Air electrode based on poly(3,4-ethylenedioxythiophene) microflower/graphene composite for superior Li–O2 batteries with excellent cycle performance

We report the use of poly(3,4-ethylenedioxythiophene) (PEDOT) microflowers as a potential electrode material for Li–O2 batteries. PEDOT is a conducting polymer that shows catalytic activity for the formation and dissociation of Li2O2 in Li–O2 cells. The microflower morphology composed of intertwined nanofibers is expected to increase the number of active sites for the redox reaction during the discharging-charging process. Although the electronic conductivity of the PEDOT microflowers is not so low, it is insufficient to achieve a large capacity for the electrode material in Li–O2 cells. Thus, a composite was prepared using 5–20 wt% of graphene as the matrix to compensate for the low conductivity of the PEDOT microflowers. The electrode employing the PEDOT microflower/graphene composite presents much higher capacity than that of the electrode employing pristine PEDOT microflowers, because of the enhanced electronic conductivity. Moreover, the electrode shows enhanced cycle performance compared to that of the electrode employing pristine graphene. This is because the microflower surface can suppress the side reactions activated by carbon, thereby reducing the accumulation of unwanted reaction products on the electrode.


Introduction
Lithium ion batteries (LIBs), which are based on the principle of intercalation chemistry, constitute an important energy-storage technology and nd widespread application in portable electronics. In particular, LIBs have enabled the transition of transportation power systems from combustion engines to electric motors. This transition is just an initial stage, but rapidly underway with an incursion of electric vehicles into the marketplace. However, the current battery chemistry based on intercalation cannot satisfy the requirements for energy-storage density in electric vehicles with sufficient travelling distance (aer a single charging process). Hence, much effort has been devoted to surpassing the limitations of existing LIBs. [1][2][3][4][5][6][7] Li-O 2 batteries (also called Li-air batteries) based on new battery chemistry, metal-molecule electrochemical redox couples, have received signicant interest owing to their remarkably high energy density. [8][9][10][11][12][13][14][15] However, the practical application of these batteries is limited by their inherent drawbacks such as high overpotential, low energy efficiency, and short cycle life, which are related to the basic reaction between the oxygen and lithium ions. [8][9][10][11][12][13][14][15] A typical non-aqueous Li-O 2 cell consists of an air electrode (cathode), organic electrolyte, and lithium anode. Upon discharging, molecular oxygen is reduced to superoxide, which combines with lithium ions at the air electrode, to form insoluble products (ideally Li 2 O 2 ). [16][17][18][19][20][21][22][23][24][25] The air electrodes act as the host matrix for the redox reactions and as storage sites for the reaction products; hence, the electrochemical performance of Li-O 2 cells is strongly dependent on the property of air electrodes. Upon charging, the previously formed reaction products should be completely dissociated so that reversible long cycling is achieved. However, in reality, the residual products gradually accumulate on the air electrode, clogging the pores on the electrode, and hindering the electrode reaction. This results in suffocation of the Li-O 2 cells, and consequently, poor cycle performance and high overpotential during cycling. [26][27][28][29][30][31][32][33] Note that most of these residual products are not Li 2 O 2 but undesired products such as Li 2 CO 3 and organics originating from the side reaction in the Li-O 2 cells. Li 2 O 2 can react with carbon in the air electrode at the Li 2 O 2 /carbon interphase to form Li 2 CO 3 . The electrolyte can decompose at high voltages too furnish organics such as CH 3 CO 2 Li and HCO 2 Li. 34-38 These unwanted products hardly dissociate even when the cell is charged at a high potential and easily accumulate on the air electrode surface. Unfortunately, these side reactions are inevitable as they are activated by carbon materials, which are commonly used for the matrix in air electrodes because of their high specic area, high electronic conductivity, and low cost.
