Issue



Breakthrough factory productivity using e-Manufacturing


09/01/2003







Overview

For several years, many in the industry have criticized the shortcomings of fab automation, including the somewhat desperate need to package it all as e-Manufacturing. Is it all coming together now? Here is a rundown of the vision being driven by fab challenges and a summary of how e-Manufacturing seems to be emerging.

To meet the needs of customers and manufacturing partners, IC makers must get new products to market quicker, support shorter customer cancellation windows, and meet the complexities of multicompany joint ventures. From a technology and manufacturing perspective, IC makers must meet the demands of increasingly complex process windows and process integration, compensate for reduced product margins, ensure that expensive capital is highly utilized, maintain the two-to-three-year process technology roadmap timeline, and transition to 300mm wafers for efficiency.

These challenges are driving the factory to meet very aggressive metrics that support customer requirements. The International Technology Roadmap for Semiconductors (ITRS) provides a long-term strategic perspective on where the industry is going and what is needed for the next 15 years. The table is a sample of factory metrics from 2003 to what is needed in 2009.

Lithography critical dimension (CD) control continues to be ever tighter with each process technology generation, while the need to maintain film thickness variance must be maintained at smaller geometries. Process requirements for high accuracy and high repeatability, while running in a "touchless" highly automated fab, are driving the need for sophisticated process control strategies and systems.

Shorter customer new and volume product delivery times translate to faster lot movement through factories. This, coupled with an increase in device complexity (i.e., more transistors, more metal layers, etc.), means that we need to accelerate the velocity of lots through the factory. The pending 2003 ITRS is discussing this in terms of days/mask layer, where a five-wafer "rocket" lot moves through the factory at a rapid 0.5 days/mask layer. Meeting these rapid cycles times is possible today for a few lots, but IC makers are grappling with the challenge of doing this with many concurrent hot lots without degrading factory performance elsewhere (i.e., velocity of normal production, utilization loss for production equipment, high labor, etc.).

Another challenge is getting high reliability from new leading-edge equipment, which is extremely important to reducing the cycle time for new process technology development. Similarly, having this reliability before the manufacturing ramp is critical to recouping high capital expenses and enabling future investment. Today's equipment often does not meet the reliability needs at these two critical phases.

Click here to enlarge image

null

Future vision

Today's 300mm wafer fab is highly automated. Key capabilities that are prevalent include the use of 100% automated material-handling systems (AMHS) for carrier delivery, fully automated control of process equipment, use of dispatching systems to determine the priority of lots processed at each tool and throughout the factory, fully automated data collection, and centrally administered recipe management.

The next generation of manufacturing will need to extend the 300mm factory with additional capabilities that meet the challenges discussed above for the 65nm, 45nm, and 32nm production environments and solve some of the challenges above in the areas of cycle time reduction, high reliability and availability, and more repeatable control of future process technologies (Fig. 1).

To improve overall factory productivity, e-Manufacturing applications are being developed to improve decision-making speed and the availability of equipment, and reduce recovery time from excursions. These applications are also automating and optimizing the execution of previously manual tasks to reduce the effort, mistakes, and associated labor, including:

  • Remote diagnostics. A supplier's ability to remotely access equipment or software from outside the factory to diagnose issues, suggest a change, configure the tool, or download new software, documentation, data, equipment constants and parameters, and leverage remote experts and company knowledge bases (i.e., documentation, expert systems, etc.);
  • Remote tool operation. Extending remote diagnostics to allow equipment or software to be run remotely in an engineering or maintenance mode;
  • Spares management. Managing spares inventory, deployment, and refresh stock across one or more factories;
  • Scheduling and dispatch. Ability to accelerate lot cycle times, coordinate reticle and lot delivery, optimize equipment utilization, and dynamically reprioritize activities according to the latest factory events (i.e., equipment down, AMHS down, excursions, etc.) or rebalancing the line to meet changing customer needs;
  • Factory error recovery. Ability to rapidly return the factory to normal operations after an error condition regarding equipment, AMHS, facilities, factory software systems, or out-of-control events; and
  • Recipe and parameter management. Ability to manage recipes, equipment constants, and adjustable equipment parameters to address challenges of real-time manufacturing and reduce factory excursions.

Complex process technology challenges are driving the industry to widely adopt and implement increasingly sophisticated and pervasive process control systems (PCS). These capabilities are needed at both lot and wafer levels in real time. The applications that must be integrated and work harmoniously at both the factory and equipment levels simultaneously include:

  • run-to-run (R2R) and advanced process control (APC),
  • statistical process control (SPC),
  • fault detection and classification (FDC),
  • integrated metrology (IM),
  • sensor integration to support PCS applications (i.e., E54 sensor bus), and
  • yield and excursion management systems.


