Interactions between Lithium Growths and Nanoporous Ceramic Separators

•Three current-dependent growth modes during Li electrodeposition are identified•Nanoporous ceramic separators can block Li growths up to a critical current density•Internal shorts at under-limiting currents are due to reaction-limited surface growth•Sudden voltage drops are signs of the metal penetration through the separator Li-ion batteries are energy-dense power sources of cell phones, laptops, and electric vehicles. The basic unit inside is a three-layer stack, i.e., anode, separator, and cathode, fully wetted by organic liquid electrolyte. Removing the ion-insertion anode materials could significantly increase the energy density of the battery. During recharge, Li ions that used to be accommodated by the anode materials will be reduced to form a Li metal anode, without dead weight and volume. The process, however, is notoriously unstable and always forms finger-like structures that can penetrate the separator to short-circuit the battery, through mechanisms more complex than the simple term “dendrite” can reveal. Depending on the current, one may generate tip-growing dendrites, root-growing whiskers, or surface-growing clusters. This study presents the accurate understanding of each growth mode, which is critical for controlling the hazardous instabilities across the entire range of working conditions. To enable lithium (Li) metal anodes with high areal capacity that can match advanced cathodes, we investigate the growth mechanisms and the tendency of the deposited metal to penetrate nanoporous ceramic separators across a range of practical current densities. Our results from realistic sandwich cells and special transparent junction cells suggest the existence of three growth modes of lithium, due to the competing reactions of lithium deposition and the solid electrolyte interphase formation. A critical current density (6 mA cm−2), ∼30% of the system-specific limiting current density, is identified as a practical safety boundary for battery design and operation, under which root-growing lithium whiskers are the dominant structure of electrodeposition and can be blocked by the nanoporous ceramic separator. Our operando experiments reveal that metal penetration of the separator does not lead to zero voltage immediately, but rather to sudden, small voltage drops, which should not be treated as benign soft shorts. To enable lithium (Li) metal anodes with high areal capacity that can match advanced cathodes, we investigate the growth mechanisms and the tendency of the deposited metal to penetrate nanoporous ceramic separators across a range of practical current densities. Our results from realistic sandwich cells and special transparent junction cells suggest the existence of three growth modes of lithium, due to the competing reactions of lithium deposition and the solid electrolyte interphase formation. A critical current density (6 mA cm−2), ∼30% of the system-specific limiting current density, is identified as a practical safety boundary for battery design and operation, under which root-growing lithium whiskers are the dominant structure of electrodeposition and can be blocked by the nanoporous ceramic separator. Our operando experiments reveal that metal penetration of the separator does not lead to zero voltage immediately, but rather to sudden, small voltage drops, which should not be treated as benign soft shorts. Developing a stable rechargeable lithium (Li) metal anode has become an urgent need for the realization of post-Li-ion batteries, including Li-O2, Li-S, and hybrid Li-flow batteries.1Manthiram A. Yu X. Wang S. Lithium battery chemistries enabled by solid-state electrolytes.Nat. Rev. Mater. 2017; 2: 16103Crossref Scopus (2241) Google Scholar, 2Xu W. Wang J. Ding F. Chen X. Nasybutin E. Zhang Y. Zhang J.-G. Lithium metal anodes for rechargeable batteries.Energ Environ. Sci. 2014; 7: 513-537Crossref Google Scholar, 3Choi J.W. Aurbach D. Promise and reality of post-lithium-ion batteries with high energy densities.Nat. Rev. Mater. 