Remote E-Diagnostics: The Future of Semiconductor Manufacturing

Part 1 Abstract

The semiconductor industry creates many of the critical components that make up the world's Information Technology (IT) infrastructure but has harnessed little of the power of today's IT solutions and their potential benefits for its own manufacturing and supply chain management processes. The intellectual power of the industry has been focused on creating and improving high-technology products and individual tool process technology rather than efficiency of the manufacturing process.

The goal of remote e-diagnostics is to leverage never-before-accessible tool state and performance information, authorized for remote transmission to the supplier, in order to enable the supplier to create models of tool performance and behavior and through analyzing this data suggest means to run the tool more efficiently. This will enable equipment suppliers to "sign up and deliver" an improved level of productivity from their tools. Transparency of tool operations is the key to improving performance.

The Problem

The unscheduled breakdown of an individual piece of manufacturing equipment is a great cost burden, as it results in downtime for dependent/related equipment in the manufacturing chain. Inefficiencies of a single piece of critical "bottleneck" process equipment can have a negative economic impact for an entire production line. Since numerous competitive equipment manufacturers supply the various discrete tools needed in a single fab, service, tool monitoring, and data collection for the tools are inefficient and cumbersome processes.

Industry Statistics

Semiconductor industry statistics show that most semiconductor capital equipment suffers at least 8% unscheduled downtime and loses another 7% to scheduled maintenance. At the January 2000 ISS, Michael Splinter, Senior Vice President and General Manager of Intel's technology and manufacturing group, estimated that each hour of downtime for a critical unit of process equipment can translate into $100,000 of lost revenue in today's chip-hungry market. In a typical fab, just reducing downtime by 1% on the 50 most critical tools can provide revenue opportunities and cost savings nearing one hundred million dollars annually.

80% of MTTRepair is due to two diagnosis issues: complex equipment and low support expertise in the factory. (Source: ITRS Factory Integration TWG Presentation May 2000.)

One way to reduce unscheduled downtime is to improve response time and repair time, and, eventually, to predict when problems will occur before they occur.

Figure 1: On Average, semiconductor process tools spend about half their time doing something productive.

A next-generation e-diagnostic system is required that brings tool-status information together with tool-diagnostic experts thus speeding up the process of helping distressed process equipment back on their feet. Moving from preventative maintenance to predictive maintenance will be enabled when the industry develops tool-failure algorithms based on long-term historical data analysis. Through e-diagnostics, suppliers working closely with fabs will monitor real-time tool status and performance data via secure networks, and thereby automate traditionally failure-prone and time-consuming problem reporting and repair procedures.

The tools are more complicated today

The semiconductor capital equipment in use today is much more complicated than the tools of five or ten years ago. The best expertise for diagnosing problems with these highly sophisticated tools resides in the companies that made them. Supplier experts, however, are not readily accessible to a semiconductor equipment engineer when tools go down.

Modern networking technology, which allows rapid communication with high bandwidth anywhere in the world, can address this problem by providing a very cost-effective means of doing virtual service and support. If you combine the need for experts, the need for improved utilization, and the connection between the vendor and customer, you end up with e-diagnostics.

Figure 2: When each vendor tries to set up their own e-diagnostic links to their customers? process tools, chaos ensues.

The Point-to-Point model

The idea of networking process tools in a semiconductor fab with experts in the tool vendor's customer service department is not unique. As with most good ideas made possible by technology, many tool vendors have developed remote e-diagnostic solutions.

The popularity of this idea is, in fact, its Achilles heel as well. A semiconductor fab contains many different pieces of process equipment provided by many different vendor companies. So many links from so many process tools to so many different vendors—each with their own protocols, link topologies and access rules—makes a nightmare for process managers who just want to make chips. When a tool goes down, the equipment engineer doesn't want to fight their way through a tangled web to reach help. They want a direct path to get help now!
The situation is difficult for vendors trying to provide that help as well. While each vendor has its own system for remote diagnostics, that system doesn't always play well with the customer's internal fab network. Thus, to run a real remote-diagnostic system for customer-service, the vendor has to modify its system to accommodate nuances of the customer's internal network. In the end, each vendor has to deal with different protocols, link topologies and access rules for each of its customers.

