DOI:10.1002/cssc.201801583 Communications

Structure, Activity,and Faradaic Efficiency of Nitrogen- Doped Porous Carbon Catalysts for Direct Electrochemical Hydrogen Peroxide Production Yanyan Sun,Shuang Li, Zarko Petar Jovanov,Denis Bernsmeier,Huan Wang, Benjamin Paul, Xingli Wang, Stefanie Kühl, and Peter Strasser*[a]

Carbon materials doped with nitrogen are active catalysts for vantages. These include simplified and in situ small-scale pro- the electrochemical two-electronoxygen reduction reaction duction, reduced energy input and waste generation,aswell (ORR) to hydrogen peroxide. Insights into the individual role of as improved process safety compared with methods using a the various chemical nitrogen functionalitiesinthe H2O2 pro- direct explosive mixture of hydrogen and oxygen.Moreover,if duction, however,haveremained scarce. Here, we explore a the two- ORR process is coupled with water electro- catalytically very active family of nitrogen-dopedporous lyzers powered by renewable power sources, chemical electric- carbon materials, prepared by direct pyrolysis of ordered mes- ity storage is intimately coupledwith the production of useful oporous carbon (CMK-3)with polyethylenimine (PEI). Voltam- chemicals.[13] Based on the considerations above, it is of para- metric rotating ring-diskanalysis in combination with chro- mountimportance to designand synthesize the ORR catalysts noamperometric bulk electrolysis measurements in electrolysis with high catalytic activity and faradaic efficiency towardH2O2 cells demonstrate apronounced effect of the applied poten- production. tials, current densities, and electrolyte pH on the H2O2 selectivi- To date, the proposed alternative ORRcatalysts for H2O2 pro- [15,17, 22–27] ty andabsoluteproduction rates. H2O2 selectivity up to 95.3% ductioninclude noble-metal-based alloys, transition- [28–32] was achieved in acidic environment, whereas the largest H2O2 metal oxides/complexes, transition-metal/C compo- 1 1 [33–35] [6,13,16,18,36–38] production rate of 570.1 mmolgÀ catalyst hÀ was observed in sites, and metal-free carbon-based materials. neutralsolution. X-ray photoemission spectroscopy (XPS) anal- Amongthem, metal-free carbon-based materials provedre- ysis suggests akey mechanistic role of pyridinic-N in the cata- peatedlytobethe most promising alternative ORR catalysts lytic process in acid, whereas graphitic-N groups appear to be for H2O2 production, largely owing to their low-cost, facile catalytically active moieties in neutraland alkaline conditions. preparation, favorable selectivity,and strongchemical stability. Our resultscontribute to the understanding and aid the ration- Considerable effortshave been devoted to the design and al design of efficient carbon-based H2O2 productioncatalysts. screening of avariety of nanostructured carbon-based materi- als with different heteroatoms such as nitrogen or sulfur.More specifically,nitrogen-doped porous carbon materials have re- The electrochemical oxygen reduction reaction (ORR) is akey ceived the most attention recently,owing to their high specific reactionfor many renewable energy conversion and storage surfacearea andelectronic conductivity,good catalytic activity, technologiessuch as rechargeable fuel cells and metal–air bat- and durability.Todate, reported synthetic strategies toward ni- teries.[1–6] Most catalyst discovery research is focused on four- trogen-doped porous carbon materials take one of two ap- electron ORR catalysts to maximize the energy conversion effi- proaches:(1) direct synthesis in the presence of carbon and ni- ciency of electrochemical energy conversion/storage devi- trogen sources through pyrolysis,[11,18,39–40] chemical vapor de- ces.[7–12] In contrast, the exploration of two-electron ORR cata- position,[41] or the newly developed electrochemical doping;[42] lysts haslong been overlooked, despite the fact that it repre- (2) nanocasting strategies using commercial SiO2 as asacrificial sents apromisingand green route for in situ electrochemical hard template with subsequent pyrolysis.[38,43–44] Although [13–19] H2O2 production. Compared with the state-of-the-artin- some promisingresultshave been achieved, there is plenty dustrial anthraquinone process[20] or the difficult and potential- room for improvement in terms of selectivity.Onone hand, ly unsafe directcatalytic synthesisfrom hydrogen and the activity and faradaicefficiency of the H2O2 producing cata- [21] oxygen, electrochemical H2O2 production proceeding lysts have remained too low for commercialization and need through two-electron oxygen reduction possesses many ad- furtherimprovement beforedeveloping practically viable devi- ces. On the other hand, there are anumber of important fun- damental aspects aboutthe H O formation process that have [a] Y. Sun, S. Li, Dr.Z.P.Jovanov,Dr. D. Bernsmeier,Dr. H. Wang, B. Paul, 2 2 X. Wang, Dr.S.Kühl,Prof. P. Strasser remained elusive:for instance, the understanding of the cata- Department of Chemistry,Chemical Engineering Division lyticallyactive sites has remained unclear,inpart owing to the Technical University of Berlin inevitable co-generation and co-existenceofchemically dis- 10623 Berlin (Germany) tinct nitrogen species(N-functionalities) within the resultant ni- E-mail:[email protected] trogen-doped carbon materials and the lack of their direct ex- Supporting Information and the ORCID identification number(s) for the [1–2, 45] author(s) of this article can be found under: perimental observations. To remedy this, more controlled https://doi.org/10.1002/cssc.201801583. synthesis methods are needed where only one type of N-func-