2016; 1: 16013Crossref Scopus (2821) Google Scholar, 4Christensen J. Albertus P. Sanchez-Carrera R.S. Lohmann T. Kozinsky B. Liedtke R. Ahmed J. Kojic A. A critical review of Li/Air batteries.J. Electrochem. Soc. 2012; 159: R1-R30Crossref Scopus (942) Google Scholar It also holds the promise to significantly increase the energy density of current Li-ion batteries by replacing the bulky graphite anode.5Albertus P. Babinec S. Litzelman S. Newman A. Status and challenges in enabling the lithium metal electrode for high-energy and low-cost rechargeable batteries.Nat. Energy. 2018; 3: 16-21Crossref Scopus (978) Google Scholar However, two major issues hinder the practical application of lithium metal anodes in rechargeable batteries. One is the internal short-circuiting of the cell by lithium metal whiskers and dendrites that can lead to catastrophic accidents. The other is the low Coulombic efficiency and therefore the short cycle life, caused by the continual consumption of active lithium and electrolyte components to form inert solid electrolyte interphase (SEI) layers on the surface of the non-uniform lithium deposits, during battery recharge. While the low Coulombic efficiency might be compensated by adding excess lithium or limiting the depth of charge/discharge, the risk of internal shorts is a serious drawback that must be solved to enable safe long-lasting rechargeable lithium metal batteries. Among the latest studies on the mechanisms of lithium dendrite penetration, Lv et al.6Lv D.P. Shao Y.Y. Lozano T. Bennett W.D. Graff G.L. Polzin B. Zhang J.G. Engelhard M.H. Saenz N.T. Henderson W.A. et al.Failure mechanism for fast-charged lithium metal batteries with liquid electrolytes.Adv. Energy Mater. 2015; 5https://doi.org/10.1002/aenm.201400993Crossref Scopus (477) Google Scholar reported that dendrite-induced short-circuiting was actually not found in coin cells using the common polyolefin separator (pore size 100–200 nm). Cell failure is more often attributed to electrolyte dry out due to continual SEI generation on the porous high-surface-area lithium deposits.7Aurbach D. Zinigrad E. Cohen Y. Teller H. A short review of failure mechanisms of lithium metal and lithiated graphite anodes in liquid electrolyte solutions.Solid State Ionics. 2002; 148: 405-416Crossref Scopus (1353) Google Scholar, 8Aurbach D. Zinigrad E. Teller H. Dan P. Factors which limit the cycle life of rechargeable lithium (metal) batteries.J. Electrochem. Soc. 2000; 147: 1274-1279Crossref Scopus (279) Google Scholar Indeed, the standard practice of using a lithium metal anode to evaluate the cycle life of a cathode material has never suffered from internal shorts. Such successes may have been enabled by the fact that the areal capacity of the lab-made cathodes are relatively low, so that only a small amount of lithium (∼1 mAh cm−2) is being deposited and dissolved in each cycle. However, when more practical cathode loadings are used, where the areal capacities are larger than 2 mAh cm−2, the behavior of lithium metal anodes becomes drastically different,9Jiao S. Zheng J. Li Q. Li X. Engelhard M.H. Cao R. Zhang J.-G. Xu W. Behavior of lithium metal anodes under various capacity utilization and high current density in lithium metal batteries.Joule. 2018; 2: 110-124Abstract Full Text Full Text PDF Scopus (215) Google Scholar and lithium penetration through nanoporous separators can easily occur.6Lv D.P. Shao Y.Y. Lozano T. Bennett W.D. Graff G.L. Polzin B. Zhang J.G. Engelhard M.H. Saenz N.T. Henderson W.A. et al.Failure mechanism for fast-charged lithium metal batteries with liquid electrolytes.Adv. Energy Mater. 2015; 5https://doi.org/10.1002/aenm.201400993Crossref Scopus (477) Google Scholar, 9Jiao S. Zheng J. Li Q. Li X. Engelhard M.H. Cao R. Zhang J.-G. Xu W. Behavior of lithium metal anodes under various capacity utilization and high current density in lithium metal batteries.Joule. 