activity–selectivity trends in the electrochemical ... · dispersed metal cations (m−n−c) are...

Activity−Selectivity Trends in the Electrochemical Production ofHydrogen Peroxide over Single-Site Metal−Nitrogen−CarbonCatalystsYanyan Sun,† Luca Silvioli,‡ Nastaran Ranjbar Sahraie,§ Wen Ju,† Jingkun Li,§ Andrea Zitolo,∥

Shuang Li,† Alexander Bagger,‡ Logi Arnarson,‡ Xingli Wang,† Tim Moeller,† Denis Bernsmeier,†

Jan Rossmeisl,*,‡ Fred́eŕic Jaouen,*,§ and Peter Strasser*,†

†Department of Chemistry, Technical University of Berlin, 10623 Berlin, Germany‡Nano-Science Center, Department of Chemistry, University of Copenhagen, Universitetsparken 5, 2100 Copenhagen, Denmark§CNRS, Universite ́ de Montpellier, ENSCM, UMR 5253, Institut Charles Gerhardt de Montpellier, 34090 Montpellier, France∥Synchrotron SOLEIL, L’Orme des Merisiers, BP 48 Saint Aubin, 91192 Gif-sur-Yvette, France

*S Supporting Information

ABSTRACT: Nitrogen-doped carbon materials featuring atomicallydispersed metal cations (M−N−C) are an emerging family of materialswith potential applications for electrocatalysis. The electrocatalytic activityof M−N−C materials toward four-electron oxygen reduction reaction(ORR) to H2O is a mainstream line of research for replacing platinum-group-metal-based catalysts at the cathode of fuel cells. However,fundamental and practical aspects of their electrocatalytic activity towardtwo-electron ORR to H2O2, a future green “dream” process for chemicalindustry, remain poorly understood. Here we combined computational andexperimental efforts to uncover the trends in electrochemical H2O2production over a series of M−N−C materials (M = Mn, Fe, Co, Ni,and Cu) exclusively comprising atomically dispersed M−Nx sites frommolecular first-principles to bench-scale electrolyzers operating at industrialcurrent density. We investigated the effect of the nature of a 3d metal within a series of M−N−C catalysts on theelectrocatalytic activity/selectivity for ORR (H2O2 and H2O products) and H2O2 reduction reaction (H2O2RR). Co−N−Ccatalyst was uncovered with outstanding H2O2 productivity considering its high ORR activity, highest H2O2 selectivity, andlowest H2O2RR activity. The activity−selectivity trend over M−N−C materials was further analyzed by density functionaltheory, providing molecular-scale understandings of experimental volcano trends for four- and two-electron ORR. Thepredicted binding energy of HO* intermediate over Co−N−C catalyst is located near the top of the volcano accounting forfavorable two-electron ORR. The industrial H2O2 productivity over Co−N−C catalyst was demonstrated in a microflow cell,exhibiting an unprecedented production rate of more than 4 mol peroxide gcatalyst

−1 h−1 at a current density of 50 mA cm−2.

■ INTRODUCTIONElectrochemical oxygen reduction reaction (ORR) to hydro-gen peroxide (H2O2, two-electron ORR) is a green and saferoute to on-site and small-scale production of peroxidecompared to the industrially established anthraquinoneprocess.1−6 Currently, the state-of-the-art ORR catalysts arebased on platinum-group metals (PGMs) and their alloys.However, for electrochemical peroxide production, suchcatalysts have two critical drawbacks, namely, high cost and,perhaps even more important, high selectivity for the four-electron ORR to H2O.

7−9 Recently, novel synthetic strategieswere reported that can orientate the ORR selectivity of PGM-based catalysts toward H2O2 instead of water. This wasachieved by shifting from metallic surfaces to the atomicdispersion of PGM elements within a carbon matrix dopedwith N or S elements (i.e., Pt−N−C or Pt−S−C

materials).1,7,8,10,11 Despite these efforts, the high cost ofPGMs, unsatisfactory beginning-of-life activity and selectivitytoward H2O2 production, as well as the risk of Pt single-atomsagglomeration during operation (forming Pt clusters that, inturn, favor four-electron ORR) are critical hindering factors ina practical application for electrochemical peroxide productionfrom O2. Therefore, it remains a grand challenge to develophighly efficient and durable PGM-free ORR catalysts forelectrochemical H2O2 production.Nitrogen-doped carbon materials have been reported as

promising alternative ORR catalysts for electrochemical H2O2production in acid medium owing to the use of abundantprecursors, facile preparation, and high selectivity.12−18

Received: May 24, 2019Published: July 15, 2019

Article

pubs.acs.org/JACSCite This: J. Am. Chem. Soc. 2019, 141, 12372−12381

© 2019 American Chemical Society 12372 DOI: 10.1021/jacs.9b05576J. Am. Chem. Soc. 2019, 141, 12372−12381

Dow

nloa

ded

via

TU

BE

RL

IN o

n Se

ptem

ber

25, 2

019

at 1

5:06

:11

(UT

C).

See

http

s://p

ubs.

acs.

org/

shar

ingg

uide

lines

for

opt

ions

on

how

to le

gitim

atel

y sh

are

publ

ishe

d ar

ticle

s.

pubs.acs.org/JACShttp://pubs.acs.org/action/showCitFormats?doi=10.1021/jacs.9b05576http://dx.doi.org/10.1021/jacs.9b05576

Although promising results have been achieved with metal-freenitrogen-doped carbons, a serious issue is their very high ORRoverpotential in acid medium. This has restricted thegeometric current density of H2O2 production in electrolyzerdevices, implying a high number of cells (or high totalgeometric area of membrane electrode assembly) and thushigh capital expenditure per rated production capacity. Theintroduction of specific transition metals into nitrogen-dopedcarbon frameworks, leading to M−Nx moieties with stabilizedand activated metal cations, is a reportedly broad strategy toovercome the very low ORR activity of pure nitrogen-dopedcarbons in acid electrolyte.19−22 Early studies resorted to thesurface functionalization of carbon materials by transition-metal macrocycles, such as Fe−, Co−, Ni−, Cu−, and Mn−porphyrins and −phthalocyanines, with a well-defined M−N4reactive site.20,21,23,24 Considerable experimental and theoreti-cal efforts have been made to better understand their catalyticactivity and selectivity toward ORR.24−30 The distinctperformance was attributed to the different binding energiesof the ORR intermediates. In particular, Fe−porphyrins andFe−phthalocyanines were found to facilitate the four-electronORR pathway (H2O production), whereas Co−porphyrinsand Co−phthalocyanines were generally found to catalyze

more predominantly the two-electron ORR pathway, morepromising for electrochemical H2O2 production.

26,31 However,the complex synthesis and related high cost of some advancedM−N4 macrocycles and, more importantly, the general poorstability of such M−N4 cores in acidic medium have prohibitedtheir commercial application for electrochemical peroxideproduction.32−34 Nevertheless, inspired by the high initialORR activity of some M−N4 macrocycles in acidic medium,various synthetic protocols have been developed to preparetransition-metal nitrogen-doped carbon materials as ORRcatalysts through the pyrolysis of catalyst precursors thatcontain optimized amounts of separate, inexpensive, precursorsof the metal, nitrogen, and carbon.35−38 Henceforth, suchpyrolyzed materials are denoted as M−N−C. Among them,most studies focused on Fe−N−C materials due to their highcatalytic activity and selectivity toward four-electronORR,35,37,39 thus achieving high energy conversion efficiencyin fuel cells and metal−air batteries. In contrast, less attentionhas been paid to M−N−C catalysts with other transitionmetals and to the understanding of the influence of the natureof the transition metal on the catalytic activity and selectivitytoward H2O2 production.

