['M A T E R I A L S   S C I E N C E\n\nA new approach to both high safety and high \nperformance of lithium-ion batteries\n\nShanhai Ge, Yongjun Leng, Teng Liu, Ryan S. Longchamps, Xiao-Guang Yang, Yue Gao, \nDaiwei Wang, Donghai Wang, Chao-Yang Wang*\n\nWe present a novel concept to achieve high performance and high safety simultaneously by passivating a Li-ion \ncell and then self-heating before use.', 'By adding a small amount of triallyl phosphate in conventional electrolytes, \nwe show that resistances of the passivated cells can increase by ~5×, thereby ensuring high safety and thermal \nstability.', 'High power before battery operation is delivered by self-heating to an elevated temperature such as \n60°C within tens of seconds.', 'The present approach of building a resistive cell with highly stable materials and then \ndelivering high power on demand through rapid thermal stimulation leads to a revolutionary route to high safety \nwhen batteries are not in use and high battery performance upon operation.', 'Copyright © 2020 \nThe Authors, some \nrights reserved; \nexclusive licensee \nAmerican Association \nfor the Advancement \nof Science.', 'No claim to \noriginal U.S. Government \nWorks.', 'Distributed \nunder a Creative \nCommons Attribution \nNonCommercial \nLicense 4.0 (CC BY-NC).', 'INTRODUCTION\nRechargeable lithium-ion batteries (LIBs) are widely used in electri-\nfied vehicles, consumer electronics, and stationary energy storage \nsystems.', 'Simultaneous realization of high safety and high energy \ndensity/performance is a perpetual pursuit.', 'Unfortunately, conven-\ntional batteries are passive devices where the performance, safety, \nand calendar/cycle life are all dictated by the electrochemical reac-\ntivity at ever-present anode/electrolyte and cathode/electrolyte \ninterfaces.', 'An inherent conflict between the reactivity and stability \nof battery materials persists at the electrode/electrolyte interfaces \n(EEIs).', 'More specifically, highly reactive electrode/electrolyte mate-\nrials provide high power and high performance but result in poor \nsafety and accelerated degradation even when the battery is not in \nuse.', 'Highly stable (i.e., less reactive) electrode/electrolyte materials \ngive rise to battery safety, low degradation, low self-discharge, and \nlong life, but such materials offer low power and performance.', 'As a \nresult, materials development for batteries has always aimed at \ntrade-offs of finding electrode and electrolyte materials that are not \ntoo reactive but also not too stable.', 'This work describes a new strategy to achieve both safe and \nenergy-dense battery (SEB) cells, as schematically sketched in Fig.', '1, \nwhere the cell resistance is plotted against the inverse of tempera-\nture.', 'First, a passivated cell is judiciously designed and built by \nusing highly stable materials and by creating exceptionally stable \nEEIs, as characterized by higher charge-transfer resistance (Rct) and \nhigher direct-current resistance (DCR).', 'This is illustrated in Fig.', '1 \nby the line from point a (the conventional Li-ion cell design) to \npoint b (the SEB cell design).', 'It follows that these resistive SEB cells \nare inherently stable and safe, as demonstrated by nail penetration, \nshort circuiting, and high-temperature storage.', 'Second, in order for \nSEB cells to deliver high power during operation, they are heated \ninstantaneously before operation, say from room temperature to 60°C, \nso as to recover acceptable operational DCR and ultimately battery \npower, going from point b (at room temperature) to point c (at an \nelevated temperature such as 60°C) in Fig.', '1.', 'This power on demand \nthrough rapid thermal stimulation is opposed to standby power in \nthe present-day LIB cells, where the high-power EEIs need to be ever \n\nElectrochemical Engine Center and Department of Mechanical Engineering, Pennsylvania \nState University, University Park, PA 16802, USA.', '*Corresponding author.', 'Email: cxw31@psu.edu\n\npresent.', 'Rapid cell heating at a speed of 1° to 5°C/s has been made \npossible by a self-heating structure discovered by Wang et\xa0al.', '(1), \nin which a micrometer-thick nickel foil is inserted in a stack of \nanode-separator-cathode assemblies for uniform and internal heating.', 'Thus, heating from room temperature to 60°C for battery operation \nconveniently takes only 10 to 20 s.\n\nTo achieve the requisite stable, robust EEIs, the additive, triallyl \nphosphate (TAP), is used in this work based on sufficient evidence \nin the literature of its ability to improve cell stability at both high- \nvoltage and high-temperature conditions.', 'For example, in 2015, \nXia et\xa0al.', '(2) demonstrated that a TAP-based electrolyte forms thick \nprotective layers on both electrodes in graphite/NCM442-based \ncells as evidenced by reduced self-discharge at high temperatures, \nreduced gas production during high-temperature cycling, and in-\ncreased Rct.', 'Although TAP has been explored previously, it is chosen \nin the present work to create resistive Li-ion cells as an example to \nillustrate the concept of realizing high safety and high performance \nsimultaneously.', 'Other ways to build resistive cells, such as using elec-\ntrode materials of low Brunauer-Emmett-Teller areas, are effective \nas well.', 'RESULTS\nAs a proof of concept, we prepare a baseline cell composed of a stan-\ndard electrolyte, 1\xa0M LiPF6 in ethylene