Towards practical Li metal batteries: Importance of separator

Subscriber access provided by UNIV OF NOTTINGHAM

Article

Towards practical Li metal batteries: Importance of separator compatibility using ionic liquid electrolytes Mojtaba Eftekharnia, Meisam Hasanpoor, Maria Forsyth, Robert Kerr, and Patrick C. Howlett ACS Appl. Energy Mater., Just Accepted Manuscript • DOI: 10.1021/acsaem.9b01175 • Publication Date (Web): 26 Aug 2019 Downloaded from pubs.acs.org on August 27, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Energy Materials

Towards practical Li metal batteries: Importance of separator compatibility using ionic liquid electrolytes Mojtaba Eftekharnia, Meisam Hasanpoor, Maria Forsyth, Robert Kerr, Patrick C. Howlett*

Institute for Frontier Materials, Deakin University, 221 Burwood Hwy, Burwood, Victoria 3126, Australia

Corresponding Author *Patrick C. Howlett. E-mail: [email protected]

ACS Paragon Plus Environment

1

ACS Applied Energy Materials 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 38

Abstract

Long-term cycling studies of high capacity Li-metal|lithium iron phosphate (LFP, 3.5 mAh/cm2) cells were carried out using two highly concentrated ionic liquid electrolytes (ILEs).

Cells

incorporating

N-propyl-N-methylpyrrolidinium bis(fluorosulfonyl)imide

(C3mpyrFSI) or triethylmethylphosphonium bis(fluorosulfonyl)imide (P1222FSI), with 50 mol% lithium bis(fluorosulfonyl)imide (LiFSI) electrolytes were shown to operate for over 180 cycles at 50 oC at a rate of C/2 (1.75 mA/cm2). The choice of separator was identified as a critical factor to enable high areal capacity cycling, with the occurrence of cell failure through a short-circuiting mechanism being particularly sensitive to separator characteristics. Several commercial separators were characterised and tested, and their performance was related to membrane properties such as the MacMullin number, pore size, and contact angle. Celgard 3000 series separators were found to support long-term cycling due to their combination of desirable nano-porosity and wettability. The most compatible cell components were assembled into a pouch cell to further demonstrate the


2

Page 3 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


feasibility of ILE incorporation into high-capacity lithium metal batteries for commercial purposes.

Keywords; Ionic liquid electrolytes, lithium metal batteries, separator, dendrites, lithium iron phosphate


3


Page 4 of 38

Introduction As society becomes more dependent on the ability to store and use energy in more demanding applications, the development of advanced energy storage technologies which are safe, durable, cost-effective and high energy density is of the highest priority. Lithium-ion batteries (LIBs) are the current technology with a strong and growing presence in stationary and mobile applications, from mobile phones to electric cars to grid-scale frequency modulation. Although LIBs enjoy the highest energy density among various commercial battery technologies, there is considerable scope to greatly increase the energy density by altering the electrode chemistries. Using Li-metal with gravimetric capacity of 3800 mAh/g as the anode instead of graphite, which has gravimetric capacity of 372 mAh/g, appears to be one of the most direct approaches to achieving a high energy-density battery.1 The safety and cycle life of Li-metal batteries, however, falls dramatically when coupled with traditional LIB organic electrolytes due to the rapid build-up of ionically resistive surface products on the lithium metal electrode which ultimately leads to capacity loss and short circuit-induced failures.2-5 Alternative electrolytes with superior safety properties are


