Size-Dependent Activity Trends Combined with in Situ X‑rayAbsorption Spectroscopy Reveal Insights into Cobalt Oxide/CarbonNanotube-Catalyzed Bifunctional Oxygen ElectrocatalysisBora Seo,†,∥ Young Jin Sa,†,∥ Jinwoo Woo,‡ Kyungjung Kwon,§ Jongnam Park,‡ Tae Joo Shin,⊥

Hu Young Jeong,*,⊥ and Sang Hoon Joo*,†,‡

†Department of Chemistry and ‡School of Energy and Chemical Engineering, Ulsan National Institute of Science and Technology(UNIST), 50 UNIST-gil, Ulsan 44919, Republic of Korea§Department of Energy & Mineral Resources Engineering, Sejong University, 209 Neungdong-ro, Seoul 05006, Republic of Korea⊥UNIST Central Research Facilities (UCRF), Ulsan National Institute of Science and Technology (UNIST), 50 UNIST-gil, Ulsan44919, Republic of Korea

*S Supporting Information

ABSTRACT: Bifunctional oxygen electrocatalysts play a vitalrole in important energy conversion and storage devices. Cost-effective, abundant, and active Co-based materials haveemerged as promising bifunctional electrocatalysts for whichidentifying catalytically active structures under reactionconditions and unraveling the structure−activity relationshipsare of critical importance. Here, we report the size-dependent(3−10 nm) structure and catalytic activity of bifunctionalcobalt oxide nanoparticle (CoOx NP) catalysts for the oxygenevolution reaction (OER) and the oxygen reduction reaction(ORR). In situ X-ray absorption spectroscopy (XAS) revealedthat the majority of NPs during OER and ORR werecomposed of the Co3O4 and CoOOH phases regardless oftheir particle sizes. The OER activity increased with decreasing NP size, which correlated to the increased oxidation state andlarger surface area in smaller NPs, whereas the ORR activity was nearly independent of NP size. These particle size-dependentcatalytic activities in conjunction with the in situ XAS results can provide insights into the CoOx-catalyzed bifunctional oxygenelectrode reactions.

KEYWORDS: cobalt oxide, size effect, bifunctional catalysis, oxygen evolution reaction, oxygen reduction reaction

With increasing demand for clean energy technologies,renewable energy conversion and storage systems have

generated tremendous interest.1,2 Bifunctional oxygen electro-catalysts that can catalyze both oxygen evolution reaction(OER) and oxygen reduction reaction (ORR) are ubiquitousand of pivotal importance in energy devices, such as unitizedregenerative fuel cells and metal−air batteries.3−12 The viabilityof these energy devices is critically dependent on the catalyticperformances of bifunctional oxygen electrocatalysts. BothOER and ORR involve the transfer of four electrons, renderingthese reactions energetically demanding and sluggish. As such,noble metal-based materials like IrO2, RuO2, and Pt with fastreaction kinetics have been used prevalently as bifunctionaloxygen electrocatalysts; however, they are costly and scarce.13,14

In this context, cost-effective and abundant transition metaloxides and hydroxides have emerged as a promising class ofcatalysts.15−24 Cobalt oxide-based bifunctional electrocatalystshave been of particular interest as economically viable andefficient bifunctional oxygen electrocatalysts.16−19,22−24

For the design of advanced cobalt oxide-based electro-catalysts, unraveling the nature of the active species andreaction mechanism is of critical importance, and under-standing the nanoscale particle size effects can provideimportant clues. The particle size effects in cobalt (oxide)-based catalysts have been established in a number of importantreactions, including Fischer−Tropsch synthesis and CO2

hydrogenation.25,26 However, such insights have not yet beengained for bifunctional oxygen electrocatalysis; only a fewsporadic works on the size dependency for respective OER orORR have been reported.27,28 More importantly, an under-standing of the size-dependent catalytic activity in symbiosiswith in situ spectroscopic characterization can further providemore compelling evidence to establish the property relation-

Received: February 24, 2016Revised: May 4, 2016Published: May 27, 2016

Research Article

pubs.acs.org/acscatalysis

© 2016 American Chemical Society 4347 DOI: 10.1021/acscatal.6b00553ACS Catal. 2016, 6, 4347−4355

ships between structure (size, shape, and/or composition) andcatalysis.29

Herein, we report the nanoscale size-dependent structureand catalytic activity of bifunctional electrocatalysts based oncobalt oxide nanoparticles (CoOx NPs) for both OER andORR for the first time. CoOx NPs with four different particlesizes, tunable from 3 to 10 nm, were synthesized and depositedon acid-treated carbon nanotubes (CNTs), affording CoOx/CNTs model catalysts to investigate bifunctional electro-catalysis in alkaline media. In situ X-ray absorption spectros-copy (XAS) analysis revealed that the composition of the size-controlled CoOx NPs was invariably Co3O4 and CoOOH witha small amount of Co(OH)2 under electrochemical OER andORR conditions. This result suggests that Co(III) species arethe key elements for OER, whereas they appear to be sideproducts generated from the oxidation of Co(II) by a peroxideintermediate during ORR. The CoOx/CNT catalysts exhibitedincreasing OER activity with decreasing NP size, which couldbe correlated with abundant surface Co(III) species and thelarge surface area of small CoOx NPs. For ORR, no particle sizedependence was found in the kinetic region; CoOx NPs mainlyplayed an auxiliary role, promoting the reduction ordisproportionation of peroxide generated from the two-electrontransfer pathway of the ORR.

■ RESULTS AND DISCUSSION

Preparation of the CoOx/CNTs model catalysts consisted of (i)the colloidal synthesis of CoOx NPs with different particlesizes,26 (ii) the attachment of prepared NPs to acid-treatedCNTs, and (iii) mild annealing to remove residual organicsurfactants around the CoOx NPs. Monodisperse, size-controlled CoOx NPs with average sizes of 3.0, 6.2, 7.4, and9.1 nm were obtained by controlling thermal decomposition

temperature of precursor (Table S1 and Figure S1). High-resolution TEM (HR-TEM) images (Figure S1e−h) demon-strated that all of the NPs consisted of the crystalline CoOphase (cubic, a = 4.22 Å).CoOx NPs were then anchored onto the acid-treated CNTs

with ultrasonication, followed by mild thermal annealing. Wechose undoped CNTs rather than N-doped CNTs as thesupport material because the presence of both CoOx andnitrogen synergistically enhances catalytic activity via theformation of Co−N bonding, potentially hampering theextraction of catalytic activity exclusively from CoOx.

