The Two-Hour Atom: A Simple Refrigerator Trick Brings 100,000-Qubit Quantum Computers Within Reach
If you want to build a quantum computer out of individual atoms suspended in laser light, you have to solve an annoying problem. The atoms keep falling out.
Not literally falling. They escape the optical tweezers that hold them in place, knocked loose by stray gas molecules drifting through the vacuum chamber. The longer you need to keep your atoms trapped (to sort them into neat arrays, to run error correction, to actually compute something), the more of them you lose. And as quantum processors grow from hundreds to thousands to tens of thousands of qubits, this problem gets worse fast.
A team at the Max Planck Institute for Quantum Optics (MPQ) in Garching, Germany just published a solution that is almost embarrassingly straightforward: they pointed a refrigerator at the problem.
In a paper posted to arXiv on July 14, Max Melchner, Immanuel Bloch, Johannes Zeiher and colleagues report a neutral-atom platform that keeps individual strontium-88 atoms trapped in optical tweezers for more than two hours. That is a record for any system with full optical access. The secret is a commercially available cryostat whose cold tip, chilled to 4 Kelvin, sits about 30 centimeters away from the atoms and silently pumps away the hydrogen gas that would otherwise knock them out of their traps.
Background & Context
Neutral-atom quantum computers are one of the leading platforms in the race toward useful quantum machines, alongside superconducting circuits and trapped ions. The idea: trap single atoms in a grid of focused laser beams (optical tweezers), then use lasers or microwaves to entangle them and perform quantum gates. Companies like QuEra, Pasqal, and PlanQC (two of the paper’s authors, Sebastian Blatt and Johannes Zeiher, are co-founders of PlanQC) are betting on this approach.
The field has been scaling rapidly. Researchers have demonstrated tweezer arrays with more than 10,000 atoms. But there is a fundamental trade-off. The better your vacuum, the longer atoms stay trapped. But to achieve ultra-high vacuum, you typically need to wrap your atoms in cold, opaque enclosures. That blocks the very laser access you need to control, image, and entangle the atoms.
Previous cryogenic atom platforms have shown impressive lifetimes (6,000 seconds from one group, 3,000 seconds from another) but at the cost of severely restricted optical access. You can have long-lived atoms, or you can have good laser access. Picking both was, until now, extremely difficult.
What the Researchers Did
The MPQ team took a different approach. Instead of wrapping the atoms in a 4K enclosure, they built a vacuum chamber mostly from grade 2 titanium (which outgasses far less hydrogen than the stainless steel used in most UHV chambers), baked it at 320°C for three weeks, and then attached a commercial closed-cycle cryostat with a 4K cold tip that extends into the vacuum chamber. The atoms sit in a fused silica glass cell, completely exposed to room-temperature optics. The cold tip is 30 centimeters away. No cryogenic windows, no in-vacuum lenses, no thermal contraction headaches.
The atoms themselves are strontium-88, loaded into a 16×16 square tweezer array with 4-micrometer spacing, formed by an 8-watt laser and a spatial light modulator. After loading, the team held the atoms for varying periods and measured how many survived. They tested four cooling strategies during the hold time: no cooling at all, continuous Sisyphus cooling, pulsed Sisyphus cooling, and pulsed sideband cooling.
| Cooling Method | Lifetime (cryo off) | Lifetime (cryo on) |
|---|---|---|
| No cooling | ~200 s | ~300 s |
| Continuous Sisyphus | ~400 s | ~1800 s |
| Pulsed Sisyphus | ~700 s | >7000 s |
| Pulsed sideband | ~800 s | >7000 s |
The numbers tell a clear story. Without cryopumping, even the best cooling protocol buys you about 13 minutes. With the cold tip at 4K, pulsed Sisyphus and pulsed sideband cooling both push lifetimes past two hours. That is a factor of ten improvement from a design that leaves every optical path wide open.
