Rechargeable lithium-ion batteries with high energy density that can be safely charged and discharged at high rates are desirable for electrified transportation and other applications1,2,3. However, the sub-optimal intercalation potentials of current anodes result in a trade-off between energy density, power and safety. Here we report that disordered rock salt4,5 Li3+xV2O5 can be used as a fast-charging anode that can reversibly cycle two lithium ions at an average voltage of about 0.6 volts versus a Li/Li+ reference electrode. The increased potential compared to graphite6,7 reduces the likelihood of lithium metal plating if proper charging controls are used, alleviating a major safety concern (short-circuiting related to Li dendrite growth). In addition, a lithium-ion battery with a disordered rock salt Li3V2O5 anode yields a cell voltage much higher than does a battery using a commercial fast-charging lithium titanate anode or other intercalation anode candidates (Li3VO4 and LiV0.5Ti0.5S2)8,9. Further, disordered rock salt Li3V2O5 can perform over 1,000 charge–discharge cycles with negligible capacity decay and exhibits exceptional rate capability, delivering over 40 per cent of its capacity in 20 seconds. We attribute the low voltage and high rate capability of disordered rock salt Li3V2O5 to a redistributive lithium intercalation mechanism with low energy barriers revealed via ab initio calculations. This low-potential, high-rate intercalation reaction can be used to identify other metal oxide anodes for fast-charging, long-life lithium-ion batteries.
Subscribe to Journal
Get full journal access for 1 year
only $3.90 per issue
All prices are NET prices.
VAT will be added later in the checkout.
Rent or Buy article
Get time limited or full article access on ReadCube.
All prices are NET prices.
The data that support the findings of this study are available from the corresponding author on reasonable request.
Armand, M. & Tarascon, J. M. Building better batteries. Nature 451, 652–657 (2008).
Larcher, D. & Tarascon, J. M. Towards greener and more sustainable batteries for electrical energy storage. Nat. Chem. 7, 19–29 (2015).
Grey, C. P. & Tarascon, J. M. Sustainability and in situ monitoring in battery development. Nat. Mater. 16, 45–56 (2017).
Lee, J. et al. Unlocking the potential of cation-disordered oxides for rechargeable lithium batteries. Science 343, 519–522 (2014).
Perez, A. J. et al. Approaching the limits of cationic and anionic electrochemical activity with the Li-rich layered rocksalt Li3IrO4. Nat. Energy 2, 954–962 (2017).
Dahn, J. R., Zheng, T., Liu, Y. & Xue, J. S. Mechanisms for lithium insertion in carbonaceous materials. Science 270, 590 (1995).
Dahn, J. R. Phase-diagram of LixC6. Phys. Rev. B 44, 9170–9177 (1991).
Li, H., Liu, X., Zhai, T., Li, D. & Zhou, H. Li3VO4: a promising insertion anode material for lithium-ion batteries. Adv. Energy Mater. 3, 428–432 (2013).
Clark, S. J., Wang, D., Armstrong, A. R. & Bruce, P. G. Li(V0.5Ti0.5)S2 as a 1 V lithium intercalation electrode. Nat. Commun. 7, 10898 (2016).
Urban, A., Matts, I., Abdellahi, A. & Ceder, G. Computational design and preparation of cation-disordered oxides for high-energy-density Li-ion batteries. Adv. Energy Mater. 6, 1600488 (2016).
Pralong, V. et al. Electrochemical synthesis of a lithium-rich rock-salt-type oxide Li5W2O7 with reversible deintercalation properties. Inorg. Chem. 53, 522–527 (2014).
Mikhailova, D. et al. Lithium insertion into Li2MoO4: reversible formation of (Li3Mo)O4 with a disordered rock-salt structure. Chem. Mater. 27, 4485–4492 (2015).
Delmas, C., Cognacauradou, H., Cocciantelli, J. M., Menetrier, M. & Doumerc, J. P. The LixV2O5 system—an overview of the structure modifications induced by the lithium intercalation. Solid State Ion. 69, 257–264 (1994).
Delmas, C. & Cognacauradou, H. Formation of the omega-type phase by lithium intercalation in (Mo, V) oxides deriving from V2O5. J. Power Sources 54, 406–410 (1995).
