Utilizing Cyclic Voltammetry to Understand the Energy Storage Mechanisms for Copper Oxide and its Graphene Oxide Hybrids as Lithium‐Ion Battery Anodes

Abstract Graphene‐based materials have been extensively researched as a means improve the electrochemical performance of transition metal oxides in Li‐ion battery applications, however an understanding of the effect of the different synthesis routes, and the factors underlying the oft‐stated better performance of the hybrid materials (compared to the pure metal oxides) is not always demonstrated. For the first time, we report a range of synthetic routes to produce graphene oxide (GO)‐coated CuO, micro‐particle/GO “bundles” as well as nano‐particulates decorated on GO sheets to enable a comparison with CuO and its carbon‐coated analogue, as confirmed using scanning electron microscopy (SEM) imaging and Raman spectroscopy. Cyclic voltammetry was utilized to probe the lithiation/delithiation mechanism of CuO by scanning at successively decreasing vertex potentials, uncovering the importance of a full reduction to Cu metal on the reduction step. The GO hybrid materials clearly show enhanced specific capacities and cycling stabilities comparative to the CuO, with the most promising material achieving a capacity of 746 mAh g−1 and capacity retention of 92 % after 30 cycles, which is the highest stable capacity quoted in literature for CuO. The simple cyclic voltammetry technique used in this work could be implemented to help further understand any conversion‐type anode materials, in turn accelerating the research and industrial development of conversion anodes.


Introduction
The performanced emands of future energy storage applications have led to considerable research on alternatives to current electrode materials and battery chemistry.A lthough Li-ion battery (LIB) capacity is limited by the cathode materials, significant effort is being expended to develop alternative anode materials to the industry standard, graphite. Graphite has been the anode materialo fc hoice for LIB since it wasfirst utilizeda s aL ii ntercalation compound by Yazami and To uzain in 1983; [1] however,t his material suffers from safety issues when used in battery systems such as Li dendrite formation at the low lithiation potential of graphite (< 0.1V vs. Li/Li + ), [2] leading to potential thermalr unaways. There are also environmental and supply concernsr elatedt ot he mining of graphite, causing multiple closures to Chinese graphite mines in 2018. [3] Transition metal oxides have been investigated extensively as an alternative LIB anode material to graphite, because of their high theoretical lithiation capacities, low cost and-in certainc ases-ubiquity and low environmental "footprint". CuO is ap romising materiali nt his regard as it fits the criteria above and has at heoretical capacity (674 mAh g À1 )a lmost double that of graphite (372 mAh g À1 ). [4] Unlike the insertion mechanism of graphite, transition metal oxidess tore Li ions via conversion reactions, which is depicted in its most general form below: This multi-electront ransfer processi sl ikely to proceed via two steps in the case of Cu, that is, reduction to Cu I ,f ollowed by formation of metallic Cu. The theoreticalc apacity of CuO has been calculated using Faraday's law assuming complete reductiont om etallic Cu. An otable fade in specific capacity of CuO anodesisreported, which has been attributedt olarge expansion/contraction of particles during lithiation/delithiation Graphene-based materials have been extensively researched as am eans improvet he electrochemical performance of transition metal oxides in Li-ion battery applications, however an understanding of the effect of the different synthesis routes,a nd the factors underlying the oft-stated better performance of the hybrid materials (compared to the pure metal oxides) is not always demonstrated. For the first time, we report ar ange of synthetic routes to produce graphene oxide (GO)-coated CuO, micro-particle/GO" bundles" as well as nano-particulates decorated on GO sheetst oe nable ac omparison with CuO and its carbon-coated analogue, as confirmed using scanning electron microscopy (SEM) imaging and Ramanspectroscopy.C yclic voltammetry was utilized to probe the lithiation/delithiation mechanism of CuO by scanninga ts uccessively decreasing vertex potentials, uncovering the importance of af ull reduction to Cu metal on the reduction step. The GO hybrid materials clearly show enhanced specificc apacities and cycling stabilities comparative to the CuO, with the mostpromisingm aterial achieving ac apacity of 746 mAh g À1 and capacityr etention of 92 %a fter 30 cycles,w hich is the highest stable capacity quoted in literature for CuO. The simple cyclic voltammetry technique used in this work could be implemented to help further understand any conversion-type anode materials, in turn accelerating the research and industrial development of conversion anodes.