4

Page 5 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


indeed required to bring rechargeable Li-metal batteries (LMBs) to market. Solid-state electrolytes have long promised to bridge this gap by acting as a mechanical barrier to lithium dendrite propagation while offering attractive economic benefits in fabrication speeds through reel-to-reel production methods. However, single-component polymer host network systems have so far failed to meet the power requirements of the market. The negligible volatility and non-flammability of room temperature ionic liquids (ILs) positions them as a ‘best-of-both-worlds’ candidate as either a liquid-state solvent or as one component in a solid polymer composite system.6-10 In addition to the inherent safety properties of ionic liquid electrolytes (ILEs), high-efficiency Li-metal cycling has been demonstrated. An average cycling efficiency of >99% has been achieved using an Nmethyl-N-alkylpyrrolidinium bis(trifluoromethanesulfonyl)imide ILE.11 More recently, promising Li cycling has been reported using the pyrrolidinium-based ILE, N-propyl-Nmethylpyrrolidinium bis(fluorosulfonyl)imide (C3mpyrFSI), with the demonstration of a Limetal|LiCoO2 (ca. 0.7 mAh/cm2) full cell.12, 13 This research also showed the advantages of employing a high concentration of lithium bis(fluorosulfonyl)imide (LiFSI) salt to enhance both rate performance and electrolyte stability.14 Girard et al. also achieved


5


Page 6 of 38

cycling efficiencies of >99.2% for lithium metal and stable Li|LiNi0.33Mn0.33Co0.33O2 (NMC111,

ca.

0.3

mAh/cm2)

cell

cycling

using

a

high

concentration

trimethylisobutylphosphonium bis(fluorosulfonyl)imide (P111i4FSI) ILE of 3.8m LiFSI.15-17 Similarly,

a

recent

study

on

the

Li-metal

cycling

performance

of

small

tetralkylphosphonium-FSI ILEs found that the triethylmethylphosphonium cation (P1222) was comparable in performance to P111i4, despite P1222FSI being an organic ionic plastic crystal at room temperature.18 These studies show that ILs have the potential to be incorporated into the next generation of Li-metal batteries. However, there are no instances of high capacity cycling reported to date. The next step in the demonstration of ILE capabilities is their incorporation into Li-metal cells where the cathode areal capacity approaches practical values of 3-4 mAh/cm2 (i.e., at the upper limit of current commercial LIBs). In this range, the compatibility of all battery components becomes an increasingly critical factor. For traditional LIB systems, the separator is generally considered to be a passive cell component that primarily acts to physically separate the two electrodes from coming into contact. It also plays an important role in determining some of the characteristics of the cell such as the safety, power and


6

Page 7 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


lifetime through choice of various parameters such as chemical and electrochemical stability, thickness, porosity, pore size, permeability, mechanical strength, wettability, thermal stability and shut-down ability.19-22 There are various types of commercial battery separator materials that have been specifically developed for LIBs using conventional organic electrolytes. The product range from Celgard is one of the most popular and widely used materials, based on porous polypropylene (PP) and polyethylene (PE). However, the role of the separator in high capacity Li-metal cells is poorly understood, whereby the plating and stripping of lithium metal at the anode surface forms a dynamic and continually evolving interface with the separator. Kirchhöfer et al. have conducted a detailed study on the wettability, conductivity and behaviour of four different RTILs, namely pyrrolidinium cations paired with either FSI or bis(trifluoromethanesulfonyl)imide (TFSI) anions, in combination with 8 commercial separators.23 The importance of wettability was also highlighted by Huie et al. whereby the performance of three commercial separators in combination with ILEs composed of various cations (piperidinium, pyrrolidinium, imidazolium, or pyridinium) and anions (tetrafluoroborate (BF4 or TFSI) was found to be a function of both the ILE conductivity


7


Page 8 of 38

and the contact angle.24 There is thus an opportunity to develop new separators for ILEs in order to improve on the generally limited rate performance. In a recent example, Evans at al. developed an electrospun polyacrylonitrile microfiber separator which exhibited a low MacMullin number when wet with C3mpyrFSI with 1.2 M LiFSI and also better wettability when compared to Whatman GF/F glass fibers. They demonstrated a superior compatibility of their new separator material with the ILE when compared to both a Celgard 2500 and Whatman GF/F separator in a Li-metal|NMC333 cell with areal capacity of less than 0.5 mAh/cm2.25 Although these studies address the important features of IL/separator combinations, they do not show the performance of these combinations in high capacity Li-metal full cells. Moreover, there are no studies which investigate separator interactions and wettability in the concentrated ILEs employed here. In this paper, we seek to demonstrate new high capacity Li-metal based cells and understand the role of the separator and determine the desirable characteristics for prolonged cycle life. This is achieved by studying the wettability and conductivity of four different commercial separators when wetted with a pyrrolidinium-based (C3mpyrFSI) and


8

Page 9 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


a phosphonium-based (P1222FSI) ILE and relating the separator characteristics to the cycling performance of high capacity Li|LFP cells.