Yin, L. et al. Extending the limits of powder diffraction analysis: diffraction parameter space, occupancy defects, and atomic form factors. Rev. Sci. Instrum. 89, 093002 (2018).
Xu, K. Electrolytes and interphases in Li-ion batteries and beyond. Chem. Rev. 114, 11503–11618 (2014).
Aldon, L. et al. Chemical and electrochemical Li-insertion into the Li4Ti5O12 spinel. Chem. Mater. 16, 5721–5725 (2004).
Liu, H. et al. Elucidating the limit of Li insertion into the spinel Li4Ti5O12. ACS Mater. Lett. 1, 96–102 (2019).
Thackeray, M. M., Johnson, P. J., de Picciotto, L. A., Bruce, P. G. & Goodenough, J. B. Electrochemical extraction of lithium from LiMn2O4. Mater. Res. Bull. 19, 179–187 (1984).
Thackeray, M. M., Mansuetto, M. F. & Bates, J. B. Structural stability of LiMn2O4 electrodes for lithium batteries. J. Power Sources 68, 153–158 (1997).
Ramana, C. V., Massot, M. & Julien, C. M. XPS and Raman spectroscopic characterization of LiMn2O4 spinels. Surf. Interface Anal. 37, 412–416 (2005).
Chaurand, P. et al. New methodological approach for the vanadium K-edge X-ray absorption near-edge structure interpretation: application to the speciation of vanadium in oxide phases from steel slag. J. Phys. Chem. B 111, 5101–5110 (2007).
Mansour, A. N., Smith, P. H., Baker, W. M., Balasubramanian, M. & McBreen, J. A comparative in situ X-ray absorption spectroscopy study of nanophase V2O5 aerogel and ambigel cathodes. J. Electrochem. Soc. 150, A403–A413 (2003).
Giorgetti, M. et al. In situ X-ray absorption spectroscopy characterization of V2O5 xerogel cathodes upon lithium intercalation. J. Electrochem. Soc. 146, 2387–2392 (1999).
Rehr, J. J., Kas, J. J., Vila, F. D., Prange, M. P. & Jorissen, K. Parameter-free calculations of X-ray spectra with FEFF9. Phys. Chem. Chem. Phys. 12, 5503–5513 (2010).
Radin, M. D. et al. Narrowing the gap between theoretical and practical capacities in Li-ion layered oxide cathode materials. Adv. Energy Mater. 7, 1602888 (2017).
Zhao, W. et al. Toward a durable solid electrolyte film on the electrodes for Li-ion batteries with high performance. Nano Energy 63, 103815 (2019).
Cheng, X. et al. Realizing superior cycling stability of Ni-rich layered cathode by combination of grain boundary engineering and surface coating. Nano Energy 62, 30–37 (2019).
Mao, Y. et al. High-voltage charging-induced strain, heterogeneity, and micro-cracks in secondary particles of a nickel-rich layered cathode material. Adv. Funct. Mater. 29, 1900247 (2019).
Thinius, S., Islam, M. M., Heitjans, P. & Bredow, T. Theoretical study of Li migration in lithium–graphite intercalation compounds with dispersion-corrected DFT methods. J. Phys. Chem. C 118, 2273–2280 (2014).
An, K. et al. First in situ lattice strains measurements under load at VULCAN. Metall. Mater. Trans. A 42, 95–99 (2011).
An, K., Wang, X. L. & Stoica, A. D. Vulcan Data Reduction And Interactive Visualization Software ORNL Report 621 (ORNL, 2012).
Larson, A. C. & Dreele, R. B. V. General Structure Analysis System (GSAS) Los Alamos National Laboratory Report (LAUR) 86-748 (LANL, 2004).
Toby, B. H. EXPGUI, a graphical user interface for GSAS. J. Appl. Crystallogr. 34, 210–213 (2001).
Borkiewicz, O. J. et al. The AMPIX electrochemical cell: a versatile apparatus for in situ X-ray scattering and spectroscopic measurements. J. Appl. Crystallogr. 45, 1261–1269 (2012).
Verde, M. G. et al. Effect of morphology and manganese valence on the voltage fade and capacity retention of Li[Li2/12Ni3/12Mn7/12]O2. ACS Appl. Mater. Interfaces 6, 18868–18877 (2014).
