Recently, researchers have calculated the theoretical limit on the speed at which optoelectronic devices can operate. How close are we to theoretical limits in electronics, what did the researchers determine, and what options will engineers have looking forward as these limits are reached?

How close are we to theoretical limits?

It seems that no matter how many times the media states that we are at the end of Moore’s Law, researchers continue to find new methods to make devices smaller and faster. Many thought that going beyond the 20nm barrier would introduce too many challenges that would make semiconductors at that scale too expensive, but looking at current research, we see that there are devices on the 5nm scale already in production.

But whether it is the maximum operating frequency of a CPU, the smallest size of a transistor, or the lowest voltage at which transistors can operate, there are indeed physical limits that will eventually be reached. Whether a device at a theoretical limit is commercially practical or not remains to be seen, but it turns out that we can calculate what these limits are.

For example, the smallest transistor made from a semiconductor cannot be smaller than the physical elements that it is made from, and so you may find that the smallest transistor will be a single atom in size (most likely a bi-directional switch controlled using a photon). We also know that the speed of electrical wavefronts in a wire limits the maximum operation frequency and can thus determine the highest bandwidth possible.

But how close are we to these current limits? As it turns out, researchers are very far away from any physical limits that would prevent a device from being reduced in size or operating at a higher speed. Even though the latest transistor technology is in the 5nm range, this is still several orders of magnitude larger than the atoms that make up transistors. The same applies to operating frequency in that the fastest CPU to date was 8.429GHz (AMD FX), using liquid nitrogen cooling, but this is nowhere near the theoretical maximum frequency of electronic circuits in general, which can well exceed 300GHz (terahertz is believed to be the upper bound). Of course, in the case of CPUs, the maximum frequency of operation depends on the length of time taken for a signal to propagate through the CPU, which has estimates of around 22GHz).

Researchers determine the theoretical limit for optoelectronic devices

If one thing is clear, the future of electronic devices will be in optoelectronics, as achieving high bandwidths is easier with light than with the electrical current. Optoelectronics will most likely involve traditional silicon devices that interface with other devices via optical links for the next few decades. This may see future PCBs incorporating photonic channels that provide high speed and electrical isolation.

Recognising the importance of optoelectronics, researchers have recently been able to determine the theoretical limit on the maximum frequency of operation of optoelectronic devices. Using mathematics and application of Heisenberg’s Uncertainty Principle, the researchers determined that the maximum frequency of operation would be one petahertz which is one million gigahertz.

To arrive at this figure, the researchers took into account the shortest laser pulse that can be generated, the exact moment when this energy is generated by electrons in a laser, and its eventual absorption and generation of current by a receiver. From there, advanced computer models allowed the researchers to determine the point where Heisenberg’s Uncertainty Principle prevents accurate measurement of laser pulses and resulting current formation.

What options will engineers have moving forward?

Even though theoretical limits exist, there are practical reasons why these may never be achieved. For example, it is theoretically possible to have a 100% efficient vehicle, but the effects of thermal radiation and friction alone will make this target outright impossible. The same applies to electronic circuits, it may be theoretically possible to have circuits operate at 1000GHz, but it will unlikely be practically possible in the realm of computing.

So, suppose the real limit of electronics really is around the corner. In that case, engineers will have to turn to alternative computation and circuit construction methods to continue the trend of creating increasingly complex devices.

In the case of limitations in transistor size, semiconductor manufacturers will turn to 3D devices (which they already are) that stack dies on top of each other. Since dies are extremely thin, a device can have its transistor count doubled while virtually showing no real increase in size or weight.

In the case of connecting ICs together, the use of optical waveguides is showing real promise. Compared to typical electrical busses found on PCBs, an optical bus would offer significantly higher bandwidth, immunity to noise, and no signal emissions. This would be particularly useful when connecting CPUs to other peripherals, including RAM, GPU, and network cards. In fact, optical communication is already being used in TVs that connect to a speaker system via an optical link instead of a traditional electrical link, as it provides immunity to noise and allows for digital information to be transmitted over long distances.

For the next few decades, it is highly unlikely that researchers will hit theoretical limits that will stop them from developing devices. The advent of 3D chips, System on Modules (SoM), customisation services, and the introduction of optical links will allow for technology to continue to exponentially increase in capability.