Reactivity of Carbon in Lithium–Oxygen Battery Positive Electrodes

Letter pubs.acs.org/NanoLett

Reactivity of Carbon in Lithium−Oxygen Battery Positive Electrodes Daniil M. Itkis,*,†,‡ Dmitry A. Semenenko,‡ Elmar Yu. Kataev,‡ Alina I. Belova,‡ Vera S. Neudachina,† Anna P. Sirotina,† Michael Hav̈ ecker,§ Detre Teschner,§ Axel Knop-Gericke,§ Pavel Dudin,∥,¶ Alexei Barinov,∥ Eugene A. Goodilin,†,‡ Yang Shao-Horn,⊥ and Lada V. Yashina† †

Department of Chemistry and ‡Department of Materials Science, Moscow State University, Moscow, 119992, Russia Fritz-Haber Institute of the Max Planck Society, Berlin, D-14195, Germany ∥ Sincrotrone Trieste S.C.p.A., Area Science Park, I-34012 Basovizza, Trieste, Italy ⊥ Materials Science and Engineering Department, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, United States §

S Supporting Information *

ABSTRACT: Unfortunately, the practical applications of Li−O2 batteries are impeded by poor rechargeability. Here, for the first time we show that superoxide radicals generated at the cathode during discharge react with carbon that contains activated double bonds or aromatics to form epoxy groups and carbonates, which limits the rechargeability of Li−O2 cells. Carbon materials with a low amount of functional groups and defects demonstrate better stability thus keeping the carbon will-o’-the-wisp lit for lithium−air batteries. KEYWORDS: Lithium−air battery, in situ APXPS, superoxide, side reactions (e.g., superoxide species) or final products attack carbon during battery operation and standby. The detailed understanding of these reaction steps could potentially assist further research on the control over such intermediate formation, migration, and decay. Here we report a first study of the electrochemical and chemical processes on the surface of carbon materials, namely reduced graphene oxide (RGO) and thermally exfoliated graphite in a cell with an all-solid-state positive electrode under true operando conditions.17,18 We find that superoxide anions attack carbon materials and then unexpectedly cause its further oxidation by molecular oxygen, which leads to the gradual formation of carbonate species. Spectroscopic observation of the operating electrochemical cells was used to reveal ORR intermediates and products formed on carbon upon discharge at different oxygen pressures. A thin film of RGO flakes (Supporting Information Figure S1) deposited onto a solid glass-ceramic NASICON-type electrolyte was employed as the working electrode, which was exposed to molecular oxygen in the ambient pressure X-ray photoelectron spectroscopy (XPS) chamber. The reference and counter electrodes were made of metallic lithium foil (see Figure 1a). The use of the solid electrolyte allows us to probe the reactivity between carbon and ORR reaction intermediates and products without the influence of their parasitic reactions with liquid electrolytes.9,13,19

A

dvanced aprotic lithium−air cells1−4 generate lithium peroxide as a final discharge product via oxygen reduction reactions (ORR). The ORR results in extra electrons on π*-antibonding orbitals of the O2 molecule and its stepwise transformation into superoxide and peroxide ions5 as intermediate species, which can be highly reactive and only stable on inorganic surfaces with no reactive centers.6 The Gibbs free energy impels the formation of a final product lattice of Li2O or Li2O24 while bulk LiO2 can not be formed above 15 K7 and should be better considered as an unstable associative complex of Li+, superoxide Ȯ 2− anion radicals, and solvent molecules.8 In the case when the solvent is not enough stable in the presence of superoxide, the latter initiates a chemical transformation as it happens with, for example, alkyl-carbonates,9,10 making them unsuitable for lithium−air batteries. Another target for the superoxide species attack is the carbon electrode surface that potentially allows chemical interactions like radical nucleophilic reactions and oxidation of olefins with activated double bonds.11,12 Unfortunately, there is a lack of direct experimental evidence on the reactivity between carbon and superoxide ions so recent studies conclude that it is the lithium peroxide that reacts with carbon and leads to carbonates as both Li2O2 and Li2CO3 can be detected by electrode postmortem analysis.13,14 This belief forces the consideration of carbon as inapplicable while its unique physical properties (such as high surface area, lightweight and high electronic conductivity) still keep graphene,15 carbon nanotubes,16 and other sp2 carbons highly attractive. Following this carbon’s will-o′-the-wisp, the future development of lithium−air batteries demands the knowledge of whether certain intermediates © 2013 American Chemical Society

