Nanoconfined growth of lithium-peroxide inside electrode pores: a noncatalytic strategy toward mitigating capacity–rechargeability trade-off in lithium–air batteries

Controlled modification of a mesoporous carbon electrode mitigates the capacity–rechargeability trade-off in Li–air batteries by pore-confined growth of nanocrystalline Li2O2.


Introduction
Highest theoretical specific energy of 3.5 kWh kg⁻ 1 makes Lithium-Air (Li-Air) battery as one of the most promising future energy storage technologies for both stationary and mobile applications. Ideally, the overall electrochemical reaction in a nonaqueous Li-Air battery involves a two-electron (2e⁻) redox of oxygen (O 2 ) in the presence of lithium (Li + ) ions (2Li + + O 2 + 2e⁻ ↔ Li 2 O 2 (s); E 0 = 2.96 V vs. Li/Li + ) leading to reversible formation (by the oxygen reduction reaction (ORR)) and decomposition (by the oxygen evolution reaction (OER)) of solid lithium peroxide (Li 2 O 2 ) during discharge (DC) and recharge (RC), respectively. 1 However, passivation of the conducting electrode by this ostensibly insulating solid Li 2 O 2 layer (both ionic and electronic conductivities in the range of 10⁻ 19 to 10⁻ 20 S cm⁻ 1 at 25 °C) and diffusion-limitation of electroactive species in the electrode-pores result in lower than theoreticallypredicted capacity, poor rechargeability, low rate capability etc. [2][3][4][5] Although increase in carbon surface area as well as pore volume accessible to both O 2 and electrolyte could increase the DC capacity (Q DC ) of the cell, growth of large insulating crystals of Li 2 O 2 with increase in Q DC often led to deterioration of rechargeability as well as poor cyclability. [6][7][8] The reason behind this low rechargeability is the limitation in charge transport through a thick Li 2 O 2 layer that results in a high RC potential > 4 V vs. Li/Li + and causes decomposition of both the nonaqueous electrolyte and carbon electrode. 4,9,10 Worsening the situation, efforts to solve rechargeability problem in Li-Air battery by solid electrocatalysts could not truly improve the ORR/OER reversibility. 11,12 Rather, the use of these "catalysts" exacerbated electrolyte/electrode instabilities. Furthermore, the soluble redox-mediators also seemed to have (in)stability issues and their use on an unprotected anode led to passivation of Li metal electrode. 13,14 Therefore, it is necessary to identify a proper noncatalytic strategy to increase the capacity without sacrificing the rechargeability of a Li-Air battery. Alternative to the usual uncontrolled growth of Li 2 O 2 on widely used carbon nanotube (CNT) or carbon particle (Ketjen Black (KB), Super P (SP) etc.) based electrodes, confining the nucleation-and-growth of Li 2 O 2 inside the framework-pores of a nanoporous carbon electrode can restrict the spatial growth of the discharge product below electron-tunnelling length and produce a nano-crystalline Li 2 O 2 with small domain size. Compared to bulk crystalline Li 2 O 2 , this spatially-confined poorly crystalline Li 2 O 2 with large interface with the carbon electrode is expected to show higher ionic and electronic conductivities that should suppress the recharge overpotential and help to achieve high round-trip energy efficiency in a Li-Air battery. 15,16 However, at the same time it is necessary to ensure facile mass-transport and enough space inside the porous framework to produce large Q DC in the cell. Subsequently, as a result of improved mass and charge transport due to nanoconfinement of Li 2 O 2 in an electrode with sufficient porosity, high rate-capability during both DC and RC of the cell can also be achieved. Based on this idea, herein, we first confirm that compared to uncontrolled growth of Li 2 O 2 on several commercial carbon electrodes, nanoconfinement of Li 2 O 2 inside a mesoporous carbon results in a much higher round-trip energy efficiency (ratio of energy densities during DC and RC respectively) that does not deteriorate with increase in depth of DC in a Li-Air battery. Further, we show that controlled increase in the electrochemically active surface area (ECSA), pore diameter and pore volume of the mesoporous carbon electrode gradually enhances Q DC by improving diffusion of electroactive species inside the mesoporous channels and simultaneously maintains an efficient rechargeability in a Li-Air battery due to decomposition of nanocrystalline Li 2 O 2 confined inside discrete pores of the electrode. It has been found that proper optimization of the physico-chemical properties of the carbon electrode could achieve ~ 2.5 times improvement in Q DC but at the same time maintained an average RC potential of ~ 3.6 V vs. Li/Li + which is 400 to 700 mV lower than that of several other commercial carbon electrodes under similar conditions. This report demonstrates a prototypical strategy of a simple physico-chemical alteration in a porous carbon electrode as a metal-catalyst/promoter-free solution toward the capacityrechargeability trade-off in Li-Air battery.

