Advanced Materials: Mapping A Path To Low-Power Devices

By Cheryl Ajluni

For many electronics devices, especially those utilized in mobile applications, achieving low power is the Holy Grail. Unfortunately this goal is one that is not easily attained. In accordance with Moore’s Law, transistor density is continuing to increase. With each scaling, transistors are being designed smaller and faster to realize increased chip performance. But the rising transistor count and their smaller size means increased static power consumption and more specifically, an increase in leakage current. Managing that leakage is crucial for reliable high-speed operation and consequently has become a critical factor in chip design.

Typically, designers will utilize low-power components, or employ some combination of low-power design techniques, design methodologies and design tradeoffs to address the power problem, but these methods can only accomplish so much. As a result, many are now turning their attention to materials science and the pursuit of an advanced material with the highest performance per watt of power dissipation to deal with current leakage.

Of course selecting the right material is no simple task. Silicon dioxide is currently the material of choice used in transistors and other microelectronic devices at the heart of today’s computers and telecommunications devices. As the transistor shrinks, so too must the thickness of its silicon dioxide layer. At some point, it simply becomes too thin to adequately control the transistor’s electrical current and ceases to function properly. A viable replacement material must therefore be able to shrink along with the transistor, while also effectively addressing the power problem.

Choosing The Right Candidate

Research and development of advanced materials is today an ongoing process at companies, public and government-funded research centers, and universities around the world. France’s CEA/Leti laboratory (www.leti.fr), for example, recently joined forces with IBM (www.ibm.com/chips) to collaborate on future nanoelectronics technology. A primary focus of this agreement is the identification of advanced materials for the production of CMOS-based microprocessors and ICs at 22nm and beyond. Both IBM and CEA/Leti bring to the table their extensive expertise in nanoscale materials and low-power CMOS (e.g., Silicon-on-Insulator (SOI) technologies), respectively.

To date, the ongoing research and development has produced an amazing array of materials. Thin films made of alloys are being developed as a potential replacement to silicon dioxide. Semiconductor nanowires are under research as a potential high-mobility material for future high-speed and low-power transistor applications, as well as future interconnect applications. Photonic materials from Cornell University, like bulk-less silicon-on-sapphire (SOS), have been shown to reduce capacitances and improve device isolation, enabling the design of low-power, high-speed CMOS circuits. In fact, Cornell researchers are now exploring the potential of designing 10-Gbps optical receivers in a 0.25-µm SOS process. And, at the University of Canterbury, scientists are looking into the use of films made of zirconia (as in cubic zirconia) for next-generation semiconductors.

Of course, these developments are just the tip of the iceberg. Some of the other interesting materials which are currently being researched include:

Carbon Nanotubes

Carbon nanotubes are molecular-scale tubes of graphitic carbon. This material is among the stiffest and strongest fibers known and offers electronic properties and other unique characteristics that make it attractive for electronic circuits. In fact, scientists from the University of Buffalo have even gone so far as to suggest that single-walled carbon nanotubes (SWCNTs)–extremely thin, hollow cylinders measuring no thicker than a single atom—may one day replace metals in millions of electronic applications (see Figure 1). Their reasoning is simple: SWCNTs are thousands of times stronger than metals and, according to their research, do not follow Joules Law. Instead, SWCNTs produce just 1% of the heat produced by traditional metals like copper.

Figure 1. Shown here is a 3-dimensional model of three types of single-walled carbon nanotubes. Graphic courtesy of wikipedia.

Researchers from the University of Illinois, in collaboration with radio frequency electronics engineers at Northrop Grumman Electronics Systems in Linthicum, Md., also believe the carbon nanotube has a future in the electronics industry. To prove it, this past year they successfully developed the world’s first all-nanotube transistor radio (see Figure 2).

Figure 2. Schematic exploded view of a radio-frequency transistor that uses parallel, aligned arrays of carbon nanotubes for the semiconductor. Its development was funded by the National Science Foundation and the U.S. Department of Energy. Image courtesy of John Roger, Founder Professor of Materials Science and Engineering at the University of Illinois.

Despite such developments, commercial applications of the material have been slow to develop due to its high production cost. In this regard, IBM’s work in the area of bio-directed assembly of nanoscale materials may prove useful. IBM’s researchers are working to couple biological systems like DNA and proteins with electronic materials like semiconductors and metals, in a way that makes them useful for fabricating nanomaterials (e.g., carbon nanotubes). Recently the company discovered that it could use isolated single strands of DNA (ssDNA) to disperse SWCNTs in aqueous solvents. It is now investigating the use of ssDNA to aid in the directed assembly of the SWCNTs into functional device architectures (see Figure 3).

Figure 3. IBM’s ongoing work in the bio-directed assembly of nanoscale materials has identified ssDNA as potentially useful in the production of SWCNTs.

High-k Dielectric Material

Intel believes it has identified a potential replacement to the transistor’s silicon dioxide gate dielectric—a high-k material known as Hafnium (Hf). High-k materials like Hf or Hafnium dioxide (HfO₂) have a high dielectric constant, enable better transistor performance and allow devices built on them to run cooler. When combined with a metal gate, they enable higher speed and less wasted power. Intel also has identified new metals to replace the polysilicon gate electrode of NMOS and PMOS transistors. These new materials, along with the right process recipe, are said to reduce gate leakage more than 100-fold, while delivering impressive transistor performance.

Organic Semiconducting Material

Researchers at Cornell University (www.engineering.cornell.edu/research/strategic-areas) are working to redefine the future of the IC; one that it feels will likely require the integration of traditional silicon electronics with organic semiconducting and optically active materials. According to researchers, these hybrids leverage the current high performance of silicon with the low cost and mechanical flexibility of organic materials. One such material is pentacene, a polycyclic aromatic hydrocarbon consisting of 5 linearly-fused benzene rings. This organic semiconductor is today a promising candidate for use in thin-film transistors.

While researchers believe that single organic molecules may one day be used to form the fundamental transistor, the marrying of inorganic silicon with organic molecules and polymers to produce the required hybrid devices is simply not yet well enough understood.

III-V Compound Semiconductor Material

III-V compound semiconductor materials are made up of elements found in the III and V columns of the periodic table. Because they feature higher electron mobility than silicon, transistors made of this material have better performance and consume less energy. Indium antimonide (InSb) is one example of an III-V material that could become a promising candidate in the development of microprocessors. Pioneered by QinetiQ, it has been used by Intel to develop quantum well (QW) transistors that demonstrate a 10x lower power consumption for the same performance as compared to today’s traditional transistors (see Figure 4).

Figure 4. A two-gate finger InSb QW transistor (QWFET) is shown here fabricated with a gate air-bridge using mesa isolation.

While the industry has yet to rally around a single material to replace silicon dioxide in transistors, research and development in the area of advanced materials science will continue unabated. Such materials will not only play a key role in addressing the current leakage that results when transistors shrink, but in the emergence of nanotechnology as well. They may even one day radically alter the way we look at the transistors and the computing chips they enable.