Although batteries fitted with a metal negative electrode are attractive for their higher energy density and lower complexity, the latter making them more easily recyclable, the threat of cell shorting by dendrites has stalled deployment of the technology1,2. Here we disclose a bidirectional, rapidly charging aluminium–chalcogen battery operating with a molten-salt electrolyte composed of NaCl–KCl–AlCl3. Formulated with high levels of AlCl3, these chloroaluminate melts contain catenated AlnCl3n+1– species, for example, Al2Cl7–, Al3Cl10– and Al4Cl13–, which with their Al–Cl–Al linkages confer facile Al3+ desolvation kinetics resulting in high faradaic exchange currents, to form the foundation for high-rate charging of the battery. This chemistry is distinguished from other aluminium batteries in the choice of a positive elemental-chalcogen electrode as opposed to various low-capacity compound formulations3,4,5,6, and in the choice of a molten-salt electrolyte as opposed to room-temperature ionic liquids that induce high polarization7,8,9,10,11,12. We show that the multi-step conversion pathway between aluminium and chalcogen allows rapid charging at up to 200C, and the battery endures hundreds of cycles at very high charging rates without aluminium dendrite formation. Importantly for scalability, the cell-level cost of the aluminium–sulfur battery is projected to be less than one-sixth that of current lithium-ion technologies. Composed of earth-abundant elements that can be ethically sourced and operated at moderately elevated temperatures just above the boiling point of water, this chemistry has all the requisites of a low-cost, rechargeable, fire-resistant, recyclable battery.
This is a preview of subscription content, access via your institution
Open Access articles citing this article.
A solution-to-solid conversion chemistry enables ultrafast-charging and long-lived molten salt aluminium batteries
Nature Communications Open Access 03 July 2023
Nano-Micro Letters Open Access 29 December 2022
Access Nature and 54 other Nature Portfolio journals
Get Nature+, our best-value online-access subscription
$29.99 / 30 days
cancel any time
Subscribe to this journal
Receive 51 print issues and online access
$199.00 per year
only $3.90 per issue
Rent or buy this article
Prices vary by article type
Prices may be subject to local taxes which are calculated during checkout
The datasets analysed and generated during the course of this study are included in the paper and its Supplementary Information.
Chen, S., Dai, F. & Cai, M. Opportunities and challenges of high-energy lithium metal batteries for electric vehicle applications. ACS Energy Lett. 5, 3140–3151 (2020).
Varzi, A. et al. Current status and future perspectives of lithium metal batteries. J. Power Sources 480, 228803 (2020).
Muldoon, J., Bucur, C. B. & Gregory, T. Quest for nonaqueous multivalent secondary batteries: magnesium and beyond. Chem. Rev. 114, 11683–11720 (2014).
Jayaprakash, N., Das, S. K. & Archer, L. A. The rechargeable aluminum-ion battery. Chem. Commun. 47, 12610–12612 (2011).
Geng, L., Lv, G., Xing, X. & Guo, J. Reversible electrochemical intercalation of aluminum in Mo6S8. Chem. Mater. 27, 4926–4929 (2015).
Kotetsu, T. et al. Reversible magnesium and aluminium ions insertion in cation-deficient anatase TiO2. Nat. Mater. 16, 1142–1148 (2017).
Xia, S., Zhang, X.-M., Huang, K., Chen, Y.-L. & Wu, Y.-T. Ionic liquid electrolytes for aluminum secondary battery: influence of organic solvents. J. Electroanal. Chem. 757, 167–175 (2015).
Gao, T. et al. A rechargeable Al/S battery with an ionic-liquid electrolyte. Angew. Chem. Int. Ed. 55, 9898–9901 (2016).
Cohn, G., Ma, L. & Archer, L. A. A novel non-aqueous aluminum sulfur battery. J. Power Sources 283, 416–422 (2015).
Yu, X. & Manthiram, A. Electrochemical energy storage with a reversible nonaqueous room-temperature aluminum-sulfur chemistry. Adv. Energy Mater. 7, 1700561 (2017).
Yu, X., Boyer, M. J., Hwang, G. S. & Manthiram, A. Room-temperature aluminum-sulfur batteries with a lithium-ion-mediated ionic liquid electrolyte. Chem 4, 586–598 (2018).
Reed, L. D., Ortiz, S. N., Xiong, M. & Menke, E. J. A rechargeable aluminium-ion battery utilizing a copper hexacyanoferrate cathode in an organic electrolyte. Chem. Commun. 51, 14397–14400 (2016).
Turcheniuk, K., Bondarev, D., Singhal, V. & Yushin, G. Ten years left to redesign lithium-ion batteries. Nature 559, 467–470 (2018).
Schoetz, T., Ponce de Leon, C., Ueda, M. & Bund, A. State of the art of rechargeable aluminum batteries in non-aqueous systems. J. Electrochem. Soc. 164, A3499–A3502 (2017).
Song, Y. et al. A long-life rechargeable Al ion battery based on molten salts. J. Mater. Chem. A 5, 1282–1291 (2017).
Chen, H. et al. Ultrafast all-climate aluminum-graphene battery with quarter-million cycle life. Sci. Adv. 3, eaao7233 (2017).
