注册 登录  
 加关注
   显示下一条  |  关闭
温馨提示!由于新浪微博认证机制调整,您的新浪微博帐号绑定已过期,请重新绑定!立即重新绑定新浪微博》  |  关闭

锂空气电池文献

博士4年的经营,看看会创造多少奇迹

 
 
 

日志

 
 

Oxygen Reduction by Lithium on Model Carbon and Oxidized Carbon Structures  

2011-09-23 13:55:27|  分类: 默认分类 |  标签: |举报 |字号 订阅

  下载LOFTER 我的照片书  |

Journal of The Electrochemical Society, 2011, Vol. 158, No. 10, pp. A1177–A1184
?2011 The Electrochemical Society. All rights reserved.


Up: Issue Table of Contents
Go to: Previous Article | Next Article
Other formats: HTML (smaller files) | PDF (2547 kB)

Ye Xu1 *,z and William A. Shelton2 ?
1Center for Nanophase Materials Sciences, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, USA
2Computer Science and Mathematics Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, USA
(Received: 17 May 2011; accepted: 26 July 2011; published online: 18 August 2011)

Li-air batteries have attracted substantial interest for their high theoretical specific energies, but the oxygen reduction reaction by Li (Li-ORR) that occurs at the carbon cathode remains poorly understood. Periodic density functional theory calculations have been performed to examine the Li-ORR on several model carbon structures, including the graphite(0001) basal plane, the (8,0) single-wall nanotube, the armchair-type edge, and a di-vacancy in the basal plane. The inertness of the basal plane limits the reversible potential of O2 reduction to 1.1 V, and slightly higher to 1.2 V on the curved nanotube. The armchair edge and di-vacancy are highly reactive and significantly oxidized at ambient conditions to various COx groups, which are reduced by Li via redox mechanisms at 1.2–1.4 V. These COx groups can also catalyze O2 reduction at up to 2.3 V (an overpotential of 0.4 V vs. the calculated equilibrium potential for bulk Li2O2 formation) by chelating and stabilizing the LiO2 intermediate. The Li-ORR on graphitic carbon, if via concerted Li+/e? transfer and involving carbon, lithium, and oxygen only, is therefore expected to initiate with the smallest overpotential at under-coordinated carbon centers that are oxidized at ambient conditions. ?2011 The Electrochemical Society



Contents
BODY OF ARTICLE
Methods
Results and Discussion
A.O2 reduction on flat and curved basal plane
B.Redox of COx groups
C.COx-catalyzed O2 reduction
Conclusions
ACKNOWLEDGMENTS
REFERENCES
FIGURES
TABLES
FOOTNOTES

Li-air batteries have gained considerable research interest recently because the Li-O2 electrochemical couple has one of the highest known specific energies. The overall reactions in non-aqueous electrolytes are

(2Li<sub>(s) + O2(g) <--> Li2O2(s),         E[convolution] = 2.96V; 2Li(s)+(1/2)O2(g) <--> Li2O(s),         E[convolution] = 2.91V)

The standard potentials are calculated based on the tabulated standard free energy of formation for the two bulk oxides, lithium peroxide (Li2O2(s)) and lithium oxide (Li2O(s)).1 Li-O2 can theoretically deliver a specific energy of 11.6 kW h/kg of Li. When the mass of oxygen, battery components, and motor efficiency are taken into account, Li-air batteries are expected to deliver 0.8–1.5 kW h/kg,2,3 which is nonetheless several times greater than what Li-ion batteries can achieve. Li-ion batteries have theoretical specific energies of ca. 0.6 kw h/kg of cathode material or 0.2 kW h/kg of battery. It has been estimated that 75 kW h of energy needs to be stored onboard to give a car a range of 200 miles on a single charge, which therefore requires 375 kg of battery, a weight that is heavier even than that of typical gasoline engines in cars. The level of transformational increase in energy storage capacity that Li-O2 promises can extend the range and increase the acceptance of battery electric vehicles, and contribute substantially to the goals of electrifying consumer fleets and reducing reliance on petroleum.4

Every component of the non-aqueous Li-air battery faces substantial challenges.5,6,7,8,9,10,11,12 In particular, prototype Li-oxygen cells, consisting of lithium metal anodes, organic solutions of Li salts as electrolytes, and air-breathing cathodes, showed severe limitations, including high overpotentials in discharge and charge and rapid loss of capacity with cycling, both of which are most likely associated with the cathode. Existing Li-oxygen cells have used carbon black or amorphous carbon5,8,13,14,15 as well as glassy carbon16 as cathodes, which have been reported to discharge at 2.5–2.7 V, or at an overpotential (eta) of 0.3 V or more vs. bulk Li2O2. For comparison, redox lithiation of oxygen functional groups on carbon (COx) has been suggested to occur at 1.5–3 V.17,18,19 While the addition of certain substances (e.g. cobalt phthalocyanine;13 oxides;2,14,15 metals16,20,21,22) to carbon cathodes has been shown to reduce the overpotential and improve capacity retention, knowing the role of carbon itself will clearly benefit the realization of rechargeable Li-air batteries, and will also provide pertinent insight for the use of carbon electrodes for other electrochemical reactions such as the oxygen reduction reaction in hydrogen fuel cells,23 redox reaction in carbon cathode-based Li ion batteries,19 and direct synthesis of hydrogen peroxide,24 as well as thermochemical reactions in general.25 Unfortunately, experiments have so far yielded little mechanistic understanding of the action of carbon in the Li-air cathode beyond what limited information that overall kinetics can reveal.

Quantum chemistry-based theoretical modeling provides a way to directly explore the details of molecular-level surface chemistry that often eludes detection, including the Li-air cathode.26,27,28 Here we report a density functional theory (DFT) based study of the oxygen reduction reaction by Li (Li-ORR), which is the discharge reaction in the non-aqueous air cathode, on clean and oxidized model carbon structures. The structure of typical carbon cathode materials is highly varied and complex, and is poorly characterized experimentally. It stands to reason that the flat basal plane (graphite(0001), henceforth abbreviated as g(0001)) is a prevalent structural motif in graphitic carbon, but it is reasonable also to expect curvature, edges, and defects to be common. We therefore consider the Li-ORR on the following model structures in this study: g(0001) basal plane (including graphene); the (8,0) single-wall nanotube (SWNT) to represent curvature; the armchair-type edge (henceforth referred to as armchair edge) of a graphene nanoribbon (GNR) to represent the edge of graphite; and a di-vacancy in graphene to represent point vacancies. No electrolyte is included in our models because the focus of this study is to understand the intrinsic oxygen reduction activity of carbon itself whereas the decomposition chemistry of common electrolytes29,30,31 can obscure this understanding. Moreover, no clear information is available for the structure of any of the electrode-electrolyte interfaces encountered in the experiments so far.

