Elucidating the Impact of Cobalt Doping on the Lithium

DOI:10.1002/celc.201600179 Articles

Elucidating the Impact of Cobalt Doping on the Lithium Storage Mechanism in Conversion/Alloying-Type Zinc Oxide Anodes Franziska Mueller,[a, b, c] Dorin Geiger,[d] UteKaiser,[d] Stefano Passerini,*[a, b] and Dominic Bresser*[a, b, e]

Herein, an in-depthinvestigation of the influence of transition- in situ electrochemical XRD analysisofthe first de-/lithiation of metal doping on the structural and electrochemical character- ZnO and Zn0.9Co0.1Orevealed the highly beneficial impact of istics of ahybrid conversion/alloying-typelithium-ionanode the transition-metal dopant on the reversible degradation of materialispresented. Therefore, pure zinc oxide (ZnO) and lithium oxide (Li2O) and suppression of zinc crystallite growth cobalt-doped ZnO (Zn0.9Co0.1O) were investigated comparative- upon lithiation;both effects are essential for greatlyimproved ly.Characterization by using X-ray diffraction (XRD), scanning electrochemical performance. As aresult, Co doping leads to 1 electron microscopy (SEM), and transmission electron micros- asubstantially increasedspecific capacityfrom 326 mAhgÀ 1 copy (TEM) confirmedthe successful incorporation of the for pure ZnO to 789 mAhgÀ for Zn0.9Co0.1Oafter 75 full cobalt (Co) dopant into the wurtzite ZnO structure, which led charge–discharge cycles. to adecreased particlesize for the doped compound. The

1. Introduction

Despite great technological interestinnanostructured zinc this inferiorelectrochemical performance of ZnO-based electro- oxide (ZnO) for alarge variety of applications, such as light- des, Pelliccione and co-workers[6] very recently performed an emitting diodes,[1] gas sensors,[2] or dye-sensitized solar cells,[3] in situ X-ray absorption fine structure (XAFS) spectroscopy their investigation as lithium-ion anode materials has been re- study.Itwas found that this substantial capacity decay mainly ported rather scarcely to date. In fact, although atheoretical originates from the formation of relatively large metallic zinc 1 specific capacity of about 988 mAhgÀ (if the reaction ZnO+ particles upon continuous de-/lithiation. Once the size of these + 3Li +3eÀ LiZn+ Li Oisconsidered to be fully reversible) particles exceeds acertain limit, the formation of ZnO accom- $ 2 certainly arouses interest, apart from afew outstanding excep- paniedbythe degradationofLi2O, that is, the conversion reactions[4] most electrochemical studies revealed poor per- tion, shows no further reversibility and only the alloying reac- formance and rapid capacity fading.[5] To explain the origin of tion of zinc with lithium takes place reversibly, which, in turn, is associated with the characteristic issues of alloyingmateri- als.[5f] [a] F. Mueller,Prof. Dr.S.Passerini, Dr.D.Bresser Helmholtz Institute Ulm (HIU) In addition, the large volume changes upon alloying–deal- Helmholtzstrasse 11,89081, Ulm (Germany) loying lead to pulverization of the active material, ongoing E-mail:[email protected] electrolyte decomposition, and finally the loss of electrical con- [email protected] tact within the composite electrode.[5f] Additionally,the de- [b] F. Mueller,Prof. Dr.S.Passerini, Dr.D.Bresser crease in electronic conductivity within the electrode owing to Karlsruher Institute of Technology(KIT) P.O. Box 3640, 76021Karlsruhe (Germany) the insulating nature of Li2Opresumably contributes to the ob- [6] [c] F. Mueller served rapid capacity fading. We have recently reported that Institute of Physical Chemistry these challenges and, in particular,the reversibility of the con- University of Muenster version reactioncan be substantially enhanced by introducing Corrensstrasse 28/30, 48149 Muenster (Germany) atransition-metaldopant into the metal oxide structure.[7] This [d] Dr.D.Geiger, Prof. Dr.U.Kaiser new conceptfollowsthe general approach of introducing CentralFacility for Electron Microscopy Group of Electron Microscopy of Materials Science ametallic element that does not alloy with lithium once re- [8] University of Ulm duced to the metallicstate, whichthusensuresasufficient 89081 Ulm (Germany) electron supply throughout the initial particles to enable the [e] Dr.D.Bresser degradation of Li2O. However,little is knownsofar about the DRF/INAC/SYMMES/PCI detailedreaction mechanism of this new class of conversion/al- CEA, UMR-5819, CEA-CNRS-UJF 17 Rue des Martyrs, 38054 Grenoble (France) loying materials. The ORCID identification number(s) for the author(s) of this article can Herein, we present an in-depth structuraland electrochemi- be found under http://dx.doi.org/10.1002/celc.201600179. cal characterization of pure ZnO and Co-doped ZnO nanoparti-

