Current status and future perspectives of lithium metal batteries
Helmholtz Institute Ulm (HIU), Helmholtzstraße 11, 89081, Ulm, Germany
Karlsruhe Institute of Technology (KIT), P.O.Box 3640, 76021, Karlsruhe, Germany
BMW Group, Petuelring 130, 80788, München, Germany
Department of Sustainable Energy Technology, SINTEF Industry, 7034, Trondheim, Norway
Electrochemical Innovation Lab, Department of Chemical Engineering, University College London, Torrington Place, London, WC1E 7JE, United Kingdom
Department of Chemical and Pharmaceutical Sciences, University of Ferrara, Via Fossato di Mortara 17, 44121, Ferrara, Italy
Fraunhofer Institute for Material and Beam Technology (IWS), Winterbergstraße 28, 01277, Dresden, Germany
Technische Universität Dresden (TUD), Bergstrasse 66, 01069, Dresden, Germany
Institute for Chemistry and Technology of Materials, Graz University of Technology, Stremayrgasse 9, 8010, Graz, Austria
Department of Information Technology and Electrical Engineering, ETH Zürich, Gloriastrasse 35, 8092, Zürich, Switzerland
IST Austria (Institute of Science and Technology Austria), Am Campus 1, 3400, Klosterneuburg, Austria
Received 15 June 2020, Revised 31 July 2020, Accepted 14 August 2020, Available online 9 September 2020.
The historical development of lithium metal batteries is briefly introduced.
General strategies for protection of Li metal anodes are reviewed.
Specific challenges of ASSBs, Li–S and Li-air batteries are extensively discussed.
Current development status is reviewed and compared to the EU SET Plan targets.
With the lithium-ion technology approaching its intrinsic limit with graphite-based anodes, Li metal is recently receiving renewed interest from the battery community as potential high capacity anode for next-generation rechargeable batteries. In this focus paper, we review the main advances in this field since the first attempts in the mid-1970s. Strategies for enabling reversible cycling and avoiding dendrite growth are thoroughly discussed, including specific applications in all-solid-state (inorganic and polymeric), Lithium–Sulfur (Li–S) and Lithium-O2 (air) batteries. A particular attention is paid to recent developments of these battery technologies and their current state with respect to the 2030 targets of the EU Integrated Strategic Energy Technology Plan (SET-Plan) Action 7.
1. – The “holy grail” Li anode: brief history, early failures and future targets of rechargeable Li-metal batteries
Since the mid-20th century, metallic Li has been of high interest for high energy density batteries. In particular, its high theoretical gravimetric capacity of 3861 mAh g−1, and the most negative standard reduction potential (−3.040 V vs. standard hydrogen electrode, SHE) render Li an attractive anode material [1,2]. The historical development of Lithium Metal Batteries (LMBs) has already been extensively covered by several recent reviews [, , ] and goes beyond the aim of this paper. Nevertheless, it is worth highlighting a few key events that determined the development of this field.
Following the pioneering work done in the late 60s and early 70s by Rüdorff, Rouxel, and co-workers on the intercalation of alkali metals in transition metal di-chalcogenides , it was Whittingham in 1976 (who was then working at Exxon) to patent the first rechargeable Li/TiS2 rechargeable chemistry . In the following years, several cathode materials have been proposed in combination with Li metal, including transition metal oxides (V2O5, V6O13) and metal selenides (NbSe3) . In the late 80s, the Canadian Moli Energy succeeded with commercializing the first rechargeable LMBs based on a molybdenum sulfide (MoS2) cathode [8,9]. Unfortunately, millions of sold cells had to soon be recalled due to frequent fire accidents . In fact, while potentially providing high gravimetric energy, the low standard reduction potential of Li lies well outside the stability window of most liquid organic electrolytes . The electrolyte is therefore reduced by the Li metal, leading to the formation of a Solid Electrolyte Interphase (SEI) [, , ]. Due to newly forming the full volume of hostless lithium during charge (i.e., Li plating) the SEI can rupture and fresh lithium is continuously exposed. The fresh lithium consumes electrolyte, deteriorates coulombic efficiency, and increases cell impedance due to the increase in SEI thickness . The ruptured SEI also provides an inhomogeneous surface during lithium plating, eventually resulting in dead lithium and dendrite formation. Sand equation states that the time for lithium dendrite formation is inverse proportional to the current density. Hence, a homogeneous distribution of the current is crucial to balance space-charge and to avoid local electric field build-up. Depending on the applied current density, dendrites either form as mossy dendrites (high current density) or needle-like dendrites (low current density) . The latter are more likely to penetrate the separator and contact the cathode, leading to