The present VLSI electronic systems rely on the Silicon MOS (metal oxide semiconductor) technology which advances will soon come to saturation. Carbon nanotubes represent an advancement in the materials technology with the potential for providing switching devices that may be faster and smaller than the present MOS devices. Carbon nanotubes are miniature tube structures with intriguing characteristics. The tube, in the normal untwisted state, conducts electricity. When twisted, the tube acts as a semiconductor.
This transistor is considered one of the greatest inventions of the twentieth century. It has helped to bring about both the information and computing age. One reason for success is its ability to decrease in size and increase in speed. This property is summarized in Moore’s law. It states that the transistor’s size will decrease exponentially while the speed will increase exponentially. Moore’s law has allowed the technology sector to progress and remain competitive.
The physical barriers arise due to the continued shrinking of the current transistor used today, the Metal-Oxide Field Effect Transistor or MOSFET. As the size shrinks, the thickness of the insulators reduces. Insulators are used to electronically isolate parts of the transistor. With the thinner insulation, the carriers can quantum-mechanically tunnel across the insulation. The result is a short circuit allowing current to flow directly from the source to drain and then drain to the body. And even though the thin gate oxide, that separates the gate from the channel. In addition, doping becomes a problem since it relies on percentages. If the total amount of atoms gets very small, then a fractional dopant atom might be required, which of course, is impossible. In addition, economic problems arise from producing and maintaining the fabrication lines. One proposed solution is the use of carbon nanotubes instead of silicon to make the transistors.
The construction and operation of CNFET are similar to the MOSFETs that we use today, thus giving them the name Carbon Nanotube Field-Effect Transistor or CNFET. Three of the most important characteristics of any transistor are speed, scalability, and power.
Speed: The carbon nanotubes unique one-dimensional nature; can utilize ballistic transport. Ballistic transport means that the mean free path is longer than the path. Thus, the charge carriers do not collide, reducing resistance to negligible levels. The result is a capability to achieve speeds of Terahertz or more, compared to today’s processors that operate at 3 gigahertz.
Scalability: A group in IBM discovered an interesting property of the CNFETs scalability. While the CNFETs improve with scaling, it is not conventional. They seem to follow the behaviour of Schottky barrier MOSFETs instead of regular MOSFETs. For this reason, the group at IBM feels that the CNFETs limits for scaling are unclear. However, they do note that, in a structured array, the CNFETs will produce enough gain and fan out for real-life applications. In addition to the CNFETs murky limits of scaling, it still will outperform silicon MOSFETs limits of scaling.
Power: The same group at IBM compared some properties of the CNFET to both a high-performance silicon MOSFET and a newer MOSFET design that utilize Silicon-On-Insulator (SOI) technology. The results are displayed in table 1.
Table 1: Comparison between MOSFET and CNT
|Circuit||FET||Delay (in pico second)||Power (in micro watt)|
|2 Input NAND||CMOS||24.32||20.67|
|2 Input NOR||CMOS||39.26||22.13|
One important difference is in I(OFF). The CNTFET has a drop of about 70% as compared to the conventional MOSFET. That emphasis on power being wasted while the transistor is off is greatly reduced. In addition, we notice that I(ON), or drive current, is larger than both technologies. In fact, it is three to four times larger. Normally, we would think that this is a bad thing. As our first instinct would mean higher power consumption. However, since the nanotube has ballistic conductance, it actually has a smaller resistance. Thus, the power consumption is the same if not smaller than the current MOSFET design. This is also supported by the two to four times increase in trans-conductance. The real big surprise is that the CNTFET is able to outperform both the current and newer technologies, despite the large gate length and gate oxide thickness. So naturally, when the CNFET design is optimized, the CNTFET will surely outperform the current technology. For these reasons, the CNTFET is a very strong contender to replace the current technology.