40−42 Moreover, establishing mean-ingful and reliable structure−activity or structure−selectivity

Figure 1. (a) Synthesis scheme of the M−N−C catalysts (M = Co, Ni, Fe, Cu, and Mn), and SEM image of the Co−N−C catalyst. (b) TEMimages of the Co−N−C catalyst. (c) XRD patterns of the M−N−C catalysts. (d and e) High-resolution Co 2p spectra and high-resolution N 1sspectra of Co−N−C catalyst. N species fitted are from low to high binding energy pyridinic-N (olive), M−Nx (orange), pyrrolic-N (magenta),graphitic-N (blue), and N-oxide (gray). Note: peaks labeled with an asterisk (*) belong to silicon substrate.49

Journal of the American Chemical Society Article

DOI: 10.1021/jacs.9b05576J. Am. Chem. Soc. 2019, 141, 12372−12381

12373

http://dx.doi.org/10.1021/jacs.9b05576

relationships for the class of M−N−C materials has been, untilrecently, extremely difficult due to the undesired coexistence oftwo completely different types of metal species in mostpyrolyzed materials, namely, atomically dispersed metal sites(M−Nx moieties) and metal aggregates (most often metallic ormetal carbide).37,39,43 Starting from these observations, itappears highly desirable to develop a universal approach forthe preparation of M−N−C catalysts that exclusively containM−Nx moieties and to establish the structure−performancerelationships for H2O2 production.In the present work, a series of M−N−C catalysts (M = Mn,

Fe, Co, Ni, and Cu) comprising exclusively M−Nx moietiesembedded in an otherwise similar structural morphology of thenitrogen-doped carbon was prepared from ZIF-8 (a Zn(II)zeolitic imidazolate framework) and 1,10-phenanthroline ascarbon and nitrogen sources and the corresponding transition-metal acetate as metal sources. The performance of the M−N−C catalysts toward both the ORR (activity and selectivity toH2O2 production) and the H2O2 reduction reaction(H2O2RR) were investigated by rotating ring-disk electrode(RRDE) and rotating disk electrode (RDE) setup, respectively.Experimentally establishing the activity and selectivity land-scape toward ORR and H2O2RR over this series of M−N−Ccatalysts with well-defined M−N4 moieties opens the door to aconsistent and reliable interpretation of experimental resultswith density functional theory (DFT) modeling, exploring therelationships between fundamental property and electro-chemical performance. Finally, laboratory-scale H2O2 produc-tion was demonstrated in a microflow cell electrolyzer reactorwith a Co−N−C cathode layer.

■ RESULTS AND DISCUSSIONThe synthetic steps for the preparation of the M−N−Ccatalysts are illustrated in Figure 1a. Briefly, the catalystprecursor consisting of ZIF-8 (Basolite Z1200 from BASF),1,10-phenanthroline, and the targeted metal acetate as metalsource for the M−Nx moieties was prepared by low-energy ballmilling in the dry state and subsequently flash-pyrolyzed in Aratmosphere to directly obtain the M−N−C catalysts withoutany postpyrolysis treatment. A reference sample labeled N−Cwas prepared identically, except that no metal acetate wasadded. The structural morphologies of the M−N−C catalystswere first characterized using scanning electron microscopy(SEM) and transmission electron microscopy (TEM). TheSEM results (Figure 1a and Figure S1) demonstrate that asimilar morphology as that of N−C is observed regardless ofthe nature of the additional metal acetate precursor withwrinkled sheet-like morphology. No metal particles could bedetected with SEM, implying that introduction of a transitionmetal in the synthesis does not result in significant structuralchanges during the pyrolysis process. In particular, the TEMmicrograph of the Co−N−C catalyst confirms the absence ofmetal nanoparticles (Figure 1b). The Brunauer−Emmett−Teller (BET) surface areas of the M−N−C catalysts were alsomeasured by nitrogen sorption isotherms, which shared asimilar range from 315 to 430 m2 g−1, which allows forexposing a high number of catalytic sites for the ORRprocess.15 Further, powder X-ray diffraction (XRD) wasperformed to identify the presence of carbon- or metal-basedcrystalline structures. The two broad diffraction peaks observedfor all M−N−C catalysts and also for N−C at around 25° and44° are assigned to amorphous carbon,44 and no characteristicpeaks related to metal-rich phases could be observed (Figure

1c), which is consistent with SEM and TEM results (Figure S1and Figure 1b).The surface compositions and electronic states of the M−

N−C catalysts were then investigated using X-ray photo-electron spectroscopy (XPS). The narrow-scan metal 2pspectra can be well divided into two regions of high and lowenergy, separated by 12−20 eV depending on the nature of themetal, and that are assigned to the 2p1/2 and 2p3/2 regions ofeach metal, respectively. The 2p1/2 and 2p3/2 regions of eachmetal were fitted with two main components each and twosatellite peaks. In Figure 1d the two main peaks in the Co 2p3/2region have positions of 780.2 and 782.6 eV and can beambiguously assigned to either the Co2+ or the Co3+ oxidationstate.45,46 However, from operando X-ray absorption near-edgestructure (XANES) spectra on the same Co−N−C material,no change in the spectra was observed in the entire potentialrange from 0.1 to 1.0 VRHE.

47 Also, a redox peak was revealedat ca. 1.25 VRHE by square-wave voltammetry, assigned to theCo3+/Co2+ redox in Co−Nx moieties. From these combinedXPS, operando XANES, and electrochemical measurements,the two main components in the Co 2p3/2 region are assignedto Co2+ in two different types of Co−Nx moieties. Meanwhile,the narrow-scan metal 2p XPS spectra for other M−N−Ccatalysts were analyzed as well, and the results are shown inFigure S2b−e. These detailed analyses of narrow-scan metal 2pXPS spectra combined with XRD patterns converge to the ideathat metal atoms in these M−N−C catalysts are well dispersedin the nitrogen-doped carbon matrix in the form of metalcations coordinated with nitrogen atoms as M−Nx moieties,which is further confirmed by metal K-edge extended X-rayabsorption fine structure (EXAFS) analysis (Figure S3 andTable S4). In addition to the investigation of the metaloxidation state, the nitrogen element was also analyzed indetail with XPS. The high-resolution N 1s XPS spectra of allM−N−C catalysts (Figure 1e and Figure S4) can be fitted withthe same set of five components assigned to different nitrogenelectronic environments, including pyridinic-N (398.5 eV),metal-coordinated M−Nx moieties (399.5 eV), pyrrolic-N(401 eV), graphitic-N (402.4 eV), as well as N-oxide (404eV).5,21,44,48 A similar overall nitrogen content (∼5 atom %) isobserved in the different M−N−C catalysts, whereas the metalcontents vary in the range from 0.29 to 0.56 atom %, except forZn whose presence is due to unevaporated Zn from the nativeZIF-8 (Table S2). Inductively coupled plasma optical emissionspectrometry (ICP-OES) results demonstrate that the bulkmetal contents (except for Zn) are in the range from 1.0 to2.65 wt %. This is systematically larger than the bulk metalcontent in the catalyst precursor, i.e., before pyrolysis (0.5 wt%, Table S3), due to a concentration effect resulting from thesignificant mass loss of phenanthroline and ZIF-8 during thepyrolysis process forming C-, N-, and Zn-based volatileproducts, while the 3d transition metals from metal acetatedo not form volatile products.On the basis of all these characterizations, we infer that these