Received: June 14, 2013 Revised: August 25, 2013 Published: September 4, 2013 4697

dx.doi.org/10.1021/nl4021649 | Nano Lett. 2013, 13, 4697−4701

Nano Letters

Letter

Figure 1. Electrochemical cell for photoemission spectroscopy under operando conditions. (a) The cell design. XPS observation of working electrode was performed though the opening in the steel plate. The electrode was externally grounded in order to avoid the shifts of the spectral features that can be caused by a potential difference between cathode and the analyzer ground.18 (b) The SEM image revealing the cathode microstructure that comprises a carbon (thermally exfoliated graphite in this image) flakes deposited onto glass-ceramic solid Li+-conducting electrolyte.

that superoxide Ȯ 2− species first formed on carbon upon the discharge can further react to degrade the carbon surface by creating oxygenated surface functional groups and defects and to form a number of reaction products like carbonates/carboxylates, as shown in Figure 3. Here we argue that the direct reaction of lithium peroxide and carbon could not be the main source of the carbonate species found recently on glassy carbon13 and carbon nanotubes.14 Though thermodynamic estimations permit such a scheme,13 almost all observations of lithium−air cell discharge reveal Li2O2 as a product easily coexisting with carbon electrodes.16,26 Using cyclic voltammetry we further found that glassy carbon electrodes do not reveal any evidence of the carbonate deposits on the surface even after a long intentional treatment with Li2O2 suspensions, while even a short contact of glassy carbon with potassium superoxide (KO2) immediately results in carbonate formation (see Supporting Information Figure S5). Recent data27 showing the presence of lithium superoxide thin layer on the surface of electrochemically formed Li2O2 suggest that Li2O2 surface can be also involved in carbon oxidation. We further performed a constant current discharge followed by an open-circuit potential (OCP) period and subsequent recharge by constant current (Figure 4a). Li 1s, O 1s, and C 1s spectra were collected as fast as possible (about 1 min per spectrum) to maintain the energy resolution and an appropriate signal-to-noise ratio. The charge that passed through the cell in this experiment was low enough so that the increase in the intensities of Li 1s and O 1s was correlated linearly with the charge (Figure 4e). With increasing discharge capacity, the epoxy groups at 286.1 eV and carbonates at 290.5 eV were found to grow simultaneously (Figure 4b), in agreement with the scheme in Figure 3. Interestingly, the amount of carbonate species was found to considerably increase at the expense of epoxy groups during the OCP period (Figure 4c), which was coupled with a small increase in the surface oxygen O 1s intensity (see Figure 4f). This observation is in agreement with the hypothesis that CO32− groups are formed from superoxide-activated sp2 lattice by its purely chemical oxidation with molecular oxygen. The lithium carbonate and peroxide produced on carbon in this study could be electrochemically oxidized and removed. Two voltage plateaus were clearly observed on the charge voltage profile in Figure 4a, which can be attributed to Li2O2 and Li2CO3 oxidation, respectively. As soon as anodic current starts to flow,

The cell demonstrated negligible discharge capacity in UHV conditions and inappreciable changes were found for the spectra of Li 1s and C 1s collected from the carbon surface (see Supporting Information Figure S3). We separately investigated time dependencies of the pristine cathode surface composition during oxygen exposure to ensure the reversibility of oxygen adsorption/desorption on/from the cathode material. Weak chemisorption was found to be a predominant process showing no irreversible changes in carbon material structure. To investigate the reactivity of ORR reaction intermediates and products with RGO, we discharged the cell under constant current in 0.1 mbar of oxygen partial pressure within the XPS chamber. Discharge was repeated after a period of open circuit (Figure 2a,b). The intensities of Li 1s and O 1s peaks (Supporting Information Figure S4) increased with increasing discharge time and charge passed through the cell (Figure 2c). In contrast, the intensity of C 1s photoemission line was reduced considerably upon discharge, indicating that the RGO flakes became covered by the ORR products, which is confirmed by the mapping of Li 1s on the electrode surface (see Supporting Information Figure S4 for more details). The analysis of the O 1s spectra reveals the presence of superoxide on the RGO surface upon discharge. Figure 2d shows differential O 1s spectra as a function of discharge capacity, which was obtained from subtracting the O 1s spectrum of the cell prior to discharge (Supporting Information Figure S4). At the onset of discharge, superoxide (Li+Ȯ 2−) located at 534.9 eV20 was found to be dominant by examining differential XPS intensities. Although superoxide was detected during ORR in organic electrolytes by Raman scattering,5 to the best of the authors’ knowledge this is the first report on the interface of molecular oxygen and a solid-state glass-ceramic electrolyte. Further discharge led to the pronounced growth of a peak positioned at 532.5 eV in the differential intensities with increasing capacity, which can be attributed to the formation of lithium peroxide20 either from further electrochemical reduction of superoxide or chemical disproportionation of superoxide, and epoxy groups21,22 and carbonates23/carboxylates.24,25 The formation of epoxy groups and carbonates/carboxylates on carbon was further supported by the C 1s spectra (Figure 2e), which reveal gradual growth of components at 286.1 eV (epoxy groups21,22) and 290.5 eV (carbon bound to three oxygen atoms23) grow during the discharge. These observations suggest 4698

