Ferroelectrics for energy-efficient electronics

It has always fascinated me how fast computers have improved over the past decades. It is still hard to grasp that a modern smartphone has more processing power than all computers existing in the 1960s combined—while running on a tiny battery.

How was this dramatic improvement of energy efficiency possible? And how much further can we push it? These questions were the ones that motivated me to pursue semi- conductor device research.

The answer to the first question is the miniaturization of transistors, which are the basic building blocks of every integrated circuit. By making transistors smaller and smaller, they also became faster, cheaper, and less energy hungry all at the same time. Without this continued miniaturization, things like smartphones, the internet, or self-driving cars would not exist today.

If we could prolong this trend, entirely new applications might become feasible, for example, fully autonomous sensor systems, which consume so little power that they can extract all of it from the environment without the need for a battery. On the other hand, the demand for more computing power will keep growing strongly in the coming decades. Therefore, we must keep the power consumption of our electronics in check, such that they can help combat climate change rather than exacerbate it.

Of course, there are physical limits to transistor miniaturization as we approach atomic dimensions. So, research on new materials for semiconductor devices is vital to further improve energy efficiency.

One example is the so-called gate insulator in a transistor, which enables the control of the electric current flowing through the device by the application of a voltage. For a long time, a thin silicon dioxide (SiO2) film was used for this purpose. However, through miniaturization, this critical layer has become only a handful of atoms thick, which means that we cannot make it any thinner. How can new materials help to overcome this issue?

My current research focuses on a particularly elegant solution to this problem: One can replace the gate insulator with a ferroelectric, which is a material with a switchable electric polarization. Ferroelectrics were first discovered more than a hundred years ago and now are used in many applications today, for example, as ceramics in ultrasonic transducers and infrared sensors or as thin films in nonvolatile memory devices. Surprisingly, if we integrate a very thin ferroelectric layer into a transistor, it can amplify the applied voltage, which means that less power will be dissipated. Such a device is called a negative capacitance transistor, which in principle can overcome the current limits to miniaturization.

However, most ferroelectric materials cannot be used to build negative capacitance transistors because they must be compatible with the manufacturing processes of state-of-the-art semiconductor devices. For example, the thin ferroelectric layer (less than ~5 nm) must be deposited conformally on a 3D substrate, which is only possible with a process called atomic layer deposition. Furthermore, the ferroelectric must endure processing temperatures as high as 500°C, without degrading its properties. Lastly, the ferroelectric material should be compatible with silicon technology and have good insulating properties. Therefore, well-understood materials like perovskite ferroelectrics cannot be used.

So far, only the relatively new class of hafnium oxide and zirconium oxide- based ferroelectrics of fluorite structure fulfill all these requirements. Indeed, both hafnium oxide and zirconium oxide (although not ferroelectric) are already used in most advanced integrated circuits. Since the first report on fluorite-structure ferroelectrics in 2011, researchers have made a lot of progress in understanding their unique properties and application potential. It has been well established that for ferroelectricity in these materials to occur, they must be stabilized in a special orthorhombic crystal phase. However, there are still a lot of open questions, especially regarding the application in ultrathin films for negative capacitance transistors.

In a recent report in MRS Bulletin, Prof. Salahuddin from the University of California, Berkeley, and I reviewed the recent status of the research on negative capacitance transistors with a focus on these fluorite-structure ferroelectrics. At the time, we concluded that while first experimental results were promising, the microscopic origin of the negative capacitance effect in these new ferroelectrics was not well understood. Furthermore, the observed improvement in transistor behavior was not as good as theory had predicted

Since then, there have been some significant advances in both our theoretical understanding and experimental progress on negative capacitance devices. On the one hand, it was shown that negative capacitance is an intrinsic property of fluorite-structure ferroelectrics, which means that in principle, large voltage amplification could be achievable.¹ While the ferroelectric films in that study were 10-nm thick, it remains to be seen if the same is true for even thinner films, which are needed for future negative capacitance transistors.

Furthermore, it was recently reported that negative capacitance does not only occur in ferroelectrics but also in materials which undergo a phase transition, such as antiferroelectrics.² This finding suggests that there might be many more materials that could be used for negative capacitance devices. In terms of demonstrating improved negative capacitance transistor behavior, there has also been significant progress. Cheema et al. recently reported negative capacitance in a fluorite-structure film as thin as 2 nm, which is ideal for transistor applications.³ Interestingly, it was found that these 2-nm fluorite-structure films consisted of a mixture of ferroelectric and antiferroelectric phases, which seems to be related to the negative capacitance effect.

These recent results are very promising for potential real-world applications, but there are still some open questions that require further investigation. For example, it is still unclear how the microstructure of a fluorite-structure ferroelectric film affects its negative capacitance behavior. Is a mixture of ferroelectric and antiferroelectric phases needed or coincidental? How can we optimize these ultrathin fluorite-structure films to obtain the maximum voltage amplification effect? What are the practical and fundamental limits to such negative capacitance devices? To answer these questions, we will need more insights from both structural characterization as well as basic theory.

Beyond these fundamental questions and further optimizations of device performance, we also need to shift our attention to the topics of manufacturability, variability, and reliability of negative capacitance transistors. In the end, we want to fabricate billions of devices at scale, typically on silicon wafers with 300-mm diameter. Because hafnium oxide and zirconium oxide-based materials are already used in semiconductor manufacturing, there seem to be no general roadblocks in terms of integration of their ferroelectric counterparts. However, because these films are polycrystalline, device-to-device variability might arise as a concern in very small transistors. In terms of reliability, first investigations are encouraging, as these new ferroelectric films seem to be at least as reliable as state-of-the-art gate insulators.³

In summary, the new class of fluorite- structure ferroelectric films shows promise for next-generation transistors with negative capacitance, which can be much more energy efficient than conventional devices. While some basic and practical questions still need to be addressed, the demonstration of first proof-of-concept devices is encouraging. Therefore, and due to the excellent compatibility of fluorite-structure ferroelectrics with advanced semiconductor manufacturing, it looks like ferroelectric gate insulators are here to stay.

Learn more about negative capacitance electronics in the following pages, which contain an excerpt from an open-access paper that I coauthored in 2021.

Read the original open-access paper here:

Progress and future prospects of negative capacitance electronics: A materials perspective