lable at ScienceDirect

Carbon 144 (2019) 259e268

Contents lists avai

Carbon

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

Sulfur bridges between Co9S8 nanoparticles and carbon nanotubesenabling robust oxygen electrocatalysis

Jialin Wang a, Hui Liu a, Yang Liu b, Wenhang Wang a, Qian Sun a, Xiaobo Wang a,Xinyu Zhao a, Han Hu a, Mingbo Wu a, *

a State Key Laboratory of Heavy Oil Processing, Institute of New Energy, College of Chemical Engineering, China University of Petroleum (East China),Qingdao, 266580, Chinab Department of Chemistry, Sungkyunkwan University (SKKU), Suwon, Republic of Korea

a r t i c l e i n f o

Article history:Received 24 October 2018Received in revised form27 November 2018Accepted 9 December 2018Available online 10 December 2018

* Corresponding author. Tel.: þ86 532 8698 3452.E-mail address: wumb@upc.edu.cn (M. Wu).

https://doi.org/10.1016/j.carbon.2018.12.0310008-6223/© 2018 Elsevier Ltd. All rights reserved.

a b s t r a c t

Electrocatalytic active species, for example, metal sulfides, have been widely combined with carbonnanomaterials for enhanced performance because of the synergetic effect arising from the interactionbetween different components. However, the binding between metal sulfides and carbon nanomaterials,which plays a key role in determining their interaction, has not been clearly elucidated. Herein, we re-ported the sulfur bridges between sulfur-doped carbon nanotubes (S-CNTs) and Co9S8 nanoparticles andtheir significant effect on the practical electrochemical performance. Compared to the physically mixedcomposites, the one with sulfur bridges between CNTs and Co9S8 shows much better performance interms of oxygen evolution (overpotential of 0.331 V at 10mA cm�2) and oxygen reduction (half-wavepotential of 0.810 V). Demonstrating that the sulfur bridges facilitate fast electron hopping fromconductive support to active species, thus contributing to the outstanding electrochemical performance.Moreover, a rechargeable Zn-Air battery based on our electrocatalyst with strong sulfur bridges deliversan outstanding performance in terms of ORR and OER bifunctional performances. The clear under-standing of the sulfur bridges between CNTs and Co9S8 is believed to shed substantial light on thedevelopment of other high-performance electrocatalysts.

© 2018 Elsevier Ltd. All rights reserved.

1. Introduction

The low cost and high activity of metal sulfides enable themoutstanding possibility as a promising alternative to precious metalfor electrocatalysis [1e3]. Nevertheless, the utilization of metalsulfides for electrocatalytic application is mainly prohibited bytheir low conductivity, easy agglomeration, and poor cyclic stabil-ity. To address these serious issues, one of the most effectivestrategy is to uniformly load the nanostructured metal sulfides onhighly conductive substrates via which the electron transfer can befacilitated and the agglomeration can be effectively suppressed[4e8]. As a result, the dramatically improved performance has beenwidely reported after rationally combining metal sulfurs with car-bon nanomaterials [9e13].

Among various kinds of carbon nanomaterials [14e17], carbonnanotubes (CNTs) have been proposed as one of themost promising

conductive substrates because of their intrinsically high conduc-tivity and excellent mechanical flexibility. Many kinds of metalsulfides with different sizes and morphologies have been blendedwith CNTs for enhanced oxygen evolution reaction (OER) and/oroxygen reduction reaction (ORR) performance [18e21]. Thisenhancement has been commonly ascribed to the ‘synergistic ef-fect’, which is believed to result from the good ‘interfacial inter-action’ between different components. Despite remarkableprogresses in utilizing the ‘synergistic effect’ achieved in the pastfew years [22e24], the efforts devoted to clearly elucidating theexact interaction mechanism are rare. As a result, it is urgentlyrequired to clarify the details behind the ‘synergistic effect’ and‘interfacial interaction’ both experimentally and theoretically, andthe good understanding of the interaction in atom level is supposedto be the key step in design and synthesis of electrocatalytsts in amore rational manner. The sulfur-doped CNTs (S-CNTs) are a pref-erable choice of substrate for loading metal sulfides [25,26].Therefore, the doped sulfurmay play a unique role in promoting theinterfacial interaction between CNTs and metal sulfides [27,28].

J. Wang et al. / Carbon 144 (2019) 259e268260

Herein, we reported the in-situ doping CNTs with sulfur andsimultaneous growth of Co9S8 nanoparticles on CNTs via a facileand efficient solvothermal process. The oxidized CNTs can afford agreat deal of active sites for metallic cation to anchor via electro-static attraction. The existence of sulfur bridges between S-CNTsand Co9S8 nanoparticles have been carefully elucidated throughstructure analysis. The composites afford an outstanding OER per-formance, and the overpotential of Co9S8/S-CNTs is 0.331 V at10mA cm�2, which is lower than 0.350 V of commercial RuO2.When evaluated as an electrocatalyst for ORR, Co9S8/S-CNTs de-livers a half-wave potential of 0.810 V, which is comparable to0.845 V of commercial Pt/C. The bifunctional catalytic performancesof Co9S8/S-CNTs are much better than many other metal sulfur andCNTs composites. Moreover, a rechargeable Zn-Air battery has beenconstructed using the Co9S8/S-CNTs as the electrocatalysts,showing a high open-circuit potential of ~1.472 V, and its peakpower density reach ~60.8mWcm�2, as well as long cycling sta-bility of 110 h. The increased performance is believed to arise fromthe strong sulfur bridges between S-CNTs and Co9S8 nanoparticles.

2. Experimental section

2.1. Materials

Carbon nanotube (CNTs), HNO3 (65%), Cobaltous Nitrate Hexa-hydrate (Co(NO3)2$6H2O), Trithiocyanuric acid Trisodillm salt(TMT-Na3) were purchased from Sinopharm Chemical ReagentCo.Ltd. Pt/C (20wt% Pt on Vulcan XC-72R), RuO2, and Nafion (5wt%)were purchased from Sigma-Aldrich. All chemicals were used asreceived without further purification.

2.2. Synthesis of CNTs

CNTs was added into HNO3 (65%) and the mixture was refluxedat 120 �C for 10 h, then the solution adequately washed withdeionized water, finally, through a freeze-drying process tocollected the CNTs.

