lable at ScienceDirect

Journal of Power Sources 287 (2015) 139e149

Contents lists avai

Journal of Power Sources

journal homepage: www.elsevier .com/locate/ jpowsour

Synergetic effect of palladiumeruthenium nanostructures for ethanolelectrooxidation in alkaline media

Evans A. Monyoncho a, b, Spyridon Ntais b, Felipo Soares b, Tom K. Woo a,Elena A. Baranova b, *

a Department of Chemistry, Centre for Catalysis Research and Innovation (CCRI), University of Ottawa, 10 Marie Curie, Ottawa, ON K1N 6N5, Canadab Department of Chemical & Biological Engineering, Centre for Catalysis Research and Innovation (CCRI), University of Ottawa, 161 Louis-Pasteur, Ottawa,ON K1N 6N5, Canada

h i g h l i g h t s

* Corresponding author.E-mail address: elena.baranova@uottawa.ca (E.A. B

http://dx.doi.org/10.1016/j.jpowsour.2015.03.1860378-7753/© 2015 Elsevier B.V. All rights reserved.

g r a p h i c a l a b s t r a c t

� Bi-phase PdxRu1�x/C nanoparticleswere prepared by a polyol method.

� Pd99Ru1/C nanoparticles have sixtimes higher mass activity comparedto Pd/C.

� Pd50Ru50/C lowers onset EOR poten-tial by 290 mV compared to Pd/C.

� Improved EOR on PdxRu1�x/C iscorrelated to the synergetic effect ofthe surface oxides.

a r t i c l e i n f o

Article history:Received 10 January 2015Received in revised form12 March 2015Accepted 27 March 2015Available online 1 April 2015

Keywords:Ethanol electrooxidationPalladiumRutheniumNanoparticlesAlkaline solutionDirect alcohol fuel cells

a b s t r a c t

Palladiumeruthenium nanoparticles supported on carbon PdxRu1�x/C (x ¼ 1, 0.99, 0.95, 0.90, 0.80, 0.50)were prepared using a polyol method for ethanol electrooxidation in alkaline media. The resultingbimetallic catalysts were found to be primarily a mix of Pd metal, Ru oxides and Pd oxides. Their elec-trocatalytic activity towards ethanol oxidation reaction (EOR) in 1M KOH was studied using cyclic vol-tammetry and chronoamperometry techniques. Addition of 1e10 at.% Ru to Pd not only lowers the onsetoxidation potential for EOR but also produces higher current densities at lower potentials compared toPd by itself. Thus, Pd90Ru10/C and Pd99Ru1/C provide the current densities of up to six times those of Pd/Cat �0.96 V and �0.67 V vs MSE, respectively. The current density at different potentials was found to bedependent on the surface composition of PdxRu1�x/C nanostructures. Pd90Ru10/C catalyst with moresurface oxides was found to be active at lower potential compared to Pd99Ru1/C with less surface oxides,which is active at higher potentials. The steady-state current densities of the two best catalysts, Pd90Ru10/C and Pd99Ru1/C, showed minimal surface deactivation from EOR intermediates/products duringchronoamperometry.

© 2015 Elsevier B.V. All rights reserved.

aranova).

1. Introduction

Direct alcohol fuel cells (DAFCs) that operate in alkalinemedia arean emerging technology starting to receive research attention asalternative electrical power sources due to the tremendous prog-ress in the development of anion exchange solid electrolytes [1e6].The advantages of using alcohols as fuel are mainly due to their

E.A. Monyoncho et al. / Journal of Power Sources 287 (2015) 139e149140

high volumetric energy density. Additionally, their storage andtransportation are cheaper than hydrogen fuel and alcohols can beeasily obtained from renewable biomass [6,7] By operating inalkaline media instead of acidic media DAFCs have low corrosive-ness and permits the use of non-noble metal electro-catalysts suchas nickel, cobalt, copper, molybdenum and iron [7,8] thusdecreasing significantly the cost. However, there are some chal-lenges hindering the full commercialization of DAFCs which needto be addressed and are the subject of recent reviews [1,6,9]. Themain challenges for DAFCs are: i) the much slower oxidation ki-netics of alcohols compared to hydrogen fueled polymer electrolytemembrane fuel cells (PEMFCs) and ii) the incomplete oxidation ofalcohols with two or more carbon atoms [1] Therefore, researchefforts are geared towards designing and developing efficientelectrocatalyst to increase electrical performance of DAFCs. Amongthe alcohols studied most are methanol, ethanol, ethylene glycol,propanol and glycerol of which ethanol is the most promising andnon-toxic candidate [7]. The state-of-the-art catalysts are able topartially oxidize ethanol to acetaldehyde and/or acetate ions, givinga maximum of four electrons. This partial oxidation of ethanoltranslates to the energy density output which is three times lowercompared towhat would be obtained if CO2 was the final product inwhich twelve electrons are exchanged. Therefore, design andimplementation of highly active and stable electrocatalyst forethanol electrooxidation is required.

Palladium has been shown as a promising electrocatalyst forethanol oxidation in alkaline media [9]. Efforts in combining Pdwith a secondmetal that can be either noble, e.g., Ru, Au, Rh or non-noble metals such as Fe, Co, Ni, Cu and Mo have been shown toimprove not only the catalytic activity but also reduce the cost ofthe membrane electrode assemblies (MEAs) [8,10e13]. Ruthenium isknown as a good promoter for alcohol oxidation and was recentlyreported to enhance ethanol oxidation on Pt in alkaline mediayielding CO2 as the major final product [14]. Recently, the interestfor bimetallic PdeRu catalyst systems for ethanol oxidation reac-tion (EOR) in alkaline media has attracted the attention of variousresearch groups [7,15e19].

