Skip to content

Deep Levels and Electron Paramagnetic Resonance Parameters of Substitutional Nitrogen in Silicon from First Principles

MetadataDetails
Publication Date2023-07-21
JournalNanomaterials
AuthorsChloĂ© Simha, Gabriela Herrero-Saboya, Luigi Giacomazzi, Layla Martin‐Samos, Anne HĂ©meryck
InstitutionsNational Research Council, University of Nova Gorica
Citations3
AnalysisFull AI Review Included

This study provides a definitive theoretical characterization of substitutional nitrogen (NSi) in silicon, resolving long-standing ambiguities regarding its structure and spectroscopic signatures (SL5 EPR center and DLTS levels).

  • Ground State Confirmation: The ground state of neutral NSi is confirmed to be an off-center configuration with C3v symmetry, rather than the high-symmetry on-center (Td) site.
  • EPR Validation (SL5 Center): Calculated Electron Paramagnetic Resonance (EPR) parameters (g-tensor and hyperfine A-tensors) for the off-center NSi configuration show excellent agreement with the experimentally observed SL5 center signature.
  • Metastability Model: The Potential Energy Surface (PES) was mapped, confirming the existence of a shallow, metastable on-center minimum (Td). The energy barrier for reorientation between equivalent off-center sites is calculated at 129 meV, consistent with experimental thermal activation estimates (0.11 ± 0.02 eV).
  • Reference Deep Levels: Highly accurate thermodynamic Charge Transition Levels (CTLs) were computed using the DFT + GW method, providing reference values for the single donor (0/+) and single acceptor (0/-) states.
  • DLTS Reassignment: The calculated CTLs (Donor: EVBM + 0.83 eV; Acceptor: ECBM - 0.55 eV) suggest that previous DLTS assignments (e.g., ECBM - 0.31 eV and ECBM - 0.08 eV) to NSi were likely incorrect.
ParameterValueUnitContext
NSi Ground State SymmetryC3v-Off-center configuration
NSi Metastable State SymmetryTd-On-center configuration
Energy Difference (Off-center vs. On-center)81meVOff-center configuration is lower in energy
Reorientation Energy Barrier129meVBarrier for N atom jump between equivalent off-center sites (via on-center Td state)
Experimental Reorientation Barrier0.11 ± 0.02eVInferred from EPR stress measurements (T > 35 K)
N-Si Bond Length (Off-center)1.86AngstromThree equivalent short bonds
N-Si Bond Length (Off-center)3.15AngstromOne elongated bond (along distortion axis)
N-Si Bond Length (On-center)2.05AngstromFour equivalent bonds
CTL Donor Level (0/+)EVBM + 0.83eVCalculated using PBE + GW approach
CTL Acceptor Level (0/-)ECBM - 0.55eVCalculated using PBE + GW approach
g-tensor Isovalue (giso)2.0062-Calculated for off-center NSi (SL5 center)
Hyperfine Aiso (29Si)-264.1MHzCalculated for Si atom along the distortion axis (off-center)
Hyperfine Aiso (14N)32.2MHzCalculated for Nitrogen impurity (off-center)

The investigation relied on advanced ab initio calculations combining Density Functional Theory (DFT) and Many-Body Perturbation Theory (GW).

  1. DFT Setup: Calculations were performed using the Quantum-ESPRESSO package, employing the Perdew-Burke-Ernzerhof (PBE) exchange-correlation functional.
  2. Supercell and Basis: A 216-atom silicon supercell was used, sampled at the Gamma (Γ) point for geometry optimization. A plane wave basis set with a kinetic energy cutoff of 84 Ry was employed.
  3. Potential Energy Surface (PES) Mapping: The Minimum Energy Path (MEP) between the on-center and off-center configurations was explored using the climbing-NEB (Nudged Elastic Band) method.
  4. EPR Parameter Calculation: Electron Paramagnetic Resonance (EPR) parameters (g-tensor and hyperfine A-tensors) were computed using the GIPAW (Gauge-Including Projector Augmented Wave) module within Quantum-ESPRESSO. A 3 x 3 x 3 Monkhorst-Pack grid was used for k-point integration to ensure convergence.
  5. Charge Transition Level (CTL) Calculation: Thermodynamic CTLs were determined using the combined DFT and GW (G0W0) approach, implemented in the ABINIT code, to accurately account for band gap corrections and self-interaction errors.
  6. GW Convergence: The GW calculations utilized the Godby-Needs plasmon-pole model, requiring 2600 bands and an energy cutoff of 30 eV for the exchange-correlation self-energy.