2.3. Synthesis of Co9S8/S-CNTs hybrid

50mg CNTs was dispersed in 50mL of deionized water underultrasonication for 1 h to form a homogeneously dispersed solu-tion. Then, 0.1988 g cobaltous nitrate hexahydrate (Co(NO3)2$6H2O)was added to the solution, then the mixed solution was heated at80 �C for 30min under stirring. 0.2489 g TMT-Na3 was dissolved in10mLH2O andwas slowly added in the solution andmaintained for1h, then it was quickly transferred into a 80mL Teflon-lined auto-clave and heated at 180 �C for 12 h. Then, the solutionwas naturallycooled down to room temperature. The resulting solid material waswashed with deionized water and ethanol in turn, followed bylyophilization. The sample was heated to 900 �C with a heating rateof 5 �C min�1 and kept for 2 h under flowing N2, then naturallycooled to room temperature, Co9S8/S-CNTs hybrids were obtained.The calcination temperaturewas 900 �C and it also named as Co9S8/S-CNTs-900.

0.1988 g Co(NO3)2$6H2O and 0.2489 g TMT-Na3 were added toprepare Co9S8/S-CNTs, which was marked as Co9S8/S-CNTs-1. Forcomparison, we turning the amounts of reactants, half of the re-actants (0.0994 g Co(NO3)2$6H2O, 0.1245 g TMT-Na3) were added toprepare catalysts, which was marked as Co9S8/S-CNTs-0.5. Then,double reactants were added (0.3976 g Co(NO3)2$6H2O, 0.4979 gTMT-Na3) to prepare Co9S8/S-CNTs-2.

2.4. Electrochemical measurements

All the electrochemical measurements for the prepared mate-rials were performed using a CHI electrochemical workstation(Chenhua Company, Shanghai, China) with a conventional three-electrode system. The three electrodes included a glassy carbonelectrode (GCE) (4.00mm in diameter with the surface area of0.1256 cm2), a saturated calomel electrode (SCE), and a platinumwire, which were used as the working electrode, reference elec-trode, and counter electrode, respectively. The catalyst suspensionwas prepared by dispersing 2mg of Co9S8/S-CNTs-900 catalyst in800 mL of ethanol and 5 mL Nafion (0.5wt%) followed by ultra-sonication for 30min to generate a uniform catalyst ink. Then,15 mLof the as-prepared catalyst ink was dropped onto the surface of aGCE and dried at the room temperature, achieving a catalystloading of about 0.294mg cm�2 during all of the electrochemicalmeasurements and all the referential samples were in the sameloading with the catalyst. The electrolyte was 0.1M KOH for theORR and 1.0M KOH for the OER. Prior to electrochemical testing,the two electrolyte was purged with Ar and O2 for about 30min,respectively. Linear sweep voltammetry (LSV) polarization curveswith a scan rate of 5mV s�1 was carried out on a rotating diskelectrode at rotating speeds of 1600 rpm with a scan rate of5mV s�1. Cyclic voltammetry (CV) experiments were conducted ona rotating disk electrode (RDE) in the non-Faradaic region at thescan rate ranging from 20 to 140mV s�1 to evaluate the apparentelectrochemical double layer capacitance (Cdl).

2.5. Fabrication and test of Zneair batteries

The rechargeable Zneair battery test was performed in a home-made electrochemical cells, where catalyst coated on carbon paper(with a geometric area of 1 cm2) and Zn plate used as the aircathode and anode, respectively. The batteries were tested in 6MKOH electrolyte containing 0.2M Zn(CH3COO)2. Dispersing 2mgcatalysts into 800 mL of ethanol and 5 mL Nafion (0.5wt%) followedby ultrasonication for 30min. Then, some amounts of catalyst inkswas dropped on the carbon paper with an average loading amountof 1mg cm�2. The battery were tested in room environment on aLAND CT2001A. The galvanostatic discharge and charge cyclingtests were evaluated by a recurrent galvanic pulse method(10mA cm�2 for 15min).

2.6. Characterization

The X-ray diffraction (XRD, X'per PROMPD) with CuKa radiation(l¼ 1.5418 Å) was used to study the phase composition of thecatalysts. The morphology and size of the catalysts were evaluatedby scanning (SEM, S-4800), transmission (TEM, 7650B, Hitachi)electron microscopy and high resolution TEM. The XPS data wereperformed on an ESCALABMK II X-ray photoelectron spectrometerusing Mg as the exciting source to investigate the structural andchemical properties of the surface constituents. The Brunaur-er�Emmett�Teller surface area and porous structures were testedby the nitrogen adsorption� desorption isotherms of the sampleswere determined by an AutoChem II 2020. TGA analyses wereconducted using a WCT-2 (TA Instruments) under air atmospherewith a temperature ramp rate of 10 �C min�1.

3. Results and discussions

3.1. Composition and structure characterizations of Co9S8/S-CNTs

The synthesis process is clearly illustrated in Fig. 1. CNTs, Cobalt(II) nitrate hexahydrate (Co(NO3)2. 6H2O) and trithiocyanuric acid

Fig. 1. Schematic illustration for synthesis of Co9S8/S-CNTs catalyst. (A colour version of this figure can be viewed online.)

J. Wang et al. / Carbon 144 (2019) 259e268 261

trisodillm salt (TMT-Na3) were chosen as the precursor. CNTs werepreoxidized for defects introduction which would facilitate theCo2þ ions attraction and sulfur doping during the hydrothermalprocess [29]. Then, a thermal treatment at 900 �C for 2 h under N2atmosphere was conducted to produce Co9S8/S-CNTs.