In the following sections, we consider the reduction of O2 on clean g(0001)/graphene and SWNT, and then the redox of COx groups on pre-oxidized g(0001), armchair edge, and di-vacancy, and finally the reduction of O2 as catalyzed by such COx groups. The approach of N?rskov and co-workers26,32 is followed in calculating the reversible potential based on the thermodynamic, not kinetic, barriers for a Li+/e? transfer step. We find that chemical inertness limits the complete O2 reduction on the basal plane, flat or curved, to ca. 1.2 V (vs. Li/Li+; same below). On the other hand, the armchair edge and di-vacancy are highly reactive and calculated to be oxidized at ambient conditions. These COx groups are reduced by Li at 1.2–1.4 V, but they are also active sites that reduce O2 at up to 2.3 V, for the 1st Li+/e? transfer forming LiO2. Thus these sites offer overpotentials for O2 reduction as small as 0.38 V vs. the calculated equilibrium potential for bulk Li2O2, and are the most likely carbon structures amongst all those considered in this study to be responsible for the reported ORR activity of carbon cathodes.

 


Methods

Spin-polarized periodic DFT calculations have been performed in the generalized gradient approximation [GGA-PBE (Ref. 33)] using the Vienna Ab initio Simulation Package (VASP).34,35,36 The core electrons are described by the PAW method,37 and the Kohn-Sham valence states (1s2s2p for Li, 2s2p for O and C) are expanded in plane wave basis sets up to a kinetic energy of 400 eV. Previously we have calculated gas-phase LiO2 (lithium superoxide), Li2O2 (lithium peroxide), LiO, and Li2O (lithium oxide) (Ref. 28) and obtained close agreement with existing experimental38,39,40,41,42,43 and DFT43 results for the geometric and vibrational properties. In addition, we also obtained close agreement between the experimental and calculated lattice parameters and energetics for bulk Li metal (bcc phase), Li2O2 (F?ppl structure44), and Li2O (antifluorite structure45). Details are provided in Ref. 28. In particular, the calculated standard free energies of formation for the bulk lithium oxides

<i>Delta Gf[convolution] = GLixO(s)[convolution]-xGLi(s)[convolution]-(y/2)GO2(g)[convolution]

under-predict the experimental values:1 for Li2O2(s), ?5.405 vs. ?5.918 eV; for Li2O(s), ?5.292 vs. ?5.826 eV. Therefore the equilibrium potentials according to the Nernst equation, U[convolution]=-((<i>Delta Gf[convolution])/(ne)), are under-estimated by 0.26–0.27 V: 2.703 vs. 2.959 V for Li2O2(s) and 2.646 vs. 2.913 V for Li2O(s). Although the cohesive energy of Li metal is accurately predicted, the energy of the gas-phase O2 is not and is instead deduced from the total energies of H2O and H2 andthe NIST gas-phase enthalpy data, following the approach of N?rskov and coworkers.46

The basal (0001) plane of graphite is modeled by semi-infinite slabs that are constructed from a (4 × 4) surface supercell (32 C atoms per layer) and consist of 1 (graphene) and 3 AB-stacked (0001) layers (Figs. 1(a)1(b)). The optimized nearest C-C bond distance is 1.42 ?, in excellent agreement with experiment.47 The distance between adjacent (0001) layers is fixed at the experimental value of 3.34 ? because the standard GGA does not capture the van der Waals interaction that primarily holds the basal planes together. The 1- and 3-layer slabs are separated from neighboring slabs in the z direction by 25.0 and 18.3 ? of vacuum, respectively. In each slab, any adsorbate together with the top layer is fully relaxed and the remaining layers (in the 3-layer slab) are held fixed at bulk positions. The effect of surface relaxation on adsorption energy (defined below) is small for atomic Li (0.17 eV more stable on the relaxed surface) but substantial for atomic O (over 1 eV more stable on the relaxed surface). The atomization energy of the graphene layer based on the DFT total energies is calculated to be ?7.86 eV per C atom. Intercalation of Li is modeled with a (3 × 3) surface supercell (18 C atoms per layer) with 3 AA-stacked basal planes (not shown in Fig. 1). Three Li atoms are inserted between the top and second basal planes only, for a local composition of LiC6, with the top layer and all the Li atoms relaxed.

Oxygen Reduction by Lithium on Model Carbon and Oxidized Carbon Structures - 伯虎 - 锂空气电池文献 Figure 1.

To explore the effects of curvature in a graphite surface, the (8,0) SWNT is used as the model surface (Fig. 1(c)), which has a radius of 3.132 ?. The SWNT is placed in an orthorhombic unit cell of 13 × 15 × 8.52 ?3 unit cell and is infinitely long in the z direction. The unit cell length in the z direction (8.52 ?) is equal to six times the C-C bond distance of 1.42 ?.

To represent fractured edges of graphite surfaces, a 9 C atom-wide (equivalently, 4 C6 rings wide) GNR with armchair edges is used. One of the two edges is capped by H atoms and the other is exposed (Fig. 1(d)). The edge-to-edge distance in the y direction (measured between C atoms) and plane-to-plane distance in the z direction between neighboring GNRs are 10.3 and 10.0 ?, respectively, to allow the edge sites to be studied in isolation. The entire GNR is relaxed except the capping H atoms, which are fixed at optimized positions obtained by initially fixing the GNR and relaxing the H atoms. Although the armchair edge is chemically less reactive than the zigzag edge,48,49,50,51 it is more stable than the zigzag edge because the C[equivalent]C triple bond character removes the dangling bonds at the armchair edge and reduces the likelihood of spontaneous reconstruction.52 As has been shown before for other species and will be seen below, the armchair edge is still significantly more reactive than the basal plane. A di-vacancy, a common type of point defect in graphite, is used to represent point defects. This model is created by removing two adjacent C atoms in a (7 × 7) graphene supercell (Fig. 1(e)).

The Brillouin zone is sampled with 5 × 5 × 1, 3 × 3 × 1, 5 × 1 × 1, and 1 × 1 × 5 Monkhorst-Pack (MP) k-point meshes for the (3 × 3) and (7 × 7) g(0001)/graphene surfaces, GNR, and (8,0) nanotube, respectively, which are found to be sufficient for the DFT total energies to converge to within 0.05 eV. For all the model carbon structures, a first-order Methfessel-Paxton scheme is used to smear the electronic states with a width of 0.1 eV.

Adsorption takes place on only one side of the g(0001) surfaces (for the 3-layer slab, the relaxed side) and on the non-H-capped edge of the GNR, with electrostatic decoupling in the z and y directions, respectively.53 For the (8,0) SWNT electrostatic decoupling is applied in the y direction. The adsorption of lithium-oxygen species on the g(0001)/graphene surfaces is considered in various high-symmetry sites or configurations only, the most stable of which are reported below. For the various oxidized carbon structures and their lithiated states at the armchair edge and the di-vacancy, we employed a heuristic annealing approach based on ab initio molecular dynamics as described previously54 to determine the minimum-energy structures.

The DFT adsorption energy of a species is calculated as DeltaE=(E+E<sub>A*ZPE)-EC-(EA+E<sub>AZPE), where E, EC, and EA are the total energies of the system, the clean carbon substrate, and the neutral adsorbate species in the gas phase, respectively, and E<sub>A*ZPE and E<sub>AZPE are the zero-point energies of species A adsorbed on the surface and free in the gas phase, respectively. Thus a negative DeltaE indicates exothermic adsorption and vice versa. Reaction activation barriers are calculated as Ea=(ETS+E<sub>TSZPE)-(EIS+E<sub>ISZPE), where IS and TS indicate the reactant and corresponding transition states, respectively. The transition states for elementary thermochemical surface reaction steps are determined using the climbing-image nudged elastic band55 and dimer56 methods.

