Ensuring Device Reliability in Automotive Electronics

More than 90% of all traffic deaths are caused by human error.¹ I’m sure you’ve all heard this statistic repeatedly over the past few years. After all, this 2016 finding by the US National Highway Traffic Safety Administration (NHTSA) is fueling the semiconductor industry’s mission to develop the automotive electronics needed to launch autonomous vehicles of various levels over the next decade.

At the same time, global climate change efforts, with many countries committing to carbon neutrality by 2050, means there is a big push, particularly in China and Europe, to replace internal combustion engines (ICEs) in automobiles with electrified vehicles (EVs), which produce zero direct emissions and are believed to produce fewer lifetime emissions than their ICE counterparts.²

Driving Semiconductor Value in Automotive Electronics

These two driving factors (pun intended), combined with the connectivity and advanced infotainment systems in vehicles, are increasing the percentage of semiconductor content on a per-vehicle basis in cars. The average semiconductor value in cars is expected to increase to $576 by 2024.³

That number is considerably higher for EVs, averaging $900 for hybrid and $900 for full-time EVs.⁴ According to the latest automotive market reports, the automotive semiconductor market will grow at an 8.3% compound annual growth rate (CAGR) from 2018-2027, totaling $76.3 billion in revenue by 2027.⁵

While those numbers are enticing, we need to remember that device reliability is critical to ensuring that autonomous vehicles and EVs actually help save, rather than cost us lives, while also reducing our carbon footprint.

There’s no doubt that there is a lot at stake here for the semiconductor industry. That’s why SEMI has established the Global Automotive Advisory Council (GAAC), to get automotive OEMs, Tier 1s, device makers and equipment and materials companies to connect, collaborate and innovate along the global automotive electronics supply chain.

I recently had the opportunity to speak with representatives of two of the member suppliers about what this means for their respective companies: Oreste Donzella, KLA, and Antoine Amade, Entegris. Both companies build and sell tools that have an impact on device reliability. KLA’s metrology and process control systems are used to detect device defects, and Entegris’s contamination control systems are used to prevent the defects from happening in the first place.

Smartphones on Wheels

We often hear, in reference to autonomous vehicles, that they are becoming a smartphone on wheels. I find this to be misleading because if your smartphone fails, it doesn’t necessarily mean the difference between life and death. You just go buy a new phone.

As electronics begin to overtake the mechanics in a vehicle’s operation, it is changing the way cars are built, as well as the importance of overall device reliability. According to Amade, by 2030, 50% of the cost of the car will be chips related, and 80% of innovation by carmakers is driven by semiconductors.

Not only that, but as Donzella explained to me, the technologies required are based on the latest and greatest technologies – artificial intelligence (AI), advanced memory, and 5G – that must all be reliable in harsh conditions. Up until now, automotive electronics contained components that were 5-10 years behind that of smartphones and lagged two to three technology nodes behind.

Automotive OEMs, who are used to working with mechanical issues, now have to understand the complexity of software and computer chips. And this is turning the automotive electronics supply chain on its head

Reliability Starts with the Chip

Both Donzella and Amade talked about the difference in reliability between a PC or smartphone, versus an automobile. Unlike with your phone or PC, “You can’t just reboot your car while you’re driving,” says Donzella.

“If chips related to the infotainment system in your car fail, its not a big deal,” says Amade. “but if it’s related to the radar, LiDar, AI, or computing data, you’ll have a problem because it will compromise the safety of the vehicle.”

And safety is what we’re here for, is it not? That’s why the industry is calling for zero defects.

The New Collaborative Approach

Amade breaks it down into killer defects and latent defects. Killer defects impact yield, because they are caught before the die is put into a system, he explained. Latent defects impact reliability, because they are not easily detected and fail in the field. “50% of failures are related to latent defects,” he said. “They don’t measure out at the end of the line.”

Often, noted Amade, device manufacturers are faced with a dilemma. Do they invest in metrology tools that do a better job detecting latent defects; a new process tool that will prevent defects from occurring; or a contamination strategy system to remove the source of the defects?

The real answer, it would seem, is an approach that takes all three into consideration, and that’s where the GAAC comes in. Developing a full ecosystem around the carmaker will make it possible to find latent defects, better understand what causes them, and how to put contamination control systems in place to reduce and possibly prevent them from going forward. Entegris calls the strategy the New Collaborative Approach. In the following video interview, Amade provides more details.

Inline Part Defect Average Testing

Even with superior contamination control, metrology and inspection are a critical part of achieving zero defects. Donzella talked about an instance when a failure in the field was traced back to a single die that had passed all the tests but still failed.

He explained that while latent defects are difficult to detect, they generally are similar to killer defects; differing only in size and proximity to design features. Total defectivity can be used to estimate the latent defectivity of a wafer. That’s why KLA instituted its Inline Part Average Testing (I-PAT), which, as Donzella explained, uses inline defect data to selectively ink-off die that has an elevated risk of latent reliability failures.

The acquisition of Orbotech has put KLA in the unique position of being the first to take metrology from the front end, through the package to the printed circuit board, says Donzella. Data-driven informatics makes it possible to share metrology, defect inspection and process data from tools across the value chain.

“KLA has developed machine-learning-based software to screen potential reliability failures at the source, to prevent the chip from getting in the car,” explained Donzella, In a recent panel discussion during CES 2020.

Donzella views SEMI’s GAAC as a way to bring the entire ecosystem together, including the OEMs, semiconductor manufacturers, and suppliers, to develop standards and methodologies that will strengthen process control inline to ensure end-to-end reliability.

“Tesla changed the way we see the automotive industry,” he said. The vertically integrated industry is changing. Success will require collaboration across the value chain.

References

Reducing Pollution with Electronic Vehicles, Office of Energy Efficiency and Renewable Energy https://www.energy.gov/eere/electricvehicles/reducing-pollution-electric-vehicles

B. Brown, Evidence Stacks up in Favor of Self-driving Cars in 2016 NHTSA Fatality report, Digital Trends, October 6, 2017 https://www.digitaltrends.com/cars/2016-nhtsa-fatality-report/

IHS Markit, Automotive Semiconductor Tracker, Q2 2019 as reported in Entegris, VW, SEMI, A New Collaborative Approach to Defectivity Challenges in the Automotive Industry, Zero Defect Conference, 2019