Customer Story

Analytics speeds time to market in high-tech manufacturing

Semiconductor manufacturer SMIC observes dramatic time and monetary savings


ChallengeDeliver advanced technical capability among an audience of varying technical abilities.
SolutionInstall JMP and JMP Pro, a product that meets engineers’ demand for an easy-to-learn, easy-to-use tool while simultaneously offering powerful graphics and analytic functionality.
ResultsSMIC reports improved quality, speed to market, time and monetary savings.

Competition is fierce in the semiconductor industry, especially in integrated circuit manufacturing. Success depends entirely on speed to market for Semiconductor Manufacturing International Corporation (SMIC), one of the leading semiconductor foundries in the world and the largest and most advanced foundry in mainland China.

"We’re developing a new generation of technology for semiconductor products, and trying to develop it – from researching the technology to fabricating the products – in the shortest cycle time," explains Siyuan Frank Yang, Corporate Member of Technical Staff at SMIC.

Speed in data analysis saves money and time

To stay ahead in a fast-paced industry like high-tech manufacturing, SMIC sought to simplify its experimentation and data mining processes. Complicated and expensive experiments had slowed progress and consumed huge costs.

The end goal? To develop a tool that delivered an advanced technical capability while also fulfilling engineers’ criteria: The software must be easy to learn and use in everyday processes.

So SMIC leadership tapped JMP to resolve obstacles like the technical learning curve without sacrificing elevated analytic functionality. "SMIC continues to innovate and succeed because we’re open to new technologies like JMP," Yang says.

Today, Yang reports that between 80 and 100 SMIC engineers are using JMP in design of experiments (DOE), modeling and fit models on a regular basis within the organization. "We’re even using JMP for advanced DOE, prediction for some circuit performance using WAT data before circuit performance is measured and analysis of highly correlated and highly dimensional data."

Collaboration drives innovation

One of SMIC’s corporate values is innovation. SMIC uses innovation to continuously and actively undertake product technology research and development, increase productivity and optimize work procedures.

One of the ways SMIC delivers on its innovation value is through a research and development collaboration group using JMP Scripting Language (JSL) to code a custom platform for multiple users to perform their daily device data analysis. "This group’s collaboration leads to greater innovation. We find that the more our engineers know about JMP – and the more they share their findings with each other – the more frequently they use JMP and the better use they make of it," Yang explains. "Most engineers here prefer JMP because it is easy to use and has powerful graphing techniques."

SMIC encourages such collaboration because it improves quality and consistency – another of its corporate promises. By building comprehensive quality and reliability assurance and control systems into its processes and services, from technology development to production, SMIC ensures a consistent level of quality and flexibility for its customers.

SMIC decides to add JMP® Pro to its analytics toolkit

With demonstrated ROI from SMIC’s new approach to DOE, there was little hesitation when Yang suggested SMIC also adopt JMP Pro for more advanced analyses. “We wanted additional advanced techniques for big data analysis,” Yang says. JMP Pro provides all the superior capabilities for interactive data visualization, exploration, analysis and communication that are the hallmarks of JMP, but with an additional library of expert analytics resources.

“There are many very useful features in JMP Pro — in particular, I’ve found some regularization methods very useful in dealing with highly dimensional and highly correlated data,” he notes.

In addition, JMP Pro offers a multitude of sophisticated techniques: predictive modeling with cross-validation using a number of different methods, modern modeling techniques, model compari­son and averaging features, advanced multivariate techniques, reliability block diagrams, covering arrays, mixed models, uplift models and advanced computational statistics methods.

The benefits of analytics trickle down to the customer

SMIC touts marked savings from equipment performance improvement with the use of the DOE capabilities of JMP, and time savings in data processing, data analysis and graphing.

“We have saved both money and time by using JMP,” Yang reflects. “The outcome—and the motivation for all of this — is that our customers benefit. The quality of our products has improved, and our products are now released from R&D stages to manufacturing much faster.” And Yang suggests that using analytics wisely can fuel growth, speed and efficiency. “The more our engineers know about JMP, the more it pays off for the company and for our customers in the long run,” he says.

Dr. Siyuan Frank Yang, an industrial statistician, is responsible for industrial statistics applications for SMIC in Shanghai, China. He leads metrology methodology development, statistical consultation and writing, and provides statistical training to SMIC companywide. Yang’s earlier work at SMIC was in the area of processing development and industrial statistics. He works alongside an MS graduate in applied statistics who focuses on JMP scripting language (JSL) programming and some statistical training.

Back to Top