Zhang, Y., Liu, S., Ji, Y., Ma, J. & Yu, H. Emerging nonaqueous aluminum-ion batteries: challenges, status, and perspectives. Adv. Mater. 30, 1706310 (2018).
Midorikawa, R. Electrolytic refining of aluminum. II. The melting point of the system AlCl3-NaCl-KCl. Denki Kagaku 23, 127–129 (1955).
Lu, G., Lai, T., He, M. & Liu, X. Experimental measurement and thermodynamic optimization of the phase diagram of LiCl-NaCl-KCl system. Chin. Sci. Bull. 65, 641–648 (2019).
Stafford, G. R. The electrodeposition of an aluminum-manganese metallic glass from molten salts. J. Electrochem. Soc. 136, 635–639 (1989).
Howie, R. C. & Macmillan, D. W. The electrodeposition of aluminium from molten aluminium chloride/sodium chloride. J. Appl. Electrochem. 2, 217–222 (1972).
Koura, N. A preliminary investigation for an Al/AlCl3-NaCl/FeS2 secondary cell. J. Electrochem. Soc. 127, 1529–1531 (1980).
Berrettoni, M., Tossici, R., Zamponi, S., Marassi, R. & Mamantov, G. A cyclic voltammetric study of the electrochemical behavior of NiS2 in molten NaCl saturated NaAlCl4 melts. J. Electrochem. Soc. 140, 969–973 (1993).
Fellner, P., Chrenková-Paučírová, M. & Matiašovský, K. Electrolytic aluminium plating in molten salt mixtures based on AlCl3 I: influence of the addition of tetramethylammonium chloride. Surface Tech. 14, 101–108 (1981).
Li, Q., Hjuler, H. A., Berg, R. W. & Bjerrum, N. J. Electrochemical deposition and dissolution of aluminum in NaAlCl4 melts: influence of MnCl2 and sulfide addition. J. Electrochem. Soc. 137, 2794 (1990).
Weber, et al. Long cycle life and dendrite-free lithium morphology in anode-free lithium pouch cells enabled by a dual-salt liquid electrolyte. Nat. Energy 4, 683 (2019).
Louli, J. et al. Diagnosing and correcting anode-free cell failure via electrolyte and morphological analysis. Nat. Energy 5, 693 (2020).
Yang, H. et al. An aluminum-sulfur battery with a fast kinetic response. Angew. Chem. Int. Ed. 57, 1989–1902 (2018).
Akdeniz, Z., Pastore, G. & Tosi, M. P. An ionic model for molecular units in molten aluminium trichloride and alkali chloroaluminates. Phys. Chem. Liq. 32, 191–209 (1996).
Boxall, L. G., Jones, H. L. & Osteryoung, R. A. Solvent equilibria of AlCl3-NaCl melts. J. Electrochem. Soc. 120, 223–231 (1973).
Griffith, K. J., Wiaderek, K. M., Cibin, G., Marbelle, L. E. & Grey, C. P. Niobium tungsten oxides for high-rate lithium-ion energy storage. Nature 591, 556–563 (2018).
Lin, M.-C. et al. An ultrafast rechargeable aluminium-ion battery. Nature 520, 325–328 (2015).
Kim, D. J. et al. Rechargeable aluminium organic batteries. Nat. Energy 4, 51–59 (2019).
Cuisinier, M. et al. Sulfur speciation in Li-S batteries determined by operando X-ray absorption spectroscopy. J. Phys. Chem. Lett. 4, 3227–3232 (2013).
Cui, Y. et al. (De)Lithiation mechanism of Li/SeSx (x = 0–7) batteries determined by in situ synchrotron X-ray diffraction and x-ray absorption spectroscopy. J. Am. Chem. Soc. 135, 8047–8056 (2013).
Wang, C. & Hussey, C. L. Aluminum anodization in the low-melting LiAlBr4-NaAlCl4-KAlCl4 molten salt. ECS Trans. 64, 257–265 (2014).
Stafford, G. R. & Haarberg, G. M. The electrodeposition of Al-Nb alloys from chloroaluminate electrolytes. Plasmas Ions 1, 35–44 (1999).
Li, Q., Hjuler, H. A., Berg, R. W. & Bjerrum, N. J. Electrochemical deposition and dissolution of aluminum in NaAlCl4 melts. Influence of MnCl2 and sulfide addition. J. Electrochem. Soc. 137, 2794–2798 (1990).
Sato, Y., Fukasawa, M., Abe, K. & Yamamura, T. Surface tensions of basic NaCl-AlCl3 and LiCl-AlCl3 binary melts. Electrochemistry (Tokyo) 67, 563–567 (1999).
Wang, K. et al. Lithium-antimony-lead liquid metal battery for grid-level energy storage. Nature 514, 348–350 (2014).
Tripathy, P. K. et al. Aluminum electroplating on steel from a fused bromide electrolyte. Surf. Coat. Tech. 258, 652–663 (2018).
Kendall, J., Crittenden, E. D. & Miller, H. K. A study of the factors influencing compound formation and solubility in fused salt mixtures. J. Am. Chem. Soc. 45, 963–996 (1923).
Midorikawa, R. Studies on electrolytic refining of aluminum at low temperature in aluminum chloride baths. IV. Measurement of the densities of the ternary molten salts AlCl3-NaCl-KCl. Denki Kagaku 23, 352–355 (1955).