The precise understanding of the NSi defect is crucial for industries relying on nitrogen-doped silicon, particularly for controlling material properties and defect engineering.

  • Advanced Silicon Crystal Growth: Nitrogen is intentionally introduced in Czochralski (CZ) and Float Zone (FZ) silicon to suppress the diffusion of self-interstitials and restrain the formation of voids, leading to higher quality wafers for microelectronics.
  • Power Electronics and Devices: Accurate CTLs (EVBM + 0.83 eV and ECBM - 0.55 eV) provide essential data for modeling carrier recombination and lifetime control in silicon power devices (e.g., high-voltage diodes, IGBTs), where deep levels dictate device performance and efficiency.
  • Mechanical Strength Enhancement: Nitrogen doping is used to enhance the mechanical strength of silicon wafers, reducing warpage and dislocation formation during high-temperature processing steps.
  • Defect Engineering: The robust correlation between the calculated off-center NSi structure and the SL5 EPR signal provides a reliable reference point for identifying and characterizing other complex nitrogen-related defects (e.g., nitrogen-vacancy complexes) that impact device functionality.
  • Photovoltaic Manufacturing: Controlling nitrogen defects is relevant for mitigating bulk lifetime degradation kinetics in silicon solar cells, contributing to improved long-term efficiency and stability.
View Original Abstract

Nitrogen is commonly implanted in silicon to suppress the diffusion of self-interstitials and the formation of voids through the creation of nitrogen-vacancy complexes and nitrogen-nitrogen pairs. Yet, identifying a specific N-related defect via spectroscopic means has proven to be non-trivial. Activation energies obtained from deep-level transient spectroscopy are often assigned to a subset of possible defects that include non-equivalent atomic structures, such as the substitutional nitrogen and the nitrogen-vacancy complex. Paramagnetic N-related defects were the object of several electron paramagnetic spectroscopy investigations which assigned the so-called SL5 signal to the presence of substitutional nitrogen (NSi). Nevertheless, its behaviour at finite temperatures has been imprecisely linked to the metastability of the NSi center. In this work, we build upon the robust identification of the SL5 signature and we establish a theoretical picture of the substitutional nitrogen. Through an understanding of its symmetry-breaking mechanism, we provide a model of its fundamental physical properties (e.g., its energy landscape) based on ab initio calculations. Moreover by including more refined density functional theory-based approaches, we calculate EPR parameters (↔g and ↔A tensors), elucidating the debate on the metastability of NSi. Finally, by computing thermodynamic charge transition levels within the GW method, we present reference values for the donor and acceptor levels of NSi.

  1. 2002 - Mechanical properties of nitrogen-doped CZ silicon crystals [Crossref]
  2. 2008 - Multiplicity of Nitrogen Species in Silicon: The Impact on Vacancy Trapping
  3. 2009 - Nitrogen diffusion and interaction with dislocations in single-crystal silicon [Crossref]
  4. 1996 - Influence of oxygen and nitrogen on point defect aggregation in silicon single crystals [Crossref]
  5. 2004 - Nitrogen interaction with vacancies in silicon [Crossref]
  6. 1975 - Nitrogen- implanted silicon. II. Electrical properties [Crossref]
  7. 1982 - Deep levels associated with nitrogen in silicon [Crossref]
  8. 1985 - Nitrogen-related deep electron traps in float zone silicon [Crossref]
  9. 2016 - Permanent annihilation of thermally activated defects which limit the lifetime of float-zone silicon [Crossref]