16,30,31

The nominal content of Co in the CoOx/CNTs was around 12wt %, as confirmed by inductively coupled plasma opticalemission spectrometry (Table S1). Scanning electron micros-copy (SEM) images (Figure S2) of the CoOx/CNTs revealedthe preserved morphology after the loading of the CoOx NPs.TEM images of the CoOx/CNTs (Figure 1a−d) indicated thatthe CoOx NPs were successfully attached and uniformlydistributed on the CNTs. The average particle sizes of theCoOx NPs in the four CoOx/CNTs were 4.3, 6.3, 7.5, and 9.5nm (Figure 1i and Table S1). The size of the smallest CoOxNPs increased from that of the pristine NPs (3.0 nm) due tothe phase transformation from metallic Co to Co3O4 during theannealing step and the smaller density of Co3O4 than that ofmetallic Co phase. Hereafter, these CoOx/CNTs are denotedas CoOx(4.3)/CNTs, CoOx(6.3)/CNTs, CoOx(7.5)/CNTs,and CoOx(9.5)/CNTs. The atomic-resolution TEM (AR-TEM) images (Figure 1e−h and insets) showed that thesmallest NPs had crystalline spinel structures of Co3O4,indicating the phase change from CoO to Co3O4 after loadingonto the CNTs, whereas the other NPs remained in the sameCoO phase, as evidenced by the AR-TEM images and thecorresponding FFT patterns.

Figure 1. Characterizations of the CoOx/CNTs. (a−d) BF-TEM images of (a) CoOx(4.3)/CNTs, (b) CoOx(6.3)/CNTs, (c) CoOx(7.5)/CNTs,and (d) CoOx(9.5)/CNTs. (e−h) Corresponding AR-TEM images and FFT patterns (insets). (i) Histograms of the particle size distributions of theCoOx NPs.

ACS Catalysis Research Article

DOI: 10.1021/acscatal.6b00553ACS Catal. 2016, 6, 4347−4355

4348

X-ray diffraction (XRD) patterns (Figure 2a) of all thesamples showed a common diffraction peak at 2θ = 25.7°coincident with those that appeared for the CNTs. The

diffraction peaks at 2θ = 36.5°, 42.8°, and 61.5° werecommensurate with those of the CoO standard appeared inCoOx(6.3)/CNTs, CoOx(7.5)/CNTs, and CoOx(9.5)/CNTs.

Figure 2. (a) Wide-angle XRD patterns of the CoOx/CNTs with patterns for the CoOx standards displayed as vertical bars. The dashed gray lineindicates the peak from the CNTs at 43°. (b) Enlarged XRD pattern in the 2θ range of 41−45°. (c) Radial distribution functions (RDFs) of the CoK-edge EXAFS spectra for the CoOx/CNTs and reference materials.

Figure 3. In situ electrochemical XAS analyses at OCV and 1.8 V (OER). (a) In situ XANES spectra of the CoOx/CNTs under OCV and an appliedpotential of 1.8 V (OER) compared with ex situ XANES spectra. (b) RDFs of the CoOx(4.3)/CNTs at OCV and 1.8 V (OER). Inset: enlargedRDFs for the first Co−O interatomic distance. (c) AR-TEM images and corresponding FFT patterns of the CoOx(9.5)/CNTs before and after theOER durability test, showing the transition of the phase from CoO to Co3O4.

Table 1. Phase Composition and Average Co Oxidation State of the CoOx NPs in the CoOx/CNTs As Determined by LinearCombination Fitting of the in Situ XANES at OCV, 1.8 V, and 0.6 V (vs RHE)

sample condition Co3O4 CoOOH CoO Co(OH)2 average Co oxidation state

CoOx(4.3)/CNTs OCV 0.61 0.20 0.00 0.19 2.61.8 V 0.55 0.42 0.01 0.02 2.80.6 V 0.26 0.56 0.00 0.18 2.8

CoOx(6.3)/CNTs OCV 0.32 0.15 0.18 0.35 2.41.8 V 0.54 0.41 0.03 0.02 2.80.6 V 0.20 0.65 0.00 0.17 2.8

CoOx(7.5)/CNTs OCV 0.19 0.30 0.09 0.43 2.41.8 V 0.44 0.50 0.00 0.06 2.80.6 V 0.44 0.50 0.00 0.06 2.8

CoOx(9.5)/CNTs OCV 0.00 0.47 0.00 0.53 2.51.8 V 0.29 0.49 0.00 0.22 2.70.6 V 0.57 0.22 0.07 0.14 2.6

ACS Catalysis Research Article

DOI: 10.1021/acscatal.6b00553ACS Catal. 2016, 6, 4347−4355

4349

In the case of the CoOx(4.3)/CNTs, different diffraction peaks(2θ = 31.8°, 36.8°, 44.9°, 59.4°, and 65.2°) were observed andcould match those of the Co3O4 standard. The radialdistribution function (RDF) from Fourier transform of k3-weighted extended X-ray absorption fine structure (EXAFS)spectrum (Figure 2b) of CoOx(4.3)/CNTs exhibited majorpeaks at 1.54, 2.48, and 3.06 Å corresponding to the Co−O,Cooct−Cooct, and Cotet−Cotet/Cotet−Cooct pairs of Co3O4,respectively.22 The other three samples showed peaks at 1.75and 2.67 Å, which were similar to those of bulk CoO. Asrevealed by TEM, XRD, and EXAFS analyses, the CoOx NPs inthe CoOx(4.3)/CNTs were mainly composed of the Co3O4phase, whereas the CoOx NPs in the other three samplesconsisted of the CoO phase.The structural changes of the CoOx NPs under OER and

