Issue of Lithium-indium Anode in High Energy and Power 1 All-Solid-State Lithium Batteries

: All-solid-state lithium batteries (ASSLBs) using sulfide solid electrolytes 4 (SSEs) offer an attractive option for energy storage applications. Lithium anode is the 5 ultimate goal for ASSLBs, but lithium-indium (Li-In) alloy anode is more widely 6 utilized in lab testing owing to the quite stable interface and elimination for the risk of 7 short circuit. However, vigorous growth of Li-In dendrites in SSE is discovered in the 8 present work when a full cell (LiNbO 3 coated LiNi 0.6 Co 0.2 Mn 0.2 O 2 //Li 6 PS 5 Cl//Li-In) is 9 cycled in high loading and high rate. Our study demonstrates that Li-In anode is 10 unstable towards SSEs at high current, which induces Li-In dendrite growth enclosing 11 electrolyte particles and eventually results in cell death after a long cycling. The 12 morphology and growth mechanism of Li-In dendrites are revealed by scanning 13 transmission electron microscopy-electron energy loss spectroscopy (STEM-EELS) 14 analysis and density function theory (DFT) calculations. Moreover, the differences 15 between Li and Li-In dendrites are systematically compared.


20
The thermal instability of conventional lithium-ion batteries (LIBs) which originated 21 from the intrinsic characteristics of liquid electrolytes causes safety issues and has 22 stability towards SSEs in ASSLBs. However, most batteries were cycled at low current 1 (< 0.5 mA cm -2 , <0.5 C) 25,26,27 , it is unclear whether or not Li-In alloy anode is still 2 stable towards SSEs at high current, which is critical for the high power applications of 3 ASSLBs. There are rare investigations to clarify the issue. 4 In the present work, we cycled a full cell at high loading (4 mAh cm -2 ) and high rate 5 (1C) to investigate the interface stability between sulfide electrolyte and Li-In anode. 6 Unexpectedly, the cell has a short circuit after 897 cycles, which is similar as using Li 7 metal anode. Combined with scanning electron microscope (SEM) and scanning 8 transmission electron microscope (STEM) observations, we discovered the growth of 9 Li-In dendrites in LPSCl solid electrolyte, which leads to a rapid capacity fading and 10 subsequent battery failure. The underlying mechanism for Li-In dendrite growth is 11 revealed by electron energy loss spectroscopy (EELS) analysis and ab initio molecular 12 dynamics (AIMD) simulations. The differences between Li and Li-In dendrites in 13 morphology and growth mechanism are also compared. Li7PS6. The electrolyte LPSCl has a high ion conductivity of 2.95×10 -3 S cm -1 at room 20 temperature (RT) as measured by electrochemical impedance spectroscopy ( Figure S2). 21 The NCM622 particles were uniformly coated by LNO with around 10 nm thickness as 22 shown in Figure S3. The SEM image of the cross section of prepared Li-In alloy is 1 shown in Figure S4 with alloy phase circled by blue dotted lines. Due to the good creep 2 property of lithium, Li metal diffuses into the In metal and the formed alloy phase is 3 uniformly distributed in indium matrix. 4 Figure 1a shows the long-term cycling of the assembled cell 5 LNO@NCM622//LPSCl//Li-In at 1C at RT with a high loading of 4 mAh cm -2 . The 6 cell maintains a stable cycling capacity and near 100% columbic efficiency during the 7 charge-discharge cycle up to 890 cycles ( Figure 1a). However, the capacity started to 8 decline after 891 cycles, and finally the discharge capacity decreased to ~ 0 at the 897th 9 cycle. Figure 1b displays the related charge-discharge voltage profile from the 891 th to 10 the 897 th cycle of the cell, in which the charge specific capacity increased gradually 11 while the corresponding discharge specific capacity decreased. At the 897 th cycle, the 12 cell was continuously charged with everlasting capacity increase accompanying a lower 13 voltage increasing rate as illustrated in Figure 1c, which indicates the appearance of 14 internal short circuit and cell death. These results are similar to those of using lithium 15 anode, which indicates that Li-In alloy anode is not forever stable for sulfide electrolyte, 16 especially after a longer charge-discharge cycle with a high current density. In order to find out the reason for the cell failure using Li-In alloy anode, we 6 conducted SEM observations for the cells with different cycling numbers.  For the cell cycled 100 times shown in Figure 2b, different from the fresh state,  In alloy grows into the electrolyte for around 20 μm and exhibits flame shape at the 5 anode-electrolyte interface. We called the Li-In anode that grows into the electrolyte as 6 Li-In dendrites. For the dead cell after 897 cycles, the Li-In alloy exhibits a striking 7 growth towards electrolyte interior with around 500 μm, nearly having a contact with cathode-electrolyte interface as demonstrated in Figure S5.Therefore, the cell failure of 3 ASSLB using Li-In anode at high current is induced by the growth of Li-In dendrites.   Figure 3c, the electrolyte is broken into smaller particles with less than 4 μm diameter. can be removed via chemical reaction. Figure 3j and 3k shows the SEM images of  In alloy anode from the oblique view at low and high magnifications, respectively.