Kresse, G. & Furthmuller, J. Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set. Phys. Rev. B 54, 11169–11186 (1996).
Blöchl, P. E. Projector augmented-wave method. Phys. Rev. B 50, 17953–17979 (1994).
Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized gradient approximation made simple. Phys. Rev. Lett. 77, 3865–3868 (1996).
Padhi, A. K., Nanjundaswamy, K. S. & Goodenough, J. B. Phospho-olivines as positive-electrode materials for rechargeable lithium batteries. J. Electrochem. Soc. 144, 1188–1194 (1997).
Liechtenstein, A. I., Anisimov, V. I. & Zaanen, J. Density-functional theory and strong-interactions—orbital ordering in Mott–Hubbard insulators. Phys. Rev. B 52, R5467–R5470 (1995).
Ling, C., Zhang, R. G. & Mizuno, F. Phase stability and its impact on the electrochemical performance of VOPO4 and LiVOPO4. J. Mater. Chem. A 2, 12330–12339 (2014).
Jain, A. et al. Commentary: The Materials Project: a materials genome approach to accelerating materials innovation. Appl. Mater. 1, 011002 (2013).
Hart, G. L. W. & Forcade, R. W. Algorithm for generating derivative structures. Phys. Rev. B 77, 224115 (2008).
Ong, S. P. et al. Python Materials Genomics (pymatgen): a robust, open-source Python library for materials analysis. Comput. Mater. Sci. 68, 314–319 (2013).
Waroquiers, D. et al. Statistical analysis of coordination environments in oxides. Chem. Mater. 29, 8346–8360 (2017).
Heyd, J., Scuseria, G. E. & Ernzerhof, M. Hybrid functionals based on a screened Coulomb potential. J. Chem. Phys. 118, 8207–8215 (2003).
Aydinol, M. K., Kohan, A. F., Ceder, G., Cho, K. & Joannopoulos, J. Ab initio study of lithium intercalation in metal oxides and metal dichalcogenides. Phys. Rev. B 56, 1354–1365 (1997).
Rozier, P. et al. Cation-disordered Li3VO4: reversible Li insertion/deinsertion mechanism for quasi Li-rich layered Li1+x[V1/2Li1/2]O2 (x = 0–1). Chem. Mater. 30, 4926–4934 (2018).
Zheng, C. et al. Automated generation and ensemble-learned matching of X-ray absorption spectra. npj Comput. Mater. 4, 12 (2018).
Part of the work used the UCSD-MTI Battery Fabrication Facility and the UCSD-Arbin Battery Testing Facility. Z.Z., Yiming Chen and S.P.O. acknowledge funding from the US Department of Energy (DOE), Office of Science, Basic Energy Sciences under award number DE-SC0012118 for the DFT thermodynamics and kinetic studies, the Data Infrastructure Building Blocks (DIBBS) Local Spectroscopy Data Infrastructure (LSDI) project funded by National Science Foundation (NSF), under award number 1640899 for the FEFF X-ray absorption spectroscopy computations, and computing resources provided by the Triton Shared Computing Cluster (TSCC) at the University of California, San Diego, the National Energy Research Scientific Computing Center (NERSC), and the Extreme Science and Engineering Discovery Environment (XSEDE) under grant ACI-1548562. The X-ray characterization work at Lawrence Berkeley National Laboratory by X.H. and R.K. was supported by the Assistant Secretary for Energy Efficiency and Renewable Energy, Vehicle Technologies Office, under the Advanced Battery Materials Research (BMR) Program of the U.S. Department of Energy under contract no. DE-AC02-05CH11231. Neutron diffraction work used resources at the Spallation Neutron Source, a DOE Office of Science User Facility operated by the Oak Ridge National Laboratory. R.Z. was supported by H.L.X.’s startup funding. This research used resources of the Center for Functional Nanomaterials, which is a US DOE Office of Science Facility, at Brookhaven National Laboratory under contract number DE-SC0012704. The work at Argonne National Laboratory was supported by the US DOE, Office of Energy Efficiency and Renewable Energy, Vehicle Technologies Office. This research used resources of the Advanced Photon Source (9-BM and 17-BM), Argonne National Laboratory, a US DOE Office of Science User Facility operated for the US DOE Office of Science by the University of Chicago Argonne, LLC, under contract number DE-AC02-06CH11357. M.L. would like to acknowledge financial support from the National Sciences and Engineering Research Council (NSERC) of Canada. This research thanks A. Yakovenko, W. Xu and K. Wiaderek for their support of the in situ XRD experiments. H.L. thanks J. Huang for assistance with the electrochemical experiment. H.L. and Z.Z. thank I.-H. Chu for suggestions on the DFT calculations.