2018; 2: 110-124Abstract Full Text Full Text PDF Scopus (215) Google Scholar It is critical to understand and predict the precise conditions for such short circuits, before one can optimize the design and operation of rechargeable metal batteries. A quantitative safety boundary, called Sand's capacity, was recently proposed based on experiments in special capillary cells.10Bai P. Li J. Brushett F.R. Bazant M.Z. Transition of lithium growth mechanisms in liquid electrolytes.Energ Environ. Sci. 2016; 9: 3221-3229Crossref Google Scholar This characteristic capacity marks the transition between two distinct growth mechanisms of lithium metal in liquid electrolytes: (1) at current densities greater than the intrinsic diffusion-limited current density (over-limiting current density), once the charged capacity exceeds Sand's capacity, lithium grows at the outermost tips of the electrode surface to form fractal “dendritic” structures; (2) at current densities lesser than the diffusion-limited current density (under-limiting current density), or before the Sand's capacity, lithium primarily grows from the root, like human hairs, to form “mossy” structures.10Bai P. Li J. Brushett F.R. Bazant M.Z. Transition of lithium growth mechanisms in liquid electrolytes.Energ Environ. Sci. 2016; 9: 3221-3229Crossref Google Scholar Because the deposits formed prior to Sand's capacity can be blocked by a nanoporous ceramic separator, whereas deposits formed afterward cannot, Sand's capacity can be considered as a safety boundary to avoid the tip-growing dendrites. However, due to the small inter-electrode separation in practical batteries, Sand's capacity appears too high to be useful. Moreover, for under-limiting, but still relatively large, current densities, lithium penetration can still occur. For this regime of reaction-limited lithium growth, more accurate critical safety boundaries must be identified. In the present study, we constructed two types of symmetrical lithium cells to investigate the safety boundaries in terms of the applied current density and areal capacity beyond which nanoporous ceramic separators can no longer block the metal growth and internal shorts become inevitable. The first set of experiments exploits a sandwich structure, which consists of lithium electrodes, an anodic aluminum oxide (AAO) separator, and polyvinylidene fluoride (PVDF) washers on both sides of the AAO to create two electrolyte-filled compartments. The second set of experiments uses a junction cell consisting of two segments of straight glass capillary and a small piece of AAO separator, all sealed in transparent epoxy resin. Using both types of experiments, we identify two critical current densities that separate three different growth modes of greatly varying safety risk. Below the first critical current density, Jcc, lithium grows primarily from the root to form whiskers, which may be attributable to the complete coverage of the lithium surface by a robust SEI layer. Beyond the second critical current density, Jlim, the system-specific diffusion-limited current density, ion depletion at the electrode surface leads to the diffusion-limited, tip-growing, dendritic lithium that can easily penetrate the AAO nanopores and short the cell. Between these two critical current densities, the SEI formation could be interrupted by the competing lithium deposition, leaving parts of the metal surface without the continuous SEI coverage, such that near isotropic surface growths start to prevail. Based on these findings, we propose a set of safety boundaries along with strategies to optimize the design of rechargeable metal batteries. Figure 1 displays the structure of the sandwich cell and the transient voltage responses to galvanostatic electrodeposition at 1 mA cm−2. At first glance, the steep voltage increase observed in Figure 1B appears similar to Sand's behavior due to ion depletion near the electrode surface, which creates a non-conductive region and necessitates an increased polarization to maintain constant current density.10Bai P. Li J. Brushett F.R. Bazant M.Z. Transition of lithium growth mechanisms in liquid electrolytes.Energ Environ. Sci. 2016; 9: 3221-3229Crossref Google Scholar However, this current density (1 mA cm−2) is well below the estimated diffusion-limited current density (20 mA cm−2), according to the dilute solution theory10Bai P. Li J. Brushett F.R. Bazant M.Z. Transition of lithium growth mechanisms in liquid electrolytes.Energ Environ. Sci. 2016; 9: 3221-3229Crossref Google Scholar Jlim = 2zcc0FDapp(taL)−1. Here, zc = 1 is the charge number of the lithium cation, c0 = 1 M is the bulk electrolyte concentration, F is the Faraday constant, Dapp = 2 × 10−6 cm2 s−1 is an averaged apparent diffusion coefficient of lithium cations in 1 M liquid electrolyte,10Bai P. Li J. Brushett