Skip to content

Epitaxial Formation of Ultrathin HfO2 on Multilayer Graphene by Sequential Oxidation

Summary

STEM tracks epitaxial ultrathin hafnia grown by sequential oxidation of epitaxial Hf on multilayer graphene, resolving a-HfOₓ → h-HfOₓ → m-HfO₂ with substrate-aligned orientations. The work reports heteroepitaxial Hf → suboxide → monoclinic HfO₂ on graphene, with STEM (epitaxy, h-HfOₓ structure) and SNOM on conductivity across phases. ReaxFF reactive MD in AMS (with hybrid fbMC/MD in places) supports the oxidation sequence and graphene-templated crystallization versus O₂ loading and temperature ramps, complementing experiment.

Methods

  • Experiment: Epitaxial Hf on multilayer graphene; controlled oxidation to a-HfOₓ, h-HfOₓ, and m-HfO₂; STEM / diffraction for epitaxial relationships; SNOM contrast between phases; additional SI materials (see journal Supporting Information).
  • ReaxFF MD (AMS): Conjugate-gradient minimization; NVT equilibration at 300 K for 100 ps (Berendsen, 100 fs damping); separate Berendsen thermostats on Hf, 4-layer graphene, and O₂ during 300 → 900 K heating at 10 K/ps with Hf/O ReaxFF interactions disabled until oxidation; then annealing at 900 K NVT for 2 ns with interactions enabled (0.1 fs time step; 10,000 fs damping on Hf and O₂, 100 fs on graphene). 18 model variants with/without graphene and O₂ density documented (Table S1).
  • Hybrid fbMC/MD (AMS): Uniform-acceptance fbMC/MD with alternating 10,000 MD and 10,000 fbMC iterations (36,000,000 each), 0.1 fs timestep, ~3.6 ns effective MD at 900 K; parameters drₘₐₓ = 0.1, imcfrq = 10,000, imcstp = 10,000, imcroo = 4 per manuscript.

1 — MD application (atomistic dynamics). ReaxFF molecular dynamics in AMS (Amsterdam Modeling Suite) on Hf/O₂/graphene compositions (18 variants in Table S1); 3D PBC; NVT stages at 300 K and 900 K with 0.1 fs time step; Berendsen thermostats on Hf, 4-layer graphene, and O₂ with stated damping; Ramped temperature to 900 K with Hf/O interactions disabled then enabled for 2 ns anneal; fbMC/MD rare-event sampling as above. N/A — no NPT barostat in the quoted NVT blocks; N/A — no static electric field; N/A — not classical replica exchange; N/A — not metadynamics in the NVT sense (use fbMC/MD label for enhanced sampling here).

2 — Force-field training — N/A to first principles (applies a published Hf/C/H/O ReaxFF stack; covalent C–Hf training limits per manuscript).

3 — Static QM — N/A (STEM/SNOM experiment + ReaxFF MD story).

4 — Review — N/A.

Findings

  • Sequential oxidation of epitaxial Hf on multilayer graphene gives a-HfOₓ → h-HfOₓ → m-HfO₂ with epitaxial alignment, with STEM-resolved displacive relations among hcp-Hf, h-HfOₓ, and m-HfO₂ relative to graphene (see main text and SI).
  • ReaxFF molecular dynamics reproduces coexistence of crystalline (hcp-Hf, h-HfOₓ, m-HfO₂) and amorphous (a-Hf, a-HfOₓ) domains and supports the proposed oxidation sequence; O₂ density and temperature ramp are swept in Table S1 variants, giving sensitivity to processing levers. The Hf/C/H/O set omits covalent graphene–Hf training, so MD should not be read as proving chemical adhesion; STEM and diffraction remain the primary experiment versus simulation comparison for epitaxy and phase labels. Limitations and outlook for interface chemistry are in ## Limitations and the manuscript.

Limitations

Hf/C/H/O ReaxFF coverage does not include explicit covalent graphene–Hf bond training; direct film–substrate bonding and graphene reconstruction are outside the model. ACS Nano details and SI tables should be checked for final published pagination.

Relevance to group

Adri C. T. van Duin (Penn State) contributes ReaxFF modeling partnered with MIT in situ TEM / growth expertise (Ross, Jaramillo).

Citations and evidence anchors