In previous work, lm-type poly(3,4-ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS), a conducting polymer, has been proposed as a candidate material for air electrodes. 39,40 Such conducting polymers can be a suitable replacement to carbon owing to their good electronic conductivity, light weight, high catalytic activity, and high stability under oxygen reduction and evolution conditions. However, lm-type PEDOT:PSS cannot provide a sufficiently large surface area for the redox reaction. In the present study, poly(3,4-ethylenedioxythiophene) (PEDOT) microowers composited with a small amount of graphene (5-20 wt%) was prepared for use as the base material for a superior air electrode. The microower-like PEDOT particles offer a larger surface area as compared to the lm-type PEDOT layer. The PEDOT microower/graphene composite is expected to serve as an effective site for the redox reaction for between the oxygen and lithium ions, in addition to offering storage space for the reaction products. Moreover, the use of this composite helps in suppressing undesired side reactions as the electrode surface predominantly comprises the PEDOT layer, and not carbon. Consequently, a superior air electrode with better cyclic performance as compared that of carbonbased electrodes can be realised.

Synthesis of PEDOT microowers and PEDOT microower/ graphene composites
First, a ternary solution was prepared, composed of surfactant sodium bis(2-ethylhexyl) sulfosuccinate (AOT, Aldrich, 96%), 7 M aqueous solution of FeCl 3 (Aldrich, 97%), and p-xylene (Aldrich, anhydrous, $99%). The molar ratio (N) of water (used to dissolve FeCl 3 ) to the surfactant was adjusted to 30. 41,42 The surfactant AOT was dissolved in 20 mL p-xylene in a beaker to obtain a concentration of 1.5 M. Then the AOT/p-xylene solution was stirred for 20 min and diffused via ultrasonic treatment for 20 min. The 7 M aqueous FeCl 3 solution was added to the above mixture with constant stirring. Then a 3,4-ethylenedioxythiophene monomer (EDOT, Aldrich, 97%) was slowly dripped into the stirred solution to prepare PEDOT microower. To prepare the PEDOT microower/graphene composites, different amounts of commercial graphene (Graphene Supermarket Co., AO-1) was immersed in the AOT solution (5, 10, 20 wt% of PEDOT microower) before adding the EDOT monomer. This solution was reacted continuously for 24 h at room temperature. The reacted nal solution was then isolated by centrifugation and was washed with ethanol and a waterethanol mixture (1 : 1 v/v) several times to remove residual materials until the supernatant was colourless and transparent. Finally, the residual product dried under vacuum at 80 C for 24 h.

Electrochemical tests
The electrodes employing pristine PEDOT microower (without graphene) and PEDOT microower/graphene composites were composed of 100% sample powder (without binder). However, the pristine graphene electrode was prepared by mixing 90 wt% pristine graphene with 10 wt% PVDF binder. The loading weight of the electrodes was adjusted to 0.3 mg AE 0.03 mg. The electrochemical performance of the electrodes was studied using a modied Swagelok cell consisting of an air electrode, a metallic Li anode, a glass lter (Whatman) separator, and an electrolyte containing 0.5 M LiTFSI and 0.5 M LiNO 3 in tetraethylene glycol dimethyl ether (TEGDME). The cells were assembled in an Ar-lled glove box and were subjected to galvanostatic cycling using a WonATech battery cycler (WBCs 3000). A potential range of 2.0-4.35 V (vs. Li/Li + ) and various current densities (20,100,200, and 400 mA g electrode À1 ) were used for the electrochemical test. All experiments were conducted under an O 2 atmosphere under ambient pressure.

Characterization of the electrode materials
Transmission electron microscopy (TEM, JEM-2100F (HR)) and scanning electron microscopy (SEM, Nova NanoSEM 450, FEI Co.) were employed to observe the shape of the PEDOT micro-ower and the PEDOT microower/graphene composites. In addition, Fourier-transform infrared spectroscopy (FTIR, Nicolet 5700, Thermo Electron Corp.) was performed to examine the formation of samples. 4-Point probe system (CMT-SR2000N, Advanced Instrument Technology) was also used to measure the resistivity of the plat electrode composed of PEDOT micro-owers or PEDOT microower/graphene composites. Furthermore, the surface morphologies of the air electrodes before and aer cycling were observed using the above-mentioned SEM and FTIR techniques. To prepare the cycled specimens for SEM and FTIR, the air electrodes were collected aer 50 cycles and washed several times with dimethyl carbonate (DMC), and then stored in vacuum chamber for 24 h. The specimens were packed in vacuum before being transferred to the instruments.

