Electrically stimulated optical spectroscopy of interface defects in wide-bandgap field-effect transistors
At a Glance
Section titled âAt a Glanceâ| Metadata | Details |
|---|---|
| Publication Date | 2023-01-31 |
| Journal | Communications Engineering |
| Authors | Maximilian W. Feil, H. Reisinger, André Kabakow, Thomas Aichinger, Christian Schleich |
| Institutions | Infineon Technologies (Austria), Siemens (Austria) |
| Citations | 12 |
| Analysis | Full AI Review Included |
Executive Summary
Section titled âExecutive SummaryâThis research introduces a powerful new techniqueâElectrically Stimulated Optical Spectroscopy (ESOS)âto characterize critical interface defects in wide-bandgap (WBG) power transistors.
- Core Value Proposition: ESOS enables the study of charge trapping mechanisms (responsible for Bias Temperature Instability, BTI, and hysteresis) in fully processed 4H-SiC MOSFETs by analyzing the light emitted during gate switching.
- Novel Methodology: The technique utilizes reverse-side detection, where the SiC substrate is etched back, providing undisturbed optical access to the SiC/SiO2 interface, unlike previous front-side methods.
- Critical Correlation: A strong, near-unity linear correlation (Pearson coefficient 0.96) was established between the measured photo charge shift (ÎQ) and the electrical threshold voltage shift (ÎVth), confirming that the radiative recombination pathway is directly linked to the performance-limiting trapped charges.
- Defect Identification: The emission spectrum (1.4 eV to 3 eV) is decomposed into ten discrete transitions, allowing the energetic position of charge transition levels (CTLs) to be determined and assigned to specific interface defects (e.g., carbon-related and nitrogen-related defects) based on Density Functional Theory (DFT) calculations.
- Mechanism Insight: The method temporally separates recombination events, showing that photon emission occurs rapidly during the gate voltage transient (rising and falling edges), providing insight into the radiative relaxation pathway that occurs alongside the dominant non-radiative multiphonon (NMP) processes.
- Engineering Impact: This optical approach provides direct access to quantum observables and structural information on interface defects, offering a unique validation tool for theoretical models and guiding process improvements to push device performance closer to the theoretical limit.
Technical Specifications
Section titled âTechnical Specificationsâ| Parameter | Value | Unit | Context |
|---|---|---|---|
| Semiconductor Material | 4H-SiC | - | Wide-bandgap power MOSFET used for testing. |
| SiC Bandgap Energy | 3.26 | eV | Energy of band-to-band recombination (not observed in the spectrum). |
| SiC/SiO2 Interface Defect Density | Up to 4 x 1012 | cm-2 eV-1 | High defect density characteristic of 4H-SiC interfaces. |
| Gate Switching Voltage Range | -10 to 10 | V | Standard continuous switching range for emission microscopy. |
| Emission Spectrum Range | 1.4 to 3 | eV | Broad band emission observed, covering the visible spectral range. |
| SiPM Internal Gain | ~5 x 106 | - | Internal gain of the Silicon Photomultiplier detector. |
| SiPM Photodetection Efficiency | ~40 | % | High efficiency due to the SiC emission being in the visible range. |
| Etched SiC Substrate Thickness | ~185 | ”m | Thickness remaining after reverse-side etching for optical access. |
| Correlation Coefficient (ÎQ vs. ÎVth) | 0.96 | - | Strong linear correlation between photo charge shift and threshold voltage shift. |
| BTI Power Law Exponent (n) | 0.062 | - | Observed dependence of photo charge shift (ÎQ) on positive gate stress time (tstress). |
| Estimated Total Photons Emitted | 2 x 104 | Photons | Estimated total photons emitted per single gate voltage transition. |
| Radiative Fraction of Trapped Charge | 3 x 10-6 | - | Fraction of trapped charges that discharge via the radiative pathway. |
| Threshold Voltage (Vth) | 1.973 | V | Real threshold voltage derived from fitting parameters. |
Key Methodologies
Section titled âKey MethodologiesâThe experiment employed a custom setup combining electrical stress with highly sensitive optical detection on commercially available 4H-SiC DMOSFETs.
-
Reverse-Side Sample Preparation:
- Commercial, packaged 4H-SiC DMOSFETs were chemically processed to remove the copper lead frame and solder.
- The reverse-side metallization was polished off, leaving the SiC epitaxial layer and substrate (~185 ”m thick) exposed for optical access. This ensures light emitted from the SiC/SiO2 interface is not absorbed by intermediate layers (like polysilicon or metal contacts).
-
Emission Microscopy (Localization):
- A nearly square gate waveform (e.g., -10 V to 10 V) was applied continuously, varying frequency from 50 kHz to 2 MHz.
- An emission microscope visualized the photon emission, confirming it was localized to the channel region underneath the gate insulator.
-
Time-Resolved Single-Transient Detection:
- A Silicon Photomultiplier (SiPM) was used to detect single photons with high sensitivity and internal gain (~5 x 106).
- This allowed the temporal separation of radiative events occurring during the rising edge (inversion to accumulation) and the falling edge (accumulation to inversion) of the gate signal.
-
Bias Dependence and Correlation:
- Short double-pulse gate voltage schemes were used to sweep the high (VH) or low (VL) voltage level while keeping the other constant.
- The integrated photo charge (Qphoto) from the SiPM peak was measured and correlated with the threshold voltage shift (ÎVth) measured using a 1 ”s delay time, establishing the linear relationship.
-
Emission Spectroscopy and Defect Assignment:
- A fiber-coupled CCD spectrometer was used to acquire the emission spectrum (1.4 eV to 3 eV) under continuous 1 MHz switching (integrated over 40 s).
- The broad spectrum was mathematically decomposed into ten distinct Gaussian peaks.
- The energy positions of these peaks were compared against theoretical charge transition levels (CTLs) derived from Density Functional Theory (DFT) simulations to assign the radiative transitions to specific interface defects (e.g., (C-C)â SiC, N-related defects).
Commercial Applications
Section titled âCommercial ApplicationsâThis advanced characterization technique directly addresses reliability and performance limitations in critical semiconductor technologies.
- High Power Conversion Systems:
- Electric Vehicles (EVs): Improving the reliability and efficiency of SiC MOSFETs used in motor drives and on-board chargers.
- Renewable Energy Inverters: Enhancing the lifetime and stability of power devices in solar and wind energy conversion systems.
- Semiconductor Manufacturing and Process Control:
- Interface Engineering: Providing quantitative feedback on the effectiveness of SiC oxidation and passivation processes (e.g., NO passivation) by identifying the specific defects created or mitigated.
- Reliability Modeling: Offering a unique method to validate and refine physical models of BTI and hysteresis, leading to more accurate lifetime predictions for WBG devices.
- Advanced Device Characterization:
- In-Situ Defect Analysis: Enabling the characterization of electrically active defects in fully processed, commercial devices, which is impossible with traditional electrical or structural methods.
- Quantum Observables: Providing direct optical access to the quantum mechanical nature of defects (charge transition levels), complementing or exceeding the capabilities of Electron Spin Resonance (ESR) or Electrically Detected Magnetic Resonance (EDMR).
View Original Abstract
Abstract Wide-bandgap semiconductors such as silicon carbide, gallium nitride, and diamond are inherently suitable for high power electronics for example in renewable energy applications and electric vehicles. Despite the high interest, the theoretical limit regarding device performance has not yet been reached for these materials. This is often due to charge trapping in defects at the semiconductor-insulator interface. Here we report a one-to-one correlation between electrically stimulated photon emission and the threshold voltage shift obtained from a fully processed commercial 4H-SiC metal-oxide-semiconductor field-effect power transistor. Based on this observation, we demonstrate that the emission spectrum contains valuable information on the energetic position of the charge transition levels of the responsible interface defects. We etch back the transistor from the reverse side in order to obtain optical access to the interface and record the emitted light. Our method opens up point defect characterization in fully processed transistors after device passivation and processing. This will lead to better understanding and improved processes and techniques, which will ultimately push the performance of these devices closer to the theoretical limit.