Li-Ion Localization and Energetics as a Function of Anode Structure
Abstract
Neutron PDF experiments on lignin-derived carbon composite anodes are combined with large-scale MD: OPLS-AA equilibration of uncharged morphologies, then ReaxFF (Li–C–H) in LAMMPS for lithiated states to map Li localization, g®, 3D density maps, and binding-energy distributions vs processing-controlled structure.
Summary¶
The study connects experimental neutron scattering (pair distribution functions, PDFs) with atomistic molecular dynamics for lignin-derived carbon composites used as Li-ion anodes. Uncharged composite morphologies (crystalline nanodomains in an amorphous matrix) are built and equilibrated with OPLS-AA; explicit hydrogens are added where needed because the nonreactive model treats some H implicitly. The reactive stage uses ReaxFF for Li–C–H chemistry with QEq charges each step, implemented in LAMMPS with Δt = 0.25 fs. Uncharged cells are equilibrated in NVT at 298 K (Nosé–Hoover thermostat) until energies plateau. Li is then added to the amorphous regions (random placement avoiding close contacts) at loadings tied to experimental capacities for slurry electrodes pyrolyzed at different temperatures. Systems are large (~10⁵–7×10⁵ atoms; box edges ~100–200 Å, Table S1 in the article). Analysis includes g®, three-dimensional atomic density maps around Li, and Li binding energy distributions separating chemisorption vs physisorption trends across density, crystallite size, crystalline volume fraction, and charge loading.
Methods¶
Force-field training / fitting. OPLS-AA is used for uncharged lignin-derived carbon composite morphologies (with explicit hydrogens added where the nonreactive treatment requires). The reactive stage uses a published ReaxFF parametrization for Li–C–H with QEq charges each step; the article does not describe a new QM refit of ReaxFF here.
MD application (atomistic dynamics). LAMMPS drives both stages. Uncharged composites are equilibrated in NVT at 298 K with a Nosé–Hoover thermostat until energies plateau. Li is then placed in amorphous regions (random placement avoiding close contacts) at loadings tied to experimental capacities for electrodes pyrolyzed at different temperatures. Reactive trajectories use Δt = 0.25 fs. Reported cells are large (~10⁵–7×10⁵ atoms; box edges ~100–200 Å, Table S1 in the article). Three-dimensional periodic boundary conditions are the standard setup for these bulk composite cells (as implied by the supercell construction in the paper). Barostat / pressure: N/A — the equilibration and lithiation stages summarized are NVT at 298 K without hydrostatic barostat control in the protocol described on this page. Electric field: N/A — not used. Enhanced sampling: N/A — conventional MD trajectories.
Static QM / DFT. N/A — not the engine for the ReaxFF production runs summarized above (DFT may appear elsewhere in the paper for reference or construction).
Review / non-simulation framing. Neutron pair distribution function (PDF) measurements on slurry electrodes are integrated with the MD workflow; the work is not a narrative review.
Findings¶
Outcomes and mechanisms. Calculated pair distribution functions (PDFs) align with neutron PDF data for uncharged and lithiated lignin-derived carbon composites; lithiation perturbs the carbon PDF only modestly on the simulated timescales, mirroring experiment. Li accumulates at interfaces between crystalline nanodomains and the amorphous matrix; H-terminated crystallite edges act as favorable, more reversible adsorption sites in the binding-energy analysis.
Comparisons. Simulation PDFs and Li localization metrics are compared directly to neutron measurements across processing-controlled morphologies (pyrolysis temperature, crystallite size, volume fraction, loading).
Sensitivity and design levers. Higher small-crystallite volume fraction increases interfacial area and, in the models, supports higher capacity via more interfacial Li hosting. Binding-energy distributions span physisorption to chemisorption depending on morphology and Li content.
Limitations and outlook (as authored). Reactive MD captures short-time lithiation response; long-term structural evolution may need longer sampling (see article caveats).
Corpus honesty. Numerical g®, density maps, and energy histograms should be taken from the DOI-linked ACS AMI article and SI; this page is a narrative guide.
Limitations¶
Reactive simulations capture short-time lithiation response; long-term structural evolution may require longer sampling. Model construction follows the Pellenq-type composite workflow referenced in the paper.
Relevance to group¶
Demonstrates ReaxFF + experiment workflow for complex disordered carbon anodes relevant to electrochemical interfaces.
Citations and evidence anchors¶
- DOI:
10.1021/acsami.6b13748