carbonate (EC)/ethyl methyl \ncarbonate (EMC) (3/7 wt) + 2 weight % (wt %) vinylene carbonate \n(VC), and a graphite anode and an NCM622 cathode.', 'Protective \nlayers of solid electrolyte interphase (SEI) on graphite and cathode \nelectrolyte interphase (CEI) are formed during initial charge/discharge \ncycles.', 'To yield low DCR and hence high power, these interfacial \nlayers are usually thin, lacking sufficient density and resilience to \nresist decomposition under abuse conditions, to suppress continuous \nreaction of the solvent EC in the SEI, or to hinder continuous oxida-\ntion of EC with oxygen released from cathode materials in the CEI, \nthereby leading to Li consumption and loss of cell capacity.', 'In contrast, \nin SEB cells, we create highly stable, flame-retardant EEIs through \nthe addition of a small amount of TAP in the standard electrolyte.', 'This electrolyte modification is accompanied by the simultaneous \nreduction of EC content, i.e., EC/EMC (1/9 wt) + 2\xa0wt % VC, intended \nfor further reduction in gas production via side reactions.', 'In this work, \nwe present results for three prototype SEB cells, identified as SEB-1, \n1 of 8\n\nGe et al., Sci.', 'Adv.', '2020; 6 : eaay7633     28 February 2020SCIENCE ADVANCES | RESEARCH ARTICLE\x0ctemperature, indicating that the SEB cell is >30× more stable and \nhence safer than the baseline cells under high-voltage charging con-\nditions.', 'Calendar life testing further shows that the passivated SEB \ncells can effectively suppress self-discharge.', 'The self-discharge cur-\nrent is measured by holding the cell voltage constant at a required \nvalue, such as 4.187 V for 100% state of charge (SOC).', 'Both base-\nline and SEB cells display a fast decrease of self-discharge current \ndensity in the beginning of calendar aging due to anode SEI layer \ngrowth.', 'Fresh SEB cells show self-discharging currents ~5× lower than \nthe baseline cell at room temperature and 50% SOC, 6× lower at room \ntemperature and 100% SOC, and ~7× lower at 60°C at both 50 and \n100% SOC.', 'After 60 days of storage, the self-discharge current be-\ngins to plateau; however, the self-discharge current in the SEB cell is \nstill 2× lower than that in the baseline cell at room temperature and \n50% SOC and 3× lower at room temperature and 100% SOC.', 'The \nself-discharge becomes 4× lower at high temperature (60°C) at both \n50 and 100% SOC, indicating that at elevated temperatures, the SEB \ncells maintain superiority in calendar life over the baseline cell.', 'A distinctive feature of SEB cells is high power on demand.', 'When batteries are not in operation, SEB cells are left idle at room \ntemperature, exhibiting high stability and safety.', 'However, upon \noperation, a SEB cell switches to high-reactivity conditions through \nrapid thermal stimulation.', 'This can be illustrated through DCR, \nwhich is inversely proportional to power performance.', 'Here, DCRs \nupon discharge and charge for SEB and baseline LIB cells are mea-\nsured at 50% SOC by a 10-s hybrid pulse power characterization \n(HPPC) method.', 'As expected, the DCR values increase substantially \nby adding a small amount of TAP in the electrolyte, as shown in \nFig.', '3\xa0(A\xa0and\xa0B), which also demonstrates that adding more TAP in \nSEB-3 (e.g., 1.5\xa0wt %) results in a further increase in DCR, owing to \nthe formation of thicker protective layers.', 'Figure\xa03C shows the relative power of SEB cells versus the base-\nline LIB, where the power of the baseline cell at room temperature \nis normalized to be unity, thus demonstrating that SEB cells can \nprovide sufficient, higher power by operating at higher tempera-\ntures.', 'At 50% SOC, SEB-1 operated at 29.2°C provides the same \npower as the baseline cell at room temperature.', 'The safest cell, SEB-3, \nwith 1.5\xa0wt % TAP additive requires an operation temperature of \n44.6°C to deliver the same power as the baseline cell at room tem-\nperature.', 'The baseline cell has a narrow temperature window of 15° \nto 35°C, whereas the SEB cells have much higher maximum operating \ntemperature without becoming excessively reactive.', 'Thus, the ne-\ncessity of higher temperature operation for power recovery does not \npose an issue for SEB cells.', 'At 50% SOC, the discharge power boost \nover the baseline Li-ion cell is 2.05, 1.81, and 1.39 for the SEB-1, \nSEB-2, and SEB-3, respectively (Fig.', '3C).', 'Moreover, at the ambient \ntemperature of 0°C, the baseline LIB cell has a relative power of \n0.38, while SEB-2 has a relative power of 1.81 when operated at \n60°C [it takes 30 s to heat up a cell from 0° to 60°C at a speed of \n2°C/s, typical of the self-heating structure of Wang et\xa0al.', '(1)].', 'This is \n~5× boost in power over the LIB cell for SEB cells working in the \nfreezing environments.', 'Generally, SEB cells perform independent \nof ambient temperatures or weather, as they are always heated up in \na matter of seconds and operate at a constant elevated temperature.', 'Although heating a cell to an elevated temperature consumes cell \nenergy, the total deliverable energy of a cell, counterintuitively, \nis not reduced.', 'As shown in fig.', 'S3, the C/3 discharge energy was \n9.62 watt-hours (Wh) for the baseline cell at room temperature and \n10.15 Wh for the SEB-3 cell at 60°C.', 'According to our previous \n2 of 8\n\nFig.', '1.', 'Principles and advantages of a SEB versus a conventional LIB.', 'DCR is \nshown to vary with the inverse of temperature for both batteries, where the upper \ncurve for the passivated SEB is always safer due to higher DCR.', 'The SEB can, however, \nachieve a similar power output to the LIB by thermal stimulation before operation, \nshown as going from point b to c.