RequestLink
MICRO
Advertiser and
Product
Information

Buyer's Guide
Buyers Guide

tom
Chip Shots blog

Greatest Hits of 2005
Greatest Hits of 2005

Featured Series
Featured Series


Web Sightings

Media Kit

Comments? Suggestions? Send us your feedback.

 

MicroMagazine.com

APC/AEC

Meeting reliability requirements for 300-mm CMP manufacturing using integrated metrology

Hirofumi Seo, Hiroshima Elpida Memory; and Ori Braitbart, Nova Measuring Instruments

A traditional barrier to the widespread adoption of integrated metrology has been the perception that it cannot be relied on by the host process tool. The 300-mm fab environment is characterized by a high level of automation and a large number of wafers in process. Hardware failures, software reliability issues, or process excursions have a very big price tag. Since a flawless process is the desired goal, any interruption, even one caused by temporary operator error, is viewed as a failure that must be eliminated. Consequently, it has become clear that for integrated metrology to be implemented and to proliferate to many toolsets, it must be highly reliable.

Considering the complex relationship between integrated metrology, the process tool, and the fab host, adopting integrated metrology is not a simple task. It requires a tight integration scheme, high-quality execution, and reliable support. As it achieves broader acceptance not only in the traditional area of chemical-mechanical polishing (CMP) but also in the etch, lithography, and chemical vapor deposition areas, reliability will come to play a more pronounced role in the selection of metrology equipment. Extensive in-house reliability testing performed before the metrology instru-
ment is shipped to the customer and subsequent integrated testing on the toolmaker’s production floor are only two prerequisites for achieving the goal of high field reliability.

The CMP production floor is a dynamic environment in which frequent modifications are made, many of which can affect metrology-tool reliability. For example, polisher and metrology-tool software updates can result in operating failures, recipes can cause malfunctions that lead to measurement failures, the product mix can cause measurement or operating failures, nonstandard tool qualifications or preventive maintenance can result in measurement or operating failures, and lack of operator skills can lead to errors.

Metrology-tool reliability must be approached as a global task that integrates the tool with the production environment, taking into account hardware, software, and operations to ensure minimal interruptions. This article presents a step-by-step methodology developed by Nova Measuring Instruments (Rehovot, Israel) for achieving high metrology reliability in the field. It includes data collected during the past two years at various customer sites, including Hiroshima Elpida Memory in Japan, a high-volume manufacturer that considers reliability to be of great importance. Nova and Elpida worked together in a continuous program that improved reliability by a factor of six within five months.

Global Methodology

A global-reliability approach involves joint collaboration between the vendor and the end-user. A schematic diagram of this approach is shown in Figure 1.

Figure 1: Schematic diagram of the global-reliability approach.

Ensuring high metrology-tool reliability before the tool is shipped to the customer is a prerequisite for achieving high reliability in the field. At the metrology vendor site, the reliability of the metrology tool is determined by performing marathon runs of single or multipe modules. To reduce the time required to complete the validation, acceleration testing is used where relevant.

As presented in Figure 2, the methodology process flow in the field involves reliability data collection, data processing, Pareto investigation, and validation of corrective actions.

Figure 2: Schematic diagram showing the flow of field reliability data.

• Reliability Data Collection. The metrology vendor collects reports of interruptions caused by the metrology unit integrated inside a CMP module. These interruptions are recorded in a log file. The log file is analyzed by the metrology vendor’s proprietary software, which extracts only the data related to interruptions caused by the metrology unit. Containing the number and the nature of the interruptions, this filtered file is sent by a field service engineer on a weekly basis to the metrology vendor’s headquarters for processing. The numbers are normalized to the number of wafers processed through the unit and are expressed in terms of the number of events per 10,000 wafers.

• Data Processing and Pareto Investigation. Using proprietary software, the filtered files are collected and processed according to tool configuration. Tools with the same configuration are benchmarked for best performance. The lower-ranked tools undergo reliability corrective actions. The ranking is done based on tool interruption Pareto charts; the items on the top of the list have a higher priority than those lower down. Benchmarking is performed at the site and the global level, ensuring that the metrology vendor can make continuous improvements. Reliability case analyses are led by integrated teams of software and hardware experts. These analyses identify root causes and initiate systematic solutions to prevent failure reoccurrences. Hence, continuous improvement in the field is achieved.