short-circuit and thermal runaway, i.e., uncontrollable exothermal reactions between the cells components, raising the cell temperature and forming highly flammable and toxic gases. The temperature increase in turn increases the reaction rate, speeding up the gas formation. Eventually the internal cell pressure leads to explosion and ignition [16,17]. This brought the safety issues of recharging LMBs to the public attention, driving the development of the much safer carbon anode, which finally resulted on what is nowadays known as the Li-ion battery (LIB) [7,18,19]. Despite the incredible commercial success of LIBs having initially set aside the development of rechargeable batteries with Li metal anodes, the topic has recently experiencing a renewed interest motivated by Li-ion technology approaching its limit. Meanwhile, the academic interest in LMBs has never waned and the understanding of beyond Li-ion systems, such as, for example, Lithium–Sulfur (Li–S) and Li–O2 batteries, has substantially advanced in the past decade [20,21]. While for Li–O2 systems many fundamental questions remain unanswered, the practical development of Li–S cells has already reached a relatively high TRL. In fact, OXIS Energy (UK) has been developing Li–S prototypes with a capacity ranging from 10 to 35 Ah, currently reaching a specific energy up to 400 Wh kg−1, which has been stated to increase shortly to 500 Wh kg−1 . OXIS Energy and Codemge recently signed a lease agreement to build the world's first Li–S manufacturing plant . In addition, plans to build Li–S batteries gigafactories in Norway are underway .
Currently, substantial efforts are made to finally benefit from the advantages of Li metal anodes in commercial rechargeable cells, especially for electric vehicles (EV) applications. As depicted in Fig. 1 , several R&D programs have been launched worldwide to accelerate this transition. Some of the most ambitious examples are the “Battery 500” (USA), “Made in China 2025” (China), and “RISING II” (Japan) [25,26]. Also in Europe, batteries are included among the key clean energy technologies of the Integrated Strategic Energy Technology Plan (SET-Plan) Action 7 [27,28]. To become competitive in the battery sector, very ambitious targets have been set for performance (energy, power and lifetime), cost, and manufacturing volume . In terms of battery chemistries, the transition to LMBs (i.e., Generation 4: all-solid-state with lithium metal; and Generation 5: Li–S and Li–O2)  is planned starting from 2025 . Overall, independently from the timeframe, it is clear that all programmes aim to reach the same target of 500 Wh kg−1. Certainly, large efforts are required to overcome the still existing challenges associated with the use of Li metal. This review comprehensively covers all these aspects.
Fig. 1. Comparison of roadmaps and targets of different R&D programs worldwide. Evolution of battery chemistry is also depicted. Plot modified from the Battery 2030+ Roadmap . Some of the data originally provided by Hong Li et al. .
2. The challenge of stabilizing Li metal anodes: general strategies
As recently discussed by Cui et al. , among all challenges identified in the past decades, two main issues need to be addressed to enable Li metal anodes: (i) the formation/disappearance of the full volume, and (ii) the high chemical reactivity.
Regarding volumetric changes, the morphology of the anode is key. Pristine Li metal foil is soft, ductile and both a good electronic and ionic conductor. Such features justify its traditional use in form of thin foil, without needing a current collector. However, a thickness change of tens of μm results from applying cathodes with practical capacities >3 mAh cm−2. To mitigate the Li interface movement during cycling, Li powder has recently been considered as alternative. Li powder particles (~20 μm in diameter) compacted into a round disc (15 MPa, Ø 15 mm) contain roughly 4.5 times the surface area of a lithium metal foil disc of the same diameter . According to the Sand equation, the increased surface area reduces the current density on the lithium surface, slowing down dendrite growth . Additionally, the porous structure can accommodate part of the volume changes upon charge/discharge in the pore volume of the electrode . However, lithium powder electrodes have significant disadvantages compared to foils as they are not freestanding and need a substrate, usually Cu-foil. The porosity of the powder electrode allows contact between the Cu and liquid electrolyte, resulting in galvanostatic corrosion (spontaneous lithium dissolution at the Cu/Li interface) . A similar effect has been seen at the Li/electrolyte interface, resulting in pits and voids. Both dissolution effects form “dead” lithium and deteriorate the lithium electrode, causing premature cell death . A solid electrolyte instead may reduce the lithium dissolution at the Cu/Li interface, but causes issues at the lithium/electrolyte interface, discussed in detail later in section 3.1.2 .
