Anion-Dependent Potential Precycling Effects on Lithium Deposition

Oct 5, 2017 - Division of Chemistry, Graduate School of Humanities and Sciences, Ochanomizu University, Ohtsuka, Bunkyo-ku, Tokyo 112-8610, Japan. ‡...
0 downloads 11 Views 945KB Size
Letter pubs.acs.org/JPCL

Cite This: J. Phys. Chem. Lett. 2017, 8, 5203-5208

Anion-Dependent Potential Precycling Effects on Lithium Deposition/Dissolution Reaction Studied by an Electrochemical Quartz Crystal Microbalance Kumar Sai Smaran,† Sae Shibata,† Asami Omachi,† Ayano Ohama,† Eika Tomizawa,† and Toshihiro Kondo*,†,‡ †

Division of Chemistry, Graduate School of Humanities and Sciences, Ochanomizu University, Ohtsuka, Bunkyo-ku, Tokyo 112-8610, Japan ‡ Global Research Center for Environment and Energy Based on Nanomaterials Science (GREEN), National Institute for Materials Science (NIMS), Namiki, Tsukuba 305-0044, Japan S Supporting Information *

ABSTRACT: The electrochemical quartz crystal microbalance technique was employed to study the initial stage of the electrodeposition and dissolution of lithium utilizing three kinds of electrolyte solutions such as LiPF6, LiTFSI, or LiFSI in tetraglyme. The native-SEI (solid− electrolyte interphase) formed by a potential prescan before lithium deposition/dissolution in all three solutions. Simultaneous additional SEI (add-SEI) deposition and its dissolution with lithium deposition and dissolution, respectively, were observed in LiPF6 and LiTFSI. Conversely, the addSEI dissolution with lithium deposition and its deposition with lithium dissolution were observed in LiFSI. Additional potential precycling resulted in the accumulation of a “pre-SEI” layer over the native-SEI layer in all of the solutions. With the pre-SEI, only lithium deposition/dissolution were significantly observed in LiTFSI and LiFSI. On the basis of the potential dependences of the mass and resistance changes, the anion-dependent effects of such a pre-SEI layer presence/absence on the lithium deposition/ dissolution processes were discussed.

T

mass change (Δm), which is obtained from the resonance frequency shift (Δf), but also the resonance resistance change (ΔR), which is induced by the surface area change (ΔA) on the electrode surface and/or density and viscosity change (Δρ and Δη, respectively) in the electrolyte solution in contact with the electrode.10−17 Using this technique, Naoi et al., Katayama and Watanabe et al., Honbo and Momose, Kanamura et al., and Aurbach et al. discussed dendrite formation resulting in the ΔA change,10 Δρ and Δη in an ionic liquid,11 alloy formation,12 the HF additive effect,13 and SEI formation and lithium cycling efficiency,14−17 respectively, during the deposition/dissolution of lithium mainly in carbonates except for an ionic liquid.11 In this study, we attempted to quantitatively understand the initial stage of the electrodeposition/dissolution process of lithium utilizing G4 with three kinds of lithium electrolytes, such as LiPF6, LiTFSI (LiN(SO2CF3)2), and LiFSI (LiN(SO2CH2F)2), which are expected as electrolytes for the LIB and LAB,10,15,18−24 employing a facile EQCM approach. A “pre-SEI layer” was introduced by potential precycling on the “native-SEI layer”; then the impact of its presence/absence on the lithium deposition/dissolution was investigated. We now report the results of Δm and ΔR occurring at the electrolyte

he significantly high theoretical specific capacity of lithium (3860 mA h g−1) and its most negative reduction potential (−3.04 V vs SHE) make lithium the most desirable anode material of future lithium-based batteries.1 Nevertheless, the lack of realizing this theoretical capacity is still a distant objective, mainly due to dendrite formation. During charging of a lithium anode, dendrites, which further lead to poor performance and short circuit, form on its surface.2−5 This phenomenon was investigated in carbonates, which were used as solvents for the conventional lithium-ion battery (LIB). A similar study in tetraglyme (CH3(CH2CH2O)4CH3; G4),6 which is expected as a solvent not only for the LIB but also for next-generation lithium-based batteries, such as the lithium−air battery (LAB), because of its high flash point, has not been sufficiently carried out. In addition to dendrites, a solid− electrolyte interphase (SEI), which is composed of organic and inorganic species generated by the decomposition of the electrolyte and/or solvent, accumulates on both the anode and cathode surfaces. An ideal SEI layer on the anode is expected to have a high and low conductivity for the lithium ion and electron, respectively. In order to suppress the dendritic growth, the ideal SEI layer should be flexible for volume expansion/shrinkage on the lithium and have an atomically flat surface without any defects that participate in nuclear formation of the lithium leading to dendrite formation.7−9 The electrochemical quartz crystal microbalance (EQCM) technique, on the other hand, provides us with not only the © 2017 American Chemical Society

Received: August 30, 2017 Accepted: October 5, 2017 Published: October 5, 2017 5203

