The Impact of Electrostatic Interactions Between Defects on the Characteristics of Random Telegraph Noise

Local electric field perturbations due to trapping mechanisms at defects: What random telegraph noise reveals

“…33 In addition, according to the IEEE International Roadmap for Devices and Systems (IRDS), 34 these V APP match the forecast scaling V DD for the next-generation devices for logic circuits. On the contrary, as discussed in our previous work, 18 for thinner oxide layers the leakage current might be dominated by direct tunneling (DT) rather than by trap-assisted tunneling (TAT) at V + s. Therefore, RTN can result from the effect of the E-field perturbation given by trapped charge at O 0 s on DT barriers seen by e − at the injecting electrode, but the same framework and 35 V + s are positively charged when empty and neutral when they capture an e − , while O 0 s are neutral when empty and negatively charged when filled by an e − . Oxygen vacancies have been identified as the defects most involved in defect-assisted charge transport, 29,30 though oxygen ions have shown to play a crucial role in resistive switching and in determining RTN in HfO 2 .…”

“…Interestingly, even if statistically less likely, when O 0 #2 gets neutral the trend τ e1,2(empty) [Fig. 7(b)-empty rhombus] is very different from the one in which #2 is full [τ e1,2(full) -red circles], the former also exhibiting a non-monotonic behavior with a (mild) positive slope between V APP = 0.45-0.7 V and a negative slope for V APP > 0.7 V. Our previous works 18,32 revealed that non-monotonic τ c /τ e trends vs V APP can be related to changes in capture source and/or emission destination of defects, demonstrating that the slope of such time constant trends cannot be unambiguously used to provide a reliable defect classification. In fact, in Fig.…”

“…3(c), the local potential at the V + is always ≥ to the V ID , since (i) within the oxide the number of V + s ( 5) is > than the number of O 0 s (4), (ii) V + s nearby to the V + of interest are positively charged, and (iii) 2 out of 4 O 0 s (namely, #3 and #4) are always neutral and therefore they do not provide negative charge (i.e., a local potential reduction at the V + ). As we highlighted in our previous works, 18,32 the local potential within the oxide is given by the overlap of the applied voltage and the possible contribution of trapped charge. Notably, the local potential varies dynamically since (in contrast to the V APP , which is kept constant over time in this study) the trapped charge contribution is given by the specific O 0 s configuration, which changes over time and with the applied voltage.…”

“…13,14 Also, we verified that O 0 #1 changes emission destination over V APP , emitting toward other traps in the range of V APP = 0.45-0.7 V and at the TE for V APP > 0.7 V, in agreement with our recent findings. 18,32 Since the τ e1,2(empty) trend has been evaluated by using Matthiessen's rule (4), i.e., considering that the trap can emit to 5 possible destinations (namely, TE, BE, TT, VB, and CB), we evaluated the emission time constants associated with each destination (but only the one associated with the emission to the TT and TE matter), and the results are reported in Fig. 7(c) for each V APP .…”

“…In fact, the latter is given by the overlap of the applied voltage (V APP ) and the (often neglected) contribution of the charge trapped at defect sites, which is instead shown to play a dominant role, especially for low applied bias. 18…”