The free energy of an adsorbed species is calculated as GA*=E-Eslab+DeltaGA*=E-Eslab+(E<sub>A*ZPE+DeltaUA*(T)-TSA*(T)). E<sub>A*ZPE, DeltaUA*, and SA* are calculated from the vibrational frequencies associated with the normal modes of the adsorbed species. Following the approach of N?rskov and co-workers,26,32 we set the potential at which a solvated Li+ ion and an electron (e?) in the electrode are in equilibrium with bulk Li equal to zero

Li<sub>(s) <--> Li+ + e-,Delta G = 0   eV   at   U = 0V

The free energy of an electron in the electrode is assumed to be linearly dependent on the electrode potential and is shifted by ?eU at a different potential U, so that for a Li atom adsorbed on the cathode side, GLi=GLi(s)-eU. For simplicity, the concentration of Li+ is taken to be 1 M. A different concentration would shift the free energy by +kTln[Li+] assuming ideal solution behavior. A concentration of 0.5 M would lower the free energy of Li+ by a negligible ?0.02 eV at room temperature. For the reason mentioned before the solvent is left unspecified, so the equilibrium represented above cannot be placed on the absolute scale and therefore cannot be directly related to the hydrogen electrode. Effects on surface species due to the electrolyte and interfacial electric field are therefore also ignored in this study.

On each carbon structure we then calculate the relative stabilities of various lithium-oxygen species as functions of the electrode potential. Knowing the relative stabilities of the possible intermediates allows us to identify possible mechanisms for the Li-ORR. Because of the approximations mentioned above, our results for reaction intermediates and pathways should be regarded as a first approximation that represents the situation where the adsorbate-surface interaction is dominant.

The Bader charge partition analysis57 was performed using the code of Henkelman et al.58 to determine the charges of individual atoms in the clusters.

 


Results and Discussion


A.O2 reduction on flat and curved basal plane

We begin by examining the interaction of Li with the g(0001) basal plane in the absence of oxygen. Li prefers to adsorb in the C6 ring center site (Fig. 2(a)) with an adsorption energy of ?0.90 eV (one Li atom per (4 × 4) unit cell) with respect to a gas-phase Li atom, which is in agreement with previous theoretical studies59,60 and considerably weaker than Li adsorption on metals.28 When referenced to bulk bcc Li metal, the adsorption energy is +0.66 eV (Table I), indicating that Li metal does not wet the g(0001) surface. The interstice between two adjacent basal planes of graphite, on the other hand, stabilizes Li better than the surface: The adsorption energy is ?0.08 eV/Li with respect to bulk Li at a local composition of LiC6 (Fig. 2(b)) The reversible potential (Urev), which makes DeltaG = 0 eV for a particular process, is 0.06 V for the intercalation of Li into graphite (Li+ + e? --> Li(int)) (Fig. 3), in agreement with the reported staging potential of less than 0.1 V for Li concentration of Li0.5C6 and higher61,62 and suggests that DFT offers excellent accuracy in the description of lithium-carbon interaction, as has been noted before.63,64 U exceeding Urev makes DeltaG > 0 and introduces a corresponding overpotential eta = DeltaG/e for a given step.32

Oxygen Reduction by Lithium on Model Carbon and Oxidized Carbon Structures - 伯虎 - 锂空气电池文献 Figure 2. Oxygen Reduction by Lithium on Model Carbon and Oxidized Carbon Structures - 伯虎 - 锂空气电池文献 Figure 3.

The perfect g(0001) plane is highly resistant to oxidation. It remains clean of oxygen at ambient conditions, and etching graphite in air requires over 500°C. In accordance with this and previous calculations,65 we find molecular O2 not to adsorb on g(0001). The reduction of O2 on g(0001), if it occurs at all, must proceed through an associative channel. The intermediates in this channel, LiO2 and LiO2Li, both adsorb via the Li (Figs. 2(c), 2(d)) with adsorption energies of ?0.29 and ?0.35 eV, respectively. The dissociation of the O-O bond in LiO2 via isomerization to OLiO has an activation barrier of 2.67 eV, considerably greater than the desorption barrier (the negative of the adsorption energy as first approximation). This stands in contrast to metal surfaces, where the lithiation of O2 activates the O-O bond.28 The relatively weak adsorption energies suggest that these Lix-O2 intermediates are highly mobile on g(0001) and have a high probability of entering a different phase (e.g., the electrolyte; solid agglomerates). LiO2 intercalated in between the top and second layers is 0.5 eV less stable than surface-adsorbed and is therefore not investigated further.

By comparing the 1-layer (graphene) and 3-layer g(0001) surfaces (Table I), we conclude that the adsorption energies of Li-oxygen species are mostly independent of the number of basal planes, and that the surfaces of graphene and graphite are chemically indistinguishable. Therefore graphene is not investigated further.

Next we deduce the mechanism of the Li-ORR on g(0001) from how the free energies of the various Li-oxygen intermediates change as a function of electrode potential (Fig. 3). At T = 298 K and pO2 = 0.1 bar, Urev = 1.12 V for the 1st lithiation step (O2(g) + Li+ + e? --> LiO2), and the subsequent steps forming LiO2Li (2nd Li+/e? transfer) and (Li2O)2 (4th Li+/e? transfer) are all downhill in free energy at this potential. Alternatively, two LiO2 molecules may disproportionate to generate LiO2Li and eject an O2, which is a chemical step with DeltaG = ?0.29 eV. The formation of weakly bound LiO2 and other Li-oxygen intermediates on g(0001) may be undesirable because they are likely to be consumed in side reactions, such as with the organic solvents used in most experimental Li-air cells to date. Whether LiO2 undergoes further reduction, disproportionation, or desorption will depend on the relative kinetic rates of these steps.

It remains to be determined whether the transfer of the first e? is in fact accompanied by the simultaneous transfer of Li+. If it does not, the driving force for the first e? transfer step should be smaller. Therefore, the 1.12 V for the direct formation of LiO2 represents an upper bound for the Urev of the reduction of molecular O2 on g(0001).

Previous studies reported that curvature increases the reactivity of the graphene surface, although the barrier for molecular O2 adsorption and dissociation remain ca. 1.5 eV with respect to gas-phase O2 in its triplet ground state even on the highly curved (8,0) SWNT.65 Both LiO2 (Fig. 4(b)) and LiO2Li (Fig. 4(c)) adsorb ca. 0.2 eV more strongly than on g(0001) (Tables I) (Ref. 66). The Urev for the reduction of O2 to LiO2 therefore increases commensurably from 1.12 to 1.24 V (Fig. 3, long dashed lines). The effect of curvature on the ORR activity therefore appears to be minute and difficult to be distinguished experimentally from the activity of the flat basal plane.

Oxygen Reduction by Lithium on Model Carbon and Oxidized Carbon Structures - 伯虎 - 锂空气电池文献 Figure 4.


