The semiconductor content in cars is projected to grow significantly as cars are being transformed into autonomous driverless vehicles. Today, a typical car contains $799 of electronics (including $353 of semiconductors) which is expected to grow to $886 ($386) by 2022. This transformation is anticipated to change the way people navigate, access information, and interact with one another . With pivotal challenges lying ahead for future automotive electronics, an evolution from semi-autonomous to fully-autonomous vehicles is expected. Designing and managing the car electronic systems to achieve those functions in mobility mode will depend heavily on complex semiconductor circuits and advanced components such as all kinds of sensors, microprocessors, microcontrollers, and mixed signal analog/RF. These electronics must have ultra-low power consumption, especially static power, superior reliability in harsh environments; and security, all while maintaining sufficient performance, high data rate, multiple and simultaneous connectivity. In this paper we will discuss enabling Silicon-On-Insulator (SOI) device technologies for the future semi and fully autonomous vehicles.
Automotive semiconductor content will continue to increase and be a driver of growth for the semiconductor industry for the next 5 years. Carsrely on greater intelligence, connectivity and sophisticated electronics; consumers demand cars that sync with smartphones and in-vehicle infotainment capability; and governments mandateimprovement in safety and fuel efficiency. Over the next decades, it is predicted that more than 100 million cars around the world will haveautonomous-driving featuresand simultaneous connectivity, both vehicle-to-vehicle and vehicle-to-infrastructure. This drivesautomotive electronic content to be more than 30% of total vehicle bill of material and continue growing .
Automotive Electronics Requirements
The requirements for automotive electronics are in general more stringent than electronics for consumer applications, smartphones, and computing. It is critical that automotive electronics must operate at a wide temperature range (-40C to 150C) and 15 years with zero failure. Cars today have progressively more computing capabilities, and various forms of semi-autonomous technologies, such as adaptive cruise control, automatic parallel parking, and collision warnings, are already widespread . More and more high performance computing capability will be needed moving forward, to enable real time analysis of data from various sensors (e.g. image, RADAR, LIDAR, audio…). With that, the adoption of Advanced Driving-Assistance Systems (ADAS) will increase to provide ultra-fast and reliable image-processing capability that can automatically take safe countermeasures such as autonomous emergency braking and pedestrian-protection systems . In addition to high performance computing, high data rate and high bandwidth connectivity are essential for a reliable crash avoidance system. Thus more complex radios with higher linearity, lower insertion loss, and better harmonics, will be needed to make autonomous driving a reality. It is important to mention that technology affordability will be key to fueling adoption. For example, with connected cars requiring RF electronics and new RADAR applications requiring high performance analog, efficient integration of the different functionalities is key for cost efficient solutions. Integration can open the way for single-chip automotive SoC solutions (e.g. Front-end and baseband processor). For easy adoption and adaptation into a wide range of vehicles, small size and low cost are key market drivers along with the required high performance levels .
SOI Device Technologies
Beside the need for numerous sensors for ADAS and for future autonomous cars, there are four highly valuable SOI based device technology solutions for the current and future automotive electronics industry; Fully Depleted SOI (FD-SOI), RF-SOI, Power-SOI and Silicon Photonics-SOI, Figure 1.
FD-SOI Technology Advantages
FD-SOI technology enables scalable planar fully depleted devices with a wide range of back bias capability to improve power and performance trade-off. This extends Moore’s Law beyond 28nm with the least design and process impact, while providing mobility booster capability and cost effective solution for low power, high performance integrated circuit solution via substrate engineering . FD-SOI devices have excellent immunity to Short Channel Effects (SCE) leading to improved sub-threshold swing and Drain-Induced Barrier Lowering (DIBL), and minimum Random Dopant Fluctuation (RDF), thanks to the undoped channel. This ensures lowest Vt variation [6, 7], resulted in excellent SRAM mismatch and stability, good analog mismatch and gain, better scalability of Vdd, SRAM cell and analog device, gain. These enable a superior digital/analog co-integration . Furthermore, FD-SOI has improved noise & noise variability due to the lower parasitic capacitance (compared to FinFET) for high speed designs and improved passives that enables LNA & VCO with improved noise figure.
Unique feature of FD-SOI on thin BOX substrate is the back-bias capability, which enables threshold voltage (Vt) tuning for better performance/power trade-off without degradation, (Figure2) and more cost effective solution than fabricating different Vt transistors using channel doping or work-function tuning.FD-SOI has superior SER reliability compared to bulk devices as shown in Figure 3 . As susceptibility of bulk devices to SER increases with scaling, SER is becoming as important to IC reliability as intrinsic failure modes. Commonly implemented redundancy schemes are no longer efficient for scaled bulk devices. FD-SOI is intrinsically more resilient to SER than bulk by at least 25x (comparing 28mn FD-SOI vs. 28nm Bulk devices). The superiority of FD-SOI is attributed to the (a) buried oxide that suppresses charge sharing, (b) SOI ultra-small sensitive volume which limits particle interactions and charge deposition, as well as (c) the SOI advantage of having no latch-up. Because of the significant SER advantage, FD-SOI will require less with less error correction and redundancy, thus simpler design.
