Data processing: measuring – calibrating – or testing – Measurement system – Statistical measurement
Reexamination Certificate
2000-03-15
2003-09-02
Hoff, Marc S. (Department: 2857)
Data processing: measuring, calibrating, or testing
Measurement system
Statistical measurement
C714S054000
Reexamination Certificate
active
06615157
ABSTRACT:
FIELD OF THE INVENTION
This invention relates generally to systems, methods, and computer program products for assessing experiment results and, more particularly, to systems, methods and computer program products for analyzing experiment results obtained in a manufacturing process to determine the effect of a control variable on various attributes of the manufacturing process.
BACKGROUND OF THE INVENTION
Large-scale producers of commercial products typically have a standard manufacturing protocol for each type of product. For example, a semiconductor chip manufacturer making 16 megabit and 64 megabit dynamic random access memory (DRAM) chips usually has a separate protocol for each of the two types of chips. Normally, the standard protocol is constantly modified so as to increase production efficiency and improve product quality.
For purposes of modification, multiple experiments are typically conducted using the standard protocol in which one or more control variables are adjusted and the experiment results are examined. Typically, there are many types of control variables in a manufacturing process that can be adjusted. For example, a semiconductor manufacturing process typically includes control variables such as critical chip dimensions, thickness of a particular metal layer, over-etch time of a particular layer, and the like. Each control variable can be set into a number of different states. Typically, two parallel experimenting processes are conducted in two different states of a particular control variable while the other control variables remain the same in the two processes. Experiment results are collected during and/or after the processes and are examined. For example, to determine the effect of the critical chip dimensions in a 16 MB DRAM chip processing procedure, two parallel processes can be conducted. The only control variable that is set different in the two processes can be the amount of exposure of the chips during the micro-imaging step. After the processes are completed, experiment results are obtained including a variety of experiment data corresponding to a plurality of attributes such as resistance values, capacitance values, yields, etc. The experiment data are analyzed and the analysis results are used as the basis for modifying the standard fabrication protocol.
Normally, a large amount of experiment data is analyzed in order to prevent diagnostic errors. In conventional methods, the experiment data is organized into tables in which the statistical analysis results for certain attributes are indicated. For example, the experiment data can be analyzed to determine the maximum value, minimum value, mean value and maximum difference value corresponding to a particular attribute. The information is included in tables. Engineers then manually examine each of the tables one by one. Normally, there are a large number of tables generated, and engineers often do not have time to examine all the tables and simply rely on personal experience to examine a portion of the tables. For example, an experienced engineer will consider the effects of adjusting the polycrystalline silicon photo on the NMOS and PMOS experiment data. Therefore, they will examine the tables concerning NMOS and PMOS but ignore the experiment data related to other attributes, e.g., the N-shaped well resistance value; which are not expected to be affected by polycrystalline silicon photo.
Conventional methods rely entirely on manual inspection and analysis of experiment data, and thus are extremely time consuming. In addition, conventional methods typically depend on personal experience, and normally only a portion of the experiment data can be examined and analyzed. Unexpected effects are often overlooked. This problem is especially serious with an inexperienced engineer. This results in inadequate consideration of the experiment data and erroneous decisions in the entire fabrication process, often causing significant losses. To aggravate this problem, very often multiple control variables are adjusted in a single experiment, and interactions between the two control variables need to be examined. Conventional methods with manual analysis are not equipped to handle the experiment data in such complex situations.
SUMMARY OF THE INVENTION
The systems, methods and computer program products of this invention can efficiently analyze experiment results, especially in complex situations in which a large number of control variables are adjusted and a large amount of experiment data needs to be analyzed. In accordance with the first aspect of the invention, a system is provided for analyzing experiment results of a plurality of attributes in a process having a control variable. The system includes a processor which compares the experiment results with a know-how database having attributes expected to be affected by changes in the control variable. In one embodiment, two parallel experimenting processes are conducted in two different states of a particular control variable while the other control variables remain the same in the two processes. The experiment results obtained from the two processes are analyzed by a processor included in the system. The processor compares the experiment data from the two processes and identifies those attributes having a statistically significant difference in their corresponding experiment data obtained in the two processes. Such attributes can be organized into an assessment database. Preferably, the processor compares the identified attributes, preferably contained in an assessment database, with a know-how database (e.g., a correlation table provided by process experts indicating the correlation between certain control variables and attributes that are expected to be affected by adjustment of the control variables). The processor then generates a conformity database containing conforming attributes, i.e., those attributes that behave as predicted in the know-how database. A non-conformity database is also generated containing non-conforming attributes, i.e., those attributes on which the effect of the change in the control variable is not the same as predicted in the know-how database.
If the non-conformity database does not contain any attributes, then all attributes for which there are differences are within the expected range. The fact that at least one attribute appears in the non-conformity database would indicate that either an as-yet undiscovered property in the process is found, or an error in the experiment has occurred.
The system may also contain a memory device. Numerical representations of experiment data, analysis results and/or the know-how, conformity, and non-conformity databases can be stored in the memory device. Alternatively the numerical representations can be stored elsewhere, such as within the processor, if so desired.
In accordance with a second aspect of this invention, a computer program product is provided. The computer program product comprises a computer readable storage medium having computer readable program code means embodied in the medium. The computer readable program code means includes computer instruction means for comparing a first and second plurality of experiment data to generate an assessment database. Attributes having a statistically significant difference in their corresponding experiment data in the first and second experiment results are identified in the assessment database. In addition, computer instruction means is also included for comparing the assessment database with a know-how database to generate a conformity database and a non-conformity database.
The computer readable storage medium may be part of the memory device, and when used, the processor of the present invention may implement the computer readable program code means to compare a first and second plurality of experiment data to generate an assessment database, identify attributes having a statistically significant difference in their corresponding experiment data in the first and second experiment results in the assessment database, and to compare
Alston & Bird LLP
Hoff Marc S.
ProMos Technologies Inc.
Raymond Edward
LandOfFree
System method and computer program product for automatically... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with System method and computer program product for automatically..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and System method and computer program product for automatically... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3028547