Results and discussions
Effect of pore-confinement of Li 2 O 2 in Li-Air cells.
In order to verify the effect of pore-confined Li 2 O 2 on suppressing RC overpotential (η RC = | E 0 -E RC |) we begin with comparing the DC/RC profiles of various carbon electrodes, such as CNT, KB, SP and mesoporous CMK-3, which have completely different surface morphologies and pore structures. While KB and SP carbons show particle-like morphology with 40-60 nm diameter, CNT has tubular morphology with an average outer diameter of 20-40 nm and length of 5-15 μm. The morphologies of these carbons are shown by the scanning electron microscopic (SEM) images in Figure S1 (a). Despite differences in morphologies, however, all these three commercial carbons give rise to inter-particle porosity with a random distribution of pore-space having diameter in the range of 50-100 nm. In contrast, CMK-3 with bead-like morphology has hexagonally ordered cylindrical mesopores (< 4 nm diameter) aligned parallel to each other inside the carbon framework. 17 Figure S1 (b) and (c) respectively show nitrogen adsorption/desorption isotherms measured at -196 ℃ and pore-size distribution by Barrett-Joyner-Halenda (BJH) method for different carbons. The pore structures are schematically represented in Figure S1 (d). At first, DC/RC of the cells were carried out in dry air (maximum dew point < -60 ℃, < 11 ppm (v/v) H 2 O) using 1 M lithium bis(trifluoromethanesulfonyl)imide (LiTFSI) in tetraethylene glycol dimethyl ether (TEGDME or tetraglyme; < 20 ppm H 2 O) as the electrolyte up to a limited Q DC of 500 mAh g⁻ 1 at the same DC/RC current density (j DC/RC ) of 100 mA g⁻ 1 . The galvanostatic DC/RC profiles in Figure 1 (a) show quite similar DC potentials for all the electrodes, but significantly different RC potentials. Among these carbons, the mesoporous CMK-3 electrode shows the lowest average RC potential of 3.6 V vs. Li/Li + , whereas the same for SP, CNT and KB are 4.28, 4.17 and 3.98 V vs. Li/Li + respectively. Figure 1 (b) further shows that CMK-3 has the highest overall round-trip energy efficiency of 74% which is 7-14% higher than that of the other carbons. During DC in the non-mesoporous carbons the inter-particle void spaces get filled with a thick film-like crystalline Li 2 O 2 (> 50 nm thick) that has very poor ionic as well as electronic conductivities and as a result, the decomposition of this insulating Li 2 O 2 on these non-mesoporous carbons requires very high η RC (usually, η RC > 1 V). On the other side, the low η RC in CMK-3 electrode can certainly be corroborated to facile decomposition of nanoconfined Li 2 O 2 inside the mesoporous channels having average pore diameter of 3.6 nm (see later).  The spatially restricted growth keeps the thickness of Li 2 O 2 well below the electron tunneling limits (~ 10 nm) and the small crystallite size results in much higher ionic conductivity compared to crystalline Li 2 O 2 . 2, 15 Therefore, circumvention of charge-transport-limitation by nanoconfinement of Li 2 O 2 should be a definite reason behind the suppression of η RC in CMK-3 electrode. The deposition of Li 2 O 2 in these carbon electrodes is shown by the SEM images in Figure S1 (e) and is schematically represented in Figure 1 (c-e). In order to verify the possibility of mass diffusion inside the mesoporous channels of CMK-3, we have precisely determined the hydrodynamic radii of the electroactive species in the electrolyte by using Stokes-Einstein equation These results confirm that the electroactive species can permeate through the porous channels of the CMK-3 electrode and take part in the electrochemical reactions inside the confined porespace. However, it should be noted that although the hydrodynamic radii of the electroactive species are smaller than the radii of the mesopores in CMK-3 and additional capillary forces help O 2 and electrolyte solution to diffuse into the mesoporous channels, the deposition of solid Li 2 O 2 continuously decreases the porosity and blocks the diffusion channels during DC. 18,19 As a result, the CMK-3 electrode shows much lower Q DC (827 mAh g⁻ 1 ) than that calculated (1920 mAh g⁻ 1 ) from full utilization of the mesopore volume (0.71 cm 3 g⁻ 1 ) (see later). Nonetheless, oxygen diffusion limitation is a general problem in air-electrodes, which are quite sensitive to the partial pressure of oxygen ( ) in the feed gas. The effect of on Q DC of the electrodes was indeed observed when DC/RC of the cells was carried out in both dry air ( 0.21) and pure oxygen ( = = 1). Unsurprisingly, the Q DC of the air-cells has been found to be much lower than that of the cells operated in pure oxygen. Figure S2 Figure S2 (c) we have shown a comparison between the Q DC of oxygen and air cells for the respective carbon electrodes and a significantly lower Q DC , roughly in the range 35-65%, has been observed for all these carbons when air instead of oxygen is used as the feed gas. Nevertheless, practical applications of these batteries require operation in ambient air with simultaneously high energy and power densities as well as efficient rechargeability. Optimistically, the high rechargeability in CMK-3 electrode is also maintained with ~ 73% round-trip efficiency which is 6-13% higher than that of other carbons under air atmosphere for a full DC up to 2 V vs. Li/Li + and the results are shown in Figure S2 (d).