M−N−C catalysts possess similar physicochemical properties,differing only in the nature of the transition metal in thereactive sites. The latter are atomically dispersed in thenitrogen-doped carbon matrix in the form of M−Nx moieties,which provide a possibility to elucidate how transition metalsaffect the catalytic activity and selectivity of M−N−C catalyststoward ORR to H2O2 production. This was evaluated in O2-saturated 0.5 M H2SO4 using the rotating ring-disk electrode(RRDE) technique with an optimal catalyst loading of 0.1 mg



12374

http://pubs.acs.org/doi/suppl/10.1021/jacs.9b05576/suppl_file/ja9b05576_si_001.pdfhttp://pubs.acs.org/doi/suppl/10.1021/jacs.9b05576/suppl_file/ja9b05576_si_001.pdfhttp://pubs.acs.org/doi/suppl/10.1021/jacs.9b05576/suppl_file/ja9b05576_si_001.pdfhttp://pubs.acs.org/doi/suppl/10.1021/jacs.9b05576/suppl_file/ja9b05576_si_001.pdfhttp://pubs.acs.org/doi/suppl/10.1021/jacs.9b05576/suppl_file/ja9b05576_si_001.pdfhttp://pubs.acs.org/doi/suppl/10.1021/jacs.9b05576/suppl_file/ja9b05576_si_001.pdfhttp://pubs.acs.org/doi/suppl/10.1021/jacs.9b05576/suppl_file/ja9b05576_si_001.pdfhttp://pubs.acs.org/doi/suppl/10.1021/jacs.9b05576/suppl_file/ja9b05576_si_001.pdfhttp://dx.doi.org/10.1021/jacs.9b05576

cm−2 (Figure S5). As displayed in Figure 2a, the Fe−N−Ccatalyst showed the highest ORR current density of 3.8 mAcm−2 at +0.1 VRHE (electrochemical potential relative to thereversible hydrogen electrode, RHE) and the most positiveonset potential (defined as the potential at the current densityof 0.01 mA cm−2) of +0.9 VRHE,

50 indicating the highest ORRcatalytic activity among the M−N−C catalysts. Nevertheless,the ring current corresponding to the oxidation of H2O2produced on the disk working electrode is slightly lower forthe Fe−N−C catalyst compared to the N−C catalyst (0.058 vs0.063 mA at +0.1 VRHE), implying that introduction of Femainly facilitates the four-electron ORR reaction pathway forH2O production. In contrast, the Co−N−C catalyst exhibits byfar the highest ring current (0.18 mA at +0.1 VRHE) with only aslightly smaller ORR current density (2.97 mA cm−2 at +0.1VRHE) and restricted negative shift of onset potential comparedto the Fe−N−C catalyst (+0.83 VRHE). The Co−N−Cmaterial thus achieves the best compromise between highcatalytic activity and high selectivity toward H2O2 production.The Mn−N−C catalyst is the third most active catalyst, withan oxygen reduction current of 2.9 mA cm−2 and a ring currentof 0.07 mA at +0.1 VRHE. The remaining two catalysts, Cu−and Ni−N−C, exhibit almost the same ring current andoxygen reduction current density as that of the N−C catalyst,indicating a minor role of Cu and Ni for improving the ORRperformance in acidic medium. The H2O2 selectivity and thenumber of electron transferred (n) during the ORR processwere also calculated (eqs S2 and S3 in the SupportingInformation). For all M−N−C catalysts, the H2O2 selectivityincreases as the applied potential is lowered. Moreover, theselectivity ranking among different catalysts at +0.1 VRHE isrepresentative for the trends seen at higher potential, as seen inFigure S6a. The advantage with a potential of 0.1 VRHE is

where the largest differences between catalysts are seen(especially, distinguishing Co−N−C from others), andanother advantage is that all catalysts (even those with lowonset potential for ORR) do significant ORR at this lowpotential. In detail, the H2O2 selectivity at 0.1 VRHE follows theorder of Fe−N−C (28%) < Cu−N−C (36%) < Mn−N−C(43%) < N−C (45%) < Ni−N−C (52%) < Co−N−C (80%),and the opposite trend for the number of electrons transferredis observed (Figure 2b), indicating the most favorable two-electron ORR pathway for H2O2 production over the Co−N−C catalyst. Besides selectivity, initial assessment of a catalyst forpractical H2O2 production capacity in a proton-exchangemembrane fuel cell (PEMFC) must address the catalyticactivity toward the electrochemical H2O2 reduction reaction(H2O2RR) since a highly active catalyst for ORR resulting in ahigh rate of peroxide production might nevertheless beunsuitable if it also can catalyze H2O2RR.

12,51,52 Therefore,the catalytic activity of the M−N−C catalysts towardelectrochemical H2O2RR was also evaluated in N2-saturated0.5 M H2SO4 containing 1 mM H2O2. The results (Figure 2cand Figure S7) reveal the trend in catalytic activity towardH2O2RR as Fe−N−C > Mn−N−C > Cu−N−C > N−C ≈Ni−N−C ≈ Co−N−C. This further confirms that Co−N−Cis the most suitable candidate for H2O2 production amongthose catalysts due to an adequate balance between high ORRactivity, high selectivity toward peroxide production during theORR process, and low activity toward H2O2RR. Moreover, thepresent Co−N−C catalyst also exhibits comparable or superiorperformance to other catalyst types previously reported forelectrochemical H2O2 production under similar conditions,including various nitrogen-doped carbons,53,54 transition-metal-based materials,55,56 and noble-metal-based materials(see Table S5).2,3,10,51

Figure 2. (a) Linear sweep voltammetry (LSV) in a rotating ring-disk electrode (RRDE) setup with the Pt ring held at +1.2 VRHE. (b) H2O2selectivity (H2O2 %) and the number of electrons (n) at +0.1 VRHE derived from RRDE data. (c) Background-corrected H2O2RR performance inN2-saturated 0.5 M H2SO4 electrolyte containing 1 mM H2O2.



12375

http://pubs.acs.org/doi/suppl/10.1021/jacs.9b05576/suppl_file/ja9b05576_si_001.pdfhttp://pubs.acs.org/doi/suppl/10.1021/jacs.9b05576/suppl_file/ja9b05576_si_001.pdfhttp://pubs.acs.org/doi/suppl/10.1021/jacs.9b05576/suppl_file/ja9b05576_si_001.pdfhttp://pubs.acs.org/doi/suppl/10.1021/jacs.9b05576/suppl_file/ja9b05576_si_001.pdfhttp://pubs.acs.org/doi/suppl/10.1021/jacs.9b05576/suppl_file/ja9b05576_si_001.pdfhttp://pubs.acs.org/doi/suppl/10.1021/jacs.9b05576/suppl_file/ja9b05576_si_001.pdfhttp://dx.doi.org/10.1021/jacs.9b05576

Density functional theory (DFT) simulations were thenconducted to understand the catalytic trends of the M−N−Ccatalysts toward ORR to H2O2 production. It is wellestablished that the electrochemical four-electron ORR toH2O involves three reaction-adsorbed intermediates includingHOO*, O*, and HO* (where the asterisk (*) denotes theactive site), whereas the two-electron ORR to H2O2 onlyinvolves a single adsorbed HOO* intermediate.1,2 Thecatalytic activity and selectivity of the catalysts toward H2O2production is thus mainly determined by the binding freeenergy of HOO* (GHOO).

