Hundred-hertz quantum circuit iteration rate in a reusable neutral-atom array

The quest to build a practical quantum computer has led scientists to explore many different ways to hold and control the smallest units of information, known as qubits. One promising approach uses individual atoms, suspended in a vacuum by beams of light, to act as these qubits. These neutral-atom arrays are attractive because they can be scaled up to hold thousands of atoms, and the atoms can be moved around with great precision. However, for years, a major bottleneck has slowed their progress. While the atoms themselves can be manipulated very quickly, the process of checking their state and preparing them for the next step has been painfully slow. Traditionally, reading the state of an atom required a method that often knocked the atom out of its trap or heated it so much that it was lost. This meant that after every single calculation, scientists had to spend hundreds of milliseconds reloading fresh atoms and rearranging them, a process that limited how many calculations could be performed in a second.

A team of researchers has now demonstrated a way to break through this speed limit, showing that a neutral-atom quantum processor can run calculations at a rate of one hundred times per second. By designing a system that can read the state of an atom without destroying it or losing it, the team enabled the same group of atoms to be used over and over again in rapid succession. This achievement moves the technology from a slow, step-by-step process into a high-speed stream of data, bringing the performance of atom-based computers much closer to the speeds needed for real-world applications.

The experiment was conducted using a specific type of atom, rubidium-87, trapped in a vacuum chamber. The researchers arranged ten of these atoms in a single line, holding each one in a tiny, focused beam of light called an optical tweezer. To perform a calculation, the team first cools the atoms and sets them to a known starting state. They then apply a series of operations, similar to flipping a switch, to change the state of the atoms. The critical challenge has always been the final step: reading the result. In older systems, this reading process involved shining light on the atoms to make them glow. While this glow reveals the state, the light also pushes the atoms, often causing them to escape their traps. If an atom is lost, the entire sequence must stop, and the system must restart from scratch, wasting valuable time.

To solve this, the researchers built a new kind of interface using a specialized chip made of glass. This chip acts like a set of tiny, invisible tunnels that collect the light emitted by each atom and guide it directly to a detector. Because the light is collected so efficiently, the team can determine the state of the atom by detecting very few photons. This gentle approach means the atoms stay cool and remain trapped. The team tested this by running a sequence of one hundred identical operations on the same ten atoms without ever reloading them. They found that after one hundred cycles, the atoms were still present with a probability of ninety-nine point seven percent. This high survival rate meant the system could keep running without the long pauses required to reload the atoms.

The result was a dramatic increase in speed. The team achieved a rate of one hundred and one calculation cycles per second, a figure known as the quantum circuit iteration rate. Even after accounting for the rare instances where an atom was lost and the data had to be discarded, the effective speed remained at seventy-four point eight cycles per second. To put this in perspective, previous systems using similar atoms were limited to fewer than ten cycles per second. The researchers also verified that the information gathered from these repeated cycles was consistent and reliable. By combining the data from all one hundred runs, they were able to measure the behavior of the atoms with much greater precision than would be possible from a single run.

This high-speed capability allowed the team to perform a standard test of the computer's quality, known as randomized benchmarking, in just thirteen minutes. In the past, performing this same test on a ten-atom array would have taken several hours because of the slow reloading process. The experiment showed that the atoms maintained a high level of accuracy, with the operations being correct more than ninety-nine percent of the time. The researchers also demonstrated that their method could be optimized to get the most useful information in the shortest amount of time. By balancing the speed of the reading process with the need to keep the atoms safe, they found a "sweet spot" where the system could gather information more than ten times faster than conventional methods, even if those methods were perfect in every other way.

The success of this experiment relies on a clever design the authors call a "volcano" architecture, which refers to the way the light paths are shaped on the chip to map the tightly packed atoms to the detectors. This setup allows each atom to be read individually and simultaneously. The researchers believe that with further improvements, such as better light collection and faster cooling techniques, this speed could eventually reach thousands of cycles per second. This would open the door to more complex calculations and the ability to correct errors in real time, which is essential for building large-scale quantum computers. The work proves that by changing how we read the atoms, rather than just trying to make the atoms themselves better, we can unlock a new level of performance for this technology.