Recent advances in the mitigation of dendrites in lithium-metal batteries Free

Since their inception in 1859 as lead acid batteries, secondary or rechargeable batteries have evolved a great deal over the years [Fig. 1(a)]. Lead acid batteries have been ubiquitous for the start-up of electric vehicles due to their low cost. However, their poor cycle life and low energy density of ∼30 Wh kg⁻¹ deems them unsuitable for replacing gasoline engines in electric vehicles. Nickel–cadmium batteries (discovered in 1899) provide a specific energy density of ∼50 Wh kg⁻¹ and are capable of fast charging and discharging (power density) while providing a much better cycle life than that of lead acid batteries. However, the presence of cadmium in these batteries raised environmental concerns, which led to the development of nickel-metal hydride batteries, where cadmium was replaced with special alloys. Hybrid electric vehicles such as the Toyota Prius commonly use such a battery system.

The advent of lithium (Li)-ion battery (LIB) technology in the early 1990s has enabled the mass-market adoption of portable electronics.¹ The emergence of electric vehicles (EVs) and renewable energy technologies which depend on LIBs as the main means of energy storage has led to a paradigm shift toward sustainable energy. State-of-the-art LIBs possess an energy density of about 250 Wh Kg⁻¹, which is still an order of magnitude lower than that of gasoline² [Fig. 1(b)]. A traditional LIB is based on the rocking-chair concept, where Li⁺ ions shuttle back and forth between the two electrodes that act as hosts for the ions on charge and discharge [Fig. 1(c)]. Such a system has been successful and exhibits longevity of up to 10 000 cycles but is limited by its relatively low energy density when compared to gasoline powered combustion engines. Thus, rapid efforts to develop secondary battery chemistries beyond Li-ion are currently in progress.^3–8,13,14

Graphite, traditionally used as the anode in a LIB, is known for its cheap cost and high cycle life but limited by its low specific capacity of 372 mAh g⁻¹. Lithium metal, possessing the highest theoretical capacity of 3860 mAh g⁻¹ and the lowest electrochemical potential (−3.04 V vs standard hydrogen electrode), is the ideal choice for the anode in a Li battery.^9,10 Utilizing Li metal as the anode can enable lithium metal battery (LMB) [Fig. 1(d)] systems to achieve energy densities of up to 1150 Wh Kg⁻¹,^11,12 inching toward that of gasoline. However, lithium metal anodes are plagued by extensive nucleation and growth of dendrites upon electrochemical cycling.^15–18 This dendritic growth of Li is associated with irreversible capacity loss, drying/degradation of the electrolyte, reduction of coulombic efficiency^19,20 but most importantly is capable of piercing the separator and short-circuiting the cell^21,22 [Fig. 2(a)]. This can cause a thermal runaway, resulting in the highly flammable organic electrolyte catching fire, which represents an unacceptable safety hazard in LMBs.

Due to its high reactivity, Li metal spontaneously forms a solid–electrolyte interphase (SEI) at its surface. Considering that organic carbonates are the electrolytes of choice in most Li battery systems, the initial SEI composition is a product of Li alkyl carbonates, Li halides, and large-molecular-weight polymers.^23,24 The SEI also consists of Li₂O and Li₂CO₃ in the presence of trace amounts of water. The structure of the SEI can be described by a “mosaic model” formed by the heterogeneous stacking of domains of various compositions²⁵ [Fig. 2(b)]. Due to this, the SEI exhibits heterogeneous Li-ion conductivity, which leads to the inhomogeneous plating and stripping of lithium metal from the metal anode surface. Additionally, volume variation due to cycling creates cracks in the SEI. This exposes the fresh Li surface, which has a lower energy barrier for Li-ion transport, and this intensifies non-uniform deposition.²⁶ Once the Li has nucleated heterogeneously, dendrite growth is self-enhanced. It can be understood that the high curvature of the nuclei has a considerably high electric field at the tips, which tend to attract more Li ions and lead to further growth of dendrites.^27–30 Thus, once dendrites are nucleated on the Li anode surface, they grow relentlessly until they pierce the separator and short the battery or consume the liquid electrolyte due to uncontrolled SEI growth. Either of these scenarios is highly undesirable and results in catastrophic battery failure.

