A prototype experiment narrows the search for invisible particles using an ingenious optical trick

Imagine trying to detect something that barely interacts with ordinary matter, emits no light, and pervades the entire galaxy. This is the challenge facing physicists searching for dark matter—the mysterious substance that comprises about 85% of all matter in the universe yet remains stubbornly invisible to our instruments. Despite overwhelming gravitational evidence for its existence, dark matter has never been directly detected in a laboratory.

Now, a collaboration of scientists working on the MADMAX (Magnetized Disc and Mirror Axion eXperiment) project has reported the first results from an innovative prototype designed to hunt for one particular dark matter candidate: dark photons. Using nothing more than three sapphire disks, a mirror, and sensitive electronics, the team has achieved a milestone that demonstrates a powerful new approach to solving one of physics' greatest mysteries.

The Dark Photon Hypothesis

Dark matter reveals itself only through gravity. Galaxies rotate too fast to hold together based on their visible matter alone. Galaxy clusters bend light more than they should. Large-scale structure formation in the universe requires far more mass than we can see. All evidence points to an invisible component that outweighs ordinary matter by more than five to one.

But what is this dark matter made of? One intriguing possibility is the dark photon, also called the hidden photon—a hypothetical particle that emerges from extensions of the Standard Model of particle physics. Just as ordinary photons are the quantum particles of light and electromagnetism, dark photons would be the particles of a separate, hidden electromagnetic field that permeates the universe.

The theory behind dark photons is elegant. If nature possesses an additional gauge symmetry beyond those in the Standard Model—something many theoretical physicists find plausible—and if this symmetry is spontaneously broken, the result would be a massive dark photon. Unlike ordinary photons, which are massless and travel at light speed, dark photons would have a tiny but nonzero mass, allowing them to move slowly enough to clump together gravitationally and form the dark matter halo surrounding our galaxy.

What makes dark photons particularly exciting is their potential connection to cosmic inflation, the period of exponential expansion thought to have occurred fractions of a second after the Big Bang. Dark photons with masses around 100 microelectronvolts (about one ten-billionth the mass of an electron) could have been produced by quantum fluctuations during inflation. If detected at this mass scale, dark photons wouldn't just solve the dark matter mystery—they would provide direct evidence for inflation and pin down the energy scale at which it occurred, offering a rare window into the universe's first moments.

The catch is that dark photons interact extraordinarily weakly with ordinary matter. The interaction strength is controlled by a parameter called the kinetic mixing angle, which describes how dark photon fields mix with ordinary electromagnetic fields. For dark photons to make up dark matter without having been detected already, this mixing must be incredibly tiny—less than one part in ten trillion.

The Detection Challenge

Detecting something that interacts so weakly presents a formidable experimental challenge. Traditional dark matter searches often look for rare collisions between dark matter particles and atomic nuclei in deep underground detectors. But for very lightweight particles like dark photons, a different approach is needed.

Dark photons would fill the galaxy as a vast cloud of quantum waves, oscillating at a frequency determined by their mass through quantum mechanics. The formula connecting mass and frequency is straightforward: a dark photon mass of 80 microelectronvolts corresponds to an oscillation frequency of about 19 gigahertz, in the microwave region of the electromagnetic spectrum. If the kinetic mixing isn't zero, these oscillating dark photon fields would act like a tiny oscillating electric current, generating real electromagnetic waves that could be detected.

The problem is scale. For typical dark photon parameters, a single metallic mirror placed in a dark photon field would emit electromagnetic power measured in attowatts—a billion times smaller than the power of a single photon hitting your eye per second. Detecting such minuscule signals requires both amplifying the emission and reducing noise to extraordinarily low levels.

Previous approaches have used carefully designed radio antennas or superconducting detectors, each with their own advantages and limitations. But all face a fundamental challenge: at higher frequencies, the wavelength of the dark photon field becomes shorter, and the effective volume where dark photons convert to detectable photons shrinks. Naively, this volume scales as the wavelength cubed, meaning that searching at ten times higher frequency requires a detector ten thousand times more sensitive—an impossible engineering challenge.

