Why Superconductors Are a Myth... So Far!

Hello, this is Kakao Ventures.

Our investment team always works closely with early-stage startups while keeping an eye on market trends. In this process, we often have questions and engage in many discussions. We believe that the more diverse the thoughts and the deeper the conversations, the better. That's why we would like to share some of our thoughts with you one by one in the future. We hope this will be of some help to entrepreneurs, investors, or anyone interested in the market.

Why Superconductors Are a Myth... So Far!

[Deep Tech Stories with Zero] 2. The Shift to Next-Generation Semiconductors

Written by Youngmoo Kim

Hi. Welcome back to the "Deep Tech Stories with Zero" series!

This is the second episode, and I already wish there were a robot or AI that could automatically read my mind and write these articles for me.

Maybe it will become possible one day. But there is an easier way to bolster one’s writing skills, though. :) Likes and subscriptions, which I now understand why are so important for YouTubers and bloggers, are really a huge encouragement for content creators like me! But I digress.

As promised in the previous episode, today I will discuss how new materials and devices are commercialized.

I think we can start by talking about a recent event. Last July, there was a global buzz surrounding a new superconductor called LK-99, announced by the Quantum Energy Research Institute.

What made LK-99 particularly exciting was its potential to exhibit superconducting properties at room temperature and atmospheric pressure.

In contrast to existing superconductors, which require extreme conditions like temperatures around -200°C to function, LK-99 seemed to work under normal conditions, making it much more feasible for commercialization.

This caused quite a sensation because a room-temperature, atmospheric-pressure superconductor could revolutionize energy usage and open up limitless possibilities, sparking a wave of internet memes.

However, despite the excitement about potentially becoming a superpower, the discovery of LK-99 turned out to be more of a temporary phenomenon, and the hype around superconductors soon faded.

I remember being personally disappointed at the thought of not being able to commute in a flying car. This - and a host of other exciting possibilities that could have happened - brings us to an important question:

“Why can’t we create superconductors?”

To answer this question, let’s try to become materials scientists aiming to develop new materials by utilizing substances with specific properties.

First, to create a room-temperature, atmospheric-pressure superconductor, we need to understand the properties of superconductors. We must know exactly what we’re trying to create before we begin.

We’ve already discussed what room temperature and atmospheric pressure mean. Superconductivity, as can be intuited from its name, refers to a material that conducts electricity with zero resistance.

Additionally, superconductors have two other properties: the Meissner effect, where a superconductor repels magnetic fields when exposed to them, and Quantum Locking, where a superconductor levitates above a magnet.

So, why do superconductors possess these three properties?

The problem is that we don’t really know why or how these properties manifest in superconductors.

Research on room-temperature, atmospheric-pressure superconductors has been ongoing for a long time, even before the Solvay Conference in the 1950s.

Despite decades of research, including Nobel Prize-winning results like the BCS Theory, Bednorz-Müller’s research, and the Ginzburg-Landau Theory, we still don’t fully understand the principles behind superconductivity.

We don’t even know whether room-temperature and atmospheric-pressure superconductors exist, or why extremely low-temperature superconductors exhibit superconductivity.

The BCS theory, which explains the phenomenon of superconductivity, assumes extremely low temperatures for it to happen and cannot explain the principles behind room-temperature, atmospheric-pressure superconductors.

The Bednorz-Müller research that proposed the possibility of relatively higher-temperature superconductors using copper oxides (like the Cu-O compounds in the diagram below) is still a “discovery” rather than a definitive scientific explanation. Simply put, we don’t know why copper oxide superconductors exhibit superconductivity at these higher temperatures and what the upper temperature limit might be - if they could ever work at room temperature as well.

I propose that for any new material to be commercialized, it must meet at least five conditions:

In the case of superconductors, although there is a need for such a material and candidate materials have been explored, the principles behind superconductivity are not fully understood. Thus, it’s challenging to achieve the market's desired "room temperature, atmospheric pressure" target, making commercialization difficult.

So, what materials have met these five conditions?

I think the most representative and successful material is the silicon wafer, used in semiconductors. Like superconductors, silicon wafers have been studied since the 1950s and have been successfully commercialized, becoming an essential part of our lives.

Here’s how they meet each of the five conditions:

First: The Need for a New Integrated Semiconductor Beyond Conventional Transistors

As discussed in the previous episode, BJTs, based on conventional transistors, were too large to improve computer performance. (You can read the previous episode through the link below for more details.)

