Editor’s Note: This article was previously published in EET China
In Chinese, the use of the word “cooking heat” is not limited to the kitchen; it can also be used to describe someone’s character and maturity.
This also applies to the semiconductor manufacturing industry.
Although a tiny chip looks quite simple, it embodies volumes of scientific knowledge. Only those who truly understand manufacturing processes and application principles know how hard it is to produce a chip, and can appreciate the beauty behind paying attention to small details.
Thanks to the unstoppable advancement of semiconductor technology, many incredible automotive functions involving semiconductor technology are being developed; for instance, advanced driver assistance systems (ADAS) are currently paving the way for self-driving cars.
Generally speaking, from now until 2023, the auto applications semiconductor market is expected to grow at a 7% compound annual growth rate, and the market’s value will increase from $35.0 billion to $54.0 billion. Thanks to the impetus from ADAS/self-driving/in-vehicle infotainment (IVI)/electric vehicle powertrain/safety applications, the value of semiconductor chips in every car is projected to rise from $375 in 2017 to $613 in 2025. During this period, the value of the ADAS applications is expected to surge, with an estimated CAGR of 19%.
But in spite of this situation, the vast majority of people are unaware of the increasingly close connection between auto electronics and semiconductor foundries.
As for GF’s auto electronics business, the AutoPro™ service package is a critical element. This service package offers the experience, quality, and reliability services for all GF’s automotive technologies. As a result, the service package can satisfy the automotive industry’s strict quality and reliability requirements, and help car manufacturers to use the power of semiconductors to enter the new “smart Internet” age.
The importance of the AutoPro service package solution lies in the way it enables all of GF’s worldwide fabs, including Dresden, Germany; Malta, New York; Singapore; and Chengdu, China, to provide modularized platforms that have passed auto specification certification for various types of automotive clients regardless of which processes they’ve selected to use (e.g., Singapore’s mainstream 180nm, 130nm, 55nm, and 40nm processes, Malta’s 14LPP/12LP/7LP FinFET, or Dresden’s 22nm FD-SOI technology).
Not surprisingly, automotive makers have even higher quality and reliability requirements than clients in other markets. That is why there is a significant importance of having IATF16949 certification.
IATF16949 certification represents confidence that the entire production process is maintained in a controllable, traceable state. It also guarantees that auto-grade IC production, testing, and screening processes have zero-defect status, and is therefore an essential indicator for automotive clients.
GF’s Dresden Fab 1 was recently completed and received its first full-scale IATF16949/ISO9001 certification, which indicates that the plant’s quality management system complies with automotive production requirements, and motor vehicle clients can obtain automotive-specification IC products from GF’s platform.
Unlike other foundries, GF has entered both FD-SOI and FinFET areas. GF’s 22FDX®, which is part of the AEC-Q100 automotive standard, has already achieved certification, and can satisfy the strict quality and performance requirements of the motor vehicle market.
We have consistently felt that the cost and complexity of the mask process in the production of 22FDX are significantly lower than in the 14nm FinFET process, and the FinFET process also cannot easily achieve the body bias needed by RF devices. As a consequence, by achieving real-time trade-offs between power consumption, performance, and cost, FD-SOI offers an ideal technology for new embedded systems with linking ability. The Internet of Things (IoT), 5G, and ADAS are the most suitable markets for FD-SOI technology. In contrast, an advanced CMOS technology such as FinFET is suitable for chips designed to offer maximum processing performance.
GF has always focused on maintaining close ties with clients and fully understanding their product needs. If a client wants to produce a high-performance processing chip, GF will recommend that they use the FinFET process; if they only want to produce a radar receiver, then the Si-Ge process will be sufficient; and if they want to produce high-resolution radar, the 22FDX process is the most appropriate. And while formulating solutions, GF also helps clients make the right choice by providing PPA analysis reports corresponding to different processes.
Taking automotive radar as an example, the RF unit of current 77-86GHz medium-/long-range automotive radar is usually based on the Si-Ge process, and the digital unit is based on the 180nm and 130nm CMOS process; as a result, the chip’s overall processing capability is poor. In comparison, GF’s 22FDX technology can provide outstanding millimeter wave performance and digital density, which can enable radar sensors based on 22FDX to provide even higher resolution and lower latency, while ensuring extremely low overall system cost. We have seen clients quickly introduce radar imaging chipsets based on 22FDX technology. These chipsets can detect objects within a range of 300 m, and offer a wide field of view with extremely high resolution.
And clients have been using GF’s mainstream CMOS process technology in the development of 77GHz short-/medium-range radar modules. These modules integrate microcontrollers, digital signal processors, SRAM, flash, and support components on individual circuit boards, and can be used to replace large radar arrays.
Of course, radar is only one way that semiconductors are being used in cars. Powertrain control is another way. At the recent Embedded World conference, Silicon Mobility displayed its Field programmable control unit (FPCU), which can be used to control electric and hybrid auto powertrains.
This element was designed to use GF’s 55LPx technology, can process information from and control sensors and actuators in real-time, and can be linked with a standard CPU on a single chip (complies with ISO 26262 ASIL-D safety standards).
A framework based on this FPCU will offer greater functionality, flexibility, and safety, and can boost powertrain control ability and performance in electric and hybrid cars. By employing hardware, not software, to quickly implement complex powertrain control algorithms, this framework can conserve energy and prolong battery life. According to Silicon Mobility, the FPCU can extend the range of electric and hybrid cars by 32%.
At present, MCUs used for air conditioning, engine, and oil system control, short-/medium-/long-range radar, ICs used for electric/hybrid car power supply management, and high-performance processors used in ADAS/self-driving systems account for the leading shares of GF auto electronics applications services. From our own observations, Chinese automotive clients tend to seek visual and self-driving processing chips, the biggest applications in the European market consist of microcontrollers, sensors, cameras, and lidar, and American clients are targeting lidar and self-driving solutions.
China is a very interesting market, and roughly 30% of semiconductor vendors in the international market are from China. However, many tier 1 auto manufacturers in China still purchase standard radar and processor chips from large motor vehicle device companies—this is the current state of affairs. Nevertheless, GF still sees great promise in the various innovative solutions that have emerged in China; one example of these is the application of visual monitoring system experience to the automotive field. Apart from providing built-in IP solutions such as MIPI interface and Can Bus, our strategy also includes a design center in China to help clients make even better use of GF platforms.