Radiant energy – Inspection of solids or liquids by charged particles – Methods
Reexamination Certificate
2001-04-05
2003-10-21
Nguyen, Kiet T. (Department: 2881)
Radiant energy
Inspection of solids or liquids by charged particles
Methods
C250S306000, C250S310000
Reexamination Certificate
active
06635872
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to methods and systems for use in defect inspection of micro-fabricated structures such as integrated circuit die on semiconductor wafers, masks or reticles for micro-fabrication, flat panel displays, micro-electromechanical (MEMs) devices and the like during and after manufacture. In particular, the invention provides methods and systems for more effectively and efficiently using defect inspection system time and for reducing the average defect inspection time required to determine defect density and other defect statistics.
2. Prior Art
Over the past decade, defect inspection to detect microscopic manufacturing defects has become a standard part of micro-fabrication manufacturing flows, especially for semiconductor wafers. Defect inspection is performed on a statistical sampling basis of the wafers in the manufacturing flow at key defect-prone or particularly defect-sensitive steps. Various types of inspection technology are in use including bright-field optical inspection with, for example, a KLA-Tencor 2138 made by KLA-Tencor of San Jose Calif., darkfield inspection with for example a KLA-Tencor AIT2 also made by KLA-Tencor. More recently e-beam inspection, with for example Odyssey 300 by Schlumberger Technologies Inc. of San Jose or a KLA-Tencor eS20 made by KLA-Tencor, has emerged has an important tool. Each type of inspection technology is usually applied at steps in the semiconductor manufacturing flow where it is best suited to the types of defects most likely to be found.
Wafer inspection systems for patterned wafer inspection usually work as follows. A high powered microscope, traditionally an optical microscope, but more recently a SEM (Scanning Electron Microscope) or electron microscope, is setup under computer control to acquire sequentially images of the area of wafers to be inspected. To minimize the overhead of wafer stage movement and settling time during the inspection process, continuous scanning motion mechanical stages are used. These stages are specifically designed to have very smooth motion in at least one scanning axis to facilitate image or pixelated contrast data acquisition. In the case of an optically based system, a TDI-CCD (Time Delay Integration-Charged Couple Device) image sensor is synchronized with the scanning motion of the continuous scanning stage to acquire images rapidly. In the case of an e-beam inspection system, the scanning motion of the beam is synchronized with the scanning stage motion to acquire images rapidly.
The image or pixelated contrast data that is acquired in this manner is then compared to reference data. Defects are found or detected where there are differences between the reference and the acquired images. The reference images may be derived from CAD data or may simply be images of neighboring cells or die on the wafer or similar wafer being inspected. The sensitivity of the defect inspection process can be controlled by adjusting the image processing parameters that are used to compare the acquired and reference images.
The economic benefits of inspection have been substantial and inspection is generally accepted as having made a significant contribution to the substantial increase in semiconductor wafer manufacturing yields seen in the 1990s.
Inspection systems are employed in a number of different applications including:
process monitoring to flag when a particular process step in the manufacturing flow has increased the number of defects produced above the level normally anticipated at that step;
problem solving by inspecting so-called short-loop wafers that have only been processed with a subset of the manufacturing process steps in order to facilitate troubleshooting and diagnosis or optimization of a particular subset of process steps and
during process development-to optimize a new manufacturing process to reduce or eliminate process-specific or systematic defect mechanisms.
In many cases, it is not only the location of individual defects that is the primary goal of defect inspection. Rather it is measuring the defect distribution over the wafer or sub region thereof, and defect density (that is, the number of defects per unit area) of all defects or of a particular type or class of defect. This information is often required to make decisions about the status of manufacturing, the severity of a problem or an issue in process development.
In a production-worthy process, typical defect densities per process step must be extremely low in order for the yield of the whole manufacturing process flow to be good. With several hundreds of individual process steps typical today in an advanced semiconductor wafer manufacturing flows, killer (critical or yield limiting) defect densities per process step usually range from 0.1-0.0001 defects/cm{circumflex over ( )}2 (defects per centimeter squared) or less.
During process development, defect densities can be much higher, for example, up to many thousands of defects/cm{circumflex over ( )}2 if the process is particularly immature or under certain troubleshooting conditions when higher defects densities may have been deliberately induced for diagnostic or characterization reasons. Such defects may include subtle non-killer defects—measured for reasons of improving quality or reliability, for example, metal voids that might migrate and cause reliability failures after the IC is in use in a system.
The throughput or speed of the inspection process is particularly critical when the inspection system is used as part of the manufacturing process flow. It is usually necessary to be able to measure or monitor defect density in under one hour in the production flow. The overall defect inspection system throughput typically scales as a function of the square of the pixel size used. For example, if the pixel linear dimension is halved in order to be able to find smaller defects, the overall inspection speed will decrease by a factor of four assuming that the pixel data rate remains constant.
This square law dependence that throughput has on pixel size is particularly critical for more advanced higher resolution inspection systems such as e-beam and ultraviolet (UV) optical where the frequent need to use smaller pixels slows the inspection speed significantly.
The slow speeds of high-resolution defect inspection, combined with the fact that the minimum wafer surface area that must be inspected to make a statistically significant measurement of defect density is not known ahead of the inspection, results in inefficient use of expensive inspection system time. To ensure a statistically useful result from inspection, relatively larger wafer surface areas must be inspected. The consequence of this inefficiency is longer inspection time, worse inspection system utilization and higher overall inspection cost.
SUMMARY OF THE INVENTION
In view of the above problems, an object of the present invention is to provide a method for reducing average defect inspection time by using in-situ statistical analysis of defect data during the operation of the inspection system to allow, under certain circumstances, time savings by early termination of some defect inspection runs. In accordance with preferred embodiments of the invention, a method of reducing defect inspection time of micro-fabricated structures by statistically analyzing defect data during inspection, the method comprising inspecting the micro-fabricated structures by acquiring pixelated contrast data from the micro-fabricated structures; detecting defects in the micro-fabricated structures by comparing the pixelated contrast data with reference data to create the defect data; calculating at least one statistic of the defect data; continuing inspection while the at least one statistic is outside a predetermined range and stopping inspection when the at least one statistic is within the predetermined range.
A method of reducing defect inspection time of micro-fabricated structures by statistically analyzing defect data during inspection, the method comprising: ins
Applied Materials Inc.
Blakely & Sokoloff, Taylor & Zafman
Nguyen Kiet T.
Talbot Christopher G.
LandOfFree
Defect inspection efficiency improvement with in-situ... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Defect inspection efficiency improvement with in-situ..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Defect inspection efficiency improvement with in-situ... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3126015