ChemElectroChem 2016, 3,1311–1319 1311 2016 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Articles cles by using XRD, SEM,TEM, cyclic voltammetry (CV), and gal- (Rockwood Lithium, battery grade) as the counter and reference vanostatic cycling techniques. Moreover,weperformed acom- electrodes. The electrodes were separated by astack of polypropy- parativeinsitu XRD analysis of Co-doped and pure ZnO nano- lene fleeces (Freudenberg FS 2190), drenched with a1m solution particles. The obtained results, which are also discussed with of LiPF6 in amixture of ethylene carbonate/diethyl carbonate (UBE; 3:7wt/wt). Because metallic lithium foil was used as the counter respecttoour previousfindings for Fe-doped ZnO,[7a] provide and reference electrodes, all potential values given herein refer to an enhanced understanding of the advantageous impact of the Li/Li + redox couple. Cyclic voltammetry and galvanostatic cy- the transition-metal dopant on the reversibility of the de-/lith- cling were performed by using aVMP3 potentiostat (BioLogic) and iation mechanism in generaland the conversion reaction in aMaccor Battery Tester 4300, respectively.All electrochemical stud- particular. In fact, doping ZnO with cobalt resultsinasubstan- ies were conducted at (20 2)8Cinapotential range of 0.01–3.0 V. Æ tially enhanced specific capacity and cycling stabilityofnano- particulate ZnO-based lithium-ion anodes. In situ XRD Analysis In situ XRD analysis of Co-doped and pure ZnO nanoparticles upon Experimental Section galvanostatic de-/lithiation was conducted by using aself-designed two-electrode in situ cell.[9] Electrodes were prepared analogously Materials Synthesis to those used for the electrochemical characterization, with an For the synthesis of Co-doped ZnO nanoparticles, stoichiometric overall composition of 65 wt%active material, 25 wt%conductive amounts of zinc(II) gluconate hydrate and cobalt(II) gluconate hy- carbon, and 10 wt%CMC. The resulting mixture was dispersed by drate (both from ABCR) were dissolved in deionized water.The re- planetary ball-milling and cast onto aberyllium (Be) disk (thickness: sulting solution, with atotal metal ion concentration of 0.2m,was 250 mm, Brush Wellman) that served concurrently as the current added dropwise under vigorous stirring to aqueous sucrose (1.2m; collector and “window” for the X-ray beam. The coated Be disk ACROS ORGANICS). After continuous stirring for an additional was dried for 30 min at 80 8Cunder ambient atmosphere and sub- 15 min at RT,the solution was heated to about 1608Ctoremove sequently at 408Cunder vacuum overnight. Metallic lithium foil the water.Concurrently,the sucrose started to become thermally served as the counter electrode. Twosheets of Whatman glass decomposed. The remaining solid precursor was further dried at fiber drenched with asolution of LiPF6 (1m,500 mL) in ethylene around 3008C, then ground and calcined for 4h under air at carbonate/diethyl carbonate (UBE;3:7 wt/wt) were used as separa- 1 tor.The assembled cell was allowed to rest for 10 hprior to the 4008C(heating rate:38CminÀ ). Pure ZnO nanoparticles were prepared by following an analogous synthesis protocol in the absence in situ XRD measurements. Simultaneous galvanostatic cycling was of the Co precursor and using an annealing temperature of 450 C. performed by using aVSP potentiostat/galvanostat (BioLogic) and 8 1 applying aspecific current of 55 (Zn Co O) or 60 mAgÀ (ZnO) 0.9 0.1 and setting the cut-off potentials to 0.01 or 3.0 V. The XRD analysis Morphological and Structural Characterization was carried out in a2q range of 25–808 with astep size of 0.027588 and atime per step of 0.65 s. Consequently,every scan Powder X-ray diffraction (XRD) patterns were recorded by using lasted exactly 30 min, which included an initial rest step of 419 s. aBRUKER D8 Advance with CuKa1 radiation (l=154 pm) and astep size of 0.00928.Scanning electron microscopy (SEM) was carried out by using aZEISS LEO 1550VP field emission electron micro- 2. Results and Discussion scope. All samples were sputtered with platinum to prevent charg- 2.1. Structural and Morphological Characterization ing effects. High-resolution transmission electron microscopy (HRTEM)was performed by using aPhilips CM 20 at 200 kV for The X-ray diffractionpatterns of pure ZnO and Zn0.9Co0.1O, pre- Zn0.9Co0.1Onanoparticles samples deposited on acopper grid sented in Figure 1, show the characteristic pattern of hexago- coated with aholey carbon film. nal-wurtzite-structuredZnO with the P63mc space group (JCPDS file card no. 01-071-6424), which indicates thesynthesis Electrochemical Characterization of phase-pure materials and the successful doping of ZnO with cobalt.The additional low-intensity reflectionsatabout 28.7, Electrodes were prepared by dissolving sodium carboxymethyl cel- 31.1, and 32.78 (marked by *) originate from the diffractionof lulose (CMC, WALOCELTM CRT 2000 PPA12, Dow Wolff Cellulosics) the Cu b radiation at the (100), (002), and (101) lattice planes, in deionized water (1.25 wt%ofCMC in solution). Conductive K 1 respectively.The Co-doped sample reveals slightly broader full carbon (Super C65 ,IMERYSGraphite and Carbon) and ZnO or width at half maximum values and lower reflection intensities Zn0.9Co0.1Onanoparticles were added, which gave arelative dry material composition of 75 wt%active material, 20 wt%conduc- compared with pure ZnO, which indicates areduced crystallite tive carbon, and 5wt% CMC. Dispersion of the obtained mixture size. These resultsare in excellent agreement with arecent in- was carried out by means of planetary ball-milling for 2h.The re- depth crystallographic analysis, reported by Giuli et al.[10] sulting slurry was cast on dendritic copper foil (SCHLENK, 99.9%), Figure 2shows the SEM images of the pristine powdery which served as acurrent collector,byusing alaboratory doctor samples. The SEM image of ZnO (Figure 2a) reveals aparticle blade (wet film thickness:120 mm). After an initial drying at RT diameter of about 40 nm, whereas the introduction of Co into overnight, disk electrodes (ø=12 mm) were punched and vacuum- the ZnO lattice resultsinadecreased particle size of around 20 dried for 24 hat1208C. The average active material mass loading 2 to 30 nm. These findings are in good agreement with the XRD was around 2.0–2.5 mgcmÀ . [7a,10] The electrochemical characterization was performed in three-elec- analysisand previousstudies. The TEM micrographs of Co- trode Swagelok-type cells assembled in an MBraun MB-200B eco doped ZnO showninFigure 3confirmthis particle size and ad- glovebox (O and H Ocontent <0.1 ppm), with lithium metal foil ditionallyrevealthe presence of slightly smaller( 10 nm) and 2 2

ChemElectroChem 2016, 3,1311–1319 www.chemelectrochem.org 1312 2016 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Articles

