High‐Performance Li–O2 Batteries with Controlled Li2O2 Growth in Graphene/Au‐Nanoparticles/Au‐Nanosheets Sandwich

The working of nonaqueous Li–O2 batteries relies on the reversible formation/decomposition of Li2O2 which is electrically insulating and reactive with carbon and electrolyte. Realizing controlled growth of Li2O2 is a prerequisite for high performance of Li–O2 batteries. In this work, a sandwich‐structured catalytic cathode is designed: graphene/Au‐nanoparticles/Au‐nanosheets (G/Au‐NP/Au‐NS) that enables controlled growth of Li2O2 spatially and structurally. It is found that thin‐layer Li2O2 (below 10 nm) can grow conformally on the surface of Au NPs confined in between graphene and Au NSs. This unique crystalline behavior of Li2O2 effectively relieves or defers the electrode deactivation with Li2O2 accumulation and largely reduces the contact of Li2O2 with graphene and electrolyte. As a result, Li–O2 batteries with the G/Au‐NP/Au‐NS cathode exhibit superior electrochemical performance. A stable cycling of battery can last 300 times at 400 mA g−1 when the capacity is limited at 500 mAh g−1. This work provides a practical design of catalytic cathodes capable of controlling Li2O2 growth.


Introduction
An increasing importance has been attached to the electrifi ed transport to meet ever pressing energy and environmental issues.  3,1500339 necessary to ensure good electronic conductivity of the electrodes and decrease the usage amount of noble metals. In this regard, a modifi cation on carbon materials is required to minimize the detrimental effects. Lu et al. provided a useful way to deactivate the active carbon defect sites through an alumina coating. [ 30 ] The active sites on carbon could also be deactivated by in situ electrochemical nitrogen doping. [ 58 ] The introduction of a component that preferably reacts with Li 2 O 2 or LiO 2 over carbon has proven to be an effective measure to lessen carboninduced negative effects. [ 59 ] In this work, we provide a unique design of binder-free catalytic cathode which was prepared in ice bath. In this cathode, Au nanoparticles (Au-NP) are sandwiched between few-layer graphene (G) and thin Au nanosheets (Au-NS), forming a G/ Au-NP/Au-NS sandwich frame. The merits of the electrode design include: (1) graphene provides the electronically conducting channels for ORR and OER; (2) Au NPs catalyze the confi ned/conformal growth of Li 2 O 2 only on the surface of Au particles; (3) Au NSs fi x the Au particles and encapsulate Li 2 O 2 -loaded Au NPs between the sandwich frame. This unique electrode design effectively alleviates or defers the deactivation of the electrode and reduces the contact of Li 2 O 2 (or LiO 2 ) with graphene and electrolyte. As a result, Li-O 2 batteries with the G/Au-NP/Au-NS cathode can sustain 300 cycles under 400 mA g −1 at a limited capacity of 500 mAh g −1 . The Li-O 2 battery can also sustain 100 cycles at a full charge/discharge mode in the cutoff voltage of 2-4.5 V. This work sheds light on the design of effi cient catalytic cathodes enabling controlled Li 2 O 2 growth aiming at high-performance Li-O 2 batteries. Figure 1 shows the architecture and working mechanism of the G/Au-NP/Au-NS electrode, where Au NPs are encapsulated in between graphene and Au NSs. For the electrode, graphene layer was deposited on Ni foam substrate by chemical vapor deposition (CVD) method, and the Au-NP/Au-NS layer was deposited on graphene by solution impregnation method in ice bath. Two forms of nanosized Au, NPs and NSs, can grow simultaneously on graphene in ice bath. The Au NPs may catalyze the growth of Au NSs at low temperature. Li 2 O 2 can realize the confi ned/conformal growth within the electrode, namely, only on the surface of Au NPs. Figure 2 a presents the scanning electron microscopy (SEM) image of porous Ni foam coated with graphene. Graphene was coated only on the skeleton of Ni foam and the porous structure of Ni is preserved for barrier-free Li-ion and oxygen transport. The enlarged view in Figure 2 b suggests that the deposited graphene is in a thin-layer form since the surface profi le of Ni can be clearly seen. The 2D characteristics peak on Raman spectrum in Figure S1a (Supporting Information) verifi es the few-layer