Attainable Gravimetric and Volumetric Energy Density of Li–S and Li

Perspective pubs.acs.org/JPCL

Attainable Gravimetric and Volumetric Energy Density of Li−S and Li Ion Battery Cells with Solid Separator-Protected Li Metal Anodes Bryan D. McCloskey* Department of Chemical and Biomolecular Engineering, University of California, Berkeley, California 94720, United States Environmental Energy Technologies Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720, United States S Supporting Information *

ABSTRACT: As a result of sulfur’s high electrochemical capacity (1675 mA h/gs), lithium−sulfur batteries have received significant attention as a potential highspecific-energy alternative to current state-of-the-art rechargeable Li ion batteries. For Li−S batteries to compete with commercially available Li ion batteries, highcapacity anodes, such as those that use Li metal, will need to be enabled to fully exploit sulfur’s high capacity. The development of Li metal anodes has focused on eliminating Coulombically inefficient and dendritic Li cycling, and to this end, an interesting direction of research is to protect Li metal by employing mechanically stiff solid-state Li+ conductors, such as garnet phase Li7La3Zr2O12 (LLZO), NASICON-type Li1+xAlxTi2−x(PO4)3 (LATP), and Li2S−P2S5 glasses (LPS), as electrode separators. Basic calculations are used to quantify useful targets for solid Li metal protective separator thickness and cost to enable Li metal batteries in general and Li−S batteries specifically. Furthermore, maximum electrolyte-to-sulfur ratios that allow Li−S batteries to compete with Li ion batteries are calculated. The results presented here suggest that controlling the complex polysulfide speciation chemistry in Li−S cells with realistic, minimal electrolyte loading presents a meaningful opportunity to develop Li−S batteries that are competitive on a specific energy basis with current state-of-the-art Li ion batteries.

A

all reported Li−S battery embodiments, a modest fraction of sulfur is not fully reduced to Li2S due to electronic conductivity limitations, partial reduction to solid Li2S2, reactions with the electrolyte, and polysulfide diffusion and reaction with the Li anode, such that the electrochemical utilization of sulfur in all Li−S cells is less than 1675 mA h/gs. Polysulfide speciation is obviously important to understand when considering solutions to these Li−S battery limitations and likely dictates the attainable sulfur utilization and sulfur loading, thereby also dictating achievable battery energy densities.

s a result of sulfur’s high theoretical specific energy and extremely low cost as a positive electrode material, Li−S batteries are currently one of many battery chemistries being explored as a potential successor to current state-of-the-art Li ion batteries.1−6 The ideal active Li−S discharge chemistry comprises Li metal oxidation at the anode and, ultimately, Li2S formation at the cathode Li → Li+ + e− (anode) 16Li+ + S8 + 16e− → 8Li 2S (cathode)

Few Li−S studies provide their employed electrolyte volume, and it is very highly recommended that this value be included in all future Li−S reports.

16Li + S8 → 8Li 2S (overall)

If sulfur were to be completely reduced to Li2S, it would have a theoretical capacity of 1675 mA h/gs. However, the equations above do not capture the full complexity of the Li−S cathode chemistry, where numerous lithium polysulfides are known to form as stable intermediates during sulfur reduction.7−11 Polysulfides are generally soluble in organic liquid electrolytes, with saturation concentrations around 10 M S (as LixS) being reported in common electrolyte solvents, such as tetrahydrofuran.12 Polysulfide formation and dissolution is in fact necessary for a Li−S battery to discharge because the electronic conductivity of elemental sulfur and Li2S is extraordinarily low;13−17 therefore, diffusion/migration of soluble polysulfides to electronically accessible portions of the cathode may afford more complete electrochemical sulfur utilization. However, in © 2015 American Chemical Society

Soluble intermediate formation is a critical difference between the active chemistry in Li−S and Li ion batteries. No soluble intermediates are necessarily formed in Li ion batteries, such that the electrolyte only supports ion Received: August 18, 2015 Accepted: October 29, 2015 Published: October 29, 2015 4581

DOI: 10.1021/acs.jpclett.5b01814 J. Phys. Chem. Lett. 2015, 6, 4581−4588

The Journal of Physical Chemistry Letters

Perspective

Figure 1. Li ion (a,b) and Li−S cells (c) studied here. The graphite anode Li ion cell is labeled LIB (C), and the Li metal anode Li ion cell is labeled LIB (Li) in the text. (d) The Li ion cathode properties used in this study. Active material properties are selected to be consistent with LiNi0.8Co0.15Al0.05O2 (NCA). Other relevant cell parameters are provided in the Supporting Information (SI), Tables S1−8. (e) The equations to calculate specific energy and energy density of all cells.