Owing to this high rechargeability, therefore, an efficient way has to be found out to further enhance the Q DC of the mesoporous electrode toward mitigating the capacity-rechargeability trade-off in Li-Air cells. In order to enhance the Q DC of the mesoporous electrode, we adopted two different strategies: one of these was to induce solution mediated DC, while the other was to strategically modify the porosity of the carbon to enhance mass diffusion inside the porous framework.

Promoter assisted increase in Q DC of mesoporous electrode.
Lithium nitrate (LiNO 3 ) and 2,5-di-tert-butyl-1,4-benzoquinone (DBBQ) were considered as the two promoters to induce solution mediated growth of Li 2 O 2 and to increase the Q DC of CMK-3 electrode. 20, 21 Specifically, we used 1 M LiNO 3 in tetraglyme and 1 M LiTFSI + 50 mM DBBQ in tetraglyme as the two electrolytes in Li-Air cells. Figure 2 (a) shows the full DC/RC curves of CMK-3 electrodes in different electrolytes with or without promoters under dry air with the cut-off conditions of 2-4.5 V vs. Li/Li + at j DC/RC of 100 mA g⁻ 1 . It is indeed found that the use of LiNO 3 and addition of DBBQ to LiTFSI as a promoter significantly increase the Q DC of CMK-3 electrodes. A quantitative comparison of Q DC in different electrolytes in Figure 2 (b) demonstrates that the Q DC of CMK-3 increases from 827 mAh g⁻ 1 in 1 M LiTFSI to 1319 mAh g⁻ 1 in 1 M LiNO 3 and 1863 mAh g⁻ 1 in 1 M LiTFSI + 50 mM DBBQ in tetraglyme electrolyte. However, this gain in Q DC due to enhanced solution mediated deposition of Li 2 O 2 is achieved at the cost of round-trip efficiency of the cells. From Figure  2 (b), the round-trip efficiency of ~ 73% in 1 M LiTFSI decreases to ~ 67% in 1 M LiNO 3 and ~ 50% in 1 M LiTFSI + 50 mM DBBQ in tetraglyme. The X-ray diffraction (XRD) patterns in Figure S3 (a) confirm the formation of Li 2 O 2 as the discharge product in all these three electrolytes and a detailed analysis of the XRD data in Figure S3 (b) shows a slight increase in domain size for the (100) reflection and an anisotropic growth of Li 2 O 2 crystals with enhanced solution mediated DC of CMK-3 electrode. Since occurrence of parasitic side reactions is thought to be a reason of poor rechargeability, we have also quantified the discharged Li 2 O 2 by titration with TiOSO 4 solution. 9 A much lower than 100% yield of Li 2 O 2 after DC definitely indicates occurrence of parasitic side reactions during DC. However, the Li 2 O 2 yields of 71±2, 74±1 and 67±2% in 1 M LiTFSI, 1 M LiNO 3 and 1 M LiTFSI + 50 mM DBBQ respectively do not show any correlation with the rechargeability of the cells. Figure S4 shows the SEM images of the CMK-3 electrodes after full DC up to 2 V vs. Li/Li + at 100 mA g⁻ 1 in these three electrolytes. Much thicker Li 2 O 2 layers on the outer surface of the electrode, owing to enhanced solution mediated deposition, have been observed after DC in LiNO 3 and DBBQ containing electrolytes compared to LiTFSI alone. It is quite reasonable to imagine that inducing solution mediated ORR does not improve the access to the mesopores. Rather, solution phase growth of Li 2 O 2 occurs in the inter-particle void spaces and outer surface of the mesoporous carbon, which nullifies the benefits of mesopores in nano-structuring Li 2 O 2 during DC. Therefore, the lower rechargeability of the cells in LiNO 3 and DBBQ containing electrolytes can be attributed to thicker Li 2 O 2 layer with slightly larger crystallite size inhibiting facile charge-transfer during RC of the cells. 22, 23 As a result, solution mediated increase in Q DC in CMK-3 electrode does not help to solve the capacity-rechargeability trade-off in Li-Air battery.