ChemSusChem 2018, 11,3388 –3395 3388  2018 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Communications tionality forms in order to establish structure–selec- tivity relationships;orelse, analysistechniques are required that allow deconvolution of the contribu- tions of the chemically distinct N-functionalities and their carbon environments in parallel. Once the cat- alytically active site or chemical motif is identified, this would prompt synthetic efforts to maximize their population (maximizing the active site densi- ty), which would boost H2O2 productionactivity and faradaic efficiency. In the present work, we have developedafacile synthetic methodfor preparingnitrogen-doped porouscarbon catalysts. We investigated the effect of applied potentials and electrolyte pH on the

H2O2 selectivity and absolute H2O2 productionyields by means of the rotatingring-disk (RRDE) and chronoamperometric bulk electrolysis tests at stationary planarelectrodes. To link the catalytic rates and efficiencies to the chemical and morpho- logical structure of the catalysts in different electro- lytes, adetailed chemical speciation of nitrogen functionalities before andafter the catalysis was conducted by X-ray photoemission spectroscopy (XPS). The nitrogen-doped porous carbon catalysts were prepared by ultrasonically mixing ordered mesoporouscarbons (CMK-3) with calculated amountsofpolyethylenimine (PEI) followed by an- nealing in N2 atmosphere. Here, PEI was chosen as the nitrogen precursor mainly because of its low Figure 1. TEM images of (a) CMK-3 and (b) PEI50CMK3_800T.(c) N2 adsorption/desorption cost and high nitrogen content up to 32.5%. Atypi- isotherms, (d) pore size distribution, and (e) Raman spectra of CMK-3 andPEI50CMK3_ 800Twith their characteristic D/G band intensity ratios. (f) The deconvolution of the N1s cal synthetic recipe involves mixingthe CMK-3 spectra of PEI50CMK3_800T with different Nspecies as shown. carbon precursor with 50 mLPEI aqueous solution (30%w/v) followed by annealing at 8008C(the re- 2 1 sulting catalystwill henceforth be denoted as PEI50CMK3_ 1682 m gÀ ,which werebeneficialfor exposing more catalyti- 800T). The morphologies of CMK-3 and PEI50CMK3_800T were cally active sites and improved mass transport during the ORR first characterizedbytransmission electron microscopy (TEM), process.[37,44,46] as shown in Figure 1a,b.After nitrogen doping, the ordered The crystallinity of nitrogen-doped porous carbon catalysts porousstructure was well retained, therebyfacilitating oxygen was investigated by powder X-ray diffraction (PXRD). For CMK- mass transport during the H2O2 productionprocess. During the 3, there are two major diffraction peaks at 248 and 42.98 (Fig- formation process of nitrogen-doped carbon,PEI filled the ure S2ainthe Supporting Information), which can be ascribed pores of CMK-3 owing to capillarysiphons and then carboniza- to be the (002)and (101)crystal planes of graphite. All cata- tion occurred, forming chemical nitrogen-doped carbon func- lysts exhibited abroad diffraction peak, revealing characteristic tionalities of CMK-3. The porousproperties and Brunauer– features of amorphous carbon (Figure S2 in the SupportingIn- Emmett–Teller (BET) surfaceareas were then evaluated by ni- formation). Raman spectroscopy was carriedout to assess the trogen adsorption/desorption isotherms. Aclear hysteresis graphitization degree and defectcontent in nitrogen-doped loop at relative pressure range between 0.4 and 0.8 and a carbon catalysts. To achieve this, we evaluated peaks at 1 1 narrow pore size distribution centered around5nm for both 1330 cmÀ and 1585 cmÀ for all catalysts (Figure 1e and Fig- CMK-3 and PEI50CMK3_800T can be observed (Figure 1c,d), ure S3 in the Supporting Information), which are attributed to confirmingthe mesoporousstructures. In addition, asimilar the Dand Gbands of carbon materials, respectively.Generally, phenomenon was also observed forothernitrogen-doped the intensity ratio of the Dband to Gband (ID/IG)isaqualita- carbon materials prepared at different conditions (Figure S1 in tive estimateofthe defect content in the doped carbon, as the Supporting Information). An exception wasthe sample the Dand Gbands are related to disordered and ordered crys- prepared at 9008C, which demonstrated acertaindegree of talline sp2 carbon domains, respectively.[13] Pyrolytic nitrogen microporous structure.Moreover,all nitrogen-doped carbon doping led to slightly increased ID/IG values, implying an in- catalysts retained extremely high BET surface areas from 726 to creaseinstructuraldisorder and defects. In addition, the slight

ChemSusChem 2018, 11,3388 –3395 www.chemsuschem.org 3389  2018 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Communications downshifts of the Gband after pyrolytic nitrogen doping sug- well (Figure 2b,Figure S4band Figure S5binthe Supporting gest that doping alteredboth the crystallinity and the elec- Information). After nitrogen doping, the H2O2 selectivity im- tronic structure of carbon nanostructures likely owing to are- provedsharply from 77.2%to97.4%, and the corresponding ductioninthe meansheet width of un-doped graphene.[47–48] averaged number of transferredper oxygen molecule