To address this issue, there was a strong need for a new integrated semiconductor, and research began on new semiconductor materials such as silicon wafers that could replace the role of huge transistors.

Second: Selecting Candidate Materials from the Periodic Table to Replace Transistors

To replace the role of transistors at the material level, we first need to understand what they do.

As mentioned in the previous episode, transistors always have three legs because computers operate on binary (0 and 1).

The 0 and 1 in a computer are determined simply by whether an electrical signal is flowing (1) or not (0).

Therefore, two legs are needed to serve as the input and output for the electrical signal, and a third leg is needed to act as an on/off switch to control the signal. This is why transistors always have three legs.

What should be noted here is that the role of the transistor is to act as an on/off switch for electrical signals.

This means that the new material for integrated semiconductors must have good electrical conductivity and be able to block the signal when needed.

So materials scientists explored substances with high electrical conductivity as well as a good ‘energy bandgap’ - which indicates effective control of electrical signals, allowing them to pass in the On state and blocking them in the Off state.

The result? These elements highlighted in red boxes in the periodic table below were selected as viable candidates.

Third: Maturing Research on the Properties of Candidate Materials

Researchers then studied the properties of these materials to understand why and how they exhibit certain behaviors.

Unlike with superconductors, scientists were able to successfully build mathematical models to explain the properties of semiconductor materials and how to control them.

And these models came with pinpoint accuracy: As shown in the diagram, researchers were able to quantify the behavior of semiconductor materials to the point where they could predict exactly how much the electrical current would change when the temperature increased by 0.001 degrees. This wasn’t just a wishy-washy ‘maybe it will move this way and that’ but a calculation that came down to precise numbers.

As a result, scientists knew how to utilize and enhance the properties of semiconductors.

Fourth: Enhancing Target Properties to Meet Commercialization Thresholds

We’re almost there.

We’ve finally figured out why and how semiconductor materials behave the way they do. Now we only need to ensure that these properties are stable, the material is easily obtainable and affordable, and its physical properties are easy to work with before bringing it to market.

For example, a material emitting radiation would certainly be unsuitable. Among the elements in the red boxes of the periodic table, carbon (C) was considered as a semiconductor material, but its form as diamond made it too hard to work with, making it difficult to etch fine circuits. Germanium (Ge) was already being used in BJT transistors, so we were looking for something with higher electrical conductivity.

That leaves us with silicon (Si), positioned between carbon and germanium.

Silicon is the second most abundant element on Earth, making it easy to obtain and affordable. It’s also thermally stable, durable, and not as hard as diamond, ****so it is more amenable to etching circuits onto it.

Thus, we decided to create silicon wafers instead of carbon or germanium wafers.

Fifth: Perfecting the Production Method for Silicon Wafers

To fully harness the electrical properties of silicon wafers, ultra-pure single-crystal silicon is required.

Fortunately, researchers were able to develop the Czochralski method, which made this possible.

Let me briefly explain this method:

First, silicon is melted in a furnace, and a small silicon seed is dipped into the liquid silicon and slowly lifted while rotating.

As the seed is lifted, the liquid ultra-pure silicon adheres to it, forming a cylindrical shape as the seed grows.

This cylindrical silicon mass is called a silicon ‘ingot’, and cutting it horizontally yields round silicon ‘wafers’.

Now we know why silicon wafers are round, even if we’re still not sure why manhole covers are round.

With these silicon wafers, integrated circuits could be made, and the semiconductors industry has since then exploded in accordance with Moore’s Law, bringing about the prosperity of humanity.

So, is this a happy ending then?

It was, until just four years ago.

The emergence of LLMs has once again shaken up the semiconductor market.

Semiconductors have been growing steadily as can be seen in the chart below, with processing capacity increasing sevenfold.

However, the computational demand driven by LLMs has simply grown in an exponential fashion, and a new solution is now in order.

Wouldn’t 1.4-nanometer semiconductors suffice?

You're right.

Semiconductor companies like Samsung, SK Hynix, and TSMC are fiercely competing to develop increasingly smaller ultra-fine process technologies: 3-nanometer chips, followed by 2-nanometer, 1.6-nanometer, 1.4-nanometer, and so on…

The use of such ultra-fine process technology is expected to improve performance.

The reason semiconductor companies are striving to make smaller chips is a continuation of the existing development trajectory in the semiconductor industry.

We have discussed how, by creating MOSFET transistors smaller than BJT transistors on silicon wafers, it became possible to perform more calculations with smaller computers.

Similarly, by making transistors even smaller than MOSFETs and embedding them on silicon wafers, smaller computers can handle even more calculations.

