catalysts Article Electrocatalytic Performance of Carbon Supported WO -Containing Pd–W Nanoalloys for Oxygen 3 Reduction Reaction in Alkaline Media NanCui,WenpengLi*,ZengfengGuo,XunXuandHongxiaZhao KeyLabofFineChemicalsinUniversitiesofShandong,InstituteofAdvancedEnergyMaterialsandChemistry, SchoolofChemistryandPharmaceuticalEngineering,QiluUniversityofTechnology(ShandongAcademy ofSciences),DaxueRoad,ChangqingDistrict,Jinan250353,China;[email protected](N.C.); [email protected](Z.G.);[email protected](X.X.);[email protected](H.Z.) * Correspondence:[email protected]@163.com;Tel.:+86-531-8963-1208 (cid:1)(cid:2)(cid:3)(cid:1)(cid:4)(cid:5)(cid:6)(cid:7)(cid:8)(cid:1) (cid:1)(cid:2)(cid:3)(cid:4)(cid:5)(cid:6)(cid:7) Received:7April2018;Accepted:21May2018;Published:24May2018 Abstract: In this paper, we report that WO containing nanoalloys exhibit stable electrocatalytic x performance in alkaline media, though bulk WO is easy to dissolve in NaOH solution. Carbon 3 supported oxide-rich Pd–W alloy nanoparticles (PdW/C) with different Pd:W atom ratios were prepared by the reduction–oxidation method. Among the catalysts, the oxide-rich Pd W /C 0.8 0.2 (Pd/W = 8:2, atom ratio) exhibits the highest catalytic activity for the oxygen reduction reaction. TheX-rayphotoelectronspectroscopydatashowsthat~40%ofPdatomsand~60%oftheWatoms areintheiroxideform. ThePd3d bindingenergyoftheoxide-richPd–Wnanoalloysishigher 5/2 than that of Pd/C, indicating the electronic structure of Pd is affected by the strong interaction betweenPdandW/WO . ComparetoPd/C,theonsetpotentialoftheoxygenreductionreaction 3 attheoxide-richPd W /Cshiftstoahigherpotential. Thecurrentdensity(mA·mgPd−1)atthe 0.8 0.2 oxide-richPd W /Cis~1.6timesofthatatPd/C.Theoxide-richPd W /Calsoexhibitshigher 0.8 0.2 0.8 0.2 catalyticstabilitythanPd/C,whichdemonstratesthatitisaprospectivecandidateforthecathodeof fuelcellsoperatingwithalkalineelectrolyte. Keywords:WO ;electrocatalysts;alkaline;Pd–Walloy;oxygenreductionreaction;reduction–oxidation 3 method 1. Introduction Thestudyoftheoxygenreductionreaction(ORR)hasahistoryofmorethanonecenturysince Grove fabricated the earliest hydrogen–oxygen fuel cell with Pt as the catalyst for ORR in 1839. Inrecentyears,thestudiesofORRhavebeenpromotedbytheincreasingdemandofcleanenergy technologylikefuelcells. Astheenergyefficiencyandbatteryvoltageofelectrochemicalcellsare limitedbytheslowkineticsoftheORR[1,2], thereisagreatneedforhighlyefficientcatalystsfor ORR.VariouselectrocatalystsforORRhavebeendeveloped,includingbutnotlimitedtoPt-based catalysts[3–5],Pd-basedcatalysts[6,7],catalystsbasedonnon-preciousmetals[8,9],catalystsbased oncarbonnanostructure/nanocomposites[10–13],catalystsbasedonmetaloxides[14,15],catalysts based on metal–organic frameworks [16,17], catalysts based on complexes [18,19], enzyme-based catalysts[20–24],metalcarbides[25–28],andsoon. AmongthecatalystsforORR,Pt-basedcatalysts are regarded as the most active catalysts [29]. However, platinum’s scarcity limits the large scale applicationofPt-basedelectrocatalysts. Palladiumhasbeenusedasoneofthealternativecandidates beingabout200timesmoreabundantintheearththanplatinum. Therehavebeensomereviewsabout Pd-basedelectrocatalysts[30–32]. TheORR[33]canbeperformedunderbothacidconditionsand alkalineconditionsinfuelcells. Ithasbeenreportedthatalkalinemediaareabenefitforthekinetics Catalysts2018,8,225;doi:10.3390/catal8060225 www.mdpi.com/journal/catalysts Catalysts2018,8,225 2of14 ofORR[34]. Inalkalinesolutions,theoxygencanbereducedthroughafour-electronpathwayora two-electronpathway[35]. AlotofnovelPd-basedelectrocatalystsforORRexist,includingcarbon ormetalsupportedPdalloys[36–38],nitrogenandsulfurco-dopedcarbonsupportedPdNicatalyst (PdNi-NS/C)[39],PdsupportedonTiO withoxygenvacancy(Pd/TiO –Vo)[40],PdWnanoparticles 2 2 supportedonsulfur-dopedgraphene(PdW/SG)[41],PdNiCu/PdNiCosupportedonnitrogendope graphene[42],PdSnCo/nitrogen-doped-graphene[43],electrochemicallyreducedgraphene-oxide supported Pd-Mn O nanoparticles [44], AuPd@PdAu alloy nanocrystals [45], three-dimensional 2 3 nitrogen-doped graphene supports for palladium nanoparticles (Pd-N/3D-GNS) [46], and so on. MostoftheresearchaboveonPd-basedelectrocatalystsforORRinalkalinemediaissupportedon graphenespeciallytreated(doping,modifying,andsoon). Althoughcarbonblackisthemostused supportfornoblemetalelectrocatalystsinfuelcells,Pd-basedelectrocatalystssupportedoncarbon black(C)forORRinalkalinemediahavebeenrarelyreportedoverthepastfewyears. Besidesthe boomofnovelsupportmaterialslikedopedgraphene,oneofthepossiblereasonsisthehighactivity ofPd/CforORRinalkalinemedia. ItwasfoundthatPd/Cexhibitssignificantlyhighactivitycloseto Pt/Cinalkalinesolutions[37,38],thereforeitisdifficultforotherelectrocatalystsforORRinalkaline mediatoexhibitmuchhigheractivitythanPd/C.Thenewstudiesaboutcarbon-blacksupportingPd basedcatalystsforORRinalkalinemediahavetofacethedifficultsituationofbeingcomparedtothe ultra-highactivecatalystPd/C. Aftertheircalculationsbasedonquantummechanics,Goddardet.al.