Figure 4. Cyclic voltammograms for ZnO:a)first sweep, b) 2nd–10th sweep; 1 and for Zn0.9Co0.1O: c) first sweep, d) 2nd–10th sweep. Sweeprate:50mVsÀ ; reversing potentials:0.01Vand 3.0 Vvs. Li/Li+. Figure 1. XRD patterns of as-synthesized ZnO nanoparticles (upper pattern in black) and Co-doped ZnO nanoparticles (lower pattern in green); the ously decreasing specific current was recorded,which is as- JCPDS reference file card no. 01-071-6424 (wurtzite structure, hexagonal signedtochargeaccumulationatthe surface and in the symmetry, P63mc space group) is given below. carbon black, employed as conductive additive. The peak at 0.70 Viscommonly ascribed to the reduction of electrolyte components at the anode/electrolyteinterface, which results in the formation of asolid–electrolyte interphase (SEI).[11] The main peak at 0.38 Vismainly relatedtothe reduction of ZnO

to form metallicZnand Li2O. At potentials below 0.2 V, the metallicZnalloys with Li. However, the slope of the baselineis rathersteep,which suggeststhat the lithiation processes mightpartially overlap. Upon delithiation, variouspeaks at 0.19, 0.29, 0.35, 0.54, and 0.68 V, one broad peak at 1.32 V, Figure 2. SEM micrographsofa)pure ZnO and b) Co-doped ZnO nanoparti- which probably consists of two overlapping peaks, and alow- cles at amagnification of 100 . intensity peak at 2.55 Vwere detected. The peaks in the low- potentialregion, that is, below 0.75 V, are relatedtothe stepwise dealloying of LiZn as previously reported in literature.[5e,12] The underlying reaction corresponding to the broad,anodic peak at about 1.32 Vremained unexplained foralong time.[5e] Huangetal.[13] recently ascribed this feature to the dealloying

of LixZn. Moreover,they correlated the low-intensitypeak at

2.55 Vtothe partially reversible degradation of Li2Oaccompa- nied by the formation of ZnO.[13] Nevertheless, from studies on tin (oxide)-based anodes, it also appears possible to assign peak e) to the partially reversible formation of Li O[7b] and Figure 3. (HR)TEM micrographs of Co-doped ZnO nanoparticles at different 2 [14] magnifications. peak f) to apartial degradation of SEI components, particularly with respecttoits absence in case of carbon-coated ZnO.[15] However,wewill discuss the assignment of peaks to larger ( 35 nm) particles. Moreover, the HRTEM micrograph specific electrochemical reactions in more detail with reference presented in Figure 3b does not showany significant lattice to the in situ XRD analysis presented hereafter.During subse- distortion, whichfurther confirms the successful incorporation quent sweeps, shown in Figure4b, new cathodic peaks at 0.96 of the cobaltdopant into the wurtzite ZnO structure. and 0.78 Varose, whereas the anodic peak at 2.55 Vcomplete- ly vanished;the latter is, in fact, in good agreement with the aforementioned studies on tin-based anodes, which assigned 2.2. Cyclic Voltammetry this peak to apartial SEI degradation.[14] In addition, peaks g),

Figure 4shows cyclic voltammograms of ZnO- and Zn0.9Co0.1O- h), b), c), and e) showashift towards slightly lower potentials based electrodes recorded between 0.01 and 3.0 V. In the first and aconsiderable intensity decrease, and peak g) completely cathodic sweep,ZnO (Figure 4a) reveals three reduction peaks disappeared after afew cycles.Moreover,peak b) split into at potentials of 0.70, 0.38, and 0.13 V. Below 0.07 V, acontinu- two peaks, which are assigned to the alloyingreaction of Zn

ChemElectroChem 2016, 3,1311–1319 www.chemelectrochem.org 1313 2016 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Articles with Li;thus, the stepwise alloying becomes more apparent. In related reflections is observed (Figure 5c;marked by the black contrast, the peaks in region d) related to the dealloying pro- arrows), which indicates an incipient reduction in the long- cesses remainrelativelyunchanged, although they reveal range order within the wurtzite structure. This structural aslightly decreasing intensity.Infact, these results revealthe change is accompanied by some electrolyte decomposition, in- often-reported irreversibility of Li2Odegradation and the limit- dicatedbyaslight change in the slope of the discharge curve, ed reversibility of the Li-alloying process of Zn.[5a–e,6] which is in good agreement with the CV results (Figure 4a). Al- In Figure 4c, the first potentiodynamic cyclic sweep for thoughsomehow expected, the initial reduction of the oxide

Zn0.9Co0.1Oispresented. Co-doped ZnO exhibits characteristics is of particular interest because it is in contrast with our earlier similar to undoped ZnO. However,the features are much less finding for Fe-doped ZnO,[7a] for which we did not observe any pronounced, which might be due to the smaller particle and initial decrease in intensity during the first scans. In fact, this crystallite size revealed by SEM (Figure 2) and XRD (Figure 1), remarkable difference regarding the structural evolution may and slightly shiftedtohigher potentials, which presumably re- support ourprevious assumption that for Fe-doped ZnO an in- sults from enhanced electronic conductivity.[16] Analogous to sertionoflithium ions into the wurtzite structure may occur ZnO, as described above, the SEI formationoccurs at about prior to the onset of the voltage plateau,favored by the pres-

0.72 V, followed by reduction of ZnO at 0.46 V. Below 0.25 V, ence of cationic vacancies in Zn0.9Fe0.1Oasproposed by Giuli electrochemical alloyingofZnwith Li, lithium insertion into et al.[10] Acareful analysisofthe different XRD patternsalso re- the conductive carbon, and charge storageatthe surface take veals the appearance of avery broad andyet tiny reflection at place. According to the standard redox potential,Coisnobler about 43.08 (marked by ared arrow). Given the indicated re- than Zn, that is, Co is reduced at higherpotentials. However, ductionofZnO accompanied by the formationofamorphous [5e,6–7,18] no clear feature related to the reduction of Co was observed Li2O, it appearsreasonable to assign this feature to the by performing CV,whichisexpected given the previousresults main reflection of metalliczinc (JCPDScard file no. 00-004- for zinc and cobalt ferrite.[17] Nonetheless, the influence of the 0831),which is also in good agreementwith the onset of the Co dopant is indicated by aslightly higher potentialofpeak a) following voltage plateau(Figure 5a) assigned to the occur- compared with pure ZnO. During the anodicsweep,the deal- rence of an equilibrium between different but simultaneously loying of Zn occurs in the potential range between 0.16 and present phases.