What They Found
The key insight is that hydrogen is the villain. Hydrogen molecules outgas from chamber walls and are notoriously hard to pump with conventional ion or getter pumps. But at 4K, hydrogen sticks. The cold tip acts as a cryopump, capturing hydrogen molecules as they drift past and preventing them from reaching the tweezer array. When the team deliberately heated the cold tip to 30K, the pressure spiked: exactly what you would expect as adsorbed hydrogen boils off. Cool it back to 4K, and the two-hour lifetimes return.
The catch is that hydrogen eventually saturates the cold surface. The team measured the lifetime decay over 16 days and found a characteristic 1/e decay time of about 10 days. Every week and a half or so, you need to regenerate the cold tip by briefly warming it to 30K, releasing the accumulated hydrogen. After an hour, you cool it back down and the pressure drops below 2.5×10⁻¹² mbar, near the detection limit of their pressure gauge.
One finding that should make quantum engineers breathe easier: the lifetime does not degrade as you increase tweezer power. This is a known headache in other cryogenic platforms, where more laser light heats the cryogenic surfaces and reduces cryopumping efficiency. At MPQ, the 30-centimeter gap between cold tip and atoms means the tweezer light never touches the cold surface. You can crank the power to accommodate larger arrays without cooking your cryopump.
Why It Matters
A two-hour atom lifetime is not just a pretty benchmark number. It directly determines how large an array you can sort into a defect-free pattern. Sorting a random distribution of atoms into an ordered array takes time proportional to the number of atoms, and the probability of preserving a defect-free array drops exponentially with array size. Every extra second of lifetime buys you more sorting time, which buys you more qubits.
The MPQ team estimates their platform can support defect-free arrays of tens of thousands of atoms at high probability. With the optical access to install large-field-of-view objectives, they see a path beyond 100,000 atoms.
The design is also unusually transferable. It works with strontium, but the approach should port directly to other atomic species. The titanium chamber and commercial cryostat combination could be replicated by other labs without the custom cryogenic engineering that has made previous cold-atom platforms so specialized. A straightforward upgrade (attaching a larger shield to increase the 4K surface area) could extend both the lifetime and the time between regenerations.
How It Could Change Our Lives
Neutral-atom quantum computers are not destined to sit in a lab forever. If the scaling trajectory holds, a machine with tens of thousands of error-corrected qubits could tackle problems in materials science, drug discovery, and logistics optimization that are impossible for classical computers. But scaling depends on solving mundane engineering problems (like keeping atoms from drifting away) before you ever get to the exotic physics.
What makes the MPQ result interesting is that it solves a key scaling bottleneck without introducing new ones. The atoms live long enough. The optics are fully accessible. The power budget works. The design can be reproduced in other labs. That combination of properties is what turns a physics experiment into a platform.
The Bigger Picture
The neutral-atom quantum computing field is moving on two parallel tracks: improving the qubits themselves (fidelity, connectivity, error correction) and improving the infrastructure that houses them (vacuum, optics, cooling). Most headlines go to the qubits. But the infrastructure wins (better vacuums, longer lifetimes, cleaner optical access) are what make qubit improvements usable at scale. The MPQ platform is an infrastructure win, and it arrives at a moment when the field is starting to think seriously about what a 100,000-qubit neutral-atom computer would actually look like.
Limitations & What’s Next
The platform does not shield atoms from room-temperature blackbody radiation, which limits the lifetime of Rydberg states, the highly excited states used for entangling gates. Fully enclosed cryostats suppress blackbody radiation and can extend Rydberg lifetimes, but at the cost of optical access. The MPQ team notes that their design remains compatible with room-temperature microwave shielding, which could partially address this. They also point out that a simple line-of-sight block between the strontium oven and the glass cell could further improve lifetimes at higher oven temperatures, which would speed up atom loading.
The regeneration cycle (heating the cold tip to 30K roughly every 10 days) is a manageable operational overhead, but not zero. Increasing the 4K surface area with a shield would stretch that interval substantially, and the team is already planning that upgrade.
📄 Source: Kumar, Festa, Grinberg et al., “A cryogenic neutral-atom platform with full optical access and 2-hour trap lifetime,” arXiv:2607.12988v1 (July 14, 2026). Max-Planck-Institut für Quantenoptik / Ludwig-Maximilians-Universität München / Munich Center for Quantum Science and Technology.