[47]predictedthatPd W 3 wasaprospectivecatalystforORR,whichwasconfirmedbyourpreviousworkforPd W inacid 0.7 0.3 media[48]. Todate,Pt-basedelectrocatalystsarecommonlyusedincommerciallyavailableelectric vehiclespoweredbyfuelcells. GoddardandhiscoworkersexaminedthecriticalbarriersoftheORR withPd WandcomparedthemtotheanalogousbarriersforPdandPt. Theresultsdemonstratedthat 3 Pd WexhibitsORRpropertiesgreatlyimprovedoverpurePdandclosetothatofpurePt. Sincethe 3 cost of Pd W is six times less than that of pure Pt, a highly efficient Pd–W system is a promising 3 candidateforfutureapplication. Inthiswork,weattempttofabricatehighperformancePd–W/C systemsforORRinalkalinemedia. Mostofthenoblemetalelectrocatalystsusedinfuelcellsarein theformofnaonoparticlessupportedoncarbon. Sincethesurfaceofmetalnanoparticlesiseasyto beoxidizedbyambientair,theeffectofoxidesinthePd-basedcatalystsforORRinalkalinemedia should be discussed. There have been some state-of-art electrocatalysts based on oxides such as Fe O @NiFexO [49]. However,astheresistanceofsemiconductoroxidesishigherthanthatofmetals, 3 4 y theoutputvoltagewilldecreasewhentheoxidesaredirectlyusedaselectrodematerialsinfuelcellsor metal–airbatteries. Combiningtheoxidewithahigh-conductivemetalineachofthenanoparticlesof catalystsisasolutionforavoidingthedecrease. Theinteractionofmetalandmetaloxidesincatalysts hasattractedresearchinterestsfordecades[50–53]. Ithasbeenreportedrecentlythatmetalandmetal oxidesinteractionsaregreatlyaffectedbythecatalyticconsequenceofelectrocatalysisreactionssuch astheoxygenreductionreaction[54]andtheethanoloxidationreaction[55,56]. BulkWO crystal 3 canbedissolvedinstrongNaOHsolutions,whichlimitsitsdirectapplicationinfuelcellsoperating withalkalineelectrolytes. Tosolvethisproblem,westartedbyseparatingtheWatomswithnoble metalssuchasPdintheatomicscalebeforetheiroxidation. Thereforethechemicalbondsattachedto mostoftheWatomsarenotW–O–WbondsbutPd–Wmetallicbonds,whicharemorestablethan W–O–Wbondsinalkalinesolutions. AccordingtotheMonteCarlosimulation[57,58],alloyclustersat thesurfaceofnano-materialssometimesexhibithigherstability. Then,wefabricatedWO -containing x Pd–Wnanoalloyswiththereduction–oxidationmethod(Scheme1). TheonsetpotentialofORRat theaspreparedoxide-richPd W /C(Pd/W=8:2, metalatomicratio)isclosetothePd/Cand 0.8 0.2 Pt/C fabricated with the chemical reduction method [36]. The ORR stability and current density (mA·mgPd−1)oftheoxide-richPd W /CarehigherthanthoseofPd/C,whichindicatesthatthe 0.8 0.2 oxide-richPd W /Cisaprospectivecandidateforthecathodeoffuelcells. 0.8 0.2 Catalysts 2018, 8, x FOR PEER REVIEW 3 of 15 Catalysts2018,8,225 3of14 Scheme1.Schematicillustrationoftheformationofcatalyst.Dimensionsarenottoscale. Scheme 1. Schematic illustration of the formation of catalyst. Dimensions are not to scale. 2. Resu2l.t sRaesnudltDs ainscdu Dssisiocunssion 2.1. Cha2r.a1c. tCehriazraatcitoernizoaftiOonx iodf eO-Rxiidceh-RPidchW P/dCWC/Cat Calaytsatlyssts TheX-Trahye dXi-frfarya cdtiifofnrac(XtioRnD ()XpRaDtt) eprnatsteorfnsP dof/ CPd(/aC) ,(oax),i doxei-driec-hricPhd P0.d60W.6W0.04.4//CC ((bb)),, PPdd00.7.W7W0.30/C.3 /(Cc), (c), Pd0.8WP0.d2/0.8CW(0d.2/)C, P(dd)0,. 9PWd00.9.W1/0C.1/C(e ()ea) raeres hshoowwnni ninF Figiguurree 11.. FFiivvee ttyyppiiccaal lddififfrfaractcitoino npepaekask osf otfheth ceatcaalytasltys sts are obsaerrev oebdseartvaedb oaut tab2o4u.8t ◦2,4.480°◦, ,404°6, ◦4,66°,8 6◦8,°8, 28◦2°i nin tthhee ddiiffffrraacctotoggrarmam. T.heT hfiveefi pveeakpse caokrrsescpoornreds ptoo tnhde to Vulcan XC-72R carbon (002) crystal face, face centered cubic (fcc) metal Pd (111), (200), (220) and theVulcanXC-72Rcarbon(002)crystalface,facecenteredcubic(fcc)metalPd(111),(200),(220)and (311) crystal plane diffraction, respectively. The XRD patterns do not show any diffraction peaks (311) crystal plane diffraction, respectively. The XRD patterns do not show any diffraction peaks corresponding to W (fcc) or WO3 indicating that most of the W atoms do not exist as an individual correspondingtoW(fcc)orWO