Wei, S. et al. Metal-sulfur battery cathodes based on PAN-sulfur composites. J. Am. Chem. Soc. 137, 12143–12152 (2015).
Salanne, M., Siqueira, L. J. A., Seitsonen, A. P., Madden, P. A. & Kirchner, B. From molten salts to room temperature ionic liquids: simulation studies on chloroaluminate systems. Faraday Discuss. 154, 171–188 (2012).
Heerman, L. & D'Olieslager, W. Potentiometric study of the solvent equilibria in AlCl3-N-n-butylpyridinium chloride melts. Inorg. Chem. 24, 4704–4707 (1985).
Øye, H. A. & Rytter, E. Raman spectra of KCl-AlCl3 melts and normal coordinate analysis of Al2Cl7-. Acta Chem. Scand. 25, 559–576 (1971).
Rytter, E., Øye, H. A., Cyvin, J., Cyvin, B. N. & Klæboe, C. P. Raman spectra of AlCl3-AlkCl and trends in species formation. J. Inorg. Nucl. Chem. 35, 1185 (1973).
Gao, J., Lowe, M. A., Kiya, Y. & Abruña, H. D. Effects of liquid electrolytes on the charge-discharge performance of rechargeable lithium/sulfur batteries: electrochemical and in-situ X-ray absorption spectroscopic studies. J. Phys. Chem. C 115, 25132–25137 (2011).
de Juan, A. & Tauler, R. Chemometrics applied to unravel multicomponent processes and mixtures: revisiting latest trends in multivariate resolution. Anal. Chim. Acta 500, 195–210 (2003).
Broux, T. et al. VIV disproportionation upon sodium extraction from Na3V2(PO4)2F3 observed by operando X-ray absorption spectroscopy and solid state NMR. J. Phys. Chem. C 121, 4103–4111 (2017).
Ravel, B. & Newville, M. ATHENA, ARTEMIS, HEPHAESTUS: data analysis for X-ray absorption spectroscopy using IFEFFIT. J. Synchrotron Radiat. 12, 537–541 (2005).
Kresse, G. & Furthmüler, J. Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set. Phys. Rev. B 54, 11169–11186 (1996).
Kresse, G. & Joubert, D. From ultrasoft pseudopotentials to the projector augmented-wave method. Phys. Rev. B 59, 1758–1775 (1999).
Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized gradient approximation made simple. Phys. Rev. Lett. 77, 3865–3868 (1996).
Harl, J., Schimka, L. & Kresse, G. Assessing the quality of the random phase approximation for lattice constants and atomization energies of solids. Phys. Rev. B 81, 115126 (2010).
Frisch, M. J. et al. Gaussian 09, Revision E.01 (Gaussian, Inc., 2009).
Chai, J.-D. & Head-Gordon, M. Long-range corrected hybrid density functionals with damped atom–atom dispersion corrections. Phys. Chem. Chem. Phys. 10, 6615–6620 (2008).
Cleaver, B. & Koronaios, P. Viscosity of the NaCl+ AlCl3 melt system, including the effect of added oxide. J. Chem. Eng. Data 39, 848–850 (1994).
Zheng, Y., Zheng, Y., Wang, Q., Wang, Z. & Tian, D. Density and viscosity of binary mixtures of 1-ethyl-3-methylimidazolium heptachlorodialuminate and tetrachloroaluminate ionic liquids. J. Chem. Eng. Data 62, 4006–4014 (2017).
Wilkes, J. S., Levisky, J. A., Wilson, R. A. & Hussey, C. L. Dialkylimidazolium chloroaluminate melts: a new class of room-temperature ionic liquids for electrochemistry, spectroscopy and synthesis. Inorg. Chem. 21, 1263–1264 (1982).
Wilkes, J. S., Frye, J. S. & Reynolds, G. F. Aluminum-27 and carbon-13 NMR studies of aluminum chloride-dialkylimidazolium chloride molten salts. Inorg. Chem. 22, 3870–3872 (1983).
Taulelle, F. & Popov, A. I. Aluminum-27 NMR study of some AlCl3− MCl molten systems. J. Solut. Chem. 15, 463–471 (1986).
Nakayama, Y. et al. Sulfone-based electrolytes for aluminium rechargeable batteries. Phys. Chem. Chem. Phys. 17, 5758–5766 (2015).
Angell, M. et al. High Coulombic efficiency aluminum-ion battery using an AlCl3-urea ionic liquid analog electrolyte. Proc. Natl Acad. Sci. USA. 114, 834–839 (2017).
Kravchyk, K. V., Wang, S., Piveteau, L. & Kovalenko, M. V. Efficient aluminum chloride–natural graphite battery. Chem. Mater. 29, 4484–4492 (2017).
Schneider, H. et al. On the electrode potentials in lithium-sulfur batteries and their solvent-dependence. J. Electrochem. Soc. 161, A1399 (2014).
Kaplan, L. H. & d’Heurle, F. M. The deposition of molybdenum and tungsten films from vapor decomposition of carbonyls. J. Electrochem. Soc. 117, 693 (1970).
Kilicaslan, A. et al. Hard titanium nitride coating deposition inside narrow tubes using pulsed DC PECVD processes. Surf. Coat. Tech. 377, 124894 (2019).