DOI: 10.1021/acs.jpclett.7b02312 J. Phys. Chem. Lett. 2017, 8, 5203−5208

Letter

The Journal of Physical Chemistry Letters

(1000 s) and 1.5 V (1500 s) were observed, which can be assigned to unavoidable trace oxygen and unavoidable moisture with anion decomposition, respectively, in all three electrolyte solutions.14 As clearly seen in Figure 1b, the cathodic current around 2.0 and 1.5 V in LiPF6-G4 was much higher than the others. Kawaguchi et al. reported that PF6− is electrochemically reduced with two electrons and three lithium ions to PF3 and produces LiF.18 Regarding the former peak, therefore, these results showed that not only the reaction of unavoidable oxygen but also the reductive reaction of PF6− take place only in LiPF6G4. It was suggested that the reduced oxygen produces LixOy by the reaction with Li+. Regarding the latter peak, on the other hand, in LiFSI-G4, it was almost negligible, suggesting that the sequence of the moisture content is LiPF6-G4 > LiTFSI-G4 > LiFSI-G4. At a potential more negative than 0.5 V (2500 s), the decomposition of anions, such as PF6−, TFSI−, and FSI−, takes place.14,23 The resultant insoluble products from the reductive decomposition of the anions were reported to be mainly LiF with PF5, Li2CO3, and POF3 from LiPF6,15,18 with Li2O, Li2SO4, Li2CO3, Li2S2O4, Li2NSO2CF3, LixC2Fy, and LiSO2CF3 from LiTFSI10,19−21,24 and with LiOH, Li2O, Li2CO3, Li2SO4, and SO2F from LiFSI.21−23 Our X-ray photoelectron spectroscopy (XPS) and infrared reflection absorption spectroscopy (IRRAS) results (Figures S1 and S2) also suggested that the majority of these byproducts exist in these native-SEI layers. In all of the solutions, both Δm and ΔR increased during the negative-going potential scan, with three distinct slopes corresponding to the cathodic current peaks around 2.0, 1.5, and 0.5 V, as already mentioned. The increase in Δm indicates that the native-SEI actually deposits on the electrode surface. From the mass per electron (MPE) values, which can be obtained from the slope plotted by Δm as a function of the charge passed in the corresponding potential region of the cathodic peak around 0.5 V, the dominant species in the nativeSEI layer was confirmed to be LiF in all of the solutions. Because both Δρ and Δη should decrease during the deposition of the decomposition species, i.e., during the native-SEI formation, the observed increase of ΔR in Figure 1d indicates that the electrode surface becomes rough associated with the native-SEI formation in all of the solutions. The sequence of the increases in Δm and ΔR during the native-SEI formation was LiTFSI-G4 > LiPF6-G4 > LiFSI-G4 and LiTFSI-G4 > LiFSI-G4 > LiPF6 -G4, respectively, suggesting that the amount and surface of the native-SEI formed in LiFSI-G4 are the lowest and slightly rougher, respectively. During the formation of the pre-SEI between 2650 and 4050 s (2 potential cycles between 0.35 and 0 V), no current peaks were observed (Figure 1b) and both Δm and ΔR gradually increased with time (Figure 1c,d), indicating that the pre-SEI forms on the native-SEI covered electrode surface with an increase in the surface roughness. Due to the following two reasons, we considered that the components of the pre-SEI layer are almost the same as those of the native-SEI layer: (i) no current peaks were observed during the pre-SEI formation and (ii) no clear difference was observed in the XPS and IRRAS between the native-SEI and pre-SEI (Figures S1 and S2). On the basis of the Δm and ΔR values during the pre-SEI formation (Figure 1c,d), the order of the amounts and roughness of the formed pre-SEI layer in the three solutions was LiTFSI-G4 > LiFSI-G4 > LiPF6-G4 and LiPF6-G4 > LiTFSI-G4 > LiFSI-G4, respectively. Both orders of the amounts and surface roughness for the pre-SEI samples,

solution/pre-SEI and/or native-SEI layers/electrode interface during the lithium deposition/dissolution processes. Before the cyclic voltammetric (CV) measurements during the lithium deposition/dissolution, we prepared two kinds of samples (see experimental details in the Supporting Information). One sample was prepared by potential scan of a copper EQCM electrode from 3.0 V (vs Li/Li+), which is close to the open-circuit potential (OCP) of ca. 3.2 V, to 0.35 V. In this case, only the native-SEI formed on the anode surface before lithium deposition, and we labeled this sample as “without preSEI”. The other sample was prepared by continuous potential cycling between 0.35 and 0 V after preparation of the nativeSEI, and this sample was labeled as “with pre-SEI”. Figure 1 shows the time dependences of the potential, current, Δm, and ΔR measured in 1 M LiPF6 in G4 (LiPF6-

Figure 1. Time dependences of the (a) potential (black), (b) current density, (c) Δm, and (d) ΔR, measured in 1 M LiPF6-G4 (red), LiTFSI-G4 (blue), and LiFSI-G4 (green) with a scan rate of 1 mV s−1. Points A and B represent starting points of CV scans (Figure 2) with and without pre-SEI, respectively.