X-ray photoelectron spectroscopy (XPS) wasalso performed decreased from 2.46 to 2.05 at +0.3 VRHE,close to the theoreti- to analyze the surface composition and electronic states of the cal value of 2for H2O2 selectivity.Inparticular,the PEI50CMK3_ nitrogen-doped carbon materials. Figure 1f shows the high- 800T catalyst achieved the highest H2O2 selectivity of 95.2– resolution N1sspectra of the nitrogen-doped carbon catalysts, 98.5%within the potential range from +0.1 to + 0.4 VRHE, which can be further divided into five peaks assigned to differ- which outperformed previously reported nitrogen-doped ent nitrogen speciesincluding pyridinic-N (398.3 eV), pyrrolic-N carbon[37–40] and noble-metal-based alloys catalysts (Table S4 in (399.9 eV), quaternary-N (400.9 eV), graphitic-N (402.1 eV), and the Supporting Information).[15,17,22–25, 53] N-oxide (404.1 eV).[49–52] The detailed content of carbon and ni- In addition to the optimization of catalyst preparation, we trogen, and different nitrogenspeciesare summarized in studied the effects of electrolyte pH on electrochemical H2O2 Table S1 and Table S2 (in the Supporting Information). Thevar- production of the PEI50CMK3_800T catalyst (Figure3). As iations in the Ncontent of the resultant catalysts are consistent showninFigure 3a,the onset potential (defined as the poten- 2 [54] with the results from element analysis(EA;Table S3 in the Sup- tial at the current density of 0.01 mA cmÀ ) of oxygen reduc- porting Information), indicating that the nitrogen dopant tion decreased in the order of KOH (0.84 VRHE) > K2SO4 amount can be tuned within the range from 0.07 %to2.04 %. (0.52 VRHE) > H2SO4 (0.49 VRHE)with the highest oxygen trans- The electrochemical activity and selectivity of the nitrogen- port-limited disk current and ring current obtained in 0.1m doped porous carboncatalysts toward ORR to H2O2 production KOH. In alkaline conditions, the Pt ring currentexhibited a was evaluated by using athree-electrode rotating ring-disk maximum HO2À production at + 0.5 VRHE of disk potential, electrode (RRDE) technique. Unlike pristine CMK-3, the N-modi- below which the HO2À production decreased gradually with a fied PEI50CMK3_800T catalystexhibitedrelatively positive negative shift of disk potentials. The effect of the nature of the onset potential, significantly increased disk currents,and ring electrolyte on the H2O2 selectivity is shown in Figure 3b.The current at the investigated range of applied potential (Fig- highest H2O2 selectivity was observed in 0.5m H2SO4,and de- ure 2a), which was also observedfor the other nitrogen-doped creasedinthe order of H2SO4 (98.5 %) > K2SO4 (89.8%) > KOH carbon catalysts prepared at different conditions (Figure S4a (83%) at +0.35 VRHE of disk potential. This trend is rationalized and Figure S5ainthe Supporting Information). This demon- by the establishment of the parallel four-electron pathway at strates that nitrogen doping resulted in an enhanced oxygen smaller potentials. reduction activity coupled with improved H2O2 selectivity. Tafel plots were evaluated from the experimental RRDE data

Quantitative H2O2 selectivities (in %) and the number of elec- in different electrolytes to verify the influence of pH on the trons transferred duringthe ORR process were evaluated, as ORR mechanism (Figure3c). Here, the kinetic current densities

(jdisk,kin)can be calculated accordingtothe Koutecky– Levich equation [Eq. (1)]:

1=j 1=j 1=j 1 disk ¼ disk; kin þ disk; lim ð Þ

where jdisk and jdisk,lim are the measured current densi- ty and the diffusion-limited currentdensity,respec- [35] tively. In addition, jdisk,lim can be obtainedbyusing the Levich equation [Eq. (2)]:

2=3 1=6 1=2 j 0:62nFD vÀ w C 2 disk; lim ¼ ð Þ

where C, F, D, n, n,and w refer to the bulk concen- 6 3 tration of oxygen (1.2 ”10À molcmÀ for pH 5–13 6 3 and 1.1”10À mol cmÀ for pH 0.3), the Faraday con- 1 stant (96 485 CmolÀ ), the diffusion coefficient of 5 2 1 oxygen (1.9 ”10À cm sÀ for pH 5–13 and 1.8” 5 2 1 10À cm sÀ for pH 0.3), the number of electrons 2 1 Figure 2. (a) Linear sweepvoltammetry performed by the rotatingring-disk electrode transferred, the kinetic viscosity (0.01 cm sÀ ), and (RRDE)techniquewhere the ring current is collectedonthe Pt ring at aconstant poten- the angularvelocity of the disk (w =2pN, N is the [35,55–57] tial of 1.2 VRHE,and (b) calculated H2O2 selectivity (%) and the number of electrons trans- linear rotation speed). The corresponding ferred (n)asafunction of electrodepotential for two materials:CMK-3 and PEI50CMK3_ 1 Tafel slopes can be calculated to be 161.3 mV decÀ 800T.Inall cases, measurements were performed in O -saturated 0.5m H SO at ascan 2 2 4 1 1 À rate of 5mVsÀ with rotatingspeed of 1600rpm at roomtemperature and the catalyst for pH 0.3, 95.7 mVdec for pH 7.0, and 2 1 loadingamountwas set to 0.05 mg cmÀ . 67.5 mV decÀ for pH 13.0, which are in accordance

ChemSusChem 2018, 11,3388 –3395 www.chemsuschem.org 3390  2018 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Communications

electrolytes, including 0.5m H2SO4 (pH 0.3), 0.1m

K2SO4 (pH 7), and 0.1m KOH (pH 13), whereby H2O2 concentration was determined by using UV/Vis ab- sorptionspectrometry.Figure4a,c,e plotsthe poten-

tiostatic H2O2 production normalized by the catalyst amount over the reactiontime at differentapplied

potentials of +0.3, +0.2, and + 0.1 VRHE.Asshown,

the H2O2 production rates increased with negative shifts of the appliedpotentialinboth acidic andal- kaline solutions.Inneutral conditions, the optimal

appliedelectrode potential was +0.2 VRHE.The H2O2

productionrate at + 0.2 VRHE decreased in 1 1 the order of K2SO4 (570.1 mmolgÀ catalyst hÀ ) > 1 1 KOH (345.5 mmol gÀ catalyst hÀ ) > H2SO4 1 1 (34.1 mmolgÀ catalyst hÀ ), indicating that the pH has a