dx.doi.org/10.1021/nl4021649 | Nano Lett. 2013, 13, 4697−4701

Nano Letters

Letter

Figure 3. Chemical transformations of superoxide species. The scheme illustrates chemical processes that are being initiated right after ORR.

Figure 2. The growth of ORR product layer on the surface of RGO. (a) The amount of charge Q passed during the cell discharge. (b) The corresponding voltage profile and discharge current pulses. The colored circles denote the points when the photoemission spectra were acquired. (c) The dependence of Li and O surface concentration increments (circles) and O/Li ratio (dashed line) on the charge passed. (d) O 1s core level differential spectra obtained by subtraction of the O 1s spectrum of pristine electrode. Colors of the experimental points designate the time when the corresponding spectrum was acquired (see panel b). The contributions at binding energies of 534.9, 531.5, and 532.5 eV are marked as Li+Ȯ 2− (superoxide), Li2O2/epoxy and carbonate, respectively. (e) The evolution of the high-resolution C 1s spectrum of a pristine cathode (the top spectrum) upon the cell discharge. Experimental data are shown as colored open circles and fitted curves are shown as solid black lines. Colors of the experimental points denote the discharge capacity shown in panel b. Spectrum deconvolution is demonstrated for a discharged RGO cathode; components that essentially increased (carbonate and epoxy) are filled with pink.

the integral intensity of C 1s line increases (Figure 4d) that is accompanied by Li 1s intensity fade (see Supporting Information Figure S6). Further charge occurs at higher potentials and leads to elimination of the carbonate component in C 1s (Figure 4d) and continuous fall in Li and O intensities. After the cell recharge, the total integral intensity of C 1s line that corresponds to surface concentration of carbon atoms is restored up to its initial state, indicating full elimination of the discharge products allowing further charge/discharge cycles (Figure 4d). The peak of carbonates was found to grow and reach saturation over one hour during OCP (Figure 5a). Moreover, the peak intensity ratio of carbonates normalized to sp2 carbon was not dependent on oxygen concentration, which suggests that neither oxygen adsorption nor its primary binding with olefin bonds is rate-limiting and the rearrangement of functional groups to form Li2CO3 and/or organic carbonate in the

Figure 4. Electrochemical and chemical processes occurring during Li−O2 cell discharge, open circuit, and recharge cycles. (a) The voltage profile during the experiment. Large overvoltages are observed due to high impedance of all-solid cathode, solid electrolyte decomposition was found to be negligible (see Supporting Information Figure S7 for details). Colored circles denote the points when C 1s spectra were collected. An inset demonstrates a close-up of open-circuit voltage profile. (b−d) The evolution of C 1s core level spectra during constant current discharge, open circuit and recharge, respectively. The colors of experimental curves correspond to those of the circles in voltage profile. Arrows show the changes in spectra occurring when time passes. (e,f) The surface composition during discharge (e) and OCP (f) cycles. 4699

dx.doi.org/10.1021/nl4021649 | Nano Lett. 2013, 13, 4697−4701

Nano Letters

Letter

leading to epoxy-groups on carbon, which are further converted into carbonates. We show that double bonds or aromatic systems activated by the presence of oxygenated functional groups and associated defects promote carbonate production. These results demonstrate that either temporary “isolation” of active superoxide ions by specially designed molecules31 with its further reduction to peroxide or utilization of carbons with low defect and functional group concentration could be the ways for “safe” carbon operation in the positive lithium−air electrodes.

■

ASSOCIATED CONTENT

* Supporting Information S

Further details of experimental methods (synthesis of RGO, assembly of the cells for in situ studies), details of photoelectron spectra analysis, and electrochemical experiments with glassy carbon electrodes. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Present Address ¶

(P.D.) Diamond Light Source Ltd., Diamond House Harwell Science and Innovation Campus, Didcot, Oxfordshire, OX11 0DE, U.K.