In this way, the performance of semiconductors has increased by increasing the number of transistors on the wafer, i.e., by enhancing integration density.

But can this trend of miniaturizing transistors through ultra-fine process technology continue indefinitely?

I personally believe we have reached the limits of enhancing semiconductor performance solely through ultra-fine process technology.

In fact, semiconductor devices like transistors have been evolving for decades. In the 100-nanometer era, we used the MOSFET (Planar FET) devices described earlier. For semiconductors using 10-nanometer-scale fine processes, next-generation FinFET devices were introduced, and for semiconductors using ultra-fine processes below 7 nanometers, GAA devices are utilized.

The core of this evolution lies in changes to the Channel structure.

As mentioned before, a transistor consists of a channel through which electricity flows and a switch that controls this flow, known as the Gate.

As shown in the diagram, the MOSFET has a flat channel where the Gate influences the flow of electrical signals.

However, as integration density increases, the channel length becomes shorter. The problem is that as the channel length shortens, the Gate’s influence on the channel diminishes.

This can result in electrical signals flowing independently of the intended On/Off state, rendering the carefully engineered semiconductor properties ineffective.

This phenomenon is known as the Short Channel Effect (SCE), and overcoming it has been the goal of semiconductor companies for the past few decades.

When we talk about the 5-nanometer, 3-nanometer, and 2-nanometer processes, we are referring to the channel length, highlighting the importance of this aspect.

To overcome the limitations of SCE, new devices have been developed.

Unlike the MOSFET, where the Gate influences the channel from only one side, the FinFET was designed to raise the channel in the shape of a shark fin, allowing the Gate to influence the channel from three sides. The GAA (Gate-All-Around) takes this a step further by wrapping the Gate around all four sides of the channel, providing even greater control.

This has allowed us to reach the 3-nanometer and 2-nanometer scales.

However, the problem is that we’ve now wrapped the channel on all four sides - further advancements in device design are becoming increasingly difficult.

Additionally, the yield rate for 2-nanometer and 3-nanometer processes currently hovers around 50%, far below the 70% threshold which is considered stable.

Even if we manage to improve yield rates and stabilize 2-nanometer technology, another issue arises.

At scales smaller than this, we enter the microscopic world where quantum mechanics, rather than classical mechanics, governs behavior.

The mathematical models of current silicon-based transistor structures are based on classical mechanics, but these models lose relevance in the quantum realm.

We lose control over the transistor itself.

So, does this mean the end of technological progress for humanity?

No, not at all.

Researchers around the world are still testing various methodologies to overcome these limitations.

If we’ve reached the limit of increasing integration density on 2D silicon wafers, we can stack them vertically!

The era of 2D Integration has ended, and the era of 3D Stacking has begun.

Alternatively, waiting for material or device-level innovations might take too long.

Instead, we could design semiconductors specialized for specific tasks, sacrificing versatility to push semiconductor performance to its extreme.

Some believe this won’t take as long as anticipated.

By leveraging AI to simulate materials at the molecular level, we might discover new materials with even better performance than expected.

Perhaps we could even unlock the secrets of superconductors which have eluded us thus far.

Kakao Ventures believes that there are still countless opportunities in the mature and colossal semiconductor market.

Especially if a strong market demand, like the emergence of LLMs, disrupts the semiconductor ecosystem, this market will be all the more worth watching—even for tiny startups.

Currently, many deep-tech startups are challenging the semiconductor market. We meticulously analyze the semiconductor industry's value chain and research methodologies, convinced that another significant hardware change is on the horizon.

As a VC investing in early-stage startups, we actually look much further into the future than would be expected.

In a world beyond AI semiconductors, who will take Nvidia's place in the Post-2D Integration and Post-LLM era?

We are constantly asking ourselves which areas to invest in and what skill sets a startup needs to survive and thrive in this ecosystem. A materials startup, for example, must meet the five conditions for commercialization mentioned above.

Even if not all skill sets are secured, that’s okay.

Kakao Ventures supports long-term, fundamental attempts to change the world rather than short-term profit models adapted to the current macroeconomic situation.

If we share an understanding of fundamental problems and solutions, we want to be a co-pilot through the challenges that follow.

In this spirit, we encourage aspiring deep-tech entrepreneurs, both domestic and abroad, to take bold and experimental risks fearlessly.

The next topic I’d like to discuss is "Next-Generation Semiconductors – HBM."

I had hoped to cover everything in this article, but it looks like we’ll have to continue in the next one!

Stay tuned.