significant effectonthe H2O2 production. Fig-

ure 4b,d,f reports the faradaic H2O2 efficiency:good

faradaic H2O2 efficiencies of 40–59 %were observed

in 0.5m H2SO4 (pH 0.3), highervalues of 37–65 %

Figure 3. Electrochemicalmeasurement of the optimized sample PEI50CMK3_800T in dif- were obtained in 0.1m K2SO4 (pH7), whereas some- ferent electrolytes:(a) linear sweep of RRDE measurement with ring current what highervalues of 51–67 %were measured in collected on the Pt ring at aconstant potential of 1.2 VRHE,and the disk geometric cur- 2 0.1m KOH (pH 13). Clearly, the long-term faradaic rent density (geometric area is 0.2475 cm ); (b) H2O2 selectivity (%);and (c) Tafel plots. H2O2 efficiency differed significantly from the H2O2 Note:the measurement was conducted in O2-saturated 0.5 m H2SO4 (pH 0.3), 0.1m K2SO4 1 (pH 7), or 0.1m KOH (pH 13) at ascan rate of 5mVsÀ with the rotating speed of selectivity obtained from voltammetric scans of the 2 1600 rpm at room temperature and the catalystloading amount is 0.05 mgcmÀ . RRDE method.

The differencebetween long-term faradaic H2O2

efficiency and the RRDE-derivedH2O2 selectivity is with the previously reported values for nitrogen-doped likely due to the longer residence times of H2O2 during bulk carbon-based ORR catalysts.[18] These resultsdemonstrate that electrolysis, resulting in its gradualchemical decomposition by 1 the ORR mechanism on the evaluated catalysts exhibits strong disproportionation (H2O2 H2O+ =2 O2)orfurther electrochem- !+ [24] pH-dependence, which affects the overall activity and selectivi- ical reduction (H O +2H +2eÀ 2H O). To validate this 2 2 ! 2 ty for peroxide production. hypothesis, the chemical disproportionation and electrochemi-

To evaluate the H2O2 production of the N-modified carbons cal reduction of H2O2 were carriedout and studied for the over extended periods of time, bulk electrolysis tests were con- PEI50CMK3_800T catalyst in the different electrolytes contain- 1 ducted in acustom-made two-chamber H-cell with different ing 2.5 mgLÀ H2O2 and 1mm H2O2 (Figure S6 in the Support-

Figure 4. (a,c,e) H2O2 production amountnormalized by the catalyst amountover the reaction time. (b, d,f) FaradaicH2O2 efficiency (FE)ofPEI50CMK3_800T catalyst in O2-saturated0.5m H2SO4 (pH 0.3) (a, b), 0.1m K2SO4 (pH 7) (c,d), and 0.1m KOH (pH 13) (e,f).

ChemSusChem 2018, 11,3388 –3395 www.chemsuschem.org 3391  2018 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Communications ing Information). As expected, disproportionation of H O was 2 2 Table 1. The nitrogen species change of PEI50CMK3_800Tbefore and observedinalkaline solution, whereas little to no H2O2 chemi- after the catalysis in different electrolytes, including 0.5m H2SO4,0.1 m cal decomposition was observed in acidic and neutral solutions K2SO4,and 0.1 m KOH. (Figure S6ainthe Supporting Information). Also, the PEI50CMK3_800T catalystexhibitedelectrochemical activity Nspecies Before [%] After [%] acidic neutralalkaline toward H2O2 reduction in neutraland alkaline solutions where- as no significant electrochemical reduction could be detected Pyridinic-N 17.2 12.13 26.37 18.95 Pyrrolic-N 22.09 28.39 30.49 43.12 in acidic solution (Figure S6binthe Supporting Information). Quaternary-N 24.12 25.28 24.59 24.92 In addition, the PEI50CMK3_800T catalystalso displayed excel- Graphitic-N 20.35 22.01 7.86 6.92 lent electrochemical stabilityatdifferent applied potentials N-oxide 16.23 12.19 10.68 6.09 over awide range of pH values during the successive electro- chemicalH2O2 production within 4h(Figure S7 in the Support- ing Information). These resultsdemonstrated that the itic-N during the ORR process.[45] This is why generally lower PEI50CMK3_800T catalystisavery promising and efficient cata- pyrrolic-N contentisexpectedafter the catalysis. However,the lyst for electrochemical H2O2 production. experiments revealed that the pyrrolic-N contentactually in- To gain more insight into the chemical stability and possibly creased after the catalysis in acidic,neutral, and alkaline solu- related mechanistic roles of the different chemical states of ni- tions. trogen during the two-electronoxygen reduction process, we To rationalize the changes in the relative content of the N investigated the quantitative changes in the chemical nitrogen species consistently,welookedatthe concomitant changes in speciesbefore and after the catalysis by using XPS. Figure 5a– their chemical environment during the ORR process in more dillustrates the XPS peak analysis of the various N-functionali- detail. Recent work has revealed that during oxygen reduction, ties before and after the catalysis at pH 0.3, pH 7, and pH 13. adsorbed oxygenated intermediates, such as Oads or OHads,may The changes of Ntypes after reaction are found to be depen- also chemically modify the immediate chemical environment dent on the electrolyte pH as significantly different changes in of the catalytically active site or structural motif of the cata- the pyridinic-N and graphitic-N contentsbefore and after the lysts.[1,45,58] With the covalentattachment of an OH group to catalysis are observed in acidic,neutral, and alkaline solutions the carbon atoms neighboring apyridinic-N, apyridonic-N spe- (Table 1). In acidic conditions, the pyridinic-N content de- cies is formed (Figure S9 in the Supporting Information), the creased, whereas graphitic-N slightly increased. In neutral and N1sbinding energy of which wasreported to upshift to alkaline solutions, the opposite is observed:the pyridinic-N 400.2 eV from the startingpyridinic-N energy of 398.8 eV, content slightly increased, whereas the graphitic-N content de- which is actually very close to the conventionally reported creased. Furthermore, pyrrolic nitrogen is generally known to binding energy of pyrrolic-N.[1,45] This is why we believe that be more sensitive to oxidationthan the pyridinic-N and graph- the decrease in pyridinic-N contentcoupled with the unex-

Figure 5. (a–d) The relative content change of various nitrogen species of the PEI50CMK3_800T catalystbefore(a) and after the catalysis in (b) pH 0.3, (c) pH 7, and (d) pH 13.