B.Redox of COx groups

It is well known that carbon materials often contain appreciable amounts of oxygen functional groups (COx), such as ether, ketone, and quinone groups,67 so the role of such oxygen groups needs to be taken into consideration. We first investigate the reduction of the COx groups themselves via redox mechanisms (COx + xLi+ + xe? --> C(OLi)x), which strictly speaking is not O2 reduction but can nonetheless occur in the carbon-lithium-oxygen electrochemical system, and therefore need to be considered.


B1.Epoxy on basal plane

Although the dissociative adsorption of O2 on g(0001) is unlikely, we consider atomic O in the case that it is generated via other means and deposited on the surface. Atomic O prefers to add itself across a C-C bond forming an epoxy group (Fig. 2(e)), pulling both C atoms up by 0.4 ?. The epoxy O is thermodynamically highly unstable compared to gas-phase O2 (cf. Fig. 3), and its equilibrium concentration on g(0001) should be vanishingly small at ambient conditions. Successive lithiation of O produces LiO and Li2O (Figs. 2(f)2(g)), both of which lie with their long axes parallel to the surface and bind through the O atom to a single C atom (instead of two C atoms as the epoxy O), pulling it upward by 0.5–0.6 ?. LiO has an adsorption energy of ?0.97 eV, and Li2O adsorbs even more weakly (DeltaE = ?0.26 eV). The calculated diffusion barrier for the epoxy O via transitioning over the top of a C atom is 0.69 eV, so LiO and Li2O, which are lithiated versions of O, are expected to diffuse with smaller barriers and therefore rapidly on g(0001) at room temperature. The first Li+/e? transfer to the epoxy O (C2 = O + Li+ + e? --> C?OLi) has a Urev = 1.19 V, whence subsequent reduction to Li2O and agglomeration to clusters (e.g., (Li2O)2; Fig. 2(h)) and bulk Li2O are all downhill in free energy (Fig. 3).


B2.Armchair edge

At the armchair edge atomic Li prefers to be chelated by two edge C atoms with an adsorption energy of ?1.03 eV per Li atom with respect to bulk Li (Table II; Fig. 5(a)) instead of being endothermic as on the basal plane. Such under-coordinated C centers are likely one of the reasons for the irreversible loss of Li that occurs in the charging cycles of Li-ion batteries using graphite anodes. We calculated that the dissociation of molecular O2 at the clean armchair edge has a negligibly small activation barrier, in line with previously reported barrier-less O2 dissociation at vacancies.49,68 Therefore the armchair edge, as well as other under-coordinated carbon sites, is almost certainly to be found in an oxidized state when exposed to even small amounts of oxygen. To determine how oxidized the armchair edge is at different conditions, we begin by exploring a number of carbon-oxygen edge structures for different C:O stoichiometries, ranging from 1 to 5 O atoms per C[equivalent]C (denoted as 2C:xO (x = 1–5) henceforth). The most stable structure for each stoichiometry is found to be epoxy (2C:1O; Fig. 5(b)), quinone (2C:2O; Fig. 5(c)), anhydride (2C:3O; Fig. 5(d)), carbonate (2C:4O; Fig. 5(e)), and lactone (2C:5O; Fig. 5(f)).

Oxygen Reduction by Lithium on Model Carbon and Oxidized Carbon Structures - 伯虎 - 锂空气电池文献 Figure 5.

Ab initio thermodynamic analysis69,70,71,72 based on the 2C:xO structures listed in Table II shows that the armchair edge is oxidized to the greatest extent considered in our study (2C:5O) relative to gas-phase O2 over a range of temperature and pressure, including ambient conditions (Fig. 6(a)), which is in complete contrast to the basal plane. We also checked the stability of the 2C:xO structures vs. gas-phase CO or CO2 by calculating the DeltaG for the oxidative etching process 2C:   xO+(<i>x/2)C-->(<i>x/2)CO2(g):71 DeltaG<0 means that less than half of a particular oxygen group would remain at ambient conditions, and vice versa. Table III suggests that thermodynamically, 2C:xO (x = 1–4) are stable, whereas 2C:5O is expected predominantly to decompose into gas-phase CO2. In the absence of 2C:5O, 2C:4O is the dominant phase across a wide range of O2 chemical potentials, including ambient conditions (Fig. 6(a)). The fact that graphitic carbon does not spontaneously combust in air indicates significant activation barriers for O2 addition to certain COx moieties, for the decomposition of COx to gas-phase CO2, or for both, which we do not examine in the present study. If a significant amount of atomic oxygen is concentrated on under-coordinated carbon sites through discharge/charge cycling, it is possible that the carbon material will be oxidatively etched away beginning at these sites.

Oxygen Reduction by Lithium on Model Carbon and Oxidized Carbon Structures - 伯虎 - 锂空气电池文献 Figure 6.

The lithiated 2C:4O, illustrated in Fig. 5(g), shows that the carbonate group has been decomposed into two carboxylate groups, and it bears no resemblance to any of the bulk Li oxides. The potential-dependent phase diagram comparing the clean, oxidized, and oxidized and lithiated armchair edge is present in Fig. 7. The Urev for the first Li+/e? transfer to each oxygen atom is found to be 1.32 V for 2C:4O-->2C:4O:4Li (Fig. 7). For comparison, we also investigated the redox of an isolated carbonyl group, 1C:1O-->1C:1O:1Li (Fig. 5(j)5(k)), for which the reversible potential is similarly 1.44 V (not shown in Fig. 7).

Oxygen Reduction by Lithium on Model Carbon and Oxidized Carbon Structures - 伯虎 - 锂空气电池文献 Figure 7.


B3.Di-vacancy in basal plane

The di-vacancy is used to represent point vacancies in graphite surfaces. In the absence of adsorbates, neighboring under-coordinated C atoms bond with each other to minimize dangling bonds, and form five-member rings.73 Li occupies the position of the missing C atom but prefers to stay out of the plane of the graphene (Fig. 8(a)). When molecular O2 is present, it dissociates with a negligible barrier49,68 and forms ether groups (C-O-C; 1V:1O; Fig. 8(b)) at even vacancies,74 followed by lactone groups (C-O-C=O; 1V:3O; Fig. 8(c)).68,71 Our calculations confirm the finding of Carlsson et al.71 that 1V:3O is more stable than 1V:1O across a wide range of O2 chemical potential (Fig. 6(b)). As in the case of the armchair edge, heavily oxidized moieties are thermodynamically favored at ambient conditions but they may be unstable with respect to decomposition to CO2. Only ether and lactone groups are stable in this regard.71 The reversible potential for the reduction of 1V:3O to 1V:3O:3Li (Fig. 8(d)) is calculated to be 1.23 V (Fig. 7).

Oxygen Reduction by Lithium on Model Carbon and Oxidized Carbon Structures - 伯虎 - 锂空气电池文献 Figure 8.

