Journal of Power Sources 294 (2015) 588e592
Contents lists available at ScienceDirect
Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour
Role of the solid electrolyte interphase on a Li metal anode in a dimethylsulfoxide-based electrolyte for a lithiumeoxygen battery Norihiro Togasaki, Toshiyuki Momma, Tetsuya Osaka* Department of Applied Chemistry, Faculty of Science and Engineering, Waseda University, 3-4-1, Okubo, Shinjuku-ku, Tokyo, 169-8555, Japan
h i g h l i g h t s
g r a p h i c a l a b s t r a c t
Inorganic compounds of a SEI enhance the cycling of a Li metal anode in the DMSO. Inorganic compounds should form on the SEI surface and interior. Organic compounds should not be a dominant component of the SEI. The inorganic layer protects the Li anode against H2O.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 5 April 2015 Received in revised form 9 June 2015 Accepted 16 June 2015 Available online xxx
The effect of the solid electrolyte interphase (SEI) on a Li anode on the chargeedischarge cycling performance in 1 M LiTFSI/dimethylsulfoxide electrolyte solution is examined by using chargeedischarge cycling. The chemical structure of the surface and interior of the SEI strongly affects the cycling performance of the anode. The observed coulombic efﬁciency is low (<45%) when organic compounds such as lithium alkyl carbonates and polycarbonate form predominantly on the surface and interior. However, when inorganic compounds such as Li2CO3, Li2O, and LiF form instead, the coulombic efﬁciency increases to >85%. This enhanced efﬁciency remains constant regardless of the O2 content and despite <1000 ppm concentration of the contaminant H2O in the electrolyte. Thus, the lithium surface should be protected by inorganic compounds prior to cycling to prevent it from undergoing side reactions with the electrolyte during cycling in the electrolyte. © 2015 Elsevier B.V. All rights reserved.
Keywords: Li metal anode SEI Inorganic layer Organic layer Dimethylsulfoxide
1. Introduction Lithiumeoxygen (LieO2) batteries are anticipated to be useful in large-scale energy storage because of their theoretically high energy density. The use of dimethylsulfoxide (DMSO) and tetraethylene glycol dimethyl ether (TEGDME) as solvents has been
* Corresponding author. E-mail address: [email protected]
(T. Osaka). http://dx.doi.org/10.1016/j.jpowsour.2015.06.092 0378-7753/© 2015 Elsevier B.V. All rights reserved.
widely studied in recent years; experiments using these electrolytes for >100 cycles have been carried out [1e3]. Compared with those in TEGDME, the Li ion conductivity, O2 diffusion coefﬁcient, discharge voltage, speciﬁc capacity, and stability of carbon cathodes in DMSO are higher [4e6]. However, the Li metal anode is unstable in DMSO [1,7]. This problem may be due to formation of an unstable solid electrolyte interphase (SEI) on the lithium surface caused by side reactions during cycling . To prevent these side reactions, a stable SEI should be formed on the lithium surface before or during cycling. To stabilize the lithium anode for a LieO2 battery using DMSO solvent, investigations have attempted to use a lithium
N. Togasaki et al. / Journal of Power Sources 294 (2015) 588e592
anode dipped in LiClO4/propylene carbonate (PC) electrolyte and another in an electrolyte containing LiNO3 [1,9]. Both approaches may be suitable for modifying the lithium surface to prevent side reactions during cycling. However, a chemical structure of the SEI that is suitable for DMSO solvents has not yet been explored. In our previous study, we found that inorganic compounds on the SEI surface, such as Li2CO3 and LiF, play an important role in preventing side reactions, thereby enhancing the cycling performance of the lithium metal anode [10e12]. Inspired by this ﬁnding, we examined the effects of SEI compounds on a lithium surface on the coulombic efﬁciency of a lithium anode in 1 M LiTFSI/DMSO electrolyte by following the procedure proposed by Koch et al. . The dependence of the coulombic efﬁciency on the amount of H2O in the electrolyte, a potential impurity in the LieO2 cell, was also examined by using lithium anodes having SEIs of different chemical structures. 