1 Given the existence of a constantscaling of 3.2 ± 0.2 eV between the binding free energy ofHOO* and HO*, the binding free energy of HO* (GHO) canalso be used as a descriptor for the two-electron ORR.1,2 As aresult, the activity and selectivity volcanoes for both the four-electron and the two-electron ORR pathways over the differentM−N−C catalysts can be constructed based on their bindingfree energy of the HO* intermediate.30 To systematicallycalculate the effect of the nature of the metal, a set of M−N−Ccatalysts featuring a single M−N4 site embedded in a graphenesheet was modeled (more details in the SupportingInformation computational section). To calculate the ORRintermediate binding energies, we applied the computationalreversible hydrogen electrode scheme,57 which does notaccount for pH effects. We reported the DFT results displayedin Figure 3 about the volcano-type relationship between the

thermodynamic limiting potential, U (i.e., the highesttheoretical potential at which all reduction reaction steps arestill energetically downhill)57 and GHO. The DFT volcanocurves are shown as solid lines, with U values shown on the lefty-axis. We chose Fe−N−C catalyst as the reference point forthe DFT study due to its highest experimental activity for four-electron ORR and widespread earlier research on thiscatalyst,58−62 which suggest the activity to derive from catalysison Fe−N4-like site. Conveniently, we report the catalyticactivity and selectivity of the M−N−C catalysts as a functionalof the binding free energy of HO* intermediate: the GHOdescriptor represents more accurately the thermodynamics ofstrong binding catalysts, therefore minimizing the uncertainty

on the reference catalyst, Fe−N−C. The theoretical insightsare correlated to experimental data shown in Figure 2a−c by asecond y-axis on the right of Figure 3 reporting ln(|jO2→H2O|)and ln(|jO2→H2O2|). We adjust the second y-axis scale to overlapFe−N−C calculated and measured activity. With this approachwe can compare the catalytic activity across the M−N−Ccatalysts and investigate whether the calculated structures holdthe experimental jO2→H2O and jO2→H2O2 partial current densitytrend. For Fe and Mn−N−C catalysts, the strong binding ofHO* results in the predominance of a four-electron pathwayover the two-electron pathway, as visible from the much lowerrelative position of their Δ relative to X symbols. From thisplot the lower experimental activity for four-electron ORR ofMn−N−C vs Fe−N−C is thus attributed to its strongerbinding of HO*. Similarly, the highest selectivity of the Co−N−C catalyst toward H2O2 is justified by its binding freeenergy for HO* intermediate with a value that places Co−N−C close to the top of the two-electron volcano, retainingcatalytic activity while promoting two-electron pathwayselectivity. The calculated activity of Co−N−C is comparableto hazardous mercury/noble metals alloys proposed in earlierresearch,1,2 albeit a direct comparison of the absolute currentsis difficult, due to differences in number of sites and surfacearea of the catalysts. On single-site catalysts the site geometrypromotes two-electron pathway selectivity at stronger oxygenbinding values than on corresponding metal catalysts, becausethe *O intermediate is destabilized relative to *OOH due toforced atop adsorption geometry.30 Meanwhile, Cu and Nibind very weakly the ORR intermediates, resulting in poorperoxide selective but virtually inactive catalysts, due to stronglimitations in forming the HOO* intermediate. We suggestthat the observed activity on Cu and Ni may originate fromgraphene edge defects and nonmetal nitrogen sites in the N−Csubstrate. Other kinetic factors may play a role that our modelfails to represent: for Co, our model underestimates theperoxide selectivity, as its GO = 2.7 ± 0.2 eV is markedly lowerthan 3.56 eV, H2O2 free energy of formation relative to H2O.We believe on Co−N−C the kinetics of peroxide desorption isfaster than the kinetics of four-electron reduction (or H2O2decomposition), which is also evidenced by the results inFigure 2c.The experimental conditions for peroxide production over

the Co−N−C catalyst were then investigated as a function ofthe electrolyte pH, with three conditions covering a stronglyalkaline, strongly acidic, and neutral pH. A significant influenceon the catalytic activity and selectivity of the Co−N−C catalysttoward H2O2 production was found (Figure 4). The onsetpotential of oxygen reduction is in the following order of pH13 (0.95 VRHE) > pH 0.3 (0.83 VRHE) > pH 7 (0.71 VRHE).Meanwhile, there were two different regions for potential-dependent oxygen reduction to peroxide production: (1) from0.8 to 0.4 VRHE, the oxygen reduction current density and ringcurrent are in the order of pH 7< pH 0.3< pH 13; (2) below0.4 VRHE, the oxygen reduction current density at pH 7 and 13are both higher than in pH 0.3, but the opposite trend isobserved for the ring current. Thus, very low potential seemsto promote selectivity to peroxide in acidic pH. The detailedH2O2 selectivity results (Figure 4b and Figure S6b)demonstrate that, in acidic medium, the H2O2 selectivityincreases as the potential is shifted negatively, whereas inneutral and alkaline media the H2O2 selectivity remains almostconstant in the whole potential range. While at 0.1 VRHE the

Figure 3. Thermodynamic relations (volcano) lines for the two-(green solid line) and four-electron ORR (black solid line). The DFTcalculated ORR onset potential values (circles) are on the left y-axis,while the experimental current densities (crosses and triangles),reported as ln(|j|), are on the right y-axis. Both are shown as functionof the chosen reaction descriptor, the DFT calculated HO* bindingfree energy (GHO).



12376

http://pubs.acs.org/doi/suppl/10.1021/jacs.9b05576/suppl_file/ja9b05576_si_001.pdfhttp://pubs.acs.org/doi/suppl/10.1021/jacs.9b05576/suppl_file/ja9b05576_si_001.pdfhttp://pubs.acs.org/doi/suppl/10.1021/jacs.9b05576/suppl_file/ja9b05576_si_001.pdfhttp://dx.doi.org/10.1021/jacs.9b05576

highest H2O2 selectivity is observed in acidic medium, theselectivity gap between pH 0.3 and pH 13 decreases as thepotential is increased to 0.3 VRHE and becomes insignificant at0.5 VRHE (Figure 4b). In addition, Tafel curves were plottedfrom the RRDE results based on the Tafel equation (eq 1)63

a j blogη = + (1)

where η is the overpotential (the difference between theapplied potential and standard potential, η = E − E0), j is thekinetic current density, and a is the Tafel slope, in order toinvestigate the influence of electrolyte pH on the catalyticORR mechanism over the Co−N−C catalyst. The kineticcurrent densities (jdisk,kin) can be first calculated based on theKoutecky−Levich equation (eq 2)12

j j j1/ 1/ 1/disk disk,kin disk,lim= + (2)

where jdisk and jdisk,lim are the measured disk current density andthe diffusion-limited current density, respectively. In thepresent system, we take the disk current density value at 0.2VRHE as the diffusion-limited current density. The calculatedTafel slopes (Figure 4c) were 60 mV dec−1 in pH 0.3, 77 mVdec−1 in pH 7.0, and 50 mV dec−1 in pH 13.0, demonstrating asimilar rate-determining step as a function of electrolyte pH.The practical H2O2 production capabilities of the Co−N−C

catalyst under bulk electrolysis conditions were first examinedin an H-cell with different electrolytes, including 0.5 M H2SO4,0.1 M K2SO4, and 0.1 M KOH (Figure S8). The highest H2O2production rate of 193.1 mmol gcatalyst

−1 h−1 is obtained in 0.1M KOH at 0.1 VRHE followed by that in 0.5 M H2SO4 (90.9mmol gcatalyst

−1 h−1) and 0.1 M K2SO4 (89.8 mmol gcatalyst−1

h−1), indicating the significant influence of electrolyte pH onthe H2O2 production capacity of the Co−N−C catalyst.Subsequently, the catalytic performance during larger scale

H2O2 production at high current densities in 0.1 M KOH wasevaluated in a commercial microflow cell (MFC, ElectroCell)electrolyzer. As displayed in Figure 5a, the MFC setupcomprises Ir−MMO on Ti plate (ElectroCell) as anode(oxygen evolution reaction) and Co−N−C-modified gasdiffusion layer (GDL) as cathode (ORR). The use of GDLelectrode here could efficiently address the mass transportproblem arising from the low concentration of dissolved O2.