0.74 V, followed by the degradation of Li2Oat1.35 V, as men- Indeed, for the following scans (Figure 5d, region B, scans 6– tioned previously.Insubsequent cycles,peaks a) and b) vanish 12), the continuous decrease and increase in intensity of the and new features appear at 0.93 and0.81 V(Figure 4d). Upon reflectionsthat correspond to ZnO and Zn0,respectively,indi- furthercycling, the peak intensities of these new features de- cates that the oxide is furtherreduced andmore metallic zinc crease and their maximaare shifted to slightly lower poten- is formed. Starting from scan 13/14, three new reflections tials. Additionally, abroad feature characterizedbyaconstantly appear at about 37.0, 41.0, and 42.78 in the selected 2q range; increasing specific current at apotentialofabout 0.5 Varises. the last one is initially present as ashoulder of the Zn0-related

However,compared with ZnO, Zn0.9Co0.1Oshows aremarkably reflection and subsequently continuously increasing in intensi- improved reversibility,particularly for peak e). ty,whereas the Zn0-relatedreflection itself vanished simultaneously.Their positions and relative intensity are in very good agreement with the JCPDS cardfile no. 01-071-9525 reported 2.3. In Situ XRD for Li0.105Zn0.895,which indicatesthat along the voltage plateau To furtherstudy the structuralchanges and phase transitions the previously formed metallic zinc also starts to alloy electro- that occur during the electrochemical lithiation and subse- chemically with lithium and thus forms aLixZn alloy phase quent delithiation reaction and to investigate the influence of with arather low lithium content, that is, x ! 1. In fact, this the cobalt dopant,insitu XRD analysis was performed on elec- finding is furthersupported by the concurrentappearance of trodes based on pure ZnO and Co-doped ZnO nanoparticles. two additional reflections at around56.5 and 67.08 (see Fig- In both cases, the electrodes were subjected to aconstant cur- ure 5b). Apparently,the reduction of the oxide and the initial rent lithiation (discharge) from OCV to 0.01 Vand, subsequent- alloyingreactionoccur almostsimultaneously, that is, at the ly,delithiation (charge)from 0.01 to 3.0 V. The resultsfor the same voltage. At the end of the voltage plateau, in scan 19, pure ZnO nanoparticles are presented in Figure 5. Figure5a the reflections thatcorrespondtothe wurtzite-structured shows the corresponding potentialprofile, which can be oxide completely vanish and only the LixZn-related reflections roughlydivided in five different regions (A, B, and Cfor the remain. discharge and Dand Efor the subsequentcharge). To provide In region C, no significant changesare observed for an overview of the evolution of the simultaneously recorded scans 19–22, that is, along the initial rather steep voltage de- XRD patterns, awaterfall diagram that shows all XRD patterns crease.Subsequently,however,the intensity of the two reflec- is presented in Figure 5b, within which these regions are indi- tions at about 37.0 and42.78 that correspond to the LixZn (x ! cated by the scans highlighted in red. Adetailed discussion of 1) phase decrease in intensity,whereas the one at around the processes that occur therein,however,isbased on the fol- 41.08 continuously increases and initially shifts to slightly larger lowing panels. Region Acovers the initial potentialdecrease 2q values and then back to slightly lower 2q values. Indeed, until the onset of the voltage plateau at about 0.55 V. For this when the extended 2q range (up to 80.08)was analyzed,itre- region (scans 1–5), adecrease in intensity of the main wurtzite- vealed that this shiftingand increase in intensity is related to

ChemElectroChem 2016, 3,1311–1319 www.chemelectrochem.org 1314 2016 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Articles

Figure 5. In situ XRD analysisofZnO:a)Galvanostatic dis-/charge profile of the first cycle,b)corresponding XRD patterns, c)–g) close-ups of selected XRD patterns.

the appearance of the fully lithiated LiZn alloy phase because that correspond to the LixZn phase remain. However, these re- from scan 2, new reflectionsalso appear at around 48.5, 60.0, flections abruptly vanish for the next scan (scan 36) as shown and 75.08.This is in excellent agreement with the JCPDS card in Figure 5g (region E), in which scan 35 is also included for file no. 03-065-3016 reported for LiZn. In addition, further lith- clarity.Instead, four new reflections are observed at about iation, that is, the formation of the fully lithiated binary LiZn 36.2, 38.8, 43.0, and 54.08,whichare in very good agreement alloy at potentials below 0.2 Valong asloped voltageplateau with the JCPDS reference reported for metallic zinc (card file is in good agreement with previous, purely electrochemical no. 00-004-0831), indicatingthat the dealloying reaction is studies on the lithium–zinc alloying reaction mechanism.[5e,12,19] completed at around0.7 Vupon charge. This finding is, in fact, Finally,inscan 28, only reflections that correspondtothe LiZn also in good agreement with the recorded specific capacity phase are present, which indicates that the lithiation reaction (Figure 5a) because the alloyingcontribution for ZnO theoreti- 1 is complete at the end of the discharge step (0.01 V). cally corresponds to 329 mAhgÀ and our assignment of the For the subsequent charge (scans29–44), the LiZn-related different peaks in the CV profile (Figure 4a andb). The subse- reflectionsare first slightly shiftedtolarger 2q values, which in- quent decrease in intensity observedfor the two reflections at dicatesthat the lattice parameters slightly decrease, presuma- around38.8 and 43.08 may be ascribed to thepartial reoxida- bly due to adecreasing lithium content in the alloy (Figure 5f, tion of metallic zinc at potentials higher than 0.7 V. [6] This is fur- scans 29–32, first slope). Beginning from scan 33 (i.e. after the ther supported by the appearance of avery broad reflection in change in the slope), the reflectionsthat correspond to the the range of 31.0 to 35.08 and the limited decrease in intensity

LixZn (x !1) phase reappear and steadily increase in intensity, of the reflection at about36.28 because it overlaps with the whereas the reflections related to the LiZn phase keep de- main reflection of the wurtzite-structured oxide.Nevertheless, creasing. In scan 35, at the end of regionD,only the reflections this reoxidation process appearstobelimited(see also Fig-