indicatingthatmostoftheWatomsdonotexistasanindividual phase, but have entered into3 the lattice of the Pd crystal. The absence of peaks for tungsten also was phase,buthaveenteredintothelatticeofthePdcrystal. Theabsenceofpeaksfortungstenalsowas found in our previous reported Pd0.7W0.3 catalyst [59] used in acid conditions. The diffraction angle foundionf o(2u2r0p) orer v(3io1u1)s crreypstoarl tpeldanPed d0i.7ffWra0c.t3iocna tpaelayksst o[5f 9th]eu sPedd inin thaec iPddcWo/nCd citaitoanlyss.tsT ihse hdigihfferra tchtiaonn thaantg le of(220)oof rth(e3 1co1r)rcesrpyostnadlinpgla Pnde/Cd icfaftraalcytsito. nThpee XaRkDs opfetahkes sPhdifti ntot ha ehiPgdheWr a/nCglcea itnadliycsattsinigs choimghpererstshioann inth at ofthectohrer edsirpeocntiodnin pgerPpde/ndCiccualtaar ltyos tt.heT hteensXilRe Dstrpeessa k[6s0s,6h1i]f.t Ttohea shizieg hoef rcaatnaglylset imndetiacla tpianrgticcloems cparne sbsei on inthedeisrteimctaiotend pweirtph eSncdheicrruelra’sr etoqutahteiotne n[6s2i]l.e Tshtere esssti[m60at,e6d1 ]p.aTrhtiecles isziezeo offc Padta/Cly, sPtdm0.6eWta0.l4/pCa, rPtdic0.l7eWs0c.3/aCn, be estimatPedd0w.8Wit0h.2/SCc haenrdre Pr’ds0.e9Wqu0.a1/tCio nw[e6r2e ].5.T6h nemes, t4im.8 antemd, p4a.5r tnicmle, s4i.z3e nomfP, da/ndC ,5P.2d nmW, res/pCec,tPivdely.W The/ C, 0.6 0.4 0.7 0.3 particle size of the oxide-rich Pd–W/C nanoparticles is smaller than that of Pd/C. Pd W /C and Pd W /C were 5.6 nm, 4.8 nm, 4.5 nm, 4.3 nm, and 5.2 nm, respectively. The 0.8 0.2 0.9 0.1 particlesizeoftheoxide-richPd–W/CnanoparticlesissmallerthanthatofPd/C. Themorphologyandparticledistribution(Figure2)ofPd/C(a,b)andoxide-richPd W /C 0.8 0.2 (c, d) were characterized by transmission electron microscope (TEM). The oxide-rich Pd W 0.8 0.2 nanoparticlesaremoreuniformlydispersedonthecarbonsurfacethanPd. Theaveragediameterof Pdnanoparticlesis5.6nmwhiletheaveragemetalparticlediameterofoxide-richPd W is4.3nm 0.8 0.2 whichareconsistentwiththeXRDresults. Figure3a–cshowshighresolutiontransmissionelectron microscopy(HRTEM)ofoxide-richPd W /Ccatalyst. ThelatticespacinginFigure3a–cis0.224nm, 0.8 0.2 0.193nmand0.263nmwhichrespectivelycorrespondtothe(111),(200)crystalplanesofface-centered cubic Pd and (220) plane of WO . The lattice fringes of WO can be found in a few nanoparticles, 3 3 whichsupportstheexistenceofWO inthePd W /Ccatalysts. AlthoughtheWO phaseisfound x 0.8 0.2 3 inafewnanoparticles,thereisnoWphaseorWO phaseinmostofthePd–Walloynanoparticles, 3 whichi sconsistentwiththeXRDpatterns. ItindicatesthatmostoftheWatomshavebeenmixed withthePdatomsduringthepreparationofPd–Walloynanoparticlesfromtheuniformmixturesof PdsaltandWsalt,thustheseparateWorWO phaseisrare. Besides,theWatomsonthesurfaceof 3 Pd–Wnanoparticlesareeasilyoxidizedbyambientair,thereforeitisdifficulttofindtheWphaseon Catalysts2018,8,225 4of14 thesurfaceofPd–Wnanoparticles. AlthoughthereisnodiffractionpeakcorrespondingtoWinthe XRDpatternsmentionedabove,theenergydispersivespectrum(EDS)oftheaspreparedoxide-rich Catalysts 2018, 8, x FOR PEER REVIEW 4 of 15 Pd W /C(Figure3d)showsthecontentofWinthePd–Wnanoalloys. 0.8 0.2 Figure 1. X-ray diffraction (XRD) patterns of Pd/C (a), oxide-rich Pd W /C (b), oxide-rich 0.6 0.4 CatalPysdts0 .270WF18i0g, .u83,/r xeC F1O(.c R)X, Po-ErxaEiyRd eRdrEifiVfcrhIaEcWPtido 0n . 8W(XR0.D2/) Cp(adtt)e,ranns doof xPidde/Cr ic(ha),P do0x.i9dWe-0r.i1c/h CP(de0).6.W0.4/C (b), oxide-rich 5 of 15 Pd0.7W0.3/C (c), oxide rich Pd0.8W0.2/C (d), and oxide rich Pd0.9W0.1/C (e). The morphology and particle distribution (Figure 2) of Pd/C (a, b) and oxide-rich Pd0.8W0.2/C (c, d) were characterized by transmission electron microscope (TEM). The oxide-rich Pd0.8W0.2 nanoparticles are more uniformly dispersed on the carbon surface than Pd. The average diameter of Pd nanoparticles is 5.6 nm while the average metal particle diameter of oxide-rich Pd0.8W0.2 is 4.3 nm which are consistent with the XRD results. Figure 3a–c shows high resolution transmission electron microscopy (HRTEM) of oxide-rich Pd0.8W0.2/C catalyst. The lattice spacing in Figure 3a–c is 0.224 nm, 0.193 nm and 0.263 nm which respectively correspond to the (111), (200) crystal planes of face-centered cubic Pd and (220) plane of WO3. The lattice fringes of WO3 can be found in a few nanoparticles, which supports the existence of WOx in the Pd0.8W0.2/C catalysts. Although the WO3 phase is found in a few nanoparticles, there is no W phase or WO3 phase in most of the Pd–W alloy nanoparticles, which is consistent with the XRD patterns. It indicates that most of the W atoms have been