Beyond CMOS: Making Way For The Next-Generation Of Semiconductor Materials

By Cheryl Ajluni

Before the advent of the cell phone, the idea of having access to a phone virtually anytime, anywhere and in a package smaller than a human hand seemed almost impossible. Today that innovation, and others like it, has become an everyday reality.

In the process it has helped spawn a technologically-driven society that continually demands more for less and waits impatiently for the next great application. Such applications are often realized in a system on chip (SoC) by system-level designers’ intent on redefining the way we live, work and play. While their creative nature plays a key role in developing new applications, they must also rely heavily on available technology. In the electronics industry, the dominant technology is the silicon complementary metal-oxide semiconductor (CMOS) transistor, which is today used to manufacture most of the world’s computer chips.

While the CMOS transistor has long dominated the industry, it is reaching its atomic limit—the point where it will no longer be able to scale down. According to the International Technology Roadmap for Semiconductors (ITRS), a worldwide organization tasked with identifying the technological challenges and needs facing the semiconductor industry, the size limit for CMOS technology may be at 5nm to 10nm and this limit may be just 10 to 15 years away.

With such a deadline looming, system-level designers face a daunting obstacle. In order to develop faster, higher-performance, lower-cost applications, they desperately need to take advantage of the benefits that come from shrinking design geometries (e.g., higher performance, smaller area and lower power). But what happens when CMOS transistors can no longer scale? A flurry of activity is now underway to answer just that question. One possible solution lies in finding a way to extend CMOS using different silicon-based techniques (e.g., silicon-on-insulator (SOI) and strained silicon). To some, the more logical option is for the semiconductor industry to simply move to new materials and technologies. Let’s take a closer look at the scope of the challenge that lies ahead and how ongoing research in materials may offer the answer the industry now desperately seeks.

The Challenge Ahead

Since their introduction, CMOS transistors have shrunk exponentially in size in keeping with Moore’s Law, becoming progressively faster and cheaper. Eventually, due to physical and technical limitations, their ability to scale will run out of steam. With those limitations growing stronger with each subsequent design shrink, the benefits traditionally associated with smaller design geometries (e.g., increased speed and reduced power consumption) are on the decline. Further complicating matters, as transistor density increases, a whole host of challenges arises stemming from the physics of such small devices and the available manufacturing methods (e.g., leakage, variability in transistor behavior, power management and timing predictability). These challenges will only get worse as design geometries move to 32nm, 22nm and beyond.

While silicon CMOS technology will continue to be evolved for as long as is possible, increasingly the electronics industry will demand materials with new and dramatically improved properties. Such materials are critical to realizing the improved performance required to enable emerging device, interconnect, manufacturing and packaging technologies in support of new applications.

Eyeing New Materials

Semiconductor devices have traditionally relied on silicon. Over time, materials like GaAs, InGaAs and heterostructures made from Si/SiGe, GaAs/AlGaAs and others have become important. As the search for new materials continues, low-dimensional variants like carbon nanotubes, semiconductor nanowires and nanodots will also warrant a closer look. Some of the interesting materials currently being researched for future high-speed and low-power transistor applications are listed below. While this is by no means a comprehensive list, it does provide a sampling of ongoing research and development in materials for semiconductor devices.

• Gallium Nitride (GaN)

GaN is a hard, mechanically-stable, semiconductor material that is used in optoelectronic, high-power and high-frequency devices such as for high-speed wireless data transmission. Transistors based on the material operate at much hotter temperatures and higher voltages than GaAs transistors. A large bandgap ensures performance which is maintained up to higher temperatures than silicon transistors.

The first GaN metal/oxide semiconductor field-effect transistor (GaN MOSFET) was experimentally demonstrated at Rensselaer Polytechnic Institute in 2008. When doped with a suitable transition metal such as manganese, GaN is a promising magnetoelectronics (e.g., spintronics) material for magnetic semiconductors. Nanotubes of GaN are proposed for applications in nanoscale electronics, optoelectronics and biochemical-sensing.

• Graphene

Graphene is a two-dimensional crystalline form of carbon—something that was believed impossible to create, let alone exist, until it was made by physicists at the University of Manchester in 2004 (see Figure 1). Currently graphene is under investigation at the Department of Energy’s Advanced Light Source (ALS) by researchers from Lawrence Berkeley National Laboratory, the University of California at San Diego and Columbia University. They are hoping to accurately measure its electrical properties by experimenting with two types of graphene samples: exfoliated graphene on a silicon-oxide/silicon substrate, and epitaxial graphene, a layer of carbon atoms chemically deposited (and chemically bonded to) a substrate of silicon carbide.

What the researchers have found thus far is that graphene has a unique band structure whereby its electrons move ballistically—without collisions—over great distances, even at room temperature. As a result, graphene electrons can conduct electrical current that is 10 to 100 times greater than in a normal silicon semiconductor at room temperature. As Zhiqiang Li, an ALS Doctoral Fellow at UCSD, explains “by applying a gate voltage to graphene that has been integrated in a gated device, one can continually control the carrier density by varying the voltage, and thus the conductivity.” This phenomenon gives rise to graphene’s practical promise and coupled with the other results of the ongoing research is pointing the way to novel practical applications, such as tunable optical modulators for communications and other nanoscale electronics.

• High-k and metal gate

High-k materials 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.

To address the leakage problems that come with shrinking transistors, Intel has identified a new High-k material, hafnium (Hf), to replace the transistor’s silicon dioxide gate dielectric, and new metals to replace the polysilicon gate electrode of NMOS and PMOS transistors. According to Intel, these new materials, along with the right process recipe, reduce gate leakage more than 100-fold, while delivering impressive transistor performance.

• III-V Transistors

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 pioneered by QinetiQ and used by Intel to jointly develop quantum well (QW) transistors (see Figure 2). Intel also currently is researching other compound semiconductor materials.

Conclusion

There is little denying the impact of silicon-based CMOS transistors on the electronics industry. Despite its widespread impact, the physical challenges of advancing CMOS technology are creating a flurry of activity. Research and development is now underway to identify not only potential technology successors, but a wide range of new materials to enable these perspective successors as well. Leveraging these advances will be critical to helping system-level engineers create the next-generation of innovative applications that will surely inspire a new way for society to live, work and play.