Can the reaction-diffusion model explain generation and recovery of interface states contributing to NBTI?

In this paper p-MOSFETs with 2.8 nm SiO₂ gate dielectrics were tested. This study shows experimental observations which contradict the RD model: (i) amount of ¿I_ep recovered in the recovery interval under dynamic NBTI remains constant with stress/recovery cycling; (ii) abrupt change in ¿I_ep, observed typically at the start of a stress or recovery cycle, is an artifact caused by interfacial oxide traps participating in the charge pumping (CP) process; (iii) stress induced ¿N_it is relatively permanent (recovery is negligible over extended period); (iv) the exponent of -0.16 for ¿I_ep is an artifact due to "charge pumping of interfacial oxide traps" (after eliminating this parasitic oxide trap component, the exponent for ANn is -0.27); (v) a consistent exponent of-0.27 for ¿N_it, is observed over a wide temperature range (this does not corroborate a dispersive transport mechanism).