The temperature dependence of is displayed in Fig. 6.11 (top) for the thinnest device () with . In the range to , the data can be perfectly fit by a logarithmic time dependence (differences would not even be visible in the plots). A slight deviation is observed for higher temperatures for , possibly due to the onset of the mechanism responsible for the long-time power-law behavior with a larger power-law exponent . This might be due to the dependence of two compound power-laws which are discussed in literature [49]. Apart from that, different temperatures can be scaled well to the data at , as shown by the dotted lines in Fig. 6.11 (top), and the indicated scaling factors marked by arrows.