Modification of mesoporous carbon for improved mass diffusion.
In contrast to promoter assisted uncontrolled growth of Li 2 O 2 , modification of the mesoporous carbon in terms of increase in the pore diameter, pore volume, ECSA and number of ORR active sites should be a better alternative not only to improve mass diffusion inside the porous structure for higher Q DC but also to maintain a controlled and restrictive growth of  27,28 For these reasons, and due to relative ease of synthesis, we synthesized N-doped CMK-3 by a hardtemplating route using phenol-urea-formaldehyde (PUF) precursor as the carbon source. By changing the concentration of urea, we synthesized four different N-doped CMK-3 carbons which are labelled as N-CMK-3-1 to -4 in the increasing order of N-content. The SEM images of the undoped and N-doped CMK-3 samples showing bulk morphologies and the corresponding elemental mapping by energy dispersive X-ray spectroscopy (EDS) identifying the surface chemical compositions are demonstrated in Figure S5 (a-e). The SEM images show similar morphologies of the CMK-3 and N-CMK-3 samples, whereas the EDS mapping of the N-doped samples confirms quite well dispersed doping of nitrogen in the carbon structure. In addition to nitrogen, high content of oxygen functionalities is also observed in all the carbon samples. In order to provide a quantitative idea about N-doping and to precisely identify the N and Ofunctionalities, we carried out X-ray photoelectron spectroscopic (XPS) analyses of all the samples. The wide-scan, O1s, N1s and C1s XPS of all the carbon materials are shown in Figure S6 (a-d) and the N/C and O/C values of all the samples are tabulated in Table 1. Deconvoluted N1s XPS show three characteristics peaks at 398.5±0.2, 400.5±0.3 and 403.8±0.3 eV, corresponding to pyridinic, pyrrolic and oxidized nitrogen, respectively, as shown in Figure S6 (c). 27 The O1s and C1s XPS of all the samples in Figure S6 (b) and (d), respectively, identify C-O (e.g., phenol, ether, epoxy etc. at C1s ~286.8±0.3 eV and O1s ~532.5±0.1 eV), C=O (e.g., carbonyl, quinone etc. at C1s 287.7±0.3 eV and O1s 530.5±0.2 eV) and -COO⁻ (e.g., carboxylic acid, ester, lactone etc. at C1s 289.3 eV) groups to be present as the oxygen functionalities. 29 The oxygens present in phenol, formaldehyde and urea are thought to be the sources of these oxygen functionalities in the carbon. The specific surface area (S BET ) of all the samples was evaluated by Brunauer-Emmett-Teller (BET) method from the N 2 adsorption/desorption data obtained at -196 ℃ and the pore structure was analyzed by Barrett-Joyner-Halenda (BJH) method. The adsorption/desorption isotherms and BJH pore size distributions are shown in Figure S7 (a) and (b) respectively and the corresponding values are summarized in Table  1. In all cases the N 2 adsorption/desorption shows a type IV isotherm which is the characteristic of a mesoporous material. 30 In case of CMK-3, a steep N 2 adsorption (capillary condensation) takes place in the relative pressure (P/P 0 ) range of 0.45-0.55, indicating presence of high pore-size uniformity in the framework mesopores. 31 The BJH pore size analysis also shows a narrow pore-size distribution with a sharp peak at 3.6 nm. Similar type IV isotherms are observed for all the N-CMK-3 samples except that the capillary condensation takes place at higher P/P 0 (> 0.6), which indicates the presence of larger mesopores since capillary condensation pressure is a function of pore diameter. 30  Please do not adjust margins Please do not adjust margins interconnected ink-bottle-shaped pore networks with a narrow distribution of pore bodies but a wide neck size distribution. 32,33 From the BJH pore size analysis it is indeed found that the N-CMK-3 carbons show bimodal porosity and as the N-content increases in the carbon framework, the pore diameter gradually increases and becomes more distributed. Since the samples were synthesized using mesoporous SBA-15 silica as the template, ideally the predominant pore diameter of all the samples should be very close to the wall thickness of the SBA-15 (~ 4.5 nm). But, due to shrinkage of the pores in the carbon framework during carbonization of the precursor and template-removal steps, the pore diameter (3.6 nm) of the CMK-3 becomes lower than the pore wall thickness of the SBA-15 template.