A New Architecture

The MADMAX collaboration has pioneered a clever solution to this scaling problem: the dielectric haloscope. Instead of relying solely on a single metallic mirror, the experiment places multiple dielectric disks—transparent materials with high refractive index—in front of the mirror. Each disk interface, where the refractive index changes, emits dark-photon-induced radiation. With careful spacing, these emissions from multiple surfaces can interfere constructively, dramatically amplifying the signal.

The prototype described in this result uses three sapphire disks, each one millimeter thick and thirty centimeters in diameter, arranged in front of an aluminum mirror. Sapphire is an excellent choice: it's transparent to microwaves, has a high dielectric constant of about 9.3, and can be manufactured to high optical quality. The disks are held parallel by mechanical spacers in a fixed configuration.

The key insight is that the signal amplification—quantified by something called the boost factor—no longer depends on wavelength in the same way as traditional detectors. By controlling the spacing between disks, researchers can tune the boost factor to resonate at specific frequencies, allowing sensitive searches across a broad range of possible dark photon masses. The boost factor achieved in this prototype peaks at 640 at a frequency of 19.48 gigahertz, meaning the signal power is amplified by that factor compared to a single mirror alone.

How does this amplification work? Think of the dark photon field as trying to shake electrons in the conducting surfaces. At each interface between materials—air to sapphire, sapphire to air, and finally the mirror—some of this shaking converts to real electromagnetic radiation. If the disks are spaced correctly, the waves emitted from different surfaces arrive at the detector in phase, adding constructively like synchronized swimmers creating a bigger splash. The waves can also bounce between disks, creating resonances that further enhance specific frequencies.

Determining the actual boost factor isn't straightforward, since you can't simply inject a known dark photon field for calibration. The MADMAX team developed an innovative measurement technique using what's called the "bead-pull method." They insert a small metal bead into the space between disks and measure how it perturbs the reflection of electromagnetic waves sent through the system using a vector network analyzer. By mapping the electric field distribution throughout the volume, they can calculate how much a dark photon field would be amplified, accounting for the finite size of the bead and other systematic effects.

Inside the Experiment

The complete experimental setup resembles a precision optical system, but operating at microwave frequencies rather than visible light. The stack of sapphire disks and mirror—collectively called the booster—sits in a radio-frequency shielded Faraday cage to minimize interference from external signals. An ellipsoidal mirror focuses the emitted radiation onto a specialized horn antenna, whose position can be precisely controlled by motorized stages.

Connected directly to the horn antenna is a low-noise amplifier, the first stage of the receiver chain. This amplifier is critical because any noise added here gets amplified along with the signal in subsequent stages. Additional amplifiers and bandpass filters boost the signal further before it reaches a mixer, which combines it with a local oscillator to shift the frequency down to a range more convenient for digitization. Finally, an FPGA-based data acquisition system records the power spectrum with a frequency resolution of 9.375 kilohertz across an intermediate frequency range spanning 2.4 gigahertz.

The careful design pays off in the system noise temperature—a measure of how much random thermal and electronic noise contaminates the measurement. The prototype achieves noise temperatures ranging from 120 to 332 kelvin, depending on frequency. This variation creates a standing-wave pattern in the noise because the less-than-perfect impedance match between the receiver and antenna causes some noise to reflect back and interfere with itself. At the main resonance frequency where the boost factor peaks, thermal radiation from the room-temperature booster contributes about 40% of the total noise, with the rest coming from the amplifier chain.

An innovative feature of the data acquisition is "local oscillator hopping." Approximately once per second, the local oscillator frequency randomly shifts by about nine megahertz. This smears out any radio frequency interference in the digitized frequency range while leaving potential dark photon signals intact, since the data is realigned in the original radio frequency band before analysis. The technique effectively suppresses interference by about thirty decibels without requiring additional hardware.

The Search and Results

The prototype ran for 16.5 days starting in late December 2023, collecting nearly 9.5 billion individual spectra. After accounting for dead time from communication delays between the data acquisition system and local oscillator—something that could be improved in future versions—the effective data-taking time was 11.7 days. The location was a shielded laboratory at DESY (Deutsches Elektronen-Synchrotron) in Hamburg, Germany.