Li|LFP cell cycling Two ionic liquid electrolytes which have previously shown promising Li-metal cycling, namely C3mpyrFSI26, 27 and P1222FSI18 each with 50 mol% LiFSI, were assembled into high capacity Li|LFP3.5 coin cells using four different separators; Separion P30, Whatman GF/A, Solupor 7P03A and Celgard 3501. The 3.5 mAh/cm2 LFP electrode was chosen as a high capacity cathode which generally exhibits excellent cycle stability28, hence these cells allow to cycle large quantities of Li metal charge with minimal degradation relating to cathode reactions. This is not to infer that ILEs are inherently electrochemically unstable when paired with high voltage cathodes, rather these high salt concentration ILEs may be an excellent avenue towards high voltage application due to the principle of improved electrochemical stability at high salt concentrations.29, 30 The challenges related to the high voltage stability of ILEs are akin to the corrosion and related challenges faced by organic electrolytes that are actively being addressed through the use of chemical


9


Page 10 of 38

additives and electrode coatings. The cells were cycled at 50 oC at C/2 for long-term cycling, with resultant voltage profiles shown in Figure 1. Cycling at 1C resulted in significantly lower cell discharge capacities of ~1.6 mAh/cm2 (see Figure S1). Cells with both Separion P30 and Whatman GF/A performed poorly, failing in the first conditioning cycles at C/20, independent of electrolyte (Figures 1a, 1b, 1f and 1g). In each case, the cells did not reach the cut-off voltage (3.8 V) during charge. Although the cells with Solupor 7P03A successfully cycled for the first two cycles at C/20, they failed after only a few cycles at C/2 (Figures 1c and 1h). Celgard 3501 showed the best cycling performance, maintaining good capacity retention and no evidence of failure until 180 and 120 cycles in C3mpyrFSI:LiFSI (1:1) and P1222FSI:LiFSI (1:1), respectively. A plot of the areal discharge capacity and columbic efficiency (CE) of Li-metal|C3mpyrFSI:LiFSI (1:1)|LFP3.5 with Celgard 3501 is shown in Figure 1e. This cell retained 92.7% of its capacity at C/2 with an average CE of 99.97% over 180 cycles. While 92.7% of the theoretical discharge capacity can be obtained when cycling at C/2 at 50 oC, room temperature cycling of an LFP2.0 electrode at C/2 only gives 40% of the theoretical discharge capacity (Figure S2). This emphasizes the importance of cycling temperature


10

Page 11 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


on these systems and the possibility for further improvement at higher operating temperatures. The P1222FSI electrolyte with Celgard 3501 performed similarly with 96% capacity retention and 99.95% average CE over 120 cycles (Figure 1j).


11


Page 12 of 38


12

Page 13 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 1. Li|LFP3.5 full cell cycling behaviour at 50oC using Separion P30, Whatman GF/A, Solupor 7P03A, Celgard 3501 separators in (a, b, c, d) C3mpyrFSI:LiFSI (1:1) and (f, g, h, i) P1222FSI:LiFSI (1:1), respectively. Areal discharge capacity and columbic efficiency versus cycle number for Li|LFP3.5 full cells with Celgard 3501 in (e) C3mpyrFSI:LiFSI (1:1), and (j) P1222FSI:LiFSI (1:1) To our knowledge, this represents one of the benchmark performances of a high areal capacity (3.5 mAh/cm2) Li-metal based cell which cycles an appreciable, although not yet entirely practical, fraction of the initial Li-metal anode capacity (~17%). A summary of the leading recent works for high capacity Li-metal cycling is shown in Table 1, with our results included for sake of easy comparison. Ultimately, both Celgard 3501 cells failed by a similar mechanism during the charge cycle (blue trace Figure 1d & 1i). Table 1. A summary of the most prominent recent works on high capacity Li-metal cycling.