\n\nSEB-2, and SEB-3 and corresponding to 0.5, 1, and 1.5\xa0wt % TAP, \nrespectively.', 'The charge-transfer resistance of the SEB cells, measured \nby electrochemical impedance spectroscopy (EIS), increases by 3× \nto 5× as compared to the baseline cell without the electrolyte addi-\ntives, as shown in Fig.', '2A.', 'The high impedance comes from the \npolymerization of TAP molecules that form thick and dense inter-\nfacial films at the surfaces of both the anode and cathode (2).', 'On \nthe anode side, the film serves as an enhanced SEI layer to stabilize \nfurther growth.', 'On the cathode side, the film hinders EC in the elec-\ntrolyte from reacting with lattice oxygen on the NCM surface at high \ntemperature or high voltage (3), as shown schematically in Fig.', '2B.', 'PolyTAP and its composites are flame-retardant materials.', 'In \naddition, the PolyTAP has electrically insulative properties and sta-\nbility at elevated temperatures (4).', 'All these properties make PolyTAP \nwell suited to enhance the safety of SEB cells under abuse conditions.', 'This is confirmed by nail penetration tests shown in Fig.', '2C, where \nthe temperature rise is only 100°C (without cell deformation, smoking, \nor fire) versus almost 1000°C (catching fire) in the baseline case \n(Fig.', '2D).', 'The superior safety performance of the SEB cell is further \nillustrated by the electrical performance of the cell during nail pen-\netration.', 'For the baseline cell, the cell voltage decreases to 0.1 V \nwithin 5 s after the nail penetration.', 'The sharp voltage drop indi-\ncates that the shorting current is extremely high due to low internal \nresistance of the baseline cell.', 'In contrast, the voltage of the SEB cell \ndrops from 4.171 to 3.085 V within 5 s after penetration, showing a \nslow and controlled discharge due to the high internal resistance of \nthe SEB cell.', 'The substantially higher resistance in the SEB cell than \nthe baseline is caused by the electrically nonconductive layers formed \non the surface of the graphite and NCM particles, as schematically \nshown in Fig.', '2B.', 'The stability and safety of SEB cells are further evident in \nhigh-voltage charge and high-temperature calendar life tests (figs.', 'S1 \nand S2).', 'The SEB cells can undergo 1254 exposures to high-voltage \nabuse [constant current (CC) charge to 4.4 V, constant voltage (CV) \nto C/20] at 40°C with a capacity retention of 80%, whereas the baseline \ncell sustains only 40 cycles at the same capacity loss and operating \n\nGe et al., Sci.', 'Adv.', '2020; 6 : eaay7633     28 February 2020SCIENCE ADVANCES | RESEARCH ARTICLE\x0cFig.', '2.', 'Experimental comparison between LIB and SEB and mechanism explanation.', '(A) Nyquist plots showing measured charge-transfer resistances of SEB cells \nversus the baseline LIB cell.', '(B) Schematic showing the in situ formed interfacial layers on the surface of graphite and NCM particles.', 'The enhanced SEI layer on graphite \nslows down EC transport though the film and suppresses further SEI growth.', 'The CEI layer hinders EC oxidation with lattice oxygen over the NCM surface at high tempera-\ntures or high voltages.', '(C and D) Cell voltage and temperature evolutions during nail penetration of a SEB cell and the baseline LIB cell along with qualitative temperature \ndistributions.', 'Both cells are 2.8-Ah pouch cells composed of the same graphite anode and NMC622 cathode materials.', 'The baseline LIB cell is filled with a standard \nelectrolyte: 1 M LiPF6 in EC/EMC (3/7 wt) + 2 wt % VC.', 'The SEB cell has the electrolyte of 1 M LiPF6 in EC/EMC (1/9 wt) + 2 wt % VC with TAP as electrolyte additive.', 'RT, \nroom temperature.', 'work (1), it takes ~1% cell energy for a 10°C temperature rise, \nmeaning that the SEB-3 cell needs 3.5% of its energy for heating from \n25° to 60°C; the remaining discharge energy is thus 9.79 Wh, which \nis 2% higher than that of the baseline cell at room temperature.', 'As mentioned, such rapid heating is achieved through the inser-\ntion of a micrometer-thick nickel sheet.', 'The impact of this addi-\ntional component on cell power density can be evaluated through \nEq.', '1, where m is the mass of the given cell and the ratio of discharge \npower evaluated through HPPC is inversely proportional to the \nratio of DCRs\n\n \n\n \n\n \n \n\n ─ \n\n─────────────── \n    \n\n (Power Density)  baseline@RT      =   (     DC  R  baseline@RT   \n (Power Density)  SEB   \n\nDC  R  SEB@60°C     )   (      m  baseline   ─ \n m  SEB      )     (1)\nThe addition of the nickel foil increases cell mass by 1.3%, while \nthe ratio of DCRs in Eq.', '1 is the relative power presented in Fig.', '3C \nat 60°C.', 'On the basis of these values, the power density is not re-\nduced but rather increased by a factor of 2.02, 1.79, and 1.37\xa0in the \ncase of SEB-1, SEB-2, and SEB-3, respectively.', 'Thus, in the case of \nSEB-3 with the highest internal resistance, the power density is still \n37% higher than the baseline cell operated at room temperature.', 'Figure\xa03D presents the relative interfacial reactivity as derived \nfrom the charge-transfer resistance of EIS results.', 'This further con-\nfirms