Figure 1. Some attributes of a leading-edge 65nm or 45nm 300mm factory.
Click here to enlarge image

null

While some, or arguably many, of these capabilities exist today, the infrastructure to seamlessly support them does not. The ability to easily tie multiple sources of data together, cross-correlate data, and rapidly detect excursion patterns or desirable patterns is a challenge. A key issue is that the infrastructure connecting equipment to automation systems was built and designed using 20-year-old technologies, like the Semiconductor Equipment Communications Standard (SECS) and Generic Equipment Model (GEM), which are difficult to configure, do not have adequate data security, and only allow point-to-point connectivity. To enable the required e-Manufacturing and process control systems, the following are needed:

  • Real-time ability to control parameters on a wafer-to-wafer basis (Semi standards E40, E90, and E94);
  • Security for remote connectivity by user, by data type, and for remote equipment operations;
  • Rich, high-quality data sampled at correct rates (up to 10kHz) to enable future advanced PCS application needs;
  • Ability to share data among many applications and with off-line yield and excursion analysis systems;
  • Equipment data acquisition (EDA) standards that enable rich, secure, standard format data for e-Manufacturing through seven Semi standards that are being defined today (see below); and
  • Data exchange between business partners (i.e., wafer, reticle, die, materials, etc.).

The move to 300mm wafers drove the use of nonproprietary and open industry standards for loadports, carriers, and interfaces to support 100% material-handling delivery within the fab. Similar equipment capabilities are needed to support the ability to do R2R process control and fault detection for all for process and metrology equipment. A significant lesson learned from 300 mm is that having third-party products (e.g., modules, libraries, software components, etc.) that enable rapid and accurate implementation of these capabilities and nonproprietary and open industry standards is a fundamental key to success.

e-Manufacturing collaboration

The approach that major industry consortia (Semiconductor Industry Association, ITRS, International Sematech [ISMT], JEITA, and Selete) have followed since the beginning of 300mm is to combine early collaboration on concepts and industry guidance with a plan for standardization of interfaces, accelerating commercialization, and testing for equipment and software compliance.

The International Equipment Engineering Capability (EEC) Guideline Collaboration is an effort to further define requirements via a series of global agreements, guidelines, and nonproprietary and open industry standards that are defined to keep the cost of equipment software for data collection and control as low as possible. Common software and interface requirements from chipmakers reduce the effort needed by equipment suppliers and third-party suppliers.

ISMT and JEITA-Selete have developed EEC guidelines [1] that strive for common interface standards and common software capabilities. The primary need is for access to equipment data. The data from process tools, metrology tools, parametric and final e-test equipment, and in-line defect metrology equipment are all needed.

All this data will be merged to form a more complete view of factory operations. The focus for EEC guidelines includes remote diagnostics, R2R control, FDC, and recipe management. Many IC makers have factory-level R2R and FDC systems. The use of third-party software and the integration of this software into factory systems will require standard interfaces, standardized data messages, and standardized feedback to process equipment (Fig. 2).


Figure 2. Identification of interfaces.
Click here to enlarge image

null

A good example of industry collaboration is the cooperation between ISMT member companies and the supplier community to collectively define remote diagnostics [2]. The fundamental purpose of remote diagnostics is to increase the availability of production and facilities equipment, reduce mean time to repair, improve ramp time, and provide significant reduction in field service resources and costs. The remote diagnostics program provides guidelines that define remote diagnostic capabilities that enable chipmakers and equipment suppliers to share the responsibility for productivity improvements. This includes nonproprietary and open industry standards to access equipment data (Fig. 2, Interface A), the establishment of communication interfaces, the ability to move data externally to the supporting organization (Fig. 2, Interface C), and the development of security guidelines.

These efforts will positively impact the effectiveness of the support offering and ultimately improve factory operations and reduce maintenance costs. It is the intent that the remote diagnostics vision not be constrained by current practical challenges or limitations (e.g., network bandwidth). Early implementations of these capabilities are being provided by suppliers and evaluated via prototyping at ISMT and its member companies. The lessons learned are incorporated into standards development and into the equipment evaluation process.

In addition, equipment suppliers are encouraged to demonstrate prototype implementations of Interface A (Fig. 2) at ISMT. The goals of the EDA Port Prototyping initiative are to verify the requirements described in standards documents, identify probable implementation barriers, and accelerate the industry's learning around remote system diagnostics and the creation and usage of the EDA port.

Nonproprietary, open industry standards

Equipment standards are important to suppliers since a single solution would meet the needs of most of its many customers, thereby reducing development and support costs. This also makes the use of third-party solutions easier and more consistent, which can further reduce cost and speed the release of new equipment. For IC makers, it means that all equipment in a factory communicates in a similar manner. For e-Manufacturing, it also provides a way to introduce mainstream technologies used by other industries into the equipment automation domain.

Some standards are currently available and form the backbone for communication within a 300mm factory. The minimum 300mm standards that must be implemented by equipment suppliers include Semi E5, E30, E37, E40, E84, E87, E90, and E94. The new EDA standards, which support e-Manufacturing, are being built upon these base standards.

