Data processing: measuring – calibrating – or testing – Measurement system – Dimensional determination
Reexamination Certificate
2000-09-08
2003-03-25
Barlow, John (Department: 2863)
Data processing: measuring, calibrating, or testing
Measurement system
Dimensional determination
C250S548000, C359S629000, C702S095000, C716S030000
Reexamination Certificate
active
06539331
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates generally to optical measurement systems. More specifically, the present invention relates to the measurement of features in semiconductor manufacturing.
BACKGROUND OF THE INVENTION
The successful manufacture of advanced sub-micron sized semiconductor devices requires the detection, measurement and evaluation of defects and imperfections as small as 1 micron on the photographic mask (photomask) used to pattern the wafer. Defect inspection and measurement techniques for masks therefore play an important role in mask making and quality assurance. Measurement of line widths is also important in producing an acceptable mask.
Thus, it is becoming increasingly important to be able to identify and to correctly size mask defects, line widths, and other features that are under 1 micron in size. Accurate sizing of these features allows masks that are below specification to be repaired, and prevents the needless and costly hold up of masks that do meet specification. However, one of the problems of assessing reticle quality at these sub-micron levels on an automatic inspection system is that the size of these features cannot always be accurately, quickly and cost-effectively measured in a production environment.
It has long been known that mask inspection tools are not measurement tools and that the size information provided by these tools has limited value. Consequently, many mask makers have incorporated measurement aids at the inspection station or move the mask to a more suitable measurement tool in order to make classification decisions. Measurement aids used at the inspection station include calipers, grids, and software based video image markers such as gates, scales, grids, boxes and circles. These aids are fairly rapid, but ultimately require the operator to “eyeball” the boundaries of the defect. This activity is very subjective and can lead to an error in the measurement of the defect.
For example, feature size is often measured by measuring the distance between opposite edges of the feature. Once a feature is identified by an inspection machine, the operator uses a video microscope and a television camera to position a cursor on one side of the feature and another cursor on the other side of the feature. The operator must judge for himself the exact boundaries of the feature and must place the cursors where he sees fit. At this point, the operator pushes a button and the software blindly computes the distance between the two cursors in order to supply a rough approximation of the dimension of the feature. This technique has many disadvantages.
Firstly, this measurement technique is operator dependent in that the operator must manually position the cursors on the boundaries of what the operator believes to be the feature. The operator may misjudge the type of a feature, its boundaries, or may simply misplace a cursor even if the feature is visible. The software then calculates the distance between the cursors, without regard for the type of feature, its true boundaries, etc. The above technique may be performed with a standard video microscope and has an accuracy of about 0.1 micron, but is completely subject to the operator's skill level and interpretation.
Another difficulty with light measurements of features less than 1 micron in size is that the wavelength of photons begins to interfere with the measurement of these 1 micron and less feature sizes. Many techniques do not adequately address the non-linearities associated with such measurements.
Alternatively, the mask may be removed from the automatic inspection tool and relocated on a more precise and repeatable measurement tool. However, this approach involves removing the mask from production, relocating the feature, and is thus impractical in a production environment. This technique is also costly, time-consuming and increases the handling risk. For example, an atomic force microscope (AFM) may be used to measure feature sizes; such a microscope is extremely accurate but is very slow, very expensive and is still subject to operator interpretation.
One approach that has been taken that uses calibration of an automatic inspection system in order to size defects is described in
Characterization Of Defect Sizing On An Automatic Inspection Station
, D. Stocker, B. Martin and J. Browne, Photomask Technology and Management (1993). One disadvantage with the approach taken in this paper is that it only provides a technique for measurement of defects of 1.5 microns and greater. Such sizes of defects would produce a linear relationship between reference sizes and actual measured sizes, and the paper does not account for defects less than 1 micron that would produce a non-linear relationship.
In general, measurement of line widths on a semiconductor mask can be broken into the two categories of isolated lines and dense lines. U.S. Pat. No. 5,966,677 describes a useful flux-area technique for measuring isolated defects or line widths, especially when they are smaller than the wavelength of light used to view them. This technique works best when at least 1.5 blur distances (the blur distance equals the wavelength in diffraction-limited optics) separates lines and/or features. Unfortunately, lines are not always separated by such distances.
The category of dense lines refers to adjacent lines (or other features) where the distance between the lines is less than about 1.5 times the wavelength being used to measure the lines. Prior art techniques for measuring line widths of dense lines are not always accurate or desirable.
In particular, line width measurement (also known as critical dimension or CD measurement) suffers increasingly from two basic limitations as feature sizes shrink. Firstly, features that are smaller then the wavelength of light become very hard to measure because their edge-to-edge measurement appears to approach the wavelength of the light used to examine them. Thus, measurement of line widths that are about the wavelength of light is difficult using prior art techniques. Secondly, measurement of a line width that has an adjoining feature (a line or other feature) that is less then about one and one-half times the wavelength of the examining radiation in distance becomes difficult. The defracted light from the neighboring feature confuses the measurement of the line width. Usually the confused measurement will be smaller then it should be because the neighboring feature will appear to increase the background light level.
FIGS. 1 and 2
illustrate problems with the conventional full-width half-maximum technique for measuring widths of dense lines.
A phenomenon is known as the optical proximity effect (OPE) influences critical dimension measurements of photographic mask and wafers in semiconductor manufacturing and has been a subject of intense interest and investigation for many years. OPE is caused by the convolution of the intensity profiles of adjacent lines and introduces errors in the determination of the line edge positions and in turn the line width. OPE is the result of any optical imaging system where adjacent line edges spread spatially such that the tail of the adjacent edge spread function if sufficiently close, influences the shape and thereby the determination of the position of the line edge in question. If the line edges for the line in question cannot be determined accurately, the line width cannot be measured accurately.
FIG. 1
shows a prior art approach which uses the full-width half-maximum technique to determine line width where the lines are isolated. Graph
10
is an intensity profile for an isolated line. Graph
10
includes an intensity axis
12
and a width axis
14
in microns. Intensity profile
16
is a profile for an isolated line such as is present on a photographic mask that is desired to be measured. Taking the 50% threshold values 18 and 20 for profile
16
easily leads to a calculation for the width of the line to be about 1 micron. For such an isolated line, measurement of its width based upon its intensity profile is re
Beyer Weaver & Thomas LLP
Le John
LandOfFree
Microscopic feature dimension measurement system does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Microscopic feature dimension measurement system, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Microscopic feature dimension measurement system will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3000663