In contrast, the presence of larger mesopores (diameter ~ 6 nm) and ink-bottle-like cavities in the N-doped carbons can be explained by incomplete filling of PUF precursors inside SBA-15 template during synthesis and partial decomposition of PUF polymer during carbonization process, leaving empty space in the carbon backbone. The CMK-3 was synthesized by polymerization of phenolformaldehyde (PF) mixture and subsequent carbonization of the PF resin. In contrast, for the synthesis of N-CMK-3 samples PUF precursor was used, where there are possibilities of several individual reactions, such as addition and condensation reactions of phenol, urea, and formaldehyde. 34 Therefore, different selfcondensation and co-condensation products can be formed during polymerization of the PUF precursor. This difference in polymerization process can affect the filling level of the SBA-15 template and produce cavities in the resulting N-doped carbons. Furthermore, addition of urea to PF resin results in poor bonding strength and lowers the stability of the resin that can decompose as well as shrink during the carbonization process. 35 As a result, the N-CMK-3 samples show larger and more distributed mesopores compared to the CMK-3 carbon. Consequently, due to this enlargement of pore diameter and formation of interconnected framework cavities, the N-CMK-3 samples exhibit higher pore volume (V BJH ) and the values are shown in Table 1. The mesoporous structure of the samples was visualized by transmission electron microscopic (TEM) images which, in Figure 3 (a-e), clearly demonstrate a hexagonally ordered mesopore arrangement for all the samples. Summarizing all these results, it is seen that N-doping of CMK-3 by addition of urea as a precursor not only increases the number of ORR active sites, but also successfully increases the porosity that can be further beneficial for Li-Air battery operations. A schematic representation of the synthetic procedure of the mesoporous carbons is shown in Figure 3 (f).

Discharge of N-CMK-3 electrodes in Li-Air cells.
All Interestingly, the values of Q DC do not follow the trend of N/C ratio and the peak of Q DC at 2111 mAh g⁻ 1 for N-CMK-3-3 and then decrease in the value to 1668 mAh g⁻ 1 for N-CMK-3-4 certainly indicate that increase in concentration of nitrogen as the ORR active sites is not the sole reason for capacity enhancement in these N-doped CMK-3 electrodes. In order to understand the reason behind the increase in Q DC upon N-doping of CMK-3 electrode, we thoroughly analyzed the pore structures of these carbon materials. Figure 4 (c) shows a decreasing trend of the S BET with increase in N/C ratio. Surprisingly this observation is contradictory to the trend of Q DC for these electrodes because intuitively, capacity should be proportional to the surface area of the electrodes. Nonetheless, the pore size and pore volume of these carbons seem consistent with Q DC values.    Increase in the pore diameter from 3.7 to ~ 6 nm is certainly expected to enhance mass transport inside the N-CMK-3 electrodes compared to undoped CMK-3. Furthermore, as shown in Figure 4 (c), with increase in N/C atomic ratio, the pore volume gradually increases, reaches at peak for N-CMK

N-CMK-3-1 N-CMK-3-2 N-CMK-3-3 N-CMK-3-4
Please do not adjust margins Please do not adjust margins network could lead to partial collapse of the carbon framework during carbonization. 34 As a result, N-CMK-3-4 breaks the increasing trend of pore volume with N/C ratio and shows smaller pore volume compared to N-CMK-3-3. Since ECSA provides a more precise estimate of the electrode-area available for electrochemical reactions, we measured and plotted the ECSA (see Supporting information for detailed calculation) of these mesoporous carbons in Figure 4 (c). Interestingly, the ECSA values do not follow the trend of S BET , rather show the similar trend of pore volume and correlate with the observed Q DC of the electrodes. This correlation can be attributed to enhanced penetration of electroactive species inside the mesoporous framework as the pore diameter and pore volume of the carbon increase. After DC of the cells in air the discharged electrodes were thoroughly analyzed by XPS, infra-red spectroscopy (FTIR) and XRD to identify the predominant DC products. Figure S8  (a-b) show XPS of discharged CMK-3 and N-CMK-3-3 electrodes, respectively. The single component Li1s spectra at ~ 55 eV and O1s spectra at ~ 532 eV confirm the deposition of Li 2 O 2 as the major DC product. 