ORR conditions were scrutinized using in situ electrochemicalXAS with a homemade spectroelectrochemical cell. Figure 3adepicts in situ Co K-edge X-ray absorption near-edgespectroscopy (XANES) spectra of the CoOx/CNTs measuredat open circuit potential (OCV) and OER potential (1.8 V vsreversible hydrogen electrode, RHE) in 0.1 M KOH electrolyte.Under the OER potential, all the samples exhibited similarXANES spectra resembling that of Co3O4 and CoOOHregardless of the CoOx NP size. We analyzed the in situXANES spectra to understand the phases comprising the CoOxNPs under each potential with linear combination fitting (LCF)analyses (Figure S3). The resulting phase compositions andaverage Co oxidation states calculated by the relative amount ofeach phase are summarized in Table 1. We found that when theOER potential was applied, the average oxidation state of Coincreased for all the samples accompanied by the increasedamount of CoOOH phase, which is believed to be the activespecies for the OER. Under OER potential, the increment inthe amount of Co3O4 in CoOx(6.3)/CNTs, CoOx(7.3)/CNTs,and CoOx(9.5)/CNTs was also observed. For the CoOx(4.3)/CNTs, a large amount of Co3O4 was already present at OCV(61%), which was partially transformed to CoOOH, leading tothe rise in the Co oxidation state. The analysis of the in situXANES indicates that the phase transformation occurred fromCo3O4 and/or CoO to Co3O4/CoOOH mixed phase in the

CoOx/CNTs. The partial phase change from CoO to Co3O4under OER conditions was further confirmed by an HR-TEMimage of the CoOx(9.5)/CNTs after the OER (Figure 3c).The structural transformation was further observed by in situ

EXAFS spectra of the CoOx(4.3)/CNTs (Figure 3b). The RDFcurve under OER potential compared to that under OCVrevealed three major changes. First, a slight shrinkage in Co−Obond length was observed under the OER potential forCoOx(4.3)/CNTs when compared to that under OCV (insetof Figure 3b). This is an indication of an increase in theoxidation state,32 possibly from Co3O4 to CoOOH. Second, theRDF peak intensities for Co−O and Co−Co increased, whichcorrespond to the increase in the amount of adjacent di-μ-oxo-bridged CoO6 octahedra composing CoOOH

33 that have beenidentified as an active phase for OER.34,35 Finally, under OERpotential, the intensity of third major RDF peak intensity at3.04 Å decreased, and the fourth major peak at 4.69 Å wasnegatively shifted, which indicated the diminishment of Co3O4characteristic peaks as well as the evolution of CoOOH-likepeaks. From these observations, we concluded that the surfacesof the CoOx NPs may be covered with a few layers of CoOOHbecause CoOOH is a thermodynamically stable phase underoxidizing potentials.36−38 We noted that the oxidation of theCoOx NPs occurred even at the OCV, as revealed by distinctchanges between ex situ XANES spectra (solid curves in Figure3a) and in situ XANES spectra at the OCV (long dashed curvesin Figure 3a). We attributed it to the intermediate range ofOCV (0.9−1.1 V vs RHE), which is between ORR and OERpotentials. This phase transformation under nonelectrocatalyticconditions is beyond of the scope of this work and is subject todetailed investigation.When the XANES spectra were measured under the ORR

potential (0.6 V vs. RHE), all of the CoOx/CNTs samplesexhibited Co3O4/CoOOH-like XANES spectra, similar to theresults of the XANES spectra taken under the OER potential,indicating the evolution of the Co(III) species during the ORR(Figure 4a). To access more detailed information, wescrutinized the XANES data with LCF analyses. The LCFanalysis of in situ XANES at 0.6 V revealed that the CoOx NPsexist in Co3O4 and CoOOH phases with a small amount of

Figure 4. In situ electrochemical XAS analyses at OCV and 0.6 V (ORR). (a) In situ XANES spectra of the CoOx/CNTs under OCV and an appliedpotential of 0.6 V (ORR) compared with ex situ XANES spectra. (b) RDFs of the CoOx/CNTs at 0.6 V.

ACS Catalysis Research Article

DOI: 10.1021/acscatal.6b00553ACS Catal. 2016, 6, 4347−4355

4350

Co(OH)2 in all samples (Table 1). The detailed numericalanalysis revealed that with increased CoOx size the portion ofCo3O4 increased whereas that of CoOOH decreased. Thecomparison of average oxidation state indicated a slightly loweroxidation state of the CoOx(9.5)/CNTs than that of the otherCoOx/CNTs samples, perhaps because of a low surface fractionof the CoOx NPs that can participate in phase transformation(Table 1).The above XANES results under ORR potential could be

further substantiated by in situ EXAFS results, which alsoidentified Co3O4 and CoOOH phases as major species (Figure4b). In the RDFs of the in situ EXAFS spectra at 0.6 V, the firsttwo peaks, corresponding to Co−O and Co−Co interatomicdistances appeared for all of the CoOx/CNTs. The other peaksat around 3.1 and 4.5 Å were also observed, which are known tooriginate from the Co3O4 and CoOOH phases (indicated bydashed lines in Figure 4b). The similar peak intensities betweenthe samples except for the CoOx(9.5)/CNTs indicated thatthey had almost identical local structure. The differentstructural properties of the CoOx(9.5)/CNTs could beattributed to a lower portion of CoOOH and larger amountof Co3O4 comprising the CoOx(9.5)/CNTs than those of theother samples, as evidenced by LCF analysis (Table 1).When the applied potential was changed from OCV to ORR

