Wenchao Xu leads the Quantum Engineering research group at Paul Scherrer Institute PSI and is a tenure-track assistant professor at ETH Zurich. Her team is developing a new hybrid architecture for quantum computers based on neutral atoms. The goal is to build large-scale arrays with thousands of data qubits using two different types of atom arrays, each optimized for different functions: an approach that has the potential to improve the correction of quantum errors, an essential element on the road to universal quantum computers. In an interview, Wenchao Xu tells us the details about their experimental setup at PSI, its potential applications, and the challenges they face.
Wenchao, what is a neutral atom-based quantum computer?
There are different ways to build quantum computers. Neutral atom-based architecture is one of the promising approaches, and it experienced rapid progress during the past few years. In this approach, individual atoms act as qubits to store quantum information in different internal states of a single atom. These atoms are cooled and trapped by lasers, which we call optical tweezers. By applying external electromagnetic fields, we can control the internal state of these atoms, enabling operations on those qubits. To scale the system up, we can assemble individually trapped atoms into a two-dimensional array, resulting in a programmable quantum processor. By controlling the interactions between atoms, we can perform multi-qubit operations, which is an essential step to construct a universal quantum computer.
What are the main advantages of this architecture?
First, all the atoms in the array are naturally identical, so it is easier to scale the system up. Second, we can create arbitrary arrangements of individually trapped atoms. This reconfigurable geometry is a unique advantage in forming highly connected qubits, opening up many opportunities for implementing complex quantum algorithms and generating large sizes of entangled states.
To have a useful quantum computer, we need a robust way to correct quantum errors, otherwise it may never actually be possible to perform better than classical computers.
You moved your lab to PSI about one year ago. What does PSI offer?
PSI is significantly investing in quantum technologies and has sufficient space for building up our platform. Moreover, at PSI, the Quantum Hub already contains superconducting qubit and trapped ion quantum computing experiments. Although my group is not part of the quantum hub, we have common interests, and we collaborate.
What is the current status of your experimental setup?
We have two vacuum systems already assembled and the progress is excellent. These systems allow us to cool atoms to near absolute zero at -273.15 degrees Celsius — a requirement for trapping and manipulating them with optical tweezers. The first one is already fully functional and currently being used to calibrate inter-atomic interactions that have not been explored before. The second system is ready for generating large-scale, dual-type atom arrays. Its long-term goal is to have many error-correctable logical qubits in order to have a robust quantum computer. The goal is to demonstrate new ways to speed up error detection and correction.
That is an ambitious goal. What are the main challenges you expect to have a sufficient number of logical qubits for running useful quantum algorithms?
There are many challenges. One major hurdle is to have a large number of high-quality physical qubits. First, having 10,000 atoms is not equivalent to having 10,000 well-controlled qubits. Achieving this level of control requires many things, including a deep understanding of interactions between atoms, and optimization on atomic state control and detection. Besides, while I'm confident that a single-unit atom array could host 10,000 physical qubits within a year or two, scaling beyond that is not trivial. We will need a modular architecture consisting of several arrays. The main challenge here is establishing quantum interconnects between remote atom arrays. Currently, there is no technically simple or scalable solution for this. We are developing a new approach where we can establish a strong atom-light interface to link remote atom arrays.
You say you use two types of atom arrays. What are the advantages with respect to using a single species?
The idea of using two types of atom arrays is to get the unique strengths of each of them. This configuration gives us the freedom to control them independently without any crosstalk. The way we control them is by sending light close to the internal transition frequencies of each atom species. This feature allows us to control one species without affecting the other and offers many possibilities regarding how to perform local quantum control and rapid quantum state measurements.
Do neutral atoms help extend coherence times as well?
Neutral atoms naturally have a really long coherence time because they are very robust against environmental noise. Therefore, it is not too difficult to achieve a coherence time in the scale of seconds. So, single-qubit coherence time does not limit us. During multi-qubit gate operations, we need to excite atoms from ground state to highly excited states. Our current bottleneck is the finite lifetime of these excited states.
It is incredible researchers can preserve coherence while shuttling atoms over a long distance. How do you do it?
It is true that this is indeed an astounding aspect. In the past, it was thought that coherence would be lost when you started moving the atoms around. But in reality, it turns out not to be that hard. We can trap atoms with optical tweezers and move them in a controlled way. When you optimize the acceleration and deceleration of the movement, atoms preserve their coherence. Usually, the atoms are in their motional ground state and, if you are careful enough, they will remain in this state throughout the whole process. It is like holding a cup of coffee while walking; you need to walk carefully so you do not spill it.
Could neutral atom-quantum computers help detect and correct errors?
To have a useful quantum computer, we need a robust way to correct quantum errors, otherwise it may never actually be possible to perform better than classical computers. However, quantum error correction is a complex task because you need multiple physical qubits to encode each logical qubit. Then, you need to figure out how to do operations between these logical qubits and detect the errors without destroying the quantum information itself.
Neutral atom-based architecture stands out doing this task. With this architecture, we can build up an arbitrary connectivity between atoms by shuttling them and implementing logical gates to carry out operations. This programmable, arbitrary connectivity is a unique feature that enables the generation of new code that shows a high performance and efficiency in performing logical quantum operations and correcting quantum errors.
What will be the main applications of neutral atom-based quantum computers?
Neutral atom-based platforms are very versatile. You can use them for either analog quantum simulation or digital quantum computation to tackle classically intractable problems in areas such as quantum materials and quantum chemistry. Also, you can use it to enhance metrology, the science of measurement. With this technology, we will be able to make quantum-enhanced precision measurements, achieving a level of accuracy that fundamentally surpasses any classical device. These high-precision measurements would help us detect subtle evidence for fundamental physics phenomena such as dark matter or gravitational waves or enhance the precision level of clocks for GPS systems, for instance.