36 Besides, The O1s spectra show additional peaks at > 532 eV that can be assigned to -CO 3 , -CO or SO 2 species resulting from parasitic reactions or adsorbed electrolyte residues on the deposited Li 2 O 2 and additionally multiple components in the deconvoluted C1s spectra also indicate presence of several functionalities, such as C-C, CH x , CO x etc., in the discharged electrodes. 36 Consistent with the XPS results, the FTIR spectra of the discharged CMK-3 and N-CMK-3 electrodes in Figure S9 identify Li 2 CO 3 and Li-carboxylates to be the parasitic products during DC. There can be different ways for the formation of parasitic side products in Li-Air cells where, in addition to electrolyte and electrode decomposition, the CO 2 (0.04%) present in air can also undergo electrochemical reaction during DC to directly produce Li 2 CO 3 as an undesired product. 37,38 Nevertheless, parasitic side reaction between the carbon electrode as well as the nonaqueous electrolyte with superoxide intermediate is a persistent problem in Li-Air battery that demands for more stable electrolytes and electrode materials. 9, 10 Figure 4 (d) and Figure S10 show the XRD patterns of the discharged CMK-3 and N-CMK-3 electrodes. These XRD patterns not only confirm Li 2 O 2 as the DC product in these electrodes, but also show significant decrease in peak intensity at high N/C ratio in the N-doped carbons. This means lower fractions of crystalline Li 2 O 2 are deposited as the N/C ratio increases. Besides, a gradual broadening of both (100) and (101) reflections of Li 2 O 2 is observed with increase in N/C ratio. The broadening of the XRD peaks indicates a decrease in the average Li 2 O 2 crystallite-size which is plotted in Figure 4 (e) for all the samples. It is found that the respective crystallite size for the (100) and (101) reflections of Li 2 O 2 decreases from ~ 8 and 7.7 nm in CMK-3 to ~ 6.1 and 4.1 nm in N-CMK-3-4. Unlike unrestricted growth of Li 2 O 2 crystals in conventional carbons without framework-porosity, here in case of these mesoporous carbons the initial nucleation takes place inside a confined porous channel that restricts the spatial growth of Li 2 O 2 . As shown earlier in this article, the hydrodynamic radii of the electroactive species are much smaller than that of the mesopores in these N-doped carbons. Furthermore, the interconnected ink-bottle pores in N-CMK-3 electrodes are expected to show more facile mass transport along the mesoporous channels. Once diffuse into the pores, these electroactive species get confined inside the nanoscopic space where heterogeneous interaction between the reactants and the electrode surface is tremendously amplified. 39 Consequently, very high interfacial area inside the mesopores significantly enhances the kinetics of the electrochemical reactions. Moreover, turbostratic structure with considerable defects and N/O-doping enhance the ORR activity in these mesoporous CMK-3 based materials compared to other commercial carbons. 26,40 We have compared the ORR activities of all the CMK-3 and N-CMK-3 electrodes by a cathodic linear sweep voltammetry (LSV) in the range 3.2 to 2 V vs. Li/Li + at a sweep rate of 0.5 mV s⁻ 1 using metallic Li as both the counter and reference electrodes. Apparently, the integrated areas under the LSV curves in Figure S11 (a) depicting the amount of charge passed for ORR show similar trend of Q DC observed in galvanostatic DC. Figure  S11 (b) shows the current densities (j ORR ) of all these electrodes at three different potentials of 2.8, 2.7 and 2.6 V vs. Li/Li + during LSV scan, where overall a gradual increase in ORR current response has been observed with increase in N/C ratio in the N-CMK-3 electrodes.