64

The electrochemical testing was carried out using thegalvanostatic method within the range from 2.5 to 65 mAcm−2 at each fixed current density for 15 min. The highestH2O2 production rate of 4.33 mol gcatalyst

−1 h−1 (Figure 5b) isachieved at a current density of 50 mA cm−2, whereas at highertotal current density, the H2O2 production rate reaches aplateau or even slightly decreases. The phenomenon may beattributed to a combination of slightly increased total currentdensity and decreased selectivity (Figure 5c). The latter is dueto competing H2O2RR at low potential, as shown in Figure 2c.Further, the stability performance of the catalyst at the currentdensity of 50 mA cm−2 was also evaluated. As observed, theconcentration of the produced H2O2 linearly increases with thereaction time for the first 3 h, indicating a constant rate ofperoxide formation (Figure 5e). For even longer duration, theincrease in H2O2 concentration with time slows down and thedecreased H2O2 FE was also observed, indicating a decreasingrate of instantaneous peroxide production (proportional to thefirst derivative of the curve). This is fully expected since theincreasing concentration of peroxide in the electrolyte leads toincreased reaction rate of peroxide reduction (increased

Figure 4. All the plots refer to Co−N−C catalyst: (a) LSV of RRDE with the ring current collected on the Pt ring at a constant potential of +1.2VRHE, and the disk geometric current density. (b) H2O2 selectivity (%) and the number of electrons transferred (n) at different applied electrodepotentials (+0.1, +0.3, and +0.5 VRHE). (c) Tafel plots. All measurements were performed in O2-saturated 0.5 M H2SO4 (pH 0.3), 0.1 M K2SO4(pH 7), and 0.1 M KOH (pH 13) at a scan rate of 5 mV s−1 with a loading amount of 0.1 mg cm−2.



12377

http://pubs.acs.org/doi/suppl/10.1021/jacs.9b05576/suppl_file/ja9b05576_si_001.pdfhttp://dx.doi.org/10.1021/jacs.9b05576

reactant concentration for H2O2RR) and H2O2 can decomposein alkaline solution overtime.65,66 Besides, the applied voltageat the cathode of MFC slowly becomes more negative as thereaction time prolongs (Figure 5d).

■ CONCLUSIONIn summary, we prepared a series of M−N−C catalysts (M =Co, Ni, Fe, Cu, and Mn) with similar structural morphologies,their only noticeable difference being the nature of the metalcation center in the catalytically active M−Nx moieties.Electrochemical measurements demonstrate that the Co−N−C catalyst exhibits a satisfactory ORR activity, the highest ORRselectivity toward H2O2 production, and the lowest H2O2RRactivity. This 3-fold combination of catalytic properties resultsin Co−N−C being the more suitable catalyst to produce H2O2among these M−N−C catalysts. Inversely, the Fe−N−Ccatalyst exhibits the highest ORR activity but the lowestselectivity toward H2O2 production. The systematic exper-imental screening of M−Nx moieties is rationalized with DFTcalculations that link experiments to the ORR volcano model.The rationale for the performance of M−N−C catalyst towardH2O2 production lies in the optimal binding of the ORRintermediates. Referencing the theoretical observations on M−N−C to a single, highly active structure (Fe−N−C) allows forthe properties of M−N−C close to the volcano top to be fairlywell represented. The binding free energy of Co−N−C placesit near the top of the H2O2 production volcano, making it anexcellent H2O2 producer, whereas the Mn−N−C and Fe−N−C bind the ORR intermediates strongly, making them H2O

producer. However, for Cu−N−C and Ni−N−C catalysts, theoverall activities are less well represented by the model,suggesting the metal center contributes to a minor extent inthe reactivity compared to its N−C defect surroundings. Thisbehavior is also observed for M−N−C catalysts in otherreactions, such as the CO2 reduction reaction. The super-imposition of experimental and theoretical data in a systematicstudy has the potential to individuate the source of catalyticperformance and pave the way toward rational design ofcatalysts in relevant reactions. In addition, the high H2O2production capacity of 4.33 mol gcatalyst

−1 h−1 over the Co−N−C catalyst could be achieved at a current density of 50 mAcm−2 in the MFC electrolyzer, showing the directions toward ascale-up of this electrochemical peroxide production technol-ogy.

■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/jacs.9b05576.

Additional material synthesis, structural characterization,and supporting electrochemical characterization (PDF)

■ AUTHOR INFORMATIONCorresponding Authors*E-mail: [email protected].*E-mail: [email protected].*E-mail: [email protected].

Figure 5. (a) Scheme of the MFC setup. (b−e) H2O2 production rate at the respective current density on a Co−N−C electrode with 0.1 mgCo−N−Ccm−2 in O2-saturated 0.1 M KOH: (b) H2O2 production rate normalized by the catalyst loading (kH2O2), (c) H2O2 faradaic efficiency (FE, %) as afunction of current density, (d) measured cell voltage with iR correction of MFC electrolyzer with time at the fixed current density of 50 mA cm−2,and (e) produced amount of H2O2 and H2O2 FE (%) with time at a current density of 50 mA cm

−2.



12378

http://pubs.acs.orghttp://pubs.acs.org/doi/abs/10.1021/jacs.9b05576http://pubs.acs.org/doi/suppl/10.1021/jacs.9b05576/suppl_file/ja9b05576_si_001.pdfmailto:[email protected]:[email protected]:[email protected]://dx.doi.org/10.1021/jacs.9b05576

ORCIDLuca Silvioli: 0000-0002-8169-2373Andrea Zitolo: 0000-0002-2187-6699Shuang Li: 0000-0001-7414-630XJan Rossmeisl: 0000-0001-7749-6567Fred́eŕic Jaouen: 0000-0001-9836-3261Peter Strasser: 0000-0002-3884-436XAuthor ContributionsAll authors have given approval to the final version of themanuscript.NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThis work was supported by FCH Joint Undertaking(CRESCENDO project, Grant Agreement No. 779366).Partial funding by the German Ministry of Economics andEnergy (BMWi) through project “ChemEFlex” (FKN0350013A) is gratefully acknowledged. ZELMI of TechnicalUniversity Berlin is acknowledged for their support with TEMmeasurements. Y.S. thanks the China Scholarship Council(CSC) and CRESCENDO project for financial support. A.B.,L.S., L.A., and J.R. acknowledge the Carlsberg Foundation(grant CF15-0165) and the Innovation Fund Denmark(ProActivE project 5160-00003B).