Yin, et al. Faradaically selective membrane for liquid metal displacement batteries. Nat. Energy 3, 127–131 (2018).
Nelson, P. A., Ahmed, S., Gallagher, K. G. & Dees, D. W. Modeling the Performance and Cost of Lithium-Ion Batteries for Electric-Drive Vehicles, 3rd edn (Argonne National Laboratory, 2019); https://doi.org/10.2172/1503280.
Eroglu, D., Zavadil, K. R. & Gallagher, K. G. Critical link between materials chemistry and cell-level design for high energy density and low cost lithium-sulfur transportation battery. J. Electrochem. Soc. 162, A982–A990 (2015).
Berg, E. J., Villevieille, C., Streich, D., Trabesinger, S. & Novak, P. Rechargeable batteries: grasping for the limits of chemistry. J. Electrochem. Soc. 162, A2486–A2475 (2015).
Liao, Q., Sun, B., Liu, Y., Sun, J. & Zhou, G. A techno‐economic analysis on NaS battery energy storage system supporting peak shaving. Int. J. Energy Res. 40, 241–247 (2016).
Boston, C. R. Density of molten AlCl3 and NaCl-AlCl3 mixtures. J. Chem. Eng. Data 11, 262–263 (1996).
Matiašovský, K., Fellner, P. & Chrenková-Paučírová, M. Density and electrical conductivity of molten NaCl-KCl-AlCl3 mixtures. Electrochim. Acta 25, 195–200 (1980).
We acknowledge financial support from the MIT Energy Initiative, the MIT Deshpande Center for Technological Innovation and ENN Group. Q.P., J.M. and X.H. thank the Peking University startup funding, the National Natural Science Foundation of China (grant no. 22075002) and National Postdoctoral Programme for Innovative Talents (grant no. BX2021002). L.M. acknowledges financial support from The NationalKey Research and Development Program of China (grant no.2020YFA0715000). B.N. and S.G. acknowledge funding from the Office of the Executive Vice President for Research and Innovation at University of Louisville. This research used resources of the National Energy Research Scientific Computing Center, a DOE Office of Science User Facility supported by the Office of Science of the US Department of Energy under Contract No. DE-AC02-05CH11231. This research used resources of the Advanced Photon Source, an Office of Science User Facility operated for the US Department of Energy (DOE) Office of Science by Argonne National Laboratory and was supported by the US DOE under Contract No. DE-AC02-06CH1135. We are grateful for discussions of the cost analysis with L. Ortiz.
D.R.S. is a co-founder of Avanti Battery Co., a company established to commercialize the aluminium–sulfur battery. D.R.S.’s role with the company is advisory; he is formally the Chief Scientific Advisor.
Peer review information
Nature thanks George Chen, Robert Dominko and Feng Li for their contribution to the peer review of this work. Peer reviewer reports are available.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Extended data figures and tables
Extended Data Fig. 1 The fundamental electrochemical behavior of the chloroaluminate melt electrolytes.
a, b, The discharge/charge voltage-time traces and the rate capability of Al-S batteries in the widely reported EMIC-AlCl3 electrolyte at 25 °C (a) and 60 °C (b). The batteries use graphene/sulfur as the positive electrode. Ionic-liquid cells show low discharge voltage (0.5V) and very large polarization even at the low rates of D/20, C/20 at 25 °C. c, d, The Al plating/stripping coulombic efficiency measurements using an Al|Ta symmetric cell configuration in the molten NaCl-KCl-AlCl3 electrolyte at a current density of (c) 20 mA cm−2 and (d) 50 mA cm−2 at 180 °C (the insets are the expanded view). e-h, the linear-sweep voltammetry (LSV) of Al|Al symmetric cells at different temperatures in EMIC-AlCl3 (e) and NaCl-KCl-AlCl3 with varied molar ratios (f-h); i, The logarithmic plot of the temperature-dependent exchange current of Al|Al symmetric cells in NaCl-KCl-AlCl3 with varied molar ratios. It is clear that as the electrolyte becomes more basic (higher percentage of alkali chloride), the exchange current I0 decreases. In another word, even with the same type of cations, the exchange current varies with composition, i.e., with Lewis acidity. Therefore, it is not the cationic component, but rather the anionic cluster component (which is affected by the choice of cations and the instant concentration = Lewis acidity) that fixes the higher I0 observed in the NaCl-KCl-AlCl3 system. j, The logarithmic plot of ionic conductivity as a function of temperature, where the activation energy Ea is derived assuming Arrhenius behavior. k, l, The LSV curves of Al|Se cells at different temperatures in NaCl-KCl-AlCl3 (f) and EMIC-AlCl3 (g) electrolytes. In all LSV plots, the voltage is expressed as the potential vs. the open circuit voltage of the cell. For exchange current calculation, regression fitting of the linear part of the LSV curves was performed and the intercept of the fitted line with the V = 0 voltage is assigned to be the exchange current.