Chen et al. and Sun et al. have compared the performance ofPdeRu and PteRu and showed that PdeRu in alkaline media isalmost four times better for EOR than Pt-Ru [7,20]. Yi et al. studiedporous PdeRu nanoparticles supported on titanium prepared byhydrothermal method [15]. They reported that Pd87Ru13 showedthe best catalytic activity towards EOR in alkaline media in meansof current density and onset potentials. They suggested that the“bifunctional mechanism” and the large surface area of Pd87Ru13are playing a critical role in the overall catalytic performance.Anindita et al. investigated nanostructured PdeRu nanoparticlessupported on carbon synthesized using sodium borohydridereduction method [16]. They prepared Pd-0.5wt% C and Pd-0.5wt%C-x wt. % Ru where x was 1, 5, 10, 20, and 50. They found that whenthe ruthenium content is 20 wt% electrocatalytic activity for EOR inalkaline media increases considerably. Bagchi et al. studied theelectrocatalytic activity of an electrodeposited PdeRu catalystsupported on nickel for EOR in alkaline media [17,19]. Theyobserved that the amount of loading and the composition of thecatalyst have a superimposed effect on ethanol electrocatalyticactivity. The thinner the electrodeposit film, the greater was thepeak current per unit mass of deposit due to greater roughnessfactor arising from small size of the crystallites. Correia et al.studied EOR using PdeRu bimetallic complexes [18]. Although, theyfound increased current density at high potentials, no improve-ment was observed on onset potential for the reaction. In the recentwork by Liang Ma et al. [21], authors reported EOR on PdeRusupported on carbon prepared by impregnation method with Rucontent ranging between 20 and 50%. Liang Ma et al. showed that

PdeRu system not only performswell in half-cell tests but also gave1.8 times higher power density compared to Pd/C in prototype fuelcell assemblies. They showed that the best performing catalyst hascomposition of 25% Ru and that the PdRu/C catalysts are promisingmaterials for ethanol oxidation in alkaline environment.

Despite all these studies on PdeRu catalyst system for EOR inalkaline media [7,11e15] and the observed promotional effect of Ruon Pd electrocatalytic activity, several key points still remain to beunderstood and clarified. Among them, the role and effect of Ruboth bulk and surface content. Similarly, the effects of surfacestructuremodifications on the catalytic activity of bimetallic PdeRunanoparticles for EOR in alkaline media are not yet fullyunderstood.

In the current work, we show the correlation of the surface-structure-composition-activity of PdxRu1�x (where x ¼ 1, 0.99,0.95, 0.90, 0.80 and 0.50) for ethanol electrooxidation in alkalinemedia. The nanoparticles were prepared using a polyol method andthen supported on carbon. The NPs were characterized by scanningtransmission electron microspectroscopy (STEM), X-ray diffraction(XRD) techniques and X-ray photoelectron spectroscopy (XPS). Theelectrocatalytic activity was investigated using cyclic voltammetry(CV) and chronoamperometry (CA) techniques.

2. Experimental

2.1. Materials

The following materials: Palladium chloride (PdCl2) anhydrous(Fisher), Ruthenium (III) chloride (RuCl3) 99.99% anhydrous (AlfaAeser), Ethylene glycol (EG) (Fisher), Potassium hydroxide (KOH)85% (EMD), Sodium hydroxide (NaOH) ACS grade (EM Science),carbon black (C) Vulcan XC-72R (Cabot), Nitrogen gas (N2) 99.9%(Linde), carbon monoxide (1000 ppm CO in He) (Linde), Nafionperfluorinated ion-exchange resin 5 wt.% solution in loweraliphatic alcohols/H2O mix containing 15e20% water from(Aldrich), and Ethanol 99.9% (Fisher), were used as received.

2.2. Catalyst preparation

The PdxRu1�x/C nanoparticles were prepared using a polyolmethod reported earlier [22,23]. In a typical synthesis, 0.25 g of themetal precursor salts (PdCl2 and RuCl3) were separately dissolvedin 50 mL of EG. Then appropriate amounts of the two salt solutionswere mixed to prepare PdxRu1�x (x ¼ 1, 0.99, 0.90, 0.95, 0.80, 0.50and 0). The solution pHwas adjusted to 8 by adding 0.06MNaOH inEG. The mixture was homogenized by stirring for 30 min at roomtemperature before refluxed at 160� for 2 h. To the resultingcolloidal NPs, appropriate amount of carbon black was added toobtain supported catalysts of 20 wt% loading. The mixture wasstirred for 48 h so as to achieve high dispersion and complete de-posit of PdeRu nanoparticles. The supported catalysts were thor-oughly washed and rinsed five times with de-ionized water(18U cm) to remove EG and salt ions through vacuum filtration anddried in the oven at 100 �C for 4 h.

2.3. Physical characterization

2.3.1. High-angle annular dark-field (HAADF) scanningtransmission electron microscopy (STEM) equipped with energy-dispersive X-ray spectroscopy (EDX)

The FEI Titan3 80e300 microscope equipped with a CEOS ab-erration corrector for the probe forming lens and a monochromaticfield-emission gunwas used to acquire HAADF-STEM images of thenanoparticles supported on carbon samples. The HAADF-STEMwasoperated at 300 k eV. The specimens were prepared by sonicating

E.A. Monyoncho et al. / Journal of Power Sources 287 (2015) 139e149 141

the as-prepared catalyst powders in ethanol. One drop of the so-lution was then placed onto a 200 mesh TEM copper grid coatedwith a lacey carbon support film (Ted Pella) and dried in air. The useof Image J software allowed for the determination of nanoparticlessize distribution.

2.3.2. X-ray diffraction (XRD)X-ray diffraction patterns were collected using a Rigaku Ultima

IV diffractometer using a Cu Ka X-ray source (l ¼ 1.54183 A, 40 kV,44mA). The diffraction patternswere recorded in the focused beamgeometry with a divergence slit of 2/3�, a scan speed of 0.17� min�1

and a scan step of 0.06� between 30� and 75�. The diffractionpatterns were collected from PdxRu1�x colloids with and withoutthe carbon support for catalyst with lower Ru content and the keyreflections were confirmed to be unchanged due to presence ofcarbon. However, for samples with more than 10 at.% Ru content,patterns from colloidal solutions were recorded to avoid thediminished peak intensities.

2.3.3. X-ray photoelectron spectroscopy (XPS)X-ray photoelectron spectroscopy measurements were per-

formed in a KRATOS Axis Ultra DLD with a Hybrid lens mode at140 W and pass energy of 20 eV using a monochromatic Al Ka. ThePd3d XPS core level spectra were analyzed using a fitting routinewhich decomposes each spectrum into individual mixed Gaus-sianeLorentzian peaks using a Shirley background subtraction overthe energy range of the fit. Their deconvolution was carried outusing doublets with spin orbit splitting 5.3 eV and intensity ratioPd3d5/2:Pd3d3/2 ¼ 3/2 [24], while a peak asymmetry was used inthe case of the Pd3d peak, which was attributed to the metallicstate as reported by Hufner et al. [25] The peak asymmetry of themetallic state was defined by using the sample of pure Pd afterreduction under H2. Binding energy scale was corrected using theC1s peak at 284.6 eV as an internal standard. The accuracy ofmeasurement of the binding energy is ±0.1 eV while that of FWHM±0.05 eV.

