts with two different surface compositions and various Pd/Co ratios were synthe- sized by an ultrasound-assisted polyol process.[58] Pd4Co/ C(core-shell) was synthesized by reacting Co(acac)2 first to form cobalt seeds with subsequent addition of Pd(acac)2 under ultrasound irradiation. Pd4Co/C(alloy) was synthesized through both Co(acac)2 and Pd(acac)2 simultaneously reacting under ul- trasound irradiation. ECSAs of the electrodes obtained from cyclic voltammetry (CV) curves (Figure 9 a) are reported in the histogram in Figure 9 c. The ECSA of the Pd4Co/C catalysts were between those of Pt/C and Pd/C. By LSV measurements (Figure 9 b), the onset potentials for the ORR were in the order of Pt/C > Pd4Co/C(core-shell) > Pd4Co/C(alloy)>Pd/C. The onset potential of Pd4Co/C(core-shell) was considerably higher than those of other palladium-containing materials and was close to that of platinum. SA values at 0.75 V by ECSA-normal- ized LSV data (Figure 9 d) were in the order of Pd/C< Pd4Co/ C(alloy)! Pt/C<Pd4Co/C(core-shell).

In some theoretical works, based on DFT, the ORR mecha- nism on PdxCo@Pd catalysts was investigated.[59-61] Tang and Henkelman calculated the energetics of oxygen dissociative adsorption on 1 nm palladium-shell nanoparticles with a series of core metals. [59] The barrier for this reaction and the binding energy of atomic oxygen correlated well with the d-band level of the surface electrons. A significant trend that holds for cata- lysts of similar structure is the Brønsted-Evans-Polanyi (BEP) relationship. This relationship linearly correlates the energy barrier of an elementary reaction to the reaction energy. A maximum in catalytic activity results from an optimal compro- mise between a low reaction barrier and a weak product bind- ing energy. Figure 10 shows the linear correlation between the saddle-point energy for O2 dissociative adsorption and the final state energy with the molecule dissociated on the sur- face.

The palladium-shelled nanoparticles follow the same BEP re- lationship as single-crystal (111) transition-metal surfaces. An ideal catalyst would have both a low barrier and weak product binding (lower right in Figure 10), but the BEP relationship does not allow this. Instead, there is an optimal trade-off be- tween the dissociation barrier and product binding near Pt(111). Gold binds oxygen weakly, but the dissociation barrier is prohibitively high. Copper and palladium have low dissocia- tion barriers, but bind oxygen too strongly for efficient reduc- tion to water. The same is true for cobalt and molybdenum ; those points are off the lower left corner of the plot. Figure 10 also shows that the core metal inside a palladium shell can be used to tune its interaction with oxygen. The monometallic palladium particle is a little more noble than the Pd(111) sur- face because it is shifted toward platinum and gold along the BEP relationship. With gold and silver noble metal cores, the palladium-shell particle becomes less noble and binds oxygen more strongly than the monometallic palladium particle. It is the reactive metal cores, copper and particularly molybdenum and cobalt, that make the palladium shell more noble and closer to Pt(111) on the BEP relationships. These core-shell structures are expected to possess high ORR activity. Zuluaga and Stolbov performed computational studies on the electron- ic structure of PdCo alloys, as well as the energetics of ad- sorption, and vibrational frequencies of the ORR intermediates on the alloy surfaces.[60] They found that the surface segrega- tion of the alloys was essential to improve the electrocatalytic properties of these materials. Indeed, the binding energy of the ORR intermediates on Pd0.75Co0.25@Pd was lower than that on Pd(111), which was favorable for the reaction, whereas the Pd0.75Co0.25 and Pd0.5Co0.5 surfaces were too reactive for the ORR. The results showed that the hybridization between d Pd and d Co states caused a low-energy shift of the d band of sur- face Pd in Pd0.75Co0.25@Pd, which caused weakening of the bonding of the intermediates. Manogaran and Hwang took the Pd3Co@Pd catalyst as a model case, in which one or two Pd overlayers were located on top of the bimetallic substrate.[61] The calculations demonstrated that the subsurface cobalt atoms assisted in facilitating the ORR by lowering the activa- tion barriers for O/OH hydrogenation with a slight increase in the O2 scission barrier. Analysis of intra- and interlayer orbital interactions in the near-surface region elucidates the synerget- ic interplay between the surface electronic structure modifica- tion, owing to the underlying cobalt atoms (interlayer ligand effect) and compressive strain caused by the Pd3Co substrate.

