At the 2018 Global CEO Summit, Soitec CEO Paul Boudre delivered a keynote speech titled "Engineered substrates – enabling the AIoT & 5G revolutions." The content covers four parts: supporting China's smart vision through engineering substrates; artificial intelligence (AI) is the catalyst of the Internet of Things - the AIoT era; engineering substrates are essential for providing 5G;summarize.
Support China's smart vision through engineering substrates
Mr. Boudre pointed out that China is fully deploying "Made in China 2025" and hopes to build China into a manufacturing powerhouse. This project covers 10 key areas, as shown in the following figure.
China has made huge investments in 5G and artificial intelligence, and we can see that these areas will have good development prospects. For example, in the infrastructure construction of 5G networks, China's investment costs exceed US$2.4 billion. China has invested in many aspects, and at the same time, the semiconductor industry has been developed as the focus of 5G construction.
China is planning to lead the 5G deployment. As shown in China Mobile's 5G route plan, it will deploy more than 10,000 sites by 2020. China will allocate up to 500MHz spectrum for 5G in the sub-6GHz band. Rapid deployment can be achieved with only a small investment in this band to meet the requirements. This is the status quo of China and the leading direction of the future.
“We want to remind everyone to choose the most suitable engineering substrate,” Boudre said. For example, in France, in order to produce very good wines, you must have a good production line, and you must choose different things according to different wines. Also pay attention to its soil condition, because good soil is a necessary condition for producing good crops and finally producing good wine. Again, this concept can be extended to the substrate material. For different semiconductor applications, there must be a very good substrate to support its advantages in achieving functionality.
When discussing the relevant engineering substrate optimization, you need to understand the history of material innovation from 1985 to the present more than 30 years. “We are constantly striving to innovate and develop, discover new materials, and continuously improve the performance of engineering substrates. It can be said that this is a revolution,” Boudre said. “From 1985 to 2015, there were a lot of developments, and many elements were added. Come in, of course, we are still developing at a high speed."
The substrate based on Smart Cut technology is shown below. “We are now combining these materials. For superimposing crystals on crystals (lattice incompatibility) and amorphous materials, we can achieve ultra-thin, highly uniform layer design. Good performance. At the same time, our technology realizes the optimization of the substrate and the integration of the insulating layer, such as SOI (insulating silicon). We apply the insulating silicon, as shown in the following analysis diagram, divided into three layers. Natural oxides, crystalline silicon and oxide layers, the intermediate crystalline silicon reaches a thickness of 3.8 nm. Very good performance can be achieved with such a design structure," Boudre explained.
Soitec offers a full range of engineered substrates covering a wide range of applications and applications, including RF front-end modules and more, for use in smartphones. “No matter what kind of smartphone you use, you can apply our related products and technologies to achieve very good performance. At the same time, we also constantly contact and communicate with our customers to get the latest information from them. To understand the problems they face in order to personalize these issues," Boudre pointed out.
Artificial intelligence (AI) is the catalyst of the Internet of Things - the era of AIoT
Thousands of projects are now centered around artificial intelligence and 5G, the next generation of core technologies for human development. The 5G network is the foundation that supports our future technological development.
What is artificial intelligence? It is actually turning data into useful information. Building such a network can create a new world for us and let us better understand the world. For example, distinguish between apples and other things, and distinguish people's voice from other vibration waves. Artificial intelligence needs to be supported by a good CPU chip to support the implementation of these technologies and final functions.
The development of artificial intelligence technology has undergone great changes and has made many breakthroughs. For example, at the maturity stage, we can even see robots that provide services like humans. At the same time, it will also affect the interaction between our humans and smartphones.
So what is the challenge of the next stage of artificial intelligence? The next challenge is the development of artificial intelligence inference to the edge, requiring a low-power solution. Artificial intelligence involves three different solutions: very edge, edge and cloud. Marginalization requires a very secure and guaranteed low power solution. "Low power consumption is a very important part of the whole equation. We still have a lot of work to do. We reduce power consumption and improve performance in various ways. I think all these efforts are worthwhile," Boudre said.
So how is the current technological progress and where should it go in the future? In the future, IoT and AI will be better integrated, which will trigger a new revolution. This new revolution requires our common participation. This revolution will also be more violent than the revolution of the past.
FD-SOI will play an important role in this revolution. The FD-SOI optimized substrate will provide energy for the fusion of AI and IoT. FD-SOI can improve energy efficiency and increase the effectiveness of the entire solution. For some technology nodes, better energy efficiency can be achieved by using body bias technology on ultra-low power. The following figure analyzes the relationship between performance and energy efficiency. FD-SOI has different energy efficiency performance under different performances. FD-SOI is a very good innovation that can help us drive the development of AI chips.
The evolution of AIoT is very fast and there are many rich applications. For example, we want to achieve always-on-line monitoring, AI theft, smart wake-up, always-on and battery-powered, local voice inference, and automatic anomaly detection. There are many examples to prove that FD-SOI is helping the AIOT revolution.
Currently, the first generation of AI chips has been introduced on the market. For example, NXP's i.MX RT600 cross-border processor is based on FD-SOI design and is suitable for machine learning and artificial intelligence functions at the edge. This new technology can bring us more possibilities in the future. Based on FD-SOI, Lattice has developed a new generation of always-on FPGA low-power machine learning inference chips. Designers can use it to design small-sized, high-performance AI devices without violating footprint and thermal management constraints. FD-SOI also has many good development projects in China.
Engineering substrates are critical to providing 5G
By then, we will have a more flexible and smarter network. LTE and WiFi will continue to exist at frequencies around 3 GHz. 5G will be deployed in the higher frequency band of millimeter waves. As the frequency increases, the wavelength will be shorter.
Compared with the current network, the 5G network will carry a lot more content. So how do the optimized substrates in today's 4G networks be applied? FD-SOI is used in many areas such as 4G bandwidth and core network in the world (blue font in the figure below). And 5G will bring us many new opportunities. In the 5G interconnected world, it is necessary to further promote the optimized substrate. In the 5G era, FD-SOI will have more applications, such as millimeter waves, which is an inevitable development choice in the 5G era. FD-SOI mmW will support core networks and access networks in the 5G era. There are many opportunities in China, and China can build a complete ecosystem in this direction.
Overall, 5G is a very complex ecosystem. Successful deployment requires the cooperation of all parties in the entire ecosystem. It needs to rely on the country's environment and needs to establish many partnerships. It needs production, distribution, engineering, and network supplier participation. . In China, China Mobile is very strong in the development of 5G. In terms of OEM, China also has many good companies, including Huawei, OPPO and ZTE. In terms of 5G fabless, IDM companies and design companies, there are also many well-known companies in China. There are also many excellent companies in foundry.
To sum up
AI and 5G are the next focus of human development. China will play an important role in helping these technologies and solutions enter the mass market. The role of optimizing the substrate is significant and can advance the development of new technologies. Throughout the value chain, we need to strengthen cooperation and promote the application of these new technologies in China.