Image analysis – Applications – Manufacturing or product inspection
Reexamination Certificate
1999-02-16
2002-09-24
Au, Amelia M. (Department: 2623)
Image analysis
Applications
Manufacturing or product inspection
C382S228000, C073S156000, C348S126000, C438S016000
Reexamination Certificate
active
06456736
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to a method for monitoring processing carried out on a semiconductor substrate, and more particularly for inspecting the critical dimensions of features formed on the semiconductor substrate using statistical sampling. The invention has particular applicability for in-line inspection of semiconductor wafers during manufacture of high-density semiconductor devices with submicron design features.
BACKGROUND ART
Current demands for high density and performance associated with ultra large scale integration require submicron features, increased transistor and circuit speeds and improved reliability. Such demands require formation of device features with high precision and uniformity, which in turn necessitates careful process monitoring, including frequent and detailed inspections of the devices while they are still in the form of semiconductor wafers.
Conventional in-process monitoring typically includes statistical process control techniques, wherein critical dimensions (CDs) of selected devices or “fields” on wafers are measured, as by a critical dimension scanning electron microscope (CD-SEM), to identify lot-to-lot variations, wafer-to-wafer variations within a lot, and field-to-field variations between devices on a single wafer. This information is thereafter used in diagnosing processing problems. The measurement of field-to-field CD variation has been found to be one of the most meaningful indicators of process control effectiveness; i.e., a large variation in CDs of comparable features of different fields on a single wafer indicates process control problems. Therefore, statistical measurement of field-to-field CD variation is typically performed as a standard inspection procedure at an automated inspection tool, such as the 7830SI or VeraSEM, available from Applied Materials of Santa Clara, Calif.
In one type of automated inspection procedure, wherein CDs of comparable features in several sample fields on a wafer are measured, sampling is dictated by a “recipe” such that the same points are sampled on every wafer under inspection. However, such fixed-field sampling leads to statistical errors, as when repeated appearances of a high systematic error data point amplifies the total error, thus inaccurately representing the errors present in the whole population. For example, due to the round shape of the wafers and the limitations of the photoresist application process steps such as coating and baking, a center-to-edge CD variation typically (i.e., systematically) exists. Thus, if one of the fixed sample fields is in an area of the wafer adversely affected by such a systematic CD variation, the error always present in that point will cause the total measured error in the wafer to increase. Furthermore, the fixed-field sampling technique requires a relatively large number of points be measured on each wafer under inspection; e.g., up to about 17 points. Such measuring requires a significant amount of time and thereby reduces production throughput.
To reduce inspection time and to avoid the inherent statistical errors of fixed-field sampling, some automatic inspection schemes employ random field sampling, wherein a predetermined number of fields (specified by the user) on each wafer under inspection are randomly selected for CD measurement. The random field selection technique yields more accurate results from fewer fields vis-a-vis fixed-field methodology.
However, because the number of randomly selected fields is a constant predetermined number, undersampling or oversampling may occur using this technique, depending on the selection of the predetermined number of fields. For example, oversampling may occur, thereby unnecessarily prolonging the inspection and reducing production throughput, when the CD variation from field to field is extremely small or nonexistent, because a lesser number of samples than the predetermined number are needed to obtain a statistically accurate result in this situation. On the other hand, undersampling may occur, leading to inaccurate statistical results, when the field-to-field CD variation is relatively large, because a greater number of samples than the predetermined number may be needed in this situation.
There exists a need for a more accurate and cost-effective methodology for in-process inspection of semiconductor wafers to provide information relating to feature CDs for statistical process control, in order to identify processes causing defects, thereby enabling early corrective action to be taken. This need is becoming more critical as the density of surface features, die sizes, and number of layers in devices increase, requiring the number of defects to be drastically reduced to attain an acceptable manufacturing yield.
SUMMARY OF THE INVENTION
At advantage of the present invention is the ability to perform inspection of features on randomly selected fields of a semiconductor wafer for statistical process control without oversampling or undersampling, thereby obtaining statistically accurate results and increasing production throughput
According to the present invention, the foregoing and other advantages are achieved in part by a method of inspecting the surface of an article having a plurality of comparable pattern units, each pattern unit having a feature, which method comprises randomly selecting a predetermined number (i.e., one or more) of the plurality of pattern units for measurement, measuring a critical dimension (CD) of the feature in each of the randomly selected pattern units to produce a first plurality of CDs, then calculating a statistical function (e.g., an average or standard deviation) using the fast plurality of CDs to produce a first statistical function value. An additional pattern unit of the plurality of pattern units is then randomly selected, the CD of the feature is measured, and the statistical function is calculated using the first plurality of CDs and the CD of the feature in the additional pattern unit to produce a current statistical function value. The inspection of the surface of the article is then terminated if the difference between the current statistical function value and the first statistical function value is less than a predetermined amount. If the current statistical function value changes the statistical function value more than the predetermined amount, the inspection continues with further random selection of additional pattern units.
Another aspect of the present invention is an apparatus for carrying out the above method.
A further aspect of the present invention is a computer-readable medium bearing instructions for inspecting the surface of an article having a plurality of comparable pattern units, each pattern unit having a feature, said instructions, when executed, being arranged to cause one or more processors to perform the steps of randomly selecting a predetermined number of the plurality of pattern units for measurement, controlling an inspection tool to measure a critical dimension (CD) of the feature in each of the randomly selected pattern units to produce a first plurality of CDs, calculating a statistical function using the first plurality of CDs to produce a first statistical function value, randomly selecting an additional pattern unit of the plurality of pattern units for measurement, controlling the inspection tool to measure the CD of the feature in the additional pattern unit, calculating the statistical function using the first plurality of CDs and the CD of the feature in the additional pattern unit to produce a current statistical function value, and terminating the inspection of the surface of the article based on the first and current statistical function values.
Additional advantages of the present invention will become readily apparent to those skilled in this art from the following detailed description, wherein only the preferred embodiment of the present invention is shown and described, simply by way of illustration of the best mode contemplated for carrying out the present invention. As will be realized,
Osborne Zoe
Su Bo
Applied Materials Inc.
Au Amelia M.
Dastouri Mehrdad
McDermott & Will & Emery
LandOfFree
Automatic field sampling for CD measurement does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Automatic field sampling for CD measurement, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Automatic field sampling for CD measurement will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2839385