The crystalline structure of Co9S8/S-CNTs was investigated byXRD, which is shown in Fig. 2a. The 2q diffraction peak around15.5�, 29.8�, 31.4�, 39.5�, 47.5�, 52.1�, 61.2� and 73.2� could beindexed to the (111), (311), (222), (331), (511), (440), (533) and (731)planes of cubic Co9S8 (PDF card no.03-065-6801) [30,31], whichindicates the formation of Co9S8 nanocomposites. The crystallinestructure of pure Co9S8 nanoparticles was also investigated, whichwas highly consistent with the standard crystalline structure ofCo9S8. The XRD characterization shows that the width of the peakgets widenwith the introduction of carbon. From Scherrer Formula,we can speculated that the crystal size of Co9S8 is much bigger than

Fig. 2. (a) XRD pattern, (b) SEM image, (c) TEM image, and (d) HRTEM image

Co9S8/S-CNTs, which can illustrate the addition of CNTs can alle-viate the aggregation of Co9S8 nanoparticles. In comparison withCo9S8, the diffraction peak had a little positive shift, which could beattributed to the crystal transfer and the interaction between Co9S8and S-CNTs.

The microstructure of Co9S8/S-CNTs was investigated by scan-ning electronmicroscopy (SEM). Themorphology (Fig. 2b) of Co9S8/S-CNTs clearly showed that Co9S8 nanoparticles are uniformlydecorated on S-CNTs with an average size of about 40 nm indiameter. Whereas, these particles seriously aggregated withoutthe introduction of CNTs as shown in Fig. S1b. Comparing Figs. S1(aand b) with Fig. 1b, the pure Co9S8 nanoparticles showed seriousaggregation and the size of the nanoparticles are not uniform.However, the Co9S8 nanoparticles decorated on S-CNTs show uni-form size.What's more, the Co9S8 nanoparticles are tightly attachedon the S-CNTs. It can be deduced that CNTs can efficiently prevent

of Co9S8/S-CNTs. (A colour version of this figure can be viewed online.)

J. Wang et al. / Carbon 144 (2019) 259e268262

the aggregation of Co9S8 nanoparticles and regulate the size ofCo9S8 nanoparticles.

Transmission electron microscopy (TEM) and high-resolutiontransmission electron microscopy (HRTEM) were employed tofurther investigate themicrostructure of Co9S8/S-CNTs. As shown inFig. 2c and d, Co9S8 nanoparticles are uniformly distributed on theS-CNTs. To investigate the influence of the adding amounts of theprecursors, the half amount of reactants (Co9S8/S-CNTs-0.5) and thedouble amount of reactants (Co9S8/S-CNTs-2) were produced. Inthe case of Co9S8/S-CNTs-0.5 (Fig. S2a), the Co9S8 nanoparticleswere sparsely embedded on the S-CNTs and their particle size aresmaller than that of Co9S8/S-CNTs. On the contrary, the seriousagglomeration of Co9S8 nanoparticles could be found in Co9S8/S-CNTs-2 (Fig. S2b), which demonstrate that proper amount of re-actants are beneficial to regulate the morphology. The HRTEMimage (Fig. 2d) clearly displays that the lattice fringes of 0.300 nmcorrespond well to the (311) planes of highly crystal Co9S8 [32].

The surface element distribution of Co9S8/S-CNTs was disclosedby energy dispersive spectra (EDS) element mappings in Fig. 3 (a,d). It is found that sulfur is not only present on the Co9S8 particlesbut also on the surface of the CNTs, which confirmthe presence ofcobalt sulfide and the successful sulfur doping into the CNTs.

The porosity and surface area of Co9S8/S-CNTs were furtherevaluated by N2 adsorption and desorption measurements. Asshown in Fig. 4a, the N2 adsorption and desorption isothermsbelong to type-IV curve with H4 hysteresis loop, which indicate theexistence of mesoporous structures. The Brunauer-Emmett-Teller(BET) surface area of Co9S8/S-CNTs was calculated to be104m2 g�1. Such a high specific surface area is beneficial to the fullexposure of the catalytic active sites. By contrast, S-CNTs has muchhigher BET surface area of 302.53m2 g�1, which is shown inFig. S3a. It can be concluded that the insertion of Co9S8 can blockthe pore structure of S-CNTs and reduce the relative BET surfacearea of S-CNTs. Furthermore, the pore size distribution curve ofCo9S8/S-CNTs (Fig. 4b) showed that the mesoporous are mainly

Fig. 3. (a) SEM image and (bed) the corresponding elemental mapping image

centered at 2.1 and 2.5 nm, and the mesoporous of S-CNTs aremainly centered at 2.4 nm (Fig. S3b).

Thermogravimetry (TG) was used to analyze the content ofdifferent components. As shown in Fig. 4c, the mild weight lossstage from 25 to 200 �C could be attributed to the removal of H2Oadsorbed on the surface of Co9S8/S-CNTs. The slight weight increasetook place from 200 to 500 �C could be attributed to the partoxidation of Co9S8 into CoSO4 [19]. At temperature higher than500 �C, the weight of residue decreased gradually due to thedecomposition of CoSO4 into Co3O4 and the oxidation of carboninto CO2. During the whole measurement range, the total weightloss was about 55wt%. When the temperature reached up to900 �C, the weight remained constant and the residue wasconfirmed to be Co3O4. Therefore, the original weight fraction ofCo9S8 was calculated to be 59.9wt% in Co9S8/S-CNTs.

To further investigate the interaction between Co9S8 and S-CNTs, the Raman spectroscopy was employed, which has beenproved to be an efficient method to manifest the intimate inter-action between the two composites (Fig. 4d). From the Ramanspectra of pure Co9S8, two strong bands at 333 and 396 cm�1 can beindexed to Co9S8. While, the position shifts of Co9S8/S-CNTs to 463and 656 cm�1 in the spectra can be clearly seen. The reason ofwhich is supposed to be the strong interaction between Co9S8 andS-CNTs. Co9S8/S-CNTs exhibits the characteristic D band(1323 cm�1), which is caused by structural defects or edges duringthe reaction. What's more, the G band (1586 cm�1) corresponds tothe first-order scattering of the E2g mode observed for sp2 carbonatoms. However, pure S-CNTSs shows two strong bands at1323 cm�1 (D band) and 1574 cm�1 (G band). Comparing withCo9S8/S-CNTs, the D band shows identical position, however, the Gband has a little negative shift, which can be attributed to theinteraction between the two composites. All these results clearlydemonstrate the intimated interaction between CNTs and TMCs.