Results and discussion
SEM and TEM images ( Fig. 1a and b) revealed that the pristine PEDOT microowers (prior to composite formation with graphene) were composed of intertwined nanobers. The surface of the particles had thorn-like protrusions, as illustrated in Fig. 1c, which provided a large surface area for the storage of reaction products during the cycling of the Li-O 2 cell. Moreover, each cluster had numerous pores that may allow access to the electrolyte containing Li ions and dissolved oxygen. The size of each PEDOT microower cluster was 200-500 nm. To determine the electrochemical performance of pristine PEDOT micro-ower, the discharge-charge capacity of the Li-O 2 cell employing these microowers was measured (Fig. 1d). The dischargecharge process proceeded normally, which is meaningful because the pure carbon-free conducting polymer acts as the electrode material (cathode) for the Li-O 2 cell. This also indicates that PEDOT microower shows catalytic activity for the redox reactions between Li ions and oxygen. The capacity reached $1200 mA h g electrode À1 at a low current density (20 mA g electrode À1 ) but decreased signicantly with an increase in the current density. This rapid capacity reduction at high current density may be attributed to the lower electronic conductivity of the PEDOT microower as compared to the carbon-based electrodes. To compensate for the low conductivity, the pristine PEDOT microowers were composited with graphene, which has high conductivity. Fig. 2 presents SEM and TEM images of the PEDOT microower/graphene composites, in which the weight ratio of graphene was controlled to 5, 10, and 20 wt%. As seen in the SEM images (Fig. 2a-c), the PEDOT microower clusters seemed to adhere to the surface of the graphene matrix. It is believed that PEDOT with slight adhesion to the graphene surface with large specic surface area is easily complexed. The surface of the composite containing 5 wt% graphene was mostly covered with PEDOT microowers. However, when the graphene content was increased to 10 and 20 wt%, the graphene bare surface without the PEDOT microowers was revealed. Fig. 2d-f show the TEM images of the composites containing 5, 10, and 20 wt% graphene. The PEDOT microowers were stably attached to the graphene surface, indicating the successful formation of the composite. The surface coverage by the PEDOT microowers clearly decreased with increasing graphene ratio. The surface of the PEDOT microower/graphene composite acts as the reaction site and provides storage space for the reaction products during the operation of the constituent Li-O 2 cell. The graphene surface covered with the microowers showed good catalytic activity for the redox reaction between Li ions and oxygen, but the bare-graphene (carbon) surface triggered unwanted side reactions such as the formation of Li 2 CO 3 and decomposition of the electrolyte. [34][35][36][37][38] The surface of the PEDOT microower showed considerable electronic conductivity and catalytic activity, as shown in Fig. 1d. Thus, composite with the PEDOT microower surface and graphene matrix may be a promising electrode material for Li-O 2 cells. Fig. 3 shows the Fourier transform infrared (FTIR) spectra of the pristine PEDOT microower and PEDOT microower/ graphene composites. As the content of graphene increased, the intensity of the absorbance peaks associated with PEDOT decreased because of low amount of PEDOT. However, peaks position of the composites was almost similar to those of the pristine PEDOT microowers, which indicated that the PEDOT structure was unchanged during composite formation.