■ REFERENCES(1) Siahrostami, S.; Verdaguer-Casadevall, A.; Karamad, M.; Deiana,D.; Malacrida, P.; Wickman, B.; Escudero-Escribano, M.; Paoli, E. A.;Frydendal, R.; Hansen, T. W.; Chorkendorff, I.; Stephens, I. E.;Rossmeisl, J. Enabling direct H2O2 production through rationalelectrocatalyst design. Nat. Mater. 2013, 12, 1137−43.(2) Verdaguer-Casadevall, A.; Deiana, D.; Karamad, M.; Siahrostami,S.; Malacrida, P.; Hansen, T. W.; Rossmeisl, J.; Chorkendorff, I.;Stephens, I. E. Trends in the electrochemical synthesis of H2O2:enhancing activity and selectivity by electrocatalytic site engineering.Nano Lett. 2014, 14, 1603−1608.(3) Zhang, Y.; Ouyang, B.; Xu, J.; Chen, S.; Rawat, R. S.; Fan, H. J.3D Porous Hierarchical Nickel-Molybdenum Nitrides Synthesized byRF Plasma as Highly Active and Stable Hydrogen-Evolution-ReactionElectrocatalysts. Adv. Energy Mater. 2016, 6, 1600221.(4) Lu, Z.; Chen, G.; Siahrostami, S.; Chen, Z.; Liu, K.; Xie, J.; Liao,L.; Wu, T.; Lin, D.; Liu, Y.; Jaramillo, T. F.; Nørskov, J. K.; Cui, Y.High-efficiency oxygen reduction to hydrogen peroxide catalysed byoxidized carbon materials. Nat. Catal. 2018, 1, 156−162.(5) Chen, S.; Chen, Z.; Siahrostami, S.; Higgins, D.; Nordlund, D.;Sokaras, D.; Kim, T. R.; Liu, Y.; Yan, X.; Nilsson, E.; Sinclair, R.;Norskov, J. K.; Jaramillo, T. F.; Bao, Z. Designing Boron NitrideIslands in Carbon Materials for Efficient Electrochemical Synthesis ofHydrogen Peroxide. J. Am. Chem. Soc. 2018, 140, 7851−7859.(6) Chen, S.; Chen, Z.; Siahrostami, S.; Kim, T. R.; Nordlund, D.;Sokaras, D.; Nowak, S.; To, J. W. F.; Higgins, D.; Sinclair, R.;Nørskov, J. K.; Jaramillo, T. F.; Bao, Z. Defective Carbon-BasedMaterials for the Electrochemical Synthesis of Hydrogen Peroxide.ACS Sustainable Chem. Eng. 2018, 6, 311−317.(7) Strasser, P.; Gliech, M.; Kuehl, S.; Moeller, T. Electrochemicalprocesses on solid shaped nanoparticles with defined facets. Chem.Soc. Rev. 2018, 47, 715−735.(8) Beermann, V.; Gocyla, M.; Kuhl, S.; Padgett, E.; Schmies, H.;Goerlin, M.; Erini, N.; Shviro, M.; Heggen, M.; Dunin-Borkowski, R.E.; Muller, D. A.; Strasser, P. Tuning the Electrocatalytic OxygenReduction Reaction Activity and Stability of Shape-Controlled Pt-NiNanoparticles by Thermal Annealing - Elucidating the Surface AtomicStructural and Compositional Changes. J. Am. Chem. Soc. 2017, 139,16536−16547.

(9) Strasser, P. Free Electrons to Molecular Bonds and Back:Closing the Energetic Oxygen Reduction (ORR)-Oxygen Evolution(OER) Cycle Using Core-Shell Nanoelectrocatalysts. Acc. Chem. Res.2016, 49, 2658−2668.(10) Yang, S.; Tak, Y. J.; Kim, J.; Soon, A.; Lee, H. Support Effects inSingle-Atom Platinum Catalysts for Electrochemical Oxygen Reduc-tion. ACS Catal. 2017, 7, 1301−1307.(11) Choi, C. H.; Kim, M.; Kwon, H. C.; Cho, S. J.; Yun, S.; Kim, H.T.; Mayrhofer, K. J.; Kim, H.; Choi, M. Tuning selectivity ofelectrochemical reactions by atomically dispersed platinum catalyst.Nat. Commun. 2016, 7, 10922.(12) Sun, Y.; Li, S.; Jovanov, Z. P.; Bernsmeier, D.; Wang, H.; Paul,B.; Wang, X.; Kuhl, S.; Strasser, P. Structure, Activity, and FaradaicEfficiency of Nitrogen-Doped Porous Carbon Catalysts for DirectElectrochemical Hydrogen Peroxide Production. ChemSusChem 2018,11, 3388−3395.(13) Park, J.; Nabae, Y.; Hayakawa, T.; Kakimoto, M.-a. HighlySelective Two-Electron Oxygen Reduction Catalyzed by MesoporousNitrogen-Doped Carbon. ACS Catal. 2014, 4, 3749−3754.(14) Iglesias, D.; Giuliani, A.; Melchionna, M.; Marchesan, S.;Criado, A.; Nasi, L.; Bevilacqua, M.; Tavagnacco, C.; Vizza, F.; Prato,M.; Fornasiero, P. N-Doped Graphitized Carbon Nanohorns as aForefront Electrocatalyst in Highly Selective O2 Reduction to H2O2.Chem. 2018, 4, 106−123.(15) Fellinger, T. P.; Hasche, F.; Strasser, P.; Antonietti, M.Mesoporous nitrogen-doped carbon for the electrocatalytic synthesisof hydrogen peroxide. J. Am. Chem. Soc. 2012, 134, 4072−4075.(16) Liu, Y.; Quan, X.; Fan, X.; Wang, H.; Chen, S. High-yieldelectrosynthesis of hydrogen peroxide from oxygen reduction byhierarchically porous carbon. Angew. Chem., Int. Ed. 2015, 54, 6837−6841.(17) Han, L.; Sun, Y.; Li, S.; Cheng, C.; Halbig, C. E.; Feicht, P.;Hübner, J. L.; Strasser, P.; Eigler, S. In-Plane Carbon Lattice-DefectRegulating Electrochemical Oxygen Reduction to Hydrogen PeroxideProduction over Nitrogen-Doped Graphene. ACS Catal. 2019, 9,1283−1288.(18) Sun, Y.; Sinev, I.; Ju, W.; Bergmann, A.; Dresp, S.; Kühl, S.;Spöri, C.; Schmies, H.; Wang, H.; Bernsmeier, D.; Paul, B.; Schmack,R.; Kraehnert, R.; Roldan Cuenya, B.; Strasser, P. EfficientElectrochemical Hydrogen Peroxide Production from MolecularOxygen on Nitrogen-Doped Mesoporous Carbon Catalysts. ACSCatal. 2018, 8, 2844−2856.(19) Cao, R.; Thapa, R.; Kim, H.; Xu, X.; Gyu Kim, M.; Li, Q.; Park,N.; Liu, M.; Cho, J. Promotion of oxygen reduction by a bio-inspiredtethered iron phthalocyanine carbon nanotube-based catalyst. Nat.Commun. 2013, 4, 2076.(20) Passard, G.; Dogutan, D. K.; Qiu, M.; Costentin, C.; Nocera, D.G. Oxygen reduction reaction promoted by manganese porphyrins.ACS Catal. 2018, 8, 8671−8679.(21) Costentin, C.; Dridi, H.; Saveant, J. M. Molecular Catalysis ofO2 Reduction by Iron Porphyrins in Water: Heterogeneous versusHomogeneous Pathways. J. Am. Chem. Soc. 2015, 137, 13535−13544.(22) Wei, P. J.; Yu, G. Q.; Naruta, Y.; Liu, J. G. Covalent grafting ofcarbon nanotubes with a biomimetic heme model compound toenhance oxygen reduction reactions. Angew. Chem., Int. Ed. 2014, 53,6659−6663.(23) McGuire, R., Jr; Dogutan, D. K.; Teets, T. S.; Suntivich, J.;Shao-Horn, Y.; Nocera, D. G. Oxygen reduction reactivity ofcobalt(ii) hangman porphyrins. Chem. Sci. 2010, 1, 411−414.(24) Zhang, W.; Lai, W.; Cao, R. Energy-Related Small MoleculeActivation Reactions: Oxygen Reduction and Hydrogen and OxygenEvolution Reactions Catalyzed by Porphyrin- and Corrole-BasedSystems. Chem. Rev. 2017, 117, 3717−3797.(25) Kim, S.; Kim, H. Oxygen reduction reaction at porphyrin-basedelectrochemical catalysts: Mechanistic effects of pH and spin statesstudied by density functional theory. Catal. Today 2017, 295, 119−124.