cycles, leadingt op ulverisation of the electrode and ad ecrease in capacity. [5] An important breakthrough in the understanding of transition metal oxides as LIB anodesw as the seminal work from Ta rascon's group, whichd emonstrated as trong correlation between increasing reversible capacity and metal oxide particle size, with the most stable performance seen when micronscale metal oxide particles were cycled, forming few nanometer diameter metal particles. [6] Moreover,t he voltage range of the cycled electrodes was found to strongly influence the cycling stability,w itht he best results obtained when the cells were cycled to the lowest (0.01 Vv s. Li)v oltage. As ubsequent study by the same group used in situ transmission electron microscopy (TEM) to elucidate aspects of the reactionm echanism which, in the case of Cu, was suggested to consist of sequential reductionsa t2 .1, 1.6, and 1.2 Vv s. Li, attributed to the formation of ap artially lithiated phase (Li x CuO, x < 0.4), CuO 2 (when x > 0.4), and Cu (when x > 0.8), respectively.Extra capacity observed, that is, for x > 2, was attributed to formation of an "organicf ilm". [7] Another importantc ontribution was made by Grey and co-workers, who used synchrotron X-ray and solidstate NMR methods leadingt hem to attribute the additional capacityt othe reduction of the Li 2 Ot oL iH. [8] Incorporation of graphene, or graphene oxide (GO), with CuO during electrode preparation has been found to increase the stable capacity and the cycling stabilityo ft he electrodes, althought he reasonsf or this improvement in performance are not always obvious. [9,10] Clearly,i ntroduction of ac onducting support is beneficial, since CuO is as emiconducting material. Graphenew ill also help to stabilizes maller scale particles of CuO, which would otherwise be likely to aggregate andh ence reduce the reversibility of the lithiation/delithiation process. An additional factor of potentialr elevance is the formationo ft he Li 2 O[ see Eq. (1)],w hich possess very low electronic conductivity, [11,12] hence further inhibitingt he reversibilityo ft he lithiation process:a gain, small particles, corresponding to thinner Li 2 O layers, would mitigate this.
The specific changes in response to graphene incorporation are dependent on the respective morphologies of the metal oxide and graphene sheets,az oo of differing geometries have been described. [13] Ar elativelye arly report by Rai et al. [14] described the formation of ar educed GO/CuO composite by a physical( milling) method. Although an initial capacity improvement was observed over the pure CuO, this decayed rapidly and within 20 cycles the capacity had fallen to the level of the pure metal oxide. Ah ydrothermal composite preparation also gave an improved capacity (ca. 500 mAh g À1 ,c ompared to ca. 300 mAh g À1 for the pure metal oxide), which showeda greater cycle stability. [15] Clearly,given the foregoing discussion, am etal oxide with high surfacea rea and intimatec ontact between metal oxide and graphene are necessary for an electrochemically reversible process. Thee ffect of CuO microstructure on cell performance has been reported for the pure CuO (graphene-free) case, [16] but to the best of our knowledge,n os ystematicc omparison between preparation method, electrode structurea nd cell performance has been reportedf or the composite case. In this work, we attempt to address this by pre-senting as ystematic comparison of the performance of CuO/ GO composite anodes prepared in three differentw ays, benchmarked against CuO and ac arbon-coated CuO. Particle morphologyi sc ompared with electrochemical performance, which includes half-cell cycling data and cyclic voltammetry as afunction of vertex potential, to assess the relationship between the extent of Cu reduction the reversibility of the reduction.

Experimental Section
Cu-GO hybrid materials were prepared in five ways. Cu 2 (OH) 2 CO 3 and GO were supplied by the industrial partner,William Blythe. The sample denoted Cu-1 consists of aC uO that was synthesized via ac alcination process of Cu 2 (OH) 2 CO 3 .I ns hort, Cu 2 (OH) 2 CO 3 (4 g) was calcined in air at 300 8Cf or 4hto produce CuO. Cu-2 was produced by mixing CuO (3 g) from the Cu-1 method and glucose (36 g) at room temperature for 4.5 h, filtering and then drying at 110 8Cf or 16 h. The dried product was then thermally treated in air at 300 8Cf or 4h.C u-3 was produced by adding the commercially supplied Cu 2 (OH) 2 CO 3 (5 g) to aG Od ispersion (100 mL, 2.5 mg mL À1 )a nd mixing for 3hat room temperature. The product was then filtered, filter cake placed into the freezer at À16 8Cf or 16 