mixed with the Pd atoms during the preparation of Pd–W alloy nanoparticles from the uniform mixtures of Pd salt and W salt, thus the separate W or WO3 phase is rare. Besides, the W atoms on the surface of Pd–W nanoparticles are easily oxidized by ambient air, therefore it is difficult to find the W phase on the surface of Pd–W nanoparticles. Although there is no diffraction peak corresponding to W in the XRD patterns mentioned above, the energy dispersive spectrum (EDS) of the as prepared oxide-rich Pd0.8W0.2/C (Figure 3d) shows the content of W in the Pd–W nanoalloys. Figure2.ThemorphologyandparticledistributionofPd/C(a,b)andoxide-richPd W /C(c,d). Figure 2. The morphology and particle distribution of Pd/C (a,b) and oxide-rich Pd0.8W0.80.2/C0.2 (c,d). Figure 3 High resolution transmission electron microscopy (HRTEM) (a–c) images and energy dispersive spectroscopy (EDS) spectra (d) of the oxide-rich Pd0.8W0.2/C catalyst. Figure 4 shows the X-ray photoelectron spectroscopy (XPS) spectra of oxide-rich Pd0.6W0.4/C, Pd0.7W0.3/C, Pd0.8W0.2/C, Pd0.9W0.1/C. The XPS spectra of Pd/C was published in our recent works [63]. All XPS curves were fitted using the Gaussian–Lorentzian (20%) method after subtracting the background with Shirley’s method. The surface composition ratios of the Pd:W elements in Catalysts 2018, 8, x FOR PEER REVIEW 5 of 15 Catalysts2018,8,225 5of14 Figure 2. The morphology and particle distribution of Pd/C (a,b) and oxide-rich Pd0.8W0.2/C (c,d). Figure 3 High resolution transmission electron microscopy (HRTEM) (a–c) images and energy Figure 3. High resolution transmission electron microscopy (HRTEM) (a–c) images and energy dispersive spectroscopy (EDS) spectra (d) of the oxide-rich Pd0.8W0.2/C catalyst. dispersivespectroscopy(EDS)spectra(d)oftheoxide-richPd W /Ccatalyst. 0.8 0.2 Figure 4 shows the X-ray photoelectron spectroscopy (XPS) spectra of oxide-rich Pd0.6W0.4/C, Pd0.7WFig0.3u/Cre, 4Pdsh0.o8Ww0s.2/tChe, PXd-r0a.9yWp0.h1/oCt.o eTlheec trXoPnSs pspeecctrtoras coofp yPd(X/CP Sw)assp epcutrbalioshfeodx idine -oriucrh rPedce0.n6Wt w0.4o/rkCs, P[6d30]..7 WAl0l .3X/PCS, cPudrv0.e8Ws w0.2e/reC f,itPtedd0. 9uWsi0n.1g/ tCh.e TGhaeusXsPiaSn–sLpoecretrnatzoiafnP (d2/0C%)w maesthpoudb laisftheerd suinbtroaucrtinregc ethnet wbaocrkkgsr[o6u3n].dA lwlXitPhS Schuirrvleeys’ws emreefithttoedd. uTsihneg sthuerfGacaeu scsoiamnp–Loosirteionntz iaranti(o2s0 %o)f mtheteh oPdda:Wfte resleumbternactsti nign the background with Shirley’s method. The surface composition ratios of the Pd:W elements in oxide-rich Pd W /C, Pd W /C, Pd W /C and Pd W /C are Pd W , Pd W , 0.6 0.4 0.7 0.3 0.8 0.2 0.9 0.1 0.57 0.43 0.70 0.30 Pd W ,Pd W ,respectively. InFigure4a,thepeaksofPd3d andPd3d correspond 0.79 0.21 0.87 0.23 5/2 3/2 toPdandPdO (0<2<y),andthePdelementispresentinallthesamplesasPdmetalandPdO . y y The binding energy of Pd 3d peaks of PdW/C catalysts respectively shifted +0.21 eV, +0.28 eV, 5/2 +0.36eV,+0.52eVcomparedwiththatofPd/C(335.6eV,thesolidline)whichindicatesadecrease ofPd3delectronicclouddensities. Thechangeoftheelectronicstructureisduetotheformationof high-valencyoxides.Figure4bisthepeakofW .Thetwopeaksat35.5and37.5eVcorrespondtoWO . 4f 3 WiththedecreaseoftungstencontentinthePd–Wnanoalloy,theatomratioofW(0):W(VI)increases. The W(0):W(VI) in Pd W /C, Pd W /C, Pd W /C and Pd W /C is W(0) :W(VI) , 0.9 0.1 0.8 0.2 0.7 0.3 0.8 0.2 0.40 0.60 W(0) :W(VI) ,W(0) :W(VI) ,W(0) :W(VI) ,respectively. Theoxidesplayakeyroleinthe 0.42 0.58 0.48 0.52 0.55 0.45 oxide-richPd–W/Celectrocatalysts. ItwasfoundthatthePdOexhibitshigherelectrocatalyticactivity andstabilityforORRthanPdinalkalinesolutions[64]. WO nanoarraysupportedoncarboncloth 3 hasbeenregardedasastate-of-artcatalystfortheoxygenevolutionreaction(OER)[65]operating inalkalinemedia. Zhangetal.[66]reportedthatPdsupportedonWO /CexhibitsahigherORR 3 activityinacidsolutionthanPd/C.ThestronginteractionbetweenPdandWO effectivelypromotes 3 the direct 4-electron pathway of the ORR at Pd. In this work, the performance of the as prepared oxide-richPd–W/CcatalystisenhancedbyboththehighORRactivityofPdOandthesynergistic effectofW/WO . 