Radiant energy – Inspection of solids or liquids by charged particles – Methods
Reexamination Certificate
1998-11-17
2001-10-16
Anderson, Bruce C. (Department: 2881)
Radiant energy
Inspection of solids or liquids by charged particles
Methods
C250S310000
Reexamination Certificate
active
06303931
ABSTRACT:
FIELD OF THE INVENTION
This invention relates to metrology tools for quality control of semiconductor wafers. In particular, the present invention concerns metrology tool for determining whether a contact hole (CH) in a resist or in an etched layer is open or closed.
BACKGROUND OF THE INVENTION
During the manufacturing process of semiconductor wafers, the so manufactured dies that are accommodated in the wafer undergo extensive quality control tests, including critical dimension scanning electronic microscope (CD-SEM) tests.
SEM is an electron microscope that utilizes a beam of electrons, accelerated to high energy and focused on the sample, to scan the sample surface, ejecting secondary (and possibly back-scattered) electrons that form the picture of the sample. The secondary electrons (referred to also as signal(s)) that are ejected (referred to also as originating) from the sample are acquired by the SEM. The acquisition step includes sampling and possibly other known per se operations so as to give rise to acquired SEM signals(s) (referred to also as acquired signal(s)). The latter are naturally constituted by discrete values.
The quality control tests including the CD-SEM tests are extremely important in order to assure high yield. Yield, in this connection, stands for the percentage of fault-free chips out of the entire manufacturing lot.
The continued reduction in design rules and increase in metalization layers creates a need for CD-SEM tests of high aspect ratio contact holes (CH), trenches and possibly other patterns (referred collectively as “features”) for sorting out the open and closed features. The known per se utilization of resist and the pertinent quality control tests will now be briefly reviewed.
Thus, during normal manufacturing process, the resist is disposed on the substrate of the wafer and is subject to exposure and development so as to convey a pattern (that corresponds to one mask of the chip'layout) to the underlying substrate. The resist includes, as a rule, a pattern of apertures and masks. According to one out of a few known etching techniques, the apertures enable a chemical substance (normally Acid) to penetrate through the apertures and etch the underlying layer of the substrate. In contrast, the mask portion protects the underlying layer against etching. The manufacturing process of resist is error prone in particular insofar as high aspect ratio features (e.g. contact holes) are concerned. More specifically, it occasionally occurs that the specified holes, do not fully pass through the resist (i.e. they constitute closed holes) with the inevitable consequence that the underlying layer is obviously not etched (regardless of which etching technique is used). As is well known, the patterns that are conveyed to the substrate eventually realize a chip that performs electrical function(s) as defined by the design of the chip. Obviously, failing to accurately convey the pattern that corresponds to the design layout to the underlying substrate (due to inter alia faulty CH in the resist) may adversely affect the chips functional performance and may even require in some cases to discard the manufactured chips. This, of course, degrades the yield score which, from commercial stand point, is intolerable.
It is therefore highly desired to test the resist before etching in order to ascertain with high degree of confidence whether CH's, are passing through i.e. open. In the case that CH's are found to be closed, the resist is discarded and a new one is manufactured. This testing procedure has the obvious advantage (as compared to the post-etching test) in that it is not destructive for the wafer, i.e. a new layer of resist will be applied to the substrate.
In a paper by Eric Solecky and Chas Archie of IBM (Eric Solecky and Chas Archie, “Contact Holes: A Challenge for Signal Collection Efficiency and Measurement Algorithms” published in the proceeding of SPIE, vol. 3050, 172-179 (1997) hereinafter the Solecky and Archie technique, it was suggested to analyze the signal that the CD-SEM acquired from an inspected CH in the resist (before etching) in order to determine whether it is open or closed.
Solecky and Archie proposed to attempt and fit a parabola to the waveform by selecting the best three parameters for the formula that define parabola. The quality of the fit, is determined by utilizing a least square fit algorithm and the resulting square of the correlation coefficient (i.e. high R
2
) is used as a classification criterion. Thus, good fit (i.e. high R
2
) implies the bottom of the CH is round and therefore the CH is closed, while a bad fit (lower R
2
) implies that the bottom of the CH is flat and therefore the CH is open. Specifically, they concluded that R
2
=0.08 differentiates between open and closed CH.
Whilst the proposed approach is basically adequate for classifying CH as open or closed, in suffers from significant shortcomings in that in some real- life scenarios it is not reliable. For a better understanding of the foregoing, attention is directed to
FIGS. 1 and 2
showing two examples where the proposed technique according to Solecky and Archie is susceptible to errors.
FIG. 1
illustrates the so called noise problem where due to low contrast to noise ratio (S/N), the distribution of the discrete values of the acquired signal (
12
) gives rise to a low (R
2
=0.6601) value. It should be noted that the signal in
FIGS. 1 and 2
is not a real signal acquired from inspected CH, but rather a synthetic signal.
Thus, in
FIG. 1
the synthetic signal is constructed by taking a signal being of genuine parabola curve (
10
) and “contaminating” it with noise (
12
). It would be expected that the Solecky and Archie technique will identify high level of correspondence to parabola (i.e. closed CH), but, regretfully, this was not the case. More specifically, according to the Solecky and Archie criterion the CH under question is open since R
2
equals 0.6601 being less than the threshold 0.8. However, since the original signal is of a parabola form, an accurate analysis should have concluded that this particular CH is in fact closed. The net effect is therefore that the bad fit (due to the noisy signal) brings about a faulty indication that the hole under inspection is an open hole.
Turning now to
FIG. 2
, there is shown another scenario referred to as the edge problem. Thus, consider an open CH with relatively steep side-walls. Naturally, due to the steepness of the side walls, most of the discrete signal values correspond to the signal that originates form the “bottom” part of the CH and not from the side walls thereof. It has been found that in noiseless signal conditions (i.e. high contrast to noise ratio S/N), the discrete signal values (e.g.
20
in
FIG. 2
) that originate from the bottom portion fit a parabolic form (e.g.
22
in
FIG. 2
) in a relatively high degree of accuracy (R
2
=0.7642) using the thereby erroneous indication that the CH under inspection is closed. As explained above, failing to correctly classify CH could degrade the yield score of the manufactured die series which is highly undesired.
There is accordingly a need in the art to provide for a technique which classifies inspected features (e.g. CH) as open or closed in a reliable manner, and in particular to provide for a technique which substantially copes with the error prone indications of the specified Solecky and Archie technique.
There is yet another need in the art to provide for new technique which enables to grade the profile of an inspected feature according to the profile's quality (constituting a profile's quality grade) which not only enables a coarse classification of the feature as open or closed, but also enables a fine classification for ranking open features according to their quality.
SUMMARY OF THE INVENTION
It should be noted that for convenience of explanation only, reference is made in the following description to contact holes (CH). Those versed in the art will readily appreciate that the invention is by
Menaker Mina
Veldman Andrei
Anderson Bruce C.
Applied Materials Inc.
Sughrue Mion Zinn Macpeak & Seas, PLLC
LandOfFree
Method for determining a profile quality grade of an... 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 determining a profile quality grade of an..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for determining a profile quality grade of an... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2598036