Electrolyte Interface in the Li–O2 Battery: Insight from

Jun 12, 2017 - Electric potential profiles, solvent structuring near the electrode surface, and salt ion distributions are presented and discussed her...
0 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/JPCC

Electrode/Electrolyte Interface in the Li−O2 Battery: Insight from Molecular Dynamics Study Artem V. Sergeev,† Alexander V. Chertovich,† Daniil M. Itkis,*,‡ Anik Sen,§ Axel Gross,§ and Alexei R. Khokhlov† †

Physics Department, Lomonosov Moscow State University, Moscow 119991, Russia Chemistry Department, Lomonosov Moscow State University, Moscow 119991, Russia § Institute of Theoretical Chemistry, Ulm University, Ulm 89081, Germany ‡

ABSTRACT: In this paper, for the first time, we report the results of molecular dynamics simulation of the electrode/electrolyte interface of a Li−O2 cathode under potentials close to experimental values in 1 M dimethyl sulfoxide (DMSO) solution of LiPF6 salt. Electric potential profiles, solvent structuring near the electrode surface, and salt ion distributions are presented and discussed here as well as potentials of mean force (PMFs) of oxygen and its reduction products. The latter would be of a great use for future theoretical studies of reaction kinetics as PMF is essentially the work term required for reaction rate constant estimations. At the electrode/ electrolyte interface under realistic potentials, oxygen anions are effectively pushed out of the reaction layer, making the second reduction of superoxide anion hardly probable. This indicates that the main cause of the electrode surface passivation should be lithium superoxide presence near the electrode surface. The way to suppress the passivation is to shift the equilibrium Ȯ −2 + Li+ ⇌ LiO2 to the side of separately solvated ions, for example, by using solvents resulting in lower free energy of the ions. This conclusion is in agreement with the hypothesis stating that high donor number solvents lead to dominant solution Li2O2 growth and significantly higher cell discharge capacities.

■

INTRODUCTION Redox couples involving alkali metals with one of the strongest and at the same time lightest oxidizersmolecular oxygen promise enormous benefits for energy storage and conversion. Li−O2 cells,1 being an implementation of this idea, might become the next generation in battery technology. At the moment, however, there are a number of issues preventing development of practical Li−O2 batteries. Cells show poor rate capability and quite low energy efficiency and cycle life, and their capacities often lie far below the theoretical expectations. One of key reasons for that is the passivation2,3 of the electrode surface by both the discharge product Li2O2, which is known to be an insulator,4,5 and byproducts formed due to side reactivity of either electrolyte solvents6 or electrode materials.7−9 Overcoming these problems requires an in-depth understanding of the processes occurring inside of the cell. Thus, intensive experimental efforts are made in this field.10−17 Two mechanisms of Li2O2 formation in the course of the oxygen reduction reaction (ORR) are proposed in the literature and summarized in Figure 1. The first one suggests oxygen reduction to Ȯ 2−,18,19 further association with the Li+-ion,20,21 and subsequent disproportionation of lithium superoxide yielding lithium peroxide.17,22 This pathway mostly leads to Li2O2 generation in bulk solution and growth of large micronsized particles.17,20,23,24 Another mechanism leading to lithium peroxide is the second electron transfer and reduction of superoxide species (Ȯ 2− to O22−). It results in product © 2017 American Chemical Society

Figure 1. Schematic representation of the possible reaction steps.

segregation on the surface and unavoidable passivation of the electrode,3,4,20,25 which in turn severely limits the cell capacity. Indeed, both pathways can compete, while the predominant one is determined by the properties of the electrolyte solution20,26 and currents/overvoltages.20,26,27 The rates of the mentioned heterogeneous reaction steps indeed depend on the structure of the electrode/solution interface, particularly on the concentration profiles of solvated Received: April 24, 2017 Revised: June 9, 2017 Published: June 12, 2017 14463

DOI: 10.1021/acs.jpcc.7b03861 J. Phys. Chem. C 2017, 121, 14463−14469

Article

The Journal of Physical Chemistry C species and the electric potential in close vicinity of the electrode surface.28 Unfortunately, experimental investigation of the electrochemical interface structure in such complicated systems turns out to be quite hard due to numerous side processes. Modeling, on the other hand, enables a controllable and independent variation of the parameters of interest, allowing one to capture general trends. Atomistic computer simulations have recently proved to be a convenient tool for investigation of an electrode/solution interfaces.29,30 The scale of interest is about several cubic nanometers of liquid phase as it was shown that species concentration oscillations near the surface span across up to 10 Å or even more.31,32 Due to high computational costs, the simulation of electrochemical interfaces is a challenging task for ab initio or density functional theory (DFT) techniques, even though some promising approaches were proposed.33 Classical molecular dynamics (MD) lack quantum phenomena such as electronic polarization of molecules, specific adsorption, bond formation or breaking, and electron transfer. On the other hand, a full atomic MD simulation is able to predict the double-layer structure at zero current condition as it mostly originates from Coulomb interactions and steric constraints. Specific adsorption of species on the electrode surface (e.g., caused by π−π interaction) if present can be accounted for by introducing an additional interaction potential based on higher-level calculations. Here, we report the results of MD simulations of an electrified interface of an air cathode with a Li+-containing dimethyl sulfoxide (DMSO)-based electrolyte. We use the potential of zero charge (p.z.c.) as a reference value to reproduce a reasonable electrode potential close to the reported experimental value. Electric potential profiles, solvent structuring near the electrode surface, and Li+/PF6− distributions are calculated and discussed together with potentials of mean force (PMFs) of oxygen and reduced oxygen species. We show that the negative surface charge of the electrode under the operation potential is high enough to effectively push oxygen anions out of the reaction layer, which makes the second electron transfer to a superoxide anion hardly possible in DMSO, in agreement with experimental observations.21 We believe that the obtained data will be useful for further theoretical studies of ORR kinetics as PMF can essentially represent the work terms required for the estimation of reaction rate constants.28,34

