Why Semiconductors are Not Just About Nanometers
Public and policy debates about semiconductors are often reduced to “nanometers” as a measure of merit. Here’s why the 21st century’s most essential resource to decarbonize and digitalize our lives and economies deserves a more nuanced debate
Semiconductors are foundational building blocks of our modern lives. Digitalization would not be possible without them, and decarbonization of our economy relies fundamentally on semiconductors to control electrical power. Not surprisingly then, semiconductor technology has been placed center stage of international politics with the goal to advance technological leadership and to increase economic resilience.
The group of stakeholders I discuss these topics with has steadily become broader: senior politicians, academics, journalists, industry leaders, students and an increasingly interested general public. What always strikes me in those discussions is the genuine curiosity to better understand semiconductors and the high-tech industry which produces them. But should I really be surprised? After all, my professional live is a never-ending learning journey about those building blocks of the modern world, which suddenly have become the subject of extensive government funding, export controls, trade discussions and global supply chain analyses. I love those discussion and can engage in them for hours.
As a simple structure, we at Infineon Technologies AG have come up with an intuitive yet powerful picture: our “Semiconductor Tree”. To insiders, it speaks for itself. Usually though, when I talk about semiconductors in our industry, I spend some time to explain and discussing it.
This short article attempts to do just that: introduce interested readers to our “Semiconductor Tree” and provide some context. I am well aware that some of you might not find the answers you are looking for — we are, after all, talking about perhaps the most highly engineered high-tech good, the most sophisticated manufacturing processes known to mankind and one of the most complex value chains the world has ever seen. But still, I hope to make a helpful contribution to the public debate and understanding of the fascination and importance of semiconductors.
Semi-conductors evolved from the first transistor switch in 1947
At its core, a semiconductor is just what it says: “a semi-conductor”. Under certain conditions it conducts an electrical current, and under other conditions it doesn’t. “Condition” can mean different things: the presence or absence of light (then you get an optical image sensor), sound (a microphone), or, most importantly, an electrical potential: this makes an electricity switch. It is the electrical switch, the transistor, which is the foundation of much of the semiconductor world.
The first transistor was demonstrated in 1947, by a team of researchers at AT&T Bell Labs. Following the decades-long exponential miniaturization-roadmap, we have now arrived at microprocessors which include billions of transistors, with structures much smaller than the wavelength of visible light.
From the single transistor different development paths emerged
Before we go there, let’s make an important distinction though:
One branch of evolution in semiconductor technology kept optimizing the individual transistor. Optimizing a single electrical switch can chiefly be done along two dimensions: faster switching (eventually at radio frequencies), and switching of ever higher electrical power. The latter branch (in the tree the bottom right hand part) is known as “power semiconductors”. Power semiconductors are typically not highly integrated — just a single switch — but designing and manufacturing highest performance power semiconductors requires sophisticated control of the underlying technologies and processes. The materials we use to manufacture power semiconductors have evolved, too: whereas basic, cost-optimized devices are still made out of silicon, more advanced, energy-efficient devices require exotic materials (so-called wide-band-gap materials) such as Silicon Carbide and Gallium Nitride. Why is this important? As societies and economies move to decarbonize, more and more will run electrically. And a power semiconductor which is 3% more efficient can make an electrical car drive 3% further, a battery-power tool run 3% longer or save 3% of energy.
The need to understand both device design, and to control processes and technology to excel in power semiconductor manufacturing leads to an industry structure where leading companies control much of the value chain in-house: the Integrated Device Manufacturer (IDM). Infineon’s power semiconductor business is an example of an IDM, with major manufacturing sites in Germany, Austria and Malaysia. Europe has a leading position in the field of power semiconductors, which is not always known in the public, but important, when we talk about resilience and global dependencies.
Another path to optimizing the humble transistor was to standardize it as much as possible, shrink it and pack it as densely as possible: this is what gave rise to advancement of other branches in our tree — memory, and micro-controllers/microprocessors.
Shrinking the transistor was the basis for wide adoption and technical progress and different business models
But let’s talk about shrinking for a moment first, because this is what the semiconductor industry is best known for:
Long before the semiconductor transistor was invented, it was clear that electrical switches could be combined to form logic circuits. And, linking enough of those circuits together, computers could be built: the race to shrink transistors and integrate as many of them as possible into an Integrated Circuit (IC) was pre-ordained even before the transistor was invented.
And what a race it is: Following the exponential shrink roadmap of Moore’s Law (the observation that the number of transistors in a dense integrated circuit doubles about every two years) leading-edge factories nowadays manufacturer features with the size of 3 nanometers (nm). For reference, the spike protein of the COVID virus is about 10 nm long, and placing three gold atoms next to each other takes about 1nm. To manufacture those types of devices, visible light can no longer be used to project the desired structures on the chip (visible light has a wavelength between 380 and 700 nm, and physics dictates that the minimum resolution of an optical system is in the order of the wavelength of the light). Manufacturing at 3nm requires extreme ultraviolet light, and the machines used to process these types of semiconductors are the most complex machines ever assembled by mankind.