hb efore freeze drying to prevent GO aggregation upon drying. The product was then calcined at 300 8Cf or 4h. Cu-4 was synthesized by incorporating 5wt% of GO into the synthesis of Cu 2 (OH) 2 CO 3 ,f iltering, then drying, the Cu 2 (OH) 2 CO 3 /GO hybrid in av acuum oven at 110 8Co vernight. The hybrid was then calcined in air at 300 8Cf or 4hto produce CuO/GO. Cu-5 was produced by dissolving Cu(NO 3 ) 2 .3H 2 O( 3.2 g) and urea (1.21 g) in deionized water (50 mL). This solution was then slowly added to aG Od ispersion (50 mL, 1.8 mg mL À1 )w ith vigorous stirring and left to mix for 20 min. The solution was then transferred to a2 00 mL Te flon-lined stainless-steel autoclave and treated at 130 8Cf or 12 h. The product was then filtered and washed until the conductance of the filtrate was below 100 mS, then placed into the freezer at À116 8Cf or 16 h. The frozen filter cake was then freeze dried to produce af ine powder of Cu 2 (OH) 2 CO 3 /GO. The composite was then calcined at 300 8Cf or 4hto produce CuO/GO The anode materials were characterized by scanning electron microscopy (SEM) performed on aF EI Quanta 650 FEG ESEM. X-ray diffraction (XRD) was performed using aP roto AXRD Benchtop with 2q scanned from 108 to 808 with Cu Ka radiation. Raman spectra of the anode materials was recorded with aR enishaw inVia Raman Microscope, using 532 nm laser excitation and a1 00 objective, operating at an intensity of 10 %. Thermal gravimetric analysis (TGA) was used to assess the relative carbon content and performed using aM ettler To ledo TGA/DSC 1S TAR e System in the presence of air.P article size was measured using aM alvern Mastersizer 2000 DLS particle sizer,w here D50 is defined as the average particle diameter. Electrochemical testing was completed in LIB half-cells in as tandard coin cell configuration versus aL if oil (all cell potentials quoted are therefore relative to this electrode). The anodes were produced by coating as lurry of Cu material, polyvinylidene difluoride (PVDF), and ac ommercial carbon black, Super-P,( 8:1:1) in Nmethyl pyrrolidinone onto aC uc urrent collector (14 mmt hickness) before drying under vacuum at 80 8Cf or 16 h. Anode discs (15 mm) were punched and the 2023-type coin cells were assembled in an Ar-filled glovebox. The electrolyte was 1 m LiPF 6 dissolved in am ixture of ethylene carbonate (EC) and dimethyl carbonate (DMC;1:1 volume ratio). The mass loading and thickness of the anode were approximately 3mgcm À2 and 60 mm, respectively. The PVDF was supplied by Kynar,S uper-P by Alfa Aesar and LiPF 6 EC/DMC by Gelon LIB Group. Cell performance was assessed via the capacity and cycling performance using aB asytec Cell Te st System at aC -rate of 0.2 C (based on the theoretical capacity of CuO), whereas cyclic voltammetry,r ecorded with an Autolab PGSTAT302N, was used to reveal some of the details of the reversibility of the anode reactions. Cyclic voltammograms are plotted as specific current (I sp ), which is defined as the current per unit mass of electrode material versus cell voltage. Figure 1s hows the XRD patternso ft he CuO-based electrode materials, which can all be indexedb yt he monoclinic symmetry (JCPDS Card No. 05 0661) described by the space group C2/c. The presence of small amounts (< 5% determinedv ia extended calcination) of Cu 2 (OH) 2 CO 3 is apparent in samples Cu-1a nd Cu-4, based on the reflectionss een at 2q = 158,1 8 8,2 4 8, and 318. [17] Although undesirable, this material is expectedt o show similar chemistry to CuO, with respectt ol ithiation. We note that previousliterature has described the use of transition metal carbonates an anode material. [18] The absence of the GO peak at 2q % 118 in Cu-3, Cu-4, and Cu-5 indicatest hat the graphene is present as mono/few layers in the graphene hybrid materials.

Results and Discussion
The carbon content of the Cu materials was confirmed with TGA as depicted in Figure 2. Cu-2 hadav ery low carbon content, lesst han 1wt%,b ased on the TGA data. In contrast, the carbon content of Cu-3 was 5.2 %, Cu-4 was 5.5 %, and Cu-5 was 9.9 %. GO was introducedi nto Cu-3 and Cu-4 at 5wt% during synthesis so this data confirms that all GO has successfully been incorporated into the material, enabling ad irect comparison of the electrochemical performance of these materials. Cu-5h as ag raphene content close to 10 %, which,a lthoughi td oes not allow direct comparison to the other hy-brids, is still an appropriate weight percentagew hen considering the expected specific capacity losses owing to reduced mass of active material, and also economic factorst hat come with introducing graphene into ab attery material.
