From the keynotes to the sessions to casual conversation, automotive electronics was a hot topic of discussion at SEMICON Europa 2017. That’s likely because Europe leads the world in automotive OEMs and is home to four of the top 10 semiconductor providers in the automotive industry. With over 22% growth in automotive ICs over the past year, it’s definitely something to talk about.
Autonomous Vehicle Readiness
Citing Yole Developpement’s market research, Maria Marced, president, TSMC Europe BV, noted that by 2045 more than 70% of all vehicles sold will integrate autonomous capabilities. Level 2 autonomy is almost here, and by 2050, 5% of cars will be level 5 ready. 2050 is a far cry from Tesla’s aggressive 2019-2020 timeline for its level 4 and 5-ready vehicles, but then again, as Marced pointed out, Tesla has been a disruptor.
Autonomous driving features aren’t the only drivers of increased electronic content. The transition from combustion to electric engines is part of that scenario. Volvo will reportedly transition all its new cars to electronic engines by 2018. By 2025, battery prices will enable a crossover of combustion car. And government mandates to ban combustion engines are in place. In France and the UK, the ban goes into effect by 2040.
Jokingly noting that park assist is her favorite feature enabled by autonomous technology, Marced said that cars are getting safer, greener, and smarter. This presents tremendous opportunities for the semiconductor industry, as the silicon content of an electronic car will be between 8-15X of the silicon content on a combustion car. What’s more, the life cycles of cars will be impacted significantly as autonomous vehicles create a paradigm shift in how we use them, which in turn will push innovation further.
The Impact of the Sharing Economy
Both Marced and Frank Rinderknecht, concept car visionary and founder of Rinspeed, talked about a shift from car ownership to a sharing economy model, thanks to a new generation of adults that view mobility as a service rather than feeding a passion for the open road.
This paradigm shift will take cars usage from 1-2 hours a day to 20 hours a day. That means rather than lasting 9-18 years, cars will only be expected to last 2-3 years. “What do we do with a PC that is nine years old?” asked Rinderknecht.
What do we do indeed? While both Rinderknecht and Marced described this as an opportunity for continued innovation and ongoing need for semiconductor content, it occurred to me that we are creating the potential for mountains of electronic waste in automotive graveyards. Additionally, what impact will the increased usage of the vehicle have on parts reliability requirements? Will cars require more frequent servicing? Will the materials used to manufacture the semiconductor content hold up? Or will requirements be relaxed as cars won’t be expected to have such long lifespans?
I took these questions to the experts.
How Do We Ensure Safety and Reliability?
Christoph Kutter, director of Fraunhofer EMFT, said the solution of what to do with the waste is simple: old cars will be recycled. In general, safety and reliability are paramount, he said. Additionally, we must ensure that fully autonomous vehicles are equipped with redundant and failsafe systems are needed to ensure functional safety of autonomous vehicles, so that in the event of a failure, the car can be pulled over and shut down in a safe way. On the topic of device reliability, he said the focus needs to be on the package, which will fail before the chip.
In his presentation during the Automotive Session, Gerald Dallmann, SGS INSTITUT FRESENIUS GmbH, talked about the importance of performing failure analysis on power devices. Two myths about understanding quality, he said, are thinking a product is safe because it was qualified, and that if the technology hasn’t changed, it’s still qualified. Dallmann noted that automotive technology and product qualification procedures still follows the AEC Q100 procedure, which states that “Components meeting these specifications are suitable for use in the harsh automotive environment without additional component level qualification testing.” He explained that failure rate assessment at device’s end of life requires proven acceleration models for Q100 qualification procedure, but there are no models available for new package types. Dallman offered a laundry list of strategies to prove and improve reliability at a low ppm level and essentially explained that testing alone is not enough to ensure quality. Rather, we need to change from AEC Q100 test-based approach to a predictive approach with failure mode assessment and degradation models.
Benedikt Ernst, of Merck, said that as a materials supplier for automotive electronics, the company is working with its automotive partners to develop reliability solutions holistically. For example, as autonomous and electric vehicles need to function in all sorts of adverse weather conditions, how to make components waterproof is a challenge. “Do we seal the entire device or use a protective coating? Or do we make the device itself waterproof?” said Ernst. “Simply providing the material is not the answer. We work with the customer to provide the best solution.” The company announced its full portfolio of materials designed to meet current automotive electronics requirements.
That said, Ernst noted that there are multiple scenarios to consider when thinking about the shift in auto usage. For example, when people share a car, they handle it differently than if they are the sole owner, and tend to be more careful with it, he explained. Additionally, there will be 1/20th of the cars on the road, which means the economic factors for automobiles will change. Do we reduce car lifespan or do we make it more reliable?
Ernst explained that with regards to qualifications for autonomous vehicles, there are still many open questions, and things we don’t know. Will there be fewer accidents? How much will we be willing to spend on highly redundant systems? Will cars drive themselves in for repairs? Will shared electric cars allow for relaxing charging requirements? Current automotive material standards were set in the 80’s and 90’s and call for 10-year reliability. If the expected lifespan is shortened, do we still need that 10-year timeline? From the perspective of a materials supplier, Ernst noted that it takes five years before the revenue shows for a newly developed material. Can the automotive industry afford this if the product lifecycle is shortened?
Luckily, noted Ernst, we don’t need to think about the full implementation yet, as we still have a long way to go. It will require close collaboration between the automotive OEMs and the suppliers to solve them. All of this, however, spells opportunity for the entire semiconductor supply chain. ~ F.v.T