also serve to protect the Li metal from parasitic reactions involving soluble polysulfides that have diffused from the cathode. This report aims to support Eroglu et al. and Hagen et al.’s conclusions and provide further guidance on necessary E/S ratios, sulfur loading and utilization, and Li metal protective layer thickness and cost to allow Li−S batteries to compete with Li ion batteries in the high-energy, rechargeable battery market. Calculation Methods. Energy densities for Li ion and Li−S batteries are compared at cell level using Faradaic calculations outlined in Figure 1e. Differences in Li−S and Li ion pack-level energy densities should be primarily influenced by differences in their cell-level energy densities because components (thermal management, casing, etc.) required for battery pack assembly should be similar for both. In this study, the achievable specific energy (or volumetric energy density) extracted from a cell is simply calculated through the product of active material loading, active material utilization, and assumed average cell voltage, divided by the total weight (or volume) of all cell components. The cell volume is calculated in the fully discharged state for the Li ion batteries and the fully charged state in Li−S batteries, which are the likely states in which each battery would be assembled. No kinetic or transport limitations are used in these calculations; however, both directly affect active material utilization and average cell operating voltage. For the Li−S results presented here, I assume an average operating voltage of either 2.0 (realistic) or 2.2 V (optimistic) for Li−S batteries, which are consistent, or slightly above, typically observed values. Furthermore, I use sulfur utilization as an adjustable parameter and note that it is a complex function of areal sulfur loading, the sulfur-to-carbon ratio in the cathode, the E/S ratio, and the current rate.6,23 I assume in the calculations that a solid-state ion conductor will be necessary to enable Li metal anodes for Li ion and Li−S batteries and can be used as the battery electrode separator without any additional electrolyte buffer layer, as is shown in Figure 1b,c. 20% excess capacity in both Li metal and graphite anodes is used for all calculations and is based on the active material capacity/utilization. Although lower than most estimates for Li metal excess, Christensen et al. argue, and I agree, that significantly more Li metal excess than 20% implies a

conductivity between the anode and cathode. Therefore, currently available Li ion cells have been engineered to reduce the necessary amount of electrolyte, with typical cells employing significantly more active electrode material than electrolyte on a weight basis.18−20 In contrast, in Li−S battery reports where electrolyte volume loading is given, significantly more electrolyte than sulfur is employed,21−23 presumably as a result of providing an electrolyte reservoir to support appropriate polysulfide dissolution. For example, Zheng et al. suggest that an electrolyte to sulfur ratio of 20:1 provides optimal sulfur utilization and cyclability.22 However, using such a high electrolyte (μL)-to-sulfur (mg) ratio (E/S ratio) will clearly be detrimental to the attainable battery energy density. A recent study by Eroglu et al. presented a cell and pack-level technoeconomic analysis on Li−S batteries and came to the conclusion that a ∼7 mg/cm2 sulfur loading and ∼60 electrolyte vol % in the cathode should be targeted.20 They emphasized that polysulfide speciation studies in electrolytestarved cells (i.e., those that approach similar electrolyte loadings as Li ion batteries) are urgently needed. A similar result was identified by Hagen et al., where they convincingly show that to compete with the energy density of current Li ion batteries, a Li−S battery must obtain 80% sulfur utilization with cathode sulfur loadings greater than 6 mg/cm2 in cells with an E/S ratio of 3 or lower.24 Furthermore, to take advantage of sulfur’s high electrochemical capacity, the sulfur cathode will need to be coupled to a high-capacity anode. Current Li ion battery anodes are composed primarily of graphitic carbon, in which Li+ can reversibly insert/deinsert during battery operation. Graphite’s Li+ storage capacity (∼330 mA h/g) unfortunately renders it unsuitable as an anode material in a Li−S battery. To this end, significant effort has been made to enable a Li metal (with a capacity of 3860 mA h/g) anode, although no long-life, rechargeable Li-metal-based batteries are commercially available.25 An interesting current direction of research to enable Li metal anodes employs mechanically rigid solid-state Li+ conductors, such as Li7La3Zr2O12 (LLZO), Li1+xAlxTi2−x(PO4)3 (LATP), and Li2S−P2S5 glasses (LPS), which impart high ion conductivity and appropriate stiffness to suppress Li metal dendrite formation.26 In a Li−S battery, these separators would 4582