a, b, Typical snapshots from the AIMD trajectories for the two electrolytes EMIC-AlCl3 (a) and NaCl-AlCl3(b); Carbon (grey), hydrogen (white), and nitrogen (blue), and sodium (green) are displayed here to highlight the steric effects of the giant EMI+ cations in the EMIC-AlCl3 electrolyte. The Cl–, AlCl4–, Al2Cl6, Al2Cl7–, and Al3Cl10– clusters are shown in purple, dark blue, gold, cyan and red respectively. c, The diffusivity of each ion calculated based on the mean square displacement derived from the AIMD simulations over a long duration of 40 ps. d, The schematic illustration of our house-designed high-temperature Raman chamber. e,f, The 27Al NMR spectra of the two electrolytes (e) and the deconvolution of the spectrum for the MCl-AlCl3 electrolyte (f). g-j, The configuration of the clusters, and the scheme of proposed first step dissociation reaction to form Al2Cl6 for Al3+ de-solvation (equations 1–4), assuming breaking of the relatively longer Al-Cl bonds, for AlCl4– (g), Al2Cl7– (h), Al3Cl10– (i), Al4Cl13– (j). Note that two AlCl3 moieties (instead of one Al2Cl6) are shown here as the product for explicit illustration of the bond breaking scheme. QC calculations do support that one Al2Cl6 is more stable than two isolated AlCl3 molecules. k, A schematic illustration of how the counter ions (Na+ vs EMI+) can impact the coulombic pulling of the Clδ– away from the Al2Cl7– cluster which in turn affects the ease of Al-Cl bond breaking and ultimately the Al3+ de-solvation kinetics. Apparently, the smaller (thus more localized charge) and closer Na+ (as seen in the coordination plot in Fig. 1e) shows higher coulombic attraction for the Clδ− within the cluster, based on Coulomb’s law. l, The configuration of the Al2Cl6 cluster and the scheme of proposed dissociation of Al2Cl6 and eventual Al3+ de-solvation, along with the overall equations for all clusters. In the configuration panels, the Al and Cl atoms are shown as brown and purple spheres, respectively. The overall reactions for the desolvation of one Al3+ out of the clusters are shown in i. To add further experimental proof of the presence of AlnCl3n+1– species (n > 2), we performed the high-temperature Raman spectroscopy on the NaCl-KCl-AlCl3 electrolyte using a house-designed air-tight sapphire-window chamber with controlled heating capability. The measurement was performed at 180 °C and further experimental details are described in the Methods sections. We do note that owing to the intense fluorescence effect from the highly concentrated Cl ions, there is a high Raman background that compromises the signal/noise ratio; but we do clearly perceive the Raman bands of interest. We observe that in addition to the Raman bands corresponding to AlCl4− (centered at 351, 184, 122 cm−1) and Al2Cl7− (centered at 312, 160 and 110 cm−1), the electrolyte shows an additional peak at 176 cm−1, which can be ascribed to higher-order polyatomic clusters, i.e. AlnCl3n+1– (n > 2)47,48. This clearly proves the existence of AlnCl3n+1– (n≥3). Due to the temperature limit related to the detector of our NMR instrument, the measurements was performed at 90 °C, the highest temperature permitted. To specifically meet the temperature requirement, a high-entropy alkali chloroaluminate melt MCl-AlCl3 (M= Li+Na+K, M:Al = 1.31; m.p. = ~75 °C). The electrolytes were not mixed with any deuterated solvent to avoid any magnetic interference and/or chemical reactions. First, we observe that the 27Al NMR spectrum of the alkali MCl-AlCl3 shows a much broader peak, spanning from 135 ppm to 70 ppm (over 65 ppm), centered at a nominal chemical shift of 101.58 ppm, in comparison to that of the EMIC-AlCl3, centered at 103.54 ppm (nominal peak width: 1300 Hz vs. 313 Hz). For the EMIC-AlCl3, the two narrow peaks at 103.54 and 95.5 ppm can be respectively ascribed to AlCl4− and Al2Cl7−.We stress that the much grater peak broadness cannot be ascribed to any difference in the viscosity of the liquids as they are reported to be rather close: around 7.69 mPa s for a NaCl-AlCl3 (2:3) electrolyte at 90 °C59, and 5.28 mPa s for an EMIC-AlCl3 (2:3) electrolyte at 80 °C60, nor to any difference in the Al ion concentration, as the MCl:AlCl3 ratio is the same (2:3). Therefore, the much broader peak for MCl-AlCl3 electrolyte can only be accounted for by the presence of multiple Al-cluster species which are in fast chemical exchange. We attempted to deconvolve the spectrum of the alkali MCl:AlCl3 with a minimal number of peaks that show reasonable peak width (800~1250 Hz) using the linear fitting approach. The fitting clearly shows three peaks – any combination of two peak yields very large fitting residual. We note that in the literature, although there is universal agreement on the chemical shift of AlCl4– (around 103 ppm), the assignment for Al2Cl7– is still unsettled61,62,63,64,65,66, and there is no report on assignment for AlnCl3n+1– (n > 2), to the best of our knowledge. Also, these reported assignments are exclusively based on room-temperature organic chloroaluminate ionic liquids, and not on high-temperature inorganic alkali chloroaluminate molten salts. Therefore, herein we take great caution when assigning the peaks. The two peaks at 103.6 ppm are ascribed to AlCl4–, while the higher field peak at 100.3 ppm is assigned to Al2Cl7–. This assignment stands on the basis that the Al nuclei in Al2Cl7– (i.e. a longer-chain AlnCl3n+1–) are more electron-shielded, intuitively due to the replacement of one electron-attracting Cl– ligand with a less polarizable AlCl4– (see the molecular structure in Fig. 1f). Following this chemical principle, we tentatively ascribe the peak at the highest field of 94.4 ppm to AlnCl3n+1– (n > 2). Note that the difference of the chemical shifts of Al2Cl7– in the MCl-AlCl3 compared to that in the EMIC-AlCl3 is probably due to the vastly different coulombic interaction of the counter cations (M+ vs EMI+).
