In a press release, Microsoft announced what they call a “historic milestone” for their Azure Quantum program – the confirmation that they were able to obtain Majorana particles in their devices. The Majorana particle was first theorized to exist in 1937 by Ettore Majorana but has been very difficult to confirm experimentally. However, Microsoft realized that if they could build a quantum computer based upon this technology, it would have some very interesting properties. Such a computer could be a topological quantum computer that would have built-in error protection because the information is encoded in the topology of the device rather than some specific physical characteristic that can be disturbed by environmental noise. This could enable qubits with error rates that have orders of magnitude lower error rates and make it much easier to build a fault tolerant quantum processor.
But what has been called the Majorana Zero Mode has been very difficult to build and verify and Microsoft has been working for a couple of decades now to show that these were real. In 2018, their team in the Netherlands thought they were able to show Majoranas existed, and they published a paper in Nature showing data that they felt proved the particle existed. But subsequent analysis showed there was an incorrect analysis of the data and the Miicrosoft team had to retract the paper, much to their embarrassment.
However, Microsoft doubled down and persisted in researching this area. They brought in additional people onto the team with contributors all over the world. They also used their massive simulation capabilities within their Azure Cloud to create computer models and simulate the physical characteristics in order to find the right materials and structures that would indeed exhibit what they call the Majorana Zero Mode and would prove that the theory was correct. Microsoft was extra careful in this effort and started by creating some criteria and published a paper describing the protocol to identify a topological superconducting phase. They then carefully reviewed the data both with their large internal team and outside experts and concluded that they had finally been able to succeed.
MIcrosoft has issued a press release, a research blog, and produced a video announcing their results, but we fear some unrealistic expectations may have been formed by general public. It is important to put these results into the proper perspective and caution that one should not believe all the hype that Microsoft will be introducing a topological based, error corrected quantum computing any time soon. Here are our concerns about expecting too much too soon.
- The Discovery Has Not Yet Gone Through a Peer Review in an Accredited Industry Journal
It’s great the Microsoft has had a great many internal people and few outside experts review the data, but it is important they try to be as transparent as possible and provide the data to the outside world. There could also be things that have been missed and it is important to have as wide of an audience look at the data as possible. To Microsoft’s credit they did publish a paper titled Protocol to identify a topological superconducting phase in a three-terminal device that describes how they will identify the existence of the Majorana. They have also indicated that they are planning to publish something in an industry journal but without giving out all their IP. As a model, we’d like to suggest something similar to the Google Quantum Supremacy paper published in Nature that sets a good example of how these types of developments can be made as transparent as possible, explained, and backed up with solid data.
- It Will Still Take Many Years for Microsoft to Build a Quantum Processor Based on the Majorana and It is Still Not a Sure Thing They Will Succeed
Demonstrating a Majorana Zero Mode is not the same as creating a qubit. Microsoft must first create a physical qubit, then combine multiple ones together to create a logical qubit and then combine millions of logical qubits together to create a processor. Large number of steps will be required to make this happen and the company is just starting the process. And there are still many unknowns that could present unanticipated problems which could substantially delay or even totally prevent the development of a reliably working processor.
- Although Topological Qubits Greatly Reduce Error Rates, They Will Not Take The Error Rates to Zero
There may be some who believe that a topological quantum computer will be completely error free and not require error correction. That is not true. It’s believed that the topological computers will great reduce the number of physical qubits per logical qubit, but no one really knows for certain by how much. Microsoft is researching something called a Floquet code for use with Majorana based qubits. There could be issues like noise in the signals from the control electronics or other unknown issues that could cause the resultant error rate to be higher than anticipated.
- Current Yields are Still Pretty Low
In the video, Dr. Chetan Nayak indicated they were only seeing the Majorana zero modes on a little more than 25% of the devices. Apparently, the fabrication process requirements to successful create the Majorana’s is quite stringent, and the yields will need to improve substantially before these structures can be used to create a reliable quantum processor.
One other thing about Microsoft’s approach that we found interesting was their intention to build a multimillion qubit processor as a single module. Most of the other approaches that we’ve heard about, including the efforts at IBM, Google, IonQ, and PsiQuantum to name a few, are looking to create these large processor clusters using multiple modules that are networked together using a mini quantum internet. There could be speed, size and other advantages to Microsoft’s monolithic approach, if works.
The other thing to mention is that Microsoft is working on two other important technologies that will be also important for their topological based quantum computer. Besides the Majorana based qubits themselves, they are also working on the aforementioned Floquet based error correction code to eliminate any remaining errors that are not corrected by the hardware. Another important technology is cryoCMOS based control electronics. For these large systems, it is imperative to find a better solution to route the control signals to the qubit chip than the rat’s nest of individual cables that are used today. By developing a low power cryoCMOS chip that can run at low temperatures the cabling can be made a lot simpler because the control chip is located very near or maybe even on the same die as the qubits.
So overall, we recognize that the Microsoft team has put in a lot of effort and hard work to get to this stage, and we congratulate them on achieving this milestone. However, we also recognize that there still is a lot of additional hard work needed to create a working processor and there is still a lot of work left to be done.
March 19, 2022