The electrochemical properties of the electrode employing the PEDOT microower/graphene composites were observed to determine the effects of the graphene matrix. For comparison, the electrode employing pristine graphene was also prepared. Hereaer, the electrode employing the pristine graphene will be referred to as the 'graphene electrode', and the electrode employing the PEDOT microower/graphene composites   containing 5, 10, and 20 wt% graphene will be referred to as 'comp-5', 'comp-10', and 'comp-20' electrodes, respectively. Fig. 4 shows the initial discharge-charge proles of the cell employing the four different electrodes at a current density of 400 mA g electrode À1 , with the capacity limited to 1500 mA h g electrode À1 . All the electrodes presented stable discharge-charge proles with a capacity of 1500 mA h g electrode À1 . Considering that the capacity of the electrode employing pristine PEDOT microowers was only $430 mA h g electrode À1 at 400 mA g electrode À1 (Fig. 1d), the introduction of graphene dramatically increased the capacity of the electrode, which could be attributed to the enhanced electronic conductivity. For comparison of electronic conductivities, at electrodes were prepared using graphene, pristine PEDOT microowers, and composites with 5, 10, and 20 wt% graphene. The resistivity of the at graphene electrode was just $0.41 U cm, while that of the at electrode fabricated using the pristine PEDOT microowers was $2.9 U cm. However, the resistivities of the at electrodes fabricated from the composites containing 5, 10, and 20 wt% graphene decreased to $1.5, 1.1, and 0.9 U cm, respectively. The average overpotentials of four electrodes were 1.07-1.26 V. Interestingly, the comp-5 electrode showed a smaller overpotential than those of the graphene and comp-20 electrodes having high electronic conductivity. This may imply that catalytic activity of the PEDOT microowers is comparable to that of graphene. The over-potential of the Li-O 2 cell is strongly associated with the dissociation of reaction products during charging, indicating that the PEDOT microowers act a better catalyst for the dissociation of Li 2 O 2 than does pristine graphene.
The cycling performance of the Li-O 2 cells employing the graphene as well as the comp-5, comp-10, and comp-20 electrodes was then evaluated at a current density of 400 mA g electrode À1 . The cell capacity was set at 1500 mA h g electrode À1 to minimize the depth of discharge. 43 As shown in Fig. 5a, the cycle life of the graphene electrode was only 76 cycles under these measurement conditions. In contrast, the electrode employing the PEDOT microower/graphene composites presented much superior cyclic performance, and especially, the comp-5 electrode maintained a constant capacity (1500 mA h g electrode À1 ) over 150 cycles. However, as the increase of the amount of graphene, the cycle life was somewhat reduced. Fig. S1 † shows the variation in the discharge-charge proles of the four electrodes during cycling. The cycle lives of the non-aqueous Li-O 2 cells are associated with several factors such as the accumulation of reaction products on the air electrode (cathode), electrolyte loss by evaporation and unwanted side reactions, and unstable Li anode. 1,8,9,21 The instability of the Li anode and evaporation of the electrolyte cannot be controlled under our experimental conditions. Thus, the cyclic performance of our electrodes is highly dependent on side reactions such as the formation of Li 2 CO 3 at the Li 2 O 2 /carbon interface and decomposition of the electrolyte during cycling. The side reactions yield unwanted products such as Li 2 CO 3 and organic materials (e.g., CH 3 CO 2 Li and HCO 2 Li), which are not dissociated during charging but accumulate on the surface of the electrode. Moreover, decomposition of the electrolyte by side reactions leads to electrolyte loss. Accordingly, the superior cyclic performance of the  electrode employing PEDOT microower/graphene composites may imply that the microowers on the electrode surface can suppress unwanted side reactions.