Extended Data Fig. 3 Morphological and elemental characterization of the Al plating in the chloroaluminate electrolytes.
a-d, The SEM images of Al plated on a Ta substrate at 180 °C in the NaCl-KCl-AlCl3 electrolytes (a,c) and in the EMIC-AlCl3 electrolyte (b,d) at a current density of 10 mA cm−2 (a,b) and 50 mA cm−2 (c,d) for an areal capacity of 5 mAh cm−2. e, f, The SEM images and the corresponding EDX mapping (Al, Cl), EDX spectra, and the quantified elemental compositions of plated Al (on a Ta substrate) performed in the NaCl-KCl-AlCl3 electrolyte at a current density of 10 mA cm−2 (e) and 50 mA cm−2 (f) for an areal capacity of 5 mAh cm−2. The EDX spectra and mapping show that to the resolution and detection limit of our EDX detector there is no Cl in the Al, which implies that the electrodeposit is pure Al metal containing no entrained electrolyte.
Extended Data Fig. 4 Cell design and the physical and electrochemical characterizations of Te and Se electrodes.
a, Schematic drawing of our in-house-made cell. b-e, Scanning electron microscopy image and energy-dispersive spectroscopy (EDS) (c), thermogravimetric analysis (d), X-ray diffraction pattern (e), transmission electron microscopy and EDS mapping (Se, N, C) (b) of the Se composite prepared by pyrolysis of polyacrylonitrile/selenium. The weight fraction of selenium in this composite is confirmed to be ~62.5% from both EDS and TGA analysis. The selenium composite has a uniform distribution of Se, N and C, where the N is from the residue of pyrolyzed polyacrylonitrile. f, The first-cycle voltage-time trace and rate performance (varying charging rate and constant discharge at D/10) of the Se-free carbon material, prepared in the same way except without selenium. g, h, Voltage-time traces (g) capacity retention (h) of Se in EMIC-AlCl3 at different temperatures and cycling rates, characterized by low capacity and fast fading even at very low cycling rates.
Extended Data Fig. 5 Electrochemical behavior of Al-Se cells fitted with molten chloroaluminate electrolytes.
(a) The GITT discharge voltage-time traces of Se electrodes (milled bulk selenium) in the two representative electrolytes, as a function of (a) measuring time and (b) of the specific capacity; measurement was performed with 20 min discharge at a C/5 rate followed by 1 h at rest; the equilibrium voltage is shown by circles in b. c, Capacity retention (discharge capacity) of Al-Se cells at rates of D/5 and C/2. d, Cell voltage (discharge and charge) as a function of cycle number for Al-Se cell running NaCl-AlCl3 electrolyte. e, Capacity retention (discharge capacity) of an Al-Se cell using ball-milled bulk Se cathode with molten NaCl-AlCl3. f, g, Voltage-time traces (f) differential capacity plots (dQ/dV) (g) of the Al-Se cells with molten NaCl-AlCl3 at different charging rates and constant discharge at D/10. h, Voltage-time traces of an Al-Se cell in EMIC-AlCl3 at different charge rates and constant discharge at D/10. The capacity and voltage time traces are determined by both the thermodynamic and kinetic factors. The GITT measurement allows one to answer what leads to the lower discharge capacity of the EMIC-AlCl3 cell. To avoid any effect of the pyrolyzed polyacrylonitrile on the potential profile, we used a bulk selenium/carbon composite without special treatment. Extended Data Fig. 5a shows that the EMIC-AlCl3 cell exhibits about the same capacity as the NaCl-AlCl3 cell; as the GITT measurement allows a close to equilibrium reaction, it is clear that the lower discharge capacity measured in galvanostatic mode (as in Fig. 2a) is due to the kinetics. Indeed, this is further supported by the fact that the voltage difference between the end of discharge pulse and the end of equilibrium period, that is, the overpotential is larger in the EMIC-AlCl3 cell than that in the NaCl-AlCl3 cell (Extended Data Fig. 5a, b). This is particularly true at high depth of discharge. Further, the quasi-equilibrium voltage time traces of the two cells are slightly different, which indicates slightly different thermodynamic energetics of the reactions. This can be explained by the different solvation energies and/or solubility for the Al2(Sen)3 intermediates by the different ions in the two electrolytes as observed for other sulfur electrochemical systems using electrolytes with varied electron donicity67. A stronger bonding of Al2(Sen)3 to the solvent (hence more stabilized) leads to higher change of Gibbs free energy of the discharge reaction, and thus a higher voltage. Also, as governed by the Nernst equation, the concentration of the Al2(Sen)3 also impacts the potential of the redox reaction. We do note that as the two cells both show a single-plateau discharge profile, any large difference in the selenium speciation mechanism or dynamics can be excluded.