2. Experimental procedure A three-electrode beaker cell was used to perform electrochemical characterization. We prepared a nickel disc (5 mm diameter, 99.99% purity) for use as the working electrode, and we used lithium foil for the reference and counter electrodes. The coulombic efﬁciency of the lithium metal anode was estimated by using chargeedischarge cycling according to Koch's method . To prepare the lithium metal anodes, 5.1C cm2 of lithium was initially electrodeposited onto a nickel substrate at 2.0 mA cm2 in the electrolyte solutions 1 M LiClO4/ethylene carbonate (EC) and diethyl carbonate (DEC) (50:50 EC/DEC volume ratio, lithium battery grade; Tomiyama Pure Chemical Industries), 1 M LiPF6/EC-DEC (50:50 EC/DEC volume ratio, lithium battery grade; Kishida Chemical), and 1 M LiNO3/DMSO (<50 ppm H2O content) that was prepared by mixing as-received LiNO3 salt (SigmaeAldrich) and dehydrated DMSO solvent (Kanto Chemical). An electrolyte solution of 1 M LiClO4/EC-DEC purged with carbon dioxide gas (<50 ppm H2O content, 99.99% purity) for 3 h was also used to prepare the Li anode. The obtained Li anodes were then transferred to an electrolyte solution of 1 M LiTFSI/DMSO (Tomiyama Chemical), which had a H2O content of <30 ppm. 1.0C cm2 of lithium was discharged (dissolved) and charged (deposited) at 2.0 mA cm2 in each subsequent cycle. Before it was transferred to the 1 M LiTFSI/ DMSO solution, the working electrode was rinsed with 1 M LiTFSI/ DMSO to remove residual solvent. A cycling test in 1 M LiClO4/ECDEC solution was also conducted for comparison. The endpoint of the cycling life was detected at a potential of 1 V vs. Li/Liþ. All chargeedischarge tests were conducted in an Ar-ﬁlled glove box with dew point below 90 C (<0.1 ppm H2O). The chemical state of elements on the Li surface was analyzed by Fourier transform infrared (FTIR) spectroscopy (Shimadzu, IR Prestige-21) and by X-ray photoelectron spectroscopy (XPS) (ULVAC-PHI, VersaProbe-II). FTIR spectra were carried out in the attenuated total reﬂection mode using a Ge prism at a resolution of 8 cm1 and an accumulation of 200 scans. Settings for XPS analysis were the same as those described in our previous paper . Prior to these measurements, all samples were washed with pure dimethyl carbonate, dried under vacuum, and then transferred to the machines without exposing them to the ambient atmosphere. 3. Results and discussion Fig. 1 shows the XPS spectra of the lithium initially deposited at 5.1C cm2 in the 1 M LiClO4/EC-DEC electrolyte with and without CO2 bubbling. To understand the details of the chemical structure of the SEI surface and interior, XPS analysis with and without Arþ etching were conducted. SEI compounds formed on the surface
layer without CO2 bubbling were markedly different from those formed with bubbling. That is, the main products of the side reactions with the electrode in the former case were lithium alkyl carbonates (ROCO2Li) and polycarbonate, the peaks of which appeared at 533.0 and 534.5 eV, respectively ; those of the side reactions with the electrode in the latter case were Li2CO3 and Li2O, the peaks of which appeared at 531.5 and 529 eV, respectively . Peak resolution of the O1s spectra with respect to ROCO2Li and polycarbonate (533.8 eV), Li2CO3 (531.5 eV), and Li2O (529.0 eV), indicates that the former SEI was composed of ROCO2Li and polycarbonate (80.2%), Li2CO3 (15.4%), and Li2O (4.4%); whereas the latter SEI was composed of Li2CO3 (61.6%), Li2O (27.1%), ROCO2Li and polycarbonate (11.3%). The Li1s spectra also show a difference in chemical structures of the SEIs, similar to that shown by the O1s spectra. Li2CO3 may form via the reaction between ROCO2Li and CO2 or H2O [10,12], and Li2O may form via the reaction between H2O and Liþ . Chemical structures of the interior of SEIs deposited in electrolytes with and without CO2 showed similar properties; both contained Li2CO3 as main constituent, as well as small amounts of ROCO2Li and polycarbonate. To support