The COx groups at the armchair edge and the di-vacancy share several common characteristics. Table II lists the average adsorption energy of an O atom in the various COx groups considered in this study. In all cases the average adsorption energy per O atom is substantially more negative than that of Li, suggesting that O can out-compete Li for direct bonding to the edge C sites. The COx groups tend to have a large number of C-O vibrational modes, some of which are summarized in Table S2 in Supplemental Information. These modes fall in approximately two groups: 1700–1850 cm?1 for carbonyl C-O stretch, and 1200–1450 cm?1 for various C-O-C stretches, in line with experimental observations.67

Overall our electrochemical modeling results indicate that the redox of these COx groups is limited to the potential range of 1.2–1.5 V, which is at the lower end of the reported redox activity on oxidized carbon (1.5–3 V).17,18,19 According to Bader charge partition analysis, the O atoms in the COx groups are highly basic so they are the preferred sites for attack by acidic Li+ ions, but it is primarily the C atoms, and not the O atoms, in the COx groups that are reduced (see Table S2 for the Bader charges).

It is interesting to compare the energetics of 2C:4O:4Li (at the armchair edge), 1V:3O:3Li (at the di-vacancy), and bulk Li2O2, states that share the same overall stoichiometry. Figure 7 shows that the stability decreases in the order of 2C:4O:4Li > 1V:3O:3Li > bulk Li2O2, suggesting that (1) the highly reactive under-coordinatedC sites significantly stabilizes Li-oxygen species, even relative to bulk Li2O2; (2) the local geometry of the carbon is important: Higher dimensionality (such as the open armchair edge) offers more bonding sites and therefore stabilizes Li-oxygen species better, whereas lower dimensionality (such as the point di-vacancy) is less stabilizing and causes such sites to be more readily oxidized back to COx upon recharge.

 


C.COx-catalyzed O2 reduction

That oxygen functionalities can enhance redox kinetics of various systems including O2 has been previously noted.75,76 Therefore we also examined the interaction of several different COx groups with LiO2, the key O2 reduction intermediate, and found that they can stabilize LiO2 better than the g(0001) basal plane via bonding between the Li and carbonyl O in the COx groups. For instance, the adsorption energy of LiO2 on an isolated epoxy O on g(0001) (Fig. 2(i)) is ?0.93 eV, vs. ?0.29 eV on g(0001) (see Table IV for comparison). Consequently, the formation of LiO2 can occur at a higher potential of 1.77 V (Fig. 3, LiO2@O). Likewise, the 2C:4O carbonate group at the armchair edge adsorbs LiO2 via a single carbonyl O (Fig. 5(h)) with a similar adsorption energy of ?1.10 eV, and the corresponding Urev for LiO2 formation is 1.86 V (Fig. 7). A pair of neighboring carbonyl O atoms can stabilize LiO2 even better: For instance, the 1V:3O group at the di-vacancy is capable of chelating LiO2 through the terminal O atom in the lactone and the side carbonyl O atom (Fig. 8(e)). The adsorption energy is ?1.57 eV, and the corresponding Urev for LiO2 formation is 2.33 V (Fig. 7). Also if we take the 2C:3O anhydride group as a different structural motif from 2C:4O at the armchair edge: A Urev of 2.25 V is found for LiO2 formation (Fig. 7) because adjacent anhydride groups can also chelate LiO2 (Fig. 5(i)). These processes occur at clearly higher potentials than the redox of the COx groups themselves by Li.

Our findings suggest that the reduction of O2, if occurring via the formation of LiO2, can be strongly promoted by COx groups. The COx-catalyzed 1st Li+/e? transfer to O2 occurs at the highest potential (i.e., the smallest overpotential, eta = 0.64 or 0.38 V vs. the tabulated or calculated equilibrium potential for bulk Li2O2, respectively) among the electrochemical processes considered in this study, via a chelating mechanism that is reminiscent of the inner-sphere catalysis of metal ion redox by carbonyl groups proposed by McCreery and co-workers.75 We therefore propose that under-coordinated carbon structures that feature a high concentration of oxygen ligands in general offer high activity for the reduction of O2 by Li.

Further steps in the reduction of O2 at these sites are the subject of ongoing investigation. Preliminary results for the 2nd Li+/e? transfer forming LiO2Li (images not shown) indicate that such a step is slightly exergonic (DeltaG = ?0.04 eV) at 1.77 V at the epoxy O center on g(0001), but is significantly endergonic at the COx sites at the armchair edge and di-vacancy because the structural models per se lack the additional COx sites for stabilizing the 2nd Li atom. It is a reflection of how dependent the Li-ORR on graphitic carbon is on the presence of multiple closely located COx groups. If a favorable local environment is absent, then at the high potential of 2.3 V the Li-ORR would terminate after only one Li+/e? transfer, leaving reactive superoxide species to undergo thermal reactions, such as with itself or with the electrolyte. The transfer of the 2nd Li+/e? would occur with greater overpotential.

 


Conclusions

Periodic DFT calculations have been performed in conjunction with thermodynamic modeling to investigate the initial stages of the oxygen reduction reaction by Li (Li-ORR) on several model graphitic carbon structures, including the flat and curved graphite basal plane (the latter represented by the (8,0) single-wall nanotube (SWNT)), the armchair-type edge of a graphene nanoribbon, which represents the edge of the basal plane, and a di-vacancy as a model of point vacancies in the basal plane. Li intercalation in between basal planes is calculated to occur at less than 0.1 V, in agreement with experiments.

Unlike metal surfaces,28 graphitic carbon presents surface structures that lie at opposite extremes of reactivity. The basal plane does not well stabilize the key intermediate, Li superoxide (LiO2), which limits the reversible potential of complete O2 reduction to 1.12 V, and to a marginally higher 1.24 V on the curved surface of the SWNT. On the other hand, the armchair edge and di-vacancy are highly reactive and are calculated to be significantly oxidized at ambient conditions, to carbonate and lactone groups respectively. Such oxidized carbon structures (COx) can be reduced by Li via redox mechanisms forming C(OLi)x at 1.2–1.4 V. The COx groups can also serve as the active sites for catalyzing O2 reduction, which can occur at 1.8–2.3 V for the formation of LiO2 because LiO2 can be chelated and stabilized by neighboring oxygen ligands. These results are summarized in Table V. Therefore, carbon itself can play an active role in the Li-ORR and provide an important context in which to interpret experimental results for carbon cathodes. The propensity to generate the superoxide species, however, needs to be taken into consideration for non-aqueous Li-air batteries using carbon cathodes and organic electrolytes.

Overall, the reduction of O2 by Li on graphitic carbon occurs with the smallest overpotentials at under-coordinated carbon centers that are oxidized under ambient conditions. When compared to our calculated equilibrium potential for bulk Li2O2, our results closely approach the reported Li-ORR activities on carbon cathodes (eta >= 0.4 V). We conclude that a high concentration of oxygen ligands can confer carbon structures high activity for O2 reduction by Li. Whether the actual discharge reaction involves carbon, oxygen, and lithium only; and whether the electrochemical steps involve the simultaneous transfer of Li+/e? to oxidants as modeled here, remain to be determined. Our study represents a first attempt at shedding light on the mechanistic details of the Li-ORR on carbon electrodes and the role that the intrinsic reactivity of the carbon plays in Li-oxygen surface electrochemistry.