Figure 5. The kinetics of carbonate production in Li−O2 cell cathode. (a) The carbon-to-carbonate conversion for RGO cathode at different oxygen pressures. α was calculated as 100%(C − C0)/Cfinal, where C is a carbonate fraction in C 1s spectra, C0 and Cfinal are measured right after discharge and when the spectra ceased to display any changes, respectively. (b) The comparison of carbonate production on RGO and TEG cathodes at 0.1 mbar oxygen pressure. Carbonate production is estimated by the increment of carbonate component with binding energy of 290.5 eV during the open-circuit period.

Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors appreciate the fruitful discussions with the members of Electronic Structure group of Fritz Haber Institute in Berlin. L.V.Y. and D.M.I. gratefully acknowledge RussianGerman Laboratory at BESSY II for the financial support. D.M.I, D.A.S., E.A.G, and A.I.B. acknowledge FM Lab llc. company for support and the equipment provided.

presence of lithium ions at room-temperature limits the conversion kinetics. The initial imperfections of sp2 carbon materials influence the surface carbonate production. Thermally exfoliated graphite (TEG) with a specific surface area of about 200−250 m2/g has a smaller amount of defects (see Supporting Information Figure S8 for Raman spectra) and oxygenated functional groups than RGO based on the C:O ratios of more than 100:1 and about 7:1 obtained from XPS data, respectively. The relative amount of carbonate produced on well-developed RGO surface (more than 400 m2/g) after 4 μAh discharge is two times higher compared to TEG, as shown in Figure 5b. This observation suggests that defects and oxygenated functional groups associated with them are potentially active sites for superoxide radical attacks to form carbonates as it is well-known for Michael-type activated double bonds in olefins and substituted aromatics.28 Recent DFT results of Xu et al.29 show that the armchair edge and divacancy, unlike basal plane, are highly reactive and can be potentially oxidized at ambient conditions to carbonates and lactones even without being activated by oxygenated functional groups. Thus defects in carbon materials can play a dual role making them vulnerable for superoxide attacks on the one hand but enhancing the catalytic activity of the material30 on the other. In this study, in situ ambient pressure XPS experiments reveal that superoxide degrades carbon and limit the use of carbon electrodes for rechargeable lithium−air batteries. Superoxide radicals formed by one-electron oxygen reduction reaction promote nucleophilic addition or electron transfer reactions

■

REFERENCES

(1) Abraham, K. M.; Jiang, Z. J. Electrochem. Soc. 1996, 143, 1. (2) Girishkumar, G.; Mccloskey, B.; Luntz, A. C.; Swanson, S.; Wilcke, W. J. Phys. Chem. Lett. 2010, 1, 2193−2203. (3) Bruce, P. G.; Freunberger, S. A.; Hardwick, L. J.; Tarascon, J.-M. Nat. Mater. 2011, 11, 19−29. (4) Lu, Y.-C.; Gallant, B. M.; Kwabi, D. G.; Harding, J. R.; Mitchell, R. R.; Whittingham, M. S.; Shao-Horn, Y. Energy Environ. Sci. 2013, 6, 750. (5) Peng, Z.; Freunberger, S. A.; Hardwick, L. J.; Chen, Y.; Giordani, V.; Bardé, F.; Novak, P.; Graham, D.; Tarascon, J.-M.; Bruce, P. G. Angew. Chem., Int. Ed. 2011, 50, 6351−6355. (6) Carter, E.; Carley, A. F.; Murphy, D. M. J. Phys. Chem. C 2007, 111, 10630−10638. (7) Lau, K. C.; Curtiss, L. A.; Greeley, J. J. Phys. Chem. C 2011, 115, 23625−23633. (8) Laoire, C. O.; Mukerjee, S.; Abraham, K. M.; Plichta, E. J.; Hendrickson, M. A. J. Phys. Chem. C 2010, 114, 9178−9186. (9) Freunberger, S. A.; Chen, Y.; Peng, Z.; Griffin, J. M.; Hardwick, L. J.; Bardé, F.; Novak, P.; Bruce, P. G. J. Am. Chem. Soc. 2011, 133, 8040−8047. (10) Herranz, J.; Garsuch, A.; Gasteiger, H. A. J. Phys. Chem. C 2012, 116, 19084−19094.