F.R. Bazant M.Z. Transition of lithium growth mechanisms in liquid electrolytes.Energ Environ. Sci. 2016; 9: 3221-3229Crossref Google Scholar ta = 0.62 is the transference number of the hexafluorophosphate anion (PF6−), and L = 300 μm is the inter-electrode distance in the sandwich cell. Therefore, the voltage spikes must come from other processes. Upon dissembling the cell post test, the lithium source electrode was almost completely consumed, and the bare surface of the stainless steel current collector was clearly visible (Figure 1C). This implies that the voltage spike was the result of the depletion of the lithium source electrode, instead of ion depletion in the electrolyte. We then doubled the amount of lithium source by using two lithium chips as the source electrode (Figure 1D). The voltage spike disappears, but is replaced by a small hump (red curve in Figure 1B), which can be attributed to the dissolution of the second lithium chip. Postmortem analyses of the one-Li-chip (Figure 1C) and two-Li-chip (Figure 1D) experiments revealed that, in the former experiment, the free-standing part of the AAO was punched through such that the edges perfectly match the inner diameter of the PVDF washer, while, in the latter experiment, the AAO was only slightly distorted and remained intact. As schematically explained in Figure 1C, depletion of the single lithium chip significantly undermined the mechanical support to the AAO at the inner rim of the washer. In the control experiment with two lithium chips (Figure 1D), dissolution occurred preferentially at the center of the second lithium chip, thus the support to AAO at the inner rim of the washer was less affected. In both cases, however, a large amount of lithium was successfully deposited inside the lower compartment before the fracture of or metal penetration into the AAO. This is consistent with the observed smooth voltage responses before the sudden voltage drop. The results suggest that lithium growths under very low current densities can be blocked by the nanoporous ceramic separator. Further experimental results from galvanostatic electrodeposition at 1 mA cm−2 and higher current densities are displayed in Figure 2. The experiments were repeated five times at each current density to determine an average response and measurement uncertainty introduced during the cell assembly. As the current density increases, the sudden voltage drop occurs sooner. We took the areal capacities at the first sudden voltage drop to be the penetration capacities (solid arrow head in Figures 2A–2E) and plotted them against their corresponding current densities in Figure 2F, where the dotted line is the theoretical areal capacity (thickness) of lithium metal that can be accommodated in the lower compartment of the sandwich cell. Figure 2F clearly shows that the lithium deposits tend to completely fill the compartment before they penetrate AAO at current densities below 6 mA cm−2. For higher current densities approaching the limiting value (20 mA cm−2), lithium deposits become increasingly penetrative, leaving the lower compartment less filled (Figure 2F). To further verify the above interpretation, we applied an over-limiting current density of 50 mA cm−2 to the sandwich cells to make the effects of lithium penetration more pronounced and easier to identify. During the deposition, three features were identified in the voltage responses (Figure 3): (1) a voltage increase due to strong concentration polarization (i.e., Sand's behavior10Bai P. Li J. Brushett F.R. Bazant M.Z. Transition of lithium growth mechanisms in liquid electrolytes.Energ Environ. Sci. 2016; 9: 3221-3229Crossref Google Scholar), (2) a linear slope, and (3) a sudden drop with a noisy tail. The linear slope could be attributed to the process of dendrite penetration, as will be discussed later in the transparent junction cell. Comparing the magnitude of the voltage and the time (capacity) accumulated awaiting the penetration, it is clear that the AAO separator with smaller pore size exhibited much higher resistance to penetration than that with larger pores. Figure 3B is an optical micrograph of the Li-penetrated AAO harvested from the short-circuited cell, on which a dark region is visible on the upper side (facing the