Professor John M. Martinis is one of the world's top quantum scientists. As a follow-up to our conversation a few months ago, I had a chance to talk to him while he was temporarily working with Michelle Simmons, CEO of Silicon Quantum Computing, in Australia.
Michelle Simmons is a highly regarded quantum physics professor at the University of New South Wales, with a list of credentials too long for this article. To name a few, she is the founder of Australia’s first quantum computing company, Silicon Quantum Computing, and in 2018, she was named Australian of the Year for her work and dedication to quantum information science.
Martinis has spent most of his research career investigating superconducting hardware. However, he welcomed the opportunity to work with Simmons and the SQC team. He said it gave him a fresh set of challenges and a hands-on chance to help advance quantum computing using SQC's relatively new technology: silicon spin qubits.
On the video call, Professor Martinis seemed more relaxed than the last time we talked. "I'm really enjoying working with Michelle Simmons," he said. "She's a great physicist. It's like talking to family. And I also think it's good for Michelle to talk to me. Although we think a lot alike, we also approach things differently. That way, we're able to get to the truth of the situation."
As for Martinis’ background, he received his Ph.D. in physics from UC Berkeley in 1985. His early research on the Josephson junction became foundational for quantum computing and quantum sensors. After working for France's atomic energy commission and later at NIST in Boulder, Colorado, he eventually moved to the University of California Santa Barbara, where he held the Worster Chair in experimental physics until 2017. While at UCSB, Martinis was the adviser to a team of graduate students and postdocs who shared his passion for building a useful quantum computer. Between 2002 and 2016, Martinis and his group published over 120 papers relating to quantum computing. There was consensus in the burgeoning quantum ecosystem that Professor Martinis and his team had developed a superior superconducting qubit. Martinis' work got the attention of Google.
Google first became interested in quantum computing in 2008. While lacking quantum expertise, it did not lack financial resources to acquire it. In 2017, Hartmut Neven, director of Google's AI group, hired Martinis and his entire UCSB quantum team. It was a good match because Google's objective coincided with Martinis' goal. They both wanted to build a quantum computer that could achieve an elusive benchmark called quantum supremacy.
At that time, Google also had an aggressive AI and machine learning research program. Nevin believed quantum computing was necessary to exploit the potential of AI and machine learning fully.
Fast track to supremacy
It took about three years to build the world's first quantum computer to achieve quantum supremacy. It was a 54-qubit specialized superconducting computer, exponentially faster than any classical superconductor computer. The quantum processor, named Sycamore, took 200 seconds to perform operations that would take a classical supercomputer thousands of years to complete. Martinis solidified his place in history on October 23, 2019, when he and his team published a paper in Nature entitled "Quantum supremacy using a programmable superconducting processor."
In my previous discussions with Professor Martinis, he attributed his success as a physicist to his laser-like focus and unwavering determination to solve problems and achieve big goals. He defines himself as a definite optimist, a rarity according to Peter Thiel's book, Zero to One. While that personality trait helped him conquer quantum supremacy, it also created friction at Google.
The next significant quantum mountain to climb meant developing a quantum computer equipped with self-correcting errors and containing thousands or even millions of high fidelity logical qubits. Despite the difficulty, Martinis and the Google team had ideas about how to do it. However, perfecting the Sycamore processor had been a long and complicated task. The pressure and intense focus on achieving quantum supremacy had adversely affected some members of his group. In early 2020, Nevin suggested that Martinis become a science adviser to the quantum hardware group rather than its leader. Because his ultimate goal was for the group to be successful, Martinis reluctantly agreed to the rearrangement. By April 2020, Martinis realized the change wasn't working, so he decided to resign. You can read the transcript of his discussion with me about his resignation here.
Simmons and the SQC team have begun working on a silicon spin quantum computer. She has developed a unique way to swap an atom of silicon in its crystal plane with a phosphorus atom. Controlling the spin of the phosphorus atom makes it useful as a qubit. SQC has also created a two-qubit gate by placing two phosphorus atoms a few nanometers apart, making it possible to take advantage of quantum tunneling. Ironically, SQC's breakthrough silicon technology was made possible by a two-decade-old IBM scanning tunneling microscope (STM) that allows individual atoms to be seen and manipulated.
A few examples such as superconducting qubits, trapped-ions and particles of light, all leverage the same quantum properties of superposition, entanglement and interference to perform quantum computations. Even though Professor Martinis' research has mainly focused on superconductors, his years of experience and acquired knowledge apply to any quantum technology.
New research domain
I have a basic understanding of SQC and its technology. I am impressed with SQC's qubits and fabrication techniques using hydrogen, phosphorus and silicon. On the call, I commented to Martinis about the complexity of SQC's technology.