Ramans pectroscopy was used to confirm the presence of graphene in the hybridm aterials. Figure3ae xhibits the three characteristic Ramana ctive modes of CuO at 275 cm À1 (A g ), 330 cm À1 (B g ), and 610 cm À1 (B g ), [19] which are present in all Raman spectra. The three GO hybrid materials all display the characteristic Da nd Gp eaks of GO. The Gp eak at 1590 cm À1 corresponds to the in-phase vibrations from the graphite lattice E 2g mode, whereas the Dp eak at 1350 cm À1 is caused by disorder on the basal plane created by the oxygen functional groups. [20,21] The morphologyo ft he CuO-based materials was compared using SEM imaging. Cu-1 ( Figure 4a)d epicts spherical secondary particles that consist of micro-rodsn ucleating from ac entral point within the particle. The spherical nature of the CuO particles is desirable for battery applications as the spheres   allow for betterp article packing within the electrode, leading to greater connection between electrode components and therefore allowing greater volumetric energy densities to be achieved.. [22] The anisotropy of the microrods could also increase the rate capability of the electrode by providing shorter pathways for Li + diffusion in and out of the particle. As expected, Cu-2 ( Figure 4b)h as av ery similar morphology to Cu-1 resulting from it being the same materialw ithalow percent content of carbonc oating.C arbon coating is ac ommon technique used to increase the performance of battery materials by enhancingt he electricalconductivity of the particles, so this materialw ill enablet he evaluation of the Cu-3 graphene composite in comparison to industry standards. [23] Cu-3 ( Figure 4c)c learly showsw rapping of the particlei n GO sheets,i ndicating that van der Waals forces are present between the carbon groups on the GO and the oxygen on the Cu allowing the hybrid particlest of orm during mixing.T his encapsulation could have a2 -fold benefit on the electrochemical performance of the materials. Firstly,b yi ncreasing the electronic conductivity of each particlee nabling an increased charge transfer throughout the electrode, and secondlyb y providing as trong graphene coating thatl imits volumetric expansion.I fr ealized,t hese effects would in turn enable greater energy densities and cycling stabilities to be achieved. [24] Cu-4 ( Figure 4d)h as av astly different morphology in comparison to Cu-3, in which the synthesis procedure has inhibited the growth of these secondary primary particles to create hybrid structures of primary micro-rod particles within GO sheets.
The segregationo ft he micro-rodsw ould further increase the 3D conductive network of graphene within the electrode, potentially enabling even greater energy densities to be achieved. [25] HybridC u-5 was synthesized via hydrothermally precipitating nanoparticles of CuCO 3 onto GO sheets as can be seen in Figure 4e.T he productiono fn anoparticles could increase the reaction kinetics for the decomposition of the usually unreactive Li 2 Oo nt he re-oxidation of Cu. [26] However,a s previously stated,astrongc orrelation between CuO particle size and cycling stabilityw as reportedi nt he work of Grugeon et al.,w ith the nanometer-scale oxide particlesh aving ad etrimental effect on cycling performance. [27] As noted in the introduction, the mechanism of CuO reduction in the LIB cell has been discussed by Debart et al., [7] using in situ TEM to aid their interpretation of the electrochemical data. Figure 5p resentst he cyclic voltammetric dataf or the CuO electrode. The small "pre-peak" seen at approximately 2.3 Vc orresponds to as imilar feature observed at as imilarp otential in Ref. [7].T he retention of CuO structurel ed to the assignment of alimited lithiation of the oxide. The second reduction feature, am uch more prominent one with ap eak potential in the range 1.2 to 1.0 V, was attributed to Cu 2 Ob yD ebart et al. By contrast, the work of Wang et al., [28] who used in situ TEM to study lithiation of CuO nanowires, suggested that the first cycle lithiation involved ac onversion from CuO to Cu, which on delithiation, disproportionates to Cu 2 Oa nd CuO.
Note that, fora ll the voltage ranges explored in Figure 5, a prominentd ifference between the first and subsequent voltammetricc ycles is not observed, which supports the mechanism proposed by Ta rascon and co-workers, [7] rather than disproportionation, on the lithiation cycle. The chargea ssociated with the discharge cycle is approximately independento fc ycle number at all vertex potentials;h owever, the reduction peaks are "sharpened"( i.e.,o ccur over an arrow potential range) by proceeding to lower vertex potentials. This sharpening could be ar esult of ag reater amount of residual Cu and Cu 2 Op resent after the oxidative sweep arisingf rom an increase in the amount of CuO reductiona tl ower potentials (as evidenced by the increased specific capacities in Figure 6a). The presence of Cu and Cu 2 Oc rystallographic phases will increaset he rate of reduction of the CuO, and the Cu will increase the conductivity of the electrode, leading to the reactions occurring within a more limited potentialand sharper voltammetric peaks.
