« PreviousUpNext »Contents
Previous: 10.2 The Bootstrapping and Cumulative Sum Algorithm    Top: 10 Advanced Step Detection Algorithm    Next: III Results

10.3 Evaluation of the Algorithm

An example of common recovery traces recorded on a \( W\times L=\SI {160}{\nano \meter }\times \SI {120}{\nano \meter } \) pMOSFET is shown in Figure 10.3.

(image)

Figure 10.3:  An example recovery trace (top) showing a single detrapping event, in this case for a defect named #1, which is clearly visible and not contaminated by RTN. However, new measurements recorded several weeks after the first dataset are strongly influenced by a new defect producing RTN (middle). The initially used naive algorithm just detects three positive changes because only positive steps are considered (middle, Simple). The BCSUM method additionally detects negative steps stemming from RTN (middle, New Full). In the spectral map (bottom), here depicted for 100 traces, the steps detected with the simple method (green) show step heights below the step level for the defect #1 which equals the step heights produced by an RTN defect. Furthermore, the large step is extracted with a wrong amplitude and so it does not contribute to the defect #1 in the spectral map. The BCSUM algorithm extracts the amplitudes corresponding to the defect #1 correctly [MWC26].

As long as the DUT contains only single emission events, which is the case for defects with \( \taue \gg \tauc \), the TDDS data can be analyzed in a straight-forward manner. However, this is not the case for signals containing single defects producing RTN. As can be seen from Figure 10.3 (middle) the initially used step detection algorithm considers steps in one direction leading to a falsely extracted step height. The BCSUM algorithm detects all steps correctly by using just two parameters.

« PreviousUpNext »Contents
Previous: 10.2 The Bootstrapping and Cumulative Sum Algorithm    Top: 10 Advanced Step Detection Algorithm    Next: III Results