DOI: 10.1021/acs.jpclett.5b01814 J. Phys. Chem. Lett. 2015, 6, 4581−4588

The Journal of Physical Chemistry Letters

Perspective

Figure 2. (a) Volumetric energy density of a LIB (Li) and LIB (C) employing a cathode described in Figure 1d. (b) The break-even solid separator thickness necessary for a Li metal battery to compete on a volumetric (W h/Lcell) and gravimetric (W h/kgcell) energy density basis with a LIB (C) cell with similar cathode-active material loading. The curves labeled LPS, LATP, and LLZO were calculated on a gravimetric energy density basis, assuming the densities of each material provided in Table S1. LIB (C) cell properties used in the calculations are provided in Table S2. A 20% Li excess is assumed in Li metal cells.

(LPS),34 may be necessary in Li ion and Li−S cells both to suppress dendrite formation and to reduce parasitic reactions between the Li and electrolyte constituents, including soluble polysulfides formed in Li−S cells. The thickness and density of these materials will obviously affect the overall energy density of the system. To identify a target for a practical solid separator thickness, I compare the gravimetric and volumetric energy density of Li ion batteries employing graphite and protected Li metal anodes in Figure 2. The batteries compared in Figure 2 are nominally identical except for the separator and anode; a 25 μm thick electrolyte-wetted porous polymer separator is used in the graphite cell, and a solid ion-conducting separator is used in the Li metal cell, with its thickness being an adjustable parameter. For the graphite anode, I assume a 95:5 graphite/ binder wt/wt ratio with 25 vol % porosity, corresponding to an anode density of 1.65 g/cm3. Given the operating potential difference of graphite and Li metal, the average discharge potential for the LIB (Li) and LIB (C) cells was assumed to be 3.7 and 3.6 V, respectively. A Li metal anode protected by a solid separator could dramatically improve the volumetric energy density of a Li ion battery over a graphite anode-based cell. For example, using the baseline cathode NCA loading of 17 mg/cm2 (Figure 2a), a solid separator of less than ∼99 μm thickness would provide a cell with greater volumetric energy density over a conventional graphite-based Li ion cell. I define the separator “break-even” thickness (e.g., 99 μm in Figure 2a) as the maximum LIB (Li) cell separator thickness that allows the cell to attain greater energy density than a conventional LIB (C) cell. The breakeven thicknesses of LLZO-, LATP-, and LPS-based separators are compared in Figure 2b on both specific energy and volumetric energy density bases as a function of active cathode material loading. The break-even thickness was found to be a strong function of active cathode material loading (and hence areal cell capacity) as electrode capacity balancing necessarily increases the thickness of the anode, which exacerbates the volume and weight difference between the Li metal and graphite anodes. For example, at 50 mg/cm2 NCA loading, a solid separator thickness of ∼240 μm or less would provide an enhancement in volumetric energy density, whereas the breakeven thickness shown in Figure 2a for a 17 mg/cm2 cathode loading is 99 μm. Furthermore, from a specific energy perspective, the break-even thickness also depends on the

large parasitic reaction that likely will not provide a long-life battery, and therefore, I assume that parasitic Li metal reactions can be appropriately controlled through engineering advances when protecting the anode with a solid ion conductor.19 I also assume that a liquid electrolyte will be necessary to ensure lowimpedance ion transport throughout the porous cathode; therefore, a liquid electrolyte is used to wet porous cathodes in both the Li ion and Li−S battery configurations. No unoccupied volume is assumed to be present in any cells studied (i.e., complete wetting of all porous materials occurs). The baseline Li ion cathode properties are outlined in Figure 1d, with active material properties similar to those of LiNi0.8Co0.15Al0.05O2.27 These properties are selected to provide a cathode with 58 μm thickness, a total dry density of 3.07 g/ cm3, and an areal capacity of 3.1 mA h/cm2, which are consistent with values for commercial electric vehicle battery cathodes.18,28,29 Certain figures use active material loading as an adjustable parameter and will be noted accordingly. Also of note is the energy density of commercially available Li ion cells employing a graphite anode. A C/LiCoO2 (C/LCO) cell has been reported to provide actual gravimetric and volumetric energy densities of 250 W h/kg and 570 W h/L,30 and a C/ LiNi0.8Co0.15Al0.05O2 (C/LNCA) cell provides 265 W h/kg and 690 W h/L.31 The energy density values calculated using the spreadsheet in the SI for C/LCO and C/LNCA cells are in good agreement with the literature values stated above. A simple cost analysis is also performed to calculate the cell material costs per kW h. These calculations only include the cell components shown in Figure 1a−c to easily allow a direct comparison between the cost of the three cells, noting that pack-level design for Li metal batteries employing thin ceramic separators is not established. A description of the calculations, along with prices for individual cell materials, is located in the SI (Table S8). All calculations used to create the figures in this Perspective have been saved in an excel spreadsheet (SI), and the reader is encouraged to explore the effect of various adjustable cell parameters on Li−S and Li ion gravimetric and volumetric energy density. All adjustable parameters and material densities used to calculate the data presented in each figure are available in the SI (Tables S1−S8). Thickness of the Solid Separator. Solid-state ion-conducting separators, such as Li7La3Zr2O12 (LLZO),32 Li 1+x Al x Ti 2−x (PO 4 ) 3 (LATP), 33 and Li 2 S−P 2 S 5 glasses 4583