source electrode). We broke the AAO across this region and investigated the morphologies by scanning electron microscopy (SEM). Indeed, very thin lithium filaments were found there, which were responsible for the sudden voltage drop. A large amount of porous lithium deposits was found beneath the AAO, i.e., in the lower compartment (Figure 1A). To correlate electrochemical responses with physical processes more clearly, we devised a transparent miniature cell with glass capillaries to visualize the operando penetration process. As seen in Figure 4B, similar features emerged in the voltage response when the same over-limiting current density of 50 mA cm−2 was employed: (1) strong polarization due to Sand's behavior, (2) a linear voltage slope, and (3) a sudden voltage drop and decay. From time point C to point F (labeled in Figure 4B), the electrode on the right was pushed backward for 11 μm, while the dendritic lithium deposits were advancing through the nanochannels of the AAO. The capacity delivered by point F was 0.97 μAh, which is equivalent to a 6.8-μm-thick disk of lithium metal within the capillary (300 μm diameter). Neglecting the volume of lithium in the AAO channels and the volume of the SEI layers covering the deposits, we estimated the porosity of the deposited lithium in our miniature cell to be 38% based on the difference in the theoretical (6.8 μm) and experimental (11 μm) lengths. Our operando experiment raises a question of how to judge an internal short by simple inspection of the electrochemical responses. As is evident in Figure 4B, lithium penetration into the AAO at point F did not lead to zero voltage. Even after physical contact is made between the lithium deposits and the source electrode (time points G through H in Figure 4B), the magnitude of the voltage appears comparable with the initial voltage in the experiment (time point C). This phenomenon is likely due to the insulating properties of the SEI layers coating the lithium deposits. Thus, such “soft shorts,” defined as the sudden small drops in the voltage response, require special attention.5Albertus P. Babinec S. Litzelman S. Newman A. Status and challenges in enabling the lithium metal electrode for high-energy and low-cost rechargeable batteries.Nat. Energy. 2018; 3: 16-21Crossref Scopus (978) Google Scholar Since the time interval from point C to point D in Figure 4B is only 2 min, the voltage variation in between will appear like a negligible voltage spike in a long-duration cycling profile. According to our results, a sudden voltage drop (time point F in Figure 4B), is indicative of lithium penetration through the separator, even though the penetrated deposits (circled part in Figure 4F) may not have touched the source electrode and the voltage appeared “normal.” This is why we defined the first sudden voltage drops in Figure 1 as the sign of the dendrite penetration, regardless of the magnitude of the voltage drop and the absolute values of the resultant voltage. By designing unique experimental cells and combining operando experiments with postmortem analyses, we clarified the details of lithium growth mechanisms in liquid electrolyte. As summarized schematically in Figure 5, lithium grows in different modes and into different morphologies/microstructures depending on the applied current densities. When a low current density is applied (<6 mA cm−2, or 30% of the limiting current density, for the sandwich cell in this work), the electrochemical reduction of electrolyte components leads to the formation of a complete coverage of robust SEI on the surface of the lithium metal anode at an early stage. Li cations can steadily diffuse across and deposit beneath the SEI layers, causing an increase in internal pressure.11Yamaki J. Tobishima S. Hayashi K. Saito K. Nemoto Y. Arakawa M. A consideration of the morphology of electrochemically deposited lithium in an organic electrolyte.J Power Sources. 