XPS data mentioned above, FTIR analysis of the lithium initially deposited at 5.1C cm2 in the 1 M LiClO4/EC-DEC electrolyte with and without CO2 bubbling was conducted. As shown in Fig. S1, absorption peaks for both electrodeposited Li surfaces at 1647, 1317, 1100, and 820 cm1, which are attributed to ROCO2Li [14,17], and those at 1520e1430 and 875 cm1, which are attributed to Li2CO3 [17,18], are observed. These spectra have similar features except for the intensities of the Li2CO3 peaks; those for lithium electrodeposited in electrolyte with CO2 bubbling are slightly higher than those for lithium electrodeposited without bubbling. FTIR spectroscopy also conﬁrmed that these SEI ﬁlms were composed of ROCO2Li and Li2CO3, and the SEI formed with CO2 has higher amount of Li2CO3 than that without CO2. The chargeedischarge test was performed in 1 M LiTFSI/DMSO electrolyte by using lithium anodes whose SEI surface structure was varied to estimate the coulombic efﬁciency of the lithium metal anode. A cycling test was also conducted in 1 M LiClO4/EC-DEC electrolyte for comparison. Fig. 2 shows the coulombic efﬁciencies of lithium anodes with two different SEI surfaces, i.e., an organiccompound-rich surface and inorganic-compound-rich surface, in electrolyte solutions of 1 M LiClO4/EC-DEC or 1 M LiTFSI/DMSO. The typical potential proﬁle during cycling in 1 M LiClO4/EC-DEC and 1 M LiTFSI/DMSO are shown in Fig. S2. Both lithium anodes had a coulombic efﬁciency of >90% in 1 M LiClO4/EC-DEC solution, and the efﬁciency increased slightly when lithium anode with an inorganic-compound-rich surface was used. This phenomenon is in agreement with our previous study . The coulombic efﬁciency of both anodes in 1 M LiTFSI/DMSO solution was found to be lower than that in 1 M LiClO4/EC-DEC electrolyte. This means that side reactions on the lithium surface in 1 M LiTFSI/DMSO occur more readily than they do in 1 M LiClO4/EC-DEC. In general, when the SEI layer breaks down after chargeedischarge cycling, a fresh lithium surface becomes exposed and reacts with the electrolyte. This reaction induces repair of the SEI layer on the lithium surface, and the repaired layer contributes to the prevention of side reactions in subsequent cycles. However, formation of a stable SEI on a lithium surface in a DMSO-based electrolyte is difﬁcult . Therefore, side reactions during cycling in 1 M LiTFSI/DMSO electrolyte tend to occur where the SEI layer has broken down, leading to lower coulombic efﬁciency of the lithium metal anode. A signiﬁcant improvement in the cycling performance in 1 M LiTFSI/DMSO was observed when lithium anode with inorganiccompound-rich surface was used in place of that with an organiccompound-rich surface. The coulombic efﬁciency obtained with the former anode was 86.3%, which is 12.5% higher than that of the
N. Togasaki et al. / Journal of Power Sources 294 (2015) 588e592
Fig. 1. O1s spectra for an electrodeposited lithium anode obtained from an electrolyte solution of 1 M LiClO4/EC-DEC (a) without and (b) with CO2 bubbling (3 h).
Fig. 2. Effects of the composition ratio of the SEI surface on the coulombic efﬁciency of a lithium metal anode in 1 M LiClO4/EC-DEC solution and in 1 M LiTFSI/DMSO solution. Coulombic efﬁciencies shown in the organic-compound-rich region and in the inorganic-compound-rich region were obtained by using lithium initially deposited in 1 M LiClO4/EC-DEC solutions without and with CO2 bubbling, respectively.
latter anode. This result suggests that the inorganic layer on the SEI surface plays an important role in preventing side reactions with the solvent DMSO. Similarly, enhanced coulombic efﬁciency (85.4%) was obtained even when an O2-saturated electrolyte solution of 1 M LiTFSI/DMSO was used for the cycling test. From this result, we conclude that dissolved O2 in the electrolyte does not affect the coulombic efﬁciency of the lithium metal anode when inorganic compounds are a dominant constituent on the lithium surface prior to cycling. Therefore, the inorganic layer composed of Li2CO3 and Li2O on the SEI surface had a passivation property.
