Open Defects in Nanometer Technologies Models, Test and Diagnosis - Models in Hardware Testing - page 11

Fig. 1.11 Experimental delay results for resistive opens with quiescent neighbors ( Arumı

et al. 2008a )

parasitic capacitances affecting the defective line are related to neighboring lines,
which may change their state when a new test pattern is applied. Thus, the effective
coupled capacitances depend on the state of the neighboring lines. The experimental
measurements in Fig. 1.12 show this phenomenon. A rising transition was transmit-
ted through the defective line for every configuration of the two neighbors (N1 and
N2) coupled to the defective line. As in the results of Fig. 1.11 , the gate voltages
of the transmission gates are controlled to obtain different resistance values. As ex-
pected, the delay is higher if the neighboring lines undergo the opposite transition
related to the defective line. However, if both neighbors undergo the same transi-
tion as the defective one, the delay variability in the defect resistance is noticeably
lower, since they help the defective line to reach the final (expected) state. When
the neighboring lines have transitions of different sign, an intermediate behavior is
observed.
The open resistance value has an important influence on the timing behavior of
the defective circuit. Thus, when the resistance of the open is significantly higher
than the on-resistance of the driving gate, i.e. R o >> R ON , the delay can be simplified
as follows:

ı R o .1 ˛/ C (1.7)

The delay increases as the open is located close to the beginning of the line (low val-
ues of '). However, this simplification is not accurate for low resistive open defects.
In these situations, a second order model must be considered. The maximum delay is
not always found at the beginning of the net, but at an intermediate location, which
is determined by the relationship between the open resistance, the on-resistance