12379

http://orcid.org/0000-0002-8169-2373http://orcid.org/0000-0002-2187-6699http://orcid.org/0000-0001-7414-630Xhttp://orcid.org/0000-0001-7749-6567http://orcid.org/0000-0001-9836-3261http://orcid.org/0000-0002-3884-436Xhttp://dx.doi.org/10.1021/jacs.9b05576

(26) Siahrostami, S.; Bjorketun, M. E.; Strasser, P.; Greeley, J.;Rossmeisl, J. Tandem cathode for proton exchange membrane fuelcells. Phys. Chem. Chem. Phys. 2013, 15, 9326−9334.(27) Mase, K.; Ohkubo, K.; Fukuzumi, S. Efficient two-electronreduction of dioxygen to hydrogen peroxide with one-electronreductants with a small overpotential catalyzed by a cobalt chlorincomplex. J. Am. Chem. Soc. 2013, 135, 2800−2808.(28) Oldacre, A. N.; Crawley, M. R.; Friedman, A. E.; Cook, T. R.Tuning the Activity of Heterogeneous Cofacial Cobalt Porphyrins forOxygen Reduction Electrocatalysis through Self-Assembly. Chem. -Eur. J. 2018, 24, 10984−10987.(29) Qu, J.; Shen, Y.; Qu, X.; Dong, S. Electrocatalytic Reduction ofOxygen at Multi-Walled Carbon Nanotubes and Cobalt PorphyrinModified Glassy Carbon Electrode. Electroanalysis 2004, 16, 1444−1450.(30) Yang, S.; Verdaguer-Casadevall, A.; Arnarson, L.; Silvioli, L.;Čolic,́ V.; Frydendal, R.; Rossmeisl, J.; Chorkendorff, I.; Stephens, I.E. L. Toward the Decentralized Electrochemical Production of H2O2:A Focus on the Catalysis. ACS Catal. 2018, 8, 4064−4081.(31) Jahan, M.; Bao, Q.; Loh, K. P. Electrocatalytically activegraphene-porphyrin MOF composite for oxygen reduction reaction. J.Am. Chem. Soc. 2012, 134, 6707−6713.(32) Wu, Z. S.; Chen, L.; Liu, J.; Parvez, K.; Liang, H.; Shu, J.;Sachdev, H.; Graf, R.; Feng, X.; Mullen, K. High-performanceelectrocatalysts for oxygen reduction derived from cobalt porphyrin-based conjugated mesoporous polymers. Adv. Mater. 2014, 26, 1450−1455.(33) Brüller, S.; Liang, H.-W.; Kramm, U. I.; Krumpfer, J. W.; Feng,X.; Müllen, K. Bimetallic porous porphyrin polymer-derived non-precious metal electrocatalysts for oxygen reduction reactions. J.Mater. Chem. A 2015, 3, 23799−23808.(34) Miller, H. A.; Bellini, M.; Oberhauser, W.; Deng, X.; Chen, H.;He, Q.; Passaponti, M.; Innocenti, M.; Yang, R.; Sun, F.; Jiang, Z.;Vizza, F. Heat treated carbon supported iron(ii)phthalocyanineoxygen reduction catalysts: elucidation of the structure-activityrelationship using X-ray absorption spectroscopy. Phys. Chem. Chem.Phys. 2016, 18, 33142−33151.(35) Leonard, N. D.; Wagner, S.; Luo, F.; Steinberg, J.; Ju, W.;Weidler, N.; Wang, H.; Kramm, U. I.; Strasser, P. Deconvolution ofUtilization, Site Density, and Turnover Frequency of Fe−Nitrogen−Carbon Oxygen Reduction Reaction Catalysts Prepared withSecondary N-Precursors. ACS Catal. 2018, 8, 1640−1647.(36) Ranjbar Sahraie, N.; Paraknowitsch, J. P.; Gobel, C.; Thomas,A.; Strasser, P. Noble-metal-free electrocatalysts with enhanced ORRperformance by task-specific functionalization of carbon using ionicliquid precursor systems. J. Am. Chem. Soc. 2014, 136, 14486−14497.(37) Wu, G.; More, K. L.; Johnston, C. M.; Zelenay, P. High-performance electrocatalysts for oxygen reduction derived frompolyaniline, iron, and cobalt. Science 2011, 332, 443−447.(38) Möller, T.; Ju, W.; Bagger, A.; Wang, X.; Luo, F.; Ngo Thanh,T.; Varela, A. S.; Rossmeisl, J.; Strasser, P. Efficient CO2 to COelectrolysis on solid Ni−N−C catalysts at industrial current densities.Energy Environ. Sci. 2019, 12, 640−647.(39) Yang, W.; Liu, X.; Yue, X.; Jia, J.; Guo, S. Bamboo-like carbonnanotube/Fe3C nanoparticle hybrids and their highly efficientcatalysis for oxygen reduction. J. Am. Chem. Soc. 2015, 137, 1436−1439.(40) Marcotte, S.; Villers, D.; Guillet, N.; Roue, L.; Dodelet, J.P.Electroreduction of oxygen on Co-based catalysts: determination ofthe parameters affecting the twoelectron transfer reaction in an acidmedium. Electrochim. Acta 2004, 50, 179−188.(41) Olson, T. S.; Pylypenko, S.; Atanassov, P.; Asazawa, K.;Yamada, K.; Tanaka, H. Anion-Exchange Membrane Fuel Cells: Dual-Site Mechanism of Oxygen Reduction Reaction in Alkaline Media onCobalt-Polypyrrole Electrocatalysts. J. Phys. Chem. C 2010, 114,5049−5059.(42) Campos, M.; Siriwatcharapiboon, W.; Potter, R. J.; Horswell, S.L. Selectivity of cobalt-based catalysts towards hydrogen peroxide