2.4. Electrochemical measurements

Cyclic voltammetry (CV) and chronoamperometry (CA) wereperformed using a BioLogic VSP potentiostat equipped with the EC-Lab software. All experiments were conducted at room tempera-ture in a customized two-compartment-cell made of Teflon. Aglassy carbon (GC) electrode of 0.1962 cm2 geometric surface areawas used as the working electrode. All potentials were measuredwith respect to mercury-mercurous sulfate (Hg/Hg2SO4, K2SO4)electrode (MSE) from Koslow scientific. The potentials are reportedverses MSE unless otherwise stated. A large surface area Pt-gauzeserved as a counter electrode. 1M KOH was used as the electro-lyte and was continuously purged with nitrogen gas. The GC-electrode was polished prior to each experiment using a solutionof 6 mm ground alumina on a polishing cloth. The catalyst inks wereprepared by dissolving 6 mg of PdxRu1�x/C powder in 1 ml of de-ionized water, 200 mL of ethanol, and 100 mL of Nafion solution.The mixture was sonicated for 10 min to form a homogeneousmixture. The ink solution (5 mL) was deposited onto the GC-electrode surface and dried in air at room temperature for 15 minand used as the working electrode.

The CV measurements were carried out between �1.4 Vand �0.4 V vs MSE at a scan rate of 20 mVs�1 in 1M KOH and 1Methanol previously reported to be the optimal conditions [26]. Thechronoamperometry measurements were performed at differentpotentials for 1.5 h. The CVs were recorded before and after each CAmeasurement to verify the stability of the catalyst. All currentdensities were normalized with respect to mass loading of Pd

(mAmg�1Pd) in order to compare the performance of the catalysts.CO stripping cyclic voltammograms were used to characterize

and compare the electrochemical surface activity of the catalysts.The electrolyte solution was first bubbled with N2 to remove anydissolved gas then saturated with CO for 20 min while holding theworking electrode potential at �1.1 V vs MSE to adsorb CO onto thecatalyst surface. The excess CO was then purged off the electrolyteusing N2 for 30 min before recording CVs at a scanning rate of25 mVs�1 [26]. All electrochemical experiments were carried out atroom temperature.

3. Results

3.1. Nanoparticles characterization

3.1.1. HAADF-STEM micrographsThe size distribution and surface morphologies of the nano-

particles were determined using HAADF-STEM. Fig. 1a shows arepresentative HAADF-STEM micrograph of Pd95Ru5/C andPd90Ru10/C catalysts with the corresponding histograms. Thenanoparticles were found to have a narrow size distribution be-tween 2 nm and 5 nm. Fig. 1b shows the micrographs of Pd/C,Pd99Ru1/C, Pd80Ru20/C and Pd50Ru50/C. The samples with �5 at.%Ru content showed poor dispersion while those with �5 at.% Rushowed better dispersion of the nanoparticles. The surfacemorphology of the nanoparticles was found to be rough (seeFigure S1), which increases the surface area of the catalyst andimproves the interaction of the adsorbate with the catalyst.

3.1.2. XRDThe X-ray diffraction patterns (XRD) of the nanoparticle are

shown in Fig. 2. The diffraction patterns are composed of face-centered-cubic (fcc) structure signature peaks at 40, 46 and 68� 2qcorrespond to (111), (200) and (220) reflections, respectively. The fccdiffraction patterns confirm that the bulk structure of Pd wasretained in the nanoparticles. As can be seen from Table 1, the 2qposition of Pd(111) peak for PdeRu catalysts is very close to pure Pd,indicating that Pd and Ru do not form an alloy. The interlayerspacing (d) for the (111) planes for Pd and PdeRu catalysts wasfound to be similar i.e 2.25 Å which confirms no alloy formationbetween Pd and Ru. These results are in agreement with the XPSdata (vide infra) and other experimental [27] and theoretical [28]reports. Therefore, PdxRu1�x/C nanoparticles formed a bi-phasecatalyst system. A peak shift of at least 0.2� 2q is required to havea lattice parameter value change by 0.01 which would be reasonablein order to conclude that an alloy is formed. In this work, the peakshifts were lower than 0.1� 2q. Since there were no diffraction peakscorresponding to crystalline Ru and no metallic Ru atoms weredetected by XPS, it is assumed that Ru exists in an amorphous phase.

The average crystallite size was estimated from peak positionsand the full width at half maximum (FWHM) of the Pd(111)reflection peak using Scherrer equation. The maximum peak in-tensity position (�2qmax) was determined from a second orderpolynomial fit to the top 20% of the experimental intensities around40� 2q. The FWHM were determined using the minimum intensitymeasured at around 55� 2q as the zero height. The Pd(111) peak2qmax, its FWHM and the calculated crystallite sizes for the cata-lysts are shown in Table 1. The crystallite size was found to decreasewith increasing Ru content but not in a linear trend.

3.1.3. XPSXPS was used to determine the elemental surface composition

and structure of the catalysts. Fig. 3 presents the deconvolutedPd3d and the corresponding Ru3p peaks for all samples and Table 2summarizes the peak positions binding energies (BE), FWHM and

Fig. 1. a: STEM images for Pd95Ru5/C (top left), Pd90Ru10/C (bottom left), and their corresponding particle size distribution histogram on the rights. b: STEM image of Pd/C, Pd99Ru1/C, Pd80Ru20/C and Pd50Ru50/C as labelled at the bottoms.