1998; 74: 219-227Crossref Scopus (368) Google Scholar At a certain point, when the pressure reaches a threshold, it will squeeze a plastic flow of lithium metal out of the SEI coat through a pinhole (Figure 5B). With very little coverage on the side walls of the protruding whisker, especially at the newly exposed part (Figure 5C), i.e., the root, further deposition leads to the root-growing phenomenon. Because the tip of whisker is being pushed from behind, penetrating through the channels of AAO is akin to threading a needle and is actually even more difficult as the whiskers are usually wider than the nanochannels, especially in ether-based electrolytes.12Zheng G.Y. Lee S.W. Liang Z. Lee H.W. Yan K. Yao H. Wang H. Li W. Chu S. Cui Y. Interconnected hollow carbon nanospheres for stable lithium metal anodes.Nat. Nanotechnol. 2014; 9: 618-623Crossref PubMed Scopus (1353) Google Scholar, 13Qian J.F. Henderson W.A. Xu W. Bhattacharya P. Engelhard M. Borodin O. Zhang J.G. High rate and stable cycling of lithium metal anode.Nat. Commun. 2015; 6: 6362Crossref PubMed Scopus (1594) Google Scholar, 14Miao R.R. Yang J. Xu Z.X. Wang J.L. Nuli Y. Sun L.M. A new ether-based electrolyte for dendrite-free lithium-metal based rechargeable batteries.Sci. Rep. 2016; 6: 21771Crossref PubMed Scopus (153) Google Scholar Instead, the growing lithium whiskers are forced to kink, elongate, and spread below the AAO, until the mechanical pressure applied by the growing deposits is sufficient to punch through the separator (Figure 1C). At the other limit, when over-limiting current densities are applied (>20 mA cm−2 for the sandwich cell in this work), instability will occur at Sand's time. Tip-growing dendritic lithium then explosively grows out to catch up the retreating concentration front in the electrolyte in order to maintain the required constant current density.10Bai P. Li J. Brushett F.R. Bazant M.Z. Transition of lithium growth mechanisms in liquid electrolytes.Energ Environ. Sci. 2016; 9: 3221-3229Crossref Google Scholar In this case, penetrating through the nanochannels becomes much easier and only takes a few minutes (Figures 3 and 4). The most interesting scenario, however, occurs when an intermediate current density is applied, under which the rate of lithium deposition becomes comparable with the rate of SEI formation. At such rates, Jcc < J < Jlim, complete coverage of robust SEI may easily be interrupted by lithium deposits (Figure 5E) that quickly form at the most favorable locations on the electrode surface. In such areas without complete SEI coverage, further deposition of lithium cations favors more isotropic surface growths. In other areas, however, the robust SEI can still trigger the growth of lithium whiskers via the mechanism described earlier. The interplay between these growth modes leads to a mossy structure. While the whiskers may still be blocked, the surface growth can penetrate the ceramic nanopores. As the current density increases, more surface growths will be favored, which promotes the metal penetration through the separator (Figure 2F). We propose that this new transition in reaction-limited surface growth reflects the different overpotential dependencies of the competing deposition and passivation reactions15Peled E. Menkin S. Review—SEI: past, present and future.J. Electrochem. Soc. 2017; 164: A1703-A1719Crossref Scopus (1009) Google Scholar, 16Peled E. Golodnitsky D. Ardel G. Advanced model for solid electrolyte interphase electrodes in liquid and polymer electrolytes.J. Electrochem. Soc. 1997; 144: L208-L210Crossref Scopus (767) Google Scholar, 17Peled E. The electrochemical behavior of alkali and alkaline earth metals in nonaqueous battery systems—the solid electrolyte interphase model.J. Electrochem. Soc. 1979; 126: 2047-2051Crossref Scopus (2138) Google Scholar: (1) electrodeposition of lithium cations to form metallic lithium (RLi); (2) reaction of anions to form the inorganic components of the SEI (RinSEI); and (3) reaction of solvents to form the organic components of the SEI (RorSEI). The first two reactions are sensitive to the local overpotential, as well as the depletion of salt concentration at higher current densities. The third reaction proceeds even in the absence of applied current and is able to passivate the electrode surface in less than 1 s,18Odziemkowski M. Irish D.E. An electrochemical study of the reactivity at the lithium electrolyte/bare lithium metal interface: I. purified electrolytes.J. Electrochem. Soc. 1992; 139: 3063-3074Crossref Scopus (76) Google Scholar apparently with negligible overpotential