The interaction of Co9S8/S-CNTs was further studied by X-rayphotoelectron spectroscopy (XPS). As shown in Fig. 5a, the

s of Co9S8/S-CNTs. (A colour version of this figure can be viewed online.)

Fig. 4. (a) Nitrogen sorption isotherm, (b) Pore-size distribution, (c) TGA curves of Co9S8/S-CNTs and (d) Raman spectra of pure Co9S8, S-CNTs and Co9S8/S-CNTs composite forcomparison. (A colour version of this figure can be viewed online.)

Fig. 5. (a) XPS spectrum of Co9S8/S-CNTs. The corresponding high resolution XPS spectra of (b) C 1s in Co9S8/S-CNTs and S-CNTs. (c) S 2p in Co9S8/S-CNTs, S-CNTs and Co9S8. (d) Co2p regions in Co9S8/S-CNTs and Co9S8. (A colour version of this figure can be viewed online.)

J. Wang et al. / Carbon 144 (2019) 259e268 263

elemental compositionwas in linewith the EDXmapping and it canbe clearly seen in Table S1. The C 1s spectra (Fig. 5b) could beresolved into four independent peaks at 284.6, 285.6, 286.7 and289.2 eV, which are related to C]CeC, CeS, CeO and C]O [33],

respectively. As shown in Fig. 5c, three different types of sulfurspecies could be seen in Co9S8. The binding energy around 161.9and 163.2 eV could be assigned to S 2p3/2 and S 2p1/2 of Co9S8[34,35]. In contrast, four different types of sulfur species could be

J. Wang et al. / Carbon 144 (2019) 259e268264

found in Co9S8/S-CNTs. The characteristic peak of CeSeC could beobserved at 165.2 eV, which demonstrates the successfully dopingof S into CNTs. The peak at 167.6 eV could be attributed to SO4

2�

species [33], which indicates the partial oxidization of sulfur on thesurface of the catalysts. A little positive shift of the binding energyof S 2p3/2 and S 2p1/2 could be seen, which illustrates the interactionof Co9S8 and S-CNTs. As shown in Fig. 5d, the Co 2p1/2 and Co 2p3/2peak could be clearly seen in Co9S8. In contrast, the strong peaks at780.9 and 781.9 eV could be assigned to Co 2p3/2 in Co9S8/S-CNTs.Moreover, the Co 2p1/2 peak could be found at 796.4 and 798.4 eV,demonstrating the existence of Co9S8 in S-CNTs. Furthermore, thebinding energy locating at 786.3 and 803.9 eV belong to theshakeup satellite peaks of Co [36,37]. The XPS results are inagreement with the previous reports [38]. All the positive shifts ofthe binding energy of S and Cowere caused by the electron transferbetween Co9S8 and S-CNTs, which could also demonstrate thestrong interaction between Co9S8 and S-CNTs.

3.2. Electrochemical performance of Co9S8/S-CNTs

Based on the structure analyses above, the strong interactionbetween Co9S8 and S-CNTs could be verified. Thus, let's seewhetherthe interaction between Co9S8 and S-CNTs can improve the effi-ciency of electrocatalytic reaction. The OER performance of pureCo9S8, S-CNT

Co9S8 mixed S-CNTs and Co9S8/S-CNTs were investigated, whichwas evaluated by the rotating disk electrode (RDE) technique at1600 rpm and at a scan rate of 5mV s�1 in Ar-saturated 1M KOHsolution, and commercial RuO2 was also tested in the same con-ditions. The linear scan voltammetry (LSV) curves can be seen inFig. 6a, pure Co9S8 and S-CNTs exhibited inferior OER catalytic ac-tivity with the overpotential of 0.537 and 0.538 V to deliver10mA cm�2, respectively. Furthermore, the Co9S8 mixed S-CNTsrequired a relatively low overpotential of 0.421 V to achieve the

Fig. 6. (a) OER polarization curves and (b) corresponding Tafel plots of Co9S8, S-CNTs, Co9S8þrate: 1600 rpm; sweep rate: 5mV s�1). (c) Electrochemical impedance spectroscopy of Co9Sshows equivalent circuit used to interpret the data. (d) Charging current density differenceCNTs. (A colour version of this figure can be viewed online.)

same current density, suggesting the physical mixture of Co9S8 andS-CNTs was able to promote the catalytic activity toward OER,which was consistent with the reported result [39]. In sharpcontrast, Co9S8/S-CNTs showed much superior OER activity andonly needed the overpotential of 0.331 V to deliver 10mA cm�2,which was even lower than that of commercial RuO2 (0.350 V). Thebest catalytic performance of Co9S8/S-CNTs can further demon-strate the promoting effect of the intimate interaction.

The OER electrocatalytic kinetics of these samples were furtherinvestigated by Tafel plots (Fig. 6b). Comparing with S-CNTs(124.5mV dec�1), Co9S8 (145.6mV dec�1) and Co9S8þS-CNTs(101.3 mV dec�1), Co9S8/S-CNTs shows the lowest Tafel slope(97.3mV dec�1). Even more, the Tafel slope of Co9S8/S-CNTs can becomparable to 88mV dec�1 of commercial RuO2 catalyst, whichindicates that Co9S8/S-CNTs has excellent catalytic performance forOER. The lower Tafel slope of Co9S8/S-CNTs hybrid reveals thehigher electrocatalytic activity [40]. The unique structure of Co9S8/S-CNTs accelerates the transfer of electrons and ions, and reducesthe free energy of the OER. Therefore, Co9S8/S-CNTs shows the bestOER catalytic activity and its overpotentional and Tafel slop aresuperior to most reported catalysts (Table S1), which is consistentwith the former results.