E.A. Monyoncho et al. / Journal of Power Sources 287 (2015) 139e149142

the relative intensities of each component and their assignments.The deconvolution of the Pd3d peak shows the existence of peaks ataround 335.5 (PdI), 336.6 (PdII), 337.7 (PdIII) and 338.6 eV (PdIV),which are attributed to palladium atoms in four different chemicalenvironments. The PdI is attributed to Pd in themetallic state, whilethe PdII is due to Pd2þ in PdO [29,30]. The PdIII is characteristic of Pdatoms in 4þ oxidation state and more specifically in PdO2 [31].Finally, the position of peak detected at higher BEs (PdIV) has beenascribed before to PdO3 [31] though its origin can be assigned toPdCl2. This last component is present in all samples implying thatsmall traces of unreacted PdCl2 exist in each sample, as confirmedby the chlorine peak detected (see Supporting information).However its relative intensity contribution does not representmorethan 9% of the surface Pd detected using XPS.

Due to the overlap of the Ru3d peak and the C1s peak, the Ru3pspectra were recorded and they are presented in Fig. 3B. The Ru3ppeak is relatively wide (3.8e4 eV, for the Ru3p3/2) and asymmetriceven if it represents an oxide [32] and for its deconvolution morestudies using reference samples are necessary. The positionthough of the recorded peaks can give important qualitative in-formation concerning the chemical environment of Ru atoms. Forthe Ru/C and the PdxRu1�x/C samples with relatively high Rucontent (10e50%) the obtained spectra are centered at around463.5 eV. This energy is characteristic of Ru4þ, e.g., in RuO2 [33].The rather high FWHM of peaks implies the existence of Ru atomsin more than one oxidation state which is a well-known phe-nomenon in literature [34]. For example, Francisco Colom inRef. [34] reports that in alkaline environment (the conditions usedin our synthesis) all the Ru atoms assume oxidation state (IV) and(VI) by spontaneous oxidation or reduction reactions. Therefore,Ru exists mainly in oxidized form such as the hydrated Ru4þ

(RuO2.nH2O) species or as the RuClx species [35]. In the case of thePdeRu samples with low content of Ru (99:1 and 99:5) the Rusignal is low. Thus for the 99:1 no Ru3p signal could be obtained(Fig. 3B, spectrum b), whereas the spectrum for 95:5 catalyst wasrecorded after prolonged acquisition which means that during the

measurement the X-ray beam have caused the partial reduction ofthe sample. The center of the Ru3p3/2 peak is shifted 1.3 eV tolower BEs and could be attributed to Ru in lower oxidation state.Though, this energy shift is considered as a result of the prolongedX-ray irradiation in order to increase the signal to noise ratio ofthis sample that caused a partial reduction and consequently theshift of the center of the peak.

To summarize, Ru in all cases is mainly in the form ofRuO2.nH2O without excluding the existence of other oxygenatedand/or chlorinated species in low percentage [35]. Using the in-tensity of the Pd3d and the Ru3p3/2 and their sensitivity factors[36] the Pd/Ru atomic ratios were calculated and the values areshown in Table 1. The obtained values are lower compared to theexperimental nominal values calculated from atomic ratios. Thisobservation indicates that surface composition is different fromthe bulk one.

From this XPS results several interesting observations can bedrawn out. As was reported earlier for PtIr nanoparticles preparedby polyol method [37] that in the case of PtIr alloys the Pt4f7/2 andthe Ir4f7/2 peaks are expected to shift to higher and lower BEs,respectively compared to the peak positions of the pure metals.This is because of the difference in the electrochemical potential ofelectrons in two metals that form an alloy, accompanied by re-hybridization of the d-band as well as the sp-band [37]. Accord-ing to our present results for PdeRu systems, no significant shift ofthe peaks attributed to the metallic state was observed. This mayindicate that Pd and Ru do not form an alloy in agreement with XRDpatterns.

For the samples with high Ru content, both PdO and PdO2 aredetected as shown in Fig. 3. However, at low Ru content Pd existsonly in one oxide form. More specifically, for the Pd99Ru1/C thedeconvolution of the Pd3d shows the presence of PdO2 only whilefor Pd95Ru5/C sample the less oxygenated species PdO was detec-ted. This observation maybe explained by the complex oxidationstates of Ru and their stability/or reduction potentials vs those ofPd. Pd is known to exist mainly in oxidation state (II) and (IV) in

Fig. 2. The XRD patterns of PdxRu1�x nanoparticles.

Table 1Summary of fcc Pd(111) characteristics and crystallite sizes for PdxRu1�x catalysts.

Catalyst 2q max Intensity (cps) FWHM �2q Crystallite size(nm)

Pd/C 40.06 890 0.55 15.4Pd99Ru1/C 40.08 286 1.22 6.9Pd95Ru5/C 40.09 66 1.27 6.7Pd90Ru10 40.09 98 1.21 7.0Pd80Ru20 40.05 80 1.08 7.8Pd50Ru50 40.05 64 1.95 4.3

E.A. Monyoncho et al. / Journal of Power Sources 287 (2015) 139e149 143

compounds with other elements, while Ru can exist in oxidationstates ranging from 0 for metal carbonyls to (VIII), and because ofthe capacity of its ions to form polynuclear complexes apparentfractional oxidation states are also known [34]. Therefore, with verysmall amounts of Ru content in Pd99Ru1/C, in the alkaline syntheticconditions used, all the Ru atoms exists in higher oxidation states,which are unstable and tends to get charge from Pd atoms/ionshence oxidizing it to Pd4þ. As the Ru content is increased otherintermediate oxidation states of Ru could be formed, which caneasily be oxidized and hence reduce Pd4þ to Pd2þ. It is interesting tonote that as the amount of Ru was increased, the mixture of Pdoxide species (PdO and PdO2) was formed, which reflects the dy-namic nature of the reaction intermediates. Consequently, as thesurface oxide species increased the surface metallic Pd speciesdecreased. The total Pd oxide species increased from 4.5% for Pd/Cto 4.6%, 13.9%, 17.6%, 22.9%, and 27.4% for Pd99Ru1/C, Pd95Ru5/C,Pd90Ru10/C, Pd80Ru20/C, and Pd50Ru50/C, respectively (see Table 2for the percentage distribution of each oxide species). This obser-vation strongly indicates that the presence of Ru affects the surfaceoxidation state of Pd.

3.1.4. EDXEnergy-dispersive X-ray spectroscopy was used qualitatively to

confirm individual elemental composition of the nanoparticles.Fig. 4 shows the representative zoomed in micrographs and theircorresponding EDX analysis of the catalysts i.e. Pd99Ru1/C,Pd80Ru20/C and Pd50Ru50/C. Please note Cu and Mo counts are ar-tefacts from the sample grid and holder, respectively. Ru was notdetected for Pd99Ru1/C which was expected because of the smallconcentration. All the other samples did show that the Pd nano-particles are enriched with Ru.