In order to improve the energy density and inch closer toward the value of gasoline, breakthrough advances are necessary to make Li metal anodes viable and bring metal battery systems to practical and safe utilization. Suppressing and controlling the growth of dendrites has taken top priority due to the safety hazards, they pose in metal batteries. Over the years, researchers have explored a range of strategies such as electrolyte engineering,^{18,27,31–36} covering the Li surface with an artificial SEI,^37–39 homogenizing Li-ion flux,^40–43 utilizing stable hosts for Li metal,^44,45 and solid electrolytes^46–60 to address this challenge [Fig. 2(c)]. In this paper, we will review some of the recent breakthrough work that shows great promise to control and eliminate dendritic growth in Li metal batteries with liquid electrolytes and offer some perspectives and directions for future work.

Electrolytes play a vital role in the formation of a stable SEI layer on the negative electrode. Commonly used Li salts such as LiPF₆, LiClO₄, LiAsF₆, and Li[FSA] dissolved in organic solvents such as ethylene carbonate (EC), diethyl carbonate (DEC), 1,2-dimethoxyethane (DME), and vinyl carbonate (VC) along with additives such as HF and fluoroethylene carbonate (FEC) have been investigated^61–70 as electrolytes in LMBs. Recent years have seen important innovations in the chemistry and composition of electrolytes for dendrite suppression, which are summarized and discussed below.

Unlike organic electrolytes, ionic liquid (IL) electrolytes are better able to suppress dendrite growth by forming an electrochemically stable SEI.^71–75 Basile et al. reported that Li metal immersed in an IL comprising the pyrrolodinium cation and N-propyly-N-methylpyrrolidinium bis(fluorosulfonyl)imide anion ([C₃mPyr⁺][FSI⁻]) for 12 days exhibits stable Li/Li symmetric cell cycling with effective suppression of Li dendrite growth.⁷¹ Furthermore, the ([C₃mPyr⁺][FSI⁻]) IL is known to possess low viscosity at room temperature and has been associated with delayed corrosion at the cathode. Hwang et al. showed that an IL electrolyte comprising of Li[FSA]⁻[1-ethyl-3-methylimidazolium], ([FSA]⁻ = bis(fluorosulfonyl)amide) presents unique properties of suppressing Li dendrite growth, dead Li accumulation, and long cycling performance from 25 to 90 °C. Exceptional full-cell performance of this electrolyte was observed using an Li₂FeP₂O₇ positive electrode with an average columbic efficiency of 99.97%⁷⁶ [Figs. 3(a) and 3(b)]. Thus, the use of ILs over conventional organic electrolytes offers great benefit in terms of dendrite suppression but the downside of ILs is their significantly increased cost.

Bai et al. presented a design strategy of achieving selective ion transport in metal-organic framework (MOF)-modified electrolytes for homogenous Li electrodeposition⁷⁷ [Fig. 3(c)]. It was observed via density functional theory (DFT)-MD studies that the TFSI⁻ anions “caged” in periodic MOF cavities and facilitate homogenous Li-ion flux with retarded dendritic growth [Fig. 3(d)]. Moreover, Yuan et al. showed that another kind of single-ion conductor in a liquid electrolyte can be prepared by the coordination of ClO₄⁻ anions in the liquid electrolyte on open metal sites of a Cu-MOF-74 matrix to realize homogenous single Li-ion transport in the electrolyte.⁷⁸ Thus, the use of MOF-modified electrolytes offers more homogeneous and uniform electrodeposition of Li when compared to traditional electrolytes, which helps to mitigate the nucleation and growth of dendrites.

Additives to electrolytes even at the ppm levels are known to greatly affect dendrite deposition morphology. Commonly investigated additives include HF,^31,32 LiF,²⁰ (C₂H₅)₄NF(HF)_4,⁷⁹ fluoroethylene carbonate,⁸⁰ 2-methylfuran,⁸¹ and various surfactants.⁸² In particular, quaternary ammonium salts such as CsPF₆ and RbPF₆ introduced as additives into the electrolyte are adsorbed on the Li metal surface and suppress Li dendrite growth.²⁷ At low concentrations, Cs⁺ and Rb⁺ ions exhibit an effective reduction potential lower than that of Li ions. During Li deposition, these additive cations form an electrostatic shield around the initial growth tip of the protuberances without reduction of the additive ions. Such an electrostatic shield forces further deposition of Li to adjacent regions of the anode and eliminates dendrite formation.²⁷ In an exciting recent report, Wang et al. showed that the addition of 0.02M thiourea (THU) to the electrolyte enables uniform and dendrite-free Li plating via a superfilling mechanism⁸³ [Fig. 3(e)]. In the initial stages of Li deposition, both Li⁺ and THU molecules are adsorbed on the surface of the Li metal substrate. THU is enriched on the concave surface rather than the convex surface of protrusions on the surface, thereby acting as a leveling agent. In such a situation, the Li⁺ tends to deposit on the concave regions, forming a level and smooth surface on the electrode.⁸³ In this way, the use of electrolyte additives offers an effective and cost-efficient means to retard and slow down dendritic growth in LMBs. That being said, it is unlikely that Li dendrites can be completely eliminated by a modification to the electrolyte alone, which leads to research in designing stable Li hosts.