potential (0.6 V), the average oxidation state increased from ∼+2.5 to ∼ +2.8, as revealed by the LCF analyses (Table 1 andFigure S3). The LCF analysis results indicated that Co(II) andCo(III) species were present in a ratio of around 1:1 at OCV,and Co(III) became the major species under the ORRpotential. Some previous works consistently suggested thatCo(II) on CoOx is relevant to catalytically active species for theORR.39−41 In another early work, theoretical calculationssuggested that the ORR on CoOOH is initiated over theCo(II) site followed by oxidation of the Co(II) to Co(III)species, which return to the Co(II) species for the next catalyticturnover.42 However, our in situ XAS result could notconfirmatively conclude that the Co(III) species evolved at0.6 V represents the catalytic intermediate in the ORR. Asdiscussed in the electrochemical analysis below, in the CoOx/CNTs, O2 is initially reduced to peroxide catalyzedpredominantly by the CNTs and subsequently the CoOx NPsdisproportionate to the peroxide. The peroxide intermediatelikely oxidizes the initial Co(II) species to Co(III) speciesconcomitant to the peroxide disproportionation.Next, the redox behavior of the samples was assessed via

cyclic voltammetry (CV) in N2-saturated 1 M KOH (Figure 5and Figure S5). The CV results showed anodic peaks I, II, andIII, which could be attributed to the oxidation of Co(II) toCo(III), the phase transition from Co3O4 to CoOOH, and theoxidation of Co(III) to Co(IV), respectively.41,43 Theconsecutive CV curves of the CoOx/CNTs, compiled fromup to 5 sequential scans, revealed the gradual decrease ofanodic peak II and nearly constant peak III, indicating anirreversible transition of surface Co3O4 to CoOOH and thereversible transition of Co(III) to Co(IV), respectively. Freiand co-workers reported that Co(III) species act as an initiatorfor the OER process.34 Stahl and co-workers suggested that thekey process in the OER is related to the reversibleinterconversion between Co(III) and Co(IV).44 The reversibleredox transition between Co(III) and Co(IV) in the CoOx/CNTs-catalyzed OER was also observed by the presence ofanodic peak III in consecutive CV scans (Figure S4), which isconsistent with these previous works. Significantly, in the first

CV curves, a negative shift of anodic peak II was observed withincreasing CoOx NP size (Figure 5), which could be associatedwith the relatively lower oxidation state of larger CoOx NPs.The result indicates that CoOx NPs in CoOx(4.3)/CNTs arepresent in a more oxidized form than in the other threesamples. They can serve more abundant Co(III) species,thereby facilitating the interconversion between Co(III) andCo(IV), which is considered as the active species in the OER.The CV results align well with the results from TEM, XRD, andEXAFS. However, the CV results are not well-correlated to thein situ XAS results because the in situ XAS spectra were takenafter applying potential for 1 h, which can induce considerablephase transformation from Co3O4/CoO to CoOOH/Co3O4 inthe CoOx NPs. This preconditioning could lead to the resultsthat the CoOx NPs mainly consisted of Co3O4/CoOOHregardless of their particle sizes (Table 1 in the originalmanuscript). Hence, the experimental conditions for in situ

Figure 5. Redox properties of CoOx/CNTs. (a) First CV curves over apotential range of 0.05−1.5 V (vs RHE) at a scan rate of 20 mV s−1 inN2-saturated 1 M KOH. (b) An enlarged view around anodic peak II(indicated by the dotted box in a). (c) Potentials at the anodic peak IIpositions versus particle size.

ACS Catalysis Research Article

DOI: 10.1021/acscatal.6b00553ACS Catal. 2016, 6, 4347−4355

4351

XAS could not reflect the initial oxidation state of Co of freshCoOx/CNTs. In contrast, under potentiodynamic conditions ofthe CV scans, the step-by-step transitions of oxidation states inCoOx NPs could be detected.Electrocatalytic OER and ORR activities were measured

using the rotating disk electrode (RDE) in 0.1 M KOH (Figure6 and Table S2). In the series of CoOx/CNTs, the OERactivities increased with decreasing NP size; the potentialsrequired to reach a current density of 10 mA cm−2 were 1.62,1.64, 1.65, and 1.68 V for the CoOx(4.3)/CNTs, CoOx(6.3)/CNTs, CoOx(7.5)/CNTs, and CoOx(9.5)/CNTs, respectively(Figure 6a). The greater OER activities in smaller CoOx NPswere likely attributed to the large active surface areas27 and theabundance of Co(III) that is the main source for the activeCo(IV) species, initiating the OER as confirmed by the CVanodic peak (Figure 5).36,44 The Tafel slopes were 69, 69, 73,and 80 mV dec−1 for the CoOx(4.3)/CNTs, CoOx(6.3)/CNTs,CoOx(7.5)/CNTs, and CoOx(9.5)/CNTs, respectively (Figure6b). A smaller Tafel slope was obtained for the smaller CoOxcatalyst, indicating more favorable reaction kinetics. In CoOx-based OER catalysts, high oxidation state Co(IV) centers havebeen suggested to play an essential role in catalyzing theOER.36,37,45,46 Active Co(IV) species arise from the oxidationof Co(III) with oxidizing potential, as evidenced by the anodicpeak III of the CVs (Figure S4).Unlike the case of the OER conditions, the CoOx/CNTs

demonstrated nearly identical activities for the ORRindependent of the CoOx NP size (Figure 6c). The onsetpotentials and diffusion-limited currents of the CoOx/CNTswere almost the same regardless of the NP sizes. Interestingly,the onset potential of the CoOx/CNTs was the same as thoseof the CNTs, and the Tafel plots of the CoOx/CNTs and

