As a candidate for post lithium ion battery technologies, rechargeable lithium metal batteries have been recognized to be attractive as energy storage systems to consumer electronic devices, power tools, and electric vehicles. However, lithium metal batteries seem to be far away from applications to such devices, because of their poor cycling and safety issues [1]. The main obstacles are that lithium is electrochemically reactive with electrolyte and lithium deposit grows like dendrites. Due to the formation of dendrites and chemical reactions with electrolyte and lithium salt, loss of lithium and electrolyte are accompanied each cycle, resulting in low Coulombic efficiency and a poor life cycle.

The life cycle of a galvanostatic cell is determined by accumulated consequences of every cycle. Among the cycles, the initial cycle, especially the first half cycle may be the most significant because lithium electrodeposits directly onto a substrate. If the substrate is lithium-free, then it is easy to understand how lithium nucleates and grows to dendrites through ex-situ analyses, and to study what variations of the cell mainly affect the first electrodepositions. The main focus of this study with electrolytes which are different in terms of ionic conductivity and viscosity is the first half cycle. The study was performed with CR2032. The electrolyte was prepared using lithium bis (fluorosulfonyl)imide (LiFSI) and 1,2-dimethoxyethane(DME) and hydrofluoroethers(HFEs).

The electrolyte having 1M LiFSI in DME was prepared and evaluated. This electrolyte is highly conductive and watery: 16mS/cm, 2.5cP. For bare Cu and Ni substrates, electrodeposited lithium was poorly uniform: some areas were not coated. This indicates that the poor electrodeposition is a result of poor nucleation. In order to improve the nucleation, two approaches were introduced: gold sputtering over the bare substrate [2] and pulsed electrodeposition [3]. The gold particle can reduce interaction barrier energy for the electrodeposition, and the pulsed mode can boost a diffusion of Li ions onto the substrate. Using these variations, the nucleation was improved to a limited extent. From this, it was assumed that for 1M LiFSI-DME, Li ions might be fast moving to local protrusions, so that Li ions might have no enough time for steps of nucleation. So, in order to restrict Li ions movement in all directions, a viscose electrolyte, 5M LiFSI-DME, having viscosity about 64cP was tested. With this electrolyte, the nucleation and growth were significantly improved, even though its ionic conductivity was four times lower than that of 1M LiFSI-DME. The improvement was similar without a gold coating or a pulsed mode.

In order to separate the two effects of ionic conductivity and viscosity, the highly concentrated electrolyte was diluted, keeping the number of Li ions in volume to be same as for 1M LiFSI-DME. Thus, a hydrofluoroether (HFE-458) was used because it cannot solvate the lithium ions but it can be well miscible with DME. One of the diluted electrolyte is 1M LiFSI in DME-HFE-458 (20:80, v:v). Compared to 1M LiFSI-DME, the diluted electrolyte is poor in ionic conductivity while similar in viscosity: 1.15mS/cm, 3.7cP. Using the diluted electrolyte, a similar result to that of the highly concentrated electrolyte was obtained. This result indicates that Li ions might slowly drift near or onto the surface of the substrate to obtain a uniform nucleation and growth.

As shown in Figure 1, the overvoltage during electrodeposition is affected by both the anode and the electrolyte. According to <em>ex-situ</em> SEM observations, higher overpotential improves the density of electrodeposits: higher the overpotential, denser the lithium deposits are formed. This observation points that a gold coating onto Cu sheet, a highly concentrated electrolyte and a mixture of HFE-458 could support the nucleation.

Figure 2 shows that when the portion of HFE-458 increases in two mixed solvents (DME and HFE-458), then the conductivity decreases while the viscosity has a peak around 50% of HFE-458. To get probable explanations to the improvement, the electrolytes were analyzed by using Gel Permeation Chromatography. This analysis revealed that the cluster size of ions in 5M LiFSI-DME was higher than that of 1M LiFSI-DME. That is why a dilution of 5M LiFSI-DME with HFE-458 provides the smaller cluster size formation. Therefore, it was concluded that the effective charge might be quasi-neutral due to tight ions paring, so that a packet of Li ions could be preferably diffusing to the substrate, leading to the uniform nucleation and growth.

References

1. W. Xu et al., Energy Environ. Sci. 7, 513-537 (2014)
2. K. Yan et al., Nature Energy 1 16010-18 (2016)
3. Q. Li et al., Sci. Adv. 3 1701246 (2017)