ChemElectroChem 2016, 3,1311–1319 www.chemelectrochem.org 1315 2016 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Articles ure 4b) because very intense reflections that correspond to reaction of the oxide upon lithiation (Figure 6c, region A), the metalliczincremain even at the end of the charge process at same structural evolution is observed:Adecrease in intensity 3.0 V. Additionally,the specific capacity obtained over this volt- of the oxide-related reflections, which indicates that for Co- age range is substantially lower than theoretically expected, doped ZnO the oxide is also immediately reduced and that no 1 given the fully reversible conversion of Li2O( 416 mAhgÀ substantial initial lithium-ion insertion into the wurtzite lattice 1 rather than 658 mAhgÀ ). Based on the above-described find- takes place. This finding is, again, in good agreement with the ings, we may qualitatively describe the multistep process as in-depthstructural analysis by Giuli et al.[10] and reveals that, in follows: contrasttoiron, cobalt is purely divalent and, therefore, does not requirethe presence of, for example, cationic vacancies to 0 A ZnO Liþ eÀ ZnO Li2O Zn balance trivalent cationsinthe oxide lattice.Inaddition, start- Þ þ þ ! þ þ 0 B ZnO Zn Liþ eÀ Lix Zn Li2O ing from scan 6, that is, at the onset of the subsequent voltage Þ þ þ þ ! þ plateau the appearance of avery broad yet tiny reflection in C Lix Zn 1 x Liþ 1 x eÀ LiZn Þ þð À Þ þð À Þ ! the range of 41.0 to 45.08 is observed. Its intensity further in- D LiZn Li Zn 1 x Liþ 1 x eÀ Þ ! x þð À Þ þð À Þ creased during the following scans (Figure 6d, region B), 0 Zn Liþ eÀ whereas the oxide-related reflections continuously decrease ! þ þ 0 0 until they completely vanish at the end of the voltage plateau E Zn Li2O Zn ZnO Liþ eÀ Þ þ ! þ þ þ (scan 22). Remarkably,however,incontrast to ZnO, no pro- In summary, nounced reflections are observed for the newly appearing re- flectioninthe range of 41.0 to 45.08 and the intensity remains

1 substantially lower in general.Infact, it appearsthat this ZnO 3Liþ 3eÀ(+LiZn Li O ^ 988 mAh gÀ þ þ À þ 2 ð ¼ Þ ratherbroad new feature maybecomposed of two reflections, which presumably correspond to the concurrent formation of Please note that the equationsfor the single regions are not metallicZn0 and Li Zn (at 43.0 and 42.08,respectively), which x balancedstoichiometrically because an adjustment with un- largely overlap. These findings provide two important pieces known general variables would certainly only complicate the of information:First, the crystallinity of the formed phases is understanding and readingsubstantially without adding any substantially lower than in ZnO (similarly to the findings for beneficial furtherinformation. Additionally,the electrode Fe-doped ZnO[7a]), which indicates that, apart from the slightly chemistry presented for each step focuses only on the reac- smaller crystallite size of the pristine oxides(see Figure 2), the tions that take place in the corresponding potentialrange. We presence of cobalt as adopant prevents the formation of large may note at this point that our findings obtained by perform- crystalline particles upon lithiation and thus favors the nano- ing in situ XRD are in excellent agreementwith the very recent crystallinity of the sample. Second, the introductionofthe [6] in situ XAFS study by Pelliccione et al., who also observed cobalt dopant appears to favor the formation of the LixZn alloy afully reversible LiZn formation in the first cycle, but observed kinetically.These two findings are, in fact, of particularinterest incompleteness of the subsequent reoxidation of metallic zinc because Su et al.[21] very recently reported that the growth of upon delithiation, which was presumably related to the forma- Zn particles and the alloyingreaction are two competing pro- tion of relatively large metallic zinc particles ( 10–20nm) and cesses that dependonthe applied current density.Wemay the corresponding extensive structuralreorganization in the thus assume that the formation of acontinuous percolating electrode, which was further supported by avery recentinsitu network of metallic cobalt leads to enhanced electron trans- TEM study.[18] port and distribution within the initial nanoparticles, which To investigate the effect of introducing the cobalt dopant favors the alloyingreaction rather than zinc crystal growth. into the wurtzite structure, we also carried out an in situ XRD Upon furtherlithiation (Figure 6e, region C), from scan 31 analysisfor Zn0.9Co0.1O-based electrodes (Figure 6). The corre- anew reflection appearsatabout 41.08,which is in excellent sponding potentialprofile is presented in Figure 6a whereas agreement with the results obtained for ZnO and indicates the the overview of the evolutionofthe recorded XRD patterns is formation of the LiZn alloy phase at apotential of less than given in Figure 6b. 0.2 V(see also Figure 6a). Again, the potentialprofile can be subdivided into five dif- This reaction is subsequently reversed upon delithiation (i.e. ferent regions(indicated by the red highlighted scans), similar charge;Figure 6f,region D), as seen for ZnO. Nevertheless, the to the analysisofZnO. Although the larger number of scansis dealloying reaction is apparently completed at relatively higher aresult of the slightly lower specific current applied (55 vs. potentials ( 1.2 Vcompared with 0.7 Vfor ZnO;see Fig- 1 60 mA gÀ for ZnO), two major differences are evident. The po- ures 5a and 6a, respectively). Also, it appears that for tential profile is generally smoother,that is, it shows less re- Zn0.9Co0.1Oinregion E(Figure 6g) some LixZn (x !1) phase remarkable changes in slope, which may originate partially from mains,asindicated by the shoulder of the reflection at about the smallerinitial crystallite size,[16a, b] and is in good agreement 43.08;the latter is assigned to the presenceofmetallic zinc, with the cyclic voltammetry results shown in Figure 4and the thoughthe rather broad nature of the reflection makesitdiffi- slightly higherpotential of the main voltage plateau upon dis- cult to clearly differentiate between the two phases. This is es- charge.The latter may be assigned to the higherreduction po- pecially true because no significant additional reflectionswere tential of cobalt compared with zinc.[20] However,for the initial observed for this sample. In either case, this rather broad re-

ChemElectroChem 2016, 3,1311–1319 www.chemelectrochem.org 1316 2016 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Articles

Figure 6. In situ XRD analysis of Zn0.9Co0.1O: a) Galvanostatic dis-/charge profile of the first cycle, b) the corresponding XRD patterns, c)–g)close-ups of selected XRD patterns.