20
Obviously, Li-In dendrites grow densely and uniformly over a wide region, like a 21 honeycomb that wraps the electrolyte particles in them.

22
Based on the above results, it can be concluded that Li-In alloy is unstable towards 1 SSEs when cycled at a high current even though it exhibits excellent stability at low 2 current. The formed Li-In dendrites can penetrate the solid electrolyte after a long cycle 3 and eventually results in short circuit and cell failure. From the top to the bottom of  In dendrites, the morphology changes from flame shape, to the stripe and then to the 5 network, which has a similar evolution in time with the increase of cycling number. It 6 should be noted that striped dendrites occupy the majority and this morphology is 7 favorable for reducing the growth rate of Li-In dendrites. More importantly, these three 8 types of Li-In dendrites have one thing in common, that is they have no significant 9 destructive effect on electrolyte structure. No apparent cracks are observed even the 10 electrolyte particles were divided into smaller particles in the In-rich layer. The 11 electrolyte layer maintains a high density overall. Li-In dendrites mainly grow along 12 the grain boundaries and have a great wettability with electrolyte particles. In order to find out the reason, STEM characterization were performed to reveal the 18 growth mechanism for Li-In dendrites. Figure 4a and 4b show the STEM images of  In dendrites at low and high magnifications, respectively. It is clearly illustrated that   Evolutions of RDFs of In-S, In-Cl and P-S for LPSCl-In interface during simulation and RDFs of 3 In-S, In-Cl and P-S for reference materials (InS, InCl) with the initial structure (before MD) 4 provided as a reference. 5 Then, first-principles calculations were further performed to investigate the interface 6 reaction between metal In and LPSCl electrolyte. The dynamic changes of LPSCl-In 7 interface were simulated using AIMD at 300 K and the structural variation was tracked 8 by radial distribution function (RDF). Figure 5c shows all the formed bonds after AIMD 9 (20 ps) as well as the interface model before AIMD (0 ps). It can be clearly found that  Moreover, due to the low content of Cl in the LPSCl electrolyte, the amount of InCl is 22 much less than that of InS, which can be reflected from the number of In-S bonds and 1 In-Cl bonds in Figure 5c. Therefore, InS should be the main reaction product.

2
EELS results and AIMD simulations demonstrates that chemical side reaction occurs 3 at the In-LPSCl interface and the generation of InS is thermodynamically favorable. 4 However, whether the growth of Li-In dendrites can continue depends not only on 5 thermodynamic favorability, but also on kinetic feasibility. The growth rate of Li-In 6 dendrites is closely related to the charging-discharging current. When the cell is cycled 7 at a high current, a large amounts of lithium ions enters into the indium matrix during 8 charging, which will induce the volume expansion of dendrite tip. Obviously, grain 9 boundary is the preferential expansion channel due to the least resistance. In addition,    Figure S6. Compared with the cell using Li-In anode, it has a much shorter  Combined with the Li-In dendrites shown above, we can find that there exist 4 significant differences between Li-In and Li dendrites. Firstly, they have different 5 growth morphologies. Lithium dendrites grow vertically, perpendicular to the anode-6 electrolyte interface. Lithium-indium dendrites grow laterally in stripes, much denser 7 and much more uniform than Li dendrites. This is because the growth of Li dendrites is 8 induced by the non-uniform Li deposition that prefers to form whiskers and Li-In 9 dendrites is caused by the volume expansion and slight interface reaction. Therefore, 10 Li-In dendrites have a slower growth rate during cycling. This inspires us that 11 converting Li deposition morphology might be a novel strategy to realize the 12 application of Li metal anode. Secondly, they have different wettability with electrolyte.

13
The growth of Li dendrites causes many cracks and voids in the electrolyte due to the 14 stress concentration and high reactivity, which leads to a loose and porous electrolyte

10
In summary, we discovered the growth of Li-In dendrites in sulfide electrolyte and 11 clarified the dendrite morphology and growth mechanism using SEM-EDX analysis, Li-In dendrites, which will provide rational guidance for testing the performance of 8 sulfide-based ASSLBs.  The data that support the findings of this study are available from the corresponding 4 author upon reasonable request.