This is the second part of ASN coverage of Day 2 of the recent SOI Workshop in Tokyo, which was dedicated to the “Convergence of IoT, Automotive through Connectivity”. Many of the presentations are now posted and freely available – click here to see the full list.
Peter Rabbeni, GlobalFoundries’ Sr. Director of 5G BizDev and Product Line Marketing focused on mmWave and why/how 5G. In his talk, Delivering on the Promise of 5G: Semiconductor Solutions for the Next Wave of Data, he pointed out that there’s not one solution for all use cases – but there is an SOI solution for all the opportunities.
You need mmWave for latency, simultaneous connectivity, energy-efficiency and mobility, he explained. mmWave addresses the trade-off between distance and data rates. In a phase array, the beam is steered, but because of atmospheric absorption, you have to do multiple beams at high frequencies.
RF-SOI technology is already found in virtually every smartphone in the world. Now, he sees two main benefits in RF-SOI (a partially depleted technology that uses “trap-rich” substrates, btw) in the move to 5G and mmWave. One is device stacking, which you can do in SOI to overcome the Johnson Limit (a tradeoff between breakdown voltage and frequency). The other comes from the benefits inherent in the substrate: high-resistivity, high-Q and isolation. It means you can have smaller arrays for each element, and fewer chips per array. That’s key: you need those smaller arrays for handsets and customer premises equipment.
Different designers are taking different approaches to RF, he notes. There are those doing FEM-centric designs, which integrate from the antenna back toward the transceiver. And then there are those that are doing integration-centric designs, which target integration from the transceiver/BB toward the antenna. The first approach is being driven by those customers with unique IP and presence in the front-end module space. The other is being driven by folks with IP and presence in the SOC space. Both will exist in some form, he contends. 45RFSOI is well aligned with the first case and focused primarily with FEM leadership performance and integration. 22FDX, on the other hand, is very well suited for transceiver/baseband ADC/DAC integration and can integrate the FEM functionality as well. Pathfinding on the FEM integration component is on-going for 22FDX.
SOI is “…the perfect solution to our needs”, said Steven Yeung, Design Manager with MIPS/Imagination Technologies in his talk, MIPS Leading Heterogenous Compute in Automotive & IP. The cost of failure is increasing he noted, citing the ISO standard 26262 for functional safety in road vehicles, and SOI “helps a lot”.
As noted in the presentation title, research powerhouse Leti sees that the Future of the Automotive Industry is Paved With SOI. Vincent Roger of Leti’s Corporate Business Development made convincing arguments as to why FD-SOI is the right solution for automotive:
you need advanced digital circuitry for all computational tasks in the automotive environment
the 3-generation node gap between automotive and consumer is closing
FD-SOI is more power efficient than planar bulk (both at 28 and 22nm) or FinFET (16nm)
it’s simpler in terms of design and IP portability than FinFET
it’s a proven solution, with better reliability and lower design costs
it addresses all performance levels and communications
it simplifies integration of control electronics for distributed sensors
Leti is actively working on getting RF capabilities in FD-SOI adopted more quickly. For example, they are developing RF models for their UTSOI-2 modeling suite for FDSOI, including back bias effects. And they’re also developing innovative basic design blocks that prove the technology validity and add new functionality.
He also sees an even bigger role for RF-SOI, the technology of choice for RF Front End Modules for connected vehicles and 5G applications. With Soitec, they’re working to keep improving existing substrates and introduce new concepts.
SOI wafer suppliers (Soitec, SEH and Simgui) are expanding capacity, said Soitec EVP Thomas Pilisczcuk. His talk, The Role of Substrates in Accelerating Mass Adoption of SOI Technologies, reviewed the various SOI substrates and partners across FD-SOI, RF-SOI, photonics, power, image sensors and more.
Soitec is launching a program called FIRST (for First Integration Ramp of SOI Toolbox) to help customers reach competitive yields fast. They are also help customers facilitate SOI integration into design and manufacturing.
There are still more talks that are now posted on the SOI Consortium website. IHS/Markit made a very interesting high-level presentation on LIDARs & Sensor Fusion ECUs Advancing ADAS Architectures Toward Automated Driving, which called for chipmakers to integrate more features. Nokia Future X Network for 5G & IoT looked at infrastructure (they use their own chipsets). ST looked at smart cities in Sensor-to-Cloud Connectivity for IoT.
Equipment makers are also eager participants in the FD-SOI ecosystem. Screen’s presentation was entitled Full Participation Within the SOI Consortium. The Applied Materials talk, Enabling SOI and IoT: An Equipment and Materials Engineering Perspective, covered how they’re working with their customers and their customers’ customers to understand the trends and enable the device roadmap.
Mark your calendars: the next workshop sponsored by the SOI Consortium will be in Shanghai this September 26th and 27th (one day is all about FD-SOI, the other about RF-SOI). You can now register or ask for an invitation: see Events on the SOI Consortium website. Last year’s Shanghai event was really dynamic and absolutely packed, so you’ll want to make sure you register early. (But if you can’t make it, you can of course read about later it in ASN!)