Glassy Li metal anode for high-performance rechargeable Li batteries
Cryogenic (scanning) transmission electron microscopy resolves electrodeposited lithium (EDLi) from early nucleation through growth, revealing disordered (“glassy”) Li at short times / selected conditions and a disorder–order transition as deposition time and current density evolve. Reactive molecular dynamics (r-MD) simulations interpret atomic-scale kinetics and critical cluster sizes for ordering.
Summary¶
The paper combines cryo-EM/cryo-TEM on EDLi across 0.1–2.5 mA cm⁻² and 5–20 min with LAMMPS r-MD using a Li-trained Reaxff (authors’ “r-MD” label) to link nucleation to spontaneous crystallization past a ~5 nm scale. Initially many deposits look amorphous/glassy; longer t and higher j yield bcc-like order; r-MD predicts a critical size for ordering in line with HRTEM-inferred domains ~5 ± 3 nm.
Methods¶
1 — MD application (r-MD, LAMMPS). Li-parametrized Reaxff as implemented; MAPS (Scienomics v4.0.1) for setup. Nose-Hoover thermostat, NVT; ~0.0534 g cm⁻³ Li density; default 0.5 fs time step (0.1 and 0.25 fs in tests per SI). Staged protocol: ramp to 500 K (0.1 ns), quench to 300 K (0.2 ns), hold 300 K (5 ns default) for nucleation in 500–1500-atom ensembles. N/A — NPT (constant-volume NVT in protocol). N/A — external electric field; N/A — metadynamics. PBC in the small Li droplet/bulk cells as in SI. N/A — hydrostatic pressure control: NVT without NPT stated for the r-MD (confirm SI for any stress protocol).
2 — Experiment (cryo-TEM/EM). EDLi on current collectors; 0.1–2.5 mA cm⁻²; 5–20 min; FFT/HRTEM; cell and electrolyte details in the article/ SI. N/A — in situ MD of full electrolyte at ab initio cost.
3 — Force-field training. N/A — the article uses a published Li r-MD model; not refit in the paper text summarized here.
4 — Static QM. N/A — not the paper’s main contribution.
5 — Review or non-simulation. N/A — research article (experiment + MD).
Findings¶
Outcomes and mechanisms. Glassy / amorphous-appearing EDLi (weak bcc FFT spots) can occur at short t / moderate j; a DOPT toward bcc increases with t and j. Crystalline patches reach ~5 ± 3 nm in an early ordering stage; r-MD supports spontaneous ordering beyond a ~5 nm-scale cluster—aligned with the TEM narrative. By ~20 min, large >50 nm bcc domains and strong diffraction appear. Reversibility in cycling tests is better for more “glassy” regimes in the data shown.
Comparisons and sensitivity. TEM/ r-MD juxtaposed; j and t are swept as levers; temperature ramps in r-MD stages.
Authored limitations. Ex situ cryo can miss some side reactions; r-MD Reaxff is not a universal potentiostatic model.
Corpus honesty. Quantitative TEM and r-MD numbers live in the VOR pdf_path.
Limitations¶
r-MD details (potential, constraints) must be read from the paper; cryo imaging may omit electrolyte side reactions not captured in ex situ snapshots.
Relevance to group¶
Uses reactive MD (as labeled by the authors) alongside high-resolution electron microscopy for Li metal morphogenesis—adjacent to ReaxFF battery-interface studies in the broader corpus.
Citations and evidence anchors¶
- DOI:
10.1038/s41563-020-0729-1