Electricity: measuring and testing – Fault detecting in electric circuits and of electric components – Of individual circuit component or element
Reexamination Certificate
2001-02-15
2003-02-25
Le, N. (Department: 2858)
Electricity: measuring and testing
Fault detecting in electric circuits and of electric components
Of individual circuit component or element
C324S765010, C438S151000
Reexamination Certificate
active
06525544
ABSTRACT:
TECHNICAL FIELD
The present invention relates to a method for predicting the lifetime of an insulating film for use in a semiconductor device and to a method for reliability testing of the device by utilizing the lifetime prediction method.
BACKGROUND ART
Hereinafter, a known lifetime prediction method for an insulating film will be described as being applied to the lifetime prediction of a gate insulating film for a MOSFET.
It should be noted that when a voltage or current value is preceded by a negative sign, that notation herein means that the potential level is lower at the gate electrode than at the substrate. Also, the “dielectric breakdown” herein means a steep rise of leakage current resulting from stressing, i.e., the generation of a hard breakdown (HBD), for a relatively thick insulating film, but means the generation of a soft breakdown (SBD) for a relatively thin insulating film. For a relatively thin insulating film, however, even if the SBD and HBD have occurred there at a time, it is supposed that only the SBD has occurred there.
First, an accelerating voltage (e.g., with an absolute value of 6 V), higher than an actual operating voltage normally applied (e.g., with an absolute value of 1.5 V), is applied to the gate electrode of a MOSFET under test, and a current flowing through its gate insulating film is measured. In this manner, a time it takes for the gate insulating film to cause a dielectric breakdown, i.e., the lifetime of the insulating film, is estimated (i.e., an accelerated test is performed). In this case, as the accelerating voltage is set even higher than the actual operating voltage, the insulating film lifetime, expected by the accelerated test, becomes even shorter than the lifetime of the insulating film under actual operating conditions.
Next, into a relationship between the voltage or electric field applied to the gate insulating film and the insulating film lifetime, i.e., a voltage-lifetime model (that should be made beforehand), the insulating film lifetime, expected by the accelerated test, is extrapolated. In this manner, the insulating film lifetime under actual operating conditions is calculated.
First Problem
Hereinafter, a first problem of the known lifetime prediction method for an insulating film will be described.
In the known lifetime prediction method, a time it takes for the insulating film to cause a dielectric breakdown should be actually measured at an accelerated test. Thus, it takes a long time to predict the lifetime of the insulating film. However, if the difference between the accelerating and actual operating voltages is increased to shorten the time for predicting the insulating film lifetime, then the expected lifetime of the insulating film will have its accuracy decreased because the voltage-lifetime model becomes less reliable. Nevertheless, if the accelerated test is performed a great number of times with the accelerating voltage set closer to the actual operating voltage to predict the lifetime of the insulating film more accurately, then it takes an even longer time to predict the lifetime of the insulating film.
Premise of Second Problem
As described above, in predicting the lifetime of an insulating film, an accurate voltage-lifetime model should be prepared.
Hereinafter, “1/E” and “E” models will be described as typical known voltage-lifetime models.
FIG. 12
illustrates a relationship between a stress voltage applied to the gate insulating film of an n-channel MOSFET and a total injected electron quantity Q
BD
, i.e., a total quantity of electrons injected into the gate insulating film before the dielectric breakdown occurs there.
FIG. 12
also illustrates a relationship between the stress voltage and a total injected hole quantity Q
P
, i.e., a total quantity of holes injected into the gate insulating film before the dielectric breakdown occurs there. In this case, the thickness of the gate insulating film is 5 nm and the gate length and width are both 100 &mgr;m. That is to say, the area of the gate insulating film is 0.01 mm
2
. In
FIG. 12
, the total injected electron and holes quantities Q
BD
and Q
P
are indicated by open and solid circles, respectively.
As shown in
FIG. 12
, the higher the stress voltage, the smaller the total injected electron quantity Q
BD
. On the other hand, the total injected hole quantity Q
P
is constant irrespective of the stress voltage. That the total injected hole quantity Q
P
becomes constant, i.e., anode hole injection model, was already reported by C. Hu et al. See I. C. Chen, S. E. Holland and C. Hu: IEEE Trans. Elec. Dev. 32 (1985) p. 413 and J. C. Lee, I. C. Chen and C. Hu: IEEE Trans. Elec. Dev. 35 (1988) p. 2268, for example.
It should be noted that when I say “constant” herein, this term also implies “substantially constant”.
According to the anode hole injection model, the dielectric breakdown is believed to occur in the following manner. First, electrons, which have been injected from the cathode (e.g., gate electrode) into the gate insulating film due to stressing, create holes in the anode (e.g., substrate). Then, the holes created are injected back into the gate insulating film, thereby producing defects in the gate insulating film and eventually causing the dielectric breakdown there. In this case, the total injected hole quantity Q
P
is believed to be constant until the dielectric breakdown occurs in the gate insulating film. The probability that the holes are created in the anode by the electrons and then injected into the gate insulating film is generally called a “quantum efficiency &ggr;”, which greatly depends on the stress voltage. Thus, since a relationship Q
P
=&ggr;×Q
BD
is met, the total injected electron quantity Q
BD
changes with the quantum efficiency &ggr;.
It is known that the lifetime T
BD
of the gate insulating film, predicted after the anode hole injection model, is exponentially proportional to the inverse of a stress electric field E
OX
(i.e., an electric field actually formed in the film due to the application of the stress voltage). For that reason, the anode hole injection model is also called a “1/E” model. The stress electric field E
OX
is given by:
E
OX
=(fraction of stress voltage applied to gate insulating film)÷(thickness of gate insulating film)
As opposed to the “1/E” model, the “E” model (see, e.g., J. W. McPherson and D. A. Baglee: Int. Rel. Phys. Symposium (1985) p. 1) supposes that the stress electric field E
OX
itself degrades the gate insulating film and finally causes the dielectric breakdown there. See also J. W. McPherson and H. C. Mogul: J. Appl. Phys. 84 (1998) pp. 1513-1523.
FIG. 13
illustrates a relationship between the stress electric field E
OX
formed in the gate insulating film and the lifetime T
BD
of the film predicted after the “1/E” model and a relationship between the stress electric field E
OX
and the lifetime T
BD
predicted after the “E” model. The data illustrated in
FIG. 13
, i.e., the lifetimes T
BD
of the gate insulating film predicted after the “1/E” and “E” models, was collected from the same gate insulating film as that used for FIG.
12
. And the predicted lifetimes of those films were fitted to each other, i.e., almost equal to each other at a stress electric field E
OX
in the range from 11 to 13 MV/cm.
As shown in
FIG. 13
, the lifetime T
BD
of the gate insulating film as predicted after the “1/E” model (see the dashed line) increases exponentially even in the semilogarithmic plot as the stress electric field E
OX
decreases. But the lifetime T
BD
of the gate insulating film as predicted after the “E” model (see the solid line) shows a negative proportionality for the stress electric field E
OX
(in the semilogarithmic plot).
Second Problem
Hereinafter, a second problem of the known method for predicting the lifetime of an insulating film after the “1/E” or “E” model, for example, will be described.
It has been impossible to definitely decide which should be regarded as the more appropriate voltage-lifetime model, “1/E” or “E”. This is partly
Cole Thomas W.
Kerveros James
Le N.
Matsushita Electric - Industrial Co., Ltd.
Nixon & Peabody LLP
LandOfFree
Method for predicting lifetime of insulating film and method... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for predicting lifetime of insulating film and method..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for predicting lifetime of insulating film and method... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3171035