Each customer having a separate link with each vendor creates a maze of links as shown in Figure 2. When trouble happens, the vendor and the fab have to find each other in this maze to solve the problem.

The Central Clearinghouse model

The Central Celaringhouse strategy eliminates many of the inefficiencies in semiconductor manufacturing by facilitating real-time information exchange between device manufacturers and their equipment and materials suppliers through the use of Internet technology. This model provides many benefits over the point-to-point model, such as: single point of entry/exit to the manufacturing facility, simplified management of network for both security and user management, easier and faster implementation, consistent interface of hardware and software and substantially lower cost than multiple point-to-point connections. The central clearinghouse model also provides for the ability to store data on a secure central hub for access by authorized parties or it provides for the routing of data to the authorized parties without central storage and access. The following diagram illustrates this point:

Figure 3: The Central Clearinghouse model brings order out of chaos by mediating the flow of e-diagnostic information between tool suppliers and semiconductor manufacturers.

Figure 3 details the Central Clearinghouse system. The system manages routing of messages sent from the fab to the appropriate vendor destination. This ensures that data meant for supplier #1 only goes to supplier #1 and not to supplier #2.

These messages all travel over the Internet through a Virtual Private Network (VPN) ensuring secure connections and data transmission. The equipment vendor similarly has just one link to maintain.

Part 2 The Central Clearinghouse model

Figure 3: The Central Clearinghouse model brings order out of chaos by mediating the flow of e-diagnostic information between tool suppliers and semiconductor manufacturers.

Figure 3 details the Central Clearinghouse system. The system manages routing of messages sent from the fab to the appropriate vendor destination. This ensures that data meant for supplier #1 only goes to supplier #1 and not to supplier #2.

These messages all travel over the Internet through a Virtual Private Network (VPN) ensuring secure connections and data transmission. The equipment vendor similarly has just one link to maintain.

Authorized Queries mean fab data is protected

The vendor's diagnostic expert has the ability to send authorized queries remotely through the network to get more information from the tool that's in trouble. Those queries have been pre-authorized by the factory, so the vendor's diagnostic experts can ask for only information that the fab's management has previously determined that they are willing to share with them. So, a diagnostician at the supplier can say: "Our tool is down and I need information. Let's run query #27." In this way, the diagnostician can get data from the tool, but the factory also has control over what data they are sending.

Semiconductor fabs using the network maintain physical control over their intellectual property because process information resides on a server at their location. Information needed by the diagnostician is sent, in response to a query, over the secure link to the supplier.

Getting Help Fast

To improve tool utilization, the first order of business is to attack the mean time to respond and mean time to repair a problem. The best way to do that is to notify someone quickly.

The following table details a typical scenario.

	Manual Reporting	Hours	Cost	Automatic Reporting - Tool Driven	Hours	Cost
1	Equipment alarm occurs; critical unit of process equipment goes down			Equipment alarm occurs; critical unit of process equipment goes down
2	Operator notices alarm and reports it to "equipment manager."	1	$100,000
3	Equipment manager performs triage to determine if this is a critical failure and contacts on-site, supplier field service technician.	1	$100,000	Equipment supplier's expert 24x7x365 global triage center calls "equipment manager", notifies them of the alarm condition, and begins to diagnose the problem.	1	$100,000
4	Technician gowns up and enters cleanroom.	1	$100,000	If the equipment does not provide enough remote diagnostic capability the technician may still need to enter the cleanroom and begin diagnosing the problem.	0-1	$0 to$100,000
5	Technician diagnoses problem. May need to contact headquarters for help.	2	$200,000	Equipment supplier's 24x7x365 global triage center works with technician to propose a solution based on tool data and comparative analysis with tools in other factories.	1	$100,000
6	Technician locates parts in stock.	1	$100,000	Technician locates parts in stock.	1	$100,000
7	Technician re-enters cleanroom and fixes equipment.	2	$200,000	Technician enters cleanroom with correct parts, latest schematics and solution in hand, and fixes equipment.	2	$200,000
8	Technician requalifies tool.	3	$300,000	Technician requalifies tool.	3	$300,000
9	Equipment back online.			Equipment back online.
		11 Hours	$1.1 Million		8 to 9 Hours	$800,000 to $900,000

The bottom line: Automatic reporting and 24x7x365 triage centers could save up to $300,000 per critical tool failure!

