The Everything New Syndrome

Options are plentiful, but so are the possible unknowns.


Technology is all about the latest features, the fastest processing, with the lowest power. While that sounds great in marketing pitch, any or all of those factors don’t necessarily equate to a better product or long-term user satisfaction.

There’s a reason semiconductor companies are conservative by nature. They want to know that when they spend tens or hundreds of millions of dollars on a chip design, it will work well enough throughout its expected lifetime. The problem is those expected lifetimes are getting longer, and the chips themselves are now collections of lots of different devices and IPs based on unique architectures designed to maximize performance per watt.

Add to that some brand new materials, and the fact that these chips are expected to work seamlessly with other systems, which may be developed independently by other companies using entirely different processes. All of this adds up to some serious challenges, made worse by the fact that the number of unknowns continues to grow, spreading across a device and frequently across multiple devices.

To say this makes test, inspection and metrology more difficult is an understatement. All three technologies are essential for reliability, but they all work best where there are known issues and proven results. Understanding how they will impact an entirely new device or system in the field, including unknown interactions over time, is yet to be proven.

There are several big shifts underway that magnify these issues even further. First, in the past, the most complex chips were used under fairly constant and well-monitored conditions. Inside of data centers, there always have been acceptable operating temperature ranges. The same is true for a smart phone, which left on the dashboard of a car in direct sunlight will alert users that it needs to cool down before it can be used. But that doesn’t work for many other edge devices, some of which may be used for safety-critical or mission-critical applications.

None of this was a concern until recently. As advanced chips find their way into cars, drones, robots, and even aerospace and military applications, conditions will vary greatly. Trying to understand what constitutes a latent defect versus a real one in the field relies heavily on simulation models as well as some educated guesswork, which are unlikely to include all the possible scenarios under which things can go wrong.

Second, heterogeneity is great from an architectural standpoint, but understanding how different components might interact when they are actually packaged together is challenging. In addition to the usual physical effects — variation, heat, electrostatic discharge, electromigration — and their impact on premature aging through processes such as electromigration, different components can shift or be damaged during the packaging process. This is particularly true as packaging density increases and line widths shrink, making delicate bumps and pillars more susceptible to any jarring or irregularies, such as nanosized particles left over during polishing and cleaning.

Third, new materials have properties that are in high demand for a variety of purposes, including everything from films and trench lines to different substrates. While some of these have been in use for many years for specialized applications, expanding that use across a broader mix of applications, in concert with other materials, and in new applications that have higher demands for utilization and/or reliability will add new unknowns into the mix — even with well-understood materials.

While existing test, inspection, and metrology equipment can identify most or all of these potential problems, that may only happen if engineering teams know what they’re looking for, including what to test for and what’s considered acceptable, and how critical various measurements need to be. But given the industry’s focus on customized solutions and new applications, that becomes difficult at best, raising the risk of failures in places where no one was looking or where they least expected.

Leave a Reply

(Note: This name will be displayed publicly)