feature of the graphene. [ 60 ] The surface profi le of Ni can still be seen after Au deposition, suggesting that the Au layer on graphene is rather thin. The magnifi ed view in Figure 2 d exhibits that the Au layer consists of Au NPs and the Au NSs. Transmission electron microscopy (TEM) images in Figure 2 e,f show that the size of Au NPs is around 100 nm. The presence of Au is confi rmed by X-ray diffraction (XRD, Figure S1c, Supporting Information), X-ray photoelectron spectrum (XPS, Figure S1b, Supporting Information), and high-resolution TEM (HRTEM, Figure S2, Supporting Information). The thickness of the Au NSs is estimated to below 2 nm from the folded domain of the Au sheets ( Figure S2b, Supporting Information). As a result, a G/Au-NP/ Au-NS sandwich has constructed by the solution impregnation method in ice bath. In contrast, only Au NPs form on graphene (G/Au-NP) when the impregnation step was performed at room temperature ( Figure S3a, Supporting Information).  Figure S4, Supporting Information), indicating that graphene itself contributes minor to the catalytic activity of G/Au-NP/Au-NS, and thus can only be considered as the conducting support for ORR/OER. Therefore, the current density and specifi c capacity of the batteries were calculated based on the weight of Au. Figure 3 b,c shows the voltage profi les and cycling performance of the G/Au-NP/Au-NS-catalyzed Li-O 2 battery at a limited capacity of 500 mAh g −1 . The capacity of 500 mAh g −1 can be maintained over 300 cycles at 400 mA g −1 . In addition, in most of these cycles, the terminal discharge voltage is over 2.5 V, indicative of a low electrode polarization with Li 2 O 2 deposition. The OER potentials are somewhat higher than expected which may be due to the relatively high current density used and sluggish transport rate of Li ions and oxygen in the presence of the Au fi lm. The batteries were also tested at lower current densities of 100 and 200 mA g −1 ( Figure S5, Supporting Information). As expected, the OER potentials can be obviously reduced at lower current densities. A stable cycling can last 170 times when the capacity was limited at 1000 mAh g −1 . A high discharge voltage of over 2.5 V is also observed for the fi rst 120 cycles at 1000 mAh g −1 , suggesting that an increased Li 2 O 2 deposition does not cause increased electrode polarization due to the unique electrode design. By contrast, the stable cycling of Li-O 2 battery with the G/Au-NP cathode could last only 28 times ( Figure S6, Supporting Information).   Note that in this rigorous cycling mode, the Li-O 2 battery can still exhibit a stable cycling. After 100 cycles, a capacity over 500 mAh g −1 is retained. The coulombic effi ciency is close to 100% during cycling which suggests the reversible growth/decomposition of Li 2 O 2 . This means that the side reactions related to electrolyte [ 61 ] or carbon [40][41][42] are not signifi cant. Previous reports showed that carbon [ 30,41 ] or noble metals [ 34,62 ] could catalytically decompose the electrolytes. As shown in Figure S7 (Supporting Information), when tested in pure argon, the battery shows a low capacity with a rapid voltage increase upon charge and a rapid voltage decrease upon discharge. It   suggests that the G/Au-NP/Au-NS electrode displays minor catalytic effect for electrolyte decomposition. Electrochemical impedance spectroscopy (EIS) was used to understand the excellent catalytic activity of G/Au-NP/Au-NS. The Nyquist plots at different states are shown in Figure 4 c. The plots are fi tted using an equivalent circuit (inset in Figure 4 c) and the fi tting results are summarized in Table S1 (Supporting Information). In the circuit, R e denotes ohm resistance of the battery components, R f and Q 1 correspond to solid-state-electrolyte resistance and relax capacitance, R ct and Q 2 represent charge transfer resistance and double layer capacitance, and Z w is related to the bulk diffusion of Li ions. [ 63 ] Note that the increase in R ct is not signifi cant after discharge, implying that Li 2 O 2 deposition does not cause obvious electrode deactivation, agreeing well with the electrochemical tests. The reduction of R ct after recharge indicates the suffi cient decomposition of Li 2 O 2 .