P.L. and H.L. report a Patent Cooperation Treaty (PCT) application filed on 16 April 2019, Anode Material For Rechargeable Li-ion Batteries, PCT Application Serial No. USl2019/27755.
Peer review information Nature thanks Yuping Wu and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.
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 phase transformation process of V2O5 upon electrochemical lithiation and joint Rietveld refinement of pristine DRS-Li3V2O5.
a, XRD of LixV2O5 (x = 0, 0.5, 1, 2, 3). The LixV2O5 was obtained by electrochemical lithiation using a Li||V2O5 cell. The cell was discharged under 0.1 A g−1 to 3.25 V, 2.5 V, 2.15 V and 1 V to generate the ε, δ, γ and ω phases. b, Voltage profile of the electrochemical lithiation of V2O5 down to 0.01 V. There are five major voltage plateaus at 3.35 V, 3.18 V, 2.35 V, 1.95 V and 0.52 V, respectively. c, Neutron diffraction pattern; d, XRD pattern. The X-ray wavelength is 0.1173 Å; the pattern is converted to the wavelength of the Cu source for comparison with the laboratory X-ray data.
a, Comparison of XRD pattern of in situ cells. Carbon paper (C paper) and Cu were used as the current collectors for the DRS-Li3V2O5 in two different AMPIX cells. The X-ray wavelength is 0.24117 Å. b, Voltage profile of the in situ cell with a carbon paper current collector. The DRS-Li3V2O5 electrodes, with active material mass loading of 5 mg cm−2, were cycled at a current density of 0.05 A g−1 between 0.1 V and 2.0 V. c, The corresponding contour plots of the DRS-Li3V2O5 (220) XRD patterns. d, Evolution of the lattice parameter a of DRS-Li3V2O5. e, The evolution of the unit cell volume of DRS-Li3V2O5.
Extended Data Fig. 3 The characterization of the SEI layer on DRS-Li3+xV2O5 and the morphologies of V2O5 and DRS-Li5V2O5.
a, TEM images of DRS-Li3+xV2O5 at 0.01 V on first discharge. The SEI layer is indicated by the red dashed lines. b, The statistical distribution of SEI thickness on the surface of DRS-Li3+xV2O5 at 0.01 V on first discharge. (counts, the times of measurements of SEI thickness; a.u., arbitrary units.) c, f, SEM images of V2O5 powders. d, g, SEM images of the V2O5 electrode. e, h, SEM images of DRS-Li5V2O5. The DRS-Li5V2O5 was prepared by lithiating the V2O5 electrode at 0.1 A g−1 until 0.01 V. The V2O5 powder shows irregular primary particle shapes with sizes ranging from 350 nm to 1 µm.
Extended Data Fig. 4 Comparison of site energies in DRS-Li3V2O5 and pseudo-binary Li3V2O5-Li5V2O5 phase diagram from DFT calculations.
a, The DFT relative site energies for the 0-TM (T1) and 1-TM (T2) tetrahedral sites in Li3V2O5. An additional Li ion was inserted in all symmetrically distinct sites in the lowest-energy configuration of the 2 × 2 × 2 supercell of Li19V13O32 (which is equivalent to Li3V2O5). The sites are ordered by increasing site energy, with the energy of the most stable site set as the zero reference. The seven lowest-energy sites (green circles) for Li insertion are 0-TM sites. The 1-TM sites (orange circles) have substantially higher energies (>387 meV; Ediff, energy difference) for Li insertion. b, Pseudo-binary Li3V2O5–Li5V2O5 compound phase diagram computed using the PBE+U functional (fu, formula unit). The two endmembers are Li3V2O5 and Li5V2O5. A finer compositional resolution of x = 1/8 increments in Li3+xV2O5 was used in the Li3V2O5–Li4V2O5 region to characterize the changes in Li distribution between the tetrahedral and octahedral sites in this region. In the Li4V2O5–Li5V2O5 region, a lower compositional resolution of x = 1/4 increments in Li3+xV2O5 was used because there are no changes in octahedral Li site occupancy in this region.