CNTs almost overlapped (Figure 6d). This result suggestedthat the CoOx NPs did not improve the intrinsic ORR activity;rather, CoOx appeared to promote the reduction ordisproportionation of peroxide species generated by theCNTs, as evidenced by the greater diffusion-limited currentsof the CoOx/CNTs compared to those of the CNTs. Inaddition, the similar Tafel slopes might have indicated that thereaction rates of the CoOx/CNTs and CNTs were limited bythe same step, which occurred on the CNTs.Therefore, we hypothesized that the ORR was initiated by

the CNTs via the two-electron transfer pathway and that theCoOx NPs mainly played an auxiliary role, promoting thereduction or disproportionation of the generated peroxide. Thefast and repetitive disproportionation of peroxide by CoOx ledto the quasi-four-electron pathway for the ORR.28,47 We againconfirmed the particle size independence of the ORR activitytrend with CoOx/CNTs with a higher CoOx loading (Table S3and Figure S5). The ORR activities of 32% CoOx(4.3)/CNTs,35% CoOx(6.3)/CNTs, and 36% CoOx(9.5)/CNTs werealmost the same regardless of their particle sizes, as evidencedby their overlapping LSV curves. It was also found that thehigh-loading 32% CoOx(4.3)/CNTs showed only marginalimprovement of the ORR activity in the kinetic region (i.e.,0.7−0.85 V) compared with the low-loading 9.4% CoOx(4.3)/CNTs. We found that the previously reported, high-perform-ance CoOx/carbon hybrid catalysts usually contained nitro-gen,16,23,30 potentially creating Co−N moieties, which areknown to be highly active species for the ORR. The rather lowORR performances of our CoOx/CNTs are rationalized by theabsence of nitrogen. We highlight the importance of preparingN-free metal oxides/carbon hybrid model catalysts to

Figure 6. Electrochemical analyses of the CoOx/CNTs. (a) Polarization curves and (b) Tafel plots for the OER. (c) Polarization curves and (d)Tafel plots for the ORR.

ACS Catalysis Research Article

DOI: 10.1021/acscatal.6b00553ACS Catal. 2016, 6, 4347−4355

4352

investigate the particle size-dependent ORR activity originatingfrom only metal oxides.Long-term stability is a critical factor for applying electro-

catalysts in practical applications. Chronopotentiometry (CP)was conducted at a current density of 5 mA cm−2 (Figure S6).The CoOx/CNTs catalysts exhibited excellent durability withlittle decay in the OER activity and maintenance of theirstructures over 400 min of operation. TEM images after the CPmeasurements revealed that the phase of the CoOx NPs wasmaintained as Co3O4 with retained particle sizes for theCoOx(4.3)/CNTs (Figure S7b). However, a long-term testunder an applied OER potential resulted in the structuralchange from CoO to Co3O4 for the largest CoOx NPs, asrevealed by the FFT patterns in Figure S7d, which wasconsistent with the in situ XAS measurement results. Inaddition, we investigated the chemical states of Co before andafter chronopotentiometry (CP) using X-ray photoelectronspectroscopy (XPS) (Figure S8). For the CoOx(4.3)/CNTs,the XPS spectra before and after the CP runs almostoverlapped. For the CoOx(9.5)/CNTs, satellite peaks around787 and 803 eV (indicated by dotted lines) disappear afterOER, indicating a decrease in the number of surface Co(II)species via the oxidation of Co(II) species to Co(III). Theresults support well the conclusions drawn from the othercharacterizations, including TEM, XRD, XAS, and CV curves.

■ CONCLUSIONS

In summary, we have investigated the size-dependent NPstructures and catalytic activities of the CoOx/CNTs forbifunctional oxygen electrocatalysis. In situ electrochemicalXAS measurements demonstrated that Co3O4 and CoOOHwere the major species regardless of the CoOx particle sizeunder both OER and ORR conditions. The OER activities ofthe CoOx/CNTs increased with decreasing particle size, whichcould be associated with the increased oxidation state and largersurface area of the smaller NPs. The ORR activity wasindependent of the CoOx NP size, revealing the auxiliary role ofCoOx on the CoOx/CNTs for the reduction or disproportio-nation of peroxide rather than the reduction of oxygen.Combining in situ XAS with electrocatalytic activity trends, wesuggested that the dominant Co(III) species are related toactive intermediates for the OER, whereas they appear to beside products generated from the oxidation of Co(II) by aperoxide intermediate during the ORR. This work can offer aplatform to explore the structural changes and reactionpathways of cobalt oxide for the rational design of advancedbifunctional oxygen electrocatalysts.

■ MATERIALS AND METHODS

Synthesis of Size-Controlled Cobalt Oxide Nano-particles. Cobalt oxide nanoparticles were synthesized via apreviously reported method26 with some modifications. For thesynthesis, standard Schlenk techniques were used, and allmanipulations with the cobalt carbonyl precursor wereperformed in a glovebox. First, 73 μL of oleic acid (99%,Sigma-Aldrich) in a 100 mL round-bottom flask were evacuatedfor 10 min and saturated with Ar. Subsequently, 7.5 mL ofanhydrous o-dichlorobenzene (DCB, 99%, Sigma-Aldrich) wereadded. The flask was equipped with a long Liebig condenser,gas-volume spacer, and gas-release line to accommodate thelarge volume of CO, which was produced upon decompositionof the carbonyl precursor. With vigorous stirring, the mixture

was heated to one of the desired temperatures (164, 168, 176,or 182 °C) at a heating rate of 5 °C min−1 under an Aratmosphere. Once stabilized at the desired temperature, 1.5 mLof Co2(CO)8 (Sigma-Aldrich) in DCB (0.5 M) was quicklyinjected into the heated solution. The transparent, brownishsolution immediately turned black, indicating the formation ofcolloidal nanoparticles. This colloidal suspension was aged for20 min prior to stopping the heating and then cooled in a flowof air. For the cobalt nanoparticles to be separated, 5 mL ofDCB and 25 mL of 2-propanol (99%, Sigma-Aldrich) wereadded to the suspension, followed by centrifugation at 8000rpm for 15 min. After discarding the supernatant, precipitatednanoparticles were dispersed in chloroform (CHCl3, Samchun).

Acid Treatment of the CNTs. First, 2.5 g of carbonnanotubes (CNTs, Carbon Nanomaterial Technology Com-pany, MR 99) were mixed with 380 g of 6 M HCl (Samchun),and the mixture was stirred at 80 °C for 12 h. The suspensionwas filtered, washed with copious amounts of deionized (DI)water until the pH of the filtrate reached 7, and dried at 60 °C.The HCl-treated CNTs were subsequently washed with 390 gof 6 M HNO3 (Samchun).