The inorganic layer improved the cycling performance of the lithium metal anode in 1 M LiTFSI/DMSO. The protective effect of this layer on a lithium anode, on LiF, and on LiNO2 (or Li2O) has also been occasionally reported in the literature [9,19]. To evaluate the effect of compounds in the SEI on the cycling performance of a lithium anode in 1 M LiTFSI/DMSO, lithium was initially electrodeposited by using electrolyte solutions of 1 M LiPF6/EC-DEC and 1 M LiNO3/DMSO to respectively form LiF and LiNO2 (or Li2O) on the lithium surface [9,10]. Fig. S3(a)e(c) and Fig. S3(d)e(f) display XPS spectra of electrodeposited lithium anodes obtained from 1 M LiPF6/EC-DEC and 1 M LiNO3/DMSO, respectively. F1s spectra of the SEI obtained from the 1 M LiPF6/ECDEC electrolyte (Fig. S3(a)) suggest that a LiF layer formed on the surface and in the interior of the SEI, as evidenced by the peak binding energy at 685 eV. Results for the Li1s spectra are similar, as shown in Fig. S3(b). O1s spectra in Fig. S3(c) indicate that the SEI obtained from 1 M LiPF6/EC-DEC was similar to that obtained from 1 M LiClO4/EC-DEC (Fig. 1(a)); that is, main compounds on the SEI surface were polycarbonate and ROCO2Li, and the main compound in the interior was Li2CO3. However, peak resolution of the Li1s spectrum of the SEI surface with respect to LiF (57.0 eV), ROCO2Li (56.3 eV), Li2CO3 (55.3 eV), and Li2O (53.7 eV) implies that the compound on the uppermost part of the SEI was LiF. N1s spectra of the SEI obtained from 1 M LiNO3/DMSO solution indicate that nitrogen compounds were absent. On the other hand, Fig. S3(e) and (f) show that Li2O formed through a reaction in the interior of the SEI; that is, the binding energies which are observed around Li1s (53.7 eV) and O1s (529.0 eV) are corresponding to Li2O. The absence of nitrogen compounds in the SEI is attributed to the solubility of LiNO2 in DMSO; LiNO2 may be soluble whereas Li2O is highly insoluble in DMSO, similar to their behavior in N,N-dimethylacetamide .
2Li þ LiNO3 /Li2 O þ LiNO2 We observed a marked difference in the interiors of SEIs obtained from 1 M LiNO3/DMSO and from other electrolytes. The ROCO2Li content of the interior of the SEI obtained from 1 M LiNO3/
N. Togasaki et al. / Journal of Power Sources 294 (2015) 588e592
DMSO was higher than that of SEIs obtained from 1 M LiClO4/ECDEC and 1 M LiPF6/EC-DEC. In the depth analysis by Arþ ion sputtering, the change of the SEI's compounds due to ion beam-induced damages has to be taken into consideration. ROCO2Li can be changed to Li2CO3 or Li2O [21,22]. However, the peak intensity of ROCO2Li in the interior of the SEI obtained from 1 M LiNO3/DMSO is higher compared to those from other electrolytes. Therefore, we assume a larger amount of ROCO2Li exists in the interior of the SEI when lithium is electrodeposited in 1 M LiNO3/DMSO. Fig. 3 shows the effects of the main compounds on the surface and in the interior of the SEI on the coulombic efﬁciency of the lithium anode in 1 M LiTFSI/DMSO electrolyte. The lithium anodes used for this cycling test were prepared by electrodeposition in 1 M LiClO4/EC-DEC electrolyte solution with and without CO2, as well as by electrodeposition in 1 M LiPF6/EC-DEC or 1 M LiNO3/DMSO. The coulombic efﬁciency (<45%) was low when the organic compound was one of the dominant components on the surface and in interior of the SEI. In contrast, the coulombic efﬁciency was >70% when inorganic compounds were the only dominant constituents in the SEI interior. The coulombic efﬁciency increased to >85% when inorganic compounds were the only dominant constituents on the SEI surface and interior. This enhanced coulombic efﬁciency was obtained regardless of the inorganic species (Li2CO3, Li2O, and LiF) used in this study. The results described above suggest that SEI compounds and their location strongly affect the coulombic efﬁciency of a lithium anode in 1 M LiTFSI/DMSO solution. Similarly, inorganic compounds play a signiﬁcant role in improving the coulombic efﬁciency in carbonate-based electrolytes [10,12]; formation of such compounds to dominant proportions on the surface and in the interior of the SEI layer leads to enhanced protection against side reactions with the electrolyte. Therefore, desolvation of Li ions should occur at the interface between the electrolyte and SEI layer to prevent side reactions; that is, any solvated lithium ions or dissociated solvents should not be in contact with the Li metal surface. However, some solvated lithium ions or solvents can reach the lithium metal surface since diffusion
Fig. 3. Effects of main compounds on the SEI surface and interior on the coulombic efﬁciency of a lithium metal anode in 1 M LiTFSI/DMSO solution. Coulombic efﬁciencies depicted from left to right were obtained from initial deposition of lithium in solutions of 1 M LiClO4/EC-DEC without CO2 bubbling, 1 M LiClO4/EC-DEC with CO2 bubbling, 1 M LiPF6/EC-DEC, and 1 M LiNO3/DMSO.