In as eparatep ublication, Laik et al. calculated the cell potentials (i.e.,t hermodynamics of reduction of metal oxides to metals,r elative to Li oxidation), showing that the reduction of ar ange of transition metal oxidesi st hermodynamically feasible, ar eductionp otentialo f2 .22 Vw as calculatedf or the case of CuO vs. Li + /Li. These authors note that all of these potentials are considerably higher than the observedv oltages (c.f. 1.0-1.2 Va bove for this process), ad iscrepancy attributed to the slow kinetics of the conversion reaction. [29] The second prominent reduction peak at 0.6 V, only visible in Figure 5a-c, is attributed to Cu 2 Or eduction to Cu. One peculiarity of the LIB system is that the only "sink" for the oxide ions, formally generated on the reductiono fC u, is the incoming Li + ion, as protons are absent andt he battery solvento nly decomposesa tp otentials below those of cuprous/cupric ion reduction.F inally,as mall feature is observed at potentials below 0.4 V: this is consistent with the additional charge attributed by Ta rascona nd co-workers [7] to the formation of an "organic film", but later reported by Grey and co-workerst ob e linked with further reduction of the lithiated products to LiH as an ew "inorganic" layer. [8] The reverse part of the cyclic voltam-mograms of Figure 5, corresponding to the delithiation process, shows that the main delithiation process takes place at  cell voltages above 2.0 V. As mall pre-peak is seen for all vertices lower than or equal to 0.9 V, that is, for all cases except Figure 5d.T arascona nd co-workers [7] attributed this feature to the reversal of the "organicf ilm" formationp rocess seen at the lowest voltages,b ut the observation of this pre-peak on the reversec ycle at av ertex potential of 0.5 V ( Figure 5c)s uggests that it is not directly relatedt ot he "extra" chargef ormation process seen at < 0.4 V. The position of this peak is, however, vertex dependent:i ts maximum lies between 1.3 and1 .6 V, with the shift to the more positive voltages seenf or the increased( more positive) vertex potentials. This indicatest hat this oxidative pre-peak could be ar esult of the reversible formation and decomposition of an solid electrolyte interface (SEI) layer,w hich typicallyf orms below 0.8 V. [30] Changing the vertex potentialt o0 .9 Vs upports this argument as it is only in this cyclic voltammogram that the oxidative pre-peak is absent.T he "organic" layer described by Ta rascon is therefore more likely to be typical SEI formation,w hich begins to form at 0.8 V, and the extra capacity at < 0.2 Vt he inorganic layer as described by Grey and co-workers. [8] The other notable feature of the "reverse" part of the CVs is the splitting of the main delithiation process into two separate voltammetric peaks, which is seen most prominently for the highest vertex (Figure 5d). The observation of a" merged"d elithiationp rocess is at variancew ith the reference [7] and suggests that the larger particles usedh ere undergo as ingle-step re-oxidation of metallic Cu to CuO. Interestingly,t he more pronounced "split" peak seen in Figure 5d suggests that the Cu 2 O speciesu ndergoes disproportionation to generate Cu and CuO, with the co-existenceo ft hese species favoring at wostep delithiation.
Ac alculation of the charge/discharge ratio from the potentiostatic data at the different cycle numbersi ss hown in Figure 6b.A tt he lower vertex potentials (0.01, 0.2, and 0.5 V) the coulombice fficiencies are in the range of 60-70%,s uggesting an incomplete conversion of Cu backt oC uO. This suggests that the degradation pathway for CuO anodes proceeds via incomplete oxidation of Cu and coupled with-possibly driven by-incomplete conversiono fL i 2 Ob ack to Li + ,r esulting in excess insulating Li 2 Or emaining at the anode. Thisr emovesL i from the system andd ecreases the electrical conductivity of the electrode.
However,a t0 .9 Vt he coulombic efficiencyi sa sh igh as 171 %o nt he 1 st cycle and 132 %o nt he 2 nd before decreasing to 97 %o nt he 10 th cycle. This extra charge gain was only measured in the potentiostatic mode;g alvanostatically the coulombic efficiency remained below 100 %o na ll cycles.C oulombic efficiencies greater than 100 %c an typically be attributed to undesired side reactions/shuttling processes, the latter being commoni no ther cell systemss uch as aL iS. [31] Although it is unlikely that shuttling would occur in this system,adissolution of CuO and platingo fC u 2 + to Cu on the electrode prior to charging would give rise to al arger portion of reduced Cu availablef or oxidation to give the two peaks as depictedi n Figure 5d.T his theory is however speculative and would require further investigation to understand the mechanism giving rise to this additional capacity.
