skip to main content

Service 3.0: Partnerships Lead to Yield Improvements

There will be more changes in device size, complexity, materials and production techniques over the next 5 years than the industry has seen in the past 15 years.

With the average cost of a modern fab running about $6 billion, chipmakers need to churn out revenues of a staggering half billion dollars monthly to realize a return on their investment. No surprise, then, that successful semiconductor manufacturing is ultimately all about achieving and maintaining optimum device performance and yield- even as the sophistication of devices and the complexity of manufacturing processes skyrocket.

In the context of this business environment, the role of “service” has evolved dramatically – from the days of traditional maintenance support to a new era where the quality of a fab’s output is a major goal. Dana Tribula, chief strategy officer and CMO for Applied’s Global Services business, calls this new service paradigm “Service 3.0.” (See the blog Evolution of the Semiconductor Service Model)

In recent years, production issues stemming from electromechanical failures have diminished as system reliability has increased. Instead, the main issues impacting yield are on the wafer and are related to process control. These have become much more subtle and thus, harder to identify and resolve.

“The good news is that no manufacturing process on earth is more closely monitored than the production of semiconductors,” said Sanjiv Mittal, general manager of Services & Consulting for Applied Global Services. “Every day a fab runs millions of unit processes through hundreds of machines, each of which is storing information from hundreds of sensors. Add in defect and electrical inspection, and this is clearly Big Data.” Hidden in that data are the key links between individual process parameters and chip performance.

It takes deep expertise to extract meaning from voluminous fab data, and so a truly effective Service 3.0 capability builds on extensive equipment and application knowledge, along with experience solving the many types of real-world problems encountered on a daily basis in every type of semiconductor production environment imaginable. For example:

  • One 300mm customer experienced a statistically significant difference in the performance of devices produced in different chambers, which in turn resulted in yield loss. There was no tool malfunction and the customer’s inline process data didn’t even show statistically significant variations among chambers. Partnered with Applied’s FabVantage consulting team, the customer achieved targeted yields through advanced data collection and analysis techniques. These led to an understanding of the problem and to the implementation of different process control limits that boosted yield. (See “Software is Key to Effective Yield Management”)
  • Another 300mm customer was experiencing a film uniformity problem among different chambers, in a production environment where there was no equipment malfunction nor control problem. The FabVantage team conducted a recipe audit and found several important differences between Applied’s best-known method (BKM) recipe and the customer’s, along with important secondary recipe factors. Together, the customer and Applied developed an optimal recipe configuration to meet the thickness uniformity specification across chambers. (See “Recipe Optimization Improves Film Uniformities”)

“Service 3.0 is Applied’s contemporary model for service support,” Mittal said. “Although yield issues themselves can be fiendishly complex, the aim of our efforts under the Service 3.0 model is straightforward: to serve as a trusted partner working hand-in-glove with customers to help them identify, understand and resolve whatever issues negatively impact their ability to quickly, predictably achieve and sustain yield.”

A former fab manager himself, Mittal said that the opportunities collaborative partnerships can bring to yield improvements are boundless. “We bring the resources and experience needed to help customers understand what a specific problem’s ‘signature’ looks like in the data, and how things can be changed to improve on-wafer results, and hence yield, as quickly as possible,” he said. “Some examples are yield-driven fault detection, yield-enabling new parts, and identification of failure modes that impact on-wafer results as well as cost and productivity.”

LOOKING FORWARD
While equipment repair, maintenance, parts and upgrades will always remain important elements of service support, helping customers achieve and sustain yields now has become a major objective. Accomplishing that will increasingly depend upon the ability to accurately characterize, model and simulate the controls, operational states, fault conditions, failure modes and other aspects of semiconductor-fabrication equipment, based on a constant stream of information gathered by embedded sensors.

In short, that’s what Applied means by Service 3.0.