that increased operational temperature effectively decreases \nthe charge-transfer resistance of SEB cells and consequently increases \nthe cell reactivity and power.', 'The reactivity boost over the baseline \n\nLIB cell is 2.06 for SEB-2.', 'Overall, both relative power and relative \nreactivity illustrate the ability of SEB cells to achieve high power on \ndemand.', 'On the other hand, the values of relative reactivity of these \ntwo cells shown in Fig.', '3D indicate that the SEB cell will be 5× safer \nand undergo 5× less aging at room temperature, which is supported \nby self-discharge and capacity retention tests displayed in fig.', 'S2.', 'Furthermore, the comparison of relative reactivity for the two cell \ntypes at 60°C indicates that the SEB cell is more than 2× less prone \nto thermal runaway.', 'Beyond safety and high power, there is an ever-increasing need \nfor long cycle life of LIBs.', 'Low-reactivity SEB cells built with highly \nstable materials offer longer calendar life (fig.', 'S2), evident from the \nslow capacity fade when the battery is in idle conditions at room \ntemperature.', 'At elevated temperatures, cycling of SEB cells is also \nstable.', 'Figure\xa04A compares capacity retention of the baseline cell \nwith SEB cells during cycling at 60°C of 1C CC charge to 4.2 V CV \ncharge till C/20 and then 1C discharge to 2.8 V. Clearly, the SEB cells \noutperform the baseline cell as evidenced by a 20% capacity loss \nat 481 cycles with visible signs of cell deformation due to gas evolu-\ntion and graphite anode swelling for the baseline cell, while SEB-3 \ncan achieve 2821 cycles before reaching 20% capacity loss.', 'This cor-\nresponds to ca.', '6× improvement in cycle life.', 'In addition, SEB-3 \nachieves 4014 cycles at 75% capacity retention while still showing \nsigns of a healthy cell capable of cycling stably (no perceptible gassing \nor lithium plating).', 'The average discharge capacity of these 4014 cycles \nis 84.2% of an equivalent full cycle (EFC).', 'Assuming a 153-mile \n3 of 8\n\nGe et al., Sci.', 'Adv.', '2020; 6 : eaay7633     28 February 2020SCIENCE ADVANCES | RESEARCH ARTICLE\x0cFig.', '3.', 'Temperature dependence of cell power performance.', '(A and B) DCRs of discharge and charge, respectively, at 50% SOC for SEB cells versus the baseline LIB cell.', '(C) Relative discharge power (DCRbaseline@RT/DCR) of a SEB cell versus the baseline LIB cell.', '(D) Relative reactivity (Rct,baseline@RT/Rct) of a SEB cell versus the baseline LIB cell, \nshowing that SEB cells operated at appropriate elevated temperatures, e.g., SEB-3 at 50°C, can deliver sufficient power at all ambient temperatures (labeled as line a), that \nSEB cells are 5× safer and less aging at room temperature (labeled as line b), and that SEB cells are 2.6× less prone to thermal runaway at 60°C (labeled as line c).', 'driving range per EFC for an electric vehicle (e.g., 2019 BMW i3), \nthe 4014 cycles mean >517,000 miles of lifetime.', 'That is more than \n5× the warranty for commercial electric cars (e.g., BMW i3, 70% \ncapacity for 8 years or 100,000 miles).', 'Improvement of cell lifetime \ncan be further demonstrated by considering the rate of capacity fade \nduring calendar aging at room temperature (i.e., the stable state), \nwhich is 7× lower than that at 60°C (the reactive state).', 'The SEB cell \nwill only be heated to the reactive state for situations requiring high \npower or fast charging.', 'The greater part of its lifetime (>90%) would \nbe spent in idle conditions (the stable state).', 'Therefore, in the field, \nthe SEB cycle life is expected to extend much beyond 4014 cycles \nbefore reaching 25% capacity loss.', 'The tremendous extension of cycle life of the SEB over the base-\nline cell may stem from the denser and more stabilized SEI layer \nformed on graphite particles and CEI layer on NCM622 particles in \nthe presence of the electrolyte additives.', 'For the baseline cell, the \nnickel-rich NCM particles are prone to microcracks along grain \nboundaries (5,\xa06) that provide gaps for electrolyte penetration and \nlead to more severe electrolyte oxidation and rock salt formation \n(7).', 'Particle cracking also sets free new, fresh surface area from \nwhich oxygen can be released (8).', 'The microcracks on NCM622 \nparticles for the baseline cell are observed after only 50 cycles \n(Fig.', '5D).', 'The formation of cracks becomes much more notable \nthroughout the micrograph domain after 956 cycles (Fig.', '5F).', 'For \nthe SEB cells, cracks on NCM particles are not observed at 50 cycles, \nand a small amount of cracks are observed after 4021 cycles (Fig.', '5J).', 'The presence of microcracks not only causes a loss of contact but \nalso accelerates NCM capacity fade.', 'For the SEB cells, the polymer \ncoating from TAP likely forms a robust CEI, reducing formation of \nmicrocracks (Fig.', '2B).', 'This is also evidenced by optical images (fig.', 'S4) \nand no observable cell deformation or swelling after 4021 cycles even \nat an elevated temperature of 60°C.', 'The use of the TAP additives altered the EEI compositions \nmarkedly.', 'We performed x-ray photoelectron spectroscopy (XPS) \non the graphite and NCM electrodes after 4021 cycles and compared \nthe EEI structure of the samples from the baseline cell after 956 cycles.', 'At the graphite anode, the TAP-containing electrolyte-derived \nSEI contains high concentrations of C, O, and P elements, compared \nto the control