Professor Martinis agreed. "Yes, the physics and fabrication with silicon are very complex and very interesting. Complexity is fine if you use it right. That's one of the things I like about Michelle Simmons' group—it can fabricate its devices atom-by-atom. I find it very clean. There are some mysteries about the technology that are not yet fully understood, and I'm working on them."
Silicon has many isotopes, but only three are stable. Each isotope differs by its number of neutrons. Silicon 29, the most common, is currently used for quantum and solid-state applications. The number of neutrons in an isotope also determines its atomic spin. In the case of SQC, silicon 29 with 15 neutrons has magnetic properties that affect the quantum state of SQC's phosphorus spin qubits. On the other hand, silicon 28, a manufactured version, only has 14 neutrons, so it has no magnetic or electrical properties to degrade SQC's phosphorus spin qubits. Silicon 28 is a better solution, but it is challenging to produce.
NIST and others are researching the use of silicon 28. I was interested in finding out if SQC and Martinis were also experimenting with that isotope. Unfortunately, Professor Martinis wasn't able to provide much detail for proprietary reasons.
"I can't say too much about what we are doing, other than SQC is making good progress on silicon 28 devices. The advantages are well known, and researchers have worked on it for years. It will be exciting to see those devices made and measured. It is needed as the next step."
Four new ideas to improve superconducting quantum computers
In October 2020, I attended a virtual Department of Energy symposium on quantum computing infrastructure. Martinis was a panel member and mentioned he had identified four major problems affecting all qubits. Unfortunately, at the time, he couldn't talk about qubit problems or solutions.
On our recent call, I asked about the four problems he had mentioned in the symposium. I expected a more definitive answer than the one Martinis gave. "There are lots of things we need to solve, but I'm not ready to talk about three of those problems yet because I'm still working through them. But I can tell you this - one of the biggest problems [in solid-state quantum computing] is cosmic rays and gamma radiation. And now I'm ready to talk in detail about it. In fact, I'm giving a seminar on it tomorrow."
Martinis explained that cosmic rays and background radiation convert into an odd phenomenon called quasiparticles that degrades qubits and affects error correction. Martinis has modeled the entire chain of events and predicts it will have a disastrous effect on error correction. However, the professor said he had developed a method to channel converted energy away from qubits.
Based on the solution to this problem, plus the other three unexplained problems, Martinis has formed a knowledge company called Quantala, for the main purpose of protecting his IP with patents. Will he eventually seek investors? He's not sure.
Wrapping up with the elephant in the room
During the call, Martinis said little about his resignation or his treatment at Google. While discussing collaboration, I mentioned how much I valued my wife's opinions because she thinks differently from me and often sees a solution that I missed.
Professor Martinis also believes that differences in thinking can be beneficial. "Collaboration and working with people is great. You get different ideas. That's what's sad about Google and me. I know I think differently than most physicists. And somehow being different put everyone off, and just created an uncomfortable atmosphere."
Based on our conversation, I believe Martinis has put the Google experience behind him. There is no question in my mind that he has moved on. There was a great deal of satisfaction in his voice when he spoke of working with SQC.
"The best part is that people now come to my office to talk about physics. Michelle and I have 1 to 2-hour lunches to talk about physics, management and leadership and even politics. That's the reason why I'm so excited and happy now. I'm sure we will continue to collaborate because we're establishing a great collaboration and friendship and mentoring each other too. It's hard to go from research to forming a company, and it's great to figure it out with each other."
Professor Martinis summed up the substance of our entire one-hour conversation when he added: "It’s great to be a scientist again. I really like that."
- A few days after our conversation, Professor Martinis published a paper that modeled the entire cosmic ray scenario, its effect on qubits, and a solution to incorporate in future qubit design. "Saving superconducting quantum processors from qubit decay and correlated errors generated by gamma and cosmic rays."
- Although Professor Martinis declined to name the three additional, long-standing quantum computing problems, Quantala’s website provides hints. Qubit decoherence, qubit readout and qubit control are blank entries beneath the cosmic ray solution and its paper on the intellectual property page. Improvements that address these three areas can be significant to quantum computing.Silicon spin qubits hold much promise. Atom qubits have the longest coherence times of a qubit in silicon and also have the highest fidelities. SQC also has the lowest electrical noise of any system yet devised to connect to a semiconductor qubit. Even though SQC's goals for increasing the number of qubits are modest, a 10-qubit machine with those characteristics could do significant computations.
- I don't believe Professor Martinis will pursue outside funding for Quantala or any other company. He doesn't appear to be interested in running a company because it would distract from his research. If someone else ran the company, and Professor Martinis could focus only on research, that might be a different matter.