Dendrite nucleation and growth can be mitigated by the use of a suitable lithiophilic host matrix.^45,84–87 In order to prevent uneven distribution of Li and provide a physical barrier for dendrite growth, confining Li plating/stripping in micro-/nanoengineered spaces are popular strategies. Alternative approaches have explored storing molten Li with a framework such as a polymeric host,⁸⁸ 3D carbon–Si frame,⁸⁹ and layered graphene oxide.⁴⁴ Ju et al. showed that using a trifluoroethanol (TFEA)-modified egg-shell membrane (ESM) (TESM) modified Cu foil can be used to control the nucleation and growth of Li⁹⁰ [Fig. 4(a)]. It was found that TESM can block lithium dendrite growth along the preferred direction and could also be incorporated in the SEI. Pristine ESM was obtained by etching CaCO₃ in eggshells with a dilute acetic acid solution, followed by freeze drying. Modifying the obtained ESM (with inevitable defects) with a one-step TFEA solvothermal treatment can reduce possible side reactions. Density functional theory calculations and inductively coupled plasma mass spectrometry (ICP-MS) adsorption techniques indicate that due to the strong binding interactions Li ions are adsorbed onto TESM. These ions are further homogeneously distributed by the 3D network provided by TESM, restraining the formation of Li dendrites.⁹⁰ 

In another work, Xu et al. fabricated a dual-functional scaffold, by evenly coating a g-C₃N₄ layer on commercial carbon cloth (CC) to find that metallic Li could uniformly deposit into the interlayer between g-C₃N₄ layer and CC fibers⁹¹ [Fig. 4(b)]. The scaffold is synthesized by soaking commercial CC into 0.2M urea solution followed by thermal condensation at 600 °C. The synergy of an artificial interphase and high specific surface area scaffold significantly restrains dendrite growth. Electrochemically, the Li/0.2 M g-C₃N₄/CC anode runs for over 1500 h with a lower overpotential than a bare Li or a Li/CC electrode.

The poor lithiophilicity of carbon cloth is the major barrier to the infusion of molten Li. Previous reports have proposed lithiophilic layers such as SnO₂, ZnO, and Ag to improve the lithiophilicity of carbon materials.^92,93 Additionally, the lithiophilic Zn is known to improve the wettability of the carbon cloth by molten Li, as the Li tends to form a solid solution with Zn.⁴⁵ Li et al. electrodeposited Zn nanoparticles on the fibers of 3D carbon cloth using ionic liquid (1-ethyl-3-methylimidazolium tetrafluoroborate) as the deposition solution.⁹⁴ Molten Li was infused into this Zn-carbon cloth to form a well-confined Li composite anode. This anode exhibits stable long term cycling performance compared to a bare Li metal anode [Fig. 4(c)].

Thus, developing lithiophilic host matrices into which the Li metal can be uniformly and safely deposited and stripped appears to be pivotal in preventing dendritic outbreaks. That being said, it must be recognized that the lithiophilic host matrix adds weight to the anode and reduces the achievable specific capacity below the theoretical maximum (3860 mAh g⁻¹) for Li metal.

The stability and homogeneity of the SEI play a vital role in the plating and stripping behavior of Li. Additionally, a low SEI/Li interface energy and insufficient mechanical strength of the SEI allow the Li dendrite to penetrate through the SEI due to large volume change during electrochemical cycling.^95,96 To address this, the use of “interfacial layers” on Li metal such as diamond-like carbon film, Al₂O_3,³⁹ HF-modified poly(dimethylsiloxane) film,⁹⁷ Li₃PO₄,⁹⁸ and interconnected hollow carbon nanospheres⁹⁹ have been explored in the literature. Below, we summarize some recent novel Li metal interfaces that have been reported to have a considerable beneficial impact on Li dendrite mitigation.

Xu et al. demonstrated by phase-field simulation that the adhesion energy of 0.5 J/m² can effectively suppress dendrite growth. They identified LiAl as a promising candidate to achieve high adhesion energy. An LiAl-rich interfacial layer is formed on the Li metal surface by treating bare Li metal with AlCl₃ in THF. The LiAl/Li interface has a high adhesion energy of ∼1.58 J/m², and it was demonstrated that this interfacial layer remarkably improves cycling stability in carbonate based electrolytes as compared to a bare Li metal electrode¹⁰⁰ [Fig. 5(a)].