dependence (despite the possibility of coupled electron transfer reactions in corrosion processes). As the applied current density increases, the associated higher overpotentials significantly promote electrodeposition (RLi). At the same time, inorganic SEI growth (RinSEI) is diminished by the progressive depletion of anions from the electrode surface, as well as electron transport limitations across the SEI layer, while organic SEI growth (RorSEI) is largely unaffected by the overpotential and proceeds slowly due to the sluggish electron transport across existing SEI. The net result is an interrupted heterogeneous SEI coverage, leading to mossy surface growth. Moreover, the competition of these reactions is sensitive to the history and state of the electrode surface. For example, lithium electrodeposition on gold electrodes in DMSO electrolyte exhibits whisker growth at high overpotentials and cauliflower-like structures at low overpotentials,19Kushima A. So K.P. Su C. Bai P. Kuriyama N. Maebashi T. Fujiwara Y. Bazant M.Z. Li J. Liquid cell transmission electron microscopy observation of lithium metal growth and dissolution: root growth, dead lithium and lithium flotsams.Nano Energy. 2017; 32: 271-279Crossref Scopus (281) Google Scholar while solvent-diffusion-limited SEI growth on graphite insertion electrodes can explain the long-term degradation of lithium ion batteries under various cycling protocols.20Pinson M.B. Bazant M.Z. Theory of SEI formation in rechargeable batteries: capacity fade, accelerated aging and lifetime prediction.J. Electrochem. Soc. 2013; 160: A243-A250Crossref Scopus (580) Google Scholar Here, we identify, for the first time, the macroscopic consequences of these new growth modes of lithium deposits on lithium electrodes for metal/separator interactions and battery safety. The mechanisms above indicate three safety boundaries that can be quantitatively defined and exploited to increase the safety of rechargeable metal batteries. (1) The critical current density Jcc (∼30% Jlim), below which the thick root-growing lithium whiskers can be easily blocked by available nanoporous ceramic10Bai P. Li J. Brushett F.R. Bazant M.Z. Transition of lithium growth mechanisms in liquid electrolytes.Energ Environ. Sci. 2016; 9: 3221-3229Crossref Google Scholar or polyolefin6Lv D.P. Shao Y.Y. Lozano T. Bennett W.D. Graff G.L. Polzin B. Zhang J.G. Engelhard M.H. Saenz N.T. Henderson W.A. et al.Failure mechanism for fast-charged lithium metal batteries with liquid electrolytes.Adv. Energy Mater. 2015; 5https://doi.org/10.1002/aenm.201400993Crossref Scopus (477) Google Scholar separators, should be experimentally identified for practical cells. (2) The electrolyte composition and separator pore sizes should be optimized synergistically, so that surface growths at higher current densities can be restrained by counteracting the SEI-modified surface energy of the lithium deposits. (3) While the practical current densities in fresh commercial cells using liquid electrolytes are well below their intrinsic limiting current densities, monitoring the deterioration of the ion transport kinetics to alert the value of the real-time limiting current density is key to avoid tip-growing lithium dendrites. While the root-growing whiskers cannot be predicted by existing mathematical models, surface-growing clusters may, in principle, be simulated by a thermodynamically consistent electrochemical phase field model. Recently, García and coworkers21Ely D.R. Jana A. García R.E. Phase field kinetics of lithium electrodeposits.J. Power Sources. 2014; 272: 581-594Crossref Scopus (91) Google Scholar, 22Jana A. Ely D.R. García R.E. Dendrite-separator interactions in lithium-based batteries.J. Power Sources. 2015; 275: 912-921Crossref Scopus (116) Google Scholar applied the classical phase field method to model the electrodeposition of lithium and investigate the interaction between the growths and a rigid porous separator, dominated by surface tension (contact angle and curvature of a pure lithium surface entering the pores). In the notation of this theory, our experiments, showing that growth at 1 mA cm−2 can be readily blocked by rigid nanopores of 100–200 nm in diameter correspond to a dimensionless current density of 50 and a dimensi