3.2. Electrochemical studies

3.2.1. CO stripping voltammetryThe catalyst surface activities were characterized by carbon

monoxide oxidation voltammograms. CO is a major poisoning in-termediate during alcohol electrooxidation reaction. Fig. 5 showsthe CO stripping voltammograms in 1M KOH at room temperature.The 1st CV cycle of the CO stripping for PdxRu1�x/C catalyst with�10 at.% Ru content showed distinct peaks (Fig. 5a) while catalystswith higher Ru loading had no peaks (Fig. 5b). The 2nd cycle showsno CO oxidation peak indicating a complete removal of CO from thecatalyst surface during the 1st cycle. The CO oxidation peak for Pd/Cand Pd99Ru1/C is at the same potential (~�0.66 V vs MSE) sug-gesting similar active sites, which can be attributed to metallic Pdbased on XPS data (Fig. 3 and Table 2). However, the presence of ashoulder (pre-peak) and higher charge density for Pd99Ru1/C in-dicates the effect of surface structure-activity modification due tothe presence of small amounts of Ru. The pre-peak shoulder showsthat the presence of Ru lowers CO oxidation potential, whileincreasing the current density. The effect of Ru is amplified with theincreasing Ru content, which shows the CO oxidation peak shift tomore negative potentials. For Pd95Ru5/C, a single almost

Fig. 3. (A) Pd3d and (B) Ru3p XPS spectra of Pd/C (a), Pd99Ru1 (b), Pd95Ru5 (c), Pd90Ru10 (d), Pd80Ru20 (e), Pd50Ru50 (f).

E.A. Monyoncho et al. / Journal of Power Sources 287 (2015) 139e149144

symmetrical shaped CO oxidation peak was observed at ~�0.83 V,while for Pd90Ru10/C catalyst, two overlapping peaks at ~�0.79 Vand �0.71 V were obtained. This finding re-enforces literature re-ports that the presence of Ru mitigates CO poisoning in Pt catalysts[38]. The absence of CO oxidation peaks for Pd80Ru20/C andPd50Ru50/C is attributed to the presence of higher Ru-oxides, whichreadily oxidizes CO at a lower potential than the adsorption po-tential of e 1.1 V used in this study, hence leading to no CO adlayeron the catalyst surfaces.We suggest themechanism to involve RuOxbecause metallic Ru is unstable in high pHs. It has also been re-ported that in alkaline conditions RuO4

� ions readily oxidize watermolecules liberating oxygen [34], which would then react withadsorbed CO molecules to form CO2. Fisher et al. had attributed theshifting of CO oxidation peak to lower potentials to the segregationof Ru on Pd, however this was done in acidic solution, wheremetallic Ru is stable for CO adsorption [39]. The insights from COoxidation graphs (shape and position of the peaks) helps indescribing surface structure-activity of the catalysts. For example,in Refs. [40e42] it was used to identify the active sites and mech-anism for CO oxidation on Pt-based electrodes.

Fig. 5 also reveals significant differences for PdxRu1�x catalystbased on their PdOx reduction peaks centered at ~�0.7 V. Pd/C didnot form significant PdOx due to lower anodic potential limitof �0.4 V we used leading to weak reduction peak during thereverse scan. As the surface Pd-oxides increase due to addition ofRu, the PdOx reduction peaks becomes stronger as shown by cat-alysts Pd99Ru1/C, Pd95Ru5/C and Pd90Ru10/C. Further increase of Rucontent changes the shape of the CV in a such a way that thecapacitive current increases, whereas PdOx reduction peakdecreases.

3.2.2. Cyclic voltammetry (CV)Cyclic voltammetry was used to study ethanol oxidation

behavior on PdxRu1�x/C. Fig. 6 shows the CVs in 1M KOH with andwithout ethanol. During the forward scan, currents starts to in-crease at ~ �0.99 V for Pd/C due to ethanol oxidation, reaches amaximum at ~ �0.66 V, followed by current decrease due to thesurface deactivation by Pd oxide coverage and/or ethanol oxidationintermediates. Addition of 1%Ru lowered onset oxidation potentialto ~�1.16 V and significantly increased the current density by more

Table 2Summary of the binding energies (BE), FWHM and the relative intensities of each of the peak components and their assignments.

Catalyst Pd3d5/2 Ru3p3/2 Pd/Ru atomic ratioa

B.E. (eV) FWHM (eV) % Relative intensities Assignment BE (eV)

Pd on C 335.5 1 95.5 Metallic Pd e e

Pd99Ru1/C 335.6 1 86.4 Metallic Pd e e

337.6 1.2 4.6 PdO2

338.7 1.4 9 PdCl2Pd95Ru5/C 335.5 1 77.2 Metallic Pd 462.3 5.4 (19)

336.7 1.2 13.9 PdO338.5 1.4 8.9 PdCl2

Pd90Ru10/C 335.4 0.9 75 Metallic Pd 463.6 4.8 (9)336.5 1.1 7.6 PdO337.7 1.3 10 PdO2

338.7 1.4 7.4 PdCl2Pd80Ru20/C 335.5 0.9 70 Metallic Pd 463.3 1.2 (4)

336.7 1.15 11.5 PdO337.7 1.3 11.4 PdO2

338.7 1.4 7.1 PdCl2Pd50Ru50/C 335.5 0.9 65 Metallic Pd 463.5 0.4 (0.96)

336.6 1.15 14 PdO337.6 1.3 13.4 PdO2

338.5 1.4 7.6 PdCl2Ru on C e e e e 463.45 e

a The values in parenthesis shows the experimental nominal values.

Fig. 4. EDX spectra (right) and their corresponding micrographs (left) for Pd99Ru1/C (top), Pd80Ru20/C (middle) and Pd50Ru50/C. Cu and Mo counts are artefacts from the sample gridand holder, respectively.

E.A. Monyoncho et al. / Journal of Power Sources 287 (2015) 139e149 145

Fig. 5. Cyclic voltammograms of the electrocatalytic oxidation of CO on PdxRu1�x/C nanoparticles recorded in 1.0 M KOH at scan rate of 25 mVs�1. Current densities are normalizedwith respect to mass loading of Pd on the electrode.