Figure6as hows the specific capacity at each potential range as af unction of cycle number.A t0 .01 Vani nitial capacity of 711mAh g À1 was obtained being greater than the theoretical( 674 mAh g À1 ), presumably as ar esult of the inorganic layer formation. This then drops to 654 mAh g À1 with ac apacity retention of 86 %a fter 30 cycles. As expected, increasing the lower vertex potentiald ecreases the specific capacity as the extent of lithiationt hat occurs in that potentialw indow is decreased. When the vertex potential was increased to 0.2 Va similar initial capacity was obtained; however,t he capacity on subsequentc ycles decreased at am uch quicker rate (62 %r etention)i nc omparison to the 0.01 Vc ell. Considering that the only known mechanism occurring below 0.2 Vi st he formation of the additional inorganic layer,t hese observations suggest that this new inorganic layer is crucial to the cycling stability. This capacity retention degrades at even faster rates at the 0.5 V( 41 %r etention) and 0.9 V( 22 %r etention) vertex potentials. An explanation for the low retention at 0.9 Vc ould be that if the disproportionation of Cu 2 Ot oC ua nd CuO does occur,t hen this decreases the fraction of Cu II available for reductiono ns ubsequentc ycles, introducing an ew degradation pathway for the electrode. Alternative reasoning could be owing to the lack of SEI layer formation, which is known to enhance the cyclabilityo fa na node materiali fastable SEI is formed. [32] Cu-2 was investigated through the same cyclic voltammetry testingp rocedure (Figure 7): the discussion of the mechanism is related to the conclusions drawn for Cu-1. Distinct differences are presenta tt he 0.01 and 0.2 Vv ertex potentials when compared to Cu-1. Firstly,t he reduction peaks at 1a nd 0.6 V are much broader and overlap to almostf orm one reduction peak. Secondly, the oxidation peaks that typically arise % 2.5 V have becomem ore separated with the Cu 2 Ot oC uO conversion becomingl ess favorable as highlighted by the shift in peak maximat oh igherp otentials. It appearst hat the carbon coating has reduced the reversibility of the conversionr eactions at the lower vertex potentials, with the electronic conductivity enhancements being outweighed by the extra impedance to Li + ion diffusion out of the particles after the initial Cu oxidation.
Acommon feature at the 0.9 V (Figure 7d)vertex is the presence of the two oxidative peaks at 2.4 and 2.8 V, further supporting the hypothesis of Cu 2 Od isproportionation.T hese peaks again give rise to ap otentiostatic coulombic efficiency (Figure 8b)g reater than 100 %o na ll cycles,d emonstrating that this feature is not an artefact of the Cu-1 material, electrode or cell, but common throughout multiple materials at the 0.9 Vv ertex.
An investigation into the cyclability of Cu-2 yieldedp oorer results than for the uncoated material at the 0.01 Vv ertex with an initial capacityo f6 86 mAh g À1 and ar etention of 80 %a fter 30 cycles.T hisconfirmsthe findings from theCVs that indicated ap oorerr eversibility at thel owestv ertexp otential.T he presence of thec arbonc oating didh owever increase thec yclability of theC uO in ther emaining 0.2( 80 %) 0.5( 70 %),a nd 0.9V (28%)w henc omparedt oC u-1, concurring with what is commonlyq uoted in literature. [24] It couldb et hatt here areu nfavor-     ( Figure 9a)d isplays as imilar shape to Cu-1, with reductionp eaks at 2.3, 1.0, and 0.6 V, the latter two shift to higher potentials with increasing cycle number,t herefore becomingm ore disfavored. This indicates that the GO coating does not give rise to a1 00 %c apacity retention. One potential improvement to the cycling stability when compared to Cu-1 is the that the peak at 0.6 Vc orresponding to the reduction of Cu 2 Ot oC ud oes not shift to al ower potentiala st he vertex potentiali ncreases from 0.01 to 0.9 V. Another feature to note in Figure 9a is the merging of the oxidation peaks instantaneously at 2.6 V, suggesting that the GO coating is enhancingt he single-step oxidation of Cu to CuO.
It was previously reported that graphene coatings can enhance the electron charge transfer and Li + diffusivity in and out of active materials; [33] therefore, the GO could be playing both roles in this system,e nabling greater accessibility to un-reactedL i 2 Oa nd an increase in reactionr ate. This peak merging is concurrent at all vertex potentials, in contrast to what is seen at 0.9 Vi nt he Cu-1 material. This suggests that the only process occurring in Figure 9d is the re-oxidation of the Cu 2 O back to CuO at 2.6 Vw ithoutt he disproportionation of the Cu 2 Ot oC uO and Cu that was proposed in Cu-1.