 


ACKNOWLEDGMENTS

This research is sponsored by the Laboratory Directed Research and Development Program of Oak Ridge National Laboratory (ORNL), which is managed by UT-Battelle, LLC, for the U. S. Department of Energy; and used computing resources of the National Center for Computational Sciences at ORNL, which is supported by the Office of Science of the U.S. Department of Energy under Contract No. DE-AC05-00OR22725.

 


REFERENCES



Auxiliary Material (EPAPS)


References
M. W. Chase, Jr., NIST-JANAF Themochemical Tables (1998). first citation in article
A. Débart, A. J. Paterson, J. Bao, and P. G. Bruce, Angew. Chem., Int. Edit., 47, 4521 (2008). first citation in article
B. Kumar, J. Kumar, R. Leese, J. P. Fellner, S. J. Rodrigues, and K. M. Abraham, J. Electrochem. Soc., 157, A50 (2010). first citation in article
Basic Research Needs for Electrical Energy Storage, Office of Basic Energy Sciences, U.S. Department of Energy, Washington, DC (2007). first citation in article
S. D. Beattie, D. M. Manolescu, and S. L. Blair, J. Electrochem. Soc., 156, A44 (2009). first citation in article
X.-H. Yang, P. He, and Y.-Y. Xia, Electrochem. Commun., 11, 1127 (2009). [Inspec] first citation in article
D. Aurbach, M. Daroux, P. Faguy, and E. Yeager, J. Electroanal. Chem., 297, 225 (1991). [ISI] first citation in article
J. Read, J. Electrochem. Soc., 149, A1190 (2002). [ISI] first citation in article
J. Read, J. Electrochem. Soc., 153, A96 (2006). first citation in article
W. Xu, J. Xiao, D. Y. Wang, J. A. Zhang, and J. G. Zhang, Electrochem. Solid-State Lett., 13, A48 (2010). first citation in article
W. K. Behl, J. Read, A. Pitt, J. Wolfenstine, and D. Foster, J. Indian Chem. Soc., 82, 1055 (2005). first citation in article
X.-H. Yang and Y.-Y. Xia, J. Solid State Electrochem., 14, 109 (2010). first citation in article
K. M. Abraham and Z. Jiang, J. Electrochem. Soc., 143, 1 (1996). [ISI] first citation in article
T. Ogasawara, A. Debart, M. Holzapfel, P. Novak, and P. G. Bruce, J. Am. Chem. Soc., 128, 1390 (2006). [MEDLINE] first citation in article
A. Débart, J. Bao, G. Armstrong, and P. G. Bruce, J. Power Sources, 174, 1177 (2007). first citation in article
Y.-C. Lu, H. A. Gasteiger, E. Crumlin, R. McGuire, and Y. Shao-Horn, J. Electrochem. Soc., 157, A1016 (2010). first citation in article
T. Le Gall, K. H. Reiman, M. C. Grossel, and J. R. Owen, J. Power Sources, 119, 316 (2003). [ISI] first citation in article
H. Chen, M. Armand, G. Demailly, F. Dolhem, P. Poizot, and J. M. Tarascon, ChemSusChem, 1, 348 (2008). first citation in article
S. W. Lee, N. Yabuuchi, B. M. Gallant, S. Chen, B. S. Kim, P. T. Hammond, and Y. Shao-Horn, Nat. Nanotechnol., 5, 531 (2010). first citation in article
A. Dobley, C. Morein, and K. Abraham, Cathode optimization for lithium-air batteries, presented at The 208th Meeting of the Electrochemical Society, Los Angeles, CA (2005); October 16-21, paper 823. first citation in article
Y.-C. Lu, H. A. Gasteiger, M. C. Parent, V. Chiloyan, and Y. Shao-Horn, Electrochem. Solid-State Lett., 13, A69 (2010). first citation in article
Y.-C. Lu, Z. Xu, H. A. Gasteiger, S. Chen, K. Hamad-Schifferli, and Y. Shao-Horn, J. Am. Chem. Soc., 132, 12170 (2010). first citation in article
K. P. Gong, F. Du, Z. H. Xia, M. Durstock, and L. M. Dai, Science, 323, 760 (2009). [MEDLINE] first citation in article
I. Yamanaka, T. Hashimoto, R. Ichihashi, and K. Otsuka, Electrochim. Acta, 53, 4824 (2008). first citation in article
D. S. Yu, E. Nagelli, F. Du, and L. M. Dai, J. Phys. Chem. Lett., 1, 2165 (2010). first citation in article
J. S. Hummelsh?j, J. Blomqvist, S. Datta, T. Vegge, J. Rossmeisl, K. S. Thygesen, A. C. Luntz, K. W. Jacobsen, and J. K. N?rskov, J. Chem. Phys., 132, 071101 (2010). [MEDLINE] first citation in article
J. Chen, J. S. Hummelsh?j, K. S. Thygesen, J. S.G. Myrdal, J. K. N?rskov, and T. Vegge, Catal. Today, 165, 2 (2011). first citation in article
Y. Xu and W. A. Shelton, J. Chem. Phys., 133, 024703 (2010). [MEDLINE] first citation in article
G. M. Veith and N. J. Dudney, J. Electrochem. Soc., 158, A658 (2011). first citation in article
J. Xiao, J. Z. Hu, D. Y. Wang, D. H. Hu, W. Xu, G. L. Graff, Z. M. Nie, J. Liu, and J. G. Zhang, J. Power Sources, 196, 5674. first citation in article
S. A. Freunberger, Y. H. Chen, Z. Q. Peng, J. M. Griffin, L. J. Hardwick, F. Barde, P. Novak, and P. G. Bruce, J. Am. Chem. Soc., 133, 8040. first citation in article
J. K. N?rskov, J. Rossmeisl, A. Logadottir, L. Lindqvist, J. R. Kitchin, T. Bligaard, and H. Jonsson, J. Phys. Chem. B, 108, 17886 (2004). first citation in article
J. J. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev. Lett., 77, 3865 (1996). first citation in article
G. Kresse and J. Furthmüller, Phys. Rev. B, 54, 11169 (1996). first citation in article
G. Kresse and J. Furthmüller, Comput. Mater. Sci., 6, 15 (1996). first citation in article
G. Kresse and J. Hafner, Phys. Rev. B, 49, 14251 (1994). first citation in article
G. Kresse and D. Joubert, Phys. Rev. B, 59, 1758 (1999). first citation in article
L. Andrews, J. Chem. Phys., 50, 4288 (1969). [ISI] first citation in article
R. C. Spiker and L. Andrews, J. Chem. Phys., 58, 702 (1973). [ISI] first citation in article
C. Yamada, M. Fujitake, and E. Hirota, J. Chem. Phys., 91, 137 (1989). [ISI] first citation in article
L. Andrews, W. Saffell, and J. T. Yustein, Chem. Phys., 189, 343 (1994). [Inspec] [ISI] first citation in article
D. Bellert and W. H. Breckenridge, J. Chem. Phys., 114, 2871 (2001). [ISI] first citation in article
X. Wang and L. Andrews, Mol. Phys., 107, 739 (2009). first citation in article
H. F?ppl, Z. Anorg. Allg. Chem., 291, 12 (1957). [ISI] first citation in article
R. Wyckoff, Crystal Structures, Interscience, New York (1963). first citation in article
J. I. Martínez, H. A. Hansen, J. Rossmeisl, and J. K. N?rskov, Phys. Rev. B, 79, 045120 (2009). first citation in article
M. S. Dresselhaus, G. Dresselhaus, and P. C. Eklund, Science of Fullerences and Carbon Nanotubes, Academic, San Diego (1996). first citation in article
N. B. Arboleda, H. Kasai, H. Nakanishi, W. A. Dino, and T. Sugimoto, J. Appl. Phys., 96, 6331 (2004). [ISI] first citation in article
A. Allouche and Y. Ferro, Carbon, 44, 3320 (2006). first citation in article
D. E. Jiang, B. G. Sumpter, and S. Dai, J. Chem. Phys., 126, 134701 (2007). [MEDLINE] first citation in article
C. Uthaisar, V. Barone, and J. E. Peralta, J. Appl. Phys., 106, 113715 (2009). first citation in article
P. Koskinen, S. Malola, and H. Hakkinen, Phys. Rev. Lett., 101, 115502 (2008). [MEDLINE] first citation in article
J. Neugebauer and M. Scheffler, Phys. Rev. B, 46, 16967 (1992). [MEDLINE] first citation in article
Y. Xu, W. A. Shelton, and W. F. Schneider, J. Phys. Chem. A, 110, 5839 (2006). [MEDLINE] first citation in article
G. Henkelman, B. P. Uberuaga, and H. Jónsson, J. Chem. Phys., 113, 9901 (2000). first citation in article
G. Henkelman and H. Jónsson, J. Chem. Phys., 111, 7010 (1999). first citation in article
R. F. W. Bader, Atoms in Molecules - A Quantum Theory, Oxford University Press, Oxford (1990). first citation in article
G. Henkelman, A. Arnaldsson, and H. Jonsson, Computational Materials Science, 36, 354 (2006). [Inspec] first citation in article
F. Valencia, A. H. Romero, F. Ancilotto, and P. L. Silvestrelli, J. Phys. Chem. B, 110, 14832 (2006). [Inspec] first citation in article
K. Rytkonen, J. Akola, and M. Manninen, Phys. Rev. B, 75, 075401 (2007). first citation in article
G.-A. Nazri and G. Pistoia, Lithium Batteries: Science and Technology, Springer, New York (2003). first citation in article
Y. F. Reynier, R. Yazami, and B. Fultz, J. Electrochem. Soc., 151, A422 (2004). [ISI] first citation in article
K. R. Kganyago and P. E. Ngoepe, Phys. Rev. B, 68, 205111 (2003). [ISI] first citation in article
J. T. Titantah, D. Lamoen, M. Schowalter, and A. Rosenauer, Carbon, 47, 2501 (2009). first citation in article
S. P. Chan, G. Chen, X. G. Gong, and Z. F. Liu, Phys. Rev. Lett., 90, 086403 (2003). [MEDLINE] first citation in article
See supplementary material at http://dx.doi.org/10.1149/1.3625620 for additional information. [EPAPS] first citation in article
T. Szabo, O. Berkesi, P. Forgo, K. Josepovits, Y. Sanakis, D. Petridis, and I. Dekany, Chem. Mater., 18, 2740 (2006). first citation in article
S. M. Lee, Y. H. Lee, Y. G. Hwang, J. R. Hahn, and H. Kang, Phys. Rev. Lett., 82, 217 (1999). first citation in article
M. V. Bollinger, K. W. Jacobsen, and J. K. N?rskov, Phys. Rev. B, 67, 085410 (2003). [ISI] first citation in article
K. Reuter and M. Scheffler, Phys. Rev. B, 65, 035406 (2002). first citation in article
J. M. Carlsson, F. Hanke, S. Linic, and M. Scheffler, Phys. Rev. Lett., 102, 166104 (2009). first citation in article
R. B. Getman, Y. Xu, and W. F. Schneider, J. Phys. Chem. C, 112, 9559 (2008). first citation in article
J. M. Carlsson and M. Scheffler, Phys. Rev. Lett., 96, 046806 (2006). [MEDLINE] first citation in article
A. Barinov, O. B. Malcioglu, S. Fabris, T. Sun, L. Gregoratti, M. Dalmiglio, and M. Kiskinova, J. Phys. Chem. C, 113, 9009 (2009). first citation in article
C. A. McDermott, K. R. Kneten, and R. L. McCreery, J. Electrochem. Soc., 140, 2593 (1993). [ISI] first citation in article
S. Ranganathan and R. L. McCreery, Anal. Chem., 73, 893 (2001). [ISI] [MEDLINE] first citation in article


