Skip to content

Growth mechanism study of boron nitride atomic layer deposition by experiment and density functional theory

Scope

ALD of BN from BCl\(_3\) or TDMAB with NH\(_3\) on Si(100): experiments map growth per cycle vs temperature; DFT identifies rate-limiting surface reactions setting the lower ALD window.

Summary

Atomic layer deposition of boron nitride must stay within a finite temperature window. Experiments compare BCl\(_3\)+NH\(_3\) vs tris-dimethylaminoborane (TDMAB)+NH\(_3\) on Si(100), measuring how growth per cycle (GPC) responds between roughly 700–900 °C. Static DFT then evaluates four representative elementary steps on OH-, Cl-, H-, and BN-terminated surfaces to locate the highest activation barrier controlling the lower temperature bound. The study is explicitly ALD-focused: precursor choice changes surface termination sequences between cycles, so GPC curves are interpreted together with stepwise ligand exchange pictures rather than as a single effective activation energy for all boron sources.

Methods

Experiments (ALD window and GPC)

  • Substrate: Si(100).
  • Precursor pairs: BCl\(_3\) + NH\(_3\) versus TDMAB + NH\(_3\) (tris(dimethylamino)borane).
  • Observables: Growth per cycle (GPC) vs temperature in an approximately 700–900 °C window (as described in the article), comparing temperature sensitivity between chemistries.

Static quantum chemistry (C)

  • Goal: Identify representative elementary steps on F-terminated / OH-terminated / H-terminated / BN-terminated Si(100) motifs relevant to BN ALD.
  • Reactions labeled (a)–(d) in the paper: (a) BCl\(_3\) on OH-terminated Si(100); (b) NH\(_3\) on Cl-terminated Si(100); © BCl\(_3\) on H-terminated BN on Si(100); (d) NH\(_3\) on Cl-terminated BN surfaces.
  • Electronic structure readout (full DFT): The PBE-family DFT setup, convergence tolerances, k-point / k-mesh sampling, slab geometries for the (a)–(d) reaction pathways on OH-, Cl-, H-, and BN-terminated Si(100) motifs, and resulting barrier energies (e.g. 30.0 kcal mol\(^{-1}\) for step ©) are reported in the primary Comput. Mater. Sci. text; DFT-D or other vdW/dispersion corrections and PAW/plane-wave / basis cutoffs are as stated there or N/A if the manuscript does not add a separate dispersion correction line in the indexed summary. N/A to NEB if the paper uses a different static path search; align with the published method name.

Findings

TDMAB + NH\(_3\) shows a steep increase in GPC with temperature in the studied window, whereas BCl\(_3\) + NH\(_3\) exhibits a weaker temperature dependence (qualitative comparison in the article).

DFT barrier assignment

Static DFT assigns an activation barrier of 30.0 kcal mol\(^{-1}\) to reaction © (BCl\(_3\) on H-terminated BN), interpreted as rate-limiting for the lower bound of the ALD temperature window. The corresponding activation temperature is about 789 °C, falling inside the experimental ALD window reported by the authors.

Mechanistic takeaway

GPC vs T behavior should be interpreted with the specific boron precursor and its surface termination sequence—not a single universal BN growth law across BCl\(_3\) vs TDMAB routes.

Limitations

DFT models idealized surfaces; reactor-scale nonuniformities are not captured. The Comp. Mater. Sci. workflow separates experimental GPC vs T trends for BCl\(_3\)+NH\(_3\) and TDMAB+NH\(_3\) from static barrier assignments on OH-, Cl-, H-, and BN-terminated Si(100) motifs; retrieval should preserve that pairing so ALD window arguments are not quoted without the DFT reaction labels (a)–(d) used in the manuscript. Surface hydroxyl/chloride coverages inferred from ideal slabs may differ from industrial reactor surfaces where contaminants and roughness shift effective barriers, stick probabilities, and local coverage.

Relevance to group

Peripheral DFT/ALD study in corpus (not ReaxFF).

Reader notes (navigation)

Citations and evidence anchors