The potentiostatic coulombic efficiency (Figure 10 c) remained between 100-104 %o na ll cycles. When taking into account the error in determining the area under the curve for separatec harge andd ischarge steps,t his result is not significant enough to conclude that extra capacity is being gained on the charge step. This suggests as trong correlation between the two-stage oxidation in Figure 5d and the extra capacity gainedo nt he charge step giving rise to a > 100 %c oulombic efficiency.A nother discrepancy for the 0.9 Vv ertex in Figure 9d,c ompared to Figure 5d,i st he emergence of ar eduction "pre-peak" at 1.5 Vo nt he 10 th cycle. This peak could be attributed to stronger interactions that form between the GO and CuO during cycling. Previous studies have shown that greater oxidative catalytic activities can be achievedf or CuO when combined with GO to form ah ybrid material. [34,35] Therefore, the Cu atoms on the surface of the CuO particle in contact with GO could be undergoing af aster rate of reduction to Cu 2 Oh ence the process is more favorable at the higher poten- tial. One final point to note is the SEI decomposition that arises on the oxidative scan. With Cu-3 there are two clearly defined oxidative peaks at 0.9 and 1.5 V( note that these are also seen with Cu-1 but are less defined) suggesting that there are two different chemicalc ompositions of SEI formingo nt he CuO, which in turn decompose at different potentials. Figure 10 apresents the cyclability of the Cu-3 hybrid material. At the lowest vertex potentialo f0 .01 Vani nitial capacity of 700 mAh g À1 was obtained with a9 5% retention of this capacity after 30 cycles, indicating that the GO coating has had as ignificant benefitt ot he cycling stability of the CuO. The slightly lower initial capacity here comparedw ith Cu-1 is most likely attributed to the lower mass of active CuO material( 95 %o f the mass of the hybrid). This increased cycling stability is also realized when cycled to the 0.2 and 0.5 Vv ertex potentials (84 %a nd 70 %r etention, respectively), supporting the above CV data that suggests that the GO enhances the reversibility of CuO reduction/oxidationa th ighervertex potentials.
An explanation for the loss in capacity for the first three vertex potentials can be determined by considering the potentiostatic coulombic efficiency (Figure 10 b), which again show variances between6 0-75 %f or all cycles suggestinga ni ncomplete conversion of Cu back to CuO, leaving unreacted Li 2 Oa t the electrode. At 0.9 Vt he capacity retention after 30 cycles is 30 %, which is only 8% higherthan for the pure Cu-1. This indicates that accelerated degradation could arise from al ack of SEI formation,t he CuO not being fully reduced to the Cu metal or via another unknown process. Figure 11 displays the CV potential rangesf or the Cu-4 hybrid that displayed ap rimary micron-sizedC uO within GO "bundles" morphology.T he CVs in Figure 11 a-c show very similar shapes to that in Figure 9a-c, indicating that similarC uO/ GO interactions are occurring, causing the reduction peak at 0.7 Vt or emain at the same potential when the vertex potential increases. In contrast to Cu-3 at the 0.9 Vv ertex (Figure 11 d), the oxidative peak at 2.6 Vi sm uch broader and could also be considered to consist of two distinct oxidative peaks with similar maxim oxidative peaks with similar maximum potentials. One other notable difference for Cu-4 is seen at the 0.9 Vv ertex potential in Figure 11 d. The additional reductive peak at 1.4 Vi sn ow visible after the earlier 5 th cycle compared to the 10 th in Figure 9d.T his can be rationalized after inspection of the morphologyo fC u-4 in Figure 4d,w hich displays am ore intimate mix of the primary micro-rods particles within aG O" bundle", whichc ould speed up the rate at which this accelerated reduction of CuO to Cu 2 Oo ccurs.
The cycling stability of Cu-4 is displayed in Figure 12 a. At the lowest0 .01 Vv ertex potential, as pecific capacity of 746 mAh g À1 was measured with a9 2% capacity retention after 30 cycles.Anumber of articles quote values > 1000 mAh g À1 for the first cycle for CuO graphene hybrids; however,t his then drops below 600 mAh g À1 after the 1 st cycle. Therefore, to the best of our knowledgew eb elieve this to be the only CuO/graphene material with as table additional capacity.