Manufacturing those type of semiconductors is very expensive (Figure 2). For it to be economically feasible, major economies of scale are crucial: producing large numbers of the same device, and processing as many of them together in one step as possible. From the first observation it follows that Integrated Circuits (ICs) manufactured at the smallest nodes require highest volume applications (e.g. mobile phones) to be economically lucrative. The second — processing as many ICs at once — lead to ever larger wafer diameters: 2, 4, 6, 8 and currently 12 inch (300mm).
From a business perspective, economies of scale favor ever larger, specialized producers. Taken together with the high degree of standardization, this gave rise to a division of labor in the industry for ICs:
· Fabless semiconductor design companies are the much less capital-intensive flip-side to the manufacturers: Those fabless companies focus solely on the design of semiconductors which are then handed over to foundries for manufacturing using standardized instructions how to do so.
· Foundries, then, focus on manufacturing highly integrated semiconductor computing devices (but much less so other semiconductor devices, as I explained above). Economies of scale favor concentration, and the largest foundries form huge manufacturing clusters. The largest, oldest and most advanced is Taiwan’s Semiconductor Manufacturing Company (TSMC). As a result of TSMCs visionary founder, support by the Taiwanese government, and the economies of scale, 90% of sub-10nm semiconductors are today manufactured in Taiwan — with obvious geopolitical implications
Before we get back to the Semiconductor Tree, a word about the impact which shrinking (remember: very expensive) has on different devices (Figure 3): So far, we talked about highly integrated devices at the smallest node sizes, and e.g. power semiconductors where advancement took place towards new materials (Silicon Carbide, Gallium Nitride). In reality though, a large number of semiconductor devices which need to combine different properties: computing, sensing, power control, memory storage (Figure 3).
Memory chips and Microcontrollers
For that, let’s get back to our semiconductor tree: Memory semiconductors are next, and they are conceptually simple to understand: they allow storage and read-out of information in the form of electrical signals (0 and 1 — what makes “digital” digital). Depending on whether storage requires a constant electrical power source in the background, or whether information is retained also without electrical power, memory is classified either as volatile or non-volatile. Memory semiconductors are shrinking at a rate just marginally below microprocessors.
While much of the public attention over the last years focused on microprocessors of the smallest node, following this shrink path is economically attractive only for highest volumes and if highest performance at the lowest power is a design requirement.
Much of the microcontrollers which control our everyday life, from washing machines to cars, airplanes and industrial control systems, are manufactured at future sizes of 22 to 40nm. Microcontrollers and device in this range are optimized for different requirements, e.g. high security and reliability, or low-power consumption. Many automotive applications rely on microcontrollers; and they will continue to play this foundationally important role in the future. European companies play a crucial role in this branch.
More exotic semiconductors link the analog to the digital world
Slightly to the left in our tree is the branch labeled “analog/mixed signal”. In brief, these semiconductors are the interface between our real world (which works with analog signals) and the digital world which is, well, digital. Often with microcontrollers at their core, the devices are optimized to also process analog electrical signals. At the highest frequencies, those analog electrical signals are electromagnetic waves, giving rise to the class of radiofrequency semiconductors: WiFi, Bluetooth or radar chips for autonomously driving cars are found in this category.
As we move further to the left, we see the wide and diverse field of (semiconductor) sensors. Temperature, magnetic fields, gas, pressure, … The number of physical effects semiconductors can sentence and convert into electrical signals is impressive. Just look at your mobile phone: the microphone, the camera, temperature and positioning — semiconductors as far as the eye can see. And, last but not least optical semiconductors emit light, just look at LEDs as the global standard for efficient lighting everywhere.
If you stayed with me on this journey through the fascinating world of semiconductors, let me offer you some concluding thoughts:
- The world of semiconductors is vast and diversified: power semiconductors, sensors, microcontrollers and -processors, chips on the smallest nodes as well as all the other categories in our semiconductor tree are based on enormous research efforts and knowhow. They gave rise to specialized industries which reflect unique properties of the products they manufacture — all based though, on the electrical transistor switch from 1947. To advance — move further away from the trunk to the leaves is a great effort — in every branch. I feel that this fact is becoming more and more prominent in the public debate and we are having a more precise discussion about the importance and development of the respective industries. In my view, this is a positive development.
- Semiconductors are in fact essential building to our personal lives, modern societies and -economies. With megatrends such as digitalization and the urgent need to decarbonize economies. This importance will only grow — with no replacement technology in sight. Some have compared their ubiquity and importance to oil, and called semiconductors the “oil of the 21st century”. An interesting comparison, considering the global and local political and societal changes which oil brought to lives and industries in the late 19th and early 20th century. No surprise then, either, that semiconductors are at the center of global politics.
— Thanks for reading and feel free to comment or contact me directly if you’d like to discuss.
I gratefully acknowledge editorial support from Andre Tauber and the original semiconductor tree was conceptualized by Jochen Hanebeck