3 Catalysts 2018, 8, x FOR PEER REVIEW 6 of 15 oxide-rich Pd0.6W0.4/C, Pd0.7W0.3/C, Pd0.8W0.2/C and Pd0.9W0.1/C are Pd0.57W0.43, Pd0.70W0.30, Pd0.79W0.21, Pd0.87W0.23, respectively. In Figure 4a, the peaks of Pd 3d5/2 and Pd 3d3/2 correspond to Pd and PdOy (0 < 2 < y), and the Pd element is present in all the samples as Pd metal and PdOy. The binding energy of Pd 3d5/2 peaks of PdW/C catalysts respectively shifted +0.21 eV, +0.28 eV, +0.36 eV, +0.52 eV compared with that of Pd/C (335.6 eV, the solid line) which indicates a decrease of Pd 3d electronic cloud densities. The change of the electronic structure is due to the formation of high-valency oxides. Figure 4b is the peak of W4f. The two peaks at 35.5 and 37.5 eV correspond to WO3. With the decrease of tungsten content in the Pd–W nanoalloy, the atom ratio of W(0):W(VI) increases. The W(0):W(VI) in Pd0.9W0.1/C, Pd0.8W0.2/C, Pd0.7W0.3/C and Pd0.8W0.2/C is W(0)0.40:W(VI)0.60, W(0)0.42:W(VI)0.58, W(0)0.48:W(VI)0.52, W(0)0.55:W(VI)0.45, respectively. The oxides play a key role in the oxide-rich Pd–W/C electrocatalysts. It was found that the PdO exhibits higher electrocatalytic activity and stability for ORR than Pd in alkaline solutions [64]. WO3 nanoarray supported on carbon cloth has been regarded as a state-of-art catalyst for the oxygen evolution reaction (OER) [65] operating in alkaline media. Zhang et.al. [66] reported that Pd supported on WO3/C exhibits a higher ORR activity in acid solution than Pd/C. The strong interaction between Pd and WO3 effectively promotes the direct 4-electron pathway of the ORR at Pd. In this work, the performance of the as Catalypsrtsep20a1r8e,d8 ,o2x25ide-rich Pd–W/C catalyst is enhanced by both the high ORR activity of PdO and th6e of14 synergistic effect of W/WO3. Figure 4. X-ray photoelectron spectroscopy (XPS) spectra of (a) Pd 3d and (b) W 4f) of oxide-rich Figure4. X-rayphotoelectronspectroscopy(XPS)spectraof(a)Pd3dand(b)W4f)ofoxide-rich PdW/C catalysts. PdW/Ccatalysts. 2.2. Electrochemical Performance 2.2. ElectrochemicalPerformance Figure 5 shows the cyclic voltammograms (CV) of Pd/C (a), oxide-rich Pd0.6W0.4/C (b), PFdi0g.7uWr0e.3/C5(cs)h, oPwd0s.8Wth0.e2/Cc y(dc)l,i cPdv0o.9Wlta0.m1/Cm (oe)g; raallm thse (CCVVs) woefreP md/eaCsu(rae)d, ion x1i dMe -NriacOhHP sdo0l.u6Wtio0n.4 a/tC a (b), Pd scWan ra/tCe (ocf) 1,0P dmV·Ws−1. A/llC th(ed p),oPtedntiaWls in /thCis( pe)a;paelrl athree qCuVotsedw weriethm reesapseucrte tdo rinev1erMsibNlea hOyHdrosgoelun tion 0.7 0.3 0.8 0.2 0.9 0.1 at aeslceactnrordaet e(RoHfE1)0. AmlVl p·so−te1n.tiAallsl vtsh. ethpeo Htegn/tHiaglOs ienletchtrisodpea wpeerrea croenqvuerotetedd tow viathlurees srpeefecrtritnog rteov tehres ible hydr ogenelectrode(RHE).Allpotentialsvs. theHg/HgOelectrodewereconvertedtovaluesreferring totheRHEusingthefollowingequation: E(RHE)=E(Hg/HgO)+0.098V+0.0591×pH.Thepeak ofhydrogenadsorption/desorptionisatabout0.23V.ThepeakofOH− adsorbedonthesurfaceof theelectrocatalystisintherangefrom0.33Vto0.53V.Theoxidationofthesurfacemetalandthe resultingreductionoftheoxideareintherangeof0.33Vto1.13V.Thepeakat~0.7Vvs. RHEchanged dramaticallydependingontheratioofPdandW.Thereareseveralfactorsaffectingthecurrent. As mentionedabove,thesize/diameterofthenanoparticleschangedwiththecontentofW.Inasmaller nanoparticle,moremetalatomslocateatthesurface,whichisabenefitfortheelectrochemicalreaction. Besides, the peak at ~0.7 V vs. RHE corresponds to the reduction of PdO at the surface of Pd–W bimetallic nanoparticles. The surface Pd content in the oxide-rich Pd–W nanoalloys changes with the Pd:W ratios. Furthermore, the peak current could also be affected by other known/unknown physicalpropertiesofthebimetallicnanoparticles. Forexample,asmentionedabove,theadditionof WchangedtheelectronstructureofPdatoms. Theelectrochemicalactivesurfacearea(EASA)can becalculatedbytheamountofchargecorrespondingtothehydrogenadsorption/desorptionregion inthecyclicvoltammetriccharacteristiccurve. However, Pdbasedcatalystssupportedoncarbon have poor clarity for hydrogen because hydrogen can penetrate into the Pd-based alloy structure. SothereductionchargeofPdOwaschosentocalculatetheEASA.TheEASAvalueofthecatalystwas calculatedbythefollowingequation: EASA=Q/QMR Catalysts 2018, 8, x FOR PEER REVIEW 7 of 15 RHE using the following equation: E(RHE) = E(Hg/HgO) + 0.098 V + 0.0591 × pH. The peak of hydrogen adsorption/desorption is at about 0.23 V. The peak of OH− adsorbed on the surface of the electrocatalyst is in the range from 0.33 V to 0.53 V. The oxidation of the surface metal and the resulting reduction of the oxide are in the range of 0.33 V to 1.13 V. The peak at ~0.7 V vs. RHE changed dramatically depending on the ratio of Pd and W. There are several factors affecting the current. As mentioned above, the size/diameter of the nanoparticles changed with the content of W. In a smaller nanoparticle, more metal atoms locate at the surface, which is a benefit for the