POWER ELECTRONICS The finFET and its significance in power electronics
The finFET is a transistor design which attempts to overcome the worst types of short-channel effect encountered by transistors while enabling chips to achieve higher performance metrics at a lower cost.
Since the breakthrough of semiconductors and throughout the history of integrated circuit design, Moore’s law—the observation that the number of transistors on a piece of silicon will double every two years—has remained the same. And it’s safe to say that this observation has held true.
With foundries developing more and more advanced process nodes to meet the demand from consumers, the transistor count on today’s advanced processors goes into the tens of billions. This is a far cry from the processors of the mid-1970s that had just a few thousand transistors, something which at the time was cutting-edge.
One of the key technology trends that has driven the semiconductor industry and made today’s chips possible is the adoption of finFET processes.
What is a finFET?
In contrast to a traditional two-dimensional planar transistor, the finFET (fin field-effect transistor) is a three-dimensional transistor with an elevated channel (‘fin’) which the gate wraps around.
The very first 25 nm finFET transistor operating on a mere 0.7 volts was demonstrated in December 2002 by TSMC. It wasn’t until 2012, however, when the first commercial 22 nm finFET became available, and subsequent improvements to the finFET architecture have allowed for great strides in enhancing performance and reducing area.
Due to their structure, finFETs generate lower leakage power and enable greater device density. They also operate at a lower voltage and offer a high drive current. All these together mean that much more performance can be packed into a smaller area, reducing cost per unit performance.
finFETs vs planar transistors (e.g., MOSFETs)
Designers opt to use finFET devices over traditional planar transistors such as MOSFETs for a variety of reasons.
Increasing computational power means increasing computational density. More transistors are of course required to achieve this, and this leads to larger chips. For practical reasons, however, it’s important to keep the area roughly the same.
One way to achieve more computational power is to shrink a transistor’s size, but as a transistor’s dimensions decrease, the distance between the drain and the source shrinks the gate electrode’s ability to control current flow in the channel region. Due to this, planar MOSFETs can suffer short-channel effects.
In short, finFET devices exhibit superior short channel behavior, have lower switching times, and a higher current density than conventional planar MOSFET technology.
finFET advantages and disadvantages
Over other transistor technologies, finFETs have several key advantages that make them ideal for use in applications where more power and performance are important:
- Better channel control
- Suppression of short-channel effects
- Faster switching speed
- Higher drain current
- Lower switching voltage
- Lower power consumption
They’re not perfect, though. Some of their disadvantages include:
- Difficult to control voltage threshold
- Three-dimensional profile leads to higher parasitics
- Very high capacitances
- High fabrication cost
Running out of steam
Although finFETs have had a great run, they’re quickly growing tired and will stop scaling. The technology isn’t expected to be much use beyond the 5 nm process. Given that many foundries have already reached this process and are steaming ahead towards 3 nm, it won’t be long until we’re moving on to its successor.
In the interim, it’s likely that some fabs will stay with the same node for longer for economic reasons. On the other hand, other companies will be forced to adopt new technologies due to the nature of their processors.
What will come after current finFETs, however, is not yet known. Many fabs are experimenting with new and novel technologies such as nanosheets whereas others are trying to find a workaround to take finFETs to 3 nm. One of these workarounds is moving to germanium materials for the p channel to provide a performance boost, but there are integration challenges.
One promising technology is the gate-all-around (GAA) transistor. This provides the most significant capacitive coupling between the gate and the channel. The problem with GAA finFETs is that it’s only a temporary solution; it may only last for a couple of decades. However, it’s likely to be the future substitute of the finFET, at least until somebody comes up with an entirely new transistor architecture.