The analysis procedure follows established techniques from previous dark matter searches but adapted for the MADMAX setup. Each saved spectrum undergoes baseline subtraction using a mathematical filter that removes slow variations in power while preserving the narrow features expected from a dark photon signal. The filter is carefully designed to maintain 91% of the signal-to-noise ratio of a true dark photon while eliminating baseline drifts caused by gain changes from room temperature variations.

After baseline subtraction, the many processed spectra are combined using optimal weights that account for the expected signal-to-noise ratio at each frequency. This combined spectrum is then cross-correlated with the expected dark photon line shape. Unlike a perfectly monochromatic signal, dark photons would have a small frequency spread because they're moving in random directions through the galaxy with velocities around 154 kilometers per second. This creates a Doppler broadening that spreads the signal across roughly five frequency bins in the digitized spectrum.

The final cross-correlated data reveals sensitivity to dark photon signals with power around two trillionths of a trillionth of a watt. Examining the data across the full measurement range from 78.62 to 83.95 microelectronvolts in dark photon mass (corresponding to 19.01 to 20.30 gigahertz in frequency), the analysis reveals no statistically significant excess that could be attributed to dark photons.

The largest observed excess has a local significance of 4.3 standard deviations—seemingly impressive until you consider that searching across thousands of frequency bins makes finding such a fluctuation somewhere entirely expected from random noise alone. Indeed, the probability of seeing at least one excess this large purely by chance is about 21%, well within expectations for thermal noise fluctuations. This is similar to flipping a coin thousands of times and noticing that somewhere in the sequence you got heads five times in a row—unlikely for any specific five flips, but almost certain to happen somewhere.

Setting New Limits

From the non-detection, the team derives exclusion limits on how strongly dark photons can mix with ordinary photons. The analysis uses a 95% confidence level criterion, meaning that if dark photons exist with parameters in the excluded region, the experiment would have had at least a 95% chance of detecting them.

The limits vary with frequency, tracking the boost factor behavior. At the peak boost frequency of 19.48 gigahertz (corresponding to a dark photon mass of 80.57 microelectronvolts), the experiment excludes kinetic mixing parameters larger than 1.1 × 10⁻¹³. Across the entire scanned range, mixing parameters above 2.7 × 10⁻¹² are excluded, assuming dark photons comprise all of the local dark matter with the standard density of 0.3 GeV per cubic centimeter.

These results represent a dramatic improvement over previous experiments at these masses. The closest previous search, called BRASS-p, used a different dish antenna approach and had probed down to mixing angles around 10⁻¹² in this frequency range. MADMAX's prototype achieves peak sensitivity almost one hundred times better—nearly three orders of magnitude improvement in just this first run with only three disks.

The improvement comes from the combination of the boost factor from the multi-disk design and the relatively long integration time. The sensitivity scales with the fourth root of observation time, meaning that doubling sensitivity through longer observations alone would require sixteen times more data-taking. The boost factor provides a more efficient path to enhanced sensitivity.

Looking Ahead

What makes this result particularly significant is that it represents just the beginning of the MADMAX program. The prototype uses only three sapphire disks at room temperature in a fixed configuration. The collaboration's plans call for scaling up dramatically: adding more disks to increase the boost factor, implementing cryogenic cooling to reduce thermal noise, and developing motorized control systems to adjust disk spacing and tune the resonance frequency.

Each of these improvements can provide substantial sensitivity gains. Adding more disks increases the conversion volume without the unfavorable scaling that plagues traditional detector designs. Computer simulations suggest that a full-scale MADMAX detector with about eighty disks could achieve boost factors exceeding ten thousand at certain frequencies, roughly fifteen times better than the current three-disk prototype.

Cryogenic operation would reduce the system noise temperature by eliminating the thermal radiation contribution from the booster and improving amplifier performance. The prototype's noise temperature of 120 to 332 kelvin could potentially drop to around twenty kelvin with liquid helium cooling, improving sensitivity by another factor of two to three.