Extended Data Fig. 5 Comparison of experimental and FEFF calculated V K-edge XANES spectra reveal the charge compensation mechanism of DRS-Li3+xV2O5.
a, The experimental V K-edge XANES spectra at different states of (dis)charge. The dashed lines are V K-edge XANES spectra of V oxides with different V oxidation states. The oxidation states of V in V2O5, VO2 and V2O3, are +5, +4 and +3, respectively. b, Comparison between experimental and FEFF-calculated V K-edge XANES spectra for Li3V2O5 and Li5V2O5. The average oxidation states of V in Li3V2O5 and Li5V2O5 are +3.5 and +2.5, respectively. c, Experimental and FEFF-calculated V K-edge XANES spectra for V2O5, VO2 and V2O3, with V oxidation states +5, +4, +3, respectively. When V is reduced from the oxidation state of +5 to +3, the experimental V K-edge XANES spectra show a decrease in P/M. FEFF-computed V K-edge XANES spectra show the same trend. d, Experimental and FEFF-calculated V K-edge XANES spectra for Li3VO4 and LiVO2. Data are from Rozier et al.49. The average oxidation states of V in Li3VO4 and LiVO2 are +5 and +3, respectively. A similar decrease in P/M is seen with reduction in V oxidation state.
a, Voltage profiles of DRS-Li3+xV2O5 over 1,000 cycles at 1 A g−1. b, Comparison of the electrochemical performance of DRS-Li3V2O5 with different voltage windows under a current density of 0.1 A g−1. c, Voltage profiles of DRS-Li3V2O5 in voltage windows of 0.01–2 V, 0.1–2 V, 0.2–2 V and 0.3–2 V, respectively. d, The initial charge voltage curve of a Li||LiNi0.8Mn0.1Co0.1O2 half-cell under 0.1 A g−1 and the discharge voltage curve of a Li||DRS-Li3V2O5 half-cell under 0.1 A g−1 were used to simulate the charge voltage profile of a DRS-Li3V2O5||LiNi0.8Mn0.1Co0.1O2 full cell. e, Voltage profiles of a Li||LiNi0.8Mn0.1Co0.1O2 half-cell under charge–discharge current densities of 0.1 A g−1 and 0.5 A g−1, with voltage window 2.8–4.3 V. The LiNi0.8Mn0.1Co0.1O2 delivered specific capacities of 190 mA h g−1 and 160 mA h g−1 under 0.1 A g−1 and 0.5 A g−1, respectively. f, Voltage profiles of a DRS-Li3V2O5||LiNi0.8Mn0.1Co0.1O2 full cell under charge–discharge current densities of 0.1 A g−1 and 0.5 A g−1, with voltage window 1.5–3.9 V. The full cell delivered specific capacities of 189 mA h g−1 and 155 mA h g−1 under 0.1 A g−1 and 0.5 A g−1, respectively. g, Discharge voltage profiles of DRS-Li3V2O5 with 30 wt% of carbon under various charge–discharge current densities, with voltage window 0.01–2.0 V. h, Discharge voltage profiles of DRS-Li3V2O5 with 10 wt% of carbon under various charge–discharge current densities, with voltage window 0.01–2.0 V. i, XRD patterns of DRS-Li3V2O5 prepared by different methods. Comparison of the electrochemical (Echem) performance of the fresh DRS-Li3V2O5 and the aged DRS-Li3V2O5 prepared by chemical (Chem) synthesis. j, Cycling stability (Ch, charge; DCh, discharge) under a current density of 0.1 A g−1. k, Representative voltage profiles of chemically synthesized DRS-Li3V2O5.