FD-SOI devices are planar devices that are fully compatible with mainstream CMOS processing, designs and EDA tools, thus enable a faster time to market and low design and process cost solution. In addition to fully leveraging conventional CMOS processes, FD-SOI process integration is simpler than Bulk, Figure 4 . The FD-SOI process saves several masks and process steps typically included for Vt tuning and for the integration of uniaxial stressors needed to boost performance in planar and FinFET bulk . Even with the drastically increasing lithography cost, such process simplification more than compensate for the SOI substrate cost, resulting in a lower overall processed wafer cost, as confirmed by foundries in Figure 3 . Last but not least, the simple process flow of FD-SOI using the existing manufacturing processes also resulted in a steeper yield-learning trend.
Whereas FD-SOI is suited well for automotive application with its low active and static power, better performance/power trade-off, excellent reliability, cost-effective and RF co-integration features, even at the high temperatures of automotive electronics
Majority of industry envisions four key capabilities in the connected vehicle: Connect inside the car, Connect to personal devices, Connect car to car and Connect to infrastructure. This will require multiconnectivity (multiple cellular, NFC, Wi-Fi, 802.11p ,…) as well as advanced connectivity and mobility management to ensure uninterrupted connection at high speed to meet latency requirements for critical applications, and to satisfy network and computing security imperatives with low power consumption [4, Cisco].
The rapid adoption of new wireless standards and the increasing demand for data bandwidth requires RF IC designers to develop devices with higher levels of integration, meeting more and more stringent specification levels. Typical SOI substrates do not have thick enough BOX to prevent the electrical field from diffusing into the substrate, inducing high-frequency signal losses, non-linearity and crosstalk which are detrimental to RF performance. To improve the insertion loss, harmonic distortion and isolation performance required for switches, the bulk base substrate of an SOI substrate was replaced by a high-resistivity base substrate known a HR-SOI . The adoption of HR-SOI wafers for RF applications has allowed monolithic integration of RF FEM, leading to smaller size, better reliability, improved performance and lower system cost [17, 18]. RF industry roadmap is converging to a second generation Trap Rich RFSOI substrates as the preferred choice due to very low RF insertion loss, low harmonic distortion along coplanar waveguide (CPW) transmission lines, and purely capacitive crosstalk close to quartz substrates as illustrated in Figure 5. The Trap Rich RF-SOI substrates meet present network standards, are currently deployed in Smartphone FEM, and open the path for more FEM future integration.
SOI is a natural way to isolate among transistors and circuit blocks, Figure 6. This is particularly advantageous for scaling and fabricating power and smart power devices/circuits. IC makers can take the benefits of complete parasitic junction removal, which results in latch-up free, die size reduction and better electromagnetic interferences immunity. This enables simpler integration for high voltage (>600V) devices and smart power circuits, which is a co-integration of Bipolar for analog, CMOS for digital, and DMOS for power and high voltage functions (BCD technology). Such integration is critical to reducing complexity and cost. In addition, SOI enables the RESURF effect (Reduced Surface Field Effect), providing the best trade-off between breakdown voltage and on-resistance of lateral devices for optimum integration of high and low breakdown voltage devices . Each application has a different motivation to using SOI. But all benefit from the SOI fundamental advantages of high reliability, high temp operation, efficiency, more integration and lower cost. All these aspects are key advantages in automotive, where ICs need to ensure very reliable operation in hostile environment. In fact, automotive electronic components must endure both very high and very low temperatures. High voltage spikes can contaminate the bus network, and cars are full electromagnetic interferences sources. This is the reason why power SOI has been increasingly used since 20 years for a wide range of automotive components such as transceivers, audio amplifiers, powertrain controls, and brightness LED drivers. We estimate that each new car built today contains an amazing average of 80 square millimeters of SOI, and more than six billion automotive ICs have been manufactured on its Power SOI substrates to date . This technology will continue to support the current and future automotive electronics for semi and fully autonomous vehicles.