ChemSusChem 2018, 11,3388 –3395 www.chemsuschem.org 3392  2018 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Communications pected increaseinpyrrolic-N content after the catalysis in in the range from 108 to 908.Brunauer–Emmett–Teller (BET) surface acidic solution is due to the conversion of pristine pyridinic-N areas were measured with Quantachrome Autosorb-1-C equip- to pyridonic-N rather than being due to arearrangementof ment, and pore size distribution can be obtained from the adsorp- the Natoms.[45] In contrast, the attachment of -OH to the tion branches of the isotherms according to cylindrical pores non- localized density functional theory.The morphologies and struc- carbon atoms next to graphitic-N was reportedtoresult in the tures of the catalysts were analyzed by transmission electron mi- downshift of nitrogen binding energy to around 400 eV,which croscopy (TEM, FEI Tecnai G2 20 S-TWIN). TEM samples were pre- again coincides with the typically reported binding energy of pared by dispersing the resultant PEIXCMK3_YTinto ethanol solu- pyrrolic-N.[58] Thisoffers aplausible explanation for the de- tion and then dropping it onto the surface of aCugrid, finally crease in graphitic-N and increase in pyrrolic-N in neutraland drying in the oven at 60 8Cfor several minutes. The element analy- alkaline solutions. Thishypothesis further implies that pyri- sis of the catalysts including C, H, and Ncontents were conducted dinic-N has acritical role as acatalytically active site in acidic with aCHN elemental analyzer (PerkinElmer,Model 2400-II). X-ray solution,whereas graphitic-N is more likelythe catalytically photoelectron spectra (XPS) were determined by using an X-ray photoelectron spectrometer (K-Alpha,Thermo Scientific). Raman active site in neutral and alkaline solutions. In other words, spectra measurements were obtained with aBruker SENTERRA pyridinic-N does the catalysis in acidic solution, whereas graph- Raman system. itic-N plays an important role in neutral and alkaline solutions. In conclusion, nitrogen-doped porouscarbon catalysts were prepared by direct pyrolysis of CMK-3 and PEI. RRDE results Electrochemical characterization and chronoamperometric bulk electrolysis measurements dem- onstrated thatappliedpotentials and electrolyte pH have a The oxygen reduction performances of the resultant catalysts were evaluated in ahome-made electrochemical cell with three-elec- pronounced effect on the H2O2 selectivity and production rate over the optimal nitrogen-doped porous carbon catalysts. In trode configuration including the rotating-ring disk electrode (RRDE, PINE Research Instrumentation, 0.2475 cm2), Pt mesh, and a acidic solution,the highest H O selectivity of 95.3%at0.1 V 2 2 RHE mercury/mercurous sulfate electrode (MSE, PAR/ AMETEK USA, cali- can be achieved whereas in neutral solution the highest H2O2 bration against areversible hydrogen electrode (RHE) in 0.5m 1 1 production rate of 570.1 mmol gÀ hÀ can be obtained. catalyst H2SO4 resulted in areference electrode potential of +0.704 VRHE)as Analyses of the chemical state trajectories of nitrogen in the working, counter,and reference , respectively.Acom- carbon catalysts suggest asignificantroleofpyridinic-N in the mercial /galvanostat (Biologic SAS, model VSP 140) catalytic process in acid, whereas graphitic-N groups appear to was used. In alkaline solution, acalibrated commercial RHE (Hydro- be catalytically active moieties in neutral and alkaline condi- flex, Gaskatel) was employed. The catalyst film electrode was pre- tions. Our resultscontribute to our understanding of carbon- pared as follows. The homogeneous catalyst inks can be first ob- tained through dispersing the catalyst powder into amixture of based H O production catalysts and also aid the rational 2 2 water,isopropanol, and Nafion solution (5 wt%, Sigma–Aldrich) design thereof. with the assistance of ultrasonication. Then, the resultant ink was dropped on the surface of the RRDE and dried in an oven at 608C for 10 min. The catalyst amount on the was Experimental Section 2 fixed at 0.05 mgcmÀ .All measurements were repeated at least Chemicals three times to ensure reproducibility. Prior to the oxygen reduction measurements, successive cycle vol- CMK-3 was purchased from East High Tech Limited. Polyethyleni- tammetry (CV) was first performed to clean the Pt ring electrode in mine (PEI) was obtained from Sigma–Aldrich. Hydrogen peroxide N2-saturated 0.5m H2SO4 electrolyte within the potential range 1 test kits were acquired from Merck. All chemicals were used with- from +0.05 to +1.2 Vatascan rate of 20 mVsÀ until steady CVs out further treatment. were obtained. Then, the background current (capacitive current)