• Validation of Corrective Actions. All corrective actions are followed up weekly as part of ongoing reliability monitoring.

Analysis of the Field Data

Figure 3 shows a typical Pareto chart of field reliability at a customer site, which can prompt corrective actions in the following areas:

Figure 3: Typical Pareto chart of field reliability at a customer site. Such data are collected over a period of several weeks.

• Communication Software Issues. The log files from the vendor’s metrology systems and the polishers in the field are analyzed in order to investigate and solve bugs. Generally, the result of this process is the installation of a new software version on the metrology unit or the process tool. In some cases, the bug may be caused by wrong settings or improperly executed mechanical adjustments, which are resolved without software changes.

• Operation Issues. Operation problems are generally related to errors caused by technicians working on the CMP modules. In such cases, corrective actions are carried out to improve operators’ skills or to simplify the use of the tools.

• Process Issues. Process variability frequently results in image-contrast variations on the wafer, which can lead to pattern-recognition failures. Such variability, which is inherent in the fab environment, requires that skilled technicians make a continuous and sustained effort to tune and optimize recipes.

Weekly Field-Data Case Studies

Case Study 1. Figure 4 presents an example of a weekly Pareto that was used to monitor the performance of a new software version at a customer site. The new software version installed on the vendor metrology tools at the site caused a bug to appear during week 7 (represented by the red bars in the chart). The software bug was analyzed and fixed in a new version of the software. After the new software version was tested at the vendor and toolmaker sites, it was installed at the customer site after week 11. Figure 4 shows that the interruptions ceased after the software bug had been eliminated.

Figure 4: Weekly Pareto chart of field reliability at a customer site showing the number of weekly interruptions per 10,000 wafers. The colors in the bars represent different root causes of the interruptions.

Case Study 2. Figure 5 presents the weekly Pareto of CMP-module interruptions that were caused by the vendor metrology unit installed on all CMP tools engaged in high-volume mass production at another customer site. An investigation was conducted to uncover the source of the interruptions, the first step of which was to determine whether the high interruption rates observed during weeks 6 and 7 were a general or a tool-specific problem. From Figure 6, it is clear that the problem was located on tool 5. Further investigation determined that a worn-out part was the root cause of the failure. After the part was replaced, the efficiency of the corrective action was validated using a Pareto chart that covered the period starting with week 8. Figure 6 shows a tool-by-tool Pareto of field reliability at the customer site during week 7. The chart identifies the problem tool.

Figure 5: Weekly Pareto chart of field reliability at a customer site showing the number of weekly interruptions per 10,000 wafers. The colors in the bars respresent different root causes of the interruptions.
Figure 6: Tool-by-tool Pareto chart of field reliability at a customer site during week 7 showing the identification of the problem tool.

Conclusion

The weekly monitoring of metrology-tool field reliability provides an efficient methodology for improving CMP tool performance and maximizing uptime in high-volume mass-production settings. This methodology is based on worldwide data collection and analysis.

This global-reliability approach ensures that bad reliability trends are detected before crash levels are reached. It also validates best-known methods and implements them worldwide. Finally, the approach measures the reliability of different sites, tools, and configurations using the same parameters so that equipment performance at semiconductor fabs throughout the world can be compared and improved. The reliability enhancements at Elpida have contributed to smooth high-volume manufacturing in the CMP module.


Hirofumi Seo is the manager of the equipment engineering department at the Hiroshima Elpida Memory fabrication facility in Japan. (Seo can be reached at info@elpida.com.)

Ori Braitbart, PhD, is director of corporate quality and reliability at Nova Measuring Instruments in Rehovot, Israel. He has more than 16 years of experience in the high-tech R&D and industrial fields. Before joining the company as a project manager in the R&D department, Braitbart was a process engineer at Intel. He received a PhD in experimental physics from the Hebrew University in Jerusalem in 1990. (Braitbart can be reached at +972 8 9387587 or ori-b@nova.co.il.)

 


MicroHome | Search | Current Issue | MicroArchives
Buyers Guide | Media Kit

Questions/comments about MICRO Magazine? E-mail us at cheynman@gmail.com.

© 2007 Tom Cheyney
All rights reserved.