Fig. 6. Cyclic voltammograms of the electrocatalytic ethanol oxidation reaction (EOR) of PdxRu1�x/C recorded in 1.0 M KOH þ 1.0 M C2H5OH (solid line) and in 1.0 M KOH (dottedline) at a scanning rate of 20 mVs�1. Current densities idem.

Table 3Summary of CVs oxidation potentials and CAs steady-state current densities after2000 s for PdxRu1�x/C catalysts.

Catalyst Pd loading (mg) Anodic E (V) i (mA mg�1 Pd)

Eonset E at imax E ¼ �0.96 V E ¼ �0.67

Pd/C 0.0046 �0.99 �0.66 0.00 5.79Pd99Ru1/C 0.0046 �1.16 �0.65 2.40 38.14Pd95Ru5/C 0.0044 �1.21 �0.64 0.86 28.87Pd90Ru10/C 0.0050 �1.23 �0.63 6.67 15.33Pd80Ru20/C 0.0039 �1.24 �0.60 4.05 6.21Pd50Ru50/C 0.0024 �1.28 �0.68 6.17 0.94

E.A. Monyoncho et al. / Journal of Power Sources 287 (2015) 139e149146

than 400% (see the differences of the y-scale on Fig. 6). Increasingthe amount of Ru shifted further the ethanol oxidation onset po-tential to lower values. The onset oxidation potentials and currentdensities for all samples are summarized in Table 3. It was foundthat increasing Ru content on Pd nanoparticles lowered the onsetpotentials for ethanol oxidation up to a maximum shift of 290 mVfor Pd50Ru50/C as shown in Fig. 6 and Table 3. A higher shift wasachieved in this work compared to 150 mV reported by Chen et al.[7]. In the reverse scan, the catalyst remains deactivated upto ~ �0.65 V then current rises sharply as more ethanol is oxidizedon the reduced surface. As shown in Fig. 5, the surface oxide

E.A. Monyoncho et al. / Journal of Power Sources 287 (2015) 139e149 147

reduction starts at ~ �0.60 V and reaches a minimum at ~ �0.70 V.Ethanol electrooxidation peak on Pd/C and Pd99Ru1/C have

similar shape and they occur at the same potential (Fig. 6) implyingsimilar surface structure active sites, which correlates well with COstripping CVs (Fig. 5). For Pd95Ru5/C to Pd50Ru50/C, an oxidationshoulder was observed at a lower potential at ~ �0.95 V. Theshoulder gets more pronounced with increasing Ru content. Thisshoulder was attributed to the synergetic effect between surfaceoxides (PdOx and RuOx) and Pd nanoparticles as discussed in Sec-tion 4.

Fig. 7 compares the forward scans of EOR on PdxRu1�x/C cata-lysts. The current densities were found to vary non-linearly as afunction of increasing Ru content within the voltage windowcovered (�1.4 V to �0.4 V). The current densities below �0.9 Vshowed different trends to the current densities above �0.9 V. Thecurrent densities at lower potentials are of great interest for fuelcell applications. Therefore, we have tabulated the current densitiesat �0.96 V in Table 3 to show the effect of Ru on the catalytic ac-tivity of Pd nanoparticles. At lower voltage (�0.96 V) the currentdensity from Pd/C is zero and it increases non-linearly with theincreasing amount of Ru to a maximum of 6.67 mA mg�1Pd forPd90Ru10/C.

3.2.3. Chronoamperometry (CA)Chronoamperometry was used to investigate the steady-state

performance of the catalysts. CA measurements were carried outat onset and peak potentials for EOR on Pd/C catalyst, i.e., �0.96 Vand �0.67 V, respectively. The behavior of PdxRu1�x/C catalysts atthese potentials provides insights into the effect of adding Ru to Pdsurface at lower and higher potentials. The steady-state currentdensity is zero at onset oxidation potential (�0.96 V) for Pd/C, whileit is at maximum at the peak oxidation potential (�0.67 V). Theresulting i-t curves at �0.96 V and at �0.67 V are shown in Fig. 8.The steady-state current densities after 2000s from the two figuresare tabulated in Table 3. The i-t curves show initial high currents,which rapidly decreases until a steady-state is established. Therapid decrease in current at the beginning is attributed to surfacecoverage by partially oxidized species, which block the surfaceactive sites of the catalyst. Therefore, the initial slopes of the i-tcurves are good indicators of the catalyst surface reaction kineticsand the extend of the catalyst surface poisoning due to adsorbedoxidized intermediates and or products. The steeper the slope ofthe i-t curves the higher the deactivation of the catalyst active sites.

Fig. 7. Cyclic voltammograms (forward scan) of PdxRu1�x/C electrocatalysts recordedin 1.0 M KOH þ 1.0 M C2H5OH at 20 mVs�1. Current densities idem. The vertical dashedlines mark the potential of the chronoamperometry experiments in Fig. 8.

4. Surface composition structure-activity correlations

We have shown that combining Pd with Ru improves the cata-lytic activity towards EOR in alkaline media. Now let us take adeeper look at what is influencing the catalytic activity of thenanoparticles tested at the relevant electrode process level, i.e., thesurface composition and morphology of the catalysts. Fig. 9 showsthe surface atomic composition of the catalysts in terms of metallicPd and PdOx as a function of Ru content and the correspondingsteady-state current densities after 2000s at �0.96 V and �0.67 Vvs MSE. The atomic composition is from XPS data in Table 2 and thecorresponding CA current densities are from Table 3. The surface Pdmetallic species decreases as the coverage with surface oxide spe-cies increases as expected. Fig. 9 provides evidence that there existsa correlation between surface composition and Ru content.Increasing the amount of Ru increases the Pd-oxide species on thenanoparticle surfaces. Since Ru was found to exist mainly in theoxidized state based on XPS data and discussion above, it is logicalto assume that the amount of Ru oxides on the surface increasedtoo as the Ru content was increased.