was collected by measuring CV in N2-saturated electrolyte within the potential range from +1.1 to +0.05 Vatascan rate of 1 Synthesis of nitrogen-doped mesoporous carbon 5mVsÀ .After that, electrochemical impedance spectroscopy (EIS) was also carried out to measure the uncompensated resistance (R) CMK-3 was first pre-treated in 2.4m HNO for 12 h, and dried at 3 of the liquid electrolyte for the subsequent iR correction. iR correc- 1008C. Then, acid-treated CMK-3 (30 mg) with different amounts of tion values ranged from afew to amaximum of 15 mV.For the H- PEI (25, 50, 100 mL) were dispersed in deionized water (2.5 mL) cell bulk electrolysis measurements, the uncompensated electro- with the assistance of grinding. The resultant mixture was an- lyte resistance was corrected at the 85%level to ensure the stabili- nealed in aquartz furnace under acontinuous flow of nitrogen gas ty of the electrochemical cell experiment, by using the Manual IR according to the following procedure:first heating to 1808Cfor compensation (MIR) technique (Biologic SAS EC Lab software). The 2h,and then heating to different annealing temperatures (600, 1 remaining fraction of the IR correction was neglected. For all acid 700, 800, and 9008C) for 4h with the rate of 58CminÀ .For sim- tests, the electrode potentials were converted to the relative to plicity,the resultant N-doped CMK-3 samples were labeled as RHE according to the following equation [Eq. (3)]: PEIXCMK3_YT(with X the PEI amount in microliters and Y is the an- nealing temperature in degrees Celsius). E RHE E MSE 0:704 iR V 3 ðÞ¼ðÞþþ À ðÞ ðÞ

Structural and composition characterization After purging with O2 for at least 30 min, linear sweep voltamme- try (LSV) was performed in O2-saturated electrolyte within the po- 1 Powder X-ray diffraction (PXRD) patterns were recorded with a tential range from +1.1 to +0.05 Vatascan rate of 5mVsÀ with

Bruker D8 Advance instrument with Cu Ka radiation (l=1.54056 Š) the rotation speed of 1600 rpm. The faradaic currents on the disk

