Bijan Moslehi, PhD, is chief technology officer and senior vice president, semiconductor technology research, for the Noblemen Group (www.noblemengroup. com), a boutique investment banking, strategic advisory, and business development firm based in San Diego, Silicon Valley, and Portland, OR. Moslehi has some 20 years' experience working in the semiconductor and semiconductor equipment industries, including stints with Hewlett-Packard, VLSI Technology, Philips, National Semiconductor, Fairchild Semiconductor, Applied Materials, KLA-Tencor, and Mattson. He can be reached at bmoslehi@noblemengroup.com.

Semiconductor manufacturing technology has entered a new and exciting era that promises to provide significant productivity and cost benefits. But this new age of chipmaking will have its share of severe challenges as well. Over the past few years, there has been slow but steady progress toward e-manufacturing, which encompasses advanced process control/advanced equipment control (APC/AEC) and e-diagnostics. Many technical, operational, economic, and business factors are driving this trend. Paramount among these are shrinking process windows and process control requirements for critical processes at the 90-nm technology node and beyond, the transition to 300-mm wafers and the challenges of processing the higher-valued substrates, rising manufacturing costs, and the quest for higher productivity.

The benefits of e-manufacturing have been well documented, widely recognized, and proven. They include significant reductions in costs, cycle times, process variances, rework rates, scrap, test wafers, and tool downtime. Other benefits are increased fab throughput, yield, device performance and speed, process capability (Cp and Cpk), tool uptime, tool utilization, and speed of capacity/yield ramps. Optimized tool-to-tool matching, work in process (WIP), tool and process qualifications, and preventive maintenance (PM) schedules have been added advantages.

Various APC point solutions have been developed and implemented by IC manufacturers, focusing on feedback and feedforward run-to-run (R2R) control as well as fault detection and classification (FDC). As defined by The International Technology Roadmap for Semiconductors (ITRS), R2R control maintains product quality by adjusting critical process parameters that control systematic tool and process variations and excursions (i.e., course corrections). With FDC, sensors respond to equipment problems by identifying faults and triggering action for equipment repairs or PMs (i.e., tool health monitoring and fault management). Integrated metrology sensors have been pushed for real-time APC with wafer-to-wafer process control and FDC with real-time fault detection and management capabilities.

Adoption of these new concepts requires a comprehensive understanding of the needs, costs, and benefits on an application-by-application basis. In theory, all processes could potentially benefit from APC, but it is prudent to initially focus on a few critical applications that absolutely require such controls. So far, APC has been successfully and broadly incorporated with integrated metrology primarily in the highly variable CMP process. Over the next few years, advanced controls will be applied to critical steps in lithography (CD, overlay), etch (CD, gate resist trimming, chamber matching), and copper damascene interconnect (resistance control), particularly in 300-mm fabs and other plants running sub-90-nm processes.

The drive toward e-manufacturing is complemented by advances in automated product flow and management. A new generation of fabwide interbay and intrabay automated material-handling systems has been introduced into 300-mm fabs. They offer direct tool-to-tool delivery capability with distributed local buffers and are supported by advanced real-time material control systems software. Once integrated into a fab's computer-integrated manufacturing network, such automation systems will essentially turn future 300-mm facilities into giant virtual processing clusters. These capabilities will create intriguing possibilities that previously have been practically impossible. Unique opportunities will arise for innovative fab and tool layout designs, novel methods of managing the product and process flow, and true facilitywide integrated yield management and real-time APC/AEC systems. One example is the ability to link competitors' tools into modular virtual clusters, such as the photo-metrology/etch-metrology/ash-wet clean-metrology sequence.

However, the goal of a "lights-out fab," which is supposed to eliminate people in fabs, remains premature and unrealistic, and will not be achievable in the foreseeable future. Many elements of the dispositioning and decision-making process can be automated in response to an out-of-control event. However, rigid automation will not be as adaptable as fab engineers and operators, who interpret complex fab information and process data, and decide the proper subsequent course of action. In addition, the reliability of current tools and fab systems is far below what would be required to even start to think about the notion of a true lights-out fab.

Despite its significant cost and productivity benefits, e-manufacturing still has a long road ahead. Even with recent limited deployments, the general industry perspective is that e-diagnostics has yet to deliver on its promise. The complexities of e-manufacturing technologies, combined with infrastructure gaps and strategic issues, have led to major implementation delays and other setbacks. Slow progress in developing standards and concerns about data and IP security have greatly contributed to these interruptions. Current fab tools lack the necessary communication links and access to internal tool and process data for effective APC implementations. In fact, some tools still fail even standard SECS-GEM tests. There is an urgent need for timely plug-and-play equipment communication and interface standards with an open architecture. Improved data quality, data encryption, firewalls, and virtual private networks (VPN) are needed for resolving data and IP security issues.

Substantial performance improvements in the CMP module have led to a strong interest in the implementation of real-time APC using integrated metrology in lithography and etch. However, current applications are still primarily limited to macrodefect inspection and resist-trimming steps. Compounding the difficulties are the very slow adoption of integrated metrology and its small niche-market status. Many suppliers have yet to achieve an acceptable return on investment for their years of effort in this area.

In fact, the broader implementation and deployment of integrated metrology has effectively been pushed out to beyond the 90-nm node. Integrated metrology systems that meet the demanding sensitivity, throughput, reliability, and performance requirements, under such harsh tool and process environmental conditions as vibration and temperature, are still immature and in many cases need to be proven and field tested. In addition, current sensors do not address the needs of many new and emerging materials and processes.

Difficulties and problems with designing integrated metrology into process tools have also taken several years to resolve. Some OEMs have chosen to offer an open tool concept, providing a standard interface that accommodates integrated metrology from different suppliers. Other toolmakers use a closed architecture. The ongoing battle for the "ownership" of the format, flow, and management of the data among tool, sensor, and software solution suppliers has had implications for fabwide, real-time APC strategy. The prospect of working with two or more suppliers to maintain a tool goes against the desires of most fabs, which prefer to deal with a sole vendor. The integration and interfacing of various components often have been very painful.

For e-manufacturing to be successfully implemented, an open fabwide APC framework for plug-and-play; availability of and compliance with standards; suitable tool and process sensors; access to relevant sensor data; and seamless integration of hardware, software, and tool/process models must take place. The entire chipmaking community needs to get behind the ongoing efforts by those industry groups working on these issues, such as International Sematech's e-manufacturing/e-diagnostics initiative and SEMI standards.