0 0 flectioncontinuously decreases in intensity upon further deli- A Zn Co O Liþ eÀ Zn Co O Co Zn Li O Þ 0:9 0:1 þ þ ! 0:9 0:1 þ þ þ 2 thiationuntil it completely vanishes at the end of the charge 0 0 0 B Zn Co O Zn Liþ eÀ Li Zn Co Zn Li O x 1 Þ 0:9 0:1 þ þ þ ! x þ þ þ 2 ð Þ process at 3.0 V. Concurrently,anew broad reflection appears 0 C Li Zn Zn Liþ eÀ LiZn within 30.0 to 37.08 and steadily increases, accompanied by Þ x þ þ þ ! 0 the appearance of an even broader reflection at around57.0 . D LiZn Lix Zn Zn Liþ eÀ 8 Þ ! þ þ þ Considering the substantially reduced crystallinity of the 0 0 E Lix Zn Zn Co Li2O Zn1 y Coy O* Liþ eÀ y 0:1 sample after the first cycledue to the extensive structuralreor- Þ þ þ þ ! À þ þ ð Þ ganization processesthat occur,itappears conceivable that In summary, these two broad reflections indicate the reappearanceof awurtzite-structured oxide phase. Moreover,the complete dis- Zn Co O 2:9Liþ 2:9eÀ appearance of the metallic zinc phase in combinationwith the 0:9 0:1 þ þ 0 1 1 0:9LiZn Co Li O ^ 962 mAh gÀ obtainedspecific capacity of 970 mAhgÀ ,that is, roughly the Ð þ þ 2 ð ¼ Þ theoretical capacity expected for the fully reversible formation 1 of LiZn and Li2O(=^ 962 mAhgÀ ), further support the forma- Please note that in this case the equations for the single re- tion of ametal oxide phase at the end of chargefor both zinc gions are not balanced stoichiometrically to facilitate the un- and cobalt. The slight excessincapacity may be well explained derstanding and readingofthe overall reaction mechanism. by the lithium storagecapability of the comprised conductive We also note that we do not have any direct evidence at this carbon,asalso observed previously.[9] Based on these results, point for the reversible formation of Co-doped ZnO (marked we propose the following multistep process for the first lithia- by *inthe above equations), not to mention the stoichiometry tion and delithiation process: of such areversibly formed phase. In fact, it may well be that

ChemElectroChem 2016, 3,1311–1319 www.chemelectrochem.org 1317 2016 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Articles

1 zinc oxide and cobalt oxide are formed instead, as reportedfor tion of the LiZnalloy (=^ 329 mAhgÀ )isconsidered. The small [17b,22] spinel-structured ZnCo2O4. Further complementary studies fraction of extra capacity maybeassigned to apartially reversi- will have to be performed to clarify this aspect. ble electrolyte decomposition, as already reported for other At this point, we briefly summarize the effect of introducing transition-metal oxides, and also the contribution from the the cobalt dopant into the wurtzite zinc oxide structure. Most conductive carbon.[26] In fact, the potential profiles of selected importantly,inparticular regarding the application of such ma- cycles (Figure 7b) show astepped shape relatedtothe gradual terials as alternative anode materials for lithium-ion batteries, de-/alloying process and no significant capacity in the high po- the presence of evenly distributed cobalt cations within the tential regime, at which the degradation of Li2Ooccurs. Like- [10] ZnO lattice enhances the reversibility of the Li2Oformation wise, Co-doped ZnO shows slight capacity fading during the and thus the reversible specific capacity.This beneficial effect initial cycles.However,after 40 cycles,Zn0.9Co0.1Oreveals presumably results from the presenceofapercolating network astable performance that provides aspecific capacity of 1 of ultrafine metallic cobalt nanoparticles formed upon lithi- 762 mAhgÀ . ation,[8a,23] which ensures the required electron supply through- The corresponding potential profiles, depictedinFigure 7c, out the Li2Omatrix.Additionally,the Co doping leads to an en- exhibit asloped profile as already stated above. Upon lithia- hanced nanocrystallinity of the lithiated sample, which con- tion, no particularfeatures can be correlated with specific elec- stricts the formation of rather large LiZn and Zn0 crystals. The trochemical reactions, which suggests acatalytic effect of Co latter aspect, in fact, appears to be of major importance foral- that resultsinthe promoted reduction of ZnO. However, in the loying materials in general,aspreviously reported for the com- low-potential range of the charge profile, features similar to mercialSn-Co-C anode materialcomposite[24] employed in the the pure ZnO are observed. More importantly,for the delithia- “Nexelion” battery (SONY), and also silicon-basedlithium-ion tion process asignificant capacity is obtainedatpotentials [25] anodes. above 1.5 Vthat is ascribed to the degradation of Li2O, in ac- cordancewith the in situ XRD analysis presented herein. To evaluate the rate capability of ZnO in comparison with 2.4. Galvanostatic Cyclingand Rate Capability Zn0.9Co0.1O, amulti-rate cycling test was conducted (Figure 8). The impact of the Co doping of ZnO on the long-term galva- The plot of the specific charge capacity versus the applied spe- nostatic cycling performanceispresented in Figure 7. In the cific current demonstratesonce more the superior capacity specific capacity versus cycle number plot (Figure 7a), the posi- provided by the Co-doped ZnO, particularly compared with tive effect of the doping in terms of achievable specific capaci- pure ZnO. Nevertheless, for very high specific currents the de- ty and first cycle irreversible capacity loss, that is, 42.6 %for liveredcapacity decreases dramatically,which indicates the

ZnO and 29.5%for Zn0.9Co0.1O, is immediately apparent. After need for enhanced electrode composite architectures,such as some formation cycles, pure ZnO shows astabilized reversible the incorporation of secondary carbon nanostructures like gra- 1 [27] capacityofabout 332 mAhgÀ in the 30th cycle, which is close phene or carbonaceous coatings. to the theoretical capacity of ZnO if only the reversible forma- 3. Conclusion Pure andCo-doped ZnO were synthesized and characterized structurally,morphologically,and electrochemically.XRD, SEM, and TEM analyses revealed areduction in the particle size due to the introduction of the dopant, which is also evident from

Figure 7. Constantcurrent cycling of ZnO- (black) and Zn0.9Co0.1O-based 1 (green)electrodes cycled at aspecific current of 48 mA gÀ (C/20); 1st cycle: 1 24 mAgÀ (C/40).a)Capacity vs. cycle number plot;b,c)the corresponding potential profiles of ZnO- and Zn0.9Co0.1O-based electrodes for selected Figure 8. Comparison of the rate capability of electrodes based on ZnO cycles (2nd,10th,20th, 30th,40th, 50th,60th, and 70th). (black, diamonds)and Co-doped ZnO (green,circles)nanoparticles.