The very low standard reduction potential of lithium is the root of its high reactivity. Even when stored under inert conditions, i.e., under argon, lithium readily reacts with trace residual atmospheric gases, resulting in a surface passivating layer . This so-called “native SEI” consists mostly of Li2O, LiOH and Li2CO3. While it enables handling of lithium metal in dry room conditions, its composition and morphology, can be influenced by production and storage conditions and is difficult to control. Meyerson et al. analysed the surface composition of a native SEI and determined a mostly inorganic surface (Li2O and Li2CO3) with organic rich veins . The inorganic sections were shown to be less reactive than the organic rich veins. Schmitz et al. additionally found Li3N and Li2C2 when analysing the native SEI, yet their work does not mention distinct morphological differences . Once the lithium electrode is exposed to the electrolyte, a “secondary SEI” forms on top of the electrode. The presence of the native SEI, and its influence on the secondary one, is often neglected in literature. This complicates a thorough understanding of the Li surface and the development of suitable surface protection strategies.
To tackle the challenges associated with lithium metal, two main approaches have been considered, as shown in Fig. 2 The first is to stabilize the lithium metal in the liquid electrolyte via a suitable SEI . The SEI requires similar properties to that applied in state-of-the-art LIBs regarding high ionic conductivity, being electronically insulating and chemically stability [38,9]. Due to the much larger volumetric changes of lithium metal compared to the graphite anode, substantially higher mechanical stability is needed. Possible SEI formation routes include: (i) electrochemical SEI formation (“in-situ” SEI) via a properly chosen electrolyte (solvent/salt/additive combination) and (ii) an artificial SEI (“ex-situ”) produced prior to cell assembly. The second approach is applying a solid instead of liquid electrolyte . The high mechanical strength of solid electrolytes, either polymeric or inorganic, should suppress dendrite growth, therefore prolonging cycle life. Additionally, solid electrolytes improve the overall cell safety. Unlike liquid organic electrolytes, they are not flammable. Yet, solid electrolytes tend to have additional issues, discussed later in section 3. Of course, a number of hybrid electrolytes resulting from the combination of these two main classes (liquid and solids) could also be employed in LMBs. As reviewed by Keller et al., possible hybridization approaches include gel polymer (liquid/polymers), quasi-solid (liquid/inorganic) and solid (polymer/inorganic) hybrid electrolytes . Nevertheless, for sake of brevity, in this section we will focus on general strategies to enable Li metal electrodes, solely in liquid cells.
Fig. 2. Schematic drawing showing the main stabilization routes for lithium metal in liquid and all-solid-state battery cells. For liquid cells, lithium metal can be stabilized with a host structure, “in-situ” SEI or “ex-situ” artificial SEI. All-solid-state cells can either use an inorganic or polymeric solid electrolyte to stabilize the lithium metal anode.
2.1. In-situ SEI with additives/electrolyte
Understanding the SEI formation process has led to thorough research towards electrolyte optimization, to derive decomposition products desirable for the SEI. Galvanostatic corrosion (spontaneous lithium dissolution at the Li/electrolyte interface) is the main driving force in the SEI formation process . Without a passivating additive the lithium dissolution at the Li/electrolyte interface will result in pits and voids, causing the formation of “dead” lithium and deterioration of the lithium electrode . Therefore, electrolyte additives have gained great interest. The formation of a SEI via electrolyte additives will initially consume some lithium of the electrode. However, this consumption is limited and will cease once the lithium electrode surface is sufficiently covered with the desired SEI. Electrolyte additives are usually divided into two main groups, reduction type and reaction type additives (Fig. 3a–i) . Reduction type additives, have a relatively high redox potential and are reduced prior to the electrolyte depletion. Their decomposition products form an insoluble film, protecting the electrode/electrolyte interface. Reduction type additives are divided into two subgroups. The first subgroup consists of reactive compounds containing an unsaturated carbon bond. These reactive monomers form an electrochemically stable and organic rich polymer layer, upon electrochemical reduction at ~0.9 V vs Li/Li+. This group of additives contains, amongst others, vinylene carbonate (VC) [42,43], fluoroethylene carbonate (FEC) , vinylene ethylene carbonate [45,46], methyl cinnamate , vinyl-containing silane-based compounds , and furan derivates . The polymerization of vinylene carbonate (VC) occurs at the carbon-carbon double bond (C
Fig. 3. Overview of the main stabilization methods for lithium metal anodes in liquid electrolyte. a)”in-situ” SEI, b) “ex-situ” artificial SEI and c) host structures. a) “in-situ” SEIs can be tailored via i) electrolyte additives or ii) ionic liquids. b) “ex-situ” artificial SEIs can be produced by i) atomic layer deposition, ii) gassing, iii) dip-coating or iv) cutting of lithium in a precursor solution. c) stabilizing host structures can consist of i) a carbon-sphere thin film, ii) a h-BN/graphene thin film, iii) hollow carbon nanospheres, iv) an ultrafine lithium seed layer or v) seeded carbon nanowires.