sample of the baseline cell (Fig.', '6 and fig.', 'S5).', 'In \naddition, a higher content of C─C species in the entire C-containing \nspecies was also found in the TAP-containing electrolyte-derived \nSEI, suggesting the decomposition of the TAP at the anode (fig.', 'S5).', 'Meanwhile, the composition of the cathode CEI was also changed.', 'With the TAP additive, the SEI layer has more C, P, and F and less \nLi and O, compared to the baseline SEI (Fig.', '6 and fig.', 'S5).', 'High \ncontents of LiF and P-containing species (O─P═O, LixPyOFz, and \nLixPyFz) (9) were found in the cathode CEI layer, owing to the use of \nthe TAP additive (Fig.', '6).', 'A detailed peak interpretation is as follows: \npeaks at 284.6, 286.1, 288.8, and 290.1 eV in the C 1s spectrum are \nattributed to C─C, C─O, O─C═O, and poly(O─C═O) (10), re-\nspectively; peaks at 684.6 eV in the F 1s spectrum are attributed to \nLiF; peaks at 686.9 eV in the F 1s spectrum and 136.7 eV in the \n4 of 8\n\nGe et al., Sci.', 'Adv.', '2020; 6 : eaay7633     28 February 2020SCIENCE ADVANCES | RESEARCH ARTICLE\x0cFig.', '4.', 'Comparison of cycling stability at 60°C.', '(A and B) Capacity retention and DCR of the SEB cells versus the baseline LIB cell during cycling at 60°C.', 'The cells are charged \nwith CCCV protocol at 1 C to 4.2 V with a cutoff current of C/20 and then discharged at 1 C to 2.8 V. (C and D) Discharge curves of the fresh SEB cell versus aged cell.', 'P 2p spectrum are attributed to O─P═O and LixPyOFz; and peaks at \n686.3 eV in the F 1s spectrum and 134.5 eV in the P 2p spectrum are \nattributed to LixPyFz.', 'In comparing O 1s spectrum of the aged baseline \n(956 cycles) and SEB-3 electrodes (4021 cycles), the peak at 529.2 eV \nfor SEB-3 cathode is effectively eliminated in comparison to that \nfor the baseline cathode (Fig.', '6).', 'This indicates that less lattice oxygen \nof NCM622 is detected for SEB-3 cathode attendant to a relatively \nthicker CEI layer.', 'This is consistent with the thicker CEI layer de-\ntected by Xia et\xa0al.', '(2) on the coated NMC442 surface in the pres-\nence of TAP additive.', 'Thus, the XPS result confirms that the TAP \nadditive leads to a thick CEI layer and, in consequence, slower crack \nevolution, less gas generation, and longer cycle life.', 'The three SEB cells show very close rates of capacity fade within \n1000 cycles (Fig.', '4A).', 'Beyond 1000 cycles, SEB-3 shows a much \nlower rate of capacity fade than the other two SEB cells, as expected \nfrom its lowest reactivity.', 'In comparison with the baseline cell, the \nstability and long cycle life of SEB cells are clearly evident, the rea-\nsons of which can be delineated from the differences in the capacity \nretention trend at various stages of aging.', 'For the baseline cell, we \nsee a sharp decrease in C/3 capacity retention during the initial \nstage and a slow decrease in the secondary stage.', 'This is primarily \nattributed to the loss of lithium inventory during the quick and slow \ngrowth of the SEI layer.', 'For the SEB cells, the capacity fade is linear \nwith cycle number, indicating that there is no quick growth in the \ninitial stage of aging as SEI layer growth is suppressed by in situ \nformation of the flame-retardant protective layer.', 'Moreover, a sharp, \nnonlinear capacity loss due to lithium plating can usually be observed \nin the baseline cell at room temperature and at low temperatures in \n\nfinal stages of cell aging (11).', 'In all SEB cells operated at 60°C, this \nlithium plating-induced capacity loss is, however, absent, indicative \nof no lithium plating in SEB cells.', 'The SEB cell free of Li plating \noffers a significant improvement in safety over conventional LIB cells.', 'There is another advantage of SEB cells promising for ultralong \ncycle life when deployed in the field.', 'While conventional LIB cells \nundergo a large environmental temperature swing, the SEB cells \nalmost always operate at a single, constant temperature (say 60°C) \nregardless of ambient temperatures and after an extremely short \nperiod of initial transition by self-heating (on the order of tens of \nseconds).', 'The latter feature guarantees minimal damage of battery \nmaterials in SEB cells caused by wide temperature variations.', 'For the baseline cell, the capacity loss at elevated temperatures is \nmainly due to SEI growth on the anode side and solvent oxidation \non the cathode side.', 'As a consequence, the DCR increases markedly \nwith cycle number (Fig.', '4B).', 'In the case of SEB cells, the DCR of the \nfresh cell is initially much larger than the baseline cell; however, its \nrate of increase is much slower due to the protective coating on both \nthe anode and cathode (Fig.', '2B).', 'Figure\xa04\xa0(C\xa0and\xa0D) shows discharge \ncurves of the fresh SEB cell versus the aged cell, respectively.', 'Be-\ncause of its DCR increase with cycle number, SEB-3 shows slight \npower fade after 2821 cycles at 60°C.', 'In contrast, the baseline cell \nshows a drastic DCR increase and, hence, substantial power loss \nwithin only 556 cycles (fig.', 'S6).', 'For all the SEB cells with TAP addi-\ntive, their DCRs increase linearly and increased additive content \nleads to higher DCR in fresh cells but slower DCR evolution with \ncycle number (Fig.', '4B).', 'The SEB cells do not produce gas during \ncycling tests, yielding more safety than the baseline cell.', 'In addition, \n5 of 8\n\nGe et al., Sci.', 'Adv.', '2020; 