Ref.

31

Electro -lyte

Cathode

Li t’ness (µm)

Q (mAh

No. Rate

/cm2)

cycle s

Temp (oC)

Sep’tor

0.6M LiTFSI/0.4M LiBOB/0.05M LiPF6 in EC/EMC (40:60)


13


Page 14 of 38

C/4 NMC11 1

charge, 200

4

C/2

300

30

100

N.S.

discharg

Celgar d 2500

e 1.0 M LiPF6 in EC/DEC (1:1) with 5% FEC 32

NMC53 2

N.S.

2.5

1C

N.S.

1M LiPF6 in FEC/DMC (1:4) 33

NMC62 2

PE 50

3.3

C/6

90

30

(Tonen )

1M LiFSI in 1,4-dioxane (DX)/1,2-dimethoxyethane (DME) (1:2) ENTEK

34

LFP

N.S.

1.4

N.S.

50

RT

ET 2026 type

This work

3.2 m LiFSI in C3mpyrFSI LFP

100

3.5

C/2

180

50

Celgar d 3501

*N.S. = not specified

Post mortem SEM and cell re-assembly To investigate the state of the cycled lithium electrode, cross-sectional SEM was carried out on a cycled lithium electrode from a coin cell which was stopped after 211 cycles


14

Page 15 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(discharged state) with a capacity retention of 87% of its initial capacity. The SEM image in Figure S3 clearly shows that there is a portion of the initial lithium metal substrate that remains underneath the surface lithium deposit. The average thickness of the remaining Li substrate is estimated to be around 60 µm, which confirms that total lithium consumption is not the cause of failure or capacity fade. The average overall electrode thickness appears to be greater than the initial 100 µm substrate, but this is consistent with a previous report which has shown the cycled, nano-structured lithium deposits can occupy approximately 50% greater volume than dense lithium metal.31 By only considering the reduction in Li metal substrate thickness and ignoring the presence of any active surface Li deposits, a lower estimate of the coulombic efficiency at the Li electrode can be calculated. Consumption of 40 µm of lithium metal corresponds to approximately 8 mAh/cm2 that has been consumed over the 211 cycles, or 0.038 mAh/cm2 per cycle. Taking an average cycled capacity of 3.2 mAh/cm2, this translates to a conservative coulombic efficiency estimate at the Li electrode of 98.8% per cycle. We have previously shown, using a similar approach, that these high concentration ILEs can


15


Page 16 of 38

achieve >99.2% efficiency in Li metal symmetrical cells under even more demanding cycling conditions (12 mA/cm2 and 6 mAh/cm2).35 In order to further investigate the cause of failure for the cells with Celgard 3501 separator, the failed Celgard 3501 cell with P1222FSI electrolyte (Figure 1i) was disassembled. Figure 2 shows photos of the components of the original failed cell after disassembly. Each component, i.e. Li-metal, separator and LFP cathode, was then reassembled in separate cells with other fresh components and replenished with extra electrolyte – for example, the recovered separator from the failed cell was assembled in a new cell with fresh Li-metal and cathode. No solvent was used to wash the cell components and they were used without any treatment.

