First-principles study of hydrogen-related defects at the a-SiO2/Si(100) interface Authors: Hong Luo, Di Qi, Zizhao Ma, Xin Guo, Hang Zhou, Yang Liu, Xu Zuo Published: 2026 ABSTRACT During the fabrication and annealing of amorphous SiO2/Si interfaces, hydrogen is commonly introduced to passivate interfacial dangling bonds, yet it may also generate electrically active hydrogen-related defects. Here, first-principles calculations are performed to systematically investigate the structural evolution and electronic properties of hydrogen-related defects at ten representative sites within the suboxide region of an amorphous SiO2/Si(100) interface under neutral and charged conditions. To eliminate spurious long-range Coulomb interactions arising from periodic boundary conditions in charged-defect calculations, electrostatic energy corrections are implemented by combining density functional theory with a classical finite-element solution of the Poisson equation. Corrected charge transition levels are subsequently determined for all defect configurations. The results suggest that electrostatic corrections can induce substantial shifts in charge transition levels, with maximum deviations reaching 0.44 eV. For most defect sites, only one thermodynamically active charge transition level is found to be located within the silicon band gap for most of the sampled defect configurations. In addition, the formation energies of hydrogen-related defects are found to span a wide range from 1.71 to 5.23 eV, exhibiting a clear dependence on the spatial position of defects within the interfacial suboxide layer. These results provide insight into the charge behavior and thermodynamic stability of hydrogen-related defects at amorphous SiO2/Si interfaces, offering an atomistic foundation for understanding their potential impact on device performance and reliability.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-026-47176-4. Introduction Silicon dioxide is one of the most important dielectric materials in modern microelectronic devices, and amorphous silicon dioxide (a-SiO2) can be readily formed through the thermal oxidation of silicon1–3. In electronic device architectures, the a-SiO2/Si interface plays a critical role in achieving high-quality gate insulation, regulating channel carrier transport, and ensuring long-term device reliability4. However, due to the inherent structural disorder of amorphous materials and the complexity of fabrication processes, a variety of point defects inevitably form in the interfacial region, which can significantly degrade the electrical performance and stability of devices5. Hydrogen has long been involved in various stages of industrial processing and is widely used to passivate electronic and optical defects in microelectronic devices6,7. At the a-SiO2/Si interface, the incorporation and behavior of hydrogen have been demonstrated to be key factors influencing interfacial properties. On the one hand, hydrogen can effectively passivate interfacial dangling bonds, such as Pb centers, by forming stable Si–H bonds, thereby reducing the interface state density8–11. On the other hand, hydrogen can also participate in the formation of various defect configurations. These hydrogen-related defects are often electrically active and can act as trapping centers for charge carriers, facilitate charge migration, and contribute to negative bias temperature instability (NBTI). Such defect-assisted processes are further exacerbated under irradiation conditions, leading to pronounced degradation of device performance and long-term reliability12–15. Consequently, elucidating the atomic-scale structural stability and charge behavior of hydrogen-related point defects has become a central scientific issue in understanding the physics of the a-SiO2/Si interface and its impact on device reliability. Among various defect characteristics, the formation energy and charge transition levels are regarded as key quantities for assessing thermodynamic stability and electrical activity16,17. The formation energy provides a quantitative measure of defect stability, while charge transition levels determine the charge state of a defect as a function of the Fermi level and are closely related to its ability to capture carriers. Therefore, accurate calculations of the formation energies and charge transition levels of hydrogen-related defects at the a-SiO2/Si interface are essential for evaluating their physical properties. For the a-SiO2/Si interface model, previous studies have demonstrated the presence of a suboxide layer at the interface, in which silicon atoms exhibit a gradual variation in oxidation states. This region forms a transitional layer characterized by multiple silicon oxidation states, where Si1+ and Si2+ species are preferentially located closer to the crystalline Si side, while Si3⁺ species are mainly distributed toward the a-SiO2 side18,19. Such a spatially nonuniform distribution of oxidation states leads to pronounced structural disorder and results in the accumulation of local interfacial stress20. Such local stress may drive the structural evolution of defects, giving rise to distinct defect configurations, some of which can introduce deep levels within the band gap. These deep-level states can significantly influence carrier trapping and transport, ultimately exerting a detrimental effect on device performance. Therefore, the influence of local stress on the structural evolution of hydrogen-related defects represents an important phenomenon that warrants systematic investigation. Oxygen vacancies are among the most prevalent and extensively investigated defect types in silicon oxide materials21–26. A substantial body of work has been devoted to the interaction between hydrogen and oxygen vacancies. For example, Blöchl systematically studied hydrogen-related reactions involving oxygen vacancies in quartz and determined their energetic, structural, charge-state, and hyperfine characteristics27. Yue et al. employed first-principles calculations to simulate the interaction between hydrogen and oxygen vacancies in bulk a-SiO2, revealing several possible hydrogen-related defect configurations28. Godet et al. investigated the behavior of protons at the Si(100)/SiO2 interface using density functional theory29, while Yao et al. further explored the energy barriers associated with proton-related defects at the a-SiO2/Si interface30. These studies collectively demonstrate that the interaction between hydrogen and oxygen vacancies is one of the key factors governing defect properties in SiO2, amorphous SiO2, and their interfacial systems. In particular, for the a-SiO2/Si interface, which constitutes a core component of transistor device structures, a systematic investigation of hydrogen, oxygen