ChemElectroChem 2016, 3,1311–1319 www.chemelectrochem.org 1318 2016 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Articles the smoother and broadened current peaks recorded by cyclic H. K. Liu, S. X. Dou, J. PowerSources 2002, 109,142 –147;c)D.Bresser, voltammetry.Toelucidate the impact of the dopantonthe E. Paillard,P.Niehoff, S. Krueger,F.Mueller,M.Winter,S.Passerini, Chem- PhysChem 2014, 15,2177 –2185;d)A.Brandt, A. Balducci, J. PowerSour- lithium storagemechanism, electrochemical in situ XRD analy- ces 2013, 230,44–49;e)C.M.Ban, Z. C. Wu, D. T. Gillaspie, L. Chen,Y.F. sis of pure ZnO and Co-doped ZnO was performed. Theresults Yan, J. L. Blackburn, A. C. Dillon, Adv.Mater. 2010, 22,E145 –E149; f) Y. J. reveal abeneficial impact of the transition-metal dopant on Zhang, Y. Yan, X. Y. Wang, G. Li, D. R. Deng, L. Jiang, C. Y. Shu, C. R. the electrochemical performance of ZnO by preventing the for- Wang, Chem. Eur.J.2014, 20,6126 –6130;g)F.Lin,D.Nordlund, T. C. Weng, Y. Zhu, C. M. Ban, R. M. Richards,H.L.Xin, Nat. Commun. 2014, 5, mation of large crystalsand enhancing the reversibility of Li2O 3358. formation.Additionally,reduction of the metal dopant presum- [9] D. Bresser,E.Paillard, R. Kloepsch, S. Krueger, M. Fiedler,R.Schmitz, D. ably leads to improved electronic conductivity within the lithi- Baither, M. Winter,S.Passerini, Adv. EnergyMater. 2013, 3,513–523. ated particles. As aresult,the reversible specific capacity initial- [10] G. Giuli, A. Trapananti, F. Mueller,D.Bresser,F.d’Acapito,S.Passerini, 1 Inorg. Chem. 2015, 54,9393 –9400. ly approaches the theoretical value of 962 mAhgÀ .Long-term [11] K. Xu, Chem. Rev. 2004, 104,4303 –4417. galvanostatic cycling further demonstrates asignificant im- [12] T. Fujieda, S. Takahashi, S. Higuchi, J. PowerSources 1992, 40,283–289. provement in the achievable specific capacity.After 75 full [13] X. H. Huang,X.H.Xia, Y. F. Yuan, F. Zhou, Electrochim. Acta 2011, 56, 4960 –4965. dis-/charge cycles, electrodes based on Zn0.9Co0.1Oshow more 1 [14] a) B. K. Guo, J. Shu,K.Tang, Y. Bai, Z. X. Wang,L.Q.Chen, J. PowerSour- than twice the capacity of pure ZnO, that is, 789 mAhgÀ ces 2008, 177,205–210;b)X.W.Guo, X. P. Fang, Y. Sun, L. Y. Shen, Z. X. 1 versus 326 mAhgÀ . Wang, L. Q. Chen, J. Power Sources 2013, 226,75–81;c)D.Bresser,F. Mueller,D.Buchholz, E. Paillard, S. Passerini, Electrochim.Acta 2014, 128,163–171. Acknowledgements [15] X. Y. Shen, D. B. Mu, S. Chen, B. R. Wu, F. Wu, ACS Appl. Mater.Interfaces 2013, 5,3118–3125. Financial support by BMW AG within the ABILE (Air Batteries with [16] a) H. Li, Y. Wei, Y. Zhang, F. Yin, C. Zhang, G. Wang,Z.Bakenov, Ionics 2016,1–7;b)J.Hassoun, G. Derrien, S. Panero, B. Scrosati, Adv.Mater. Ionic Liquid Electrolytes) project is gratefully acknowledged. D.B. 2008, 20,3169 –3175;c)I.A.Courtney,J.R.Dahn, J. Electrochem. Soc. would like to thank the CEA NanoSciences Programme within the 1997, 144,2943 –2948. ELLIPSEproject (14.4.1) and theCEA/EU Enhanced Eurotalents [17] a) Y. Ding, Y. F. Yang, H. X. Shao, Electrochim. Acta 2011, 56,9433 –9438; Fellowshipfor financial support. b) N. Du, Y. F. Xu, H. Zhang, J. X. Yu,C.X.Zhai, D. R. Yang, Inorg. Chem. 2011, 50,3320 –3324. [18] Y. F. Zhang, Z. Y. Wang, Y. J. Li, K. J. Zhao, Mech.Mater. 2015, 91,313 – Keywords: cobalt · doping · lithium-ion batteries · 322. [19] J. Q. Wang, P. King, R. A. Huggins, Solid State Ionics 1986, 20,185 –189. nanoparticles · zinc oxide [20] A. J. Bard, R. Parsons, J. Jordan, Standard Potentials in Aqueous Solution, MarcelDekkerInc.,New York &Basel, 1985. [1] a) A. Tsukazaki, A. Ohtomo, T. Onuma, M. Ohtani, T. Makino, M. Sumiya, [21] Q. M. Su, Z. M. Dong, J. Zhang, G. H. Du, B. S. Xu, Nanotechnology 2013, K. Ohtani, S. F. Chichibu, S. Fuke,Y.Segawa, H. Ohno, H. Koinuma, M. 24,255705. Kawasaki, Nat. Mater. 2005, 4,42–46;b)D.I.Son, B. W. Kwon, D. H. [22] Y. Sharma,N.Sharma, G. V. S. Rao, B. V. R. Chowdari, Adv. Funct. Mater. Park, W. S. Seo,Y.Yi, B. Angadi, C. L. Lee, W. K. Choi, Nat. Nanotechnol. 