During stress/measure experiments BTI was observed to recover already when the gate voltage changes from stress to recovery bias. During this switching cycle, which is approximately in our setup, the threshold voltage shift produced by single defects can not be evaluated. Nonetheless, as charge trapping is very sensitive to the device temperature, cryogenic temperatures might be used to shift the charge trapping events which occur in the switching phase into our experimental window. This would allow to study the trapping kinetics of single defects with ultra-fast emission time constants. However, it has to be noted that charge capture is also strongly temperature dependent. At cryogenic temperatures, the charge capture time increases too, and as a consequence only a reduced number of single defects might become charged. To circumvent this limitation, a sophisticated setup which allows to stress at high temperatures and to recovery a low temperatures would be required.
By analyzing the measured recovery data the charge capture and emission events of single defects can be extracted and afterwards explained using the four-state NMP model. In our SiGe investigations the simulated charge capture and emission times were calculated considering the four-state NMP model for charge transitions between the metal gate and the single defect, and charge transitions between the conducting channel and the single defect. In order to substantiate the simulated transition rates for charge trapping between the single defect and the metal-gate, single-defect SILC and conventional charge trapping visible in the drain-source current has to be monitored. Therefore the TMI has to be extended to monitor the gate and drain-source currents with very high measurement resolutions simultaneously.
In order to provide the whole spectra of well-established semiconductor characterization techniques the TMI can be extended to perform C(V) measurements and charge pumping measurements. For both methods the hardware and software has to be adjusted which, due to the modular design, can be achieved with reasonable effort. Furthermore, an extension of the sampling unit using an FPGA would (i) improve the sampling frequency and (ii) increase the available data memory. Both features would allow the TMI to properly resolve RTN signals with very short capture and emission times over a very long time. Additionally, the fast-VT method could also be integrated into the TMI.
In summary, the TMI is a powerful measurement instrument which provides all necessary features for the successful characterization of single defects. Due to its modular design, which is separated into the voltage unit, data acquisition unit and device connector unit, the TMI can be individually adjusted to the experimental requirements. Furthermore, the TMI can be used in combination with computer-controlled furnaces and modern, partially semi-automatic, probestations. The external components are thereby controlled by a jobserver, which also organizes the measurement queues. In future investigations the large list of features already provided by the developed framework is planned to be further extended by commonly used characterization techniques like charge-pumping or C(V) measurements. Nonetheless, the TMI together with the developed software toolset can be already considered as a state-of-the art semiconductor characterization framework.« PreviousUpNext »Contents