Equipment data acquisition

To meet the needs of equipment engineering capabilities, there is a need to improve production equipment data collection in terms of accuracy, security, quality, and timeliness of the data. It also allows the semiconductor industry to leverage mainstream computing technology, such as extensible markup language (XML), web services description language (WSDL), and simple object access protocol (SOAP), to eliminate semiconductor industry-specific technologies. There are seven standards being developed that facilitate these improvements:

  • Common Equipment Model (Semi E120) provides a common standardized description for equipment and enables all of the other standards below. This was approved in October 2002.
  • Equipment Data Acquisition (Semi PR8) is an intermediate specification to create an initial and simple equipment interface for e-Manufacturing that jump-starts the industry using mainstream computing technologies, such as SOAP over Hypertext Transfer Protocol (HTTP). This was approved in December 2002.
  • Equipment Client Authentication and Authorization (Semi Doc 3507) provides security methods to control which applications are permitted to communicate with equipment as well as which equipment services applications are permitted to use intellectual property protection.
  • Equipment Self-Description (Semi E125) provides a means for applications to discover via software the physical structure, data items, events, and exceptions of the equipment. It will enable automated data collection setup for equipment and automated methods for keeping available tool data up-to-date over paper documentation. This was approved in March 2003.
  • Data Collection Management (Semi Doc 3509) provides a means for applications to organize all data needs (trace, exception, event) into logical, named units that can be individually activated and deactivated. Anticipated benefits include simplified setup of data collection and detection of equipment performance issues, among others.
  • Provisional Specification for Equipment Performance Tracking (EPT, Semi E116) defines equipment behavior states and data required to track basic equipment performance.
  • XML Message Structures (Semi E128) provides a standard structure to enable reliable asynchronous exchange of messages using common protocols such as SOAP and XML protocols. The XML Message Structures complement EDA Standards by providing a consistent way to orchestrate required request-replay and event message conversations with mainstream messaging technologies.

Accelerating commercial implementations

Practically speaking, success is measured by the existence of deployed e-Manufacturing solutions that meet the needs of IC makers and suppliers. Ensuring that capabilities are available in equipment and in commercially available building blocks and end-user applications is extremely important to the entire process. Thus, we must all be involved with the following: informing the supplier community of the unity of their customers in requiring these standards; educating suppliers and users about the details of standards to ensure consistent implementations are created; encouraging early implementation through monetary or marketing incentives to equipment suppliers (these implementations identify gaps with standards and enable them to be corrected faster); providing forums for suppliers to ask questions about standards and to gain consensus on interpretations for implementations; developing conformance testing or certification processes to help measure (and thus improve) the quality of implementations; working with suppliers on improving their implementations; and feeding back issues and problems to the standards process to support continuing improvement and clarification.

The real success of these activities occurs when guidelines and standards have been implemented and are being successfully used in factories.

The next steps

The industry has been very successful in its move to highly automated systems that support 100% carrier delivery for 300mm fabs. Continued attention must be given to wafer tracking and control capabilities that are crucial to high-mix, flexible production, especially with regard to compliance with Semi E40, E90, and E94. These are key enablers to more advanced capabilities.

Similar methods must be applied to the e-Manufacturing revolution to get breakthrough capabilities and performance that meet 65nm production challenges and beyond. These capabilities include pervasive remote diagnostics, enabling rich, high-speed data; implementing more advanced process control systems (R2R, FDC, SPC, IM) that leverage this data to meet process needs at 65nm and below; and other productivity improvements to reduce capital and wafer costs. Equipment must provide rich, high-quality data using EDA (E125, doc 3507, doc 3509, etc.) standard formats and behaviors, key to enabling infrastructure and needed on all equipment by 1H05.

e-Manufacturing applications should allow IC makers to pick and implement modular, best-of-breed products from many suppliers without being locked into a specific architecture. It must also include the option of working in a hosted, stand-alone, or embedded environment. To realize these breakthrough capabilities and meet industry challenges, IC makers and equipment suppliers must aggressively finalize standards now, iteratively implement and refine e-Manufacturing applications, and ensure solid, tested, and fully featured capabilities are delivered by 2005.

Blaine Crandell, Texas Instruments, Dallas, Texas
Michael Passow, IBM, Fishkill, New York
Jeff Pettinato, Intel, Chandler, Arizona
Harvey Wohlwend, International Sematech, Austin, Texas

References

  1. EEC Guidelines, www.sematech.org/public
    esources/ediag/guidelines/.eecguide_rev2.5.pdf.
  2. e-Diagnostics Guidebook, www.sematech.org/public
    esources/ediag/guidelines/guidelines.htm.

Blaine Crandell is a solutions architect and distinguished member of the technical staff at Texas Instruments.

Michael Passow manages sensor integration and advanced process control at IBM.

Jeff Pettinato manages an Equipment and Process Control Systems department at Intel.

Harvey Wohlwend is a manager at International Sematech, 2706 Montopolis Dr., Austin, TX 78741; ph 512/356-7536, fax 512/356-7631, e-mail [email protected].