Figure 2. Simulation cell snapshot. Solvent molecules are in gray. The width of the region filled with electrolyte is 80 Å. The charge density on the electrode is σ = 9.7 μC/cm2.

correction (particle−particle−particle-mesh (pppm)40) was used to calculate Columbic interactions beyond a 12 Å cutoff. van der Waals interactions were also calculated with a 12 Å cutoff. All productive runs were carried out at 298 K in the canonical (NVT) ensemble using the Nosé−Hoover thermostat. The solution consisted of 512 DMSO molecules with 40 Li+ and 40 PF6− ions that correspond to about a 1 M salt concertation. A flexible all-atom model of DMSO developed by Strader and Feller41 was adopted. Force field parameters for Li+ and PF6− ions designed for organic solution were taken from recent work by Kumar and Seminario.42 Lorentz−Berthelot mixing rules were applied. Although carbon materials are most often used in Li−O2 cells, we here consider a noble metal electrode, namely, gold, as it is often used in studies focused on ORR mechanisms.20,43,44 To describe the electrode, we chose the GolP model45,46 of the Au(111) surface that efficiently takes into account polarization (image charge) effects. Gold plates in the simulations were composed of four atomic layers. In the case of nonzero electrode charge, a relatively small charge was added to every surface gold atom of one electrode and balanced by an opposite charge at the other electrode. PMF calculations were carried out for the following species: O2, Ȯ 2−, LiO2, and Li+. The averaged force acting on the species of interest was calculated at a set of distances from the electrode surface (the plane containing the centers of gold atoms composing the outer layer). For each distance, the molecule was bound by a harmonic potential (k = 1500 kcal/ mol/Å2) to a plane parallel to the electrode surface and located at the distance of interest. On average, the net force exerted by the surroundings should be balanced by the force of the harmonic bond. Therefore, the force exerted by the bond was used to calculate the mean force acting on the molecule. For the calculation, at each distance, the system was equilibrated for 0.5 ns, and then, the force was averaged over a 1 ns simulation. After that, the molecule was dragged to another distance. The planes were located in the range of distances from 3.2 to 20 Å from the surface. The spatial step was 0.5 Å for distances greater than 8 and 0.2 Å for smaller distances. At least two runs were performed: forward (moving the molecule toward the surface) and backward. The averaged force data were than interpolated and integrated to obtain the PMF. The equilibrium O−O bond length of the O2 molecule was set to 1.21 Å in accordance with experimental data.47 Although the experimental bond length for Ȯ 2− is estimated to be 1.33 Å,48 we used the same value as that for O2 as it did not affect the calculations much. The stretching force constant set to 1694 kcal/mol/Å2 was derived from experimental vibrational

■

COMPUTATIONAL DETAILS All MD simulations were performed using the large-scale atomic/molecular massively parallel simulator (LAMMPS) package.35 The VMD software36 was used for visualization and postprocessing of simulation data including the calculation of electric potential profiles by means of PME electrostatic extension.37 The simulation cell was designed as a parallel-plate capacitor (Figure 2) filled with electrolyte solution. Plates were perpendicular to the z axis and their x/y dimensions were fixed to 29.30 × 30.45 Å. The distance between plates was 80 Å. The simulation cell volume was chosen to reproduce the density obtained from bulk solution simulation (1.176 ± 0.006 g/cm3) in the 20 Å thick middle region of the cell. Periodic boundary conditions were applied; unwanted interactions between the replicas in the z direction were eliminated by using corresponding routines within the LAMMPS package.38,39 For that reason, a 220 Å vacuum region was added and the electrostatic interaction between the replicas was then estimated and subtracted. Long-range 14464

DOI: 10.1021/acs.jpcc.7b03861 J. Phys. Chem. C 2017, 121, 14463−14469

Article

The Journal of Physical Chemistry C frequencies.49 Lennard-Jones (LJ) parameters for the oxygen atom in O2, Ȯ 2−, and LiO2 have been taken to be the same as those for the CHARMM50 amide oxygen. The lithium LJ parameters for the LiO2 molecule were the same as that for the Li+ ion. The geometry and partial charges of the LiO2 associate were obtained with the help of DFT calculations using Gaussian 09 software.51 The geometry optimization was carried out using the B3LYP functional and 6-31++G** basis set followed by single-point and frequency calculations with the aug-cc-pVTZ basis set. Partial atomic charges were fit to the electrostatic potential. Li−O bond constants were derived from a frequency analysis. Parameters were calculated for both LiO2 isolated in vacuum and for LiO2 in solvent. The IEFPCM scheme was used in the latter case.