Extended Data Fig. 6 Analysis of the basic components that comprise the Se K-edge X-ray absorption spectra during operando discharge at 170 °C.
a, b, The XRD pattern (a) and SEM image (b) of the selenium composite used for XAFS measurement; the absence of diffraction peaks indicates the non-crystalline nature of Se. c, The magnified XANES spectra in a representative region that shows the isosbestic points shared by the first few but all not all spectra.d, e, The plots of logarithmic variance (d) and eigenvalue (e) of each component for the principal component analysis (PCA), showing that at least 3 components are required to capture > 99.99% of the XAFS data set. f, The plots of first five PCA-derived components (Components 3–5 magnified in the inset); while Components 1, 2, 3 show meaningful curvature, Component 4 and beyond start to lose significance. g, The k-space EXAFS k2χ(k) oscillations of three identified basic components using MCR-ALS method, with the input data being either energy (E) space or frequency (k) space data sets. Notably, the three identified components are the same, irrespective of the type of input data set, validating the utilized analysis procedure. h-j, Fitting results of the EXAFS data (top: magnitude and real-part of Fourier transforms of k3χ(k); bottom: the corresponding k3χ(k) oscillations) for the standard selenium Se0 measured at 170 °C (h) and crystalline Na2Se2 measured at 25 °C (i) and the summary of the fitting parameters (j). Na2Se2 was synthesized by reaction of Na-naphthalene with selenium (see Methods). The window in the FT panels is the fit range used and the window in the EXAFS panels is the k-space data range used for the FT. The paths used for fitting Na2Se2 are from the Se-Se and Se-Na coordinations calculated from the crystalline Na2Se2. The fitting of Se0 allows us to determine the amplitude reduction factor (S02: 0.973) which is a beam-related constant and is used for fitting all spectra collected in this study. The methodology of fitting the Na2Se2 standard validates taking multiple paths from crystalline data to fit the first (and second) Se shell of unknown spectra as will be shown below.
a, Component A is fit very well using Se-Se correlations, as in crystalline Se0 (P 31 2 1). b-e, The fit for component C using different models and summary of the fitting parameters (f). The top of each panel is the magnitude and real-part of Fourier transforms (FT) of k3χ(k), and the bottom is the corresponding k3χ(k) oscillation. The window in the FT panels is the fit range used and the window in the EXAFS panels is the k-space data range used for the FT. The fitting R-range is from 1.3 to 2.5 Å, except that when the longer Se-Na path is included, the R-range is increased from 1.3 to 3 Å. All fits were carried out in R-space. Evaluation of the various fits shows that the Component C is best modelled by using selenium-aluminium correlation exclusively (fit b). g-i, The attempted EXAFS fitting for component B using just one path: Se-Se (a) or Se-Al (b), and summary of the fitting parameters (c). Evaluation of all attempted fits shows that Component B is best modelled by using both Se-Al and Se-Se correlations. Importantly, we note that the R of Se-Al in Component B is different from that in Component C (2.494 vs. 2.397 Å), proving that Component B is distinguished from Component C and is thus indeed a necessary component.
Extended Data Fig. 8 Further analysis of the reaction mechanism on discharge of Al-Se cells fitted with molten NaCl-AlCl3 electrolyte.
a, The χred2 of linear combination fittings using three components (χred2best, bottom) and using only two of the three components (ratio over χred2best, top). The significantly higher χred2 using only two components proves that three components are necessary to capture the information of the entire data set; the almost constant χred2best across all scans indicates further that 3 components faithfully reproduce the entire data set, in complete agreement with the PCA analysis (Extended Data Fig. 6d, e). More importantly, the low χred2 of using Components A and B for scans 1–6 (χred2/ χred2best ~1) indicates that only A and B are present in these 6 scans, and similarly, the high χred2 of using B and C for scans 10–15 (χred2/ χred2best = 4~11) indicates that A is necessary and present in these scans. b, The operando X-ray diffraction patterns of the crystalline Se cathode during discharge at D/10 and 180 °C, featuring the Se (101) and Se (010) peaks. The Se peaks do not disappear until the end of discharge. c, d, Nyquist plots of the in-situ measured impedance data on an Al|Se cell, as a function of the state of discharge (SOD) from bottom to top (c); the evolution of charge transfer resistance (Rct) and electrolyte resistance (Rs) of the cell, as fitted from the Nyquist plots (d). For consistency amongst all impedance data, circle fitting of the Nyquist plots on the high-frequency semicircle was performed (instead of finding one equivalent circuit for all), and the first intercept (of the fitted circle with the real-axis) is regarded as Rs and the distance between the two intercepts is regarded as Rct. The error bars are the standard deviations based on three measurements at the corresponding SOD. The Rs remains constant at around 1.5 ohm cm2 for all SOD, indicating there is minimal change of the electrolyte composition and dissolution of formed species. The Rct experiences initial decrease (SOD: 0–30%, due to formation of partially soluble Al2(Sen)3 species), slight increase (SOD: 30–55%, probably due to initial formation of Al2Se3), stabilization (SOD: 55–85%), and eventual increase (SOD: 85–100%, eventual formation of Al2Se3). This is consistent with the trend in the proposed reaction mechanism in Fig. 3e. e, The picture of Al2(Se6)3 (targeted stoichiometry) and selenium in the molten NaCl-AlCl3 electrolyte placed on a hot plate (around 180 °C); partial solubility is confirmed by the darkened color. The nominal Al2(Se6)3 was prepared by mixing the targeted amount of Se and Al2Se3 in the melt; thus, it may not reflect the exact ratio and serves only as demonstration of partial solubility of such compounds. Note that elemental selenium appears to have some solubility as well. f-h, the ex-situ XAFS studies on the fast-charged selenium electrodes: the XAFS spectra of fully discharged and charged electrodes collected operando and ex-situ for comparison (f); as the fast-charging study has to be performed ex-situ due to the limited time resolution of the XAS scan (~10 min), the very similar XAFS features using two measurement approaches indicates that the chemical states of selenium species remain unchanged if we stop and cool down the warm cell at a certain SOC (the electrolyte freezes). The ex-situ Se K-edge XAFS spectra of the selenium electrodes charged at 20C (g) and 50C (h) respectively; the spectra were collected using transmission mode on the retrieved fast charged electrode, and fitted by linear combination fittings using the three principal components. By quantifying the components using linear combination fitting, we observed that the cathode recharged at 20C contains 10.5% of the Al2(Sen)3 and 89.5% of Se0, and the one recharged at 50C contains 17.3% of the Al2(Sen)3 and 82.7% of Se0. No component C was observed in both electrodes.