electrochemical reaction. Besides, the peak at ~0.7 V vs. RHE corresponds to the reduction of PdO at the surface of Pd–W bimetallic nanoparticles. The surface Pd content in the oxide-rich Pd–W nanoalloys changes with the Pd:W ratios. Furthermore, the peak current could also be affected by other known/unknown physical properties of the bimetallic nanoparticles. For example, as mentioned above, the addition of W changed the electron structure of Pd atoms. The electrochemical active surface area (EASA) can be calculated by the amount of charge corresponding to the hydrogen adsorption/desorption region in the cyclic voltammetric characteristic curve. However, Pd based catalysts supported on carbon have poor clarity for hydrogen because hydrogen can penetrate into the Pd-based alloy structure. So the reduction charge of PdO was chosen to calculate the EASA. The EASA value of the catalyst was calculated by Catalytshtse2 f0o1l8l,o8w,2i2n5g equation: 7of14 EASA = Q/QMR whereQisthecoulombsofthereductionofpalladiumoxideoverarangeof0.47Vto0.87V,andQMR where Q is the coulombs of the reduction of palladium oxide over a range of 0.47 V to 0.87 V, and is the charge required to reduce a monolayer of PdO. The single-layer palladium oxide reduction QMR is the charge required to reduce a monolayer of PdO. The single-layer palladium oxide charrgeedwucatsio4n.0 5chCa/rgme2 .wTahse 4E.A05S AC/omf2P. dT/hCe (Ea)A,SoAxi doef- rPicdh/CP d(0a.6)W, o0x.4i/dCe-r(ibc)h, oPxdid0.e6W-ri0c.4h/CP d(b0.)7,W o0x.i3d/eC-r(icc)h, oxidPe-dr0i.c7WhP0.3d/C0.8 (Wc),0 .o2/xiCde(e-r)icahn dPdo0x.8iWde0-.2r/iCch (eP)d a0n.9dW o0x.1i/dCe-r(idc)h wPedr0e.9W7.80.15/Ccm (d2), 8w.7e5rec m7.82,59 c.m912,c 8m.725, 1c1m.82, c9m.921, andc1m9.29, 1cm1.82 crmes2p, eacntdiv 1e9ly.9. cm2 respectively. Figure5.Cyclicvoltammograms(CV)ofPd/C(a),oxide-richPd W /C(b),oxide-richPd W /C 0.6 0.4 0.7 0.3 (c),oFxiigdue-rrei c5h. PCdy0c.l8iWc v0o.2l/taCm(md)o,gorxaimdes- r(iCchV)P odf0 .P9Wd/C0.1 (/aC), (oex)iidne1-rMichN PadO0.6HWs0o.4/luCt i(obn),. oSxcaidner-aritceh1 P0dm0.V7W·s0−.3/1C. (c), oxide-rich Pd0.8W0.2/C (d), oxide-rich Pd0.9W0.1/C (e) in 1 M NaOH solution. Scan rate 10 mV·s−1. Figure6displaysthelinearsweepvoltammetry(LSV)ofPd/C(a),oxide-richPd W /C(b), 0.6 0.4 Pd W F/igCur(ec )6, PddisplWays t/hCe l(idn)e,arP dsweWep v/olCtam(e)mceatrtayl y(LstSsV.) Tohf ePLd/SCV (cau),r ovxeisdwe-reirceh mPdea0.6sWur0e.4d/Ca (tba), 0.7 0.3 0.8 0.2 0.9 0.1 rotatPindg0.7Wdi0s.3k/Ce l(ec)c,t rPodd0.e8W(R0.2D/CE )(dw), iPthd0a.9Wro0.t1a/Cti n(eg) scpateaeldystosf. T2h00e 0LSr/Vm cuinrvaens dweares cmanearsautreedo fat1 a mroVta·tsi−n1g. Theedliesckt reolelyctteroidse0 .(1RMDEN) waOitHh as orlouttaitoinngs astpuereadte odf w20it0h0 Or/m.iCno amnpda ar esdcawn irthatPe do/f C1 ,mthVe·so−n1.s Tethpe oelteecnttrioallyatet 2 theo xide-richPd W /Cshiftedtowardahigherpotential. Whichisconsistentwiththeprediction 0.8 0.2 ofthehighactivityofPd W[47]. Asmentionedabove,thePd W nanoparticlesexhibitthesmallest 3 0.8 0.2 sizeandthelargestEASA,whichisabenefitfortheirelectrocatalyticperformanceofORR. Chronoamperometry(CA)curvesareoftenusedtoevaluatethestabilityofelectrocatalysts[67,68]. ElectrocatalyticstabilityofPd/Candoxide-richPd W CcatalystswerecharacterizedbyCAat 0.8 0.2 0.57 V vs. RHE in 0.1 M NaOH solution (Figure 7). At the beginning the CA current decreased rapidly, then the current density of each catalyst became relatively stable. The rapid decrease in theinitial1or2smaybethechargingcurrentattheworkingelectrode(WE).Atthebeginningof theCAmeasurements,thepotentialattheWErapidlyshiftedtothesetpotential(0.57Vvs. RHE). TherapidshiftofpotentialcausedachargingcurrentattheWE.Althoughtheoxide-richPd W /C 0.8 0.2 catalystexhibitshigherelectrocatalyticstabilitythanPd/C,itisstilldifficulttodrawaconclusionof whetherthecompositionofthecatalystisunchangedduringtheORRmeasurements. Catalystswith constantcompositionssuchaspurePd[69]orpurePt[70]sometimesexhibitpoorelectrocatalytic stability. Therefore the high electrocatalytic stability does not mean constant composition. It can beseenfromFigure6thatPd W /Calsoexhibitshighactivity. Thatmeansevenifahalfofthe 0.9 0.1 W/WO de-alloyedfromthePd W nanoalloys,thePd–Wcatalystsstillmaintainhighactivity. The 3 0.8 0.2 currentdensity(mA·mgPd−1)attheoxide-richPd W /Cismorethan1.6timesofthatatPd/C. 0.8 0.2 Catalysts 2018, 8, x FOR PEER REVIEW 8 of 15 is 0.1 M NaOH solution saturated with O2. Compared with Pd/C, the onset potential at the oxide-rich Pd0.8W0.2/C shifted toward a higher potential. Which is consistent with the prediction of the high activity of Pd3W [47]. As mentioned above, the Pd0.8W0.2 nanoparticles exhibit the smallest Catalysts2018,8,225 8of14 size and the largest EASA, which is a benefit for their electrocatalytic performance of ORR. Figure 6. Linear sweep voltammetry (LSV) of Pd/C (a), oxide-rich Pd W /C (b), oxide-rich Catalysts 2018, 8, x FOR PEER REVIEW 0.6 0.4 9 of 15 Pd W /C (c), oxide-rich Pd W /C (d), oxide-rich Pd W /C (e) in 0.1 M NaOH solution 0.7 0.3 0.8 0.2 0.9 0.1 saturatedwithO .Rotatingspeed2000r/min.Scanrate1mV·s−1. 