■

RESULTS It is well-known that the electrode potential strongly influences the structure of the electrode/solution interface.31,52 According to experimental data,53 the p.z.c. of a gold electrode in DMSO is +0.39 V vs SHE (+3.43 V vs Li+/Li), while the standard potential of O2/Ȯ 2− is reported to be −0.78 V vs SCE54 (+2.50 V vs Li+/Li), and the common discharge potential range of Li− O2 cells is 2.4−2.8 V vs Li+/Li. As the potential of interest (“operation potential” here) is about 1 V below the p.z.c., it would be unjustified to neglect the potential aspect in the simulations. Therefore, we ran a series of simulations with varying charge of the electrode plates and calculated electric potential profiles across the cell (Figure 3). A negative charge of up to −0.05 e0

Figure 4. Orientational order parameters (red and blue solid lines) and normalized solvent concentration (black dashed) profiles near the neutral (A) and negatively charged (B) surfaces. (C) Explanatory sketches illustrating the geometry of the DMSO molecule and its connection with orientational order parameters.

(calculated using mass center coordinates) normalized to the bulk value is depicted by the dashed black line. The oscillation of the solvent concentration near the surface is a common behavior for a solid/liquid interface,31,32 and there is no significant difference between charged and uncharged cases. Orientational order parameters P1 = ⟨cos(θ)⟩ and P2 = ⟨(3/ 2) cos2(θ) − (1/2)⟩, where θ is the angle between the z direction and normal to the plane containing S and C atoms of the DMSO molecule, are also plotted in Figure 4. In the case of negative surface charge (Figure 4b), one can observe a prominent orientation of solvent molecules (⟨θ⟩ → 0) in such a way that S atoms and CH3 groups lie on the electrode surface while O atoms point out toward the solution (Figure 4c). The orientation is much weaker near the noncharged surface (Figure 4a). Thus, solvent ordering in the first molecular layer at least partially originates from Coulomb interaction of the charged surface with the positively charged S atoms and CH3 groups and negatively charged O atoms. This effect can noticeably influence electrochemical reaction rates as oriented solvent molecules can lose the ability to effectively solvate ions.

Figure 3. Electric potential profiles near the left electrode at different negative surface charge values. Atomic planes of the gold electrode are indicated by dash−dot lines.

was added to every surface gold atom of the left (lower z) electrode. The electric potential converged to its bulk value within 3 molecular layers (∼15 Å). A surface charge of −σ = −0.045 e0/Au = −9.7 μC/cm2 yielded an electrode potential about 1 V below the potential of bulk solution that corresponds to the difference between the p.z.c. and the experimental ORR potential. To evaluate the influence of the electrode charge, we have compared the solvent structuring (Figure 4) and salt ion distributions (Figure 5) in the cases of zero and relevant negative surface charges. In Figure 4, the solvent concentration 14465

DOI: 10.1021/acs.jpcc.7b03861 J. Phys. Chem. C 2017, 121, 14463−14469

Article

The Journal of Physical Chemistry C

O2 cathode and the DMSO-based solution. Besides the expected attraction of salt cations and repulsion of anions, a complete exclusion of both cations and anions from the first molecular layer of DMSO occurs. It is accompanied by a prominent reorientation of DMSO molecules in the first layer. We’d like to note that the positions of the ion concentration peaks were practically the same at a surface charge value as small as −4.3 μC/cm2, or −0.02 e0/Au atom, though the peak amplitudes were smaller. Thus, some possible error of estimated surface charge that is supposed to be relevant to the electrode operation potential is not crucial for the conclusions drawn based on the data calculated at −σ = −0.045 e0/Au = −9.7 μC/cm2. To calculate the PMF, we first performed DFT calculations and determined partial charges on the lithium superoxide. As seen in Table 1, the partial charges strongly depend on the Table 1. Sets of LiO2 Molecule Parameters Derived from DFT Calculations Carried out in Vacuum (LiO2_vac) and Employing an Implicit Solvent Model (LiO2_DMSO) O−O length, Å O−O kb, kcal/mol/Å2 Li−O length, Å Li−O kb, kcal/mol/Å2 O charge, e0 Li charge, e0

Figure 5. Salt ion concentration profiles near the neutral (A) and negatively charged (B) surfaces.