formation during the reduction of oxygen. Catal. Today 2013, 202,135−143.(43) Zhu, C.; Fu, S.; Shi, Q.; Du, D.; Lin, Y. Single-AtomElectrocatalysts. Angew. Chem., Int. Ed. 2017, 56, 13944−13960.(44) Ju, W.; Bagger, A.; Hao, G. P.; Varela, A. S.; Sinev, I.; Bon, V.;Roldan Cuenya, B.; Kaskel, S.; Rossmeisl, J.; Strasser, P. Under-standing activity and selectivity of metal-nitrogen-doped carboncatalysts for electrochemical reduction of CO2. Nat. Commun. 2017,8, 944.(45) Xue, Y.; Zuo, Z.; Li, Y.; Liu, H.; Li, Y. Graphdiyne-SupportedNiCo2S4 Nanowires: A Highly Active and Stable 3D BifunctionalElectrode Material. Small 2017, 13, 1700936.(46) Sivanantham, A.; Ganesan, P.; Shanmugam, S. HierarchicalNiCo2S4 Nanowire Arrays Supported on Ni Foam: An Efficient andDurable Bifunctional Electrocatalyst for Oxygen and HydrogenEvolution Reactions. Adv. Funct. Mater. 2016, 26, 4661−4672.(47) Zitolo, A.; Ranjbar-Sahraie, N.; Mineva, T.; Li, J.; Jia, Q.;Stamatin, S.; Harrington, G. F.; Lyth, S. M.; Krtil, P.; Mukerjee, S.;Fonda, E.; Jaouen, F. Identification of catalytic sites in cobalt-nitrogen-carbon materials for the oxygen reduction reaction. Nat.Commun. 2017, 8, 957.(48) Varela, A. S.; Ju, W.; Strasser, P. Molecular Nitrogen-CarbonCatalysts, Solid Metal Organic Framework Catalysts, and SolidMetal/Nitrogen-Doped Carbon (MNC) Catalysts for the Electro-chemical CO2 Reduction. Adv. Energy Mater. 2018, 8, 1703614.(49) Weng, M.-H.; Pan, C.-T.; Huang, C.-W.; Yang, R.-Y. Improvingthe Microstructure and Electrical Properties of Aluminum InducedPolysilicon Thin Films Using Silicon Nitride Capping Layer. J.Nanomater. 2014, 2014, 1−9.(50) Shui, J.; Chen, C.; Grabstanowicz, L.; Zhao, D.; Liu, D. J.Highly efficient nonprecious metal catalyst prepared with metal-organic framework in a continuous carbon nanofibrous network. Proc.Natl. Acad. Sci. U. S. A. 2015, 112, 10629−10634.(51) Choi, C. H.; Kwon, H. C.; Yook, S.; Shin, H.; Kim, H.; Choi,M. Hydrogen Peroxide Synthesis via Enhanced Two-Electron OxygenReduction Pathway on Carbon-Coated Pt Surface. J. Phys. Chem. C2014, 118, 30063−30070.(52) Jaouen, F.; Dodelet, J.-P. O2 Reduction Mechanism on Non-Noble Metal Catalysts for PEM Fuel Cells. Part I: Experimental Ratesof O2 Electroreduction, H2O2 Electroreduction, and H2O2 Dis-proportionation. J. Phys. Chem. C 2009, 113, 15422−15432.(53) Yang, Y.; He, F.; Shen, Y.; Chen, X.; Mei, H.; Liu, S.; Zhang, Y.A biomass derived N/C-catalyst for the electrochemical production ofhydrogen peroxide. Chem. Commun. 2017, 53, 9994−9997.(54) Hasche,́ F.; Oezaslan, M.; Strasser, P.; Fellinger, T.-P.Electrocatalytic hydrogen peroxide formation on mesoporous non-metal nitrogen-doped carbon catalyst. J. Energy Chem. 2016, 25, 251−257.(55) Bonakdarpour, A.; Esau, D.; Cheng, H.; Wang, A.; Gyenge, E.;Wilkinson, D. P. Preparation and electrochemical studies of metal−carbon composite catalysts for small-scale electrosynthesis of H2O2.Electrochim. Acta 2011, 56, 9074−9081.(56) Rojas-Carbonell, S.; Santoro, C.; Serov, A.; Atanassov, P.Transition metal-nitrogen-carbon catalysts for oxygen reductionreaction in neutral electrolyte. Electrochem. Commun. 2017, 75, 38−42.(57) Nørskov, J. K.; Rossmeisl, J.; Logadottir, A.; Lindqvist, L.;Kitchin, J. R.; Bligaard, T.; Jońsson, H. Origin of the Overpotential forOxygen Reduction at a Fuel-Cell Cathode. J. Phys. Chem. B 2004, 108,17886−17892.(58) Holst-Olesen, K.; Silvioli, L.; Rossmeisl, J.; Arenz, M. EnhancedOxygen Reduction Reaction on Fe/N/C Catalyst in Acetate BufferElectrolyte. ACS Catal. 2019, 9, 3082−3089.(59) Kramm, U. I.; Herranz, J.; Larouche, N.; Arruda, T. M.;Lefevre, M.; Jaouen, F.; Bogdanoff, P.; Fiechter, S.; Abs-Wurmbach,I.; Mukerjee, S.; Dodelet, J. P. Structure of the catalytic sites in Fe/N/C-catalysts for O2-reduction in PEM fuel cells. Phys. Chem. Chem.Phys. 2012, 14, 11673−11688.



12380


(60) Zitolo, A.; Goellner, V.; Armel, V.; Sougrati, M. T.; Mineva, T.;Stievano, L.; Fonda, E.; Jaouen, F. Identification of catalytic sites foroxygen reduction in iron- and nitrogen-doped graphene materials.Nat. Mater. 2015, 14, 937−942.(61) Sa, Y. J.; Seo, D. J.; Woo, J.; Lim, J. T.; Cheon, J. Y.; Yang, S. Y.;Lee, J. M.; Kang, D.; Shin, T. J.; Shin, H. S.; Jeong, H. Y.; Kim, C. S.;Kim, M. G.; Kim, T. Y.; Joo, S. H. A General Approach to PreferentialFormation of Active Fe-Nx Sites in Fe-N/C Electrocatalysts forEfficient Oxygen Reduction Reaction. J. Am. Chem. Soc. 2016, 138,15046−15056.(62) Lefevre, M.; Proietti, E.; Jaouen, F.; Dodelet, J. P. Iron-basedcatalysts with improved oxygen reduction activity in polymerelectrolyte fuel cells. Science 2009, 324, 71−4.(63) Shinagawa, T.; Garcia-Esparza, A. T.; Takanabe, K. Insight onTafel slopes from a microkinetic analysis of aqueous electrocatalysisfor energy conversion. Sci. Rep. 2015, 5, 13801.(64) Reis, R. M.; Beati, A. A. G. F.; Rocha, R. S.; Assumpca̧õ, M. H.M. T.; Santos, M. C.; Bertazzoli, R.; Lanza, M. R. V. Use of GasDiffusion Electrode for the In Situ Generation of Hydrogen Peroxidein an Electrochemical Flow-By Reactor. Ind. Eng. Chem. Res. 2012, 51,649−654.(65) Nicoll, W. D.; Smith, A. F. Stability of Dilute Alkaline Solutionsof Hydrogen Peroxide. Ind. Eng. Chem. 1955, 47, 2548−2554.(66) Cota, H. M.; Katan, T.; Chin, M.; Schoenweis, F. J.Decomposition of Dilute Hydrogen Peroxide in Alkaline Solutions.Nature 1964, 203, 1281−1281.

■ NOTE ADDED AFTER ASAP PUBLICATIONDue to a production error, this paper was published on theWeb on July 29, 2019, with the incorrect artwork for Figure 3.Figure 3 was replaced, and the corrected version was repostedon July 30, 2019.



12381

activity–selectivity trends in the electrochemical ... · dispersed metal cations (m−n−c) are...

Documents