Results and Discussion
To further clarify the superior catalytic performance of the G/ Au-NP/Au-NS electrode, SEM and TEM observations were conducted on the discharged electrodes. As shown Figure 5 a, the surface profi les are maintained after discharge without the formation of any cracks. The enlarged image in Figure 5 Figure S8a,b, Supporting Information) indicate that the dominant discharge product is Li 2 O 2 although a small amount of Li 2 CO 3 also forms. The formation of Li 2 CO 3 can be attributed to the decomposition of electrolyte. Note that repeated cycling does not lead to the accumulation of Li 2 CO 3 obviously as seen in Figure S8 (Supporting Information), which can explain the good cycling performance of the battery. As seen in Figure S9 (Supporting Information), the battery with G/Au-NP/ Au-NS electrode can sustain 245 cycles at 100 mA g −1 , corresponding to a long working period of up to 102 d, indicating that increasing working time does not lead to obvious Li 2 CO 3 accumulation and that the electrolyte is relatively stable during cycling. It should be noted that although Li 2 O 2 and Li 2 CO 3 can be qualitatively detected by XPS, the quantitative information regarding the effi ciency of Li 2 O 2 formation needs more efficient analysis tool such as differential electrochemical mass spectrometry system. [ 35 ] Importantly, Li 2 O 2 can grow along the surface profi les of the Au NPs, which clearly indicates that Au NPs do catalyze the confi ned and conformal growth of Li 2 O 2 . This growth behavior of Li 2 O 2 is favorable considering the facts that the contact between Li 2 O 2 (or LiO 2 ) with graphene can be minimized ( Figure S8c  volume expansion of electrode with Li 2 O 2 accumulation can be largely avoided. In addition, Li 2 O 2 with a thin-layer structure can be easily decomposed upon recharge. [ 18,63,64 ] In the G/Au-NP/Au-NS electrode, graphene participates in the catalytic reactions by providing the conducting channels, [ 65,66 ] although its catalytic activity is weak as mentioned above. The poor catalytic activity of Au NSs for Li 2 O 2 growth may be due to the insuffi cient electron transfer since the Au NPs that bridge graphene with Au NSs are covered with insulating Li 2 O 2 . In this case, the Au NSs may act as the separator to reduce the contact of Li 2 O 2 (or LiO 2 ) with electrolyte, and as the fi xer to stabilize the G/Au-NP electrode. This can explain the considerably improved cycling stability of the G/Au-NP/ Au-NS-catalyzed Li-O 2 battery compared with the G/Au-NP-catalyzed battery. SEM images in Figure 6 a,b show that the microstructure of the G/Au-NP/Au-NS electrode can be retained after the recharge process. The decomposition of Li 2 O 2 upon recharge can be confi rmed by Li 1s and O 1s XPS ( Figure S8a,b, Supporting Information). In contrast, Li 2 CO 3 is remained after charge, suggesting that the decomposition of Li 2 CO 3 is difficult at the applied voltage ( Figure S8b-d, Supporting Information). It was also noticed that the ether peak at around 286. 3 eV appears after the fi rst discharge, and its intensity increases with cycling, suggesting increased amount of decomposition products although the tetraethylene glycol dimethyl ether (TEGDME) electrolyte is relatively stable. Similar result was found in other work using TEGDME as electrolyte. [ 67,68 ] No obvious cracks were generated in Au NSs after recharge, indicating the robustness of the Au NSs. As shown in Figure 6 c,d, no aggregation of Au NPs occurs because of the immobilization effect of the Au NSs, ensuring the durability of catalytic activity of the electrodes and long cycle life of the batteries.