Note that a strong ordering of the first solvent layer that we found was demonstrated for an ideally flat electrode surface. It should still persist for the surfaces with curvature radius much greater than the molecular layer thickness (∼5 Å). Additional studies should be, however, conducted to check the influence of surface defects and angstrom-scale roughness. Some corresponding attempts have been made recently,55 but the electrode potential was not taken into account in these studies. One should also keep in mind that the adsorption of DMSO molecules on the electrode surface can be influenced by quantum effects that are not considered in MD simulations. Therefore, further periodic DFT calculations of the DMSO layer on the electrode surface under the potential are also required. The ion distributions are compared in Figure 5. There are only small oscillations of ion concentrations near the neutral surface. Note that there is a substantial concentration of salt ions inside of the first molecular layer at distances closer than 5 Å to the surface (a ∼3 Å wide layer is empty due to the Pauli repulsion included in the LJ potential). In the case of negative surface charge, a pronounced peak of Li+ concentration can be observed near the electrode. It is located at z = 6 Å, just between the first and second molecular layers of the solvent (see DMSO concentration peaks in Figure 4) rather than next to the negatively charged surface, as one could expect relying on Coulomb attraction. The effect of exclusion of Li+ ions from the first molecular layer was as well observed in the simulation results at a smaller surface charge value (particularly −4.3 μC/ cm2 or −0.02 e0/Au atom). The PF6− anions are almost completely pushed out beyond the second molecular layer of the solvent. Thus, our calculations support the notion that the negative surface charge at the electrode operation potential causes significant structural changes (in comparison to zero surface charge) of the electrode/electrolyte interface between the Li/

LiO2_vac

LiO2_DMSO

1.35 1810 1.77 175 −0.365 +0.73

1.35 1600 1.91 82 −0.45 +0.90

dielectric permittivity of the medium. Calculations for vacuum showed that associated Ȯ 2− and Li+ ions partially share their charge, significantly reducing the overall dipole moment. When the surrounding solvent was taken into account implicitly by means of the polarizable continuum model (PCM), the ions retained most of their initial charge. PCM is a good approximation for solvation in bulk liquid, but it should be noted that dielectric properties of the solvent change at the solid/liquid interface.56 It is reasonable to assume that the lack of orientational freedom of the DMSO molecules in the first layer weakens the ability to screen ion charges, that is, it lowers the local dielectric permittivity. Therefore, we performed PMF calculations using two sets of LiO2 parameters based on vacuum (LiO2_vac) and implicit solvent (LiO2_DMSO) DFT calculations, expecting the true values to lie in between these limiting cases. The calculated PMF profiles are presented in Figure 6. The neutral O2 molecule can reach the electrode surface by overcoming a free energy barrier of about 2.5kbT. The PMFs of Ȯ 2− and Li+ exhibit a steep rise starting from somewhere in between the first and second molecular layers (z = 6 Å). This confirms that the intercalation of ions in the first DMSO layer is improbable. The same is true for the LiO2 bound ion pair described by the LiO2_DMSO set of parameters (see Table 1). Using parameters calculated for vacuum, (LiO2_vac) results in a weaker repulsion from the surface in comparison to LiO2_DMSO due to reduced atomic charges. The PMF of Ȯ 2− near the neutral surface (dashed blue line) does not rise significantly. Thus, ion exclusion from the first molecular layer is caused by the negative charge of the electrode surface. As can be seen from the PMF profiles, the free energy of both the superoxide anion and lithium superoxide in the first molecular layer is extremely high; therefore, we can conclude 14466

DOI: 10.1021/acs.jpcc.7b03861 J. Phys. Chem. C 2017, 121, 14463−14469

Article

The Journal of Physical Chemistry C

represents species concentration profiles normalized to the bulk value. Substantial peaks of lithium superoxide (for both LiO2_vac and LiO2_DMSO parameter sets) are located at around z = 7 Å, while the Ȯ 2− concentration is negligible at distances smaller than 10 Å. The electron transfer probability decays exponentially with distance κet(z) = κ0 exp(−z/λet); the characteristic length λet is about 1 Å.28,57 The presence of a lithium superoxide concentration peak at 3 Å closer to the electrode (7 Å from the surface) indicates that LiO2 reduction is kinetically favorable in comparison with the second reduction of the unassociated superoxide anion (Ȯ −2 + e−⇌ O2− 2 ). The free energy change of the reduction of the neutral LiO2 associate should be lower than that of Ȯ 2− to O22− reduction, which is supported by the experimentally observed shift of the second reduction peak upon addition of lithium salt.20 Thus, both kinetically and thermodynamically, the second electron transfer to the superoxide anion is less probable than the reduction of superoxide after its association with the lithium cation, in line with experimental reports.21

■

DISCUSSION As the passivation of the electrode surface by lithium peroxide can quickly shut down the Li−O2 cell, we are here discussing ways to prevent the fast passivation of the electrode surface by Li2O2 in light of the MD simulation results reported above. The high rate of Li+ and Ȯ 2− association leads to an increased concentration of LiO2 near the electrode surface. This subsequently not only promotes faster disproportionation to lithium peroxide in the vicinity of electrode but also increases the rate of the second electron transfer, yielding again Li2O2, which passivates the surface. Slowing the association rate will shift the LiO2 concentration peak toward the bulk solution, thus diminishing the passivation rate. While it was already experimentally demonstrated that solvents with higher solvation ability for both cations and anions (i.e., higher donor and acceptor numbers) inhibit Li+ and Ȯ 2− association and thus favor solution-phase lithium peroxide formation,14,20,26 we suggest one additional way to suppress Li2O2 precipitation on the surface. We assume that lithium and superoxide ion association can be suppressed by minimizing the Li+ concentration in the region where Ȯ 2− is generated. Our calculations show that Li+ ions are concentrated right in that region as a response to negative surface charge, which is screened by these cations. We speculate that addition of electrochemically inactive cations to the electrolyte can result in substitution of Li+ in the vicinity of the interface that in turn can potentially suppress second electron transfer and disproportionation as Ȯ 2− will preferably diffuse away and couple with Li+ pushed toward the bulk solution.