Most importantly, implementing tunable disk spacing would enable scanning across the full design range of MADMAX, from 40 to 400 microelectronvolts in dark photon (or axion) mass. The prototype used fixed spacing optimized for a narrow range, but motorized control systems under development by the collaboration would allow adjusting the resonance peak to any desired frequency. This would enable both targeted searches at specific motivated mass values and broad surveys across large mass ranges.

The technique is also relevant for axion searches, MADMAX's primary target. Axions are another lightweight dark matter candidate, hypothetical particles arising from solutions to the strong CP problem in quantum chromodynamics. Unlike dark photons, which don't require an external magnetic field, axions convert to photons only in the presence of strong magnetic fields. The same dielectric haloscope design works for axions by adding a large magnet, with the disk enhancement providing sensitivity to higher axion masses than traditional cavity-based searches can readily access.

Broader Implications

This result exemplifies a broader trend in particle physics: the expansion of the experimental toolkit for exploring very lightweight dark matter candidates. For decades, direct detection efforts focused primarily on weakly interacting massive particles (WIMPs) with masses around the proton mass, roughly one billion times heavier than the dark photons sought here. While WIMP searches continue and have achieved remarkable sensitivity, the lack of definitive signals has motivated exploring lighter candidates.

The mass range probed by MADMAX and similar experiments—tens to hundreds of microelectronvolts—occupies a sweet spot for several theoretical reasons. It's where certain production mechanisms during inflation naturally generate sufficient dark matter. It's also where "axiverse" scenarios, which postulate many light pseudoscalar particles arising from string theory, predict observable candidates. And perhaps most compellingly, it's a range that was previously unexplored simply because the experimental techniques didn't exist.

The development of dielectric haloscopes represents genuine technological innovation. The concept emerged only in 2013, and experimental demonstrations followed years later. This prototype is the first to show the technique working at these higher frequencies and with multiple disks, validating the fundamental approach and paving the way for more ambitious implementations.

The non-detection doesn't rule out dark photons as dark matter candidates—the parameter space is vast, and this experiment probed only a small slice. But it does meaningfully constrain where dark photons could hide. Each experiment that excludes a portion of parameter space forces theorists to refine their models and experimentalists to develop new strategies. The interplay between theoretical motivation and experimental capability drives progress in the field.

The Search Continues

What distinguishes successful dark matter searches is not necessarily finding the particle—that would be transformative, but remains elusive across all approaches—but rather demonstrating new methods that open previously inaccessible territory. The MADMAX prototype accomplishes this goal while simultaneously achieving best-in-class sensitivity at its target frequencies.

For the collaboration, this result marks a proof of principle. The boost factor measurements validated their modeling and analysis techniques. The data acquisition and noise characterization performed as expected. Radio frequency interference mitigation strategies proved effective. All systems worked together to achieve the design sensitivity, building confidence for the next phases of the experiment.

The path forward is clear: more disks, lower temperatures, tunable operation, and eventually, longer observation times scanning across the full frequency range. Each improvement compounds with the others, promising sensitivity to kinetic mixing parameters as small as 10⁻¹⁶ or better—nearly a million times more sensitive than current limits at some frequencies.

Whether dark photons or axions constitute dark matter remains unknown. But experiments like MADMAX are asking the question in new ways, using innovative techniques to probe regions of parameter space that previous generations of experiments could not reach. In the search for dark matter, creativity in experimental design matters as much as theoretical motivation.

The sapphire disks in this prototype, arranged with millimeter precision and read out by sophisticated microwave receivers, represent humanity's ongoing effort to illuminate the dark universe. They turned up no dark photons this time, but they blazed a trail for more powerful searches to come. In fundamental physics, understanding what's not there is often the first step toward discovering what is.

Credit & Disclaimer: This article is a popular science summary written to make peer-reviewed research accessible to a broad audience. All scientific facts, findings, and conclusions presented here are drawn directly and accurately from the original research paper. Readers are strongly encouraged to consult the full research article for complete data, methodologies, and scientific detail. The article can be accessed through https://doi.org/10.1103/PhysRevLett.134.151004