中文 English

Extreme Quality Semiconductor Manufacturing, Part 1: Automotive

Automotive industry trends and the innovations in defect control needed to meet stringent quality requirements.

popularity

By Ben Tsai and Cathy Perry Sullivan

Across the full range of semiconductor device types and design nodes, there is a drive to produce chips with significantly higher quality. Automotive, IoT and other industrial applications require chips that achieve very high reliability over a long period of time, and some of these chips must maintain reliable performance while operating in an environment of temperature and humidity fluctuations, vibrations or other harsh conditions. Leading-edge semiconductor ICs with ≤5nm design nodes, gate all around (GAA) or other 3D architectures, and 1000+ process steps require careful control of variability to achieve power and performance targets. Extreme quality semiconductor manufacturing innovations are essential to achieving variability and defectivity control so that fabs can produce chips that meet strict reliability and performance standards.

This article focuses on automotive industry trends and innovations needed to meet the stringent quality requirements for automotive ICs. The second article in this series will look at quality trends for future node semiconductors.

Automotive
The automotive industry continues to integrate more complex automated driver assist, safety and infotainment features, and to move towards electric engines and full autonomous driving capability. As advancements in connectivity, electrification and autonomous driving are made, the number of semiconductor chips in a vehicle increases. Depending on the make and model, a new car can have 6000 – 10,000 chips per vehicle,1,2 with the electronics subsystems accounting for ~35% of the vehicle’s cost.3 These chips include processors, memory devices, RF devices, LEDs, power devices and MEMS, covering design nodes from 4Xnm to <1Xnm that are produced in 200mm and 300mm semiconductor fabs. The automotive industry’s extensive use of semiconductors, and reliance on semiconductors for future innovation, is reflected in the fact that automotive is the fastest growing semiconductor segment at >2x the industry rate.4


The semiconductor content in vehicles is increasing to support electrification, connectivity and autonomous driving. (Image: KLA Corporation)

With thousands of chips in a vehicle, many in mission-critical functions, there is a new focus on semiconductor quality. If just one critical chip fails in the field, it can cause expensive repairs, damage to the automakers’ reputation, and even personal injuries or loss of life. Fundamentally, an in-field chip failure is a reliability problem. When the chip moved from the semiconductor manufacturer into the supply chain, it was functional and had passed standard performance and quality control tests, including electrical tests and burn-in tests. Yet, it was unable to perform reliably under the fluctuating operating conditions (heat, cold, vibrations, snow, rain, etc.) of a vehicle. Unlike consumer semiconductors used in applications such as smartphones, automotive semiconductors need to meet higher reliability standards in a variable environment over an extensive period—5 to 10 years or even longer. These criteria are driving the need for extreme quality control in semiconductor manufacturing.

Latent Defects: Up to now, the primary focus of automotive semiconductor manufacturers has been on becoming better at sorting out chips that have a high likelihood of reliability problems in the field, without wastefully throwing away too many good chips. In other words, fabs are now optimizing for reliability in addition to optimizing for yield. The chips with a higher probability of reliability problems are more likely to have latent defects. Latent defects are typically process-related defects that have a size or location that doesn’t kill the die, or that lie in an untested area of the die.5 The operating environment of a vehicle activates the latent defect, causing the chip to malfunction or fail.


The size or location characteristics of latent defects (left) do not cause the chip to fail. Within the extreme operating environment of a vehicle (heat, cold, vibrations, humidity), a latent defect can be activated (right), causing the chip to malfunction or fail. (Image: KLA Corporation)

An effective way of finding and removing chips with latent defects is to decrease parametric and defectivity margins. Decreasing parametric margins means requiring that the chips not only function, but also operate within a tighter parametric window. Decreasing defectivity margins means setting the acceptable defect size to be smaller than what has proven to be yield-killer defects. To find more subtle parametric variations or smaller defects, fabs need to implement higher sensitivity process control strategies 6 – either by increasing the process control tool’s recipe sensitivity or by utilizing inspection and metrology systems designed to detect smaller defects or variations. With more capable process control systems, automotive fabs can detect, monitor and control the latent defects that might otherwise cause premature chip reliability failures.