6 : eaay7633     28 February 2020SCIENCE ADVANCES | RESEARCH ARTICLE\x0ction.', 'SEB cells initially contains 10\xa0wt % EC in the electrolyte.', 'Some \nEC is consumed during the formation cycle, resulting in an EC \ncontent much less than 10% in formed SEB cells.', 'This is advanta-\ngeous since the rate of gas generation in an EC-less electrolyte \nwould be lower than that in an electrolyte with high EC content.', 'With the introduction of new materials into the electrolyte, the \neffect on cell cost, weight, and fabrication should be evaluated.', 'The \nelectrolyte additive, TAP, has a comparable price and density when \ncompared to current standard solvents; thus, no notable material \ncost difference is expected with the introduction of SEB electrolytes.', 'Unlike superconcentrated electrolytes (14), the electrolytes with \nTAP do not increase the viscosity compared to the standard electro-\nlyte.', 'From a fabrication perspective, the SEB electrolytes will also \nadd no additional cost due to the similarity in processing during \nand after introduction into a cell.', 'Last, the SEB cells offer an important benefit associated with \nthermal management of a battery pack.', 'When high power is re-\nquired, the SEB cells are to be heated internally (1) and operated at \nelevated temperatures.', 'Assuming the environmental temperature is \n25°C, and the SEB and baseline cells operate at 60° and 30°C, re-\nspectively, the SEB provides a temperature difference driving heat \ndissipation that is 7× larger than the baseline case.', 'Further, the \nSEB cell has lower DCR at its operational temperature of 60°C \n(17.1 ohm·cm2 for SEB with 1\xa0wt % TAP) than the baseline cell at \n30°C (25.3 ohm·cm2), indicating ~1.5× lower heat generation at the \nsame current.', 'The combination of these two factors eases the burden \nof thermal management by a factor of approximately 10 for SEB cells.', 'Fig.', '5.', 'SEM micrographs of the pristine, aged electrodes for the baseline and \nSEB-3 cells.', '(A) Pristine anode.', '(B) Pristine cathode.', '(C) Baseline anode after 50 cycles.', '(D) Baseline cathode after 50 cycles.', '(E) Baseline anode after 956 cycles.', '(F) Baseline \ncathode after 956 cycles.', '(G) SEB-3 anode after 50 cycles.', '(H) SEB-3 cathode after 50 cycles.', '(I) SEB-3 anode after 4021 cycles.', '(J) SEB-3 cathode after 4021 cycles.', 'ETD,  Everhart- \nThornley Detector; HV, electron accelerating voltage; WD, working distance; HFW, \nhorizontal field width.', 'cells containing TAP produce less gas during formation than the \nbaseline cell (2,\xa012).', 'Because of the high-voltage tolerance of SEB cells, when charged \nto a high voltage of 4.4 V as compared to 4.2 V, the SEB cell dis-\ncharge capacity increases 12.7%, and its discharge energy increases \n14.5% (fig.', 'S1B).', 'Thus, the high-voltage tolerance can be used to \nincrease cell energy density.', 'The electrolytes for SEB cells were formulated by reducing EC \ncontent and adding TAP as an additive.', 'Although EC is an essential \nsolvent for SEI layer formation, it also leads to gas generation, \nespecially at high voltage (13).', 'Calendar aging testing shows that \nhigh-temperature and high-SOC conditions accelerate capacity fade \nand increase in internal resistance while also promoting gas genera-\n\nDISCUSSION\nOverall, the novel SEB offers high safety and stability at idle condi-\ntions, demonstrated through mechanical (nail penetration), electrical \n(high-voltage charge), and thermal abuse (high-temperature storage).', 'The stability of the SEB-2 cell under high voltage leads to 14.5% \nextra energy when it is charged to 4.4 V. It delivers 81% more dis-\ncharge power and 65% more charge power while operating at \n60°C.', 'Furthermore, SEB-3 achieved >4000 cycles at 60°C while still \nshowing signs of a healthy cell to continue operation.', 'In addition, \nthe high operational temperature facilitates thermal management.', 'All these key advantages demonstrated here (high safety, high power \non demand, long life, and easy thermal management) make SEB \ncells highly promising for widespread application to, e.g., the \nnext-generation electric vehicles.', 'In the context of electric vehicles, introduction of the SEB yields \na substantial gain in energy density and cost reduction at the pack \nlevel due to (i) simplified or entirely eliminated thermal manage-\nment system; (ii) enhanced safety of SEB cells, thus enabling removal \nof some or all safety devices installed in a battery pack; and (iii) \nsimplified battery management system due to SEB cells always \noperating at a single, constant temperature, rendering estimation of \nSOC, state of power, state of health, etc.', 'oversimplified.', 'We believe \nthat 25 to 30% gain in energy density and 30% reduction in cost for \nSEB battery packs are possible without introducing new chemistry \nor cell manufacturing.', 'More broadly, the proposed strategy of passivating a battery cell \nfor safety in idle conditions and then self-heating for high power \nbefore operation has a profound impact on future directions of battery \nmaterials development.', 'Within the SEB framework, the high reac-\ntivity of active materials and electrolytes as well as of EEIs is no longer \n6 of 8\n\nGe et al., Sci.', 'Adv.', '2020; 6 : eaay7633     28 February 2020SCIENCE ADVANCES | RESEARCH ARTICLE\x0cFig.', '6.', 'XPS core spectra comparison for aged baseline and SEB-3 electrodes.', 'The graphite