Figure 2. Li|LFP full cell (P1222FSI electrolyte) components after failure (cycling history in Figure 1i). (a) Lithium anode on a spacer, and (b) Celgard 3501 separator


16

Page 17 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Voltage profiles of re-assembled cells are presented in Figure 3 and clearly show that the new cell with used separator failed quickly (2nd cycle at C/20), while the other two cells with used electrodes cycled for more than 50 cycles without any sign of failure. The capacity of the cell with cycled lithium (Figure 3a) maintained a diminished capacity of around 2.7 mAh/cm2. In this instance, the high stability of the cycling suggests that the reduction in capacity is incurred during the disassembly and reassembly procedure. The evolution of a resistive interphase layer is supported by EIS measurements taken during cycling (Figure S4), and would also explain the primary reason for the capacity fade over the first few hundred cycles observed in all cells. EIS measurements for this cell show two clear frequency-resolved cell impedance components in the form of depressed semicircles that strongly resemble the components seen previously for Li|Li symmetric cells using our electrolytes.35, 36 Given that the impedance spectra appear to be dominated by the lithium electrode (and the known stability of the LFP electrode), we can attribute the increase in impedance of each component observed during the course of cycling to a build-up of electrode-electrolyte decomposition products (eg LiF, Li2CO3).


17


Page 18 of 38

. These results indicate that, despite it being the best separator studied here, the Celgard 3501 is still susceptible to cell failure which occurs via a similar process as observed with the other separators.

Figure 3. Voltage profiles and coulombic efficiencies of re-assembled Li|LFP3.5 cells using P1222FSI:LiFSI (1:1) electrolyte and a (a) cycled Li-metal electrode, (b) cycled separator, (c) cycled LFP electrode. Further investigation into the cause of the separator-related failure mechanism was performed by SEM imaging of separator cross-sections, shown in Figure 4. The separator cross-section undergoes dramatic changes as a result of cell cycling. For Celgard 3501,


18

Page 19 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


the region around 10 μm closest to the anode is now filled with solid products – presumably decomposition products accrued during repeated cycling. In the case of Solupor 7P03A, the effect is much more pronounced, with solid products penetrating the entire thickness of the separator to reach the surface of the cathode side. The large pore size not providing any physical barrier to dendrite propagation is a likely reason for the observed short-circuiting behavior that occurs with this separator.

Figure 4. Cross-sectional SEM images of (a) Celgard 3501, and (b) Solupor 7P03A both before and after cycling (cycling history corresponds to cells in Figure 1 with C3mpyrFSI:LiFSI (1:1) ILE)


19


Page 20 of 38

MacMullin number (NM) and separator wetting The ionic conductivity of the ILE-filled separators was measured both before and after cycling in order to determine how the transport properties of the electrolyte correlate to the cell failures observed during cycling. The MacMullin number, NM, is a ratio that can be used to express the change in electrolyte ionic conductivity when a separator is introduced. A higher NM number corresponds to a larger decrease in conductivity when a separator is present. NM for a traditional carbonate-based battery electrolyte paired with a commercial battery separator such as Celgard 2500 typically falls in the range of 5-20.37 Similar values are seen for the ILEs when paired with the various separators used here. The NM after cycling can further provide an indication of whether the pores of the separator have become filled and blocked due to a build-up of lithium degradation products. The NM in Table 2 of the three separators lie within a very close range of 2-5 for C3mpyrFSI, suggesting good separator wetting and only a marginal increase in ionic impedance. After cycling, NM increased by a factor of 1.5 for Whatman GF/A, 1.6 for Solupor 7P03A, and 4.3 for Celgard 3501. It is important to consider that the Whatman and Solupor cells failed within a handful of cycles whereas the C3mpyrFSI cycled for 180


20

Page 21 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


cycles prior to failure (Figure 1). Looking at the SEM cross-section of Celgard 3501 and Solupor 7P03A separators after cycling in Figure 4, it is clear that there is a build-up of products which could reasonably explain the increase in ionic impedance. There are two notable differences when comparing the NM for the P1222FSI ILE. Firstly, there is a dramatic decrease in NM after cycling for the Whatman GF/A and Solupor 7P03A. This could be ascribed to a shortening of the effective electrode separation via the growth of active deposits into the membrane pores. However, it is not yet clear why this effect was only observed for the P1222FSI and not the C3mpyrFSI. The second major difference is the larger initial NM of 24, which is particularly surprising given that the contact

angle

is

not

significantly

different.