4.4. Stability of PdCo Catalysts Among different degradation phenomena, the durability of the catalyst is one of the major issues in PEMFC technology. Liu and Manthiram performed a repetitive potential cycling (RPC) on partially alloyed PdCo catalysts and the variation in the surface oxide reduction peak (cathodic peak) area with the number of cycles was used to estimate the durability of the catalysts.[24,44] The ratio of the cathodic peak area in the 15th cycle to that in the 7th cycle (D) was used to compare the durability of PdCo catalysts. As seen in Figure 11, D monoto- nously decreased with increasing nominal cobalt content from 0 to 50 at % Co. Thus, it seems that PdCo catalysts are less stable than palladium. However, the lower stability of PdCo catalysts does not depend directly on the cobalt content in the catalyst, but on a decrease of the PdCo particle size with increasing cobalt content in the catalysts. Upon RPC, small par- ticles are less stable than large particles : both cobalt and palla- dium dissolve out simultaneously from small alloy particles and palladium re-deposits on the surfaces of large alloy parti- cles (Ostwald ripening), as observed also for PtCo catalysts.[62] As the particle size decreases with increasing cobalt content in the catalyst, the number of small particles dissolved increases with increasing amounts of cobalt ; this gives rise to a signifi- cant decrease in the surface area, which explains the decrease of the ORR activity and durability with increasing cobalt con- tent in PdCo catalysts. If D is normalized by taking into ac- count the particle size (Dn = D dPd/dPdCo, in which dPd (4.0 nm) and dPdCo are the crystallite sizes of the Pd and PdCo catalysts, respectively), the obtained durability (ca. 75 %) is almost inde- pendent of cobalt present in the catalyst, as shown in Figure 11. In agreement with this result, at a fixed cobalt con- tent in the catalyst (30 at %) and particle size (ca. 8 nm), the durability of PdCo catalysts (ca. 90 %) was independent of the amount of cobalt alloyed.[39] Conversely, by RPC on Pd and PdCo of comparable sizes, it was observed that PdCo was more stable than Pd ;[39, 63] however, the durability of PdCo was lower than that of PdFe and PdNi.[34] Mustain et al. eval- uated the stability of PdCo (3 :1) catalysts in a single PEMFC by measuring the open-circuit cell voltage (OCV) and current density at 0.8 V versus NHE for 25 h.[64] The cell performance appears to be very stable within the test period. The OCV de- creased by only 3 %, whereas the performance at 0.8 V versus NHE dropped by approximately 10 %; this indicated the excel- lent stability characteristics of the PdCo catalyst. However, the slight decrease in performance could indicate that some leaching occurred. Wang et al. investigated the durability of a core-shell PdCo@Pd/C catalyst by RPC between 0.05 and 1.1 V versus RHE.[51, 56] The large decrease in the PdCo@Pd/C ECSA after 50 cycles indicated the poor stability of the core- shell structure.

Kim et al. compared the durability of Pd70Co30 catalysts with core-shell and alloy structures by RPC between 0.6 and 1.1 V versus RHE for 3000 cycles.[27] As seen in Figure 12, after dura- bility tests, for the Pd70Co30 alloy structure (obtained by ther- mal treatment at 700 8C), the ORR onset potential was only slightly shifted to a lower potential, whereas for the Pd70Co30 core-shell structure (obtained by thermal treatment at 300 8C) the ORR onset potential considerably shifted at lower poten- tial. STEM mapping was used to identify structural changes after the durability tests. For the Pd70Co30 core-shell catalyst, the diameters of Pd and Co mapping became almost the same after the test, which implied that Pd in the surface-rich phase was seriously dissolved and the core-shell structure collapsed. On the other hand, the alloyed catalysts showed little structur- al changes after the durability tests. Thus, it seems that palladi- um dissolution, owing to the small particle size, is the main factor that governs the durability of the PdCo core-shell structure.

The stability of PdCo catalysts can be enhanced by adding a third metal, such as molybdenum or tungsten : the incorpora- tion of molybdenum in PdCo to form a ternary system im- proved the stability of the catalysts without a decrease of the ORR. Raghuveer et al. found that the PdCoMo catalyst with a Pd/Co/Mo atomic ratio of 70 :20 :10 possessed a higher activi- ty than that of Pt/C and, unlike the PdCoAu catalyst, excel- lent chemical stability.[65] Stability tests under direct ethanol fuel cell (DEFC) operating conditions for 50 h indicated higher stability of PdCoMo/C (70 :20 :10) than that of Pt/C.[66] During durability tests, the cell with heat-treated PdCoMo/C showed a stable voltage with low polarization losses, whereas the commercial Pt/C catalyst showed considerably high polari- zation losses. Brace et al. investigated palladium-based ternary alloys for the ORR with, at the same time, improved stability relative to the parent binary alloys.[67] Among different ternary catalysts, the PdCoW alloys were the most suitable as ORR cat- alysts and were more stable than the PdCo alloys. The most stable alloys were composed of more than 60 at % Pd, less than 40 at % Co, and less than 20 at % W.