and NCM622 electrodes are taken from the baseline cell after 956 cycles \nand the SEB-3 cell after 4021 cycles.', 'a requirement.', 'The only requirement is stability at elevated tempera-\ntures.', 'Such a design strategy will lead to dormant cell performance \nat low temperatures or even low power at room temperature, but all \nthese deficiencies can be circumvented by the self-heating structure \n(1), which will take a SEB cell from low or room temperature to an \nelevated temperature suitable for high power output in tens of sec-\nonds.', 'On the other hand, these SEB cells enjoy ultrahigh safety and \nultralow degradation under all scenarios as well as robust cycling \nstability at elevated temperatures.', 'We hope that this unconventional \nstrategy will unleash a vast class of new materials for development \nof a disruptive generation of LIBs.', 'MATERIALS AND METHODS\nWe fabricated 2.8-Ah pouch cells using LiNi0.6Co0.2Mn0.2O2 (Umicore) \nfor cathodes and graphite (Nippon Carbon) for anodes.', 'The capacity \nratio of negative to positive electrode, or NP ratio, was designed at \n1.2.', 'The 2.8-Ah pouch cell contains a stack of 20 anode and 19 cath-\node layers.', 'A Celgard-2325 separator of 25 \uf06dm in thickness was \nused.', 'The loadings of NMC622 on the positive electrode and graphite \non the negative electrode were 10.5 and 6.6 mg/cm2, respectively.', 'The cathodes were prepared by coating an N-methyl-2-pyrrolidone- \nbased–based slurry onto 15-\uf06dm-thick Al foil, whose dry material \nconsists of NCM622 (91.5\xa0wt %), Super-P (TIMCAL) (4.1\xa0wt %), \n\nand polyvinylidene fluoride (Arkema) (4.4\xa0wt %) as a binder.', 'The \nanodes were prepared by coating deionized water–based slurry onto \n10-\uf06dm-thick Cu foil, whose dry material consists of graphite (95.4\xa0wt \n%), Super-P (1.0\xa0wt %), styrene-butadiene rubber (Zeon) (2.2\xa0wt %), \nand carboxymethyl cellulose (Dai-Ichi Kogyo Seiyaku) (1.4\xa0wt %).', 'One molar of LiPF6 dissolved in EC/EMC (3:7 by wt) + 2\xa0wt % \nVC was used as control electrolyte (BASF).', 'One molar of LiPF6 \ndissolved in a mixture of EC/EMC + 2\xa0wt % VC was mixed in-house.', 'To build SEB cells, 0.5 to 1.5\xa0wt % TAP were mixed into the conven-\ntional electrolyte as additives.', 'Each pouch cell has a 110\xa0mm × 56\xa0mm footprint area, weighs 63 g, \nand has 2.8-Ah nominal capacity with a specific energy of 166 Wh/kg \nand an energy density of 310 Wh per liter.', 'Discharge performance \nof the baseline and SEB cells at room temperatures is shown in fig.', 'S7 \nas a function of C-rate.', 'Cycle aging tests of the pouch cells were performed using a Land \ninstrument battery testing system (Model CT2001B, Land Instru-\nments).', 'A forced-air oven was used to control different ambient \ntemperatures.', 'For each aging cycle, the cell was charged to 4.2 V at \na constant current of 2.8 A (1C-rate) and then charged at a constant \nvoltage of 4.2 V until the current decreased to 0.14 A (C/20).', 'After \nresting for 5 min, the cell was discharged to 2.8 V at a constant cur-\nrent of 2.8 A (1C-rate) followed by a final rest period of 5\xa0min.', 'When the aging cycle number reached a specific value (e.g., 403, \n7 of 8\n\nGe et al., Sci.', 'Adv.', '2020; 6 : eaay7633     28 February 2020SCIENCE ADVANCES | RESEARCH ARTICLE\x0c1006 cycles), the cell was cycled at a charge and discharge rate of C/3 \nto determine the capacity (designated as C/3 capacity) of the cell.', 'For impedance tests at different temperatures, the cells were fully \ncharged and then discharged at a rate of C/3 to 90% SOC.', 'Imped-\nance testing was performed with an AC voltage amplitude of 5 mV \nin the frequency range of 50\xa0kHz to 0.005\xa0Hz.', 'For DCR test, the cells \nwere fully charged and then discharged to 50% SOC at C/3-rate.', 'A \ndischarge rate of 5C and a charge rate of 3.75C were used to deter-\nmine the value of DCRDischarge and DCRCharge.', 'Calendar aging tests were performed at different ambient tem-\nperatures and SOCs.', 'The forced-air oven was used to control differ-\nent ambient temperatures.', 'The cell voltage was kept constant, and \nthe current was collected.', 'When the calendar aging time reached a \nspecific value (e.g., 25, 60, 120, and 180 days), the cell was cycled at \na charge and discharge rate of C/3 to determine capacity of the cell.', 'Then, impedance and DCR tests were conducted under the same \nconditions as that for the cycle-aged cells.', 'For the nail penetration test, the cell was fully charged (1C CCCV \ncharge with a cutoff voltage of 4.2 V and a cutoff current of C/20).', 'Thermocouples were placed at 10\xa0mm to the geometry center of the \ncell and at the negative tab of the cell.', 'The nail diameter is 5 mm, made \nof heat-resisting steel (point angle of the nail is 60°; nail surface is clean, \nwithout rust or oil).', 'Rate of penetration was at 30 mm/s; the nail \npenetrated through the geometrical center of the electrode plane \nperpendicularly and stayed inside the cell.', 'The observation time was 1 hour \nuntil the cell cooled down and the cell voltage dropped to nearly zero.', 'The SEM and XPS analyses were performed by first extracting \nthe electrode samples from the fully discharged graphite/NCM622 \npouch cells after cycling and washing 3× with EMC.', 'XPS tests were \nconducted on a PHI VersaProbe II Scanning XPS Microprobe.', 'The \nsamples were loaded in a glove box and transferred into the instru-\nment through a vacuum transfer vessel.', 'SEM imaging was performed \non an FEI Nova NanoSEM 630 SEM instrument.', 'SUPPLEMENTARY MATERIALS\nSupplementary material for this article is available at http://advances.sciencemag.org/cgi/\ncontent/full/6/9/eaay7633/DC1\nFig.', 'S1.', 'Cycling behavior of LIB and SEB under over-charge conditions.', 'Fig.', 'S2.', 'Calendar life testing for LIB and SEB.', 'Fig.', 'S3.', 'Comparison of cell discharge performance.', 'Fig.', 'S4.', 'Optical comparison of fresh and aged electrodes.', 'Fig.', 'S5.', 'Elemental concentration of EEI layers.', 'Fig.', 'S6.', 'Discharge curves of the fresh baseline cell versus the aged cell.', 'Fig.', 'S7.', 'Comparison of C-rate discharge curves for the baseline LIB cell and the three SEB cells \nat room temperature.', 'REFERENCES AND NOTES\n  1.', 'C.