Conclusion
In summary, a highly effi cient catalytic cathode of G/Au-NP/ Au-NS was prepared by a facile impregnation approach in ice bath. In this electrode, Li 2 O 2 realizes the conformal growth and crystallizes into thin-layer form on the surface of Au NPs, graphene supplies the conducting channels for ORR/OER, and Au NSs stabilizes the G/Au-NP structure and reduces the contact of Li 2 O 2 (or LiO 2 ) with electrolyte. The unique architecture of the G/Au-NP/Au-NS electrode enables controlled Li 2 O 2 growth, easy Li 2 O 2 decomposition, and reduced side reactions, leading to excellent electrochemical properties of Li-O 2 batteries. At a limited capacity of 500 mAh g −1 , the battery can sustain a stable cycling over 300 times at 400 mA g −1 . The Li-O 2 battery can maintain a capacity of 500 mAh g −1 after 100 cycles at 800 mA g −1 when it was tested in a full charge/discharge mode in the cutoff voltage of 2-4.5 V. The results show that the controlled growth of Li 2 O 2 can be realized by optimizing the electrode design and that both architecture and component of the electrodes are important to achieve high performance of Li-O 2 batteries.

Experimental Section
Electrodes Preparation and Characterization : Graphene was deposited on Ni foam by a CVD method as described previously. [ 63 ] The loading of graphene on Ni is 0.6 mg cm −2 . The Ni-supported G/Au-NP/Au-NS electrodes were prepared by a solution impregnation method. Briefl y, the graphene-loaded Ni foam pieces were immersed in a beaker containing HAuCl 4 ·3H 2 O aqueous solution (0.16 mg mL −1 ) and the beaker was placed in an ice bath for 3 h. The Ni-supported G/Au-NP/Au-NS electrodes were then rinsed with distilled water, dried at 60 °C for 10 h in vacuum and heated at 300 °C for 2 h in argon. For comparison, the Ni-supported G/Au-NP electrodes were also prepared by a similar route with the impregnation step performed at room temperature. The loading of Αu on graphene is 0.5-0.6 mg cm −2 . The materials deposited on Ni foam were checked by XRD on a Rigaku D/Max-2550pc diffractometer with Cu K α radiation ( λ = 1.541 Å). The structure of graphene on Ni was analyzed by Raman spectrum on a Jobin-Yvon Labor Raman HR-800 system with 514.5 nm Ar-ion laser. The morphologies of the pristine, discharged and recharged electrodes were observed by fi eld-emission SEM on an S-4800 microscope. The microstructure of electrode components exfoliated from the electrodes was characterized by TEM on a JEM 2100F microscope. The chemical states of the elements in the electrodes were examined by XPS on a Kratos Axis Ultra-DLD spectrometer with Al K α radiation ( hν = 1486.6 eV). The electrodes after cycling were carefully handled before the XPS measurements. First, the batteries after charge or discharge were dissembled in the argon-fi lled glove box. The electrodes were then washed by 1,2-dimethoxyethane (DME) three times to remove TEGDME and LiClO 4 suffi ciently on the electrode surface. The DME was then removed by resting the electrodes in the small chamber of the glove box under evacuation for 15 min. After that, the electrodes were fi xed on the XPS holder and sealed in a plastics box under Ar atmosphere in the glove box. For XPS tests, the electrodes were taken out from the box and transferred to the XPS chamber as quickly as possible (within 2 min). The XPS chamber was then quickly evacuated.
Li-O 2 Batteries Assembly and Electrochemical Measurements : Coin-type Li-O 2 batteries were assembled in an argon-fi lled glove box using Li foils as anodes, Ni-supported G/Au-NP/Au-NS (or graphene, G/Au-NP) as cathodes (0.36 cm 2 ), Celgard C480 porous fi lms as separators, and 1 M LiClO 4 in TEGDME (Sigma-Aldrich, treated with molecular sieve before use) as electrolyte. The cathodes were dried at 80 °C under vacuum for 10 h prior to assembly of the batteries. The assembled batteries were then purged with O 2 for 10 min and stayed at open voltage circuit for 5 h to achieve an equilibrium. Galvanostatic cycling was conducted on a Neware battery cycler (Shenzhen, China) over a voltage range of 2.0-4.5 V (vs Li/Li + ). The specifi c capacity (mAh g −1 ) and current density (mA g −1 ) of the G/Au-NP/Au-NS and G/Au-NP electrodes were calculated based on the weight of Au. The discharge and charge are referred to lithiation and delithiation, respectively. EIS was recorded on a VersaSTAT3 electrochemistry workstation (Princeton Applied Research) by applying an ac voltage of 10 mV amplitude in a frequency range from 10 −2 to 10 5 Hz. The electrochemical tests were all carried out at room temperature.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.