5. Summary and Outlook The addition of cobalt affects the activity for oxygen reduction of palladium. The SA for the ORR of PdCo depends on the amount of Co alloyed and goes through a maximum. The amount of Co alloyed at the maximum ORR activity is approxi- mately 9 at % and corresponds to the optimum PdPd dis- tance. Considering that only part of cobalt present in the cata- lyst is alloyed with palladium, the amount of cobalt in the cata- lyst at the maximum ORR activity, generally around 25 at %, but depending on the synthesis and thermal treatment, is higher than the amount of cobalt alloyed. The presence of cobalt also affects the PdCo particle size, which decreases with increasing cobalt content in the catalyst, increasing in this way the active surface area, so the cobalt content in the cata- lyst at the maximum ORR MA can be slightly higher than the cobalt content at the maximum ORR SA.

The ORR activity of core-shell PdxCo@Pd (0 x< 1) catalysts was evaluated by experimental and theoretical works. The ORR activity of PdxCo@Pd catalysts was higher than that of the al- loyed PdCo catalysts and comparable to that of Pt/C. The ORR activity was correlated with the energy of the d-band center of the palladium monolayer, which, in turn, depends on the characteristics of the PdxCo subsurface layer.

The durability of PdCo alloy catalysts at fixed particle sizes is higher or, at least, the same as that of palladium, however, if we take into account of the effect of cobalt on the particle size, increasing the amount of cobalt decreases the stability of the catalyst. Considering that the ORR activity increases with decreasing particle size and, conversely, the durability decreas- es with increasing particle size, a good compromise between activity and durability of PdCo alloy catalysts can be obtained by modulating the amount of cobalt in the catalyst. A way to improve the durability of alloyed PdCo catalysts without a de- crease in the ORR activity is the addition of a third metal to the PdCo structure through the formation of a stable PdCoM alloy.

Poor stability of the PdCo core-shell structure, which was lower than that of alloyed PdCo, was observed, owing to the dissolution of the Pd layer on the catalyst surface. Future work should be addressed to improve the stability of the core-shell structure, without affecting the catalytic activity.

[1] E. Antolini, Energy Environ. Sci. 2009, 2, 915 - 931.

[2] J. J. Salvador-Pascual, S. Citaln-Cigarroa, O. Solorza-Feria, J. Power Sour- ces 2007, 172, 229 - 234.

[3] E. Antolini, I. R. C. Salgado, M. J. Giz, E. R. Gonzalez, Int. J. Hydrogen Energy 2005, 30, 1213 - 1220.

[4] J. L. Fernndez, D. A. Walsh, A. J. Bard, J. Am. Chem. Soc. 2005, 127, 357-365.

[5] K. R. Lee, Y. Jung, S. I. Woo, ACS Comb. Sci. 2012, 14, 10 - 16.

[6] C. Bianchini, P. K. Shen, Chem. Rev. 2009, 109, 4183 - 4206.

[7] M. Shao, J. Power Sources 2011, 196, 2433 - 2444.

[8] T. Mallt, J. Petr, S. Szab, L. Marczis, J. Electroanal. Chem. 1986 , 208, 169-173.

[9] Y. Matsuo, J. Phys. Soc. Jpn. 1972, 32, 972 - 978.

[10] I. Matolinova, S. Fabik, K. MaSek, L. Sedlcek, T. Skla, K. Veltrusk, V. Ma- tolin, Vacuum 2003, 71, 41 - 45.

[11] M. Krawczyk, J. W. Sobczak, Appl. Surf. Sci. 2004, 235, 49 - 52.

[12] L. Arroyo-Ramirez, R. Montano-Serrano, T. Luna-Pineda, F. R. Romn, R. G. Raptis, C. R. Cabrera, ACS Appl. Mater. Interfaces 2013, 5, 11603 - 11612.

[13] K. Oishi, O. Savadogo, Electrochim. Acta 2013, 98, 225 - 238.

[14] W. Juszczyk, Z. Karpinski, D. Lomot, J. Pielaszek, Z. Paal, A. Y. Stakheev, J. Catal. 1993, 142, 617 - 629.

[15] W. Juszczyk, Z. Karpinski, J. Pielaszek, Z. Paal, J. Catal. 1993, 143, 583 - 593.