Fab Process Quality: Automotive semiconductor manufacturers are adopting an increased quality mentality in order to prevent chip reliability issues in the field. For example, continuous improvement programs reduce the random defectivity introduced by process tools, while more stringent characterization and monitoring strategies ensure that the process tools are in top operating condition.7 Rather than exclusively focusing on optimizing fab processes for yield, IC manufacturers need to shift to running processes under the best possible conditions to achieve reliability standards. This quality mentality may increase fab costs in the short term but will result in long term savings from delivering the higher reliability chips required by automakers.

Part Average Testing: Beyond optimizing fab quality by reducing overall process defectivity, automotive fabs can benefit from implementing new die screening methods to prevent potential reliability failures from escaping the fab. A new inline technique, called I-PAT (Inline Defect Part Average Testing), uses inline defect information to identify die at risk for reliability problems in the fab.8 I-PAT looks at the defect population that results from stacking defects detected at multiple, critical process steps into one composite inspection result, thereby revealing die that have high defectivity when all process steps are taken into account. Die with defectivity levels that lie outside the normal distribution of the population have a higher probability of latent defects and can be excluded from the automotive supply chain.


Automotive process control and die screening methodologies that help automotive semiconductor fabs achieve Zero Defect standards. (Image: KLA Corporation)

Future Innovations: As automotive electronics continue to increase in complexity, it is possible that the semiconductor industry will introduce changes in automotive chip architectures to ensure reliability. As an example, consider redundancy, which is necessary for critical automotive subsystems in case a failure happens.9,10 If the semiconductor chips are the most critical part with the most risk for failure, then instead of relying on one processor to make the critical decisions, it may be worthwhile to build three processors into the chip that all operate at the same time. The results of all three processors would be used for critical decisions – essentially by voting. Then, if one processor is hit by a cosmic ray particle causing a bit flip that gives a wrong answer, or if a latent defect is activated causing a processor failure, the other two processors will still give the correct answer. With today’s low transistor cost, using a smaller design node or having a slightly larger chip size creates the possibility of having built-in fault tolerance without a drastic increase in chip cost.

Simulation software tools for automotive electronics could also integrate more capabilities to enable design for reliability. These automotive simulation capabilities could start in-house and then develop into an independent EDA-type industry for automotive electronics in the future.

Extreme quality manufacturing for automotive electronics is still at an early stage. Going forward, the automotive semiconductor industry will develop new methodologies for traceability, with an abundance of data generated during the manufacturing process, to help remove risky die from the supply stream and to help drive the process improvements needed to eliminate latent defects. In doing so, semiconductor fabs are likely to differentiate themselves by establishing their own extreme quality manufacturing processes for automotive ICs, while automakers will differentiate themselves by establishing trusted semiconductor supplier partnerships that help produce more reliable electronics and safer vehicles.

References

  1. https://www.audi-mediacenter.com/en/techday-quality-9568/semiconductor-lab-9572
  2. Audi AG, Semiconductor Strategy PSCP, SEMICON Korea, January 2018.
  3. https://www.statista.com/statistics/277931/automotive-electronics-cost-as-a-share-of-total-car-cost-worldwide/
  4. Source: Gartner
  5. Price, Sutherland and Rathert, “Process Watch: The (Automotive) Problem With Semiconductors,” Solid State Technology, January 2018.
  6. Price, Sutherland, Rathert, McCormack and Saville, “Process Watch: Automotive Defect Sensitivity Requirements,” Solid State Technology, August 2018.
  7. Price, Sutherland and Rathert, “Process Watch: Baseline Yield Predicts Baseline Reliability,” Solid State Technology, March 2018.
  8. Price, Sutherland and Rathert, “Process Watch: A Statistical Approach to Improving Chip Reliability,” Semiconductor Digest, September 2019.
  9. https://www.bosch.com/stories/redundant-systems-automated-driving/
  10. https://blog.nxp.com/automotive/automotive-functional-safety-the-evolution-of-fail-safe-to-fail-operational-architecture

Cathy Perry Sullivan, Ph.D., is a technical marketing manager at KLA.



Leave a Reply


(Note: This name will be displayed publicly)