^{1}, P. M. Lenahan

^{1,a)}and A. J. Lelis

^{2}

### Abstract

We have identified a magnetic resonance spectrum associated with minority carrier lifetime killing defects in device quality SiC through magnetic resonance measurements in bipolar junction transistors using spin dependent recombination (SDR). The SDR spectrum has nine distinguishable lines; it is, within experimental error, essentially isotropic with four distinguishable pairs of side peaks symmetric about the strong center line. The line shape is, within experimental error, independent of bias voltage and recombination current. The large amplitude and spacing of the inner pair of side peaks and three more widely separated pairs of side peaks are not consistent with either a simple silicon or carbonvacancy or a carbon or silicon antisite. This indicates that the lifetime killing defect is not a simple defect but a defect aggregate. The spectrum is consistent with a multidefect cluster with an electron spin . (The observed spectrum has not been reported previously in the magnetic resonance literature on SiC.) A fairly strong argument can be made in terms of a first order model linking the SDR spectrum to a divacancy or possibly a vacancy/antisite pair. The SDR amplitude versus gate voltage is semiquantitatively consistent with a very simple model in which the defect is uniformly distributed within the depletion region of the base/collector junction and is also the dominating recombination center. The large relative amplitude of the SDR response is more nearly consistent with a Kaplan–Solomon–Mott-like model for spin dependent recombination than the Lepine model.

The work at Pennsylvania State University was supported by the U.S. Army Research Laboratory, Adelphi, MD.

I. INTRODUCTION

II. EXPERIMENTAL DETAILS

III. RESULTS

A. The spectrum

B. The SDR response to junction bias

C. Interpretation of SDR results: Provisional identification of the defect

D. Conventional ESR studies of divacancies and (carbon)vacancy antisites

E. The relevance of the SDR results to DLTS studies

IV. CONCLUSIONS

### Key Topics

- Silicon
- 30.0
- Vacancies
- 22.0
- Carbon
- 17.0
- Electron paramagnetic resonance spectroscopy
- 16.0
- Magnetic resonance
- 15.0

## Figures

SDR spectrum of a SiC BJT with magnetic field oriented perpendicular to the crystalline -axis.

SDR spectrum of a SiC BJT with magnetic field oriented perpendicular to the crystalline -axis.

SDR spectrum of a SiC BJT with magnetic field oriented parallel to the crystalline -axis.

SDR spectrum of a SiC BJT with magnetic field oriented parallel to the crystalline -axis.

This figure illustrates a map of values plotted as a function of the angle between the magnetic field and directions in the SiC. The transistor was rotated about three perpendicular axes. The three axes correspond approximately to the , the , and the [0001] directions. Each of the three axes corresponds to one of the maps as indicated in the figure. In the top map, labeled , we plot as a function of the angle between the [0001] direction and the magnetic field. In the middle map, labeled , we plot as a function of the angle between the [0001] direction and the magnetic field. In the bottom map, labeled [0001], we plot as a function of the angle between the direction and the magnetic field. The important conclusion to draw from this figure is that, within our experimental error, ±0.0003, the is independent of magnetic field orientation. It should be emphasized that some weak anisotropy may be present; if so, however, it is below the precision of our measurements.

This figure illustrates a map of values plotted as a function of the angle between the magnetic field and directions in the SiC. The transistor was rotated about three perpendicular axes. The three axes correspond approximately to the , the , and the [0001] directions. Each of the three axes corresponds to one of the maps as indicated in the figure. In the top map, labeled , we plot as a function of the angle between the [0001] direction and the magnetic field. In the middle map, labeled , we plot as a function of the angle between the [0001] direction and the magnetic field. In the bottom map, labeled [0001], we plot as a function of the angle between the direction and the magnetic field. The important conclusion to draw from this figure is that, within our experimental error, ±0.0003, the is independent of magnetic field orientation. It should be emphasized that some weak anisotropy may be present; if so, however, it is below the precision of our measurements.

Derivative of the SDR spectrum shown in Fig. 1.

Derivative of the SDR spectrum shown in Fig. 1.

Top trace is the base-collector junction current vs forward bias voltage. The lower trace is SDR amplitude vs base-collector junction forward bias voltage.

Top trace is the base-collector junction current vs forward bias voltage. The lower trace is SDR amplitude vs base-collector junction forward bias voltage.

Calculated recombination current (top) vs SDR (bottom) as functions of junction bias.

Calculated recombination current (top) vs SDR (bottom) as functions of junction bias.

Ball and stick models of the (a) divacancy and (b) carbon vacancy/antisite defects.

Ball and stick models of the (a) divacancy and (b) carbon vacancy/antisite defects.

The dotted line illustrates the integral of SDR spectra observed in Fig. 1. The solid line illustrates the expected integrated intensity with hyperfine parameter assumptions: a hyperfine coupling constant of 11 G for the nine nearest neighbor silicons, a coupling of 43 G for the first neighbor silicons, and 34 G for the second neighbor carbons. The numbers in the figure (0.071, 0.064, 0.031,…, correspond to the relative contribution of each Gaussian to the total spectrum.

The dotted line illustrates the integral of SDR spectra observed in Fig. 1. The solid line illustrates the expected integrated intensity with hyperfine parameter assumptions: a hyperfine coupling constant of 11 G for the nine nearest neighbor silicons, a coupling of 43 G for the first neighbor silicons, and 34 G for the second neighbor carbons. The numbers in the figure (0.071, 0.064, 0.031,…, correspond to the relative contribution of each Gaussian to the total spectrum.

Comparison of (a) the SDR spectrum obtained for the SiC BJT and (b) and the expected spectrum with the same isotropic hyperfine coupling constants utilized in Fig. 8. (Trace b in this figure is the derivative of the solid line corresponding to the trace in Fig. 8.)

Comparison of (a) the SDR spectrum obtained for the SiC BJT and (b) and the expected spectrum with the same isotropic hyperfine coupling constants utilized in Fig. 8. (Trace b in this figure is the derivative of the solid line corresponding to the trace in Fig. 8.)

Comparison of the second derivative of (a) the SDR spectrum and (b) the second derivative SDR spectrum based on the same parameters utilized in Figs. 8 and 9. (Trace b in this figure is the derivative of the model calculation trace of 9b.)

Comparison of the second derivative of (a) the SDR spectrum and (b) the second derivative SDR spectrum based on the same parameters utilized in Figs. 8 and 9. (Trace b in this figure is the derivative of the model calculation trace of 9b.)

Article metrics loading...

Full text loading...

Commenting has been disabled for this content