The electrochemical impedance spectroscopy measurement(100 kHze0.1 Hz) was used to better demonstrate the electrodeinterface and reaction kinetics (Fig. 6c). The Nyquist plots werefitted with the equivalent circuit model, which could be seen in theinset of Fig. 6c. Here, Rs represents the solution resistance of elec-trolyte, while the Rc represents the charge transfer resistanceduring OER. The good catalytic activity could be attributed to thelow resistance of charge transfer from the electron-rich S-CNTs toCo9S8. It could be concluded that Co9S8/S-CNTs has much smallercharge transfer resistance than those of Co9S8, S-CNTs and Co9S8þS-CNTs. The lower electron transfer resistance means the betterconductivity, which could be ascribed to the strong coupling

S-CNTs, Co9S8/S-CNTs, and RuO2 electrodes in Ar-saturated 1M KOH solution (rotation8/S-CNTs, Co9S8þS-CNTs, Co9S8 and S-CNTs at �0.238 V (100 kHz-0.1 Hz) and the inset(DJ ¼ (Ja � Jc)/2) plotted against scan rate of Co9S8/S-CNTs, Co9S8þS-CNTs, Co9S8 and S-

J. Wang et al. / Carbon 144 (2019) 259e268 265

between Co9S8 and S-CNTs, which can faster the process of chargetransformation.

To further illustrate the good catalytic activity of Co9S8/S-CNTs,the electrochemically active surface area (ECSA) was calculated(Fig. 6d). The ECSA is a criterion to measure the catalytic efficiencyand it is linearly proportional to the double-layer capacitances (Cdl)of the electrocatalysts. The cyclic voltammograms (CVs) measure-ments were performed in the potential range of 1.223e1.323 V (noelectrode rotation) at different scan rates ranging from 20mVS�1 to120mV S�1 (Fig. S5) [41]. The Cdl of Co9S8/S-CNTs is computed to be49.71mF cm�2, which is much higher than Co9S8 (0.69mF cm�2), S-CNTs (0.85mF cm�2) and Co9S8þS-CNTs (9.44 mF cm�2), and thehigher Cdl is more beneficial to the catalytic reaction.

The addition of reactants will influence the morphology andstructure of the products, and good morphology is critical inboosting the number of catalytic active sites. Therefore, the influ-ence of the adding amounts was investigated, which could be seenin Fig. S4a. Co9S8/S-CNTs-1 shows the best OER catalytic activitywith the need of 0.331V overpotential to deliver 10mA cm�2. Itshows much negative potential than Co9S8/S-CNTs-0.5 (0.379 V)and Co9S8/S-CNTs-2 (0.516 V). Comparing with Co9S8/S-CNTs-0.5and Co9S8/S-CNTs-2, the good catalytic activity of Co9S8/S-CNTs-1could be attributed to the less aggregation and the uniformdispersion of Co9S8 nanoparticles, which could be clearly seen inFigs. S2(a and b). Besides the good morphology, the appropriateproportion of the reactants is also an essential factor to improve thecatalytic efficiency.

Due to donating electrons to the carbon skeleton, the doped Scan greatly enhance the conductivity of S-CNTs, which is beneficialto the ORR process [10,42]. Transition metals have various valencesin transition-metal sulfides and multivalent metal compounds arepromoters in electrocatalysis reactions. Furthermore, they can

Fig. 7. (a) ORR polarization curves and (b) corresponding Tafel plots of Co9S8 nanoparticlessolution (rotation rate: 1600 rpm; sweep rate: 5mV s�1). (c) Stability of the Co9S8/S-CNTs andCNTs and RuO2þPt/C at 1600 rpm in 0.1 M KOH with a scan rate of 5 mV s�1. (A colour ve

afford donor�acceptor sites to facilitate the adsorption of oxygenand fasten the rate of electron transfer to accelerate the ORR elec-trocatalytic activity [43]. Consequently, the rotating disk electrode(RDE) measurement was used to examine the ORR electrocatalyticperformance of Co9S8/S-CNTs in Fig. 7a. The tests were carried outin 0.1M KOH electrolyte at a scan rate of 5mV s�1. Remarkably,Co9S8/S-CNTs shows the best ORR performance with a half-wavepotential of 0.810 V, which is more positive than those of S-CNTs(0.768 V), Co9S8 (0.424 V), Co9S8þS-CNTs (0.612 V). The perfor-mance of Co9S8/S-CNTs is comparable to the commercial Pt/C(0.845 V). The good catalytic activity could be attributed to thestrong interaction, fast electron-transfer rate and short electrondiffusion distance between Co9S8 and S-CNTs. Moreover, Co9S8/S-CNTs-1 shows much better ORR activity than those of Co9S8/S-CNTs-0.5 and Co9S8/S-CNTs-2. As exhibited in Fig. S4, Co9S8/S-CNTs-1 exhibits an E1/2 of 0.810 V, which is positive than Co9S8/S-CNTs-0.5 (0.651 V) and Co9S8/S-CNTs-2 (0.686 V) (Fig. S4b), which dem-onstrates that the loading amounts is also a crucial factor to in-fluence the ORR activity.

The corresponding Tafel slope is shown in Fig. 7b. The Tafel slopeof Co9S8/S-CNTs is �32.86mV dec�1, which is much lower than S-CNTs (�67.17mV dec�1), Co9S8 (�140.1mV dec�1), Co9S8þS-CNTs(�59.79 mV dec�1) and it is even lower than the commercial Pt/C(�57.23mV dec�1). What's more, its half-wave potential and Tafelslop is better than reported catalysts (Table S2).

The stability is also a critical criterion to estimate the perfor-mance of the catalysts. As exhibited in Fig. 7c, the long-term sta-bility of Co9S8/S-CNTs was examined by chronoamper-ometry (i-t)measurements. Co9S8/S-CNTs can maintain about 88.7% of theinitial current after 10 h continuous operation. In contrast, a sharpdecrease of 50.0% of the initial current was calculated for com-mercial Pt/C, which clearly demonstrate that Co9S8/S-CNTs has

, S-CNTs, Co9S8þS-CNTs, Co9S8/S-CNTs, and Pt/C electrodes in O2-saturated 0.1M KOHPt/C electrodes measured in 0.1M KOH. (d) Oxygen electrode performance of Co9S8/S-

rsion of this figure can be viewed online.)

J. Wang et al. / Carbon 144 (2019) 259e268266

excellent durability. Furthermore, the tolerance of Co9S8/S-CNTsand Pt/C toward methanol crossover was evaluated. As demon-strated in Fig. S6, when 3Mmethanol was added into O2-saturated0.1M KOH solution in the 300th second, Pt/C had an obviousdecrease of current density. However, Co9S8/S-CNTs electrodecatalyst exhibited almost no current density decrease, which indi-cate that Co9S8/S-CNTs catalyst has higher resistance to methanolcrossover than Pt/C.