The cause of this additional capacity could be 2-fold:( i) the reduction of LiOH by Li + on CuO surface to form LiH and Li 2 O and (ii)the intercalation of Li + between the GO sheetsa sd emonstrated by Manthiram and co-workers, [36] which allows for capacities of over 800 mAh g À1 to be obtained as ar esult of increasedi nterlayer spacing. This Li + intercalation however should be visible on the CV as ac learly defined oxidative peak at 0.25 V; this is not observed in Figure 11 a. The extra capacity must be attributed to the inorganic layer formation occurring below 0.2 V. Aj ustification as to why this higher capacity was obtained for Cu-4 and not Cu-3 could be particles ize related. Cu-4 has am uch lower d50 (median value of the particle size distribution;h ere:8 .5 mm) comparedt oC u-3 (26 mm), which could lead to ag reat accessibility to the surface of the particles, allowing for al arger inorganic layer to be formed and therefore greater lithium storage.    Figure 11 di sa lso in agreement with this increased galvanostatic cycling stability.Areason for this could be relatedt ot he GO acting as an artificial SEI as ar esult of the primary particle coveragea ss een in Figure4d. It is commonk nowledget hat the formation of as table SEI layer can enhancet he cyclability of anode materials such as graphite and silicon; [32] however,a t the 0.9 Vv ertex we expect no SEI formation. Chen and coworkers demonstrated the use of GO-basedm aterials to form artificial SEI layers on transition metal oxidesf or use in LIBs, [37] so this type of artificial SEI could also have formed in Cu-4. We would not see this in the previous composite materials as the carbon/graphene coatings wereo nly presento nt he surface of the secondary particles and therefore would be an incomplete coverageo fa ll the availablep articles urface for SEI formation. It can also be noted that the potentiostatic coulombic efficiency (Figure 12 b) at the 0.9 Vv ertex remains above 100 %a fter the first cycle, furthers upporting the hypothesis of enhanced stability. Figure 13 displays the cyclic voltammetry for final hybrid material: Cu-5, which possessed the morphologyo fn ano-sized CuO particles deposited on the surface of GO sheets. Generally, the CVs show as imilar shapet ot he other graphene hybridsi n having little dependence of peak positiono nv ertex potential, and in the mergingo ft he oxidativep eak at 2.35 V. The cyclic voltammograms do appear to be more reversible than all materials in Figure 13 a-c,; however,t his stability collapses at the 0.9 Vv ertex. The peak maximum of the merged oxidative peak (Figure 13 d) is much lower than in the other CuO materials, suggesting the oxidation of Cu 2 Ob ack to CuO is more favorable in Cu-5 in the first cycle. However,i ns ubsequentc ycles the delithiation peak shifts back to 2.5 Vast he overall storage capacityd ecreases.I tc an also be noted that at this 0.9 V vertex we do not see the additional reductive peak at 1.5 V that was observed in Cu-3 and Cu-4, indicating al ess intimate mix of the CuO and GO. The coulombic efficiency at this vertex again exceeds 100 %, reaching 109 %i nt he 1 st cycleb efore decreasings teadily to 84 %b yt he 10 th cycle in agreement with the other four materials. Thisinstabilitya t0.9 Vcould be attributed to an incomplete coverage of CuO nanoparticles, which seem to have deposited onto the GO sheets rather than be embedded within aG Os tructure as can be confirmed for Cu-4.
The initial capacity of the 0.01 Vv ertex potentialw as 665 mAh g À1 and retained 91 %o ft his capacity after 30 cycles ( Figure 14). This is the lowest capacity achieved of all materials presented in this work;however,thesecannot be directly compared because of the higherc ontent of GO present in Cu-5, leadingt oa no verall decrease in active materialm ass. Despite having as imilar morphology to Cu-4, the cycling stability of Cu-5 is poorer at most vertex potentials (70 %a t0 .2 V, 50 %a t 0.5 V, and 35 %a t0 .9 V), which is in agreement with the Ta rascon article [27] demonstrating ap oorer cyclability with decreasedp article size.

Conclusions
We describe as ystematic comparison of CuO/graphene oxide (GO) composite materials for use as anodesi nL i-ion batteries synthesized via coating,p recipitation,a nd hydrothermalr outes to produce av ariety of morphologies, alongside ac omparison with ap ure CuO and ac arbon-coated CuO as is the current industry standard. Experimentally we have realized clear performance enhancements in both specific capacity and cycle stabilityt hrough the incorporation of GO in all three routes. Whereas aG Oc oating increased the cyclability of the CuO/GO hybrid,i tw as the precipitation route that produced the highest stable specific capacity know in literature at 746 mAh g À1 at 0.2 C. Using cyclic voltammetry at four different voltage ranges we have also further probedt he lithiation/delithiationm echanism of CuO, which has in turn enhanced understandingo f why the graphene hybrids possessgreater electrochemical performances in Li-ion battery applications.T he advanced CuO/ GO material offers ap otential environmentally benign alternative to the graphite anode as wella sam ethod to investigate the mechanism of transition metal oxides using as imple cyclic voltammetry technique.