We have also identified the overpotential (η = | 2.96 -E |, V) required to reach the j ORR of 0.05, 0.1 and 0.15 mA cm⁻ 2 in all these undoped and N-doped CMK-3 electrodes and the results in Figure  S11 (c) consistently show that with increase in N/C ratio the overpotential to reach respective j ORR becomes lower. Furthermore, the Tafel plots in Figure S11 (d) yield Tafel slopes of 473, 370, 383, 373 and 443 mV dec⁻ 1 for CMK-3 and N-CMK-3-1 to -4 electrodes respectively and these values unequivocally demonstrate an enhancement in ORR activities of the N-CMK-3 electrodes compared to the undoped one. As a result of this good electrocatalytic activity, the strong interaction between depositing Li 2 O 2 and the surface of the N-doped electrode leads to growth of a poorly crystalline thinfilm-like Li 2 O 2 on the surface of the carbons, even outside of the porous framework (SEM in Figure S12). 41 Therefore, despite higher Q DC in N-CMK-3 electrodes, owing to the structural effect of the porous framework as well as the electrocatalytic activities of the Nsites, the size of individual Li 2 O 2 crystallites remains small.

Recharge of N-CMK-3 electrodes in Li-Air cells.
It is important to note from Figure 4 (a) and (b) that the increase in Q DC in the case of N-CMK-3 electrodes does not affect the rechargeability and overall round-trip efficiency of the cells. The average RC potential for all the CMK-3 and N-CMK-3 electrodes has been calculated to be < 3.65 V vs. Li/Li + and Figure 4 (b) shows that > 70% round-trip efficiency is maintained in all the N-doped electrodes despite up to ~ 2.5 times increase in Q DC during full DC/RC in air. These results demonstrate a successful mitigation of capacityrechargeability trade-off in Li-Air battery. A closer look at the galvanostatic curves reveals that the RC at these electrodes starts at potentials very close to 3 V vs. Li/Li + , remains below 3.5 V vs. Li/Li + up to 50% state of charge (SOC) and goes above 4 V vs. Li/Li + at 70-75% SOC. Online electrochemical mass spectrometry (OEMS) of N-CMK-3-3 as the representative electrode shows the trend of gas evolution during RC of a cell with N-doped mesoporous carbon. Correlating the RC curve of N-CMK-3-3 in Figure 5 (a) with OEMS data in Figure 5 (b) it is found that O 2 gas starts evolving as soon as RC of the cell begins and the onset of O 2 evolution is very close to 3 V vs. Li/Li + . The gas evolution trend further shows that as the RC progresses, copious amount of O 2 is evolved below 3.5 V vs. Li/Li + . These observations give direct evidence of a facile decomposition of Li 2 O 2 below 3.5 V vs. Li/Li + in case of N-CMK-3-3 electrode. However, gradual increase in RC potential and large amount of carbon dioxide (CO 2 ) evolution at the later part of RC (> 4 V vs. Li/Li + ) indicate decomposition of deposited parasitic side products, such as lithium carbonate and lithium carboxylate, as evidenced by XPS and FTIR observations. Nevertheless, when compared with OEMS data of CNT electrode in Figure S13   In order to ascertain OER catalytic activity trend with N/C ratio, we have carried out anodic LSV (0.05 mV s⁻ 1 ) of these electrodes after reduction up to Q DC of 0.5 mAh cm⁻ 2 as shown in Figure S15 (a) that exhibits two pronounced oxidation peaks, one centered around 3.4 V vs. Li/Li + and another above 4 V vs. Li/Li + . In order to compare the OER activities of these electrodes, we have analyzed the LSV curves in the η RC range below 250 mV. The onset potentials for OER of all the mesoporous electrodes are found to be quite similar and close to 3 V vs. Li/Li + . However, interestingly, the current responses at different overpotentials (see Figure S15 (b)) and vice versa (see Figure S15 (c)) do not show any correlation with N/C ratio. Therefore, the direct role of N-sites for catalyzing OER cannot be confirmed in these electrodes. Rather, these results evince the predominant effect of nanoconfined Li 2 O 2 in suppressing η RC of the cells. Although Li 2 O 2 is thought to be a wide-bandgap insulator (band gaps of 5-6.4 eV), electrochemically formed Li 2 O 2 usually possesses several defect sites that act as the charge carriers and improve the overall conductivity. 4   atoms resulting in higher mobility of Li + in the Li 2 O 2 crystal. 41 A comparative impedance analysis of discharged KB and N-CMK-3-3 electrodes in Figure S16 indeed shows more facile charge transport for the latter electrode and supports the hypothesis presented in this work. In case of N and O-doped carbons, the electron-withdrawing hetero atoms can activate neighboring C atoms for stronger adsorption of Li and O 2 and induce deposition of disordered or poorly-crystalline Li 2 O 2 as the discharged product, which is consistent with the XRD results in Figure 4 (d) and (e). 43 A prior DFT calculation showed that during DC of the cell the binding of depositing Li 2 O 2 units on top of the initially nucleated disordered Li 2 O 2 layer becomes weak with lower than expected coordination number and continues building up disordered structure. 16 Therefore, in the cases of N-CMK-3 electrodes depositions of the poorly crystalline Li 2 O 2 having weaker bonding, lower coordination number and improved ionic/electronic conductivities continue despite increase in Q DC and maintain low η RC during RC of the cells.