ChemSusChem 2018, 11,3388 –3395 www.chemsuschem.org 3393  2018 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Communications electrode during the ORR process are obtained by subtraction of [1] D. Guo, R. Shibuya, C. Akiba, S. Saji, T. Kondo, J. Nakamura, Science 2016, 351,361 –365. the corresponding capacitive current. The H2O2 amount produced at the disk electrode was evaluated by using the ring current at a [2] H. B. Yang, J. Miao, S. F. Hung,J.Chen, H. B. Tao, X. Wang, L. Zhang, R. fixed ring potential of +1.2 V .The collection efficiency(N)ofthe Chen, J. Gao, H. M. Chen, L. Dai, B. Liu, Sci. Adv. 2016, 2,e1501122. RHE [3] A. Zitolo, N. Ranjbar-Sahraie,T.Mineva, J. Li, Q. Jia, S. Stamatin,G.F.Har- ring electrode was 37%. The H O selectivity (expressed as H O %, 2 2 2 2 rington, S. M. Lyth, P. Krtil, S. Mukerjee, E. Fonda, F. Jaouen, Nat. also referred to as the fraction of H2O2 produced) of the catalysts Commun. 2017, 8,957. and the number of electrons transferred (n)were calculated ac- [4] A. Zitolo, V. Goellner,V.Armel, M. T. Sougrati, T. Mineva, L. Stievano, E. [59] cording to the following equations [Eqs. (4)–(5)]: Fonda, F. Jaouen, Nat. Mater. 2015, 14,937–942. [5] H. Fei, J. Dong, Y. Feng, C. S. Allen, C. Wan, B. Volosskiy,M.Li, Z. Zhao, Y. H O % 200 I = N I I 4 Wang, H. Sun, P. An, W. Chen, Z. Guo, C. Lee, D. Chen,I.Shakir,M.Liu, T. 2 2 ¼ Â ring ð Â jjdisk þringÞ ð Þ Hu, Y. Li, A. I. Kirkland,X.Duan, Y. Huang, Nat. Catal. 2018, 1,63–72. Iring [6] J. Wang, Z. Huang,W.Liu, C. Chang, H. Tang, Z. Li, W. Chen,C.Jia, T. n 4 Idisk = Idisk 5 ¼ jjðjjþ N Þ ð Þ Yao, S. Wei, Y. Wu, Y. Li, J. Am. Chem. Soc. 2017, 139,17281 –17284. [7] N. D. Leonard, S. Wagner,F.Luo, J. Steinberg, W. Ju, N. Weidler,H. Wang, U. I. Kramm, P. Strasser, ACS Catal. 2018, 8,1640–1647. where Iring is the positive ring current and Idisk is the negative disk [8] X. Ge,A.Sumboja, D.Wuu, T. An, B. Li,F.W.T.Goh, T. S. A. Hor,Y.Zong, current. Note that Equation (4) reports H2O2 selectivity,and not effi- ciencies, because faradaic processes other than oxygen reduction Z. Liu, ACS Catal. 2015, 5,4643 –4667. [9] X. X. Wang, D. A. Cullen, Y. T. Pan, S. Hwang, M. Wang, Z. Feng, J. Wang, are not taken into consideration. M. H. Engelhard, H. Zhang, Y. He, Y. Shao,D.Su, K. L. More, J. S. Spende- Bulk H2O2 production was further evaluated by means of chro- low,G.Wu, Adv.Mater. 2018, 30,1706758. noamperometry at different applied working electrode potentials [10] P. Song, M. Luo, X. Liu, W. Xing,W.Xu, Z. Jiang,L.Gu, Adv.Funct. Mater. (+0.1, +0.2, and +0.3 VRHE)inacustomhome-made two-chamber 2017, 27,1700802. H-cell with apre-treated Nafion membrane as separator in acidic [11] S. Li, C. Cheng, X. Zhao, J. Schmidt, A. Thomas, Angew.Chem. Int. Ed. solution,[13] anion exchange membrane (Selemion ASV,AGC Engi- 2018, 57,1856 –1862; Angew.Chem. 2018, 130,1874–1880. neering Co.,LTD) in neutral solution, and Selemion AHO in alkaline [12] S. Li, C. Cheng, H. W. Liang, X. Feng, A. Thomas, Adv.Mater. 2017, 29, 1700707. solution. The influences of different electrolytes (0.5m H2SO4,0.1m K SO ,and 0.1m KOH) on H O production were also investigated. [13] Y. Y. Sun, I. Sinev,W.Ju, A. Bergmann,S.Dresp,S.Kuhl, C. Spori, H. 2 4 2 2 Schmies, H. Wang, D. Bernsmeier,B.Paul, R. Schmack, R. Kraehnert, B. UV/Vis photometric method was used to measure the produced RoldanCuenya, P. Strasser, ACS Catal. 2018, 8,2844 –2856. H2O2 amount at regular time intervals. Sample solution aliquots [14] S. Chen, Z. Chen,S.Siahrostami, T. R. Kim, D. Nordlund, D. Sokaras, S. were collected from the cathode compartment of the H-cell and Nowak, J. W. F. To,D.Higgins, R. Sinclair, J. K. Nørskov,T.F.Jaramillo,Z. analyzed by using acommercial hydrogen peroxide test kit (Merck Bao, ACS Sustainable Chem. Eng. 2017, 6,311 –317. KGaA, Germany). The faradaic H2O2 efficiency(FE) was calculated [15] Z. Zheng, Y. H. Ng, D. W. Wang, R. Amal, Adv. Mater. 2016, 28,9949 – from the H2O2 yield against the total quantity of charge passed 9955. [Eq. (6)]: [16] Y. Liu, X. Quan, X. Fan, H. Wang,S.Chen, Angew.Chem. Int. Ed. 2015, 54, 6837 –6841; Angew.Chem. 2015, 127,6941 –6945. [17] S. Yang, J. Kim, Y. J. Tak, A. Soon, H. Lee, Angew.Chem. Int. Ed. 2016, 55, H2O2 FE; % 2CVF=Q 6 ðÞ¼ ð Þ 2058 –2062; Angew.Chem. 2016, 128,2098 –2102. [18] D. Iglesias, A. Giuliani, M. Melchionna, S. Marchesan, A. Criado, L. Nasi, 1 where C is the H2O2 concentration (molLÀ ), V is the volume of M. Bevilacqua, C. Tavagnacco, F. Vizza, M. Prato,P.Fornasiero, Chem 1 electrolyte (L), F is the Faraday constant (CmolÀ ), and Q is the 2018, 4,106 –123. passed charge amount (C). The passed charge amount can be ob- [19] W. P. Mounfield, A. Garg, Y. Shao-Horn, Y. Romµn-Leshkov, Chem 2018, tained as the integral of current from t=0tot=t (time of polariza- 4,18–19. tion at agiven applied potential) according to the following equa- [20] J. M. Campos-Martin, G. Blanco-Brieva, J. L. Fierro, Angew.Chem.Int. Ed. t 2006, 45,6962 –6984; Angew.Chem. 2006, 118,7116–7139. tion: Q It; Eapplied dt. ¼ 0 [21] C. Samanta, Appl. Catal. A 2008, 350,133 –149. R ÀÁ [22] A. Verdaguer-Casadevall, D. Deiana, M. Karamad, S. Siahrostami, P. Mala- crida, T. W. Hansen,J.Rossmeisl, I. Chorkendorff, I. E. Stephens, Nano Lett. 2014, 14,1603 –1608. [23] J. S. Jirkovsky,I.Panas, E. Ahlberg, M. Halasa,S.Romani, D. J. Schiffrin, J. Acknowledgements Am. Chem. Soc. 2011, 133,19432–19441. [24] C. H. Choi, H. C. Kwon, S. Yook, H. Shin, H. Kim, M. Choi, J. Phys. Chem. C 2014, 118,30063 –30070. This work received funding from the German Federal Ministry of [25] S. Siahrostami, A. Verdaguer-Casadevall, M. Karamad, D. Deiana, P. Mala- Education and Research (Bundesministerium fürBildung und For- crida, B. Wickman, M. Escudero-Escribano, E. A. Paoli, R. Frydendal, T. W. schung, BMBF) under the grant 03SF0531B “HT-LINKED”. ZELMI of Hansen,I.Chorkendorff, I. E. Stephens,J.Rossmeisl, Nat. Mater. 2013, 12,1137–1143. Technical University Berlin is acknowledged for their supportwith [26] E. Pizzutilo, O. Kasian, C. H. Choi, S. Cherevko, G. J. Hutchings, K. J. J. TEM measurements. Yanyan Sun thanks the China Scholarship Mayrhofer,S.J.Freakley, Chem. Phys.Lett. 2017, 683,436–442. Council (CSC) for financial support. [27] C. H. Choi, M. Kim, H. C. Kwon, S. J. Cho, S. Yun, H. T. Kim, K. J. Mayrhof- er,H.Kim, M. Choi, Nat. Commun. 2016, 7,10922. [28] S. Siahrostami, M. E. Bjorketun, P. Strasser,J.Greeley,J.Rossmeisl, Phys. Conflict of interest Chem. Chem. Phys. 2013, 15,9326 –9334. [29] X.-H. Yan, Z. Meng, J. Liu, T. Xue, Mater.Lett. 2018, 217,171 –173. [30] M. A. Ghanem,A.M.Al-Mayouf,M.N.Shaddad, F. Marken, Electrochim. The authors declare no conflict of interest. Acta 2015, 174,557–562. [31] W. R. P. Barros, Q. Wei, G. Zhang, S. Sun,M.R.V.Lanza, A. C. Tavares, Electrochim. Acta 2015, 162,263–270. Keywords: electrolysis · faradaic efficiency · hydrogen [32] F. Xu, T. Song, Y. Xu, Y. Chen,S.Zhu, S. Shen, J. Rare Earths 2009, 27, peroxide · oxygen reduction · porouscarbon 128–133.