G4), 1 M LiTFSI in G4 (LiTFSI-G4), and 1 M LiFSI in G4 (LiFSI-G4), during the preparation of both the with and without pre-SEI samples. During the negative-going potential scan from 3.0 to 0.35 V, corresponding to a time dependence from 0 to 2650 s in Figure 1a, cathodic peaks around 2.0 V 5204

DOI: 10.1021/acs.jpclett.7b02312 J. Phys. Chem. Lett. 2017, 8, 5203−5208

Letter

The Journal of Physical Chemistry Letters

Figure 2. CVs and potential dependences of Δm and ΔR with (red) and without (blue) pre-SEI in the first scan, measured in (a) LiPF6 in G4, (b) LiTFSI in G4, and (c) LiFSI in G4 with a scan rate of 1 mV s−1.

Figure 3. Δm values as a function of anodic (blue) and cathodic (red) charge in (a,b) LiPF6-G4, (c,d) LiTFSI-G4, and (e,f) LiFSI-G4. (a,c,e) Without the pre-SEI; (b,d,f) with the pre-SEI.

however, were LiTFSI-G4 > LiFSI-G4 > LiPF6-G4, indicating that the surface with the pre-SEI prepared in LiPF6-G4 and LiFSI-G4 is the roughest and smoothest, respectively. Figure 2 shows the CVs and potential dependences of Δm and ΔR of the first scan with and without the pre-SEI measured in the three electrolyte solutions. In all of the solutions both

with and without the pre-SEI, the cathodic and anodic currents due to lithium deposition and dissolution, respectively, started to flow at ca. 0.05 V in the negative-going scan and at ca. 0 V in the positive-going scan, respectively. The current densities due to the lithium deposition/dissolution with the pre-SEI were slightly lower than those without the pre-SEI. Because it is 5205

DOI: 10.1021/acs.jpclett.7b02312 J. Phys. Chem. Lett. 2017, 8, 5203−5208

Letter

The Journal of Physical Chemistry Letters Table 1. MPE Values Obtained from CVs in Five Cycles LiPF6-G4 without pre-SEI b

cycle number

depo

1

57.6 41.1

2

35.0 19.8 22.8 16.5 20.6 15.7 16.2 19.9

3 4 5

dis

b

20.2 34.9 26.5 (7.8) 19.5 9.6 16.3 9.7 18.6 9.1 15.4

LiTFSI-G4 with pre-SEI b

dis

depo

20.9 31.5

13.9 27.9

11.7 15.1 11.4 15.4 11.6 18.6 11.6 19.7

8.2

98.0 69.8 41.3 63.5 24.3 47.0 16.1 57.0 17.1 44.5 17.4

depo

b

without pre-SEI

9.0 10.1 10.6

b

LiFSI-G4 with pre-SEI

b

depo

29.6 46.5 9.5 19.8 8.4 15.4 8.3 16.0 8.0 17.7

dis

b

without pre-SEI b

depo

disb

8.2

8.0

5.9

3.7

10.8

9.0

6.2

5.3

4.2

3.2

6.1

5.6

5.2

5.8

3.8

2.7

5.3

5.5

5.1

5.8

3.6

2.9

5.1

5.7

6.1

6.5

3.2

2.8

4.6

5.5

dis

b

with pre-SEI depob

dis

b

a

The MPE values close to 7 (between 5 and 9), much higher than 7 (>9), and much lower than 7 ( 9): The add-SEI formation and its dissolution take place along with lithium deposition and dissolution. (c) MPE ≪ 7 (i.e., < 5): Aurbach et al. interpreted an MPE lower than 7 as follows.16 During lithium deposition, (i) the capacitive charge consumption is not negligible as

compared to the charge transfer that leads to an increase in Δm, or (ii) lithium deposition leads to massive dendrite formation. During lithium dissolution, the electrode is poorly passivated, and then massive reduction (decomposition) of electrolytes takes place, leading to a Δm increase, which counter balances the possible mass decrease due to lithium dissolution. Multiple MPE values reflect more than one distinct electrochemical process. On the basis of this knowledge, we interpreted the present results as described below. Multiple/ higher MPE values, especially in the early scans, without the pre-SEI in LiPF6-G4 and LiTFSI-G4 and with the pre-SEI in LiPF6-G4 indicated deposition/dissolution of the add-SEI layer besides lithium due to the undesirable side reactions. In these cases, relatively high ΔR changes (Figure 3) were observed as a result of the lithium dendrite (lithium nucleation) formation, leading to formation of the add-SEI. On the contrary, anomalously low MPE values were obtained without the preSEI in LiFSI-G4 for all cycles except for the first deposition, in which the MPE is close to 7. It indicates that only lithium is electrodeposited on the electrode without any add-SEI formation in the first deposition. This result shows that the native-SEI formed in LiFSI-G4 is not thick enough to fully cover the electrode surface and that the FSI− is more electrochemically stable than the other anions on the copper electrode. During the first dissolution, however, FSI− is directly decomposed by the deposited lithium,25−27 which is only covered with a relatively thin native-SEI and/or add-SEI layers, as Aurbach et al. reported.16 This indicates that during the dissolution the deposited lithium is not completely dissolved and the add-SEI layer forms on the lithium, leading to a low MPE (