Lessons from the Semiconductor Industry – Germanium versus Silicon



December 23, 2017, marks the 70th anniversary of the invention of the transistor at Bell Labs in 1947.  The semiconductor industry has gone through a profound evolution since that time, and there may be lessons that we can learn from its history that could provide vision on how the quantum computing industry will evolve.

The first semiconductors were built using germanium and this material dominated transistor production for the next 10 years.  Bell Labs licensed the technology a few years after the invention and many companies started production using the germanium material.   The early killer apps for these devices were battery powered hearing aids and transistorized radios.   In 1958, Fairchild Semiconductor started first commercial production of a silicon transistor.  Although silicon may have initially been harder to work with, it offered superior properties with regards to leakage current and thermal characteristics and proved to be the material of choice for future semiconductors.  Within five years, production of silicon transistors and silicon integrated circuits overtook germanium and today the industry produces over a sextillion (1021) silicon transistors every year.

So what does this have to do with quantum computing?  Today, the most popular technology for building qubits is a superconducting based technology originally created at Yale University.  As shown in our Qubit Technology Scorecard, there are 17 organizations developing superconducting qubits.  A key reason this technology is so popular is because it can leverage the heavy investments in processing techniques and manufacturing equipment that have been made by the semiconductor industry during the past 70 years.

But waiting in the wings there are other, less mature technologies that may someday overtake the superconducting technology. These technologies include ion traps, photonic, topological, NV diamond, quantum dot, and perhaps others. The reasons one of these other technologies may eventually win out could be due to superior qubit quality, better qubit connectivity, and for some of the technologies, the potential elimination of the dilution refrigerator.

At this time we can’t predict which one of these potential alternate technologies may turn out to be the best.  In fact, we can’t even predict when or if any of these will ever turn into the qubit technology of choice.  The big challenge will be to evolve the architectures, manufacturing techniques, and equipment such that the qubits can be built in large scale with good quality.  The cost per qubit won’t necessary need to be the same as a superconducting technology, but it should be close enough such that the technical advantages outweigh any cost disadvantages these qubits may have. If this does happen, it may still take many years. But we are still in the early innings of quantum computing and just like we have seen with semiconductors there is sure to be a dramatic evolution in the quantum computing technology in the decades ahead.

Leave a Reply

Your email address will not be published. Required fields are marked *


*


This site uses Akismet to reduce spam. Learn how your comment data is processed.