FIGURES


Oxygen Reduction by Lithium on Model Carbon and Oxidized Carbon Structures - 伯虎 - 锂空气电池文献 Full figure (41 kB)

Fig. 1. GGA-PBE optimized structural models in top (upper panels) and side (lower panels) views for: (a) graphene; (b) 3-layer g(0001); (c) (8,0) SWNT; (d) GNR with armchair edge; (e) graphene with a di-vacancy. Unit cells, axes, and two types of site are as labeled. First citation in article


Oxygen Reduction by Lithium on Model Carbon and Oxidized Carbon Structures - 伯虎 - 锂空气电池文献 Full figure (105 kB)

Fig. 2. Minimum-energy geometries in top (upper panels) and side (lower panels) views for Li-O species on g(0001): (a) Li (adsorbed); (b) Li (intercalated); (c) LiO2; (d) LiO2Li; (e) O; (f) LiO; (g) Li2O; (h) (Li2O)2; (i) LiO2@O. Grey, white, and red spheres represent C, Li, and O atoms respectively. Only top two layers of graphite are shown. First citation in article


Oxygen Reduction by Lithium on Model Carbon and Oxidized Carbon Structures - 伯虎 - 锂空气电池文献 Full figure (63 kB)

Fig. 3. The potential-dependent lithium-oxygen surface phase diagram for g(0001) and (8,0) SWNT. Zero energy (x-axis) corresponds to clean carbon structures, one O2 molecule at the gas-phase conditions of T = 298 K and pO2 = 0.1 bar, and four pairs of Li+/e; thus DeltaG=GA*-GO2(g)-4GLi. Lines are labeled by the respective phases. Solid lines are phases on g(0001); long dashed lines with circle symbols are phases on the SWNT; and dotted lines are bulk phases (“s”). “Li(int)” indicates intercalated Li. First citation in article


Oxygen Reduction by Lithium on Model Carbon and Oxidized Carbon Structures - 伯虎 - 锂空气电池文献 Full figure (34 kB)

Fig. 4. Minimum-energy geometries for (a) Li; (b) LiO2; (c) LiO2Li on (8,0) SWNT. Grey, white, and red spheres represent C, Li, and O atoms respectively. First citation in article


Oxygen Reduction by Lithium on Model Carbon and Oxidized Carbon Structures - 伯虎 - 锂空气电池文献 Full figure (80 kB)