Organic and inorganic hybrid SEIs have been also developed utilizing metal halides as electrolyte additives [60,61]. AlI3, for example, is able to stabilize the lithium metal anode surface by a multi-step, synergistic reaction. The initial reduction of the AlI3 salt leads to the formation of a stable LiI layer on top of the lithium metal surface, reducing the activation barrier for Li+ transport across the electrode/electrolyte interphase. Additionally, aluminum metal will form the previously mentioned intermetallic alloy phase, suppressing dendrite growth. Finally Al3+, a strong Lewis acid, is an excellent initiator of the 1,3-Dioxolane (DOL) polymerization, producing a thin, protective, polymeric film on the lithium metal surface. The polymeric film protects from further unwanted side-reactions with the electrolyte, while maintaining a high Li+ conductivity.
LiAsF6 has also been investigated as lithium salt additive for organic carbonate based electrolytes . It is reduced in the electrolyte, forming a LixAs alloy phase and LiF on the lithium anode, positively affecting lithium deposition and the surface morphology . Overall, halogenated lithium salt additives are beneficial for improving long-term cyclability of LMBs. Lithium halides (LiF, LiBr and LiI) suppress dendrite formation. Even without good salt solubility, the anions (F−, Br− and I−) adsorb on the lithium surface and enhance the surface mobility of lithium ions [64,65]. Since halide salts cannot be reduced any further, they reduce or prevent reactions of lithium with other electrolyte components.
Ionic liquids (IL) have also been investigated as SEI precursors, yet many ionic liquids are not stable towards lithium metal (Fig. 3a–ii) [66,67]. Generally, ionic liquids are reduced at a more positive potential with respect to the potential of lithium plating. Adding a lithium salt, such as LiBF4, LiPF6 and LiTFSI to an IL is beneficial. By using either the FSI− or TFSI− anion, the stability window of the electrolyte is extended and it can be combined with lithium metal [66,67]. Since ionic liquids do not contain solvents, the anion plays the deciding role in the SEI formation and can be tailored accordingly. In the case of LiFSI-IL, the SEI consists of LiF, Li2O, LiOH and FSI− decomposition products . Once the cell is cycled, additional species associated with the cation are present.
Another example of safe electrolyte worth to be mentioned is the 1.2 M LiFSI in a mixture of triethyl phosphate (TEP) and bis(2,2,2-trifluoroethyl) ether (BTFE) reported by Chen at al . Besides being non-flammable, it produces a much thinner and dense SEI on Li metal compared to conventional carbonates, thus mitigating its continuous corrosion, which results in less surface being available for SEI formation and other parasitic reactions. As shown by Niu et al. , when employed in a 1 Ah Li|NMC622 pouch cell a gravimetric energy of 300 Wh kg−1, this electrolyte substantially mitigates cell swelling under applied external pressure.
2.2. Artificial SEI
As a measure to prevent dendrite formation and ensure long-term cycling stability artificial SEIs have been of particular interest. The artificial SEI is the passivate layer formed on top of the lithium metal anode before coming into contact with the electrolyte (Fig. 3b). Depending on the processing method, the artificial SEI forms on top of pristine lithium or the native SEI. Stabilizing the anode surface before cycling allows the regulation of the SEI considering the thickness, homogeneity and conformity. Artificial SEIs specifically for lithium metal electrodes are often formed by atomic-layer deposition, aeration or coating in a liquid [, , ].