-Y.', 'Wang, G. Zhang, S. Ge, T. Xu, Y. Ji, X.-G. Yang, Y. Leng, Lithium-ion battery structure \n\nthat self-heats at low temperatures.', 'Nature 529, 515–518 (2016).', '2.', 'J. Xia, L. Madec, L. Ma, L. D. Ellis, W. Qiu, K. J. Nelson, Z. Lu, J. R. Dahn, Study of triallyl \n\nphosphate as an electrolyte additive for high voltage lithium-ion cells.', 'J.', 'Power Sources \n295, 203–211 (2015).', '3.', 'R. Jung, M. Metzger, F. Maglia, C. Stinner, H. A. Gasteiger, Oxygen release and its effect \non the cycling stability of LiNixMnyCozO2 (NMC) cathode materials for Li-ion batteries.', 'J. Electrochem.', 'Soc.', '164, A1361–A1377 (2017).', '4.', 'S. Gaan, G. Sun, Effect of phosphorus flame retardants on thermo-oxidative \n\ndecomposition of cotton.', 'Polym.', 'Degrad.', 'Stab.', '92, 968–974 (2007).', '5.', 'L. de Biasi, A. O. Kondrakov, H. Geßwein, T. Brezesinski, P. Hartmann, J. Janek, Between \nscylla and charybdis: Balancing among structural stability and energy density of layered \nNCM cathode materials for advanced lithium-ion batteries.', 'J. Phys.', 'Chem.', 'C 121, \n26163–26171 (2017).', '6.', 'H.-H. Ryu, K.-J.', 'Park, C. S. Yoon, Y.-K. Sun, Capacity fading of Ni-rich Li[NixCoyMn1−x−y]O2 \n(0.6 ≤ x ≤ 0.95) cathodes for high-energy-density lithium-ion batteries: Bulk or surface \ndegradation?', 'Chem.', 'Mater.', '30, 1155–1163 (2018).', '7.', 'C.-H. Shen, Q. Wang, H.-J.', 'Chen, C.-G. Shi, H.-Y.', 'Zhang, L. Huang, J.-T. Li, S.-G. Sun, In situ \nmultitechnical investigation into capacity fading of high-voltage LiNi0.5Co0.2Mn0.3O2.', 'ACS Appl.', 'Mater.', 'Interfaces 8, 35323–35335 (2016).', '8.', 'R. Jung, F. Linsenmann, R. Thomas, J. Wandt, S. Solchenbach, F. Maglia, C. Stinner, \n\nM. Tromp, H. A. Gasteiger, Nickel, manganese, and cobalt dissolution from Ni-rich NMC \nand their effects on NMC622-graphite cells.', 'J. Electrochem.', 'Soc.', '166, A378–A389 \n(2019).', '9.', 'Y. Gao, Z. Yan, J. L. Gray, X.', 'He, D. Wang, T. Chen, Q. Huang, Y. C. Li, H. Wang, S. H. Kim, \n\nT. E. Mallouk, D. Wang, Polymer–inorganic solid–electrolyte interphase for stable \nlithium metal batteries under lean electrolyte conditions.', 'Nat.', 'Mater.', '18, 384–389 \n(2019).', '10.', 'Y. Gao, Y. Zhao, Y. C. Li, Q. Huang, T. E. Mallouk, D. Wang, Interfacial chemistry regulation \n\nvia a skin-grafting strategy enables high-performance lithium-metal batteries.', 'J.', 'Am.', 'Chem.', 'Soc.', '139, 15288–15291 (2017).', '11.', 'Y. Leng, S. Ge, D. Marple, X.-G. Yang, C. Bauer, P. Lamp, C.-Y.', 'Wang, Electrochemical \n\ncycle-life characterization of high energy lithium-ion cells with thick Li(Ni0.6Mn0.2Co0.2)O2 \nand graphite electrodes.', 'J. Electrochem.', 'Soc.', '164, A1037–A1049 (2017).', '12.', 'Q. Shi, G. Sheng, Electrolyte for lithium ion battery.', 'CN103594729A (2014).', '13.', 'L. Ma, S. L. Glazier, R. Petibon, J. Xia, J. M. Peters, Q. Liu, J. Allen, R. N. C. Doig, J. R. Dahn,  \n\nA guide to ethylene carbonate-free electrolyte making for Li-ion cells.', 'J. Electrochem.', 'Soc.', '164, A5008–A5018 (2017).', '14.', 'Y. Yamada, J. Wang, S. Ko, E. Watanabe, A. Yamada, Advances and issues in developing \n\nsalt-concentrated battery electrolytes.', 'Nat.', 'Energy 4, 269–280 (2019).', 'Acknowledgments \nFunding: This work was partially supported by U.S. Department of Energy’s Office of Energy \nEfficiency and Renewable Energy under award number DE-EE0008447.', 'Author contributions: \nS.G. and C.-Y.W.', 'developed the concept and wrote the manuscript.', 'S.G. and R.S.L.', 'designed \nand built the cells.', 'S.G. built the test stand and carried out the performance characterization.', 'T.L.', 'carried out the nail penetration test.', 'Y.L.', 'performed the impedance analysis.', 'Y.G.', 'and \nDaiwei Wang carried out the XPS and SEM analysis.', 'All authors contributed to development of \nthe manuscript and to discussions as the project developed.', 'Competing interests: The \nauthors declare that they have no competing interests.', 'Data and materials availability: All \ndata needed to evaluate the conclusions in the paper are present in the paper and/or the \nSupplementary Materials.', 'Additional data related to this paper may be requested from  \nthe authors.', 'Submitted 16 July 2019\nAccepted 6 December 2019\nPublished 28 February 2020\n10.1126/sciadv.aay7633\n\nCitation: S. Ge, Y. Leng, T. Liu, R. S. Longchamps, X.-G. Yang, Y. Gao, D. Wang, D. Wang, C.-Y.', 'Wang, \nA new approach to both high safety and high performance of lithium-ion batteries.', 'Sci.', 'Adv.', '6, \neaay7633 (2020).', '8 of 8\n\nGe et al., Sci.', 'Adv.', '2020; 6 : eaay7633     28 February 2020SCIENCE ADVANCES | RESEARCH ARTICLE']