Alert notification and escalation: Field rep is automatically paged, emailed or phoned by the network due to tool down situation. Escalations occur if down time continues past specified length.

Remote e-diagnostics: Field rep and support staff at supplier location can analyze problem via the network to determine probable cause prior to dispatching correct staff with correct parts to customer site. This can eliminate need to "gown-up" and enter fab to diagnose problem, saving at least 1 to 1.5 hours ($100K-$150K).

Reference knowledge base and best practices: Service staff will access knowledge base and best practices databases to determine corrective action. Due to central location, if this problem has happened at any location in the field, service staff will be informed!

Field service rep. works on tool: Correct parts and manuals are made available to get the tool back up quickly.

Billing and spare parts inventory control: Access to service contract and BOM information via the network can be utilized for correct billing and spare parts management. Time and materials information can be fed directly to supplier's billing system.

Operator training: The network will identify which operator was on shift at the time of the failure. This may lead to actions such as learning the tool state prior to failure and assigning the operator additional training.

Information feedback: Data can be fed back from the network to service support staff at the supplier location(s) to populate the knowledge base and best practices databases.

Results: Mean Time To Respond and Mean Time To Repair are reduced.

The Benefits of Diagnostic Data

When new fab construction costs $1B to $2B, semiconductor manufacturers have to get a very rapid return on investment. Downtime rapidly becomes very expensive.

In addition to the 15% taken up by unscheduled downtime and scheduled maintenance, the typical tool spends 27% of its time on set up and running test wafers, and 14% of its time just sitting around doing nothing. About half of a tool's life is spent doing something productive!

With semiconductor equipment productivity at around 50%, there is a big opportunity to increase it by applying proactive asset management. This means building more intelligent process tools and optimizing their performance. The Internet and secure networks will help make this goal a reality through the exchange of information between the fabs and their tool suppliers.

Through the secure network, tool vendors can gather and analyze real-time status and parametric data from equipment operating in the field. This data can give them better information about how their products perform. With better knowledge, they can develop optimization guidelines and predictive maintenance practices to keep the tools up and running at peak utilization.

Today, most device manufacturers do not share production data with their suppliers. They receive reactive support from suppliers' field service organizations. They don't get the kind of proactive support that could help them increase tool utilization and avoid problems.

Tomorrow's suppliers will need to be kept in-the-loop in real-time in order to ensure peak efficiency for their tools.

Data Security

A worldwide network supporting the far-flung operations of chip companies and their tools suppliers must, of course, be very secure and auditable. A secure, neutral information clearinghouse is essential to the industry. The security model should provide firewall-to-firewall authenticated and encrypted data streams, proof positive identification of both sender and receiver, prevention of data tampering or modification, and blocking of any fraudulent data. Additionally, any system like this should be audited not only by the constituents but by one of the big five accounting firms.

Conclusions

The IC manufacturer has complete control over what data can be collected within the fab and what can pass through the firewall from the hub to the network. A contractual agreement between the tool supplier and the IC manufacturer will define all intellectual property issues, such as what data can be published by a tool or sent to the hub and beyond.

The time is now to connect semiconductor manufacturer tools to their suppliers in order to improve tool productivity and significantly enhance the relationship between tool vendors and semiconductor manufacturers, by fostering communication and collaboration. The network will enable tool vendors to capture and analyze in-depth both tool status and performance information, helping reduce downtime, improve throughput, and respond more quickly when problems occur.

Biography

Chris Saso
AvantCom
510-403-2106
[email protected]

Chris Saso has worked in the semiconductor industry since 1991. He has extensive background in semiconductor manufacturing factory automation, applying equipment control technologies and SECS/GEM communications standards to three product lines during his work at Watkins-Johnson. Prior to joining AvantCom Network Saso spent over four years at GW Associates, Inc. the leading provider of SECS/GEM software solutions. Saso provided technical sales and marketing expertise as well training and product requirements generation in his job as Technical Sales Manager. At AvantCom Network Saso is facilitating the paradigm shift that is taking place in the industry about how semiconductor manufacturing will be reshaped through the use of e-diagnostics and information exchanges between equipment suppliers and their customers.