Fig. 5. Minimum-energy geometries in top (upper panels) and side (lower panels) views for Li-O species at the armchair edge: (a) Li; (b) epoxy (2C:1O); (c) quinone (2C:2O); (d) anhydride (2C:3O); (e) carbonate (2C:4O); (f) lactone (2C:5O); (g) 2C:4O:4Li; (h) LiO2@2C:4O; (i) LiO2@2C:3O; (j) lone carbonyl (1C:1O); (k) 1C:1O:1Li. Grey, white, red, and small white spheres represent C, Li, O, and H atoms respectively. First citation in article


Oxygen Reduction by Lithium on Model Carbon and Oxidized Carbon Structures - 伯虎 - 锂空气电池文献 Full figure (43 kB)

Fig. 6. Carbon-oxygen phase diagrams for (a) the armchair edge of a GNR; (b) a di-vacancy in graphene. The scales at the top indicate pO2 at given temperatures (in 10x bar) that corresponds to the chemical potential of O2 (?O2). The free energy contributions for the surface C-O groups are calculated at 298.15 K, so the pressure correspondence is approximate for the other temperatures. First citation in article


Oxygen Reduction by Lithium on Model Carbon and Oxidized Carbon Structures - 伯虎 - 锂空气电池文献 Full figure (55 kB)

Fig. 7. The potential-dependent lithium-oxygen surface phase diagram for the armchair edge and di-vacancy. Zero energy (x-axis) corresponds to clean carbon structures, one O2 molecule at the gas-phase conditions of T = 298 K and pO2 = 0.1 bar, and four pairs of Li+/e; thus DeltaG=GA*-GO2(g)-4GLi. Lines are labeled by the respective phases. Solid lines are phases at the armchair edge; long dashed lines with circle symbols are phases at the di-vacancy; and dotted lines are bulk phases (“s”). First citation in article


Oxygen Reduction by Lithium on Model Carbon and Oxidized Carbon Structures - 伯虎 - 锂空气电池文献 Full figure (60 kB)

Fig. 8. Minimum-energy geometries in top (upper panels) and side (lower panels) views for Li-O species at the di-vacancy: (a) Li; (b) ether (1V:1O); (c) lactone (1V:3O); (d) 1V:3O:3Li; (e) LiO2@1V:3O. Grey, white, and red spheres represent C, Li, and O atoms respectively. First citation in article


TABLES

Table I. Minimum-energy adsorption configurations and corresponding adsorption energies (DeltaE, in eV per adsorbate) of lithium and oxygen species on graphene, g(0001), and (8,0) SWNT.a
configuration graphene 3-layer g(0001) (8,0) SWNT
Li (adsorbed)b C6 ring +0.82 +0.66 +0.32
Li (intercalated)b C6 ring - ?0.08c -
LiO2 Li-C6 ring ?0.30 ?0.29 ?0.46
LiO2Li Li-C2 bridge ?0.34 ?0.35 ?0.55
Od epoxy ?1.89 ?1.90 ?3.24
LiO O-C top ?0.98 ?0.97 ?1.61
Li2O O-C top ?0.28 ?0.26 ?1.11
(Li2O)<sub>2e cluster - ?1.50 -
Li2O<sub>(s)e bulk - ?4.22 -
aCoverage is one adsorbate per unit cell unless otherwise indicated; see Fig. 2 for the configurations. C6 ring and C top sites are labeled in Fig. 1. DeltaE include ZPE corrections. ZPE corrections, free energycorrections, principal vibrational modes, Bader charges, and magnetic moments are listed in Table S1 in Supplemental Information. “-” indicates calculation not done.
bWith respect to bulk bcc Li metal.
cThree Li per (3 × 3) unit cell for LiC6.
dWith respect to gas-phase O atom.
eDeltaE is per Li2O unit.
First citation in article

Table II. Minimum-energy lithiated and oxidized carbon structures and corresponding average adsorption energies (DeltaE, in eV per Li or O) at the armchair edge and the di-vacancy.a
configuration DeltaE
armchair edge
Lib C-C chelated ?1.03
2C:1Oc epoxy ?3.92
2C:2O quinone ?5.62
2C:3O anhydride ?5.61
2C:4O carbonate ?5.33
2C:5O lactone ?5.02
di-vacancy
Lib C-C chelated ?0.53
1V:1O ether ?6.58
1V:3O lactone ?4.63
aSee Fig. 5 for the structures. DeltaE include ZPE corrections. ZPE corrections, free energy corrections, principal vibrational modes, and Bader charges are listed in Table S2 in Supplemental Information. There is no magnetic moment on any of these species.
bWith respect to bulk bcc Li metal.
cAll O adsorption energies with respect to gas-phase O atom.
First citation in article

Table III. Reaction free energies (DeltaG, in eV) for the decomposition of the various COx at the armchair edge.a
DeltaG
2C:2O + C--> CO2(g) +0.71
2C:3O + 3/2·C--> 3/2·CO2(g) +1.01
2C:4O + 2·C--> 2·CO2(g) +0.24
2C:5O + 5/2·C--> 5/2·CO2(g) ?1.25
aEnergy of C is that of a C atom in an infinite graphene sheet. pCO2 is taken to be 0.001 bar, which is the typical concentration of CO2 in the atmosphere. T = 298.15 K.
First citation in article

Table IV. Adsorption energies of LiO2 (DeltaE, in eV per LiO2) at various sites.a
site DeltaE
clean g(0001) ?0.29
g(0001), epoxy O ?0.93
armchair edge, 2C:4O ?1.11
armchair edge, 2C:3O ?1.50
di-vacancy, 1V:3O ?1.57
aDeltaE include ZPE corrections. ZPE corrections, free energy corrections, O-O bond lengths and vibrational frequencies, Bader charges, and magnetic moments are listed in Table S3 in Supplemental Information.
First citation in article

Table V. Summary of the reversible potentials (in V) for O2 reduction and COx redox reactions on graphitic carbon.a
g(0001) SWNT armchair edge di-vacancy
O2 reduction 1.12 1.24 - -
COx redox 1.19 - 1.32; 1.44 1.23
COx-catalyzed O2 reduction 1.77 - 1.86; 2.25 2.33
a“-” indicates calculation not done.
First citation in article


FOOTNOTES

*Electrochemical Society Active Member.

?Present address: Environmental Molecular Sciences Laboratory, Pacific Northwest National Laboratory, Richland, WA 95352, USA.

zE-mail: xuy2@ornl.gov


 

Up: Issue Table of Contents
Go to: Previous Article | Next Article
Other formats: HTML (smaller files) | PDF (2547 kB)

  评论这张
 
阅读(1043)| 评论(0)
推荐 转载

历史上的今天

评论

<#--最新日志,群博日志--> <#--推荐日志--> <#--引用记录--> <#--博主推荐--> <#--随机阅读--> <#--首页推荐--> <#--历史上的今天--> <#--被推荐日志--> <#--上一篇,下一篇--> <#-- 热度 --> <#-- 网易新闻广告 --> <#--右边模块结构--> <#--评论模块结构--> <#--引用模块结构--> <#--博主发起的投票-->
 
 
 
 
 
 
 
 
 
 
 
 
 
 

页脚

网易公司版权所有 ©1997-2017