vacancy interactions within the interfacial suboxide layer is crucial for understanding the formation and evolution mechanisms of interface defects. In this work, oxygen vacancies were generated by removing oxygen atoms from Si–O–Si units at ten distinct sites within the suboxide layer of a defect-free a-SiO2/Si interface model. Hydrogen atoms were subsequently introduced at the vacancy sites to allow full interaction with the oxygen vacancies. We first examine the fully relaxed configurations of hydrogen–oxygen vacancy complexes at different interfacial sites under neutral and charged conditions. To properly account for spurious long-range Coulomb interactions in charged-defect calculations, electrostatic energy corrections are evaluated by solving the Poisson equation using a finite-element approach. Based on the corrected total energies, charge transition levels are determined for the considered defect configurations. Finally, the formation energies of hydrogen-related defects in different charge states are calculated to assess their thermodynamic stability. Methods First-principles calculation All first-principles calculations are conducted based on density functional theory (DFT) as implemented in the Vienna Ab initio Simulation Package (VASP)31–33. The exchange–correlation effects are calculated by using the Perdew-Burke-Ernzerhof (PBE) functional within the generalized gradient approximation (GGA) framework. Electron–ion interactions are considered by using the projector augmented wave (PAW) method. The plane-wave cutoff energy of 520 eV was adopted to ensure reliable total energy convergence. Due to the large size of the amorphous supercell, only the Γ-point is employed for Brillouin zone sampling34. Spin-polarized calculations were performed to accurately describe potential magnetic contributions arising from unpaired electrons. Geometry optimizations are carried out using the conjugate gradient method with a maximum of 300 ionic steps. Structural relaxation was deemed converged when the maximum force on each atom was below 0.05 eV/Å. For the electronic self-consistent field (SCF) calculations, the energy convergence criterion was set to 10-5 eV, and the number of electronic steps was limited to 300.To evaluate the electronic properties, Gaussian smearing with a width of 0.01 eV was used. To overcome the systematic underestimation of the band gap inherent to the PBE functional and to improve the accuracy of electronic energy calculations, the Heyd–Scuseria–Ernzerhof(HSE) hybrid functional was further employed for electronic structure calculations based on the PBE-relaxed equilibrium geometries35. The structure model of amorphous SiO2/Si(100) interface model are obtained by Li et al.36, with a supercell size of 15.24 Å × 15.24 Å × 40.00 Å. The calculation of formation energy Point defects can have a profound impact on the performance of electronic devices. The formation energy of a defect provides a direct measure of its structural stability. Density functional theory is commonly employed to calculate the formation energies of interfacial defects, as it enables an accurate quantum-mechanical description of defect-related electronic properties, including band structure, density of states, and charge distribution. Due to limitations in computational resources, DFT calculations are typically performed using relatively small supercells combined with periodic boundary conditions to approximate infinitely extended interfaces in an efficient manner. However, when defects carry a net charge, the use of periodic boundary conditions gives rise to Coulomb interactions between the defect and its periodic images, which can significantly compromise the accuracy of the calculated formation energies37. Specifically, the long-range Coulomb repulsion between charged defects and their periodic replicas in finite supercells introduces nonphysical energy shifts38. In addition, defect-state wave functions may overlap with their image states, further affecting the charge distribution and the positions of defect levels within the band structure. Together, these effects lead to systematic errors in both defect formation energies and charge transition levels. To address these issues, appropriate correction schemes must be introduced to eliminate the spurious long-range Coulomb interactions arising from periodic boundary conditions, thereby recovering the physically meaningful properties of charged defects in the isolated limit. The corrected formation energy of a charged point defect is given by the following expression37,39,1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}^{f}\left({X}^{q}\right)={E}^{tot}\left({X}^{q}\right)-{E}_{tot}\left(pristine\right)+{E}_{q}^{corr}+q\left({\epsilon }_{vbm}^{pristine}+{\epsilon }_{F}-\Delta {V}_{0/p}\right)-\sum_{i}{n}_{i}{\mu }_{i}$$\end{document} Where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}^{f}\left({X}^{q}\right)$$\end{document} is the formation energy of the defect in charge state,\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}^{tot}\left({X}^{q}\right)$$\end{document} is the total energy of the supercell containing the defect,\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{tot}\left(pristine\right)$$\end{document} is the total energy of the corresponding pristine (defect-free) supercell,\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{q}^{corr}$$\end{document} is the electrostatic correction that accounts for the spurious interactions between the charged defect and its periodic images, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\epsilon }_{F}$$\end{document} is the Fermi level referenced to the valence band maximum (VBM), \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\epsilon }_{vbm}^{pristine}$$\end{document} is the energy of the valence band maximum (VBM) in the pristine (defect-free) supercell,\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta {V}_{0/p}$$\end{document} is the potential alignment term between the neutral defect supercell and the pristine supercell,\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${n}_{i}$$\end{document} denotes the number of atoms of species that are added (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${n}_{i}$$\end{document}>0) or removed (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${n}_{i}$$\end{document}<0) to form the defect,\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mu }_{i}$$\end{document} is the chemical potential of species;and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$q$$\end{document} is the charge state of the defect. The calculation formula for the electrostatic energy correction term is as follows.2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{q}^{corr}={E}_{q}^{lattice}-q\Delta {V}_{q-0/m}$$\end{document} Where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{q}^{lattice}$$\end{document} is the lattice energy, and its calculation formula is as follows.3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{q}^{lattice}={E}_{q}^{isolated}-{E}_{q}^{period}$$\end{document} Among them, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{q}^{period}$$\end{document} represents the electrostatic energy calculated using the supercell model with periodic boundary conditions, while \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{q}^{isolated}$$\end{document} is the isolated energy obtained by proportionally enlarging the supercell model. The calculation formula for \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{q}^{period}$$\end{document} is as follows,4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{q}^{period}=\frac{1}{2}\int {\rho }^{m}\left(\overrightarrow{r}\right){V}_{q}^{per,m}\left(\overrightarrow{r}\right)dr$$\end{document} Where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\rho }^{m}$$\end{document} represents the spatial charge density of the model, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{q}^{per,m}$$\end{document} represents the electrostatic potential of the model. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{q}^{per,m}$$\end{document} can be solved numerically using the finite element method to solve the Poisson equation, thereby obtaining the electrostatic potential inside the model. The calculation formula is as follows39,40,5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\nabla \cdot \left[\varepsilon \left(\overrightarrow{r}\right)\nabla {V}_{q}^{per,m}\left(r\right)\right]=-4\pi {\rho }^{m}\left(\overrightarrow{r}\right)$$\end{document} Where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon$$\end{document} represents the dielectric constant distribution of the model, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\rho }^{m}$$\end{document} represents the spatial charge density distribution of the model. In reciprocal space, the electrostatic potential \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{q}^{per,m}$$\end{document} of the model can be solved according to formula (5). The detailed formulations used to calculate \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\rho }^{m}$$\end{document} are provided in SI.1 and SI. 2. In formula (2), the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta {V}_{q-0/m}$$\end{document} term aligns the reference potential between DFT calculations and model calculations. The calculation formula for this term is as follows,6\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta {V}_{q-0/m=}{(V}_{q}^{DFT}-{V}_{0}^{DFT})-{V}_{q}^{per,m}$$\end{document} In formula (1), \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta {V}_{0/p}$$\end{document} represents the difference between the electrostatic potential of the neutral defect and the electrostatic potential of the original supercell without the defect. The calculation formula for this term is as follows,7\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta {V}_{0/p}={V}_{0}^{DFT}-{V}_{p}^{DFT}$$\end{document} The charge transition levels is defined as the Fermi energy at which the formation energies of two charge states q1 and q2 are equal. Its value is given by the following formula,8\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon \left(q1/q2\right)=\frac{{E}^{f}\left({X}^{q1}\right)-{E}^{f}\left({X}^{q2}\right)}{q2-q1}$$\end{document} On the basis of the formation energy electrostatic correction term, the charge transition level can be further expressed according to formula (8) as follows,9\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon \left(+/0\right)={E}_{tot}\left(0\right)-{E}_{tot}\left(+\right)-{E}_{corr}^{+}-q\left({\epsilon }_{vbm}^{pristine}-\Delta {V}_{0/p}\right)$$\end{document}10\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon \left(0/-\right)={E}_{tot}\left(-\right)-{E}_{tot}\left(0\right)+{E}_{corr}^{-}+q\left({\epsilon }_{vbm}^{pristine}-\Delta {V}_{0/p}\right)$$\end{document} Where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{tot}\left(q\right)$$\end{document} represents the total energy of the supercell containing the defect under different charge states (q=0, 1, -1). \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{corr}^{q}$$\end{document} is the electrostatic correction term for the charged defect, which can be obtained from formulas (2) and (3), and can also be written as,11\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{q}^{corr}={E}_{q}^{isolated}-{E}_{q}^{period}-q\Delta {V}_{q-0/m}$$\end{document} The calculation procedure In this work, considering the complexity of the system, we combined first-principles calculations with classical finite-element methods to achieve accurate evaluations of the correction terms. This methodological framework has been thoroughly validated in previous calculations of other defect-related properties in the a-SiO2/Si interface model, confirming its suitability and reliability for addressing charged defect problems at a-SiO2/Si interfaces36,41. The general procedure for calculating the formation energy correction of charged defects is usually shown in Fig. 1. The yellow box represents computations performed by Density Functional Theory (DFT)42,43, the green box represents computations performed by COMSOL, and the pink box represents data processing carried out by Origin.Fig. 1Calculation procedure for the formation energy of charged defects. Calculation procedure for the formation energy of charged defects. The general steps of the calculation are as follows, (1) Use VASP to calculate the planar average electrostatic potential \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{p}^{DFT}$$\end{document} of the pristine supercell without defects. (2) Use VASP to calculate the planar average electrostatic potential \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{0}^{DFT}$$\end{document} of the supercell with neutral defects. (3) Use VASP to calculate the planar average electrostatic potential \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{q}^{DFT}$$\end{document} of the supercell with charged defects. (4) A classical electrostatic model was constructed based on the atomic coordinates of the relaxed defect structure, the spatial distribution of the dielectric constant, the supercell dimensions, and the standard deviation of the Gaussian charge distribution derived from the defect-induced charge density. Using this information, the model electrostatic potential \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{q}^{per,m}$$\end{document} under periodic boundary conditions was computed with COMSOL Multiphysics.The electrostatic energy under periodic boundary