Synthesis of the CoOx/CNTs. The CoOx nanoparticles(NPs) loaded on the CNTs (CoOx/CNTs) were synthesizedas follows. First, 350 mg of CNTs were dispersed in 50 mL ofCHCl3 in a 100 mL Erlenmeyer flask. The flask was coveredwith Parafilm to avoid evaporation of the CHCl3. After stirringfor 15 min, 38.9 mg of the as-prepared cobalt nanoparticles,dispersed in CHCl3, were added dropwise to the solution.Subsequent sonication in ice water for 3 h led to thehomogeneous dispersion of CoOx nanoparticles on CNTs.The product was collected by centrifugation and dried at 60 °C.Finally, the surfactants surrounding the nanoparticles wereremoved following a previously reported method.48 The driedpowder was annealed by raising the temperature from roomtemperature to 185 °C for 2 h and maintained at thattemperature for 5 h under air. For comparative purposes, acid-treated CNTs without CoOx were also annealed at 185 °C for 5h in the same manner as above.

Synthesis of CoOOH. CoOOH was synthesized for use asa reference material for the X-ray absorption spectroscopystudy, as previously reported.49 First, Co(OH)2 powder(Sigma-Aldrich) was dispersed in 40 mL of DI water. Next,10 mL of 8 M NaOH was added dropwise, and 4 mL of H2O2(30%, Sigma-Aldrich) was subsequently added with vigorousstirring. This reaction produces an explosive amount of O2 gas.The resulting dark-brown-colored suspension was stirred at 45°C for 28 h. The product was filtered, washed with DI waterseveral times, and dried overnight at 60 °C. The resultingCoOOH was found to be phase-pure with large crystallite sizeas revealed by XRD (Figure S9).

Characterization Methods. Scanning electron microscopy(SEM) analysis was conducted on a Hitachi S-4800 scanningelectron microscope operating at 10 kV. High-resolutiontransmission electron microscopy (HR-TEM) images weretaken on a JEOL JEM-2100 electron microscope at anacceleration voltage of 200 kV. Atomic-resolution TEM (AR-TEM) images were taken with a low-voltage sphericalaberration-corrected TEM (FEI Titan3 G2 60−300 with animage Cs corrector) with an acceleration voltage of 80 kV. X-ray powder diffraction (XRD) patterns were obtained with ahigh power X-ray diffractometer (Rigaku) equipped with CuKα radiation and operated at 40 kV and 200 mA. Wide-angleXRD patterns were measured in a 2θ range from 10° to 80° at a

ACS Catalysis Research Article

DOI: 10.1021/acscatal.6b00553ACS Catal. 2016, 6, 4347−4355

4353

scan rate of 4° min−1. The Co-metal content of each samplewas analyzed using an inductively coupled plasma opticalemission spectrometry (ICP-OES) analyzer (Varian). The ICP-OES analysis results are summarized in Table S1.X-ray Absorption Spectroscopy Experiments. X-ray

absorption spectroscopy (XAS) experiments were conductedon the Beamlines 6D and 10C of the Pohang AcceleratorLaboratory (PAL) in South Korea with a beam energy andcurrent of 3 GeV and 300 mA, respectively. X-ray photonenergy was monochromatized with a Si(111) double-crystalmonochromator, which was detuned by around 15 and 30% atthe 6D and 10C beamlines, respectively, to remove high-orderharmonics. In situ XAS spectra were obtained by using ahomemade spectroelectrochemical cell in fluorescence mode.Catalyst ink was dropped and dried on a piece of carbon fiberpaper and contacted with the electrolyte (0.1 M KOH) in thecell. In situ XAS measurement at OCV was first conducted, andthe subsequent XAS scan was then performed after applyingORR (0.6 V vs RHE, iR-corrected) or OER (1.8 V) potentialfor 1 h to give enough time for phase transformation.Background removal and normalization of the spectra werecarried out by using IFEFFIT (Athena) software.50

Electrochemical Measurements. Electrochemical charac-terization of the catalysts was performed using an IviumStatelectrochemical analyzer at room temperature (25 °C) andatmospheric pressure using a three-electrode system. A graphitecounter electrode and a Hg/HgO reference electrode wereused. All potentials in this report were converted to thereversible hydrogen electrode (RHE) scale.A rotating ring-disk electrode (RRDE) composed of a glassy

carbon (GC) disk (4 mm in diameter) and a Pt ring was usedas a working electrode. The RRDE was polished with a 1.0 μmalumina suspension and then with a 0.3 μm suspension togenerate a mirror finish. The catalyst ink was prepared bymixing catalyst (7.5 mg), neutralized Nafion (0.2 mL), DI water(0.1 mL), and absolute ethanol (0.9 mL) with sonication for atleast 1 h. Neutralized Nafion was prepared by mixing 0.1 MNaOH (99.99%, Sigma-Aldrich) and Nafion (5 wt %, Sigma-Aldrich) in a ratio of 1:2 (v:v), considering the protonconcentration of Nafion (∼0.05 M), to minimize anytransformation of the catalyst during the ink preparation.51