Si Photonics is a promising low cost, high density, high bandwidth interconnect solution for future high bandwidth data transmission (Data centers à Chip-level). Silicon photonics enables monolithic integration of optical components with electronics using standard semiconductor fabrication techniques and SOI substrates. SOI substrates provide the optical platform enabling high density optical interconnects along with optical modules/functions and other Si electronics. The motivation is to build very high throughput interconnect networks, similar to what is used in telecommunication, but at smaller footprint and far lower cost. Photonics is currently being deployed in metro and datacenters for 100G interconnect technologies . In automotive silicon photonics could also be a game changer for in-vehicle networking. When we look at autonomous cars, they will look more and more like a “mini” data center surrounded by sensors (cameras, RADAR, LIDAR, ultrasound…), all in a hostile environment. The challenges of the in-vehicle network linking all sensors to the central computing chip regarding speed and reliability will therefore be very significant. Silicon photonics, because of far less sensitivity than standard electronics regarding the specific constraints of car environment, could be a key technology to meet these challenges.
Today, cars have an average of 60-100 sensors on board. With the transformation to partially and fully autonomous vehicles, the number of sensors is projected to reach 200 sensors per car . SOI technology can further simplify MEMS process, manufacturing control and design. One example, is pre-etched cavities in SOI wafers, which can simplifying the process flow, allowing optimized geometries key for sensitive devices as well as permitting access to thin membranes (e.g. gyroscopes, resonators). In addition to providing more freedom in design and integration; SOI-based MEMS solutions have several advantages, including; ability to withstand high pressure and temperature, long product lifetime, and small die size . Some examples of SOI based MEMS devices include, gyro sensor for automotive control  and Gas sensors for combustion optimization and emission monitoring .
Cameras are the fastest growing sensors for ADAS applications . SOI is suitable for Backside Illuminated (BSI) image sensors that provide the most direct path for light into the pixel, providing enhanced efficiency, minimum distortion and higher yield enabled by the robust SOI etch stop . Other types of sensors such as Time-Of-Flight (TOF) sensors are also being deployed in cars to enable ADAS. For such sensors, SOI can enhance the efficiency at near infrared wavelength, while maintaining short timing resolution at low power consumption .
Security, reliability, high performance and low power electronics are essential for the adoption of partially to fully autonomous vehicles. Designing and managing the car electronic systems to achieve the needed security and reliability depends on key enabling functions. We showed how SOI based devices provide reliable low cost integrated solutions with low power, high performance for RF, analog and digital circuits. In addition to the current SOI smart power and several sensor technologies. These devices are available today, but will continue improving in performance and integration level to provide the best reliability, wide range of temperature operation, low power, high performance, and low cost SoC solutions for enabling future autonomous driverless vehicles.
 Strategy Analytics, Automotive Electronics System Demand Forecast 2013 to 2022, (2015)
 M. Claudel, Full speed ahead: How the driverless car could transform cities, McKinney & Company (2015)
 PWC Analysis
 L. Reger, ISSCC, p. 28-30, (2016)
 M. Sadaka et al., Solid State Technology, July 2015
 O. Weber et al., IEDM, p.1-4, (2008)
 L. Grenouillet et al., IEDM, p.3.6.1-3.6.4, (2012)
 S. Le Tual et al., ISSCC, p. 32-383, (2014)
 P. Roche, et al., IEDM, p.31.1.1-31.1.4, (2013)
 P. Magashak et al., DAC (2014)
 P. Flatresse, S3S Conference short course, Monterey, 2013
 B.-Y. Nguyen et al., Advanced Substrate News, (2014)
 C. Didier, White Paper: RF-SOI Wafer Characterization, Soitec (2015)
 J.-P. Raskin et al., IEEE Trans. on Electron Devices, 44, p. 2252 (1997)
 A. Larie et Al., ISSCC, p.48-50, 2015
 Yole Développement: RF Filters, PAs, Antenna Switches &Tunability for Cellular Handsets, (2014)
 J. Le Meil, et. al., VSLI TSA, (2015)
 J.-P. Raskin et al., IEEE Trans Electron Devices, v.44, #12, p. 2252-2261, (1997)
 D. Lederer et al., Solid-State Electronics, 47 (11), p. 1927 (2003)
 Piet Wessels, NXP, SOI Conference 2007 Shengdong Hu et al, 2008
 SOITEC receives best quality award from NXP (http://www.soitec.com/pdf/nxp_award_press_release_may_2016_final.pdf)
 Cisco CPAK for 100-Gbps Solutions (http://www.cisco.com/c/en/us/products/collateral/optical-networking/ons-15454-series-multiservice-provisioning-platforms/white_paper_c11-727398_031813.html)
 Automotive Sensors and Electronics Expo 2016
 E. Mounier, ASN, (2011)
 Y. Nonomura, et al., Sensors and Actuators A: Physical, v. 132(1), p. 42–46, (2006)
 F. Udreaet, et al., CAS, v.1, p. 3 – 10, (2012)
 R. S. Edelstein, IISW (2011)
 E. Charbon, IEDM, p. 10.2.1 – 10.2.4, (2014)