This

important

difference

in

electrolyte|separator interaction warrants further investigation to understand and further the design of both separator and electrolyte, where the potential benefits for device performance are obvious. A similar increase in the NM by a factor of 3.6 occurred after cycling. It is also interesting to compare the contact angles between Celgard 3501 and Solupor 7P03A. The Celgard 3501 wet completely after several minutes while the droplet


21


Page 22 of 38

remained on the surface of the Solupor 7P03A for over 1 hour (see Figure S5). These NM and contact angle measurements show that the primary cause of the short circuiting failure mechanism is mostly related to the separator pore size. Table 2. MacMullin number (NM) and contact angle of separators before and after cycling

NM

NM

Contact angle

before cycling

after cycling

(± 5o)

Separion P30

-

-

23

C3mpyrFSI:LiFSI

Whatman GF/A

2

3

0

(1:1)

Solupor 7P03A

5

8

94

Celgard 3501

3

13

77

Separion P30

-

-

14

P1222FSI:LiFSI

Whatman GF/A

3

1

0

(1:1)

Solupor 7P03A

8

0.3

99

Celgard 3501

24

86

76

Separator

Effect of cycling capacity on failure Despite having shown that Solupor 7P30A is incompatible with the ILEs and high capacity cycling, there have been previous reports which have shown stable and long-life


22

Page 23 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Li-metal cycling in ILEs. Lewandowski et al. reported the cycling performance of Limetal|C3mpyrFSI with 0.5 mol/kg LiTFSI |LFP cell with Solupor separator that cycled for over 400 cycles.38 Yoon et al. also used Solupor 7P03A in Li|C3mpyrFSI with 3.2 m LiFSI|LiCoO2 cell and reported no failure.12 In both studies, the areal capacity of the cathode was less than 0.7 mAh/cm2. To gain an understanding of how sensitive the separator-related failure mechanism is to the areal discharge capacity, the long-term cycling stability was measured using lower capacity LFP cathode (2 mAh/cm2). The cycling data shown in Figure 5 shows no improvement in full cell behaviour with Separion P30 and Whatman GF/A. However, the performance of the cell with Solupor 7P03A is notably improved whereby the areal discharge capacity is close to the nominal capacity and the cell cycled for 26 cycles. The effect of cathode capacity on cycle life is more apparent in the case of Celgard 3501, with Figure 5f showing areal discharge capacity and CE versus cycle. Here, the cell retained 94.4% of its capacity over 280 cycles with an average CE of 99.94% - 100 cycles more than the high capacity electrode. After 280 cycles, the cell cycled with poor CE for another 116 cycles at which point experiment was stopped as the capacity dropped to under 80% of its initial capacity. Thus this work shows


23


Page 24 of 38

the importance of investigating materials combinations under realistic conditions for these cells as well as determining the failure mechanism and the critical component in order to reach the higher energy density devices desired for next generation batteries.

Figure 5. Li|LFP2.0 full cell cycling behaviour at 50 oC in C3mpyrFSI:LiFSI (1:1) with (a) Separion P30, (b) Whatman GF/A, (c) Solupor 7P03A, and (d) Celgard 3501. Li|LFP2.0


24

Page 25 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


full cell Columbic efficiency and areal discharge capacity versus cycle number in C3mpyrFSI:LiFSI (1:1) with (e) Solupor 7P30A and (f) Celgard 3501.

Cell cycling experiments were performed on other Celgard separators in order to rank the separator characteristics in terms of their effects on cycling performance. According to the publicly available Celgard product information sheet, the Celgard 3xxx-series separators are identical to the 2xxx series counterparts but for the addition of a surfactant coating to improve wettability. Celgard 2500, which is structurally identical to Celgard 3500/3501, was thus chosen to directly measure the strength of the surfactant effect. Conversely, Celgard 3401 has the surfactant coating but has a smaller pore size (0.043 μm) and a lower porosity (41%) than Celgard 3501 (0.064 μm, 55%). The cycling performance of Celgard 3401, shown in Figure S6, was almost identical to Celgard 3501, while the Celgard 2500 failed to cycle any charge, indicating poor ionic connection between the electrodes. This definitively shows that the surfactant is critical in these ILE systems, while the exact role of separator pore size remains unclear.