Next, 3 μL of the catalyst ink was pipetted with a microsyringeand deposited onto the GC electrode and dried at 70 °C for 2min. The catalyst loading was 0.15 mg cm−2.For the redox properties of the samples to be investigated,

cyclic voltammetry (CV) from 0.05 to 1.50 V (vs RHE) wasconducted in N2-saturated 1 M KOH at a scan rate of 20 mVs−1 (Figure 5 and Figure S5). Before the activity measurements,electrochemical impedance spectroscopy (EIS) was conductedaround an open circuit potential with a potential amplitude of10 mV from 10000 to 1 Hz. Series resistance was determined ata high frequency intercept on the x axis (real part) of the EISspectra, which was used to correct the iR-drop. Theelectrocatalytic activity toward the oxygen evolution reaction(OER) was obtained from 10 CV cycles in the range of 1.2 to1.8 V (vs RHE) at a scan rate of 20 mV s−1 with an electroderotating at 1600 rpm. The OER measurement wasindependently carried out three times, and the cathodic andanodic currents were averaged. The averaged and iR-compensated 10th CV scans are displayed in the report. Linearsweep voltammetry (LSV) polarization curves for the oxygenreduction reaction (ORR) were obtained by sweeping thepotential from 1.1 to 0.2 V (vs RHE) at a scan rate of 5 mV s−1

in O2-saturated 0.1 M KOH with O2 bubbling at a rotatingspeed of 1600 rpm. The ORR measurements wereindependently repeated three times, and the averaged and iR-compensated data are presented.For the evaluation of the kinetics for the ORR, the kinetic

current was extracted from the equation

= +I I I1 1 1

K L

where I, IK, and IL are the measured current, kinetic current,and diffusion-limited current, respectively.The logarithmic plot of the kinetic current density (just the

current density in the case of the OER) versus the overpotentialgives a linear Tafel plot

η = − +b J b Jlog( ) log( )K 0

where η, b, JK, and J0 are the applied overpotential, Tafel slope,kinetic current density, and exchange current density,respectively.

■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acscatal.6b00553.

Average particle sizes, OER and ORR activities, Cocontents, TEM images and FFT patterns, SEM images,XANES and XPS spectra, CV and chronopotentiometrycurves, and XRD patterns (PDF)

■ AUTHOR INFORMATIONCorresponding Authors*E-mail: hulex@unist.ac.kr (H.Y.J.).*E-mail: shjoo@unist.ac.kr (S.H.J.).Author Contributions∥B.S. and Y.J.S. contributed equally to this work.NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThis work was supported by the National Research Foundation(NRF) of Korea (2013R1A1A2012960, 2015M1A2A2056560),the KETEP funded by the Ministry of Trade, Industry, &Energy (MOTIE, 20133030011320), and the KEIT funded bythe MOTIE (10050509). B.S. and Y.J.S. acknowledge theGlobal Ph.D. Fellowship (2013H1A2A1032647 to B.S. and2013H1A2A1032644 to Y.J.S.). The EXAFS experimentsperformed at the Beamlines 6D and 10C of Pohang AcceleratorLaboratory (PAL) were supported in part by the Ministry ofEducation and POSTECH.

■ REFERENCES(1) Cook, T. R.; Dogutan, D. K.; Reece, S. Y.; Surendranath, Y.;Teets, T. S.; Nocera, D. G. Chem. Rev. 2010, 110, 6474−6502.(2) Chu, S.; Majumdar, A. Nature 2012, 488, 294−303.(3) Park, S.; Shao, Y.; Liu, J.; Wang, Y. Energy Environ. Sci. 2012, 5,9331−9344.(4) Cao, R.; Lee, J.-S.; Liu, M.; Cho, J. Adv. Energy Mater. 2012, 2,816−829.(5) Katsounaros, I.; Cherevko, S.; Zeradjanin, A. R.; Mayrhofer, K. J.J. Angew. Chem., Int. Ed. 2014, 53, 102−121.(6) Jiao, Y.; Zheng, Y.; Jaroniec, M.; Qiao, S. Z. Chem. Soc. Rev. 2015,44, 2060−2086.