in the inorganic layer occurs not only through the crystalline regions and but also through the grain boundary . Therefore, the inorganic layer must be of a certain thickness to prevent side reactions with the electrolyte. Surface morphology of a deposited Li is also related to the coulombic efﬁciency, and its smooth surface produces the better cycling performance due to the decrease of the active area for side reactions. Formation of the inorganic species (Li2CO3, Li2O, and LiF) on the Li anode is reported to induce the smoother surface since they provide a more uniform current distribution [12,19]. In this study, surface morphology may also affect the coulombic efﬁciency as well as protective function by inorganic species. In a LieO2 battery, pure oxygen gas with low water content is used; however, accumulation of water from the oxygen gas in the electrolyte is a concern during long-term cycling. In particular, water from the atmosphere may affect the performance of the LieO2 cell, especially the lithium anode, when the electrolyte quantity is small, as in a coin cell or in a Swagelok cell. Thus, we evaluated the inﬂuence of the H2O content in 1 M LiTFSI/DMSO solution on the coulombic efﬁciency of the lithium metal anode. Fig. 4 shows the dependence of the coulombic efﬁciency of the anode with various SEI compounds on the H2O content of a 1 M LiTFSI/DMSO solution. Prior to cycling in 1 M LiTFSI/DMSO, electrodeposited lithium anodes were prepared by using 1 M LiClO4/ EC-DEC solutions with and without CO2 bubbling. The resulting SEI surface from the former solution was mainly composed of inorganic compounds such as Li2CO3 and Li2O, and that from the latter solution was mainly composed of organic compounds such as ROCO2Li and polycarbonate (Fig. 1). The coulombic efﬁciency of the lithium anode with SEI surface consisting of organic compounds decreased with the increase in H2O content of the electrolyte, reaching 0% at a H2O content of 3000 ppm. This decrease may be due to the reaction between the Li anode and H2O. In contrast, the coulombic efﬁciency of the lithium anode with SEI surface consisting of inorganic compounds remained constant at a H2O content of 1000 ppm in the electrolyte, decreasing to ~70% at a H2O content of 3000 ppm. This protective function of the inorganic SEI layer against H2O for the lithium anode was similarly observed in our previous study in which stable cycling performance was attained at
Fig. 4. Dependence of the coulombic efﬁciency of a lithium metal anode on the H2O content of 1 M LiTFSI/DMSO electrolyte. Black and white dots respectively indicate results from tests using deposited lithium anode obtained from 1 M LiClO4/EC-DEC electrolyte with CO2 bubbling for 3 h and without CO2 bubbling.
N. Togasaki et al. / Journal of Power Sources 294 (2015) 588e592
a H2O content of 3000 ppm in 1 M LiClO4/PC electrolyte . The difference between DMSO and PC-based electrolyte may be due to the difﬁculty of restoring the SEI during cycling in DMSO, as previously mentioned. Nevertheless, passivation by the inorganic SEI layer against H2O contamination in the DMSO-based electrolyte was still conﬁrmed. 4. Conclusion Compared with organic compounds such as ROCO2Li and polycarbonate, inorganic compounds such as Li2CO3, Li2O, and LiF in the SEI provide improved protection against side reactions with 1 M LiTFSI/DMSO electrolyte solutions. Thus, inorganic compounds should be formed on the surface and in the interior of the SEI to dominant proportions to improve the coulombic efﬁciency of a lithium anode, and vice versa for organic compounds. Inorganic compounds in the SEI also protect against H2O contamination at a level of 1000 ppm of 1 M LiTFSI/DMSO solution. Acknowledgments This work was partially supported by the Grants for Excellent Graduate Schools (Practical Chemical Wisdom) of MEXT, Japan. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.jpowsour.2015.06.092. References  Z.Q. Peng, S.A. Freunberger, Y.H. Chen, P.G. Bruce, Science 337 (2012) 563e566.
 M.M.O. Thotiyl, S.A. Freunberger, Z.Q. Peng, Y.H. Chen, Z. Liu, P.G. Bruce, Nat. Mater. 12 (2013) 1049e1055.  H.G. Jung, J. Hassoun, J.B. Park, Y.K. Sun, B. Scrosati, Nat. Chem. 4 (2012) 579e585.  C.O. Laoire, S. Mukerjee, K.M. Abraham, E.J. Plichta, M.A. Hendrickson, J. Phys. Chem. C 114 (2010) 9178e9186.  D. Xu, Z.L. Wang, J.J. Xu, L.L. Zhang, X.B. Zhang, Chem. Commun. 48 (2012) 6948e6950.  M.M.O. Thotiyl, S.A. Freunberger, Z.Q. Peng, P.G. Bruce, J. Am. Chem. Soc. 135 (2013) 494e500.  Y.H. Chen, S.A. Freunberger, Z.Q. Peng, O. Fontaine, P.G. Bruce, Nat. Chem. 5 (2013) 489e494.  M. Roberts, R. Younesi, W. Richardson, J. Liu, T. Gustafsson, J.F. Zhu, K. Edstrom, ECS Electrochem. Lett. 3 (2014) A62eA65.  B. Sun, X.D. Huang, S.Q. Chen, J.Q. Zhang, G.X. Wang, RSC Adv. 4 (2014) 11115e11120.  N. Togasaki, T. Momma, T. Osaka, J. Power Sources 261 (2014) 23e27.  T. Osaka, T. Momma, Y. Matsumoto, Y. Uchida, J. Electrochem. Soc. 144 (1997) 1709e1713.  T. Osaka, T. Momma, T. Tajima, Y. Matsumoto, J. Electrochem. Soc. 142 (1995) 1057e1060.  J.L. Goldman, R.M. Mank, J.H. Young, V.R. Koch, J. Electrochem. Soc. 127 (1980) 1460e1467.  H. Ota, Y. Sakata, Y. Otake, K. Shima, M. Ue, J. Yamaki, J. Electrochem. Soc. 151 (2004) A1778eA1788.  H. Ota, Y. Sakata, X.M. Wang, J. Sasahara, E. Yasukawa, J. Electrochem. Soc. 151 (2004) A437eA446.  D. Aurbach, M. Daroux, P. Faguy, E. Yeager, J. Electroanal. Chem. 297 (1991) 225e244.  D. Aurbach, M.D. Levi, E. Levi, A. Schechter, J. Phys. Chem. B 101 (1997) 2195e2206.  K. Morigaki, A. Ohta, J. Power Sources 76 (1998) 159e166.  Z. Takehara, J. Power Sources 68 (1997) 82e86.  V. Giordani, W. Walker, V.S. Bryantsev, J. Uddin, G.V. Chase, D. Addison, J. Electrochem. Soc. 160 (2013) A1544eA1550.  A.M. Andersson, A. Henningson, H. Siegbahn, U. Jansson, K. Edstrom, J. Power Sources 119 (2003) 522e527.  P. Kubiak, M. wolfahrt-Mehrens, K. Edstrom, M. Morcrette, Review on ageing mechanisms of different Li-ion batteries for automotive applications, JPS Power D-12e03691.  H. Iddir, L.A. Curtiss, J. Phys. Chem. C 114 (2010) 20903e20906.