4700

dx.doi.org/10.1021/nl4021649 | Nano Lett. 2013, 13, 4697−4701

Nano Letters

Letter

(11) Danen, W. C.; Warner, R. J.; Arudi, R. L. In Organic free radicals: a symposium; ACS Symposium Series 69; American Chemical Society: Washington, DC, 1978; pp 244−257. (12) Frimer, A. A.; Rosenthal, I. Photochem. Photobiol. 1978, 28, 711−717. (13) McCloskey, B. D.; Speidel, A.; Scheffler, R.; Miller, D. C.; Viswanathan, V.; Hummelsho̷ , J. S.; No̷ rskov, J. K.; Luntz, A. C. J. Phys. Chem. Lett. 2012, 3, 997−1001. (14) Gallant, B. M.; Mitchell, R. R.; Kwabi, D. G.; Zhou, J.; Zuin, L.; Thompson, C. V.; Shao-Horn, Y. J. Phys. Chem. C 2012, 116, 20800− 20805. (15) Xiao, J.; Mei, D.; Li, X.; Xu, W.; Wang, D.; Graff, G. L.; Bennett, W. D.; Nie, Z.; Saraf, L. V.; Aksay, I. A.; Liu, J.; Zhang, J.-G. Nano Lett. 2011, 11, 5071−5078. (16) Mitchell, R. R.; Gallant, B. M.; Thompson, C. V.; Shao-Horn, Y. Energy Environ. Sci. 2011, 4, 2952. (17) Lu, Y.-C.; Crumlin, E. J.; Veith, G. M.; Harding, J. R.; Mutoro, E.; Baggetto, L.; Dudney, N. J.; Liu, Z.; Shao-Horn, Y. Sci. Rep. 2012, 2, 715. (18) Zhang, C.; Grass, M. E.; McDaniel, A. H.; DeCaluwe, S. C.; Gabaly, F. E.; Liu, Z.; McCarty, K. F.; Farrow, R. L.; Linne, M. A.; Hussain, Z.; Jackson, G. S.; Bluhm, H.; Eichhorn, B. W. Nat. Mater. 2010, 9, 944−949. (19) McCloskey, B. D.; Bethune, D. S.; Shelby, R. M.; Mori, T.; Scheffler, R.; Speidel, A.; Sherwood, M.; Luntz, A. C. J. Phys. Chem. Lett. 2012, 3, 3043−3047. (20) Qiu, S. L.; Lin, C. L.; Chen, J.; Strongin, M. Phys. Rev. B 1989, 39, 6194. (21) Barinov, A.; Malcioǧlu, O. B.; Fabris, S.; Sun, T.; Gregoratti, L.; Dalmiglio, M.; Kiskinova, M. J. Phys. Chem. C 2009, 113, 9009−9013. (22) Ganguly, A.; Sharma, S.; Papakonstantinou, P.; Hamilton, J. J. Phys. Chem. C 2011, 115, 17009−17019. (23) Ensling, D.; Thissen, A.; Jaegermann, W. Appl. Surf. Sci. 2008, 255, 2517−2523. (24) Dedryvère, R.; Laruelle, S.; Grugeon, S.; Gireaud, L.; Tarascon, J. M.; Gonbeau, D. J. Electrochem. Soc. 2005, 152, A689. (25) Zhuang, G.; Chen, Y.; Ross, P. N. Langmuir 1999, 15, 1470− 1479. (26) Black, R.; Lee, J.-H.; Adams, B.; Mims, C. A.; Nazar, L. F. Angew. Chem., Int. Ed. 2012, 52, 392−396. (27) Gallant, B. M.; Kwabi, D. G.; Mitchell, R. R.; Zhou, J.; Thompson, C. V.; Shao-Horn, Y. Energy Environ. Sci. 2013, 1−33. (28) Frimer, A. A. In The Chemistry of Functional Groups, Peroxides; Patai, S., Ed.; John Wiley & Sons: New York, 1983; pp 429−461. (29) Xu, Y.; Shelton, W. A. J. Electrochem. Soc. 2011, 158, A1177. (30) Nakanishi, S.; Mizuno, F.; Abe, T.; Iba, H. Electrochemistry 2012, 80, 783−786. (31) Lopez, N.; Graham, D. J.; McGuire, R.; Alliger, G. E.; ShaoHorn, Y.; Cummins, C. C.; Nocera, D. G. Science 2012, 335, 450−453.

4701

dx.doi.org/10.1021/nl4021649 | Nano Lett. 2013, 13, 4697−4701