2007, 17,2855 –2861. 2012, 7,465 –471. [23] a) F. Wang,R.Robert, N. A. Chernova, N. Pereira,F.Omenya, F. Badway, [2] a) N. Han, L. Y. Chai, Q. Wang, Y. J. Tian,P.Y.Deng, Y. F. Chen, Sens.Ac- X. Hua, M. Ruotolo, R. G. Zhang, L. J. Wu,V.Volkov,D.Su, B. Key,M.S. tuatorsB2010, 147,525–530;b)J.X.Wang, X. W. Sun, Y. Yang,H. Whittingharn, C. P. Grey,G.G.Amatucci, Y. M. Zhu, J. Graetz, J. Am. Huang,Y.C.Lee, O. K. Tan, L. Vayssieres, Nanotechnology 2006, 17, Chem. Soc. 2011, 133,18828–18836;b)F.Wang, H. C. Yu,M.H.Chen, 4995 –4998. L. J. Wu, N. Pereira, K. Thornton, A. Vander Ven, Y. M. Zhu, G. G. Amatuc- [3] a) M. Grätzel, Nature 2001, 414,338 –344;b)M.Law,L.E.Greene, J. C. ci, J. Graetz, Nat. Commun. 2012, 3,1201. Johnson, R. Saykally,P.D.Yang, Nat. Mater. 2005, 4,455 –459. [24] a) A. D. W. Todd, R. E. Mar,J.R.Dahn, J. Electrochem. Soc. 2006, 153, [4] a) G. Z. Yang, H. W. Song, H. Cui, Y. C. Liu, C. X. Wang, Nano Energy 2013, A1998–A2005;b)J.R.Dahn, R. E. Mar,A.Abouzeid, J. Electrochem. Soc. 2,579–585;b)N.Li, S. X. Jin, Q. Y. Liao, C. X. Wang, ACS Appl. Mater.In- 2006, 153,A361–A365;c)A.D.W.Todd, R. E. Mar,J.R.Dahn, J. Electro- terfaces 2014, 6,20590 –20596;c)Y.Sun, G. Z. Yang, H. Cui, J. Wang, chem. Soc. 2007, 154,A597–A604;d)Q.Fan, P. J. Chupas, M. S. Whit- C. X. Wang, ACS Appl. Mater.Interfaces 2015, 7,15230 –15239;d)O.B. tingham, Electrochem. Solid-State Lett. 2007, 10,A274 –A278. Chae, S. Park, J. H. Ryu, S. M. Oh, J. Electrochem. Soc. 2013, 160,A11 – [25] a) M. N. Obrovac, L. Christensen, Electrochem.Solid-State Lett. 2004, 7, A14;e)G.L.Xu, Y. Li, T. Y. Ma, Y. Ren, H. H. Wang, L. F. Wang, J. G. Wen, A93–A96;b)B.Key,M.Morcrette, J. M. Tarascon, C. P. Grey, J. Am. D. Miller,K.Amine, Z. H. Chen, Nano Energy 2015, 18,253–264;f)B.Os- Chem. Soc. 2011, 133,503–512. chmann, M. N. Tahir, F. Mueller,D.Bresser,I.Lieberwirth, W. Tremel, S. [26] a) F. Mueller, D. Bresser,N.Minderjahn,J.Kalhoff,S.Menne, S. Krueger, Passerini, R. Zentel, Macromol. Rapid Commun. 2015, 36,1075 –1082. M. Winter,S.Passerini, Dalton Trans. 2014, 43,15013 –15021; b) B. Liu, J. [5] a) F. Belliard, P. A. Connor,J.T.S.Irvine, Ionics 1999, 5,450 –454;b)M. Zhang, X. F. Wang, G. Chen,D.Chen, C. W. Zhou, G. Z. Shen, Nano Lett. Ahmad,Y.Y.Shi, A. Nisar, H. Y. Sun, W. C. Shen, M. Wei, J. Zhu, J. Mater. 2012, 12,3005 –3011; c) S. Laruelle, S. Grugeon, P. Poizot, M. Dolle, L. Chem. 2011, 21,7723–7729;c)H.B.Wang, Q. M. Pan, Y. X. Cheng, J. W. Dupont, J. M. Tarascon, J. Electrochem. Soc. 2002, 149,A627–A634;d)S. Zhao, G. P. Yin, Electrochim. Acta 2009, 54,2851–2855;d)K.T.Park, F. Grugeon, S. Laruelle, L. Dupont, J. M. Tarascon, Solid State Sci. 2003, 5, Xia, S. W. Kim, S. B. Kim, T. Song,U.Paik, W. I. Park, J. Phys. Chem.C 895–904;e)A.Ponrouch,P.L.Taberna, P. Simon, M. R. Palacin, Electro- 2013, 117,1037–1043;e)F.Belliard, J. T. S. Irvine, J. Power Sources 2001, chim. Acta 2012, 61,13–18. 97–98,219 –222;f)W.J.Zhang, J. PowerSources 2011, 196,13–24. [27] a) R. Raccichini, A. Varzi, S. Passerini, B. Scrosati, Nat. Mater. 2015, 14, [6] C. J. Pelliccione,Y.J.Ding, E. V. Timofeeva, C. U. Segre, J. Electrochem. 271–279;b)Y.R.Zhong, M. Yang, X. L. Zhou,Z.Zhou, Mater.Horiz. Soc. 2015, 162,A1935 –A1939. 2015, 2,553 –566. [7] a) D. Bresser,F.Mueller,M.Fiedler,S.Krueger,R.Kloepsch,D.Baither,M. Winter,E.Paillard, S. Passerini, Chem. Mater. 2013, 25,4977–4985;b)F. Mueller,D.Bresser,V.S.K.Chakravadhanula,S.Passerini, J. PowerSour- ces 2015, 299,398–402. Manuscript received:April 6, 2016 [8] a) P. Poizot, S. Laruelle, S. Grugeon, L. Dupont, J. M. Tarascon, Nature AcceptedArticle published:May 19, 2016 2000, 407,496 –499;b)G.X.Wang, Y. Chen, K. Konstantinov, M. Lindsay, Final Article published:June 17, 2016

ChemElectroChem 2016, 3,1311–1319 www.chemelectrochem.org 1319 2016 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim