NO3– Dual-Anion Electrolyte for Suppressing

Jul 17, 2017 - ACS Applied Materials & Interfaces. Advanced .... GREEN, National Institute for Materials Science, 1-1 Namiki, Tsukuba 305-0044, Japan...
0 downloads 3 Views 7MB Size
Research Article www.acsami.org

Highly Efficient Br−/NO3− Dual-Anion Electrolyte for Suppressing Charging Instabilities of Li−O2 Batteries Xing Xin, Kimihiko Ito, and Yoshimi Kubo* GREEN, National Institute for Materials Science, 1-1 Namiki, Tsukuba 305-0044, Japan S Supporting Information *

ABSTRACT: The main issues with Li−O2 batteries are the high overpotential at the cathode and the dendrite formation at the anode during charging. Various types of redox mediators (RMs) have been proposed to reduce the charging voltage. However, the RMs tend to lose their activity during cycling owing to not only decomposition reactions but also undesirable discharge (shuttle effect) at the Li metal anode. Moreover, the dendrite growth of the Li metal anode is not resolved by merely adding RMs to the electrolytes. Here we report a simple yet highly effective method to reduce the charge overpotential while protecting the Li metal anode by incorporating LiBr and LiNO3 in a tetraglyme solvent as the electrolyte for Li−O2 cells. The Br−/Br3− couple acts as an RM to oxidize the discharge product Li2O2 at the cathode, whereas the NO3− anion oxidizes the Li metal surface to prevent the shuttle reaction. In this work, we found that both anions work synergistically in the mixed Br−/NO3− electrolyte to dramatically suppress both parasitic reactions and dendrite formation by generating a solid Li2O thin film on the Li metal anode. As a result, the charge voltage was reduced to below 3.6 V over 40 cycles. The O2 evolution during charging was more than 80% of the theoretical value, and CO2 emission during charging was negligible. After cycling, the Li metal anode showed smooth surfaces with no indication of dendrite formation. These observations clearly demonstrate that the Br−/NO3− dual-anion electrolyte can solve the problems associated with both the overpotential at the cathode and the dendrite formation at the anode. KEYWORDS: Li−O2 battery, Li-halides, redox mediators, dendritic growth, electrolyte, SEI film



INTRODUCTION Nonaqueous Li−O2 batteries have the highest theoretical specific energy (11 700 Wh per kg) among all battery systems.1−4 One of the key challenges that prevents the practical application of nonaqueous Li−O2 batteries is the high overpotential during the charging process, i.e., oxygen evolution reaction (OER), to oxidize the discharge product Li2O2. The charge voltage typically increases to 4.2−4.5 V, which not only results in low round-trip efficiency but also induces side reactions that lead to poor cycling stability.5,6 Thus, the O2 recovery during charging is as low as 50−70% of the theoretical value, and CO2 is typically emitted at the final stage of charging.7−9 To reduce the charge voltage to an acceptable range, recent studies have focused on soluble-type catalysts or redox mediators (RMs), which promote the charge transfer between the electrode surface and Li2O2.10−29 The RM is first oxidized on the cathode to RM+, which in turn oxidizes the Li2O2 to return to the RM. Therefore, the redox potential (Eredox) of the RM/RM+ couple largely determines the charge voltage. Various types of RMs have been reported, such as TTF (tetrathiafulvalene, Eredox ∼ 3.6 V),10−14 TEMPO (2,2,6,6tetramethylpiperidinyloxyl, ∼3.74 V),15−18 TDPA (tris[4(diethylaminophenyl]amine, ∼3.1 V),19 FePc (iron phthalocyanine, ∼3.65 V),20 and Li-halides (LiI and LiBr, ∼2.9−3.55 V and ∼3.5−3.9 V).21−25 All of these RMs lead to a substantial reduction in the charge voltage from 4.2−4.5 V to 3.3−3.7 V. © 2017 American Chemical Society

Nevertheless, most organic RMs such as TTF, TEMPO, and TDPA quickly lose their activity: the charge voltage increases within the initial several cycles, and CO2 due to side reactions is emitted at the final stage of each charging.13−15,19 Inorganic RMs like Li-halides also have some problems. Although LiIadded electrolytes often exhibit the lowest charge voltages, they easily induce side reactions to form LiOH.23,25 LiBr-added electrolytes can improve the cell performance more than LiI, but still emit CO2 at the end of the charge.24 In addition, the RM mechanism essentially involves instability that RM+ could be reduced to RM on the Li metal anode (shuttle effect).30 Therefore, to use RMs in practice, the Li metal surface must be effectively protected from reactions with RMs.31,32 Furthermore, the dendritic growth of Li metal should also be avoided for the reversibility of the battery.33−35 In this regard, LiNO3 is used as an electrolyte additive in Li−S batteries to prevent reactions between Li metal and soluble polysulfides by stabilizing the solid-electrolyte interphase (SEI).36 Recently, Uddin et al. found that adding LiNO3 in the electrolyte of a Li− O2 battery also promotes the formation of stable SEI films on the Li metal anode.37 However, whether this SEI film can Received: April 24, 2017 Accepted: July 17, 2017 Published: July 17, 2017 25976

DOI: 10.1021/acsami.7b05692 ACS Appl. Mater. Interfaces 2017, 9, 25976−25984

Research Article

ACS Applied Materials & Interfaces

Figure 1. Discharge performance and morphology. (a) First discharge profiles of the RGO cathodes (area of 2 cm2) operated at 0.1 mA to a cutoff voltage of 2.5 V. (b) Corresponding XRD patterns of the RGO cathodes after discharge. SEM images of the RGO cathodes after discharge in different electrolytes with tetraglyme solvent: (c) 1 M LiCF3SO3; (d) 0.05 M LiI−1 M LiCF3SO3; (e) 0.05 M LiBr−1 M LiCF3SO3; (f) 1 M LiNO3; (g) 0.05 M LiI−1 M LiNO3; (h) 0.05 M LiBr−1 M LiNO3.



RESULTS Electrochemical Properties. We examined six electrolytes with/without 50 mM Li-halides (LiI and LiBr) in standard 1 M LiCF3SO3/tetraglyme and 1 M LiNO3/tetraglyme electrolytes. The first discharge profiles of Li−O2 cells with reduced graphene oxide (RGO) cathodes are shown in Figure 1a. All cells delivered nearly the same discharge plateau of approximately 2.75 V during the stable discharging process, which is in accordance with previous reports on the negligible influence of RMs on the discharge voltage.10,15,25,26 However, the discharge capacities showed a remarkable dependence on electrolytes. In standard 1 M LiCF3SO3 electrolyte systems, adding Li-halides considerably decreased the discharge capacity, which was similar to other RM systems, such as TEMPO.15 Note that LiBr addition resulted in discharge capacities that were two times higher than that for LiI. For a 1 M LiNO3 electrolyte without any Li-halide addition, the cell presented a comparable yet slightly lower capacity than that of the standard LiCF3SO3 cell (15.7 mAh for LiNO3 and 17.2 mAh for LiCF3SO3 at a cutoff voltage of 2.5 V). This result is inconsistent with a previous report in which the discharge capacity is greatly enhanced by addition of LiNO3 to the electrolyte.38 The discrepancy might be due to the differences in solvents (tetraglyme vs dimethoxyethane) and/or cathode materials (graphene vs carbon black). The discharge capacity

suppress the shuttle reactions of the RM and/or mitigate the dendrite growth on the Li metal anode remains unclear. In this work, we develop a simple yet highly effective method to reduce the charge overpotential while protecting the Li metal anode by incorporating LiBr and LiNO3 in a tetraglyme solvent as the electrolyte for Li−O2 cells. In this case, the Br−/Br3− couple acts as an RM to oxidize the discharge product Li2O2 at the cathode, whereas the NO3− anion oxidizes the Li metal surface to prevent the shuttle reaction. Interestingly, we found that the both anions work synergistically in the mixed Br−/ NO3− electrolyte to dramatically suppress both parasitic reactions and dendrite formation by generating a solid Li2O thin film on the Li metal anode. As a result, the charge voltage was reduced to below 3.6 V over 40 cycles. Differential electrochemical mass spectrometry (DEMS) measurement revealed that the O2 evolution during charging was more than 80% of the theoretical value and that almost no CO2 was emitted during charging. Further tests demonstrated that side reactions were largely suppressed during cycling. The Li metal anode after cycling exhibited rather smooth surfaces covered with solid Li2O thin layers without any indication of dendrite formation. These observations clearly demonstrate that the Br−/NO3− dual-anion electrolyte can simultaneously solve both the problem of overpotential at the cathode and the problem of dendrite formation at the anode. 25977

DOI: 10.1021/acsami.7b05692 ACS Appl. Mater. Interfaces 2017, 9, 25976−25984

Research Article

ACS Applied Materials & Interfaces was similarly decreased by adding Li-halides to the LiNO3 electrolyte but less remarkably decreased compared to the LiCF3SO3-based electrolyte. X-ray diffraction (XRD) patterns (Figure 1b) of the discharged RGO cathodes exhibited peaks characteristic of Li2O2 crystals except for the LiI−LiCF3SO3 system. No other byproduct peaks, such as LiOH, were detected. The absence of Li2O2 peaks for the LiI−LiCF3SO3 electrolyte can be attributed to the formation of an amorphouslike structure, which was also confirmed in the Ketjenblack (KB) cathode for the same electrolyte and discharge capacity (Figure S1). The surface morphologies of the discharged RGO cathodes were observed by scanning electron microscopy (SEM). The cathode surface from the 1 M LiCF3SO3 cell was covered with typical toroidal crystals having diameters of 100− 200 nm, as shown in Figure 1c. By contrast, micrometer-sized (approximately 1 μm) particles composed of flower-like clusters (see Figure 1f) can be found in the 1 M LiNO3 cell. The flowerlike clusters are composed of thin nanosheets, which are closely connected to each other. In addition, flower-like particles of considerably smaller size were observed in the LiBr−LiCF3SO3, LiI−LiNO3, and LiBr−LiNO3 cells (Figure 1e,g,h). However, the cathode surface in the LiI−LiCF3SO3 electrolyte was covered only by films composed of thin platelets without clear particle edges (Figure 1d). The charging process after full discharge (see Figure 1a) is displayed in Figure S2. Although the charge voltage increases considerably to 4 V for the standard 1 M LiCF3SO3 electrolyte, it is considerably suppressed to approximately 3.5 V for other electrolyte systems. This clearly demonstrates the RM effects of these electrolytes. Figure S3 shows the morphology of the RGO cathodes after full recharge. All cathodes are shown to be back to the initial sheet-like structure of graphene. Next, we adopted a fixed capacity of 1 mAh to examine the RM effects in charging in further detail, as shown in Figure 2a. The standard 1 M LiCF3SO3 cell shows the highest charge voltage with a plateau of ∼4 V and an end point of >4.3 V. The 1 M LiNO3 cell presents a lower voltage profile, increasing from 3.5 to 4.0 V. By contrast, the LiI-containing cells exhibit the lowest charge profiles, with values below ∼3.6 V in the entire range of charging, showing initial plateaus at ∼3.2 V in accordance with the I−/I3− redox potential. The LiBr-containing cells exhibit even higher plateaus at ∼3.5 V because of the higher redox potential of Br−/Br3−,24,25 followed by a gradual increase to ∼3.8 V. The reduction reactions of the remaining I3− and Br3− in the electrolytes are clearly indicated in the second discharge profiles as plateaus at ∼3.0 and ∼3.3 V, respectively. This means that up to 10% of the halide ions remained inactive as an RM after becoming oxidized during the charging process in the present experimental conditions. The voltammetric responses of the cells under an oxygen atmosphere after 1 mAh discharge are shown in Figure 2b. The oxidation current for the 1 M LiCF3SO3 cell gradually increases to 4 V without exhibiting remarkable peaks. The 1 M LiNO3 cell exhibits a rapid increase in current above 3.5 V, which suggests a possible RM effect of this electrolyte, likely due to the NO 2 − /NO 2 redox reaction.29,39 The addition of Li-halides (LiX) to these electrolytes produces new peaks corresponding to the X−/X3− redox reaction. The LiI-containing cells show oxidation and reduction peaks at approximately 3.15 and 3.0 V, respectively. They also exhibit reduction peaks at ∼3.7 V corresponding to the reduction of I2 to I3−, which clearly demonstrates the formation of I2 during charging. By contrast, the LiBrcontaining cells exhibit a pair of redox peaks at approximately

Figure 2. Electrochemical performance. (a) First discharge/charge and second discharge profiles of the KB cathodes (area of 2 cm2) operated at 0.2 mA with a fixed capacity of 1 mAh in different electrolytes. (b) Cyclic voltammetry plots recorded at a scan rate of 0.1 mV s−1 under an oxygen atmosphere from OCV to 4.2 V after discharge to 1 mAh in different electrolytes.

3.3 V but no peaks for the Br2 reduction reaction. This result is quite reasonable considering the higher Br3−/Br2 redox potential of 4.1−4.2 V.24,25 The main oxidation peaks spanning from ∼3.3 to ∼4 V represent the oxidation of Li2O2, and are remarkably enhanced by the existence of RMs. The peak position for the LiI-containing cell is ∼0.1 V lower than that for the corresponding LiBr cell. In addition, the peak positions for the LiNO3-based cell are ∼0.1 V higher than those for the corresponding LiCF3SO3-based cell. The third oxidation peaks at approximately 3.8−4.2 V observed for the LiCF3SO3-based electrolytes appear to be due to the decomposition of byproducts, including lithium carbonate. Gas Evolution and Cycle Stability. Gas evolution during the charging process was measured for these electrolytes using DEMS. Without any Li-halide addition, the O2 evolution from the standard 1 M LiCF3SO3 cell presents a typical wavy profile, as shown in Figure 3a, which is in agreement with previous reports.7,40,41 The average O2 evolution rate was as low as 60% of that for the ideal two-electron reaction (O2/2e−). In addition, as shown in Figure S4a, a small amount of carbon monoxide (CO) was detected at the beginning of charging, and a significant amount of carbon dioxide (CO2) was evolved when the charge voltage increased to more than 4 V at the end of the charging process. The evolution of CO is likely due to the decomposition of tetraglyme,5,6,24 whereas the significant increase in CO2 evolution at the final stage of charge can be attributed to the decomposition of byproducts, such as Li2CO3 .5,7,42,43 The addition of Li-halides to LiCF3 SO3 increased the O2 evolution rate slightly and suppressed the CO2 evolution to some extent (Figure 3b,c). However, the O2 25978

DOI: 10.1021/acsami.7b05692 ACS Appl. Mater. Interfaces 2017, 9, 25976−25984

Research Article

ACS Applied Materials & Interfaces

Figure 3. In situ DEMS measurement. Galvanostatic discharge−charge voltage curves (up) and the corresponding DEMS results of O2 and CO2 evolution during charge in different electrolytes with tetraglyme solvent: (a) 1 M LiCF3SO3; (b) 0.05 M LiI−1 M LiCF3SO3; (c) 0.05 M LiBr−1 M LiCF3SO3; (d) 1 M LiNO3; (e) 0.05 M LiI−1 M LiNO3; and (f) 0.05 M LiBr−1 M LiNO3. The measurements were taken using carbon nanotube (CNT) cathodes (area of 2 cm2) at a current of 0.1 mA with a discharge capacity of 1.5 mAh. Red lines (e−/O2 = 2) indicate the theoretical O2 evolution rate according to the two-electron reaction.

degradation might be due to the formation of I2 molecules, as indicated by the discharge plateau at 3.7 V, which will aggressively react with the cell components.23,25 By contrast, the LiBr-containing cells exhibited rather stable plateaus at ∼3.5 V over 20 cycles, although the voltage increased to ∼4 V at the end of each charge. The LiBr−LiNO3 cell (Figure S5f) exhibited slightly lower voltages compared to the those of the LiBr−LiCF3SO3 cell (Figure S5c). The concentrations of anions (CF3SO3−, NO3−, I−, Br−) in these electrolytes were determined by ion chromatography before and after cycling, as shown in Figure S6. The I− concentration in the LiCF3SO3 electrolyte is remarkably decreased with cycling, which suggests the considerable formation of I2 at high voltage and its evaporation during cycling (Figure S6c). On the contrary, the Br− concentrations are rather stable during cycling (Figure S6e,f), likely because the Br2 formation is avoided. Characterization of Li Metal Anode. The surfaces of the Li metal anodes were surveyed using SEM and energy dispersive X-ray spectrometry (EDS), as shown in Figure S7. The surface becomes considerably rough after 20 cycles in both the LiCF3SO3 and LiNO3 electrolytes. The roughness is reduced by the addition of Li-halides, particularly LiBr. The EDS analysis showed only C and O peaks, without any signals

evolution rate was considerably improved by changing LiCF3SO3 to LiNO3. The DEMS profile of the Li−O2 cell with only a LiNO3 electrolyte presents a higher O2 evolution rate of ∼80% during charging, as shown in Figure 3d, whereas CO2 is still detected when the voltage increases at the end of the charge. The CO2 evolution is nearly entirely suppressed by adding Li-halides to the LiNO3 electrolyte, as shown in Figure 3e,f. In particular, the LiBr−LiNO3 electrolyte exhibits a rather high constant O2 evolution (∼80%) throughout the entire charging process (Figure 3f). Furthermore, liquid chromatography−mass spectrometry (LC−MS) analysis of the sidereaction products (polymerized ethers) in electrolyte after cycling with longer capacity (Table S1) demonstrates that sidereactions are largely suppressed by combining Li-halides and LiNO3. The cycle stability of these electrolytes was tested using commercial KB as a cathode material at 0.2 mA with a limited capacity of 1 mAh, as shown in Figure S5. After 20 cycles, the charge voltage of the standard LiCF3SO3 cell exceeded 4.5 V, whereas that of the LiNO3 cell was 4.3 V at the end of charge. Although the LiI-containing cells exhibited the smallest overpotential in the first cycle, their charge profiles deteriorated rapidly during cycling, as shown in Figure S5b,e. Such 25979

DOI: 10.1021/acsami.7b05692 ACS Appl. Mater. Interfaces 2017, 9, 25976−25984

Research Article

ACS Applied Materials & Interfaces

Figure 4. Characteristics of the Li anodes after cycling. (a) XPS survey spectra for the surfaces of the cycled Li-foil anodes and fresh (as-received) Li foil. XPS depth profiles of C, O, and Li compositions for Li anodes after cycling (b) in 0.05 M LiBr−1 M LiNO3 electrolyte and (c) in 0.05 M LiI−1 M LiNO3 electrolyte. SEM images of the cross section of the (d) fresh Li foil and cycled Li foil in (e) 1 M LiCF3SO3, (f) 1 M LiNO3, and (g) 0.05 M LiBr−1 M LiNO3 electrolytes with tetraglyme solvent. The cells (area of 2 cm2) using KB cathodes were cycled 20 times at a current of 0.2 mA with a fixed capacity of 1 mAh.

Figure 5. SEM images and EBSD mapping of the Li foil in different electrolytes with tetraglyme solvent after discharge (1 and 3) and after recharge (2 and 4). Parts 1 and 2 are the SEM images. Parts 3 and 4 are EBSD mapping images (inverse pole figure maps from Z direction). (a) Tested in 1 M LiCF3SO3; (b) tested in 1 M LiNO3; (c) tested in 0.05 M LiBr−1 M LiNO3. The Li-foil anodes (area of 2 cm2) were discharged/recharged at a current of 0.1 mA with a fixed capacity of 4 mAh, corresponding to the Li dissolution/deposition thickness of 10 μm.

C 1s peaks for the LiI−LiNO3 electrolyte exhibit further increased intensities. The depth profiles of compositions for these Li anode foils are shown in Figure 4b,c. The Li foil in the LiBr−LiNO 3 cell presents a nearly constant high Li concentration of ∼80 at. % for regions deeper than 20 nm. Considering that small amounts of C and O are mainly due to the deposition and reactions during Ar sputtering, the Li foil appears to be quite pure except for the oxidized surface layer, with a thickness of