The simulation procedure is analogous to the experimental procedure. Transient
pulses are applied to the device terminals. After the simulation, lasting a few
signal periods, the time averages of the interface effective generation of
electrons and holes
,
and of the electron and hole components in the terminal currents
,
,
,
,
and
are
calculated. The averaging is performed in the period for which the
periodic-steady-state condition is fulfilled:
where is the charge-pumping current. Note that we simulate an aperiodic
signal. Therefore, the steady-state solution at the chosen
starting bias, which is set as an initial condition for the transient
simulation, is not equal to the periodic-steady-state condition at this bias
point. For the large-signal techniques the simulation of
to
periods
is sufficient to perform a proper averaging in time. However, in some problems
the simulation of several signal periods is necessary to obtain the
periodic-steady-state solution, as e.g. for the front-back interface
coupling in SOI devices [357], the charge pumping in LDD
devices [168] (gate-corner/LDD-region electric-field fringing analyzed
in Section 3.5.3 and Appendix E) and for the
small-rectangular-pulse technique [495][494]. The averaging in time
must be consistent with the discretization of the continuity
equations 3.1 and 3.2, because of the
time-discretization error
. Comparing the calculated DC electron and hole components in the terminal
currents and the interface net generation rates with each other, the deviations
from the `ideal charge-pumping theory' can be extracted for some particular
problem. For example, the differences between the total
and
indicate the presence of the geometric
current component (Section 3.4). The transient currents
are of order
, while the charge-pumping currents
are of
order
. It is evident that very accurate terminal current calculation
becomes indispensible for the presented approach. The method of weighting
functions is used to accurately calculate the terminal currents in transient
cases [324]. Because of the error in the current calculation, the
gate current ranges
after an averaging in time
(in the periodic-steady-state condition). Details on the current calculation
are given in [191].