Figure 6. PMF of the different species near the negatively charge surface (solid lines) and PMF of the superoxide anion near the neutral surface (dashed line).

that adsorption and consequent disproportionation of these species directly on the surface is doubtful. We further use calculated PMFs to derive concentration profiles for the species of interest. Concentration profiles near the electrode surface were estimated based on the Boltzmann distribution (cX(z)/cX,bulk = exp(−PMFX(z)/kbT)) as direct calculations are impossible due to a very low concentration (1 mM, less than 1 molecule per simulation cell). We validated this approach by comparing the PMF-based Li+ concentration profile with the directly calculated one (Figure 5B, dashed line). A good quantitative agreement was achieved. Figure 7

■

CONCLUSIONS We performed all-atom MD simulations of the electrified interface between a gold electrode and 1 M LiPF6 solution in DMSO in order to derive PMFs for oxygen and reduced oxygen species generated in Li−O2 cells. The electrode potential was simulated using a polarizable model of the Au(111) surface, while the p.z.c. was used as a reference to find the value of the surface charge that corresponds to the cathode operation potential of about +2.4−2.6 V vs Li+/Li. We found that at such potentials, besides the expected attraction of cations and repulsion of anions, complete exclusion of both Li+ and PF6− from the first molecular layer of DMSO occurs. It is

Figure 7. Concentration profiles calculated using the PMF (solid lines). Li+ concentration profile (dashed line) obtained directly from the simulation. Electron transfer probability decays with distance from the electrode. 14467

DOI: 10.1021/acs.jpcc.7b03861 J. Phys. Chem. C 2017, 121, 14463−14469

Article

The Journal of Physical Chemistry C

Conductivity in Li2O2 and Its Role in Determining Capacity Limitations in Non-Aqueous Li-O2 Batteries. J. Chem. Phys. 2011, 135 (21), 214704. (5) Radin, M. D.; Siegel, D. J. Charge Transport in Lithium Peroxide: Relevance for Rechargeable Metal−air Batteries. Energy Environ. Sci. 2013, 6 (8), 2370. (6) Freunberger, S. A.; Chen, Y.; Drewett, N. E.; Hardwick, L. J.; Bardé, F.; Bruce, P. G. The Lithium-Oxygen Battery with Ether-Based Electrolytes. Angew. Chem., Int. Ed. 2011, 50 (37), 8609−8613. (7) Itkis, D. M.; Semenenko, D. A.; Kataev, E. Y.; Belova, A. I.; Neudachina, V. S.; Sirotina, A. P.; Hävecker, M.; Teschner, D.; KnopGericke, A.; Dudin, P.; et al. Reactivity of Carbon in Lithium−Oxygen Battery Positive Electrodes. Nano Lett. 2013, 13 (10), 4697−4701. (8) Kozmenkova, A. Y.; Kataev, E. Y.; Belova, A. I.; Amati, M.; Gregoratti, L.; Velasco-Vélez, J.; Knop-Gericke, A.; Senkovsky, B.; Vyalikh, D. V.; Itkis, D. M.; et al. Tuning Surface Chemistry of TiC Electrodes for Lithium−Air Batteries. Chem. Mater. 2016, 28 (22), 8248−8255. (9) McCloskey, B. D.; Speidel, A.; Scheffler, R.; Miller, D. C.; Viswanathan, V.; Hummelshøj, J. S.; Nørskov, J. K.; Luntz, A. C. Twin Problems of Interfacial Carbonate Formation in Nonaqueous Li-O2 Batteries. J. Phys. Chem. Lett. 2012, 3 (8), 997−1001. (10) Zhai, D.; Lau, K. C.; Wang, H. H.; Wen, J.; Miller, D. J.; Lu, J.; Kang, F.; Li, B.; Yang, W.; Gao, J.; et al. Interfacial Effects on Lithium Superoxide Disproportionation in Li-O2 Batteries. Nano Lett. 2015, 15 (2), 1041−1046. (11) Lu, Y.-C.; Gasteiger, H. a.; Crumlin, E.; McGuire, R.; ShaoHorn, Y. Electrocatalytic Activity Studies of Select Metal Surfaces and Implications in Li-Air Batteries. J. Electrochem. Soc. 2010, 157 (9), A1016. (12) Yang, J.; Zhai, D.; Wang, H.-H.; Lau, K. C.; Schlueter, J. A.; Du, P.; Myers, D. J.; Sun, Y.-K.; Curtiss, L. A.; Amine, K. Evidence for Lithium Superoxide-like Species in the Discharge Product of a Li−O2 Battery. Phys. Chem. Chem. Phys. 2013, 15 (11), 3764. (13) Abraham, K. M. Electrolyte-Directed Reactions of the Oxygen Electrode in Lithium-Air Batteries. J. Electrochem. Soc. 2015, 162 (2), A3021−A3031. (14) Laoire, C. O.; Mukerjee, S.; Abraham, K. M.; Plichta, E. J.; Hendrickson, M. A. Influence of Nonaqueous Solvents on the Electrochemistry of Oxygen in the Rechargeable Lithium−Air Battery. J. Phys. Chem. C 2010, 114 (19), 9178−9186. (15) Zhai, D.; Wang, H.-H.; Lau, K. C.; Gao, J.; Redfern, P. C.; Kang, F.; Li, B.; Indacochea, E.; Das, U.; Sun, H.-H.; et al. Raman Evidence for Late Stage Disproportionation in a Li−O2 Battery. J. Phys. Chem. Lett. 2014, 5 (15), 2705−2710. (16) Black, R.; Oh, S. H.; Lee, J.-H.; Yim, T.; Adams, B.; Nazar, L. F. Screening for Superoxide Reactivity in Li-O2 Batteries: Effect on Li2O2 /LiOH Crystallization. J. Am. Chem. Soc. 2012, 134 (6), 2902−2905. (17) Zhai, D.; Wang, H.-H.; Yang, J.; Lau, K. C.; Li, K.; Amine, K.; Curtiss, L. A. Disproportionation in Li−O2 Batteries Based on a Large Surface Area Carbon Cathode. J. Am. Chem. Soc. 2013, 135 (41), 15364−15372. (18) Sawyer, D. T.; Gibian, M. J. The Chemistry of Superoxide Ion. Tetrahedron 1979, 35 (12), 1471−1481. (19) Herranz, J.; Garsuch, A.; Gasteiger, H. A. Using Rotating Ring Disc Electrode Voltammetry to Quantify the Superoxide Radical Stability of Aprotic Li−Air Battery Electrolytes. J. Phys. Chem. C 2012, 116 (36), 19084−19094. (20) Johnson, L.; Li, C.; Liu, Z.; Chen, Y.; Freunberger, S. a; Ashok, P. C.; Praveen, B. B.; Dholakia, K.; Tarascon, J.-M.; Bruce, P. G. The Role of LiO2 Solubility in O2 Reduction in Aprotic Solvents and Its Consequences for Li-O2 Batteries. Nat. Chem. 2014, 6 (12), 1091− 1099. (21) Belova, A. I.; Kwabi, D. G.; Yashina, L. V.; Shao-Horn, Y.; Itkis, D. M. On the Mechanism of Oxygen Reduction in Aprotic Li-Air Batteries: The Role of Carbon Electrode Surface Structure. J. Phys. Chem. C 2017, 121, 1569. (22) Zakharchenko, T. K.; Kozmenkova, A. Y.; Itkis, D. M.; Goodilin, E. A. Lithium Peroxide Crystal Clusters as a Natural Growth Feature