2 Figure 6. Linear sweep voltammetry (LSV) of Pd/C (a), oxide-rich Pd0.6W0.4/C (b), oxide-rich Pd0.7W0.3/C (c), oxide-rich Pd0.8W0.2/C (d), oxide-rich Pd0.9W0.1/C (e) in 0.1 M NaOH solution saturated with O2. Rotating speed 2000 r/min. Scan rate 1 mV·s−1. Chronoamperometry (CA) curves are often used to evaluate the stability of electrocatalysts [67,68]. Electrocatalytic stability of Pd/C and oxide-rich Pd0.8W0.2C catalysts were characterized by CA at 0.57 V vs. RHE in 0.1 M NaOH solution (Figure 7). At the beginning the CA current decreased rapidly, then the current density of each catalyst became relatively stable. The rapid decrease in the initial 1 or 2 s may be the charging current at the working electrode (WE). At the beginning of the CA measurements, the potential at the WE rapidly shifted to the set potential (0.57 V vs. RHE). The rapid shift of potential caused a charging current at the WE. Although the oxide-rich Pd0.8W0.2/C catalyst exhibits higher electrocatalytic stability than Pd/C, it is still difficult to draw a conclusion of whether the composition of the catalyst is unchanged during the ORR measurements. Catalysts with constant compositions such as pure Pd [69] or pure Pt [70] sometimes exhibit poor electrocatalytic stability. Therefore the high electrocatalytic stability does not mean constant composition. It can be seen from Figure 6 that Pd0.9W0.1/C also exhibits high activity. That means even if a half of the W/WO3 de-alloyed from the Pd0.8W0.2 nanoalloys, the Pd–W catalysts still maintain high activity. The current density (mA·mg Pd−1) at the oxide-rich Pd0.8W0.2/C is more than 1.6 times of that at Pd/C. Figure7. ElectrocatalyticstabilityofPd/C(a)andoxide-richPd W /C(b)in0.1mol/LNaOH Figure 7. Electrocatalytic stability of Pd/C (a) and oxide-rich P0.d80.8W0.20.2/C (b) in 0.1 mol/L NaOH solutionsaturatedwithO .Potential0.57Vvs.RHE.Rotatingspeed2000r/min. solution saturated with2 O2. Potential 0.57 V vs. RHE. Rotating speed 2000 r/min. FiguFrigeu8res h8o swhsowths etheele cetlreoctcrhoecmheimcailcailm ipmepdeadnacneces psepcetrcotrsocsocpoypy( E(EISI)S)o offP Pdd//CC aanndd ooxxiiddee--rriicchh Pd0.P8dW0.80W.2/0.C2/Cc actaatlaylystssts..E EISISm meeaassuurreemmeennttss wweerree ppeerrffoorrmmeedd ffrroomm 11 HHzz ttoo 110000 kkHHzz wwiitthh 55 mmVV ssiiggnnaallss. . TheTsheem isceimrcilceiracrlce raardci ursadiniutsh einh itghhe -fhriegqhu-efrnecqyureengciyo nrergeiporne sreenptrsetsheentcsh athrge ecthraarngsef etrrarenssifsetra nrecseis(tRaCncTe) . Itca(nRCbeTs).e eItn ctahna bteth seeecnh atrhgaet ttrhaen cshfearrgrees tirsatannsfceer orfesthisetaonxcied eo-fr tihche Poxdi0d.8eW-ri0c.2h/ PCdi0s.8Wsim0.2i/lCa rist ositmhailtaor ftoP dth/aCt . oFf iPgdu/rCe. 9showstheKoutecky–Levichplotsofoxide-richPd W /Ccatalyst, thetransferred 0.8 0.2 electronFniugmurbee 9rs sohfoOws itnhOe RKRouwteecrkeyd–eLteevrimchin peldotbsy otfh eoxKidoeu-treicchk yP–dL0e.8vWic0h.2/Ceq cuaattailoyns:t, the transferred 2 electron numbers of O2 in ORR were determined by the Koutecky–Levich equation: BB=00..22nnFF(cid:0)D(OD2(cid:1)O2/23)ν2−/31/6C1O/26CO2 where B is the reciprocal of the slope of the Koutecky–Levich plots, n is the number of electrons transferred, F is the Faraday constant, D is the diffusion coefficient of O2 in 0.1 M NaOH, ν is the viscosity of the electrolyte, and C is the saturated concentration of O2 in 0.1 M NaOH [71]. Based on the Koutecky–Levich plots and the Koutecky–Levich equation, the calculated number of electrons transferred is 3.8. Thus there is mainly a 4-electron mechanism for the ORR at the oxide-rich Pd0.8W0.2/C catalyst. Catalysts2018,8,225 9of14 where B is the reciprocal of the slope of the Koutecky–Levich plots, n is the number of electrons transferred, F is the Faraday constant, D is the diffusion coefficient of O in 0.1 M NaOH, ν is 2 the viscosity of the electrolyte, and C is the saturated concentration of O in 0.1 M NaOH [71]. 