[16] F. B. Noronha, M. Schmal, C. Nicot, B. Moraweck, R. Frty, J. Catal. 1997, 168, 42 - 50.

[17] F. B. Noronha, M. Schmal, R. Frty, G. Bergeret, B. Moraweck, J. Catal. 1999, 186, 20 -30.

[18] W. Wang, D. Zheng, C. Du, Z. Zou, X. Zhang, B. Xia, H. Yang, D. L. Akins, J. Power Sources 2007 , 167, 243 - 249.

[19] X. Li, Q. Huang, Z. Zou, B. Xia, H. Yang, Electrochim. Acta 2008, 53, 6662-6667.

[20] W. E. Mustain, J. Prakash, J. Power Sources 2007 , 170, 28 - 37.

[21] R. N. Singh, C. S. Sharma, ETASR - Eng. Tech. Appl. Sci. Res. 2012, 2, 295 - 301.

[22] H. Gharibi, F. Golmohammadi, M. Kheirmand, Electrochim. Acta 2013, 89, 212-221.

[23] K. Ishida, T. Nishizawa, J. Phase Equilib. 1991, 12, 83 - 87.

[24] H. Liu, A. Manthiram, Appl. Catal. B 2009, 90, 184.

[25] W. Juszczyk, Z. Karpinski, Z. Pal, J. Pielaszek, Stud. Surf. Sci. Catal. 1993, 75, 1843 - 1846.

[26] M. Krawczyk, L. Zommer, B. Lesiak, A. Jablonski, Surf. Interface Anal. 1997, 25, 356 - 365.

[27] D. S. Kim, J. H. Kim, I. K. Jeong, J. K. Choi, Y. T. Kim, J. Catal. 2012, 290, 65-78.

[28] Y. C. Wei, C. W. Liu, K. W. Wang, ChemPhysChem 2010, 11, 3078 - 3085.

[29] H. Borchert, B. Jrgens, T. Nowitzki, P. Behrend, Y. Borchert, V. Zielasek, S. Giorgio, C. R. Henry, M. Bumer, J. Catal. 2008, 256, 24 - 36.

[30] L. Zhang, K. Lee, J. Zhang, Electrochim. Acta 2007, 52, 3088 - 3094.

[31] S. Tominaka, T. Momma, T. Osaka, Electrochim. Acta 2008, 53, 4679 - 4686.

[32] F. Mavr, R. K. Anand, D. R. Laws, K. F. Chow, B. Y. Chang, J. A. Crooks, R. M. Crooks, Anal. Chem. 2010, 82, 8766 - 8774.

[33] E. S. Smotkin, R. R. Daz-Morales, Annu. Rev. Mater. Res. 2003, 33, 557 - 581.

[34] A. J. Bard, J. Am. Chem. Soc. 2010, 132, 7559 - 7567.

[35] S. E. Fosdick, R. M. Crooks, J. Am. Chem. Soc. 2012, 134, 863- 866.

[36] S. E. Fosdick, S. P. Berglund, C. Buddie Mullins, R. M. Crooks, Anal. Chem. 2013, 85, 2493 - 2499.

[37] O. Savadogo, K. Lee, K. Oishi, S. Mitsushima, N. Kamiya, K. I. Ota, Electro- chem. Commun. 2004, 6, 105 - 109.

[38] K. Lee, O. Savadogo, A. Ishihara, S. Mitsushima, N. Kamiya, K. I. Ota, J. Electrochem. Soc. 2006, 153 , A20 - A24.

[39] F. I. Pires, H. M. Villullas, Int. J. Hydrogen Energy 2012, 37, 17052 - 17059.

[40] N. Alexeyeva, A. Sarapuu, K. Tammeveski, F. J. Vidal-Iglesias, J. Solla- Gulln, J. M. Feliu, Electrochim. Acta 2011, 56, 6702 - 6708.

[41] Y. Suo, L. Zhuang, J. Lu, Angew. Chem. Int. Ed. 2007, 46, 2862 - 2864 ; Angew. Chem. 2007, 119, 2920 -2922.

[42] J. L. Fernndez, J. M. White, Y. Sun, W. Tang, G. Henkelman, A. J. Bard, Langmuir 2006, 22, 10426- 10431.

[43] C. L. Lin, C. M. Snchez-Snchez, A. J. Bard, Electrochem. Solid-State Lett. 2008, 11, B136 - B139.

[44] H. Liu, A. Manthiram, Electrochem. Commun. 2008, 10, 740 - 744.

[45] S. Mukerjee, S. Srinivasan, M. P. Soriaga, J. McBreen, J. Electrochem. Soc. 1995, 142, 1409 - 1422.

