As of 2020 MOCVD technology has gained widespread use across the globe. The majority of Compound Semiconductor (CS) based products are made employing MOCVD processes. The CS market is growing by double digit numbers, as the areas of possible applications are widening. However, CS-based products still face growth limitations in important market sectors such as microprocessor technology, commodity electronics and to some extent also power and RF electronics. They have gained limited penetration depth in special applications where other solutions simply do not work.
The situation has started to change with the introduction of GaN-based devices. GaN technology gained momentum around 2009 due to the rapidly expanding usage of so-called white LEDs the market volume of which has been growing continuously. By now, ten years later, an improvement in LED efficiency and further technology developments have enabled a phase transition towards micro-LEDs. It cannot be excluded that within the next one or two decades yet another phase transition might occur. The broadly spread usage of white LEDs stimulated the development and expansion not only of MOCVD equipment, but the whole corresponding infrastructure as well. The metal-organic materials applied in MOCVD gas supplies have become more or less commodity goods, especially in countries with well-established semiconductor industry.
Nevertheless until now CS based devices face the following dilemma: They are chosen due to their superior functionality and performance whenever other technology options fail whereby their higher production costs are grudgingly accepted. However, the rule that everything that can be done in Si will be done in Si has not yet been broken. Si-based technology is decades older than its CS counterpart. It is more mature, has a much higher market volume and, most importantly, is cheaper in production per unit base. It dominates electron-based device fabrication, non-surprisingly, but also the manufacturing of photo-electronic devices, especially of photovoltaic applications, where it remains almost without alternative when comparing the power/cost ratios of available technology options. Nothing else comes even close. Unfortunately for CS materials, only them possess a direct bandgap which improves photon-to-electron conversion by a huge factor.
What would it take to create a completely different market situation where everything that at the moment Si-based technology is used for would be done employing CS type materials? We firmly believe that the only obstacle preventing such a decisive shift towards CS materials are too high production costs of the latter. Many semiconductor companies are in principle on the look-out for alternatives to “jump off the Si needle”. There exists, for example, a need for higher operation frequencies for CPUs at lower power consumption. CS-based devices would be able to provide these desired features yet the potential product end price prevents their replacing of Si-based devices. Over the past 40 years customers have gotten used to pay the same prices for a doubling in performance every time they are ready for the next upgrade. The necessity to pay more for improved performance and/or lower power consumption will stop them from purchasing the next upgrade. As a matter of fact, the motivation to buy a faster PC or laptop has declined significantly during the past decade. A slow-down of new, enlarged and more complex software releases has become rare as modern multi-core CPUs are just too powerful.
However, the arrival and expansion of new technology sectors will help to enhance the need for CS compound devices. The Internet-of-Things (IoT) market drives power consumption demand to the lowest possible values. State-of-the-art mobile phones which have less and less to do with phones of the old times are actually the most widespread use of IoT devices, although they are generally not recognized as such. On-battery time remains on the list of most crucial features since the introduction of the first smartphone. Another technology-driver on the rise is Artificial Intelligence (AI) which requires the highest computing power possible while remaining within a reasonable range of electric power consumption. IoT and AI as well as other related areas such as the implementation of autonomous driving could be forces pushing CS technology back to the game.
In our opinion improvements in the CS area have to occur, otherwise the upcoming surge in demand derived from the above mentioned innovative technologies will be served as usual by Si-based devices. Therefore we started Epistrome LLC in order to promote a strategic inflection point as mentioned in the book ‘Only the Paranoid Survive’ written by the former INTEL CEO Andrew S. Grove. Our main target is MOCVD processing, being a major portion of CS device fabrication. We are not the first to see the problem, but so far MOCVD processes were considered untouchable or not flexible enough for an upgrade. Most researchers focused instead on replacing MOCVD for a given material or process step. Such an approach does not seem to promise real improvement. It will lead to technology fragmentation while technology integration should be the real goal.
We are not wizards and we fully understand that a significant cost reduction will be only achieved if a mental shift regarding the assessment and definition of the real processing requirements will happen. Otherwise improvements may still occur, although on a relatively limited scale. Generally speaking, we have to ask ourselves: If MOCVD technology will still exist in 10 - 20 years, how would it look like? Would it be the same as we know it today? That is the question our company is looking forward to answer.