at a given operation bias value can help to evaluate the performance of a device if gain is a restricting factor at the given frequency. However, several other factors than this small-signal quantity have to be considered to successfully design a high-power structure.
The physical origin for the bias dependence can by understood looking at the bias dependence of the three constituting elements , , and . Aggressive scaling to obtain high value at low bias ( = 1V) does not necessarily increase the values for higher , e.g. = 3 V or 5 V, but can, on the opposite, decrease them. The most important parameter for this decrease is the gate length itself. Fig. 7.16 shows the measured dependence of the decrease of / in a logarithmic scale versus the gate length . The data were consistently measured for one technology, which is similar to technology C, on one wafer. If the product were a constant as a function of , a straight line would be observed. However, the decrease enhances for shorter gate length 0.3 m, which is the proof for short channel effects.
Fig. 7.17 shows the dependence of on the contact situation given in Fig. 3.25. A difference can be observed for a device which is otherwise not changed. The direct contacting leads to increased at low due to the lower contact resistance. However, for increasing bias the decrease in Case I is relatively stronger, so that for higher bias Case II appears more useful.
Fig. 7.17 is a strong hint to the importance of RST and consequently parasitic charge modulation as a function of bias. Mostly, the term modulation efficiency ME, as given in (4.3) is used to compare different materials system, Fig. 7.17 shows the importance of ME as a function of bias within the same materials system. A comparison of Technology A and Technology C, which mainly differ in the contact situation, justifies the results shown in Fig. 4.4.