2.4 Charge Pumping

For the assessment of interface traps, which are situated at the interface between the oxide and the substrate along the channel of a MOSFET, the charge pumping method (CP) is often the measurement method of choice. The setup of CP after [4344] is described on basis of Fig. 2.10 (left). Source and drain of the transistor are shortened and biased at a certain reverse voltage with respect to the substrate. To perform a CP measurement the gate is stepped between accumulation and inversion by a pulse generator. Pulsing towards inversion deeply depletes the surface and minority carriers (holes in the case of the depicted pMOS) are injected from the source and drain regions into the channel where they can be captured by the interface states. When going back from inversion towards accumulation, the mobile minority carriers drift back to source and drain due to the reverse bias. The charges trapped at the interface are too slow to follow, but will recombine with the majority carriers of the substrate (electrons). The resulting recombination current, the charge pumping current Icp   , is measured at the bulk. Icp   is proportional to the number of interface traps, i.e. CP directly measures the amount of interface states.

By additionally varying parameters like amplitude, frequency, rise and fall time of the pulses and the temperature of the device under test, not only the number of interface traps, but also their energetic and spatial distributions can be determined, which gives very useful information of NBTI related degradation [4546474849].

Unfortunately, the characterization by CP measurements following NBTI stress, especially the characterization of the fast recovery behavior, is extremely challenging because the CP technique inherently relies on a bias switch into accumulation. Consequently, it is unclear whether the often observed weak recovery in CP data is a consequence of the fact that interface states do not recover or whether this is an artifact of the measurement technique brought about by the strong bias switch. Moreover, the necessary averaging of many pulses implies a rather large measurement delay, which is also not favorable when trying to access early recovery.


PIC


Figure 2.10: Left: Setup of the charge pumping measurement technique to indirectly determine the amount of interface states. A gate pulse sweep is applied ranging from accumulation to inversion. Thereby, the minority carriers that got trapped by the interface states during inversion recombine with the majority carriers during accumulation. This recombination current is then proportional to the interface state amount. Right: Typical NBTI stress and relaxation measurement of the charge pumping current Icp   using the OFIT technique with a duty cycle of 50%  . Each symbol in the upper graph consists of some hundred averaged pulses as schematically displayed in the lower graph for some points. After the reference measurement, where VG,low = Vrel   is pulsed across Vacc   and back periodically, VG,low   is set to the stress level Vstr   , in order to continuously stress the device. In this way the CP measurement and the application of stress are carried out consecutively with only interrupting the stress by the very short gate pulse. During relaxation VG,low = Vrel   . Constant slopes of the stress and relaxation pulses have to be ensured [50].