(image) (image) [Previous] [Next]

Degradation of Electrical Parameters of Power Semiconductor Devices – Process Influences and Modeling

2.2 Oxide thickness dependence

First, the dependence of NBTI and positive BTI (PBTI) of pMOSFETs on the gate oxide thickness is investigated. The threshold voltage drifts can be directly compared considering the thickness dependence of the drift magnitude. The shift of the threshold voltage (math image) depends inversely on the gate oxide capacitance. Furthermore, because of \( \gls {Cox}=\gls {epsSiO2}/\gls {dox} \), (math image) scales linearly with the oxide thickness. With consideration of (1.1) the normalized threshold voltage shift [Rei+08; Pob+10; Pob+11a],

(2.1) \begin{equation} \gls {dVthnorm} = \frac {\gls {dVth}}{\gls {dox}}, \end{equation}

accounts properly for the oxide thickness, assuming [Rei+08; Pob+10] that all charges created through BTS reside close to the active device Si-SiO2 interface. The recovery voltage is the threshold voltage of the particular device. This ensures an equivalent hole density at the interface and thus equivalent recovery conditions for all investigated oxide thicknesses.

For a comparison of CP measurements of devices with different oxide thicknesses it is important to ensure that the high and low levels of the gate pulse are sufficiently large such that always the maximum CP current is measured. Constant rising and falling slopes provide an equivalent probed energy interval for all devices [Gro+84; ANG08] as already described in Section 1.2.2.

Taking these considerations into account, the impact of N- and PBTI on n \( ^{++} \) and p \( ^{++} \) poly gate pMOSFETs can be compared as shown in Fig. 2.2 and Fig. 2.3, respectively.


Fig. 2.2: SiO2 equivalent oxide thickness, EOT = εSiO2 /Cox (EOT) dependence of the degradation following N- (blue closed symbols) or PBTS (red open symbols) for pMOSFETs with an n \( ^{++} \) doped poly gate [Pob+11a]. Minimum delay charge pumping measurements (upper plot) and threshold voltage shifts (lower plot) measured at the same chuck temperature of 50 °C.


Fig. 2.3: EOT dependence of the degradation following N- or PBTS for pMOSFET with a p \( ^{++} \) doped poly gate [Pob+11a]. Charge pumping measurements (upper plot) and threshold voltage shifts (lower plot) measured at different chuck temperatures.

Both N- and PBTI are independent of the EOT for the n \( ^{++} \) poly gated pMOSFETs in Fig. 2.2. This is consistent with earlier work [Pob+10; Rei+08; Pob+11a]. These results indicate that the degradation is not due to impact ionization [DCA93] or anode hole injection [SH94; DiM00; Hua+03], since those mechanisms would strongly depend on the voltage across the oxide and not on the electric oxide field. Although PBTS causes a large change in the CP current, it results in a vanishingly small (math image) shift measured by the change in the drain current.

A similar result is found for the p \( ^{++} \) poly gate pMOSFETs following NBTS, for which no consistent dependence on the EOT is observed in Fig. 2.3. In contrast, following PBTS, the degradation decreases strongly with increasing EOT. Remarkably, the drift polarity changes its sign between 6 nm and 26 nm. The thick oxide device experiences rather a positive drift after PBTS indicating the creation of negative charges [Pob+11a]. From this observation a different degradation mechanism is suspected to occur for PBTS on p \( ^{++} \)/pMOSFETs with thin gate oxides.

In order to better understand the different behavior for the PBTI on p \( ^{++} \)/pMOSFETs the dependence on the stress oxide field is investigated.