DOI: 10.1021/acs.jpclett.5b01814 J. Phys. Chem. Lett. 2015, 6, 4581−4588

The Journal of Physical Chemistry Letters

Perspective

Figure 3. An “ideal” 100 wt % sulfur cathode with 100% utilization. Cell specific energy (a) and volumetric energy density (b) as a function of areal sulfur loading. The numbers on each curve represent the electrolyte (μL)-to-sulfur (mg) (E/S) ratio. Gravimetric or volumetric energy densities of Li ion batteries employing either a graphite anode (LIB (C)) or a protected Li metal anode (LIB (Li)) are denoted as dashed lines. Cell properties used in the calculations are provided in Tables S2−S4.

Li metal anode will have to approach those of the more conventional graphite anode and liquid electrolyte-wetted porous polymer separator, as will be discussed later. Polyplus and Sion, as well as many other researchers, have made significant progress in enabling protected Li metal anodes.21,33,35 Figure 3 presents Faradaic calculations of Li−S cell energy densities when employing a 100% sulfur cathode (i.e., no carbon or binder), assuming 100% S utilization (1675 mA h/g), and at various E/S ratios. This cathode configuration is highly idealized and will likely never be achievable given the issues with S conductivity, but is studied as an optimistic comparison to a Li ion battery. As a point of reference, two Li ion batteries, both with the cathode described in Figure 1d, are shown in Figure 3; (a) LIB (Li) employs a 40 μm LATP separator with a 20% excess Li metal anode, and (b) LIB (C) employs a 25 μm thick electrolyte-wetted porous separator with a 20% excess capacity graphite anode. Importantly, as the areal sulfur loading increases, the energy density of the cell reaches an asymptotic limit at a given E/S ratio. Focusing first on the comparison between Li−S cells and LIB (C) cells, an E/S ratio of 11.1 is identified as the maximum allowable value for a Li−S cell to compete with the commercially available LIB (C) cells on a specific energy basis. That is, no Li−S configuration with an E/S ratio of greater than 11.1 will surpass the specific energy of a graphiteanode-based Li ion battery, even at infinite S loading, 100% S utilization, and a 2.2 V average discharge potential. An 11.1 E/S ratio is therefore the absolute maximum ratio recommended for future Li−S studies. Furthermore, assuming an E/S ratio of 0 and 100% utilization of a 100% S cathode allows the calculation of a lower S loading bound, which is found to be 0.7 mg/cm2, below which a Li−S battery cannot compete against a Li ion battery in terms of specific energy. When comparing the same calculations for volumetric energy density, an E/S ratio of 4.0 is the maximum allowable value, once again assuming an ideal 100% S cathode at 100% utilization. In an encouraging note, these E/S ratios (11.1 and 4.0) correspond to total sulfur concentrations of 2.8 and 7.7 M, respectively, well below the polysulfide saturation limit of known Li−S electrolytes. Of course, a 100% S cathode achieving 100% utilization is likely unattainable given the insulating properties of both S and Li2S. Calculations on a more realistic cell are therefore provided in Figure 4 (75:20:5 S/C/binder (wt) cathode, 60% S

separator’s density and hence the difference between the various separator materials. Given the rather low density and very thin format of an electrolyte-wetted polymeric porous separator, high-density separator materials (e.g., LLZO) will have to be manufactured to low thicknesses (