Skip to content

Development of a ReaxFF reactive force field for lithium ion conducting solid electrolyte Li1+xAlxTi2−x(PO4)3 (LATP)

Evidence and attribution

Authority of statements

Prose sections below (Summary, Methods, Findings, etc.) are curated summaries of the publication identified by doi, title, and pdf_path in the front matter above. They are not new primary claims by this wiki.

For definitive numerical values, reaction schemes, and interpretations, use the peer-reviewed article (and optional records under normalized/papers/ when present)—not this page alone.

Summary

This work develops and benchmarks a ReaxFF parameterization for NASICON-type Li1+xAlxTi2−x(PO4)3 (LATP), a solid electrolyte candidate for all-solid-state Li-ion batteries. Parameters are trained against DFT data on equations of state, heats of formation for relevant oxides and phosphates, and Li diffusion barriers in TiO2 and LiTi2(PO4)3 (LTP). ReaxFF reproduces structural trends with composition—including site preference of Li near Al versus Ti and lattice contraction with Al substitution—and explores Li transport and ionic conductivity from 300–1100 K. Hybrid MC/MD is used to sample disordered LATP configurations; at x = 0.5 the model yields conductivity in closer agreement with experiment than lower-x structures, illustrating composition-sensitive transport and the utility of reactive MD for solid electrolyte screening.

Methods

Force-field training (ReaxFF for Li–Al–Ti–P–O). The authors develop a ReaxFF reactive force field for NASICON-type Li\(_{1+x}\)Al\(_x\)Ti\(_{2-x}\)(PO\(_4\))\(_3\) (LATP), starting from parent ReaxFF libraries for oxides/phosphates and extending interaction classes needed for Li transport in the Ti/Al-substituted framework. QM reference data come from DFT calculations of equations of state, heats of formation for reference crystals (Li\(_x\)TiO\(_2\), Al\(_2\)TiO\(_5\), LiAlO\(_2\), AlPO\(_4\), Li\(_3\)PO\(_4\), LiTi\(_2\)(PO\(_4\))\(_3\) (LTP)), and Li migration barriers in TiO\(_2\) and LTP via vacancy and interstitial pathways. Optimization follows the standard ReaxFF parameter fitting workflow described in PCCP (least-squares-style minimization of QM vs force-field errors across the training set). Validation / reference data additionally include experimental lattice trends used to judge Al substitution effects.

MD application + hybrid MC/MD. Engine: molecular dynamics with the fitted ReaxFF potential, combined with hybrid Monte Carlo / MD sampling to explore disordered Li\(_{1+x}\)Al\(_x\)Ti\(_{2-x}\)(PO\(_4\))\(_3\) arrangements (reactive MD in the sense of variable Li site occupancy moves plus dynamics). System: periodic supercells of LATP compositions spanning the \(x\) range studied (atom counts in article tables). PBC: three-dimensional periodic boundaries. Ensemble: production ionic conductivity trajectories are canonical NVT-style MD segments as reported in PCCP (NPT details N/A — not emphasized in the abstract-level summary on this page). Temperature: ionic conductivity and diffusion analyses span 300–1100 K as reported. Timestep / thermostat / barostat / production length: N/A — explicit fs timestep, thermostat family, and multi-ns production tables should be copied from the PCCP Methods rather than inferred from this summary. Pressure: N/A — not emphasized in the excerpted abstract-style summary on this page. Electric field: N/A — not used. Enhanced sampling: hybrid MC/MD beyond straight MD; N/A — umbrella / metadynamics not indicated in the indexed excerpt.

Findings

LiTi\(_2\)(PO\(_4\))\(_3\) (LTP) shows low room-temperature conductivity (\(\sim 5.9\times 10^{-5}\) S cm\(^{-1}\)). Substitution toward LATP at \(x=0.2\) raises conductivity modestly (\(\sim 8.4\times 10^{-5}\) S cm\(^{-1}\)) but remains below reported values for \(x\approx 0.3\)–\(0.5\). Hybrid MC/MD sampling at \(x=0.5\) produces a thermodynamically stable arrangement with conductivity \(\sim 7.4\times 10^{-4}\) S cm\(^{-1}\) near 300 K—about an order of magnitude above LTP and the \(x=0.2\) composition, matching the order of magnitude of independent measurements (\(\sim 2.5\times 10^{-4}\) S cm\(^{-1}\)). The strong composition dependence of conductivity highlights solid-solution disorder in NASICON-type electrolytes and supports hybrid MC/MD as a practical tool for sampling these configurations.

Across 300–1100 K, the study tracks Li diffusion mechanisms and ionic conductivity trends in LTP vs LATP, emphasizing that experimental conductivities on the order of \(10^{-4}\)–\(10^{-3}\) S cm\(^{-1}\) for \(x \approx 0.3\)–\(0.5\) are only approached in the ReaxFF model once a hybrid MC/MD-sampled \(x=0.5\) arrangement is used, while a simpler \(x=0.2\) realization remains sub-experimental. The authors argue this supports both the transferability of the LATP force field for solid electrolyte screening and the sensitivity of conductivity to solid-solution disorder. Limitations / outlook: conductivity remains sensitive to structural realization and sampling length scales, as noted under ## Limitations. Corpus honesty: tabulated conductivity values here follow the article abstract language mirrored in normalized/extracts/2018shin-physical-che-development-reaxff_p1-2.txt; confirm against pdf_path for final typesetting.

Limitations

  • Conductivity still depends strongly on structural realization and composition; reaching experimental values requires careful sampling and may need further refinement of training data or dynamics length scales.
  • High-temperature MD may not capture all long-time defect equilibria relevant near room temperature.

Relevance to group

Core example of ReaxFF parameterization for battery-relevant ceramic electrolytes with direct van Duin group authorship; connects reactive FF development to measurable transport properties for solid-state battery materials.

Citations and evidence anchors

  • DOI: 10.1039/c8cp03586e
  • Primary numerical results and fitting rationale: early sections and Results in PCCP paper; see normalized extract normalized/extracts/2018shin-physical-che-development-reaxff_p1-2.txt for text-aligned pointers.

Reader notes (navigation)