accompanied by prominent reorientation of the DMSO molecules in the first layer. Highly polar LiO2 is also pushed away from the surface. Hence, we speculate that adsorption and further disproportionation of superoxide anions and lithium superoxide directly on the surface is unlikely. The value of the negative surface charge at the electrode operation potential turned out to be enough to push Ȯ 2− beyond the 10 Å range from the electrode surface, which suppresses electron transfer from the electrode. On the other hand, the lithium superoxide concentration peak found at a distance of 7 Å from the electrode enables effective electron transfer. These findings support that the second electron transfer is possible only after the association of Ȯ 2− and Li+. On the basis of our findings, we suppose that not only high donor and acceptor number solvents, which solvate both Li+ and Ȯ 2− and thus inhibit its association in the vicinity of the interface, can suppress electrode passivation and favor solutionphase Li2O2 formation. We propose that addition of supporting electrolyte with inactive cations can result in the substitution of Li+ near the interface, pushing its association reaction with O2− away from the electrode and thus minimizing Li2O2 generation close to the electrode surface. Although Li+ can be hardly displaced by other metal cations due to its high chemical hardness in terms of hard−soft acid and base theory, some other organic cations possessing strong specific interactions with the electrode surface can potentially replace lithium ions in the vicinity of the interface.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Artem V. Sergeev: 0000-0001-6572-0197 Daniil M. Itkis: 0000-0002-6363-6669 Axel Gross: 0000-0003-4037-7331 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS All simulations were performed on the cluster of the Supercomputing Center of Lomonosov Moscow State University.58 Financial support was provided by the Centre for Electrochemical Energy of Skolkovo Institute of Science and Technology and Russian Foundation for Basic Research (Grant # 14- 29-04101). This work was in part supported by RFBR (Project No. 17-53-12056). The authors gratefully acknowledge Dr. Sergey Kislenko for his advice on molecular dynamics simulations.