Comparison of rate capabilities of different electrodes in Li-Air cells.
Retention of high Q DC together with low η RC at high current densities is significantly important for the development of practical Li-Air batteries. Figure S17  among these carbon electrodes, N-CMK-3-3 and KB show much better retention of gravimetric energy at high power compared to the other two electrodes. Higher surface area (both S BET and ECSA) and pore volume of N-CMK-3-3 and KB than those of CNT and SP can be attributed to this higher energy and power capabilities. Therefore, consistent with previous report, high surface area and large pore volume in a carbon electrode seem to be essential requirements to achieve simultaneously high energy and power densities during DC of a Li-Air battery. 5 However, Figure 6 (b) shows that the trend of DC performances in these electrodes does not correlate to the rechargeability of the cell. Considering rechargeability, the performance of N-CMK-3-3 profoundly stands out from the other three electrodes. The N-CMK-3-3 electrode consistently maintains a round-trip energy efficiency of > 70% at high j DC up to 500 mA g⁻ 1 , which is 5-15% higher than those of KB, CNT and SP. This significantly higher rechargeability of N-CMK-3-3 electrode at high rate can again be ascribed to the formation of spatially restricted poorly crystalline Li 2 O 2 with better Li + mobilities that can transport charges more rapidly compared to bulk crystalline Li 2 O 2 formed in other electrodes. 15 These results indicate that high surface area and large pore volume coupled with mesoporous framework help N-CMK-3-3 electrode to produce high capacity with efficient rechargeability even at high current rates during both DC and RC, respectively. The efficient rechargeability of N-CMK-3-3 electrode was tested over repeated cycles and was also compared with the cyclability of KB electrode under same conditions and the results are shown in Figure  S18. The cycling data further show better performance for the N-CMK-3-3 electrode in suppressing the η RC over several cycles. However, deterioration in cycling stability is evident for both these electrodes and this can be attributed to parasitic side reactions as explained before.

Conclusion
In summary, we have successfully demonstrated a noncatalytic strategy to improve the discharge capacity of a Li-Air battery without sacrificing the rechargeability and overall energy efficiency of the cell. To do so, we have chosen a mesoporous carbon as the electrode material that offers a confined pore-space for spatially restricted growth of Li 2 O 2 . Unlike uncontrolled growth of Li 2 O 2 in conventional high surface area electrodes, the confined growth inside the mesopores produces nano-structured Li 2 O 2 that is benefited with high mobility of Li + and other charge carriers for facile decomposition during recharge. We have further shown that controlled increase in electrochemically active surface area, pore width and pore volume of the carbon, while maintaining the mesoporous framework, can significantly enhance the discharge capacity and at the same time retains high rechargeability. In our study, we followed a wet chemical pathway for the synthesis of mesoporous carbons using mesoporous silica as the hard template that was impregnated with phenolformaldehyde mixture with or without urea as the carbon precursor.
Our results have shown that addition of urea not only introduces nitrogen as a dopant in the carbon skeleton but also increases the porosity in the framework. For a certain dopant concentration, the porosity reaches a maximum and the electrode produces a discharge capacity that is nearly 2.5 times higher than that of the undoped carbon. Interestingly, despite this significant increase in discharge capacity, the electrode still maintains a round-trip energy efficiency of > 70% which is up to 15% higher than that in the conventional electrodes. Detailed investigations have found that, although enhanced porosity accommodates larger amount of Li 2 O 2 and gives higher discharge capacity, the mesoporous framework in the Ndoped carbons produces spatially-restricted nano-sized and poorly crystalline Li 2 O 2 that decomposes at low recharge potential. Our study demonstrates a prototypical strategy of electrode modification for controlled growth of Li 2 O 2 in order to achieve high capacity with simultaneously high rechargeability and provides crucial insights into electrode design for future development of efficient Li-Air batteries.

Author contributions
A. D. conceived the research, synthesized the materials, carried out the electrochemical measurements and wrote the manuscript. K. I. carried out the OEMS experiments and Y. K. supervised the overall research and contributed to scientific discussions and writing of the manuscript.