Welcome reader!

Let me briefly introduce myself. My name is Maciej, and I’m a quantum computer developer at Oxford Ionics.

It seems the quantum space has gotten busy in the recent years, eh? When I entered the field in 2014 as a Master’s student in Oxford’s Trapped Ion Quantum Information group, the entrance criterion was something like “he’s heard about qubits and opamps”. By the end of my PhD at ETH Zurich in 2021, we’d welcome a dozen or so new students each year, many of them experienced with qiskit, Lindbladt operators and randomised benchmarking. Crazy times!

Building a quantum computer is my largest scientific obsession. Interestingly, it is very easy to spend your time ‘working in quantum computing’ without actually trying to build a quantum computer very much. This happens in a variety of ways, and is usually a necessity of academic life. For example, early on in my PhD, we struggled for a long time to extend the capability of our ion trap beyond simple single-ion operations. Instead of pushing very hard to implement two-ion experiments and beyond - what you’d need to do quantum computing - we developed several interesting ideas for encoding and manupulating a qutrit (quantum three-level system). We then spent a following year implementing two experiments on quantum foundations with a single qutrit.

Was that time well spent? Some of it for sure was. It was an opportunity for me to properly understand many more subtle aspects of quantum operations, to benchmark the strength of the noise sources in our system, and to expand the control system capabilities. It generated two papers, which was important for everyone involved. However, it was a very inefficient way of expanding our quantum computing toolbox. Over 50% of time was spent digging through quantum foundations literature, trying to tell sense from nonesense. The qutrit operations worked pretty much just as well at the end of month 1 as they did at the end of month 12, and the noise benchmarking was rather inconclusive (i.e. “we’re limited by system vibrations in some complicated way”). And we delayed major system reworks, taking data over and over waiting for a high-stability period. Worst of all, very little of this capability was ever used again, and qutrit control seems of limited utility to quantum computing overall (I’ll write about this some time in the future).

This time stands in a very strong contrast to the time later in the PhD, when I got a chance to work on ion traps with integrated optics. If you want to build a quantum computer with trapped ions, photonic integration is a necessity. Earlier work from MIT and MIT LL demonstrated single-qubit operations with trap-integrated light, and the major question was whether these methods are compatible with noise and power requirements of two-qubit gates. We tackled this challenge head-on, demonstrating for the first time that low-error entanglement with integrated optics is indeed possible. The ion trap world took notice, and optical integration is now part of the roadmap for big trapped-ion QC players like IonQ and Honeywell (and Oxford Ionics, obviously). Together with beautiful work on integrated detectors from NIST and MIT LL, fully integrated ion traps suddently don’t seem so hard.

On a personal level, this work was fun, because it was aimed directly at the goal of QC. We weren’t working in a limbo, and there was no need to try to convince the community that this work was important. It also worked out ok - a nice addition, but frankly not a prerequisite for the fun. It reaffirmed my conviction that I would like to work “to build a quantum computer”, not just “in quantum computing” or “towards quantum computation”. It just so happens that this line of work became a thing in the last few years, and that’s what brought me to Oxford Ionics.