Co9S8/S-CNTs exhibited prominent bifunctional OER and ORRperformance under alkaline conditions (Fig. 7d). As a bifunctionalcatalyst, the overall oxygen electrode activity could be evaluated bythe potential gap (DE¼ Ej¼10 - E1/2), (E1/2, half-wave potential andEj¼10, potential required at an current density of 10mA cm�2) ofOER and ORR metrics [44]. The smaller value of DE indicates thecloser of the catalyst to an ideal reversible oxygen electrode. The DEvalue of Co9S8/S-CNTs nanocomposites was calculated to be only0.751 V, which was only a little more than that of Pt/C þ RuO2catalyst (0.735 V), suggesting that Co9S8/S-CNTs is an ideal bifunc-tional catalyst.

3.3. Performance of rechargeable Zn-air battery

Owing to the excellent bifunctional catalytic activity, Co9S8/S-CNTs was used to be the cathode material of the conventionalrechargeable Zn-air battery. In order to demonstrate the practicalapplication of Co9S8/S-CNTs, the physical mixture of commercial Pt/C and RuO2 catalysts (Pt/C þ RuO2) were used to be the referencebattery. As shown in Fig. 8a, a Zn foil was used as the anode, thebifunctional electrocatalyst coated on carbon paper was employedas the metal-free cathode, and 6.0 M KOHþ0.2 M zinc acetate wereused as the liquid electrolyte. The battery was assembled asdescribed in the experimental section. Fig. 8 illustrates the

Fig. 8. (a) Schematic of a liquid rechargeable Zneair battery. (b) Open-circuit potential of alarization curves and the corresponding power density curves of Zn-air batteries of Co9S8/S-Con the air electrode was 1mg cm�2). (e) Galvanostatic dischargeecharge cycling (30min pdensity of 10mA cm�2. (f) Photograph of the Zn-air batteries and corresponding open-circl

performance of the Zn-air battery under real battery condition,which possesses a high open-circuit potential (1.472 V) (Fig. 8b andf). Fig. 8c gives the polarization plots and corresponding powerdensity curves of the two catalysts. Moreover, the discharge po-larization curves show that the peak power density of Co9S8/S-CNTscalculated to be 74.75mWcm�2 is outperforming than35.97mWcm�2 of precious metal catalysts Pt/C þ RuO2. Thedischarge and charge polarization curves of Co9S8/S-CNTs could beseen in Fig. 8d, which shows a charge and discharge potentials of1.52 V and 1.12 V at 20mA cm�2, respectively. The performance iseven better than the mixture of Pt/C þ RuO2 (1.97 and 1.04 V).Fig. 8e shows a good long-term cycling stability without obviousloss of both charge and discharge voltage over 110 h cycling test at aconstant current density of 10mA cm�2, which is much better thanPt/C þ RuO2. Therefore, the performance of Co9S8/S-CNTs is supe-rior to those of Pt/C þ RuO2 based Zn-air battery.

The origin of the outstanding bifunctional activity of Co9S8/S-CNTs is elucidated from the enhanced electrochemical surface-active area, the well-dispersed nanostructures, the in situ grownfeature on the highly conductive S-CNTs, and the strong interactionbetween Co9S8 and S-CNTs. The good stability can be attributed tothat S-CNTs can protect the Co9S8 from aggregating, improve theutilization of active sites and accelerate the transfer efficiency ofelectrons. All of these advantages favor the attractive electro-catalytic performance of Co9S8/S-CNTs.

4. Conclusion

In conclusion, bifunctional Co9S8/S-CNTs were prepared via afacile method. The mechanism and significance of the sulfurbridges between sulfur-doped carbon nanotubes (S-CNTs) andCo9S8 nanoparticles on the practical electrochemical performance

Zneair battery using the Co9S8/S-CNTs catalyst for the air cathode. (c) Discharge po-NTs and Pt/C þ RuO2. (d) Charge and discharge polarization curves (the catalyst loadinger cycle) of the Co9S8/S-CNTs Zneair battery at a constant dischargeecharge currente potential (OCP). (A colour version of this figure can be viewed online.)

J. Wang et al. / Carbon 144 (2019) 259e268 267

were clarified. A synergistic effect between Co9S8 nanoparticles andS-CNTs effectively improve the bifunctional electrocatalytic per-formance of the compounds. Owing to the excellent bifunctionalactivity of the Co9S8/S-CNTs catalyst, the as-made rechargeableliquid Zneair battery displays favorable performance. It is believedthat TMCs/carbon composites will have extensive application inelectrocatalytic reaction, super capacitors, lithium-ion batteries,and so on.

Acknowladgements

This work was supported by the National Natural ScienceFoundation of China (Nos. 51372277, 51572296, U1662113); TheFundamental Research Funds for the Central Universities(15CX08005A). The Financial Support from Taishan Scholar Project,Scientific Research and Technology Development Project of Petro-china Co., LTD (2016B-2004(GF))

Appendix A. Supplementary data

Supplementary data to this article can be found online athttps://doi.org/10.1016/j.carbon.2018.12.031.

References

[1] D. Xiong, Q. Zhang, S.M. Thalluri, J. Xu, W. Li, X. Fu, L. Liu, One-step fabricationof monolithic electrodes comprising Co9S8 particles supported on cobalt foamfor efficient and durable oxygen evolution reaction, Chemistry 23 (36) (2017)8749e8755.

[2] J. Bai, T. Meng, D. Guo, S. Wang, B. Mao, M. Cao, Co9S8@MoS2 coreeshellheterostructures as trifunctional electrocatalysts for overall water splittingand Zneair batteries, ACS Appl. Mater. Interfaces 10 (2018) 1678e1689.

[3] J. Yang, Y.U. Chang, X. Fan, S. Liang, S. Li, H. Huang, Z. Ling, C. Hao, J. Qiu,Electroactive edge site-enriched nickel-cobalt sulfide into graphene frame-works for high-performance asymmetric supercapacitors, Energy Environ. Sci.9 (4) (2016) 1299e1307.