Atomic layer deposition (ALD) is an advanced thin-film fabrication technique, producing homogenous, conform, and ultra-thin films at temperatures below the melting point of lithium (Fig. 3b–i) . The surface film needs to be as thin as possible to preserve high ionic conductivity, but be thick enough to protect the lithium metal surface. ALD films based on Al2O3 result in the lithiation of Al2O3 and the formation of a stable, ionically conductive LixAl2O3 alloy layer . According to Qin et al. the lithiation degree of a lithium aluminate layer increases upon consecutive cycling, which may be beneficial to guarantee a more homogeneous Li diffusion. Ultimately, it cannot be excluded that a Li–Al alloy is also formed . Kozen et al. showed that a 14 nm thick film only contains the LixAl2O3 alloy phase in the 6 nm closest to the lithium metal surface. The top 8 nm consist of Al2O3 and undergo lithiation upon cycling, resulting in a pure LixAl2O3 alloy layer . Combined with a sulfidic solid electrolyte the ALD Al2O3 protective layer prevents self-discharge during the rest period and reduces capacity loss by 40% after 100 cycles . A subsequent study by Kazyak et al. showed the beneficial effect of a significantly thinner ALD Al2O3 film of only 2–3 nm . This film was beneficial for suppressing dendrite propagation and doubled the lifetime of lithium metal electrodes before short-circuiting. Despite the reduction of the Al concentration on the lithium metal surface, the more homogenous current distribution on the surface reduces dendrite growth significantly. Chen et al. used low temperature ALD (150 °C) to produce a homogenous, high purity (>99%) LiF film on top of a lithium metal surface. The LiF layer thickness is tailored by increasing its thickness by 0.8 Å per ALD cycle. Its high shear modulus (58 GPa) suppressed dendrite growth and increased cycle life by four times in comparison to uncoated lithium electrodes, whilst showing a high Coulombic efficiency of 99.5% .
Another method of creating an artificial SEI is via reaction of lithium metal with gaseous species (Fig. 3b–ii). The treatment with N2 at room temperature results in a stable and dense Li3N protective film . Wu et al. produced a highly conductive Li3N layer with a thickness of 159 nm. The protective layer effectively prevents side reactions between lithium metal and the electrolyte whilst Li3N, due to its high lithium ion conductivity, provides barely any resistance towards lithium ion mitigation . After 100 cycles the passivating layer is still stable and without cracks. Importantly, the exposure time of lithium to N2 is the deciding factor towards performance and stability of the passivating Li3N film. Alternatively, CO2 has been used to passivate the lithium metal surface. Lithium exposure to a CO2 atmosphere at room temperature leads to the electrode being coated with a Li2CO3 layer . The protective layer improved the ionic conductivity and resistance compared to the native SEI on lithium. For the Li2CO3 layer formation, the native SEI has to be removed from the lithium surface via mechanical brushing. Without this step, the surface film would be dominated by Li2O, resulting in reduced ionic conductivity. The high lithium ion exchange rate for Li2CO3 is based on the charge centre in the carbonate shifting from one oxygen atom to another, due to orbital interaction and charge delocalization .Due to low ionic resistance, the Li2CO3 layer itself is relatively stable and withstands high current densities of 20 mA cm−2 without cracking . Sulfur gas has been also used to produce a stable Li2S layer on lithium metal electrodes . The gas phase reaction at elevated temperature (170 °C) forms a homogenous and conductive layer. Due to its certain ionic conductivity (10−5 S cm−1), the Li2S layer can mitigate inhomogeneous lithium ion flux. Upon cycling the artificial SEI preserves its protective function by converting into a layered SEI, containing RCO2Li, Li2CO3, sulfonates and a Li2S/Li2S2 mixture. The Li2S protective triples the cycle life compared to unprotected lithium at 2 mA cm−2.
Additionally, an artificial SEI can be fabricated by exposing lithium metal to selected liquid chemicals. One method is dip-coating lithium metal in appropriate SEI precursors as initially proposed by Schechter et al.  (Fig. 3b–iii). For example, dip-coating with polyphosphoric acid solution (0.4 wt% in DMSO) leads to the formation of an artificial Li3PO4 SEI layer . This method replaces the native SEI on the lithium surface with a uniform Li3PO4 SEI, showing excellent chemical stability, a high Young's modulus (10–11 GPa) and high lithium ion conductivity. Dip-coating lithium metal in a metal chloride solution (MClx in THF, M = In, As, Bi, Zn) forms a LixMy alloy phase on the lithium surface . This method utilizes the high lithium ion conductivity of the alloy phase and lithium ion from the underlying lithium metal. The formation of electronically insulating LiCl compensated the bulk alloy layer being electronically conductive, by establishing an electric field across the surface film, driving lithium mitigating through the protective layer. The layer prevents lithium reduction on the surface and suppresses dendrite growth sufficiently, allowing stable cycling at high current densities (2 mA cm−2) for up to 1000 h. Using a dip-coating procedure to fabricate the artificial SEI has one major drawback: it produces the artificial SEI on top of the native SEI, making it difficult to unambiguously assign electrochemical properties. Furthermore, the composition of the native SEI depends on the lithium provider and storage conditions and can vary between lithium batches. Cutting the lithium directly in the precursor solution ensures the artificial SEI being produced on top of pristine lithium and enables improved investigation of the artificial SEI  (Fig. 3b–iv). This method was developed and used by Ding et al. to form a protective layer based on 1-pentylamine in pentane ......