To evaluate the effect of the PEDOT microowers on the surface of composite, the graphene, comp-5, 10 and 20 electrodes were analysed using SEM and FTIR spectroscopy. Fig. 6 presents the SEM images of the graphene, comp-5, 10, 20 electrodes before and aer 50 cycles (charged state) at 400 mA g electrode À1 with a limited capacity of 1500 mA h g electrode À1 . The surface morphology of the electrodes before the cycles (Fig. 6a, c, e and g) was similar to that of the powder samples shown in Fig. 2. Fig. S2 † shows the SEM images of the electrodes aer the initial discharge. The surface of the electrodes seemed to be covered by a lm composed of the reaction products. As shown in Fig. S1a and S1b, † the graphene surface was still smooth, implying that the reaction products were homogeneously attached to the surface of the graphene electrode. In contrast, the surface of the graphene electrode aer 50 cycles displayed thick and cluster-type reaction products, as indicated by the yellow circles in Fig. 6b. Ideally, when the electrode is charged state, most of the reaction products are expected to dissociate from its surface. However, a considerable amount of reaction products was accumulated on the graphene electrode, because  of the insufficient cycle life (Fig. 5a). On the contrary, no such reaction products were seen on the surface of the comp-5 electrode even aer 50 cycles (Fig. 6d). While we cannot exclude the possibility that reaction products could have accumulated on the surface of the PEDOT microower electrode, large heterogeneous reaction products observed with the graphene electrode were not seen in the present case. In contrast, some cluster-type reaction products appeared on the surface of the comp-20 electrode aer 50 cycles (yellow circles in Fig. 6h) because of the predominant graphene content. The FTIR spectra of the four electrodes aer 50 cycles (charged state) were also analysed to identify the presence of residual reaction products during cycling. As shown in the le column of Fig. 7a, several absorbance peaks were observed in the spectrum of the graphene electrode (aer 50 cycles) in the 400-700 and 1350-1700 cm À1 (marked with +) region. These peaks are attributable to unwanted products such as CH 3 CO 2 Li and Li 2 CO 3 , which presenting that the most of the residual reaction products are attributed to the accumulation of unwanted reaction products derived from side reactions. The right column of Fig. 7a illustrates the formation of unwanted reaction products on the surface of the graphene electrode. The FTIR spectrum of the comp-5 electrode aer 50 cycles also contained peaks due to these products, the peak intensities were signicantly lower than those for the graphene electrode (le column of Fig. 7b). This result indicates that the PEDOT microower layer effectively reduces the accumulation of unwanted reaction products on the electrode surface. As illustrated in the right column of Fig. 7b, the PEDOT microower covering the graphene matrix prevents direct contact between carbon and Li 2 O 2 and/or the electrolyte, thus suppressing the side reactions at the carbon/Li 2 O 2 and carbon/electrolyte interfaces. This in turn reduces the formation of residual reaction products and leads to enhanced cyclic performance, as shown in Fig. 5b. However, in the spectra of the comp-20 electrode aer 50 cycles, the intensity of the peaks related to the unwanted reaction products increased compared to that for the comp-5 electrode (le column of Fig. 7d), implying that the amount of residual products increased with the graphene content.
This result could be explained by the area of the surface covered with microowers in the PEDOT microower/graphene composite. As shown in Fig. 2, most of the surface of the comp-5 electrode was protected from the reactive Li 2 O 2 and electrolyte by the PEDOT microower clusters. However, a considerable portion of the surface of the comp-20 electrode was graphene (carbon), so the exposed carbon surface activated side reactions, resulting in the accumulation of unwanted products on the electrode, as illustrated in the right column of the Fig. 7d. This could explain the inferior cyclic performance of the comp-20 electrode compared to that of the comp-5 or 10 electrodes (Fig. 5). Thus, the electrode surface covered with PEDOT microowers would be preferable for obtaining Li-O 2 cells with excellent cycle performance.

Conclusions
In this article, PEDOT microowers are introduced as a potential electrode material for Li-O 2 batteries. The electrodes with pristine PEDOT microowers showed stable discharge-charge proles in the Li-O 2 cells, but their capacity was very small due to the low electronic conductivity. In contrast, the composite composed of the graphene matrix covered with PEDOT micro-owers presented notable capacity and low overpotential. The cycle performance of the electrodes employing the PEDOT microower/graphene composite was superior to that of the electrode employing pristine graphene.
Specially, PEDOT microower/graphene composite containing 5 wt% graphene was the most suitable for the material of air electrodes, within our experimental conditions. As a reference, the comparison of the cyclic performance of the Li-O 2 cells containing carbon-based composites was summarized in the Table 1. As per SEM and FTIR analysis, the residual reaction products derived from side reactions were decreased due to the  coverage of the electrode surface by the PEDOT microowers. As illustrated in Fig. 8, unwanted reaction products are easily accumulated on the surface of graphene because the carbon surface activates the side reactions. However, the surface of the PEDOT microowers can suppress such side reaction and reduce the formation of residual reaction products on the electrode, leading to enhanced cycle performance. In this context, the composite in which a large portion of the graphene surface was exposed inferior cycle performance as compared to that whose surface mostly covered by PEDOT microowers.

Conflicts of interest
There are no conicts to declare.