Get Ahead of the Curve with Faster Ramps, Better Yields
By Gary Dagastine
Advanced data gathering and analysis, local and remote service and engineering expertise, global supply chains and integrated automation solutions are among the essential tools for enabling both greenfield and existing fabs to quickly achieve production goals, maximize profitability and capitalize on market opportunities.
Semiconductor wafer manufacturing, assembly and test have never been so complex.
At the leading edge, integrated circuits now have architectures so highly advanced that even slight variations in their manufacturing processes can lead to poor device performance and low yields. Suboptimal fab operation and supply chain inefficiencies, meanwhile, lead to more scrap, higher costs, and to the misuse of resources including the most important one of all – time.
It’s not easy at the trailing edge either, because the need to take advantage of new, pre-validated manufacturing applications for greater flexibility has never been greater. Billions of electronic devices must be produced in just the next few years for evolving applications like autonomous vehicles, Internet of Things (IoT) systems, artificial intelligence, 5G connectivity, augmented/virtual reality and others.
An added dimension is that in many locations, the expertise, infrastructure and proven applications necessary for timely success are very thin—whether it’s establishing a greenfield fab or upgrading an existing operation.
For example, China has set a strategic goal to greatly increase self-sufficiency in semiconductors under a “Made in China 2025” program. However, setting up these greenfield fabs quickly to serve tight market windows while meeting device, output and economic requirements are major challenges.
It’s difficult to see how all of this can be accomplished without the use of advanced digital analytics, expert local- and remote support, global supply chain resources, and fully integrated automation solutions to enable effective computer-integrated manufacturing (CIM). A way to think about the problem analytically is to visualize two curves (see figure 1).
Figure 1. Ramp curve for technology change (left); and operational characteristic curve for output. These curves help manufacturers see how to enhance performance by repeatedly adjusting their operations for better results.
The curve on the left represents the optimal progression of a fab’s ability to manufacture, or ramp, a given semiconductor technology. It shows yield/output versus time across all phases of manufacturing: rising steeply through startup and initial ramp; high and stable during high-volume manufacturing; and finally structured for cost control as the technology becomes mature.
The curve on the right shows a factory’s manufacturing output, or operational characteristics (OC). The OC curve compares the time it takes to “flow” a product through a factory to the overall percentage of factory utilization.
These curves are idealized because in reality, progress isn’t achieved so smoothly. Technology ramps don’t always go as planned, learning takes place continually, results must be evaluated along the way, and so forth, so many discrete bursts of analysis and readjustment are needed to bring manufacturing performance closer to the ideal.
Together, these curves are useful guides to understanding how to achieve maximum flexibility with a given set of business priorities and manufacturing resources. Fully integrated automation solutions and technology-enabled services—including secure data gathering and analysis for remote support—help manufacturers do this faster and better, and with significant cost improvement.
For additional information contact firstname.lastname@example.org.
See related articles in this issue of Nanochip Express: “Computational Process Control Solutions to Serve Fabs in China,” and “Smart from the Start: Fully-Automated Fabs Deliver Competitive Advantage.”