ACS Catalysis Research Article

DOI: 10.1021/acscatal.6b00553ACS Catal. 2016, 6, 4347−4355

4354

(7) Zhang, J.; Zhao, Z.; Xia, Z.; Dai, L. Nat. Nanotechnol. 2015, 10,444−452.(8) Duan, J.; Chen, S.; Jaroniec, M.; Qiao, S. Z. ACS Catal. 2015, 5,5207−5234.(9) Mitlitsky, F.; Myers, B.; Weisberg, A. H. Energy Fuels 1998, 12,56−71.(10) Ng, J. W. D.; Gorlin, Y.; Hatsukade, T.; Jaramillo, T. F. Adv.Energy Mater. 2013, 3, 1545−1550.(11) Bruce, P. G.; Freunberger, S. A.; Hardwick, L. J.; Tarascon, J.-M.Nat. Mater. 2012, 11, 19−29.(12) Cheng, F.; Chen, J. Chem. Soc. Rev. 2012, 41, 2172−2192.(13) Ioroi, T.; Kitazawa, N.; Yasuda, K.; Yamamoto, Y.; Takenaka, H.J. Electrochem. Soc. 2000, 147, 2018−2022.(14) Chen, G.; Bare, S. R.; Mallouk, T. E. J. Electrochem. Soc. 2002,149, A1092−A1099.(15) Gorlin, Y.; Jaramillo, T. F. J. Am. Chem. Soc. 2010, 132, 13612−13614.(16) Liang, Y.; Li, Y.; Wang, H.; Zhou, J.; Wang, J.; Regier, T.; Dai,H. Nat. Mater. 2011, 10, 780−786.(17) Cheng, F.; Shen, J.; Peng, B.; Pan, Y.; Tao, Z.; Chen, J. Nat.Chem. 2011, 3, 79−84.(18) Maiyalagan, T.; Jarvis, K. A.; Therese, S.; Ferreira, P. J.;Manthiram, A. Nat. Commun. 2014, 5, 3949.(19) Indra, A.; Menezes, P. W.; Sahraie, N. R.; Bergmann, A.; Das,C.; Tallarida, M.; Schmeiβer, D.; Strasser, P.; Driess, M. J. Am. Chem.Soc. 2014, 136, 17530−17536.(20) Jung, J.-I.; Jeong, H. Y.; Lee, J.-S.; Kim, M. G.; Cho, J. Angew.Chem., Int. Ed. 2014, 53, 4582−4586.(21) Chen, S.; Duan, J.; Bian, P.; Tang, Y.; Zheng, R.; Qiao, S.-Z. Adv.Energy Mater. 2015, 5, 1500936.(22) Sa, Y. J.; Kwon, K.; Cheon, J. Y.; Kleitz, F.; Joo, S. H. J. Mater.Chem. A 2013, 1, 9992−10001.(23) Masa, J.; Xia, W.; Sinev, I.; Zhao, A.; Sun, Z.; Grutzke, S.; Weide,P.; Muhler, M.; Schuhmann, W. Angew. Chem., Int. Ed. 2014, 53,8508−8512.(24) Ma, T. Y.; Dai, S.; Jaroniec, M.; Qiao, S. Z. J. Am. Chem. Soc.2014, 136, 13925−13931.(25) den Breejen, J. P.; Radstake, P. B.; Bezemer, G. L.; Bitter, J. H.;Frøseth, V.; Holmen, A.; de Jong, K. P. J. Am. Chem. Soc. 2009, 131,7197−7203.(26) Iablokov, V.; Beaumont, S. K.; Alayoglu, S.; Pushkarev, V. V.;Specht, C.; Gao, J.; Alivisatos, A. P.; Kruse, N.; Somorjai, G. A. NanoLett. 2012, 12, 3091−3096.(27) Esswein, A. J.; McMurdo, M. J.; Ross, P. N.; Bell, A. T.; Tilley,T. D. J. Phys. Chem. C 2009, 113, 15068−15072.(28) Liu, J.; Jiang, L.; Zhang, B.; Jin, J.; Su, D. S.; Wang, S.; Sun, G.ACS Catal. 2014, 4, 2998−3001.(29) Rosen, J.; Hutchings, G. S.; Lu, Q.; Forest, R. V.; Moore, A.;Jiao, F. ACS Catal. 2015, 5, 4586−4591.(30) Olson, T. S.; Pylypenko, S.; Atanassov, P.; Asazawa, K.; Yamada,K.; Tanaka, H. J. Phys. Chem. C 2010, 114, 5049−5059.(31) Jin, H.; Wang, J.; Su, D.; Wei, Z.; Pang, Z.; Wang, Y. J. Am.Chem. Soc. 2015, 137, 2688−2694.(32) Friebel, D.; Louie, M. W.; Bajdich, M.; Sanwald, K. E.; Cai, Y.;Wise, A. M.; Cheng, M.-J.; Sokaras, D.; Weng, T.-C.; Alonso-Mori, R.;Davis, R. C.; Bargar, J. R.; Nørskov, J. K.; Nilsson, A.; Bell, A. T. J. Am.Chem. Soc. 2015, 137, 1305−1313.(33) Risch, M.; Zaharieva, K. I.; Gerencser, L.; Chernev, P.; Dau, H. J.Am. Chem. Soc. 2009, 131, 6936−6937.(34) Zhang, M.; de Respinis, M.; Frei, H. Nat. Chem. 2014, 6, 362−367.(35) Plaisance, C. P.; van Santen, R. A. J. Am. Chem. Soc. 2015, 137,14660−14672.(36) Liu, Y.-C.; Koza, J. A.; Switzer, J. A. Electrochim. Acta 2014, 140,359−365.(37) Yeo, B. S.; Bell, A. T. J. Am. Chem. Soc. 2011, 133, 5587−5593.(38) Subbaraman, R.; Tripkovic, D.; Chang, K.-C.; Strmcnik, D.;Paulikas, A. P.; Hirunsit, P.; Chan, M.; Greeley, J.; Stamenkovic, V.;Markovic, N. M. Nat. Mater. 2012, 11, 550−557.

(39) Xiao, J.; Kuang, Q.; Yang, S.; Xiao, F.; Wang, S.; Guo, L. Sci. Rep.2013, 3, 2300.(40) Menezes, P. W.; Indra, A.; Gonzalez-Flores, D.; Sahraie, N. R.;Zaharieva, I.; Schwarze, M.; Strasser, P.; Dau, H.; Driess, M. ACSCatal. 2015, 5, 2017−2027.(41) Liang, Y.; Wang, H.; Diao, P.; Chang, W.; Hong, G.; Li, Y.;Gong, M.; Xie, L.; Zhou, J.; Wang, J.; Regier, T. Z.; Wei, F.; Dai, H. J.Am. Chem. Soc. 2012, 134, 15849−15857.(42) Johnson Wass, J. R. T.; Panas, I.; Asbjornsson, J.; Ahlberg, E. J.Electroanal. Chem. 2007, 599, 295−312.(43) Wu, L.; Li, Q.; Wu, C. H.; Zhu, H.; Mendoza-Garcia, A.; Shen,B.; Guo, J.; Sun, S. J. Am. Chem. Soc. 2015, 137, 7071−7074.(44) Gerken, J. B.; McAlpin, J. G.; Chen, J. Y. C.; Rigsby, M. L.;Casey, W. H.; Britt, R. D.; Stahl, S. S. J. Am. Chem. Soc. 2011, 133,14431−14442.(45) McAlpin, J. G.; Surendranath, Y.; Dinca, M.; Stich, T. A.; Stoian,S. A.; Casey, W. H.; Nocera, D. G.; Britt, R. D. J. Am. Chem. Soc. 2010,132, 6882−6883.(46) Song, F.; Hu, X. J. Am. Chem. Soc. 2014, 136, 16481−16484.(47) Liu, J.; Jiang, L.; Tang, Q.; Zhang, B.; Su, D. S.; Wang, S.; Sun,G. ChemSusChem 2012, 5, 2315−2318.(48) Li, D.; Wang, C.; Tripkovic, D.; Sun, S.; Markovic, N. M.;Stamenkovic, V. R. ACS Catal. 2012, 2, 1358−1362.(49) Yang, J.; Liu, H.; Martens, W. N.; Frost, R. L. J. Phys. Chem. C2010, 114, 111−119.(50) Ravel, B.; Newville, M. J. Synchrotron Radiat. 2005, 12, 537−541.(51) Suntivich, J.; Gasteiger, H. A.; Yabuuchi, N.; Shao-Horn, Y. J.Electrochem. Soc. 2010, 157, B1263−B1268.

ACS Catalysis Research Article

DOI: 10.1021/acscatal.6b00553ACS Catal. 2016, 6, 4347−4355

4355