Liu et al. reported that a high ϒE product (ϒ—Li/SEI interface energy and E—Young's modulus of the SEI) improves the mobility of Li atoms along the interphase and suppresses Li dendrite growth.^101,102 DFT calculations reveal that SrF₂ has a high value of the ϒE product. They formed a SrF₂-rich SEI on Li-11 wt. % Sr alloy anodes in an ether based electrolyte. The high reducing capability of Li–Sr enables the effective formation of a SrF₂-rich interphase when immersed in a fluorine based electrolyte. Such an electrode exhibits a stable cycling performance and a high columbic efficiency of 99.42% in lithium metal batteries¹⁰³ [Fig. 5(b)].

Qi et al. report an interfacial layer based on a polymer of intrinsic microporosity (PIM) that inhibits dendrite formation and growth.¹⁰⁴ This polymer interfacial layer is shown to possess uniform 3D interconnected pores to facilitate uniform conduction of Li⁺ ions and a high Young's modulus, sufficient to suppress Li dendrite growth. The polymer consists of dimethylbiphenyl (DMBP) structural units fused together by a Troger's base linkage (TB) to yield a PIM-DMBP-TB interfacial layer. This layer also acts as a sieve, selectively allowing the transport of less solvated Li⁺ ions instead of the fully solvated ones, thereby suppressing side reactions between the electrolyte and Li metal¹⁰⁴ [Fig. 5(c)].

Based on the above discussion, it is clear that interfacial layers deposited directly on the surface of the Li metal electrode or on the current collector can have a powerful effect in inhibiting nucleation and growth of dendrites. This concept of engineering an artificial SEI layer that is customized to suppress dendrites causes a minimal reduction in the gravimetric and volumetric energy density. That being said, fail-safe (i.e., 100%) prevention of dendritic outbreaks is difficult to guarantee with interfacial engineering alone.

Separators are a key component of batteries, and their thermal stability and wettability directly affect the electrochemical performance of batteries.^105–109 Reports have shown that using polymer electrolytes incorporated with inorganic particles demonstrate superior wettability as well as mechanical and thermal stability at high temperatures.^110–112 Non-uniform pore distribution in the separator causes disordered diffusion of Li⁺ ions during plating/stripping, which leads to the formation and growth of Li dendrites. Zhao et al. fabricated a poly(vinylidene fluoride-co-hexafluoropropylene) (PVDF-HFP)/Li-montmorillonite (MMT) composite separator that exhibits enhanced electrolyte wettability, thermal stability, and electrochemical stability [Fig. 6(a)]. The parallel interlayers in the separator enable the uniform flow of Li⁺ ions, thereby suppressing dendrite growth.¹¹³ Rao et al. demonstrated that a purely inorganic separator comprised of hydroxyapatite (HAP) nanowires can suppress Li dendrite growth owing to its uniform porosity, which facilitates Li⁺ diffusion and reduces interfacial resistance [Fig. 6(b)]. Additionally, the Ca present in the HAP separator can effectively consume the F⁻ produced in the electrolyte to suppress attack due to HF to yield improved stability at high temperatures.¹¹⁴ 

The above strategies involve internal alterations to the battery components to suppress and control dendrite growth. However, these alterations introduce additional parameters that affect cell stability as well as increase battery manufacturing cost. Recently, it has been demonstrated that Li metal dendrites can in fact be suppressed without making any changes to the internal cell components. Shen et al. demonstrated the use of an external magnetic field to obtain uniform Li⁺ ion distribution during plating/stripping to suppress dendrite growth. Such a magnetohydrodynamics effect changes the direction of charged species inducing convection of the electrolyte, improving ion distribution and mass transfer, and decreasing the concentration gradient of Li⁺ ions¹¹⁵ [Fig. 6(c)].

In previous sections, we have summarized various interesting and innovative approaches to mitigating and suppressing Li dendrites in Li metal batteries. It should, however, be noted that the nucleation and growth of Li dendrites are kinetically highly favorable and, therefore, complete (i.e., 100%) elimination of dendrites is unlikely. A possible solution to this challenge could like in the physical healing of dendrites once they have formed. In recent work, our group has studied the evolution of dendrite morphologies at varying current densities and observed that beyond a threshold current density, the local rise in temperature can trigger extensive surface diffusion to heal the dendrites^116–118 [Fig. 7(a)]. In this way, an internal temperature field within the cell can be imposed on the dendritic surface to “heal” away from any dendrites that may have formed during cell cycling. We have used detailed computational thermal modeling and density functional theory (DFT) calculations to predict the temperature rise in the dendrites and the temperature driven surface diffusion phenomena that is responsible for the planarization or flattening of the dendrites [Fig. 7(b)]. This healing of Li dendrites has been demonstrated in full cells using Li metal anode and lithium iron phosphate based cathodes in a carbonate electrolyte [Fig. 7(a)] and in a Li–Sr environment. Additionally, this healing of dendrites was also demonstrated in potassium metal batteries,¹¹⁸ confirming that the healing phenomenon is not restricted exclusively to Li.