■

REFERENCES

(1) Aurbach, D.; McCloskey, B. D.; Nazar, L. F.; Bruce, P. G. Advances in Understanding Mechanisms Underpinning Lithium−air Batteries. Nat. Energy 2016, 1 (9), 16128. (2) Albertus, P.; Girishkumar, G.; McCloskey, B.; Sánchez-Carrera, R. S.; Kozinsky, B.; Christensen, J.; Luntz, A. C. Identifying Capacity Limitations in the Li/Oxygen Battery Using Experiments and Modeling. J. Electrochem. Soc. 2011, 158, A343. (3) McCloskey, B. D.; Scheffler, R.; Speidel, A.; Girishkumar, G.; Luntz, A. C. On the Mechanism of Nonaqueous Li-O2 Electrochemistry on C and Its Kinetic Overpotentials: Some Implications for Li-Air Batteries. J. Phys. Chem. C 2012, 116 (45), 23897−23905. (4) Viswanathan, V.; Thygesen, K. S.; Hummelshøj, J. S.; Nørskov, J. K.; Girishkumar, G.; McCloskey, B. D.; Luntz, A. C. Electrical 14468

DOI: 10.1021/acs.jpcc.7b03861 J. Phys. Chem. C 2017, 121, 14463−14469

Article

The Journal of Physical Chemistry C of Discharge Products in Li−O2 Cells. Beilstein J. Nanotechnol. 2013, 4, 758−762. (23) Mitchell, R. R.; Gallant, B. M.; Thompson, C. V.; Shao-Horn, Y. All-Carbon-Nanofiber Electrodes for High-Energy Rechargeable Li− O2 Batteries. Energy Environ. Sci. 2011, 4 (8), 2952. (24) Mitchell, R. R.; Gallant, B. M.; Shao-Horn, Y.; Thompson, C. V. Mechanisms of Morphological Evolution of Li2O2 Particles during Electrochemical Growth. J. Phys. Chem. Lett. 2013, 4 (7), 1060−1064. (25) Wen, R.; Hong, M.; Byon, H. R. In Situ AFM Imaging of Li−O2 Electrochemical Reaction on Highly Oriented Pyrolytic Graphite with Ether-Based Electrolyte. J. Am. Chem. Soc. 2013, 135 (29), 10870− 10876. (26) Kwabi, D. G.; Bryantsev, V. S.; Batcho, T. P.; Itkis, D. M.; Thompson, C. V.; Shao-Horn, Y. Experimental and Computational Analysis of the Solvent-Dependent O2 /Li+-O2− Redox Couple: Standard Potentials, Coupling Strength, and Implications for LithiumOxygen Batteries. Angew. Chem., Int. Ed. 2016, 55 (9), 3129−3134. (27) Adams, B. D.; Radtke, C.; Black, R.; Trudeau, M. L.; Zaghib, K.; Nazar, L. F. Current Density Dependence of Peroxide Formation in the Li−O2 Battery and Its Effect on Charge. Energy Environ. Sci. 2013, 6, 1772. (28) Nikitina, V. A.; Kislenko, S. A.; Nazmutdinov, R. R.; Bronshtein, M. D.; Tsirlina, G. A.; Nikitina, V. A.; Kislenko, S. A.; Nazmutdinov, R. R.; Bronshtein, M. D.; Tsirlina, G. A.; et al. Ferrocene/Ferrocenium Redox Couple at Au(111)/Ionic Liquid and Au(111)/Acetonitrile Interfaces: A Molecular-Level View at the Elementary Act. J. Phys. Chem. C 2014, 118 (12), 6151−6164. (29) Burt, R.; Birkett, G.; Zhao, X. S. A Review of Molecular Modelling of Electric Double Layer Capacitors. Phys. Chem. Chem. Phys. 2014, 16 (14), 6519. (30) Hughes, Z.; Walsh, T. R. Computational Chemistry for Graphene-Based Energy Applications: Progress and Challenges. Nanoscale 2015, 7, 6883. (31) Kislenko, S. A.; Amirov, R. H.; Samoylov, I. S. Molecular Dynamics Simulation of the Electrical Double Layer in Ionic Liquids. J. Phys.: Conf. Ser. 2013, 418, 012021. (32) Xing, L.; Vatamanu, J.; Borodin, O.; Smith, G. D.; Bedrov, D. Electrode/Electrolyte Interface in Sulfolane-Based Electrolytes for Li Ion Batteries: A Molecular Dynamics Simulation Study. J. Phys. Chem. C 2012, 116 (45), 23871−23881. (33) Schnur, S.; Groß, A. Challenges in the First-Principles Description of Reactions in Electrocatalysis. Catal. Today 2011, 165 (1), 129−137. (34) Nazmutdinov, R. R.; Bronshtein, M. D.; Zinkicheva, T. T.; Glukhov, D. V. Modeling of Electron Transfer across Electrochemical Interfaces: State-of-the Art and Challenges for Quantum and Computational Chemistry. Int. J. Quantum Chem. 2016, 116, 189−201. (35) Plimpton, S. Fast Parallel Algorithms for Short-Range Molecular Dynamics. J. Comput. Phys. 1995, 117 (1), 1−19. (36) Humphrey, W.; Dalke, A.; Schulten, K. VMD: Visual Molecular Dynamics. J. Mol. Graphics 1996, 14 (1), 33−38. (37) Aksimentiev, A.; Schulten, K. Imaging α-Hemolysin with Molecular Dynamics: Ionic Conductance, Osmotic Permeability, and the Electrostatic Potential Map. Biophys. J. 2005, 88 (6), 3745−3761. (38) Yeh, I.-C.; Berkowitz, M. L. Ewald Summation for Systems with Slab Geometry. J. Chem. Phys. 1999, 111 (7), 3155. (39) Ballenegger, V.; Arnold, A.; Cerdá, J. J. Simulations of NonNeutral Slab Systems with Long-Range Electrostatic Interactions in Two-Dimensional Periodic Boundary Conditions. J. Chem. Phys. 2009, 131 (9), 094107. (40) Hockney, R. W.; Eastwood, J. W. Computer Simulation Using Particles; Adam Hilger: New York, 1988. (41) Strader, M. L.; Feller, S. E. A Flexible All-Atom Model of Dimethyl Sulfoxide for Molecular Dynamics Simulations. J. Phys. Chem. A 2002, 106 (6), 1074−1080. (42) Kumar, N.; Seminario, J. M. Lithium-Ion Model Behavior in an Ethylene Carbonate Electrolyte Using Molecular Dynamics. J. Phys. Chem. C 2016, 120 (30), 16322−16332.