a, SEM images of pristine LiNi0.8Mn0.1Co0.1O2, showing agglomerated spherical secondary particles with diameters ranging from 6 µm to 25 µm. b, c, SEM images of the LiNi0.8Mn0.1Co0.1O2 from the DRS-Li3V2O5||LiNi0.8Mn0.1Co0.1O2 full cell after 1,000 cycles. The cycled LiNi0.8Mn0.1Co0.1O2 exhibits different degrees of cracking on their secondary particles. d–f, XPS of the separator from the cycled DRS-Li3V2O5||LiNi0.8Mn0.1Co0.1O2 full cell: d, Ni 2p region; e, Mn 2p region; f, Co 2p region. The separator is from the DRS-Li3V2O5||LiNi0.8Mn0.1Co0.1O2 full cell after 1,000 cycles. The Ni 2p peak is clearly presented, while no signal appears in the Mn 2p and Co 2p regions. (CPS, counts per second.) g–j, XPS of the DRS-Li3V2O5 electrode from the same full cell before and after etching: g, Ni 2p region; h, O 1s and V 2p region; i, Mn 2p region; j, Co 2p region. Similar to the separator, there is only Ni on the anode side, further confirming the dissolution of Ni from the cathode. k–n, XPS of the LiNi0.8Mn0.1Co0.1O2 electrode from the same full cell: k, Ni 2p region; l, O 1s and V 2p region; m, Mn 2p region; n, Co 2p region. The lack of V signal on the cathode side suggests that the DRS-Li3V2O5 anode does not suffer from metal dissolution.
Extended Data Fig. 8 NEB barriers categorized by their mechanism at the start (x ≈0) and end (x ≈ 2) of the discharge in Li3+xV2O5.
Panels a–c refer to Li interstitial migration barriers in Li19V13O32 (equivalent to Li3V2O5) and panels d and e refer to Li vacancy migration in Li32V13O32 (equivalent to Li5V2O5). a, Concerted Li migration barriers in Li3V2O5 based on four representative configurations. Five paths from four orderings contribute to super-low NEB barriers ranging from 166 meV to 290 meV. The hopping type is opposing T1-O-T1, which refers to cooperative hops between two T1 (0-TM) tetrahedral sites through an octahedral site. The relative positions between the initial and final tetrahedral sites are opposing versus the central octahedral site. b, Concerted Li migration barriers in Li3V2O5 based on four representative configurations. Seven paths from four orderings contribute to relatively low NEB barriers ranging from 204 meV to 435 meV. The hopping type is corner-sharing T1-O-T1, which refers to cooperative hops between two T1 (0-TM) tetrahedral sites through an octahedral site. The relative positions between the initial and final tetrahedral sites are corner-shared with each other. c, Direct Li migration barriers in Li3V2O5 based on four representative configurations. Four paths from four orderings contribute to high NEB barriers ranging from 634 meV to 1,049 meV. The hopping type is edge-sharing T1-T1, which refers to the direct hops between two nearest edge-sharing T1 (0-TM) tetrahedral sites. d, Vacancy migration barriers in the lowest-energy configuration of Li5V2O5. Direct tetrahedron-to-tetrahedron (t-t) hops with super-low NEB barriers ranging from 181 meV to 310 meV. e, Hops by the t-o-t mechanism refer to the migration from one tetrahedron to the other through an empty face-shared octahedron. The barriers from 703 meV to 1,109 meV are much higher than for the direct t-t mechanism, which makes this mechanism unfavourable.
Extended Data Fig. 9 Four representative configurations of Li19V13O32 (equivalent to Li3V2O5) with low energies and the most stable ordered structures of Li3+xV2O5 (x = 0, 1, 2) after structure enumeration.
The configurations a to d are obtained from 2 × 2 × 2 supercells of the rock salt cubic conventional cell. Octahedral sites are fully occupied by Li/V atoms (green octahedra are LiO6; red octahedra are VO6). These representative configurations were used for NEB calculations at the start of discharge. e, LiO4 tetrahedra and LiO6 octahedra are shown as blue and green polyhedral, respectively, when red ones are for VO6 octahedra. From the structure of Li3V2O5 to Li4V2O5, the majority of octahedral Li transfer to tetrahedral Li, which is consistent with site occupancy results in Fig. 2c. When Li continues to be inserted, the number of LiO6 octahedra remain the same and LiO4 tetrahedra keep increasing.
About this article
Cite this article
Liu, H., Zhu, Z., Yan, Q. et al. A disordered rock salt anode for fast-charging lithium-ion batteries. Nature 585, 63–67 (2020). https://doi.org/10.1038/s41586-020-2637-6