edfas.org 11 ELECTRONIC DEVICE FAILURE ANALYSIS | VOLUME 27 NO. 1 the S0 versus τ0 plot may additionally be used to confirm that it is about the same trap over the entire investigated temperature range. Examples of S0 versus τ0 are plotted in Figs. 4b and 5b. In Fig. 6 Sv,GR(f0,T)f0 versus the temperature is given for the trap identified in Fig. 5, permitting an estimate of the trap volume density. IDENTIFIED TRAPS IN ACTIVE DEVICES Table 1 summarizes the main identified traps in different 16 nm and below state-of-the-art MOSFET devices in the temperature range of 80 K to 340 K. The main differences between UTBOX technologies are related to the gate stack and the resulting equivalent oxide thickness (EOT), 2.6 nm, 2.1 nm, and 1.9 nm, respectively.[16-18] The standard UTBOX have the channel orientation <100> while for the rotated UTBOX the channel orientation is <110>. In reference 19 the resulting EOT is also 2.6 nm, but the gate stack consists of 2.5 nm SiON on top of 1 nm interfacial SiO2 instead of 2.5 nm HfSiO with 60% Hf +1.5 nm SiO2, as recorded in reference 16. The nature of the traps is generally established by using results from the standard DLTS analysis.[25-28] The estimated activation energy and capture cross section of the T1 and T2 traps may suggest the impact of a hydrogen related center, the divacancy-hydrogen V2H trap (∆E of 0.45 eV and a sn in the 10-17 cm2 range[25,26]) for T1 and VOH trap (∆E of 0.32 eV and a sn in the 10-15 cm2 range[25,26]) for T2. The third identified trap (T3) presents an activation energy and capture cross section close to those reported in the literature for the phosphorus-vacancy complex trap V-P (∆E of 0.44 eV and a sn in the 10-14 -10-15 cm2 range[25,27]). (b) (b) (a) Fig. 4 (a) Arrhenius plot and (b) estimated GR plateau (S0) versus the relaxation time constant (τ0) corresponding to the most pronounced GR contribution of spectra illustrated in Fig. 2. A least-squares linear fit with the experimental data permits to estimate the energy difference between the conduction band energy and the trap energy (ΔE = EC – ET) (from the slope) and the capture cross section σn (from the y-intercept) (Fig. 4a); employing Eq 5, the surface trap density may be estimated from the linear dependency of the S0 with τ0 (Fig. 4b). Fig. 5 (a) Arrhenius plot and (b) estimated GR plateau (S0) versus the relaxation time constant (τ0) for the device of Fig. 2 operated at a constant applied drain current polarization of 2 µA (VDS =50 mV) in order to guarantee a constant quasi-Fermi level in a temperature range of 300 K to 330 K (a)
RkJQdWJsaXNoZXIy MTYyMzk3NQ==