From the CVs (Figs. 6 and 7), a correlation would be made fromthe single EOR peak at �0.67 V, which is almost symmetric, forPd99Ru1/C and Pd/C suggesting that their electrocatalytic activityoccurred on similar surface active sites. Fig. 9 shows that the sur-face composition for Pd/C and Pd99Ru1/C catalyst to be ~95.5% and~86.4% in metallic Pd state, respectively. Hence, the increased EORcurrent density for Pd99Ru1/C is due to presence of Ru ions in vi-cinity to metallic Pd. However, recent theoretical studies haveshown that it is the adsorbed OH on Pd surface rather than Pdatoms which is the active center for the EOR [43] Therefore, EORpeak at �0.67 V, which is present in all samples, would be assignedto surface metallic Pd active sites. Increasing the amount of Ru inthe nanoparticles was found to diminish this maximum currentpeak, imax, by up to >55% for catalyst with more than 5%Ru content,a trend consistent with the decreasing amount of surface metallicPd shown in Fig. 9. Traces of Ru at 1% were found to enhanceethanol electrooxidation leading to more than 400% imax comparedto Pd/C (see Fig. 6 and note the y-scale difference). Increasing Rucontent further decreased surface metallic Pd species from 86% inPd99Ru1/C to 65% in Pd50Ru50/C. Consequently, the steady-statecurrent densities decreased at �0.67 V as shown in Fig. 9b. It isimportant to note that Pd90Ru10/C showed highest mass activityat �0.96 V (6.67 mA mg�1Pd) while Pd99Ru1/C showed highestmass activity at �0.67 V (38.14 mA mg�1Pd) after 2000s whichhighlights the differences in extrinsic and intrinsic properties of thecatalysts.

The forward scans on Fig. 7 shows that Ru content between 1and 20% improves Pd activity towards EOR at lower potential(�0.96 V) which is further supported by CA analysis as shown inFig. 9a. Fig. 9 demonstrates that at lower potentials the currentdensities have an increasing trend as Ru content is increased andvice versa at higher potentials with a limiting activity at 10 at.% Rucontent. This limiting surface composition ratio is in good agree-ment with values reported in literature [15,16]. For example, Yi Q.et al. reported an optimal ratio of 13% Ru for PdeRu nanoparticlesprepared by hydrothermal method [15]. Anindita et al. reported theoptimum ratio of 20% Ru for PdeRu nanoparticles synthesized viasodium borohydride reduction method [16]. Interestingly, Chenet al. reported a higher optimum ratio of 50% Ru for PdeRu catalystsprepared by impregnation and reduction method [7]. However,Liang Ma et al. using PdeRu nanoparticles prepared by impregna-tion and sodium borohydride reduction method reported thatcatalysts with 25% Ru performed well [21]. But we have shown inthis work that the promotional effect of Ru on EOR on Pd is greatlydepended on the oxidation potential. It is important to note that

Fig. 8. Chronoamperometric curves of PdxRu1�x/C electroctalysts at (a) E ¼ �0.96 V and (b) E ¼ �0.67 V vs MSE recorded in 1.0 M KOH þ 1.0 M C2H5OH. Current densities idem.

Fig. 9. Surface atomic composition of PdxR1�x/C from XPS data (black squares and red triangleseleft axis) and the corresponding steady-state current density at E ¼ �0.96 V (panela) and at E ¼ �0.67 V (panel b) vs MSE (green triangleseright axis). Original data from Tables 2 and 3. (For interpretation of the references to colour in this figure legend, the readeris referred to the web version of this article.)

E.A. Monyoncho et al. / Journal of Power Sources 287 (2015) 139e149148

these ratios are based on the nominal values and not the actualsurface composition of the catalyst.

The appearance of EOR shoulder at lower potentials shown inFigs. 6 and 7 would be attributed to the synergetic effect betweenthe surface oxide species (Ru-oxides and Pd-oxides) andmetallic Pdspecies. Although Ru-oxide species, particularly hydrous Ru oxide(RuOx.nH2O), had been reported to be the active site for methanoloxidation [44], it is not the only contributing factor in the currentcase. This is because Ru/C catalyst with 100% Ru-oxides surfacecomposition was found to be a poor catalyst for EOR as shown inthe Supporting information (Figure S2). The current densities forEOR showed increasing trend at lower potentials (Fig. 9a) as thesurface oxides were increased and plateaus at 10 at.% Ru which isthe limiting composition for a better performing catalyst at lowerpotential. At � 20%Ru content, Pd nanoparticles are completelycovered with Ru oxides which are poor EOR catalyst alone. There-fore, we propose that the synergetic effect is better expressed whenthe boundaries between surface oxides and metallic Pd form aninterface with the electrolyte solution during electrooxidationreaction.

5. Conclusions

In conclusion, we have shown that PdxRu1�x/C nanoparticlessynthesized by a polyol method form a bi-phase catalyst system. Noalloying between Pd and Ru is formed as revealed by XPS and XRD.The nanoparticle size distribution was found to range between 2and 5 nm and the catalysts have a rough topology as show fromSTEM micrographs. The bulk nanoparticles were found to have anfcc structure indicating that they were mainly composed of Pdmetal from XRD patterns. XPS and EDX analysis shows that thenanoparticle surfaces are composed of different percentages ofmetallic Pd species, Pd-oxide species, and Ru-oxide species. The Pd-

oxide species were found to increase while the metallic Pd speciesdecreased with the increasing amount of Ru content, a phenome-non attributed the complex oxidation states of Ru which rangesfrom (0) to (VII).

We report that the synergetic effect between surface oxidespecies (PdOx and RuOx) on Pd nanoparticles lowers EOR potentialand 10 at.%Ru is the limiting PdxRu1�x/C composition for goodperformance at lower potential. The Pd90Ru10/C and Pd99Ru1/Cwere found to be the best catalyst systems which produced morethan four times higher mass activity (current density per mass ofPd) compared to pure Pd at �0.96 V and �0.67 V vs MSE, respec-tively. The i-t curve slopes show that the two catalysts have lowersurface deactivation from EOR intermediates/products. This workshows that Ru improves ethanol oxidation on Pd nanoparticles inalkaline media and the current densities are depended on theoxidation potential. CO stripping studies show that PdeRu combi-nation mitigates CO poisoning in alkaline media which is a majorchallenge in acidic media. Therefore, PdeRu is a promising bime-tallic combination that can be optimized for use in ambient con-ditions DAFCs.

Acknowledgment

Authors would like to thank Dr. Martin Couillard from NationalResearch Council for his help with HAADF-STEM micrographs. Thiswork was supported by the Natural Science and EngineeringResearch Council (NSERC), and the University of Ottawa, Canada.