ChemSusChem 2018, 11,3388 –3395 www.chemsuschem.org 3394  2018 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Communications

[33] I. Yamanaka, S. Tazawa,T.Murayama, T. Iwasaki,S.Takenaka, ChemSu- [48] R. Lv,Q.Li, A. R. Botello-Mendez, T. Hayashi, B. Wang, A. Berkdemir,Q. sChem 2010, 3,59–62. Hao,A.L.Elias, R. Cruz-Silva, H. R. Gutierrez, Y. A. Kim, H. Muramatsu, J. [34] I. Yamanaka, R. Ichihashi, T. Iwasaki, N. Nishimura, T. Murayama, W. Zhu, M. Endo, H. Terrones, J. C. Charlier,M.Pan, M. Terrones, Sci. Rep. Ueda, S. Takenaka, Electrochim. Acta 2013, 108,321 –329. 2012, 2,586. [35] A. Bonakdarpour, D. Esau, H. Cheng, A. Wang, E. Gyenge, D. P. Wilkin- [49] T. Sharifi, G. Hu, X. Jia, T. Wagberg, ACS Nano 2012, 6,8904 –8912. son, Electrochim.Acta 2011, 56,9074 –9081. [50] W. He, C. Jiang, J. Wang, L. Lu, Angew.Chem. Int. Ed. 2014, 53,9503 – [36] K. Zhao, Y. Su, X. Quan,Y.Liu, S. Chen,H.Yu, J. Catal. 2018, 357,118– 9507; Angew.Chem. 2014, 126,9657 –9661. 126. [51] W. Ju, A. Bagger,G.P.Hao, A. S. Varela, I. Sinev,V.Bon,B.Roldan Cue- [37] F. HaschØ,M.Oezaslan, P. Strasser,T.-P.Fellinger, J. Energy Chem. 2016, nya, S. Kaskel, J. Rossmeisl, P. Strasser, Nat. Commun. 2017, 8,944. 25,251–257. [52] A. S. Varela, W. Ju, P. Strasser, Adv. Energy Mater. 2018,1703614. [38] J. Park,Y.Nabae, T. Hayakawa, M.-a. Kakimoto, ACS Catal. 2014, 4, [53] F. V. E. dos Reis, V. S. Antonin, P. Hammer,M.C.Santos, P. H. C. Camargo, 3749 –3754. J. Catal. 2015, 326,100 –106. [39] Y. Yang, F. He, Y. Shen, X. Chen, H. Mei, S. Liu, Y. Zhang, Chem. Commun. [54] J. Shui,C.Chen, L. Grabstanowicz, D. Zhao, D. J. Liu, Proc. Natl. Acad. 2017, 53,9994 –9997. Sci. USA 2015, 112,10629 –10634. [40] Y.-H. Lee, F. Li, K.-H. Chang, C.-C. Hu, T. Ohsaka, Appl. Catal. B 2012, 126, [55] Z. Gong, G. Zhang, S. Wang, J. Chem. 2013, 2013,1–9. 208–214. [56] G. Yang, W. Choi, X. Pu, C. Yu, EnergyEnviron. Sci. 2015, 8,1799–1807. [41] L. Qu, Y. Liu, J. B. Baek,L.Dai, ACS Nano 2010, 4,1321 –1326. [57] I. Kruusenberg, N. Alexeyeva,K.Tammeveski, Carbon 2009, 47,651 – [42] A. M. Abdelkader,D.J.Fray, Nanoscale 2017, 9,14548 –14557. 658. [43] V. Perazzolo, C. Durante, R. Pilot, A. Paduano, J. Zheng, G. A. Rizzi, A. [58] M. Scardamaglia, T. Susi, C. Struzzi, R. Snyders, G. DiSanto,L.Petaccia, Martucci,G.Granozzi, A. Gennaro, Carbon 2015, 95,949 –963. C. Bittencourt, Sci. Rep. 2017, 7,7960. [44] T. P. Fellinger,F.Hasche, P. Strasser, M. Antonietti, J. Am. Chem. Soc. [59] U. A. Paulus,T.J.Schmidt, H. A. Gasteiger,R.J.Behm, J. Electroanal. 2012, 134,4072 –4075. Chem. 2001, 495,134–145. [45] T. Xing, Y. Zheng,L.H.Li, B. C. Cowie,D.Gunzelmann,S.Z.Qiao, S. Huang,Y.Chen, ACS Nano 2014, 8,6856 –6862. [46] X. Liu, Y. Zhou,W.Zhou, L. Li, S. Huang, S. Chen, Nanoscale 2015, 7, 6136 –6142. Manuscript received:July 13, 2018 [47] Z. J. Han, C. Huang,S.S.Meysami, D. Piche, D. H. Seo,S.Pineda, A. T. Revised manuscript received:August 12, 2018 Murdock, P. S. Bruce, P. S. Grant, N. Grobert, Carbon 2018, 126,305 – Accepted manuscript online:August 13, 2018 312. Version of record online:September 5, 2018

ChemSusChem 2018, 11,3388 –3395 www.chemsuschem.org 3395  2018 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim