^{1,a)}, P. M. Lenahan

^{1}, A. T. Krishnan

^{2}and S. Krishnan

^{2}

### Abstract

We demonstrate voltage controlled spin dependent tunneling in 1.2 nm effective oxide thickness silicon oxynitride films. Our observations introduce a simple method to link point defect structure and energy levels in a very direct way in materials of great technological importance. We obtain defect energy level resolution by exploiting the enormous difference between the capacitance of the very thin dielectric and the capacitance of the depletion layer of moderately doped silicon. The simplicity of the technique and the robust character of the response make it, at least potentially, of widespread utility in the understanding of defects important in solid state electronics. Since the specific defect observed is generated by high electric field stressing, an important device instability in present day integrated circuitry, the observations are of considerable importance for present day technology. Since the observations involve inherent high sensitivity and tunneling, and since the process can be turned on and off with the application of a narrow range of voltage, our results may also be relevant to the development of spin based quantum computing.

Work at Penn State has been supported by Texas Instruments through Semiconductor Research Corporation Custom Funding.

I. INTRODUCTION

II. EXPERIMENTAL DETAILS

III. RESULTS

IV. ANALYSIS OF THE RESULTS

V. ON THE MECHANISM OF SDT

VI. CONCLUSIONS

### Key Topics

- Tunneling
- 34.0
- Dielectrics
- 27.0
- Dielectric thin films
- 23.0
- Crystal defects
- 14.0
- Paramagnetism
- 12.0

## Figures

curves before and after high electric field stressing.

curves before and after high electric field stressing.

vs , where is the gate current density prestress and is the gate current density poststress minus . The peak in the curve is caused by a trap assisted tunneling current in the stressed measurement of Fig. 1.

vs , where is the gate current density prestress and is the gate current density poststress minus . The peak in the curve is caused by a trap assisted tunneling current in the stressed measurement of Fig. 1.

Representative SDT measurement taken with biased to correspond to the peak in the curve of Fig. 2. The measurement was taken with the magnetic field parallel to the Si/dielectric interface normal. The zero crossing .

Representative SDT measurement taken with biased to correspond to the peak in the curve of Fig. 2. The measurement was taken with the magnetic field parallel to the Si/dielectric interface normal. The zero crossing .

In this trace the sample is rotated in the magnetic field so that the Si/dielectric interface normal is perpendicular to the magnetic field. Note that the spectrum zero crossing g does not change, within experimental error, from the g with the interface normal parallel to the magnetic field as shown in Fig. 3.

In this trace the sample is rotated in the magnetic field so that the Si/dielectric interface normal is perpendicular to the magnetic field. Note that the spectrum zero crossing g does not change, within experimental error, from the g with the interface normal parallel to the magnetic field as shown in Fig. 3.

Comparison between the normalized SDT intensities as a function of (a) and the vs (b) plot of Fig. 2. The normalization of (a) is achieved by dividing the spin dependent modification to the tunneling current by the total dc current (I). The SDT response very closely follows the characteristic trap assisted tunneling peak of (b).

Comparison between the normalized SDT intensities as a function of (a) and the vs (b) plot of Fig. 2. The normalization of (a) is achieved by dividing the spin dependent modification to the tunneling current by the total dc current (I). The SDT response very closely follows the characteristic trap assisted tunneling peak of (b).

SDT spin dependent modification to the tunneling current as a function of . Note that it peaks at about indicating the peak at in the SDT is shifted downward because direct tunneling overwhelms the trap assisted tunneling process at higher voltages.

SDT spin dependent modification to the tunneling current as a function of . Note that it peaks at about indicating the peak at in the SDT is shifted downward because direct tunneling overwhelms the trap assisted tunneling process at higher voltages.

Energy band diagrams for the sample at three different values of . Note that the only plausible explanation for the tunneling current must involve electron tunneling through defects with levels corresponding to the range of the silicon band gap. The simplified sketch illustrates two dielectric defect levels, consistent with the experimental result.

Energy band diagrams for the sample at three different values of . Note that the only plausible explanation for the tunneling current must involve electron tunneling through defects with levels corresponding to the range of the silicon band gap. The simplified sketch illustrates two dielectric defect levels, consistent with the experimental result.

(a) The SDT response as a function of interface , (b) a crude schematic representation of K center DOS, and (c) a cartoon representation of the charge states of the K centers.

(a) The SDT response as a function of interface , (b) a crude schematic representation of K center DOS, and (c) a cartoon representation of the charge states of the K centers.

Schematic illustration of the DOS for an array of precisely identical defects with precisely identical energy levels.

Schematic illustration of the DOS for an array of precisely identical defects with precisely identical energy levels.

(a) A more physically reasonable DOS in which each of the levels of Fig. 9 is broadened to take into account disorder. (b) The SDT response from the levels of (a). (c) Schematic illustration of the derivative of the SDT amplitude vs energy response of (b). (d) The absolute value of the derivative (c). The plot illustrated in (d) is, as discussed in the text, an approximation of the defect DOS.

(a) A more physically reasonable DOS in which each of the levels of Fig. 9 is broadened to take into account disorder. (b) The SDT response from the levels of (a). (c) Schematic illustration of the derivative of the SDT amplitude vs energy response of (b). (d) The absolute value of the derivative (c). The plot illustrated in (d) is, as discussed in the text, an approximation of the defect DOS.

SDT signal intensity vs square root of microwave power. Note that the signal intensity does not saturate at the highest power level available in our measurements. This indicates that far higher sensitivities are possible.

SDT signal intensity vs square root of microwave power. Note that the signal intensity does not saturate at the highest power level available in our measurements. This indicates that far higher sensitivities are possible.

Article metrics loading...

Full text loading...

Commenting has been disabled for this content