Cited by 96Silicon Carbide Semiconductor TechnologiesSemiconductor materials and devicesSemiconductor materials and interfaces

An electrically detected magnetic resonance study of performance limiting defects in SiC metal oxide semiconductor field effect transistors

Cochrane, C. J., Lenahan, P. M., Lelis, A. J.·2011·Journal of Applied Physics

Abstract

In this study, we utilize electrically detected magnetic resonance (EDMR) techniques and electrical measurements to study defects in SiC based metal oxide semiconductor field effect transistors (MOSFETs). We compare results on a series of SiC MOSFETs prepared with significantly different processing parameters. The EDMR is detected through spin dependent recombination (SDR) in most cases. However, in some devices at a fairly high negative bias, the EDMR likely also involves spin dependent trap-assisted tunneling (SDT) between defects on both sides of the SiC/SiO2 interface. At least three different defects have been detected in the magnetic resonance measurements. The defects observed include two at the SiC/SiO2 interface or on the SiC side of the SiC/SiO2 interface: one is very likely a vacancy center with a distribution which extends into the bulk of the SiC and the other is likely a “dangling bond” defect. A third defect, located on the SiO2 side of the SiC/SiO2 interface, has a spectrum very similar to that previously reported for an oxygen deficient silicon coupled to a hydrogen atom. In nearly all cases, we observe a strong dominating single line EDMR spectrum with an isotropic g≈2.0027. In some samples, this strong central line is accompanied by two pairs of considerably weaker side peaks which we link to hyperfine interactions with nearby Si and C atoms. The pattern is physically reasonable for a silicon vacancy in SiC. We therefore tentatively assign it to a silicon vacancy or silicon vacancy associated defect in the SiC. In one set of devices with very high interface trap density we observe another dominating spectrum with g∥=2.0026 and g⊥=2.0010 with the symmetry axis coincident with the [0001] and nearly the SiC/SiO2 interface normal. We ascribe this EDMR spectrum to a “dangling bond” defect. A third EDMR spectrum shows up in some devices at a fairly large negative gate bias. The phase of this spectrum is quite consistently opposite to that of the SDR detected EDMR at other biases. Part of this inverted phase spectrum involves two narrow lines which are separated by ≈10.5 G. Since the center responsible for this spectrum is almost certainly in the SiO2, it is likely due to the so called 10.4 G doublet center, an unpaired electron residing on an oxygen deficient silicon atom coupled to a hydrogen in SiO2. The likely presence of one oxygen deficient silicon defect suggests that other oxygen deficient silicon atom defect sites in the oxide may also be important in SiC/SiO2 devices. Oxygen deficient silicon defects in SiO2 are typically called E′ centers. Our results collectively demonstrate considerable complexity in both the chemical composition and physical distribution of performance limiting defects in SiC transistors, with defects observed on both sides of the SiC/SiO2 interface. Our results most strongly indicate that fairly high densities of intrinsic deep-level defects, likely due to a Si vacancy or a closely related defect, extend into the bulk of the SiC in all but one of the devices prepared utilizing a fairly wide range of processing parameters.

Integrity signals

Registry

Crossref · 10.1063/1.3530600

Cites

29 of 29 references verified0 citation errors · 0 not checkable

Statistics

Recomputing p-values needs the paper’s full text.

Claims the paper makes

This paper’s full text isn’t openly available, so its claims and self-check weren’t run. The registry and citation checks still apply.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • The interface defect density of SiC/SiO2 is two times higher than Si/SiO2 due to the presence of carbon atoms during thermal oxidation.

    “Also, the interface defect density of SiC/SiO 2 is two times higher than Si/SiO 2 due to the presence of carbon atoms during the thermal oxidation [50], [51].”

Cites

Flagged first, then by citations

Cited by

Sorted by citations

Authors
Cochrane, C. J., Lenahan, P. M., & Lelis, A. J.
Affiliations
  • Pennsylvania State University (US)
  • DEVCOM Army Research Laboratory (US)
Venue
Journal of Applied Physics
Published
2011-01-01
Volume
109
Issue
1
ISSN
  • 0021-8979 (ISSN-L)
  • 1089-7550
OpenAlex
W2027548420
Cited by
96
References
29
Citations by year
2026: 4 · 2025: 3 · 2024: 7 · 2023: 9 · 2022: 5 · 2021: 3 · 2020: 11 · 2019: 7 · 2018: 5 · 2017: 9 · 2016: 5 · 2015: 1 · 2014: 7 · 2013: 5 · 2012: 8
FWCI
6.05
Type
journal article
Language
en
Open access
closed
Locations
Journal of Applied Physics (publishedVersion)
Funders
  • Pennsylvania State University
  • Army Research Laboratory
Topics
  • Physical Sciences › Engineering › Electrical and Electronic Engineering › Silicon Carbide Semiconductor Technologies
  • Physical Sciences › Engineering › Electrical and Electronic Engineering › Semiconductor materials and devices
  • Physical Sciences › Physics and Astronomy › Atomic and Molecular Physics, and Optics › Semiconductor materials and interfaces
Keywords
SiC MOSFET, electrically detected magnetic resonance, spin-dependent recombination, SiO2/SiC interface, interface trap density, silicon vacancy, dangling bond defects, hyperfine interaction, g-factor anisotropy, hydrogen

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review