2 BasCeadtaloynsts t2h01e8,K 8,o xu FtOeRck PyEE–RL eRvEVicIEhWp l otsandtheKoutecky–Levichequation,thecalculatednu10m obf e1r5 of electroCnatsaltyrstas n20s1f8e,r 8r,e xd FOisR 3P.E8E.RT RhEuVsIEtWh e reismainlya4-electronmechanismfortheORRattheox10i doef -1r5ic h Pd W /Ccatalyst. 0.8 0.2 FiguFirgFeui8gr.ueE r8eI. S 8E.oI ESf IoPSf d oP/fd CP/dCa/ Cnand adno dox xiodixdeie-dr-eri-cicrhihc PhPd dP00d..880WW.8W00.20./2.2C//C Cc acctaaatlatyalysltyssts.s t.s . FigFuirgFeui9gr.eu Kr9e.o 9Ku. otKeucotkeucytek–cyLk–eyLv–eiLvcehivchpic lhpo ltposltoso tfos foo oxf xiodixdeied-r-eri-circhihcP hP ddP0d0..880W.W8W00.20./2.2C//C Cc accataataltylaysltsy.t s. t. 3. M3. aMteartiearlisa alsn adn Md Methetohdosd s 3.1. Preparation and Characterization of the Catalysts 3.1. Preparation and Characterization of the Catalysts PdPCdl2C lw2 aws aps upruchrcahseadse dfr ofrmom S iSnionpohpahramrm C Chehmemiciacla lR Reaegaegnent tC Coo.,. ,L Ltdtd. . (S(Shhaanngghhaai,i , CChhiinnaa)).. TThhee Vulcan carbon powder XC-72R was obtained from Cabot Corporation (Cabot Corp., Billerica, MA, Vulcan carbon powder XC-72R was obtained from Cabot Corporation (Cabot Corp., Billerica, MA, USA). Nafion solution (5%) was obtained from DuPont (Delaware, DE, USA). All other chemicals USA). Nafion solution (5%) was obtained from DuPont (Delaware, DE, USA). All other chemicals Catalysts2018,8,225 10of14 3. MaterialsandMethods 3.1. PreparationandCharacterizationoftheCatalysts PdCl waspurchasedfromSinopharmChemicalReagentCo.,Ltd. (Shanghai,China). TheVulcan 2 carbon powder XC-72R was obtained from Cabot Corporation (Cabot Corp., Billerica, MA, USA). Nafion solution (5%) was obtained from DuPont (Delaware, DE, USA). All other chemicals were of analytical grade and used as acquired. Triple-distilled water was used through-out. The WO 3 containingPd–Wcatalystswerepreparedwiththereduction–oxidationprocedures,schematically illustratedinScheme1. Pd/CandPdW/Ccatalystswithametalloadingof20wt%werepreparedbytheNaBH chemical 4 reductionmethod(Scheme1,step1)thatweusedbefore[72]. PdCl andNa WO wereusedasthe 2 2 4 precursors. Electrocatalystswithdifferentatomicratioswerecontrolledbythemolarratioofmetal precursors. ThePd–WnanoalloysareeasilyoxidizedinambientairandformedtheWO -contained 3 Pd–W/Ccatalysts(Scheme1,step2). TheXRDwascarriedoutbyaBrukerD8advanceX-raydiffractometer(BRUKERAXSGMBH, Karlsruhe,Germany)operatingat40keVand30mAwithCuKαradiationsource,λ=0.15406nm. The TEM/HRTEM images were obtained on a JEOL JEM-2100 transmission electron microscope (JEOL,Tokyo,Japan). Thecontentofmetalelementsonthesurfaceofthesampleswasanalyzedby EDS.ThepresenceofthemetalwasexcitedbyX-rayphotoelectronspectroscopy(XPS)usingAlKα X-rayradiationonanESCALAB250(ThermoFisherSCIENTIFIC,Waltham,MA,USA)spectrometer. PeakfittingusingGaussian/Lorentzian(20%Gaussian)methodafterbackgroundsubtractionusing Shirley’smethod[73]. 3.2. ElectrochemicalMeasurements TheelectrochemicalmeasurementswereperformedwithCHI832BandCHI660Eelectrochemical workstations(CHIInstruments,Austin,TX,USA)andaconventionalthree-electrodeelectrochemical cell. Acarbon-rodwasusedastheauxiliaryelectrode. AHg/HgOelectrodewasusedasthereference electrode. The working electrode was prepared by the following procedure: The glassy carbon electrode(GCE,3mmindiameter,LANLIKE,Tianjin,China)wascarefullypolishedwith0.05µm alumina(Al O )powder,andwashedwithtriple-distilledwaterbeforeuse. Tenmgofthecatalyst 2 3 powderinamixtureof0.5mLwaterand0.5mLethanolwasultrasonicatedfor15mintopreparethe inkforcatalysts. TwentyµL(2µL×10times)oftheinkwasdroppedontheGCE.ThreeµLofNafion solution(5wt.%)wasdroppedonthesurfaceaftertheinkwasdriedinair. 4. Conclusions The WO containing oxide-rich Pd–W/C catalysts were successful fabricated by reduction– 3 oxidationprocedures. Theaspreparedoxide-richPd W /Ccatalystsexhibithighelectrocatalytic 0.8 0.2 activityandstability. Thisdemonstratesthattheaspreparedoxide-richPd W /Cisaprospective 0.8 0.2 candidateforthecathodeoffuelcellsoperatingwithalkalineelectrolyte. AuthorContributions:ThecorrespondingauthorW.L.isthedirectoroftheotherauthors.W.L.andN.C.designed theexperiments.N.C.andZ.G.carriedouttheexperiments;X.X.andH.Z.analyzedtheexperimentalresults. Acknowledgments: This work was supported by the Natural Science Foundation of Shandong Province (ZR2016BM31). We are greatly appreciate the contributions of Guang Dong, Xin Han, Haoquan Zhang, ShuzhengXu,PeipeiYuandMingchenQinfortheirkindlyrepeatedtheexperimentstomakesurethedatais repeatable,andproofreadingthespellings. ConflictsofInterest:Theauthorsdeclarenoconflictofinterest.