Extended Data Fig. 9 The electrochemical behaviour of Al-S cells fitted with molten chloroaluminate electrolytes.
a, The TGA plot and the representative TEM image of the S/graphene composite, which contains 50 wt% of sulfur. b, c The first-cycle voltage-time trace (b) and capacity retention (c) of the graphene that is used to prepare the sulfur composite. Capacity retention was measured at various charging rates and at constant discharge rate of D/2. d, the differential capacity plot (dQ/dV) of the Al-S cell in molten NaCl-KCl-AlCl3 at different rates at 110 °C. e, Voltage-time traces of an Al-S cell in EMIC-AlCl3 at different rates at 110 °C. f, g, Capacity retention (discharge capacity, f) and the corresponding voltage-time traces (g) of an Al-S cell using KB/S composite in NaCl-KCl-AlCl3 at 110 °C, at constant discharge rate of D/2. h, i, The discharge voltage-time traces of Al-S cells in NaCl-KCl-AlCl3 (h) and EMIC-AlCl3 (i), at various discharge rates (D/5 to 20D) and at constant charge rate of C/2. j, k, Two representative surface SEM images of the Al negative electrode in Al-S cells after the charging rate-capability measurement. The electrode was thouroughly washed by acetonitrile in the glovebox before imaging.
Extended Data Fig. 10 The practical attributes of Al-S batteries fitted with molten alkali chloroaluminate electrolyte.
a, The discharge capacity retention (specific capacity and areal capacity) of an Al-S cell using sulfur elecctrodes with 12.0 mg cm−2 areal loading in molten NaCl-KCl-ACl3 at a D/5 and C/5 rate. b-e, The representative SEM images of the prepared 3D interconnected reduced graphene oxide/sulfur composite (rGO/S) (b-c), and the voltage-time traces (d) and charging rate capability at constant discharge rate of D/2 (e) using rGO-S electrode with an areal sulfur loading of 7.1 mg cm−2. f, A schematic illustration of the proposed tri-layer structure of a large-scale Al-S battery that is modified from the architecture of the liquid metal battery. g-j, the cost breakdown of the representative cell chemistry (Al-S battery, Li-S, graphite-NMC622 and graphite-LiFePO4); the cost of electrodes includes the associated carbon and binder where applicable; the percentage is rounded up to integral digit (the details of calculation is shown in Extended Data Table 1). k, The fast-charge performance of an Al-S battery fitted with a low-grade (food packaging foil) Al-foil negative electrode, foretelling major cost reduction. Our current use of Mo current collector is for proof-of-concept demonstration of the Al-S battery using alkali chloroaluminate melt, and Mo foil is not an ideal positive current collector for a practical battery. Modifications can be Mo coated Al foil (Mo coating by vapor deposition: 300~550 nm thick)68, TiN coated Al foil (TiN coating by pulsed DC plasma enhanced CVD)69,70, or non-graphitic carbon foil. In our estimate calculation, the model of Mo coated Al foil (Mo: 0.5 μm thick) was used. Therefore, on the package level, our Al-S battery is expected to cost about 1/4 and 1/5 that of the graphite-LiFePO4 and graphite-NMC622 battery (20.8 vs 80.3 and 94.7 USD$ kWh−1; unit price of each material/component is based on the latest market and as accurate as possible). As the price is the latest market price or the projected price when market price is unavailable, the cost may fluctuate and contain errors, and we do expect further modifications when it comes to the manufacturing. Nevertheless, we believe that this chemistry offers a great economical advantage.
About this article
Cite this article
Pang, Q., Meng, J., Gupta, S. et al. Fast-charging aluminium–chalcogen batteries resistant to dendritic shorting. Nature 608, 704–711 (2022). https://doi.org/10.1038/s41586-022-04983-9
This article is cited by
A solution-to-solid conversion chemistry enables ultrafast-charging and long-lived molten salt aluminium batteries
Nature Communications (2023)
Nature Sustainability (2023)
Nature Sustainability (2023)
Nano-Micro Letters (2023)