[4] L. Jiao, Y.X. Zhou, H.L. Jiang, Metal-organic framework-based CoP/reducedgraphene oxide: high-performance bifunctional electrocatalyst for overallwater splitting, Chem. Sci. 7 (2016) 1690e1695.

[5] L. Yang, Z. Cai, L. Hao, L. Ran, X. Xu, Y. Dai, S. Pan, B. Jing, J. Zou, Increase ofstructural defects by N doping in MoS2 cross-linked with N-doped CNTs/carbon for enhancing charge transfer in oxygen reduction, Electrochim. Acta192 (2016) 22e29.

[6] X. Gan, R. Lv, X. Wang, Z. Zhang, K. Fujisawa, Y. Lei, Z.H. Huang, M. Terrones,F. Kang, Pyrolytic carbon supported alloying metal dichalcogenides as free-standing electrodes for efficient hydrogen evolution, Carbon 132 (2018)512e519.

[7] J. Zhu, W. Sun, D. Yang, Y. Zhang, H.H. Hoon, H. Zhang, Q. Yan, Multifunctionalarchitectures constructing of PANI nanoneedle arrays on MoS2 thin nano-sheets for high-energy supercapacitors, Small 11 (33) (2015) 4123.

[8] T.J. Kang, S. Fang, M.E. Kozlov, C.S. Haines, L. Na, H.K. Yong, Y. Chen,R.H. Baughman, Electrical power from nanotube and graphene electro-chemical thermal energy harvesters, Adv. Funct. Mater. 22 (3) (2012)477e489.

[9] S. Huang, Y. Meng, S. He, A. Goswami, Q. Wu, J. Li, S. Tong, T. Asefa, M. Wu, N-,O-, and S-tridoped carbon-encapsulated Co9S8 nanomaterials: efficientbifunctional electrocatalysts for overall water splitting, Adv. Funct. Mater. 27(2017), 1606585.

[10] X.X. Ma, X.H. Dai, X. He, Co9S8 modified N, S, and P ternary-doped 3D gra-phene aerogels as A high performance electrocatalyst for both oxygenreduction reaction and oxygen evolution reaction, ACS Sustain. Chem. Eng. 5(2017) 9848e9857.

[11] B. Chen, G. Ma, Y. Zhu, J. Wang, W. Xiong, Y. Xia, Metal-organic-framework-derived bi-metallic sulfide on N, S-codoped porous carbon nanocomposites asmultifunctional electrocatalysts, J. Power Sources 334 (2016) 112e119.

[12] C. Liu, K. Wang, X. Zheng, X. Liu, Q. Liang, Z. Chen, Rational design of MoSe2-NiSe@carbon heteronanostructures for efficient electrocatalytic hydrogenevolution in both acidic and alkaline media, Carbon 139 (2018) 1e9.

[13] C. Yu, X. Meng, X. Song, S. Liang, Q. Dong, G. Wang, C. Hao, X. Yang, T. Ma,P.M. Ajayan, Graphene-mediated highly-dispersed MoS2 nanosheets withenhanced triiodide reduction activity for dye-sensitized solar cells, Carbon100 (2016) 474e483.

[14] C. Yu, X. Han, Z. Liu, C. Zhao, H. Huang, J. Yang, Y. Niu, J. Qiu, An effectivegraphene confined strategy to construct active edge sites-enriched nano-sheets with enhanced oxygen evolution, Carbon 126 (2017) 437e442.

[15] Z. Jixin, Y. Dan, Y. Zongyou, Y. Qingyu, Z. Hua, Graphene and graphene-basedmaterials for energy storage applications, Small 10 (17) (2015) 3480e3498.

[16] C. Yu, Z. Liu, X. Han, H. Huang, C. Zhao, J. Yang, J. Qiu, NiCo-layered doublehydroxides vertically assembled on carbon fiber papers as binder-free high-active electrocatalysts for water oxidation, Carbon 110 (2016) 1e7.

[17] Y. Huang, J. Liang, Y. Chen, An overview of the applications of graphene-basedmaterials in supercapacitors, Small 8 (12) (2012) 1805e1834.

[18] R. Jin, J. Liu, Y. Xu, G. Li, G. Chen, Solvothermal synthesis and excellent elec-trochemical performance of polycrystalline rose-like Co9S8 hierarchical ar-chitectures, J. Mater. Chem. 1 (2013) 7995.

[19] M. Li, H. Zhou, W. Yang, L. Chen, Z. Huang, N. Zhang, C. Fu, Y. Kuang, Co9S8nanoparticles embedded in a N, S co-doped graphene-unzipped carbonnanotube composite as a high performance electrocatalyst for the hydrogenevolution reaction, J. Mater. Chem. 5 (2016) 1014e1021.

[20] W. Chen, Y. Li, X. Han, X. Zhao, Y. Zhao, Self-assembled Co9S8/RGO-CNTinterconnected architecture as composite electrode for supercapacitors, RSCAdv. 7 (2017) 6835e6841.

[21] Q. Wang, Y. Fan, K. Wang, H. Shen, G. Li, H. Fu, Y. She, Hierarchical tubularstructures composed of CoPx and carbon nanotubes: highly effective elec-trocatalyst for oxygen reduction, Carbon 130 (2018) 241e249.

[22] C. Ye, L. Zhang, C. Guo, D. Li, A. Vasileff, H. Wang, S.Z. Qiao, A 3D hybrid ofchemically coupled nickel sulfide and hollow carbon spheres for high per-formance lithiumesulfur batteries, Adv. Funct. Mater. 27 (2017), 1702524.

[23] J. Zhou, H. Song, L. Ma, X. Chen, Magnetite/graphene nanosheet composites:interfacial interaction and its impact on the durable high-rate performance inlithium-ion batteries, RSC Adv. 1 (2011) 782e791.

[24] K.S. Subrahmanyam, A.K. Manna, S.K. Pati, C.N.R. Rao, A study of graphenedecorated with metal nanoparticles, Chem. Phys. Lett. 497 (2010) 70e75.

[25] J. Xie, H. Qu, J. Xin, X. Zhang, G. Cui, X. Zhang, J. Bao, B. Tang, Y. Xie, Defect-richMoS2 nanowall catalyst for efficient hydrogen evolution reaction, Nano Res.10 (2017) 1178e1188.

