TREND Transistors: Carbon now outperforming silicon

| |
1 Star2 Stars3 Stars4 Stars5 Stars

Since their discovery, carbon nanotubes have continued to fire the imagination. Now they also seem to be slowly but surely taking over from silicon as a transistor material.

A twenty-year old dream has now become reality. For the first time, transistors made from carbon nanotubes can outperform the top dog silicon. Scientists at the University of Wisconsin-Madison (UW Madison) recently reported they had achieved 1.9 times higher currents than with silicon transistors.

The fact that the microscopically small tube-shaped carbon structures have outstanding electrical and thermal properties isn’t new and is largely due to the “miracle material” graphene. Nowhere do electrons move so quickly and impact-free than in the two-dimensional modification of carbon. When rolled, they form carbon nanotubes (CNT).

Nanotubes (Image: Mstroeck/wiki commons)
Like graphene, the walls of carbon nanotubes consist only of carbon, with the atoms taking on a honeycomb-like structure with six corners and having three binding partners. (Image: Mstroeck/Wiki Commons).

As opposed to the two-dimensional graphene, they also have significant semiconductor properties. But unfortunately, that’s not all. The three different production processes (gas decomposition, arc and laser deposition) enable the “growth” of semiconductor and also metallic conducting tubes. Measuring up to several millimeters in length, their diameter varies between 0.4 and 100 nm, depending on whether they are single-walled CNT or multi-walled CNT.

The battle of the short circuit

While nanotubes with metallic conductivity are desirable for interconnects in microchips, transistors need only the semiconductor variant. And they have to be very accurately aligned.

Using polymers, scientists at UW Madison were able to separate the semiconducting CNTs from the conducting ones and thus keep metallic “contamination” below 0.01 percent. The correct alignment is achieved with a process called floating evaporative self-assembly, which is based on the self-organizing properties of the nanotubes.

Carbon nanotubes with bottlenecks

When, finally, the scientists had created almost pure semiconductor nanotubes with all their wonderful properties, the next problems arose. The current has to exit the transistor again via contact surfaces, and with the dimensions and resistance differences between nanotubes and electrodes, that was the next challenge.

The scientists in Wisconsin overcame this problem by removing the polymers, which not only separated the conducting from the semiconducting tubes, but also acted as an insulator between the electrodes and the tubes, exactly at this position.

IBM scientists solved the problem a year earlier by fusing the carbon and metal atoms of the contact by a chemical method. This allowed the contacts to shrink to a ten millionth of a millimeter (10 nm) with no negative effects on the performance of the carbon transistors. As a result, IBM again forecast that nanotube technology will be mature by 2020.

In Germany, the Institute for Semiconductor Technology and Nanoelectronics (IHTN) at Darmstadt University of Technology is investigating the production and integration of CNTs in CMOS technology. The aim is to link the nano dimension to the macroscopic environment. In 2010, functioning CNT field effect transistors with diameters of 1-3 nm were produced through controlled growth on a silicon substrate prepared using CMOS technology.

First nanotube computer

At the end of 2013, a team of scientists at Stanford University announced that it had developed the first computer with a carbon nanotube processor. In terms of functions, the prototype with 178 transistors is comparable with a computer from around 1955, said Prof. Franz Kreupl from Munich University of Technology (TUM).

Carbon Nanotubes
3-D chip connects four layers with standard vias, with the bottom being standard CMOS, the top being carbon-nanotube logic transistors, and the middle two layers of resistive random access memory (RRAM). (SOURCE: Stanford University, Mitra/Wong Lab).

The scientists eliminated wrongly aligned nanotubes with an algorithm and the conductive tubes “vaporized” them. This involved subjecting the processor to an even current, which mainly heated the conducting tubes. The semiconducting ones remained.

Six months ago, the scientists went one better and presented a complete CNT processor with a new 3D design. Memory and processor are united on a small area, which considerably shortens runtime between the two. This should achieve up to 1000 times faster computing speeds.

Nanoelectronics driving ahead

However, it will be several years before nanotube computers are really able to replace silicon, even though after 25 years of nanoelectronics progress is more obvious and efforts are increasing.

For example, the German government only recently decided on the Nanotechnology Plan of Action 2020, which aims to pool all nanotechnology activities across different disciplines. This year, the German Federal Ministry of Education and Research (BMBF) made €190 million available for this. In Germany, about 2,200 establishments and networks from industry, science and various trade associations are active in the area of nanotechnology.

However, according to Lux Research, China, driven by government subsidies, is the leader when it comes to carbon nanotube research. In terms of production, China has caught up with the competition and will share in what will be a $560 million market by 2025. But multi-walled CNTs will make up the lion’s share of the industry, such as in Li-ion batteries and electronic packaging.

The tiny tubes have a lot of uses. Or as Richard Feynman named his famous presentation in 1959: There’s Plenty of Room at the Bottom.


logo electronica eToday, integrated circuits are used in practically every electronic device, and they have revolutionized the world of electronics. At electronica, you will find everthing from power semiconductors, microprocessors, diodes to transistors.