(43) Khan, A.; Zhao, C. Oxygen Reduction Reactions in Aprotic Ionic Liquids Based Mixed Electrolytes for High Performance of Li-O2 Batteries. ACS Sustainable Chem. Eng. 2016, 4, 506. (44) Aurbach, D.; Daroux, M.; Faguy, P.; Yeager, E. The Electrochemistry of Noble Metal Electrodes in Aprotic Organic Solvents Containing Lithium Salts. J. Electroanal. Chem. Interfacial Electrochem. 1991, 297 (1), 225−244. (45) Wright, L. B.; Rodger, P. M.; Corni, S.; Walsh, T. R. GolPCHARMM: First-Principles Based Force Fields for the Interaction of Proteins with Au(111) and Au(100). J. Chem. Theory Comput. 2013, 9 (3), 1616−1630. (46) Iori, F.; Corni, S. Including Image Charge Effects in the Molecular Dynamics Simulations of Molecules on Metal Surfaces. J. Comput. Chem. 2008, 29 (10), 1656−1666. (47) Huber, K.-P. Molecular Spectra and Molecular Structure: IV. Constants of Diatomic Molecules; Springer Science & Business Media, 2013. (48) Krupenie, P. H. The Spectrum of Molecular Oxygen. J. Phys. Chem. Ref. Data 1972, 1 (2), 423−534. (49) Irikura, K. K. Experimental Vibrational Zero-Point Energies: Diatomic Molecules. J. Phys. Chem. Ref. Data 2007, 36 (2), 389−397. (50) Vanommeslaeghe, K.; Hatcher, E.; Acharya, C.; Kundu, S.; Zhong, S.; Shim, J.; Darian, E.; Guvench, O.; Lopes, P.; Vorobyov, I.; et al. CHARMM General Force Field: A Force Field for Drug-like Molecules Compatible with the CHARMM All-Atom Additive Biological Force Fields. J. Comput. Chem. 2009, NA−NA. (51) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A.; et al. Gaussian 09, revision E.01; Gaussian Inc.: Wallingford, CT, 2009 (52) Vatamanu, J.; Borodin, O.; Smith, G. D. Molecular Insights into the Potential and Temperature Dependences of the Differential Capacitance of a Room-Temperature Ionic Liquid at Graphite Electrodes. J. Am. Chem. Soc. 2010, 132 (42), 14825−14833. (53) Trasatti, S.; Lust, E. The Potential of Zero Charge. Mod. Asp. Electrochem. 1999, 329 (33), 1−215. (54) Sawyer, D. T.; Chiericato, G.; Angelis, C. T.; Nanni, E. J.; Tsuchiya, T. Effects of Media and Electrode Materials on the Electrochemical Reduction of Dioxygen. Anal. Chem. 1982, 54 (11), 1720−1724. (55) Pavlov, S.; Kislenko, S. Effects of Carbon Surface Topography on the Electrode/Electrolyte Interface Structure and Relevance to LiAir Batteries. Phys. Chem. Chem. Phys. 2016, 18, 30830. (56) Bonthuis, D. J.; Gekle, S.; Netz, R. R. Dielectric Profile of Interfacial Water and Its Effect on Double-Layer Capacitance. Phys. Rev. Lett. 2011, 107 (16), 1−5. (57) Khoshtariya, D. E.; Dolidze, T. D.; van Eldik, R. Multiple Mechanisms for Electron Transfer at Metal/Self-Assembled Monolayer/Room-Temperature Ionic Liquid Junctions: Dynamical Arrest versus Frictional Control and Non-Adiabaticity. Chem. - Eur. J. 2009, 15 (21), 5254−5262. (58) Sadovnichy, V.; Tikhonravov, A.; Voevodin, V.; Opanasenko, V. “Lomonosov”: Supercomputing at Moscow State University. In Contemporary High Performance Computing: From Petascale toward Exascale; Vetter, J. S., Ed.; CRC Press: Boca Raton, FL, 2013; pp 283− 308.

14469

DOI: 10.1021/acs.jpcc.7b03861 J. Phys. Chem. C 2017, 121, 14463−14469