[46] L. Zhang, K. Lee, J. Zhang, Electrochim. Acta 2007, 52, 7964 - 7971.

[47] M. H. Shao, K. Sasaki, R. R. Adzic, J. Am. Chem. Soc. 2006, 128, 3526 - 3527.

[48] T. B. Flanagan, W. A. Oates, Annu. Rev. Mater. Sci. 1991, 21, 269 - 304.

[49] E. A. Owen, D. M. Jones, Proc. Phys. Soc. London Sect. B 1954, 67, 456 - 466.

[50] Y.-C. Wei, C.-W. Liu, Y.-W. Chang, C.-M. Lai, P.-Y. Lim, L.-D. Tsai, K.-W. Wang, Int. J. Hydrogen Energy 2010 , 35, 1864 - 1871.

[51] Y.-C. Wei, C.-W. Liu, H.-W. Lee, S.-R. Chung, S.-L. Lee, T.-S. Chan, J.-F. Lee, K.-W. Wang, Int. J. Hydrogen Energy 2011, 36, 3789 - 3802.

[52] R. Rahul, R. K. Singh, M. Neergat, J. Electroanal. Chem. 2014 , 712, 223 - 229.

[53] R. R. Adzic, J. Zhang, K. Sasaki, M. B. Vukmirovic, M. Shao, J. X. Wang, A. U. Nilekar, M. Mavrikakis, J. A. Valerio, F. Uribe, Top. Catal. 2007 , 46, 249-262.

[54] J. Zhang, M. B. Vukmirovic, Y. Xu, M. Mavrikakis, R. R. Adzic, Angew. Chem. Int. Ed. 2005, 44, 2132 - 2135; Angew. Chem. 2005, 117, 2170 - 2173.

[55] M. H. Shao, T. Huang, P. Liu, J. Zhang, K. Sasaki, M. B. Vukmirovic, R. R. Adzic, Langmuir 2006, 22, 10409- 10415.

[56] D. Wang, H. L. Xin, Y. Yu, H. Wang, E. Rus, D. A. Muller, H. D. AbruÇa, J. Am. Chem. Soc. 2010, 132, 17664 - 17666.

[57] D. Wang, H. L. Xin, H. Wang, Y. Yu, E. Rus, D. A. Muller, F. J. DiSalvo, H. D. AbruÇa, Chem. Mater. 2012, 24, 2274 - 2281.

[58] J.-H. Jang, C. Pak, Y.-U. Kwon, J. Power Sources 2012, 201, 179 - 183.

[59] W. Tang, G. Henkelman, J. Chem. Phys. 2009, 130, 194504.

[60] S. Zuluaga, S. Stolbov, J. Chem. Phys. 2011, 135 , 134702.

[61] D. Manogaran, G. S. Hwang, Phys. Chem. Chem. Phys. 2013, 15, 12118 - 12123.

[62] M. Watanabe, K. Tsurumi, T. Mizukami, T. Nakamura, P. Stonehart, J. Elec- trochem. Soc. 1994, 141, 2659- 2668.

[63] M. R. Tarasevich, D. V. Novikov, G. V. Zhutaeva, V. A. Bogdanovskaya, L. A. Reznikova, N. A. Kapustina, V. V. Batrakov, Prot. Met. Phys. Chem. Surf. 2009, 45, 782 - 786.

[64] W. E. Mustain, K. Kepler, J. Prakash, Electrochim. Acta 2007, 52, 2102 - 2108.

[65] V. Raghuveer, A. Manthiram, A. J. Bard, J. Phys. Chem. B 2005, 109, 22909-22912.

[66] Ch. V. Rao, B. Viswanathan, Electrochim. Acta 2010, 55, 3002 - 3007.

[67] K. Brace, B. Hayden, C. Lee, T. Le Gall, European patent EP 2 381 518A1, 2011.

Received : February 25, 2014 Published online on May 20, 2014 Ermete Antolini*[a] [a] Prof. E. Antolini Scuola di Scienza dei Materiali Via 25 aprile 22, 16016 Cogoleto, Genoa (Italy) E-mail : [email protected] Ermete Antolini is Research Professor at Scuola Scienza Materiali, Genoa, Italy. After receiving his PhD in chemis- try from the University of Genova, Italy, he worked at Ansaldo Research, Genoa, and ENEA, Rome. He worked as a Visiting Professor at the Universidade de Sao Paulo, Sao Carlos, Brazil. His re- search interests focus on the develop- ment of materials for heterogeneous catalysis with an emphasis on support- ed catalysts for low-temperature fuel cells.

(c) 2014 Blackwell Publishing Ltd.

[ Back To TMCnet.com's Homepage ]