Rice University physicists used a grain-of-sand-sized magnet levitated above a superconductor cooled to near absolute zero to search for ultraheavy dark matter. Results presented in September 2026 found no signal but probed particles approximately 10 million times heavier than those examined in previous levitated-particle experiments.

A tiny magnet floating above a superconductor has given physicists a way to test an unusually heavy set of dark-matter candidates. The detector did not identify dark matter. What it demonstrated was that a small mechanical sensor, kept exceptionally quiet and watched for long enough, could investigate masses far beyond the reach of earlier levitated-particle searches.

Rice University announced the results on September 4, 2026, describing work presented at a particle-physics and cosmology conference. The collaboration’s research preprint, posted in August, sets out the experiment and its limits. Its advance is a roughly ten-million-fold extension in the dark-matter mass range accessible to levitated sensors, not the discovery of a particle ten million times heavier than a previously detected one.

A floating target waiting for a push

The POLONAISE experiment uses a permanent magnet on the scale of a grain of sand, held in a superconducting tantalum trap. The levitated assembly weighs 0.356 milligrams and includes a small glass sphere attached to the magnetic material. With no ordinary mechanical support rubbing against the moving target, tiny forces can leave measurable traces in its motion.

The cold surroundings help make superconducting operation possible. A superconducting pick-up coil senses changes in magnetic flux as the magnet moves, and an exceptionally sensitive device called a SQUID reads the resulting signal. The instrument therefore watches motion rather than waiting for a flash of light or an electrical recoil inside a conventional particle detector.

The target was small, but the effort to keep it still was substantial.

According to the paper, the cryostat rested on a 25-tonne concrete block supported by pneumatic dampers. Inside, a suspended system further reduced vibrations. Pumps and cooling machinery were mechanically separated as far as practicable, because ordinary laboratory disturbances could imitate the impulse being sought.

Why heavier dark matter needs a different search

The analysis considered dark matter that could interact with neutrons through a hypothetical new force carried by a very light mediator particle. If such a dark-matter object passed close to the sensor, the combined response of many constituents in the magnet could produce a detectable momentum kick.

This is a particular interaction model, not a detector for every conceivable form of dark matter. The force’s strength and range affect whether a passing object would move the target enough to register. The researchers also modelled whether interactions with the atmosphere would slow incoming particles before they reached the apparatus.

Mass introduces another difficulty. For a fixed amount of dark matter in a region, making each object heavier means there are fewer objects passing through. A search for rare, massive visitors needs a suitable target and sufficient observation time, as well as sensitivity to a single encounter.

The paper reports sensitivity across roughly nine orders of magnitude in candidate mass. Its upper reach extends about seven orders of magnitude beyond earlier optical-levitation searches, whose suspended targets were much lighter. Seven orders of magnitude is the factor of ten million in Rice’s announcement.

A month of monitoring, nine days of selected data

Measurements ran from December 22, 2025, to January 21, 2026. The team selected overnight periods, between 7 p.m. and 7 a.m. local time, when external activity was expected to be lower. It also removed calibration periods and dates affected by apparatus work or a deliberate vibration test.

That left 219.66 hours of usable exposure, equivalent to about 9.15 days. The distinction matters: a month-long observing run does not automatically provide a full month of equally useful measurements. The published limits account for the selected exposure and for the probability of recognising an impulse of a given size.

Eight candidate impulses survived the main selection.

They were consistent with instrumental disturbances, and their behaviour correlated with laboratory activity. The researchers had no reliable model that would let them subtract those disturbances as a known background. Instead, they treated every surviving candidate as potentially due to dark matter when calculating conservative upper limits.

What a null result can establish

The outcome therefore does not mean the magnet never moved. It means that the observations supplied no confirmed evidence for the proposed dark-matter interaction. The remaining events could still be used to ask how strong such an interaction could be before the model predicted more activity than the data allowed.

The collaboration reported exclusions at the 95 per cent confidence level for specified combinations of particle mass, mediator properties and interaction strength. Those qualifications define the result. They cannot be discarded to claim that ultraheavy dark matter as an entire category has been ruled out.

There are boundaries at both ends of the search. A particle must deliver enough momentum to cross the detection threshold. At very high masses, encounters become so rare that the available exposure cannot support an exclusion. Strong interactions can also make the atmosphere an obstacle, while a sufficiently short force range weakens the response of the macroscopic target.

From repeating waves to isolated encounters

The same broader technology had already supported a different investigation. In Rice’s June 2025 account of an ultralight-dark-matter search, researchers looked for a repeating force associated with a wave-like field. That experiment also found no evidence for its target signal.

The ultraheavy analysis asks a different question of the motion: whether a passing object gave the sensor an isolated kick. Reusing sensitive levitation technology does not make those two hypotheses interchangeable, but it shows how one experimental platform can address widely separated possibilities.

Rice says the team plans further cooling, longer measurements and multiple levitated magnets. Several sensors could help distinguish a possible passing particle from a disturbance affecting the laboratory. For now, the achievement is a measured expansion of experimental reach: a small floating magnet has tested a region that earlier levitated targets could not cover.