[26] A. Zhu, P. Tan, L. Qiao, Y. Liu, Y. Ma, J. Pan, Sulphur and nitrogen dual-dopedmesoporous carbon hybrid coupling with graphite coated cobalt and cobaltsulfide nanoparticles: rational synthesis and advanced multifunctional elec-trochemical properties, J. Colloid Interface Sci. 509 (2018) 254e264.

[27] C.B. Ma, X. Qi, B. Chen, S. Bao, Z. Yin, X.J. Wu, Z. Luo, J. Wei, H.L. Zhang,H. Zhang, MoS2 nanoflower-decorated reduced graphene oxide paper forhigh-performance hydrogen evolution reaction, Nanoscale 6 (2014)5624e5629.

[28] Y. Lin, L. Yang, Y. Zhang, H. Jiang, Z. Xiao, C. Wu, G. Zhang, J. Jiang, L. Song,Defective carboneCoP nanoparticles hybrids with interfacial charges polari-zation for efficient bifunctional oxygen electrocatalysis, Adv. Energy Materials8 (2018), 1703623.

[29] P. Ganesan, M. Prabu, J. Sanetuntikul, S. Shanmugam, Cobalt sulfide nano-particles grown on nitrogen and sulfur codoped graphene oxide: an efficientelectrocatalyst for oxygen reduction and evolution reactions, ACS Catal. 5(2015) 3625e3637.

[30] J. Long, Y. Gong, J. Lin, Metaleorganic framework e derived Co9S8@CoS@CoO@C nanoparticles as efficient electro- and photo-catalysts for the oxygen evo-lution reaction, J. Mater. Chem. 5 (2017) 10495e10509.

[31] H. Yuan, Q. Jiao, J. Liu, X. Liu, H. Yang, Y. Zhao, Q. Wu, D. Shi, H. Li, Ultrathin-walled Co9S8 nanotube/reduced graphene oxide composite as an efficientelectrocatalyst for the reduction of triiodide, J. Power Sources 336 (2016)132e142.

[32] H. Zhu, J. Zhang, R. Yanzhang, M. Du, Q. Wang, G. Gao, J. Wu, G. Wu, M. Zhang,B. Liu, When cubic cobalt sulfide meets layered molybdenum disulfide: a core-shell system toward synergetic electrocatalytic water splitting, Adv. Mater. 27(2015) 4752e4759.

[33] L.L. Feng, G.D. Li, Y. Liu, Y. Wu, H. Chen, Y. Wang, Y.C. Zou, D. Wang, X. Zou,Carbon-armored Co9S8 nanoparticles as all-pH efficient and durable H2-evolvielectrocatalysts, ACS Appl. Mater. Interfaces 7 (2015) 980e998.

[34] L. Han, W. Qin, J. Zhou, J. Jian, S. Lu, X. Wu, G. Fan, P. Gao, B. Liu, Chemicalgrafting of Co9S8 onto C60 for hydrogen spillover and storage, Nanoscale 9(2017) 5141e5147.

[35] X.i. Ma, Y. Su, X.u. He, Use of cobalt polyphthalocyanine and graphene asprecursors to construct an efficient Co9S8/N,S-G electrocatalyst for the oxygenelectrode reaction in harsh media, ChemCatChem 9 (2017) 308e315.

[36] Q. Wang, L. Jiao, H. Du, Y. Si, Y. Wang, H. Yuan, Co3S4 hollow nanospheresgrown on graphene as advanced electrode materials for supercapacitors,J. Mater. Chem. 22 (2012) 21387e21391.

[37] L.L. Feng, M. Fan, Y. Wu, Y. Liu, G.D. Li, H. Chen, W. Chen, D. Wang, X. Zou,Metallic Co9S8 nanosheets grown on carbon cloth as efficient binder-freeelectrocatalysts for the hydrogen evolution reaction in neutral media,J. Mater. Chem. 4 (2016) 6860e6867.

[38] S. Liu, M. Tong, G. Liu, X. Zhang, Z. Wang, G. Wang, W. Cai, H. Zhang, H. Zhao,S,N-Containing Co-MOF derived Co9S8@S,N-doped carbon materials as effi-cient oxygen electrocatalysts and supercapacitor electrode materials, Inorg.Chem. Front. 4 (2016) 491e498.

[39] X. Cao, X. Zheng, J. Tian, C. Jin, K. Ke, R. Yang, Cobalt sulfide embedded inporous nitrogen-doped carbon as a bifunctional electrocatalyst for oxygenreduction and evolution reactions, Electrochim. Acta 191 (2016) 776e783.

[40] W. Wang, L. Kuai, W. Cao, M. Huttula, S. Ollikkala, T. Ahopelto, A.P. Honkanen,S. Huotari, M. Yu, B. Geng, Mass - production of mesoporous MnCo2O4 spinelswith manganese(IV) -and cobalt(II) - rich surfaces for superior bifunctionaloxygen electrocatalysis, Angew. Chem. Int. Ed. 56 (2017) 14977e14981.

[41] C.C. Mccrory, S. Jung, J.C. Peters, T.F. Jaramillo, Benchmarking heterogeneouselectrocatalysts for the oxygen evolution reaction, J. Am. Chem. Soc. 135

J. Wang et al. / Carbon 144 (2019) 259e268268

(2013) 16977e16987.[42] J. Duan, S. Chen, M. Jaroniec, S.Z. Qiao, Heteroatom-doped graphene-based

materials for energy-relevant electrocatalytic processes, ACS Catal. 5 (2015)5207e5234.

[43] A. Pendashteh, J. Palma, M. Anderson, R. Marcilla, NiCoMnO4 nanoparticles onN-doped graphene: highly efficient bifunctional electrocatalyst for oxygen

reduction/evolution reactions, Appl.Catal. B Environ. 201 (2017) 241e252.[44] G.L. Tian, Q. Zhang, B. Zhang, Y.G. Jin, J.Q. Huang, D.S. Su, F. Wei, Toward full

exposure of “active sites”: nanocarbon electrocatalyst with surface enrichednitrogen for superior oxygen reduction and evolution reactivity, Adv. Funct.Mater. 24 (2015) 5956e5961.