Beyond Silicon: A Look at Alternative Semiconductor Materials

Shela Aboud

Feb 20, 2024 / 4 min read

This article was originally published in Fierce Electronics.

Since 1954, silicon has been at the foundation of advanced technology: its position as the basis of our electronics infrastructure is seen as widely unassailable. Yet guided as it is by Moore's law, silicon has its limits. This means that since its first application as a semiconductor, the industry has been looking at alternative materials that might take silicon’s place.

While no material has proven to be as diverse as silicon (and the industry's level of investment in silicon means it is here to stay), the fact remains that many types of semiconductor materials exist and can be utilized to provide value to different areas of microelectronics. It is interesting to keep in mind that the first transistor was not silicon but germanium; alternative materials can and do play an important role alongside silicon today. This is especially true as consumer technologies continue to shift toward electrification. Experts in semiconductor materials know that it is not a question of supplanting one material with another but rather selecting a material that is best suited to the application at hand (e.g., matches its requirements for performance, efficiency, robustness, etc.).

Here are two promising alternative semiconductor materials being used today, what applications they are best used for, and the pros and cons of each.

alternative semiconductor materials

Solutions for Extreme Environments

Wide-bandgap semiconductors are a class of materials that offer a range of advantages over silicon. These materials can operate at higher voltages and temperatures, serving as critical enablers of innovation by opening up higher communication channels and functioning in a wider range of environments that are sometimes extreme. Electronics applications in power and RF can benefit from these wide-bandgap materials by allowing for faster, smaller, more efficient, and more reliable device designs.

Two prominent wide-bandgap semiconductor materials in use today are gallium nitride (GaN) and silicon carbide (SiC). Both materials came to commercial prominence in the late 1980s with the development of blue LEDs and were made possible because of the size of their bandgap. (Bandgap is the distance between the valence band of electrons and the conduction band.) Advances in GaN growth processes over SiC ultimately led to its widespread adoption in optical technologies, such as Blu-ray DVD players, and resulted in a Nobel Prize in Physics in 2014.  

Besides having a wide bandgap, these materials also have a very high thermal conductivity which means they can dissipate heat more effectively which results in more efficient devices as lower temperatures optimize device performance. The ability to withstand higher electric fields and high temperatures makes these materials particularly attractive in power electronics applications, especially for the design of inverters, power supplies, and motor drives. Automotive applications such as electric vehicles (EVs) and plug-in, hybrid vehicles benefit greatly from the characteristics that GaN and SiC bring to power devices.

In addition to its application in LEDs and power electronics, GaN is also an important material in high-frequency devices like RF amplifiers and has led to improvements in wireless communication and 5G networks.

Meanwhile, SiC is a very hard material and possesses a high mechanical stability. Combined with its low material cost, SiC is valuable in many different industries, including for abrasive and cutting tools.

High Potential, High Complexity

Of course, many aspects of developing GaN and SiC for use in the semiconductor industry are more complex than traditional silicon. The main issues standing in the way of GaN's widespread adoption are its reliability and cost.

While GaN's 3.4eV (versus silicon's 1.12eV) bandgap makes it well-suited to high-power and high-frequency devices, it is prone to defects and dislocations during the growth process and device reliability can be problematic. This also makes it difficult and very expensive to grow large GaN-based wafers. To overcome this, many researchers’ focus on developing ways to integrate GaN onto silicon wafers, which means bringing two different crystal structures together in a way that avoids dislocations and defects. It is no easy task and often leads to wafer cracking.

As for SiC, its characteristics (namely its hardness and its fragility) make it challenging to produce. The material needs higher temperatures as well as a lot of time and energy for the crystal to grow and be processed. Surface defects can be challenging to find during inspections, especially with the widely used 4H-SiC crystalline structure, which is noted for its high transparency and high refractive index.

A New Era of Semiconductor Simulation

GaN and SiC are just two of many promising additions to the world of semiconductor materials. We expect to see a continual stream of new materials and material applications that utilize new types of physics, such as in the field of alternative memory device designs that take advantage of physics in the form of spin-driven or ferroelectricity, or even phase-change materials.

Efforts to identify 2D materials beyond graphene are yielding new classes of materials such as transition-metal dichalcogenide monolayers (TMD), which open the door to new devices. There is also the growing space of neuromorphic computing, which has profound implications for the workings of devices and computer architecture. Finally, there are cryogenic temperature applications that could help improve the electric footprint of data centers, which is becoming even more problematic with the exploding AI applications and may lead to entirely different classes of materials.

In all cases, we are talking about additions (rather than replacements) to silicon. Current and evolving technology is changing the rules around scaling. As we enter the system technology co-optimization (STCO) era, in which designs are disaggregated so components can be built more cost-effectively and then re-assembled for higher performance, we must be ready for a new level of experimentation.

Continue Reading