conditions, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{q}^{period}$$\end{document}, was obtained from the classical model. To approximate the electrostatic energy under isolated boundary conditions, the supercell was uniformly scaled using a multiplicative factor \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\lambda$$\end{document}, increasing from \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\lambda =2}^{0}$$\end{document} to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\lambda =2}^{10}$$\end{document}, such that the lattice parameters were scaled as \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$(a,b,c\to \lambda a,\lambda b,\lambda c)$$\end{document}, where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$a,b,c$$\end{document} are the original lattice vectors.The extrapolated energy at infinite cell size yields the isolated electrostatic energy \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{q}^{isolated}$$\end{document}, and the lattice correction term is then calculated as \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{q}^{isolated}-{E}_{q}^{period}$$\end{document}. (5) Calculate the electrostatic potential alignment term \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta {V}_{0/p}$$\end{document} between the supercell with neutral defects and the original supercell without defects using formula (7). (6) Calculate the reference potential alignment term \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta {V}_{q-0/m}$$\end{document} between DFT calculations and model calculations using formula (6). Results and discussion In this study, hydrogen-related defects were constructed based on a defect-free a-SiO2/Si interface supercell, as shown in Fig. 2(a). Oxygen vacancies were first generated by removing oxygen atoms from selected Si–O–Si units within the interfacial suboxide layer. Subsequently, hydrogen atoms were introduced in the vicinity of these oxygen vacancies, allowing them to fully interact with the vacancy sites and form hydrogen-related defect configurations. The spatial locations of the defects within the supercell are illustrated in Fig. 2(b). In addition, Fig. 2(c) presents the relative positions of the ten selected defect sites within the interfacial suboxide region, highlighting their distribution across the amorphous interface.Fig. 2(a) Schematic illustration of the defect-free a-SiO2/Si interface model, viewed along the (101) direction. Silicon, oxygen, and hydrogen atoms are represented by blue, red, and white spheres, respectively. (b) Relative spatial distribution of the defect sites within the supercell; the shaded region denotes the interfacial suboxide layer. (c) Relative positions of the different defect sites within the suboxide region of the interface. (a) Schematic illustration of the defect-free a-SiO2/Si interface model, viewed along the (101) direction. Silicon, oxygen, and hydrogen atoms are represented by blue, red, and white spheres, respectively. (b) Relative spatial distribution of the defect sites within the supercell; the shaded region denotes the interfacial suboxide layer. (c) Relative positions of the different defect sites within the suboxide region of the interface. To facilitate clear identification and discussion of the calculated results, each defect is labeled according to the index of the oxygen atom removed from the pristine interface model. The defect configurations formed at each interfacial site under different charge states are summarized in Table 1. A detailed analysis of the corresponding atomic structures and their structural evolution is presented in Defect Structure Variation Across Charge States .Table 1Defect configurations formed at different sites under various charge states.Defect configurationDefect sitePositive charge stateNeutral charge stateNegative charge stateSi-H-Si5,11,19,24,31,44,6211,3111,31Si-H--Si22,32,68Si-H···Si5,19,22,24,32,44,62,6819,22,32,44,68fivefold-coordinated silicon5,24,62 Table 1Defect configurations formed at different sites under various charge states.Defect configurationDefect sitePositive charge stateNeutral charge stateNegative charge stateSi-H-Si5,11,19,24,31,44,6211,3111,31Si-H--Si22,32,68Si-H···Si5,19,22,24,32,44,62,6819,22,32,44,68fivefold-coordinated silicon5,24,62 Defect configurations formed at different sites under various charge states. Defect structure variation across charge states In the positively charged state In the positively charged defect systems, full structural relaxation was performed for ten distinct defect sites. All stable configurations can be classified into two categories: symmetric positive hydrogen bridges Si-H-Si(HB+) and asymmetric positive hydrogen bridges Si-H···Si(a-HB+). Among them, seven defect sites relax into the symmetric Si-H-Si configuration, as illustrated in Fig. 3(a). In this structure, the hydrogen atom forms covalent bonds with two neighboring silicon atoms, giving rise to an approximately symmetric hydrogen-bridge geometry. Specifically, the projected density of states (PDOS) was calculated to decompose the total density of states into contributions from individual atoms (Si、O and H), allowing the microscopic origin of defect-related electronic states to be identified. Projection density of states (PDOS) analysis, shown in Fig. 4(a), suggests that this symmetric hydrogen-bridge configuration introduces defect states near the conduction band minimum (CBM) of silicon, with the dominant contributions arising from Si and O atoms. The corresponding Si–H bond lengths range from 1.59 to 1.73 Å, in good agreement with previous first-principles studies of positive hydrogen bridge defects in silicon oxide systems27,28,44. It is worth noting that, in contrast to bulk quartz, where the HB+ bond angle is relatively fixed at approximately 148.4°, the hydrogen-bridge bond angle in the interfacial suboxide region exhibits a wide range from 135.8° to 176.8°. This broad distribution suggests that structural disorder and inhomogeneous local stress inherent to the amorphous interfacial suboxide layer on the defect geometry.Fig. 3Local atomic structures (left) and the corresponding two-dimensional charge density distributions (right) of representative hydrogen-related defects at the a-SiO2/Si interface under different charge states: (a) symmetric hydrogen-bridge Si-H-Si configuration in the positive charge state; (b) asymmetric hydrogen-bridge Si–H--Si configuration in the positive charge state; (c) Si–H···Si configuration in the neutral charge state; (d) fivefold-coordinated silicon defect configuration formed in the negative charge state.Fig. 4Projection density of states of representative hydrogen-related defects under different charge states using the HSE Functional: (a) symmetric hydrogen-bridge Si–H–Si configuration(HB+) in the positive charge state; (b) asymmetric hydrogen-bridge Si–H--Si configuration in the positive charge state; (c) Si–H–Si configuration in the neutral charge state; (d) Si–H···Si configuration in the neutral charge state; (e) Si–H···Si configuration in the negative charge state; (f) fivefold-coordinated silicon defect configuration formed in the negative charge state. The Fermi level is referenced to 0 eV. Insets show the corresponding atomic structures of each defect configuration. Local atomic structures (left) and the corresponding two-dimensional charge density distributions (right) of representative hydrogen-related defects at the a-SiO2/Si interface under different charge states: (a) symmetric hydrogen-bridge Si-H-Si configuration in the positive charge state; (b) asymmetric hydrogen-bridge Si–H--Si configuration in the positive charge state; (c) Si–H···Si configuration in the neutral charge state; (d) fivefold-coordinated silicon defect configuration formed in the negative charge state. Projection density of states of representative hydrogen-related defects under different charge states using the HSE Functional: (a) symmetric hydrogen-bridge Si–H–Si configuration(HB+) in the positive charge state; (b) asymmetric hydrogen-bridge Si–H--Si configuration in the positive charge state; (c) Si–H–Si configuration in the neutral charge state; (d) Si–H···Si configuration in the neutral charge state; (e) Si–H···Si configuration in the negative charge state; (f) fivefold-coordinated silicon defect configuration formed in the negative charge state. The Fermi level is referenced to 0 eV. Insets show the corresponding atomic structures of each defect configuration. In addition, three defect sites relax into an asymmetric positive hydrogen-bridge Si-H--Si configuration(a-HB+) under the positively charged condition, as shown in the left panel of Fig. 3(b). In this configuration, the hydrogen atom forms a stable covalent bond with only one silicon atom, while interacting weakly with the second silicon atom. This weak interaction corresponds to a significantly elongated Si--H separation, where the notation “--" denotes a nonbonding interaction. The two-dimensional charge density distribution (right panel of Fig. 3(b)) clearly shows the absence of appreciable bonding charge accumulation between the hydrogen atom and the second silicon atom. Correspondingly, PDOS analysis (Fig. 4(b)) reveals that this asymmetric hydrogen-bridge configuration introduces localized defect states within the silicon band gap. Detailed information on the structural parameters can be found in SI. 3.1. Overall, under positively charged conditions, the interaction between H+ and oxygen vacancies in the suboxide region of the a-SiO2/Si interface is found to preferentially form symmetric and asymmetric hydrogen-bridge configurations among the sampled defect sites. Compared with bulk a-SiO2, where additional defect motifs such as sp2-coordinated silicon and four-membered ring puckering structures may also form28, the interfacial environment may favor these hydrogen-bridge configurations, This behavior may be associated with the combined effects of local stress distribution and structural disorder in the suboxide layer. In the neutral state Among the ten neutral defect cases considered in this study, eight sites relax into the Si-H···Si configuration, while the remaining two sites form a Si-H-Si defect structure. For these two neutral Si-H-Si(HB0) cases, the relaxed geometry is similar to that of the symmetric hydrogen bridge observed in the positively charged state, where the hydrogen atom forms covalent bonds with both neighboring silicon atoms. However, as indicated by the PDOS shown in Fig. 4(c), the neutral hydrogen bridge does not introduce any defect states within the silicon band gap. For the eight Si-H···Si configurations, as illustrated in the left panel of Fig. 3(c), the hydrogen atom bonds to only one silicon atom Si(1), while the bond between hydrogen and the second silicon atom is broken, resulting in a dangling-bond configuration. In contrast to the asymmetric hydrogen bridge a-HB+ observed in the positively charged system, no appreciable interaction exists between the hydrogen atom and the threefold-coordinated silicon atom in the Si-H···Si configuration. This feature is clearly evidenced by the two-dimensional charge density profiles shown in the right panel of Fig. 3(c), where significant charge accumulation is observed only along the Si(1)–H bond, while a pronounced charge depletion region separates hydrogen from the second silicon atom. From an electronic-structure perspective, the PDOS presented in Fig. 4(d) reveals that the Si-H···Si configuration introduces two spin-split defect levels within the silicon band gap. These defect states are predominantly derived from the electronic states of silicon and oxygen atoms. Detailed information on the structural parameters can be found in SI. 3.2. In summary, under neutral conditions, the interaction between H⁰ and oxygen vacancies in the suboxide layer of the a-SiO2/Si interface is found to result in two distinct defect configurations among the sampled cases. The neutral Si–H–Si hydrogen bridge does not appear to introduce defect states within the silicon band gap, whereas the Si–H···Si configuration is found to give rise to localized gap states that may act as carrier trapping centers. Notably, in contrast to bulk a-SiO2, where the reaction between hydrogen and neutral oxygen vacancy dimers yields exclusively Si-H···Si defects28, the interfacial suboxide environment appears to favor the stabilization of neutral hydrogen bridge configurations. In the negatively charged state Under negatively charged conditions, the ten defect cases relax into three distinct types of stable configurations. Specifically, five cases evolve into Si-H···Si configurations, three cases form fivefold-coordinated silicon defects, and the remaining two cases stabilize as Si-H-Si structures. For the five Si-H···Si configurations obtained in the negative charge state, their structural characteristics are essentially identical to those of the neutral Si-H···Si defects. As shown in the PDOS in Fig. 4(e), these defects also introduce electronic states within the silicon band gap, with the dominant contributions arising from silicon and oxygen atoms. In addition, two defect sites relax into Si-H-Si configurations(HB-) under negative charging, which are structurally similar to the hydrogen bridge defects observed in the positively charged state, where the hydrogen atom forms covalent bonds with both neighboring silicon atoms. Notably, three defect sites in the negatively charged system evolve into fivefold-coordinated silicon configurations after full structural relaxation. As illustrated in the left panel of Fig. 3(d), the two silicon atoms adjacent to the oxygen vacancy approach each other and form a Si–Si dimer, while the hydrogen atom binds to one of the silicon atoms, giving rise to a fivefold-coordinated defect center. The corresponding two-dimensional charge density profiles shown in the right panel of Fig. 3(d) clearly indicate a bonding interaction between Si(1) and Si(2), confirming the formation of the dimer structure. In contrast, the hydrogen atom forms a bond only with Si(1), and no significant charge interaction is observed between hydrogen and Si(2), indicating the absence of a stable bonding state. Detailed information on the structural parameters can be found in SI. 3.3. The PDOS analysis presented in Fig. 4(f) indicates that this fivefold-coordinated silicon defect introduces defect levels within the silicon band gap. Although silicon, oxygen, and hydrogen atoms all contribute to these defect states, the dominant contribution originates from silicon atoms. This behavior differs from that of the fivefold-coordinated silicon defects formed by negatively charged hydrogen–oxygen vacancy complexes in quartz (SiH+SiSi(5)-)27, where the threefold-coordinated silicon atom typically captures an electron and subsequently bonds with a backward silicon atom, leading to an irregular fivefold coordination. In contrast, at the a-SiO2/Si(100) interface, the structural disorder inherent to the amorphous network and the presence of localized interfacial stress may favor structural reconstruction of the threefold-coordinated silicon atom toward the Si-H moiety, which may contribute to the formation of a fivefold-coordinated silicon defect. Calculation of electrostatic correction The detailed procedure for calculating the electrostatic correction of charged defects is described in Section “The calculation procedure”. Owing to the relatively large spatial extent of the a-SiO2/Si interface system and the pronounced dielectric inhomogeneity along the interface normal direction, solving the electrostatic potential directly in reciprocal space becomes highly nontrivial. To address this issue, a finite-element-based approach is innovatively employed in this work to solve the Poisson equation and evaluate the electrostatic correction term for charged defects. Specifically, the spatially resolved dielectric profile \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon \left(z\right)$$\end{document} and the Gaussian charge distribution \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\rho \left(r\right)$$\end{document} of the relaxed defect structures are extracted from first-principles calculations and subsequently imported into the finite-element analysis software COMSOL Multiphysics to construct a classical electrostatic model, as illustrated in Fig. 5(a). The related analysis of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\rho }^{m}$$\end{document} is provided in SI. 1 and SI. 2. Under periodic boundary conditions, the simulation cell is uniformly expanded by a scaling factor λ ranging from 20 to 210, allowing the electrostatic energy to converge with increasing system size. In parallel, the same uniform expansion procedure is performed under grounded boundary conditions to construct an isolated-boundary model. As the scaling factor increases, the electrostatic energies obtained under both boundary conditions converge to the same limiting value, as shown in Fig. 5(c). This converged value is defined as the electrostatic energy of the isolated system \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{q}^{isolated}$$\end{document}.Fig.5(a) The distribution of the electrostatic potential given the dielectric constant distribution ϵ(z) and the Gaussian charge distribution ρ(r). (b) The difference between the planar average electrostatic potential calculated by DFT and the model electrostatic potential solved by COMSOL, and (c) The isolated energy of the model solved under periodic boundary conditions and ground boundary conditions. (a) The distribution of the electrostatic potential given the dielectric constant distribution ϵ(z) and the Gaussian charge distribution ρ(r). (b) The difference between the planar average electrostatic potential calculated by DFT and the model electrostatic potential solved by COMSOL, and (c) The isolated energy of the model solved under periodic boundary conditions and ground boundary conditions. Because the reference potential in first-principles calculations differs from that used in the classical finite-element model, a potential alignment procedure is required. As shown in Fig. 5(b), the potential alignment term \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{q-0/m}$$\end{document} is obtained by comparing the planar-averaged electrostatic potential difference between the positively charged and neutral systems far from the defect region in the DFT calculations with the corresponding electrostatic potential from the finite-element model. Using this approach, the electrostatic correction terms for the ten distinct defect sites are summarized in Table 2.Table 2Electrostatic correction energies and charge transition levels ε(+/0) and ε(0/−) for the ten different defect sites.Defect site\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{corr}^{+}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{corr}^{-}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon \left(+/0\right)$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon \left(0/-\right)$$\end{document}50.290.35-0.421.12110.240.250.531.52190.420.420.291.67220.440.42-0.261.75240.280.270.280.77310.250.260.621.52320.290.32-0.051.16440.270.34-0.390.91620.290.34-0.181.05680.360.370.351.62All values are given in eV. Table 2Electrostatic correction energies and charge transition levels ε(+/0) and ε(0/−) for the ten different defect sites.Defect site\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{corr}^{+}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{corr}^{-}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon \left(+/0\right)$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon \left(0/-\right)$$\end{document}50.290.35-0.421.12110.240.250.531.52190.420.420.291.67220.440.42-0.261.75240.280.270.280.77310.250.260.621.52320.290.32-0.051.16440.270.34-0.390.91620.290.34-0.181.05680.360.370.351.62All values are given in eV. Electrostatic correction energies and charge transition levels ε(+/0) and ε(0/−) for the ten different defect sites. All values are given in eV. It is worth noting that previous studies reported electrostatic correction terms of approximately 0.48 eV for charge transition levels in quartz27and about 0.35 eV for hydrogen bridge defects in bulk a-SiO245. In contrast, within the suboxide layer of the a-SiO2/Si interface, the electrostatic correction terms exhibit noticeable site dependence, which may be related to variations in local stress and local atomic configurations. After applying the finite-size electrostatic corrections, the charge transition levels of some defects are found to shift significantly compared with the uncorrected values, with energy shifts ranging from approximately 0.24 to 0.44 eV. These results suggest that long-range Coulomb interactions induced by periodic boundary conditions can play an important role in charged defect calculations at the interface and should be carefully considered. Charge transition levels After applying the electrostatic energy corrections, the charge transition levels (CTLs) of ten distinct defect sites were evaluated. For most defect configurations, only a single charge transition level falls within the silicon band gap, while the other transition level lies close to or merges into either the valence band maximum (VBM) or the conduction band minimum (CBM). This behavior indicates that hydrogen-related defects at the interface predominantly act as single effective trap centers. As shown in Fig. 6, at defect site 24, both \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon \left(+/0\right)$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon \left(0/-\right)$$\end{document} are located within the silicon band gap, at 0.28 eV and 0.77 eV, respectively, representing a rare case where two charge transition levels coexist inside the gap. In contrast, the remaining defect sites exhibit more asymmetric level distributions. Specifically, for sites 11, 19, 31, and 68, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon \left(+/0\right)$$\end{document} lies within the band gap, whereas \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon \left(0/-\right)$$\end{document} is positioned above the CBM. For sites 5, 32, 35, 44, and 62, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon \left(0/-\right)$$\end{document} falls inside the gap, while \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon \left(+/0\right)$$\end{document} is located below the VBM. At site 22, neither \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon \left(+/0\right)$$\end{document} nor \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon \left(0/-\right)$$\end{document} enters the silicon band gap; instead, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon \left(+/0\right)$$\end{document} lies below the VBM and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon \left(0/-\right)$$\end{document} above the CBM.Fig. 6Charge transition levels for ten distinct defect sites. The red horizontal bars denote the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon \left(+/0\right)$$\end{document} transition levels, while the black horizontal bars represent the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon \left(0/-\right)$$\end{document} transition levels. The valence band maximum (VBM) and band gap are referenced to bulk silicon36,41. Charge transition levels for ten distinct defect sites. The red horizontal bars denote the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon \left(+/0\right)$$\end{document} transition levels, while the black horizontal bars represent the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon \left(0/-\right)$$\end{document} transition levels. The valence band maximum (VBM) and band gap are referenced to bulk silicon36,41. Although the absolute energies of the charge transition levels show noticeable site-to-site variations, their overall distribution pattern remains consistent across the interfacial suboxide region. This observation suggests that local structural disorder and stress variations primarily modulate the precise energetic positions of the CTLs, without fundamentally altering their alignment relative to the silicon band edges. In bulk SiO2, Si–H–Si defects typically exhibit \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon \left(+/0\right)$$\end{document} within the silicon band gap and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon \left(0/-\right)$$\end{document} near the valence band edge46, while in bulk a-SiO2 the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon \left(0/-\right)$$\end{document} levels are commonly found close to the conduction band edge45. In contrast, within the suboxide layer of the a-SiO2/Si interface, the combined effects of local stress accumulation and atomic-scale structural disorder may contribute to a more pronounced site dependence of the CTLs. Nevertheless, for most of the sampled interfacial defect sites, only a single transition level is found to be introduced into the silicon band gap. Overall, under different charge states, hydrogen-related defects tend to undergo structural reconstruction that shifts one of the charge transition levels toward the silicon band edges, thereby reducing the total energy of the system. This characteristic implies that such defects are more likely to exhibit pronounced carrier trapping activity for a specific carrier type (either electrons or holes), rather than functioning as bipolar deep-level traps affecting both carriers simultaneously. The calculation of formation energy Charged point defects play an important role in determining the physical properties of materials, particularly in semiconductor devices, where they can strongly modify the electronic structure, carrier concentrations, and chemical stability. The computational formalism used to evaluate defect formation energies is described in Section “The calculation of formation energy” . For defects in the neutral charge state, no net charge is introduced into the system, and consequently no spurious long-range Coulomb interactions arise from the use of periodic boundary conditions. Therefore, electrostatic corrections are not required when calculating the formation energies of neutral defects. The formation energy of a neutral defect is given by,12\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}^{f}\left({X}^{0}\right)={E}^{tot}\left({X}^{0}\right)-{E}_{tot}\left(pristine\right)-\sum_{i}{n}_{i}{\mu }_{i}$$\end{document} Where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}^{f}\left({X}^{0}\right)$$\end{document} denotes the formation energy of the defect in the neutral charge state, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}^{tot}\left({X}^{0}\right)$$\end{document} is the total energy of the supercell containing the neutral defect. Figure 7(a) presents the variation of defect formation energies as a function of the Fermi level for different defect sites. Owing to the amorphous nature of the suboxide layer at the a-SiO2/Si interface, as well as the site-dependent local stress environment, the calculated formation energies exhibit