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.jpowsour.2015.03.186.

E.A. Monyoncho et al. / Journal of Power Sources 287 (2015) 139e149 149

References

[1] M.Z.F. Kamarudin, S.K. Kamarudin, M.S. Masdar, W.R.W. Daud, Int. J. HydrogenEnergy 38 (2013) 9438.

[2] G. Andreadis, V. Stergiopoulos, S. Song, P. Tsiakaras, Appl. Catal. B 100 (2010)157.

[3] N. Fujiwara, Z. Siroma, S. Yamazaki, T. Ioroi, H. Senoh, K. Yasuda, J. PowerSources 185 (2008) 621.

[4] A.Y. Tsivadze, M.R. Tarasevich, V.N. Andreev, V.A. Bogdanovskaya, Russ. J. Gen.Chem. 77 (2007) 783.

[5] E. Antolini, E.R. Gonzalez, J. Power Sources 195 (2010) 3431.[6] E.H. Yu, U. Krewer, K. Scott, Energies 3 (2010) 1499.[7] Y. Chen, L. Zhuang, J. Lu, Chin. J. Catal. 28 (2007) 870.[8] M. Ammam, E.B. Easton, J. Power Sources 215 (2012) 188.[9] C. Bianchini, P.K. Shen, Chem. Rev. 109 (2009) 4183.

[10] Y.-Y. Feng, Z.-H. Liu, Y. Xu, P. Wang, W.-H. Wang, D.-S. Kong, J. Power Sources232 (2013) 99.

[11] V. Bambagioni, C. Bianchini, A. Marchionni, J. Filippi, F. Vizza, J. Teddy, P. Serp,M. Zhiani, J. Power Sources 190 (2009) 241.

[12] Y. Wang, T.S. Nguyen, X. Liu, X. Wang, J. Power Sources 195 (2010) 2619.[13] C. Xu, P. Kang Shen, Y. Liu, J. Power Sources 164 (2007) 527.[14] V.D. Colle, A. Bern�a, G. Tremiliosi-Filho, E. Herrero, J.M. Feliu, Phys. Chem.

Chem. Phys. 10 (2008) 3766.[15] Q. Yi, F. Niu, L. Song, X. Liu, H. Nie, Electroanalysis 23 (2011) 2232.[16] R. Anindita, Awasthi, Madhu, R.N. Singh, Open Catal. J. 4 (2011) 88.[17] J. Bagchi, S.K. Bhattacharya, Transit. Met. Chem. 33 (2008) 113.[18] M.C. Correia, A. Moghieb, S.K. Goforth, L. McElwee-White, ECS Trans. 19

(2009) 13.[19] J. Bagchi, S.K. Bhattacharya, Transit. Met. Chem. 32 (2007) 47.[20] S. Sun, Z. Jusys, R.J. Behm, J. Power Sources 231 (2013) 122.[21] L. Ma, H. He, A. Hsu, R. Chen, J. Power Sources 241 (2013) 696.[22] E.A. Baranova, T. Amir, P.H.J. Mercier, B. Patarachao, D. Wang, Y. Page, J. Appl.

Electrochem 40 (2010) 1767.[23] M. Figlarz, F. Fievet, J.P. Lagier, Reduction of Metal Compounds to Metal

Powders by Polyols, 1984.[24] G. Johansson, J. Hedman, A. Berndtsson, M. Klasson, R. Nilsson, J. Electron

Spectrosc. Relat. Phenom. 2 (1973) 295.[25] S. Huefner, G.K. Wertheim, Phys. Rev. B 11 (1975) 678.[26] F. Miao, B. Tao, Electrochim. Acta 56 (2011) 6709.[27] J. Storm, R.M. Lambert, N. Memmel, J. Onsgaard, E. Taglauer, Surf. Sci. 436

(1999) 259.[28] S. Curtarolo, D. Morgan, G. Ceder, Calphad 29 (2005) 163.[29] H. Tollefsen, E.O. Laastad, S. Raaen, Surf. Sci. 603 (2009) 197.[30] M. Brun, A. Berther, J.C. Bertolini, J. Electron Spectrosc. Relat. Phenom. 104

(1999) 55.[31] J.M. Tura, P. Regull, L. Victori, de C. Dolors M, SIA, Surf. Interface Anal. 11

(1988) 447.[32] H.J. Lewerenz, S. Stucki, R. Koetz, Surf. Sci. 126 (1983) 463.[33] Y. Liang, J. Li, Q.-C. Xu, R.-Z. Hu, J.-D. Lin, D.-W. Liao, J. Alloy. Compd. 465

(2008) 296.[34] Standard Potentials in Aqueous Solution, M. Dekker, New York, 1985. New

York, 1985.[35] A.S. Arico, V. Baglio, B. Di A, E. Modica, P.L. Antonucci, V. Antonucci,

J. Electroanal. Chem. 557 (2003) 167.[36] D. Briggs, M.P. Seah, Practical Surface Analysis: Auger and X-Ray Photoelec-

tron Spectroscopy, Wiley, New York, 1996.[37] A. Allagui, M. Oudah, X. Tuaev, S. Ntais, F. Almomani, E.A. Baranova, Int. J.

Hydrogen Energy 38 (2013) 2455.[38] J. Thepkaew, S. Therdthianwong, A. Therdthianwong, A. Kucernak,

N. Wongyao, Int. J. Hydrogen Energy 38 (2013) 9454.[39] J.M. Fisher, N. Cabello-Moreno, E. Christian, D. Thompsett, Electrochem. Solid-

State Lett. 12 (2009) B77.[40] A. L�opez-Cudero, �A. Cuesta, C. Guti�errez, J. Electroanal. Chem. 586 (2006) 204.[41] A. L�opez-Cudero, A. Cuesta, C. Guti�errez, J. Electroanal. Chem. 579 (2005) 1.[42] A. Cuesta, A. Couto, A. Rincon, M.C. Perez, A. Lopez-Cudero, C. Gutierrez,

J. Electroanal. Chem. 586 (2006) 184.[43] T. Sheng, W.-F. Lin, C. Hardacre, P. Hu, J. Phys. Chem. C 118 (2014) 5762.[44] D.R. Rolison, P.L. Hagans, K.E. Swider, J.W. Long, Langmuir 15 (1999) 774.