25


Page 26 of 38

Finally, in order to evaluate the performance of high capacity Li metal|LFP battery in an ILE at more commercially relevant scales and cell formats, a Li metal|P1222FSI:LiFSI (1:1)|LFP3.5 cell with Celgard 3501 was assembled in a 15 cm2 electrode area pouch cell format. Details of the pouch cell construction and performance characteristics can be found in Table S1 and Figure S7 in the supporting information. The voltage profile and areal discharge capacity and CE versus cycle number are shown in Figures 6a and b. Stable capacity retention for 90 cycles demonstrates the applicability of these ILEs with high LiFSI content as electrolytes for practical LMBs when they are coupled with compatible components. An unstable capacity, CE profile, and early failure of these pouch cells after 90 cycles was observed. This is believed to be related to non-uniform pressure and/or current distribution over the Li-metal electrode area (Figure S8). This is an ongoing area of investigation.


26

Page 27 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 6. Li|LFP3.5 Pouch cell cycling behaviour in P1222FSI-LiFSI (1:1). (a) chargedischarge profiles, (b) columbic efficiency and areal discharge capacity.

Conclusion It has been shown that the choice of separator plays a critical role in the cycle life of high capacity lithium metal cells by determining the onset of a short-circuiting failure mechanism. The compatibility of four commercial separators, namely Celgard 3501, Solupor 7P03A, Whatman GF/A and Separion P30, was determined by cycling studies in Li|LFP3.5 cells assembled using two ILEs, namely C3mpyrFSI and P1222FSI with 50 mol% LiFSI. Long-term cycling of Li|LFP full cells was only achieved with the Celgard 3501 separator. Li metal|P1222FSI:LiFSI (1:1)|LFP3.5 cell with Celgard 3501 were also


27


Page 28 of 38

assembled in the pouch cell format and its performance once again demonstrated the feasibility of using concentrated ILEs in high capacity Li metal batteries. SEM imaging showed that short circuiting is the root cause of high capacity Li|LFP full cell failure. The MacMullin number and contact angle measurements of electrolyte-imbibed separators showed that Celgard 3501 does not show superior transport or wetting properties when compared to the Whatman GF/A. The superior performance of this type of separator in preventing the occurrence of short circuiting is attributed to its small pore size that obstructs Li dendrite growth. This finding has major implications for materials design considerations for new separators for lithium metal batteries. It is anticipated that further separator design can contribute to the realisation of practical Li metal batteries when coupled with high-safety, high stability superconcentrated ionic liquid electrolytes.

Experimental methods and materials Electrolytes were prepared by adding LiFSI (CoorsTek) to either N-propyl-Nmethylpyrrolidinium bis(fluorosulfonyl)imide (C3mpyrFSI, University of Wollongong) or triethylmethylphosphonium bis(fluorosulfonyl)imide (P1222FSI, Boron Molecular) with a


28

Page 29 of 38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


LiFSI:IL molar ratio of 1:1. The composition and structure of the synthesised ionic liquids was first confirmed using H-NMR and MS. From differential scanning calorimetry, the melting point of C3mpyrFSI was -8 oC (-9 oC in literature26 from Suzhou Fluolyte) and for P1222FSI it was 46.5 oC (47 oC in literature39). ICP-MS was used to measure the potassium content (45 ppm for C3mpyrFSI and 280 ppm for P1222FSI). The measured bromide content of C3mpyrFSI was 98 ppm using ion-selective electrode. All ILs were vacuumdried on a Schlenk line at

Towards practical Li metal batteries: Importance of separator

Recommend Documents