Theoretical studies by Hong and Viswanathan reveal the importance of electrochemical reaction barrier as a parameter and the effects of thermal shock and diffusion barriers on dendrite growth and healing.¹¹⁹ Recently, Wang et al. used a strategy involving asymmetrical bidirectional current mode (ABCM) of charging, which tends to “heal” dendrites that might be forming during a charge cycle¹²⁰ [Fig. 7(c)]. Such self-heating induced healing of dendrites eliminates the risk of thermal runaway caused by battery shorting and serves as a reliable strategy for safe utilization of secondary metal batteries.

In this review, we have discussed the need and urgency to develop chemistries beyond Li-ion to match energy demands provided by gasoline. Utilizing Li metal as the anode, battery systems are capable of reaching energy densities closer to that of gasoline. We discuss the problems associated with using a Li metal anode, focusing on the nucleation and growth of dendrites during electrochemical cycling. In order to suppress and control the growth of Li dendrites, we have outlined various strategies and recent advances to optimize the electrochemical performance of Li metal anodes. Although tremendous progress has been made to address dendritic growth, low columbic efficiency, volume expansion of the metal anode, and safety issues, there is still a significant need for improvement in order to realize the commercial utilization of LMBs.

The stability and composition of the SEI play an important role in the performance of LMBs. Tuning the composition of the SEI via electrolyte additives and artificial coatings to protect the Li metal anode are shown to be promising avenues. Plasticity and adhesion of these layers on the Li metal anode prove to be the main concern, which researchers need to further improve. Developing improved electrolytes show effective suppression of dendrite growth but compromises such as the reduction in the operating voltage window and increased material cost must be taken into consideration. Moreover, polymers and oligomers should be explored as possible coatings. Most of the coatings are currently designed as single layers but multi-layered and multi-functional coatings should be synthesized to increase the efficacy of coatings and tackle multiple problems of the Li metal anode. Using stable hosts for Li metal, homogenizing the Li⁺ ion flux, minimizing the high volume change and deposition of physical barriers for dendrite growth are all effective strategies. However, a loss of gravimetric and volumetric energy density as well as an increase in battery manufacturing cost is inevitable in many of these approaches, which can reduce their attractiveness.

One of the main challenges of effectively realizing strategies for Li metal anodes is in a full cell, when the Li metal anode is paired with cathodes such as lithium metal oxides. These cathodes are known to have their own unique problems such as cation dissolution and shuttling of species from the cathode.¹²¹ A change in the electrolyte environment due to the dissolution of species from the cathode is known to affect the plating/stripping performance of Li metal. Developing strategies to effectively combat the problems of the cathode and the anode are of paramount importance. This requires an engineering system (i.e., cell and pack) level perspective rather than simply looking at the anode or cathode in isolation.

Several smart designs such as dendrite detection,¹²² battery shut down mechanisms,¹²³ and thermoresponsive flame-retardant release¹²⁴ have been previously demonstrated using novel engineering separators and current collectors. Further safety strategies by incorporating smart materials into battery design would be worth exploring. Additional concerns that need to be addressed are electrolyte drying and the optimization of the above strategies when scaling up to larger form-factor pouch and cylindrical cells. What works in a coin cell need not always work in a scaled up pouch, prismatic, or cylindrical cell. Therefore, form-factor scalability is a critical issue that must be addressed. Finally, to realize the utilization of Li metal anode in LMBs, it is essential to consider the battery pack as an integrated system. For such a complex (engineered) system, it is unlikely that there is a single magic bullet or solution, making it necessary to explore a combination of strategies to work in synergy to make Li metal anodes viable in commercial systems. We look forward to the exploration and exciting advances in the field of LMBs in the years to come.

N.K. acknowledges support from the USA National Science Foundation (Award No. 1922633), and John A. Clark and Edward T. Crossan endowed chair professorship at the Rensselaer Polytechnic Institute.

Data sharing is not applicable to this article as no new data were created or analyzed in this study.