pronounced site-to-site variations. For defect sites 11, 19, 31, and 68, the defects are thermodynamically stable in the positive charge state within the silicon band gap. As the Fermi level increases, the stable charge state transitions from positive to neutral. In contrast, for sites 5, 32, 35, 44, and 62, the defects are initially stable in the neutral charge state and subsequently transform into the negative charge state as the Fermi level rises. At site 24, the defect undergoes successive charge-state transitions from positive to neutral and finally to negative, exhibiting a complete sequence of charge-state evolution. For site 22, however, the defect remains thermodynamically stable in the neutral charge state throughout the entire silicon band gap.Fig. 7(a) Formation energies of defects at ten different sites as a function of the Fermi level. (b) Formation energies of defects at different sites for different charge states, calculated at \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\epsilon }_{F}$$\end{document} = VBM. Red, blue, and black symbols denote the formation energies of positively charged, neutral, and negatively charged defects, respectively. The horizontal lines in the corresponding colors indicate the average formation energy for each charge state. (a) Formation energies of defects at ten different sites as a function of the Fermi level. (b) Formation energies of defects at different sites for different charge states, calculated at \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\epsilon }_{F}$$\end{document} = VBM. Red, blue, and black symbols denote the formation energies of positively charged, neutral, and negatively charged defects, respectively. The horizontal lines in the corresponding colors indicate the average formation energy for each charge state. As shown in Fig. 7(b), the formation energies of the hydrogen-related defects at different sites, calculated at \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\epsilon }_{F}$$\end{document} = VBM, are found to be mainly distributed in the range of approximately 1.71~5.23 eV. Overall, the formation energies of neutral and positively charged defects are systematically lower than those of negatively charged defects. Specifically, the average formation energies are 2.45 eV for the positively charged state, 2.52 eV for the neutral state, and significantly higher at 3.83 eV for the negatively charged state. or a given charge state, noticeable variations in formation energy are observed among different defect sites. By correlating these results with the spatial distribution of defect sites within the interfacial suboxide layer shown in Fig. 2(c), a positional dependence can be identified: defects located closer to the a-SiO2 side exhibit higher formation energies, whereas those situated nearer to the Si substrate possess lower formation energies and are may therefore be more thermodynamically stable. This trend suggests that the thermodynamic stability of hydrogen-related defects in the interfacial suboxide layer may be influenced by their spatial location across the a-SiO2/Si interface. Conclusion In this work, first-principles calculations were employed to systematically investigate the stable configurations, formation energies, and charge transition levels of hydrogen-related defects in the suboxide region of the amorphous SiO2/Si(100) interface. We have summarized a computational workflow for formation energy correction in inhomogeneous dielectric systems, specifically for the amorphous a-SiO2/Si interface. To eliminate the spurious long-range Coulomb interactions arising from periodic boundary conditions in charged defect calculations, a quantitative electrostatic correction scheme was implemented by combining density functional theory with a classical finite-element solution of the Poisson equation. The resulting electrostatic correction energies range from approximately 0.24 to 0.44 eV, yielding physically meaningful and quantitatively reliable charge transition levels. Structural analyses suggest pronounced charge-state-dependent evolution of hydrogen-related defects at the interface. Under positively charged conditions, hydrogen is found to preferentially form symmetric Si–H–Si bridge configurations among the sampled defect sites. In the neutral charge state, Si-H···Si configurations are found to energetically favored and may introduce localized defect states within the silicon band gap. In the negatively charged state, in addition to Si-H-Si and Si-H···Si structures, fivefold-coordinated silicon defect configurations can also form. After electrostatic corrections, the calculated charge transition levels suggest that, for most of the sampled interfacial hydrogen-related defects, only a single transition level appears within the silicon band gap, while the other transition level is located near the valence-band maximum or conduction-band minimum. This behavior suggests that such defects may predominantly act as single effective carrier traps for either electrons or holes, rather than as multi-level deep traps active for both carrier types. Formation energy calculations further indicate significant site-dependent variations in defect stability. Notably, the formation energies exhibit a correlation with the spatial location of the defect within the interfacial suboxide layer: defects located closer to the amorphous SiO2 side tend to less thermodynamically stable, whereas those nearer to the silicon substrate exhibit lower formation energies. This trend suggests that the thermodynamic stability of hydrogen-related defects may be influenced by local structural disorder and stress distribution. Overall, by elucidating the interplay between atomic structure, charge-state evolution, and defect energetics within the sampled configurations, this work provides insight into hydrogen-related defects at amorphous SiO2/Si interfaces. It should be noted that the present study is based on a limited number of representative defect configurations, which may impose certain limitations on the statistical representativeness of the results. Therefore, the conclusions drawn in this work should be interpreted with appropriate caution and are intended to reflect indicative trends rather than universally applicable rules. Future investigations involving a broader sampling of defect configurations are required to enable more comprehensive statistical analyses and to establish more generally applicable trends. Nevertheless, the present results offer useful theoretical insight into the role of hydrogen in silicon-based microelectronic devices and its potential impact on interfacial electronic properties and long-term device reliability. Supplementary Information Supplementary Information. Supplementary Information.